adapt glm quant

Signed-off-by: whx-sjtu <[email protected]>

Author: whx-sjtu

vllm_ascend/quantization/quant_config.py

@@ -210,7 +210,20 @@ def get_scaled_act_names(self) -> List[str]:
             "gate_proj",
             "up_proj",
         ],
-    }
+    },
+    "glm4_moe": {
+        "qkv_proj": [
+            "q_proj",
+            "k_proj",
+            "v_proj",
+        ],
+        "gate_up_proj": [
+            "gate_proj",
+            "up_proj",
+        ],
+        "experts":
+        ["experts.0.gate_proj", "experts.0.up_proj", "experts.0.down_proj"]
+    },
 }