Questions about the code in FedST-join

Hello, I would like to know how to implement the unified style transfer model in FedST-join, especially using the one-hot encoding domain vector to identify the client's operation. I did not find the corresponding content in ./scripts/train_test.sh, and which file should I look for the specific code implementation?