Export model to onnx and save its initializer with several independent files

deployment
1

Is it possible to export the torch model to the ONNX model and save its weights with several independent files? Like what Accelerator.save_model do?