aten::native_group_norm_backward is not currently supported on the DML backend and will fall back to run on the CPU

[Inokinoki]

Got user warning but trainning failed:

/***/venv/lib/python3.10/site-packages/torch/autograd/graph.py:769: UserWarning: The operator 'aten::native_group_norm_backward' is not currently supported on the DML backend and will fall back to run on the CPU. This may have performance implications. (Triggered internally at /__w/1/s/pytorch-directml-plugin/torch_directml/csrc/dml/dml_cpu_fallback.cpp:15.)
  return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
Traceback (most recent call last):
  File "/***/sd-scripts/train_network.py", line 1246, in <module>
    trainer.train(args)
  File "/***/sd-scripts/train_network.py", line 1009, in train
    accelerator.backward(loss)
  File "/***/venv/lib/python3.10/site-packages/accelerate/accelerator.py", line 1905, in backward
    loss.backward(**kwargs)
  File "/***/venv/lib/python3.10/site-packages/torch/_tensor.py", line 521, in backward
    torch.autograd.backward(
  File "/***/venv/lib/python3.10/site-packages/torch/autograd/__init__.py", line 289, in backward
    _engine_run_backward(
  File "/***/venv/lib/python3.10/site-packages/torch/autograd/graph.py", line 769, in _engine_run_backward
    return Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
RuntimeError: Unknown error -2005270523

Then it failed.

I checked that there is already aten::native_group_norm implementation in DirectML:

202	aten::native_group_norm	[torch.float16, torch.float32]

But not backward