Warning: Grad strides do not match bucket view strides DDP. (Can impact performance)

Warning when attempting to start training:

...\lib\site-packages\torch\autograd\__init__.py:251: UserWarning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param's strides changed since DDP was constructed.  This is not an error, but may impair performance.
grad.sizes() = [64, 1, 4], strides() = [4, 1, 1]
bucket_view.sizes() = [64, 1, 4], strides() = [4, 4, 1] (Triggered internally at C:\actions-runner\_work\pytorch\pytorch\builder\windows\pytorch\torch\csrc\distributed\c10d\reducer.cpp:334.)
  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass

Maybe this can help https://github.com/pytorch/pytorch/issues/47163 & https://github.com/noahzn/Lite-Mono/issues/43

Start with the reported autograd warning in the training path and compare its grad and bucket_view strides with PyTorch issue 47163 and the linked Lite-Mono issue. Reproduce the warning during training, identify the affected parameter or operation, and confirm that the warning or its performance impact is resolved.