Running regular PyTorch with PT-XLA CUDA Container

[codeislife99]

With the PT-XLA containers for CUDA, it is only possible to run PT-XLA. If one tries to run regular PyTorch, then we get
Torch not compiled with CUDA enabled. Is there any change that can be made to the Dockerfile so that both regular CUDA PyTorch and PT-XLA can be run on the same container ?

[jysohn23]

I don't think we've tested it but you could try enabling this for https://github.com/pytorch/pytorch/blob/3d6aa3a2f6e67140a21ca1e33fab089a57600fe7/setup.py#L14:

xla/docker/Dockerfile

Line 20 in c297289

ENV USE_CUDA "0"

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.