pymlg
pymlg copied to clipboard
added cuda support for all pytorch mlg operations
tests passing on a GeForce GTX 1650 Mobile / Max-Q w/ CUDA 12.8
- [ ] Still need to add complex tensor/ndarray support for complex step jacobian. not currently sure how to test that?
@angadbajwa thanks for doing this man! the automated testing environment doesn't seem to have a GPU, which will cause the unit tests to fail. I suppose you could just use torch.cuda.is_available in the unit tests themselves for now, so that it doesn't run in the github environment. not ideal but good enough