Error running train_vimeo90k.py

[Liming-belief]

RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [6, 32, 112, 112]], which is output 0 of AsStridedBackward0, is at version 1; expected version 0 instead.

May I ask what is causing this and how can I make the necessary modifications

[ltkong218]

What experiment environment are you using? (PyTorch, CUDA version and GPU card)

[jackyin68]

Have the same question, video size should be fixed? and what?

[Siziff]

SOOOO? Has this problem been solved? (pytorch == 2.1.2+cu121, CUDA == 12.2 , GPU only one v100 16 GB )

[ltkong218]

This problem has been solved as #31.

[Siziff]

@ltkong218 I installed the same versions as yours and it worked. Thanks :)