IFRNet icon indicating copy to clipboard operation
IFRNet copied to clipboard

Published 20 hours ago • verified publisher icon ltkong218

Error running train_vimeo90k.py

Open Liming-belief opened this issue 2 years ago • 5 comments

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

Liming-belief avatar May 08 '23 05:05 Liming-belief

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

ltkong218 avatar Jun 12 '23 01:06 ltkong218

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

jackyin68 avatar Dec 21 '23 09:12 jackyin68

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

Siziff avatar Feb 12 '24 12:02 Siziff

This problem has been solved as https://github.com/ltkong218/IFRNet/pull/31.

ltkong218 avatar Feb 19 '24 12:02 ltkong218

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

Siziff avatar Feb 19 '24 12:02 Siziff