geekac

@osho-agyeya I guess you run the code under Windows platform, if so, set the parameter [num_workers=0] when loading data by torch.utils.data.DataLoader()... It's multiprocess problem, if you use Ubuntu, it disappears.

I solved this problem by downgrade package timm version from 0.9.16 to 0.9.2 And, I have tested that the last version is 0.9.10 which do not print this warning.

Thank you, Dmitrii. Base on your guidance, I wrote the right code for the same size of data. But, to learn more, does that means for diffierent size of inputs,...