train error

[pheotry]

How to deal with this problem.

File "d:\code\tao\torchrs\src\train.py", line 305, in
main(config)
File "d:\code\tao\torchrs\src\train.py", line 291, in main
train_and_get_best_model(misr_model, regis_model, optimizer, dataloaders, baseline_cpsnrs, config)
File "d:\code\tao\torchrs\src\train.py", line 175, in train_and_get_best_model
srs_shifted = apply_shifts(regis_model, srs, shifts, device)[:, 0]
File "d:\code\tao\torchrs\src\train.py", line 57, in apply_shifts
new_images = shiftNet.transform(thetas, images, device=device)
File "d:\code\tao\torchrs\src\DeepNetworks\ShiftNet.py", line 96, in transform
new_I = lanczos_shift(img=I.transpose(0, 1),
File "d:\code\tao\torchrs\src\DeepNetworks\lanczos.py", line 97, in lanczos_shift
I_s = torch.conv1d(I_padded,
RuntimeError: Expected 2D (unbatched) or 3D (batched) input to conv1d, but got input of size: [1, 1, 202, 202]

[mafizurrahman]

Hi, did you solve the problem?

[qqhe-frank]

I had the same problem, but the issue was only with the version of pytorch, I was previously using pytorch version 1.13. And using pytorch 1.10,python3.6 successfully eliminated the problem.