annotated_deep_learning_paper_implementations icon indicating copy to clipboard operation
annotated_deep_learning_paper_implementations copied to clipboard
verified publisher icon labmlai

Unet error in DDPM

Open TwinkleStarst opened this issue 1 year ago • 4 comments

When training with the celeba dataset, the program encountered an error. Snipaste_2024-08-14_15-17-03

TwinkleStarst avatar Aug 14 '24 07:08 TwinkleStarst

Yes, I am getting the same one. Did you find any solution? I'll let you know if I do.

Paschalis-D avatar Aug 18 '24 16:08 Paschalis-D

same issue.Have you solved the problem? When I use dataset MNIST,it works well.

baizhu0414 avatar Oct 26 '24 14:10 baizhu0414

I'd like to work on this issue. Can I take it?

NisargUpadhyayIITJ avatar Mar 16 '25 05:03 NisargUpadhyayIITJ

DownSample use Conv2d with kernel_size (3, 3), but UpSample use ConvTranspose2d with kernel_size (4, 4) which results in the (H, W) of x larger than s. I add a F.pad operation before torch.cat

G23yl avatar Jul 30 '25 04:07 G23yl