search

Zj-BinXia / ENLCA

This project is official implementation of 'Efficient Non-Local Contrastive Attention for Image Super-Resolution', AAAI2022
86 stars 7 forks source link

想请教一下在训练enlcn模型时，psnr值为nan #15

Open Zc229 opened 1 year ago

Zc229 commented 1 year ago

Preparing loss function: 1.000 * L1 [Epoch 1] Learning rate: 1.00e-4

Evaluation: 0it [00:01, ?it/s] [Set14 x4] PSNR: nan (Best: nan @epoch 1) Forward: 2.35s

jiaweimmiao commented 9 months ago

@Zc229 我也遇到了相同的问题，请问您的问题已经解决了吗？可以分享一下解决方案么？