search

wyf0912 / SinSR

[CVPR 2024] SinSR: Diffusion-Based Image Super-Resolution in a Single Step
Other
327 stars 19 forks source link

Question about reproduction. #9

Closed jiaxiangc closed 6 months ago

jiaxiangc commented 6 months ago

Does your inverse loss (in paper) decrease？ I try to reproduce it, but it always be 1.0.

wyf0912 commented 6 months ago

The inverse loss remains unchanged is abnormal.

Did you make any changes to the code and could you post the training log?

The inverse loss can decrease normally when I trained the models.

wyf0912 commented 6 months ago

Besides, you can check the inverse loss of the provided checkpoints. I think the inverse loss in checkpoints shall be much smaller than 1.