JingyunLiang / SwinIR

SwinIR: Image Restoration Using Swin Transformer (official repository)

Home Page:https://arxiv.org/abs/2108.10257

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

about charbonnierloss

jiaaihhy opened this issue · comments

charbonnierloos have a extra parameter eps, in paper, eps is 1e-3, its true use is (1e-3) ^ 2, but in your code May be you dont take ^2 operations. I Dont know its inference is important?

This is a minor mistake. eps^2=1e-3 and eps^2=1e-9 lead to the same results in my experiments. If you have to choose, I suggest you use eps^2=1e-9 (a common practice).

I have fixed it in KAIR.

thankyou

Feel free to open it if you have more questions.

ezoic increase your site revenue