zimenglan

hi @PkuRainBow, why try to change the hard_ratio?

@PkuRainBow i found that changing the `batchsize` to `50` the loss won't decrease on `CUB200_2011` dataset

hi @KimSoybean do you still have the plan to release the code and model? thanks.

+1

hi @allenu2 @zxr8192 have u downloaded the datasets?

any progress updates?

wait for the train code.

hi @MasterBin-IIAU any update?

> This should be work, I covert operations on weight to input(x), then GFPGAN model will be constant. 这种方法实测有效，我将对weight的操作等效修改为对x的操作，这样卷积的权重就是固定参数，模型就会从动态转换为静态。 > > ``` > class ModulatedConv2d(nn.Module): > def __init__(self, > in_channels,...

+1