Zhang Yuefeng

It's the problem related to the setting of hyperparameters. To stay tune with the pretrained model, we should set `options/config_hifacegan.py`: ``` netG='lipspade' #ngf = 64 # set to 48 for...

> @ZhangYuef thanks for raising this issue! Could you provide a reference for MS-SSIM? Hi @jni , thanks for reply! MS-SSIM is also an image quality assessment method which was...

> We don't currently have this, but from the reference it looks like this would not be too difficult to implement with some refactoring of the existing single scale function....

> In my case, the values of RGB channels in result picture come so close (e.g. 20/23/32) which contributes to its gray outlook. Any advice to avoid this problem?

> > > In my case, the values of RGB channels in result picture come so close (e.g. 20/23/32) which contributes to its gray outlook. > > > > >...

Same problem, did you solve it? Refer #467

> Have you successfully built DCNv2？I think the error is caused by "You're using a third-party extension module written in C, and that extension module has crashed." You can try...

> Hey, > there seems to be an inversion in the expected result. > The films with the highest number of reviews on the original basis (popularity =% of users)...

> Hi! > I run this: https://github.com/NicolasHug/Surprise/blob/master/examples/top_n_recommendations.py > which gives me inverted results Hi @fegarat , would you mind to offer the code you use to calculate `popularity`? I am...

Hi @fegarat , This `popularity` metric is unusual here. Since in the example [link](https://github.com/NicolasHug/Surprise/blob/master/examples/top_n_recommendations.py) we use `SVD` and aim to predict each user's rating of the movie, we may not...