Gaurav Parmar

This should not be the case because generated images are often quantized to uint8 before computing the FID score. (because real images are not real values). The different in FID...

Hi, Thank you for pointing this out! I have addressed in the new release of the library. Regards, Gaurav

We have not implemented resizing which can propagate gradient in this repository. There are some related libraries that might be of interest [ResizeRight](https://github.com/assafshocher/ResizeRight), [kornia](https://kornia.readthedocs.io/en/latest/index.html). Although we have not fully evaluated...

Hi, Thank you for pointing this out. However, the difference in runtime should not be this large. Could you provide the code snippet you are using to compute the FID?...

Hi, Could you share the script you use for computing the Clean-FID score for StyleGANv2? It is difficult to tell how the processing of images may be different that might...

Hi, Thank you for taking a look at the issue and the corresponding pull request. I have merged the PR and the new clean-fid release should have this issue addressed....

Hi @Yves907, Thank you for pointing out the issue! Can you show the contents of this file? > home/sll/miniconda3/envs/yanjie_torch/lib/python3.6/site-packages/cleanfid/features.py This will better help me debug why this runtime error is...

It looks like the Inception model was not downloaded correctly. Can you check if the file `/tmp/inception-2015-12-05.pt` exists?

Thank you for pointing these details out! I have noticed that sometime the download of the inception file gets corrupted if the internet connection is interrupted midway. I will add...

Hi, Thank you for pointing it out. This is because when the FID stats are loaded, the library checks for the lowercase version of the string. I will address this...