alias-free-gan-pytorch icon indicating copy to clipboard operation
alias-free-gan-pytorch copied to clipboard

Published 20 hours ago • verified publisher icon rosinality

Training Speed

Open nicolas-dufour opened this issue 3 years ago • 1 comments

Hi! The authors from the paper said that they needed to use a custom conda kernel to get acceptable performance (They say it's 20x faster than with pytorch primitives. I noticed u didn't use a custom cuda object. How is your performance compared to StyleGAN2?

Thanks!

nicolas-dufour avatar Jul 27 '21 10:07 nicolas-dufour

Config-T (3x3 conv) is quite similar to stylegan2, and Config-R (1x1 conv) is about 1.5x slower.

rosinality avatar Jul 27 '21 13:07 rosinality