search

advimman / CIPS

Official repository for the paper "Image Generators with Conditionally-Independent Pixel Synthesis" (CVPR2021, Oral)
MIT License
208 stars 36 forks source link

Quick question about training time #1

Closed greeneggsandyaml closed 3 years ago

greeneggsandyaml commented 3 years ago

Hello authors,

Thank you for your paper -- it looks very interesting! I had a quick question about training time: how long (and with how many GPUs) does it take you to train the pretrained models provided in the repo? Do you find that the method is slow or fast relative to StyleGAN2?

I look forward to playing around with the repo over the next week!

Best, greeneggsandyaml

avecplezir commented 3 years ago

Hi!

Thanks for the question. With our implementation, CIPSis 4.3x times slower than StyleGANv2 and has 3.3x moreMACs (at 256x resolution). We trained our model on eight P40 GPUs and showed approximately 3.2 mln images to the discriminator during training. It took about two weeks to get these checkpoints.

Best wishes, Ivan

greeneggsandyaml commented 3 years ago

Thank you for the response! I really appreciate it, and I'm closing the issue.