Aran Komatsuzaki   @arankomatsuzaki   6/8/2021       

Learning to Efficiently Sample from Diffusion Probabilistic Models Discovers inference time schedules requiring as few as 32 refinement steps, while sacrificing less than 0.1 bits per dimension compared to the default 4,000 steps used on ImageNet 64x64. https://t.co/cPjuCNKqh8

 Reply  0     Retweet   17      Like   95





Posted by Aran Komatsuzaki