stable-fast
stable-fast copied to clipboard
componentwise config
enable vae.encode compile: A100 speedup
- original VAE: 1 step LCM sdv1.5 - 24ms speedup! (0.062s -> 0.038s) (I think it doesn't actually work though)
- taesd: (4step LCM sdv1.5) - 2ms (0.058s -> 0.056s)
@jon-chuang vae.encode now is compiled through https://github.com/chengzeyi/stable-fast/commit/c152145a4fe3061afc21cc58bf892eeae60c142f