evelynmitchell / alignprop Goto Github PK
View Code? Open in Web Editor NEWThis project forked from mihirp1998/alignprop
AlignProp uses direct reward backpropogation for the alignment of large-scale text-to-image diffusion models. Our method is 25x more sample and compute efficient than reinforcement learning methods (PPO) for finetuning Stable Diffusion
Home Page: https://align-prop.github.io/
License: MIT License