lucidrains/PaLM-rlhf-pytorch

Implementation of RLHF (Reinforcement Learning with Human Feedback) on top of the PaLM architecture. Maybe I'll add retrieval functionality too, à la RETRO