4 ms·Rethinking the Role of PPO in RLHF2 points by skilled 3y agoturtleyacht 3y agoReinforcement Learning with Human Feedback (RHLF) Proximal Policy Optimization (PPO)