A arxiv.org Training Language Models to Self-Correct via Reinforcement Learning 230 points weirdcat 2 years ago 92 comments
