2 ms·Sentence-Level Reward Model Can Better Aligning LLM from Human Preference1 points by PaulHoule 2y ago