HF Daily Papers·2026-10-02 04:00· 2026-10-02AI 评分28

MetaRubric：面向评分标准强化学习的可学习奖励方法

MetaRubric: Learning to Reward for Rubric-Based Reinforcement Learning

AI 导读

MetaRubric 提出

来源：HF Daily Papers · huggingface.co

数据与训练论文研究 #论文/研究 #数据/训练 #arXiv #Hugging Face