Nathan Lambert 评 score-centering RL 论文的重要性采样表
Was reading the score-centering RL paper to add some notes on it the my book, and they have a great table in the appendix on the different importance sampling b
AI 导读
Nathan Lambert 表示，他为写书笔记读了 score-centering RL 论文，认为其附录中关于 RL 中基于重要性采样的训练-推理不一致方法的表格很出色。他还认为，自 GRPO 兴起以来，这类方法一直是 RL 算法研究的最大焦点。
正文
Was reading the score-centering RL paper to add some notes on it the my book, and they have a great table in the appendix on the different importance sampling based train-inference mismatch methods in RL. These have been the biggest focus of RL algorithm research since the GRPO takeoff, imo.