arXiv 2610.08077 · cs.AI · 2026-10-06
Self-Retrospection Distillation: Turning Post-hoc Experiences into Prior Foresight
Haoxiang Zhang, Qinglin Chen, Hiroaki Hayashi, Zhuofeng Li, Siming Zhang, Jiaxin Zhang, Jixuan Chen, Fang Wu 等 12 人
Salesforce AI Research
在最新榜 #68 2026-W41 第 4 名 本站收录于 2026-10-21
一句话要点
自我回顾蒸馏把事后经验转成事前预判，弥补 RLVR 在奖励全相同时信号消失的问题。
Self-Retrospection Distillation turns hindsight into foresight supervision, filling the gap when RLVR rewards are all identical.
阅读 arXiv 原文 PDF Hugging Face 讨论 代码仓库
主题与学科
同主题：强化学习
- Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy Distillation ▲218
- LoGRA: Scaling LLM Reinforcement Learning with Low-Rank Gradient Sketches ▲152
- Sharpening Tax in Post-Training ▲111
- RULER: Instance-aware Rubric Rewards for SVG Generation ▲104
- TPPO: Tracking-Value PPO For Language Model Reasoning ▲96
数据说明
元数据来自 arXiv API（已核实）；点赞来自 Hugging Face 每日论文；引用来自 Semantic Scholar。最近刷新 2026-10-22。
要点由本站根据摘要自写，不转载摘要原文，请阅读原文。