作者位次 #4/5NeurIPS 2026Poster
Cornerstones or Stumbling Blocks? Deciphering the Rock Tokens in On-Policy Distillation
Yuxuan Jiang · Runchao Li · Shubhashis Roy Dipta · Dawei Li · Zhao Yang
强化学习与后训练OpenReview
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Yuxuan Jiang · Runchao Li · Shubhashis Roy Dipta · Dawei Li · Zhao Yang
Yuxuan Jiang · Dawei Li · Francis Ferraro
While Large Reasoning Models (LRMs) excel at complex tasks via long Chain-of-Thought (CoT) reasoning, their outputs are often excessively verbose, leading to inefficiency. This problem is amplified when the student’s long-form reasoning mismatches the concise outputs of smaller teacher models—com...