一作ACL 2026Poster· Natural Language Processing
Harnessing Negative Signals: Reinforcement Distillation from Teacher Data for LLM Reasoning
Shuyao Xu · Cheng Peng · Jiangxuan Long · Weidi Xu · Wei Chu · Yuan Qi
摘要
Recent advances in model distillation show that data from advanced reasoning models can effectively train smaller student models. However, standard practices discard incorrect reasoning traces—valuable, yet underutilized data. This paper addresses the critical question: How can both positive and...