作者位次 #7/16ACL 2026Poster· Natural Language Processing
Which Reasoning Trajectories Teach Students to Reason Better? A Simple Metric of Informative Alignment
Yuming Yang · Mingyoung Lai · Wanxu Zhao · Xiaoran Fan · Zhiheng Xi · Mingqi Wu · Chiyue Huang · Jun Zhao · Haijun Lv · Jian Tong · Yunhua Zhou · Yicheng Zou · Qipeng Guo · Tao Gui · Qi Zhang · Xuanjing Huang
摘要
Long chain-of-thought (CoT) trajectories provide rich supervision signals for distilling reasoning from teacher to student LLMs. However, both prior work and our experiments show that trajectories from stronger teachers do not necessarily yield better students, highlighting the importance of data...