作者位次 #4/9ACL 2026Poster· Natural Language Processing
Learning from Mistakes: Negative Reasoning Samples Enhance Out-of-Domain Generalization
Tian Xueyun · MingHua Ma · Bingbing Xu · Nuoyan Lyu · Wei Li · Heng Dong · Zheng Chu · Yuanzhuo Wang · Huawei Shen
摘要
Supervised fine-tuning (SFT) on chain-of-thought (CoT) trajectories demonstrations is a common approach for enabling reasoning in large language models. Standard practices typically only retain trajectories with correct final answers ( positives ) while ignoring the rest ( negatives ). We argue t...