作者位次 #2/4ICLR 2026Poster· Theory / Interpretability and Visualization
Characterizing and Mitigating Reasoning Drift in Large Language Models
Yufeng Zhang · Xuepeng Wang · Lingxiang Wu · Jinqiao Wang
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。
Yufeng Zhang · Xuepeng Wang · Lingxiang Wu · Jinqiao Wang
Yufeng Zhang · Xuepeng Wang · Lingxiang Wu · Jinqiao Wang
Large language models (LLMs) are prone to distraction by contextual information during reasoning. Previous work primarily focuses on improving the generation of the next token while overlooking the potential bias introduced by existing premises. We propose a novel decoding method to mitigate such...