Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution 文章

ArXiv CS.AI2026-06-09NEWSen作者: Xiaoou Liu, Tiejin Chen, Dengjia Zhang, Yaqing Wang, Lu Cheng, Hua Wei
查看原文 →
关系图谱

Diagnosing Multi-step Reasoning Failures in Black-box LLMs via Stepwise Confidence Attribution · 相关技术

相关技术

Large language modelInformation Bottleneck principle