arXiv Artificial Intelligence· Zirui Ding (Institute of Computing Technology, Chinese Academy of Sciences, University of Chinese Academy of Sciences), Xinxin Fan (University of Chinese Academy of Sciences, State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences), Hao Yuan (Institute of Computing Technology, Chinese Academy of Sciences, ShanghaiTech University), Shaoye Luo (University of Chinese Academy of Sciences, State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences), Men Niu (Institute of Computing Technology, Chinese Academy of Sciences, University of Chinese Academy of Sciences), Yunfeng Lu (Beihang University)·· 19 小时前AI 评分31
CauTrace：揭示 LLM 推理失败根源的白盒诊断框架
Uncovering Why LLMs Fail: From Internal Mechanisms to Adversarial Reasoning Trajectories
AI 导读
中科院计算所等团队提出白盒诊断框架 CauTrace，从功能状态、表征传播、时序状态协调与推理轨迹三个层面剖析 LLM 失败根源。在 Qwen 与 LLaMA 系列模型上，失败轨迹表现出支持状态更新时 Attention-MLP 一致性更弱、验证活动相对承诺更少，具体表征特征与时序偏移因架构而异。留出集修复与引导测试在部分模型和特定设置下产生与主张一致的操作信号。
来源：arXiv Artificial Intelligence · arxiv.org