揭示大模型幻觉成因，通过子序列关联追踪错误源头
Why and How LLMs Hallucinate: Connecting the Dots with Subsequence Associations
- 将大模型视为子序列嵌入器，用线性层编码输入输出关联
- 提出追踪算法，在随机输入中识别导致幻觉的关键子序列
- 方法比传统归因技术更准，且与训练数据证据一致
大语言模型常产生偏离事实或上下文的幻觉内容，其成因复杂难以诊断。本文提出一种子序列关联框架，系统追踪幻觉根源。核心观点是：当主导性幻觉关联强于真实关联时，幻觉即发生。理论与实证分析表明，解码器仅的Transformer实质为子序列嵌入模型，线性层编码输入-输出关联。我们设计追踪算法，通过分析随机输入上下文中的幻觉概率，识别因果子序列。实验显示，该方法在定位幻觉原因上优于标准归因技术，并与模型训练语料证据相符。本工作提供幻觉统一视角与可靠追踪分析框架。
原文摘要 · Abstract (English)
Large language models (LLMs) frequently generate hallucinations-content that deviates from factual accuracy or provided context-posing challenges for diagnosis due to the complex interplay of underlying causes. This paper introduces a subsequence association framework to systematically trace and understand hallucinations. Our key insight is that hallucinations arise when dominant hallucinatory associations outweigh faithful ones. Through theoretical and empirical analyses, we demonstrate that decoder-only transformers effectively function as subsequence embedding models, with linear layers encoding input-output associations. We propose a tracing algorithm that identifies causal subsequences by analyzing hallucination probabilities across randomized input contexts. Experiments show our method outperforms standard attribution techniques in identifying hallucination causes and aligns with evidence from the model's training corpus. This work provides a unified perspective on hallucinations and a robust framework for their tracing and analysis.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。