外部观察者可能看得更清楚：通过隐藏状态探测实现大型语言模型的跨模型跨度级幻觉检测

中文摘要 AI 辅助

随着大型语言模型（LLMs）日益成为基础推理引擎，其产生幻觉的倾向仍然是一个关键漏洞。虽然最近的内部状态探测为缓慢的外部检索系统提供了一种有前景的替代方案，但它们大多将幻觉检测简化为逐词二元分类任务，未能捕捉语义漂移的结构化、序列化边界。在此，我们提出一种内部隐藏状态框架，用于细粒度的跨度级幻觉检测。通过检查逐层激活模式，我们试图检测LLM生成中幻觉的确切起始和延续标记。我们的实验表明，该方法成功隔离了幻觉起始，尽管存在极端类别不平衡，但在精确率-召回率AUC上相比随机基线取得了显著改进。最终，我们提出一种新颖的跨模型检测框架，其中一个模型观察由另一个模型生成所引发的内部表示。我们发现，外部观察者可以匹配或超过生成者对其自身幻觉起始的自我检测，即使观察者是较小的模型，这表明自我检测并非起始定位的上限。

英文摘要

As Large Language Models (LLMs) increasingly serve as foundational reasoning engines, their tendency to hallucinate remains a critical vulnerability. While recent internal state probes offer a promising alternative to slow external retrieval systems, they largely reduce hallucination detection to a token-wise binary classification task, failing to capture the structured, sequential boundaries of semantic drift. Here, we introduce an internal hidden state framework for fine-grained, span-level hallucination detection. By inspecting layer-wise activation patterns, we attempt to detect the exact hallucination onset and continuation tokens in an LLM generation. Our experiments show that this approach successfully isolates hallucination onsets, achieving substantial improvements in Precision-Recall AUC over random baselines despite extreme class imbalance. Ultimately, we propose a novel cross-model detection framework in which one model observes the internal representations elicited by another model's generation. We find that an external observer can match or exceed a generator's self-detection of its own hallucination onsets, including when the observer is the smaller model, suggesting that self-detection is not the ceiling for onset localisation.

发表机构

• LIX, École Polytechnique（巴黎综合理工学院 LIX）
• School of Electrical and Electronic Engineering, Nanyang Technological University（南洋理工大学电气与电子工程学院）
• LIPN, Sorbonne Paris Nord（巴黎北索邦大学 LIPN）

机构由 AI 辅助整理，请以论文原文为准。

补充信息

备注

12 pages, 2 figures, 9 tables