arXiv cs.AI· Yedidia Agnimo, Anna Korba, Annabelle Blangero, Nicolas Chesneau, Karteek Alahari·· 3 天前AI 评分41
LLM 不确定性估计器与模型幻觉的对齐关系研究
How Do Uncertainty Estimators Align with Hallucination Occurrence in LLMs?
AI 导读
研究系统评估了 46 种不确定性估计器在内在幻觉与外在幻觉场景下的表现，覆盖四个 benchmark 和多个 LLM。结果显示不确定性与幻觉之间的关联差异很大且往往较弱，取决于幻觉场景和所评估的模型。这一结果对将不确定性直接作为幻觉信号的做法提出了质疑。
正文 · 原文
Abstract:Large language models (LLMs) can hallucinate, but do model uncertainty estimators reliably reveal when they do? Although uncertainty estimators are increasingly used to identify unreliable generations, their relationship with hallucination remains poorly understood and is often taken for granted. We systematically evaluate 46 uncertainty estimators across complementary hallucination settings, covering both intrinsic hallucinations, where outputs are unsupported by the provided context, and extrinsic hallucinations, where claims are unsupported by expected model knowledge. Across four benchmarks and multiple LLMs, we find that the association between uncertainty and hallucination varies widely and is often weak, depending on the hallucination setting and the model under evaluation. These results challenge the use of uncertainty as a direct signal of hallucination and clarify when it provides useful information.
| Comments: | 50 pages, 11 figures, 18 tables |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Machine Learning (stat.ML) |
| Cite as: | arXiv:2605.27016 [cs.CL] |
| (or arXiv:2605.27016v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2605.27016 arXiv-issued DOI via DataCite |
Submission history
From: Yedidia Agnimo [view email]
[v1]
Tue, 26 May 2026 13:34:54 UTC (1,278 KB)
[v2]
Mon, 12 Oct 2026 08:14:23 UTC (1,124 KB)
来源：arXiv cs.AI · arxiv.org