arXiv cs.AI· Xiaoyu Liang, Virginia Ceccatelli, Aliai Eusebi, Tobias Braun, Nils Loose, Carlo Malagnino, Lorenzo Cavallaro·· 3 小时前AI 评分41
LLM 不确定性量化在哪里被观测？基于机制可解释性的四种观测视角解析
Where Is Uncertainty Observed? A Mechanistic Understanding of Uncertainty Quantification in LLMs
AI 导读
一项 arXiv 研究从机制可解释性角度梳理 LLM 不确定性量化（UQ），提出不确定性在模型计算中的观测位置问题。作者定义结构、表征、分布、语义四种机制观测视角，通过残差流将其按信息量排序，为针对不同目标选择 UQ 方法提供原则，并梳理现有方法与评测协议，指出连接不确定性估计与模型内部计算的开放问题。
正文 · 原文
Abstract:Knowing when a large language model's (LLM's) response can be trusted is critical for its safe use. Uncertainty quantification (UQ) can help LLMs recognize if their responses may be unreliable. Existing taxonomies organize UQ methods by uncertainty source or estimation mechanism. They leave unresolved the question of where uncertainty is observed in the model's computation. The answer matters because the observation point determines whether an estimator has access to enough information to recover a given target, and how well it can recover it. Drawing on mechanistic interpretability, we define four mechanistic observation lenses: the structural lens observes component updates, the representational lens observes hidden states, the distributional lens observes token probabilities, and the semantical lens observes distributions over meanings. We relate these lenses through the residual stream and order them by the information they carry, which gives a principle for selecting UQ methods for different targets. We survey existing methods through the four lenses. Furthermore, we distinguish evaluation targets, criteria, and metrics to clarify different UQ evaluation protocols. Finally, we identify open problems and research directions to connect uncertainty estimation to the internal computation and downstream use of LLMs.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2611.15735 [cs.CL] |
| (or arXiv:2611.15735v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.15735 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xiaoyu Liang [view email]
[v1]
Wed, 11 Nov 2026 17:14:57 UTC (1,863 KB)
来源：arXiv cs.AI · arxiv.org