arXiv 2509.107532025-09-12cs.LGcs.AI

HalluField：基于场论建模检测 LLM 幻觉

HalluField: Detecting LLM Hallucinations via Field-Theoretic Modeling

Center for Nonlinear Studies（非线性研究中心）；Los Alamos National Laboratory（洛斯阿拉莫斯国家实验室）；T-1 Los Alamos National Laboratory（洛斯阿拉莫斯国家实验室T-1部门）；Applied Mathematics University of Colorado Boulder（科罗拉多大学博尔德分校应用数学系）；T-4 Los Alamos National Laboratory（洛斯阿拉莫斯国家实验室T-4部门）

Minh Vu, Brian K. Tran, Syed A. Shah, Geigh Zollicoffer, Nhat Hoang-Xuan, Manish Bhattarai

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出 HalluField，一种基于参数化变分原理和热力学的场论方法，通过分析模型输出 logits 的能量与熵分布来量化语义稳定性，无需微调即可实现最先进的 LLM 幻觉检测。

摘要：

大型语言模型（LLMs）展现出令人印象深刻的推理与问答能力。然而，它们经常产生不准确或不可靠的内容，即所谓的幻觉。这种不可靠性严重限制了其在高风险应用中的部署。因此，对通用方法来检测 LLM 幻觉的需求日益增长。在本工作中，我们提出了 HalluField，一种基于参数化变分原理和热力学的新型场论幻觉检测方法。受热力学启发，HalluField 将 LLM 对给定查询和温度设置的响应建模为离散似然 token 路径的集合，每条路径具有相应的能量和熵。通过分析在不同温度和似然度变化下，能量和熵分布如何随 token 路径变化，HalluField 量化了响应的语义稳定性。随后，通过识别该能量景观中的不稳定或 erratic 行为来检测幻觉。HalluField 计算高效且极具实用性：它直接作用于模型的输出 logits，无需微调或辅助神经网络。值得注意的是，该方法建立在有原则的物理解释之上，类比于热力学第一定律。显著的是，通过这一物理透镜建模 LLM 行为，HalluField 在跨模型和数据集上实现了最先进的幻觉检测性能。

英文摘要：

Large Language Models (LLMs) exhibit impressive reasoning and question-answering capabilities. However, they often produce inaccurate or unreliable content known as hallucinations. This unreliability significantly limits their deployment in high-stakes applications. Thus, there is a growing need for a general-purpose method to detect hallucinations in LLMs. In this work, we introduce HalluField, a novel field-theoretic approach for hallucination detection based on a parametrized variational principle and thermodynamics. Inspired by thermodynamics, HalluField models an LLM's response to a given query and temperature setting as a collection of discrete likelihood token paths, each associated with a corresponding energy and entropy. By analyzing how energy and entropy distributions vary across token paths under changes in temperature and likelihood, HalluField quantifies the semantic stability of a response. Hallucinations are then detected by identifying unstable or erratic behavior in this energy landscape. HalluField is computationally efficient and highly practical: it operates directly on the model's output logits without requiring fine-tuning or auxiliary neural networks. Notably, the method is grounded in a principled physical interpretation, drawing analogies to the first law of thermodynamics. Remarkably, by modeling LLM behavior through this physical lens, HalluField achieves state-of-the-art hallucination detection performance across models and datasets.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。