让大模型自己解释推理过程，更准地判断回答可信度
CoT-UQ: Improving Response-wise Uncertainty Quantification in LLMs with Chain-of-Thought
- 用思维链提取每步推理关键词，评估对答案重要性
- 在逻辑数学任务上比现有方法平均提升5.9%准确率
- 适合需要可靠判断的场景，如医疗、金融决策
大型语言模型（LLMs）在多项任务中表现优异，但在生成回答时难以准确量化不确定性，导致难以识别错误信息并保障决策可靠性。现有不确定性量化（UQ）方法多为提示级而非响应级，通常需多次采样生成回复，计算成本高。此外，大模型在使用推理步骤时往往过于自信。本文提出CoT-UQ，一种基于思维链（Chain-of-Thought, CoT）的响应级不确定性量化框架，将模型内在推理能力融入不确定性评估过程。该方法在推理过程中提取每一步的关键词，并评估其对最终答案的重要性，再聚合这些信息生成最终不确定性估计。我们在Llama系列模型（8B至13B）上开展广泛实验，涵盖逻辑与数学推理任务。结果表明，CoT-UQ显著优于现有方法，在平均AUROC上提升5.9%。代码已公开于：https://github.com/ZBox1005/CoT-UQ。
原文摘要 · Abstract (English)
Large language models (LLMs) excel in many tasks but struggle to accurately quantify uncertainty in their generated responses. This limitation makes it challenging to detect misinformation and ensure reliable decision-making. Existing uncertainty quantification (UQ) methods for LLMs are primarily prompt-wise rather than response-wise, often requiring multiple response samples, which incurs high computational costs. Moreover, LLMs have been shown to be overconfident, particularly when using reasoning steps to derive their answers. In this work, we propose CoT-UQ, a response-wise UQ framework that integrates LLMs' inherent reasoning capabilities through Chain-of-Thought (CoT) into the UQ process. CoT-UQ captures critical information during inference by extracting keywords from each reasoning step and assessing their importance to the final answer. This key reasoning information is then aggregated to produce a final uncertainty estimate. We conduct extensive experiments based on Llama Family with model sizes varying from 8B to 13B across logical and mathematical reasoning tasks. Experimental results demonstrate that CoT-UQ significantly outperforms existing UQ methods, achieving an average improvement of 5.9% AUROC compared to current UQ methods. The code is available at: https://github.com/ZBox1005/CoT-UQ.
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。