通过不确定性分析，提前识别大模型推理中的错误路径。
Tracing Uncertainty in Language Model "Reasoning"
- 将推理过程视为动态状态，用不确定性特征描述其演变轨迹。
- 仅用前几百个词元就能预测答案正确性，准确率最高达0.807 AUROC。
- 正确推理路径的不确定性下降更陡且非线性，适合用于错误预警。
大模型推理（如思维链）虽能提升基准性能，但其内在机制仍不清晰。本文从不确定性量化视角出发，将模型生成的中间文本序列视为演化中的模型状态，提出用一组小特征（如不确定性信号的斜率、线性度）构成不确定性轨迹轮廓来总结每条推理路径。在GSM8K和ProntoQA数据集上，五种大模型的实验表明，该轮廓可预测最终答案正确性，最大AUROC达0.807，显著优于现有方法。仅使用完整轨迹前几百个词元即达到0.801的AUROC，表明错误可在早期被检测。对正确与错误路径的对比显示，正确路径的不确定性下降更陡峭且非线性。结果表明，基于不确定性决策的分析框架为理解大模型推理生成过程提供了原理性视角。
原文摘要 · Abstract (English)
Language model (LM) "reasoning", commonly described as Chain-of-Thought or test-time scaling, often improves benchmark performance, but the dynamics underlying this process remain poorly understood. We study these dynamics through the lens of uncertainty quantification by treating the "reasoning" traces, the intermediate token sequences generated by LMs, as evolving model states. We summarize each trace by an uncertainty trace profile: a small set of features describing the shape of the uncertainty signal over its trace, such as its slope and linearity. We find that across five LMs evaluated on GSM8K and ProntoQA, these profiles predict whether a trace yields a correct final answer with AUROC up to 0.807, improving markedly on recent related work. We reach AUROC 0.801 using only the first few hundred tokens of full traces, suggesting that errors can be detected early in the generation. A detailed comparison of correct and incorrect traces further reveals qualitatively distinct uncertainty profiles, with correct traces showing a steeper and less linear decline in uncertainty. Together, the results suggest that our method, grounded in decision-making under uncertainty, provides a principled lens for studying the generative process underlying LM "reasoning".
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。