推理模型能更准确地表达其置信度

摘要

尽管大型语言模型（LLMs）具备强大能力，它们往往难以准确传达其置信度，这使得评估其可能出错的情况变得困难，从而限制了其可靠性。在本研究中，我们证明了推理模型——即那些进行扩展链式思维（CoT）推理的LLMs——不仅在问题解决上表现更优，还能更精确地表达其置信度。具体而言，我们在六个数据集上对六种推理模型进行了基准测试，发现它们在36种设定中有33种情况下，其置信度校准严格优于非推理模型。深入分析表明，这些校准上的提升源于推理模型的慢思考行为，如探索替代方法和回溯，这些行为使它们能够在CoT过程中动态调整置信度，使其逐步变得更加准确。特别地，我们发现随着CoT的展开，推理模型的校准度持续提高，这一趋势在非推理模型中并未观察到。此外，若从CoT中移除慢思考行为，校准度会显著下降。最后，我们指出这些优势并非推理模型独有——通过上下文学习引导非推理模型进行慢思考，它们同样能从中获益。

English

Despite their strengths, large language models (LLMs) often fail to communicate their confidence accurately, making it difficult to assess when they might be wrong and limiting their reliability. In this work, we demonstrate that reasoning models-LLMs that engage in extended chain-of-thought (CoT) reasoning-exhibit superior performance not only in problem-solving but also in accurately expressing their confidence. Specifically, we benchmark six reasoning models across six datasets and find that they achieve strictly better confidence calibration than their non-reasoning counterparts in 33 out of the 36 settings. Our detailed analysis reveals that these gains in calibration stem from the slow thinking behaviors of reasoning models-such as exploring alternative approaches and backtracking-which enable them to adjust their confidence dynamically throughout their CoT, making it progressively more accurate. In particular, we find that reasoning models become increasingly better calibrated as their CoT unfolds, a trend not observed in non-reasoning models. Moreover, removing slow thinking behaviors from the CoT leads to a significant drop in calibration. Lastly, we show that these gains are not exclusive to reasoning models-non-reasoning models also benefit when guided to perform slow thinking via in-context learning.

推理模型能更准确地表达其置信度

Reasoning Models Better Express Their Confidence

摘要

Summary

Support

Support