arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.00306cs.LG
收藏

当链式思维帮助何时:马尔可夫视角

When does Chain-of-Thought Help: A Markovian Perspective

New York University(纽约大学)

Zihan Wang, Yijun Dong, Qi Lei

AI总结:

本文从马尔可夫视角分析了链式思维在不同任务中的有效性,揭示了转移对齐性是决定CoT效果的关键因素,并通过合成基准验证了理论预测。

摘要:

链式思维(CoT)提示是一种在推理过程中广泛使用的技术,用于提高推理能力,但其在不同任务中的收益不均。我们通过将逐步推理轨迹建模为马尔可夫链来分析何时和为什么CoT有效。每个中间步骤是一个状态,步骤之间的依赖性通过转移核来捕捉。我们的理论将转移对齐性,即实例是否共享共同的逐步转移核,作为CoT有效性的关键决定因素。当步骤间的转移相同,CoT减少推理时间的样本复杂度:较少的上下文样本轨迹足以恢复最终决策。相反,当步骤间的转移不同,这些收益可能消失。我们进一步量化了中间步骤中的噪声如何调节CoT的收益。除了理论之外,我们设计了合成基准,以隔离这些因素来补充对现实任务的先前结果,并通过实验证证我们的预测。

英文摘要:

Chain-of-Thought (CoT) prompting is a widely used inference-time technique for improving reasoning, yet its gains are uneven across tasks. We analyze when and why CoT helps by modeling the step-wise reasoning trajectory as a Markov chain. Each intermediate step is a state and the dependence between steps is captured by a transition kernel. Our theory identifies transition alignment, whether instances share a common step-wise transition kernel, as the key determinant of CoT's effectiveness. When transitions are identical across steps, CoT reduces inference-time sample complexity: fewer context sample trajectories suffice to recover the final decision. In contrast, when transitions differ across steps, these gains can vanish. We further quantify how noise in intermediate steps modulates CoT's benefit. Beyond theory, we design synthetic benchmarks that isolate these factors to complement prior results on real-world tasks and to empirically validate our predictions.

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。

↑