思维状态：面向思维树的结构化行动模板

STATe-of-Thoughts: Structured Action Templates for Tree-of-Thoughts

February 15, 2026

作者: Zachary Bamberger, Till R. Saenger, Gilad Morad, Ofra Amir, Brandon M. Stewart, Amir Feder

cs.AI

摘要

诸如最佳N采样（Best-of-N）和思维树（Tree-of-Thoughts）这类推理时计算（ITC）方法，旨在生成兼具高质量与多样性的输出候选，但其采用的高温采样策略往往难以实现有意义的输出多样性。此外，现有ITC方法对推理过程的控制能力有限，这反过来限制了其可解释性。我们提出STATe-of-Thoughts（STATe）——一种可解释的ITC方法，通过对高层推理模式进行搜索来实现优化。STATe用离散且可解释的文本干预替代随机采样：控制器选择编码高层推理决策的动作，生成器基于这些决策生成推理步骤，评估器则对候选结果评分以引导搜索。这种结构化方法具有三大优势：首先，动作引导的文本干预比基于温度的采样能产生更丰富的响应多样性；其次，在论证生成的案例研究中，STATe显式的动作序列能捕捉对输出质量具有高度预测性的可解释特征；最后，通过分析性能与动作选择的关联性，我们能识别动作空间中具有潜力但尚未探索的区域，并直接引导生成过程朝向这些区域。综合来看，这些成果确立了STATe作为生成高质量、多样化且可解释文本的实用框架。我们的框架已在https://github.com/zbambergerNLP/state-of-thoughts 开源。

English

Inference-Time-Compute (ITC) methods like Best-of-N and Tree-of-Thoughts are meant to produce output candidates that are both high-quality and diverse, but their use of high-temperature sampling often fails to achieve meaningful output diversity. Moreover, existing ITC methods offer limited control over how to perform reasoning, which in turn limits their explainability. We present STATe-of-Thoughts (STATe), an interpretable ITC method that searches over high-level reasoning patterns. STATe replaces stochastic sampling with discrete and interpretable textual interventions: a controller selects actions encoding high-level reasoning choices, a generator produces reasoning steps conditioned on those choices, and an evaluator scores candidates to guide search. This structured approach yields three main advantages. First, action-guided textual interventions produce greater response diversity than temperature-based sampling. Second, in a case study on argument generation, STATe's explicit action sequences capture interpretable features that are highly predictive of output quality. Third, estimating the association between performance and action choices allows us to identify promising yet unexplored regions of the action space and steer generation directly toward them. Together, these results establish STATe as a practical framework for generating high-quality, diverse, and interpretable text. Our framework is available at https://github.com/zbambergerNLP/state-of-thoughts.