STATe-of-Thoughts: Structured Action Templates for Tree-of-Thoughts
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866914432281477120 |
|---|---|
| author | Bamberger, Zachary Saenger, Till R. Morad, Gilad Amir, Ofra Stewart, Brandon M. Feder, Amir |
| author_facet | Bamberger, Zachary Saenger, Till R. Morad, Gilad Amir, Ofra Stewart, Brandon M. Feder, Amir |
| contents | 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 interpretability. 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 reliably influence LLM generations and 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 toward them. Together, these results establish STATe as both a practical framework for diverse and controllable text generation, and as a tool for understanding the reasoning patterns that drive performance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_14265 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | STATe-of-Thoughts: Structured Action Templates for Tree-of-Thoughts Bamberger, Zachary Saenger, Till R. Morad, Gilad Amir, Ofra Stewart, Brandon M. Feder, Amir Computation and Language Machine Learning I.2.7; I.2.11; H.1.2 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 interpretability. 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 reliably influence LLM generations and 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 toward them. Together, these results establish STATe as both a practical framework for diverse and controllable text generation, and as a tool for understanding the reasoning patterns that drive performance. |
| title | STATe-of-Thoughts: Structured Action Templates for Tree-of-Thoughts |
| topic | Computation and Language Machine Learning I.2.7; I.2.11; H.1.2 |
| url | https://arxiv.org/abs/2602.14265 |