arXiv:2605.28566v1 Announce Type: cross 
Abstract: Large Language Models (LLMs) have demonstrated remarkable reasoning capabilities, yet their standard generation process -- auto-regressive token prediction -- is inherently myopic and prone to cascading errors. To address this, the Tree-of-Thoughts (ToT) framework creates a search space over intermediate reasoning steps, allowing search models to explore, look ahead, and backtrack. However, current ToT research remains fragmented across Natural Language Processing and Automated Planning communities, often using inconsistent terminology and ad-hoc implementations. Consequently, we synthesize the ToT landscape through a unified taxonomy based on classical heuristic search terminology. We map LLM-based reasoning to classical search components: state representation (granularity of thoughts), successor generation (prompting operators), and heuristic evaluation (self-assessment of progress). We analyze existing work within the context of our taxonomy and identify emerging design patterns: systematic search (Best-First Search) for shallow, deterministic tasks and lookahead-heavy strategies (DFS, MCTS) for deep multi-step reasoning. We conclude by identifying open algorithmic challenges at the intersection of heuristic search and LLM reasoning, and call on the heuristic search community to engage with this emerging domain.

تم تقديم إطار Tree-of-Thoughts (ToT) كحل للقيود المفروضة على نماذج اللغة الكبيرة (LLMs)، والتي غالبًا ما تظهر تفكيرًا قصير النظر وأخطاء متتالية أثناء التنبؤ بالتوكنات بشكل تلقائي. يتيح هذا الإطار مساحة بحث منظمة حول خطوات التفكير الوسيطة، مما يمكّن النماذج من الاستكشاف والتطلع إلى الأمام والعودة بشكل فعال.

El marco Tree-of-Thoughts (ToT) se ha introducido como una solución a las limitaciones de los Modelos de Lenguaje Grande (LLMs), que a menudo exhiben un razonamiento miope y errores en cascada durante la predicción de tokens autorregresiva. Este marco permite un espacio de búsqueda estructurado sobre los pasos intermedios de razonamiento, lo que permite a los modelos explorar, anticipar y retroceder de manera efectiva.

Le cadre Tree-of-Thoughts (ToT) a été introduit comme une solution aux limitations des grands modèles de langage (LLM), qui présentent souvent un raisonnement myope et des erreurs en cascade lors de la prédiction de tokens auto-régressive. Ce cadre permet un espace de recherche structuré sur les étapes intermédiaires de raisonnement, permettant aux modèles d'explorer, de prévoir et de revenir en arrière efficacement.

The Tree-of-Thoughts (ToT) framework has been introduced as a solution to the limitations of Large Language Models (LLMs), which often exhibit myopic reasoning and cascading errors during auto-regressive token prediction. This framework allows for a structured search space over intermediate reasoning steps, enabling models to explore, look ahead, and backtrack effectively.