The CoT Encyclopedia: Analyzing, Predicting, and Controlling how a Reasoning Model will Think

TL;DR

The paper introduces the CoT Encyclopedia, a framework for analyzing, predicting, and guiding reasoning strategies in large language models, leading to improved interpretability and performance in chain-of-thought reasoning.

Contribution

It presents a novel bottom-up approach to automatically categorize and interpret diverse reasoning behaviors in models, surpassing prior human-intuition-based methods.

Findings

Model reasoning strategies can be effectively categorized and predicted.

Training data format significantly influences reasoning behavior.

Understanding strategies enables performance improvements.

Abstract

Long chain-of-thought (CoT) is an essential ingredient in effective usage of modern large language models, but our understanding of the reasoning strategies underlying these capabilities remains limited. While some prior works have attempted to categorize CoTs using predefined strategy types, such approaches are constrained by human intuition and fail to capture the full diversity of model behaviors. In this work, we introduce the CoT Encyclopedia, a bottom-up framework for analyzing and steering model reasoning. Our method automatically extracts diverse reasoning criteria from model-generated CoTs, embeds them into a semantic space, clusters them into representative categories, and derives contrastive rubrics to interpret reasoning behavior. Human evaluations show that this framework produces more interpretable and comprehensive analyses than existing methods. Moreover, we demonstrate that this understanding enables performance gains: we can predict which strategy a model is likely to use and guide it toward more effective alternatives. Finally, we provide practical insights, such as that training data format (e.g., free-form vs. multiple-choice) has a far greater impact on reasoning behavior than data domain, underscoring the importance of format-aware model design.