The CoT Encyclopedia: Analyzing, Predicting, and Controlling how a Reasoning Model will Think
This work addresses the problem of improving interpretability and control over reasoning in large language models for researchers and practitioners, though it is incremental as it builds on existing CoT categorization efforts.
The paper tackled the limited understanding of reasoning strategies in chain-of-thought (CoT) usage by introducing the CoT Encyclopedia, a bottom-up framework that automatically analyzes and steers model reasoning, resulting in more interpretable analyses and enabling performance gains through strategy prediction and guidance.
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.