Grounded Chess Reasoning in Language Models via Master Distillation

2026-03-20

AI Summary

This work addresses the challenge of limited reasoning interpretability in language models within domains where expert systems—such as chess engines—excel despite scarce training data. The authors propose the “Master Distillation” framework, which, for the first time, translates the complete reasoning chains of expert systems into structured, natural-language step-by-step explanations. These explanations are then leveraged in a training pipeline integrating supervised fine-tuning, reinforcement learning, and topic-balanced sampling. The resulting 4B-parameter model, C1, achieves 48.1% accuracy on chess reasoning tasks—surpassing all open-source models and most closed-source counterparts—while using two orders of magnitude fewer tokens, thereby enabling efficient, interpretable, and strategically robust reasoning.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Natural Language Processing: (Large) Language Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Language models often lack grounded reasoning capabilities in specialized domains where training data is scarce but bespoke systems excel. We introduce a general framework for distilling expert system reasoning into natural language chain-of-thought explanations, enabling compact models to acquire domain expertise and the ability to generate faithful, grounded explanations. Rather than distilling only final outputs, we capture the full reasoning process, transforming opaque expert computations into transparent, step-by-step explanations. We demonstrate this approach in chess, a canonical reasoning domain where language models continue to underperform. Our 4B parameter model, C1, advances from a near-zero baseline to 48.1% accuracy, outperforming all open-source models and most frontier proprietary systems. Notably, C1 surpasses its distillation teacher and generates solutions in two orders of magnitude fewer tokens than baselines. Unlike prior neural chess approaches that predict only best moves, C1 generates explainable solutions revealing strategic reasoning. Our pipeline combines supervised fine-tuning and reinforcement learning with theme-balanced data sampling for comprehensive tactical coverage. Master Distillation demonstrates how to inject expert-level knowledge into compact models for under-optimized domains, offering a recipe for unlocking RLVR where LLMs lack sufficient base capabilities.

Problem

• grounded reasoning
• language models
• expert systems
• chess
• domain expertise

Innovation

• Master Distillation
• Chain-of-Thought Reasoning
• Expert System Distillation
• Explainable AI
• Language Model Alignment