UniMaia: Steering Chess Policies with Language for Human-like Play

2026-05-26

AI Summary

This work investigates how to achieve natural language–based semantic controllability in chess policy networks without compromising their performance. To this end, the authors propose a framework that avoids end-to-end multimodal training by leveraging a parameter-efficient text encoder and a ControlNet-style conditioning mechanism to modulate a frozen Lc0 policy network via prompting, enabling flexible control over semantic attributes such as opening preferences and playing strength. They further introduce UniMaia-Aux, an auxiliary temporal prediction objective, to enhance behavioral modeling. Combining a large-scale metadata-augmented dataset with an automated prompt generation pipeline, the method achieves state-of-the-art accuracy across multiple semantic control benchmarks while maintaining competitive performance on human move prediction tasks.

Abstract

Recent advances in large language models have enabled natural language to serve as a flexible interface for controlling complex systems, but often at the cost of large-scale multimodal training or weakened domain-specific inductive biases. In structured decision-making domains such as chess, specialized policy networks achieve strong performance but lack semantic controllability, while prompt-conditioned language models are more flexible yet typically exhibit weaker domain grounding. We propose $\textbf{UniMaia}$, a framework for prompt-conditioned policy modulation that adapts a frozen Lc0-based chess policy network using a parameter-efficient text encoder and a ControlNet-style conditioning mechanism. UniMaia enables semantic control over gameplay, including opening selection and player strength, while preserving the pretrained policy representations. We further introduce $\textbf{UniMaia-Aux}$, which incorporates auxiliary temporal conditioning and behavioral prediction objectives. To support this work, we construct a large-scale metadata-augmented Lichess dataset, develop a semi-automated prompt-generation pipeline, and introduce benchmarks spanning both prompt-conditioned and metadata-conditioned settings. UniMaia achieves state-of-the-art expected accuracy on several prompt-conditioned benchmarks and competitive top-move accuracy on general instruction-following tasks, while remaining competitive with dedicated metadata-conditioned approaches on human move prediction benchmarks. UniMaia-Aux further improves expected accuracy and behavioral modeling across several evaluation settings, with modest trade-offs in top-move accuracy. Overall, our results demonstrate that prompt-conditioned control of domain-specific policy networks is feasible without end-to-end multimodal training, while highlighting trade-offs between controllability and predictive performance.

Problem

• semantic controllability
• chess policy
• language-conditioned control
• human-like play
• domain-specific policy

Innovation

• prompt-conditioned control
• parameter-efficient adaptation
• chess policy modulation
• ControlNet-style conditioning
• behavioral prediction