用自然语言控制多智能体系统，让小模型也能高效运行
MinionsLLM: a Task-adaptive Framework For The Training and Control of Multi-Agent Systems Through Natural Language
- 将大模型与行为树结合，通过语法生成数据优化指令理解
- 细调后语法正确率达92.6%，任务性能平均提升33%
- 适合资源受限场景，小模型效果提升显著
本文提出MinionsLLM，一种将大语言模型（LLMs）与行为树（BTs）及形式文法结合的新型框架，实现对任意用户定义环境中的多智能体系统的自然语言控制。该框架提供标准化接口定义环境、智能体与行为基元，并引入两种合成数据集生成方法（Method A 和 Method B），用于微调大模型以提升语法正确性与语义任务相关性。我们使用谷歌Gemma 3模型系列在三个参数规模（1B、4B、12B）上验证方法，结果表明：Method B使语法正确率提升至92.6%，平均任务性能较基线提高33%。值得注意的是，小模型在微调后收益最大，为在资源受限环境下部署轻量级本地化大模型提供了可行路径。框架及所有资源均已开源，支持可复现性与后续研究。
原文摘要 · Abstract (English)
This paper presents MinionsLLM, a novel framework that integrates Large Language Models (LLMs) with Behavior Trees (BTs) and Formal Grammars to enable natural language control of multi-agent systems within arbitrary, user-defined environments. MinionsLLM provides standardized interfaces for defining environments, agents, and behavioral primitives, and introduces two synthetic dataset generation methods (Method A and Method B) to fine-tune LLMs for improved syntactic validity and semantic task relevance. We validate our approach using Google's Gemma 3 model family at three parameter scales (1B, 4B, and 12B) and demonstrate substantial gains: Method B increases syntactic validity to 92.6% and achieves a mean task performance improvement of 33% over baseline. Notably, our experiments show that smaller models benefit most from fine-tuning, suggesting promising directions for deploying compact, locally hosted LLMs in resource-constrained multi-agent control scenarios. The framework and all resources are released open-source to support reproducibility and future research.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。