Large language models (LLMs) have advanced code generation from single-function tasks to competitive-programming problems, but existing multi-agent solutions either rely on costly large-scale (>30B) models or collapse when downsized to small open-source models. We present MapCoder-Lite, a framework for distilling the complex reasoning of large, multi-agent coding systems into a single 7B model. Our contribution is a novel, three-pillar methodology that synergistically generates, refines, and encodes multi-agent knowledge: (i) pass-based trajectory distillation from strong LLMs fixes format fragility in retrieval and reduces failures in debugging, (ii) supervisor-guided correction with global feedback strengthens planning and coding agents, and (iii) agent-wise LoRA fine-tuning delivers memory-efficient specialisation. Comprehensive evaluation on xCodeEval, APPS, and CodeContests shows that MapCoder-Lite more than doubles xCodeEval accuracy (from 13.2% to 28.3%), eliminates all format failures, while reducing GPU memory and token-generation time by 4x compared to a 32B model. It also achieves over 10% gains on simpler coding benchmarks, demonstrating broad improvements beyond competitive programming. These results demonstrate that careful agent-wise fine-tuning unleashes high-quality multi-agent coding on a small language model. Our code is publicly available at https://github.com/aiha-lab/MapCoder-Lite.

翻译：大语言模型（LLM）在代码生成方面已从单函数任务发展到可解决竞赛编程问题，但现有的多智能体解决方案要么依赖成本高昂的大规模（>300亿参数）模型，要么在缩小至小型开源模型时性能崩溃。我们提出了MapCoder-Lite，这是一个将复杂的大型多智能体编码系统推理能力蒸馏至单个70亿参数模型的框架。我们的核心贡献是一种新颖的三支柱协同方法，用于生成、精炼和编码多智能体知识：（i）基于轮次的轨迹蒸馏从强LLM中提取知识，修复检索中的格式脆弱性并减少调试失败；（ii）采用全局反馈的监督器引导校正机制，增强规划与编码智能体的能力；（iii）基于智能体的LoRA微调实现内存高效的专业化。在xCodeEval、APPS和CodeContests上的综合评估表明，MapCoder-Lite将xCodeEval准确率提升一倍以上（从13.2%至28.3%），完全消除了格式错误，同时与320亿参数模型相比，GPU内存占用和令牌生成时间减少至四分之一。该模型在简单代码基准测试中也实现了超过10%的性能提升，证明了其在竞赛编程之外的广泛改进。这些结果表明，经过精细的智能体级微调，小型语言模型也能释放出高质量的多智能体编码能力。我们的代码已公开于https://github.com/aiha-lab/MapCoder-Lite。