用多智能体迭代纠错提升孟加拉语转代码准确率
PyBangla at BLP-2025 Task 2: Enhancing Bangla-to-Python Code Generation with Iterative Self-Correction and Multilingual Agents
- 构建基于思维-代码-观察循环的多智能体框架
- Qwen3-8B在开发集上达94.0%准确率，盲测71.6%
- 适合低资源语言代码生成研究者参考
大语言模型在英文指令生成代码方面表现优异，但在低资源语言领域进展有限。本文针对孟加拉语到Python代码的转换问题，提出BanglaCodeAct框架，采用多智能体提示与迭代自校正机制。不同于依赖任务特化微调的方法，该框架利用开源多语言LLM，在思维-代码-观察循环中动态生成、测试并优化代码。我们在mHumanEval数据集上评估多个小参数开源LLM的效果，结果表明：部署于BanglaCodeAct时，Qwen3-8B在开发集上达到94.0%的pass@1准确率，在盲测集上为71.6%。该成果建立了孟加拉语转代码的新基准，展示了多智能体推理在低资源语言代码生成中的潜力。实验脚本已公开于github.com/jahidulzaid/PyBanglaCodeActAgent。
原文摘要 · Abstract (English)
LLMs excel at code generation from English prompts, but this progress has not extended to low-resource languages. We address Bangla-to-Python code generation by introducing BanglaCodeAct, an agent-based framework that leverages multi-agent prompting and iterative self-correction. Unlike prior approaches relying on task-specific fine-tuning, BanglaCodeAct employs an open-source multilingual LLM within a Thought-Code-Observation loop, enabling dynamic generation, testing, and refinement of code from Bangla instructions. We benchmark several small-parameter open-source LLMs and evaluate their effectiveness on the mHumanEval dataset for Bangla NL2Code. Our results show that Qwen3-8B, when deployed with BanglaCodeAct, achieves the best performance, with pass@1 accuracy of 94.0\% on the development set and 71.6\% on the blind test set. These results establish a new benchmark for Bangla-to-Python translation and highlight the potential of agent-based reasoning for reliable code generation in low-resource languages. Experimental scripts are publicly available at github.com/jahidulzaid/PyBanglaCodeActAgent.
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。