[论文] MineValiCoder: Reliable Code Generation with Test Case Quality Mining ...

论文概要

研究领域: ML
作者: Zhen Zhao, Qihang Yang, Feifei Dai, Xiangfang Li, Bo Li
发布时间: 2026-07-24
arXiv: 2607.22471

中文摘要

基于大型语言模型(LLM)的测试驱动开发(TDD)推进了自动化代码生成。然而，现有方法严重依赖人工设计的测试用例，且当仅有自然语言需求时无法有效运作。虽然最近的工作实现了自动测试生成，但它往往忽视了LLM的固有随机性，导致两个关键缺陷：错误测试产生误导性反馈扭曲代码优化，而质量参差的测试用例产生冲突的评估信号阻碍可靠代码选择。为解决这些挑战，本文提出MineValiCoder，一个基于测试用例质量与代码质量相互增强的协作闭环TDD框架。MineValiCoder包含三个模块。测试用例质量挖掘(TCQM)模块通过自验证过滤错误测试用例，提供可靠的优化监督。并行TDD精化模块使用经验证的测试用例反馈迭代优化代码并生成多样化的高质量代码候选。基于二分图的代码-测试相互验证(BiCoTeV)模块动态建模代码-测试交互并执行相互验证评分以实现稳定可靠的最优代码选择。跨四个LLM和主流基准测试的大量评估表明，MineValiCoder显著优于最先进方法。具体而言，它在HumanEval上达到96.34%的Pass@1分数，在MBPP上为87.40%，在APPS上为64.00%，在LiveCodeBench上为51.33%。这些结果证明了MineValiCoder在缓解LLM随机性和提高自动化代码生成可靠性方面的有效性。

原文摘要

Large Language Model (LLM)-based Test-Driven Development (TDD) has advanced automated code generation. However, existing approaches depend heavily on human-crafted test cases and cannot operate effectively when only natural-language requirements are available. Although recent work enables automatic test generation, it often overlooks the inherent stochasticity of LLMs, leading to two key defects: faulty tests generate misleading feedback that distorts code optimization, while mixed-quality test cases produce conflicting evaluation signals that hinder reliable code selection. To address these challenges, we propose MineValiCoder, a collaborative closed-loop TDD framework based on the mutual reinforcement of test-case quality and code quality. MineValiCoder comprises three modules. The Test ...

自动采集于 2026-07-28

#论文 #arXiv #ML #小凯