PairCoder++:结对编程作为验证代码驱动的多模态与结构化工件生成的通用范式
PairCoder++: Pair Programming as a Universal Paradigm for Verified Code-Driven Multimodal and Structured-Artifact Generation
AI总结:
提出PairCoder框架,通过驱动者和导航者两个智能体结对编程,利用工具链反馈验证代码生成,在17个基准测试中显著提升可验证工件的生成质量。
摘要:
代码是大语言模型生成结构化工件的媒介:图表、科学图形、矢量图形、CAD模型、3D场景和硬件设计都是通过编写程序产生的。在这种模式下,单次推理是脆弱的,因为决定工件是否存在的编译器、渲染器或模拟器对模型不可见。我们提出了PairCoder,它将审查扎根于工具链,并实现为两个智能体结对编程:驱动者智能体编写程序,导航者智能体根据验证证据(诊断信息、执行结果以及当前工件与目标工件的渲染对比)进行审查,当错误持续存在时两者交换角色。在来自三个供应商的七个模型的17个公开基准测试中,PairCoder在几乎所有工件可验证的基准测试上都有改进,基于完整的官方指标套件而不仅仅是执行(例如,Blender场景可执行性从0.20提升到0.78;每个模型上的TikZ编译率提升10到30个百分点),成本是单模型成本的2.9到9.2倍(总体约7倍)。改进集中在工具链提供信息丰富的预言机且基线留有提升空间的地方,而在预言机较弱的方法中则持平或略有下降;我们将结对编程定位为验证代码驱动生成的可靠方法。
英文摘要:
Code is the medium through which large language models generate structured artifacts: charts, scientific figures, vector graphics, CAD models, 3D scenes, and hardware designs are all produced by writing programs. In this regime single pass inference is brittle, because the compiler, renderer, or simulator that decides whether the artifact exists is invisible to the model. We present PairCoder, which grounds review in the toolchain and realizes it as two agent pair programming: a Driver agent writes the program, a Navigator agent reviews it against verification evidence (diagnostics, execution results, and renderings of the current artifact beside the target), and the two switch roles when errors persist. Across 17 public benchmarks and seven models from three vendors, PairCoder improves essentially every benchmark whose artifact is verifiable, on full official metric suites rather than execution alone (for example, Blender scene executability 0.20 to 0.78; TikZ compile rate up 10 to 30 points on every model), at 2.9 to 9.2 times single model cost (about 7 times overall). The improvements concentrate where the toolchain provides an informative oracle and the baseline leaves headroom, and the method ties or mildly regresses where the oracle is weak; we frame pair programming as a reliable recipe for verified code driven generation.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。