ReproAgent：基于合约引导的论文到代码复现
ReproAgent: Contract-Guided Paper-to-Code Reproduction
浏览论文内容
中文总结 AI 辅助
ReproAgent是基于合约引导的四阶段论文到代码复现流水线，在PaperBench Code-Dev基准上，其在两种骨干模型的同架构支架中取得最高平均得分，相关产物已公开。
中文摘要 AI 辅助
论文到代码复现要求科学AI智能体将研究论文转化为可执行代码仓库，以保留论文的方法、协议和产物。这一任务颇具难度，因为规范存在分散性：算法、指标和产物等明确的论文内容常因智能体的长轨迹而丢失，而框架默认值、从相关工作继承的约定等隐含细节则未在论文中提及。我们提出ReproAgent，这是一个围绕持久实现合约构建的四阶段Prepare--Plan--Generate--Repair流水线，包含两个通道：实现需求通道将论文片段转化为代码义务，参考证据通道从相关仓库检索内容和结构证据。两者均绑定到工作包，投影为文件级合约，并在生成和修复阶段使用。在PaperBench Code-Dev基准上，ReproAgent在Claude-Sonnet-4.5和Gemini-3-Flash两种骨干模型的同架构支架中取得最高平均得分。端到端通道消融实验和单论文案例均验证了两个通道的贡献，代码和实验产物已公开。
英文摘要
Paper-to-code reproduction asks scientific AI agents to turn research papers into executable repositories that preserve the paper's method, protocol and artifacts. This is difficult because the specification is split: explicit paper content such as algorithms, metrics and artifacts is often lost across long agent trajectories, while implicit details such as framework defaults and conventions inherited from related work are absent from the paper. We introduce ReproAgent, a four-stage Prepare--Plan--Generate--Repair pipeline built around a persistent implementation contract with two channels: an implementation-requirement channel that turns paper snippets into code obligations, and a reference-evidence channel that retrieves content and structure evidence from related repositories. Both are bound to work packages, projected into file-level contracts, and consumed across generation and repair. On PaperBench Code-Dev, ReproAgent reaches the highest mean score among same-backbone scaffolds under both Claude-Sonnet-4.5 and Gemini-3-Flash. End-to-end channel ablations and per-paper cases support the contribution of both channels. Code and experimental artifacts are publicly available.
发表机构
- Beihang University（北京航空航天大学）
- Shanghai Jiao Tong University（上海交通大学）
- Peking University（北京大学）
- Minzu University of China（中央民族大学）
- Zhongguancun Academy（中关村科学院）
机构由 AI 辅助整理，请以论文原文为准。
