Copilot-Code-Dataset

We have constructed a dataset of code generated by Copilot from real open-source projects on the Github platform. This dataset covers four mainstream languages: Go, Java, JavaScript, and Python, and includes 157 code repositories and 727 code files. It will be continuously updated to provide a reusable data foundation for subsequent empirical research involving multiple languages, granularities, and dimensions.我们构建了一个收录GitHub Copilot（Copilot）从GitHub平台真实开源项目中生成代码的数据集。本数据集涵盖Go、Java、JavaScript、Python四种主流编程语言，包含157个代码仓库与727个代码文件。本数据集将持续更新，旨在为后续涉及多语言、多粒度、多维度的实证研究提供可复用的数据基础。