JetBrains-Research/EnvBench

EnvBench基准数据集，包含来自GitHub Search的仓库统计数据、多种数据分割、Git仓库、README文件、GitHub Actions工作流程和仓库文件列表。数据集针对JVM和Python构建工具进行了筛选，包括单一代码构建工具在根目录的配置文件、无Docker配置文件的仓库，以及无法通过确定性脚本正确配置的仓库。

The EnvBench benchmark dataset contains repository statistics from GitHub Search, various data splits, Git repositories, README files, GitHub Actions workflows, and lists of files in the repositories. The dataset is filtered for JVM and Python build tools, including repositories with a single build tool configuration in the root directory, repositories without Docker configuration files, and repositories that cannot be correctly configured by a deterministic script.