Sellopale/SWE-PolyBench_500

SWE-PolyBench是一个多语言仓库级软件工程基准测试数据集。目前包含四种编程语言：Python、Java、Javascript和Typescript，每种语言的实例数量分别为：Javascript 1017个、Typescript 729个、Python 199个、Java 165个。数据集分为三个子集：完整数据集、分层抽样数据集（500个实例）和已验证数据集（394个实例）。数据集结构包括多个列，如实例ID、补丁、仓库信息、基础提交、问题描述、创建日期、测试补丁、问题陈述等，用于评估编码代理/模型的性能。数据集文本主要为英文，采用MIT许可证。

SWE-PolyBench is a multi-language repo-level software engineering benchmark. Currently it includes 4 languages: Python, Java, Javascript, and Typescript, with the number of instances in each language being: Javascript: 1017, Typescript: 729, Python: 199, Java: 165. There are three datasets available: the full dataset, a stratified sampled dataset with 500 instances, and a verified dataset with 394 instances. The dataset structure includes columns such as instance_id, patch, repo, base_commit, hints_text, created_at, test_patch, problem_statement, etc., for evaluating the performance of coding agents/models. The text of the dataset is primarily English, and it is licensed under MIT.