devngho/the-stack-mini

该数据集包含代码相关的数据，特征包括代码的哈希值（hexsha）、大小（size）、扩展名（ext）、语言（lang）、最大星标仓库路径（max_stars_repo_path）、最大星标仓库名称（max_stars_repo_name）、最大星标仓库头部哈希值（max_stars_repo_head_hexsha）、最大星标仓库许可证（max_stars_repo_licenses）、最大星标数量（max_stars_count）、最大星标仓库星标事件最小和最大时间（max_stars_repo_stars_event_min_datetime和max_stars_repo_stars_event_max_datetime）、最大问题仓库路径（max_issues_repo_path）、最大问题仓库名称（max_issues_repo_name）、最大问题仓库头部哈希值（max_issues_repo_head_hexsha）、最大问题仓库许可证（max_issues_repo_licenses）、最大问题数量（max_issues_count）、最大问题仓库问题事件最小和最大时间（max_issues_repo_issues_event_min_datetime和max_issues_repo_issues_event_max_datetime）、最大分叉仓库路径（max_forks_repo_path）、最大分叉仓库名称（max_forks_repo_name）、最大分叉仓库头部哈希值（max_forks_repo_head_hexsha）、最大分叉仓库许可证（max_forks_repo_licenses）、最大分叉数量（max_forks_count）、最大分叉仓库分叉事件最小和最大时间（max_forks_repo_forks_event_min_datetime和max_forks_repo_forks_event_max_datetime）、内容（content）、平均行长度（avg_line_length）、最大行长度（max_line_length）和字母数字比例（alphanum_fraction）。数据集分为训练集，包含6219883个样本，总大小为39589908892字节。数据来源于bigcode/the-stack-dedup，任务类别为文本生成和填充掩码，语言为代码，许可证为其他。

This dataset contains code-related data, with features including code hash (hexsha), size, extension (ext), language (lang), path to the repository with the most stars (max_stars_repo_path), name of the repository with the most stars (max_stars_repo_name), head hash of the repository with the most stars (max_stars_repo_head_hexsha), licenses of the repository with the most stars (max_stars_repo_licenses), number of stars for the repository with the most stars (max_stars_count), minimum and maximum datetime of star events for the repository with the most stars (max_stars_repo_stars_event_min_datetime and max_stars_repo_stars_event_max_datetime), path to the repository with the most issues (max_issues_repo_path), name of the repository with the most issues (max_issues_repo_name), head hash of the repository with the most issues (max_issues_repo_head_hexsha), licenses of the repository with the most issues (max_issues_repo_licenses), number of issues for the repository with the most issues (max_issues_count), minimum and maximum datetime of issue events for the repository with the most issues (max_issues_repo_issues_event_min_datetime and max_issues_repo_issues_event_max_datetime), path to the repository with the most forks (max_forks_repo_path), name of the repository with the most forks (max_forks_repo_name), head hash of the repository with the most forks (max_forks_repo_head_hexsha), licenses of the repository with the most forks (max_forks_repo_licenses), number of forks for the repository with the most forks (max_forks_count), minimum and maximum datetime of fork events for the repository with the most forks (max_forks_repo_forks_event_min_datetime and max_forks_repo_forks_event_max_datetime), content, average line length (avg_line_length), maximum line length (max_line_length), and alphanumeric fraction (alphanum_fraction). The dataset is divided into a training set containing 6219883 samples with a total size of 39589908892 bytes. The data is sourced from bigcode/the-stack-dedup, with task categories including text generation and fill-mask, language as code, and license as other.