NickIBrody/ruby-code-corpus

一个大型的Ruby源代码语料库，收集自GitHub上的公共仓库。包含294,074个文件，来自3,920多个Ruby仓库，这些仓库按星标数量排名（作为质量信号）。数据集经过了过滤，移除了小于200字节的文件（琐碎/空文件），并且只包含宽松许可证（MIT、Apache-2.0、BSD、ISC、Ruby）。文件类型分布为约95%的.rb文件，约37%的测试文件（spec/test，有意保留，对训练有用），约58%的生产代码。数据集字段包括来源、仓库、仓库URL、文件路径、语言、许可证、星标数量、分支名称、文件大小和原始源代码内容。

A large corpus of Ruby source code collected from public GitHub repositories. Contains 294,074 files from 3,920+ Ruby repositories on GitHub, sourced from repos ranked by star count (quality signal). Filtered to remove files under 200 bytes (trivial/empty files) and only includes permissive licenses: MIT, Apache-2.0, BSD, ISC, Ruby. File composition is approximately 95% .rb files, ~37% test files (spec/test — kept intentionally, useful for training), and ~58% production code. Dataset fields include source, repo, repo_url, path, language, license, stars, ref, size_bytes, and raw source content.