This dataset contains metadata and source code of 9,000 public domain (cc0 or unlicense) licensed GitHub repositories that have more than 25 stars.
The dataset was created by scraping the GitHub API and downloading the repositories, so long as they are under 100mb.
The dataset can be used for various natural language processing and software engineering tasks, such as code summarization, code generation, code search, code analysis, etc.… See the full description on the dataset page:
https://huggingface.co/datasets/KoalaAI/GitHub-CC0.
