An extensive new archive of academic literature is now available as an open dataset on the Hugging Face platform. The 16 TB collection, shared by user @secemp9, provides access to 3,148,796 papers from the arXiv repository, including every version of the documents in LaTeX, PDF, PostScript, and HTML formats.

The project organizes the complete arXiv site into a single downloadable resource for scientific workflows. This release encompasses the full history of papers hosted on the original repository, providing a unified source for all previously published versions of the academic works.