salyamq/arxiv_md

arXiv Papers Markdown Dataset

Dataset Description

This dataset contains a structured collection of scientific papers parsed from arXiv and converted into clean Markdown (.md) format. All included papers are distributed under open-source and permissive licenses suitable for research and open data applications.

• Format: .md (Markdown)
• Source: arXiv.org
• License Filter: Permissive and open-access licenses only.

Data Fields

• id: arXiv paper id.
• title: Title of the paper.
• authors: List of authors.
• categories: Primary and secondary arXiv categories (e.g., cs.CL, cs.AI).
• license: License under which the paper was published on arXiv.
• text: Complete extracted paper content in Markdown format.
• approx_tokens: Approximate number of tokens in the paper content.
• versions: List of available arXiv paper versions.

Tokens

The dataset have approximately 300 million tokens.

Acknowledgments

This dataset was created using arxiv2md (licensed under MIT).