This dataset consists of Markdown-formatted texts of mathematical research papers generated by various open-source and proprietary LLMs for my thesis in the summer of 2025.
It was firstly generated in TeX format, then compiled into PDF and then extacted with OCR model (well, that was the experiment setup).
