AI data mining

MathPile is a mathematics-centric corpus of approximately 9.5 billion tokens, drawing mathematics content from textbooks (including lecture notes), arXiv, Wikipedia, ProofWiki, StackExchange, and the web for K-12, college, graduate level, and mathematics competitions. MathPile's data is of high quality and has rich data documentation to enhance transparency and give users the ability to use data flexibly. In terms of licensing, MathPile follows the BY-NC-SA 4.0 license agreement and plans to release a commercially available version soon.