Mathematics for and by Large Language Models

Explore the intersection of mathematics and large language models through this comprehensive conference featuring six expert presentations totaling nearly six hours of content. Delve into the mathematical foundations underlying LLMs and discover how these powerful AI systems can be leveraged to advance mathematical research. Learn about synthetic data generation and its impact on model scaling, examine methods for grounding LLMs in executable environments, and investigate three fundamental problems in deep learning mathematics. Understand how mathematical tasks can be framed as translation problems and explore the importance of training data distributions in mathematical reasoning. Gain insights into the mathematical structures that LLMs implicitly compute and discover practical applications of machine learning in mathematical research. This conference brings together leading researchers from Meta AI Research, Harvard University, IHES, and other prestigious institutions to foster dialogue between the LLM community and the broader mathematical research community, addressing both theoretical understanding and practical applications in this rapidly evolving field.