Benchmarking & automating mathematics.
ML researcher and mathematician focused on benchmarking and automating advanced mathematical problem-solving.
- University of Oxford, DPhil under T. Lukasiewicz
- Tim Gowers' group at Cambridge University
- Fields Model Initiative, benchXiv, and AIMO Prize
For AI benchmarks what arXiv is for preprints — starting with rigorous mathematics benchmarks for language models.
benchxiv.org →
Fields Model Initiative Training LLMs for mathematics that converge to Fields Medal performance. A non-profit initiative.
fieldsmodel.org →
Managing the AI Mathematical Olympiad competition series with Terence Tao, Tim Gowers, and the AIMO Advisors.
aimoprize.com →Roles
In the media
Selected talks
More links
Research
Open-source theorem prover for Euclidean geometry The mathematical datasets we need to get math copilots Language Models as Science Tutors LLM vs ITP at the NeurIPS MATH-AI workshop Mathematical Capabilities of ChatGPTMedia Archive
AI Scholar report ZDNet: risk in not knowing what OpenAI is building ZDNet: customized benchmarks and openness Synced Review reportOlder Talks
Aizawa group, University of Tokyo / NII LOD 2023 conference Neuromonster 2023 Tim Gowers' group talkTeaching
Supervised master's theses on language models, including Philip Vonderlind's work on language models and formal games.
Demonstrator for Oxford's Deep Learning in Healthcare course, plus earlier teaching in analysis, linear algebra, and machine learning.