Olympiad-level formal mathematical reasoning is achieved with reinforcement learning, with performance reaching a score equivalent to that of a silver medallist at the 2024 International Mathematical Olympiad competition.
- Thomas Hubert
- Rishi Mehta
- David Silver