Post
2327
π Exciting News! At the end of September, we're releasing an upgraded version of the MCTSr algorithm, LLaMA-O1, as an open-source fork of the OpenAI Strawberry project. This enhanced tool specializes in tackling the most challenging Mathematical Olympiad tasks, including:
A remarkable score of 8/30 in AIME 2024 π
An impressive 26.7% in GPQA Diamond π
A solid 58% in AMC 2023 π
Stay tuned for the release and get ready to explore the cutting edge of mathematical problem-solving! #OpenSource #MathOlympiad #Innovation
Former vision:
Accessing GPT-4 level Mathematical Olympiad Solutions via Monte Carlo Tree Self-refine with LLaMa-3 8B (2406.07394)
A remarkable score of 8/30 in AIME 2024 π
An impressive 26.7% in GPQA Diamond π
A solid 58% in AMC 2023 π
Stay tuned for the release and get ready to explore the cutting edge of mathematical problem-solving! #OpenSource #MathOlympiad #Innovation
Former vision:
Accessing GPT-4 level Mathematical Olympiad Solutions via Monte Carlo Tree Self-refine with LLaMa-3 8B (2406.07394)