Math Olympiad Problem Formalizations for Theorem Proving

MathOlympiadBench contains human-verified formalizations of Olympiad-level mathematical competition problems. The dataset was created by Goedel-LM and introduced in a paper published in 2025. It sources problems from the Compfiles and IMOSLLean4 repository.

Use Cases

• Training language models to generate formal proofs from natural language problem statements using the formalized theorems.
• Benchmarking the performance of automated theorem provers on complex, competition-level mathematical problems.
• Studying the structure and difficulty of formalized Olympiad problems for curriculum design in AI reasoning.

Strengths

• Contains human-verified formalizations, ensuring correctness for benchmarking.
• Problems are sourced from established mathematical competition repositories like Compfiles and IMOSLLean4.

Limitations

• The specific number of problems and formalizations is not provided, making size assessment difficult.
• Potential bias towards specific types of Olympiad problems available in the source repositories.

Provenance

Source

Compfiles and IMOSLLean4 repository.

Collection Method

Human-verified formalization of existing competition problems.

Freshness

Dataset was last updated on August 6, 2025.