ProofNet#: A Lean 4 Benchmark for Undergraduate Mathematics Formalization

ProofNet# is a port of the ProofNet benchmark for autoformalization and formal proving of undergraduate-level mathematics. The dataset contains 371 examples, each consisting of a formal theorem. It was created by PAug and last updated on March 24, 2025.

Use Cases

• Benchmarking autoformalization models based on the 371 undergraduate-level mathematics examples.
• Evaluating formal theorem-proving systems using the Lean 4 compatible theorems.
• Training AI models for mathematical reasoning on a structured set of formal statements.

Strengths

• Contains 371 specific examples for benchmarking.
• Explicitly compatible with Lean versions v4.7.0 through v4.16.0-rc2.
• Includes fixes and a comparison with previous ports, suggesting active maintenance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

PAug on Hugging Face.

Collection Method

Port of the original ProofNet benchmark to Lean 4.

Freshness

Last updated 2025-03-24 17:16:58; freshness should be verified.