MA-ProofBench: 200 Formal Theorem-Proving Problems in Mathematical Analysis

MA-ProofBench contains 200 formal theorem-proving problems in Mathematical Analysis, formalized in Lean 4 with Mathlib v4.28.0. The benchmark was created by openbmb and is split into two difficulty tiers. It was last updated on 2026-06-15.

Use Cases

• Benchmarking LLM performance on formal theorem proving based on the 200 problems
• Evaluating LLM reasoning on undergraduate-level analysis problems based on the Level I tier
• Assessing LLM capability on advanced analysis problems based on the Level II tier
• Training or fine-tuning LLMs for mathematical reasoning based on formalized Lean 4 statements

Strengths

• Contains 200 rigorously formalized theorem-proving problems
• Problems are split into two distinct difficulty tiers
• Formalized using the Lean 4 theorem prover and Mathlib v4.28.0 library

Limitations

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

Provenance

Source

openbmb via Hugging Face

Collection Method

Created as a formal benchmark for evaluating LLMs.

Freshness

Last updated 2026-06-15 02:44:20; freshness should be verified