pith. sign in

arxiv: 2505.02735 · v1 · pith:SU7KDQC7new · submitted 2025-05-05 · 💻 cs.AI · cs.LG

FormalMATH: Benchmarking Formal Mathematical Reasoning of Large Language Models

classification 💻 cs.AI cs.LG
keywords reasoningformalformalmathmathematicalmodelsalgebraautoformalizationbenchmark
0
0 comments X
read the original abstract

Formal mathematical reasoning remains a critical challenge for artificial intelligence, hindered by limitations of existing benchmarks in scope and scale. To address this, we present FormalMATH, a large-scale Lean4 benchmark comprising 5,560 formally verified problems spanning from high-school Olympiad challenges to undergraduate-level theorems across diverse domains (e.g., algebra, applied mathematics, calculus, number theory, and discrete mathematics). To mitigate the inefficiency of manual formalization, we introduce a novel human-in-the-loop autoformalization pipeline that integrates: (1) specialized large language models (LLMs) for statement autoformalization, (2) multi-LLM semantic verification, and (3) negation-based disproof filtering strategies using off-the-shelf LLM-based provers. This approach reduces expert annotation costs by retaining 72.09% of statements before manual verification while ensuring fidelity to the original natural-language problems. Our evaluation of state-of-the-art LLM-based theorem provers reveals significant limitations: even the strongest models achieve only 16.46% success rate under practical sampling budgets, exhibiting pronounced domain bias (e.g., excelling in algebra but failing in calculus) and over-reliance on simplified automation tactics. Notably, we identify a counterintuitive inverse relationship between natural-language solution guidance and proof success in chain-of-thought reasoning scenarios, suggesting that human-written informal reasoning introduces noise rather than clarity in the formal reasoning settings. We believe that FormalMATH provides a robust benchmark for benchmarking formal mathematical reasoning.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 11 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. TheoremBench: Evaluating LLMs on Theorem Proving in Formal Mathematics

    cs.AI 2026-06 unverdicted novelty 8.0

    TheoremBench is a Lean4 benchmark of classical theorems in main and premised forms that evaluates LLM provers on partial progress, coverage, and token efficiency rather than binary success on competition problems.

  2. LAMP: Lean-based Agentic framework with MCP and Proof Repair

    cs.LO 2026-06 conditional novelty 7.0

    LAMP achieves 96.7% success generating verified Lean proofs for 90 Combinatorics on Words theorems by coordinating Planner, Builder, and Verifier agents with a CoW ontology accessed through Model Context Protocol.

  3. FVSpec: Real-World Property-Based Tests as Lean Challenges

    cs.SE 2026-05 conditional novelty 7.0

    A new benchmark of 9,415 Lean 4 specifications derived from 2,772 scraped Python property-based tests, plus a three-agent LLM transpilation pipeline and proof-generation baselines.

  4. CAM-Bench: A Benchmark for Computational and Applied Mathematics in Lean

    cs.AI 2026-05 accept novelty 7.0

    CAM-Bench is a new Lean 4 theorem-proving benchmark of 1,000 problems in computational and applied mathematics, built from textbook exercises using a dependency-recovery pipeline to reconstruct local context.

  5. Lean-GAP: A Dataset of Formalized Graduate Algebra Problems

    cs.LO 2026-05 unverdicted novelty 6.0

    Lean-GAP is a dataset of 430 graduate algebra problems formalized in Lean 4 from Dummit and Foote, with a described pipeline for autoformalization and verification plus analysis of challenges.

  6. OProver: A Unified Framework for Agentic Formal Theorem Proving

    cs.CL 2026-05 unverdicted novelty 6.0

    OProver-32B achieves top Pass@32 scores on MiniF2F, ProverBench, and PutnamBench by combining continued pretraining with iterative agentic proving, retrieval, SFT on repairs, and RL on unresolved cases using a 6.86M-p...

  7. Forge: Quality-Aware Reinforcement Learning for NP-Hard Optimization in LLMs

    cs.AI 2026-05 unverdicted novelty 6.0

    OPT-BENCH trains LLMs on NP-hard optimization via quality-aware RLVR, achieving 93.1% success rate and 46.6% quality ratio on Qwen2.5-7B while outperforming GPT-4o and transferring gains to other domains.

  8. Aristotle: IMO-level Automated Theorem Proving

    cs.AI 2025-10 unverdicted novelty 6.0

    Aristotle reaches gold-medal-equivalent performance on 2025 IMO problems via integrated Lean proof search, informal lemma formalization, and a dedicated geometry solver.

  9. MedConclusion: A Benchmark for Biomedical Conclusion Generation from Structured Abstracts

    cs.CL 2026-04 unverdicted novelty 5.0

    MedConclusion is a 5.7M-instance benchmark dataset for generating biomedical conclusions from structured PubMed abstracts, with LLM evaluations showing conclusion writing differs from summarization and that judge choi...

  10. BV-Blend: Uncertainty-Weighted Historical Baselines for Stable Critic-Free RL with Verifiable Rewards

    cs.AI 2026-06 unverdicted novelty 4.0

    BV-Blend blends prompt-local and semantic-cluster historical reward statistics via SEM-derived weights to stabilize critic-free RL advantage estimation.

  11. Rethinking Wireless Communications through Formal Mathematical AI Reasoning

    eess.SP 2026-04 unverdicted novelty 4.0

    Proposes a three-layer framework using formal AI reasoning for verification, derivation, and discovery in wireless communications theory.