Saved in:
Bibliographic Details
Main Authors: Barkallah, Slim, Bailey, Luke, Wen, Kaiyue, Abouzaid, Mohammed, Ma, Tengyu
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.20531
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910239396200448
author Barkallah, Slim
Bailey, Luke
Wen, Kaiyue
Abouzaid, Mohammed
Ma, Tengyu
author_facet Barkallah, Slim
Bailey, Luke
Wen, Kaiyue
Abouzaid, Mohammed
Ma, Tengyu
contents Reliable verification of proofs remains a bottleneck for training and evaluating AI systems on hard mathematical reasoning. Fully formal proofs, in languages like Lean, are easy to verify because they are unambiguous and modular. Most proofs, particularly those written by AI systems, have neither property, and translating them into formal languages remains challenging in many frontier math settings. We propose Pseudo-Formalization (PF), a proof format that captures the modularity and precision of formal proofs while retaining the flexibility of natural language. A Pseudo-Formal proof is decomposed into self-contained modules, each stating its premises, conclusion, and proof in natural language. To verify the correctness of a regular natural language proof, an LLM translates it to Pseudo-Formal and then verifies each module independently, an algorithm we call Block Verification (BV). We evaluate PF+BV on two benchmarks spanning olympiad and research-level mathematics, where it pareto-dominates LLM-as-judge baselines on error-finding precision and recall. To support future work, we release our research-level proof verification benchmark ArxivMathGradingBench.
format Preprint
id arxiv_https___arxiv_org_abs_2605_20531
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Pseudo-Formalization for Automatic Proof Verification
Barkallah, Slim
Bailey, Luke
Wen, Kaiyue
Abouzaid, Mohammed
Ma, Tengyu
Logic in Computer Science
Machine Learning
Reliable verification of proofs remains a bottleneck for training and evaluating AI systems on hard mathematical reasoning. Fully formal proofs, in languages like Lean, are easy to verify because they are unambiguous and modular. Most proofs, particularly those written by AI systems, have neither property, and translating them into formal languages remains challenging in many frontier math settings. We propose Pseudo-Formalization (PF), a proof format that captures the modularity and precision of formal proofs while retaining the flexibility of natural language. A Pseudo-Formal proof is decomposed into self-contained modules, each stating its premises, conclusion, and proof in natural language. To verify the correctness of a regular natural language proof, an LLM translates it to Pseudo-Formal and then verifies each module independently, an algorithm we call Block Verification (BV). We evaluate PF+BV on two benchmarks spanning olympiad and research-level mathematics, where it pareto-dominates LLM-as-judge baselines on error-finding precision and recall. To support future work, we release our research-level proof verification benchmark ArxivMathGradingBench.
title Pseudo-Formalization for Automatic Proof Verification
topic Logic in Computer Science
Machine Learning
url https://arxiv.org/abs/2605.20531