miniF2F-Lean Revisited: Reviewing Limitations and Charting a Path Forward
Azim Ospanov, Farzan Farnia, Roozbeh Yousefzadeh
cs.AI
Nov 5, 2025 · v1
TL;DR
Audits and corrects the Lean miniF2F benchmark, releasing miniF2F-v2 with verified formal and informal statements and proofs.
Abstract
We perform a thorough analysis of the formal and informal statements in the miniF2F benchmark from the perspective of an AI system that is tasked to participate in a math Olympiad consisting of the problems in miniF2F. In such setting, the model has to read and comprehend the problems in natural language, formalize them in Lean language, then proceed with proving the problems, and it will get credit for each problem if the formal proof corresponds to the original informal statement presented to the model. Our evaluation results reveal that the best accuracy of such pipeline can be about 36% using the SoTA models in the literature, considerably lower than the individual SoTA accuracies, 97% and 69% reported in the autoformalization and theorem proving literature. Analyzing the failure modes, we trace back a considerable portion of this drop to discrepancies between the formal and informal statements for more than half of the problems in miniF2F. We proceed with correcting all the errors, discrepancies and simplifications in formal and informal statements, and present the miniF2F-v2 with fully verified formal and informal statements and proofs. Evaluating the full theorem proving pipeline on miniF2F-v2 leads to the best accuracy of 70%, a significant improvement from the 40% on the original miniF2F, yet indicating considerable misalignment between the autoformalization models and theorem provers. Our deep analysis suggests that a higher quality benchmark can help the community better evaluate progress in the field of formal reasoning and also better diagnose the failure and success modes of autoformalization and theorem proving models. Our dataset is available at
https://github.com/roozbeh-yz/miniF2F_v2.
Problem
The miniF2F benchmark is widely used to evaluate formal theorem provers, but discrepancies between its formal and informal statements affect more than half the problems, leading to misleading accuracy measurements for end-to-end proving pipelines.
Approach
The authors perform a thorough analysis of both formal and informal statements in miniF2F from the perspective of an AI system tasked with reading informal problems, formalizing them in Lean, and proving them. They identify and correct errors, discrepancies, and simplifications across all problems, producing miniF2F-v2 with fully verified formal and informal statements and proofs. The full theorem-proving pipeline is then re-evaluated on the corrected benchmark.
Results
The best end-to-end pipeline accuracy on original miniF2F is about 36%, far below the individual SoTA of 97% for autoformalization and 69% for theorem proving. Evaluating on miniF2F-v2 yields 70% best accuracy (up from 40% on original miniF2F), while still indicating considerable misalignment between autoformalization models and theorem provers.
