MiniF-2F

The 'MiniF-2F' dataset/benchmark contains formal proofs and is used to evaluate the robustness of large language models in generating formalizations from semantically similar paraphrased natural language statements.

Papers using MiniF-2F (5)

Goedel-Prover: A Frontier Model for Open-Source Automated Theorem Proving2025 · 1 citesEvaluating Autoformalization Robustness via Semantically Similar Paraphrasing2025MPS-Prover: Advancing Stepwise Theorem Proving by Multi-Perspective Search and Data Curation2025Scaling up Multi-Turn Off-Policy RL and Multi-Agent Tree Search for LLM Step-Provers2025EconProver: Towards More Economical Test-Time Scaling for Automated Theorem Proving2025