LeanFormalizer_SFT / README.md
Inuyasha2023ch's picture
Update README.md
98430bc verified
metadata
license: apache-2.0
datasets:
  - SJTULean/LeanStatement_SFT
language:
  - en
result:
  dataset:
    type: Formalizer Lean
    name: miniF2F
  metrics:
    - name: pass@1
      type: pass@1
      value: 0.9406
      verified: false
base_model:
  - Qwen/Qwen2.5-7B-Instruct

LeanFormalizer_SFT

Based on Qwen2.5-7b and trained on our LeanStatement_SFT dataset, our model achieves state-of-the-art performance in formal mathematics verification, with a pass@1 compilation success rate of 94.1% (459/488) on MiniF2F and 75.4% (282/374) on ProofNet benchmarks as of December 2024(while our PPO model achieves 95.3% on MiniF2F and 76.0% on Proofnet). These results demonstrate the effectiveness of scaling laws in mathematical formalization using the Lean theorem prover.