MPS-Prover, a stepwise Lean prover with curated training data and multi-perspective tree search, reports 75.82% on miniF2F and 32.97% on ProofNet, a new 7B-class step-level state of the art.
Deepmath-deep sequence models for premise selection
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
MPS-Prover: Advancing Stepwise Theorem Proving by Multi-Perspective Search and Data Curation
MPS-Prover, a stepwise Lean prover with curated training data and multi-perspective tree search, reports 75.82% on miniF2F and 32.97% on ProofNet, a new 7B-class step-level state of the art.