Training a First-Order Theorem Prover from Synthetic Data
원문 초록
A major challenge in applying machine learning to automated theorem proving is the scarcity of training data, which is a key ingredient in training successful deep learning models. To tackle this problem, we propose an approach that relies on training purely with synthetically generated theorems, without any human data aside from axioms. We use these theorems to train a neurally-guided saturation-based prover. Our neural prover outperforms the state-of-the-art E-prover on this synthetic data in both time and search steps, and shows significant transfer to the unseen human-written theorems from the TPTP library, where it solves 72\% of first-order problems without equality.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
CriticLean: Critic-Guided Reinforcement Learning for Mathematical Formalization
Safe: Enhancing Mathematical Reasoning in Large Language Models via Retrospective Step-aware Formal Verification
LeanExplore: A search engine for Lean 4 declarations
Autoformalization in the Era of Large Language Models: A Survey
DeepTheorem: Advancing LLM Reasoning for Theorem Proving Through Natural Language and Reinforcement Learning
Enumerate-Conjecture-Prove: Formally Solving Answer-Construction Problems in Math Competitions
토론
이 논문과 연결된 포럼 글이 아직 없습니다.