AI Daybook
AI Daybook
오늘의 지면
논문연구·논문

daVinci-Prover, 형식 정리 증명에 에이전트 중심 학습 적용

daVinci-Prover: Agent-Native Mid-Training for Formal Theorem Proving

매체
Takara TLDR
게시
중요도
3/5
화제 지수
—

한국어 요약

연구팀이 형식 정리 증명 모델 향상을 위해 분해·탐색·자체 검증·수정을 포함한 337억 토큰의 에이전트 중심 중간 학습 코퍼스를 구성했다. Qwen3-8B 기준 MiniF2F·FATE-M·ProofNet에서 평균 13.7점 향상을 달성했고, Qwen3.5 시리즈에 적용하면 MiniF2F에서 거의 포화 성능을 낸다.

원문 발췌 · English

Advances in large language models (LLMs) are rapidly transforming formal theorem proving. Yet further progress remains constrained by the lack of a transparent and effective data recipe for systematically improving theorem-proving models. In this work, we argue that effective agentic provers require strong mathematical skill priors to ground theorem proving and agentic trajectories to enable long-horizon, interactive proof construction. To this end, we synthesize an agent-native 33.7B-token mid-training corpus covering decomposition, search, self-verification, and revision, together with verifier-grounded agentic trajectories. We further introduce the daVinci-Prover harness, which enables models to construct proofs and interact with a Lean verifier in an isolated workspace, and use it to collect approximately 36K high-quality agent trajectories for supervised fine-tuning. Our mid-training recipe yields an average improvement of 13.7 points over the matched Qwen3-8B SFT-only baseline across MiniF2F, FATE-M, and ProofNet. When applied to the Qwen3.5 series, our approach also yields strong performance across theorem-proving benchmarks and achieves near-saturating MiniF2F performance with memory compaction. The results highlight agent-native mid-training as an effective foundation for developing stronger interactive theorem-proving agents.

#llm#agent#theorem-proving
원문 기사 읽기
전체 보기 →