按年被提到
20252025 · 1 次2026
数学、逻辑与理论›AI 数学与定理证明·arXiv:2502.00212 · cs.LG

STP: Self-play LLM Theorem Provers with Iterative Conjecturing and Proving

Kefan Dong, Tengyu Ma · 2025
开始阅读 ↗PDFalphaXiv 讨论
被提及 · 2 人首次 2025-02-08 · 最近 2026-07-05
2不同的人
2活跃月份
2讨论串
4被引(OpenAlex)
arXiv 摘要(候选层不写导读)

A fundamental challenge in formal theorem proving by LLMs is the lack of high-quality training data. Although reinforcement learning or expert iteration partially mitigates this issue by alternating between LLM generating proofs and finetuning them on correctly generated ones, performance quickly plateaus due to the scarcity of correct proofs (sparse rewards). To keep improving the models with limited data, we draw inspiration from mathematicians, who continuously develop new results, partly by proposing novel conjectures or exercises (which are often variants of known results) and attempting to solve them. We design the Self-play Theorem Prover (STP) that simultaneously takes on two roles, conjecturer and prover, each providing training signals to the other. The conjecturer is trained iteratively on previously generated conjectures that are barely provable by the current prover, which incentivizes it to generate increasingly challenging conjectures over time. The prover attempts to prove the conjectures with standard expert iteration. We evaluate STP with both Lean and Isabelle formal versifiers. With 51.3 billion tokens generated during the training in Lean, STP proves 28.5% of the statements in the LeanWorkbook dataset, doubling the previous best result of 13.2% achieved through expert iteration. The final model achieves state-of-the-art performance among whole-proof generation methods on miniF2F-test (65.0%, pass@3200), Proofnet-test (23.9%, pass@3200) and PutnamBench (8/644, pass@3200). We release our code, model, and dataset in this URL: https://github.com/kfdong/STP.

候选层:至少 2 个不同的人、在 2 个不同月份提到过,还没到主书架的 3 人 3 个月。没有做立场标注,所以只写“被提及”。

HN 书评

推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及
评论anuramat · 2026-07

“> this year or next maybe; I meant "in a single year eventually", not this year necessarily > what research transformers were published in 2017, and nobody anticipated agentic coding back then; but if I had to guess: 1. lean self play as in https://arxiv.org/abs/2502.00212 and https://arxiv.org/abs/2606.13473 2. "recursive self improvement"/autoresea…”

It Still Can't Do My Job: Four Years of Moving Goalposts (2022–2026)原评论 ↗
被提交为 story · 1 次