数学与形式推理 颠覆级 暂无讲解视频
发表时间
2026-02-09
arXiv
2602.08384

收录解读

- 分级:`颠覆性` - 正式标题:`Towards Real-World Industrial-Scale Verification: LLM-Driven Theorem Proving on seL4` - 原文:`2026-02-09-M3_AutoReal_seL4-Towards_Real_World_Industrial_Scale_Verification_LLM_Driven_Theorem_Proving_on_s.pdf` - 抽取:`extracted.md`

## 重写摘要

这篇论文最重要的一点,是把 LLM 驱动的证明从学术玩具 benchmark 推进到真实工业级形式化验证项目 seL4。作者提出 AutoReal,并基于该方法微调得到一个可本地部署的 7B 级证明器。系统结合了两类关键增强:一是 CoT 风格的证明训练,使模型不仅给 proof script,也学习步骤间的推理逻辑;二是上下文增强,把项目内部的证明背景显式注入模型。

论文在 seL4 重要理论集的 660 个定理上报告了 51.67% 的成功率。这个结果的意义不是“数字本身有多高”,而是它证明了本地、小模型、面向真实工程系统的证明器已经开始可用,而不再依赖超大闭源模型。

## 为什么重要

形式化验证能不能在工业里落地,关键不在 miniF2F,而在 seL4、协议、内核、编译器这类真实工程对象。AutoReal 属于把“数学能力”翻译成“工程工具”的关键过桥工作。

## 局限

需要继续确认它到底是推理能力提升,还是更强的上下文拼接和库风格拟合。真正的外部验证,必须看它在非 seL4 项目上的迁移表现。

原始摘要与中文对照

中文对照翻译

摘要。形式化方法(FM)可靠但应用成本高昂,在seL4等工业级项目中,尤其是在定理证明方面,通常需要数年专家级的努力。大型语言模型(LLMs)的最新进展使得自动化定理证明变得越来越可行。然而,大多数现有工作侧重于miniF2F等面向数学的基准测试,对真实世界验证项目的评估有限。少数考虑工业级验证的研究大多依赖于拥有数千亿参数的闭源模型,这些模型无法在本地部署并产生高昂的使用成本。在本文中,我们提出了AutoReal,一种用于真实世界工业级系统的LLM驱动定理证明方法,支持轻量级本地部署。我们以seL4-Isabelle验证项目作为代表性和挑战性的案例研究,对AutoReal进行了评估。AutoReal包含两项关键改进:(1) 基于思维链(CoT)的证明训练,它教会LLM证明步骤背后的推理,并支持在证明的同时提供分步解释;(2) 上下文增强,它利用项目中的证明上下文来增强LLM驱动的证明。基于AutoReal方法,我们微调了一个基础模型,获得了AutoReal-Prover,一个用于工业级定理证明的紧凑型7B规模证明器。AutoReal-Prover在来自seL4指定重要理论的660个定理上,在所有10个seL4证明类别中,实现了51.67%的证明成功率,显著优于之前在seL4上的尝试(27.06%)。为了评估泛化能力,我们将AutoReal-Prover进一步应用于形式化证明档案库(AFP)中的三个安全相关项目:CRYSTALS-Kyber_Security、RSAPSS和Elliptic_Curves_Group_Law,涵盖所有451个定理,并取得了53.88%的证明成功率。总的来说,这项工作推动了LLM驱动的定理证明在真实世界工业级验证中的应用。

原始摘要

Abstract. Formal methods (FM) are reliable but costly to apply, often requiring years of expert effort in industrial-scale projects such as seL4, especially for theorem proving. Recent advances in large language models (LLMs) have made automated theorem proving increasingly feasible. How- ever, most prior work focuses on mathematics-oriented benchmarks such as miniF2F, with limited evaluation on real-world verification projects. The few studies that consider industrial-scale verification mostly rely on closed-source models with hundreds of billions of parameters, which cannot be locally deployed and incur substantial usage costs. In this paper, we propose AutoReal, an LLM-driven theorem proving method for real-world industrial-scale systems with support for lightweight local de- ployment. We evaluate AutoReal on the seL4-Isabelle verification project as a representative and challenging case study. AutoReal incorporates two key improvements: (1) chain-of-thought (CoT)-based proof training, which teaches the LLM the reasoning behind proof steps and enables step-wise explanations alongside proofs, and (2) context augmentation, which leverages proof context from the project to enhance LLM-driven proving. Based on the AutoReal methodology, we fine-tune a base model to obtain AutoReal-Prover, a compact 7B-scale prover for industrial-scale theorem proving. AutoReal-Prover achieves a 51.67% proof success rate on 660 theorems from seL4-designated Important Theories across all 10 seL4 proof categories, substantially outperforming prior attempts on seL4 (27.06%). To evaluate generalization, we further apply AutoReal-Prover to three security-related projects from the Archive of Formal Proofs (AFP): CRYSTALS-Kyber_Security, RSAPSS, and Elliptic_Curves_Group_Law, covering all 451 theorems and achieving a proof success rate of 53.88%. Overall, this work advances the application of LLM-driven theorem prov- ing in real-world industrial-scale verification.

相关论文

链接