arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2511.17833cs.AIcs.SE
收藏

学习调试:基于LLM的知识树用于解决RTL断言失败

Learning to Debug: LLM-Organized Knowledge Trees for Solving RTL Assertion Failures

NVIDIA

Yunsheng Bai, Haoxing Ren

AI总结:

GROVE通过构建基于LLM的知识树,有效解决RTL断言失败问题,提升调试效率和准确性。

摘要:

调试是现代硬件验证中的主要成本,其中断言失败是频率最高且最昂贵的故障类型之一。尽管大语言模型(LLMs)显示出潜力,但它们往往无法捕捉工程师应用的精确且可重用的专业知识,导致响应不准确。我们提出了GROVE,一种分层的知识管理框架,该框架学习并组织可重用的调试专业知识,形成一个基于LLM的知识树,用于解决断言失败。GROVE从先前案例中提炼调试知识,并将其组织成可配置深度的垂直树,每个节点编码一个简洁的知识条目和显式的适用条件。在训练过程中,GROVE使用并行、梯度自由的循环,通过学习案例,让LLM提出树的修改作为结构化的JSON编辑。在测试时,执行一种预算感知的迭代缩放,以导航树,检索一小部分适用的知识条目,指导基础LLM的假设生成和修复提案。在一系列断言失败案例上评估,GROVE在pass@1和pass@5上均表现出一致的提升,证明了结构化知识演化的价值。

英文摘要:

Debugging is the dominant cost in modern hardware verification, where assertion failures are among the most frequent and expensive to resolve. While Large Language Models (LLMs) show promise, they often fail to capture the precise, reusable expertise that engineers apply, leading to inaccurate responses. We propose GROVE, a hierarchical knowledge management framework that learns and organizes reusable debugging expertise into an LLM-organized knowledge tree for solving assertion failures. GROVE distills debugging knowledge from prior cases and organizes it into a vertical tree of configurable depth, with each node encoding a concise knowledge item and explicit applicability conditions. During training, GROVE uses a parallel, gradient-free loop where an LLM proposes tree modifications as structured JSON edits by learning from the cases. At test time, a budget-aware iterative zoom is performed to navigate the tree, retrieving a small set of applicable knowledge items that guide a base LLM's hypothesis generation and fix proposals. Evaluated on a suite of assertion-failure cases, GROVE delivers consistent gains in pass@1 and pass@5, demonstrating the value of structured knowledge evolution.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑