GRAML：图证据与多任务学习提升LLM漏洞检测

1. 12:00
   arXiv · Software Engineering

   GRAML：面向 LLM 软件漏洞检测的图引导推理与多任务学习

   论文提出 GRAML 框架，将图证据、漏洞描述生成与多任务训练结合，用于 LLM 软件漏洞检测。GRAML 对 C/C++ 程序做静态分析，提取关键源码行与类型化行关系作为结构证据，引导 GPT-5 经 ToT-VR 生成漏洞描述，并与 Detection、Localization、Assessment 样本构建统一四任务数据集。

Current heat 9·Comparable peak 10（Oct 8）·Comparable change over 24 hours –

The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.