Reproduce “GRAML: Graph-Grounded Reasoning and Multi-Task Learning for LLM-Based Software Vulnerability Detection” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ruohan Li, Miaoqing Tian, Haipeng Qu, Ruobing Jiang
GRAML maps code relationships from vulnerable lines into training explanations and auxiliary supervision. It trains a detector that uses source code alone at inference. Graph evidence and CVE information guide training-data construction. CodeLlama-7B reaches 68.70% average F1 across an ID test set and six out-of-distribution datasets, compared with 65.06% for DeepSeek-V4-Flash. The method is evaluated for C/C++.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.