Reproduce “Reva: Real-Time Vulnerability Detection through Contextual Probing of Large Language Models” 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
Amine Lbath, Massih-Reza Amini
REVA screens functions for vulnerabilities by passing each, with program context, through an LLM once and using a trained probe to read its internal code representations instead of requesting a verdict. On PrimeVul’s held-out test set it reached 60.0% accuracy versus VulTriage’s 59.2% with GLM-5.3-Flash; REVA took 1.33 seconds per function versus 362, but the accuracy difference was inconclusive.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.