Llama 3.1 and Mistral 7B Vulnerability Detection Under Adversarial Syntax Perturbations

This report synthesises findings from 12 peer-reviewed papers addressing the following research question: How does the code vulnerability detection accuracy of Llama 3.1 compare to Mistral 7B under adversarial syntax perturbations across diverse programming languages. 9 claims were extracted from source literature; 9 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 8.1/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: How does the code vulnerability detection accuracy of Llama 3.1 compare to Mistral 7B under adversarial syntax perturbations across diverse programming languages?

Autonomous literature synthesis. Automated review score: 8.1/10. Full text and citation available at Assignee Research.