GRACE-LLaVA-1.5-7B-INT4 Robustness to Adversarial Visual Perturbations in Vision-Language Tasks

This report synthesises findings from 4 peer-reviewed papers addressing the following research question: How does the robustness of GRACE-LLaVA-1.5-7B-INT4 hold against adversarial visual perturbations on language understanding tasks compared to unquantized vision-language models. 10 claims were extracted from source literature; 10 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 8.8/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: How does the robustness of GRACE-LLaVA-1.5-7B-INT4 hold against adversarial visual perturbations on language understanding tasks compared to unquantized vision-language models?

Autonomous literature synthesis. Automated review score: 8.8/10. Full text and citation available at Assignee Research.