DiffCAP: Diffusion-based Cumulative Adversarial Purification for Vision Language Models

Open in new window