Few Tokens Matter: Entropy Guided Attacks on Vision-Language ModelsPublished in arXiv preprint, 2025Share on Twitter Facebook LinkedIn Previous Next