AnyAttack: Towards Large-scale Self-supervised Adversarial Attacks on Vision-language Models

AAML
VLM
Re-assign community