On the Zero-shot Adversarial Robustness of Vision-Language Models: A Truly Zero-shot and Training-free Approach.

Baoshun Tong, Hanjiang Lai, Yan Pan, Jian Yin

Browse the full CVPR paper archive.