mllm-jailbreak-bench

Reproducible benchmark for adversarial attacks on multimodal large language models