This if the official implementation for paper: SneakyPrompt: Jailbreaking Text-to-image Generative Models

python main.py --target='sd' --method='rl' --reward_mode='clip' --threshold=0.26 --len_subword=10 --q_limit=60 --safety='ti_sd'

You can change the parameters follow the choices in 'search.py'. The adversarial prompts and statistic results (xx.csv) will be saved under '/results', and the generated images will be saved under '/figure'

python evaluate.py --path='PATH OF xx.csv'

Please cite our paper if you find this repo useful.

@inproceedings{yang2023sneakyprompt,
      title={SneakyPrompt: Jailbreaking Text-to-image Generative Models},
      author={Yuchen Yang and Bo Hui and Haolin Yuan and Neil Gong and Yinzhi Cao},
      year={2024},
      booktitle={Proceedings of the IEEE Symposium on Security and Privacy (Oakland)}
}