Red-teaming on Violence prompts
0Failure Rate (FR)UnlearnDiffAtk
Evaluation Results
| Method | Links | |
|---|---|---|
| UnlearnDiffAtkSafe T2I Model=RECE, Attack Access=white-box2024.11 | 0 | |
| PEZSafe T2I Model=Receler, Attack Access=black-box2024.11 | 0.74 | |
| PEZSafe T2I Model=AdvUnlearn, Attack Access=black-box2024.11 | 1.85 | |
| Ring-A-BellSafe T2I Model=AdvUnlearn, Attack Access=black-box2024.11 | 3.2 | |
| PEZSafe T2I Model=SLD-MAX, Attack Access=black-box2024.11 | 3.7 | |
| SneakyPromptSafe T2I Model=AdvUnlearn, Attack Access=black-box2024.11 | 4.05 | |
| PEZSafe T2I Model=ESD, Attack Access=black-box2024.11 | 4.63 | |
| Ring-A-BellSafe T2I Model=ESD, Attack Access=black-box2024.11 | 4.8 | |
| P4DSafe T2I Model=SLD-MAX, Attack Access=white-box2024.11 | 5.56 | |
| Ring-A-BellSafe T2I Model=SLD-MAX, Attack Access=black-box2024.11 | 5.6 | |
| Ring-A-BellSafe T2I Model=Receler, Attack Access=black-box2024.11 | 5.6 | |
| Ring-A-BellSafe T2I Model=RECE, Attack Access=black-box2024.11 | 5.6 | |
| MMA-DiffusionSafe T2I Model=AdvUnlearn, Attack Access=white-box2024.11 | 6.02 | |
| Flirt-ScoringSafe T2I Model=RECE, Attack Access=black-box2024.11 | 6.02 | |
| MMA-DiffusionSafe T2I Model=Receler, Attack Access=white-box2024.11 | 6.38 | |
| MMA-DiffusionSafe T2I Model=SLD-MAX, Attack Access=white-box2024.11 | 6.94 | |
| SneakyPromptSafe T2I Model=Receler, Attack Access=black-box2024.11 | 7.01 | |
| P4DSafe T2I Model=SAFREE, Attack Access=white-box2024.11 | 7.41 | |
| UnlearnDiffAtkSafe T2I Model=SLD-MAX, Attack Access=white-box2024.11 | 7.41 | |
| SneakyPromptSafe T2I Model=SAFREE, Attack Access=black-box2024.11 | 9.72 | |
| MMA-DiffusionSafe T2I Model=ESD, Attack Access=white-box2024.11 | 11.11 | |
| Flirt-ScoringSafe T2I Model=AdvUnlearn, Attack Access=black-box2024.11 | 11.57 | |
| MMA-DiffusionSafe T2I Model=SAFREE, Attack Access=white-box2024.11 | 13.43 | |
| P4DSafe T2I Model=AdvUnlearn, Attack Access=white-box2024.11 | 13.89 | |
| P4DSafe T2I Model=RECE, Attack Access=white-box2024.11 | 13.89 | |
| UnlearnDiffAtkSafe T2I Model=AdvUnlearn, Attack Access=white-box2024.11 | 14.35 | |
| PEZSafe T2I Model=SAFREE, Attack Access=black-box2024.11 | 14.81 | |
| UnlearnDiffAtkSafe T2I Model=SAFREE, Attack Access=white-box2024.11 | 15.74 | |
| SneakyPromptSafe T2I Model=RECE, Attack Access=black-box2024.11 | 17.13 | |
| SneakyPromptSafe T2I Model=ESD, Attack Access=black-box2024.11 | 17.14 | |
| SneakyPromptSafe T2I Model=SLD-MAX, Attack Access=black-box2024.11 | 19.16 | |
| MMA-DiffusionSafe T2I Model=RECE, Attack Access=white-box2024.11 | 22.69 | |
| ICERSafe T2I Model=AdvUnlearn, Attack Access=black-box2024.11 | 23.3 | |
| Ring-A-BellSafe T2I Model=SAFREE, Attack Access=black-box2024.11 | 24 | |
| Flirt-ScoringSafe T2I Model=Receler, Attack Access=black-box2024.11 | 24.07 | |
| Flirt-ScoringSafe T2I Model=SAFREE, Attack Access=black-box2024.11 | 27.78 | |
| Flirt-ScoringSafe T2I Model=SLD-MAX, Attack Access=black-box2024.11 | 30.09 | |
| P4DSafe T2I Model=Receler, Attack Access=white-box2024.11 | 30.56 | |
| UnlearnDiffAtkSafe T2I Model=Receler, Attack Access=white-box2024.11 | 30.56 | |
| PEZSafe T2I Model=RECE, Attack Access=black-box2024.11 | 36.57 | |
| P4DSafe T2I Model=ESD, Attack Access=white-box2024.11 | 37.96 | |
| UnlearnDiffAtkSafe T2I Model=ESD, Attack Access=white-box2024.11 | 38.89 | |
| Flirt-ScoringSafe T2I Model=ESD, Attack Access=black-box2024.11 | 38.89 | |
| ICERSafe T2I Model=Receler, Attack Access=black-box2024.11 | 42.75 | |
| ICERSafe T2I Model=SLD-MAX, Attack Access=black-box2024.11 | 44.91 | |
| ICERSafe T2I Model=SAFREE, Attack Access=black-box2024.11 | 47.45 | |
| ICERSafe T2I Model=RECE, Attack Access=black-box2024.11 | 58.11 | |
| ICERSafe T2I Model=ESD, Attack Access=black-box2024.11 | 61.73 |