Harmful Prompt Detection on SneakyPrompt
100PrecisionDetoxify (Orig)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Detoxify (Orig)2026.04 | 100 | 28 | 44 | |
| Latent Guard2026.04 | 91 | 41 | 57 | |
| HyPE2026.04 | 68 | 93 | 78 | |
| NSFW-Classifier2026.04 | 67 | 93 | 78 | |
| GuardT2I2026.04 | 52 | 95 | 66 | |
| DiffGuard2026.04 | 46 | 85 | 60 |