Prompt Reconstruction Defense (TokenInfer attack) on WikiText2
97.54TRANo Protection
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| No ProtectionAttack Target=Response2026.02 | 97.54 | 0.9401 | |
| No ProtectionAttack Target=Prompt2026.02 | 97.21 | 0.9444 | |
| SPDAttack Target=Response2026.02 | 96.87 | 0.9375 | |
| CapeAttack Target=Response2026.02 | 84.65 | 0.8906 | |
| CapeAttack Target=Prompt2026.02 | 84.55 | 0.8984 | |
| TalariaAttack Target=Prompt2026.02 | 0.47 | 0.4299 | |
| TalariaAttack Target=Response2026.02 | 0.38 | 0.4242 |