Prompt Reconstruction Defense (TokenInfer attack) on GPT-samples
97.68TRANo Protection
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| No ProtectionAttack Target=Response2026.02 | 97.68 | 94.56 | |
| No ProtectionAttack Target=Prompt2026.02 | 97.53 | 94.21 | |
| SPDAttack Target=Response2026.02 | 97.24 | 94.13 | |
| CapeAttack Target=Response2026.02 | 86.1 | 83.89 | |
| CapeAttack Target=Prompt2026.02 | 85.89 | 83.34 | |
| TalariaAttack Target=Prompt2026.02 | 2.11 | 45.35 | |
| TalariaAttack Target=Response2026.02 | 1.97 | 44.63 |