Prompt Injection Detection on SafeGuard
82.65AccuracyML
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MLEvaluation Protocol=Zero-shot2026.06 | 82.65 | 82.65 | 90.38 | 9.62 | |
| ProtectAI v1Evaluation Protocol=Zero-shot2026.06 | 81.53 | 57.24 | 40.79 | 59.21 | |
| LWVEvaluation Protocol=Zero-shot2026.06 | 58.16 | 58.16 | 91.79 | 8.21 | |
| NeMoGuardEvaluation Protocol=Zero-shot2026.06 | 45.22 | 47.62 | 21.27 | 78.73 |