False Positive Rate Detection on WildGuard benign
3.44FPRGate
Evaluation results
| Method | Links | |
|---|---|---|
| Gatematched_operating_point=true2026.06 | 3.44 | |
| LlamaGuard 32026.06 | 13.84 | |
| GPT-4o2026.06 | 19.76 | |
| Lakera Guard2026.06 | 37.4 | |
| ImProGuard / PGuard2026.06 | 53.88 | |
| ProtectAI Deberta v22026.06 | 54.92 | |
| Llama-2-chat2026.06 | 59.93 | |
| Deepset (deberta-v3-base-injection)2026.06 | 88.01 | |
| Prompt-Guard-v1-8B2026.06 | 88.22 | |
| Meta Prompt Guard v12026.06 | 99.11 |