Response Classification on Public Safety Benchmarks Response Suite
89.9BeaverT ScoreBeaverDam
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| BeaverDamModel Size=7B2026.03 | 89.9 | 72.1 | 58.4 | 72.2 | 63.4 | 71.2 | |
| MD-JudgeModel Size=7B2026.03 | 86.7 | 64.7 | 81.4 | 81.2 | 76.8 | 78.2 | |
| WildGuardModel Size=7B2026.03 | 84.1 | 64.2 | 86.3 | 83.5 | 75.8 | 78.8 | |
| EXPGUARDModel Size=7B2026.03 | 81.8 | 64.1 | 85.5 | 82.7 | 78.3 | 78.5 | |
| HarmBench-MistralModel Size=7B2026.03 | 75.1 | 52.6 | 86.8 | 56.9 | 60.1 | 66.3 | |
| Aegis-Guard-DModel Size=7B2026.03 | 74.9 | 59.1 | 62.5 | 79.5 | 50.3 | 65.3 | |
| Aegis-Guard-PModel Size=7B2026.03 | 73.9 | 54.9 | 60.8 | 80.8 | 56.8 | 65.4 | |
| HarmBench-LlamaModel Size=13B2026.03 | 72 | 59.8 | 84.3 | 61.2 | 45.7 | 64.6 | |
| Llama-Guard2Model Size=8B2026.03 | 71.8 | 51.6 | 77.8 | 67.8 | 66.5 | 67.1 | |
| Llama-Guard3Model Size=8B2026.03 | 68 | 45.1 | 84.5 | 66 | 70.5 | 66.8 | |
| Llama-GuardModel Size=7B2026.03 | 67.1 | 48.4 | 52 | 70.4 | 50.5 | 57.7 | |
| ShieldGemmaModel Size=9B2026.03 | 66.6 | 51.4 | 53.7 | 73.5 | 44.6 | 58 | |
| AzureModel Size=-2026.03 | 34.7 | 20.1 | 35.2 | 37.1 | 26.6 | 30.7 | |
| PerspectiveModel Size=-2026.03 | 19.3 | 13.9 | 7.1 | 28.1 | 4.1 | 14.5 | |
| DetoxifyModel Size=-2026.03 | 18.4 | 15.9 | 5.7 | 34.3 | 2.1 | 15.3 | |
| OpenAI ModerationModel Size=-2026.03 | 15.7 | 9.7 | 20.6 | 84.1 | 37.1 | 33.4 |