Policy Responsiveness on LlavaGuard v1 (test)
98.75PERCompAgent
Evaluation Results
| Method | Links | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CompAgentCategory=O6: Weapons or Substance Abuse2025.10 | 98.75 | 97.3 | 98.02 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=O6: Weapons or Substance Abuse2025.10 | 98.75 | 94.41 | 96.53 | — | — | — | — | — | — | — | — | — | — | |
| CompAgentCategory=O8: Animal Cruelty2025.10 | 98.57 | 91.61 | 94.96 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=O8: Animal Cruelty2025.10 | 98.57 | 85.36 | 91.49 | — | — | — | — | — | — | — | — | — | — | |
| CompAgentCategory=O1: Hate, Humiliation, Harassment2025.10 | 97.5 | 95.14 | 96.3 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=O1: Hate, Humiliation, Harassment2025.10 | 96.25 | 94.36 | 95.29 | — | — | — | — | — | — | — | — | — | — | |
| CompAgentCategory=O2: Violence, Harm, or Cruelty2025.10 | 95.56 | 91.31 | 93.39 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=O2: Violence, Harm, or Cruelty2025.10 | 95.56 | 87.54 | 91.37 | — | — | — | — | — | — | — | — | — | — | |
| CompAgentCategory=Overall2025.10 | 93.98 | 94.61 | 94.29 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=Overall2025.10 | 93.38 | 92.87 | 93.13 | — | — | — | — | — | — | — | — | — | — | |
| CompAgentCategory=O9: Disasters or Emergencies2025.10 | 92.5 | 92.94 | 92.72 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=O9: Disasters or Emergencies2025.10 | 92.5 | 92 | 92.25 | — | — | — | — | — | — | — | — | — | — | |
| CompAgentCategory=O7: Self-Harm2025.10 | 91.67 | 85.94 | 88.71 | — | — | — | — | — | — | — | — | — | — | |
| CompAgentCategory=O3: Sexual Content2025.10 | 90 | 86.71 | 88.32 | — | — | — | — | — | — | — | — | — | — | |
| CompAgentCategory=O4: Nudity Content2025.10 | 90 | 93.75 | 91.84 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=O3: Sexual Content2025.10 | 90 | 86.71 | 88.32 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=O4: Nudity Content2025.10 | 90 | 91.67 | 90.83 | — | — | — | — | — | — | — | — | — | — | |
| CompAgentCategory=O5: Criminal Planning2025.10 | 89.33 | 89.67 | 89.5 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=O7: Self-Harm2025.10 | 88.33 | 81.77 | 84.93 | — | — | — | — | — | — | — | — | — | — | |
| LlavaGuardCategory=O5: Criminal Planning2025.10 | 88 | 86.83 | 87.41 | — | — | — | — | — | — | — | — | — | — |