LlavaGuard
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Unsafe content detection | LlavaGuard | Accuracy82 | 14 | |
| Prompt Classification | LlavaGuard Image Prompt | F1 Score0.752 | 7 | |
| Out-of-Taxonomy Risk Detection | LlavaGuard | F1 Score66.87 | 4 | |
| OOD safety category inference (Stage 2) | LlavaGuard | Reward Mean13.28 | 4 |
Showing 4 of 4 rows