Content Moderation on Laion5B UnsafeBench (test)
72HateGGuard
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GGuard2025.12 | 72 | 70.5 | 77.5 | 61.11 | 79.8 | 72.3 | 71 | 85.1 | 46.1 | 71.5 | 81.4 | 72 | |
| GGuard2025.12 | 67 | 69 | 73.8 | 65.9 | 64.2 | 71 | 67.7 | 74.8 | 68.2 | 67.7 | 70.1 | 73.8 | |
| GPT-4VEvaluation Source=UnsafeBench dataset2025.12 | 56 | 70.6 | 77.4 | 55.7 | 86.6 | 72.4 | 89.7 | 82.7 | 60.5 | 40.5 | 71.8 | 71.2 | |
| GPT-4VEvaluation Source=UnsafeBench dataset2025.12 | 42 | 68.1 | 73.8 | 59 | 84.7 | 83.9 | 78 | 67.3 | 78 | 49.2 | 53.7 | 70.9 |