Content Safety Detection on HUS-II HPAA-applied
0Hate Detection RateSG-9B
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SG-9BIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 0 | 0 | 0 | 0 | 0 | |
| Enkrypt AIIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 0 | 0 | 0 | 0 | 0 | |
| LG3-8BIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 0 | 0 | 0 | 0 | 0 | |
| Amazon-CIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 1.47 | 0 | 2.78 | 4.55 | 3.62 | |
| G2FIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 1.47 | 0 | 0 | 0 | 0.72 | |
| PAIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 1.52 | 0.89 | 5.88 | 2.33 | 1.52 | |
| OmniIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 2.94 | 1.75 | 2.78 | 0 | 1.45 | |
| GPT-4oIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 4.41 | 2.63 | 2.78 | 2.27 | 2.9 | |
| Azure AIIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 11.76 | 14.04 | 16.67 | 13.64 | 11.59 | |
| G-2.5-FLIteration number=1, Attack budget=1, Human exposure rate=92%, Attack configuration=M6-W-Hi2026.06 | 11.76 | 7.89 | 8.33 | 6.82 | 10.14 |