Harmful content detection on Ours CADD-based synthetic
65.55AccuracyPerspective API
Evaluation Results
| Method | Links | |
|---|---|---|
| Perspective API2026.04 | 65.55 | |
| OpenAI Moderation2026.04 | 37.17 | |
| LlamaGuard-12026.04 | 20.83 | |
| LlamaGuard-22026.04 | 0.77 |
Loading the SOTA2 catalog…