Harmful content detection on CADD DeepSeek generations
60.13AccuracyPerspective API
Evaluation Results
| Method | Links | |
|---|---|---|
| Perspective API2026.04 | 60.13 | |
| OpenAI Moderation2026.04 | 37.72 | |
| LlamaGuard-12026.04 | 31.28 | |
| LlamaGuard-22026.04 | 8.86 |
Loading the SOTA2 catalog…