Content Safety on SafetyBench (test)
89.3Macro F1-ScoreGPT-3.5
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-3.52026.06 | 89.3 | |
| Qwen1.5-72BParameters=72B2026.06 | 87.7 | |
| Yuvion-32BParameters=32B2026.06 | 87.5 | |
| Qwen2-32BParameters=32B2026.06 | 86.2 | |
| DeepSeek-67BParameters=67B2026.06 | 85.9 | |
| Qwen2-8BParameters=8B2026.06 | 82.4 | |
| Yuvion-8BParameters=8B2026.06 | 81.6 | |
| Qwen2-72BParameters=72B2026.06 | 78.4 |