Safety Classification on Generic Synthetic Self-harm (test)
0.19Violation RateLlamaGuard-3-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| LlamaGuard-3-8BParameters=8B2026.05 | 0.19 | |
| ShieldGemma-9BParameters=9B2026.05 | 0.56 | |
| Qwen3Guard-4BParameters=4B2026.05 | 4.16 |
Loading the SOTA2 catalog…