Safety Classification on OR-Bench
83.1F1 ScoreCSRM
Evaluation Results
| Method | Links | |
|---|---|---|
| CSRMprotocol=response-classification2026.05 | 83.1 | |
| LlamaGuard-3-8Bprotocol=response-classification2026.05 | 78.4 | |
| MistralModel Version=Mistral-7B-Instruct-v0.32025.01 | 77 | |
| ShieldGemma-9Bprotocol=response-classification2026.05 | 64.4 | |
| Llama 3Model Version=Llama-3.1-8B-Instruct2025.01 | 63 | |
| Llama-3.1-8B-Instprotocol=response-classification2026.05 | 63 | |
| Zephyr RMUModel Version=Zephyr_RMU2025.01 | 62 | |
| OSS-Safeguard-20Bprotocol=response-classification2026.05 | 46.9 |