Safety Classification on DynaBench (test)
75.8F1 ScoreCSRM
Evaluation Results
| Method | Links | |
|---|---|---|
| CSRMseverity augmentation=true, configurable safety configuration augmentation=true2026.05 | 75.8 | |
| Llama-3.1-8B-Inst2026.05 | 73.4 | |
| DynaGuard-8BChain-of-Thought=true2026.05 | 73.1 | |
| CSRMseverity augmentation=true, configurable safety configuration augmentation=false2026.05 | 69.2 | |
| CSRMseverity augmentation=false, configurable safety configuration augmentation=false2026.05 | 66.8 | |
| OSS-Safeguard-20B-High2026.05 | 66.4 | |
| Llama-3.1-70B-InstChain-of-Thought=true2026.05 | 58.3 | |
| Qwen3-30BThinking Mode=true2026.05 | 57.6 | |
| ShieldGemma-9B2026.05 | 38.2 | |
| LlamaGuard-3-8B2026.05 | 13.1 |