Harmlessness Evaluation on DNA (test)
100Tie RateSHARD
Evaluation Results
| Method | Links | |
|---|---|---|
| SHARDModel=Qwen3.5 27B2026.06 | 100 | |
| Benign IntentModel=Qwen3.5 27B2026.06 | 100 | |
| SHARDModel=Qwen3.5 9B2026.06 | 100 | |
| Benign IntentModel=Qwen3.5 9B2026.06 | 100 | |
| SHARDModel=Phi-42026.06 | 99.9 | |
| Benign IntentModel=Phi-42026.06 | 99.8 | |
| SHARDModel=Mistral 24B2026.06 | 99.7 | |
| SHARDModel=Llama 3.1 8B2026.06 | 99.6 | |
| Benign IntentModel=Llama 3.1 8B2026.06 | 99.6 | |
| Benign IntentModel=Mistral 24B2026.06 | 99.6 | |
| SHARDModel=Llama 3.3 70B2026.06 | 99.5 | |
| Benign IntentModel=Llama 3.3 70B2026.06 | 99 | |
| SHARDModel=Mistral 7B2026.06 | 96.9 | |
| Benign IntentModel=Mistral 7B2026.06 | 96.9 |