Harmfulness Evaluation on HarmfulQ
0Harmfulness RateNLCf/800 step
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NLCf/800 stepSteps=800, Loss=NLCf2026.05 | 0 | 0.6 | |
| EMD/800 stepSteps=800, Loss=EMD2026.05 | 0 | 0.5 | |
| Mistral MoE-XLArchitecture=MoE-XL2026.05 | 4 | 0.004 | |
| Qwen-3B MoE-XLArchitecture=MoE-XL2026.05 | 6 | 0.006 | |
| Mistral2026.05 | 20 | 2.8 | |
| Qwen-3B2026.05 | 25 | 0.015 |