Safety & Helpfulness Evaluation on XSTest
74.8XSTest ScoreDefault
Evaluation Results
| Method | Links | |
|---|---|---|
| DefaultModel=Qwen2.5-14B2026.02 | 74.8 | |
| CBModel=Llama3-8B2026.02 | 67.2 | |
| DefaultModel=Llama3-8B2026.02 | 66 | |
| CATModel=Qwen2.5-14B2026.02 | 54.4 | |
| Diffusion-onlyModel=Llama3-8B2026.02 | 53.6 | |
| CATModel=Llama3-8B2026.02 | 46.4 | |
| DATModel=Llama3-8B2026.02 | 46.4 | |
| DATModel=Qwen2.5-14B2026.02 | 46.4 | |
| MixAT-GCGModel=Llama3-8B2026.02 | 45.6 | |
| LATModel=Llama3-8B2026.02 | 44.4 | |
| MixAT-GCGModel=Qwen2.5-14B2026.02 | 38.8 |