Jailbreak Attack on WildGuard (test)
82.64ASROriginal Weight
Evaluation Results
| Method | Links | |
|---|---|---|
| Original WeightModel=Llama3 8B Instruct, Attack Protocol=Weight Orthogonalization2025.04 | 82.64 | |
| NPOModel=Llama3 8B Instruct, Attack Protocol=Weight Orthogonalization2025.04 | 81.04 | |
| Original Weight ModelModel=Qwen2.5 14B Instruct2025.04 | 33.11 | |
| CB*Model=Llama3 8B Instruct, Attack Protocol=Weight Orthogonalization2025.04 | 30.95 | |
| Original Weight ModelModel=Gemma2 2B Instruct2025.04 | 28.7 | |
| REPBENDModel=Llama3 8B Instruct, Attack Protocol=Weight Orthogonalization2025.04 | 7.48 | |
| REPBENDModel=Qwen2.5 14B Instruct2025.04 | 6.67 | |
| REPBENDModel=Gemma2 2B Instruct2025.04 | 1.34 |