Jailbreaking on JBB n = 100
93ASRSTEER
Evaluation Results
| Method | Links | |
|---|---|---|
| STEERTarget Model=Gemma-7B, Iterations=@82026.07 | 93 | |
| STEERTarget Model=GLM-4-9B, Iterations=@82026.07 | 93 | |
| GCGTarget Model=Mistral-7B, Iterations=@82026.07 | 92 | |
| STEERTarget Model=Mistral-7B, Iterations=@82026.07 | 92 | |
| CSRTTarget Model=Mistral-7B, Iterations=@82026.07 | 90 | |
| GCGTarget Model=Gemma-7B, Iterations=@82026.07 | 89 | |
| CSRTTarget Model=GLM-4-9B, Iterations=@82026.07 | 89 | |
| CSRTTarget Model=Gemma-7B, Iterations=@82026.07 | 88 | |
| STEERTarget Model=Qwen3-8B, Iterations=@82026.07 | 85 | |
| STEERTarget Model=Llama-3-8B, Iterations=@82026.07 | 83 | |
| STEERTarget Model=DeepSeek-R1, Iterations=@82026.07 | 80 | |
| CSRTTarget Model=Llama-3-8B, Iterations=@82026.07 | 55 | |
| GCGTarget Model=Qwen3-8B, Iterations=@82026.07 | 53 | |
| GCGTarget Model=DeepSeek-R1, Iterations=@82026.07 | 47 | |
| CSRTTarget Model=Qwen3-8B, Iterations=@82026.07 | 47 | |
| CSRTTarget Model=DeepSeek-R1, Iterations=@82026.07 | 44 | |
| GCGTarget Model=Llama-3-8B, Iterations=@82026.07 | 41 | |
| GCGTarget Model=GLM-4-9B, Iterations=@82026.07 | 39 |