Jailbreaking on JailBreak
1LG4 ASRAutoDAN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AutoDANTarget Model=google/gemma-3-1b-it, Decoding=greedy, max_new_tokens=5122026.04 | 1 | 0.12 | |
| No attackTarget Model=google/gemma-3-1b-it, Decoding=greedy, max_new_tokens=5122026.04 | 5 | 0.42 | |
| Steering-SAETarget Model=google/gemma-3-1b-it, Decoding=greedy, max_new_tokens=5122026.04 | 6 | 0.57 | |
| GCGTarget Model=google/gemma-3-1b-it, Decoding=greedy, max_new_tokens=5122026.04 | 12 | 0.62 | |
| PAPTarget Model=google/gemma-3-1b-it, Decoding=greedy, max_new_tokens=5122026.04 | 15.6 | 1.27 | |
| Refusal-DirectionTarget Model=google/gemma-3-1b-it, Decoding=greedy, max_new_tokens=5122026.04 | 35 | 0.66 | |
| CRaFTTarget Model=google/gemma-3-1b-it, Decoding=greedy, max_new_tokens=5122026.04 | 52 | 2.57 | |
| Refusal-SAETarget Model=google/gemma-3-1b-it, Decoding=greedy, max_new_tokens=5122026.04 | 53 | 1.87 |