Jailbreak Mitigation on GCG
0GCG ASRGoal
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| GoalModel=Mistral-7B2026.04 | 0 | — | — | — | — | — | |
| ASGUARD2025.09 | 1 | 96.7 | 59.5 | 68.3 | 76 | 45 | |
| SFTData Mixture=30/702025.09 | 2 | 66.1 | 2.27 | 67.3 | 13.8 | 13.4 | |
| RepBend2025.09 | 5 | 94.3 | 50.4 | 68.1 | 73.3 | 41.7 | |
| CBModel=Mistral-7B2026.04 | 14 | — | — | — | — | — | |
| Llama-3.1-8B-InstructBackbone=Llama-3.1, Parameters=8B, Mode=Instruct2025.09 | 15 | 88.5 | 28.9 | 68.2 | — | — | |
| BeamModel=Mistral-7B2026.04 | 17.5 | — | — | — | — | — | |
| BaseModel=Mistral-7B2026.04 | 53.2 | — | — | — | — | — |