Jailbreak Attack on JailbreakBench
0ASR@10DR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DRTarget Model=Llama-2-7b-chat2024.04 | 0 | — | — | — | |
| HumanTarget Model=Llama-2-7b-chat2024.04 | 0 | — | — | — | |
| DRTarget Model=Llama-2-13b-chat2024.04 | 0 | — | — | — | |
| HumanTarget Model=Llama-3-8B-Instruct2024.04 | 0 | — | — | — | |
| DRTarget Model=Gemma-2-9b-it2024.04 | 0 | — | — | — | |
| OriginalTarget Model=OLMoE-1B-7B2026.02 | 0 | — | — | — | |
| SHIPSTarget Model=DeepSeek-V2-Lite2026.02 | 0 | — | — | — | |
| TAPTarget Model=Llama-2-7b-chat2024.04 | 1 | — | — | — | |
| TAPTarget Model=Llama-2-13b-chat2024.04 | 1 | — | — | — | |
| HumanTarget Model=Llama-2-13b-chat2024.04 | 1 | — | — | — | |
| HumanTarget Model=Llama-3.1-8B-Instruct2024.04 | 1 | — | — | — | |
| DRTarget Model=Llama-3.1-8B-Instruct2024.04 | 2 | — | — | — | |
| DRTarget Model=Llama-3-8B-Instruct2024.04 | 4 | — | — | — | |
| DRTarget Model=Vicuna-13b-v1.52024.04 | 4 | — | — | — | |
| DRTarget Model=Qwen2.5-7B-Instruct2024.04 | 5 | — | — | — | |
| PAIRTarget Model=Llama-3.1-8B-Instruct2024.04 | 6 | — | — | — | |
| TAPTarget Model=Llama-3.1-8B-Instruct2024.04 | 7 | — | — | — | |
| DRTarget Model=Qwen2-7B-Chat2024.04 | 7 | — | — | — | |
| OriginalTarget Model=Qwen1.5-MoE-A2.7B2026.02 | 7 | — | — | — | |
| SAFExTarget Model=OLMoE-1B-7B2026.02 | 7 | — | — | — | |
| TAPTarget Model=Llama-3-8B-Instruct2024.04 | 8 | — | — | — | |
| PAIRTarget Model=Llama-2-13b-chat2024.04 | 9 | — | — | — | |
| PAIRTarget Model=Llama-2-7b-chat2024.04 | 10 | — | — | — | |
| DRTarget Model=Vicuna-7b-v1.32024.04 | 11 | — | — | — | |
| DRTarget Model=Vicuna-7b-v1.52024.04 | 11 | — | — | — | |
| PAIRTarget Model=Llama-3-8B-Instruct2024.04 | 14 | — | — | — | |
| SteerMoETarget Model=OLMoE-1B-7B2026.02 | 14 | — | — | — | |
| TAPTarget Model=OLMoE-1B-7B2026.02 | 14 | — | — | — | |
| RSTarget Model=Llama-2-7b-chat2024.04 | 15 | — | — | — | |
| OriginalTarget Model=DeepSeek-V2-Lite2026.02 | 15 | — | — | — | |
| OriginalTarget Model=Average2026.02 | 16 | — | — | — | |
| SHIPSTarget Model=OLMoE-1B-7B2026.02 | 16 | — | — | — | |
| RSTarget Model=Llama-2-13b-chat2024.04 | 21 | — | — | — | |
| PAIRTarget Model=Gemma-2-9b-it2024.04 | 24 | — | — | — | |
| GCGTarget Model=OLMoE-1B-7B2026.02 | 24 | — | — | — | |
| TAPTarget Model=Gemma-2-9b-it2024.04 | 26 | — | — | — | |
| SHIPSTarget Model=Average2026.02 | 29 | — | — | — | |
| SAFExTarget Model=DeepSeek-V2-Lite2026.02 | 31 | — | — | — | |
| SAFExTarget Model=Average2026.02 | 31 | — | — | — | |
| SHIPSTarget Model=Qwen1.5-MoE-A2.7B2026.02 | 33 | — | — | — | |
| TAPTarget Model=Qwen2.5-7B-Instruct2024.04 | 34 | — | — | — | |
| Dataset-Level Top-5 RoSaisTarget Model=OLMoE-1B-7B2026.02 | 34 | — | — | — | |
| TAPTarget Model=Mixtral-8x7B2026.02 | 36 | — | — | — | |
| Dataset-Level Top-5 RoSaisTarget Model=Qwen1.5-MoE-A2.7B2026.02 | 37 | — | — | — | |
| GCGTarget Model=DeepSeek-V2-Lite2026.02 | 38 | — | — | — | |
| DRTarget Model=Mistral-7B-Instruct-v0.22024.04 | 39 | — | — | — | |
| SAFExTarget Model=Qwen1.5-MoE-A2.7B2026.02 | 39 | — | — | — | |
| OriginalTarget Model=Mixtral-8x7B2026.02 | 40 | — | — | — | |
| TAPTarget Model=Vicuna-13b-v1.52024.04 | 41 | — | — | — | |
| PAIRTarget Model=Qwen2-7B-Chat2024.04 | 42 | — | — | — | |
| PAIRTarget Model=Qwen2.5-7B-Instruct2024.04 | 44 | — | — | — | |
| DRTarget Model=Mistral-7B-Instruct-v0.32024.04 | 44 | — | — | — | |
| Sample-Level Top-5 RoSaisTarget Model=OLMoE-1B-7B2026.02 | 45 | — | — | — | |
| GCGTarget Model=Average2026.02 | 46 | — | — | — | |
| SAFExTarget Model=Mixtral-8x7B2026.02 | 46 | — | — | — | |
| SteerMoETarget Model=Qwen1.5-MoE-A2.7B2026.02 | 46 | — | — | — | |
| Sample-Level Top-5 RoSaisTarget Model=DeepSeek-V2-Lite2026.02 | 46 | — | — | — | |
| PAIRTarget Model=Vicuna-13b-v1.52024.04 | 47 | — | — | — | |
| TAPTarget Model=Average2026.02 | 47 | — | — | — | |
| TAPTarget Model=Qwen2-7B-Chat2024.04 | 49 | — | — | — | |
| SteerMoETarget Model=Average2026.02 | 50 | — | — | — | |
| TAPTarget Model=Vicuna-7b-v1.52024.04 | 51 | — | — | — | |
| PAIRTarget Model=Mistral-7B-Instruct-v0.22024.04 | 52 | — | — | — | |
| PAIRTarget Model=Mistral-7B-Instruct-v0.32024.04 | 52 | — | — | — | |
| PAIRTarget Model=Vicuna-7b-v1.32024.04 | 54 | — | — | — | |
| TAPTarget Model=Vicuna-7b-v1.32024.04 | 55 | — | — | — | |
| Dataset-Level Top-5 RoSaisTarget Model=Average2026.02 | 55 | — | — | — | |
| TAPTarget Model=Mistral-7B-Instruct-v0.32024.04 | 57 | — | — | — | |
| GCGTarget Model=Llama-3.1-8B-Instruct2024.04 | 58 | — | — | — | |
| PAIRTarget Model=Vicuna-7b-v1.52024.04 | 58 | — | — | — | |
| GCGTarget Model=Qwen1.5-MoE-A2.7B2026.02 | 58 | — | — | — | |
| Sample-Level Top-5 RoSaisTarget Model=Average2026.02 | 59 | — | — | — | |
| TAPTarget Model=DeepSeek-V2-Lite2026.02 | 60 | — | — | — | |
| TAPTarget Model=Mistral-7B-Instruct-v0.22024.04 | 61 | — | — | — | |
| GCGTarget Model=Mixtral-8x7B2026.02 | 63 | — | — | — | |
| Sample-Level Top-5 RoSaisTarget Model=Qwen1.5-MoE-A2.7B2026.02 | 63 | — | — | — | |
| RSTarget Model=Llama-3.1-8B-Instruct2024.04 | 64 | — | — | — | |
| SteerMoETarget Model=DeepSeek-V2-Lite2026.02 | 66 | — | — | — | |
| SHIPSTarget Model=Mixtral-8x7B2026.02 | 67 | — | — | — | |
| Dataset-Level Top-5 RoSaisTarget Model=Mixtral-8x7B2026.02 | 68 | — | — | — | |
| HumanTarget Model=Qwen2.5-7B-Instruct2024.04 | 70 | — | — | — | |
| SteerMoETarget Model=Mixtral-8x7B2026.02 | 73 | — | — | — | |
| GCGTarget Model=Llama-3-8B-Instruct2024.04 | 74 | — | — | — | |
| HumanTarget Model=Qwen2-7B-Chat2024.04 | 74 | — | — | — | |
| GCGTarget Model=Llama-2-7b-chat2024.04 | 76 | — | — | — | |
| TAPTarget Model=Qwen1.5-MoE-A2.7B2026.02 | 77 | — | — | — | |
| Dataset-Level Top-5 RoSaisTarget Model=DeepSeek-V2-Lite2026.02 | 79 | — | — | — | |
| GCGTarget Model=Llama-2-13b-chat2024.04 | 80 | — | — | — | |
| DSNTarget Model=Llama-3.1-8B-Instruct2024.04 | 81 | — | — | — | |
| GCGTarget Model=Vicuna-7b-v1.32024.04 | 81 | — | — | — | |
| Sample-Level Top-5 RoSaisTarget Model=Mixtral-8x7B2026.02 | 81 | — | — | — | |
| RSTarget Model=Llama-3-8B-Instruct2024.04 | 83 | — | — | — | |
| RS_self-transferTarget Model=Llama-2-7b-chat2024.04 | 84 | — | — | — | |
| F-SOURTarget Model=OLMoE-1B-7B2026.02 | 86 | — | — | — | |
| HumanTarget Model=Vicuna-7b-v1.52024.04 | 87 | — | — | — | |
| GCGTarget Model=Gemma-2-9b-it2024.04 | 88 | — | — | — | |
| HumanTarget Model=Vicuna-7b-v1.32024.04 | 88 | — | — | — | |
| GCGTarget Model=Vicuna-7b-v1.52024.04 | 88 | — | — | — | |
| F-SOURTarget Model=Qwen1.5-MoE-A2.7B2026.02 | 88 | — | — | — | |
| RS_self-transferTarget Model=Llama-3-8B-Instruct2024.04 | 89 | — | — | — |