List your company

Jailbreaking on AdvBench Sub

4.73BERT ScoreNaive
1.18362.10433.0253.9457May 17, 2025

Evaluation results

MethodLinks
NaiveModel=Llama3-8B-CB2025.054.731.850.65
LINTModel=Llama3-8B-CB2025.054.571.960.73
GCGModel=Llama3-8B-CB2025.054.362.190.5
FLIPModel=Llama3-8B-CB2025.054.32.080.42
JULIModel=Qwen2.5-1.5B2025.054.154.423.31
JULIModel=Llama3-8B-CB2025.054.082.382.04
EDModel=Llama3-8B-CB2025.053.993.732.54
BaseModel=Llama3-8B-CB2025.053.911.380.54
JULIModel=Llama3-3B-INST2025.053.94.272.85
Emulated AlignmentModel=Llama3-3B-INST2025.053.843.52.65
Emulated AlignmentModel=Llama3-8B-INST2025.053.843.812.42
Emulated AlignmentModel=Llama2-7B-CHAT2025.053.682.541.5
FLIPModel=Llama3-8B-INST2025.053.532.770.77
FLIPModel=Llama3-3B-INST2025.053.522.880.73
JULIModel=Llama2-7B-CHAT2025.053.353.852.38
GCGModel=Qwen2.5-1.5B2025.053.342.812
NaiveModel=Qwen2.5-1.5B2025.053.271.810.96
FLIPModel=Qwen2.5-1.5B2025.053.252.350.69
JULIModel=Llama3-8B-INST2025.053.123.622.45
Emulated AlignmentModel=Qwen2.5-1.5B2025.053.041.230.46
LINTModel=Qwen2.5-1.5B2025.0533.691.65
WTSModel=Llama3-8B-CB2025.052.991.580.5
WTSModel=Qwen2.5-1.5B2025.052.893.151.73
BaseModel=Qwen2.5-1.5B2025.052.851.160.08
LINTModel=Llama3-8B-INST2025.052.73.51.84
BaseModel=Llama3-8B-INST2025.052.691.420.42
LINTModel=Llama2-7B-CHAT2025.052.673.51.35
NaiveModel=Llama3-8B-INST2025.052.461.810.85
NaiveModel=Llama2-7B-CHAT2025.052.422.812.19
GCGModel=Llama3-3B-INST2025.052.232.581.85
FLIPModel=Llama2-7B-CHAT2025.052.210.02
GCGModel=Llama3-8B-INST2025.052.131.850.92
NaiveModel=Llama3-3B-INST2025.052.051.460.54
WTSModel=Llama3-8B-INST2025.052.052.231.27
WTSModel=Llama3-3B-INST2025.051.971.810.85
LINTModel=Llama3-3B-INST2025.051.772.691.12
BaseModel=Llama3-3B-INST2025.051.731.580.46
WTSModel=Llama2-7B-CHAT2025.051.671.620.42
GCGModel=Llama2-7B-CHAT2025.051.351.380.42
BaseModel=Llama2-7B-CHAT2025.051.321.040.15