List your company

Adversarial Attack on AdvBench 80/20 (test)

82.7ASR@1ATOXIA
-2.26819.79141.8563.909Oct 8, 2025

Evaluation results

MethodLinks
ATOXIATarget LLM=VICUNA-7B2025.1082.792.34.53
ATOXIATarget LLM=MISTRAL-7B2025.1073.199.254.42
RED-BANDIT (ϵ-GREEDY)Target LLM=VICUNA-7B2025.1071.21001.85
RED-BANDIT (UCB)Target LLM=MISTRAL-7B2025.1070.21002.31
RED-BANDIT (UCB)Target LLM=VICUNA-7B2025.1063.51002.37
AUTODAN-UNIVERSALTarget LLM=VICUNA-7B2025.1063.284.976.33
RED-BANDIT (ϵ-GREEDY)Target LLM=MISTRAL-7B2025.1061.51003.49
ADVPROMPTER-WARMSTARTTarget LLM=MISTRAL-7B2025.1058.795.941.16
ADVPROMPTERTarget LLM=MISTRAL-7B2025.1054.396.141.6
AUTODAN-UNIVERSALTarget LLM=MISTRAL-7B2025.1051.986.557.41
GCG-UNIVERSALTarget LLM=MISTRAL-7B2025.1046.299—
RED-BANDIT (ϵ-GREEDY)Target LLM=LLAMA2-7B2025.1039.496.23.5
GCG-UNIVERSALTarget LLM=VICUNA-7B2025.1036.782.7—
ADVPROMPTER-WARMSTARTTarget LLM=VICUNA-7B2025.1035.685.613.02
RED-BANDIT (UCB)Target LLM=LLAMA2-7B2025.1033.7992.4
ADVPROMPTERTarget LLM=VICUNA-7B2025.1033.487.512.09
ATOXIATarget LLM=LLAMA2-7B2025.1018.341.45.8
ADVPROMPTER-WARMSTARTTarget LLM=LLAMA2-7B2025.1012.546.1158.5
GCG-UNIVERSALTarget LLM=LLAMA2-7B2025.1012.1—
AUTODAN-UNIVERSALTarget LLM=LLAMA2-7B2025.1012.1373.72
ADVPROMPTERTarget LLM=LLAMA2-7B2025.1017.786.8