List your company
ResearchFormerly WizWand

Defense against adversarial prompts on Ring-A-Bell (test)

0Attack Success RateGLoCE
-3.925222.569949.06575.5601Jul 1, 2026
Updated 4mo ago

Evaluation results

MethodLinks
GLoCEBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.070
AdvUnlearnBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.070.93
DESBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.070.93
SAGEBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.071.87
GuardT2IBase Model=SDv1.5, Evaluation Classifier=filtering accuracy2026.073.1
SalUnBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.073.74
OpenAIBase Model=SDv1.5, Evaluation Classifier=filtering accuracy2026.0719.26
UCEBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.0721.5
ESDBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.0726.17
Latent GuardBase Model=SDv1.5, Evaluation Classifier=filtering accuracy2026.0743.1
MicrosoftBase Model=SDv1.5, Evaluation Classifier=filtering accuracy2026.0744.02
SAFREEBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.0776.64
SPMBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.0791.59
SLD-strongBase Model=SDv1.5, Evaluation Classifier=NudeNet2026.0797.2
SDv1.5Base Model=SDv1.5, Evaluation Classifier=NudeNet2026.0798.13