List your company

Malicious Fine-tuning Attack on Advbench

84.5Attack Success Rate (ASR)Vaccine
15.54833.44951.3569.251Jun 6, 2026

Evaluation results

MethodLinks
VaccineTest-time Attack Dataset=Toxic-DPO2026.0684.5
VaccineTest-time Attack Dataset=PKU-SafeRLHF-unsafe2026.0678.5
BoosterTest-time Attack Dataset=Toxic-DPO2026.0677.7
SFTTest-time Attack Dataset=Toxic-DPO2026.0676.1
VaccineTest-time Attack Dataset=Beavertails-unsafe2026.0672.6
PatcherTest-time Attack Dataset=Toxic-DPO2026.0671.4
SFTTest-time Attack Dataset=PKU-SafeRLHF-unsafe2026.0670.4
BoosterTest-time Attack Dataset=PKU-SafeRLHF-unsafe2026.0668.2
BoosterTest-time Attack Dataset=Beavertails-unsafe2026.0657.4
SFTTest-time Attack Dataset=Beavertails-unsafe2026.0656
PatcherTest-time Attack Dataset=PKU-SafeRLHF-unsafe2026.0637.1
PatcherTest-time Attack Dataset=Beavertails-unsafe2026.0618.2