List your company
ResearchFormerly WizWand

Safety Evaluation on MM-SafetyBench v1.0 (test)

0.6ASRLLaVA-1.5-7B + Mixed-SFT
-3.11421.955547.02572.0945Mar 14, 2025
Updated 8mo ago

Evaluation results

MethodLinks
LLaVA-1.5-7B + Mixed-SFTAttack Status=Before, Tuning Strategy=Full-parameter2025.030.6
LLaVA-1.5-7B + Posthoc-SFTAttack Status=Before, Tuning Strategy=Full-parameter2025.030.6
LLaVA-1.5-7B-LoRA + Mixed-SFTAttack Status=Before, Tuning Strategy=LoRA2025.030.6
LLaVA-1.5-7B-LoRA + Posthoc-SFTAttack Status=Before, Tuning Strategy=LoRA2025.030.6
LLaVA-1.5-7B + RMU-UnlearningAttack Status=Before, Tuning Strategy=Full-parameter2025.032.98
LLaVA-1.5-7B-LoRA + NPO-UnlearningAttack Status=Before, Tuning Strategy=LoRA2025.034.17
LLaVA-1.5-7B + NPO-UnlearningAttack Status=Before, Tuning Strategy=Full-parameter2025.034.76
LLaVA-1.5-7B-LoRA + RMU-UnlearningAttack Status=Before, Tuning Strategy=LoRA2025.035.36
LLaVA-1.5-7B + RMU-UnlearningAttack Status=After, Tuning Strategy=Full-parameter2025.0317.26
LLaVA-1.5-7B-LoRA + RMU-UnlearningAttack Status=After, Tuning Strategy=LoRA2025.0319.05
LLaVA-1.5-7B + NPO-UnlearningAttack Status=After, Tuning Strategy=Full-parameter2025.0320.24
LLaVA-1.5-7B-LoRA + NPO-UnlearningAttack Status=After, Tuning Strategy=LoRA2025.0323.81
LLaVA-1.5-7B + Posthoc-SFTAttack Status=After, Tuning Strategy=Full-parameter2025.0340.48
LLaVA-1.5-7B-LoRA + Posthoc-SFTAttack Status=After, Tuning Strategy=LoRA2025.0341.07
LLaVA-1.5-7BAttack Status=Before, Tuning Strategy=Full-parameter2025.0348.81
LLaVA-1.5-7B + Mixed-SFTAttack Status=After, Tuning Strategy=Full-parameter2025.0348.81
LLaVA-1.5-7B + Unsafe-FilterAttack Status=Before, Tuning Strategy=Full-parameter2025.0350.6
LLaVA-1.5-7B-LoRAAttack Status=Before, Tuning Strategy=LoRA2025.0357.74
LLaVA-1.5-7B-LoRA + Unsafe-FilterAttack Status=Before, Tuning Strategy=LoRA2025.0358.93
LLaVA-1.5-7B-LoRA + Mixed-SFTAttack Status=After, Tuning Strategy=LoRA2025.0363.69
LLaVA-1.5-7B + Unsafe-FilterAttack Status=After, Tuning Strategy=Full-parameter2025.0390.28
LLaVA-1.5-7B-LoRA + Unsafe-FilterAttack Status=After, Tuning Strategy=LoRA2025.0391.07
LLaVA-1.5-7BAttack Status=After, Tuning Strategy=Full-parameter2025.0391.27
LLaVA-1.5-7B-LoRAAttack Status=After, Tuning Strategy=LoRA2025.0393.45