List your company

Jailbreak Attack Robustness on Role-Play jailbreak attack

2,108DSRTGA-7B
-51.04509.481,0701,630.52Oct 16, 2024

Evaluation results

MethodLinks
TGA-7Bparameters=7B2024.102,108
Unlearn-FigS2024.101,348
Qwen-VL-chatvariant=chat2024.10455
InstructBlip2024.10295
LLaVA-v1.5version=v1.52024.1067
LLaVA-v1.6version=v1.62024.1066
BLIP-22024.1032