List your company

Style Manipulation Attack on Llama3-8B-Inst

1.481ScoreBITE
-0.059240.340630.74051.14037May 24, 2026

Evaluation results

MethodLinks
BITEOptimization Steps=25, Attack Type=Black-box2026.051.481
JudgeDeceiverOptimization Steps=600, Attack Type=White-box2026.051
JudgeDeceiverOptimization Steps=200, Attack Type=White-box2026.050.818
JudgeDeceiverOptimization Steps=300, Attack Type=White-box2026.050.818
JudgeDeceiverOptimization Steps=100, Attack Type=White-box2026.050.727
JudgeDeceiverOptimization Steps=25, Attack Type=White-box2026.050