List your company

Red-teaming on Llama-3.3-70B

0.558Attack Success Rate (ASR)Adaptive Instruction Composition
0.1680.269250.37050.47175Apr 22, 2026

Evaluation results

MethodLinks
Adaptive Instruction CompositionBandit Type=Aggressive, lambda=0.012026.040.5585,579
Adaptive Instruction CompositionBandit Type=Subtle, lambda=0.012026.040.2472,465
WildTeamingAttack Strategy=WildTeaming2026.040.1831,830