List your company

Instruction Following on AlpacaLlama

82.97Task PerformanceClean
71.06274.153577.24580.3365Dec 22, 2025

Evaluation results

MethodLinks
CleanAttack Method=Clean2025.1282.97—0—
Causal DetoxifyAttack Method=Causal Detoxify2025.1281.560.78170.34520.2753
Two-StepAttack Method=Two-Step2025.1281.130.86290.78170.4424
FusionAttack Method=Fusion2025.1280.980.63950.38580.2331
OverpoisonAttack Method=Overpoison2025.1278.560.84770.74110.4401
AdaptiveAttack Method=Adaptive Poison2025.1275.040.89850.65990.3975
Extreme PoisonAttack Method=Extreme Poison2025.1271.520.93910.74920.4371