List your company

Adversarial Attack on LIAR ICL NEW

30.35Attack Success RateAgentic Adversarial Rewriting (Full History)
0.83488.497416.1623.8226Apr 26, 2026

Evaluation results

MethodLinks
Agentic Adversarial Rewriting (Full History)Model Access=Direct Target Access, Optimization=Prompt Optimization, History=Full History2026.0430.35
Agentic Adversarial Rewriting (Previous Step)Model Access=Direct Target Access, Optimization=Prompt Optimization, History=Previous Step2026.0425
Agentic Adversarial Rewriting (Attacker Only)Model Access=Direct Target Access, Optimization=None2026.0420.11
CLAREModel Access=RoBERTa Surrogate2026.043.9
TextFoolerModel Access=RoBERTa Surrogate2026.043.09
TextBuggerModel Access=RoBERTa Surrogate2026.042.8
DeepWordBugModel Access=RoBERTa Surrogate2026.041.97