List your company

Adversarial Attack on LIAR-NEW Verifact

40.34Attack Success RateAgentic Adversarial Rewriting (Full History)
-0.095210.402420.931.3976Apr 26, 2026

Evaluation results

MethodLinks
Agentic Adversarial Rewriting (Full History)Model Access=Direct Target Access, Optimization=Prompt Optimization, History=Full History2026.0440.34
Agentic Adversarial Rewriting (Previous Step)Model Access=Direct Target Access, Optimization=Prompt Optimization, History=Previous Step2026.0439.56
Agentic Adversarial Rewriting (Attacker Only)Model Access=Direct Target Access, Optimization=None2026.0435.92
TextFoolerModel Access=RoBERTa Surrogate2026.042.18
TextBuggerModel Access=RoBERTa Surrogate2026.041.94
CLAREModel Access=RoBERTa Surrogate2026.041.46
DeepWordBugModel Access=RoBERTa Surrogate2026.041.46