List your company

Over-refusal evaluation on OR-Bench-Hard-1K (test)

60.6F1 ScoreRobust Harmful Features detector
44.68848.81952.9557.081Jun 26, 2026

Evaluation results

MethodLinks
Robust Harmful Features detectorBase Model=Llama-3-8B2026.0660.641.8
VanillaBase Model=Llama-3-8B2026.0657.132.9
GradientCuffBase Model=Llama-3-8B2026.06515.3
CASTBase Model=Llama-3-8B2026.0645.329