List your company

Safety Evaluation on XSTest Overrefusal

99.5Not Overrefuse RateHelpful-PPO
95.3496.4297.598.58Nov 2, 2024

Evaluation results

MethodLinks
Helpful-PPO2024.1199.5100
RBR-PPO2024.1199.599.5
Human-PPO2024.1195.595.5