Modelpedia
Work in progress
About
Findings
Models
Concepts
Methods
Datasets
Sources
Light
Dark
HarmBench Classifier
text · discriminative ·
anchor
Findings
IC-309
As zero-shot safety judges, GPT-4o achieves 78.9% Cohen's kappa agreement with human annotators while Llama-3-8B-instruct (39.0%) and Mistral-7B-instruct-v0.2 (53.9%) perform substantially worse