Human review
Adjudication
Review low-confidence or high-variance model evaluations with prompt context, peer output, rubric metadata, and a structured decision.
Queue
0 items require review.
Queue clear
No evaluations currently require manual adjudication.
Current evaluation
Prompt, output, expected answer, rubric, and peer variance.
No evaluation selected
Select a queued item to inspect prompt context, output, rubric, and peer variance.
Decision
Persist a structured score override and review note.