Benchmark Adjudication

Assign reviewer benchmark labels to retrospective cases without touching live workflow state.

This view is benchmark-only. It lets a reviewer add explicit target labels, confidence, and rationale over completion-gated retrospective cases so the advisory ranking pilot can be judged against a sharper gold set than workflow-derived labels alone.

Back to benchmark view Open advisory benchmark pilot Export adjudications JSON Export adjudications CSV

Cases analysed

Completion-gated retrospective cases in the current benchmark window.

Adjudicated cases

Cases with an explicit reviewer benchmark label saved.

Bundle window

500

Recent bundles requested for retrospective adjudication review.

Export rows

Adjudicated cases available for benchmark export right now.

Evaluation window

Review 200 bundles Review 500 bundles Review 1000 bundles

No benchmark-ready retrospective cases are available yet.