Evaluations
Evaluation observability dashboard
Inspect real exported qrels-backed evaluation runs, aggregate metrics, per-query results, trace links, and report JSON.
Snapshot source /demo-data/evaluations.json
Public evaluation snapshot
Metrics, query results, and reports are real exported outputs from completed local evaluation runs.
Metric comparison
Recent runs with real Recall@10, MRR@10, and NDCG@10 values.
No numeric metric values available for charting.
Latency comparison
Recent runs with real measured latency fields.
No latency values available for charting.
Evaluation runs
0 runs loaded from snapshot exports.
No evaluation runs returned for the current filters.
Select an evaluation run
No run is selected.
No fake run details are shown when no real run is selected.