?
Show eval result
Opens one run's aggregates, with its worst queries on request.
admin.dataSourceEvals.getRun(runId, { queries?, worst? })
Opens one run's full aggregates: recall, MRR, NDCG, latency, per-tag breakdown, and the retrieval-branch split. --worst surfaces the queries that scored lowest, which is where a retrieval problem shows itself.
Parameters
runId
string
RequiredRun id (positional).
--queries
flag
Include per-query results.
--worst
number
Show the N worst-scoring queries.
CLI
remy-admin datasources eval result run_abc123 --worst 20