Skip to content
Longterm Wiki
Updated 2026-04-29HistoryData
Page StatusDocumentationDashboard
Edited 1 day ago
Content0/12
SummaryScheduleEntityEdit history
Tables0Diagrams0Int. links0/ ~5Ext. links0Footnotes0References0Quotes0Accuracy0

Benchmark Quarantine

Triage queue for safety-benchmark ingester rows whose raw model name could not be resolved to a known entity. Promote rows by adding an alias via crux tb model-aliases sync (which will then auto-resolve and let the next ingester run promote the row to benchmark_results); reject obvious noise by setting status = rejected with a reason.

Pending

0

Resolved

0

Rejected

0

Total

0

Queue is clear

No benchmark ingester rows are waiting for triage. New rows will appear here when an ingester encounters a raw model name that doesn't match any model_aliases entry.