EVERGREEN / EVIDENCE

Accuracy and performance dashboard

Every number on this page is generated from committed files by tools/build-dashboard-data.ts. The data is data.json; its inputs are benchmarks/thresholds.json and the public corpus. Nothing on this page is typed by hand.

Last full run

Results per detector

Level is the promise recorded in benchmarks/thresholds.json. Earned level is the highest level the counted corpus evidence supports under decision 0013. Where a level is higher than its earned level, the row says so.

Evidence required to earn a level

The thresholds an earned level is measured against, from decision 0013.

Definitions

What each metric means, how it is computed, and the file it comes from.

Limitations