Shared eval dashboard

Loading…

Evals in test set Latest version · evals passed Previous Older

Each bar is how many test cases this feature had that day. The set grows as new cases are found, and steps up when a batch is promoted to a new set version. Each line is how many of that day's cases a pipeline version passed. Cases a version skips count as failed. Every number here is replayed from an append-only, hash-chained ledger that you can verify in your browser.