Evals & regression gate
Quality per suite against the promoted baselines, refreshed by the six-hourly sweep, CI and post-deploy runs. The gate compares each run to the floor a human promoted — it never moves on its own.
Loading…
Quality per suite against the promoted baselines, refreshed by the six-hourly sweep, CI and post-deploy runs. The gate compares each run to the floor a human promoted — it never moves on its own.
Loading…