No usage records yet.
This screen reads the pipeline's usage ledger — figures appear after the first pipeline run records per-call usage.
Calibration0 reviewed
| Band | Decisions | Reviewed | Approved | Corrected | Rejected | Overturn rate |
|---|---|---|---|---|---|---|
| 0.10-0.15 | 2 | 0 | 0 | 0 | 0 | insufficient sample (n=0 < 5) |
| 0.80-0.85 | 5 | 0 | 0 | 0 | 0 | insufficient sample (n=0 < 5) |
| 0.85-0.90 | 2 | 0 | 0 | 0 | 0 | insufficient sample (n=0 < 5) |
| 0.90-0.95 | 1 | 0 | 0 | 0 | 0 | insufficient sample (n=0 < 5) |
10 decisions · 10 banded · 0 human-reviewed · 0 auto-shipped (excluded)
Auto-shipped decisions carry no human outcome and are EXCLUDED from every overturn rate above. Unreviewed is never treated as correct. Bands below the minimum sample report no rate. This report measures calibration only; theta is a human setting and is unchanged by it.