Repository navigation
Import authenticated US calibration baseline - #225
Merged
Merged
postil-dev / postil/review
succeeded
Oct 20, 2026 in 32s
No merge-relevant findings
Postil reviewed this change and found nothing that affects the merge decision.
1 suppressed
Correct the aggregate gate-verdict correctness
bench/baseline-us.json:193
Suppressed: below the configured confidence threshold.
Severity error, confidence 0.34.
The ten gateVerdictCorrect values on lines 37, 53, 69, 85, 101, 117, 133, 149, 165, and 181 sum to 592. With 10 runs of 70 cases, that is 592/700 = 0.8457142857142857, not the recorded 0.8357142857142857 (585/700). This makes the committed benchmark baseline report a false correctness rate; recompute the aggregate from the per-run values or correct the per-run data before merging.
Loading