Certify the Build M US Populace release as the certified default, replacing Build J:
- New:
populace-us-2024-buildm-sparse-rmloss100-98bf731-20260717T075101Z (published 2026-07-17, immutable tag on policyengine/populace-us; latest.json deliberately untouched)
- Outgoing:
populace-us-2024-buildj-sparse-rmloss100-75d5add-20260710T094201Z
- Model:
policyengine-us==1.764.6 (unchanged; compatibility_basis: built_with_model_package)
Build M sparse vs Build J on the release scorecard: final loss 0.02666 vs 0.03084, within-10% 90.21% vs 89.00%, 5,667 vs 5,510 compiled targets, all 14 release gates passed (export-mass parity 0/30 after the reviewed exclusions in populace#433/#442; SSI swap delta 464,107 within cap 654,017; reform-coverage smoke fully green including the keogh probe at +$118.1M under the populace#446 selection-mass protection).
Runbook: docs/engineering/runbooks/build-m-us-populace-certification.md. The merge of the certification PR is the default flip and stays with the maintainer.
Certify the Build M US Populace release as the certified default, replacing Build J:
populace-us-2024-buildm-sparse-rmloss100-98bf731-20260717T075101Z(published 2026-07-17, immutable tag onpolicyengine/populace-us;latest.jsondeliberately untouched)populace-us-2024-buildj-sparse-rmloss100-75d5add-20260710T094201Zpolicyengine-us==1.764.6(unchanged;compatibility_basis: built_with_model_package)Build M sparse vs Build J on the release scorecard: final loss 0.02666 vs 0.03084, within-10% 90.21% vs 89.00%, 5,667 vs 5,510 compiled targets, all 14 release gates passed (export-mass parity 0/30 after the reviewed exclusions in populace#433/#442; SSI swap delta 464,107 within cap 654,017; reform-coverage smoke fully green including the keogh probe at +$118.1M under the populace#446 selection-mass protection).
Runbook:
docs/engineering/runbooks/build-m-us-populace-certification.md. The merge of the certification PR is the default flip and stays with the maintainer.