Skip to content

Certify Build M US Populace as the certified default#475

Merged
MaxGhenis merged 1 commit into
mainfrom
certify-us-buildm
Jul 17, 2026
Merged

Certify Build M US Populace as the certified default#475
MaxGhenis merged 1 commit into
mainfrom
certify-us-buildm

Conversation

@MaxGhenis

Copy link
Copy Markdown
Contributor

Fixes #474

Certifies Build M as the US Populace certified default, replacing Build J, per docs/engineering/runbooks/build-m-us-populace-certification.md (the #470 pattern).

  • BUILD_M_RELEASE_ID = populace-us-2024-buildm-sparse-rmloss100-98bf731-20260717T075101Z (immutable tag on policyengine/populace-us; latest.json untouched — merging this PR is the flip decision)
  • MODEL_VERSION = policyengine-us==1.764.6 (unchanged; compatibility_basis: built_with_model_packagepyproject.toml and household snapshots untouched, the Certify US Populace default: Build J 57k (rmloss100) #470 common case)
  • Outgoing: populace-us-2024-buildj-sparse-rmloss100-75d5add-20260710T094201Z

Release quality (Build M sparse vs Build J)

metric Build M (98bf731) Build J (75d5add)
final loss 0.02666 0.03084
within-10% 90.21% 89.00%
targets compiled 5,667 5,510
export-mass parity PASS 0/30 PASS 0/35
gates ALL PASS (14) ALL PASS (14)

Gate battery notes: SSI swap delta 464,107 within the populace#431 cap (654,017), reform-coverage smoke fully green — keogh probe at +$118.1M under the populace#446 selection-mass protection (populace#445 adjudication), farm probes green under the populace#437 "either" sign and the populace#442 parity exclusion (populace#441 adjudication).

Changes (exactly the runbook's file set)

  • src/policyengine/data/bundle/manifest.json — rewritten by scripts/certify_data_release.py from the Build M release manifest (sha 25ee62fb…)
  • src/policyengine/data/bundle/us.trace.tro.jsonld + uk.trace.tro.jsonld — regenerated with --include-tros; the UK sidecar's two sha fields update because it binds the same (changed) bundle manifest — not a degraded regeneration (no limited markers; sidecar-binding tests pass)
  • tests/test_release_manifests.py, tests/test_models.py, tests/test_us_regions.py — the three pinned release-id constants
  • changelog.d/certify-us-populace-us-2024-buildm-sparse-rmloss100-98bf731-20260717T075101Z.changed.md — auto-written by the certify script

Verification

  • scripts/bundle.py check exits 0; local-area overlay (populace_us_2024_acs_local) preserved and asserted
  • make format / make lint clean
  • tests/test_release_manifests.py + tests/test_certify_data_release.py: 70 passed
  • tests/test_models.py + tests/test_us_regions.py: green against the published tag (results in thread)

Draft on purpose: do not merge without the maintainer's explicit flip decision.

🤖 Generated with Claude Code

Fixes #474.

BUILD_M_RELEASE_ID populace-us-2024-buildm-sparse-rmloss100-98bf731-
20260717T075101Z replaces populace-us-2024-buildj-sparse-rmloss100-
75d5add-20260710T094201Z; policyengine-us stays 1.764.6
(built_with_model_package), so pyproject and household snapshots are
untouched - the #470 common case, per the Build M runbook.

Build M sparse vs Build J: 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. The UK TRO sidecar's two sha fields update because it
binds the same (changed) bundle manifest; not a degraded regeneration.

Verified: bundle.py check green; local-area overlay preserved;
manifest+certify suites 70 passed; models+US regions 57 passed against
the published tag; overlay selection 7 passed; format+lint clean.

Merging this PR is the default-flip decision and stays with the
maintainer.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@MaxGhenis
MaxGhenis marked this pull request as ready for review July 17, 2026 11:03
@MaxGhenis
MaxGhenis merged commit 3f18265 into main Jul 17, 2026
24 checks passed
@MaxGhenis
MaxGhenis deleted the certify-us-buildm branch July 17, 2026 14:55
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Certify Build M US Populace (populace-us-2024-buildm-sparse-rmloss100-98bf731) as the certified default

1 participant