Skip to content

Fill release readiness from score smoke artifacts#37

Merged
MaxGhenis merged 1 commit into
mainfrom
codex/dashboard-release-readiness-from-score-smoke
May 28, 2026
Merged

Fill release readiness from score smoke artifacts#37
MaxGhenis merged 1 commit into
mainfrom
codex/dashboard-release-readiness-from-score-smoke

Conversation

@MaxGhenis
Copy link
Copy Markdown
Contributor

@MaxGhenis MaxGhenis commented May 28, 2026

Summary

  • synthesize tier-level release-readiness rows from scores.json artifacts that already have runtime_smoke_loader.json
  • keep full mp300k_artifact_gates.json reports authoritative when present
  • treat score+smoke-only evidence as incomplete until a full gate report exists, so it can surface candidates without certifying release readiness
  • derive candidate_beats_baseline from losses when older score payloads omit the boolean

Checks

  • uv run --python 3.13 --extra dev ruff check src/microplex_us/pipelines/dashboard.py tests/pipelines/test_dashboard.py
  • uv run --python 3.13 --extra dev pytest -q tests/pipelines/test_dashboard.py

Dashboard verification

With the shared artifact root, /Users/maxghenis/CosilicoAI/microplex-us/artifacts/microplex_dashboard_current.json reports:

  • mp-120k: release_ready, loss 0.09356981757194573 vs eCPS 0.16636673113001124
  • mp-152k: blocked by artifact_size and runtime
  • mp-300k: blocked on the legacy/patched surface by size/runtime/benchmark manifest gates

@MaxGhenis MaxGhenis force-pushed the codex/dashboard-release-readiness-from-score-smoke branch from d751da9 to f503f78 Compare May 28, 2026 11:09
@MaxGhenis MaxGhenis merged commit 7a961fb into main May 28, 2026
3 checks passed
@MaxGhenis MaxGhenis deleted the codex/dashboard-release-readiness-from-score-smoke branch May 28, 2026 11:11
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.

1 participant