proved yet: nolocal download§5 Fail + Incomplete
E1 §5 success-criteria scorecard
Official gate-by-gate scorecard from measured aggregates. Do not flip support_claim.
Public downloads
These are measured files from the lab — scorecards, trimmed summaries, figures, and short literature briefs. Finished runs are not the same as proved.
Measured public mirrors — scorecards, trimmed summaries, figures, lit briefs. Completing a grid ≠ clearing pre-registered gates.
site/artifacts/.
No public GitHub artifacts repo was found for aigrowthlabsus-debug — we do not invent GitHub URLs. Full multi-MB E1 JSON is not mirrored; use the trimmed summary.
Official gate-by-gate scorecard from measured aggregates. Do not flip support_claim.
Measured fields only: n_cells, partial, accept/fail tallies, matched_wall keys, finished_utc. Full multi-MB JSON not mirrored.
Focused patch (not full 1920 redo). Claim stays false until every §5 gate clears.
Recovery macro and accept-by-law plots from the finished 1920-cell grid.
U>R>D measured; U lift survives chemsys ban. Fingerprint cannot alone flip success.
Exploratory calibration probe on fingerprint secondary track.
Diversity arm, leakage, pilot, ablation, light hit-rate (measured runs).
Diversity arm, structure pilot, leakage diag, feature ablation (measured secondary track).
Cross-track 48h, InsightSR vs PhySO, Prop-01/02/03 watches, resume + rolling watchlist.
