UNI Universal Natural Intelligence

Wiki · Hierarchical Active Inference

Corrected-Cell Result Index

Hierarchical Active Inference · hierarchical-aif/reports/CORRECTED-CELL-RESULT-INDEX.md @ b909801f3db4 (hierarchical-aif/motor-stack) — opens the published snapshot 8b4b5935bcba

How to read this page

A Plain and a Clear version of this page have not been written yet. What follows is the document itself.

Precise — the source document

This is the document. Rendered from the repository at the commit above, with nothing rewritten for the web. A gate re-renders it on every deploy and fails the build if a single byte differs.

Date: 2026-07-22 · Gate: H-AIF-G5 (all four landed) · One row per corrected B4 cell. Every hash below is verified by sha256sum -c against its sidecar. No P-level is raised by any cell.

Index

field B4C01 B4C02 B4C10 B4C11
cell B4C01_SYNTHETIC_PARAMETER_RECOVERY B4C02_MISSPECIFIED_WORLDS B4C10_M4_STRUCTURAL_IDENTIFIABILITY B4C11_M7_STRUCTURAL_IDENTIFIABILITY
planned_N 200/gen × 5 = 1000 sims 200/gen × 3 = 600 sims N_boot 2000 N_boot 2000
actual_N 1000 sims (0 failed) 600 sims (0 failed) 2000 boot (1994 completed / 6 failed) 2000 boot (2000 completed / 0 failed)
status ELIGIBLE_FOR_FROZEN_VERDICT ELIGIBLE_FOR_FROZEN_VERDICT complete, resourceBoundPartial:false ELIGIBLE_FOR_FROZEN_VERDICT
result_hash 8256cb120376f46bf00b537d2ad665f9d50335e3eed3727a1299129195dfb65b 0633988dbfd690c0c0d12075dba4e0d8c25ddd178125064bd01fbdaf4629e398 959a00e974641eca1c0d6f3c2f7322b8c6c8411f68ca953c7e169492c4a53dde 564a5b0f4461c83fe2c8792a2304dc136799b0923d887a27a60572361a851b53
report_path reports/B4C01-CORRECTED-FULL-REPORT.md reports/B4C02-CORRECTED-FULL-REPORT.md reports/B4C10-CORRECTED-FULL-REPORT.md reports/B4C11-CORRECTED-FULL-REPORT.md
prediction_commit 28ce738 b9b5670 b9b5670 897c8ab
prospectivity_status SATISFIED — committed with the cell never having run at any N (cleanest in the batch) SATISFIED — committed 3 h 17 min before the observation (mid-run-commit caveat recorded) NOT_SATISFIED — prediction and result share commit b9b5670; strict ancestry structurally unattainable (D9) SATISFIED — committed ~19.4 h before the observation
old_status NOT_RUN (RESOURCE_BOUND, claimed 250–400 h) NOT_RUN (RESOURCE_BOUND, claimed 150–250 h) RESOURCE_BOUND_PARTIAL (100/2000, 5%) submitted U4_OK on 30/2000 under a defective bootstrap — WITHDRAWN
corrected_status NOT_ESTABLISHED — failing criterion M0_EXPONENTIAL:self_win (0.290); all four parameterised generators withinTolerance GENERATOR-SPECIFIC — gensWithM2overM3 = 1 of 3 IDENTIFIED_ON_THIS_COHORT — U2/U3/U4 all OK at full N U4_OK — collapseFraction 0.0055 < 0.25; group counts 80/80
prediction_outcome frozen PASS REFUTED; my committed directional prediction hit the primary structural claim (6 of 9 sub-items) frozen GENERATOR-ROBUST_ADVERSE REFUTED frozen UNIDENTIFIED_OR_WEAK REFUTED (within C10 scope) frozen PROFILE_FLAT_OR_WEAK REFUTED; my committed §7.3 directional prediction HIT all 5 sub-commitments
defect_impacted D3 CLOSED (deterministic corrected run); D4 superseded (reason text); D2 measured 16.83 h D3 CLOSED; D2 measured 8.17 h D2 measured 2.3 h; D1/D3 do not apply (M4 pooled train_y, arithmetic seeds) D1 CLOSED_BY_CORRECTED_RERUN; D2 measured 21.12 h
P-level mapping P0/P1 integrity; constrains interpretation of P3. Cannot move P6. Synthetic data — none OBSERVED constrains interpretation of the adverse P3 result (shape-artifact weakened). No mechanism P1/P6 scoped: M4 identifiability on one cohort. Identifiability ≠ correctness ≠ mechanism P6 for C11 U4 re-established on the corrected run only. No other scope; no level raised
allowed wording "sanity floor not met", "power-limited / nesting-blind at this holdout size", "not established" "generator-specific", "the adverse result is not generic heavy-tailed shape", "not established" "supports M4 identifiability on this cohort under the frozen U2/U3/U4 criteria" "U4 re-established on the corrected run only", "the withdrawn U4_OK remains withdrawn"
forbidden wording any claim this tests the motor-stack AIF model or the mark process; "M3 is vindicated"; "M2 is the UNI model" "mechanism demonstrated"; "M3 is vindicated"; any transfer of a synthetic result to the real cohort as observation "M4 is the correct model"; "mixture mechanism confirmed"; transfer to M7/C11/parity any diagnostic-proves-U4 phrasing; transfer of B4C10's M4 result to C11; any mechanism/parity claim

Cross-cutting facts

  • Every dataset in B4C01 and B4C02 is SYNTHETIC. Nothing in those two cells may be labelled OBSERVED. B4C10 and B4C11 are identifiability checks on frozen fits, not new observations either.
  • Three of four cells are genuinely prospective by commit graph (C01, C02, C11). B4C10 cannot be — prediction and result entered in one commit — and is pinned NOT_SATISFIED by D9, not relabelled.
  • No cell raised a P-level. P8 remains FULL_PARITY = false; first unsatisfied level P4.

NEXT_ACT = update the defect closure ledger to its final freeze state, then write PHASE-HIERARCHICAL-AIF-CORRECTED-CLOSURE.md

sha256 f6af4731606d54d3 — of the original file, so what was ingested stays checkable.