| Age | Commit message (Collapse) | Author |
|---|
|
Note 42 is a self-contained write-from document: full statements and
proofs for Thm 1-4 / Prop 1-3 / Cor 1-2, every result with exact
numbers and configs cross-checked against the CSVs, the negative
results, and the honesty boundaries.
Audit finding while building it: the "E_B[e0]=0.4378 +/- 0.008 vs
0.4403" headline (cited in notes 29/35/36/40/41 and the old draft) is
not reproducible from the current moments CSV. Retired in favor of the
reproducible 6x6 table (max |error| <= 0.008, predicted span 0.15-0.45).
Living docs updated; note 37 sec 4 records the retirement.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
|
Root README orients the repo; notes/README reindexed by status
(writing pipeline / authoritative note per result / six historical
phases with superseded markers); scripts/README gains a status table.
Note 40 = reproduction manifest (claim -> command -> outputs -> note;
everything unlisted is not citable). Note 41 = the binding paper plan:
AAAI-27 constraints (7pp + refs, appendix not guaranteed read),
narrative, section/page budget, frozen theorem and figure numbering,
related-work list with per-cite rationale, writing rules, weekly
schedule to the 07-20/07-27 deadlines, and risk register.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
|
SGD teacher sweep (widths 8-96, 24 FA runs/width): train gap soft and
monotone (0.249 -> 0.0069) while the test gap changes sign — FA
generalizes better than BP at intermediate width (-0.120 +/- 0.013 at
w=32) and settles slightly positive once both fit. Scopes the paper's
optimization-gap claims and preempts the generalization question.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
|
New script validates both no-fit claims on MNIST-subset MLPs. e0
theorem: 16 configs, hidden share 0.39-0.82, measured E_B[e0] within
0.0001-0.0089 of prediction (512 real-backward draws each). Estimator:
96 trajectories, velocity MAE 0.00196 (3.9% of mean gap), bias -0.0006,
beating fixed K(0) everywhere; kernels computed via a layerwise gram
identity self-tested against autograd at machine precision.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
|
Note 36 consolidates the full evidence chain (numbers, figure sources,
negative results, claim discipline) as the paper-writing base; note 37
records the two audit fixes; superseded banners on notes 00/03; README
index updated. Superseded alignment_recovery_dynamics probe removed.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
