| Age | Commit message (Collapse) | Author |
|---|
|
Panel 1a: rename to Backward rules; DFA drawn as direct side-to-side
arrows from the error to each hidden layer. Panel 1b: replace the dual
y-axis with stacked guarantee/cost sub-axes. Figure 2: reader-facing
panel titles; legend renames (linear extrapolation, relinearization).
Track the regenerated figure-source CSVs (verified bit-exact against
the frozen paper numbers) so figures can be rebuilt without rerunning
experiments.
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phases with superseded markers); scripts/README gains a status table.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>
|
|
The sampler previously drew e0 through the analytic linear form <C,B>,
making the Gaussian KS test near-tautological. It now runs the real FA
backward pass per feedback draw and reports the max deviation from the
linear identity (<= 8.9e-16). Rerun reproduces the published table.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
|
Closed-form gap law note (34) + consolidated snapshot (35), with the
comparison script reading the dense run's config/gaps from its output
dir and evaluating the closed form on matched data and init seed plus a
multi-init band. Corrected headline: corr(log,log) 0.933 matched /
0.982 five-init geometric mean; lam_min ensemble corr 0.987.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
