Section 7.2.5, “Scope and provenance of the reported result”Scope and provenance: one configuration, seed 0, run 20260214_120801_cd18fc9f
Empirical result Supported by saved experimental runs, not by a proof.
1 Paper text
Scope and provenance of the reported result. This controlled supervised run shows a large advantage for transition information at , , and seed 0. It does not prove worst-case mastery, efficient self-play learning, or the asymptotic circuit lower bound. The checkpoint CSV is identified by run suffix 20260214_120801_cd18fc9f; the architecture and evaluation convention above follow its associated saved notebook. We report this documented configuration and do not make an unsupported claim of robustness across a hyperparameter or board-size sweep. Multiple independent seeds, matched-parameter controls and out-of-distribution evaluations remain useful follow-up experiments.
Generated from paper_jmlr.tex, lines 1133–1135.
2 Explain this
The scope of the restoration result: one configuration (, , examples, seed 0), identified by its run suffix and saved notebook. No robustness across hyperparameters, board sizes or seeds is claimed; independent seeds, matched-parameter controls and out-of-distribution tests are listed as useful follow-ups.
3 How this is formalised
How this is covered. A provenance statement. The run records are retained with the local review materials and are not part of this edition.
| Paper anchor | par-scope-and-provenance-of-the-reported-result-p1 (website anchor; the manuscript has no label here) |
|---|---|
| Displayed as | Section 7.2.5, “Scope and provenance of the reported result” |
| Claim | Scope and provenance: one configuration, seed 0, run 20260214_120801_cd18fc9f |
| Coverage category | Empirical result |
| Statement source | paper_jmlr.tex lines 1133–1135; SHA-256 of the normalised block source 2d632590847724c5… |
| Explanations | explain/c-exp-provenance, formal/c-exp-provenance |
| Build evidence | fresh isolated rebuild, audit report |
4 Lean statement and source
No Lean statement corresponds to this claim. This is an empirical statement about saved experimental runs. Lean proofs do not certify training runs, curves or checkpoints.
Build evidence: fresh isolated rebuild of all 38 bundled modules with Lean (version 4.26.0, started 2026-10-10T15:21:56Z, finished 2026-10-10T15:27:00Z: 38/38 modules compiled, 1588 axiom reports, no errors. See Build for the commands and the independent audit of 2026-10-10 (dated evidence).