Figure 6Saved test accuracy at the ten checkpoints
Empirical result Supported by saved experimental runs, not by a proof.
1 Paper text
Generated from paper_jmlr.tex, lines 1118–1124.
2 Explain this
Test accuracy at the ten saved checkpoints. The checkpoints are sparse, so they confirm the improvement beyond the plateau but do not show its timing; the denser validation curve does.
3 How this is formalised
How this is covered. Empirical. Replotted from the saved checkpoint test-accuracy CSV (20 rows; SHA-256
766f9934272d…) without changes to the values; converted to SVG for this edition without re-plotting.
| Paper anchor | fig:testacc-H20k4 (manuscript label) |
|---|---|
| Displayed as | Figure 6 |
| Claim | Saved test accuracy at the ten checkpoints |
| Coverage category | Empirical result |
| Statement source | paper_jmlr.tex lines 1118–1124; SHA-256 of the normalised block source b7b14ba64b1c6ca5… |
| Explanations | explain/fig:testacc-H20k4, formal/fig:testacc-H20k4 |
| 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).