Skip to content

site: finish the 2.0 correction — negatives note and blog table #163

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ronaldtse merged 1 commit into from
Sep 7, 2026
Merged

site: finish the 2.0 correction — negatives note and blog table #163

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 3 additions & 2 deletions src/content/blog/2026-09-05-frontier-closed.adoc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@ bootstrap interval on the gap to the teacher:
|from-scratch 30M, every lever |30M |73.95 |[70.22, 71.09]
|1.0 rung |300M |8.26 |[4.6, 5.7]
|lite rung (enc 12→6) |190M |5.78 |[3.03, 3.49]
|2.0 rung |300M |4.82 |[2.36, 2.82]
|2.0 rung |300M |5.08 (corrected 2026-09-05) |[2.36, 2.82]
|2.1 rung |300M |4.57 |[1.91, 2.35]
|teacher r7 |580M |2.29 |—
|===
Expand All @@ -52,7 +52,8 @@ per-step confidence.
The residual gap between the 2.1 student and its teacher had one live
attribution left: classical-domain coverage. Both tests of it failed.
Swapping news-domain training units for classical Tashkeela at constant
budget made the student worse (−0.98pp). Adding five times the
budget made the student worse (+0.72pp over the corrected 2.0
control). Adding five times the
classical corpus — the maximal version of the lever — left it
statistically flat (4.82 vs 4.57, intervals overlapping). The residual
is not a coverage deficit the corpus can reach; it lives in the
Expand Down
16 changes: 11 additions & 5 deletions src/pages/ml.astro
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -338,11 +338,17 @@ const clientModels = [
</ul>
<p class="lb-note">
What didn't move it — every rung pre-registered, measured, and kept in the log:
multi-token-prediction auxiliary 5.09 (+0.26pp); classical-register swap at constant
budget 5.81 (+0.98pp); register add at matched epochs 4.82 (flat — it cancelled the epoch
gain); on-policy GKD 6.00 (+1.18pp, the worst rung); depth-halved 5.78 (63% of the
parameters — shipped anyway as the browser tier). At SFT convergence, supervision quality
dominates. The ladder is closed.
multi-token-prediction auxiliary 5.09 (+0.26pp as measured vs the
then-published control — level with the corrected 5.08); classical-register swap at
constant budget 5.81 (+0.98pp as measured; +0.72pp over the corrected control); register
add at matched epochs 4.82 (statistically flat against news-only 4.57, intervals
overlapping — it sits between the corrected control and the news-only rung); on-policy
GKD 6.00 (+1.18pp as measured; +1.43pp over the rung it was meant to improve — the worst
measured); depth-halved 5.78 (63% of the parameters — shipped anyway as the browser
tier). At SFT convergence, supervision quality dominates. The ladder is closed. The
control's originally published 4.8218 did not reproduce and was withdrawn on
2026-09-05; two independent re-measurements, one at the artifact level on the exact
released bytes, agree on 5.08.
</p>
<p class="lb-note">
Hebrew, same discipline: 16.43% DER on Biblical Hebrew, where the modern-Hebrew state of
Expand Down