viz: floor->ceiling as two normalized panels (best vs control vs reference)

Browse Source

Rework per feedback: hack and solve are not opposites, so they get separate
floor->ceiling axes (each 0=floor..1=ceiling) rather than sharing a zero -- this
also stops solve (range ~0.13-0.22) being squished next to hack (0-0.61).
Minimal: routeV per-token (best) vs random-V (direction control) vs the SGTM
gradient-routing paper placed on the same floor->ceiling % axis (approx, LM task).

Reads: hack suppression 93% best / 84% control / ~98% reference (9pp = direction
signal); solve gained +17% / -17% / ~95% (far from ceiling -- model barely learns
to solve in 60 steps). Moved out/plots -> out/figs.

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>

...

This commit is contained in:

wassname

2026-06-09 09:55:03 +00:00

co-authored by Claudypoo

parent 7d08ad2acd

commit 34a2eec704

5 changed files with 49 additions and 53 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

out/plots/floor_ceiling.csv → out/figs/floor_ceiling.csv

out/plots/floor_ceiling.pdf → out/figs/floor_ceiling.pdf

BIN

Binary file not shown.

out/figs/floor_ceiling.png

BIN

Binary file not shown.

Side by Side After Width:  |  Height:  |  Size: 84 KiB

out/plots/floor_ceiling.png

BIN

Binary file not shown.

Side by Side Before Width:  |  Height:  |  Size: 154 KiB