Skip to content

Commit 0433136

Browse files
author
ClassicLarry
committed
Add isoflop curvature stability criterion to promotion gates
1 parent c333125 commit 0433136

1 file changed

Lines changed: 2 additions & 0 deletions

File tree

experiments/grug/moe/README.md

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -104,6 +104,8 @@ Changes can be promoted to this recipe when they demonstrate:
104104
2. **Lower projected c4_en BPB at 1e21 and 1e23 FLOPs**, using the scaling-law
105105
fit above (L∞ pinned at 1.6 for Paloma macro). Re-fit the power law on the
106106
candidate's ladder and compare projections head-to-head.
107+
3. **Low curvature around the minimum of each isoflop curve** — stable
108+
behavior across under- and over-trained regimes.
107109

108110
Most promotable changes will land in one of three files:
109111

0 commit comments

Comments
 (0)