ACE extracts invariant algorithmic cores shared across independently trained transformers

burny_tech · x · 2026-11-20

Joshua S. Schiffman's paper introduces Algorithmic Core Extraction (ACE): identifying compact subspaces that are necessary and sufficient for a task and recur across independently trained models. Key findings:

The upshot: training selects behavior, not circuitry, and ACE shifts study from what transformers happen to do to what they must do. Nearly everything runs on a laptop.

Original post →

More from Models

Models channel →