The Long Delay to Arithmetic Generalization: When Learned Representations Outrun Behavior
Laura Gomezjurado · 2026
Read The Long Delay to Arithmetic Generalization: When Learned Representations Outrun Behavior by Laura Gomezjurado (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-13082.txt.
Description
Gomezjurado studies grokking, the long delay before a transformer trained on arithmetic suddenly generalizes, in encoder-decoder models doing one-step Collatz prediction. She finds the encoder organizes parity and residue structure early, while output accuracy stays near chance far longer, pointing to a decoder bottleneck. Transplanting a trained encoder speeds grokking 2.75 times, freezing a converged encoder removes the plateau, and numeral base shapes learnability, with base 24 reaching 99.8 percent while binary fails.
Catalog record
| CREATOR | Laura Gomezjurado |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 54614 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.13082