Scale Determines Whether Language Models Organize Representation Geometry for Prediction

Weilun Xu · 2026

Read Scale Determines Whether Language Models Organize Representation Geometry for Prediction by Weilun Xu (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-2605-17084.txt.

Description

In his 2026 paper, Xu studies how the inner layout of a language model lines up with its word-predicting output layer. He tests a new score, Subspace PGA, across ten models of varied size and family. Mid-layer layouts track prediction closely, but small models lose that link late in training while large ones keep it. A few strong directions drift off course and hide the pattern, and removing them restores it.

Catalog record

CREATORWeilun Xu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE54199
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.17084

Browse the full library · About the Tower of Babel