AI RESEARCH
Categorical Perception in Large Language Model Hidden States: Structural Warping at Digit-Count Boundaries
arXiv CS.AI
•
ArXi:2603.28258v1 Announce Type: cross Categorical perception (CP) -- enhanced discriminability at category boundaries -- is among the most studied phenomena in perceptual psychology. This paper reports that analogous geometric warping occurs in the hidden-state representations of large language models (LLMs) processing Arabic numerals.