Preprint
Jul 2026
Laguerre Geometry for Interpreting Large Language Models
Geometric Lens, a training-free, hyperparameter-free method for reading out the exact concept a hidden vector encodes at any layer, and Laguerre Autoencoder, a 2D visualizer that renders both the decision geometry and a model's full reasoning trajectory in one view are developed.
Chunwei Ma, Russ Wolfinger
· 1 citation