Showing cs.LGShow all
2 papers · 1 filter
cs.LG2026
Near-Floor Geometry Is Generic: Leverage Dispersion in Trained Overcomplete Codes
Hector Borobia, Elies Seguí-Mas, Guillermina Tormo-Carbó
An overcomplete code packs features into dimensions, so linear readout of one feature picks up cross-talk from the others, bounded below by the rank-trace floor $W(F,d)=(…
cs.LG2026
How Pruning Reshapes Features: Sparse Autoencoder Analysis of Weight-Pruned Language Models
Hector Borobia, Elies Seguí-Mas, Guillermina Tormo-Carbó
Weight pruning is a standard technique for compressing large language models, yet its effect on learned internal representations remains poorly understood. We present the first sys…