SEP 10, 2026 · PREPRINT
LILA: Calibration-Free Structured Pruning of Large Language Models via Latent Spectral Geometry
arXiv
This is an unrefereed arXiv preprint describing a novel computational method for neural network compression; it reports empirical results and theoretical analysis but has not undergone peer review.
Reported
Zero-shot accuracy improvement vs…1.57 pp
Zero-shot accuracy improvement vs…6.0 pp
Post-LoRA accuracy margin vs Slic…0.48 pp