Information-Theoretic Perspectives on Optimizers

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Tan, Zhiquan, Huang, Weiran
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915176551284736
author Tan, Zhiquan
Huang, Weiran
author_facet Tan, Zhiquan
Huang, Weiran
contents The interplay of optimizers and architectures in neural networks is complicated and hard to understand why some optimizers work better on some specific architectures. In this paper, we find that the traditionally used sharpness metric does not fully explain the intricate interplay and introduces information-theoretic metrics called entropy gap to better help analyze. It is found that both sharpness and entropy gap affect the performance, including the optimization dynamic and generalization. We further use information-theoretic tools to understand a recently proposed optimizer called Lion and find ways to improve it.
format Preprint
id arxiv_https___arxiv_org_abs_2502_20763
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Information-Theoretic Perspectives on Optimizers
Tan, Zhiquan
Huang, Weiran
Machine Learning
The interplay of optimizers and architectures in neural networks is complicated and hard to understand why some optimizers work better on some specific architectures. In this paper, we find that the traditionally used sharpness metric does not fully explain the intricate interplay and introduces information-theoretic metrics called entropy gap to better help analyze. It is found that both sharpness and entropy gap affect the performance, including the optimization dynamic and generalization. We further use information-theoretic tools to understand a recently proposed optimizer called Lion and find ways to improve it.
title Information-Theoretic Perspectives on Optimizers
topic Machine Learning
url https://arxiv.org/abs/2502.20763