Characterization and Mitigation of Training Instabilities in Microscaling Formats

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Su, Huangyuan, Kwun, Mujin, Gil, Stephanie, Kakade, Sham, Anand, Nikhil
Format:	Preprint
Published:	2025
Subjects:	Machine Learning Hardware Architecture
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Internet

https://arxiv.org/abs/2506.20752