Training LLMs with Fault Tolerant HSDP on 100,000 GPUs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Salpekar, Omkar, Varma, Rohan, Yu, Kenny, Ivanov, Vladimir, Wang, Yang, Sharif, Ahmed, Si, Min, Xu, Shawn, Tian, Feng, Zheng, Shengbao, Rice, Tristan, Garg, Ankush, Peng, Shangfu, Siravara, Shreyas, Fu, Wenyin, de Castro, Rodrigo, Gangidi, Adithya, Obraztsov, Andrey, Narang, Sharan, Edunov, Sergey, Naumov, Maxim, Tang, Chunqiang, Oldham, Mathew
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!

Similar Items