Merak: An Efficient Distributed DNN Training Framework With Automated 3D Parallelism for Giant Foundation Models

Author:

Lai Zhiquan1ORCID,Li Shengwei1ORCID,Tang Xudong1,Ge Keshi1ORCID,Liu Weijie1ORCID,Duan Yabo1,Qiao Linbo1ORCID,Li Dongsheng1ORCID

Affiliation:

1. National Laboratory for Parallel and Distributed Processing, College of Computer, National University of Defense Technology, Changsha, Hunan, China

Funder

National Key R&D Program of China

National Natural Science Foundation of China

Publisher

Institute of Electrical and Electronics Engineers (IEEE)

Subject

Computational Theory and Mathematics,Hardware and Architecture,Signal Processing

Reference52 articles.

1. PyTorch: An imperative style, high-performance deep learning library;paszke;Proc Adv Neural Inf Process Syst,2019

2. End-to-end adaptive distributed training on paddlepaddle;ao,2021

3. DeepSpeed

4. Transformers: State-of-the-Art Natural Language Processing

5. Openwebtext corpus;gokaslan,2019

Cited by 22 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. DELTA: Memory-Efficient Training via Dynamic Fine-Grained Recomputation and Swapping;ACM Transactions on Architecture and Code Optimization;2024-08-20

2. Proactive Caching With Distributed Deep Reinforcement Learning in 6G Cloud-Edge Collaboration Computing;IEEE Transactions on Parallel and Distributed Systems;2024-08

3. Distributed Foundation Models for Multi-Modal Learning in 6G Wireless Networks;IEEE Wireless Communications;2024-06

4. Distributed Training Optimization for DCU;2024 5th International Conference on Information Science, Parallel and Distributed Systems (ISPDS);2024-05-31

5. Construction of Raining System Based on 3D Virtual Simulation;2024 Second International Conference on Data Science and Information System (ICDSIS);2024-05-17

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3