I agree that this site is using cookies. You can find further informations
here
.
X
Login
Merkliste (
0
)
Home
About us
Home About us
Our history
Profile
Press & public relations
Friends
The library in figures
Exhibitions
Projects
Training, internships, careers
Films
Services & Information
Home Services & Information
Lending and interlibrary loans
Returns and renewals
Training and library tours
My Account
Library cards
New to the library?
Download Information
Opening hours
Learning spaces
PC, WLAN, copy, scan and print
Catalogs and collections
Home Catalogs and Collections
Rare books and manuscripts
Digital collections
Subject Areas
Our sites
Home Our sites
Central Library
Law Library (Juridicum)
BB Business and Economics (BB11)
BB Physics and Electrical Engineering
TB Engineering and Social Sciences
TB Economics and Nautical Sciences
TB Music
TB Art & Design
TB Bremerhaven
Contact the library
Home Contact the library
Staff Directory
Open access & publishing
Home Open access & publishing
Reference management: Citavi & RefWorks
Publishing documents
Open Access in Bremen
zur Desktop-Version
Toggle navigation
Merkliste
1 Ergebnisse
1
Hippie: A Data-Paralleled Pipeline Approach to Improve Memo..:
, In:
50th International Conference on Parallel Processing
,
Ye, Xiangyu
;
Lai, Zhiquan
;
Li, Shengwei
... - p. 1-10 , 2021
Link:
https://dl.acm.org/doi/10.1145/3472456.3472497
RT T1 <cTitle>50th International Conference on Parallel Processing</cTitle>: T1 <Title>Hippie: A Data-Paralleled Pipeline Approach to Improve Memory-Efficiency and Scalability for Large DNN Training</Title> UL https://suche.suub.uni-bremen.de/peid=acm-3472497&Exemplar=1&LAN=DE A1 Ye, Xiangyu A1 Lai, Zhiquan A1 Li, Shengwei A1 Cai, Lei A1 Sun, Ding A1 Qiao, Linbo A1 Li, Dongsheng PB ACM YR 2021 K1 deep learning K1 distributed training K1 hybrid parallelism K1 memory efficiency SP 1 OP 10 LK http://dx.doi.org/https://dl.acm.org/doi/10.1145/3472456.3472497 DO https://dl.acm.org/doi/10.1145/3472456.3472497 SF ELIB - SuUB Bremen
Export
RefWorks (nur Desktop-Version!)
Flow
(Zuerst in
Flow
einloggen, dann importieren)
