Cloud computing data compression for allreduce in deep learning

In deep learning, and in particular, for data compression for allreduce in deep learning, a gradient may be compressed for synchronization in a data parallel deep neural network training for allreduce by sharing a consensus vector between each node in a plurality of nodes to ensure identical indexin...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Finkler, Ulrich, Zhang, Wei, Cho, Minsik
Format: Patent
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!