19–23 Sept 2022
Prague, Czech Republic
Europe/Amsterdam timezone

Distributed Deep Learning by Horovod - a new EOSC service

22 Sept 2022, 16:55
20m
Opal, Topaz

Opal, Topaz

Speaker

Dr Jozsef Kovacs (SZTAKI)

Description

The "Distributed Deep Learning by Horovod" EOSC service provides the infrastructure, resources and libraries to its users in order to perform effective distributed training of deep neural networks. Access to a ready-to-use Horovod cluster is provided through Jupyterlab. The talk will give a short introduction of the service developed and recently onboarded by the Neanias EU project and supported by the Hungarian ELKH cloud.

Topic EOSC Compute Platform

Primary author

Dr Jozsef Kovacs (SZTAKI)

Co-authors

Mr Attila Farkas (SZTAKI) Mr Krisztian Pora (SZTAKI) Dr Robert Lovas (SZTAKI)

Presentation materials