Introducing Dynamic Training


Dynamic Training is an open-source deep learning project that allows you to add and remove EC2 Spot or reserved instances elastically in the middle of training with no loss in accuracy, helping you increase compute utilization and reduce training cost. Dynamic Training is currently available for Apache MXNet on AWS, with more frameworks coming soon. In this session, learn more about how Dynamic Training works and how to get started with it. Follow Along: https://github.com/awslabs/dynamic-training-with-apache-mxnet-on-aws

Catch up on the excitement of re:Invent 2018 with the AWS launchpad featuring launch announcements, demos of newly launched technology, interviews with expert guests and live Q&A. AWS re:Invent is a tech education conference for the global cloud computing community hosted by Amazon Web Services. See all recordings of the AWS Launchpad at re:Invent here: https://www.youtube.com/playlist?list=PLhr1KZpdzukc0WXQruGVXTiNPtct-LLaa and learn more about AWS live streaming here: https://aws.amazon.com/twitch.


Duration: 20:24
Publisher: Amazon Web Services
You can watch this video also at the source.