When using the cloud version of Deep Learning DS, there is no need for any local installation. Otherwise, you can install any component of Deep Learning DS on-premises.
Deep Learning DS comprises three parts:
Deep Learning DS Server (dlds-server): Server application for datasets, annotations, model information, training, and evaluation results.
Deep Learning DS Training Worker (dlds-worker): Worker performing the training and evaluation of deep neural networks.
Deep Learning DS Client (dlds-client): Library and CLI for interacting with the APIs of Deep Learning DS.
All parts can be installed on-premises. Furthermore, there is no need to install all parts on one machine. The components are designed to work individually. You can even combine the cloud version of Deep Learning DS with a training worker on-premises. For instance, you can host your own training worker on a local machine or using a cloud provider.
For a complete on-premises setup, follow the installation instructions of the Deep Learning DS Server and Deep Learning DS Training Worker. The Deep Learning DS Client is included within the package of the Training Worker.