In this course, we provide an overview of end-to-end deep learning with the latest version of Tensorflow/Keras. It covers the basic concepts to define models with Keras and data pipelines with Tensorflow’s „Dataset“, and to visualize the results with Tensorboard while training. If training on one node or GPU is insufficient, we show how to scale up/out distributed training onto multiple compute nodes and GPUs with Horovod. Furthermore, we provide an introduction to scikit-learn, with an overview of different machine learning algorithms it provides and how to utilize it on GPUs with H2O4GPU. Furthermore, we provide an introduction to scikit-learn, with an overview of different machine learning algorithms it provides and how to utilize it on GPUs with H2O4GPU and NVIDIA cuML. The training courses consist of hands-on exercises to be run directly on IT4I's infrastructure. 

A short start-up guide to get started with scikit-learn, Tensorflow and Pytorch frameworks for IT4Innovations and LUMI systems conclude the day.

[exercises, however, will only be done on IT4I infrastructure]

This training will be held in a hybrid mode.