eth-cscs / pytorch-training

PyTorch training at CSCS

Hands-on Introduction to Deep Learning with PyTorch

Course Description

This course is intended as a quick introduction of fundamental concepts of deep learning, covering neural network basics, training methods, as well as a few examples of specific applications such as convolutional neural networks for computer vision and the transformer model for natural language processing.

More specifically, the following topics will be covered:

• Fundamentals of neural networks
• Training deep learning models: stochastic gradient decent, optimizers, loss functions, regularization, etc.
• Convolutional Neural Networks (CNNs) for computer vision: basics of CNNs, image classification and generation
• Natural Language Processing (NLP) with transformers: basics of NLP, the transformer model, attention mechanism, etc.

The lessons will blend theory with hands-on practice, using PyTorch for practical exercises. We will run these sessions on the Piz Daint supercomputer at CSCS.