
This workshop is designed as an intensive 6-hour Pytorch bootcamp. We will start with small programming exercises designed to quickly teach you key programming skills specific to Pytorch. We will then build up to larger tasks, such as building a model from scratch, designing your own training loop, and building dynamic models.
We will focus on the skills that are most important to scientists working in fields outside machine learning, such as data management, model design and experimentation. For the assignments we will use Pytorch, but the main ideas will translate also to other frameworks like Tensorflow, Keras or Jax/Haiku.
The workshop is in English. A free lunch is provided.
Who is this workshop for
This workshop is for researchers, research support staff and master students of the University of Amsterdam. You understand the basics and have some Pytorch programming experience, and want to use Deep Learning methods for your research or production work.
What you will learn
- Intermediary to advanced Pytorch application using Python
- Programming exercises that will help you elevate your programming skills
- How to build a model from scratch, design your own training loop and build dynamic models
- Data management, model design and experimentation
Pre-requisites for attendees
- A basic understanding of Python and Pytorch is expected. Participants who do not know Python or Pytorch can register, but may not fully benefit from the excercises
- You have a Google account
- All material is provided in Google Colab worksheets, so no software needs to be installed beforehand
- Optional: you can find materials of the previous Deep Learning (advanced) workshop on the DSC-GitHub here
Time and location
Location: Bétalounge, B 1.25 (max. 20 participants) Science Park 904, Amsterdam.
Time: 10:00-16:00
Workshop Instructor
Peter Bloem is assistant professor at the Vrije Universiteit Amsterdam. He is specialised in the combination of Machine Learning and Knowledge Representation.
Registration
Registration is free of cost, with prioritization to members of the Data Science Centre and UvA-researchers.
If you have questions, you can contact us at dsc@uva.nl with the subject ‘Deep Learning Bootcamp, 21st of April’.