This is a tutorial presented during a workshop at the TheWebConf 2021.

It consists of 7 interactive Jupyter notebooks.

The links below open the notebooks in Google Colab. This way, you don't have to install anything on your computer and you get a GPU for free.

1. Basics
2. Data
3. Autograd
4. Optimization
5. Modules
6. CNN
7. Transfer Learning

This notebook covers the basic features of Google Colab.

Check out these others tutorials and courses if you are hungry to learn more:

• Official tutorials: pytorch.org/tutorials. There are many!
• Step-by-step tutorial: towardsdatascience.com
• PyTorch for deep learning (.py files): yunjey/pytorch-tutorial
• PyTorch for deep learning (notebooks): yandexdataschool/Practical_DL

Binder is an alterantive way to run remote notebooks that does not require a Google account.

If you want to run the notebooks locally, you can use conda. The following instructions should work on Linux/Mac OS, Windows might require slight adaptations.

If you have not installed it yet, you can download it from Anaconda (Python 3.6 version).

Verify that it is installed by running

conda -V

Make sure your conda installation is up-to-date:

conda update conda

Now clone the repository:

git clone https://github.com/theevann/webconf-pytorch-workshop.git
cd webconf-pytorch-workshop

The available conda environments can be listed using

conda env list

Let's now create a new environment called 'webconf-pytorch'.

conda env create -f environments.yml

After the environment has been created, you can activate it by

source activate webconf-pytorch

Now start the Jupyter notebook by running

jupyter notebook

The environment can similarly deactivated by

source deactivate