Deep Learning in Practice: Environment for Practical Sessions
Three possibilities to run the exercises:
Locally on your computer (see below to install the required libraries),
or on a cloud service:
a public one: provided by CentraleSupelec / DataIA / SaclAI-school (more info in TP1)
a private one: Google Colab, requiring a (possibly temporary) Google account.
Possible mini-guide for installing standard deep learning libraries and running Jupyter for newbies:
install anaconda (this will allow you to install python libraries, locally in your home directory [no root rights needed]).
optionally, create a conda environment
install the required libraries with the following command line: conda install pytorch jupyter matplotlib
and if you also need the computer vision part of PyTorch: conda install -c pytorch pytorchvision
download the exercise instructions
run the Jupyter notebook: just type jupyter notebook in a terminal, which will open a window or tab in your favorite web browser; there, just select the exercise file. Check that you can run the instructions with code (tip: there's a button to stop running it if needed).
get familiar with the Python (3) language and with PyTorch (deep learning platform). Even though we won't need advanced knowledge of them here, it's a good investment.