by Marco Buzzelli, Luigi Celona, Flavio Piccoli, and Simone Zini
Designed for education purposes. Please do not distribute without permission.
This colaboratory (colab) will help you get prepared for the rest of the practical sessions at M2L 2021. Here you will get familiar with the environment and tools used in the rest of the practical sessions.
We strongly encourage you to cover the Colab and JAX tutorials.
After that, if you want to know a bit more about the basics, you can browse the other notebooks (Numpy and Plotting).
This tutorial teaches you about colab and its main features. You will need to know this for the rest of the labs as you will write all code in colab.
Open the file Lab_01_Intro_Colab.ipynb to access the colab.
This colab is designed as an all-encompassing guide to get started with the fundamentals of using Tensorflow/JAX/Haiku in a Google Colab environment.
Open the file Lab_02_Intro_JAX.ipynb to access the colab.
This colab introduces you to numpy, the python package we use for computing. Topics such as
- array creation
- operations on arrays
- indexing and selection on arrays
- broadcasting
are covered.
By the end of this colab you will have written a function to generate datasets for learning the NXOR function.
Open the file Lab_03_Intro_Numpy.ipynb to access the colab.
In this colab we generate the plot included with the definition of the NXOR function. In the process we use some more features of numpy.
Other than the plot above, we also see how to use matplotlib to
- draw line plots so we can visualize training curves later, and
- display images, or galleries of images so we can visualize datasets and the output of learnt models.
Open the file Lab_04_Intro_Plotting.ipynb to access the colab.