capitolmuckrakr / aistudio-workshops

Workshops created by the Quartz AI Studio

AI Studio Workshops

This code supports workshops taught by members of the Quartz AI Studio for getting journalists learning machine learning.

Resources

• Quartz AI Studio: https://qz.ai
• fast.ai: https://fast.ai
• practical deep learning: https://course.fast.ai/

A note about notebooks

These workshops exist in Jupyter Notebooks (formerly IPython notebooks, which is why notebook files end in .ipynb). Because we'll be doing machine learning, we also need a GPU, which is a fast parallel-processor that speeds up the math we use for training models.

At the time of this writing, the Google Colaboratory platform provides both a notebook enviroment and a GPU for free. So these notebooks are designed to work particularly well with Google Colab.

If you know how to spin up another platform – such as Amazon's EC2 – the notebooks should work there, too.

Getting started, with Google Colab

Here's how to get started with these workshops using Google Colaboatory.

• Go to Google Colaboratory.
• In the top bar of the welcome window, pick "Github."
• Enter quartz on the long blue line and press Return.

• From the list that appears, make sure aistudio-workshops is the selected repository and then click on the notebook for the lesson you'd like.

• For many (though not all) of these lessons, we'll want to turn on the GPU. From the "Runtime" menu, pick "Change runtime type."

• Then form the "Hardware accellerator" dropdown, pick "GPU."

• You want to run the first code cell in the notebook, by tapping the "play" button on the cell that includes the code ## ALL GOOGLE COLAB USERS RUN THIS CELL

• You may get one or two warnings, which you can safely dismiss:

You're all set!