macloo / aistudio-workshops

Workshops created by the Quartz AI Studio

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

AI Studio Workshops

Welcome

The code here supports workshops taught by members of the Quartz, helping journalists learning machine learning.

The key materials are in the notebooks folder. More on that below.

Videos

Many of the notebooks in the notebooks folder pair up with a set of videos originally recorded for an online class provided by the UT Knight Center for Journalism in the Americas. They are now online for free!

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.

Choose Github

  • 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.

Pick the notebook matching the lesson

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

Pick change runtime type

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

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

Play the first code cell of the notebook

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

Click through the first warning

Click through the second warning

You're all set!

More resources

  • Examples, walk-throughs and other materials are available at the Quartz AI Studio website.
  • Many of the projects here use the great library made by fast.ai
  • Some of the notebooks are also based on what we learned taking fast.ai's great practical deep learning class, which you should consider, too!

About

Workshops created by the Quartz AI Studio

License:MIT License


Languages

Language:Jupyter Notebook 100.0%