Skip to content

Home

Jump to bottom
Matt Callanan edited this page Mar 15, 2019 · 4 revisions

The Fast.ai Brisbane 2019 Study Group.

Schedule

This page documents how to join the Fast.ai Study Group running out of the Brisbane Expedia office starting March 2019. Feel free to fork this repo to create your own study group.

What is Fast.ai Study Group?

The group is a series of sessions (both in-person and video-conferenced over BlueJeans) run weekly.  It follows the excellent Practical Deep Learning for Coders which has been taken by thousands of people worldwide.  The fast.ai course is refreshed each year. We are currently following the 2019 version (recorded in October 2018) - it exposes group members to cutting edge research and practical applications.

The purpose of the group is to provide motivation, camaraderie, and assistance to anyone keen to learn.

People joining the group have many reasons, including:

  • I want to understand what Deep Learning is
  • I want to learn how to think differently about how to solve problems
  • I have a background in ML and want to refresh my skillset
  • I've tried to do this on my own but I need more motivation to get through it!

Pre-requisites

The course recommends you have had exposure to at least 1 year of coding (course uses Python) and basic high school mathematics. No prior knowledge of Machine Learning or Deep Learning is required.

The course is aimed at deep learning newbies but gets going very quickly. It uses a "top-down" approach to learning - i.e. in Lesson 1 you are building a deep learning neural network to classify images of dogs and cats - something which was very hard to do a few years ago!

From the course itself:

You might be surprised by what you don't need to become a top deep learning practitioner. You need one year of coding experience, a GPU and appropriate software (see below), and that's it. You don't need much data, you don't need university-level math, and you don't need a giant data center. For more on this, see our article: What you need to do deep learning.

https://course.fast.ai/

Can I Do the Course by Myself?

Yes! The online course is very much designed to be self-guided - but where's the fun in doing it yourself?!  Study groups are really all about motivation, camaraderie, and helping each other learn.

Group Member Instructions

The group started on Thursday 14th March 2019 - but it's not too late to join! It will run every Thursday at 12:30pm AEST (02:30 UTC). See Schedule.

From the course itself:

The easiest way to get started is to just start watching the first video right now! On the sidebar just click "Lessons" and then click on lesson 1, and you'll be on your way.

For an overview of the topics that are covered in the course, see: Practical Deep Learning for Coders 2019

Joining Instructions

Lessons

Coding Guide

The code is run via Jupyter notebooks using the fast.ai Python library (and other libraries such as pytorch, numpy, sklearn, matplotlib, pandas).  The course requires a GPU to run the code in a cloud environment.

Start getting familiar with running Jupyter by reading the fast.ai Getting Started guide.  It guides you through a range of cloud options and a tutorial on running the initial Jupyter tutorial.

Our group has had the following experiences with cloud options:

Cloud Result Cost Notes Best Contact
AWS EC2 Good $0.90/hr Matt Callanan
GCP Good Free for new users (up to $300) Use a personal GCP account. See https://course.fast.ai/start_gcp.html David Sutherland
Colab Mixed Free Mixed Results (make sure you tick the GPU option or you'll waste a lot of time!)
AWS SageMaker Negative Free (via Code Academy funding) Lesson 1 notebook failed to run to completion

Feel free to try your own and update the above table.

How Can I Run my own Fast.ai Study Group?

Feel free to fork this repository for your own group.

Copy and change this page and the Schedule page to match

Clone this wiki locally