Skip to content

Latest commit

 

History

History
executable file
·
82 lines (56 loc) · 2.28 KB

HOWTO-2.md

File metadata and controls

executable file
·
82 lines (56 loc) · 2.28 KB
Raw

HOWTO 2 of the Machine Learning Committee of SIIM.

Intended Audience: People who have already setup a Docker runtime engine and have an account on DockerHub, and have run a turnkey ML DOcker

Table of Contents

  1. Background
  2. Installing git and getting a Github account
  3. Cloning a git repository
  4. Modifying the code on your local branch
  5. Submitting the code back to the repo owner

Chapter 1

In HOWTO-1 you learned how to take a turnkey ML Docker and run it on your system. Now, we assume you want to make improvements/changes to it. To do this, you will need the source code used to build the Docker by the original authors

Enter git ...

Chapter 2

Most people in the open source community use a site called Github to share their code with others. This requires the authors to have git running locally on their PC, and to have an account on Github to post their code for others to see and get

You can get a github account here https://github.com/

And install git to your local system from here https://git-scm.com/downloads

Chapter 3

In HOWTO-1 you grabbed a Docker that was already built and pulled it onto your PC to run. But once you have a github account and git running locally, you can download the source code for a DOcker and build it yourself. This also means you can modify it

In HOWTO-1 we went to a site where Dr. Shih had his pre-built Docker (dockerhub). To get his source code so can modify and build it yourself, go here and clone it (green button) https://github.com/ImagingInformatics/machine-learning

This will actually get you the entire SiiM-MLC repository, including the file you are reading right now!

Chapter 3

Let's modify the source. The first thing you will likely want is to enable the GPU version of Tensorflow on your computer (if it supports it). The place to do that is the line in the DOckerfile that installs keras. Instead of

RUN conda install keras

try using

RUN conda install keras-gpu

And save the new DOckerfile. Now rebuild your Jupyter DOcker by typing "make build" on the command line.

Chapter 4

And now that you've made massive improvements to Dr. S code you of course want to submit it back to him to update his repo. To do so learn about "git pull" requests here https://www.git-scm.com/docs/git-pull