Colab

Open https://colab.research.google.com/github/ to browse your GitHub repo (or any public that contain Jupyter notebooks)
Load the notebook in Colab
Sometimes the notebook references other files in the github repo. Add the following to your Colab notebook to download the repo (replace bitdribble/LDL with <github_user>/<repo>):
```
# Colab starts shell in /content
!rm -rf LDL
!git clone https://github.com/Bitdribble/LDL.git
```
Sometimes you need extra python modules installed. Create a colab_requirements.txt in your git repo, and load it from the notebook:
```
!pip install -r /content/LDL/colab_requirements.txt
```
Sometimes you need to download data. Invoke a shell command from your git repo to do that:
```
!/content/LDL/data/mnist/download_mnist.sh
```
Before running the rest of the notebook, sometimes you need to cd to a folder. We use % so the command is executed in the notebook environment, using ! would execute it in its own environment:
```
%cd /content/LDL/stand_alone
```
Run the rest of the notebook, and fix any remaining issues
When done, click File->Save a Copy in GitHub. Add a Open in Colab badge.
- I save notebooks in a separate top-level folder called colab, so I don’t break the source notebooks.
- You need to grant Colab permission to your GitHub account.
For an example setup, see https://github.com/bitdribble/LDL

In Colab, use !shell_cmd to execute a shell command.
To execute it in the Colab environment, rather than as a subcommand, do %shell_cmd. This is useful when changing directory, for example.