Jupyter notebooks have been a huge productivity boost for machine learning researchers. They are a great way to share code, and to share ideas. Still, it’s only been recently that they have introduced support for realtime collaboration. I have written this post as a tutorial to help you get started with realtime collaboration using Jupyter lab on AWS and GCP.

According to gpt3:

## What is Jupyter lab?

Jupyter lab is a Jupyter notebook extension that allows you to run Jupyter notebooks in realtime. It is a Jupyter notebook server that runs on your local machine, and it is connected to a Jupyter notebook server on a remote machine.

## What is realtime collaboration?

Realtime collaboration is the ability to run Jupyter notebooks in realtime. Jupyter notebooks are run on your local machine, and Jupyter notebooks are connected to a remote Jupyter notebook server. The remote Jupyter notebook server is the one that is running on your machine.

## What is GCP?

    Google Cloud Platform is a cloud computing platform that provides a set of open-source software tools for developers to build, deploy, and manage their cloud computing resources.

## What is AWS?
    
    Amazon Web Services is a cloud computing platform that provides a set of open-source software tools for developers to build, deploy, and manage their cloud computing resources.

AWS

  1. Create an ec2 instance with the deep learning image
  2. Create a security group that allows all inbound traffic on ports 22 (ssh) and 8888 (jupyter lab) and allows all outbound traffic on all ports.
  3. Open a local terminal and ssh into the instance
  4. Run the following command to start jupyter lab 
    jupyter lab --ip=0.0.0.0 --port=8888 --collaborative
  5. Copy and paste the URL that the jupyter lab server pits out into your browser but replace https://localhost with the ephemeral ip address of the instance. Make sure to use the port 8888.
  6. Troubleshooting: sometimes, I have to explicitly reassign the security group.
  7. Terminate the instance when you are done.

GCP

  1. Create a GCP project and request a GPU quote increase
  2. Create a GCP compute engine instance with the deep learning image
  3. Run the following command to start jupyter lab 
    jupyter lab --ip=0.0.0.0 --port=8888 --collaborative
  4. Copy and paste the URL that the jupyter lab server pits out into your browser but replace https://localhost with the ephemeral ip address of the instance. Make sure to use the port 8888.
  5. Troubleshooting: sometimes, I have to explicitly run a network analysis on port 8888 on the instance.
  6. Delete the instance when you are done.