Skip to content


Collaborate in near real time on Jupyter notebooks to develop, troubleshoot, pair program, and tutor. Publish parametrized notebooks, or AppBooks, which automatically generate a form that allows everyone, including domain experts, to train a model, while seamlessly tracking parameters, metrics, and models.

Quick start

You can create a notebook by clicking on Notebook on the sidebar, and choose from the Server Options available. A fresh Jupyter server will be started for you with the most popular libraries already installed, and hassle free GPU support.

Start a notebook server

This might take a few moments but we're always working to improve the user experience, especially for users with unreliable internet connections, as are the developers of Jupyter.

Multiple checkpoints

You can easily go back and forth to different versions of your notebook by clicking on File, then click on Revert to a checkpoint in the menu. A dialogue titled Choose a checkpoint will appear. Select the checkpoint, and click OK. A confirmation message will appear, and you can click Revert to confirm.

Revert to a checkpoint


Work with other users on the same notebook in near real-time by sharing a link to your notebook, or choosing who to share it with. Once they open the link, you will see each other's cursors and changes to the notebook as they happen. This is useful when one needs help to troubleshoot a notebook, or when multiple users work on different parts of the notebook.

Click on the button that has the hint Share notebook Start a notebook server, a dialog box titled Share will appear, and you will have two ways to share the notebook:

  1. Click on the button labeled Get link
  2. Choose a Permission:
    • The Read only permission allows the user to make changes but not save them.
    • The Edit permission allows the user to make and save changes.
  3. Click on the button labeled Copy to copy the link
  4. You can now share the link via email or instant messaging.

Invite a specific user

  1. Click on the button labeled Invite
  2. Pick a user by clicking on the drop down list
  3. Click Next to choose permissions
  4. Choose a Permission:

    • The Read only permission allows the user to make changes but not save them.
    • The Edit permission allows the user to make and save changes.
  5. Click OK

Invite a user to collaborate on a notebook

Long-running notebooks

Working with a notebook to explore data is fun, but what if you want to run a compute intensive training job on larger data? You can schedule your notebooks to be executed with more compute and be immune to network issues, or your browser being closed. You can choose whether to overwrite your notebook with the results, or save the resulting notebook to another one.

Run a notebook asynchronously

You can even watch your notebook's output on your mobile phone as it runs.

Launch it, watch it on your mobile device

Simply click on the Schedule notebook from the toolbar. Choose a Docker image you want your notebook to be executed on from the dialog box, and specify the notebook name under which the results will be saved. If you want to overwrite the notebook, you can leave the default name.

A dialog will appear informing you that your notebook has been scheduled. You can visit the list to see the progress, status, and position in the queue as there may be notebooks of yours or other users that are being run.

Running notebooks asynchronously is extremely useful in several scenarios:

  • Multiple users no longer compete for limited compute resources
  • Accessing notebook outputs without having to keep the browser open, which is important for any non trivial work.
  • Your notebook runs are tracked, so you know how you got to that awesome model you now have.

Publishing an AppBook

AppBooks are applications generated from notebooks. They are a refinement on parametrized notebooks in that they present an auto-generated form with variables in the code that you decide to expose to users, and enable running the notebook with the new values without changing the notebook's code. They also allow you to deploy a model, or build a Docker image for it in one click.

To publish an AppBook, click on the Publish button in the notebook's toolbar. If you want to expose a parameter to the AppBook user, simply assign a default value to it.

Publish an AppBook

Read more on AppBooks

Using Git repositories

A good way to get started is to clone a popular Git repository that contains notebooks for machine learning.

Cloning with jupyter-git

The notebook server has the jupyterlab-git extension pre-installed, which allows you to work with Git repositories right from the notebook's interface with the following steps:

  • Click on the button with the Git icon and the Git Clone hint
  • Paste the repository link into the modal dialog's text field
  • Click on the modal dialog's CLONE button

Clone a Git repository from the Notebook interface

Cloning with Git

You can clone a repository with git by clicking on Terminal in the notebook's Launcher, then using the git command line interface to clone a repository on the web pseudo terminal that appears.

Cloning popular repositories from the command line
# courses
git clone

# Fran├žois Chollet's "Deep Learning with Python" notebooks
git clone

# Lex Fridman's "MIT Deep Learning" courses notebooks
git clone

Clone a Git repository from the pseudo terminal