Jupyter

This topic contains the necessary information for how to run Jupyter notebooks on the student cluster.

Jupyter Hub

Jupyter notebooks should be started through the JupyterHub at

Select the course and environment to spawn a notebook server. For environments created using our instructions below or for TAs, the remaining time before the server auto-terminates will be displayed in the bottom status bar.

Your server will keep running and consume time unless you stop it manually or it runs out of time. Simply closing the browser window/tab does not stop the server.

Course Environments

After signing in you can start one of the provided environments for your courses or use your own (see below). Depending on the course you have access to one GPU or not. Each course comes with it's own time budget available to you.

DIY Environment

You can set up your own environment on one of the login nodes. Simply run the following script:

/cluster/jupyter/setup-for-home.sh

This will create a minimal Jupyter environment in the folder jupyter in your home directory. To use the environment use the full path to the bin folder in that directory as the custom environment path.

To select a particular version of python run the script like this:

PYTHON=3.11 /cluster/jupyter/setup-for-home.sh

Adding Packages

To add additional packages simply activate the environment:

. ~/jupyter/bin/activate

After that use pip --no-cache-dir to install any python package that you need. If you install PyTorch or modules requiring Cuda then follow the instructions here.


Page URL: https://isg.inf.ethz.ch/bin/view/Main/HelpClusterComputingStudentClusterJupyter
2025-04-01
© 2025 Eidgenössische Technische Hochschule Zürich