Jupyter Lab

VS Code allows you to run a Jupyer Lab notebook on a compute node, optionally with a GPU, and connect to the notebook, all inside of your browser.

Usage

Click on the Jupyter tab in the left sidebar

You should see a page similar to the screenshot below.

First you need to configure your Jupyter Lab job's resources. See our remote desktop guide for more details.

A key addition here is the Conda environment input. Clicking on this reveals a dropdown menu of Conda environments.

The top environment(s) are global environments for all M3 users. So far, we have:

Environment name	Description
dsks_2024.06	"Data Science Kitchen Sink", a general purpose Python environment with commond data science packages installed. See the MLeRP page.

Below these prebuilt environments, you will see your own Conda environments that have jupyterlab installed in them. You must have run

conda install jupyterlab

inside your Conda environment in order for it to appear here.

Once you are happy with your configuration, click Launch. Once your job has started running, click Connect to open a new tab with Jupyter Lab running on a compute node. Wait up to a minute or so for it to initialise.