Jupyter Notebooks is a powerful computational notebook tool that lets you create and share your code. JupyterLab uses Jupyter Notebook and takes it much further. JupyterLab is an interactive web-based development environment for working with Jupyter Notebooks, text editors, terminals, and custom components.
You can arrange multiple Notebooks, documents, and activities in a single interface using tabs and splitters. Also, JupyterLab has a modular extensible architecture that lets you customize your development environment. A Notebook is just one of the components that you can work on in JupyterLab.
To create a new JupyterLab instance:
- Log in to your Neu.ro dashboard.
- Click RUN A JOB in the IDEs widget.
- Select Jupyterlab in the top drop-down menu, a preset in the bottom drop-down menu, and click RUN.
When you start JupyterLab, the platform storage is attached as /var/storage. All data created during a lab session in this folder persists and can be used later.
All JupyterLab instances are jobs that run on GPU-small preset. You must kill the JupyterLab session whenever you are done; else, it consumes GPU hours.
The Jupyter UI has the following areas:
- Tools area: The tools area lists the most commonly used commands.
- Work area: The work area hosts the launcher and the tabs that you are currently working on.
- Menu: The menu bar includes all available options for JupyterLab.
The tools area lists a host of options available for JupyterLab, such as:
- File Browser for managing files in the storage.
- Sessions for managing a list of active sessions.
- Commands for viewing the commands available for the current state.
- Properties for managing the properties for the current selection.
- Tabs to view the list of tabs open.
- Extension Manager.
You can manage your JupyterLab instances from the Neu.ro dashboard. The JupyterLab instances are listed in the Running Jobs section, and their names start with jupyter-lab-.
You can view the details of your JupyterLab instance by clicking on its job ID.
There are a few ways you can open an instance of JupyterLab:
- Click HTTP URL in the Running Jobs section on your Dashboard.
- Clicking OPEN RUNNING in the IDEs widget on your Dashboard.
- Clicking HTTP URL in the drop-down menu for the corresponding job in the job list.
You can kill the JupyterLab instance using the same drop-down menu. Note that closing the tab does not kill the instance.
Alternatively, you can kill the instance from the JupyterLab instance using the File > Shut Down option.
Killing a JupyterLab instance
An instance is automatically terminated 24 hours after initiation.