Notebooks run within a Docker container so they are lightweight, portable, and easy to work with. You can start your Notebook with one of our pre-configured templates or use a custom docker image.
Your runtime behaves like a traditional operating system: You can download files and install and run any package with standard tools available in Linux e.g. wget, pip, etc. from within the provided terminal or within a Jupyter cell. Any changes to your environment are persisted across multiple sessions.
When you launch a Notebook, you can provide a workspace which is a source of notebook (.ipynb) or other files that will be downloaded and added to your Notebook instance when it starts. Common workspace sources include GitHub repos or individual files stored on GitHub.
Open the Notebooks tab to create a new notebook. After naming your notebook, the next step is to choose a pre-configured template or custom container. We'll select PyTorch which includes the PyTorch framework and the necessary NVIDIA libraries to enable GPU support. You can learn about using custom containers here.
In the next step, choose the machine type. Here you can choose from the managed service or your private cluster instances if you have created a private cluster. Select the Free GPU machine to access free GPU instances. On this step, you can also select auto-shutdown which is especially useful when using paid instances. Finally, you can toggle the Notebook to be public or private.
The last (optional) step is to customize the Workspace and the Container. The Workspace option provides the ability to pull down files into the notebook when it's created. A common example is to pull a GitHub repository (see image below). The container option allows you to provide a custom Docker image from a public or private container registry.
The final step is to click start notebook.
The Notebook interface
On the files tab, you can create, rename, and delete files and folders. Once your Notebook is online, you can edit and run notebook or other code files.
Jupyter operations, such as adding and removing cells, are fully supported. You can right-click on cells to view additional options.
The Jupyter interface
Standard code files are also supported with full syntax highlighting.