AI Workbench Applications
Applications (apps) are programs and tools available to use with NVIDIA AI Workbench. The following are the types of apps in AI Workbench:
Type |
Description |
Examples |
Where it runs |
|---|---|---|---|
| Web App | An application with a browser-based user interface. | JupyterLab, TensorBoard | Web apps run in the project container. |
| Process | A command or script that performs a task without an interface. | A shell script that downloads a dataset or model weights. | Processes run in the project container without exposing any network access. Processes usually rely on code in the project repository. |
| Native App | An application installed on your computer that you launch outside of the project container. | VS Code | Native apps run on your local system. Native apps might not have access to the project container. |
Use this documentation to learn about the following:
Each of your projects has its own applications, and you can start and stop them in the project window of the desktop app or by using the CLI.
To start and stop apps in the AI Workbench desktop app, click apps running in the status bar, or click Environment and scroll to the Applications section, and then toggle apps on or off.
To start apps by using the CLI, use the start and stop commands.
Most built-in apps come pre-installed in the NVIDIA-provided containers. All the default base images for AI Workbench have JupyterLab installed. The Pytorch and Tensorflow bases have TensorBoard installed.
You only need to add a built-in app to your project if you are using your own custom container that doesn’t have it installed. To add a built-in app to your project, click Environment, scroll to the Applications section, and then click Add.
In some cases, VS Code requires additional configuration. For details, see VS Code in AI Workbench.
Custom apps require you to specify information to start, stop, and use the app. Although custom apps are more complicated to configure, they offer greater flexibility and control over your workflow. To learn how to add a custom app to a project, see Walkthrough: Add a Custom Application.