NVIDIA AI Workbench

It’s a developer toolkit for data science, machine learning and AI that streamlines and integrates git, containers and IDEs. It’s free and you can install it in minutes on your local or remote systems.

It provides features and automation for:

  • Simplifying setup on local and remote machines.

  • Streamlining for working with GPU-enabled development environments.

  • Reproducibility and portability with integration for Git platforms like GitHub and GitLab.

  • Handling containers and connecting to registries like NVIDIA’s NGC.

  • Working with IDEs like JupyterLab and VS Code.

Supported Operating Systems

  • Windows 11

  • Windows 10 (build 19041 or higher)

  • Ubuntu 22.04

  • macOS (Monterrey (12) or higher)

operating_systems.png

It’s for interactive development and computational work in JupyterLab and VS Code.

You can use it to:

  • Develop in JupyterLab or VS Code with containerized, GPU-enabled environments.

  • Work on your laptop or remote machines within the same user interface.

  • Clone fully functioning examples for things like RAG and fine-tuning with models like LLaMa 2 and Mistral.

  • Collaborate and distribute your work through Git platforms like GitHub and GitLab.

dev_envs.png

You can install and use it on your laptop, workstations, servers, cloud instances and VMs.

Deployment Pattern

  • You must first install it locally on your laptop or main daily driver. This is your Local AI Workbench.

  • Then you can install it on remote machines and connect them to your Local AI Workbench.

  • You can have multiple Remote AI Workbench Locations, but you should have only one Local AI Workbench.

Supported GPUs and Systems

  • Consumer and professional GPUs on laptops and workstations.

  • Datacenter GPUs on servers and in the cloud.

heterogeneous.png

Previous NVIDIA AI Workbench
Next Locations
© Copyright 2024, NVIDIA. Last updated on Mar 18, 2024.