Nvidia wants to clear the thicket around the numerous open source tools for AI development. AI Workbench is a web interface that combines numerous repositories from Github, its own NGC and the new partner Hugging Face.
Advertisement
AI Workbench loads a desired image generator or large language model (LLM) as a base. The tool can be used to generate content, but also to adapt the data records. As an example in the Siggraph keynote, Nvidia boss Jensen Huang showed a workflow with Stable Diffusion XL.
The image generator should create an image with a Huang action figure in space. Because the training set does not know Huang, Stable Diffusion XL rendered any man. A handful of pictures with Huang as an action figure were then added to the training set, after which new, suitable pictures could be generated. A chatbot with your own data should be designed in a similar way.
Seamless switching between PC and cloud
The biggest highlight apart from the simplified operation is the possibility of having content generated on different systems. You can start on your own desktop PC or notebook and move the project to a more powerful workstation or cloud platform if you wish.
This works with just a few clicks. AI Workbench adapts the runtime environments and containers accordingly. Users should not have any additional work with the change.
Advertisement
Interested parties can currently register for early access via Nvidia’s website. AI Workbench will be released in autumn. From then on, the tool requires a Hugging Face membership with cloud access – Nvidia and Hugging Face are launching a new subscription, but are not yet revealing the price. AI Workbench requires an RTX graphics card in your own PC; as a browser application, it runs on both Windows and Linux.
Recommended Editorial Content
With your consent, an external YouTube video (Google Ireland Limited) will be loaded here.
Always load YouTube video Load YouTube video now
(mma)
To home page
#Nvidia #Workbench #Flexibles #KITool #Browser