Nvidia Wants to Help Customers With Their Own AI
Nvidia announced some cloud services at its GPU Technology Conference that will allow companies to develop and run their own generative AI models.
These are two specific new cloud services. Nvidia NeMo is for language models, and Nvidia Picasso is for generating images, video and 3D models. The intention is that companies adapt the models to their needs and train them on their own data to be used for specific purposes, according to a press release.
The company offers pre-trained data processing algorithms and frameworks within the service, APIs, and support. Once the customer has trained a model, he can run it in Nvidia’s cloud or other cloud services. Both services are currently being tested for specific users who must first apply.
It is striking that Nvidia also immediately announces partnerships. For example, it will work with Getty Images for ‘responsible’ generative AI that converts text into images and video. The responsible part here revolves around copyright, where there are big question marks (and lawsuits) with quite a few generative AI.
Nvidia says it is creating a system where royalties generated by the AI models go to the original artists. The company is also said to partner with another image bank, Shutterstock, to build an image generator around 3D objects.