NVIDIA partners with Hugging Face to offer cloud-based AI training

In light of the current trends in autonomous generative artificial intelligence and metaverse applications, NVIDIA announced at SIGGRAPH 2023, held in Los Angeles, a collaboration with Hugging Face, a platform that allows users to share machine learning models and datasets. Furthermore, NVIDIA unveiled its NVIDIA AI Enterprise 4.0 and AI Workbench software tools designed specifically for autonomous generative AI applications. In the realm of metaverse applications, NVIDIA continues to drive the development of the OpenUSD alliance, launching autonomous generative AI services corresponding to OpenUSD, and establishing a partnership with image library provider Shutterstock to create richer 3D imagery through autonomous generative AI.

In collaboration with Hugging Face, NVIDIA will utilize its DGX Cloud supercomputing hosting service to facilitate businesses in effortlessly executing AI training tasks via machine learning models and datasets shared on the Hugging Face service platform. This will enable rapid implementation of trained AI technology in business services, thereby reducing the time and cost required to develop AI technology applications.

Additionally, in response to the growing demand for autonomous generative AI development among enterprises, the newly updated NVIDIA AI Enterprise 4.0 software tools integrate numerous resources for autonomous generative AI applications. These tools can interface with software services such as ServiceNow, Snowflake, Domino Data Lab, and ClearML, and operate on servers under Dell, HPE, Lenovo, Supermicro, or cloud platforms like AWS, Google Cloud, Azure, and Oracle.

The AI Workbench software tools can be paired with NVIDIA’s GPU products to establish, adjust, or deploy various autonomous generative AI technologies, allowing the transfer of content between device endpoints, data centers, or cloud platforms. It is anticipated to be made available to businesses this autumn.

Regarding metaverse applications, NVIDIA, after recently announcing the establishment of the OpenUSD alliance with Apple, Adobe, and other industry players to promote 3D content development standards, has declared the launch of three autonomous generative AI tools operated by Omniverse Cloud. These include ChatUSD, which automatically generates Python code to render USD content via text commands; RunUSD, which can convert USD content into interactive real-time ray tracing effects; and DeepSearch, which enables searching for untagged object content via text. This simplifies development complexity and workflow, accelerating the spread of USD content within Omniverse Cloud applications.

In collaboration with image library provider Shutterstock, NVIDIA will combine its NVIDIA Picasso autonomous generative AI visual application with Shutterstock’s Shutterstock.AI service, powered by OpenAI’s DALL·E text-to-image technology. This simplification of generating high dynamic range imaging (HDRI) 360-degree panoramic images allows creators to more swiftly produce exquisite 3D scenes, which can subsequently be used in games, images, or films.

Through the utilization of autonomous generative AI technology, NVIDIA anticipates accelerating the advancement of a new generation of image content creation applications.