New NVIDIA AI Foundation Models

Pre-trained models

The NVIDIA NGC catalog and Hugging Face, as well as the Microsoft Azure AI model catalog, now provide NVIDIA AI Foundation Models for free experimentation.

Companies today have access to a whole smorgasbord of free, open-source large language models (LLMs). Because of factors like compatibility, security, and the data needed to train the models, developers tasked with creating bespoke generative AI applications may feel overwhelmed by the sheer volume of available options.

With NVIDIA AI Foundation Models, an enterprise-grade collection of pre-trained models, programmers can get a head start on integrating bespoke generative AI into business applications.

Read Top 20 Uses of Artificial Intelligence In Cloud Computing For 2024

Accelerating Innovation with NVIDIA-Optimized Base Models

The NVIDIA AI Foundation Models are accessible from any web browser via an easy-to-use interface or application programming interface (API). In addition, businesses may access these models via NVIDIA AI Foundation Endpoints to evaluate model functionality directly within their software.

Popular community models like Llama 2 and Stable Diffusion XL are also available in a format that simplifies the integration of developers’ data. Additionally, models have been tuned using NVIDIA TensorRT-LLM to give the maximum throughput and lowest latency and to run at scale on any NVIDIA GPU-accelerated stack. When using TensorRT-LLM to improve the Llama 2 model, for instance, performance is substantially doubled on NVIDIA H100.

Read: Top 15 AI Trends In 5G Technology

What is new?

The models are a good place to begin when creating safe, production-ready generative AI applications for your business. They were developed using ethically sourced datasets and performed at a level equivalent to much bigger models. Because of this, they are superb for use in large corporations.

The Nemotron-3 8B models stand out due to their multilingual capabilities. In addition to English, German, Russian, Spanish, French, Japanese, Chinese, Korean, Italian, and Dutch, the models are pre-trained to understand over 50 more languages.

Businesses that want to implement generative AI across their operations will require the services of an AI foundry so that they may tailor the models to their specific needs. Including the NVIDIA AI Foundation Models, the NVIDIA NeMo framework and tools, and the NVIDIA DGX Cloud AI supercomputing services, NVIDIA has created an AI foundry. Together, they form a comprehensive business solution for developing unique AI generative models.

NVIDIA AI Enterprise software provides enterprise-grade security, stability, and support for custom-built models, and those models may be deployed almost anywhere on accelerated computing.