Exciting Updates from Nvidia Revealed at SIGGRAPH
Aside from showcasing the RTX Pro server, Nvidia made waves at the recent SIGGRAPH event with the introduction of two new additions to its Nemotron model family. The Nemotron Nano 2 and Llama Nemotron Super 1.5 are equipped with advanced reasoning capabilities, aimed at revolutionizing AI technology by enabling smarter AI agents.
The Nemotron family consists of enterprise-ready, open large language models (LLM) specifically designed to enhance agentic AI for tasks that demand sophisticated reasoning, instruction following, coding, tool use, and multimodal understanding. These models boast high accuracy relative to their size, particularly in scientific reasoning, coding, tool use, instruction following, and chat applications, according to Nvidia.
Nvidia’s Nemotron Nano 2 stands out with up to six times higher token generation throughput compared to similar models, while the Llama Nemotron Super 1.5 excels in reasoning accuracy, making it ideal for handling complex enterprise tasks. In addition, Nvidia is empowering robotics and machines with new AI models that can process visual information and make informed decisions based on what they “see.”
One of the latest innovations from Nvidia is the Cosmos Reason, a customizable 7 billion-parameter reasoning Vision Language Model (VLM) that allows robots and vision agents to interpret visual data and make decisions akin to human thinking. This breakthrough technology equips robotics with the ability to analyze and react to their surroundings, marking a significant advancement in the field of artificial intelligence.
