Contextual AI, founded by Douwe Kiela, has improved Retrieval-A performance optimization by up to 10x. This achievement builds on the company's founders' seminal research. The improvement was made before OpenAI's release of ChatGPT in 2022. Kiela and his team at Facebook recognized that large language models would face data freshness issues.

Source: https://blogs.nvidia.com/

NVIDIA has introduced Multi-LoRA Support, a technique for fine-tuning large language models (LLMs) to fit the needs of various applications. This feature is part of the NVIDIA RTX AI Toolkit, designed for RTX AI PCs and workstations. LoR allows developers to adapt LLMs for specific use cases, such as productivity tools, digital assistants, and non-playable characters in video games.

Source: https://blogs.nvidia.com/

Mike Pritchard is the director of climate simulation research for NVIDIA Research and a professor at the University of California, Irvine. He has led a research lab studying climate dynamics and next-generation atmospheric simulation algorithms since 2013. In 2022, he joined NVIDIA and directs Earth-2 research, managing a team exploring the potential of AI to advance weather and climate simulation.

Source: https://blogs.nvidia.com/

Jason Paul has been at NVIDIA since 2003, responsible for bringing to market various GeForce GPUs, including GeForce 6600, 7600, 7900, 8800, GTX 200, GTX 400, GTX 500, and GTX 600. He is currently in charge of NVIDIA's new gaming portable, SHIELD. Jason holds an MBA from Stanford University and a BA from UCLA.

Source: https://blogs.nvidia.com/

Matt Wuebbling has been with NVIDIA since 2006, serving as the director of product marketing for Tegra. He previously worked in the notebook business, launching NVIDIA's Optimus Technology. Prior to NVIDIA, Matt held leadership positions in engineering and marketing with established Silicon Valley companies and startups. He received a BS in computer engineering from Northwestern University and an MBA from the University of Michigan Ross School of Business.

Source: https://blogs.nvidia.com/

NVIDIA has introduced the NIM Agent Blueprint, a new framework for developing generative AI applications. The blueprint is powered by advanced open-source foundation models and agentic AI, which can improve efficiency and autonomy of AI workflows. Enterprises can use models like Google Gemma, Llama 3.1 405B, Microsoft Phi, Mixtral, and Nemotron to develop their own AI applications. These applications can support business growth and boost productivity.

Source: https://blogs.nvidia.com/

NVIDIA released the NIM Agent Blueprint for generative AI-based virtual screening, aiming to make the drug discovery process faster and smarter. Benchling, Dotmatics, Terray, TetraScience, and Cadence Molecular Sciences will use NVIDIA NIM microservices and NIM Agent Blueprints to push the boundaries of drug discovery. This innovative approach reduces the time and cost of developing life-saving drugs, enabling quicker access to critical treatments for patients.

Source: https://blogs.nvidia.com/

NVIDIA has launched four new microservices called NIM, which enable developers to build and deploy high-performing generative AI applications. These microservices support popular community models tailored to meet regional needs, enhancing user interactions through accurate understanding and improved responses based on local languages and cultural heritage. The move is in support of nations pursuing sovereign AI, which involves producing artificial intelligence using their own computing infrastructure, data, workforce, and business networks to ensure AI systems align with local values, laws, and interests.

Source: https://blogs.nvidia.com/

NVIDIA and its partners achieved the best overall AI training performance and submitted the most entries across all benchmarks, with 90% of all submissions coming from the ecosystem.

Source: https://blogs.nvidia.com/

Source: https://blogs.nvidia.com/