News
Inside NVIDIA Halos for Robotics: A Full-Stack Functional Safety System for Physical AI
1+ day, 9+ hour ago (883+ words) Physical AI—robots working autonomously alongside people in factories, warehouses, hospitals, and homes—is arriving faster than most expected....
Building AI Agents for AR Glasses and XR Devices with NVIDIA XR AI
5+ day, 16+ hour ago (792+ words) Developers building for AR glasses and wearable devices face an infrastructure gap. The hardware is ready, but creating AI experiences requires integrating live camera and microphone streams, multimodal AI models, enterprise data, tool use, deployment infrastructure, and device-specific runtimes. NVIDIA…...
Build Your Own Transaction Foundation Model for Financial Intelligence
5+ day, 18+ hour ago (1134+ words) Every swipe, transfer, and payment on a modern financial network encodes a pattern of human behavior. Transaction data is one of the richest signals an enterprise owns. Yet most production use cases for such tabular data still depend on hand-engineered…...
Build On-Device AI Companions with the NVIDIA ACE Game Agent SDK and Unreal Engine 5 Plugins
5+ day, 20+ hour ago (791+ words) Building on-device AI NPCs requires not only choosing the right AI models, but also integrating them into a seamless, high-performance pipeline. Developers must overcome hurdles to deliver a viable experience, such as minimizing conversational latency for instantaneous responses, syncing global…...
How to Optimize Transformer-Based Models for Low-Precision Training
6+ day, 22+ hour ago (464+ words) Suppose you're working with a 5 B-parameter model such as Codon FM 5 B. It will have a config such as: Your training configuration is: The benchmark tool can then take these hyperparameters directly and then use a single command to derive GEMM…...
NVIDIA Blackwell Tops MLPerf Training 6. 0 with Industry-Leading Scale and Performance
5+ day, 23+ hour ago (479+ words) NVIDIA delivered a clean sweep in MLPerf Training v6. 0, the latest edition of industry-standard AI training benchmarks developed by the MLCommons consortium. NVIDIA achieved the fastest time to train at scale, and also delivered the highest performance when normalized on a…...
Fine-Tuning Biological Foundation Models with Lo RA Using NVIDIA Bio Ne Mo Recipes
6+ day, 20+ hour ago (1195+ words) Foundation models are reshaping computational biology. Pretrained on massive corpora of protein or genomic sequences, models such as ESM2 (a protein language model) and Evo 2 (a DNA language model) capture statistical regularities of biological sequences. These transfer well to a wide…...
Pretrained to Imagine, Fine-Tuned to Act: The Rise of World-Action Models
1+ week, 12+ hour ago (1580+ words) But the ideas behind WAMs are not new. Early WAMs like Uni Pi [10] proposed essentially this approach back in 2023. So why did it take several years for the paradigm to enter the robot foundation model mainstream, and where does it…...
NVIDIA Achieves Leading Agentic Coding Performance on First Agentic AI Benchmark
1+ week, 2+ day ago (340+ words) AI agents have fundamentally changed the complexity of inference workloads. Until now, the industry has struggled to define a standard for measuring how inference systems perform under these conditions. Artificial Analysis Agent Perf (AA-Agent Perf) offers the industry's first multi-vendor…...
Deploy Long-Context Reasoning and Agentic Workflows with Mini Max M3 on NVIDIA Accelerated Infrastructure
1+ week, 2+ day ago (421+ words) NVIDIA Developer Deploy Long-Context Reasoning and Agentic Workflows with Mini Max M3 on NVIDIA Accelerated Infrastructure - Mini Max M3, a 428 B parameter Mixture-of-Experts model with 1 M-token context and native multimodality, leverages NVIDIA Blackwell infrastructure to unify text, vision, and code tasks, supporting…...