Volume 341, Issue 4 · 15 articles from 2 issues
NVIDIA details the quantization-aware distillation pipeline used to build Nemotron 3.5 Lightning in NVFP4 precision. Read more
NVIDIA examines generative recommenders as a shift in how recommendation systems are trained and served at scale. Read more
NVIDIA introduces SkillEvaluator for measuring how well agent skills supply the context agents need to work efficiently. Read more
NVIDIA presents AdaptGrow, a GPU-accelerated matrix factorization method for clustering financial instruments. Read more
NVIDIA shows how AI coding agents plus the ALCHEMI toolkit lower the barrier to running atomistic materials simulations. Read more
NVIDIA argues AI factories are the defining infrastructure of the AI era and sets out what securing them requires. Read more
NVIDIA DSX MaxLPS combines dynamic power allocation and performance tuning to raise AI output per available watt. Read more
NVIDIA shows how to post-train Cosmos 3 Edge world models so robot control policies run on onboard compute. Read more
NVIDIA shows how the Holoscan CLI, skills and AI coding agents speed development of real-time edge AI applications. Read more
NVIDIA cuML and cuVS added multi-GPU support for UMAP, executing over 870 GB of vectors in eight minutes. Read more
NVIDIA FLARE orchestrates federated training for vision-language models across institutions that cannot pool data. Read more
NVIDIA's Vera CPU, its first processor designed for AI agents, has begun shipping at scale across the AI ecosystem. Read more
NVIDIA Dynamo's Shadow Engine Recovery avoids cold restarts when an LLM engine process fails. Read more
NVIDIA argues custom XPU silicon must be deployed within full AI factory infrastructure to deliver on token economics. Read more
TensorRT Model Connect reduces open model deployment from custom conversion code to two commands. Read more