The Daily Digest

Faster Scientific Simulations with New FP64 Algorithms in NVIDIA cuBLAS | NVIDIA Technical Blog

Many scientific simulations, from molecular chemistry to materials science, spend the majority of their execution time computing high-precision dense general…
CUDA Toolkit 13.4.2 introduces Ozaki-II and 2M algorithms that accelerate double-precision GEMM on integer Tensor Cores while preserving full FP64…
Ozaki-II reduces the number of integer multiplications compared to Ozaki-I by using the Chinese Remainder Theorem…

developer.nvidia.com12h ago

Run GPU-Powered Black Hole Interferometry with NVIDIA cuPhoton | NVIDIA Technical Blog

Imaging black holes is challenging because they are distant and appear tiny in the sky—for example…
NVIDIA cuPhoton .xBeam accelerates radio interferometry correlation for the Event Horizon Telescope by processing voltage recordings into visibilities on GPUs.
The FX correlator ingests VDIF frames, applies delay and fringe corrections, channelizes signals via FFT, cross-multiplies station pairs…

developer.nvidia.com13h ago

Run Native DLA Inference in Python on NVIDIA Jetson AGX Orin Using cuDLA | NVIDIA Technical Blog

AI applications at the edge often need to run multiple neural networks within limited compute and power budgets. NVIDIA Deep…
NVIDIA Deep Learning Accelerator (DLA) on Jetson AGX Orin executes supported neural network operations to improve power efficiency and free…
The cuda.bindings.cudla module exposes the cuDLA C++ API to Python…

developer.nvidia.com15h ago

Build a Robot Pick-and-Place Simulation with SimReady Assets

A Blog post by NVIDIA on Hugging Face
Back to Articles Enterprise + Article Published November 13…
For robotics workflows, SimReady means an OpenUSD asset has been prepared with the properties needed for a defined simulation use…

huggingface.co16h ago

XYBench: Helping with the problem behind the question

A Blog post by Ai2 on Hugging Face
Back to Articles Enterprise Article Published November 13…
We’re releasing XYBench , a benchmark of 8K+ questions that tests how AI assistants respond to requests based on misconceptions…

huggingface.co17h ago

Adapt Human Motion for Robots with NVIDIA SOMA Retargeter | NVIDIA Technical Blog

Human motion offers roboticists a great starting point for teaching humanoid robots how to move. A human demonstration captures…
NVIDIA SOMA Retargeter adapts human motion data to the proportions and joint structure of specific humanoid robots using the shared…
The retargeting pipeline scales motion targets to the robot, solves inverse kinematics with Newton built on NVIDIA Warp…

developer.nvidia.com17h ago

When One GPU Isn’t Enough: Serving GLM-5.3 Across Two NVIDIA DGX Stations with vLLM | NVIDIA Technical Blog

Running a frontier-scale model locally means keeping all its weights in GPU memory…
GLM-5.3 NVFP4 , a quantized mixture-of-experts model with 753 billion parameters and 40 billion activated per token…
Two NVIDIA DGX Stations connected over ConnectX-8 RDMA serve the model through one OpenAI-compatible endpoint using tensor parallelism (TP=2)…

developer.nvidia.com17h ago

llama.cpp v1.0.0 : A stronger foundation for the local AI ecosystem

A Blog post by ggml-org on Hugging Face
llama.cpp makes it possible to run multi-modal LLMs across a wide range of hardware - from consumer-grade local machines…
Through quantization, optimized backends, and a portable C/C++ implementation…

huggingface.co20h ago

Yang-Mills, OpenAI proof, and the mass gap problem

A Blog post by David Louapre on Hugging Face
A few days ago, OpenAI made an announcement about the existence of Quantum Chromodynamics (QCD) and a unified completion…
It sounds like a physics problem but this is actually the first half of the Yang-Mills existence and mass gap…

huggingface.co24h ago

Which Weather Model Is Most Accurate? We Score Them Every Day, in the Open

A Blog post by FINAL_Bench on Hugging Face
Back to Articles Community Article Published November 13…
A grid operator decides how much power to buy.

huggingface.co24h ago

LLMON: An LLM-Native Markup Language, now in Granite 4.1 3B and 8B

A Blog post by IBM Granite on Hugging Face
Back to Articles Enterprise Article Published November 12…
Over a long trajectory, the context fills with numerous types of components: a system prompt, past conversation turns…

huggingface.co2d ago

On-Device Speech Recognition and Text-to-Speech in Google AI Edge Gallery

A Blog post by LiteRT Community (FKA TFLite) on Hugging Face
Back to Articles Community Article Published November 12…
Today, we are introducing on-device Automatic Speech Recognition (ASR) and Text-to-Speech (TTS) capabilities in Google AI Edge Gallery .

huggingface.co2d ago

Introducing Salesforce SOV-Transcribe-1: #1 Open-Weight Model Built for Real-World Speech

A Blog post by Salesforce Sovereign Intelligence on Hugging Face
Back to Articles Community Article Published November 12…
SOV-Transcribe-1 — our first open-weight speech-augmented language model, with 4 billion active parameters (26 billion total)…

huggingface.co2d ago

ZONOS3-Mini: Open-Source Real-Time Voice Cloning & Steerable TTS in 100M Active Parameters

A Blog post by Zyphra on Hugging Face
Back to Articles Community Article Published November 12…
We are excited to release ZONOS3-Mini , our next-generation open-source real-time Text-To-Speech (TTS) model for multilingual speech synthesis…

huggingface.co2d ago

The Definitive Fine-Tuning Guide for 3D Generation Models

We’re on a journey to advance and democratize artificial intelligence through open source and open science.
Back to Articles Published November 12, 2026 Update on GitHub Upvote 8 +2 Suvaditya Mukherjee suvadityamuk Follow merve merve Follow…
If you are interested in fine-tuning, you can use our scripts with few arguments (assuming you updated your dataset…

huggingface.co2d ago

Validating GPU Platforms with Runtime AI Inventory via the NVIDIA DSX AI Factory | NVIDIA Technical Blog

As teams deploy and scale AI workloads using NVIDIA Dynamo, Kubeflow, or Slurm…
NVIDIA AI Cluster Runtime (AICR) now includes a qualified AI bill of materials (AIBOM) component built on the open source…
AICR generates version-locked recipes and reproducible deployment bundles for GPU-accelerated Kubernetes…

developer.nvidia.com3d ago

BotPsy: On-Device Personal Assistant with a Sub-Billion-Parameter Agentic Language Model

A Blog post by QVAC on Hugging Face
Frontier cloud LLMs carry high operational costs that can fluctuate unexpectedly, and relying on such third-party models introduces vulnerabilities…
Furthermore, proprietary models offer limited customization for specific workflows, while posing risks to privacy, data control, and transparency.

huggingface.co3d ago

Token Pooling-Aware Training for Multi-Vector Embedding Models

We’re on a journey to advance and democratize artificial intelligence through open source and open science.
Back to Articles Published November 11, 2026 Update on GitHub Upvote 18 +12 Tom Aarsen tomaarsen Follow Multi-vector embedding models…
This makes them wonderfully expressive for search, but those document vectors add up quickly.

huggingface.co3d ago

MotionGraphicsBench: Evaluating Motion Graphics Generation with Programmatic Verification

A Blog post by Figma on Hugging Face
Back to Articles Community Article Published November 10…
For each of our 100 tasks , we give a model a complete input static SVG and a motion edit…

huggingface.co4d ago

Running AV Reasoning Models Within Onboard Compute and Latency Limits | NVIDIA Technical Blog

Autonomous vehicle (AV) developers need reasoning models that can operate within the compute and latency constraints of an onboard system.
NVIDIA Alpamayo 2 Edge is a 2B-parameter adaptive reasoning vision-language-action model designed for real-time operation on NVIDIA DRIVE AGX Thor.
The model generates Chain-of-Causation reasoning only when needed and uses a diffusion action expert to produce driving trajectories from surround-camera…

developer.nvidia.com4d ago

Bring Your Own Expert! Extending MoE Models with New Domain Expertise

A Blog post by TNG Technology Consulting GmbH on Hugging Face
Back to Articles Community Article Published November 10…
This is called a Mixture-of-Experts (MoE) architecture [1] [2] .

huggingface.co4d ago

Agent Substrate: Advancing Kubernetes for Agents at Scale | NVIDIA Technical Blog

AI agents spend most of their time idle, waiting on a human prompt…
Open source Agent Substrate pairs with NVIDIA Dynamo and KAI Scheduler in the NVIDIA DSX AI factory platform to coordinate…
Agent Substrate multiplexes many agent sessions over a small pool of warm Kubernetes worker Pods…

developer.nvidia.com4d ago

Manage Kubernetes Clusters for GPU Infrastructure with NVIDIA Kubernetes Engine | NVIDIA Technical Blog

NVIDIA is announcing two open source projects for building and operating managed Kubernetes services on GPU infrastructure—NVIDIA Kubernetes Engine (NVKE)…
NVIDIA DSX AI Factory Platform introduces two open source projects, NVIDIA Kubernetes Engine (NVKE) and NVIDIA Kubernetes Recipes (NVKR)…
NVKR builds reproducible node images, container images, and packages from version-pinned sources…

developer.nvidia.com4d ago

Start and Restore GPU Workloads in Seconds with NVIDIA Snapshot | NVIDIA Technical Blog

NVIDIA Snapshot, a new open-source component of the NVIDIA DSX AI Factory Platform…
NVIDIA Snapshot is an open-source checkpoint and restore system for GPU workloads on Kubernetes that captures fully initialized CPU…
Restoring from a Snapshot artifact brings inference replicas online in 3.6 to 42.7 seconds across tested configurations…

developer.nvidia.com4d ago

The Daily Digest by email

AI