hardware

GLiNER2.5-Decide (DeBERTa-v3-large, 340M) converted to LiteRT: fp16 weights, answers identical to the official fp32 model on 361/361 test requests, 66–70 ms per request on a Galaxy S26 GPU at 128 toke

Let me get this straight. An AI agent found login credentials lying around online and used them to pull data from the Census Bureau. It tried to break into the Education Department's civil rights offi

Rubin HBM Despec = Nvidia cutting coke Fentanyl Grade Compute: Why a GB300 Rack Will Out-Price Blow by Weight in 2030 A GB300 NVL72 weighs roughly 1,580 kg fully populated and recent purchase orders p

arXiv:2609.29648v1 Announce Type: new Abstract: Video object detection on edge devices runs computationally expensive detectors over long frame streams, causing high energy consumption and sustained G

arXiv:2609.30198v1 Announce Type: new Abstract: Latent neural surrogate solvers, or latent dynamics models, accelerate simulations of time-dependent physical systems by evolving a compressed latent sp

arXiv:2609.29395v1 Announce Type: cross Abstract: Training with a fixed global batch limits how many distributed clients can provide examples in any one step. We examine a way to use additional server

The rise of AI-native cloud provider CoreWeave Inc. is part of the greater story emerging around operationalizing AI. As enterprises shift from training to inference, neoclouds such as CoreWeave are p

everybody talks about the economics of the ai data center build, the biggest capex buildout in human history not nearly enough people are stepping back to just marvel at the technology itself i just m

arXiv:2609.29999v1 Announce Type: cross Abstract: Post-training quantization of vision--language models (VLMs) is typically assessed through aggregate task accuracy and memory savings, but preserving

Every rent-vs-buy thread I read has confident people on both sides, but not many actually show the numbers. So I finally ran the numbers for our own decision. Posting the working here in case it is us

arXiv:2609.30059v1 Announce Type: cross Abstract: Deep learning inference and training performance depends critically on GPU kernel efficiency. Modern compilers such as PyTorch Inductor automatically

arXiv:2609.29121v1 Announce Type: cross Abstract: Audio-Visual Semantic Segmentation (AVSS) aims to identify, segment, and classify sound-emitting objects in video frames. Previous Transformer-based A

arXiv:2603.06471v3 Announce Type: replace Abstract: Acquiring per-frame annotations for ultrasound videos is costly and requires clinical expertise, limiting learning-based analysis. We study cross-vi

arXiv:2609.29044v1 Announce Type: cross Abstract: Pilot-aided channel estimation is a decisive block in orthogonal frequency-division multiplexing (OFDM) receivers for both 5G New Radio (5G-NR) and Lo

arXiv:2609.29157v1 Announce Type: new Abstract: To achieve general-purpose autonomy over long horizons, a robot needs to maintain spatial environment information that supports a variety of tasks: geom

arXiv:2602.05145v2 Announce Type: replace-cross Abstract: Speculative decoding can substantially accelerate LLM inference, but realizing its benefits in practice is challenging due to evolving workloa

arXiv:2609.30222v1 Announce Type: cross Abstract: Existing point tracking models face a fundamental tradeoff: they can either track a sparse set of query points over long horizons, or track all points

arXiv:2609.28923v1 Announce Type: new Abstract: Few-step autoregressive (AR) video diffusion enables low-latency streaming generation, but existing post-training methods predominantly rely on Distribu

arXiv:2609.29106v1 Announce Type: cross Abstract: 3D foundation models recover video cameras and geometry in one forward pass, but some of the strongest are up to scale. Joint people-scene reconstruct

arXiv:2609.27232v1 Announce Type: new Abstract: Scaling laws have guided large-model development in computer vision and natural language processing, but the relationships among data, model size, and c

arXiv:2609.27342v1 Announce Type: new Abstract: Cutting changes both the shape and topology of deformable objects, making accurate simulation challenging for robotic manipulation. A simulator must tra

arXiv:2609.27581v1 Announce Type: cross Abstract: Step Law gives power-law formulas for the optimal peak learning rate eta* and batch size B* when pre-training language models. It was calibrated on mo

arXiv:2609.27547v1 Announce Type: new Abstract: Embodied reinforcement learning (RL) improves model capabilities with a pipeline of environment simulation, action generation, and model updates. These

Mixture‑of‑Experts (MoE) models scale capacity more efficiently than dense transformers by activating only a subset of expert subnetworks per token. NVIDIA’s Transformer Engine optimizes MoE training

Fentanyl Grade Compute: Why a GB300 Rack Will Out-Price Blow by Weight in 2030 A GB300 NVL72 weighs roughly 1,580 kg fully populated and recent purchase orders put it at 5M per rack. That is 3,165/kg.

Ok, so the Mac M5 Ultra (256GB) hit the market, but the only publicly available benchmarks material are flashy YouTube 'clown influencers' videos. We need serious numbers to evaluate whether Apple’s s

When COVID-19 emerged, scientists had a crucial advantage: Decades of prior research on coronaviruses meant they understood the virus’ key proteins well enough to design vaccines in record time. The n

As Jensen Huang puts it, AI can help fight climate change - but only if it inflicts 'an enormous amount of pain and suffering' first. The Nvidia CEO discussed the future of energy and AI's impact on o

Rafe Rosner-Uddin / Financial Times: Q&A with Cisco President and CPO Jeetu Patel on its multibillion-dollar networking deals for AI infrastructure, competition from partners like Nvidia, and more — T

arXiv:2609.27209v1 Announce Type: cross Abstract: Subgraph sampling reduces the training cost of large-scale graph neural networks, but sampling criteria may overlook the geometric roles of edges. We

arXiv:2609.28208v1 Announce Type: new Abstract: Tabular foundation models face a feature-side scaling dilemma: full-width pairwise mixing grows quadratically with the number of columns, whereas featur

arXiv:2609.25624v1 Announce Type: cross Abstract: Large language model (LLM) outputs are expected to be reproducible under greedy decoding, yet in practice the same model, prompt, and software stack p

Basecamp Research Ltd. today announced that it has raised 140 million in funding from a group of prominent investors. S32, a fund affiliated with Google LLC co-founder Bill Maris, led the Series C rou

arXiv:2609.26315v1 Announce Type: new Abstract: Orchard robots need maps that preserve small but semantically important structures such as trunks, trellises, and fruit. 3D Gaussian Splatting (3DGS) SL

NVIDIA AI Day Singapore, which takes place Sept. 22-23 at the Raffles City Convention Centre, is offering attendees opportunities to explore the hands-on training, expert-led sessions and advanced too

Coachella of AI is here. Still some tickets available at http://summit.runway.com. Line up is ready: Adarsh Kulkarni, Foundry Robotics Alexander Toshev, Wayve Anastasis Germanidis, Runway Andy Chen, R

arXiv:2609.24456v1 Announce Type: cross Abstract: Distributed reinforcement learning (RL) scales training by parallelizing actors and learners around an Experience Buffer. As RL workloads grow, howeve

arXiv:2609.26796v1 Announce Type: new Abstract: Diffusion Large Language Models (dLLMs) have recently emerged as a promising alternative to autoregressive LLMs by enabling non-autoregressive text gene

arXiv:2609.26621v1 Announce Type: new Abstract: Greedy decoding from large language models is commonly treated as deterministic. We show it is not precision-invariant: the same model, prompt, and deco

Interesting paper on agent memory stored as a linked markdown wiki. Lots of great ideas and insights if you work with LLM Wikis. Wikis are useful for agents because each page holds dense text and the

NV‑Reason‑CT is a purpose‑built vision‑language model for volumetric CT that uses a full 3‑D Vision Transformer encoder coupled with a Qwen 3.5‑4B language model to generate radiologist‑style chain‑of

John Reynolds / Tech.eu: London-based AI life sciences startup Basecamp Research raised a 140M Series C led by S32, with participation from Anthropic, taking its total funding to 225M — Basecamp says

Kijai put his magic to good use to improve your MiniMax H3 workflows. The video VAE now encodes up to ~2.2x faster and decodes ~1.4-2.7x faster on Nvidia GPU. Checkout the blog post how to use it: htt

NodeWright is an open‑source, Kubernetes‑native package manager that declaratively configures host operating systems on GPU clusters while preserving workloads; its operator cordons nodes, drains pend

When Sakeena Fiza describes her work as a validation engineer at NVIDIA, she does so in terms more befitting a detective story than a world-class engineering lab. “Validation engineers look in the sha

Financial Times: Sources and Nscale's US SEC filings: ByteDance accounted for nearly 75% of Nscale's sales in 2025 and used Nscale's facility in Norway to access Nvidia AI chips — Chinese tech giant w