H200 GPUs
PulseAugur coverage of H200 GPUs — every cluster mentioning H200 GPUs across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
AIRoA-0 system learns complex manipulation from human handheld demos
Researchers have developed AIRoA-0, a bimanual manipulation system trained on over 16,000 hours of human demonstrations recorded using a handheld gripper, without any robot-specific data during training. This system can…
-
Wave Forcing enables efficient streaming video generation
Researchers have introduced Wave Forcing (WF), a novel framework designed for efficient and high-quality continuous long-video generation. This method co-designs the generation, training, and execution processes to enab…
-
New Astra weather model uses less compute, beats ECMWF ensemble
Researchers have developed Astra, a novel probabilistic weather forecasting model that achieves high resolution with significantly reduced computational resources. Astra utilizes a two-sided transport attention mechanis…
-
Astra model achieves high-resolution weather forecasts with reduced compute
A new probabilistic weather forecasting model named Astra has been developed, achieving high-resolution predictions with significantly reduced computational resources. Astra utilizes a novel two-sided transport attentio…
-
Meta releases open-weight multimodal coding model Muse Spark Open
Meta has released Muse Spark Open, an open-weight, multimodal mixture-of-experts model designed for agentic coding tasks. This large-scale model can process images, audio, and video, and is trained on real-world develop…
-
OpenAI revenue surges, Gemini gets LMCache, Llama 4 leaks, Anthropic expands agent use · 1 source tracked
OpenAI has reported significant revenue growth, driven by enterprise adoption of GPT-5 and autonomous agents, with projections to double revenue by 2027. Google DeepMind has introduced LMCache for Gemini models, reducin…
-
TERRA framework tackles memory demands for high-resolution AI Earth modeling
Researchers have developed TERRA, a new framework designed to handle the memory-intensive demands of training high-resolution AI-based Earth forecasting models. TERRA introduces Sampling-Aware Window, Sequence, and Tens…
-
New Data-Centric Parallel method speeds up variable-length sequence training
Researchers have developed a new technique called Data-Centric Parallel (DCP) to address the computational challenges of training deep learning models on variable-length sequences. DCP dynamically adjusts runtime settin…
-
STAGE framework synthesizes LLM execution graphs for distributed workloads · 2 sources tracked
A new framework called STAGE has been developed to synthesize high-fidelity execution graphs for large language models (LLMs) and Mixture-of-Experts (MoEs). This framework aims to optimize distributed AI workloads by mo…
-
Prime Intellect releases open framework for training trillion-parameter MoE models
Prime Intellect has launched prime-rl 0.6.0, an open framework designed for training large Mixture-of-Experts (MoE) models using agentic reinforcement learning. This new system successfully trained the GLM-5 model on so…
-
Nvidia chips smuggled to China and Russia despite US export controls
U.S. authorities are investigating multiple cases of advanced Nvidia GPUs and other semiconductor technology being illegally smuggled to China and Russia, circumventing export controls. These efforts involve sophisticat…
-
Hugging Face and AWS Detail Foundation Model Infrastructure
Hugging Face and AWS have collaborated to detail the infrastructure required for training and running large foundation models. The blog post outlines a layered architecture, emphasizing the interplay between AWS's compu…