Tag
NVIDIA highlights how Nordic countries are rapidly building out advanced AI computing infrastructure, with major deployments by Spotify, ABB Robotics, Elisa, and Swedish consortium Sferical One on NVIDIA DGX SuperPODs and Nemotron models, projecting 1GW of AI cloud capacity in the region by 2027.
提出训练无关的解码算法 One Block Ahead (OBA)，在块扩散语言模型中并发去噪当前块与下一块并传递采样器状态，在 DiffusionGemma 上将每次前向传递的 token 数提升 42.3%、单请求吞吐提升 16.3%。
NVIDIA CEO Jensen Huang said in a Bloomberg interview that he would not sell GPUs to AI labs deemed irresponsible, signaling a potential shift in how chip suppliers view their role in AI safety oversight.
NVIDIA 担任 PyTorch Conference 北美站赞助商，公布 12 场演讲与 25 个海报环节，涵盖编译器与 CUDA Graphs、FP4/FP8 低精度训练、分布式系统以及物理与智能体 AI 等方向。会议将于 10 月 20-21 日在圣何塞会议中心举行。
NVIDIA researchers present AlphaNIST, a paper exploring how emergent assembly skills arise from binary task rewards in robotic manipulation, accompanied by a project page and arXiv preprint.
回顾Nvidia于1999年10月11日发布GeForce 256——首款被其称为GPU的产品，以及Sega当年500万美元投资如何帮助Nvidia渡过难关并最终击败3dfx的历史故事。
The UK just hosted a major AI day with little attention: NVIDIA pledged $250M in compute for AI safety research with Oxford, UCL and Imperial, while AI deployments accelerate across the NHS, Danaher's AI-run lab, Liverpool's 'City Brain', and UK banks like Lloyds and Starling — highlighting NVIDIA's dominance of the AI stack and the UK's growing role in building it.
NVIDIA's Run:AI AI workload and GPU orchestration platform is now available under the Apache-2.0 license as an open-source monorepo, including its control plane, cluster services, CLI, and MCP server for managing GPU resources on Kubernetes.
NVIDIA 宣布多项英国生态合作与投资，包括五年2.5亿美元的 agentic 安全研究投入、与 Earthshot Prize 的合作，以及 Fuse Energy、Locai Labs、OneAdvanced、Speechmatics、Danaher、利物浦城市等企业与机构在能源、医疗、智慧城市等领域的全面部署。
欧洲多城市与 NVIDIA 合作，利用加速计算、Cosmos、Nemotron 和 Omniverse 构建物理 AI 智能体与数字孪生，用于城市管理、公共设施运维、交通与能源优化等场景。
NVIDIA 将完整的 Run:ai 平台开源，作为 NVIDIA DSX AI Factory 平台的一部分，提供生产级 AI 工作负载与 GPU 编排能力，包括可独立采用的 KAI Scheduler、GPU Fractioning Operator、KAI Resource Management 和 Workload Map 四大组件。
Clement Delangue 表达了 Hugging Face 与 NVIDIA 未来在法国扩大合作的意愿，认为法国凭借人才、活力和强大的开源 AI 生态，有望成为全球第三大数据中心与 AI 大国。
Nvidia's DLSS 5 will debut in Hogwarts Legacy on November 5th with significantly toned-down facial and environmental alterations after gamer backlash, alongside an October update making the model twice as fast on RTX 50-series GPUs with 24% fewer parameters.
LangChain 于 Interrupt 大会举办 Agent Arena 活动，由 NVIDIA 赞助：参与者可通过玩俄罗斯方块训练智能体、用 LangSmith LLM Gateway 选择模型，并在排行榜上观看智能体对竞争霸。
NVIDIA and its ecosystem partners are rolling out new software features, models, and performance optimizations ahead of the RTX Spark laptop launch on Oct 16, including agent apps like GitHub Copilot and Hermes, the Nemotron 3.5 Bolt 74B model, Adobe app optimizations, DLSS 5, and full developer framework support for local AI on the PC superchip.
一位用户分享了购买两台 Nvidia DGX Spark（约6100美元/台）的经历，用于学习和编程工作流，并询问 GLM 5.3 flash 的本地部署配方以及 Nvidia 论坛是否是获取相关资讯的最佳渠道。
法国企业正基于 NVIDIA 技术在航天、金融、公共服务和科研等领域大规模部署智能体计算，涵盖主权 AI 基础设施、开源模型及跨行业应用。
A post floats the idea of using AI to reverse-engineer NVIDIA's CUDA into an open-source implementation, inspired by AI-assisted game modding, and highlights ongoing frustration among AMD Radeon users over proprietary CUDA lock-in.
一位YouTuber通过NVLink载板将四块9年前的NVIDIA Tesla V100 GPU组合起来，在Qwen3.8-Flash-Next推理基准测试中，性能超越了DGX Spark和Apple M5 Ultra，最高领先幅度达230%。
NVIDIA的研究提出Semi-OPD替代方案，发现在初始学生的离线rollout上进行蒸馏通常优于on-policy蒸馏，在17组教师-学生配对中14次胜出，最高提升13.6%准确率和11.4倍训练加速。