←搜索结果

搜索“Hugging Face”

按匹配度 · 共 174 条结果 清除搜索

星期四

  1. NVIDIA Developer · Generative AI精选

    NVIDIA 在 GB300 NVL72 上部署 Qwen3.8-2.4T-A95B，单 GPU 吞吐超 4K tokens/s

    NVIDIA 发布在 GB300 NVL72 上部署阿里开源权重模型 Qwen3.8-2.4T-A95B 的方案，该模型总参数 2.4T、每 token 激活 95B，采用细粒度 MoE 与全注意力加线性注意力混合架构，上下文窗口最高 100 万 tokens。在 FP8 精度下 Day 0 即实现单 GPU 超 4K tokens/s、单用户超 350 tokens/s，NVFP4 优化预计带来进一步提升。SGLang、vLLM 和 NVIDIA Dynamo 提供开源推理配方，也可通过 NVIDIA NIM 容器部署，并可用 NeMo AutoModel 做全量 SFT 或 LoRA 微调。

    timizations. NVIDIA NeMo AutoModel supports post-training the model for domain-specific use cases with full supervised fine-tuning or memory-efficient LoRA fine-tuning directly on Hugging Face checkpoints. Open-source inference recipes are available for SGLang, vLLM, and NVIDIA Dynamo, and the model can also be deployed via a model-free NVIDIA NIM container from NVIDIA NGC. Next Steps Try the model on DeepInfra for immediate hosted inference access. Download weights from Hugging Face to run the

    匹配字段正文

    #lab#产品更新#开源生态#部署/工程

星期三

  1. The Next Platform

    Nvidia 发布 Nemotron 3.5 Lightning 模型与 NeMo Switchyard 路由库

    Nvidia 发布 Nemotron 3.5 Lightning 开源模型和 NeMo Switchyard 路由库，前者面向持续运行的智能体任务，官方称在同级中实现 4 倍吞吐、完成智能体任务速度提升 30%，并可在 Jetson、GeForce RTX、DGX Spark 和 DGX Station 上本地运行。Switchyard 让开发者自定义模型池、路由标准和策略，把推理密集型步骤交给前沿模型、把执行类任务交给 Lightning，官方称其以 Anthropic Opus 4.8 约三分之一成本取得更好准确率。Nvidia 同时公开训练数据与方法，并发布用于训练编码智能体的开放数据集 Nemotron-RL-Agentic-Terminal-Pivot。

    ving deeper into other areas of generative and agentic AI, including security. Soon after reports of OpenAI’s GPT-5.6-Sol and another unreleased frontier model autonomously breach Hugging Face, Nvidia led the development of a consortium of more than three dozen companies launched the Open Secure AI Alliance, arguing that protections used to defend infrastructure from AI threats need to be built on open models and tools that all defenders can adapt and deploy. The consortium now includes more tha

    匹配字段正文

    #hot-signal#产品更新#Agent#开源生态

星期二

  1. LMSYS Blog

    LMSYS 发布 SpecForge v0.3.0：分离式与同置式推测解码训练栈及新 SpecBundle 草稿模型

    LMSYS 发布 SpecForge v0.3.0，将目标模型推理与草稿模型训练解耦为生产者与消费者两个资源池，在 8×H20 上以 3 台 SGLang 捕获服务器加 5 个训练 worker 的拓扑，端到端训练吞吐较此前同置实现提升约 10%。新运行时统一支持 EAGLE3、EAGLE3.1、P-EAGLE、DFlash、Domino 和 DSpark 多种推测解码方法，并新增训练与推理一致性校验流程。同时发布 SpecBundle 草稿模型集合，11 个 checkpoint 中有 9 个来自蚂蚁集团 AQ、RadixArk、招商银行、Domino 作者及社区成员，全部仅用开放数据训练。

    SpecForge online-disaggregated training architecture

    t alongside these — contributions of new targets and new algorithms are both welcome. Released models and performance All checkpoints are published in the SpecBundle collection on Hugging Face: Results for a subset of these models are shown below, grouped by algorithm. EAGLE3 Figure 3. Three EAGLE3 draft models at the same drafting configuration (3 steps, top-k 1, 4 draft tokens): output throughput against the autoregressive baseline, with the speedup labelled above each bar. Step-3.5-Flash and

    匹配字段正文

    #company#开源/仓库#部署/工程#数据/训练

星期三

  1. NVIDIA Developer · Generative AI精选

    NVIDIA 发布 Isaac for Healthcare 医疗物理仿真框架

    NVIDIA 发布开源 GPU 加速的 Medical Physics Simulation 框架，作为 Isaac for Healthcare 的一部分，用于生成解剖数字孪生、模拟器械与解剖结构交互并训练强化学习策略。其中 Endoluminal Simulation Module 已正式可用，用 XPBD 求解器实时模拟柔性导管在患者特异性血管中的导航；Surgical Simulation Module 处于早期访问，可在单张消费级 GPU 上以超过 30 帧每秒模拟软组织变形与触觉反馈。Cosmos-H 系列生成式世界模型可从机器人动作生成手术视频，Cosmos-H-Dreams 经 FlashDreams 推理引擎实现超过 30 帧每秒的交互式仿真。

    atient-specific simulations. Run Cosmos-H Dreams interactive simulation from the GitHub repository with pretrained checkpoints. Download Cosmos-H Dreams model checkpoints from the Hugging Face checkpoint for generative surgical simulation. Powered by NVIDIA Nemotron. AI-generated content may summarize information incompletely. Verify important information. Learn more Unlike autonomous driving or industrial robotics, healthcare robotics can’t rely on internet-scale data collection or unlimited re

    匹配字段正文

    #lab#教程/实践#部署/工程#开源生态

星期四

  1. Together AI Blog

    Together AI 更新开源权重模型推理平台并开启自定义训练内测

    Together AI 发布 Dedicated Model Inference 推理平台更新，支持在同一稳定端点后运行多个部署，并提供 canary、蓝绿和滚动更新及按阈值自动回滚、A/B 与影子流量测试、按 inflight 请求或 GPU 利用率等指标自动扩缩容。平台重建了模型缓存与分发层，内部测试中多个前沿模型的冷启动速度约提升 4 倍，同时提供组织级 Prometheus 端点接入自有可观测工具。Together AI 还开启自定义训练内测，涵盖全权重与 LoRA 强化学习及监督微调，checkpoint 可直接部署到推理服务，内测名额有限。

    e new versions into production. Deploy models from Together’s model platform, or bring your own open-weight or fine-tuned model. You can upload full model weights or adapters from Hugging Face, S3, or your local machine. The platform is designed to support the full lifecycle of a model, from an early fine-tune through the versions that eventually serve production traffic. The model itself is only one part of the deployment. Hardware type, quantization, tensor parallelism, speculative decoding, a

    匹配字段正文

    #company#产品更新#部署/工程#开源生态

星期三

  1. Prime Intellect精选

    Prime Intellect 整合 23 个智能体任务集，提供 36.5 万+ 环境用于 SWE、终端与搜索 RL

    Prime Intellect 将 23 个智能体任务集整合到统一的 taskset API 之下，共约 36.5 万个任务，包括 20 多种语言的约 19.8 万个软件工程任务、约 2.86 万个终端任务和约 13.76 万个搜索任务，可直接用于评测和 RL 训练。整合保留各任务集原有评分路径，统一了配置接口、镜像注册表和完整性标准，评分材料在打分前对智能体隐藏。团队还对数据集做了 gold-patch 与 no-op 验证并重新上传清洗版本，例如 R2E-Gym-Subset-Verified 保留 4,578 行中的 4,522 行。

    Scaling Agentic RL: 365,000+ Environments for SWE, Terminal, and Search

    ges. Every task runs from a prebuilt Prime image and is gold-validated against the upstream solutions All our filtered and verified SWE re-uploads live in the SWE RL collection on Hugging Face, each with a dataset card documenting exactly what changed vs upstream, the validation methodology, and exclusions preserved for audit. Terminal Tasks Terminal-Lego (Yang et al., Huawei / HKU - paper, terminal lego v1) is 13,800 Docker-verified Terminal-Bench-style tasks built by the SWE-Lego team from rea

    匹配字段正文

    #company#开源/仓库#Agent#数据/训练

星期三

  1. LMSYS Blog精选

    SGLang 与 Miles 为多模态模型 Inkling 提供 Day-0 支持

    LMSYS 与 Thinking Machines 团队合作，为 975B 参数多模态模型 Inkling 在 SGLang 和 Miles 上提供 Day-0 支持。SGLang 针对其短卷积、相对位置注意力和共享专家 sink 的 MoE 架构做了专门优化，在 Nvidia Blackwell GPU 上实现最高 71.7k tok/s 输入吞吐和 171.0 tok/s 单用户解码速度，并支持投机解码、PD 分离、多 LoRA 服务与 HiCache。Miles 通过定制 Megatron 后端支持全参数与 LoRA 强化学习，覆盖文本和视觉语言任务。

    attention. ShortConv gathers the full sequence and returns each rank's local slice An Inkling model bridge handles checkpoint conversion in both directions: it shards the released Hugging Face weights for Megatron training, then reconstructs the layouts required by checkpoint export and rollout weight updates. The same backend drives both full-parameter and LoRA RL. Full-Parameter RL Built on the native Inkling backend and parallel stack above, Miles runs full-parameter GRPO by updating every mo

    匹配字段正文

    #company#开源/仓库#多模态#部署/工程
  2. Modal 官方工程博客精选

    Thinking Machines 的 Inkling 上线 Modal，975B 参数多模态模型支持 Day 0 托管

    Thinking Machines 发布通用多模态模型 Inkling，可接受文本、图像和音频输入并生成文本输出，Modal 以 Day 0 合作将其作为按 token 计费的托管端点上线。Inkling 是 MoE Transformer，总参数 975B、激活 41B，上下文窗口 1M token，原生支持音频与视觉。Modal 为其定制了 DFlash 推测解码器，在 8x B200 的智能体负载上实现每用户 250 tokens/s、每 GPU 2.5M TPM，交互性比模型内置推测路径快 67%。

    k diffusion drafter generates a whole block of tokens in one parallel forward pass, conditioned on features from the target model. And in the latest draft models we’ve released on Hugging Face, we’re making a similar bet towards local attention, just like Inkling. We're able to push DFlash in a few ways to maximize speed and compute efficiency (without losing accuracy): All local attention. Every drafter layer uses sliding window attention, with no global attention at all. The drafter can afford

    匹配字段正文

    #company#产品更新#多模态#部署/工程

星期二

  1. LMSYS Blog

    LMSYS 为 Miles RL 框架提供 ROCm 支持，可在 AMD Instinct GPU 上运行

    LMSYS 宣布开源 RL 框架 Miles 支持 ROCm，可在 AMD Instinct MI300/350 系列 GPU 上原生运行分布式 rollout 与 GRPO 训练。Miles 基于 SGLang 与 Slime 生态，将 rollout 生成与 Megatron 模型优化解耦，并提供预装 SGLang 和 Megatron-LM 的 ROCm 容器。在单节点 8 卡 MI300X 上训练 Qwen3-30B-A3B 时，平均单步耗时 388.50s，其中 rollout 生成占 152.79s，AIME 准确率从 0.665（step 19）升至 0.729（step 139）。

    mage tag based on your hardware: The container includes SGLang and Megatron-LM preinstalled. Install Miles and download assets Download the example model and datasets: Convert the Hugging Face checkpoint to Megatron format: Launch RL training This launches a full RL pipeline: Ray cluster initialization; SGLang rollout workers; GRPO training loop; On-policy rollout → train → update cycle Experiments and Performance Agentic task training: multiturn Multi-turn agents are quickly becoming the defaul

    匹配字段正文

    #company#开源/仓库#部署/工程#数据/训练

星期二

  1. Prime Intellect精选

    Prime Intellect 发布 Lab 全栈训练平台

    Prime Intellect 发布 Lab，把 Environments Hub、Hosted Training 和 Hosted Evaluations 整合为面向智能体后训练的全栈平台，覆盖大规模智能体 RL、推理与评测。平台以环境为核心，包含任务数据集、模型 harness 和评分 rubric，当前 Hosted Training 支持基于 prime-rl 的 LoRA 智能体 RL，后续将加入 SFT、DPO 等算法。私有测试期间已完成 3000 多次 RL 运行，Environments Hub 累计有 250 多位创作者构建 1000 多个环境、下载量超过 10 万次。

    Introducing Lab: The Full-Stack Platform for Training your Own Models

    f currently available models for Hosted Training, do: Beyond our own INTELLECT-3 model, many awesome small and large models are available from organizations such as NVIDIA, Arcee, Hugging Face, Allen AI, Z.AI, Alibaba Qwen, and many more launching soon. We've also added experimental multimodality support and have deployed our first models to train on environments with image input. Runs are configured via a simple .toml file: To start a training, run: You can then view logs, training results, and

    匹配字段正文

    #company#产品更新#Agent#数据/训练

星期五

  1. Qwen Blog

    Qwen3-TTS 更新：49 种音色、10 种语言与 9 种方言

    Qwen 团队发布 Qwen3-TTS-Flash 更新，支持 49 种以上音色、10 种语言和 9 种方言的语音合成，可通过 Qwen API 调用。官方称在 MiniMax TTS 多语言测试集上，其平均词错误率低于 MiniMax、ElevenLabs 和 GPT-4o-Audio-Preview。相比上一版本，模型根据文本自适应调整语速和韵律的能力明显提升，更接近真人语音。

    Qwen Blog Publication About Try Qwen Chat Qwen3-TTS Update! 49 Timbres + 10 Languages + 9 Dialects December 5, 2025 · 6 min · 1086 words · Qwen Team | Translations: 简体中文 DASHSCOPE HUGGING FACE DEMO MODELSCOPE DEMO Qwen3-TTS-Flash is a flagship text-to-speech model that supports multi-timbre, multi-lingual, and multi-dialect speech synthesis. It aims to produce natural and expressive speech and is available via Qwen API . Major Improvements: Richer Timbres Support : Qwen3-TTS offers over 49 high-

    匹配字段正文

    #company#模型发布#语音#多模态

星期四

  1. vLLM 官方博客

    vLLM Semantic Router 推进 LLM 智能推理路由

    vLLM Semantic Router 为 vLLM 推理引擎提供开源的意图感知路由层，按查询复杂度选择快速路径或 CoT 推理模式。试验中，该设计使准确率提高约 10%、延迟降低约 50%、token 减少约 50%，并支持通过 Kubernetes 和 Envoy 部署。v0.1 路线图包括语义缓存、多因素路由、性能测试和路由策略可视化。

    determine routing paths; Smart Routing: Simple queries → "fast path" inference; Complex queries → "Chain-of-Thought" reasoning mode; High-Performance Engine: Written in Rust using Hugging Face Candle, it delivers high concurrency and zero-copy inference; Cloud-Native Integration: Works out-of-the-box with Kubernetes and Envoy via the ext proc plugin In trials, this design yielded: 10% higher accuracy; 50% lower latency; 50% fewer tokens In business and economics domains, gains exceeded 20% accur

    匹配字段正文

    #engineering#产品更新#推理#部署/工程

星期二

  1. vLLM 官方博客

    vLLM 新增对 gpt-oss 的支持

    vLLM 现已支持在 NVIDIA Blackwell、Hopper 以及 AMD MI300x、MI355x GPU 上运行 gpt-oss。其针对 MXFP4 MoE、高效注意力和混合 KV cache 集成了专用内核与内存管理方案，并通过 OpenAI Responses API、gpt-oss toolkit 或外部 MCP 工具服务器支持工具调用。

    . The AMD team contributed significantly to the integration of the model on their devices: Hongxia Yang, Ali Zaidy, with great support from Peng Sun, Vinayak Gokhale, Andy Luo The Hugging Face team continues to be amazing at building an open source ecosystem: Lysandre, Hugo, Marc, vb, Arthur, Mohamed, Andrien. Finally, we want to thank all the partners that leveraged vLLM in some way and delivered valuable feedback and improvements to this effort: AWS, Cloudflare, Snowflake, Databricks, Together

    匹配字段正文

    #engineering#产品更新#部署/工程#MCP/工具调用

星期五

  1. Prime Intellect精选

    Prime Intellect 启动 INTELLECT-1：首个全球分布式训练的 10B 参数模型

    Prime Intellect 发布 INTELLECT-1，称其为首个在全球分布式节点上训练的 10B 参数模型，并开放算力贡献入口。该模型基于 Llama-3 架构，训练数据混合 55% Fineweb-edu、20% DLCM、20% Stack v2 和 5% OpenWebMath，经 Llama-3 tokenizer 处理后超过 6 万亿 token。配套发布的分布式训练框架 Prime 支持容错训练、节点动态上下线和 int8 伪梯度通信，官方称生产训练中每 100 步同步一次，通信仅占总训练时间的 1-2%。

    INTELLECT–1: Launching the First Globally-Distributed Training of a 10B Parameter Model

    ep to ensure AGI will be open-source, transparent, and accessible. Launch Partners and Contributors We are excited and grateful to be joined by leading open-source AI players like Hugging Face, SemiAnalysis, Arcee, Hyperbolic, Olas, Akash, Schelling AI and many others contributing compute to this training run. Anyone can now contribute resources to advance open-source AI through our platform and later on also more easily with their own hardware. Paradigm Shift for Distributed Training. As Jack C

    匹配字段正文

    #company#模型发布#数据/训练#开源生态