[{"data":1,"prerenderedAt":488},["ShallowReactive",2],{"content-query-ROsC6SPZoF":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"body":13,"_type":482,"_id":483,"_source":484,"_file":485,"_stem":486,"_extension":487},"/news/zh/1957","zh",false,"","重磅！Canonical加入MindSpore社区，带来首个预集成国内AI开源框架的Charmed Kubeflow新版本","Canonical发布了其名下Kubeflow企业版端到端MLOps平台Charmed Kubeflow的1.6新版本，正式支持开源深度学习框架昇思MindSpore，并签署CCLA协议，加入昇思MindSpore社区。","2022-11-17","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/11/27/45a7bb68d2cf4f76b4923e66904880d5.png","news",{"type":14,"children":15,"toc":479},"root",[16,24,33,38,51,58,65,70,75,83,88,93,101,106,111,116,124,132,137,147,152,157,165,170,178,194,204,209,214,219,227,232,240,245,250,258,266,271,278,283,290,295,300,307,312,320,330,337,342,350,361,366,377,386,397,404,412,422,438,448,464],{"type":17,"tag":18,"props":19,"children":21},"element","h1",{"id":20},"重磅canonical加入mindspore社区带来首个预集成国内ai开源框架的charmed-kubeflow新版本",[22],{"type":23,"value":8},"text",{"type":17,"tag":25,"props":26,"children":27},"p",{},[28],{"type":17,"tag":29,"props":30,"children":32},"img",{"alt":7,"src":31},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/11/27/1acebfaa5f0e41aabf3f4dc13a7b5e6c.gif",[],{"type":17,"tag":25,"props":34,"children":35},{},[36],{"type":23,"value":37},"**2022年11月8日至9日，Open Source Experience(OSXP 2022)大会在法国巴黎拉开了帷幕。**OSXP是欧洲开源软件行业的年度重要会议, 汇集4500多名专业人士，本次会议深入探讨了AI/ML、数字孪生和区块链等主题。",{"type":17,"tag":25,"props":39,"children":40},{},[41,43,49],{"type":23,"value":42},"大会期间作为著名Linux发行版Ubuntu的发行商Canonical，发布了其名下Kubeflow企业版端到端MLOps平台",{"type":17,"tag":44,"props":45,"children":46},"strong",{},[47],{"type":23,"value":48},"Charmed Kubeflow的1.6新版本，正式支持开源深度学习框架昇思MindSpore，并签署CCLA协议",{"type":23,"value":50},"，加入昇思MindSpore社区。这意味着昇思MindSpore全球化拓展的新里程碑，同时也显示出Charmed Kubeflow生态圈的持续成长与成熟。",{"type":17,"tag":25,"props":52,"children":53},{},[54],{"type":17,"tag":29,"props":55,"children":57},{"alt":7,"src":56},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/11/27/db5d6544e9aa4c019d9f46147e126c8c.png",[],{"type":17,"tag":25,"props":59,"children":60},{},[61],{"type":17,"tag":29,"props":62,"children":64},{"alt":7,"src":63},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/11/27/1e61cd18a9874772801b07516c8368eb.png",[],{"type":17,"tag":25,"props":66,"children":67},{},[68],{"type":23,"value":69},"Canonical公司由企业家马克·沙特尔沃思创建，支持并创建了开源软件（FOSS）和一些旨在改善自由软件开发者和贡献者的工具，其最知名的产品为Linux发行版Ubuntu，为开源社区的发展发挥了重要作用。Canonical的使命是“让开源软件可供各个地方的人使用。我们认为，促进创新的最佳之道就是为创新者提供他们所需的技术。”",{"type":17,"tag":25,"props":71,"children":72},{},[73],{"type":23,"value":74},"昇思MindSpore是华为开源的新一代全场景AI框架，支持端、边、云全场景灵活部署，开创全新的AI编程范式，降低AI开发门槛，旨在实现开发友好、运行高效、部署灵活三大目标，同时着力构筑面向全球的人工智能开源社区，推动人工智能软硬件应用生态繁荣发展。",{"type":17,"tag":25,"props":76,"children":77},{},[78],{"type":17,"tag":44,"props":79,"children":80},{},[81],{"type":23,"value":82},"何为Charmed Kubeflow？",{"type":17,"tag":25,"props":84,"children":85},{},[86],{"type":23,"value":87},"企业级端到端MLOps平台Charmed Kubeflow可以将数据科学工作流中的各个步骤转换成Kubernetes任务。Charmed Kubeflow是Kubeflow上游项目的官方发行版，为数据科学家和机器学习工程师提供简单、轻便、可扩展的机器学习Deployment。",{"type":17,"tag":25,"props":89,"children":90},{},[91],{"type":23,"value":92},"提供构成KubeFlow最新版本的30多个应用程序和服务，并且让Kubeflow的部署更快，更简单。Charmed Kubeflow应用广泛，如使用Notebooks进行实验、利用Kubeflow流水线进行训练以及使用Katib进行调优。",{"type":17,"tag":25,"props":94,"children":95},{},[96],{"type":17,"tag":44,"props":97,"children":98},{},[99],{"type":23,"value":100},"为何选择MindSpore?",{"type":17,"tag":25,"props":102,"children":103},{},[104],{"type":23,"value":105},"Kubeflow与MindSpore的原生集成为模型开发、执行和部署提供统一接口以及端到端AI能力。MindSpore提供AI原生执行模式，帮助Kubeflow充分利用昇腾AI硬件算力。",{"type":17,"tag":25,"props":107,"children":108},{},[109],{"type":23,"value":110},"MindSpore针对华为昇腾/GPU/CPU等硬件进行深度优化，支持云上、边缘和移动场景的快速部署，工程师可针对不同应用场景开发多种不同方案。MindSpore在基于Web的交互式计算平台Jupyter Notebook上运行。在创建Notebook时，工程师可从默认的Jupyter Lab镜像列表中选择MindSpore镜像，分配一定的CPU资源，启动一个可以运行MindSpore及MindSpore Vision套件代码的Notebook实例，进行机器学习实验。",{"type":17,"tag":25,"props":112,"children":113},{},[114],{"type":23,"value":115},"通过与MindSpore的原生集成和与华为在开源项目上的合作，Canonical为其MLOps生态系统提供了越来越丰富的工具集。Charmed Kubeflow集成多种AI/ML工具，包括提供中央模型注册表的MLFlow以及方便数据流处理的Spark。而与MindSpore的集成则能提供统一编程、统一算子、模型自动分区以及动静态计算图等多种方案特性。",{"type":17,"tag":25,"props":117,"children":118},{},[119],{"type":17,"tag":44,"props":120,"children":121},{},[122],{"type":23,"value":123},"如何在Kubeflow上使用MindSpore?",{"type":17,"tag":25,"props":125,"children":126},{},[127],{"type":17,"tag":44,"props":128,"children":129},{},[130],{"type":23,"value":131},"1、安装配置Charmed Kubeflow",{"type":17,"tag":25,"props":133,"children":134},{},[135],{"type":23,"value":136},"安装Charmed Kubeflow需用到Ubuntu 20.04或更高版本的系统，要求至少32G可用内存以及50G硬盘空间。部署过程中，同时安装MicroK8s和Juju。当Charmed Kubeflow部署完成，所有Kubeflow组件激活后，执行以下命令升级jupyter-ui为最新版本，以获得MindSpore Notebook镜像。",{"type":17,"tag":138,"props":139,"children":141},"pre",{"code":140},"juju refresh jupyter-ui --channel=latest/edge \n",[142],{"type":17,"tag":143,"props":144,"children":145},"code",{"__ignoreMap":7},[146],{"type":23,"value":140},{"type":17,"tag":25,"props":148,"children":149},{},[150],{"type":23,"value":151},"Kubeflow安装后，执行以下命令进行配置。",{"type":17,"tag":25,"props":153,"children":154},{},[155],{"type":23,"value":156},"配置接入地址：",{"type":17,"tag":138,"props":158,"children":160},{"code":159},"Juju config dex-auth public-url=http://xxx.xxx.xxx..xxx.nip.io\n\njuju config oidc-gatekeeper public url=http://xxx.xxx.xxx..xxx.nip.io\n",[161],{"type":17,"tag":143,"props":162,"children":163},{"__ignoreMap":7},[164],{"type":23,"value":159},{"type":17,"tag":25,"props":166,"children":167},{},[168],{"type":23,"value":169},"配置接入Kubeflow的用户名及密码：",{"type":17,"tag":138,"props":171,"children":173},{"code":172},"juju config dex-auth static-username=admin\n\njuju config dex-auth static-password=****\n",[174],{"type":17,"tag":143,"props":175,"children":176},{"__ignoreMap":7},[177],{"type":23,"value":172},{"type":17,"tag":25,"props":179,"children":180},{},[181,183,192],{"type":23,"value":182},"登录",{"type":17,"tag":184,"props":185,"children":189},"a",{"href":186,"rel":187},"http://xxx.xxx.xxx..xxx.nip.io%EF%BC%8C%E8%BE%93%E5%85%A5%E8%AE%BE%E7%BD%AE%E7%9A%84%E5%AF%86%E7%A0%81%E6%8E%A5%E5%85%A5%E5%AE%8C%E6%95%B4%E7%9A%84Kubeflow",[188],"nofollow",[190],{"type":23,"value":191},"http://xxx.xxx.xxx..xxx.nip.io，输入设置的密码接入完整的Kubeflow",{"type":23,"value":193}," 仪表盘(Dashboard)。",{"type":17,"tag":25,"props":195,"children":196},{},[197,199],{"type":23,"value":198},"**2、**",{"type":17,"tag":44,"props":200,"children":201},{},[202],{"type":23,"value":203},"新建MicroK8s集群",{"type":17,"tag":25,"props":205,"children":206},{},[207],{"type":23,"value":208},"由于Charmed Kubeflow只能部署在x86环境，集成MindSpore的Charmed Kubeflow需要一个包含两个节点的Kubernetes集群：x86节点用来部署Charmed Kubeflow, ARM节点（最好带有Ascend AI处理器）用来调动工作节点。（本节所述步骤为MicroK8s独有。）",{"type":17,"tag":25,"props":210,"children":211},{},[212],{"type":23,"value":213},"x86节点作为主节点使用，支持Charmed Kubeflow Deployment和Kubernetes控制面；ARM节点作为备节点使用，支持工作节点调度。",{"type":17,"tag":25,"props":215,"children":216},{},[217],{"type":23,"value":218},"在主备节点上安装 MicroK8s 1.22。",{"type":17,"tag":138,"props":220,"children":222},{"code":221},"sudo snap install microk8s --channel 1.22/stable –classic\n",[223],{"type":17,"tag":143,"props":224,"children":225},{"__ignoreMap":7},[226],{"type":23,"value":221},{"type":17,"tag":25,"props":228,"children":229},{},[230],{"type":23,"value":231},"添加备节点至Kubernetes集群中，在主节点上执行如下命令：",{"type":17,"tag":138,"props":233,"children":235},{"code":234},"microk8s add-node\n",[236],{"type":17,"tag":143,"props":237,"children":238},{"__ignoreMap":7},[239],{"type":23,"value":234},{"type":17,"tag":25,"props":241,"children":242},{},[243],{"type":23,"value":244},"注：会返回部分连接指令。在目标集群的MicroK8s实例上执行这些指令，无须在执行add-node命令的节点上执行。执行microk8s kubectl get no查询最近添加的节点。详细信息参见“添加节点”。",{"type":17,"tag":25,"props":246,"children":247},{},[248],{"type":23,"value":249},"给备节点打上标签，在主节点上执行如下命令：",{"type":17,"tag":138,"props":251,"children":253},{"code":252},"microk8s kubectl label nodes  kubernetes.io/arch=arm64\n",[254],{"type":17,"tag":143,"props":255,"children":256},{"__ignoreMap":7},[257],{"type":23,"value":252},{"type":17,"tag":25,"props":259,"children":260},{},[261],{"type":17,"tag":44,"props":262,"children":263},{},[264],{"type":23,"value":265},"3、 运行MindSpore支持的Notebook",{"type":17,"tag":25,"props":267,"children":268},{},[269],{"type":23,"value":270},"在Charmed Kubeflow中接入MindSpore，需使用本机激活的Notebook。在Charmed Kubeflow仪表盘上的Notebook列表中，选择“创建Notebook”（Create Notebook）,输入Notebook名称“MindSpore”,镜像选择“jupyterlab”，在列表中选择“mindspore/jupyter-mindspore”，分配合适的CPU和内容资源，最后点击“创建”（Create）。",{"type":17,"tag":25,"props":272,"children":273},{},[274],{"type":17,"tag":29,"props":275,"children":277},{"alt":7,"src":276},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/11/27/948daf23cb0e4167b9bc5d25f94d86a5.png",[],{"type":17,"tag":25,"props":279,"children":280},{},[281],{"type":23,"value":282},"在Notebook界面，很快就能看到创建完成的提示。",{"type":17,"tag":25,"props":284,"children":285},{},[286],{"type":17,"tag":29,"props":287,"children":289},{"alt":7,"src":288},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/11/27/f8e8e2921ae64495968ef27f67afe9e1.png",[],{"type":17,"tag":25,"props":291,"children":292},{},[293],{"type":23,"value":294},"点击“连接”（Connect），在另一个浏览器tab页面打开Notebook。这里我们可以使用MindSpore官网提供的Notebook来上传手写数字识别Notebook。",{"type":17,"tag":25,"props":296,"children":297},{},[298],{"type":23,"value":299},"点击“执行”（Execute）, 这个Notebook就能直接运行，不用额外去安装MindSpore及Vision套件。",{"type":17,"tag":25,"props":301,"children":302},{},[303],{"type":17,"tag":29,"props":304,"children":306},{"alt":7,"src":305},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/11/27/a019a53c23e84968aa3664346e5eb304.png",[],{"type":17,"tag":25,"props":308,"children":309},{},[310],{"type":23,"value":311},"一旦MindSpore镜像在你所创建的Charmed Kubeflow Notebook中激活，便能开启AI/ML项目啦！MindSpore支持多种用例，计算机视觉会是个不错的选择。",{"type":17,"tag":25,"props":313,"children":314},{},[315],{"type":17,"tag":44,"props":316,"children":317},{},[318],{"type":23,"value":319},"生态合作",{"type":17,"tag":25,"props":321,"children":322},{},[323,325],{"type":23,"value":324},"2022年11月，",{"type":17,"tag":44,"props":326,"children":327},{},[328],{"type":23,"value":329},"Canonical Limited 签署CCLA（Corporate Contributor License Agreement，企业贡献者许可协议书），正式加入昇思MindSpore开源社区。",{"type":17,"tag":25,"props":331,"children":332},{},[333],{"type":17,"tag":29,"props":334,"children":336},{"alt":7,"src":335},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/11/27/0946a7ab6ad349e6b51268261967eabf.png",[],{"type":17,"tag":25,"props":338,"children":339},{},[340],{"type":23,"value":341},"未来，Canonical 将进一步繁荣昇思生态，深化双方社区项目合作，共同推动开源生态发展，构建繁荣的人工智能产业生态圈。",{"type":17,"tag":25,"props":343,"children":344},{},[345],{"type":17,"tag":44,"props":346,"children":347},{},[348],{"type":23,"value":349},"参考链接",{"type":17,"tag":25,"props":351,"children":352},{},[353,355],{"type":23,"value":354},"[1]",{"type":17,"tag":184,"props":356,"children":359},{"href":357,"rel":358},"https://ubuntu.com/blog/open-source-mlops-ai-framework",[188],[360],{"type":23,"value":357},{"type":17,"tag":25,"props":362,"children":363},{},[364],{"type":23,"value":365},"[2]Hand's on guide：",{"type":17,"tag":25,"props":367,"children":368},{},[369,375],{"type":17,"tag":184,"props":370,"children":373},{"href":371,"rel":372},"https://ubuntu.com/blog/kubeflow-mindspore-integration-guide",[188],[374],{"type":23,"value":371},{"type":23,"value":376}," [3]Tutorial:",{"type":17,"tag":25,"props":378,"children":379},{},[380],{"type":17,"tag":184,"props":381,"children":384},{"href":382,"rel":383},"https://charmed-kubeflow.io/docs/mindspore",[188],[385],{"type":23,"value":382},{"type":17,"tag":25,"props":387,"children":388},{},[389,391],{"type":23,"value":390},"[4]",{"type":17,"tag":184,"props":392,"children":395},{"href":393,"rel":394},"https://zhuanlan.zhihu.com/p/568398333",[188],[396],{"type":23,"value":393},{"type":17,"tag":25,"props":398,"children":399},{},[400],{"type":17,"tag":29,"props":401,"children":403},{"alt":7,"src":402},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/10/24/f5482b1c286f4581b7f3deb4b69c8d1a.jpg",[],{"type":17,"tag":25,"props":405,"children":406},{},[407],{"type":17,"tag":44,"props":408,"children":409},{},[410],{"type":23,"value":411},"MindSpore官方资料",{"type":17,"tag":25,"props":413,"children":414},{},[415,420],{"type":17,"tag":44,"props":416,"children":417},{},[418],{"type":23,"value":419},"官方QQ群",{"type":23,"value":421}," : 486831414",{"type":17,"tag":25,"props":423,"children":424},{},[425,430,432],{"type":17,"tag":44,"props":426,"children":427},{},[428],{"type":23,"value":429},"官网",{"type":23,"value":431},"：",{"type":17,"tag":184,"props":433,"children":436},{"href":434,"rel":435},"https://www.mindspore.cn/",[188],[437],{"type":23,"value":434},{"type":17,"tag":25,"props":439,"children":440},{},[441,446],{"type":17,"tag":44,"props":442,"children":443},{},[444],{"type":23,"value":445},"Gitee",{"type":23,"value":447}," : https : //gitee.com/mindspore/mindspore",{"type":17,"tag":25,"props":449,"children":450},{},[451,456,458],{"type":17,"tag":44,"props":452,"children":453},{},[454],{"type":23,"value":455},"GitHub",{"type":23,"value":457}," : ",{"type":17,"tag":184,"props":459,"children":462},{"href":460,"rel":461},"https://github.com/mindspore-ai/mindspore",[188],[463],{"type":23,"value":460},{"type":17,"tag":25,"props":465,"children":466},{},[467,472,473],{"type":17,"tag":44,"props":468,"children":469},{},[470],{"type":23,"value":471},"论坛",{"type":23,"value":431},{"type":17,"tag":184,"props":474,"children":477},{"href":475,"rel":476},"https://bbs.huaweicloud.com/forum/forum-1076-1.html",[188],[478],{"type":23,"value":475},{"title":7,"searchDepth":480,"depth":480,"links":481},4,[],"markdown","content:news:zh:1957.md","content","news/zh/1957.md","news/zh/1957","md",1776506057247]