[{"data":1,"prerenderedAt":375},["ShallowReactive",2],{"content-query-OOU3KxCmOk":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"body":13,"_type":369,"_id":370,"_source":371,"_file":372,"_stem":373,"_extension":374},"/news/zh/1075","zh",false,"","本周Parallel SIG分享会——MindSpore的并行策略详解，尽在周六晚！","本周六晚走起！","2022-03-16","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/03/17/4c3be6a72c484ae78e57442fd53eaf32.png","news",{"type":14,"children":15,"toc":366},"root",[16,24,33,42,52,57,67,75,87,92,103,114,130,138,146,154,159,167,177,185,193,200,205,212,221,229,237,244,249,256,265,273,278,285,292,300,310,325,340,350],{"type":17,"tag":18,"props":19,"children":21},"element","h1",{"id":20},"本周parallel-sig分享会mindspore的并行策略详解尽在周六晚",[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/2022/03/17/a8a30b4d60bf40e7ba454999ab7ff050.gif",[],{"type":17,"tag":25,"props":34,"children":35},{},[36],{"type":17,"tag":37,"props":38,"children":39},"strong",{},[40],{"type":23,"value":41},"Parallel SIG",{"type":17,"tag":25,"props":43,"children":44},{},[45,47],{"type":23,"value":46},"近年来，神经网络模型的规模持续增长，从千亿突破至万亿，甚至朝着十万亿乃至百万亿的方向发展。",{"type":17,"tag":37,"props":48,"children":49},{},[50],{"type":23,"value":51},"如此大规模的参数量，单块加速设备的算力和存储能力完全无法满足需求，因此利用分布式加速大模型的训练和推理是一种趋势。",{"type":17,"tag":25,"props":53,"children":54},{},[55],{"type":23,"value":56},"而大模型的训练往往需要成百上千张加速芯片来完成任务，比如华为使用上千张Ascend 910训练PanGu-alpha。**那如何利用计算中心成百上千张AI加速芯片高效地进行训练，并行是其中的核心技术。**并行技术提升AI加速芯片的利用率，让大模型的训练时间被约束在可接受范围内。",{"type":17,"tag":25,"props":58,"children":59},{},[60,65],{"type":17,"tag":37,"props":61,"children":62},{},[63],{"type":23,"value":64},"并行策略",{"type":23,"value":66},"作为MindSpore的关键特性，被业界广泛使用。此次Parallel SIG（special interest groups）会议，我们将一起讨论与MindSpore并行策略相关的课题。",{"type":17,"tag":25,"props":68,"children":69},{},[70],{"type":17,"tag":37,"props":71,"children":72},{},[73],{"type":23,"value":74},"会议详情",{"type":17,"tag":25,"props":76,"children":77},{},[78,80,85],{"type":23,"value":79},"01 ",{"type":17,"tag":37,"props":81,"children":82},{},[83],{"type":23,"value":84},"会议时间",{"type":23,"value":86},"：",{"type":17,"tag":25,"props":88,"children":89},{},[90],{"type":23,"value":91},"2022年**3月19日（周六）**19:00-21:00",{"type":17,"tag":25,"props":93,"children":94},{},[95,97,102],{"type":23,"value":96},"02 ",{"type":17,"tag":37,"props":98,"children":99},{},[100],{"type":23,"value":101},"会议链接",{"type":23,"value":86},{"type":17,"tag":25,"props":104,"children":105},{},[106],{"type":17,"tag":107,"props":108,"children":112},"a",{"href":109,"rel":110},"https://meeting.tencent.com/dm/3uSnNqriRHZ4",[111],"nofollow",[113],{"type":23,"value":109},{"type":17,"tag":25,"props":115,"children":116},{},[117,119,124,125],{"type":23,"value":118},"03 ",{"type":17,"tag":37,"props":120,"children":121},{},[122],{"type":23,"value":123},"会议ID",{"type":23,"value":86},{"type":17,"tag":37,"props":126,"children":127},{},[128],{"type":23,"value":129},"245-887-799",{"type":17,"tag":25,"props":131,"children":132},{},[133],{"type":17,"tag":37,"props":134,"children":135},{},[136],{"type":23,"value":137},"议题一",{"type":17,"tag":25,"props":139,"children":140},{},[141],{"type":17,"tag":37,"props":142,"children":143},{},[144],{"type":23,"value":145},"MindSpore并行策略之sharding propagation",{"type":17,"tag":25,"props":147,"children":148},{},[149],{"type":17,"tag":37,"props":150,"children":151},{},[152],{"type":23,"value":153},"01 时间",{"type":17,"tag":25,"props":155,"children":156},{},[157],{"type":23,"value":158},"19:00-19:40",{"type":17,"tag":25,"props":160,"children":161},{},[162],{"type":17,"tag":37,"props":163,"children":164},{},[165],{"type":23,"value":166},"02 演讲人",{"type":17,"tag":25,"props":168,"children":169},{},[170,175],{"type":17,"tag":37,"props":171,"children":172},{},[173],{"type":23,"value":174},"达叔",{"type":23,"value":176},"，MindSpore并行组核心开发成员",{"type":17,"tag":25,"props":178,"children":179},{},[180],{"type":17,"tag":37,"props":181,"children":182},{},[183],{"type":23,"value":184},"议题二",{"type":17,"tag":25,"props":186,"children":187},{},[188],{"type":17,"tag":37,"props":189,"children":190},{},[191],{"type":23,"value":192},"MindSpore算子级并行用法",{"type":17,"tag":25,"props":194,"children":195},{},[196],{"type":17,"tag":37,"props":197,"children":198},{},[199],{"type":23,"value":153},{"type":17,"tag":25,"props":201,"children":202},{},[203],{"type":23,"value":204},"19:40-20:20",{"type":17,"tag":25,"props":206,"children":207},{},[208],{"type":17,"tag":37,"props":209,"children":210},{},[211],{"type":23,"value":166},{"type":17,"tag":25,"props":213,"children":214},{},[215,220],{"type":17,"tag":37,"props":216,"children":217},{},[218],{"type":23,"value":219},"大总管",{"type":23,"value":176},{"type":17,"tag":25,"props":222,"children":223},{},[224],{"type":17,"tag":37,"props":225,"children":226},{},[227],{"type":23,"value":228},"议题三",{"type":17,"tag":25,"props":230,"children":231},{},[232],{"type":17,"tag":37,"props":233,"children":234},{},[235],{"type":23,"value":236},"基于MindSpore训练MoE",{"type":17,"tag":25,"props":238,"children":239},{},[240],{"type":17,"tag":37,"props":241,"children":242},{},[243],{"type":23,"value":153},{"type":17,"tag":25,"props":245,"children":246},{},[247],{"type":23,"value":248},"20:20-21:00",{"type":17,"tag":25,"props":250,"children":251},{},[252],{"type":17,"tag":37,"props":253,"children":254},{},[255],{"type":23,"value":166},{"type":17,"tag":25,"props":257,"children":258},{},[259,264],{"type":17,"tag":37,"props":260,"children":261},{},[262],{"type":23,"value":263},"王博",{"type":23,"value":176},{"type":17,"tag":25,"props":266,"children":267},{},[268],{"type":17,"tag":37,"props":269,"children":270},{},[271],{"type":23,"value":272},"加入Parallel SIG",{"type":17,"tag":25,"props":274,"children":275},{},[276],{"type":23,"value":277},"扫描二维码",{"type":17,"tag":25,"props":279,"children":280},{},[281],{"type":17,"tag":29,"props":282,"children":284},{"alt":7,"src":283},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/03/17/0c00da9a1c6549bcbc61627e7ea8d188.jpg",[],{"type":17,"tag":25,"props":286,"children":287},{},[288],{"type":17,"tag":29,"props":289,"children":291},{"alt":7,"src":290},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/03/17/2715c402afd34d018ebd96e874299adb.jpg",[],{"type":17,"tag":25,"props":293,"children":294},{},[295],{"type":17,"tag":37,"props":296,"children":297},{},[298],{"type":23,"value":299},"MindSpore官方资料",{"type":17,"tag":25,"props":301,"children":302},{},[303,308],{"type":17,"tag":37,"props":304,"children":305},{},[306],{"type":23,"value":307},"官方QQ群",{"type":23,"value":309}," : 486831414",{"type":17,"tag":25,"props":311,"children":312},{},[313,318,319],{"type":17,"tag":37,"props":314,"children":315},{},[316],{"type":23,"value":317},"官网",{"type":23,"value":86},{"type":17,"tag":107,"props":320,"children":323},{"href":321,"rel":322},"https://www.mindspore.cn/",[111],[324],{"type":23,"value":321},{"type":17,"tag":25,"props":326,"children":327},{},[328,333,334],{"type":17,"tag":37,"props":329,"children":330},{},[331],{"type":23,"value":332},"论坛",{"type":23,"value":86},{"type":17,"tag":107,"props":335,"children":338},{"href":336,"rel":337},"https://bbs.huaweicloud.com/forum/forum-1076-1.html",[111],[339],{"type":23,"value":336},{"type":17,"tag":25,"props":341,"children":342},{},[343,348],{"type":17,"tag":37,"props":344,"children":345},{},[346],{"type":23,"value":347},"Gitee",{"type":23,"value":349}," : https : //gitee.com/mindspore/mindspore",{"type":17,"tag":25,"props":351,"children":352},{},[353,358,360],{"type":17,"tag":37,"props":354,"children":355},{},[356],{"type":23,"value":357},"GitHub",{"type":23,"value":359}," : ",{"type":17,"tag":107,"props":361,"children":364},{"href":362,"rel":363},"https://github.com/mindspore-ai/mindspore",[111],[365],{"type":23,"value":362},{"title":7,"searchDepth":367,"depth":367,"links":368},4,[],"markdown","content:news:zh:1075.md","content","news/zh/1075.md","news/zh/1075","md",1776506047702]