[{"data":1,"prerenderedAt":223},["ShallowReactive",2],{"content-query-YdedymzTHJ":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"category":13,"body":14,"_type":217,"_id":218,"_source":219,"_file":220,"_stem":221,"_extension":222},"/technology-blogs/zh/1665","zh",false,"","【MindSpore易点通】官网使用指南<3>资源和基础模型","本月昇思MindSpore又又又又又更新啦！已经是V1.8版本了，大家快来体验一下！","2022-07-28","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2022/08/02/3e9ea4db487a404bb3aee2da4786cf5d.png","technology-blogs","基础知识",{"type":15,"children":16,"toc":214},"root",[17,26,32,37,50,55,64,73,78,83,92,97,102,107,112,117,122,130,135,148,161,166,174,179,184,189,194,199,204,209],{"type":18,"tag":19,"props":20,"children":22},"element","h1",{"id":21},"mindspore易点通官网使用指南3资源和基础模型",[23],{"type":24,"value":25},"text","【MindSpore易点通】官网使用指南\u003C3>资源和基础模型",{"type":18,"tag":27,"props":28,"children":29},"p",{},[30],{"type":24,"value":31},"本文时间：2022年7月",{"type":18,"tag":27,"props":33,"children":34},{},[35],{"type":24,"value":36},"基于当前v1.8版本",{"type":18,"tag":27,"props":38,"children":39},{},[40,42],{"type":24,"value":41},"官网地址：",{"type":18,"tag":43,"props":44,"children":48},"a",{"href":45,"rel":46},"https://mindspore.cn/",[47],"nofollow",[49],{"type":24,"value":45},{"type":18,"tag":27,"props":51,"children":52},{},[53],{"type":24,"value":54},"大家好，本篇是官网使用指南系列的第3篇啦，本次我们分享的是资源和基础模型板块，往期篇章请看这里：",{"type":18,"tag":27,"props":56,"children":57},{},[58],{"type":18,"tag":43,"props":59,"children":62},{"href":60,"rel":61},"https://bbs.huaweicloud.com/forum/thread-192262-1-1.html",[47],[63],{"type":24,"value":60},{"type":18,"tag":27,"props":65,"children":66},{},[67],{"type":18,"tag":43,"props":68,"children":71},{"href":69,"rel":70},"https://bbs.huaweicloud.com/forum/thread-193766-1-1.html",[47],[72],{"type":24,"value":69},{"type":18,"tag":27,"props":74,"children":75},{},[76],{"type":24,"value":77},"对于文中有需要补充说明的部分，您也可以在下方留言哦~~",{"type":18,"tag":19,"props":79,"children":81},{"id":80},"资源",[82],{"type":24,"value":80},{"type":18,"tag":27,"props":84,"children":85},{},[86],{"type":18,"tag":87,"props":88,"children":91},"img",{"alt":89,"src":90},"资源概述.PNG","https://bbs-img.huaweicloud.com/data/forums/attachment/forum/20227/28/1658937854390899458.PNG",[],{"type":18,"tag":27,"props":93,"children":94},{},[95],{"type":24,"value":96},"本版块主要是提供工具、课程、书籍、视频等资源供大家开始MindSpore学习。",{"type":18,"tag":27,"props":98,"children":99},{},[100],{"type":24,"value":101},"特别要提一下的是，v1.8版本新增了拓展工具MindSpore Dev Toolkit，它是一个面向MindSpore开发者的开发套件。提供包括工程管理、安装环境、智能代码补全、智能知识搜索、算子互搜等一系列全方位辅助功能。全面提升MindSpore框架的易用性。还有其他工具如算子对照表和Mindconverter。其中算子对照表当前提供PyTorch、TensorFlow与MindSpore算子的API对照信息。方便您快速找到已有网络算子的替代方案快速将网络迁移到我们的MindSpore上。而MindConverter是一款模型迁移工具，可将PyTorch(ONNX)或Tensorflow(PB)模型快速迁移到MindSpore框架下使用。模型文件（ONNX/PB）包含网络模型结构（network）与权重信息（weights），迁移后将生成MindSpore框架下的模型定义脚本（model.py）与权重文件（ckpt）。",{"type":18,"tag":27,"props":103,"children":104},{},[105],{"type":24,"value":106},"提供的模型与样例包含Model Zoo和预训练模型（Hub）",{"type":18,"tag":27,"props":108,"children":109},{},[110],{"type":24,"value":111},"Model Zoo其中为您提供可直接调用的昇思MindSpore支持的网络，包括LeNet、LSTM、BERT、ResNet等，避免您重复进行定义，提升开发效率。可以通过import mindspore.model_zoo.xxx等直接调用典型的神经网络。",{"type":18,"tag":27,"props":113,"children":114},{},[115],{"type":24,"value":116},"而MindSpore Hub中，官方对于典型的大型网络提供已训练好的网络模型文件，用户可以用来进行模型推理、微调、迁移学习等，避免花费大量算力和时间重头训练。",{"type":18,"tag":27,"props":118,"children":119},{},[120],{"type":24,"value":121},"课程认证中，包含了初中高级课程资料，视频，PDF等形式方便您的学习！无论您是哪个级别的开发者，都能从中有所收获。",{"type":18,"tag":27,"props":123,"children":124},{},[125],{"type":18,"tag":87,"props":126,"children":129},{"alt":127,"src":128},"选择课程.PNG","https://bbs-img.huaweicloud.com/data/forums/attachment/forum/20227/28/1658937875851738827.PNG",[],{"type":18,"tag":27,"props":131,"children":132},{},[133],{"type":24,"value":134},"还提供不同级别工程师职业认证，通过考试获得证书，让自己脱颖而出。",{"type":18,"tag":27,"props":136,"children":137},{},[138,140,146],{"type":24,"value":139},"应用案例提供企业、高校和开发者的众多开发案例。您可以借此了解企业、高校、开发者如何使用升思MindSpore解决问题、技术研究、开发应用等。其中小编比较印象深刻的、也是之前推过一个有趣的开发者案例，对于小白来说非常友好的《AI诗人》",{"type":18,"tag":43,"props":141,"children":144},{"href":142,"rel":143},"https://zhuanlan.zhihu.com/p/254937902",[47],[145],{"type":24,"value":142},{"type":24,"value":147},"。大家可以闲暇的时候体验一下。",{"type":18,"tag":27,"props":149,"children":150},{},[151,153,159],{"type":24,"value":152},"当然，顶流的学术论文我们也是有整理和汇总的，均发布在资源->学术论文板块中，想要充电请访问",{"type":18,"tag":43,"props":154,"children":157},{"href":155,"rel":156},"https://www.mindspore.cn/resources/papers",[47],[158],{"type":24,"value":155},{"type":24,"value":160},"。",{"type":18,"tag":27,"props":162,"children":163},{},[164],{"type":24,"value":165},"最后！超丰富奖金来啦！虽说搞深度学习是一门艺术，但艺术家也是要吃饭的哈哈。金钱的滋味真的不亚于寒冬腊月风雪夜归后的一碗热馄饨汤。我们也提供了丰富的竞赛活动，当前有很多也在进行中！了解了MindSpore之后，不妨加入其中挑战一下自己吧！",{"type":18,"tag":27,"props":167,"children":168},{},[169],{"type":18,"tag":87,"props":170,"children":173},{"alt":171,"src":172},"竞赛活动.PNG","https://bbs-img.huaweicloud.com/data/forums/attachment/forum/20227/28/1658937898822332273.PNG",[],{"type":18,"tag":19,"props":175,"children":177},{"id":176},"基础模型",[178],{"type":24,"value":176},{"type":18,"tag":27,"props":180,"children":181},{},[182],{"type":24,"value":183},"当前的基础模型我们上线了两个，分别是大名鼎鼎的鹏城实验室的——鹏程.盘古和中科院自动化研究所的——紫东.太初。",{"type":18,"tag":27,"props":185,"children":186},{},[187],{"type":24,"value":188},"偷个懒贴个简介给大家看看：",{"type":18,"tag":27,"props":190,"children":191},{},[192],{"type":24,"value":193},"鹏程.盘古：业界首个千亿级参数中文自然语言处理大模型，可支持知识问答、知识检索、知识推理、阅读理解等丰富的下游应用。「鹏城.盘古」由以鹏城实验室为首的技术团队联合攻关，首次基于“鹏城云脑Ⅱ”和国产MindSpore框架的自动混合并行模式实现在2048卡算力集群上的大规模分布式训练，训练出业界首个2000亿参数以中文为核心的预训练生成语言模型。鹏程·盘古α预训练模型支持丰富的场景应用，在知识问答、知识检索、知识推理、阅读理解等文本生成领域表现突出，具备很强的小样本学习能力。",{"type":18,"tag":27,"props":195,"children":196},{},[197],{"type":24,"value":198},"紫东.太初：业界首个三模态千亿参数大模型，支持文本、视觉、语音不同模态间的高效协同，可支撑影视创作、工业质检、智能驾驶等产业应用。",{"type":18,"tag":27,"props":200,"children":201},{},[202],{"type":24,"value":203},"我们的初心是原生支持基础创新模型，加速科研创新与行业应用落地。",{"type":18,"tag":27,"props":205,"children":206},{},[207],{"type":24,"value":208},"本月昇思MindSpore又又又又又更新啦！已经是V1.8版本了，大家快来体验一下！欢迎友友们补充和建议，直接评论在下方，系列还没结束，我们下期再会。",{"type":18,"tag":27,"props":210,"children":211},{},[212],{"type":24,"value":213},"See you ~~~~",{"title":7,"searchDepth":215,"depth":215,"links":216},4,[],"markdown","content:technology-blogs:zh:1665.md","content","technology-blogs/zh/1665.md","technology-blogs/zh/1665","md",1776506114780]