MLflow模型注册表示例| AWS上的数据库</标题> <link rel="stylesheet" href="//www.neidfyre.com/docs/docs/applications/_static/css/theme.css" type="text/css"> <link rel="stylesheet" href="//www.neidfyre.com/docs/docs/applications/_static/css/custom.css" type="text/css"> <link rel="stylesheet" href="//www.neidfyre.com/docs/docs/applications/_static/css/algolia.css" type="text/css"> <link rel="stylesheet" href="//www.neidfyre.com/docs/docs/applications/_static/css/cloud-provider-selector.css" type="text/css"> <link rel="stylesheet" href="https://databricks.searchunify.com/resources/search_clients_custom/7dc8d13f-90bb-11e9-98a5-06d762ad9a62/searchbox.css"> <link rel="stylesheet" href="https://databricks.searchunify.com/resources/search_clients_custom/7dc8d13f-90bb-11e9-98a5-06d762ad9a62/external_styles.css"> <link rel="stylesheet" href="https://databricks.searchunify.com/resources/search_clients_custom/7dc8d13f-90bb-11e9-98a5-06d762ad9a62/feedback.css"> </head> <body class="wy-body-for-nav MLflow Model Registry example" role="document">  <noscript> <iframe src="https://www.neidfyre.com/ns.html?id=GTM-T85FQ33" height="0" width="0" style="display:none;visibility:hidden"></iframe> </noscript>  <nav class="wy-nav-top header su_header" role="navigation" aria-label="top navigation"> <nav class="wy-nav-top header su_header" role="navigation" aria-label="top navigation"> <div class="container-logo"> <ul class="mobile-menu-toggle"> <li class="menu-toggle"><i data-toggle="wy-nav-top" class="wy-nav-top-menu-button db-icon db-icon-menu pull-left"></i><a href="//www.neidfyre.com/" class="wy-nav-top-logo"><img src="//www.neidfyre.com/docs/docs/applications/_static/databricks-logo-mobile.png" width="137" height="21" alt="砖＂></a></li> </ul> <ul class="su_nav-menu"> <li class="menu-toggle"><i data-toggle="wy-nav-top" class="wy-nav-top-menu-button db-icon db-icon-menu pull-left"></i><a href="//www.neidfyre.com/" class="wy-nav-top-logo"><img src="//www.neidfyre.com/docs/docs/applications/_static/databricks-logo-mobile.png" width="137" height="21" alt="砖＂></a></li> </ul> </div> <div class="su_nav-right"> <ul class="su_link-mobile">  </ul> <ul class="right-try-list"> <li><a href="//www.neidfyre.com/help">支持</a></li> <li><a href="mailto:doc-feedback@www.neidfyre.com?subject=Documentation Feedback">反馈</a></li> <li class="try-databrick-item"><a href="//www.neidfyre.com/docs/try-databricks">试着砖</a></li> </ul> </div> </nav> </nav> <div class="su_sub-header"> <div class="container"> <div class="su_sub-header-inner"> <div class="su_subnav-menu"> <ul> <li><a href="//www.neidfyre.com/help/s/">帮助中心</a></li> <li class="active"><a href="//www.neidfyre.com/docs/">文档</a></li> <li><a href="//www.neidfyre.com/kb/">知识库</a></li> </ul> </div> <div class="su_subnav-menu-right"> <div id="auto" style="width: 100%"> <div ng-controller="SearchautoController"> <div bind-html-compile="autocompleteHtml"> <form class="su__search-box-1"> <input class="su__search-input" type="search" name="Search box" id="su__search-b" placeholder="Search here"> <button class="button button-success" type="submit" on> <svg width="20" height="20" viewbox="0 0 28 28"> <path fill="#fff" d="M1688.27,61.271a1.327,1.327,0,0,1,0,1.859l-1.54,1.531a1.25,1.25,0,0,1-.93.383,1.282,1.282,0,0,1-.93-0.383l-5.46-5.469a1.254,1.254,0,0,1-.39-0.93V57.388a11.4,11.4,0,0,1-12.71.875,11.5,11.5,0,0,1-4.13-4.129,11.438,11.438,0,0,1,0-11.43,11.5,11.5,0,0,1,4.13-4.129,11.431,11.431,0,0,1,11.43,0,11.5,11.5,0,0,1,4.13,4.129,11.4,11.4,0,0,1-.88,12.715h0.88a1.282,1.282,0,0,1,.93.383Zm-16.25-5.852a6.952,6.952,0,0,0,3.53-.93,6.772,6.772,0,0,0,2.54-2.543,7.154,7.154,0,0,0,0-7.055,6.772,6.772,0,0,0-2.54-2.543,7.147,7.147,0,0,0-7.05,0,6.8,6.8,0,0,0-2.55,2.543,7.154,7.154,0,0,0,0,7.055,6.8,6.8,0,0,0,2.55,2.543A6.916,6.916,0,0,0,1672.02,55.419Z" transform="translate(-1660.66 -37.031)"></path> </svg></button> </form> </div> </div> </div> </div> <div> <select name="cloud provider selector" id="cloud-provider-selector"><option value="aws＂selected>亚马逊网络服务</option><option value="azure＂>微软Azure</option><option value="gcp＂>谷歌云平台bob体育客户端下载</option></select> </div> </div> </div> </div> <page class="js-page-container"> <nav data-toggle="wy-nav-shift" class="wy-nav-side su_nav-side"> <div class="wy-side-scroll"> <div class="wy-side-nav-search"></div> <div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation"> <a href="//www.neidfyre.com/docs/docs/applications/index.html" class="main-navigation-home">AWS上的数据</a> <p class="caption"><span class="caption-text">开始</span></p> <ul> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/introduction/index.html">什么是数据库?</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/getting-started/index.html">开始</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/getting-started/tutorials/index.html">教程和最佳实践</a></li> </ul> <p class="caption"><span class="caption-text">用户指南</span></p> <ul class="current"> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/workspace-index.html">数据科学与工程</a></li> <li class="toctree-l1 current"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/index.html">机器学习</a> <ul class="current"> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/tutorial/ml-quickstart.html">试试Databricks机器学习</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/data-preparation.html">准备数据和环境</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/train-model/index.html">火车模型</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/track-model-development/index.html">航迹模型开发</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/manage-model-lifecycle/index.html">管理模型生命周期</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/model-inference/index.html">为推理部署模型</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/model-export/mleap-model-export.html">导出和导入模型</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/reference-solutions/index.html">参考解决方案</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/feature-store/index.html">Databricks功能商店</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/automl/index.html">砖AutoML</a></li> <li class="toctree-l2 current"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/index.html">MLflow</a> <ul class="current"> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/quick-start.html">快速入门</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/tracking.html">跟踪机器学习训练运行</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/models.html">记录、加载、注册和部署MLflow模型</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/projects.html">在数据库上运行MLflow项目</a></li> <li class="toctree-l3 current"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/model-registry.html">MLflow模型注册表上的数据</a> <ul class="current"> <li class="toctree-l4 current"><a class="current reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#">MLflow模型注册表示例</a></li> </ul></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/serverless-real-time-inference.html">使用无服务器实时推断的模型服务</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/migrate-and-enable-serverless-real-time-inference.html">迁移到无服务器实时推理</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/create-manage-serverless-model-endpoints.html">创建和管理无服务器实时推断端点</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/private-libraries-serverless-model-serving.html">使用无服务器实时推理的自定义Python库</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/serverless-model-serving-custom-artifacts.html">打包无服务器实时推理的自定义工件和文件</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/model-serving.html">经典MLflow模型服务于数据库</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/databricks-autologging.html">砖Autologging</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/migrate-mlflow-objects.html">在工作空间之间复制MLflow对象</a></li> <li class="toctree-l3"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/end-to-end-example.html">教程:Databricks上的端到端ML模型</a></li> </ul></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/machine-learning/mlops/mlops-workflow.html">MLOps</a></li> <li class="toctree-l2"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/integrations/graphframes/index.html">GraphFrames</a></li> </ul></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/sql/index.html">砖的SQL</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/lakehouse/index.html">数据lakehouse</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/data/index.html">数据发现</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/ingestion/index.html">数据摄取</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/delta/index.html">三角洲湖</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/dev-tools/index.html">开发人员工具</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/integrations/index.html">集成</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/partner-connect/index.html">合作伙伴联系</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/partners/partners.html">砖的合作伙伴bob体育外网下载</a></li> </ul> <p class="caption"><span class="caption-text">管理指南</span></p> <ul> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/administration-guide/index.html">帐户和工作空间管理</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/workspace/index.html">浏览工作区</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/security/index.html">安全性和遵从性</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/data-governance/index.html">数据治理</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/data-sharing/index.html">数据共享(Delta sharing)</a></li> </ul> <p class="caption"><span class="caption-text">参考指南</span></p> <ul> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/reference/api.html">API参考</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/reference/sql/index.html">SQL参考</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/languages/index.html">特定于语言的概述</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/spark/index.html">Apache Spark简介</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/reference/command.html">CLI和实用程序</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/error-messages/index.html">错误消息</a></li> </ul> <p class="caption"><span class="caption-text">资源</span></p> <ul> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/release-notes/index.html">发布说明</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/optimizations/index.html">优化和性能</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/resources/index.html">资源</a></li> <li class="toctree-l1"><a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/archive/index.html">文档归档</a></li> </ul> </div> <div role="contentinfo"> <p class="build_info" data-last-edit="January 04, 2023">2023年1月12日更新</p> <p><a id="feedbacklink" href="mailto:doc-feedback@www.neidfyre.com?subject=Documentation Feedback">给我们反馈</a></p> </div> </div> </nav> <main class="wy-grid-for-nav su_nav-grid"> <section data-toggle="wy-nav-shift" class="wy-nav-content-wrap"> <div class="wy-nav-content su__nav_content"> <div class="rst-content"> <div role="navigation" aria-label="breadcrumbs navigation" class="wy-breadcrumbs-wrapper"> <ul class="wy-breadcrumbs"> <li><a href="//www.neidfyre.com/docs/docs/applications/index.html">文档</a><span class="db-icon db-icon-chevron-right"></span></li> <li><a href="//www.neidfyre.com/docs/docs/applications/machine-learning/index.html">Databricks机器学习指南</a><span class="db-icon db-icon-chevron-right"></span></li> <li><a href="//www.neidfyre.com/docs/docs/applications/mlflow/index.html">MLflow指南</a><span class="db-icon db-icon-chevron-right"></span></li> <li><a href="//www.neidfyre.com/docs/docs/applications/mlflow/model-registry.html">MLflow模型注册表上的数据</a><span class="db-icon db-icon-chevron-right"></span></li> <li>MLflow模型注册表示例</li> <li class="wy-breadcrumbs-aside"></li> </ul> </div> <div role="main" class="document" itemscope itemtype="http://schema.org/Article"> <div itemprop="articleBody"> <div class="section" id="mlflow-model-registry-example"> <h1>MLflow模型注册表示例<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#mlflow-model-registry-example" title=""></a></h1> <p>这个例子说明了如何使用MLflow模型注册表来构建一个机器学习应用程序，预测风电场的每日功率输出。这个例子展示了如何:</p> <ul class="simple"> <li><p>使用MLflow跟踪和记录模型</p></li> <li><p>向模型注册中心注册模型</p></li> <li><p>描述模型并进行模型版本阶段转换</p></li> <li><p>将已注册的模型与生产应用程序集成</p></li> <li><p>在Model Registry中搜索和发现模型</p></li> <li><p>存档和删除模型</p></li> </ul> <p>本文描述了如何使用MLflow Tracking和MLflow Model Registry ui和api执行这些步骤。</p> <p>有关使用MLflow跟踪和注册表api执行所有这些步骤的笔记本，请参阅<a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#notebook"><span class="std std-ref">注册表范例笔记本</span></a>．</p> <div class="section" id="load-dataset-train-model-and-track-with-mlflow-tracking"> <h2>加载数据集，训练模型，并跟踪MLflow跟踪<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#load-dataset-train-model-and-track-with-mlflow-tracking" title=""></a></h2> <p>在您可以在model Registry中注册一个模型之前，您必须首先训练并记录模型<a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/models.html"><span class="doc">模型</span></a>在一次<a class="reference internal" href="//www.neidfyre.com/docs/docs/applications/mlflow/tracking.html"><span class="std std-ref">实验运行</span></a>．本节展示如何加载风电场数据集、训练模型并将训练运行记录到MLflow。</p> <div class="section" id="load-dataset"> <h3>加载数据集<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#load-dataset" title=""></a></h3> <p>下面的代码加载一个数据集，其中包含美国风电场的天气数据和功率输出信息。数据集包含<code class="docutils literal notranslate"><span class="pre">风</span><span class="pre">方向</span></code>，<code class="docutils literal notranslate"><span class="pre">风</span><span class="pre">速度</span></code>,<code class="docutils literal notranslate"><span class="pre">空气</span><span class="pre">温度</span></code>特征每六小时采样一次<code class="docutils literal notranslate"><span class="pre">00:00</span></code>，一次在<code class="docutils literal notranslate"><span class="pre">08:00</span></code>，一到<code class="docutils literal notranslate"><span class="pre">16:00</span></code>)，以及每日总功率输出(<code class="docutils literal notranslate"><span class="pre">权力</span></code>)。</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="kn">进口</span><span class="nn">熊猫</span><span class="k">作为</span><span class="nn">pd</span><span class="n">wind_farm_data</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">read_csv</span><span class="p">（</span><span class="s2">“https://github.com/dbczumar/model-registry-demo-notebook/raw/master/dataset/windfarm_data.csv”</span><span class="p">，</span><span class="n">index_col</span><span class="o">＝</span><span class="mi">0</span><span class="p">）</span><span class="k">def</span><span class="nf">get_training_data</span><span class="p">（）：</span><span class="n">training_data</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">DataFrame</span><span class="p">（</span><span class="n">wind_farm_data</span><span class="p">［</span><span class="s2">“2014-01-01”</span><span class="p">：</span><span class="s2">“2018-01-01”</span><span class="p">])</span><span class="n">X</span><span class="o">＝</span><span class="n">training_data</span><span class="o">．</span><span class="n">下降</span><span class="p">（</span><span class="n">列</span><span class="o">＝</span><span class="s2">“权力”</span><span class="p">）</span><span class="n">y</span><span class="o">＝</span><span class="n">training_data</span><span class="p">［</span><span class="s2">“权力”</span><span class="p">］</span><span class="k">返回</span><span class="n">X</span><span class="p">，</span><span class="n">y</span><span class="k">def</span><span class="nf">get_validation_data</span><span class="p">（）：</span><span class="n">validation_data</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">DataFrame</span><span class="p">（</span><span class="n">wind_farm_data</span><span class="p">［</span><span class="s2">“2018-01-01”</span><span class="p">：</span><span class="s2">“2019-01-01”</span><span class="p">])</span><span class="n">X</span><span class="o">＝</span><span class="n">validation_data</span><span class="o">．</span><span class="n">下降</span><span class="p">（</span><span class="n">列</span><span class="o">＝</span><span class="s2">“权力”</span><span class="p">）</span><span class="n">y</span><span class="o">＝</span><span class="n">validation_data</span><span class="p">［</span><span class="s2">“权力”</span><span class="p">］</span><span class="k">返回</span><span class="n">X</span><span class="p">，</span><span class="n">y</span><span class="k">def</span><span class="nf">get_weather_and_forecast</span><span class="p">（）：</span><span class="n">format_date</span><span class="o">＝</span><span class="k">λ</span><span class="n">pd_date</span><span class="p">：</span><span class="n">pd_date</span><span class="o">．</span><span class="n">日期</span><span class="p">（）</span><span class="o">．</span><span class="n">strftime</span><span class="p">（</span><span class="s2">“% Y - % m</span><span class="si">% d</span><span class="s2">＂</span><span class="p">）</span><span class="n">今天</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">时间戳</span><span class="p">（</span><span class="s1">“今天”</span><span class="p">）</span><span class="o">．</span><span class="n">正常化</span><span class="p">（）</span><span class="n">week_ago</span><span class="o">＝</span><span class="n">今天</span><span class="o">-</span><span class="n">pd</span><span class="o">．</span><span class="n">Timedelta</span><span class="p">（</span><span class="n">天</span><span class="o">＝</span><span class="mi">5</span><span class="p">）</span><span class="n">week_later</span><span class="o">＝</span><span class="n">今天</span><span class="o">+</span><span class="n">pd</span><span class="o">．</span><span class="n">Timedelta</span><span class="p">（</span><span class="n">天</span><span class="o">＝</span><span class="mi">5</span><span class="p">）</span><span class="n">past_power_output</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">DataFrame</span><span class="p">（</span><span class="n">wind_farm_data</span><span class="p">）［</span><span class="n">format_date</span><span class="p">（</span><span class="n">week_ago</span><span class="p">)：</span><span class="n">format_date</span><span class="p">（</span><span class="n">今天</span><span class="p">）]</span><span class="n">weather_and_forecast</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">DataFrame</span><span class="p">（</span><span class="n">wind_farm_data</span><span class="p">）［</span><span class="n">format_date</span><span class="p">（</span><span class="n">week_ago</span><span class="p">)：</span><span class="n">format_date</span><span class="p">（</span><span class="n">week_later</span><span class="p">）]</span><span class="k">如果</span><span class="nb">len</span><span class="p">（</span><span class="n">weather_and_forecast</span><span class="p">）</span><span class="o"><</span><span class="mi">10</span><span class="p">：</span><span class="n">past_power_output</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">DataFrame</span><span class="p">（</span><span class="n">wind_farm_data</span><span class="p">）</span><span class="o">．</span><span class="n">iloc</span><span class="p">［</span><span class="o">-</span><span class="mi">10</span><span class="p">：</span><span class="o">-</span><span class="mi">5</span><span class="p">］</span><span class="n">weather_and_forecast</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">DataFrame</span><span class="p">（</span><span class="n">wind_farm_data</span><span class="p">）</span><span class="o">．</span><span class="n">iloc</span><span class="p">［</span><span class="o">-</span><span class="mi">10</span><span class="p">:]</span><span class="k">返回</span><span class="n">weather_and_forecast</span><span class="o">．</span><span class="n">下降</span><span class="p">（</span><span class="n">列</span><span class="o">＝</span><span class="s2">“权力”</span><span class="p">),</span><span class="n">past_power_output</span><span class="p">［</span><span class="s2">“权力”</span><span class="p">］</span></pre> </div> </div> </div> <div class="section" id="train-model"> <h3>火车模型<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#train-model" title=""></a></h3> <p>下面的代码使用TensorFlow Keras训练一个神经网络，根据数据集中的天气特征预测功率输出。MLflow用于跟踪模型的超参数、性能指标、源代码和工件。</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="k">def</span><span class="nf">train_keras_model</span><span class="p">（</span><span class="n">X</span><span class="p">，</span><span class="n">y</span><span class="p">)：</span><span class="kn">进口</span><span class="nn">tensorflow.keras</span><span class="kn">从</span><span class="nn">tensorflow.keras.models</span><span class="kn">进口</span><span class="n">顺序</span><span class="kn">从</span><span class="nn">tensorflow.keras.layers</span><span class="kn">进口</span><span class="n">密集的</span><span class="n">模型</span><span class="o">＝</span><span class="n">顺序</span><span class="p">（）</span><span class="n">模型</span><span class="o">．</span><span class="n">添加</span><span class="p">（</span><span class="n">密集的</span><span class="p">（</span><span class="mi">One hundred.</span><span class="p">，</span><span class="n">input_shape</span><span class="o">＝</span><span class="p">（</span><span class="n">X_train</span><span class="o">．</span><span class="n">形状</span><span class="p">［</span><span class="o">-</span><span class="mi">1</span><span class="p">),),</span><span class="n">激活</span><span class="o">＝</span><span class="s2">“relu”</span><span class="p">，</span><span class="n">的名字</span><span class="o">＝</span><span class="s2">“hidden_layer”</span><span class="p">）)</span><span class="n">模型</span><span class="o">．</span><span class="n">添加</span><span class="p">（</span><span class="n">密集的</span><span class="p">（</span><span class="mi">1</span><span class="p">）)</span><span class="n">模型</span><span class="o">．</span><span class="n">编译</span><span class="p">（</span><span class="n">损失</span><span class="o">＝</span><span class="s2">“mse”</span><span class="p">，</span><span class="n">优化器</span><span class="o">＝</span><span class="s2">“亚当”</span><span class="p">）</span><span class="n">模型</span><span class="o">．</span><span class="n">适合</span><span class="p">（</span><span class="n">X_train</span><span class="p">，</span><span class="n">y_train</span><span class="p">，</span><span class="n">时代</span><span class="o">＝</span><span class="mi">One hundred.</span><span class="p">，</span><span class="n">batch_size</span><span class="o">＝</span><span class="mi">64</span><span class="p">，</span><span class="n">validation_split</span><span class="o">＝</span><span class="mf">．2</span><span class="p">）</span><span class="k">返回</span><span class="n">模型</span><span class="kn">进口</span><span class="nn">mlflow</span><span class="n">X_train</span><span class="p">，</span><span class="n">y_train</span><span class="o">＝</span><span class="n">get_training_data</span><span class="p">（）</span><span class="k">与</span><span class="n">mlflow</span><span class="o">．</span><span class="n">start_run</span><span class="p">（）：</span><span class="c1">#自动捕获模型的参数，度量，工件，</span><span class="c1">#和' autolog() '函数的源代码</span><span class="n">mlflow</span><span class="o">．</span><span class="n">tensorflow</span><span class="o">．</span><span class="n">autolog</span><span class="p">（）</span><span class="n">train_keras_model</span><span class="p">（</span><span class="n">X_train</span><span class="p">，</span><span class="n">y_train</span><span class="p">）</span><span class="n">run_id</span><span class="o">＝</span><span class="n">mlflow</span><span class="o">．</span><span class="n">active_run</span><span class="p">（）</span><span class="o">．</span><span class="n">信息</span><span class="o">．</span><span class="n">run_id</span></pre> </div> </div> </div> </div> <div class="section" id="register-and-manage-the-model-using-the-mlflow-ui"> <h2>使用MLflow UI注册和管理模型<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#register-and-manage-the-model-using-the-mlflow-ui" title=""></a></h2> <div class="contents local topic" id="in-this-section"> <p class="topic-title first">本节:</p> <ul class="simple"> <li><p><a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#create-a-new-registered-model" id="id2">创建一个新的注册模型</a></p></li> <li><p><a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#explore-the-model-registry-ui" id="id3">探索Model Registry UI</a></p></li> <li><p><a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#add-model-descriptions" id="id4">添加模型描述</a></p></li> <li><p><a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#transition-a-model-version" id="id5">转换模型版本</a></p></li> </ul> </div> <div class="section" id="create-a-new-registered-model"> <h3><a class="toc-backref" href="//www.neidfyre.com/docs/applications/mlflow/#id2">创建一个新的注册模型</a><a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#create-a-new-registered-model" title=""></a></h3> <ol class="arabic"> <li><p>控件导航到MLflow实验运行侧栏<strong>实验</strong>图标<img alt="实验的图标＂src="//www.neidfyre.com/docs/docs/applications/_images/experiment.png">在Databricks笔记本的右侧栏。</p> <div class="figure align-default"> <img alt="分栏＂src="//www.neidfyre.com/docs/docs/applications/_images/notebook-toolbar.png"> </div></li> <li><p>找到TensorFlow Keras模型训练会话对应的MLflow Run，在MLflow Run UI中单击<strong>查看运行详情</strong>图标。</p></li> <li><p>在MLflow UI中，向下滚动到<strong>工件</strong>部分，并单击指定的目录<strong>模型</strong>．单击<strong>注册模式</strong>显示的按钮。</p> <div class="figure align-default"> <img alt="注册模式＂src="//www.neidfyre.com/docs/docs/applications/_images/mlflow_ui_register_model.png"> </div></li> <li><p>选择<strong>创建新模型</strong>从下拉菜单中，输入如下型号名称:<code class="docutils literal notranslate"><span class="pre">power-forecasting-model</span></code>．</p></li> <li><p>点击<strong>注册</strong>．这将注册一个名为<code class="docutils literal notranslate"><span class="pre">power-forecasting-model</span></code>并创建一个新的模型版本:<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>．</p> <div class="figure align-default"> <img alt="新型号＂src="//www.neidfyre.com/docs/docs/applications/_images/register_model_confirm.png"> </div><p>过了一会儿，MLflow UI显示到新注册模型的链接。按照这个链接在MLflow model Registry UI中打开新的模型版本。</p></li> </ol> </div> <div class="section" id="explore-the-model-registry-ui"> <h3><a class="toc-backref" href="//www.neidfyre.com/docs/applications/mlflow/#id3">探索Model Registry UI</a><a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#explore-the-model-registry-ui" title=""></a></h3> <p>MLflow model Registry UI中的模型版本页提供了关于<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>已注册的预测模型的信息，包括其作者、创建时间和当前阶段。</p> <div class="figure align-default"> <img alt="模型版本页面＂src="//www.neidfyre.com/docs/docs/applications/_images/registry_version_page.png"> </div> <p>模型版本页还提供了<strong>源运行</strong>链接，它将打开用于在MLflow Run UI中创建模型的MLflow Run。在MLflow Run UI中，可以访问<strong>源</strong>notebook链接查看用于训练模型的Databricks笔记本的快照。</p> <div class="figure align-default"> <img alt="源运行＂src="//www.neidfyre.com/docs/docs/applications/_images/source_run_link.png"> </div> <div class="figure align-default"> <img alt="源的笔记本＂src="//www.neidfyre.com/docs/docs/applications/_images/source_notebook_link.png"> </div> <p>要导航回MLflow模型注册表，请单击<img alt="模型图标＂src="//www.neidfyre.com/docs/docs/applications/_images/models-icon.png"><strong>模型</strong>在侧边栏。</p> <p>生成的MLflow Model Registry主页显示Databricks工作区中所有已注册模型的列表，包括它们的版本和阶段。</p> <p>单击<strong>power-forecasting-model</strong>链接以打开已注册模型页面，该页面显示预测模型的所有版本。</p> </div> <div class="section" id="add-model-descriptions"> <h3><a class="toc-backref" href="//www.neidfyre.com/docs/applications/mlflow/#id4">添加模型描述</a><a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#add-model-descriptions" title=""></a></h3> <p>您可以向已注册的模型和模型版本添加描述。注册模型描述对于记录适用于多个模型版本的信息很有用(例如，建模问题和数据集的一般概述)。模型版本描述对于详细描述特定模型版本的独特属性(例如，用于开发模型的方法和算法)是有用的。</p> <ol class="arabic"> <li><p>向已注册的功率预测模型添加高级描述。单击<img alt="编辑图标＂src="//www.neidfyre.com/docs/docs/applications/_images/edit-icon.png">图标，并输入如下描述:</p> <div class="highlight-text notranslate"> <div class="highlight"> <pre><span></span>该模型根据天气数据预测风电场的输出功率。天气数据包括三个特征:风速、风向和气温。</pre> </div> </div> <div class="figure align-default"> <img alt="添加模型描述＂src="//www.neidfyre.com/docs/docs/applications/_images/model_description.png"> </div></li> <li><p>点击<strong>保存</strong>．</p></li> <li><p>单击<strong>版本1</strong>链接从已注册的模型页面导航回模型版本页面。</p></li> <li><p>单击<img alt="编辑图标＂src="//www.neidfyre.com/docs/docs/applications/_images/edit-icon.png">图标，并输入如下描述:</p> <div class="highlight-text notranslate"> <div class="highlight"> <pre><span></span>这个模型版本是使用TensorFlow Keras构建的。它是一个有一个隐藏层的前馈神经网络。</pre> </div> </div> <div class="figure align-default"> <img alt="添加模型版本说明＂src="//www.neidfyre.com/docs/docs/applications/_images/model_version_description.png"> </div></li> <li><p>点击<strong>保存</strong>．</p></li> </ol> </div> <div class="section" id="transition-a-model-version"> <h3><a class="toc-backref" href="//www.neidfyre.com/docs/applications/mlflow/#id5">转换模型版本</a><a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#transition-a-model-version" title=""></a></h3> <p>MLflow模型注册表定义了几个模型阶段:<strong>没有一个</strong>，<strong>暂存</strong>，<strong>生产</strong>,<code class="docutils literal notranslate"><span class="pre">存档</span></code>．每个阶段都有独特的意义。例如,<strong>暂存</strong>用于模型测试，而<strong>生产</strong>用于已完成测试或审查流程并已部署到应用程序的模型。</p> <ol class="arabic"> <li><p>单击<strong>阶段</strong>按钮，显示可用的模型阶段和可用的阶段转换选项的列表。</p></li> <li><p>选择<strong>过渡到->生产</strong>并按<strong>好吧</strong>在阶段转换确认窗口中将模型转换到<strong>生产</strong>．</p> <div class="figure align-default"> <img alt="向生产过渡＂src="//www.neidfyre.com/docs/docs/applications/_images/stage_transition_prod.png"> </div><p>在模型版本过渡到<strong>生产</strong>，当前阶段显示在UI中，活动日志中添加了一个条目来反映转换。</p> <div class="figure align-default"> <img alt="生产阶段＂src="//www.neidfyre.com/docs/docs/applications/_images/stage_production.png"> </div> <div class="figure align-default"> <img alt="模型版活动＂src="//www.neidfyre.com/docs/docs/applications/_images/activity_production.png"> </div></li> </ol> <p>MLflow模型注册表允许多个模型版本共享相同的阶段。当按阶段引用模型时，模型注册表使用最新的模型版本(具有最大版本ID的模型版本)。注册的模型页面显示了特定模型的所有版本。</p> <div class="figure align-default"> <img alt="注册模型页面＂src="//www.neidfyre.com/docs/docs/applications/_images/model_registry_versions.png"> </div> </div> </div> <div class="section" id="register-and-manage-the-model-using-the--mlflow-api"> <h2>使用MLflow API注册和管理模型<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#register-and-manage-the-model-using-the--mlflow-api" title=""></a></h2> <div class="contents local topic" id="id1"> <p class="topic-title first">本节:</p> <ul class="simple"> <li><p><a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#define-the-models-name-programmatically" id="id6">以编程方式定义模型名称</a></p></li> <li><p><a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#register-the-model" id="id7">注册模型</a></p></li> <li><p><a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#add-model-and-model-version-descriptions-using-the-api" id="id8">使用API添加模型和模型版本描述</a></p></li> <li><p><a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#transition-a-model-version-and-retrieve-details-using-the-api" id="id9">转换模型版本并使用API检索细节</a></p></li> </ul> </div> <div class="section" id="define-the-models-name-programmatically"> <h3><a class="toc-backref" href="//www.neidfyre.com/docs/applications/mlflow/#id6">以编程方式定义模型名称</a><a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#define-the-models-name-programmatically" title=""></a></h3> <p>现在已经注册并转换了模型<strong>生产</strong>，您可以使用MLflow编程api引用它。定义注册模型的名称，如下所示:</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="n">model_name</span><span class="o">＝</span><span class="s2">“power-forecasting-model”</span></pre> </div> </div> </div> <div class="section" id="register-the-model"> <h3><a class="toc-backref" href="//www.neidfyre.com/docs/applications/mlflow/#id7">注册模型</a><a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#register-the-model" title=""></a></h3> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="n">model_name</span><span class="o">＝</span><span class="n">get_model_name</span><span class="p">（）</span><span class="kn">进口</span><span class="nn">mlflow</span><span class="c1">MLflow自动记录函数存储TensorFlow Keras模型的默认路径</span><span class="n">artifact_path</span><span class="o">＝</span><span class="s2">“模型”</span><span class="n">model_uri</span><span class="o">＝</span><span class="s2">”:/</span><span class="si">{run_id}</span><span class="s2">/</span><span class="si">{artifact_path}</span><span class="s2">＂</span><span class="o">．</span><span class="n">格式</span><span class="p">（</span><span class="n">run_id</span><span class="o">＝</span><span class="n">run_id</span><span class="p">，</span><span class="n">artifact_path</span><span class="o">＝</span><span class="n">artifact_path</span><span class="p">）</span><span class="n">model_details</span><span class="o">＝</span><span class="n">mlflow</span><span class="o">．</span><span class="n">register_model</span><span class="p">（</span><span class="n">model_uri</span><span class="o">＝</span><span class="n">model_uri</span><span class="p">，</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_name</span><span class="p">）</span><span class="kn">进口</span><span class="nn">时间</span><span class="kn">从</span><span class="nn">mlflow.tracking.client</span><span class="kn">进口</span><span class="n">MlflowClient</span><span class="kn">从</span><span class="nn">mlflow.entities.model_registry.model_version_status</span><span class="kn">进口</span><span class="n">ModelVersionStatus</span><span class="c1">#等待模型准备好</span><span class="k">def</span><span class="nf">wait_until_ready</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="n">model_version</span><span class="p">)：</span><span class="n">客户端</span><span class="o">＝</span><span class="n">MlflowClient</span><span class="p">（）</span><span class="k">为</span><span class="n">＿</span><span class="ow">在</span><span class="nb">范围</span><span class="p">（</span><span class="mi">10</span><span class="p">)：</span><span class="n">model_version_details</span><span class="o">＝</span><span class="n">客户端</span><span class="o">．</span><span class="n">get_model_version</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_name</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="n">model_version</span><span class="p">，</span><span class="p">）</span><span class="n">状态</span><span class="o">＝</span><span class="n">ModelVersionStatus</span><span class="o">．</span><span class="n">from_string</span><span class="p">（</span><span class="n">model_version_details</span><span class="o">．</span><span class="n">状态</span><span class="p">）</span><span class="nb">打印</span><span class="p">（</span><span class="s2">”模型状态:</span><span class="si">% s</span><span class="s2">＂</span><span class="o">％</span><span class="n">ModelVersionStatus</span><span class="o">．</span><span class="n">to_string</span><span class="p">（</span><span class="n">状态</span><span class="p">）)</span><span class="k">如果</span><span class="n">状态</span><span class="o">= =</span><span class="n">ModelVersionStatus</span><span class="o">．</span><span class="n">准备好了</span><span class="p">：</span><span class="k">打破</span><span class="n">时间</span><span class="o">．</span><span class="n">睡眠</span><span class="p">（</span><span class="mi">1</span><span class="p">）</span><span class="n">wait_until_ready</span><span class="p">（</span><span class="n">model_details</span><span class="o">．</span><span class="n">的名字</span><span class="p">，</span><span class="n">model_details</span><span class="o">．</span><span class="n">版本</span><span class="p">）</span></pre> </div> </div> </div> <div class="section" id="add-model-and-model-version-descriptions-using-the-api"> <h3><a class="toc-backref" href="//www.neidfyre.com/docs/applications/mlflow/#id8">使用API添加模型和模型版本描述</a><a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#add-model-and-model-version-descriptions-using-the-api" title=""></a></h3> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="kn">从</span><span class="nn">mlflow.tracking.client</span><span class="kn">进口</span><span class="n">MlflowClient</span><span class="n">客户端</span><span class="o">＝</span><span class="n">MlflowClient</span><span class="p">（）</span><span class="n">客户端</span><span class="o">．</span><span class="n">update_registered_model</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_details</span><span class="o">．</span><span class="n">的名字</span><span class="p">，</span><span class="n">描述</span><span class="o">＝</span><span class="s2">“该模型根据天气数据预测风电场的输出功率。天气数据包括三个特征:风速、风向和气温。”</span><span class="p">）</span><span class="n">客户端</span><span class="o">．</span><span class="n">update_model_version</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_details</span><span class="o">．</span><span class="n">的名字</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="n">model_details</span><span class="o">．</span><span class="n">版本</span><span class="p">，</span><span class="n">描述</span><span class="o">＝</span><span class="s2">“这个模型版本是使用TensorFlow Keras构建的。这是一个有一个隐藏层的前馈神经网络。”</span><span class="p">）</span></pre> </div> </div> </div> <div class="section" id="transition-a-model-version-and-retrieve-details-using-the-api"> <h3><a class="toc-backref" href="//www.neidfyre.com/docs/applications/mlflow/#id9">转换模型版本并使用API检索细节</a><a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#transition-a-model-version-and-retrieve-details-using-the-api" title=""></a></h3> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="n">客户端</span><span class="o">．</span><span class="n">transition_model_version_stage</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_details</span><span class="o">．</span><span class="n">的名字</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="n">model_details</span><span class="o">．</span><span class="n">版本</span><span class="p">，</span><span class="n">阶段</span><span class="o">＝</span><span class="s1">“生产”</span><span class="p">，</span><span class="p">）</span><span class="n">model_version_details</span><span class="o">＝</span><span class="n">客户端</span><span class="o">．</span><span class="n">get_model_version</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_details</span><span class="o">．</span><span class="n">的名字</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="n">model_details</span><span class="o">．</span><span class="n">版本</span><span class="p">，</span><span class="p">）</span><span class="nb">打印</span><span class="p">（</span><span class="s2">“目前的模型阶段是:</span><span class="si">{阶段}</span><span class="s2">’”</span><span class="o">．</span><span class="n">格式</span><span class="p">（</span><span class="n">阶段</span><span class="o">＝</span><span class="n">model_version_details</span><span class="o">．</span><span class="n">current_stage</span><span class="p">）)</span><span class="n">latest_version_info</span><span class="o">＝</span><span class="n">客户端</span><span class="o">．</span><span class="n">get_latest_versions</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="n">阶段</span><span class="o">＝</span><span class="p">［</span><span class="s2">“生产”</span><span class="p">])</span><span class="n">latest_production_version</span><span class="o">＝</span><span class="n">latest_version_info</span><span class="p">［</span><span class="mi">0</span><span class="p">］</span><span class="o">．</span><span class="n">版本</span><span class="nb">打印</span><span class="p">（</span><span class="s2">“该模型的最新生产版本”</span><span class="si">% s</span><span class="s2">' is '</span><span class="si">% s</span><span class="s2">’。”</span><span class="o">％</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="n">latest_production_version</span><span class="p">）)</span></pre> </div> </div> </div> </div> <div class="section" id="load-versions-of-the-registered-model-using-the-api"> <h2>使用API加载已注册模型的版本<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#load-versions-of-the-registered-model-using-the-api" title=""></a></h2> <p>MLflow Models组件定义了用于从多个机器学习框架加载模型的函数。例如,<code class="docutils literal notranslate"><span class="pre">mlflow.tensorflow.load_model ()</span></code>用于加载以MLflow格式保存的TensorFlow模型，以及<code class="docutils literal notranslate"><span class="pre">mlflow.sklearn.load_model ()</span></code>用于加载以MLflow格式保存的scikit-learn模型。</p> <p>这些函数可以从MLflow模型注册表加载模型。</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="kn">进口</span><span class="nn">mlflow.pyfunc</span><span class="n">model_version_uri</span><span class="o">＝</span><span class="s2">“模型:/</span><span class="si">{model_name}</span><span class="s2">/ 1”</span><span class="o">．</span><span class="n">格式</span><span class="p">（</span><span class="n">model_name</span><span class="o">＝</span><span class="n">model_name</span><span class="p">）</span><span class="nb">打印</span><span class="p">（</span><span class="s2">从URI加载已注册的模型版本:</span><span class="si">{model_uri}</span><span class="s2">’”</span><span class="o">．</span><span class="n">格式</span><span class="p">（</span><span class="n">model_uri</span><span class="o">＝</span><span class="n">model_version_uri</span><span class="p">）)</span><span class="n">model_version_1</span><span class="o">＝</span><span class="n">mlflow</span><span class="o">．</span><span class="n">pyfunc</span><span class="o">．</span><span class="n">load_model</span><span class="p">（</span><span class="n">model_version_uri</span><span class="p">）</span><span class="n">model_production_uri</span><span class="o">＝</span><span class="s2">“模型:/</span><span class="si">{model_name}</span><span class="s2">/生产”</span><span class="o">．</span><span class="n">格式</span><span class="p">（</span><span class="n">model_name</span><span class="o">＝</span><span class="n">model_name</span><span class="p">）</span><span class="nb">打印</span><span class="p">（</span><span class="s2">从URI加载已注册的模型版本:</span><span class="si">{model_uri}</span><span class="s2">’”</span><span class="o">．</span><span class="n">格式</span><span class="p">（</span><span class="n">model_uri</span><span class="o">＝</span><span class="n">model_production_uri</span><span class="p">）)</span><span class="n">model_production</span><span class="o">＝</span><span class="n">mlflow</span><span class="o">．</span><span class="n">pyfunc</span><span class="o">．</span><span class="n">load_model</span><span class="p">（</span><span class="n">model_production_uri</span><span class="p">）</span></pre> </div> </div> </div> <div class="section" id="forecast-power-output-with-the-production-model"> <h2>用生产模型预测功率输出<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#forecast-power-output-with-the-production-model" title=""></a></h2> <p>在本节中，生产模型用于评估风电场的天气预报数据。的<code class="docutils literal notranslate"><span class="pre">forecast_power ()</span></code>应用程序从指定阶段加载最新版本的预测模型，并使用它预测未来五天的发电量。</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="k">def</span><span class="nf">情节</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="n">model_stage</span><span class="p">，</span><span class="n">model_version</span><span class="p">，</span><span class="n">power_predictions</span><span class="p">，</span><span class="n">past_power_output</span><span class="p">)：</span><span class="kn">进口</span><span class="nn">熊猫</span><span class="k">作为</span><span class="nn">pd</span><span class="kn">进口</span><span class="nn">matplotlib.dates</span><span class="k">作为</span><span class="nn">mdat</span><span class="kn">从</span><span class="nn">matplotlib</span><span class="kn">进口</span><span class="n">pyplot</span><span class="k">作为</span><span class="n">plt</span><span class="n">指数</span><span class="o">＝</span><span class="n">power_predictions</span><span class="o">．</span><span class="n">指数</span><span class="n">无花果</span><span class="o">＝</span><span class="n">plt</span><span class="o">．</span><span class="n">数字</span><span class="p">（</span><span class="n">figsize</span><span class="o">＝</span><span class="p">（</span><span class="mi">11</span><span class="p">，</span><span class="mi">7</span><span class="p">）)</span><span class="n">斧头</span><span class="o">＝</span><span class="n">无花果</span><span class="o">．</span><span class="n">add_subplot</span><span class="p">（</span><span class="mi">111</span><span class="p">）</span><span class="n">斧头</span><span class="o">．</span><span class="n">set_xlabel</span><span class="p">（</span><span class="s2">“日期”</span><span class="p">，</span><span class="n">大小</span><span class="o">＝</span><span class="mi">20.</span><span class="p">，</span><span class="n">labelpad</span><span class="o">＝</span><span class="mi">20.</span><span class="p">）</span><span class="n">斧头</span><span class="o">．</span><span class="n">set_ylabel</span><span class="p">（</span><span class="s2">“权力</span><span class="se">\ n</span><span class="s2">输出</span><span class="se">\ n</span><span class="s2">(MW)”</span><span class="p">，</span><span class="n">大小</span><span class="o">＝</span><span class="mi">20.</span><span class="p">，</span><span class="n">labelpad</span><span class="o">＝</span><span class="mi">60</span><span class="p">，</span><span class="n">旋转</span><span class="o">＝</span><span class="mi">0</span><span class="p">）</span><span class="n">斧头</span><span class="o">．</span><span class="n">tick_params</span><span class="p">（</span><span class="n">轴</span><span class="o">＝</span><span class="s1">“两个”</span><span class="p">，</span><span class="n">哪一个</span><span class="o">＝</span><span class="s1">“主要的”</span><span class="p">，</span><span class="n">labelsize</span><span class="o">＝</span><span class="mi">17</span><span class="p">）</span><span class="n">斧头</span><span class="o">．</span><span class="n">xaxis</span><span class="o">．</span><span class="n">set_major_formatter</span><span class="p">（</span><span class="n">mdat</span><span class="o">．</span><span class="n">dateformat</span><span class="p">（</span><span class="s1">' % m /</span><span class="si">% d</span><span class="s1">＇</span><span class="p">）)</span><span class="n">斧头</span><span class="o">．</span><span class="n">情节</span><span class="p">（</span><span class="n">指数</span><span class="p">[:</span><span class="nb">len</span><span class="p">（</span><span class="n">past_power_output</span><span class="p">)),</span><span class="n">past_power_output</span><span class="p">，</span><span class="n">标签</span><span class="o">＝</span><span class="s2">“真正的”</span><span class="p">，</span><span class="n">颜色</span><span class="o">＝</span><span class="s2">“红色”</span><span class="p">，</span><span class="n">α</span><span class="o">＝</span><span class="mf">0．5</span><span class="p">，</span><span class="n">线宽</span><span class="o">＝</span><span class="mi">4</span><span class="p">）</span><span class="n">斧头</span><span class="o">．</span><span class="n">情节</span><span class="p">（</span><span class="n">指数</span><span class="p">，</span><span class="n">power_predictions</span><span class="o">．</span><span class="n">挤压</span><span class="p">（）,</span><span class="s2">”——“</span><span class="p">，</span><span class="n">标签</span><span class="o">＝</span><span class="s2">“由‘</span><span class="si">% s</span><span class="s2">＇</span><span class="se">\ n</span><span class="s2">在舞台上</span><span class="si">% s</span><span class="s2">”(版本</span><span class="si">% d</span><span class="s2">）"</span><span class="o">％</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="n">model_stage</span><span class="p">，</span><span class="n">model_version</span><span class="p">),</span><span class="n">颜色</span><span class="o">＝</span><span class="s2">“蓝色”</span><span class="p">，</span><span class="n">线宽</span><span class="o">＝</span><span class="mi">3.</span><span class="p">）</span><span class="n">斧头</span><span class="o">．</span><span class="n">set_ylim</span><span class="p">（</span><span class="n">ymin</span><span class="o">＝</span><span class="mi">0</span><span class="p">，</span><span class="n">ymax</span><span class="o">＝</span><span class="nb">马克斯</span><span class="p">（</span><span class="mi">3500</span><span class="p">，</span><span class="nb">int</span><span class="p">（</span><span class="nb">马克斯</span><span class="p">（</span><span class="n">power_predictions</span><span class="o">．</span><span class="n">值</span><span class="p">）</span><span class="o">＊</span><span class="mf">1.3</span><span class="p">）））</span><span class="n">斧头</span><span class="o">．</span><span class="n">传说</span><span class="p">（</span><span class="n">字形大小</span><span class="o">＝</span><span class="mi">14</span><span class="p">）</span><span class="n">plt</span><span class="o">．</span><span class="n">标题</span><span class="p">（</span><span class="s2">风电场输出功率及预测</span><span class="p">，</span><span class="n">大小</span><span class="o">＝</span><span class="mi">24</span><span class="p">，</span><span class="n">垫</span><span class="o">＝</span><span class="mi">20.</span><span class="p">）</span><span class="n">plt</span><span class="o">．</span><span class="n">tight_layout</span><span class="p">（）</span><span class="n">显示</span><span class="p">（</span><span class="n">plt</span><span class="o">．</span><span class="n">显示</span><span class="p">（））</span><span class="k">def</span><span class="nf">forecast_power</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="n">model_stage</span><span class="p">)：</span><span class="kn">从</span><span class="nn">mlflow.tracking.client</span><span class="kn">进口</span><span class="n">MlflowClient</span><span class="n">客户端</span><span class="o">＝</span><span class="n">MlflowClient</span><span class="p">（）</span><span class="n">model_version</span><span class="o">＝</span><span class="n">客户端</span><span class="o">．</span><span class="n">get_latest_versions</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="n">阶段</span><span class="o">＝</span><span class="p">［</span><span class="n">model_stage</span><span class="p">)) (</span><span class="mi">0</span><span class="p">］</span><span class="o">．</span><span class="n">版本</span><span class="n">model_uri</span><span class="o">＝</span><span class="s2">“模型:/</span><span class="si">{model_name}</span><span class="s2">/</span><span class="si">{model_stage}</span><span class="s2">＂</span><span class="o">．</span><span class="n">格式</span><span class="p">（</span><span class="n">model_name</span><span class="o">＝</span><span class="n">model_name</span><span class="p">，</span><span class="n">model_stage</span><span class="o">＝</span><span class="n">model_stage</span><span class="p">）</span><span class="n">模型</span><span class="o">＝</span><span class="n">mlflow</span><span class="o">．</span><span class="n">pyfunc</span><span class="o">．</span><span class="n">load_model</span><span class="p">（</span><span class="n">model_uri</span><span class="p">）</span><span class="n">weather_data</span><span class="p">，</span><span class="n">past_power_output</span><span class="o">＝</span><span class="n">get_weather_and_forecast</span><span class="p">（）</span><span class="n">power_predictions</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">DataFrame</span><span class="p">（</span><span class="n">模型</span><span class="o">．</span><span class="n">预测</span><span class="p">（</span><span class="n">weather_data</span><span class="p">）)</span><span class="n">power_predictions</span><span class="o">．</span><span class="n">指数</span><span class="o">＝</span><span class="n">pd</span><span class="o">．</span><span class="n">to_datetime</span><span class="p">（</span><span class="n">weather_data</span><span class="o">．</span><span class="n">指数</span><span class="p">）</span><span class="nb">打印</span><span class="p">（</span><span class="n">power_predictions</span><span class="p">）</span><span class="n">情节</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="n">model_stage</span><span class="p">，</span><span class="nb">int</span><span class="p">（</span><span class="n">model_version</span><span class="p">),</span><span class="n">power_predictions</span><span class="p">，</span><span class="n">past_power_output</span><span class="p">）</span></pre> </div> </div> </div> <div class="section" id="create-a-new-model-version"> <h2>创建一个新的模型版本<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#create-a-new-model-version" title=""></a></h2> <p>经典的机器学习技术对功率预测也很有效。下面的代码使用scikit-learn训练一个随机森林模型，并通过<code class="docutils literal notranslate"><span class="pre">mlflow.sklearn.log_model ()</span></code>函数。</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="kn">进口</span><span class="nn">mlflow.sklearn</span><span class="kn">从</span><span class="nn">sklearn.ensemble</span><span class="kn">进口</span><span class="n">RandomForestRegressor</span><span class="kn">从</span><span class="nn">sklearn.metrics</span><span class="kn">进口</span><span class="n">mean_squared_error</span><span class="k">与</span><span class="n">mlflow</span><span class="o">．</span><span class="n">start_run</span><span class="p">（）：</span><span class="n">n_estimators</span><span class="o">＝</span><span class="mi">300</span><span class="n">mlflow</span><span class="o">．</span><span class="n">log_param</span><span class="p">（</span><span class="s2">“n_estimators”</span><span class="p">，</span><span class="n">n_estimators</span><span class="p">）</span><span class="n">rand_forest</span><span class="o">＝</span><span class="n">RandomForestRegressor</span><span class="p">（</span><span class="n">n_estimators</span><span class="o">＝</span><span class="n">n_estimators</span><span class="p">）</span><span class="n">rand_forest</span><span class="o">．</span><span class="n">适合</span><span class="p">（</span><span class="n">X_train</span><span class="p">，</span><span class="n">y_train</span><span class="p">）</span><span class="n">val_x</span><span class="p">，</span><span class="n">val_y</span><span class="o">＝</span><span class="n">get_validation_data</span><span class="p">（）</span><span class="n">均方误差</span><span class="o">＝</span><span class="n">mean_squared_error</span><span class="p">（</span><span class="n">rand_forest</span><span class="o">．</span><span class="n">预测</span><span class="p">（</span><span class="n">val_x</span><span class="p">),</span><span class="n">val_y</span><span class="p">）</span><span class="nb">打印</span><span class="p">（</span><span class="s2">“验证MSE:</span><span class="si">% d</span><span class="s2">＂</span><span class="o">％</span><span class="n">均方误差</span><span class="p">）</span><span class="n">mlflow</span><span class="o">．</span><span class="n">log_metric</span><span class="p">（</span><span class="s2">“mse”</span><span class="p">，</span><span class="n">均方误差</span><span class="p">）</span><span class="c1">#指定' mlflow.sklearn.log_model() '的' registered_model_name '参数</span><span class="c1">函数将模型注册到MLflow模型注册表中。这个自动</span><span class="c1">#创建一个新的模型版本</span><span class="n">mlflow</span><span class="o">．</span><span class="n">sklearn</span><span class="o">．</span><span class="n">log_model</span><span class="p">（</span><span class="n">sk_model</span><span class="o">＝</span><span class="n">rand_forest</span><span class="p">，</span><span class="n">artifact_path</span><span class="o">＝</span><span class="s2">“sklearn-model”</span><span class="p">，</span><span class="n">registered_model_name</span><span class="o">＝</span><span class="n">model_name</span><span class="p">，</span><span class="p">）</span></pre> </div> </div> <div class="section" id="fetch-the-new-model-version-id-using-mlflow-model-registry-search"> <h3>使用MLflow model Registry搜索获取新的模型版本ID<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#fetch-the-new-model-version-id-using-mlflow-model-registry-search" title=""></a></h3> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="kn">从</span><span class="nn">mlflow.tracking.client</span><span class="kn">进口</span><span class="n">MlflowClient</span><span class="n">客户端</span><span class="o">＝</span><span class="n">MlflowClient</span><span class="p">（）</span><span class="n">model_version_infos</span><span class="o">＝</span><span class="n">客户端</span><span class="o">．</span><span class="n">search_model_versions</span><span class="p">（</span><span class="s2">"name = '</span><span class="si">% s</span><span class="s2">’”</span><span class="o">％</span><span class="n">model_name</span><span class="p">）</span><span class="n">new_model_version</span><span class="o">＝</span><span class="nb">马克斯</span><span class="p">([</span><span class="n">model_version_info</span><span class="o">．</span><span class="n">版本</span><span class="k">为</span><span class="n">model_version_info</span><span class="ow">在</span><span class="n">model_version_infos</span><span class="p">])</span><span class="n">wait_until_ready</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="n">new_model_version</span><span class="p">）</span></pre> </div> </div> </div> <div class="section" id="add-a-description-to-the-new-model-version"> <h3>为新模型版本添加描述<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#add-a-description-to-the-new-model-version" title=""></a></h3> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="n">客户端</span><span class="o">．</span><span class="n">update_model_version</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_name</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="n">new_model_version</span><span class="p">，</span><span class="n">描述</span><span class="o">＝</span><span class="s2">“这个模型版本是一个随机森林，包含100棵用scikit-learn训练过的决策树。”</span><span class="p">）</span></pre> </div> </div> </div> <div class="section" id="transition-the-new-model-version-to-staging-and-test-the-model"> <h3>将新的模型版本转换到登台并测试模型<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#transition-the-new-model-version-to-staging-and-test-the-model" title=""></a></h3> <p>在将模型部署到生产应用程序之前，最佳实践通常是在登台环境中对其进行测试。下面的代码将新模型版本转换为<strong>暂存</strong>并评估其性能。</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="n">客户端</span><span class="o">．</span><span class="n">transition_model_version_stage</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_name</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="n">new_model_version</span><span class="p">，</span><span class="n">阶段</span><span class="o">＝</span><span class="s2">“暂存”</span><span class="p">，</span><span class="p">）</span><span class="n">forecast_power</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="s2">“暂存”</span><span class="p">）</span></pre> </div> </div> </div> <div class="section" id="deploy-the-new-model-version-to-production"> <h3>将新的模型版本部署到生产中<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#deploy-the-new-model-version-to-production" title=""></a></h3> <p>在验证新模型版本在分段中表现良好之后，下面的代码将模型转换为<strong>生产</strong>并使用完全相同的应用程序代码<a class="reference internal" href="//www.neidfyre.com/docs/applications/mlflow/#forecast-power-output-with-the-production-model"><span class="std std-ref">用生产模型预测功率输出</span></a>部分，以产生电力预测。</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="n">客户端</span><span class="o">．</span><span class="n">transition_model_version_stage</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_name</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="n">new_model_version</span><span class="p">，</span><span class="n">阶段</span><span class="o">＝</span><span class="s2">“生产”</span><span class="p">，</span><span class="p">）</span><span class="n">forecast_power</span><span class="p">（</span><span class="n">model_name</span><span class="p">，</span><span class="s2">“生产”</span><span class="p">）</span></pre> </div> </div> <p>现在有两个模型版本的预测模型<strong>生产</strong>阶段:Keras模型训练的模型版本和scikit-learn训练的模型版本。</p> <div class="figure align-default"> <img alt="产品型号版本＂src="//www.neidfyre.com/docs/docs/applications/_images/multiple_prod_stage.png"> </div> <div class="admonition note"> <p class="admonition-title">请注意</p> <p>在按阶段引用模型时，MLflow model model Registry自动使用最新的生产版本。这使您可以在不更改任何应用程序代码的情况下更新生产模型。</p> </div> </div> </div> <div class="section" id="archive-and-delete-models"> <h2>存档和删除模型<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#archive-and-delete-models" title=""></a></h2> <p>当模型版本不再使用时，您可以将其存档或删除。你也可以删除整个注册模型;这将删除其所有关联的模型版本。</p> <div class="section" id="archive-version-1-of-the-power-forecasting-model"> <h3>存档<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>功率预测模型<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#archive-version-1-of-the-power-forecasting-model" title=""></a></h3> <p>存档<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>功率预测模型，因为它不再被使用。您可以在MLflow Model Registry UI中或通过MLflow API归档模型。</p> </div> <div class="section" id="archive-version-1-in-the-mlflow-ui"> <h3>存档<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>在MLflow UI中<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#archive-version-1-in-the-mlflow-ui" title=""></a></h3> <p>存档<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>功率预测模型:</p> <ol class="arabic"> <li><p>在MLflow model Registry UI中打开相应的模型版本页面:</p> <div class="figure align-default"> <img alt="转换为存档＂src="//www.neidfyre.com/docs/docs/applications/_images/stage_transition_archived.png"> </div></li> <li><p>单击<strong>阶段</strong>按钮,选择<strong>过渡到->已归档</strong>：</p> <div class="figure align-default"> <img alt="归档阶段＂src="//www.neidfyre.com/docs/docs/applications/_images/confirm_archived_transition.png"> </div></li> <li><p>新闻<strong>好吧</strong>在阶段转换确认窗口。</p> <div class="figure align-default"> <img alt="存档模型版本＂src="//www.neidfyre.com/docs/docs/applications/_images/stage_archived.png"> </div></li> </ol> <div class="section" id="archive-version-1-using-the-mlflow-api"> <h4>存档<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>使用MLflow API<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#archive-version-1-using-the-mlflow-api" title=""></a></h4> <p>下面的代码使用<code class="docutils literal notranslate"><span class="pre">MlflowClient.update_model_version ()</span></code>存档功能<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>功率预测模型。</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="kn">从</span><span class="nn">mlflow.tracking.client</span><span class="kn">进口</span><span class="n">MlflowClient</span><span class="n">客户端</span><span class="o">＝</span><span class="n">MlflowClient</span><span class="p">（）</span><span class="n">客户端</span><span class="o">．</span><span class="n">transition_model_version_stage</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_name</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="mi">1</span><span class="p">，</span><span class="n">阶段</span><span class="o">＝</span><span class="s2">“存档”</span><span class="p">，</span><span class="p">）</span></pre> </div> </div> </div> <div class="section" id="delete-version-1-of-the-power-forecasting-model"> <h4>删除<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>功率预测模型<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#delete-version-1-of-the-power-forecasting-model" title=""></a></h4> <p>您还可以使用MLflow UI或MLflow API删除模型版本。</p> <div class="admonition warning"> <p class="admonition-title">警告</p> <p>模型版本的删除是永久性的，无法恢复。</p> </div> <div class="section" id="delete-version-1-in-the-mlflow-ui"> <h5>删除<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>在MLflow UI中<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#delete-version-1-in-the-mlflow-ui" title=""></a></h5> <p>删除<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>功率预测模型:</p> <ol class="arabic"> <li><p>在MLflow model Registry UI中打开相应的模型版本页面。</p> <div class="figure align-default"> <img alt="删除模型版本＂src="//www.neidfyre.com/docs/docs/applications/_images/delete_version.png"> </div></li> <li><p>选择版本标识符旁边的下拉箭头，单击<strong>删除</strong>．</p></li> </ol> </div> <div class="section" id="delete-version-1-using-the-mlflow-api"> <h5>删除<code class="docutils literal notranslate"><span class="pre">版本</span><span class="pre">1</span></code>使用MLflow API<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#delete-version-1-using-the-mlflow-api" title=""></a></h5> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="n">客户端</span><span class="o">．</span><span class="n">delete_model_version</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_name</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="mi">1</span><span class="p">，</span><span class="p">）</span></pre> </div> </div> </div> <div class="section" id="delete-the-model-using-the-mlflow-api"> <h5>使用MLflow API删除模型<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#delete-the-model-using-the-mlflow-api" title=""></a></h5> <p>您必须首先将所有剩余的模型版本阶段转换到<strong>没有一个</strong>或<strong>存档</strong>．</p> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="kn">从</span><span class="nn">mlflow.tracking.client</span><span class="kn">进口</span><span class="n">MlflowClient</span><span class="n">客户端</span><span class="o">＝</span><span class="n">MlflowClient</span><span class="p">（）</span><span class="n">客户端</span><span class="o">．</span><span class="n">transition_model_version_stage</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_name</span><span class="p">，</span><span class="n">版本</span><span class="o">＝</span><span class="mi">2</span><span class="p">，</span><span class="n">阶段</span><span class="o">＝</span><span class="s2">“存档”</span><span class="p">，</span><span class="p">）</span></pre> </div> </div> <div class="highlight-python notranslate"> <div class="highlight"> <pre><span></span><span class="n">客户端</span><span class="o">．</span><span class="n">delete_registered_model</span><span class="p">（</span><span class="n">的名字</span><span class="o">＝</span><span class="n">model_name</span><span class="p">）</span></pre> </div> </div> </div> </div> </div> </div> <div class="section" id="notebook"> <h2>笔记本<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#notebook" title=""></a></h2> <div class="embedded-notebook-section section" id="mlflow-model-registry-example-notebook"> <span id="_staticnotebooksmlflowmlflow-model-registry-example"></span> <h3>MLflow模型注册表示例笔记本<a class="headerlink" href="//www.neidfyre.com/docs/applications/mlflow/#mlflow-model-registry-example-notebook" title=""></a></h3> <div class="embedded-notebook"> <p class="notebook-import-help"><a href="//www.neidfyre.com/docs/_static/notebooks/mlflow/mlflow-model-registry-example.html" target="_blank">在新标签页打开笔记本</a><button class="embedded-notebook-copy-to-clipboard" copy-path-to-clipboard-on-click="//www.neidfyre.com/docs/_static/notebooks/mlflow/mlflow-model-registry-example.html"><img src="//www.neidfyre.com/docs/_static/clippy.svg" alt="复制到剪贴板＂>复制导入链接</button></p> <div class="embedded-notebook-container"> <div class="loading-spinner"></div> <iframe data-src="//www.neidfyre.com/docs/_static/notebooks/mlflow/mlflow-model-registry-example.html" id="-3864000911301923380" height="1000px" width="100%" style="overflow-y:hidden;" scrolling="no"></iframe> </div> </div> </div> <p></p> <p></p> <p></p> <p></p> <p></p> <p></p> <p></p> <p></p> <p></p> <p></p> <p></p> <p></p> </div> </div> </div> </div> <div class="suapp-rating"> <div id="suPageRateApp"> <su-app></su-app> </div> </div> <footer> <hr> <div role="contentinfo"> <p class="copyright">bob下载地址©Databricks 2023。版权所有。Apache、Apache Spark、Spark和Spark标志均为<a href="http://www.apache.org/">Apache软件基金会</a>．</p> <p><a id="feedbacklink" href="mailto:doc-feedback@www.neidfyre.com?subject=Documentation Feedback">给我们反馈</a>|<a href="//www.neidfyre.com/docs/privacy-policy">隐私政策</a>|<a href="//www.neidfyre.com/docs/terms-of-use">使用条款</a></p> </div> </footer> </div> </div> </section> </main> </page>  <link href="https://cdn.jsdelivr.net/npm/select2@4.1.0-rc.0/dist/css/select2.min.css" rel="stylesheet">  </body> </html>