| <!DOCTYPE html> |
| |
| <html xmlns="http://www.w3.org/1999/xhtml"> |
| <head> |
| <meta charset="utf-8" /> |
| <meta charset="utf-8"> |
| <meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"> |
| <meta http-equiv="x-ua-compatible" content="ie=edge"> |
| <style> |
| .dropdown { |
| position: relative; |
| display: inline-block; |
| } |
| |
| .dropdown-content { |
| display: none; |
| position: absolute; |
| background-color: #f9f9f9; |
| min-width: 160px; |
| box-shadow: 0px 8px 16px 0px rgba(0,0,0,0.2); |
| padding: 12px 16px; |
| z-index: 1; |
| text-align: left; |
| } |
| |
| .dropdown:hover .dropdown-content { |
| display: block; |
| } |
| |
| .dropdown-option:hover { |
| color: #FF4500; |
| } |
| |
| .dropdown-option-active { |
| color: #FF4500; |
| font-weight: lighter; |
| } |
| |
| .dropdown-option { |
| color: #000000; |
| font-weight: lighter; |
| } |
| |
| .dropdown-header { |
| color: #FFFFFF; |
| display: inline-flex; |
| } |
| |
| .dropdown-caret { |
| width: 18px; |
| height: 54px; |
| } |
| |
| .dropdown-caret-path { |
| fill: #FFFFFF; |
| } |
| </style> |
| |
| <title>gluon.nn — Apache MXNet documentation</title> |
| |
| <link rel="stylesheet" href="../../../_static/basic.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../_static/pygments.css" type="text/css" /> |
| <link rel="stylesheet" type="text/css" href="../../../_static/mxnet.css" /> |
| <link rel="stylesheet" href="../../../_static/material-design-lite-1.3.0/material.blue-deep_orange.min.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../_static/sphinx_materialdesign_theme.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../_static/fontawesome/all.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../_static/fonts.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../_static/feedback.css" type="text/css" /> |
| <script id="documentation_options" data-url_root="../../../" src="../../../_static/documentation_options.js"></script> |
| <script src="../../../_static/jquery.js"></script> |
| <script src="../../../_static/underscore.js"></script> |
| <script src="../../../_static/doctools.js"></script> |
| <script src="../../../_static/language_data.js"></script> |
| <script src="../../../_static/matomo_analytics.js"></script> |
| <script src="../../../_static/autodoc.js"></script> |
| <script crossorigin="anonymous" integrity="sha256-Ae2Vz/4ePdIu6ZyI/5ZGsYnb+m0JlOmKPjt6XZ9JJkA=" src="https://cdnjs.cloudflare.com/ajax/libs/require.js/2.3.4/require.min.js"></script> |
| <script async="async" src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/latest.js?config=TeX-AMS-MML_HTMLorMML"></script> |
| <script type="text/x-mathjax-config">MathJax.Hub.Config({"tex2jax": {"inlineMath": [["$", "$"], ["\\(", "\\)"]], "processEscapes": true, "ignoreClass": "document", "processClass": "math|output_area"}})</script> |
| <script src="../../../_static/sphinx_materialdesign_theme.js"></script> |
| <link rel="shortcut icon" href="../../../_static/mxnet-icon.png"/> |
| <link rel="index" title="Index" href="../../../genindex.html" /> |
| <link rel="search" title="Search" href="../../../search.html" /> |
| <link rel="next" title="gluon.rnn" href="../rnn/index.html" /> |
| <link rel="prev" title="gluon.model_zoo.vision" href="../model_zoo/index.html" /> |
| </head> |
| <body><header class="site-header" role="banner"> |
| <div class="wrapper"> |
| <a class="site-title" rel="author" href="/"><img |
| src="../../../_static/mxnet_logo.png" class="site-header-logo"></a> |
| <nav class="site-nav"> |
| <input type="checkbox" id="nav-trigger" class="nav-trigger"/> |
| <label for="nav-trigger"> |
| <span class="menu-icon"> |
| <svg viewBox="0 0 18 15" width="18px" height="15px"> |
| <path d="M18,1.484c0,0.82-0.665,1.484-1.484,1.484H1.484C0.665,2.969,0,2.304,0,1.484l0,0C0,0.665,0.665,0,1.484,0 h15.032C17.335,0,18,0.665,18,1.484L18,1.484z M18,7.516C18,8.335,17.335,9,16.516,9H1.484C0.665,9,0,8.335,0,7.516l0,0 c0-0.82,0.665-1.484,1.484-1.484h15.032C17.335,6.031,18,6.696,18,7.516L18,7.516z M18,13.516C18,14.335,17.335,15,16.516,15H1.484 C0.665,15,0,14.335,0,13.516l0,0c0-0.82,0.665-1.483,1.484-1.483h15.032C17.335,12.031,18,12.695,18,13.516L18,13.516z"/> |
| </svg> |
| </span> |
| </label> |
| |
| <div class="trigger"> |
| <a class="page-link" href="/get_started">Get Started</a> |
| <a class="page-link" href="/features">Features</a> |
| <a class="page-link" href="/ecosystem">Ecosystem</a> |
| <a class="page-link page-current" href="/api">Docs & Tutorials</a> |
| <a class="page-link" href="/trusted_by">Trusted By</a> |
| <a class="page-link" href="https://github.com/apache/mxnet">GitHub</a> |
| <div class="dropdown" style="min-width:100px"> |
| <span class="dropdown-header">Apache |
| <svg class="dropdown-caret" viewBox="0 0 32 32" class="icon icon-caret-bottom" aria-hidden="true"><path class="dropdown-caret-path" d="M24 11.305l-7.997 11.39L8 11.305z"></path></svg> |
| </span> |
| <div class="dropdown-content" style="min-width:250px"> |
| <a href="https://www.apache.org/foundation/">Apache Software Foundation</a> |
| <a href="https://incubator.apache.org/">Apache Incubator</a> |
| <a href="https://www.apache.org/licenses/">License</a> |
| <a href="/versions/1.9.1/api/faq/security.html">Security</a> |
| <a href="https://privacy.apache.org/policies/privacy-policy-public.html">Privacy</a> |
| <a href="https://www.apache.org/events/current-event">Events</a> |
| <a href="https://www.apache.org/foundation/sponsorship.html">Sponsorship</a> |
| <a href="https://www.apache.org/foundation/thanks.html">Thanks</a> |
| </div> |
| </div> |
| <div class="dropdown"> |
| <span class="dropdown-header">master |
| <svg class="dropdown-caret" viewBox="0 0 32 32" class="icon icon-caret-bottom" aria-hidden="true"><path class="dropdown-caret-path" d="M24 11.305l-7.997 11.39L8 11.305z"></path></svg> |
| </span> |
| <div class="dropdown-content"> |
| <a class="dropdown-option-active" href="/versions/master/">master</a><br> |
| <a class="dropdown-option" href="/versions/1.9.1/">1.9.1</a><br> |
| <a class="dropdown-option" href="/versions/1.8.0/">1.8.0</a><br> |
| <a class="dropdown-option" href="/versions/1.7.0/">1.7.0</a><br> |
| <a class="dropdown-option" href="/versions/1.6.0/">1.6.0</a><br> |
| <a class="dropdown-option" href="/versions/1.5.0/">1.5.0</a><br> |
| <a class="dropdown-option" href="/versions/1.4.1/">1.4.1</a><br> |
| <a class="dropdown-option" href="/versions/1.3.1/">1.3.1</a><br> |
| <a class="dropdown-option" href="/versions/1.2.1/">1.2.1</a><br> |
| <a class="dropdown-option" href="/versions/1.1.0/">1.1.0</a><br> |
| <a class="dropdown-option" href="/versions/1.0.0/">1.0.0</a><br> |
| <a class="dropdown-option" href="/versions/0.12.1/">0.12.1</a><br> |
| <a class="dropdown-option" href="/versions/0.11.0/">0.11.0</a> |
| </div> |
| </div> |
| </div> |
| </nav> |
| </div> |
| </header> |
| <div class="mdl-layout mdl-js-layout mdl-layout--fixed-header mdl-layout--fixed-drawer"><header class="mdl-layout__header mdl-layout__header--waterfall "> |
| <div class="mdl-layout__header-row"> |
| |
| <nav class="mdl-navigation breadcrumb"> |
| <a class="mdl-navigation__link" href="../../index.html">Python API</a><i class="material-icons">navigate_next</i> |
| <a class="mdl-navigation__link" href="../index.html">mxnet.gluon</a><i class="material-icons">navigate_next</i> |
| <a class="mdl-navigation__link is-active">gluon.nn</a> |
| </nav> |
| <div class="mdl-layout-spacer"></div> |
| <nav class="mdl-navigation"> |
| |
| <form class="form-inline pull-sm-right" action="../../../search.html" method="get"> |
| <div class="mdl-textfield mdl-js-textfield mdl-textfield--expandable mdl-textfield--floating-label mdl-textfield--align-right"> |
| <label id="quick-search-icon" class="mdl-button mdl-js-button mdl-button--icon" for="waterfall-exp"> |
| <i class="material-icons">search</i> |
| </label> |
| <div class="mdl-textfield__expandable-holder"> |
| <input class="mdl-textfield__input" type="text" name="q" id="waterfall-exp" placeholder="Search" /> |
| <input type="hidden" name="check_keywords" value="yes" /> |
| <input type="hidden" name="area" value="default" /> |
| </div> |
| </div> |
| <div class="mdl-tooltip" data-mdl-for="quick-search-icon"> |
| Quick search |
| </div> |
| </form> |
| |
| <a id="button-show-github" |
| href="https://github.com/apache/mxnet/edit/master/docs/python_docs/python/api/gluon/nn/index.rst" class="mdl-button mdl-js-button mdl-button--icon"> |
| <i class="material-icons">edit</i> |
| </a> |
| <div class="mdl-tooltip" data-mdl-for="button-show-github"> |
| Edit on Github |
| </div> |
| </nav> |
| </div> |
| <div class="mdl-layout__header-row header-links"> |
| <div class="mdl-layout-spacer"></div> |
| <nav class="mdl-navigation"> |
| </nav> |
| </div> |
| </header><header class="mdl-layout__drawer"> |
| |
| <div class="globaltoc"> |
| <span class="mdl-layout-title toc">Table Of Contents</span> |
| |
| |
| |
| <nav class="mdl-navigation"> |
| <ul class="current"> |
| <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/index.html">Python Tutorials</a><ul> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/getting-started/index.html">Getting Started</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/index.html">Crash Course</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/0-introduction.html">Introduction</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/1-nparray.html">Step 1: Manipulate data with NP on MXNet</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/2-create-nn.html">Step 2: Create a neural network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/3-autograd.html">Step 3: Automatic differentiation with autograd</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/4-components.html">Step 4: Necessary components that are not in the network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-datasets.html">Step 5: <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s and <code class="docutils literal notranslate"><span class="pre">DataLoader</span></code></a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-datasets.html#Using-own-data-with-included-Datasets">Using own data with included <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-datasets.html#Using-your-own-data-with-custom-Datasets">Using your own data with custom <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-datasets.html#New-in-MXNet-2.0:-faster-C++-backend-dataloaders">New in MXNet 2.0: faster C++ backend dataloaders</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/6-train-nn.html">Step 6: Train a Neural Network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/7-use-gpus.html">Step 7: Load and Run a NN using GPU</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/to-mxnet/index.html">Moving to MXNet from Other Frameworks</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/to-mxnet/pytorch.html">PyTorch vs Apache MXNet</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/gluon_from_experiment_to_deployment.html">Gluon: from experiment to deployment</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/gluon_migration_guide.html">Gluon2.0: Migration Guide</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/logistic_regression_explained.html">Logistic regression explained</a></li> |
| <li class="toctree-l3"><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/image/mnist.html">MNIST</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/packages/index.html">Packages</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/autograd/index.html">Automatic Differentiation</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/gluon/index.html">Gluon</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/index.html">Blocks</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/custom-layer.html">Custom Layers</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/hybridize.html">Hybridize</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/init.html">Initialization</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/naming.html">Parameter and Block Naming</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/nn.html">Layers and Blocks</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/parameters.html">Parameter Management</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/activations/activations.html">Activation Blocks</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/data/index.html">Data Tutorials</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/data_augmentation.html">Image Augmentation</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/datasets.html">Gluon <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s and <code class="docutils literal notranslate"><span class="pre">DataLoader</span></code></a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/datasets.html#Using-own-data-with-included-Datasets">Using own data with included <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/datasets.html#Using-own-data-with-custom-Datasets">Using own data with custom <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/datasets.html#Appendix:-Upgrading-from-Module-DataIter-to-Gluon-DataLoader">Appendix: Upgrading from Module <code class="docutils literal notranslate"><span class="pre">DataIter</span></code> to Gluon <code class="docutils literal notranslate"><span class="pre">DataLoader</span></code></a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/image/index.html">Image Tutorials</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/image/info_gan.html">Image similarity search with InfoGAN</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/image/mnist.html">Handwritten Digit Recognition</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/loss/index.html">Losses</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/loss/custom-loss.html">Custom Loss Blocks</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/loss/kl_divergence.html">Kullback-Leibler (KL) Divergence</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/loss/loss.html">Loss functions</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/text/index.html">Text Tutorials</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/text/gnmt.html">Google Neural Machine Translation</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/text/transformer.html">Machine Translation with Transformer</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/training/index.html">Training</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/training/fit_api_tutorial.html">MXNet Gluon Fit API</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/training/trainer.html">Trainer</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/training/learning_rates/index.html">Learning Rates</a><ul> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/gluon/training/learning_rates/learning_rate_finder.html">Learning Rate Finder</a></li> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/gluon/training/learning_rates/learning_rate_schedules.html">Learning Rate Schedules</a></li> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/gluon/training/learning_rates/learning_rate_schedules_advanced.html">Advanced Learning Rate Schedules</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/training/normalization/index.html">Normalization Blocks</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/kvstore/index.html">KVStore</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/kvstore/kvstore.html">Distributed Key-Value Store</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/legacy/index.html">Legacy</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/index.html">NDArray</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/01-ndarray-intro.html">An Intro: Manipulate Data the MXNet Way with NDArray</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/02-ndarray-operations.html">NDArray Operations</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/03-ndarray-contexts.html">NDArray Contexts</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/gotchas_numpy_in_mxnet.html">Gotchas using NumPy in Apache MXNet</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/sparse/index.html">Tutorials</a><ul> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/sparse/csr.html">CSRNDArray - NDArray in Compressed Sparse Row Storage Format</a></li> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/sparse/row_sparse.html">RowSparseNDArray - NDArray for Sparse Gradient Updates</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/np/index.html">What is NP on MXNet</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/np/cheat-sheet.html">The NP on MXNet cheat sheet</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/np/np-vs-numpy.html">Differences between NP on MXNet and NumPy</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/onnx/index.html">ONNX</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/onnx/fine_tuning_gluon.html">Fine-tuning an ONNX model</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/onnx/inference_on_onnx_model.html">Running inference on MXNet/Gluon from an ONNX model</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/deploy/export/onnx.html">Export ONNX Models</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/optimizer/index.html">Optimizers</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/viz/index.html">Visualization</a><ul> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/faq/visualize_graph">Visualize networks</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/performance/index.html">Performance</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/performance/compression/index.html">Compression</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/compression/int8.html">Deploy with int-8</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/faq/float16">Float16</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/faq/gradient_compression">Gradient Compression</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://gluon-cv.mxnet.io/build/examples_deployment/int8_inference.html">GluonCV with Quantized Models</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/performance/backend/index.html">Accelerated Backend Tools</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/dnnl/index.html">oneDNN</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/dnnl/dnnl_readme.html">Install MXNet with oneDNN</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/dnnl/dnnl_quantization.html">oneDNN Quantization</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/dnnl/dnnl_quantization_inc.html">Improving accuracy with Intel® Neural Compressor</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/tvm.html">Use TVM</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/profiler.html">Profiling MXNet Models</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/amp.html">Using AMP: Automatic Mixed Precision</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/deploy/index.html">Deployment</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/deploy/export/index.html">Export</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/export/onnx.html">Exporting to ONNX format</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://gluon-cv.mxnet.io/build/examples_deployment/export_network.html">Export Gluon CV Models</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Save / Load Parameters</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/deploy/inference/index.html">Inference</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/inference/cpp.html">Deploy into C++</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/inference/image_classification_jetson.html">Image Classication using pretrained ResNet-50 model on Jetson module</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/deploy/run-on-aws/index.html">Run on AWS</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/run-on-aws/use_ec2.html">Run on an EC2 Instance</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/run-on-aws/use_sagemaker.html">Run on Amazon SageMaker</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/run-on-aws/cloud.html">MXNet on the Cloud</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/extend/index.html">Extend</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/extend/customop.html">Custom Numpy Operators</a></li> |
| <li class="toctree-l3"><a class="reference external" href="https://mxnet.apache.org/api/faq/new_op">New Operator Creation</a></li> |
| <li class="toctree-l3"><a class="reference external" href="https://mxnet.apache.org/api/faq/add_op_in_backend">New Operator in MXNet Backend</a></li> |
| <li class="toctree-l3"><a class="reference external" href="https://mxnet.apache.org/api/faq/using_rtc">Using RTC for CUDA kernels</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l1 current"><a class="reference internal" href="../../index.html">Python API</a><ul class="current"> |
| <li class="toctree-l2"><a class="reference internal" href="../../np/index.html">mxnet.np</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../np/arrays.html">Array objects</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/arrays.ndarray.html">The N-dimensional array (<code class="xref py py-class docutils literal notranslate"><span class="pre">ndarray</span></code>)</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/arrays.indexing.html">Indexing</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../np/routines.html">Routines</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.array-creation.html">Array creation routines</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.eye.html">mxnet.np.eye</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.empty.html">mxnet.np.empty</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.full.html">mxnet.np.full</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.identity.html">mxnet.np.identity</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ones.html">mxnet.np.ones</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ones_like.html">mxnet.np.ones_like</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.zeros.html">mxnet.np.zeros</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.zeros_like.html">mxnet.np.zeros_like</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.array.html">mxnet.np.array</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.copy.html">mxnet.np.copy</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arange.html">mxnet.np.arange</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linspace.html">mxnet.np.linspace</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.logspace.html">mxnet.np.logspace</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.meshgrid.html">mxnet.np.meshgrid</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tril.html">mxnet.np.tril</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.array-manipulation.html">Array manipulation routines</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.reshape.html">mxnet.np.reshape</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ravel.html">mxnet.np.ravel</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ndarray.flatten.html">mxnet.np.ndarray.flatten</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.swapaxes.html">mxnet.np.swapaxes</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ndarray.T.html">mxnet.np.ndarray.T</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.transpose.html">mxnet.np.transpose</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.moveaxis.html">mxnet.np.moveaxis</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.rollaxis.html">mxnet.np.rollaxis</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.expand_dims.html">mxnet.np.expand_dims</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.squeeze.html">mxnet.np.squeeze</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.broadcast_to.html">mxnet.np.broadcast_to</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.broadcast_arrays.html">mxnet.np.broadcast_arrays</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.atleast_1d.html">mxnet.np.atleast_1d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.atleast_2d.html">mxnet.np.atleast_2d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.atleast_3d.html">mxnet.np.atleast_3d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.concatenate.html">mxnet.np.concatenate</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.stack.html">mxnet.np.stack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.dstack.html">mxnet.np.dstack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.vstack.html">mxnet.np.vstack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.column_stack.html">mxnet.np.column_stack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.hstack.html">mxnet.np.hstack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.split.html">mxnet.np.split</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.hsplit.html">mxnet.np.hsplit</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.vsplit.html">mxnet.np.vsplit</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.array_split.html">mxnet.np.array_split</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.dsplit.html">mxnet.np.dsplit</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tile.html">mxnet.np.tile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.repeat.html">mxnet.np.repeat</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.unique.html">mxnet.np.unique</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.delete.html">mxnet.np.delete</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.insert.html">mxnet.np.insert</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.append.html">mxnet.np.append</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.resize.html">mxnet.np.resize</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.trim_zeros.html">mxnet.np.trim_zeros</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.reshape.html">mxnet.np.reshape</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.flip.html">mxnet.np.flip</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.roll.html">mxnet.np.roll</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.rot90.html">mxnet.np.rot90</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fliplr.html">mxnet.np.fliplr</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.flipud.html">mxnet.np.flipud</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.io.html">Input and output</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.genfromtxt.html">mxnet.np.genfromtxt</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ndarray.tolist.html">mxnet.np.ndarray.tolist</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.set_printoptions.html">mxnet.np.set_printoptions</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.linalg.html">Linear algebra (<code class="xref py py-mod docutils literal notranslate"><span class="pre">numpy.linalg</span></code>)</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.dot.html">mxnet.np.dot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.vdot.html">mxnet.np.vdot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.inner.html">mxnet.np.inner</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.outer.html">mxnet.np.outer</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tensordot.html">mxnet.np.tensordot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.einsum.html">mxnet.np.einsum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.multi_dot.html">mxnet.np.linalg.multi_dot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.matmul.html">mxnet.np.matmul</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.matrix_power.html">mxnet.np.linalg.matrix_power</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.kron.html">mxnet.np.kron</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.svd.html">mxnet.np.linalg.svd</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.cholesky.html">mxnet.np.linalg.cholesky</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.qr.html">mxnet.np.linalg.qr</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.eig.html">mxnet.np.linalg.eig</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.eigh.html">mxnet.np.linalg.eigh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.eigvals.html">mxnet.np.linalg.eigvals</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.eigvalsh.html">mxnet.np.linalg.eigvalsh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.norm.html">mxnet.np.linalg.norm</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.trace.html">mxnet.np.trace</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.cond.html">mxnet.np.linalg.cond</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.det.html">mxnet.np.linalg.det</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.matrix_rank.html">mxnet.np.linalg.matrix_rank</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.slogdet.html">mxnet.np.linalg.slogdet</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.solve.html">mxnet.np.linalg.solve</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.tensorsolve.html">mxnet.np.linalg.tensorsolve</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.lstsq.html">mxnet.np.linalg.lstsq</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.inv.html">mxnet.np.linalg.inv</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.pinv.html">mxnet.np.linalg.pinv</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.tensorinv.html">mxnet.np.linalg.tensorinv</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.math.html">Mathematical functions</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sin.html">mxnet.np.sin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cos.html">mxnet.np.cos</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tan.html">mxnet.np.tan</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arcsin.html">mxnet.np.arcsin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arccos.html">mxnet.np.arccos</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arctan.html">mxnet.np.arctan</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.degrees.html">mxnet.np.degrees</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.radians.html">mxnet.np.radians</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.hypot.html">mxnet.np.hypot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arctan2.html">mxnet.np.arctan2</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.deg2rad.html">mxnet.np.deg2rad</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.rad2deg.html">mxnet.np.rad2deg</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.unwrap.html">mxnet.np.unwrap</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sinh.html">mxnet.np.sinh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cosh.html">mxnet.np.cosh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tanh.html">mxnet.np.tanh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arcsinh.html">mxnet.np.arcsinh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arccosh.html">mxnet.np.arccosh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arctanh.html">mxnet.np.arctanh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.rint.html">mxnet.np.rint</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fix.html">mxnet.np.fix</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.floor.html">mxnet.np.floor</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ceil.html">mxnet.np.ceil</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.trunc.html">mxnet.np.trunc</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.around.html">mxnet.np.around</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.round_.html">mxnet.np.round_</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sum.html">mxnet.np.sum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.prod.html">mxnet.np.prod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cumsum.html">mxnet.np.cumsum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanprod.html">mxnet.np.nanprod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nansum.html">mxnet.np.nansum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cumprod.html">mxnet.np.cumprod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nancumprod.html">mxnet.np.nancumprod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nancumsum.html">mxnet.np.nancumsum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.diff.html">mxnet.np.diff</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ediff1d.html">mxnet.np.ediff1d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cross.html">mxnet.np.cross</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.trapz.html">mxnet.np.trapz</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.exp.html">mxnet.np.exp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.expm1.html">mxnet.np.expm1</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.log.html">mxnet.np.log</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.log10.html">mxnet.np.log10</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.log2.html">mxnet.np.log2</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.log1p.html">mxnet.np.log1p</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.logaddexp.html">mxnet.np.logaddexp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.i0.html">mxnet.np.i0</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ldexp.html">mxnet.np.ldexp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.signbit.html">mxnet.np.signbit</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.copysign.html">mxnet.np.copysign</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.frexp.html">mxnet.np.frexp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.spacing.html">mxnet.np.spacing</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.lcm.html">mxnet.np.lcm</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.gcd.html">mxnet.np.gcd</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.add.html">mxnet.np.add</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.reciprocal.html">mxnet.np.reciprocal</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.negative.html">mxnet.np.negative</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.divide.html">mxnet.np.divide</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.power.html">mxnet.np.power</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.subtract.html">mxnet.np.subtract</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.mod.html">mxnet.np.mod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.multiply.html">mxnet.np.multiply</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.true_divide.html">mxnet.np.true_divide</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.remainder.html">mxnet.np.remainder</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.positive.html">mxnet.np.positive</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.float_power.html">mxnet.np.float_power</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fmod.html">mxnet.np.fmod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.modf.html">mxnet.np.modf</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.divmod.html">mxnet.np.divmod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.floor_divide.html">mxnet.np.floor_divide</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.clip.html">mxnet.np.clip</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sqrt.html">mxnet.np.sqrt</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cbrt.html">mxnet.np.cbrt</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.square.html">mxnet.np.square</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.absolute.html">mxnet.np.absolute</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sign.html">mxnet.np.sign</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.maximum.html">mxnet.np.maximum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.minimum.html">mxnet.np.minimum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fabs.html">mxnet.np.fabs</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.heaviside.html">mxnet.np.heaviside</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fmax.html">mxnet.np.fmax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fmin.html">mxnet.np.fmin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nan_to_num.html">mxnet.np.nan_to_num</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.interp.html">mxnet.np.interp</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/random/index.html">np.random</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.choice.html">mxnet.np.random.choice</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.shuffle.html">mxnet.np.random.shuffle</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.normal.html">mxnet.np.random.normal</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.uniform.html">mxnet.np.random.uniform</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.rand.html">mxnet.np.random.rand</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.randint.html">mxnet.np.random.randint</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.beta.html">mxnet.np.random.beta</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.chisquare.html">mxnet.np.random.chisquare</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.exponential.html">mxnet.np.random.exponential</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.f.html">mxnet.np.random.f</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.gamma.html">mxnet.np.random.gamma</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.gumbel.html">mxnet.np.random.gumbel</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.laplace.html">mxnet.np.random.laplace</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.logistic.html">mxnet.np.random.logistic</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.lognormal.html">mxnet.np.random.lognormal</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.multinomial.html">mxnet.np.random.multinomial</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.multivariate_normal.html">mxnet.np.random.multivariate_normal</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.pareto.html">mxnet.np.random.pareto</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.power.html">mxnet.np.random.power</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.rayleigh.html">mxnet.np.random.rayleigh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.weibull.html">mxnet.np.random.weibull</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.sort.html">Sorting, searching, and counting</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ndarray.sort.html">mxnet.np.ndarray.sort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sort.html">mxnet.np.sort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.lexsort.html">mxnet.np.lexsort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argsort.html">mxnet.np.argsort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.msort.html">mxnet.np.msort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.partition.html">mxnet.np.partition</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argpartition.html">mxnet.np.argpartition</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argmax.html">mxnet.np.argmax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argmin.html">mxnet.np.argmin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanargmax.html">mxnet.np.nanargmax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanargmin.html">mxnet.np.nanargmin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argwhere.html">mxnet.np.argwhere</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nonzero.html">mxnet.np.nonzero</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.flatnonzero.html">mxnet.np.flatnonzero</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.where.html">mxnet.np.where</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.searchsorted.html">mxnet.np.searchsorted</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.extract.html">mxnet.np.extract</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.count_nonzero.html">mxnet.np.count_nonzero</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.statistics.html">Statistics</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.min.html">mxnet.np.min</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.max.html">mxnet.np.max</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.amin.html">mxnet.np.amin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.amax.html">mxnet.np.amax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanmin.html">mxnet.np.nanmin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanmax.html">mxnet.np.nanmax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ptp.html">mxnet.np.ptp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.percentile.html">mxnet.np.percentile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanpercentile.html">mxnet.np.nanpercentile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.quantile.html">mxnet.np.quantile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanquantile.html">mxnet.np.nanquantile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.mean.html">mxnet.np.mean</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.std.html">mxnet.np.std</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.var.html">mxnet.np.var</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.median.html">mxnet.np.median</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.average.html">mxnet.np.average</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanmedian.html">mxnet.np.nanmedian</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanstd.html">mxnet.np.nanstd</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanvar.html">mxnet.np.nanvar</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.corrcoef.html">mxnet.np.corrcoef</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.correlate.html">mxnet.np.correlate</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cov.html">mxnet.np.cov</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.histogram.html">mxnet.np.histogram</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.histogram2d.html">mxnet.np.histogram2d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.histogramdd.html">mxnet.np.histogramdd</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.bincount.html">mxnet.np.bincount</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.histogram_bin_edges.html">mxnet.np.histogram_bin_edges</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.digitize.html">mxnet.np.digitize</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../npx/index.html">NPX: NumPy Neural Network Extension</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.set_np.html">mxnet.npx.set_np</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.reset_np.html">mxnet.npx.reset_np</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.cpu.html">mxnet.npx.cpu</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.cpu_pinned.html">mxnet.npx.cpu_pinned</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.gpu.html">mxnet.npx.gpu</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.gpu_memory_info.html">mxnet.npx.gpu_memory_info</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.current_device.html">mxnet.npx.current_device</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.num_gpus.html">mxnet.npx.num_gpus</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.activation.html">mxnet.npx.activation</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.batch_norm.html">mxnet.npx.batch_norm</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.convolution.html">mxnet.npx.convolution</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.dropout.html">mxnet.npx.dropout</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.embedding.html">mxnet.npx.embedding</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.fully_connected.html">mxnet.npx.fully_connected</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.layer_norm.html">mxnet.npx.layer_norm</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.pooling.html">mxnet.npx.pooling</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.rnn.html">mxnet.npx.rnn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.leaky_relu.html">mxnet.npx.leaky_relu</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.multibox_detection.html">mxnet.npx.multibox_detection</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.multibox_prior.html">mxnet.npx.multibox_prior</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.multibox_target.html">mxnet.npx.multibox_target</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.roi_pooling.html">mxnet.npx.roi_pooling</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.sigmoid.html">mxnet.npx.sigmoid</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.relu.html">mxnet.npx.relu</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.smooth_l1.html">mxnet.npx.smooth_l1</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.softmax.html">mxnet.npx.softmax</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.log_softmax.html">mxnet.npx.log_softmax</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.topk.html">mxnet.npx.topk</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.waitall.html">mxnet.npx.waitall</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.load.html">mxnet.npx.load</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.save.html">mxnet.npx.save</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.one_hot.html">mxnet.npx.one_hot</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.pick.html">mxnet.npx.pick</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.reshape_like.html">mxnet.npx.reshape_like</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.batch_flatten.html">mxnet.npx.batch_flatten</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.batch_dot.html">mxnet.npx.batch_dot</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.gamma.html">mxnet.npx.gamma</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.sequence_mask.html">mxnet.npx.sequence_mask</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2 current"><a class="reference internal" href="../index.html">mxnet.gluon</a><ul class="current"> |
| <li class="toctree-l3"><a class="reference internal" href="../block.html">gluon.Block</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../hybrid_block.html">gluon.HybridBlock</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../symbol_block.html">gluon.SymbolBlock</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../constant.html">gluon.Constant</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../parameter.html">gluon.Parameter</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../trainer.html">gluon.Trainer</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../contrib/index.html">gluon.contrib</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../data/index.html">gluon.data</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../data/vision/index.html">data.vision</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../data/vision/datasets/index.html">vision.datasets</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../data/vision/transforms/index.html">vision.transforms</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../loss/index.html">gluon.loss</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../metric/index.html">gluon.metric</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../model_zoo/index.html">gluon.model_zoo.vision</a></li> |
| <li class="toctree-l3 current"><a class="current reference internal" href="#">gluon.nn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../rnn/index.html">gluon.rnn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../utils/index.html">gluon.utils</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../autograd/index.html">mxnet.autograd</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../initializer/index.html">mxnet.initializer</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../optimizer/index.html">mxnet.optimizer</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../lr_scheduler/index.html">mxnet.lr_scheduler</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html">KVStore: Communication for Distributed Training</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html#horovod">Horovod</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.Horovod.html">mxnet.kvstore.Horovod</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html#byteps">BytePS</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.BytePS.html">mxnet.kvstore.BytePS</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html#kvstore-interface">KVStore Interface</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.KVStore.html">mxnet.kvstore.KVStore</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.KVStoreBase.html">mxnet.kvstore.KVStoreBase</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.KVStoreServer.html">mxnet.kvstore.KVStoreServer</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../contrib/index.html">mxnet.contrib</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/io/index.html">contrib.io</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/ndarray/index.html">contrib.ndarray</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/onnx/index.html">contrib.onnx</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/quantization/index.html">contrib.quantization</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/symbol/index.html">contrib.symbol</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/tensorboard/index.html">contrib.tensorboard</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/tensorrt/index.html">contrib.tensorrt</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/text/index.html">contrib.text</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../legacy/index.html">Legacy</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/callback/index.html">mxnet.callback</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/image/index.html">mxnet.image</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/io/index.html">mxnet.io</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/ndarray/index.html">mxnet.ndarray</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/ndarray.html">ndarray</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/contrib/index.html">ndarray.contrib</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/image/index.html">ndarray.image</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/linalg/index.html">ndarray.linalg</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/op/index.html">ndarray.op</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/random/index.html">ndarray.random</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/register/index.html">ndarray.register</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/sparse/index.html">ndarray.sparse</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/utils/index.html">ndarray.utils</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/recordio/index.html">mxnet.recordio</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/symbol/index.html">mxnet.symbol</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/symbol.html">symbol</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/contrib/index.html">symbol.contrib</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/image/index.html">symbol.image</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/linalg/index.html">symbol.linalg</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/op/index.html">symbol.op</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/random/index.html">symbol.random</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/register/index.html">symbol.register</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/sparse/index.html">symbol.sparse</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/visualization/index.html">mxnet.visualization</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../device/index.html">mxnet.device</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../engine/index.html">mxnet.engine</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../executor/index.html">mxnet.executor</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore_server/index.html">mxnet.kvstore_server</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../profiler/index.html">mxnet.profiler</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../rtc/index.html">mxnet.rtc</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../runtime/index.html">mxnet.runtime</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../runtime/generated/mxnet.runtime.Feature.html">mxnet.runtime.Feature</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../runtime/generated/mxnet.runtime.Features.html">mxnet.runtime.Features</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../runtime/generated/mxnet.runtime.feature_list.html">mxnet.runtime.feature_list</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../test_utils/index.html">mxnet.test_utils</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../util/index.html">mxnet.util</a></li> |
| </ul> |
| </li> |
| </ul> |
| |
| </nav> |
| |
| </div> |
| |
| </header> |
| <main class="mdl-layout__content" tabIndex="0"> |
| <header class="mdl-layout__drawer"> |
| |
| <div class="globaltoc"> |
| <span class="mdl-layout-title toc">Table Of Contents</span> |
| |
| |
| |
| <nav class="mdl-navigation"> |
| <ul class="current"> |
| <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/index.html">Python Tutorials</a><ul> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/getting-started/index.html">Getting Started</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/index.html">Crash Course</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/0-introduction.html">Introduction</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/1-nparray.html">Step 1: Manipulate data with NP on MXNet</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/2-create-nn.html">Step 2: Create a neural network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/3-autograd.html">Step 3: Automatic differentiation with autograd</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/4-components.html">Step 4: Necessary components that are not in the network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-datasets.html">Step 5: <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s and <code class="docutils literal notranslate"><span class="pre">DataLoader</span></code></a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-datasets.html#Using-own-data-with-included-Datasets">Using own data with included <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-datasets.html#Using-your-own-data-with-custom-Datasets">Using your own data with custom <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-datasets.html#New-in-MXNet-2.0:-faster-C++-backend-dataloaders">New in MXNet 2.0: faster C++ backend dataloaders</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/6-train-nn.html">Step 6: Train a Neural Network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/7-use-gpus.html">Step 7: Load and Run a NN using GPU</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/to-mxnet/index.html">Moving to MXNet from Other Frameworks</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/to-mxnet/pytorch.html">PyTorch vs Apache MXNet</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/gluon_from_experiment_to_deployment.html">Gluon: from experiment to deployment</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/gluon_migration_guide.html">Gluon2.0: Migration Guide</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/getting-started/logistic_regression_explained.html">Logistic regression explained</a></li> |
| <li class="toctree-l3"><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/image/mnist.html">MNIST</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/packages/index.html">Packages</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/autograd/index.html">Automatic Differentiation</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/gluon/index.html">Gluon</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/index.html">Blocks</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/custom-layer.html">Custom Layers</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/hybridize.html">Hybridize</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/init.html">Initialization</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/naming.html">Parameter and Block Naming</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/nn.html">Layers and Blocks</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/parameters.html">Parameter Management</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/blocks/activations/activations.html">Activation Blocks</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/data/index.html">Data Tutorials</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/data_augmentation.html">Image Augmentation</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/datasets.html">Gluon <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s and <code class="docutils literal notranslate"><span class="pre">DataLoader</span></code></a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/datasets.html#Using-own-data-with-included-Datasets">Using own data with included <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/datasets.html#Using-own-data-with-custom-Datasets">Using own data with custom <code class="docutils literal notranslate"><span class="pre">Dataset</span></code>s</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/datasets.html#Appendix:-Upgrading-from-Module-DataIter-to-Gluon-DataLoader">Appendix: Upgrading from Module <code class="docutils literal notranslate"><span class="pre">DataIter</span></code> to Gluon <code class="docutils literal notranslate"><span class="pre">DataLoader</span></code></a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/image/index.html">Image Tutorials</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/image/info_gan.html">Image similarity search with InfoGAN</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/image/mnist.html">Handwritten Digit Recognition</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/loss/index.html">Losses</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/loss/custom-loss.html">Custom Loss Blocks</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/loss/kl_divergence.html">Kullback-Leibler (KL) Divergence</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/loss/loss.html">Loss functions</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/text/index.html">Text Tutorials</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/text/gnmt.html">Google Neural Machine Translation</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/text/transformer.html">Machine Translation with Transformer</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/gluon/training/index.html">Training</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/training/fit_api_tutorial.html">MXNet Gluon Fit API</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/training/trainer.html">Trainer</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/training/learning_rates/index.html">Learning Rates</a><ul> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/gluon/training/learning_rates/learning_rate_finder.html">Learning Rate Finder</a></li> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/gluon/training/learning_rates/learning_rate_schedules.html">Learning Rate Schedules</a></li> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/gluon/training/learning_rates/learning_rate_schedules_advanced.html">Advanced Learning Rate Schedules</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/training/normalization/index.html">Normalization Blocks</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/kvstore/index.html">KVStore</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/kvstore/kvstore.html">Distributed Key-Value Store</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/legacy/index.html">Legacy</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/index.html">NDArray</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/01-ndarray-intro.html">An Intro: Manipulate Data the MXNet Way with NDArray</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/02-ndarray-operations.html">NDArray Operations</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/03-ndarray-contexts.html">NDArray Contexts</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/gotchas_numpy_in_mxnet.html">Gotchas using NumPy in Apache MXNet</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/sparse/index.html">Tutorials</a><ul> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/sparse/csr.html">CSRNDArray - NDArray in Compressed Sparse Row Storage Format</a></li> |
| <li class="toctree-l6"><a class="reference internal" href="../../../tutorials/packages/legacy/ndarray/sparse/row_sparse.html">RowSparseNDArray - NDArray for Sparse Gradient Updates</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/np/index.html">What is NP on MXNet</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/np/cheat-sheet.html">The NP on MXNet cheat sheet</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/np/np-vs-numpy.html">Differences between NP on MXNet and NumPy</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/onnx/index.html">ONNX</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/onnx/fine_tuning_gluon.html">Fine-tuning an ONNX model</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/onnx/inference_on_onnx_model.html">Running inference on MXNet/Gluon from an ONNX model</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/deploy/export/onnx.html">Export ONNX Models</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/optimizer/index.html">Optimizers</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/packages/viz/index.html">Visualization</a><ul> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/faq/visualize_graph">Visualize networks</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/performance/index.html">Performance</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/performance/compression/index.html">Compression</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/compression/int8.html">Deploy with int-8</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/faq/float16">Float16</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/faq/gradient_compression">Gradient Compression</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://gluon-cv.mxnet.io/build/examples_deployment/int8_inference.html">GluonCV with Quantized Models</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/performance/backend/index.html">Accelerated Backend Tools</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/dnnl/index.html">oneDNN</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/dnnl/dnnl_readme.html">Install MXNet with oneDNN</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/dnnl/dnnl_quantization.html">oneDNN Quantization</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/dnnl/dnnl_quantization_inc.html">Improving accuracy with Intel® Neural Compressor</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/tvm.html">Use TVM</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/profiler.html">Profiling MXNet Models</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/amp.html">Using AMP: Automatic Mixed Precision</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/deploy/index.html">Deployment</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/deploy/export/index.html">Export</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/export/onnx.html">Exporting to ONNX format</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://gluon-cv.mxnet.io/build/examples_deployment/export_network.html">Export Gluon CV Models</a></li> |
| <li class="toctree-l4"><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Save / Load Parameters</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/deploy/inference/index.html">Inference</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/inference/cpp.html">Deploy into C++</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/inference/image_classification_jetson.html">Image Classication using pretrained ResNet-50 model on Jetson module</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/deploy/run-on-aws/index.html">Run on AWS</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/run-on-aws/use_ec2.html">Run on an EC2 Instance</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/run-on-aws/use_sagemaker.html">Run on Amazon SageMaker</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/run-on-aws/cloud.html">MXNet on the Cloud</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../../tutorials/extend/index.html">Extend</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../../tutorials/extend/customop.html">Custom Numpy Operators</a></li> |
| <li class="toctree-l3"><a class="reference external" href="https://mxnet.apache.org/api/faq/new_op">New Operator Creation</a></li> |
| <li class="toctree-l3"><a class="reference external" href="https://mxnet.apache.org/api/faq/add_op_in_backend">New Operator in MXNet Backend</a></li> |
| <li class="toctree-l3"><a class="reference external" href="https://mxnet.apache.org/api/faq/using_rtc">Using RTC for CUDA kernels</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l1 current"><a class="reference internal" href="../../index.html">Python API</a><ul class="current"> |
| <li class="toctree-l2"><a class="reference internal" href="../../np/index.html">mxnet.np</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../np/arrays.html">Array objects</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/arrays.ndarray.html">The N-dimensional array (<code class="xref py py-class docutils literal notranslate"><span class="pre">ndarray</span></code>)</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/arrays.indexing.html">Indexing</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../np/routines.html">Routines</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.array-creation.html">Array creation routines</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.eye.html">mxnet.np.eye</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.empty.html">mxnet.np.empty</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.full.html">mxnet.np.full</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.identity.html">mxnet.np.identity</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ones.html">mxnet.np.ones</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ones_like.html">mxnet.np.ones_like</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.zeros.html">mxnet.np.zeros</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.zeros_like.html">mxnet.np.zeros_like</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.array.html">mxnet.np.array</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.copy.html">mxnet.np.copy</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arange.html">mxnet.np.arange</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linspace.html">mxnet.np.linspace</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.logspace.html">mxnet.np.logspace</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.meshgrid.html">mxnet.np.meshgrid</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tril.html">mxnet.np.tril</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.array-manipulation.html">Array manipulation routines</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.reshape.html">mxnet.np.reshape</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ravel.html">mxnet.np.ravel</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ndarray.flatten.html">mxnet.np.ndarray.flatten</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.swapaxes.html">mxnet.np.swapaxes</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ndarray.T.html">mxnet.np.ndarray.T</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.transpose.html">mxnet.np.transpose</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.moveaxis.html">mxnet.np.moveaxis</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.rollaxis.html">mxnet.np.rollaxis</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.expand_dims.html">mxnet.np.expand_dims</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.squeeze.html">mxnet.np.squeeze</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.broadcast_to.html">mxnet.np.broadcast_to</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.broadcast_arrays.html">mxnet.np.broadcast_arrays</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.atleast_1d.html">mxnet.np.atleast_1d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.atleast_2d.html">mxnet.np.atleast_2d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.atleast_3d.html">mxnet.np.atleast_3d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.concatenate.html">mxnet.np.concatenate</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.stack.html">mxnet.np.stack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.dstack.html">mxnet.np.dstack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.vstack.html">mxnet.np.vstack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.column_stack.html">mxnet.np.column_stack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.hstack.html">mxnet.np.hstack</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.split.html">mxnet.np.split</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.hsplit.html">mxnet.np.hsplit</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.vsplit.html">mxnet.np.vsplit</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.array_split.html">mxnet.np.array_split</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.dsplit.html">mxnet.np.dsplit</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tile.html">mxnet.np.tile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.repeat.html">mxnet.np.repeat</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.unique.html">mxnet.np.unique</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.delete.html">mxnet.np.delete</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.insert.html">mxnet.np.insert</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.append.html">mxnet.np.append</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.resize.html">mxnet.np.resize</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.trim_zeros.html">mxnet.np.trim_zeros</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.reshape.html">mxnet.np.reshape</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.flip.html">mxnet.np.flip</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.roll.html">mxnet.np.roll</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.rot90.html">mxnet.np.rot90</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fliplr.html">mxnet.np.fliplr</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.flipud.html">mxnet.np.flipud</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.io.html">Input and output</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.genfromtxt.html">mxnet.np.genfromtxt</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ndarray.tolist.html">mxnet.np.ndarray.tolist</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.set_printoptions.html">mxnet.np.set_printoptions</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.linalg.html">Linear algebra (<code class="xref py py-mod docutils literal notranslate"><span class="pre">numpy.linalg</span></code>)</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.dot.html">mxnet.np.dot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.vdot.html">mxnet.np.vdot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.inner.html">mxnet.np.inner</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.outer.html">mxnet.np.outer</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tensordot.html">mxnet.np.tensordot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.einsum.html">mxnet.np.einsum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.multi_dot.html">mxnet.np.linalg.multi_dot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.matmul.html">mxnet.np.matmul</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.matrix_power.html">mxnet.np.linalg.matrix_power</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.kron.html">mxnet.np.kron</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.svd.html">mxnet.np.linalg.svd</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.cholesky.html">mxnet.np.linalg.cholesky</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.qr.html">mxnet.np.linalg.qr</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.eig.html">mxnet.np.linalg.eig</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.eigh.html">mxnet.np.linalg.eigh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.eigvals.html">mxnet.np.linalg.eigvals</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.eigvalsh.html">mxnet.np.linalg.eigvalsh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.norm.html">mxnet.np.linalg.norm</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.trace.html">mxnet.np.trace</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.cond.html">mxnet.np.linalg.cond</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.det.html">mxnet.np.linalg.det</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.matrix_rank.html">mxnet.np.linalg.matrix_rank</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.slogdet.html">mxnet.np.linalg.slogdet</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.solve.html">mxnet.np.linalg.solve</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.tensorsolve.html">mxnet.np.linalg.tensorsolve</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.lstsq.html">mxnet.np.linalg.lstsq</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.inv.html">mxnet.np.linalg.inv</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.pinv.html">mxnet.np.linalg.pinv</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.linalg.tensorinv.html">mxnet.np.linalg.tensorinv</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.math.html">Mathematical functions</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sin.html">mxnet.np.sin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cos.html">mxnet.np.cos</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tan.html">mxnet.np.tan</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arcsin.html">mxnet.np.arcsin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arccos.html">mxnet.np.arccos</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arctan.html">mxnet.np.arctan</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.degrees.html">mxnet.np.degrees</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.radians.html">mxnet.np.radians</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.hypot.html">mxnet.np.hypot</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arctan2.html">mxnet.np.arctan2</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.deg2rad.html">mxnet.np.deg2rad</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.rad2deg.html">mxnet.np.rad2deg</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.unwrap.html">mxnet.np.unwrap</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sinh.html">mxnet.np.sinh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cosh.html">mxnet.np.cosh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.tanh.html">mxnet.np.tanh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arcsinh.html">mxnet.np.arcsinh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arccosh.html">mxnet.np.arccosh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.arctanh.html">mxnet.np.arctanh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.rint.html">mxnet.np.rint</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fix.html">mxnet.np.fix</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.floor.html">mxnet.np.floor</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ceil.html">mxnet.np.ceil</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.trunc.html">mxnet.np.trunc</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.around.html">mxnet.np.around</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.round_.html">mxnet.np.round_</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sum.html">mxnet.np.sum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.prod.html">mxnet.np.prod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cumsum.html">mxnet.np.cumsum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanprod.html">mxnet.np.nanprod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nansum.html">mxnet.np.nansum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cumprod.html">mxnet.np.cumprod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nancumprod.html">mxnet.np.nancumprod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nancumsum.html">mxnet.np.nancumsum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.diff.html">mxnet.np.diff</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ediff1d.html">mxnet.np.ediff1d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cross.html">mxnet.np.cross</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.trapz.html">mxnet.np.trapz</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.exp.html">mxnet.np.exp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.expm1.html">mxnet.np.expm1</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.log.html">mxnet.np.log</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.log10.html">mxnet.np.log10</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.log2.html">mxnet.np.log2</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.log1p.html">mxnet.np.log1p</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.logaddexp.html">mxnet.np.logaddexp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.i0.html">mxnet.np.i0</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ldexp.html">mxnet.np.ldexp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.signbit.html">mxnet.np.signbit</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.copysign.html">mxnet.np.copysign</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.frexp.html">mxnet.np.frexp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.spacing.html">mxnet.np.spacing</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.lcm.html">mxnet.np.lcm</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.gcd.html">mxnet.np.gcd</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.add.html">mxnet.np.add</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.reciprocal.html">mxnet.np.reciprocal</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.negative.html">mxnet.np.negative</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.divide.html">mxnet.np.divide</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.power.html">mxnet.np.power</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.subtract.html">mxnet.np.subtract</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.mod.html">mxnet.np.mod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.multiply.html">mxnet.np.multiply</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.true_divide.html">mxnet.np.true_divide</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.remainder.html">mxnet.np.remainder</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.positive.html">mxnet.np.positive</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.float_power.html">mxnet.np.float_power</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fmod.html">mxnet.np.fmod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.modf.html">mxnet.np.modf</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.divmod.html">mxnet.np.divmod</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.floor_divide.html">mxnet.np.floor_divide</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.clip.html">mxnet.np.clip</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sqrt.html">mxnet.np.sqrt</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cbrt.html">mxnet.np.cbrt</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.square.html">mxnet.np.square</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.absolute.html">mxnet.np.absolute</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sign.html">mxnet.np.sign</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.maximum.html">mxnet.np.maximum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.minimum.html">mxnet.np.minimum</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fabs.html">mxnet.np.fabs</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.heaviside.html">mxnet.np.heaviside</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fmax.html">mxnet.np.fmax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.fmin.html">mxnet.np.fmin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nan_to_num.html">mxnet.np.nan_to_num</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.interp.html">mxnet.np.interp</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/random/index.html">np.random</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.choice.html">mxnet.np.random.choice</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.shuffle.html">mxnet.np.random.shuffle</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.normal.html">mxnet.np.random.normal</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.uniform.html">mxnet.np.random.uniform</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.rand.html">mxnet.np.random.rand</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.randint.html">mxnet.np.random.randint</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.beta.html">mxnet.np.random.beta</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.chisquare.html">mxnet.np.random.chisquare</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.exponential.html">mxnet.np.random.exponential</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.f.html">mxnet.np.random.f</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.gamma.html">mxnet.np.random.gamma</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.gumbel.html">mxnet.np.random.gumbel</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.laplace.html">mxnet.np.random.laplace</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.logistic.html">mxnet.np.random.logistic</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.lognormal.html">mxnet.np.random.lognormal</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.multinomial.html">mxnet.np.random.multinomial</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.multivariate_normal.html">mxnet.np.random.multivariate_normal</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.pareto.html">mxnet.np.random.pareto</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.power.html">mxnet.np.random.power</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.rayleigh.html">mxnet.np.random.rayleigh</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/random/generated/mxnet.np.random.weibull.html">mxnet.np.random.weibull</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.sort.html">Sorting, searching, and counting</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ndarray.sort.html">mxnet.np.ndarray.sort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.sort.html">mxnet.np.sort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.lexsort.html">mxnet.np.lexsort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argsort.html">mxnet.np.argsort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.msort.html">mxnet.np.msort</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.partition.html">mxnet.np.partition</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argpartition.html">mxnet.np.argpartition</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argmax.html">mxnet.np.argmax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argmin.html">mxnet.np.argmin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanargmax.html">mxnet.np.nanargmax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanargmin.html">mxnet.np.nanargmin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.argwhere.html">mxnet.np.argwhere</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nonzero.html">mxnet.np.nonzero</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.flatnonzero.html">mxnet.np.flatnonzero</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.where.html">mxnet.np.where</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.searchsorted.html">mxnet.np.searchsorted</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.extract.html">mxnet.np.extract</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.count_nonzero.html">mxnet.np.count_nonzero</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../np/routines.statistics.html">Statistics</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.min.html">mxnet.np.min</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.max.html">mxnet.np.max</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.amin.html">mxnet.np.amin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.amax.html">mxnet.np.amax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanmin.html">mxnet.np.nanmin</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanmax.html">mxnet.np.nanmax</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.ptp.html">mxnet.np.ptp</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.percentile.html">mxnet.np.percentile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanpercentile.html">mxnet.np.nanpercentile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.quantile.html">mxnet.np.quantile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanquantile.html">mxnet.np.nanquantile</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.mean.html">mxnet.np.mean</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.std.html">mxnet.np.std</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.var.html">mxnet.np.var</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.median.html">mxnet.np.median</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.average.html">mxnet.np.average</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanmedian.html">mxnet.np.nanmedian</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanstd.html">mxnet.np.nanstd</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.nanvar.html">mxnet.np.nanvar</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.corrcoef.html">mxnet.np.corrcoef</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.correlate.html">mxnet.np.correlate</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.cov.html">mxnet.np.cov</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.histogram.html">mxnet.np.histogram</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.histogram2d.html">mxnet.np.histogram2d</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.histogramdd.html">mxnet.np.histogramdd</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.bincount.html">mxnet.np.bincount</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.histogram_bin_edges.html">mxnet.np.histogram_bin_edges</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../np/generated/mxnet.np.digitize.html">mxnet.np.digitize</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../npx/index.html">NPX: NumPy Neural Network Extension</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.set_np.html">mxnet.npx.set_np</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.reset_np.html">mxnet.npx.reset_np</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.cpu.html">mxnet.npx.cpu</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.cpu_pinned.html">mxnet.npx.cpu_pinned</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.gpu.html">mxnet.npx.gpu</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.gpu_memory_info.html">mxnet.npx.gpu_memory_info</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.current_device.html">mxnet.npx.current_device</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.num_gpus.html">mxnet.npx.num_gpus</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.activation.html">mxnet.npx.activation</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.batch_norm.html">mxnet.npx.batch_norm</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.convolution.html">mxnet.npx.convolution</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.dropout.html">mxnet.npx.dropout</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.embedding.html">mxnet.npx.embedding</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.fully_connected.html">mxnet.npx.fully_connected</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.layer_norm.html">mxnet.npx.layer_norm</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.pooling.html">mxnet.npx.pooling</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.rnn.html">mxnet.npx.rnn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.leaky_relu.html">mxnet.npx.leaky_relu</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.multibox_detection.html">mxnet.npx.multibox_detection</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.multibox_prior.html">mxnet.npx.multibox_prior</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.multibox_target.html">mxnet.npx.multibox_target</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.roi_pooling.html">mxnet.npx.roi_pooling</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.sigmoid.html">mxnet.npx.sigmoid</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.relu.html">mxnet.npx.relu</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.smooth_l1.html">mxnet.npx.smooth_l1</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.softmax.html">mxnet.npx.softmax</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.log_softmax.html">mxnet.npx.log_softmax</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.topk.html">mxnet.npx.topk</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.waitall.html">mxnet.npx.waitall</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.load.html">mxnet.npx.load</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.save.html">mxnet.npx.save</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.one_hot.html">mxnet.npx.one_hot</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.pick.html">mxnet.npx.pick</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.reshape_like.html">mxnet.npx.reshape_like</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.batch_flatten.html">mxnet.npx.batch_flatten</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.batch_dot.html">mxnet.npx.batch_dot</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.gamma.html">mxnet.npx.gamma</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../npx/generated/mxnet.npx.sequence_mask.html">mxnet.npx.sequence_mask</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2 current"><a class="reference internal" href="../index.html">mxnet.gluon</a><ul class="current"> |
| <li class="toctree-l3"><a class="reference internal" href="../block.html">gluon.Block</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../hybrid_block.html">gluon.HybridBlock</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../symbol_block.html">gluon.SymbolBlock</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../constant.html">gluon.Constant</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../parameter.html">gluon.Parameter</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../trainer.html">gluon.Trainer</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../contrib/index.html">gluon.contrib</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../data/index.html">gluon.data</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../data/vision/index.html">data.vision</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../data/vision/datasets/index.html">vision.datasets</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../data/vision/transforms/index.html">vision.transforms</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../loss/index.html">gluon.loss</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../metric/index.html">gluon.metric</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../model_zoo/index.html">gluon.model_zoo.vision</a></li> |
| <li class="toctree-l3 current"><a class="current reference internal" href="#">gluon.nn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../rnn/index.html">gluon.rnn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../utils/index.html">gluon.utils</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../autograd/index.html">mxnet.autograd</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../initializer/index.html">mxnet.initializer</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../optimizer/index.html">mxnet.optimizer</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../lr_scheduler/index.html">mxnet.lr_scheduler</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html">KVStore: Communication for Distributed Training</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html#horovod">Horovod</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.Horovod.html">mxnet.kvstore.Horovod</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html#byteps">BytePS</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.BytePS.html">mxnet.kvstore.BytePS</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html#kvstore-interface">KVStore Interface</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.KVStore.html">mxnet.kvstore.KVStore</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.KVStoreBase.html">mxnet.kvstore.KVStoreBase</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../kvstore/generated/mxnet.kvstore.KVStoreServer.html">mxnet.kvstore.KVStoreServer</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../contrib/index.html">mxnet.contrib</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/io/index.html">contrib.io</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/ndarray/index.html">contrib.ndarray</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/onnx/index.html">contrib.onnx</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/quantization/index.html">contrib.quantization</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/symbol/index.html">contrib.symbol</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/tensorboard/index.html">contrib.tensorboard</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/tensorrt/index.html">contrib.tensorrt</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../contrib/text/index.html">contrib.text</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../legacy/index.html">Legacy</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/callback/index.html">mxnet.callback</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/image/index.html">mxnet.image</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/io/index.html">mxnet.io</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/ndarray/index.html">mxnet.ndarray</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/ndarray.html">ndarray</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/contrib/index.html">ndarray.contrib</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/image/index.html">ndarray.image</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/linalg/index.html">ndarray.linalg</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/op/index.html">ndarray.op</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/random/index.html">ndarray.random</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/register/index.html">ndarray.register</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/sparse/index.html">ndarray.sparse</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/ndarray/utils/index.html">ndarray.utils</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/recordio/index.html">mxnet.recordio</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/symbol/index.html">mxnet.symbol</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/symbol.html">symbol</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/contrib/index.html">symbol.contrib</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/image/index.html">symbol.image</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/linalg/index.html">symbol.linalg</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/op/index.html">symbol.op</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/random/index.html">symbol.random</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/register/index.html">symbol.register</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../legacy/symbol/sparse/index.html">symbol.sparse</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../legacy/visualization/index.html">mxnet.visualization</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../device/index.html">mxnet.device</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../engine/index.html">mxnet.engine</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../executor/index.html">mxnet.executor</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore_server/index.html">mxnet.kvstore_server</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../profiler/index.html">mxnet.profiler</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../rtc/index.html">mxnet.rtc</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../runtime/index.html">mxnet.runtime</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../runtime/generated/mxnet.runtime.Feature.html">mxnet.runtime.Feature</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../runtime/generated/mxnet.runtime.Features.html">mxnet.runtime.Features</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../runtime/generated/mxnet.runtime.feature_list.html">mxnet.runtime.feature_list</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../test_utils/index.html">mxnet.test_utils</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../util/index.html">mxnet.util</a></li> |
| </ul> |
| </li> |
| </ul> |
| |
| </nav> |
| |
| </div> |
| |
| </header> |
| |
| <div class="document"> |
| <div class="page-content" role="main"> |
| |
| <div class="section" id="gluon-nn"> |
| <h1>gluon.nn<a class="headerlink" href="#gluon-nn" title="Permalink to this headline">¶</a></h1> |
| <p>Gluon provides a large number of build-in neural network layers in the following |
| two modules:</p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#module-mxnet.gluon.nn" title="mxnet.gluon.nn"><code class="xref py py-obj docutils literal notranslate"><span class="pre">mxnet.gluon.nn</span></code></a></p></td> |
| <td><p>Neural network layers.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p>We group all layers in these two modules according to their categories.</p> |
| <div class="section" id="sequential-containers"> |
| <h2>Sequential Containers<a class="headerlink" href="#sequential-containers" title="Permalink to this headline">¶</a></h2> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Sequential" title="mxnet.gluon.nn.Sequential"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Sequential</span></code></a></p></td> |
| <td><p>Stacks Blocks sequentially.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential" title="mxnet.gluon.nn.HybridSequential"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.HybridSequential</span></code></a></p></td> |
| <td><p>Stacks HybridBlocks sequentially.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="concatenation-containers"> |
| <h2>Concatenation Containers<a class="headerlink" href="#concatenation-containers" title="Permalink to this headline">¶</a></h2> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate" title="mxnet.gluon.nn.Concatenate"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Concatenate</span></code></a></p></td> |
| <td><p>Lays <cite>Block</cite> s concurrently.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate" title="mxnet.gluon.nn.HybridConcatenate"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.HybridConcatenate</span></code></a></p></td> |
| <td><p>Lays <cite>HybridBlock</cite> s concurrently.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="basic-layers"> |
| <h2>Basic Layers<a class="headerlink" href="#basic-layers" title="Permalink to this headline">¶</a></h2> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense" title="mxnet.gluon.nn.Dense"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Dense</span></code></a></p></td> |
| <td><p>Just your regular densely-connected NN layer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation" title="mxnet.gluon.nn.Activation"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Activation</span></code></a></p></td> |
| <td><p>Applies an activation function to input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout" title="mxnet.gluon.nn.Dropout"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Dropout</span></code></a></p></td> |
| <td><p>Applies Dropout to the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten" title="mxnet.gluon.nn.Flatten"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Flatten</span></code></a></p></td> |
| <td><p>Flattens the input to two dimensional.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Lambda" title="mxnet.gluon.nn.Lambda"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Lambda</span></code></a></p></td> |
| <td><p>Wraps an operator or an expression as a Block object.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda" title="mxnet.gluon.nn.HybridLambda"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.HybridLambda</span></code></a></p></td> |
| <td><p>Wraps an operator or an expression as a HybridBlock object.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity" title="mxnet.gluon.nn.Identity"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Identity</span></code></a></p></td> |
| <td><p>Block that passes through the input directly.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="convolutional-layers"> |
| <h2>Convolutional Layers<a class="headerlink" href="#convolutional-layers" title="Permalink to this headline">¶</a></h2> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv1D" title="mxnet.gluon.nn.Conv1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Conv1D</span></code></a></p></td> |
| <td><p>1D convolution layer (e.g.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv2D" title="mxnet.gluon.nn.Conv2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Conv2D</span></code></a></p></td> |
| <td><p>2D convolution layer (e.g.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv3D" title="mxnet.gluon.nn.Conv3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Conv3D</span></code></a></p></td> |
| <td><p>3D convolution layer (e.g.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv1DTranspose" title="mxnet.gluon.nn.Conv1DTranspose"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Conv1DTranspose</span></code></a></p></td> |
| <td><p>Transposed 1D convolution layer (sometimes called Deconvolution).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv2DTranspose" title="mxnet.gluon.nn.Conv2DTranspose"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Conv2DTranspose</span></code></a></p></td> |
| <td><p>Transposed 2D convolution layer (sometimes called Deconvolution).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv3DTranspose" title="mxnet.gluon.nn.Conv3DTranspose"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Conv3DTranspose</span></code></a></p></td> |
| <td><p>Transposed 3D convolution layer (sometimes called Deconvolution).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution" title="mxnet.gluon.nn.DeformableConvolution"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.DeformableConvolution</span></code></a></p></td> |
| <td><p>2-D Deformable Convolution v_1 (Dai, 2017).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution" title="mxnet.gluon.nn.ModulatedDeformableConvolution"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.ModulatedDeformableConvolution</span></code></a></p></td> |
| <td><p>2-D Deformable Convolution v2 (Dai, 2018).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="pixel-shuffle-layers"> |
| <h2>Pixel Shuffle Layers<a class="headerlink" href="#pixel-shuffle-layers" title="Permalink to this headline">¶</a></h2> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D" title="mxnet.gluon.nn.PixelShuffle1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.PixelShuffle1D</span></code></a></p></td> |
| <td><p>Pixel-shuffle layer for upsampling in 1 dimension.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D" title="mxnet.gluon.nn.PixelShuffle2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.PixelShuffle2D</span></code></a></p></td> |
| <td><p>Pixel-shuffle layer for upsampling in 2 dimensions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D" title="mxnet.gluon.nn.PixelShuffle3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.PixelShuffle3D</span></code></a></p></td> |
| <td><p>Pixel-shuffle layer for upsampling in 3 dimensions.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="pooling-layers"> |
| <h2>Pooling Layers<a class="headerlink" href="#pooling-layers" title="Permalink to this headline">¶</a></h2> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.MaxPool1D" title="mxnet.gluon.nn.MaxPool1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.MaxPool1D</span></code></a></p></td> |
| <td><p>Max pooling operation for one dimensional data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.MaxPool2D" title="mxnet.gluon.nn.MaxPool2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.MaxPool2D</span></code></a></p></td> |
| <td><p>Max pooling operation for two dimensional (spatial) data.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.MaxPool3D" title="mxnet.gluon.nn.MaxPool3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.MaxPool3D</span></code></a></p></td> |
| <td><p>Max pooling operation for 3D data (spatial or spatio-temporal).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.AvgPool1D" title="mxnet.gluon.nn.AvgPool1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.AvgPool1D</span></code></a></p></td> |
| <td><p>Average pooling operation for temporal data.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.AvgPool2D" title="mxnet.gluon.nn.AvgPool2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.AvgPool2D</span></code></a></p></td> |
| <td><p>Average pooling operation for spatial data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.AvgPool3D" title="mxnet.gluon.nn.AvgPool3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.AvgPool3D</span></code></a></p></td> |
| <td><p>Average pooling operation for 3D data (spatial or spatio-temporal).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalMaxPool1D" title="mxnet.gluon.nn.GlobalMaxPool1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.GlobalMaxPool1D</span></code></a></p></td> |
| <td><p>Gloabl max pooling operation for one dimensional (temporal) data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalMaxPool2D" title="mxnet.gluon.nn.GlobalMaxPool2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.GlobalMaxPool2D</span></code></a></p></td> |
| <td><p>Global max pooling operation for two dimensional (spatial) data.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalMaxPool3D" title="mxnet.gluon.nn.GlobalMaxPool3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.GlobalMaxPool3D</span></code></a></p></td> |
| <td><p>Global max pooling operation for 3D data (spatial or spatio-temporal).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalAvgPool1D" title="mxnet.gluon.nn.GlobalAvgPool1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.GlobalAvgPool1D</span></code></a></p></td> |
| <td><p>Global average pooling operation for temporal data.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalAvgPool2D" title="mxnet.gluon.nn.GlobalAvgPool2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.GlobalAvgPool2D</span></code></a></p></td> |
| <td><p>Global average pooling operation for spatial data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalAvgPool3D" title="mxnet.gluon.nn.GlobalAvgPool3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.GlobalAvgPool3D</span></code></a></p></td> |
| <td><p>Global average pooling operation for 3D data (spatial or spatio-temporal).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D" title="mxnet.gluon.nn.ReflectionPad2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.ReflectionPad2D</span></code></a></p></td> |
| <td><p>Pads the input tensor using the reflection of the input boundary.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="normalization-layers"> |
| <h2>Normalization Layers<a class="headerlink" href="#normalization-layers" title="Permalink to this headline">¶</a></h2> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.BatchNorm" title="mxnet.gluon.nn.BatchNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.BatchNorm</span></code></a></p></td> |
| <td><p>Batch normalization layer (Ioffe and Szegedy, 2014).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm" title="mxnet.gluon.nn.InstanceNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.InstanceNorm</span></code></a></p></td> |
| <td><p>Applies instance normalization to the n-dimensional input array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm" title="mxnet.gluon.nn.LayerNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.LayerNorm</span></code></a></p></td> |
| <td><p>Applies layer normalization to the n-dimensional input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm" title="mxnet.gluon.nn.SyncBatchNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.SyncBatchNorm</span></code></a></p></td> |
| <td><p>Cross-GPU Synchronized Batch normalization (SyncBN)</p></td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="embedding-layers"> |
| <h2>Embedding Layers<a class="headerlink" href="#embedding-layers" title="Permalink to this headline">¶</a></h2> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding" title="mxnet.gluon.nn.Embedding"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Embedding</span></code></a></p></td> |
| <td><p>Turns non-negative integers (indexes/tokens) into dense vectors of fixed size.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="advanced-activation-layers"> |
| <h2>Advanced Activation Layers<a class="headerlink" href="#advanced-activation-layers" title="Permalink to this headline">¶</a></h2> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU" title="mxnet.gluon.nn.LeakyReLU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.LeakyReLU</span></code></a></p></td> |
| <td><p>Leaky version of a Rectified Linear Unit.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU" title="mxnet.gluon.nn.PReLU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.PReLU</span></code></a></p></td> |
| <td><p>Parametric leaky version of a Rectified Linear Unit.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU" title="mxnet.gluon.nn.ELU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.ELU</span></code></a></p></td> |
| <td><p>Exponential Linear Unit (ELU)</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU" title="mxnet.gluon.nn.SELU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.SELU</span></code></a></p></td> |
| <td><p>Scaled Exponential Linear Unit (SELU)</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish" title="mxnet.gluon.nn.Swish"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.Swish</span></code></a></p></td> |
| <td><p>Swish Activation function (SiLU with a hyperparameter)</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU" title="mxnet.gluon.nn.SiLU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.SiLU</span></code></a></p></td> |
| <td><p>Sigmoid Linear Units</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU" title="mxnet.gluon.nn.GELU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nn.GELU</span></code></a></p></td> |
| <td><p>Gaussian Exponential Linear Unit (GELU)</p></td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="module-mxnet.gluon.nn"> |
| <span id="api-reference"></span><h2>API Reference<a class="headerlink" href="#module-mxnet.gluon.nn" title="Permalink to this headline">¶</a></h2> |
| <p>Neural network layers.</p> |
| <p><strong>Classes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation" title="mxnet.gluon.nn.Activation"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Activation</span></code></a>(activation, **kwargs)</p></td> |
| <td><p>Applies an activation function to input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.AvgPool1D" title="mxnet.gluon.nn.AvgPool1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">AvgPool1D</span></code></a>([pool_size, strides, padding, …])</p></td> |
| <td><p>Average pooling operation for temporal data.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.AvgPool2D" title="mxnet.gluon.nn.AvgPool2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">AvgPool2D</span></code></a>([pool_size, strides, padding, …])</p></td> |
| <td><p>Average pooling operation for spatial data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.AvgPool3D" title="mxnet.gluon.nn.AvgPool3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">AvgPool3D</span></code></a>([pool_size, strides, padding, …])</p></td> |
| <td><p>Average pooling operation for 3D data (spatial or spatio-temporal).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.BatchNorm" title="mxnet.gluon.nn.BatchNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">BatchNorm</span></code></a>([axis, momentum, epsilon, center, …])</p></td> |
| <td><p>Batch normalization layer (Ioffe and Szegedy, 2014).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Block</span></code></a>()</p></td> |
| <td><p>Base class for all neural network layers and models.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate" title="mxnet.gluon.nn.Concatenate"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Concatenate</span></code></a>([axis])</p></td> |
| <td><p>Lays <cite>Block</cite> s concurrently.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv1D" title="mxnet.gluon.nn.Conv1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Conv1D</span></code></a>(channels, kernel_size[, strides, …])</p></td> |
| <td><p>1D convolution layer (e.g. temporal convolution).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv1DTranspose" title="mxnet.gluon.nn.Conv1DTranspose"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Conv1DTranspose</span></code></a>(channels, kernel_size[, …])</p></td> |
| <td><p>Transposed 1D convolution layer (sometimes called Deconvolution).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv2D" title="mxnet.gluon.nn.Conv2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Conv2D</span></code></a>(channels, kernel_size[, strides, …])</p></td> |
| <td><p>2D convolution layer (e.g. spatial convolution over images).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv2DTranspose" title="mxnet.gluon.nn.Conv2DTranspose"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Conv2DTranspose</span></code></a>(channels, kernel_size[, …])</p></td> |
| <td><p>Transposed 2D convolution layer (sometimes called Deconvolution).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv3D" title="mxnet.gluon.nn.Conv3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Conv3D</span></code></a>(channels, kernel_size[, strides, …])</p></td> |
| <td><p>3D convolution layer (e.g. spatial convolution over volumes).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Conv3DTranspose" title="mxnet.gluon.nn.Conv3DTranspose"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Conv3DTranspose</span></code></a>(channels, kernel_size[, …])</p></td> |
| <td><p>Transposed 3D convolution layer (sometimes called Deconvolution).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution" title="mxnet.gluon.nn.DeformableConvolution"><code class="xref py py-obj docutils literal notranslate"><span class="pre">DeformableConvolution</span></code></a>(channels[, …])</p></td> |
| <td><p>2-D Deformable Convolution v_1 (Dai, 2017).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense" title="mxnet.gluon.nn.Dense"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Dense</span></code></a>(units[, activation, use_bias, …])</p></td> |
| <td><p>Just your regular densely-connected NN layer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout" title="mxnet.gluon.nn.Dropout"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Dropout</span></code></a>(rate[, axes])</p></td> |
| <td><p>Applies Dropout to the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU" title="mxnet.gluon.nn.ELU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ELU</span></code></a>([alpha])</p></td> |
| <td><p>Exponential Linear Unit (ELU)</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding" title="mxnet.gluon.nn.Embedding"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Embedding</span></code></a>(input_dim, output_dim[, dtype, …])</p></td> |
| <td><p>Turns non-negative integers (indexes/tokens) into dense vectors of fixed size.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten" title="mxnet.gluon.nn.Flatten"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Flatten</span></code></a>(**kwargs)</p></td> |
| <td><p>Flattens the input to two dimensional.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU" title="mxnet.gluon.nn.GELU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GELU</span></code></a>([approximation])</p></td> |
| <td><p>Gaussian Exponential Linear Unit (GELU)</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalAvgPool1D" title="mxnet.gluon.nn.GlobalAvgPool1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GlobalAvgPool1D</span></code></a>([layout])</p></td> |
| <td><p>Global average pooling operation for temporal data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalAvgPool2D" title="mxnet.gluon.nn.GlobalAvgPool2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GlobalAvgPool2D</span></code></a>([layout])</p></td> |
| <td><p>Global average pooling operation for spatial data.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalAvgPool3D" title="mxnet.gluon.nn.GlobalAvgPool3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GlobalAvgPool3D</span></code></a>([layout])</p></td> |
| <td><p>Global average pooling operation for 3D data (spatial or spatio-temporal).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalMaxPool1D" title="mxnet.gluon.nn.GlobalMaxPool1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GlobalMaxPool1D</span></code></a>([layout])</p></td> |
| <td><p>Gloabl max pooling operation for one dimensional (temporal) data.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalMaxPool2D" title="mxnet.gluon.nn.GlobalMaxPool2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GlobalMaxPool2D</span></code></a>([layout])</p></td> |
| <td><p>Global max pooling operation for two dimensional (spatial) data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GlobalMaxPool3D" title="mxnet.gluon.nn.GlobalMaxPool3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GlobalMaxPool3D</span></code></a>([layout])</p></td> |
| <td><p>Global max pooling operation for 3D data (spatial or spatio-temporal).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm" title="mxnet.gluon.nn.GroupNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GroupNorm</span></code></a>([num_groups, epsilon, center, …])</p></td> |
| <td><p>Applies group normalization to the n-dimensional input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-obj docutils literal notranslate"><span class="pre">HybridBlock</span></code></a>()</p></td> |
| <td><p><cite>HybridBlock</cite> supports forwarding with both Symbol and NDArray.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate" title="mxnet.gluon.nn.HybridConcatenate"><code class="xref py py-obj docutils literal notranslate"><span class="pre">HybridConcatenate</span></code></a>([axis])</p></td> |
| <td><p>Lays <cite>HybridBlock</cite> s concurrently.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda" title="mxnet.gluon.nn.HybridLambda"><code class="xref py py-obj docutils literal notranslate"><span class="pre">HybridLambda</span></code></a>(function)</p></td> |
| <td><p>Wraps an operator or an expression as a HybridBlock object.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential" title="mxnet.gluon.nn.HybridSequential"><code class="xref py py-obj docutils literal notranslate"><span class="pre">HybridSequential</span></code></a>()</p></td> |
| <td><p>Stacks HybridBlocks sequentially.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity" title="mxnet.gluon.nn.Identity"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Identity</span></code></a>()</p></td> |
| <td><p>Block that passes through the input directly.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm" title="mxnet.gluon.nn.InstanceNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">InstanceNorm</span></code></a>([axis, epsilon, center, scale, …])</p></td> |
| <td><p>Applies instance normalization to the n-dimensional input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Lambda" title="mxnet.gluon.nn.Lambda"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Lambda</span></code></a>(function)</p></td> |
| <td><p>Wraps an operator or an expression as a Block object.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm" title="mxnet.gluon.nn.LayerNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">LayerNorm</span></code></a>([axis, epsilon, center, scale, …])</p></td> |
| <td><p>Applies layer normalization to the n-dimensional input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU" title="mxnet.gluon.nn.LeakyReLU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">LeakyReLU</span></code></a>(alpha, **kwargs)</p></td> |
| <td><p>Leaky version of a Rectified Linear Unit.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.MaxPool1D" title="mxnet.gluon.nn.MaxPool1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">MaxPool1D</span></code></a>([pool_size, strides, padding, …])</p></td> |
| <td><p>Max pooling operation for one dimensional data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.MaxPool2D" title="mxnet.gluon.nn.MaxPool2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">MaxPool2D</span></code></a>([pool_size, strides, padding, …])</p></td> |
| <td><p>Max pooling operation for two dimensional (spatial) data.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.MaxPool3D" title="mxnet.gluon.nn.MaxPool3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">MaxPool3D</span></code></a>([pool_size, strides, padding, …])</p></td> |
| <td><p>Max pooling operation for 3D data (spatial or spatio-temporal).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution" title="mxnet.gluon.nn.ModulatedDeformableConvolution"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ModulatedDeformableConvolution</span></code></a>(channels[, …])</p></td> |
| <td><p>2-D Deformable Convolution v2 (Dai, 2018).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU" title="mxnet.gluon.nn.PReLU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">PReLU</span></code></a>([alpha_initializer, in_channels])</p></td> |
| <td><p>Parametric leaky version of a Rectified Linear Unit.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D" title="mxnet.gluon.nn.PixelShuffle1D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">PixelShuffle1D</span></code></a>(factor)</p></td> |
| <td><p>Pixel-shuffle layer for upsampling in 1 dimension.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D" title="mxnet.gluon.nn.PixelShuffle2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">PixelShuffle2D</span></code></a>(factor)</p></td> |
| <td><p>Pixel-shuffle layer for upsampling in 2 dimensions.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D" title="mxnet.gluon.nn.PixelShuffle3D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">PixelShuffle3D</span></code></a>(factor)</p></td> |
| <td><p>Pixel-shuffle layer for upsampling in 3 dimensions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D" title="mxnet.gluon.nn.ReflectionPad2D"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ReflectionPad2D</span></code></a>([padding])</p></td> |
| <td><p>Pads the input tensor using the reflection of the input boundary.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU" title="mxnet.gluon.nn.SELU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SELU</span></code></a>(**kwargs)</p></td> |
| <td><p>Scaled Exponential Linear Unit (SELU)</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Sequential" title="mxnet.gluon.nn.Sequential"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Sequential</span></code></a>()</p></td> |
| <td><p>Stacks Blocks sequentially.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU" title="mxnet.gluon.nn.SiLU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SiLU</span></code></a>(**kwargs)</p></td> |
| <td><p>Sigmoid Linear Units</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish" title="mxnet.gluon.nn.Swish"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Swish</span></code></a>([beta])</p></td> |
| <td><p>Swish Activation function (SiLU with a hyperparameter)</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SymbolBlock" title="mxnet.gluon.nn.SymbolBlock"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SymbolBlock</span></code></a>(outputs, inputs[, params])</p></td> |
| <td><p>Construct block from symbol.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm" title="mxnet.gluon.nn.SyncBatchNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SyncBatchNorm</span></code></a>([in_channels, num_devices, …])</p></td> |
| <td><p>Cross-GPU Synchronized Batch normalization (SyncBN)</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Activation"> |
| <em class="property">class </em><code class="sig-name descname">Activation</code><span class="sig-paren">(</span><em class="sig-param">activation</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#Activation"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Activation" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Applies an activation function to input.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>activation</strong> (<em>str</em>) – Name of activation function to use. |
| See <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.Activation" title="mxnet.ndarray.Activation"><code class="xref py py-func docutils literal notranslate"><span class="pre">Activation()</span></code></a> for available choices.</p> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.apply" title="mxnet.gluon.nn.Activation.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.collect_params" title="mxnet.gluon.nn.Activation.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.export" title="mxnet.gluon.nn.Activation.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.forward" title="mxnet.gluon.nn.Activation.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.hybridize" title="mxnet.gluon.nn.Activation.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.infer_shape" title="mxnet.gluon.nn.Activation.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.infer_type" title="mxnet.gluon.nn.Activation.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.initialize" title="mxnet.gluon.nn.Activation.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.load" title="mxnet.gluon.nn.Activation.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.load_dict" title="mxnet.gluon.nn.Activation.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.load_parameters" title="mxnet.gluon.nn.Activation.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.optimize_for" title="mxnet.gluon.nn.Activation.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.register_forward_hook" title="mxnet.gluon.nn.Activation.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.register_forward_pre_hook" title="mxnet.gluon.nn.Activation.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.register_op_hook" title="mxnet.gluon.nn.Activation.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.reset_ctx" title="mxnet.gluon.nn.Activation.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.reset_device" title="mxnet.gluon.nn.Activation.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.save" title="mxnet.gluon.nn.Activation.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.save_parameters" title="mxnet.gluon.nn.Activation.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.setattr" title="mxnet.gluon.nn.Activation.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.share_parameters" title="mxnet.gluon.nn.Activation.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.summary" title="mxnet.gluon.nn.Activation.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.zero_grad" title="mxnet.gluon.nn.Activation.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Activation.params" title="mxnet.gluon.nn.Activation.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#Activation.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Activation.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.Activation.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.Activation.forward" title="mxnet.gluon.nn.Activation.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.Activation.forward" title="mxnet.gluon.nn.Activation.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Activation.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Activation.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.AvgPool1D"> |
| <em class="property">class </em><code class="sig-name descname">AvgPool1D</code><span class="sig-paren">(</span><em class="sig-param">pool_size=2</em>, <em class="sig-param">strides=None</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">layout='NCW'</em>, <em class="sig-param">ceil_mode=False</em>, <em class="sig-param">count_include_pad=True</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#AvgPool1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.AvgPool1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Average pooling operation for temporal data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>pool_size</strong> (<em>int</em>) – Size of the average pooling windows.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em>, or </em><em>None</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and out (‘NCW’ or ‘NWC’). |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. padding is applied on ‘W’ dimension.</p></li> |
| <li><p><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When <cite>True</cite>, will use ceil instead of floor to compute the output shape.</p></li> |
| <li><p><strong>count_include_pad</strong> (<em>bool</em><em>, </em><em>default True</em>) – When ‘False’, will exclude padding elements when computing the average value.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, out_width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. out_width is calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="o">-</span><span class="n">pool_size</span><span class="p">)</span><span class="o">/</span><span class="n">strides</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.AvgPool2D"> |
| <em class="property">class </em><code class="sig-name descname">AvgPool2D</code><span class="sig-paren">(</span><em class="sig-param">pool_size=(2</em>, <em class="sig-param">2)</em>, <em class="sig-param">strides=None</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">ceil_mode=False</em>, <em class="sig-param">layout='NCHW'</em>, <em class="sig-param">count_include_pad=True</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#AvgPool2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.AvgPool2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Average pooling operation for spatial data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>pool_size</strong> (<em>int</em><em> or </em><em>list/tuple of 2 ints</em><em>,</em>) – Size of the average pooling windows.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em>, </em><em>list/tuple of 2 ints</em><em>, or </em><em>None.</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>list/tuple of 2 ints</em><em>,</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and out (‘NCHW’ or ‘NHWC’). |
| ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, and width |
| dimensions respectively. padding is applied on ‘H’ and ‘W’ dimension.</p></li> |
| <li><p><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When True, will use ceil instead of floor to compute the output shape.</p></li> |
| <li><p><strong>count_include_pad</strong> (<em>bool</em><em>, </em><em>default True</em>) – When ‘False’, will exclude padding elements when computing the average value.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.AvgPool3D"> |
| <em class="property">class </em><code class="sig-name descname">AvgPool3D</code><span class="sig-paren">(</span><em class="sig-param">pool_size=(2</em>, <em class="sig-param">2</em>, <em class="sig-param">2)</em>, <em class="sig-param">strides=None</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">ceil_mode=False</em>, <em class="sig-param">layout='NCDHW'</em>, <em class="sig-param">count_include_pad=True</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#AvgPool3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.AvgPool3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Average pooling operation for 3D data (spatial or spatio-temporal).</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>pool_size</strong> (<em>int</em><em> or </em><em>list/tuple of 3 ints</em><em>,</em>) – Size of the average pooling windows.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em>, </em><em>list/tuple of 3 ints</em><em>, or </em><em>None.</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>list/tuple of 3 ints</em><em>,</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and out (‘NCDHW’ or ‘NDHWC’). |
| ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, width and |
| depth dimensions respectively. padding is applied on ‘D’, ‘H’ and ‘W’ |
| dimension.</p></li> |
| <li><p><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When True, will use ceil instead of floor to compute the output shape.</p></li> |
| <li><p><strong>count_include_pad</strong> (<em>bool</em><em>, </em><em>default True</em>) – When ‘False’, will exclude padding elements when computing the average value.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, out_depth, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>. |
| out_depth, out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_depth</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">depth</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True,</cite> ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.BatchNorm"> |
| <em class="property">class </em><code class="sig-name descname">BatchNorm</code><span class="sig-paren">(</span><em class="sig-param">axis=1</em>, <em class="sig-param">momentum=0.9</em>, <em class="sig-param">epsilon=1e-05</em>, <em class="sig-param">center=True</em>, <em class="sig-param">scale=True</em>, <em class="sig-param">use_global_stats=False</em>, <em class="sig-param">beta_initializer='zeros'</em>, <em class="sig-param">gamma_initializer='ones'</em>, <em class="sig-param">running_mean_initializer='zeros'</em>, <em class="sig-param">running_variance_initializer='ones'</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#BatchNorm"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.BatchNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.basic_layers._BatchNorm</span></code></p> |
| <p>Batch normalization layer (Ioffe and Szegedy, 2014). |
| Normalizes the input at each batch, i.e. applies a transformation |
| that maintains the mean activation close to 0 and the activation |
| standard deviation close to 1.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>default 1</em>) – The axis that should be normalized. This is typically the channels |
| (C) axis. For instance, after a <cite>Conv2D</cite> layer with <cite>layout=’NCHW’</cite>, |
| set <cite>axis=1</cite> in <cite>BatchNorm</cite>. If <cite>layout=’NHWC’</cite>, then set <cite>axis=3</cite>.</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>default 0.9</em>) – Momentum for the moving average.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>default 1e-5</em>) – Small float added to variance to avoid dividing by zero.</p></li> |
| <li><p><strong>center</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, add offset of <cite>beta</cite> to normalized tensor. |
| If False, <cite>beta</cite> is ignored.</p></li> |
| <li><p><strong>scale</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, multiply by <cite>gamma</cite>. If False, <cite>gamma</cite> is not used. |
| When the next layer is linear (also e.g. <cite>nn.relu</cite>), |
| this can be disabled since the scaling |
| will be done by the next layer.</p></li> |
| <li><p><strong>use_global_stats</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, use global moving statistics instead of local batch-norm. This will force |
| change batch-norm into a scale shift operator. |
| If False, use local batch-norm.</p></li> |
| <li><p><strong>beta_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the beta weight.</p></li> |
| <li><p><strong>gamma_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the gamma weight.</p></li> |
| <li><p><strong>running_mean_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the running mean.</p></li> |
| <li><p><strong>running_variance_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the running variance.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – Number of channels (feature maps) in input data. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Block"> |
| <em class="property">class </em><code class="sig-name descname">Block</code><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">object</span></code></p> |
| <p>Base class for all neural network layers and models. Your models should |
| subclass this class.</p> |
| <p><a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> can be nested recursively in a tree structure. You can create and |
| assign child <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> as regular attributes:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">mxnet</span> <span class="k">as</span> <span class="nn">mx</span> |
| <span class="kn">from</span> <span class="nn">mxnet.gluon</span> <span class="kn">import</span> <span class="n">Block</span><span class="p">,</span> <span class="n">nn</span> |
| |
| <span class="k">class</span> <span class="nc">Model</span><span class="p">(</span><span class="n">Block</span><span class="p">):</span> |
| <span class="k">def</span> <span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">):</span> |
| <span class="nb">super</span><span class="p">(</span><span class="n">Model</span><span class="p">,</span> <span class="bp">self</span><span class="p">)</span><span class="o">.</span><span class="fm">__init__</span><span class="p">(</span><span class="o">**</span><span class="n">kwargs</span><span class="p">)</span> |
| <span class="bp">self</span><span class="o">.</span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="bp">self</span><span class="o">.</span><span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| |
| <span class="k">def</span> <span class="nf">forward</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">x</span><span class="p">):</span> |
| <span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">npx</span><span class="o">.</span><span class="n">relu</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">dense0</span><span class="p">(</span><span class="n">x</span><span class="p">))</span> |
| <span class="k">return</span> <span class="n">mx</span><span class="o">.</span><span class="n">npx</span><span class="o">.</span><span class="n">relu</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">dense1</span><span class="p">(</span><span class="n">x</span><span class="p">))</span> |
| |
| <span class="n">model</span> <span class="o">=</span> <span class="n">Model</span><span class="p">()</span> |
| <span class="n">model</span><span class="o">.</span><span class="n">initialize</span><span class="p">(</span><span class="n">device</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">))</span> |
| <span class="n">model</span><span class="p">(</span><span class="n">mx</span><span class="o">.</span><span class="n">np</span><span class="o">.</span><span class="n">zeros</span><span class="p">((</span><span class="mi">10</span><span class="p">,</span> <span class="mi">10</span><span class="p">),</span> <span class="n">device</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">)))</span> |
| </pre></div> |
| </div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.apply" title="mxnet.gluon.nn.Block.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.cast" title="mxnet.gluon.nn.Block.cast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cast</span></code></a>(dtype)</p></td> |
| <td><p>Cast this Block to use another data type.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.collect_params" title="mxnet.gluon.nn.Block.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.forward" title="mxnet.gluon.nn.Block.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(*args)</p></td> |
| <td><p>Overrides to implement forward computation using <code class="xref py py-class docutils literal notranslate"><span class="pre">NDArray</span></code>.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.hybridize" title="mxnet.gluon.nn.Block.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active])</p></td> |
| <td><p>Please refer description of HybridBlock hybridize().</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.initialize" title="mxnet.gluon.nn.Block.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.load" title="mxnet.gluon.nn.Block.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.load_dict" title="mxnet.gluon.nn.Block.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.load_parameters" title="mxnet.gluon.nn.Block.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.register_child" title="mxnet.gluon.nn.Block.register_child"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_child</span></code></a>(block[, name])</p></td> |
| <td><p>Registers block as a child of self.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.register_forward_hook" title="mxnet.gluon.nn.Block.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.register_forward_pre_hook" title="mxnet.gluon.nn.Block.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.register_op_hook" title="mxnet.gluon.nn.Block.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install callback monitor.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.reset_ctx" title="mxnet.gluon.nn.Block.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.reset_device" title="mxnet.gluon.nn.Block.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.save" title="mxnet.gluon.nn.Block.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.save_parameters" title="mxnet.gluon.nn.Block.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.setattr" title="mxnet.gluon.nn.Block.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.share_parameters" title="mxnet.gluon.nn.Block.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.summary" title="mxnet.gluon.nn.Block.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.zero_grad" title="mxnet.gluon.nn.Block.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Block.params" title="mxnet.gluon.nn.Block.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p>Child <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> assigned this way will be registered and <a class="reference internal" href="#mxnet.gluon.nn.Block.collect_params" title="mxnet.gluon.nn.Block.collect_params"><code class="xref py py-meth docutils literal notranslate"><span class="pre">collect_params()</span></code></a> |
| will collect their Parameters recursively. You can also manually register |
| child blocks with <a class="reference internal" href="#mxnet.gluon.nn.Block.register_child" title="mxnet.gluon.nn.Block.register_child"><code class="xref py py-meth docutils literal notranslate"><span class="pre">register_child()</span></code></a>.</p> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.apply"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.cast"> |
| <code class="sig-name descname">cast</code><span class="sig-paren">(</span><em class="sig-param">dtype</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.cast"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.cast" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Cast this Block to use another data type.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>dtype</strong> (<em>str</em><em> or </em><em>numpy.dtype</em>) – The new data type.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.collect_params"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides to implement forward computation using <code class="xref py py-class docutils literal notranslate"><span class="pre">NDArray</span></code>. Only |
| accepts positional arguments.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>*args</strong> (<em>list of NDArray</em>) – Input tensors.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.hybridize"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Please refer description of HybridBlock hybridize().</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.initialize"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.load"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.load_dict"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.load_parameters"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.Block.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.register_child"> |
| <code class="sig-name descname">register_child</code><span class="sig-paren">(</span><em class="sig-param">block</em>, <em class="sig-param">name=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.register_child"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.register_child" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers block as a child of self. <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> s assigned to self as |
| attributes will be registered automatically.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.register_forward_hook"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.Block.forward" title="mxnet.gluon.nn.Block.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.register_forward_pre_hook"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.Block.forward" title="mxnet.gluon.nn.Block.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.register_op_hook"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install callback monitor.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.reset_ctx"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">Block.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.reset_device"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.save"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.save_parameters"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.setattr"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.share_parameters"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.summary"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Block.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#Block.zero_grad"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Block.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Concatenate"> |
| <em class="property">class </em><code class="sig-name descname">Concatenate</code><span class="sig-paren">(</span><em class="sig-param">axis=-1</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Concatenate"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Concatenate" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.basic_layers.Sequential</span></code></p> |
| <p>Lays <cite>Block</cite> s concurrently.</p> |
| <p>This block feeds its input to all children blocks, and |
| produce the output by concatenating all the children blocks’ outputs |
| on the specified axis.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">net</span> <span class="o">=</span> <span class="n">Concatenate</span><span class="p">()</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">activation</span><span class="o">=</span><span class="s1">'relu'</span><span class="p">))</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">))</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">Identity</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.add" title="mxnet.gluon.nn.Concatenate.add"><code class="xref py py-obj docutils literal notranslate"><span class="pre">add</span></code></a>(*blocks)</p></td> |
| <td><p>Adds block on top of the stack.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.apply" title="mxnet.gluon.nn.Concatenate.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.cast" title="mxnet.gluon.nn.Concatenate.cast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cast</span></code></a>(dtype)</p></td> |
| <td><p>Cast this Block to use another data type.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.collect_params" title="mxnet.gluon.nn.Concatenate.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.forward" title="mxnet.gluon.nn.Concatenate.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides to implement forward computation using <code class="xref py py-class docutils literal notranslate"><span class="pre">NDArray</span></code>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.hybridize" title="mxnet.gluon.nn.Concatenate.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active])</p></td> |
| <td><p>Activates or deactivates <cite>HybridBlock</cite> s recursively.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.initialize" title="mxnet.gluon.nn.Concatenate.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.load" title="mxnet.gluon.nn.Concatenate.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.load_dict" title="mxnet.gluon.nn.Concatenate.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.load_parameters" title="mxnet.gluon.nn.Concatenate.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.register_child" title="mxnet.gluon.nn.Concatenate.register_child"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_child</span></code></a>(block[, name])</p></td> |
| <td><p>Registers block as a child of self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.register_forward_hook" title="mxnet.gluon.nn.Concatenate.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.register_forward_pre_hook" title="mxnet.gluon.nn.Concatenate.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.register_op_hook" title="mxnet.gluon.nn.Concatenate.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install callback monitor.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.reset_ctx" title="mxnet.gluon.nn.Concatenate.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.reset_device" title="mxnet.gluon.nn.Concatenate.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.save" title="mxnet.gluon.nn.Concatenate.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.save_parameters" title="mxnet.gluon.nn.Concatenate.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.setattr" title="mxnet.gluon.nn.Concatenate.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.share_parameters" title="mxnet.gluon.nn.Concatenate.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.summary" title="mxnet.gluon.nn.Concatenate.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.zero_grad" title="mxnet.gluon.nn.Concatenate.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Concatenate.params" title="mxnet.gluon.nn.Concatenate.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>axis</strong> (<em>int</em><em>, </em><em>default -1</em>) – The axis on which to concatenate the outputs.</p> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.add"> |
| <code class="sig-name descname">add</code><span class="sig-paren">(</span><em class="sig-param">*blocks</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.add" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Adds block on top of the stack.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.cast"> |
| <code class="sig-name descname">cast</code><span class="sig-paren">(</span><em class="sig-param">dtype</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.cast" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Cast this Block to use another data type.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>dtype</strong> (<em>str</em><em> or </em><em>numpy.dtype</em>) – The new data type.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Concatenate.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides to implement forward computation using <code class="xref py py-class docutils literal notranslate"><span class="pre">NDArray</span></code>. Only |
| accepts positional arguments.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>*args</strong> (<em>list of NDArray</em>) – Input tensors.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <cite>HybridBlock</cite> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>string</em>) – Additional flags for hybridized operator.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.register_child"> |
| <code class="sig-name descname">register_child</code><span class="sig-paren">(</span><em class="sig-param">block</em>, <em class="sig-param">name=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.register_child" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers block as a child of self. <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> s assigned to self as |
| attributes will be registered automatically.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.Concatenate.forward" title="mxnet.gluon.nn.Concatenate.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.Concatenate.forward" title="mxnet.gluon.nn.Concatenate.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install callback monitor.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">Block.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Concatenate.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Concatenate.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv1D"> |
| <em class="property">class </em><code class="sig-name descname">Conv1D</code><span class="sig-paren">(</span><em class="sig-param">channels</em>, <em class="sig-param">kernel_size</em>, <em class="sig-param">strides=1</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">dilation=1</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">layout='NCW'</em>, <em class="sig-param">activation=None</em>, <em class="sig-param">use_bias=True</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">bias_initializer='zeros'</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Conv</span></code></p> |
| <p>1D convolution layer (e.g. temporal convolution).</p> |
| <p>This layer creates a convolution kernel that is convolved |
| with the layer input over a single spatial (or temporal) dimension |
| to produce a tensor of outputs. |
| If <cite>use_bias</cite> is True, a bias vector is created and added to the outputs. |
| Finally, if <cite>activation</cite> is not <cite>None</cite>, |
| it is applied to the outputs as well.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</p></li> |
| <li><p><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 1 int</em>) – Specifies the dimensions of the convolution window.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 1 int</em><em>,</em>) – Specify the strides of the convolution.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 1 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</p></li> |
| <li><p><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 1 int</em>) – Specifies the dilation rate to use for dilated convolution.</p></li> |
| <li><p><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and weight. Only supports ‘NCW’ layout for now. |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. Convolution is applied on the ‘W’ dimension.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| <li><p><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../../npx/generated/mxnet.npx.activation.html#mxnet.npx.activation" title="mxnet.npx.activation"><code class="xref py py-func docutils literal notranslate"><span class="pre">activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</p></li> |
| <li><p><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</p></li> |
| <li><p><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</p></li> |
| <li><p><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, out_width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. out_width is calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="o">-</span><span class="n">dilation</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv1DTranspose"> |
| <em class="property">class </em><code class="sig-name descname">Conv1DTranspose</code><span class="sig-paren">(</span><em class="sig-param">channels</em>, <em class="sig-param">kernel_size</em>, <em class="sig-param">strides=1</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">output_padding=0</em>, <em class="sig-param">dilation=1</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">layout='NCW'</em>, <em class="sig-param">activation=None</em>, <em class="sig-param">use_bias=True</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">bias_initializer='zeros'</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv1DTranspose"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv1DTranspose" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Conv</span></code></p> |
| <p>Transposed 1D convolution layer (sometimes called Deconvolution).</p> |
| <p>The need for transposed convolutions generally arises |
| from the desire to use a transformation going in the opposite direction |
| of a normal convolution, i.e., from something that has the shape of the |
| output of some convolution to something that has the shape of its input |
| while maintaining a connectivity pattern that is compatible with |
| said convolution.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</p></li> |
| <li><p><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 1 int</em>) – Specifies the dimensions of the convolution window.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 1 int</em>) – Specify the strides of the convolution.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 1 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</p></li> |
| <li><p><strong>output_padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 1 int</em>) – Controls the amount of implicit zero-paddings on both sides of the |
| output for output_padding number of points for each dimension.</p></li> |
| <li><p><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 1 int</em>) – Controls the spacing between the kernel points; also known as the |
| a trous algorithm</p></li> |
| <li><p><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and weight. Only supports ‘NCW’ layout for now. |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. Convolution is applied on the ‘W’ dimension.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| <li><p><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../../npx/generated/mxnet.npx.activation.html#mxnet.npx.activation" title="mxnet.npx.activation"><code class="xref py py-func docutils literal notranslate"><span class="pre">activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</p></li> |
| <li><p><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</p></li> |
| <li><p><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</p></li> |
| <li><p><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, out_width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. out_width is calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_width</span> <span class="o">=</span> <span class="p">(</span><span class="n">width</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="o">+</span><span class="n">kernel_size</span><span class="o">+</span><span class="n">output_padding</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv2D"> |
| <em class="property">class </em><code class="sig-name descname">Conv2D</code><span class="sig-paren">(</span><em class="sig-param">channels</em>, <em class="sig-param">kernel_size</em>, <em class="sig-param">strides=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">padding=(0</em>, <em class="sig-param">0)</em>, <em class="sig-param">dilation=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">layout='NCHW'</em>, <em class="sig-param">activation=None</em>, <em class="sig-param">use_bias=True</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">bias_initializer='zeros'</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Conv</span></code></p> |
| <p>2D convolution layer (e.g. spatial convolution over images).</p> |
| <p>This layer creates a convolution kernel that is convolved |
| with the layer input to produce a tensor of |
| outputs. If <cite>use_bias</cite> is True, |
| a bias vector is created and added to the outputs. Finally, if |
| <cite>activation</cite> is not <cite>None</cite>, it is applied to the outputs as well.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</p></li> |
| <li><p><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 2 int</em>) – Specifies the dimensions of the convolution window.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 2 int</em><em>,</em>) – Specify the strides of the convolution.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 2 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</p></li> |
| <li><p><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 2 int</em>) – Specifies the dilation rate to use for dilated convolution.</p></li> |
| <li><p><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and weight. Only supports ‘NCHW’ and ‘NHWC’ |
| layout for now. ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, |
| and width dimensions respectively. Convolution is applied on the ‘H’ and |
| ‘W’ dimensions.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| <li><p><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../../npx/generated/mxnet.npx.activation.html#mxnet.npx.activation" title="mxnet.npx.activation"><code class="xref py py-func docutils literal notranslate"><span class="pre">activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</p></li> |
| <li><p><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</p></li> |
| <li><p><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</p></li> |
| <li><p><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv2DTranspose"> |
| <em class="property">class </em><code class="sig-name descname">Conv2DTranspose</code><span class="sig-paren">(</span><em class="sig-param">channels</em>, <em class="sig-param">kernel_size</em>, <em class="sig-param">strides=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">padding=(0</em>, <em class="sig-param">0)</em>, <em class="sig-param">output_padding=(0</em>, <em class="sig-param">0)</em>, <em class="sig-param">dilation=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">layout='NCHW'</em>, <em class="sig-param">activation=None</em>, <em class="sig-param">use_bias=True</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">bias_initializer='zeros'</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv2DTranspose"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv2DTranspose" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Conv</span></code></p> |
| <p>Transposed 2D convolution layer (sometimes called Deconvolution).</p> |
| <p>The need for transposed convolutions generally arises |
| from the desire to use a transformation going in the opposite direction |
| of a normal convolution, i.e., from something that has the shape of the |
| output of some convolution to something that has the shape of its input |
| while maintaining a connectivity pattern that is compatible with |
| said convolution.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</p></li> |
| <li><p><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 2 int</em>) – Specifies the dimensions of the convolution window.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 2 int</em>) – Specify the strides of the convolution.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 2 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</p></li> |
| <li><p><strong>output_padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 2 int</em>) – Controls the amount of implicit zero-paddings on both sides of the |
| output for output_padding number of points for each dimension.</p></li> |
| <li><p><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 2 int</em>) – Controls the spacing between the kernel points; also known as the |
| a trous algorithm</p></li> |
| <li><p><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and weight. Only supports ‘NCHW’ and ‘NHWC’ |
| layout for now. ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, |
| and width dimensions respectively. Convolution is applied on the ‘H’ and |
| ‘W’ dimensions.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| <li><p><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../../npx/generated/mxnet.npx.activation.html#mxnet.npx.activation" title="mxnet.npx.activation"><code class="xref py py-func docutils literal notranslate"><span class="pre">activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</p></li> |
| <li><p><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</p></li> |
| <li><p><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</p></li> |
| <li><p><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="p">(</span><span class="n">height</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="p">(</span><span class="n">width</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv3D"> |
| <em class="property">class </em><code class="sig-name descname">Conv3D</code><span class="sig-paren">(</span><em class="sig-param">channels</em>, <em class="sig-param">kernel_size</em>, <em class="sig-param">strides=(1</em>, <em class="sig-param">1</em>, <em class="sig-param">1)</em>, <em class="sig-param">padding=(0</em>, <em class="sig-param">0</em>, <em class="sig-param">0)</em>, <em class="sig-param">dilation=(1</em>, <em class="sig-param">1</em>, <em class="sig-param">1)</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">layout='NCDHW'</em>, <em class="sig-param">activation=None</em>, <em class="sig-param">use_bias=True</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">bias_initializer='zeros'</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Conv</span></code></p> |
| <p>3D convolution layer (e.g. spatial convolution over volumes).</p> |
| <p>This layer creates a convolution kernel that is convolved |
| with the layer input to produce a tensor of |
| outputs. If <cite>use_bias</cite> is <cite>True</cite>, |
| a bias vector is created and added to the outputs. Finally, if |
| <cite>activation</cite> is not <cite>None</cite>, it is applied to the outputs as well.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</p></li> |
| <li><p><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dimensions of the convolution window.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em><em>,</em>) – Specify the strides of the convolution.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 3 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</p></li> |
| <li><p><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dilation rate to use for dilated convolution.</p></li> |
| <li><p><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and weight. Only supports ‘NCDHW’ and ‘NDHWC’ |
| layout for now. ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, |
| width and depth dimensions respectively. Convolution is applied on the ‘D’, |
| ‘H’ and ‘W’ dimensions.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| <li><p><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../../npx/generated/mxnet.npx.activation.html#mxnet.npx.activation" title="mxnet.npx.activation"><code class="xref py py-func docutils literal notranslate"><span class="pre">activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</p></li> |
| <li><p><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</p></li> |
| <li><p><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</p></li> |
| <li><p><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, out_depth, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>. |
| out_depth, out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_depth</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">depth</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv3DTranspose"> |
| <em class="property">class </em><code class="sig-name descname">Conv3DTranspose</code><span class="sig-paren">(</span><em class="sig-param">channels</em>, <em class="sig-param">kernel_size</em>, <em class="sig-param">strides=(1</em>, <em class="sig-param">1</em>, <em class="sig-param">1)</em>, <em class="sig-param">padding=(0</em>, <em class="sig-param">0</em>, <em class="sig-param">0)</em>, <em class="sig-param">output_padding=(0</em>, <em class="sig-param">0</em>, <em class="sig-param">0)</em>, <em class="sig-param">dilation=(1</em>, <em class="sig-param">1</em>, <em class="sig-param">1)</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">layout='NCDHW'</em>, <em class="sig-param">activation=None</em>, <em class="sig-param">use_bias=True</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">bias_initializer='zeros'</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv3DTranspose"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv3DTranspose" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Conv</span></code></p> |
| <p>Transposed 3D convolution layer (sometimes called Deconvolution).</p> |
| <p>The need for transposed convolutions generally arises |
| from the desire to use a transformation going in the opposite direction |
| of a normal convolution, i.e., from something that has the shape of the |
| output of some convolution to something that has the shape of its input |
| while maintaining a connectivity pattern that is compatible with |
| said convolution.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</p></li> |
| <li><p><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dimensions of the convolution window.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specify the strides of the convolution.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 3 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</p></li> |
| <li><p><strong>output_padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 3 int</em>) – Controls the amount of implicit zero-paddings on both sides of the |
| output for output_padding number of points for each dimension.</p></li> |
| <li><p><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Controls the spacing between the kernel points; also known as the |
| a trous algorithm.</p></li> |
| <li><p><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and weight. Only supports ‘NCDHW’ and ‘NDHWC’ |
| layout for now. ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, |
| width and depth dimensions respectively. Convolution is applied on the ‘D’, |
| ‘H’ and ‘W’ dimensions.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| <li><p><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../../npx/generated/mxnet.npx.activation.html#mxnet.npx.activation" title="mxnet.npx.activation"><code class="xref py py-func docutils literal notranslate"><span class="pre">activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</p></li> |
| <li><p><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</p></li> |
| <li><p><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</p></li> |
| <li><p><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, out_depth, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>. |
| out_depth, out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_depth</span> <span class="o">=</span> <span class="p">(</span><span class="n">depth</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> |
| <span class="n">out_height</span> <span class="o">=</span> <span class="p">(</span><span class="n">height</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="p">(</span><span class="n">width</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution"> |
| <em class="property">class </em><code class="sig-name descname">DeformableConvolution</code><span class="sig-paren">(</span><em class="sig-param">channels</em>, <em class="sig-param">kernel_size=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">strides=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">padding=(0</em>, <em class="sig-param">0)</em>, <em class="sig-param">dilation=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">num_deformable_group=1</em>, <em class="sig-param">layout='NCHW'</em>, <em class="sig-param">use_bias=True</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">activation=None</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">bias_initializer='zeros'</em>, <em class="sig-param">offset_weight_initializer='zeros'</em>, <em class="sig-param">offset_bias_initializer='zeros'</em>, <em class="sig-param">offset_use_bias=True</em>, <em class="sig-param">op_name='DeformableConvolution'</em>, <em class="sig-param">adj=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#DeformableConvolution"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>2-D Deformable Convolution v_1 (Dai, 2017). |
| Normal Convolution uses sampling points in a regular grid, while the sampling |
| points of Deformablem Convolution can be offset. The offset is learned with a |
| separate convolution layer during the training. Both the convolution layer for |
| generating the output features and the offsets are included in this gluon layer.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>channels</strong> (<em>int</em><em>,</em>) – The dimensionality of the output space |
| i.e. the number of output channels in the convolution.</p></li> |
| <li><p><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 2 ints</em><em>, </em><em>(</em><em>Default value =</em><em> (</em><em>1</em><em>,</em><em>1</em><em>)</em><em>)</em>) – Specifies the dimensions of the convolution window.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 2 ints</em><em>, </em><em>(</em><em>Default value =</em><em> (</em><em>1</em><em>,</em><em>1</em><em>)</em><em>)</em>) – Specifies the strides of the convolution.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>tuple/list of 2 ints</em><em>, </em><em>(</em><em>Default value =</em><em> (</em><em>0</em><em>,</em><em>0</em><em>)</em><em>)</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points.</p></li> |
| <li><p><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 2 ints</em><em>, </em><em>(</em><em>Default value =</em><em> (</em><em>1</em><em>,</em><em>1</em><em>)</em><em>)</em>) – Specifies the dilation rate to use for dilated convolution.</p></li> |
| <li><p><strong>groups</strong> (<em>int</em><em>, </em><em>(</em><em>Default value = 1</em><em>)</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two convolution |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</p></li> |
| <li><p><strong>num_deformable_group</strong> (<em>int</em><em>, </em><em>(</em><em>Default value = 1</em><em>)</em>) – Number of deformable group partitions.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>(</em><em>Default value = NCHW</em><em>)</em>) – Dimension ordering of data and weight. Can be ‘NCW’, ‘NWC’, ‘NCHW’, |
| ‘NHWC’, ‘NCDHW’, ‘NDHWC’, etc. ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for |
| batch, channel, height, width and depth dimensions respectively. |
| Convolution is performed over ‘D’, ‘H’, and ‘W’ dimensions.</p></li> |
| <li><p><strong>use_bias</strong> (<em>bool</em><em>, </em><em>(</em><em>Default value = True</em><em>)</em>) – Whether the layer for generating the output features uses a bias vector.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>(</em><em>Default value = 0</em><em>)</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and input channels will be inferred from the shape of input data.</p></li> |
| <li><p><strong>activation</strong> (<em>str</em><em>, </em><em>(</em><em>Default value = None</em><em>)</em>) – Activation function to use. See <a class="reference internal" href="../../npx/generated/mxnet.npx.activation.html#mxnet.npx.activation" title="mxnet.npx.activation"><code class="xref py py-func docutils literal notranslate"><span class="pre">activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</p></li> |
| <li><p><strong>weight_initializer</strong> (str or <cite>Initializer</cite>, (Default value = None)) – Initializer for the <cite>weight</cite> weights matrix for the convolution layer |
| for generating the output features.</p></li> |
| <li><p><strong>bias_initializer</strong> (str or <cite>Initializer</cite>, (Default value = zeros)) – Initializer for the bias vector for the convolution layer |
| for generating the output features.</p></li> |
| <li><p><strong>offset_weight_initializer</strong> (str or <cite>Initializer</cite>, (Default value = zeros)) – Initializer for the <cite>weight</cite> weights matrix for the convolution layer |
| for generating the offset.</p></li> |
| <li><p><strong>offset_bias_initializer</strong> (str or <cite>Initializer</cite>, (Default value = zeros),) – Initializer for the bias vector for the convolution layer |
| for generating the offset.</p></li> |
| <li><p><strong>offset_use_bias</strong> (<em>bool</em><em>, </em><em>(</em><em>Default value = True</em><em>)</em>) – Whether the layer for generating the offset uses a bias vector.</p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><p><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </p></li> |
| <li><p><strong>Outputs</strong> – <ul> |
| <li><p><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.apply" title="mxnet.gluon.nn.DeformableConvolution.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.collect_params" title="mxnet.gluon.nn.DeformableConvolution.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.export" title="mxnet.gluon.nn.DeformableConvolution.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.forward" title="mxnet.gluon.nn.DeformableConvolution.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.hybridize" title="mxnet.gluon.nn.DeformableConvolution.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.infer_shape" title="mxnet.gluon.nn.DeformableConvolution.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(x)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.infer_type" title="mxnet.gluon.nn.DeformableConvolution.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.initialize" title="mxnet.gluon.nn.DeformableConvolution.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.load" title="mxnet.gluon.nn.DeformableConvolution.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.load_dict" title="mxnet.gluon.nn.DeformableConvolution.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.load_parameters" title="mxnet.gluon.nn.DeformableConvolution.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.optimize_for" title="mxnet.gluon.nn.DeformableConvolution.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.pre_infer_offset_weight" title="mxnet.gluon.nn.DeformableConvolution.pre_infer_offset_weight"><code class="xref py py-obj docutils literal notranslate"><span class="pre">pre_infer_offset_weight</span></code></a>()</p></td> |
| <td><p>Pre-infer the shape of offsite weight parameter based on kernel size,</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.pre_infer_weight" title="mxnet.gluon.nn.DeformableConvolution.pre_infer_weight"><code class="xref py py-obj docutils literal notranslate"><span class="pre">pre_infer_weight</span></code></a>()</p></td> |
| <td><p>Pre-infer the shape of weight parameter based on kernel size, group size and channels</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.register_forward_hook" title="mxnet.gluon.nn.DeformableConvolution.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.register_forward_pre_hook" title="mxnet.gluon.nn.DeformableConvolution.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.register_op_hook" title="mxnet.gluon.nn.DeformableConvolution.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.reset_ctx" title="mxnet.gluon.nn.DeformableConvolution.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.reset_device" title="mxnet.gluon.nn.DeformableConvolution.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.save" title="mxnet.gluon.nn.DeformableConvolution.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.save_parameters" title="mxnet.gluon.nn.DeformableConvolution.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.setattr" title="mxnet.gluon.nn.DeformableConvolution.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.share_parameters" title="mxnet.gluon.nn.DeformableConvolution.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.summary" title="mxnet.gluon.nn.DeformableConvolution.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.zero_grad" title="mxnet.gluon.nn.DeformableConvolution.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.params" title="mxnet.gluon.nn.DeformableConvolution.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#DeformableConvolution.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#DeformableConvolution.infer_shape"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.pre_infer_offset_weight"> |
| <code class="sig-name descname">pre_infer_offset_weight</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#DeformableConvolution.pre_infer_offset_weight"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.pre_infer_offset_weight" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Pre-infer the shape of offsite weight parameter based on kernel size, |
| group size and offset channels</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.pre_infer_weight"> |
| <code class="sig-name descname">pre_infer_weight</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#DeformableConvolution.pre_infer_weight"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.pre_infer_weight" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Pre-infer the shape of weight parameter based on kernel size, group size and channels</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.forward" title="mxnet.gluon.nn.DeformableConvolution.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.DeformableConvolution.forward" title="mxnet.gluon.nn.DeformableConvolution.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.DeformableConvolution.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.DeformableConvolution.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Dense"> |
| <em class="property">class </em><code class="sig-name descname">Dense</code><span class="sig-paren">(</span><em class="sig-param">units</em>, <em class="sig-param">activation=None</em>, <em class="sig-param">use_bias=True</em>, <em class="sig-param">flatten=True</em>, <em class="sig-param">dtype='float32'</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">bias_initializer='zeros'</em>, <em class="sig-param">in_units=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Dense"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Dense" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Just your regular densely-connected NN layer.</p> |
| <p><cite>Dense</cite> implements the operation: |
| <cite>output = activation(dot(input, weight.T) + bias)</cite> |
| where <cite>activation</cite> is the element-wise activation function |
| passed as the <cite>activation</cite> argument, <cite>weight</cite> is a weights matrix |
| created by the layer, and <cite>bias</cite> is a bias vector created by the layer |
| (only applicable if <cite>use_bias</cite> is <cite>True</cite>).</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>units</strong> (<em>int</em>) – Dimensionality of the output space.</p></li> |
| <li><p><strong>activation</strong> (<em>str</em>) – Activation function to use. See help on <cite>Activation</cite> layer. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</p></li> |
| <li><p><strong>use_bias</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether the layer uses a bias vector.</p></li> |
| <li><p><strong>flatten</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether the input tensor should be flattened. |
| If true, all but the first axis of input data are collapsed together. |
| If false, all but the last axis of input data are kept the same, and the transformation |
| applies on the last axis.</p></li> |
| <li><p><strong>dtype</strong> (<em>str</em><em> or </em><em>np.dtype</em><em>, </em><em>default 'float32'</em>) – Data type of output embeddings.</p></li> |
| <li><p><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>kernel</cite> weights matrix.</p></li> |
| <li><p><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</p></li> |
| <li><p><strong>in_units</strong> (<em>int</em><em>, </em><em>optional</em>) – Size of the input data. If not specified, initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_units</cite> |
| will be inferred from the shape of input data.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.apply" title="mxnet.gluon.nn.Dense.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.collect_params" title="mxnet.gluon.nn.Dense.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.export" title="mxnet.gluon.nn.Dense.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.forward" title="mxnet.gluon.nn.Dense.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.hybridize" title="mxnet.gluon.nn.Dense.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.infer_shape" title="mxnet.gluon.nn.Dense.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(x, *args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.infer_type" title="mxnet.gluon.nn.Dense.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.initialize" title="mxnet.gluon.nn.Dense.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.load" title="mxnet.gluon.nn.Dense.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.load_dict" title="mxnet.gluon.nn.Dense.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.load_parameters" title="mxnet.gluon.nn.Dense.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.optimize_for" title="mxnet.gluon.nn.Dense.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.register_forward_hook" title="mxnet.gluon.nn.Dense.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.register_forward_pre_hook" title="mxnet.gluon.nn.Dense.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.register_op_hook" title="mxnet.gluon.nn.Dense.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.reset_ctx" title="mxnet.gluon.nn.Dense.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.reset_device" title="mxnet.gluon.nn.Dense.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.save" title="mxnet.gluon.nn.Dense.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.save_parameters" title="mxnet.gluon.nn.Dense.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.setattr" title="mxnet.gluon.nn.Dense.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.share_parameters" title="mxnet.gluon.nn.Dense.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.summary" title="mxnet.gluon.nn.Dense.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.zero_grad" title="mxnet.gluon.nn.Dense.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dense.params" title="mxnet.gluon.nn.Dense.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: if <cite>flatten</cite> is True, <cite>data</cite> should be a tensor with shape |
| <cite>(batch_size, x1, x2, …, xn)</cite>, where x1 * x2 * … * xn is equal to |
| <cite>in_units</cite>. If <cite>flatten</cite> is False, <cite>data</cite> should have shape |
| <cite>(x1, x2, …, xn, in_units)</cite>.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: if <cite>flatten</cite> is True, <cite>out</cite> will be a tensor with shape |
| <cite>(batch_size, units)</cite>. If <cite>flatten</cite> is False, <cite>out</cite> will have shape |
| <cite>(x1, x2, …, xn, units)</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Dense.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Dense.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Dense.infer_shape"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Dense.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.Dense.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.Dense.forward" title="mxnet.gluon.nn.Dense.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.Dense.forward" title="mxnet.gluon.nn.Dense.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dense.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dense.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Dropout"> |
| <em class="property">class </em><code class="sig-name descname">Dropout</code><span class="sig-paren">(</span><em class="sig-param">rate</em>, <em class="sig-param">axes=()</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Dropout"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Dropout" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Applies Dropout to the input.</p> |
| <p>Dropout consists in randomly setting a fraction <cite>rate</cite> of input units |
| to 0 at each update during training time, which helps prevent overfitting.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>rate</strong> (<em>float</em>) – Fraction of the input units to drop. Must be a number between 0 and 1.</p></li> |
| <li><p><strong>axes</strong> (<em>tuple of int</em><em>, </em><em>default</em><em> (</em><em>)</em>) – The axes on which dropout mask is shared. If empty, regular dropout is applied.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.apply" title="mxnet.gluon.nn.Dropout.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.collect_params" title="mxnet.gluon.nn.Dropout.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.export" title="mxnet.gluon.nn.Dropout.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.forward" title="mxnet.gluon.nn.Dropout.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.hybridize" title="mxnet.gluon.nn.Dropout.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.infer_shape" title="mxnet.gluon.nn.Dropout.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.infer_type" title="mxnet.gluon.nn.Dropout.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.initialize" title="mxnet.gluon.nn.Dropout.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.load" title="mxnet.gluon.nn.Dropout.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.load_dict" title="mxnet.gluon.nn.Dropout.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.load_parameters" title="mxnet.gluon.nn.Dropout.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.optimize_for" title="mxnet.gluon.nn.Dropout.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.register_forward_hook" title="mxnet.gluon.nn.Dropout.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.register_forward_pre_hook" title="mxnet.gluon.nn.Dropout.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.register_op_hook" title="mxnet.gluon.nn.Dropout.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.reset_ctx" title="mxnet.gluon.nn.Dropout.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.reset_device" title="mxnet.gluon.nn.Dropout.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.save" title="mxnet.gluon.nn.Dropout.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.save_parameters" title="mxnet.gluon.nn.Dropout.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.setattr" title="mxnet.gluon.nn.Dropout.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.share_parameters" title="mxnet.gluon.nn.Dropout.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.summary" title="mxnet.gluon.nn.Dropout.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.zero_grad" title="mxnet.gluon.nn.Dropout.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Dropout.params" title="mxnet.gluon.nn.Dropout.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="http://www.cs.toronto.edu/~rsalakhu/papers/srivastava14a.pdf">Dropout: A Simple Way to Prevent Neural Networks from Overfitting</a></p> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Dropout.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Dropout.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.Dropout.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.Dropout.forward" title="mxnet.gluon.nn.Dropout.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.Dropout.forward" title="mxnet.gluon.nn.Dropout.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Dropout.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Dropout.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.ELU"> |
| <em class="property">class </em><code class="sig-name descname">ELU</code><span class="sig-paren">(</span><em class="sig-param">alpha=1.0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#ELU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ELU" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <dl class="simple"> |
| <dt>Exponential Linear Unit (ELU)</dt><dd><p>“Fast and Accurate Deep Network Learning by Exponential Linear Units”, Clevert et al, 2016 |
| <a class="reference external" href="https://arxiv.org/abs/1511.07289">https://arxiv.org/abs/1511.07289</a> |
| Published as a conference paper at ICLR 2016</p> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.apply" title="mxnet.gluon.nn.ELU.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.collect_params" title="mxnet.gluon.nn.ELU.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.export" title="mxnet.gluon.nn.ELU.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.forward" title="mxnet.gluon.nn.ELU.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.hybridize" title="mxnet.gluon.nn.ELU.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.infer_shape" title="mxnet.gluon.nn.ELU.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.infer_type" title="mxnet.gluon.nn.ELU.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.initialize" title="mxnet.gluon.nn.ELU.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.load" title="mxnet.gluon.nn.ELU.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.load_dict" title="mxnet.gluon.nn.ELU.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.load_parameters" title="mxnet.gluon.nn.ELU.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.optimize_for" title="mxnet.gluon.nn.ELU.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.register_forward_hook" title="mxnet.gluon.nn.ELU.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.register_forward_pre_hook" title="mxnet.gluon.nn.ELU.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.register_op_hook" title="mxnet.gluon.nn.ELU.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.reset_ctx" title="mxnet.gluon.nn.ELU.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.reset_device" title="mxnet.gluon.nn.ELU.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.save" title="mxnet.gluon.nn.ELU.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.save_parameters" title="mxnet.gluon.nn.ELU.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.setattr" title="mxnet.gluon.nn.ELU.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.share_parameters" title="mxnet.gluon.nn.ELU.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.summary" title="mxnet.gluon.nn.ELU.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.zero_grad" title="mxnet.gluon.nn.ELU.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ELU.params" title="mxnet.gluon.nn.ELU.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>alpha</strong> (<em>float</em>) – The alpha parameter as described by Clevert et al, 2016</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#ELU.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ELU.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.ELU.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.ELU.forward" title="mxnet.gluon.nn.ELU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.ELU.forward" title="mxnet.gluon.nn.ELU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ELU.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ELU.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Embedding"> |
| <em class="property">class </em><code class="sig-name descname">Embedding</code><span class="sig-paren">(</span><em class="sig-param">input_dim</em>, <em class="sig-param">output_dim</em>, <em class="sig-param">dtype='float32'</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">sparse_grad=False</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Embedding"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Embedding" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Turns non-negative integers (indexes/tokens) into dense vectors |
| of fixed size. eg. [4, 20] -> [[0.25, 0.1], [0.6, -0.2]]</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>if <cite>sparse_grad</cite> is set to True, the gradient w.r.t weight will be |
| sparse. Only a subset of optimizers support sparse gradients, including SGD, |
| AdaGrad and Adam. By default lazy updates is turned on, which may perform |
| differently from standard updates. For more details, please check the |
| Optimization API at: |
| <a class="reference external" href="https://mxnet.apache.org/versions/master/api/python/docs/api/optimizer/index.html">https://mxnet.apache.org/versions/master/api/python/docs/api/optimizer/index.html</a></p> |
| </div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.apply" title="mxnet.gluon.nn.Embedding.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.collect_params" title="mxnet.gluon.nn.Embedding.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.export" title="mxnet.gluon.nn.Embedding.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.forward" title="mxnet.gluon.nn.Embedding.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.hybridize" title="mxnet.gluon.nn.Embedding.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.infer_shape" title="mxnet.gluon.nn.Embedding.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.infer_type" title="mxnet.gluon.nn.Embedding.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.initialize" title="mxnet.gluon.nn.Embedding.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.load" title="mxnet.gluon.nn.Embedding.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.load_dict" title="mxnet.gluon.nn.Embedding.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.load_parameters" title="mxnet.gluon.nn.Embedding.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.optimize_for" title="mxnet.gluon.nn.Embedding.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.register_forward_hook" title="mxnet.gluon.nn.Embedding.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.register_forward_pre_hook" title="mxnet.gluon.nn.Embedding.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.register_op_hook" title="mxnet.gluon.nn.Embedding.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.reset_ctx" title="mxnet.gluon.nn.Embedding.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.reset_device" title="mxnet.gluon.nn.Embedding.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.save" title="mxnet.gluon.nn.Embedding.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.save_parameters" title="mxnet.gluon.nn.Embedding.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.setattr" title="mxnet.gluon.nn.Embedding.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.share_parameters" title="mxnet.gluon.nn.Embedding.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.summary" title="mxnet.gluon.nn.Embedding.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.zero_grad" title="mxnet.gluon.nn.Embedding.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Embedding.params" title="mxnet.gluon.nn.Embedding.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>input_dim</strong> (<em>int</em>) – Size of the vocabulary, i.e. maximum integer index + 1.</p></li> |
| <li><p><strong>output_dim</strong> (<em>int</em>) – Dimension of the dense embedding.</p></li> |
| <li><p><strong>dtype</strong> (<em>str</em><em> or </em><em>np.dtype</em><em>, </em><em>default 'float32'</em>) – Data type of output embeddings.</p></li> |
| <li><p><strong>weight_initializer</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Initializer for the <cite>embeddings</cite> matrix.</p></li> |
| <li><p><strong>sparse_grad</strong> (<em>bool</em>) – If True, gradient w.r.t. weight will be a ‘row_sparse’ NDArray.</p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><p><strong>data</strong>: (N-1)-D tensor with shape: <cite>(x1, x2, …, xN-1)</cite>.</p></li> |
| </ul> |
| </p></li> |
| <li><p><strong>Output</strong> – <ul> |
| <li><p><strong>out</strong>: N-D tensor with shape: <cite>(x1, x2, …, xN-1, output_dim)</cite>.</p></li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Embedding.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Embedding.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.Embedding.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.Embedding.forward" title="mxnet.gluon.nn.Embedding.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.Embedding.forward" title="mxnet.gluon.nn.Embedding.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Embedding.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Embedding.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Flatten"> |
| <em class="property">class </em><code class="sig-name descname">Flatten</code><span class="sig-paren">(</span><em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Flatten"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Flatten" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Flattens the input to two dimensional.</p> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape <cite>(N, x1, x2, …, xn)</cite></p></li> |
| </ul> |
| </dd> |
| <dt>Output:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: 2D tensor with shape: <cite>(N, x1 cdot x2 cdot … cdot xn)</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.apply" title="mxnet.gluon.nn.Flatten.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.collect_params" title="mxnet.gluon.nn.Flatten.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.export" title="mxnet.gluon.nn.Flatten.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.forward" title="mxnet.gluon.nn.Flatten.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.hybridize" title="mxnet.gluon.nn.Flatten.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.infer_shape" title="mxnet.gluon.nn.Flatten.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.infer_type" title="mxnet.gluon.nn.Flatten.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.initialize" title="mxnet.gluon.nn.Flatten.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.load" title="mxnet.gluon.nn.Flatten.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.load_dict" title="mxnet.gluon.nn.Flatten.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.load_parameters" title="mxnet.gluon.nn.Flatten.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.optimize_for" title="mxnet.gluon.nn.Flatten.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.register_forward_hook" title="mxnet.gluon.nn.Flatten.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.register_forward_pre_hook" title="mxnet.gluon.nn.Flatten.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.register_op_hook" title="mxnet.gluon.nn.Flatten.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.reset_ctx" title="mxnet.gluon.nn.Flatten.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.reset_device" title="mxnet.gluon.nn.Flatten.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.save" title="mxnet.gluon.nn.Flatten.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.save_parameters" title="mxnet.gluon.nn.Flatten.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.setattr" title="mxnet.gluon.nn.Flatten.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.share_parameters" title="mxnet.gluon.nn.Flatten.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.summary" title="mxnet.gluon.nn.Flatten.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.zero_grad" title="mxnet.gluon.nn.Flatten.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Flatten.params" title="mxnet.gluon.nn.Flatten.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Flatten.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Flatten.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.Flatten.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.Flatten.forward" title="mxnet.gluon.nn.Flatten.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.Flatten.forward" title="mxnet.gluon.nn.Flatten.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Flatten.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Flatten.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GELU"> |
| <em class="property">class </em><code class="sig-name descname">GELU</code><span class="sig-paren">(</span><em class="sig-param">approximation='erf'</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#GELU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GELU" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <dl class="simple"> |
| <dt>Gaussian Exponential Linear Unit (GELU)</dt><dd><p>“Gaussian Error Linear Units (GELUs)”, Hendrycks et al, 2016 |
| <a class="reference external" href="https://arxiv.org/abs/1606.08415">https://arxiv.org/abs/1606.08415</a></p> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.apply" title="mxnet.gluon.nn.GELU.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.collect_params" title="mxnet.gluon.nn.GELU.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.export" title="mxnet.gluon.nn.GELU.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.forward" title="mxnet.gluon.nn.GELU.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.hybridize" title="mxnet.gluon.nn.GELU.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.infer_shape" title="mxnet.gluon.nn.GELU.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.infer_type" title="mxnet.gluon.nn.GELU.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.initialize" title="mxnet.gluon.nn.GELU.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.load" title="mxnet.gluon.nn.GELU.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.load_dict" title="mxnet.gluon.nn.GELU.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.load_parameters" title="mxnet.gluon.nn.GELU.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.optimize_for" title="mxnet.gluon.nn.GELU.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.register_forward_hook" title="mxnet.gluon.nn.GELU.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.register_forward_pre_hook" title="mxnet.gluon.nn.GELU.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.register_op_hook" title="mxnet.gluon.nn.GELU.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.reset_ctx" title="mxnet.gluon.nn.GELU.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.reset_device" title="mxnet.gluon.nn.GELU.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.save" title="mxnet.gluon.nn.GELU.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.save_parameters" title="mxnet.gluon.nn.GELU.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.setattr" title="mxnet.gluon.nn.GELU.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.share_parameters" title="mxnet.gluon.nn.GELU.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.summary" title="mxnet.gluon.nn.GELU.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.zero_grad" title="mxnet.gluon.nn.GELU.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GELU.params" title="mxnet.gluon.nn.GELU.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>approximation</strong> (<em>string</em>) – Which approximation of GELU calculation to use (erf or tanh).</p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </p></li> |
| <li><p><strong>Outputs</strong> – <ul> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#GELU.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GELU.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.GELU.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.GELU.forward" title="mxnet.gluon.nn.GELU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.GELU.forward" title="mxnet.gluon.nn.GELU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GELU.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GELU.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalAvgPool1D"> |
| <em class="property">class </em><code class="sig-name descname">GlobalAvgPool1D</code><span class="sig-paren">(</span><em class="sig-param">layout='NCW'</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalAvgPool1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalAvgPool1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Global average pooling operation for temporal data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and out (‘NCW’ or ‘NWC’). |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. padding is applied on ‘W’ dimension.</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, 1)</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalAvgPool2D"> |
| <em class="property">class </em><code class="sig-name descname">GlobalAvgPool2D</code><span class="sig-paren">(</span><em class="sig-param">layout='NCHW'</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalAvgPool2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalAvgPool2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Global average pooling operation for spatial data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and out (‘NCHW’ or ‘NHWC’). |
| ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, and width |
| dimensions respectively.</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, 1, 1)</cite> when <cite>layout</cite> is <cite>NCHW</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalAvgPool3D"> |
| <em class="property">class </em><code class="sig-name descname">GlobalAvgPool3D</code><span class="sig-paren">(</span><em class="sig-param">layout='NCDHW'</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalAvgPool3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalAvgPool3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Global average pooling operation for 3D data (spatial or spatio-temporal).</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and out (‘NCDHW’ or ‘NDHWC’). |
| ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, width and |
| depth dimensions respectively. padding is applied on ‘D’, ‘H’ and ‘W’ |
| dimension.</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, 1, 1, 1)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalMaxPool1D"> |
| <em class="property">class </em><code class="sig-name descname">GlobalMaxPool1D</code><span class="sig-paren">(</span><em class="sig-param">layout='NCW'</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalMaxPool1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalMaxPool1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Gloabl max pooling operation for one dimensional (temporal) data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and out (‘NCW’ or ‘NWC’). |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. Pooling is applied on the W dimension.</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, 1)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalMaxPool2D"> |
| <em class="property">class </em><code class="sig-name descname">GlobalMaxPool2D</code><span class="sig-paren">(</span><em class="sig-param">layout='NCHW'</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalMaxPool2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalMaxPool2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Global max pooling operation for two dimensional (spatial) data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and out (‘NCHW’ or ‘NHWC’). |
| ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, and width |
| dimensions respectively. padding is applied on ‘H’ and ‘W’ dimension.</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, 1, 1)</cite> when <cite>layout</cite> is <cite>NCHW</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalMaxPool3D"> |
| <em class="property">class </em><code class="sig-name descname">GlobalMaxPool3D</code><span class="sig-paren">(</span><em class="sig-param">layout='NCDHW'</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalMaxPool3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalMaxPool3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Global max pooling operation for 3D data (spatial or spatio-temporal).</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and out (‘NCDHW’ or ‘NDHWC’). |
| ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, width and |
| depth dimensions respectively. padding is applied on ‘D’, ‘H’ and ‘W’ |
| dimension.</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, 1, 1, 1)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GroupNorm"> |
| <em class="property">class </em><code class="sig-name descname">GroupNorm</code><span class="sig-paren">(</span><em class="sig-param">num_groups=1</em>, <em class="sig-param">epsilon=1e-05</em>, <em class="sig-param">center=True</em>, <em class="sig-param">scale=True</em>, <em class="sig-param">beta_initializer='zeros'</em>, <em class="sig-param">gamma_initializer='ones'</em>, <em class="sig-param">in_channels=0</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#GroupNorm"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Applies group normalization to the n-dimensional input array. |
| This operator takes an n-dimensional input array where the leftmost 2 axis are |
| <cite>batch</cite> and <cite>channel</cite> respectively:</p> |
| <div class="math notranslate nohighlight"> |
| \[x = x.reshape((N, num_groups, C // num_groups, ...)) |
| axis = (2, ...) |
| out = \frac{x - mean[x, axis]}{ \sqrt{Var[x, axis] + \epsilon}} * gamma + beta\]</div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.apply" title="mxnet.gluon.nn.GroupNorm.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.collect_params" title="mxnet.gluon.nn.GroupNorm.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.export" title="mxnet.gluon.nn.GroupNorm.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.forward" title="mxnet.gluon.nn.GroupNorm.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(data)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.hybridize" title="mxnet.gluon.nn.GroupNorm.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.infer_shape" title="mxnet.gluon.nn.GroupNorm.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(data, *args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.infer_type" title="mxnet.gluon.nn.GroupNorm.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.initialize" title="mxnet.gluon.nn.GroupNorm.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.load" title="mxnet.gluon.nn.GroupNorm.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.load_dict" title="mxnet.gluon.nn.GroupNorm.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.load_parameters" title="mxnet.gluon.nn.GroupNorm.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.optimize_for" title="mxnet.gluon.nn.GroupNorm.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.register_forward_hook" title="mxnet.gluon.nn.GroupNorm.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.register_forward_pre_hook" title="mxnet.gluon.nn.GroupNorm.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.register_op_hook" title="mxnet.gluon.nn.GroupNorm.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.reset_ctx" title="mxnet.gluon.nn.GroupNorm.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.reset_device" title="mxnet.gluon.nn.GroupNorm.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.save" title="mxnet.gluon.nn.GroupNorm.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.save_parameters" title="mxnet.gluon.nn.GroupNorm.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.setattr" title="mxnet.gluon.nn.GroupNorm.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.share_parameters" title="mxnet.gluon.nn.GroupNorm.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.summary" title="mxnet.gluon.nn.GroupNorm.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.zero_grad" title="mxnet.gluon.nn.GroupNorm.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.params" title="mxnet.gluon.nn.GroupNorm.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>num_groups</strong> (<em>int</em><em>, </em><em>default 1</em>) – Number of groups to separate the channel axis into.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>default 1e-5</em>) – Small float added to variance to avoid dividing by zero.</p></li> |
| <li><p><strong>center</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, add offset of <cite>beta</cite> to normalized tensor. |
| If False, <cite>beta</cite> is ignored.</p></li> |
| <li><p><strong>scale</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, multiply by <cite>gamma</cite>. If False, <cite>gamma</cite> is not used.</p></li> |
| <li><p><strong>beta_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the beta weight.</p></li> |
| <li><p><strong>gamma_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the gamma weight.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with shape (N, C, …).</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://arxiv.org/pdf/1803.08494.pdf">Group Normalization</a></p> |
| <p class="rubric">Examples</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="c1"># Input of shape (2, 3, 4)</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">np</span><span class="o">.</span><span class="n">array</span><span class="p">([[[</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">],</span> |
| <span class="go"> [ 4, 5, 6, 7],</span> |
| <span class="go"> [ 8, 9, 10, 11]],</span> |
| <span class="go"> [[12, 13, 14, 15],</span> |
| <span class="go"> [16, 17, 18, 19],</span> |
| <span class="go"> [20, 21, 22, 23]]])</span> |
| <span class="gp">>>> </span><span class="c1"># Group normalization is calculated with the above formula</span> |
| <span class="gp">>>> </span><span class="n">layer</span> <span class="o">=</span> <span class="n">GroupNorm</span><span class="p">()</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="o">.</span><span class="n">initialize</span><span class="p">(</span><span class="n">device</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> |
| <span class="go">[[[-1.5932543 -1.3035717 -1.0138891 -0.7242065]</span> |
| <span class="go"> [-0.4345239 -0.1448413 0.1448413 0.4345239]</span> |
| <span class="go"> [ 0.7242065 1.0138891 1.3035717 1.5932543]]</span> |
| <span class="go"> [[-1.5932543 -1.3035717 -1.0138891 -0.7242065]</span> |
| <span class="go"> [-0.4345239 -0.1448413 0.1448413 0.4345239]</span> |
| <span class="go"> [ 0.7242065 1.0138891 1.3035717 1.5932543]]]</span> |
| </pre></div> |
| </div> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">data</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#GroupNorm.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">data</em>, <em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#GroupNorm.infer_shape"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.forward" title="mxnet.gluon.nn.GroupNorm.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.GroupNorm.forward" title="mxnet.gluon.nn.GroupNorm.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.GroupNorm.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.GroupNorm.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.HybridBlock"> |
| <em class="property">class </em><code class="sig-name descname">HybridBlock</code><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.Block</span></code></p> |
| <p><cite>HybridBlock</cite> supports forwarding with both Symbol and NDArray.</p> |
| <p><cite>HybridBlock</cite> is similar to <cite>Block</cite>, with a few differences:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">mxnet</span> <span class="k">as</span> <span class="nn">mx</span> |
| <span class="kn">from</span> <span class="nn">mxnet.gluon</span> <span class="kn">import</span> <span class="n">HybridBlock</span><span class="p">,</span> <span class="n">nn</span> |
| |
| <span class="k">class</span> <span class="nc">Model</span><span class="p">(</span><span class="n">HybridBlock</span><span class="p">):</span> |
| <span class="k">def</span> <span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">):</span> |
| <span class="nb">super</span><span class="p">(</span><span class="n">Model</span><span class="p">,</span> <span class="bp">self</span><span class="p">)</span><span class="o">.</span><span class="fm">__init__</span><span class="p">(</span><span class="o">**</span><span class="n">kwargs</span><span class="p">)</span> |
| <span class="bp">self</span><span class="o">.</span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="bp">self</span><span class="o">.</span><span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| |
| <span class="k">def</span> <span class="nf">forward</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">x</span><span class="p">):</span> |
| <span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">npx</span><span class="o">.</span><span class="n">relu</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">dense0</span><span class="p">(</span><span class="n">x</span><span class="p">))</span> |
| <span class="k">return</span> <span class="n">mx</span><span class="o">.</span><span class="n">npx</span><span class="o">.</span><span class="n">relu</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">dense1</span><span class="p">(</span><span class="n">x</span><span class="p">))</span> |
| |
| <span class="n">model</span> <span class="o">=</span> <span class="n">Model</span><span class="p">()</span> |
| <span class="n">model</span><span class="o">.</span><span class="n">initialize</span><span class="p">(</span><span class="n">device</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">))</span> |
| <span class="n">model</span><span class="o">.</span><span class="n">hybridize</span><span class="p">()</span> |
| <span class="n">model</span><span class="p">(</span><span class="n">mx</span><span class="o">.</span><span class="n">np</span><span class="o">.</span><span class="n">zeros</span><span class="p">((</span><span class="mi">10</span><span class="p">,</span> <span class="mi">10</span><span class="p">),</span> <span class="n">device</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">)))</span> |
| </pre></div> |
| </div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.cast" title="mxnet.gluon.nn.HybridBlock.cast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cast</span></code></a>(dtype)</p></td> |
| <td><p>Cast this Block to use another data type.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.forward" title="mxnet.gluon.nn.HybridBlock.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x, *args)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.hybridize" title="mxnet.gluon.nn.HybridBlock.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.infer_shape" title="mxnet.gluon.nn.HybridBlock.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.infer_type" title="mxnet.gluon.nn.HybridBlock.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.optimize_for" title="mxnet.gluon.nn.HybridBlock.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.register_child" title="mxnet.gluon.nn.HybridBlock.register_child"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_child</span></code></a>(block[, name])</p></td> |
| <td><p>Registers block as a child of self.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.register_op_hook" title="mxnet.gluon.nn.HybridBlock.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.reset_ctx" title="mxnet.gluon.nn.HybridBlock.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.reset_device" title="mxnet.gluon.nn.HybridBlock.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p>Forward computation in <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> must be static to work with <code class="xref py py-class docutils literal notranslate"><span class="pre">Symbol</span></code> s, |
| i.e. you cannot call <code class="xref py py-meth docutils literal notranslate"><span class="pre">NDArray.asnumpy()</span></code>, <code class="xref py py-attr docutils literal notranslate"><span class="pre">NDArray.shape</span></code>, |
| <code class="xref py py-attr docutils literal notranslate"><span class="pre">NDArray.dtype</span></code>, <cite>NDArray</cite> indexing (<cite>x[i]</cite>) etc on tensors. |
| Also, you cannot use branching or loop logic that bases on non-constant |
| expressions like random numbers or intermediate results, since they change |
| the graph structure for each iteration.</p> |
| <p>Before activating with <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.hybridize" title="mxnet.gluon.nn.HybridBlock.hybridize"><code class="xref py py-meth docutils literal notranslate"><span class="pre">hybridize()</span></code></a>, <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> works just like normal |
| <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>. After activation, <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> will create a symbolic graph |
| representing the forward computation and cache it. On subsequent forwards, |
| the cached graph will be used instead of <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.forward" title="mxnet.gluon.nn.HybridBlock.forward"><code class="xref py py-meth docutils literal notranslate"><span class="pre">forward()</span></code></a>.</p> |
| <p>Please see references for detailed tutorial.</p> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/versions/master/api/python/docs/tutorials/packages/gluon/blocks/hybridize.html">Hybridize - A Hybrid of Imperative and Symbolic Programming</a></p> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.cast"> |
| <code class="sig-name descname">cast</code><span class="sig-paren">(</span><em class="sig-param">dtype</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.cast"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.cast" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Cast this Block to use another data type.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>dtype</strong> (<em>str</em><em> or </em><em>numpy.dtype</em>) – The new data type.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.export"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.hybridize"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.infer_shape"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.infer_type"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.optimize_for"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.register_child"> |
| <code class="sig-name descname">register_child</code><span class="sig-paren">(</span><em class="sig-param">block</em>, <em class="sig-param">name=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.register_child"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.register_child" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers block as a child of self. <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> s assigned to self as |
| attributes will be registered automatically.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.register_op_hook"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.reset_ctx"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridBlock.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#HybridBlock.reset_device"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridBlock.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate"> |
| <em class="property">class </em><code class="sig-name descname">HybridConcatenate</code><span class="sig-paren">(</span><em class="sig-param">axis=-1</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#HybridConcatenate"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.basic_layers.HybridSequential</span></code></p> |
| <p>Lays <cite>HybridBlock</cite> s concurrently.</p> |
| <p>This block feeds its input to all children blocks, and |
| produce the output by concatenating all the children blocks’ outputs |
| on the specified axis.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">net</span> <span class="o">=</span> <span class="n">HybridConcatenate</span><span class="p">()</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">activation</span><span class="o">=</span><span class="s1">'relu'</span><span class="p">))</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">))</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">Identity</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.add" title="mxnet.gluon.nn.HybridConcatenate.add"><code class="xref py py-obj docutils literal notranslate"><span class="pre">add</span></code></a>(*blocks)</p></td> |
| <td><p>Adds block on top of the stack.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.apply" title="mxnet.gluon.nn.HybridConcatenate.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.collect_params" title="mxnet.gluon.nn.HybridConcatenate.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.export" title="mxnet.gluon.nn.HybridConcatenate.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.forward" title="mxnet.gluon.nn.HybridConcatenate.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.hybridize" title="mxnet.gluon.nn.HybridConcatenate.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.infer_shape" title="mxnet.gluon.nn.HybridConcatenate.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.infer_type" title="mxnet.gluon.nn.HybridConcatenate.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.initialize" title="mxnet.gluon.nn.HybridConcatenate.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.load" title="mxnet.gluon.nn.HybridConcatenate.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.load_dict" title="mxnet.gluon.nn.HybridConcatenate.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.load_parameters" title="mxnet.gluon.nn.HybridConcatenate.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.optimize_for" title="mxnet.gluon.nn.HybridConcatenate.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.register_forward_hook" title="mxnet.gluon.nn.HybridConcatenate.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.register_forward_pre_hook" title="mxnet.gluon.nn.HybridConcatenate.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.register_op_hook" title="mxnet.gluon.nn.HybridConcatenate.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.reset_ctx" title="mxnet.gluon.nn.HybridConcatenate.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.reset_device" title="mxnet.gluon.nn.HybridConcatenate.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.save" title="mxnet.gluon.nn.HybridConcatenate.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.save_parameters" title="mxnet.gluon.nn.HybridConcatenate.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.setattr" title="mxnet.gluon.nn.HybridConcatenate.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.share_parameters" title="mxnet.gluon.nn.HybridConcatenate.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.summary" title="mxnet.gluon.nn.HybridConcatenate.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.zero_grad" title="mxnet.gluon.nn.HybridConcatenate.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.params" title="mxnet.gluon.nn.HybridConcatenate.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>axis</strong> (<em>int</em><em>, </em><em>default -1</em>) – The axis on which to concatenate the outputs.</p> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.add"> |
| <code class="sig-name descname">add</code><span class="sig-paren">(</span><em class="sig-param">*blocks</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.add" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Adds block on top of the stack.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#HybridConcatenate.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.forward" title="mxnet.gluon.nn.HybridConcatenate.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.HybridConcatenate.forward" title="mxnet.gluon.nn.HybridConcatenate.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridConcatenate.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridConcatenate.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.HybridLambda"> |
| <em class="property">class </em><code class="sig-name descname">HybridLambda</code><span class="sig-paren">(</span><em class="sig-param">function</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#HybridLambda"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Wraps an operator or an expression as a HybridBlock object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>function</strong> (<em>str</em><em> or </em><em>function</em>) – <p>Function used in lambda must be one of the following: |
| 1) The name of an operator that is available in both symbol and ndarray. For example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">block</span> <span class="o">=</span> <span class="n">HybridLambda</span><span class="p">(</span><span class="s1">'tanh'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <ol class="arabic" start="2"> |
| <li><p>A function that conforms to <code class="docutils literal notranslate"><span class="pre">def</span> <span class="pre">function(F,</span> <span class="pre">data,</span> <span class="pre">*args)</span></code>. For example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">block</span> <span class="o">=</span> <span class="n">HybridLambda</span><span class="p">(</span><span class="k">lambda</span> <span class="n">F</span><span class="p">,</span> <span class="n">x</span><span class="p">:</span> <span class="n">F</span><span class="o">.</span><span class="n">LeakyReLU</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">slope</span><span class="o">=</span><span class="mf">0.1</span><span class="p">))</span> |
| </pre></div> |
| </div> |
| </li> |
| </ol> |
| </p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><dl class="simple"> |
| <dt>** <em>args *</em>: one or more input data. First argument must be symbol or ndarray. Their </dt><dd><p>shapes depend on the function.</p> |
| </dd> |
| </dl> |
| </li> |
| </ul> |
| </p></li> |
| <li><p><strong>Output</strong> – <ul> |
| <li><p>** <em>outputs *</em>: one or more output data. Their shapes depend on the function.</p></li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.apply" title="mxnet.gluon.nn.HybridLambda.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.collect_params" title="mxnet.gluon.nn.HybridLambda.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.export" title="mxnet.gluon.nn.HybridLambda.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.forward" title="mxnet.gluon.nn.HybridLambda.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x, *args)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.hybridize" title="mxnet.gluon.nn.HybridLambda.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.infer_shape" title="mxnet.gluon.nn.HybridLambda.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.infer_type" title="mxnet.gluon.nn.HybridLambda.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.initialize" title="mxnet.gluon.nn.HybridLambda.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.load" title="mxnet.gluon.nn.HybridLambda.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.load_dict" title="mxnet.gluon.nn.HybridLambda.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.load_parameters" title="mxnet.gluon.nn.HybridLambda.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.optimize_for" title="mxnet.gluon.nn.HybridLambda.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.register_forward_hook" title="mxnet.gluon.nn.HybridLambda.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.register_forward_pre_hook" title="mxnet.gluon.nn.HybridLambda.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.register_op_hook" title="mxnet.gluon.nn.HybridLambda.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.reset_ctx" title="mxnet.gluon.nn.HybridLambda.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.reset_device" title="mxnet.gluon.nn.HybridLambda.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.save" title="mxnet.gluon.nn.HybridLambda.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.save_parameters" title="mxnet.gluon.nn.HybridLambda.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.setattr" title="mxnet.gluon.nn.HybridLambda.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.share_parameters" title="mxnet.gluon.nn.HybridLambda.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.summary" title="mxnet.gluon.nn.HybridLambda.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.zero_grad" title="mxnet.gluon.nn.HybridLambda.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.params" title="mxnet.gluon.nn.HybridLambda.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#HybridLambda.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.forward" title="mxnet.gluon.nn.HybridLambda.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.HybridLambda.forward" title="mxnet.gluon.nn.HybridLambda.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridLambda.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.HybridSequential"> |
| <em class="property">class </em><code class="sig-name descname">HybridSequential</code><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#HybridSequential"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Stacks HybridBlocks sequentially.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">net</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">HybridSequential</span><span class="p">()</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">activation</span><span class="o">=</span><span class="s1">'relu'</span><span class="p">))</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">))</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">hybridize</span><span class="p">()</span> |
| </pre></div> |
| </div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.add" title="mxnet.gluon.nn.HybridSequential.add"><code class="xref py py-obj docutils literal notranslate"><span class="pre">add</span></code></a>(*blocks)</p></td> |
| <td><p>Adds block on top of the stack.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.apply" title="mxnet.gluon.nn.HybridSequential.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.collect_params" title="mxnet.gluon.nn.HybridSequential.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.export" title="mxnet.gluon.nn.HybridSequential.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.forward" title="mxnet.gluon.nn.HybridSequential.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x, *args)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.hybridize" title="mxnet.gluon.nn.HybridSequential.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.infer_shape" title="mxnet.gluon.nn.HybridSequential.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.infer_type" title="mxnet.gluon.nn.HybridSequential.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.initialize" title="mxnet.gluon.nn.HybridSequential.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.load" title="mxnet.gluon.nn.HybridSequential.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.load_dict" title="mxnet.gluon.nn.HybridSequential.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.load_parameters" title="mxnet.gluon.nn.HybridSequential.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.optimize_for" title="mxnet.gluon.nn.HybridSequential.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.register_forward_hook" title="mxnet.gluon.nn.HybridSequential.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.register_forward_pre_hook" title="mxnet.gluon.nn.HybridSequential.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.register_op_hook" title="mxnet.gluon.nn.HybridSequential.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.reset_ctx" title="mxnet.gluon.nn.HybridSequential.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.reset_device" title="mxnet.gluon.nn.HybridSequential.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.save" title="mxnet.gluon.nn.HybridSequential.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.save_parameters" title="mxnet.gluon.nn.HybridSequential.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.setattr" title="mxnet.gluon.nn.HybridSequential.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.share_parameters" title="mxnet.gluon.nn.HybridSequential.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.summary" title="mxnet.gluon.nn.HybridSequential.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.zero_grad" title="mxnet.gluon.nn.HybridSequential.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.params" title="mxnet.gluon.nn.HybridSequential.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.add"> |
| <code class="sig-name descname">add</code><span class="sig-paren">(</span><em class="sig-param">*blocks</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#HybridSequential.add"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.add" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Adds block on top of the stack.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#HybridSequential.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.forward" title="mxnet.gluon.nn.HybridSequential.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.HybridSequential.forward" title="mxnet.gluon.nn.HybridSequential.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.HybridSequential.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.HybridSequential.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Identity"> |
| <em class="property">class </em><code class="sig-name descname">Identity</code><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Identity"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Identity" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Block that passes through the input directly.</p> |
| <p>This block can be used in conjunction with HybridConcatenate |
| block for residual connection.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">net</span> <span class="o">=</span> <span class="n">HybridConcatenate</span><span class="p">()</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">activation</span><span class="o">=</span><span class="s1">'relu'</span><span class="p">))</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">))</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">Identity</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.apply" title="mxnet.gluon.nn.Identity.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.collect_params" title="mxnet.gluon.nn.Identity.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.export" title="mxnet.gluon.nn.Identity.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.forward" title="mxnet.gluon.nn.Identity.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.hybridize" title="mxnet.gluon.nn.Identity.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.infer_shape" title="mxnet.gluon.nn.Identity.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.infer_type" title="mxnet.gluon.nn.Identity.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.initialize" title="mxnet.gluon.nn.Identity.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.load" title="mxnet.gluon.nn.Identity.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.load_dict" title="mxnet.gluon.nn.Identity.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.load_parameters" title="mxnet.gluon.nn.Identity.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.optimize_for" title="mxnet.gluon.nn.Identity.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.register_forward_hook" title="mxnet.gluon.nn.Identity.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.register_forward_pre_hook" title="mxnet.gluon.nn.Identity.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.register_op_hook" title="mxnet.gluon.nn.Identity.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.reset_ctx" title="mxnet.gluon.nn.Identity.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.reset_device" title="mxnet.gluon.nn.Identity.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.save" title="mxnet.gluon.nn.Identity.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.save_parameters" title="mxnet.gluon.nn.Identity.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.setattr" title="mxnet.gluon.nn.Identity.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.share_parameters" title="mxnet.gluon.nn.Identity.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.summary" title="mxnet.gluon.nn.Identity.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.zero_grad" title="mxnet.gluon.nn.Identity.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Identity.params" title="mxnet.gluon.nn.Identity.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Identity.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Identity.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.Identity.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.Identity.forward" title="mxnet.gluon.nn.Identity.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.Identity.forward" title="mxnet.gluon.nn.Identity.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Identity.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Identity.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.InstanceNorm"> |
| <em class="property">class </em><code class="sig-name descname">InstanceNorm</code><span class="sig-paren">(</span><em class="sig-param">axis=1</em>, <em class="sig-param">epsilon=1e-05</em>, <em class="sig-param">center=True</em>, <em class="sig-param">scale=False</em>, <em class="sig-param">beta_initializer='zeros'</em>, <em class="sig-param">gamma_initializer='ones'</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#InstanceNorm"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Applies instance normalization to the n-dimensional input array. |
| This operator takes an n-dimensional input array where (n>2) and normalizes |
| the input using the following formula:</p> |
| <div class="math notranslate nohighlight"> |
| \[ \begin{align}\begin{aligned}\bar{C} = \{i \mid i \neq 0, i \neq axis\}\\out = \frac{x - mean[data, \bar{C}]}{ \sqrt{Var[data, \bar{C}]} + \epsilon} |
| * gamma + beta\end{aligned}\end{align} \]</div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.apply" title="mxnet.gluon.nn.InstanceNorm.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.collect_params" title="mxnet.gluon.nn.InstanceNorm.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.export" title="mxnet.gluon.nn.InstanceNorm.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.forward" title="mxnet.gluon.nn.InstanceNorm.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.hybridize" title="mxnet.gluon.nn.InstanceNorm.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.infer_shape" title="mxnet.gluon.nn.InstanceNorm.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(x, *args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.infer_type" title="mxnet.gluon.nn.InstanceNorm.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.initialize" title="mxnet.gluon.nn.InstanceNorm.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.load" title="mxnet.gluon.nn.InstanceNorm.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.load_dict" title="mxnet.gluon.nn.InstanceNorm.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.load_parameters" title="mxnet.gluon.nn.InstanceNorm.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.optimize_for" title="mxnet.gluon.nn.InstanceNorm.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.register_forward_hook" title="mxnet.gluon.nn.InstanceNorm.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.register_forward_pre_hook" title="mxnet.gluon.nn.InstanceNorm.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.register_op_hook" title="mxnet.gluon.nn.InstanceNorm.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.reset_ctx" title="mxnet.gluon.nn.InstanceNorm.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.reset_device" title="mxnet.gluon.nn.InstanceNorm.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.save" title="mxnet.gluon.nn.InstanceNorm.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.save_parameters" title="mxnet.gluon.nn.InstanceNorm.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.setattr" title="mxnet.gluon.nn.InstanceNorm.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.share_parameters" title="mxnet.gluon.nn.InstanceNorm.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.summary" title="mxnet.gluon.nn.InstanceNorm.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.zero_grad" title="mxnet.gluon.nn.InstanceNorm.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.params" title="mxnet.gluon.nn.InstanceNorm.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>default 1</em>) – The axis that will be excluded in the normalization process. This is typically the channels |
| (C) axis. For instance, after a <cite>Conv2D</cite> layer with <cite>layout=’NCHW’</cite>, |
| set <cite>axis=1</cite> in <cite>InstanceNorm</cite>. If <cite>layout=’NHWC’</cite>, then set <cite>axis=3</cite>. Data will be |
| normalized along axes excluding the first axis and the axis given.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>default 1e-5</em>) – Small float added to variance to avoid dividing by zero.</p></li> |
| <li><p><strong>center</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, add offset of <cite>beta</cite> to normalized tensor. |
| If False, <cite>beta</cite> is ignored.</p></li> |
| <li><p><strong>scale</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, multiply by <cite>gamma</cite>. If False, <cite>gamma</cite> is not used. |
| When the next layer is linear (also e.g. <cite>nn.relu</cite>), |
| this can be disabled since the scaling |
| will be done by the next layer.</p></li> |
| <li><p><strong>beta_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the beta weight.</p></li> |
| <li><p><strong>gamma_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the gamma weight.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – Number of channels (feature maps) in input data. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://arxiv.org/abs/1607.08022">Instance Normalization: The Missing Ingredient for Fast Stylization</a></p> |
| <p class="rubric">Examples</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="c1"># Input of shape (2,1,2)</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">np</span><span class="o">.</span><span class="n">array</span><span class="p">([[[</span> <span class="mf">1.1</span><span class="p">,</span> <span class="mf">2.2</span><span class="p">]],</span> |
| <span class="gp">... </span> <span class="p">[[</span> <span class="mf">3.3</span><span class="p">,</span> <span class="mf">4.4</span><span class="p">]]])</span> |
| <span class="gp">>>> </span><span class="c1"># Instance normalization is calculated with the above formula</span> |
| <span class="gp">>>> </span><span class="n">layer</span> <span class="o">=</span> <span class="n">InstanceNorm</span><span class="p">()</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="o">.</span><span class="n">initialize</span><span class="p">(</span><span class="n">device</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> |
| <span class="go">[[[-0.99998355 0.99998331]]</span> |
| <span class="go"> [[-0.99998319 0.99998361]]]</span> |
| </pre></div> |
| </div> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#InstanceNorm.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#InstanceNorm.infer_shape"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.forward" title="mxnet.gluon.nn.InstanceNorm.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm.forward" title="mxnet.gluon.nn.InstanceNorm.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.InstanceNorm.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Lambda"> |
| <em class="property">class </em><code class="sig-name descname">Lambda</code><span class="sig-paren">(</span><em class="sig-param">function</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Lambda"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Lambda" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.Block</span></code></p> |
| <p>Wraps an operator or an expression as a Block object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>function</strong> (<em>str</em><em> or </em><em>function</em>) – <p>Function used in lambda must be one of the following: |
| 1) the name of an operator that is available in ndarray. For example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">block</span> <span class="o">=</span> <span class="n">Lambda</span><span class="p">(</span><span class="s1">'tanh'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <ol class="arabic" start="2"> |
| <li><p>a function that conforms to <code class="docutils literal notranslate"><span class="pre">def</span> <span class="pre">function(*args)</span></code>. For example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">block</span> <span class="o">=</span> <span class="n">Lambda</span><span class="p">(</span><span class="k">lambda</span> <span class="n">x</span><span class="p">:</span> <span class="n">npx</span><span class="o">.</span><span class="n">leaky_relu</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">slope</span><span class="o">=</span><span class="mf">0.1</span><span class="p">))</span> |
| </pre></div> |
| </div> |
| </li> |
| </ol> |
| </p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><p>** <em>args *</em>: one or more input data. Their shapes depend on the function.</p></li> |
| </ul> |
| </p></li> |
| <li><p><strong>Output</strong> – <ul> |
| <li><p>** <em>outputs *</em>: one or more output data. Their shapes depend on the function.</p></li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Lambda.forward" title="mxnet.gluon.nn.Lambda.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(*args)</p></td> |
| <td><p>Overrides to implement forward computation using <code class="xref py py-class docutils literal notranslate"><span class="pre">NDArray</span></code>.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Lambda.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Lambda.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Lambda.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides to implement forward computation using <code class="xref py py-class docutils literal notranslate"><span class="pre">NDArray</span></code>. Only |
| accepts positional arguments.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>*args</strong> (<em>list of NDArray</em>) – Input tensors.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.LayerNorm"> |
| <em class="property">class </em><code class="sig-name descname">LayerNorm</code><span class="sig-paren">(</span><em class="sig-param">axis=-1</em>, <em class="sig-param">epsilon=1e-05</em>, <em class="sig-param">center=True</em>, <em class="sig-param">scale=True</em>, <em class="sig-param">beta_initializer='zeros'</em>, <em class="sig-param">gamma_initializer='ones'</em>, <em class="sig-param">in_channels=0</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#LayerNorm"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Applies layer normalization to the n-dimensional input array. |
| This operator takes an n-dimensional input array and normalizes |
| the input using the given axis:</p> |
| <div class="math notranslate nohighlight"> |
| \[out = \frac{x - mean[data, axis]}{ \sqrt{Var[data, axis] + \epsilon}} * gamma + beta\]</div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.apply" title="mxnet.gluon.nn.LayerNorm.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.collect_params" title="mxnet.gluon.nn.LayerNorm.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.export" title="mxnet.gluon.nn.LayerNorm.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.forward" title="mxnet.gluon.nn.LayerNorm.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(data)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.hybridize" title="mxnet.gluon.nn.LayerNorm.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.infer_shape" title="mxnet.gluon.nn.LayerNorm.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(data, *args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.infer_type" title="mxnet.gluon.nn.LayerNorm.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.initialize" title="mxnet.gluon.nn.LayerNorm.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.load" title="mxnet.gluon.nn.LayerNorm.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.load_dict" title="mxnet.gluon.nn.LayerNorm.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.load_parameters" title="mxnet.gluon.nn.LayerNorm.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.optimize_for" title="mxnet.gluon.nn.LayerNorm.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.register_forward_hook" title="mxnet.gluon.nn.LayerNorm.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.register_forward_pre_hook" title="mxnet.gluon.nn.LayerNorm.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.register_op_hook" title="mxnet.gluon.nn.LayerNorm.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.reset_ctx" title="mxnet.gluon.nn.LayerNorm.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.reset_device" title="mxnet.gluon.nn.LayerNorm.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.save" title="mxnet.gluon.nn.LayerNorm.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.save_parameters" title="mxnet.gluon.nn.LayerNorm.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.setattr" title="mxnet.gluon.nn.LayerNorm.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.share_parameters" title="mxnet.gluon.nn.LayerNorm.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.summary" title="mxnet.gluon.nn.LayerNorm.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.zero_grad" title="mxnet.gluon.nn.LayerNorm.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.params" title="mxnet.gluon.nn.LayerNorm.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>default -1</em>) – The axis that should be normalized. This is typically the axis of the channels.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>default 1e-5</em>) – Small float added to variance to avoid dividing by zero.</p></li> |
| <li><p><strong>center</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, add offset of <cite>beta</cite> to normalized tensor. |
| If False, <cite>beta</cite> is ignored.</p></li> |
| <li><p><strong>scale</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, multiply by <cite>gamma</cite>. If False, <cite>gamma</cite> is not used.</p></li> |
| <li><p><strong>beta_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the beta weight.</p></li> |
| <li><p><strong>gamma_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the gamma weight.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – Number of channels (feature maps) in input data. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://arxiv.org/pdf/1607.06450.pdf">Layer Normalization</a></p> |
| <p class="rubric">Examples</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="c1"># Input of shape (2, 5)</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">np</span><span class="o">.</span><span class="n">array</span><span class="p">([[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">5</span><span class="p">],</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">2</span><span class="p">]])</span> |
| <span class="gp">>>> </span><span class="c1"># Layer normalization is calculated with the above formula</span> |
| <span class="gp">>>> </span><span class="n">layer</span> <span class="o">=</span> <span class="n">LayerNorm</span><span class="p">()</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="o">.</span><span class="n">initialize</span><span class="p">(</span><span class="n">device</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> |
| <span class="go">[[-1.41421 -0.707105 0. 0.707105 1.41421 ]</span> |
| <span class="go"> [-1.2247195 -1.2247195 0.81647956 0.81647956 0.81647956]]</span> |
| </pre></div> |
| </div> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">data</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#LayerNorm.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">data</em>, <em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#LayerNorm.infer_shape"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.forward" title="mxnet.gluon.nn.LayerNorm.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.LayerNorm.forward" title="mxnet.gluon.nn.LayerNorm.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LayerNorm.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.LeakyReLU"> |
| <em class="property">class </em><code class="sig-name descname">LeakyReLU</code><span class="sig-paren">(</span><em class="sig-param">alpha</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#LeakyReLU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Leaky version of a Rectified Linear Unit.</p> |
| <p>It allows a small gradient when the unit is not active</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}f\left(x\right) = \left\{ |
| \begin{array}{lr} |
| \alpha x & : x \lt 0 \\ |
| x & : x \geq 0 \\ |
| \end{array} |
| \right.\\\end{split}\]</div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.apply" title="mxnet.gluon.nn.LeakyReLU.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.collect_params" title="mxnet.gluon.nn.LeakyReLU.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.export" title="mxnet.gluon.nn.LeakyReLU.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.forward" title="mxnet.gluon.nn.LeakyReLU.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.hybridize" title="mxnet.gluon.nn.LeakyReLU.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.infer_shape" title="mxnet.gluon.nn.LeakyReLU.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.infer_type" title="mxnet.gluon.nn.LeakyReLU.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.initialize" title="mxnet.gluon.nn.LeakyReLU.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.load" title="mxnet.gluon.nn.LeakyReLU.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.load_dict" title="mxnet.gluon.nn.LeakyReLU.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.load_parameters" title="mxnet.gluon.nn.LeakyReLU.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.optimize_for" title="mxnet.gluon.nn.LeakyReLU.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.register_forward_hook" title="mxnet.gluon.nn.LeakyReLU.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.register_forward_pre_hook" title="mxnet.gluon.nn.LeakyReLU.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.register_op_hook" title="mxnet.gluon.nn.LeakyReLU.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.reset_ctx" title="mxnet.gluon.nn.LeakyReLU.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.reset_device" title="mxnet.gluon.nn.LeakyReLU.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.save" title="mxnet.gluon.nn.LeakyReLU.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.save_parameters" title="mxnet.gluon.nn.LeakyReLU.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.setattr" title="mxnet.gluon.nn.LeakyReLU.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.share_parameters" title="mxnet.gluon.nn.LeakyReLU.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.summary" title="mxnet.gluon.nn.LeakyReLU.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.zero_grad" title="mxnet.gluon.nn.LeakyReLU.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.params" title="mxnet.gluon.nn.LeakyReLU.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>alpha</strong> (<em>float</em>) – slope coefficient for the negative half axis. Must be >= 0.</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#LeakyReLU.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.forward" title="mxnet.gluon.nn.LeakyReLU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU.forward" title="mxnet.gluon.nn.LeakyReLU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.LeakyReLU.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.MaxPool1D"> |
| <em class="property">class </em><code class="sig-name descname">MaxPool1D</code><span class="sig-paren">(</span><em class="sig-param">pool_size=2</em>, <em class="sig-param">strides=None</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">layout='NCW'</em>, <em class="sig-param">ceil_mode=False</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#MaxPool1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.MaxPool1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Max pooling operation for one dimensional data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>pool_size</strong> (<em>int</em>) – Size of the max pooling windows.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em>, or </em><em>None</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and out (‘NCW’ or ‘NWC’). |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. Pooling is applied on the W dimension.</p></li> |
| <li><p><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When <cite>True</cite>, will use ceil instead of floor to compute the output shape.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, out_width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. out_width is calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="o">-</span><span class="n">pool_size</span><span class="p">)</span><span class="o">/</span><span class="n">strides</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.MaxPool2D"> |
| <em class="property">class </em><code class="sig-name descname">MaxPool2D</code><span class="sig-paren">(</span><em class="sig-param">pool_size=(2</em>, <em class="sig-param">2)</em>, <em class="sig-param">strides=None</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">layout='NCHW'</em>, <em class="sig-param">ceil_mode=False</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#MaxPool2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.MaxPool2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Max pooling operation for two dimensional (spatial) data.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>pool_size</strong> (<em>int</em><em> or </em><em>list/tuple of 2 ints</em><em>,</em>) – Size of the max pooling windows.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em>, </em><em>list/tuple of 2 ints</em><em>, or </em><em>None.</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>list/tuple of 2 ints</em><em>,</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and out (‘NCHW’ or ‘NHWC’). |
| ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, and width |
| dimensions respectively. padding is applied on ‘H’ and ‘W’ dimension.</p></li> |
| <li><p><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When <cite>True</cite>, will use ceil instead of floor to compute the output shape.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.MaxPool3D"> |
| <em class="property">class </em><code class="sig-name descname">MaxPool3D</code><span class="sig-paren">(</span><em class="sig-param">pool_size=(2</em>, <em class="sig-param">2</em>, <em class="sig-param">2)</em>, <em class="sig-param">strides=None</em>, <em class="sig-param">padding=0</em>, <em class="sig-param">ceil_mode=False</em>, <em class="sig-param">layout='NCDHW'</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#MaxPool3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.MaxPool3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.conv_layers._Pooling</span></code></p> |
| <p>Max pooling operation for 3D data (spatial or spatio-temporal).</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>pool_size</strong> (<em>int</em><em> or </em><em>list/tuple of 3 ints</em><em>,</em>) – Size of the max pooling windows.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em>, </em><em>list/tuple of 3 ints</em><em>, or </em><em>None.</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>list/tuple of 3 ints</em><em>,</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and out (‘NCDHW’ or ‘NDHWC’). |
| ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, width and |
| depth dimensions respectively. padding is applied on ‘D’, ‘H’ and ‘W’ |
| dimension.</p></li> |
| <li><p><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When <cite>True</cite>, will use ceil instead of floor to compute the output shape.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, out_depth, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCDHW</cite>. |
| out_depth, out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_depth</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">depth</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution"> |
| <em class="property">class </em><code class="sig-name descname">ModulatedDeformableConvolution</code><span class="sig-paren">(</span><em class="sig-param">channels</em>, <em class="sig-param">kernel_size=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">strides=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">padding=(0</em>, <em class="sig-param">0)</em>, <em class="sig-param">dilation=(1</em>, <em class="sig-param">1)</em>, <em class="sig-param">groups=1</em>, <em class="sig-param">num_deformable_group=1</em>, <em class="sig-param">layout='NCHW'</em>, <em class="sig-param">use_bias=True</em>, <em class="sig-param">in_channels=0</em>, <em class="sig-param">activation=None</em>, <em class="sig-param">weight_initializer=None</em>, <em class="sig-param">bias_initializer='zeros'</em>, <em class="sig-param">offset_weight_initializer='zeros'</em>, <em class="sig-param">offset_bias_initializer='zeros'</em>, <em class="sig-param">offset_use_bias=True</em>, <em class="sig-param">op_name='ModulatedDeformableConvolution'</em>, <em class="sig-param">adj=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#ModulatedDeformableConvolution"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>2-D Deformable Convolution v2 (Dai, 2018).</p> |
| <p>The modulated deformable convolution operation is described in <a class="reference external" href="https://arxiv.org/abs/1811.11168">https://arxiv.org/abs/1811.11168</a></p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>channels</strong> (<em>int</em><em>,</em>) – The dimensionality of the output space |
| i.e. the number of output channels in the convolution.</p></li> |
| <li><p><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 2 ints</em><em>, </em><em>(</em><em>Default value =</em><em> (</em><em>1</em><em>,</em><em>1</em><em>)</em><em>)</em>) – Specifies the dimensions of the convolution window.</p></li> |
| <li><p><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 2 ints</em><em>, </em><em>(</em><em>Default value =</em><em> (</em><em>1</em><em>,</em><em>1</em><em>)</em><em>)</em>) – Specifies the strides of the convolution.</p></li> |
| <li><p><strong>padding</strong> (<em>int</em><em> or </em><em>tuple/list of 2 ints</em><em>, </em><em>(</em><em>Default value =</em><em> (</em><em>0</em><em>,</em><em>0</em><em>)</em><em>)</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points.</p></li> |
| <li><p><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 2 ints</em><em>, </em><em>(</em><em>Default value =</em><em> (</em><em>1</em><em>,</em><em>1</em><em>)</em><em>)</em>) – Specifies the dilation rate to use for dilated convolution.</p></li> |
| <li><p><strong>groups</strong> (<em>int</em><em>, </em><em>(</em><em>Default value = 1</em><em>)</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two convolution |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</p></li> |
| <li><p><strong>num_deformable_group</strong> (<em>int</em><em>, </em><em>(</em><em>Default value = 1</em><em>)</em>) – Number of deformable group partitions.</p></li> |
| <li><p><strong>layout</strong> (<em>str</em><em>, </em><em>(</em><em>Default value = NCHW</em><em>)</em>) – Dimension ordering of data and weight. Can be ‘NCW’, ‘NWC’, ‘NCHW’, |
| ‘NHWC’, ‘NCDHW’, ‘NDHWC’, etc. ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for |
| batch, channel, height, width and depth dimensions respectively. |
| Convolution is performed over ‘D’, ‘H’, and ‘W’ dimensions.</p></li> |
| <li><p><strong>use_bias</strong> (<em>bool</em><em>, </em><em>(</em><em>Default value = True</em><em>)</em>) – Whether the layer for generating the output features uses a bias vector.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>(</em><em>Default value = 0</em><em>)</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and input channels will be inferred from the shape of input data.</p></li> |
| <li><p><strong>activation</strong> (<em>str</em><em>, </em><em>(</em><em>Default value = None</em><em>)</em>) – Activation function to use. See <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.Activation" title="mxnet.ndarray.Activation"><code class="xref py py-func docutils literal notranslate"><span class="pre">Activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</p></li> |
| <li><p><strong>weight_initializer</strong> (str or <cite>Initializer</cite>, (Default value = None)) – Initializer for the <cite>weight</cite> weights matrix for the convolution layer |
| for generating the output features.</p></li> |
| <li><p><strong>bias_initializer</strong> (str or <cite>Initializer</cite>, (Default value = zeros)) – Initializer for the bias vector for the convolution layer |
| for generating the output features.</p></li> |
| <li><p><strong>offset_weight_initializer</strong> (str or <cite>Initializer</cite>, (Default value = zeros)) – Initializer for the <cite>weight</cite> weights matrix for the convolution layer |
| for generating the offset.</p></li> |
| <li><p><strong>offset_bias_initializer</strong> (str or <cite>Initializer</cite>, (Default value = zeros),) – Initializer for the bias vector for the convolution layer |
| for generating the offset.</p></li> |
| <li><p><strong>offset_use_bias</strong> (<em>bool</em><em>, </em><em>(</em><em>Default value = True</em><em>)</em>) – Whether the layer for generating the offset uses a bias vector.</p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><p><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| For other layouts shape is permuted accordingly.</p></li> |
| </ul> |
| </p></li> |
| <li><p><strong>Outputs</strong> – <ul> |
| <li><p><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCHW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.apply" title="mxnet.gluon.nn.ModulatedDeformableConvolution.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.collect_params" title="mxnet.gluon.nn.ModulatedDeformableConvolution.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.export" title="mxnet.gluon.nn.ModulatedDeformableConvolution.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.forward" title="mxnet.gluon.nn.ModulatedDeformableConvolution.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.hybridize" title="mxnet.gluon.nn.ModulatedDeformableConvolution.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.infer_shape" title="mxnet.gluon.nn.ModulatedDeformableConvolution.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(x)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.infer_type" title="mxnet.gluon.nn.ModulatedDeformableConvolution.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.initialize" title="mxnet.gluon.nn.ModulatedDeformableConvolution.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.load" title="mxnet.gluon.nn.ModulatedDeformableConvolution.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.load_dict" title="mxnet.gluon.nn.ModulatedDeformableConvolution.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.load_parameters" title="mxnet.gluon.nn.ModulatedDeformableConvolution.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.optimize_for" title="mxnet.gluon.nn.ModulatedDeformableConvolution.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.pre_infer_offset_weight" title="mxnet.gluon.nn.ModulatedDeformableConvolution.pre_infer_offset_weight"><code class="xref py py-obj docutils literal notranslate"><span class="pre">pre_infer_offset_weight</span></code></a>()</p></td> |
| <td><p>Pre-infer the shape of offsite weight parameter based on kernel size,</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.pre_infer_weight" title="mxnet.gluon.nn.ModulatedDeformableConvolution.pre_infer_weight"><code class="xref py py-obj docutils literal notranslate"><span class="pre">pre_infer_weight</span></code></a>()</p></td> |
| <td><p>Pre-infer the shape of weight parameter based on kernel size, group size and channels</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.register_forward_hook" title="mxnet.gluon.nn.ModulatedDeformableConvolution.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.register_forward_pre_hook" title="mxnet.gluon.nn.ModulatedDeformableConvolution.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.register_op_hook" title="mxnet.gluon.nn.ModulatedDeformableConvolution.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.reset_ctx" title="mxnet.gluon.nn.ModulatedDeformableConvolution.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.reset_device" title="mxnet.gluon.nn.ModulatedDeformableConvolution.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.save" title="mxnet.gluon.nn.ModulatedDeformableConvolution.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.save_parameters" title="mxnet.gluon.nn.ModulatedDeformableConvolution.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.setattr" title="mxnet.gluon.nn.ModulatedDeformableConvolution.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.share_parameters" title="mxnet.gluon.nn.ModulatedDeformableConvolution.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.summary" title="mxnet.gluon.nn.ModulatedDeformableConvolution.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.zero_grad" title="mxnet.gluon.nn.ModulatedDeformableConvolution.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.params" title="mxnet.gluon.nn.ModulatedDeformableConvolution.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#ModulatedDeformableConvolution.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#ModulatedDeformableConvolution.infer_shape"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.pre_infer_offset_weight"> |
| <code class="sig-name descname">pre_infer_offset_weight</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#ModulatedDeformableConvolution.pre_infer_offset_weight"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.pre_infer_offset_weight" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Pre-infer the shape of offsite weight parameter based on kernel size, |
| group size and offset channels</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.pre_infer_weight"> |
| <code class="sig-name descname">pre_infer_weight</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#ModulatedDeformableConvolution.pre_infer_weight"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.pre_infer_weight" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Pre-infer the shape of weight parameter based on kernel size, group size and channels</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.forward" title="mxnet.gluon.nn.ModulatedDeformableConvolution.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.forward" title="mxnet.gluon.nn.ModulatedDeformableConvolution.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ModulatedDeformableConvolution.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ModulatedDeformableConvolution.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.PReLU"> |
| <em class="property">class </em><code class="sig-name descname">PReLU</code><span class="sig-paren">(</span><em class="sig-param">alpha_initializer=<mxnet.initializer.Constant object></em>, <em class="sig-param">in_channels=1</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#PReLU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.PReLU" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Parametric leaky version of a Rectified Linear Unit. |
| <<a class="reference external" href="https://arxiv.org/abs/1502.01852">https://arxiv.org/abs/1502.01852</a>>`_ paper.</p> |
| <p>It learns a gradient when the unit is not active</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}f\left(x\right) = \left\{ |
| \begin{array}{lr} |
| \alpha x & : x \lt 0 \\ |
| x & : x \geq 0 \\ |
| \end{array} |
| \right.\\\end{split}\]</div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.apply" title="mxnet.gluon.nn.PReLU.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.collect_params" title="mxnet.gluon.nn.PReLU.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.export" title="mxnet.gluon.nn.PReLU.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.forward" title="mxnet.gluon.nn.PReLU.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.hybridize" title="mxnet.gluon.nn.PReLU.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.infer_shape" title="mxnet.gluon.nn.PReLU.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.infer_type" title="mxnet.gluon.nn.PReLU.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.initialize" title="mxnet.gluon.nn.PReLU.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.load" title="mxnet.gluon.nn.PReLU.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.load_dict" title="mxnet.gluon.nn.PReLU.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.load_parameters" title="mxnet.gluon.nn.PReLU.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.optimize_for" title="mxnet.gluon.nn.PReLU.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.register_forward_hook" title="mxnet.gluon.nn.PReLU.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.register_forward_pre_hook" title="mxnet.gluon.nn.PReLU.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.register_op_hook" title="mxnet.gluon.nn.PReLU.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.reset_ctx" title="mxnet.gluon.nn.PReLU.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.reset_device" title="mxnet.gluon.nn.PReLU.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.save" title="mxnet.gluon.nn.PReLU.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.save_parameters" title="mxnet.gluon.nn.PReLU.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.setattr" title="mxnet.gluon.nn.PReLU.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.share_parameters" title="mxnet.gluon.nn.PReLU.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.summary" title="mxnet.gluon.nn.PReLU.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.zero_grad" title="mxnet.gluon.nn.PReLU.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PReLU.params" title="mxnet.gluon.nn.PReLU.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p>where alpha is a learned parameter.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>alpha_initializer</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Initializer for the <cite>embeddings</cite> matrix.</p></li> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 1</em>) – Number of channels (alpha parameters) to learn. Can either be 1 |
| or <cite>n</cite> where <cite>n</cite> is the size of the second dimension of the input |
| tensor.</p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </p></li> |
| <li><p><strong>Outputs</strong> – <ul> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#PReLU.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.PReLU.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.PReLU.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.PReLU.forward" title="mxnet.gluon.nn.PReLU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.PReLU.forward" title="mxnet.gluon.nn.PReLU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PReLU.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PReLU.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D"> |
| <em class="property">class </em><code class="sig-name descname">PixelShuffle1D</code><span class="sig-paren">(</span><em class="sig-param">factor</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#PixelShuffle1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Pixel-shuffle layer for upsampling in 1 dimension.</p> |
| <p>Pixel-shuffling is the operation of taking groups of values along |
| the <em>channel</em> dimension and regrouping them into blocks of pixels |
| along the <code class="docutils literal notranslate"><span class="pre">W</span></code> dimension, thereby effectively multiplying that dimension |
| by a constant factor in size.</p> |
| <p>For example, a feature map of shape <span class="math notranslate nohighlight">\((fC, W)\)</span> is reshaped |
| into <span class="math notranslate nohighlight">\((C, fW)\)</span> by forming little value groups of size <span class="math notranslate nohighlight">\(f\)</span> |
| and arranging them in a grid of size <span class="math notranslate nohighlight">\(W\)</span>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>factor</strong> (<em>int</em><em> or </em><em>1-tuple of int</em>) – Upsampling factor, applied to the <code class="docutils literal notranslate"><span class="pre">W</span></code> dimension.</p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><p><strong>data</strong>: Tensor of shape <code class="docutils literal notranslate"><span class="pre">(N,</span> <span class="pre">f*C,</span> <span class="pre">W)</span></code>.</p></li> |
| </ul> |
| </p></li> |
| <li><p><strong>Outputs</strong> – <ul> |
| <li><p><strong>out</strong>: Tensor of shape <code class="docutils literal notranslate"><span class="pre">(N,</span> <span class="pre">C,</span> <span class="pre">W*f)</span></code>.</p></li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.apply" title="mxnet.gluon.nn.PixelShuffle1D.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.collect_params" title="mxnet.gluon.nn.PixelShuffle1D.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.export" title="mxnet.gluon.nn.PixelShuffle1D.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.forward" title="mxnet.gluon.nn.PixelShuffle1D.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Perform pixel-shuffling on the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.hybridize" title="mxnet.gluon.nn.PixelShuffle1D.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.infer_shape" title="mxnet.gluon.nn.PixelShuffle1D.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.infer_type" title="mxnet.gluon.nn.PixelShuffle1D.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.initialize" title="mxnet.gluon.nn.PixelShuffle1D.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.load" title="mxnet.gluon.nn.PixelShuffle1D.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.load_dict" title="mxnet.gluon.nn.PixelShuffle1D.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.load_parameters" title="mxnet.gluon.nn.PixelShuffle1D.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.optimize_for" title="mxnet.gluon.nn.PixelShuffle1D.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.register_forward_hook" title="mxnet.gluon.nn.PixelShuffle1D.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.register_forward_pre_hook" title="mxnet.gluon.nn.PixelShuffle1D.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.register_op_hook" title="mxnet.gluon.nn.PixelShuffle1D.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.reset_ctx" title="mxnet.gluon.nn.PixelShuffle1D.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.reset_device" title="mxnet.gluon.nn.PixelShuffle1D.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.save" title="mxnet.gluon.nn.PixelShuffle1D.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.save_parameters" title="mxnet.gluon.nn.PixelShuffle1D.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.setattr" title="mxnet.gluon.nn.PixelShuffle1D.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.share_parameters" title="mxnet.gluon.nn.PixelShuffle1D.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.summary" title="mxnet.gluon.nn.PixelShuffle1D.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.zero_grad" title="mxnet.gluon.nn.PixelShuffle1D.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.params" title="mxnet.gluon.nn.PixelShuffle1D.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p class="rubric">Examples</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">pxshuf</span> <span class="o">=</span> <span class="n">PixelShuffle1D</span><span class="p">(</span><span class="mi">2</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">np</span><span class="o">.</span><span class="n">zeros</span><span class="p">((</span><span class="mi">1</span><span class="p">,</span> <span class="mi">8</span><span class="p">,</span> <span class="mi">3</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">pxshuf</span><span class="p">(</span><span class="n">x</span><span class="p">)</span><span class="o">.</span><span class="n">shape</span> |
| <span class="go">(1, 4, 6)</span> |
| </pre></div> |
| </div> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#PixelShuffle1D.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Perform pixel-shuffling on the input.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.forward" title="mxnet.gluon.nn.PixelShuffle1D.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle1D.forward" title="mxnet.gluon.nn.PixelShuffle1D.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle1D.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle1D.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D"> |
| <em class="property">class </em><code class="sig-name descname">PixelShuffle2D</code><span class="sig-paren">(</span><em class="sig-param">factor</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#PixelShuffle2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Pixel-shuffle layer for upsampling in 2 dimensions.</p> |
| <p>Pixel-shuffling is the operation of taking groups of values along |
| the <em>channel</em> dimension and regrouping them into blocks of pixels |
| along the <code class="docutils literal notranslate"><span class="pre">H</span></code> and <code class="docutils literal notranslate"><span class="pre">W</span></code> dimensions, thereby effectively multiplying |
| those dimensions by a constant factor in size.</p> |
| <p>For example, a feature map of shape <span class="math notranslate nohighlight">\((f^2 C, H, W)\)</span> is reshaped |
| into <span class="math notranslate nohighlight">\((C, fH, fW)\)</span> by forming little <span class="math notranslate nohighlight">\(f \times f\)</span> blocks |
| of pixels and arranging them in an <span class="math notranslate nohighlight">\(H \times W\)</span> grid.</p> |
| <p>Pixel-shuffling together with regular convolution is an alternative, |
| learnable way of upsampling an image by arbitrary factors. It is reported |
| to help overcome checkerboard artifacts that are common in upsampling with |
| transposed convolutions (also called deconvolutions). See the paper |
| <a class="reference external" href="https://arxiv.org/abs/1609.05158">Real-Time Single Image and Video Super-Resolution Using an Efficient |
| Sub-Pixel Convolutional Neural Network</a> |
| for further details.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>factor</strong> (<em>int</em><em> or </em><em>2-tuple of int</em>) – Upsampling factors, applied to the <code class="docutils literal notranslate"><span class="pre">H</span></code> and <code class="docutils literal notranslate"><span class="pre">W</span></code> dimensions, |
| in that order.</p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><p><strong>data</strong>: Tensor of shape <code class="docutils literal notranslate"><span class="pre">(N,</span> <span class="pre">f1*f2*C,</span> <span class="pre">H,</span> <span class="pre">W)</span></code>.</p></li> |
| </ul> |
| </p></li> |
| <li><p><strong>Outputs</strong> – <ul> |
| <li><p><strong>out</strong>: Tensor of shape <code class="docutils literal notranslate"><span class="pre">(N,</span> <span class="pre">C,</span> <span class="pre">H*f1,</span> <span class="pre">W*f2)</span></code>.</p></li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.apply" title="mxnet.gluon.nn.PixelShuffle2D.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.collect_params" title="mxnet.gluon.nn.PixelShuffle2D.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.export" title="mxnet.gluon.nn.PixelShuffle2D.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.forward" title="mxnet.gluon.nn.PixelShuffle2D.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Perform pixel-shuffling on the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.hybridize" title="mxnet.gluon.nn.PixelShuffle2D.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.infer_shape" title="mxnet.gluon.nn.PixelShuffle2D.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.infer_type" title="mxnet.gluon.nn.PixelShuffle2D.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.initialize" title="mxnet.gluon.nn.PixelShuffle2D.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.load" title="mxnet.gluon.nn.PixelShuffle2D.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.load_dict" title="mxnet.gluon.nn.PixelShuffle2D.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.load_parameters" title="mxnet.gluon.nn.PixelShuffle2D.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.optimize_for" title="mxnet.gluon.nn.PixelShuffle2D.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.register_forward_hook" title="mxnet.gluon.nn.PixelShuffle2D.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.register_forward_pre_hook" title="mxnet.gluon.nn.PixelShuffle2D.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.register_op_hook" title="mxnet.gluon.nn.PixelShuffle2D.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.reset_ctx" title="mxnet.gluon.nn.PixelShuffle2D.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.reset_device" title="mxnet.gluon.nn.PixelShuffle2D.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.save" title="mxnet.gluon.nn.PixelShuffle2D.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.save_parameters" title="mxnet.gluon.nn.PixelShuffle2D.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.setattr" title="mxnet.gluon.nn.PixelShuffle2D.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.share_parameters" title="mxnet.gluon.nn.PixelShuffle2D.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.summary" title="mxnet.gluon.nn.PixelShuffle2D.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.zero_grad" title="mxnet.gluon.nn.PixelShuffle2D.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.params" title="mxnet.gluon.nn.PixelShuffle2D.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p class="rubric">Examples</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">pxshuf</span> <span class="o">=</span> <span class="n">PixelShuffle2D</span><span class="p">((</span><span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">np</span><span class="o">.</span><span class="n">zeros</span><span class="p">((</span><span class="mi">1</span><span class="p">,</span> <span class="mi">12</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">5</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">pxshuf</span><span class="p">(</span><span class="n">x</span><span class="p">)</span><span class="o">.</span><span class="n">shape</span> |
| <span class="go">(1, 2, 6, 15)</span> |
| </pre></div> |
| </div> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#PixelShuffle2D.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Perform pixel-shuffling on the input.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.forward" title="mxnet.gluon.nn.PixelShuffle2D.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle2D.forward" title="mxnet.gluon.nn.PixelShuffle2D.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle2D.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle2D.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D"> |
| <em class="property">class </em><code class="sig-name descname">PixelShuffle3D</code><span class="sig-paren">(</span><em class="sig-param">factor</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#PixelShuffle3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Pixel-shuffle layer for upsampling in 3 dimensions.</p> |
| <p>Pixel-shuffling (or voxel-shuffling in 3D) is the operation of taking |
| groups of values along the <em>channel</em> dimension and regrouping them into |
| blocks of voxels along the <code class="docutils literal notranslate"><span class="pre">D</span></code>, <code class="docutils literal notranslate"><span class="pre">H</span></code> and <code class="docutils literal notranslate"><span class="pre">W</span></code> dimensions, thereby |
| effectively multiplying those dimensions by a constant factor in size.</p> |
| <p>For example, a feature map of shape <span class="math notranslate nohighlight">\((f^3 C, D, H, W)\)</span> is reshaped |
| into <span class="math notranslate nohighlight">\((C, fD, fH, fW)\)</span> by forming little <span class="math notranslate nohighlight">\(f \times f \times f\)</span> |
| blocks of voxels and arranging them in a <span class="math notranslate nohighlight">\(D \times H \times W\)</span> grid.</p> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.apply" title="mxnet.gluon.nn.PixelShuffle3D.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.collect_params" title="mxnet.gluon.nn.PixelShuffle3D.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.export" title="mxnet.gluon.nn.PixelShuffle3D.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.forward" title="mxnet.gluon.nn.PixelShuffle3D.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Perform pixel-shuffling on the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.hybridize" title="mxnet.gluon.nn.PixelShuffle3D.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.infer_shape" title="mxnet.gluon.nn.PixelShuffle3D.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.infer_type" title="mxnet.gluon.nn.PixelShuffle3D.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.initialize" title="mxnet.gluon.nn.PixelShuffle3D.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.load" title="mxnet.gluon.nn.PixelShuffle3D.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.load_dict" title="mxnet.gluon.nn.PixelShuffle3D.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.load_parameters" title="mxnet.gluon.nn.PixelShuffle3D.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.optimize_for" title="mxnet.gluon.nn.PixelShuffle3D.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.register_forward_hook" title="mxnet.gluon.nn.PixelShuffle3D.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.register_forward_pre_hook" title="mxnet.gluon.nn.PixelShuffle3D.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.register_op_hook" title="mxnet.gluon.nn.PixelShuffle3D.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.reset_ctx" title="mxnet.gluon.nn.PixelShuffle3D.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.reset_device" title="mxnet.gluon.nn.PixelShuffle3D.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.save" title="mxnet.gluon.nn.PixelShuffle3D.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.save_parameters" title="mxnet.gluon.nn.PixelShuffle3D.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.setattr" title="mxnet.gluon.nn.PixelShuffle3D.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.share_parameters" title="mxnet.gluon.nn.PixelShuffle3D.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.summary" title="mxnet.gluon.nn.PixelShuffle3D.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.zero_grad" title="mxnet.gluon.nn.PixelShuffle3D.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.params" title="mxnet.gluon.nn.PixelShuffle3D.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p>Pixel-shuffling together with regular convolution is an alternative, |
| learnable way of upsampling an image by arbitrary factors. It is reported |
| to help overcome checkerboard artifacts that are common in upsampling with |
| transposed convolutions (also called deconvolutions). See the paper |
| <a class="reference external" href="https://arxiv.org/abs/1609.05158">Real-Time Single Image and Video Super-Resolution Using an Efficient |
| Sub-Pixel Convolutional Neural Network</a> |
| for further details.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>factor</strong> (<em>int</em><em> or </em><em>3-tuple of int</em>) – Upsampling factors, applied to the <code class="docutils literal notranslate"><span class="pre">D</span></code>, <code class="docutils literal notranslate"><span class="pre">H</span></code> and <code class="docutils literal notranslate"><span class="pre">W</span></code> |
| dimensions, in that order.</p></li> |
| <li><p><strong>Inputs</strong> – <ul> |
| <li><p><strong>data</strong>: Tensor of shape <code class="docutils literal notranslate"><span class="pre">(N,</span> <span class="pre">f1*f2*f3*C,</span> <span class="pre">D,</span> <span class="pre">H,</span> <span class="pre">W)</span></code>.</p></li> |
| </ul> |
| </p></li> |
| <li><p><strong>Outputs</strong> – <ul> |
| <li><p><strong>out</strong>: Tensor of shape <code class="docutils literal notranslate"><span class="pre">(N,</span> <span class="pre">C,</span> <span class="pre">D*f1,</span> <span class="pre">H*f2,</span> <span class="pre">W*f3)</span></code>.</p></li> |
| </ul> |
| </p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">pxshuf</span> <span class="o">=</span> <span class="n">PixelShuffle3D</span><span class="p">((</span><span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">np</span><span class="o">.</span><span class="n">zeros</span><span class="p">((</span><span class="mi">1</span><span class="p">,</span> <span class="mi">48</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">7</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">pxshuf</span><span class="p">(</span><span class="n">x</span><span class="p">)</span><span class="o">.</span><span class="n">shape</span> |
| <span class="go">(1, 2, 6, 15, 28)</span> |
| </pre></div> |
| </div> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#PixelShuffle3D.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Perform pixel-shuffling on the input.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.forward" title="mxnet.gluon.nn.PixelShuffle3D.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.PixelShuffle3D.forward" title="mxnet.gluon.nn.PixelShuffle3D.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.PixelShuffle3D.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.PixelShuffle3D.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D"> |
| <em class="property">class </em><code class="sig-name descname">ReflectionPad2D</code><span class="sig-paren">(</span><em class="sig-param">padding=0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#ReflectionPad2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Pads the input tensor using the reflection of the input boundary.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>padding</strong> (<em>int</em>) – An integer padding size</p> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.apply" title="mxnet.gluon.nn.ReflectionPad2D.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.collect_params" title="mxnet.gluon.nn.ReflectionPad2D.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.export" title="mxnet.gluon.nn.ReflectionPad2D.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.forward" title="mxnet.gluon.nn.ReflectionPad2D.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Use pad operator in numpy extension module,</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.hybridize" title="mxnet.gluon.nn.ReflectionPad2D.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.infer_shape" title="mxnet.gluon.nn.ReflectionPad2D.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.infer_type" title="mxnet.gluon.nn.ReflectionPad2D.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.initialize" title="mxnet.gluon.nn.ReflectionPad2D.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.load" title="mxnet.gluon.nn.ReflectionPad2D.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.load_dict" title="mxnet.gluon.nn.ReflectionPad2D.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.load_parameters" title="mxnet.gluon.nn.ReflectionPad2D.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.optimize_for" title="mxnet.gluon.nn.ReflectionPad2D.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.register_forward_hook" title="mxnet.gluon.nn.ReflectionPad2D.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.register_forward_pre_hook" title="mxnet.gluon.nn.ReflectionPad2D.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.register_op_hook" title="mxnet.gluon.nn.ReflectionPad2D.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.reset_ctx" title="mxnet.gluon.nn.ReflectionPad2D.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.reset_device" title="mxnet.gluon.nn.ReflectionPad2D.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.save" title="mxnet.gluon.nn.ReflectionPad2D.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.save_parameters" title="mxnet.gluon.nn.ReflectionPad2D.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.setattr" title="mxnet.gluon.nn.ReflectionPad2D.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.share_parameters" title="mxnet.gluon.nn.ReflectionPad2D.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.summary" title="mxnet.gluon.nn.ReflectionPad2D.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.zero_grad" title="mxnet.gluon.nn.ReflectionPad2D.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.params" title="mxnet.gluon.nn.ReflectionPad2D.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with the shape <span class="math notranslate nohighlight">\((N, C, H_{in}, W_{in})\)</span>.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul> |
| <li><p><strong>out</strong>: output tensor with the shape <span class="math notranslate nohighlight">\((N, C, H_{out}, W_{out})\)</span>, where</p> |
| <div class="math notranslate nohighlight"> |
| \[ \begin{align}\begin{aligned}H_{out} = H_{in} + 2 \cdot padding\\W_{out} = W_{in} + 2 \cdot padding\end{aligned}\end{align} \]</div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">m</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">ReflectionPad2D</span><span class="p">(</span><span class="mi">3</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="nb">input</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">normal</span><span class="p">(</span><span class="n">size</span><span class="o">=</span><span class="p">(</span><span class="mi">16</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">224</span><span class="p">,</span> <span class="mi">224</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">output</span> <span class="o">=</span> <span class="n">m</span><span class="p">(</span><span class="nb">input</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#ReflectionPad2D.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Use pad operator in numpy extension module, |
| which has backward support for reflect mode</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.forward" title="mxnet.gluon.nn.ReflectionPad2D.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D.forward" title="mxnet.gluon.nn.ReflectionPad2D.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.SELU"> |
| <em class="property">class </em><code class="sig-name descname">SELU</code><span class="sig-paren">(</span><em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#SELU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SELU" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <dl class="simple"> |
| <dt>Scaled Exponential Linear Unit (SELU)</dt><dd><p>“Self-Normalizing Neural Networks”, Klambauer et al, 2017 |
| <a class="reference external" href="https://arxiv.org/abs/1706.02515">https://arxiv.org/abs/1706.02515</a></p> |
| </dd> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.apply" title="mxnet.gluon.nn.SELU.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.collect_params" title="mxnet.gluon.nn.SELU.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.export" title="mxnet.gluon.nn.SELU.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.forward" title="mxnet.gluon.nn.SELU.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.hybridize" title="mxnet.gluon.nn.SELU.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.infer_shape" title="mxnet.gluon.nn.SELU.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.infer_type" title="mxnet.gluon.nn.SELU.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.initialize" title="mxnet.gluon.nn.SELU.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.load" title="mxnet.gluon.nn.SELU.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.load_dict" title="mxnet.gluon.nn.SELU.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.load_parameters" title="mxnet.gluon.nn.SELU.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.optimize_for" title="mxnet.gluon.nn.SELU.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.register_forward_hook" title="mxnet.gluon.nn.SELU.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.register_forward_pre_hook" title="mxnet.gluon.nn.SELU.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.register_op_hook" title="mxnet.gluon.nn.SELU.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.reset_ctx" title="mxnet.gluon.nn.SELU.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.reset_device" title="mxnet.gluon.nn.SELU.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.save" title="mxnet.gluon.nn.SELU.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.save_parameters" title="mxnet.gluon.nn.SELU.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.setattr" title="mxnet.gluon.nn.SELU.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.share_parameters" title="mxnet.gluon.nn.SELU.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.summary" title="mxnet.gluon.nn.SELU.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.zero_grad" title="mxnet.gluon.nn.SELU.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SELU.params" title="mxnet.gluon.nn.SELU.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#SELU.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SELU.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.SELU.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.SELU.forward" title="mxnet.gluon.nn.SELU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.SELU.forward" title="mxnet.gluon.nn.SELU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SELU.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SELU.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Sequential"> |
| <em class="property">class </em><code class="sig-name descname">Sequential</code><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Sequential"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Sequential" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.Block</span></code></p> |
| <p>Stacks Blocks sequentially.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">net</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Sequential</span><span class="p">()</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="n">activation</span><span class="o">=</span><span class="s1">'relu'</span><span class="p">))</span> |
| <span class="n">net</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">))</span> |
| </pre></div> |
| </div> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Sequential.add" title="mxnet.gluon.nn.Sequential.add"><code class="xref py py-obj docutils literal notranslate"><span class="pre">add</span></code></a>(*blocks)</p></td> |
| <td><p>Adds block on top of the stack.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Sequential.forward" title="mxnet.gluon.nn.Sequential.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x, *args)</p></td> |
| <td><p>Overrides to implement forward computation using <code class="xref py py-class docutils literal notranslate"><span class="pre">NDArray</span></code>.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Sequential.hybridize" title="mxnet.gluon.nn.Sequential.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active])</p></td> |
| <td><p>Activates or deactivates <cite>HybridBlock</cite> s recursively.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Sequential.add"> |
| <code class="sig-name descname">add</code><span class="sig-paren">(</span><em class="sig-param">*blocks</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Sequential.add"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Sequential.add" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Adds block on top of the stack.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Sequential.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Sequential.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Sequential.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides to implement forward computation using <code class="xref py py-class docutils literal notranslate"><span class="pre">NDArray</span></code>. Only |
| accepts positional arguments.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>*args</strong> (<em>list of NDArray</em>) – Input tensors.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Sequential.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Sequential.hybridize"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Sequential.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <cite>HybridBlock</cite> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>string</em>) – Additional flags for hybridized operator.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.SiLU"> |
| <em class="property">class </em><code class="sig-name descname">SiLU</code><span class="sig-paren">(</span><em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#SiLU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SiLU" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <dl class="simple"> |
| <dt>Sigmoid Linear Units</dt><dd><p>Originally proposed “Gaussian Error Linear Units (GELUs)”, Hendrycks et al, 2016 |
| <a class="reference external" href="https://arxiv.org/abs/1606.08415">https://arxiv.org/abs/1606.08415</a></p> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.apply" title="mxnet.gluon.nn.SiLU.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.collect_params" title="mxnet.gluon.nn.SiLU.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.export" title="mxnet.gluon.nn.SiLU.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.forward" title="mxnet.gluon.nn.SiLU.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.hybridize" title="mxnet.gluon.nn.SiLU.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.infer_shape" title="mxnet.gluon.nn.SiLU.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.infer_type" title="mxnet.gluon.nn.SiLU.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.initialize" title="mxnet.gluon.nn.SiLU.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.load" title="mxnet.gluon.nn.SiLU.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.load_dict" title="mxnet.gluon.nn.SiLU.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.load_parameters" title="mxnet.gluon.nn.SiLU.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.optimize_for" title="mxnet.gluon.nn.SiLU.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.register_forward_hook" title="mxnet.gluon.nn.SiLU.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.register_forward_pre_hook" title="mxnet.gluon.nn.SiLU.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.register_op_hook" title="mxnet.gluon.nn.SiLU.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.reset_ctx" title="mxnet.gluon.nn.SiLU.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.reset_device" title="mxnet.gluon.nn.SiLU.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.save" title="mxnet.gluon.nn.SiLU.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.save_parameters" title="mxnet.gluon.nn.SiLU.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.setattr" title="mxnet.gluon.nn.SiLU.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.share_parameters" title="mxnet.gluon.nn.SiLU.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.summary" title="mxnet.gluon.nn.SiLU.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.zero_grad" title="mxnet.gluon.nn.SiLU.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SiLU.params" title="mxnet.gluon.nn.SiLU.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>beta</strong> (<em>float</em>) – silu(x) = x * sigmoid(x)</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#SiLU.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SiLU.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.SiLU.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.SiLU.forward" title="mxnet.gluon.nn.SiLU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.SiLU.forward" title="mxnet.gluon.nn.SiLU.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SiLU.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SiLU.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Swish"> |
| <em class="property">class </em><code class="sig-name descname">Swish</code><span class="sig-paren">(</span><em class="sig-param">beta=1.0</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#Swish"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Swish" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <dl class="simple"> |
| <dt>Swish Activation function (SiLU with a hyperparameter)</dt><dd><p><a class="reference external" href="https://arxiv.org/pdf/1710.05941.pdf">https://arxiv.org/pdf/1710.05941.pdf</a></p> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.apply" title="mxnet.gluon.nn.Swish.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.collect_params" title="mxnet.gluon.nn.Swish.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.export" title="mxnet.gluon.nn.Swish.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.forward" title="mxnet.gluon.nn.Swish.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.hybridize" title="mxnet.gluon.nn.Swish.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.infer_shape" title="mxnet.gluon.nn.Swish.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.infer_type" title="mxnet.gluon.nn.Swish.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.initialize" title="mxnet.gluon.nn.Swish.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.load" title="mxnet.gluon.nn.Swish.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.load_dict" title="mxnet.gluon.nn.Swish.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.load_parameters" title="mxnet.gluon.nn.Swish.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.optimize_for" title="mxnet.gluon.nn.Swish.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.register_forward_hook" title="mxnet.gluon.nn.Swish.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.register_forward_pre_hook" title="mxnet.gluon.nn.Swish.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.register_op_hook" title="mxnet.gluon.nn.Swish.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.reset_ctx" title="mxnet.gluon.nn.Swish.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.reset_device" title="mxnet.gluon.nn.Swish.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.save" title="mxnet.gluon.nn.Swish.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.save_parameters" title="mxnet.gluon.nn.Swish.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.setattr" title="mxnet.gluon.nn.Swish.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.share_parameters" title="mxnet.gluon.nn.Swish.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.summary" title="mxnet.gluon.nn.Swish.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.zero_grad" title="mxnet.gluon.nn.Swish.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.Swish.params" title="mxnet.gluon.nn.Swish.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>beta</strong> (<em>float</em>) – swish(x) = x * sigmoid(beta*x)</p> |
| </dd> |
| </dl> |
| <dl class="simple"> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#Swish.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Swish.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.Swish.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.Swish.forward" title="mxnet.gluon.nn.Swish.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.Swish.forward" title="mxnet.gluon.nn.Swish.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.Swish.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.Swish.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.SymbolBlock"> |
| <em class="property">class </em><code class="sig-name descname">SymbolBlock</code><span class="sig-paren">(</span><em class="sig-param">outputs</em>, <em class="sig-param">inputs</em>, <em class="sig-param">params=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#SymbolBlock"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SymbolBlock" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.block.HybridBlock</span></code></p> |
| <p>Construct block from symbol. This is useful for using pre-trained models |
| as feature extractors. For example, you may want to extract the output |
| from fc2 layer in AlexNet.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>outputs</strong> (<a class="reference internal" href="../../legacy/symbol/symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em> or </em><em>list of Symbol</em>) – The desired output for SymbolBlock.</p></li> |
| <li><p><strong>inputs</strong> (<a class="reference internal" href="../../legacy/symbol/symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em> or </em><em>list of Symbol</em>) – The Variables in output’s argument that should be used as inputs.</p></li> |
| <li><p><strong>params</strong> (<em>dict</em>) – Parameter dictionary for arguments and auxililary states of outputs |
| that are not inputs.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SymbolBlock.cast" title="mxnet.gluon.nn.SymbolBlock.cast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cast</span></code></a>(dtype)</p></td> |
| <td><p>Cast this Block to use another data type.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SymbolBlock.forward" title="mxnet.gluon.nn.SymbolBlock.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x, *args)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SymbolBlock.imports" title="mxnet.gluon.nn.SymbolBlock.imports"><code class="xref py py-obj docutils literal notranslate"><span class="pre">imports</span></code></a>(symbol_file, input_names[, …])</p></td> |
| <td><p>Import model previously saved by <cite>gluon.HybridBlock.export</cite> as a <cite>gluon.SymbolBlock</cite> for use in Gluon.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SymbolBlock.infer_shape" title="mxnet.gluon.nn.SymbolBlock.infer_shape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_shape</span></code></a>(*args)</p></td> |
| <td><p>Infers shape of Parameters from inputs.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p class="rubric">Examples</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="c1"># To extract the feature from fc1 and fc2 layers of AlexNet:</span> |
| <span class="gp">>>> </span><span class="n">alexnet</span> <span class="o">=</span> <span class="n">gluon</span><span class="o">.</span><span class="n">model_zoo</span><span class="o">.</span><span class="n">vision</span><span class="o">.</span><span class="n">alexnet</span><span class="p">(</span><span class="n">pretrained</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">device</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">())</span> |
| <span class="gp">>>> </span><span class="n">inputs</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">sym</span><span class="o">.</span><span class="n">var</span><span class="p">(</span><span class="s1">'data'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">out</span> <span class="o">=</span> <span class="n">alexnet</span><span class="p">(</span><span class="n">inputs</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">internals</span> <span class="o">=</span> <span class="n">out</span><span class="o">.</span><span class="n">get_internals</span><span class="p">()</span> |
| <span class="gp">>>> </span><span class="nb">print</span><span class="p">(</span><span class="n">internals</span><span class="o">.</span><span class="n">list_outputs</span><span class="p">())</span> |
| <span class="go">['data', ..., 'features_9_act_fwd_output', ..., 'features_11_act_fwd_output', ...]</span> |
| <span class="gp">>>> </span><span class="n">outputs</span> <span class="o">=</span> <span class="p">[</span><span class="n">internals</span><span class="p">[</span><span class="s1">'features_9_act_fwd_output'</span><span class="p">],</span> |
| <span class="go"> internals['features_11_act_fwd_output']]</span> |
| <span class="gp">>>> </span><span class="c1"># Create SymbolBlock that shares parameters with alexnet</span> |
| <span class="gp">>>> </span><span class="n">feat_model</span> <span class="o">=</span> <span class="n">gluon</span><span class="o">.</span><span class="n">SymbolBlock</span><span class="p">(</span><span class="n">outputs</span><span class="p">,</span> <span class="n">inputs</span><span class="p">,</span> <span class="n">params</span><span class="o">=</span><span class="n">alexnet</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">nd</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">normal</span><span class="p">(</span><span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">16</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">224</span><span class="p">,</span> <span class="mi">224</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="nb">print</span><span class="p">(</span><span class="n">feat_model</span><span class="p">(</span><span class="n">x</span><span class="p">))</span> |
| </pre></div> |
| </div> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SymbolBlock.cast"> |
| <code class="sig-name descname">cast</code><span class="sig-paren">(</span><em class="sig-param">dtype</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#SymbolBlock.cast"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SymbolBlock.cast" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Cast this Block to use another data type.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>dtype</strong> (<em>str</em><em> or </em><em>numpy.dtype</em>) – The new data type.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SymbolBlock.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#SymbolBlock.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SymbolBlock.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SymbolBlock.imports"> |
| <em class="property">static </em><code class="sig-name descname">imports</code><span class="sig-paren">(</span><em class="sig-param">symbol_file</em>, <em class="sig-param">input_names</em>, <em class="sig-param">param_file=None</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#SymbolBlock.imports"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SymbolBlock.imports" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Import model previously saved by <cite>gluon.HybridBlock.export</cite> |
| as a <cite>gluon.SymbolBlock</cite> for use in Gluon.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>symbol_file</strong> (<em>str</em>) – Path to symbol file.</p></li> |
| <li><p><strong>input_names</strong> (<em>list of str</em>) – List of input variable names</p></li> |
| <li><p><strong>param_file</strong> (<em>str</em><em>, </em><em>optional</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>default None</em>) – The device to initialize <cite>gluon.SymbolBlock</cite> on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><cite>gluon.SymbolBlock</cite> loaded from symbol and parameter files.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol_block.html#mxnet.gluon.SymbolBlock" title="mxnet.gluon.SymbolBlock">gluon.SymbolBlock</a></p> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">net1</span> <span class="o">=</span> <span class="n">gluon</span><span class="o">.</span><span class="n">model_zoo</span><span class="o">.</span><span class="n">vision</span><span class="o">.</span><span class="n">resnet18_v1</span><span class="p">(</span><span class="n">pretrained</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">net1</span><span class="o">.</span><span class="n">hybridize</span><span class="p">()</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">nd</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">normal</span><span class="p">(</span><span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">32</span><span class="p">,</span> <span class="mi">32</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">out1</span> <span class="o">=</span> <span class="n">net1</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">net1</span><span class="o">.</span><span class="n">export</span><span class="p">(</span><span class="s1">'net1'</span><span class="p">,</span> <span class="n">epoch</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> |
| <span class="go">>>></span> |
| <span class="gp">>>> </span><span class="n">net2</span> <span class="o">=</span> <span class="n">gluon</span><span class="o">.</span><span class="n">SymbolBlock</span><span class="o">.</span><span class="n">imports</span><span class="p">(</span> |
| <span class="gp">... </span> <span class="s1">'net1-symbol.json'</span><span class="p">,</span> <span class="p">[</span><span class="s1">'data'</span><span class="p">],</span> <span class="s1">'net1-0001.params'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">out2</span> <span class="o">=</span> <span class="n">net2</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SymbolBlock.infer_shape"> |
| <code class="sig-name descname">infer_shape</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/block.html#SymbolBlock.infer_shape"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SymbolBlock.infer_shape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers shape of Parameters from inputs.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm"> |
| <em class="property">class </em><code class="sig-name descname">SyncBatchNorm</code><span class="sig-paren">(</span><em class="sig-param">in_channels=0</em>, <em class="sig-param">num_devices=None</em>, <em class="sig-param">momentum=0.9</em>, <em class="sig-param">epsilon=1e-05</em>, <em class="sig-param">center=True</em>, <em class="sig-param">scale=True</em>, <em class="sig-param">use_global_stats=False</em>, <em class="sig-param">beta_initializer='zeros'</em>, <em class="sig-param">gamma_initializer='ones'</em>, <em class="sig-param">running_mean_initializer='zeros'</em>, <em class="sig-param">running_variance_initializer='ones'</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#SyncBatchNorm"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.nn.basic_layers.BatchNorm</span></code></p> |
| <p>Cross-GPU Synchronized Batch normalization (SyncBN)</p> |
| <p>Standard BN <a class="footnote-reference brackets" href="#id57" id="id55">1</a> implementation only normalize the data within each device. |
| SyncBN normalizes the input within the whole mini-batch. |
| We follow the implementation described in the paper <a class="footnote-reference brackets" href="#id58" id="id56">2</a>.</p> |
| <p>Note: Current implementation of SyncBN does not support FP16 training. |
| For FP16 inference, use standard nn.BatchNorm instead of SyncBN.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – Number of channels (feature maps) in input data. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</p></li> |
| <li><p><strong>num_devices</strong> (<em>int</em><em>, </em><em>default number of visible GPUs</em>) – </p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>default 0.9</em>) – Momentum for the moving average.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>default 1e-5</em>) – Small float added to variance to avoid dividing by zero.</p></li> |
| <li><p><strong>center</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, add offset of <cite>beta</cite> to normalized tensor. |
| If False, <cite>beta</cite> is ignored.</p></li> |
| <li><p><strong>scale</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, multiply by <cite>gamma</cite>. If False, <cite>gamma</cite> is not used. |
| When the next layer is linear (also e.g. <cite>nn.relu</cite>), |
| this can be disabled since the scaling |
| will be done by the next layer.</p></li> |
| <li><p><strong>use_global_stats</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, use global moving statistics instead of local batch-norm. This will force |
| change batch-norm into a scale shift operator. |
| If False, use local batch-norm.</p></li> |
| <li><p><strong>beta_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the beta weight.</p></li> |
| <li><p><strong>gamma_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the gamma weight.</p></li> |
| <li><p><strong>running_mean_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the running mean.</p></li> |
| <li><p><strong>running_variance_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the running variance.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p><strong>Methods</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.apply" title="mxnet.gluon.nn.SyncBatchNorm.apply"><code class="xref py py-obj docutils literal notranslate"><span class="pre">apply</span></code></a>(fn)</p></td> |
| <td><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.collect_params" title="mxnet.gluon.nn.SyncBatchNorm.collect_params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">collect_params</span></code></a>([select])</p></td> |
| <td><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> which match some given regular expressions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.export" title="mxnet.gluon.nn.SyncBatchNorm.export"><code class="xref py py-obj docutils literal notranslate"><span class="pre">export</span></code></a>(path[, epoch, remove_amp_cast])</p></td> |
| <td><p>Export HybridBlock to json format that can be loaded by <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.forward" title="mxnet.gluon.nn.SyncBatchNorm.forward"><code class="xref py py-obj docutils literal notranslate"><span class="pre">forward</span></code></a>(x)</p></td> |
| <td><p>Overrides the forward computation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.hybridize" title="mxnet.gluon.nn.SyncBatchNorm.hybridize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hybridize</span></code></a>([active, partition_if_dynamic, …])</p></td> |
| <td><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.infer_type" title="mxnet.gluon.nn.SyncBatchNorm.infer_type"><code class="xref py py-obj docutils literal notranslate"><span class="pre">infer_type</span></code></a>(*args)</p></td> |
| <td><p>Infers data type of Parameters from inputs.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.initialize" title="mxnet.gluon.nn.SyncBatchNorm.initialize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">initialize</span></code></a>([init, device, verbose, force_reinit])</p></td> |
| <td><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.load" title="mxnet.gluon.nn.SyncBatchNorm.load"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load</span></code></a>(prefix)</p></td> |
| <td><p>Load a model saved using the <cite>save</cite> API</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.load_dict" title="mxnet.gluon.nn.SyncBatchNorm.load_dict"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_dict</span></code></a>(param_dict[, device, …])</p></td> |
| <td><p>Load parameters from dict</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.load_parameters" title="mxnet.gluon.nn.SyncBatchNorm.load_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">load_parameters</span></code></a>(filename[, device, …])</p></td> |
| <td><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.optimize_for" title="mxnet.gluon.nn.SyncBatchNorm.optimize_for"><code class="xref py py-obj docutils literal notranslate"><span class="pre">optimize_for</span></code></a>(x, *args[, backend, clear, …])</p></td> |
| <td><p>Partitions the current HybridBlock and optimizes it for a given backend without executing a forward pass.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.register_forward_hook" title="mxnet.gluon.nn.SyncBatchNorm.register_forward_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward hook on the block.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.register_forward_pre_hook" title="mxnet.gluon.nn.SyncBatchNorm.register_forward_pre_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_forward_pre_hook</span></code></a>(hook)</p></td> |
| <td><p>Registers a forward pre-hook on the block.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.register_op_hook" title="mxnet.gluon.nn.SyncBatchNorm.register_op_hook"><code class="xref py py-obj docutils literal notranslate"><span class="pre">register_op_hook</span></code></a>(callback[, monitor_all])</p></td> |
| <td><p>Install op hook for block recursively.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.reset_ctx" title="mxnet.gluon.nn.SyncBatchNorm.reset_ctx"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_ctx</span></code></a>(ctx)</p></td> |
| <td><p>This function has been deprecated.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.reset_device" title="mxnet.gluon.nn.SyncBatchNorm.reset_device"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_device</span></code></a>(device)</p></td> |
| <td><p>Re-assign all Parameters to other devices.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.save" title="mxnet.gluon.nn.SyncBatchNorm.save"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save</span></code></a>(prefix)</p></td> |
| <td><p>Save the model architecture and parameters to load again later</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.save_parameters" title="mxnet.gluon.nn.SyncBatchNorm.save_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">save_parameters</span></code></a>(filename[, deduplicate])</p></td> |
| <td><p>Save parameters to file.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.setattr" title="mxnet.gluon.nn.SyncBatchNorm.setattr"><code class="xref py py-obj docutils literal notranslate"><span class="pre">setattr</span></code></a>(name, value)</p></td> |
| <td><p>Set an attribute to a new value for all Parameters.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.share_parameters" title="mxnet.gluon.nn.SyncBatchNorm.share_parameters"><code class="xref py py-obj docutils literal notranslate"><span class="pre">share_parameters</span></code></a>(shared)</p></td> |
| <td><p>Share parameters recursively inside the model.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.summary" title="mxnet.gluon.nn.SyncBatchNorm.summary"><code class="xref py py-obj docutils literal notranslate"><span class="pre">summary</span></code></a>(*inputs)</p></td> |
| <td><p>Print the summary of the model’s output and parameters.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.zero_grad" title="mxnet.gluon.nn.SyncBatchNorm.zero_grad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zero_grad</span></code></a>()</p></td> |
| <td><p>Sets all Parameters’ gradient buffer to 0.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <p><strong>Attributes</strong></p> |
| <table class="longtable docutils align-default"> |
| <colgroup> |
| <col style="width: 10%" /> |
| <col style="width: 90%" /> |
| </colgroup> |
| <tbody> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.params" title="mxnet.gluon.nn.SyncBatchNorm.params"><code class="xref py py-obj docutils literal notranslate"><span class="pre">params</span></code></a></p></td> |
| <td><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its children’s parameters).</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl> |
| <dt>Inputs:</dt><dd><ul class="simple"> |
| <li><p><strong>data</strong>: input tensor with arbitrary shape.</p></li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt><dd><ul class="simple"> |
| <li><p><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</p></li> |
| </ul> |
| </dd> |
| <dt>Reference:</dt><dd><dl class="footnote brackets"> |
| <dt class="label" id="id57"><span class="brackets"><a class="fn-backref" href="#id55">1</a></span></dt> |
| <dd><p>Ioffe, Sergey, and Christian Szegedy. “Batch normalization: Accelerating deep network training by reducing internal covariate shift.” <em>ICML 2015</em></p> |
| </dd> |
| <dt class="label" id="id58"><span class="brackets"><a class="fn-backref" href="#id56">2</a></span></dt> |
| <dd><p>Hang Zhang, Kristin Dana, Jianping Shi, Zhongyue Zhang, Xiaogang Wang, Ambrish Tyagi, and Amit Agrawal. “Context Encoding for Semantic Segmentation.” <em>CVPR 2018</em></p> |
| </dd> |
| </dl> |
| </dd> |
| </dl> |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.apply"> |
| <code class="sig-name descname">apply</code><span class="sig-paren">(</span><em class="sig-param">fn</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.apply" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies <code class="docutils literal notranslate"><span class="pre">fn</span></code> recursively to every child block as well as self.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>fn</strong> (<em>callable</em>) – Function to be applied to each submodule, of form <cite>fn(block)</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.collect_params"> |
| <code class="sig-name descname">collect_params</code><span class="sig-paren">(</span><em class="sig-param">select=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.collect_params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> containing this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and all of its |
| children’s Parameters(default), also can returns the select <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code> |
| which match some given regular expressions.</p> |
| <p>For example, collect the specified parameters in [‘conv1.weight’, ‘conv1.bias’, ‘fc.weight’, |
| ‘fc.bias’]:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'conv1.weight|conv1.bias|fc.weight|fc.bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or collect all parameters whose names end with ‘weight’ or ‘bias’, this can be done |
| using regular expressions:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">collect_params</span><span class="p">(</span><span class="s1">'.*weight|.*bias'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>select</strong> (<em>str</em>) – regular expressions</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>The selected <code class="xref py py-class docutils literal notranslate"><span class="pre">Dict</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.export"> |
| <code class="sig-name descname">export</code><span class="sig-paren">(</span><em class="sig-param">path</em>, <em class="sig-param">epoch=0</em>, <em class="sig-param">remove_amp_cast=True</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.export" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Export HybridBlock to json format that can be loaded by |
| <cite>gluon.SymbolBlock.imports</cite> or the C++ interface.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When there are only one input, it will have name <cite>data</cite>. When there |
| Are more than one inputs, they will be named as <cite>data0</cite>, <cite>data1</cite>, etc.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>path</strong> (<em>str</em><em> or </em><em>None</em>) – Path to save model. Two files <cite>path-symbol.json</cite> and <cite>path-xxxx.params</cite> |
| will be created, where xxxx is the 4 digits epoch number. |
| If None, do not export to file but return Python Symbol object and |
| corresponding dictionary of parameters.</p></li> |
| <li><p><strong>epoch</strong> (<em>int</em>) – Epoch number of saved model.</p></li> |
| <li><p><strong>remove_amp_cast</strong> (<em>bool</em><em>, </em><em>optional</em>) – Whether to remove the amp_cast and amp_multicast operators, before saving the model.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p><ul class="simple"> |
| <li><p><strong>symbol_filename</strong> (<em>str</em>) – Filename to which model symbols were saved, including <cite>path</cite> prefix.</p></li> |
| <li><p><strong>params_filename</strong> (<em>str</em>) – Filename to which model parameters were saved, including <cite>path</cite> prefix.</p></li> |
| </ul> |
| </p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.forward"> |
| <code class="sig-name descname">forward</code><span class="sig-paren">(</span><em class="sig-param">x</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#SyncBatchNorm.forward"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.forward" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Overrides the forward computation. Arguments must be |
| <code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.numpy.ndarray</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.hybridize"> |
| <code class="sig-name descname">hybridize</code><span class="sig-paren">(</span><em class="sig-param">active=True</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.hybridize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Activates or deactivates <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock" title="mxnet.gluon.nn.HybridBlock"><code class="xref py py-class docutils literal notranslate"><span class="pre">HybridBlock</span></code></a> s recursively. Has no effect on |
| non-hybrid children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>active</strong> (<em>bool</em><em>, </em><em>default True</em>) – Whether to turn hybrid on or off.</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.infer_type"> |
| <code class="sig-name descname">infer_type</code><span class="sig-paren">(</span><em class="sig-param">*args</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.infer_type" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Infers data type of Parameters from inputs.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.initialize"> |
| <code class="sig-name descname">initialize</code><span class="sig-paren">(</span><em class="sig-param">init=<mxnet.initializer.Uniform object></em>, <em class="sig-param">device=None</em>, <em class="sig-param">verbose=False</em>, <em class="sig-param">force_reinit=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.initialize" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Initializes <code class="xref py py-class docutils literal notranslate"><span class="pre">Parameter</span></code> s of this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a> and its children.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>init</strong> (<a class="reference internal" href="../../initializer/index.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Global default Initializer to be used when <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> is <code class="docutils literal notranslate"><span class="pre">None</span></code>. |
| Otherwise, <code class="xref py py-meth docutils literal notranslate"><span class="pre">Parameter.init()</span></code> takes precedence.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em>) – Keeps a copy of Parameters on one or many device(s).</p></li> |
| <li><p><strong>verbose</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to verbosely print out details on initialization.</p></li> |
| <li><p><strong>force_reinit</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to force re-initialization if parameter is already initialized.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.load"> |
| <code class="sig-name descname">load</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.load" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load a model saved using the <cite>save</cite> API</p> |
| <p>Reconfigures a model using the saved configuration. This function |
| does not regenerate the model architecture. It resets each Block’s |
| parameter UUIDs as they were when saved in order to match the names of the |
| saved parameters.</p> |
| <p>This function assumes the Blocks in the model were created in the same |
| order they were when the model was saved. This is because each Block is |
| uniquely identified by Block class name and a unique ID in order (since |
| its an OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph (Symbol & inputs) and settings are |
| restored if it had been hybridized before saving.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for loading this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.load_dict"> |
| <code class="sig-name descname">load_dict</code><span class="sig-paren">(</span><em class="sig-param">param_dict</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.load_dict" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from dict</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>param_dict</strong> (<em>dict</em>) – Dictionary containing model parameters</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em>, </em><em>optional</em>) – Device context on which the memory is allocated. Default is |
| <cite>mxnet.device.current_device()</cite>.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represented in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this dict.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.load_parameters"> |
| <code class="sig-name descname">load_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">device=None</em>, <em class="sig-param">allow_missing=False</em>, <em class="sig-param">ignore_extra=False</em>, <em class="sig-param">cast_dtype=False</em>, <em class="sig-param">dtype_source='current'</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.load_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Load parameters from file previously saved by <cite>save_parameters</cite>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to parameter file.</p></li> |
| <li><p><strong>device</strong> (<a class="reference internal" href="../../device/index.html#mxnet.device.Device" title="mxnet.device.Device"><em>Device</em></a><em> or </em><em>list of Device</em><em>, </em><em>default cpu</em><em>(</em><em>)</em>) – Device(s) to initialize loaded parameters on.</p></li> |
| <li><p><strong>allow_missing</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently skip loading parameters not represents in the file.</p></li> |
| <li><p><strong>ignore_extra</strong> (<em>bool</em><em>, </em><em>default False</em>) – Whether to silently ignore parameters from the file that are not |
| present in this Block.</p></li> |
| <li><p><strong>cast_dtype</strong> (<em>bool</em><em>, </em><em>default False</em>) – Cast the data type of the NDArray loaded from the checkpoint to the dtype |
| provided by the Parameter if any.</p></li> |
| <li><p><strong>dtype_source</strong> (<em>str</em><em>, </em><em>default 'current'</em>) – must be in {‘current’, ‘saved’} |
| Only valid if cast_dtype=True, specify the source of the dtype for casting |
| the parameters</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.optimize_for"> |
| <code class="sig-name descname">optimize_for</code><span class="sig-paren">(</span><em class="sig-param">x</em>, <em class="sig-param">*args</em>, <em class="sig-param">backend=None</em>, <em class="sig-param">clear=False</em>, <em class="sig-param">partition_if_dynamic=True</em>, <em class="sig-param">static_alloc=False</em>, <em class="sig-param">static_shape=False</em>, <em class="sig-param">inline_limit=2</em>, <em class="sig-param">forward_bulk_size=None</em>, <em class="sig-param">backward_bulk_size=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.optimize_for" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Partitions the current HybridBlock and optimizes it for a given backend |
| without executing a forward pass. Modifies the HybridBlock in-place.</p> |
| <p>Immediately partitions a HybridBlock using the specified backend. Combines |
| the work done in the hybridize API with part of the work done in the forward |
| pass without calling the CachedOp. Can be used in place of hybridize, |
| afterwards <cite>export</cite> can be called or inference can be run. See README.md in |
| example/extensions/lib_subgraph/README.md for more details.</p> |
| <p class="rubric">Examples</p> |
| <p># partition and then export to file |
| block.optimize_for(x, backend=’myPart’) |
| block.export(‘partitioned’)</p> |
| <p># partition and then run inference |
| block.optimize_for(x, backend=’myPart’) |
| block(x)</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – first input to model</p></li> |
| <li><p><strong>*args</strong> (<a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – other inputs to model</p></li> |
| <li><p><strong>backend</strong> (<em>str</em>) – The name of backend, as registered in <cite>SubgraphBackendRegistry</cite>, default None</p></li> |
| <li><p><strong>backend_opts</strong> (<em>dict of user-specified options to pass to the backend for partitioning</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| <li><p><strong>clear</strong> (<em>bool</em><em>, </em><em>default False</em>) – clears any previous optimizations</p></li> |
| <li><p><strong>partition_if_dynamic</strong> (<em>bool</em><em>, </em><em>default False</em>) – whether to partition the graph when dynamic shape op exists</p></li> |
| <li><p><strong>static_alloc</strong> (<em>bool</em><em>, </em><em>default False</em>) – Statically allocate memory to improve speed. Memory usage may increase.</p></li> |
| <li><p><strong>static_shape</strong> (<em>bool</em><em>, </em><em>default False</em>) – Optimize for invariant input shapes between iterations. Must also |
| set static_alloc to True. Change of input shapes is still allowed |
| but slower.</p></li> |
| <li><p><strong>inline_limit</strong> (<em>optional int</em><em>, </em><em>default 2</em>) – Maximum number of operators that can be inlined.</p></li> |
| <li><p><strong>forward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during forward pass.</p></li> |
| <li><p><strong>backward_bulk_size</strong> (<em>optional int</em><em>, </em><em>default None</em>) – Segment size of bulk execution during backward pass.</p></li> |
| <li><p><strong>**kwargs</strong> (<em>The backend options</em><em>, </em><em>optional</em>) – Passed on to <cite>PrePartition</cite> and <cite>PostPartition</cite> functions of <cite>SubgraphProperty</cite></p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.params"> |
| <em class="property">property </em><code class="sig-name descname">params</code><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.params" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns this <a class="reference internal" href="#mxnet.gluon.nn.Block" title="mxnet.gluon.nn.Block"><code class="xref py py-class docutils literal notranslate"><span class="pre">Block</span></code></a>’s parameter dictionary (does not include its |
| children’s parameters).</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.register_forward_hook"> |
| <code class="sig-name descname">register_forward_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.register_forward_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward hook on the block.</p> |
| <p>The hook function is called immediately after <a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.forward" title="mxnet.gluon.nn.SyncBatchNorm.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input, output) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.register_forward_pre_hook"> |
| <code class="sig-name descname">register_forward_pre_hook</code><span class="sig-paren">(</span><em class="sig-param">hook</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.register_forward_pre_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Registers a forward pre-hook on the block.</p> |
| <p>The hook function is called immediately before <a class="reference internal" href="#mxnet.gluon.nn.SyncBatchNorm.forward" title="mxnet.gluon.nn.SyncBatchNorm.forward"><code class="xref py py-func docutils literal notranslate"><span class="pre">forward()</span></code></a>. |
| It should not modify the input or output.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>hook</strong> (<em>callable</em>) – The forward hook function of form <cite>hook(block, input) -> None</cite>.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.gluon.utils.HookHandle</span></code></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.register_op_hook"> |
| <code class="sig-name descname">register_op_hook</code><span class="sig-paren">(</span><em class="sig-param">callback</em>, <em class="sig-param">monitor_all=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.register_op_hook" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Install op hook for block recursively.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>callback</strong> (<em>function</em>) – Function called to inspect the values of the intermediate outputs |
| of blocks after hybridization. It takes 3 parameters: |
| name of the tensor being inspected (str) |
| name of the operator producing or consuming that tensor (str) |
| tensor being inspected (NDArray).</p></li> |
| <li><p><strong>monitor_all</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, monitor both input and output, otherwise monitor output only.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.reset_ctx"> |
| <code class="sig-name descname">reset_ctx</code><span class="sig-paren">(</span><em class="sig-param">ctx</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.reset_ctx" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This function has been deprecated. Please refer to <code class="docutils literal notranslate"><span class="pre">HybridBlock.reset_device</span></code>.</p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.reset_device"> |
| <code class="sig-name descname">reset_device</code><span class="sig-paren">(</span><em class="sig-param">device</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.reset_device" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Re-assign all Parameters to other devices. If the Block is hybridized, it will reset the _cached_op_args.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>device</strong> (Device or list of Device, default <code class="xref py py-meth docutils literal notranslate"><span class="pre">device.current_device()</span></code>.) – Assign Parameter to given device. If device is a list of Device, a |
| copy will be made for each device.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.save"> |
| <code class="sig-name descname">save</code><span class="sig-paren">(</span><em class="sig-param">prefix</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.save" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save the model architecture and parameters to load again later</p> |
| <p>Saves the model architecture as a nested dictionary where each Block |
| in the model is a dictionary and its children are sub-dictionaries.</p> |
| <p>Each Block is uniquely identified by Block class name and a unique ID. |
| We save each Block’s parameter UUID to restore later in order to match |
| the saved parameters.</p> |
| <p>Recursively traverses a Block’s children in order (since its an |
| OrderedDict) and uses the unique ID to denote that specific Block.</p> |
| <p>Assumes that the model is created in an identical order every time. |
| If the model is not able to be recreated deterministically do not |
| use this set of APIs to save/load your model.</p> |
| <p>For HybridBlocks, the cached_graph is saved (Symbol & inputs) if |
| it has already been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>prefix</strong> (<em>str</em>) – The prefix to use in filenames for saving this model: |
| <prefix>-model.json and <prefix>-model.params</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.save_parameters"> |
| <code class="sig-name descname">save_parameters</code><span class="sig-paren">(</span><em class="sig-param">filename</em>, <em class="sig-param">deduplicate=False</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.save_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Save parameters to file.</p> |
| <p>Saved parameters can only be loaded with <cite>load_parameters</cite>. Note that this |
| method only saves parameters, not model structure. If you want to save |
| model structures, please use <a class="reference internal" href="#mxnet.gluon.nn.HybridBlock.export" title="mxnet.gluon.nn.HybridBlock.export"><code class="xref py py-meth docutils literal notranslate"><span class="pre">HybridBlock.export()</span></code></a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>filename</strong> (<em>str</em>) – Path to file.</p></li> |
| <li><p><strong>deduplicate</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, save shared parameters only once. Otherwise, if a Block |
| contains multiple sub-blocks that share parameters, each of the |
| shared parameters will be separately saved for every sub-block.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://mxnet.apache.org/api/python/docs/tutorials/packages/gluon/blocks/save_load_params.html">Saving and Loading Gluon Models</a></p> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.setattr"> |
| <code class="sig-name descname">setattr</code><span class="sig-paren">(</span><em class="sig-param">name</em>, <em class="sig-param">value</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.setattr" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set an attribute to a new value for all Parameters.</p> |
| <p>For example, set grad_req to null if you don’t need gradient w.r.t a |
| model’s Parameters:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'grad_req'</span><span class="p">,</span> <span class="s1">'null'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>or change the learning rate multiplier:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">model</span><span class="o">.</span><span class="n">setattr</span><span class="p">(</span><span class="s1">'lr_mult'</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>name</strong> (<em>str</em>) – Name of the attribute.</p></li> |
| <li><p><strong>value</strong> (<em>valid type for attribute name</em>) – The new value for the attribute.</p></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.share_parameters"> |
| <code class="sig-name descname">share_parameters</code><span class="sig-paren">(</span><em class="sig-param">shared</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.share_parameters" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Share parameters recursively inside the model.</p> |
| <p>For example, if you want <code class="docutils literal notranslate"><span class="pre">dense1</span></code> to share <code class="docutils literal notranslate"><span class="pre">dense0</span></code>’s weights, you can do:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense0</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Dense</span><span class="p">(</span><span class="mi">20</span><span class="p">)</span> |
| <span class="n">dense1</span><span class="o">.</span><span class="n">share_parameters</span><span class="p">(</span><span class="n">dense0</span><span class="o">.</span><span class="n">collect_params</span><span class="p">())</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>which equals to</dt><dd><p>dense1.weight = dense0.weight |
| dense1.bias = dense0.bias</p> |
| </dd> |
| </dl> |
| <p>Note that unlike the <cite>load_parameters</cite> or <cite>load_dict</cite> functions, |
| <cite>share_parameters</cite> results in the <cite>Parameter</cite> object being shared (or |
| tied) between the models, whereas <cite>load_parameters</cite> or <cite>load_dict</cite> only |
| set the value of the data dictionary of a model. If you call |
| <cite>load_parameters</cite> or <cite>load_dict</cite> after <cite>share_parameters</cite>, the loaded |
| value will be reflected in all networks that use the shared (or tied) |
| <cite>Parameter</cite> object.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>shared</strong> (<em>Dict</em>) – Dict of the shared parameters.</p> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p></p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p>this block</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.summary"> |
| <code class="sig-name descname">summary</code><span class="sig-paren">(</span><em class="sig-param">*inputs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.summary" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Print the summary of the model’s output and parameters.</p> |
| <p>The network must have been initialized, and must not have been hybridized.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><p><strong>inputs</strong> (<em>object</em>) – Any input that the model supports. For any tensor in the input, only |
| <a class="reference internal" href="../../legacy/ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><code class="xref py py-class docutils literal notranslate"><span class="pre">mxnet.ndarray.NDArray</span></code></a> is supported.</p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="method"> |
| <dt id="mxnet.gluon.nn.SyncBatchNorm.zero_grad"> |
| <code class="sig-name descname">zero_grad</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.gluon.nn.SyncBatchNorm.zero_grad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all Parameters’ gradient buffer to 0.</p> |
| </dd></dl> |
| |
| </dd></dl> |
| |
| </div> |
| </div> |
| |
| |
| <hr class="feedback-hr-top" /> |
| <div class="feedback-container"> |
| <div class="feedback-question">Did this page help you?</div> |
| <div class="feedback-answer-container"> |
| <div class="feedback-answer yes-link" data-response="yes">Yes</div> |
| <div class="feedback-answer no-link" data-response="no">No</div> |
| </div> |
| <div class="feedback-thank-you">Thanks for your feedback!</div> |
| </div> |
| <hr class="feedback-hr-bottom" /> |
| </div> |
| <div class="side-doc-outline"> |
| <div class="side-doc-outline--content"> |
| <div class="localtoc"> |
| <p class="caption"> |
| <span class="caption-text">Table Of Contents</span> |
| </p> |
| <ul> |
| <li><a class="reference internal" href="#">gluon.nn</a><ul> |
| <li><a class="reference internal" href="#sequential-containers">Sequential Containers</a></li> |
| <li><a class="reference internal" href="#concatenation-containers">Concatenation Containers</a></li> |
| <li><a class="reference internal" href="#basic-layers">Basic Layers</a></li> |
| <li><a class="reference internal" href="#convolutional-layers">Convolutional Layers</a></li> |
| <li><a class="reference internal" href="#pixel-shuffle-layers">Pixel Shuffle Layers</a></li> |
| <li><a class="reference internal" href="#pooling-layers">Pooling Layers</a></li> |
| <li><a class="reference internal" href="#normalization-layers">Normalization Layers</a></li> |
| <li><a class="reference internal" href="#embedding-layers">Embedding Layers</a></li> |
| <li><a class="reference internal" href="#advanced-activation-layers">Advanced Activation Layers</a></li> |
| <li><a class="reference internal" href="#module-mxnet.gluon.nn">API Reference</a></li> |
| </ul> |
| </li> |
| </ul> |
| |
| </div> |
| </div> |
| </div> |
| |
| <div class="clearer"></div> |
| </div><div class="pagenation"> |
| <a id="button-prev" href="../model_zoo/index.html" class="mdl-button mdl-js-button mdl-js-ripple-effect mdl-button--colored" role="botton" accesskey="P"> |
| <i class="pagenation-arrow-L fas fa-arrow-left fa-lg"></i> |
| <div class="pagenation-text"> |
| <span class="pagenation-direction">Previous</span> |
| <div>gluon.model_zoo.vision</div> |
| </div> |
| </a> |
| <a id="button-next" href="../rnn/index.html" class="mdl-button mdl-js-button mdl-js-ripple-effect mdl-button--colored" role="botton" accesskey="N"> |
| <i class="pagenation-arrow-R fas fa-arrow-right fa-lg"></i> |
| <div class="pagenation-text"> |
| <span class="pagenation-direction">Next</span> |
| <div>gluon.rnn</div> |
| </div> |
| </a> |
| </div> |
| <footer class="site-footer h-card"> |
| <div class="wrapper"> |
| <div class="row"> |
| <div class="col-4"> |
| <h4 class="footer-category-title">Resources</h4> |
| <ul class="contact-list"> |
| <li><a href="https://lists.apache.org/list.html?dev@mxnet.apache.org">Mailing list</a> <a class="u-email" href="mailto:dev-subscribe@mxnet.apache.org">(subscribe)</a></li> |
| <li><a href="https://discuss.mxnet.io">MXNet Discuss forum</a></li> |
| <li><a href="https://github.com/apache/mxnet/issues">Github Issues</a></li> |
| <li><a href="https://github.com/apache/mxnet/projects">Projects</a></li> |
| <li><a href="https://cwiki.apache.org/confluence/display/MXNET/Apache+MXNet+Home">Developer Wiki</a></li> |
| <li><a href="/community">Contribute To MXNet</a></li> |
| </ul> |
| </div> |
| |
| <div class="col-4"><ul class="social-media-list"><li><a href="https://github.com/apache/mxnet"><svg class="svg-icon"><use xlink:href="../../../_static/minima-social-icons.svg#github"></use></svg> <span class="username">apache/mxnet</span></a></li><li><a href="https://www.twitter.com/apachemxnet"><svg class="svg-icon"><use xlink:href="../../../_static/minima-social-icons.svg#twitter"></use></svg> <span class="username">apachemxnet</span></a></li><li><a href="https://youtube.com/apachemxnet"><svg class="svg-icon"><use xlink:href="../../../_static/minima-social-icons.svg#youtube"></use></svg> <span class="username">apachemxnet</span></a></li></ul> |
| </div> |
| |
| <div class="col-4 footer-text"> |
| <p>A flexible and efficient library for deep learning.</p> |
| </div> |
| </div> |
| </div> |
| </footer> |
| |
| <footer class="site-footer2"> |
| <div class="wrapper"> |
| <div class="row"> |
| <div class="col-3"> |
| <img src="../../../_static/apache_incubator_logo.png" class="footer-logo col-2"> |
| </div> |
| <div class="footer-bottom-warning col-9"> |
| <p>Apache MXNet is an effort undergoing incubation at <a href="http://www.apache.org/">The Apache Software Foundation</a> (ASF), <span style="font-weight:bold">sponsored by the <i>Apache Incubator</i></span>. Incubation is required |
| of all newly accepted projects until a further review indicates that the infrastructure, |
| communications, and decision making process have stabilized in a manner consistent with other |
| successful ASF projects. While incubation status is not necessarily a reflection of the completeness |
| or stability of the code, it does indicate that the project has yet to be fully endorsed by the ASF. |
| </p><p>"Copyright © 2017-2018, The Apache Software Foundation Apache MXNet, MXNet, Apache, the Apache |
| feather, and the Apache MXNet project logo are either registered trademarks or trademarks of the |
| Apache Software Foundation."</p> |
| </div> |
| </div> |
| </div> |
| </footer> |
| |
| </body> |
| </html> |
