| <!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; |
| } |
| |
| .dropdown-caret-path { |
| fill: #FFFFFF; |
| } |
| </style> |
| |
| <title>symbol.op — 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> |
| <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="symbol.random" href="../random/index.html" /> |
| <link rel="prev" title="symbol.linalg" href="../linalg/index.html" /> |
| </head> |
| <body><header class="site-header" role="banner"> |
| <div class="wrapper"> |
| <a class="site-title" rel="author" href="/versions/1.9.1/"><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="/versions/1.9.1/get_started">Get Started</a> |
| <a class="page-link" href="/versions/1.9.1/features">Features</a> |
| <a class="page-link" href="/versions/1.9.1/ecosystem">Ecosystem</a> |
| <a class="page-link page-current" href="/versions/1.9.1/api">Docs & Tutorials</a> |
| <a class="page-link" href="/versions/1.9.1/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">1.9.1 |
| <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" href="/">master</a><br> |
| <a class="dropdown-option-active" 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.symbol</a><i class="material-icons">navigate_next</i> |
| <a class="mdl-navigation__link is-active">symbol.op</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-source" |
| class="mdl-button mdl-js-button mdl-button--icon" |
| href="../../../_sources/api/symbol/op/index.rst" rel="nofollow"> |
| <i class="material-icons">code</i> |
| </a> |
| <div class="mdl-tooltip" data-mdl-for="button-show-source"> |
| Show Source |
| </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/1-ndarray.html">Manipulate data with <code class="docutils literal notranslate"><span class="pre">ndarray</span></code></a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/2-nn.html">Create a neural network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/3-autograd.html">Automatic differentiation with <code class="docutils literal notranslate"><span class="pre">autograd</span></code></a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/4-train.html">Train the neural network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-predict.html">Predict with a pre-trained model</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/6-use_gpus.html">Use GPUs</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/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/custom_layer_beginners.html">Customer Layers (Beginners)</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/data_augmentation.html#Spatial-Augmentation">Spatial Augmentation</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/data_augmentation.html#Color-Augmentation">Color Augmentation</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/data_augmentation.html#Composed-Augmentations">Composed Augmentations</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/image-augmentation.html">Image Augmentation</a></li> |
| <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> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/image/pretrained_models.html">Using pre-trained models in MXNet</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/ndarray/index.html">NDArray</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/01-ndarray-intro.html">An Intro: Manipulate Data the MXNet Way with NDArray</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/02-ndarray-operations.html">NDArray Operations</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/03-ndarray-contexts.html">NDArray Contexts</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/gotchas_numpy_in_mxnet.html">Gotchas using NumPy in Apache MXNet</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/index.html">Tutorials</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/csr.html">CSRNDArray - NDArray in Compressed Sparse Row Storage Format</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/row_sparse.html">RowSparseNDArray - NDArray for Sparse Gradient Updates</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/train.html">Train a Linear Regression Model with Sparse Symbols</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/train_gluon.html">Sparse NDArrays with Gluon</a></li> |
| </ul> |
| </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 internal" href="../../../tutorials/packages/onnx/super_resolution.html">Importing an ONNX model into MXNet</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/mkldnn/index.html">Intel MKL-DNN</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/mkldnn/mkldnn_quantization.html">Quantize with MKL-DNN backend</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/mkldnn/mkldnn_quantization.html#Improving-accuracy-with-Intel®-Neural-Compressor">Improving accuracy with Intel® Neural Compressor</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/mkldnn/mkldnn_readme.html">Install MXNet with MKL-DNN</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/tensorrt/index.html">TensorRT</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/tensorrt/tensorrt.html">Optimizing Deep Learning Computation Graphs with TensorRT</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> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/inference/scala.html">Deploy into a Java or Scala Environment</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/inference/wine_detector.html">Real-time Object Detection with MXNet On The Raspberry Pi</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/custom_layer.html">Custom Layers</a></li> |
| <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> |
| </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="../../ndarray/index.html">mxnet.ndarray</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/ndarray.html">ndarray</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/contrib/index.html">ndarray.contrib</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/image/index.html">ndarray.image</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/linalg/index.html">ndarray.linalg</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/op/index.html">ndarray.op</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/random/index.html">ndarray.random</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/register/index.html">ndarray.register</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/sparse/index.html">ndarray.sparse</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/utils/index.html">ndarray.utils</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../gluon/index.html">mxnet.gluon</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/block.html">gluon.Block</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/hybrid_block.html">gluon.HybridBlock</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/symbol_block.html">gluon.SymbolBlock</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/constant.html">gluon.Constant</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/parameter.html">gluon.Parameter</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/parameter_dict.html">gluon.ParameterDict</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/trainer.html">gluon.Trainer</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/contrib/index.html">gluon.contrib</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/data/index.html">gluon.data</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../gluon/data/vision/index.html">data.vision</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../gluon/data/vision/datasets/index.html">vision.datasets</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../gluon/data/vision/transforms/index.html">vision.transforms</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/loss/index.html">gluon.loss</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/model_zoo/index.html">gluon.model_zoo.vision</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/nn/index.html">gluon.nn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/rnn/index.html">gluon.rnn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/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="../../metric/index.html">mxnet.metric</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html">mxnet.kvstore</a></li> |
| <li class="toctree-l2 current"><a class="reference internal" href="../index.html">mxnet.symbol</a><ul class="current"> |
| <li class="toctree-l3"><a class="reference internal" href="../symbol.html">symbol</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../contrib/index.html">symbol.contrib</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../image/index.html">symbol.image</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../linalg/index.html">symbol.linalg</a></li> |
| <li class="toctree-l3 current"><a class="current reference internal" href="#">symbol.op</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../random/index.html">symbol.random</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../register/index.html">symbol.register</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../sparse/index.html">symbol.sparse</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../module/index.html">mxnet.module</a></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/autograd/index.html">contrib.autograd</a></li> |
| <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="../../mxnet/index.html">mxnet</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/attribute/index.html">mxnet.attribute</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/base/index.html">mxnet.base</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/callback/index.html">mxnet.callback</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/context/index.html">mxnet.context</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/engine/index.html">mxnet.engine</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/executor/index.html">mxnet.executor</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/executor_manager/index.html">mxnet.executor_manager</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/image/index.html">mxnet.image</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/io/index.html">mxnet.io</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/kvstore_server/index.html">mxnet.kvstore_server</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/libinfo/index.html">mxnet.libinfo</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/log/index.html">mxnet.log</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/model/index.html">mxnet.model</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/monitor/index.html">mxnet.monitor</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/name/index.html">mxnet.name</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/notebook/index.html">mxnet.notebook</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/operator/index.html">mxnet.operator</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/profiler/index.html">mxnet.profiler</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/random/index.html">mxnet.random</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/recordio/index.html">mxnet.recordio</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/registry/index.html">mxnet.registry</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/rtc/index.html">mxnet.rtc</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/runtime/index.html">mxnet.runtime</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/test_utils/index.html">mxnet.test_utils</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/torch/index.html">mxnet.torch</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/util/index.html">mxnet.util</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/visualization/index.html">mxnet.visualization</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| </ul> |
| |
| </nav> |
| |
| </div> |
| |
| </header> |
| <main class="mdl-layout__content" tabIndex="0"> |
| |
| <script type="text/javascript" src="../../../_static/sphinx_materialdesign_theme.js "></script> |
| <script type="text/javascript" src="../../../_static/feedback.js"></script> |
| <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/1-ndarray.html">Manipulate data with <code class="docutils literal notranslate"><span class="pre">ndarray</span></code></a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/2-nn.html">Create a neural network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/3-autograd.html">Automatic differentiation with <code class="docutils literal notranslate"><span class="pre">autograd</span></code></a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/4-train.html">Train the neural network</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/5-predict.html">Predict with a pre-trained model</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/getting-started/crash-course/6-use_gpus.html">Use GPUs</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/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/custom_layer_beginners.html">Customer Layers (Beginners)</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/data_augmentation.html#Spatial-Augmentation">Spatial Augmentation</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/data_augmentation.html#Color-Augmentation">Color Augmentation</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/data/data_augmentation.html#Composed-Augmentations">Composed Augmentations</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/image-augmentation.html">Image Augmentation</a></li> |
| <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> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/gluon/image/pretrained_models.html">Using pre-trained models in MXNet</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/ndarray/index.html">NDArray</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/01-ndarray-intro.html">An Intro: Manipulate Data the MXNet Way with NDArray</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/02-ndarray-operations.html">NDArray Operations</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/03-ndarray-contexts.html">NDArray Contexts</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/gotchas_numpy_in_mxnet.html">Gotchas using NumPy in Apache MXNet</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/index.html">Tutorials</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/csr.html">CSRNDArray - NDArray in Compressed Sparse Row Storage Format</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/row_sparse.html">RowSparseNDArray - NDArray for Sparse Gradient Updates</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/train.html">Train a Linear Regression Model with Sparse Symbols</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/packages/ndarray/sparse/train_gluon.html">Sparse NDArrays with Gluon</a></li> |
| </ul> |
| </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 internal" href="../../../tutorials/packages/onnx/super_resolution.html">Importing an ONNX model into MXNet</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/mkldnn/index.html">Intel MKL-DNN</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/mkldnn/mkldnn_quantization.html">Quantize with MKL-DNN backend</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/mkldnn/mkldnn_quantization.html#Improving-accuracy-with-Intel®-Neural-Compressor">Improving accuracy with Intel® Neural Compressor</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/mkldnn/mkldnn_readme.html">Install MXNet with MKL-DNN</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/performance/backend/tensorrt/index.html">TensorRT</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../../tutorials/performance/backend/tensorrt/tensorrt.html">Optimizing Deep Learning Computation Graphs with TensorRT</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> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/inference/scala.html">Deploy into a Java or Scala Environment</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="../../../tutorials/deploy/inference/wine_detector.html">Real-time Object Detection with MXNet On The Raspberry Pi</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/custom_layer.html">Custom Layers</a></li> |
| <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> |
| </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="../../ndarray/index.html">mxnet.ndarray</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/ndarray.html">ndarray</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/contrib/index.html">ndarray.contrib</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/image/index.html">ndarray.image</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/linalg/index.html">ndarray.linalg</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/op/index.html">ndarray.op</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/random/index.html">ndarray.random</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/register/index.html">ndarray.register</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/sparse/index.html">ndarray.sparse</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../ndarray/utils/index.html">ndarray.utils</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../gluon/index.html">mxnet.gluon</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/block.html">gluon.Block</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/hybrid_block.html">gluon.HybridBlock</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/symbol_block.html">gluon.SymbolBlock</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/constant.html">gluon.Constant</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/parameter.html">gluon.Parameter</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/parameter_dict.html">gluon.ParameterDict</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/trainer.html">gluon.Trainer</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/contrib/index.html">gluon.contrib</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/data/index.html">gluon.data</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="../../gluon/data/vision/index.html">data.vision</a><ul> |
| <li class="toctree-l5"><a class="reference internal" href="../../gluon/data/vision/datasets/index.html">vision.datasets</a></li> |
| <li class="toctree-l5"><a class="reference internal" href="../../gluon/data/vision/transforms/index.html">vision.transforms</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/loss/index.html">gluon.loss</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/model_zoo/index.html">gluon.model_zoo.vision</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/nn/index.html">gluon.nn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/rnn/index.html">gluon.rnn</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../gluon/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="../../metric/index.html">mxnet.metric</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../kvstore/index.html">mxnet.kvstore</a></li> |
| <li class="toctree-l2 current"><a class="reference internal" href="../index.html">mxnet.symbol</a><ul class="current"> |
| <li class="toctree-l3"><a class="reference internal" href="../symbol.html">symbol</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../contrib/index.html">symbol.contrib</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../image/index.html">symbol.image</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../linalg/index.html">symbol.linalg</a></li> |
| <li class="toctree-l3 current"><a class="current reference internal" href="#">symbol.op</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../random/index.html">symbol.random</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../register/index.html">symbol.register</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../sparse/index.html">symbol.sparse</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../module/index.html">mxnet.module</a></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/autograd/index.html">contrib.autograd</a></li> |
| <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="../../mxnet/index.html">mxnet</a><ul> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/attribute/index.html">mxnet.attribute</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/base/index.html">mxnet.base</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/callback/index.html">mxnet.callback</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/context/index.html">mxnet.context</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/engine/index.html">mxnet.engine</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/executor/index.html">mxnet.executor</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/executor_manager/index.html">mxnet.executor_manager</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/image/index.html">mxnet.image</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/io/index.html">mxnet.io</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/kvstore_server/index.html">mxnet.kvstore_server</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/libinfo/index.html">mxnet.libinfo</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/log/index.html">mxnet.log</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/model/index.html">mxnet.model</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/monitor/index.html">mxnet.monitor</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/name/index.html">mxnet.name</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/notebook/index.html">mxnet.notebook</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/operator/index.html">mxnet.operator</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/profiler/index.html">mxnet.profiler</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/random/index.html">mxnet.random</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/recordio/index.html">mxnet.recordio</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/registry/index.html">mxnet.registry</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/rtc/index.html">mxnet.rtc</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/runtime/index.html">mxnet.runtime</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/test_utils/index.html">mxnet.test_utils</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/torch/index.html">mxnet.torch</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/util/index.html">mxnet.util</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../../mxnet/visualization/index.html">mxnet.visualization</a></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| </ul> |
| |
| </nav> |
| |
| </div> |
| |
| </header> |
| |
| <div class="document"> |
| <div class="page-content" role="main"> |
| |
| <div class="section" id="module-mxnet.symbol.op"> |
| <span id="symbol-op"></span><h1>symbol.op<a class="headerlink" href="#module-mxnet.symbol.op" title="Permalink to this headline">¶</a></h1> |
| <p>Backend ops in mxnet.symbol namespace.</p> |
| <p><strong>Functions</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.symbol.op.Activation" title="mxnet.symbol.op.Activation"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Activation</span></code></a>([data, act_type, name, attr, out])</p></td> |
| <td><p>Applies an activation function element-wise to the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.BatchNorm" title="mxnet.symbol.op.BatchNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">BatchNorm</span></code></a>([data, gamma, beta, moving_mean, …])</p></td> |
| <td><p>Batch normalization.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.BatchNorm_v1" title="mxnet.symbol.op.BatchNorm_v1"><code class="xref py py-obj docutils literal notranslate"><span class="pre">BatchNorm_v1</span></code></a>([data, gamma, beta, eps, …])</p></td> |
| <td><p>Batch normalization.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.BilinearSampler" title="mxnet.symbol.op.BilinearSampler"><code class="xref py py-obj docutils literal notranslate"><span class="pre">BilinearSampler</span></code></a>([data, grid, cudnn_off, …])</p></td> |
| <td><p>Applies bilinear sampling to input feature map.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.BlockGrad" title="mxnet.symbol.op.BlockGrad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">BlockGrad</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Stops gradient computation.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.CTCLoss" title="mxnet.symbol.op.CTCLoss"><code class="xref py py-obj docutils literal notranslate"><span class="pre">CTCLoss</span></code></a>([data, label, data_lengths, …])</p></td> |
| <td><p>Connectionist Temporal Classification Loss.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.Cast" title="mxnet.symbol.op.Cast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Cast</span></code></a>([data, dtype, name, attr, out])</p></td> |
| <td><p>Casts all elements of the input to a new type.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.Concat" title="mxnet.symbol.op.Concat"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Concat</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Joins input arrays along a given axis.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.Convolution" title="mxnet.symbol.op.Convolution"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Convolution</span></code></a>([data, weight, bias, kernel, …])</p></td> |
| <td><p>Compute <em>N</em>-D convolution on <em>(N+2)</em>-D input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.Convolution_v1" title="mxnet.symbol.op.Convolution_v1"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Convolution_v1</span></code></a>([data, weight, bias, kernel, …])</p></td> |
| <td><p>This operator is DEPRECATED.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.Correlation" title="mxnet.symbol.op.Correlation"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Correlation</span></code></a>([data1, data2, kernel_size, …])</p></td> |
| <td><p>Applies correlation to inputs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.Crop" title="mxnet.symbol.op.Crop"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Crop</span></code></a>(*data, **kwargs)</p></td> |
| <td><p><div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><cite>Crop</cite> is deprecated. Use <cite>slice</cite> instead.</p> |
| </div> |
| </p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.Custom" title="mxnet.symbol.op.Custom"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Custom</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Apply a custom operator implemented in a frontend language (like Python).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.Deconvolution" title="mxnet.symbol.op.Deconvolution"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Deconvolution</span></code></a>([data, weight, bias, kernel, …])</p></td> |
| <td><p>Computes 1D or 2D transposed convolution (aka fractionally strided convolution) of the input tensor.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.Dropout" title="mxnet.symbol.op.Dropout"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Dropout</span></code></a>([data, p, mode, axes, cudnn_off, …])</p></td> |
| <td><p>Applies dropout operation to input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.ElementWiseSum" title="mxnet.symbol.op.ElementWiseSum"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ElementWiseSum</span></code></a>(*args, **kwargs)</p></td> |
| <td><p>Adds all input arguments element-wise.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.Embedding" title="mxnet.symbol.op.Embedding"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Embedding</span></code></a>([data, weight, input_dim, …])</p></td> |
| <td><p>Maps integer indices to vector representations (embeddings).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.Flatten" title="mxnet.symbol.op.Flatten"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Flatten</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Flattens the input array into a 2-D array by collapsing the higher dimensions.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.FullyConnected" title="mxnet.symbol.op.FullyConnected"><code class="xref py py-obj docutils literal notranslate"><span class="pre">FullyConnected</span></code></a>([data, weight, bias, …])</p></td> |
| <td><p>Applies a linear transformation: <span class="math notranslate nohighlight">\(Y = XW^T + b\)</span>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.GridGenerator" title="mxnet.symbol.op.GridGenerator"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GridGenerator</span></code></a>([data, transform_type, …])</p></td> |
| <td><p>Generates 2D sampling grid for bilinear sampling.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.GroupNorm" title="mxnet.symbol.op.GroupNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">GroupNorm</span></code></a>([data, gamma, beta, num_groups, …])</p></td> |
| <td><p>Group normalization.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.IdentityAttachKLSparseReg" title="mxnet.symbol.op.IdentityAttachKLSparseReg"><code class="xref py py-obj docutils literal notranslate"><span class="pre">IdentityAttachKLSparseReg</span></code></a>([data, …])</p></td> |
| <td><p>Apply a sparse regularization to the output a sigmoid activation function.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.InstanceNorm" title="mxnet.symbol.op.InstanceNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">InstanceNorm</span></code></a>([data, gamma, beta, eps, name, …])</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.symbol.op.L2Normalization" title="mxnet.symbol.op.L2Normalization"><code class="xref py py-obj docutils literal notranslate"><span class="pre">L2Normalization</span></code></a>([data, eps, mode, name, …])</p></td> |
| <td><p>Normalize the input array using the L2 norm.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.LRN" title="mxnet.symbol.op.LRN"><code class="xref py py-obj docutils literal notranslate"><span class="pre">LRN</span></code></a>([data, alpha, beta, knorm, nsize, name, …])</p></td> |
| <td><p>Applies local response normalization to the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.LayerNorm" title="mxnet.symbol.op.LayerNorm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">LayerNorm</span></code></a>([data, gamma, beta, axis, eps, …])</p></td> |
| <td><p>Layer normalization.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.LeakyReLU" title="mxnet.symbol.op.LeakyReLU"><code class="xref py py-obj docutils literal notranslate"><span class="pre">LeakyReLU</span></code></a>([data, gamma, act_type, slope, …])</p></td> |
| <td><p>Applies Leaky rectified linear unit activation element-wise to the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.LinearRegressionOutput" title="mxnet.symbol.op.LinearRegressionOutput"><code class="xref py py-obj docutils literal notranslate"><span class="pre">LinearRegressionOutput</span></code></a>([data, label, …])</p></td> |
| <td><p>Computes and optimizes for squared loss during backward propagation.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.LogisticRegressionOutput" title="mxnet.symbol.op.LogisticRegressionOutput"><code class="xref py py-obj docutils literal notranslate"><span class="pre">LogisticRegressionOutput</span></code></a>([data, label, …])</p></td> |
| <td><p>Applies a logistic function to the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.MAERegressionOutput" title="mxnet.symbol.op.MAERegressionOutput"><code class="xref py py-obj docutils literal notranslate"><span class="pre">MAERegressionOutput</span></code></a>([data, label, …])</p></td> |
| <td><p>Computes mean absolute error of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.MakeLoss" title="mxnet.symbol.op.MakeLoss"><code class="xref py py-obj docutils literal notranslate"><span class="pre">MakeLoss</span></code></a>([data, grad_scale, valid_thresh, …])</p></td> |
| <td><p>Make your own loss function in network construction.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.Pad" title="mxnet.symbol.op.Pad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Pad</span></code></a>([data, mode, pad_width, constant_value, …])</p></td> |
| <td><p>Pads an input array with a constant or edge values of the array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.Pooling" title="mxnet.symbol.op.Pooling"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Pooling</span></code></a>([data, kernel, pool_type, …])</p></td> |
| <td><p>Performs pooling on the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.Pooling_v1" title="mxnet.symbol.op.Pooling_v1"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Pooling_v1</span></code></a>([data, kernel, pool_type, …])</p></td> |
| <td><p>This operator is DEPRECATED.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.RNN" title="mxnet.symbol.op.RNN"><code class="xref py py-obj docutils literal notranslate"><span class="pre">RNN</span></code></a>([data, parameters, state, state_cell, …])</p></td> |
| <td><p>Applies recurrent layers to input data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.ROIPooling" title="mxnet.symbol.op.ROIPooling"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ROIPooling</span></code></a>([data, rois, pooled_size, …])</p></td> |
| <td><p>Performs region of interest(ROI) pooling on the input array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.Reshape" title="mxnet.symbol.op.Reshape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Reshape</span></code></a>([data, shape, reverse, …])</p></td> |
| <td><p>Reshapes the input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.SVMOutput" title="mxnet.symbol.op.SVMOutput"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SVMOutput</span></code></a>([data, label, margin, …])</p></td> |
| <td><p>Computes support vector machine based transformation of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.SequenceLast" title="mxnet.symbol.op.SequenceLast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SequenceLast</span></code></a>([data, sequence_length, …])</p></td> |
| <td><p>Takes the last element of a sequence.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.SequenceMask" title="mxnet.symbol.op.SequenceMask"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SequenceMask</span></code></a>([data, sequence_length, …])</p></td> |
| <td><p>Sets all elements outside the sequence to a constant value.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.SequenceReverse" title="mxnet.symbol.op.SequenceReverse"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SequenceReverse</span></code></a>([data, sequence_length, …])</p></td> |
| <td><p>Reverses the elements of each sequence.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.SliceChannel" title="mxnet.symbol.op.SliceChannel"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SliceChannel</span></code></a>([data, num_outputs, axis, …])</p></td> |
| <td><p>Splits an array along a particular axis into multiple sub-arrays.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.Softmax" title="mxnet.symbol.op.Softmax"><code class="xref py py-obj docutils literal notranslate"><span class="pre">Softmax</span></code></a>([data, label, grad_scale, …])</p></td> |
| <td><p>Computes the gradient of cross entropy loss with respect to softmax output.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.SoftmaxActivation" title="mxnet.symbol.op.SoftmaxActivation"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SoftmaxActivation</span></code></a>([data, mode, name, attr, out])</p></td> |
| <td><p>Applies softmax activation to input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.SoftmaxOutput" title="mxnet.symbol.op.SoftmaxOutput"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SoftmaxOutput</span></code></a>([data, label, grad_scale, …])</p></td> |
| <td><p>Computes the gradient of cross entropy loss with respect to softmax output.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.SpatialTransformer" title="mxnet.symbol.op.SpatialTransformer"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SpatialTransformer</span></code></a>([data, loc, …])</p></td> |
| <td><p>Applies a spatial transformer to input feature map.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.SwapAxis" title="mxnet.symbol.op.SwapAxis"><code class="xref py py-obj docutils literal notranslate"><span class="pre">SwapAxis</span></code></a>([data, dim1, dim2, name, attr, out])</p></td> |
| <td><p>Interchanges two axes of an array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.UpSampling" title="mxnet.symbol.op.UpSampling"><code class="xref py py-obj docutils literal notranslate"><span class="pre">UpSampling</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Upsamples the given input data.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.abs" title="mxnet.symbol.op.abs"><code class="xref py py-obj docutils literal notranslate"><span class="pre">abs</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise absolute value of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.adam_update" title="mxnet.symbol.op.adam_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">adam_update</span></code></a>([weight, grad, mean, var, lr, …])</p></td> |
| <td><p>Update function for Adam optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.add_n" title="mxnet.symbol.op.add_n"><code class="xref py py-obj docutils literal notranslate"><span class="pre">add_n</span></code></a>(*args, **kwargs)</p></td> |
| <td><p>Adds all input arguments element-wise.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.all_finite" title="mxnet.symbol.op.all_finite"><code class="xref py py-obj docutils literal notranslate"><span class="pre">all_finite</span></code></a>([data, init_output, name, attr, out])</p></td> |
| <td><p>Check if all the float numbers in the array are finite (used for AMP)</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.amp_cast" title="mxnet.symbol.op.amp_cast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">amp_cast</span></code></a>([data, dtype, name, attr, out])</p></td> |
| <td><p>Cast function between low precision float/FP32 used by AMP.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.amp_multicast" title="mxnet.symbol.op.amp_multicast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">amp_multicast</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Cast function used by AMP, that casts its inputs to the common widest type.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.arccos" title="mxnet.symbol.op.arccos"><code class="xref py py-obj docutils literal notranslate"><span class="pre">arccos</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise inverse cosine of the input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.arccosh" title="mxnet.symbol.op.arccosh"><code class="xref py py-obj docutils literal notranslate"><span class="pre">arccosh</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns the element-wise inverse hyperbolic cosine of the input array, computed element-wise.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.arcsin" title="mxnet.symbol.op.arcsin"><code class="xref py py-obj docutils literal notranslate"><span class="pre">arcsin</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise inverse sine of the input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.arcsinh" title="mxnet.symbol.op.arcsinh"><code class="xref py py-obj docutils literal notranslate"><span class="pre">arcsinh</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns the element-wise inverse hyperbolic sine of the input array, computed element-wise.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.arctan" title="mxnet.symbol.op.arctan"><code class="xref py py-obj docutils literal notranslate"><span class="pre">arctan</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise inverse tangent of the input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.arctanh" title="mxnet.symbol.op.arctanh"><code class="xref py py-obj docutils literal notranslate"><span class="pre">arctanh</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns the element-wise inverse hyperbolic tangent of the input array, computed element-wise.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.argmax" title="mxnet.symbol.op.argmax"><code class="xref py py-obj docutils literal notranslate"><span class="pre">argmax</span></code></a>([data, axis, keepdims, name, attr, out])</p></td> |
| <td><p>Returns indices of the maximum values along an axis.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.argmax_channel" title="mxnet.symbol.op.argmax_channel"><code class="xref py py-obj docutils literal notranslate"><span class="pre">argmax_channel</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns argmax indices of each channel from the input array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.argmin" title="mxnet.symbol.op.argmin"><code class="xref py py-obj docutils literal notranslate"><span class="pre">argmin</span></code></a>([data, axis, keepdims, name, attr, out])</p></td> |
| <td><p>Returns indices of the minimum values along an axis.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.argsort" title="mxnet.symbol.op.argsort"><code class="xref py py-obj docutils literal notranslate"><span class="pre">argsort</span></code></a>([data, axis, is_ascend, dtype, …])</p></td> |
| <td><p>Returns the indices that would sort an input array along the given axis.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.batch_dot" title="mxnet.symbol.op.batch_dot"><code class="xref py py-obj docutils literal notranslate"><span class="pre">batch_dot</span></code></a>([lhs, rhs, transpose_a, …])</p></td> |
| <td><p>Batchwise dot product.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.batch_take" title="mxnet.symbol.op.batch_take"><code class="xref py py-obj docutils literal notranslate"><span class="pre">batch_take</span></code></a>([a, indices, name, attr, out])</p></td> |
| <td><p>Takes elements from a data batch.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_add" title="mxnet.symbol.op.broadcast_add"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_add</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns element-wise sum of the input arrays with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_axes" title="mxnet.symbol.op.broadcast_axes"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_axes</span></code></a>([data, axis, size, name, …])</p></td> |
| <td><p>Broadcasts the input array over particular axes.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_axis" title="mxnet.symbol.op.broadcast_axis"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_axis</span></code></a>([data, axis, size, name, …])</p></td> |
| <td><p>Broadcasts the input array over particular axes.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_div" title="mxnet.symbol.op.broadcast_div"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_div</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns element-wise division of the input arrays with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_equal" title="mxnet.symbol.op.broadcast_equal"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_equal</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns the result of element-wise <strong>equal to</strong> (==) comparison operation with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_greater" title="mxnet.symbol.op.broadcast_greater"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_greater</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns the result of element-wise <strong>greater than</strong> (>) comparison operation with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_greater_equal" title="mxnet.symbol.op.broadcast_greater_equal"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_greater_equal</span></code></a>([lhs, rhs, name, …])</p></td> |
| <td><p>Returns the result of element-wise <strong>greater than or equal to</strong> (>=) comparison operation with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_hypot" title="mxnet.symbol.op.broadcast_hypot"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_hypot</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns the hypotenuse of a right angled triangle, given its “legs” with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_lesser" title="mxnet.symbol.op.broadcast_lesser"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_lesser</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns the result of element-wise <strong>lesser than</strong> (<) comparison operation with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_lesser_equal" title="mxnet.symbol.op.broadcast_lesser_equal"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_lesser_equal</span></code></a>([lhs, rhs, name, …])</p></td> |
| <td><p>Returns the result of element-wise <strong>lesser than or equal to</strong> (<=) comparison operation with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_like" title="mxnet.symbol.op.broadcast_like"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_like</span></code></a>([lhs, rhs, lhs_axes, …])</p></td> |
| <td><p>Broadcasts lhs to have the same shape as rhs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_logical_and" title="mxnet.symbol.op.broadcast_logical_and"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_logical_and</span></code></a>([lhs, rhs, name, …])</p></td> |
| <td><p>Returns the result of element-wise <strong>logical and</strong> with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_logical_or" title="mxnet.symbol.op.broadcast_logical_or"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_logical_or</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns the result of element-wise <strong>logical or</strong> with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_logical_xor" title="mxnet.symbol.op.broadcast_logical_xor"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_logical_xor</span></code></a>([lhs, rhs, name, …])</p></td> |
| <td><p>Returns the result of element-wise <strong>logical xor</strong> with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_maximum" title="mxnet.symbol.op.broadcast_maximum"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_maximum</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns element-wise maximum of the input arrays with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_minimum" title="mxnet.symbol.op.broadcast_minimum"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_minimum</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns element-wise minimum of the input arrays with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_minus" title="mxnet.symbol.op.broadcast_minus"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_minus</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns element-wise difference of the input arrays with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_mod" title="mxnet.symbol.op.broadcast_mod"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_mod</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns element-wise modulo of the input arrays with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_mul" title="mxnet.symbol.op.broadcast_mul"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_mul</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns element-wise product of the input arrays with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_not_equal" title="mxnet.symbol.op.broadcast_not_equal"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_not_equal</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns the result of element-wise <strong>not equal to</strong> (!=) comparison operation with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_plus" title="mxnet.symbol.op.broadcast_plus"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_plus</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns element-wise sum of the input arrays with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_power" title="mxnet.symbol.op.broadcast_power"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_power</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns result of first array elements raised to powers from second array, element-wise with broadcasting.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_sub" title="mxnet.symbol.op.broadcast_sub"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_sub</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Returns element-wise difference of the input arrays with broadcasting.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.broadcast_to" title="mxnet.symbol.op.broadcast_to"><code class="xref py py-obj docutils literal notranslate"><span class="pre">broadcast_to</span></code></a>([data, shape, name, attr, out])</p></td> |
| <td><p>Broadcasts the input array to a new shape.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.cast" title="mxnet.symbol.op.cast"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cast</span></code></a>([data, dtype, name, attr, out])</p></td> |
| <td><p>Casts all elements of the input to a new type.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.cast_storage" title="mxnet.symbol.op.cast_storage"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cast_storage</span></code></a>([data, stype, name, attr, out])</p></td> |
| <td><p>Casts tensor storage type to the new type.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.cbrt" title="mxnet.symbol.op.cbrt"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cbrt</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise cube-root value of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.ceil" title="mxnet.symbol.op.ceil"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ceil</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise ceiling of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.choose_element_0index" title="mxnet.symbol.op.choose_element_0index"><code class="xref py py-obj docutils literal notranslate"><span class="pre">choose_element_0index</span></code></a>([data, index, axis, …])</p></td> |
| <td><p>Picks elements from an input array according to the input indices along the given axis.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.clip" title="mxnet.symbol.op.clip"><code class="xref py py-obj docutils literal notranslate"><span class="pre">clip</span></code></a>([data, a_min, a_max, name, attr, out])</p></td> |
| <td><p>Clips (limits) the values in an array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.col2im" title="mxnet.symbol.op.col2im"><code class="xref py py-obj docutils literal notranslate"><span class="pre">col2im</span></code></a>([data, output_size, kernel, stride, …])</p></td> |
| <td><p>Combining the output column matrix of im2col back to image array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.concat" title="mxnet.symbol.op.concat"><code class="xref py py-obj docutils literal notranslate"><span class="pre">concat</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Joins input arrays along a given axis.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.cos" title="mxnet.symbol.op.cos"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cos</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Computes the element-wise cosine of the input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.cosh" title="mxnet.symbol.op.cosh"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cosh</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns the hyperbolic cosine of the input array, computed element-wise.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.crop" title="mxnet.symbol.op.crop"><code class="xref py py-obj docutils literal notranslate"><span class="pre">crop</span></code></a>([data, begin, end, step, name, attr, out])</p></td> |
| <td><p>Slices a region of the array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.ctc_loss" title="mxnet.symbol.op.ctc_loss"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ctc_loss</span></code></a>([data, label, data_lengths, …])</p></td> |
| <td><p>Connectionist Temporal Classification Loss.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.cumsum" title="mxnet.symbol.op.cumsum"><code class="xref py py-obj docutils literal notranslate"><span class="pre">cumsum</span></code></a>([a, axis, dtype, name, attr, out])</p></td> |
| <td><p>Return the cumulative sum of the elements along a given axis.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.degrees" title="mxnet.symbol.op.degrees"><code class="xref py py-obj docutils literal notranslate"><span class="pre">degrees</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Converts each element of the input array from radians to degrees.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.depth_to_space" title="mxnet.symbol.op.depth_to_space"><code class="xref py py-obj docutils literal notranslate"><span class="pre">depth_to_space</span></code></a>([data, block_size, name, …])</p></td> |
| <td><p>Rearranges(permutes) data from depth into blocks of spatial data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.diag" title="mxnet.symbol.op.diag"><code class="xref py py-obj docutils literal notranslate"><span class="pre">diag</span></code></a>([data, k, axis1, axis2, name, attr, out])</p></td> |
| <td><p>Extracts a diagonal or constructs a diagonal array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.dot" title="mxnet.symbol.op.dot"><code class="xref py py-obj docutils literal notranslate"><span class="pre">dot</span></code></a>([lhs, rhs, transpose_a, transpose_b, …])</p></td> |
| <td><p>Dot product of two arrays.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.elemwise_add" title="mxnet.symbol.op.elemwise_add"><code class="xref py py-obj docutils literal notranslate"><span class="pre">elemwise_add</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Adds arguments element-wise.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.elemwise_div" title="mxnet.symbol.op.elemwise_div"><code class="xref py py-obj docutils literal notranslate"><span class="pre">elemwise_div</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Divides arguments element-wise.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.elemwise_mul" title="mxnet.symbol.op.elemwise_mul"><code class="xref py py-obj docutils literal notranslate"><span class="pre">elemwise_mul</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Multiplies arguments element-wise.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.elemwise_sub" title="mxnet.symbol.op.elemwise_sub"><code class="xref py py-obj docutils literal notranslate"><span class="pre">elemwise_sub</span></code></a>([lhs, rhs, name, attr, out])</p></td> |
| <td><p>Subtracts arguments element-wise.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.erf" title="mxnet.symbol.op.erf"><code class="xref py py-obj docutils literal notranslate"><span class="pre">erf</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise gauss error function of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.erfinv" title="mxnet.symbol.op.erfinv"><code class="xref py py-obj docutils literal notranslate"><span class="pre">erfinv</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise inverse gauss error function of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.exp" title="mxnet.symbol.op.exp"><code class="xref py py-obj docutils literal notranslate"><span class="pre">exp</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise exponential value of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.expand_dims" title="mxnet.symbol.op.expand_dims"><code class="xref py py-obj docutils literal notranslate"><span class="pre">expand_dims</span></code></a>([data, axis, name, attr, out])</p></td> |
| <td><p>Inserts a new axis of size 1 into the array shape For example, given <code class="docutils literal notranslate"><span class="pre">x</span></code> with shape <code class="docutils literal notranslate"><span class="pre">(2,3,4)</span></code>, then <code class="docutils literal notranslate"><span class="pre">expand_dims(x,</span> <span class="pre">axis=1)</span></code> will return a new array with shape <code class="docutils literal notranslate"><span class="pre">(2,1,3,4)</span></code>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.expm1" title="mxnet.symbol.op.expm1"><code class="xref py py-obj docutils literal notranslate"><span class="pre">expm1</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns <code class="docutils literal notranslate"><span class="pre">exp(x)</span> <span class="pre">-</span> <span class="pre">1</span></code> computed element-wise on the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.fill_element_0index" title="mxnet.symbol.op.fill_element_0index"><code class="xref py py-obj docutils literal notranslate"><span class="pre">fill_element_0index</span></code></a>([lhs, mhs, rhs, name, …])</p></td> |
| <td><p>Fill one element of each line(row for python, column for R/Julia) in lhs according to index indicated by rhs and values indicated by mhs.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.fix" title="mxnet.symbol.op.fix"><code class="xref py py-obj docutils literal notranslate"><span class="pre">fix</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise rounded value to the nearest integer towards zero of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.flatten" title="mxnet.symbol.op.flatten"><code class="xref py py-obj docutils literal notranslate"><span class="pre">flatten</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Flattens the input array into a 2-D array by collapsing the higher dimensions.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.flip" title="mxnet.symbol.op.flip"><code class="xref py py-obj docutils literal notranslate"><span class="pre">flip</span></code></a>([data, axis, name, attr, out])</p></td> |
| <td><p>Reverses the order of elements along given axis while preserving array shape.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.floor" title="mxnet.symbol.op.floor"><code class="xref py py-obj docutils literal notranslate"><span class="pre">floor</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise floor of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.ftml_update" title="mxnet.symbol.op.ftml_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ftml_update</span></code></a>([weight, grad, d, v, z, lr, …])</p></td> |
| <td><p>The FTML optimizer described in <em>FTML - Follow the Moving Leader in Deep Learning</em>, available at <a class="reference external" href="http://proceedings.mlr.press/v70/zheng17a/zheng17a.pdf">http://proceedings.mlr.press/v70/zheng17a/zheng17a.pdf</a>.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.ftrl_update" title="mxnet.symbol.op.ftrl_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ftrl_update</span></code></a>([weight, grad, z, n, lr, …])</p></td> |
| <td><p>Update function for Ftrl optimizer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.gamma" title="mxnet.symbol.op.gamma"><code class="xref py py-obj docutils literal notranslate"><span class="pre">gamma</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns the gamma function (extension of the factorial function to the reals), computed element-wise on the input array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.gammaln" title="mxnet.symbol.op.gammaln"><code class="xref py py-obj docutils literal notranslate"><span class="pre">gammaln</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise log of the absolute value of the gamma function of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.gather_nd" title="mxnet.symbol.op.gather_nd"><code class="xref py py-obj docutils literal notranslate"><span class="pre">gather_nd</span></code></a>([data, indices, name, attr, out])</p></td> |
| <td><p>Gather elements or slices from <cite>data</cite> and store to a tensor whose shape is defined by <cite>indices</cite>.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.hard_sigmoid" title="mxnet.symbol.op.hard_sigmoid"><code class="xref py py-obj docutils literal notranslate"><span class="pre">hard_sigmoid</span></code></a>([data, alpha, beta, name, …])</p></td> |
| <td><p>Computes hard sigmoid of x element-wise.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.identity" title="mxnet.symbol.op.identity"><code class="xref py py-obj docutils literal notranslate"><span class="pre">identity</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns a copy of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.im2col" title="mxnet.symbol.op.im2col"><code class="xref py py-obj docutils literal notranslate"><span class="pre">im2col</span></code></a>([data, kernel, stride, dilate, pad, …])</p></td> |
| <td><p>Extract sliding blocks from input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.khatri_rao" title="mxnet.symbol.op.khatri_rao"><code class="xref py py-obj docutils literal notranslate"><span class="pre">khatri_rao</span></code></a>(*args, **kwargs)</p></td> |
| <td><p>Computes the Khatri-Rao product of the input matrices.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.lamb_update_phase1" title="mxnet.symbol.op.lamb_update_phase1"><code class="xref py py-obj docutils literal notranslate"><span class="pre">lamb_update_phase1</span></code></a>([weight, grad, mean, …])</p></td> |
| <td><p>Phase I of lamb update it performs the following operations and returns g:.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.lamb_update_phase2" title="mxnet.symbol.op.lamb_update_phase2"><code class="xref py py-obj docutils literal notranslate"><span class="pre">lamb_update_phase2</span></code></a>([weight, g, r1, r2, lr, …])</p></td> |
| <td><p>Phase II of lamb update it performs the following operations and updates grad.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_det" title="mxnet.symbol.op.linalg_det"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_det</span></code></a>([A, name, attr, out])</p></td> |
| <td><p>Compute the determinant of a matrix.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_extractdiag" title="mxnet.symbol.op.linalg_extractdiag"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_extractdiag</span></code></a>([A, offset, name, attr, out])</p></td> |
| <td><p>Extracts the diagonal entries of a square matrix.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_extracttrian" title="mxnet.symbol.op.linalg_extracttrian"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_extracttrian</span></code></a>([A, offset, lower, …])</p></td> |
| <td><p>Extracts a triangular sub-matrix from a square matrix.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_gelqf" title="mxnet.symbol.op.linalg_gelqf"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_gelqf</span></code></a>([A, name, attr, out])</p></td> |
| <td><p>LQ factorization for general matrix.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_gemm" title="mxnet.symbol.op.linalg_gemm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_gemm</span></code></a>([A, B, C, transpose_a, …])</p></td> |
| <td><p>Performs general matrix multiplication and accumulation.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_gemm2" title="mxnet.symbol.op.linalg_gemm2"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_gemm2</span></code></a>([A, B, transpose_a, …])</p></td> |
| <td><p>Performs general matrix multiplication.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_inverse" title="mxnet.symbol.op.linalg_inverse"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_inverse</span></code></a>([A, name, attr, out])</p></td> |
| <td><p>Compute the inverse of a matrix.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_makediag" title="mxnet.symbol.op.linalg_makediag"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_makediag</span></code></a>([A, offset, name, attr, out])</p></td> |
| <td><p>Constructs a square matrix with the input as diagonal.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_maketrian" title="mxnet.symbol.op.linalg_maketrian"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_maketrian</span></code></a>([A, offset, lower, name, …])</p></td> |
| <td><p>Constructs a square matrix with the input representing a specific triangular sub-matrix.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_potrf" title="mxnet.symbol.op.linalg_potrf"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_potrf</span></code></a>([A, name, attr, out])</p></td> |
| <td><p>Performs Cholesky factorization of a symmetric positive-definite matrix.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_potri" title="mxnet.symbol.op.linalg_potri"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_potri</span></code></a>([A, name, attr, out])</p></td> |
| <td><p>Performs matrix inversion from a Cholesky factorization.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_slogdet" title="mxnet.symbol.op.linalg_slogdet"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_slogdet</span></code></a>([A, name, attr, out])</p></td> |
| <td><p>Compute the sign and log of the determinant of a matrix.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_sumlogdiag" title="mxnet.symbol.op.linalg_sumlogdiag"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_sumlogdiag</span></code></a>([A, name, attr, out])</p></td> |
| <td><p>Computes the sum of the logarithms of the diagonal elements of a square matrix.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_syrk" title="mxnet.symbol.op.linalg_syrk"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_syrk</span></code></a>([A, transpose, alpha, name, …])</p></td> |
| <td><p>Multiplication of matrix with its transpose.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_trmm" title="mxnet.symbol.op.linalg_trmm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_trmm</span></code></a>([A, B, transpose, rightside, …])</p></td> |
| <td><p>Performs multiplication with a lower triangular matrix.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.linalg_trsm" title="mxnet.symbol.op.linalg_trsm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">linalg_trsm</span></code></a>([A, B, transpose, rightside, …])</p></td> |
| <td><p>Solves matrix equation involving a lower triangular matrix.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.log" title="mxnet.symbol.op.log"><code class="xref py py-obj docutils literal notranslate"><span class="pre">log</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise Natural logarithmic value of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.log10" title="mxnet.symbol.op.log10"><code class="xref py py-obj docutils literal notranslate"><span class="pre">log10</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise Base-10 logarithmic value of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.log1p" title="mxnet.symbol.op.log1p"><code class="xref py py-obj docutils literal notranslate"><span class="pre">log1p</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise <code class="docutils literal notranslate"><span class="pre">log(1</span> <span class="pre">+</span> <span class="pre">x)</span></code> value of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.log2" title="mxnet.symbol.op.log2"><code class="xref py py-obj docutils literal notranslate"><span class="pre">log2</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise Base-2 logarithmic value of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.log_softmax" title="mxnet.symbol.op.log_softmax"><code class="xref py py-obj docutils literal notranslate"><span class="pre">log_softmax</span></code></a>([data, axis, temperature, …])</p></td> |
| <td><p>Computes the log softmax of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.logical_not" title="mxnet.symbol.op.logical_not"><code class="xref py py-obj docutils literal notranslate"><span class="pre">logical_not</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns the result of logical NOT (!) function</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.make_loss" title="mxnet.symbol.op.make_loss"><code class="xref py py-obj docutils literal notranslate"><span class="pre">make_loss</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Make your own loss function in network construction.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.max" title="mxnet.symbol.op.max"><code class="xref py py-obj docutils literal notranslate"><span class="pre">max</span></code></a>([data, axis, keepdims, exclude, name, …])</p></td> |
| <td><p>Computes the max of array elements over given axes.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.max_axis" title="mxnet.symbol.op.max_axis"><code class="xref py py-obj docutils literal notranslate"><span class="pre">max_axis</span></code></a>([data, axis, keepdims, exclude, …])</p></td> |
| <td><p>Computes the max of array elements over given axes.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.mean" title="mxnet.symbol.op.mean"><code class="xref py py-obj docutils literal notranslate"><span class="pre">mean</span></code></a>([data, axis, keepdims, exclude, name, …])</p></td> |
| <td><p>Computes the mean of array elements over given axes.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.min" title="mxnet.symbol.op.min"><code class="xref py py-obj docutils literal notranslate"><span class="pre">min</span></code></a>([data, axis, keepdims, exclude, name, …])</p></td> |
| <td><p>Computes the min of array elements over given axes.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.min_axis" title="mxnet.symbol.op.min_axis"><code class="xref py py-obj docutils literal notranslate"><span class="pre">min_axis</span></code></a>([data, axis, keepdims, exclude, …])</p></td> |
| <td><p>Computes the min of array elements over given axes.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.moments" title="mxnet.symbol.op.moments"><code class="xref py py-obj docutils literal notranslate"><span class="pre">moments</span></code></a>([data, axes, keepdims, name, attr, out])</p></td> |
| <td><p>Calculate the mean and variance of <cite>data</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.mp_lamb_update_phase1" title="mxnet.symbol.op.mp_lamb_update_phase1"><code class="xref py py-obj docutils literal notranslate"><span class="pre">mp_lamb_update_phase1</span></code></a>([weight, grad, mean, …])</p></td> |
| <td><p>Mixed Precision version of Phase I of lamb update it performs the following operations and returns g:.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.mp_lamb_update_phase2" title="mxnet.symbol.op.mp_lamb_update_phase2"><code class="xref py py-obj docutils literal notranslate"><span class="pre">mp_lamb_update_phase2</span></code></a>([weight, g, r1, r2, …])</p></td> |
| <td><p>Mixed Precision version Phase II of lamb update it performs the following operations and updates grad.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.mp_nag_mom_update" title="mxnet.symbol.op.mp_nag_mom_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">mp_nag_mom_update</span></code></a>([weight, grad, mom, …])</p></td> |
| <td><p>Update function for multi-precision Nesterov Accelerated Gradient( NAG) optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.mp_sgd_mom_update" title="mxnet.symbol.op.mp_sgd_mom_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">mp_sgd_mom_update</span></code></a>([weight, grad, mom, …])</p></td> |
| <td><p>Updater function for multi-precision sgd optimizer</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.mp_sgd_update" title="mxnet.symbol.op.mp_sgd_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">mp_sgd_update</span></code></a>([weight, grad, weight32, lr, …])</p></td> |
| <td><p>Updater function for multi-precision sgd optimizer</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.multi_all_finite" title="mxnet.symbol.op.multi_all_finite"><code class="xref py py-obj docutils literal notranslate"><span class="pre">multi_all_finite</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Check if all the float numbers in all the arrays are finite (used for AMP)</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.multi_lars" title="mxnet.symbol.op.multi_lars"><code class="xref py py-obj docutils literal notranslate"><span class="pre">multi_lars</span></code></a>([lrs, weights_sum_sq, …])</p></td> |
| <td><p>Compute the LARS coefficients of multiple weights and grads from their sums of square”</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.multi_mp_sgd_mom_update" title="mxnet.symbol.op.multi_mp_sgd_mom_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">multi_mp_sgd_mom_update</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Momentum update function for multi-precision Stochastic Gradient Descent (SGD) optimizer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.multi_mp_sgd_update" title="mxnet.symbol.op.multi_mp_sgd_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">multi_mp_sgd_update</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Update function for multi-precision Stochastic Gradient Descent (SDG) optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.multi_sgd_mom_update" title="mxnet.symbol.op.multi_sgd_mom_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">multi_sgd_mom_update</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Momentum update function for Stochastic Gradient Descent (SGD) optimizer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.multi_sgd_update" title="mxnet.symbol.op.multi_sgd_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">multi_sgd_update</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Update function for Stochastic Gradient Descent (SDG) optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.multi_sum_sq" title="mxnet.symbol.op.multi_sum_sq"><code class="xref py py-obj docutils literal notranslate"><span class="pre">multi_sum_sq</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Compute the sums of squares of multiple arrays</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.nag_mom_update" title="mxnet.symbol.op.nag_mom_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nag_mom_update</span></code></a>([weight, grad, mom, lr, …])</p></td> |
| <td><p>Update function for Nesterov Accelerated Gradient( NAG) optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.nanprod" title="mxnet.symbol.op.nanprod"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nanprod</span></code></a>([data, axis, keepdims, exclude, …])</p></td> |
| <td><p>Computes the product of array elements over given axes treating Not a Numbers (<code class="docutils literal notranslate"><span class="pre">NaN</span></code>) as one.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.nansum" title="mxnet.symbol.op.nansum"><code class="xref py py-obj docutils literal notranslate"><span class="pre">nansum</span></code></a>([data, axis, keepdims, exclude, …])</p></td> |
| <td><p>Computes the sum of array elements over given axes treating Not a Numbers (<code class="docutils literal notranslate"><span class="pre">NaN</span></code>) as zero.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.negative" title="mxnet.symbol.op.negative"><code class="xref py py-obj docutils literal notranslate"><span class="pre">negative</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Numerical negative of the argument, element-wise.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.norm" title="mxnet.symbol.op.norm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">norm</span></code></a>([data, ord, axis, out_dtype, keepdims, …])</p></td> |
| <td><p>Computes the norm on an NDArray.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.normal" title="mxnet.symbol.op.normal"><code class="xref py py-obj docutils literal notranslate"><span class="pre">normal</span></code></a>([loc, scale, shape, ctx, dtype, …])</p></td> |
| <td><p>Draw random samples from a normal (Gaussian) distribution.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.one_hot" title="mxnet.symbol.op.one_hot"><code class="xref py py-obj docutils literal notranslate"><span class="pre">one_hot</span></code></a>([indices, depth, on_value, …])</p></td> |
| <td><p>Returns a one-hot array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.ones_like" title="mxnet.symbol.op.ones_like"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ones_like</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Return an array of ones with the same shape and type as the input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.pad" title="mxnet.symbol.op.pad"><code class="xref py py-obj docutils literal notranslate"><span class="pre">pad</span></code></a>([data, mode, pad_width, constant_value, …])</p></td> |
| <td><p>Pads an input array with a constant or edge values of the array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.pick" title="mxnet.symbol.op.pick"><code class="xref py py-obj docutils literal notranslate"><span class="pre">pick</span></code></a>([data, index, axis, keepdims, mode, …])</p></td> |
| <td><p>Picks elements from an input array according to the input indices along the given axis.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.preloaded_multi_mp_sgd_mom_update" title="mxnet.symbol.op.preloaded_multi_mp_sgd_mom_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">preloaded_multi_mp_sgd_mom_update</span></code></a>(*data, …)</p></td> |
| <td><p>Momentum update function for multi-precision Stochastic Gradient Descent (SGD) optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.preloaded_multi_mp_sgd_update" title="mxnet.symbol.op.preloaded_multi_mp_sgd_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">preloaded_multi_mp_sgd_update</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Update function for multi-precision Stochastic Gradient Descent (SDG) optimizer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.preloaded_multi_sgd_mom_update" title="mxnet.symbol.op.preloaded_multi_sgd_mom_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">preloaded_multi_sgd_mom_update</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Momentum update function for Stochastic Gradient Descent (SGD) optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.preloaded_multi_sgd_update" title="mxnet.symbol.op.preloaded_multi_sgd_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">preloaded_multi_sgd_update</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Update function for Stochastic Gradient Descent (SDG) optimizer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.prod" title="mxnet.symbol.op.prod"><code class="xref py py-obj docutils literal notranslate"><span class="pre">prod</span></code></a>([data, axis, keepdims, exclude, name, …])</p></td> |
| <td><p>Computes the product of array elements over given axes.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.radians" title="mxnet.symbol.op.radians"><code class="xref py py-obj docutils literal notranslate"><span class="pre">radians</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Converts each element of the input array from degrees to radians.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_exponential" title="mxnet.symbol.op.random_exponential"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_exponential</span></code></a>([lam, shape, ctx, dtype, …])</p></td> |
| <td><p>Draw random samples from an exponential distribution.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_gamma" title="mxnet.symbol.op.random_gamma"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_gamma</span></code></a>([alpha, beta, shape, ctx, …])</p></td> |
| <td><p>Draw random samples from a gamma distribution.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_generalized_negative_binomial" title="mxnet.symbol.op.random_generalized_negative_binomial"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_generalized_negative_binomial</span></code></a>([mu, …])</p></td> |
| <td><p>Draw random samples from a generalized negative binomial distribution.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_negative_binomial" title="mxnet.symbol.op.random_negative_binomial"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_negative_binomial</span></code></a>([k, p, shape, ctx, …])</p></td> |
| <td><p>Draw random samples from a negative binomial distribution.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_normal" title="mxnet.symbol.op.random_normal"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_normal</span></code></a>([loc, scale, shape, ctx, …])</p></td> |
| <td><p>Draw random samples from a normal (Gaussian) distribution.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_pdf_dirichlet" title="mxnet.symbol.op.random_pdf_dirichlet"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_pdf_dirichlet</span></code></a>([sample, alpha, …])</p></td> |
| <td><p>Computes the value of the PDF of <em>sample</em> of Dirichlet distributions with parameter <em>alpha</em>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_pdf_exponential" title="mxnet.symbol.op.random_pdf_exponential"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_pdf_exponential</span></code></a>([sample, lam, …])</p></td> |
| <td><p>Computes the value of the PDF of <em>sample</em> of exponential distributions with parameters <em>lam</em> (rate).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_pdf_gamma" title="mxnet.symbol.op.random_pdf_gamma"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_pdf_gamma</span></code></a>([sample, alpha, beta, …])</p></td> |
| <td><p>Computes the value of the PDF of <em>sample</em> of gamma distributions with parameters <em>alpha</em> (shape) and <em>beta</em> (rate).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_pdf_generalized_negative_binomial" title="mxnet.symbol.op.random_pdf_generalized_negative_binomial"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_pdf_generalized_negative_binomial</span></code></a>([…])</p></td> |
| <td><p>Computes the value of the PDF of <em>sample</em> of generalized negative binomial distributions with parameters <em>mu</em> (mean) and <em>alpha</em> (dispersion).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_pdf_negative_binomial" title="mxnet.symbol.op.random_pdf_negative_binomial"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_pdf_negative_binomial</span></code></a>([sample, k, p, …])</p></td> |
| <td><p>Computes the value of the PDF of samples of negative binomial distributions with parameters <em>k</em> (failure limit) and <em>p</em> (failure probability).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_pdf_normal" title="mxnet.symbol.op.random_pdf_normal"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_pdf_normal</span></code></a>([sample, mu, sigma, …])</p></td> |
| <td><p>Computes the value of the PDF of <em>sample</em> of normal distributions with parameters <em>mu</em> (mean) and <em>sigma</em> (standard deviation).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_pdf_poisson" title="mxnet.symbol.op.random_pdf_poisson"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_pdf_poisson</span></code></a>([sample, lam, is_log, …])</p></td> |
| <td><p>Computes the value of the PDF of <em>sample</em> of Poisson distributions with parameters <em>lam</em> (rate).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_pdf_uniform" title="mxnet.symbol.op.random_pdf_uniform"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_pdf_uniform</span></code></a>([sample, low, high, …])</p></td> |
| <td><p>Computes the value of the PDF of <em>sample</em> of uniform distributions on the intervals given by <em>[low,high)</em>.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_poisson" title="mxnet.symbol.op.random_poisson"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_poisson</span></code></a>([lam, shape, ctx, dtype, …])</p></td> |
| <td><p>Draw random samples from a Poisson distribution.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_randint" title="mxnet.symbol.op.random_randint"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_randint</span></code></a>([low, high, shape, ctx, …])</p></td> |
| <td><p>Draw random samples from a discrete uniform distribution.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.random_uniform" title="mxnet.symbol.op.random_uniform"><code class="xref py py-obj docutils literal notranslate"><span class="pre">random_uniform</span></code></a>([low, high, shape, ctx, …])</p></td> |
| <td><p>Draw random samples from a uniform distribution.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.ravel_multi_index" title="mxnet.symbol.op.ravel_multi_index"><code class="xref py py-obj docutils literal notranslate"><span class="pre">ravel_multi_index</span></code></a>([data, shape, name, attr, out])</p></td> |
| <td><p>Converts a batch of index arrays into an array of flat indices.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.rcbrt" title="mxnet.symbol.op.rcbrt"><code class="xref py py-obj docutils literal notranslate"><span class="pre">rcbrt</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise inverse cube-root value of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.reciprocal" title="mxnet.symbol.op.reciprocal"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reciprocal</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns the reciprocal of the argument, element-wise.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.relu" title="mxnet.symbol.op.relu"><code class="xref py py-obj docutils literal notranslate"><span class="pre">relu</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Computes rectified linear activation.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.repeat" title="mxnet.symbol.op.repeat"><code class="xref py py-obj docutils literal notranslate"><span class="pre">repeat</span></code></a>([data, repeats, axis, name, attr, out])</p></td> |
| <td><p>Repeats elements of an array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.reset_arrays" title="mxnet.symbol.op.reset_arrays"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reset_arrays</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Set to zero multiple arrays</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.reshape" title="mxnet.symbol.op.reshape"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reshape</span></code></a>([data, shape, reverse, …])</p></td> |
| <td><p>Reshapes the input array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.reshape_like" title="mxnet.symbol.op.reshape_like"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reshape_like</span></code></a>([lhs, rhs, lhs_begin, lhs_end, …])</p></td> |
| <td><p>Reshape some or all dimensions of <cite>lhs</cite> to have the same shape as some or all dimensions of <cite>rhs</cite>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.reverse" title="mxnet.symbol.op.reverse"><code class="xref py py-obj docutils literal notranslate"><span class="pre">reverse</span></code></a>([data, axis, name, attr, out])</p></td> |
| <td><p>Reverses the order of elements along given axis while preserving array shape.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.rint" title="mxnet.symbol.op.rint"><code class="xref py py-obj docutils literal notranslate"><span class="pre">rint</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise rounded value to the nearest integer of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.rmsprop_update" title="mxnet.symbol.op.rmsprop_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">rmsprop_update</span></code></a>([weight, grad, n, lr, …])</p></td> |
| <td><p>Update function for <cite>RMSProp</cite> optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.rmspropalex_update" title="mxnet.symbol.op.rmspropalex_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">rmspropalex_update</span></code></a>([weight, grad, n, g, …])</p></td> |
| <td><p>Update function for RMSPropAlex optimizer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.round" title="mxnet.symbol.op.round"><code class="xref py py-obj docutils literal notranslate"><span class="pre">round</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise rounded value to the nearest integer of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.rsqrt" title="mxnet.symbol.op.rsqrt"><code class="xref py py-obj docutils literal notranslate"><span class="pre">rsqrt</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise inverse square-root value of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.sample_exponential" title="mxnet.symbol.op.sample_exponential"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sample_exponential</span></code></a>([lam, shape, dtype, …])</p></td> |
| <td><p>Concurrent sampling from multiple exponential distributions with parameters lambda (rate).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.sample_gamma" title="mxnet.symbol.op.sample_gamma"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sample_gamma</span></code></a>([alpha, beta, shape, dtype, …])</p></td> |
| <td><p>Concurrent sampling from multiple gamma distributions with parameters <em>alpha</em> (shape) and <em>beta</em> (scale).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.sample_generalized_negative_binomial" title="mxnet.symbol.op.sample_generalized_negative_binomial"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sample_generalized_negative_binomial</span></code></a>([mu, …])</p></td> |
| <td><p>Concurrent sampling from multiple generalized negative binomial distributions with parameters <em>mu</em> (mean) and <em>alpha</em> (dispersion).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.sample_multinomial" title="mxnet.symbol.op.sample_multinomial"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sample_multinomial</span></code></a>([data, shape, get_prob, …])</p></td> |
| <td><p>Concurrent sampling from multiple multinomial distributions.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.sample_negative_binomial" title="mxnet.symbol.op.sample_negative_binomial"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sample_negative_binomial</span></code></a>([k, p, shape, …])</p></td> |
| <td><p>Concurrent sampling from multiple negative binomial distributions with parameters <em>k</em> (failure limit) and <em>p</em> (failure probability).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.sample_normal" title="mxnet.symbol.op.sample_normal"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sample_normal</span></code></a>([mu, sigma, shape, dtype, …])</p></td> |
| <td><p>Concurrent sampling from multiple normal distributions with parameters <em>mu</em> (mean) and <em>sigma</em> (standard deviation).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.sample_poisson" title="mxnet.symbol.op.sample_poisson"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sample_poisson</span></code></a>([lam, shape, dtype, name, …])</p></td> |
| <td><p>Concurrent sampling from multiple Poisson distributions with parameters lambda (rate).</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.sample_uniform" title="mxnet.symbol.op.sample_uniform"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sample_uniform</span></code></a>([low, high, shape, dtype, …])</p></td> |
| <td><p>Concurrent sampling from multiple uniform distributions on the intervals given by <em>[low,high)</em>.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.scatter_nd" title="mxnet.symbol.op.scatter_nd"><code class="xref py py-obj docutils literal notranslate"><span class="pre">scatter_nd</span></code></a>([data, indices, shape, name, …])</p></td> |
| <td><p>Scatters data into a new tensor according to indices.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.sgd_mom_update" title="mxnet.symbol.op.sgd_mom_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sgd_mom_update</span></code></a>([weight, grad, mom, lr, …])</p></td> |
| <td><p>Momentum update function for Stochastic Gradient Descent (SGD) optimizer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.sgd_update" title="mxnet.symbol.op.sgd_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sgd_update</span></code></a>([weight, grad, lr, wd, …])</p></td> |
| <td><p>Update function for Stochastic Gradient Descent (SGD) optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.shape_array" title="mxnet.symbol.op.shape_array"><code class="xref py py-obj docutils literal notranslate"><span class="pre">shape_array</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns a 1D int64 array containing the shape of data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.shuffle" title="mxnet.symbol.op.shuffle"><code class="xref py py-obj docutils literal notranslate"><span class="pre">shuffle</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Randomly shuffle the elements.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.sigmoid" title="mxnet.symbol.op.sigmoid"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sigmoid</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Computes sigmoid of x element-wise.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.sign" title="mxnet.symbol.op.sign"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sign</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise sign of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.signsgd_update" title="mxnet.symbol.op.signsgd_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">signsgd_update</span></code></a>([weight, grad, lr, wd, …])</p></td> |
| <td><p>Update function for SignSGD optimizer.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.signum_update" title="mxnet.symbol.op.signum_update"><code class="xref py py-obj docutils literal notranslate"><span class="pre">signum_update</span></code></a>([weight, grad, mom, lr, …])</p></td> |
| <td><p>SIGN momentUM (Signum) optimizer.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.sin" title="mxnet.symbol.op.sin"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sin</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Computes the element-wise sine of the input array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.sinh" title="mxnet.symbol.op.sinh"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sinh</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns the hyperbolic sine of the input array, computed element-wise.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.size_array" title="mxnet.symbol.op.size_array"><code class="xref py py-obj docutils literal notranslate"><span class="pre">size_array</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns a 1D int64 array containing the size of data.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.slice" title="mxnet.symbol.op.slice"><code class="xref py py-obj docutils literal notranslate"><span class="pre">slice</span></code></a>([data, begin, end, step, name, attr, out])</p></td> |
| <td><p>Slices a region of the array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.slice_axis" title="mxnet.symbol.op.slice_axis"><code class="xref py py-obj docutils literal notranslate"><span class="pre">slice_axis</span></code></a>([data, axis, begin, end, name, …])</p></td> |
| <td><p>Slices along a given axis.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.slice_like" title="mxnet.symbol.op.slice_like"><code class="xref py py-obj docutils literal notranslate"><span class="pre">slice_like</span></code></a>([data, shape_like, axes, name, …])</p></td> |
| <td><p>Slices a region of the array like the shape of another array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.smooth_l1" title="mxnet.symbol.op.smooth_l1"><code class="xref py py-obj docutils literal notranslate"><span class="pre">smooth_l1</span></code></a>([data, scalar, name, attr, out])</p></td> |
| <td><p>Calculate Smooth L1 Loss(lhs, scalar) by summing</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.softmax" title="mxnet.symbol.op.softmax"><code class="xref py py-obj docutils literal notranslate"><span class="pre">softmax</span></code></a>([data, length, axis, temperature, …])</p></td> |
| <td><p>Applies the softmax function.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.softmax_cross_entropy" title="mxnet.symbol.op.softmax_cross_entropy"><code class="xref py py-obj docutils literal notranslate"><span class="pre">softmax_cross_entropy</span></code></a>([data, label, name, …])</p></td> |
| <td><p>Calculate cross entropy of softmax output and one-hot label.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.softmin" title="mxnet.symbol.op.softmin"><code class="xref py py-obj docutils literal notranslate"><span class="pre">softmin</span></code></a>([data, axis, temperature, dtype, …])</p></td> |
| <td><p>Applies the softmin function.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.softsign" title="mxnet.symbol.op.softsign"><code class="xref py py-obj docutils literal notranslate"><span class="pre">softsign</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Computes softsign of x element-wise.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.sort" title="mxnet.symbol.op.sort"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sort</span></code></a>([data, axis, is_ascend, name, attr, out])</p></td> |
| <td><p>Returns a sorted copy of an input array along the given axis.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.space_to_depth" title="mxnet.symbol.op.space_to_depth"><code class="xref py py-obj docutils literal notranslate"><span class="pre">space_to_depth</span></code></a>([data, block_size, name, …])</p></td> |
| <td><p>Rearranges(permutes) blocks of spatial data into depth.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.split" title="mxnet.symbol.op.split"><code class="xref py py-obj docutils literal notranslate"><span class="pre">split</span></code></a>([data, num_outputs, axis, …])</p></td> |
| <td><p>Splits an array along a particular axis into multiple sub-arrays.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.sqrt" title="mxnet.symbol.op.sqrt"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sqrt</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise square-root value of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.square" title="mxnet.symbol.op.square"><code class="xref py py-obj docutils literal notranslate"><span class="pre">square</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns element-wise squared value of the input.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.squeeze" title="mxnet.symbol.op.squeeze"><code class="xref py py-obj docutils literal notranslate"><span class="pre">squeeze</span></code></a>([data, axis, name, attr, out])</p></td> |
| <td><p>Remove single-dimensional entries from the shape of an array.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.stack" title="mxnet.symbol.op.stack"><code class="xref py py-obj docutils literal notranslate"><span class="pre">stack</span></code></a>(*data, **kwargs)</p></td> |
| <td><p>Join a sequence of arrays along a new axis.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.stop_gradient" title="mxnet.symbol.op.stop_gradient"><code class="xref py py-obj docutils literal notranslate"><span class="pre">stop_gradient</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Stops gradient computation.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.sum" title="mxnet.symbol.op.sum"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sum</span></code></a>([data, axis, keepdims, exclude, name, …])</p></td> |
| <td><p>Computes the sum of array elements over given axes.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.sum_axis" title="mxnet.symbol.op.sum_axis"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sum_axis</span></code></a>([data, axis, keepdims, exclude, …])</p></td> |
| <td><p>Computes the sum of array elements over given axes.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.swapaxes" title="mxnet.symbol.op.swapaxes"><code class="xref py py-obj docutils literal notranslate"><span class="pre">swapaxes</span></code></a>([data, dim1, dim2, name, attr, out])</p></td> |
| <td><p>Interchanges two axes of an array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.take" title="mxnet.symbol.op.take"><code class="xref py py-obj docutils literal notranslate"><span class="pre">take</span></code></a>([a, indices, axis, mode, name, attr, out])</p></td> |
| <td><p>Takes elements from an input array along the given axis.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.tan" title="mxnet.symbol.op.tan"><code class="xref py py-obj docutils literal notranslate"><span class="pre">tan</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Computes the element-wise tangent of the input array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.tanh" title="mxnet.symbol.op.tanh"><code class="xref py py-obj docutils literal notranslate"><span class="pre">tanh</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Returns the hyperbolic tangent of the input array, computed element-wise.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.tile" title="mxnet.symbol.op.tile"><code class="xref py py-obj docutils literal notranslate"><span class="pre">tile</span></code></a>([data, reps, name, attr, out])</p></td> |
| <td><p>Repeats the whole array multiple times.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.topk" title="mxnet.symbol.op.topk"><code class="xref py py-obj docutils literal notranslate"><span class="pre">topk</span></code></a>([data, axis, k, ret_typ, is_ascend, …])</p></td> |
| <td><p>Returns the indices of the top <em>k</em> elements in an input array along the given axis (by default).</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.transpose" title="mxnet.symbol.op.transpose"><code class="xref py py-obj docutils literal notranslate"><span class="pre">transpose</span></code></a>([data, axes, name, attr, out])</p></td> |
| <td><p>Permutes the dimensions of an array.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.trunc" title="mxnet.symbol.op.trunc"><code class="xref py py-obj docutils literal notranslate"><span class="pre">trunc</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Return the element-wise truncated value of the input.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.uniform" title="mxnet.symbol.op.uniform"><code class="xref py py-obj docutils literal notranslate"><span class="pre">uniform</span></code></a>([low, high, shape, ctx, dtype, …])</p></td> |
| <td><p>Draw random samples from a uniform distribution.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.unravel_index" title="mxnet.symbol.op.unravel_index"><code class="xref py py-obj docutils literal notranslate"><span class="pre">unravel_index</span></code></a>([data, shape, name, attr, out])</p></td> |
| <td><p>Converts an array of flat indices into a batch of index arrays.</p></td> |
| </tr> |
| <tr class="row-even"><td><p><a class="reference internal" href="#mxnet.symbol.op.where" title="mxnet.symbol.op.where"><code class="xref py py-obj docutils literal notranslate"><span class="pre">where</span></code></a>([condition, x, y, name, attr, out])</p></td> |
| <td><p>Return the elements, either from x or y, depending on the condition.</p></td> |
| </tr> |
| <tr class="row-odd"><td><p><a class="reference internal" href="#mxnet.symbol.op.zeros_like" title="mxnet.symbol.op.zeros_like"><code class="xref py py-obj docutils literal notranslate"><span class="pre">zeros_like</span></code></a>([data, name, attr, out])</p></td> |
| <td><p>Return an array of zeros with the same shape, type and storage type as the input array.</p></td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Activation"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Activation</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">act_type=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Activation" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies an activation function element-wise to the input.</p> |
| <p>The following activation functions are supported:</p> |
| <ul class="simple"> |
| <li><p><cite>relu</cite>: Rectified Linear Unit, <span class="math notranslate nohighlight">\(y = max(x, 0)\)</span></p></li> |
| <li><p><cite>sigmoid</cite>: <span class="math notranslate nohighlight">\(y = \frac{1}{1 + exp(-x)}\)</span></p></li> |
| <li><p><cite>tanh</cite>: Hyperbolic tangent, <span class="math notranslate nohighlight">\(y = \frac{exp(x) - exp(-x)}{exp(x) + exp(-x)}\)</span></p></li> |
| <li><p><cite>softrelu</cite>: Soft ReLU, or SoftPlus, <span class="math notranslate nohighlight">\(y = log(1 + exp(x))\)</span></p></li> |
| <li><p><cite>softsign</cite>: <span class="math notranslate nohighlight">\(y = \frac{x}{1 + abs(x)}\)</span></p></li> |
| </ul> |
| <p>Defined in src/operator/nn/activation.cc:L164</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>act_type</strong> (<em>{'relu'</em><em>, </em><em>'sigmoid'</em><em>, </em><em>'softrelu'</em><em>, </em><em>'softsign'</em><em>, </em><em>'tanh'}</em><em>, </em><em>required</em>) – Activation function to be applied.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <p>A one-hidden-layer MLP with ReLU activation:</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">data</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'data'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">mlp</span> <span class="o">=</span> <span class="n">FullyConnected</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">data</span><span class="p">,</span> <span class="n">num_hidden</span><span class="o">=</span><span class="mi">128</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'proj'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">mlp</span> <span class="o">=</span> <span class="n">Activation</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">mlp</span><span class="p">,</span> <span class="n">act_type</span><span class="o">=</span><span class="s1">'relu'</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'activation'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">mlp</span> <span class="o">=</span> <span class="n">FullyConnected</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">mlp</span><span class="p">,</span> <span class="n">num_hidden</span><span class="o">=</span><span class="mi">10</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'mlp'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">mlp</span> |
| <span class="go"><Symbol mlp></span> |
| </pre></div> |
| </div> |
| <p>ReLU activation</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">test_suites</span> <span class="o">=</span> <span class="p">[</span> |
| <span class="gp">... </span><span class="p">(</span><span class="s1">'relu'</span><span class="p">,</span> <span class="k">lambda</span> <span class="n">x</span><span class="p">:</span> <span class="n">np</span><span class="o">.</span><span class="n">maximum</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="mi">0</span><span class="p">)),</span> |
| <span class="gp">... </span><span class="p">(</span><span class="s1">'sigmoid'</span><span class="p">,</span> <span class="k">lambda</span> <span class="n">x</span><span class="p">:</span> <span class="mi">1</span> <span class="o">/</span> <span class="p">(</span><span class="mi">1</span> <span class="o">+</span> <span class="n">np</span><span class="o">.</span><span class="n">exp</span><span class="p">(</span><span class="o">-</span><span class="n">x</span><span class="p">))),</span> |
| <span class="gp">... </span><span class="p">(</span><span class="s1">'tanh'</span><span class="p">,</span> <span class="k">lambda</span> <span class="n">x</span><span class="p">:</span> <span class="n">np</span><span class="o">.</span><span class="n">tanh</span><span class="p">(</span><span class="n">x</span><span class="p">)),</span> |
| <span class="gp">... </span><span class="p">(</span><span class="s1">'softrelu'</span><span class="p">,</span> <span class="k">lambda</span> <span class="n">x</span><span class="p">:</span> <span class="n">np</span><span class="o">.</span><span class="n">log</span><span class="p">(</span><span class="mi">1</span> <span class="o">+</span> <span class="n">np</span><span class="o">.</span><span class="n">exp</span><span class="p">(</span><span class="n">x</span><span class="p">)))</span> |
| <span class="gp">... </span><span class="p">]</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">random_arrays</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="k">for</span> <span class="n">act_type</span><span class="p">,</span> <span class="n">numpy_impl</span> <span class="ow">in</span> <span class="n">test_suites</span><span class="p">:</span> |
| <span class="gp">... </span><span class="n">op</span> <span class="o">=</span> <span class="n">Activation</span><span class="p">(</span><span class="n">act_type</span><span class="o">=</span><span class="n">act_type</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'act'</span><span class="p">)</span> |
| <span class="gp">... </span><span class="n">y</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">simple_forward</span><span class="p">(</span><span class="n">op</span><span class="p">,</span> <span class="n">act_data</span><span class="o">=</span><span class="n">x</span><span class="p">)</span> |
| <span class="gp">... </span><span class="n">y_np</span> <span class="o">=</span> <span class="n">numpy_impl</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> |
| <span class="gp">... </span><span class="nb">print</span><span class="p">(</span><span class="s1">'</span><span class="si">%s</span><span class="s1">: </span><span class="si">%s</span><span class="s1">'</span> <span class="o">%</span> <span class="p">(</span><span class="n">act_type</span><span class="p">,</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">almost_equal</span><span class="p">(</span><span class="n">y</span><span class="p">,</span> <span class="n">y_np</span><span class="p">)))</span> |
| <span class="go">relu: True</span> |
| <span class="go">sigmoid: True</span> |
| <span class="go">tanh: True</span> |
| <span class="go">softrelu: True</span> |
| </pre></div> |
| </div> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.BatchNorm"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">BatchNorm</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">gamma=None</em>, <em class="sig-param">beta=None</em>, <em class="sig-param">moving_mean=None</em>, <em class="sig-param">moving_var=None</em>, <em class="sig-param">eps=_Null</em>, <em class="sig-param">momentum=_Null</em>, <em class="sig-param">fix_gamma=_Null</em>, <em class="sig-param">use_global_stats=_Null</em>, <em class="sig-param">output_mean_var=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">cudnn_off=_Null</em>, <em class="sig-param">min_calib_range=_Null</em>, <em class="sig-param">max_calib_range=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.BatchNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Batch normalization.</p> |
| <p>Normalizes a data batch by mean and variance, and applies a scale <code class="docutils literal notranslate"><span class="pre">gamma</span></code> as |
| well as offset <code class="docutils literal notranslate"><span class="pre">beta</span></code>.</p> |
| <p>Assume the input has more than one dimension and we normalize along axis 1. |
| We first compute the mean and variance along this axis:</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}data\_mean[i] = mean(data[:,i,:,...]) \\ |
| data\_var[i] = var(data[:,i,:,...])\end{split}\]</div> |
| <p>Then compute the normalized output, which has the same shape as input, as following:</p> |
| <div class="math notranslate nohighlight"> |
| \[out[:,i,:,...] = \frac{data[:,i,:,...] - data\_mean[i]}{\sqrt{data\_var[i]+\epsilon}} * gamma[i] + beta[i]\]</div> |
| <p>Both <em>mean</em> and <em>var</em> returns a scalar by treating the input as a vector.</p> |
| <p>Assume the input has size <em>k</em> on axis 1, then both <code class="docutils literal notranslate"><span class="pre">gamma</span></code> and <code class="docutils literal notranslate"><span class="pre">beta</span></code> |
| have shape <em>(k,)</em>. If <code class="docutils literal notranslate"><span class="pre">output_mean_var</span></code> is set to be true, then outputs both <code class="docutils literal notranslate"><span class="pre">data_mean</span></code> and |
| the inverse of <code class="docutils literal notranslate"><span class="pre">data_var</span></code>, which are needed for the backward pass. Note that gradient of these |
| two outputs are blocked.</p> |
| <p>Besides the inputs and the outputs, this operator accepts two auxiliary |
| states, <code class="docutils literal notranslate"><span class="pre">moving_mean</span></code> and <code class="docutils literal notranslate"><span class="pre">moving_var</span></code>, which are <em>k</em>-length |
| vectors. They are global statistics for the whole dataset, which are updated |
| by:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">moving_mean</span> <span class="o">=</span> <span class="n">moving_mean</span> <span class="o">*</span> <span class="n">momentum</span> <span class="o">+</span> <span class="n">data_mean</span> <span class="o">*</span> <span class="p">(</span><span class="mi">1</span> <span class="o">-</span> <span class="n">momentum</span><span class="p">)</span> |
| <span class="n">moving_var</span> <span class="o">=</span> <span class="n">moving_var</span> <span class="o">*</span> <span class="n">momentum</span> <span class="o">+</span> <span class="n">data_var</span> <span class="o">*</span> <span class="p">(</span><span class="mi">1</span> <span class="o">-</span> <span class="n">momentum</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>If <code class="docutils literal notranslate"><span class="pre">use_global_stats</span></code> is set to be true, then <code class="docutils literal notranslate"><span class="pre">moving_mean</span></code> and |
| <code class="docutils literal notranslate"><span class="pre">moving_var</span></code> are used instead of <code class="docutils literal notranslate"><span class="pre">data_mean</span></code> and <code class="docutils literal notranslate"><span class="pre">data_var</span></code> to compute |
| the output. It is often used during inference.</p> |
| <p>The parameter <code class="docutils literal notranslate"><span class="pre">axis</span></code> specifies which axis of the input shape denotes |
| the ‘channel’ (separately normalized groups). The default is 1. Specifying -1 sets the channel |
| axis to be the last item in the input shape.</p> |
| <p>Both <code class="docutils literal notranslate"><span class="pre">gamma</span></code> and <code class="docutils literal notranslate"><span class="pre">beta</span></code> are learnable parameters. But if <code class="docutils literal notranslate"><span class="pre">fix_gamma</span></code> is true, |
| then set <code class="docutils literal notranslate"><span class="pre">gamma</span></code> to 1 and its gradient to 0.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>When <code class="docutils literal notranslate"><span class="pre">fix_gamma</span></code> is set to True, no sparse support is provided. If <code class="docutils literal notranslate"><span class="pre">fix_gamma</span> <span class="pre">is</span></code> set to False, |
| the sparse tensors will fallback.</p> |
| </div> |
| <p>Defined in src/operator/nn/batch_norm.cc:L608</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to batch normalization</p></li> |
| <li><p><strong>gamma</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – gamma array</p></li> |
| <li><p><strong>beta</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – beta array</p></li> |
| <li><p><strong>moving_mean</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – running mean of input</p></li> |
| <li><p><strong>moving_var</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – running variance of input</p></li> |
| <li><p><strong>eps</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=0.0010000000474974513</em>) – Epsilon to prevent div 0. Must be no less than CUDNN_BN_MIN_EPSILON defined in cudnn.h when using cudnn (usually 1e-5)</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.899999976</em>) – Momentum for moving average</p></li> |
| <li><p><strong>fix_gamma</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Fix gamma while training</p></li> |
| <li><p><strong>use_global_stats</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether use global moving statistics instead of local batch-norm. This will force change batch-norm into a scale shift operator.</p></li> |
| <li><p><strong>output_mean_var</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Output the mean and inverse std</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Specify which shape axis the channel is specified</p></li> |
| <li><p><strong>cudnn_off</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Do not select CUDNN operator, if available</p></li> |
| <li><p><strong>min_calib_range</strong> (<em>float</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – The minimum scalar value in the form of float32 obtained through calibration. If present, it will be used to by quantized batch norm op to calculate primitive scale.Note: this calib_range is to calib bn output.</p></li> |
| <li><p><strong>max_calib_range</strong> (<em>float</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – The maximum scalar value in the form of float32 obtained through calibration. If present, it will be used to by quantized batch norm op to calculate primitive scale.Note: this calib_range is to calib bn output.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.BatchNorm_v1"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">BatchNorm_v1</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">gamma=None</em>, <em class="sig-param">beta=None</em>, <em class="sig-param">eps=_Null</em>, <em class="sig-param">momentum=_Null</em>, <em class="sig-param">fix_gamma=_Null</em>, <em class="sig-param">use_global_stats=_Null</em>, <em class="sig-param">output_mean_var=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.BatchNorm_v1" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Batch normalization.</p> |
| <p>This operator is DEPRECATED. Perform BatchNorm on the input.</p> |
| <p>Normalizes a data batch by mean and variance, and applies a scale <code class="docutils literal notranslate"><span class="pre">gamma</span></code> as |
| well as offset <code class="docutils literal notranslate"><span class="pre">beta</span></code>.</p> |
| <p>Assume the input has more than one dimension and we normalize along axis 1. |
| We first compute the mean and variance along this axis:</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}data\_mean[i] = mean(data[:,i,:,...]) \\ |
| data\_var[i] = var(data[:,i,:,...])\end{split}\]</div> |
| <p>Then compute the normalized output, which has the same shape as input, as following:</p> |
| <div class="math notranslate nohighlight"> |
| \[out[:,i,:,...] = \frac{data[:,i,:,...] - data\_mean[i]}{\sqrt{data\_var[i]+\epsilon}} * gamma[i] + beta[i]\]</div> |
| <p>Both <em>mean</em> and <em>var</em> returns a scalar by treating the input as a vector.</p> |
| <p>Assume the input has size <em>k</em> on axis 1, then both <code class="docutils literal notranslate"><span class="pre">gamma</span></code> and <code class="docutils literal notranslate"><span class="pre">beta</span></code> |
| have shape <em>(k,)</em>. If <code class="docutils literal notranslate"><span class="pre">output_mean_var</span></code> is set to be true, then outputs both <code class="docutils literal notranslate"><span class="pre">data_mean</span></code> and |
| <code class="docutils literal notranslate"><span class="pre">data_var</span></code> as well, which are needed for the backward pass.</p> |
| <p>Besides the inputs and the outputs, this operator accepts two auxiliary |
| states, <code class="docutils literal notranslate"><span class="pre">moving_mean</span></code> and <code class="docutils literal notranslate"><span class="pre">moving_var</span></code>, which are <em>k</em>-length |
| vectors. They are global statistics for the whole dataset, which are updated |
| by:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">moving_mean</span> <span class="o">=</span> <span class="n">moving_mean</span> <span class="o">*</span> <span class="n">momentum</span> <span class="o">+</span> <span class="n">data_mean</span> <span class="o">*</span> <span class="p">(</span><span class="mi">1</span> <span class="o">-</span> <span class="n">momentum</span><span class="p">)</span> |
| <span class="n">moving_var</span> <span class="o">=</span> <span class="n">moving_var</span> <span class="o">*</span> <span class="n">momentum</span> <span class="o">+</span> <span class="n">data_var</span> <span class="o">*</span> <span class="p">(</span><span class="mi">1</span> <span class="o">-</span> <span class="n">momentum</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>If <code class="docutils literal notranslate"><span class="pre">use_global_stats</span></code> is set to be true, then <code class="docutils literal notranslate"><span class="pre">moving_mean</span></code> and |
| <code class="docutils literal notranslate"><span class="pre">moving_var</span></code> are used instead of <code class="docutils literal notranslate"><span class="pre">data_mean</span></code> and <code class="docutils literal notranslate"><span class="pre">data_var</span></code> to compute |
| the output. It is often used during inference.</p> |
| <p>Both <code class="docutils literal notranslate"><span class="pre">gamma</span></code> and <code class="docutils literal notranslate"><span class="pre">beta</span></code> are learnable parameters. But if <code class="docutils literal notranslate"><span class="pre">fix_gamma</span></code> is true, |
| then set <code class="docutils literal notranslate"><span class="pre">gamma</span></code> to 1 and its gradient to 0.</p> |
| <p>There’s no sparse support for this operator, and it will exhibit problematic behavior if used with |
| sparse tensors.</p> |
| <p>Defined in src/operator/batch_norm_v1.cc:L94</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to batch normalization</p></li> |
| <li><p><strong>gamma</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – gamma array</p></li> |
| <li><p><strong>beta</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – beta array</p></li> |
| <li><p><strong>eps</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.00100000005</em>) – Epsilon to prevent div 0</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.899999976</em>) – Momentum for moving average</p></li> |
| <li><p><strong>fix_gamma</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Fix gamma while training</p></li> |
| <li><p><strong>use_global_stats</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether use global moving statistics instead of local batch-norm. This will force change batch-norm into a scale shift operator.</p></li> |
| <li><p><strong>output_mean_var</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Output All,normal mean and var</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.BilinearSampler"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">BilinearSampler</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">grid=None</em>, <em class="sig-param">cudnn_off=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.BilinearSampler" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies bilinear sampling to input feature map.</p> |
| <p>Bilinear Sampling is the key of [NIPS2015] “Spatial Transformer Networks”. The usage of the operator is very similar to remap function in OpenCV, |
| except that the operator has the backward pass.</p> |
| <p>Given <span class="math notranslate nohighlight">\(data\)</span> and <span class="math notranslate nohighlight">\(grid\)</span>, then the output is computed by</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}x_{src} = grid[batch, 0, y_{dst}, x_{dst}] \\ |
| y_{src} = grid[batch, 1, y_{dst}, x_{dst}] \\ |
| output[batch, channel, y_{dst}, x_{dst}] = G(data[batch, channel, y_{src}, x_{src})\end{split}\]</div> |
| <p><span class="math notranslate nohighlight">\(x_{dst}\)</span>, <span class="math notranslate nohighlight">\(y_{dst}\)</span> enumerate all spatial locations in <span class="math notranslate nohighlight">\(output\)</span>, and <span class="math notranslate nohighlight">\(G()\)</span> denotes the bilinear interpolation kernel. |
| The out-boundary points will be padded with zeros.The shape of the output will be (data.shape[0], data.shape[1], grid.shape[2], grid.shape[3]).</p> |
| <p>The operator assumes that <span class="math notranslate nohighlight">\(data\)</span> has ‘NCHW’ layout and <span class="math notranslate nohighlight">\(grid\)</span> has been normalized to [-1, 1].</p> |
| <p>BilinearSampler often cooperates with GridGenerator which generates sampling grids for BilinearSampler. |
| GridGenerator supports two kinds of transformation: <code class="docutils literal notranslate"><span class="pre">affine</span></code> and <code class="docutils literal notranslate"><span class="pre">warp</span></code>. |
| If users want to design a CustomOp to manipulate <span class="math notranslate nohighlight">\(grid\)</span>, please firstly refer to the code of GridGenerator.</p> |
| <p>Example 1:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1">## Zoom out data two times</span> |
| <span class="n">data</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">4</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">6</span><span class="p">],</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">8</span><span class="p">,</span> <span class="mi">9</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">1</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">0</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="n">affine_matrix</span> <span class="o">=</span> <span class="n">array</span><span class="p">([[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">]])</span> |
| |
| <span class="n">affine_matrix</span> <span class="o">=</span> <span class="n">reshape</span><span class="p">(</span><span class="n">affine_matrix</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">6</span><span class="p">))</span> |
| |
| <span class="n">grid</span> <span class="o">=</span> <span class="n">GridGenerator</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">affine_matrix</span><span class="p">,</span> <span class="n">transform_type</span><span class="o">=</span><span class="s1">'affine'</span><span class="p">,</span> <span class="n">target_shape</span><span class="o">=</span><span class="p">(</span><span class="mi">4</span><span class="p">,</span> <span class="mi">4</span><span class="p">))</span> |
| |
| <span class="n">out</span> <span class="o">=</span> <span class="n">BilinearSampler</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">grid</span><span class="p">)</span> |
| |
| <span class="n">out</span> |
| <span class="p">[[[[</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">0</span><span class="p">,</span> <span class="mf">3.5</span><span class="p">,</span> <span class="mf">6.5</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">0</span><span class="p">,</span> <span class="mf">1.25</span><span class="p">,</span> <span class="mf">2.5</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Example 2:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1">## shift data horizontally by -1 pixel</span> |
| |
| <span class="n">data</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">4</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">6</span><span class="p">],</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">8</span><span class="p">,</span> <span class="mi">9</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">1</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">0</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="n">warp_maxtrix</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">1</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="p">[</span><span class="mi">1</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">1</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">1</span><span class="p">,</span> <span class="mi">1</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">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]]])</span> |
| |
| <span class="n">grid</span> <span class="o">=</span> <span class="n">GridGenerator</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">warp_matrix</span><span class="p">,</span> <span class="n">transform_type</span><span class="o">=</span><span class="s1">'warp'</span><span class="p">)</span> |
| <span class="n">out</span> <span class="o">=</span> <span class="n">BilinearSampler</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">grid</span><span class="p">)</span> |
| |
| <span class="n">out</span> |
| <span class="p">[[[[</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">6</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">8</span><span class="p">,</span> <span class="mi">8</span><span class="p">,</span> <span class="mi">9</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</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">3</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/bilinear_sampler.cc:L255</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the BilinearsamplerOp.</p></li> |
| <li><p><strong>grid</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input grid to the BilinearsamplerOp.grid has two channels: x_src, y_src</p></li> |
| <li><p><strong>cudnn_off</strong> (<em>boolean</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – whether to turn cudnn off</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.BlockGrad"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">BlockGrad</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.BlockGrad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Stops gradient computation.</p> |
| <p>Stops the accumulated gradient of the inputs from flowing through this operator |
| in the backward direction. In other words, this operator prevents the contribution |
| of its inputs to be taken into account for computing gradients.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">v1</span> <span class="o">=</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="n">v2</span> <span class="o">=</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="n">a</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'a'</span><span class="p">)</span> |
| <span class="n">b</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'b'</span><span class="p">)</span> |
| <span class="n">b_stop_grad</span> <span class="o">=</span> <span class="n">stop_gradient</span><span class="p">(</span><span class="mi">3</span> <span class="o">*</span> <span class="n">b</span><span class="p">)</span> |
| <span class="n">loss</span> <span class="o">=</span> <span class="n">MakeLoss</span><span class="p">(</span><span class="n">b_stop_grad</span> <span class="o">+</span> <span class="n">a</span><span class="p">)</span> |
| |
| <span class="n">executor</span> <span class="o">=</span> <span class="n">loss</span><span class="o">.</span><span class="n">simple_bind</span><span class="p">(</span><span class="n">ctx</span><span class="o">=</span><span class="n">cpu</span><span class="p">(),</span> <span class="n">a</span><span class="o">=</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="n">b</span><span class="o">=</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="n">executor</span><span class="o">.</span><span class="n">forward</span><span class="p">(</span><span class="n">is_train</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">a</span><span class="o">=</span><span class="n">v1</span><span class="p">,</span> <span class="n">b</span><span class="o">=</span><span class="n">v2</span><span class="p">)</span> |
| <span class="n">executor</span><span class="o">.</span><span class="n">outputs</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">5.</span><span class="p">]</span> |
| |
| <span class="n">executor</span><span class="o">.</span><span class="n">backward</span><span class="p">()</span> |
| <span class="n">executor</span><span class="o">.</span><span class="n">grad_arrays</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L325</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.CTCLoss"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">CTCLoss</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">label=None</em>, <em class="sig-param">data_lengths=None</em>, <em class="sig-param">label_lengths=None</em>, <em class="sig-param">use_data_lengths=_Null</em>, <em class="sig-param">use_label_lengths=_Null</em>, <em class="sig-param">blank_label=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.CTCLoss" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Connectionist Temporal Classification Loss.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The existing alias <code class="docutils literal notranslate"><span class="pre">contrib_CTCLoss</span></code> is deprecated.</p> |
| </div> |
| <p>The shapes of the inputs and outputs:</p> |
| <ul class="simple"> |
| <li><p><strong>data</strong>: <cite>(sequence_length, batch_size, alphabet_size)</cite></p></li> |
| <li><p><strong>label</strong>: <cite>(batch_size, label_sequence_length)</cite></p></li> |
| <li><p><strong>out</strong>: <cite>(batch_size)</cite></p></li> |
| </ul> |
| <p>The <cite>data</cite> tensor consists of sequences of activation vectors (without applying softmax), |
| with i-th channel in the last dimension corresponding to i-th label |
| for i between 0 and alphabet_size-1 (i.e always 0-indexed). |
| Alphabet size should include one additional value reserved for blank label. |
| When <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"first"</span></code>, the <code class="docutils literal notranslate"><span class="pre">0</span></code>-th channel is be reserved for |
| activation of blank label, or otherwise if it is “last”, <code class="docutils literal notranslate"><span class="pre">(alphabet_size-1)</span></code>-th channel should be |
| reserved for blank label.</p> |
| <p><code class="docutils literal notranslate"><span class="pre">label</span></code> is an index matrix of integers. When <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"first"</span></code>, |
| the value 0 is then reserved for blank label, and should not be passed in this matrix. Otherwise, |
| when <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"last"</span></code>, the value <cite>(alphabet_size-1)</cite> is reserved for blank label.</p> |
| <p>If a sequence of labels is shorter than <em>label_sequence_length</em>, use the special |
| padding value at the end of the sequence to conform it to the correct |
| length. The padding value is <cite>0</cite> when <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"first"</span></code>, and <cite>-1</cite> otherwise.</p> |
| <p>For example, suppose the vocabulary is <cite>[a, b, c]</cite>, and in one batch we have three sequences |
| ‘ba’, ‘cbb’, and ‘abac’. When <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"first"</span></code>, we can index the labels as |
| <cite>{‘a’: 1, ‘b’: 2, ‘c’: 3}</cite>, and we reserve the 0-th channel for blank label in data tensor. |
| The resulting <cite>label</cite> tensor should be padded to be:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mi">3</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">0</span><span class="p">],</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">1</span><span class="p">,</span> <span class="mi">3</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>When <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"last"</span></code>, we can index the labels as |
| <cite>{‘a’: 0, ‘b’: 1, ‘c’: 2}</cite>, and we reserve the channel index 3 for blank label in data tensor. |
| The resulting <cite>label</cite> tensor should be padded to be:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">1</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="p">[</span><span class="mi">2</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="o">-</span><span class="mi">1</span><span class="p">],</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">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p><code class="docutils literal notranslate"><span class="pre">out</span></code> is a list of CTC loss values, one per example in the batch.</p> |
| <p>See <em>Connectionist Temporal Classification: Labelling Unsegmented |
| Sequence Data with Recurrent Neural Networks</em>, A. Graves <em>et al</em>. for more |
| information on the definition and the algorithm.</p> |
| <p>Defined in src/operator/nn/ctc_loss.cc:L100</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input ndarray</p></li> |
| <li><p><strong>label</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Ground-truth labels for the loss.</p></li> |
| <li><p><strong>data_lengths</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lengths of data for each of the samples. Only required when use_data_lengths is true.</p></li> |
| <li><p><strong>label_lengths</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lengths of labels for each of the samples. Only required when use_label_lengths is true.</p></li> |
| <li><p><strong>use_data_lengths</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether the data lenghts are decided by <cite>data_lengths</cite>. If false, the lengths are equal to the max sequence length.</p></li> |
| <li><p><strong>use_label_lengths</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether the label lenghts are decided by <cite>label_lengths</cite>, or derived from <cite>padding_mask</cite>. If false, the lengths are derived from the first occurrence of the value of <cite>padding_mask</cite>. The value of <cite>padding_mask</cite> is <code class="docutils literal notranslate"><span class="pre">0</span></code> when first CTC label is reserved for blank, and <code class="docutils literal notranslate"><span class="pre">-1</span></code> when last label is reserved for blank. See <cite>blank_label</cite>.</p></li> |
| <li><p><strong>blank_label</strong> (<em>{'first'</em><em>, </em><em>'last'}</em><em>,</em><em>optional</em><em>, </em><em>default='first'</em>) – Set the label that is reserved for blank label.If “first”, 0-th label is reserved, and label values for tokens in the vocabulary are between <code class="docutils literal notranslate"><span class="pre">1</span></code> and <code class="docutils literal notranslate"><span class="pre">alphabet_size-1</span></code>, and the padding mask is <code class="docutils literal notranslate"><span class="pre">-1</span></code>. If “last”, last label value <code class="docutils literal notranslate"><span class="pre">alphabet_size-1</span></code> is reserved for blank label instead, and label values for tokens in the vocabulary are between <code class="docutils literal notranslate"><span class="pre">0</span></code> and <code class="docutils literal notranslate"><span class="pre">alphabet_size-2</span></code>, and the padding mask is <code class="docutils literal notranslate"><span class="pre">0</span></code>.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Cast"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Cast</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Cast" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Casts all elements of the input to a new type.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><code class="docutils literal notranslate"><span class="pre">Cast</span></code> is deprecated. Use <code class="docutils literal notranslate"><span class="pre">cast</span></code> instead.</p> |
| </div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">cast</span><span class="p">([</span><span class="mf">0.9</span><span class="p">,</span> <span class="mf">1.3</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="s1">'int32'</span><span class="p">)</span> <span class="o">=</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="n">cast</span><span class="p">([</span><span class="mf">1e20</span><span class="p">,</span> <span class="mf">11.1</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="s1">'float16'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="n">inf</span><span class="p">,</span> <span class="mf">11.09375</span><span class="p">]</span> |
| <span class="n">cast</span><span class="p">([</span><span class="mi">300</span><span class="p">,</span> <span class="mf">11.1</span><span class="p">,</span> <span class="mf">10.9</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">3</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="s1">'uint8'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mi">44</span><span class="p">,</span> <span class="mi">11</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">255</span><span class="p">,</span> <span class="mi">253</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L664</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'bfloat16'</em><em>, </em><em>'bool'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'</em><em>, </em><em>'int8'</em><em>, </em><em>'uint8'}</em><em>, </em><em>required</em>) – Output data type.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Concat"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Concat</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Concat" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Joins input arrays along a given axis.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><cite>Concat</cite> is deprecated. Use <cite>concat</cite> instead.</p> |
| </div> |
| <p>The dimensions of the input arrays should be the same except the axis along |
| which they will be concatenated. |
| The dimension of the output array along the concatenated axis will be equal |
| to the sum of the corresponding dimensions of the input arrays.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">concat</span></code> output depends on storage types of inputs</p> |
| <ul class="simple"> |
| <li><p>concat(csr, csr, …, csr, dim=0) = csr</p></li> |
| <li><p>otherwise, <code class="docutils literal notranslate"><span class="pre">concat</span></code> generates output with default storage</p></li> |
| </ul> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="n">y</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">3</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">4</span><span class="p">],[</span><span class="mi">5</span><span class="p">,</span><span class="mi">5</span><span class="p">]]</span> |
| <span class="n">z</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">6</span><span class="p">,</span><span class="mi">6</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span><span class="mi">7</span><span class="p">],[</span><span class="mi">8</span><span class="p">,</span><span class="mi">8</span><span class="p">]]</span> |
| |
| <span class="n">concat</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">y</span><span class="p">,</span><span class="n">z</span><span class="p">,</span><span class="n">dim</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]</span> |
| |
| <span class="n">Note</span> <span class="n">that</span> <span class="n">you</span> <span class="n">cannot</span> <span class="n">concat</span> <span class="n">x</span><span class="p">,</span><span class="n">y</span><span class="p">,</span><span class="n">z</span> <span class="n">along</span> <span class="n">dimension</span> <span class="mi">1</span> <span class="n">since</span> <span class="n">dimension</span> |
| <span class="mi">0</span> <span class="ow">is</span> <span class="ow">not</span> <span class="n">the</span> <span class="n">same</span> <span class="k">for</span> <span class="nb">all</span> <span class="n">the</span> <span class="nb">input</span> <span class="n">arrays</span><span class="o">.</span> |
| |
| <span class="n">concat</span><span class="p">(</span><span class="n">y</span><span class="p">,</span><span class="n">z</span><span class="p">,</span><span class="n">dim</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/nn/concat.cc:L384 |
| This function support variable length of positional input.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – List of arrays to concatenate</p></li> |
| <li><p><strong>dim</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – the dimension to be concated.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <p>Concat two (or more) inputs along a specific dimension:</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">a</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'a'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">b</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'b'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">c</span> <span class="o">=</span> <span class="n">Concat</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">dim</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'my-concat'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">c</span> |
| <span class="go"><Symbol my-concat></span> |
| <span class="gp">>>> </span><span class="n">SymbolDoc</span><span class="o">.</span><span class="n">get_output_shape</span><span class="p">(</span><span class="n">c</span><span class="p">,</span> <span class="n">a</span><span class="o">=</span><span class="p">(</span><span class="mi">128</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">),</span> <span class="n">b</span><span class="o">=</span><span class="p">(</span><span class="mi">128</span><span class="p">,</span> <span class="mi">15</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">))</span> |
| <span class="go">{'my-concat_output': (128L, 25L, 3L, 3L)}</span> |
| </pre></div> |
| </div> |
| <p>Note the shape should be the same except on the dimension that is being |
| concatenated.</p> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Convolution"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Convolution</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">weight=None</em>, <em class="sig-param">bias=None</em>, <em class="sig-param">kernel=_Null</em>, <em class="sig-param">stride=_Null</em>, <em class="sig-param">dilate=_Null</em>, <em class="sig-param">pad=_Null</em>, <em class="sig-param">num_filter=_Null</em>, <em class="sig-param">num_group=_Null</em>, <em class="sig-param">workspace=_Null</em>, <em class="sig-param">no_bias=_Null</em>, <em class="sig-param">cudnn_tune=_Null</em>, <em class="sig-param">cudnn_off=_Null</em>, <em class="sig-param">layout=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Convolution" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Compute <em>N</em>-D convolution on <em>(N+2)</em>-D input.</p> |
| <p>In the 2-D convolution, given input data with shape <em>(batch_size, |
| channel, height, width)</em>, the output is computed by</p> |
| <div class="math notranslate nohighlight"> |
| \[out[n,i,:,:] = bias[i] + \sum_{j=0}^{channel} data[n,j,:,:] \star |
| weight[i,j,:,:]\]</div> |
| <p>where <span class="math notranslate nohighlight">\(\star\)</span> is the 2-D cross-correlation operator.</p> |
| <p>For general 2-D convolution, the shapes are</p> |
| <ul class="simple"> |
| <li><p><strong>data</strong>: <em>(batch_size, channel, height, width)</em></p></li> |
| <li><p><strong>weight</strong>: <em>(num_filter, channel, kernel[0], kernel[1])</em></p></li> |
| <li><p><strong>bias</strong>: <em>(num_filter,)</em></p></li> |
| <li><p><strong>out</strong>: <em>(batch_size, num_filter, out_height, out_width)</em>.</p></li> |
| </ul> |
| <p>Define:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">f</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">k</span><span class="p">,</span><span class="n">p</span><span class="p">,</span><span class="n">s</span><span class="p">,</span><span class="n">d</span><span class="p">)</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">x</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">p</span><span class="o">-</span><span class="n">d</span><span class="o">*</span><span class="p">(</span><span class="n">k</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">s</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>then we have:</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">f</span><span class="p">(</span><span class="n">height</span><span class="p">,</span> <span class="n">kernel</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">pad</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">stride</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">dilate</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="n">f</span><span class="p">(</span><span class="n">width</span><span class="p">,</span> <span class="n">kernel</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">pad</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">stride</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">dilate</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span> |
| </pre></div> |
| </div> |
| <p>If <code class="docutils literal notranslate"><span class="pre">no_bias</span></code> is set to be true, then the <code class="docutils literal notranslate"><span class="pre">bias</span></code> term is ignored.</p> |
| <p>The default data <code class="docutils literal notranslate"><span class="pre">layout</span></code> is <em>NCHW</em>, namely <em>(batch_size, channel, height, |
| width)</em>. We can choose other layouts such as <em>NWC</em>.</p> |
| <p>If <code class="docutils literal notranslate"><span class="pre">num_group</span></code> is larger than 1, denoted by <em>g</em>, then split the input <code class="docutils literal notranslate"><span class="pre">data</span></code> |
| evenly into <em>g</em> parts along the channel axis, and also evenly split <code class="docutils literal notranslate"><span class="pre">weight</span></code> |
| along the first dimension. Next compute the convolution on the <em>i</em>-th part of |
| the data with the <em>i</em>-th weight part. The output is obtained by concatenating all |
| the <em>g</em> results.</p> |
| <p>1-D convolution does not have <em>height</em> dimension but only <em>width</em> in space.</p> |
| <ul class="simple"> |
| <li><p><strong>data</strong>: <em>(batch_size, channel, width)</em></p></li> |
| <li><p><strong>weight</strong>: <em>(num_filter, channel, kernel[0])</em></p></li> |
| <li><p><strong>bias</strong>: <em>(num_filter,)</em></p></li> |
| <li><p><strong>out</strong>: <em>(batch_size, num_filter, out_width)</em>.</p></li> |
| </ul> |
| <p>3-D convolution adds an additional <em>depth</em> dimension besides <em>height</em> and |
| <em>width</em>. The shapes are</p> |
| <ul class="simple"> |
| <li><p><strong>data</strong>: <em>(batch_size, channel, depth, height, width)</em></p></li> |
| <li><p><strong>weight</strong>: <em>(num_filter, channel, kernel[0], kernel[1], kernel[2])</em></p></li> |
| <li><p><strong>bias</strong>: <em>(num_filter,)</em></p></li> |
| <li><p><strong>out</strong>: <em>(batch_size, num_filter, out_depth, out_height, out_width)</em>.</p></li> |
| </ul> |
| <p>Both <code class="docutils literal notranslate"><span class="pre">weight</span></code> and <code class="docutils literal notranslate"><span class="pre">bias</span></code> are learnable parameters.</p> |
| <p>There are other options to tune the performance.</p> |
| <ul class="simple"> |
| <li><p><strong>cudnn_tune</strong>: enable this option leads to higher startup time but may give |
| faster speed. Options are</p> |
| <ul> |
| <li><p><strong>off</strong>: no tuning</p></li> |
| <li><p><strong>limited_workspace</strong>:run test and pick the fastest algorithm that doesn’t |
| exceed workspace limit.</p></li> |
| <li><p><strong>fastest</strong>: pick the fastest algorithm and ignore workspace limit.</p></li> |
| <li><p><strong>None</strong> (default): the behavior is determined by environment variable |
| <code class="docutils literal notranslate"><span class="pre">MXNET_CUDNN_AUTOTUNE_DEFAULT</span></code>. 0 for off, 1 for limited workspace |
| (default), 2 for fastest.</p></li> |
| </ul> |
| </li> |
| <li><p><strong>workspace</strong>: A large number leads to more (GPU) memory usage but may improve |
| the performance.</p></li> |
| </ul> |
| <p>Defined in src/operator/nn/convolution.cc:L475</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the ConvolutionOp.</p></li> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight matrix.</p></li> |
| <li><p><strong>bias</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Bias parameter.</p></li> |
| <li><p><strong>kernel</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – Convolution kernel size: (w,), (h, w) or (d, h, w)</p></li> |
| <li><p><strong>stride</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Convolution stride: (w,), (h, w) or (d, h, w). Defaults to 1 for each dimension.</p></li> |
| <li><p><strong>dilate</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Convolution dilate: (w,), (h, w) or (d, h, w). Defaults to 1 for each dimension.</p></li> |
| <li><p><strong>pad</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Zero pad for convolution: (w,), (h, w) or (d, h, w). Defaults to no padding.</p></li> |
| <li><p><strong>num_filter</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>required</em>) – Convolution filter(channel) number</p></li> |
| <li><p><strong>num_group</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Number of group partitions.</p></li> |
| <li><p><strong>workspace</strong> (<em>long</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=1024</em>) – Maximum temporary workspace allowed (MB) in convolution.This parameter has two usages. When CUDNN is not used, it determines the effective batch size of the convolution kernel. When CUDNN is used, it controls the maximum temporary storage used for tuning the best CUDNN kernel when <cite>limited_workspace</cite> strategy is used.</p></li> |
| <li><p><strong>no_bias</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to disable bias parameter.</p></li> |
| <li><p><strong>cudnn_tune</strong> (<em>{None</em><em>, </em><em>'fastest'</em><em>, </em><em>'limited_workspace'</em><em>, </em><em>'off'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – Whether to pick convolution algo by running performance test.</p></li> |
| <li><p><strong>cudnn_off</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Turn off cudnn for this layer.</p></li> |
| <li><p><strong>layout</strong> (<em>{None</em><em>, </em><em>'NCDHW'</em><em>, </em><em>'NCHW'</em><em>, </em><em>'NCW'</em><em>, </em><em>'NDHWC'</em><em>, </em><em>'NHWC'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – Set layout for input, output and weight. Empty for |
| default layout: NCW for 1d, NCHW for 2d and NCDHW for 3d.NHWC and NDHWC are only supported on GPU.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Convolution_v1"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Convolution_v1</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">weight=None</em>, <em class="sig-param">bias=None</em>, <em class="sig-param">kernel=_Null</em>, <em class="sig-param">stride=_Null</em>, <em class="sig-param">dilate=_Null</em>, <em class="sig-param">pad=_Null</em>, <em class="sig-param">num_filter=_Null</em>, <em class="sig-param">num_group=_Null</em>, <em class="sig-param">workspace=_Null</em>, <em class="sig-param">no_bias=_Null</em>, <em class="sig-param">cudnn_tune=_Null</em>, <em class="sig-param">cudnn_off=_Null</em>, <em class="sig-param">layout=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Convolution_v1" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This operator is DEPRECATED. Apply convolution to input then add a bias.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the ConvolutionV1Op.</p></li> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight matrix.</p></li> |
| <li><p><strong>bias</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Bias parameter.</p></li> |
| <li><p><strong>kernel</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – convolution kernel size: (h, w) or (d, h, w)</p></li> |
| <li><p><strong>stride</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – convolution stride: (h, w) or (d, h, w)</p></li> |
| <li><p><strong>dilate</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – convolution dilate: (h, w) or (d, h, w)</p></li> |
| <li><p><strong>pad</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – pad for convolution: (h, w) or (d, h, w)</p></li> |
| <li><p><strong>num_filter</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>required</em>) – convolution filter(channel) number</p></li> |
| <li><p><strong>num_group</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Number of group partitions. Equivalent to slicing input into num_group |
| partitions, apply convolution on each, then concatenate the results</p></li> |
| <li><p><strong>workspace</strong> (<em>long</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=1024</em>) – Maximum temporary workspace allowed for convolution (MB).This parameter determines the effective batch size of the convolution kernel, which may be smaller than the given batch size. Also, the workspace will be automatically enlarged to make sure that we can run the kernel with batch_size=1</p></li> |
| <li><p><strong>no_bias</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to disable bias parameter.</p></li> |
| <li><p><strong>cudnn_tune</strong> (<em>{None</em><em>, </em><em>'fastest'</em><em>, </em><em>'limited_workspace'</em><em>, </em><em>'off'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – Whether to pick convolution algo by running performance test. |
| Leads to higher startup time but may give faster speed. Options are: |
| ‘off’: no tuning |
| ‘limited_workspace’: run test and pick the fastest algorithm that doesn’t exceed workspace limit. |
| ‘fastest’: pick the fastest algorithm and ignore workspace limit. |
| If set to None (default), behavior is determined by environment |
| variable MXNET_CUDNN_AUTOTUNE_DEFAULT: 0 for off, |
| 1 for limited workspace (default), 2 for fastest.</p></li> |
| <li><p><strong>cudnn_off</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Turn off cudnn for this layer.</p></li> |
| <li><p><strong>layout</strong> (<em>{None</em><em>, </em><em>'NCDHW'</em><em>, </em><em>'NCHW'</em><em>, </em><em>'NDHWC'</em><em>, </em><em>'NHWC'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – Set layout for input, output and weight. Empty for |
| default layout: NCHW for 2d and NCDHW for 3d.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Correlation"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Correlation</code><span class="sig-paren">(</span><em class="sig-param">data1=None</em>, <em class="sig-param">data2=None</em>, <em class="sig-param">kernel_size=_Null</em>, <em class="sig-param">max_displacement=_Null</em>, <em class="sig-param">stride1=_Null</em>, <em class="sig-param">stride2=_Null</em>, <em class="sig-param">pad_size=_Null</em>, <em class="sig-param">is_multiply=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Correlation" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies correlation to inputs.</p> |
| <p>The correlation layer performs multiplicative patch comparisons between two feature maps.</p> |
| <p>Given two multi-channel feature maps <span class="math notranslate nohighlight">\(f_{1}, f_{2}\)</span>, with <span class="math notranslate nohighlight">\(w\)</span>, <span class="math notranslate nohighlight">\(h\)</span>, and <span class="math notranslate nohighlight">\(c\)</span> being their width, height, and number of channels, |
| the correlation layer lets the network compare each patch from <span class="math notranslate nohighlight">\(f_{1}\)</span> with each patch from <span class="math notranslate nohighlight">\(f_{2}\)</span>.</p> |
| <p>For now we consider only a single comparison of two patches. The ‘correlation’ of two patches centered at <span class="math notranslate nohighlight">\(x_{1}\)</span> in the first map and |
| <span class="math notranslate nohighlight">\(x_{2}\)</span> in the second map is then defined as:</p> |
| <div class="math notranslate nohighlight"> |
| \[c(x_{1}, x_{2}) = \sum_{o \in [-k,k] \times [-k,k]} <f_{1}(x_{1} + o), f_{2}(x_{2} + o)>\]</div> |
| <p>for a square patch of size <span class="math notranslate nohighlight">\(K:=2k+1\)</span>.</p> |
| <p>Note that the equation above is identical to one step of a convolution in neural networks, but instead of convolving data with a filter, it convolves data with other |
| data. For this reason, it has no training weights.</p> |
| <p>Computing <span class="math notranslate nohighlight">\(c(x_{1}, x_{2})\)</span> involves <span class="math notranslate nohighlight">\(c * K^{2}\)</span> multiplications. Comparing all patch combinations involves <span class="math notranslate nohighlight">\(w^{2}*h^{2}\)</span> such computations.</p> |
| <p>Given a maximum displacement <span class="math notranslate nohighlight">\(d\)</span>, for each location <span class="math notranslate nohighlight">\(x_{1}\)</span> it computes correlations <span class="math notranslate nohighlight">\(c(x_{1}, x_{2})\)</span> only in a neighborhood of size <span class="math notranslate nohighlight">\(D:=2d+1\)</span>, |
| by limiting the range of <span class="math notranslate nohighlight">\(x_{2}\)</span>. We use strides <span class="math notranslate nohighlight">\(s_{1}, s_{2}\)</span>, to quantize <span class="math notranslate nohighlight">\(x_{1}\)</span> globally and to quantize <span class="math notranslate nohighlight">\(x_{2}\)</span> within the neighborhood |
| centered around <span class="math notranslate nohighlight">\(x_{1}\)</span>.</p> |
| <p>The final output is defined by the following expression:</p> |
| <div class="math notranslate nohighlight"> |
| \[out[n, q, i, j] = c(x_{i, j}, x_{q})\]</div> |
| <p>where <span class="math notranslate nohighlight">\(i\)</span> and <span class="math notranslate nohighlight">\(j\)</span> enumerate spatial locations in <span class="math notranslate nohighlight">\(f_{1}\)</span>, and <span class="math notranslate nohighlight">\(q\)</span> denotes the <span class="math notranslate nohighlight">\(q^{th}\)</span> neighborhood of <span class="math notranslate nohighlight">\(x_{i,j}\)</span>.</p> |
| <p>Defined in src/operator/correlation.cc:L197</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data1</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data1 to the correlation.</p></li> |
| <li><p><strong>data2</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data2 to the correlation.</p></li> |
| <li><p><strong>kernel_size</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – kernel size for Correlation must be an odd number</p></li> |
| <li><p><strong>max_displacement</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Max displacement of Correlation</p></li> |
| <li><p><strong>stride1</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – stride1 quantize data1 globally</p></li> |
| <li><p><strong>stride2</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – stride2 quantize data2 within the neighborhood centered around data1</p></li> |
| <li><p><strong>pad_size</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – pad for Correlation</p></li> |
| <li><p><strong>is_multiply</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – operation type is either multiplication or subduction</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Crop"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Crop</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Crop" title="Permalink to this definition">¶</a></dt> |
| <dd><div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><cite>Crop</cite> is deprecated. Use <cite>slice</cite> instead.</p> |
| </div> |
| <p>Crop the 2nd and 3rd dim of input data, with the corresponding size of h_w or |
| with width and height of the second input symbol, i.e., with one input, we need h_w to |
| specify the crop height and width, otherwise the second input symbol’s size will be used</p> |
| <p>Defined in src/operator/crop.cc:L49 |
| This function support variable length of positional input.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em> or </em><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Tensor or List of Tensors, the second input will be used as crop_like shape reference</p></li> |
| <li><p><strong>offset</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>0</em><em>,</em><em>0</em><em>]</em>) – crop offset coordinate: (y, x)</p></li> |
| <li><p><strong>h_w</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>0</em><em>,</em><em>0</em><em>]</em>) – crop height and width: (h, w)</p></li> |
| <li><p><strong>center_crop</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to true, then it will use be the center_crop,or it will crop using the shape of crop_like</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Custom"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Custom</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Custom" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Apply a custom operator implemented in a frontend language (like Python).</p> |
| <p>Custom operators should override required methods like <cite>forward</cite> and <cite>backward</cite>. |
| The custom operator must be registered before it can be used. |
| Please check the tutorial here: <a class="reference external" href="https://mxnet.incubator.apache.org/api/faq/new_op">https://mxnet.incubator.apache.org/api/faq/new_op</a></p> |
| <p>Defined in src/operator/custom/custom.cc:L546</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Input data for the custom operator.</p></li> |
| <li><p><strong>op_type</strong> (<em>string</em>) – Name of the custom operator. This is the name that is passed to <cite>mx.operator.register</cite> to register the operator.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Deconvolution"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Deconvolution</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">weight=None</em>, <em class="sig-param">bias=None</em>, <em class="sig-param">kernel=_Null</em>, <em class="sig-param">stride=_Null</em>, <em class="sig-param">dilate=_Null</em>, <em class="sig-param">pad=_Null</em>, <em class="sig-param">adj=_Null</em>, <em class="sig-param">target_shape=_Null</em>, <em class="sig-param">num_filter=_Null</em>, <em class="sig-param">num_group=_Null</em>, <em class="sig-param">workspace=_Null</em>, <em class="sig-param">no_bias=_Null</em>, <em class="sig-param">cudnn_tune=_Null</em>, <em class="sig-param">cudnn_off=_Null</em>, <em class="sig-param">layout=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Deconvolution" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes 1D or 2D transposed convolution (aka fractionally strided convolution) of the input tensor. This operation can be seen as the gradient of Convolution operation with respect to its input. Convolution usually reduces the size of the input. Transposed convolution works the other way, going from a smaller input to a larger output while preserving the connectivity pattern.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input tensor to the deconvolution operation.</p></li> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weights representing the kernel.</p></li> |
| <li><p><strong>bias</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Bias added to the result after the deconvolution operation.</p></li> |
| <li><p><strong>kernel</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – Deconvolution kernel size: (w,), (h, w) or (d, h, w). This is same as the kernel size used for the corresponding convolution</p></li> |
| <li><p><strong>stride</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The stride used for the corresponding convolution: (w,), (h, w) or (d, h, w). Defaults to 1 for each dimension.</p></li> |
| <li><p><strong>dilate</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Dilation factor for each dimension of the input: (w,), (h, w) or (d, h, w). Defaults to 1 for each dimension.</p></li> |
| <li><p><strong>pad</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The amount of implicit zero padding added during convolution for each dimension of the input: (w,), (h, w) or (d, h, w). <code class="docutils literal notranslate"><span class="pre">(kernel-1)/2</span></code> is usually a good choice. If <cite>target_shape</cite> is set, <cite>pad</cite> will be ignored and a padding that will generate the target shape will be used. Defaults to no padding.</p></li> |
| <li><p><strong>adj</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Adjustment for output shape: (w,), (h, w) or (d, h, w). If <cite>target_shape</cite> is set, <cite>adj</cite> will be ignored and computed accordingly.</p></li> |
| <li><p><strong>target_shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Shape of the output tensor: (w,), (h, w) or (d, h, w).</p></li> |
| <li><p><strong>num_filter</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>required</em>) – Number of output filters.</p></li> |
| <li><p><strong>num_group</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Number of groups partition.</p></li> |
| <li><p><strong>workspace</strong> (<em>long</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=512</em>) – Maximum temporary workspace allowed (MB) in deconvolution.This parameter has two usages. When CUDNN is not used, it determines the effective batch size of the deconvolution kernel. When CUDNN is used, it controls the maximum temporary storage used for tuning the best CUDNN kernel when <cite>limited_workspace</cite> strategy is used.</p></li> |
| <li><p><strong>no_bias</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Whether to disable bias parameter.</p></li> |
| <li><p><strong>cudnn_tune</strong> (<em>{None</em><em>, </em><em>'fastest'</em><em>, </em><em>'limited_workspace'</em><em>, </em><em>'off'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – Whether to pick convolution algorithm by running performance test.</p></li> |
| <li><p><strong>cudnn_off</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Turn off cudnn for this layer.</p></li> |
| <li><p><strong>layout</strong> (<em>{None</em><em>, </em><em>'NCDHW'</em><em>, </em><em>'NCHW'</em><em>, </em><em>'NCW'</em><em>, </em><em>'NDHWC'</em><em>, </em><em>'NHWC'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – Set layout for input, output and weight. Empty for default layout, NCW for 1d, NCHW for 2d and NCDHW for 3d.NHWC and NDHWC are only supported on GPU.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Dropout"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Dropout</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">p=_Null</em>, <em class="sig-param">mode=_Null</em>, <em class="sig-param">axes=_Null</em>, <em class="sig-param">cudnn_off=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Dropout" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies dropout operation to input array.</p> |
| <ul class="simple"> |
| <li><p>During training, each element of the input is set to zero with probability p. |
| The whole array is rescaled by <span class="math notranslate nohighlight">\(1/(1-p)\)</span> to keep the expected |
| sum of the input unchanged.</p></li> |
| <li><p>During testing, this operator does not change the input if mode is ‘training’. |
| If mode is ‘always’, the same computaion as during training will be applied.</p></li> |
| </ul> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">random</span><span class="o">.</span><span class="n">seed</span><span class="p">(</span><span class="mi">998</span><span class="p">)</span> |
| <span class="n">input_array</span> <span class="o">=</span> <span class="n">array</span><span class="p">([[</span><span class="mf">3.</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">2.</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.4</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">]])</span> |
| <span class="n">a</span> <span class="o">=</span> <span class="n">symbol</span><span class="o">.</span><span class="n">Variable</span><span class="p">(</span><span class="s1">'a'</span><span class="p">)</span> |
| <span class="n">dropout</span> <span class="o">=</span> <span class="n">symbol</span><span class="o">.</span><span class="n">Dropout</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">p</span> <span class="o">=</span> <span class="mf">0.2</span><span class="p">)</span> |
| <span class="n">executor</span> <span class="o">=</span> <span class="n">dropout</span><span class="o">.</span><span class="n">simple_bind</span><span class="p">(</span><span class="n">a</span> <span class="o">=</span> <span class="n">input_array</span><span class="o">.</span><span class="n">shape</span><span class="p">)</span> |
| |
| <span class="c1">## If training</span> |
| <span class="n">executor</span><span class="o">.</span><span class="n">forward</span><span class="p">(</span><span class="n">is_train</span> <span class="o">=</span> <span class="kc">True</span><span class="p">,</span> <span class="n">a</span> <span class="o">=</span> <span class="n">input_array</span><span class="p">)</span> |
| <span class="n">executor</span><span class="o">.</span><span class="n">outputs</span> |
| <span class="p">[[</span> <span class="mf">3.75</span> <span class="mf">0.625</span> <span class="o">-</span><span class="mf">0.</span> <span class="mf">2.5</span> <span class="mf">8.75</span> <span class="p">]</span> |
| <span class="p">[</span> <span class="mf">2.5</span> <span class="o">-</span><span class="mf">0.5</span> <span class="mf">8.75</span> <span class="mf">3.75</span> <span class="mf">0.</span> <span class="p">]]</span> |
| |
| <span class="c1">## If testing</span> |
| <span class="n">executor</span><span class="o">.</span><span class="n">forward</span><span class="p">(</span><span class="n">is_train</span> <span class="o">=</span> <span class="kc">False</span><span class="p">,</span> <span class="n">a</span> <span class="o">=</span> <span class="n">input_array</span><span class="p">)</span> |
| <span class="n">executor</span><span class="o">.</span><span class="n">outputs</span> |
| <span class="p">[[</span> <span class="mf">3.</span> <span class="mf">0.5</span> <span class="o">-</span><span class="mf">0.5</span> <span class="mf">2.</span> <span class="mf">7.</span> <span class="p">]</span> |
| <span class="p">[</span> <span class="mf">2.</span> <span class="o">-</span><span class="mf">0.4</span> <span class="mf">7.</span> <span class="mf">3.</span> <span class="mf">0.2</span> <span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/nn/dropout.cc:L95</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array to which dropout will be applied.</p></li> |
| <li><p><strong>p</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.5</em>) – Fraction of the input that gets dropped out during training time.</p></li> |
| <li><p><strong>mode</strong> (<em>{'always'</em><em>, </em><em>'training'}</em><em>,</em><em>optional</em><em>, </em><em>default='training'</em>) – Whether to only turn on dropout during training or to also turn on for inference.</p></li> |
| <li><p><strong>axes</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Axes for variational dropout kernel.</p></li> |
| <li><p><strong>cudnn_off</strong> (<em>boolean</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to turn off cudnn in dropout operator. This option is ignored if axes is specified.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <p>Apply dropout to corrupt input as zero with probability 0.2:</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">data</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'data'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">data_dp</span> <span class="o">=</span> <span class="n">Dropout</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">data</span><span class="p">,</span> <span class="n">p</span><span class="o">=</span><span class="mf">0.2</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">100</span><span class="p">,</span> <span class="mi">100</span><span class="p">)</span> <span class="c1"># take larger shapes to be more statistical stable</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="n">shape</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">op</span> <span class="o">=</span> <span class="n">Dropout</span><span class="p">(</span><span class="n">p</span><span class="o">=</span><span class="mf">0.5</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'dp'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="c1"># dropout is identity during testing</span> |
| <span class="gp">>>> </span><span class="n">y</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">simple_forward</span><span class="p">(</span><span class="n">op</span><span class="p">,</span> <span class="n">dp_data</span><span class="o">=</span><span class="n">x</span><span class="p">,</span> <span class="n">is_train</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">test_utils</span><span class="o">.</span><span class="n">almost_equal</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> |
| <span class="go">True</span> |
| <span class="gp">>>> </span><span class="n">y</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">simple_forward</span><span class="p">(</span><span class="n">op</span><span class="p">,</span> <span class="n">dp_data</span><span class="o">=</span><span class="n">x</span><span class="p">,</span> <span class="n">is_train</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="c1"># expectation is (approximately) unchanged</span> |
| <span class="gp">>>> </span><span class="n">np</span><span class="o">.</span><span class="n">abs</span><span class="p">(</span><span class="n">x</span><span class="o">.</span><span class="n">mean</span><span class="p">()</span> <span class="o">-</span> <span class="n">y</span><span class="o">.</span><span class="n">mean</span><span class="p">())</span> <span class="o"><</span> <span class="mf">0.1</span> |
| <span class="go">True</span> |
| <span class="gp">>>> </span><span class="nb">set</span><span class="p">(</span><span class="n">np</span><span class="o">.</span><span class="n">unique</span><span class="p">(</span><span class="n">y</span><span class="p">))</span> <span class="o">==</span> <span class="nb">set</span><span class="p">([</span><span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">])</span> |
| <span class="go">True</span> |
| </pre></div> |
| </div> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.ElementWiseSum"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">ElementWiseSum</code><span class="sig-paren">(</span><em class="sig-param">*args</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.ElementWiseSum" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Adds all input arguments element-wise.</p> |
| <div class="math notranslate nohighlight"> |
| \[add\_n(a_1, a_2, ..., a_n) = a_1 + a_2 + ... + a_n\]</div> |
| <p><code class="docutils literal notranslate"><span class="pre">add_n</span></code> is potentially more efficient than calling <code class="docutils literal notranslate"><span class="pre">add</span></code> by <cite>n</cite> times.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">add_n</span></code> output depends on storage types of inputs</p> |
| <ul class="simple"> |
| <li><p>add_n(row_sparse, row_sparse, ..) = row_sparse</p></li> |
| <li><p>add_n(default, csr, default) = default</p></li> |
| <li><p>add_n(any input combinations longer than 4 (>4) with at least one default type) = default</p></li> |
| <li><p>otherwise, <code class="docutils literal notranslate"><span class="pre">add_n</span></code> falls all inputs back to default storage and generates default storage</p></li> |
| </ul> |
| <p>Defined in src/operator/tensor/elemwise_sum.cc:L155 |
| This function support variable length of positional input.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>args</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Positional input arguments</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Embedding"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Embedding</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">weight=None</em>, <em class="sig-param">input_dim=_Null</em>, <em class="sig-param">output_dim=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">sparse_grad=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Embedding" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Maps integer indices to vector representations (embeddings).</p> |
| <p>This operator maps words to real-valued vectors in a high-dimensional space, |
| called word embeddings. These embeddings can capture semantic and syntactic properties of the words. |
| For example, it has been noted that in the learned embedding spaces, similar words tend |
| to be close to each other and dissimilar words far apart.</p> |
| <p>For an input array of shape (d1, …, dK), |
| the shape of an output array is (d1, …, dK, output_dim). |
| All the input values should be integers in the range [0, input_dim).</p> |
| <p>If the input_dim is ip0 and output_dim is op0, then shape of the embedding weight matrix must be |
| (ip0, op0).</p> |
| <p>When “sparse_grad” is False, if any index mentioned is too large, it is replaced by the index that |
| addresses the last vector in an embedding matrix. |
| When “sparse_grad” is True, an error will be raised if invalid indices are found.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">input_dim</span> <span class="o">=</span> <span class="mi">4</span> |
| <span class="n">output_dim</span> <span class="o">=</span> <span class="mi">5</span> |
| |
| <span class="o">//</span> <span class="n">Each</span> <span class="n">row</span> <span class="ow">in</span> <span class="n">weight</span> <span class="n">matrix</span> <span class="n">y</span> <span class="n">represents</span> <span class="n">a</span> <span class="n">word</span><span class="o">.</span> <span class="n">So</span><span class="p">,</span> <span class="n">y</span> <span class="o">=</span> <span class="p">(</span><span class="n">w0</span><span class="p">,</span><span class="n">w1</span><span class="p">,</span><span class="n">w2</span><span class="p">,</span><span class="n">w3</span><span class="p">)</span> |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">,</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">15.</span><span class="p">,</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">,</span> <span class="mf">19.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">Input</span> <span class="n">array</span> <span class="n">x</span> <span class="n">represents</span> <span class="n">n</span><span class="o">-</span><span class="n">grams</span><span class="p">(</span><span class="mi">2</span><span class="o">-</span><span class="n">gram</span><span class="p">)</span><span class="o">.</span> <span class="n">So</span><span class="p">,</span> <span class="n">x</span> <span class="o">=</span> <span class="p">[(</span><span class="n">w1</span><span class="p">,</span><span class="n">w3</span><span class="p">),</span> <span class="p">(</span><span class="n">w0</span><span class="p">,</span><span class="n">w2</span><span class="p">)]</span> |
| <span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">Mapped</span> <span class="nb">input</span> <span class="n">x</span> <span class="n">to</span> <span class="n">its</span> <span class="n">vector</span> <span class="n">representation</span> <span class="n">y</span><span class="o">.</span> |
| <span class="n">Embedding</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">5</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">15.</span><span class="p">,</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">,</span> <span class="mf">19.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">,</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of weight can be either row_sparse or default.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>If “sparse_grad” is set to True, the storage type of gradient w.r.t weights will be |
| “row_sparse”. Only a subset of optimizers support sparse gradients, including SGD, AdaGrad |
| and Adam. Note that 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.incubator.apache.org/api/python/optimization/optimization.html">https://mxnet.incubator.apache.org/api/python/optimization/optimization.html</a></p> |
| </div> |
| <p>Defined in src/operator/tensor/indexing_op.cc:L597</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array to the embedding operator.</p></li> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The embedding weight matrix.</p></li> |
| <li><p><strong>input_dim</strong> (<em>int</em><em>, </em><em>required</em>) – Vocabulary size of the input indices.</p></li> |
| <li><p><strong>output_dim</strong> (<em>int</em><em>, </em><em>required</em>) – Dimension of the embedding vectors.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'bfloat16'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'</em><em>, </em><em>'int8'</em><em>, </em><em>'uint8'}</em><em>,</em><em>optional</em><em>, </em><em>default='float32'</em>) – Data type of weight.</p></li> |
| <li><p><strong>sparse_grad</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Compute row sparse gradient in the backward calculation. If set to True, the grad’s storage type is row_sparse.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <p>Assume we want to map the 26 English alphabet letters to 16-dimensional |
| vectorial representations.</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">vocabulary_size</span> <span class="o">=</span> <span class="mi">26</span> |
| <span class="gp">>>> </span><span class="n">embed_dim</span> <span class="o">=</span> <span class="mi">16</span> |
| <span class="gp">>>> </span><span class="n">seq_len</span><span class="p">,</span> <span class="n">batch_size</span> <span class="o">=</span> <span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="mi">64</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="nb">input</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'letters'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">op</span> <span class="o">=</span> <span class="n">Embedding</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="nb">input</span><span class="p">,</span> <span class="n">input_dim</span><span class="o">=</span><span class="n">vocabulary_size</span><span class="p">,</span> <span class="n">output_dim</span><span class="o">=</span><span class="n">embed_dim</span><span class="p">,</span> |
| <span class="go">...name='embed')</span> |
| <span class="gp">>>> </span><span class="n">SymbolDoc</span><span class="o">.</span><span class="n">get_output_shape</span><span class="p">(</span><span class="n">op</span><span class="p">,</span> <span class="n">letters</span><span class="o">=</span><span class="p">(</span><span class="n">seq_len</span><span class="p">,</span> <span class="n">batch_size</span><span class="p">))</span> |
| <span class="go">{'embed_output': (10L, 64L, 16L)}</span> |
| </pre></div> |
| </div> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">vocab_size</span><span class="p">,</span> <span class="n">embed_dim</span> <span class="o">=</span> <span class="p">(</span><span class="mi">26</span><span class="p">,</span> <span class="mi">16</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">batch_size</span> <span class="o">=</span> <span class="mi">12</span> |
| <span class="gp">>>> </span><span class="n">word_vecs</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">random_arrays</span><span class="p">((</span><span class="n">vocab_size</span><span class="p">,</span> <span class="n">embed_dim</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">op</span> <span class="o">=</span> <span class="n">Embedding</span><span class="p">(</span><span class="n">name</span><span class="o">=</span><span class="s1">'embed'</span><span class="p">,</span> <span class="n">input_dim</span><span class="o">=</span><span class="n">vocab_size</span><span class="p">,</span> <span class="n">output_dim</span><span class="o">=</span><span class="n">embed_dim</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">x</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">choice</span><span class="p">(</span><span class="n">vocab_size</span><span class="p">,</span> <span class="n">batch_size</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">y</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">simple_forward</span><span class="p">(</span><span class="n">op</span><span class="p">,</span> <span class="n">embed_data</span><span class="o">=</span><span class="n">x</span><span class="p">,</span> <span class="n">embed_weight</span><span class="o">=</span><span class="n">word_vecs</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">y_np</span> <span class="o">=</span> <span class="n">word_vecs</span><span class="p">[</span><span class="n">x</span><span class="p">]</span> |
| <span class="gp">>>> </span><span class="n">test_utils</span><span class="o">.</span><span class="n">almost_equal</span><span class="p">(</span><span class="n">y</span><span class="p">,</span> <span class="n">y_np</span><span class="p">)</span> |
| <span class="go">True</span> |
| </pre></div> |
| </div> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Flatten"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Flatten</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Flatten" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Flattens the input array into a 2-D array by collapsing the higher dimensions. |
| .. note:: <cite>Flatten</cite> is deprecated. Use <cite>flatten</cite> instead. |
| For an input array with shape <code class="docutils literal notranslate"><span class="pre">(d1,</span> <span class="pre">d2,</span> <span class="pre">...,</span> <span class="pre">dk)</span></code>, <cite>flatten</cite> operation reshapes |
| the input array into an output array of shape <code class="docutils literal notranslate"><span class="pre">(d1,</span> <span class="pre">d2*...*dk)</span></code>. |
| Note that the behavior of this function is different from numpy.ndarray.flatten, |
| which behaves similar to mxnet.ndarray.reshape((-1,)). |
| Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</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="p">[</span><span class="mi">4</span><span class="p">,</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">7</span><span class="p">,</span><span class="mi">8</span><span class="p">,</span><span class="mi">9</span><span class="p">]</span> |
| <span class="p">],</span> |
| <span class="p">[</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="p">[</span><span class="mi">4</span><span class="p">,</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">7</span><span class="p">,</span><span class="mi">8</span><span class="p">,</span><span class="mi">9</span><span class="p">]</span> |
| <span class="p">]],</span> |
| <span class="n">flatten</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L249</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <p>Flatten is usually applied before <cite>FullyConnected</cite>, to reshape the 4D tensor |
| produced by convolutional layers to 2D matrix:</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">data</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'data'</span><span class="p">)</span> <span class="c1"># say this is 4D from some conv/pool</span> |
| <span class="gp">>>> </span><span class="n">flatten</span> <span class="o">=</span> <span class="n">Flatten</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">data</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'flat'</span><span class="p">)</span> <span class="c1"># now this is 2D</span> |
| <span class="gp">>>> </span><span class="n">SymbolDoc</span><span class="o">.</span><span class="n">get_output_shape</span><span class="p">(</span><span class="n">flatten</span><span class="p">,</span> <span class="n">data</span><span class="o">=</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="go">{'flat_output': (2L, 60L)}</span> |
| </pre></div> |
| </div> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">test_dims</span> <span class="o">=</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">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">),</span> <span class="p">(</span><span class="mi">2</span><span class="p">,)]</span> |
| <span class="gp">>>> </span><span class="n">op</span> <span class="o">=</span> <span class="n">Flatten</span><span class="p">(</span><span class="n">name</span><span class="o">=</span><span class="s1">'flat'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="k">for</span> <span class="n">dims</span> <span class="ow">in</span> <span class="n">test_dims</span><span class="p">:</span> |
| <span class="gp">... </span><span class="n">x</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">random_arrays</span><span class="p">(</span><span class="n">dims</span><span class="p">)</span> |
| <span class="gp">... </span><span class="n">y</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">simple_forward</span><span class="p">(</span><span class="n">op</span><span class="p">,</span> <span class="n">flat_data</span><span class="o">=</span><span class="n">x</span><span class="p">)</span> |
| <span class="gp">... </span><span class="n">y_np</span> <span class="o">=</span> <span class="n">x</span><span class="o">.</span><span class="n">reshape</span><span class="p">((</span><span class="n">dims</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">np</span><span class="o">.</span><span class="n">prod</span><span class="p">(</span><span class="n">dims</span><span class="p">[</span><span class="mi">1</span><span class="p">:])</span><span class="o">.</span><span class="n">astype</span><span class="p">(</span><span class="s1">'int32'</span><span class="p">)))</span> |
| <span class="gp">... </span><span class="nb">print</span><span class="p">(</span><span class="s1">'</span><span class="si">%s</span><span class="s1">: </span><span class="si">%s</span><span class="s1">'</span> <span class="o">%</span> <span class="p">(</span><span class="n">dims</span><span class="p">,</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">almost_equal</span><span class="p">(</span><span class="n">y</span><span class="p">,</span> <span class="n">y_np</span><span class="p">)))</span> |
| <span class="go">(2, 3, 4, 5): True</span> |
| <span class="go">(2, 3): True</span> |
| <span class="go">(2,): True</span> |
| </pre></div> |
| </div> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.FullyConnected"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">FullyConnected</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">weight=None</em>, <em class="sig-param">bias=None</em>, <em class="sig-param">num_hidden=_Null</em>, <em class="sig-param">no_bias=_Null</em>, <em class="sig-param">flatten=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.FullyConnected" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies a linear transformation: <span class="math notranslate nohighlight">\(Y = XW^T + b\)</span>.</p> |
| <p>If <code class="docutils literal notranslate"><span class="pre">flatten</span></code> is set to be true, then the shapes are:</p> |
| <ul class="simple"> |
| <li><p><strong>data</strong>: <cite>(batch_size, x1, x2, …, xn)</cite></p></li> |
| <li><p><strong>weight</strong>: <cite>(num_hidden, x1 * x2 * … * xn)</cite></p></li> |
| <li><p><strong>bias</strong>: <cite>(num_hidden,)</cite></p></li> |
| <li><p><strong>out</strong>: <cite>(batch_size, num_hidden)</cite></p></li> |
| </ul> |
| <p>If <code class="docutils literal notranslate"><span class="pre">flatten</span></code> is set to be false, then the shapes are:</p> |
| <ul class="simple"> |
| <li><p><strong>data</strong>: <cite>(x1, x2, …, xn, input_dim)</cite></p></li> |
| <li><p><strong>weight</strong>: <cite>(num_hidden, input_dim)</cite></p></li> |
| <li><p><strong>bias</strong>: <cite>(num_hidden,)</cite></p></li> |
| <li><p><strong>out</strong>: <cite>(x1, x2, …, xn, num_hidden)</cite></p></li> |
| </ul> |
| <p>The learnable parameters include both <code class="docutils literal notranslate"><span class="pre">weight</span></code> and <code class="docutils literal notranslate"><span class="pre">bias</span></code>.</p> |
| <p>If <code class="docutils literal notranslate"><span class="pre">no_bias</span></code> is set to be true, then the <code class="docutils literal notranslate"><span class="pre">bias</span></code> term is ignored.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The sparse support for FullyConnected is limited to forward evaluation with <cite>row_sparse</cite> |
| weight and bias, where the length of <cite>weight.indices</cite> and <cite>bias.indices</cite> must be equal |
| to <cite>num_hidden</cite>. This could be useful for model inference with <cite>row_sparse</cite> weights |
| trained with importance sampling or noise contrastive estimation.</p> |
| <p>To compute linear transformation with ‘csr’ sparse data, sparse.dot is recommended instead |
| of sparse.FullyConnected.</p> |
| </div> |
| <p>Defined in src/operator/nn/fully_connected.cc:L286</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data.</p></li> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight matrix.</p></li> |
| <li><p><strong>bias</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Bias parameter.</p></li> |
| <li><p><strong>num_hidden</strong> (<em>int</em><em>, </em><em>required</em>) – Number of hidden nodes of the output.</p></li> |
| <li><p><strong>no_bias</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to disable bias parameter.</p></li> |
| <li><p><strong>flatten</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Whether to collapse all but the first axis of the input data tensor.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <p>Construct a fully connected operator with target dimension 512.</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">data</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'data'</span><span class="p">)</span> <span class="c1"># or some constructed NN</span> |
| <span class="gp">>>> </span><span class="n">op</span> <span class="o">=</span> <span class="n">FullyConnected</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">data</span><span class="p">,</span> |
| <span class="gp">... </span><span class="n">num_hidden</span><span class="o">=</span><span class="mi">512</span><span class="p">,</span> |
| <span class="gp">... </span><span class="n">name</span><span class="o">=</span><span class="s1">'FC1'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">op</span> |
| <span class="go"><Symbol FC1></span> |
| <span class="gp">>>> </span><span class="n">SymbolDoc</span><span class="o">.</span><span class="n">get_output_shape</span><span class="p">(</span><span class="n">op</span><span class="p">,</span> <span class="n">data</span><span class="o">=</span><span class="p">(</span><span class="mi">128</span><span class="p">,</span> <span class="mi">100</span><span class="p">))</span> |
| <span class="go">{'FC1_output': (128L, 512L)}</span> |
| </pre></div> |
| </div> |
| <p>A simple 3-layer MLP with ReLU activation:</p> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">net</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'data'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="k">for</span> <span class="n">i</span><span class="p">,</span> <span class="n">dim</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">([</span><span class="mi">128</span><span class="p">,</span> <span class="mi">64</span><span class="p">]):</span> |
| <span class="gp">... </span><span class="n">net</span> <span class="o">=</span> <span class="n">FullyConnected</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">net</span><span class="p">,</span> <span class="n">num_hidden</span><span class="o">=</span><span class="n">dim</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'FC</span><span class="si">%d</span><span class="s1">'</span> <span class="o">%</span> <span class="n">i</span><span class="p">)</span> |
| <span class="gp">... </span><span class="n">net</span> <span class="o">=</span> <span class="n">Activation</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">net</span><span class="p">,</span> <span class="n">act_type</span><span class="o">=</span><span class="s1">'relu'</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'ReLU</span><span class="si">%d</span><span class="s1">'</span> <span class="o">%</span> <span class="n">i</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="c1"># 10-class predictor (e.g. MNIST)</span> |
| <span class="gp">>>> </span><span class="n">net</span> <span class="o">=</span> <span class="n">FullyConnected</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">net</span><span class="p">,</span> <span class="n">num_hidden</span><span class="o">=</span><span class="mi">10</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'pred'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">net</span> |
| <span class="go"><Symbol pred></span> |
| </pre></div> |
| </div> |
| <div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">dim_in</span><span class="p">,</span> <span class="n">dim_out</span> <span class="o">=</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="p">,</span> <span class="n">w</span><span class="p">,</span> <span class="n">b</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">random_arrays</span><span class="p">((</span><span class="mi">10</span><span class="p">,</span> <span class="n">dim_in</span><span class="p">),</span> <span class="p">(</span><span class="n">dim_out</span><span class="p">,</span> <span class="n">dim_in</span><span class="p">),</span> <span class="p">(</span><span class="n">dim_out</span><span class="p">,))</span> |
| <span class="gp">>>> </span><span class="n">op</span> <span class="o">=</span> <span class="n">FullyConnected</span><span class="p">(</span><span class="n">num_hidden</span><span class="o">=</span><span class="n">dim_out</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s1">'FC'</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">out</span> <span class="o">=</span> <span class="n">test_utils</span><span class="o">.</span><span class="n">simple_forward</span><span class="p">(</span><span class="n">op</span><span class="p">,</span> <span class="n">FC_data</span><span class="o">=</span><span class="n">x</span><span class="p">,</span> <span class="n">FC_weight</span><span class="o">=</span><span class="n">w</span><span class="p">,</span> <span class="n">FC_bias</span><span class="o">=</span><span class="n">b</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="c1"># numpy implementation of FullyConnected</span> |
| <span class="gp">>>> </span><span class="n">out_np</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">dot</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">w</span><span class="o">.</span><span class="n">T</span><span class="p">)</span> <span class="o">+</span> <span class="n">b</span> |
| <span class="gp">>>> </span><span class="n">test_utils</span><span class="o">.</span><span class="n">almost_equal</span><span class="p">(</span><span class="n">out</span><span class="p">,</span> <span class="n">out_np</span><span class="p">)</span> |
| <span class="go">True</span> |
| </pre></div> |
| </div> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.GridGenerator"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">GridGenerator</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">transform_type=_Null</em>, <em class="sig-param">target_shape=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.GridGenerator" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Generates 2D sampling grid for bilinear sampling.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the function.</p></li> |
| <li><p><strong>transform_type</strong> (<em>{'affine'</em><em>, </em><em>'warp'}</em><em>, </em><em>required</em>) – The type of transformation. For <cite>affine</cite>, input data should be an affine matrix of size (batch, 6). For <cite>warp</cite>, input data should be an optical flow of size (batch, 2, h, w).</p></li> |
| <li><p><strong>target_shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>0</em><em>,</em><em>0</em><em>]</em>) – Specifies the output shape (H, W). This is required if transformation type is <cite>affine</cite>. If transformation type is <cite>warp</cite>, this parameter is ignored.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.GroupNorm"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">GroupNorm</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">gamma=None</em>, <em class="sig-param">beta=None</em>, <em class="sig-param">num_groups=_Null</em>, <em class="sig-param">eps=_Null</em>, <em class="sig-param">output_mean_var=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.GroupNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Group normalization.</p> |
| <p>The input channels are separated into <code class="docutils literal notranslate"><span class="pre">num_groups</span></code> groups, each containing <code class="docutils literal notranslate"><span class="pre">num_channels</span> <span class="pre">/</span> <span class="pre">num_groups</span></code> channels. |
| The mean and standard-deviation are calculated separately over the each group.</p> |
| <div class="math notranslate nohighlight"> |
| \[data = data.reshape((N, num_groups, C // num_groups, ...)) |
| out = \frac{data - mean(data, axis)}{\sqrt{var(data, axis) + \epsilon}} * gamma + beta\]</div> |
| <p>Both <code class="docutils literal notranslate"><span class="pre">gamma</span></code> and <code class="docutils literal notranslate"><span class="pre">beta</span></code> are learnable parameters.</p> |
| <p>Defined in src/operator/nn/group_norm.cc:L76</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data</p></li> |
| <li><p><strong>gamma</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – gamma array</p></li> |
| <li><p><strong>beta</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – beta array</p></li> |
| <li><p><strong>num_groups</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Total number of groups.</p></li> |
| <li><p><strong>eps</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=9.99999975e-06</em>) – An <cite>epsilon</cite> parameter to prevent division by 0.</p></li> |
| <li><p><strong>output_mean_var</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Output the mean and std calculated along the given axis.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.IdentityAttachKLSparseReg"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">IdentityAttachKLSparseReg</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">sparseness_target=_Null</em>, <em class="sig-param">penalty=_Null</em>, <em class="sig-param">momentum=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.IdentityAttachKLSparseReg" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Apply a sparse regularization to the output a sigmoid activation function.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data.</p></li> |
| <li><p><strong>sparseness_target</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.100000001</em>) – The sparseness target</p></li> |
| <li><p><strong>penalty</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.00100000005</em>) – The tradeoff parameter for the sparseness penalty</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.899999976</em>) – The momentum for running average</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.InstanceNorm"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">InstanceNorm</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">gamma=None</em>, <em class="sig-param">beta=None</em>, <em class="sig-param">eps=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.InstanceNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies instance normalization to the n-dimensional input array.</p> |
| <p>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"> |
| \[out = \frac{x - mean[data]}{ \sqrt{Var[data]} + \epsilon} * gamma + beta\]</div> |
| <p>This layer is similar to batch normalization layer (<cite>BatchNorm</cite>) |
| with two differences: first, the normalization is |
| carried out per example (instance), not over a batch. Second, the |
| same normalization is applied both at test and train time. This |
| operation is also known as <cite>contrast normalization</cite>.</p> |
| <p>If the input data is of shape [batch, channel, spacial_dim1, spacial_dim2, …], |
| <cite>gamma</cite> and <cite>beta</cite> parameters must be vectors of shape [channel].</p> |
| <p>This implementation is based on this paper <a class="footnote-reference brackets" href="#id2" id="id1">1</a></p> |
| <dl class="footnote brackets"> |
| <dt class="label" id="id2"><span class="brackets"><a class="fn-backref" href="#id1">1</a></span></dt> |
| <dd><p>Instance Normalization: The Missing Ingredient for Fast Stylization, |
| D. Ulyanov, A. Vedaldi, V. Lempitsky, 2016 (arXiv:1607.08022v2).</p> |
| </dd> |
| </dl> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="o">//</span> <span class="n">Input</span> <span class="n">of</span> <span class="n">shape</span> <span class="p">(</span><span class="mi">2</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="n">x</span> <span class="o">=</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="p">[[</span> <span class="mf">3.3</span><span class="p">,</span> <span class="mf">4.4</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">gamma</span> <span class="n">parameter</span> <span class="n">of</span> <span class="n">length</span> <span class="mi">1</span> |
| <span class="n">gamma</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.5</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">beta</span> <span class="n">parameter</span> <span class="n">of</span> <span class="n">length</span> <span class="mi">1</span> |
| <span class="n">beta</span> <span class="o">=</span> <span class="p">[</span><span class="mf">0.5</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Instance</span> <span class="n">normalization</span> <span class="ow">is</span> <span class="n">calculated</span> <span class="k">with</span> <span class="n">the</span> <span class="n">above</span> <span class="n">formula</span> |
| <span class="n">InstanceNorm</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">gamma</span><span class="p">,</span><span class="n">beta</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="o">-</span><span class="mf">0.997527</span> <span class="p">,</span> <span class="mf">1.99752665</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="o">-</span><span class="mf">0.99752653</span><span class="p">,</span> <span class="mf">1.99752724</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/instance_norm.cc:L94</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – An n-dimensional input array (n > 2) of the form [batch, channel, spatial_dim1, spatial_dim2, …].</p></li> |
| <li><p><strong>gamma</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – A vector of length ‘channel’, which multiplies the normalized input.</p></li> |
| <li><p><strong>beta</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – A vector of length ‘channel’, which is added to the product of the normalized input and the weight.</p></li> |
| <li><p><strong>eps</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.00100000005</em>) – An <cite>epsilon</cite> parameter to prevent division by 0.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.L2Normalization"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">L2Normalization</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">eps=_Null</em>, <em class="sig-param">mode=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.L2Normalization" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Normalize the input array using the L2 norm.</p> |
| <p>For 1-D NDArray, it computes:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">out</span> <span class="o">=</span> <span class="n">data</span> <span class="o">/</span> <span class="n">sqrt</span><span class="p">(</span><span class="nb">sum</span><span class="p">(</span><span class="n">data</span> <span class="o">**</span> <span class="mi">2</span><span class="p">)</span> <span class="o">+</span> <span class="n">eps</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>For N-D NDArray, if the input array has shape (N, N, …, N),</p> |
| <p>with <code class="docutils literal notranslate"><span class="pre">mode</span></code> = <code class="docutils literal notranslate"><span class="pre">instance</span></code>, it normalizes each instance in the multidimensional |
| array by its L2 norm.:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="mf">0.</span><span class="o">..</span><span class="n">N</span> |
| <span class="n">out</span><span class="p">[</span><span class="n">i</span><span class="p">,:,:,</span><span class="o">...</span><span class="p">,:]</span> <span class="o">=</span> <span class="n">data</span><span class="p">[</span><span class="n">i</span><span class="p">,:,:,</span><span class="o">...</span><span class="p">,:]</span> <span class="o">/</span> <span class="n">sqrt</span><span class="p">(</span><span class="nb">sum</span><span class="p">(</span><span class="n">data</span><span class="p">[</span><span class="n">i</span><span class="p">,:,:,</span><span class="o">...</span><span class="p">,:]</span> <span class="o">**</span> <span class="mi">2</span><span class="p">)</span> <span class="o">+</span> <span class="n">eps</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>with <code class="docutils literal notranslate"><span class="pre">mode</span></code> = <code class="docutils literal notranslate"><span class="pre">channel</span></code>, it normalizes each channel in the array by its L2 norm.:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="mf">0.</span><span class="o">..</span><span class="n">N</span> |
| <span class="n">out</span><span class="p">[:,</span><span class="n">i</span><span class="p">,:,</span><span class="o">...</span><span class="p">,:]</span> <span class="o">=</span> <span class="n">data</span><span class="p">[:,</span><span class="n">i</span><span class="p">,:,</span><span class="o">...</span><span class="p">,:]</span> <span class="o">/</span> <span class="n">sqrt</span><span class="p">(</span><span class="nb">sum</span><span class="p">(</span><span class="n">data</span><span class="p">[:,</span><span class="n">i</span><span class="p">,:,</span><span class="o">...</span><span class="p">,:]</span> <span class="o">**</span> <span class="mi">2</span><span class="p">)</span> <span class="o">+</span> <span class="n">eps</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>with <code class="docutils literal notranslate"><span class="pre">mode</span></code> = <code class="docutils literal notranslate"><span class="pre">spatial</span></code>, it normalizes the cross channel norm for each position |
| in the array by its L2 norm.:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="k">for</span> <span class="n">dim</span> <span class="ow">in</span> <span class="mf">2.</span><span class="o">..</span><span class="n">N</span> |
| <span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="mf">0.</span><span class="o">..</span><span class="n">N</span> |
| <span class="n">out</span><span class="p">[</span><span class="o">.....</span><span class="p">,</span><span class="n">i</span><span class="p">,</span><span class="o">...</span><span class="p">]</span> <span class="o">=</span> <span class="n">take</span><span class="p">(</span><span class="n">out</span><span class="p">,</span> <span class="n">indices</span><span class="o">=</span><span class="n">i</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="n">dim</span><span class="p">)</span> <span class="o">/</span> <span class="n">sqrt</span><span class="p">(</span><span class="nb">sum</span><span class="p">(</span><span class="n">take</span><span class="p">(</span><span class="n">out</span><span class="p">,</span> <span class="n">indices</span><span class="o">=</span><span class="n">i</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="n">dim</span><span class="p">)</span> <span class="o">**</span> <span class="mi">2</span><span class="p">)</span> <span class="o">+</span> <span class="n">eps</span><span class="p">)</span> |
| <span class="o">-</span><span class="n">dim</span><span class="o">-</span> |
| </pre></div> |
| </div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span><span class="mi">4</span><span class="p">]],</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="p">[</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">]]]</span> |
| |
| <span class="n">L2Normalization</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">mode</span><span class="o">=</span><span class="s1">'instance'</span><span class="p">)</span> |
| <span class="o">=</span><span class="p">[[[</span> <span class="mf">0.18257418</span> <span class="mf">0.36514837</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.54772252</span> <span class="mf">0.73029673</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">0.24077171</span> <span class="mf">0.24077171</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.60192931</span> <span class="mf">0.72231513</span><span class="p">]]]</span> |
| |
| <span class="n">L2Normalization</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">mode</span><span class="o">=</span><span class="s1">'channel'</span><span class="p">)</span> |
| <span class="o">=</span><span class="p">[[[</span> <span class="mf">0.31622776</span> <span class="mf">0.44721359</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.94868326</span> <span class="mf">0.89442718</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">0.37139067</span> <span class="mf">0.31622776</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.92847669</span> <span class="mf">0.94868326</span><span class="p">]]]</span> |
| |
| <span class="n">L2Normalization</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">mode</span><span class="o">=</span><span class="s1">'spatial'</span><span class="p">)</span> |
| <span class="o">=</span><span class="p">[[[</span> <span class="mf">0.44721359</span> <span class="mf">0.89442718</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.60000002</span> <span class="mf">0.80000001</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">0.70710677</span> <span class="mf">0.70710677</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.6401844</span> <span class="mf">0.76822126</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/l2_normalization.cc:L195</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array to normalize.</p></li> |
| <li><p><strong>eps</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1.00000001e-10</em>) – A small constant for numerical stability.</p></li> |
| <li><p><strong>mode</strong> (<em>{'channel'</em><em>, </em><em>'instance'</em><em>, </em><em>'spatial'}</em><em>,</em><em>optional</em><em>, </em><em>default='instance'</em>) – Specify the dimension along which to compute L2 norm.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.LRN"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">LRN</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">alpha=_Null</em>, <em class="sig-param">beta=_Null</em>, <em class="sig-param">knorm=_Null</em>, <em class="sig-param">nsize=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.LRN" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies local response normalization to the input.</p> |
| <p>The local response normalization layer performs “lateral inhibition” by normalizing |
| over local input regions.</p> |
| <p>If <span class="math notranslate nohighlight">\(a_{x,y}^{i}\)</span> is the activity of a neuron computed by applying kernel <span class="math notranslate nohighlight">\(i\)</span> at position |
| <span class="math notranslate nohighlight">\((x, y)\)</span> and then applying the ReLU nonlinearity, the response-normalized |
| activity <span class="math notranslate nohighlight">\(b_{x,y}^{i}\)</span> is given by the expression:</p> |
| <div class="math notranslate nohighlight"> |
| \[b_{x,y}^{i} = \frac{a_{x,y}^{i}}{\Bigg({k + \frac{\alpha}{n} \sum_{j=max(0, i-\frac{n}{2})}^{min(N-1, i+\frac{n}{2})} (a_{x,y}^{j})^{2}}\Bigg)^{\beta}}\]</div> |
| <p>where the sum runs over <span class="math notranslate nohighlight">\(n\)</span> “adjacent” kernel maps at the same spatial position, and <span class="math notranslate nohighlight">\(N\)</span> is the total |
| number of kernels in the layer.</p> |
| <p>Defined in src/operator/nn/lrn.cc:L157</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to LRN</p></li> |
| <li><p><strong>alpha</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=9.99999975e-05</em>) – The variance scaling parameter <span class="math notranslate nohighlight">\(lpha\)</span> in the LRN expression.</p></li> |
| <li><p><strong>beta</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.75</em>) – The power parameter <span class="math notranslate nohighlight">\(eta\)</span> in the LRN expression.</p></li> |
| <li><p><strong>knorm</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=2</em>) – The parameter <span class="math notranslate nohighlight">\(k\)</span> in the LRN expression.</p></li> |
| <li><p><strong>nsize</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>required</em>) – normalization window width in elements.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.LayerNorm"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">LayerNorm</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">gamma=None</em>, <em class="sig-param">beta=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">eps=_Null</em>, <em class="sig-param">output_mean_var=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.LayerNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Layer normalization.</p> |
| <p>Normalizes the channels of the input tensor by mean and variance, and applies a scale <code class="docutils literal notranslate"><span class="pre">gamma</span></code> as |
| well as offset <code class="docutils literal notranslate"><span class="pre">beta</span></code>.</p> |
| <p>Assume the input has more than one dimension and we normalize along axis 1. |
| We first compute the mean and variance along this axis and then |
| compute the normalized output, which has the same shape as input, as following:</p> |
| <div class="math notranslate nohighlight"> |
| \[out = \frac{data - mean(data, axis)}{\sqrt{var(data, axis) + \epsilon}} * gamma + beta\]</div> |
| <p>Both <code class="docutils literal notranslate"><span class="pre">gamma</span></code> and <code class="docutils literal notranslate"><span class="pre">beta</span></code> are learnable parameters.</p> |
| <p>Unlike BatchNorm and InstanceNorm, the <em>mean</em> and <em>var</em> are computed along the channel dimension.</p> |
| <p>Assume the input has size <em>k</em> on axis 1, then both <code class="docutils literal notranslate"><span class="pre">gamma</span></code> and <code class="docutils literal notranslate"><span class="pre">beta</span></code> |
| have shape <em>(k,)</em>. If <code class="docutils literal notranslate"><span class="pre">output_mean_var</span></code> is set to be true, then outputs both <code class="docutils literal notranslate"><span class="pre">data_mean</span></code> and |
| <code class="docutils literal notranslate"><span class="pre">data_std</span></code>. Note that no gradient will be passed through these two outputs.</p> |
| <p>The parameter <code class="docutils literal notranslate"><span class="pre">axis</span></code> specifies which axis of the input shape denotes |
| the ‘channel’ (separately normalized groups). The default is -1, which sets the channel |
| axis to be the last item in the input shape.</p> |
| <p>Defined in src/operator/nn/layer_norm.cc:L201</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to layer normalization</p></li> |
| <li><p><strong>gamma</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – gamma array</p></li> |
| <li><p><strong>beta</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – beta array</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='-1'</em>) – The axis to perform layer normalization. Usually, this should be be axis of the channel dimension. Negative values means indexing from right to left.</p></li> |
| <li><p><strong>eps</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=9.99999975e-06</em>) – An <cite>epsilon</cite> parameter to prevent division by 0.</p></li> |
| <li><p><strong>output_mean_var</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Output the mean and std calculated along the given axis.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.LeakyReLU"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">LeakyReLU</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">gamma=None</em>, <em class="sig-param">act_type=_Null</em>, <em class="sig-param">slope=_Null</em>, <em class="sig-param">lower_bound=_Null</em>, <em class="sig-param">upper_bound=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.LeakyReLU" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies Leaky rectified linear unit activation element-wise to the input.</p> |
| <p>Leaky ReLUs attempt to fix the “dying ReLU” problem by allowing a small <cite>slope</cite> |
| when the input is negative and has a slope of one when input is positive.</p> |
| <p>The following modified ReLU Activation functions are supported:</p> |
| <ul class="simple"> |
| <li><p><em>elu</em>: Exponential Linear Unit. <cite>y = x > 0 ? x : slope * (exp(x)-1)</cite></p></li> |
| <li><p><em>selu</em>: Scaled Exponential Linear Unit. <cite>y = lambda * (x > 0 ? x : alpha * (exp(x) - 1))</cite> where |
| <em>lambda = 1.0507009873554804934193349852946</em> and <em>alpha = 1.6732632423543772848170429916717</em>.</p></li> |
| <li><p><em>leaky</em>: Leaky ReLU. <cite>y = x > 0 ? x : slope * x</cite></p></li> |
| <li><p><em>prelu</em>: Parametric ReLU. This is same as <em>leaky</em> except that <cite>slope</cite> is learnt during training.</p></li> |
| <li><p><em>rrelu</em>: Randomized ReLU. same as <em>leaky</em> but the <cite>slope</cite> is uniformly and randomly chosen from |
| <em>[lower_bound, upper_bound)</em> for training, while fixed to be |
| <em>(lower_bound+upper_bound)/2</em> for inference.</p></li> |
| </ul> |
| <p>Defined in src/operator/leaky_relu.cc:L162</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to activation function.</p></li> |
| <li><p><strong>gamma</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to activation function.</p></li> |
| <li><p><strong>act_type</strong> (<em>{'elu'</em><em>, </em><em>'gelu'</em><em>, </em><em>'leaky'</em><em>, </em><em>'prelu'</em><em>, </em><em>'rrelu'</em><em>, </em><em>'selu'}</em><em>,</em><em>optional</em><em>, </em><em>default='leaky'</em>) – Activation function to be applied.</p></li> |
| <li><p><strong>slope</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.25</em>) – Init slope for the activation. (For leaky and elu only)</p></li> |
| <li><p><strong>lower_bound</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.125</em>) – Lower bound of random slope. (For rrelu only)</p></li> |
| <li><p><strong>upper_bound</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.333999991</em>) – Upper bound of random slope. (For rrelu only)</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.LinearRegressionOutput"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">LinearRegressionOutput</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">label=None</em>, <em class="sig-param">grad_scale=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.LinearRegressionOutput" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes and optimizes for squared loss during backward propagation. |
| Just outputs <code class="docutils literal notranslate"><span class="pre">data</span></code> during forward propagation.</p> |
| <p>If <span class="math notranslate nohighlight">\(\hat{y}_i\)</span> is the predicted value of the i-th sample, and <span class="math notranslate nohighlight">\(y_i\)</span> is the corresponding target value, |
| then the squared loss estimated over <span class="math notranslate nohighlight">\(n\)</span> samples is defined as</p> |
| <p><span class="math notranslate nohighlight">\(\text{SquaredLoss}(\textbf{Y}, \hat{\textbf{Y}} ) = \frac{1}{n} \sum_{i=0}^{n-1} \lVert \textbf{y}_i - \hat{\textbf{y}}_i \rVert_2\)</span></p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>Use the LinearRegressionOutput as the final output layer of a net.</p> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">label</span></code> can be <code class="docutils literal notranslate"><span class="pre">default</span></code> or <code class="docutils literal notranslate"><span class="pre">csr</span></code></p> |
| <ul class="simple"> |
| <li><p>LinearRegressionOutput(default, default) = default</p></li> |
| <li><p>LinearRegressionOutput(default, csr) = default</p></li> |
| </ul> |
| <p>By default, gradients of this loss function are scaled by factor <cite>1/m</cite>, where m is the number of regression outputs of a training example. |
| The parameter <cite>grad_scale</cite> can be used to change this scale to <cite>grad_scale/m</cite>.</p> |
| <p>Defined in src/operator/regression_output.cc:L92</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the function.</p></li> |
| <li><p><strong>label</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input label to the function.</p></li> |
| <li><p><strong>grad_scale</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scale the gradient by a float factor</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.LogisticRegressionOutput"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">LogisticRegressionOutput</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">label=None</em>, <em class="sig-param">grad_scale=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.LogisticRegressionOutput" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies a logistic function to the input.</p> |
| <p>The logistic function, also known as the sigmoid function, is computed as |
| <span class="math notranslate nohighlight">\(\frac{1}{1+exp(-\textbf{x})}\)</span>.</p> |
| <p>Commonly, the sigmoid is used to squash the real-valued output of a linear model |
| <span class="math notranslate nohighlight">\(wTx+b\)</span> into the [0,1] range so that it can be interpreted as a probability. |
| It is suitable for binary classification or probability prediction tasks.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>Use the LogisticRegressionOutput as the final output layer of a net.</p> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">label</span></code> can be <code class="docutils literal notranslate"><span class="pre">default</span></code> or <code class="docutils literal notranslate"><span class="pre">csr</span></code></p> |
| <ul class="simple"> |
| <li><p>LogisticRegressionOutput(default, default) = default</p></li> |
| <li><p>LogisticRegressionOutput(default, csr) = default</p></li> |
| </ul> |
| <p>The loss function used is the Binary Cross Entropy Loss:</p> |
| <p><span class="math notranslate nohighlight">\(-{(y\log(p) + (1 - y)\log(1 - p))}\)</span></p> |
| <p>Where <cite>y</cite> is the ground truth probability of positive outcome for a given example, and <cite>p</cite> the probability predicted by the model. By default, gradients of this loss function are scaled by factor <cite>1/m</cite>, where m is the number of regression outputs of a training example. |
| The parameter <cite>grad_scale</cite> can be used to change this scale to <cite>grad_scale/m</cite>.</p> |
| <p>Defined in src/operator/regression_output.cc:L152</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the function.</p></li> |
| <li><p><strong>label</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input label to the function.</p></li> |
| <li><p><strong>grad_scale</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scale the gradient by a float factor</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.MAERegressionOutput"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">MAERegressionOutput</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">label=None</em>, <em class="sig-param">grad_scale=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.MAERegressionOutput" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes mean absolute error of the input.</p> |
| <p>MAE is a risk metric corresponding to the expected value of the absolute error.</p> |
| <p>If <span class="math notranslate nohighlight">\(\hat{y}_i\)</span> is the predicted value of the i-th sample, and <span class="math notranslate nohighlight">\(y_i\)</span> is the corresponding target value, |
| then the mean absolute error (MAE) estimated over <span class="math notranslate nohighlight">\(n\)</span> samples is defined as</p> |
| <p><span class="math notranslate nohighlight">\(\text{MAE}(\textbf{Y}, \hat{\textbf{Y}} ) = \frac{1}{n} \sum_{i=0}^{n-1} \lVert \textbf{y}_i - \hat{\textbf{y}}_i \rVert_1\)</span></p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>Use the MAERegressionOutput as the final output layer of a net.</p> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">label</span></code> can be <code class="docutils literal notranslate"><span class="pre">default</span></code> or <code class="docutils literal notranslate"><span class="pre">csr</span></code></p> |
| <ul class="simple"> |
| <li><p>MAERegressionOutput(default, default) = default</p></li> |
| <li><p>MAERegressionOutput(default, csr) = default</p></li> |
| </ul> |
| <p>By default, gradients of this loss function are scaled by factor <cite>1/m</cite>, where m is the number of regression outputs of a training example. |
| The parameter <cite>grad_scale</cite> can be used to change this scale to <cite>grad_scale/m</cite>.</p> |
| <p>Defined in src/operator/regression_output.cc:L120</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the function.</p></li> |
| <li><p><strong>label</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input label to the function.</p></li> |
| <li><p><strong>grad_scale</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scale the gradient by a float factor</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.MakeLoss"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">MakeLoss</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">grad_scale=_Null</em>, <em class="sig-param">valid_thresh=_Null</em>, <em class="sig-param">normalization=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.MakeLoss" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Make your own loss function in network construction.</p> |
| <p>This operator accepts a customized loss function symbol as a terminal loss and |
| the symbol should be an operator with no backward dependency. |
| The output of this function is the gradient of loss with respect to the input data.</p> |
| <p>For example, if you are a making a cross entropy loss function. Assume <code class="docutils literal notranslate"><span class="pre">out</span></code> is the |
| predicted output and <code class="docutils literal notranslate"><span class="pre">label</span></code> is the true label, then the cross entropy can be defined as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">cross_entropy</span> <span class="o">=</span> <span class="n">label</span> <span class="o">*</span> <span class="n">log</span><span class="p">(</span><span class="n">out</span><span class="p">)</span> <span class="o">+</span> <span class="p">(</span><span class="mi">1</span> <span class="o">-</span> <span class="n">label</span><span class="p">)</span> <span class="o">*</span> <span class="n">log</span><span class="p">(</span><span class="mi">1</span> <span class="o">-</span> <span class="n">out</span><span class="p">)</span> |
| <span class="n">loss</span> <span class="o">=</span> <span class="n">MakeLoss</span><span class="p">(</span><span class="n">cross_entropy</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>We will need to use <code class="docutils literal notranslate"><span class="pre">MakeLoss</span></code> when we are creating our own loss function or we want to |
| combine multiple loss functions. Also we may want to stop some variables’ gradients |
| from backpropagation. See more detail in <code class="docutils literal notranslate"><span class="pre">BlockGrad</span></code> or <code class="docutils literal notranslate"><span class="pre">stop_gradient</span></code>.</p> |
| <p>In addition, we can give a scale to the loss by setting <code class="docutils literal notranslate"><span class="pre">grad_scale</span></code>, |
| so that the gradient of the loss will be rescaled in the backpropagation.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>This operator should be used as a Symbol instead of NDArray.</p> |
| </div> |
| <p>Defined in src/operator/make_loss.cc:L70</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array.</p></li> |
| <li><p><strong>grad_scale</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Gradient scale as a supplement to unary and binary operators</p></li> |
| <li><p><strong>valid_thresh</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – clip each element in the array to 0 when it is less than <code class="docutils literal notranslate"><span class="pre">valid_thresh</span></code>. This is used when <code class="docutils literal notranslate"><span class="pre">normalization</span></code> is set to <code class="docutils literal notranslate"><span class="pre">'valid'</span></code>.</p></li> |
| <li><p><strong>normalization</strong> (<em>{'batch'</em><em>, </em><em>'null'</em><em>, </em><em>'valid'}</em><em>,</em><em>optional</em><em>, </em><em>default='null'</em>) – If this is set to null, the output gradient will not be normalized. If this is set to batch, the output gradient will be divided by the batch size. If this is set to valid, the output gradient will be divided by the number of valid input elements.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Pad"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Pad</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">mode=_Null</em>, <em class="sig-param">pad_width=_Null</em>, <em class="sig-param">constant_value=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Pad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Pads an input array with a constant or edge values of the array.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><cite>Pad</cite> is deprecated. Use <cite>pad</cite> instead.</p> |
| </div> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>Current implementation only supports 4D and 5D input arrays with padding applied |
| only on axes 1, 2 and 3. Expects axes 4 and 5 in <cite>pad_width</cite> to be zero.</p> |
| </div> |
| <p>This operation pads an input array with either a <cite>constant_value</cite> or edge values |
| along each axis of the input array. The amount of padding is specified by <cite>pad_width</cite>.</p> |
| <p><cite>pad_width</cite> is a tuple of integer padding widths for each axis of the format |
| <code class="docutils literal notranslate"><span class="pre">(before_1,</span> <span class="pre">after_1,</span> <span class="pre">...</span> <span class="pre">,</span> <span class="pre">before_N,</span> <span class="pre">after_N)</span></code>. The <cite>pad_width</cite> should be of length <code class="docutils literal notranslate"><span class="pre">2*N</span></code> |
| where <code class="docutils literal notranslate"><span class="pre">N</span></code> is the number of dimensions of the array.</p> |
| <p>For dimension <code class="docutils literal notranslate"><span class="pre">N</span></code> of the input array, <code class="docutils literal notranslate"><span class="pre">before_N</span></code> and <code class="docutils literal notranslate"><span class="pre">after_N</span></code> indicates how many values |
| to add before and after the elements of the array along dimension <code class="docutils literal notranslate"><span class="pre">N</span></code>. |
| The widths of the higher two dimensions <code class="docutils literal notranslate"><span class="pre">before_1</span></code>, <code class="docutils literal notranslate"><span class="pre">after_1</span></code>, <code class="docutils literal notranslate"><span class="pre">before_2</span></code>, |
| <code class="docutils literal notranslate"><span class="pre">after_2</span></code> must be 0.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[[</span> <span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span> <span class="mf">5.</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">7.</span> <span class="mf">8.</span> <span class="mf">9.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">10.</span> <span class="mf">11.</span> <span class="mf">12.</span><span class="p">]]]</span> |
| |
| |
| <span class="p">[[[</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">13.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">14.</span> <span class="mf">15.</span> <span class="mf">16.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">17.</span> <span class="mf">18.</span> <span class="mf">19.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">20.</span> <span class="mf">21.</span> <span class="mf">22.</span><span class="p">]]]]</span> |
| |
| <span class="n">pad</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">mode</span><span class="o">=</span><span class="s2">"edge"</span><span class="p">,</span> <span class="n">pad_width</span><span class="o">=</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</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">1</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="o">=</span> |
| |
| <span class="p">[[[[</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">3.</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">3.</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">5.</span> <span class="mf">6.</span> <span class="mf">6.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">5.</span> <span class="mf">6.</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">7.</span> <span class="mf">7.</span> <span class="mf">8.</span> <span class="mf">9.</span> <span class="mf">9.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">7.</span> <span class="mf">7.</span> <span class="mf">8.</span> <span class="mf">9.</span> <span class="mf">9.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">10.</span> <span class="mf">10.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">12.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">10.</span> <span class="mf">10.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">12.</span><span class="p">]]]</span> |
| |
| |
| <span class="p">[[[</span> <span class="mf">11.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">13.</span> <span class="mf">13.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">11.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">13.</span> <span class="mf">13.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">14.</span> <span class="mf">14.</span> <span class="mf">15.</span> <span class="mf">16.</span> <span class="mf">16.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">14.</span> <span class="mf">14.</span> <span class="mf">15.</span> <span class="mf">16.</span> <span class="mf">16.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">17.</span> <span class="mf">17.</span> <span class="mf">18.</span> <span class="mf">19.</span> <span class="mf">19.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">17.</span> <span class="mf">17.</span> <span class="mf">18.</span> <span class="mf">19.</span> <span class="mf">19.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">20.</span> <span class="mf">20.</span> <span class="mf">21.</span> <span class="mf">22.</span> <span class="mf">22.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">20.</span> <span class="mf">20.</span> <span class="mf">21.</span> <span class="mf">22.</span> <span class="mf">22.</span><span class="p">]]]]</span> |
| |
| <span class="n">pad</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">mode</span><span class="o">=</span><span class="s2">"constant"</span><span class="p">,</span> <span class="n">constant_value</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">pad_width</span><span class="o">=</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</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">1</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="o">=</span> |
| |
| <span class="p">[[[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">3.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">4.</span> <span class="mf">5.</span> <span class="mf">6.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">7.</span> <span class="mf">8.</span> <span class="mf">9.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">10.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]]</span> |
| |
| |
| <span class="p">[[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">13.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">14.</span> <span class="mf">15.</span> <span class="mf">16.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">17.</span> <span class="mf">18.</span> <span class="mf">19.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">20.</span> <span class="mf">21.</span> <span class="mf">22.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/pad.cc:L765</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – An n-dimensional input array.</p></li> |
| <li><p><strong>mode</strong> (<em>{'constant'</em><em>, </em><em>'edge'</em><em>, </em><em>'reflect'}</em><em>, </em><em>required</em>) – Padding type to use. “constant” pads with <cite>constant_value</cite> “edge” pads using the edge values of the input array “reflect” pads by reflecting values with respect to the edges.</p></li> |
| <li><p><strong>pad_width</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – Widths of the padding regions applied to the edges of each axis. It is a tuple of integer padding widths for each axis of the format <code class="docutils literal notranslate"><span class="pre">(before_1,</span> <span class="pre">after_1,</span> <span class="pre">...</span> <span class="pre">,</span> <span class="pre">before_N,</span> <span class="pre">after_N)</span></code>. It should be of length <code class="docutils literal notranslate"><span class="pre">2*N</span></code> where <code class="docutils literal notranslate"><span class="pre">N</span></code> is the number of dimensions of the array.This is equivalent to pad_width in numpy.pad, but flattened.</p></li> |
| <li><p><strong>constant_value</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The value used for padding when <cite>mode</cite> is “constant”.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Pooling"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Pooling</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">kernel=_Null</em>, <em class="sig-param">pool_type=_Null</em>, <em class="sig-param">global_pool=_Null</em>, <em class="sig-param">cudnn_off=_Null</em>, <em class="sig-param">pooling_convention=_Null</em>, <em class="sig-param">stride=_Null</em>, <em class="sig-param">pad=_Null</em>, <em class="sig-param">p_value=_Null</em>, <em class="sig-param">count_include_pad=_Null</em>, <em class="sig-param">layout=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Pooling" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Performs pooling on the input.</p> |
| <p>The shapes for 1-D pooling are</p> |
| <ul class="simple"> |
| <li><p><strong>data</strong> and <strong>out</strong>: <em>(batch_size, channel, width)</em> (NCW layout) or |
| <em>(batch_size, width, channel)</em> (NWC layout),</p></li> |
| </ul> |
| <p>The shapes for 2-D pooling are</p> |
| <ul> |
| <li><p><strong>data</strong> and <strong>out</strong>: <em>(batch_size, channel, height, width)</em> (NCHW layout) or |
| <em>(batch_size, height, width, channel)</em> (NHWC layout),</p> |
| <blockquote> |
| <div><p>out_height = f(height, kernel[0], pad[0], stride[0]) |
| out_width = f(width, kernel[1], pad[1], stride[1])</p> |
| </div></blockquote> |
| </li> |
| </ul> |
| <p>The definition of <em>f</em> depends on <code class="docutils literal notranslate"><span class="pre">pooling_convention</span></code>, which has two options:</p> |
| <ul> |
| <li><p><strong>valid</strong> (default):</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">f</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">k</span><span class="p">,</span> <span class="n">p</span><span class="p">,</span> <span class="n">s</span><span class="p">)</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">x</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">p</span><span class="o">-</span><span class="n">k</span><span class="p">)</span><span class="o">/</span><span class="n">s</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| <li><p><strong>full</strong>, which is compatible with Caffe:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">f</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">k</span><span class="p">,</span> <span class="n">p</span><span class="p">,</span> <span class="n">s</span><span class="p">)</span> <span class="o">=</span> <span class="n">ceil</span><span class="p">((</span><span class="n">x</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">p</span><span class="o">-</span><span class="n">k</span><span class="p">)</span><span class="o">/</span><span class="n">s</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| <p>When <code class="docutils literal notranslate"><span class="pre">global_pool</span></code> is set to be true, then global pooling is performed. It will reset |
| <code class="docutils literal notranslate"><span class="pre">kernel=(height,</span> <span class="pre">width)</span></code> and set the appropiate padding to 0.</p> |
| <p>Three pooling options are supported by <code class="docutils literal notranslate"><span class="pre">pool_type</span></code>:</p> |
| <ul class="simple"> |
| <li><p><strong>avg</strong>: average pooling</p></li> |
| <li><p><strong>max</strong>: max pooling</p></li> |
| <li><p><strong>sum</strong>: sum pooling</p></li> |
| <li><p><strong>lp</strong>: Lp pooling</p></li> |
| </ul> |
| <p>For 3-D pooling, an additional <em>depth</em> dimension is added before |
| <em>height</em>. Namely the input data and output will have shape <em>(batch_size, channel, depth, |
| height, width)</em> (NCDHW layout) or <em>(batch_size, depth, height, width, channel)</em> (NDHWC layout).</p> |
| <p>Notes on Lp pooling:</p> |
| <p>Lp pooling was first introduced by this paper: <a class="reference external" href="https://arxiv.org/pdf/1204.3968.pdf">https://arxiv.org/pdf/1204.3968.pdf</a>. |
| L-1 pooling is simply sum pooling, while L-inf pooling is simply max pooling. |
| We can see that Lp pooling stands between those two, in practice the most common value for p is 2.</p> |
| <p>For each window <code class="docutils literal notranslate"><span class="pre">X</span></code>, the mathematical expression for Lp pooling is:</p> |
| <p><span class="math notranslate nohighlight">\(f(X) = \sqrt[p]{\sum_{x}^{X} x^p}\)</span></p> |
| <p>Defined in src/operator/nn/pooling.cc:L416</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the pooling operator.</p></li> |
| <li><p><strong>kernel</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Pooling kernel size: (y, x) or (d, y, x)</p></li> |
| <li><p><strong>pool_type</strong> (<em>{'avg'</em><em>, </em><em>'lp'</em><em>, </em><em>'max'</em><em>, </em><em>'sum'}</em><em>,</em><em>optional</em><em>, </em><em>default='max'</em>) – Pooling type to be applied.</p></li> |
| <li><p><strong>global_pool</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Ignore kernel size, do global pooling based on current input feature map.</p></li> |
| <li><p><strong>cudnn_off</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Turn off cudnn pooling and use MXNet pooling operator.</p></li> |
| <li><p><strong>pooling_convention</strong> (<em>{'full'</em><em>, </em><em>'same'</em><em>, </em><em>'valid'}</em><em>,</em><em>optional</em><em>, </em><em>default='valid'</em>) – Pooling convention to be applied.</p></li> |
| <li><p><strong>stride</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Stride: for pooling (y, x) or (d, y, x). Defaults to 1 for each dimension.</p></li> |
| <li><p><strong>pad</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Pad for pooling: (y, x) or (d, y, x). Defaults to no padding.</p></li> |
| <li><p><strong>p_value</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – Value of p for Lp pooling, can be 1 or 2, required for Lp Pooling.</p></li> |
| <li><p><strong>count_include_pad</strong> (<em>boolean</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Only used for AvgPool, specify whether to count padding elements for averagecalculation. For example, with a 5*5 kernel on a 3*3 corner of a image,the sum of the 9 valid elements will be divided by 25 if this is set to true,or it will be divided by 9 if this is set to false. Defaults to true.</p></li> |
| <li><p><strong>layout</strong> (<em>{None</em><em>, </em><em>'NCDHW'</em><em>, </em><em>'NCHW'</em><em>, </em><em>'NCW'</em><em>, </em><em>'NDHWC'</em><em>, </em><em>'NHWC'</em><em>, </em><em>'NWC'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – Set layout for input and output. Empty for |
| default layout: NCW for 1d, NCHW for 2d and NCDHW for 3d.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Pooling_v1"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Pooling_v1</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">kernel=_Null</em>, <em class="sig-param">pool_type=_Null</em>, <em class="sig-param">global_pool=_Null</em>, <em class="sig-param">pooling_convention=_Null</em>, <em class="sig-param">stride=_Null</em>, <em class="sig-param">pad=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Pooling_v1" title="Permalink to this definition">¶</a></dt> |
| <dd><p>This operator is DEPRECATED. |
| Perform pooling on the input.</p> |
| <p>The shapes for 2-D pooling is</p> |
| <ul> |
| <li><p><strong>data</strong>: <em>(batch_size, channel, height, width)</em></p></li> |
| <li><p><strong>out</strong>: <em>(batch_size, num_filter, out_height, out_width)</em>, with:</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">f</span><span class="p">(</span><span class="n">height</span><span class="p">,</span> <span class="n">kernel</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">pad</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">stride</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="n">f</span><span class="p">(</span><span class="n">width</span><span class="p">,</span> <span class="n">kernel</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">pad</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">stride</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| <p>The definition of <em>f</em> depends on <code class="docutils literal notranslate"><span class="pre">pooling_convention</span></code>, which has two options:</p> |
| <ul> |
| <li><p><strong>valid</strong> (default):</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">f</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">k</span><span class="p">,</span> <span class="n">p</span><span class="p">,</span> <span class="n">s</span><span class="p">)</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">x</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">p</span><span class="o">-</span><span class="n">k</span><span class="p">)</span><span class="o">/</span><span class="n">s</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| <li><p><strong>full</strong>, which is compatible with Caffe:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">f</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">k</span><span class="p">,</span> <span class="n">p</span><span class="p">,</span> <span class="n">s</span><span class="p">)</span> <span class="o">=</span> <span class="n">ceil</span><span class="p">((</span><span class="n">x</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">p</span><span class="o">-</span><span class="n">k</span><span class="p">)</span><span class="o">/</span><span class="n">s</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| <p>But <code class="docutils literal notranslate"><span class="pre">global_pool</span></code> is set to be true, then do a global pooling, namely reset |
| <code class="docutils literal notranslate"><span class="pre">kernel=(height,</span> <span class="pre">width)</span></code>.</p> |
| <p>Three pooling options are supported by <code class="docutils literal notranslate"><span class="pre">pool_type</span></code>:</p> |
| <ul class="simple"> |
| <li><p><strong>avg</strong>: average pooling</p></li> |
| <li><p><strong>max</strong>: max pooling</p></li> |
| <li><p><strong>sum</strong>: sum pooling</p></li> |
| </ul> |
| <p>1-D pooling is special case of 2-D pooling with <em>weight=1</em> and |
| <em>kernel[1]=1</em>.</p> |
| <p>For 3-D pooling, an additional <em>depth</em> dimension is added before |
| <em>height</em>. Namely the input data will have shape <em>(batch_size, channel, depth, |
| height, width)</em>.</p> |
| <p>Defined in src/operator/pooling_v1.cc:L103</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the pooling operator.</p></li> |
| <li><p><strong>kernel</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – pooling kernel size: (y, x) or (d, y, x)</p></li> |
| <li><p><strong>pool_type</strong> (<em>{'avg'</em><em>, </em><em>'max'</em><em>, </em><em>'sum'}</em><em>,</em><em>optional</em><em>, </em><em>default='max'</em>) – Pooling type to be applied.</p></li> |
| <li><p><strong>global_pool</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Ignore kernel size, do global pooling based on current input feature map.</p></li> |
| <li><p><strong>pooling_convention</strong> (<em>{'full'</em><em>, </em><em>'valid'}</em><em>,</em><em>optional</em><em>, </em><em>default='valid'</em>) – Pooling convention to be applied.</p></li> |
| <li><p><strong>stride</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – stride: for pooling (y, x) or (d, y, x)</p></li> |
| <li><p><strong>pad</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – pad for pooling: (y, x) or (d, y, x)</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.RNN"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">RNN</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">parameters=None</em>, <em class="sig-param">state=None</em>, <em class="sig-param">state_cell=None</em>, <em class="sig-param">sequence_length=None</em>, <em class="sig-param">state_size=_Null</em>, <em class="sig-param">num_layers=_Null</em>, <em class="sig-param">bidirectional=_Null</em>, <em class="sig-param">mode=_Null</em>, <em class="sig-param">p=_Null</em>, <em class="sig-param">state_outputs=_Null</em>, <em class="sig-param">projection_size=_Null</em>, <em class="sig-param">lstm_state_clip_min=_Null</em>, <em class="sig-param">lstm_state_clip_max=_Null</em>, <em class="sig-param">lstm_state_clip_nan=_Null</em>, <em class="sig-param">use_sequence_length=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.RNN" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies recurrent layers to input data. Currently, vanilla RNN, LSTM and GRU are |
| implemented, with both multi-layer and bidirectional support.</p> |
| <p>When the input data is of type float32 and the environment variables MXNET_CUDA_ALLOW_TENSOR_CORE |
| and MXNET_CUDA_TENSOR_OP_MATH_ALLOW_CONVERSION are set to 1, this operator will try to use |
| pseudo-float16 precision (float32 math with float16 I/O) precision in order to use |
| Tensor Cores on suitable NVIDIA GPUs. This can sometimes give significant speedups.</p> |
| <p><strong>Vanilla RNN</strong></p> |
| <p>Applies a single-gate recurrent layer to input X. Two kinds of activation function are supported: |
| ReLU and Tanh.</p> |
| <p>With ReLU activation function:</p> |
| <div class="math notranslate nohighlight"> |
| \[h_t = relu(W_{ih} * x_t + b_{ih} + W_{hh} * h_{(t-1)} + b_{hh})\]</div> |
| <p>With Tanh activtion function:</p> |
| <div class="math notranslate nohighlight"> |
| \[h_t = \tanh(W_{ih} * x_t + b_{ih} + W_{hh} * h_{(t-1)} + b_{hh})\]</div> |
| <p>Reference paper: Finding structure in time - Elman, 1988. |
| <a class="reference external" href="https://crl.ucsd.edu/~elman/Papers/fsit.pdf">https://crl.ucsd.edu/~elman/Papers/fsit.pdf</a></p> |
| <p><strong>LSTM</strong></p> |
| <p>Long Short-Term Memory - Hochreiter, 1997. <a class="reference external" href="http://www.bioinf.jku.at/publications/older/2604.pdf">http://www.bioinf.jku.at/publications/older/2604.pdf</a></p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}\begin{array}{ll} |
| i_t = \mathrm{sigmoid}(W_{ii} x_t + b_{ii} + W_{hi} h_{(t-1)} + b_{hi}) \\ |
| f_t = \mathrm{sigmoid}(W_{if} x_t + b_{if} + W_{hf} h_{(t-1)} + b_{hf}) \\ |
| g_t = \tanh(W_{ig} x_t + b_{ig} + W_{hc} h_{(t-1)} + b_{hg}) \\ |
| o_t = \mathrm{sigmoid}(W_{io} x_t + b_{io} + W_{ho} h_{(t-1)} + b_{ho}) \\ |
| c_t = f_t * c_{(t-1)} + i_t * g_t \\ |
| h_t = o_t * \tanh(c_t) |
| \end{array}\end{split}\]</div> |
| <p>With the projection size being set, LSTM could use the projection feature to reduce the parameters |
| size and give some speedups without significant damage to the accuracy.</p> |
| <p>Long Short-Term Memory Based Recurrent Neural Network Architectures for Large Vocabulary Speech |
| Recognition - Sak et al. 2014. <a class="reference external" href="https://arxiv.org/abs/1402.1128">https://arxiv.org/abs/1402.1128</a></p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}\begin{array}{ll} |
| i_t = \mathrm{sigmoid}(W_{ii} x_t + b_{ii} + W_{ri} r_{(t-1)} + b_{ri}) \\ |
| f_t = \mathrm{sigmoid}(W_{if} x_t + b_{if} + W_{rf} r_{(t-1)} + b_{rf}) \\ |
| g_t = \tanh(W_{ig} x_t + b_{ig} + W_{rc} r_{(t-1)} + b_{rg}) \\ |
| o_t = \mathrm{sigmoid}(W_{io} x_t + b_{o} + W_{ro} r_{(t-1)} + b_{ro}) \\ |
| c_t = f_t * c_{(t-1)} + i_t * g_t \\ |
| h_t = o_t * \tanh(c_t) |
| r_t = W_{hr} h_t |
| \end{array}\end{split}\]</div> |
| <p><strong>GRU</strong></p> |
| <p>Gated Recurrent Unit - Cho et al. 2014. <a class="reference external" href="http://arxiv.org/abs/1406.1078">http://arxiv.org/abs/1406.1078</a></p> |
| <p>The definition of GRU here is slightly different from paper but compatible with CUDNN.</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}\begin{array}{ll} |
| r_t = \mathrm{sigmoid}(W_{ir} x_t + b_{ir} + W_{hr} h_{(t-1)} + b_{hr}) \\ |
| z_t = \mathrm{sigmoid}(W_{iz} x_t + b_{iz} + W_{hz} h_{(t-1)} + b_{hz}) \\ |
| n_t = \tanh(W_{in} x_t + b_{in} + r_t * (W_{hn} h_{(t-1)}+ b_{hn})) \\ |
| h_t = (1 - z_t) * n_t + z_t * h_{(t-1)} \\ |
| \end{array}\end{split}\]</div> |
| <p>Defined in src/operator/rnn.cc:L375</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to RNN</p></li> |
| <li><p><strong>parameters</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Vector of all RNN trainable parameters concatenated</p></li> |
| <li><p><strong>state</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – initial hidden state of the RNN</p></li> |
| <li><p><strong>state_cell</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – initial cell state for LSTM networks (only for LSTM)</p></li> |
| <li><p><strong>sequence_length</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Vector of valid sequence lengths for each element in batch. (Only used if use_sequence_length kwarg is True)</p></li> |
| <li><p><strong>state_size</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>required</em>) – size of the state for each layer</p></li> |
| <li><p><strong>num_layers</strong> (<em>int</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>required</em>) – number of stacked layers</p></li> |
| <li><p><strong>bidirectional</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – whether to use bidirectional recurrent layers</p></li> |
| <li><p><strong>mode</strong> (<em>{'gru'</em><em>, </em><em>'lstm'</em><em>, </em><em>'rnn_relu'</em><em>, </em><em>'rnn_tanh'}</em><em>, </em><em>required</em>) – the type of RNN to compute</p></li> |
| <li><p><strong>p</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – drop rate of the dropout on the outputs of each RNN layer, except the last layer.</p></li> |
| <li><p><strong>state_outputs</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to have the states as symbol outputs.</p></li> |
| <li><p><strong>projection_size</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – size of project size</p></li> |
| <li><p><strong>lstm_state_clip_min</strong> (<em>double</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Minimum clip value of LSTM states. This option must be used together with lstm_state_clip_max.</p></li> |
| <li><p><strong>lstm_state_clip_max</strong> (<em>double</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Maximum clip value of LSTM states. This option must be used together with lstm_state_clip_min.</p></li> |
| <li><p><strong>lstm_state_clip_nan</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to stop NaN from propagating in state by clipping it to min/max. If clipping range is not specified, this option is ignored.</p></li> |
| <li><p><strong>use_sequence_length</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to true, this layer takes in an extra input parameter <cite>sequence_length</cite> to specify variable length sequence</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.ROIPooling"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">ROIPooling</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">rois=None</em>, <em class="sig-param">pooled_size=_Null</em>, <em class="sig-param">spatial_scale=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.ROIPooling" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Performs region of interest(ROI) pooling on the input array.</p> |
| <p>ROI pooling is a variant of a max pooling layer, in which the output size is fixed and |
| region of interest is a parameter. Its purpose is to perform max pooling on the inputs |
| of non-uniform sizes to obtain fixed-size feature maps. ROI pooling is a neural-net |
| layer mostly used in training a <cite>Fast R-CNN</cite> network for object detection.</p> |
| <p>This operator takes a 4D feature map as an input array and region proposals as <cite>rois</cite>, |
| then it pools over sub-regions of input and produces a fixed-sized output array |
| regardless of the ROI size.</p> |
| <p>To crop the feature map accordingly, you can resize the bounding box coordinates |
| by changing the parameters <cite>rois</cite> and <cite>spatial_scale</cite>.</p> |
| <p>The cropped feature maps are pooled by standard max pooling operation to a fixed size output |
| indicated by a <cite>pooled_size</cite> parameter. batch_size will change to the number of region |
| bounding boxes after <cite>ROIPooling</cite>.</p> |
| <p>The size of each region of interest doesn’t have to be perfectly divisible by |
| the number of pooling sections(<cite>pooled_size</cite>).</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span>x = [[[[ 0., 1., 2., 3., 4., 5.], |
| [ 6., 7., 8., 9., 10., 11.], |
| [ 12., 13., 14., 15., 16., 17.], |
| [ 18., 19., 20., 21., 22., 23.], |
| [ 24., 25., 26., 27., 28., 29.], |
| [ 30., 31., 32., 33., 34., 35.], |
| [ 36., 37., 38., 39., 40., 41.], |
| [ 42., 43., 44., 45., 46., 47.]]]] |
| |
| // region of interest i.e. bounding box coordinates. |
| y = [[0,0,0,4,4]] |
| |
| // returns array of shape (2,2) according to the given roi with max pooling. |
| ROIPooling(x, y, (2,2), 1.0) = [[[[ 14., 16.], |
| [ 26., 28.]]]] |
| |
| // region of interest is changed due to the change in `spacial_scale` parameter. |
| ROIPooling(x, y, (2,2), 0.7) = [[[[ 7., 9.], |
| [ 19., 21.]]]] |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/roi_pooling.cc:L224</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array to the pooling operator, a 4D Feature maps</p></li> |
| <li><p><strong>rois</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Bounding box coordinates, a 2D array of [[batch_index, x1, y1, x2, y2]], where (x1, y1) and (x2, y2) are top left and bottom right corners of designated region of interest. <cite>batch_index</cite> indicates the index of corresponding image in the input array</p></li> |
| <li><p><strong>pooled_size</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – ROI pooling output shape (h,w)</p></li> |
| <li><p><strong>spatial_scale</strong> (<em>float</em><em>, </em><em>required</em>) – Ratio of input feature map height (or w) to raw image height (or w). Equals the reciprocal of total stride in convolutional layers</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Reshape"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Reshape</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">reverse=_Null</em>, <em class="sig-param">target_shape=_Null</em>, <em class="sig-param">keep_highest=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Reshape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Reshapes the input array. |
| .. note:: <code class="docutils literal notranslate"><span class="pre">Reshape</span></code> is deprecated, use <code class="docutils literal notranslate"><span class="pre">reshape</span></code> |
| Given an array and a shape, this function returns a copy of the array in the new shape. |
| The shape is a tuple of integers such as (2,3,4). The size of the new shape should be same as the size of the input array. |
| Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">reshape</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="n">shape</span><span class="o">=</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="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span><span class="mi">4</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Some dimensions of the shape can take special values from the set {0, -1, -2, -3, -4}. The significance of each is explained below: |
| - <code class="docutils literal notranslate"><span class="pre">0</span></code> copy this dimension from the input to the output shape.</p> |
| <blockquote> |
| <div><p>Example:: |
| - input shape = (2,3,4), shape = (4,0,2), output shape = (4,3,2) |
| - input shape = (2,3,4), shape = (2,0,0), output shape = (2,3,4)</p> |
| </div></blockquote> |
| <ul class="simple"> |
| <li><p><code class="docutils literal notranslate"><span class="pre">-1</span></code> infers the dimension of the output shape by using the remainder of the input dimensions |
| keeping the size of the new array same as that of the input array. |
| At most one dimension of shape can be -1. |
| Example:: |
| - input shape = (2,3,4), shape = (6,1,-1), output shape = (6,1,4) |
| - input shape = (2,3,4), shape = (3,-1,8), output shape = (3,1,8) |
| - input shape = (2,3,4), shape=(-1,), output shape = (24,)</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">-2</span></code> copy all/remainder of the input dimensions to the output shape. |
| Example:: |
| - input shape = (2,3,4), shape = (-2,), output shape = (2,3,4) |
| - input shape = (2,3,4), shape = (2,-2), output shape = (2,3,4) |
| - input shape = (2,3,4), shape = (-2,1,1), output shape = (2,3,4,1,1)</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">-3</span></code> use the product of two consecutive dimensions of the input shape as the output dimension. |
| Example:: |
| - input shape = (2,3,4), shape = (-3,4), output shape = (6,4) |
| - input shape = (2,3,4,5), shape = (-3,-3), output shape = (6,20) |
| - input shape = (2,3,4), shape = (0,-3), output shape = (2,12) |
| - input shape = (2,3,4), shape = (-3,-2), output shape = (6,4)</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">-4</span></code> split one dimension of the input into two dimensions passed subsequent to -4 in shape (can contain -1). |
| Example:: |
| - input shape = (2,3,4), shape = (-4,1,2,-2), output shape =(1,2,3,4) |
| - input shape = (2,3,4), shape = (2,-4,-1,3,-2), output shape = (2,1,3,4)</p></li> |
| </ul> |
| <dl class="simple"> |
| <dt>If the argument <cite>reverse</cite> is set to 1, then the special values are inferred from right to left.</dt><dd><p>Example:: |
| - without reverse=1, for input shape = (10,5,4), shape = (-1,0), output shape would be (40,5) |
| - with reverse=1, output shape will be (50,4).</p> |
| </dd> |
| </dl> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L174</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to reshape.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The target shape</p></li> |
| <li><p><strong>reverse</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true then the special values are inferred from right to left</p></li> |
| <li><p><strong>target_shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – (Deprecated! Use <code class="docutils literal notranslate"><span class="pre">shape</span></code> instead.) Target new shape. One and only one dim can be 0, in which case it will be inferred from the rest of dims</p></li> |
| <li><p><strong>keep_highest</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – (Deprecated! Use <code class="docutils literal notranslate"><span class="pre">shape</span></code> instead.) Whether keep the highest dim unchanged.If set to true, then the first dim in target_shape is ignored,and always fixed as input</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.SVMOutput"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">SVMOutput</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">label=None</em>, <em class="sig-param">margin=_Null</em>, <em class="sig-param">regularization_coefficient=_Null</em>, <em class="sig-param">use_linear=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.SVMOutput" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes support vector machine based transformation of the input.</p> |
| <p>This tutorial demonstrates using SVM as output layer for classification instead of softmax: |
| <a class="reference external" href="https://github.com/apache/mxnet/tree/v1.x/example/svm_mnist">https://github.com/apache/mxnet/tree/v1.x/example/svm_mnist</a>.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data for SVM transformation.</p></li> |
| <li><p><strong>label</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Class label for the input data.</p></li> |
| <li><p><strong>margin</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – The loss function penalizes outputs that lie outside this margin. Default margin is 1.</p></li> |
| <li><p><strong>regularization_coefficient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Regularization parameter for the SVM. This balances the tradeoff between coefficient size and error.</p></li> |
| <li><p><strong>use_linear</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to use L1-SVM objective. L2-SVM objective is used by default.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.SequenceLast"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">SequenceLast</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">sequence_length=None</em>, <em class="sig-param">use_sequence_length=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.SequenceLast" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Takes the last element of a sequence.</p> |
| <p>This function takes an n-dimensional input array of the form |
| [max_sequence_length, batch_size, other_feature_dims] and returns a (n-1)-dimensional array |
| of the form [batch_size, other_feature_dims].</p> |
| <p>Parameter <cite>sequence_length</cite> is used to handle variable-length sequences. <cite>sequence_length</cite> should be |
| an input array of positive ints of dimension [batch_size]. To use this parameter, |
| set <cite>use_sequence_length</cite> to <cite>True</cite>, otherwise each example in the batch is assumed |
| to have the max sequence length.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>Alternatively, you can also use <cite>take</cite> operator.</p> |
| </div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">,</span> <span class="mf">15.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">19.</span><span class="p">,</span> <span class="mf">20.</span><span class="p">,</span> <span class="mf">21.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">22.</span><span class="p">,</span> <span class="mf">23.</span><span class="p">,</span> <span class="mf">24.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">25.</span><span class="p">,</span> <span class="mf">26.</span><span class="p">,</span> <span class="mf">27.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">returns</span> <span class="n">last</span> <span class="n">sequence</span> <span class="n">when</span> <span class="n">sequence_length</span> <span class="n">parameter</span> <span class="ow">is</span> <span class="ow">not</span> <span class="n">used</span> |
| <span class="n">SequenceLast</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">19.</span><span class="p">,</span> <span class="mf">20.</span><span class="p">,</span> <span class="mf">21.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">22.</span><span class="p">,</span> <span class="mf">23.</span><span class="p">,</span> <span class="mf">24.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">25.</span><span class="p">,</span> <span class="mf">26.</span><span class="p">,</span> <span class="mf">27.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">sequence_length</span> <span class="ow">is</span> <span class="n">used</span> |
| <span class="n">SequenceLast</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">sequence_length</span><span class="o">=</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">1</span><span class="p">],</span> <span class="n">use_sequence_length</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> |
| <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">sequence_length</span> <span class="ow">is</span> <span class="n">used</span> |
| <span class="n">SequenceLast</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">sequence_length</span><span class="o">=</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="n">use_sequence_length</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> |
| <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">,</span> <span class="mf">15.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">25.</span><span class="p">,</span> <span class="mf">26.</span><span class="p">,</span> <span class="mf">27.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/sequence_last.cc:L105</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – n-dimensional input array of the form [max_sequence_length, batch_size, other_feature_dims] where n>2</p></li> |
| <li><p><strong>sequence_length</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – vector of sequence lengths of the form [batch_size]</p></li> |
| <li><p><strong>use_sequence_length</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to true, this layer takes in an extra input parameter <cite>sequence_length</cite> to specify variable length sequence</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – The sequence axis. Only values of 0 and 1 are currently supported.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.SequenceMask"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">SequenceMask</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">sequence_length=None</em>, <em class="sig-param">use_sequence_length=_Null</em>, <em class="sig-param">value=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.SequenceMask" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Sets all elements outside the sequence to a constant value.</p> |
| <p>This function takes an n-dimensional input array of the form |
| [max_sequence_length, batch_size, other_feature_dims] and returns an array of the same shape.</p> |
| <p>Parameter <cite>sequence_length</cite> is used to handle variable-length sequences. <cite>sequence_length</cite> |
| should be an input array of positive ints of dimension [batch_size]. |
| To use this parameter, set <cite>use_sequence_length</cite> to <cite>True</cite>, |
| otherwise each example in the batch is assumed to have the max sequence length and |
| this operator works as the <cite>identity</cite> operator.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">,</span> <span class="mf">15.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">Batch</span> <span class="mi">1</span> |
| <span class="n">B1</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">,</span> <span class="mf">15.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">Batch</span> <span class="mi">2</span> |
| <span class="n">B2</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">works</span> <span class="k">as</span> <span class="n">identity</span> <span class="n">operator</span> <span class="n">when</span> <span class="n">sequence_length</span> <span class="n">parameter</span> <span class="ow">is</span> <span class="ow">not</span> <span class="n">used</span> |
| <span class="n">SequenceMask</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">,</span> <span class="mf">15.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">sequence_length</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="n">means</span> <span class="mi">1</span> <span class="n">of</span> <span class="n">each</span> <span class="n">batch</span> <span class="n">will</span> <span class="n">be</span> <span class="n">kept</span> |
| <span class="o">//</span> <span class="ow">and</span> <span class="n">other</span> <span class="n">rows</span> <span class="n">are</span> <span class="n">masked</span> <span class="k">with</span> <span class="n">default</span> <span class="n">mask</span> <span class="n">value</span> <span class="o">=</span> <span class="mi">0</span> |
| <span class="n">SequenceMask</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">sequence_length</span><span class="o">=</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="n">use_sequence_length</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> |
| <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">sequence_length</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="n">means</span> <span class="mi">2</span> <span class="n">of</span> <span class="n">batch</span> <span class="n">B1</span> <span class="ow">and</span> <span class="mi">3</span> <span class="n">of</span> <span class="n">batch</span> <span class="n">B2</span> <span class="n">will</span> <span class="n">be</span> <span class="n">kept</span> |
| <span class="o">//</span> <span class="ow">and</span> <span class="n">other</span> <span class="n">rows</span> <span class="n">are</span> <span class="n">masked</span> <span class="k">with</span> <span class="n">value</span> <span class="o">=</span> <span class="mi">1</span> |
| <span class="n">SequenceMask</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">sequence_length</span><span class="o">=</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="n">use_sequence_length</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">value</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> |
| <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/sequence_mask.cc:L185</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – n-dimensional input array of the form [max_sequence_length, batch_size, other_feature_dims] where n>2</p></li> |
| <li><p><strong>sequence_length</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – vector of sequence lengths of the form [batch_size]</p></li> |
| <li><p><strong>use_sequence_length</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to true, this layer takes in an extra input parameter <cite>sequence_length</cite> to specify variable length sequence</p></li> |
| <li><p><strong>value</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The value to be used as a mask.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – The sequence axis. Only values of 0 and 1 are currently supported.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.SequenceReverse"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">SequenceReverse</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">sequence_length=None</em>, <em class="sig-param">use_sequence_length=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.SequenceReverse" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Reverses the elements of each sequence.</p> |
| <p>This function takes an n-dimensional input array of the form [max_sequence_length, batch_size, other_feature_dims] |
| and returns an array of the same shape.</p> |
| <p>Parameter <cite>sequence_length</cite> is used to handle variable-length sequences. |
| <cite>sequence_length</cite> should be an input array of positive ints of dimension [batch_size]. |
| To use this parameter, set <cite>use_sequence_length</cite> to <cite>True</cite>, |
| otherwise each example in the batch is assumed to have the max sequence length.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">,</span> <span class="mf">15.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">Batch</span> <span class="mi">1</span> |
| <span class="n">B1</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">,</span> <span class="mf">15.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">Batch</span> <span class="mi">2</span> |
| <span class="n">B2</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">returns</span> <span class="n">reverse</span> <span class="n">sequence</span> <span class="n">when</span> <span class="n">sequence_length</span> <span class="n">parameter</span> <span class="ow">is</span> <span class="ow">not</span> <span class="n">used</span> |
| <span class="n">SequenceReverse</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">,</span> <span class="mf">15.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">sequence_length</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="n">means</span> <span class="mi">2</span> <span class="n">rows</span> <span class="n">of</span> |
| <span class="o">//</span> <span class="n">both</span> <span class="n">batch</span> <span class="n">B1</span> <span class="ow">and</span> <span class="n">B2</span> <span class="n">will</span> <span class="n">be</span> <span class="nb">reversed</span><span class="o">.</span> |
| <span class="n">SequenceReverse</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">sequence_length</span><span class="o">=</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="n">use_sequence_length</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> |
| <span class="p">[[[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">14.</span><span class="p">,</span> <span class="mf">15.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">sequence_length</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="n">means</span> <span class="mi">2</span> <span class="n">of</span> <span class="n">batch</span> <span class="n">B2</span> <span class="ow">and</span> <span class="mi">3</span> <span class="n">of</span> <span class="n">batch</span> <span class="n">B3</span> |
| <span class="o">//</span> <span class="n">will</span> <span class="n">be</span> <span class="nb">reversed</span><span class="o">.</span> |
| <span class="n">SequenceReverse</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">sequence_length</span><span class="o">=</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="n">use_sequence_length</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> |
| <span class="p">[[[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">17.</span><span class="p">,</span> <span class="mf">18.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mi">14</span><span class="p">,</span> <span class="mf">15.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/sequence_reverse.cc:L121</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – n-dimensional input array of the form [max_sequence_length, batch_size, other dims] where n>2</p></li> |
| <li><p><strong>sequence_length</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – vector of sequence lengths of the form [batch_size]</p></li> |
| <li><p><strong>use_sequence_length</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to true, this layer takes in an extra input parameter <cite>sequence_length</cite> to specify variable length sequence</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – The sequence axis. Only 0 is currently supported.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.SliceChannel"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">SliceChannel</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">num_outputs=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">squeeze_axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.SliceChannel" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Splits an array along a particular axis into multiple sub-arrays.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><code class="docutils literal notranslate"><span class="pre">SliceChannel</span></code> is deprecated. Use <code class="docutils literal notranslate"><span class="pre">split</span></code> instead.</p> |
| </div> |
| <p><strong>Note</strong> that <cite>num_outputs</cite> should evenly divide the length of the axis |
| along which to split the array.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">5.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">]]]</span> |
| <span class="n">x</span><span class="o">.</span><span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">3</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="n">split</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">num_outputs</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span> <span class="o">//</span> <span class="n">a</span> <span class="nb">list</span> <span class="n">of</span> <span class="mi">2</span> <span class="n">arrays</span> <span class="k">with</span> <span class="n">shape</span> <span class="p">(</span><span class="mi">3</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="n">y</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">3.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">5.</span><span class="p">]]]</span> |
| |
| <span class="p">[[[</span> <span class="mf">2.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">4.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">6.</span><span class="p">]]]</span> |
| |
| <span class="n">y</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">3</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="n">z</span> <span class="o">=</span> <span class="n">split</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">num_outputs</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> <span class="o">//</span> <span class="n">a</span> <span class="nb">list</span> <span class="n">of</span> <span class="mi">3</span> <span class="n">arrays</span> <span class="k">with</span> <span class="n">shape</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">1</span><span class="p">)</span> |
| <span class="n">z</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]]</span> |
| |
| <span class="p">[[[</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">]]]</span> |
| |
| <span class="p">[[[</span> <span class="mf">5.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">]]]</span> |
| |
| <span class="n">z</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</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">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p><cite>squeeze_axis=1</cite> removes the axis with length 1 from the shapes of the output arrays. |
| <strong>Note</strong> that setting <cite>squeeze_axis</cite> to <code class="docutils literal notranslate"><span class="pre">1</span></code> removes axis with length 1 only |
| along the <cite>axis</cite> which it is split. |
| Also <cite>squeeze_axis</cite> can be set to true only if <code class="docutils literal notranslate"><span class="pre">input.shape[axis]</span> <span class="pre">==</span> <span class="pre">num_outputs</span></code>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">z</span> <span class="o">=</span> <span class="n">split</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">num_outputs</span><span class="o">=</span><span class="mi">3</span><span class="p">,</span> <span class="n">squeeze_axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">//</span> <span class="n">a</span> <span class="nb">list</span> <span class="n">of</span> <span class="mi">3</span> <span class="n">arrays</span> <span class="k">with</span> <span class="n">shape</span> <span class="p">(</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span> |
| <span class="n">z</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">5.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">]]</span> |
| <span class="n">z</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">2</span> <span class="p">,</span><span class="mi">1</span> <span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/slice_channel.cc:L106</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>num_outputs</strong> (<em>int</em><em>, </em><em>required</em>) – Number of splits. Note that this should evenly divide the length of the <cite>axis</cite>.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Axis along which to split.</p></li> |
| <li><p><strong>squeeze_axis</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true, Removes the axis with length 1 from the shapes of the output arrays. <strong>Note</strong> that setting <cite>squeeze_axis</cite> to <code class="docutils literal notranslate"><span class="pre">true</span></code> removes axis with length 1 only along the <cite>axis</cite> which it is split. Also <cite>squeeze_axis</cite> can be set to <code class="docutils literal notranslate"><span class="pre">true</span></code> only if <code class="docutils literal notranslate"><span class="pre">input.shape[axis]</span> <span class="pre">==</span> <span class="pre">num_outputs</span></code>.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.Softmax"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">Softmax</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">label=None</em>, <em class="sig-param">grad_scale=_Null</em>, <em class="sig-param">ignore_label=_Null</em>, <em class="sig-param">multi_output=_Null</em>, <em class="sig-param">use_ignore=_Null</em>, <em class="sig-param">preserve_shape=_Null</em>, <em class="sig-param">normalization=_Null</em>, <em class="sig-param">out_grad=_Null</em>, <em class="sig-param">smooth_alpha=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.Softmax" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the gradient of cross entropy loss with respect to softmax output.</p> |
| <ul> |
| <li><p>This operator computes the gradient in two steps. |
| The cross entropy loss does not actually need to be computed.</p> |
| <ul class="simple"> |
| <li><p>Applies softmax function on the input array.</p></li> |
| <li><p>Computes and returns the gradient of cross entropy loss w.r.t. the softmax output.</p></li> |
| </ul> |
| </li> |
| <li><p>The softmax function, cross entropy loss and gradient is given by:</p> |
| <ul> |
| <li><p>Softmax Function:</p> |
| <div class="math notranslate nohighlight"> |
| \[\text{softmax}(x)_i = \frac{exp(x_i)}{\sum_j exp(x_j)}\]</div> |
| </li> |
| <li><p>Cross Entropy Function:</p> |
| <div class="math notranslate nohighlight"> |
| \[\text{CE(label, output)} = - \sum_i \text{label}_i \log(\text{output}_i)\]</div> |
| </li> |
| <li><p>The gradient of cross entropy loss w.r.t softmax output:</p> |
| <div class="math notranslate nohighlight"> |
| \[\text{gradient} = \text{output} - \text{label}\]</div> |
| </li> |
| </ul> |
| </li> |
| <li><p>During forward propagation, the softmax function is computed for each instance in the input array.</p> |
| <p>For general <em>N</em>-D input arrays with shape <span class="math notranslate nohighlight">\((d_1, d_2, ..., d_n)\)</span>. The size is |
| <span class="math notranslate nohighlight">\(s=d_1 \cdot d_2 \cdot \cdot \cdot d_n\)</span>. We can use the parameters <cite>preserve_shape</cite> |
| and <cite>multi_output</cite> to specify the way to compute softmax:</p> |
| <ul class="simple"> |
| <li><p>By default, <cite>preserve_shape</cite> is <code class="docutils literal notranslate"><span class="pre">false</span></code>. This operator will reshape the input array |
| into a 2-D array with shape <span class="math notranslate nohighlight">\((d_1, \frac{s}{d_1})\)</span> and then compute the softmax function for |
| each row in the reshaped array, and afterwards reshape it back to the original shape |
| <span class="math notranslate nohighlight">\((d_1, d_2, ..., d_n)\)</span>.</p></li> |
| <li><p>If <cite>preserve_shape</cite> is <code class="docutils literal notranslate"><span class="pre">true</span></code>, the softmax function will be computed along |
| the last axis (<cite>axis</cite> = <code class="docutils literal notranslate"><span class="pre">-1</span></code>).</p></li> |
| <li><p>If <cite>multi_output</cite> is <code class="docutils literal notranslate"><span class="pre">true</span></code>, the softmax function will be computed along |
| the second axis (<cite>axis</cite> = <code class="docutils literal notranslate"><span class="pre">1</span></code>).</p></li> |
| </ul> |
| </li> |
| <li><p>During backward propagation, the gradient of cross-entropy loss w.r.t softmax output array is computed. |
| The provided label can be a one-hot label array or a probability label array.</p> |
| <ul> |
| <li><p>If the parameter <cite>use_ignore</cite> is <code class="docutils literal notranslate"><span class="pre">true</span></code>, <cite>ignore_label</cite> can specify input instances |
| with a particular label to be ignored during backward propagation. <strong>This has no effect when |
| softmax `output` has same shape as `label`</strong>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">data</span> <span class="o">=</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">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="mi">2</span><span class="p">],[</span><span class="mi">3</span><span class="p">,</span><span class="mi">3</span><span class="p">,</span><span class="mi">3</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">4</span><span class="p">,</span><span class="mi">4</span><span class="p">,</span><span class="mi">4</span><span class="p">]]</span> |
| <span class="n">label</span> <span class="o">=</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span><span class="mi">0</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="n">ignore_label</span> <span class="o">=</span> <span class="mi">1</span> |
| <span class="n">SoftmaxOutput</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">data</span><span class="p">,</span> <span class="n">label</span> <span class="o">=</span> <span class="n">label</span><span class="p">,</span>\ |
| <span class="n">multi_output</span><span class="o">=</span><span class="n">true</span><span class="p">,</span> <span class="n">use_ignore</span><span class="o">=</span><span class="n">true</span><span class="p">,</span>\ |
| <span class="n">ignore_label</span><span class="o">=</span><span class="n">ignore_label</span><span class="p">)</span> |
| <span class="c1">## forward softmax output</span> |
| <span class="p">[[</span> <span class="mf">0.0320586</span> <span class="mf">0.08714432</span> <span class="mf">0.23688284</span> <span class="mf">0.64391428</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="p">]]</span> |
| <span class="c1">## backward gradient output</span> |
| <span class="p">[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="p">]</span> |
| <span class="p">[</span><span class="o">-</span><span class="mf">0.75</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="o">-</span><span class="mf">0.75</span> <span class="mf">0.25</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="o">-</span><span class="mf">0.75</span><span class="p">]]</span> |
| <span class="c1">## notice that the first row is all 0 because label[0] is 1, which is equal to ignore_label.</span> |
| </pre></div> |
| </div> |
| </li> |
| <li><p>The parameter <cite>grad_scale</cite> can be used to rescale the gradient, which is often used to |
| give each loss function different weights.</p></li> |
| <li><p>This operator also supports various ways to normalize the gradient by <cite>normalization</cite>, |
| The <cite>normalization</cite> is applied if softmax output has different shape than the labels. |
| The <cite>normalization</cite> mode can be set to the followings:</p> |
| <ul class="simple"> |
| <li><p><code class="docutils literal notranslate"><span class="pre">'null'</span></code>: do nothing.</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">'batch'</span></code>: divide the gradient by the batch size.</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">'valid'</span></code>: divide the gradient by the number of instances which are not ignored.</p></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| </ul> |
| <p>Defined in src/operator/softmax_output.cc:L242</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array.</p></li> |
| <li><p><strong>label</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Ground truth label.</p></li> |
| <li><p><strong>grad_scale</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scales the gradient by a float factor.</p></li> |
| <li><p><strong>ignore_label</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – The instances whose <cite>labels</cite> == <cite>ignore_label</cite> will be ignored during backward, if <cite>use_ignore</cite> is set to <code class="docutils literal notranslate"><span class="pre">true</span></code>).</p></li> |
| <li><p><strong>multi_output</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to <code class="docutils literal notranslate"><span class="pre">true</span></code>, the softmax function will be computed along axis <code class="docutils literal notranslate"><span class="pre">1</span></code>. This is applied when the shape of input array differs from the shape of label array.</p></li> |
| <li><p><strong>use_ignore</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to <code class="docutils literal notranslate"><span class="pre">true</span></code>, the <cite>ignore_label</cite> value will not contribute to the backward gradient.</p></li> |
| <li><p><strong>preserve_shape</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to <code class="docutils literal notranslate"><span class="pre">true</span></code>, the softmax function will be computed along the last axis (<code class="docutils literal notranslate"><span class="pre">-1</span></code>).</p></li> |
| <li><p><strong>normalization</strong> (<em>{'batch'</em><em>, </em><em>'null'</em><em>, </em><em>'valid'}</em><em>,</em><em>optional</em><em>, </em><em>default='null'</em>) – Normalizes the gradient.</p></li> |
| <li><p><strong>out_grad</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Multiplies gradient with output gradient element-wise.</p></li> |
| <li><p><strong>smooth_alpha</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Constant for computing a label smoothed version of cross-entropyfor the backwards pass. This constant gets subtracted from theone-hot encoding of the gold label and distributed uniformly toall other labels.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.SoftmaxActivation"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">SoftmaxActivation</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">mode=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.SoftmaxActivation" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies softmax activation to input. This is intended for internal layers.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>This operator has been deprecated, please use <cite>softmax</cite>.</p> |
| </div> |
| <p>If <cite>mode</cite> = <code class="docutils literal notranslate"><span class="pre">instance</span></code>, this operator will compute a softmax for each instance in the batch. |
| This is the default mode.</p> |
| <p>If <cite>mode</cite> = <code class="docutils literal notranslate"><span class="pre">channel</span></code>, this operator will compute a k-class softmax at each position |
| of each instance, where <cite>k</cite> = <code class="docutils literal notranslate"><span class="pre">num_channel</span></code>. This mode can only be used when the input array |
| has at least 3 dimensions. |
| This can be used for <cite>fully convolutional network</cite>, <cite>image segmentation</cite>, etc.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">input_array</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">array</span><span class="p">([[</span><span class="mf">3.</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">],</span> |
| <span class="gp">>>> </span> <span class="p">[</span><span class="mf">2.</span><span class="p">,</span> <span class="o">-</span><span class="mf">.4</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">]])</span> |
| <span class="gp">>>> </span><span class="n">softmax_act</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">SoftmaxActivation</span><span class="p">(</span><span class="n">input_array</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="nb">print</span> <span class="n">softmax_act</span><span class="o">.</span><span class="n">asnumpy</span><span class="p">()</span> |
| <span class="go">[[ 1.78322066e-02 1.46375655e-03 5.38485940e-04 6.56010211e-03 9.73605454e-01]</span> |
| <span class="go"> [ 6.56221947e-03 5.95310994e-04 9.73919690e-01 1.78379621e-02 1.08472735e-03]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/nn/softmax_activation.cc:L58</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>mode</strong> (<em>{'channel'</em><em>, </em><em>'instance'}</em><em>,</em><em>optional</em><em>, </em><em>default='instance'</em>) – Specifies how to compute the softmax. If set to <code class="docutils literal notranslate"><span class="pre">instance</span></code>, it computes softmax for each instance. If set to <code class="docutils literal notranslate"><span class="pre">channel</span></code>, It computes cross channel softmax for each position of each instance.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.SoftmaxOutput"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">SoftmaxOutput</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">label=None</em>, <em class="sig-param">grad_scale=_Null</em>, <em class="sig-param">ignore_label=_Null</em>, <em class="sig-param">multi_output=_Null</em>, <em class="sig-param">use_ignore=_Null</em>, <em class="sig-param">preserve_shape=_Null</em>, <em class="sig-param">normalization=_Null</em>, <em class="sig-param">out_grad=_Null</em>, <em class="sig-param">smooth_alpha=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.SoftmaxOutput" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the gradient of cross entropy loss with respect to softmax output.</p> |
| <ul> |
| <li><p>This operator computes the gradient in two steps. |
| The cross entropy loss does not actually need to be computed.</p> |
| <ul class="simple"> |
| <li><p>Applies softmax function on the input array.</p></li> |
| <li><p>Computes and returns the gradient of cross entropy loss w.r.t. the softmax output.</p></li> |
| </ul> |
| </li> |
| <li><p>The softmax function, cross entropy loss and gradient is given by:</p> |
| <ul> |
| <li><p>Softmax Function:</p> |
| <div class="math notranslate nohighlight"> |
| \[\text{softmax}(x)_i = \frac{exp(x_i)}{\sum_j exp(x_j)}\]</div> |
| </li> |
| <li><p>Cross Entropy Function:</p> |
| <div class="math notranslate nohighlight"> |
| \[\text{CE(label, output)} = - \sum_i \text{label}_i \log(\text{output}_i)\]</div> |
| </li> |
| <li><p>The gradient of cross entropy loss w.r.t softmax output:</p> |
| <div class="math notranslate nohighlight"> |
| \[\text{gradient} = \text{output} - \text{label}\]</div> |
| </li> |
| </ul> |
| </li> |
| <li><p>During forward propagation, the softmax function is computed for each instance in the input array.</p> |
| <p>For general <em>N</em>-D input arrays with shape <span class="math notranslate nohighlight">\((d_1, d_2, ..., d_n)\)</span>. The size is |
| <span class="math notranslate nohighlight">\(s=d_1 \cdot d_2 \cdot \cdot \cdot d_n\)</span>. We can use the parameters <cite>preserve_shape</cite> |
| and <cite>multi_output</cite> to specify the way to compute softmax:</p> |
| <ul class="simple"> |
| <li><p>By default, <cite>preserve_shape</cite> is <code class="docutils literal notranslate"><span class="pre">false</span></code>. This operator will reshape the input array |
| into a 2-D array with shape <span class="math notranslate nohighlight">\((d_1, \frac{s}{d_1})\)</span> and then compute the softmax function for |
| each row in the reshaped array, and afterwards reshape it back to the original shape |
| <span class="math notranslate nohighlight">\((d_1, d_2, ..., d_n)\)</span>.</p></li> |
| <li><p>If <cite>preserve_shape</cite> is <code class="docutils literal notranslate"><span class="pre">true</span></code>, the softmax function will be computed along |
| the last axis (<cite>axis</cite> = <code class="docutils literal notranslate"><span class="pre">-1</span></code>).</p></li> |
| <li><p>If <cite>multi_output</cite> is <code class="docutils literal notranslate"><span class="pre">true</span></code>, the softmax function will be computed along |
| the second axis (<cite>axis</cite> = <code class="docutils literal notranslate"><span class="pre">1</span></code>).</p></li> |
| </ul> |
| </li> |
| <li><p>During backward propagation, the gradient of cross-entropy loss w.r.t softmax output array is computed. |
| The provided label can be a one-hot label array or a probability label array.</p> |
| <ul> |
| <li><p>If the parameter <cite>use_ignore</cite> is <code class="docutils literal notranslate"><span class="pre">true</span></code>, <cite>ignore_label</cite> can specify input instances |
| with a particular label to be ignored during backward propagation. <strong>This has no effect when |
| softmax `output` has same shape as `label`</strong>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">data</span> <span class="o">=</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">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="mi">2</span><span class="p">],[</span><span class="mi">3</span><span class="p">,</span><span class="mi">3</span><span class="p">,</span><span class="mi">3</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">4</span><span class="p">,</span><span class="mi">4</span><span class="p">,</span><span class="mi">4</span><span class="p">]]</span> |
| <span class="n">label</span> <span class="o">=</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span><span class="mi">0</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="n">ignore_label</span> <span class="o">=</span> <span class="mi">1</span> |
| <span class="n">SoftmaxOutput</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">data</span><span class="p">,</span> <span class="n">label</span> <span class="o">=</span> <span class="n">label</span><span class="p">,</span>\ |
| <span class="n">multi_output</span><span class="o">=</span><span class="n">true</span><span class="p">,</span> <span class="n">use_ignore</span><span class="o">=</span><span class="n">true</span><span class="p">,</span>\ |
| <span class="n">ignore_label</span><span class="o">=</span><span class="n">ignore_label</span><span class="p">)</span> |
| <span class="c1">## forward softmax output</span> |
| <span class="p">[[</span> <span class="mf">0.0320586</span> <span class="mf">0.08714432</span> <span class="mf">0.23688284</span> <span class="mf">0.64391428</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="p">]]</span> |
| <span class="c1">## backward gradient output</span> |
| <span class="p">[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="p">]</span> |
| <span class="p">[</span><span class="o">-</span><span class="mf">0.75</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="o">-</span><span class="mf">0.75</span> <span class="mf">0.25</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="mf">0.25</span> <span class="o">-</span><span class="mf">0.75</span><span class="p">]]</span> |
| <span class="c1">## notice that the first row is all 0 because label[0] is 1, which is equal to ignore_label.</span> |
| </pre></div> |
| </div> |
| </li> |
| <li><p>The parameter <cite>grad_scale</cite> can be used to rescale the gradient, which is often used to |
| give each loss function different weights.</p></li> |
| <li><p>This operator also supports various ways to normalize the gradient by <cite>normalization</cite>, |
| The <cite>normalization</cite> is applied if softmax output has different shape than the labels. |
| The <cite>normalization</cite> mode can be set to the followings:</p> |
| <ul class="simple"> |
| <li><p><code class="docutils literal notranslate"><span class="pre">'null'</span></code>: do nothing.</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">'batch'</span></code>: divide the gradient by the batch size.</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">'valid'</span></code>: divide the gradient by the number of instances which are not ignored.</p></li> |
| </ul> |
| </li> |
| </ul> |
| </li> |
| </ul> |
| <p>Defined in src/operator/softmax_output.cc:L242</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array.</p></li> |
| <li><p><strong>label</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Ground truth label.</p></li> |
| <li><p><strong>grad_scale</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scales the gradient by a float factor.</p></li> |
| <li><p><strong>ignore_label</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – The instances whose <cite>labels</cite> == <cite>ignore_label</cite> will be ignored during backward, if <cite>use_ignore</cite> is set to <code class="docutils literal notranslate"><span class="pre">true</span></code>).</p></li> |
| <li><p><strong>multi_output</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to <code class="docutils literal notranslate"><span class="pre">true</span></code>, the softmax function will be computed along axis <code class="docutils literal notranslate"><span class="pre">1</span></code>. This is applied when the shape of input array differs from the shape of label array.</p></li> |
| <li><p><strong>use_ignore</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to <code class="docutils literal notranslate"><span class="pre">true</span></code>, the <cite>ignore_label</cite> value will not contribute to the backward gradient.</p></li> |
| <li><p><strong>preserve_shape</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set to <code class="docutils literal notranslate"><span class="pre">true</span></code>, the softmax function will be computed along the last axis (<code class="docutils literal notranslate"><span class="pre">-1</span></code>).</p></li> |
| <li><p><strong>normalization</strong> (<em>{'batch'</em><em>, </em><em>'null'</em><em>, </em><em>'valid'}</em><em>,</em><em>optional</em><em>, </em><em>default='null'</em>) – Normalizes the gradient.</p></li> |
| <li><p><strong>out_grad</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Multiplies gradient with output gradient element-wise.</p></li> |
| <li><p><strong>smooth_alpha</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Constant for computing a label smoothed version of cross-entropyfor the backwards pass. This constant gets subtracted from theone-hot encoding of the gold label and distributed uniformly toall other labels.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.SpatialTransformer"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">SpatialTransformer</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">loc=None</em>, <em class="sig-param">target_shape=_Null</em>, <em class="sig-param">transform_type=_Null</em>, <em class="sig-param">sampler_type=_Null</em>, <em class="sig-param">cudnn_off=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.SpatialTransformer" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies a spatial transformer to input feature map.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the SpatialTransformerOp.</p></li> |
| <li><p><strong>loc</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – localisation net, the output dim should be 6 when transform_type is affine. You shold initialize the weight and bias with identity tranform.</p></li> |
| <li><p><strong>target_shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>0</em><em>,</em><em>0</em><em>]</em>) – output shape(h, w) of spatial transformer: (y, x)</p></li> |
| <li><p><strong>transform_type</strong> (<em>{'affine'}</em><em>, </em><em>required</em>) – transformation type</p></li> |
| <li><p><strong>sampler_type</strong> (<em>{'bilinear'}</em><em>, </em><em>required</em>) – sampling type</p></li> |
| <li><p><strong>cudnn_off</strong> (<em>boolean</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – whether to turn cudnn off</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.SwapAxis"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">SwapAxis</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">dim1=_Null</em>, <em class="sig-param">dim2=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.SwapAxis" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Interchanges two axes of an array.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span> <span class="n">x</span> <span class="o">=</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="n">swapaxes</span><span class="p">(</span><span class="n">x</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="o">=</span> <span class="p">[[</span> <span class="mi">1</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">2</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">3</span><span class="p">]]</span> |
| |
| <span class="n">x</span> <span class="o">=</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="p">[</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">]],</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">6</span><span class="p">,</span> <span class="mi">7</span><span class="p">]]]</span> <span class="o">//</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="n">array</span> |
| |
| <span class="n">swapaxes</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">6</span><span class="p">]],</span> |
| <span class="p">[[</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">5</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">7</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/swapaxis.cc:L69</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array.</p></li> |
| <li><p><strong>dim1</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – the first axis to be swapped.</p></li> |
| <li><p><strong>dim2</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – the second axis to be swapped.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.UpSampling"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">UpSampling</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.UpSampling" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Upsamples the given input data.</p> |
| <p>Two algorithms (<code class="docutils literal notranslate"><span class="pre">sample_type</span></code>) are available for upsampling:</p> |
| <ul class="simple"> |
| <li><p>Nearest Neighbor</p></li> |
| <li><p>Bilinear</p></li> |
| </ul> |
| <p><strong>Nearest Neighbor Upsampling</strong></p> |
| <p>Input data is expected to be NCHW.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]]]]</span> |
| |
| <span class="n">UpSampling</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">scale</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">sample_type</span><span class="o">=</span><span class="s1">'nearest'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]]]]</span> |
| </pre></div> |
| </div> |
| <p><strong>Bilinear Upsampling</strong></p> |
| <p>Uses <cite>deconvolution</cite> algorithm under the hood. You need provide both input data and the kernel.</p> |
| <p>Input data is expected to be NCHW.</p> |
| <p><cite>num_filter</cite> is expected to be same as the number of channels.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]]]]</span> |
| |
| <span class="n">w</span> <span class="o">=</span> <span class="p">[[[[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]]]]</span> |
| |
| <span class="n">UpSampling</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">w</span><span class="p">,</span> <span class="n">scale</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">sample_type</span><span class="o">=</span><span class="s1">'bilinear'</span><span class="p">,</span> <span class="n">num_filter</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[[</span><span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">2.</span> <span class="mf">2.</span> <span class="mf">2.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">2.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">2.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">2.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">2.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">2.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">2.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">2.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">2.</span><span class="p">]</span> |
| <span class="p">[</span><span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">2.</span> <span class="mf">2.</span> <span class="mf">2.</span> <span class="mf">1.</span><span class="p">]]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/nn/upsampling.cc:L172 |
| This function support variable length of positional input.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Array of tensors to upsample. For bilinear upsampling, there should be 2 inputs - 1 data and 1 weight.</p></li> |
| <li><p><strong>scale</strong> (<em>int</em><em>, </em><em>required</em>) – Up sampling scale</p></li> |
| <li><p><strong>num_filter</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – Input filter. Only used by bilinear sample_type.Since bilinear upsampling uses deconvolution, num_filters is set to the number of channels.</p></li> |
| <li><p><strong>sample_type</strong> (<em>{'bilinear'</em><em>, </em><em>'nearest'}</em><em>, </em><em>required</em>) – upsampling method</p></li> |
| <li><p><strong>multi_input_mode</strong> (<em>{'concat'</em><em>, </em><em>'sum'}</em><em>,</em><em>optional</em><em>, </em><em>default='concat'</em>) – How to handle multiple input. concat means concatenate upsampled images along the channel dimension. sum means add all images together, only available for nearest neighbor upsampling.</p></li> |
| <li><p><strong>workspace</strong> (<em>long</em><em> (</em><em>non-negative</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=512</em>) – Tmp workspace for deconvolution (MB)</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.abs"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">abs</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.abs" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise absolute value of the input.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="nb">abs</span><span class="p">([</span><span class="o">-</span><span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">3</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">3</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">abs</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>abs(default) = default</p></li> |
| <li><p>abs(row_sparse) = row_sparse</p></li> |
| <li><p>abs(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L720</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.adam_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">adam_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">mean=None</em>, <em class="sig-param">var=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">beta1=_Null</em>, <em class="sig-param">beta2=_Null</em>, <em class="sig-param">epsilon=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">lazy_update=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.adam_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for Adam optimizer. Adam is seen as a generalization |
| of AdaGrad.</p> |
| <p>Adam update consists of the following steps, where g represents gradient and m, v |
| are 1st and 2nd order moment estimates (mean and variance).</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}g_t = \nabla J(W_{t-1})\\ |
| m_t = \beta_1 m_{t-1} + (1 - \beta_1) g_t\\ |
| v_t = \beta_2 v_{t-1} + (1 - \beta_2) g_t^2\\ |
| W_t = W_{t-1} - \alpha \frac{ m_t }{ \sqrt{ v_t } + \epsilon }\end{split}\]</div> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">m</span> <span class="o">=</span> <span class="n">beta1</span><span class="o">*</span><span class="n">m</span> <span class="o">+</span> <span class="p">(</span><span class="mi">1</span><span class="o">-</span><span class="n">beta1</span><span class="p">)</span><span class="o">*</span><span class="n">grad</span> |
| <span class="n">v</span> <span class="o">=</span> <span class="n">beta2</span><span class="o">*</span><span class="n">v</span> <span class="o">+</span> <span class="p">(</span><span class="mi">1</span><span class="o">-</span><span class="n">beta2</span><span class="p">)</span><span class="o">*</span><span class="p">(</span><span class="n">grad</span><span class="o">**</span><span class="mi">2</span><span class="p">)</span> |
| <span class="n">w</span> <span class="o">+=</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="n">m</span> <span class="o">/</span> <span class="p">(</span><span class="n">sqrt</span><span class="p">(</span><span class="n">v</span><span class="p">)</span> <span class="o">+</span> <span class="n">epsilon</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>However, if grad’s storage type is <code class="docutils literal notranslate"><span class="pre">row_sparse</span></code>, <code class="docutils literal notranslate"><span class="pre">lazy_update</span></code> is True and the storage |
| type of weight is the same as those of m and v, |
| only the row slices whose indices appear in grad.indices are updated (for w, m and v):</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="k">for</span> <span class="n">row</span> <span class="ow">in</span> <span class="n">grad</span><span class="o">.</span><span class="n">indices</span><span class="p">:</span> |
| <span class="n">m</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">=</span> <span class="n">beta1</span><span class="o">*</span><span class="n">m</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">+</span> <span class="p">(</span><span class="mi">1</span><span class="o">-</span><span class="n">beta1</span><span class="p">)</span><span class="o">*</span><span class="n">grad</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> |
| <span class="n">v</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">=</span> <span class="n">beta2</span><span class="o">*</span><span class="n">v</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">+</span> <span class="p">(</span><span class="mi">1</span><span class="o">-</span><span class="n">beta2</span><span class="p">)</span><span class="o">*</span><span class="p">(</span><span class="n">grad</span><span class="p">[</span><span class="n">row</span><span class="p">]</span><span class="o">**</span><span class="mi">2</span><span class="p">)</span> |
| <span class="n">w</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">+=</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="n">m</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">/</span> <span class="p">(</span><span class="n">sqrt</span><span class="p">(</span><span class="n">v</span><span class="p">[</span><span class="n">row</span><span class="p">])</span> <span class="o">+</span> <span class="n">epsilon</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/optimizer_op.cc:L687</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>mean</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Moving mean</p></li> |
| <li><p><strong>var</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Moving variance</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>beta1</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.899999976</em>) – The decay rate for the 1st moment estimates.</p></li> |
| <li><p><strong>beta2</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.999000013</em>) – The decay rate for the 2nd moment estimates.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=9.99999994e-09</em>) – A small constant for numerical stability.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>lazy_update</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – If true, lazy updates are applied if gradient’s stype is row_sparse and all of w, m and v have the same stype</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.add_n"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">add_n</code><span class="sig-paren">(</span><em class="sig-param">*args</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.add_n" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Adds all input arguments element-wise.</p> |
| <div class="math notranslate nohighlight"> |
| \[add\_n(a_1, a_2, ..., a_n) = a_1 + a_2 + ... + a_n\]</div> |
| <p><code class="docutils literal notranslate"><span class="pre">add_n</span></code> is potentially more efficient than calling <code class="docutils literal notranslate"><span class="pre">add</span></code> by <cite>n</cite> times.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">add_n</span></code> output depends on storage types of inputs</p> |
| <ul class="simple"> |
| <li><p>add_n(row_sparse, row_sparse, ..) = row_sparse</p></li> |
| <li><p>add_n(default, csr, default) = default</p></li> |
| <li><p>add_n(any input combinations longer than 4 (>4) with at least one default type) = default</p></li> |
| <li><p>otherwise, <code class="docutils literal notranslate"><span class="pre">add_n</span></code> falls all inputs back to default storage and generates default storage</p></li> |
| </ul> |
| <p>Defined in src/operator/tensor/elemwise_sum.cc:L155 |
| This function support variable length of positional input.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>args</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Positional input arguments</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.all_finite"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">all_finite</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">init_output=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.all_finite" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Check if all the float numbers in the array are finite (used for AMP)</p> |
| <p>Defined in src/operator/contrib/all_finite.cc:L100</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../../ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Array</p></li> |
| <li><p><strong>init_output</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Initialize output to 1.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.amp_cast"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">amp_cast</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.amp_cast" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Cast function between low precision float/FP32 used by AMP.</p> |
| <p>It casts only between low precision float/FP32 and does not do anything for other types.</p> |
| <p>Defined in src/operator/tensor/amp_cast.cc:L125</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'bfloat16'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'</em><em>, </em><em>'int8'</em><em>, </em><em>'uint8'}</em><em>, </em><em>required</em>) – Output data type.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.amp_multicast"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">amp_multicast</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.amp_multicast" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Cast function used by AMP, that casts its inputs to the common widest type.</p> |
| <p>It casts only between low precision float/FP32 and does not do anything for other types.</p> |
| <p>Defined in src/operator/tensor/amp_cast.cc:L169</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Weights</p></li> |
| <li><p><strong>num_outputs</strong> (<em>int</em><em>, </em><em>required</em>) – Number of input/output pairs to be casted to the widest type.</p></li> |
| <li><p><strong>cast_narrow</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to cast to the narrowest type</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.arccos"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">arccos</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.arccos" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise inverse cosine of the input array.</p> |
| <p>The input should be in range <cite>[-1, 1]</cite>. |
| The output is in the closed interval <span class="math notranslate nohighlight">\([0, \pi]\)</span></p> |
| <div class="math notranslate nohighlight"> |
| \[arccos([-1, -.707, 0, .707, 1]) = [\pi, 3\pi/4, \pi/2, \pi/4, 0]\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">arccos</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L233</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.arccosh"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">arccosh</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.arccosh" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the element-wise inverse hyperbolic cosine of the input array, computed element-wise.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">arccosh</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L535</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.arcsin"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">arcsin</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.arcsin" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise inverse sine of the input array.</p> |
| <p>The input should be in the range <cite>[-1, 1]</cite>. |
| The output is in the closed interval of [<span class="math notranslate nohighlight">\(-\pi/2\)</span>, <span class="math notranslate nohighlight">\(\pi/2\)</span>].</p> |
| <div class="math notranslate nohighlight"> |
| \[arcsin([-1, -.707, 0, .707, 1]) = [-\pi/2, -\pi/4, 0, \pi/4, \pi/2]\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">arcsin</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>arcsin(default) = default</p></li> |
| <li><p>arcsin(row_sparse) = row_sparse</p></li> |
| <li><p>arcsin(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L187</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.arcsinh"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">arcsinh</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.arcsinh" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the element-wise inverse hyperbolic sine of the input array, computed element-wise.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">arcsinh</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>arcsinh(default) = default</p></li> |
| <li><p>arcsinh(row_sparse) = row_sparse</p></li> |
| <li><p>arcsinh(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L494</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.arctan"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">arctan</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.arctan" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise inverse tangent of the input array.</p> |
| <p>The output is in the closed interval <span class="math notranslate nohighlight">\([-\pi/2, \pi/2]\)</span></p> |
| <div class="math notranslate nohighlight"> |
| \[arctan([-1, 0, 1]) = [-\pi/4, 0, \pi/4]\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">arctan</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>arctan(default) = default</p></li> |
| <li><p>arctan(row_sparse) = row_sparse</p></li> |
| <li><p>arctan(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L282</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.arctanh"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">arctanh</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.arctanh" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the element-wise inverse hyperbolic tangent of the input array, computed element-wise.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">arctanh</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>arctanh(default) = default</p></li> |
| <li><p>arctanh(row_sparse) = row_sparse</p></li> |
| <li><p>arctanh(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L579</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.argmax"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">argmax</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.argmax" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns indices of the maximum values along an axis.</p> |
| <p>In the case of multiple occurrences of maximum values, the indices corresponding to the first occurrence |
| are returned.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">argmax</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">0</span> |
| <span class="n">argmax</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">argmax</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> |
| <span class="n">argmax</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">argmax</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> <span class="n">keeping</span> <span class="n">same</span> <span class="n">dims</span> <span class="k">as</span> <span class="n">an</span> <span class="nb">input</span> <span class="n">array</span> |
| <span class="n">argmax</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">keepdims</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L51</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – The axis along which to perform the reduction. Negative values means indexing from right to left. <code class="docutils literal notranslate"><span class="pre">Requires</span> <span class="pre">axis</span> <span class="pre">to</span> <span class="pre">be</span> <span class="pre">set</span> <span class="pre">as</span> <span class="pre">int,</span> <span class="pre">because</span> <span class="pre">global</span> <span class="pre">reduction</span> <span class="pre">is</span> <span class="pre">not</span> <span class="pre">supported</span> <span class="pre">yet.</span></code></p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axis is left in the result as dimension with size one.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.argmax_channel"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">argmax_channel</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.argmax_channel" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns argmax indices of each channel from the input array.</p> |
| <p>The result will be an NDArray of shape (num_channel,).</p> |
| <p>In case of multiple occurrences of the maximum values, the indices corresponding to the first occurrence |
| are returned.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]]</span> |
| |
| <span class="n">argmax_channel</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L96</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.argmin"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">argmin</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.argmin" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns indices of the minimum values along an axis.</p> |
| <p>In the case of multiple occurrences of minimum values, the indices corresponding to the first occurrence |
| are returned.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">argmin</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">0</span> |
| <span class="n">argmin</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">argmin</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> |
| <span class="n">argmin</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">argmin</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> <span class="n">keeping</span> <span class="n">same</span> <span class="n">dims</span> <span class="k">as</span> <span class="n">an</span> <span class="nb">input</span> <span class="n">array</span> |
| <span class="n">argmin</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">keepdims</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L76</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – The axis along which to perform the reduction. Negative values means indexing from right to left. <code class="docutils literal notranslate"><span class="pre">Requires</span> <span class="pre">axis</span> <span class="pre">to</span> <span class="pre">be</span> <span class="pre">set</span> <span class="pre">as</span> <span class="pre">int,</span> <span class="pre">because</span> <span class="pre">global</span> <span class="pre">reduction</span> <span class="pre">is</span> <span class="pre">not</span> <span class="pre">supported</span> <span class="pre">yet.</span></code></p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axis is left in the result as dimension with size one.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.argsort"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">argsort</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">is_ascend=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.argsort" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the indices that would sort an input array along the given axis.</p> |
| <p>This function performs sorting along the given axis and returns an array of indices having same shape |
| as an input array that index data in sorted order.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.4</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">sort</span> <span class="n">along</span> <span class="n">axis</span> <span class="o">-</span><span class="mi">1</span> |
| <span class="n">argsort</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">sort</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">0</span> |
| <span class="n">argsort</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">flatten</span> <span class="ow">and</span> <span class="n">then</span> <span class="n">sort</span> |
| <span class="n">argsort</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="kc">None</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/ordering_op.cc:L184</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='-1'</em>) – Axis along which to sort the input tensor. If not given, the flattened array is used. Default is -1.</p></li> |
| <li><p><strong>is_ascend</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Whether to sort in ascending or descending order.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'</em><em>, </em><em>'uint8'}</em><em>,</em><em>optional</em><em>, </em><em>default='float32'</em>) – DType of the output indices. It is only valid when ret_typ is “indices” or “both”. An error will be raised if the selected data type cannot precisely represent the indices.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.batch_dot"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">batch_dot</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">transpose_a=_Null</em>, <em class="sig-param">transpose_b=_Null</em>, <em class="sig-param">forward_stype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.batch_dot" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Batchwise dot product.</p> |
| <p><code class="docutils literal notranslate"><span class="pre">batch_dot</span></code> is used to compute dot product of <code class="docutils literal notranslate"><span class="pre">x</span></code> and <code class="docutils literal notranslate"><span class="pre">y</span></code> when <code class="docutils literal notranslate"><span class="pre">x</span></code> and |
| <code class="docutils literal notranslate"><span class="pre">y</span></code> are data in batch, namely N-D (N >= 3) arrays in shape of <cite>(B0, …, B_i, :, :)</cite>.</p> |
| <p>For example, given <code class="docutils literal notranslate"><span class="pre">x</span></code> with shape <cite>(B_0, …, B_i, N, M)</cite> and <code class="docutils literal notranslate"><span class="pre">y</span></code> with shape |
| <cite>(B_0, …, B_i, M, K)</cite>, the result array will have shape <cite>(B_0, …, B_i, N, K)</cite>, |
| which is computed by:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">batch_dot</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">y</span><span class="p">)[</span><span class="n">b_0</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">b_i</span><span class="p">,</span> <span class="p">:,</span> <span class="p">:]</span> <span class="o">=</span> <span class="n">dot</span><span class="p">(</span><span class="n">x</span><span class="p">[</span><span class="n">b_0</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">b_i</span><span class="p">,</span> <span class="p">:,</span> <span class="p">:],</span> <span class="n">y</span><span class="p">[</span><span class="n">b_0</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">b_i</span><span class="p">,</span> <span class="p">:,</span> <span class="p">:])</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/dot.cc:L127</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The first input</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The second input</p></li> |
| <li><p><strong>transpose_a</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true then transpose the first input before dot.</p></li> |
| <li><p><strong>transpose_b</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true then transpose the second input before dot.</p></li> |
| <li><p><strong>forward_stype</strong> (<em>{None</em><em>, </em><em>'csr'</em><em>, </em><em>'default'</em><em>, </em><em>'row_sparse'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – The desired storage type of the forward output given by user, if thecombination of input storage types and this hint does not matchany implemented ones, the dot operator will perform fallback operationand still produce an output of the desired storage type.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.batch_take"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">batch_take</code><span class="sig-paren">(</span><em class="sig-param">a=None</em>, <em class="sig-param">indices=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.batch_take" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Takes elements from a data batch.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><cite>batch_take</cite> is deprecated. Use <cite>pick</cite> instead.</p> |
| </div> |
| <p>Given an input array of shape <code class="docutils literal notranslate"><span class="pre">(d0,</span> <span class="pre">d1)</span></code> and indices of shape <code class="docutils literal notranslate"><span class="pre">(i0,)</span></code>, the result will be |
| an output array of shape <code class="docutils literal notranslate"><span class="pre">(i0,)</span></code> with:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">output</span><span class="p">[</span><span class="n">i</span><span class="p">]</span> <span class="o">=</span> <span class="nb">input</span><span class="p">[</span><span class="n">i</span><span class="p">,</span> <span class="n">indices</span><span class="p">[</span><span class="n">i</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">takes</span> <span class="n">elements</span> <span class="k">with</span> <span class="n">specified</span> <span class="n">indices</span> |
| <span class="n">batch_take</span><span class="p">(</span><span class="n">x</span><span class="p">,</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">0</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span> <span class="mf">4.</span> <span class="mf">5.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/indexing_op.cc:L835</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>a</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array</p></li> |
| <li><p><strong>indices</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The index array</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_add"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_add</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_add" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise sum of the input arrays with broadcasting.</p> |
| <p><cite>broadcast_plus</cite> is an alias to the function <cite>broadcast_add</cite>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_add</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_plus</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Supported sparse operations:</p> |
| <blockquote> |
| <div><p>broadcast_add(csr, dense(1D)) = dense |
| broadcast_add(dense(1D), csr) = dense</p> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L57</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_axes"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_axes</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">size=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_axes" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Broadcasts the input array over particular axes.</p> |
| <p>Broadcasting is allowed on axes with size 1, such as from <cite>(2,1,3,1)</cite> to |
| <cite>(2,8,3,9)</cite>. Elements will be duplicated on the broadcasted axes.</p> |
| <p><cite>broadcast_axes</cite> is an alias to the function <cite>broadcast_axis</cite>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="o">//</span> <span class="n">given</span> <span class="n">x</span> <span class="n">of</span> <span class="n">shape</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">1</span><span class="p">)</span> |
| <span class="n">x</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">broadcast</span> <span class="n">x</span> <span class="n">on</span> <span class="n">on</span> <span class="n">axis</span> <span class="mi">2</span> |
| <span class="n">broadcast_axis</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">size</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]]</span> |
| <span class="o">//</span> <span class="n">broadcast</span> <span class="n">x</span> <span class="n">on</span> <span class="n">on</span> <span class="n">axes</span> <span class="mi">0</span> <span class="ow">and</span> <span class="mi">2</span> |
| <span class="n">broadcast_axis</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span><span class="mi">2</span><span class="p">),</span> <span class="n">size</span><span class="o">=</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="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]],</span> |
| <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L92</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The axes to perform the broadcasting.</p></li> |
| <li><p><strong>size</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Target sizes of the broadcasting axes.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_axis"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_axis</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">size=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_axis" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Broadcasts the input array over particular axes.</p> |
| <p>Broadcasting is allowed on axes with size 1, such as from <cite>(2,1,3,1)</cite> to |
| <cite>(2,8,3,9)</cite>. Elements will be duplicated on the broadcasted axes.</p> |
| <p><cite>broadcast_axes</cite> is an alias to the function <cite>broadcast_axis</cite>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="o">//</span> <span class="n">given</span> <span class="n">x</span> <span class="n">of</span> <span class="n">shape</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">1</span><span class="p">)</span> |
| <span class="n">x</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">broadcast</span> <span class="n">x</span> <span class="n">on</span> <span class="n">on</span> <span class="n">axis</span> <span class="mi">2</span> |
| <span class="n">broadcast_axis</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">size</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]]</span> |
| <span class="o">//</span> <span class="n">broadcast</span> <span class="n">x</span> <span class="n">on</span> <span class="n">on</span> <span class="n">axes</span> <span class="mi">0</span> <span class="ow">and</span> <span class="mi">2</span> |
| <span class="n">broadcast_axis</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span><span class="mi">2</span><span class="p">),</span> <span class="n">size</span><span class="o">=</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="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]],</span> |
| <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L92</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The axes to perform the broadcasting.</p></li> |
| <li><p><strong>size</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Target sizes of the broadcasting axes.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_div"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_div</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_div" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise division of the input arrays with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_div</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Supported sparse operations:</p> |
| <blockquote> |
| <div><p>broadcast_div(csr, dense(1D)) = csr</p> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L186</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_equal"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_equal</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_equal" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of element-wise <strong>equal to</strong> (==) comparison operation with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_equal</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L45</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_greater"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_greater</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_greater" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of element-wise <strong>greater than</strong> (>) comparison operation with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_greater</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L81</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_greater_equal"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_greater_equal</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_greater_equal" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of element-wise <strong>greater than or equal to</strong> (>=) comparison operation with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_greater_equal</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L99</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_hypot"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_hypot</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_hypot" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the hypotenuse of a right angled triangle, given its “legs” |
| with broadcasting.</p> |
| <p>It is equivalent to doing <span class="math notranslate nohighlight">\(sqrt(x_1^2 + x_2^2)\)</span>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_hypot</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]]</span> |
| |
| <span class="n">z</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_hypot</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">z</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L157</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_lesser"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_lesser</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_lesser" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of element-wise <strong>lesser than</strong> (<) comparison operation with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_lesser</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L117</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_lesser_equal"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_lesser_equal</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_lesser_equal" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of element-wise <strong>lesser than or equal to</strong> (<=) comparison operation with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_lesser_equal</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L135</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_like"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_like</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">lhs_axes=_Null</em>, <em class="sig-param">rhs_axes=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_like" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Broadcasts lhs to have the same shape as rhs.</p> |
| <p>Broadcasting is a mechanism that allows NDArrays to perform arithmetic operations |
| with arrays of different shapes efficiently without creating multiple copies of arrays. |
| Also see, <a class="reference external" href="https://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">Broadcasting</a> for more explanation.</p> |
| <p>Broadcasting is allowed on axes with size 1, such as from <cite>(2,1,3,1)</cite> to |
| <cite>(2,8,3,9)</cite>. Elements will be duplicated on the broadcasted axes.</p> |
| <p>For example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">broadcast_like</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="p">[[</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">,</span><span class="mi">7</span><span class="p">],[</span><span class="mi">7</span><span class="p">,</span><span class="mi">8</span><span class="p">,</span><span class="mi">9</span><span class="p">]])</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]])</span> |
| |
| <span class="n">broadcast_like</span><span class="p">([</span><span class="mi">9</span><span class="p">],</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="n">lhs_axes</span><span class="o">=</span><span class="p">(</span><span class="mi">0</span><span class="p">,),</span> <span class="n">rhs_axes</span><span class="o">=</span><span class="p">(</span><span class="o">-</span><span class="mi">1</span><span class="p">,))</span> <span class="o">=</span> <span class="p">[</span><span class="mi">9</span><span class="p">,</span><span class="mi">9</span><span class="p">,</span><span class="mi">9</span><span class="p">,</span><span class="mi">9</span><span class="p">,</span><span class="mi">9</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L178</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input.</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input.</p></li> |
| <li><p><strong>lhs_axes</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Axes to perform broadcast on in the first input array</p></li> |
| <li><p><strong>rhs_axes</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Axes to copy from the second input array</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_logical_and"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_logical_and</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_logical_and" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of element-wise <strong>logical and</strong> with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_logical_and</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L153</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_logical_or"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_logical_or</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_logical_or" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of element-wise <strong>logical or</strong> with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_logical_or</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L171</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_logical_xor"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_logical_xor</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_logical_xor" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of element-wise <strong>logical xor</strong> with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_logical_xor</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L189</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_maximum"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_maximum</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_maximum" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise maximum of the input arrays with broadcasting.</p> |
| <p>This function compares two input arrays and returns a new array having the element-wise maxima.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_maximum</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L80</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_minimum"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_minimum</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_minimum" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise minimum of the input arrays with broadcasting.</p> |
| <p>This function compares two input arrays and returns a new array having the element-wise minima.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_maximum</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L116</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_minus"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_minus</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_minus" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise difference of the input arrays with broadcasting.</p> |
| <p><cite>broadcast_minus</cite> is an alias to the function <cite>broadcast_sub</cite>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_sub</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_minus</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Supported sparse operations:</p> |
| <blockquote> |
| <div><p>broadcast_sub/minus(csr, dense(1D)) = dense |
| broadcast_sub/minus(dense(1D), csr) = dense</p> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L105</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_mod"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_mod</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_mod" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise modulo of the input arrays with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_mod</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L221</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_mul"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_mul</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_mul" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise product of the input arrays with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_mul</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Supported sparse operations:</p> |
| <blockquote> |
| <div><p>broadcast_mul(csr, dense(1D)) = csr</p> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L145</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_not_equal"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_not_equal</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_not_equal" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of element-wise <strong>not equal to</strong> (!=) comparison operation with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_not_equal</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L63</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_plus"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_plus</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_plus" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise sum of the input arrays with broadcasting.</p> |
| <p><cite>broadcast_plus</cite> is an alias to the function <cite>broadcast_add</cite>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_add</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_plus</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Supported sparse operations:</p> |
| <blockquote> |
| <div><p>broadcast_add(csr, dense(1D)) = dense |
| broadcast_add(dense(1D), csr) = dense</p> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L57</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_power"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_power</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_power" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns result of first array elements raised to powers from second array, element-wise with broadcasting.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_power</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L44</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_sub"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_sub</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_sub" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise difference of the input arrays with broadcasting.</p> |
| <p><cite>broadcast_minus</cite> is an alias to the function <cite>broadcast_sub</cite>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_sub</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="n">broadcast_minus</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Supported sparse operations:</p> |
| <blockquote> |
| <div><p>broadcast_sub/minus(csr, dense(1D)) = dense |
| broadcast_sub/minus(dense(1D), csr) = dense</p> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L105</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input to the function</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input to the function</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.broadcast_to"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">broadcast_to</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.broadcast_to" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Broadcasts the input array to a new shape.</p> |
| <p>Broadcasting is a mechanism that allows NDArrays to perform arithmetic operations |
| with arrays of different shapes efficiently without creating multiple copies of arrays. |
| Also see, <a class="reference external" href="https://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">Broadcasting</a> for more explanation.</p> |
| <p>Broadcasting is allowed on axes with size 1, such as from <cite>(2,1,3,1)</cite> to |
| <cite>(2,8,3,9)</cite>. Elements will be duplicated on the broadcasted axes.</p> |
| <p>For example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">broadcast_to</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="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]])</span> |
| </pre></div> |
| </div> |
| <p>The dimension which you do not want to change can also be kept as <cite>0</cite> which means copy the original value. |
| So with <cite>shape=(2,0)</cite>, we will obtain the same result as in the above example.</p> |
| <p>Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L116</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The shape of the desired array. We can set the dim to zero if it’s same as the original. E.g <cite>A = broadcast_to(B, shape=(10, 0, 0))</cite> has the same meaning as <cite>A = broadcast_axis(B, axis=0, size=10)</cite>.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.cast"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">cast</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.cast" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Casts all elements of the input to a new type.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><code class="docutils literal notranslate"><span class="pre">Cast</span></code> is deprecated. Use <code class="docutils literal notranslate"><span class="pre">cast</span></code> instead.</p> |
| </div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">cast</span><span class="p">([</span><span class="mf">0.9</span><span class="p">,</span> <span class="mf">1.3</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="s1">'int32'</span><span class="p">)</span> <span class="o">=</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="n">cast</span><span class="p">([</span><span class="mf">1e20</span><span class="p">,</span> <span class="mf">11.1</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="s1">'float16'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="n">inf</span><span class="p">,</span> <span class="mf">11.09375</span><span class="p">]</span> |
| <span class="n">cast</span><span class="p">([</span><span class="mi">300</span><span class="p">,</span> <span class="mf">11.1</span><span class="p">,</span> <span class="mf">10.9</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">3</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="s1">'uint8'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mi">44</span><span class="p">,</span> <span class="mi">11</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">255</span><span class="p">,</span> <span class="mi">253</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L664</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'bfloat16'</em><em>, </em><em>'bool'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'</em><em>, </em><em>'int8'</em><em>, </em><em>'uint8'}</em><em>, </em><em>required</em>) – Output data type.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.cast_storage"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">cast_storage</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">stype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.cast_storage" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Casts tensor storage type to the new type.</p> |
| <p>When an NDArray with default storage type is cast to csr or row_sparse storage, |
| the result is compact, which means:</p> |
| <ul class="simple"> |
| <li><p>for csr, zero values will not be retained</p></li> |
| <li><p>for row_sparse, row slices of all zeros will not be retained</p></li> |
| </ul> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">cast_storage</span></code> output depends on stype parameter:</p> |
| <ul class="simple"> |
| <li><p>cast_storage(csr, ‘default’) = default</p></li> |
| <li><p>cast_storage(row_sparse, ‘default’) = default</p></li> |
| <li><p>cast_storage(default, ‘csr’) = csr</p></li> |
| <li><p>cast_storage(default, ‘row_sparse’) = row_sparse</p></li> |
| <li><p>cast_storage(csr, ‘csr’) = csr</p></li> |
| <li><p>cast_storage(row_sparse, ‘row_sparse’) = row_sparse</p></li> |
| </ul> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dense</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="c1"># cast to row_sparse storage type</span> |
| <span class="n">rsp</span> <span class="o">=</span> <span class="n">cast_storage</span><span class="p">(</span><span class="n">dense</span><span class="p">,</span> <span class="s1">'row_sparse'</span><span class="p">)</span> |
| <span class="n">rsp</span><span class="o">.</span><span class="n">indices</span> <span class="o">=</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="n">rsp</span><span class="o">.</span><span class="n">values</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]]</span> |
| |
| <span class="c1"># cast to csr storage type</span> |
| <span class="n">csr</span> <span class="o">=</span> <span class="n">cast_storage</span><span class="p">(</span><span class="n">dense</span><span class="p">,</span> <span class="s1">'csr'</span><span class="p">)</span> |
| <span class="n">csr</span><span class="o">.</span><span class="n">indices</span> <span class="o">=</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">]</span> |
| <span class="n">csr</span><span class="o">.</span><span class="n">values</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="n">csr</span><span class="o">.</span><span class="n">indptr</span> <span class="o">=</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">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/cast_storage.cc:L71</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input.</p></li> |
| <li><p><strong>stype</strong> (<em>{'csr'</em><em>, </em><em>'default'</em><em>, </em><em>'row_sparse'}</em><em>, </em><em>required</em>) – Output storage type.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.cbrt"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">cbrt</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.cbrt" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise cube-root value of the input.</p> |
| <div class="math notranslate nohighlight"> |
| \[cbrt(x) = \sqrt[3]{x}\]</div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">cbrt</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="o">-</span><span class="mi">125</span><span class="p">])</span> <span class="o">=</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="o">-</span><span class="mi">5</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">cbrt</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>cbrt(default) = default</p></li> |
| <li><p>cbrt(row_sparse) = row_sparse</p></li> |
| <li><p>cbrt(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L270</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.ceil"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">ceil</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.ceil" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise ceiling of the input.</p> |
| <p>The ceil of the scalar x is the smallest integer i, such that i >= x.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">ceil</span><span class="p">([</span><span class="o">-</span><span class="mf">2.1</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.9</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">,</span> <span class="mf">1.9</span><span class="p">,</span> <span class="mf">2.1</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">2.</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">ceil</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>ceil(default) = default</p></li> |
| <li><p>ceil(row_sparse) = row_sparse</p></li> |
| <li><p>ceil(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L817</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.choose_element_0index"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">choose_element_0index</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">index=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">mode=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.choose_element_0index" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Picks elements from an input array according to the input indices along the given axis.</p> |
| <p>Given an input array of shape <code class="docutils literal notranslate"><span class="pre">(d0,</span> <span class="pre">d1)</span></code> and indices of shape <code class="docutils literal notranslate"><span class="pre">(i0,)</span></code>, the result will be |
| an output array of shape <code class="docutils literal notranslate"><span class="pre">(i0,)</span></code> with:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">output</span><span class="p">[</span><span class="n">i</span><span class="p">]</span> <span class="o">=</span> <span class="nb">input</span><span class="p">[</span><span class="n">i</span><span class="p">,</span> <span class="n">indices</span><span class="p">[</span><span class="n">i</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>By default, if any index mentioned is too large, it is replaced by the index that addresses |
| the last element along an axis (the <cite>clip</cite> mode).</p> |
| <p>This function supports n-dimensional input and (n-1)-dimensional indices arrays.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">picks</span> <span class="n">elements</span> <span class="k">with</span> <span class="n">specified</span> <span class="n">indices</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">0</span> |
| <span class="n">pick</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="o">=</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">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">picks</span> <span class="n">elements</span> <span class="k">with</span> <span class="n">specified</span> <span class="n">indices</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> |
| <span class="n">pick</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="o">=</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">0</span><span class="p">],</span> <span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">picks</span> <span class="n">elements</span> <span class="k">with</span> <span class="n">specified</span> <span class="n">indices</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> <span class="n">using</span> <span class="s1">'wrap'</span> <span class="n">mode</span> |
| <span class="o">//</span> <span class="n">to</span> <span class="n">place</span> <span class="n">indicies</span> <span class="n">that</span> <span class="n">would</span> <span class="n">normally</span> <span class="n">be</span> <span class="n">out</span> <span class="n">of</span> <span class="n">bounds</span> |
| <span class="n">pick</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="o">=</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">2</span><span class="p">],</span> <span class="mi">1</span><span class="p">,</span> <span class="n">mode</span><span class="o">=</span><span class="s1">'wrap'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">picks</span> <span class="n">elements</span> <span class="k">with</span> <span class="n">specified</span> <span class="n">indices</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> <span class="ow">and</span> <span class="n">dims</span> <span class="n">are</span> <span class="n">maintained</span> |
| <span class="n">pick</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="n">keepdims</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L150</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array</p></li> |
| <li><p><strong>index</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The index array</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='-1'</em>) – int or None. The axis to picking the elements. Negative values means indexing from right to left. If is <cite>None</cite>, the elements in the index w.r.t the flattened input will be picked.</p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true, the axis where we pick the elements is left in the result as dimension with size one.</p></li> |
| <li><p><strong>mode</strong> (<em>{'clip'</em><em>, </em><em>'wrap'}</em><em>,</em><em>optional</em><em>, </em><em>default='clip'</em>) – Specify how out-of-bound indices behave. Default is “clip”. “clip” means clip to the range. So, if all indices mentioned are too large, they are replaced by the index that addresses the last element along an axis. “wrap” means to wrap around.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.clip"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">clip</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">a_min=_Null</em>, <em class="sig-param">a_max=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.clip" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Clips (limits) the values in an array. |
| Given an interval, values outside the interval are clipped to the interval edges. |
| Clipping <code class="docutils literal notranslate"><span class="pre">x</span></code> between <cite>a_min</cite> and <cite>a_max</cite> would be:: |
| .. math:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">clip</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">a_min</span><span class="p">,</span> <span class="n">a_max</span><span class="p">)</span> <span class="o">=</span> \<span class="nb">max</span><span class="p">(</span>\<span class="nb">min</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">a_max</span><span class="p">),</span> <span class="n">a_min</span><span class="p">))</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>Example::</dt><dd><p>x = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9] |
| clip(x,1,8) = [ 1., 1., 2., 3., 4., 5., 6., 7., 8., 8.]</p> |
| </dd> |
| </dl> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">clip</span></code> output depends on storage types of inputs and the a_min, a_max parameter values:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>clip(default) = default</p></li> |
| <li><p>clip(row_sparse, a_min <= 0, a_max >= 0) = row_sparse</p></li> |
| <li><p>clip(csr, a_min <= 0, a_max >= 0) = csr</p></li> |
| <li><p>clip(row_sparse, a_min < 0, a_max < 0) = default</p></li> |
| <li><p>clip(row_sparse, a_min > 0, a_max > 0) = default</p></li> |
| <li><p>clip(csr, a_min < 0, a_max < 0) = csr</p></li> |
| <li><p>clip(csr, a_min > 0, a_max > 0) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L676</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array.</p></li> |
| <li><p><strong>a_min</strong> (<em>float</em><em>, </em><em>required</em>) – Minimum value</p></li> |
| <li><p><strong>a_max</strong> (<em>float</em><em>, </em><em>required</em>) – Maximum value</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.col2im"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">col2im</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">output_size=_Null</em>, <em class="sig-param">kernel=_Null</em>, <em class="sig-param">stride=_Null</em>, <em class="sig-param">dilate=_Null</em>, <em class="sig-param">pad=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.col2im" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Combining the output column matrix of im2col back to image array.</p> |
| <p>Like <a class="reference internal" href="../../ndarray/ndarray.html#mxnet.ndarray.im2col" title="mxnet.ndarray.im2col"><code class="xref py py-class docutils literal notranslate"><span class="pre">im2col</span></code></a>, this operator is also used in the vanilla convolution |
| implementation. Despite the name, col2im is not the reverse operation of im2col. Since there |
| may be overlaps between neighbouring sliding blocks, the column elements cannot be directly |
| put back into image. Instead, they are accumulated (i.e., summed) in the input image |
| just like the gradient computation, so col2im is the gradient of im2col and vice versa.</p> |
| <p>Using the notation in im2col, given an input column array of shape |
| <span class="math notranslate nohighlight">\((N, C \times \prod(\text{kernel}), W)\)</span>, this operator accumulates the column elements |
| into output array of shape <span class="math notranslate nohighlight">\((N, C, \text{output_size}[0], \text{output_size}[1], \dots)\)</span>. |
| Only 1-D, 2-D and 3-D of spatial dimension is supported in this operator.</p> |
| <p>Defined in src/operator/nn/im2col.cc:L181</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array to combine sliding blocks.</p></li> |
| <li><p><strong>output_size</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – The spatial dimension of image array: (w,), (h, w) or (d, h, w).</p></li> |
| <li><p><strong>kernel</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – Sliding kernel size: (w,), (h, w) or (d, h, w).</p></li> |
| <li><p><strong>stride</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The stride between adjacent sliding blocks in spatial dimension: (w,), (h, w) or (d, h, w). Defaults to 1 for each dimension.</p></li> |
| <li><p><strong>dilate</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The spacing between adjacent kernel points: (w,), (h, w) or (d, h, w). Defaults to 1 for each dimension.</p></li> |
| <li><p><strong>pad</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The zero-value padding size on both sides of spatial dimension: (w,), (h, w) or (d, h, w). Defaults to no padding.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.concat"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">concat</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.concat" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Joins input arrays along a given axis.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><cite>Concat</cite> is deprecated. Use <cite>concat</cite> instead.</p> |
| </div> |
| <p>The dimensions of the input arrays should be the same except the axis along |
| which they will be concatenated. |
| The dimension of the output array along the concatenated axis will be equal |
| to the sum of the corresponding dimensions of the input arrays.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">concat</span></code> output depends on storage types of inputs</p> |
| <ul class="simple"> |
| <li><p>concat(csr, csr, …, csr, dim=0) = csr</p></li> |
| <li><p>otherwise, <code class="docutils literal notranslate"><span class="pre">concat</span></code> generates output with default storage</p></li> |
| </ul> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="n">y</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">3</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">4</span><span class="p">],[</span><span class="mi">5</span><span class="p">,</span><span class="mi">5</span><span class="p">]]</span> |
| <span class="n">z</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">6</span><span class="p">,</span><span class="mi">6</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span><span class="mi">7</span><span class="p">],[</span><span class="mi">8</span><span class="p">,</span><span class="mi">8</span><span class="p">]]</span> |
| |
| <span class="n">concat</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">y</span><span class="p">,</span><span class="n">z</span><span class="p">,</span><span class="n">dim</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]</span> |
| |
| <span class="n">Note</span> <span class="n">that</span> <span class="n">you</span> <span class="n">cannot</span> <span class="n">concat</span> <span class="n">x</span><span class="p">,</span><span class="n">y</span><span class="p">,</span><span class="n">z</span> <span class="n">along</span> <span class="n">dimension</span> <span class="mi">1</span> <span class="n">since</span> <span class="n">dimension</span> |
| <span class="mi">0</span> <span class="ow">is</span> <span class="ow">not</span> <span class="n">the</span> <span class="n">same</span> <span class="k">for</span> <span class="nb">all</span> <span class="n">the</span> <span class="nb">input</span> <span class="n">arrays</span><span class="o">.</span> |
| |
| <span class="n">concat</span><span class="p">(</span><span class="n">y</span><span class="p">,</span><span class="n">z</span><span class="p">,</span><span class="n">dim</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/nn/concat.cc:L384 |
| This function support variable length of positional input.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – List of arrays to concatenate</p></li> |
| <li><p><strong>dim</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – the dimension to be concated.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.cos"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">cos</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.cos" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the element-wise cosine of the input array.</p> |
| <p>The input should be in radians (<span class="math notranslate nohighlight">\(2\pi\)</span> rad equals 360 degrees).</p> |
| <div class="math notranslate nohighlight"> |
| \[cos([0, \pi/4, \pi/2]) = [1, 0.707, 0]\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">cos</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L90</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.cosh"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">cosh</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.cosh" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the hyperbolic cosine of the input array, computed element-wise.</p> |
| <div class="math notranslate nohighlight"> |
| \[cosh(x) = 0.5\times(exp(x) + exp(-x))\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">cosh</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L409</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.crop"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">crop</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">begin=_Null</em>, <em class="sig-param">end=_Null</em>, <em class="sig-param">step=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.crop" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Slices a region of the array. |
| .. note:: <code class="docutils literal notranslate"><span class="pre">crop</span></code> is deprecated. Use <code class="docutils literal notranslate"><span class="pre">slice</span></code> instead. |
| This function returns a sliced array between the indices given |
| by <cite>begin</cite> and <cite>end</cite> with the corresponding <cite>step</cite>. |
| For an input array of <code class="docutils literal notranslate"><span class="pre">shape=(d_0,</span> <span class="pre">d_1,</span> <span class="pre">...,</span> <span class="pre">d_n-1)</span></code>, |
| slice operation with <code class="docutils literal notranslate"><span class="pre">begin=(b_0,</span> <span class="pre">b_1...b_m-1)</span></code>, |
| <code class="docutils literal notranslate"><span class="pre">end=(e_0,</span> <span class="pre">e_1,</span> <span class="pre">...,</span> <span class="pre">e_m-1)</span></code>, and <code class="docutils literal notranslate"><span class="pre">step=(s_0,</span> <span class="pre">s_1,</span> <span class="pre">...,</span> <span class="pre">s_m-1)</span></code>, |
| where m <= n, results in an array with the shape |
| <code class="docutils literal notranslate"><span class="pre">(|e_0-b_0|/|s_0|,</span> <span class="pre">...,</span> <span class="pre">|e_m-1-b_m-1|/|s_m-1|,</span> <span class="pre">d_m,</span> <span class="pre">...,</span> <span class="pre">d_n-1)</span></code>. |
| The resulting array’s <em>k</em>-th dimension contains elements |
| from the <em>k</em>-th dimension of the input array starting |
| from index <code class="docutils literal notranslate"><span class="pre">b_k</span></code> (inclusive) with step <code class="docutils literal notranslate"><span class="pre">s_k</span></code> |
| until reaching <code class="docutils literal notranslate"><span class="pre">e_k</span></code> (exclusive). |
| If the <em>k</em>-th elements are <cite>None</cite> in the sequence of <cite>begin</cite>, <cite>end</cite>, |
| and <cite>step</cite>, the following rule will be used to set default values. |
| If <cite>s_k</cite> is <cite>None</cite>, set <cite>s_k=1</cite>. If <cite>s_k > 0</cite>, set <cite>b_k=0</cite>, <cite>e_k=d_k</cite>; |
| else, set <cite>b_k=d_k-1</cite>, <cite>e_k=-1</cite>. |
| The storage type of <code class="docutils literal notranslate"><span class="pre">slice</span></code> output depends on storage types of inputs |
| - slice(csr) = csr |
| - otherwise, <code class="docutils literal notranslate"><span class="pre">slice</span></code> generates output with default storage |
| .. note:: When input data storage type is csr, it only supports</p> |
| <blockquote> |
| <div><p>step=(), or step=(None,), or step=(1,) to generate a csr output. |
| For other step parameter values, it falls back to slicing |
| a dense tensor.</p> |
| </div></blockquote> |
| <dl class="simple"> |
| <dt>Example::</dt><dd><dl class="simple"> |
| <dt>x = [[ 1., 2., 3., 4.],</dt><dd><p>[ 5., 6., 7., 8.], |
| [ 9., 10., 11., 12.]]</p> |
| </dd> |
| <dt>slice(x, begin=(0,1), end=(2,4)) = [[ 2., 3., 4.],</dt><dd><p>[ 6., 7., 8.]]</p> |
| </dd> |
| <dt>slice(x, begin=(None, 0), end=(None, 3), step=(-1, 2)) = [[9., 11.],</dt><dd><p>[5., 7.], |
| [1., 3.]]</p> |
| </dd> |
| </dl> |
| </dd> |
| </dl> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L481</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Source input</p></li> |
| <li><p><strong>begin</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – starting indices for the slice operation, supports negative indices.</p></li> |
| <li><p><strong>end</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – ending indices for the slice operation, supports negative indices.</p></li> |
| <li><p><strong>step</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – step for the slice operation, supports negative values.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.ctc_loss"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">ctc_loss</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">label=None</em>, <em class="sig-param">data_lengths=None</em>, <em class="sig-param">label_lengths=None</em>, <em class="sig-param">use_data_lengths=_Null</em>, <em class="sig-param">use_label_lengths=_Null</em>, <em class="sig-param">blank_label=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.ctc_loss" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Connectionist Temporal Classification Loss.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The existing alias <code class="docutils literal notranslate"><span class="pre">contrib_CTCLoss</span></code> is deprecated.</p> |
| </div> |
| <p>The shapes of the inputs and outputs:</p> |
| <ul class="simple"> |
| <li><p><strong>data</strong>: <cite>(sequence_length, batch_size, alphabet_size)</cite></p></li> |
| <li><p><strong>label</strong>: <cite>(batch_size, label_sequence_length)</cite></p></li> |
| <li><p><strong>out</strong>: <cite>(batch_size)</cite></p></li> |
| </ul> |
| <p>The <cite>data</cite> tensor consists of sequences of activation vectors (without applying softmax), |
| with i-th channel in the last dimension corresponding to i-th label |
| for i between 0 and alphabet_size-1 (i.e always 0-indexed). |
| Alphabet size should include one additional value reserved for blank label. |
| When <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"first"</span></code>, the <code class="docutils literal notranslate"><span class="pre">0</span></code>-th channel is be reserved for |
| activation of blank label, or otherwise if it is “last”, <code class="docutils literal notranslate"><span class="pre">(alphabet_size-1)</span></code>-th channel should be |
| reserved for blank label.</p> |
| <p><code class="docutils literal notranslate"><span class="pre">label</span></code> is an index matrix of integers. When <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"first"</span></code>, |
| the value 0 is then reserved for blank label, and should not be passed in this matrix. Otherwise, |
| when <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"last"</span></code>, the value <cite>(alphabet_size-1)</cite> is reserved for blank label.</p> |
| <p>If a sequence of labels is shorter than <em>label_sequence_length</em>, use the special |
| padding value at the end of the sequence to conform it to the correct |
| length. The padding value is <cite>0</cite> when <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"first"</span></code>, and <cite>-1</cite> otherwise.</p> |
| <p>For example, suppose the vocabulary is <cite>[a, b, c]</cite>, and in one batch we have three sequences |
| ‘ba’, ‘cbb’, and ‘abac’. When <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"first"</span></code>, we can index the labels as |
| <cite>{‘a’: 1, ‘b’: 2, ‘c’: 3}</cite>, and we reserve the 0-th channel for blank label in data tensor. |
| The resulting <cite>label</cite> tensor should be padded to be:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mi">3</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">0</span><span class="p">],</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">1</span><span class="p">,</span> <span class="mi">3</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>When <cite>blank_label</cite> is <code class="docutils literal notranslate"><span class="pre">"last"</span></code>, we can index the labels as |
| <cite>{‘a’: 0, ‘b’: 1, ‘c’: 2}</cite>, and we reserve the channel index 3 for blank label in data tensor. |
| The resulting <cite>label</cite> tensor should be padded to be:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">1</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="p">[</span><span class="mi">2</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="o">-</span><span class="mi">1</span><span class="p">],</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">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p><code class="docutils literal notranslate"><span class="pre">out</span></code> is a list of CTC loss values, one per example in the batch.</p> |
| <p>See <em>Connectionist Temporal Classification: Labelling Unsegmented |
| Sequence Data with Recurrent Neural Networks</em>, A. Graves <em>et al</em>. for more |
| information on the definition and the algorithm.</p> |
| <p>Defined in src/operator/nn/ctc_loss.cc:L100</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input ndarray</p></li> |
| <li><p><strong>label</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Ground-truth labels for the loss.</p></li> |
| <li><p><strong>data_lengths</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lengths of data for each of the samples. Only required when use_data_lengths is true.</p></li> |
| <li><p><strong>label_lengths</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lengths of labels for each of the samples. Only required when use_label_lengths is true.</p></li> |
| <li><p><strong>use_data_lengths</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether the data lenghts are decided by <cite>data_lengths</cite>. If false, the lengths are equal to the max sequence length.</p></li> |
| <li><p><strong>use_label_lengths</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether the label lenghts are decided by <cite>label_lengths</cite>, or derived from <cite>padding_mask</cite>. If false, the lengths are derived from the first occurrence of the value of <cite>padding_mask</cite>. The value of <cite>padding_mask</cite> is <code class="docutils literal notranslate"><span class="pre">0</span></code> when first CTC label is reserved for blank, and <code class="docutils literal notranslate"><span class="pre">-1</span></code> when last label is reserved for blank. See <cite>blank_label</cite>.</p></li> |
| <li><p><strong>blank_label</strong> (<em>{'first'</em><em>, </em><em>'last'}</em><em>,</em><em>optional</em><em>, </em><em>default='first'</em>) – Set the label that is reserved for blank label.If “first”, 0-th label is reserved, and label values for tokens in the vocabulary are between <code class="docutils literal notranslate"><span class="pre">1</span></code> and <code class="docutils literal notranslate"><span class="pre">alphabet_size-1</span></code>, and the padding mask is <code class="docutils literal notranslate"><span class="pre">-1</span></code>. If “last”, last label value <code class="docutils literal notranslate"><span class="pre">alphabet_size-1</span></code> is reserved for blank label instead, and label values for tokens in the vocabulary are between <code class="docutils literal notranslate"><span class="pre">0</span></code> and <code class="docutils literal notranslate"><span class="pre">alphabet_size-2</span></code>, and the padding mask is <code class="docutils literal notranslate"><span class="pre">0</span></code>.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.cumsum"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">cumsum</code><span class="sig-paren">(</span><em class="sig-param">a=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.cumsum" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Return the cumulative sum of the elements along a given axis.</p> |
| <p>Defined in src/operator/numpy/np_cumsum.cc:L70</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>a</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input ndarray</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – Axis along which the cumulative sum is computed. The default (None) is to compute the cumsum over the flattened array.</p></li> |
| <li><p><strong>dtype</strong> (<em>{None</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'</em><em>, </em><em>'int8'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – Type of the returned array and of the accumulator in which the elements are summed. If dtype is not specified, it defaults to the dtype of a, unless a has an integer dtype with a precision less than that of the default platform integer. In that case, the default platform integer is used.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.degrees"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">degrees</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.degrees" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Converts each element of the input array from radians to degrees.</p> |
| <div class="math notranslate nohighlight"> |
| \[degrees([0, \pi/2, \pi, 3\pi/2, 2\pi]) = [0, 90, 180, 270, 360]\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">degrees</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>degrees(default) = default</p></li> |
| <li><p>degrees(row_sparse) = row_sparse</p></li> |
| <li><p>degrees(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L332</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.depth_to_space"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">depth_to_space</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">block_size=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.depth_to_space" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Rearranges(permutes) data from depth into blocks of spatial data. |
| Similar to ONNX DepthToSpace operator: |
| <a class="reference external" href="https://github.com/onnx/onnx/blob/master/docs/Operators.md#DepthToSpace">https://github.com/onnx/onnx/blob/master/docs/Operators.md#DepthToSpace</a>. |
| The output is a new tensor where the values from depth dimension are moved in spatial blocks |
| to height and width dimension. The reverse of this operation is <code class="docutils literal notranslate"><span class="pre">space_to_depth</span></code>. |
| .. math:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span>\<span class="n">begin</span><span class="p">{</span><span class="n">gather</span><span class="o">*</span><span class="p">}</span> |
| <span class="n">x</span> \<span class="n">prime</span> <span class="o">=</span> <span class="n">reshape</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="p">[</span><span class="n">N</span><span class="p">,</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">,</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">,</span> <span class="n">C</span> <span class="o">/</span> <span class="p">(</span><span class="n">block</span>\<span class="n">_size</span> <span class="o">^</span> <span class="mi">2</span><span class="p">),</span> <span class="n">H</span> <span class="o">*</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">,</span> <span class="n">W</span> <span class="o">*</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">])</span> \\ |
| <span class="n">x</span> \<span class="n">prime</span> \<span class="n">prime</span> <span class="o">=</span> <span class="n">transpose</span><span class="p">(</span><span class="n">x</span> \<span class="n">prime</span><span class="p">,</span> <span class="p">[</span><span class="mi">0</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">1</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">2</span><span class="p">])</span> \\ |
| <span class="n">y</span> <span class="o">=</span> <span class="n">reshape</span><span class="p">(</span><span class="n">x</span> \<span class="n">prime</span> \<span class="n">prime</span><span class="p">,</span> <span class="p">[</span><span class="n">N</span><span class="p">,</span> <span class="n">C</span> <span class="o">/</span> <span class="p">(</span><span class="n">block</span>\<span class="n">_size</span> <span class="o">^</span> <span class="mi">2</span><span class="p">),</span> <span class="n">H</span> <span class="o">*</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">,</span> <span class="n">W</span> <span class="o">*</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">])</span> |
| \<span class="n">end</span><span class="p">{</span><span class="n">gather</span><span class="o">*</span><span class="p">}</span> |
| </pre></div> |
| </div> |
| <p>where <span class="math notranslate nohighlight">\(x\)</span> is an input tensor with default layout as <span class="math notranslate nohighlight">\([N, C, H, W]\)</span>: [batch, channels, height, width] |
| and <span class="math notranslate nohighlight">\(y\)</span> is the output tensor of layout <span class="math notranslate nohighlight">\([N, C / (block\_size ^ 2), H * block\_size, W * block\_size]\)</span> |
| Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="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">6</span><span class="p">,</span> <span class="mi">7</span><span class="p">,</span> <span class="mi">8</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">9</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">11</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mi">12</span><span class="p">,</span> <span class="mi">13</span><span class="p">,</span> <span class="mi">14</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">15</span><span class="p">,</span> <span class="mi">16</span><span class="p">,</span> <span class="mi">17</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mi">18</span><span class="p">,</span> <span class="mi">19</span><span class="p">,</span> <span class="mi">20</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">21</span><span class="p">,</span> <span class="mi">22</span><span class="p">,</span> <span class="mi">23</span><span class="p">]]]]</span> |
| <span class="n">depth_to_space</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="mi">2</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">6</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">7</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">8</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">12</span><span class="p">,</span> <span class="mi">18</span><span class="p">,</span> <span class="mi">13</span><span class="p">,</span> <span class="mi">19</span><span class="p">,</span> <span class="mi">14</span><span class="p">,</span> <span class="mi">20</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">9</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">11</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">15</span><span class="p">,</span> <span class="mi">21</span><span class="p">,</span> <span class="mi">16</span><span class="p">,</span> <span class="mi">22</span><span class="p">,</span> <span class="mi">17</span><span class="p">,</span> <span class="mi">23</span><span class="p">]]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L971</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input ndarray</p></li> |
| <li><p><strong>block_size</strong> (<em>int</em><em>, </em><em>required</em>) – Blocks of [block_size. block_size] are moved</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.diag"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">diag</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">k=_Null</em>, <em class="sig-param">axis1=_Null</em>, <em class="sig-param">axis2=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.diag" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Extracts a diagonal or constructs a diagonal array.</p> |
| <p><code class="docutils literal notranslate"><span class="pre">diag</span></code>’s behavior depends on the input array dimensions:</p> |
| <ul> |
| <li><p>1-D arrays: constructs a 2-D array with the input as its diagonal, all other elements are zero.</p></li> |
| <li><p>N-D arrays: extracts the diagonals of the sub-arrays with axes specified by <code class="docutils literal notranslate"><span class="pre">axis1</span></code> and <code class="docutils literal notranslate"><span class="pre">axis2</span></code>. |
| The output shape would be decided by removing the axes numbered <code class="docutils literal notranslate"><span class="pre">axis1</span></code> and <code class="docutils literal notranslate"><span class="pre">axis2</span></code> from the |
| input shape and appending to the result a new axis with the size of the diagonals in question.</p> |
| <p>For example, when the input shape is <cite>(2, 3, 4, 5)</cite>, <code class="docutils literal notranslate"><span class="pre">axis1</span></code> and <code class="docutils literal notranslate"><span class="pre">axis2</span></code> are 0 and 2 |
| respectively and <code class="docutils literal notranslate"><span class="pre">k</span></code> is 0, the resulting shape would be <cite>(3, 5, 2)</cite>.</p> |
| </li> |
| </ul> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="p">[</span><span class="mi">4</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">]]</span> |
| |
| <span class="n">diag</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">5</span><span class="p">]</span> |
| |
| <span class="n">diag</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">k</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">6</span><span class="p">]</span> |
| |
| <span class="n">diag</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">k</span><span class="o">=-</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mi">4</span><span class="p">]</span> |
| |
| <span class="n">x</span> <span class="o">=</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="n">diag</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">3</span><span class="p">]]</span> |
| |
| <span class="n">diag</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">k</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</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">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]</span> |
| |
| <span class="n">diag</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">k</span><span class="o">=-</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]</span> |
| |
| <span class="n">x</span> <span class="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]],</span> |
| |
| <span class="p">[[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">8</span><span class="p">]]]</span> |
| |
| <span class="n">diag</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">7</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">8</span><span class="p">]]</span> |
| |
| <span class="n">diag</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">k</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">3</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">4</span><span class="p">]]</span> |
| |
| <span class="n">diag</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis1</span><span class="o">=-</span><span class="mi">2</span><span class="p">,</span> <span class="n">axis2</span><span class="o">=-</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">8</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/diag_op.cc:L86</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input ndarray</p></li> |
| <li><p><strong>k</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – Diagonal in question. The default is 0. Use k>0 for diagonals above the main diagonal, and k<0 for diagonals below the main diagonal. If input has shape (S0 S1) k must be between -S0 and S1</p></li> |
| <li><p><strong>axis1</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – The first axis of the sub-arrays of interest. Ignored when the input is a 1-D array.</p></li> |
| <li><p><strong>axis2</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – The second axis of the sub-arrays of interest. Ignored when the input is a 1-D array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.dot"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">dot</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">transpose_a=_Null</em>, <em class="sig-param">transpose_b=_Null</em>, <em class="sig-param">forward_stype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.dot" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Dot product of two arrays.</p> |
| <p><code class="docutils literal notranslate"><span class="pre">dot</span></code>’s behavior depends on the input array dimensions:</p> |
| <ul> |
| <li><p>1-D arrays: inner product of vectors</p></li> |
| <li><p>2-D arrays: matrix multiplication</p></li> |
| <li><p>N-D arrays: a sum product over the last axis of the first input and the first |
| axis of the second input</p> |
| <p>For example, given 3-D <code class="docutils literal notranslate"><span class="pre">x</span></code> with shape <cite>(n,m,k)</cite> and <code class="docutils literal notranslate"><span class="pre">y</span></code> with shape <cite>(k,r,s)</cite>, the |
| result array will have shape <cite>(n,m,r,s)</cite>. It is computed by:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">dot</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">y</span><span class="p">)[</span><span class="n">i</span><span class="p">,</span><span class="n">j</span><span class="p">,</span><span class="n">a</span><span class="p">,</span><span class="n">b</span><span class="p">]</span> <span class="o">=</span> <span class="nb">sum</span><span class="p">(</span><span class="n">x</span><span class="p">[</span><span class="n">i</span><span class="p">,</span><span class="n">j</span><span class="p">,:]</span><span class="o">*</span><span class="n">y</span><span class="p">[:,</span><span class="n">a</span><span class="p">,</span><span class="n">b</span><span class="p">])</span> |
| </pre></div> |
| </div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="n">reshape</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="mi">4</span><span class="p">,</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">,</span><span class="mi">7</span><span class="p">],</span> <span class="n">shape</span><span class="o">=</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="n">y</span> <span class="o">=</span> <span class="n">reshape</span><span class="p">([</span><span class="mi">7</span><span class="p">,</span><span class="mi">6</span><span class="p">,</span><span class="mi">5</span><span class="p">,</span><span class="mi">4</span><span class="p">,</span><span class="mi">3</span><span class="p">,</span><span class="mi">2</span><span class="p">,</span><span class="mi">1</span><span class="p">,</span><span class="mi">0</span><span class="p">],</span> <span class="n">shape</span><span class="o">=</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="n">dot</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">y</span><span class="p">)[</span><span class="mi">0</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">1</span><span class="p">]</span> <span class="o">=</span> <span class="mi">0</span> |
| <span class="nb">sum</span><span class="p">(</span><span class="n">x</span><span class="p">[</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</span><span class="p">,:]</span><span class="o">*</span><span class="n">y</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="o">=</span> <span class="mi">0</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">dot</span></code> output depends on storage types of inputs, transpose option and |
| forward_stype option for output storage type. Implemented sparse operations include:</p> |
| <ul class="simple"> |
| <li><p>dot(default, default, transpose_a=True/False, transpose_b=True/False) = default</p></li> |
| <li><p>dot(csr, default, transpose_a=True) = default</p></li> |
| <li><p>dot(csr, default, transpose_a=True) = row_sparse</p></li> |
| <li><p>dot(csr, default) = default</p></li> |
| <li><p>dot(csr, row_sparse) = default</p></li> |
| <li><p>dot(default, csr) = csr (CPU only)</p></li> |
| <li><p>dot(default, csr, forward_stype=’default’) = default</p></li> |
| <li><p>dot(default, csr, transpose_b=True, forward_stype=’default’) = default</p></li> |
| </ul> |
| <p>If the combination of input storage types and forward_stype does not match any of the |
| above patterns, <code class="docutils literal notranslate"><span class="pre">dot</span></code> will fallback and generate output with default storage.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>If the storage type of the lhs is “csr”, the storage type of gradient w.r.t rhs will be |
| “row_sparse”. Only a subset of optimizers support sparse gradients, including SGD, AdaGrad |
| and Adam. Note that 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.incubator.apache.org/api/python/optimization/optimization.html">https://mxnet.incubator.apache.org/api/python/optimization/optimization.html</a></p> |
| </div> |
| <p>Defined in src/operator/tensor/dot.cc:L77</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The first input</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The second input</p></li> |
| <li><p><strong>transpose_a</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true then transpose the first input before dot.</p></li> |
| <li><p><strong>transpose_b</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true then transpose the second input before dot.</p></li> |
| <li><p><strong>forward_stype</strong> (<em>{None</em><em>, </em><em>'csr'</em><em>, </em><em>'default'</em><em>, </em><em>'row_sparse'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – The desired storage type of the forward output given by user, if thecombination of input storage types and this hint does not matchany implemented ones, the dot operator will perform fallback operationand still produce an output of the desired storage type.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.elemwise_add"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">elemwise_add</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.elemwise_add" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Adds arguments element-wise.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">elemwise_add</span></code> output depends on storage types of inputs</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>elemwise_add(row_sparse, row_sparse) = row_sparse</p></li> |
| <li><p>elemwise_add(csr, csr) = csr</p></li> |
| <li><p>elemwise_add(default, csr) = default</p></li> |
| <li><p>elemwise_add(csr, default) = default</p></li> |
| <li><p>elemwise_add(default, rsp) = default</p></li> |
| <li><p>elemwise_add(rsp, default) = default</p></li> |
| <li><p>otherwise, <code class="docutils literal notranslate"><span class="pre">elemwise_add</span></code> generates output with default storage</p></li> |
| </ul> |
| </div></blockquote> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – first input</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – second input</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.elemwise_div"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">elemwise_div</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.elemwise_div" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Divides arguments element-wise.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">elemwise_div</span></code> output is always dense</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – first input</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – second input</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.elemwise_mul"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">elemwise_mul</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.elemwise_mul" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Multiplies arguments element-wise.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">elemwise_mul</span></code> output depends on storage types of inputs</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>elemwise_mul(default, default) = default</p></li> |
| <li><p>elemwise_mul(row_sparse, row_sparse) = row_sparse</p></li> |
| <li><p>elemwise_mul(default, row_sparse) = row_sparse</p></li> |
| <li><p>elemwise_mul(row_sparse, default) = row_sparse</p></li> |
| <li><p>elemwise_mul(csr, csr) = csr</p></li> |
| <li><p>otherwise, <code class="docutils literal notranslate"><span class="pre">elemwise_mul</span></code> generates output with default storage</p></li> |
| </ul> |
| </div></blockquote> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – first input</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – second input</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.elemwise_sub"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">elemwise_sub</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.elemwise_sub" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Subtracts arguments element-wise.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">elemwise_sub</span></code> output depends on storage types of inputs</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>elemwise_sub(row_sparse, row_sparse) = row_sparse</p></li> |
| <li><p>elemwise_sub(csr, csr) = csr</p></li> |
| <li><p>elemwise_sub(default, csr) = default</p></li> |
| <li><p>elemwise_sub(csr, default) = default</p></li> |
| <li><p>elemwise_sub(default, rsp) = default</p></li> |
| <li><p>elemwise_sub(rsp, default) = default</p></li> |
| <li><p>otherwise, <code class="docutils literal notranslate"><span class="pre">elemwise_sub</span></code> generates output with default storage</p></li> |
| </ul> |
| </div></blockquote> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – first input</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – second input</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.erf"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">erf</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.erf" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise gauss error function of the input.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">erf</span><span class="p">([</span><span class="mi">0</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">10.</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="mf">0.</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.8427</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L886</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.erfinv"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">erfinv</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.erfinv" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise inverse gauss error function of the input.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">erfinv</span><span class="p">([</span><span class="mi">0</span><span class="p">,</span> <span class="mf">0.5</span><span class="o">.</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="mf">0.</span><span class="p">,</span> <span class="mf">0.4769</span><span class="p">,</span> <span class="o">-</span><span class="n">inf</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L908</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.exp"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">exp</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.exp" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise exponential value of the input.</p> |
| <div class="math notranslate nohighlight"> |
| \[exp(x) = e^x \approx 2.718^x\]</div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">exp</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="o">=</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">2.71828175</span><span class="p">,</span> <span class="mf">7.38905621</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">exp</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L64</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.expand_dims"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">expand_dims</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.expand_dims" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Inserts a new axis of size 1 into the array shape |
| For example, given <code class="docutils literal notranslate"><span class="pre">x</span></code> with shape <code class="docutils literal notranslate"><span class="pre">(2,3,4)</span></code>, then <code class="docutils literal notranslate"><span class="pre">expand_dims(x,</span> <span class="pre">axis=1)</span></code> |
| will return a new array with shape <code class="docutils literal notranslate"><span class="pre">(2,1,3,4)</span></code>.</p> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L394</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Source input</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>required</em>) – Position where new axis is to be inserted. Suppose that the input <cite>NDArray</cite>’s dimension is <cite>ndim</cite>, the range of the inserted axis is <cite>[-ndim, ndim]</cite></p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.expm1"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">expm1</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.expm1" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns <code class="docutils literal notranslate"><span class="pre">exp(x)</span> <span class="pre">-</span> <span class="pre">1</span></code> computed element-wise on the input.</p> |
| <p>This function provides greater precision than <code class="docutils literal notranslate"><span class="pre">exp(x)</span> <span class="pre">-</span> <span class="pre">1</span></code> for small values of <code class="docutils literal notranslate"><span class="pre">x</span></code>.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">expm1</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>expm1(default) = default</p></li> |
| <li><p>expm1(row_sparse) = row_sparse</p></li> |
| <li><p>expm1(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L244</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.fill_element_0index"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">fill_element_0index</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">mhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.fill_element_0index" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Fill one element of each line(row for python, column for R/Julia) in lhs according to index indicated by rhs and values indicated by mhs. This function assume rhs uses 0-based index.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../../ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Left operand to the function.</p></li> |
| <li><p><strong>mhs</strong> (<a class="reference internal" href="../../ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Middle operand to the function.</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../../ndarray/ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Right operand to the function.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.fix"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">fix</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.fix" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise rounded value to the nearest integer towards zero of the input.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">fix</span><span class="p">([</span><span class="o">-</span><span class="mf">2.1</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.9</span><span class="p">,</span> <span class="mf">1.9</span><span class="p">,</span> <span class="mf">2.1</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">2.</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">fix</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>fix(default) = default</p></li> |
| <li><p>fix(row_sparse) = row_sparse</p></li> |
| <li><p>fix(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L874</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.flatten"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">flatten</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.flatten" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Flattens the input array into a 2-D array by collapsing the higher dimensions. |
| .. note:: <cite>Flatten</cite> is deprecated. Use <cite>flatten</cite> instead. |
| For an input array with shape <code class="docutils literal notranslate"><span class="pre">(d1,</span> <span class="pre">d2,</span> <span class="pre">...,</span> <span class="pre">dk)</span></code>, <cite>flatten</cite> operation reshapes |
| the input array into an output array of shape <code class="docutils literal notranslate"><span class="pre">(d1,</span> <span class="pre">d2*...*dk)</span></code>. |
| Note that the behavior of this function is different from numpy.ndarray.flatten, |
| which behaves similar to mxnet.ndarray.reshape((-1,)). |
| Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</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="p">[</span><span class="mi">4</span><span class="p">,</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">7</span><span class="p">,</span><span class="mi">8</span><span class="p">,</span><span class="mi">9</span><span class="p">]</span> |
| <span class="p">],</span> |
| <span class="p">[</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="p">[</span><span class="mi">4</span><span class="p">,</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">7</span><span class="p">,</span><span class="mi">8</span><span class="p">,</span><span class="mi">9</span><span class="p">]</span> |
| <span class="p">]],</span> |
| <span class="n">flatten</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L249</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.flip"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">flip</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.flip" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Reverses the order of elements along given axis while preserving array shape. |
| Note: reverse and flip are equivalent. We use reverse in the following examples. |
| Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">]]</span> |
| <span class="n">reverse</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]</span> |
| <span class="n">reverse</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">9.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L831</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data array</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – The axis which to reverse elements.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.floor"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">floor</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.floor" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise floor of the input.</p> |
| <p>The floor of the scalar x is the largest integer i, such that i <= x.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">floor</span><span class="p">([</span><span class="o">-</span><span class="mf">2.1</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.9</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">,</span> <span class="mf">1.9</span><span class="p">,</span> <span class="mf">2.1</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">3.</span><span class="p">,</span> <span class="o">-</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">floor</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>floor(default) = default</p></li> |
| <li><p>floor(row_sparse) = row_sparse</p></li> |
| <li><p>floor(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L836</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.ftml_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">ftml_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">d=None</em>, <em class="sig-param">v=None</em>, <em class="sig-param">z=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">beta1=_Null</em>, <em class="sig-param">beta2=_Null</em>, <em class="sig-param">epsilon=_Null</em>, <em class="sig-param">t=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_grad=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.ftml_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>The FTML optimizer described in |
| <em>FTML - Follow the Moving Leader in Deep Learning</em>, |
| available at <a class="reference external" href="http://proceedings.mlr.press/v70/zheng17a/zheng17a.pdf">http://proceedings.mlr.press/v70/zheng17a/zheng17a.pdf</a>.</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}g_t = \nabla J(W_{t-1})\\ |
| v_t = \beta_2 v_{t-1} + (1 - \beta_2) g_t^2\\ |
| d_t = \frac{ 1 - \beta_1^t }{ \eta_t } (\sqrt{ \frac{ v_t }{ 1 - \beta_2^t } } + \epsilon) |
| \sigma_t = d_t - \beta_1 d_{t-1} |
| z_t = \beta_1 z_{ t-1 } + (1 - \beta_1^t) g_t - \sigma_t W_{t-1} |
| W_t = - \frac{ z_t }{ d_t }\end{split}\]</div> |
| <p>Defined in src/operator/optimizer_op.cc:L639</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>d</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Internal state <code class="docutils literal notranslate"><span class="pre">d_t</span></code></p></li> |
| <li><p><strong>v</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Internal state <code class="docutils literal notranslate"><span class="pre">v_t</span></code></p></li> |
| <li><p><strong>z</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Internal state <code class="docutils literal notranslate"><span class="pre">z_t</span></code></p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate.</p></li> |
| <li><p><strong>beta1</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.600000024</em>) – Generally close to 0.5.</p></li> |
| <li><p><strong>beta2</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.999000013</em>) – Generally close to 1.</p></li> |
| <li><p><strong>epsilon</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=9.9999999392252903e-09</em>) – Epsilon to prevent div 0.</p></li> |
| <li><p><strong>t</strong> (<em>int</em><em>, </em><em>required</em>) – Number of update.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.ftrl_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">ftrl_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">z=None</em>, <em class="sig-param">n=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">lamda1=_Null</em>, <em class="sig-param">beta=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.ftrl_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for Ftrl optimizer. |
| Referenced from <em>Ad Click Prediction: a View from the Trenches</em>, available at |
| <a class="reference external" href="http://dl.acm.org/citation.cfm?id=2488200">http://dl.acm.org/citation.cfm?id=2488200</a>.</p> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">rescaled_grad</span> <span class="o">=</span> <span class="n">clip</span><span class="p">(</span><span class="n">grad</span> <span class="o">*</span> <span class="n">rescale_grad</span><span class="p">,</span> <span class="n">clip_gradient</span><span class="p">)</span> |
| <span class="n">z</span> <span class="o">+=</span> <span class="n">rescaled_grad</span> <span class="o">-</span> <span class="p">(</span><span class="n">sqrt</span><span class="p">(</span><span class="n">n</span> <span class="o">+</span> <span class="n">rescaled_grad</span><span class="o">**</span><span class="mi">2</span><span class="p">)</span> <span class="o">-</span> <span class="n">sqrt</span><span class="p">(</span><span class="n">n</span><span class="p">))</span> <span class="o">*</span> <span class="n">weight</span> <span class="o">/</span> <span class="n">learning_rate</span> |
| <span class="n">n</span> <span class="o">+=</span> <span class="n">rescaled_grad</span><span class="o">**</span><span class="mi">2</span> |
| <span class="n">w</span> <span class="o">=</span> <span class="p">(</span><span class="n">sign</span><span class="p">(</span><span class="n">z</span><span class="p">)</span> <span class="o">*</span> <span class="n">lamda1</span> <span class="o">-</span> <span class="n">z</span><span class="p">)</span> <span class="o">/</span> <span class="p">((</span><span class="n">beta</span> <span class="o">+</span> <span class="n">sqrt</span><span class="p">(</span><span class="n">n</span><span class="p">))</span> <span class="o">/</span> <span class="n">learning_rate</span> <span class="o">+</span> <span class="n">wd</span><span class="p">)</span> <span class="o">*</span> <span class="p">(</span><span class="nb">abs</span><span class="p">(</span><span class="n">z</span><span class="p">)</span> <span class="o">></span> <span class="n">lamda1</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>If w, z and n are all of <code class="docutils literal notranslate"><span class="pre">row_sparse</span></code> storage type, |
| only the row slices whose indices appear in grad.indices are updated (for w, z and n):</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="k">for</span> <span class="n">row</span> <span class="ow">in</span> <span class="n">grad</span><span class="o">.</span><span class="n">indices</span><span class="p">:</span> |
| <span class="n">rescaled_grad</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">=</span> <span class="n">clip</span><span class="p">(</span><span class="n">grad</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">*</span> <span class="n">rescale_grad</span><span class="p">,</span> <span class="n">clip_gradient</span><span class="p">)</span> |
| <span class="n">z</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">+=</span> <span class="n">rescaled_grad</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">-</span> <span class="p">(</span><span class="n">sqrt</span><span class="p">(</span><span class="n">n</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">+</span> <span class="n">rescaled_grad</span><span class="p">[</span><span class="n">row</span><span class="p">]</span><span class="o">**</span><span class="mi">2</span><span class="p">)</span> <span class="o">-</span> <span class="n">sqrt</span><span class="p">(</span><span class="n">n</span><span class="p">[</span><span class="n">row</span><span class="p">]))</span> <span class="o">*</span> <span class="n">weight</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">/</span> <span class="n">learning_rate</span> |
| <span class="n">n</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">+=</span> <span class="n">rescaled_grad</span><span class="p">[</span><span class="n">row</span><span class="p">]</span><span class="o">**</span><span class="mi">2</span> |
| <span class="n">w</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">=</span> <span class="p">(</span><span class="n">sign</span><span class="p">(</span><span class="n">z</span><span class="p">[</span><span class="n">row</span><span class="p">])</span> <span class="o">*</span> <span class="n">lamda1</span> <span class="o">-</span> <span class="n">z</span><span class="p">[</span><span class="n">row</span><span class="p">])</span> <span class="o">/</span> <span class="p">((</span><span class="n">beta</span> <span class="o">+</span> <span class="n">sqrt</span><span class="p">(</span><span class="n">n</span><span class="p">[</span><span class="n">row</span><span class="p">]))</span> <span class="o">/</span> <span class="n">learning_rate</span> <span class="o">+</span> <span class="n">wd</span><span class="p">)</span> <span class="o">*</span> <span class="p">(</span><span class="nb">abs</span><span class="p">(</span><span class="n">z</span><span class="p">[</span><span class="n">row</span><span class="p">])</span> <span class="o">></span> <span class="n">lamda1</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/optimizer_op.cc:L875</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>z</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – z</p></li> |
| <li><p><strong>n</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Square of grad</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>lamda1</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.00999999978</em>) – The L1 regularization coefficient.</p></li> |
| <li><p><strong>beta</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Per-Coordinate Learning Rate beta.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.gamma"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">gamma</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.gamma" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the gamma function (extension of the factorial function to the reals), computed element-wise on the input array.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">gamma</span></code> output is always dense</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.gammaln"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">gammaln</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.gammaln" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise log of the absolute value of the gamma function of the input.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">gammaln</span></code> output is always dense</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.gather_nd"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">gather_nd</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">indices=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.gather_nd" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Gather elements or slices from <cite>data</cite> and store to a tensor whose |
| shape is defined by <cite>indices</cite>.</p> |
| <p>Given <cite>data</cite> with shape <cite>(X_0, X_1, …, X_{N-1})</cite> and indices with shape |
| <cite>(M, Y_0, …, Y_{K-1})</cite>, the output will have shape <cite>(Y_0, …, Y_{K-1}, X_M, …, X_{N-1})</cite>, |
| where <cite>M <= N</cite>. If <cite>M == N</cite>, output shape will simply be <cite>(Y_0, …, Y_{K-1})</cite>.</p> |
| <p>The elements in output is defined as follows:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">output</span><span class="p">[</span><span class="n">y_0</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">y_</span><span class="p">{</span><span class="n">K</span><span class="o">-</span><span class="mi">1</span><span class="p">},</span> <span class="n">x_M</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">x_</span><span class="p">{</span><span class="n">N</span><span class="o">-</span><span class="mi">1</span><span class="p">}]</span> <span class="o">=</span> <span class="n">data</span><span class="p">[</span><span class="n">indices</span><span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="n">y_0</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">y_</span><span class="p">{</span><span class="n">K</span><span class="o">-</span><span class="mi">1</span><span class="p">}],</span> |
| <span class="o">...</span><span class="p">,</span> |
| <span class="n">indices</span><span class="p">[</span><span class="n">M</span><span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="n">y_0</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">y_</span><span class="p">{</span><span class="n">K</span><span class="o">-</span><span class="mi">1</span><span class="p">}],</span> |
| <span class="n">x_M</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">x_</span><span class="p">{</span><span class="n">N</span><span class="o">-</span><span class="mi">1</span><span class="p">}]</span> |
| </pre></div> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">data</span> <span class="o">=</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="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">]]</span> |
| <span class="n">indices</span> <span class="o">=</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">0</span><span class="p">],</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">0</span><span class="p">]]</span> |
| <span class="n">gather_nd</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">indices</span><span class="p">)</span> <span class="o">=</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">0</span><span class="p">]</span> |
| |
| <span class="n">data</span> <span class="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]],</span> <span class="p">[[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">8</span><span class="p">]]]</span> |
| <span class="n">indices</span> <span class="o">=</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="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]</span> |
| <span class="n">gather_nd</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">indices</span><span class="p">)</span> <span class="o">=</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="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</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>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – data</p></li> |
| <li><p><strong>indices</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – indices</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.hard_sigmoid"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">hard_sigmoid</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">alpha=_Null</em>, <em class="sig-param">beta=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.hard_sigmoid" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes hard sigmoid of x element-wise.</p> |
| <div class="math notranslate nohighlight"> |
| \[y = max(0, min(1, alpha * x + beta))\]</div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L161</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>alpha</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.200000003</em>) – Slope of hard sigmoid</p></li> |
| <li><p><strong>beta</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.5</em>) – Bias of hard sigmoid.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.identity"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">identity</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.identity" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a copy of the input.</p> |
| <p>From:src/operator/tensor/elemwise_unary_op_basic.cc:244</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.im2col"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">im2col</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">kernel=_Null</em>, <em class="sig-param">stride=_Null</em>, <em class="sig-param">dilate=_Null</em>, <em class="sig-param">pad=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.im2col" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Extract sliding blocks from input array.</p> |
| <p>This operator is used in vanilla convolution implementation to transform the sliding |
| blocks on image to column matrix, then the convolution operation can be computed |
| by matrix multiplication between column and convolution weight. Due to the close |
| relation between im2col and convolution, the concept of <strong>kernel</strong>, <strong>stride</strong>, |
| <strong>dilate</strong> and <strong>pad</strong> in this operator are inherited from convolution operation.</p> |
| <p>Given the input data of shape <span class="math notranslate nohighlight">\((N, C, *)\)</span>, where <span class="math notranslate nohighlight">\(N\)</span> is the batch size, |
| <span class="math notranslate nohighlight">\(C\)</span> is the channel size, and <span class="math notranslate nohighlight">\(*\)</span> is the arbitrary spatial dimension, |
| the output column array is always with shape <span class="math notranslate nohighlight">\((N, C \times \prod(\text{kernel}), W)\)</span>, |
| where <span class="math notranslate nohighlight">\(C \times \prod(\text{kernel})\)</span> is the block size, and <span class="math notranslate nohighlight">\(W\)</span> is the |
| block number which is the spatial size of the convolution output with same input parameters. |
| Only 1-D, 2-D and 3-D of spatial dimension is supported in this operator.</p> |
| <p>Defined in src/operator/nn/im2col.cc:L99</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array to extract sliding blocks.</p></li> |
| <li><p><strong>kernel</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – Sliding kernel size: (w,), (h, w) or (d, h, w).</p></li> |
| <li><p><strong>stride</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The stride between adjacent sliding blocks in spatial dimension: (w,), (h, w) or (d, h, w). Defaults to 1 for each dimension.</p></li> |
| <li><p><strong>dilate</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The spacing between adjacent kernel points: (w,), (h, w) or (d, h, w). Defaults to 1 for each dimension.</p></li> |
| <li><p><strong>pad</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The zero-value padding size on both sides of spatial dimension: (w,), (h, w) or (d, h, w). Defaults to no padding.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.khatri_rao"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">khatri_rao</code><span class="sig-paren">(</span><em class="sig-param">*args</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.khatri_rao" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the Khatri-Rao product of the input matrices.</p> |
| <p>Given a collection of <span class="math notranslate nohighlight">\(n\)</span> input matrices,</p> |
| <div class="math notranslate nohighlight"> |
| \[A_1 \in \mathbb{R}^{M_1 \times M}, \ldots, A_n \in \mathbb{R}^{M_n \times N},\]</div> |
| <p>the (column-wise) Khatri-Rao product is defined as the matrix,</p> |
| <div class="math notranslate nohighlight"> |
| \[X = A_1 \otimes \cdots \otimes A_n \in \mathbb{R}^{(M_1 \cdots M_n) \times N},\]</div> |
| <p>where the <span class="math notranslate nohighlight">\(k\)</span> th column is equal to the column-wise outer product |
| <span class="math notranslate nohighlight">\({A_1}_k \otimes \cdots \otimes {A_n}_k\)</span> where <span class="math notranslate nohighlight">\({A_i}_k\)</span> is the kth |
| column of the ith matrix.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">A</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">array</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="gp">>>> </span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="o">-</span><span class="mi">3</span><span class="p">]])</span> |
| <span class="gp">>>> </span><span class="n">B</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">array</span><span class="p">([[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</span> |
| <span class="gp">>>> </span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">5</span><span class="p">],</span> |
| <span class="gp">>>> </span> <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">6</span><span class="p">]])</span> |
| <span class="gp">>>> </span><span class="n">C</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">khatri_rao</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">B</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="nb">print</span><span class="p">(</span><span class="n">C</span><span class="o">.</span><span class="n">asnumpy</span><span class="p">())</span> |
| <span class="go">[[ 1. -4.]</span> |
| <span class="go"> [ 2. -5.]</span> |
| <span class="go"> [ 3. -6.]</span> |
| <span class="go"> [ 2. -12.]</span> |
| <span class="go"> [ 4. -15.]</span> |
| <span class="go"> [ 6. -18.]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/contrib/krprod.cc:L108 |
| This function support variable length of positional input.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>args</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Positional input matrices</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.lamb_update_phase1"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">lamb_update_phase1</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">mean=None</em>, <em class="sig-param">var=None</em>, <em class="sig-param">beta1=_Null</em>, <em class="sig-param">beta2=_Null</em>, <em class="sig-param">epsilon=_Null</em>, <em class="sig-param">t=_Null</em>, <em class="sig-param">bias_correction=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.lamb_update_phase1" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Phase I of lamb update it performs the following operations and returns g:.</p> |
| <p>Link to paper: <a class="reference external" href="https://arxiv.org/pdf/1904.00962.pdf">https://arxiv.org/pdf/1904.00962.pdf</a></p> |
| <div class="math notranslate nohighlight"> |
| \[ \begin{align}\begin{aligned}\begin{gather*} |
| grad = grad * rescale_grad |
| if (grad < -clip_gradient) |
| then |
| grad = -clip_gradient |
| if (grad > clip_gradient) |
| then |
| grad = clip_gradient\\mean = beta1 * mean + (1 - beta1) * grad; |
| variance = beta2 * variance + (1. - beta2) * grad ^ 2;\\if (bias_correction) |
| then |
| mean_hat = mean / (1. - beta1^t); |
| var_hat = var / (1 - beta2^t); |
| g = mean_hat / (var_hat^(1/2) + epsilon) + wd * weight; |
| else |
| g = mean / (var_data^(1/2) + epsilon) + wd * weight; |
| \end{gather*}\end{aligned}\end{align} \]</div> |
| <p>Defined in src/operator/optimizer_op.cc:L952</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>mean</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Moving mean</p></li> |
| <li><p><strong>var</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Moving variance</p></li> |
| <li><p><strong>beta1</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.899999976</em>) – The decay rate for the 1st moment estimates.</p></li> |
| <li><p><strong>beta2</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.999000013</em>) – The decay rate for the 2nd moment estimates.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=9.99999997e-07</em>) – A small constant for numerical stability.</p></li> |
| <li><p><strong>t</strong> (<em>int</em><em>, </em><em>required</em>) – Index update count.</p></li> |
| <li><p><strong>bias_correction</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Whether to use bias correction.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>required</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.lamb_update_phase2"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">lamb_update_phase2</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">g=None</em>, <em class="sig-param">r1=None</em>, <em class="sig-param">r2=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">lower_bound=_Null</em>, <em class="sig-param">upper_bound=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.lamb_update_phase2" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Phase II of lamb update it performs the following operations and updates grad.</p> |
| <p>Link to paper: <a class="reference external" href="https://arxiv.org/pdf/1904.00962.pdf">https://arxiv.org/pdf/1904.00962.pdf</a></p> |
| <div class="math notranslate nohighlight"> |
| \[ \begin{align}\begin{aligned}\begin{gather*} |
| if (lower_bound >= 0) |
| then |
| r1 = max(r1, lower_bound) |
| if (upper_bound >= 0) |
| then |
| r1 = max(r1, upper_bound)\\if (r1 == 0 or r2 == 0) |
| then |
| lr = lr |
| else |
| lr = lr * (r1/r2) |
| weight = weight - lr * g |
| \end{gather*}\end{aligned}\end{align} \]</div> |
| <p>Defined in src/operator/optimizer_op.cc:L991</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>g</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Output of lamb_update_phase 1</p></li> |
| <li><p><strong>r1</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – r1</p></li> |
| <li><p><strong>r2</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – r2</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>lower_bound</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Lower limit of norm of weight. If lower_bound <= 0, Lower limit is not set</p></li> |
| <li><p><strong>upper_bound</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Upper limit of norm of weight. If upper_bound <= 0, Upper limit is not set</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_det"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_det</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_det" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Compute the determinant of a matrix. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, <em>A</em> is a square matrix. We compute:</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>det(A)</em></p> |
| </div></blockquote> |
| <p>If <em>n>2</em>, <em>det</em> is performed separately on the trailing two dimensions |
| for all inputs (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>There is no gradient backwarded when A is non-invertible (which is |
| equivalent to det(A) = 0) because zero is rarely hit upon in float |
| point computation and the Jacobi’s formula on determinant gradient |
| is not computationally efficient when A is non-invertible.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">determinant</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> <span class="p">[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]]</span> |
| <span class="n">det</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">5.</span><span class="p">]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">determinant</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> <span class="p">[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]]</span> |
| <span class="n">det</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">5.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L974</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of square matrix</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_extractdiag"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_extractdiag</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">offset=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_extractdiag" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Extracts the diagonal entries of a square matrix. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, then <em>A</em> represents a single square matrix which diagonal elements get extracted as a 1-dimensional tensor.</p> |
| <p>If <em>n>2</em>, then <em>A</em> represents a batch of square matrices on the trailing two dimensions. The extracted diagonals are returned as an <em>n-1</em>-dimensional tensor.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">diagonal</span> <span class="n">extraction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">3.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]]</span> |
| |
| <span class="n">extractdiag</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]</span> |
| |
| <span class="n">extractdiag</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">2.0</span><span class="p">]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">diagonal</span> <span class="n">extraction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">3.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">5.0</span><span class="p">,</span> <span class="mf">6.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">7.0</span><span class="p">,</span> <span class="mf">8.0</span><span class="p">]]]</span> |
| |
| <span class="n">extractdiag</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">5.0</span><span class="p">,</span> <span class="mf">8.0</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L494</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of square matrices</p></li> |
| <li><p><strong>offset</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – Offset of the diagonal versus the main diagonal. 0 corresponds to the main diagonal, a negative/positive value to diagonals below/above the main diagonal.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_extracttrian"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_extracttrian</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">offset=_Null</em>, <em class="sig-param">lower=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_extracttrian" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Extracts a triangular sub-matrix from a square matrix. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, then <em>A</em> represents a single square matrix from which a triangular sub-matrix is extracted as a 1-dimensional tensor.</p> |
| <p>If <em>n>2</em>, then <em>A</em> represents a batch of square matrices on the trailing two dimensions. The extracted triangular sub-matrices are returned as an <em>n-1</em>-dimensional tensor.</p> |
| <p>The <em>offset</em> and <em>lower</em> parameters determine the triangle to be extracted:</p> |
| <ul class="simple"> |
| <li><p>When <em>offset = 0</em> either the lower or upper triangle with respect to the main diagonal is extracted depending on the value of parameter <em>lower</em>.</p></li> |
| <li><p>When <em>offset = k > 0</em> the upper triangle with respect to the k-th diagonal above the main diagonal is extracted.</p></li> |
| <li><p>When <em>offset = k < 0</em> the lower triangle with respect to the k-th diagonal below the main diagonal is extracted.</p></li> |
| </ul> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">triagonal</span> <span class="n">extraction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">3.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]]</span> |
| |
| <span class="n">extracttrian</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]</span> |
| <span class="n">extracttrian</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">lower</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]</span> |
| <span class="n">extracttrian</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">2.0</span><span class="p">]</span> |
| <span class="n">extracttrian</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">3.0</span><span class="p">]</span> |
| |
| <span class="n">Batch</span> <span class="n">triagonal</span> <span class="n">extraction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">3.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">5.0</span><span class="p">,</span> <span class="mf">6.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">7.0</span><span class="p">,</span> <span class="mf">8.0</span><span class="p">]]]</span> |
| |
| <span class="n">extracttrian</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">5.0</span><span class="p">,</span> <span class="mf">7.0</span><span class="p">,</span> <span class="mf">8.0</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L604</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of square matrices</p></li> |
| <li><p><strong>offset</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – Offset of the diagonal versus the main diagonal. 0 corresponds to the main diagonal, a negative/positive value to diagonals below/above the main diagonal.</p></li> |
| <li><p><strong>lower</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Refer to the lower triangular matrix if lower=true, refer to the upper otherwise. Only relevant when offset=0</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_gelqf"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_gelqf</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_gelqf" title="Permalink to this definition">¶</a></dt> |
| <dd><p>LQ factorization for general matrix. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, we compute the LQ factorization (LAPACK <em>gelqf</em>, followed by <em>orglq</em>). <em>A</em> |
| must have shape <em>(x, y)</em> with <em>x <= y</em>, and must have full rank <em>=x</em>. The LQ |
| factorization consists of <em>L</em> with shape <em>(x, x)</em> and <em>Q</em> with shape <em>(x, y)</em>, so |
| that:</p> |
| <blockquote> |
| <div><p><em>A</em> = <em>L</em> * <em>Q</em></p> |
| </div></blockquote> |
| <p>Here, <em>L</em> is lower triangular (upper triangle equal to zero) with nonzero diagonal, |
| and <em>Q</em> is row-orthonormal, meaning that</p> |
| <blockquote> |
| <div><p><em>Q</em> * <em>Q</em><sup>T</sup></p> |
| </div></blockquote> |
| <p>is equal to the identity matrix of shape <em>(x, x)</em>.</p> |
| <p>If <em>n>2</em>, <em>gelqf</em> is performed separately on the trailing two dimensions for all |
| inputs (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">LQ</span> <span class="n">factorization</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]</span> |
| <span class="n">Q</span><span class="p">,</span> <span class="n">L</span> <span class="o">=</span> <span class="n">gelqf</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> |
| <span class="n">Q</span> <span class="o">=</span> <span class="p">[[</span><span class="o">-</span><span class="mf">0.26726124</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.53452248</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.80178373</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.87287156</span><span class="p">,</span> <span class="mf">0.21821789</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.43643578</span><span class="p">]]</span> |
| <span class="n">L</span> <span class="o">=</span> <span class="p">[[</span><span class="o">-</span><span class="mf">3.74165739</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span><span class="o">-</span><span class="mf">8.55235974</span><span class="p">,</span> <span class="mf">1.96396101</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">LQ</span> <span class="n">factorization</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> <span class="p">[</span><span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]]]</span> |
| <span class="n">Q</span><span class="p">,</span> <span class="n">L</span> <span class="o">=</span> <span class="n">gelqf</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> |
| <span class="n">Q</span> <span class="o">=</span> <span class="p">[[[</span><span class="o">-</span><span class="mf">0.26726124</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.53452248</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.80178373</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.87287156</span><span class="p">,</span> <span class="mf">0.21821789</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.43643578</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="o">-</span><span class="mf">0.50257071</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.57436653</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.64616234</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.7620735</span><span class="p">,</span> <span class="mf">0.05862104</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.64483142</span><span class="p">]]]</span> |
| <span class="n">L</span> <span class="o">=</span> <span class="p">[[[</span><span class="o">-</span><span class="mf">3.74165739</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span><span class="o">-</span><span class="mf">8.55235974</span><span class="p">,</span> <span class="mf">1.96396101</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="o">-</span><span class="mf">13.92838828</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span><span class="o">-</span><span class="mf">19.09768702</span><span class="p">,</span> <span class="mf">0.52758934</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L797</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of input matrices to be factorized</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_gemm"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_gemm</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">B=None</em>, <em class="sig-param">C=None</em>, <em class="sig-param">transpose_a=_Null</em>, <em class="sig-param">transpose_b=_Null</em>, <em class="sig-param">alpha=_Null</em>, <em class="sig-param">beta=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_gemm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Performs general matrix multiplication and accumulation. |
| Input are tensors <em>A</em>, <em>B</em>, <em>C</em>, each of dimension <em>n >= 2</em> and having the same shape |
| on the leading <em>n-2</em> dimensions.</p> |
| <p>If <em>n=2</em>, the BLAS3 function <em>gemm</em> is performed:</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>alpha</em> * <em>op</em>(<em>A</em>) * <em>op</em>(<em>B</em>) + <em>beta</em> * <em>C</em></p> |
| </div></blockquote> |
| <p>Here, <em>alpha</em> and <em>beta</em> are scalar parameters, and <em>op()</em> is either the identity or |
| matrix transposition (depending on <em>transpose_a</em>, <em>transpose_b</em>).</p> |
| <p>If <em>n>2</em>, <em>gemm</em> is performed separately for a batch of matrices. The column indices of the matrices |
| are given by the last dimensions of the tensors, the row indices by the axis specified with the <em>axis</em> |
| parameter. By default, the trailing two dimensions will be used for matrix encoding.</p> |
| <p>For a non-default axis parameter, the operation performed is equivalent to a series of swapaxes/gemm/swapaxes |
| calls. For example let <em>A</em>, <em>B</em>, <em>C</em> be 5 dimensional tensors. Then gemm(<em>A</em>, <em>B</em>, <em>C</em>, axis=1) is equivalent |
| to the following without the overhead of the additional swapaxis operations:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">A1</span> <span class="o">=</span> <span class="n">swapaxes</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">dim1</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">dim2</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> |
| <span class="n">B1</span> <span class="o">=</span> <span class="n">swapaxes</span><span class="p">(</span><span class="n">B</span><span class="p">,</span> <span class="n">dim1</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">dim2</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> |
| <span class="n">C</span> <span class="o">=</span> <span class="n">swapaxes</span><span class="p">(</span><span class="n">C</span><span class="p">,</span> <span class="n">dim1</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">dim2</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> |
| <span class="n">C</span> <span class="o">=</span> <span class="n">gemm</span><span class="p">(</span><span class="n">A1</span><span class="p">,</span> <span class="n">B1</span><span class="p">,</span> <span class="n">C</span><span class="p">)</span> |
| <span class="n">C</span> <span class="o">=</span> <span class="n">swapaxis</span><span class="p">(</span><span class="n">C</span><span class="p">,</span> <span class="n">dim1</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">dim2</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>When the input data is of type float32 and the environment variables MXNET_CUDA_ALLOW_TENSOR_CORE |
| and MXNET_CUDA_TENSOR_OP_MATH_ALLOW_CONVERSION are set to 1, this operator will try to use |
| pseudo-float16 precision (float32 math with float16 I/O) precision in order to use |
| Tensor Cores on suitable NVIDIA GPUs. This can sometimes give significant speedups.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">multiply</span><span class="o">-</span><span class="n">add</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]</span> |
| <span class="n">B</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]</span> |
| <span class="n">C</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]</span> |
| <span class="n">gemm</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">B</span><span class="p">,</span> <span class="n">C</span><span class="p">,</span> <span class="n">transpose_b</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">2.0</span><span class="p">,</span> <span class="n">beta</span><span class="o">=</span><span class="mf">10.0</span><span class="p">)</span> |
| <span class="o">=</span> <span class="p">[[</span><span class="mf">14.0</span><span class="p">,</span> <span class="mf">14.0</span><span class="p">,</span> <span class="mf">14.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">14.0</span><span class="p">,</span> <span class="mf">14.0</span><span class="p">,</span> <span class="mf">14.0</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">multiply</span><span class="o">-</span><span class="n">add</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">]]]</span> |
| <span class="n">B</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">]]]</span> |
| <span class="n">C</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">10.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.01</span><span class="p">]]]</span> |
| <span class="n">gemm</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">B</span><span class="p">,</span> <span class="n">C</span><span class="p">,</span> <span class="n">transpose_b</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">2.0</span> <span class="p">,</span> <span class="n">beta</span><span class="o">=</span><span class="mf">10.0</span><span class="p">)</span> |
| <span class="o">=</span> <span class="p">[[[</span><span class="mf">104.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.14</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L88</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of input matrices</p></li> |
| <li><p><strong>B</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of input matrices</p></li> |
| <li><p><strong>C</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of input matrices</p></li> |
| <li><p><strong>transpose_a</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Multiply with transposed of first input (A).</p></li> |
| <li><p><strong>transpose_b</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Multiply with transposed of second input (B).</p></li> |
| <li><p><strong>alpha</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scalar factor multiplied with A*B.</p></li> |
| <li><p><strong>beta</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scalar factor multiplied with C.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='-2'</em>) – Axis corresponding to the matrix rows.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_gemm2"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_gemm2</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">B=None</em>, <em class="sig-param">transpose_a=_Null</em>, <em class="sig-param">transpose_b=_Null</em>, <em class="sig-param">alpha=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_gemm2" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Performs general matrix multiplication. |
| Input are tensors <em>A</em>, <em>B</em>, each of dimension <em>n >= 2</em> and having the same shape |
| on the leading <em>n-2</em> dimensions.</p> |
| <p>If <em>n=2</em>, the BLAS3 function <em>gemm</em> is performed:</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>alpha</em> * <em>op</em>(<em>A</em>) * <em>op</em>(<em>B</em>)</p> |
| </div></blockquote> |
| <p>Here <em>alpha</em> is a scalar parameter and <em>op()</em> is either the identity or the matrix |
| transposition (depending on <em>transpose_a</em>, <em>transpose_b</em>).</p> |
| <p>If <em>n>2</em>, <em>gemm</em> is performed separately for a batch of matrices. The column indices of the matrices |
| are given by the last dimensions of the tensors, the row indices by the axis specified with the <em>axis</em> |
| parameter. By default, the trailing two dimensions will be used for matrix encoding.</p> |
| <p>For a non-default axis parameter, the operation performed is equivalent to a series of swapaxes/gemm/swapaxes |
| calls. For example let <em>A</em>, <em>B</em> be 5 dimensional tensors. Then gemm(<em>A</em>, <em>B</em>, axis=1) is equivalent to |
| the following without the overhead of the additional swapaxis operations:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">A1</span> <span class="o">=</span> <span class="n">swapaxes</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">dim1</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">dim2</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> |
| <span class="n">B1</span> <span class="o">=</span> <span class="n">swapaxes</span><span class="p">(</span><span class="n">B</span><span class="p">,</span> <span class="n">dim1</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">dim2</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> |
| <span class="n">C</span> <span class="o">=</span> <span class="n">gemm2</span><span class="p">(</span><span class="n">A1</span><span class="p">,</span> <span class="n">B1</span><span class="p">)</span> |
| <span class="n">C</span> <span class="o">=</span> <span class="n">swapaxis</span><span class="p">(</span><span class="n">C</span><span class="p">,</span> <span class="n">dim1</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">dim2</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>When the input data is of type float32 and the environment variables MXNET_CUDA_ALLOW_TENSOR_CORE |
| and MXNET_CUDA_TENSOR_OP_MATH_ALLOW_CONVERSION are set to 1, this operator will try to use |
| pseudo-float16 precision (float32 math with float16 I/O) precision in order to use |
| Tensor Cores on suitable NVIDIA GPUs. This can sometimes give significant speedups.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">multiply</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]</span> |
| <span class="n">B</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]</span> |
| <span class="n">gemm2</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">B</span><span class="p">,</span> <span class="n">transpose_b</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">2.0</span><span class="p">)</span> |
| <span class="o">=</span> <span class="p">[[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">multiply</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">]]]</span> |
| <span class="n">B</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">]]]</span> |
| <span class="n">gemm2</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">B</span><span class="p">,</span> <span class="n">transpose_b</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">2.0</span><span class="p">)</span> |
| <span class="o">=</span> <span class="p">[[[</span><span class="mf">4.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.04</span> <span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L162</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of input matrices</p></li> |
| <li><p><strong>B</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of input matrices</p></li> |
| <li><p><strong>transpose_a</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Multiply with transposed of first input (A).</p></li> |
| <li><p><strong>transpose_b</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Multiply with transposed of second input (B).</p></li> |
| <li><p><strong>alpha</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scalar factor multiplied with A*B.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='-2'</em>) – Axis corresponding to the matrix row indices.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_inverse"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_inverse</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_inverse" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Compute the inverse of a matrix. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, <em>A</em> is a square matrix. We compute:</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>A</em><sup>-1</sup></p> |
| </div></blockquote> |
| <p>If <em>n>2</em>, <em>inverse</em> is performed separately on the trailing two dimensions |
| for all inputs (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">inverse</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> <span class="p">[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]]</span> |
| <span class="n">inverse</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="o">-</span><span class="mf">0.6</span><span class="p">,</span> <span class="mf">0.8</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.4</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.2</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">inverse</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> <span class="p">[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> <span class="p">[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]]</span> |
| <span class="n">inverse</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="o">-</span><span class="mf">0.6</span><span class="p">,</span> <span class="mf">0.8</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.4</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.2</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="o">-</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.5</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L919</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of square matrix</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_makediag"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_makediag</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">offset=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_makediag" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Constructs a square matrix with the input as diagonal. |
| Input is a tensor <em>A</em> of dimension <em>n >= 1</em>.</p> |
| <p>If <em>n=1</em>, then <em>A</em> represents the diagonal entries of a single square matrix. This matrix will be returned as a 2-dimensional tensor. |
| If <em>n>1</em>, then <em>A</em> represents a batch of diagonals of square matrices. The batch of diagonal matrices will be returned as an <em>n+1</em>-dimensional tensor.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">diagonal</span> <span class="n">matrix</span> <span class="n">construction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">]</span> |
| |
| <span class="n">makediag</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">]]</span> |
| |
| <span class="n">makediag</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">diagonal</span> <span class="n">matrix</span> <span class="n">construction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">3.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]]</span> |
| |
| <span class="n">makediag</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">3.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L546</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of diagonal entries</p></li> |
| <li><p><strong>offset</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – Offset of the diagonal versus the main diagonal. 0 corresponds to the main diagonal, a negative/positive value to diagonals below/above the main diagonal.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_maketrian"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_maketrian</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">offset=_Null</em>, <em class="sig-param">lower=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_maketrian" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Constructs a square matrix with the input representing a specific triangular sub-matrix. |
| This is basically the inverse of <em>linalg.extracttrian</em>. Input is a tensor <em>A</em> of dimension <em>n >= 1</em>.</p> |
| <p>If <em>n=1</em>, then <em>A</em> represents the entries of a triangular matrix which is lower triangular if <em>offset<0</em> or <em>offset=0</em>, <em>lower=true</em>. The resulting matrix is derived by first constructing the square |
| matrix with the entries outside the triangle set to zero and then adding <em>offset</em>-times an additional |
| diagonal with zero entries to the square matrix.</p> |
| <p>If <em>n>1</em>, then <em>A</em> represents a batch of triangular sub-matrices. The batch of corresponding square matrices is returned as an <em>n+1</em>-dimensional tensor.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">construction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">]</span> |
| |
| <span class="n">maketrian</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">]]</span> |
| |
| <span class="n">maketrian</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">lower</span><span class="o">=</span><span class="n">false</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">]]</span> |
| |
| <span class="n">maketrian</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">offset</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">]]</span> |
| <span class="n">maketrian</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">offset</span><span class="o">=-</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">construction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">5.0</span><span class="p">,</span> <span class="mf">6.0</span><span class="p">]]</span> |
| |
| <span class="n">maketrian</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">5.0</span><span class="p">,</span> <span class="mf">6.0</span><span class="p">]]]</span> |
| |
| <span class="n">maketrian</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">offset</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">3.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">,</span> <span class="mf">5.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">6.0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L672</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of triangular matrices stored as vectors</p></li> |
| <li><p><strong>offset</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – Offset of the diagonal versus the main diagonal. 0 corresponds to the main diagonal, a negative/positive value to diagonals below/above the main diagonal.</p></li> |
| <li><p><strong>lower</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Refer to the lower triangular matrix if lower=true, refer to the upper otherwise. Only relevant when offset=0</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_potrf"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_potrf</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_potrf" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Performs Cholesky factorization of a symmetric positive-definite matrix. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, the Cholesky factor <em>B</em> of the symmetric, positive definite matrix <em>A</em> is |
| computed. <em>B</em> is triangular (entries of upper or lower triangle are all zero), has |
| positive diagonal entries, and:</p> |
| <blockquote> |
| <div><p><em>A</em> = <em>B</em> * <em>B</em><sup>T</sup> if <em>lower</em> = <em>true</em> |
| <em>A</em> = <em>B</em><sup>T</sup> * <em>B</em> if <em>lower</em> = <em>false</em></p> |
| </div></blockquote> |
| <p>If <em>n>2</em>, <em>potrf</em> is performed separately on the trailing two dimensions for all inputs |
| (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">factorization</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">4.25</span><span class="p">]]</span> |
| <span class="n">potrf</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">factorization</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">4.25</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">16.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">17.0</span><span class="p">]]]</span> |
| <span class="n">potrf</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L213</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of input matrices to be decomposed</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_potri"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_potri</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_potri" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Performs matrix inversion from a Cholesky factorization. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, <em>A</em> is a triangular matrix (entries of upper or lower triangle are all zero) |
| with positive diagonal. We compute:</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>A</em><sup>-T</sup> * <em>A</em><sup>-1</sup> if <em>lower</em> = <em>true</em> |
| <em>out</em> = <em>A</em><sup>-1</sup> * <em>A</em><sup>-T</sup> if <em>lower</em> = <em>false</em></p> |
| </div></blockquote> |
| <p>In other words, if <em>A</em> is the Cholesky factor of a symmetric positive definite matrix |
| <em>B</em> (obtained by <em>potrf</em>), then</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>B</em><sup>-1</sup></p> |
| </div></blockquote> |
| <p>If <em>n>2</em>, <em>potri</em> is performed separately on the trailing two dimensions for all inputs |
| (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>Use this operator only if you are certain you need the inverse of <em>B</em>, and |
| cannot use the Cholesky factor <em>A</em> (<em>potrf</em>), together with backsubstitution |
| (<em>trsm</em>). The latter is numerically much safer, and also cheaper.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">inverse</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">]]</span> |
| <span class="n">potri</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.26563</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.0625</span><span class="p">],</span> <span class="p">[</span><span class="o">-</span><span class="mf">0.0625</span><span class="p">,</span> <span class="mf">0.25</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">inverse</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]]]</span> |
| <span class="n">potri</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">0.26563</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.0625</span><span class="p">],</span> <span class="p">[</span><span class="o">-</span><span class="mf">0.0625</span><span class="p">,</span> <span class="mf">0.25</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">0.06641</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.01562</span><span class="p">],</span> <span class="p">[</span><span class="o">-</span><span class="mf">0.01562</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span><span class="mi">0625</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L274</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of lower triangular matrices</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_slogdet"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_slogdet</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_slogdet" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Compute the sign and log of the determinant of a matrix. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, <em>A</em> is a square matrix. We compute:</p> |
| <blockquote> |
| <div><p><em>sign</em> = <em>sign(det(A))</em> |
| <em>logabsdet</em> = <em>log(abs(det(A)))</em></p> |
| </div></blockquote> |
| <p>If <em>n>2</em>, <em>slogdet</em> is performed separately on the trailing two dimensions |
| for all inputs (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The gradient is not properly defined on sign, so the gradient of |
| it is not backwarded.</p> |
| </div> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>No gradient is backwarded when A is non-invertible. Please see |
| the docs of operator det for detail.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">signed</span> <span class="n">log</span> <span class="n">determinant</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]</span> |
| <span class="n">sign</span><span class="p">,</span> <span class="n">logabsdet</span> <span class="o">=</span> <span class="n">slogdet</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> |
| <span class="n">sign</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.</span><span class="p">]</span> |
| <span class="n">logabsdet</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.609438</span><span class="p">]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">signed</span> <span class="n">log</span> <span class="n">determinant</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> <span class="p">[</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]]]</span> |
| <span class="n">sign</span><span class="p">,</span> <span class="n">logabsdet</span> <span class="o">=</span> <span class="n">slogdet</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> |
| <span class="n">sign</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.</span><span class="p">]</span> |
| <span class="n">logabsdet</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.609438</span><span class="p">,</span> <span class="o">-</span><span class="n">inf</span><span class="p">,</span> <span class="mf">1.609438</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L1033</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of square matrix</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_sumlogdiag"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_sumlogdiag</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_sumlogdiag" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the sum of the logarithms of the diagonal elements of a square matrix. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, <em>A</em> must be square with positive diagonal entries. We sum the natural |
| logarithms of the diagonal elements, the result has shape (1,).</p> |
| <p>If <em>n>2</em>, <em>sumlogdiag</em> is performed separately on the trailing two dimensions for all |
| inputs (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">reduction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">7.0</span><span class="p">]]</span> |
| <span class="n">sumlogdiag</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.9459</span><span class="p">]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">reduction</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">7.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">3.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mf">17.0</span><span class="p">]]]</span> |
| <span class="n">sumlogdiag</span><span class="p">(</span><span class="n">A</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.9459</span><span class="p">,</span> <span class="mf">3.9318</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L444</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of square matrices</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_syrk"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_syrk</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">transpose=_Null</em>, <em class="sig-param">alpha=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_syrk" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Multiplication of matrix with its transpose. |
| Input is a tensor <em>A</em> of dimension <em>n >= 2</em>.</p> |
| <p>If <em>n=2</em>, the operator performs the BLAS3 function <em>syrk</em>:</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>alpha</em> * <em>A</em> * <em>A</em><sup>T</sup></p> |
| </div></blockquote> |
| <p>if <em>transpose=False</em>, or</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>alpha</em> * <em>A</em><sup>T</sup> * <em>A</em></p> |
| </div></blockquote> |
| <p>if <em>transpose=True</em>.</p> |
| <p>If <em>n>2</em>, <em>syrk</em> is performed separately on the trailing two dimensions for all |
| inputs (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">multiply</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]</span> |
| <span class="n">syrk</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">1.</span><span class="p">,</span> <span class="n">transpose</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span> |
| <span class="o">=</span> <span class="p">[[</span><span class="mf">14.</span><span class="p">,</span> <span class="mf">32.</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">32.</span><span class="p">,</span> <span class="mf">77.</span><span class="p">]]</span> |
| <span class="n">syrk</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">1.</span><span class="p">,</span> <span class="n">transpose</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> |
| <span class="o">=</span> <span class="p">[[</span><span class="mf">17.</span><span class="p">,</span> <span class="mf">22.</span><span class="p">,</span> <span class="mf">27.</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">22.</span><span class="p">,</span> <span class="mf">29.</span><span class="p">,</span> <span class="mf">36.</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">27.</span><span class="p">,</span> <span class="mf">36.</span><span class="p">,</span> <span class="mf">45.</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">multiply</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">]]]</span> |
| <span class="n">syrk</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">2.</span><span class="p">,</span> <span class="n">transpose</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">4.</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.04</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L729</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of input matrices</p></li> |
| <li><p><strong>transpose</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Use transpose of input matrix.</p></li> |
| <li><p><strong>alpha</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scalar factor to be applied to the result.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_trmm"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_trmm</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">B=None</em>, <em class="sig-param">transpose=_Null</em>, <em class="sig-param">rightside=_Null</em>, <em class="sig-param">lower=_Null</em>, <em class="sig-param">alpha=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_trmm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Performs multiplication with a lower triangular matrix. |
| Input are tensors <em>A</em>, <em>B</em>, each of dimension <em>n >= 2</em> and having the same shape |
| on the leading <em>n-2</em> dimensions.</p> |
| <p>If <em>n=2</em>, <em>A</em> must be triangular. The operator performs the BLAS3 function |
| <em>trmm</em>:</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>alpha</em> * <em>op</em>(<em>A</em>) * <em>B</em></p> |
| </div></blockquote> |
| <p>if <em>rightside=False</em>, or</p> |
| <blockquote> |
| <div><p><em>out</em> = <em>alpha</em> * <em>B</em> * <em>op</em>(<em>A</em>)</p> |
| </div></blockquote> |
| <p>if <em>rightside=True</em>. Here, <em>alpha</em> is a scalar parameter, and <em>op()</em> is either the |
| identity or the matrix transposition (depending on <em>transpose</em>).</p> |
| <p>If <em>n>2</em>, <em>trmm</em> is performed separately on the trailing two dimensions for all inputs |
| (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">triangular</span> <span class="n">matrix</span> <span class="n">multiply</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]</span> |
| <span class="n">B</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]</span> |
| <span class="n">trmm</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">B</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">2.0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">triangular</span> <span class="n">matrix</span> <span class="n">multiply</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]]</span> |
| <span class="n">B</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">]]]</span> |
| <span class="n">trmm</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">B</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">2.0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L332</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of lower triangular matrices</p></li> |
| <li><p><strong>B</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of matrices</p></li> |
| <li><p><strong>transpose</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Use transposed of the triangular matrix</p></li> |
| <li><p><strong>rightside</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Multiply triangular matrix from the right to non-triangular one.</p></li> |
| <li><p><strong>lower</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – True if the triangular matrix is lower triangular, false if it is upper triangular.</p></li> |
| <li><p><strong>alpha</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scalar factor to be applied to the result.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.linalg_trsm"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">linalg_trsm</code><span class="sig-paren">(</span><em class="sig-param">A=None</em>, <em class="sig-param">B=None</em>, <em class="sig-param">transpose=_Null</em>, <em class="sig-param">rightside=_Null</em>, <em class="sig-param">lower=_Null</em>, <em class="sig-param">alpha=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.linalg_trsm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Solves matrix equation involving a lower triangular matrix. |
| Input are tensors <em>A</em>, <em>B</em>, each of dimension <em>n >= 2</em> and having the same shape |
| on the leading <em>n-2</em> dimensions.</p> |
| <p>If <em>n=2</em>, <em>A</em> must be triangular. The operator performs the BLAS3 function |
| <em>trsm</em>, solving for <em>out</em> in:</p> |
| <blockquote> |
| <div><p><em>op</em>(<em>A</em>) * <em>out</em> = <em>alpha</em> * <em>B</em></p> |
| </div></blockquote> |
| <p>if <em>rightside=False</em>, or</p> |
| <blockquote> |
| <div><p><em>out</em> * <em>op</em>(<em>A</em>) = <em>alpha</em> * <em>B</em></p> |
| </div></blockquote> |
| <p>if <em>rightside=True</em>. Here, <em>alpha</em> is a scalar parameter, and <em>op()</em> is either the |
| identity or the matrix transposition (depending on <em>transpose</em>).</p> |
| <p>If <em>n>2</em>, <em>trsm</em> is performed separately on the trailing two dimensions for all inputs |
| (batch mode).</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The operator supports float32 and float64 data types only.</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">Single</span> <span class="n">matrix</span> <span class="n">solve</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]</span> |
| <span class="n">B</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]]</span> |
| <span class="n">trsm</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">B</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">0.5</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]</span> |
| |
| <span class="n">Batch</span> <span class="n">matrix</span> <span class="n">solve</span> |
| <span class="n">A</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]],</span> <span class="p">[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]]]</span> |
| <span class="n">B</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">,</span> <span class="mf">4.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">8.0</span><span class="p">,</span> <span class="mf">8.0</span><span class="p">,</span> <span class="mf">8.0</span><span class="p">]]]</span> |
| <span class="n">trsm</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">B</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">0.5</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.0</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/la_op.cc:L395</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>A</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of lower triangular matrices</p></li> |
| <li><p><strong>B</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Tensor of matrices</p></li> |
| <li><p><strong>transpose</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Use transposed of the triangular matrix</p></li> |
| <li><p><strong>rightside</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Multiply triangular matrix from the right to non-triangular one.</p></li> |
| <li><p><strong>lower</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – True if the triangular matrix is lower triangular, false if it is upper triangular.</p></li> |
| <li><p><strong>alpha</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Scalar factor to be applied to the result.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.log"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">log</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.log" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise Natural logarithmic value of the input.</p> |
| <p>The natural logarithm is logarithm in base <em>e</em>, so that <code class="docutils literal notranslate"><span class="pre">log(exp(x))</span> <span class="pre">=</span> <span class="pre">x</span></code></p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">log</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L77</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.log10"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">log10</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.log10" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise Base-10 logarithmic value of the input.</p> |
| <p><code class="docutils literal notranslate"><span class="pre">10**log10(x)</span> <span class="pre">=</span> <span class="pre">x</span></code></p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">log10</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L94</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.log1p"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">log1p</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.log1p" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise <code class="docutils literal notranslate"><span class="pre">log(1</span> <span class="pre">+</span> <span class="pre">x)</span></code> value of the input.</p> |
| <p>This function is more accurate than <code class="docutils literal notranslate"><span class="pre">log(1</span> <span class="pre">+</span> <span class="pre">x)</span></code> for small <code class="docutils literal notranslate"><span class="pre">x</span></code> so that |
| <span class="math notranslate nohighlight">\(1+x\approx 1\)</span></p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">log1p</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>log1p(default) = default</p></li> |
| <li><p>log1p(row_sparse) = row_sparse</p></li> |
| <li><p>log1p(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L199</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.log2"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">log2</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.log2" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise Base-2 logarithmic value of the input.</p> |
| <p><code class="docutils literal notranslate"><span class="pre">2**log2(x)</span> <span class="pre">=</span> <span class="pre">x</span></code></p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">log2</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L106</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.log_softmax"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">log_softmax</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">temperature=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">use_length=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.log_softmax" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the log softmax of the input. |
| This is equivalent to computing softmax followed by log.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></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">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="mf">.1</span><span class="p">])</span> |
| <span class="gp">>>> </span><span class="n">mx</span><span class="o">.</span><span class="n">nd</span><span class="o">.</span><span class="n">log_softmax</span><span class="p">(</span><span class="n">x</span><span class="p">)</span><span class="o">.</span><span class="n">asnumpy</span><span class="p">()</span> |
| <span class="go">array([-1.41702998, -0.41702995, -2.31702995], dtype=float32)</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">array</span><span class="p">(</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="mf">.1</span><span class="p">],[</span><span class="mf">.1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">]]</span> <span class="p">)</span> |
| <span class="gp">>>> </span><span class="n">mx</span><span class="o">.</span><span class="n">nd</span><span class="o">.</span><span class="n">log_softmax</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span><span class="o">.</span><span class="n">asnumpy</span><span class="p">()</span> |
| <span class="go">array([[-0.34115392, -0.69314718, -1.24115396],</span> |
| <span class="go"> [-1.24115396, -0.69314718, -0.34115392]], dtype=float32)</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>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='-1'</em>) – The axis along which to compute softmax.</p></li> |
| <li><p><strong>temperature</strong> (<em>double</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Temperature parameter in softmax</p></li> |
| <li><p><strong>dtype</strong> (<em>{None</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to the same as input’s dtype if not defined (dtype=None).</p></li> |
| <li><p><strong>use_length</strong> (<em>boolean</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to use the length input as a mask over the data input.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.logical_not"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">logical_not</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.logical_not" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the result of logical NOT (!) function</p> |
| <p class="rubric">Example</p> |
| <p>logical_not([-2., 0., 1.]) = [0., 1., 0.]</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.make_loss"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">make_loss</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.make_loss" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Make your own loss function in network construction.</p> |
| <p>This operator accepts a customized loss function symbol as a terminal loss and |
| the symbol should be an operator with no backward dependency. |
| The output of this function is the gradient of loss with respect to the input data.</p> |
| <p>For example, if you are a making a cross entropy loss function. Assume <code class="docutils literal notranslate"><span class="pre">out</span></code> is the |
| predicted output and <code class="docutils literal notranslate"><span class="pre">label</span></code> is the true label, then the cross entropy can be defined as:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">cross_entropy</span> <span class="o">=</span> <span class="n">label</span> <span class="o">*</span> <span class="n">log</span><span class="p">(</span><span class="n">out</span><span class="p">)</span> <span class="o">+</span> <span class="p">(</span><span class="mi">1</span> <span class="o">-</span> <span class="n">label</span><span class="p">)</span> <span class="o">*</span> <span class="n">log</span><span class="p">(</span><span class="mi">1</span> <span class="o">-</span> <span class="n">out</span><span class="p">)</span> |
| <span class="n">loss</span> <span class="o">=</span> <span class="n">make_loss</span><span class="p">(</span><span class="n">cross_entropy</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>We will need to use <code class="docutils literal notranslate"><span class="pre">make_loss</span></code> when we are creating our own loss function or we want to |
| combine multiple loss functions. Also we may want to stop some variables’ gradients |
| from backpropagation. See more detail in <code class="docutils literal notranslate"><span class="pre">BlockGrad</span></code> or <code class="docutils literal notranslate"><span class="pre">stop_gradient</span></code>.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">make_loss</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>make_loss(default) = default</p></li> |
| <li><p>make_loss(row_sparse) = row_sparse</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L358</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.max"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">max</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.max" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the max of array elements over given axes.</p> |
| <p>Defined in src/operator/tensor/./broadcast_reduce_op.h:L31</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.max_axis"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">max_axis</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.max_axis" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the max of array elements over given axes.</p> |
| <p>Defined in src/operator/tensor/./broadcast_reduce_op.h:L31</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.mean"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">mean</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.mean" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the mean of array elements over given axes.</p> |
| <p>Defined in src/operator/tensor/./broadcast_reduce_op.h:L83</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.min"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">min</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.min" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the min of array elements over given axes.</p> |
| <p>Defined in src/operator/tensor/./broadcast_reduce_op.h:L46</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.min_axis"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">min_axis</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.min_axis" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the min of array elements over given axes.</p> |
| <p>Defined in src/operator/tensor/./broadcast_reduce_op.h:L46</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.moments"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">moments</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axes=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.moments" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Calculate the mean and variance of <cite>data</cite>.</p> |
| <p>The mean and variance are calculated by aggregating the contents of data across axes. |
| If x is 1-D and axes = [0] this is just the mean and variance of a vector.</p> |
| <p class="rubric">Example</p> |
| <p>x = [[1, 2, 3], [4, 5, 6]] |
| mean, var = moments(data=x, axes=[0]) |
| mean = [2.5, 3.5, 4.5] |
| var = [2.25, 2.25, 2.25] |
| mean, var = moments(data=x, axes=[1]) |
| mean = [2.0, 5.0] |
| var = [0.66666667, 0.66666667] |
| mean, var = moments(data=x, axis=[0, 1]) |
| mean = [3.5] |
| var = [2.9166667]</p> |
| <p>Defined in src/operator/nn/moments.cc:L53</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input ndarray</p></li> |
| <li><p><strong>axes</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Array of ints. Axes along which to compute mean and variance.</p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – produce moments with the same dimensionality as the input.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.mp_lamb_update_phase1"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">mp_lamb_update_phase1</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">mean=None</em>, <em class="sig-param">var=None</em>, <em class="sig-param">weight32=None</em>, <em class="sig-param">beta1=_Null</em>, <em class="sig-param">beta2=_Null</em>, <em class="sig-param">epsilon=_Null</em>, <em class="sig-param">t=_Null</em>, <em class="sig-param">bias_correction=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.mp_lamb_update_phase1" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Mixed Precision version of Phase I of lamb update |
| it performs the following operations and returns g:.</p> |
| <blockquote> |
| <div><p>Link to paper: <a class="reference external" href="https://arxiv.org/pdf/1904.00962.pdf">https://arxiv.org/pdf/1904.00962.pdf</a></p> |
| <div class="math notranslate nohighlight"> |
| \[ \begin{align}\begin{aligned}\begin{gather*} |
| grad32 = grad(float16) * rescale_grad |
| if (grad < -clip_gradient) |
| then |
| grad = -clip_gradient |
| if (grad > clip_gradient) |
| then |
| grad = clip_gradient\\mean = beta1 * mean + (1 - beta1) * grad; |
| variance = beta2 * variance + (1. - beta2) * grad ^ 2;\\if (bias_correction) |
| then |
| mean_hat = mean / (1. - beta1^t); |
| var_hat = var / (1 - beta2^t); |
| g = mean_hat / (var_hat^(1/2) + epsilon) + wd * weight32; |
| else |
| g = mean / (var_data^(1/2) + epsilon) + wd * weight32; |
| \end{gather*}\end{aligned}\end{align} \]</div> |
| </div></blockquote> |
| <p>Defined in src/operator/optimizer_op.cc:L1032</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>mean</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Moving mean</p></li> |
| <li><p><strong>var</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Moving variance</p></li> |
| <li><p><strong>weight32</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight32</p></li> |
| <li><p><strong>beta1</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.899999976</em>) – The decay rate for the 1st moment estimates.</p></li> |
| <li><p><strong>beta2</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.999000013</em>) – The decay rate for the 2nd moment estimates.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=9.99999997e-07</em>) – A small constant for numerical stability.</p></li> |
| <li><p><strong>t</strong> (<em>int</em><em>, </em><em>required</em>) – Index update count.</p></li> |
| <li><p><strong>bias_correction</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Whether to use bias correction.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>required</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.mp_lamb_update_phase2"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">mp_lamb_update_phase2</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">g=None</em>, <em class="sig-param">r1=None</em>, <em class="sig-param">r2=None</em>, <em class="sig-param">weight32=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">lower_bound=_Null</em>, <em class="sig-param">upper_bound=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.mp_lamb_update_phase2" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Mixed Precision version Phase II of lamb update |
| it performs the following operations and updates grad.</p> |
| <blockquote> |
| <div><p>Link to paper: <a class="reference external" href="https://arxiv.org/pdf/1904.00962.pdf">https://arxiv.org/pdf/1904.00962.pdf</a></p> |
| <div class="math notranslate nohighlight"> |
| \[ \begin{align}\begin{aligned}\begin{gather*} |
| if (lower_bound >= 0) |
| then |
| r1 = max(r1, lower_bound) |
| if (upper_bound >= 0) |
| then |
| r1 = max(r1, upper_bound)\\if (r1 == 0 or r2 == 0) |
| then |
| lr = lr |
| else |
| lr = lr * (r1/r2) |
| weight32 = weight32 - lr * g |
| weight(float16) = weight32 |
| \end{gather*}\end{aligned}\end{align} \]</div> |
| </div></blockquote> |
| <p>Defined in src/operator/optimizer_op.cc:L1074</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>g</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Output of mp_lamb_update_phase 1</p></li> |
| <li><p><strong>r1</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – r1</p></li> |
| <li><p><strong>r2</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – r2</p></li> |
| <li><p><strong>weight32</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight32</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>lower_bound</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Lower limit of norm of weight. If lower_bound <= 0, Lower limit is not set</p></li> |
| <li><p><strong>upper_bound</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Upper limit of norm of weight. If upper_bound <= 0, Upper limit is not set</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.mp_nag_mom_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">mp_nag_mom_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">mom=None</em>, <em class="sig-param">weight32=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">momentum=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.mp_nag_mom_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for multi-precision Nesterov Accelerated Gradient( NAG) optimizer.</p> |
| <p>Defined in src/operator/optimizer_op.cc:L744</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>mom</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Momentum</p></li> |
| <li><p><strong>weight32</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight32</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The decay rate of momentum estimates at each epoch.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.mp_sgd_mom_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">mp_sgd_mom_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">mom=None</em>, <em class="sig-param">weight32=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">momentum=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">lazy_update=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.mp_sgd_mom_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Updater function for multi-precision sgd optimizer</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>mom</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Momentum</p></li> |
| <li><p><strong>weight32</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight32</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The decay rate of momentum estimates at each epoch.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>lazy_update</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – If true, lazy updates are applied if gradient’s stype is row_sparse and both weight and momentum have the same stype</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.mp_sgd_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">mp_sgd_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">weight32=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">lazy_update=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.mp_sgd_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Updater function for multi-precision sgd optimizer</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – gradient</p></li> |
| <li><p><strong>weight32</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight32</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>lazy_update</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – If true, lazy updates are applied if gradient’s stype is row_sparse.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.multi_all_finite"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">multi_all_finite</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.multi_all_finite" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Check if all the float numbers in all the arrays are finite (used for AMP)</p> |
| <p>Defined in src/operator/contrib/all_finite.cc:L132</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Arrays</p></li> |
| <li><p><strong>num_arrays</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of arrays.</p></li> |
| <li><p><strong>init_output</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Initialize output to 1.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.multi_lars"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">multi_lars</code><span class="sig-paren">(</span><em class="sig-param">lrs=None</em>, <em class="sig-param">weights_sum_sq=None</em>, <em class="sig-param">grads_sum_sq=None</em>, <em class="sig-param">wds=None</em>, <em class="sig-param">eta=_Null</em>, <em class="sig-param">eps=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.multi_lars" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Compute the LARS coefficients of multiple weights and grads from their sums of square”</p> |
| <p>Defined in src/operator/contrib/multi_lars.cc:L36</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lrs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Learning rates to scale by LARS coefficient</p></li> |
| <li><p><strong>weights_sum_sq</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – sum of square of weights arrays</p></li> |
| <li><p><strong>grads_sum_sq</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – sum of square of gradients arrays</p></li> |
| <li><p><strong>wds</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – weight decays</p></li> |
| <li><p><strong>eta</strong> (<em>float</em><em>, </em><em>required</em>) – LARS eta</p></li> |
| <li><p><strong>eps</strong> (<em>float</em><em>, </em><em>required</em>) – LARS eps</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Gradient rescaling factor</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.multi_mp_sgd_mom_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">multi_mp_sgd_mom_update</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.multi_mp_sgd_mom_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Momentum update function for multi-precision Stochastic Gradient Descent (SGD) optimizer.</p> |
| <p>Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below:</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-1} - \alpha * \nabla J(W_{t-1})\\ |
| W_t = W_{t-1} + v_t\end{split}\]</div> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">v</span> <span class="o">=</span> <span class="n">momentum</span> <span class="o">*</span> <span class="n">v</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="n">gradient</span> |
| <span class="n">weight</span> <span class="o">+=</span> <span class="n">v</span> |
| </pre></div> |
| </div> |
| <p>Where the parameter <code class="docutils literal notranslate"><span class="pre">momentum</span></code> is the decay rate of momentum estimates at each epoch.</p> |
| <p>Defined in src/operator/optimizer_op.cc:L471</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Weights</p></li> |
| <li><p><strong>lrs</strong> (<em>tuple of <float></em><em>, </em><em>required</em>) – Learning rates.</p></li> |
| <li><p><strong>wds</strong> (<em>tuple of <float></em><em>, </em><em>required</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The decay rate of momentum estimates at each epoch.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>num_weights</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of updated weights.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.multi_mp_sgd_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">multi_mp_sgd_update</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.multi_mp_sgd_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for multi-precision Stochastic Gradient Descent (SDG) optimizer.</p> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">weight</span> <span class="o">=</span> <span class="n">weight</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="p">(</span><span class="n">gradient</span> <span class="o">+</span> <span class="n">wd</span> <span class="o">*</span> <span class="n">weight</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/optimizer_op.cc:L416</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Weights</p></li> |
| <li><p><strong>lrs</strong> (<em>tuple of <float></em><em>, </em><em>required</em>) – Learning rates.</p></li> |
| <li><p><strong>wds</strong> (<em>tuple of <float></em><em>, </em><em>required</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>num_weights</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of updated weights.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.multi_sgd_mom_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">multi_sgd_mom_update</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.multi_sgd_mom_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Momentum update function for Stochastic Gradient Descent (SGD) optimizer.</p> |
| <p>Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below:</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-1} - \alpha * \nabla J(W_{t-1})\\ |
| W_t = W_{t-1} + v_t\end{split}\]</div> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">v</span> <span class="o">=</span> <span class="n">momentum</span> <span class="o">*</span> <span class="n">v</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="n">gradient</span> |
| <span class="n">weight</span> <span class="o">+=</span> <span class="n">v</span> |
| </pre></div> |
| </div> |
| <p>Where the parameter <code class="docutils literal notranslate"><span class="pre">momentum</span></code> is the decay rate of momentum estimates at each epoch.</p> |
| <p>Defined in src/operator/optimizer_op.cc:L373</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Weights, gradients and momentum</p></li> |
| <li><p><strong>lrs</strong> (<em>tuple of <float></em><em>, </em><em>required</em>) – Learning rates.</p></li> |
| <li><p><strong>wds</strong> (<em>tuple of <float></em><em>, </em><em>required</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The decay rate of momentum estimates at each epoch.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>num_weights</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of updated weights.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.multi_sgd_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">multi_sgd_update</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.multi_sgd_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for Stochastic Gradient Descent (SDG) optimizer.</p> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">weight</span> <span class="o">=</span> <span class="n">weight</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="p">(</span><span class="n">gradient</span> <span class="o">+</span> <span class="n">wd</span> <span class="o">*</span> <span class="n">weight</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/optimizer_op.cc:L328</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Weights</p></li> |
| <li><p><strong>lrs</strong> (<em>tuple of <float></em><em>, </em><em>required</em>) – Learning rates.</p></li> |
| <li><p><strong>wds</strong> (<em>tuple of <float></em><em>, </em><em>required</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>num_weights</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of updated weights.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.multi_sum_sq"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">multi_sum_sq</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.multi_sum_sq" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Compute the sums of squares of multiple arrays</p> |
| <p>Defined in src/operator/contrib/multi_sum_sq.cc:L35</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Arrays</p></li> |
| <li><p><strong>num_arrays</strong> (<em>int</em><em>, </em><em>required</em>) – number of input arrays.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.nag_mom_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">nag_mom_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">mom=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">momentum=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.nag_mom_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for Nesterov Accelerated Gradient( NAG) optimizer. |
| It updates the weights using the following formula,</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}v_t = \gamma v_{t-1} + \eta * \nabla J(W_{t-1} - \gamma v_{t-1})\\ |
| W_t = W_{t-1} - v_t\end{split}\]</div> |
| <p>Where |
| <span class="math notranslate nohighlight">\(\eta\)</span> is the learning rate of the optimizer |
| <span class="math notranslate nohighlight">\(\gamma\)</span> is the decay rate of the momentum estimate |
| <span class="math notranslate nohighlight">\(\v_t\)</span> is the update vector at time step <cite>t</cite> |
| <span class="math notranslate nohighlight">\(\W_t\)</span> is the weight vector at time step <cite>t</cite></p> |
| <p>Defined in src/operator/optimizer_op.cc:L725</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>mom</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Momentum</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The decay rate of momentum estimates at each epoch.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.nanprod"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">nanprod</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.nanprod" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the product of array elements over given axes treating Not a Numbers (<code class="docutils literal notranslate"><span class="pre">NaN</span></code>) as one.</p> |
| <p>Defined in src/operator/tensor/broadcast_reduce_prod_value.cc:L46</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.nansum"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">nansum</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.nansum" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the sum of array elements over given axes treating Not a Numbers (<code class="docutils literal notranslate"><span class="pre">NaN</span></code>) as zero.</p> |
| <p>Defined in src/operator/tensor/broadcast_reduce_sum_value.cc:L101</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.negative"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">negative</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.negative" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Numerical negative of the argument, element-wise.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">negative</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>negative(default) = default</p></li> |
| <li><p>negative(row_sparse) = row_sparse</p></li> |
| <li><p>negative(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.norm"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">norm</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">ord=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">out_dtype=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.norm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the norm on an NDArray.</p> |
| <p>This operator computes the norm on an NDArray with the specified axis, depending |
| on the value of the ord parameter. By default, it computes the L2 norm on the entire |
| array. Currently only ord=2 supports sparse ndarrays.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]],</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="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">]]]</span> |
| |
| <span class="n">norm</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="nb">ord</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">3.1622777</span> <span class="mf">4.472136</span> <span class="p">]</span> |
| <span class="p">[</span><span class="mf">5.3851647</span> <span class="mf">6.3245554</span><span class="p">]]</span> |
| |
| <span class="n">norm</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="nb">ord</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">4.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]</span> |
| |
| <span class="n">rsp</span> <span class="o">=</span> <span class="n">x</span><span class="o">.</span><span class="n">cast_storage</span><span class="p">(</span><span class="s1">'row_sparse'</span><span class="p">)</span> |
| |
| <span class="n">norm</span><span class="p">(</span><span class="n">rsp</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">5.47722578</span><span class="p">]</span> |
| |
| <span class="n">csr</span> <span class="o">=</span> <span class="n">x</span><span class="o">.</span><span class="n">cast_storage</span><span class="p">(</span><span class="s1">'csr'</span><span class="p">)</span> |
| |
| <span class="n">norm</span><span class="p">(</span><span class="n">csr</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">5.47722578</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_norm_value.cc:L88</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>ord</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='2'</em>) – Order of the norm. Currently ord=1 and ord=2 is supported.</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <dl class="simple"> |
| <dt>The axis or axes along which to perform the reduction.</dt><dd><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>. |
| If <cite>axis</cite> is int, a reduction is performed on a particular axis. |
| If <cite>axis</cite> is a 2-tuple, it specifies the axes that hold 2-D matrices, |
| and the matrix norms of these matrices are computed.</p> |
| </dd> |
| </dl> |
| </p></li> |
| <li><p><strong>out_dtype</strong> (<em>{None</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'</em><em>, </em><em>'int8'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – The data type of the output.</p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axis is left in the result as dimension with size one.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.normal"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">normal</code><span class="sig-paren">(</span><em class="sig-param">loc=_Null</em>, <em class="sig-param">scale=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.normal" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from a normal (Gaussian) distribution.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The existing alias <code class="docutils literal notranslate"><span class="pre">normal</span></code> is deprecated.</p> |
| </div> |
| <p>Samples are distributed according to a normal distribution parametrized by <em>loc</em> (mean) and <em>scale</em> |
| (standard deviation).</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">normal</span><span class="p">(</span><span class="n">loc</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">scale</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">1.89171135</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.16881478</span><span class="p">],</span> |
| <span class="p">[</span><span class="o">-</span><span class="mf">1.23474145</span><span class="p">,</span> <span class="mf">1.55807114</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L112</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>loc</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Mean of the distribution.</p></li> |
| <li><p><strong>scale</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Standard deviation of the distribution.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.one_hot"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">one_hot</code><span class="sig-paren">(</span><em class="sig-param">indices=None</em>, <em class="sig-param">depth=_Null</em>, <em class="sig-param">on_value=_Null</em>, <em class="sig-param">off_value=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.one_hot" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a one-hot array.</p> |
| <p>The locations represented by <cite>indices</cite> take value <cite>on_value</cite>, while all |
| other locations take value <cite>off_value</cite>.</p> |
| <p><cite>one_hot</cite> operation with <cite>indices</cite> of shape <code class="docutils literal notranslate"><span class="pre">(i0,</span> <span class="pre">i1)</span></code> and <cite>depth</cite> of <code class="docutils literal notranslate"><span class="pre">d</span></code> would result |
| in an output array of shape <code class="docutils literal notranslate"><span class="pre">(i0,</span> <span class="pre">i1,</span> <span class="pre">d)</span></code> with:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">output</span><span class="p">[</span><span class="n">i</span><span class="p">,</span><span class="n">j</span><span class="p">,:]</span> <span class="o">=</span> <span class="n">off_value</span> |
| <span class="n">output</span><span class="p">[</span><span class="n">i</span><span class="p">,</span><span class="n">j</span><span class="p">,</span><span class="n">indices</span><span class="p">[</span><span class="n">i</span><span class="p">,</span><span class="n">j</span><span class="p">]]</span> <span class="o">=</span> <span class="n">on_value</span> |
| </pre></div> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">one_hot</span><span class="p">([</span><span class="mi">1</span><span class="p">,</span><span class="mi">0</span><span class="p">,</span><span class="mi">2</span><span class="p">,</span><span class="mi">0</span><span class="p">],</span> <span class="mi">3</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span> <span class="mf">1.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="n">one_hot</span><span class="p">([</span><span class="mi">1</span><span class="p">,</span><span class="mi">0</span><span class="p">,</span><span class="mi">2</span><span class="p">,</span><span class="mi">0</span><span class="p">],</span> <span class="mi">3</span><span class="p">,</span> <span class="n">on_value</span><span class="o">=</span><span class="mi">8</span><span class="p">,</span> <span class="n">off_value</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> |
| <span class="n">dtype</span><span class="o">=</span><span class="s1">'int32'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">1</span> <span class="mi">8</span> <span class="mi">1</span><span class="p">]</span> |
| <span class="p">[</span><span class="mi">8</span> <span class="mi">1</span> <span class="mi">1</span><span class="p">]</span> |
| <span class="p">[</span><span class="mi">1</span> <span class="mi">1</span> <span class="mi">8</span><span class="p">]</span> |
| <span class="p">[</span><span class="mi">8</span> <span class="mi">1</span> <span class="mi">1</span><span class="p">]]</span> |
| |
| <span class="n">one_hot</span><span class="p">([[</span><span class="mi">1</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">0</span><span class="p">],[</span><span class="mi">2</span><span class="p">,</span><span class="mi">0</span><span class="p">]],</span> <span class="mi">3</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">0.</span> <span class="mf">1.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">0.</span> <span class="mf">1.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/indexing_op.cc:L882</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>indices</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – array of locations where to set on_value</p></li> |
| <li><p><strong>depth</strong> (<em>int</em><em>, </em><em>required</em>) – Depth of the one hot dimension.</p></li> |
| <li><p><strong>on_value</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – The value assigned to the locations represented by indices.</p></li> |
| <li><p><strong>off_value</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The value assigned to the locations not represented by indices.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'bfloat16'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'</em><em>, </em><em>'int8'</em><em>, </em><em>'uint8'}</em><em>,</em><em>optional</em><em>, </em><em>default='float32'</em>) – DType of the output</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.ones_like"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">ones_like</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.ones_like" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Return an array of ones with the same shape and type |
| as the input array.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="n">ones_like</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</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>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.pad"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">pad</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">mode=_Null</em>, <em class="sig-param">pad_width=_Null</em>, <em class="sig-param">constant_value=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.pad" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Pads an input array with a constant or edge values of the array.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><cite>Pad</cite> is deprecated. Use <cite>pad</cite> instead.</p> |
| </div> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>Current implementation only supports 4D and 5D input arrays with padding applied |
| only on axes 1, 2 and 3. Expects axes 4 and 5 in <cite>pad_width</cite> to be zero.</p> |
| </div> |
| <p>This operation pads an input array with either a <cite>constant_value</cite> or edge values |
| along each axis of the input array. The amount of padding is specified by <cite>pad_width</cite>.</p> |
| <p><cite>pad_width</cite> is a tuple of integer padding widths for each axis of the format |
| <code class="docutils literal notranslate"><span class="pre">(before_1,</span> <span class="pre">after_1,</span> <span class="pre">...</span> <span class="pre">,</span> <span class="pre">before_N,</span> <span class="pre">after_N)</span></code>. The <cite>pad_width</cite> should be of length <code class="docutils literal notranslate"><span class="pre">2*N</span></code> |
| where <code class="docutils literal notranslate"><span class="pre">N</span></code> is the number of dimensions of the array.</p> |
| <p>For dimension <code class="docutils literal notranslate"><span class="pre">N</span></code> of the input array, <code class="docutils literal notranslate"><span class="pre">before_N</span></code> and <code class="docutils literal notranslate"><span class="pre">after_N</span></code> indicates how many values |
| to add before and after the elements of the array along dimension <code class="docutils literal notranslate"><span class="pre">N</span></code>. |
| The widths of the higher two dimensions <code class="docutils literal notranslate"><span class="pre">before_1</span></code>, <code class="docutils literal notranslate"><span class="pre">after_1</span></code>, <code class="docutils literal notranslate"><span class="pre">before_2</span></code>, |
| <code class="docutils literal notranslate"><span class="pre">after_2</span></code> must be 0.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[[</span> <span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span> <span class="mf">5.</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">7.</span> <span class="mf">8.</span> <span class="mf">9.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">10.</span> <span class="mf">11.</span> <span class="mf">12.</span><span class="p">]]]</span> |
| |
| |
| <span class="p">[[[</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">13.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">14.</span> <span class="mf">15.</span> <span class="mf">16.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">17.</span> <span class="mf">18.</span> <span class="mf">19.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">20.</span> <span class="mf">21.</span> <span class="mf">22.</span><span class="p">]]]]</span> |
| |
| <span class="n">pad</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">mode</span><span class="o">=</span><span class="s2">"edge"</span><span class="p">,</span> <span class="n">pad_width</span><span class="o">=</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</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">1</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="o">=</span> |
| |
| <span class="p">[[[[</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">3.</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">3.</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">5.</span> <span class="mf">6.</span> <span class="mf">6.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span> <span class="mf">4.</span> <span class="mf">5.</span> <span class="mf">6.</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">7.</span> <span class="mf">7.</span> <span class="mf">8.</span> <span class="mf">9.</span> <span class="mf">9.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">7.</span> <span class="mf">7.</span> <span class="mf">8.</span> <span class="mf">9.</span> <span class="mf">9.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">10.</span> <span class="mf">10.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">12.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">10.</span> <span class="mf">10.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">12.</span><span class="p">]]]</span> |
| |
| |
| <span class="p">[[[</span> <span class="mf">11.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">13.</span> <span class="mf">13.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">11.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">13.</span> <span class="mf">13.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">14.</span> <span class="mf">14.</span> <span class="mf">15.</span> <span class="mf">16.</span> <span class="mf">16.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">14.</span> <span class="mf">14.</span> <span class="mf">15.</span> <span class="mf">16.</span> <span class="mf">16.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">17.</span> <span class="mf">17.</span> <span class="mf">18.</span> <span class="mf">19.</span> <span class="mf">19.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">17.</span> <span class="mf">17.</span> <span class="mf">18.</span> <span class="mf">19.</span> <span class="mf">19.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">20.</span> <span class="mf">20.</span> <span class="mf">21.</span> <span class="mf">22.</span> <span class="mf">22.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">20.</span> <span class="mf">20.</span> <span class="mf">21.</span> <span class="mf">22.</span> <span class="mf">22.</span><span class="p">]]]]</span> |
| |
| <span class="n">pad</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">mode</span><span class="o">=</span><span class="s2">"constant"</span><span class="p">,</span> <span class="n">constant_value</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">pad_width</span><span class="o">=</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</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">1</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="o">=</span> |
| |
| <span class="p">[[[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">3.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">4.</span> <span class="mf">5.</span> <span class="mf">6.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">7.</span> <span class="mf">8.</span> <span class="mf">9.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">10.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]]</span> |
| |
| |
| <span class="p">[[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">11.</span> <span class="mf">12.</span> <span class="mf">13.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">14.</span> <span class="mf">15.</span> <span class="mf">16.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">17.</span> <span class="mf">18.</span> <span class="mf">19.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">20.</span> <span class="mf">21.</span> <span class="mf">22.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/pad.cc:L765</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – An n-dimensional input array.</p></li> |
| <li><p><strong>mode</strong> (<em>{'constant'</em><em>, </em><em>'edge'</em><em>, </em><em>'reflect'}</em><em>, </em><em>required</em>) – Padding type to use. “constant” pads with <cite>constant_value</cite> “edge” pads using the edge values of the input array “reflect” pads by reflecting values with respect to the edges.</p></li> |
| <li><p><strong>pad_width</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – Widths of the padding regions applied to the edges of each axis. It is a tuple of integer padding widths for each axis of the format <code class="docutils literal notranslate"><span class="pre">(before_1,</span> <span class="pre">after_1,</span> <span class="pre">...</span> <span class="pre">,</span> <span class="pre">before_N,</span> <span class="pre">after_N)</span></code>. It should be of length <code class="docutils literal notranslate"><span class="pre">2*N</span></code> where <code class="docutils literal notranslate"><span class="pre">N</span></code> is the number of dimensions of the array.This is equivalent to pad_width in numpy.pad, but flattened.</p></li> |
| <li><p><strong>constant_value</strong> (<em>double</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The value used for padding when <cite>mode</cite> is “constant”.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.pick"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">pick</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">index=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">mode=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.pick" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Picks elements from an input array according to the input indices along the given axis.</p> |
| <p>Given an input array of shape <code class="docutils literal notranslate"><span class="pre">(d0,</span> <span class="pre">d1)</span></code> and indices of shape <code class="docutils literal notranslate"><span class="pre">(i0,)</span></code>, the result will be |
| an output array of shape <code class="docutils literal notranslate"><span class="pre">(i0,)</span></code> with:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">output</span><span class="p">[</span><span class="n">i</span><span class="p">]</span> <span class="o">=</span> <span class="nb">input</span><span class="p">[</span><span class="n">i</span><span class="p">,</span> <span class="n">indices</span><span class="p">[</span><span class="n">i</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>By default, if any index mentioned is too large, it is replaced by the index that addresses |
| the last element along an axis (the <cite>clip</cite> mode).</p> |
| <p>This function supports n-dimensional input and (n-1)-dimensional indices arrays.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">picks</span> <span class="n">elements</span> <span class="k">with</span> <span class="n">specified</span> <span class="n">indices</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">0</span> |
| <span class="n">pick</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="o">=</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">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">picks</span> <span class="n">elements</span> <span class="k">with</span> <span class="n">specified</span> <span class="n">indices</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> |
| <span class="n">pick</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="o">=</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">0</span><span class="p">],</span> <span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">picks</span> <span class="n">elements</span> <span class="k">with</span> <span class="n">specified</span> <span class="n">indices</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> <span class="n">using</span> <span class="s1">'wrap'</span> <span class="n">mode</span> |
| <span class="o">//</span> <span class="n">to</span> <span class="n">place</span> <span class="n">indicies</span> <span class="n">that</span> <span class="n">would</span> <span class="n">normally</span> <span class="n">be</span> <span class="n">out</span> <span class="n">of</span> <span class="n">bounds</span> |
| <span class="n">pick</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="o">=</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">2</span><span class="p">],</span> <span class="mi">1</span><span class="p">,</span> <span class="n">mode</span><span class="o">=</span><span class="s1">'wrap'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">picks</span> <span class="n">elements</span> <span class="k">with</span> <span class="n">specified</span> <span class="n">indices</span> <span class="n">along</span> <span class="n">axis</span> <span class="mi">1</span> <span class="ow">and</span> <span class="n">dims</span> <span class="n">are</span> <span class="n">maintained</span> |
| <span class="n">pick</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="n">keepdims</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L150</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array</p></li> |
| <li><p><strong>index</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The index array</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='-1'</em>) – int or None. The axis to picking the elements. Negative values means indexing from right to left. If is <cite>None</cite>, the elements in the index w.r.t the flattened input will be picked.</p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true, the axis where we pick the elements is left in the result as dimension with size one.</p></li> |
| <li><p><strong>mode</strong> (<em>{'clip'</em><em>, </em><em>'wrap'}</em><em>,</em><em>optional</em><em>, </em><em>default='clip'</em>) – Specify how out-of-bound indices behave. Default is “clip”. “clip” means clip to the range. So, if all indices mentioned are too large, they are replaced by the index that addresses the last element along an axis. “wrap” means to wrap around.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.preloaded_multi_mp_sgd_mom_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">preloaded_multi_mp_sgd_mom_update</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.preloaded_multi_mp_sgd_mom_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Momentum update function for multi-precision Stochastic Gradient Descent (SGD) optimizer.</p> |
| <p>Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below:</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-1} - \alpha * \nabla J(W_{t-1})\\ |
| W_t = W_{t-1} + v_t\end{split}\]</div> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">v</span> <span class="o">=</span> <span class="n">momentum</span> <span class="o">*</span> <span class="n">v</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="n">gradient</span> |
| <span class="n">weight</span> <span class="o">+=</span> <span class="n">v</span> |
| </pre></div> |
| </div> |
| <p>Where the parameter <code class="docutils literal notranslate"><span class="pre">momentum</span></code> is the decay rate of momentum estimates at each epoch.</p> |
| <p>Defined in src/operator/contrib/preloaded_multi_sgd.cc:L199</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Weights, gradients, momentums, learning rates and weight decays</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The decay rate of momentum estimates at each epoch.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>num_weights</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of updated weights.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.preloaded_multi_mp_sgd_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">preloaded_multi_mp_sgd_update</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.preloaded_multi_mp_sgd_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for multi-precision Stochastic Gradient Descent (SDG) optimizer.</p> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">weight</span> <span class="o">=</span> <span class="n">weight</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="p">(</span><span class="n">gradient</span> <span class="o">+</span> <span class="n">wd</span> <span class="o">*</span> <span class="n">weight</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/contrib/preloaded_multi_sgd.cc:L139</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Weights, gradients, learning rates and weight decays</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>num_weights</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of updated weights.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.preloaded_multi_sgd_mom_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">preloaded_multi_sgd_mom_update</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.preloaded_multi_sgd_mom_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Momentum update function for Stochastic Gradient Descent (SGD) optimizer.</p> |
| <p>Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below:</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-1} - \alpha * \nabla J(W_{t-1})\\ |
| W_t = W_{t-1} + v_t\end{split}\]</div> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">v</span> <span class="o">=</span> <span class="n">momentum</span> <span class="o">*</span> <span class="n">v</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="n">gradient</span> |
| <span class="n">weight</span> <span class="o">+=</span> <span class="n">v</span> |
| </pre></div> |
| </div> |
| <p>Where the parameter <code class="docutils literal notranslate"><span class="pre">momentum</span></code> is the decay rate of momentum estimates at each epoch.</p> |
| <p>Defined in src/operator/contrib/preloaded_multi_sgd.cc:L90</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Weights, gradients, momentum, learning rates and weight decays</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The decay rate of momentum estimates at each epoch.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>num_weights</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of updated weights.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.preloaded_multi_sgd_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">preloaded_multi_sgd_update</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.preloaded_multi_sgd_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for Stochastic Gradient Descent (SDG) optimizer.</p> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">weight</span> <span class="o">=</span> <span class="n">weight</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="p">(</span><span class="n">gradient</span> <span class="o">+</span> <span class="n">wd</span> <span class="o">*</span> <span class="n">weight</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/contrib/preloaded_multi_sgd.cc:L41</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Weights, gradients, learning rates and weight decays</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>num_weights</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of updated weights.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.prod"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">prod</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.prod" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the product of array elements over given axes.</p> |
| <p>Defined in src/operator/tensor/./broadcast_reduce_op.h:L30</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.radians"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">radians</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.radians" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Converts each element of the input array from degrees to radians.</p> |
| <div class="math notranslate nohighlight"> |
| \[radians([0, 90, 180, 270, 360]) = [0, \pi/2, \pi, 3\pi/2, 2\pi]\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">radians</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>radians(default) = default</p></li> |
| <li><p>radians(row_sparse) = row_sparse</p></li> |
| <li><p>radians(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L351</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_exponential"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_exponential</code><span class="sig-paren">(</span><em class="sig-param">lam=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_exponential" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from an exponential distribution.</p> |
| <p>Samples are distributed according to an exponential distribution parametrized by <em>lambda</em> (rate).</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">exponential</span><span class="p">(</span><span class="n">lam</span><span class="o">=</span><span class="mi">4</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">0.0097189</span> <span class="p">,</span> <span class="mf">0.08999364</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.04146638</span><span class="p">,</span> <span class="mf">0.31715935</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L136</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lam</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Lambda parameter (rate) of the exponential distribution.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_gamma"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_gamma</code><span class="sig-paren">(</span><em class="sig-param">alpha=_Null</em>, <em class="sig-param">beta=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_gamma" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from a gamma distribution.</p> |
| <p>Samples are distributed according to a gamma distribution parametrized by <em>alpha</em> (shape) and <em>beta</em> (scale).</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">gamma</span><span class="p">(</span><span class="n">alpha</span><span class="o">=</span><span class="mi">9</span><span class="p">,</span> <span class="n">beta</span><span class="o">=</span><span class="mf">0.5</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">7.10486984</span><span class="p">,</span> <span class="mf">3.37695289</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.91697288</span><span class="p">,</span> <span class="mf">3.65933681</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L124</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>alpha</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Alpha parameter (shape) of the gamma distribution.</p></li> |
| <li><p><strong>beta</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Beta parameter (scale) of the gamma distribution.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_generalized_negative_binomial"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_generalized_negative_binomial</code><span class="sig-paren">(</span><em class="sig-param">mu=_Null</em>, <em class="sig-param">alpha=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_generalized_negative_binomial" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from a generalized negative binomial distribution.</p> |
| <p>Samples are distributed according to a generalized negative binomial distribution parametrized by |
| <em>mu</em> (mean) and <em>alpha</em> (dispersion). <em>alpha</em> is defined as <em>1/k</em> where <em>k</em> is the failure limit of the |
| number of unsuccessful experiments (generalized to real numbers). |
| Samples will always be returned as a floating point data type.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">generalized_negative_binomial</span><span class="p">(</span><span class="n">mu</span><span class="o">=</span><span class="mf">2.0</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">0.3</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L178</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>mu</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Mean of the negative binomial distribution.</p></li> |
| <li><p><strong>alpha</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Alpha (dispersion) parameter of the negative binomial distribution.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_negative_binomial"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_negative_binomial</code><span class="sig-paren">(</span><em class="sig-param">k=_Null</em>, <em class="sig-param">p=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_negative_binomial" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from a negative binomial distribution.</p> |
| <p>Samples are distributed according to a negative binomial distribution parametrized by |
| <em>k</em> (limit of unsuccessful experiments) and <em>p</em> (failure probability in each experiment). |
| Samples will always be returned as a floating point data type.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">negative_binomial</span><span class="p">(</span><span class="n">k</span><span class="o">=</span><span class="mi">3</span><span class="p">,</span> <span class="n">p</span><span class="o">=</span><span class="mf">0.4</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L163</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>k</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Limit of unsuccessful experiments.</p></li> |
| <li><p><strong>p</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Failure probability in each experiment.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_normal"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_normal</code><span class="sig-paren">(</span><em class="sig-param">loc=_Null</em>, <em class="sig-param">scale=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_normal" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from a normal (Gaussian) distribution.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The existing alias <code class="docutils literal notranslate"><span class="pre">normal</span></code> is deprecated.</p> |
| </div> |
| <p>Samples are distributed according to a normal distribution parametrized by <em>loc</em> (mean) and <em>scale</em> |
| (standard deviation).</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">normal</span><span class="p">(</span><span class="n">loc</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">scale</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">1.89171135</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.16881478</span><span class="p">],</span> |
| <span class="p">[</span><span class="o">-</span><span class="mf">1.23474145</span><span class="p">,</span> <span class="mf">1.55807114</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L112</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>loc</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Mean of the distribution.</p></li> |
| <li><p><strong>scale</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Standard deviation of the distribution.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_pdf_dirichlet"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_pdf_dirichlet</code><span class="sig-paren">(</span><em class="sig-param">sample=None</em>, <em class="sig-param">alpha=None</em>, <em class="sig-param">is_log=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_pdf_dirichlet" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the value of the PDF of <em>sample</em> of |
| Dirichlet distributions with parameter <em>alpha</em>.</p> |
| <p>The shape of <em>alpha</em> must match the leftmost subshape of <em>sample</em>. That is, <em>sample</em> |
| can have the same shape as <em>alpha</em>, in which case the output contains one density per |
| distribution, or <em>sample</em> can be a tensor of tensors with that shape, in which case |
| the output is a tensor of densities such that the densities at index <em>i</em> in the output |
| are given by the samples at index <em>i</em> in <em>sample</em> parameterized by the value of <em>alpha</em> |
| at index <em>i</em>.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">random_pdf_dirichlet</span><span class="p">(</span><span class="n">sample</span><span class="o">=</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">3</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="n">alpha</span><span class="o">=</span><span class="p">[</span><span class="mf">2.5</span><span class="p">,</span> <span class="mf">2.5</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[</span><span class="mf">38.413498</span><span class="p">,</span> <span class="mf">199.60245</span><span class="p">,</span> <span class="mf">564.56085</span><span class="p">]</span> |
| |
| <span class="n">sample</span> <span class="o">=</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="p">[</span><span class="mi">10</span><span class="p">,</span> <span class="mi">20</span><span class="p">,</span> <span class="mi">30</span><span class="p">],</span> <span class="p">[</span><span class="mi">100</span><span class="p">,</span> <span class="mi">200</span><span class="p">,</span> <span class="mi">300</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.3</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.01</span><span class="p">,</span> <span class="mf">0.02</span><span class="p">,</span> <span class="mf">0.03</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.001</span><span class="p">,</span> <span class="mf">0.002</span><span class="p">,</span> <span class="mf">0.003</span><span class="p">]]]</span> |
| |
| <span class="n">random_pdf_dirichlet</span><span class="p">(</span><span class="n">sample</span><span class="o">=</span><span class="n">sample</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="p">[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.4</span><span class="p">,</span> <span class="mf">0.9</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">2.3257459e-02</span><span class="p">,</span> <span class="mf">5.8420084e-04</span><span class="p">,</span> <span class="mf">1.4674458e-05</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">9.2589635e-01</span><span class="p">,</span> <span class="mf">3.6860607e+01</span><span class="p">,</span> <span class="mf">1.4674468e+03</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/pdf_op.cc:L315</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>sample</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Samples from the distributions.</p></li> |
| <li><p><strong>alpha</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Concentration parameters of the distributions.</p></li> |
| <li><p><strong>is_log</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set, compute the density of the log-probability instead of the probability.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_pdf_exponential"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_pdf_exponential</code><span class="sig-paren">(</span><em class="sig-param">sample=None</em>, <em class="sig-param">lam=None</em>, <em class="sig-param">is_log=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_pdf_exponential" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the value of the PDF of <em>sample</em> of |
| exponential distributions with parameters <em>lam</em> (rate).</p> |
| <p>The shape of <em>lam</em> must match the leftmost subshape of <em>sample</em>. That is, <em>sample</em> |
| can have the same shape as <em>lam</em>, in which case the output contains one density per |
| distribution, or <em>sample</em> can be a tensor of tensors with that shape, in which case |
| the output is a tensor of densities such that the densities at index <em>i</em> in the output |
| are given by the samples at index <em>i</em> in <em>sample</em> parameterized by the value of <em>lam</em> |
| at index <em>i</em>.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">random_pdf_exponential</span><span class="p">(</span><span class="n">sample</span><span class="o">=</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="n">lam</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.36787945</span><span class="p">,</span> <span class="mf">0.13533528</span><span class="p">,</span> <span class="mf">0.04978707</span><span class="p">]]</span> |
| |
| <span class="n">sample</span> <span class="o">=</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="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="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="n">random_pdf_exponential</span><span class="p">(</span><span class="n">sample</span><span class="o">=</span><span class="n">sample</span><span class="p">,</span> <span class="n">lam</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">,</span><span class="mf">0.5</span><span class="p">,</span><span class="mf">0.25</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.36787945</span><span class="p">,</span> <span class="mf">0.13533528</span><span class="p">,</span> <span class="mf">0.04978707</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.30326533</span><span class="p">,</span> <span class="mf">0.18393973</span><span class="p">,</span> <span class="mf">0.11156508</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.1947002</span><span class="p">,</span> <span class="mf">0.15163267</span><span class="p">,</span> <span class="mf">0.11809164</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/pdf_op.cc:L304</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>sample</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Samples from the distributions.</p></li> |
| <li><p><strong>lam</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lambda (rate) parameters of the distributions.</p></li> |
| <li><p><strong>is_log</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set, compute the density of the log-probability instead of the probability.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_pdf_gamma"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_pdf_gamma</code><span class="sig-paren">(</span><em class="sig-param">sample=None</em>, <em class="sig-param">alpha=None</em>, <em class="sig-param">beta=None</em>, <em class="sig-param">is_log=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_pdf_gamma" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the value of the PDF of <em>sample</em> of |
| gamma distributions with parameters <em>alpha</em> (shape) and <em>beta</em> (rate).</p> |
| <p><em>alpha</em> and <em>beta</em> must have the same shape, which must match the leftmost subshape |
| of <em>sample</em>. That is, <em>sample</em> can have the same shape as <em>alpha</em> and <em>beta</em>, in which |
| case the output contains one density per distribution, or <em>sample</em> can be a tensor |
| of tensors with that shape, in which case the output is a tensor of densities such that |
| the densities at index <em>i</em> in the output are given by the samples at index <em>i</em> in <em>sample</em> |
| parameterized by the values of <em>alpha</em> and <em>beta</em> at index <em>i</em>.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">random_pdf_gamma</span><span class="p">(</span><span class="n">sample</span><span class="o">=</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="n">alpha</span><span class="o">=</span><span class="p">[</span><span class="mi">5</span><span class="p">],</span> <span class="n">beta</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.01532831</span><span class="p">,</span> <span class="mf">0.09022352</span><span class="p">,</span> <span class="mf">0.16803136</span><span class="p">,</span> <span class="mf">0.19536681</span><span class="p">,</span> <span class="mf">0.17546739</span><span class="p">]]</span> |
| |
| <span class="n">sample</span> <span class="o">=</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">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="mi">6</span><span class="p">],</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="mi">6</span><span class="p">,</span> <span class="mi">7</span><span class="p">]]</span> |
| |
| <span class="n">random_pdf_gamma</span><span class="p">(</span><span class="n">sample</span><span class="o">=</span><span class="n">sample</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="p">[</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">,</span><span class="mi">7</span><span class="p">],</span> <span class="n">beta</span><span class="o">=</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">1</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.01532831</span><span class="p">,</span> <span class="mf">0.09022352</span><span class="p">,</span> <span class="mf">0.16803136</span><span class="p">,</span> <span class="mf">0.19536681</span><span class="p">,</span> <span class="mf">0.17546739</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.03608941</span><span class="p">,</span> <span class="mf">0.10081882</span><span class="p">,</span> <span class="mf">0.15629345</span><span class="p">,</span> <span class="mf">0.17546739</span><span class="p">,</span> <span class="mf">0.16062315</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.05040941</span><span class="p">,</span> <span class="mf">0.10419563</span><span class="p">,</span> <span class="mf">0.14622283</span><span class="p">,</span> <span class="mf">0.16062315</span><span class="p">,</span> <span class="mf">0.14900276</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/pdf_op.cc:L302</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>sample</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Samples from the distributions.</p></li> |
| <li><p><strong>alpha</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Alpha (shape) parameters of the distributions.</p></li> |
| <li><p><strong>is_log</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set, compute the density of the log-probability instead of the probability.</p></li> |
| <li><p><strong>beta</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Beta (scale) parameters of the distributions.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_pdf_generalized_negative_binomial"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_pdf_generalized_negative_binomial</code><span class="sig-paren">(</span><em class="sig-param">sample=None</em>, <em class="sig-param">mu=None</em>, <em class="sig-param">alpha=None</em>, <em class="sig-param">is_log=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_pdf_generalized_negative_binomial" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the value of the PDF of <em>sample</em> of |
| generalized negative binomial distributions with parameters <em>mu</em> (mean) |
| and <em>alpha</em> (dispersion). This can be understood as a reparameterization of |
| the negative binomial, where <em>k</em> = <em>1 / alpha</em> and <em>p</em> = <em>1 / (mu * alpha + 1)</em>.</p> |
| <p><em>mu</em> and <em>alpha</em> must have the same shape, which must match the leftmost subshape |
| of <em>sample</em>. That is, <em>sample</em> can have the same shape as <em>mu</em> and <em>alpha</em>, in which |
| case the output contains one density per distribution, or <em>sample</em> can be a tensor |
| of tensors with that shape, in which case the output is a tensor of densities such that |
| the densities at index <em>i</em> in the output are given by the samples at index <em>i</em> in <em>sample</em> |
| parameterized by the values of <em>mu</em> and <em>alpha</em> at index <em>i</em>.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">random_pdf_generalized_negative_binomial</span><span class="p">(</span><span class="n">sample</span><span class="o">=</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="n">alpha</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">mu</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.25</span><span class="p">,</span> <span class="mf">0.125</span><span class="p">,</span> <span class="mf">0.0625</span><span class="p">,</span> <span class="mf">0.03125</span><span class="p">]]</span> |
| |
| <span class="n">sample</span> <span class="o">=</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="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="n">random_pdf_generalized_negative_binomial</span><span class="p">(</span><span class="n">sample</span><span class="o">=</span><span class="n">sample</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mf">0.6666</span><span class="p">],</span> <span class="n">mu</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.25</span><span class="p">,</span> <span class="mf">0.125</span><span class="p">,</span> <span class="mf">0.0625</span><span class="p">,</span> <span class="mf">0.03125</span> <span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.26517063</span><span class="p">,</span> <span class="mf">0.16573331</span><span class="p">,</span> <span class="mf">0.09667706</span><span class="p">,</span> <span class="mf">0.05437994</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/pdf_op.cc:L313</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>sample</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Samples from the distributions.</p></li> |
| <li><p><strong>mu</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Means of the distributions.</p></li> |
| <li><p><strong>is_log</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set, compute the density of the log-probability instead of the probability.</p></li> |
| <li><p><strong>alpha</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Alpha (dispersion) parameters of the distributions.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_pdf_negative_binomial"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_pdf_negative_binomial</code><span class="sig-paren">(</span><em class="sig-param">sample=None</em>, <em class="sig-param">k=None</em>, <em class="sig-param">p=None</em>, <em class="sig-param">is_log=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_pdf_negative_binomial" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the value of the PDF of samples of |
| negative binomial distributions with parameters <em>k</em> (failure limit) and <em>p</em> (failure probability).</p> |
| <p><em>k</em> and <em>p</em> must have the same shape, which must match the leftmost subshape |
| of <em>sample</em>. That is, <em>sample</em> can have the same shape as <em>k</em> and <em>p</em>, in which |
| case the output contains one density per distribution, or <em>sample</em> can be a tensor |
| of tensors with that shape, in which case the output is a tensor of densities such that |
| the densities at index <em>i</em> in the output are given by the samples at index <em>i</em> in <em>sample</em> |
| parameterized by the values of <em>k</em> and <em>p</em> at index <em>i</em>.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">random_pdf_negative_binomial</span><span class="p">(</span><span class="n">sample</span><span class="o">=</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="n">k</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">p</span><span class="o">=</span><span class="n">a</span><span class="p">[</span><span class="mf">0.5</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.25</span><span class="p">,</span> <span class="mf">0.125</span><span class="p">,</span> <span class="mf">0.0625</span><span class="p">,</span> <span class="mf">0.03125</span><span class="p">]]</span> |
| |
| <span class="c1"># Note that k may be real-valued</span> |
| <span class="n">sample</span> <span class="o">=</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="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="n">random_pdf_negative_binomial</span><span class="p">(</span><span class="n">sample</span><span class="o">=</span><span class="n">sample</span><span class="p">,</span> <span class="n">k</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">],</span> <span class="n">p</span><span class="o">=</span><span class="p">[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.25</span><span class="p">,</span> <span class="mf">0.125</span><span class="p">,</span> <span class="mf">0.0625</span><span class="p">,</span> <span class="mf">0.03125</span> <span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.26516506</span><span class="p">,</span> <span class="mf">0.16572815</span><span class="p">,</span> <span class="mf">0.09667476</span><span class="p">,</span> <span class="mf">0.05437956</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/pdf_op.cc:L309</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>sample</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Samples from the distributions.</p></li> |
| <li><p><strong>k</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Limits of unsuccessful experiments.</p></li> |
| <li><p><strong>is_log</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set, compute the density of the log-probability instead of the probability.</p></li> |
| <li><p><strong>p</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Failure probabilities in each experiment.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_pdf_normal"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_pdf_normal</code><span class="sig-paren">(</span><em class="sig-param">sample=None</em>, <em class="sig-param">mu=None</em>, <em class="sig-param">sigma=None</em>, <em class="sig-param">is_log=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_pdf_normal" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the value of the PDF of <em>sample</em> of |
| normal distributions with parameters <em>mu</em> (mean) and <em>sigma</em> (standard deviation).</p> |
| <p><em>mu</em> and <em>sigma</em> must have the same shape, which must match the leftmost subshape |
| of <em>sample</em>. That is, <em>sample</em> can have the same shape as <em>mu</em> and <em>sigma</em>, in which |
| case the output contains one density per distribution, or <em>sample</em> can be a tensor |
| of tensors with that shape, in which case the output is a tensor of densities such that |
| the densities at index <em>i</em> in the output are given by the samples at index <em>i</em> in <em>sample</em> |
| parameterized by the values of <em>mu</em> and <em>sigma</em> at index <em>i</em>.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">sample</span> <span class="o">=</span> <span class="p">[[</span><span class="o">-</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="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="n">random_pdf_normal</span><span class="p">(</span><span class="n">sample</span><span class="o">=</span><span class="n">sample</span><span class="p">,</span> <span class="n">mu</span><span class="o">=</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">sigma</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.05399097</span><span class="p">,</span> <span class="mf">0.24197073</span><span class="p">,</span> <span class="mf">0.3989423</span><span class="p">,</span> <span class="mf">0.24197073</span><span class="p">,</span> <span class="mf">0.05399097</span><span class="p">]]</span> |
| |
| <span class="n">random_pdf_normal</span><span class="p">(</span><span class="n">sample</span><span class="o">=</span><span class="n">sample</span><span class="o">*</span><span class="mi">2</span><span class="p">,</span> <span class="n">mu</span><span class="o">=</span><span class="p">[</span><span class="mi">0</span><span class="p">,</span><span class="mi">0</span><span class="p">],</span> <span class="n">sigma</span><span class="o">=</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="o">=</span> |
| <span class="p">[[</span><span class="mf">0.05399097</span><span class="p">,</span> <span class="mf">0.24197073</span><span class="p">,</span> <span class="mf">0.3989423</span><span class="p">,</span> <span class="mf">0.24197073</span><span class="p">,</span> <span class="mf">0.05399097</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.12098537</span><span class="p">,</span> <span class="mf">0.17603266</span><span class="p">,</span> <span class="mf">0.19947115</span><span class="p">,</span> <span class="mf">0.17603266</span><span class="p">,</span> <span class="mf">0.12098537</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/pdf_op.cc:L299</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>sample</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Samples from the distributions.</p></li> |
| <li><p><strong>mu</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Means of the distributions.</p></li> |
| <li><p><strong>is_log</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set, compute the density of the log-probability instead of the probability.</p></li> |
| <li><p><strong>sigma</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Standard deviations of the distributions.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_pdf_poisson"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_pdf_poisson</code><span class="sig-paren">(</span><em class="sig-param">sample=None</em>, <em class="sig-param">lam=None</em>, <em class="sig-param">is_log=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_pdf_poisson" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the value of the PDF of <em>sample</em> of |
| Poisson distributions with parameters <em>lam</em> (rate).</p> |
| <p>The shape of <em>lam</em> must match the leftmost subshape of <em>sample</em>. That is, <em>sample</em> |
| can have the same shape as <em>lam</em>, in which case the output contains one density per |
| distribution, or <em>sample</em> can be a tensor of tensors with that shape, in which case |
| the output is a tensor of densities such that the densities at index <em>i</em> in the output |
| are given by the samples at index <em>i</em> in <em>sample</em> parameterized by the value of <em>lam</em> |
| at index <em>i</em>.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">random_pdf_poisson</span><span class="p">(</span><span class="n">sample</span><span class="o">=</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="n">lam</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span> <span class="o">=</span> |
| <span class="p">[[</span><span class="mf">0.36787945</span><span class="p">,</span> <span class="mf">0.36787945</span><span class="p">,</span> <span class="mf">0.18393973</span><span class="p">,</span> <span class="mf">0.06131324</span><span class="p">]]</span> |
| |
| <span class="n">sample</span> <span class="o">=</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="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="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="n">random_pdf_poisson</span><span class="p">(</span><span class="n">sample</span><span class="o">=</span><span class="n">sample</span><span class="p">,</span> <span class="n">lam</span><span class="o">=</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="o">=</span> |
| <span class="p">[[</span><span class="mf">0.36787945</span><span class="p">,</span> <span class="mf">0.36787945</span><span class="p">,</span> <span class="mf">0.18393973</span><span class="p">,</span> <span class="mf">0.06131324</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.13533528</span><span class="p">,</span> <span class="mf">0.27067056</span><span class="p">,</span> <span class="mf">0.27067056</span><span class="p">,</span> <span class="mf">0.18044704</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.04978707</span><span class="p">,</span> <span class="mf">0.14936121</span><span class="p">,</span> <span class="mf">0.22404182</span><span class="p">,</span> <span class="mf">0.22404182</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/pdf_op.cc:L306</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>sample</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Samples from the distributions.</p></li> |
| <li><p><strong>lam</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lambda (rate) parameters of the distributions.</p></li> |
| <li><p><strong>is_log</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set, compute the density of the log-probability instead of the probability.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_pdf_uniform"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_pdf_uniform</code><span class="sig-paren">(</span><em class="sig-param">sample=None</em>, <em class="sig-param">low=None</em>, <em class="sig-param">high=None</em>, <em class="sig-param">is_log=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_pdf_uniform" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the value of the PDF of <em>sample</em> of |
| uniform distributions on the intervals given by <em>[low,high)</em>.</p> |
| <p><em>low</em> and <em>high</em> must have the same shape, which must match the leftmost subshape |
| of <em>sample</em>. That is, <em>sample</em> can have the same shape as <em>low</em> and <em>high</em>, in which |
| case the output contains one density per distribution, or <em>sample</em> can be a tensor |
| of tensors with that shape, in which case the output is a tensor of densities such that |
| the densities at index <em>i</em> in the output are given by the samples at index <em>i</em> in <em>sample</em> |
| parameterized by the values of <em>low</em> and <em>high</em> at index <em>i</em>.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">random_pdf_uniform</span><span class="p">(</span><span class="n">sample</span><span class="o">=</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="n">low</span><span class="o">=</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">high</span><span class="o">=</span><span class="p">[</span><span class="mi">10</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">]</span> |
| |
| <span class="n">sample</span> <span class="o">=</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="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="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="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="n">low</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]</span> |
| <span class="n">high</span> <span class="o">=</span> <span class="p">[[</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">10</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">15</span><span class="p">,</span> <span class="mi">20</span><span class="p">]]</span> |
| <span class="n">random_pdf_uniform</span><span class="p">(</span><span class="n">sample</span><span class="o">=</span><span class="n">sample</span><span class="p">,</span> <span class="n">low</span><span class="o">=</span><span class="n">low</span><span class="p">,</span> <span class="n">high</span><span class="o">=</span><span class="n">high</span><span class="p">)</span> <span class="o">=</span> |
| <span class="p">[[[</span><span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.2</span> <span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.1</span> <span class="p">]],</span> |
| <span class="p">[[</span><span class="mf">0.06667</span><span class="p">,</span> <span class="mf">0.06667</span><span class="p">,</span> <span class="mf">0.06667</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.05</span><span class="p">,</span> <span class="mf">0.05</span><span class="p">,</span> <span class="mf">0.05</span> <span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/pdf_op.cc:L297</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>sample</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Samples from the distributions.</p></li> |
| <li><p><strong>low</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lower bounds of the distributions.</p></li> |
| <li><p><strong>is_log</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If set, compute the density of the log-probability instead of the probability.</p></li> |
| <li><p><strong>high</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Upper bounds of the distributions.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_poisson"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_poisson</code><span class="sig-paren">(</span><em class="sig-param">lam=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_poisson" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from a Poisson distribution.</p> |
| <p>Samples are distributed according to a Poisson distribution parametrized by <em>lambda</em> (rate). |
| Samples will always be returned as a floating point data type.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">poisson</span><span class="p">(</span><span class="n">lam</span><span class="o">=</span><span class="mi">4</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L149</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lam</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Lambda parameter (rate) of the Poisson distribution.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_randint"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_randint</code><span class="sig-paren">(</span><em class="sig-param">low=_Null</em>, <em class="sig-param">high=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_randint" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from a discrete uniform distribution.</p> |
| <p>Samples are uniformly distributed over the half-open interval <em>[low, high)</em> |
| (includes <em>low</em>, but excludes <em>high</em>).</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">randint</span><span class="p">(</span><span class="n">low</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">high</span><span class="o">=</span><span class="mi">5</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">1</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L193</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>low</strong> (<em>long</em><em>, </em><em>required</em>) – Lower bound of the distribution.</p></li> |
| <li><p><strong>high</strong> (<em>long</em><em>, </em><em>required</em>) – Upper bound of the distribution.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to int32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.random_uniform"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">random_uniform</code><span class="sig-paren">(</span><em class="sig-param">low=_Null</em>, <em class="sig-param">high=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.random_uniform" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from a uniform distribution.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The existing alias <code class="docutils literal notranslate"><span class="pre">uniform</span></code> is deprecated.</p> |
| </div> |
| <p>Samples are uniformly distributed over the half-open interval <em>[low, high)</em> |
| (includes <em>low</em>, but excludes <em>high</em>).</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">uniform</span><span class="p">(</span><span class="n">low</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">high</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">0.60276335</span><span class="p">,</span> <span class="mf">0.85794562</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.54488319</span><span class="p">,</span> <span class="mf">0.84725171</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L95</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>low</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Lower bound of the distribution.</p></li> |
| <li><p><strong>high</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Upper bound of the distribution.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.ravel_multi_index"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">ravel_multi_index</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.ravel_multi_index" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Converts a batch of index arrays into an array of flat indices. The operator follows numpy conventions so a single multi index is given by a column of the input matrix. The leading dimension may be left unspecified by using -1 as placeholder.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">A</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">3</span><span class="p">,</span><span class="mi">6</span><span class="p">,</span><span class="mi">6</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="mi">1</span><span class="p">]]</span> |
| <span class="n">ravel</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">7</span><span class="p">,</span><span class="mi">6</span><span class="p">))</span> <span class="o">=</span> <span class="p">[</span><span class="mi">22</span><span class="p">,</span><span class="mi">41</span><span class="p">,</span><span class="mi">37</span><span class="p">]</span> |
| <span class="n">ravel</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="o">-</span><span class="mi">1</span><span class="p">,</span><span class="mi">6</span><span class="p">))</span> <span class="o">=</span> <span class="p">[</span><span class="mi">22</span><span class="p">,</span><span class="mi">41</span><span class="p">,</span><span class="mi">37</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/ravel.cc:L41</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Batch of multi-indices</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the array into which the multi-indices apply.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.rcbrt"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">rcbrt</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.rcbrt" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise inverse cube-root value of the input.</p> |
| <div class="math notranslate nohighlight"> |
| \[rcbrt(x) = 1/\sqrt[3]{x}\]</div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">rcbrt</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="o">-</span><span class="mi">125</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="o">-</span><span class="mf">0.2</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L323</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.reciprocal"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">reciprocal</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.reciprocal" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the reciprocal of the argument, element-wise.</p> |
| <p>Calculates 1/x.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">reciprocal</span><span class="p">([</span><span class="o">-</span><span class="mi">2</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="mf">1.6</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.33333334</span><span class="p">,</span> <span class="mf">0.625</span><span class="p">,</span> <span class="mf">5.0</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L43</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.relu"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">relu</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.relu" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes rectified linear activation.</p> |
| <div class="math notranslate nohighlight"> |
| \[max(features, 0)\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">relu</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>relu(default) = default</p></li> |
| <li><p>relu(row_sparse) = row_sparse</p></li> |
| <li><p>relu(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L85</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.repeat"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">repeat</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">repeats=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.repeat" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Repeats elements of an array. |
| By default, <code class="docutils literal notranslate"><span class="pre">repeat</span></code> flattens the input array into 1-D and then repeats the |
| elements:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="p">[</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]]</span> |
| <span class="n">repeat</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">repeats</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <dl class="simple"> |
| <dt>The parameter <code class="docutils literal notranslate"><span class="pre">axis</span></code> specifies the axis along which to perform repeat::</dt><dd><dl class="simple"> |
| <dt>repeat(x, repeats=2, axis=1) = [[ 1., 1., 2., 2.],</dt><dd><p>[ 3., 3., 4., 4.]]</p> |
| </dd> |
| <dt>repeat(x, repeats=2, axis=0) = [[ 1., 2.],</dt><dd><p>[ 1., 2.], |
| [ 3., 4.], |
| [ 3., 4.]]</p> |
| </dd> |
| <dt>repeat(x, repeats=2, axis=-1) = [[ 1., 1., 2., 2.],</dt><dd><p>[ 3., 3., 4., 4.]]</p> |
| </dd> |
| </dl> |
| </dd> |
| </dl> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L743</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data array</p></li> |
| <li><p><strong>repeats</strong> (<em>int</em><em>, </em><em>required</em>) – The number of repetitions for each element.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – The axis along which to repeat values. The negative numbers are interpreted counting from the backward. By default, use the flattened input array, and return a flat output array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.reset_arrays"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">reset_arrays</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.reset_arrays" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Set to zero multiple arrays</p> |
| <p>Defined in src/operator/contrib/reset_arrays.cc:L35</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – Arrays</p></li> |
| <li><p><strong>num_arrays</strong> (<em>int</em><em>, </em><em>required</em>) – number of input arrays.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.reshape"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">reshape</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">reverse=_Null</em>, <em class="sig-param">target_shape=_Null</em>, <em class="sig-param">keep_highest=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.reshape" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Reshapes the input array. |
| .. note:: <code class="docutils literal notranslate"><span class="pre">Reshape</span></code> is deprecated, use <code class="docutils literal notranslate"><span class="pre">reshape</span></code> |
| Given an array and a shape, this function returns a copy of the array in the new shape. |
| The shape is a tuple of integers such as (2,3,4). The size of the new shape should be same as the size of the input array. |
| Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">reshape</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="n">shape</span><span class="o">=</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="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span><span class="mi">4</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Some dimensions of the shape can take special values from the set {0, -1, -2, -3, -4}. The significance of each is explained below: |
| - <code class="docutils literal notranslate"><span class="pre">0</span></code> copy this dimension from the input to the output shape.</p> |
| <blockquote> |
| <div><p>Example:: |
| - input shape = (2,3,4), shape = (4,0,2), output shape = (4,3,2) |
| - input shape = (2,3,4), shape = (2,0,0), output shape = (2,3,4)</p> |
| </div></blockquote> |
| <ul class="simple"> |
| <li><p><code class="docutils literal notranslate"><span class="pre">-1</span></code> infers the dimension of the output shape by using the remainder of the input dimensions |
| keeping the size of the new array same as that of the input array. |
| At most one dimension of shape can be -1. |
| Example:: |
| - input shape = (2,3,4), shape = (6,1,-1), output shape = (6,1,4) |
| - input shape = (2,3,4), shape = (3,-1,8), output shape = (3,1,8) |
| - input shape = (2,3,4), shape=(-1,), output shape = (24,)</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">-2</span></code> copy all/remainder of the input dimensions to the output shape. |
| Example:: |
| - input shape = (2,3,4), shape = (-2,), output shape = (2,3,4) |
| - input shape = (2,3,4), shape = (2,-2), output shape = (2,3,4) |
| - input shape = (2,3,4), shape = (-2,1,1), output shape = (2,3,4,1,1)</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">-3</span></code> use the product of two consecutive dimensions of the input shape as the output dimension. |
| Example:: |
| - input shape = (2,3,4), shape = (-3,4), output shape = (6,4) |
| - input shape = (2,3,4,5), shape = (-3,-3), output shape = (6,20) |
| - input shape = (2,3,4), shape = (0,-3), output shape = (2,12) |
| - input shape = (2,3,4), shape = (-3,-2), output shape = (6,4)</p></li> |
| <li><p><code class="docutils literal notranslate"><span class="pre">-4</span></code> split one dimension of the input into two dimensions passed subsequent to -4 in shape (can contain -1). |
| Example:: |
| - input shape = (2,3,4), shape = (-4,1,2,-2), output shape =(1,2,3,4) |
| - input shape = (2,3,4), shape = (2,-4,-1,3,-2), output shape = (2,1,3,4)</p></li> |
| </ul> |
| <dl class="simple"> |
| <dt>If the argument <cite>reverse</cite> is set to 1, then the special values are inferred from right to left.</dt><dd><p>Example:: |
| - without reverse=1, for input shape = (10,5,4), shape = (-1,0), output shape would be (40,5) |
| - with reverse=1, output shape will be (50,4).</p> |
| </dd> |
| </dl> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L174</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to reshape.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – The target shape</p></li> |
| <li><p><strong>reverse</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true then the special values are inferred from right to left</p></li> |
| <li><p><strong>target_shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – (Deprecated! Use <code class="docutils literal notranslate"><span class="pre">shape</span></code> instead.) Target new shape. One and only one dim can be 0, in which case it will be inferred from the rest of dims</p></li> |
| <li><p><strong>keep_highest</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – (Deprecated! Use <code class="docutils literal notranslate"><span class="pre">shape</span></code> instead.) Whether keep the highest dim unchanged.If set to true, then the first dim in target_shape is ignored,and always fixed as input</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.reshape_like"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">reshape_like</code><span class="sig-paren">(</span><em class="sig-param">lhs=None</em>, <em class="sig-param">rhs=None</em>, <em class="sig-param">lhs_begin=_Null</em>, <em class="sig-param">lhs_end=_Null</em>, <em class="sig-param">rhs_begin=_Null</em>, <em class="sig-param">rhs_end=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.reshape_like" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Reshape some or all dimensions of <cite>lhs</cite> to have the same shape as some or all dimensions of <cite>rhs</cite>.</p> |
| <p>Returns a <strong>view</strong> of the <cite>lhs</cite> array with a new shape without altering any data.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="mi">6</span><span class="p">]</span> |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="o">-</span><span class="mi">4</span><span class="p">],</span> <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</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="n">reshape_like</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</span> <span class="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>More precise control over how dimensions are inherited is achieved by specifying slices over the <cite>lhs</cite> and <cite>rhs</cite> array dimensions. Only the sliced <cite>lhs</cite> dimensions are reshaped to the <cite>rhs</cite> sliced dimensions, with the non-sliced <cite>lhs</cite> dimensions staying the same.</p> |
| <blockquote> |
| <div><p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="o">-</span> <span class="n">lhs</span> <span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">30</span><span class="p">,</span><span class="mi">7</span><span class="p">),</span> <span class="n">rhs</span> <span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">15</span><span class="p">,</span><span class="mi">2</span><span class="p">,</span><span class="mi">4</span><span class="p">),</span> <span class="n">lhs_begin</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">lhs_end</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">rhs_begin</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">rhs_end</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">output</span> <span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">15</span><span class="p">,</span><span class="mi">2</span><span class="p">,</span><span class="mi">7</span><span class="p">)</span> |
| <span class="o">-</span> <span class="n">lhs</span> <span class="n">shape</span> <span class="o">=</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="n">rhs</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">15</span><span class="p">,</span><span class="mi">4</span><span class="p">),</span> <span class="n">lhs_begin</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">lhs_end</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">rhs_begin</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">rhs_end</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">output</span> <span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">15</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| </div></blockquote> |
| <p>Negative indices are supported, and <cite>None</cite> can be used for either <cite>lhs_end</cite> or <cite>rhs_end</cite> to indicate the end of the range.</p> |
| <blockquote> |
| <div><p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="o">-</span> <span class="n">lhs</span> <span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">30</span><span class="p">,</span> <span class="mi">12</span><span class="p">),</span> <span class="n">rhs</span> <span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">4</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">3</span><span class="p">),</span> <span class="n">lhs_begin</span><span class="o">=-</span><span class="mi">1</span><span class="p">,</span> <span class="n">lhs_end</span><span class="o">=</span><span class="kc">None</span><span class="p">,</span> <span class="n">rhs_begin</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">rhs_end</span><span class="o">=</span><span class="kc">None</span><span class="p">,</span> <span class="n">output</span> <span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">30</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">3</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L511</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – First input.</p></li> |
| <li><p><strong>rhs</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Second input.</p></li> |
| <li><p><strong>lhs_begin</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – Defaults to 0. The beginning index along which the lhs dimensions are to be reshaped. Supports negative indices.</p></li> |
| <li><p><strong>lhs_end</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – Defaults to None. The ending index along which the lhs dimensions are to be used for reshaping. Supports negative indices.</p></li> |
| <li><p><strong>rhs_begin</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – Defaults to 0. The beginning index along which the rhs dimensions are to be used for reshaping. Supports negative indices.</p></li> |
| <li><p><strong>rhs_end</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='None'</em>) – Defaults to None. The ending index along which the rhs dimensions are to be used for reshaping. Supports negative indices.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.reverse"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">reverse</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.reverse" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Reverses the order of elements along given axis while preserving array shape. |
| Note: reverse and flip are equivalent. We use reverse in the following examples. |
| Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">]]</span> |
| <span class="n">reverse</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">9.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]</span> |
| <span class="n">reverse</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">9.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L831</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data array</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – The axis which to reverse elements.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.rint"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">rint</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.rint" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise rounded value to the nearest integer of the input.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <ul class="simple"> |
| <li><p>For input <code class="docutils literal notranslate"><span class="pre">n.5</span></code> <code class="docutils literal notranslate"><span class="pre">rint</span></code> returns <code class="docutils literal notranslate"><span class="pre">n</span></code> while <code class="docutils literal notranslate"><span class="pre">round</span></code> returns <code class="docutils literal notranslate"><span class="pre">n+1</span></code>.</p></li> |
| <li><p>For input <code class="docutils literal notranslate"><span class="pre">-n.5</span></code> both <code class="docutils literal notranslate"><span class="pre">rint</span></code> and <code class="docutils literal notranslate"><span class="pre">round</span></code> returns <code class="docutils literal notranslate"><span class="pre">-n-1</span></code>.</p></li> |
| </ul> |
| </div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">rint</span><span class="p">([</span><span class="o">-</span><span class="mf">1.5</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.9</span><span class="p">,</span> <span class="mf">1.9</span><span class="p">,</span> <span class="mf">2.1</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="o">-</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">rint</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>rint(default) = default</p></li> |
| <li><p>rint(row_sparse) = row_sparse</p></li> |
| <li><p>rint(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L798</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.rmsprop_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">rmsprop_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">n=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">gamma1=_Null</em>, <em class="sig-param">epsilon=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">clip_weights=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.rmsprop_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for <cite>RMSProp</cite> optimizer.</p> |
| <p><cite>RMSprop</cite> is a variant of stochastic gradient descent where the gradients are |
| divided by a cache which grows with the sum of squares of recent gradients?</p> |
| <p><cite>RMSProp</cite> is similar to <cite>AdaGrad</cite>, a popular variant of <cite>SGD</cite> which adaptively |
| tunes the learning rate of each parameter. <cite>AdaGrad</cite> lowers the learning rate for |
| each parameter monotonically over the course of training. |
| While this is analytically motivated for convex optimizations, it may not be ideal |
| for non-convex problems. <cite>RMSProp</cite> deals with this heuristically by allowing the |
| learning rates to rebound as the denominator decays over time.</p> |
| <p>Define the Root Mean Square (RMS) error criterion of the gradient as |
| <span class="math notranslate nohighlight">\(RMS[g]_t = \sqrt{E[g^2]_t + \epsilon}\)</span>, where <span class="math notranslate nohighlight">\(g\)</span> represents |
| gradient and <span class="math notranslate nohighlight">\(E[g^2]_t\)</span> is the decaying average over past squared gradient.</p> |
| <p>The <span class="math notranslate nohighlight">\(E[g^2]_t\)</span> is given by:</p> |
| <div class="math notranslate nohighlight"> |
| \[E[g^2]_t = \gamma * E[g^2]_{t-1} + (1-\gamma) * g_t^2\]</div> |
| <p>The update step is</p> |
| <div class="math notranslate nohighlight"> |
| \[\theta_{t+1} = \theta_t - \frac{\eta}{RMS[g]_t} g_t\]</div> |
| <p>The RMSProp code follows the version in |
| <a class="reference external" href="http://www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slides_lec6.pdf">http://www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slides_lec6.pdf</a> |
| Tieleman & Hinton, 2012.</p> |
| <p>Hinton suggests the momentum term <span class="math notranslate nohighlight">\(\gamma\)</span> to be 0.9 and the learning rate |
| <span class="math notranslate nohighlight">\(\eta\)</span> to be 0.001.</p> |
| <p>Defined in src/operator/optimizer_op.cc:L796</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>n</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – n</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>gamma1</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.949999988</em>) – The decay rate of momentum estimates.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=9.99999994e-09</em>) – A small constant for numerical stability.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>clip_weights</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip weights to the range of [-clip_weights, clip_weights] If clip_weights <= 0, weight clipping is turned off. weights = max(min(weights, clip_weights), -clip_weights).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.rmspropalex_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">rmspropalex_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">n=None</em>, <em class="sig-param">g=None</em>, <em class="sig-param">delta=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">gamma1=_Null</em>, <em class="sig-param">gamma2=_Null</em>, <em class="sig-param">epsilon=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">clip_weights=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.rmspropalex_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for RMSPropAlex optimizer.</p> |
| <p><cite>RMSPropAlex</cite> is non-centered version of <cite>RMSProp</cite>.</p> |
| <p>Define <span class="math notranslate nohighlight">\(E[g^2]_t\)</span> is the decaying average over past squared gradient and |
| <span class="math notranslate nohighlight">\(E[g]_t\)</span> is the decaying average over past gradient.</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}E[g^2]_t = \gamma_1 * E[g^2]_{t-1} + (1 - \gamma_1) * g_t^2\\ |
| E[g]_t = \gamma_1 * E[g]_{t-1} + (1 - \gamma_1) * g_t\\ |
| \Delta_t = \gamma_2 * \Delta_{t-1} - \frac{\eta}{\sqrt{E[g^2]_t - E[g]_t^2 + \epsilon}} g_t\\\end{split}\]</div> |
| <p>The update step is</p> |
| <div class="math notranslate nohighlight"> |
| \[\theta_{t+1} = \theta_t + \Delta_t\]</div> |
| <p>The RMSPropAlex code follows the version in |
| <a class="reference external" href="http://arxiv.org/pdf/1308.0850v5.pdf">http://arxiv.org/pdf/1308.0850v5.pdf</a> Eq(38) - Eq(45) by Alex Graves, 2013.</p> |
| <p>Graves suggests the momentum term <span class="math notranslate nohighlight">\(\gamma_1\)</span> to be 0.95, <span class="math notranslate nohighlight">\(\gamma_2\)</span> |
| to be 0.9 and the learning rate <span class="math notranslate nohighlight">\(\eta\)</span> to be 0.0001.</p> |
| <p>Defined in src/operator/optimizer_op.cc:L835</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>n</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – n</p></li> |
| <li><p><strong>g</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – g</p></li> |
| <li><p><strong>delta</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – delta</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>gamma1</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.949999988</em>) – Decay rate.</p></li> |
| <li><p><strong>gamma2</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0.899999976</em>) – Decay rate.</p></li> |
| <li><p><strong>epsilon</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=9.99999994e-09</em>) – A small constant for numerical stability.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>clip_weights</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip weights to the range of [-clip_weights, clip_weights] If clip_weights <= 0, weight clipping is turned off. weights = max(min(weights, clip_weights), -clip_weights).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.round"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">round</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.round" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise rounded value to the nearest integer of the input.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="nb">round</span><span class="p">([</span><span class="o">-</span><span class="mf">1.5</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.9</span><span class="p">,</span> <span class="mf">1.9</span><span class="p">,</span> <span class="mf">2.1</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="o">-</span><span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">round</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>round(default) = default</p></li> |
| <li><p>round(row_sparse) = row_sparse</p></li> |
| <li><p>round(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L777</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.rsqrt"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">rsqrt</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.rsqrt" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise inverse square-root value of the input.</p> |
| <div class="math notranslate nohighlight"> |
| \[rsqrt(x) = 1/\sqrt{x}\]</div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">rsqrt</span><span class="p">([</span><span class="mi">4</span><span class="p">,</span><span class="mi">9</span><span class="p">,</span><span class="mi">16</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.33333334</span><span class="p">,</span> <span class="mf">0.25</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">rsqrt</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L221</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sample_exponential"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sample_exponential</code><span class="sig-paren">(</span><em class="sig-param">lam=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sample_exponential" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Concurrent sampling from multiple |
| exponential distributions with parameters lambda (rate).</p> |
| <p>The parameters of the distributions are provided as an input array. |
| Let <em>[s]</em> be the shape of the input array, <em>n</em> be the dimension of <em>[s]</em>, <em>[t]</em> |
| be the shape specified as the parameter of the operator, and <em>m</em> be the dimension |
| of <em>[t]</em>. Then the output will be a <em>(n+m)</em>-dimensional array with shape <em>[s]x[t]</em>.</p> |
| <p>For any valid <em>n</em>-dimensional index <em>i</em> with respect to the input array, <em>output[i]</em> |
| will be an <em>m</em>-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input value at index <em>i</em>. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input array.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">lam</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">8.5</span> <span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">single</span> <span class="n">sample</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_exponential</span><span class="p">(</span><span class="n">lam</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.51837951</span><span class="p">,</span> <span class="mf">0.09994757</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">vector</span> <span class="n">containing</span> <span class="n">two</span> <span class="n">samples</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_exponential</span><span class="p">(</span><span class="n">lam</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">2</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.51837951</span><span class="p">,</span> <span class="mf">0.19866663</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.09994757</span><span class="p">,</span> <span class="mf">0.50447971</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/multisample_op.cc:L283</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lam</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lambda (rate) parameters of the distributions.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Shape to be sampled from each random distribution.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sample_gamma"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sample_gamma</code><span class="sig-paren">(</span><em class="sig-param">alpha=None</em>, <em class="sig-param">beta=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sample_gamma" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Concurrent sampling from multiple |
| gamma distributions with parameters <em>alpha</em> (shape) and <em>beta</em> (scale).</p> |
| <p>The parameters of the distributions are provided as input arrays. |
| Let <em>[s]</em> be the shape of the input arrays, <em>n</em> be the dimension of <em>[s]</em>, <em>[t]</em> |
| be the shape specified as the parameter of the operator, and <em>m</em> be the dimension |
| of <em>[t]</em>. Then the output will be a <em>(n+m)</em>-dimensional array with shape <em>[s]x[t]</em>.</p> |
| <p>For any valid <em>n</em>-dimensional index <em>i</em> with respect to the input arrays, <em>output[i]</em> |
| will be an <em>m</em>-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index <em>i</em>. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">alpha</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">2.5</span> <span class="p">]</span> |
| <span class="n">beta</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.7</span> <span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">single</span> <span class="n">sample</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_gamma</span><span class="p">(</span><span class="n">alpha</span><span class="p">,</span> <span class="n">beta</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.</span> <span class="p">,</span> <span class="mf">2.25797319</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">vector</span> <span class="n">containing</span> <span class="n">two</span> <span class="n">samples</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_gamma</span><span class="p">(</span><span class="n">alpha</span><span class="p">,</span> <span class="n">beta</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">2</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span> <span class="p">,</span> <span class="mf">0.</span> <span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.25797319</span><span class="p">,</span> <span class="mf">1.70734084</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/multisample_op.cc:L281</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>alpha</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Alpha (shape) parameters of the distributions.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Shape to be sampled from each random distribution.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>beta</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Beta (scale) parameters of the distributions.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sample_generalized_negative_binomial"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sample_generalized_negative_binomial</code><span class="sig-paren">(</span><em class="sig-param">mu=None</em>, <em class="sig-param">alpha=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sample_generalized_negative_binomial" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Concurrent sampling from multiple |
| generalized negative binomial distributions with parameters <em>mu</em> (mean) and <em>alpha</em> (dispersion).</p> |
| <p>The parameters of the distributions are provided as input arrays. |
| Let <em>[s]</em> be the shape of the input arrays, <em>n</em> be the dimension of <em>[s]</em>, <em>[t]</em> |
| be the shape specified as the parameter of the operator, and <em>m</em> be the dimension |
| of <em>[t]</em>. Then the output will be a <em>(n+m)</em>-dimensional array with shape <em>[s]x[t]</em>.</p> |
| <p>For any valid <em>n</em>-dimensional index <em>i</em> with respect to the input arrays, <em>output[i]</em> |
| will be an <em>m</em>-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index <em>i</em>. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays.</p> |
| <p>Samples will always be returned as a floating point data type.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">mu</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">2.0</span><span class="p">,</span> <span class="mf">2.5</span> <span class="p">]</span> |
| <span class="n">alpha</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">0.1</span> <span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">single</span> <span class="n">sample</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_generalized_negative_binomial</span><span class="p">(</span><span class="n">mu</span><span class="p">,</span> <span class="n">alpha</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">vector</span> <span class="n">containing</span> <span class="n">two</span> <span class="n">samples</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_generalized_negative_binomial</span><span class="p">(</span><span class="n">mu</span><span class="p">,</span> <span class="n">alpha</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">2</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/multisample_op.cc:L292</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>mu</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Means of the distributions.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Shape to be sampled from each random distribution.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>alpha</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Alpha (dispersion) parameters of the distributions.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sample_multinomial"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sample_multinomial</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">get_prob=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sample_multinomial" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Concurrent sampling from multiple multinomial distributions.</p> |
| <p><em>data</em> is an <em>n</em> dimensional array whose last dimension has length <em>k</em>, where |
| <em>k</em> is the number of possible outcomes of each multinomial distribution. This |
| operator will draw <em>shape</em> samples from each distribution. If shape is empty |
| one sample will be drawn from each distribution.</p> |
| <p>If <em>get_prob</em> is true, a second array containing log likelihood of the drawn |
| samples will also be returned. This is usually used for reinforcement learning |
| where you can provide reward as head gradient for this array to estimate |
| gradient.</p> |
| <p>Note that the input distribution must be normalized, i.e. <em>data</em> must sum to |
| 1 along its last axis.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">probs</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.4</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.4</span><span class="p">,</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">single</span> <span class="n">sample</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_multinomial</span><span class="p">(</span><span class="n">probs</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">0</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">vector</span> <span class="n">containing</span> <span class="n">two</span> <span class="n">samples</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_multinomial</span><span class="p">(</span><span class="n">probs</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">2</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">4</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">requests</span> <span class="n">log</span> <span class="n">likelihood</span> |
| <span class="n">sample_multinomial</span><span class="p">(</span><span class="n">probs</span><span class="p">,</span> <span class="n">get_prob</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.3</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>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Distribution probabilities. Must sum to one on the last axis.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Shape to be sampled from each random distribution.</p></li> |
| <li><p><strong>get_prob</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to also return the log probability of sampled result. This is usually used for differentiating through stochastic variables, e.g. in reinforcement learning.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'uint8'}</em><em>,</em><em>optional</em><em>, </em><em>default='int32'</em>) – DType of the output in case this can’t be inferred.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sample_negative_binomial"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sample_negative_binomial</code><span class="sig-paren">(</span><em class="sig-param">k=None</em>, <em class="sig-param">p=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sample_negative_binomial" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Concurrent sampling from multiple |
| negative binomial distributions with parameters <em>k</em> (failure limit) and <em>p</em> (failure probability).</p> |
| <p>The parameters of the distributions are provided as input arrays. |
| Let <em>[s]</em> be the shape of the input arrays, <em>n</em> be the dimension of <em>[s]</em>, <em>[t]</em> |
| be the shape specified as the parameter of the operator, and <em>m</em> be the dimension |
| of <em>[t]</em>. Then the output will be a <em>(n+m)</em>-dimensional array with shape <em>[s]x[t]</em>.</p> |
| <p>For any valid <em>n</em>-dimensional index <em>i</em> with respect to the input arrays, <em>output[i]</em> |
| will be an <em>m</em>-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index <em>i</em>. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays.</p> |
| <p>Samples will always be returned as a floating point data type.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">k</span> <span class="o">=</span> <span class="p">[</span> <span class="mi">20</span><span class="p">,</span> <span class="mi">49</span> <span class="p">]</span> |
| <span class="n">p</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.4</span> <span class="p">,</span> <span class="mf">0.77</span> <span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">single</span> <span class="n">sample</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_negative_binomial</span><span class="p">(</span><span class="n">k</span><span class="p">,</span> <span class="n">p</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">15.</span><span class="p">,</span> <span class="mf">16.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">vector</span> <span class="n">containing</span> <span class="n">two</span> <span class="n">samples</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_negative_binomial</span><span class="p">(</span><span class="n">k</span><span class="p">,</span> <span class="n">p</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">2</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">15.</span><span class="p">,</span> <span class="mf">50.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">16.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/multisample_op.cc:L288</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>k</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Limits of unsuccessful experiments.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Shape to be sampled from each random distribution.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>p</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Failure probabilities in each experiment.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sample_normal"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sample_normal</code><span class="sig-paren">(</span><em class="sig-param">mu=None</em>, <em class="sig-param">sigma=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sample_normal" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Concurrent sampling from multiple |
| normal distributions with parameters <em>mu</em> (mean) and <em>sigma</em> (standard deviation).</p> |
| <p>The parameters of the distributions are provided as input arrays. |
| Let <em>[s]</em> be the shape of the input arrays, <em>n</em> be the dimension of <em>[s]</em>, <em>[t]</em> |
| be the shape specified as the parameter of the operator, and <em>m</em> be the dimension |
| of <em>[t]</em>. Then the output will be a <em>(n+m)</em>-dimensional array with shape <em>[s]x[t]</em>.</p> |
| <p>For any valid <em>n</em>-dimensional index <em>i</em> with respect to the input arrays, <em>output[i]</em> |
| will be an <em>m</em>-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index <em>i</em>. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">mu</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">2.5</span> <span class="p">]</span> |
| <span class="n">sigma</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">3.7</span> <span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">single</span> <span class="n">sample</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_normal</span><span class="p">(</span><span class="n">mu</span><span class="p">,</span> <span class="n">sigma</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">0.56410581</span><span class="p">,</span> <span class="mf">0.95934606</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">vector</span> <span class="n">containing</span> <span class="n">two</span> <span class="n">samples</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_normal</span><span class="p">(</span><span class="n">mu</span><span class="p">,</span> <span class="n">sigma</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">2</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span><span class="o">-</span><span class="mf">0.56410581</span><span class="p">,</span> <span class="mf">0.2928229</span> <span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.95934606</span><span class="p">,</span> <span class="mf">4.48287058</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/multisample_op.cc:L278</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>mu</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Means of the distributions.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Shape to be sampled from each random distribution.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>sigma</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Standard deviations of the distributions.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sample_poisson"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sample_poisson</code><span class="sig-paren">(</span><em class="sig-param">lam=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sample_poisson" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Concurrent sampling from multiple |
| Poisson distributions with parameters lambda (rate).</p> |
| <p>The parameters of the distributions are provided as an input array. |
| Let <em>[s]</em> be the shape of the input array, <em>n</em> be the dimension of <em>[s]</em>, <em>[t]</em> |
| be the shape specified as the parameter of the operator, and <em>m</em> be the dimension |
| of <em>[t]</em>. Then the output will be a <em>(n+m)</em>-dimensional array with shape <em>[s]x[t]</em>.</p> |
| <p>For any valid <em>n</em>-dimensional index <em>i</em> with respect to the input array, <em>output[i]</em> |
| will be an <em>m</em>-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input value at index <em>i</em>. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input array.</p> |
| <p>Samples will always be returned as a floating point data type.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">lam</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">8.5</span> <span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">single</span> <span class="n">sample</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_poisson</span><span class="p">(</span><span class="n">lam</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">13.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">vector</span> <span class="n">containing</span> <span class="n">two</span> <span class="n">samples</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_poisson</span><span class="p">(</span><span class="n">lam</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">2</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">13.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/multisample_op.cc:L285</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>lam</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lambda (rate) parameters of the distributions.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Shape to be sampled from each random distribution.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sample_uniform"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sample_uniform</code><span class="sig-paren">(</span><em class="sig-param">low=None</em>, <em class="sig-param">high=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sample_uniform" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Concurrent sampling from multiple |
| uniform distributions on the intervals given by <em>[low,high)</em>.</p> |
| <p>The parameters of the distributions are provided as input arrays. |
| Let <em>[s]</em> be the shape of the input arrays, <em>n</em> be the dimension of <em>[s]</em>, <em>[t]</em> |
| be the shape specified as the parameter of the operator, and <em>m</em> be the dimension |
| of <em>[t]</em>. Then the output will be a <em>(n+m)</em>-dimensional array with shape <em>[s]x[t]</em>.</p> |
| <p>For any valid <em>n</em>-dimensional index <em>i</em> with respect to the input arrays, <em>output[i]</em> |
| will be an <em>m</em>-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index <em>i</em>. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">low</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">2.5</span> <span class="p">]</span> |
| <span class="n">high</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">3.7</span> <span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">single</span> <span class="n">sample</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_uniform</span><span class="p">(</span><span class="n">low</span><span class="p">,</span> <span class="n">high</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">0.40451524</span><span class="p">,</span> <span class="mf">3.18687344</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Draw</span> <span class="n">a</span> <span class="n">vector</span> <span class="n">containing</span> <span class="n">two</span> <span class="n">samples</span> <span class="k">for</span> <span class="n">each</span> <span class="n">distribution</span> |
| <span class="n">sample_uniform</span><span class="p">(</span><span class="n">low</span><span class="p">,</span> <span class="n">high</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">2</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.40451524</span><span class="p">,</span> <span class="mf">0.18017688</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.18687344</span><span class="p">,</span> <span class="mf">3.68352246</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/multisample_op.cc:L276</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>low</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Lower bounds of the distributions.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Shape to be sampled from each random distribution.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>high</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Upper bounds of the distributions.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.scatter_nd"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">scatter_nd</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">indices=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.scatter_nd" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Scatters data into a new tensor according to indices.</p> |
| <p>Given <cite>data</cite> with shape <cite>(Y_0, …, Y_{K-1}, X_M, …, X_{N-1})</cite> and indices with shape |
| <cite>(M, Y_0, …, Y_{K-1})</cite>, the output will have shape <cite>(X_0, X_1, …, X_{N-1})</cite>, |
| where <cite>M <= N</cite>. If <cite>M == N</cite>, data shape should simply be <cite>(Y_0, …, Y_{K-1})</cite>.</p> |
| <p>The elements in output is defined as follows:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">output</span><span class="p">[</span><span class="n">indices</span><span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="n">y_0</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">y_</span><span class="p">{</span><span class="n">K</span><span class="o">-</span><span class="mi">1</span><span class="p">}],</span> |
| <span class="o">...</span><span class="p">,</span> |
| <span class="n">indices</span><span class="p">[</span><span class="n">M</span><span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="n">y_0</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">y_</span><span class="p">{</span><span class="n">K</span><span class="o">-</span><span class="mi">1</span><span class="p">}],</span> |
| <span class="n">x_M</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">x_</span><span class="p">{</span><span class="n">N</span><span class="o">-</span><span class="mi">1</span><span class="p">}]</span> <span class="o">=</span> <span class="n">data</span><span class="p">[</span><span class="n">y_0</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">y_</span><span class="p">{</span><span class="n">K</span><span class="o">-</span><span class="mi">1</span><span class="p">},</span> <span class="n">x_M</span><span class="p">,</span> <span class="o">...</span><span class="p">,</span> <span class="n">x_</span><span class="p">{</span><span class="n">N</span><span class="o">-</span><span class="mi">1</span><span class="p">}]</span> |
| </pre></div> |
| </div> |
| <p>all other entries in output are 0.</p> |
| <div class="admonition warning"> |
| <p class="admonition-title">Warning</p> |
| <p>If the indices have duplicates, the result will be non-deterministic and |
| the gradient of <cite>scatter_nd</cite> will not be correct!!</p> |
| </div> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">data</span> <span class="o">=</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">0</span><span class="p">]</span> |
| <span class="n">indices</span> <span class="o">=</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">0</span><span class="p">],</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">0</span><span class="p">]]</span> |
| <span class="n">shape</span> <span class="o">=</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="n">scatter_nd</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">indices</span><span class="p">,</span> <span class="n">shape</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</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="n">data</span> <span class="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]],</span> <span class="p">[[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">8</span><span class="p">]]]</span> |
| <span class="n">indices</span> <span class="o">=</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="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">]]</span> |
| <span class="n">shape</span> <span class="o">=</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="mi">2</span><span class="p">)</span> |
| <span class="n">scatter_nd</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">indices</span><span class="p">,</span> <span class="n">shape</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">]],</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="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]]],</span> |
| |
| <span class="p">[[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">]],</span> |
| |
| <span class="p">[[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">8</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>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – data</p></li> |
| <li><p><strong>indices</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – indices</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – Shape of output.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sgd_mom_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sgd_mom_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">mom=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">momentum=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">lazy_update=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sgd_mom_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Momentum update function for Stochastic Gradient Descent (SGD) optimizer.</p> |
| <p>Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below:</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-1} - \alpha * \nabla J(W_{t-1})\\ |
| W_t = W_{t-1} + v_t\end{split}\]</div> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">v</span> <span class="o">=</span> <span class="n">momentum</span> <span class="o">*</span> <span class="n">v</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="n">gradient</span> |
| <span class="n">weight</span> <span class="o">+=</span> <span class="n">v</span> |
| </pre></div> |
| </div> |
| <p>Where the parameter <code class="docutils literal notranslate"><span class="pre">momentum</span></code> is the decay rate of momentum estimates at each epoch.</p> |
| <p>However, if grad’s storage type is <code class="docutils literal notranslate"><span class="pre">row_sparse</span></code>, <code class="docutils literal notranslate"><span class="pre">lazy_update</span></code> is True and weight’s storage |
| type is the same as momentum’s storage type, |
| only the row slices whose indices appear in grad.indices are updated (for both weight and momentum):</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="k">for</span> <span class="n">row</span> <span class="ow">in</span> <span class="n">gradient</span><span class="o">.</span><span class="n">indices</span><span class="p">:</span> |
| <span class="n">v</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">=</span> <span class="n">momentum</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">*</span> <span class="n">v</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="n">gradient</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> |
| <span class="n">weight</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">+=</span> <span class="n">v</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/optimizer_op.cc:L564</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>mom</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Momentum</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The decay rate of momentum estimates at each epoch.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>lazy_update</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – If true, lazy updates are applied if gradient’s stype is row_sparse and both weight and momentum have the same stype</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sgd_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sgd_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">lazy_update=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sgd_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for Stochastic Gradient Descent (SGD) optimizer.</p> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">weight</span> <span class="o">=</span> <span class="n">weight</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="p">(</span><span class="n">gradient</span> <span class="o">+</span> <span class="n">wd</span> <span class="o">*</span> <span class="n">weight</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>However, if gradient is of <code class="docutils literal notranslate"><span class="pre">row_sparse</span></code> storage type and <code class="docutils literal notranslate"><span class="pre">lazy_update</span></code> is True, |
| only the row slices whose indices appear in grad.indices are updated:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="k">for</span> <span class="n">row</span> <span class="ow">in</span> <span class="n">gradient</span><span class="o">.</span><span class="n">indices</span><span class="p">:</span> |
| <span class="n">weight</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">=</span> <span class="n">weight</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="p">(</span><span class="n">gradient</span><span class="p">[</span><span class="n">row</span><span class="p">]</span> <span class="o">+</span> <span class="n">wd</span> <span class="o">*</span> <span class="n">weight</span><span class="p">[</span><span class="n">row</span><span class="p">])</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/optimizer_op.cc:L523</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>lazy_update</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – If true, lazy updates are applied if gradient’s stype is row_sparse.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.shape_array"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">shape_array</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.shape_array" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a 1D int64 array containing the shape of data.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">shape_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="p">[</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">,</span><span class="mi">7</span><span class="p">,</span><span class="mi">8</span><span class="p">]])</span> <span class="o">=</span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span><span class="mi">4</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L573</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input Array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.shuffle"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">shuffle</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.shuffle" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Randomly shuffle the elements.</p> |
| <p>This shuffles the array along the first axis. |
| The order of the elements in each subarray does not change. |
| For example, if a 2D array is given, the order of the rows randomly changes, |
| but the order of the elements in each row does not change.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Data to be shuffled.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sigmoid"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sigmoid</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sigmoid" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes sigmoid of x element-wise.</p> |
| <div class="math notranslate nohighlight"> |
| \[y = 1 / (1 + exp(-x))\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">sigmoid</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L119</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sign"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sign</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sign" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise sign of the input.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">sign</span><span class="p">([</span><span class="o">-</span><span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">3</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">1</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">sign</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>sign(default) = default</p></li> |
| <li><p>sign(row_sparse) = row_sparse</p></li> |
| <li><p>sign(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L758</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.signsgd_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">signsgd_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.signsgd_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Update function for SignSGD optimizer.</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}g_t = \nabla J(W_{t-1})\\ |
| W_t = W_{t-1} - \eta_t \text{sign}(g_t)\end{split}\]</div> |
| <p>It updates the weights using:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">weight</span> <span class="o">=</span> <span class="n">weight</span> <span class="o">-</span> <span class="n">learning_rate</span> <span class="o">*</span> <span class="n">sign</span><span class="p">(</span><span class="n">gradient</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <ul class="simple"> |
| <li><p>sparse ndarray not supported for this optimizer yet.</p></li> |
| </ul> |
| </div> |
| <p>Defined in src/operator/optimizer_op.cc:L62</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.signum_update"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">signum_update</code><span class="sig-paren">(</span><em class="sig-param">weight=None</em>, <em class="sig-param">grad=None</em>, <em class="sig-param">mom=None</em>, <em class="sig-param">lr=_Null</em>, <em class="sig-param">momentum=_Null</em>, <em class="sig-param">wd=_Null</em>, <em class="sig-param">rescale_grad=_Null</em>, <em class="sig-param">clip_gradient=_Null</em>, <em class="sig-param">wd_lh=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.signum_update" title="Permalink to this definition">¶</a></dt> |
| <dd><p>SIGN momentUM (Signum) optimizer.</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}g_t = \nabla J(W_{t-1})\\ |
| m_t = \beta m_{t-1} + (1 - \beta) g_t\\ |
| W_t = W_{t-1} - \eta_t \text{sign}(m_t)\end{split}\]</div> |
| <dl class="simple"> |
| <dt>It updates the weights using::</dt><dd><p>state = momentum * state + (1-momentum) * gradient |
| weight = weight - learning_rate * sign(state)</p> |
| </dd> |
| </dl> |
| <p>Where the parameter <code class="docutils literal notranslate"><span class="pre">momentum</span></code> is the decay rate of momentum estimates at each epoch.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <ul class="simple"> |
| <li><p>sparse ndarray not supported for this optimizer yet.</p></li> |
| </ul> |
| </div> |
| <p>Defined in src/operator/optimizer_op.cc:L91</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>weight</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Weight</p></li> |
| <li><p><strong>grad</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Gradient</p></li> |
| <li><p><strong>mom</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Momentum</p></li> |
| <li><p><strong>lr</strong> (<em>float</em><em>, </em><em>required</em>) – Learning rate</p></li> |
| <li><p><strong>momentum</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The decay rate of momentum estimates at each epoch.</p></li> |
| <li><p><strong>wd</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Weight decay augments the objective function with a regularization term that penalizes large weights. The penalty scales with the square of the magnitude of each weight.</p></li> |
| <li><p><strong>rescale_grad</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Rescale gradient to grad = rescale_grad*grad.</p></li> |
| <li><p><strong>clip_gradient</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=-1</em>) – Clip gradient to the range of [-clip_gradient, clip_gradient] If clip_gradient <= 0, gradient clipping is turned off. grad = max(min(grad, clip_gradient), -clip_gradient).</p></li> |
| <li><p><strong>wd_lh</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – The amount of weight decay that does not go into gradient/momentum calculationsotherwise do weight decay algorithmically only.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sin"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sin</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sin" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the element-wise sine of the input array.</p> |
| <p>The input should be in radians (<span class="math notranslate nohighlight">\(2\pi\)</span> rad equals 360 degrees).</p> |
| <div class="math notranslate nohighlight"> |
| \[sin([0, \pi/4, \pi/2]) = [0, 0.707, 1]\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">sin</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>sin(default) = default</p></li> |
| <li><p>sin(row_sparse) = row_sparse</p></li> |
| <li><p>sin(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L47</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sinh"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sinh</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sinh" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the hyperbolic sine of the input array, computed element-wise.</p> |
| <div class="math notranslate nohighlight"> |
| \[sinh(x) = 0.5\times(exp(x) - exp(-x))\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">sinh</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>sinh(default) = default</p></li> |
| <li><p>sinh(row_sparse) = row_sparse</p></li> |
| <li><p>sinh(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L371</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.size_array"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">size_array</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.size_array" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a 1D int64 array containing the size of data.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">size_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="p">[</span><span class="mi">5</span><span class="p">,</span><span class="mi">6</span><span class="p">,</span><span class="mi">7</span><span class="p">,</span><span class="mi">8</span><span class="p">]])</span> <span class="o">=</span> <span class="p">[</span><span class="mi">8</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L624</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input Array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.slice"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">slice</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">begin=_Null</em>, <em class="sig-param">end=_Null</em>, <em class="sig-param">step=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.slice" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Slices a region of the array. |
| .. note:: <code class="docutils literal notranslate"><span class="pre">crop</span></code> is deprecated. Use <code class="docutils literal notranslate"><span class="pre">slice</span></code> instead. |
| This function returns a sliced array between the indices given |
| by <cite>begin</cite> and <cite>end</cite> with the corresponding <cite>step</cite>. |
| For an input array of <code class="docutils literal notranslate"><span class="pre">shape=(d_0,</span> <span class="pre">d_1,</span> <span class="pre">...,</span> <span class="pre">d_n-1)</span></code>, |
| slice operation with <code class="docutils literal notranslate"><span class="pre">begin=(b_0,</span> <span class="pre">b_1...b_m-1)</span></code>, |
| <code class="docutils literal notranslate"><span class="pre">end=(e_0,</span> <span class="pre">e_1,</span> <span class="pre">...,</span> <span class="pre">e_m-1)</span></code>, and <code class="docutils literal notranslate"><span class="pre">step=(s_0,</span> <span class="pre">s_1,</span> <span class="pre">...,</span> <span class="pre">s_m-1)</span></code>, |
| where m <= n, results in an array with the shape |
| <code class="docutils literal notranslate"><span class="pre">(|e_0-b_0|/|s_0|,</span> <span class="pre">...,</span> <span class="pre">|e_m-1-b_m-1|/|s_m-1|,</span> <span class="pre">d_m,</span> <span class="pre">...,</span> <span class="pre">d_n-1)</span></code>. |
| The resulting array’s <em>k</em>-th dimension contains elements |
| from the <em>k</em>-th dimension of the input array starting |
| from index <code class="docutils literal notranslate"><span class="pre">b_k</span></code> (inclusive) with step <code class="docutils literal notranslate"><span class="pre">s_k</span></code> |
| until reaching <code class="docutils literal notranslate"><span class="pre">e_k</span></code> (exclusive). |
| If the <em>k</em>-th elements are <cite>None</cite> in the sequence of <cite>begin</cite>, <cite>end</cite>, |
| and <cite>step</cite>, the following rule will be used to set default values. |
| If <cite>s_k</cite> is <cite>None</cite>, set <cite>s_k=1</cite>. If <cite>s_k > 0</cite>, set <cite>b_k=0</cite>, <cite>e_k=d_k</cite>; |
| else, set <cite>b_k=d_k-1</cite>, <cite>e_k=-1</cite>. |
| The storage type of <code class="docutils literal notranslate"><span class="pre">slice</span></code> output depends on storage types of inputs |
| - slice(csr) = csr |
| - otherwise, <code class="docutils literal notranslate"><span class="pre">slice</span></code> generates output with default storage |
| .. note:: When input data storage type is csr, it only supports</p> |
| <blockquote> |
| <div><p>step=(), or step=(None,), or step=(1,) to generate a csr output. |
| For other step parameter values, it falls back to slicing |
| a dense tensor.</p> |
| </div></blockquote> |
| <dl class="simple"> |
| <dt>Example::</dt><dd><dl class="simple"> |
| <dt>x = [[ 1., 2., 3., 4.],</dt><dd><p>[ 5., 6., 7., 8.], |
| [ 9., 10., 11., 12.]]</p> |
| </dd> |
| <dt>slice(x, begin=(0,1), end=(2,4)) = [[ 2., 3., 4.],</dt><dd><p>[ 6., 7., 8.]]</p> |
| </dd> |
| <dt>slice(x, begin=(None, 0), end=(None, 3), step=(-1, 2)) = [[9., 11.],</dt><dd><p>[5., 7.], |
| [1., 3.]]</p> |
| </dd> |
| </dl> |
| </dd> |
| </dl> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L481</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Source input</p></li> |
| <li><p><strong>begin</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – starting indices for the slice operation, supports negative indices.</p></li> |
| <li><p><strong>end</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – ending indices for the slice operation, supports negative indices.</p></li> |
| <li><p><strong>step</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – step for the slice operation, supports negative values.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.slice_axis"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">slice_axis</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">begin=_Null</em>, <em class="sig-param">end=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.slice_axis" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Slices along a given axis. |
| Returns an array slice along a given <cite>axis</cite> starting from the <cite>begin</cite> index |
| to the <cite>end</cite> index. |
| Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">9.</span><span class="p">,</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]]</span> |
| <span class="n">slice_axis</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">begin</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">end</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">9.</span><span class="p">,</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]]</span> |
| <span class="n">slice_axis</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">begin</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">end</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">9.</span><span class="p">,</span> <span class="mf">10.</span><span class="p">]]</span> |
| <span class="n">slice_axis</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">begin</span><span class="o">=-</span><span class="mi">3</span><span class="p">,</span> <span class="n">end</span><span class="o">=-</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L570</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Source input</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>required</em>) – Axis along which to be sliced, supports negative indexes.</p></li> |
| <li><p><strong>begin</strong> (<em>int</em><em>, </em><em>required</em>) – The beginning index along the axis to be sliced, supports negative indexes.</p></li> |
| <li><p><strong>end</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>required</em>) – The ending index along the axis to be sliced, supports negative indexes.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.slice_like"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">slice_like</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">shape_like=None</em>, <em class="sig-param">axes=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.slice_like" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Slices a region of the array like the shape of another array. |
| This function is similar to <code class="docutils literal notranslate"><span class="pre">slice</span></code>, however, the <cite>begin</cite> are always <cite>0`s |
| and `end</cite> of specific axes are inferred from the second input <cite>shape_like</cite>. |
| Given the second <cite>shape_like</cite> input of <code class="docutils literal notranslate"><span class="pre">shape=(d_0,</span> <span class="pre">d_1,</span> <span class="pre">...,</span> <span class="pre">d_n-1)</span></code>, |
| a <code class="docutils literal notranslate"><span class="pre">slice_like</span></code> operator with default empty <cite>axes</cite>, it performs the |
| following operation: |
| `` out = slice(input, begin=(0, 0, …, 0), end=(d_0, d_1, …, d_n-1))``. |
| When <cite>axes</cite> is not empty, it is used to speficy which axes are being sliced. |
| Given a 4-d input data, <code class="docutils literal notranslate"><span class="pre">slice_like</span></code> operator with <code class="docutils literal notranslate"><span class="pre">axes=(0,</span> <span class="pre">2,</span> <span class="pre">-1)</span></code> |
| will perform the following operation: |
| `` out = slice(input, begin=(0, 0, 0, 0), end=(d_0, None, d_2, d_3))``. |
| Note that it is allowed to have first and second input with different dimensions, |
| however, you have to make sure the <cite>axes</cite> are specified and not exceeding the |
| dimension limits. |
| For example, given <cite>input_1</cite> with <code class="docutils literal notranslate"><span class="pre">shape=(2,3,4,5)</span></code> and <cite>input_2</cite> with |
| <code class="docutils literal notranslate"><span class="pre">shape=(1,2,3)</span></code>, it is not allowed to use: |
| `` out = slice_like(a, b)`` because ndim of <cite>input_1</cite> is 4, and ndim of <cite>input_2</cite> |
| is 3. |
| The following is allowed in this situation: |
| `` out = slice_like(a, b, axes=(0, 2))`` |
| Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">9.</span><span class="p">,</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">,</span> <span class="mf">12.</span><span class="p">]]</span> |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">]]</span> |
| <span class="n">slice_like</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">]]</span> |
| <span class="n">slice_like</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="n">axes</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">]]</span> |
| <span class="n">slice_like</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="n">axes</span><span class="o">=</span><span class="p">(</span><span class="mi">0</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]</span> |
| <span class="n">slice_like</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="n">axes</span><span class="o">=</span><span class="p">(</span><span class="o">-</span><span class="mi">1</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">9.</span><span class="p">,</span> <span class="mf">10.</span><span class="p">,</span> <span class="mf">11.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L624</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Source input</p></li> |
| <li><p><strong>shape_like</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Shape like input</p></li> |
| <li><p><strong>axes</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – List of axes on which input data will be sliced according to the corresponding size of the second input. By default will slice on all axes. Negative axes are supported.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.smooth_l1"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">smooth_l1</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">scalar=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.smooth_l1" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Calculate Smooth L1 Loss(lhs, scalar) by summing</p> |
| <div class="math notranslate nohighlight"> |
| \[\begin{split}f(x) = |
| \begin{cases} |
| (\sigma x)^2/2,& \text{if }x < 1/\sigma^2\\ |
| |x|-0.5/\sigma^2,& \text{otherwise} |
| \end{cases}\end{split}\]</div> |
| <p>where <span class="math notranslate nohighlight">\(x\)</span> is an element of the tensor <em>lhs</em> and <span class="math notranslate nohighlight">\(\sigma\)</span> is the scalar.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">smooth_l1</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="o">=</span> <span class="p">[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">,</span> <span class="mf">2.5</span><span class="p">,</span> <span class="mf">3.5</span><span class="p">]</span> |
| <span class="n">smooth_l1</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="n">scalar</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">,</span> <span class="mf">2.5</span><span class="p">,</span> <span class="mf">3.5</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_binary_scalar_op_extended.cc:L108</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – source input</p></li> |
| <li><p><strong>scalar</strong> (<em>float</em>) – scalar input</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.softmax"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">softmax</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">length=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">temperature=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">use_length=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.softmax" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies the softmax function.</p> |
| <p>The resulting array contains elements in the range (0,1) and the elements along the given axis sum up to 1.</p> |
| <div class="math notranslate nohighlight"> |
| \[softmax(\mathbf{z/t})_j = \frac{e^{z_j/t}}{\sum_{k=1}^K e^{z_k/t}}\]</div> |
| <p>for <span class="math notranslate nohighlight">\(j = 1, ..., K\)</span></p> |
| <p>t is the temperature parameter in softmax function. By default, t equals 1.0</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">softmax</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.5</span> <span class="mf">0.5</span> <span class="mf">0.5</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">0.5</span> <span class="mf">0.5</span> <span class="mf">0.5</span><span class="p">]]</span> |
| |
| <span class="n">softmax</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.33333334</span><span class="p">,</span> <span class="mf">0.33333334</span><span class="p">,</span> <span class="mf">0.33333334</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.33333334</span><span class="p">,</span> <span class="mf">0.33333334</span><span class="p">,</span> <span class="mf">0.33333334</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/nn/softmax.cc:L135</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>length</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The length array.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='-1'</em>) – The axis along which to compute softmax.</p></li> |
| <li><p><strong>temperature</strong> (<em>double</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Temperature parameter in softmax</p></li> |
| <li><p><strong>dtype</strong> (<em>{None</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to the same as input’s dtype if not defined (dtype=None).</p></li> |
| <li><p><strong>use_length</strong> (<em>boolean</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to use the length input as a mask over the data input.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.softmax_cross_entropy"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">softmax_cross_entropy</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">label=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.softmax_cross_entropy" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Calculate cross entropy of softmax output and one-hot label.</p> |
| <ul> |
| <li><p>This operator computes the cross entropy in two steps: |
| - Applies softmax function on the input array. |
| - Computes and returns the cross entropy loss between the softmax output and the labels.</p></li> |
| <li><p>The softmax function and cross entropy loss is given by:</p> |
| <ul class="simple"> |
| <li><p>Softmax Function:</p></li> |
| </ul> |
| <div class="math notranslate nohighlight"> |
| \[\text{softmax}(x)_i = \frac{exp(x_i)}{\sum_j exp(x_j)}\]</div> |
| <ul class="simple"> |
| <li><p>Cross Entropy Function:</p></li> |
| </ul> |
| <div class="math notranslate nohighlight"> |
| \[\text{CE(label, output)} = - \sum_i \text{label}_i \log(\text{output}_i)\]</div> |
| </li> |
| </ul> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="p">[</span><span class="mi">11</span><span class="p">,</span> <span class="mi">7</span><span class="p">,</span> <span class="mi">5</span><span class="p">]]</span> |
| |
| <span class="n">label</span> <span class="o">=</span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">]</span> |
| |
| <span class="n">softmax</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.09003057</span><span class="p">,</span> <span class="mf">0.24472848</span><span class="p">,</span> <span class="mf">0.66524094</span><span class="p">],</span> |
| <span class="p">[</span><span class="mf">0.97962922</span><span class="p">,</span> <span class="mf">0.01794253</span><span class="p">,</span> <span class="mf">0.00242826</span><span class="p">]]</span> |
| |
| <span class="n">softmax_cross_entropy</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">label</span><span class="p">)</span> <span class="o">=</span> <span class="o">-</span> <span class="n">log</span><span class="p">(</span><span class="mf">0.66524084</span><span class="p">)</span> <span class="o">-</span> <span class="n">log</span><span class="p">(</span><span class="mf">0.97962922</span><span class="p">)</span> <span class="o">=</span> <span class="mf">0.4281871</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/loss_binary_op.cc:L58</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data</p></li> |
| <li><p><strong>label</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input label</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.softmin"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">softmin</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">temperature=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">use_length=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.softmin" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies the softmin function.</p> |
| <p>The resulting array contains elements in the range (0,1) and the elements along the given axis sum |
| up to 1.</p> |
| <div class="math notranslate nohighlight"> |
| \[softmin(\mathbf{z/t})_j = \frac{e^{-z_j/t}}{\sum_{k=1}^K e^{-z_k/t}}\]</div> |
| <p>for <span class="math notranslate nohighlight">\(j = 1, ..., K\)</span></p> |
| <p>t is the temperature parameter in softmax function. By default, t equals 1.0</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span> <span class="mf">2.</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">3.</span> <span class="mf">2.</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">softmin</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.88079703</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.11920292</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.11920292</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.88079703</span><span class="p">]]</span> |
| |
| <span class="n">softmin</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.66524094</span><span class="p">,</span> <span class="mf">0.24472848</span><span class="p">,</span> <span class="mf">0.09003057</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.09003057</span><span class="p">,</span> <span class="mf">0.24472848</span><span class="p">,</span> <span class="mf">0.66524094</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/nn/softmin.cc:L56</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='-1'</em>) – The axis along which to compute softmax.</p></li> |
| <li><p><strong>temperature</strong> (<em>double</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Temperature parameter in softmax</p></li> |
| <li><p><strong>dtype</strong> (<em>{None</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to the same as input’s dtype if not defined (dtype=None).</p></li> |
| <li><p><strong>use_length</strong> (<em>boolean</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to use the length input as a mask over the data input.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.softsign"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">softsign</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.softsign" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes softsign of x element-wise.</p> |
| <div class="math notranslate nohighlight"> |
| \[y = x / (1 + abs(x))\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">softsign</span></code> output is always dense</p> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L191</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sort"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sort</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">is_ascend=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sort" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns a sorted copy of an input array along the given axis.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">1</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">sorts</span> <span class="n">along</span> <span class="n">the</span> <span class="n">last</span> <span class="n">axis</span> |
| <span class="n">sort</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">flattens</span> <span class="ow">and</span> <span class="n">then</span> <span class="n">sorts</span> |
| <span class="n">sort</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="kc">None</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">sorts</span> <span class="n">along</span> <span class="n">the</span> <span class="n">first</span> <span class="n">axis</span> |
| <span class="n">sort</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="ow">in</span> <span class="n">a</span> <span class="n">descend</span> <span class="n">order</span> |
| <span class="n">sort</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">is_ascend</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/ordering_op.cc:L132</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='-1'</em>) – Axis along which to choose sort the input tensor. If not given, the flattened array is used. Default is -1.</p></li> |
| <li><p><strong>is_ascend</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Whether to sort in ascending or descending order.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.space_to_depth"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">space_to_depth</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">block_size=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.space_to_depth" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Rearranges(permutes) blocks of spatial data into depth. |
| Similar to ONNX SpaceToDepth operator: |
| <a class="reference external" href="https://github.com/onnx/onnx/blob/master/docs/Operators.md#SpaceToDepth">https://github.com/onnx/onnx/blob/master/docs/Operators.md#SpaceToDepth</a> |
| The output is a new tensor where the values from height and width dimension are |
| moved to the depth dimension. The reverse of this operation is <code class="docutils literal notranslate"><span class="pre">depth_to_space</span></code>. |
| .. math:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span>\<span class="n">begin</span><span class="p">{</span><span class="n">gather</span><span class="o">*</span><span class="p">}</span> |
| <span class="n">x</span> \<span class="n">prime</span> <span class="o">=</span> <span class="n">reshape</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="p">[</span><span class="n">N</span><span class="p">,</span> <span class="n">C</span><span class="p">,</span> <span class="n">H</span> <span class="o">/</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">,</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">,</span> <span class="n">W</span> <span class="o">/</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">,</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">])</span> \\ |
| <span class="n">x</span> \<span class="n">prime</span> \<span class="n">prime</span> <span class="o">=</span> <span class="n">transpose</span><span class="p">(</span><span class="n">x</span> \<span class="n">prime</span><span class="p">,</span> <span class="p">[</span><span class="mi">0</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">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">4</span><span class="p">])</span> \\ |
| <span class="n">y</span> <span class="o">=</span> <span class="n">reshape</span><span class="p">(</span><span class="n">x</span> \<span class="n">prime</span> \<span class="n">prime</span><span class="p">,</span> <span class="p">[</span><span class="n">N</span><span class="p">,</span> <span class="n">C</span> <span class="o">*</span> <span class="p">(</span><span class="n">block</span>\<span class="n">_size</span> <span class="o">^</span> <span class="mi">2</span><span class="p">),</span> <span class="n">H</span> <span class="o">/</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">,</span> <span class="n">W</span> <span class="o">/</span> <span class="n">block</span>\<span class="n">_size</span><span class="p">])</span> |
| \<span class="n">end</span><span class="p">{</span><span class="n">gather</span><span class="o">*</span><span class="p">}</span> |
| </pre></div> |
| </div> |
| <p>where <span class="math notranslate nohighlight">\(x\)</span> is an input tensor with default layout as <span class="math notranslate nohighlight">\([N, C, H, W]\)</span>: [batch, channels, height, width] |
| and <span class="math notranslate nohighlight">\(y\)</span> is the output tensor of layout <span class="math notranslate nohighlight">\([N, C * (block\_size ^ 2), H / block\_size, W / block\_size]\)</span> |
| Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">6</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">7</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">8</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">12</span><span class="p">,</span> <span class="mi">18</span><span class="p">,</span> <span class="mi">13</span><span class="p">,</span> <span class="mi">19</span><span class="p">,</span> <span class="mi">14</span><span class="p">,</span> <span class="mi">20</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">9</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">11</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">15</span><span class="p">,</span> <span class="mi">21</span><span class="p">,</span> <span class="mi">16</span><span class="p">,</span> <span class="mi">22</span><span class="p">,</span> <span class="mi">17</span><span class="p">,</span> <span class="mi">23</span><span class="p">]]]]</span> |
| <span class="n">space_to_depth</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="mi">2</span><span class="p">)</span> <span class="o">=</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="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">6</span><span class="p">,</span> <span class="mi">7</span><span class="p">,</span> <span class="mi">8</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">9</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">11</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mi">12</span><span class="p">,</span> <span class="mi">13</span><span class="p">,</span> <span class="mi">14</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">15</span><span class="p">,</span> <span class="mi">16</span><span class="p">,</span> <span class="mi">17</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mi">18</span><span class="p">,</span> <span class="mi">19</span><span class="p">,</span> <span class="mi">20</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">21</span><span class="p">,</span> <span class="mi">22</span><span class="p">,</span> <span class="mi">23</span><span class="p">]]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L1018</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input ndarray</p></li> |
| <li><p><strong>block_size</strong> (<em>int</em><em>, </em><em>required</em>) – Blocks of [block_size. block_size] are moved</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.split"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">split</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">num_outputs=_Null</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">squeeze_axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.split" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Splits an array along a particular axis into multiple sub-arrays.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><code class="docutils literal notranslate"><span class="pre">SliceChannel</span></code> is deprecated. Use <code class="docutils literal notranslate"><span class="pre">split</span></code> instead.</p> |
| </div> |
| <p><strong>Note</strong> that <cite>num_outputs</cite> should evenly divide the length of the axis |
| along which to split the array.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">5.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">]]]</span> |
| <span class="n">x</span><span class="o">.</span><span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">3</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span> |
| |
| <span class="n">y</span> <span class="o">=</span> <span class="n">split</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">num_outputs</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span> <span class="o">//</span> <span class="n">a</span> <span class="nb">list</span> <span class="n">of</span> <span class="mi">2</span> <span class="n">arrays</span> <span class="k">with</span> <span class="n">shape</span> <span class="p">(</span><span class="mi">3</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="n">y</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">3.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">5.</span><span class="p">]]]</span> |
| |
| <span class="p">[[[</span> <span class="mf">2.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">4.</span><span class="p">]]</span> |
| <span class="p">[[</span> <span class="mf">6.</span><span class="p">]]]</span> |
| |
| <span class="n">y</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">3</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="n">z</span> <span class="o">=</span> <span class="n">split</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">num_outputs</span><span class="o">=</span><span class="mi">3</span><span class="p">)</span> <span class="o">//</span> <span class="n">a</span> <span class="nb">list</span> <span class="n">of</span> <span class="mi">3</span> <span class="n">arrays</span> <span class="k">with</span> <span class="n">shape</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">1</span><span class="p">)</span> |
| <span class="n">z</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]]</span> |
| |
| <span class="p">[[[</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">]]]</span> |
| |
| <span class="p">[[[</span> <span class="mf">5.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">]]]</span> |
| |
| <span class="n">z</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</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">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p><cite>squeeze_axis=1</cite> removes the axis with length 1 from the shapes of the output arrays. |
| <strong>Note</strong> that setting <cite>squeeze_axis</cite> to <code class="docutils literal notranslate"><span class="pre">1</span></code> removes axis with length 1 only |
| along the <cite>axis</cite> which it is split. |
| Also <cite>squeeze_axis</cite> can be set to true only if <code class="docutils literal notranslate"><span class="pre">input.shape[axis]</span> <span class="pre">==</span> <span class="pre">num_outputs</span></code>.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">z</span> <span class="o">=</span> <span class="n">split</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">num_outputs</span><span class="o">=</span><span class="mi">3</span><span class="p">,</span> <span class="n">squeeze_axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">//</span> <span class="n">a</span> <span class="nb">list</span> <span class="n">of</span> <span class="mi">3</span> <span class="n">arrays</span> <span class="k">with</span> <span class="n">shape</span> <span class="p">(</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span> |
| <span class="n">z</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">3.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">5.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">6.</span><span class="p">]]</span> |
| <span class="n">z</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">shape</span> <span class="o">=</span> <span class="p">(</span><span class="mi">2</span> <span class="p">,</span><span class="mi">1</span> <span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/slice_channel.cc:L106</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>num_outputs</strong> (<em>int</em><em>, </em><em>required</em>) – Number of splits. Note that this should evenly divide the length of the <cite>axis</cite>.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Axis along which to split.</p></li> |
| <li><p><strong>squeeze_axis</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If true, Removes the axis with length 1 from the shapes of the output arrays. <strong>Note</strong> that setting <cite>squeeze_axis</cite> to <code class="docutils literal notranslate"><span class="pre">true</span></code> removes axis with length 1 only along the <cite>axis</cite> which it is split. Also <cite>squeeze_axis</cite> can be set to <code class="docutils literal notranslate"><span class="pre">true</span></code> only if <code class="docutils literal notranslate"><span class="pre">input.shape[axis]</span> <span class="pre">==</span> <span class="pre">num_outputs</span></code>.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sqrt"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sqrt</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sqrt" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise square-root value of the input.</p> |
| <div class="math notranslate nohighlight"> |
| \[\textrm{sqrt}(x) = \sqrt{x}\]</div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">sqrt</span><span class="p">([</span><span class="mi">4</span><span class="p">,</span> <span class="mi">9</span><span class="p">,</span> <span class="mi">16</span><span class="p">])</span> <span class="o">=</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> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">sqrt</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>sqrt(default) = default</p></li> |
| <li><p>sqrt(row_sparse) = row_sparse</p></li> |
| <li><p>sqrt(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L170</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.square"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">square</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.square" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns element-wise squared value of the input.</p> |
| <div class="math notranslate nohighlight"> |
| \[square(x) = x^2\]</div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">square</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="o">=</span> <span class="p">[</span><span class="mi">4</span><span class="p">,</span> <span class="mi">9</span><span class="p">,</span> <span class="mi">16</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">square</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>square(default) = default</p></li> |
| <li><p>square(row_sparse) = row_sparse</p></li> |
| <li><p>square(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L119</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.squeeze"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">squeeze</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.squeeze" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Remove single-dimensional entries from the shape of an array. |
| Same behavior of defining the output tensor shape as numpy.squeeze for the most of cases. |
| See the following note for exception. |
| Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">data</span> <span class="o">=</span> <span class="p">[[[</span><span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mi">2</span><span class="p">]]]</span> |
| <span class="n">squeeze</span><span class="p">(</span><span class="n">data</span><span class="p">)</span> <span class="o">=</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="n">squeeze</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mi">2</span><span class="p">]]</span> |
| <span class="n">squeeze</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span> <span class="o">=</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="n">squeeze</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">))</span> <span class="o">=</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> |
| </pre></div> |
| </div> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The output of this operator will keep at least one dimension not removed. For example, |
| squeeze([[[4]]]) = [4], while in numpy.squeeze, the output will become a scalar.</p> |
| </div> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – data to squeeze</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Selects a subset of the single-dimensional entries in the shape. If an axis is selected with shape entry greater than one, an error is raised.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.stack"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">stack</code><span class="sig-paren">(</span><em class="sig-param">*data</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.stack" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Join a sequence of arrays along a new axis. |
| The axis parameter specifies the index of the new axis in the dimensions of the |
| result. For example, if axis=0 it will be the first dimension and if axis=-1 it |
| will be the last dimension. |
| Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="n">y</span> <span class="o">=</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="n">stack</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]]</span> |
| <span class="n">stack</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</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="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">4</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>This function support variable length of positional input.</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a><em>[</em><em>]</em>) – List of arrays to stack</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – The axis in the result array along which the input arrays are stacked.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.stop_gradient"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">stop_gradient</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.stop_gradient" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Stops gradient computation.</p> |
| <p>Stops the accumulated gradient of the inputs from flowing through this operator |
| in the backward direction. In other words, this operator prevents the contribution |
| of its inputs to be taken into account for computing gradients.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">v1</span> <span class="o">=</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="n">v2</span> <span class="o">=</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="n">a</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'a'</span><span class="p">)</span> |
| <span class="n">b</span> <span class="o">=</span> <span class="n">Variable</span><span class="p">(</span><span class="s1">'b'</span><span class="p">)</span> |
| <span class="n">b_stop_grad</span> <span class="o">=</span> <span class="n">stop_gradient</span><span class="p">(</span><span class="mi">3</span> <span class="o">*</span> <span class="n">b</span><span class="p">)</span> |
| <span class="n">loss</span> <span class="o">=</span> <span class="n">MakeLoss</span><span class="p">(</span><span class="n">b_stop_grad</span> <span class="o">+</span> <span class="n">a</span><span class="p">)</span> |
| |
| <span class="n">executor</span> <span class="o">=</span> <span class="n">loss</span><span class="o">.</span><span class="n">simple_bind</span><span class="p">(</span><span class="n">ctx</span><span class="o">=</span><span class="n">cpu</span><span class="p">(),</span> <span class="n">a</span><span class="o">=</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="n">b</span><span class="o">=</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="n">executor</span><span class="o">.</span><span class="n">forward</span><span class="p">(</span><span class="n">is_train</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">a</span><span class="o">=</span><span class="n">v1</span><span class="p">,</span> <span class="n">b</span><span class="o">=</span><span class="n">v2</span><span class="p">)</span> |
| <span class="n">executor</span><span class="o">.</span><span class="n">outputs</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">5.</span><span class="p">]</span> |
| |
| <span class="n">executor</span><span class="o">.</span><span class="n">backward</span><span class="p">()</span> |
| <span class="n">executor</span><span class="o">.</span><span class="n">grad_arrays</span> |
| <span class="p">[</span> <span class="mf">0.</span> <span class="mf">0.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">1.</span> <span class="mf">1.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L325</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sum"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sum</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sum" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the sum of array elements over given axes.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><cite>sum</cite> and <cite>sum_axis</cite> are equivalent. |
| For ndarray of csr storage type summation along axis 0 and axis 1 is supported. |
| Setting keepdims or exclude to True will cause a fallback to dense operator.</p> |
| </div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">data</span> <span class="o">=</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="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">],</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="p">[[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</span> <span class="p">[</span><span class="mi">4</span><span class="p">,</span> <span class="mi">3</span><span class="p">],</span> <span class="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">2</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">3</span><span class="p">]]]</span> |
| |
| <span class="nb">sum</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> |
| <span class="p">[[</span> <span class="mf">4.</span> <span class="mf">8.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">10.</span> <span class="mf">9.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">21.</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="nb">sum</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</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="p">[</span> <span class="mf">12.</span> <span class="mf">19.</span> <span class="mf">27.</span><span class="p">]</span> |
| |
| <span class="n">data</span> <span class="o">=</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">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">3</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="p">[</span><span class="mi">4</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]</span> |
| |
| <span class="n">csr</span> <span class="o">=</span> <span class="n">cast_storage</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="s1">'csr'</span><span class="p">)</span> |
| |
| <span class="nb">sum</span><span class="p">(</span><span class="n">csr</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> |
| <span class="p">[</span> <span class="mf">8.</span> <span class="mf">3.</span> <span class="mf">1.</span><span class="p">]</span> |
| |
| <span class="nb">sum</span><span class="p">(</span><span class="n">csr</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> |
| <span class="p">[</span> <span class="mf">3.</span> <span class="mf">4.</span> <span class="mf">5.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_sum_value.cc:L66</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.sum_axis"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">sum_axis</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">keepdims=_Null</em>, <em class="sig-param">exclude=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.sum_axis" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the sum of array elements over given axes.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p><cite>sum</cite> and <cite>sum_axis</cite> are equivalent. |
| For ndarray of csr storage type summation along axis 0 and axis 1 is supported. |
| Setting keepdims or exclude to True will cause a fallback to dense operator.</p> |
| </div> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">data</span> <span class="o">=</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="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">],</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="p">[[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</span> <span class="p">[</span><span class="mi">4</span><span class="p">,</span> <span class="mi">3</span><span class="p">],</span> <span class="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">2</span><span class="p">]],</span> |
| <span class="p">[[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">3</span><span class="p">]]]</span> |
| |
| <span class="nb">sum</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> |
| <span class="p">[[</span> <span class="mf">4.</span> <span class="mf">8.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">10.</span> <span class="mf">9.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">21.</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="nb">sum</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</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="p">[</span> <span class="mf">12.</span> <span class="mf">19.</span> <span class="mf">27.</span><span class="p">]</span> |
| |
| <span class="n">data</span> <span class="o">=</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">0</span><span class="p">],</span> |
| <span class="p">[</span><span class="mi">3</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="p">[</span><span class="mi">4</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]</span> |
| |
| <span class="n">csr</span> <span class="o">=</span> <span class="n">cast_storage</span><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="s1">'csr'</span><span class="p">)</span> |
| |
| <span class="nb">sum</span><span class="p">(</span><span class="n">csr</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> |
| <span class="p">[</span> <span class="mf">8.</span> <span class="mf">3.</span> <span class="mf">1.</span><span class="p">]</span> |
| |
| <span class="nb">sum</span><span class="p">(</span><span class="n">csr</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> |
| <span class="p">[</span> <span class="mf">3.</span> <span class="mf">4.</span> <span class="mf">5.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/broadcast_reduce_sum_value.cc:L66</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>axis</strong> (<em>Shape</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – <p>The axis or axes along which to perform the reduction.</p> |
| <blockquote> |
| <div><p>The default, <cite>axis=()</cite>, will compute over all elements into a |
| scalar array with shape <cite>(1,)</cite>.</p> |
| <p>If <cite>axis</cite> is int, a reduction is performed on a particular axis.</p> |
| <p>If <cite>axis</cite> is a tuple of ints, a reduction is performed on all the axes |
| specified in the tuple.</p> |
| <p>If <cite>exclude</cite> is true, reduction will be performed on the axes that are |
| NOT in axis instead.</p> |
| <p>Negative values means indexing from right to left.</p> |
| </div></blockquote> |
| </p></li> |
| <li><p><strong>keepdims</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – If this is set to <cite>True</cite>, the reduced axes are left in the result as dimension with size one.</p></li> |
| <li><p><strong>exclude</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to perform reduction on axis that are NOT in axis instead.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.swapaxes"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">swapaxes</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">dim1=_Null</em>, <em class="sig-param">dim2=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.swapaxes" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Interchanges two axes of an array.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span> <span class="n">x</span> <span class="o">=</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="n">swapaxes</span><span class="p">(</span><span class="n">x</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="o">=</span> <span class="p">[[</span> <span class="mi">1</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">2</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">3</span><span class="p">]]</span> |
| |
| <span class="n">x</span> <span class="o">=</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="p">[</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">]],</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">6</span><span class="p">,</span> <span class="mi">7</span><span class="p">]]]</span> <span class="o">//</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="n">array</span> |
| |
| <span class="n">swapaxes</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">6</span><span class="p">]],</span> |
| <span class="p">[[</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">5</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">7</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/swapaxis.cc:L69</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input array.</p></li> |
| <li><p><strong>dim1</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – the first axis to be swapped.</p></li> |
| <li><p><strong>dim2</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – the second axis to be swapped.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.take"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">take</code><span class="sig-paren">(</span><em class="sig-param">a=None</em>, <em class="sig-param">indices=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">mode=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.take" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Takes elements from an input array along the given axis.</p> |
| <p>This function slices the input array along a particular axis with the provided indices.</p> |
| <p>Given data tensor of rank r >= 1, and indices tensor of rank q, gather entries of the axis |
| dimension of data (by default outer-most one as axis=0) indexed by indices, and concatenates them |
| in an output tensor of rank q + (r - 1).</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[</span><span class="mf">4.</span> <span class="mf">5.</span> <span class="mf">6.</span><span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">Trivial</span> <span class="n">case</span><span class="p">,</span> <span class="n">take</span> <span class="n">the</span> <span class="n">second</span> <span class="n">element</span> <span class="n">along</span> <span class="n">the</span> <span class="n">first</span> <span class="n">axis</span><span class="o">.</span> |
| |
| <span class="n">take</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="p">[</span><span class="mi">1</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">5.</span> <span class="p">]</span> |
| |
| <span class="o">//</span> <span class="n">The</span> <span class="n">other</span> <span class="n">trivial</span> <span class="n">case</span><span class="p">,</span> <span class="n">axis</span><span class="o">=-</span><span class="mi">1</span><span class="p">,</span> <span class="n">take</span> <span class="n">the</span> <span class="n">third</span> <span class="n">element</span> <span class="n">along</span> <span class="n">the</span> <span class="n">first</span> <span class="n">axis</span> |
| |
| <span class="n">take</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="p">[</span><span class="mi">3</span><span class="p">],</span> <span class="n">axis</span><span class="o">=-</span><span class="mi">1</span><span class="p">,</span> <span class="n">mode</span><span class="o">=</span><span class="s1">'clip'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[</span> <span class="mf">6.</span> <span class="p">]</span> |
| |
| <span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">In</span> <span class="n">this</span> <span class="n">case</span> <span class="n">we</span> <span class="n">will</span> <span class="n">get</span> <span class="n">rows</span> <span class="mi">0</span> <span class="ow">and</span> <span class="mi">1</span><span class="p">,</span> <span class="n">then</span> <span class="mi">1</span> <span class="ow">and</span> <span class="mf">2.</span> <span class="n">Along</span> <span class="n">axis</span> <span class="mi">0</span> |
| |
| <span class="n">take</span><span class="p">(</span><span class="n">x</span><span class="p">,</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">1</span><span class="p">,</span><span class="mi">2</span><span class="p">]])</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]],</span> |
| |
| <span class="p">[[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]]]</span> |
| |
| <span class="o">//</span> <span class="n">In</span> <span class="n">this</span> <span class="n">case</span> <span class="n">we</span> <span class="n">will</span> <span class="n">get</span> <span class="n">rows</span> <span class="mi">0</span> <span class="ow">and</span> <span class="mi">1</span><span class="p">,</span> <span class="n">then</span> <span class="mi">1</span> <span class="ow">and</span> <span class="mi">2</span> <span class="p">(</span><span class="n">calculated</span> <span class="n">by</span> <span class="n">wrapping</span> <span class="n">around</span><span class="p">)</span><span class="o">.</span> |
| <span class="o">//</span> <span class="n">Along</span> <span class="n">axis</span> <span class="mi">1</span> |
| |
| <span class="n">take</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">3</span><span class="p">],</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">2</span><span class="p">]],</span> <span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">mode</span><span class="o">=</span><span class="s1">'wrap'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span> <span class="mf">2.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">2.</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">3.</span> <span class="mf">4.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">4.</span> <span class="mf">3.</span><span class="p">]]</span> |
| |
| <span class="p">[[</span> <span class="mf">5.</span> <span class="mf">6.</span><span class="p">]</span> |
| <span class="p">[</span> <span class="mf">6.</span> <span class="mf">5.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">take</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>take(default, default) = default</p></li> |
| <li><p>take(csr, default, axis=0) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/indexing_op.cc:L776</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>a</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>indices</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The indices of the values to be extracted.</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='0'</em>) – The axis of input array to be taken.For input tensor of rank r, it could be in the range of [-r, r-1]</p></li> |
| <li><p><strong>mode</strong> (<em>{'clip'</em><em>, </em><em>'raise'</em><em>, </em><em>'wrap'}</em><em>,</em><em>optional</em><em>, </em><em>default='clip'</em>) – Specify how out-of-bound indices bahave. Default is “clip”. “clip” means clip to the range. So, if all indices mentioned are too large, they are replaced by the index that addresses the last element along an axis. “wrap” means to wrap around. “raise” means to raise an error when index out of range.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.tan"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">tan</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.tan" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Computes the element-wise tangent of the input array.</p> |
| <p>The input should be in radians (<span class="math notranslate nohighlight">\(2\pi\)</span> rad equals 360 degrees).</p> |
| <div class="math notranslate nohighlight"> |
| \[tan([0, \pi/4, \pi/2]) = [0, 1, -inf]\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">tan</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>tan(default) = default</p></li> |
| <li><p>tan(row_sparse) = row_sparse</p></li> |
| <li><p>tan(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L140</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.tanh"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">tanh</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.tanh" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Returns the hyperbolic tangent of the input array, computed element-wise.</p> |
| <div class="math notranslate nohighlight"> |
| \[tanh(x) = sinh(x) / cosh(x)\]</div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">tanh</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>tanh(default) = default</p></li> |
| <li><p>tanh(row_sparse) = row_sparse</p></li> |
| <li><p>tanh(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L451</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.tile"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">tile</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">reps=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.tile" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Repeats the whole array multiple times. |
| If <code class="docutils literal notranslate"><span class="pre">reps</span></code> has length <em>d</em>, and input array has dimension of <em>n</em>. There are |
| three cases: |
| - <strong>n=d</strong>. Repeat <em>i</em>-th dimension of the input by <code class="docutils literal notranslate"><span class="pre">reps[i]</span></code> times:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]]</span> |
| <span class="n">tile</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">reps</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <ul> |
| <li><p><strong>n>d</strong>. <code class="docutils literal notranslate"><span class="pre">reps</span></code> is promoted to length <em>n</em> by pre-pending 1’s to it. Thus for |
| an input shape <code class="docutils literal notranslate"><span class="pre">(2,3)</span></code>, <code class="docutils literal notranslate"><span class="pre">repos=(2,)</span></code> is treated as <code class="docutils literal notranslate"><span class="pre">(1,2)</span></code>:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">tile</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">reps</span><span class="o">=</span><span class="p">(</span><span class="mi">2</span><span class="p">,))</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| </li> |
| <li><p><strong>n<d</strong>. The input is promoted to be d-dimensional by prepending new axes. So a |
| shape <code class="docutils literal notranslate"><span class="pre">(2,2)</span></code> array is promoted to <code class="docutils literal notranslate"><span class="pre">(1,2,2)</span></code> for 3-D replication:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">tile</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">reps</span><span class="o">=</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">3</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]],</span> |
| <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L795</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data array</p></li> |
| <li><p><strong>reps</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>required</em>) – The number of times for repeating the tensor a. Each dim size of reps must be a positive integer. If reps has length d, the result will have dimension of max(d, a.ndim); If a.ndim < d, a is promoted to be d-dimensional by prepending new axes. If a.ndim > d, reps is promoted to a.ndim by pre-pending 1’s to it.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.topk"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">topk</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axis=_Null</em>, <em class="sig-param">k=_Null</em>, <em class="sig-param">ret_typ=_Null</em>, <em class="sig-param">is_ascend=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.topk" title="Permalink to this definition">¶</a></dt> |
| <dd><dl class="simple"> |
| <dt>Returns the indices of the top <em>k</em> elements in an input array along the given</dt><dd><p>axis (by default). |
| If ret_type is set to ‘value’ returns the value of top <em>k</em> elements (instead of indices). |
| In case of ret_type = ‘both’, both value and index would be returned. |
| The returned elements will be sorted.</p> |
| </dd> |
| </dl> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.4</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">returns</span> <span class="n">an</span> <span class="n">index</span> <span class="n">of</span> <span class="n">the</span> <span class="n">largest</span> <span class="n">element</span> <span class="n">on</span> <span class="n">last</span> <span class="n">axis</span> |
| <span class="n">topk</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">returns</span> <span class="n">the</span> <span class="n">value</span> <span class="n">of</span> <span class="n">top</span><span class="o">-</span><span class="mi">2</span> <span class="n">largest</span> <span class="n">elements</span> <span class="n">on</span> <span class="n">last</span> <span class="n">axis</span> |
| <span class="n">topk</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">ret_typ</span><span class="o">=</span><span class="s1">'value'</span><span class="p">,</span> <span class="n">k</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.4</span><span class="p">,</span> <span class="mf">0.3</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">returns</span> <span class="n">the</span> <span class="n">value</span> <span class="n">of</span> <span class="n">top</span><span class="o">-</span><span class="mi">2</span> <span class="n">smallest</span> <span class="n">elements</span> <span class="n">on</span> <span class="n">last</span> <span class="n">axis</span> |
| <span class="n">topk</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">ret_typ</span><span class="o">=</span><span class="s1">'value'</span><span class="p">,</span> <span class="n">k</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">is_ascend</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.2</span> <span class="p">,</span> <span class="mf">0.3</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.1</span> <span class="p">,</span> <span class="mf">0.2</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">returns</span> <span class="n">the</span> <span class="n">value</span> <span class="n">of</span> <span class="n">top</span><span class="o">-</span><span class="mi">2</span> <span class="n">largest</span> <span class="n">elements</span> <span class="n">on</span> <span class="n">axis</span> <span class="mi">0</span> |
| <span class="n">topk</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">ret_typ</span><span class="o">=</span><span class="s1">'value'</span><span class="p">,</span> <span class="n">k</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.4</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">]]</span> |
| |
| <span class="o">//</span> <span class="n">flattens</span> <span class="ow">and</span> <span class="n">then</span> <span class="n">returns</span> <span class="nb">list</span> <span class="n">of</span> <span class="n">both</span> <span class="n">values</span> <span class="ow">and</span> <span class="n">indices</span> |
| <span class="n">topk</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">ret_typ</span><span class="o">=</span><span class="s1">'both'</span><span class="p">,</span> <span class="n">k</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">0.4</span><span class="p">,</span> <span class="mf">0.3</span><span class="p">],</span> <span class="p">[</span> <span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">]]</span> <span class="p">,</span> <span class="p">[[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/ordering_op.cc:L67</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array</p></li> |
| <li><p><strong>axis</strong> (<em>int</em><em> or </em><em>None</em><em>, </em><em>optional</em><em>, </em><em>default='-1'</em>) – Axis along which to choose the top k indices. If not given, the flattened array is used. Default is -1.</p></li> |
| <li><p><strong>k</strong> (<em>int</em><em>, </em><em>optional</em><em>, </em><em>default='1'</em>) – Number of top elements to select, should be always smaller than or equal to the element number in the given axis. A global sort is performed if set k < 1.</p></li> |
| <li><p><strong>ret_typ</strong> (<em>{'both'</em><em>, </em><em>'indices'</em><em>, </em><em>'mask'</em><em>, </em><em>'value'}</em><em>,</em><em>optional</em><em>, </em><em>default='indices'</em>) – The return type. |
| “value” means to return the top k values, “indices” means to return the indices of the top k values, “mask” means to return a mask array containing 0 and 1. 1 means the top k values. “both” means to return a list of both values and indices of top k elements.</p></li> |
| <li><p><strong>is_ascend</strong> (<em>boolean</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Whether to choose k largest or k smallest elements. Top K largest elements will be chosen if set to false.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'</em><em>, </em><em>'int32'</em><em>, </em><em>'int64'</em><em>, </em><em>'uint8'}</em><em>,</em><em>optional</em><em>, </em><em>default='float32'</em>) – DType of the output indices when ret_typ is “indices” or “both”. An error will be raised if the selected data type cannot precisely represent the indices.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.transpose"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">transpose</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">axes=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.transpose" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Permutes the dimensions of an array. |
| Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="p">[</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]]</span> |
| <span class="n">transpose</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]]</span> |
| <span class="n">x</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">]],</span> |
| <span class="p">[[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]]</span> |
| <span class="n">transpose</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">5.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">7.</span><span class="p">]],</span> |
| <span class="p">[[</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">4.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]]</span> |
| <span class="n">transpose</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">axes</span><span class="o">=</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span><span class="mi">0</span><span class="p">,</span><span class="mi">2</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">5.</span><span class="p">,</span> <span class="mf">6.</span><span class="p">]],</span> |
| <span class="p">[[</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">7.</span><span class="p">,</span> <span class="mf">8.</span><span class="p">]]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/matrix_op.cc:L327</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Source input</p></li> |
| <li><p><strong>axes</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=</em><em>[</em><em>]</em>) – Target axis order. By default the axes will be inverted.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.trunc"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">trunc</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.trunc" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Return the element-wise truncated value of the input.</p> |
| <p>The truncated value of the scalar x is the nearest integer i which is closer to |
| zero than x is. In short, the fractional part of the signed number x is discarded.</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">trunc</span><span class="p">([</span><span class="o">-</span><span class="mf">2.1</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.9</span><span class="p">,</span> <span class="mf">1.5</span><span class="p">,</span> <span class="mf">1.9</span><span class="p">,</span> <span class="mf">2.1</span><span class="p">])</span> <span class="o">=</span> <span class="p">[</span><span class="o">-</span><span class="mf">2.</span><span class="p">,</span> <span class="o">-</span><span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">trunc</span></code> output depends upon the input storage type:</p> |
| <blockquote> |
| <div><ul class="simple"> |
| <li><p>trunc(default) = default</p></li> |
| <li><p>trunc(row_sparse) = row_sparse</p></li> |
| <li><p>trunc(csr) = csr</p></li> |
| </ul> |
| </div></blockquote> |
| <p>Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L856</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input array.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.uniform"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">uniform</code><span class="sig-paren">(</span><em class="sig-param">low=_Null</em>, <em class="sig-param">high=_Null</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">ctx=_Null</em>, <em class="sig-param">dtype=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.uniform" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Draw random samples from a uniform distribution.</p> |
| <div class="admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>The existing alias <code class="docutils literal notranslate"><span class="pre">uniform</span></code> is deprecated.</p> |
| </div> |
| <p>Samples are uniformly distributed over the half-open interval <em>[low, high)</em> |
| (includes <em>low</em>, but excludes <em>high</em>).</p> |
| <p>Example:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">uniform</span><span class="p">(</span><span class="n">low</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span> <span class="n">high</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</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="o">=</span> <span class="p">[[</span> <span class="mf">0.60276335</span><span class="p">,</span> <span class="mf">0.85794562</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.54488319</span><span class="p">,</span> <span class="mf">0.84725171</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/random/sample_op.cc:L95</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>low</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=0</em>) – Lower bound of the distribution.</p></li> |
| <li><p><strong>high</strong> (<em>float</em><em>, </em><em>optional</em><em>, </em><em>default=1</em>) – Upper bound of the distribution.</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the output.</p></li> |
| <li><p><strong>ctx</strong> (<em>string</em><em>, </em><em>optional</em><em>, </em><em>default=''</em>) – Context of output, in format [cpu|gpu|cpu_pinned](n). Only used for imperative calls.</p></li> |
| <li><p><strong>dtype</strong> (<em>{'None'</em><em>, </em><em>'float16'</em><em>, </em><em>'float32'</em><em>, </em><em>'float64'}</em><em>,</em><em>optional</em><em>, </em><em>default='None'</em>) – DType of the output in case this can’t be inferred. Defaults to float32 if not defined (dtype=None).</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.unravel_index"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">unravel_index</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">shape=_Null</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.unravel_index" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Converts an array of flat indices into a batch of index arrays. The operator follows numpy conventions so a single multi index is given by a column of the output matrix. The leading dimension may be left unspecified by using -1 as placeholder.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">A</span> <span class="o">=</span> <span class="p">[</span><span class="mi">22</span><span class="p">,</span><span class="mi">41</span><span class="p">,</span><span class="mi">37</span><span class="p">]</span> |
| <span class="n">unravel</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">7</span><span class="p">,</span><span class="mi">6</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">3</span><span class="p">,</span><span class="mi">6</span><span class="p">,</span><span class="mi">6</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="mi">1</span><span class="p">]]</span> |
| <span class="n">unravel</span><span class="p">(</span><span class="n">A</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="o">-</span><span class="mi">1</span><span class="p">,</span><span class="mi">6</span><span class="p">))</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">3</span><span class="p">,</span><span class="mi">6</span><span class="p">,</span><span class="mi">6</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="mi">1</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/ravel.cc:L67</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Array of flat indices</p></li> |
| <li><p><strong>shape</strong> (<em>Shape</em><em>(</em><em>tuple</em><em>)</em><em>, </em><em>optional</em><em>, </em><em>default=None</em>) – Shape of the array into which the multi-indices apply.</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.where"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">where</code><span class="sig-paren">(</span><em class="sig-param">condition=None</em>, <em class="sig-param">x=None</em>, <em class="sig-param">y=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.where" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Return the elements, either from x or y, depending on the condition.</p> |
| <p>Given three ndarrays, condition, x, and y, return an ndarray with the elements from x or y, |
| depending on the elements from condition are true or false. x and y must have the same shape. |
| If condition has the same shape as x, each element in the output array is from x if the |
| corresponding element in the condition is true, and from y if false.</p> |
| <p>If condition does not have the same shape as x, it must be a 1D array whose size is |
| the same as x’s first dimension size. Each row of the output array is from x’s row |
| if the corresponding element from condition is true, and from y’s row if false.</p> |
| <p>Note that all non-zero values are interpreted as <code class="docutils literal notranslate"><span class="pre">True</span></code> in condition.</p> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</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="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]]</span> |
| <span class="n">y</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">8</span><span class="p">]]</span> |
| <span class="n">cond</span> <span class="o">=</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="p">[</span><span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">]]</span> |
| |
| <span class="n">where</span><span class="p">(</span><span class="n">cond</span><span class="p">,</span> <span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">8</span><span class="p">]]</span> |
| |
| <span class="n">csr_cond</span> <span class="o">=</span> <span class="n">cast_storage</span><span class="p">(</span><span class="n">cond</span><span class="p">,</span> <span class="s1">'csr'</span><span class="p">)</span> |
| |
| <span class="n">where</span><span class="p">(</span><span class="n">csr_cond</span><span class="p">,</span> <span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">8</span><span class="p">]]</span> |
| </pre></div> |
| </div> |
| <p>Defined in src/operator/tensor/control_flow_op.cc:L56</p> |
| <dl class="field-list simple"> |
| <dt class="field-odd">Parameters</dt> |
| <dd class="field-odd"><ul class="simple"> |
| <li><p><strong>condition</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – condition array</p></li> |
| <li><p><strong>x</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – </p></li> |
| <li><p><strong>y</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – </p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| <dl class="function"> |
| <dt id="mxnet.symbol.op.zeros_like"> |
| <code class="sig-prename descclassname">mxnet.symbol.op.</code><code class="sig-name descname">zeros_like</code><span class="sig-paren">(</span><em class="sig-param">data=None</em>, <em class="sig-param">name=None</em>, <em class="sig-param">attr=None</em>, <em class="sig-param">out=None</em>, <em class="sig-param">**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.symbol.op.zeros_like" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Return an array of zeros with the same shape, type and storage type |
| as the input array.</p> |
| <p>The storage type of <code class="docutils literal notranslate"><span class="pre">zeros_like</span></code> output depends on the storage type of the input</p> |
| <ul class="simple"> |
| <li><p>zeros_like(row_sparse) = row_sparse</p></li> |
| <li><p>zeros_like(csr) = csr</p></li> |
| <li><p>zeros_like(default) = default</p></li> |
| </ul> |
| <p>Examples:</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">]]</span> |
| |
| <span class="n">zeros_like</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">],</span> |
| <span class="p">[</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</span><span class="p">,</span> <span class="mf">0.</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>data</strong> (<a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – The input</p></li> |
| <li><p><strong>name</strong> (<em>string</em><em>, </em><em>optional.</em>) – Name of the resulting symbol.</p></li> |
| </ul> |
| </dd> |
| <dt class="field-even">Returns</dt> |
| <dd class="field-even"><p>The result symbol.</p> |
| </dd> |
| <dt class="field-odd">Return type</dt> |
| <dd class="field-odd"><p><a class="reference internal" href="../symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol">Symbol</a></p> |
| </dd> |
| </dl> |
| </dd></dl> |
| |
| </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> |
| </div> |
| |
| <div class="clearer"></div> |
| </div><div class="pagenation"> |
| <a id="button-prev" href="../linalg/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>symbol.linalg</div> |
| </div> |
| </a> |
| <a id="button-next" href="../random/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>symbol.random</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 class="u-email" href="mailto:dev@mxnet.apache.org">Dev list</a></li> |
| <li><a class="u-email" href="mailto:user@mxnet.apache.org">User mailing list</a></li> |
| <li><a href="https://cwiki.apache.org/confluence/display/MXNET/Apache+MXNet+Home">Developer Wiki</a></li> |
| <li><a href="https://issues.apache.org/jira/projects/MXNET/issues">Jira Tracker</a></li> |
| <li><a href="https://github.com/apache/mxnet/labels/Roadmap">Github Roadmap</a></li> |
| <li><a href="https://medium.com/apache-mxnet">Blog</a></li> |
| <li><a href="https://discuss.mxnet.io">Forum</a></li> |
| <li><a href="/community/contribute">Contribute</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> |
