| <!DOCTYPE html> |
| |
| <!--- |
| Licensed to the Apache Software Foundation (ASF) under one |
| or more contributor license agreements. See the NOTICE file |
| distributed with this work for additional information |
| regarding copyright ownership. The ASF licenses this file |
| to you under the Apache License, Version 2.0 (the |
| "License"); you may not use this file except in compliance |
| with the License. You may obtain a copy of the License at |
| http://www.apache.org/licenses/LICENSE-2.0 |
| Unless required by applicable law or agreed to in writing, |
| software distributed under the License is distributed on an |
| "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY |
| KIND, either express or implied. See the License for the |
| specific language governing permissions and limitations |
| under the License. |
| --> |
| |
| <html lang=" en"><head> |
| <meta charset="utf-8"> |
| <meta http-equiv="X-UA-Compatible" content="IE=edge"> |
| <meta name="viewport" content="width=device-width, initial-scale=1"> |
| <link href="/versions/1.9.1/assets/img/mxnet-icon.png" rel="icon" type="image/png"><!-- Begin Jekyll SEO tag v2.6.1 --> |
| <title>Convert from Caffe to MXNet | Apache MXNet</title> |
| <meta name="generator" content="Jekyll v3.8.6" /> |
| <meta property="og:title" content="Convert from Caffe to MXNet" /> |
| <meta property="og:locale" content="en_US" /> |
| <meta name="description" content="A flexible and efficient library for deep learning." /> |
| <meta property="og:description" content="A flexible and efficient library for deep learning." /> |
| <link rel="canonical" href="https://mxnet.apache.org/versions/1.9.1/api/faq/caffe" /> |
| <meta property="og:url" content="https://mxnet.apache.org/versions/1.9.1/api/faq/caffe" /> |
| <meta property="og:site_name" content="Apache MXNet" /> |
| <script type="application/ld+json"> |
| {"description":"A flexible and efficient library for deep learning.","headline":"Convert from Caffe to MXNet","@type":"WebPage","url":"https://mxnet.apache.org/versions/1.9.1/api/faq/caffe","@context":"https://schema.org"}</script> |
| <!-- End Jekyll SEO tag --> |
| <link rel="stylesheet" href="/versions/1.9.1/assets/docsearch.min.css" /><link rel="stylesheet" href="/versions/1.9.1/assets/main.css"><link type="application/atom+xml" rel="alternate" href="https://mxnet.apache.org/versions/1.9.1/feed.xml" title="Apache MXNet" /><!-- Matomo --> |
| <script> |
| var _paq = window._paq = window._paq || []; |
| /* tracker methods like "setCustomDimension" should be called before "trackPageView" */ |
| /* We explicitly disable cookie tracking to avoid privacy issues */ |
| _paq.push(['disableCookies']); |
| _paq.push(['trackPageView']); |
| _paq.push(['enableLinkTracking']); |
| (function() { |
| var u="https://analytics.apache.org/"; |
| _paq.push(['setTrackerUrl', u+'matomo.php']); |
| _paq.push(['setSiteId', '23']); |
| var d=document, g=d.createElement('script'), s=d.getElementsByTagName('script')[0]; |
| g.async=true; g.src=u+'matomo.js'; s.parentNode.insertBefore(g,s); |
| })(); |
| </script> |
| <!-- End Matomo Code --> |
| |
| <script src="/versions/1.9.1/assets/js/jquery-3.3.1.min.js"></script> |
| <script src="/versions/1.9.1/assets/js/docsearch.min.js"></script><script src="/versions/1.9.1/assets/js/globalSearch.js" defer></script> |
| <script src="/versions/1.9.1/assets/js/clipboard.js" defer></script> |
| <script src="/versions/1.9.1/assets/js/copycode.js" defer></script></head> |
| <body><header class="site-header" role="banner"> |
| |
| <script> |
| $(document).ready(function () { |
| |
| // HEADER OPACITY LOGIC |
| |
| function opacity_header() { |
| var value = "rgba(4,140,204," + ($(window).scrollTop() / 300 + 0.4) + ")" |
| $('.site-header').css("background-color", value) |
| } |
| |
| $(window).scroll(function () { |
| opacity_header() |
| }) |
| opacity_header(); |
| |
| // MENU SELECTOR LOGIC |
| $('.page-link').each( function () { |
| if (window.location.href.includes(this.href)) { |
| $(this).addClass("page-current"); |
| } |
| }); |
| }) |
| </script> |
| <div class="wrapper"> |
| <a class="site-title" rel="author" href="/versions/1.9.1/"><img |
| src="/versions/1.9.1/assets/img/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="gs-search-border"> |
| <div id="gs-search-icon"></div> |
| <form id="global-search-form"> |
| <input id="global-search" type="text" title="Search" placeholder="Search" /> |
| <div id="global-search-dropdown-container"> |
| <button class="gs-current-version btn" type="button" data-toggle="dropdown"> |
| <span id="gs-current-version-label">1.9.1</span> |
| <svg class="gs-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> |
| </button> |
| <ul class="gs-opt-group gs-version-dropdown"> |
| |
| |
| <li class="gs-opt gs-versions">master</li> |
| |
| |
| |
| <li class="gs-opt gs-versions active">1.9.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.8.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.7.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.6.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.5.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.4.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.3.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.2.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.1.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.0.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">0.12.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">0.11.0</li> |
| |
| |
| </ul> |
| </div> |
| <span id="global-search-close">x</span> |
| </form> |
| </div> |
| <div class="trigger"> |
| <div id="global-search-mobile-border"> |
| <div id="gs-search-icon-mobile"></div> |
| <input id="global-search-mobile" placeholder="Search..." type="text"/> |
| <div id="global-search-dropdown-container-mobile"> |
| <button class="gs-current-version-mobile btn" type="button" data-toggle="dropdown"> |
| <svg class="gs-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> |
| </button> |
| <ul class="gs-opt-group gs-version-dropdown-mobile"> |
| |
| |
| <li class="gs-opt gs-versions">master</li> |
| |
| |
| |
| <li class="gs-opt gs-versions active">1.9.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.8.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.7.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.6.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.5.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.4.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.3.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.2.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.1.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">1.0.0</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">0.12.1</li> |
| |
| |
| |
| <li class="gs-opt gs-versions">0.11.0</li> |
| |
| |
| </ul> |
| </div> |
| </div> |
| <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" 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://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 href="/">master</a> |
| <a class="dropdown-option-active" href="/versions/1.9.1/">1.9.1</a> |
| <a href="/versions/1.8.0/">1.8.0</a> |
| <a href="/versions/1.7.0/">1.7.0</a> |
| <a href="/versions/1.6.0/">1.6.0</a> |
| <a href="/versions/1.5.0/">1.5.0</a> |
| <a href="/versions/1.4.1/">1.4.1</a> |
| <a href="/versions/1.3.1/">1.3.1</a> |
| <a href="/versions/1.2.1/">1.2.1</a> |
| <a href="/versions/1.1.0/">1.1.0</a> |
| <a href="/versions/1.0.0/">1.0.0</a> |
| <a href="/versions/0.12.1/">0.12.1</a> |
| <a href="/versions/0.11.0/">0.11.0</a> |
| </div> |
| </div> |
| </div> |
| </nav> |
| </div> |
| </header> |
| <main class="page-content" aria-label="Content"> |
| <script> |
| |
| </script> |
| <article class="post"> |
| |
| <header class="post-header wrapper"> |
| <h1 class="post-title">Convert from Caffe to MXNet</h1> |
| <h3></h3></header> |
| |
| <div class="post-content"> |
| <div class="wrapper"> |
| <div class="row"> |
| <div class="col-3 docs-side-bar"> |
| <h3 style="text-transform: capitalize; padding-left:10px">faq</h3> |
| <ul> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/add_op_in_backend">A Beginner's Guide to Implementing Operators in MXNet Backend</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/caffe">Convert from Caffe to MXNet</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/cloud">MXNet on the Cloud</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/distributed_training">Distributed Training in MXNet</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/env_var">Environment Variables</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/float16">Float16</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/large_tensor_support">Using MXNet with Large Tensor Support</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/model_parallel_lstm">Model Parallel</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/multi_device">Data Parallelism with Multiple CPU/GPUs on MXNet</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/new_op">Create New Operators</a></li> |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/nnpack">NNPACK for Multi-Core CPU Support in MXNet</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/perf">Some Tips for Improving MXNet Performance</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/recordio">Create a Dataset Using RecordIO</a></li> |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/s3_integration">Use data from S3 for training</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/security">MXNet Security Best Practices</a></li> |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/smart_device">Deep Learning at the Edge</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/visualize_graph">Visualize Neural Networks</a></li> |
| <!-- page-category --> |
| |
| |
| <li><a href="/versions/1.9.1/api/faq/why_mxnet">Why MXNet came to be?</a></li> |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| |
| <!-- page-category --> |
| <!-- resource-p --> |
| </ul> |
| </div> |
| <div class="col-9"> |
| <!--- Licensed to the Apache Software Foundation (ASF) under one --> |
| |
| <!--- or more contributor license agreements. See the NOTICE file --> |
| |
| <!--- distributed with this work for additional information --> |
| |
| <!--- regarding copyright ownership. The ASF licenses this file --> |
| |
| <!--- to you under the Apache License, Version 2.0 (the --> |
| |
| <!--- "License"); you may not use this file except in compliance --> |
| |
| <!--- with the License. You may obtain a copy of the License at --> |
| |
| <!--- http://www.apache.org/licenses/LICENSE-2.0 --> |
| |
| <!--- Unless required by applicable law or agreed to in writing, --> |
| |
| <!--- software distributed under the License is distributed on an --> |
| |
| <!--- "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY --> |
| |
| <!--- KIND, either express or implied. See the License for the --> |
| |
| <!--- specific language governing permissions and limitations --> |
| |
| <!--- under the License. --> |
| |
| <h1 id="how-to-convert-from-caffe-to-mxnet">How to | Convert from Caffe to MXNet</h1> |
| |
| <p>Key topics covered include the following:</p> |
| |
| <ul> |
| <li><a href="#converting-caffe-trained-models-to-mxnet">Converting Caffe trained models to MXNet</a></li> |
| <li><a href="#calling-caffe-operators-in-mxnet">Calling Caffe operators in MXNet</a></li> |
| </ul> |
| |
| <h2 id="converting-caffe-trained-models-to-mxnet">Converting Caffe trained models to MXNet</h2> |
| |
| <p>The converting tool is available at |
| <a href="https://github.com/apache/mxnet/tree/v1.x/tools/caffe_converter">tools/caffe_converter</a>. On |
| the remaining of this section, we assume we are on the <code>tools/caffe_converter</code> |
| directory.</p> |
| |
| <h3 id="how-to-build">How to build</h3> |
| |
| <p>If Caffe's python package is installed, namely we can run <code>import caffe</code> in |
| python, then we are ready to go.</p> |
| |
| <p>For example, we can used |
| <a href="https://aws.amazon.com/marketplace/pp/B06VSPXKDX">AWS Deep Learning AMI</a> with |
| both Caffe and MXNet installed.</p> |
| |
| <p>Otherwise we can install the |
| <a href="https://developers.google.com/protocol-buffers/?hl=en">Google protobuf</a> |
| compiler and its python binding. It is easier to install, but may be slower |
| during running.</p> |
| |
| <ol> |
| <li><p>Install the compiler:</p> |
| |
| <ul> |
| <li>Linux: install <code>protobuf-compiler</code> e.g. <code>sudo apt-get install |
| protobuf-compiler</code> for Ubuntu and <code>sudo yum install protobuf-compiler</code> for |
| Redhat/Fedora.</li> |
| <li>Windows: Download the win32 build of |
| <a href="https://github.com/google/protobuf/releases">protobuf</a>. Make sure to |
| download the version that corresponds to the version of the python binding |
| on the next step. Extract to any location then add that location to your |
| <code>PATH</code></li> |
| <li>Mac OS X: <code>brew install protobuf</code></li> |
| </ul></li> |
| <li><p>Install the python binding by either <code>conda install -c conda-forge protobuf</code> |
| or <code>pip install protobuf</code>.</p></li> |
| <li><p>Compile Caffe proto definition. Run <code>make</code> in Linux or Mac OS X, or |
| <code>make_win32.bat</code> in Windows</p></li> |
| </ol> |
| |
| <h3 id="how-to-use">How to use</h3> |
| |
| <p>There are three tools:</p> |
| |
| <ul> |
| <li><code>convert_symbol.py</code> : convert Caffe model definition in protobuf into MXNet's |
| Symbol in JSON format.</li> |
| <li><code>convert_model.py</code> : convert Caffe model parameters into MXNet's NDArray format</li> |
| <li><code>convert_mean.py</code> : convert Caffe input mean file into MXNet's NDArray format</li> |
| </ul> |
| |
| <p>In addition, there are two tools: |
| - <code>convert_caffe_modelzoo.py</code> : download and convert models from Caffe model |
| zoo. |
| - <code>test_converter.py</code> : test the converted models by checking the prediction |
| accuracy.</p> |
| |
| <h2 id="calling-caffe-operators-in-mxnet">Calling Caffe operators in MXNet</h2> |
| |
| <p>Besides converting Caffe models, MXNet supports calling most Caffe operators, |
| including network layer, data layer, and loss function, directly. It is |
| particularly useful if there are customized operators implemented in Caffe, then |
| we do not need to re-implement them in MXNet.</p> |
| |
| <h3 id="how-to-install">How to install</h3> |
| |
| <p>This feature requires Caffe. In particular, we need to re-compile Caffe before |
| <a href="https://github.com/BVLC/caffe/pull/4527">PR #4527</a> is merged into Caffe. There |
| are the steps of how to rebuild Caffe:</p> |
| |
| <ol> |
| <li>Download <a href="https://github.com/BVLC/caffe">Caffe</a>. E.g. <code>git clone |
| https://github.com/BVLC/caffe</code></li> |
| <li>Download the |
| <a href="https://github.com/BVLC/caffe/pull/4527.patch">patch for the MXNet interface</a> |
| and apply to Caffe. E.g. |
| <code>bash |
| cd caffe && wget https://github.com/BVLC/caffe/pull/4527.patch && git apply 4527.patch |
| </code></li> |
| <li>Build and install Caffe by following the |
| <a href="https://caffe.berkeleyvision.org/installation.html">official guide</a>.</li> |
| </ol> |
| |
| <p>Next we need to compile MXNet with Caffe supports</p> |
| |
| <ol> |
| <li>Copy <code>make/config.mk</code> (for Linux) or <code>make/osx.mk</code> |
| (for Mac) into the MXNet root folder as <code>config.mk</code> if you have not done it yet</li> |
| <li>Open the copied <code>config.mk</code> and uncomment these two lines |
| <code>bash |
| CAFFE_PATH = $(HOME)/caffe |
| MXNET_PLUGINS += plugin/caffe/caffe.mk |
| </code> |
| Modify <code>CAFFE_PATH</code> to your Caffe installation, if necessary.</li> |
| <li>Then build with 8 threads <code>make clean && make -j8</code>.</li> |
| </ol> |
| |
| <h3 id="how-to-use">How to use</h3> |
| |
| <p>This Caffe plugin adds three components into MXNet:</p> |
| |
| <ul> |
| <li><code>sym.CaffeOp</code> : Caffe neural network layer</li> |
| <li><code>sym.CaffeLoss</code> : Caffe loss functions</li> |
| <li><code>io.CaffeDataIter</code> : Caffe data layer</li> |
| </ul> |
| |
| <h4 id="use-sym-caffeop">Use <code>sym.CaffeOp</code></h4> |
| |
| <p>The following example shows the definition of a 10 classes multi-layer perceptron:</p> |
| <div class="highlight"><pre><code class="language-Python" data-lang="Python">data = mx.sym.Variable('data') |
| fc1 = mx.sym.CaffeOp(data_0=data, num_weight=2, name='fc1', prototxt="layer{type:\"InnerProduct\" inner_product_param{num_output: 128} }") |
| act1 = mx.sym.CaffeOp(data_0=fc1, prototxt="layer{type:\"TanH\"}") |
| fc2 = mx.sym.CaffeOp(data_0=act1, num_weight=2, name='fc2', prototxt="layer{type:\"InnerProduct\" inner_product_param{num_output: 64} }") |
| act2 = mx.sym.CaffeOp(data_0=fc2, prototxt="layer{type:\"TanH\"}") |
| fc3 = mx.sym.CaffeOp(data_0=act2, num_weight=2, name='fc3', prototxt="layer{type:\"InnerProduct\" inner_product_param{num_output: 10}}") |
| mlp = mx.sym.SoftmaxOutput(data=fc3, name='softmax') |
| </code></pre></div> |
| <p>Let's break it down. First, <code>data = mx.sym.Variable('data')</code> defines a variable |
| as a placeholder for input. Then, it's fed through Caffe operators with <code>fc1 = |
| mx.sym.CaffeOp(...)</code>. <code>CaffeOp</code> accepts several arguments:</p> |
| |
| <ul> |
| <li>The inputs to Caffe operators are named as <code>data_i</code> for <em>i=0, ..., num_data-1</em></li> |
| <li><code>num_data</code> is the number of inputs. In default it is 1, and therefore |
| skipped in the above example.</li> |
| <li><code>num_out</code> is the number of outputs. In default it is 1 and also skipped.</li> |
| <li><code>num_weight</code> is the number of weights (<code>blobs_</code>). Its default value is 0. We |
| need to explicitly specify it for a non-zero value.</li> |
| <li><code>prototxt</code> is the protobuf configuration string.</li> |
| </ul> |
| |
| <h4 id="use-sym-caffeloss">Use <code>sym.CaffeLoss</code></h4> |
| |
| <p>Using Caffe loss is similar. |
| We can replace the MXNet loss with Caffe loss. |
| We can replace</p> |
| |
| <p>Replacing the last line of the above example with the following two lines we can |
| call Caffe loss instead of MXNet loss.</p> |
| <div class="highlight"><pre><code class="language-Python" data-lang="Python">label = mx.sym.Variable('softmax_label') |
| mlp = mx.sym.CaffeLoss(data=fc3, label=label, grad_scale=1, name='softmax', prototxt="layer{type:\"SoftmaxWithLoss\"}") |
| </code></pre></div> |
| <p>Similar to <code>CaffeOp</code>, <code>CaffeLoss</code> has arguments <code>num_data</code> (2 in default) and |
| <code>num_out</code> (1 in default). But there are two differences</p> |
| |
| <ol> |
| <li>Inputs are <code>data</code> and <code>label</code>. And we need to explicitly create a variable |
| placeholder for label, which is implicitly done in MXNet loss.</li> |
| <li><code>grad_scale</code> is the weight of this loss.</li> |
| </ol> |
| |
| <h4 id="use-io-caffedataiter">Use <code>io.CaffeDataIter</code></h4> |
| |
| <p>We can also wrap a Caffe data layer into MXNet's data iterator. Below is an |
| example for creating a data iterator for MNIST</p> |
| <div class="highlight"><pre><code class="language-python" data-lang="python"><span class="n">train</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">io</span><span class="o">.</span><span class="n">CaffeDataIter</span><span class="p">(</span> |
| <span class="n">prototxt</span> <span class="o">=</span> |
| <span class="s">'layer { </span><span class="se">\ |
| </span><span class="s"> name: "mnist" </span><span class="se">\ |
| </span><span class="s"> type: "Data" </span><span class="se">\ |
| </span><span class="s"> top: "data" </span><span class="se">\ |
| </span><span class="s"> top: "label" </span><span class="se">\ |
| </span><span class="s"> include { </span><span class="se">\ |
| </span><span class="s"> phase: TEST </span><span class="se">\ |
| </span><span class="s"> } </span><span class="se">\ |
| </span><span class="s"> transform_param { </span><span class="se">\ |
| </span><span class="s"> scale: 0.00390625 </span><span class="se">\ |
| </span><span class="s"> } </span><span class="se">\ |
| </span><span class="s"> data_param { </span><span class="se">\ |
| </span><span class="s"> source: "caffe/examples/mnist/mnist_test_lmdb" </span><span class="se">\ |
| </span><span class="s"> batch_size: 100 </span><span class="se">\ |
| </span><span class="s"> backend: LMDB </span><span class="se">\ |
| </span><span class="s"> } </span><span class="se">\ |
| </span><span class="s"> }'</span><span class="p">,</span> |
| <span class="n">flat</span> <span class="o">=</span> <span class="n">flat</span><span class="p">,</span> |
| <span class="n">num_examples</span> <span class="o">=</span> <span class="mi">60000</span><span class="p">,</span> |
| <span class="p">)</span> |
| </code></pre></div> |
| <h3 id="put-it-all-together">Put it all together</h3> |
| |
| <p>The complete example is available at |
| <a href="https://github.com/apache/mxnet/blob/v1.x/example/caffe/">example/caffe</a></p> |
| |
| </div> |
| </div> |
| |
| </div> |
| </div> |
| |
| </article> |
| |
| </main><footer class="site-footer h-card"> |
| <div class="wrapper"> |
| <div class="row"> |
| <div class="col-4"> |
| <h4 class="footer-category-title">Resources</h4> |
| <ul class="contact-list"> |
| <li><a href="/versions/1.9.1/community/contribute#mxnet-dev-communications">Mailing lists</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="/versions/1.9.1/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="/versions/1.9.1/assets/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="/versions/1.9.1/assets/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="/versions/1.9.1/assets/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="/versions/1.9.1/assets/img/asf_logo.svg" class="footer-logo col-2"> |
| </div> |
| <div class="footer-bottom-warning col-9"> |
| </p><p>"Copyright © 2017-2022, The Apache Software Foundation. Licensed under the Apache License, Version 2.0. 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> |
