Skip to content

P
Paddle

机器未来 / Paddle
与 Fork 源项目一致

Fork自 PaddlePaddle / Paddle

通知 1
Star 1
Fork 0
P
Paddle
切换分支/标签
查找文件 普通视图历史永久链接Permalink
auto_gradient_check.html 24.0 KB
Newer Older
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
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 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 


<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
  <meta charset="utf-8">
  
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
  
  <title>Auto Gradient Checker Design &mdash; PaddlePaddle  文档</title>
  

  
  

  

  
  
    

  

  
  
    <link rel="stylesheet" href="../_static/css/theme.css" type="text/css" />
  

  
  
        <link rel="index" title="索引"
              href="../genindex.html"/>
        <link rel="search" title="搜索" href="../search.html"/>
    <link rel="top" title="PaddlePaddle  文档" href="../index.html"/> 

  <link rel="stylesheet" href="https://cdn.jsdelivr.net/perfect-scrollbar/0.6.14/css/perfect-scrollbar.min.css" type="text/css" />
  <link rel="stylesheet" href="../_static/css/override.css" type="text/css" />
  <script>
  var _hmt = _hmt || [];
  (function() {
    var hm = document.createElement("script");
    hm.src = "//hm.baidu.com/hm.js?b9a314ab40d04d805655aab1deee08ba";
    var s = document.getElementsByTagName("script")[0]; 
    s.parentNode.insertBefore(hm, s);
  })();
  </script>

  

  
  <script src="../_static/js/modernizr.min.js"></script>

</head>

<body class="wy-body-for-nav" role="document">

  
  <header class="site-header">
    <div class="site-logo">
      <a href="/"><img src="../_static/images/PP_w.png"></a>
    </div>
    <div class="site-nav-links">
      <div class="site-menu">
        <a class="fork-on-github" href="https://github.com/PaddlePaddle/Paddle" target="_blank"><i class="fa fa-github"></i>Fork me on Github</a>
        <div class="language-switcher dropdown">
          <a type="button" data-toggle="dropdown">
            <span>English</span>
            <i class="fa fa-angle-up"></i>
            <i class="fa fa-angle-down"></i>
          </a>
          <ul class="dropdown-menu">
            <li><a href="/doc_cn">中文</a></li>
            <li><a href="/doc">English</a></li>
          </ul>
        </div>
        <ul class="site-page-links">
          <li><a href="/">Home</a></li>
        </ul>
      </div>
      <div class="doc-module">
        
        <ul>
<li class="toctree-l1"><a class="reference internal" href="../getstarted/index_cn.html">新手入门</a></li>
<li class="toctree-l1"><a class="reference internal" href="../howto/index_cn.html">进阶指南</a></li>
<li class="toctree-l1"><a class="reference internal" href="../api/index_cn.html">API</a></li>
<li class="toctree-l1"><a class="reference internal" href="../faq/index_cn.html">FAQ</a></li>
</ul>

        
<div role="search">
  <form id="rtd-search-form" class="wy-form" action="../search.html" method="get">
    <input type="text" name="q" placeholder="Search docs" />
    <input type="hidden" name="check_keywords" value="yes" />
    <input type="hidden" name="area" value="default" />
  </form>
</div>        
      </div>
    </div>
  </header>
  
  <div class="main-content-wrap">

    
    <nav class="doc-menu-vertical" role="navigation">
        
          
          <ul>
<li class="toctree-l1"><a class="reference internal" href="../getstarted/index_cn.html">新手入门</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../getstarted/build_and_install/index_cn.html">安装与编译</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../getstarted/build_and_install/docker_install_cn.html">PaddlePaddle的Docker容器使用方式</a></li>
<li class="toctree-l3"><a class="reference internal" href="../getstarted/build_and_install/cmake/build_from_source_cn.html">PaddlePaddle的编译选项</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../getstarted/concepts/use_concepts_cn.html">基本使用概念</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../howto/index_cn.html">进阶指南</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../howto/usage/cmd_parameter/index_cn.html">设置命令行参数</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../howto/usage/cmd_parameter/use_case_cn.html">使用案例</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/usage/cmd_parameter/arguments_cn.html">参数概述</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/usage/cmd_parameter/detail_introduction_cn.html">细节描述</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../howto/usage/cluster/cluster_train_cn.html">运行分布式训练</a></li>
<li class="toctree-l2"><a class="reference internal" href="../howto/usage/k8s/k8s_basis_cn.html">Kubernetes 简介</a></li>
<li class="toctree-l2"><a class="reference internal" href="../howto/usage/k8s/k8s_cn.html">Kubernetes单机训练</a></li>
<li class="toctree-l2"><a class="reference internal" href="../howto/usage/k8s/k8s_distributed_cn.html">Kubernetes分布式训练</a></li>
T
Deploy to GitHub Pages: 111812fa3b41797a3940bbcd51e51ff21b799ffe  
Travis CI 已提交
129 
<li class="toctree-l2"><a class="reference internal" href="../howto/dev/build_cn.html">编译PaddlePaddle和运行单元测试</a></li>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 
<li class="toctree-l2"><a class="reference internal" href="../howto/dev/write_docs_cn.html">如何贡献/修改文档</a></li>
<li class="toctree-l2"><a class="reference internal" href="../howto/dev/contribute_to_paddle_cn.html">如何贡献代码</a></li>
<li class="toctree-l2"><a class="reference internal" href="../howto/deep_model/rnn/index_cn.html">RNN相关模型</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../howto/deep_model/rnn/rnn_config_cn.html">RNN配置</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/deep_model/rnn/recurrent_group_cn.html">Recurrent Group教程</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/deep_model/rnn/hierarchical_layer_cn.html">支持双层序列作为输入的Layer</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/deep_model/rnn/hrnn_rnn_api_compare_cn.html">单双层RNN API对比介绍</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../howto/optimization/gpu_profiling_cn.html">GPU性能分析与调优</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../api/index_cn.html">API</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../api/v2/model_configs.html">模型配置</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../api/v2/config/activation.html">Activation</a></li>
<li class="toctree-l3"><a class="reference internal" href="../api/v2/config/layer.html">Layers</a></li>
<li class="toctree-l3"><a class="reference internal" href="../api/v2/config/evaluators.html">Evaluators</a></li>
<li class="toctree-l3"><a class="reference internal" href="../api/v2/config/optimizer.html">Optimizer</a></li>
<li class="toctree-l3"><a class="reference internal" href="../api/v2/config/pooling.html">Pooling</a></li>
<li class="toctree-l3"><a class="reference internal" href="../api/v2/config/networks.html">Networks</a></li>
<li class="toctree-l3"><a class="reference internal" href="../api/v2/config/attr.html">Parameter Attribute</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../api/v2/data.html">数据访问</a></li>
<li class="toctree-l2"><a class="reference internal" href="../api/v2/run_logic.html">训练与应用</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../faq/index_cn.html">FAQ</a></li>
</ul>

        
    </nav>
    
    <section class="doc-content-wrap">

      

 







<div role="navigation" aria-label="breadcrumbs navigation">
  <ul class="wy-breadcrumbs">
      
    <li>Auto Gradient Checker Design</li>
  </ul>
</div>
      
      <div class="wy-nav-content" id="doc-content">
        <div class="rst-content">
          <div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
           <div itemprop="articleBody">
            
  <div class="section" id="auto-gradient-checker-design">
<span id="auto-gradient-checker-design"></span><h1>Auto Gradient Checker Design<a class="headerlink" href="#auto-gradient-checker-design" title="永久链接至标题"></a></h1>
</div>
<div class="section" id="backgraound">
<span id="backgraound"></span><h1>Backgraound：<a class="headerlink" href="#backgraound" title="永久链接至标题"></a></h1>
<ul class="simple">
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
193 
<li>Generally, it is easy to check whether the forward computation of an Operator is correct or not. However, backpropagation is a notoriously difficult algorithm to debug and get right:<ol>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
194 195 196 197 198 
<li>you should get the right backpropagation formula according to the forward computation.</li>
<li>you should implement it right in CPP.</li>
<li>it&#8217;s difficult to prepare test data.</li>
</ol>
</li>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
199 200 
<li>Auto gradient checking gets a numerical gradient by forward Operator and use it as a reference of the backward Operator&#8217;s result. It has several advantages:<ol>
<li>numerical gradient checker only need forward operator.</li>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
201 202 203 204 205 206 207 
<li>user only need to prepare the input data for forward Operator.</li>
</ol>
</li>
</ul>
</div>
<div class="section" id="mathematical-theory">
<span id="mathematical-theory"></span><h1>Mathematical Theory<a class="headerlink" href="#mathematical-theory" title="永久链接至标题"></a></h1>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
208 
<p>The following two document from Stanford has a detailed explanation of how to get numerical gradient and why it&#8217;s useful.</p>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
209 210 211 212 213 214 215 216 217 218 219 
<div class="toctree-wrapper compound">
<ul>
<li class="toctree-l1"><a class="reference external" href="http://deeplearning.stanford.edu/wiki/index.php/Gradient_checking_and_advanced_optimization">Gradient checking and advanced optimization(en)</a></li>
<li class="toctree-l1"><a class="reference external" href="http://ufldl.stanford.edu/wiki/index.php/%E6%A2%AF%E5%BA%A6%E6%A3%80%E9%AA%8C%E4%B8%8E%E9%AB%98%E7%BA%A7%E4%BC%98%E5%8C%96">Gradient checking and advanced optimization(cn)</a></li>
</ul>
</div>
</div>
<div class="section" id="numeric-gradient-implementation">
<span id="numeric-gradient-implementation"></span><h1>Numeric Gradient Implementation<a class="headerlink" href="#numeric-gradient-implementation" title="永久链接至标题"></a></h1>
<div class="section" id="python-interface">
<span id="python-interface"></span><h2>Python Interface<a class="headerlink" href="#python-interface" title="永久链接至标题"></a></h2>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
220 
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="k">def</span> <span class="nf">get_numerical_gradient</span><span class="p">(</span><span class="n">op</span><span class="p">,</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
221 222 223 224 225 226 227 228 229 
                         <span class="n">input_values</span><span class="p">,</span>
                         <span class="n">output_name</span><span class="p">,</span>
                         <span class="n">input_to_check</span><span class="p">,</span>
                         <span class="n">delta</span><span class="o">=</span><span class="mf">0.005</span><span class="p">,</span>
                         <span class="n">local_scope</span><span class="o">=</span><span class="bp">None</span><span class="p">):</span>
    <span class="sd">&quot;&quot;&quot;</span>
<span class="sd">    Get Numeric Gradient for an operator&#39;s input.</span>

<span class="sd">    :param op: C++ operator instance, could be an network</span>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
230 231 
<span class="sd">    :param input_values: The input variables. Should be an dictionary, whose key is</span>
<span class="sd">    variable name, and value is numpy array.</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
232 
<span class="sd">    :param output_name: The final output variable name.</span>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
233 
<span class="sd">    :param input_to_check: The input variable with respect to which to compute the gradient.</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
234 235 
<span class="sd">    :param delta: The perturbation value for numeric gradient method. The</span>
<span class="sd">    smaller delta is, the more accurate result will get. But if that delta is</span>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
236 
<span class="sd">     too small, it will suffer from numerical stability problem.</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
237 238 239 240 241 242 243 244 245 246 
<span class="sd">    :param local_scope: The local scope used for get_numeric_gradient.</span>
<span class="sd">    :return: The gradient array in numpy format.</span>
<span class="sd">    &quot;&quot;&quot;</span>
</pre></div>
</div>
</div>
<div class="section" id="explaination">
<span id="explaination"></span><h2>Explaination:<a class="headerlink" href="#explaination" title="永久链接至标题"></a></h2>
<ul class="simple">
<li>Why need <code class="docutils literal"><span class="pre">output_name</span></code><ul>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
247 
<li>An Operator may have multiple Output, one can get independent gradient from each Output. So caller should specify the name of the output variable.</li>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
248 249 250 
</ul>
</li>
<li>Why need <code class="docutils literal"><span class="pre">input_to_check</span></code><ul>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
251 
<li>One operator may have multiple inputs. Gradient Op can calculate the gradient of these inputs at the same time. But Numeric Gradient needs to calculate them one by one. So <code class="docutils literal"><span class="pre">get_numeric_gradient</span></code> is designed to calculate the gradient for one input. If you need to compute multiple inputs, you can call <code class="docutils literal"><span class="pre">get_numeric_gradient</span></code> multiple times.</li>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
252 253 254 255 256 257 
</ul>
</li>
</ul>
</div>
<div class="section" id="core-algorithm-implementation">
<span id="core-algorithm-implementation"></span><h2>Core Algorithm Implementation<a class="headerlink" href="#core-algorithm-implementation" title="永久链接至标题"></a></h2>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
258 259 
<div class="highlight-python"><div class="highlight"><pre><span></span>    <span class="c1"># we only compute gradient of one element a time.</span>
    <span class="c1"># we use a for loop to compute the gradient of each element.</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
260 
    <span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">xrange</span><span class="p">(</span><span class="n">tensor_size</span><span class="p">):</span>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
261 
        <span class="c1"># get one input element by its index i.</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
262 263 
        <span class="n">origin</span> <span class="o">=</span> <span class="n">tensor_to_check</span><span class="o">.</span><span class="n">get_float_element</span><span class="p">(</span><span class="n">i</span><span class="p">)</span>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
264 
        <span class="c1"># add delta to it, run op and then get the new value of the result tensor.</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
265 266 267 268 
        <span class="n">x_pos</span> <span class="o">=</span> <span class="n">origin</span> <span class="o">+</span> <span class="n">delta</span>
        <span class="n">tensor_to_check</span><span class="o">.</span><span class="n">set_float_element</span><span class="p">(</span><span class="n">i</span><span class="p">,</span> <span class="n">x_pos</span><span class="p">)</span>
        <span class="n">y_pos</span> <span class="o">=</span> <span class="n">get_output</span><span class="p">()</span>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
269 
        <span class="c1"># plus delta to this element, run op and get the new value of the result tensor.</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 
        <span class="n">x_neg</span> <span class="o">=</span> <span class="n">origin</span> <span class="o">-</span> <span class="n">delta</span>
        <span class="n">tensor_to_check</span><span class="o">.</span><span class="n">set_float_element</span><span class="p">(</span><span class="n">i</span><span class="p">,</span> <span class="n">x_neg</span><span class="p">)</span>
        <span class="n">y_neg</span> <span class="o">=</span> <span class="n">get_output</span><span class="p">()</span>

        <span class="c1"># restore old value</span>
        <span class="n">tensor_to_check</span><span class="o">.</span><span class="n">set_float_element</span><span class="p">(</span><span class="n">i</span><span class="p">,</span> <span class="n">origin</span><span class="p">)</span>

        <span class="c1"># compute the gradient of this element and store it into a numpy array.</span>
        <span class="n">gradient_flat</span><span class="p">[</span><span class="n">i</span><span class="p">]</span> <span class="o">=</span> <span class="p">(</span><span class="n">y_pos</span> <span class="o">-</span> <span class="n">y_neg</span><span class="p">)</span> <span class="o">/</span> <span class="n">delta</span> <span class="o">/</span> <span class="mi">2</span>

    <span class="c1"># reshape the gradient result to the shape of the source tensor.</span>
    <span class="k">return</span> <span class="n">gradient_flat</span><span class="o">.</span><span class="n">reshape</span><span class="p">(</span><span class="n">tensor_to_check</span><span class="o">.</span><span class="n">get_dims</span><span class="p">())</span>
</pre></div>
</div>
</div>
</div>
<div class="section" id="auto-graident-checker-framework">
<span id="auto-graident-checker-framework"></span><h1>Auto Graident Checker Framework<a class="headerlink" href="#auto-graident-checker-framework" title="永久链接至标题"></a></h1>
<p>Each Operator Kernel has three kinds of Gradient:</p>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
289 290 291 292 
<ol class="simple">
<li>Numerical gradient</li>
<li>CPU kernel gradient</li>
<li>GPU kernel gradient (if supported)</li>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
293 
</ol>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
294 295 296 297 298 
<p>The numerical gradient only relies on forward Operator. So we use the numerical gradient as the reference value. And the gradient checking is performed in the following three steps:</p>
<ol class="simple">
<li>calculate the numerical gradient</li>
<li>calculate CPU kernel gradient with the backward Operator and compare it with the numerical gradient</li>
<li>calculate GPU kernel gradient with the backward Operator and compare it with the numeric gradient (if supported)</li>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 
</ol>
<div class="section" id="python-interface">
<span id="id1"></span><h2>Python Interface<a class="headerlink" href="#python-interface" title="永久链接至标题"></a></h2>
<div class="highlight-python"><div class="highlight"><pre><span></span>    <span class="k">def</span> <span class="nf">check_grad</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span>
                   <span class="n">forward_op</span><span class="p">,</span>
                   <span class="n">input_vars</span><span class="p">,</span>
                   <span class="n">inputs_to_check</span><span class="p">,</span>
                   <span class="n">output_name</span><span class="p">,</span>
                   <span class="n">no_grad_set</span><span class="o">=</span><span class="bp">None</span><span class="p">,</span>
                   <span class="n">only_cpu</span><span class="o">=</span><span class="bp">False</span><span class="p">,</span>
                   <span class="n">max_relative_error</span><span class="o">=</span><span class="mf">0.005</span><span class="p">):</span>
        <span class="sd">&quot;&quot;&quot;</span>
<span class="sd">        :param forward_op: used to create backward_op</span>
<span class="sd">        :param input_vars: numpy value of input variable. The following</span>
<span class="sd">            computation will use these variables.</span>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
314 315 
<span class="sd">        :param inputs_to_check: the input variable with respect to which to compute the gradient.</span>
<span class="sd">        :param output_name: The final output variable name.</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
316 317 318 319 320 321 322 323 324 325 
<span class="sd">        :param max_relative_error: The relative tolerance parameter.</span>
<span class="sd">        :param no_grad_set: used when create backward ops</span>
<span class="sd">        :param only_cpu: only compute and check gradient on cpu kernel.</span>
<span class="sd">        :return:</span>
<span class="sd">        &quot;&quot;&quot;</span>
</pre></div>
</div>
</div>
<div class="section" id="how-to-check-if-two-numpy-array-is-close-enough">
<span id="how-to-check-if-two-numpy-array-is-close-enough"></span><h2>How to check if two numpy array is close enough?<a class="headerlink" href="#how-to-check-if-two-numpy-array-is-close-enough" title="永久链接至标题"></a></h2>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
326 327 
<p>if <code class="docutils literal"><span class="pre">abs_numerical_grad</span></code> is nearly zero, then use abs error for numerical_grad</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">numerical_grad</span> <span class="o">=</span> <span class="o">...</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
328 329 
<span class="n">operator_grad</span> <span class="o">=</span> <span class="n">numpy</span><span class="o">.</span><span class="n">array</span><span class="p">(</span><span class="n">scope</span><span class="o">.</span><span class="n">find_var</span><span class="p">(</span><span class="n">grad_var_name</span><span class="p">(</span><span class="n">name</span><span class="p">))</span><span class="o">.</span><span class="n">get_tensor</span><span class="p">())</span>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
330 331 
<span class="n">abs_numerical_grad</span> <span class="o">=</span> <span class="n">numpy</span><span class="o">.</span><span class="n">abs</span><span class="p">(</span><span class="n">numerical_grad</span><span class="p">)</span>
<span class="c1"># if abs_numerical_grad is nearly zero, then use abs error for numeric_grad, not relative</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
332 
<span class="c1"># error.</span>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
333 
<span class="n">abs_numerical_grad</span><span class="p">[</span><span class="n">abs_numerical_grad</span> <span class="o">&lt;</span> <span class="mf">1e-3</span><span class="p">]</span> <span class="o">=</span> <span class="mi">1</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
334
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
335 
<span class="n">diff_mat</span> <span class="o">=</span> <span class="n">numpy</span><span class="o">.</span><span class="n">abs</span><span class="p">(</span><span class="n">abs_numerical_grad</span> <span class="o">-</span> <span class="n">operator_grad</span><span class="p">)</span> <span class="o">/</span> <span class="n">abs_numerical_grad</span>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
336 337 338 339 340 
<span class="n">max_diff</span> <span class="o">=</span> <span class="n">numpy</span><span class="o">.</span><span class="n">max</span><span class="p">(</span><span class="n">diff_mat</span><span class="p">)</span>
</pre></div>
</div>
<div class="section" id="notes">
<span id="notes"></span><h3>Notes：<a class="headerlink" href="#notes" title="永久链接至标题"></a></h3>
T
Deploy to GitHub Pages: 63bc2ff876724016062223bc3c966a4db43ae9c5  
Travis CI 已提交
341 
<p>The Input data for auto gradient checker should be reasonable to avoid numerical  stability problem.</p>
T
Deploy to GitHub Pages: 0dc75ab3ef6bcf07bc93e08f8fa4bcd02503cbc4  
Travis CI 已提交
342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 
</div>
<div class="section" id="refs">
<span id="refs"></span><h3>Refs:<a class="headerlink" href="#refs" title="永久链接至标题"></a></h3>
<div class="toctree-wrapper compound">
<ul>
<li class="toctree-l1"><a class="reference external" href="http://deeplearning.stanford.edu/wiki/index.php/Gradient_checking_and_advanced_optimization">Gradient checking and advanced optimization(en)</a></li>
<li class="toctree-l1"><a class="reference external" href="http://ufldl.stanford.edu/wiki/index.php/%E6%A2%AF%E5%BA%A6%E6%A3%80%E9%AA%8C%E4%B8%8E%E9%AB%98%E7%BA%A7%E4%BC%98%E5%8C%96">Gradient checking and advanced optimization(cn)</a></li>
</ul>
</div>
</div>
</div>
</div>


           </div>
          </div>
          <footer>
  

  <hr/>

  <div role="contentinfo">
    <p>
        &copy; Copyright 2016, PaddlePaddle developers.

    </p>
  </div>
  Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/snide/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>. 

</footer>

        </div>
      </div>

    </section>

  </div>
  


  

    <script type="text/javascript">
        var DOCUMENTATION_OPTIONS = {
            URL_ROOT:'../',
            VERSION:'',
            COLLAPSE_INDEX:false,
            FILE_SUFFIX:'.html',
            HAS_SOURCE:  true,
            SOURCELINK_SUFFIX: ".txt",
        };
    </script>
      <script type="text/javascript" src="../_static/jquery.js"></script>
      <script type="text/javascript" src="../_static/underscore.js"></script>
      <script type="text/javascript" src="../_static/doctools.js"></script>
      <script type="text/javascript" src="../_static/translations.js"></script>
      <script type="text/javascript" src="https://cdn.bootcss.com/mathjax/2.7.0/MathJax.js"></script>
       
  

  
  
    <script type="text/javascript" src="../_static/js/theme.js"></script>
  
  
  <script src="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.7/js/bootstrap.min.js" integrity="sha384-Tc5IQib027qvyjSMfHjOMaLkfuWVxZxUPnCJA7l2mCWNIpG9mGCD8wGNIcPD7Txa" crossorigin="anonymous"></script>
  <script src="https://cdn.jsdelivr.net/perfect-scrollbar/0.6.14/js/perfect-scrollbar.jquery.min.js"></script>
  <script src="../_static/js/paddle_doc_init.js"></script> 

</body>
</html>