error_clip.html



<!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>Error Clip &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"/> 
<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">

  <div class="wy-grid-for-nav">

    
    <nav data-toggle="wy-nav-shift" class="wy-nav-side">
      <div class="wy-side-scroll">
        <div class="wy-side-nav-search">
          

            <a href="../index_cn.html" class="icon icon-home"> PaddlePaddle
          

          </a>

          
<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 class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
          
<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/quickstart_cn.html">快速开始</a></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="../build_and_install/index_cn.html">安装与编译</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../build_and_install/pip_install_cn.html">使用pip安装</a></li>
<li class="toctree-l2"><a class="reference internal" href="../build_and_install/docker_install_cn.html">使用Docker安装运行</a></li>
<li class="toctree-l2"><a class="reference internal" href="../build_and_install/build_from_source_cn.html">从源码编译</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../howto/index_cn.html">进阶使用</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../howto/cmd_parameter/index_cn.html">命令行参数设置</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../howto/cmd_parameter/use_case_cn.html">使用案例</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/cmd_parameter/arguments_cn.html">参数概述</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/cmd_parameter/detail_introduction_cn.html">细节描述</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../howto/cluster/index_cn.html">分布式训练</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../howto/cluster/preparations_cn.html">环境准备</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/cluster/cmd_argument_cn.html">启动参数说明</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/cluster/multi_cluster/index_cn.html">在不同集群中运行</a><ul>
<li class="toctree-l4"><a class="reference internal" href="../howto/cluster/multi_cluster/k8s_cn.html">Kubernetes单机训练</a></li>
<li class="toctree-l4"><a class="reference internal" href="../howto/cluster/multi_cluster/k8s_distributed_cn.html">Kubernetes分布式训练</a></li>
<li class="toctree-l4"><a class="reference internal" href="../howto/cluster/multi_cluster/openmpi_cn.html">在OpenMPI集群中启动训练</a></li>
<li class="toctree-l4"><a class="reference internal" href="../howto/cluster/multi_cluster/fabric_cn.html">使用fabric启动集群训练</a></li>
<li class="toctree-l4"><a class="reference internal" href="../howto/cluster/multi_cluster/k8s_aws_cn.html">Kubernetes on AWS</a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../howto/capi/index_cn.html">C-API预测库</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../howto/capi/compile_paddle_lib_cn.html">安装与编译C-API预测库</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/capi/organization_of_the_inputs_cn.html">输入/输出数据组织</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/capi/workflow_of_capi_cn.html">C-API使用流程</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../howto/rnn/index_cn.html">RNN模型</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../howto/rnn/rnn_config_cn.html">RNN配置</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/rnn/recurrent_group_cn.html">Recurrent Group教程</a></li>
<li class="toctree-l3"><a class="reference internal" href="../howto/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/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性能调优</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../dev/index_cn.html">开发标准</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../dev/contribute_to_paddle_cn.html">如何贡献代码</a></li>
<li class="toctree-l2"><a class="reference internal" href="../dev/write_docs_cn.html">如何贡献文档</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../faq/index_cn.html">FAQ</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../faq/build_and_install/index_cn.html">编译安装与单元测试</a></li>
<li class="toctree-l2"><a class="reference internal" href="../faq/model/index_cn.html">模型配置</a></li>
<li class="toctree-l2"><a class="reference internal" href="../faq/parameter/index_cn.html">参数设置</a></li>
<li class="toctree-l2"><a class="reference internal" href="../faq/local/index_cn.html">本地训练与预测</a></li>
<li class="toctree-l2"><a class="reference internal" href="../faq/cluster/index_cn.html">集群训练与预测</a></li>
</ul>
</li>
</ul>

</nav>

        </div>
      </div>
    </nav>

    <section data-toggle="wy-nav-shift" class="wy-nav-content-wrap">

      
      <nav class="wy-nav-top" role="navigation" aria-label="top navigation">
        <i data-toggle="wy-nav-top" class="fa fa-bars"></i>
        <a href="../index_cn.html">PaddlePaddle</a>
      </nav>


      <div class="wy-nav-content">
        <div class="rst-content">
          

<div role="navigation" aria-label="breadcrumbs navigation">
  <ul class="wy-breadcrumbs">
    <li><a href="../index_cn.html">Docs</a> &raquo;</li>
      
    <li>Error Clip</li>
      <li class="wy-breadcrumbs-aside">
        
          
            <a href="../_sources/design/error_clip.md.txt" rel="nofollow"> View page source</a>
          
        
      </li>
  </ul>
  <hr/>
</div>
          <div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
           <div itemprop="articleBody">
            
  <div class="section" id="error-clip">
<span id="error-clip"></span><h1>Error Clip<a class="headerlink" href="#error-clip" title="永久链接至标题">¶</a></h1>
<div class="section" id="overview">
<span id="overview"></span><h2>Overview<a class="headerlink" href="#overview" title="永久链接至标题">¶</a></h2>
<p>Error clip is widely used in model training to prevent gradient exploding. It takes some specific rules to adjust variables&#8217; gradients and prevent them from being too large. With it, values of a gradient will be checked before they are taken by the next <code class="docutils literal"><span class="pre">grad_op</span></code> and be shrunk if necessary.</p>
</div>
<div class="section" id="usage">
<span id="usage"></span><h2>Usage<a class="headerlink" href="#usage" title="永久链接至标题">¶</a></h2>
<p>Users are allowed to assign different error clip methods or attributes to different <code class="docutils literal"><span class="pre">Variable</span></code>s. Users can specify it as a parameter of <code class="docutils literal"><span class="pre">Variable</span></code>&#8216;s constructor:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">var</span> <span class="o">=</span> <span class="n">framework</span><span class="o">.</span><span class="n">Variable</span><span class="p">(</span><span class="o">...</span><span class="p">,</span> <span class="n">error_clip</span><span class="o">=</span><span class="n">myErrorClip</span><span class="p">,</span> <span class="o">...</span><span class="p">)</span>
</pre></div>
</div>
<p>The default value of <code class="docutils literal"><span class="pre">error_clip</span></code> is <code class="docutils literal"><span class="pre">None</span></code>, which means no error clip is employed. When it&#8217;s not <code class="docutils literal"><span class="pre">None</span></code>, it should take an object of <code class="docutils literal"><span class="pre">BaseErrorClipAttr</span></code>&#8216;s derived class. So far, <code class="docutils literal"><span class="pre">BaseErrorClipAttr</span></code> has only one derived class: <code class="docutils literal"><span class="pre">ErrorClipByValue</span></code>, whose constructor is:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">ErrorClipByValue</span><span class="p">(</span><span class="nb">max</span><span class="p">,</span> <span class="nb">min</span><span class="o">=</span><span class="bp">None</span><span class="p">)</span>
</pre></div>
</div>
<p><code class="docutils literal"><span class="pre">max</span></code> and <code class="docutils literal"><span class="pre">min</span></code> represent the maximal and minimal clip threshold respectively. In backward pass, all values of <code class="docutils literal"><span class="pre">var</span></code>&#8216;s gradient greater than <code class="docutils literal"><span class="pre">max</span></code> or less than <code class="docutils literal"><span class="pre">min</span></code> will be clipped to <code class="docutils literal"><span class="pre">max</span></code> and <code class="docutils literal"><span class="pre">min</span></code> respectively. When the <code class="docutils literal"><span class="pre">min</span></code> is None, the minimal threshold will be assigned with <code class="docutils literal"><span class="pre">-max</span></code> automatically.</p>
<p>So we can enable the error clip with threshold <code class="docutils literal"><span class="pre">[-5.0,</span> <span class="pre">5.0]</span></code> for variable <code class="docutils literal"><span class="pre">var</span></code> by:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">var</span> <span class="o">=</span> <span class="n">framework</span><span class="o">.</span><span class="n">Variable</span><span class="p">(</span><span class="o">...</span><span class="p">,</span> <span class="n">error_clip</span><span class="o">=</span><span class="n">ErrorClipByValue</span><span class="p">(</span><span class="nb">max</span><span class="o">=</span><span class="mf">5.0</span><span class="p">),</span> <span class="o">...</span><span class="p">)</span>
</pre></div>
</div>
</div>
<div class="section" id="implementation">
<span id="implementation"></span><h2>Implementation<a class="headerlink" href="#implementation" title="永久链接至标题">¶</a></h2>
<p>The <code class="docutils literal"><span class="pre">BaseErrorClipAttr</span></code> and its derived class <code class="docutils literal"><span class="pre">ErrorClipByValue</span></code> are defined in <em>clip.py</em>.</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="k">class</span> <span class="nc">BaseErrorClipAttr</span><span class="p">(</span><span class="nb">object</span><span class="p">):</span>
    <span class="k">def</span> <span class="nf">append_clip_op</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">block</span><span class="p">,</span> <span class="n">grad_name</span><span class="p">):</span>
        <span class="k">raise</span> <span class="ne">NotImplementedError</span><span class="p">()</span>


<span class="k">class</span> <span class="nc">ErrorClipByValue</span><span class="p">(</span><span class="n">BaseErrorClipAttr</span><span class="p">):</span>
    <span class="k">def</span> <span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="nb">max</span><span class="p">,</span> <span class="nb">min</span><span class="o">=</span><span class="bp">None</span><span class="p">):</span>
        <span class="nb">max</span> <span class="o">=</span> <span class="nb">float</span><span class="p">(</span><span class="nb">max</span><span class="p">)</span>
        <span class="k">if</span> <span class="nb">min</span> <span class="ow">is</span> <span class="bp">None</span><span class="p">:</span>
            <span class="nb">min</span> <span class="o">=</span> <span class="o">-</span><span class="nb">max</span>
        <span class="k">else</span><span class="p">:</span>
            <span class="nb">min</span> <span class="o">=</span> <span class="nb">float</span><span class="p">(</span><span class="nb">min</span><span class="p">)</span>
        <span class="bp">self</span><span class="o">.</span><span class="n">max</span> <span class="o">=</span> <span class="nb">max</span>
        <span class="bp">self</span><span class="o">.</span><span class="n">min</span> <span class="o">=</span> <span class="nb">min</span>

    <span class="k">def</span> <span class="nf">append_clip_op</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">block</span><span class="p">,</span> <span class="n">grad_name</span><span class="p">):</span>
        <span class="n">clip_op_desc</span> <span class="o">=</span> <span class="n">block</span><span class="o">.</span><span class="n">desc</span><span class="o">.</span><span class="n">append_op</span><span class="p">()</span>
        <span class="n">clip_op_desc</span><span class="o">.</span><span class="n">set_type</span><span class="p">(</span><span class="s2">&quot;clip&quot;</span><span class="p">)</span>
        <span class="n">clip_op_desc</span><span class="o">.</span><span class="n">set_input</span><span class="p">(</span><span class="s2">&quot;X&quot;</span><span class="p">,</span> <span class="p">[</span><span class="n">grad_name</span><span class="p">])</span>
        <span class="n">clip_op_desc</span><span class="o">.</span><span class="n">set_output</span><span class="p">(</span><span class="s2">&quot;Out&quot;</span><span class="p">,</span> <span class="p">[</span><span class="n">grad_name</span><span class="p">])</span>
        <span class="n">clip_op_desc</span><span class="o">.</span><span class="n">set_attr</span><span class="p">(</span><span class="s2">&quot;min&quot;</span><span class="p">,</span> <span class="bp">self</span><span class="o">.</span><span class="n">min</span><span class="p">)</span>
        <span class="n">clip_op_desc</span><span class="o">.</span><span class="n">set_attr</span><span class="p">(</span><span class="s2">&quot;max&quot;</span><span class="p">,</span> <span class="bp">self</span><span class="o">.</span><span class="n">max</span><span class="p">)</span>
</pre></div>
</div>
<p>The <code class="docutils literal"><span class="pre">BaseErrorClipAttr</span></code> have one main member functions: <code class="docutils literal"><span class="pre">append_clip_op(self,</span> <span class="pre">block,</span> <span class="pre">grad_name)</span></code>.</p>
<p>This function is used to create a <code class="docutils literal"><span class="pre">clip_op</span></code> and append it to the end of given <code class="docutils literal"><span class="pre">block</span></code>. For different error clip algorithm require different <code class="docutils literal"><span class="pre">clip_op</span></code>, the function is defined as virtual in the base class. All derived classes must implement their own versions of this function.</p>
<p>These <code class="docutils literal"><span class="pre">clip_op</span></code>s should be inserted after <code class="docutils literal"><span class="pre">grad_op</span></code>s whose output gradients need to be clipped. It is equivalent to appending some <code class="docutils literal"><span class="pre">clip_op</span></code>s to the end of the target block every time a new <code class="docutils literal"><span class="pre">grad_op</span></code> is added.</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="k">for</span> <span class="n">op_desc</span> <span class="ow">in</span> <span class="n">grad_op_descs</span><span class="p">:</span>
        <span class="n">new_op_desc</span> <span class="o">=</span> <span class="n">target_block</span><span class="o">.</span><span class="n">desc</span><span class="o">.</span><span class="n">append_op</span><span class="p">()</span>
        <span class="n">new_op_desc</span><span class="o">.</span><span class="n">copy_from</span><span class="p">(</span><span class="n">op_desc</span><span class="p">)</span>
        <span class="n">callback</span><span class="p">(</span><span class="n">block</span><span class="o">=</span><span class="n">target_block</span><span class="p">,</span> <span class="n">context</span><span class="o">=</span><span class="n">grad_to_var</span><span class="p">)</span>
</pre></div>
</div>
<p>Here we employ a callback function to complete this kind of jobs. In <code class="docutils literal"><span class="pre">_append_backward_ops_</span></code> function, each time after a <code class="docutils literal"><span class="pre">grad_op</span></code> is added to the <code class="docutils literal"><span class="pre">target_block</span></code>, a callback function is invoked. The logic of <code class="docutils literal"><span class="pre">clip_op</span></code> appending can be implemented inside the callback function.</p>
<p>The callback function for <code class="docutils literal"><span class="pre">clip_op</span></code> appending is defined in <em>clip.py</em>:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="k">def</span> <span class="nf">error_clip_callback</span><span class="p">(</span><span class="n">block</span><span class="p">,</span> <span class="n">context</span><span class="p">):</span>
    <span class="c1"># the context is a grad_to_var map</span>
    <span class="n">grad_to_var</span> <span class="o">=</span> <span class="n">context</span>
    <span class="n">op_desc</span> <span class="o">=</span> <span class="n">block</span><span class="o">.</span><span class="n">desc</span><span class="o">.</span><span class="n">op</span><span class="p">(</span><span class="n">block</span><span class="o">.</span><span class="n">desc</span><span class="o">.</span><span class="n">op_size</span><span class="p">()</span> <span class="o">-</span> <span class="mi">1</span><span class="p">)</span>
    <span class="k">for</span> <span class="n">grad_n</span> <span class="ow">in</span> <span class="nb">filter</span><span class="p">(</span><span class="k">lambda</span> <span class="n">n</span><span class="p">:</span> <span class="n">grad_to_var</span><span class="o">.</span><span class="n">has_key</span><span class="p">(</span><span class="n">n</span><span class="p">),</span>
                         <span class="n">op_desc</span><span class="o">.</span><span class="n">output_arg_names</span><span class="p">()):</span>
        <span class="n">fwd_var</span> <span class="o">=</span> <span class="n">block</span><span class="o">.</span><span class="n">var_recursive</span><span class="p">(</span><span class="n">grad_to_var</span><span class="p">[</span><span class="n">grad_n</span><span class="p">])</span>
        <span class="n">error_clip</span> <span class="o">=</span> <span class="nb">getattr</span><span class="p">(</span><span class="n">fwd_var</span><span class="p">,</span> <span class="s2">&quot;error_clip&quot;</span><span class="p">,</span> <span class="bp">None</span><span class="p">)</span>
        <span class="k">if</span> <span class="ow">not</span> <span class="p">(</span><span class="n">error_clip</span> <span class="ow">is</span> <span class="bp">None</span> <span class="ow">or</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">error_clip</span><span class="p">,</span>
                                                 <span class="n">BaseErrorClipAttr</span><span class="p">)):</span>
            <span class="k">raise</span> <span class="ne">TypeError</span><span class="p">(</span>
                <span class="s2">&quot;Variable&#39;s error_clip should be an instance of BaseErrorClipAttr or None.&quot;</span>
            <span class="p">)</span>
        <span class="k">if</span> <span class="n">error_clip</span> <span class="ow">is</span> <span class="ow">not</span> <span class="bp">None</span><span class="p">:</span>
            <span class="n">error_clip</span><span class="o">.</span><span class="n">append_clip_op</span><span class="p">(</span><span class="n">block</span><span class="p">,</span> <span class="n">grad_n</span><span class="p">)</span>
</pre></div>
</div>
<p>This function takes a <code class="docutils literal"><span class="pre">block</span></code> and a <code class="docutils literal"><span class="pre">context</span></code>(which is actually a grad_to_var map) as inputs. It checks each output of the last <code class="docutils literal"><span class="pre">OpDesc</span></code> in the <code class="docutils literal"><span class="pre">block</span></code>. Notice that the last <code class="docutils literal"><span class="pre">OpDesc</span></code> of the <code class="docutils literal"><span class="pre">block</span></code> must be a <code class="docutils literal"><span class="pre">grad_op</span></code> and its outputs must be some forward variables&#8217; gradients. If an output gradient&#8217;s corresponding forward variable has an attribute of <code class="docutils literal"><span class="pre">error_clip</span></code>, <code class="docutils literal"><span class="pre">error_clip_callback</span></code> will call the <code class="docutils literal"><span class="pre">error_clip</span></code>&#8216;s <code class="docutils literal"><span class="pre">append_clip_op</span></code> function to append the required <code class="docutils literal"><span class="pre">clip_op</span></code> into the <code class="docutils literal"><span class="pre">block</span></code>.</p>
</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
        };
    </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 type="text/javascript">
      jQuery(function () {
          SphinxRtdTheme.StickyNav.enable();
      });
  </script>
   

</body>
</html>