Skip to content

P
Paddle

BaiXuePrincess / 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
index.html 21.3 KB
Newer Older
Y
Update docs  
Yu Yang 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 
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
  "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">


<html xmlns="http://www.w3.org/1999/xhtml">
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
    
    <title>Chinese Word Embedding Model Tutorial &mdash; PaddlePaddle  documentation</title>
    
    <link rel="stylesheet" href="../../_static/classic.css" type="text/css" />
    <link rel="stylesheet" href="../../_static/pygments.css" type="text/css" />
    
    <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="https://cdn.mathjax.org/mathjax/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
    <link rel="top" title="PaddlePaddle  documentation" href="../../index.html" />
    <link rel="up" title="Examples and demos" href="../index.html" />
Y
Upload docs  
Yu Yang 已提交
29 
    <link rel="next" title="Cluster Train" href="../../cluster/index.html" />
Y
Update docs  
Yu Yang 已提交
30 31 32 33 34 35 36 37 38 39 40 41 42 
    <link rel="prev" title="Model Zoo - ImageNet" href="../imagenet_model/resnet_model.html" /> 
  </head>
  <body role="document">
    <div class="related" role="navigation" aria-label="related navigation">
      <h3>Navigation</h3>
      <ul>
        <li class="right" style="margin-right: 10px">
          <a href="../../genindex.html" title="General Index"
             accesskey="I">index</a></li>
        <li class="right" >
          <a href="../../py-modindex.html" title="Python Module Index"
             >modules</a> |</li>
        <li class="right" >
Y
Upload docs  
Yu Yang 已提交
43 
          <a href="../../cluster/index.html" title="Cluster Train"
Y
Update docs  
Yu Yang 已提交
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 
             accesskey="N">next</a> |</li>
        <li class="right" >
          <a href="../imagenet_model/resnet_model.html" title="Model Zoo - ImageNet"
             accesskey="P">previous</a> |</li>
        <li class="nav-item nav-item-0"><a href="../../index.html">PaddlePaddle  documentation</a> &raquo;</li>
          <li class="nav-item nav-item-1"><a href="../index.html" accesskey="U">Examples and demos</a> &raquo;</li> 
      </ul>
    </div>  

    <div class="document">
      <div class="documentwrapper">
        <div class="bodywrapper">
          <div class="body" role="main">
            
  <div class="section" id="chinese-word-embedding-model-tutorial">
<span id="chinese-word-embedding-model-tutorial"></span><h1>Chinese Word Embedding Model Tutorial<a class="headerlink" href="#chinese-word-embedding-model-tutorial" title="Permalink to this headline"></a></h1>
<hr class="docutils" />
<p>This tutorial is to guide you through the process of using a Pretrained Chinese Word Embedding Model in the PaddlePaddle standard format.</p>
<p>We thank &#64;lipeng for the pull request that defined the model schemas and pretrained the models.</p>
<div class="section" id="introduction">
<span id="introduction"></span><h2>Introduction<a class="headerlink" href="#introduction" title="Permalink to this headline"></a></h2>
<div class="section" id="chinese-word-dictionary">
<span id="chinese-word-dictionary"></span><h3>Chinese Word Dictionary<a class="headerlink" href="#chinese-word-dictionary" title="Permalink to this headline"></a></h3>
<p>Our Chinese-word dictionary is created on Baidu ZhiDao and Baidu Baike by using in-house word segmentor. For example, the participle of &#8220;《红楼梦》&#8221; is &#8220;《&#8221;，&#8221;红楼梦&#8221;，&#8221;》&#8221;，and &#8220;《红楼梦》&#8221;. Our dictionary (using UTF-8 format) has has two columns: word and its frequency. The total word count is 3206325, including 3 special token:</p>
<ul class="simple">
<li><code class="docutils literal"><span class="pre">&lt;s&gt;</span></code>: the start of a sequence</li>
<li><code class="docutils literal"><span class="pre">&lt;e&gt;</span></code>: the end of a sequence</li>
<li><code class="docutils literal"><span class="pre">&lt;unk&gt;</span></code>: a word not included in dictionary</li>
</ul>
</div>
<div class="section" id="pretrained-chinese-word-embedding-model">
<span id="pretrained-chinese-word-embedding-model"></span><h3>Pretrained Chinese Word Embedding Model<a class="headerlink" href="#pretrained-chinese-word-embedding-model" title="Permalink to this headline"></a></h3>
<p>Inspired by paper <a class="reference external" href="http://www.jmlr.org/papers/volume3/bengio03a/bengio03a.pdf">A Neural Probabilistic Language Model</a>, our model architecture (<strong>Embedding joint of six words-&gt;FullyConnect-&gt;SoftMax</strong>) is as following graph. And for our dictionary, we pretrain four models with different word vector dimenstions, i.e 32, 64, 128, 256.
<center><img alt="" src="../../_images/neural-n-gram-model.png" /></center>
<center>Figure 1. neural-n-gram-model</center></p>
</div>
<div class="section" id="download-and-extract">
<span id="download-and-extract"></span><h3>Download and Extract<a class="headerlink" href="#download-and-extract" title="Permalink to this headline"></a></h3>
<p>To download and extract our dictionary and pretrained model, run the following commands.</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>cd $PADDLE_ROOT/demo/model_zoo/embedding
./pre_DictAndModel.sh
</pre></div>
</div>
</div>
</div>
<div class="section" id="chinese-paraphrasing-example">
<span id="chinese-paraphrasing-example"></span><h2>Chinese Paraphrasing Example<a class="headerlink" href="#chinese-paraphrasing-example" title="Permalink to this headline"></a></h2>
<p>We provide a paraphrasing task to show the usage of pretrained Chinese Word Dictionary and Embedding Model.</p>
<div class="section" id="data-preparation-and-preprocess">
<span id="data-preparation-and-preprocess"></span><h3>Data Preparation and Preprocess<a class="headerlink" href="#data-preparation-and-preprocess" title="Permalink to this headline"></a></h3>
<p>First, run the following commands to download and extract the in-house dataset. The dataset (using UTF-8 format) has 20 training samples, 5 testing samples and 2 generating samples.</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>cd $PADDLE_ROOT/demo/seqToseq/data
./paraphrase_data.sh
</pre></div>
</div>
<p>Second, preprocess data and build dictionary on train data by running the following commands, and the preprocessed dataset is stored in <code class="docutils literal"><span class="pre">$PADDLE_SOURCE_ROOT/demo/seqToseq/data/pre-paraphrase</span></code>:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>cd $PADDLE_ROOT/demo/seqToseq/
python preprocess.py -i data/paraphrase [--mergeDict]
</pre></div>
</div>
<ul class="simple">
<li><code class="docutils literal"><span class="pre">--mergeDict</span></code>: if using this option, the source and target dictionary are merged, i.e, two dictionaries have the same context. Here, as source and target data are all chinese words, this option can be used.</li>
</ul>
</div>
<div class="section" id="user-specified-embedding-model">
<span id="user-specified-embedding-model"></span><h3>User Specified Embedding Model<a class="headerlink" href="#user-specified-embedding-model" title="Permalink to this headline"></a></h3>
<p>The general command of extracting desired parameters from the pretrained embedding model based on user dictionary is:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>cd $PADDLE_ROOT/demo/model_zoo/embedding
python extract_para.py --preModel PREMODEL --preDict PREDICT --usrModel USRMODEL--usrDict USRDICT -d DIM
</pre></div>
</div>
<ul class="simple">
<li><code class="docutils literal"><span class="pre">--preModel</span> <span class="pre">PREMODEL</span></code>: the name of pretrained embedding model</li>
<li><code class="docutils literal"><span class="pre">--preDict</span> <span class="pre">PREDICT</span></code>: the name of pretrained dictionary</li>
<li><code class="docutils literal"><span class="pre">--usrModel</span> <span class="pre">USRMODEL</span></code>: the name of extracted embedding model</li>
<li><code class="docutils literal"><span class="pre">--usrDict</span> <span class="pre">USRDICT</span></code>: the name of user specified dictionary</li>
<li><code class="docutils literal"><span class="pre">-d</span> <span class="pre">DIM</span></code>: dimension of parameter</li>
</ul>
<p>Here, you can simply run the command:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>cd $PADDLE_ROOT/demo/seqToseq/data/
Y
Upload docs  
Yu Yang 已提交
124 
./paraphrase_model.sh
Y
Update docs  
Yu Yang 已提交
125 126 127 
</pre></div>
</div>
<p>And you will see following embedding model structure:</p>
Y
Upload docs  
Yu Yang 已提交
128 
<div class="highlight-python"><div class="highlight"><pre><span></span>paraphrase_model
Y
Update docs  
Yu Yang 已提交
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 
|--- _source_language_embedding
|--- _target_language_embedding
</pre></div>
</div>
</div>
<div class="section" id="training-model-in-paddlepaddle">
<span id="training-model-in-paddlepaddle"></span><h3>Training Model in PaddlePaddle<a class="headerlink" href="#training-model-in-paddlepaddle" title="Permalink to this headline"></a></h3>
<p>First, create a model config file, see example <code class="docutils literal"><span class="pre">demo/seqToseq/paraphrase/train.conf</span></code>:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">seqToseq_net</span> <span class="kn">import</span> <span class="o">*</span>
<span class="n">is_generating</span> <span class="o">=</span> <span class="bp">False</span>

<span class="c1">################## Data Definition #####################</span>
<span class="n">train_conf</span> <span class="o">=</span> <span class="n">seq_to_seq_data</span><span class="p">(</span><span class="n">data_dir</span> <span class="o">=</span> <span class="s2">&quot;./data/pre-paraphrase&quot;</span><span class="p">,</span>
                             <span class="n">job_mode</span> <span class="o">=</span> <span class="n">job_mode</span><span class="p">)</span>

<span class="c1">############## Algorithm Configuration ##################</span>
<span class="n">settings</span><span class="p">(</span>
      <span class="n">learning_method</span> <span class="o">=</span> <span class="n">AdamOptimizer</span><span class="p">(),</span>
      <span class="n">batch_size</span> <span class="o">=</span> <span class="mi">50</span><span class="p">,</span>
      <span class="n">learning_rate</span> <span class="o">=</span> <span class="mf">5e-4</span><span class="p">)</span>

<span class="c1">################# Network configure #####################</span>
<span class="n">gru_encoder_decoder</span><span class="p">(</span><span class="n">train_conf</span><span class="p">,</span> <span class="n">is_generating</span><span class="p">,</span> <span class="n">word_vector_dim</span> <span class="o">=</span> <span class="mi">32</span><span class="p">)</span>
</pre></div>
</div>
<p>This config is almost the same as <code class="docutils literal"><span class="pre">demo/seqToseq/translation/train.conf</span></code>.</p>
<p>Then, train the model by running the command:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>cd $PADDLE_SOURCE_ROOT/demo/seqToseq/paraphrase
./train.sh
</pre></div>
</div>
<p>where <code class="docutils literal"><span class="pre">train.sh</span></code> is almost the same as <code class="docutils literal"><span class="pre">demo/seqToseq/translation/train.sh</span></code>, the only difference is following two command arguments:</p>
<ul class="simple">
Y
Upload docs  
Yu Yang 已提交
162 
<li><code class="docutils literal"><span class="pre">--init_model_path</span></code>: path of the initialization model, here is <code class="docutils literal"><span class="pre">data/paraphrase_model</span></code></li>
Y
Update docs  
Yu Yang 已提交
163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 
<li><code class="docutils literal"><span class="pre">--load_missing_parameter_strategy</span></code>: operations when model file is missing, here use a normal distibution to initialize the other parameters except for the embedding layer</li>
</ul>
<p>For users who want to understand the dataset format, model architecture and training procedure in detail, please refer to <a class="reference external" href="demo/embedding_model/text_generation.md">Text generation Tutorial</a>.</p>
</div>
</div>
<div class="section" id="optional-function">
<span id="optional-function"></span><h2>Optional Function<a class="headerlink" href="#optional-function" title="Permalink to this headline"></a></h2>
<div class="section" id="embedding-parameters-observation">
<span id="embedding-parameters-observation"></span><h3>Embedding Parameters Observation<a class="headerlink" href="#embedding-parameters-observation" title="Permalink to this headline"></a></h3>
<p>For users who want to observe the embedding parameters, this function can convert a PaddlePaddle binary embedding model to a text model by running the command:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>cd $PADDLE_ROOT/demo/model_zoo/embedding
python paraconvert.py --b2t -i INPUT -o OUTPUT -d DIM
</pre></div>
</div>
<ul class="simple">
<li><code class="docutils literal"><span class="pre">-i</span> <span class="pre">INPUT</span></code>: the name of input binary embedding model</li>
<li><code class="docutils literal"><span class="pre">-o</span> <span class="pre">OUTPUT</span></code>: the name of output text embedding model</li>
<li><code class="docutils literal"><span class="pre">-d</span> <span class="pre">DIM</span></code>: the dimension of parameter</li>
</ul>
<p>You will see parameters like this in output text model:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="mi">0</span><span class="p">,</span><span class="mi">4</span><span class="p">,</span><span class="mi">32156096</span>
<span class="o">-</span><span class="mf">0.7845433</span><span class="p">,</span><span class="mf">1.1937413</span><span class="p">,</span><span class="o">-</span><span class="mf">0.1704215</span><span class="p">,</span><span class="mf">0.4154715</span><span class="p">,</span><span class="mf">0.9566584</span><span class="p">,</span><span class="o">-</span><span class="mf">0.5558153</span><span class="p">,</span><span class="o">-</span><span class="mf">0.2503305</span><span class="p">,</span> <span class="o">......</span>
<span class="mf">0.0000909</span><span class="p">,</span><span class="mf">0.0009465</span><span class="p">,</span><span class="o">-</span><span class="mf">0.0008813</span><span class="p">,</span><span class="o">-</span><span class="mf">0.0008428</span><span class="p">,</span><span class="mf">0.0007879</span><span class="p">,</span><span class="mf">0.0000183</span><span class="p">,</span><span class="mf">0.0001984</span><span class="p">,</span> <span class="o">......</span>
<span class="o">......</span>
</pre></div>
</div>
<ul class="simple">
<li>1st line is <strong>PaddlePaddle format file head</strong>, it has 3 attributes:<ul>
<li>version of PaddlePaddle, here is 0</li>
<li>sizeof(float), here is 4</li>
<li>total number of parameter, here is 32156096</li>
</ul>
</li>
<li>Other lines print the paramters (assume <code class="docutils literal"><span class="pre">&lt;dim&gt;</span></code> = 32)<ul>
<li>each line print 32 paramters splitted by &#8216;,&#8217;</li>
<li>there is 32156096/32 = 1004877 lines, meaning there is 1004877 embedding words</li>
</ul>
</li>
</ul>
</div>
<div class="section" id="embedding-parameters-revision">
<span id="embedding-parameters-revision"></span><h3>Embedding Parameters Revision<a class="headerlink" href="#embedding-parameters-revision" title="Permalink to this headline"></a></h3>
<p>For users who want to revise the embedding parameters, this function can convert a revised text embedding model to a PaddlePaddle binary model by running the command:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>cd $PADDLE_ROOT/demo/model_zoo/embedding
python paraconvert.py --t2b -i INPUT -o OUTPUT
</pre></div>
</div>
<ul class="simple">
<li><code class="docutils literal"><span class="pre">-i</span> <span class="pre">INPUT</span></code>: the name of input text embedding model.</li>
<li><code class="docutils literal"><span class="pre">-o</span> <span class="pre">OUTPUT</span></code>: the name of output binary embedding model</li>
</ul>
<p>Note that the format of input text model is as follows:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="o">-</span><span class="mf">0.7845433</span><span class="p">,</span><span class="mf">1.1937413</span><span class="p">,</span><span class="o">-</span><span class="mf">0.1704215</span><span class="p">,</span><span class="mf">0.4154715</span><span class="p">,</span><span class="mf">0.9566584</span><span class="p">,</span><span class="o">-</span><span class="mf">0.5558153</span><span class="p">,</span><span class="o">-</span><span class="mf">0.2503305</span><span class="p">,</span> <span class="o">......</span>
<span class="mf">0.0000909</span><span class="p">,</span><span class="mf">0.0009465</span><span class="p">,</span><span class="o">-</span><span class="mf">0.0008813</span><span class="p">,</span><span class="o">-</span><span class="mf">0.0008428</span><span class="p">,</span><span class="mf">0.0007879</span><span class="p">,</span><span class="mf">0.0000183</span><span class="p">,</span><span class="mf">0.0001984</span><span class="p">,</span> <span class="o">......</span>
<span class="o">......</span>
</pre></div>
</div>
<ul class="simple">
<li>there is no file header in 1st line</li>
<li>each line stores parameters for one word, the separator is commas &#8216;,&#8217;</li>
</ul>
</div>
</div>
</div>


          </div>
        </div>
      </div>
      <div class="sphinxsidebar" role="navigation" aria-label="main navigation">
        <div class="sphinxsidebarwrapper">
  <h3><a href="../../index.html">Table Of Contents</a></h3>
  <ul>
<li><a class="reference internal" href="#">Chinese Word Embedding Model Tutorial</a><ul>
<li><a class="reference internal" href="#introduction">Introduction</a><ul>
<li><a class="reference internal" href="#chinese-word-dictionary">Chinese Word Dictionary</a></li>
<li><a class="reference internal" href="#pretrained-chinese-word-embedding-model">Pretrained Chinese Word Embedding Model</a></li>
<li><a class="reference internal" href="#download-and-extract">Download and Extract</a></li>
</ul>
</li>
<li><a class="reference internal" href="#chinese-paraphrasing-example">Chinese Paraphrasing Example</a><ul>
<li><a class="reference internal" href="#data-preparation-and-preprocess">Data Preparation and Preprocess</a></li>
<li><a class="reference internal" href="#user-specified-embedding-model">User Specified Embedding Model</a></li>
<li><a class="reference internal" href="#training-model-in-paddlepaddle">Training Model in PaddlePaddle</a></li>
</ul>
</li>
<li><a class="reference internal" href="#optional-function">Optional Function</a><ul>
<li><a class="reference internal" href="#embedding-parameters-observation">Embedding Parameters Observation</a></li>
<li><a class="reference internal" href="#embedding-parameters-revision">Embedding Parameters Revision</a></li>
</ul>
</li>
</ul>
</li>
</ul>

  <h4>Previous topic</h4>
  <p class="topless"><a href="../imagenet_model/resnet_model.html"
                        title="previous chapter">Model Zoo - ImageNet</a></p>
  <h4>Next topic</h4>
Y
Upload docs  
Yu Yang 已提交
262 263 
  <p class="topless"><a href="../../cluster/index.html"
                        title="next chapter">Cluster Train</a></p>
Y
Update docs  
Yu Yang 已提交
264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 
  <div role="note" aria-label="source link">
    <h3>This Page</h3>
    <ul class="this-page-menu">
      <li><a href="../../_sources/demo/embedding_model/index.txt"
            rel="nofollow">Show Source</a></li>
    </ul>
   </div>
<div id="searchbox" style="display: none" role="search">
  <h3>Quick search</h3>
    <form class="search" action="../../search.html" method="get">
      <input type="text" name="q" />
      <input type="submit" value="Go" />
      <input type="hidden" name="check_keywords" value="yes" />
      <input type="hidden" name="area" value="default" />
    </form>
    <p class="searchtip" style="font-size: 90%">
    Enter search terms or a module, class or function name.
    </p>
</div>
<script type="text/javascript">$('#searchbox').show(0);</script>
        </div>
      </div>
      <div class="clearer"></div>
    </div>
    <div class="related" role="navigation" aria-label="related navigation">
      <h3>Navigation</h3>
      <ul>
        <li class="right" style="margin-right: 10px">
          <a href="../../genindex.html" title="General Index"
             >index</a></li>
        <li class="right" >
          <a href="../../py-modindex.html" title="Python Module Index"
             >modules</a> |</li>
        <li class="right" >
Y
Upload docs  
Yu Yang 已提交
298 
          <a href="../../cluster/index.html" title="Cluster Train"
Y
Update docs  
Yu Yang 已提交
299 300 301 302 303 304 305 306 307 308 309 310 311 312 
             >next</a> |</li>
        <li class="right" >
          <a href="../imagenet_model/resnet_model.html" title="Model Zoo - ImageNet"
             >previous</a> |</li>
        <li class="nav-item nav-item-0"><a href="../../index.html">PaddlePaddle  documentation</a> &raquo;</li>
          <li class="nav-item nav-item-1"><a href="../index.html" >Examples and demos</a> &raquo;</li> 
      </ul>
    </div>
    <div class="footer" role="contentinfo">
        &copy; Copyright 2016, PaddlePaddle developers.
      Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.3.5.
    </div>
  </body>
</html>