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
sequence_decoder.html 35.8 KB
Newer Older
T
Deploy to GitHub Pages: 53cb4df0a2b9deb6b1fd5e9c8e2027c4ad27b352  
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 


<!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>Design: Sequence Decoder Generating LoDTensors &mdash; PaddlePaddle  documentation</title>
  

  
  

  

  
  
    

  

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

  
  
        <link rel="index" title="Index"
              href="../../genindex.html"/>
        <link rel="search" title="Search" href="../../search.html"/>
    <link rel="top" title="PaddlePaddle  documentation" 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_en.html">GET STARTED</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../howto/index_en.html">HOW TO</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../api/index_en.html">API</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../mobile/index_en.html">MOBILE</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_en.html">GET STARTED</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../getstarted/build_and_install/index_en.html">Install and Build</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../getstarted/build_and_install/docker_install_en.html">PaddlePaddle in Docker Containers</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../getstarted/build_and_install/build_from_source_en.html">Installing from Sources</a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../howto/index_en.html">HOW TO</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/cmd_parameter/index_en.html">Set Command-line Parameters</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../howto/usage/cmd_parameter/use_case_en.html">Use Case</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../howto/usage/cmd_parameter/arguments_en.html">Argument Outline</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../howto/usage/cmd_parameter/detail_introduction_en.html">Detail Description</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/cluster/cluster_train_en.html">PaddlePaddle Distributed Training</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/k8s/k8s_en.html">Paddle On Kubernetes</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/usage/k8s/k8s_aws_en.html">Distributed PaddlePaddle Training on AWS with Kubernetes</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/dev/build_en.html">Build PaddlePaddle from Source Code and Run Unit Test</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/dev/new_layer_en.html">Write New Layers</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/dev/contribute_to_paddle_en.html">Contribute Code</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/deep_model/rnn/index_en.html">RNN Models</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../howto/deep_model/rnn/rnn_config_en.html">RNN Configuration</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../howto/optimization/gpu_profiling_en.html">Tune GPU Performance</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../api/index_en.html">API</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../api/v2/model_configs.html">Model Configuration</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">Data Reader Interface and DataSets</a><ul>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/data/data_reader.html">Data Reader Interface</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/data/image.html">Image Interface</a></li>
<li class="toctree-l3"><a class="reference internal" href="../../api/v2/data/dataset.html">Dataset</a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../../api/v2/run_logic.html">Training and Inference</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../mobile/index_en.html">MOBILE</a><ul>
<li class="toctree-l2"><a class="reference internal" href="../../mobile/cross_compiling_for_android_en.html">Build PaddlePaddle for Android</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../mobile/cross_compiling_for_raspberry_en.html">Build PaddlePaddle for Raspberry Pi</a></li>
</ul>
</li>
</ul>

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

      

 







<div role="navigation" aria-label="breadcrumbs navigation">
  <ul class="wy-breadcrumbs">
      
    <li>Design: Sequence Decoder Generating LoDTensors</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="design-sequence-decoder-generating-lodtensors">
<span id="design-sequence-decoder-generating-lodtensors"></span><h1>Design: Sequence Decoder Generating LoDTensors<a class="headerlink" href="#design-sequence-decoder-generating-lodtensors" title="Permalink to this headline"></a></h1>
<p>In tasks such as machine translation and image to text,
a <a class="reference external" href="https://github.com/PaddlePaddle/book/blob/develop/08.machine_translation/README.md">sequence decoder</a> is necessary to generate sequences.</p>
<p>This documentation describes how to implement the sequence decoder as an operator.</p>
<div class="section" id="beam-search-based-decoder">
<span id="beam-search-based-decoder"></span><h2>Beam Search based Decoder<a class="headerlink" href="#beam-search-based-decoder" title="Permalink to this headline"></a></h2>
<p>The <a class="reference external" href="https://en.wikipedia.org/wiki/Beam_search">beam search algorithm</a> is necessary when generating sequences,
it is a heuristic search algorithm that explores the paths by expanding the most promising node in a limited set.</p>
<p>In the old version of PaddlePaddle, a C++ class <code class="docutils literal"><span class="pre">RecurrentGradientMachine</span></code> implements the general sequence decoder based on beam search,
due to the complexity, the implementation relays on a lot of special data structures,
quite trivial and hard to be customized by users.</p>
<p>There are a lot of heuristic tricks in the sequence generation tasks,
so the flexibility of sequence decoder is very important to users.</p>
<p>During PaddlePaddle&#8217;s refactoring work,
some new concept is proposed such as <a class="reference external" href="https://github.com/PaddlePaddle/Paddle/blob/develop/paddle/framework/lod_tensor.md">LoDTensor</a> and <a class="reference external" href="https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/tensor_array.md">TensorArray</a> that can better support sequence usage,
and they can help to make the implementation of beam search based sequence decoder <strong>more transparent and modular</strong> .</p>
<p>For example, the RNN sates, candidates IDs and probabilities of beam search can be represented as <code class="docutils literal"><span class="pre">LoDTensors</span></code>;
the selected candidate&#8217;s IDs in each time step can be stored in a <code class="docutils literal"><span class="pre">TensorArray</span></code>, and <code class="docutils literal"><span class="pre">Packed</span></code> to the sentences translated.</p>
</div>
<div class="section" id="changing-lod-s-absolute-offset-to-relative-offsets">
<span id="changing-lod-s-absolute-offset-to-relative-offsets"></span><h2>Changing LoD&#8217;s absolute offset to relative offsets<a class="headerlink" href="#changing-lod-s-absolute-offset-to-relative-offsets" title="Permalink to this headline"></a></h2>
<p>The current <code class="docutils literal"><span class="pre">LoDTensor</span></code> is designed to store levels of variable-length sequences,
it stores several arrays of integers each represents a level.</p>
<p>The integers in each level represents the begin and end (not inclusive) offset of a sequence <strong>in the underlying tensor</strong>,
let&#8217;s call this format the <strong>absolute-offset LoD</strong> for clear.</p>
<p>The relative-offset LoD can fast retrieve any sequence but fails to represent empty sequences, for example, a two-level LoD is as follows</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">9</span><span class="p">]</span>
 <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">9</span><span class="p">]]</span>
</pre></div>
</div>
<p>The first level tells that there are two sequences:</p>
<ul class="simple">
<li>the first&#8217;s offset is <code class="docutils literal"><span class="pre">[0,</span> <span class="pre">3)</span></code></li>
<li>the second&#8217;s offset is <code class="docutils literal"><span class="pre">[3,</span> <span class="pre">9)</span></code></li>
</ul>
<p>while on the second level, there are several empty sequences that both begin and end at <code class="docutils literal"><span class="pre">3</span></code>.
It is impossible to tell how many empty second-level sequences exist in the first-level sequences.</p>
<p>There are many scenarios that relay on empty sequence representation,
such as machine translation or image to text, one instance has no translations or the empty candidate set for a prefix.</p>
<p>So let&#8217;s introduce another format of LoD,
it stores <strong>the offsets of the lower level sequences</strong> and is called <strong>relative-offset</strong> LoD.</p>
<p>For example, to represent the same sequences of the above data</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">6</span><span class="p">]</span>
 <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">9</span><span class="p">]]</span>
</pre></div>
</div>
<p>the first level represents that there are two sequences,
their offsets in the second-level LoD is <code class="docutils literal"><span class="pre">[0,</span> <span class="pre">3)</span></code> and <code class="docutils literal"><span class="pre">[3,</span> <span class="pre">5)</span></code>.</p>
<p>The second level is the same with the relative offset example because the lower level is a tensor.
It is easy to find out the second sequence in the first-level LoD has two empty sequences.</p>
<p>The following demos are based on relative-offset LoD.</p>
</div>
<div class="section" id="usage-in-a-simple-machine-translation-model">
<span id="usage-in-a-simple-machine-translation-model"></span><h2>Usage in a simple machine translation model<a class="headerlink" href="#usage-in-a-simple-machine-translation-model" title="Permalink to this headline"></a></h2>
<p>Let&#8217;s start from a simple machine translation model that is simplified from <a class="reference external" href="https://github.com/PaddlePaddle/book/tree/develop/08.machine_translation">machine translation chapter</a> to draw a simple blueprint of what a sequence decoder can do and how to use it.</p>
<p>The model has an encoder that learns the semantic vector from a sequence,
and a decoder which uses the sequence decoder to generate new sentences.</p>
<p><strong>Encoder</strong></p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">paddle</span> <span class="kn">as</span> <span class="nn">pd</span>

<span class="n">dict_size</span> <span class="o">=</span> <span class="mi">8000</span>
<span class="n">source_dict_size</span> <span class="o">=</span> <span class="n">dict_size</span>
<span class="n">target_dict_size</span> <span class="o">=</span> <span class="n">dict_size</span>
<span class="n">word_vector_dim</span> <span class="o">=</span> <span class="mi">128</span>
<span class="n">encoder_dim</span> <span class="o">=</span> <span class="mi">128</span>
<span class="n">decoder_dim</span> <span class="o">=</span> <span class="mi">128</span>
<span class="n">beam_size</span> <span class="o">=</span> <span class="mi">5</span>
<span class="n">max_length</span> <span class="o">=</span> <span class="mi">120</span>

<span class="c1"># encoder</span>
<span class="n">src_word_id</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">data</span><span class="p">(</span>
    <span class="n">name</span><span class="o">=</span><span class="s1">&#39;source_language_word&#39;</span><span class="p">,</span>
    <span class="nb">type</span><span class="o">=</span><span class="n">pd</span><span class="o">.</span><span class="n">data</span><span class="o">.</span><span class="n">integer_value_sequence</span><span class="p">(</span><span class="n">source_dict_dim</span><span class="p">))</span>
<span class="n">src_embedding</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">embedding</span><span class="p">(</span><span class="n">size</span><span class="o">=</span><span class="n">source_dict_size</span><span class="p">,</span> <span class="n">size</span><span class="o">=</span><span class="n">word_vector_dim</span><span class="p">)</span>

<span class="n">src_word_vec</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">lookup</span><span class="p">(</span><span class="n">src_embedding</span><span class="p">,</span> <span class="n">src_word_id</span><span class="p">)</span>

<span class="n">encoder_out_seq</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">gru</span><span class="p">(</span><span class="nb">input</span><span class="o">=</span><span class="n">src_word_vec</span><span class="p">,</span> <span class="n">size</span><span class="o">=</span><span class="n">encoder_dim</span><span class="p">)</span>

<span class="n">encoder_ctx</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">last_seq</span><span class="p">(</span><span class="n">encoder_out_seq</span><span class="p">)</span>
<span class="c1"># encoder_ctx_proj is the learned semantic vector</span>
<span class="n">encoder_ctx_proj</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">fc</span><span class="p">(</span>
    <span class="n">encoder_ctx</span><span class="p">,</span> <span class="n">size</span><span class="o">=</span><span class="n">decoder_dim</span><span class="p">,</span> <span class="n">act</span><span class="o">=</span><span class="n">pd</span><span class="o">.</span><span class="n">activation</span><span class="o">.</span><span class="n">Tanh</span><span class="p">(),</span> <span class="n">bias</span><span class="o">=</span><span class="bp">None</span><span class="p">)</span>
</pre></div>
</div>
<p><strong>Decoder</strong></p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="k">def</span> <span class="nf">generate</span><span class="p">():</span>
    <span class="n">decoder</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">while_loop</span><span class="p">()</span>
    <span class="k">with</span> <span class="n">decoder</span><span class="o">.</span><span class="n">step</span><span class="p">():</span>
        <span class="n">decoder_mem</span> <span class="o">=</span> <span class="n">decoder</span><span class="o">.</span><span class="n">memory</span><span class="p">(</span><span class="n">init</span><span class="o">=</span><span class="n">encoder_ctx</span><span class="p">)</span>  <span class="c1"># mark the memory</span>
        <span class="n">generated_ids</span> <span class="o">=</span> <span class="n">decoder</span><span class="o">.</span><span class="n">memory</span><span class="p">()</span> <span class="c1"># TODO init to batch_size &lt;s&gt;s</span>
        <span class="n">generated_scores</span> <span class="o">=</span> <span class="n">decoder</span><span class="o">.</span><span class="n">memory</span><span class="p">()</span> <span class="c1"># TODO init to batch_size 1s or 0s</span>

        <span class="n">target_word</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">lookup</span><span class="p">(</span><span class="n">trg_embedding</span><span class="p">,</span> <span class="n">gendrated_ids</span><span class="p">)</span>
        <span class="c1"># expand encoder_ctx&#39;s batch to fit target_word&#39;s lod</span>
        <span class="c1"># for example</span>
        <span class="c1"># decoder_mem.lod is</span>
        <span class="c1"># [[0 1 3],</span>
        <span class="c1">#  [0 1 3 6]]</span>
        <span class="c1"># its tensor content is [a1 a2 a3 a4 a5]</span>
        <span class="c1"># which means there are 2 sentences to translate</span>
        <span class="c1">#   - the first sentence has 1 translation prefixes, the offsets are [0, 1)</span>
        <span class="c1">#   - the second sentence has 2 translation prefixes, the offsets are [1, 3) and [3, 6)</span>
        <span class="c1"># the target_word.lod is </span>
        <span class="c1"># [[0, 1, 6]</span>
        <span class="c1">#  [0, 2, 4, 7, 9 12]]</span>
        <span class="c1"># which means 2 sentences to translate, each has 1 and 5 prefixes</span>
        <span class="c1"># the first prefix has 2 candidates</span>
        <span class="c1"># the following has 2, 3, 2, 3 candidates</span>
        <span class="c1"># the encoder_ctx_expanded&#39;s content will be</span>
        <span class="c1"># [a1 a1 a2 a2 a3 a3 a3 a4 a4 a5 a5 a5]</span>
        <span class="n">encoder_ctx_expanded</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">lod_expand</span><span class="p">(</span><span class="n">encoder_ctx</span><span class="p">,</span> <span class="n">target_word</span><span class="p">)</span>
        <span class="n">decoder_input</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">fc</span><span class="p">(</span>
            <span class="n">act</span><span class="o">=</span><span class="n">pd</span><span class="o">.</span><span class="n">activation</span><span class="o">.</span><span class="n">Linear</span><span class="p">(),</span>
            <span class="nb">input</span><span class="o">=</span><span class="p">[</span><span class="n">target_word</span><span class="p">,</span> <span class="n">encoder_ctx</span><span class="p">],</span>
            <span class="n">size</span><span class="o">=</span><span class="mi">3</span> <span class="o">*</span> <span class="n">decoder_dim</span><span class="p">)</span>
        <span class="n">gru_out</span><span class="p">,</span> <span class="n">cur_mem</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">gru_step</span><span class="p">(</span>
            <span class="n">decoder_input</span><span class="p">,</span> <span class="n">mem</span><span class="o">=</span><span class="n">decoder_mem</span><span class="p">,</span> <span class="n">size</span><span class="o">=</span><span class="n">decoder_dim</span><span class="p">)</span>
        <span class="n">scores</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">fc</span><span class="p">(</span>
            <span class="n">gru_out</span><span class="p">,</span>
            <span class="n">size</span><span class="o">=</span><span class="n">trg_dic_size</span><span class="p">,</span>
            <span class="n">bias</span><span class="o">=</span><span class="bp">None</span><span class="p">,</span>
            <span class="n">act</span><span class="o">=</span><span class="n">pd</span><span class="o">.</span><span class="n">activation</span><span class="o">.</span><span class="n">Softmax</span><span class="p">())</span>
        <span class="c1"># K is an config</span>
        <span class="n">topk_scores</span><span class="p">,</span> <span class="n">topk_ids</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">top_k</span><span class="p">(</span><span class="n">scores</span><span class="p">,</span> <span class="n">K</span><span class="p">)</span>
        <span class="n">topk_generated_scores</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">add_scalar</span><span class="p">(</span><span class="n">topk_scores</span><span class="p">,</span> <span class="n">generated_scores</span><span class="p">)</span>

        <span class="n">selected_ids</span><span class="p">,</span> <span class="n">selected_generation_scores</span> <span class="o">=</span> <span class="n">decoder</span><span class="o">.</span><span class="n">beam_search</span><span class="p">(</span>
            <span class="n">topk_ids</span><span class="p">,</span> <span class="n">topk_generated_scores</span><span class="p">)</span>

        <span class="c1"># update the states</span>
        <span class="n">decoder_mem</span><span class="o">.</span><span class="n">update</span><span class="p">(</span><span class="n">cur_mem</span><span class="p">)</span>  <span class="c1"># tells how to update state</span>
        <span class="n">generated_ids</span><span class="o">.</span><span class="n">update</span><span class="p">(</span><span class="n">selected_ids</span><span class="p">)</span>
        <span class="n">generated_scores</span><span class="o">.</span><span class="n">update</span><span class="p">(</span><span class="n">selected_generation_scores</span><span class="p">)</span>

        <span class="n">decoder</span><span class="o">.</span><span class="n">output</span><span class="p">(</span><span class="n">selected_ids</span><span class="p">)</span>
        <span class="n">decoder</span><span class="o">.</span><span class="n">output</span><span class="p">(</span><span class="n">selected_generation_scores</span><span class="p">)</span>

<span class="n">translation_ids</span><span class="p">,</span> <span class="n">translation_scores</span> <span class="o">=</span> <span class="n">decoder</span><span class="p">()</span>
</pre></div>
</div>
<p>The <code class="docutils literal"><span class="pre">decoder.beam_search</span></code> is a operator that given the candidates and the scores of translations including the candidates,
return the result of the beam search algorithm.</p>
<p>In this way, users can customize anything on the inputs or outputs of beam search, for example, two ways to prune some translation prefixes</p>
<ol class="simple">
<li>meke the correspondind elements in <code class="docutils literal"><span class="pre">topk_generated_scores</span></code> zero or some small values, beam_search will discard this candidate.</li>
<li>remove some specific candidate in <code class="docutils literal"><span class="pre">selected_ids</span></code></li>
<li>get the final <code class="docutils literal"><span class="pre">translation_ids</span></code>, remove the translation sequence in it.</li>
</ol>
<p>The implementation of sequence decoder can reuse the C++ class <a class="reference external" href="https://github.com/Superjom/Paddle/blob/68cac3c0f8451fe62a4cdf156747d6dc0ee000b3/paddle/operators/dynamic_recurrent_op.h#L30">RNNAlgorithm</a>,
so the python syntax is quite similar to a <a class="reference external" href="https://github.com/Superjom/Paddle/blob/68cac3c0f8451fe62a4cdf156747d6dc0ee000b3/doc/design/block.md#blocks-with-for-and-rnnop">RNN</a>.</p>
<p>Both of them are two-level <code class="docutils literal"><span class="pre">LoDTensors</span></code></p>
<ul class="simple">
<li>the first level represents <code class="docutils literal"><span class="pre">batch_size</span></code> of (source) sentences;</li>
<li>the second level represents the candidate ID sets for translation prefix.</li>
</ul>
<p>for example, 3 source sentences to translate, and has 2, 3, 1 candidates.</p>
<p>Unlike an RNN, in sequence decoder, the previous state and the current state have different LoD and shape,
a <code class="docutils literal"><span class="pre">lod_expand</span></code> operator is used to expand the LoD of the previous state to fit the current state.</p>
<p>For example, the previous state</p>
<ul class="simple">
<li>LoD is <code class="docutils literal"><span class="pre">[0,</span> <span class="pre">1,</span> <span class="pre">3][0,</span> <span class="pre">2,</span> <span class="pre">5,</span> <span class="pre">6]</span></code></li>
<li>content of tensor is <code class="docutils literal"><span class="pre">a1</span> <span class="pre">a2</span> <span class="pre">b1</span> <span class="pre">b2</span> <span class="pre">b3</span> <span class="pre">c1</span></code></li>
</ul>
<p>the current state stored in <code class="docutils literal"><span class="pre">encoder_ctx_expanded</span></code></p>
<ul class="simple">
<li>LoD is <code class="docutils literal"><span class="pre">[0,</span> <span class="pre">2,</span> <span class="pre">7][0</span> <span class="pre">3</span> <span class="pre">5</span> <span class="pre">8</span> <span class="pre">9</span> <span class="pre">11</span> <span class="pre">11]</span></code></li>
<li>the content is<ul>
<li>a1 a1 a1 (a1 has 3 candidates, so the state should be copied 3 times for each candidates)</li>
<li>a2 a2</li>
<li>b1 b1 b1</li>
<li>b2</li>
<li>b3 b3</li>
<li>None (c1 has 0 candidates, so c1 is dropped)</li>
</ul>
</li>
</ul>
<p>Benefit from the relative offset LoD, empty candidate set can be represented naturally.</p>
<p>the status in each time step can be stored in <code class="docutils literal"><span class="pre">TensorArray</span></code>, and <code class="docutils literal"><span class="pre">Pack</span></code>ed to a final LoDTensor, the corresponding syntax is</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">decoder</span><span class="o">.</span><span class="n">output</span><span class="p">(</span><span class="n">selected_ids</span><span class="p">)</span>
<span class="n">decoder</span><span class="o">.</span><span class="n">output</span><span class="p">(</span><span class="n">selected_generation_scores</span><span class="p">)</span>
</pre></div>
</div>
<p>the <code class="docutils literal"><span class="pre">selected_ids</span></code> is the candidate ids for the prefixes,
it will be <code class="docutils literal"><span class="pre">Packed</span></code> by <code class="docutils literal"><span class="pre">TensorArray</span></code> to a two-level <code class="docutils literal"><span class="pre">LoDTensor</span></code>,
the first level represents the source sequences,
the second level represents generated sequences.</p>
<p>Pack the <code class="docutils literal"><span class="pre">selected_scores</span></code> will get a <code class="docutils literal"><span class="pre">LoDTensor</span></code> that stores scores of each candidate of translations.</p>
<p>Pack the <code class="docutils literal"><span class="pre">selected_generation_scores</span></code> will get a <code class="docutils literal"><span class="pre">LoDTensor</span></code>, and each tail is the probability of the translation.</p>
</div>
<div class="section" id="lod-and-shape-changes-during-decoding">
<span id="lod-and-shape-changes-during-decoding"></span><h2>LoD and shape changes during decoding<a class="headerlink" href="#lod-and-shape-changes-during-decoding" title="Permalink to this headline"></a></h2>
<p align="center">
  <img src="./images/LOD-and-shape-changes-during-decoding.jpg"/>
</p><p>According the image above, the only phrase to change LoD is beam search.</p>
</div>
<div class="section" id="beam-search-design">
<span id="beam-search-design"></span><h2>Beam search design<a class="headerlink" href="#beam-search-design" title="Permalink to this headline"></a></h2>
<p>The beam search algorthm will be implemented as one method of the sequence decoder, it has 3 inputs</p>
<ol class="simple">
<li><code class="docutils literal"><span class="pre">topk_ids</span></code>, top K candidate ids for each prefix.</li>
<li><code class="docutils literal"><span class="pre">topk_scores</span></code>, the corresponding scores for <code class="docutils literal"><span class="pre">topk_ids</span></code></li>
<li><code class="docutils literal"><span class="pre">generated_scores</span></code>, the score of the prefixes.</li>
</ol>
<p>All of the are LoDTensors, so that the sequence affilication is clear.
Beam search will keep a beam for each prefix and select a smaller candidate set for each prefix.</p>
<p>It will return three variables</p>
<ol class="simple">
<li><code class="docutils literal"><span class="pre">selected_ids</span></code>, the final candidate beam search function selected for the next step.</li>
<li><code class="docutils literal"><span class="pre">selected_scores</span></code>, the scores for the candidates.</li>
<li><code class="docutils literal"><span class="pre">generated_scores</span></code>, the updated scores for each prefixes (with the new candidates appended).</li>
</ol>
</div>
<div class="section" id="introducing-the-lod-based-pack-and-unpack-methods-in-tensorarray">
<span id="introducing-the-lod-based-pack-and-unpack-methods-in-tensorarray"></span><h2>Introducing the LoD-based <code class="docutils literal"><span class="pre">Pack</span></code> and <code class="docutils literal"><span class="pre">Unpack</span></code> methods in <code class="docutils literal"><span class="pre">TensorArray</span></code><a class="headerlink" href="#introducing-the-lod-based-pack-and-unpack-methods-in-tensorarray" title="Permalink to this headline"></a></h2>
<p>The <code class="docutils literal"><span class="pre">selected_ids</span></code>, <code class="docutils literal"><span class="pre">selected_scores</span></code> and <code class="docutils literal"><span class="pre">generated_scores</span></code> are LoDTensors,
and they exist in each time step,
so it is natural to store them in arrays.</p>
<p>Currently, PaddlePaddle has a module called <code class="docutils literal"><span class="pre">TensorArray</span></code> which can store an array of tensors,
the results of beam search are better to store in a <code class="docutils literal"><span class="pre">TensorArray</span></code>.</p>
<p>The <code class="docutils literal"><span class="pre">Pack</span></code> and <code class="docutils literal"><span class="pre">UnPack</span></code> in <code class="docutils literal"><span class="pre">TensorArray</span></code> are used to package tensors in the array to a <code class="docutils literal"><span class="pre">LoDTensor</span></code> or split the <code class="docutils literal"><span class="pre">LoDTensor</span></code> to an array of tensors.
It needs some extensions to support pack or unpack an array of <code class="docutils literal"><span class="pre">LoDTensors</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,
            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="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.0/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></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>