Deploy to GitHub Pages: 68c68ec9

3a9a6be9 · Travis CI · ceed547e · 3a9a6be9 · 3a9a6be9 · 3a9a6be9
8 changed file
--- a/develop/doc/api/v2/config/layer.html
+++ b/develop/doc/api/v2/config/layer.html
@@ -1782,10 +1782,10 @@ each timestep of which is also a sequence.</li>
 </ul>
 <p>Accordingly, AggregateLevel supports two modes:</p>
 <ul class="simple">
-<li><code class="code docutils literal"><span class="pre">AggregateLevel.EACH_TIMESTEP</span></code> means the aggregation acts on each
+<li><code class="code docutils literal"><span class="pre">AggregateLevel.TO_NO_SEQUENCE</span></code> means the aggregation acts on each
 timestep of a sequence, both <code class="code docutils literal"><span class="pre">SUB_SEQUENCE</span></code> and <code class="code docutils literal"><span class="pre">SEQUENCE</span></code> will
 be aggregated to <code class="code docutils literal"><span class="pre">NO_SEQUENCE</span></code>.</li>
-<li><code class="code docutils literal"><span class="pre">AggregateLevel.EACH_SEQUENCE</span></code> means the aggregation acts on each
+<li><code class="code docutils literal"><span class="pre">AggregateLevel.TO_SEQUENCE</span></code> means the aggregation acts on each
 sequence of a nested sequence, <code class="code docutils literal"><span class="pre">SUB_SEQUENCE</span></code> will be aggregated to
 <code class="code docutils literal"><span class="pre">SEQUENCE</span></code>.</li>
 </ul>
@@ -1801,7 +1801,7 @@ sequence of a nested sequence, <code class="code docutils literal"><span class="
 <p>The example usage is:</p>
 <div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">seq_pool</span> <span class="o">=</span> <span class="n">pooling</span><span class="p">(</span><span class="nb">input</span><span class="o">=</span><span class="n">layer</span><span class="p">,</span>
                         <span class="n">pooling_type</span><span class="o">=</span><span class="n">AvgPooling</span><span class="p">(),</span>
-                         <span class="n">agg_level</span><span class="o">=</span><span class="n">AggregateLevel</span><span class="o">.</span><span class="n">EACH_SEQUENCE</span><span class="p">)</span>
+                         <span class="n">agg_level</span><span class="o">=</span><span class="n">AggregateLevel</span><span class="o">.</span><span class="n">TO_NO_SEQUENCE</span><span class="p">)</span>
 </pre></div>
 </div>
 <table class="docutils field-list" frame="void" rules="none">
@@ -1809,8 +1809,8 @@ sequence of a nested sequence, <code class="code docutils literal"><span class="
 <col class="field-body" />
 <tbody valign="top">
 <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
-<li><strong>agg_level</strong> (<em>AggregateLevel</em>) &#8211; AggregateLevel.EACH_TIMESTEP or
+<li><strong>agg_level</strong> (<em>AggregateLevel</em>) &#8211; AggregateLevel.TO_NO_SEQUENCE or
-AggregateLevel.EACH_SEQUENCE</li>
+AggregateLevel.TO_SEQUENCE</li>
 <li><strong>name</strong> (<em>basestring</em>) &#8211; layer name.</li>
 <li><strong>input</strong> (<em>paddle.v2.config_base.Layer</em>) &#8211; input layer name.</li>
 <li><strong>pooling_type</strong> (<em>BasePoolingType|None</em>) &#8211; Type of pooling, MaxPooling(default), AvgPooling,
@@ -2057,11 +2057,11 @@ convolution neural network, and before recurrent neural network.</p>
 <dd><p>Please refer to AggregateLevel first.</p>
 <p>ExpandLevel supports two modes:</p>
 <ul class="simple">
-<li><code class="code docutils literal"><span class="pre">ExpandLevel.FROM_TIMESTEP</span></code> means the expandation acts on each
+<li><code class="code docutils literal"><span class="pre">ExpandLevel.FROM_NO_SEQUENCE</span></code> means the expansion acts on
-timestep of a sequence, <code class="code docutils literal"><span class="pre">NO_SEQUENCE</span></code> will be expanded to
+<code class="code docutils literal"><span class="pre">NO_SEQUENCE</span></code>, which will be expanded to
 <code class="code docutils literal"><span class="pre">SEQUENCE</span></code> or <code class="code docutils literal"><span class="pre">SUB_SEQUENCE</span></code>.</li>
-<li><code class="code docutils literal"><span class="pre">ExpandLevel.FROM_SEQUENCE</span></code> means the expandation acts on each
+<li><code class="code docutils literal"><span class="pre">ExpandLevel.FROM_SEQUENCE</span></code> means the expansion acts on
-sequence of a nested sequence, <code class="code docutils literal"><span class="pre">SEQUENCE</span></code> will be expanded to
+<code class="code docutils literal"><span class="pre">SEQUENCE</span></code>, which will be expanded to
 <code class="code docutils literal"><span class="pre">SUB_SEQUENCE</span></code>.</li>
 </ul>
 </dd></dl>
@@ -2077,7 +2077,7 @@ sequence is one) to sequence data.&#8221;</p>
 <p>The example usage is:</p>
 <div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">expand</span> <span class="o">=</span> <span class="n">expand</span><span class="p">(</span><span class="nb">input</span><span class="o">=</span><span class="n">layer1</span><span class="p">,</span>
                      <span class="n">expand_as</span><span class="o">=</span><span class="n">layer2</span><span class="p">,</span>
-                      <span class="n">expand_level</span><span class="o">=</span><span class="n">ExpandLevel</span><span class="o">.</span><span class="n">FROM_TIMESTEP</span><span class="p">)</span>
+                      <span class="n">expand_level</span><span class="o">=</span><span class="n">ExpandLevel</span><span class="o">.</span><span class="n">FROM_NO_SEQUENCE</span><span class="p">)</span>
 </pre></div>
 </div>
 <table class="docutils field-list" frame="void" rules="none">

--- a/develop/doc/searchindex.js
+++ b/develop/doc/searchindex.js
--- a/develop/doc_cn/_sources/howto/deep_model/rnn/hierarchical_layer_cn.rst.txt
+++ b/develop/doc_cn/_sources/howto/deep_model/rnn/hierarchical_layer_cn.rst.txt
@@ -28,17 +28,17 @@ pooling 的使用示例如下，详细见 :ref:`api_v2.layer_pooling` 配置API
        seq_pool = pooling(input=layer,
                           pooling_type=pooling.Max(),
-                           agg_level=AggregateLevel.EACH_SEQUENCE)
+                           agg_level=AggregateLevel.TO_SEQUENCE)
 - `pooling_type` 目前支持两种，分别是：pooling.Max()和pooling.Avg()。
- `agg_level=AggregateLevel.EACH_TIMESTEP` 时（默认值）：
+- `agg_level=AggregateLevel.TO_NO_SEQUENCE` 时（默认值）：
  - 作用：双层序列经过运算变成一个0层序列，或单层序列经过运算变成一个0层序列
  - 输入：一个双层序列，或一个单层序列
  - 输出：一个0层序列，即整个输入序列（单层或双层）的平均值（或最大值）
- `agg_level=AggregateLevel.EACH_SEQUENCE` 时：
+- `agg_level=AggregateLevel.TO_SEQUENCE` 时：
  - 作用：一个双层序列经过运算变成一个单层序列
  - 输入：必须是一个双层序列
@@ -52,15 +52,15 @@ last_seq 的使用示例如下（ :ref:`api_v2.layer_first_seq` 类似），详
 ..	code-block:: bash
        last = last_seq(input=layer,
-                        agg_level=AggregateLevel.EACH_SEQUENCE)
+                        agg_level=AggregateLevel.TO_SEQUENCE)
- `agg_level=AggregateLevel.EACH_TIMESTEP` 时（默认值）：
+- `agg_level=AggregateLevel.TO_NO_SEQUENCE` 时（默认值）：
  - 作用：一个双层序列经过运算变成一个0层序列，或一个单层序列经过运算变成一个0层序列
  - 输入：一个双层序列或一个单层序列
  - 输出：一个0层序列，即整个输入序列（双层或者单层）最后一个，或第一个元素。
- `agg_level=AggregateLevel.EACH_SEQUENCE` 时：
+- `agg_level=AggregateLevel.TO_SEQUENCE` 时：
  - 作用：一个双层序列经过运算变成一个单层序列
  - 输入：必须是一个双层序列
  - 输出：一个单层序列，其中每个元素是双层序列中每个subseq最后一个（或第一个）元素。
@@ -74,9 +74,9 @@ expand 的使用示例如下，详细见 :ref:`api_v2.layer_expand` 配置API。
        ex = expand(input=layer1,
                    expand_as=layer2,
-                    expand_level=ExpandLevel.FROM_TIMESTEP)
+                    expand_level=ExpandLevel.FROM_NO_SEQUENCE)
- `expand_level=ExpandLevel.FROM_TIMESTEP` 时（默认值）：
+- `expand_level=ExpandLevel.FROM_NO_SEQUENCE` 时（默认值）：
  - 作用：一个0层序列经过运算扩展成一个单层序列，或者一个双层序列
  - 输入：layer1必须是一个0层序列，是待扩展的数据；layer2 可以是一个单层序列，或者是一个双层序列，提供扩展的长度信息

--- a/develop/doc_cn/_sources/howto/deep_model/rnn/hrnn_rnn_api_compare_cn.rst.txt
+++ b/develop/doc_cn/_sources/howto/deep_model/rnn/hrnn_rnn_api_compare_cn.rst.txt
@@ -81,7 +81,7 @@
  * 在本例中，我们将原始数据的每一组，通过\ :code:`recurrent_group`\ 进行拆解，拆解成的每一句话再通过一个LSTM网络。这和单层RNN的配置是等价的。
-* 与单层RNN的配置类似，我们只需要使用LSTM encode成的最后一个向量。所以对\ :code:`recurrent_group`\ 进行了\ :code:`last_seq`\ 操作。但和单层RNN不同，我们是对每一个子序列取最后一个元素，因此\ :code:`agg_level=AggregateLevel.EACH_SEQUENCE`\ 。
+* 与单层RNN的配置类似，我们只需要使用LSTM encode成的最后一个向量。所以对\ :code:`recurrent_group`\ 进行了\ :code:`last_seq`\ 操作。但和单层RNN不同，我们是对每一个子序列取最后一个元素，因此\ :code:`agg_level=AggregateLevel.TO_SEQUENCE`\ 。
 * 至此，\ :code:`lstm_last`\ 便和单层RNN配置中的\ :code:`lstm_last`\ 具有相同的结果了。

--- a/develop/doc_cn/api/v2/config/layer.html
+++ b/develop/doc_cn/api/v2/config/layer.html
@@ -1789,10 +1789,10 @@ each timestep of which is also a sequence.</li>
 </ul>
 <p>Accordingly, AggregateLevel supports two modes:</p>
 <ul class="simple">
-<li><code class="code docutils literal"><span class="pre">AggregateLevel.EACH_TIMESTEP</span></code> means the aggregation acts on each
+<li><code class="code docutils literal"><span class="pre">AggregateLevel.TO_NO_SEQUENCE</span></code> means the aggregation acts on each
 timestep of a sequence, both <code class="code docutils literal"><span class="pre">SUB_SEQUENCE</span></code> and <code class="code docutils literal"><span class="pre">SEQUENCE</span></code> will
 be aggregated to <code class="code docutils literal"><span class="pre">NO_SEQUENCE</span></code>.</li>
-<li><code class="code docutils literal"><span class="pre">AggregateLevel.EACH_SEQUENCE</span></code> means the aggregation acts on each
+<li><code class="code docutils literal"><span class="pre">AggregateLevel.TO_SEQUENCE</span></code> means the aggregation acts on each
 sequence of a nested sequence, <code class="code docutils literal"><span class="pre">SUB_SEQUENCE</span></code> will be aggregated to
 <code class="code docutils literal"><span class="pre">SEQUENCE</span></code>.</li>
 </ul>
@@ -1808,7 +1808,7 @@ sequence of a nested sequence, <code class="code docutils literal"><span class="
 <p>The example usage is:</p>
 <div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">seq_pool</span> <span class="o">=</span> <span class="n">pooling</span><span class="p">(</span><span class="nb">input</span><span class="o">=</span><span class="n">layer</span><span class="p">,</span>
                         <span class="n">pooling_type</span><span class="o">=</span><span class="n">AvgPooling</span><span class="p">(),</span>
-                         <span class="n">agg_level</span><span class="o">=</span><span class="n">AggregateLevel</span><span class="o">.</span><span class="n">EACH_SEQUENCE</span><span class="p">)</span>
+                         <span class="n">agg_level</span><span class="o">=</span><span class="n">AggregateLevel</span><span class="o">.</span><span class="n">TO_NO_SEQUENCE</span><span class="p">)</span>
 </pre></div>
 </div>
 <table class="docutils field-list" frame="void" rules="none">
@@ -1816,8 +1816,8 @@ sequence of a nested sequence, <code class="code docutils literal"><span class="
 <col class="field-body" />
 <tbody valign="top">
 <tr class="field-odd field"><th class="field-name">参数:</th><td class="field-body"><ul class="first simple">
-<li><strong>agg_level</strong> (<em>AggregateLevel</em>) &#8211; AggregateLevel.EACH_TIMESTEP or
+<li><strong>agg_level</strong> (<em>AggregateLevel</em>) &#8211; AggregateLevel.TO_NO_SEQUENCE or
-AggregateLevel.EACH_SEQUENCE</li>
+AggregateLevel.TO_SEQUENCE</li>
 <li><strong>name</strong> (<em>basestring</em>) &#8211; layer name.</li>
 <li><strong>input</strong> (<em>paddle.v2.config_base.Layer</em>) &#8211; input layer name.</li>
 <li><strong>pooling_type</strong> (<em>BasePoolingType|None</em>) &#8211; Type of pooling, MaxPooling(default), AvgPooling,
@@ -2064,11 +2064,11 @@ convolution neural network, and before recurrent neural network.</p>
 <dd><p>Please refer to AggregateLevel first.</p>
 <p>ExpandLevel supports two modes:</p>
 <ul class="simple">
-<li><code class="code docutils literal"><span class="pre">ExpandLevel.FROM_TIMESTEP</span></code> means the expandation acts on each
+<li><code class="code docutils literal"><span class="pre">ExpandLevel.FROM_NO_SEQUENCE</span></code> means the expansion acts on
-timestep of a sequence, <code class="code docutils literal"><span class="pre">NO_SEQUENCE</span></code> will be expanded to
+<code class="code docutils literal"><span class="pre">NO_SEQUENCE</span></code>, which will be expanded to
 <code class="code docutils literal"><span class="pre">SEQUENCE</span></code> or <code class="code docutils literal"><span class="pre">SUB_SEQUENCE</span></code>.</li>
-<li><code class="code docutils literal"><span class="pre">ExpandLevel.FROM_SEQUENCE</span></code> means the expandation acts on each
+<li><code class="code docutils literal"><span class="pre">ExpandLevel.FROM_SEQUENCE</span></code> means the expansion acts on
-sequence of a nested sequence, <code class="code docutils literal"><span class="pre">SEQUENCE</span></code> will be expanded to
+<code class="code docutils literal"><span class="pre">SEQUENCE</span></code>, which will be expanded to
 <code class="code docutils literal"><span class="pre">SUB_SEQUENCE</span></code>.</li>
 </ul>
 </dd></dl>
@@ -2084,7 +2084,7 @@ sequence is one) to sequence data.&#8221;</p>
 <p>The example usage is:</p>
 <div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">expand</span> <span class="o">=</span> <span class="n">expand</span><span class="p">(</span><span class="nb">input</span><span class="o">=</span><span class="n">layer1</span><span class="p">,</span>
                      <span class="n">expand_as</span><span class="o">=</span><span class="n">layer2</span><span class="p">,</span>
-                      <span class="n">expand_level</span><span class="o">=</span><span class="n">ExpandLevel</span><span class="o">.</span><span class="n">FROM_TIMESTEP</span><span class="p">)</span>
+                      <span class="n">expand_level</span><span class="o">=</span><span class="n">ExpandLevel</span><span class="o">.</span><span class="n">FROM_NO_SEQUENCE</span><span class="p">)</span>
 </pre></div>
 </div>
 <table class="docutils field-list" frame="void" rules="none">

--- a/develop/doc_cn/howto/deep_model/rnn/hierarchical_layer_cn.html
+++ b/develop/doc_cn/howto/deep_model/rnn/hierarchical_layer_cn.html
@@ -221,18 +221,18 @@
 <p>pooling 的使用示例如下，详细见 <a class="reference internal" href="../../../api/v2/config/layer.html#api-v2-layer-pooling"><span class="std std-ref">pooling</span></a> 配置API。</p>
 <div class="highlight-bash"><div class="highlight"><pre><span></span><span class="nv">seq_pool</span> <span class="o">=</span> pooling<span class="o">(</span><span class="nv">input</span><span class="o">=</span>layer,
                   <span class="nv">pooling_type</span><span class="o">=</span>pooling.Max<span class="o">()</span>,
-                   <span class="nv">agg_level</span><span class="o">=</span>AggregateLevel.EACH_SEQUENCE<span class="o">)</span>
+                   <span class="nv">agg_level</span><span class="o">=</span>AggregateLevel.TO_SEQUENCE<span class="o">)</span>
 </pre></div>
 </div>
 <ul class="simple">
 <li><cite>pooling_type</cite> 目前支持两种，分别是：pooling.Max()和pooling.Avg()。</li>
-<li><cite>agg_level=AggregateLevel.EACH_TIMESTEP</cite> 时（默认值）：<ul>
+<li><cite>agg_level=AggregateLevel.TO_NO_SEQUENCE</cite> 时（默认值）：<ul>
 <li>作用：双层序列经过运算变成一个0层序列，或单层序列经过运算变成一个0层序列</li>
 <li>输入：一个双层序列，或一个单层序列</li>
 <li>输出：一个0层序列，即整个输入序列（单层或双层）的平均值（或最大值）</li>
 </ul>
 </li>
-<li><cite>agg_level=AggregateLevel.EACH_SEQUENCE</cite> 时：<ul>
+<li><cite>agg_level=AggregateLevel.TO_SEQUENCE</cite> 时：<ul>
 <li>作用：一个双层序列经过运算变成一个单层序列</li>
 <li>输入：必须是一个双层序列</li>
 <li>输出：一个单层序列，序列的每个元素是原来双层序列每个subseq元素的平均值（或最大值）</li>
@@ -244,17 +244,17 @@
 <h2><a class="toc-backref" href="#id5">last_seq 和 first_seq</a><a class="headerlink" href="#last-seq-first-seq" title="永久链接至标题">¶</a></h2>
 <p>last_seq 的使用示例如下（ <a class="reference internal" href="../../../api/v2/config/layer.html#api-v2-layer-first-seq"><span class="std std-ref">first_seq</span></a> 类似），详细见 <a class="reference internal" href="../../../api/v2/config/layer.html#api-v2-layer-last-seq"><span class="std std-ref">last_seq</span></a> 配置API。</p>
 <div class="highlight-bash"><div class="highlight"><pre><span></span><span class="nv">last</span> <span class="o">=</span> last_seq<span class="o">(</span><span class="nv">input</span><span class="o">=</span>layer,
-                <span class="nv">agg_level</span><span class="o">=</span>AggregateLevel.EACH_SEQUENCE<span class="o">)</span>
+                <span class="nv">agg_level</span><span class="o">=</span>AggregateLevel.TO_SEQUENCE<span class="o">)</span>
 </pre></div>
 </div>
 <ul class="simple">
-<li><cite>agg_level=AggregateLevel.EACH_TIMESTEP</cite> 时（默认值）：<ul>
+<li><cite>agg_level=AggregateLevel.TO_NO_SEQUENCE</cite> 时（默认值）：<ul>
 <li>作用：一个双层序列经过运算变成一个0层序列，或一个单层序列经过运算变成一个0层序列</li>
 <li>输入：一个双层序列或一个单层序列</li>
 <li>输出：一个0层序列，即整个输入序列（双层或者单层）最后一个，或第一个元素。</li>
 </ul>
 </li>
-<li><cite>agg_level=AggregateLevel.EACH_SEQUENCE</cite> 时：
+<li><cite>agg_level=AggregateLevel.TO_SEQUENCE</cite> 时：
 - 作用：一个双层序列经过运算变成一个单层序列
 - 输入：必须是一个双层序列
 - 输出：一个单层序列，其中每个元素是双层序列中每个subseq最后一个（或第一个）元素。</li>
@@ -265,11 +265,11 @@
 <p>expand 的使用示例如下，详细见 <a class="reference internal" href="../../../api/v2/config/layer.html#api-v2-layer-expand"><span class="std std-ref">ExpandLevel</span></a> 配置API。</p>
 <div class="highlight-bash"><div class="highlight"><pre><span></span><span class="nv">ex</span> <span class="o">=</span> expand<span class="o">(</span><span class="nv">input</span><span class="o">=</span>layer1,
            <span class="nv">expand_as</span><span class="o">=</span>layer2,
-            <span class="nv">expand_level</span><span class="o">=</span>ExpandLevel.FROM_TIMESTEP<span class="o">)</span>
+            <span class="nv">expand_level</span><span class="o">=</span>ExpandLevel.FROM_NO_SEQUENCE<span class="o">)</span>
 </pre></div>
 </div>
 <ul class="simple">
-<li><cite>expand_level=ExpandLevel.FROM_TIMESTEP</cite> 时（默认值）：<ul>
+<li><cite>expand_level=ExpandLevel.FROM_NO_SEQUENCE</cite> 时（默认值）：<ul>
 <li>作用：一个0层序列经过运算扩展成一个单层序列，或者一个双层序列</li>
 <li>输入：layer1必须是一个0层序列，是待扩展的数据；layer2 可以是一个单层序列，或者是一个双层序列，提供扩展的长度信息</li>
 <li>输出：一个单层序列或一个双层序列，输出序列的类型（双层序列或单层序列）和序列中含有元素的数目同 layer2 一致。若输出是单层序列，单层序列的每个元素（0层序列），都是对layer1元素的拷贝；若输出是双层序列，双层序列每个subseq中每个元素（0层序列），都是对layer1元素的拷贝</li>

--- a/develop/doc_cn/howto/deep_model/rnn/hrnn_rnn_api_compare_cn.html
+++ b/develop/doc_cn/howto/deep_model/rnn/hrnn_rnn_api_compare_cn.html
@@ -419,7 +419,7 @@
 <li>在本例中，我们将原始数据的每一组，通过<code class="code docutils literal"><span class="pre">recurrent_group</span></code>进行拆解，拆解成的每一句话再通过一个LSTM网络。这和单层RNN的配置是等价的。</li>
 </ul>
 </li>
-<li>与单层RNN的配置类似，我们只需要使用LSTM encode成的最后一个向量。所以对<code class="code docutils literal"><span class="pre">recurrent_group</span></code>进行了<code class="code docutils literal"><span class="pre">last_seq</span></code>操作。但和单层RNN不同，我们是对每一个子序列取最后一个元素，因此<code class="code docutils literal"><span class="pre">agg_level=AggregateLevel.EACH_SEQUENCE</span></code>。</li>
+<li>与单层RNN的配置类似，我们只需要使用LSTM encode成的最后一个向量。所以对<code class="code docutils literal"><span class="pre">recurrent_group</span></code>进行了<code class="code docutils literal"><span class="pre">last_seq</span></code>操作。但和单层RNN不同，我们是对每一个子序列取最后一个元素，因此<code class="code docutils literal"><span class="pre">agg_level=AggregateLevel.TO_SEQUENCE</span></code>。</li>
 <li>至此，<code class="code docutils literal"><span class="pre">lstm_last</span></code>便和单层RNN配置中的<code class="code docutils literal"><span class="pre">lstm_last</span></code>具有相同的结果了。</li>
 </ul>
 <div class="highlight-python"><table class="highlighttable"><tr><td class="linenos"><div class="linenodiv"><pre> 1
@@ -472,7 +472,7 @@
 </span><span class="hll">    <span class="nb">input</span><span class="o">=</span><span class="n">SubsequenceInput</span><span class="p">(</span><span class="n">emb_group</span><span class="p">),</span> <span class="n">step</span><span class="o">=</span><span class="n">lstm_group</span><span class="p">,</span> <span class="n">name</span><span class="o">=</span><span class="s2">&quot;lstm_nest_group&quot;</span><span class="p">)</span>
 </span><span class="hll"><span class="c1"># hasSubseq -&gt;(seqlastins) seq</span>
 </span><span class="hll"><span class="n">lstm_last</span> <span class="o">=</span> <span class="n">last_seq</span><span class="p">(</span>
-</span><span class="hll">    <span class="nb">input</span><span class="o">=</span><span class="n">lstm_nest_group</span><span class="p">,</span> <span class="n">agg_level</span><span class="o">=</span><span class="n">AggregateLevel</span><span class="o">.</span><span class="n">EACH_SEQUENCE</span><span class="p">)</span>
+</span><span class="hll">    <span class="nb">input</span><span class="o">=</span><span class="n">lstm_nest_group</span><span class="p">,</span> <span class="n">agg_level</span><span class="o">=</span><span class="n">AggregateLevel</span><span class="o">.</span><span class="n">TO_SEQUENCE</span><span class="p">)</span>
 </span><span class="hll">
 </span><span class="c1"># seq -&gt;(expand) hasSubseq</span>
 </pre></div>

--- a/develop/doc_cn/searchindex.js
+++ b/develop/doc_cn/searchindex.js