<!-- HTML header for doxygen 1.8.14-->
<!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/xhtml;charset=UTF-8"/>
<meta http-equiv="X-UA-Compatible" content="IE=9"/>
<meta name="generator" content="Doxygen 1.8.11"/>
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<title>Caffe2 - Python API: caffe2/python/lstm_benchmark.py Source File</title>
<link href="tabs.css" rel="stylesheet" type="text/css"/>
<link rel="icon" href="/static/favicon.png" type="image/x-icon">
<script type="text/javascript" src="jquery.js"></script>
<script type="text/javascript" src="dynsections.js"></script>
<link href="search/search.css" rel="stylesheet" type="text/css"/>
<script type="text/javascript" src="search/searchdata.js"></script>
<script type="text/javascript" src="search/search.js"></script>
<script type="text/javascript">
$(document).ready(function() { init_search(); });
</script>
<link href="stylesheet.css" rel="stylesheet" type="text/css" />
<link href="main.css" rel="stylesheet" type="text/css"/>
</head>
<body>
<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
<div id="titlearea">
<table cellspacing="0" cellpadding="0">
<tbody>
<tr style="height: 56px;">
<td id="projectlogo" width="56"><a href="/"><img alt="Logo" src="Caffe2-with-name-55-tall.png"/></a></td>
<td id="projectalign" style="padding-left: 0.5em;">
<div id="projectname">Caffe2 - Python API
</div>
<div id="projectbrief">A deep learning, cross platform ML framework</div>
</td>
</tr>
</tbody>
</table>
</div>
<!-- end header part -->
<!-- Generated by Doxygen 1.8.11 -->
<script type="text/javascript">
var searchBox = new SearchBox("searchBox", "search",false,'Search');
</script>
<div id="navrow1" class="tabs">
<ul class="tablist">
<li><a href="namespaces.html"><span>Packages</span></a></li>
<li><a href="annotated.html"><span>Classes</span></a></li>
<li class="current"><a href="files.html"><span>Files</span></a></li>
<li><a href="/doxygen-c/html/classes.html"><span>C++ API</span></a></li>
<li><a href="/doxygen-python/html/annotated.html"><span>Python API</span></a></li>
<li><a href="https://github.com/caffe2/caffe2"><span>GitHub</span></a></li>
<li>
<div id="MSearchBox" class="MSearchBoxInactive">
<span class="left">
<img id="MSearchSelect" src="search/mag_sel.png"
onmouseover="return searchBox.OnSearchSelectShow()"
onmouseout="return searchBox.OnSearchSelectHide()"
alt=""/>
<input type="text" id="MSearchField" value="Search" accesskey="S"
onfocus="searchBox.OnSearchFieldFocus(true)"
onblur="searchBox.OnSearchFieldFocus(false)"
onkeyup="searchBox.OnSearchFieldChange(event)"/>
</span><span class="right">
<a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
</span>
</div>
</li>
</ul>
</div>
<div id="navrow2" class="tabs2">
<ul class="tablist">
<li><a href="files.html"><span>File List</span></a></li>
</ul>
</div>
<!-- window showing the filter options -->
<div id="MSearchSelectWindow"
onmouseover="return searchBox.OnSearchSelectShow()"
onmouseout="return searchBox.OnSearchSelectHide()"
onkeydown="return searchBox.OnSearchSelectKey(event)">
</div>
<!-- iframe showing the search results (closed by default) -->
<div id="MSearchResultsWindow">
<iframe src="javascript:void(0)" frameborder="0"
name="MSearchResults" id="MSearchResults">
</iframe>
</div>
<div id="nav-path" class="navpath">
<ul>
<li class="navelem"><a class="el" href="dir_20697b8f204bdfcab31e6b1a416f3ab8.html">caffe2</a></li><li class="navelem"><a class="el" href="dir_f4fda25fc5253eea1ed54677ae5fa2de.html">python</a></li> </ul>
</div>
</div><!-- top -->
<div class="header">
<div class="headertitle">
<div class="title">lstm_benchmark.py</div> </div>
</div><!--header-->
<div class="contents">
<div class="fragment"><div class="line"><a name="l00001"></a><span class="lineno"> 1</span> <span class="comment">## @package lstm_benchmark</span></div><div class="line"><a name="l00002"></a><span class="lineno"> 2</span> <span class="comment"># Module caffe2.python.lstm_benchmark</span></div><div class="line"><a name="l00003"></a><span class="lineno"> 3</span> <span class="keyword">from</span> __future__ <span class="keyword">import</span> absolute_import</div><div class="line"><a name="l00004"></a><span class="lineno"> 4</span> <span class="keyword">from</span> __future__ <span class="keyword">import</span> division</div><div class="line"><a name="l00005"></a><span class="lineno"> 5</span> <span class="keyword">from</span> __future__ <span class="keyword">import</span> print_function</div><div class="line"><a name="l00006"></a><span class="lineno"> 6</span> <span class="keyword">from</span> __future__ <span class="keyword">import</span> unicode_literals</div><div class="line"><a name="l00007"></a><span class="lineno"> 7</span> </div><div class="line"><a name="l00008"></a><span class="lineno"> 8</span> <span class="keyword">from</span> caffe2.proto <span class="keyword">import</span> caffe2_pb2</div><div class="line"><a name="l00009"></a><span class="lineno"> 9</span> <span class="keyword">from</span> <a class="code" href="namespacecaffe2_1_1python.html">caffe2.python</a> <span class="keyword">import</span> workspace, core, utils, rnn_cell, model_helper</div><div class="line"><a name="l00010"></a><span class="lineno"> 10</span> <span class="keyword">from</span> <a class="code" href="namespacecaffe2_1_1python.html">caffe2.python</a> <span class="keyword">import</span> recurrent</div><div class="line"><a name="l00011"></a><span class="lineno"> 11</span> </div><div class="line"><a name="l00012"></a><span class="lineno"> 12</span> <span class="keyword">import</span> argparse</div><div class="line"><a name="l00013"></a><span class="lineno"> 13</span> <span class="keyword">import</span> numpy <span class="keyword">as</span> np</div><div class="line"><a name="l00014"></a><span class="lineno"> 14</span> <span class="keyword">import</span> time</div><div class="line"><a name="l00015"></a><span class="lineno"> 15</span> </div><div class="line"><a name="l00016"></a><span class="lineno"> 16</span> <span class="keyword">import</span> logging</div><div class="line"><a name="l00017"></a><span class="lineno"> 17</span> </div><div class="line"><a name="l00018"></a><span class="lineno"> 18</span> logging.basicConfig()</div><div class="line"><a name="l00019"></a><span class="lineno"> 19</span> log = logging.getLogger(<span class="stringliteral">"lstm_bench"</span>)</div><div class="line"><a name="l00020"></a><span class="lineno"> 20</span> log.setLevel(logging.DEBUG)</div><div class="line"><a name="l00021"></a><span class="lineno"> 21</span> </div><div class="line"><a name="l00022"></a><span class="lineno"> 22</span> </div><div class="line"><a name="l00023"></a><span class="lineno"> 23</span> <span class="keyword">def </span>generate_data(T, shape, num_labels, fixed_shape):</div><div class="line"><a name="l00024"></a><span class="lineno"> 24</span>  <span class="stringliteral">'''</span></div><div class="line"><a name="l00025"></a><span class="lineno"> 25</span> <span class="stringliteral"> Fill a queue with input data</span></div><div class="line"><a name="l00026"></a><span class="lineno"> 26</span> <span class="stringliteral"> '''</span></div><div class="line"><a name="l00027"></a><span class="lineno"> 27</span>  log.info(<span class="stringliteral">"Generating T={} sequence batches"</span>.format(T))</div><div class="line"><a name="l00028"></a><span class="lineno"> 28</span> </div><div class="line"><a name="l00029"></a><span class="lineno"> 29</span>  generate_input_init_net = core.Net(<span class="stringliteral">'generate_input_init'</span>)</div><div class="line"><a name="l00030"></a><span class="lineno"> 30</span>  queue = generate_input_init_net.CreateBlobsQueue(</div><div class="line"><a name="l00031"></a><span class="lineno"> 31</span>  [], <span class="stringliteral">"inputqueue"</span>, num_blobs=1, capacity=T,</div><div class="line"><a name="l00032"></a><span class="lineno"> 32</span>  )</div><div class="line"><a name="l00033"></a><span class="lineno"> 33</span>  label_queue = generate_input_init_net.CreateBlobsQueue(</div><div class="line"><a name="l00034"></a><span class="lineno"> 34</span>  [], <span class="stringliteral">"labelqueue"</span>, num_blobs=1, capacity=T,</div><div class="line"><a name="l00035"></a><span class="lineno"> 35</span>  )</div><div class="line"><a name="l00036"></a><span class="lineno"> 36</span> </div><div class="line"><a name="l00037"></a><span class="lineno"> 37</span>  workspace.RunNetOnce(generate_input_init_net)</div><div class="line"><a name="l00038"></a><span class="lineno"> 38</span>  generate_input_net = core.Net(<span class="stringliteral">'generate_input'</span>)</div><div class="line"><a name="l00039"></a><span class="lineno"> 39</span> </div><div class="line"><a name="l00040"></a><span class="lineno"> 40</span>  generate_input_net.EnqueueBlobs([queue, <span class="stringliteral">"scratch"</span>], [<span class="stringliteral">"scratch"</span>])</div><div class="line"><a name="l00041"></a><span class="lineno"> 41</span>  generate_input_net.EnqueueBlobs([label_queue, <span class="stringliteral">"label_scr"</span>], [<span class="stringliteral">"label_scr"</span>])</div><div class="line"><a name="l00042"></a><span class="lineno"> 42</span>  np.random.seed(2603)</div><div class="line"><a name="l00043"></a><span class="lineno"> 43</span> </div><div class="line"><a name="l00044"></a><span class="lineno"> 44</span>  entry_counts = []</div><div class="line"><a name="l00045"></a><span class="lineno"> 45</span>  <span class="keywordflow">for</span> t <span class="keywordflow">in</span> range(T):</div><div class="line"><a name="l00046"></a><span class="lineno"> 46</span>  <span class="keywordflow">if</span> (t % (max(10, T // 10)) == 0):</div><div class="line"><a name="l00047"></a><span class="lineno"> 47</span>  print(<span class="stringliteral">"Generating data {}/{}"</span>.format(t, T))</div><div class="line"><a name="l00048"></a><span class="lineno"> 48</span>  <span class="comment"># Randomize the seqlength</span></div><div class="line"><a name="l00049"></a><span class="lineno"> 49</span>  random_shape = (</div><div class="line"><a name="l00050"></a><span class="lineno"> 50</span>  [np.random.randint(1, shape[0])] + shape[1:]</div><div class="line"><a name="l00051"></a><span class="lineno"> 51</span>  <span class="keywordflow">if</span> t > 0 <span class="keywordflow">and</span> <span class="keywordflow">not</span> fixed_shape <span class="keywordflow">else</span> shape</div><div class="line"><a name="l00052"></a><span class="lineno"> 52</span>  )</div><div class="line"><a name="l00053"></a><span class="lineno"> 53</span>  X = np.random.rand(*random_shape).astype(np.float32)</div><div class="line"><a name="l00054"></a><span class="lineno"> 54</span>  batch_size = random_shape[1]</div><div class="line"><a name="l00055"></a><span class="lineno"> 55</span>  L = num_labels * batch_size</div><div class="line"><a name="l00056"></a><span class="lineno"> 56</span>  labels = (np.random.rand(random_shape[0]) * L).astype(np.int32)</div><div class="line"><a name="l00057"></a><span class="lineno"> 57</span>  workspace.FeedBlob(<span class="stringliteral">"scratch"</span>, X)</div><div class="line"><a name="l00058"></a><span class="lineno"> 58</span>  workspace.FeedBlob(<span class="stringliteral">"label_scr"</span>, labels)</div><div class="line"><a name="l00059"></a><span class="lineno"> 59</span>  workspace.RunNetOnce(generate_input_net.Proto())</div><div class="line"><a name="l00060"></a><span class="lineno"> 60</span>  entry_counts.append(random_shape[0] * random_shape[1])</div><div class="line"><a name="l00061"></a><span class="lineno"> 61</span> </div><div class="line"><a name="l00062"></a><span class="lineno"> 62</span>  log.info(<span class="stringliteral">"Finished data generation"</span>)</div><div class="line"><a name="l00063"></a><span class="lineno"> 63</span> </div><div class="line"><a name="l00064"></a><span class="lineno"> 64</span>  <span class="keywordflow">return</span> queue, label_queue, entry_counts</div><div class="line"><a name="l00065"></a><span class="lineno"> 65</span> </div><div class="line"><a name="l00066"></a><span class="lineno"> 66</span> </div><div class="line"><a name="l00067"></a><span class="lineno"> 67</span> <span class="keyword">def </span>create_model(args, queue, label_queue, input_shape):</div><div class="line"><a name="l00068"></a><span class="lineno"> 68</span>  model = model_helper.ModelHelper(name=<span class="stringliteral">"LSTM_bench"</span>)</div><div class="line"><a name="l00069"></a><span class="lineno"> 69</span>  seq_lengths, target = \</div><div class="line"><a name="l00070"></a><span class="lineno"> 70</span>  model.net.AddExternalInputs(</div><div class="line"><a name="l00071"></a><span class="lineno"> 71</span>  <span class="stringliteral">'seq_lengths'</span>,</div><div class="line"><a name="l00072"></a><span class="lineno"> 72</span>  <span class="stringliteral">'target'</span>,</div><div class="line"><a name="l00073"></a><span class="lineno"> 73</span>  )</div><div class="line"><a name="l00074"></a><span class="lineno"> 74</span> </div><div class="line"><a name="l00075"></a><span class="lineno"> 75</span>  input_blob = model.net.DequeueBlobs(queue, <span class="stringliteral">"input_data"</span>)</div><div class="line"><a name="l00076"></a><span class="lineno"> 76</span>  labels = model.net.DequeueBlobs(label_queue, <span class="stringliteral">"label"</span>)</div><div class="line"><a name="l00077"></a><span class="lineno"> 77</span> </div><div class="line"><a name="l00078"></a><span class="lineno"> 78</span>  init_blobs = []</div><div class="line"><a name="l00079"></a><span class="lineno"> 79</span>  <span class="keywordflow">if</span> args.implementation <span class="keywordflow">in</span> [<span class="stringliteral">"own"</span>, <span class="stringliteral">"static"</span>, <span class="stringliteral">"static_dag"</span>]:</div><div class="line"><a name="l00080"></a><span class="lineno"> 80</span>  T = <span class="keywordtype">None</span></div><div class="line"><a name="l00081"></a><span class="lineno"> 81</span>  <span class="keywordflow">if</span> <span class="stringliteral">"static"</span> <span class="keywordflow">in</span> args.implementation:</div><div class="line"><a name="l00082"></a><span class="lineno"> 82</span>  <span class="keyword">assert</span> args.fixed_shape, \</div><div class="line"><a name="l00083"></a><span class="lineno"> 83</span>  <span class="stringliteral">"Random input length is not static RNN compatible"</span></div><div class="line"><a name="l00084"></a><span class="lineno"> 84</span>  T = args.seq_length</div><div class="line"><a name="l00085"></a><span class="lineno"> 85</span>  print(<span class="stringliteral">"Using static RNN of size {}"</span>.format(T))</div><div class="line"><a name="l00086"></a><span class="lineno"> 86</span> </div><div class="line"><a name="l00087"></a><span class="lineno"> 87</span>  <span class="keywordflow">for</span> i <span class="keywordflow">in</span> range(args.num_layers):</div><div class="line"><a name="l00088"></a><span class="lineno"> 88</span>  hidden_init, cell_init = model.net.AddExternalInputs(</div><div class="line"><a name="l00089"></a><span class="lineno"> 89</span>  <span class="stringliteral">"hidden_init_{}"</span>.format(i),</div><div class="line"><a name="l00090"></a><span class="lineno"> 90</span>  <span class="stringliteral">"cell_init_{}"</span>.format(i)</div><div class="line"><a name="l00091"></a><span class="lineno"> 91</span>  )</div><div class="line"><a name="l00092"></a><span class="lineno"> 92</span>  init_blobs.extend([hidden_init, cell_init])</div><div class="line"><a name="l00093"></a><span class="lineno"> 93</span> </div><div class="line"><a name="l00094"></a><span class="lineno"> 94</span>  output, last_hidden, _, last_state = rnn_cell.LSTM(</div><div class="line"><a name="l00095"></a><span class="lineno"> 95</span>  model=model,</div><div class="line"><a name="l00096"></a><span class="lineno"> 96</span>  input_blob=input_blob,</div><div class="line"><a name="l00097"></a><span class="lineno"> 97</span>  seq_lengths=seq_lengths,</div><div class="line"><a name="l00098"></a><span class="lineno"> 98</span>  initial_states=init_blobs,</div><div class="line"><a name="l00099"></a><span class="lineno"> 99</span>  dim_in=args.input_dim,</div><div class="line"><a name="l00100"></a><span class="lineno"> 100</span>  dim_out=[args.hidden_dim] * args.num_layers,</div><div class="line"><a name="l00101"></a><span class="lineno"> 101</span>  scope=<span class="stringliteral">"lstm1"</span>,</div><div class="line"><a name="l00102"></a><span class="lineno"> 102</span>  memory_optimization=args.memory_optimization,</div><div class="line"><a name="l00103"></a><span class="lineno"> 103</span>  forward_only=args.forward_only,</div><div class="line"><a name="l00104"></a><span class="lineno"> 104</span>  drop_states=<span class="keyword">True</span>,</div><div class="line"><a name="l00105"></a><span class="lineno"> 105</span>  return_last_layer_only=<span class="keyword">True</span>,</div><div class="line"><a name="l00106"></a><span class="lineno"> 106</span>  static_rnn_unroll_size=T,</div><div class="line"><a name="l00107"></a><span class="lineno"> 107</span>  )</div><div class="line"><a name="l00108"></a><span class="lineno"> 108</span> </div><div class="line"><a name="l00109"></a><span class="lineno"> 109</span>  <span class="keywordflow">if</span> <span class="stringliteral">"dag"</span> <span class="keywordflow">in</span> args.implementation:</div><div class="line"><a name="l00110"></a><span class="lineno"> 110</span>  print(<span class="stringliteral">"Using DAG net type"</span>)</div><div class="line"><a name="l00111"></a><span class="lineno"> 111</span>  model.net.Proto().type = <span class="stringliteral">'dag'</span></div><div class="line"><a name="l00112"></a><span class="lineno"> 112</span>  model.net.Proto().num_workers = 4</div><div class="line"><a name="l00113"></a><span class="lineno"> 113</span> </div><div class="line"><a name="l00114"></a><span class="lineno"> 114</span>  <span class="keywordflow">elif</span> args.implementation == <span class="stringliteral">"cudnn"</span>:</div><div class="line"><a name="l00115"></a><span class="lineno"> 115</span>  <span class="comment"># We need to feed a placeholder input so that RecurrentInitOp</span></div><div class="line"><a name="l00116"></a><span class="lineno"> 116</span>  <span class="comment"># can infer the dimensions.</span></div><div class="line"><a name="l00117"></a><span class="lineno"> 117</span>  init_blobs = model.net.AddExternalInputs(<span class="stringliteral">"hidden_init"</span>, <span class="stringliteral">"cell_init"</span>)</div><div class="line"><a name="l00118"></a><span class="lineno"> 118</span>  model.param_init_net.ConstantFill([], input_blob, shape=input_shape)</div><div class="line"><a name="l00119"></a><span class="lineno"> 119</span>  output, last_hidden, _ = rnn_cell.cudnn_LSTM(</div><div class="line"><a name="l00120"></a><span class="lineno"> 120</span>  model=model,</div><div class="line"><a name="l00121"></a><span class="lineno"> 121</span>  input_blob=input_blob,</div><div class="line"><a name="l00122"></a><span class="lineno"> 122</span>  initial_states=init_blobs,</div><div class="line"><a name="l00123"></a><span class="lineno"> 123</span>  dim_in=args.input_dim,</div><div class="line"><a name="l00124"></a><span class="lineno"> 124</span>  dim_out=args.hidden_dim,</div><div class="line"><a name="l00125"></a><span class="lineno"> 125</span>  scope=<span class="stringliteral">"cudnnlstm"</span>,</div><div class="line"><a name="l00126"></a><span class="lineno"> 126</span>  num_layers=args.num_layers,</div><div class="line"><a name="l00127"></a><span class="lineno"> 127</span>  )</div><div class="line"><a name="l00128"></a><span class="lineno"> 128</span> </div><div class="line"><a name="l00129"></a><span class="lineno"> 129</span>  <span class="keywordflow">else</span>:</div><div class="line"><a name="l00130"></a><span class="lineno"> 130</span>  <span class="keyword">assert</span> <span class="keyword">False</span>, <span class="stringliteral">"Unknown implementation"</span></div><div class="line"><a name="l00131"></a><span class="lineno"> 131</span> </div><div class="line"><a name="l00132"></a><span class="lineno"> 132</span>  weights = model.net.UniformFill(labels, <span class="stringliteral">"weights"</span>)</div><div class="line"><a name="l00133"></a><span class="lineno"> 133</span>  softmax, loss = model.net.SoftmaxWithLoss(</div><div class="line"><a name="l00134"></a><span class="lineno"> 134</span>  [model.Flatten(output), labels, weights],</div><div class="line"><a name="l00135"></a><span class="lineno"> 135</span>  [<span class="stringliteral">'softmax'</span>, <span class="stringliteral">'loss'</span>],</div><div class="line"><a name="l00136"></a><span class="lineno"> 136</span>  )</div><div class="line"><a name="l00137"></a><span class="lineno"> 137</span> </div><div class="line"><a name="l00138"></a><span class="lineno"> 138</span>  <span class="keywordflow">if</span> <span class="keywordflow">not</span> args.forward_only:</div><div class="line"><a name="l00139"></a><span class="lineno"> 139</span>  model.AddGradientOperators([loss])</div><div class="line"><a name="l00140"></a><span class="lineno"> 140</span> </div><div class="line"><a name="l00141"></a><span class="lineno"> 141</span>  <span class="comment"># carry states over</span></div><div class="line"><a name="l00142"></a><span class="lineno"> 142</span>  <span class="keywordflow">for</span> init_blob <span class="keywordflow">in</span> init_blobs:</div><div class="line"><a name="l00143"></a><span class="lineno"> 143</span>  model.net.Copy(last_hidden, init_blob)</div><div class="line"><a name="l00144"></a><span class="lineno"> 144</span> </div><div class="line"><a name="l00145"></a><span class="lineno"> 145</span>  sz = args.hidden_dim</div><div class="line"><a name="l00146"></a><span class="lineno"> 146</span>  <span class="keywordflow">if</span> args.implementation == <span class="stringliteral">"cudnn"</span>:</div><div class="line"><a name="l00147"></a><span class="lineno"> 147</span>  sz *= args.num_layers</div><div class="line"><a name="l00148"></a><span class="lineno"> 148</span>  workspace.FeedBlob(init_blob, np.zeros(</div><div class="line"><a name="l00149"></a><span class="lineno"> 149</span>  [1, args.batch_size, sz], dtype=np.float32</div><div class="line"><a name="l00150"></a><span class="lineno"> 150</span>  ))</div><div class="line"><a name="l00151"></a><span class="lineno"> 151</span> </div><div class="line"><a name="l00152"></a><span class="lineno"> 152</span>  <span class="keywordflow">if</span> args.rnn_executor:</div><div class="line"><a name="l00153"></a><span class="lineno"> 153</span>  <span class="keywordflow">for</span> op <span class="keywordflow">in</span> model.net.Proto().op:</div><div class="line"><a name="l00154"></a><span class="lineno"> 154</span>  <span class="keywordflow">if</span> op.type.startswith(<span class="stringliteral">'RecurrentNetwork'</span>):</div><div class="line"><a name="l00155"></a><span class="lineno"> 155</span>  recurrent.set_rnn_executor_config(</div><div class="line"><a name="l00156"></a><span class="lineno"> 156</span>  op,</div><div class="line"><a name="l00157"></a><span class="lineno"> 157</span>  num_threads=args.rnn_executor_num_threads,</div><div class="line"><a name="l00158"></a><span class="lineno"> 158</span>  max_cuda_streams=args.rnn_executor_max_cuda_streams,</div><div class="line"><a name="l00159"></a><span class="lineno"> 159</span>  )</div><div class="line"><a name="l00160"></a><span class="lineno"> 160</span>  <span class="keywordflow">return</span> model, output</div><div class="line"><a name="l00161"></a><span class="lineno"> 161</span> </div><div class="line"><a name="l00162"></a><span class="lineno"> 162</span> </div><div class="line"><a name="l00163"></a><span class="lineno"> 163</span> <span class="keyword">def </span>Caffe2LSTM(args):</div><div class="line"><a name="l00164"></a><span class="lineno"> 164</span>  T = args.data_size // args.batch_size</div><div class="line"><a name="l00165"></a><span class="lineno"> 165</span> </div><div class="line"><a name="l00166"></a><span class="lineno"> 166</span>  input_blob_shape = [args.seq_length, args.batch_size, args.input_dim]</div><div class="line"><a name="l00167"></a><span class="lineno"> 167</span>  queue, label_queue, entry_counts = generate_data(T // args.seq_length,</div><div class="line"><a name="l00168"></a><span class="lineno"> 168</span>  input_blob_shape,</div><div class="line"><a name="l00169"></a><span class="lineno"> 169</span>  args.hidden_dim,</div><div class="line"><a name="l00170"></a><span class="lineno"> 170</span>  args.fixed_shape)</div><div class="line"><a name="l00171"></a><span class="lineno"> 171</span> </div><div class="line"><a name="l00172"></a><span class="lineno"> 172</span>  workspace.FeedBlob(</div><div class="line"><a name="l00173"></a><span class="lineno"> 173</span>  <span class="stringliteral">"seq_lengths"</span>,</div><div class="line"><a name="l00174"></a><span class="lineno"> 174</span>  np.array([args.seq_length] * args.batch_size, dtype=np.int32)</div><div class="line"><a name="l00175"></a><span class="lineno"> 175</span>  )</div><div class="line"><a name="l00176"></a><span class="lineno"> 176</span> </div><div class="line"><a name="l00177"></a><span class="lineno"> 177</span>  model, output = create_model(args, queue, label_queue, input_blob_shape)</div><div class="line"><a name="l00178"></a><span class="lineno"> 178</span> </div><div class="line"><a name="l00179"></a><span class="lineno"> 179</span>  workspace.RunNetOnce(model.param_init_net)</div><div class="line"><a name="l00180"></a><span class="lineno"> 180</span>  workspace.CreateNet(model.net)</div><div class="line"><a name="l00181"></a><span class="lineno"> 181</span> </div><div class="line"><a name="l00182"></a><span class="lineno"> 182</span>  start_time = time.time()</div><div class="line"><a name="l00183"></a><span class="lineno"> 183</span>  num_iters = T // args.seq_length</div><div class="line"><a name="l00184"></a><span class="lineno"> 184</span>  total_iters = 0</div><div class="line"><a name="l00185"></a><span class="lineno"> 185</span> </div><div class="line"><a name="l00186"></a><span class="lineno"> 186</span>  <span class="comment"># Run the Benchmark</span></div><div class="line"><a name="l00187"></a><span class="lineno"> 187</span>  log.info(<span class="stringliteral">"------ Warming up ------"</span>)</div><div class="line"><a name="l00188"></a><span class="lineno"> 188</span>  workspace.RunNet(model.net.Proto().name)</div><div class="line"><a name="l00189"></a><span class="lineno"> 189</span> </div><div class="line"><a name="l00190"></a><span class="lineno"> 190</span>  <span class="keywordflow">if</span> (args.gpu):</div><div class="line"><a name="l00191"></a><span class="lineno"> 191</span>  log.info(<span class="stringliteral">"Memory stats:"</span>)</div><div class="line"><a name="l00192"></a><span class="lineno"> 192</span>  stats = utils.GetGPUMemoryUsageStats()</div><div class="line"><a name="l00193"></a><span class="lineno"> 193</span>  log.info(<span class="stringliteral">"GPU memory:\t{} MB"</span>.format(stats[<span class="stringliteral">'max_total'</span>] / 1024 / 1024))</div><div class="line"><a name="l00194"></a><span class="lineno"> 194</span> </div><div class="line"><a name="l00195"></a><span class="lineno"> 195</span>  log.info(<span class="stringliteral">"------ Starting benchmark ------"</span>)</div><div class="line"><a name="l00196"></a><span class="lineno"> 196</span>  start_time = time.time()</div><div class="line"><a name="l00197"></a><span class="lineno"> 197</span>  last_time = time.time()</div><div class="line"><a name="l00198"></a><span class="lineno"> 198</span>  <span class="keywordflow">for</span> iteration <span class="keywordflow">in</span> range(1, num_iters, args.iters_to_report):</div><div class="line"><a name="l00199"></a><span class="lineno"> 199</span>  iters_once = min(args.iters_to_report, num_iters - iteration)</div><div class="line"><a name="l00200"></a><span class="lineno"> 200</span>  total_iters += iters_once</div><div class="line"><a name="l00201"></a><span class="lineno"> 201</span>  workspace.RunNet(model.net.Proto().name, iters_once)</div><div class="line"><a name="l00202"></a><span class="lineno"> 202</span> </div><div class="line"><a name="l00203"></a><span class="lineno"> 203</span>  new_time = time.time()</div><div class="line"><a name="l00204"></a><span class="lineno"> 204</span>  log.info(</div><div class="line"><a name="l00205"></a><span class="lineno"> 205</span>  <span class="stringliteral">"Iter: {} / {}. Entries Per Second: {}k."</span>.format(</div><div class="line"><a name="l00206"></a><span class="lineno"> 206</span>  iteration,</div><div class="line"><a name="l00207"></a><span class="lineno"> 207</span>  num_iters,</div><div class="line"><a name="l00208"></a><span class="lineno"> 208</span>  np.sum(entry_counts[iteration:iteration + iters_once]) /</div><div class="line"><a name="l00209"></a><span class="lineno"> 209</span>  (new_time - last_time) // 100 / 10,</div><div class="line"><a name="l00210"></a><span class="lineno"> 210</span>  )</div><div class="line"><a name="l00211"></a><span class="lineno"> 211</span>  )</div><div class="line"><a name="l00212"></a><span class="lineno"> 212</span>  last_time = new_time</div><div class="line"><a name="l00213"></a><span class="lineno"> 213</span> </div><div class="line"><a name="l00214"></a><span class="lineno"> 214</span>  log.info(<span class="stringliteral">"Done. Total EPS excluding 1st iteration: {}k {}"</span>.format(</div><div class="line"><a name="l00215"></a><span class="lineno"> 215</span>  np.sum(entry_counts[1:]) / (time.time() - start_time) // 100 / 10,</div><div class="line"><a name="l00216"></a><span class="lineno"> 216</span>  <span class="stringliteral">" (with RNN executor)"</span> <span class="keywordflow">if</span> args.rnn_executor <span class="keywordflow">else</span> <span class="stringliteral">""</span>,</div><div class="line"><a name="l00217"></a><span class="lineno"> 217</span>  ))</div><div class="line"><a name="l00218"></a><span class="lineno"> 218</span> </div><div class="line"><a name="l00219"></a><span class="lineno"> 219</span>  <span class="keywordflow">if</span> (args.gpu):</div><div class="line"><a name="l00220"></a><span class="lineno"> 220</span>  log.info(<span class="stringliteral">"Memory stats:"</span>)</div><div class="line"><a name="l00221"></a><span class="lineno"> 221</span>  stats = utils.GetGPUMemoryUsageStats()</div><div class="line"><a name="l00222"></a><span class="lineno"> 222</span>  log.info(<span class="stringliteral">"GPU memory:\t{} MB"</span>.format(stats[<span class="stringliteral">'max_total'</span>] / 1024 / 1024))</div><div class="line"><a name="l00223"></a><span class="lineno"> 223</span>  <span class="keywordflow">if</span> (stats[<span class="stringliteral">'max_total'</span>] != stats[<span class="stringliteral">'total'</span>]):</div><div class="line"><a name="l00224"></a><span class="lineno"> 224</span>  log.warning(</div><div class="line"><a name="l00225"></a><span class="lineno"> 225</span>  <span class="stringliteral">"Max usage differs from current total usage: {} > {}"</span>.</div><div class="line"><a name="l00226"></a><span class="lineno"> 226</span>  format(stats[<span class="stringliteral">'max_total'</span>], stats[<span class="stringliteral">'total'</span>])</div><div class="line"><a name="l00227"></a><span class="lineno"> 227</span>  )</div><div class="line"><a name="l00228"></a><span class="lineno"> 228</span>  log.warning(<span class="stringliteral">"This means that costly deallocations occured."</span>)</div><div class="line"><a name="l00229"></a><span class="lineno"> 229</span> </div><div class="line"><a name="l00230"></a><span class="lineno"> 230</span>  <span class="keywordflow">return</span> time.time() - start_time</div><div class="line"><a name="l00231"></a><span class="lineno"> 231</span> </div><div class="line"><a name="l00232"></a><span class="lineno"> 232</span> </div><div class="line"><a name="l00233"></a><span class="lineno"> 233</span> @utils.debug</div><div class="line"><a name="l00234"></a><span class="lineno"> 234</span> <span class="keyword">def </span>Benchmark(args):</div><div class="line"><a name="l00235"></a><span class="lineno"> 235</span>  <span class="keywordflow">return</span> Caffe2LSTM(args)</div><div class="line"><a name="l00236"></a><span class="lineno"> 236</span> </div><div class="line"><a name="l00237"></a><span class="lineno"> 237</span> </div><div class="line"><a name="l00238"></a><span class="lineno"> 238</span> <span class="keyword">def </span>GetArgumentParser():</div><div class="line"><a name="l00239"></a><span class="lineno"> 239</span>  parser = argparse.ArgumentParser(description=<span class="stringliteral">"LSTM benchmark."</span>)</div><div class="line"><a name="l00240"></a><span class="lineno"> 240</span> </div><div class="line"><a name="l00241"></a><span class="lineno"> 241</span>  parser.add_argument(</div><div class="line"><a name="l00242"></a><span class="lineno"> 242</span>  <span class="stringliteral">"--hidden_dim"</span>,</div><div class="line"><a name="l00243"></a><span class="lineno"> 243</span>  type=int,</div><div class="line"><a name="l00244"></a><span class="lineno"> 244</span>  default=800,</div><div class="line"><a name="l00245"></a><span class="lineno"> 245</span>  help=<span class="stringliteral">"Hidden dimension"</span>,</div><div class="line"><a name="l00246"></a><span class="lineno"> 246</span>  )</div><div class="line"><a name="l00247"></a><span class="lineno"> 247</span>  parser.add_argument(</div><div class="line"><a name="l00248"></a><span class="lineno"> 248</span>  <span class="stringliteral">"--input_dim"</span>,</div><div class="line"><a name="l00249"></a><span class="lineno"> 249</span>  type=int,</div><div class="line"><a name="l00250"></a><span class="lineno"> 250</span>  default=40,</div><div class="line"><a name="l00251"></a><span class="lineno"> 251</span>  help=<span class="stringliteral">"Input dimension"</span>,</div><div class="line"><a name="l00252"></a><span class="lineno"> 252</span>  )</div><div class="line"><a name="l00253"></a><span class="lineno"> 253</span>  parser.add_argument(</div><div class="line"><a name="l00254"></a><span class="lineno"> 254</span>  <span class="stringliteral">"--batch_size"</span>,</div><div class="line"><a name="l00255"></a><span class="lineno"> 255</span>  type=int,</div><div class="line"><a name="l00256"></a><span class="lineno"> 256</span>  default=128,</div><div class="line"><a name="l00257"></a><span class="lineno"> 257</span>  help=<span class="stringliteral">"The batch size."</span></div><div class="line"><a name="l00258"></a><span class="lineno"> 258</span>  )</div><div class="line"><a name="l00259"></a><span class="lineno"> 259</span>  parser.add_argument(</div><div class="line"><a name="l00260"></a><span class="lineno"> 260</span>  <span class="stringliteral">"--seq_length"</span>,</div><div class="line"><a name="l00261"></a><span class="lineno"> 261</span>  type=int,</div><div class="line"><a name="l00262"></a><span class="lineno"> 262</span>  default=20,</div><div class="line"><a name="l00263"></a><span class="lineno"> 263</span>  help=<span class="stringliteral">"Max sequence length"</span></div><div class="line"><a name="l00264"></a><span class="lineno"> 264</span>  )</div><div class="line"><a name="l00265"></a><span class="lineno"> 265</span>  parser.add_argument(</div><div class="line"><a name="l00266"></a><span class="lineno"> 266</span>  <span class="stringliteral">"--data_size"</span>,</div><div class="line"><a name="l00267"></a><span class="lineno"> 267</span>  type=int,</div><div class="line"><a name="l00268"></a><span class="lineno"> 268</span>  default=1000000,</div><div class="line"><a name="l00269"></a><span class="lineno"> 269</span>  help=<span class="stringliteral">"Number of data points to generate"</span></div><div class="line"><a name="l00270"></a><span class="lineno"> 270</span>  )</div><div class="line"><a name="l00271"></a><span class="lineno"> 271</span>  parser.add_argument(</div><div class="line"><a name="l00272"></a><span class="lineno"> 272</span>  <span class="stringliteral">"--iters_to_report"</span>,</div><div class="line"><a name="l00273"></a><span class="lineno"> 273</span>  type=int,</div><div class="line"><a name="l00274"></a><span class="lineno"> 274</span>  default=20,</div><div class="line"><a name="l00275"></a><span class="lineno"> 275</span>  help=<span class="stringliteral">"Number of iteration to report progress"</span></div><div class="line"><a name="l00276"></a><span class="lineno"> 276</span>  )</div><div class="line"><a name="l00277"></a><span class="lineno"> 277</span>  parser.add_argument(</div><div class="line"><a name="l00278"></a><span class="lineno"> 278</span>  <span class="stringliteral">"--gpu"</span>,</div><div class="line"><a name="l00279"></a><span class="lineno"> 279</span>  action=<span class="stringliteral">"store_true"</span>,</div><div class="line"><a name="l00280"></a><span class="lineno"> 280</span>  help=<span class="stringliteral">"Run all on GPU"</span>,</div><div class="line"><a name="l00281"></a><span class="lineno"> 281</span>  )</div><div class="line"><a name="l00282"></a><span class="lineno"> 282</span>  parser.add_argument(</div><div class="line"><a name="l00283"></a><span class="lineno"> 283</span>  <span class="stringliteral">"--implementation"</span>,</div><div class="line"><a name="l00284"></a><span class="lineno"> 284</span>  type=str,</div><div class="line"><a name="l00285"></a><span class="lineno"> 285</span>  default=<span class="stringliteral">"own"</span>,</div><div class="line"><a name="l00286"></a><span class="lineno"> 286</span>  help=<span class="stringliteral">"'cudnn', 'own', 'static' or 'static_dag'"</span>,</div><div class="line"><a name="l00287"></a><span class="lineno"> 287</span>  )</div><div class="line"><a name="l00288"></a><span class="lineno"> 288</span>  parser.add_argument(</div><div class="line"><a name="l00289"></a><span class="lineno"> 289</span>  <span class="stringliteral">"--fixed_shape"</span>,</div><div class="line"><a name="l00290"></a><span class="lineno"> 290</span>  action=<span class="stringliteral">"store_true"</span>,</div><div class="line"><a name="l00291"></a><span class="lineno"> 291</span>  help=(<span class="stringliteral">"Whether to randomize shape of input batches. "</span></div><div class="line"><a name="l00292"></a><span class="lineno"> 292</span>  <span class="stringliteral">"Static RNN requires fixed shape"</span>),</div><div class="line"><a name="l00293"></a><span class="lineno"> 293</span>  )</div><div class="line"><a name="l00294"></a><span class="lineno"> 294</span>  parser.add_argument(</div><div class="line"><a name="l00295"></a><span class="lineno"> 295</span>  <span class="stringliteral">"--memory_optimization"</span>,</div><div class="line"><a name="l00296"></a><span class="lineno"> 296</span>  action=<span class="stringliteral">"store_true"</span>,</div><div class="line"><a name="l00297"></a><span class="lineno"> 297</span>  help=<span class="stringliteral">"Whether to use memory optimized LSTM or not"</span>,</div><div class="line"><a name="l00298"></a><span class="lineno"> 298</span>  )</div><div class="line"><a name="l00299"></a><span class="lineno"> 299</span>  parser.add_argument(</div><div class="line"><a name="l00300"></a><span class="lineno"> 300</span>  <span class="stringliteral">"--forward_only"</span>,</div><div class="line"><a name="l00301"></a><span class="lineno"> 301</span>  action=<span class="stringliteral">"store_true"</span>,</div><div class="line"><a name="l00302"></a><span class="lineno"> 302</span>  help=<span class="stringliteral">"Whether to run only forward pass"</span></div><div class="line"><a name="l00303"></a><span class="lineno"> 303</span>  )</div><div class="line"><a name="l00304"></a><span class="lineno"> 304</span>  parser.add_argument(</div><div class="line"><a name="l00305"></a><span class="lineno"> 305</span>  <span class="stringliteral">"--num_layers"</span>,</div><div class="line"><a name="l00306"></a><span class="lineno"> 306</span>  type=int,</div><div class="line"><a name="l00307"></a><span class="lineno"> 307</span>  default=1,</div><div class="line"><a name="l00308"></a><span class="lineno"> 308</span>  help=<span class="stringliteral">"Number of LSTM layers. All output dimensions are going to be"</span></div><div class="line"><a name="l00309"></a><span class="lineno"> 309</span>  <span class="stringliteral">"of hidden_dim size"</span>,</div><div class="line"><a name="l00310"></a><span class="lineno"> 310</span>  )</div><div class="line"><a name="l00311"></a><span class="lineno"> 311</span>  parser.add_argument(</div><div class="line"><a name="l00312"></a><span class="lineno"> 312</span>  <span class="stringliteral">"--rnn_executor"</span>,</div><div class="line"><a name="l00313"></a><span class="lineno"> 313</span>  action=<span class="stringliteral">"store_true"</span>,</div><div class="line"><a name="l00314"></a><span class="lineno"> 314</span>  help=<span class="stringliteral">"Whether to use RNN executor"</span></div><div class="line"><a name="l00315"></a><span class="lineno"> 315</span>  )</div><div class="line"><a name="l00316"></a><span class="lineno"> 316</span>  parser.add_argument(</div><div class="line"><a name="l00317"></a><span class="lineno"> 317</span>  <span class="stringliteral">"--rnn_executor_num_threads"</span>,</div><div class="line"><a name="l00318"></a><span class="lineno"> 318</span>  type=int,</div><div class="line"><a name="l00319"></a><span class="lineno"> 319</span>  default=<span class="keywordtype">None</span>,</div><div class="line"><a name="l00320"></a><span class="lineno"> 320</span>  help=<span class="stringliteral">"Number of threads used by CPU RNN Executor"</span></div><div class="line"><a name="l00321"></a><span class="lineno"> 321</span>  )</div><div class="line"><a name="l00322"></a><span class="lineno"> 322</span>  parser.add_argument(</div><div class="line"><a name="l00323"></a><span class="lineno"> 323</span>  <span class="stringliteral">"--rnn_executor_max_cuda_streams"</span>,</div><div class="line"><a name="l00324"></a><span class="lineno"> 324</span>  type=int,</div><div class="line"><a name="l00325"></a><span class="lineno"> 325</span>  default=<span class="keywordtype">None</span>,</div><div class="line"><a name="l00326"></a><span class="lineno"> 326</span>  help=<span class="stringliteral">"Maximum number of CUDA streams used by RNN executor on GPU"</span></div><div class="line"><a name="l00327"></a><span class="lineno"> 327</span>  )</div><div class="line"><a name="l00328"></a><span class="lineno"> 328</span>  <span class="keywordflow">return</span> parser</div><div class="line"><a name="l00329"></a><span class="lineno"> 329</span> </div><div class="line"><a name="l00330"></a><span class="lineno"> 330</span> </div><div class="line"><a name="l00331"></a><span class="lineno"> 331</span> <span class="keywordflow">if</span> __name__ == <span class="stringliteral">'__main__'</span>:</div><div class="line"><a name="l00332"></a><span class="lineno"> 332</span>  args, extra_args = GetArgumentParser().parse_known_args()</div><div class="line"><a name="l00333"></a><span class="lineno"> 333</span> </div><div class="line"><a name="l00334"></a><span class="lineno"> 334</span>  rnn_executor_opt = 1 <span class="keywordflow">if</span> args.rnn_executor <span class="keywordflow">else</span> 0</div><div class="line"><a name="l00335"></a><span class="lineno"> 335</span> </div><div class="line"><a name="l00336"></a><span class="lineno"> 336</span>  workspace.GlobalInit([</div><div class="line"><a name="l00337"></a><span class="lineno"> 337</span>  <span class="stringliteral">'caffe2'</span>,</div><div class="line"><a name="l00338"></a><span class="lineno"> 338</span>  <span class="stringliteral">'--caffe2_log_level=0'</span>,</div><div class="line"><a name="l00339"></a><span class="lineno"> 339</span>  <span class="stringliteral">'--caffe2_print_blob_sizes_at_exit=0'</span>,</div><div class="line"><a name="l00340"></a><span class="lineno"> 340</span>  <span class="stringliteral">'--caffe2_rnn_executor={}'</span>.format(rnn_executor_opt),</div><div class="line"><a name="l00341"></a><span class="lineno"> 341</span>  <span class="stringliteral">'--caffe2_gpu_memory_tracking=1'</span>] + extra_args)</div><div class="line"><a name="l00342"></a><span class="lineno"> 342</span> </div><div class="line"><a name="l00343"></a><span class="lineno"> 343</span>  device = core.DeviceOption(</div><div class="line"><a name="l00344"></a><span class="lineno"> 344</span>  caffe2_pb2.CUDA <span class="keywordflow">if</span> args.gpu <span class="keywordflow">else</span> caffe2_pb2.CPU, 4)</div><div class="line"><a name="l00345"></a><span class="lineno"> 345</span> </div><div class="line"><a name="l00346"></a><span class="lineno"> 346</span>  with core.DeviceScope(device):</div><div class="line"><a name="l00347"></a><span class="lineno"> 347</span>  Benchmark(args)</div><div class="ttc" id="namespacecaffe2_1_1python_html"><div class="ttname"><a href="namespacecaffe2_1_1python.html">caffe2.python</a></div><div class="ttdef"><b>Definition:</b> <a href="python_2____init_____8py_source.html#l00001">__init__.py:1</a></div></div>
</div><!-- fragment --></div><!-- contents -->
<!-- HTML footer for doxygen 1.8.14-->
<!-- start footer part -->
<hr class="footer"/><address class="footer"><small>
Generated on Thu Apr 19 2018 13:04:04 for Caffe2 - Python API by  <a href="http://www.doxygen.org/index.html">
<img class="footer" src="doxygen.png" alt="doxygen"/>
</a> 1.8.11
</small></address>
<div class="footerContainer">
<div id="footer_wrap" class="wrapper footerWrapper">
<div class="footerBlocks">
<div id="fb_oss" class="footerSection fbOpenSourceFooter">
<svg class="facebookOSSLogoSvg" viewBox="0 0 1133.9 1133.9" x="0px" y="0px" height=50 width=50>
<g>
<path class="logoRing outerRing" d="M 498.3 3.7 c 153.6 88.9 307.3 177.7 461.1 266.2 c 7.6 4.4 10.3 9.1 10.3 17.8 c -0.3 179.1 -0.2 358.3 0 537.4 c 0 8.1 -2.4 12.8 -9.7 17.1 c -154.5 88.9 -308.8 178.1 -462.9 267.5 c -9 5.2 -15.5 5.3 -24.6 0.1 c -153.9 -89.2 -307.9 -178 -462.1 -266.8 C 3 838.8 0 833.9 0 825.1 c 0.3 -179.1 0.2 -358.3 0 -537.4 c 0 -8.6 2.6 -13.6 10.2 -18 C 164.4 180.9 318.4 92 472.4 3 C 477 -1.5 494.3 -0.7 498.3 3.7 Z M 48.8 555.3 c 0 79.9 0.2 159.9 -0.2 239.8 c -0.1 10 3 15.6 11.7 20.6 c 137.2 78.8 274.2 157.8 411 237.3 c 9.9 5.7 17 5.7 26.8 0.1 c 137.5 -79.8 275.2 -159.2 412.9 -238.5 c 7.4 -4.3 10.5 -8.9 10.5 -17.8 c -0.3 -160.2 -0.3 -320.5 0 -480.7 c 0 -8.8 -2.8 -13.6 -10.3 -18 C 772.1 218 633.1 137.8 494.2 57.4 c -6.5 -3.8 -11.5 -4.5 -18.5 -0.5 C 336.8 137.4 197.9 217.7 58.8 297.7 c -7.7 4.4 -10.2 9.2 -10.2 17.9 C 48.9 395.5 48.8 475.4 48.8 555.3 Z" />
<path class="logoRing middleRing" d="M 184.4 555.9 c 0 -33.3 -1 -66.7 0.3 -100 c 1.9 -48 24.1 -86 64.7 -110.9 c 54.8 -33.6 110.7 -65.5 167 -96.6 c 45.7 -25.2 92.9 -24.7 138.6 1 c 54.4 30.6 108.7 61.5 162.2 93.7 c 44 26.5 67.3 66.8 68 118.4 c 0.9 63.2 0.9 126.5 0 189.7 c -0.7 50.6 -23.4 90.7 -66.6 116.9 c -55 33.4 -110.8 65.4 -167.1 96.5 c -43.4 24 -89 24.2 -132.3 0.5 c -57.5 -31.3 -114.2 -64 -170 -98.3 c -41 -25.1 -62.9 -63.7 -64.5 -112.2 C 183.5 621.9 184.3 588.9 184.4 555.9 Z M 232.9 556.3 c 0 29.5 0.5 59.1 -0.1 88.6 c -0.8 39.2 16.9 67.1 50.2 86.2 c 51.2 29.4 102.2 59.2 153.4 88.4 c 31.4 17.9 63.6 18.3 95 0.6 c 53.7 -30.3 107.1 -61.2 160.3 -92.5 c 29.7 -17.5 45 -44.5 45.3 -78.8 c 0.6 -61.7 0.5 -123.5 0 -185.2 c -0.3 -34.4 -15.3 -61.5 -44.9 -79 C 637.7 352.6 583 320.8 527.9 290 c -27.5 -15.4 -57.2 -16.1 -84.7 -0.7 c -56.9 31.6 -113.4 64 -169.1 97.6 c -26.4 15.9 -40.7 41.3 -41.1 72.9 C 232.6 491.9 232.9 524.1 232.9 556.3 Z" />
<path class="logoRing innerRing" d="M 484.9 424.4 c 69.8 -2.8 133.2 57.8 132.6 132 C 617 630 558.5 688.7 484.9 689.1 c -75.1 0.4 -132.6 -63.6 -132.7 -132.7 C 352.1 485 413.4 421.5 484.9 424.4 Z M 401.3 556.7 c -3.4 37.2 30.5 83.6 83 84.1 c 46.6 0.4 84.8 -37.6 84.9 -84 c 0.1 -46.6 -37.2 -84.4 -84.2 -84.6 C 432.2 472.1 397.9 518.3 401.3 556.7 Z" />
</g>
</svg>
<h2>Facebook Open Source</h2>
</div>
<div class="footerSection">
<a class="footerLink" href="https://code.facebook.com/projects/" target="_blank">Open Source Projects</a>
<a class="footerLink" href="https://github.com/facebook/" target="_blank">GitHub</a>
<a class="footerLink" href="https://twitter.com/fbOpenSource" target="_blank">Twitter</a>
</div>
<div class="footerSection rightAlign">
<a class="footerLink" href="https://github.com/caffe2/caffe2" target="_blank">Contribute to this project on GitHub</a>
</div>
</div>
</div>
</div>
<script type="text/javascript" src="/js/jekyll-link-anchors.js"></script>
<script>
(function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
(i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
})(window,document,'script','//www.google-analytics.com/analytics.js','ga');
ga('create', '{{ site.gacode }}', 'auto');
ga('send', 'pageview');
</script>
</body>
</html>