Deci-AI
/
super-gradients
connected to https://github.com/Deci-AI/super-gradients.git


  
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

	
            <!DOCTYPE html>
<html class="writer-html5" lang="en" >
<head>
  <meta charset="utf-8" />
  <meta name="viewport" content="width=device-width, initial-scale=1.0" />
  <title>super_gradients.training.utils.distributed_training_utils &mdash; SuperGradients 1.0 documentation</title>
      <link rel="stylesheet" href="../../../../_static/pygments.css" type="text/css" />
      <link rel="stylesheet" href="../../../../_static/css/theme.css" type="text/css" />
      <link rel="stylesheet" href="../../../../_static/graphviz.css" type="text/css" />
  <!--[if lt IE 9]>
    <script src="../../../../_static/js/html5shiv.min.js"></script>
  <![endif]-->
  
        <script data-url_root="../../../../" id="documentation_options" src="../../../../_static/documentation_options.js"></script>
        <script src="../../../../_static/jquery.js"></script>
        <script src="../../../../_static/underscore.js"></script>
        <script src="../../../../_static/doctools.js"></script>
    <script src="../../../../_static/js/theme.js"></script>
    <link rel="index" title="Index" href="../../../../genindex.html" />
    <link rel="search" title="Search" href="../../../../search.html" /> 
</head>

<body class="wy-body-for-nav"> 
  <div class="wy-grid-for-nav">
    <nav data-toggle="wy-nav-shift" class="wy-nav-side">
      <div class="wy-side-scroll">
        <div class="wy-side-nav-search" >
            <a href="../../../../index.html" class="icon icon-home"> SuperGradients
          </a>
<div role="search">
  <form id="rtd-search-form" class="wy-form" action="../../../../search.html" method="get">
    <input type="text" name="q" placeholder="Search docs" />
    <input type="hidden" name="check_keywords" value="yes" />
    <input type="hidden" name="area" value="default" />
  </form>
</div>
        </div><div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="Navigation menu">
              <p class="caption"><span class="caption-text">SuperGradients</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../../../../welcome.html">SuperGradients</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../../super_gradients.common.html">Common</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../../super_gradients.training.html">Training</a></li>
</ul>

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

    <section data-toggle="wy-nav-shift" class="wy-nav-content-wrap"><nav class="wy-nav-top" aria-label="Mobile navigation menu" >
          <i data-toggle="wy-nav-top" class="fa fa-bars"></i>
          <a href="../../../../index.html">SuperGradients</a>
      </nav>

      <div class="wy-nav-content">
        <div class="rst-content">
          <div role="navigation" aria-label="Page navigation">
  <ul class="wy-breadcrumbs">
      <li><a href="../../../../index.html" class="icon icon-home"></a> &raquo;</li>
          <li><a href="../../../index.html">Module code</a> &raquo;</li>
      <li>super_gradients.training.utils.distributed_training_utils</li>
      <li class="wy-breadcrumbs-aside">
      </li>
  </ul>
  <hr/>
</div>
          <div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
           <div itemprop="articleBody">
             
  <h1>Source code for super_gradients.training.utils.distributed_training_utils</h1><div class="highlight"><pre>
<span></span><span class="kn">import</span> <span class="nn">torch</span>
<span class="kn">from</span> <span class="nn">torch.cuda.amp</span> <span class="kn">import</span> <span class="n">autocast</span>
<span class="kn">import</span> <span class="nn">torch.nn</span> <span class="k">as</span> <span class="nn">nn</span>
<span class="kn">import</span> <span class="nn">itertools</span>


<div class="viewcode-block" id="distributed_all_reduce_tensor_average"><a class="viewcode-back" href="../../../../super_gradients.training.utils.html#super_gradients.training.utils.distributed_training_utils.distributed_all_reduce_tensor_average">[docs]</a><span class="k">def</span> <span class="nf">distributed_all_reduce_tensor_average</span><span class="p">(</span><span class="n">tensor</span><span class="p">,</span> <span class="n">n</span><span class="p">):</span>
    <span class="sd">&quot;&quot;&quot;</span>
<span class="sd">    This method performs a reduce operation on multiple nodes running distributed training</span>
<span class="sd">    It first sums all of the results and then divides the summation</span>
<span class="sd">    :param tensor:  The tensor to perform the reduce operation for</span>
<span class="sd">    :param n:  Number of nodes</span>
<span class="sd">    :return:   Averaged tensor from all of the nodes</span>
<span class="sd">    &quot;&quot;&quot;</span>
    <span class="n">rt</span> <span class="o">=</span> <span class="n">tensor</span><span class="o">.</span><span class="n">clone</span><span class="p">()</span>
    <span class="n">torch</span><span class="o">.</span><span class="n">distributed</span><span class="o">.</span><span class="n">all_reduce</span><span class="p">(</span><span class="n">rt</span><span class="p">,</span> <span class="n">op</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">distributed</span><span class="o">.</span><span class="n">ReduceOp</span><span class="o">.</span><span class="n">SUM</span><span class="p">)</span>
    <span class="n">rt</span> <span class="o">/=</span> <span class="n">n</span>
    <span class="k">return</span> <span class="n">rt</span></div>


<div class="viewcode-block" id="reduce_results_tuple_for_ddp"><a class="viewcode-back" href="../../../../super_gradients.training.utils.html#super_gradients.training.utils.distributed_training_utils.reduce_results_tuple_for_ddp">[docs]</a><span class="k">def</span> <span class="nf">reduce_results_tuple_for_ddp</span><span class="p">(</span><span class="n">validation_results_tuple</span><span class="p">,</span> <span class="n">device</span><span class="p">):</span>
    <span class="sd">&quot;&quot;&quot;Gather all validation tuples from the various devices and average them&quot;&quot;&quot;</span>
    <span class="n">validation_results_list</span> <span class="o">=</span> <span class="nb">list</span><span class="p">(</span><span class="n">validation_results_tuple</span><span class="p">)</span>
    <span class="k">for</span> <span class="n">i</span><span class="p">,</span> <span class="n">validation_result</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">(</span><span class="n">validation_results_list</span><span class="p">):</span>
        <span class="n">validation_results_list</span><span class="p">[</span><span class="n">i</span><span class="p">]</span> <span class="o">=</span> <span class="n">distributed_all_reduce_tensor_average</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="n">validation_result</span><span class="p">)</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="n">device</span><span class="p">),</span>
                                                                           <span class="n">torch</span><span class="o">.</span><span class="n">distributed</span><span class="o">.</span><span class="n">get_world_size</span><span class="p">())</span>
    <span class="n">validation_results_tuple</span> <span class="o">=</span> <span class="nb">tuple</span><span class="p">(</span><span class="n">validation_results_list</span><span class="p">)</span>
    <span class="k">return</span> <span class="n">validation_results_tuple</span></div>


<div class="viewcode-block" id="MultiGPUModeAutocastWrapper"><a class="viewcode-back" href="../../../../super_gradients.training.utils.html#super_gradients.training.utils.distributed_training_utils.MultiGPUModeAutocastWrapper">[docs]</a><span class="k">class</span> <span class="nc">MultiGPUModeAutocastWrapper</span><span class="p">():</span>
    <span class="k">def</span> <span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">func</span><span class="p">):</span>
        <span class="bp">self</span><span class="o">.</span><span class="n">func</span> <span class="o">=</span> <span class="n">func</span>

    <span class="k">def</span> <span class="fm">__call__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="o">*</span><span class="n">args</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">):</span>
        <span class="k">with</span> <span class="n">autocast</span><span class="p">():</span>
            <span class="n">out</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">func</span><span class="p">(</span><span class="o">*</span><span class="n">args</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">)</span>
        <span class="k">return</span> <span class="n">out</span></div>


<div class="viewcode-block" id="scaled_all_reduce"><a class="viewcode-back" href="../../../../super_gradients.training.utils.html#super_gradients.training.utils.distributed_training_utils.scaled_all_reduce">[docs]</a><span class="k">def</span> <span class="nf">scaled_all_reduce</span><span class="p">(</span><span class="n">tensors</span><span class="p">:</span> <span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">,</span> <span class="n">num_gpus</span><span class="p">:</span> <span class="nb">int</span><span class="p">):</span>
    <span class="sd">&quot;&quot;&quot;</span>
<span class="sd">    Performs the scaled all_reduce operation on the provided tensors.</span>
<span class="sd">    The input tensors are modified in-place.</span>
<span class="sd">    Currently supports only the sum</span>
<span class="sd">    reduction operator.</span>
<span class="sd">    The reduced values are scaled by the inverse size of the</span>
<span class="sd">    process group (equivalent to num_gpus).</span>
<span class="sd">    &quot;&quot;&quot;</span>
    <span class="c1"># There is no need for reduction in the single-proc case</span>
    <span class="k">if</span> <span class="n">num_gpus</span> <span class="o">==</span> <span class="mi">1</span><span class="p">:</span>
        <span class="k">return</span> <span class="n">tensors</span>

    <span class="c1"># Queue the reductions</span>
    <span class="n">reductions</span> <span class="o">=</span> <span class="p">[]</span>
    <span class="k">for</span> <span class="n">tensor</span> <span class="ow">in</span> <span class="n">tensors</span><span class="p">:</span>
        <span class="n">reduction</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">distributed</span><span class="o">.</span><span class="n">all_reduce</span><span class="p">(</span><span class="n">tensor</span><span class="p">,</span> <span class="n">async_op</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
        <span class="n">reductions</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">reduction</span><span class="p">)</span>

    <span class="c1"># Wait for reductions to finish</span>
    <span class="k">for</span> <span class="n">reduction</span> <span class="ow">in</span> <span class="n">reductions</span><span class="p">:</span>
        <span class="n">reduction</span><span class="o">.</span><span class="n">wait</span><span class="p">()</span>

    <span class="c1"># Scale the results</span>
    <span class="k">for</span> <span class="n">tensor</span> <span class="ow">in</span> <span class="n">tensors</span><span class="p">:</span>
        <span class="n">tensor</span><span class="o">.</span><span class="n">mul_</span><span class="p">(</span><span class="mf">1.0</span> <span class="o">/</span> <span class="n">num_gpus</span><span class="p">)</span>
    <span class="k">return</span> <span class="n">tensors</span></div>


<div class="viewcode-block" id="compute_precise_bn_stats"><a class="viewcode-back" href="../../../../super_gradients.training.utils.html#super_gradients.training.utils.distributed_training_utils.compute_precise_bn_stats">[docs]</a><span class="nd">@torch</span><span class="o">.</span><span class="n">no_grad</span><span class="p">()</span>
<span class="k">def</span> <span class="nf">compute_precise_bn_stats</span><span class="p">(</span><span class="n">model</span><span class="p">:</span> <span class="n">nn</span><span class="o">.</span><span class="n">Module</span><span class="p">,</span> <span class="n">loader</span><span class="p">:</span> <span class="n">torch</span><span class="o">.</span><span class="n">utils</span><span class="o">.</span><span class="n">data</span><span class="o">.</span><span class="n">DataLoader</span><span class="p">,</span> <span class="n">precise_bn_batch_size</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span> <span class="n">num_gpus</span><span class="p">:</span> <span class="nb">int</span><span class="p">):</span>
    <span class="sd">&#39;&#39;&#39;</span>
<span class="sd">    :param model:                   The model being trained (ie: SgModel.net)</span>
<span class="sd">    :param loader:                  Training dataloader (ie: SgModel.train_loader)</span>
<span class="sd">    :param precise_bn_batch_size:   The effective batch size we want to calculate the batchnorm on. For example, if we are training a model</span>
<span class="sd">                                    on 8 gpus, with a batch of 128 on each gpu, a good rule of thumb would be to give it 8192</span>
<span class="sd">                                    (ie: effective_batch_size * num_gpus = batch_per_gpu * num_gpus * num_gpus).</span>
<span class="sd">                                    If precise_bn_batch_size is not provided in the training_params, the latter heuristic</span>
<span class="sd">                                    will be taken.</span>
<span class="sd">    param num_gpus:                 The number of gpus we are training on</span>
<span class="sd">    &#39;&#39;&#39;</span>

    <span class="c1"># Compute the number of minibatches to use</span>
    <span class="n">num_iter</span> <span class="o">=</span> <span class="nb">int</span><span class="p">(</span><span class="n">precise_bn_batch_size</span> <span class="o">/</span> <span class="p">(</span><span class="n">loader</span><span class="o">.</span><span class="n">batch_size</span> <span class="o">*</span> <span class="n">num_gpus</span><span class="p">))</span> <span class="k">if</span> <span class="n">precise_bn_batch_size</span> <span class="k">else</span> <span class="n">num_gpus</span>
    <span class="n">num_iter</span> <span class="o">=</span> <span class="nb">min</span><span class="p">(</span><span class="n">num_iter</span><span class="p">,</span> <span class="nb">len</span><span class="p">(</span><span class="n">loader</span><span class="p">))</span>

    <span class="c1"># Retrieve the BN layers</span>
    <span class="n">bns</span> <span class="o">=</span> <span class="p">[</span><span class="n">m</span> <span class="k">for</span> <span class="n">m</span> <span class="ow">in</span> <span class="n">model</span><span class="o">.</span><span class="n">modules</span><span class="p">()</span> <span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">m</span><span class="p">,</span> <span class="n">torch</span><span class="o">.</span><span class="n">nn</span><span class="o">.</span><span class="n">BatchNorm2d</span><span class="p">)]</span>

    <span class="c1"># Initialize BN stats storage for computing mean(mean(batch)) and mean(var(batch))</span>
    <span class="n">running_means</span> <span class="o">=</span> <span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">zeros_like</span><span class="p">(</span><span class="n">bn</span><span class="o">.</span><span class="n">running_mean</span><span class="p">)</span> <span class="k">for</span> <span class="n">bn</span> <span class="ow">in</span> <span class="n">bns</span><span class="p">]</span>
    <span class="n">running_vars</span> <span class="o">=</span> <span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">zeros_like</span><span class="p">(</span><span class="n">bn</span><span class="o">.</span><span class="n">running_var</span><span class="p">)</span> <span class="k">for</span> <span class="n">bn</span> <span class="ow">in</span> <span class="n">bns</span><span class="p">]</span>

    <span class="c1"># Remember momentum values</span>
    <span class="n">momentums</span> <span class="o">=</span> <span class="p">[</span><span class="n">bn</span><span class="o">.</span><span class="n">momentum</span> <span class="k">for</span> <span class="n">bn</span> <span class="ow">in</span> <span class="n">bns</span><span class="p">]</span>

    <span class="c1"># Set momentum to 1.0 to compute BN stats that only reflect the current batch</span>
    <span class="k">for</span> <span class="n">bn</span> <span class="ow">in</span> <span class="n">bns</span><span class="p">:</span>
        <span class="n">bn</span><span class="o">.</span><span class="n">momentum</span> <span class="o">=</span> <span class="mf">1.0</span>

    <span class="c1"># Average the BN stats for each BN layer over the batches</span>
    <span class="k">for</span> <span class="n">inputs</span><span class="p">,</span> <span class="n">_labels</span> <span class="ow">in</span> <span class="n">itertools</span><span class="o">.</span><span class="n">islice</span><span class="p">(</span><span class="n">loader</span><span class="p">,</span> <span class="n">num_iter</span><span class="p">):</span>
        <span class="n">model</span><span class="p">(</span><span class="n">inputs</span><span class="o">.</span><span class="n">cuda</span><span class="p">())</span>
        <span class="k">for</span> <span class="n">i</span><span class="p">,</span> <span class="n">bn</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">(</span><span class="n">bns</span><span class="p">):</span>
            <span class="n">running_means</span><span class="p">[</span><span class="n">i</span><span class="p">]</span> <span class="o">+=</span> <span class="n">bn</span><span class="o">.</span><span class="n">running_mean</span> <span class="o">/</span> <span class="n">num_iter</span>
            <span class="n">running_vars</span><span class="p">[</span><span class="n">i</span><span class="p">]</span> <span class="o">+=</span> <span class="n">bn</span><span class="o">.</span><span class="n">running_var</span> <span class="o">/</span> <span class="n">num_iter</span>

    <span class="c1"># Sync BN stats across GPUs (no reduction if 1 GPU used)</span>
    <span class="n">running_means</span> <span class="o">=</span> <span class="n">scaled_all_reduce</span><span class="p">(</span><span class="n">running_means</span><span class="p">,</span> <span class="n">num_gpus</span><span class="o">=</span><span class="n">num_gpus</span><span class="p">)</span>
    <span class="n">running_vars</span> <span class="o">=</span> <span class="n">scaled_all_reduce</span><span class="p">(</span><span class="n">running_vars</span><span class="p">,</span> <span class="n">num_gpus</span><span class="o">=</span><span class="n">num_gpus</span><span class="p">)</span>

    <span class="c1"># Set BN stats and restore original momentum values</span>
    <span class="k">for</span> <span class="n">i</span><span class="p">,</span> <span class="n">bn</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">(</span><span class="n">bns</span><span class="p">):</span>
        <span class="n">bn</span><span class="o">.</span><span class="n">running_mean</span> <span class="o">=</span> <span class="n">running_means</span><span class="p">[</span><span class="n">i</span><span class="p">]</span>
        <span class="n">bn</span><span class="o">.</span><span class="n">running_var</span> <span class="o">=</span> <span class="n">running_vars</span><span class="p">[</span><span class="n">i</span><span class="p">]</span>
        <span class="n">bn</span><span class="o">.</span><span class="n">momentum</span> <span class="o">=</span> <span class="n">momentums</span><span class="p">[</span><span class="n">i</span><span class="p">]</span></div>
</pre></div>

           </div>
          </div>
          <footer>

  <hr/>

  <div role="contentinfo">
    <p>&#169; Copyright 2021, SuperGradients team.</p>
  </div>

  Built with <a href="https://www.sphinx-doc.org/">Sphinx</a> using a
    <a href="https://github.com/readthedocs/sphinx_rtd_theme">theme</a>
    provided by <a href="https://readthedocs.org">Read the Docs</a>.
   

</footer>
        </div>
      </div>
    </section>
  </div>
  <script>
      jQuery(function () {
          SphinxRtdTheme.Navigation.enable(true);
      });
  </script> 

</body>
</html>