mirror of
https://github.com/wassname/pytorch-lightning.git
synced 2026-09-10 12:21:57 +08:00
Deployed ed31417 with MkDocs version: 1.0.4
This commit is contained in:
William Falcon
@@ -0,0 +1,303 @@
|
||||
<!DOCTYPE html>
|
||||
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
|
||||
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
|
||||
<head>
|
||||
<meta charset="utf-8">
|
||||
<meta http-equiv="X-UA-Compatible" content="IE=edge">
|
||||
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
||||
|
||||
|
||||
<link rel="shortcut icon" href="../../img/favicon.ico">
|
||||
<title>Training Loop - Pytorch lightning Documentation</title>
|
||||
<link href='https://fonts.googleapis.com/css?family=Lato:400,700|Roboto+Slab:400,700|Inconsolata:400,700' rel='stylesheet' type='text/css'>
|
||||
|
||||
<link rel="stylesheet" href="../../css/theme.css" type="text/css" />
|
||||
<link rel="stylesheet" href="../../css/theme_extra.css" type="text/css" />
|
||||
<link rel="stylesheet" href="//cdnjs.cloudflare.com/ajax/libs/highlight.js/9.12.0/styles/github.min.css">
|
||||
|
||||
<script>
|
||||
// Current page data
|
||||
var mkdocs_page_name = "Training Loop";
|
||||
var mkdocs_page_input_path = "Trainer/Training Loop.md";
|
||||
var mkdocs_page_url = null;
|
||||
</script>
|
||||
|
||||
<script src="../../js/jquery-2.1.1.min.js" defer></script>
|
||||
<script src="../../js/modernizr-2.8.3.min.js" defer></script>
|
||||
<script src="//cdnjs.cloudflare.com/ajax/libs/highlight.js/9.12.0/highlight.min.js"></script>
|
||||
<script>hljs.initHighlightingOnLoad();</script>
|
||||
|
||||
</head>
|
||||
|
||||
<body class="wy-body-for-nav" role="document">
|
||||
|
||||
<div class="wy-grid-for-nav">
|
||||
|
||||
|
||||
<nav data-toggle="wy-nav-shift" class="wy-nav-side stickynav">
|
||||
<div class="wy-side-nav-search">
|
||||
<a href="../.." class="icon icon-home"> Pytorch lightning Documentation</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" title="Type search term here" />
|
||||
</form>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
|
||||
<ul class="current">
|
||||
|
||||
|
||||
<li class="toctree-l1">
|
||||
|
||||
<a class="" href="../..">PYTORCH-LIGHTNING DOCUMENTATION</a>
|
||||
</li>
|
||||
|
||||
<li class="toctree-l1">
|
||||
|
||||
<span class="caption-text">Pytorch Lightning</span>
|
||||
<ul class="subnav">
|
||||
<li class="">
|
||||
|
||||
<a class="" href="../../Pytorch-Lightning/LightningModule/">Lightning module</a>
|
||||
</li>
|
||||
</ul>
|
||||
</li>
|
||||
|
||||
<li class="toctree-l1">
|
||||
|
||||
<span class="caption-text">Trainer</span>
|
||||
<ul class="subnav">
|
||||
<li class="">
|
||||
|
||||
<a class="" href="../">Trainer</a>
|
||||
</li>
|
||||
<li class="">
|
||||
|
||||
<a class="" href="../Checkpointing/">Checkpointing</a>
|
||||
</li>
|
||||
<li class="">
|
||||
|
||||
<a class="" href="../Distributed training/">Distributed training</a>
|
||||
</li>
|
||||
<li class="">
|
||||
|
||||
<a class="" href="../SLURM Managed Cluster/">SLURM Managed Cluster</a>
|
||||
</li>
|
||||
<li class=" current">
|
||||
|
||||
<a class="current" href="./">Training Loop</a>
|
||||
<ul class="subnav">
|
||||
|
||||
<li class="toctree-l3"><a href="#accumulated-gradients">Accumulated gradients</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#anneal-learning-rate">Anneal Learning rate</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#check-gpu-usage">Check GPU usage</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#check-which-gradients-are-nan">Check which gradients are nan</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#check-validation-every-n-epochs">Check validation every n epochs</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#display-metrics-in-progress-bar">Display metrics in progress bar</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#display-the-parameter-count-by-layer">Display the parameter count by layer</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#force-training-for-min-or-max-epochs">Force training for min or max epochs</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#inspect-gradient-norms">Inspect gradient norms</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#make-model-overfit-on-subset-of-data">Make model overfit on subset of data</a></li>
|
||||
|
||||
|
||||
<li class="toctree-l3"><a href="#set-how-much-of-the-training-set-to-check">Set how much of the training set to check</a></li>
|
||||
|
||||
|
||||
</ul>
|
||||
</li>
|
||||
<li class="">
|
||||
|
||||
<a class="" href="../Vaildation loop/">Vaildation loop</a>
|
||||
</li>
|
||||
</ul>
|
||||
</li>
|
||||
|
||||
</ul>
|
||||
</div>
|
||||
|
||||
</nav>
|
||||
|
||||
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap">
|
||||
|
||||
|
||||
<nav class="wy-nav-top" role="navigation" aria-label="top navigation">
|
||||
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
|
||||
<a href="../..">Pytorch lightning Documentation</a>
|
||||
</nav>
|
||||
|
||||
|
||||
<div class="wy-nav-content">
|
||||
<div class="rst-content">
|
||||
<div role="navigation" aria-label="breadcrumbs navigation">
|
||||
<ul class="wy-breadcrumbs">
|
||||
<li><a href="../..">Docs</a> »</li>
|
||||
|
||||
|
||||
|
||||
<li>Trainer »</li>
|
||||
|
||||
|
||||
|
||||
<li>Training Loop</li>
|
||||
<li class="wy-breadcrumbs-aside">
|
||||
|
||||
<a href="https://github.com/williamFalcon/pytorch-lightning/edit/master/docs/Trainer/Training Loop.md"
|
||||
class="icon icon-github"> Edit on GitHub</a>
|
||||
|
||||
</li>
|
||||
</ul>
|
||||
<hr/>
|
||||
</div>
|
||||
<div role="main">
|
||||
<div class="section">
|
||||
|
||||
<p>The asdf</p>
|
||||
<hr />
|
||||
<h4 id="accumulated-gradients">Accumulated gradients</h4>
|
||||
<p>Accumulated gradients runs K small batches of size N before doing a backwards pass. The effect is a large effective batch size of size KxN. </p>
|
||||
<pre><code class="python"># DEFAULT (ie: no accumulated grads)
|
||||
trainer = Trainer(accumulate_grad_batches=1)
|
||||
</code></pre>
|
||||
|
||||
<hr />
|
||||
<h4 id="anneal-learning-rate">Anneal Learning rate</h4>
|
||||
<p>Cut the learning rate by 10 at every epoch listed in this list.</p>
|
||||
<pre><code class="python"># DEFAULT (don't anneal)
|
||||
trainer = Trainer(lr_scheduler_milestones=None)
|
||||
|
||||
# cut LR by 10 at 100, 200, and 300 epochs
|
||||
trainer = Trainer(lr_scheduler_milestones=[100, 200, 300])
|
||||
</code></pre>
|
||||
|
||||
<hr />
|
||||
<h4 id="check-gpu-usage">Check GPU usage</h4>
|
||||
<p>Lightning automatically logs gpu usage to the test tube logs. It'll only do it at the metric logging interval, so it doesn't slow down training.</p>
|
||||
<hr />
|
||||
<h4 id="check-which-gradients-are-nan">Check which gradients are nan</h4>
|
||||
<p>This option prints a list of tensors with nan gradients.</p>
|
||||
<pre><code class="python"># DEFAULT
|
||||
trainer = Trainer(print_nan_grads=False)
|
||||
</code></pre>
|
||||
|
||||
<hr />
|
||||
<h4 id="check-validation-every-n-epochs">Check validation every n epochs</h4>
|
||||
<p>If you have a small dataset you might want to check validation every n epochs</p>
|
||||
<pre><code class="python"># DEFAULT
|
||||
trainer = Trainer(check_val_every_n_epoch=1)
|
||||
</code></pre>
|
||||
|
||||
<hr />
|
||||
<h4 id="display-metrics-in-progress-bar">Display metrics in progress bar</h4>
|
||||
<pre><code class="python"># DEFAULT
|
||||
trainer = Trainer(progress_bar=True)
|
||||
</code></pre>
|
||||
|
||||
<hr />
|
||||
<h4 id="display-the-parameter-count-by-layer">Display the parameter count by layer</h4>
|
||||
<p>By default lightning prints a list of parameters <em>and submodules</em> when it starts training.</p>
|
||||
<hr />
|
||||
<h4 id="force-training-for-min-or-max-epochs">Force training for min or max epochs</h4>
|
||||
<p>It can be useful to force training for a minimum number of epochs or limit to a max number</p>
|
||||
<pre><code class="python"># DEFAULT
|
||||
trainer = Trainer(min_nb_epochs=1, max_nb_epochs=1000)
|
||||
</code></pre>
|
||||
|
||||
<hr />
|
||||
<h4 id="inspect-gradient-norms">Inspect gradient norms</h4>
|
||||
<p>Looking at grad norms can help you figure out where training might be going wrong.</p>
|
||||
<pre><code class="python"># DEFAULT (-1 doesn't track norms)
|
||||
trainer = Trainer(track_grad_norm=-1)
|
||||
|
||||
# track the LP norm (P=2 here)
|
||||
trainer = Trainer(track_grad_norm=2)
|
||||
</code></pre>
|
||||
|
||||
<hr />
|
||||
<h4 id="make-model-overfit-on-subset-of-data">Make model overfit on subset of data</h4>
|
||||
<p>A useful debugging trick is to make your model overfit a tiny fraction of the data.</p>
|
||||
<pre><code class="python"># DEFAULT don't overfit (ie: normal training)
|
||||
trainer = Trainer(overfit_pct=0.0)
|
||||
|
||||
# overfit on 1% of data
|
||||
trainer = Trainer(overfit_pct=0.01)
|
||||
</code></pre>
|
||||
|
||||
<hr />
|
||||
<h4 id="set-how-much-of-the-training-set-to-check">Set how much of the training set to check</h4>
|
||||
<p>If you don't want to check 100% of the validation set (for debugging or if it's huge), set this flag</p>
|
||||
<pre><code class="python"># DEFAULT
|
||||
trainer = Trainer(train_percent_check=1.0)
|
||||
|
||||
# check 10% only
|
||||
trainer = Trainer(train_percent_check=0.1)
|
||||
</code></pre>
|
||||
|
||||
</div>
|
||||
</div>
|
||||
<footer>
|
||||
|
||||
<div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
|
||||
|
||||
<a href="../Vaildation loop/" class="btn btn-neutral float-right" title="Vaildation loop">Next <span class="icon icon-circle-arrow-right"></span></a>
|
||||
|
||||
|
||||
<a href="../SLURM Managed Cluster/" class="btn btn-neutral" title="SLURM Managed Cluster"><span class="icon icon-circle-arrow-left"></span> Previous</a>
|
||||
|
||||
</div>
|
||||
|
||||
|
||||
<hr/>
|
||||
|
||||
<div role="contentinfo">
|
||||
<!-- Copyright etc -->
|
||||
|
||||
</div>
|
||||
|
||||
Built with <a href="http://www.mkdocs.org">MkDocs</a> using a <a href="https://github.com/snide/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
|
||||
</footer>
|
||||
|
||||
</div>
|
||||
</div>
|
||||
|
||||
</section>
|
||||
|
||||
</div>
|
||||
|
||||
<div class="rst-versions" role="note" style="cursor: pointer">
|
||||
<span class="rst-current-version" data-toggle="rst-current-version">
|
||||
|
||||
<a href="https://github.com/williamFalcon/pytorch-lightning/" class="fa fa-github" style="float: left; color: #fcfcfc"> GitHub</a>
|
||||
|
||||
|
||||
<span><a href="../SLURM Managed Cluster/" style="color: #fcfcfc;">« Previous</a></span>
|
||||
|
||||
|
||||
<span style="margin-left: 15px"><a href="../Vaildation loop/" style="color: #fcfcfc">Next »</a></span>
|
||||
|
||||
</span>
|
||||
</div>
|
||||
<script>var base_url = '../..';</script>
|
||||
<script src="../../js/theme.js" defer></script>
|
||||
<script src="../../search/main.js" defer></script>
|
||||
|
||||
</body>
|
||||
</html>
|
||||
Reference in New Issue
Block a user