mirror of
https://github.com/AntoineHX/smart_augmentation.git
synced 2025-05-03 11:40:46 +02:00
299 lines
12 KiB
HTML
299 lines
12 KiB
HTML
<!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 1.8.13"/>
|
|
<meta name="viewport" content="width=device-width, initial-scale=1"/>
|
|
<title>My Project: train_utils Namespace Reference</title>
|
|
<link href="tabs.css" rel="stylesheet" type="text/css"/>
|
|
<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>
|
|
<link href="doxygen.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="projectalign" style="padding-left: 0.5em;">
|
|
<div id="projectname">My Project
|
|
</div>
|
|
</td>
|
|
</tr>
|
|
</tbody>
|
|
</table>
|
|
</div>
|
|
<!-- end header part -->
|
|
<!-- Generated by Doxygen 1.8.13 -->
|
|
<script type="text/javascript">
|
|
var searchBox = new SearchBox("searchBox", "search",false,'Search');
|
|
</script>
|
|
<script type="text/javascript" src="menudata.js"></script>
|
|
<script type="text/javascript" src="menu.js"></script>
|
|
<script type="text/javascript">
|
|
$(function() {
|
|
initMenu('',true,false,'search.php','Search');
|
|
$(document).ready(function() { init_search(); });
|
|
});
|
|
</script>
|
|
<div id="main-nav"></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><!-- top -->
|
|
<div class="header">
|
|
<div class="summary">
|
|
<a href="#func-members">Functions</a> </div>
|
|
<div class="headertitle">
|
|
<div class="title">train_utils Namespace Reference</div> </div>
|
|
</div><!--header-->
|
|
<div class="contents">
|
|
<table class="memberdecls">
|
|
<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="func-members"></a>
|
|
Functions</h2></td></tr>
|
|
<tr class="memitem:a3021b0f6d08103a5d6b30ec48bd257f4"><td class="memItemLeft" align="right" valign="top">def </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetrain__utils.html#a3021b0f6d08103a5d6b30ec48bd257f4">test</a> (model)</td></tr>
|
|
<tr class="separator:a3021b0f6d08103a5d6b30ec48bd257f4"><td class="memSeparator" colspan="2"> </td></tr>
|
|
<tr class="memitem:a2743f44f4251820afaf234e17a1b8dd6"><td class="memItemLeft" align="right" valign="top">def </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetrain__utils.html#a2743f44f4251820afaf234e17a1b8dd6">compute_vaLoss</a> (model, dl_it, dl)</td></tr>
|
|
<tr class="separator:a2743f44f4251820afaf234e17a1b8dd6"><td class="memSeparator" colspan="2"> </td></tr>
|
|
<tr class="memitem:a62ad8259931264e8b17ececfc96abadf"><td class="memItemLeft" align="right" valign="top">def </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetrain__utils.html#a62ad8259931264e8b17ececfc96abadf">train_classic</a> (model, opt_param, epochs=1, print_freq=1)</td></tr>
|
|
<tr class="separator:a62ad8259931264e8b17ececfc96abadf"><td class="memSeparator" colspan="2"> </td></tr>
|
|
<tr class="memitem:a10789b14974c3f8232a458edd3af821b"><td class="memItemLeft" align="right" valign="top">def </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetrain__utils.html#a10789b14974c3f8232a458edd3af821b">run_dist_dataugV3</a> (model, opt_param, epochs=1, inner_it=1, dataug_epoch_start=0, print_freq=1, KLdiv=1, hp_opt=False, save_sample_freq=None)</td></tr>
|
|
<tr class="separator:a10789b14974c3f8232a458edd3af821b"><td class="memSeparator" colspan="2"> </td></tr>
|
|
</table>
|
|
<a name="details" id="details"></a><h2 class="groupheader">Detailed Description</h2>
|
|
<div class="textblock"><pre class="fragment">Utilities function for training.</pre> </div><h2 class="groupheader">Function Documentation</h2>
|
|
<a id="a2743f44f4251820afaf234e17a1b8dd6"></a>
|
|
<h2 class="memtitle"><span class="permalink"><a href="#a2743f44f4251820afaf234e17a1b8dd6">◆ </a></span>compute_vaLoss()</h2>
|
|
|
|
<div class="memitem">
|
|
<div class="memproto">
|
|
<table class="memname">
|
|
<tr>
|
|
<td class="memname">def train_utils.compute_vaLoss </td>
|
|
<td>(</td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>model</em>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>dl_it</em>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>dl</em> </td>
|
|
</tr>
|
|
<tr>
|
|
<td></td>
|
|
<td>)</td>
|
|
<td></td><td></td>
|
|
</tr>
|
|
</table>
|
|
</div><div class="memdoc">
|
|
<pre class="fragment">Evaluate a model on a batch of data.
|
|
|
|
Args:
|
|
model (nn.Module): Model to evaluate.
|
|
dl_it (Iterator): Data loader iterator.
|
|
dl (DataLoader): Data loader.
|
|
|
|
Returns:
|
|
(Tensor) Loss on a single batch of data.
|
|
</pre>
|
|
</div>
|
|
</div>
|
|
<a id="a10789b14974c3f8232a458edd3af821b"></a>
|
|
<h2 class="memtitle"><span class="permalink"><a href="#a10789b14974c3f8232a458edd3af821b">◆ </a></span>run_dist_dataugV3()</h2>
|
|
|
|
<div class="memitem">
|
|
<div class="memproto">
|
|
<table class="memname">
|
|
<tr>
|
|
<td class="memname">def train_utils.run_dist_dataugV3 </td>
|
|
<td>(</td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>model</em>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>opt_param</em>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>epochs</em> = <code>1</code>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>inner_it</em> = <code>1</code>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>dataug_epoch_start</em> = <code>0</code>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>print_freq</em> = <code>1</code>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>KLdiv</em> = <code>1</code>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>hp_opt</em> = <code>False</code>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>save_sample_freq</em> = <code>None</code> </td>
|
|
</tr>
|
|
<tr>
|
|
<td></td>
|
|
<td>)</td>
|
|
<td></td><td></td>
|
|
</tr>
|
|
</table>
|
|
</div><div class="memdoc">
|
|
<pre class="fragment">Training of an augmented model with higher.
|
|
|
|
This function is intended to be used with Augmented_model containing an Higher_model (see dataug.py).
|
|
Ex : Augmented_model(Data_augV5(...), Higher_model(model))
|
|
|
|
Training loss can either be computed directly from augmented inputs (KLdiv=0).
|
|
However, it is recommended to use the KLdiv loss computation, inspired from UDA, which combine original and augmented inputs to compute the loss (KLdiv>0).
|
|
See : https://github.com/google-research/uda
|
|
|
|
Args:
|
|
model (nn.Module): Augmented model to train.
|
|
opt_param (dict): Dictionnary containing optimizers parameters.
|
|
epochs (int): Number of epochs to perform. (default: 1)
|
|
inner_it (int): Number of inner iteration before a meta-step. 0 inner iteration means there's no meta-step. (default: 1)
|
|
dataug_epoch_start (int): Epoch when to start data augmentation. (default: 0)
|
|
print_freq (int): Number of epoch between display of the state of training. If set to None, no display will be done. (default:1)
|
|
KLdiv (float): Proportion of the KLdiv loss added to the supervised loss. If set to 0, the loss is classicly computed on augmented inputs. (default: 1)
|
|
hp_opt (bool): Wether to learn inner optimizer parameters. (default: False)
|
|
save_sample_freq (int): Number of epochs between saves of samples of data. If set to None, only one save would be done at the end of the training. (default: None)
|
|
|
|
Returns:
|
|
(list) Logs of training. Each items is a dict containing results of an epoch.
|
|
</pre>
|
|
</div>
|
|
</div>
|
|
<a id="a3021b0f6d08103a5d6b30ec48bd257f4"></a>
|
|
<h2 class="memtitle"><span class="permalink"><a href="#a3021b0f6d08103a5d6b30ec48bd257f4">◆ </a></span>test()</h2>
|
|
|
|
<div class="memitem">
|
|
<div class="memproto">
|
|
<table class="memname">
|
|
<tr>
|
|
<td class="memname">def train_utils.test </td>
|
|
<td>(</td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>model</em></td><td>)</td>
|
|
<td></td>
|
|
</tr>
|
|
</table>
|
|
</div><div class="memdoc">
|
|
<pre class="fragment">Evaluate a model on test data.
|
|
|
|
Args:
|
|
model (nn.Module): Model to test.
|
|
|
|
Returns:
|
|
(float, Tensor) Returns the accuracy and test loss of the model.
|
|
</pre>
|
|
</div>
|
|
</div>
|
|
<a id="a62ad8259931264e8b17ececfc96abadf"></a>
|
|
<h2 class="memtitle"><span class="permalink"><a href="#a62ad8259931264e8b17ececfc96abadf">◆ </a></span>train_classic()</h2>
|
|
|
|
<div class="memitem">
|
|
<div class="memproto">
|
|
<table class="memname">
|
|
<tr>
|
|
<td class="memname">def train_utils.train_classic </td>
|
|
<td>(</td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>model</em>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>opt_param</em>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>epochs</em> = <code>1</code>, </td>
|
|
</tr>
|
|
<tr>
|
|
<td class="paramkey"></td>
|
|
<td></td>
|
|
<td class="paramtype"> </td>
|
|
<td class="paramname"><em>print_freq</em> = <code>1</code> </td>
|
|
</tr>
|
|
<tr>
|
|
<td></td>
|
|
<td>)</td>
|
|
<td></td><td></td>
|
|
</tr>
|
|
</table>
|
|
</div><div class="memdoc">
|
|
<pre class="fragment">Classic training of a model.
|
|
|
|
Args:
|
|
model (nn.Module): Model to train.
|
|
opt_param (dict): Dictionnary containing optimizers parameters.
|
|
epochs (int): Number of epochs to perform. (default: 1)
|
|
print_freq (int): Number of epoch between display of the state of training. If set to None, no display will be done. (default:1)
|
|
|
|
Returns:
|
|
(list) Logs of training. Each items is a dict containing results of an epoch.
|
|
</pre>
|
|
</div>
|
|
</div>
|
|
</div><!-- contents -->
|
|
<!-- start footer part -->
|
|
<hr class="footer"/><address class="footer"><small>
|
|
Generated by  <a href="http://www.doxygen.org/index.html">
|
|
<img class="footer" src="doxygen.png" alt="doxygen"/>
|
|
</a> 1.8.13
|
|
</small></address>
|
|
</body>
|
|
</html>
|