AntoineH/smart_augmentation
1
0
Fork 0
mirror of https://github.com/AntoineHX/smart_augmentation.git synced 2025-05-03 11:40:46 +02:00
Code Issues Projects Releases Packages Wiki Activity
smart_augmentation/higher/doc/html/classdataug_1_1Higher__model.html
Antoine Harlé b89dac9084 Changes since Teledyne
2024-08-20 11:53:35 +02:00

345 lines
14 KiB
HTML
Raw Blame History

<!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: dataug.Higher_model Class 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 id="nav-path" class="navpath">
<ul>
<li class="navelem"><a class="el" href="namespacedataug.html">dataug</a></li><li class="navelem"><a class="el" href="classdataug_1_1Higher__model.html">Higher_model</a></li> </ul>
</div>
</div><!-- top -->
<div class="header">
<div class="summary">
<a href="#pub-methods">Public Member Functions</a> &#124;
<a href="classdataug_1_1Higher__model-members.html">List of all members</a> </div>
<div class="headertitle">
<div class="title">dataug.Higher_model Class Reference</div> </div>
</div><!--header-->
<div class="contents">
<div class="dynheader">
Inheritance diagram for dataug.Higher_model:</div>
<div class="dyncontent">
<div class="center"><img src="classdataug_1_1Higher__model__inherit__graph.png" border="0" usemap="#dataug_8Higher__model_inherit__map" alt="Inheritance graph"/></div>
<center><span class="legend">[<a href="graph_legend.html">legend</a>]</span></center></div>
<div class="dynheader">
Collaboration diagram for dataug.Higher_model:</div>
<div class="dyncontent">
<div class="center"><img src="classdataug_1_1Higher__model__coll__graph.png" border="0" usemap="#dataug_8Higher__model_coll__map" alt="Collaboration graph"/></div>
<center><span class="legend">[<a href="graph_legend.html">legend</a>]</span></center></div>
<table class="memberdecls">
<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="pub-methods"></a>
Public Member Functions</h2></td></tr>
<tr class="memitem:addbc61b8165d39818598c4cb969c2841"><td class="memItemLeft" align="right" valign="top">def&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classdataug_1_1Higher__model.html#addbc61b8165d39818598c4cb969c2841">__init__</a> (self, model)</td></tr>
<tr class="separator:addbc61b8165d39818598c4cb969c2841"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ab6a29e0bff528d896c17925793b1cbbb"><td class="memItemLeft" align="right" valign="top">def&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classdataug_1_1Higher__model.html#ab6a29e0bff528d896c17925793b1cbbb">get_diffopt</a> (self, opt, grad_callback=None, track_higher_grads=True)</td></tr>
<tr class="separator:ab6a29e0bff528d896c17925793b1cbbb"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ad0d2caf5b673e5ea7179eb7e5603903c"><td class="memItemLeft" align="right" valign="top">def&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classdataug_1_1Higher__model.html#ad0d2caf5b673e5ea7179eb7e5603903c">forward</a> (self, x)</td></tr>
<tr class="separator:ad0d2caf5b673e5ea7179eb7e5603903c"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:a1bd8bee3d758a2a0f54eb37ddea857c8"><td class="memItemLeft" align="right" valign="top">def&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classdataug_1_1Higher__model.html#a1bd8bee3d758a2a0f54eb37ddea857c8">detach_</a> (self)</td></tr>
<tr class="separator:a1bd8bee3d758a2a0f54eb37ddea857c8"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:a58dbc71a89657f1700c7c19f5e097e53"><td class="memItemLeft" align="right" valign="top">def&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classdataug_1_1Higher__model.html#a58dbc71a89657f1700c7c19f5e097e53">state_dict</a> (self)</td></tr>
<tr class="separator:a58dbc71a89657f1700c7c19f5e097e53"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:a2586084ef45ef8e08aa310a5cdd07bfe"><td class="memItemLeft" align="right" valign="top">def&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classdataug_1_1Higher__model.html#a2586084ef45ef8e08aa310a5cdd07bfe">__getitem__</a> (self, key)</td></tr>
<tr class="separator:a2586084ef45ef8e08aa310a5cdd07bfe"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:a662c66a8fc13fddd2f5fe21610801c41"><td class="memItemLeft" align="right" valign="top">def&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="classdataug_1_1Higher__model.html#a662c66a8fc13fddd2f5fe21610801c41">__str__</a> (self)</td></tr>
<tr class="separator:a662c66a8fc13fddd2f5fe21610801c41"><td class="memSeparator" colspan="2">&#160;</td></tr>
</table>
<a name="details" id="details"></a><h2 class="groupheader">Detailed Description</h2>
<div class="textblock"><pre class="fragment">Model wrapper for higher gradient tracking.
Keep in memory the orginial model and it's functionnal, higher, version.
Might not be needed anymore if Higher implement detach for fmodel.
see : https://github.com/facebookresearch/higher
TODO: Get rid of the original model if not needed by user.
Attributes:
_name (string): Name of the model.
_mods (nn.ModuleDict): Models (Orginial and Higher version).
</pre> </div><h2 class="groupheader">Constructor &amp; Destructor Documentation</h2>
<a id="addbc61b8165d39818598c4cb969c2841"></a>
<h2 class="memtitle"><span class="permalink"><a href="#addbc61b8165d39818598c4cb969c2841">&#9670;&nbsp;</a></span>__init__()</h2>
<div class="memitem">
<div class="memproto">
<table class="memname">
<tr>
<td class="memname">def dataug.Higher_model.__init__ </td>
<td>(</td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>self</em>, </td>
</tr>
<tr>
<td class="paramkey"></td>
<td></td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>model</em>&#160;</td>
</tr>
<tr>
<td></td>
<td>)</td>
<td></td><td></td>
</tr>
</table>
</div><div class="memdoc">
<pre class="fragment">Init Higher_model.
Args:
model (nn.Module): Network for which higher gradients can be tracked.
</pre>
</div>
</div>
<h2 class="groupheader">Member Function Documentation</h2>
<a id="a2586084ef45ef8e08aa310a5cdd07bfe"></a>
<h2 class="memtitle"><span class="permalink"><a href="#a2586084ef45ef8e08aa310a5cdd07bfe">&#9670;&nbsp;</a></span>__getitem__()</h2>
<div class="memitem">
<div class="memproto">
<table class="memname">
<tr>
<td class="memname">def dataug.Higher_model.__getitem__ </td>
<td>(</td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>self</em>, </td>
</tr>
<tr>
<td class="paramkey"></td>
<td></td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>key</em>&#160;</td>
</tr>
<tr>
<td></td>
<td>)</td>
<td></td><td></td>
</tr>
</table>
</div><div class="memdoc">
<pre class="fragment">Access to modules
Args:
key (string): Name of the module to access.
Returns:
nn.Module.
</pre>
</div>
</div>
<a id="a662c66a8fc13fddd2f5fe21610801c41"></a>
<h2 class="memtitle"><span class="permalink"><a href="#a662c66a8fc13fddd2f5fe21610801c41">&#9670;&nbsp;</a></span>__str__()</h2>
<div class="memitem">
<div class="memproto">
<table class="memname">
<tr>
<td class="memname">def dataug.Higher_model.__str__ </td>
<td>(</td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>self</em></td><td>)</td>
<td></td>
</tr>
</table>
</div><div class="memdoc">
<pre class="fragment">Name of the module
Returns:
String containing the name of the module.
</pre>
</div>
</div>
<a id="a1bd8bee3d758a2a0f54eb37ddea857c8"></a>
<h2 class="memtitle"><span class="permalink"><a href="#a1bd8bee3d758a2a0f54eb37ddea857c8">&#9670;&nbsp;</a></span>detach_()</h2>
<div class="memitem">
<div class="memproto">
<table class="memname">
<tr>
<td class="memname">def dataug.Higher_model.detach_ </td>
<td>(</td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>self</em></td><td>)</td>
<td></td>
</tr>
</table>
</div><div class="memdoc">
<pre class="fragment">Detach from the graph.
Needed to limit the number of state kept in memory.
</pre>
</div>
</div>
<a id="ad0d2caf5b673e5ea7179eb7e5603903c"></a>
<h2 class="memtitle"><span class="permalink"><a href="#ad0d2caf5b673e5ea7179eb7e5603903c">&#9670;&nbsp;</a></span>forward()</h2>
<div class="memitem">
<div class="memproto">
<table class="memname">
<tr>
<td class="memname">def dataug.Higher_model.forward </td>
<td>(</td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>self</em>, </td>
</tr>
<tr>
<td class="paramkey"></td>
<td></td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>x</em>&#160;</td>
</tr>
<tr>
<td></td>
<td>)</td>
<td></td><td></td>
</tr>
</table>
</div><div class="memdoc">
<pre class="fragment">Main method of the model.
Args:
x (Tensor): Batch of data.
Returns:
Tensor : Output of the network. Should be logits.
</pre>
</div>
</div>
<a id="ab6a29e0bff528d896c17925793b1cbbb"></a>
<h2 class="memtitle"><span class="permalink"><a href="#ab6a29e0bff528d896c17925793b1cbbb">&#9670;&nbsp;</a></span>get_diffopt()</h2>
<div class="memitem">
<div class="memproto">
<table class="memname">
<tr>
<td class="memname">def dataug.Higher_model.get_diffopt </td>
<td>(</td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>self</em>, </td>
</tr>
<tr>
<td class="paramkey"></td>
<td></td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>opt</em>, </td>
</tr>
<tr>
<td class="paramkey"></td>
<td></td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>grad_callback</em> = <code>None</code>, </td>
</tr>
<tr>
<td class="paramkey"></td>
<td></td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>track_higher_grads</em> = <code>True</code>&#160;</td>
</tr>
<tr>
<td></td>
<td>)</td>
<td></td><td></td>
</tr>
</table>
</div><div class="memdoc">
<pre class="fragment">Get a differentiable version of an Optimizer.
Higher/Differentiable optimizer required to be used for higher gradient tracking.
Usage : diffopt.step(loss) == (opt.zero_grad, loss.backward, opt.step)
Be warry that if track_higher_grads is set to True, a new state of the model would be saved each time diffopt.step() is called.
Thus increasing memory consumption. The detach_() method should be called to reset the gradient tape and prevent memory saturation.
Args:
opt (torch.optim): Optimizer to make differentiable.
grad_callback (fct(grads)=grads): Function applied to the list of gradients parameters (ex: clipping). (default: None)
track_higher_grads (bool): Wether higher gradient are tracked. If True, the graph/states will be retained to allow backpropagation. (default: True)
Returns:
(Higher.DifferentiableOptimizer): Differentiable version of the optimizer.
</pre>
</div>
</div>
<a id="a58dbc71a89657f1700c7c19f5e097e53"></a>
<h2 class="memtitle"><span class="permalink"><a href="#a58dbc71a89657f1700c7c19f5e097e53">&#9670;&nbsp;</a></span>state_dict()</h2>
<div class="memitem">
<div class="memproto">
<table class="memname">
<tr>
<td class="memname">def dataug.Higher_model.state_dict </td>
<td>(</td>
<td class="paramtype">&#160;</td>
<td class="paramname"><em>self</em></td><td>)</td>
<td></td>
</tr>
</table>
</div><div class="memdoc">
<pre class="fragment">Returns a dictionary containing a whole state of the module.
</pre>
</div>
</div>
<hr/>The documentation for this class was generated from the following file:<ul>
<li>/home/antoineh/github/smart_augmentation/higher/smart_aug/dataug.py</li>
</ul>
</div><!-- contents -->
<!-- start footer part -->
<hr class="footer"/><address class="footer"><small>
Generated by &#160;<a href="http://www.doxygen.org/index.html">
<img class="footer" src="doxygen.png" alt="doxygen"/>
</a> 1.8.13
</small></address>
</body>
</html>
Reference in a new issue View git blame Copy permalink