| <!DOCTYPE html > |
| <html> |
| <head> |
| <title>NDArrayBase - MXNet - org.apache.mxnet.NDArrayBase</title> |
| <meta name="description" content="NDArrayBase - MXNet - org.apache.mxnet.NDArrayBase" /> |
| <meta name="keywords" content="NDArrayBase MXNet org.apache.mxnet.NDArrayBase" /> |
| <meta http-equiv="content-type" content="text/html; charset=UTF-8" /> |
| |
| <link href="../../../lib/template.css" media="screen" type="text/css" rel="stylesheet" /> |
| <link href="../../../lib/diagrams.css" media="screen" type="text/css" rel="stylesheet" id="diagrams-css" /> |
| <script type="text/javascript" src="../../../lib/jquery.js" id="jquery-js"></script> |
| <script type="text/javascript" src="../../../lib/jquery-ui.js"></script> |
| <script type="text/javascript" src="../../../lib/template.js"></script> |
| <script type="text/javascript" src="../../../lib/tools.tooltip.js"></script> |
| |
| <script type="text/javascript"> |
| if(top === self) { |
| var url = '../../../index.html'; |
| var hash = 'org.apache.mxnet.NDArrayBase'; |
| var anchor = window.location.hash; |
| var anchor_opt = ''; |
| if (anchor.length >= 1) |
| anchor_opt = '@' + anchor.substring(1); |
| window.location.href = url + '#' + hash + anchor_opt; |
| } |
| </script> |
| |
| </head> |
| <body class="type"> |
| <div id="definition"> |
| <img alt="Class" src="../../../lib/class_big.png" /> |
| <p id="owner"><a href="../../package.html" class="extype" name="org">org</a>.<a href="../package.html" class="extype" name="org.apache">apache</a>.<a href="package.html" class="extype" name="org.apache.mxnet">mxnet</a></p> |
| <h1>NDArrayBase</h1><h3><span class="morelinks"><div>Related Doc: |
| <a href="package.html" class="extype" name="org.apache.mxnet">package mxnet</a> |
| </div></span></h3><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| </div> |
| |
| <h4 id="signature" class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">class</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">NDArrayBase</span><span class="result"> extends <span class="extype" name="scala.AnyRef">AnyRef</span></span> |
| </span> |
| </h4> |
| |
| <div id="comment" class="fullcommenttop"><div class="toggleContainer block"> |
| <span class="toggle">Linear Supertypes</span> |
| <div class="superTypes hiddenContent"><span class="extype" name="scala.AnyRef">AnyRef</span>, <span class="extype" name="scala.Any">Any</span></div> |
| </div><div class="toggleContainer block"> |
| <span class="toggle">Known Subclasses</span> |
| <div class="subClasses hiddenContent"><a href="NDArray$.html" class="extype" name="org.apache.mxnet.NDArray">NDArray</a></div> |
| </div></div> |
| |
| |
| <div id="mbrsel"> |
| <div id="textfilter"><span class="pre"></span><span class="input"><input id="mbrsel-input" type="text" accesskey="/" /></span><span class="post"></span></div> |
| <div id="order"> |
| <span class="filtertype">Ordering</span> |
| <ol> |
| |
| <li class="alpha in"><span>Alphabetic</span></li> |
| <li class="inherit out"><span>By Inheritance</span></li> |
| </ol> |
| </div> |
| <div id="ancestors"> |
| <span class="filtertype">Inherited<br /> |
| </span> |
| <ol id="linearization"> |
| <li class="in" name="org.apache.mxnet.NDArrayBase"><span>NDArrayBase</span></li><li class="in" name="scala.AnyRef"><span>AnyRef</span></li><li class="in" name="scala.Any"><span>Any</span></li> |
| </ol> |
| </div><div id="ancestors"> |
| <span class="filtertype"></span> |
| <ol> |
| <li class="hideall out"><span>Hide All</span></li> |
| <li class="showall in"><span>Show All</span></li> |
| </ol> |
| </div> |
| <div id="visbl"> |
| <span class="filtertype">Visibility</span> |
| <ol><li class="public in"><span>Public</span></li><li class="all out"><span>All</span></li></ol> |
| </div> |
| </div> |
| |
| <div id="template"> |
| <div id="allMembers"> |
| <div id="constructors" class="members"> |
| <h3>Instance Constructors</h3> |
| <ol><li name="org.apache.mxnet.NDArrayBase#<init>" visbl="pub" data-isabs="false" fullComment="no" group="Ungrouped"> |
| <a id="<init>():org.apache.mxnet.NDArrayBase"></a> |
| <a id="<init>:NDArrayBase"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier"></span> |
| <span class="kind">new</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">NDArrayBase</span><span class="params">()</span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@<init>():org.apache.mxnet.NDArrayBase" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| |
| </li></ol> |
| </div> |
| |
| |
| |
| <div id="values" class="values members"> |
| <h3>Abstract Value Members</h3> |
| <ol><li name="org.apache.mxnet.NDArrayBase#Activation" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Activation(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Activation(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Activation</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Activation(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies an activation function element-wise to the input. |
| |
| The following activation functions are supported: |
| |
| - `relu`: Rectified Linear <span class="std">Unit</span>, :math:`y = max(x, <span class="num">0</span>)` |
| - `sigmoid`: :math:`y = \frac{<span class="num">1</span>}{<span class="num">1</span> + exp(-x)}` |
| - `tanh`: Hyperbolic tangent, :math:`y = \frac{exp(x) - exp(-x)}{exp(x) + exp(-x)}` |
| - `softrelu`: Soft ReLU, or SoftPlus, :math:`y = log(<span class="num">1</span> + exp(x))` |
| - `softsign`: :math:`y = \frac{x}{<span class="num">1</span> + abs(x)}` |
| |
| |
| |
| Defined in src/operator/nn/activation.cc:L164</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Activation" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Activation(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Activation(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Activation</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Activation(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies an activation function element-wise to the input. |
| |
| The following activation functions are supported: |
| |
| - `relu`: Rectified Linear <span class="std">Unit</span>, :math:`y = max(x, <span class="num">0</span>)` |
| - `sigmoid`: :math:`y = \frac{<span class="num">1</span>}{<span class="num">1</span> + exp(-x)}` |
| - `tanh`: Hyperbolic tangent, :math:`y = \frac{exp(x) - exp(-x)}{exp(x) + exp(-x)}` |
| - `softrelu`: Soft ReLU, or SoftPlus, :math:`y = log(<span class="num">1</span> + exp(x))` |
| - `softsign`: :math:`y = \frac{x}{<span class="num">1</span> + abs(x)}` |
| |
| |
| |
| Defined in src/operator/nn/activation.cc:L164</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#BatchNorm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="BatchNorm(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="BatchNorm(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">BatchNorm</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@BatchNorm(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Batch normalization. |
| |
| Normalizes a data batch by mean and variance, and applies a scale ``gamma`` as |
| well as offset ``beta``. |
| |
| Assume the input has more than one dimension and we normalize along axis <span class="num">1.</span> |
| We first compute the mean and variance along <span class="kw">this</span> axis: |
| |
| .. math:: |
| |
| data\_mean[i] = mean(data[:,i,:,...]) \\ |
| data\_var[i] = <span class="kw">var</span>(data[:,i,:,...]) |
| |
| Then compute the normalized output, which has the same shape as input, as following: |
| |
| .. math:: |
| |
| out[:,i,:,...] = \frac{data[:,i,:,...] - data\_mean[i]}{\sqrt{data\_var[i]+\epsilon}} * gamma[i] + beta[i] |
| |
| Both *mean* and *<span class="kw">var</span>* returns a scalar by treating the input as a vector. |
| |
| Assume the input has size *k* on axis <span class="num">1</span>, then both ``gamma`` and ``beta`` |
| have shape *(k,)*. If ``output_mean_var`` is set to be <span class="kw">true</span>, then outputs both ``data_mean`` and |
| the inverse of ``data_var``, which are needed <span class="kw">for</span> the backward pass. Note that gradient of these |
| two outputs are blocked. |
| |
| Besides the inputs and the outputs, <span class="kw">this</span> operator accepts two auxiliary |
| states, ``moving_mean`` and ``moving_var``, which are *k*-length |
| vectors. They are global statistics <span class="kw">for</span> the whole dataset, which are updated |
| by:: |
| |
| moving_mean = moving_mean * momentum + data_mean * (<span class="num">1</span> - momentum) |
| moving_var = moving_var * momentum + data_var * (<span class="num">1</span> - momentum) |
| |
| If ``use_global_stats`` is set to be <span class="kw">true</span>, then ``moving_mean`` and |
| ``moving_var`` are used instead of ``data_mean`` and ``data_var`` to compute |
| the output. It is often used during inference. |
| |
| The parameter ``axis`` specifies which axis of the input shape denotes |
| the 'channel' (separately normalized groups). The default is <span class="num">1.</span> Specifying -<span class="num">1</span> sets the channel |
| axis to be the last item in the input shape. |
| |
| Both ``gamma`` and ``beta`` are learnable parameters. But <span class="kw">if</span> ``fix_gamma`` is <span class="kw">true</span>, |
| then set ``gamma`` to <span class="num">1</span> and its gradient to <span class="num">0.</span> |
| |
| .. Note:: |
| When ``fix_gamma`` is set to True, no sparse support is provided. If ``fix_gamma is`` set to False, |
| the sparse tensors will fallback. |
| |
| |
| |
| Defined in src/operator/nn/batch_norm.cc:L608</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#BatchNorm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="BatchNorm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="BatchNorm(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">BatchNorm</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@BatchNorm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Batch normalization. |
| |
| Normalizes a data batch by mean and variance, and applies a scale ``gamma`` as |
| well as offset ``beta``. |
| |
| Assume the input has more than one dimension and we normalize along axis <span class="num">1.</span> |
| We first compute the mean and variance along <span class="kw">this</span> axis: |
| |
| .. math:: |
| |
| data\_mean[i] = mean(data[:,i,:,...]) \\ |
| data\_var[i] = <span class="kw">var</span>(data[:,i,:,...]) |
| |
| Then compute the normalized output, which has the same shape as input, as following: |
| |
| .. math:: |
| |
| out[:,i,:,...] = \frac{data[:,i,:,...] - data\_mean[i]}{\sqrt{data\_var[i]+\epsilon}} * gamma[i] + beta[i] |
| |
| Both *mean* and *<span class="kw">var</span>* returns a scalar by treating the input as a vector. |
| |
| Assume the input has size *k* on axis <span class="num">1</span>, then both ``gamma`` and ``beta`` |
| have shape *(k,)*. If ``output_mean_var`` is set to be <span class="kw">true</span>, then outputs both ``data_mean`` and |
| the inverse of ``data_var``, which are needed <span class="kw">for</span> the backward pass. Note that gradient of these |
| two outputs are blocked. |
| |
| Besides the inputs and the outputs, <span class="kw">this</span> operator accepts two auxiliary |
| states, ``moving_mean`` and ``moving_var``, which are *k*-length |
| vectors. They are global statistics <span class="kw">for</span> the whole dataset, which are updated |
| by:: |
| |
| moving_mean = moving_mean * momentum + data_mean * (<span class="num">1</span> - momentum) |
| moving_var = moving_var * momentum + data_var * (<span class="num">1</span> - momentum) |
| |
| If ``use_global_stats`` is set to be <span class="kw">true</span>, then ``moving_mean`` and |
| ``moving_var`` are used instead of ``data_mean`` and ``data_var`` to compute |
| the output. It is often used during inference. |
| |
| The parameter ``axis`` specifies which axis of the input shape denotes |
| the 'channel' (separately normalized groups). The default is <span class="num">1.</span> Specifying -<span class="num">1</span> sets the channel |
| axis to be the last item in the input shape. |
| |
| Both ``gamma`` and ``beta`` are learnable parameters. But <span class="kw">if</span> ``fix_gamma`` is <span class="kw">true</span>, |
| then set ``gamma`` to <span class="num">1</span> and its gradient to <span class="num">0.</span> |
| |
| .. Note:: |
| When ``fix_gamma`` is set to True, no sparse support is provided. If ``fix_gamma is`` set to False, |
| the sparse tensors will fallback. |
| |
| |
| |
| Defined in src/operator/nn/batch_norm.cc:L608</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#BatchNorm_v1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="BatchNorm_v1(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="BatchNorm_v1(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">BatchNorm_v1</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@BatchNorm_v1(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Batch normalization. |
| |
| This operator is DEPRECATED. Perform BatchNorm on the input. |
| |
| Normalizes a data batch by mean and variance, and applies a scale ``gamma`` as |
| well as offset ``beta``. |
| |
| Assume the input has more than one dimension and we normalize along axis <span class="num">1.</span> |
| We first compute the mean and variance along <span class="kw">this</span> axis: |
| |
| .. math:: |
| |
| data\_mean[i] = mean(data[:,i,:,...]) \\ |
| data\_var[i] = <span class="kw">var</span>(data[:,i,:,...]) |
| |
| Then compute the normalized output, which has the same shape as input, as following: |
| |
| .. math:: |
| |
| out[:,i,:,...] = \frac{data[:,i,:,...] - data\_mean[i]}{\sqrt{data\_var[i]+\epsilon}} * gamma[i] + beta[i] |
| |
| Both *mean* and *<span class="kw">var</span>* returns a scalar by treating the input as a vector. |
| |
| Assume the input has size *k* on axis <span class="num">1</span>, then both ``gamma`` and ``beta`` |
| have shape *(k,)*. If ``output_mean_var`` is set to be <span class="kw">true</span>, then outputs both ``data_mean`` and |
| ``data_var`` as well, which are needed <span class="kw">for</span> the backward pass. |
| |
| Besides the inputs and the outputs, <span class="kw">this</span> operator accepts two auxiliary |
| states, ``moving_mean`` and ``moving_var``, which are *k*-length |
| vectors. They are global statistics <span class="kw">for</span> the whole dataset, which are updated |
| by:: |
| |
| moving_mean = moving_mean * momentum + data_mean * (<span class="num">1</span> - momentum) |
| moving_var = moving_var * momentum + data_var * (<span class="num">1</span> - momentum) |
| |
| If ``use_global_stats`` is set to be <span class="kw">true</span>, then ``moving_mean`` and |
| ``moving_var`` are used instead of ``data_mean`` and ``data_var`` to compute |
| the output. It is often used during inference. |
| |
| Both ``gamma`` and ``beta`` are learnable parameters. But <span class="kw">if</span> ``fix_gamma`` is <span class="kw">true</span>, |
| then set ``gamma`` to <span class="num">1</span> and its gradient to <span class="num">0.</span> |
| |
| There's no sparse support <span class="kw">for</span> <span class="kw">this</span> operator, and it will exhibit problematic behavior <span class="kw">if</span> used <span class="kw">with</span> |
| sparse tensors. |
| |
| |
| |
| Defined in src/operator/batch_norm_v1.cc:L94</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#BatchNorm_v1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="BatchNorm_v1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="BatchNorm_v1(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">BatchNorm_v1</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@BatchNorm_v1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Batch normalization. |
| |
| This operator is DEPRECATED. Perform BatchNorm on the input. |
| |
| Normalizes a data batch by mean and variance, and applies a scale ``gamma`` as |
| well as offset ``beta``. |
| |
| Assume the input has more than one dimension and we normalize along axis <span class="num">1.</span> |
| We first compute the mean and variance along <span class="kw">this</span> axis: |
| |
| .. math:: |
| |
| data\_mean[i] = mean(data[:,i,:,...]) \\ |
| data\_var[i] = <span class="kw">var</span>(data[:,i,:,...]) |
| |
| Then compute the normalized output, which has the same shape as input, as following: |
| |
| .. math:: |
| |
| out[:,i,:,...] = \frac{data[:,i,:,...] - data\_mean[i]}{\sqrt{data\_var[i]+\epsilon}} * gamma[i] + beta[i] |
| |
| Both *mean* and *<span class="kw">var</span>* returns a scalar by treating the input as a vector. |
| |
| Assume the input has size *k* on axis <span class="num">1</span>, then both ``gamma`` and ``beta`` |
| have shape *(k,)*. If ``output_mean_var`` is set to be <span class="kw">true</span>, then outputs both ``data_mean`` and |
| ``data_var`` as well, which are needed <span class="kw">for</span> the backward pass. |
| |
| Besides the inputs and the outputs, <span class="kw">this</span> operator accepts two auxiliary |
| states, ``moving_mean`` and ``moving_var``, which are *k*-length |
| vectors. They are global statistics <span class="kw">for</span> the whole dataset, which are updated |
| by:: |
| |
| moving_mean = moving_mean * momentum + data_mean * (<span class="num">1</span> - momentum) |
| moving_var = moving_var * momentum + data_var * (<span class="num">1</span> - momentum) |
| |
| If ``use_global_stats`` is set to be <span class="kw">true</span>, then ``moving_mean`` and |
| ``moving_var`` are used instead of ``data_mean`` and ``data_var`` to compute |
| the output. It is often used during inference. |
| |
| Both ``gamma`` and ``beta`` are learnable parameters. But <span class="kw">if</span> ``fix_gamma`` is <span class="kw">true</span>, |
| then set ``gamma`` to <span class="num">1</span> and its gradient to <span class="num">0.</span> |
| |
| There's no sparse support <span class="kw">for</span> <span class="kw">this</span> operator, and it will exhibit problematic behavior <span class="kw">if</span> used <span class="kw">with</span> |
| sparse tensors. |
| |
| |
| |
| Defined in src/operator/batch_norm_v1.cc:L94</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#BilinearSampler" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="BilinearSampler(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="BilinearSampler(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">BilinearSampler</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@BilinearSampler(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies bilinear sampling to input feature map. |
| |
| Bilinear Sampling is the key of [NIPS2015] \<span class="lit">"Spatial Transformer Networks\". The usage of the operator is very similar to remap function in OpenCV, |
| except that the operator has the backward pass. |
| |
| Given :math:`data` and :math:`grid`, then the output is computed by |
| |
| .. math:: |
| x_{src} = grid[batch, 0, y_{dst}, x_{dst}] \\ |
| y_{src} = grid[batch, 1, y_{dst}, x_{dst}] \\ |
| output[batch, channel, y_{dst}, x_{dst}] = G(data[batch, channel, y_{src}, x_{src}) |
| |
| :math:`x_{dst}`, :math:`y_{dst}` enumerate all spatial locations in :math:`output`, and :math:`G()` denotes the bilinear interpolation kernel. |
| The out-boundary points will be padded with zeros.The shape of the output will be (data.shape[0], data.shape[1], grid.shape[2], grid.shape[3]). |
| |
| The operator assumes that :math:`data` has 'NCHW' layout and :math:`grid` has been normalized to [-1, 1]. |
| |
| BilinearSampler often cooperates with GridGenerator which generates sampling grids for BilinearSampler. |
| GridGenerator supports two kinds of transformation: ``affine`` and ``warp``. |
| If users want to design a CustomOp to manipulate :math:`grid`, please firstly refer to the code of GridGenerator. |
| |
| Example 1:: |
| |
| ## Zoom out data two times |
| data = array(`[ [`[ [1, 4, 3, 6], |
| [1, 8, 8, 9], |
| [0, 4, 1, 5], |
| [1, 0, 1, 3] ] ] ]) |
| |
| affine_matrix = array(`[ [2, 0, 0], |
| [0, 2, 0] ]) |
| |
| affine_matrix = reshape(affine_matrix, shape=(1, 6)) |
| |
| grid = GridGenerator(data=affine_matrix, transform_type='affine', target_shape=(4, 4)) |
| |
| out = BilinearSampler(data, grid) |
| |
| out |
| `[ [`[ [ 0, 0, 0, 0], |
| [ 0, 3.5, 6.5, 0], |
| [ 0, 1.25, 2.5, 0], |
| [ 0, 0, 0, 0] ] ] |
| |
| |
| Example 2:: |
| |
| ## shift data horizontally by -1 pixel |
| |
| data = array(`[ [`[ [1, 4, 3, 6], |
| [1, 8, 8, 9], |
| [0, 4, 1, 5], |
| [1, 0, 1, 3] ] ] ]) |
| |
| warp_maxtrix = array(`[ [`[ [1, 1, 1, 1], |
| [1, 1, 1, 1], |
| [1, 1, 1, 1], |
| [1, 1, 1, 1] ], |
| `[ [0, 0, 0, 0], |
| [0, 0, 0, 0], |
| [0, 0, 0, 0], |
| [0, 0, 0, 0] ] ] ]) |
| |
| grid = GridGenerator(data=warp_matrix, transform_type='warp') |
| out = BilinearSampler(data, grid) |
| |
| out |
| `[ [`[ [ 4, 3, 6, 0], |
| [ 8, 8, 9, 0], |
| [ 4, 1, 5, 0], |
| [ 0, 1, 3, 0] ] ] |
| |
| |
| Defined in src/operator/bilinear_sampler.cc:L255</span></pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#BilinearSampler" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="BilinearSampler(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="BilinearSampler(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">BilinearSampler</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@BilinearSampler(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies bilinear sampling to input feature map. |
| |
| Bilinear Sampling is the key of [NIPS2015] \<span class="lit">"Spatial Transformer Networks\". The usage of the operator is very similar to remap function in OpenCV, |
| except that the operator has the backward pass. |
| |
| Given :math:`data` and :math:`grid`, then the output is computed by |
| |
| .. math:: |
| x_{src} = grid[batch, 0, y_{dst}, x_{dst}] \\ |
| y_{src} = grid[batch, 1, y_{dst}, x_{dst}] \\ |
| output[batch, channel, y_{dst}, x_{dst}] = G(data[batch, channel, y_{src}, x_{src}) |
| |
| :math:`x_{dst}`, :math:`y_{dst}` enumerate all spatial locations in :math:`output`, and :math:`G()` denotes the bilinear interpolation kernel. |
| The out-boundary points will be padded with zeros.The shape of the output will be (data.shape[0], data.shape[1], grid.shape[2], grid.shape[3]). |
| |
| The operator assumes that :math:`data` has 'NCHW' layout and :math:`grid` has been normalized to [-1, 1]. |
| |
| BilinearSampler often cooperates with GridGenerator which generates sampling grids for BilinearSampler. |
| GridGenerator supports two kinds of transformation: ``affine`` and ``warp``. |
| If users want to design a CustomOp to manipulate :math:`grid`, please firstly refer to the code of GridGenerator. |
| |
| Example 1:: |
| |
| ## Zoom out data two times |
| data = array(`[ [`[ [1, 4, 3, 6], |
| [1, 8, 8, 9], |
| [0, 4, 1, 5], |
| [1, 0, 1, 3] ] ] ]) |
| |
| affine_matrix = array(`[ [2, 0, 0], |
| [0, 2, 0] ]) |
| |
| affine_matrix = reshape(affine_matrix, shape=(1, 6)) |
| |
| grid = GridGenerator(data=affine_matrix, transform_type='affine', target_shape=(4, 4)) |
| |
| out = BilinearSampler(data, grid) |
| |
| out |
| `[ [`[ [ 0, 0, 0, 0], |
| [ 0, 3.5, 6.5, 0], |
| [ 0, 1.25, 2.5, 0], |
| [ 0, 0, 0, 0] ] ] |
| |
| |
| Example 2:: |
| |
| ## shift data horizontally by -1 pixel |
| |
| data = array(`[ [`[ [1, 4, 3, 6], |
| [1, 8, 8, 9], |
| [0, 4, 1, 5], |
| [1, 0, 1, 3] ] ] ]) |
| |
| warp_maxtrix = array(`[ [`[ [1, 1, 1, 1], |
| [1, 1, 1, 1], |
| [1, 1, 1, 1], |
| [1, 1, 1, 1] ], |
| `[ [0, 0, 0, 0], |
| [0, 0, 0, 0], |
| [0, 0, 0, 0], |
| [0, 0, 0, 0] ] ] ]) |
| |
| grid = GridGenerator(data=warp_matrix, transform_type='warp') |
| out = BilinearSampler(data, grid) |
| |
| out |
| `[ [`[ [ 4, 3, 6, 0], |
| [ 8, 8, 9, 0], |
| [ 4, 1, 5, 0], |
| [ 0, 1, 3, 0] ] ] |
| |
| |
| Defined in src/operator/bilinear_sampler.cc:L255</span></pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#BlockGrad" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="BlockGrad(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="BlockGrad(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">BlockGrad</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@BlockGrad(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Stops gradient computation. |
| |
| Stops the accumulated gradient of the inputs from flowing through <span class="kw">this</span> operator |
| in the backward direction. In other words, <span class="kw">this</span> operator prevents the contribution |
| of its inputs to be taken into account <span class="kw">for</span> computing gradients. |
| |
| Example:: |
| |
| v1 = [<span class="num">1</span>, <span class="num">2</span>] |
| v2 = [<span class="num">0</span>, <span class="num">1</span>] |
| a = Variable(<span class="lit">'a'</span>) |
| b = Variable(<span class="lit">'b'</span>) |
| b_stop_grad = stop_gradient(<span class="num">3</span> * b) |
| loss = MakeLoss(b_stop_grad + a) |
| |
| executor = loss.simple_bind(ctx=cpu(), a=(<span class="num">1</span>,<span class="num">2</span>), b=(<span class="num">1</span>,<span class="num">2</span>)) |
| executor.forward(is_train=True, a=v1, b=v2) |
| executor.outputs |
| [ <span class="num">1.</span> <span class="num">5.</span>] |
| |
| executor.backward() |
| executor.grad_arrays |
| [ <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L325</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#BlockGrad" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="BlockGrad(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="BlockGrad(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">BlockGrad</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@BlockGrad(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Stops gradient computation. |
| |
| Stops the accumulated gradient of the inputs from flowing through <span class="kw">this</span> operator |
| in the backward direction. In other words, <span class="kw">this</span> operator prevents the contribution |
| of its inputs to be taken into account <span class="kw">for</span> computing gradients. |
| |
| Example:: |
| |
| v1 = [<span class="num">1</span>, <span class="num">2</span>] |
| v2 = [<span class="num">0</span>, <span class="num">1</span>] |
| a = Variable(<span class="lit">'a'</span>) |
| b = Variable(<span class="lit">'b'</span>) |
| b_stop_grad = stop_gradient(<span class="num">3</span> * b) |
| loss = MakeLoss(b_stop_grad + a) |
| |
| executor = loss.simple_bind(ctx=cpu(), a=(<span class="num">1</span>,<span class="num">2</span>), b=(<span class="num">1</span>,<span class="num">2</span>)) |
| executor.forward(is_train=True, a=v1, b=v2) |
| executor.outputs |
| [ <span class="num">1.</span> <span class="num">5.</span>] |
| |
| executor.backward() |
| executor.grad_arrays |
| [ <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L325</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#CTCLoss" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="CTCLoss(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="CTCLoss(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">CTCLoss</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@CTCLoss(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Connectionist Temporal Classification Loss. |
| |
| .. note:: The existing alias ``contrib_CTCLoss`` is deprecated. |
| |
| The shapes of the inputs and outputs: |
| |
| - **data**: `(sequence_length, batch_size, alphabet_size)` |
| - **label**: `(batch_size, label_sequence_length)` |
| - **out**: `(batch_size)` |
| |
| The `data` tensor consists of sequences of activation vectors (without applying softmax), |
| <span class="kw">with</span> i-th channel in the last dimension corresponding to i-th label |
| <span class="kw">for</span> i between <span class="num">0</span> and alphabet_size-<span class="num">1</span> (i.e always <span class="num">0</span>-indexed). |
| Alphabet size should include one additional value reserved <span class="kw">for</span> blank label. |
| When `blank_label` is ``<span class="lit">"first"</span>``, the ``<span class="num">0</span>``-th channel is be reserved <span class="kw">for</span> |
| activation of blank label, or otherwise <span class="kw">if</span> it is <span class="lit">"last"</span>, ``(alphabet_size-<span class="num">1</span>)``-th channel should be |
| reserved <span class="kw">for</span> blank label. |
| |
| ``label`` is an index matrix of integers. When `blank_label` is ``<span class="lit">"first"</span>``, |
| the value <span class="num">0</span> is then reserved <span class="kw">for</span> blank label, and should not be passed in <span class="kw">this</span> matrix. Otherwise, |
| when `blank_label` is ``<span class="lit">"last"</span>``, the value `(alphabet_size-<span class="num">1</span>)` is reserved <span class="kw">for</span> blank label. |
| |
| If a sequence of labels is shorter than *label_sequence_length*, use the special |
| padding value at the end of the sequence to conform it to the correct |
| length. The padding value is `<span class="num">0</span>` when `blank_label` is ``<span class="lit">"first"</span>``, and `-<span class="num">1</span>` otherwise. |
| |
| For example, suppose the vocabulary is `[a, b, c]`, and in one batch we have three sequences |
| <span class="lit">'ba'</span>, <span class="lit">'cbb'</span>, and <span class="lit">'abac'</span>. When `blank_label` is ``<span class="lit">"first"</span>``, we can index the labels as |
| `{<span class="lit">'a'</span>: <span class="num">1</span>, <span class="lit">'b'</span>: <span class="num">2</span>, <span class="lit">'c'</span>: <span class="num">3</span>}`, and we reserve the <span class="num">0</span>-th channel <span class="kw">for</span> blank label in data tensor. |
| The resulting `label` tensor should be padded to be:: |
| |
| `[ [<span class="num">2</span>, <span class="num">1</span>, <span class="num">0</span>, <span class="num">0</span>], [<span class="num">3</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">0</span>], [<span class="num">1</span>, <span class="num">2</span>, <span class="num">1</span>, <span class="num">3</span>] ] |
| |
| When `blank_label` is ``<span class="lit">"last"</span>``, we can index the labels as |
| `{<span class="lit">'a'</span>: <span class="num">0</span>, <span class="lit">'b'</span>: <span class="num">1</span>, <span class="lit">'c'</span>: <span class="num">2</span>}`, and we reserve the channel index <span class="num">3</span> <span class="kw">for</span> blank label in data tensor. |
| The resulting `label` tensor should be padded to be:: |
| |
| `[ [<span class="num">1</span>, <span class="num">0</span>, -<span class="num">1</span>, -<span class="num">1</span>], [<span class="num">2</span>, <span class="num">1</span>, <span class="num">1</span>, -<span class="num">1</span>], [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>, <span class="num">2</span>] ] |
| |
| ``out`` is a list of CTC loss values, one per example in the batch. |
| |
| See *Connectionist Temporal Classification: Labelling Unsegmented |
| Sequence Data <span class="kw">with</span> Recurrent Neural Networks*, A. Graves *et al*. <span class="kw">for</span> more |
| information on the definition and the algorithm. |
| |
| |
| |
| Defined in src/operator/nn/ctc_loss.cc:L100</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#CTCLoss" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="CTCLoss(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="CTCLoss(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">CTCLoss</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@CTCLoss(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Connectionist Temporal Classification Loss. |
| |
| .. note:: The existing alias ``contrib_CTCLoss`` is deprecated. |
| |
| The shapes of the inputs and outputs: |
| |
| - **data**: `(sequence_length, batch_size, alphabet_size)` |
| - **label**: `(batch_size, label_sequence_length)` |
| - **out**: `(batch_size)` |
| |
| The `data` tensor consists of sequences of activation vectors (without applying softmax), |
| <span class="kw">with</span> i-th channel in the last dimension corresponding to i-th label |
| <span class="kw">for</span> i between <span class="num">0</span> and alphabet_size-<span class="num">1</span> (i.e always <span class="num">0</span>-indexed). |
| Alphabet size should include one additional value reserved <span class="kw">for</span> blank label. |
| When `blank_label` is ``<span class="lit">"first"</span>``, the ``<span class="num">0</span>``-th channel is be reserved <span class="kw">for</span> |
| activation of blank label, or otherwise <span class="kw">if</span> it is <span class="lit">"last"</span>, ``(alphabet_size-<span class="num">1</span>)``-th channel should be |
| reserved <span class="kw">for</span> blank label. |
| |
| ``label`` is an index matrix of integers. When `blank_label` is ``<span class="lit">"first"</span>``, |
| the value <span class="num">0</span> is then reserved <span class="kw">for</span> blank label, and should not be passed in <span class="kw">this</span> matrix. Otherwise, |
| when `blank_label` is ``<span class="lit">"last"</span>``, the value `(alphabet_size-<span class="num">1</span>)` is reserved <span class="kw">for</span> blank label. |
| |
| If a sequence of labels is shorter than *label_sequence_length*, use the special |
| padding value at the end of the sequence to conform it to the correct |
| length. The padding value is `<span class="num">0</span>` when `blank_label` is ``<span class="lit">"first"</span>``, and `-<span class="num">1</span>` otherwise. |
| |
| For example, suppose the vocabulary is `[a, b, c]`, and in one batch we have three sequences |
| <span class="lit">'ba'</span>, <span class="lit">'cbb'</span>, and <span class="lit">'abac'</span>. When `blank_label` is ``<span class="lit">"first"</span>``, we can index the labels as |
| `{<span class="lit">'a'</span>: <span class="num">1</span>, <span class="lit">'b'</span>: <span class="num">2</span>, <span class="lit">'c'</span>: <span class="num">3</span>}`, and we reserve the <span class="num">0</span>-th channel <span class="kw">for</span> blank label in data tensor. |
| The resulting `label` tensor should be padded to be:: |
| |
| `[ [<span class="num">2</span>, <span class="num">1</span>, <span class="num">0</span>, <span class="num">0</span>], [<span class="num">3</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">0</span>], [<span class="num">1</span>, <span class="num">2</span>, <span class="num">1</span>, <span class="num">3</span>] ] |
| |
| When `blank_label` is ``<span class="lit">"last"</span>``, we can index the labels as |
| `{<span class="lit">'a'</span>: <span class="num">0</span>, <span class="lit">'b'</span>: <span class="num">1</span>, <span class="lit">'c'</span>: <span class="num">2</span>}`, and we reserve the channel index <span class="num">3</span> <span class="kw">for</span> blank label in data tensor. |
| The resulting `label` tensor should be padded to be:: |
| |
| `[ [<span class="num">1</span>, <span class="num">0</span>, -<span class="num">1</span>, -<span class="num">1</span>], [<span class="num">2</span>, <span class="num">1</span>, <span class="num">1</span>, -<span class="num">1</span>], [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>, <span class="num">2</span>] ] |
| |
| ``out`` is a list of CTC loss values, one per example in the batch. |
| |
| See *Connectionist Temporal Classification: Labelling Unsegmented |
| Sequence Data <span class="kw">with</span> Recurrent Neural Networks*, A. Graves *et al*. <span class="kw">for</span> more |
| information on the definition and the algorithm. |
| |
| |
| |
| Defined in src/operator/nn/ctc_loss.cc:L100</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Cast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Cast(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Cast(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Cast</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Cast(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Casts all elements of the input to a <span class="kw">new</span> <span class="kw">type</span>. |
| |
| .. note:: ``Cast`` is deprecated. Use ``cast`` instead. |
| |
| Example:: |
| |
| cast([<span class="num">0.9</span>, <span class="num">1.3</span>], dtype=<span class="lit">'int32'</span>) = [<span class="num">0</span>, <span class="num">1</span>] |
| cast([<span class="num">1</span>e20, <span class="num">11.1</span>], dtype='float16') = [inf, <span class="num">11.09375</span>] |
| cast([<span class="num">300</span>, <span class="num">11.1</span>, <span class="num">10.9</span>, -<span class="num">1</span>, -<span class="num">3</span>], dtype=<span class="lit">'uint8'</span>) = [<span class="num">44</span>, <span class="num">11</span>, <span class="num">10</span>, <span class="num">255</span>, <span class="num">253</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L664</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Cast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Cast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Cast(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Cast</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Cast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Casts all elements of the input to a <span class="kw">new</span> <span class="kw">type</span>. |
| |
| .. note:: ``Cast`` is deprecated. Use ``cast`` instead. |
| |
| Example:: |
| |
| cast([<span class="num">0.9</span>, <span class="num">1.3</span>], dtype=<span class="lit">'int32'</span>) = [<span class="num">0</span>, <span class="num">1</span>] |
| cast([<span class="num">1</span>e20, <span class="num">11.1</span>], dtype='float16') = [inf, <span class="num">11.09375</span>] |
| cast([<span class="num">300</span>, <span class="num">11.1</span>, <span class="num">10.9</span>, -<span class="num">1</span>, -<span class="num">3</span>], dtype=<span class="lit">'uint8'</span>) = [<span class="num">44</span>, <span class="num">11</span>, <span class="num">10</span>, <span class="num">255</span>, <span class="num">253</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L664</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Concat" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Concat(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Concat(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Concat</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Concat(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Joins input arrays along a given axis. |
| |
| .. note:: `Concat` is deprecated. Use `concat` instead. |
| |
| The dimensions of the input arrays should be the same except the axis along |
| which they will be concatenated. |
| The dimension of the output array along the concatenated axis will be equal |
| to the sum of the corresponding dimensions of the input arrays. |
| |
| The storage <span class="kw">type</span> of ``concat`` output depends on storage types of inputs |
| |
| - concat(csr, csr, ..., csr, dim=<span class="num">0</span>) = csr |
| - otherwise, ``concat`` generates output <span class="kw">with</span> default storage |
| |
| Example:: |
| |
| x = `[ [<span class="num">1</span>,<span class="num">1</span>],[<span class="num">2</span>,<span class="num">2</span>] ] |
| y = `[ [<span class="num">3</span>,<span class="num">3</span>],[<span class="num">4</span>,<span class="num">4</span>],[<span class="num">5</span>,<span class="num">5</span>] ] |
| z = `[ [<span class="num">6</span>,<span class="num">6</span>], [<span class="num">7</span>,<span class="num">7</span>],[<span class="num">8</span>,<span class="num">8</span>] ] |
| |
| concat(x,y,z,dim=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>], |
| [ <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">7.</span>], |
| [ <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| Note that you cannot concat x,y,z along dimension <span class="num">1</span> since dimension |
| <span class="num">0</span> is not the same <span class="kw">for</span> all the input arrays. |
| |
| concat(y,z,dim=<span class="num">1</span>) = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>, <span class="num">7.</span>, <span class="num">7.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/concat.cc:L384</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Concat" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Concat(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Concat(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Concat</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Concat(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Joins input arrays along a given axis. |
| |
| .. note:: `Concat` is deprecated. Use `concat` instead. |
| |
| The dimensions of the input arrays should be the same except the axis along |
| which they will be concatenated. |
| The dimension of the output array along the concatenated axis will be equal |
| to the sum of the corresponding dimensions of the input arrays. |
| |
| The storage <span class="kw">type</span> of ``concat`` output depends on storage types of inputs |
| |
| - concat(csr, csr, ..., csr, dim=<span class="num">0</span>) = csr |
| - otherwise, ``concat`` generates output <span class="kw">with</span> default storage |
| |
| Example:: |
| |
| x = `[ [<span class="num">1</span>,<span class="num">1</span>],[<span class="num">2</span>,<span class="num">2</span>] ] |
| y = `[ [<span class="num">3</span>,<span class="num">3</span>],[<span class="num">4</span>,<span class="num">4</span>],[<span class="num">5</span>,<span class="num">5</span>] ] |
| z = `[ [<span class="num">6</span>,<span class="num">6</span>], [<span class="num">7</span>,<span class="num">7</span>],[<span class="num">8</span>,<span class="num">8</span>] ] |
| |
| concat(x,y,z,dim=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>], |
| [ <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">7.</span>], |
| [ <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| Note that you cannot concat x,y,z along dimension <span class="num">1</span> since dimension |
| <span class="num">0</span> is not the same <span class="kw">for</span> all the input arrays. |
| |
| concat(y,z,dim=<span class="num">1</span>) = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>, <span class="num">7.</span>, <span class="num">7.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/concat.cc:L384</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Convolution" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Convolution(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Convolution(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Convolution</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Convolution(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute *N*-D convolution on *(N+<span class="num">2</span>)*-D input. |
| |
| In the <span class="num">2</span>-D convolution, given input data <span class="kw">with</span> shape *(batch_size, |
| channel, height, width)*, the output is computed by |
| |
| .. math:: |
| |
| out[n,i,:,:] = bias[i] + \sum_{j=<span class="num">0</span>}^{channel} data[n,j,:,:] \star |
| weight[i,j,:,:] |
| |
| where :math:`\star` is the <span class="num">2</span>-D cross-correlation operator. |
| |
| For general <span class="num">2</span>-D convolution, the shapes are |
| |
| - **data**: *(batch_size, channel, height, width)* |
| - **weight**: *(num_filter, channel, kernel[<span class="num">0</span>], kernel[<span class="num">1</span>])* |
| - **bias**: *(num_filter,)* |
| - **out**: *(batch_size, num_filter, out_height, out_width)*. |
| |
| Define:: |
| |
| f(x,k,p,s,d) = floor((x+<span class="num">2</span>*p-d*(k-<span class="num">1</span>)-<span class="num">1</span>)/s)+<span class="num">1</span> |
| |
| then we have:: |
| |
| out_height=f(height, kernel[<span class="num">0</span>], pad[<span class="num">0</span>], stride[<span class="num">0</span>], dilate[<span class="num">0</span>]) |
| out_width=f(width, kernel[<span class="num">1</span>], pad[<span class="num">1</span>], stride[<span class="num">1</span>], dilate[<span class="num">1</span>]) |
| |
| If ``no_bias`` is set to be <span class="kw">true</span>, then the ``bias`` term is ignored. |
| |
| The default data ``layout`` is *NCHW*, namely *(batch_size, channel, height, |
| width)*. We can choose other layouts such as *NWC*. |
| |
| If ``num_group`` is larger than <span class="num">1</span>, denoted by *g*, then split the input ``data`` |
| evenly into *g* parts along the channel axis, and also evenly split ``weight`` |
| along the first dimension. Next compute the convolution on the *i*-th part of |
| the data <span class="kw">with</span> the *i*-th weight part. The output is obtained by concatenating all |
| the *g* results. |
| |
| <span class="num">1</span>-D convolution does not have *height* dimension but only *width* in space. |
| |
| - **data**: *(batch_size, channel, width)* |
| - **weight**: *(num_filter, channel, kernel[<span class="num">0</span>])* |
| - **bias**: *(num_filter,)* |
| - **out**: *(batch_size, num_filter, out_width)*. |
| |
| <span class="num">3</span>-D convolution adds an additional *depth* dimension besides *height* and |
| *width*. The shapes are |
| |
| - **data**: *(batch_size, channel, depth, height, width)* |
| - **weight**: *(num_filter, channel, kernel[<span class="num">0</span>], kernel[<span class="num">1</span>], kernel[<span class="num">2</span>])* |
| - **bias**: *(num_filter,)* |
| - **out**: *(batch_size, num_filter, out_depth, out_height, out_width)*. |
| |
| Both ``weight`` and ``bias`` are learnable parameters. |
| |
| There are other options to tune the performance. |
| |
| - **cudnn_tune**: enable <span class="kw">this</span> option leads to higher startup time but may give |
| faster speed. Options are |
| |
| - **off**: no tuning |
| - **limited_workspace**:run test and pick the fastest algorithm that doesn't |
| exceed workspace limit. |
| - **fastest**: pick the fastest algorithm and ignore workspace limit. |
| - **<span class="std">None</span>** (default): the behavior is determined by environment variable |
| ``MXNET_CUDNN_AUTOTUNE_DEFAULT``. <span class="num">0</span> <span class="kw">for</span> off, <span class="num">1</span> <span class="kw">for</span> limited workspace |
| (default), <span class="num">2</span> <span class="kw">for</span> fastest. |
| |
| - **workspace**: A large number leads to more (GPU) memory usage but may improve |
| the performance. |
| |
| |
| |
| Defined in src/operator/nn/convolution.cc:L475</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Convolution" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Convolution(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Convolution(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Convolution</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Convolution(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute *N*-D convolution on *(N+<span class="num">2</span>)*-D input. |
| |
| In the <span class="num">2</span>-D convolution, given input data <span class="kw">with</span> shape *(batch_size, |
| channel, height, width)*, the output is computed by |
| |
| .. math:: |
| |
| out[n,i,:,:] = bias[i] + \sum_{j=<span class="num">0</span>}^{channel} data[n,j,:,:] \star |
| weight[i,j,:,:] |
| |
| where :math:`\star` is the <span class="num">2</span>-D cross-correlation operator. |
| |
| For general <span class="num">2</span>-D convolution, the shapes are |
| |
| - **data**: *(batch_size, channel, height, width)* |
| - **weight**: *(num_filter, channel, kernel[<span class="num">0</span>], kernel[<span class="num">1</span>])* |
| - **bias**: *(num_filter,)* |
| - **out**: *(batch_size, num_filter, out_height, out_width)*. |
| |
| Define:: |
| |
| f(x,k,p,s,d) = floor((x+<span class="num">2</span>*p-d*(k-<span class="num">1</span>)-<span class="num">1</span>)/s)+<span class="num">1</span> |
| |
| then we have:: |
| |
| out_height=f(height, kernel[<span class="num">0</span>], pad[<span class="num">0</span>], stride[<span class="num">0</span>], dilate[<span class="num">0</span>]) |
| out_width=f(width, kernel[<span class="num">1</span>], pad[<span class="num">1</span>], stride[<span class="num">1</span>], dilate[<span class="num">1</span>]) |
| |
| If ``no_bias`` is set to be <span class="kw">true</span>, then the ``bias`` term is ignored. |
| |
| The default data ``layout`` is *NCHW*, namely *(batch_size, channel, height, |
| width)*. We can choose other layouts such as *NWC*. |
| |
| If ``num_group`` is larger than <span class="num">1</span>, denoted by *g*, then split the input ``data`` |
| evenly into *g* parts along the channel axis, and also evenly split ``weight`` |
| along the first dimension. Next compute the convolution on the *i*-th part of |
| the data <span class="kw">with</span> the *i*-th weight part. The output is obtained by concatenating all |
| the *g* results. |
| |
| <span class="num">1</span>-D convolution does not have *height* dimension but only *width* in space. |
| |
| - **data**: *(batch_size, channel, width)* |
| - **weight**: *(num_filter, channel, kernel[<span class="num">0</span>])* |
| - **bias**: *(num_filter,)* |
| - **out**: *(batch_size, num_filter, out_width)*. |
| |
| <span class="num">3</span>-D convolution adds an additional *depth* dimension besides *height* and |
| *width*. The shapes are |
| |
| - **data**: *(batch_size, channel, depth, height, width)* |
| - **weight**: *(num_filter, channel, kernel[<span class="num">0</span>], kernel[<span class="num">1</span>], kernel[<span class="num">2</span>])* |
| - **bias**: *(num_filter,)* |
| - **out**: *(batch_size, num_filter, out_depth, out_height, out_width)*. |
| |
| Both ``weight`` and ``bias`` are learnable parameters. |
| |
| There are other options to tune the performance. |
| |
| - **cudnn_tune**: enable <span class="kw">this</span> option leads to higher startup time but may give |
| faster speed. Options are |
| |
| - **off**: no tuning |
| - **limited_workspace**:run test and pick the fastest algorithm that doesn't |
| exceed workspace limit. |
| - **fastest**: pick the fastest algorithm and ignore workspace limit. |
| - **<span class="std">None</span>** (default): the behavior is determined by environment variable |
| ``MXNET_CUDNN_AUTOTUNE_DEFAULT``. <span class="num">0</span> <span class="kw">for</span> off, <span class="num">1</span> <span class="kw">for</span> limited workspace |
| (default), <span class="num">2</span> <span class="kw">for</span> fastest. |
| |
| - **workspace**: A large number leads to more (GPU) memory usage but may improve |
| the performance. |
| |
| |
| |
| Defined in src/operator/nn/convolution.cc:L475</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Convolution_v1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Convolution_v1(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Convolution_v1(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Convolution_v1</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Convolution_v1(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>This operator is DEPRECATED. Apply convolution to input then add a bias.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Convolution_v1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Convolution_v1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Convolution_v1(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Convolution_v1</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Convolution_v1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>This operator is DEPRECATED. Apply convolution to input then add a bias.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Correlation" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Correlation(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Correlation(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Correlation</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Correlation(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies correlation to inputs. |
| |
| The correlation layer performs multiplicative patch comparisons between two feature maps. |
| |
| Given two multi-channel feature maps :math:`f_{<span class="num">1</span>}, f_{<span class="num">2</span>}`, <span class="kw">with</span> :math:`w`, :math:`h`, and :math:`c` being their width, height, and number of channels, |
| the correlation layer lets the network compare each patch from :math:`f_{<span class="num">1</span>}` <span class="kw">with</span> each patch from :math:`f_{<span class="num">2</span>}`. |
| |
| For now we consider only a single comparison of two patches. The 'correlation' of two patches centered at :math:`x_{<span class="num">1</span>}` in the first map and |
| :math:`x_{<span class="num">2</span>}` in the second map is then defined as: |
| |
| .. math:: |
| |
| c(x_{<span class="num">1</span>}, x_{<span class="num">2</span>}) = \sum_{o \in [-k,k] \times [-k,k]} <f_{<span class="num">1</span>}(x_{<span class="num">1</span>} + o), f_{<span class="num">2</span>}(x_{<span class="num">2</span>} + o)> |
| |
| <span class="kw">for</span> a square patch of size :math:`K:=<span class="num">2</span>k+<span class="num">1</span>`. |
| |
| Note that the equation above is identical to one step of a convolution in neural networks, but instead of convolving data <span class="kw">with</span> a filter, it convolves data <span class="kw">with</span> other |
| data. For <span class="kw">this</span> reason, it has no training weights. |
| |
| Computing :math:`c(x_{<span class="num">1</span>}, x_{<span class="num">2</span>})` involves :math:`c * K^{<span class="num">2</span>}` multiplications. Comparing all patch combinations involves :math:`w^{<span class="num">2</span>}*h^{<span class="num">2</span>}` such computations. |
| |
| Given a maximum displacement :math:`d`, <span class="kw">for</span> each location :math:`x_{<span class="num">1</span>}` it computes correlations :math:`c(x_{<span class="num">1</span>}, x_{<span class="num">2</span>})` only in a neighborhood of size :math:`D:=<span class="num">2</span>d+<span class="num">1</span>`, |
| by limiting the range of :math:`x_{<span class="num">2</span>}`. We use strides :math:`s_{<span class="num">1</span>}, s_{<span class="num">2</span>}`, to quantize :math:`x_{<span class="num">1</span>}` globally and to quantize :math:`x_{<span class="num">2</span>}` within the neighborhood |
| centered around :math:`x_{<span class="num">1</span>}`. |
| |
| The <span class="kw">final</span> output is defined by the following expression: |
| |
| .. math:: |
| out[n, q, i, j] = c(x_{i, j}, x_{q}) |
| |
| where :math:`i` and :math:`j` enumerate spatial locations in :math:`f_{<span class="num">1</span>}`, and :math:`q` denotes the :math:`q^{th}` neighborhood of :math:`x_{i,j}`. |
| |
| |
| Defined in src/operator/correlation.cc:L197</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Correlation" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Correlation(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Correlation(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Correlation</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Correlation(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies correlation to inputs. |
| |
| The correlation layer performs multiplicative patch comparisons between two feature maps. |
| |
| Given two multi-channel feature maps :math:`f_{<span class="num">1</span>}, f_{<span class="num">2</span>}`, <span class="kw">with</span> :math:`w`, :math:`h`, and :math:`c` being their width, height, and number of channels, |
| the correlation layer lets the network compare each patch from :math:`f_{<span class="num">1</span>}` <span class="kw">with</span> each patch from :math:`f_{<span class="num">2</span>}`. |
| |
| For now we consider only a single comparison of two patches. The 'correlation' of two patches centered at :math:`x_{<span class="num">1</span>}` in the first map and |
| :math:`x_{<span class="num">2</span>}` in the second map is then defined as: |
| |
| .. math:: |
| |
| c(x_{<span class="num">1</span>}, x_{<span class="num">2</span>}) = \sum_{o \in [-k,k] \times [-k,k]} <f_{<span class="num">1</span>}(x_{<span class="num">1</span>} + o), f_{<span class="num">2</span>}(x_{<span class="num">2</span>} + o)> |
| |
| <span class="kw">for</span> a square patch of size :math:`K:=<span class="num">2</span>k+<span class="num">1</span>`. |
| |
| Note that the equation above is identical to one step of a convolution in neural networks, but instead of convolving data <span class="kw">with</span> a filter, it convolves data <span class="kw">with</span> other |
| data. For <span class="kw">this</span> reason, it has no training weights. |
| |
| Computing :math:`c(x_{<span class="num">1</span>}, x_{<span class="num">2</span>})` involves :math:`c * K^{<span class="num">2</span>}` multiplications. Comparing all patch combinations involves :math:`w^{<span class="num">2</span>}*h^{<span class="num">2</span>}` such computations. |
| |
| Given a maximum displacement :math:`d`, <span class="kw">for</span> each location :math:`x_{<span class="num">1</span>}` it computes correlations :math:`c(x_{<span class="num">1</span>}, x_{<span class="num">2</span>})` only in a neighborhood of size :math:`D:=<span class="num">2</span>d+<span class="num">1</span>`, |
| by limiting the range of :math:`x_{<span class="num">2</span>}`. We use strides :math:`s_{<span class="num">1</span>}, s_{<span class="num">2</span>}`, to quantize :math:`x_{<span class="num">1</span>}` globally and to quantize :math:`x_{<span class="num">2</span>}` within the neighborhood |
| centered around :math:`x_{<span class="num">1</span>}`. |
| |
| The <span class="kw">final</span> output is defined by the following expression: |
| |
| .. math:: |
| out[n, q, i, j] = c(x_{i, j}, x_{q}) |
| |
| where :math:`i` and :math:`j` enumerate spatial locations in :math:`f_{<span class="num">1</span>}`, and :math:`q` denotes the :math:`q^{th}` neighborhood of :math:`x_{i,j}`. |
| |
| |
| Defined in src/operator/correlation.cc:L197</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Crop" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Crop(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Crop(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Crop</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Crop(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>.. note:: `Crop` is deprecated. Use `slice` instead. |
| |
| Crop the <span class="num">2</span>nd and <span class="num">3</span>rd dim of input data, <span class="kw">with</span> the corresponding size of h_w or |
| <span class="kw">with</span> width and height of the second input symbol, i.e., <span class="kw">with</span> one input, we need h_w to |
| specify the crop height and width, otherwise the second input symbol's size will be used |
| |
| |
| Defined in src/operator/crop.cc:L49</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Crop" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Crop(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Crop(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Crop</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Crop(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>.. note:: `Crop` is deprecated. Use `slice` instead. |
| |
| Crop the <span class="num">2</span>nd and <span class="num">3</span>rd dim of input data, <span class="kw">with</span> the corresponding size of h_w or |
| <span class="kw">with</span> width and height of the second input symbol, i.e., <span class="kw">with</span> one input, we need h_w to |
| specify the crop height and width, otherwise the second input symbol's size will be used |
| |
| |
| Defined in src/operator/crop.cc:L49</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Custom" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Custom(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Custom(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Custom</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Custom(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Apply a custom operator implemented in a frontend language (like Python). |
| |
| Custom operators should <span class="kw">override</span> required methods like `forward` and `backward`. |
| The custom operator must be registered before it can be used. |
| Please check the tutorial here: https:<span class="cmt">//mxnet.incubator.apache.org/api/faq/new_op</span> |
| |
| |
| |
| Defined in src/operator/custom/custom.cc:L546</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Custom" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Custom(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Custom(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Custom</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Custom(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Apply a custom operator implemented in a frontend language (like Python). |
| |
| Custom operators should <span class="kw">override</span> required methods like `forward` and `backward`. |
| The custom operator must be registered before it can be used. |
| Please check the tutorial here: https:<span class="cmt">//mxnet.incubator.apache.org/api/faq/new_op</span> |
| |
| |
| |
| Defined in src/operator/custom/custom.cc:L546</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Deconvolution" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Deconvolution(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Deconvolution(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Deconvolution</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Deconvolution(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes <span class="num">1</span>D or <span class="num">2</span>D transposed convolution (aka fractionally strided convolution) of the input tensor. This operation can be seen as the gradient of Convolution operation <span class="kw">with</span> respect to its input. Convolution usually reduces the size of the input. Transposed convolution works the other way, going from a smaller input to a larger output <span class="kw">while</span> preserving the connectivity pattern.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Deconvolution" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Deconvolution(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Deconvolution(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Deconvolution</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Deconvolution(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes <span class="num">1</span>D or <span class="num">2</span>D transposed convolution (aka fractionally strided convolution) of the input tensor. This operation can be seen as the gradient of Convolution operation <span class="kw">with</span> respect to its input. Convolution usually reduces the size of the input. Transposed convolution works the other way, going from a smaller input to a larger output <span class="kw">while</span> preserving the connectivity pattern.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Dropout" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Dropout(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Dropout(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Dropout</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Dropout(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies dropout operation to input array. |
| |
| - During training, each element of the input is set to zero <span class="kw">with</span> probability p. |
| The whole array is rescaled by :math:`<span class="num">1</span>/(<span class="num">1</span>-p)` to keep the expected |
| sum of the input unchanged. |
| |
| - During testing, <span class="kw">this</span> operator does not change the input <span class="kw">if</span> mode is 'training'. |
| If mode is <span class="lit">'always'</span>, the same computaion as during training will be applied. |
| |
| Example:: |
| |
| random.seed(<span class="num">998</span>) |
| input_array = array(`[ [<span class="num">3.</span>, <span class="num">0.5</span>, -<span class="num">0.5</span>, <span class="num">2.</span>, <span class="num">7.</span>], |
| [<span class="num">2.</span>, -<span class="num">0.4</span>, <span class="num">7.</span>, <span class="num">3.</span>, <span class="num">0.2</span>] ]) |
| a = symbol.Variable(<span class="lit">'a'</span>) |
| dropout = symbol.Dropout(a, p = <span class="num">0.2</span>) |
| executor = dropout.simple_bind(a = input_array.shape) |
| |
| ## If training |
| executor.forward(is_train = True, a = input_array) |
| executor.outputs |
| `[ [ <span class="num">3.75</span> <span class="num">0.625</span> -<span class="num">0.</span> <span class="num">2.5</span> <span class="num">8.75</span> ] |
| [ <span class="num">2.5</span> -<span class="num">0.5</span> <span class="num">8.75</span> <span class="num">3.75</span> <span class="num">0.</span> ] ] |
| |
| ## If testing |
| executor.forward(is_train = False, a = input_array) |
| executor.outputs |
| `[ [ <span class="num">3.</span> <span class="num">0.5</span> -<span class="num">0.5</span> <span class="num">2.</span> <span class="num">7.</span> ] |
| [ <span class="num">2.</span> -<span class="num">0.4</span> <span class="num">7.</span> <span class="num">3.</span> <span class="num">0.2</span> ] ] |
| |
| |
| Defined in src/operator/nn/dropout.cc:L95</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Dropout" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Dropout(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Dropout(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Dropout</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Dropout(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies dropout operation to input array. |
| |
| - During training, each element of the input is set to zero <span class="kw">with</span> probability p. |
| The whole array is rescaled by :math:`<span class="num">1</span>/(<span class="num">1</span>-p)` to keep the expected |
| sum of the input unchanged. |
| |
| - During testing, <span class="kw">this</span> operator does not change the input <span class="kw">if</span> mode is 'training'. |
| If mode is <span class="lit">'always'</span>, the same computaion as during training will be applied. |
| |
| Example:: |
| |
| random.seed(<span class="num">998</span>) |
| input_array = array(`[ [<span class="num">3.</span>, <span class="num">0.5</span>, -<span class="num">0.5</span>, <span class="num">2.</span>, <span class="num">7.</span>], |
| [<span class="num">2.</span>, -<span class="num">0.4</span>, <span class="num">7.</span>, <span class="num">3.</span>, <span class="num">0.2</span>] ]) |
| a = symbol.Variable(<span class="lit">'a'</span>) |
| dropout = symbol.Dropout(a, p = <span class="num">0.2</span>) |
| executor = dropout.simple_bind(a = input_array.shape) |
| |
| ## If training |
| executor.forward(is_train = True, a = input_array) |
| executor.outputs |
| `[ [ <span class="num">3.75</span> <span class="num">0.625</span> -<span class="num">0.</span> <span class="num">2.5</span> <span class="num">8.75</span> ] |
| [ <span class="num">2.5</span> -<span class="num">0.5</span> <span class="num">8.75</span> <span class="num">3.75</span> <span class="num">0.</span> ] ] |
| |
| ## If testing |
| executor.forward(is_train = False, a = input_array) |
| executor.outputs |
| `[ [ <span class="num">3.</span> <span class="num">0.5</span> -<span class="num">0.5</span> <span class="num">2.</span> <span class="num">7.</span> ] |
| [ <span class="num">2.</span> -<span class="num">0.4</span> <span class="num">7.</span> <span class="num">3.</span> <span class="num">0.2</span> ] ] |
| |
| |
| Defined in src/operator/nn/dropout.cc:L95</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ElementWiseSum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ElementWiseSum(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ElementWiseSum(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ElementWiseSum</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ElementWiseSum(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Adds all input arguments element-wise. |
| |
| .. math:: |
| add\_n(a_1, a_2, ..., a_n) = a_1 + a_2 + ... + a_n |
| |
| ``add_n`` is potentially more efficient than calling ``add`` by `n` times. |
| |
| The storage <span class="kw">type</span> of ``add_n`` output depends on storage types of inputs |
| |
| - add_n(row_sparse, row_sparse, ..) = row_sparse |
| - add_n(default, csr, default) = default |
| - add_n(any input combinations longer than <span class="num">4</span> (><span class="num">4</span>) <span class="kw">with</span> at least one default <span class="kw">type</span>) = default |
| - otherwise, ``add_n`` falls all inputs back to default storage and generates default storage |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_sum.cc:L155</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ElementWiseSum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ElementWiseSum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ElementWiseSum(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ElementWiseSum</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ElementWiseSum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Adds all input arguments element-wise. |
| |
| .. math:: |
| add\_n(a_1, a_2, ..., a_n) = a_1 + a_2 + ... + a_n |
| |
| ``add_n`` is potentially more efficient than calling ``add`` by `n` times. |
| |
| The storage <span class="kw">type</span> of ``add_n`` output depends on storage types of inputs |
| |
| - add_n(row_sparse, row_sparse, ..) = row_sparse |
| - add_n(default, csr, default) = default |
| - add_n(any input combinations longer than <span class="num">4</span> (><span class="num">4</span>) <span class="kw">with</span> at least one default <span class="kw">type</span>) = default |
| - otherwise, ``add_n`` falls all inputs back to default storage and generates default storage |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_sum.cc:L155</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Embedding" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Embedding(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Embedding(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Embedding</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Embedding(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Maps integer indices to vector representations (embeddings). |
| |
| This operator maps words to real-valued vectors in a high-dimensional space, |
| called word embeddings. These embeddings can capture semantic and syntactic properties of the words. |
| For example, it has been noted that in the learned embedding spaces, similar words tend |
| to be close to each other and dissimilar words far apart. |
| |
| For an input array of shape (d1, ..., dK), |
| the shape of an output array is (d1, ..., dK, output_dim). |
| All the input values should be integers in the range [<span class="num">0</span>, input_dim). |
| |
| If the input_dim is ip0 and output_dim is op0, then shape of the embedding weight matrix must be |
| (ip0, op0). |
| |
| When <span class="lit">"sparse_grad"</span> is False, <span class="kw">if</span> any index mentioned is too large, it is replaced by the index that |
| addresses the last vector in an embedding matrix. |
| When <span class="lit">"sparse_grad"</span> is True, an error will be raised <span class="kw">if</span> invalid indices are found. |
| |
| Examples:: |
| |
| input_dim = <span class="num">4</span> |
| output_dim = <span class="num">5</span> |
| |
| <span class="cmt">// Each row in weight matrix y represents a word. So, y = (w0,w1,w2,w3)</span> |
| y = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>, <span class="num">13.</span>, <span class="num">14.</span>], |
| [ <span class="num">15.</span>, <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>, <span class="num">19.</span>] ] |
| |
| <span class="cmt">// Input array x represents n-grams(2-gram). So, x = [(w1,w3), (w0,w2)]</span> |
| x = `[ [ <span class="num">1.</span>, <span class="num">3.</span>], |
| [ <span class="num">0.</span>, <span class="num">2.</span>] ] |
| |
| <span class="cmt">// Mapped input x to its vector representation y.</span> |
| Embedding(x, y, <span class="num">4</span>, <span class="num">5</span>) = `[ `[ [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">15.</span>, <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>, <span class="num">19.</span>] ], |
| |
| `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>, <span class="num">13.</span>, <span class="num">14.</span>] ] ] |
| |
| |
| The storage <span class="kw">type</span> of weight can be either row_sparse or default. |
| |
| .. Note:: |
| |
| If <span class="lit">"sparse_grad"</span> is set to True, the storage <span class="kw">type</span> of gradient w.r.t weights will be |
| <span class="lit">"row_sparse"</span>. Only a subset of optimizers support sparse gradients, including SGD, AdaGrad |
| and Adam. Note that by default <span class="kw">lazy</span> updates is turned on, which may perform differently |
| from standard updates. For more details, please check the Optimization API at: |
| https:<span class="cmt">//mxnet.incubator.apache.org/api/python/optimization/optimization.html</span> |
| |
| |
| |
| Defined in src/operator/tensor/indexing_op.cc:L597</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Embedding" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Embedding(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Embedding(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Embedding</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Embedding(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Maps integer indices to vector representations (embeddings). |
| |
| This operator maps words to real-valued vectors in a high-dimensional space, |
| called word embeddings. These embeddings can capture semantic and syntactic properties of the words. |
| For example, it has been noted that in the learned embedding spaces, similar words tend |
| to be close to each other and dissimilar words far apart. |
| |
| For an input array of shape (d1, ..., dK), |
| the shape of an output array is (d1, ..., dK, output_dim). |
| All the input values should be integers in the range [<span class="num">0</span>, input_dim). |
| |
| If the input_dim is ip0 and output_dim is op0, then shape of the embedding weight matrix must be |
| (ip0, op0). |
| |
| When <span class="lit">"sparse_grad"</span> is False, <span class="kw">if</span> any index mentioned is too large, it is replaced by the index that |
| addresses the last vector in an embedding matrix. |
| When <span class="lit">"sparse_grad"</span> is True, an error will be raised <span class="kw">if</span> invalid indices are found. |
| |
| Examples:: |
| |
| input_dim = <span class="num">4</span> |
| output_dim = <span class="num">5</span> |
| |
| <span class="cmt">// Each row in weight matrix y represents a word. So, y = (w0,w1,w2,w3)</span> |
| y = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>, <span class="num">13.</span>, <span class="num">14.</span>], |
| [ <span class="num">15.</span>, <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>, <span class="num">19.</span>] ] |
| |
| <span class="cmt">// Input array x represents n-grams(2-gram). So, x = [(w1,w3), (w0,w2)]</span> |
| x = `[ [ <span class="num">1.</span>, <span class="num">3.</span>], |
| [ <span class="num">0.</span>, <span class="num">2.</span>] ] |
| |
| <span class="cmt">// Mapped input x to its vector representation y.</span> |
| Embedding(x, y, <span class="num">4</span>, <span class="num">5</span>) = `[ `[ [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">15.</span>, <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>, <span class="num">19.</span>] ], |
| |
| `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>, <span class="num">13.</span>, <span class="num">14.</span>] ] ] |
| |
| |
| The storage <span class="kw">type</span> of weight can be either row_sparse or default. |
| |
| .. Note:: |
| |
| If <span class="lit">"sparse_grad"</span> is set to True, the storage <span class="kw">type</span> of gradient w.r.t weights will be |
| <span class="lit">"row_sparse"</span>. Only a subset of optimizers support sparse gradients, including SGD, AdaGrad |
| and Adam. Note that by default <span class="kw">lazy</span> updates is turned on, which may perform differently |
| from standard updates. For more details, please check the Optimization API at: |
| https:<span class="cmt">//mxnet.incubator.apache.org/api/python/optimization/optimization.html</span> |
| |
| |
| |
| Defined in src/operator/tensor/indexing_op.cc:L597</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Flatten" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Flatten(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Flatten(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Flatten</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Flatten(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Flattens the input array into a <span class="num">2</span>-D array by collapsing the higher dimensions. |
| .. note:: `Flatten` is deprecated. Use `flatten` instead. |
| For an input array <span class="kw">with</span> shape ``(d1, d2, ..., dk)``, `flatten` operation reshapes |
| the input array into an output array of shape ``(d1, d2*...*dk)``. |
| Note that the behavior of <span class="kw">this</span> function is different from numpy.ndarray.flatten, |
| which behaves similar to mxnet.ndarray.reshape((-<span class="num">1</span>,)). |
| Example:: |
| x = `[ [ |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>], |
| [<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] |
| ], |
| [ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>], |
| [<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] |
| ] ], |
| flatten(x) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L249</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Flatten" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Flatten(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Flatten(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Flatten</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Flatten(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Flattens the input array into a <span class="num">2</span>-D array by collapsing the higher dimensions. |
| .. note:: `Flatten` is deprecated. Use `flatten` instead. |
| For an input array <span class="kw">with</span> shape ``(d1, d2, ..., dk)``, `flatten` operation reshapes |
| the input array into an output array of shape ``(d1, d2*...*dk)``. |
| Note that the behavior of <span class="kw">this</span> function is different from numpy.ndarray.flatten, |
| which behaves similar to mxnet.ndarray.reshape((-<span class="num">1</span>,)). |
| Example:: |
| x = `[ [ |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>], |
| [<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] |
| ], |
| [ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>], |
| [<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] |
| ] ], |
| flatten(x) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L249</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#FullyConnected" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="FullyConnected(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="FullyConnected(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">FullyConnected</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@FullyConnected(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies a linear transformation: :math:`Y = XW^T + b`. |
| |
| If ``flatten`` is set to be <span class="kw">true</span>, then the shapes are: |
| |
| - **data**: `(batch_size, x1, x2, ..., xn)` |
| - **weight**: `(num_hidden, x1 * x2 * ... * xn)` |
| - **bias**: `(num_hidden,)` |
| - **out**: `(batch_size, num_hidden)` |
| |
| If ``flatten`` is set to be <span class="kw">false</span>, then the shapes are: |
| |
| - **data**: `(x1, x2, ..., xn, input_dim)` |
| - **weight**: `(num_hidden, input_dim)` |
| - **bias**: `(num_hidden,)` |
| - **out**: `(x1, x2, ..., xn, num_hidden)` |
| |
| The learnable parameters include both ``weight`` and ``bias``. |
| |
| If ``no_bias`` is set to be <span class="kw">true</span>, then the ``bias`` term is ignored. |
| |
| .. Note:: |
| |
| The sparse support <span class="kw">for</span> FullyConnected is limited to forward evaluation <span class="kw">with</span> `row_sparse` |
| weight and bias, where the length of `weight.indices` and `bias.indices` must be equal |
| to `num_hidden`. This could be useful <span class="kw">for</span> model inference <span class="kw">with</span> `row_sparse` weights |
| trained <span class="kw">with</span> importance sampling or noise contrastive estimation. |
| |
| To compute linear transformation <span class="kw">with</span> <span class="lit">'csr'</span> sparse data, sparse.dot is recommended instead |
| of sparse.FullyConnected. |
| |
| |
| |
| Defined in src/operator/nn/fully_connected.cc:L286</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#FullyConnected" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="FullyConnected(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="FullyConnected(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">FullyConnected</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@FullyConnected(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies a linear transformation: :math:`Y = XW^T + b`. |
| |
| If ``flatten`` is set to be <span class="kw">true</span>, then the shapes are: |
| |
| - **data**: `(batch_size, x1, x2, ..., xn)` |
| - **weight**: `(num_hidden, x1 * x2 * ... * xn)` |
| - **bias**: `(num_hidden,)` |
| - **out**: `(batch_size, num_hidden)` |
| |
| If ``flatten`` is set to be <span class="kw">false</span>, then the shapes are: |
| |
| - **data**: `(x1, x2, ..., xn, input_dim)` |
| - **weight**: `(num_hidden, input_dim)` |
| - **bias**: `(num_hidden,)` |
| - **out**: `(x1, x2, ..., xn, num_hidden)` |
| |
| The learnable parameters include both ``weight`` and ``bias``. |
| |
| If ``no_bias`` is set to be <span class="kw">true</span>, then the ``bias`` term is ignored. |
| |
| .. Note:: |
| |
| The sparse support <span class="kw">for</span> FullyConnected is limited to forward evaluation <span class="kw">with</span> `row_sparse` |
| weight and bias, where the length of `weight.indices` and `bias.indices` must be equal |
| to `num_hidden`. This could be useful <span class="kw">for</span> model inference <span class="kw">with</span> `row_sparse` weights |
| trained <span class="kw">with</span> importance sampling or noise contrastive estimation. |
| |
| To compute linear transformation <span class="kw">with</span> <span class="lit">'csr'</span> sparse data, sparse.dot is recommended instead |
| of sparse.FullyConnected. |
| |
| |
| |
| Defined in src/operator/nn/fully_connected.cc:L286</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#GridGenerator" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="GridGenerator(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="GridGenerator(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">GridGenerator</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@GridGenerator(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Generates <span class="num">2</span>D sampling grid <span class="kw">for</span> bilinear sampling.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#GridGenerator" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="GridGenerator(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="GridGenerator(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">GridGenerator</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@GridGenerator(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Generates <span class="num">2</span>D sampling grid <span class="kw">for</span> bilinear sampling.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#GroupNorm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="GroupNorm(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="GroupNorm(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">GroupNorm</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@GroupNorm(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Group normalization. |
| |
| The input channels are separated into ``num_groups`` groups, each containing ``num_channels / num_groups`` channels. |
| The mean and standard-deviation are calculated separately over the each group. |
| |
| .. math:: |
| |
| data = data.reshape((N, num_groups, C <span class="cmt">// num_groups, ...))</span> |
| out = \frac{data - mean(data, axis)}{\sqrt{<span class="kw">var</span>(data, axis) + \epsilon}} * gamma + beta |
| |
| Both ``gamma`` and ``beta`` are learnable parameters. |
| |
| |
| |
| Defined in src/operator/nn/group_norm.cc:L76</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#GroupNorm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="GroupNorm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="GroupNorm(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">GroupNorm</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@GroupNorm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Group normalization. |
| |
| The input channels are separated into ``num_groups`` groups, each containing ``num_channels / num_groups`` channels. |
| The mean and standard-deviation are calculated separately over the each group. |
| |
| .. math:: |
| |
| data = data.reshape((N, num_groups, C <span class="cmt">// num_groups, ...))</span> |
| out = \frac{data - mean(data, axis)}{\sqrt{<span class="kw">var</span>(data, axis) + \epsilon}} * gamma + beta |
| |
| Both ``gamma`` and ``beta`` are learnable parameters. |
| |
| |
| |
| Defined in src/operator/nn/group_norm.cc:L76</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#IdentityAttachKLSparseReg" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="IdentityAttachKLSparseReg(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="IdentityAttachKLSparseReg(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">IdentityAttachKLSparseReg</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@IdentityAttachKLSparseReg(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Apply a sparse regularization to the output a sigmoid activation function.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#IdentityAttachKLSparseReg" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="IdentityAttachKLSparseReg(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="IdentityAttachKLSparseReg(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">IdentityAttachKLSparseReg</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@IdentityAttachKLSparseReg(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Apply a sparse regularization to the output a sigmoid activation function.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#InstanceNorm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="InstanceNorm(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="InstanceNorm(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">InstanceNorm</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@InstanceNorm(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies instance normalization to the n-dimensional input array. |
| |
| This operator takes an n-dimensional input array where (n><span class="num">2</span>) and normalizes |
| the input using the following formula: |
| |
| .. math:: |
| |
| out = \frac{x - mean[data]}{ \sqrt{Var[data]} + \epsilon} * gamma + beta |
| |
| This layer is similar to batch normalization layer (`BatchNorm`) |
| <span class="kw">with</span> two differences: first, the normalization is |
| carried out per example (instance), not over a batch. Second, the |
| same normalization is applied both at test and train time. This |
| operation is also known as `contrast normalization`. |
| |
| If the input data is of shape [batch, channel, spacial_dim1, spacial_dim2, ...], |
| `gamma` and `beta` parameters must be vectors of shape [channel]. |
| |
| This implementation is based on <span class="kw">this</span> paper [<span class="num">1</span>]_ |
| |
| .. [<span class="num">1</span>] Instance Normalization: The Missing Ingredient <span class="kw">for</span> Fast Stylization, |
| D. Ulyanov, A. Vedaldi, V. Lempitsky, <span class="num">2016</span> (arXiv:<span class="num">1607.08022</span>v2). |
| |
| Examples:: |
| |
| <span class="cmt">// Input of shape (2,1,2)</span> |
| x = `[ `[ [ <span class="num">1.1</span>, <span class="num">2.2</span>] ], |
| `[ [ <span class="num">3.3</span>, <span class="num">4.4</span>] ] ] |
| |
| <span class="cmt">// gamma parameter of length 1</span> |
| gamma = [<span class="num">1.5</span>] |
| |
| <span class="cmt">// beta parameter of length 1</span> |
| beta = [<span class="num">0.5</span>] |
| |
| <span class="cmt">// Instance normalization is calculated with the above formula</span> |
| InstanceNorm(x,gamma,beta) = `[ `[ [-<span class="num">0.997527</span> , <span class="num">1.99752665</span>] ], |
| `[ [-<span class="num">0.99752653</span>, <span class="num">1.99752724</span>] ] ] |
| |
| |
| |
| Defined in src/operator/instance_norm.cc:L94</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#InstanceNorm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="InstanceNorm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="InstanceNorm(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">InstanceNorm</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@InstanceNorm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies instance normalization to the n-dimensional input array. |
| |
| This operator takes an n-dimensional input array where (n><span class="num">2</span>) and normalizes |
| the input using the following formula: |
| |
| .. math:: |
| |
| out = \frac{x - mean[data]}{ \sqrt{Var[data]} + \epsilon} * gamma + beta |
| |
| This layer is similar to batch normalization layer (`BatchNorm`) |
| <span class="kw">with</span> two differences: first, the normalization is |
| carried out per example (instance), not over a batch. Second, the |
| same normalization is applied both at test and train time. This |
| operation is also known as `contrast normalization`. |
| |
| If the input data is of shape [batch, channel, spacial_dim1, spacial_dim2, ...], |
| `gamma` and `beta` parameters must be vectors of shape [channel]. |
| |
| This implementation is based on <span class="kw">this</span> paper [<span class="num">1</span>]_ |
| |
| .. [<span class="num">1</span>] Instance Normalization: The Missing Ingredient <span class="kw">for</span> Fast Stylization, |
| D. Ulyanov, A. Vedaldi, V. Lempitsky, <span class="num">2016</span> (arXiv:<span class="num">1607.08022</span>v2). |
| |
| Examples:: |
| |
| <span class="cmt">// Input of shape (2,1,2)</span> |
| x = `[ `[ [ <span class="num">1.1</span>, <span class="num">2.2</span>] ], |
| `[ [ <span class="num">3.3</span>, <span class="num">4.4</span>] ] ] |
| |
| <span class="cmt">// gamma parameter of length 1</span> |
| gamma = [<span class="num">1.5</span>] |
| |
| <span class="cmt">// beta parameter of length 1</span> |
| beta = [<span class="num">0.5</span>] |
| |
| <span class="cmt">// Instance normalization is calculated with the above formula</span> |
| InstanceNorm(x,gamma,beta) = `[ `[ [-<span class="num">0.997527</span> , <span class="num">1.99752665</span>] ], |
| `[ [-<span class="num">0.99752653</span>, <span class="num">1.99752724</span>] ] ] |
| |
| |
| |
| Defined in src/operator/instance_norm.cc:L94</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#L2Normalization" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="L2Normalization(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="L2Normalization(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">L2Normalization</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@L2Normalization(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Normalize the input array using the L2 norm. |
| |
| For <span class="num">1</span>-D NDArray, it computes:: |
| |
| out = data / sqrt(sum(data ** <span class="num">2</span>) + eps) |
| |
| For N-D NDArray, <span class="kw">if</span> the input array has shape (N, N, ..., N), |
| |
| <span class="kw">with</span> ``mode`` = ``instance``, it normalizes each instance in the multidimensional |
| array by its L2 norm.:: |
| |
| <span class="kw">for</span> i in <span class="num">0.</span>..N |
| out[i,:,:,...,:] = data[i,:,:,...,:] / sqrt(sum(data[i,:,:,...,:] ** <span class="num">2</span>) + eps) |
| |
| <span class="kw">with</span> ``mode`` = ``channel``, it normalizes each channel in the array by its L2 norm.:: |
| |
| <span class="kw">for</span> i in <span class="num">0.</span>..N |
| out[:,i,:,...,:] = data[:,i,:,...,:] / sqrt(sum(data[:,i,:,...,:] ** <span class="num">2</span>) + eps) |
| |
| <span class="kw">with</span> ``mode`` = ``spatial``, it normalizes the cross channel norm <span class="kw">for</span> each position |
| in the array by its L2 norm.:: |
| |
| <span class="kw">for</span> dim in <span class="num">2.</span>..N |
| <span class="kw">for</span> i in <span class="num">0.</span>..N |
| out[.....,i,...] = take(out, indices=i, axis=dim) / sqrt(sum(take(out, indices=i, axis=dim) ** <span class="num">2</span>) + eps) |
| -dim- |
| |
| Example:: |
| |
| x = `[ `[ [<span class="num">1</span>,<span class="num">2</span>], |
| [<span class="num">3</span>,<span class="num">4</span>] ], |
| `[ [<span class="num">2</span>,<span class="num">2</span>], |
| [<span class="num">5</span>,<span class="num">6</span>] ] ] |
| |
| L2Normalization(x, mode='instance') |
| =`[ `[ [ <span class="num">0.18257418</span> <span class="num">0.36514837</span>] |
| [ <span class="num">0.54772252</span> <span class="num">0.73029673</span>] ] |
| `[ [ <span class="num">0.24077171</span> <span class="num">0.24077171</span>] |
| [ <span class="num">0.60192931</span> <span class="num">0.72231513</span>] ] ] |
| |
| L2Normalization(x, mode='channel') |
| =`[ `[ [ <span class="num">0.31622776</span> <span class="num">0.44721359</span>] |
| [ <span class="num">0.94868326</span> <span class="num">0.89442718</span>] ] |
| `[ [ <span class="num">0.37139067</span> <span class="num">0.31622776</span>] |
| [ <span class="num">0.92847669</span> <span class="num">0.94868326</span>] ] ] |
| |
| L2Normalization(x, mode='spatial') |
| =`[ `[ [ <span class="num">0.44721359</span> <span class="num">0.89442718</span>] |
| [ <span class="num">0.60000002</span> <span class="num">0.80000001</span>] ] |
| `[ [ <span class="num">0.70710677</span> <span class="num">0.70710677</span>] |
| [ <span class="num">0.6401844</span> <span class="num">0.76822126</span>] ] ] |
| |
| |
| |
| Defined in src/operator/l2_normalization.cc:L195</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#L2Normalization" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="L2Normalization(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="L2Normalization(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">L2Normalization</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@L2Normalization(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Normalize the input array using the L2 norm. |
| |
| For <span class="num">1</span>-D NDArray, it computes:: |
| |
| out = data / sqrt(sum(data ** <span class="num">2</span>) + eps) |
| |
| For N-D NDArray, <span class="kw">if</span> the input array has shape (N, N, ..., N), |
| |
| <span class="kw">with</span> ``mode`` = ``instance``, it normalizes each instance in the multidimensional |
| array by its L2 norm.:: |
| |
| <span class="kw">for</span> i in <span class="num">0.</span>..N |
| out[i,:,:,...,:] = data[i,:,:,...,:] / sqrt(sum(data[i,:,:,...,:] ** <span class="num">2</span>) + eps) |
| |
| <span class="kw">with</span> ``mode`` = ``channel``, it normalizes each channel in the array by its L2 norm.:: |
| |
| <span class="kw">for</span> i in <span class="num">0.</span>..N |
| out[:,i,:,...,:] = data[:,i,:,...,:] / sqrt(sum(data[:,i,:,...,:] ** <span class="num">2</span>) + eps) |
| |
| <span class="kw">with</span> ``mode`` = ``spatial``, it normalizes the cross channel norm <span class="kw">for</span> each position |
| in the array by its L2 norm.:: |
| |
| <span class="kw">for</span> dim in <span class="num">2.</span>..N |
| <span class="kw">for</span> i in <span class="num">0.</span>..N |
| out[.....,i,...] = take(out, indices=i, axis=dim) / sqrt(sum(take(out, indices=i, axis=dim) ** <span class="num">2</span>) + eps) |
| -dim- |
| |
| Example:: |
| |
| x = `[ `[ [<span class="num">1</span>,<span class="num">2</span>], |
| [<span class="num">3</span>,<span class="num">4</span>] ], |
| `[ [<span class="num">2</span>,<span class="num">2</span>], |
| [<span class="num">5</span>,<span class="num">6</span>] ] ] |
| |
| L2Normalization(x, mode='instance') |
| =`[ `[ [ <span class="num">0.18257418</span> <span class="num">0.36514837</span>] |
| [ <span class="num">0.54772252</span> <span class="num">0.73029673</span>] ] |
| `[ [ <span class="num">0.24077171</span> <span class="num">0.24077171</span>] |
| [ <span class="num">0.60192931</span> <span class="num">0.72231513</span>] ] ] |
| |
| L2Normalization(x, mode='channel') |
| =`[ `[ [ <span class="num">0.31622776</span> <span class="num">0.44721359</span>] |
| [ <span class="num">0.94868326</span> <span class="num">0.89442718</span>] ] |
| `[ [ <span class="num">0.37139067</span> <span class="num">0.31622776</span>] |
| [ <span class="num">0.92847669</span> <span class="num">0.94868326</span>] ] ] |
| |
| L2Normalization(x, mode='spatial') |
| =`[ `[ [ <span class="num">0.44721359</span> <span class="num">0.89442718</span>] |
| [ <span class="num">0.60000002</span> <span class="num">0.80000001</span>] ] |
| `[ [ <span class="num">0.70710677</span> <span class="num">0.70710677</span>] |
| [ <span class="num">0.6401844</span> <span class="num">0.76822126</span>] ] ] |
| |
| |
| |
| Defined in src/operator/l2_normalization.cc:L195</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LRN" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LRN(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LRN(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LRN</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LRN(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies local response normalization to the input. |
| |
| The local response normalization layer performs <span class="lit">"lateral inhibition"</span> by normalizing |
| over local input regions. |
| |
| If :math:`a_{x,y}^{i}` is the activity of a neuron computed by applying kernel :math:`i` at position |
| :math:`(x, y)` and then applying the ReLU nonlinearity, the response-normalized |
| activity :math:`b_{x,y}^{i}` is given by the expression: |
| |
| .. math:: |
| b_{x,y}^{i} = \frac{a_{x,y}^{i}}{\Bigg({k + \frac{\alpha}{n} \sum_{j=max(<span class="num">0</span>, i-\frac{n}{<span class="num">2</span>})}^{min(N-<span class="num">1</span>, i+\frac{n}{<span class="num">2</span>})} (a_{x,y}^{j})^{<span class="num">2</span>}}\Bigg)^{\beta}} |
| |
| where the sum runs over :math:`n` <span class="lit">"adjacent"</span> kernel maps at the same spatial position, and :math:`N` is the total |
| number of kernels in the layer. |
| |
| |
| |
| Defined in src/operator/nn/lrn.cc:L157</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LRN" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LRN(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LRN(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LRN</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LRN(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies local response normalization to the input. |
| |
| The local response normalization layer performs <span class="lit">"lateral inhibition"</span> by normalizing |
| over local input regions. |
| |
| If :math:`a_{x,y}^{i}` is the activity of a neuron computed by applying kernel :math:`i` at position |
| :math:`(x, y)` and then applying the ReLU nonlinearity, the response-normalized |
| activity :math:`b_{x,y}^{i}` is given by the expression: |
| |
| .. math:: |
| b_{x,y}^{i} = \frac{a_{x,y}^{i}}{\Bigg({k + \frac{\alpha}{n} \sum_{j=max(<span class="num">0</span>, i-\frac{n}{<span class="num">2</span>})}^{min(N-<span class="num">1</span>, i+\frac{n}{<span class="num">2</span>})} (a_{x,y}^{j})^{<span class="num">2</span>}}\Bigg)^{\beta}} |
| |
| where the sum runs over :math:`n` <span class="lit">"adjacent"</span> kernel maps at the same spatial position, and :math:`N` is the total |
| number of kernels in the layer. |
| |
| |
| |
| Defined in src/operator/nn/lrn.cc:L157</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LayerNorm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LayerNorm(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LayerNorm(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LayerNorm</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LayerNorm(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Layer normalization. |
| |
| Normalizes the channels of the input tensor by mean and variance, and applies a scale ``gamma`` as |
| well as offset ``beta``. |
| |
| Assume the input has more than one dimension and we normalize along axis <span class="num">1.</span> |
| We first compute the mean and variance along <span class="kw">this</span> axis and then |
| compute the normalized output, which has the same shape as input, as following: |
| |
| .. math:: |
| |
| out = \frac{data - mean(data, axis)}{\sqrt{<span class="kw">var</span>(data, axis) + \epsilon}} * gamma + beta |
| |
| Both ``gamma`` and ``beta`` are learnable parameters. |
| |
| Unlike BatchNorm and InstanceNorm, the *mean* and *<span class="kw">var</span>* are computed along the channel dimension. |
| |
| Assume the input has size *k* on axis <span class="num">1</span>, then both ``gamma`` and ``beta`` |
| have shape *(k,)*. If ``output_mean_var`` is set to be <span class="kw">true</span>, then outputs both ``data_mean`` and |
| ``data_std``. Note that no gradient will be passed through these two outputs. |
| |
| The parameter ``axis`` specifies which axis of the input shape denotes |
| the 'channel' (separately normalized groups). The default is -<span class="num">1</span>, which sets the channel |
| axis to be the last item in the input shape. |
| |
| |
| |
| Defined in src/operator/nn/layer_norm.cc:L201</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LayerNorm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LayerNorm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LayerNorm(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LayerNorm</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LayerNorm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Layer normalization. |
| |
| Normalizes the channels of the input tensor by mean and variance, and applies a scale ``gamma`` as |
| well as offset ``beta``. |
| |
| Assume the input has more than one dimension and we normalize along axis <span class="num">1.</span> |
| We first compute the mean and variance along <span class="kw">this</span> axis and then |
| compute the normalized output, which has the same shape as input, as following: |
| |
| .. math:: |
| |
| out = \frac{data - mean(data, axis)}{\sqrt{<span class="kw">var</span>(data, axis) + \epsilon}} * gamma + beta |
| |
| Both ``gamma`` and ``beta`` are learnable parameters. |
| |
| Unlike BatchNorm and InstanceNorm, the *mean* and *<span class="kw">var</span>* are computed along the channel dimension. |
| |
| Assume the input has size *k* on axis <span class="num">1</span>, then both ``gamma`` and ``beta`` |
| have shape *(k,)*. If ``output_mean_var`` is set to be <span class="kw">true</span>, then outputs both ``data_mean`` and |
| ``data_std``. Note that no gradient will be passed through these two outputs. |
| |
| The parameter ``axis`` specifies which axis of the input shape denotes |
| the 'channel' (separately normalized groups). The default is -<span class="num">1</span>, which sets the channel |
| axis to be the last item in the input shape. |
| |
| |
| |
| Defined in src/operator/nn/layer_norm.cc:L201</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LeakyReLU" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LeakyReLU(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LeakyReLU(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LeakyReLU</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LeakyReLU(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies Leaky rectified linear unit activation element-wise to the input. |
| |
| Leaky ReLUs attempt to fix the <span class="lit">"dying ReLU"</span> problem by allowing a small `slope` |
| when the input is negative and has a slope of one when input is positive. |
| |
| The following modified ReLU Activation functions are supported: |
| |
| - *elu*: Exponential Linear <span class="std">Unit</span>. `y = x > <span class="num">0</span> ? x : slope * (exp(x)-<span class="num">1</span>)` |
| - *selu*: Scaled Exponential Linear <span class="std">Unit</span>. `y = lambda * (x > <span class="num">0</span> ? x : alpha * (exp(x) - <span class="num">1</span>))` where |
| *lambda = <span class="num">1.0507009873554804934193349852946</span>* and *alpha = <span class="num">1.6732632423543772848170429916717</span>*. |
| - *leaky*: Leaky ReLU. `y = x > <span class="num">0</span> ? x : slope * x` |
| - *prelu*: Parametric ReLU. This is same as *leaky* except that `slope` is learnt during training. |
| - *rrelu*: Randomized ReLU. same as *leaky* but the `slope` is uniformly and randomly chosen from |
| *[lower_bound, upper_bound)* <span class="kw">for</span> training, <span class="kw">while</span> fixed to be |
| *(lower_bound+upper_bound)/<span class="num">2</span>* <span class="kw">for</span> inference. |
| |
| |
| |
| Defined in src/operator/leaky_relu.cc:L162</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LeakyReLU" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LeakyReLU(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LeakyReLU(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LeakyReLU</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LeakyReLU(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies Leaky rectified linear unit activation element-wise to the input. |
| |
| Leaky ReLUs attempt to fix the <span class="lit">"dying ReLU"</span> problem by allowing a small `slope` |
| when the input is negative and has a slope of one when input is positive. |
| |
| The following modified ReLU Activation functions are supported: |
| |
| - *elu*: Exponential Linear <span class="std">Unit</span>. `y = x > <span class="num">0</span> ? x : slope * (exp(x)-<span class="num">1</span>)` |
| - *selu*: Scaled Exponential Linear <span class="std">Unit</span>. `y = lambda * (x > <span class="num">0</span> ? x : alpha * (exp(x) - <span class="num">1</span>))` where |
| *lambda = <span class="num">1.0507009873554804934193349852946</span>* and *alpha = <span class="num">1.6732632423543772848170429916717</span>*. |
| - *leaky*: Leaky ReLU. `y = x > <span class="num">0</span> ? x : slope * x` |
| - *prelu*: Parametric ReLU. This is same as *leaky* except that `slope` is learnt during training. |
| - *rrelu*: Randomized ReLU. same as *leaky* but the `slope` is uniformly and randomly chosen from |
| *[lower_bound, upper_bound)* <span class="kw">for</span> training, <span class="kw">while</span> fixed to be |
| *(lower_bound+upper_bound)/<span class="num">2</span>* <span class="kw">for</span> inference. |
| |
| |
| |
| Defined in src/operator/leaky_relu.cc:L162</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LinearRegressionOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LinearRegressionOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LinearRegressionOutput(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LinearRegressionOutput</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LinearRegressionOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes and optimizes <span class="kw">for</span> squared loss during backward propagation. |
| Just outputs ``data`` during forward propagation. |
| |
| If :math:`\hat{y}_i` is the predicted value of the i-th sample, and :math:`y_i` is the corresponding target value, |
| then the squared loss estimated over :math:`n` samples is defined as |
| |
| :math:`\text{SquaredLoss}(\textbf{Y}, \hat{\textbf{Y}} ) = \frac{<span class="num">1</span>}{n} \sum_{i=<span class="num">0</span>}^{n-<span class="num">1</span>} \lVert \textbf{y}_i - \hat{\textbf{y}}_i \rVert_2` |
| |
| .. note:: |
| Use the LinearRegressionOutput as the <span class="kw">final</span> output layer of a net. |
| |
| The storage <span class="kw">type</span> of ``label`` can be ``default`` or ``csr`` |
| |
| - LinearRegressionOutput(default, default) = default |
| - LinearRegressionOutput(default, csr) = default |
| |
| By default, gradients of <span class="kw">this</span> loss function are scaled by factor `<span class="num">1</span>/m`, where m is the number of regression outputs of a training example. |
| The parameter `grad_scale` can be used to change <span class="kw">this</span> scale to `grad_scale/m`. |
| |
| |
| |
| Defined in src/operator/regression_output.cc:L92</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LinearRegressionOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LinearRegressionOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LinearRegressionOutput(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LinearRegressionOutput</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LinearRegressionOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes and optimizes <span class="kw">for</span> squared loss during backward propagation. |
| Just outputs ``data`` during forward propagation. |
| |
| If :math:`\hat{y}_i` is the predicted value of the i-th sample, and :math:`y_i` is the corresponding target value, |
| then the squared loss estimated over :math:`n` samples is defined as |
| |
| :math:`\text{SquaredLoss}(\textbf{Y}, \hat{\textbf{Y}} ) = \frac{<span class="num">1</span>}{n} \sum_{i=<span class="num">0</span>}^{n-<span class="num">1</span>} \lVert \textbf{y}_i - \hat{\textbf{y}}_i \rVert_2` |
| |
| .. note:: |
| Use the LinearRegressionOutput as the <span class="kw">final</span> output layer of a net. |
| |
| The storage <span class="kw">type</span> of ``label`` can be ``default`` or ``csr`` |
| |
| - LinearRegressionOutput(default, default) = default |
| - LinearRegressionOutput(default, csr) = default |
| |
| By default, gradients of <span class="kw">this</span> loss function are scaled by factor `<span class="num">1</span>/m`, where m is the number of regression outputs of a training example. |
| The parameter `grad_scale` can be used to change <span class="kw">this</span> scale to `grad_scale/m`. |
| |
| |
| |
| Defined in src/operator/regression_output.cc:L92</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LogisticRegressionOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LogisticRegressionOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LogisticRegressionOutput(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LogisticRegressionOutput</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LogisticRegressionOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies a logistic function to the input. |
| |
| The logistic function, also known as the sigmoid function, is computed as |
| :math:`\frac{<span class="num">1</span>}{<span class="num">1</span>+exp(-\textbf{x})}`. |
| |
| Commonly, the sigmoid is used to squash the real-valued output of a linear model |
| :math:`wTx+b` into the [<span class="num">0</span>,<span class="num">1</span>] range so that it can be interpreted as a probability. |
| It is suitable <span class="kw">for</span> binary classification or probability prediction tasks. |
| |
| .. note:: |
| Use the LogisticRegressionOutput as the <span class="kw">final</span> output layer of a net. |
| |
| The storage <span class="kw">type</span> of ``label`` can be ``default`` or ``csr`` |
| |
| - LogisticRegressionOutput(default, default) = default |
| - LogisticRegressionOutput(default, csr) = default |
| |
| The loss function used is the Binary Cross Entropy Loss: |
| |
| :math:`-{(y\log(p) + (<span class="num">1</span> - y)\log(<span class="num">1</span> - p))}` |
| |
| Where `y` is the ground truth probability of positive outcome <span class="kw">for</span> a given example, and `p` the probability predicted by the model. By default, gradients of <span class="kw">this</span> loss function are scaled by factor `<span class="num">1</span>/m`, where m is the number of regression outputs of a training example. |
| The parameter `grad_scale` can be used to change <span class="kw">this</span> scale to `grad_scale/m`. |
| |
| |
| |
| Defined in src/operator/regression_output.cc:L152</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#LogisticRegressionOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="LogisticRegressionOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="LogisticRegressionOutput(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">LogisticRegressionOutput</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@LogisticRegressionOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies a logistic function to the input. |
| |
| The logistic function, also known as the sigmoid function, is computed as |
| :math:`\frac{<span class="num">1</span>}{<span class="num">1</span>+exp(-\textbf{x})}`. |
| |
| Commonly, the sigmoid is used to squash the real-valued output of a linear model |
| :math:`wTx+b` into the [<span class="num">0</span>,<span class="num">1</span>] range so that it can be interpreted as a probability. |
| It is suitable <span class="kw">for</span> binary classification or probability prediction tasks. |
| |
| .. note:: |
| Use the LogisticRegressionOutput as the <span class="kw">final</span> output layer of a net. |
| |
| The storage <span class="kw">type</span> of ``label`` can be ``default`` or ``csr`` |
| |
| - LogisticRegressionOutput(default, default) = default |
| - LogisticRegressionOutput(default, csr) = default |
| |
| The loss function used is the Binary Cross Entropy Loss: |
| |
| :math:`-{(y\log(p) + (<span class="num">1</span> - y)\log(<span class="num">1</span> - p))}` |
| |
| Where `y` is the ground truth probability of positive outcome <span class="kw">for</span> a given example, and `p` the probability predicted by the model. By default, gradients of <span class="kw">this</span> loss function are scaled by factor `<span class="num">1</span>/m`, where m is the number of regression outputs of a training example. |
| The parameter `grad_scale` can be used to change <span class="kw">this</span> scale to `grad_scale/m`. |
| |
| |
| |
| Defined in src/operator/regression_output.cc:L152</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#MAERegressionOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="MAERegressionOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="MAERegressionOutput(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">MAERegressionOutput</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@MAERegressionOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes mean absolute error of the input. |
| |
| MAE is a risk metric corresponding to the expected value of the absolute error. |
| |
| If :math:`\hat{y}_i` is the predicted value of the i-th sample, and :math:`y_i` is the corresponding target value, |
| then the mean absolute error (MAE) estimated over :math:`n` samples is defined as |
| |
| :math:`\text{MAE}(\textbf{Y}, \hat{\textbf{Y}} ) = \frac{<span class="num">1</span>}{n} \sum_{i=<span class="num">0</span>}^{n-<span class="num">1</span>} \lVert \textbf{y}_i - \hat{\textbf{y}}_i \rVert_1` |
| |
| .. note:: |
| Use the MAERegressionOutput as the <span class="kw">final</span> output layer of a net. |
| |
| The storage <span class="kw">type</span> of ``label`` can be ``default`` or ``csr`` |
| |
| - MAERegressionOutput(default, default) = default |
| - MAERegressionOutput(default, csr) = default |
| |
| By default, gradients of <span class="kw">this</span> loss function are scaled by factor `<span class="num">1</span>/m`, where m is the number of regression outputs of a training example. |
| The parameter `grad_scale` can be used to change <span class="kw">this</span> scale to `grad_scale/m`. |
| |
| |
| |
| Defined in src/operator/regression_output.cc:L120</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#MAERegressionOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="MAERegressionOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="MAERegressionOutput(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">MAERegressionOutput</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@MAERegressionOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes mean absolute error of the input. |
| |
| MAE is a risk metric corresponding to the expected value of the absolute error. |
| |
| If :math:`\hat{y}_i` is the predicted value of the i-th sample, and :math:`y_i` is the corresponding target value, |
| then the mean absolute error (MAE) estimated over :math:`n` samples is defined as |
| |
| :math:`\text{MAE}(\textbf{Y}, \hat{\textbf{Y}} ) = \frac{<span class="num">1</span>}{n} \sum_{i=<span class="num">0</span>}^{n-<span class="num">1</span>} \lVert \textbf{y}_i - \hat{\textbf{y}}_i \rVert_1` |
| |
| .. note:: |
| Use the MAERegressionOutput as the <span class="kw">final</span> output layer of a net. |
| |
| The storage <span class="kw">type</span> of ``label`` can be ``default`` or ``csr`` |
| |
| - MAERegressionOutput(default, default) = default |
| - MAERegressionOutput(default, csr) = default |
| |
| By default, gradients of <span class="kw">this</span> loss function are scaled by factor `<span class="num">1</span>/m`, where m is the number of regression outputs of a training example. |
| The parameter `grad_scale` can be used to change <span class="kw">this</span> scale to `grad_scale/m`. |
| |
| |
| |
| Defined in src/operator/regression_output.cc:L120</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#MakeLoss" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="MakeLoss(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="MakeLoss(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">MakeLoss</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@MakeLoss(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Make your own loss function in network construction. |
| |
| This operator accepts a customized loss function symbol as a terminal loss and |
| the symbol should be an operator <span class="kw">with</span> no backward dependency. |
| The output of <span class="kw">this</span> function is the gradient of loss <span class="kw">with</span> respect to the input data. |
| |
| For example, <span class="kw">if</span> you are a making a cross entropy loss function. Assume ``out`` is the |
| predicted output and ``label`` is the <span class="kw">true</span> label, then the cross entropy can be defined as:: |
| |
| cross_entropy = label * log(out) + (<span class="num">1</span> - label) * log(<span class="num">1</span> - out) |
| loss = MakeLoss(cross_entropy) |
| |
| We will need to use ``MakeLoss`` when we are creating our own loss function or we want to |
| combine multiple loss functions. Also we may want to stop some variables' gradients |
| from backpropagation. See more detail in ``BlockGrad`` or ``stop_gradient``. |
| |
| In addition, we can give a scale to the loss by setting ``grad_scale``, |
| so that the gradient of the loss will be rescaled in the backpropagation. |
| |
| .. note:: This operator should be used as a <span class="std">Symbol</span> instead of NDArray. |
| |
| |
| |
| Defined in src/operator/make_loss.cc:L70</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#MakeLoss" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="MakeLoss(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="MakeLoss(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">MakeLoss</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@MakeLoss(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Make your own loss function in network construction. |
| |
| This operator accepts a customized loss function symbol as a terminal loss and |
| the symbol should be an operator <span class="kw">with</span> no backward dependency. |
| The output of <span class="kw">this</span> function is the gradient of loss <span class="kw">with</span> respect to the input data. |
| |
| For example, <span class="kw">if</span> you are a making a cross entropy loss function. Assume ``out`` is the |
| predicted output and ``label`` is the <span class="kw">true</span> label, then the cross entropy can be defined as:: |
| |
| cross_entropy = label * log(out) + (<span class="num">1</span> - label) * log(<span class="num">1</span> - out) |
| loss = MakeLoss(cross_entropy) |
| |
| We will need to use ``MakeLoss`` when we are creating our own loss function or we want to |
| combine multiple loss functions. Also we may want to stop some variables' gradients |
| from backpropagation. See more detail in ``BlockGrad`` or ``stop_gradient``. |
| |
| In addition, we can give a scale to the loss by setting ``grad_scale``, |
| so that the gradient of the loss will be rescaled in the backpropagation. |
| |
| .. note:: This operator should be used as a <span class="std">Symbol</span> instead of NDArray. |
| |
| |
| |
| Defined in src/operator/make_loss.cc:L70</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Pad" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Pad(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Pad(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Pad</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Pad(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Pads an input array <span class="kw">with</span> a constant or edge values of the array. |
| |
| .. note:: `Pad` is deprecated. Use `pad` instead. |
| |
| .. note:: Current implementation only supports <span class="num">4</span>D and <span class="num">5</span>D input arrays <span class="kw">with</span> padding applied |
| only on axes <span class="num">1</span>, <span class="num">2</span> and <span class="num">3.</span> Expects axes <span class="num">4</span> and <span class="num">5</span> in `pad_width` to be zero. |
| |
| This operation pads an input array <span class="kw">with</span> either a `constant_value` or edge values |
| along each axis of the input array. The amount of padding is specified by `pad_width`. |
| |
| `pad_width` is a tuple of integer padding widths <span class="kw">for</span> each axis of the format |
| ``(before_1, after_1, ... , before_N, after_N)``. The `pad_width` should be of length ``<span class="num">2</span>*N`` |
| where ``N`` is the number of dimensions of the array. |
| |
| For dimension ``N`` of the input array, ``before_N`` and ``after_N`` indicates how many values |
| to add before and after the elements of the array along dimension ``N``. |
| The widths of the higher two dimensions ``before_1``, ``after_1``, ``before_2``, |
| ``after_2`` must be <span class="num">0.</span> |
| |
| Example:: |
| |
| x = `[ [`[ [ <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span>] |
| [ <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span>] ] |
| |
| `[ [ <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span>] |
| [ <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span>] |
| [ <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span>] ] |
| |
| `[ [ <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span>] |
| [ <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span>] ] ] ] |
| |
| pad(x,mode=<span class="lit">"edge"</span>, pad_width=(<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>)) = |
| |
| `[ [`[ [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">3.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">3.</span>] |
| [ <span class="num">4.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">6.</span>] |
| [ <span class="num">4.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">6.</span>] ] |
| |
| `[ [ <span class="num">7.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">9.</span>] |
| [ <span class="num">7.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">9.</span>] |
| [ <span class="num">10.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">12.</span>] |
| [ <span class="num">10.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">12.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">11.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">13.</span>] |
| [ <span class="num">11.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">13.</span>] |
| [ <span class="num">14.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">16.</span>] |
| [ <span class="num">14.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">16.</span>] ] |
| |
| `[ [ <span class="num">17.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">19.</span>] |
| [ <span class="num">17.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">19.</span>] |
| [ <span class="num">20.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">22.</span>] |
| [ <span class="num">20.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">22.</span>] ] ] ] |
| |
| pad(x, mode=<span class="lit">"constant"</span>, constant_value=<span class="num">0</span>, pad_width=(<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>)) = |
| |
| `[ [`[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] ] |
| |
| |
| |
| |
| Defined in src/operator/pad.cc:L765</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Pad" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Pad(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Pad(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Pad</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Pad(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Pads an input array <span class="kw">with</span> a constant or edge values of the array. |
| |
| .. note:: `Pad` is deprecated. Use `pad` instead. |
| |
| .. note:: Current implementation only supports <span class="num">4</span>D and <span class="num">5</span>D input arrays <span class="kw">with</span> padding applied |
| only on axes <span class="num">1</span>, <span class="num">2</span> and <span class="num">3.</span> Expects axes <span class="num">4</span> and <span class="num">5</span> in `pad_width` to be zero. |
| |
| This operation pads an input array <span class="kw">with</span> either a `constant_value` or edge values |
| along each axis of the input array. The amount of padding is specified by `pad_width`. |
| |
| `pad_width` is a tuple of integer padding widths <span class="kw">for</span> each axis of the format |
| ``(before_1, after_1, ... , before_N, after_N)``. The `pad_width` should be of length ``<span class="num">2</span>*N`` |
| where ``N`` is the number of dimensions of the array. |
| |
| For dimension ``N`` of the input array, ``before_N`` and ``after_N`` indicates how many values |
| to add before and after the elements of the array along dimension ``N``. |
| The widths of the higher two dimensions ``before_1``, ``after_1``, ``before_2``, |
| ``after_2`` must be <span class="num">0.</span> |
| |
| Example:: |
| |
| x = `[ [`[ [ <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span>] |
| [ <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span>] ] |
| |
| `[ [ <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span>] |
| [ <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span>] |
| [ <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span>] ] |
| |
| `[ [ <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span>] |
| [ <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span>] ] ] ] |
| |
| pad(x,mode=<span class="lit">"edge"</span>, pad_width=(<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>)) = |
| |
| `[ [`[ [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">3.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">3.</span>] |
| [ <span class="num">4.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">6.</span>] |
| [ <span class="num">4.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">6.</span>] ] |
| |
| `[ [ <span class="num">7.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">9.</span>] |
| [ <span class="num">7.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">9.</span>] |
| [ <span class="num">10.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">12.</span>] |
| [ <span class="num">10.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">12.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">11.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">13.</span>] |
| [ <span class="num">11.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">13.</span>] |
| [ <span class="num">14.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">16.</span>] |
| [ <span class="num">14.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">16.</span>] ] |
| |
| `[ [ <span class="num">17.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">19.</span>] |
| [ <span class="num">17.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">19.</span>] |
| [ <span class="num">20.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">22.</span>] |
| [ <span class="num">20.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">22.</span>] ] ] ] |
| |
| pad(x, mode=<span class="lit">"constant"</span>, constant_value=<span class="num">0</span>, pad_width=(<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>)) = |
| |
| `[ [`[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] ] |
| |
| |
| |
| |
| Defined in src/operator/pad.cc:L765</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Pooling" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Pooling(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Pooling(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Pooling</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Pooling(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs pooling on the input. |
| |
| The shapes <span class="kw">for</span> <span class="num">1</span>-D pooling are |
| |
| - **data** and **out**: *(batch_size, channel, width)* (NCW layout) or |
| *(batch_size, width, channel)* (NWC layout), |
| |
| The shapes <span class="kw">for</span> <span class="num">2</span>-D pooling are |
| |
| - **data** and **out**: *(batch_size, channel, height, width)* (NCHW layout) or |
| *(batch_size, height, width, channel)* (NHWC layout), |
| |
| out_height = f(height, kernel[<span class="num">0</span>], pad[<span class="num">0</span>], stride[<span class="num">0</span>]) |
| out_width = f(width, kernel[<span class="num">1</span>], pad[<span class="num">1</span>], stride[<span class="num">1</span>]) |
| |
| The definition of *f* depends on ``pooling_convention``, which has two options: |
| |
| - **valid** (default):: |
| |
| f(x, k, p, s) = floor((x+<span class="num">2</span>*p-k)/s)+<span class="num">1</span> |
| |
| - **full**, which is compatible <span class="kw">with</span> Caffe:: |
| |
| f(x, k, p, s) = ceil((x+<span class="num">2</span>*p-k)/s)+<span class="num">1</span> |
| |
| When ``global_pool`` is set to be <span class="kw">true</span>, then global pooling is performed. It will reset |
| ``kernel=(height, width)`` and set the appropiate padding to <span class="num">0.</span> |
| |
| Three pooling options are supported by ``pool_type``: |
| |
| - **avg**: average pooling |
| - **max**: max pooling |
| - **sum**: sum pooling |
| - **lp**: Lp pooling |
| |
| For <span class="num">3</span>-D pooling, an additional *depth* dimension is added before |
| *height*. Namely the input data and output will have shape *(batch_size, channel, depth, |
| height, width)* (NCDHW layout) or *(batch_size, depth, height, width, channel)* (NDHWC layout). |
| |
| Notes on Lp pooling: |
| |
| Lp pooling was first introduced by <span class="kw">this</span> paper: https:<span class="cmt">//arxiv.org/pdf/1204.3968.pdf.</span> |
| L-<span class="num">1</span> pooling is simply sum pooling, <span class="kw">while</span> L-inf pooling is simply max pooling. |
| We can see that Lp pooling stands between those two, in practice the most common value <span class="kw">for</span> p is <span class="num">2.</span> |
| |
| For each window ``X``, the mathematical expression <span class="kw">for</span> Lp pooling is: |
| |
| :math:`f(X) = \sqrt[p]{\sum_{x}^{X} x^p}` |
| |
| |
| |
| Defined in src/operator/nn/pooling.cc:L416</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Pooling" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Pooling(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Pooling(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Pooling</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Pooling(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs pooling on the input. |
| |
| The shapes <span class="kw">for</span> <span class="num">1</span>-D pooling are |
| |
| - **data** and **out**: *(batch_size, channel, width)* (NCW layout) or |
| *(batch_size, width, channel)* (NWC layout), |
| |
| The shapes <span class="kw">for</span> <span class="num">2</span>-D pooling are |
| |
| - **data** and **out**: *(batch_size, channel, height, width)* (NCHW layout) or |
| *(batch_size, height, width, channel)* (NHWC layout), |
| |
| out_height = f(height, kernel[<span class="num">0</span>], pad[<span class="num">0</span>], stride[<span class="num">0</span>]) |
| out_width = f(width, kernel[<span class="num">1</span>], pad[<span class="num">1</span>], stride[<span class="num">1</span>]) |
| |
| The definition of *f* depends on ``pooling_convention``, which has two options: |
| |
| - **valid** (default):: |
| |
| f(x, k, p, s) = floor((x+<span class="num">2</span>*p-k)/s)+<span class="num">1</span> |
| |
| - **full**, which is compatible <span class="kw">with</span> Caffe:: |
| |
| f(x, k, p, s) = ceil((x+<span class="num">2</span>*p-k)/s)+<span class="num">1</span> |
| |
| When ``global_pool`` is set to be <span class="kw">true</span>, then global pooling is performed. It will reset |
| ``kernel=(height, width)`` and set the appropiate padding to <span class="num">0.</span> |
| |
| Three pooling options are supported by ``pool_type``: |
| |
| - **avg**: average pooling |
| - **max**: max pooling |
| - **sum**: sum pooling |
| - **lp**: Lp pooling |
| |
| For <span class="num">3</span>-D pooling, an additional *depth* dimension is added before |
| *height*. Namely the input data and output will have shape *(batch_size, channel, depth, |
| height, width)* (NCDHW layout) or *(batch_size, depth, height, width, channel)* (NDHWC layout). |
| |
| Notes on Lp pooling: |
| |
| Lp pooling was first introduced by <span class="kw">this</span> paper: https:<span class="cmt">//arxiv.org/pdf/1204.3968.pdf.</span> |
| L-<span class="num">1</span> pooling is simply sum pooling, <span class="kw">while</span> L-inf pooling is simply max pooling. |
| We can see that Lp pooling stands between those two, in practice the most common value <span class="kw">for</span> p is <span class="num">2.</span> |
| |
| For each window ``X``, the mathematical expression <span class="kw">for</span> Lp pooling is: |
| |
| :math:`f(X) = \sqrt[p]{\sum_{x}^{X} x^p}` |
| |
| |
| |
| Defined in src/operator/nn/pooling.cc:L416</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Pooling_v1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Pooling_v1(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Pooling_v1(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Pooling_v1</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Pooling_v1(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>This operator is DEPRECATED. |
| Perform pooling on the input. |
| |
| The shapes <span class="kw">for</span> <span class="num">2</span>-D pooling is |
| |
| - **data**: *(batch_size, channel, height, width)* |
| - **out**: *(batch_size, num_filter, out_height, out_width)*, <span class="kw">with</span>:: |
| |
| out_height = f(height, kernel[<span class="num">0</span>], pad[<span class="num">0</span>], stride[<span class="num">0</span>]) |
| out_width = f(width, kernel[<span class="num">1</span>], pad[<span class="num">1</span>], stride[<span class="num">1</span>]) |
| |
| The definition of *f* depends on ``pooling_convention``, which has two options: |
| |
| - **valid** (default):: |
| |
| f(x, k, p, s) = floor((x+<span class="num">2</span>*p-k)/s)+<span class="num">1</span> |
| |
| - **full**, which is compatible <span class="kw">with</span> Caffe:: |
| |
| f(x, k, p, s) = ceil((x+<span class="num">2</span>*p-k)/s)+<span class="num">1</span> |
| |
| But ``global_pool`` is set to be <span class="kw">true</span>, then <span class="kw">do</span> a global pooling, namely reset |
| ``kernel=(height, width)``. |
| |
| Three pooling options are supported by ``pool_type``: |
| |
| - **avg**: average pooling |
| - **max**: max pooling |
| - **sum**: sum pooling |
| |
| <span class="num">1</span>-D pooling is special <span class="kw">case</span> of <span class="num">2</span>-D pooling <span class="kw">with</span> *weight=<span class="num">1</span>* and |
| *kernel[<span class="num">1</span>]=<span class="num">1</span>*. |
| |
| For <span class="num">3</span>-D pooling, an additional *depth* dimension is added before |
| *height*. Namely the input data will have shape *(batch_size, channel, depth, |
| height, width)*. |
| |
| |
| |
| Defined in src/operator/pooling_v1.cc:L103</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Pooling_v1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Pooling_v1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Pooling_v1(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Pooling_v1</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Pooling_v1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>This operator is DEPRECATED. |
| Perform pooling on the input. |
| |
| The shapes <span class="kw">for</span> <span class="num">2</span>-D pooling is |
| |
| - **data**: *(batch_size, channel, height, width)* |
| - **out**: *(batch_size, num_filter, out_height, out_width)*, <span class="kw">with</span>:: |
| |
| out_height = f(height, kernel[<span class="num">0</span>], pad[<span class="num">0</span>], stride[<span class="num">0</span>]) |
| out_width = f(width, kernel[<span class="num">1</span>], pad[<span class="num">1</span>], stride[<span class="num">1</span>]) |
| |
| The definition of *f* depends on ``pooling_convention``, which has two options: |
| |
| - **valid** (default):: |
| |
| f(x, k, p, s) = floor((x+<span class="num">2</span>*p-k)/s)+<span class="num">1</span> |
| |
| - **full**, which is compatible <span class="kw">with</span> Caffe:: |
| |
| f(x, k, p, s) = ceil((x+<span class="num">2</span>*p-k)/s)+<span class="num">1</span> |
| |
| But ``global_pool`` is set to be <span class="kw">true</span>, then <span class="kw">do</span> a global pooling, namely reset |
| ``kernel=(height, width)``. |
| |
| Three pooling options are supported by ``pool_type``: |
| |
| - **avg**: average pooling |
| - **max**: max pooling |
| - **sum**: sum pooling |
| |
| <span class="num">1</span>-D pooling is special <span class="kw">case</span> of <span class="num">2</span>-D pooling <span class="kw">with</span> *weight=<span class="num">1</span>* and |
| *kernel[<span class="num">1</span>]=<span class="num">1</span>*. |
| |
| For <span class="num">3</span>-D pooling, an additional *depth* dimension is added before |
| *height*. Namely the input data will have shape *(batch_size, channel, depth, |
| height, width)*. |
| |
| |
| |
| Defined in src/operator/pooling_v1.cc:L103</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#RNN" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="RNN(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="RNN(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">RNN</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@RNN(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies recurrent layers to input data. Currently, vanilla RNN, LSTM and GRU are |
| implemented, <span class="kw">with</span> both multi-layer and bidirectional support. |
| |
| When the input data is of <span class="kw">type</span> float32 and the environment variables MXNET_CUDA_ALLOW_TENSOR_CORE |
| and MXNET_CUDA_TENSOR_OP_MATH_ALLOW_CONVERSION are set to <span class="num">1</span>, <span class="kw">this</span> operator will <span class="kw">try</span> to use |
| pseudo-float16 precision (float32 math <span class="kw">with</span> float16 I/O) precision in order to use |
| Tensor Cores on suitable NVIDIA GPUs. This can sometimes give significant speedups. |
| |
| **Vanilla RNN** |
| |
| Applies a single-gate recurrent layer to input X. Two kinds of activation function are supported: |
| ReLU and Tanh. |
| |
| With ReLU activation function: |
| |
| .. math:: |
| h_t = relu(W_{ih} * x_t + b_{ih} + W_{hh} * h_{(t-<span class="num">1</span>)} + b_{hh}) |
| |
| With Tanh activtion function: |
| |
| .. math:: |
| h_t = \tanh(W_{ih} * x_t + b_{ih} + W_{hh} * h_{(t-<span class="num">1</span>)} + b_{hh}) |
| |
| Reference paper: Finding structure in time - Elman, <span class="num">1988.</span> |
| https:<span class="cmt">//crl.ucsd.edu/~elman/Papers/fsit.pdf</span> |
| |
| **LSTM** |
| |
| <span class="std">Long</span> <span class="std">Short</span>-Term Memory - Hochreiter, <span class="num">1997.</span> http:<span class="cmt">//www.bioinf.jku.at/publications/older/2604.pdf</span> |
| |
| .. math:: |
| \begin{array}{ll} |
| i_t = \mathrm{sigmoid}(W_{ii} x_t + b_{ii} + W_{hi} h_{(t-<span class="num">1</span>)} + b_{hi}) \\ |
| f_t = \mathrm{sigmoid}(W_{<span class="kw">if</span>} x_t + b_{<span class="kw">if</span>} + W_{hf} h_{(t-<span class="num">1</span>)} + b_{hf}) \\ |
| g_t = \tanh(W_{ig} x_t + b_{ig} + W_{hc} h_{(t-<span class="num">1</span>)} + b_{hg}) \\ |
| o_t = \mathrm{sigmoid}(W_{io} x_t + b_{io} + W_{ho} h_{(t-<span class="num">1</span>)} + b_{ho}) \\ |
| c_t = f_t * c_{(t-<span class="num">1</span>)} + i_t * g_t \\ |
| h_t = o_t * \tanh(c_t) |
| \end{array} |
| |
| With the projection size being set, LSTM could use the projection feature to reduce the parameters |
| size and give some speedups without significant damage to the accuracy. |
| |
| <span class="std">Long</span> <span class="std">Short</span>-Term Memory Based Recurrent Neural Network Architectures <span class="kw">for</span> Large Vocabulary Speech |
| Recognition - Sak et al. <span class="num">2014.</span> https:<span class="cmt">//arxiv.org/abs/1402.1128</span> |
| |
| .. math:: |
| \begin{array}{ll} |
| i_t = \mathrm{sigmoid}(W_{ii} x_t + b_{ii} + W_{ri} r_{(t-<span class="num">1</span>)} + b_{ri}) \\ |
| f_t = \mathrm{sigmoid}(W_{<span class="kw">if</span>} x_t + b_{<span class="kw">if</span>} + W_{rf} r_{(t-<span class="num">1</span>)} + b_{rf}) \\ |
| g_t = \tanh(W_{ig} x_t + b_{ig} + W_{rc} r_{(t-<span class="num">1</span>)} + b_{rg}) \\ |
| o_t = \mathrm{sigmoid}(W_{io} x_t + b_{o} + W_{ro} r_{(t-<span class="num">1</span>)} + b_{ro}) \\ |
| c_t = f_t * c_{(t-<span class="num">1</span>)} + i_t * g_t \\ |
| h_t = o_t * \tanh(c_t) |
| r_t = W_{hr} h_t |
| \end{array} |
| |
| **GRU** |
| |
| Gated Recurrent <span class="std">Unit</span> - Cho et al. <span class="num">2014.</span> http:<span class="cmt">//arxiv.org/abs/1406.1078</span> |
| |
| The definition of GRU here is slightly different from paper but compatible <span class="kw">with</span> CUDNN. |
| |
| .. math:: |
| \begin{array}{ll} |
| r_t = \mathrm{sigmoid}(W_{ir} x_t + b_{ir} + W_{hr} h_{(t-<span class="num">1</span>)} + b_{hr}) \\ |
| z_t = \mathrm{sigmoid}(W_{iz} x_t + b_{iz} + W_{hz} h_{(t-<span class="num">1</span>)} + b_{hz}) \\ |
| n_t = \tanh(W_{in} x_t + b_{in} + r_t * (W_{hn} h_{(t-<span class="num">1</span>)}+ b_{hn})) \\ |
| h_t = (<span class="num">1</span> - z_t) * n_t + z_t * h_{(t-<span class="num">1</span>)} \\ |
| \end{array} |
| |
| |
| Defined in src/operator/rnn.cc:L375</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#RNN" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="RNN(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="RNN(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">RNN</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@RNN(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies recurrent layers to input data. Currently, vanilla RNN, LSTM and GRU are |
| implemented, <span class="kw">with</span> both multi-layer and bidirectional support. |
| |
| When the input data is of <span class="kw">type</span> float32 and the environment variables MXNET_CUDA_ALLOW_TENSOR_CORE |
| and MXNET_CUDA_TENSOR_OP_MATH_ALLOW_CONVERSION are set to <span class="num">1</span>, <span class="kw">this</span> operator will <span class="kw">try</span> to use |
| pseudo-float16 precision (float32 math <span class="kw">with</span> float16 I/O) precision in order to use |
| Tensor Cores on suitable NVIDIA GPUs. This can sometimes give significant speedups. |
| |
| **Vanilla RNN** |
| |
| Applies a single-gate recurrent layer to input X. Two kinds of activation function are supported: |
| ReLU and Tanh. |
| |
| With ReLU activation function: |
| |
| .. math:: |
| h_t = relu(W_{ih} * x_t + b_{ih} + W_{hh} * h_{(t-<span class="num">1</span>)} + b_{hh}) |
| |
| With Tanh activtion function: |
| |
| .. math:: |
| h_t = \tanh(W_{ih} * x_t + b_{ih} + W_{hh} * h_{(t-<span class="num">1</span>)} + b_{hh}) |
| |
| Reference paper: Finding structure in time - Elman, <span class="num">1988.</span> |
| https:<span class="cmt">//crl.ucsd.edu/~elman/Papers/fsit.pdf</span> |
| |
| **LSTM** |
| |
| <span class="std">Long</span> <span class="std">Short</span>-Term Memory - Hochreiter, <span class="num">1997.</span> http:<span class="cmt">//www.bioinf.jku.at/publications/older/2604.pdf</span> |
| |
| .. math:: |
| \begin{array}{ll} |
| i_t = \mathrm{sigmoid}(W_{ii} x_t + b_{ii} + W_{hi} h_{(t-<span class="num">1</span>)} + b_{hi}) \\ |
| f_t = \mathrm{sigmoid}(W_{<span class="kw">if</span>} x_t + b_{<span class="kw">if</span>} + W_{hf} h_{(t-<span class="num">1</span>)} + b_{hf}) \\ |
| g_t = \tanh(W_{ig} x_t + b_{ig} + W_{hc} h_{(t-<span class="num">1</span>)} + b_{hg}) \\ |
| o_t = \mathrm{sigmoid}(W_{io} x_t + b_{io} + W_{ho} h_{(t-<span class="num">1</span>)} + b_{ho}) \\ |
| c_t = f_t * c_{(t-<span class="num">1</span>)} + i_t * g_t \\ |
| h_t = o_t * \tanh(c_t) |
| \end{array} |
| |
| With the projection size being set, LSTM could use the projection feature to reduce the parameters |
| size and give some speedups without significant damage to the accuracy. |
| |
| <span class="std">Long</span> <span class="std">Short</span>-Term Memory Based Recurrent Neural Network Architectures <span class="kw">for</span> Large Vocabulary Speech |
| Recognition - Sak et al. <span class="num">2014.</span> https:<span class="cmt">//arxiv.org/abs/1402.1128</span> |
| |
| .. math:: |
| \begin{array}{ll} |
| i_t = \mathrm{sigmoid}(W_{ii} x_t + b_{ii} + W_{ri} r_{(t-<span class="num">1</span>)} + b_{ri}) \\ |
| f_t = \mathrm{sigmoid}(W_{<span class="kw">if</span>} x_t + b_{<span class="kw">if</span>} + W_{rf} r_{(t-<span class="num">1</span>)} + b_{rf}) \\ |
| g_t = \tanh(W_{ig} x_t + b_{ig} + W_{rc} r_{(t-<span class="num">1</span>)} + b_{rg}) \\ |
| o_t = \mathrm{sigmoid}(W_{io} x_t + b_{o} + W_{ro} r_{(t-<span class="num">1</span>)} + b_{ro}) \\ |
| c_t = f_t * c_{(t-<span class="num">1</span>)} + i_t * g_t \\ |
| h_t = o_t * \tanh(c_t) |
| r_t = W_{hr} h_t |
| \end{array} |
| |
| **GRU** |
| |
| Gated Recurrent <span class="std">Unit</span> - Cho et al. <span class="num">2014.</span> http:<span class="cmt">//arxiv.org/abs/1406.1078</span> |
| |
| The definition of GRU here is slightly different from paper but compatible <span class="kw">with</span> CUDNN. |
| |
| .. math:: |
| \begin{array}{ll} |
| r_t = \mathrm{sigmoid}(W_{ir} x_t + b_{ir} + W_{hr} h_{(t-<span class="num">1</span>)} + b_{hr}) \\ |
| z_t = \mathrm{sigmoid}(W_{iz} x_t + b_{iz} + W_{hz} h_{(t-<span class="num">1</span>)} + b_{hz}) \\ |
| n_t = \tanh(W_{in} x_t + b_{in} + r_t * (W_{hn} h_{(t-<span class="num">1</span>)}+ b_{hn})) \\ |
| h_t = (<span class="num">1</span> - z_t) * n_t + z_t * h_{(t-<span class="num">1</span>)} \\ |
| \end{array} |
| |
| |
| Defined in src/operator/rnn.cc:L375</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ROIPooling" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ROIPooling(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ROIPooling(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ROIPooling</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ROIPooling(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs region of interest(ROI) pooling on the input array. |
| |
| ROI pooling is a variant of a max pooling layer, in which the output size is fixed and |
| region of interest is a parameter. Its purpose is to perform max pooling on the inputs |
| of non-uniform sizes to obtain fixed-size feature maps. ROI pooling is a neural-net |
| layer mostly used in training a `Fast R-CNN` network <span class="kw">for</span> <span class="kw">object</span> detection. |
| |
| This operator takes a <span class="num">4</span>D feature map as an input array and region proposals as `rois`, |
| then it pools over sub-regions of input and produces a fixed-sized output array |
| regardless of the ROI size. |
| |
| To crop the feature map accordingly, you can resize the bounding box coordinates |
| by changing the parameters `rois` and `spatial_scale`. |
| |
| The cropped feature maps are pooled by standard max pooling operation to a fixed size output |
| indicated by a `pooled_size` parameter. batch_size will change to the number of region |
| bounding boxes after `ROIPooling`. |
| |
| The size of each region of interest doesn't have to be perfectly divisible by |
| the number of pooling sections(`pooled_size`). |
| |
| Example:: |
| |
| x = `[ [`[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>], |
| [ <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>], |
| [ <span class="num">12.</span>, <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>, <span class="num">16.</span>, <span class="num">17.</span>], |
| [ <span class="num">18.</span>, <span class="num">19.</span>, <span class="num">20.</span>, <span class="num">21.</span>, <span class="num">22.</span>, <span class="num">23.</span>], |
| [ <span class="num">24.</span>, <span class="num">25.</span>, <span class="num">26.</span>, <span class="num">27.</span>, <span class="num">28.</span>, <span class="num">29.</span>], |
| [ <span class="num">30.</span>, <span class="num">31.</span>, <span class="num">32.</span>, <span class="num">33.</span>, <span class="num">34.</span>, <span class="num">35.</span>], |
| [ <span class="num">36.</span>, <span class="num">37.</span>, <span class="num">38.</span>, <span class="num">39.</span>, <span class="num">40.</span>, <span class="num">41.</span>], |
| [ <span class="num">42.</span>, <span class="num">43.</span>, <span class="num">44.</span>, <span class="num">45.</span>, <span class="num">46.</span>, <span class="num">47.</span>] ] ] ] |
| |
| <span class="cmt">// region of interest i.e. bounding box coordinates.</span> |
| y = `[ [<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">4</span>,<span class="num">4</span>] ] |
| |
| <span class="cmt">// returns array of shape (2,2) according to the given roi with max pooling.</span> |
| ROIPooling(x, y, (<span class="num">2</span>,<span class="num">2</span>), <span class="num">1.0</span>) = `[ [`[ [ <span class="num">14.</span>, <span class="num">16.</span>], |
| [ <span class="num">26.</span>, <span class="num">28.</span>] ] ] ] |
| |
| <span class="cmt">// region of interest is changed due to the change in `spacial_scale` parameter.</span> |
| ROIPooling(x, y, (<span class="num">2</span>,<span class="num">2</span>), <span class="num">0.7</span>) = `[ [`[ [ <span class="num">7.</span>, <span class="num">9.</span>], |
| [ <span class="num">19.</span>, <span class="num">21.</span>] ] ] ] |
| |
| |
| |
| Defined in src/operator/roi_pooling.cc:L224</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ROIPooling" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ROIPooling(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ROIPooling(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ROIPooling</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ROIPooling(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs region of interest(ROI) pooling on the input array. |
| |
| ROI pooling is a variant of a max pooling layer, in which the output size is fixed and |
| region of interest is a parameter. Its purpose is to perform max pooling on the inputs |
| of non-uniform sizes to obtain fixed-size feature maps. ROI pooling is a neural-net |
| layer mostly used in training a `Fast R-CNN` network <span class="kw">for</span> <span class="kw">object</span> detection. |
| |
| This operator takes a <span class="num">4</span>D feature map as an input array and region proposals as `rois`, |
| then it pools over sub-regions of input and produces a fixed-sized output array |
| regardless of the ROI size. |
| |
| To crop the feature map accordingly, you can resize the bounding box coordinates |
| by changing the parameters `rois` and `spatial_scale`. |
| |
| The cropped feature maps are pooled by standard max pooling operation to a fixed size output |
| indicated by a `pooled_size` parameter. batch_size will change to the number of region |
| bounding boxes after `ROIPooling`. |
| |
| The size of each region of interest doesn't have to be perfectly divisible by |
| the number of pooling sections(`pooled_size`). |
| |
| Example:: |
| |
| x = `[ [`[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>], |
| [ <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>], |
| [ <span class="num">12.</span>, <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>, <span class="num">16.</span>, <span class="num">17.</span>], |
| [ <span class="num">18.</span>, <span class="num">19.</span>, <span class="num">20.</span>, <span class="num">21.</span>, <span class="num">22.</span>, <span class="num">23.</span>], |
| [ <span class="num">24.</span>, <span class="num">25.</span>, <span class="num">26.</span>, <span class="num">27.</span>, <span class="num">28.</span>, <span class="num">29.</span>], |
| [ <span class="num">30.</span>, <span class="num">31.</span>, <span class="num">32.</span>, <span class="num">33.</span>, <span class="num">34.</span>, <span class="num">35.</span>], |
| [ <span class="num">36.</span>, <span class="num">37.</span>, <span class="num">38.</span>, <span class="num">39.</span>, <span class="num">40.</span>, <span class="num">41.</span>], |
| [ <span class="num">42.</span>, <span class="num">43.</span>, <span class="num">44.</span>, <span class="num">45.</span>, <span class="num">46.</span>, <span class="num">47.</span>] ] ] ] |
| |
| <span class="cmt">// region of interest i.e. bounding box coordinates.</span> |
| y = `[ [<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">4</span>,<span class="num">4</span>] ] |
| |
| <span class="cmt">// returns array of shape (2,2) according to the given roi with max pooling.</span> |
| ROIPooling(x, y, (<span class="num">2</span>,<span class="num">2</span>), <span class="num">1.0</span>) = `[ [`[ [ <span class="num">14.</span>, <span class="num">16.</span>], |
| [ <span class="num">26.</span>, <span class="num">28.</span>] ] ] ] |
| |
| <span class="cmt">// region of interest is changed due to the change in `spacial_scale` parameter.</span> |
| ROIPooling(x, y, (<span class="num">2</span>,<span class="num">2</span>), <span class="num">0.7</span>) = `[ [`[ [ <span class="num">7.</span>, <span class="num">9.</span>], |
| [ <span class="num">19.</span>, <span class="num">21.</span>] ] ] ] |
| |
| |
| |
| Defined in src/operator/roi_pooling.cc:L224</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Reshape" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Reshape(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Reshape(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Reshape</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Reshape(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reshapes the input array. |
| .. note:: ``Reshape`` is deprecated, use ``reshape`` |
| Given an array and a shape, <span class="kw">this</span> function returns a copy of the array in the <span class="kw">new</span> shape. |
| The shape is a tuple of integers such as (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>). The size of the <span class="kw">new</span> shape should be same as the size of the input array. |
| Example:: |
| reshape([<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [<span class="num">1</span>,<span class="num">2</span>], [<span class="num">3</span>,<span class="num">4</span>] ] |
| <span class="std">Some</span> dimensions of the shape can take special values from the set {<span class="num">0</span>, -<span class="num">1</span>, -<span class="num">2</span>, -<span class="num">3</span>, -<span class="num">4</span>}. The significance of each is explained below: |
| - ``<span class="num">0</span>`` copy <span class="kw">this</span> dimension from the input to the output shape. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">4</span>,<span class="num">0</span>,<span class="num">2</span>), output shape = (<span class="num">4</span>,<span class="num">3</span>,<span class="num">2</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,<span class="num">0</span>,<span class="num">0</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - ``-<span class="num">1</span>`` infers the dimension of the output shape by using the remainder of the input dimensions |
| keeping the size of the <span class="kw">new</span> array same as that of the input array. |
| At most one dimension of shape can be -<span class="num">1.</span> |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">6</span>,<span class="num">1</span>,-<span class="num">1</span>), output shape = (<span class="num">6</span>,<span class="num">1</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">3</span>,-<span class="num">1</span>,<span class="num">8</span>), output shape = (<span class="num">3</span>,<span class="num">1</span>,<span class="num">8</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape=(-<span class="num">1</span>,), output shape = (<span class="num">24</span>,) |
| - ``-<span class="num">2</span>`` copy all/remainder of the input dimensions to the output shape. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">2</span>,), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,-<span class="num">2</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">2</span>,<span class="num">1</span>,<span class="num">1</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">1</span>,<span class="num">1</span>) |
| - ``-<span class="num">3</span>`` use the product of two consecutive dimensions of the input shape as the output dimension. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">3</span>,<span class="num">4</span>), output shape = (<span class="num">6</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>), shape = (-<span class="num">3</span>,-<span class="num">3</span>), output shape = (<span class="num">6</span>,<span class="num">20</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">0</span>,-<span class="num">3</span>), output shape = (<span class="num">2</span>,<span class="num">12</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">3</span>,-<span class="num">2</span>), output shape = (<span class="num">6</span>,<span class="num">4</span>) |
| - ``-<span class="num">4</span>`` split one dimension of the input into two dimensions passed subsequent to -<span class="num">4</span> in shape (can contain -<span class="num">1</span>). |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">4</span>,<span class="num">1</span>,<span class="num">2</span>,-<span class="num">2</span>), output shape =(<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,-<span class="num">4</span>,-<span class="num">1</span>,<span class="num">3</span>,-<span class="num">2</span>), output shape = (<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">4</span>) |
| If the argument `reverse` is set to <span class="num">1</span>, then the special values are inferred from right to left. |
| Example:: |
| - without reverse=<span class="num">1</span>, <span class="kw">for</span> input shape = (<span class="num">10</span>,<span class="num">5</span>,<span class="num">4</span>), shape = (-<span class="num">1</span>,<span class="num">0</span>), output shape would be (<span class="num">40</span>,<span class="num">5</span>) |
| - <span class="kw">with</span> reverse=<span class="num">1</span>, output shape will be (<span class="num">50</span>,<span class="num">4</span>). |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L174</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Reshape" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Reshape(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Reshape(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Reshape</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Reshape(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reshapes the input array. |
| .. note:: ``Reshape`` is deprecated, use ``reshape`` |
| Given an array and a shape, <span class="kw">this</span> function returns a copy of the array in the <span class="kw">new</span> shape. |
| The shape is a tuple of integers such as (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>). The size of the <span class="kw">new</span> shape should be same as the size of the input array. |
| Example:: |
| reshape([<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [<span class="num">1</span>,<span class="num">2</span>], [<span class="num">3</span>,<span class="num">4</span>] ] |
| <span class="std">Some</span> dimensions of the shape can take special values from the set {<span class="num">0</span>, -<span class="num">1</span>, -<span class="num">2</span>, -<span class="num">3</span>, -<span class="num">4</span>}. The significance of each is explained below: |
| - ``<span class="num">0</span>`` copy <span class="kw">this</span> dimension from the input to the output shape. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">4</span>,<span class="num">0</span>,<span class="num">2</span>), output shape = (<span class="num">4</span>,<span class="num">3</span>,<span class="num">2</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,<span class="num">0</span>,<span class="num">0</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - ``-<span class="num">1</span>`` infers the dimension of the output shape by using the remainder of the input dimensions |
| keeping the size of the <span class="kw">new</span> array same as that of the input array. |
| At most one dimension of shape can be -<span class="num">1.</span> |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">6</span>,<span class="num">1</span>,-<span class="num">1</span>), output shape = (<span class="num">6</span>,<span class="num">1</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">3</span>,-<span class="num">1</span>,<span class="num">8</span>), output shape = (<span class="num">3</span>,<span class="num">1</span>,<span class="num">8</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape=(-<span class="num">1</span>,), output shape = (<span class="num">24</span>,) |
| - ``-<span class="num">2</span>`` copy all/remainder of the input dimensions to the output shape. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">2</span>,), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,-<span class="num">2</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">2</span>,<span class="num">1</span>,<span class="num">1</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">1</span>,<span class="num">1</span>) |
| - ``-<span class="num">3</span>`` use the product of two consecutive dimensions of the input shape as the output dimension. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">3</span>,<span class="num">4</span>), output shape = (<span class="num">6</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>), shape = (-<span class="num">3</span>,-<span class="num">3</span>), output shape = (<span class="num">6</span>,<span class="num">20</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">0</span>,-<span class="num">3</span>), output shape = (<span class="num">2</span>,<span class="num">12</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">3</span>,-<span class="num">2</span>), output shape = (<span class="num">6</span>,<span class="num">4</span>) |
| - ``-<span class="num">4</span>`` split one dimension of the input into two dimensions passed subsequent to -<span class="num">4</span> in shape (can contain -<span class="num">1</span>). |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">4</span>,<span class="num">1</span>,<span class="num">2</span>,-<span class="num">2</span>), output shape =(<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,-<span class="num">4</span>,-<span class="num">1</span>,<span class="num">3</span>,-<span class="num">2</span>), output shape = (<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">4</span>) |
| If the argument `reverse` is set to <span class="num">1</span>, then the special values are inferred from right to left. |
| Example:: |
| - without reverse=<span class="num">1</span>, <span class="kw">for</span> input shape = (<span class="num">10</span>,<span class="num">5</span>,<span class="num">4</span>), shape = (-<span class="num">1</span>,<span class="num">0</span>), output shape would be (<span class="num">40</span>,<span class="num">5</span>) |
| - <span class="kw">with</span> reverse=<span class="num">1</span>, output shape will be (<span class="num">50</span>,<span class="num">4</span>). |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L174</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SVMOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SVMOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SVMOutput(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SVMOutput</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SVMOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes support vector machine based transformation of the input. |
| |
| This tutorial demonstrates using SVM as output layer <span class="kw">for</span> classification instead of softmax: |
| https:<span class="cmt">//github.com/apache/mxnet/tree/v1.x/example/svm_mnist.</span></pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SVMOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SVMOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SVMOutput(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SVMOutput</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SVMOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes support vector machine based transformation of the input. |
| |
| This tutorial demonstrates using SVM as output layer <span class="kw">for</span> classification instead of softmax: |
| https:<span class="cmt">//github.com/apache/mxnet/tree/v1.x/example/svm_mnist.</span></pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SequenceLast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SequenceLast(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SequenceLast(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SequenceLast</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SequenceLast(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Takes the last element of a sequence. |
| |
| This function takes an n-dimensional input array of the form |
| [max_sequence_length, batch_size, other_feature_dims] and returns a (n-<span class="num">1</span>)-dimensional array |
| of the form [batch_size, other_feature_dims]. |
| |
| Parameter `sequence_length` is used to handle variable-length sequences. `sequence_length` should be |
| an input array of positive ints of dimension [batch_size]. To use <span class="kw">this</span> parameter, |
| set `use_sequence_length` to `True`, otherwise each example in the batch is assumed |
| to have the max sequence length. |
| |
| .. note:: Alternatively, you can also use `take` operator. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ], |
| |
| `[ [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>], |
| [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ], |
| |
| `[ [ <span class="num">19.</span>, <span class="num">20.</span>, <span class="num">21.</span>], |
| [ <span class="num">22.</span>, <span class="num">23.</span>, <span class="num">24.</span>], |
| [ <span class="num">25.</span>, <span class="num">26.</span>, <span class="num">27.</span>] ] ] |
| |
| <span class="cmt">// returns last sequence when sequence_length parameter is not used</span> |
| SequenceLast(x) = `[ [ <span class="num">19.</span>, <span class="num">20.</span>, <span class="num">21.</span>], |
| [ <span class="num">22.</span>, <span class="num">23.</span>, <span class="num">24.</span>], |
| [ <span class="num">25.</span>, <span class="num">26.</span>, <span class="num">27.</span>] ] |
| |
| <span class="cmt">// sequence_length is used</span> |
| SequenceLast(x, sequence_length=[<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>], use_sequence_length=True) = |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| |
| <span class="cmt">// sequence_length is used</span> |
| SequenceLast(x, sequence_length=[<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], use_sequence_length=True) = |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">25.</span>, <span class="num">26.</span>, <span class="num">27.</span>] ] |
| |
| |
| |
| Defined in src/operator/sequence_last.cc:L105</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SequenceLast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SequenceLast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SequenceLast(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SequenceLast</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SequenceLast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Takes the last element of a sequence. |
| |
| This function takes an n-dimensional input array of the form |
| [max_sequence_length, batch_size, other_feature_dims] and returns a (n-<span class="num">1</span>)-dimensional array |
| of the form [batch_size, other_feature_dims]. |
| |
| Parameter `sequence_length` is used to handle variable-length sequences. `sequence_length` should be |
| an input array of positive ints of dimension [batch_size]. To use <span class="kw">this</span> parameter, |
| set `use_sequence_length` to `True`, otherwise each example in the batch is assumed |
| to have the max sequence length. |
| |
| .. note:: Alternatively, you can also use `take` operator. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ], |
| |
| `[ [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>], |
| [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ], |
| |
| `[ [ <span class="num">19.</span>, <span class="num">20.</span>, <span class="num">21.</span>], |
| [ <span class="num">22.</span>, <span class="num">23.</span>, <span class="num">24.</span>], |
| [ <span class="num">25.</span>, <span class="num">26.</span>, <span class="num">27.</span>] ] ] |
| |
| <span class="cmt">// returns last sequence when sequence_length parameter is not used</span> |
| SequenceLast(x) = `[ [ <span class="num">19.</span>, <span class="num">20.</span>, <span class="num">21.</span>], |
| [ <span class="num">22.</span>, <span class="num">23.</span>, <span class="num">24.</span>], |
| [ <span class="num">25.</span>, <span class="num">26.</span>, <span class="num">27.</span>] ] |
| |
| <span class="cmt">// sequence_length is used</span> |
| SequenceLast(x, sequence_length=[<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>], use_sequence_length=True) = |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| |
| <span class="cmt">// sequence_length is used</span> |
| SequenceLast(x, sequence_length=[<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], use_sequence_length=True) = |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">25.</span>, <span class="num">26.</span>, <span class="num">27.</span>] ] |
| |
| |
| |
| Defined in src/operator/sequence_last.cc:L105</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SequenceMask" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SequenceMask(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SequenceMask(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SequenceMask</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SequenceMask(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Sets all elements outside the sequence to a constant value. |
| |
| This function takes an n-dimensional input array of the form |
| [max_sequence_length, batch_size, other_feature_dims] and returns an array of the same shape. |
| |
| Parameter `sequence_length` is used to handle variable-length sequences. `sequence_length` |
| should be an input array of positive ints of dimension [batch_size]. |
| To use <span class="kw">this</span> parameter, set `use_sequence_length` to `True`, |
| otherwise each example in the batch is assumed to have the max sequence length and |
| <span class="kw">this</span> operator works as the `identity` operator. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| <span class="cmt">// Batch 1</span> |
| B1 = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>] ] |
| |
| <span class="cmt">// Batch 2</span> |
| B2 = `[ [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] |
| |
| <span class="cmt">// works as identity operator when sequence_length parameter is not used</span> |
| SequenceMask(x) = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| <span class="cmt">// sequence_length [1,1] means 1 of each batch will be kept</span> |
| <span class="cmt">// and other rows are masked with default mask value = 0</span> |
| SequenceMask(x, sequence_length=[<span class="num">1</span>,<span class="num">1</span>], use_sequence_length=True) = |
| `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ], |
| |
| `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] ] |
| |
| <span class="cmt">// sequence_length [2,3] means 2 of batch B1 and 3 of batch B2 will be kept</span> |
| <span class="cmt">// and other rows are masked with value = 1</span> |
| SequenceMask(x, sequence_length=[<span class="num">2</span>,<span class="num">3</span>], use_sequence_length=True, value=<span class="num">1</span>) = |
| `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| |
| |
| Defined in src/operator/sequence_mask.cc:L185</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SequenceMask" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SequenceMask(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SequenceMask(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SequenceMask</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SequenceMask(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Sets all elements outside the sequence to a constant value. |
| |
| This function takes an n-dimensional input array of the form |
| [max_sequence_length, batch_size, other_feature_dims] and returns an array of the same shape. |
| |
| Parameter `sequence_length` is used to handle variable-length sequences. `sequence_length` |
| should be an input array of positive ints of dimension [batch_size]. |
| To use <span class="kw">this</span> parameter, set `use_sequence_length` to `True`, |
| otherwise each example in the batch is assumed to have the max sequence length and |
| <span class="kw">this</span> operator works as the `identity` operator. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| <span class="cmt">// Batch 1</span> |
| B1 = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>] ] |
| |
| <span class="cmt">// Batch 2</span> |
| B2 = `[ [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] |
| |
| <span class="cmt">// works as identity operator when sequence_length parameter is not used</span> |
| SequenceMask(x) = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| <span class="cmt">// sequence_length [1,1] means 1 of each batch will be kept</span> |
| <span class="cmt">// and other rows are masked with default mask value = 0</span> |
| SequenceMask(x, sequence_length=[<span class="num">1</span>,<span class="num">1</span>], use_sequence_length=True) = |
| `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ], |
| |
| `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] ] |
| |
| <span class="cmt">// sequence_length [2,3] means 2 of batch B1 and 3 of batch B2 will be kept</span> |
| <span class="cmt">// and other rows are masked with value = 1</span> |
| SequenceMask(x, sequence_length=[<span class="num">2</span>,<span class="num">3</span>], use_sequence_length=True, value=<span class="num">1</span>) = |
| `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| |
| |
| Defined in src/operator/sequence_mask.cc:L185</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SequenceReverse" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SequenceReverse(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SequenceReverse(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SequenceReverse</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SequenceReverse(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reverses the elements of each sequence. |
| |
| This function takes an n-dimensional input array of the form [max_sequence_length, batch_size, other_feature_dims] |
| and returns an array of the same shape. |
| |
| Parameter `sequence_length` is used to handle variable-length sequences. |
| `sequence_length` should be an input array of positive ints of dimension [batch_size]. |
| To use <span class="kw">this</span> parameter, set `use_sequence_length` to `True`, |
| otherwise each example in the batch is assumed to have the max sequence length. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| <span class="cmt">// Batch 1</span> |
| B1 = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>] ] |
| |
| <span class="cmt">// Batch 2</span> |
| B2 = `[ [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] |
| |
| <span class="cmt">// returns reverse sequence when sequence_length parameter is not used</span> |
| SequenceReverse(x) = `[ `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ] ] |
| |
| <span class="cmt">// sequence_length [2,2] means 2 rows of</span> |
| <span class="cmt">// both batch B1 and B2 will be reversed.</span> |
| SequenceReverse(x, sequence_length=[<span class="num">2</span>,<span class="num">2</span>], use_sequence_length=True) = |
| `[ `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| <span class="cmt">// sequence_length [2,3] means 2 of batch B2 and 3 of batch B3</span> |
| <span class="cmt">// will be reversed.</span> |
| SequenceReverse(x, sequence_length=[<span class="num">2</span>,<span class="num">3</span>], use_sequence_length=True) = |
| `[ `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ], |
| |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14</span>, <span class="num">15.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ] ] |
| |
| |
| |
| Defined in src/operator/sequence_reverse.cc:L121</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SequenceReverse" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SequenceReverse(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SequenceReverse(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SequenceReverse</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SequenceReverse(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reverses the elements of each sequence. |
| |
| This function takes an n-dimensional input array of the form [max_sequence_length, batch_size, other_feature_dims] |
| and returns an array of the same shape. |
| |
| Parameter `sequence_length` is used to handle variable-length sequences. |
| `sequence_length` should be an input array of positive ints of dimension [batch_size]. |
| To use <span class="kw">this</span> parameter, set `use_sequence_length` to `True`, |
| otherwise each example in the batch is assumed to have the max sequence length. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| <span class="cmt">// Batch 1</span> |
| B1 = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>] ] |
| |
| <span class="cmt">// Batch 2</span> |
| B2 = `[ [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] |
| |
| <span class="cmt">// returns reverse sequence when sequence_length parameter is not used</span> |
| SequenceReverse(x) = `[ `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ], |
| |
| `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ] ] |
| |
| <span class="cmt">// sequence_length [2,2] means 2 rows of</span> |
| <span class="cmt">// both batch B1 and B2 will be reversed.</span> |
| SequenceReverse(x, sequence_length=[<span class="num">2</span>,<span class="num">2</span>], use_sequence_length=True) = |
| `[ `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14.</span>, <span class="num">15.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ] ] |
| |
| <span class="cmt">// sequence_length [2,3] means 2 of batch B2 and 3 of batch B3</span> |
| <span class="cmt">// will be reversed.</span> |
| SequenceReverse(x, sequence_length=[<span class="num">2</span>,<span class="num">3</span>], use_sequence_length=True) = |
| `[ `[ [ <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">16.</span>, <span class="num">17.</span>, <span class="num">18.</span>] ], |
| |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ], |
| |
| `[ [ <span class="num">13.</span>, <span class="num">14</span>, <span class="num">15.</span>], |
| [ <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ] ] |
| |
| |
| |
| Defined in src/operator/sequence_reverse.cc:L121</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SliceChannel" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SliceChannel(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SliceChannel(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SliceChannel</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SliceChannel(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Splits an array along a particular axis into multiple sub-arrays. |
| |
| .. note:: ``SliceChannel`` is deprecated. Use ``split`` instead. |
| |
| **Note** that `num_outputs` should evenly divide the length of the axis |
| along which to split the array. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] |
| `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] |
| `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] ] |
| x.shape = (<span class="num">3</span>, <span class="num">2</span>, <span class="num">1</span>) |
| |
| y = split(x, axis=<span class="num">1</span>, num_outputs=<span class="num">2</span>) <span class="cmt">// a list of 2 arrays with shape (3, 1, 1)</span> |
| y = `[ `[ [ <span class="num">1.</span>] ] |
| `[ [ <span class="num">3.</span>] ] |
| `[ [ <span class="num">5.</span>] ] ] |
| |
| `[ `[ [ <span class="num">2.</span>] ] |
| `[ [ <span class="num">4.</span>] ] |
| `[ [ <span class="num">6.</span>] ] ] |
| |
| y[<span class="num">0</span>].shape = (<span class="num">3</span>, <span class="num">1</span>, <span class="num">1</span>) |
| |
| z = split(x, axis=<span class="num">0</span>, num_outputs=<span class="num">3</span>) <span class="cmt">// a list of 3 arrays with shape (1, 2, 1)</span> |
| z = `[ `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] ] |
| |
| `[ `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] ] |
| |
| `[ `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] ] |
| |
| z[<span class="num">0</span>].shape = (<span class="num">1</span>, <span class="num">2</span>, <span class="num">1</span>) |
| |
| `squeeze_axis=<span class="num">1</span>` removes the axis <span class="kw">with</span> length <span class="num">1</span> from the shapes of the output arrays. |
| **Note** that setting `squeeze_axis` to ``<span class="num">1</span>`` removes axis <span class="kw">with</span> length <span class="num">1</span> only |
| along the `axis` which it is split. |
| Also `squeeze_axis` can be set to <span class="kw">true</span> only <span class="kw">if</span> ``input.shape[axis] == num_outputs``. |
| |
| Example:: |
| |
| z = split(x, axis=<span class="num">0</span>, num_outputs=<span class="num">3</span>, squeeze_axis=<span class="num">1</span>) <span class="cmt">// a list of 3 arrays with shape (2, 1)</span> |
| z = `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] |
| |
| `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] |
| |
| `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] |
| z[<span class="num">0</span>].shape = (<span class="num">2</span> ,<span class="num">1</span> ) |
| |
| |
| |
| Defined in src/operator/slice_channel.cc:L106</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SliceChannel" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SliceChannel(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SliceChannel(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SliceChannel</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SliceChannel(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Splits an array along a particular axis into multiple sub-arrays. |
| |
| .. note:: ``SliceChannel`` is deprecated. Use ``split`` instead. |
| |
| **Note** that `num_outputs` should evenly divide the length of the axis |
| along which to split the array. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] |
| `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] |
| `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] ] |
| x.shape = (<span class="num">3</span>, <span class="num">2</span>, <span class="num">1</span>) |
| |
| y = split(x, axis=<span class="num">1</span>, num_outputs=<span class="num">2</span>) <span class="cmt">// a list of 2 arrays with shape (3, 1, 1)</span> |
| y = `[ `[ [ <span class="num">1.</span>] ] |
| `[ [ <span class="num">3.</span>] ] |
| `[ [ <span class="num">5.</span>] ] ] |
| |
| `[ `[ [ <span class="num">2.</span>] ] |
| `[ [ <span class="num">4.</span>] ] |
| `[ [ <span class="num">6.</span>] ] ] |
| |
| y[<span class="num">0</span>].shape = (<span class="num">3</span>, <span class="num">1</span>, <span class="num">1</span>) |
| |
| z = split(x, axis=<span class="num">0</span>, num_outputs=<span class="num">3</span>) <span class="cmt">// a list of 3 arrays with shape (1, 2, 1)</span> |
| z = `[ `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] ] |
| |
| `[ `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] ] |
| |
| `[ `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] ] |
| |
| z[<span class="num">0</span>].shape = (<span class="num">1</span>, <span class="num">2</span>, <span class="num">1</span>) |
| |
| `squeeze_axis=<span class="num">1</span>` removes the axis <span class="kw">with</span> length <span class="num">1</span> from the shapes of the output arrays. |
| **Note** that setting `squeeze_axis` to ``<span class="num">1</span>`` removes axis <span class="kw">with</span> length <span class="num">1</span> only |
| along the `axis` which it is split. |
| Also `squeeze_axis` can be set to <span class="kw">true</span> only <span class="kw">if</span> ``input.shape[axis] == num_outputs``. |
| |
| Example:: |
| |
| z = split(x, axis=<span class="num">0</span>, num_outputs=<span class="num">3</span>, squeeze_axis=<span class="num">1</span>) <span class="cmt">// a list of 3 arrays with shape (2, 1)</span> |
| z = `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] |
| |
| `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] |
| |
| `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] |
| z[<span class="num">0</span>].shape = (<span class="num">2</span> ,<span class="num">1</span> ) |
| |
| |
| |
| Defined in src/operator/slice_channel.cc:L106</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Softmax" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Softmax(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Softmax(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Softmax</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Softmax(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the gradient of cross entropy loss <span class="kw">with</span> respect to softmax output. |
| |
| - This operator computes the gradient in two steps. |
| The cross entropy loss does not actually need to be computed. |
| |
| - Applies softmax function on the input array. |
| - Computes and returns the gradient of cross entropy loss w.r.t. the softmax output. |
| |
| - The softmax function, cross entropy loss and gradient is given by: |
| |
| - Softmax <span class="std">Function</span>: |
| |
| .. math:: \text{softmax}(x)_i = \frac{exp(x_i)}{\sum_j exp(x_j)} |
| |
| - Cross Entropy <span class="std">Function</span>: |
| |
| .. math:: \text{CE(label, output)} = - \sum_i \text{label}_i \log(\text{output}_i) |
| |
| - The gradient of cross entropy loss w.r.t softmax output: |
| |
| .. math:: \text{gradient} = \text{output} - \text{label} |
| |
| - During forward propagation, the softmax function is computed <span class="kw">for</span> each instance in the input array. |
| |
| For general *N*-D input arrays <span class="kw">with</span> shape :math:`(d_1, d_2, ..., d_n)`. The size is |
| :math:`s=d_1 \cdot d_2 \cdot \cdot \cdot d_n`. We can use the parameters `preserve_shape` |
| and `multi_output` to specify the way to compute softmax: |
| |
| - By default, `preserve_shape` is ``<span class="kw">false</span>``. This operator will reshape the input array |
| into a <span class="num">2</span>-D array <span class="kw">with</span> shape :math:`(d_1, \frac{s}{d_1})` and then compute the softmax function <span class="kw">for</span> |
| each row in the reshaped array, and afterwards reshape it back to the original shape |
| :math:`(d_1, d_2, ..., d_n)`. |
| - If `preserve_shape` is ``<span class="kw">true</span>``, the softmax function will be computed along |
| the last axis (`axis` = ``-<span class="num">1</span>``). |
| - If `multi_output` is ``<span class="kw">true</span>``, the softmax function will be computed along |
| the second axis (`axis` = ``<span class="num">1</span>``). |
| |
| - During backward propagation, the gradient of cross-entropy loss w.r.t softmax output array is computed. |
| The provided label can be a one-hot label array or a probability label array. |
| |
| - If the parameter `use_ignore` is ``<span class="kw">true</span>``, `ignore_label` can specify input instances |
| <span class="kw">with</span> a particular label to be ignored during backward propagation. **This has no effect when |
| softmax `output` has same shape as `label`**. |
| |
| Example:: |
| |
| data = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>],[<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>],[<span class="num">3</span>,<span class="num">3</span>,<span class="num">3</span>,<span class="num">3</span>],[<span class="num">4</span>,<span class="num">4</span>,<span class="num">4</span>,<span class="num">4</span>] ] |
| label = [<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>,<span class="num">3</span>] |
| ignore_label = <span class="num">1</span> |
| SoftmaxOutput(data=data, label = label,\ |
| multi_output=<span class="kw">true</span>, use_ignore=<span class="kw">true</span>,\ |
| ignore_label=ignore_label) |
| ## forward softmax output |
| `[ [ <span class="num">0.0320586</span> <span class="num">0.08714432</span> <span class="num">0.23688284</span> <span class="num">0.64391428</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] ] |
| ## backward gradient output |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> ] |
| [-<span class="num">0.75</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> -<span class="num">0.75</span> <span class="num">0.25</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> -<span class="num">0.75</span>] ] |
| ## notice that the first row is all <span class="num">0</span> because label[<span class="num">0</span>] is <span class="num">1</span>, which is equal to ignore_label. |
| |
| - The parameter `grad_scale` can be used to rescale the gradient, which is often used to |
| give each loss function different weights. |
| |
| - This operator also supports various ways to normalize the gradient by `normalization`, |
| The `normalization` is applied <span class="kw">if</span> softmax output has different shape than the labels. |
| The `normalization` mode can be set to the followings: |
| |
| - ``<span class="lit">'null'</span>``: <span class="kw">do</span> nothing. |
| - ``<span class="lit">'batch'</span>``: divide the gradient by the batch size. |
| - ``<span class="lit">'valid'</span>``: divide the gradient by the number of instances which are not ignored. |
| |
| |
| |
| Defined in src/operator/softmax_output.cc:L242</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#Softmax" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="Softmax(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="Softmax(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">Softmax</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@Softmax(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the gradient of cross entropy loss <span class="kw">with</span> respect to softmax output. |
| |
| - This operator computes the gradient in two steps. |
| The cross entropy loss does not actually need to be computed. |
| |
| - Applies softmax function on the input array. |
| - Computes and returns the gradient of cross entropy loss w.r.t. the softmax output. |
| |
| - The softmax function, cross entropy loss and gradient is given by: |
| |
| - Softmax <span class="std">Function</span>: |
| |
| .. math:: \text{softmax}(x)_i = \frac{exp(x_i)}{\sum_j exp(x_j)} |
| |
| - Cross Entropy <span class="std">Function</span>: |
| |
| .. math:: \text{CE(label, output)} = - \sum_i \text{label}_i \log(\text{output}_i) |
| |
| - The gradient of cross entropy loss w.r.t softmax output: |
| |
| .. math:: \text{gradient} = \text{output} - \text{label} |
| |
| - During forward propagation, the softmax function is computed <span class="kw">for</span> each instance in the input array. |
| |
| For general *N*-D input arrays <span class="kw">with</span> shape :math:`(d_1, d_2, ..., d_n)`. The size is |
| :math:`s=d_1 \cdot d_2 \cdot \cdot \cdot d_n`. We can use the parameters `preserve_shape` |
| and `multi_output` to specify the way to compute softmax: |
| |
| - By default, `preserve_shape` is ``<span class="kw">false</span>``. This operator will reshape the input array |
| into a <span class="num">2</span>-D array <span class="kw">with</span> shape :math:`(d_1, \frac{s}{d_1})` and then compute the softmax function <span class="kw">for</span> |
| each row in the reshaped array, and afterwards reshape it back to the original shape |
| :math:`(d_1, d_2, ..., d_n)`. |
| - If `preserve_shape` is ``<span class="kw">true</span>``, the softmax function will be computed along |
| the last axis (`axis` = ``-<span class="num">1</span>``). |
| - If `multi_output` is ``<span class="kw">true</span>``, the softmax function will be computed along |
| the second axis (`axis` = ``<span class="num">1</span>``). |
| |
| - During backward propagation, the gradient of cross-entropy loss w.r.t softmax output array is computed. |
| The provided label can be a one-hot label array or a probability label array. |
| |
| - If the parameter `use_ignore` is ``<span class="kw">true</span>``, `ignore_label` can specify input instances |
| <span class="kw">with</span> a particular label to be ignored during backward propagation. **This has no effect when |
| softmax `output` has same shape as `label`**. |
| |
| Example:: |
| |
| data = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>],[<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>],[<span class="num">3</span>,<span class="num">3</span>,<span class="num">3</span>,<span class="num">3</span>],[<span class="num">4</span>,<span class="num">4</span>,<span class="num">4</span>,<span class="num">4</span>] ] |
| label = [<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>,<span class="num">3</span>] |
| ignore_label = <span class="num">1</span> |
| SoftmaxOutput(data=data, label = label,\ |
| multi_output=<span class="kw">true</span>, use_ignore=<span class="kw">true</span>,\ |
| ignore_label=ignore_label) |
| ## forward softmax output |
| `[ [ <span class="num">0.0320586</span> <span class="num">0.08714432</span> <span class="num">0.23688284</span> <span class="num">0.64391428</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] ] |
| ## backward gradient output |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> ] |
| [-<span class="num">0.75</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> -<span class="num">0.75</span> <span class="num">0.25</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> -<span class="num">0.75</span>] ] |
| ## notice that the first row is all <span class="num">0</span> because label[<span class="num">0</span>] is <span class="num">1</span>, which is equal to ignore_label. |
| |
| - The parameter `grad_scale` can be used to rescale the gradient, which is often used to |
| give each loss function different weights. |
| |
| - This operator also supports various ways to normalize the gradient by `normalization`, |
| The `normalization` is applied <span class="kw">if</span> softmax output has different shape than the labels. |
| The `normalization` mode can be set to the followings: |
| |
| - ``<span class="lit">'null'</span>``: <span class="kw">do</span> nothing. |
| - ``<span class="lit">'batch'</span>``: divide the gradient by the batch size. |
| - ``<span class="lit">'valid'</span>``: divide the gradient by the number of instances which are not ignored. |
| |
| |
| |
| Defined in src/operator/softmax_output.cc:L242</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SoftmaxActivation" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SoftmaxActivation(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SoftmaxActivation(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SoftmaxActivation</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SoftmaxActivation(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies softmax activation to input. This is intended <span class="kw">for</span> internal layers. |
| |
| .. note:: |
| |
| This operator has been deprecated, please use `softmax`. |
| |
| If `mode` = ``instance``, <span class="kw">this</span> operator will compute a softmax <span class="kw">for</span> each instance in the batch. |
| This is the default mode. |
| |
| If `mode` = ``channel``, <span class="kw">this</span> operator will compute a k-<span class="kw">class</span> softmax at each position |
| of each instance, where `k` = ``num_channel``. This mode can only be used when the input array |
| has at least <span class="num">3</span> dimensions. |
| This can be used <span class="kw">for</span> `fully convolutional network`, `image segmentation`, etc. |
| |
| Example:: |
| |
| >>> input_array = mx.nd.array(`[ [<span class="num">3.</span>, <span class="num">0.5</span>, -<span class="num">0.5</span>, <span class="num">2.</span>, <span class="num">7.</span>], |
| >>> [<span class="num">2.</span>, -<span class="num">.4</span>, <span class="num">7.</span>, <span class="num">3.</span>, <span class="num">0.2</span>] ]) |
| >>> softmax_act = mx.nd.SoftmaxActivation(input_array) |
| >>> print softmax_act.asnumpy() |
| `[ [ <span class="num">1.78322066e-02</span> <span class="num">1.46375655e-03</span> <span class="num">5.38485940e-04</span> <span class="num">6.56010211e-03</span> <span class="num">9.73605454e-01</span>] |
| [ <span class="num">6.56221947e-03</span> <span class="num">5.95310994e-04</span> <span class="num">9.73919690e-01</span> <span class="num">1.78379621e-02</span> <span class="num">1.08472735e-03</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/softmax_activation.cc:L58</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SoftmaxActivation" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SoftmaxActivation(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SoftmaxActivation(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SoftmaxActivation</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SoftmaxActivation(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies softmax activation to input. This is intended <span class="kw">for</span> internal layers. |
| |
| .. note:: |
| |
| This operator has been deprecated, please use `softmax`. |
| |
| If `mode` = ``instance``, <span class="kw">this</span> operator will compute a softmax <span class="kw">for</span> each instance in the batch. |
| This is the default mode. |
| |
| If `mode` = ``channel``, <span class="kw">this</span> operator will compute a k-<span class="kw">class</span> softmax at each position |
| of each instance, where `k` = ``num_channel``. This mode can only be used when the input array |
| has at least <span class="num">3</span> dimensions. |
| This can be used <span class="kw">for</span> `fully convolutional network`, `image segmentation`, etc. |
| |
| Example:: |
| |
| >>> input_array = mx.nd.array(`[ [<span class="num">3.</span>, <span class="num">0.5</span>, -<span class="num">0.5</span>, <span class="num">2.</span>, <span class="num">7.</span>], |
| >>> [<span class="num">2.</span>, -<span class="num">.4</span>, <span class="num">7.</span>, <span class="num">3.</span>, <span class="num">0.2</span>] ]) |
| >>> softmax_act = mx.nd.SoftmaxActivation(input_array) |
| >>> print softmax_act.asnumpy() |
| `[ [ <span class="num">1.78322066e-02</span> <span class="num">1.46375655e-03</span> <span class="num">5.38485940e-04</span> <span class="num">6.56010211e-03</span> <span class="num">9.73605454e-01</span>] |
| [ <span class="num">6.56221947e-03</span> <span class="num">5.95310994e-04</span> <span class="num">9.73919690e-01</span> <span class="num">1.78379621e-02</span> <span class="num">1.08472735e-03</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/softmax_activation.cc:L58</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SoftmaxOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SoftmaxOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SoftmaxOutput(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SoftmaxOutput</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SoftmaxOutput(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the gradient of cross entropy loss <span class="kw">with</span> respect to softmax output. |
| |
| - This operator computes the gradient in two steps. |
| The cross entropy loss does not actually need to be computed. |
| |
| - Applies softmax function on the input array. |
| - Computes and returns the gradient of cross entropy loss w.r.t. the softmax output. |
| |
| - The softmax function, cross entropy loss and gradient is given by: |
| |
| - Softmax <span class="std">Function</span>: |
| |
| .. math:: \text{softmax}(x)_i = \frac{exp(x_i)}{\sum_j exp(x_j)} |
| |
| - Cross Entropy <span class="std">Function</span>: |
| |
| .. math:: \text{CE(label, output)} = - \sum_i \text{label}_i \log(\text{output}_i) |
| |
| - The gradient of cross entropy loss w.r.t softmax output: |
| |
| .. math:: \text{gradient} = \text{output} - \text{label} |
| |
| - During forward propagation, the softmax function is computed <span class="kw">for</span> each instance in the input array. |
| |
| For general *N*-D input arrays <span class="kw">with</span> shape :math:`(d_1, d_2, ..., d_n)`. The size is |
| :math:`s=d_1 \cdot d_2 \cdot \cdot \cdot d_n`. We can use the parameters `preserve_shape` |
| and `multi_output` to specify the way to compute softmax: |
| |
| - By default, `preserve_shape` is ``<span class="kw">false</span>``. This operator will reshape the input array |
| into a <span class="num">2</span>-D array <span class="kw">with</span> shape :math:`(d_1, \frac{s}{d_1})` and then compute the softmax function <span class="kw">for</span> |
| each row in the reshaped array, and afterwards reshape it back to the original shape |
| :math:`(d_1, d_2, ..., d_n)`. |
| - If `preserve_shape` is ``<span class="kw">true</span>``, the softmax function will be computed along |
| the last axis (`axis` = ``-<span class="num">1</span>``). |
| - If `multi_output` is ``<span class="kw">true</span>``, the softmax function will be computed along |
| the second axis (`axis` = ``<span class="num">1</span>``). |
| |
| - During backward propagation, the gradient of cross-entropy loss w.r.t softmax output array is computed. |
| The provided label can be a one-hot label array or a probability label array. |
| |
| - If the parameter `use_ignore` is ``<span class="kw">true</span>``, `ignore_label` can specify input instances |
| <span class="kw">with</span> a particular label to be ignored during backward propagation. **This has no effect when |
| softmax `output` has same shape as `label`**. |
| |
| Example:: |
| |
| data = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>],[<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>],[<span class="num">3</span>,<span class="num">3</span>,<span class="num">3</span>,<span class="num">3</span>],[<span class="num">4</span>,<span class="num">4</span>,<span class="num">4</span>,<span class="num">4</span>] ] |
| label = [<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>,<span class="num">3</span>] |
| ignore_label = <span class="num">1</span> |
| SoftmaxOutput(data=data, label = label,\ |
| multi_output=<span class="kw">true</span>, use_ignore=<span class="kw">true</span>,\ |
| ignore_label=ignore_label) |
| ## forward softmax output |
| `[ [ <span class="num">0.0320586</span> <span class="num">0.08714432</span> <span class="num">0.23688284</span> <span class="num">0.64391428</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] ] |
| ## backward gradient output |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> ] |
| [-<span class="num">0.75</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> -<span class="num">0.75</span> <span class="num">0.25</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> -<span class="num">0.75</span>] ] |
| ## notice that the first row is all <span class="num">0</span> because label[<span class="num">0</span>] is <span class="num">1</span>, which is equal to ignore_label. |
| |
| - The parameter `grad_scale` can be used to rescale the gradient, which is often used to |
| give each loss function different weights. |
| |
| - This operator also supports various ways to normalize the gradient by `normalization`, |
| The `normalization` is applied <span class="kw">if</span> softmax output has different shape than the labels. |
| The `normalization` mode can be set to the followings: |
| |
| - ``<span class="lit">'null'</span>``: <span class="kw">do</span> nothing. |
| - ``<span class="lit">'batch'</span>``: divide the gradient by the batch size. |
| - ``<span class="lit">'valid'</span>``: divide the gradient by the number of instances which are not ignored. |
| |
| |
| |
| Defined in src/operator/softmax_output.cc:L242</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SoftmaxOutput" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SoftmaxOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SoftmaxOutput(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SoftmaxOutput</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SoftmaxOutput(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the gradient of cross entropy loss <span class="kw">with</span> respect to softmax output. |
| |
| - This operator computes the gradient in two steps. |
| The cross entropy loss does not actually need to be computed. |
| |
| - Applies softmax function on the input array. |
| - Computes and returns the gradient of cross entropy loss w.r.t. the softmax output. |
| |
| - The softmax function, cross entropy loss and gradient is given by: |
| |
| - Softmax <span class="std">Function</span>: |
| |
| .. math:: \text{softmax}(x)_i = \frac{exp(x_i)}{\sum_j exp(x_j)} |
| |
| - Cross Entropy <span class="std">Function</span>: |
| |
| .. math:: \text{CE(label, output)} = - \sum_i \text{label}_i \log(\text{output}_i) |
| |
| - The gradient of cross entropy loss w.r.t softmax output: |
| |
| .. math:: \text{gradient} = \text{output} - \text{label} |
| |
| - During forward propagation, the softmax function is computed <span class="kw">for</span> each instance in the input array. |
| |
| For general *N*-D input arrays <span class="kw">with</span> shape :math:`(d_1, d_2, ..., d_n)`. The size is |
| :math:`s=d_1 \cdot d_2 \cdot \cdot \cdot d_n`. We can use the parameters `preserve_shape` |
| and `multi_output` to specify the way to compute softmax: |
| |
| - By default, `preserve_shape` is ``<span class="kw">false</span>``. This operator will reshape the input array |
| into a <span class="num">2</span>-D array <span class="kw">with</span> shape :math:`(d_1, \frac{s}{d_1})` and then compute the softmax function <span class="kw">for</span> |
| each row in the reshaped array, and afterwards reshape it back to the original shape |
| :math:`(d_1, d_2, ..., d_n)`. |
| - If `preserve_shape` is ``<span class="kw">true</span>``, the softmax function will be computed along |
| the last axis (`axis` = ``-<span class="num">1</span>``). |
| - If `multi_output` is ``<span class="kw">true</span>``, the softmax function will be computed along |
| the second axis (`axis` = ``<span class="num">1</span>``). |
| |
| - During backward propagation, the gradient of cross-entropy loss w.r.t softmax output array is computed. |
| The provided label can be a one-hot label array or a probability label array. |
| |
| - If the parameter `use_ignore` is ``<span class="kw">true</span>``, `ignore_label` can specify input instances |
| <span class="kw">with</span> a particular label to be ignored during backward propagation. **This has no effect when |
| softmax `output` has same shape as `label`**. |
| |
| Example:: |
| |
| data = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>],[<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>],[<span class="num">3</span>,<span class="num">3</span>,<span class="num">3</span>,<span class="num">3</span>],[<span class="num">4</span>,<span class="num">4</span>,<span class="num">4</span>,<span class="num">4</span>] ] |
| label = [<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>,<span class="num">3</span>] |
| ignore_label = <span class="num">1</span> |
| SoftmaxOutput(data=data, label = label,\ |
| multi_output=<span class="kw">true</span>, use_ignore=<span class="kw">true</span>,\ |
| ignore_label=ignore_label) |
| ## forward softmax output |
| `[ [ <span class="num">0.0320586</span> <span class="num">0.08714432</span> <span class="num">0.23688284</span> <span class="num">0.64391428</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> ] ] |
| ## backward gradient output |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> ] |
| [-<span class="num">0.75</span> <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> -<span class="num">0.75</span> <span class="num">0.25</span>] |
| [ <span class="num">0.25</span> <span class="num">0.25</span> <span class="num">0.25</span> -<span class="num">0.75</span>] ] |
| ## notice that the first row is all <span class="num">0</span> because label[<span class="num">0</span>] is <span class="num">1</span>, which is equal to ignore_label. |
| |
| - The parameter `grad_scale` can be used to rescale the gradient, which is often used to |
| give each loss function different weights. |
| |
| - This operator also supports various ways to normalize the gradient by `normalization`, |
| The `normalization` is applied <span class="kw">if</span> softmax output has different shape than the labels. |
| The `normalization` mode can be set to the followings: |
| |
| - ``<span class="lit">'null'</span>``: <span class="kw">do</span> nothing. |
| - ``<span class="lit">'batch'</span>``: divide the gradient by the batch size. |
| - ``<span class="lit">'valid'</span>``: divide the gradient by the number of instances which are not ignored. |
| |
| |
| |
| Defined in src/operator/softmax_output.cc:L242</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SpatialTransformer" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SpatialTransformer(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SpatialTransformer(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SpatialTransformer</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SpatialTransformer(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies a spatial transformer to input feature map.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SpatialTransformer" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SpatialTransformer(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SpatialTransformer(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SpatialTransformer</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SpatialTransformer(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies a spatial transformer to input feature map.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SwapAxis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SwapAxis(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SwapAxis(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SwapAxis</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SwapAxis(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Interchanges two axes of an array. |
| |
| Examples:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ]) |
| swapaxes(x, <span class="num">0</span>, <span class="num">1</span>) = `[ [ <span class="num">1</span>], |
| [ <span class="num">2</span>], |
| [ <span class="num">3</span>] ] |
| |
| x = `[ `[ [ <span class="num">0</span>, <span class="num">1</span>], |
| [ <span class="num">2</span>, <span class="num">3</span>] ], |
| `[ [ <span class="num">4</span>, <span class="num">5</span>], |
| [ <span class="num">6</span>, <span class="num">7</span>] ] ] <span class="cmt">// (2,2,2) array</span> |
| |
| swapaxes(x, <span class="num">0</span>, <span class="num">2</span>) = `[ `[ [ <span class="num">0</span>, <span class="num">4</span>], |
| [ <span class="num">2</span>, <span class="num">6</span>] ], |
| `[ [ <span class="num">1</span>, <span class="num">5</span>], |
| [ <span class="num">3</span>, <span class="num">7</span>] ] ] |
| |
| |
| Defined in src/operator/swapaxis.cc:L69</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#SwapAxis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="SwapAxis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="SwapAxis(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">SwapAxis</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@SwapAxis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Interchanges two axes of an array. |
| |
| Examples:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ]) |
| swapaxes(x, <span class="num">0</span>, <span class="num">1</span>) = `[ [ <span class="num">1</span>], |
| [ <span class="num">2</span>], |
| [ <span class="num">3</span>] ] |
| |
| x = `[ `[ [ <span class="num">0</span>, <span class="num">1</span>], |
| [ <span class="num">2</span>, <span class="num">3</span>] ], |
| `[ [ <span class="num">4</span>, <span class="num">5</span>], |
| [ <span class="num">6</span>, <span class="num">7</span>] ] ] <span class="cmt">// (2,2,2) array</span> |
| |
| swapaxes(x, <span class="num">0</span>, <span class="num">2</span>) = `[ `[ [ <span class="num">0</span>, <span class="num">4</span>], |
| [ <span class="num">2</span>, <span class="num">6</span>] ], |
| `[ [ <span class="num">1</span>, <span class="num">5</span>], |
| [ <span class="num">3</span>, <span class="num">7</span>] ] ] |
| |
| |
| Defined in src/operator/swapaxis.cc:L69</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#UpSampling" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="UpSampling(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="UpSampling(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">UpSampling</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@UpSampling(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Upsamples the given input data. |
| |
| Two algorithms (``sample_type``) are available <span class="kw">for</span> upsampling: |
| |
| - Nearest Neighbor |
| - Bilinear |
| |
| **Nearest Neighbor Upsampling** |
| |
| Input data is expected to be NCHW. |
| |
| Example:: |
| |
| x = `[ [`[ [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] ] ] |
| |
| UpSampling(x, scale=<span class="num">2</span>, sample_type='nearest') = `[ [`[ [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] ] ] |
| |
| **Bilinear Upsampling** |
| |
| Uses `deconvolution` algorithm under the hood. You need provide both input data and the kernel. |
| |
| Input data is expected to be NCHW. |
| |
| `num_filter` is expected to be same as the number of channels. |
| |
| Example:: |
| |
| x = `[ [`[ [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] ] ] |
| |
| w = `[ [`[ [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] ] ] |
| |
| UpSampling(x, w, scale=<span class="num">2</span>, sample_type='bilinear', num_filter=<span class="num">1</span>) = `[ [`[ [<span class="num">1.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">1.</span>] |
| [<span class="num">2.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">2.</span>] |
| [<span class="num">2.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">2.</span>] |
| [<span class="num">2.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">2.</span>] |
| [<span class="num">2.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">2.</span>] |
| [<span class="num">1.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">1.</span>] ] ] ] |
| |
| |
| Defined in src/operator/nn/upsampling.cc:L172</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#UpSampling" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="UpSampling(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="UpSampling(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">UpSampling</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@UpSampling(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Upsamples the given input data. |
| |
| Two algorithms (``sample_type``) are available <span class="kw">for</span> upsampling: |
| |
| - Nearest Neighbor |
| - Bilinear |
| |
| **Nearest Neighbor Upsampling** |
| |
| Input data is expected to be NCHW. |
| |
| Example:: |
| |
| x = `[ [`[ [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] ] ] |
| |
| UpSampling(x, scale=<span class="num">2</span>, sample_type='nearest') = `[ [`[ [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] ] ] |
| |
| **Bilinear Upsampling** |
| |
| Uses `deconvolution` algorithm under the hood. You need provide both input data and the kernel. |
| |
| Input data is expected to be NCHW. |
| |
| `num_filter` is expected to be same as the number of channels. |
| |
| Example:: |
| |
| x = `[ [`[ [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] ] ] |
| |
| w = `[ [`[ [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [<span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] ] ] |
| |
| UpSampling(x, w, scale=<span class="num">2</span>, sample_type='bilinear', num_filter=<span class="num">1</span>) = `[ [`[ [<span class="num">1.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">1.</span>] |
| [<span class="num">2.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">2.</span>] |
| [<span class="num">2.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">2.</span>] |
| [<span class="num">2.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">2.</span>] |
| [<span class="num">2.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">4.</span> <span class="num">2.</span>] |
| [<span class="num">1.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">2.</span> <span class="num">1.</span>] ] ] ] |
| |
| |
| Defined in src/operator/nn/upsampling.cc:L172</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#abs" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="abs(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="abs(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">abs</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@abs(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise absolute value of the input. |
| |
| Example:: |
| |
| abs([-<span class="num">2</span>, <span class="num">0</span>, <span class="num">3</span>]) = [<span class="num">2</span>, <span class="num">0</span>, <span class="num">3</span>] |
| |
| The storage <span class="kw">type</span> of ``abs`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - abs(default) = default |
| - abs(row_sparse) = row_sparse |
| - abs(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L720</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#abs" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="abs(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="abs(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">abs</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@abs(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise absolute value of the input. |
| |
| Example:: |
| |
| abs([-<span class="num">2</span>, <span class="num">0</span>, <span class="num">3</span>]) = [<span class="num">2</span>, <span class="num">0</span>, <span class="num">3</span>] |
| |
| The storage <span class="kw">type</span> of ``abs`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - abs(default) = default |
| - abs(row_sparse) = row_sparse |
| - abs(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L720</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#adam_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="adam_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="adam_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">adam_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@adam_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Adam optimizer. Adam is seen as a generalization |
| of AdaGrad. |
| |
| Adam update consists of the following steps, where g represents gradient and m, v |
| are <span class="num">1</span>st and <span class="num">2</span>nd order moment estimates (mean and variance). |
| |
| .. math:: |
| |
| g_t = \nabla J(W_{t-<span class="num">1</span>})\\ |
| m_t = \beta_1 m_{t-<span class="num">1</span>} + (<span class="num">1</span> - \beta_1) g_t\\ |
| v_t = \beta_2 v_{t-<span class="num">1</span>} + (<span class="num">1</span> - \beta_2) g_t^<span class="num">2</span>\\ |
| W_t = W_{t-<span class="num">1</span>} - \alpha \frac{ m_t }{ \sqrt{ v_t } + \epsilon } |
| |
| It updates the weights using:: |
| |
| m = beta1*m + (<span class="num">1</span>-beta1)*grad |
| v = beta2*v + (<span class="num">1</span>-beta2)*(grad**<span class="num">2</span>) |
| w += - learning_rate * m / (sqrt(v) + epsilon) |
| |
| However, <span class="kw">if</span> grad's storage <span class="kw">type</span> is ``row_sparse``, ``lazy_update`` is True and the storage |
| <span class="kw">type</span> of weight is the same as those of m and v, |
| only the row slices whose indices appear in grad.indices are updated (<span class="kw">for</span> w, m and v):: |
| |
| <span class="kw">for</span> row in grad.indices: |
| m[row] = beta1*m[row] + (<span class="num">1</span>-beta1)*grad[row] |
| v[row] = beta2*v[row] + (<span class="num">1</span>-beta2)*(grad[row]**<span class="num">2</span>) |
| w[row] += - learning_rate * m[row] / (sqrt(v[row]) + epsilon) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L687</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#adam_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="adam_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="adam_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">adam_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@adam_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Adam optimizer. Adam is seen as a generalization |
| of AdaGrad. |
| |
| Adam update consists of the following steps, where g represents gradient and m, v |
| are <span class="num">1</span>st and <span class="num">2</span>nd order moment estimates (mean and variance). |
| |
| .. math:: |
| |
| g_t = \nabla J(W_{t-<span class="num">1</span>})\\ |
| m_t = \beta_1 m_{t-<span class="num">1</span>} + (<span class="num">1</span> - \beta_1) g_t\\ |
| v_t = \beta_2 v_{t-<span class="num">1</span>} + (<span class="num">1</span> - \beta_2) g_t^<span class="num">2</span>\\ |
| W_t = W_{t-<span class="num">1</span>} - \alpha \frac{ m_t }{ \sqrt{ v_t } + \epsilon } |
| |
| It updates the weights using:: |
| |
| m = beta1*m + (<span class="num">1</span>-beta1)*grad |
| v = beta2*v + (<span class="num">1</span>-beta2)*(grad**<span class="num">2</span>) |
| w += - learning_rate * m / (sqrt(v) + epsilon) |
| |
| However, <span class="kw">if</span> grad's storage <span class="kw">type</span> is ``row_sparse``, ``lazy_update`` is True and the storage |
| <span class="kw">type</span> of weight is the same as those of m and v, |
| only the row slices whose indices appear in grad.indices are updated (<span class="kw">for</span> w, m and v):: |
| |
| <span class="kw">for</span> row in grad.indices: |
| m[row] = beta1*m[row] + (<span class="num">1</span>-beta1)*grad[row] |
| v[row] = beta2*v[row] + (<span class="num">1</span>-beta2)*(grad[row]**<span class="num">2</span>) |
| w[row] += - learning_rate * m[row] / (sqrt(v[row]) + epsilon) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L687</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#add_n" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="add_n(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="add_n(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">add_n</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@add_n(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Adds all input arguments element-wise. |
| |
| .. math:: |
| add\_n(a_1, a_2, ..., a_n) = a_1 + a_2 + ... + a_n |
| |
| ``add_n`` is potentially more efficient than calling ``add`` by `n` times. |
| |
| The storage <span class="kw">type</span> of ``add_n`` output depends on storage types of inputs |
| |
| - add_n(row_sparse, row_sparse, ..) = row_sparse |
| - add_n(default, csr, default) = default |
| - add_n(any input combinations longer than <span class="num">4</span> (><span class="num">4</span>) <span class="kw">with</span> at least one default <span class="kw">type</span>) = default |
| - otherwise, ``add_n`` falls all inputs back to default storage and generates default storage |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_sum.cc:L155</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#add_n" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="add_n(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="add_n(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">add_n</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@add_n(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Adds all input arguments element-wise. |
| |
| .. math:: |
| add\_n(a_1, a_2, ..., a_n) = a_1 + a_2 + ... + a_n |
| |
| ``add_n`` is potentially more efficient than calling ``add`` by `n` times. |
| |
| The storage <span class="kw">type</span> of ``add_n`` output depends on storage types of inputs |
| |
| - add_n(row_sparse, row_sparse, ..) = row_sparse |
| - add_n(default, csr, default) = default |
| - add_n(any input combinations longer than <span class="num">4</span> (><span class="num">4</span>) <span class="kw">with</span> at least one default <span class="kw">type</span>) = default |
| - otherwise, ``add_n`` falls all inputs back to default storage and generates default storage |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_sum.cc:L155</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#all_finite" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="all_finite(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="all_finite(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">all_finite</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@all_finite(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Check <span class="kw">if</span> all the float numbers in the array are finite (used <span class="kw">for</span> AMP) |
| |
| |
| Defined in src/operator/contrib/all_finite.cc:L100</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#all_finite" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="all_finite(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="all_finite(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">all_finite</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@all_finite(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Check <span class="kw">if</span> all the float numbers in the array are finite (used <span class="kw">for</span> AMP) |
| |
| |
| Defined in src/operator/contrib/all_finite.cc:L100</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#amp_cast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="amp_cast(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="amp_cast(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">amp_cast</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@amp_cast(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Cast function between low precision float/FP32 used by AMP. |
| |
| It casts only between low precision float/FP32 and does not <span class="kw">do</span> anything <span class="kw">for</span> other types. |
| |
| |
| Defined in src/operator/tensor/amp_cast.cc:L125</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#amp_cast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="amp_cast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="amp_cast(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">amp_cast</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@amp_cast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Cast function between low precision float/FP32 used by AMP. |
| |
| It casts only between low precision float/FP32 and does not <span class="kw">do</span> anything <span class="kw">for</span> other types. |
| |
| |
| Defined in src/operator/tensor/amp_cast.cc:L125</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#amp_multicast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="amp_multicast(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="amp_multicast(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">amp_multicast</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@amp_multicast(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Cast function used by AMP, that casts its inputs to the common widest <span class="kw">type</span>. |
| |
| It casts only between low precision float/FP32 and does not <span class="kw">do</span> anything <span class="kw">for</span> other types. |
| |
| |
| |
| Defined in src/operator/tensor/amp_cast.cc:L169</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#amp_multicast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="amp_multicast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="amp_multicast(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">amp_multicast</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@amp_multicast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Cast function used by AMP, that casts its inputs to the common widest <span class="kw">type</span>. |
| |
| It casts only between low precision float/FP32 and does not <span class="kw">do</span> anything <span class="kw">for</span> other types. |
| |
| |
| |
| Defined in src/operator/tensor/amp_cast.cc:L169</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arccos" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arccos(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arccos(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arccos</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arccos(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse cosine of the input array. |
| |
| The input should be in range `[-<span class="num">1</span>, <span class="num">1</span>]`. |
| The output is in the closed interval :math:`[<span class="num">0</span>, \pi]` |
| |
| .. math:: |
| arccos([-<span class="num">1</span>, -<span class="num">.707</span>, <span class="num">0</span>, <span class="num">.707</span>, <span class="num">1</span>]) = [\pi, <span class="num">3</span>\pi/<span class="num">4</span>, \pi/<span class="num">2</span>, \pi/<span class="num">4</span>, <span class="num">0</span>] |
| |
| The storage <span class="kw">type</span> of ``arccos`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L233</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arccos" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arccos(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arccos(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arccos</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arccos(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse cosine of the input array. |
| |
| The input should be in range `[-<span class="num">1</span>, <span class="num">1</span>]`. |
| The output is in the closed interval :math:`[<span class="num">0</span>, \pi]` |
| |
| .. math:: |
| arccos([-<span class="num">1</span>, -<span class="num">.707</span>, <span class="num">0</span>, <span class="num">.707</span>, <span class="num">1</span>]) = [\pi, <span class="num">3</span>\pi/<span class="num">4</span>, \pi/<span class="num">2</span>, \pi/<span class="num">4</span>, <span class="num">0</span>] |
| |
| The storage <span class="kw">type</span> of ``arccos`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L233</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arccosh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arccosh(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arccosh(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arccosh</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arccosh(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the element-wise inverse hyperbolic cosine of the input array, \ |
| computed element-wise. |
| |
| The storage <span class="kw">type</span> of ``arccosh`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L535</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arccosh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arccosh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arccosh(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arccosh</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arccosh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the element-wise inverse hyperbolic cosine of the input array, \ |
| computed element-wise. |
| |
| The storage <span class="kw">type</span> of ``arccosh`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L535</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arcsin" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arcsin(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arcsin(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arcsin</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arcsin(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse sine of the input array. |
| |
| The input should be in the range `[-<span class="num">1</span>, <span class="num">1</span>]`. |
| The output is in the closed interval of [:math:`-\pi/<span class="num">2</span>`, :math:`\pi/<span class="num">2</span>`]. |
| |
| .. math:: |
| arcsin([-<span class="num">1</span>, -<span class="num">.707</span>, <span class="num">0</span>, <span class="num">.707</span>, <span class="num">1</span>]) = [-\pi/<span class="num">2</span>, -\pi/<span class="num">4</span>, <span class="num">0</span>, \pi/<span class="num">4</span>, \pi/<span class="num">2</span>] |
| |
| The storage <span class="kw">type</span> of ``arcsin`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - arcsin(default) = default |
| - arcsin(row_sparse) = row_sparse |
| - arcsin(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L187</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arcsin" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arcsin(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arcsin(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arcsin</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arcsin(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse sine of the input array. |
| |
| The input should be in the range `[-<span class="num">1</span>, <span class="num">1</span>]`. |
| The output is in the closed interval of [:math:`-\pi/<span class="num">2</span>`, :math:`\pi/<span class="num">2</span>`]. |
| |
| .. math:: |
| arcsin([-<span class="num">1</span>, -<span class="num">.707</span>, <span class="num">0</span>, <span class="num">.707</span>, <span class="num">1</span>]) = [-\pi/<span class="num">2</span>, -\pi/<span class="num">4</span>, <span class="num">0</span>, \pi/<span class="num">4</span>, \pi/<span class="num">2</span>] |
| |
| The storage <span class="kw">type</span> of ``arcsin`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - arcsin(default) = default |
| - arcsin(row_sparse) = row_sparse |
| - arcsin(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L187</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arcsinh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arcsinh(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arcsinh(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arcsinh</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arcsinh(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the element-wise inverse hyperbolic sine of the input array, \ |
| computed element-wise. |
| |
| The storage <span class="kw">type</span> of ``arcsinh`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - arcsinh(default) = default |
| - arcsinh(row_sparse) = row_sparse |
| - arcsinh(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L494</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arcsinh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arcsinh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arcsinh(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arcsinh</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arcsinh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the element-wise inverse hyperbolic sine of the input array, \ |
| computed element-wise. |
| |
| The storage <span class="kw">type</span> of ``arcsinh`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - arcsinh(default) = default |
| - arcsinh(row_sparse) = row_sparse |
| - arcsinh(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L494</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arctan" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arctan(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arctan(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arctan</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arctan(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse tangent of the input array. |
| |
| The output is in the closed interval :math:`[-\pi/<span class="num">2</span>, \pi/<span class="num">2</span>]` |
| |
| .. math:: |
| arctan([-<span class="num">1</span>, <span class="num">0</span>, <span class="num">1</span>]) = [-\pi/<span class="num">4</span>, <span class="num">0</span>, \pi/<span class="num">4</span>] |
| |
| The storage <span class="kw">type</span> of ``arctan`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - arctan(default) = default |
| - arctan(row_sparse) = row_sparse |
| - arctan(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L282</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arctan" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arctan(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arctan(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arctan</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arctan(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse tangent of the input array. |
| |
| The output is in the closed interval :math:`[-\pi/<span class="num">2</span>, \pi/<span class="num">2</span>]` |
| |
| .. math:: |
| arctan([-<span class="num">1</span>, <span class="num">0</span>, <span class="num">1</span>]) = [-\pi/<span class="num">4</span>, <span class="num">0</span>, \pi/<span class="num">4</span>] |
| |
| The storage <span class="kw">type</span> of ``arctan`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - arctan(default) = default |
| - arctan(row_sparse) = row_sparse |
| - arctan(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L282</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arctanh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arctanh(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arctanh(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arctanh</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arctanh(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the element-wise inverse hyperbolic tangent of the input array, \ |
| computed element-wise. |
| |
| The storage <span class="kw">type</span> of ``arctanh`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - arctanh(default) = default |
| - arctanh(row_sparse) = row_sparse |
| - arctanh(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L579</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#arctanh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="arctanh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="arctanh(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">arctanh</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@arctanh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the element-wise inverse hyperbolic tangent of the input array, \ |
| computed element-wise. |
| |
| The storage <span class="kw">type</span> of ``arctanh`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - arctanh(default) = default |
| - arctanh(row_sparse) = row_sparse |
| - arctanh(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L579</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#argmax" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="argmax(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="argmax(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">argmax</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@argmax(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns indices of the maximum values along an axis. |
| |
| In the <span class="kw">case</span> of multiple occurrences of maximum values, the indices corresponding to the first occurrence |
| are returned. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>] ] |
| |
| <span class="cmt">// argmax along axis 0</span> |
| argmax(x, axis=<span class="num">0</span>) = [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] |
| |
| <span class="cmt">// argmax along axis 1</span> |
| argmax(x, axis=<span class="num">1</span>) = [ <span class="num">2.</span>, <span class="num">2.</span>] |
| |
| <span class="cmt">// argmax along axis 1 keeping same dims as an input array</span> |
| argmax(x, axis=<span class="num">1</span>, keepdims=True) = `[ [ <span class="num">2.</span>], |
| [ <span class="num">2.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L51</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#argmax" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="argmax(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="argmax(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">argmax</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@argmax(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns indices of the maximum values along an axis. |
| |
| In the <span class="kw">case</span> of multiple occurrences of maximum values, the indices corresponding to the first occurrence |
| are returned. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>] ] |
| |
| <span class="cmt">// argmax along axis 0</span> |
| argmax(x, axis=<span class="num">0</span>) = [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] |
| |
| <span class="cmt">// argmax along axis 1</span> |
| argmax(x, axis=<span class="num">1</span>) = [ <span class="num">2.</span>, <span class="num">2.</span>] |
| |
| <span class="cmt">// argmax along axis 1 keeping same dims as an input array</span> |
| argmax(x, axis=<span class="num">1</span>, keepdims=True) = `[ [ <span class="num">2.</span>], |
| [ <span class="num">2.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L51</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#argmax_channel" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="argmax_channel(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="argmax_channel(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">argmax_channel</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@argmax_channel(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns argmax indices of each channel from the input array. |
| |
| The result will be an NDArray of shape (num_channel,). |
| |
| In <span class="kw">case</span> of multiple occurrences of the maximum values, the indices corresponding to the first occurrence |
| are returned. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>] ] |
| |
| argmax_channel(x) = [ <span class="num">2.</span>, <span class="num">2.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L96</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#argmax_channel" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="argmax_channel(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="argmax_channel(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">argmax_channel</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@argmax_channel(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns argmax indices of each channel from the input array. |
| |
| The result will be an NDArray of shape (num_channel,). |
| |
| In <span class="kw">case</span> of multiple occurrences of the maximum values, the indices corresponding to the first occurrence |
| are returned. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>] ] |
| |
| argmax_channel(x) = [ <span class="num">2.</span>, <span class="num">2.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L96</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#argmin" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="argmin(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="argmin(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">argmin</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@argmin(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns indices of the minimum values along an axis. |
| |
| In the <span class="kw">case</span> of multiple occurrences of minimum values, the indices corresponding to the first occurrence |
| are returned. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>] ] |
| |
| <span class="cmt">// argmin along axis 0</span> |
| argmin(x, axis=<span class="num">0</span>) = [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] |
| |
| <span class="cmt">// argmin along axis 1</span> |
| argmin(x, axis=<span class="num">1</span>) = [ <span class="num">0.</span>, <span class="num">0.</span>] |
| |
| <span class="cmt">// argmin along axis 1 keeping same dims as an input array</span> |
| argmin(x, axis=<span class="num">1</span>, keepdims=True) = `[ [ <span class="num">0.</span>], |
| [ <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L76</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#argmin" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="argmin(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="argmin(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">argmin</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@argmin(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns indices of the minimum values along an axis. |
| |
| In the <span class="kw">case</span> of multiple occurrences of minimum values, the indices corresponding to the first occurrence |
| are returned. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>] ] |
| |
| <span class="cmt">// argmin along axis 0</span> |
| argmin(x, axis=<span class="num">0</span>) = [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] |
| |
| <span class="cmt">// argmin along axis 1</span> |
| argmin(x, axis=<span class="num">1</span>) = [ <span class="num">0.</span>, <span class="num">0.</span>] |
| |
| <span class="cmt">// argmin along axis 1 keeping same dims as an input array</span> |
| argmin(x, axis=<span class="num">1</span>, keepdims=True) = `[ [ <span class="num">0.</span>], |
| [ <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L76</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#argsort" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="argsort(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="argsort(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">argsort</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@argsort(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the indices that would sort an input array along the given axis. |
| |
| This function performs sorting along the given axis and returns an array of indices having same shape |
| as an input array that index data in sorted order. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.3</span>, <span class="num">0.2</span>, <span class="num">0.4</span>], |
| [ <span class="num">0.1</span>, <span class="num">0.3</span>, <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// sort along axis -1</span> |
| argsort(x) = `[ [ <span class="num">1.</span>, <span class="num">0.</span>, <span class="num">2.</span>], |
| [ <span class="num">0.</span>, <span class="num">2.</span>, <span class="num">1.</span>] ] |
| |
| <span class="cmt">// sort along axis 0</span> |
| argsort(x, axis=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">0.</span>, <span class="num">1.</span>] |
| [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| <span class="cmt">// flatten and then sort</span> |
| argsort(x, axis=<span class="std">None</span>) = [ <span class="num">3.</span>, <span class="num">1.</span>, <span class="num">5.</span>, <span class="num">0.</span>, <span class="num">4.</span>, <span class="num">2.</span>] |
| |
| |
| Defined in src/operator/tensor/ordering_op.cc:L184</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#argsort" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="argsort(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="argsort(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">argsort</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@argsort(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the indices that would sort an input array along the given axis. |
| |
| This function performs sorting along the given axis and returns an array of indices having same shape |
| as an input array that index data in sorted order. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.3</span>, <span class="num">0.2</span>, <span class="num">0.4</span>], |
| [ <span class="num">0.1</span>, <span class="num">0.3</span>, <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// sort along axis -1</span> |
| argsort(x) = `[ [ <span class="num">1.</span>, <span class="num">0.</span>, <span class="num">2.</span>], |
| [ <span class="num">0.</span>, <span class="num">2.</span>, <span class="num">1.</span>] ] |
| |
| <span class="cmt">// sort along axis 0</span> |
| argsort(x, axis=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">0.</span>, <span class="num">1.</span>] |
| [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| <span class="cmt">// flatten and then sort</span> |
| argsort(x, axis=<span class="std">None</span>) = [ <span class="num">3.</span>, <span class="num">1.</span>, <span class="num">5.</span>, <span class="num">0.</span>, <span class="num">4.</span>, <span class="num">2.</span>] |
| |
| |
| Defined in src/operator/tensor/ordering_op.cc:L184</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#batch_dot" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="batch_dot(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="batch_dot(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">batch_dot</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@batch_dot(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Batchwise dot product. |
| |
| ``batch_dot`` is used to compute dot product of ``x`` and ``y`` when ``x`` and |
| ``y`` are data in batch, namely N-D (N >= <span class="num">3</span>) arrays in shape of `(B0, ..., B_i, :, :)`. |
| |
| For example, given ``x`` <span class="kw">with</span> shape `(B_0, ..., B_i, N, M)` and ``y`` <span class="kw">with</span> shape |
| `(B_0, ..., B_i, M, K)`, the result array will have shape `(B_0, ..., B_i, N, K)`, |
| which is computed by:: |
| |
| batch_dot(x,y)[b_0, ..., b_i, :, :] = dot(x[b_0, ..., b_i, :, :], y[b_0, ..., b_i, :, :]) |
| |
| |
| |
| Defined in src/operator/tensor/dot.cc:L127</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#batch_dot" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="batch_dot(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="batch_dot(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">batch_dot</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@batch_dot(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Batchwise dot product. |
| |
| ``batch_dot`` is used to compute dot product of ``x`` and ``y`` when ``x`` and |
| ``y`` are data in batch, namely N-D (N >= <span class="num">3</span>) arrays in shape of `(B0, ..., B_i, :, :)`. |
| |
| For example, given ``x`` <span class="kw">with</span> shape `(B_0, ..., B_i, N, M)` and ``y`` <span class="kw">with</span> shape |
| `(B_0, ..., B_i, M, K)`, the result array will have shape `(B_0, ..., B_i, N, K)`, |
| which is computed by:: |
| |
| batch_dot(x,y)[b_0, ..., b_i, :, :] = dot(x[b_0, ..., b_i, :, :], y[b_0, ..., b_i, :, :]) |
| |
| |
| |
| Defined in src/operator/tensor/dot.cc:L127</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#batch_take" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="batch_take(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="batch_take(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">batch_take</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@batch_take(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Takes elements from a data batch. |
| |
| .. note:: |
| `batch_take` is deprecated. Use `pick` instead. |
| |
| Given an input array of shape ``(d0, d1)`` and indices of shape ``(i0,)``, the result will be |
| an output array of shape ``(i0,)`` <span class="kw">with</span>:: |
| |
| output[i] = input[i, indices[i] ] |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] |
| |
| <span class="cmt">// takes elements with specified indices</span> |
| batch_take(x, [<span class="num">0</span>,<span class="num">1</span>,<span class="num">0</span>]) = [ <span class="num">1.</span> <span class="num">4.</span> <span class="num">5.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/indexing_op.cc:L835</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#batch_take" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="batch_take(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="batch_take(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">batch_take</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@batch_take(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Takes elements from a data batch. |
| |
| .. note:: |
| `batch_take` is deprecated. Use `pick` instead. |
| |
| Given an input array of shape ``(d0, d1)`` and indices of shape ``(i0,)``, the result will be |
| an output array of shape ``(i0,)`` <span class="kw">with</span>:: |
| |
| output[i] = input[i, indices[i] ] |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] |
| |
| <span class="cmt">// takes elements with specified indices</span> |
| batch_take(x, [<span class="num">0</span>,<span class="num">1</span>,<span class="num">0</span>]) = [ <span class="num">1.</span> <span class="num">4.</span> <span class="num">5.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/indexing_op.cc:L835</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_add" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_add(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_add(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_add</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_add(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise sum of the input arrays <span class="kw">with</span> broadcasting. |
| |
| `broadcast_plus` is an alias to the function `broadcast_add`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_add(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| broadcast_plus(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_add(csr, dense(<span class="num">1</span>D)) = dense |
| broadcast_add(dense(<span class="num">1</span>D), csr) = dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L57</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_add" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_add(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_add(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_add</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_add(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise sum of the input arrays <span class="kw">with</span> broadcasting. |
| |
| `broadcast_plus` is an alias to the function `broadcast_add`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_add(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| broadcast_plus(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_add(csr, dense(<span class="num">1</span>D)) = dense |
| broadcast_add(dense(<span class="num">1</span>D), csr) = dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L57</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_axes" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_axes(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_axes(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_axes</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_axes(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Broadcasts the input array over particular axes. |
| |
| Broadcasting is allowed on axes <span class="kw">with</span> size <span class="num">1</span>, such as from `(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">1</span>)` to |
| `(<span class="num">2</span>,<span class="num">8</span>,<span class="num">3</span>,<span class="num">9</span>)`. Elements will be duplicated on the broadcasted axes. |
| |
| `broadcast_axes` is an alias to the function `broadcast_axis`. |
| |
| Example:: |
| |
| <span class="cmt">// given x of shape (1,2,1)</span> |
| x = `[ `[ [ <span class="num">1.</span>], |
| [ <span class="num">2.</span>] ] ] |
| |
| <span class="cmt">// broadcast x on on axis 2</span> |
| broadcast_axis(x, axis=<span class="num">2</span>, size=<span class="num">3</span>) = `[ `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] ] |
| <span class="cmt">// broadcast x on on axes 0 and 2</span> |
| broadcast_axis(x, axis=(<span class="num">0</span>,<span class="num">2</span>), size=(<span class="num">2</span>,<span class="num">3</span>)) = `[ `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ], |
| `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L92</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_axes" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_axes(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_axes(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_axes</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_axes(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Broadcasts the input array over particular axes. |
| |
| Broadcasting is allowed on axes <span class="kw">with</span> size <span class="num">1</span>, such as from `(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">1</span>)` to |
| `(<span class="num">2</span>,<span class="num">8</span>,<span class="num">3</span>,<span class="num">9</span>)`. Elements will be duplicated on the broadcasted axes. |
| |
| `broadcast_axes` is an alias to the function `broadcast_axis`. |
| |
| Example:: |
| |
| <span class="cmt">// given x of shape (1,2,1)</span> |
| x = `[ `[ [ <span class="num">1.</span>], |
| [ <span class="num">2.</span>] ] ] |
| |
| <span class="cmt">// broadcast x on on axis 2</span> |
| broadcast_axis(x, axis=<span class="num">2</span>, size=<span class="num">3</span>) = `[ `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] ] |
| <span class="cmt">// broadcast x on on axes 0 and 2</span> |
| broadcast_axis(x, axis=(<span class="num">0</span>,<span class="num">2</span>), size=(<span class="num">2</span>,<span class="num">3</span>)) = `[ `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ], |
| `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L92</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_axis(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_axis</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Broadcasts the input array over particular axes. |
| |
| Broadcasting is allowed on axes <span class="kw">with</span> size <span class="num">1</span>, such as from `(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">1</span>)` to |
| `(<span class="num">2</span>,<span class="num">8</span>,<span class="num">3</span>,<span class="num">9</span>)`. Elements will be duplicated on the broadcasted axes. |
| |
| `broadcast_axes` is an alias to the function `broadcast_axis`. |
| |
| Example:: |
| |
| <span class="cmt">// given x of shape (1,2,1)</span> |
| x = `[ `[ [ <span class="num">1.</span>], |
| [ <span class="num">2.</span>] ] ] |
| |
| <span class="cmt">// broadcast x on on axis 2</span> |
| broadcast_axis(x, axis=<span class="num">2</span>, size=<span class="num">3</span>) = `[ `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] ] |
| <span class="cmt">// broadcast x on on axes 0 and 2</span> |
| broadcast_axis(x, axis=(<span class="num">0</span>,<span class="num">2</span>), size=(<span class="num">2</span>,<span class="num">3</span>)) = `[ `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ], |
| `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L92</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_axis(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_axis</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Broadcasts the input array over particular axes. |
| |
| Broadcasting is allowed on axes <span class="kw">with</span> size <span class="num">1</span>, such as from `(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">1</span>)` to |
| `(<span class="num">2</span>,<span class="num">8</span>,<span class="num">3</span>,<span class="num">9</span>)`. Elements will be duplicated on the broadcasted axes. |
| |
| `broadcast_axes` is an alias to the function `broadcast_axis`. |
| |
| Example:: |
| |
| <span class="cmt">// given x of shape (1,2,1)</span> |
| x = `[ `[ [ <span class="num">1.</span>], |
| [ <span class="num">2.</span>] ] ] |
| |
| <span class="cmt">// broadcast x on on axis 2</span> |
| broadcast_axis(x, axis=<span class="num">2</span>, size=<span class="num">3</span>) = `[ `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] ] |
| <span class="cmt">// broadcast x on on axes 0 and 2</span> |
| broadcast_axis(x, axis=(<span class="num">0</span>,<span class="num">2</span>), size=(<span class="num">2</span>,<span class="num">3</span>)) = `[ `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ], |
| `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L92</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_div" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_div(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_div(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_div</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_div(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise division of the input arrays <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">6.</span>, <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">6.</span>, <span class="num">6.</span>, <span class="num">6.</span>] ] |
| |
| y = `[ [ <span class="num">2.</span>], |
| [ <span class="num">3.</span>] ] |
| |
| broadcast_div(x, y) = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">3.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_div(csr, dense(<span class="num">1</span>D)) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L186</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_div" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_div(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_div(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_div</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_div(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise division of the input arrays <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">6.</span>, <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">6.</span>, <span class="num">6.</span>, <span class="num">6.</span>] ] |
| |
| y = `[ [ <span class="num">2.</span>], |
| [ <span class="num">3.</span>] ] |
| |
| broadcast_div(x, y) = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">3.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_div(csr, dense(<span class="num">1</span>D)) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L186</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_equal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_equal(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_equal(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_equal</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_equal(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **equal to** (==) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_equal(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L45</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_equal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_equal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_equal(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_equal</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_equal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **equal to** (==) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_equal(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L45</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_greater" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_greater(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_greater(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_greater</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_greater(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **greater than** (>) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_greater(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L81</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_greater" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_greater(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_greater(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_greater</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_greater(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **greater than** (>) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_greater(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L81</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_greater_equal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_greater_equal(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_greater_equal(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_greater_equal</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_greater_equal(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **greater than or equal to** (>=) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_greater_equal(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L99</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_greater_equal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_greater_equal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_greater_equal(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_greater_equal</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_greater_equal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **greater than or equal to** (>=) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_greater_equal(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L99</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_hypot" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_hypot(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_hypot(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_hypot</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_hypot(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre> Returns the hypotenuse of a right angled triangle, given its <span class="lit">"legs"</span> |
| <span class="kw">with</span> broadcasting. |
| |
| It is equivalent to doing :math:`sqrt(x_1^<span class="num">2</span> + x_2^<span class="num">2</span>)`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">3.</span>] ] |
| |
| y = `[ [ <span class="num">4.</span>], |
| [ <span class="num">4.</span>] ] |
| |
| broadcast_hypot(x, y) = `[ [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">5.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">5.</span>] ] |
| |
| z = `[ [ <span class="num">0.</span>], |
| [ <span class="num">4.</span>] ] |
| |
| broadcast_hypot(x, z) = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">3.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">5.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L157</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_hypot" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_hypot(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_hypot(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_hypot</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_hypot(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre> Returns the hypotenuse of a right angled triangle, given its <span class="lit">"legs"</span> |
| <span class="kw">with</span> broadcasting. |
| |
| It is equivalent to doing :math:`sqrt(x_1^<span class="num">2</span> + x_2^<span class="num">2</span>)`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">3.</span>] ] |
| |
| y = `[ [ <span class="num">4.</span>], |
| [ <span class="num">4.</span>] ] |
| |
| broadcast_hypot(x, y) = `[ [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">5.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">5.</span>] ] |
| |
| z = `[ [ <span class="num">0.</span>], |
| [ <span class="num">4.</span>] ] |
| |
| broadcast_hypot(x, z) = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">3.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">5.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L157</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_lesser" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_lesser(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_lesser(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_lesser</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_lesser(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **lesser than** (<) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_lesser(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L117</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_lesser" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_lesser(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_lesser(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_lesser</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_lesser(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **lesser than** (<) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_lesser(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L117</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_lesser_equal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_lesser_equal(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_lesser_equal(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_lesser_equal</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_lesser_equal(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **lesser than or equal to** (<=) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_lesser_equal(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L135</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_lesser_equal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_lesser_equal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_lesser_equal(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_lesser_equal</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_lesser_equal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **lesser than or equal to** (<=) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_lesser_equal(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L135</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_like(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_like</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Broadcasts lhs to have the same shape as rhs. |
| |
| Broadcasting is a mechanism that allows NDArrays to perform arithmetic operations |
| <span class="kw">with</span> arrays of different shapes efficiently without creating multiple copies of arrays. |
| Also see, `Broadcasting <https:<span class="cmt">//docs.scipy.org/doc/numpy/user/basics.broadcasting.html>`_ for more explanation.</span> |
| |
| Broadcasting is allowed on axes <span class="kw">with</span> size <span class="num">1</span>, such as from `(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">1</span>)` to |
| `(<span class="num">2</span>,<span class="num">8</span>,<span class="num">3</span>,<span class="num">9</span>)`. Elements will be duplicated on the broadcasted axes. |
| |
| For example:: |
| |
| broadcast_like(`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ], `[ [<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>],[<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] ]) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] ]) |
| |
| broadcast_like([<span class="num">9</span>], [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>], lhs_axes=(<span class="num">0</span>,), rhs_axes=(-<span class="num">1</span>,)) = [<span class="num">9</span>,<span class="num">9</span>,<span class="num">9</span>,<span class="num">9</span>,<span class="num">9</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L178</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_like(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_like</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Broadcasts lhs to have the same shape as rhs. |
| |
| Broadcasting is a mechanism that allows NDArrays to perform arithmetic operations |
| <span class="kw">with</span> arrays of different shapes efficiently without creating multiple copies of arrays. |
| Also see, `Broadcasting <https:<span class="cmt">//docs.scipy.org/doc/numpy/user/basics.broadcasting.html>`_ for more explanation.</span> |
| |
| Broadcasting is allowed on axes <span class="kw">with</span> size <span class="num">1</span>, such as from `(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">1</span>)` to |
| `(<span class="num">2</span>,<span class="num">8</span>,<span class="num">3</span>,<span class="num">9</span>)`. Elements will be duplicated on the broadcasted axes. |
| |
| For example:: |
| |
| broadcast_like(`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ], `[ [<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>],[<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] ]) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] ]) |
| |
| broadcast_like([<span class="num">9</span>], [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>], lhs_axes=(<span class="num">0</span>,), rhs_axes=(-<span class="num">1</span>,)) = [<span class="num">9</span>,<span class="num">9</span>,<span class="num">9</span>,<span class="num">9</span>,<span class="num">9</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L178</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_logical_and" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_logical_and(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_logical_and(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_logical_and</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_logical_and(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **logical and** <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_logical_and(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L153</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_logical_and" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_logical_and(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_logical_and(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_logical_and</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_logical_and(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **logical and** <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_logical_and(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L153</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_logical_or" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_logical_or(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_logical_or(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_logical_or</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_logical_or(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **logical or** <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| y = `[ [ <span class="num">1.</span>], |
| [ <span class="num">0.</span>] ] |
| |
| broadcast_logical_or(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L171</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_logical_or" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_logical_or(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_logical_or(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_logical_or</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_logical_or(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **logical or** <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| y = `[ [ <span class="num">1.</span>], |
| [ <span class="num">0.</span>] ] |
| |
| broadcast_logical_or(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L171</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_logical_xor" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_logical_xor(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_logical_xor(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_logical_xor</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_logical_xor(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **logical xor** <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| y = `[ [ <span class="num">1.</span>], |
| [ <span class="num">0.</span>] ] |
| |
| broadcast_logical_xor(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L189</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_logical_xor" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_logical_xor(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_logical_xor(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_logical_xor</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_logical_xor(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **logical xor** <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| y = `[ [ <span class="num">1.</span>], |
| [ <span class="num">0.</span>] ] |
| |
| broadcast_logical_xor(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L189</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_maximum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_maximum(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_maximum(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_maximum</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_maximum(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise maximum of the input arrays <span class="kw">with</span> broadcasting. |
| |
| This function compares two input arrays and returns a <span class="kw">new</span> array having the element-wise maxima. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_maximum(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L80</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_maximum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_maximum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_maximum(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_maximum</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_maximum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise maximum of the input arrays <span class="kw">with</span> broadcasting. |
| |
| This function compares two input arrays and returns a <span class="kw">new</span> array having the element-wise maxima. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_maximum(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L80</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_minimum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_minimum(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_minimum(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_minimum</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_minimum(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise minimum of the input arrays <span class="kw">with</span> broadcasting. |
| |
| This function compares two input arrays and returns a <span class="kw">new</span> array having the element-wise minima. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_maximum(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L116</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_minimum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_minimum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_minimum(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_minimum</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_minimum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise minimum of the input arrays <span class="kw">with</span> broadcasting. |
| |
| This function compares two input arrays and returns a <span class="kw">new</span> array having the element-wise minima. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_maximum(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L116</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_minus" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_minus(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_minus(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_minus</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_minus(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise difference of the input arrays <span class="kw">with</span> broadcasting. |
| |
| `broadcast_minus` is an alias to the function `broadcast_sub`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_sub(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| broadcast_minus(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_sub/minus(csr, dense(<span class="num">1</span>D)) = dense |
| broadcast_sub/minus(dense(<span class="num">1</span>D), csr) = dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L105</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_minus" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_minus(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_minus(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_minus</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_minus(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise difference of the input arrays <span class="kw">with</span> broadcasting. |
| |
| `broadcast_minus` is an alias to the function `broadcast_sub`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_sub(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| broadcast_minus(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_sub/minus(csr, dense(<span class="num">1</span>D)) = dense |
| broadcast_sub/minus(dense(<span class="num">1</span>D), csr) = dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L105</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_mod" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_mod(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_mod(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_mod</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_mod(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise modulo of the input arrays <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">8.</span>, <span class="num">8.</span>, <span class="num">8.</span>], |
| [ <span class="num">8.</span>, <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| y = `[ [ <span class="num">2.</span>], |
| [ <span class="num">3.</span>] ] |
| |
| broadcast_mod(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L221</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_mod" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_mod(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_mod(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_mod</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_mod(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise modulo of the input arrays <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">8.</span>, <span class="num">8.</span>, <span class="num">8.</span>], |
| [ <span class="num">8.</span>, <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| y = `[ [ <span class="num">2.</span>], |
| [ <span class="num">3.</span>] ] |
| |
| broadcast_mod(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L221</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_mul" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_mul(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_mul(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_mul</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_mul(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise product of the input arrays <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_mul(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_mul(csr, dense(<span class="num">1</span>D)) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L145</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_mul" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_mul(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_mul(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_mul</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_mul(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise product of the input arrays <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_mul(x, y) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_mul(csr, dense(<span class="num">1</span>D)) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L145</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_not_equal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_not_equal(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_not_equal(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_not_equal</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_not_equal(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **not equal to** (!=) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_not_equal(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L63</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_not_equal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_not_equal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_not_equal(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_not_equal</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_not_equal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of element-wise **not equal to** (!=) comparison operation <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_not_equal(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_logic.cc:L63</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_plus" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_plus(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_plus(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_plus</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_plus(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise sum of the input arrays <span class="kw">with</span> broadcasting. |
| |
| `broadcast_plus` is an alias to the function `broadcast_add`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_add(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| broadcast_plus(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_add(csr, dense(<span class="num">1</span>D)) = dense |
| broadcast_add(dense(<span class="num">1</span>D), csr) = dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L57</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_plus" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_plus(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_plus(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_plus</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_plus(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise sum of the input arrays <span class="kw">with</span> broadcasting. |
| |
| `broadcast_plus` is an alias to the function `broadcast_add`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_add(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| broadcast_plus(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_add(csr, dense(<span class="num">1</span>D)) = dense |
| broadcast_add(dense(<span class="num">1</span>D), csr) = dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L57</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_power" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_power(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_power(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_power</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_power(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns result of first array elements raised to powers from second array, element-wise <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_power(x, y) = `[ [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>, <span class="num">4.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L44</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_power" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_power(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_power(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_power</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_power(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns result of first array elements raised to powers from second array, element-wise <span class="kw">with</span> broadcasting. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_power(x, y) = `[ [ <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>, <span class="num">4.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_extended.cc:L44</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_sub" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_sub(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_sub(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_sub</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_sub(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise difference of the input arrays <span class="kw">with</span> broadcasting. |
| |
| `broadcast_minus` is an alias to the function `broadcast_sub`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_sub(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| broadcast_minus(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_sub/minus(csr, dense(<span class="num">1</span>D)) = dense |
| broadcast_sub/minus(dense(<span class="num">1</span>D), csr) = dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L105</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_sub" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_sub(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_sub(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_sub</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_sub(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise difference of the input arrays <span class="kw">with</span> broadcasting. |
| |
| `broadcast_minus` is an alias to the function `broadcast_sub`. |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| y = `[ [ <span class="num">0.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| broadcast_sub(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| broadcast_minus(x, y) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| Supported sparse operations: |
| |
| broadcast_sub/minus(csr, dense(<span class="num">1</span>D)) = dense |
| broadcast_sub/minus(dense(<span class="num">1</span>D), csr) = dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_broadcast_op_basic.cc:L105</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_to" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_to(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_to(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_to</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_to(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Broadcasts the input array to a <span class="kw">new</span> shape. |
| |
| Broadcasting is a mechanism that allows NDArrays to perform arithmetic operations |
| <span class="kw">with</span> arrays of different shapes efficiently without creating multiple copies of arrays. |
| Also see, `Broadcasting <https:<span class="cmt">//docs.scipy.org/doc/numpy/user/basics.broadcasting.html>`_ for more explanation.</span> |
| |
| Broadcasting is allowed on axes <span class="kw">with</span> size <span class="num">1</span>, such as from `(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">1</span>)` to |
| `(<span class="num">2</span>,<span class="num">8</span>,<span class="num">3</span>,<span class="num">9</span>)`. Elements will be duplicated on the broadcasted axes. |
| |
| For example:: |
| |
| broadcast_to(`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ], shape=(<span class="num">2</span>,<span class="num">3</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] ]) |
| |
| The dimension which you <span class="kw">do</span> not want to change can also be kept as `<span class="num">0</span>` which means copy the original value. |
| So <span class="kw">with</span> `shape=(<span class="num">2</span>,<span class="num">0</span>)`, we will obtain the same result as in the above example. |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L116</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#broadcast_to" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="broadcast_to(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="broadcast_to(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">broadcast_to</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@broadcast_to(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Broadcasts the input array to a <span class="kw">new</span> shape. |
| |
| Broadcasting is a mechanism that allows NDArrays to perform arithmetic operations |
| <span class="kw">with</span> arrays of different shapes efficiently without creating multiple copies of arrays. |
| Also see, `Broadcasting <https:<span class="cmt">//docs.scipy.org/doc/numpy/user/basics.broadcasting.html>`_ for more explanation.</span> |
| |
| Broadcasting is allowed on axes <span class="kw">with</span> size <span class="num">1</span>, such as from `(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">1</span>)` to |
| `(<span class="num">2</span>,<span class="num">8</span>,<span class="num">3</span>,<span class="num">9</span>)`. Elements will be duplicated on the broadcasted axes. |
| |
| For example:: |
| |
| broadcast_to(`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ], shape=(<span class="num">2</span>,<span class="num">3</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] ]) |
| |
| The dimension which you <span class="kw">do</span> not want to change can also be kept as `<span class="num">0</span>` which means copy the original value. |
| So <span class="kw">with</span> `shape=(<span class="num">2</span>,<span class="num">0</span>)`, we will obtain the same result as in the above example. |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_value.cc:L116</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cast(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cast(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cast</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cast(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Casts all elements of the input to a <span class="kw">new</span> <span class="kw">type</span>. |
| |
| .. note:: ``Cast`` is deprecated. Use ``cast`` instead. |
| |
| Example:: |
| |
| cast([<span class="num">0.9</span>, <span class="num">1.3</span>], dtype=<span class="lit">'int32'</span>) = [<span class="num">0</span>, <span class="num">1</span>] |
| cast([<span class="num">1</span>e20, <span class="num">11.1</span>], dtype='float16') = [inf, <span class="num">11.09375</span>] |
| cast([<span class="num">300</span>, <span class="num">11.1</span>, <span class="num">10.9</span>, -<span class="num">1</span>, -<span class="num">3</span>], dtype=<span class="lit">'uint8'</span>) = [<span class="num">44</span>, <span class="num">11</span>, <span class="num">10</span>, <span class="num">255</span>, <span class="num">253</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L664</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cast" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cast(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cast</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cast(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Casts all elements of the input to a <span class="kw">new</span> <span class="kw">type</span>. |
| |
| .. note:: ``Cast`` is deprecated. Use ``cast`` instead. |
| |
| Example:: |
| |
| cast([<span class="num">0.9</span>, <span class="num">1.3</span>], dtype=<span class="lit">'int32'</span>) = [<span class="num">0</span>, <span class="num">1</span>] |
| cast([<span class="num">1</span>e20, <span class="num">11.1</span>], dtype='float16') = [inf, <span class="num">11.09375</span>] |
| cast([<span class="num">300</span>, <span class="num">11.1</span>, <span class="num">10.9</span>, -<span class="num">1</span>, -<span class="num">3</span>], dtype=<span class="lit">'uint8'</span>) = [<span class="num">44</span>, <span class="num">11</span>, <span class="num">10</span>, <span class="num">255</span>, <span class="num">253</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L664</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cast_storage" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cast_storage(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cast_storage(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cast_storage</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cast_storage(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Casts tensor storage <span class="kw">type</span> to the <span class="kw">new</span> <span class="kw">type</span>. |
| |
| When an NDArray <span class="kw">with</span> default storage <span class="kw">type</span> is cast to csr or row_sparse storage, |
| the result is compact, which means: |
| |
| - <span class="kw">for</span> csr, zero values will not be retained |
| - <span class="kw">for</span> row_sparse, row slices of all zeros will not be retained |
| |
| The storage <span class="kw">type</span> of ``cast_storage`` output depends on stype parameter: |
| |
| - cast_storage(csr, 'default') = default |
| - cast_storage(row_sparse, 'default') = default |
| - cast_storage(default, <span class="lit">'csr'</span>) = csr |
| - cast_storage(default, 'row_sparse') = row_sparse |
| - cast_storage(csr, <span class="lit">'csr'</span>) = csr |
| - cast_storage(row_sparse, 'row_sparse') = row_sparse |
| |
| Example:: |
| |
| dense = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">2.</span>, <span class="num">0.</span>, <span class="num">3.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| # cast to row_sparse storage <span class="kw">type</span> |
| rsp = cast_storage(dense, 'row_sparse') |
| rsp.indices = [<span class="num">0</span>, <span class="num">1</span>] |
| rsp.values = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">2.</span>, <span class="num">0.</span>, <span class="num">3.</span>] ] |
| |
| # cast to csr storage <span class="kw">type</span> |
| csr = cast_storage(dense, <span class="lit">'csr'</span>) |
| csr.indices = [<span class="num">1</span>, <span class="num">0</span>, <span class="num">2</span>] |
| csr.values = [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| csr.indptr = [<span class="num">0</span>, <span class="num">1</span>, <span class="num">3</span>, <span class="num">3</span>, <span class="num">3</span>] |
| |
| |
| |
| Defined in src/operator/tensor/cast_storage.cc:L71</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cast_storage" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cast_storage(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cast_storage(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cast_storage</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cast_storage(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Casts tensor storage <span class="kw">type</span> to the <span class="kw">new</span> <span class="kw">type</span>. |
| |
| When an NDArray <span class="kw">with</span> default storage <span class="kw">type</span> is cast to csr or row_sparse storage, |
| the result is compact, which means: |
| |
| - <span class="kw">for</span> csr, zero values will not be retained |
| - <span class="kw">for</span> row_sparse, row slices of all zeros will not be retained |
| |
| The storage <span class="kw">type</span> of ``cast_storage`` output depends on stype parameter: |
| |
| - cast_storage(csr, 'default') = default |
| - cast_storage(row_sparse, 'default') = default |
| - cast_storage(default, <span class="lit">'csr'</span>) = csr |
| - cast_storage(default, 'row_sparse') = row_sparse |
| - cast_storage(csr, <span class="lit">'csr'</span>) = csr |
| - cast_storage(row_sparse, 'row_sparse') = row_sparse |
| |
| Example:: |
| |
| dense = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">2.</span>, <span class="num">0.</span>, <span class="num">3.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| # cast to row_sparse storage <span class="kw">type</span> |
| rsp = cast_storage(dense, 'row_sparse') |
| rsp.indices = [<span class="num">0</span>, <span class="num">1</span>] |
| rsp.values = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">2.</span>, <span class="num">0.</span>, <span class="num">3.</span>] ] |
| |
| # cast to csr storage <span class="kw">type</span> |
| csr = cast_storage(dense, <span class="lit">'csr'</span>) |
| csr.indices = [<span class="num">1</span>, <span class="num">0</span>, <span class="num">2</span>] |
| csr.values = [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| csr.indptr = [<span class="num">0</span>, <span class="num">1</span>, <span class="num">3</span>, <span class="num">3</span>, <span class="num">3</span>] |
| |
| |
| |
| Defined in src/operator/tensor/cast_storage.cc:L71</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cbrt" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cbrt(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cbrt(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cbrt</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cbrt(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise cube-root value of the input. |
| |
| .. math:: |
| cbrt(x) = \sqrt[<span class="num">3</span>]{x} |
| |
| Example:: |
| |
| cbrt([<span class="num">1</span>, <span class="num">8</span>, -<span class="num">125</span>]) = [<span class="num">1</span>, <span class="num">2</span>, -<span class="num">5</span>] |
| |
| The storage <span class="kw">type</span> of ``cbrt`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - cbrt(default) = default |
| - cbrt(row_sparse) = row_sparse |
| - cbrt(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L270</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cbrt" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cbrt(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cbrt(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cbrt</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cbrt(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise cube-root value of the input. |
| |
| .. math:: |
| cbrt(x) = \sqrt[<span class="num">3</span>]{x} |
| |
| Example:: |
| |
| cbrt([<span class="num">1</span>, <span class="num">8</span>, -<span class="num">125</span>]) = [<span class="num">1</span>, <span class="num">2</span>, -<span class="num">5</span>] |
| |
| The storage <span class="kw">type</span> of ``cbrt`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - cbrt(default) = default |
| - cbrt(row_sparse) = row_sparse |
| - cbrt(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L270</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ceil" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ceil(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ceil(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ceil</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ceil(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise ceiling of the input. |
| |
| The ceil of the scalar x is the smallest integer i, such that i >= x. |
| |
| Example:: |
| |
| ceil([-<span class="num">2.1</span>, -<span class="num">1.9</span>, <span class="num">1.5</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, -<span class="num">1.</span>, <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| |
| The storage <span class="kw">type</span> of ``ceil`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - ceil(default) = default |
| - ceil(row_sparse) = row_sparse |
| - ceil(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L817</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ceil" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ceil(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ceil(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ceil</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ceil(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise ceiling of the input. |
| |
| The ceil of the scalar x is the smallest integer i, such that i >= x. |
| |
| Example:: |
| |
| ceil([-<span class="num">2.1</span>, -<span class="num">1.9</span>, <span class="num">1.5</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, -<span class="num">1.</span>, <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| |
| The storage <span class="kw">type</span> of ``ceil`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - ceil(default) = default |
| - ceil(row_sparse) = row_sparse |
| - ceil(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L817</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#choose_element_0index" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="choose_element_0index(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="choose_element_0index(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">choose_element_0index</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@choose_element_0index(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Picks elements from an input array according to the input indices along the given axis. |
| |
| Given an input array of shape ``(d0, d1)`` and indices of shape ``(i0,)``, the result will be |
| an output array of shape ``(i0,)`` <span class="kw">with</span>:: |
| |
| output[i] = input[i, indices[i] ] |
| |
| By default, <span class="kw">if</span> any index mentioned is too large, it is replaced by the index that addresses |
| the last element along an axis (the `clip` mode). |
| |
| This function supports n-dimensional input and (n-<span class="num">1</span>)-dimensional indices arrays. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] |
| |
| <span class="cmt">// picks elements with specified indices along axis 0</span> |
| pick(x, y=[<span class="num">0</span>,<span class="num">1</span>], <span class="num">0</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1</span> |
| pick(x, y=[<span class="num">0</span>,<span class="num">1</span>,<span class="num">0</span>], <span class="num">1</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>, <span class="num">5.</span>] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1 using 'wrap' mode</span> |
| <span class="cmt">// to place indicies that would normally be out of bounds</span> |
| pick(x, y=[<span class="num">2</span>,-<span class="num">1</span>,-<span class="num">2</span>], <span class="num">1</span>, mode=<span class="lit">'wrap'</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>, <span class="num">5.</span>] |
| |
| y = `[ [ <span class="num">1.</span>], |
| [ <span class="num">0.</span>], |
| [ <span class="num">2.</span>] ] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1 and dims are maintained</span> |
| pick(x, y, <span class="num">1</span>, keepdims=True) = `[ [ <span class="num">2.</span>], |
| [ <span class="num">3.</span>], |
| [ <span class="num">6.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L150</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#choose_element_0index" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="choose_element_0index(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="choose_element_0index(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">choose_element_0index</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@choose_element_0index(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Picks elements from an input array according to the input indices along the given axis. |
| |
| Given an input array of shape ``(d0, d1)`` and indices of shape ``(i0,)``, the result will be |
| an output array of shape ``(i0,)`` <span class="kw">with</span>:: |
| |
| output[i] = input[i, indices[i] ] |
| |
| By default, <span class="kw">if</span> any index mentioned is too large, it is replaced by the index that addresses |
| the last element along an axis (the `clip` mode). |
| |
| This function supports n-dimensional input and (n-<span class="num">1</span>)-dimensional indices arrays. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] |
| |
| <span class="cmt">// picks elements with specified indices along axis 0</span> |
| pick(x, y=[<span class="num">0</span>,<span class="num">1</span>], <span class="num">0</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1</span> |
| pick(x, y=[<span class="num">0</span>,<span class="num">1</span>,<span class="num">0</span>], <span class="num">1</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>, <span class="num">5.</span>] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1 using 'wrap' mode</span> |
| <span class="cmt">// to place indicies that would normally be out of bounds</span> |
| pick(x, y=[<span class="num">2</span>,-<span class="num">1</span>,-<span class="num">2</span>], <span class="num">1</span>, mode=<span class="lit">'wrap'</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>, <span class="num">5.</span>] |
| |
| y = `[ [ <span class="num">1.</span>], |
| [ <span class="num">0.</span>], |
| [ <span class="num">2.</span>] ] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1 and dims are maintained</span> |
| pick(x, y, <span class="num">1</span>, keepdims=True) = `[ [ <span class="num">2.</span>], |
| [ <span class="num">3.</span>], |
| [ <span class="num">6.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L150</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#clip" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="clip(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="clip(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">clip</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@clip(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Clips (limits) the values in an array. |
| Given an interval, values outside the interval are clipped to the interval edges. |
| Clipping ``x`` between `a_min` and `a_max` would be:: |
| .. math:: |
| clip(x, a_min, a_max) = \max(\min(x, a_max), a_min)) |
| Example:: |
| x = [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>, <span class="num">7</span>, <span class="num">8</span>, <span class="num">9</span>] |
| clip(x,<span class="num">1</span>,<span class="num">8</span>) = [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">8.</span>] |
| The storage <span class="kw">type</span> of ``clip`` output depends on storage types of inputs and the a_min, a_max \ |
| parameter values: |
| - clip(default) = default |
| - clip(row_sparse, a_min <= <span class="num">0</span>, a_max >= <span class="num">0</span>) = row_sparse |
| - clip(csr, a_min <= <span class="num">0</span>, a_max >= <span class="num">0</span>) = csr |
| - clip(row_sparse, a_min < <span class="num">0</span>, a_max < <span class="num">0</span>) = default |
| - clip(row_sparse, a_min > <span class="num">0</span>, a_max > <span class="num">0</span>) = default |
| - clip(csr, a_min < <span class="num">0</span>, a_max < <span class="num">0</span>) = csr |
| - clip(csr, a_min > <span class="num">0</span>, a_max > <span class="num">0</span>) = csr |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L676</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#clip" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="clip(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="clip(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">clip</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@clip(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Clips (limits) the values in an array. |
| Given an interval, values outside the interval are clipped to the interval edges. |
| Clipping ``x`` between `a_min` and `a_max` would be:: |
| .. math:: |
| clip(x, a_min, a_max) = \max(\min(x, a_max), a_min)) |
| Example:: |
| x = [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>, <span class="num">7</span>, <span class="num">8</span>, <span class="num">9</span>] |
| clip(x,<span class="num">1</span>,<span class="num">8</span>) = [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">8.</span>] |
| The storage <span class="kw">type</span> of ``clip`` output depends on storage types of inputs and the a_min, a_max \ |
| parameter values: |
| - clip(default) = default |
| - clip(row_sparse, a_min <= <span class="num">0</span>, a_max >= <span class="num">0</span>) = row_sparse |
| - clip(csr, a_min <= <span class="num">0</span>, a_max >= <span class="num">0</span>) = csr |
| - clip(row_sparse, a_min < <span class="num">0</span>, a_max < <span class="num">0</span>) = default |
| - clip(row_sparse, a_min > <span class="num">0</span>, a_max > <span class="num">0</span>) = default |
| - clip(csr, a_min < <span class="num">0</span>, a_max < <span class="num">0</span>) = csr |
| - clip(csr, a_min > <span class="num">0</span>, a_max > <span class="num">0</span>) = csr |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L676</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#col2im" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="col2im(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="col2im(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">col2im</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@col2im(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Combining the output column matrix of im2col back to image array. |
| |
| Like :<span class="kw">class</span>:`~mxnet.ndarray.im2col`, <span class="kw">this</span> operator is also used in the vanilla convolution |
| implementation. Despite the name, col2im is not the reverse operation of im2col. Since there |
| may be overlaps between neighbouring sliding blocks, the column elements cannot be directly |
| put back into image. Instead, they are accumulated (i.e., summed) in the input image |
| just like the gradient computation, so col2im is the gradient of im2col and vice versa. |
| |
| Using the notation in im2col, given an input column array of shape |
| :math:`(N, C \times \prod(\text{kernel}), W)`, <span class="kw">this</span> operator accumulates the column elements |
| into output array of shape :math:`(N, C, \text{output_size}[<span class="num">0</span>], \text{output_size}[<span class="num">1</span>], \dots)`. |
| Only <span class="num">1</span>-D, <span class="num">2</span>-D and <span class="num">3</span>-D of spatial dimension is supported in <span class="kw">this</span> operator. |
| |
| |
| |
| Defined in src/operator/nn/im2col.cc:L181</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#col2im" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="col2im(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="col2im(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">col2im</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@col2im(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Combining the output column matrix of im2col back to image array. |
| |
| Like :<span class="kw">class</span>:`~mxnet.ndarray.im2col`, <span class="kw">this</span> operator is also used in the vanilla convolution |
| implementation. Despite the name, col2im is not the reverse operation of im2col. Since there |
| may be overlaps between neighbouring sliding blocks, the column elements cannot be directly |
| put back into image. Instead, they are accumulated (i.e., summed) in the input image |
| just like the gradient computation, so col2im is the gradient of im2col and vice versa. |
| |
| Using the notation in im2col, given an input column array of shape |
| :math:`(N, C \times \prod(\text{kernel}), W)`, <span class="kw">this</span> operator accumulates the column elements |
| into output array of shape :math:`(N, C, \text{output_size}[<span class="num">0</span>], \text{output_size}[<span class="num">1</span>], \dots)`. |
| Only <span class="num">1</span>-D, <span class="num">2</span>-D and <span class="num">3</span>-D of spatial dimension is supported in <span class="kw">this</span> operator. |
| |
| |
| |
| Defined in src/operator/nn/im2col.cc:L181</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#concat" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="concat(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="concat(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">concat</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@concat(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Joins input arrays along a given axis. |
| |
| .. note:: `Concat` is deprecated. Use `concat` instead. |
| |
| The dimensions of the input arrays should be the same except the axis along |
| which they will be concatenated. |
| The dimension of the output array along the concatenated axis will be equal |
| to the sum of the corresponding dimensions of the input arrays. |
| |
| The storage <span class="kw">type</span> of ``concat`` output depends on storage types of inputs |
| |
| - concat(csr, csr, ..., csr, dim=<span class="num">0</span>) = csr |
| - otherwise, ``concat`` generates output <span class="kw">with</span> default storage |
| |
| Example:: |
| |
| x = `[ [<span class="num">1</span>,<span class="num">1</span>],[<span class="num">2</span>,<span class="num">2</span>] ] |
| y = `[ [<span class="num">3</span>,<span class="num">3</span>],[<span class="num">4</span>,<span class="num">4</span>],[<span class="num">5</span>,<span class="num">5</span>] ] |
| z = `[ [<span class="num">6</span>,<span class="num">6</span>], [<span class="num">7</span>,<span class="num">7</span>],[<span class="num">8</span>,<span class="num">8</span>] ] |
| |
| concat(x,y,z,dim=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>], |
| [ <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">7.</span>], |
| [ <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| Note that you cannot concat x,y,z along dimension <span class="num">1</span> since dimension |
| <span class="num">0</span> is not the same <span class="kw">for</span> all the input arrays. |
| |
| concat(y,z,dim=<span class="num">1</span>) = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>, <span class="num">7.</span>, <span class="num">7.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/concat.cc:L384</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#concat" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="concat(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="concat(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">concat</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@concat(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Joins input arrays along a given axis. |
| |
| .. note:: `Concat` is deprecated. Use `concat` instead. |
| |
| The dimensions of the input arrays should be the same except the axis along |
| which they will be concatenated. |
| The dimension of the output array along the concatenated axis will be equal |
| to the sum of the corresponding dimensions of the input arrays. |
| |
| The storage <span class="kw">type</span> of ``concat`` output depends on storage types of inputs |
| |
| - concat(csr, csr, ..., csr, dim=<span class="num">0</span>) = csr |
| - otherwise, ``concat`` generates output <span class="kw">with</span> default storage |
| |
| Example:: |
| |
| x = `[ [<span class="num">1</span>,<span class="num">1</span>],[<span class="num">2</span>,<span class="num">2</span>] ] |
| y = `[ [<span class="num">3</span>,<span class="num">3</span>],[<span class="num">4</span>,<span class="num">4</span>],[<span class="num">5</span>,<span class="num">5</span>] ] |
| z = `[ [<span class="num">6</span>,<span class="num">6</span>], [<span class="num">7</span>,<span class="num">7</span>],[<span class="num">8</span>,<span class="num">8</span>] ] |
| |
| concat(x,y,z,dim=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">3.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>], |
| [ <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">7.</span>], |
| [ <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| Note that you cannot concat x,y,z along dimension <span class="num">1</span> since dimension |
| <span class="num">0</span> is not the same <span class="kw">for</span> all the input arrays. |
| |
| concat(y,z,dim=<span class="num">1</span>) = `[ [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">6.</span>, <span class="num">6.</span>], |
| [ <span class="num">4.</span>, <span class="num">4.</span>, <span class="num">7.</span>, <span class="num">7.</span>], |
| [ <span class="num">5.</span>, <span class="num">5.</span>, <span class="num">8.</span>, <span class="num">8.</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/concat.cc:L384</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cos" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cos(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cos(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cos</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cos(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the element-wise cosine of the input array. |
| |
| The input should be in radians (:math:`<span class="num">2</span>\pi` rad equals <span class="num">360</span> degrees). |
| |
| .. math:: |
| cos([<span class="num">0</span>, \pi/<span class="num">4</span>, \pi/<span class="num">2</span>]) = [<span class="num">1</span>, <span class="num">0.707</span>, <span class="num">0</span>] |
| |
| The storage <span class="kw">type</span> of ``cos`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L90</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cos" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cos(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cos(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cos</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cos(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the element-wise cosine of the input array. |
| |
| The input should be in radians (:math:`<span class="num">2</span>\pi` rad equals <span class="num">360</span> degrees). |
| |
| .. math:: |
| cos([<span class="num">0</span>, \pi/<span class="num">4</span>, \pi/<span class="num">2</span>]) = [<span class="num">1</span>, <span class="num">0.707</span>, <span class="num">0</span>] |
| |
| The storage <span class="kw">type</span> of ``cos`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L90</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cosh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cosh(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cosh(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cosh</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cosh(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the hyperbolic cosine of the input array, computed element-wise. |
| |
| .. math:: |
| cosh(x) = <span class="num">0.5</span>\times(exp(x) + exp(-x)) |
| |
| The storage <span class="kw">type</span> of ``cosh`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L409</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cosh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cosh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cosh(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cosh</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cosh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the hyperbolic cosine of the input array, computed element-wise. |
| |
| .. math:: |
| cosh(x) = <span class="num">0.5</span>\times(exp(x) + exp(-x)) |
| |
| The storage <span class="kw">type</span> of ``cosh`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L409</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#crop" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="crop(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="crop(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">crop</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@crop(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Slices a region of the array. |
| .. note:: ``crop`` is deprecated. Use ``slice`` instead. |
| This function returns a sliced array between the indices given |
| by `begin` and `end` <span class="kw">with</span> the corresponding `step`. |
| For an input array of ``shape=(d_0, d_1, ..., d_n-<span class="num">1</span>)``, |
| slice operation <span class="kw">with</span> ``begin=(b_0, b_1...b_m-<span class="num">1</span>)``, |
| ``end=(e_0, e_1, ..., e_m-<span class="num">1</span>)``, and ``step=(s_0, s_1, ..., s_m-<span class="num">1</span>)``, |
| where m <= n, results in an array <span class="kw">with</span> the shape |
| ``(|e_0-b_0|/|s_0|, ..., |e_m-<span class="num">1</span>-b_m-<span class="num">1</span>|/|s_m-<span class="num">1</span>|, d_m, ..., d_n-<span class="num">1</span>)``. |
| The resulting array's *k*-th dimension contains elements |
| from the *k*-th dimension of the input array starting |
| from index ``b_k`` (inclusive) <span class="kw">with</span> step ``s_k`` |
| until reaching ``e_k`` (exclusive). |
| If the *k*-th elements are `<span class="std">None</span>` in the sequence of `begin`, `end`, |
| and `step`, the following rule will be used to set default values. |
| If `s_k` is `<span class="std">None</span>`, set `s_k=<span class="num">1</span>`. If `s_k > <span class="num">0</span>`, set `b_k=<span class="num">0</span>`, `e_k=d_k`; |
| <span class="kw">else</span>, set `b_k=d_k-<span class="num">1</span>`, `e_k=-<span class="num">1</span>`. |
| The storage <span class="kw">type</span> of ``slice`` output depends on storage types of inputs |
| - slice(csr) = csr |
| - otherwise, ``slice`` generates output <span class="kw">with</span> default storage |
| .. note:: When input data storage <span class="kw">type</span> is csr, it only supports |
| step=(), or step=(<span class="std">None</span>,), or step=(<span class="num">1</span>,) to generate a csr output. |
| For other step parameter values, it falls back to slicing |
| a dense tensor. |
| Example:: |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| slice(x, begin=(<span class="num">0</span>,<span class="num">1</span>), end=(<span class="num">2</span>,<span class="num">4</span>)) = `[ [ <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>] ] |
| slice(x, begin=(<span class="std">None</span>, <span class="num">0</span>), end=(<span class="std">None</span>, <span class="num">3</span>), step=(-<span class="num">1</span>, <span class="num">2</span>)) = `[ [<span class="num">9.</span>, <span class="num">11.</span>], |
| [<span class="num">5.</span>, <span class="num">7.</span>], |
| [<span class="num">1.</span>, <span class="num">3.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L481</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#crop" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="crop(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="crop(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">crop</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@crop(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Slices a region of the array. |
| .. note:: ``crop`` is deprecated. Use ``slice`` instead. |
| This function returns a sliced array between the indices given |
| by `begin` and `end` <span class="kw">with</span> the corresponding `step`. |
| For an input array of ``shape=(d_0, d_1, ..., d_n-<span class="num">1</span>)``, |
| slice operation <span class="kw">with</span> ``begin=(b_0, b_1...b_m-<span class="num">1</span>)``, |
| ``end=(e_0, e_1, ..., e_m-<span class="num">1</span>)``, and ``step=(s_0, s_1, ..., s_m-<span class="num">1</span>)``, |
| where m <= n, results in an array <span class="kw">with</span> the shape |
| ``(|e_0-b_0|/|s_0|, ..., |e_m-<span class="num">1</span>-b_m-<span class="num">1</span>|/|s_m-<span class="num">1</span>|, d_m, ..., d_n-<span class="num">1</span>)``. |
| The resulting array's *k*-th dimension contains elements |
| from the *k*-th dimension of the input array starting |
| from index ``b_k`` (inclusive) <span class="kw">with</span> step ``s_k`` |
| until reaching ``e_k`` (exclusive). |
| If the *k*-th elements are `<span class="std">None</span>` in the sequence of `begin`, `end`, |
| and `step`, the following rule will be used to set default values. |
| If `s_k` is `<span class="std">None</span>`, set `s_k=<span class="num">1</span>`. If `s_k > <span class="num">0</span>`, set `b_k=<span class="num">0</span>`, `e_k=d_k`; |
| <span class="kw">else</span>, set `b_k=d_k-<span class="num">1</span>`, `e_k=-<span class="num">1</span>`. |
| The storage <span class="kw">type</span> of ``slice`` output depends on storage types of inputs |
| - slice(csr) = csr |
| - otherwise, ``slice`` generates output <span class="kw">with</span> default storage |
| .. note:: When input data storage <span class="kw">type</span> is csr, it only supports |
| step=(), or step=(<span class="std">None</span>,), or step=(<span class="num">1</span>,) to generate a csr output. |
| For other step parameter values, it falls back to slicing |
| a dense tensor. |
| Example:: |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| slice(x, begin=(<span class="num">0</span>,<span class="num">1</span>), end=(<span class="num">2</span>,<span class="num">4</span>)) = `[ [ <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>] ] |
| slice(x, begin=(<span class="std">None</span>, <span class="num">0</span>), end=(<span class="std">None</span>, <span class="num">3</span>), step=(-<span class="num">1</span>, <span class="num">2</span>)) = `[ [<span class="num">9.</span>, <span class="num">11.</span>], |
| [<span class="num">5.</span>, <span class="num">7.</span>], |
| [<span class="num">1.</span>, <span class="num">3.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L481</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ctc_loss" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ctc_loss(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ctc_loss(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ctc_loss</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ctc_loss(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Connectionist Temporal Classification Loss. |
| |
| .. note:: The existing alias ``contrib_CTCLoss`` is deprecated. |
| |
| The shapes of the inputs and outputs: |
| |
| - **data**: `(sequence_length, batch_size, alphabet_size)` |
| - **label**: `(batch_size, label_sequence_length)` |
| - **out**: `(batch_size)` |
| |
| The `data` tensor consists of sequences of activation vectors (without applying softmax), |
| <span class="kw">with</span> i-th channel in the last dimension corresponding to i-th label |
| <span class="kw">for</span> i between <span class="num">0</span> and alphabet_size-<span class="num">1</span> (i.e always <span class="num">0</span>-indexed). |
| Alphabet size should include one additional value reserved <span class="kw">for</span> blank label. |
| When `blank_label` is ``<span class="lit">"first"</span>``, the ``<span class="num">0</span>``-th channel is be reserved <span class="kw">for</span> |
| activation of blank label, or otherwise <span class="kw">if</span> it is <span class="lit">"last"</span>, ``(alphabet_size-<span class="num">1</span>)``-th channel should be |
| reserved <span class="kw">for</span> blank label. |
| |
| ``label`` is an index matrix of integers. When `blank_label` is ``<span class="lit">"first"</span>``, |
| the value <span class="num">0</span> is then reserved <span class="kw">for</span> blank label, and should not be passed in <span class="kw">this</span> matrix. Otherwise, |
| when `blank_label` is ``<span class="lit">"last"</span>``, the value `(alphabet_size-<span class="num">1</span>)` is reserved <span class="kw">for</span> blank label. |
| |
| If a sequence of labels is shorter than *label_sequence_length*, use the special |
| padding value at the end of the sequence to conform it to the correct |
| length. The padding value is `<span class="num">0</span>` when `blank_label` is ``<span class="lit">"first"</span>``, and `-<span class="num">1</span>` otherwise. |
| |
| For example, suppose the vocabulary is `[a, b, c]`, and in one batch we have three sequences |
| <span class="lit">'ba'</span>, <span class="lit">'cbb'</span>, and <span class="lit">'abac'</span>. When `blank_label` is ``<span class="lit">"first"</span>``, we can index the labels as |
| `{<span class="lit">'a'</span>: <span class="num">1</span>, <span class="lit">'b'</span>: <span class="num">2</span>, <span class="lit">'c'</span>: <span class="num">3</span>}`, and we reserve the <span class="num">0</span>-th channel <span class="kw">for</span> blank label in data tensor. |
| The resulting `label` tensor should be padded to be:: |
| |
| `[ [<span class="num">2</span>, <span class="num">1</span>, <span class="num">0</span>, <span class="num">0</span>], [<span class="num">3</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">0</span>], [<span class="num">1</span>, <span class="num">2</span>, <span class="num">1</span>, <span class="num">3</span>] ] |
| |
| When `blank_label` is ``<span class="lit">"last"</span>``, we can index the labels as |
| `{<span class="lit">'a'</span>: <span class="num">0</span>, <span class="lit">'b'</span>: <span class="num">1</span>, <span class="lit">'c'</span>: <span class="num">2</span>}`, and we reserve the channel index <span class="num">3</span> <span class="kw">for</span> blank label in data tensor. |
| The resulting `label` tensor should be padded to be:: |
| |
| `[ [<span class="num">1</span>, <span class="num">0</span>, -<span class="num">1</span>, -<span class="num">1</span>], [<span class="num">2</span>, <span class="num">1</span>, <span class="num">1</span>, -<span class="num">1</span>], [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>, <span class="num">2</span>] ] |
| |
| ``out`` is a list of CTC loss values, one per example in the batch. |
| |
| See *Connectionist Temporal Classification: Labelling Unsegmented |
| Sequence Data <span class="kw">with</span> Recurrent Neural Networks*, A. Graves *et al*. <span class="kw">for</span> more |
| information on the definition and the algorithm. |
| |
| |
| |
| Defined in src/operator/nn/ctc_loss.cc:L100</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ctc_loss" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ctc_loss(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ctc_loss(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ctc_loss</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ctc_loss(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Connectionist Temporal Classification Loss. |
| |
| .. note:: The existing alias ``contrib_CTCLoss`` is deprecated. |
| |
| The shapes of the inputs and outputs: |
| |
| - **data**: `(sequence_length, batch_size, alphabet_size)` |
| - **label**: `(batch_size, label_sequence_length)` |
| - **out**: `(batch_size)` |
| |
| The `data` tensor consists of sequences of activation vectors (without applying softmax), |
| <span class="kw">with</span> i-th channel in the last dimension corresponding to i-th label |
| <span class="kw">for</span> i between <span class="num">0</span> and alphabet_size-<span class="num">1</span> (i.e always <span class="num">0</span>-indexed). |
| Alphabet size should include one additional value reserved <span class="kw">for</span> blank label. |
| When `blank_label` is ``<span class="lit">"first"</span>``, the ``<span class="num">0</span>``-th channel is be reserved <span class="kw">for</span> |
| activation of blank label, or otherwise <span class="kw">if</span> it is <span class="lit">"last"</span>, ``(alphabet_size-<span class="num">1</span>)``-th channel should be |
| reserved <span class="kw">for</span> blank label. |
| |
| ``label`` is an index matrix of integers. When `blank_label` is ``<span class="lit">"first"</span>``, |
| the value <span class="num">0</span> is then reserved <span class="kw">for</span> blank label, and should not be passed in <span class="kw">this</span> matrix. Otherwise, |
| when `blank_label` is ``<span class="lit">"last"</span>``, the value `(alphabet_size-<span class="num">1</span>)` is reserved <span class="kw">for</span> blank label. |
| |
| If a sequence of labels is shorter than *label_sequence_length*, use the special |
| padding value at the end of the sequence to conform it to the correct |
| length. The padding value is `<span class="num">0</span>` when `blank_label` is ``<span class="lit">"first"</span>``, and `-<span class="num">1</span>` otherwise. |
| |
| For example, suppose the vocabulary is `[a, b, c]`, and in one batch we have three sequences |
| <span class="lit">'ba'</span>, <span class="lit">'cbb'</span>, and <span class="lit">'abac'</span>. When `blank_label` is ``<span class="lit">"first"</span>``, we can index the labels as |
| `{<span class="lit">'a'</span>: <span class="num">1</span>, <span class="lit">'b'</span>: <span class="num">2</span>, <span class="lit">'c'</span>: <span class="num">3</span>}`, and we reserve the <span class="num">0</span>-th channel <span class="kw">for</span> blank label in data tensor. |
| The resulting `label` tensor should be padded to be:: |
| |
| `[ [<span class="num">2</span>, <span class="num">1</span>, <span class="num">0</span>, <span class="num">0</span>], [<span class="num">3</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">0</span>], [<span class="num">1</span>, <span class="num">2</span>, <span class="num">1</span>, <span class="num">3</span>] ] |
| |
| When `blank_label` is ``<span class="lit">"last"</span>``, we can index the labels as |
| `{<span class="lit">'a'</span>: <span class="num">0</span>, <span class="lit">'b'</span>: <span class="num">1</span>, <span class="lit">'c'</span>: <span class="num">2</span>}`, and we reserve the channel index <span class="num">3</span> <span class="kw">for</span> blank label in data tensor. |
| The resulting `label` tensor should be padded to be:: |
| |
| `[ [<span class="num">1</span>, <span class="num">0</span>, -<span class="num">1</span>, -<span class="num">1</span>], [<span class="num">2</span>, <span class="num">1</span>, <span class="num">1</span>, -<span class="num">1</span>], [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>, <span class="num">2</span>] ] |
| |
| ``out`` is a list of CTC loss values, one per example in the batch. |
| |
| See *Connectionist Temporal Classification: Labelling Unsegmented |
| Sequence Data <span class="kw">with</span> Recurrent Neural Networks*, A. Graves *et al*. <span class="kw">for</span> more |
| information on the definition and the algorithm. |
| |
| |
| |
| Defined in src/operator/nn/ctc_loss.cc:L100</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cumsum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cumsum(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cumsum(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cumsum</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cumsum(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return the cumulative sum of the elements along a given axis. |
| |
| Defined in src/operator/numpy/np_cumsum.cc:L70</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#cumsum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="cumsum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="cumsum(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">cumsum</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@cumsum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return the cumulative sum of the elements along a given axis. |
| |
| Defined in src/operator/numpy/np_cumsum.cc:L70</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#degrees" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="degrees(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="degrees(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">degrees</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@degrees(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Converts each element of the input array from radians to degrees. |
| |
| .. math:: |
| degrees([<span class="num">0</span>, \pi/<span class="num">2</span>, \pi, <span class="num">3</span>\pi/<span class="num">2</span>, <span class="num">2</span>\pi]) = [<span class="num">0</span>, <span class="num">90</span>, <span class="num">180</span>, <span class="num">270</span>, <span class="num">360</span>] |
| |
| The storage <span class="kw">type</span> of ``degrees`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - degrees(default) = default |
| - degrees(row_sparse) = row_sparse |
| - degrees(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L332</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#degrees" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="degrees(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="degrees(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">degrees</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@degrees(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Converts each element of the input array from radians to degrees. |
| |
| .. math:: |
| degrees([<span class="num">0</span>, \pi/<span class="num">2</span>, \pi, <span class="num">3</span>\pi/<span class="num">2</span>, <span class="num">2</span>\pi]) = [<span class="num">0</span>, <span class="num">90</span>, <span class="num">180</span>, <span class="num">270</span>, <span class="num">360</span>] |
| |
| The storage <span class="kw">type</span> of ``degrees`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - degrees(default) = default |
| - degrees(row_sparse) = row_sparse |
| - degrees(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L332</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#depth_to_space" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="depth_to_space(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="depth_to_space(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">depth_to_space</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@depth_to_space(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Rearranges(permutes) data from depth into blocks of spatial data. |
| Similar to ONNX DepthToSpace operator: |
| https:<span class="cmt">//github.com/onnx/onnx/blob/master/docs/Operators.md#DepthToSpace.</span> |
| The output is a <span class="kw">new</span> tensor where the values from depth dimension are moved in spatial blocks |
| to height and width dimension. The reverse of <span class="kw">this</span> operation is ``space_to_depth``. |
| .. math:: |
| \begin{gather*} |
| x \prime = reshape(x, [N, block\_size, block\_size, C / (block\_size ^ <span class="num">2</span>), H * block\_size, W * block\_size]) \\ |
| x \prime \prime = transpose(x \prime, [<span class="num">0</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">1</span>, <span class="num">5</span>, <span class="num">2</span>]) \\ |
| y = reshape(x \prime \prime, [N, C / (block\_size ^ <span class="num">2</span>), H * block\_size, W * block\_size]) |
| \end{gather*} |
| where :math:`x` is an input tensor <span class="kw">with</span> default layout as :math:`[N, C, H, W]`: [batch, channels, height, width] |
| and :math:`y` is the output tensor of layout :math:`[N, C / (block\_size ^ <span class="num">2</span>), H * block\_size, W * block\_size]` |
| Example:: |
| x = `[ [`[ [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>] ], |
| `[ [<span class="num">6</span>, <span class="num">7</span>, <span class="num">8</span>], |
| [<span class="num">9</span>, <span class="num">10</span>, <span class="num">11</span>] ], |
| `[ [<span class="num">12</span>, <span class="num">13</span>, <span class="num">14</span>], |
| [<span class="num">15</span>, <span class="num">16</span>, <span class="num">17</span>] ], |
| `[ [<span class="num">18</span>, <span class="num">19</span>, <span class="num">20</span>], |
| [<span class="num">21</span>, <span class="num">22</span>, <span class="num">23</span>] ] ] ] |
| depth_to_space(x, <span class="num">2</span>) = `[ [`[ [<span class="num">0</span>, <span class="num">6</span>, <span class="num">1</span>, <span class="num">7</span>, <span class="num">2</span>, <span class="num">8</span>], |
| [<span class="num">12</span>, <span class="num">18</span>, <span class="num">13</span>, <span class="num">19</span>, <span class="num">14</span>, <span class="num">20</span>], |
| [<span class="num">3</span>, <span class="num">9</span>, <span class="num">4</span>, <span class="num">10</span>, <span class="num">5</span>, <span class="num">11</span>], |
| [<span class="num">15</span>, <span class="num">21</span>, <span class="num">16</span>, <span class="num">22</span>, <span class="num">17</span>, <span class="num">23</span>] ] ] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L971</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#depth_to_space" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="depth_to_space(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="depth_to_space(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">depth_to_space</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@depth_to_space(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Rearranges(permutes) data from depth into blocks of spatial data. |
| Similar to ONNX DepthToSpace operator: |
| https:<span class="cmt">//github.com/onnx/onnx/blob/master/docs/Operators.md#DepthToSpace.</span> |
| The output is a <span class="kw">new</span> tensor where the values from depth dimension are moved in spatial blocks |
| to height and width dimension. The reverse of <span class="kw">this</span> operation is ``space_to_depth``. |
| .. math:: |
| \begin{gather*} |
| x \prime = reshape(x, [N, block\_size, block\_size, C / (block\_size ^ <span class="num">2</span>), H * block\_size, W * block\_size]) \\ |
| x \prime \prime = transpose(x \prime, [<span class="num">0</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">1</span>, <span class="num">5</span>, <span class="num">2</span>]) \\ |
| y = reshape(x \prime \prime, [N, C / (block\_size ^ <span class="num">2</span>), H * block\_size, W * block\_size]) |
| \end{gather*} |
| where :math:`x` is an input tensor <span class="kw">with</span> default layout as :math:`[N, C, H, W]`: [batch, channels, height, width] |
| and :math:`y` is the output tensor of layout :math:`[N, C / (block\_size ^ <span class="num">2</span>), H * block\_size, W * block\_size]` |
| Example:: |
| x = `[ [`[ [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>] ], |
| `[ [<span class="num">6</span>, <span class="num">7</span>, <span class="num">8</span>], |
| [<span class="num">9</span>, <span class="num">10</span>, <span class="num">11</span>] ], |
| `[ [<span class="num">12</span>, <span class="num">13</span>, <span class="num">14</span>], |
| [<span class="num">15</span>, <span class="num">16</span>, <span class="num">17</span>] ], |
| `[ [<span class="num">18</span>, <span class="num">19</span>, <span class="num">20</span>], |
| [<span class="num">21</span>, <span class="num">22</span>, <span class="num">23</span>] ] ] ] |
| depth_to_space(x, <span class="num">2</span>) = `[ [`[ [<span class="num">0</span>, <span class="num">6</span>, <span class="num">1</span>, <span class="num">7</span>, <span class="num">2</span>, <span class="num">8</span>], |
| [<span class="num">12</span>, <span class="num">18</span>, <span class="num">13</span>, <span class="num">19</span>, <span class="num">14</span>, <span class="num">20</span>], |
| [<span class="num">3</span>, <span class="num">9</span>, <span class="num">4</span>, <span class="num">10</span>, <span class="num">5</span>, <span class="num">11</span>], |
| [<span class="num">15</span>, <span class="num">21</span>, <span class="num">16</span>, <span class="num">22</span>, <span class="num">17</span>, <span class="num">23</span>] ] ] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L971</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#diag" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="diag(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="diag(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">diag</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@diag(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Extracts a diagonal or constructs a diagonal array. |
| |
| ``diag``'s behavior depends on the input array dimensions: |
| |
| - <span class="num">1</span>-D arrays: constructs a <span class="num">2</span>-D array <span class="kw">with</span> the input as its diagonal, all other elements are zero. |
| - N-D arrays: extracts the diagonals of the sub-arrays <span class="kw">with</span> axes specified by ``axis1`` and ``axis2``. |
| The output shape would be decided by removing the axes numbered ``axis1`` and ``axis2`` from the |
| input shape and appending to the result a <span class="kw">new</span> axis <span class="kw">with</span> the size of the diagonals in question. |
| |
| For example, when the input shape is `(<span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>)`, ``axis1`` and ``axis2`` are <span class="num">0</span> and <span class="num">2</span> |
| respectively and ``k`` is <span class="num">0</span>, the resulting shape would be `(<span class="num">3</span>, <span class="num">5</span>, <span class="num">2</span>)`. |
| |
| Examples:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], |
| [<span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>] ] |
| |
| diag(x) = [<span class="num">1</span>, <span class="num">5</span>] |
| |
| diag(x, k=<span class="num">1</span>) = [<span class="num">2</span>, <span class="num">6</span>] |
| |
| diag(x, k=-<span class="num">1</span>) = [<span class="num">4</span>] |
| |
| x = [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] |
| |
| diag(x) = `[ [<span class="num">1</span>, <span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">2</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>, <span class="num">3</span>] ] |
| |
| diag(x, k=<span class="num">1</span>) = `[ [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>, <span class="num">2</span>], |
| [<span class="num">0</span>, <span class="num">0</span>, <span class="num">0</span>] ] |
| |
| diag(x, k=-<span class="num">1</span>) = `[ [<span class="num">0</span>, <span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">1</span>, <span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">2</span>, <span class="num">0</span>] ] |
| |
| x = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ], |
| |
| `[ [<span class="num">5</span>, <span class="num">6</span>], |
| [<span class="num">7</span>, <span class="num">8</span>] ] ] |
| |
| diag(x) = `[ [<span class="num">1</span>, <span class="num">7</span>], |
| [<span class="num">2</span>, <span class="num">8</span>] ] |
| |
| diag(x, k=<span class="num">1</span>) = `[ [<span class="num">3</span>], |
| [<span class="num">4</span>] ] |
| |
| diag(x, axis1=-<span class="num">2</span>, axis2=-<span class="num">1</span>) = `[ [<span class="num">1</span>, <span class="num">4</span>], |
| [<span class="num">5</span>, <span class="num">8</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/diag_op.cc:L86</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#diag" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="diag(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="diag(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">diag</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@diag(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Extracts a diagonal or constructs a diagonal array. |
| |
| ``diag``'s behavior depends on the input array dimensions: |
| |
| - <span class="num">1</span>-D arrays: constructs a <span class="num">2</span>-D array <span class="kw">with</span> the input as its diagonal, all other elements are zero. |
| - N-D arrays: extracts the diagonals of the sub-arrays <span class="kw">with</span> axes specified by ``axis1`` and ``axis2``. |
| The output shape would be decided by removing the axes numbered ``axis1`` and ``axis2`` from the |
| input shape and appending to the result a <span class="kw">new</span> axis <span class="kw">with</span> the size of the diagonals in question. |
| |
| For example, when the input shape is `(<span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>)`, ``axis1`` and ``axis2`` are <span class="num">0</span> and <span class="num">2</span> |
| respectively and ``k`` is <span class="num">0</span>, the resulting shape would be `(<span class="num">3</span>, <span class="num">5</span>, <span class="num">2</span>)`. |
| |
| Examples:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], |
| [<span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>] ] |
| |
| diag(x) = [<span class="num">1</span>, <span class="num">5</span>] |
| |
| diag(x, k=<span class="num">1</span>) = [<span class="num">2</span>, <span class="num">6</span>] |
| |
| diag(x, k=-<span class="num">1</span>) = [<span class="num">4</span>] |
| |
| x = [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] |
| |
| diag(x) = `[ [<span class="num">1</span>, <span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">2</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>, <span class="num">3</span>] ] |
| |
| diag(x, k=<span class="num">1</span>) = `[ [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>, <span class="num">2</span>], |
| [<span class="num">0</span>, <span class="num">0</span>, <span class="num">0</span>] ] |
| |
| diag(x, k=-<span class="num">1</span>) = `[ [<span class="num">0</span>, <span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">1</span>, <span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">2</span>, <span class="num">0</span>] ] |
| |
| x = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ], |
| |
| `[ [<span class="num">5</span>, <span class="num">6</span>], |
| [<span class="num">7</span>, <span class="num">8</span>] ] ] |
| |
| diag(x) = `[ [<span class="num">1</span>, <span class="num">7</span>], |
| [<span class="num">2</span>, <span class="num">8</span>] ] |
| |
| diag(x, k=<span class="num">1</span>) = `[ [<span class="num">3</span>], |
| [<span class="num">4</span>] ] |
| |
| diag(x, axis1=-<span class="num">2</span>, axis2=-<span class="num">1</span>) = `[ [<span class="num">1</span>, <span class="num">4</span>], |
| [<span class="num">5</span>, <span class="num">8</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/diag_op.cc:L86</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#dot" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="dot(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="dot(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">dot</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@dot(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Dot product of two arrays. |
| |
| ``dot``'s behavior depends on the input array dimensions: |
| |
| - <span class="num">1</span>-D arrays: inner product of vectors |
| - <span class="num">2</span>-D arrays: matrix multiplication |
| - N-D arrays: a sum product over the last axis of the first input and the first |
| axis of the second input |
| |
| For example, given <span class="num">3</span>-D ``x`` <span class="kw">with</span> shape `(n,m,k)` and ``y`` <span class="kw">with</span> shape `(k,r,s)`, the |
| result array will have shape `(n,m,r,s)`. It is computed by:: |
| |
| dot(x,y)[i,j,a,b] = sum(x[i,j,:]*y[:,a,b]) |
| |
| Example:: |
| |
| x = reshape([<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>], shape=(<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>)) |
| y = reshape([<span class="num">7</span>,<span class="num">6</span>,<span class="num">5</span>,<span class="num">4</span>,<span class="num">3</span>,<span class="num">2</span>,<span class="num">1</span>,<span class="num">0</span>], shape=(<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>)) |
| dot(x,y)[<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>] = <span class="num">0</span> |
| sum(x[<span class="num">0</span>,<span class="num">0</span>,:]*y[:,<span class="num">1</span>,<span class="num">1</span>]) = <span class="num">0</span> |
| |
| The storage <span class="kw">type</span> of ``dot`` output depends on storage types of inputs, transpose option and |
| forward_stype option <span class="kw">for</span> output storage <span class="kw">type</span>. Implemented sparse operations include: |
| |
| - dot(default, default, transpose_a=True/False, transpose_b=True/False) = default |
| - dot(csr, default, transpose_a=True) = default |
| - dot(csr, default, transpose_a=True) = row_sparse |
| - dot(csr, default) = default |
| - dot(csr, row_sparse) = default |
| - dot(default, csr) = csr (CPU only) |
| - dot(default, csr, forward_stype='default') = default |
| - dot(default, csr, transpose_b=True, forward_stype='default') = default |
| |
| If the combination of input storage types and forward_stype does not <span class="kw">match</span> any of the |
| above patterns, ``dot`` will fallback and generate output <span class="kw">with</span> default storage. |
| |
| .. Note:: |
| |
| If the storage <span class="kw">type</span> of the lhs is <span class="lit">"csr"</span>, the storage <span class="kw">type</span> of gradient w.r.t rhs will be |
| <span class="lit">"row_sparse"</span>. Only a subset of optimizers support sparse gradients, including SGD, AdaGrad |
| and Adam. Note that by default <span class="kw">lazy</span> updates is turned on, which may perform differently |
| from standard updates. For more details, please check the Optimization API at: |
| https:<span class="cmt">//mxnet.incubator.apache.org/api/python/optimization/optimization.html</span> |
| |
| |
| |
| Defined in src/operator/tensor/dot.cc:L77</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#dot" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="dot(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="dot(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">dot</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@dot(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Dot product of two arrays. |
| |
| ``dot``'s behavior depends on the input array dimensions: |
| |
| - <span class="num">1</span>-D arrays: inner product of vectors |
| - <span class="num">2</span>-D arrays: matrix multiplication |
| - N-D arrays: a sum product over the last axis of the first input and the first |
| axis of the second input |
| |
| For example, given <span class="num">3</span>-D ``x`` <span class="kw">with</span> shape `(n,m,k)` and ``y`` <span class="kw">with</span> shape `(k,r,s)`, the |
| result array will have shape `(n,m,r,s)`. It is computed by:: |
| |
| dot(x,y)[i,j,a,b] = sum(x[i,j,:]*y[:,a,b]) |
| |
| Example:: |
| |
| x = reshape([<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>], shape=(<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>)) |
| y = reshape([<span class="num">7</span>,<span class="num">6</span>,<span class="num">5</span>,<span class="num">4</span>,<span class="num">3</span>,<span class="num">2</span>,<span class="num">1</span>,<span class="num">0</span>], shape=(<span class="num">2</span>,<span class="num">2</span>,<span class="num">2</span>)) |
| dot(x,y)[<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>] = <span class="num">0</span> |
| sum(x[<span class="num">0</span>,<span class="num">0</span>,:]*y[:,<span class="num">1</span>,<span class="num">1</span>]) = <span class="num">0</span> |
| |
| The storage <span class="kw">type</span> of ``dot`` output depends on storage types of inputs, transpose option and |
| forward_stype option <span class="kw">for</span> output storage <span class="kw">type</span>. Implemented sparse operations include: |
| |
| - dot(default, default, transpose_a=True/False, transpose_b=True/False) = default |
| - dot(csr, default, transpose_a=True) = default |
| - dot(csr, default, transpose_a=True) = row_sparse |
| - dot(csr, default) = default |
| - dot(csr, row_sparse) = default |
| - dot(default, csr) = csr (CPU only) |
| - dot(default, csr, forward_stype='default') = default |
| - dot(default, csr, transpose_b=True, forward_stype='default') = default |
| |
| If the combination of input storage types and forward_stype does not <span class="kw">match</span> any of the |
| above patterns, ``dot`` will fallback and generate output <span class="kw">with</span> default storage. |
| |
| .. Note:: |
| |
| If the storage <span class="kw">type</span> of the lhs is <span class="lit">"csr"</span>, the storage <span class="kw">type</span> of gradient w.r.t rhs will be |
| <span class="lit">"row_sparse"</span>. Only a subset of optimizers support sparse gradients, including SGD, AdaGrad |
| and Adam. Note that by default <span class="kw">lazy</span> updates is turned on, which may perform differently |
| from standard updates. For more details, please check the Optimization API at: |
| https:<span class="cmt">//mxnet.incubator.apache.org/api/python/optimization/optimization.html</span> |
| |
| |
| |
| Defined in src/operator/tensor/dot.cc:L77</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#elemwise_add" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="elemwise_add(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="elemwise_add(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">elemwise_add</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@elemwise_add(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Adds arguments element-wise. |
| |
| The storage <span class="kw">type</span> of ``elemwise_add`` output depends on storage types of inputs |
| |
| - elemwise_add(row_sparse, row_sparse) = row_sparse |
| - elemwise_add(csr, csr) = csr |
| - elemwise_add(default, csr) = default |
| - elemwise_add(csr, default) = default |
| - elemwise_add(default, rsp) = default |
| - elemwise_add(rsp, default) = default |
| - otherwise, ``elemwise_add`` generates output <span class="kw">with</span> default storage</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#elemwise_add" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="elemwise_add(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="elemwise_add(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">elemwise_add</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@elemwise_add(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Adds arguments element-wise. |
| |
| The storage <span class="kw">type</span> of ``elemwise_add`` output depends on storage types of inputs |
| |
| - elemwise_add(row_sparse, row_sparse) = row_sparse |
| - elemwise_add(csr, csr) = csr |
| - elemwise_add(default, csr) = default |
| - elemwise_add(csr, default) = default |
| - elemwise_add(default, rsp) = default |
| - elemwise_add(rsp, default) = default |
| - otherwise, ``elemwise_add`` generates output <span class="kw">with</span> default storage</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#elemwise_div" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="elemwise_div(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="elemwise_div(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">elemwise_div</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@elemwise_div(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Divides arguments element-wise. |
| |
| The storage <span class="kw">type</span> of ``elemwise_div`` output is always dense</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#elemwise_div" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="elemwise_div(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="elemwise_div(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">elemwise_div</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@elemwise_div(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Divides arguments element-wise. |
| |
| The storage <span class="kw">type</span> of ``elemwise_div`` output is always dense</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#elemwise_mul" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="elemwise_mul(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="elemwise_mul(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">elemwise_mul</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@elemwise_mul(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Multiplies arguments element-wise. |
| |
| The storage <span class="kw">type</span> of ``elemwise_mul`` output depends on storage types of inputs |
| |
| - elemwise_mul(default, default) = default |
| - elemwise_mul(row_sparse, row_sparse) = row_sparse |
| - elemwise_mul(default, row_sparse) = row_sparse |
| - elemwise_mul(row_sparse, default) = row_sparse |
| - elemwise_mul(csr, csr) = csr |
| - otherwise, ``elemwise_mul`` generates output <span class="kw">with</span> default storage</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#elemwise_mul" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="elemwise_mul(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="elemwise_mul(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">elemwise_mul</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@elemwise_mul(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Multiplies arguments element-wise. |
| |
| The storage <span class="kw">type</span> of ``elemwise_mul`` output depends on storage types of inputs |
| |
| - elemwise_mul(default, default) = default |
| - elemwise_mul(row_sparse, row_sparse) = row_sparse |
| - elemwise_mul(default, row_sparse) = row_sparse |
| - elemwise_mul(row_sparse, default) = row_sparse |
| - elemwise_mul(csr, csr) = csr |
| - otherwise, ``elemwise_mul`` generates output <span class="kw">with</span> default storage</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#elemwise_sub" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="elemwise_sub(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="elemwise_sub(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">elemwise_sub</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@elemwise_sub(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Subtracts arguments element-wise. |
| |
| The storage <span class="kw">type</span> of ``elemwise_sub`` output depends on storage types of inputs |
| |
| - elemwise_sub(row_sparse, row_sparse) = row_sparse |
| - elemwise_sub(csr, csr) = csr |
| - elemwise_sub(default, csr) = default |
| - elemwise_sub(csr, default) = default |
| - elemwise_sub(default, rsp) = default |
| - elemwise_sub(rsp, default) = default |
| - otherwise, ``elemwise_sub`` generates output <span class="kw">with</span> default storage</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#elemwise_sub" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="elemwise_sub(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="elemwise_sub(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">elemwise_sub</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@elemwise_sub(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Subtracts arguments element-wise. |
| |
| The storage <span class="kw">type</span> of ``elemwise_sub`` output depends on storage types of inputs |
| |
| - elemwise_sub(row_sparse, row_sparse) = row_sparse |
| - elemwise_sub(csr, csr) = csr |
| - elemwise_sub(default, csr) = default |
| - elemwise_sub(csr, default) = default |
| - elemwise_sub(default, rsp) = default |
| - elemwise_sub(rsp, default) = default |
| - otherwise, ``elemwise_sub`` generates output <span class="kw">with</span> default storage</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#erf" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="erf(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="erf(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">erf</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@erf(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise gauss error function of the input. |
| |
| Example:: |
| |
| erf([<span class="num">0</span>, -<span class="num">1.</span>, <span class="num">10.</span>]) = [<span class="num">0.</span>, -<span class="num">0.8427</span>, <span class="num">1.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L886</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#erf" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="erf(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="erf(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">erf</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@erf(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise gauss error function of the input. |
| |
| Example:: |
| |
| erf([<span class="num">0</span>, -<span class="num">1.</span>, <span class="num">10.</span>]) = [<span class="num">0.</span>, -<span class="num">0.8427</span>, <span class="num">1.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L886</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#erfinv" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="erfinv(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="erfinv(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">erfinv</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@erfinv(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse gauss error function of the input. |
| |
| Example:: |
| |
| erfinv([<span class="num">0</span>, <span class="num">0.5</span>., -<span class="num">1.</span>]) = [<span class="num">0.</span>, <span class="num">0.4769</span>, -inf] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L908</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#erfinv" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="erfinv(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="erfinv(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">erfinv</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@erfinv(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse gauss error function of the input. |
| |
| Example:: |
| |
| erfinv([<span class="num">0</span>, <span class="num">0.5</span>., -<span class="num">1.</span>]) = [<span class="num">0.</span>, <span class="num">0.4769</span>, -inf] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L908</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#exp" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="exp(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="exp(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">exp</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@exp(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise exponential value of the input. |
| |
| .. math:: |
| exp(x) = e^x \approx <span class="num">2.718</span>^x |
| |
| Example:: |
| |
| exp([<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>]) = [<span class="num">1.</span>, <span class="num">2.71828175</span>, <span class="num">7.38905621</span>] |
| |
| The storage <span class="kw">type</span> of ``exp`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L64</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#exp" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="exp(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="exp(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">exp</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@exp(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise exponential value of the input. |
| |
| .. math:: |
| exp(x) = e^x \approx <span class="num">2.718</span>^x |
| |
| Example:: |
| |
| exp([<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>]) = [<span class="num">1.</span>, <span class="num">2.71828175</span>, <span class="num">7.38905621</span>] |
| |
| The storage <span class="kw">type</span> of ``exp`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L64</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#expand_dims" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="expand_dims(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="expand_dims(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">expand_dims</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@expand_dims(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Inserts a <span class="kw">new</span> axis of size <span class="num">1</span> into the array shape |
| For example, given ``x`` <span class="kw">with</span> shape ``(<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>)``, then ``expand_dims(x, axis=<span class="num">1</span>)`` |
| will <span class="kw">return</span> a <span class="kw">new</span> array <span class="kw">with</span> shape ``(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">4</span>)``. |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L394</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#expand_dims" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="expand_dims(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="expand_dims(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">expand_dims</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@expand_dims(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Inserts a <span class="kw">new</span> axis of size <span class="num">1</span> into the array shape |
| For example, given ``x`` <span class="kw">with</span> shape ``(<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>)``, then ``expand_dims(x, axis=<span class="num">1</span>)`` |
| will <span class="kw">return</span> a <span class="kw">new</span> array <span class="kw">with</span> shape ``(<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">4</span>)``. |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L394</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#expm1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="expm1(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="expm1(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">expm1</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@expm1(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns ``exp(x) - <span class="num">1</span>`` computed element-wise on the input. |
| |
| This function provides greater precision than ``exp(x) - <span class="num">1</span>`` <span class="kw">for</span> small values of ``x``. |
| |
| The storage <span class="kw">type</span> of ``expm1`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - expm1(default) = default |
| - expm1(row_sparse) = row_sparse |
| - expm1(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L244</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#expm1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="expm1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="expm1(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">expm1</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@expm1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns ``exp(x) - <span class="num">1</span>`` computed element-wise on the input. |
| |
| This function provides greater precision than ``exp(x) - <span class="num">1</span>`` <span class="kw">for</span> small values of ``x``. |
| |
| The storage <span class="kw">type</span> of ``expm1`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - expm1(default) = default |
| - expm1(row_sparse) = row_sparse |
| - expm1(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L244</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#fill_element_0index" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="fill_element_0index(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="fill_element_0index(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">fill_element_0index</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@fill_element_0index(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Fill one element of each line(row <span class="kw">for</span> python, column <span class="kw">for</span> R/Julia) in lhs according to index indicated by rhs and values indicated by mhs. This function assume rhs uses <span class="num">0</span>-based index.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#fill_element_0index" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="fill_element_0index(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="fill_element_0index(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">fill_element_0index</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@fill_element_0index(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Fill one element of each line(row <span class="kw">for</span> python, column <span class="kw">for</span> R/Julia) in lhs according to index indicated by rhs and values indicated by mhs. This function assume rhs uses <span class="num">0</span>-based index.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#fix" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="fix(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="fix(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">fix</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@fix(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise rounded value to the nearest \ |
| integer towards zero of the input. |
| |
| Example:: |
| |
| fix([-<span class="num">2.1</span>, -<span class="num">1.9</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, -<span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``fix`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - fix(default) = default |
| - fix(row_sparse) = row_sparse |
| - fix(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L874</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#fix" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="fix(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="fix(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">fix</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@fix(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise rounded value to the nearest \ |
| integer towards zero of the input. |
| |
| Example:: |
| |
| fix([-<span class="num">2.1</span>, -<span class="num">1.9</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, -<span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``fix`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - fix(default) = default |
| - fix(row_sparse) = row_sparse |
| - fix(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L874</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#flatten" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="flatten(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="flatten(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">flatten</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@flatten(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Flattens the input array into a <span class="num">2</span>-D array by collapsing the higher dimensions. |
| .. note:: `Flatten` is deprecated. Use `flatten` instead. |
| For an input array <span class="kw">with</span> shape ``(d1, d2, ..., dk)``, `flatten` operation reshapes |
| the input array into an output array of shape ``(d1, d2*...*dk)``. |
| Note that the behavior of <span class="kw">this</span> function is different from numpy.ndarray.flatten, |
| which behaves similar to mxnet.ndarray.reshape((-<span class="num">1</span>,)). |
| Example:: |
| x = `[ [ |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>], |
| [<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] |
| ], |
| [ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>], |
| [<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] |
| ] ], |
| flatten(x) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L249</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#flatten" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="flatten(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="flatten(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">flatten</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@flatten(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Flattens the input array into a <span class="num">2</span>-D array by collapsing the higher dimensions. |
| .. note:: `Flatten` is deprecated. Use `flatten` instead. |
| For an input array <span class="kw">with</span> shape ``(d1, d2, ..., dk)``, `flatten` operation reshapes |
| the input array into an output array of shape ``(d1, d2*...*dk)``. |
| Note that the behavior of <span class="kw">this</span> function is different from numpy.ndarray.flatten, |
| which behaves similar to mxnet.ndarray.reshape((-<span class="num">1</span>,)). |
| Example:: |
| x = `[ [ |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>], |
| [<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] |
| ], |
| [ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">4</span>,<span class="num">5</span>,<span class="num">6</span>], |
| [<span class="num">7</span>,<span class="num">8</span>,<span class="num">9</span>] |
| ] ], |
| flatten(x) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L249</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#flip" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="flip(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="flip(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">flip</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@flip(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reverses the order of elements along given axis <span class="kw">while</span> preserving array shape. |
| Note: reverse and flip are equivalent. We use reverse in the following examples. |
| Examples:: |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| reverse(x, axis=<span class="num">0</span>) = `[ [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] |
| reverse(x, axis=<span class="num">1</span>) = `[ [ <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">9.</span>, <span class="num">8.</span>, <span class="num">7.</span>, <span class="num">6.</span>, <span class="num">5.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L831</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#flip" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="flip(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="flip(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">flip</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@flip(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reverses the order of elements along given axis <span class="kw">while</span> preserving array shape. |
| Note: reverse and flip are equivalent. We use reverse in the following examples. |
| Examples:: |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| reverse(x, axis=<span class="num">0</span>) = `[ [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] |
| reverse(x, axis=<span class="num">1</span>) = `[ [ <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">9.</span>, <span class="num">8.</span>, <span class="num">7.</span>, <span class="num">6.</span>, <span class="num">5.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L831</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#floor" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="floor(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="floor(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">floor</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@floor(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise floor of the input. |
| |
| The floor of the scalar x is the largest integer i, such that i <= x. |
| |
| Example:: |
| |
| floor([-<span class="num">2.1</span>, -<span class="num">1.9</span>, <span class="num">1.5</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">3.</span>, -<span class="num">2.</span>, <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``floor`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - floor(default) = default |
| - floor(row_sparse) = row_sparse |
| - floor(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L836</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#floor" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="floor(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="floor(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">floor</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@floor(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise floor of the input. |
| |
| The floor of the scalar x is the largest integer i, such that i <= x. |
| |
| Example:: |
| |
| floor([-<span class="num">2.1</span>, -<span class="num">1.9</span>, <span class="num">1.5</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">3.</span>, -<span class="num">2.</span>, <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``floor`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - floor(default) = default |
| - floor(row_sparse) = row_sparse |
| - floor(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L836</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ftml_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ftml_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ftml_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ftml_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ftml_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>The FTML optimizer described in |
| *FTML - Follow the Moving Leader in Deep Learning*, |
| available at http:<span class="cmt">//proceedings.mlr.press/v70/zheng17a/zheng17a.pdf.</span> |
| |
| .. math:: |
| |
| g_t = \nabla J(W_{t-<span class="num">1</span>})\\ |
| v_t = \beta_2 v_{t-<span class="num">1</span>} + (<span class="num">1</span> - \beta_2) g_t^<span class="num">2</span>\\ |
| d_t = \frac{ <span class="num">1</span> - \beta_1^t }{ \eta_t } (\sqrt{ \frac{ v_t }{ <span class="num">1</span> - \beta_2^t } } + \epsilon) |
| \sigma_t = d_t - \beta_1 d_{t-<span class="num">1</span>} |
| z_t = \beta_1 z_{ t-<span class="num">1</span> } + (<span class="num">1</span> - \beta_1^t) g_t - \sigma_t W_{t-<span class="num">1</span>} |
| W_t = - \frac{ z_t }{ d_t } |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L639</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ftml_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ftml_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ftml_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ftml_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ftml_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>The FTML optimizer described in |
| *FTML - Follow the Moving Leader in Deep Learning*, |
| available at http:<span class="cmt">//proceedings.mlr.press/v70/zheng17a/zheng17a.pdf.</span> |
| |
| .. math:: |
| |
| g_t = \nabla J(W_{t-<span class="num">1</span>})\\ |
| v_t = \beta_2 v_{t-<span class="num">1</span>} + (<span class="num">1</span> - \beta_2) g_t^<span class="num">2</span>\\ |
| d_t = \frac{ <span class="num">1</span> - \beta_1^t }{ \eta_t } (\sqrt{ \frac{ v_t }{ <span class="num">1</span> - \beta_2^t } } + \epsilon) |
| \sigma_t = d_t - \beta_1 d_{t-<span class="num">1</span>} |
| z_t = \beta_1 z_{ t-<span class="num">1</span> } + (<span class="num">1</span> - \beta_1^t) g_t - \sigma_t W_{t-<span class="num">1</span>} |
| W_t = - \frac{ z_t }{ d_t } |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L639</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ftrl_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ftrl_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ftrl_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ftrl_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ftrl_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Ftrl optimizer. |
| Referenced from *Ad Click Prediction: a View from the Trenches*, available at |
| http:<span class="cmt">//dl.acm.org/citation.cfm?id=2488200.</span> |
| |
| It updates the weights using:: |
| |
| rescaled_grad = clip(grad * rescale_grad, clip_gradient) |
| z += rescaled_grad - (sqrt(n + rescaled_grad**<span class="num">2</span>) - sqrt(n)) * weight / learning_rate |
| n += rescaled_grad**<span class="num">2</span> |
| w = (sign(z) * lamda1 - z) / ((beta + sqrt(n)) / learning_rate + wd) * (abs(z) > lamda1) |
| |
| If w, z and n are all of ``row_sparse`` storage <span class="kw">type</span>, |
| only the row slices whose indices appear in grad.indices are updated (<span class="kw">for</span> w, z and n):: |
| |
| <span class="kw">for</span> row in grad.indices: |
| rescaled_grad[row] = clip(grad[row] * rescale_grad, clip_gradient) |
| z[row] += rescaled_grad[row] - (sqrt(n[row] + rescaled_grad[row]**<span class="num">2</span>) - sqrt(n[row])) * weight[row] / learning_rate |
| n[row] += rescaled_grad[row]**<span class="num">2</span> |
| w[row] = (sign(z[row]) * lamda1 - z[row]) / ((beta + sqrt(n[row])) / learning_rate + wd) * (abs(z[row]) > lamda1) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L875</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ftrl_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ftrl_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ftrl_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ftrl_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ftrl_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Ftrl optimizer. |
| Referenced from *Ad Click Prediction: a View from the Trenches*, available at |
| http:<span class="cmt">//dl.acm.org/citation.cfm?id=2488200.</span> |
| |
| It updates the weights using:: |
| |
| rescaled_grad = clip(grad * rescale_grad, clip_gradient) |
| z += rescaled_grad - (sqrt(n + rescaled_grad**<span class="num">2</span>) - sqrt(n)) * weight / learning_rate |
| n += rescaled_grad**<span class="num">2</span> |
| w = (sign(z) * lamda1 - z) / ((beta + sqrt(n)) / learning_rate + wd) * (abs(z) > lamda1) |
| |
| If w, z and n are all of ``row_sparse`` storage <span class="kw">type</span>, |
| only the row slices whose indices appear in grad.indices are updated (<span class="kw">for</span> w, z and n):: |
| |
| <span class="kw">for</span> row in grad.indices: |
| rescaled_grad[row] = clip(grad[row] * rescale_grad, clip_gradient) |
| z[row] += rescaled_grad[row] - (sqrt(n[row] + rescaled_grad[row]**<span class="num">2</span>) - sqrt(n[row])) * weight[row] / learning_rate |
| n[row] += rescaled_grad[row]**<span class="num">2</span> |
| w[row] = (sign(z[row]) * lamda1 - z[row]) / ((beta + sqrt(n[row])) / learning_rate + wd) * (abs(z[row]) > lamda1) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L875</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#gamma" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="gamma(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="gamma(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">gamma</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@gamma(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the gamma function (extension of the factorial function \ |
| to the reals), computed element-wise on the input array. |
| |
| The storage <span class="kw">type</span> of ``gamma`` output is always dense</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#gamma" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="gamma(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="gamma(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">gamma</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@gamma(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the gamma function (extension of the factorial function \ |
| to the reals), computed element-wise on the input array. |
| |
| The storage <span class="kw">type</span> of ``gamma`` output is always dense</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#gammaln" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="gammaln(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="gammaln(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">gammaln</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@gammaln(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise log of the absolute value of the gamma function \ |
| of the input. |
| |
| The storage <span class="kw">type</span> of ``gammaln`` output is always dense</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#gammaln" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="gammaln(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="gammaln(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">gammaln</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@gammaln(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise log of the absolute value of the gamma function \ |
| of the input. |
| |
| The storage <span class="kw">type</span> of ``gammaln`` output is always dense</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#gather_nd" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="gather_nd(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="gather_nd(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">gather_nd</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@gather_nd(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Gather elements or slices from `data` and store to a tensor whose |
| shape is defined by `indices`. |
| |
| Given `data` <span class="kw">with</span> shape `(X_0, X_1, ..., X_{N-<span class="num">1</span>})` and indices <span class="kw">with</span> shape |
| `(M, Y_0, ..., Y_{K-<span class="num">1</span>})`, the output will have shape `(Y_0, ..., Y_{K-<span class="num">1</span>}, X_M, ..., X_{N-<span class="num">1</span>})`, |
| where `M <= N`. If `M == N`, output shape will simply be `(Y_0, ..., Y_{K-<span class="num">1</span>})`. |
| |
| The elements in output is defined as follows:: |
| |
| output[y_0, ..., y_{K-<span class="num">1</span>}, x_M, ..., x_{N-<span class="num">1</span>}] = data[indices[<span class="num">0</span>, y_0, ..., y_{K-<span class="num">1</span>}], |
| ..., |
| indices[M-<span class="num">1</span>, y_0, ..., y_{K-<span class="num">1</span>}], |
| x_M, ..., x_{N-<span class="num">1</span>}] |
| |
| Examples:: |
| |
| data = `[ [<span class="num">0</span>, <span class="num">1</span>], [<span class="num">2</span>, <span class="num">3</span>] ] |
| indices = `[ [<span class="num">1</span>, <span class="num">1</span>, <span class="num">0</span>], [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>] ] |
| gather_nd(data, indices) = [<span class="num">2</span>, <span class="num">3</span>, <span class="num">0</span>] |
| |
| data = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">4</span>] ], `[ [<span class="num">5</span>, <span class="num">6</span>], [<span class="num">7</span>, <span class="num">8</span>] ] ] |
| indices = `[ [<span class="num">0</span>, <span class="num">1</span>], [<span class="num">1</span>, <span class="num">0</span>] ] |
| gather_nd(data, indices) = `[ [<span class="num">3</span>, <span class="num">4</span>], [<span class="num">5</span>, <span class="num">6</span>] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#gather_nd" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="gather_nd(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="gather_nd(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">gather_nd</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@gather_nd(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Gather elements or slices from `data` and store to a tensor whose |
| shape is defined by `indices`. |
| |
| Given `data` <span class="kw">with</span> shape `(X_0, X_1, ..., X_{N-<span class="num">1</span>})` and indices <span class="kw">with</span> shape |
| `(M, Y_0, ..., Y_{K-<span class="num">1</span>})`, the output will have shape `(Y_0, ..., Y_{K-<span class="num">1</span>}, X_M, ..., X_{N-<span class="num">1</span>})`, |
| where `M <= N`. If `M == N`, output shape will simply be `(Y_0, ..., Y_{K-<span class="num">1</span>})`. |
| |
| The elements in output is defined as follows:: |
| |
| output[y_0, ..., y_{K-<span class="num">1</span>}, x_M, ..., x_{N-<span class="num">1</span>}] = data[indices[<span class="num">0</span>, y_0, ..., y_{K-<span class="num">1</span>}], |
| ..., |
| indices[M-<span class="num">1</span>, y_0, ..., y_{K-<span class="num">1</span>}], |
| x_M, ..., x_{N-<span class="num">1</span>}] |
| |
| Examples:: |
| |
| data = `[ [<span class="num">0</span>, <span class="num">1</span>], [<span class="num">2</span>, <span class="num">3</span>] ] |
| indices = `[ [<span class="num">1</span>, <span class="num">1</span>, <span class="num">0</span>], [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>] ] |
| gather_nd(data, indices) = [<span class="num">2</span>, <span class="num">3</span>, <span class="num">0</span>] |
| |
| data = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">4</span>] ], `[ [<span class="num">5</span>, <span class="num">6</span>], [<span class="num">7</span>, <span class="num">8</span>] ] ] |
| indices = `[ [<span class="num">0</span>, <span class="num">1</span>], [<span class="num">1</span>, <span class="num">0</span>] ] |
| gather_nd(data, indices) = `[ [<span class="num">3</span>, <span class="num">4</span>], [<span class="num">5</span>, <span class="num">6</span>] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#hard_sigmoid" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="hard_sigmoid(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="hard_sigmoid(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">hard_sigmoid</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@hard_sigmoid(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes hard sigmoid of x element-wise. |
| |
| .. math:: |
| y = max(<span class="num">0</span>, min(<span class="num">1</span>, alpha * x + beta)) |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L161</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#hard_sigmoid" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="hard_sigmoid(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="hard_sigmoid(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">hard_sigmoid</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@hard_sigmoid(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes hard sigmoid of x element-wise. |
| |
| .. math:: |
| y = max(<span class="num">0</span>, min(<span class="num">1</span>, alpha * x + beta)) |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L161</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#identity" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="identity(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="identity(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">identity</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@identity(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a copy of the input. |
| |
| From:src/operator/tensor/elemwise_unary_op_basic.cc:<span class="num">244</span></pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#identity" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="identity(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="identity(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">identity</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@identity(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a copy of the input. |
| |
| From:src/operator/tensor/elemwise_unary_op_basic.cc:<span class="num">244</span></pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#im2col" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="im2col(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="im2col(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">im2col</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@im2col(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Extract sliding blocks from input array. |
| |
| This operator is used in vanilla convolution implementation to transform the sliding |
| blocks on image to column matrix, then the convolution operation can be computed |
| by matrix multiplication between column and convolution weight. Due to the close |
| relation between im2col and convolution, the concept of **kernel**, **stride**, |
| **dilate** and **pad** in <span class="kw">this</span> operator are inherited from convolution operation. |
| |
| Given the input data of shape :math:`(N, C, *)`, where :math:`N` is the batch size, |
| :math:`C` is the channel size, and :math:`*` is the arbitrary spatial dimension, |
| the output column array is always <span class="kw">with</span> shape :math:`(N, C \times \prod(\text{kernel}), W)`, |
| where :math:`C \times \prod(\text{kernel})` is the block size, and :math:`W` is the |
| block number which is the spatial size of the convolution output <span class="kw">with</span> same input parameters. |
| Only <span class="num">1</span>-D, <span class="num">2</span>-D and <span class="num">3</span>-D of spatial dimension is supported in <span class="kw">this</span> operator. |
| |
| |
| |
| Defined in src/operator/nn/im2col.cc:L99</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#im2col" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="im2col(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="im2col(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">im2col</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@im2col(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Extract sliding blocks from input array. |
| |
| This operator is used in vanilla convolution implementation to transform the sliding |
| blocks on image to column matrix, then the convolution operation can be computed |
| by matrix multiplication between column and convolution weight. Due to the close |
| relation between im2col and convolution, the concept of **kernel**, **stride**, |
| **dilate** and **pad** in <span class="kw">this</span> operator are inherited from convolution operation. |
| |
| Given the input data of shape :math:`(N, C, *)`, where :math:`N` is the batch size, |
| :math:`C` is the channel size, and :math:`*` is the arbitrary spatial dimension, |
| the output column array is always <span class="kw">with</span> shape :math:`(N, C \times \prod(\text{kernel}), W)`, |
| where :math:`C \times \prod(\text{kernel})` is the block size, and :math:`W` is the |
| block number which is the spatial size of the convolution output <span class="kw">with</span> same input parameters. |
| Only <span class="num">1</span>-D, <span class="num">2</span>-D and <span class="num">3</span>-D of spatial dimension is supported in <span class="kw">this</span> operator. |
| |
| |
| |
| Defined in src/operator/nn/im2col.cc:L99</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#khatri_rao" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="khatri_rao(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="khatri_rao(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">khatri_rao</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@khatri_rao(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the Khatri-Rao product of the input matrices. |
| |
| Given a collection of :math:`n` input matrices, |
| |
| .. math:: |
| A_1 \in \mathbb{R}^{M_1 \times M}, \ldots, A_n \in \mathbb{R}^{M_n \times N}, |
| |
| the (column-wise) Khatri-Rao product is defined as the matrix, |
| |
| .. math:: |
| X = A_1 \otimes \cdots \otimes A_n \in \mathbb{R}^{(M_1 \cdots M_n) \times N}, |
| |
| where the :math:`k` th column is equal to the column-wise outer product |
| :math:`{A_1}_k \otimes \cdots \otimes {A_n}_k` where :math:`{A_i}_k` is the kth |
| column of the ith matrix. |
| |
| Example:: |
| |
| >>> A = mx.nd.array(`[ [<span class="num">1</span>, -<span class="num">1</span>], |
| >>> [<span class="num">2</span>, -<span class="num">3</span>] ]) |
| >>> B = mx.nd.array(`[ [<span class="num">1</span>, <span class="num">4</span>], |
| >>> [<span class="num">2</span>, <span class="num">5</span>], |
| >>> [<span class="num">3</span>, <span class="num">6</span>] ]) |
| >>> C = mx.nd.khatri_rao(A, B) |
| >>> print(C.asnumpy()) |
| `[ [ <span class="num">1.</span> -<span class="num">4.</span>] |
| [ <span class="num">2.</span> -<span class="num">5.</span>] |
| [ <span class="num">3.</span> -<span class="num">6.</span>] |
| [ <span class="num">2.</span> -<span class="num">12.</span>] |
| [ <span class="num">4.</span> -<span class="num">15.</span>] |
| [ <span class="num">6.</span> -<span class="num">18.</span>] ] |
| |
| |
| |
| Defined in src/operator/contrib/krprod.cc:L108</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#khatri_rao" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="khatri_rao(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="khatri_rao(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">khatri_rao</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@khatri_rao(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the Khatri-Rao product of the input matrices. |
| |
| Given a collection of :math:`n` input matrices, |
| |
| .. math:: |
| A_1 \in \mathbb{R}^{M_1 \times M}, \ldots, A_n \in \mathbb{R}^{M_n \times N}, |
| |
| the (column-wise) Khatri-Rao product is defined as the matrix, |
| |
| .. math:: |
| X = A_1 \otimes \cdots \otimes A_n \in \mathbb{R}^{(M_1 \cdots M_n) \times N}, |
| |
| where the :math:`k` th column is equal to the column-wise outer product |
| :math:`{A_1}_k \otimes \cdots \otimes {A_n}_k` where :math:`{A_i}_k` is the kth |
| column of the ith matrix. |
| |
| Example:: |
| |
| >>> A = mx.nd.array(`[ [<span class="num">1</span>, -<span class="num">1</span>], |
| >>> [<span class="num">2</span>, -<span class="num">3</span>] ]) |
| >>> B = mx.nd.array(`[ [<span class="num">1</span>, <span class="num">4</span>], |
| >>> [<span class="num">2</span>, <span class="num">5</span>], |
| >>> [<span class="num">3</span>, <span class="num">6</span>] ]) |
| >>> C = mx.nd.khatri_rao(A, B) |
| >>> print(C.asnumpy()) |
| `[ [ <span class="num">1.</span> -<span class="num">4.</span>] |
| [ <span class="num">2.</span> -<span class="num">5.</span>] |
| [ <span class="num">3.</span> -<span class="num">6.</span>] |
| [ <span class="num">2.</span> -<span class="num">12.</span>] |
| [ <span class="num">4.</span> -<span class="num">15.</span>] |
| [ <span class="num">6.</span> -<span class="num">18.</span>] ] |
| |
| |
| |
| Defined in src/operator/contrib/krprod.cc:L108</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#lamb_update_phase1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="lamb_update_phase1(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="lamb_update_phase1(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">lamb_update_phase1</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@lamb_update_phase1(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Phase I of lamb update it performs the following operations and returns g:. |
| |
| Link to paper: https:<span class="cmt">//arxiv.org/pdf/1904.00962.pdf</span> |
| |
| .. math:: |
| \begin{gather*} |
| grad = grad * rescale_grad |
| <span class="kw">if</span> (grad < -clip_gradient) |
| then |
| grad = -clip_gradient |
| <span class="kw">if</span> (grad > clip_gradient) |
| then |
| grad = clip_gradient |
| |
| mean = beta1 * mean + (<span class="num">1</span> - beta1) * grad; |
| variance = beta2 * variance + (<span class="num">1.</span> - beta2) * grad ^ <span class="num">2</span>; |
| |
| <span class="kw">if</span> (bias_correction) |
| then |
| mean_hat = mean / (<span class="num">1.</span> - beta1^t); |
| var_hat = <span class="kw">var</span> / (<span class="num">1</span> - beta2^t); |
| g = mean_hat / (var_hat^(<span class="num">1</span>/<span class="num">2</span>) + epsilon) + wd * weight; |
| <span class="kw">else</span> |
| g = mean / (var_data^(<span class="num">1</span>/<span class="num">2</span>) + epsilon) + wd * weight; |
| \end{gather*} |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L952</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#lamb_update_phase1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="lamb_update_phase1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="lamb_update_phase1(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">lamb_update_phase1</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@lamb_update_phase1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Phase I of lamb update it performs the following operations and returns g:. |
| |
| Link to paper: https:<span class="cmt">//arxiv.org/pdf/1904.00962.pdf</span> |
| |
| .. math:: |
| \begin{gather*} |
| grad = grad * rescale_grad |
| <span class="kw">if</span> (grad < -clip_gradient) |
| then |
| grad = -clip_gradient |
| <span class="kw">if</span> (grad > clip_gradient) |
| then |
| grad = clip_gradient |
| |
| mean = beta1 * mean + (<span class="num">1</span> - beta1) * grad; |
| variance = beta2 * variance + (<span class="num">1.</span> - beta2) * grad ^ <span class="num">2</span>; |
| |
| <span class="kw">if</span> (bias_correction) |
| then |
| mean_hat = mean / (<span class="num">1.</span> - beta1^t); |
| var_hat = <span class="kw">var</span> / (<span class="num">1</span> - beta2^t); |
| g = mean_hat / (var_hat^(<span class="num">1</span>/<span class="num">2</span>) + epsilon) + wd * weight; |
| <span class="kw">else</span> |
| g = mean / (var_data^(<span class="num">1</span>/<span class="num">2</span>) + epsilon) + wd * weight; |
| \end{gather*} |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L952</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#lamb_update_phase2" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="lamb_update_phase2(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="lamb_update_phase2(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">lamb_update_phase2</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@lamb_update_phase2(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Phase II of lamb update it performs the following operations and updates grad. |
| |
| Link to paper: https:<span class="cmt">//arxiv.org/pdf/1904.00962.pdf</span> |
| |
| .. math:: |
| \begin{gather*} |
| <span class="kw">if</span> (lower_bound >= <span class="num">0</span>) |
| then |
| r1 = max(r1, lower_bound) |
| <span class="kw">if</span> (upper_bound >= <span class="num">0</span>) |
| then |
| r1 = max(r1, upper_bound) |
| |
| <span class="kw">if</span> (r1 == <span class="num">0</span> or r2 == <span class="num">0</span>) |
| then |
| lr = lr |
| <span class="kw">else</span> |
| lr = lr * (r1/r2) |
| weight = weight - lr * g |
| \end{gather*} |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L991</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#lamb_update_phase2" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="lamb_update_phase2(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="lamb_update_phase2(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">lamb_update_phase2</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@lamb_update_phase2(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Phase II of lamb update it performs the following operations and updates grad. |
| |
| Link to paper: https:<span class="cmt">//arxiv.org/pdf/1904.00962.pdf</span> |
| |
| .. math:: |
| \begin{gather*} |
| <span class="kw">if</span> (lower_bound >= <span class="num">0</span>) |
| then |
| r1 = max(r1, lower_bound) |
| <span class="kw">if</span> (upper_bound >= <span class="num">0</span>) |
| then |
| r1 = max(r1, upper_bound) |
| |
| <span class="kw">if</span> (r1 == <span class="num">0</span> or r2 == <span class="num">0</span>) |
| then |
| lr = lr |
| <span class="kw">else</span> |
| lr = lr * (r1/r2) |
| weight = weight - lr * g |
| \end{gather*} |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L991</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_det" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_det(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_det(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_det</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_det(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the determinant of a matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* is a square matrix. We compute: |
| |
| *out* = *det(A)* |
| |
| If *n><span class="num">2</span>*, *det* is performed separately on the trailing two dimensions |
| <span class="kw">for</span> all inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| .. note:: There is no gradient backwarded when A is non-invertible (which is |
| equivalent to det(A) = <span class="num">0</span>) because zero is rarely hit upon in float |
| point computation and the Jacobi's formula on determinant gradient |
| is not computationally efficient when A is non-invertible. |
| |
| Examples:: |
| |
| Single matrix determinant |
| A = `[ [<span class="num">1.</span>, <span class="num">4.</span>], [<span class="num">2.</span>, <span class="num">3.</span>] ] |
| det(A) = [-<span class="num">5.</span>] |
| |
| Batch matrix determinant |
| A = `[ `[ [<span class="num">1.</span>, <span class="num">4.</span>], [<span class="num">2.</span>, <span class="num">3.</span>] ], |
| `[ [<span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">1.</span>, <span class="num">4.</span>] ] ] |
| det(A) = [-<span class="num">5.</span>, <span class="num">5.</span>] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L974</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_det" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_det(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_det(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_det</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_det(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the determinant of a matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* is a square matrix. We compute: |
| |
| *out* = *det(A)* |
| |
| If *n><span class="num">2</span>*, *det* is performed separately on the trailing two dimensions |
| <span class="kw">for</span> all inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| .. note:: There is no gradient backwarded when A is non-invertible (which is |
| equivalent to det(A) = <span class="num">0</span>) because zero is rarely hit upon in float |
| point computation and the Jacobi's formula on determinant gradient |
| is not computationally efficient when A is non-invertible. |
| |
| Examples:: |
| |
| Single matrix determinant |
| A = `[ [<span class="num">1.</span>, <span class="num">4.</span>], [<span class="num">2.</span>, <span class="num">3.</span>] ] |
| det(A) = [-<span class="num">5.</span>] |
| |
| Batch matrix determinant |
| A = `[ `[ [<span class="num">1.</span>, <span class="num">4.</span>], [<span class="num">2.</span>, <span class="num">3.</span>] ], |
| `[ [<span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">1.</span>, <span class="num">4.</span>] ] ] |
| det(A) = [-<span class="num">5.</span>, <span class="num">5.</span>] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L974</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_extractdiag" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_extractdiag(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_extractdiag(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_extractdiag</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_extractdiag(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Extracts the diagonal entries of a square matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, then *A* represents a single square matrix which diagonal elements get extracted as a <span class="num">1</span>-dimensional tensor. |
| |
| If *n><span class="num">2</span>*, then *A* represents a batch of square matrices on the trailing two dimensions. The extracted diagonals are returned as an *n-<span class="num">1</span>*-dimensional tensor. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix diagonal extraction |
| A = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ] |
| |
| extractdiag(A) = [<span class="num">1.0</span>, <span class="num">4.0</span>] |
| |
| extractdiag(A, <span class="num">1</span>) = [<span class="num">2.0</span>] |
| |
| Batch matrix diagonal extraction |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ], |
| `[ [<span class="num">5.0</span>, <span class="num">6.0</span>], |
| [<span class="num">7.0</span>, <span class="num">8.0</span>] ] ] |
| |
| extractdiag(A) = `[ [<span class="num">1.0</span>, <span class="num">4.0</span>], |
| [<span class="num">5.0</span>, <span class="num">8.0</span>] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L494</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_extractdiag" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_extractdiag(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_extractdiag(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_extractdiag</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_extractdiag(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Extracts the diagonal entries of a square matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, then *A* represents a single square matrix which diagonal elements get extracted as a <span class="num">1</span>-dimensional tensor. |
| |
| If *n><span class="num">2</span>*, then *A* represents a batch of square matrices on the trailing two dimensions. The extracted diagonals are returned as an *n-<span class="num">1</span>*-dimensional tensor. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix diagonal extraction |
| A = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ] |
| |
| extractdiag(A) = [<span class="num">1.0</span>, <span class="num">4.0</span>] |
| |
| extractdiag(A, <span class="num">1</span>) = [<span class="num">2.0</span>] |
| |
| Batch matrix diagonal extraction |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ], |
| `[ [<span class="num">5.0</span>, <span class="num">6.0</span>], |
| [<span class="num">7.0</span>, <span class="num">8.0</span>] ] ] |
| |
| extractdiag(A) = `[ [<span class="num">1.0</span>, <span class="num">4.0</span>], |
| [<span class="num">5.0</span>, <span class="num">8.0</span>] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L494</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_extracttrian" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_extracttrian(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_extracttrian(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_extracttrian</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_extracttrian(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Extracts a triangular sub-matrix from a square matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, then *A* represents a single square matrix from which a triangular sub-matrix is extracted as a <span class="num">1</span>-dimensional tensor. |
| |
| If *n><span class="num">2</span>*, then *A* represents a batch of square matrices on the trailing two dimensions. The extracted triangular sub-matrices are returned as an *n-<span class="num">1</span>*-dimensional tensor. |
| |
| The *offset* and *lower* parameters determine the triangle to be extracted: |
| |
| - When *offset = <span class="num">0</span>* either the lower or upper triangle <span class="kw">with</span> respect to the main diagonal is extracted depending on the value of parameter *lower*. |
| - When *offset = k > <span class="num">0</span>* the upper triangle <span class="kw">with</span> respect to the k-th diagonal above the main diagonal is extracted. |
| - When *offset = k < <span class="num">0</span>* the lower triangle <span class="kw">with</span> respect to the k-th diagonal below the main diagonal is extracted. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single triagonal extraction |
| A = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ] |
| |
| extracttrian(A) = [<span class="num">1.0</span>, <span class="num">3.0</span>, <span class="num">4.0</span>] |
| extracttrian(A, lower=False) = [<span class="num">1.0</span>, <span class="num">2.0</span>, <span class="num">4.0</span>] |
| extracttrian(A, <span class="num">1</span>) = [<span class="num">2.0</span>] |
| extracttrian(A, -<span class="num">1</span>) = [<span class="num">3.0</span>] |
| |
| Batch triagonal extraction |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ], |
| `[ [<span class="num">5.0</span>, <span class="num">6.0</span>], |
| [<span class="num">7.0</span>, <span class="num">8.0</span>] ] ] |
| |
| extracttrian(A) = `[ [<span class="num">1.0</span>, <span class="num">3.0</span>, <span class="num">4.0</span>], |
| [<span class="num">5.0</span>, <span class="num">7.0</span>, <span class="num">8.0</span>] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L604</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_extracttrian" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_extracttrian(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_extracttrian(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_extracttrian</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_extracttrian(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Extracts a triangular sub-matrix from a square matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, then *A* represents a single square matrix from which a triangular sub-matrix is extracted as a <span class="num">1</span>-dimensional tensor. |
| |
| If *n><span class="num">2</span>*, then *A* represents a batch of square matrices on the trailing two dimensions. The extracted triangular sub-matrices are returned as an *n-<span class="num">1</span>*-dimensional tensor. |
| |
| The *offset* and *lower* parameters determine the triangle to be extracted: |
| |
| - When *offset = <span class="num">0</span>* either the lower or upper triangle <span class="kw">with</span> respect to the main diagonal is extracted depending on the value of parameter *lower*. |
| - When *offset = k > <span class="num">0</span>* the upper triangle <span class="kw">with</span> respect to the k-th diagonal above the main diagonal is extracted. |
| - When *offset = k < <span class="num">0</span>* the lower triangle <span class="kw">with</span> respect to the k-th diagonal below the main diagonal is extracted. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single triagonal extraction |
| A = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ] |
| |
| extracttrian(A) = [<span class="num">1.0</span>, <span class="num">3.0</span>, <span class="num">4.0</span>] |
| extracttrian(A, lower=False) = [<span class="num">1.0</span>, <span class="num">2.0</span>, <span class="num">4.0</span>] |
| extracttrian(A, <span class="num">1</span>) = [<span class="num">2.0</span>] |
| extracttrian(A, -<span class="num">1</span>) = [<span class="num">3.0</span>] |
| |
| Batch triagonal extraction |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ], |
| `[ [<span class="num">5.0</span>, <span class="num">6.0</span>], |
| [<span class="num">7.0</span>, <span class="num">8.0</span>] ] ] |
| |
| extracttrian(A) = `[ [<span class="num">1.0</span>, <span class="num">3.0</span>, <span class="num">4.0</span>], |
| [<span class="num">5.0</span>, <span class="num">7.0</span>, <span class="num">8.0</span>] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L604</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_gelqf" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_gelqf(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_gelqf(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_gelqf</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_gelqf(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>LQ factorization <span class="kw">for</span> general matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, we compute the LQ factorization (LAPACK *gelqf*, followed by *orglq*). *A* |
| must have shape *(x, y)* <span class="kw">with</span> *x <= y*, and must have full rank *=x*. The LQ |
| factorization consists of *L* <span class="kw">with</span> shape *(x, x)* and *Q* <span class="kw">with</span> shape *(x, y)*, so |
| that: |
| |
| *A* = *L* \* *Q* |
| |
| Here, *L* is lower triangular (upper triangle equal to zero) <span class="kw">with</span> nonzero diagonal, |
| and *Q* is row-orthonormal, meaning that |
| |
| *Q* \* *Q*\ :sup:`T` |
| |
| is equal to the identity matrix of shape *(x, x)*. |
| |
| If *n><span class="num">2</span>*, *gelqf* is performed separately on the trailing two dimensions <span class="kw">for</span> all |
| inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single LQ factorization |
| A = `[ [<span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ] |
| Q, L = gelqf(A) |
| Q = `[ [-<span class="num">0.26726124</span>, -<span class="num">0.53452248</span>, -<span class="num">0.80178373</span>], |
| [<span class="num">0.87287156</span>, <span class="num">0.21821789</span>, -<span class="num">0.43643578</span>] ] |
| L = `[ [-<span class="num">3.74165739</span>, <span class="num">0.</span>], |
| [-<span class="num">8.55235974</span>, <span class="num">1.96396101</span>] ] |
| |
| Batch LQ factorization |
| A = `[ `[ [<span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| `[ [<span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], [<span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] ] |
| Q, L = gelqf(A) |
| Q = `[ `[ [-<span class="num">0.26726124</span>, -<span class="num">0.53452248</span>, -<span class="num">0.80178373</span>], |
| [<span class="num">0.87287156</span>, <span class="num">0.21821789</span>, -<span class="num">0.43643578</span>] ], |
| `[ [-<span class="num">0.50257071</span>, -<span class="num">0.57436653</span>, -<span class="num">0.64616234</span>], |
| [<span class="num">0.7620735</span>, <span class="num">0.05862104</span>, -<span class="num">0.64483142</span>] ] ] |
| L = `[ `[ [-<span class="num">3.74165739</span>, <span class="num">0.</span>], |
| [-<span class="num">8.55235974</span>, <span class="num">1.96396101</span>] ], |
| `[ [-<span class="num">13.92838828</span>, <span class="num">0.</span>], |
| [-<span class="num">19.09768702</span>, <span class="num">0.52758934</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L797</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_gelqf" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_gelqf(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_gelqf(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_gelqf</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_gelqf(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>LQ factorization <span class="kw">for</span> general matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, we compute the LQ factorization (LAPACK *gelqf*, followed by *orglq*). *A* |
| must have shape *(x, y)* <span class="kw">with</span> *x <= y*, and must have full rank *=x*. The LQ |
| factorization consists of *L* <span class="kw">with</span> shape *(x, x)* and *Q* <span class="kw">with</span> shape *(x, y)*, so |
| that: |
| |
| *A* = *L* \* *Q* |
| |
| Here, *L* is lower triangular (upper triangle equal to zero) <span class="kw">with</span> nonzero diagonal, |
| and *Q* is row-orthonormal, meaning that |
| |
| *Q* \* *Q*\ :sup:`T` |
| |
| is equal to the identity matrix of shape *(x, x)*. |
| |
| If *n><span class="num">2</span>*, *gelqf* is performed separately on the trailing two dimensions <span class="kw">for</span> all |
| inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single LQ factorization |
| A = `[ [<span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ] |
| Q, L = gelqf(A) |
| Q = `[ [-<span class="num">0.26726124</span>, -<span class="num">0.53452248</span>, -<span class="num">0.80178373</span>], |
| [<span class="num">0.87287156</span>, <span class="num">0.21821789</span>, -<span class="num">0.43643578</span>] ] |
| L = `[ [-<span class="num">3.74165739</span>, <span class="num">0.</span>], |
| [-<span class="num">8.55235974</span>, <span class="num">1.96396101</span>] ] |
| |
| Batch LQ factorization |
| A = `[ `[ [<span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ], |
| `[ [<span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], [<span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] ] |
| Q, L = gelqf(A) |
| Q = `[ `[ [-<span class="num">0.26726124</span>, -<span class="num">0.53452248</span>, -<span class="num">0.80178373</span>], |
| [<span class="num">0.87287156</span>, <span class="num">0.21821789</span>, -<span class="num">0.43643578</span>] ], |
| `[ [-<span class="num">0.50257071</span>, -<span class="num">0.57436653</span>, -<span class="num">0.64616234</span>], |
| [<span class="num">0.7620735</span>, <span class="num">0.05862104</span>, -<span class="num">0.64483142</span>] ] ] |
| L = `[ `[ [-<span class="num">3.74165739</span>, <span class="num">0.</span>], |
| [-<span class="num">8.55235974</span>, <span class="num">1.96396101</span>] ], |
| `[ [-<span class="num">13.92838828</span>, <span class="num">0.</span>], |
| [-<span class="num">19.09768702</span>, <span class="num">0.52758934</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L797</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_gemm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_gemm(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_gemm(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_gemm</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_gemm(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs general matrix multiplication and accumulation. |
| Input are tensors *A*, *B*, *C*, each of dimension *n >= <span class="num">2</span>* and having the same shape |
| on the leading *n-<span class="num">2</span>* dimensions. |
| |
| If *n=<span class="num">2</span>*, the BLAS3 function *gemm* is performed: |
| |
| *out* = *alpha* \* *op*\ (*A*) \* *op*\ (*B*) + *beta* \* *C* |
| |
| Here, *alpha* and *beta* are scalar parameters, and *op()* is either the identity or |
| matrix transposition (depending on *transpose_a*, *transpose_b*). |
| |
| If *n><span class="num">2</span>*, *gemm* is performed separately <span class="kw">for</span> a batch of matrices. The column indices of the matrices |
| are given by the last dimensions of the tensors, the row indices by the axis specified <span class="kw">with</span> the *axis* |
| parameter. By default, the trailing two dimensions will be used <span class="kw">for</span> matrix encoding. |
| |
| For a non-default axis parameter, the operation performed is equivalent to a series of swapaxes/gemm/swapaxes |
| calls. For example let *A*, *B*, *C* be <span class="num">5</span> dimensional tensors. Then gemm(*A*, *B*, *C*, axis=<span class="num">1</span>) is equivalent |
| to the following without the overhead of the additional swapaxis operations:: |
| |
| A1 = swapaxes(A, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| B1 = swapaxes(B, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| C = swapaxes(C, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| C = gemm(A1, B1, C) |
| C = swapaxis(C, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| |
| When the input data is of <span class="kw">type</span> float32 and the environment variables MXNET_CUDA_ALLOW_TENSOR_CORE |
| and MXNET_CUDA_TENSOR_OP_MATH_ALLOW_CONVERSION are set to <span class="num">1</span>, <span class="kw">this</span> operator will <span class="kw">try</span> to use |
| pseudo-float16 precision (float32 math <span class="kw">with</span> float16 I/O) precision in order to use |
| Tensor Cores on suitable NVIDIA GPUs. This can sometimes give significant speedups. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix multiply-add |
| A = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| B = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| C = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| gemm(A, B, C, transpose_b=True, alpha=<span class="num">2.0</span>, beta=<span class="num">10.0</span>) |
| = `[ [<span class="num">14.0</span>, <span class="num">14.0</span>, <span class="num">14.0</span>], [<span class="num">14.0</span>, <span class="num">14.0</span>, <span class="num">14.0</span>] ] |
| |
| Batch matrix multiply-add |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| B = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| C = `[ `[ [<span class="num">10.0</span>] ], `[ [<span class="num">0.01</span>] ] ] |
| gemm(A, B, C, transpose_b=True, alpha=<span class="num">2.0</span> , beta=<span class="num">10.0</span>) |
| = `[ `[ [<span class="num">104.0</span>] ], `[ [<span class="num">0.14</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L88</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_gemm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_gemm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_gemm(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_gemm</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_gemm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs general matrix multiplication and accumulation. |
| Input are tensors *A*, *B*, *C*, each of dimension *n >= <span class="num">2</span>* and having the same shape |
| on the leading *n-<span class="num">2</span>* dimensions. |
| |
| If *n=<span class="num">2</span>*, the BLAS3 function *gemm* is performed: |
| |
| *out* = *alpha* \* *op*\ (*A*) \* *op*\ (*B*) + *beta* \* *C* |
| |
| Here, *alpha* and *beta* are scalar parameters, and *op()* is either the identity or |
| matrix transposition (depending on *transpose_a*, *transpose_b*). |
| |
| If *n><span class="num">2</span>*, *gemm* is performed separately <span class="kw">for</span> a batch of matrices. The column indices of the matrices |
| are given by the last dimensions of the tensors, the row indices by the axis specified <span class="kw">with</span> the *axis* |
| parameter. By default, the trailing two dimensions will be used <span class="kw">for</span> matrix encoding. |
| |
| For a non-default axis parameter, the operation performed is equivalent to a series of swapaxes/gemm/swapaxes |
| calls. For example let *A*, *B*, *C* be <span class="num">5</span> dimensional tensors. Then gemm(*A*, *B*, *C*, axis=<span class="num">1</span>) is equivalent |
| to the following without the overhead of the additional swapaxis operations:: |
| |
| A1 = swapaxes(A, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| B1 = swapaxes(B, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| C = swapaxes(C, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| C = gemm(A1, B1, C) |
| C = swapaxis(C, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| |
| When the input data is of <span class="kw">type</span> float32 and the environment variables MXNET_CUDA_ALLOW_TENSOR_CORE |
| and MXNET_CUDA_TENSOR_OP_MATH_ALLOW_CONVERSION are set to <span class="num">1</span>, <span class="kw">this</span> operator will <span class="kw">try</span> to use |
| pseudo-float16 precision (float32 math <span class="kw">with</span> float16 I/O) precision in order to use |
| Tensor Cores on suitable NVIDIA GPUs. This can sometimes give significant speedups. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix multiply-add |
| A = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| B = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| C = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| gemm(A, B, C, transpose_b=True, alpha=<span class="num">2.0</span>, beta=<span class="num">10.0</span>) |
| = `[ [<span class="num">14.0</span>, <span class="num">14.0</span>, <span class="num">14.0</span>], [<span class="num">14.0</span>, <span class="num">14.0</span>, <span class="num">14.0</span>] ] |
| |
| Batch matrix multiply-add |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| B = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| C = `[ `[ [<span class="num">10.0</span>] ], `[ [<span class="num">0.01</span>] ] ] |
| gemm(A, B, C, transpose_b=True, alpha=<span class="num">2.0</span> , beta=<span class="num">10.0</span>) |
| = `[ `[ [<span class="num">104.0</span>] ], `[ [<span class="num">0.14</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L88</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_gemm2" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_gemm2(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_gemm2(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_gemm2</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_gemm2(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs general matrix multiplication. |
| Input are tensors *A*, *B*, each of dimension *n >= <span class="num">2</span>* and having the same shape |
| on the leading *n-<span class="num">2</span>* dimensions. |
| |
| If *n=<span class="num">2</span>*, the BLAS3 function *gemm* is performed: |
| |
| *out* = *alpha* \* *op*\ (*A*) \* *op*\ (*B*) |
| |
| Here *alpha* is a scalar parameter and *op()* is either the identity or the matrix |
| transposition (depending on *transpose_a*, *transpose_b*). |
| |
| If *n><span class="num">2</span>*, *gemm* is performed separately <span class="kw">for</span> a batch of matrices. The column indices of the matrices |
| are given by the last dimensions of the tensors, the row indices by the axis specified <span class="kw">with</span> the *axis* |
| parameter. By default, the trailing two dimensions will be used <span class="kw">for</span> matrix encoding. |
| |
| For a non-default axis parameter, the operation performed is equivalent to a series of swapaxes/gemm/swapaxes |
| calls. For example let *A*, *B* be <span class="num">5</span> dimensional tensors. Then gemm(*A*, *B*, axis=<span class="num">1</span>) is equivalent to |
| the following without the overhead of the additional swapaxis operations:: |
| |
| A1 = swapaxes(A, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| B1 = swapaxes(B, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| C = gemm2(A1, B1) |
| C = swapaxis(C, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| |
| When the input data is of <span class="kw">type</span> float32 and the environment variables MXNET_CUDA_ALLOW_TENSOR_CORE |
| and MXNET_CUDA_TENSOR_OP_MATH_ALLOW_CONVERSION are set to <span class="num">1</span>, <span class="kw">this</span> operator will <span class="kw">try</span> to use |
| pseudo-float16 precision (float32 math <span class="kw">with</span> float16 I/O) precision in order to use |
| Tensor Cores on suitable NVIDIA GPUs. This can sometimes give significant speedups. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix multiply |
| A = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| B = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| gemm2(A, B, transpose_b=True, alpha=<span class="num">2.0</span>) |
| = `[ [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ] |
| |
| Batch matrix multiply |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| B = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| gemm2(A, B, transpose_b=True, alpha=<span class="num">2.0</span>) |
| = `[ `[ [<span class="num">4.0</span>] ], `[ [<span class="num">0.04</span> ] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L162</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_gemm2" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_gemm2(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_gemm2(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_gemm2</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_gemm2(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs general matrix multiplication. |
| Input are tensors *A*, *B*, each of dimension *n >= <span class="num">2</span>* and having the same shape |
| on the leading *n-<span class="num">2</span>* dimensions. |
| |
| If *n=<span class="num">2</span>*, the BLAS3 function *gemm* is performed: |
| |
| *out* = *alpha* \* *op*\ (*A*) \* *op*\ (*B*) |
| |
| Here *alpha* is a scalar parameter and *op()* is either the identity or the matrix |
| transposition (depending on *transpose_a*, *transpose_b*). |
| |
| If *n><span class="num">2</span>*, *gemm* is performed separately <span class="kw">for</span> a batch of matrices. The column indices of the matrices |
| are given by the last dimensions of the tensors, the row indices by the axis specified <span class="kw">with</span> the *axis* |
| parameter. By default, the trailing two dimensions will be used <span class="kw">for</span> matrix encoding. |
| |
| For a non-default axis parameter, the operation performed is equivalent to a series of swapaxes/gemm/swapaxes |
| calls. For example let *A*, *B* be <span class="num">5</span> dimensional tensors. Then gemm(*A*, *B*, axis=<span class="num">1</span>) is equivalent to |
| the following without the overhead of the additional swapaxis operations:: |
| |
| A1 = swapaxes(A, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| B1 = swapaxes(B, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| C = gemm2(A1, B1) |
| C = swapaxis(C, dim1=<span class="num">1</span>, dim2=<span class="num">3</span>) |
| |
| When the input data is of <span class="kw">type</span> float32 and the environment variables MXNET_CUDA_ALLOW_TENSOR_CORE |
| and MXNET_CUDA_TENSOR_OP_MATH_ALLOW_CONVERSION are set to <span class="num">1</span>, <span class="kw">this</span> operator will <span class="kw">try</span> to use |
| pseudo-float16 precision (float32 math <span class="kw">with</span> float16 I/O) precision in order to use |
| Tensor Cores on suitable NVIDIA GPUs. This can sometimes give significant speedups. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix multiply |
| A = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| B = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| gemm2(A, B, transpose_b=True, alpha=<span class="num">2.0</span>) |
| = `[ [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ] |
| |
| Batch matrix multiply |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| B = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| gemm2(A, B, transpose_b=True, alpha=<span class="num">2.0</span>) |
| = `[ `[ [<span class="num">4.0</span>] ], `[ [<span class="num">0.04</span> ] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L162</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_inverse" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_inverse(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_inverse(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_inverse</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_inverse(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the inverse of a matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* is a square matrix. We compute: |
| |
| *out* = *A*\ :sup:`-<span class="num">1</span>` |
| |
| If *n><span class="num">2</span>*, *inverse* is performed separately on the trailing two dimensions |
| <span class="kw">for</span> all inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix inverse |
| A = `[ [<span class="num">1.</span>, <span class="num">4.</span>], [<span class="num">2.</span>, <span class="num">3.</span>] ] |
| inverse(A) = `[ [-<span class="num">0.6</span>, <span class="num">0.8</span>], [<span class="num">0.4</span>, -<span class="num">0.2</span>] ] |
| |
| Batch matrix inverse |
| A = `[ `[ [<span class="num">1.</span>, <span class="num">4.</span>], [<span class="num">2.</span>, <span class="num">3.</span>] ], |
| `[ [<span class="num">1.</span>, <span class="num">3.</span>], [<span class="num">2.</span>, <span class="num">4.</span>] ] ] |
| inverse(A) = `[ `[ [-<span class="num">0.6</span>, <span class="num">0.8</span>], [<span class="num">0.4</span>, -<span class="num">0.2</span>] ], |
| `[ [-<span class="num">2.</span>, <span class="num">1.5</span>], [<span class="num">1.</span>, -<span class="num">0.5</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L919</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_inverse" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_inverse(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_inverse(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_inverse</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_inverse(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the inverse of a matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* is a square matrix. We compute: |
| |
| *out* = *A*\ :sup:`-<span class="num">1</span>` |
| |
| If *n><span class="num">2</span>*, *inverse* is performed separately on the trailing two dimensions |
| <span class="kw">for</span> all inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix inverse |
| A = `[ [<span class="num">1.</span>, <span class="num">4.</span>], [<span class="num">2.</span>, <span class="num">3.</span>] ] |
| inverse(A) = `[ [-<span class="num">0.6</span>, <span class="num">0.8</span>], [<span class="num">0.4</span>, -<span class="num">0.2</span>] ] |
| |
| Batch matrix inverse |
| A = `[ `[ [<span class="num">1.</span>, <span class="num">4.</span>], [<span class="num">2.</span>, <span class="num">3.</span>] ], |
| `[ [<span class="num">1.</span>, <span class="num">3.</span>], [<span class="num">2.</span>, <span class="num">4.</span>] ] ] |
| inverse(A) = `[ `[ [-<span class="num">0.6</span>, <span class="num">0.8</span>], [<span class="num">0.4</span>, -<span class="num">0.2</span>] ], |
| `[ [-<span class="num">2.</span>, <span class="num">1.5</span>], [<span class="num">1.</span>, -<span class="num">0.5</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L919</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_makediag" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_makediag(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_makediag(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_makediag</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_makediag(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Constructs a square matrix <span class="kw">with</span> the input as diagonal. |
| Input is a tensor *A* of dimension *n >= <span class="num">1</span>*. |
| |
| If *n=<span class="num">1</span>*, then *A* represents the diagonal entries of a single square matrix. This matrix will be returned as a <span class="num">2</span>-dimensional tensor. |
| If *n><span class="num">1</span>*, then *A* represents a batch of diagonals of square matrices. The batch of diagonal matrices will be returned as an *n+<span class="num">1</span>*-dimensional tensor. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single diagonal matrix construction |
| A = [<span class="num">1.0</span>, <span class="num">2.0</span>] |
| |
| makediag(A) = `[ [<span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">0.0</span>, <span class="num">2.0</span>] ] |
| |
| makediag(A, <span class="num">1</span>) = `[ [<span class="num">0.0</span>, <span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">2.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>] ] |
| |
| Batch diagonal matrix construction |
| A = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ] |
| |
| makediag(A) = `[ `[ [<span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">0.0</span>, <span class="num">2.0</span>] ], |
| `[ [<span class="num">3.0</span>, <span class="num">0.0</span>], |
| [<span class="num">0.0</span>, <span class="num">4.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L546</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_makediag" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_makediag(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_makediag(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_makediag</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_makediag(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Constructs a square matrix <span class="kw">with</span> the input as diagonal. |
| Input is a tensor *A* of dimension *n >= <span class="num">1</span>*. |
| |
| If *n=<span class="num">1</span>*, then *A* represents the diagonal entries of a single square matrix. This matrix will be returned as a <span class="num">2</span>-dimensional tensor. |
| If *n><span class="num">1</span>*, then *A* represents a batch of diagonals of square matrices. The batch of diagonal matrices will be returned as an *n+<span class="num">1</span>*-dimensional tensor. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single diagonal matrix construction |
| A = [<span class="num">1.0</span>, <span class="num">2.0</span>] |
| |
| makediag(A) = `[ [<span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">0.0</span>, <span class="num">2.0</span>] ] |
| |
| makediag(A, <span class="num">1</span>) = `[ [<span class="num">0.0</span>, <span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">2.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>] ] |
| |
| Batch diagonal matrix construction |
| A = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">3.0</span>, <span class="num">4.0</span>] ] |
| |
| makediag(A) = `[ `[ [<span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">0.0</span>, <span class="num">2.0</span>] ], |
| `[ [<span class="num">3.0</span>, <span class="num">0.0</span>], |
| [<span class="num">0.0</span>, <span class="num">4.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L546</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_maketrian" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_maketrian(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_maketrian(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_maketrian</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_maketrian(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Constructs a square matrix <span class="kw">with</span> the input representing a specific triangular sub-matrix. |
| This is basically the inverse of *linalg.extracttrian*. Input is a tensor *A* of dimension *n >= <span class="num">1</span>*. |
| |
| If *n=<span class="num">1</span>*, then *A* represents the entries of a triangular matrix which is lower triangular <span class="kw">if</span> *offset<<span class="num">0</span>* or *offset=<span class="num">0</span>*, *lower=<span class="kw">true</span>*. The resulting matrix is derived by first constructing the square |
| matrix <span class="kw">with</span> the entries outside the triangle set to zero and then adding *offset*-times an additional |
| diagonal <span class="kw">with</span> zero entries to the square matrix. |
| |
| If *n><span class="num">1</span>*, then *A* represents a batch of triangular sub-matrices. The batch of corresponding square matrices is returned as an *n+<span class="num">1</span>*-dimensional tensor. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix construction |
| A = [<span class="num">1.0</span>, <span class="num">2.0</span>, <span class="num">3.0</span>] |
| |
| maketrian(A) = `[ [<span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">2.0</span>, <span class="num">3.0</span>] ] |
| |
| maketrian(A, lower=<span class="kw">false</span>) = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">0.0</span>, <span class="num">3.0</span>] ] |
| |
| maketrian(A, offset=<span class="num">1</span>) = `[ [<span class="num">0.0</span>, <span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">3.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>] ] |
| maketrian(A, offset=-<span class="num">1</span>) = `[ [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>], |
| [<span class="num">1.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>], |
| [<span class="num">2.0</span>, <span class="num">3.0</span>, <span class="num">0.0</span>] ] |
| |
| Batch matrix construction |
| A = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>, <span class="num">3.0</span>], |
| [<span class="num">4.0</span>, <span class="num">5.0</span>, <span class="num">6.0</span>] ] |
| |
| maketrian(A) = `[ `[ [<span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">2.0</span>, <span class="num">3.0</span>] ], |
| `[ [<span class="num">4.0</span>, <span class="num">0.0</span>], |
| [<span class="num">5.0</span>, <span class="num">6.0</span>] ] ] |
| |
| maketrian(A, offset=<span class="num">1</span>) = `[ `[ [<span class="num">0.0</span>, <span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">3.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>] ], |
| `[ [<span class="num">0.0</span>, <span class="num">4.0</span>, <span class="num">5.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">6.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L672</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_maketrian" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_maketrian(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_maketrian(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_maketrian</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_maketrian(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Constructs a square matrix <span class="kw">with</span> the input representing a specific triangular sub-matrix. |
| This is basically the inverse of *linalg.extracttrian*. Input is a tensor *A* of dimension *n >= <span class="num">1</span>*. |
| |
| If *n=<span class="num">1</span>*, then *A* represents the entries of a triangular matrix which is lower triangular <span class="kw">if</span> *offset<<span class="num">0</span>* or *offset=<span class="num">0</span>*, *lower=<span class="kw">true</span>*. The resulting matrix is derived by first constructing the square |
| matrix <span class="kw">with</span> the entries outside the triangle set to zero and then adding *offset*-times an additional |
| diagonal <span class="kw">with</span> zero entries to the square matrix. |
| |
| If *n><span class="num">1</span>*, then *A* represents a batch of triangular sub-matrices. The batch of corresponding square matrices is returned as an *n+<span class="num">1</span>*-dimensional tensor. |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix construction |
| A = [<span class="num">1.0</span>, <span class="num">2.0</span>, <span class="num">3.0</span>] |
| |
| maketrian(A) = `[ [<span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">2.0</span>, <span class="num">3.0</span>] ] |
| |
| maketrian(A, lower=<span class="kw">false</span>) = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">0.0</span>, <span class="num">3.0</span>] ] |
| |
| maketrian(A, offset=<span class="num">1</span>) = `[ [<span class="num">0.0</span>, <span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">3.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>] ] |
| maketrian(A, offset=-<span class="num">1</span>) = `[ [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>], |
| [<span class="num">1.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>], |
| [<span class="num">2.0</span>, <span class="num">3.0</span>, <span class="num">0.0</span>] ] |
| |
| Batch matrix construction |
| A = `[ [<span class="num">1.0</span>, <span class="num">2.0</span>, <span class="num">3.0</span>], |
| [<span class="num">4.0</span>, <span class="num">5.0</span>, <span class="num">6.0</span>] ] |
| |
| maketrian(A) = `[ `[ [<span class="num">1.0</span>, <span class="num">0.0</span>], |
| [<span class="num">2.0</span>, <span class="num">3.0</span>] ], |
| `[ [<span class="num">4.0</span>, <span class="num">0.0</span>], |
| [<span class="num">5.0</span>, <span class="num">6.0</span>] ] ] |
| |
| maketrian(A, offset=<span class="num">1</span>) = `[ `[ [<span class="num">0.0</span>, <span class="num">1.0</span>, <span class="num">2.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">3.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>] ], |
| `[ [<span class="num">0.0</span>, <span class="num">4.0</span>, <span class="num">5.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">6.0</span>], |
| [<span class="num">0.0</span>, <span class="num">0.0</span>, <span class="num">0.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L672</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_potrf" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_potrf(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_potrf(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_potrf</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_potrf(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs Cholesky factorization of a symmetric positive-definite matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, the Cholesky factor *B* of the symmetric, positive definite matrix *A* is |
| computed. *B* is triangular (entries of upper or lower triangle are all zero), has |
| positive diagonal entries, and: |
| |
| *A* = *B* \* *B*\ :sup:`T` <span class="kw">if</span> *lower* = *<span class="kw">true</span>* |
| *A* = *B*\ :sup:`T` \* *B* <span class="kw">if</span> *lower* = *<span class="kw">false</span>* |
| |
| If *n><span class="num">2</span>*, *potrf* is performed separately on the trailing two dimensions <span class="kw">for</span> all inputs |
| (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix factorization |
| A = `[ [<span class="num">4.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">4.25</span>] ] |
| potrf(A) = `[ [<span class="num">2.0</span>, <span class="num">0</span>], [<span class="num">0.5</span>, <span class="num">2.0</span>] ] |
| |
| Batch matrix factorization |
| A = `[ `[ [<span class="num">4.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">4.25</span>] ], `[ [<span class="num">16.0</span>, <span class="num">4.0</span>], [<span class="num">4.0</span>, <span class="num">17.0</span>] ] ] |
| potrf(A) = `[ `[ [<span class="num">2.0</span>, <span class="num">0</span>], [<span class="num">0.5</span>, <span class="num">2.0</span>] ], `[ [<span class="num">4.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">4.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L213</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_potrf" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_potrf(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_potrf(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_potrf</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_potrf(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs Cholesky factorization of a symmetric positive-definite matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, the Cholesky factor *B* of the symmetric, positive definite matrix *A* is |
| computed. *B* is triangular (entries of upper or lower triangle are all zero), has |
| positive diagonal entries, and: |
| |
| *A* = *B* \* *B*\ :sup:`T` <span class="kw">if</span> *lower* = *<span class="kw">true</span>* |
| *A* = *B*\ :sup:`T` \* *B* <span class="kw">if</span> *lower* = *<span class="kw">false</span>* |
| |
| If *n><span class="num">2</span>*, *potrf* is performed separately on the trailing two dimensions <span class="kw">for</span> all inputs |
| (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix factorization |
| A = `[ [<span class="num">4.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">4.25</span>] ] |
| potrf(A) = `[ [<span class="num">2.0</span>, <span class="num">0</span>], [<span class="num">0.5</span>, <span class="num">2.0</span>] ] |
| |
| Batch matrix factorization |
| A = `[ `[ [<span class="num">4.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">4.25</span>] ], `[ [<span class="num">16.0</span>, <span class="num">4.0</span>], [<span class="num">4.0</span>, <span class="num">17.0</span>] ] ] |
| potrf(A) = `[ `[ [<span class="num">2.0</span>, <span class="num">0</span>], [<span class="num">0.5</span>, <span class="num">2.0</span>] ], `[ [<span class="num">4.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">4.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L213</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_potri" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_potri(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_potri(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_potri</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_potri(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs matrix inversion from a Cholesky factorization. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* is a triangular matrix (entries of upper or lower triangle are all zero) |
| <span class="kw">with</span> positive diagonal. We compute: |
| |
| *out* = *A*\ :sup:`-T` \* *A*\ :sup:`-<span class="num">1</span>` <span class="kw">if</span> *lower* = *<span class="kw">true</span>* |
| *out* = *A*\ :sup:`-<span class="num">1</span>` \* *A*\ :sup:`-T` <span class="kw">if</span> *lower* = *<span class="kw">false</span>* |
| |
| In other words, <span class="kw">if</span> *A* is the Cholesky factor of a symmetric positive definite matrix |
| *B* (obtained by *potrf*), then |
| |
| *out* = *B*\ :sup:`-<span class="num">1</span>` |
| |
| If *n><span class="num">2</span>*, *potri* is performed separately on the trailing two dimensions <span class="kw">for</span> all inputs |
| (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| .. note:: Use <span class="kw">this</span> operator only <span class="kw">if</span> you are certain you need the inverse of *B*, and |
| cannot use the Cholesky factor *A* (*potrf*), together <span class="kw">with</span> backsubstitution |
| (*trsm*). The latter is numerically much safer, and also cheaper. |
| |
| Examples:: |
| |
| Single matrix inverse |
| A = `[ [<span class="num">2.0</span>, <span class="num">0</span>], [<span class="num">0.5</span>, <span class="num">2.0</span>] ] |
| potri(A) = `[ [<span class="num">0.26563</span>, -<span class="num">0.0625</span>], [-<span class="num">0.0625</span>, <span class="num">0.25</span>] ] |
| |
| Batch matrix inverse |
| A = `[ `[ [<span class="num">2.0</span>, <span class="num">0</span>], [<span class="num">0.5</span>, <span class="num">2.0</span>] ], `[ [<span class="num">4.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">4.0</span>] ] ] |
| potri(A) = `[ `[ [<span class="num">0.26563</span>, -<span class="num">0.0625</span>], [-<span class="num">0.0625</span>, <span class="num">0.25</span>] ], |
| `[ [<span class="num">0.06641</span>, -<span class="num">0.01562</span>], [-<span class="num">0.01562</span>, <span class="num">0</span>,<span class="num">0625</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L274</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_potri" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_potri(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_potri(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_potri</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_potri(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs matrix inversion from a Cholesky factorization. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* is a triangular matrix (entries of upper or lower triangle are all zero) |
| <span class="kw">with</span> positive diagonal. We compute: |
| |
| *out* = *A*\ :sup:`-T` \* *A*\ :sup:`-<span class="num">1</span>` <span class="kw">if</span> *lower* = *<span class="kw">true</span>* |
| *out* = *A*\ :sup:`-<span class="num">1</span>` \* *A*\ :sup:`-T` <span class="kw">if</span> *lower* = *<span class="kw">false</span>* |
| |
| In other words, <span class="kw">if</span> *A* is the Cholesky factor of a symmetric positive definite matrix |
| *B* (obtained by *potrf*), then |
| |
| *out* = *B*\ :sup:`-<span class="num">1</span>` |
| |
| If *n><span class="num">2</span>*, *potri* is performed separately on the trailing two dimensions <span class="kw">for</span> all inputs |
| (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| .. note:: Use <span class="kw">this</span> operator only <span class="kw">if</span> you are certain you need the inverse of *B*, and |
| cannot use the Cholesky factor *A* (*potrf*), together <span class="kw">with</span> backsubstitution |
| (*trsm*). The latter is numerically much safer, and also cheaper. |
| |
| Examples:: |
| |
| Single matrix inverse |
| A = `[ [<span class="num">2.0</span>, <span class="num">0</span>], [<span class="num">0.5</span>, <span class="num">2.0</span>] ] |
| potri(A) = `[ [<span class="num">0.26563</span>, -<span class="num">0.0625</span>], [-<span class="num">0.0625</span>, <span class="num">0.25</span>] ] |
| |
| Batch matrix inverse |
| A = `[ `[ [<span class="num">2.0</span>, <span class="num">0</span>], [<span class="num">0.5</span>, <span class="num">2.0</span>] ], `[ [<span class="num">4.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">4.0</span>] ] ] |
| potri(A) = `[ `[ [<span class="num">0.26563</span>, -<span class="num">0.0625</span>], [-<span class="num">0.0625</span>, <span class="num">0.25</span>] ], |
| `[ [<span class="num">0.06641</span>, -<span class="num">0.01562</span>], [-<span class="num">0.01562</span>, <span class="num">0</span>,<span class="num">0625</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L274</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_slogdet" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_slogdet(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_slogdet(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_slogdet</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_slogdet(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the sign and log of the determinant of a matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* is a square matrix. We compute: |
| |
| *sign* = *sign(det(A))* |
| *logabsdet* = *log(abs(det(A)))* |
| |
| If *n><span class="num">2</span>*, *slogdet* is performed separately on the trailing two dimensions |
| <span class="kw">for</span> all inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| .. note:: The gradient is not properly defined on sign, so the gradient of |
| it is not backwarded. |
| .. note:: No gradient is backwarded when A is non-invertible. Please see |
| the docs of operator det <span class="kw">for</span> detail. |
| |
| Examples:: |
| |
| Single matrix signed log determinant |
| A = `[ [<span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">1.</span>, <span class="num">4.</span>] ] |
| sign, logabsdet = slogdet(A) |
| sign = [<span class="num">1.</span>] |
| logabsdet = [<span class="num">1.609438</span>] |
| |
| Batch matrix signed log determinant |
| A = `[ `[ [<span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">1.</span>, <span class="num">4.</span>] ], |
| `[ [<span class="num">1.</span>, <span class="num">2.</span>], [<span class="num">2.</span>, <span class="num">4.</span>] ], |
| `[ [<span class="num">1.</span>, <span class="num">2.</span>], [<span class="num">4.</span>, <span class="num">3.</span>] ] ] |
| sign, logabsdet = slogdet(A) |
| sign = [<span class="num">1.</span>, <span class="num">0.</span>, -<span class="num">1.</span>] |
| logabsdet = [<span class="num">1.609438</span>, -inf, <span class="num">1.609438</span>] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L1033</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_slogdet" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_slogdet(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_slogdet(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_slogdet</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_slogdet(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the sign and log of the determinant of a matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* is a square matrix. We compute: |
| |
| *sign* = *sign(det(A))* |
| *logabsdet* = *log(abs(det(A)))* |
| |
| If *n><span class="num">2</span>*, *slogdet* is performed separately on the trailing two dimensions |
| <span class="kw">for</span> all inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| .. note:: The gradient is not properly defined on sign, so the gradient of |
| it is not backwarded. |
| .. note:: No gradient is backwarded when A is non-invertible. Please see |
| the docs of operator det <span class="kw">for</span> detail. |
| |
| Examples:: |
| |
| Single matrix signed log determinant |
| A = `[ [<span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">1.</span>, <span class="num">4.</span>] ] |
| sign, logabsdet = slogdet(A) |
| sign = [<span class="num">1.</span>] |
| logabsdet = [<span class="num">1.609438</span>] |
| |
| Batch matrix signed log determinant |
| A = `[ `[ [<span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">1.</span>, <span class="num">4.</span>] ], |
| `[ [<span class="num">1.</span>, <span class="num">2.</span>], [<span class="num">2.</span>, <span class="num">4.</span>] ], |
| `[ [<span class="num">1.</span>, <span class="num">2.</span>], [<span class="num">4.</span>, <span class="num">3.</span>] ] ] |
| sign, logabsdet = slogdet(A) |
| sign = [<span class="num">1.</span>, <span class="num">0.</span>, -<span class="num">1.</span>] |
| logabsdet = [<span class="num">1.609438</span>, -inf, <span class="num">1.609438</span>] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L1033</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_sumlogdiag" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_sumlogdiag(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_sumlogdiag(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_sumlogdiag</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_sumlogdiag(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the sum of the logarithms of the diagonal elements of a square matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* must be square <span class="kw">with</span> positive diagonal entries. We sum the natural |
| logarithms of the diagonal elements, the result has shape (<span class="num">1</span>,). |
| |
| If *n><span class="num">2</span>*, *sumlogdiag* is performed separately on the trailing two dimensions <span class="kw">for</span> all |
| inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix reduction |
| A = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">7.0</span>] ] |
| sumlogdiag(A) = [<span class="num">1.9459</span>] |
| |
| Batch matrix reduction |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">7.0</span>] ], `[ [<span class="num">3.0</span>, <span class="num">0</span>], [<span class="num">0</span>, <span class="num">17.0</span>] ] ] |
| sumlogdiag(A) = [<span class="num">1.9459</span>, <span class="num">3.9318</span>] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L444</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_sumlogdiag" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_sumlogdiag(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_sumlogdiag(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_sumlogdiag</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_sumlogdiag(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the sum of the logarithms of the diagonal elements of a square matrix. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, *A* must be square <span class="kw">with</span> positive diagonal entries. We sum the natural |
| logarithms of the diagonal elements, the result has shape (<span class="num">1</span>,). |
| |
| If *n><span class="num">2</span>*, *sumlogdiag* is performed separately on the trailing two dimensions <span class="kw">for</span> all |
| inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix reduction |
| A = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">7.0</span>] ] |
| sumlogdiag(A) = [<span class="num">1.9459</span>] |
| |
| Batch matrix reduction |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">7.0</span>] ], `[ [<span class="num">3.0</span>, <span class="num">0</span>], [<span class="num">0</span>, <span class="num">17.0</span>] ] ] |
| sumlogdiag(A) = [<span class="num">1.9459</span>, <span class="num">3.9318</span>] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L444</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_syrk" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_syrk(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_syrk(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_syrk</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_syrk(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Multiplication of matrix <span class="kw">with</span> its transpose. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, the operator performs the BLAS3 function *syrk*: |
| |
| *out* = *alpha* \* *A* \* *A*\ :sup:`T` |
| |
| <span class="kw">if</span> *transpose=False*, or |
| |
| *out* = *alpha* \* *A*\ :sup:`T` \ \* *A* |
| |
| <span class="kw">if</span> *transpose=True*. |
| |
| If *n><span class="num">2</span>*, *syrk* is performed separately on the trailing two dimensions <span class="kw">for</span> all |
| inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix multiply |
| A = `[ [<span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ] |
| syrk(A, alpha=<span class="num">1.</span>, transpose=False) |
| = `[ [<span class="num">14.</span>, <span class="num">32.</span>], |
| [<span class="num">32.</span>, <span class="num">77.</span>] ] |
| syrk(A, alpha=<span class="num">1.</span>, transpose=True) |
| = `[ [<span class="num">17.</span>, <span class="num">22.</span>, <span class="num">27.</span>], |
| [<span class="num">22.</span>, <span class="num">29.</span>, <span class="num">36.</span>], |
| [<span class="num">27.</span>, <span class="num">36.</span>, <span class="num">45.</span>] ] |
| |
| Batch matrix multiply |
| A = `[ `[ [<span class="num">1.</span>, <span class="num">1.</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| syrk(A, alpha=<span class="num">2.</span>, transpose=False) = `[ `[ [<span class="num">4.</span>] ], `[ [<span class="num">0.04</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L729</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_syrk" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_syrk(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_syrk(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_syrk</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_syrk(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Multiplication of matrix <span class="kw">with</span> its transpose. |
| Input is a tensor *A* of dimension *n >= <span class="num">2</span>*. |
| |
| If *n=<span class="num">2</span>*, the operator performs the BLAS3 function *syrk*: |
| |
| *out* = *alpha* \* *A* \* *A*\ :sup:`T` |
| |
| <span class="kw">if</span> *transpose=False*, or |
| |
| *out* = *alpha* \* *A*\ :sup:`T` \ \* *A* |
| |
| <span class="kw">if</span> *transpose=True*. |
| |
| If *n><span class="num">2</span>*, *syrk* is performed separately on the trailing two dimensions <span class="kw">for</span> all |
| inputs (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix multiply |
| A = `[ [<span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>], [<span class="num">4.</span>, <span class="num">5.</span>, <span class="num">6.</span>] ] |
| syrk(A, alpha=<span class="num">1.</span>, transpose=False) |
| = `[ [<span class="num">14.</span>, <span class="num">32.</span>], |
| [<span class="num">32.</span>, <span class="num">77.</span>] ] |
| syrk(A, alpha=<span class="num">1.</span>, transpose=True) |
| = `[ [<span class="num">17.</span>, <span class="num">22.</span>, <span class="num">27.</span>], |
| [<span class="num">22.</span>, <span class="num">29.</span>, <span class="num">36.</span>], |
| [<span class="num">27.</span>, <span class="num">36.</span>, <span class="num">45.</span>] ] |
| |
| Batch matrix multiply |
| A = `[ `[ [<span class="num">1.</span>, <span class="num">1.</span>] ], `[ [<span class="num">0.1</span>, <span class="num">0.1</span>] ] ] |
| syrk(A, alpha=<span class="num">2.</span>, transpose=False) = `[ `[ [<span class="num">4.</span>] ], `[ [<span class="num">0.04</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L729</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_trmm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_trmm(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_trmm(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_trmm</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_trmm(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs multiplication <span class="kw">with</span> a lower triangular matrix. |
| Input are tensors *A*, *B*, each of dimension *n >= <span class="num">2</span>* and having the same shape |
| on the leading *n-<span class="num">2</span>* dimensions. |
| |
| If *n=<span class="num">2</span>*, *A* must be triangular. The operator performs the BLAS3 function |
| *trmm*: |
| |
| *out* = *alpha* \* *op*\ (*A*) \* *B* |
| |
| <span class="kw">if</span> *rightside=False*, or |
| |
| *out* = *alpha* \* *B* \* *op*\ (*A*) |
| |
| <span class="kw">if</span> *rightside=True*. Here, *alpha* is a scalar parameter, and *op()* is either the |
| identity or the matrix transposition (depending on *transpose*). |
| |
| If *n><span class="num">2</span>*, *trmm* is performed separately on the trailing two dimensions <span class="kw">for</span> all inputs |
| (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single triangular matrix multiply |
| A = `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| B = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| trmm(A, B, alpha=<span class="num">2.0</span>) = `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ] |
| |
| Batch triangular matrix multiply |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] ] |
| B = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.5</span>, <span class="num">0.5</span>, <span class="num">0.5</span>], [<span class="num">0.5</span>, <span class="num">0.5</span>, <span class="num">0.5</span>] ] ] |
| trmm(A, B, alpha=<span class="num">2.0</span>) = `[ `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ], |
| `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L332</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_trmm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_trmm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_trmm(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_trmm</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_trmm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Performs multiplication <span class="kw">with</span> a lower triangular matrix. |
| Input are tensors *A*, *B*, each of dimension *n >= <span class="num">2</span>* and having the same shape |
| on the leading *n-<span class="num">2</span>* dimensions. |
| |
| If *n=<span class="num">2</span>*, *A* must be triangular. The operator performs the BLAS3 function |
| *trmm*: |
| |
| *out* = *alpha* \* *op*\ (*A*) \* *B* |
| |
| <span class="kw">if</span> *rightside=False*, or |
| |
| *out* = *alpha* \* *B* \* *op*\ (*A*) |
| |
| <span class="kw">if</span> *rightside=True*. Here, *alpha* is a scalar parameter, and *op()* is either the |
| identity or the matrix transposition (depending on *transpose*). |
| |
| If *n><span class="num">2</span>*, *trmm* is performed separately on the trailing two dimensions <span class="kw">for</span> all inputs |
| (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single triangular matrix multiply |
| A = `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| B = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| trmm(A, B, alpha=<span class="num">2.0</span>) = `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ] |
| |
| Batch triangular matrix multiply |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] ] |
| B = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">0.5</span>, <span class="num">0.5</span>, <span class="num">0.5</span>], [<span class="num">0.5</span>, <span class="num">0.5</span>, <span class="num">0.5</span>] ] ] |
| trmm(A, B, alpha=<span class="num">2.0</span>) = `[ `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ], |
| `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L332</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_trsm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_trsm(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_trsm(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_trsm</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_trsm(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Solves matrix equation involving a lower triangular matrix. |
| Input are tensors *A*, *B*, each of dimension *n >= <span class="num">2</span>* and having the same shape |
| on the leading *n-<span class="num">2</span>* dimensions. |
| |
| If *n=<span class="num">2</span>*, *A* must be triangular. The operator performs the BLAS3 function |
| *trsm*, solving <span class="kw">for</span> *out* in: |
| |
| *op*\ (*A*) \* *out* = *alpha* \* *B* |
| |
| <span class="kw">if</span> *rightside=False*, or |
| |
| *out* \* *op*\ (*A*) = *alpha* \* *B* |
| |
| <span class="kw">if</span> *rightside=True*. Here, *alpha* is a scalar parameter, and *op()* is either the |
| identity or the matrix transposition (depending on *transpose*). |
| |
| If *n><span class="num">2</span>*, *trsm* is performed separately on the trailing two dimensions <span class="kw">for</span> all inputs |
| (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix solve |
| A = `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| B = `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ] |
| trsm(A, B, alpha=<span class="num">0.5</span>) = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| |
| Batch matrix solve |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] ] |
| B = `[ `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ], |
| `[ [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>], [<span class="num">8.0</span>, <span class="num">8.0</span>, <span class="num">8.0</span>] ] ] |
| trsm(A, B, alpha=<span class="num">0.5</span>) = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ], |
| `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L395</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#linalg_trsm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="linalg_trsm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="linalg_trsm(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">linalg_trsm</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@linalg_trsm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Solves matrix equation involving a lower triangular matrix. |
| Input are tensors *A*, *B*, each of dimension *n >= <span class="num">2</span>* and having the same shape |
| on the leading *n-<span class="num">2</span>* dimensions. |
| |
| If *n=<span class="num">2</span>*, *A* must be triangular. The operator performs the BLAS3 function |
| *trsm*, solving <span class="kw">for</span> *out* in: |
| |
| *op*\ (*A*) \* *out* = *alpha* \* *B* |
| |
| <span class="kw">if</span> *rightside=False*, or |
| |
| *out* \* *op*\ (*A*) = *alpha* \* *B* |
| |
| <span class="kw">if</span> *rightside=True*. Here, *alpha* is a scalar parameter, and *op()* is either the |
| identity or the matrix transposition (depending on *transpose*). |
| |
| If *n><span class="num">2</span>*, *trsm* is performed separately on the trailing two dimensions <span class="kw">for</span> all inputs |
| (batch mode). |
| |
| .. note:: The operator supports float32 and float64 data types only. |
| |
| Examples:: |
| |
| Single matrix solve |
| A = `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| B = `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ] |
| trsm(A, B, alpha=<span class="num">0.5</span>) = `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ] |
| |
| Batch matrix solve |
| A = `[ `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ], `[ [<span class="num">1.0</span>, <span class="num">0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>] ] ] |
| B = `[ `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>] ], |
| `[ [<span class="num">4.0</span>, <span class="num">4.0</span>, <span class="num">4.0</span>], [<span class="num">8.0</span>, <span class="num">8.0</span>, <span class="num">8.0</span>] ] ] |
| trsm(A, B, alpha=<span class="num">0.5</span>) = `[ `[ [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>], [<span class="num">1.0</span>, <span class="num">1.0</span>, <span class="num">1.0</span>] ], |
| `[ [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>], [<span class="num">2.0</span>, <span class="num">2.0</span>, <span class="num">2.0</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/la_op.cc:L395</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise Natural logarithmic value of the input. |
| |
| The natural logarithm is logarithm in base *e*, so that ``log(exp(x)) = x`` |
| |
| The storage <span class="kw">type</span> of ``log`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L77</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise Natural logarithmic value of the input. |
| |
| The natural logarithm is logarithm in base *e*, so that ``log(exp(x)) = x`` |
| |
| The storage <span class="kw">type</span> of ``log`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L77</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log10" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log10(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log10(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log10</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log10(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise Base-<span class="num">10</span> logarithmic value of the input. |
| |
| ``<span class="num">10</span>**log10(x) = x`` |
| |
| The storage <span class="kw">type</span> of ``log10`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L94</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log10" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log10(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log10(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log10</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log10(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise Base-<span class="num">10</span> logarithmic value of the input. |
| |
| ``<span class="num">10</span>**log10(x) = x`` |
| |
| The storage <span class="kw">type</span> of ``log10`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L94</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log1p" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log1p(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log1p(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log1p</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log1p(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise ``log(<span class="num">1</span> + x)`` value of the input. |
| |
| This function is more accurate than ``log(<span class="num">1</span> + x)`` <span class="kw">for</span> small ``x`` so that |
| :math:`<span class="num">1</span>+x\approx <span class="num">1</span>` |
| |
| The storage <span class="kw">type</span> of ``log1p`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - log1p(default) = default |
| - log1p(row_sparse) = row_sparse |
| - log1p(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L199</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log1p" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log1p(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log1p(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log1p</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log1p(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise ``log(<span class="num">1</span> + x)`` value of the input. |
| |
| This function is more accurate than ``log(<span class="num">1</span> + x)`` <span class="kw">for</span> small ``x`` so that |
| :math:`<span class="num">1</span>+x\approx <span class="num">1</span>` |
| |
| The storage <span class="kw">type</span> of ``log1p`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - log1p(default) = default |
| - log1p(row_sparse) = row_sparse |
| - log1p(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L199</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log2" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log2(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log2(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log2</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log2(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise Base-<span class="num">2</span> logarithmic value of the input. |
| |
| ``<span class="num">2</span>**log2(x) = x`` |
| |
| The storage <span class="kw">type</span> of ``log2`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L106</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log2" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log2(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log2(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log2</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log2(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise Base-<span class="num">2</span> logarithmic value of the input. |
| |
| ``<span class="num">2</span>**log2(x) = x`` |
| |
| The storage <span class="kw">type</span> of ``log2`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_logexp.cc:L106</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log_softmax" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log_softmax(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log_softmax(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log_softmax</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log_softmax(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the log softmax of the input. |
| This is equivalent to computing softmax followed by log. |
| |
| Examples:: |
| |
| >>> x = mx.nd.array([<span class="num">1</span>, <span class="num">2</span>, <span class="num">.1</span>]) |
| >>> mx.nd.log_softmax(x).asnumpy() |
| array([-<span class="num">1.41702998</span>, -<span class="num">0.41702995</span>, -<span class="num">2.31702995</span>], dtype=float32) |
| |
| >>> x = mx.nd.array( `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">.1</span>],[<span class="num">.1</span>, <span class="num">2</span>, <span class="num">1</span>] ] ) |
| >>> mx.nd.log_softmax(x, axis=<span class="num">0</span>).asnumpy() |
| array(`[ [-<span class="num">0.34115392</span>, -<span class="num">0.69314718</span>, -<span class="num">1.24115396</span>], |
| [-<span class="num">1.24115396</span>, -<span class="num">0.69314718</span>, -<span class="num">0.34115392</span>] ], dtype=float32)</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#log_softmax" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="log_softmax(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="log_softmax(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">log_softmax</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@log_softmax(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the log softmax of the input. |
| This is equivalent to computing softmax followed by log. |
| |
| Examples:: |
| |
| >>> x = mx.nd.array([<span class="num">1</span>, <span class="num">2</span>, <span class="num">.1</span>]) |
| >>> mx.nd.log_softmax(x).asnumpy() |
| array([-<span class="num">1.41702998</span>, -<span class="num">0.41702995</span>, -<span class="num">2.31702995</span>], dtype=float32) |
| |
| >>> x = mx.nd.array( `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">.1</span>],[<span class="num">.1</span>, <span class="num">2</span>, <span class="num">1</span>] ] ) |
| >>> mx.nd.log_softmax(x, axis=<span class="num">0</span>).asnumpy() |
| array(`[ [-<span class="num">0.34115392</span>, -<span class="num">0.69314718</span>, -<span class="num">1.24115396</span>], |
| [-<span class="num">1.24115396</span>, -<span class="num">0.69314718</span>, -<span class="num">0.34115392</span>] ], dtype=float32)</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#logical_not" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="logical_not(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="logical_not(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">logical_not</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@logical_not(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of logical NOT (!) function |
| |
| Example: |
| logical_not([-<span class="num">2.</span>, <span class="num">0.</span>, <span class="num">1.</span>]) = [<span class="num">0.</span>, <span class="num">1.</span>, <span class="num">0.</span>]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#logical_not" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="logical_not(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="logical_not(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">logical_not</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@logical_not(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the result of logical NOT (!) function |
| |
| Example: |
| logical_not([-<span class="num">2.</span>, <span class="num">0.</span>, <span class="num">1.</span>]) = [<span class="num">0.</span>, <span class="num">1.</span>, <span class="num">0.</span>]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#make_loss" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="make_loss(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="make_loss(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">make_loss</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@make_loss(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Make your own loss function in network construction. |
| |
| This operator accepts a customized loss function symbol as a terminal loss and |
| the symbol should be an operator <span class="kw">with</span> no backward dependency. |
| The output of <span class="kw">this</span> function is the gradient of loss <span class="kw">with</span> respect to the input data. |
| |
| For example, <span class="kw">if</span> you are a making a cross entropy loss function. Assume ``out`` is the |
| predicted output and ``label`` is the <span class="kw">true</span> label, then the cross entropy can be defined as:: |
| |
| cross_entropy = label * log(out) + (<span class="num">1</span> - label) * log(<span class="num">1</span> - out) |
| loss = make_loss(cross_entropy) |
| |
| We will need to use ``make_loss`` when we are creating our own loss function or we want to |
| combine multiple loss functions. Also we may want to stop some variables' gradients |
| from backpropagation. See more detail in ``BlockGrad`` or ``stop_gradient``. |
| |
| The storage <span class="kw">type</span> of ``make_loss`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - make_loss(default) = default |
| - make_loss(row_sparse) = row_sparse |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L358</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#make_loss" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="make_loss(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="make_loss(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">make_loss</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@make_loss(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Make your own loss function in network construction. |
| |
| This operator accepts a customized loss function symbol as a terminal loss and |
| the symbol should be an operator <span class="kw">with</span> no backward dependency. |
| The output of <span class="kw">this</span> function is the gradient of loss <span class="kw">with</span> respect to the input data. |
| |
| For example, <span class="kw">if</span> you are a making a cross entropy loss function. Assume ``out`` is the |
| predicted output and ``label`` is the <span class="kw">true</span> label, then the cross entropy can be defined as:: |
| |
| cross_entropy = label * log(out) + (<span class="num">1</span> - label) * log(<span class="num">1</span> - out) |
| loss = make_loss(cross_entropy) |
| |
| We will need to use ``make_loss`` when we are creating our own loss function or we want to |
| combine multiple loss functions. Also we may want to stop some variables' gradients |
| from backpropagation. See more detail in ``BlockGrad`` or ``stop_gradient``. |
| |
| The storage <span class="kw">type</span> of ``make_loss`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - make_loss(default) = default |
| - make_loss(row_sparse) = row_sparse |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L358</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#max" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="max(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="max(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">max</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@max(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the max of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L31</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#max" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="max(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="max(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">max</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@max(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the max of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L31</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#max_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="max_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="max_axis(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">max_axis</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@max_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the max of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L31</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#max_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="max_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="max_axis(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">max_axis</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@max_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the max of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L31</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mean" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mean(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mean(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mean</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mean(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the mean of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L83</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mean" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mean(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mean(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mean</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mean(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the mean of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L83</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#min" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="min(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="min(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">min</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@min(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the min of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L46</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#min" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="min(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="min(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">min</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@min(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the min of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L46</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#min_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="min_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="min_axis(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">min_axis</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@min_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the min of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L46</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#min_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="min_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="min_axis(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">min_axis</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@min_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the min of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L46</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#moments" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="moments(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="moments(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">moments</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@moments(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Calculate the mean and variance of `data`. |
| |
| The mean and variance are calculated by aggregating the contents of data across axes. |
| If x is <span class="num">1</span>-D and axes = [<span class="num">0</span>] <span class="kw">this</span> is just the mean and variance of a vector. |
| |
| Example: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], [<span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>] ] |
| mean, <span class="kw">var</span> = moments(data=x, axes=[<span class="num">0</span>]) |
| mean = [<span class="num">2.5</span>, <span class="num">3.5</span>, <span class="num">4.5</span>] |
| <span class="kw">var</span> = [<span class="num">2.25</span>, <span class="num">2.25</span>, <span class="num">2.25</span>] |
| mean, <span class="kw">var</span> = moments(data=x, axes=[<span class="num">1</span>]) |
| mean = [<span class="num">2.0</span>, <span class="num">5.0</span>] |
| <span class="kw">var</span> = [<span class="num">0.66666667</span>, <span class="num">0.66666667</span>] |
| mean, <span class="kw">var</span> = moments(data=x, axis=[<span class="num">0</span>, <span class="num">1</span>]) |
| mean = [<span class="num">3.5</span>] |
| <span class="kw">var</span> = [<span class="num">2.9166667</span>] |
| |
| |
| |
| Defined in src/operator/nn/moments.cc:L53</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#moments" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="moments(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="moments(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">moments</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@moments(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Calculate the mean and variance of `data`. |
| |
| The mean and variance are calculated by aggregating the contents of data across axes. |
| If x is <span class="num">1</span>-D and axes = [<span class="num">0</span>] <span class="kw">this</span> is just the mean and variance of a vector. |
| |
| Example: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], [<span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>] ] |
| mean, <span class="kw">var</span> = moments(data=x, axes=[<span class="num">0</span>]) |
| mean = [<span class="num">2.5</span>, <span class="num">3.5</span>, <span class="num">4.5</span>] |
| <span class="kw">var</span> = [<span class="num">2.25</span>, <span class="num">2.25</span>, <span class="num">2.25</span>] |
| mean, <span class="kw">var</span> = moments(data=x, axes=[<span class="num">1</span>]) |
| mean = [<span class="num">2.0</span>, <span class="num">5.0</span>] |
| <span class="kw">var</span> = [<span class="num">0.66666667</span>, <span class="num">0.66666667</span>] |
| mean, <span class="kw">var</span> = moments(data=x, axis=[<span class="num">0</span>, <span class="num">1</span>]) |
| mean = [<span class="num">3.5</span>] |
| <span class="kw">var</span> = [<span class="num">2.9166667</span>] |
| |
| |
| |
| Defined in src/operator/nn/moments.cc:L53</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_lamb_update_phase1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_lamb_update_phase1(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_lamb_update_phase1(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_lamb_update_phase1</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_lamb_update_phase1(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Mixed Precision version of Phase I of lamb update |
| it performs the following operations and returns g:. |
| |
| Link to paper: https:<span class="cmt">//arxiv.org/pdf/1904.00962.pdf</span> |
| |
| .. math:: |
| \begin{gather*} |
| grad32 = grad(float16) * rescale_grad |
| <span class="kw">if</span> (grad < -clip_gradient) |
| then |
| grad = -clip_gradient |
| <span class="kw">if</span> (grad > clip_gradient) |
| then |
| grad = clip_gradient |
| |
| mean = beta1 * mean + (<span class="num">1</span> - beta1) * grad; |
| variance = beta2 * variance + (<span class="num">1.</span> - beta2) * grad ^ <span class="num">2</span>; |
| |
| <span class="kw">if</span> (bias_correction) |
| then |
| mean_hat = mean / (<span class="num">1.</span> - beta1^t); |
| var_hat = <span class="kw">var</span> / (<span class="num">1</span> - beta2^t); |
| g = mean_hat / (var_hat^(<span class="num">1</span>/<span class="num">2</span>) + epsilon) + wd * weight32; |
| <span class="kw">else</span> |
| g = mean / (var_data^(<span class="num">1</span>/<span class="num">2</span>) + epsilon) + wd * weight32; |
| \end{gather*} |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L1032</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_lamb_update_phase1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_lamb_update_phase1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_lamb_update_phase1(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_lamb_update_phase1</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_lamb_update_phase1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Mixed Precision version of Phase I of lamb update |
| it performs the following operations and returns g:. |
| |
| Link to paper: https:<span class="cmt">//arxiv.org/pdf/1904.00962.pdf</span> |
| |
| .. math:: |
| \begin{gather*} |
| grad32 = grad(float16) * rescale_grad |
| <span class="kw">if</span> (grad < -clip_gradient) |
| then |
| grad = -clip_gradient |
| <span class="kw">if</span> (grad > clip_gradient) |
| then |
| grad = clip_gradient |
| |
| mean = beta1 * mean + (<span class="num">1</span> - beta1) * grad; |
| variance = beta2 * variance + (<span class="num">1.</span> - beta2) * grad ^ <span class="num">2</span>; |
| |
| <span class="kw">if</span> (bias_correction) |
| then |
| mean_hat = mean / (<span class="num">1.</span> - beta1^t); |
| var_hat = <span class="kw">var</span> / (<span class="num">1</span> - beta2^t); |
| g = mean_hat / (var_hat^(<span class="num">1</span>/<span class="num">2</span>) + epsilon) + wd * weight32; |
| <span class="kw">else</span> |
| g = mean / (var_data^(<span class="num">1</span>/<span class="num">2</span>) + epsilon) + wd * weight32; |
| \end{gather*} |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L1032</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_lamb_update_phase2" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_lamb_update_phase2(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_lamb_update_phase2(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_lamb_update_phase2</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_lamb_update_phase2(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Mixed Precision version Phase II of lamb update |
| it performs the following operations and updates grad. |
| |
| Link to paper: https:<span class="cmt">//arxiv.org/pdf/1904.00962.pdf</span> |
| |
| .. math:: |
| \begin{gather*} |
| <span class="kw">if</span> (lower_bound >= <span class="num">0</span>) |
| then |
| r1 = max(r1, lower_bound) |
| <span class="kw">if</span> (upper_bound >= <span class="num">0</span>) |
| then |
| r1 = max(r1, upper_bound) |
| |
| <span class="kw">if</span> (r1 == <span class="num">0</span> or r2 == <span class="num">0</span>) |
| then |
| lr = lr |
| <span class="kw">else</span> |
| lr = lr * (r1/r2) |
| weight32 = weight32 - lr * g |
| weight(float16) = weight32 |
| \end{gather*} |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L1074</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_lamb_update_phase2" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_lamb_update_phase2(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_lamb_update_phase2(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_lamb_update_phase2</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_lamb_update_phase2(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Mixed Precision version Phase II of lamb update |
| it performs the following operations and updates grad. |
| |
| Link to paper: https:<span class="cmt">//arxiv.org/pdf/1904.00962.pdf</span> |
| |
| .. math:: |
| \begin{gather*} |
| <span class="kw">if</span> (lower_bound >= <span class="num">0</span>) |
| then |
| r1 = max(r1, lower_bound) |
| <span class="kw">if</span> (upper_bound >= <span class="num">0</span>) |
| then |
| r1 = max(r1, upper_bound) |
| |
| <span class="kw">if</span> (r1 == <span class="num">0</span> or r2 == <span class="num">0</span>) |
| then |
| lr = lr |
| <span class="kw">else</span> |
| lr = lr * (r1/r2) |
| weight32 = weight32 - lr * g |
| weight(float16) = weight32 |
| \end{gather*} |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L1074</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_nag_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_nag_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_nag_mom_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_nag_mom_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_nag_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> multi-precision Nesterov Accelerated Gradient( NAG) optimizer. |
| |
| |
| Defined in src/operator/optimizer_op.cc:L744</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_nag_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_nag_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_nag_mom_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_nag_mom_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_nag_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> multi-precision Nesterov Accelerated Gradient( NAG) optimizer. |
| |
| |
| Defined in src/operator/optimizer_op.cc:L744</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_sgd_mom_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_sgd_mom_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Updater function <span class="kw">for</span> multi-precision sgd optimizer</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_sgd_mom_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_sgd_mom_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Updater function <span class="kw">for</span> multi-precision sgd optimizer</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_sgd_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_sgd_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Updater function <span class="kw">for</span> multi-precision sgd optimizer</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#mp_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="mp_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="mp_sgd_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">mp_sgd_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@mp_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Updater function <span class="kw">for</span> multi-precision sgd optimizer</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_all_finite" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_all_finite(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_all_finite(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_all_finite</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_all_finite(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Check <span class="kw">if</span> all the float numbers in all the arrays are finite (used <span class="kw">for</span> AMP) |
| |
| |
| Defined in src/operator/contrib/all_finite.cc:L132</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_all_finite" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_all_finite(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_all_finite(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_all_finite</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_all_finite(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Check <span class="kw">if</span> all the float numbers in all the arrays are finite (used <span class="kw">for</span> AMP) |
| |
| |
| Defined in src/operator/contrib/all_finite.cc:L132</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_lars" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_lars(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_lars(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_lars</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_lars(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the LARS coefficients of multiple weights and grads from their sums of square<span class="lit">" |
| |
| |
| Defined in src/operator/contrib/multi_lars.cc:L36</span></pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_lars" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_lars(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_lars(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_lars</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_lars(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the LARS coefficients of multiple weights and grads from their sums of square<span class="lit">" |
| |
| |
| Defined in src/operator/contrib/multi_lars.cc:L36</span></pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_mp_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_mp_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_mp_sgd_mom_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_mp_sgd_mom_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_mp_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> multi-precision Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L471</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_mp_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_mp_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_mp_sgd_mom_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_mp_sgd_mom_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_mp_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> multi-precision Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L471</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_mp_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_mp_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_mp_sgd_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_mp_sgd_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_mp_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> multi-precision Stochastic Gradient Descent (SDG) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L416</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_mp_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_mp_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_mp_sgd_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_mp_sgd_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_mp_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> multi-precision Stochastic Gradient Descent (SDG) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L416</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_sgd_mom_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_sgd_mom_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L373</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_sgd_mom_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_sgd_mom_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L373</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_sgd_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_sgd_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Stochastic Gradient Descent (SDG) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L328</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_sgd_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_sgd_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Stochastic Gradient Descent (SDG) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L328</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_sum_sq" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_sum_sq(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_sum_sq(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_sum_sq</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_sum_sq(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the sums of squares of multiple arrays |
| |
| |
| Defined in src/operator/contrib/multi_sum_sq.cc:L35</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#multi_sum_sq" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="multi_sum_sq(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="multi_sum_sq(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">multi_sum_sq</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@multi_sum_sq(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Compute the sums of squares of multiple arrays |
| |
| |
| Defined in src/operator/contrib/multi_sum_sq.cc:L35</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#nag_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="nag_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="nag_mom_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">nag_mom_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@nag_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Nesterov Accelerated Gradient( NAG) optimizer. |
| It updates the weights using the following formula, |
| |
| .. math:: |
| v_t = \gamma v_{t-<span class="num">1</span>} + \eta * \nabla J(W_{t-<span class="num">1</span>} - \gamma v_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} - v_t |
| |
| Where |
| :math:`\eta` is the learning rate of the optimizer |
| :math:`\gamma` is the decay rate of the momentum estimate |
| :math:`\v_t` is the update vector at time step `t` |
| :math:`\W_t` is the weight vector at time step `t` |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L725</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#nag_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="nag_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="nag_mom_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">nag_mom_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@nag_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Nesterov Accelerated Gradient( NAG) optimizer. |
| It updates the weights using the following formula, |
| |
| .. math:: |
| v_t = \gamma v_{t-<span class="num">1</span>} + \eta * \nabla J(W_{t-<span class="num">1</span>} - \gamma v_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} - v_t |
| |
| Where |
| :math:`\eta` is the learning rate of the optimizer |
| :math:`\gamma` is the decay rate of the momentum estimate |
| :math:`\v_t` is the update vector at time step `t` |
| :math:`\W_t` is the weight vector at time step `t` |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L725</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#nanprod" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="nanprod(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="nanprod(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">nanprod</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@nanprod(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the product of array elements over given axes treating Not a Numbers (``NaN``) as one. |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_prod_value.cc:L46</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#nanprod" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="nanprod(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="nanprod(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">nanprod</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@nanprod(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the product of array elements over given axes treating Not a Numbers (``NaN``) as one. |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_prod_value.cc:L46</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#nansum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="nansum(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="nansum(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">nansum</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@nansum(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the sum of array elements over given axes treating Not a Numbers (``NaN``) as zero. |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_sum_value.cc:L101</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#nansum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="nansum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="nansum(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">nansum</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@nansum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the sum of array elements over given axes treating Not a Numbers (``NaN``) as zero. |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_sum_value.cc:L101</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#negative" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="negative(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="negative(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">negative</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@negative(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Numerical negative of the argument, element-wise. |
| |
| The storage <span class="kw">type</span> of ``negative`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - negative(default) = default |
| - negative(row_sparse) = row_sparse |
| - negative(csr) = csr</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#negative" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="negative(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="negative(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">negative</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@negative(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Numerical negative of the argument, element-wise. |
| |
| The storage <span class="kw">type</span> of ``negative`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - negative(default) = default |
| - negative(row_sparse) = row_sparse |
| - negative(csr) = csr</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#norm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="norm(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="norm(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">norm</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@norm(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the norm on an NDArray. |
| |
| This operator computes the norm on an NDArray <span class="kw">with</span> the specified axis, depending |
| on the value of the ord parameter. By default, it computes the L2 norm on the entire |
| array. Currently only ord=<span class="num">2</span> supports sparse ndarrays. |
| |
| Examples:: |
| |
| x = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ], |
| `[ [<span class="num">2</span>, <span class="num">2</span>], |
| [<span class="num">5</span>, <span class="num">6</span>] ] ] |
| |
| norm(x, ord=<span class="num">2</span>, axis=<span class="num">1</span>) = `[ [<span class="num">3.1622777</span> <span class="num">4.472136</span> ] |
| [<span class="num">5.3851647</span> <span class="num">6.3245554</span>] ] |
| |
| norm(x, ord=<span class="num">1</span>, axis=<span class="num">1</span>) = `[ [<span class="num">4.</span>, <span class="num">6.</span>], |
| [<span class="num">7.</span>, <span class="num">8.</span>] ] |
| |
| rsp = x.cast_storage('row_sparse') |
| |
| norm(rsp) = [<span class="num">5.47722578</span>] |
| |
| csr = x.cast_storage(<span class="lit">'csr'</span>) |
| |
| norm(csr) = [<span class="num">5.47722578</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_norm_value.cc:L88</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#norm" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="norm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="norm(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">norm</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@norm(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the norm on an NDArray. |
| |
| This operator computes the norm on an NDArray <span class="kw">with</span> the specified axis, depending |
| on the value of the ord parameter. By default, it computes the L2 norm on the entire |
| array. Currently only ord=<span class="num">2</span> supports sparse ndarrays. |
| |
| Examples:: |
| |
| x = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ], |
| `[ [<span class="num">2</span>, <span class="num">2</span>], |
| [<span class="num">5</span>, <span class="num">6</span>] ] ] |
| |
| norm(x, ord=<span class="num">2</span>, axis=<span class="num">1</span>) = `[ [<span class="num">3.1622777</span> <span class="num">4.472136</span> ] |
| [<span class="num">5.3851647</span> <span class="num">6.3245554</span>] ] |
| |
| norm(x, ord=<span class="num">1</span>, axis=<span class="num">1</span>) = `[ [<span class="num">4.</span>, <span class="num">6.</span>], |
| [<span class="num">7.</span>, <span class="num">8.</span>] ] |
| |
| rsp = x.cast_storage('row_sparse') |
| |
| norm(rsp) = [<span class="num">5.47722578</span>] |
| |
| csr = x.cast_storage(<span class="lit">'csr'</span>) |
| |
| norm(csr) = [<span class="num">5.47722578</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_norm_value.cc:L88</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#normal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="normal(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="normal(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">normal</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@normal(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a normal (Gaussian) distribution. |
| |
| .. note:: The existing alias ``normal`` is deprecated. |
| |
| Samples are distributed according to a normal distribution parametrized by *loc* (mean) and *scale* |
| (standard deviation). |
| |
| Example:: |
| |
| normal(loc=<span class="num">0</span>, scale=<span class="num">1</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">1.89171135</span>, -<span class="num">1.16881478</span>], |
| [-<span class="num">1.23474145</span>, <span class="num">1.55807114</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L112</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#normal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="normal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="normal(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">normal</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@normal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a normal (Gaussian) distribution. |
| |
| .. note:: The existing alias ``normal`` is deprecated. |
| |
| Samples are distributed according to a normal distribution parametrized by *loc* (mean) and *scale* |
| (standard deviation). |
| |
| Example:: |
| |
| normal(loc=<span class="num">0</span>, scale=<span class="num">1</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">1.89171135</span>, -<span class="num">1.16881478</span>], |
| [-<span class="num">1.23474145</span>, <span class="num">1.55807114</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L112</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#one_hot" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="one_hot(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="one_hot(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">one_hot</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@one_hot(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a one-hot array. |
| |
| The locations represented by `indices` take value `on_value`, <span class="kw">while</span> all |
| other locations take value `off_value`. |
| |
| `one_hot` operation <span class="kw">with</span> `indices` of shape ``(i0, i1)`` and `depth` of ``d`` would result |
| in an output array of shape ``(i0, i1, d)`` <span class="kw">with</span>:: |
| |
| output[i,j,:] = off_value |
| output[i,j,indices[i,j] ] = on_value |
| |
| Examples:: |
| |
| one_hot([<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>,<span class="num">0</span>], <span class="num">3</span>) = `[ [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">1.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| one_hot([<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>,<span class="num">0</span>], <span class="num">3</span>, on_value=<span class="num">8</span>, off_value=<span class="num">1</span>, |
| dtype=<span class="lit">'int32'</span>) = `[ [<span class="num">1</span> <span class="num">8</span> <span class="num">1</span>] |
| [<span class="num">8</span> <span class="num">1</span> <span class="num">1</span>] |
| [<span class="num">1</span> <span class="num">1</span> <span class="num">8</span>] |
| [<span class="num">8</span> <span class="num">1</span> <span class="num">1</span>] ] |
| |
| one_hot(`[ [<span class="num">1</span>,<span class="num">0</span>],[<span class="num">1</span>,<span class="num">0</span>],[<span class="num">2</span>,<span class="num">0</span>] ], <span class="num">3</span>) = `[ `[ [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">1.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/indexing_op.cc:L882</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#one_hot" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="one_hot(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="one_hot(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">one_hot</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@one_hot(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a one-hot array. |
| |
| The locations represented by `indices` take value `on_value`, <span class="kw">while</span> all |
| other locations take value `off_value`. |
| |
| `one_hot` operation <span class="kw">with</span> `indices` of shape ``(i0, i1)`` and `depth` of ``d`` would result |
| in an output array of shape ``(i0, i1, d)`` <span class="kw">with</span>:: |
| |
| output[i,j,:] = off_value |
| output[i,j,indices[i,j] ] = on_value |
| |
| Examples:: |
| |
| one_hot([<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>,<span class="num">0</span>], <span class="num">3</span>) = `[ [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">1.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| one_hot([<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>,<span class="num">0</span>], <span class="num">3</span>, on_value=<span class="num">8</span>, off_value=<span class="num">1</span>, |
| dtype=<span class="lit">'int32'</span>) = `[ [<span class="num">1</span> <span class="num">8</span> <span class="num">1</span>] |
| [<span class="num">8</span> <span class="num">1</span> <span class="num">1</span>] |
| [<span class="num">1</span> <span class="num">1</span> <span class="num">8</span>] |
| [<span class="num">8</span> <span class="num">1</span> <span class="num">1</span>] ] |
| |
| one_hot(`[ [<span class="num">1</span>,<span class="num">0</span>],[<span class="num">1</span>,<span class="num">0</span>],[<span class="num">2</span>,<span class="num">0</span>] ], <span class="num">3</span>) = `[ `[ [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">1.</span>] |
| [ <span class="num">1.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/indexing_op.cc:L882</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ones_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ones_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ones_like(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ones_like</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ones_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return an array of ones <span class="kw">with</span> the same shape and <span class="kw">type</span> |
| as the input array. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| ones_like(x) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ones_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ones_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ones_like(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ones_like</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ones_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return an array of ones <span class="kw">with</span> the same shape and <span class="kw">type</span> |
| as the input array. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| |
| ones_like(x) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#pad" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="pad(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="pad(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">pad</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@pad(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Pads an input array <span class="kw">with</span> a constant or edge values of the array. |
| |
| .. note:: `Pad` is deprecated. Use `pad` instead. |
| |
| .. note:: Current implementation only supports <span class="num">4</span>D and <span class="num">5</span>D input arrays <span class="kw">with</span> padding applied |
| only on axes <span class="num">1</span>, <span class="num">2</span> and <span class="num">3.</span> Expects axes <span class="num">4</span> and <span class="num">5</span> in `pad_width` to be zero. |
| |
| This operation pads an input array <span class="kw">with</span> either a `constant_value` or edge values |
| along each axis of the input array. The amount of padding is specified by `pad_width`. |
| |
| `pad_width` is a tuple of integer padding widths <span class="kw">for</span> each axis of the format |
| ``(before_1, after_1, ... , before_N, after_N)``. The `pad_width` should be of length ``<span class="num">2</span>*N`` |
| where ``N`` is the number of dimensions of the array. |
| |
| For dimension ``N`` of the input array, ``before_N`` and ``after_N`` indicates how many values |
| to add before and after the elements of the array along dimension ``N``. |
| The widths of the higher two dimensions ``before_1``, ``after_1``, ``before_2``, |
| ``after_2`` must be <span class="num">0.</span> |
| |
| Example:: |
| |
| x = `[ [`[ [ <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span>] |
| [ <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span>] ] |
| |
| `[ [ <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span>] |
| [ <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span>] |
| [ <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span>] ] |
| |
| `[ [ <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span>] |
| [ <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span>] ] ] ] |
| |
| pad(x,mode=<span class="lit">"edge"</span>, pad_width=(<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>)) = |
| |
| `[ [`[ [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">3.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">3.</span>] |
| [ <span class="num">4.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">6.</span>] |
| [ <span class="num">4.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">6.</span>] ] |
| |
| `[ [ <span class="num">7.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">9.</span>] |
| [ <span class="num">7.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">9.</span>] |
| [ <span class="num">10.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">12.</span>] |
| [ <span class="num">10.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">12.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">11.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">13.</span>] |
| [ <span class="num">11.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">13.</span>] |
| [ <span class="num">14.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">16.</span>] |
| [ <span class="num">14.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">16.</span>] ] |
| |
| `[ [ <span class="num">17.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">19.</span>] |
| [ <span class="num">17.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">19.</span>] |
| [ <span class="num">20.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">22.</span>] |
| [ <span class="num">20.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">22.</span>] ] ] ] |
| |
| pad(x, mode=<span class="lit">"constant"</span>, constant_value=<span class="num">0</span>, pad_width=(<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>)) = |
| |
| `[ [`[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] ] |
| |
| |
| |
| |
| Defined in src/operator/pad.cc:L765</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#pad" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="pad(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="pad(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">pad</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@pad(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Pads an input array <span class="kw">with</span> a constant or edge values of the array. |
| |
| .. note:: `Pad` is deprecated. Use `pad` instead. |
| |
| .. note:: Current implementation only supports <span class="num">4</span>D and <span class="num">5</span>D input arrays <span class="kw">with</span> padding applied |
| only on axes <span class="num">1</span>, <span class="num">2</span> and <span class="num">3.</span> Expects axes <span class="num">4</span> and <span class="num">5</span> in `pad_width` to be zero. |
| |
| This operation pads an input array <span class="kw">with</span> either a `constant_value` or edge values |
| along each axis of the input array. The amount of padding is specified by `pad_width`. |
| |
| `pad_width` is a tuple of integer padding widths <span class="kw">for</span> each axis of the format |
| ``(before_1, after_1, ... , before_N, after_N)``. The `pad_width` should be of length ``<span class="num">2</span>*N`` |
| where ``N`` is the number of dimensions of the array. |
| |
| For dimension ``N`` of the input array, ``before_N`` and ``after_N`` indicates how many values |
| to add before and after the elements of the array along dimension ``N``. |
| The widths of the higher two dimensions ``before_1``, ``after_1``, ``before_2``, |
| ``after_2`` must be <span class="num">0.</span> |
| |
| Example:: |
| |
| x = `[ [`[ [ <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span>] |
| [ <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span>] ] |
| |
| `[ [ <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span>] |
| [ <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span>] |
| [ <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span>] ] |
| |
| `[ [ <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span>] |
| [ <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span>] ] ] ] |
| |
| pad(x,mode=<span class="lit">"edge"</span>, pad_width=(<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>)) = |
| |
| `[ [`[ [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">3.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">3.</span>] |
| [ <span class="num">4.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">6.</span>] |
| [ <span class="num">4.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">6.</span>] ] |
| |
| `[ [ <span class="num">7.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">9.</span>] |
| [ <span class="num">7.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">9.</span>] |
| [ <span class="num">10.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">12.</span>] |
| [ <span class="num">10.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">12.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">11.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">13.</span>] |
| [ <span class="num">11.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">13.</span>] |
| [ <span class="num">14.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">16.</span>] |
| [ <span class="num">14.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">16.</span>] ] |
| |
| `[ [ <span class="num">17.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">19.</span>] |
| [ <span class="num">17.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">19.</span>] |
| [ <span class="num">20.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">22.</span>] |
| [ <span class="num">20.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">22.</span>] ] ] ] |
| |
| pad(x, mode=<span class="lit">"constant"</span>, constant_value=<span class="num">0</span>, pad_width=(<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">0</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>)) = |
| |
| `[ [`[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">7.</span> <span class="num">8.</span> <span class="num">9.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">10.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] |
| |
| |
| `[ `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">11.</span> <span class="num">12.</span> <span class="num">13.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">14.</span> <span class="num">15.</span> <span class="num">16.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] |
| |
| `[ [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">17.</span> <span class="num">18.</span> <span class="num">19.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">20.</span> <span class="num">21.</span> <span class="num">22.</span> <span class="num">0.</span>] |
| [ <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span> <span class="num">0.</span>] ] ] ] |
| |
| |
| |
| |
| Defined in src/operator/pad.cc:L765</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#pick" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="pick(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="pick(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">pick</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@pick(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Picks elements from an input array according to the input indices along the given axis. |
| |
| Given an input array of shape ``(d0, d1)`` and indices of shape ``(i0,)``, the result will be |
| an output array of shape ``(i0,)`` <span class="kw">with</span>:: |
| |
| output[i] = input[i, indices[i] ] |
| |
| By default, <span class="kw">if</span> any index mentioned is too large, it is replaced by the index that addresses |
| the last element along an axis (the `clip` mode). |
| |
| This function supports n-dimensional input and (n-<span class="num">1</span>)-dimensional indices arrays. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] |
| |
| <span class="cmt">// picks elements with specified indices along axis 0</span> |
| pick(x, y=[<span class="num">0</span>,<span class="num">1</span>], <span class="num">0</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1</span> |
| pick(x, y=[<span class="num">0</span>,<span class="num">1</span>,<span class="num">0</span>], <span class="num">1</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>, <span class="num">5.</span>] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1 using 'wrap' mode</span> |
| <span class="cmt">// to place indicies that would normally be out of bounds</span> |
| pick(x, y=[<span class="num">2</span>,-<span class="num">1</span>,-<span class="num">2</span>], <span class="num">1</span>, mode=<span class="lit">'wrap'</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>, <span class="num">5.</span>] |
| |
| y = `[ [ <span class="num">1.</span>], |
| [ <span class="num">0.</span>], |
| [ <span class="num">2.</span>] ] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1 and dims are maintained</span> |
| pick(x, y, <span class="num">1</span>, keepdims=True) = `[ [ <span class="num">2.</span>], |
| [ <span class="num">3.</span>], |
| [ <span class="num">6.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L150</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#pick" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="pick(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="pick(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">pick</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@pick(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Picks elements from an input array according to the input indices along the given axis. |
| |
| Given an input array of shape ``(d0, d1)`` and indices of shape ``(i0,)``, the result will be |
| an output array of shape ``(i0,)`` <span class="kw">with</span>:: |
| |
| output[i] = input[i, indices[i] ] |
| |
| By default, <span class="kw">if</span> any index mentioned is too large, it is replaced by the index that addresses |
| the last element along an axis (the `clip` mode). |
| |
| This function supports n-dimensional input and (n-<span class="num">1</span>)-dimensional indices arrays. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] |
| |
| <span class="cmt">// picks elements with specified indices along axis 0</span> |
| pick(x, y=[<span class="num">0</span>,<span class="num">1</span>], <span class="num">0</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1</span> |
| pick(x, y=[<span class="num">0</span>,<span class="num">1</span>,<span class="num">0</span>], <span class="num">1</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>, <span class="num">5.</span>] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1 using 'wrap' mode</span> |
| <span class="cmt">// to place indicies that would normally be out of bounds</span> |
| pick(x, y=[<span class="num">2</span>,-<span class="num">1</span>,-<span class="num">2</span>], <span class="num">1</span>, mode=<span class="lit">'wrap'</span>) = [ <span class="num">1.</span>, <span class="num">4.</span>, <span class="num">5.</span>] |
| |
| y = `[ [ <span class="num">1.</span>], |
| [ <span class="num">0.</span>], |
| [ <span class="num">2.</span>] ] |
| |
| <span class="cmt">// picks elements with specified indices along axis 1 and dims are maintained</span> |
| pick(x, y, <span class="num">1</span>, keepdims=True) = `[ [ <span class="num">2.</span>], |
| [ <span class="num">3.</span>], |
| [ <span class="num">6.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_op_index.cc:L150</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#preloaded_multi_mp_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="preloaded_multi_mp_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="preloaded_multi_mp_sgd_mom_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">preloaded_multi_mp_sgd_mom_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@preloaded_multi_mp_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> multi-precision Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| |
| |
| Defined in src/operator/contrib/preloaded_multi_sgd.cc:L199</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#preloaded_multi_mp_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="preloaded_multi_mp_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="preloaded_multi_mp_sgd_mom_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">preloaded_multi_mp_sgd_mom_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@preloaded_multi_mp_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> multi-precision Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| |
| |
| Defined in src/operator/contrib/preloaded_multi_sgd.cc:L199</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#preloaded_multi_mp_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="preloaded_multi_mp_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="preloaded_multi_mp_sgd_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">preloaded_multi_mp_sgd_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@preloaded_multi_mp_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> multi-precision Stochastic Gradient Descent (SDG) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| |
| |
| Defined in src/operator/contrib/preloaded_multi_sgd.cc:L139</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#preloaded_multi_mp_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="preloaded_multi_mp_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="preloaded_multi_mp_sgd_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">preloaded_multi_mp_sgd_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@preloaded_multi_mp_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> multi-precision Stochastic Gradient Descent (SDG) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| |
| |
| Defined in src/operator/contrib/preloaded_multi_sgd.cc:L139</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#preloaded_multi_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="preloaded_multi_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="preloaded_multi_sgd_mom_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">preloaded_multi_sgd_mom_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@preloaded_multi_sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| |
| |
| Defined in src/operator/contrib/preloaded_multi_sgd.cc:L90</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#preloaded_multi_sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="preloaded_multi_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="preloaded_multi_sgd_mom_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">preloaded_multi_sgd_mom_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@preloaded_multi_sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| |
| |
| Defined in src/operator/contrib/preloaded_multi_sgd.cc:L90</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#preloaded_multi_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="preloaded_multi_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="preloaded_multi_sgd_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">preloaded_multi_sgd_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@preloaded_multi_sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Stochastic Gradient Descent (SDG) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| |
| |
| Defined in src/operator/contrib/preloaded_multi_sgd.cc:L41</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#preloaded_multi_sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="preloaded_multi_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="preloaded_multi_sgd_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">preloaded_multi_sgd_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@preloaded_multi_sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Stochastic Gradient Descent (SDG) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| |
| |
| Defined in src/operator/contrib/preloaded_multi_sgd.cc:L41</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#prod" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="prod(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="prod(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">prod</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@prod(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the product of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L30</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#prod" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="prod(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="prod(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">prod</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@prod(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the product of array elements over given axes. |
| |
| Defined in src/operator/tensor/./broadcast_reduce_op.h:L30</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#radians" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="radians(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="radians(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">radians</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@radians(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Converts each element of the input array from degrees to radians. |
| |
| .. math:: |
| radians([<span class="num">0</span>, <span class="num">90</span>, <span class="num">180</span>, <span class="num">270</span>, <span class="num">360</span>]) = [<span class="num">0</span>, \pi/<span class="num">2</span>, \pi, <span class="num">3</span>\pi/<span class="num">2</span>, <span class="num">2</span>\pi] |
| |
| The storage <span class="kw">type</span> of ``radians`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - radians(default) = default |
| - radians(row_sparse) = row_sparse |
| - radians(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L351</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#radians" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="radians(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="radians(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">radians</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@radians(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Converts each element of the input array from degrees to radians. |
| |
| .. math:: |
| radians([<span class="num">0</span>, <span class="num">90</span>, <span class="num">180</span>, <span class="num">270</span>, <span class="num">360</span>]) = [<span class="num">0</span>, \pi/<span class="num">2</span>, \pi, <span class="num">3</span>\pi/<span class="num">2</span>, <span class="num">2</span>\pi] |
| |
| The storage <span class="kw">type</span> of ``radians`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - radians(default) = default |
| - radians(row_sparse) = row_sparse |
| - radians(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L351</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_exponential" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_exponential(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_exponential(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_exponential</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_exponential(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from an exponential distribution. |
| |
| Samples are distributed according to an exponential distribution parametrized by *lambda* (rate). |
| |
| Example:: |
| |
| exponential(lam=<span class="num">4</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">0.0097189</span> , <span class="num">0.08999364</span>], |
| [ <span class="num">0.04146638</span>, <span class="num">0.31715935</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L136</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_exponential" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_exponential(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_exponential(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_exponential</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_exponential(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from an exponential distribution. |
| |
| Samples are distributed according to an exponential distribution parametrized by *lambda* (rate). |
| |
| Example:: |
| |
| exponential(lam=<span class="num">4</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">0.0097189</span> , <span class="num">0.08999364</span>], |
| [ <span class="num">0.04146638</span>, <span class="num">0.31715935</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L136</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_gamma" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_gamma(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_gamma(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_gamma</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_gamma(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a gamma distribution. |
| |
| Samples are distributed according to a gamma distribution parametrized by *alpha* (shape) and *beta* (scale). |
| |
| Example:: |
| |
| gamma(alpha=<span class="num">9</span>, beta=<span class="num">0.5</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">7.10486984</span>, <span class="num">3.37695289</span>], |
| [ <span class="num">3.91697288</span>, <span class="num">3.65933681</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L124</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_gamma" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_gamma(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_gamma(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_gamma</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_gamma(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a gamma distribution. |
| |
| Samples are distributed according to a gamma distribution parametrized by *alpha* (shape) and *beta* (scale). |
| |
| Example:: |
| |
| gamma(alpha=<span class="num">9</span>, beta=<span class="num">0.5</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">7.10486984</span>, <span class="num">3.37695289</span>], |
| [ <span class="num">3.91697288</span>, <span class="num">3.65933681</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L124</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_generalized_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_generalized_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_generalized_negative_binomial(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_generalized_negative_binomial</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_generalized_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a generalized negative binomial distribution. |
| |
| Samples are distributed according to a generalized negative binomial distribution parametrized by |
| *mu* (mean) and *alpha* (dispersion). *alpha* is defined as *<span class="num">1</span>/k* where *k* is the failure limit of the |
| number of unsuccessful experiments (generalized to real numbers). |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Example:: |
| |
| generalized_negative_binomial(mu=<span class="num">2.0</span>, alpha=<span class="num">0.3</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">2.</span>, <span class="num">1.</span>], |
| [ <span class="num">6.</span>, <span class="num">4.</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L178</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_generalized_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_generalized_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_generalized_negative_binomial(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_generalized_negative_binomial</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_generalized_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a generalized negative binomial distribution. |
| |
| Samples are distributed according to a generalized negative binomial distribution parametrized by |
| *mu* (mean) and *alpha* (dispersion). *alpha* is defined as *<span class="num">1</span>/k* where *k* is the failure limit of the |
| number of unsuccessful experiments (generalized to real numbers). |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Example:: |
| |
| generalized_negative_binomial(mu=<span class="num">2.0</span>, alpha=<span class="num">0.3</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">2.</span>, <span class="num">1.</span>], |
| [ <span class="num">6.</span>, <span class="num">4.</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L178</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_negative_binomial(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_negative_binomial</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a negative binomial distribution. |
| |
| Samples are distributed according to a negative binomial distribution parametrized by |
| *k* (limit of unsuccessful experiments) and *p* (failure probability in each experiment). |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Example:: |
| |
| negative_binomial(k=<span class="num">3</span>, p=<span class="num">0.4</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">4.</span>, <span class="num">7.</span>], |
| [ <span class="num">2.</span>, <span class="num">5.</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L163</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_negative_binomial(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_negative_binomial</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a negative binomial distribution. |
| |
| Samples are distributed according to a negative binomial distribution parametrized by |
| *k* (limit of unsuccessful experiments) and *p* (failure probability in each experiment). |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Example:: |
| |
| negative_binomial(k=<span class="num">3</span>, p=<span class="num">0.4</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">4.</span>, <span class="num">7.</span>], |
| [ <span class="num">2.</span>, <span class="num">5.</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L163</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_normal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_normal(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_normal(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_normal</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_normal(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a normal (Gaussian) distribution. |
| |
| .. note:: The existing alias ``normal`` is deprecated. |
| |
| Samples are distributed according to a normal distribution parametrized by *loc* (mean) and *scale* |
| (standard deviation). |
| |
| Example:: |
| |
| normal(loc=<span class="num">0</span>, scale=<span class="num">1</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">1.89171135</span>, -<span class="num">1.16881478</span>], |
| [-<span class="num">1.23474145</span>, <span class="num">1.55807114</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L112</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_normal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_normal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_normal(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_normal</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_normal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a normal (Gaussian) distribution. |
| |
| .. note:: The existing alias ``normal`` is deprecated. |
| |
| Samples are distributed according to a normal distribution parametrized by *loc* (mean) and *scale* |
| (standard deviation). |
| |
| Example:: |
| |
| normal(loc=<span class="num">0</span>, scale=<span class="num">1</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">1.89171135</span>, -<span class="num">1.16881478</span>], |
| [-<span class="num">1.23474145</span>, <span class="num">1.55807114</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L112</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_dirichlet" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_dirichlet(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_dirichlet(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_dirichlet</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_dirichlet(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| Dirichlet distributions <span class="kw">with</span> parameter *alpha*. |
| |
| The shape of *alpha* must <span class="kw">match</span> the leftmost subshape of *sample*. That is, *sample* |
| can have the same shape as *alpha*, in which <span class="kw">case</span> the output contains one density per |
| distribution, or *sample* can be a tensor of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> |
| the output is a tensor of densities such that the densities at index *i* in the output |
| are given by the samples at index *i* in *sample* parameterized by the value of *alpha* |
| at index *i*. |
| |
| Examples:: |
| |
| random_pdf_dirichlet(sample=`[ [<span class="num">1</span>,<span class="num">2</span>],[<span class="num">2</span>,<span class="num">3</span>],[<span class="num">3</span>,<span class="num">4</span>] ], alpha=[<span class="num">2.5</span>, <span class="num">2.5</span>]) = |
| [<span class="num">38.413498</span>, <span class="num">199.60245</span>, <span class="num">564.56085</span>] |
| |
| sample = `[ `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], [<span class="num">10</span>, <span class="num">20</span>, <span class="num">30</span>], [<span class="num">100</span>, <span class="num">200</span>, <span class="num">300</span>] ], |
| `[ [<span class="num">0.1</span>, <span class="num">0.2</span>, <span class="num">0.3</span>], [<span class="num">0.01</span>, <span class="num">0.02</span>, <span class="num">0.03</span>], [<span class="num">0.001</span>, <span class="num">0.002</span>, <span class="num">0.003</span>] ] ] |
| |
| random_pdf_dirichlet(sample=sample, alpha=[<span class="num">0.1</span>, <span class="num">0.4</span>, <span class="num">0.9</span>]) = |
| `[ [<span class="num">2.3257459e-02</span>, <span class="num">5.8420084e-04</span>, <span class="num">1.4674458e-05</span>], |
| [<span class="num">9.2589635e-01</span>, <span class="num">3.6860607e+01</span>, <span class="num">1.4674468e+03</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L315</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_dirichlet" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_dirichlet(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_dirichlet(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_dirichlet</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_dirichlet(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| Dirichlet distributions <span class="kw">with</span> parameter *alpha*. |
| |
| The shape of *alpha* must <span class="kw">match</span> the leftmost subshape of *sample*. That is, *sample* |
| can have the same shape as *alpha*, in which <span class="kw">case</span> the output contains one density per |
| distribution, or *sample* can be a tensor of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> |
| the output is a tensor of densities such that the densities at index *i* in the output |
| are given by the samples at index *i* in *sample* parameterized by the value of *alpha* |
| at index *i*. |
| |
| Examples:: |
| |
| random_pdf_dirichlet(sample=`[ [<span class="num">1</span>,<span class="num">2</span>],[<span class="num">2</span>,<span class="num">3</span>],[<span class="num">3</span>,<span class="num">4</span>] ], alpha=[<span class="num">2.5</span>, <span class="num">2.5</span>]) = |
| [<span class="num">38.413498</span>, <span class="num">199.60245</span>, <span class="num">564.56085</span>] |
| |
| sample = `[ `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], [<span class="num">10</span>, <span class="num">20</span>, <span class="num">30</span>], [<span class="num">100</span>, <span class="num">200</span>, <span class="num">300</span>] ], |
| `[ [<span class="num">0.1</span>, <span class="num">0.2</span>, <span class="num">0.3</span>], [<span class="num">0.01</span>, <span class="num">0.02</span>, <span class="num">0.03</span>], [<span class="num">0.001</span>, <span class="num">0.002</span>, <span class="num">0.003</span>] ] ] |
| |
| random_pdf_dirichlet(sample=sample, alpha=[<span class="num">0.1</span>, <span class="num">0.4</span>, <span class="num">0.9</span>]) = |
| `[ [<span class="num">2.3257459e-02</span>, <span class="num">5.8420084e-04</span>, <span class="num">1.4674458e-05</span>], |
| [<span class="num">9.2589635e-01</span>, <span class="num">3.6860607e+01</span>, <span class="num">1.4674468e+03</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L315</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_exponential" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_exponential(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_exponential(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_exponential</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_exponential(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| exponential distributions <span class="kw">with</span> parameters *lam* (rate). |
| |
| The shape of *lam* must <span class="kw">match</span> the leftmost subshape of *sample*. That is, *sample* |
| can have the same shape as *lam*, in which <span class="kw">case</span> the output contains one density per |
| distribution, or *sample* can be a tensor of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> |
| the output is a tensor of densities such that the densities at index *i* in the output |
| are given by the samples at index *i* in *sample* parameterized by the value of *lam* |
| at index *i*. |
| |
| Examples:: |
| |
| random_pdf_exponential(sample=`[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ], lam=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.36787945</span>, <span class="num">0.13533528</span>, <span class="num">0.04978707</span>] ] |
| |
| sample = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ] |
| |
| random_pdf_exponential(sample=sample, lam=[<span class="num">1</span>,<span class="num">0.5</span>,<span class="num">0.25</span>]) = |
| `[ [<span class="num">0.36787945</span>, <span class="num">0.13533528</span>, <span class="num">0.04978707</span>], |
| [<span class="num">0.30326533</span>, <span class="num">0.18393973</span>, <span class="num">0.11156508</span>], |
| [<span class="num">0.1947002</span>, <span class="num">0.15163267</span>, <span class="num">0.11809164</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L304</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_exponential" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_exponential(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_exponential(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_exponential</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_exponential(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| exponential distributions <span class="kw">with</span> parameters *lam* (rate). |
| |
| The shape of *lam* must <span class="kw">match</span> the leftmost subshape of *sample*. That is, *sample* |
| can have the same shape as *lam*, in which <span class="kw">case</span> the output contains one density per |
| distribution, or *sample* can be a tensor of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> |
| the output is a tensor of densities such that the densities at index *i* in the output |
| are given by the samples at index *i* in *sample* parameterized by the value of *lam* |
| at index *i*. |
| |
| Examples:: |
| |
| random_pdf_exponential(sample=`[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ], lam=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.36787945</span>, <span class="num">0.13533528</span>, <span class="num">0.04978707</span>] ] |
| |
| sample = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ] |
| |
| random_pdf_exponential(sample=sample, lam=[<span class="num">1</span>,<span class="num">0.5</span>,<span class="num">0.25</span>]) = |
| `[ [<span class="num">0.36787945</span>, <span class="num">0.13533528</span>, <span class="num">0.04978707</span>], |
| [<span class="num">0.30326533</span>, <span class="num">0.18393973</span>, <span class="num">0.11156508</span>], |
| [<span class="num">0.1947002</span>, <span class="num">0.15163267</span>, <span class="num">0.11809164</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L304</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_gamma" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_gamma(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_gamma(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_gamma</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_gamma(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| gamma distributions <span class="kw">with</span> parameters *alpha* (shape) and *beta* (rate). |
| |
| *alpha* and *beta* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *alpha* and *beta*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *alpha* and *beta* at index *i*. |
| |
| Examples:: |
| |
| random_pdf_gamma(sample=`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>] ], alpha=[<span class="num">5</span>], beta=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.01532831</span>, <span class="num">0.09022352</span>, <span class="num">0.16803136</span>, <span class="num">0.19536681</span>, <span class="num">0.17546739</span>] ] |
| |
| sample = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>], |
| [<span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>], |
| [<span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>, <span class="num">7</span>] ] |
| |
| random_pdf_gamma(sample=sample, alpha=[<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>], beta=[<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>]) = |
| `[ [<span class="num">0.01532831</span>, <span class="num">0.09022352</span>, <span class="num">0.16803136</span>, <span class="num">0.19536681</span>, <span class="num">0.17546739</span>], |
| [<span class="num">0.03608941</span>, <span class="num">0.10081882</span>, <span class="num">0.15629345</span>, <span class="num">0.17546739</span>, <span class="num">0.16062315</span>], |
| [<span class="num">0.05040941</span>, <span class="num">0.10419563</span>, <span class="num">0.14622283</span>, <span class="num">0.16062315</span>, <span class="num">0.14900276</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L302</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_gamma" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_gamma(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_gamma(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_gamma</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_gamma(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| gamma distributions <span class="kw">with</span> parameters *alpha* (shape) and *beta* (rate). |
| |
| *alpha* and *beta* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *alpha* and *beta*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *alpha* and *beta* at index *i*. |
| |
| Examples:: |
| |
| random_pdf_gamma(sample=`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>] ], alpha=[<span class="num">5</span>], beta=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.01532831</span>, <span class="num">0.09022352</span>, <span class="num">0.16803136</span>, <span class="num">0.19536681</span>, <span class="num">0.17546739</span>] ] |
| |
| sample = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>], |
| [<span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>], |
| [<span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>, <span class="num">7</span>] ] |
| |
| random_pdf_gamma(sample=sample, alpha=[<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>], beta=[<span class="num">1</span>,<span class="num">1</span>,<span class="num">1</span>]) = |
| `[ [<span class="num">0.01532831</span>, <span class="num">0.09022352</span>, <span class="num">0.16803136</span>, <span class="num">0.19536681</span>, <span class="num">0.17546739</span>], |
| [<span class="num">0.03608941</span>, <span class="num">0.10081882</span>, <span class="num">0.15629345</span>, <span class="num">0.17546739</span>, <span class="num">0.16062315</span>], |
| [<span class="num">0.05040941</span>, <span class="num">0.10419563</span>, <span class="num">0.14622283</span>, <span class="num">0.16062315</span>, <span class="num">0.14900276</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L302</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_generalized_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_generalized_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_generalized_negative_binomial(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_generalized_negative_binomial</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_generalized_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| generalized negative binomial distributions <span class="kw">with</span> parameters *mu* (mean) |
| and *alpha* (dispersion). This can be understood as a reparameterization of |
| the negative binomial, where *k* = *<span class="num">1</span> / alpha* and *p* = *<span class="num">1</span> / (mu \* alpha + <span class="num">1</span>)*. |
| |
| *mu* and *alpha* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *mu* and *alpha*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *mu* and *alpha* at index *i*. |
| |
| Examples:: |
| |
| random_pdf_generalized_negative_binomial(sample=`[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>] ], alpha=[<span class="num">1</span>], mu=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.25</span>, <span class="num">0.125</span>, <span class="num">0.0625</span>, <span class="num">0.03125</span>] ] |
| |
| sample = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>] ] |
| random_pdf_generalized_negative_binomial(sample=sample, alpha=[<span class="num">1</span>, <span class="num">0.6666</span>], mu=[<span class="num">1</span>, <span class="num">1.5</span>]) = |
| `[ [<span class="num">0.25</span>, <span class="num">0.125</span>, <span class="num">0.0625</span>, <span class="num">0.03125</span> ], |
| [<span class="num">0.26517063</span>, <span class="num">0.16573331</span>, <span class="num">0.09667706</span>, <span class="num">0.05437994</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L313</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_generalized_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_generalized_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_generalized_negative_binomial(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_generalized_negative_binomial</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_generalized_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| generalized negative binomial distributions <span class="kw">with</span> parameters *mu* (mean) |
| and *alpha* (dispersion). This can be understood as a reparameterization of |
| the negative binomial, where *k* = *<span class="num">1</span> / alpha* and *p* = *<span class="num">1</span> / (mu \* alpha + <span class="num">1</span>)*. |
| |
| *mu* and *alpha* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *mu* and *alpha*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *mu* and *alpha* at index *i*. |
| |
| Examples:: |
| |
| random_pdf_generalized_negative_binomial(sample=`[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>] ], alpha=[<span class="num">1</span>], mu=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.25</span>, <span class="num">0.125</span>, <span class="num">0.0625</span>, <span class="num">0.03125</span>] ] |
| |
| sample = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>] ] |
| random_pdf_generalized_negative_binomial(sample=sample, alpha=[<span class="num">1</span>, <span class="num">0.6666</span>], mu=[<span class="num">1</span>, <span class="num">1.5</span>]) = |
| `[ [<span class="num">0.25</span>, <span class="num">0.125</span>, <span class="num">0.0625</span>, <span class="num">0.03125</span> ], |
| [<span class="num">0.26517063</span>, <span class="num">0.16573331</span>, <span class="num">0.09667706</span>, <span class="num">0.05437994</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L313</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_negative_binomial(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_negative_binomial</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of samples of |
| negative binomial distributions <span class="kw">with</span> parameters *k* (failure limit) and *p* (failure probability). |
| |
| *k* and *p* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *k* and *p*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *k* and *p* at index *i*. |
| |
| Examples:: |
| |
| random_pdf_negative_binomial(sample=`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>] ], k=[<span class="num">1</span>], p=a[<span class="num">0.5</span>]) = |
| `[ [<span class="num">0.25</span>, <span class="num">0.125</span>, <span class="num">0.0625</span>, <span class="num">0.03125</span>] ] |
| |
| # Note that k may be real-valued |
| sample = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>] ] |
| random_pdf_negative_binomial(sample=sample, k=[<span class="num">1</span>, <span class="num">1.5</span>], p=[<span class="num">0.5</span>, <span class="num">0.5</span>]) = |
| `[ [<span class="num">0.25</span>, <span class="num">0.125</span>, <span class="num">0.0625</span>, <span class="num">0.03125</span> ], |
| [<span class="num">0.26516506</span>, <span class="num">0.16572815</span>, <span class="num">0.09667476</span>, <span class="num">0.05437956</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L309</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_negative_binomial(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_negative_binomial</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of samples of |
| negative binomial distributions <span class="kw">with</span> parameters *k* (failure limit) and *p* (failure probability). |
| |
| *k* and *p* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *k* and *p*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *k* and *p* at index *i*. |
| |
| Examples:: |
| |
| random_pdf_negative_binomial(sample=`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>] ], k=[<span class="num">1</span>], p=a[<span class="num">0.5</span>]) = |
| `[ [<span class="num">0.25</span>, <span class="num">0.125</span>, <span class="num">0.0625</span>, <span class="num">0.03125</span>] ] |
| |
| # Note that k may be real-valued |
| sample = `[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], |
| [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>] ] |
| random_pdf_negative_binomial(sample=sample, k=[<span class="num">1</span>, <span class="num">1.5</span>], p=[<span class="num">0.5</span>, <span class="num">0.5</span>]) = |
| `[ [<span class="num">0.25</span>, <span class="num">0.125</span>, <span class="num">0.0625</span>, <span class="num">0.03125</span> ], |
| [<span class="num">0.26516506</span>, <span class="num">0.16572815</span>, <span class="num">0.09667476</span>, <span class="num">0.05437956</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L309</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_normal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_normal(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_normal(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_normal</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_normal(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| normal distributions <span class="kw">with</span> parameters *mu* (mean) and *sigma* (standard deviation). |
| |
| *mu* and *sigma* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *mu* and *sigma*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *mu* and *sigma* at index *i*. |
| |
| Examples:: |
| |
| sample = `[ [-<span class="num">2</span>, -<span class="num">1</span>, <span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>] ] |
| random_pdf_normal(sample=sample, mu=[<span class="num">0</span>], sigma=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.05399097</span>, <span class="num">0.24197073</span>, <span class="num">0.3989423</span>, <span class="num">0.24197073</span>, <span class="num">0.05399097</span>] ] |
| |
| random_pdf_normal(sample=sample*<span class="num">2</span>, mu=[<span class="num">0</span>,<span class="num">0</span>], sigma=[<span class="num">1</span>,<span class="num">2</span>]) = |
| `[ [<span class="num">0.05399097</span>, <span class="num">0.24197073</span>, <span class="num">0.3989423</span>, <span class="num">0.24197073</span>, <span class="num">0.05399097</span>], |
| [<span class="num">0.12098537</span>, <span class="num">0.17603266</span>, <span class="num">0.19947115</span>, <span class="num">0.17603266</span>, <span class="num">0.12098537</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L299</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_normal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_normal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_normal(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_normal</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_normal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| normal distributions <span class="kw">with</span> parameters *mu* (mean) and *sigma* (standard deviation). |
| |
| *mu* and *sigma* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *mu* and *sigma*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *mu* and *sigma* at index *i*. |
| |
| Examples:: |
| |
| sample = `[ [-<span class="num">2</span>, -<span class="num">1</span>, <span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>] ] |
| random_pdf_normal(sample=sample, mu=[<span class="num">0</span>], sigma=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.05399097</span>, <span class="num">0.24197073</span>, <span class="num">0.3989423</span>, <span class="num">0.24197073</span>, <span class="num">0.05399097</span>] ] |
| |
| random_pdf_normal(sample=sample*<span class="num">2</span>, mu=[<span class="num">0</span>,<span class="num">0</span>], sigma=[<span class="num">1</span>,<span class="num">2</span>]) = |
| `[ [<span class="num">0.05399097</span>, <span class="num">0.24197073</span>, <span class="num">0.3989423</span>, <span class="num">0.24197073</span>, <span class="num">0.05399097</span>], |
| [<span class="num">0.12098537</span>, <span class="num">0.17603266</span>, <span class="num">0.19947115</span>, <span class="num">0.17603266</span>, <span class="num">0.12098537</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L299</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_poisson" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_poisson(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_poisson(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_poisson</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_poisson(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| Poisson distributions <span class="kw">with</span> parameters *lam* (rate). |
| |
| The shape of *lam* must <span class="kw">match</span> the leftmost subshape of *sample*. That is, *sample* |
| can have the same shape as *lam*, in which <span class="kw">case</span> the output contains one density per |
| distribution, or *sample* can be a tensor of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> |
| the output is a tensor of densities such that the densities at index *i* in the output |
| are given by the samples at index *i* in *sample* parameterized by the value of *lam* |
| at index *i*. |
| |
| Examples:: |
| |
| random_pdf_poisson(sample=`[ [<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ], lam=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.36787945</span>, <span class="num">0.36787945</span>, <span class="num">0.18393973</span>, <span class="num">0.06131324</span>] ] |
| |
| sample = `[ [<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ] |
| |
| random_pdf_poisson(sample=sample, lam=[<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>]) = |
| `[ [<span class="num">0.36787945</span>, <span class="num">0.36787945</span>, <span class="num">0.18393973</span>, <span class="num">0.06131324</span>], |
| [<span class="num">0.13533528</span>, <span class="num">0.27067056</span>, <span class="num">0.27067056</span>, <span class="num">0.18044704</span>], |
| [<span class="num">0.04978707</span>, <span class="num">0.14936121</span>, <span class="num">0.22404182</span>, <span class="num">0.22404182</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L306</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_poisson" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_poisson(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_poisson(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_poisson</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_poisson(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| Poisson distributions <span class="kw">with</span> parameters *lam* (rate). |
| |
| The shape of *lam* must <span class="kw">match</span> the leftmost subshape of *sample*. That is, *sample* |
| can have the same shape as *lam*, in which <span class="kw">case</span> the output contains one density per |
| distribution, or *sample* can be a tensor of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> |
| the output is a tensor of densities such that the densities at index *i* in the output |
| are given by the samples at index *i* in *sample* parameterized by the value of *lam* |
| at index *i*. |
| |
| Examples:: |
| |
| random_pdf_poisson(sample=`[ [<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ], lam=[<span class="num">1</span>]) = |
| `[ [<span class="num">0.36787945</span>, <span class="num">0.36787945</span>, <span class="num">0.18393973</span>, <span class="num">0.06131324</span>] ] |
| |
| sample = `[ [<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>], |
| [<span class="num">0</span>,<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>] ] |
| |
| random_pdf_poisson(sample=sample, lam=[<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>]) = |
| `[ [<span class="num">0.36787945</span>, <span class="num">0.36787945</span>, <span class="num">0.18393973</span>, <span class="num">0.06131324</span>], |
| [<span class="num">0.13533528</span>, <span class="num">0.27067056</span>, <span class="num">0.27067056</span>, <span class="num">0.18044704</span>], |
| [<span class="num">0.04978707</span>, <span class="num">0.14936121</span>, <span class="num">0.22404182</span>, <span class="num">0.22404182</span>] ] |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L306</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_uniform" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_uniform(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_uniform(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_uniform</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_uniform(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| uniform distributions on the intervals given by *[low,high)*. |
| |
| *low* and *high* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *low* and *high*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *low* and *high* at index *i*. |
| |
| Examples:: |
| |
| random_pdf_uniform(sample=`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>] ], low=[<span class="num">0</span>], high=[<span class="num">10</span>]) = [<span class="num">0.1</span>, <span class="num">0.1</span>, <span class="num">0.1</span>, <span class="num">0.1</span>] |
| |
| sample = `[ `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], |
| [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ], |
| `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], |
| [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ] ] |
| low = `[ [<span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>] ] |
| high = `[ [ <span class="num">5</span>, <span class="num">10</span>], |
| [<span class="num">15</span>, <span class="num">20</span>] ] |
| random_pdf_uniform(sample=sample, low=low, high=high) = |
| `[ `[ [<span class="num">0.2</span>, <span class="num">0.2</span>, <span class="num">0.2</span> ], |
| [<span class="num">0.1</span>, <span class="num">0.1</span>, <span class="num">0.1</span> ] ], |
| `[ [<span class="num">0.06667</span>, <span class="num">0.06667</span>, <span class="num">0.06667</span>], |
| [<span class="num">0.05</span>, <span class="num">0.05</span>, <span class="num">0.05</span> ] ] ] |
| |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L297</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_pdf_uniform" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_pdf_uniform(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_pdf_uniform(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_pdf_uniform</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_pdf_uniform(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the value of the PDF of *sample* of |
| uniform distributions on the intervals given by *[low,high)*. |
| |
| *low* and *high* must have the same shape, which must <span class="kw">match</span> the leftmost subshape |
| of *sample*. That is, *sample* can have the same shape as *low* and *high*, in which |
| <span class="kw">case</span> the output contains one density per distribution, or *sample* can be a tensor |
| of tensors <span class="kw">with</span> that shape, in which <span class="kw">case</span> the output is a tensor of densities such that |
| the densities at index *i* in the output are given by the samples at index *i* in *sample* |
| parameterized by the values of *low* and *high* at index *i*. |
| |
| Examples:: |
| |
| random_pdf_uniform(sample=`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>] ], low=[<span class="num">0</span>], high=[<span class="num">10</span>]) = [<span class="num">0.1</span>, <span class="num">0.1</span>, <span class="num">0.1</span>, <span class="num">0.1</span>] |
| |
| sample = `[ `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], |
| [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ], |
| `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], |
| [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ] ] |
| low = `[ [<span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>] ] |
| high = `[ [ <span class="num">5</span>, <span class="num">10</span>], |
| [<span class="num">15</span>, <span class="num">20</span>] ] |
| random_pdf_uniform(sample=sample, low=low, high=high) = |
| `[ `[ [<span class="num">0.2</span>, <span class="num">0.2</span>, <span class="num">0.2</span> ], |
| [<span class="num">0.1</span>, <span class="num">0.1</span>, <span class="num">0.1</span> ] ], |
| `[ [<span class="num">0.06667</span>, <span class="num">0.06667</span>, <span class="num">0.06667</span>], |
| [<span class="num">0.05</span>, <span class="num">0.05</span>, <span class="num">0.05</span> ] ] ] |
| |
| |
| |
| Defined in src/operator/random/pdf_op.cc:L297</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_poisson" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_poisson(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_poisson(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_poisson</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_poisson(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a Poisson distribution. |
| |
| Samples are distributed according to a Poisson distribution parametrized by *lambda* (rate). |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Example:: |
| |
| poisson(lam=<span class="num">4</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">5.</span>, <span class="num">2.</span>], |
| [ <span class="num">4.</span>, <span class="num">6.</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L149</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_poisson" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_poisson(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_poisson(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_poisson</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_poisson(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a Poisson distribution. |
| |
| Samples are distributed according to a Poisson distribution parametrized by *lambda* (rate). |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Example:: |
| |
| poisson(lam=<span class="num">4</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">5.</span>, <span class="num">2.</span>], |
| [ <span class="num">4.</span>, <span class="num">6.</span>] ] |
| |
| |
| Defined in src/operator/random/sample_op.cc:L149</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_randint" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_randint(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_randint(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_randint</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_randint(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a discrete uniform distribution. |
| |
| Samples are uniformly distributed over the half-open interval *[low, high)* |
| (includes *low*, but excludes *high*). |
| |
| Example:: |
| |
| randint(low=<span class="num">0</span>, high=<span class="num">5</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">0</span>, <span class="num">2</span>], |
| [ <span class="num">3</span>, <span class="num">1</span>] ] |
| |
| |
| |
| Defined in src/operator/random/sample_op.cc:L193</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_randint" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_randint(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_randint(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_randint</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_randint(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a discrete uniform distribution. |
| |
| Samples are uniformly distributed over the half-open interval *[low, high)* |
| (includes *low*, but excludes *high*). |
| |
| Example:: |
| |
| randint(low=<span class="num">0</span>, high=<span class="num">5</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">0</span>, <span class="num">2</span>], |
| [ <span class="num">3</span>, <span class="num">1</span>] ] |
| |
| |
| |
| Defined in src/operator/random/sample_op.cc:L193</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_uniform" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_uniform(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_uniform(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_uniform</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_uniform(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a uniform distribution. |
| |
| .. note:: The existing alias ``uniform`` is deprecated. |
| |
| Samples are uniformly distributed over the half-open interval *[low, high)* |
| (includes *low*, but excludes *high*). |
| |
| Example:: |
| |
| uniform(low=<span class="num">0</span>, high=<span class="num">1</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">0.60276335</span>, <span class="num">0.85794562</span>], |
| [ <span class="num">0.54488319</span>, <span class="num">0.84725171</span>] ] |
| |
| |
| |
| Defined in src/operator/random/sample_op.cc:L95</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#random_uniform" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="random_uniform(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="random_uniform(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">random_uniform</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@random_uniform(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a uniform distribution. |
| |
| .. note:: The existing alias ``uniform`` is deprecated. |
| |
| Samples are uniformly distributed over the half-open interval *[low, high)* |
| (includes *low*, but excludes *high*). |
| |
| Example:: |
| |
| uniform(low=<span class="num">0</span>, high=<span class="num">1</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">0.60276335</span>, <span class="num">0.85794562</span>], |
| [ <span class="num">0.54488319</span>, <span class="num">0.84725171</span>] ] |
| |
| |
| |
| Defined in src/operator/random/sample_op.cc:L95</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ravel_multi_index" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ravel_multi_index(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ravel_multi_index(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ravel_multi_index</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ravel_multi_index(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Converts a batch of index arrays into an array of flat indices. The operator follows numpy conventions so a single multi index is given by a column of the input matrix. The leading dimension may be left unspecified by using -<span class="num">1</span> as placeholder. |
| |
| Examples:: |
| |
| A = `[ [<span class="num">3</span>,<span class="num">6</span>,<span class="num">6</span>],[<span class="num">4</span>,<span class="num">5</span>,<span class="num">1</span>] ] |
| ravel(A, shape=(<span class="num">7</span>,<span class="num">6</span>)) = [<span class="num">22</span>,<span class="num">41</span>,<span class="num">37</span>] |
| ravel(A, shape=(-<span class="num">1</span>,<span class="num">6</span>)) = [<span class="num">22</span>,<span class="num">41</span>,<span class="num">37</span>] |
| |
| |
| |
| Defined in src/operator/tensor/ravel.cc:L41</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#ravel_multi_index" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="ravel_multi_index(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="ravel_multi_index(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ravel_multi_index</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ravel_multi_index(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Converts a batch of index arrays into an array of flat indices. The operator follows numpy conventions so a single multi index is given by a column of the input matrix. The leading dimension may be left unspecified by using -<span class="num">1</span> as placeholder. |
| |
| Examples:: |
| |
| A = `[ [<span class="num">3</span>,<span class="num">6</span>,<span class="num">6</span>],[<span class="num">4</span>,<span class="num">5</span>,<span class="num">1</span>] ] |
| ravel(A, shape=(<span class="num">7</span>,<span class="num">6</span>)) = [<span class="num">22</span>,<span class="num">41</span>,<span class="num">37</span>] |
| ravel(A, shape=(-<span class="num">1</span>,<span class="num">6</span>)) = [<span class="num">22</span>,<span class="num">41</span>,<span class="num">37</span>] |
| |
| |
| |
| Defined in src/operator/tensor/ravel.cc:L41</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rcbrt" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rcbrt(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rcbrt(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rcbrt</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rcbrt(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse cube-root value of the input. |
| |
| .. math:: |
| rcbrt(x) = <span class="num">1</span>/\sqrt[<span class="num">3</span>]{x} |
| |
| Example:: |
| |
| rcbrt([<span class="num">1</span>,<span class="num">8</span>,-<span class="num">125</span>]) = [<span class="num">1.0</span>, <span class="num">0.5</span>, -<span class="num">0.2</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L323</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rcbrt" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rcbrt(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rcbrt(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rcbrt</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rcbrt(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse cube-root value of the input. |
| |
| .. math:: |
| rcbrt(x) = <span class="num">1</span>/\sqrt[<span class="num">3</span>]{x} |
| |
| Example:: |
| |
| rcbrt([<span class="num">1</span>,<span class="num">8</span>,-<span class="num">125</span>]) = [<span class="num">1.0</span>, <span class="num">0.5</span>, -<span class="num">0.2</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L323</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reciprocal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reciprocal(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reciprocal(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reciprocal</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reciprocal(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the reciprocal of the argument, element-wise. |
| |
| Calculates <span class="num">1</span>/x. |
| |
| Example:: |
| |
| reciprocal([-<span class="num">2</span>, <span class="num">1</span>, <span class="num">3</span>, <span class="num">1.6</span>, <span class="num">0.2</span>]) = [-<span class="num">0.5</span>, <span class="num">1.0</span>, <span class="num">0.33333334</span>, <span class="num">0.625</span>, <span class="num">5.0</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L43</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reciprocal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reciprocal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reciprocal(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reciprocal</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reciprocal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the reciprocal of the argument, element-wise. |
| |
| Calculates <span class="num">1</span>/x. |
| |
| Example:: |
| |
| reciprocal([-<span class="num">2</span>, <span class="num">1</span>, <span class="num">3</span>, <span class="num">1.6</span>, <span class="num">0.2</span>]) = [-<span class="num">0.5</span>, <span class="num">1.0</span>, <span class="num">0.33333334</span>, <span class="num">0.625</span>, <span class="num">5.0</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L43</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#relu" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="relu(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="relu(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">relu</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@relu(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes rectified linear activation. |
| |
| .. math:: |
| max(features, <span class="num">0</span>) |
| |
| The storage <span class="kw">type</span> of ``relu`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - relu(default) = default |
| - relu(row_sparse) = row_sparse |
| - relu(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L85</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#relu" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="relu(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="relu(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">relu</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@relu(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes rectified linear activation. |
| |
| .. math:: |
| max(features, <span class="num">0</span>) |
| |
| The storage <span class="kw">type</span> of ``relu`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - relu(default) = default |
| - relu(row_sparse) = row_sparse |
| - relu(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L85</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#repeat" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="repeat(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="repeat(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">repeat</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@repeat(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Repeats elements of an array. |
| By default, ``repeat`` flattens the input array into <span class="num">1</span>-D and then repeats the |
| elements:: |
| x = `[ [ <span class="num">1</span>, <span class="num">2</span>], |
| [ <span class="num">3</span>, <span class="num">4</span>] ] |
| repeat(x, repeats=<span class="num">2</span>) = [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">4.</span>] |
| The parameter ``axis`` specifies the axis along which to perform repeat:: |
| repeat(x, repeats=<span class="num">2</span>, axis=<span class="num">1</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">4.</span>] ] |
| repeat(x, repeats=<span class="num">2</span>, axis=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>] ] |
| repeat(x, repeats=<span class="num">2</span>, axis=-<span class="num">1</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">4.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L743</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#repeat" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="repeat(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="repeat(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">repeat</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@repeat(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Repeats elements of an array. |
| By default, ``repeat`` flattens the input array into <span class="num">1</span>-D and then repeats the |
| elements:: |
| x = `[ [ <span class="num">1</span>, <span class="num">2</span>], |
| [ <span class="num">3</span>, <span class="num">4</span>] ] |
| repeat(x, repeats=<span class="num">2</span>) = [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">4.</span>] |
| The parameter ``axis`` specifies the axis along which to perform repeat:: |
| repeat(x, repeats=<span class="num">2</span>, axis=<span class="num">1</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">4.</span>] ] |
| repeat(x, repeats=<span class="num">2</span>, axis=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>] ] |
| repeat(x, repeats=<span class="num">2</span>, axis=-<span class="num">1</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">4.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L743</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reset_arrays" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reset_arrays(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reset_arrays(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reset_arrays</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reset_arrays(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre><span class="std">Set</span> to zero multiple arrays |
| |
| |
| Defined in src/operator/contrib/reset_arrays.cc:L35</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reset_arrays" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reset_arrays(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reset_arrays(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reset_arrays</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reset_arrays(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre><span class="std">Set</span> to zero multiple arrays |
| |
| |
| Defined in src/operator/contrib/reset_arrays.cc:L35</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reshape" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reshape(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reshape(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reshape</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reshape(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reshapes the input array. |
| .. note:: ``Reshape`` is deprecated, use ``reshape`` |
| Given an array and a shape, <span class="kw">this</span> function returns a copy of the array in the <span class="kw">new</span> shape. |
| The shape is a tuple of integers such as (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>). The size of the <span class="kw">new</span> shape should be same as the size of the input array. |
| Example:: |
| reshape([<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [<span class="num">1</span>,<span class="num">2</span>], [<span class="num">3</span>,<span class="num">4</span>] ] |
| <span class="std">Some</span> dimensions of the shape can take special values from the set {<span class="num">0</span>, -<span class="num">1</span>, -<span class="num">2</span>, -<span class="num">3</span>, -<span class="num">4</span>}. The significance of each is explained below: |
| - ``<span class="num">0</span>`` copy <span class="kw">this</span> dimension from the input to the output shape. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">4</span>,<span class="num">0</span>,<span class="num">2</span>), output shape = (<span class="num">4</span>,<span class="num">3</span>,<span class="num">2</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,<span class="num">0</span>,<span class="num">0</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - ``-<span class="num">1</span>`` infers the dimension of the output shape by using the remainder of the input dimensions |
| keeping the size of the <span class="kw">new</span> array same as that of the input array. |
| At most one dimension of shape can be -<span class="num">1.</span> |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">6</span>,<span class="num">1</span>,-<span class="num">1</span>), output shape = (<span class="num">6</span>,<span class="num">1</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">3</span>,-<span class="num">1</span>,<span class="num">8</span>), output shape = (<span class="num">3</span>,<span class="num">1</span>,<span class="num">8</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape=(-<span class="num">1</span>,), output shape = (<span class="num">24</span>,) |
| - ``-<span class="num">2</span>`` copy all/remainder of the input dimensions to the output shape. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">2</span>,), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,-<span class="num">2</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">2</span>,<span class="num">1</span>,<span class="num">1</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">1</span>,<span class="num">1</span>) |
| - ``-<span class="num">3</span>`` use the product of two consecutive dimensions of the input shape as the output dimension. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">3</span>,<span class="num">4</span>), output shape = (<span class="num">6</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>), shape = (-<span class="num">3</span>,-<span class="num">3</span>), output shape = (<span class="num">6</span>,<span class="num">20</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">0</span>,-<span class="num">3</span>), output shape = (<span class="num">2</span>,<span class="num">12</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">3</span>,-<span class="num">2</span>), output shape = (<span class="num">6</span>,<span class="num">4</span>) |
| - ``-<span class="num">4</span>`` split one dimension of the input into two dimensions passed subsequent to -<span class="num">4</span> in shape (can contain -<span class="num">1</span>). |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">4</span>,<span class="num">1</span>,<span class="num">2</span>,-<span class="num">2</span>), output shape =(<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,-<span class="num">4</span>,-<span class="num">1</span>,<span class="num">3</span>,-<span class="num">2</span>), output shape = (<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">4</span>) |
| If the argument `reverse` is set to <span class="num">1</span>, then the special values are inferred from right to left. |
| Example:: |
| - without reverse=<span class="num">1</span>, <span class="kw">for</span> input shape = (<span class="num">10</span>,<span class="num">5</span>,<span class="num">4</span>), shape = (-<span class="num">1</span>,<span class="num">0</span>), output shape would be (<span class="num">40</span>,<span class="num">5</span>) |
| - <span class="kw">with</span> reverse=<span class="num">1</span>, output shape will be (<span class="num">50</span>,<span class="num">4</span>). |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L174</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reshape" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reshape(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reshape(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reshape</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reshape(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reshapes the input array. |
| .. note:: ``Reshape`` is deprecated, use ``reshape`` |
| Given an array and a shape, <span class="kw">this</span> function returns a copy of the array in the <span class="kw">new</span> shape. |
| The shape is a tuple of integers such as (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>). The size of the <span class="kw">new</span> shape should be same as the size of the input array. |
| Example:: |
| reshape([<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [<span class="num">1</span>,<span class="num">2</span>], [<span class="num">3</span>,<span class="num">4</span>] ] |
| <span class="std">Some</span> dimensions of the shape can take special values from the set {<span class="num">0</span>, -<span class="num">1</span>, -<span class="num">2</span>, -<span class="num">3</span>, -<span class="num">4</span>}. The significance of each is explained below: |
| - ``<span class="num">0</span>`` copy <span class="kw">this</span> dimension from the input to the output shape. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">4</span>,<span class="num">0</span>,<span class="num">2</span>), output shape = (<span class="num">4</span>,<span class="num">3</span>,<span class="num">2</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,<span class="num">0</span>,<span class="num">0</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - ``-<span class="num">1</span>`` infers the dimension of the output shape by using the remainder of the input dimensions |
| keeping the size of the <span class="kw">new</span> array same as that of the input array. |
| At most one dimension of shape can be -<span class="num">1.</span> |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">6</span>,<span class="num">1</span>,-<span class="num">1</span>), output shape = (<span class="num">6</span>,<span class="num">1</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">3</span>,-<span class="num">1</span>,<span class="num">8</span>), output shape = (<span class="num">3</span>,<span class="num">1</span>,<span class="num">8</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape=(-<span class="num">1</span>,), output shape = (<span class="num">24</span>,) |
| - ``-<span class="num">2</span>`` copy all/remainder of the input dimensions to the output shape. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">2</span>,), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,-<span class="num">2</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">2</span>,<span class="num">1</span>,<span class="num">1</span>), output shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">1</span>,<span class="num">1</span>) |
| - ``-<span class="num">3</span>`` use the product of two consecutive dimensions of the input shape as the output dimension. |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">3</span>,<span class="num">4</span>), output shape = (<span class="num">6</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>), shape = (-<span class="num">3</span>,-<span class="num">3</span>), output shape = (<span class="num">6</span>,<span class="num">20</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">0</span>,-<span class="num">3</span>), output shape = (<span class="num">2</span>,<span class="num">12</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">3</span>,-<span class="num">2</span>), output shape = (<span class="num">6</span>,<span class="num">4</span>) |
| - ``-<span class="num">4</span>`` split one dimension of the input into two dimensions passed subsequent to -<span class="num">4</span> in shape (can contain -<span class="num">1</span>). |
| Example:: |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (-<span class="num">4</span>,<span class="num">1</span>,<span class="num">2</span>,-<span class="num">2</span>), output shape =(<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>) |
| - input shape = (<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>), shape = (<span class="num">2</span>,-<span class="num">4</span>,-<span class="num">1</span>,<span class="num">3</span>,-<span class="num">2</span>), output shape = (<span class="num">2</span>,<span class="num">1</span>,<span class="num">3</span>,<span class="num">4</span>) |
| If the argument `reverse` is set to <span class="num">1</span>, then the special values are inferred from right to left. |
| Example:: |
| - without reverse=<span class="num">1</span>, <span class="kw">for</span> input shape = (<span class="num">10</span>,<span class="num">5</span>,<span class="num">4</span>), shape = (-<span class="num">1</span>,<span class="num">0</span>), output shape would be (<span class="num">40</span>,<span class="num">5</span>) |
| - <span class="kw">with</span> reverse=<span class="num">1</span>, output shape will be (<span class="num">50</span>,<span class="num">4</span>). |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L174</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reshape_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reshape_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reshape_like(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reshape_like</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reshape_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reshape some or all dimensions of `lhs` to have the same shape as some or all dimensions of `rhs`. |
| |
| Returns a **view** of the `lhs` array <span class="kw">with</span> a <span class="kw">new</span> shape without altering any data. |
| |
| Example:: |
| |
| x = [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>] |
| y = `[ [<span class="num">0</span>, -<span class="num">4</span>], [<span class="num">3</span>, <span class="num">2</span>], [<span class="num">2</span>, <span class="num">2</span>] ] |
| reshape_like(x, y) = `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">4</span>], [<span class="num">5</span>, <span class="num">6</span>] ] |
| |
| More precise control over how dimensions are inherited is achieved by specifying \ |
| slices over the `lhs` and `rhs` array dimensions. Only the sliced `lhs` dimensions \ |
| are reshaped to the `rhs` sliced dimensions, <span class="kw">with</span> the non-sliced `lhs` dimensions staying the same. |
| |
| Examples:: |
| |
| - lhs shape = (<span class="num">30</span>,<span class="num">7</span>), rhs shape = (<span class="num">15</span>,<span class="num">2</span>,<span class="num">4</span>), lhs_begin=<span class="num">0</span>, lhs_end=<span class="num">1</span>, rhs_begin=<span class="num">0</span>, rhs_end=<span class="num">2</span>, output shape = (<span class="num">15</span>,<span class="num">2</span>,<span class="num">7</span>) |
| - lhs shape = (<span class="num">3</span>, <span class="num">5</span>), rhs shape = (<span class="num">1</span>,<span class="num">15</span>,<span class="num">4</span>), lhs_begin=<span class="num">0</span>, lhs_end=<span class="num">2</span>, rhs_begin=<span class="num">1</span>, rhs_end=<span class="num">2</span>, output shape = (<span class="num">15</span>) |
| |
| Negative indices are supported, and `<span class="std">None</span>` can be used <span class="kw">for</span> either `lhs_end` or `rhs_end` to indicate the end of the range. |
| |
| Example:: |
| |
| - lhs shape = (<span class="num">30</span>, <span class="num">12</span>), rhs shape = (<span class="num">4</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">3</span>), lhs_begin=-<span class="num">1</span>, lhs_end=<span class="std">None</span>, rhs_begin=<span class="num">1</span>, rhs_end=<span class="std">None</span>, output shape = (<span class="num">30</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">3</span>) |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L511</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reshape_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reshape_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reshape_like(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reshape_like</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reshape_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reshape some or all dimensions of `lhs` to have the same shape as some or all dimensions of `rhs`. |
| |
| Returns a **view** of the `lhs` array <span class="kw">with</span> a <span class="kw">new</span> shape without altering any data. |
| |
| Example:: |
| |
| x = [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>, <span class="num">6</span>] |
| y = `[ [<span class="num">0</span>, -<span class="num">4</span>], [<span class="num">3</span>, <span class="num">2</span>], [<span class="num">2</span>, <span class="num">2</span>] ] |
| reshape_like(x, y) = `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">4</span>], [<span class="num">5</span>, <span class="num">6</span>] ] |
| |
| More precise control over how dimensions are inherited is achieved by specifying \ |
| slices over the `lhs` and `rhs` array dimensions. Only the sliced `lhs` dimensions \ |
| are reshaped to the `rhs` sliced dimensions, <span class="kw">with</span> the non-sliced `lhs` dimensions staying the same. |
| |
| Examples:: |
| |
| - lhs shape = (<span class="num">30</span>,<span class="num">7</span>), rhs shape = (<span class="num">15</span>,<span class="num">2</span>,<span class="num">4</span>), lhs_begin=<span class="num">0</span>, lhs_end=<span class="num">1</span>, rhs_begin=<span class="num">0</span>, rhs_end=<span class="num">2</span>, output shape = (<span class="num">15</span>,<span class="num">2</span>,<span class="num">7</span>) |
| - lhs shape = (<span class="num">3</span>, <span class="num">5</span>), rhs shape = (<span class="num">1</span>,<span class="num">15</span>,<span class="num">4</span>), lhs_begin=<span class="num">0</span>, lhs_end=<span class="num">2</span>, rhs_begin=<span class="num">1</span>, rhs_end=<span class="num">2</span>, output shape = (<span class="num">15</span>) |
| |
| Negative indices are supported, and `<span class="std">None</span>` can be used <span class="kw">for</span> either `lhs_end` or `rhs_end` to indicate the end of the range. |
| |
| Example:: |
| |
| - lhs shape = (<span class="num">30</span>, <span class="num">12</span>), rhs shape = (<span class="num">4</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">3</span>), lhs_begin=-<span class="num">1</span>, lhs_end=<span class="std">None</span>, rhs_begin=<span class="num">1</span>, rhs_end=<span class="std">None</span>, output shape = (<span class="num">30</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">3</span>) |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L511</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reverse" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reverse(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reverse(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reverse</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reverse(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reverses the order of elements along given axis <span class="kw">while</span> preserving array shape. |
| Note: reverse and flip are equivalent. We use reverse in the following examples. |
| Examples:: |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| reverse(x, axis=<span class="num">0</span>) = `[ [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] |
| reverse(x, axis=<span class="num">1</span>) = `[ [ <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">9.</span>, <span class="num">8.</span>, <span class="num">7.</span>, <span class="num">6.</span>, <span class="num">5.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L831</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#reverse" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="reverse(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="reverse(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">reverse</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@reverse(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Reverses the order of elements along given axis <span class="kw">while</span> preserving array shape. |
| Note: reverse and flip are equivalent. We use reverse in the following examples. |
| Examples:: |
| x = `[ [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>] ] |
| reverse(x, axis=<span class="num">0</span>) = `[ [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>, <span class="num">9.</span>], |
| [ <span class="num">0.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] |
| reverse(x, axis=<span class="num">1</span>) = `[ [ <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">0.</span>], |
| [ <span class="num">9.</span>, <span class="num">8.</span>, <span class="num">7.</span>, <span class="num">6.</span>, <span class="num">5.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L831</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rint" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rint(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rint(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rint</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rint(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise rounded value to the nearest integer of the input. |
| |
| .. note:: |
| - For input ``n.<span class="num">5</span>`` ``rint`` returns ``n`` <span class="kw">while</span> ``round`` returns ``n+<span class="num">1</span>``. |
| - For input ``-n.<span class="num">5</span>`` both ``rint`` and ``round`` returns ``-n-<span class="num">1</span>``. |
| |
| Example:: |
| |
| rint([-<span class="num">1.5</span>, <span class="num">1.5</span>, -<span class="num">1.9</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, <span class="num">1.</span>, -<span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``rint`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - rint(default) = default |
| - rint(row_sparse) = row_sparse |
| - rint(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L798</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rint" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rint(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rint(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rint</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rint(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise rounded value to the nearest integer of the input. |
| |
| .. note:: |
| - For input ``n.<span class="num">5</span>`` ``rint`` returns ``n`` <span class="kw">while</span> ``round`` returns ``n+<span class="num">1</span>``. |
| - For input ``-n.<span class="num">5</span>`` both ``rint`` and ``round`` returns ``-n-<span class="num">1</span>``. |
| |
| Example:: |
| |
| rint([-<span class="num">1.5</span>, <span class="num">1.5</span>, -<span class="num">1.9</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, <span class="num">1.</span>, -<span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``rint`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - rint(default) = default |
| - rint(row_sparse) = row_sparse |
| - rint(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L798</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rmsprop_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rmsprop_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rmsprop_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rmsprop_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rmsprop_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> `RMSProp` optimizer. |
| |
| `RMSprop` is a variant of stochastic gradient descent where the gradients are |
| divided by a cache which grows <span class="kw">with</span> the sum of squares of recent gradients? |
| |
| `RMSProp` is similar to `AdaGrad`, a popular variant of `SGD` which adaptively |
| tunes the learning rate of each parameter. `AdaGrad` lowers the learning rate <span class="kw">for</span> |
| each parameter monotonically over the course of training. |
| While <span class="kw">this</span> is analytically motivated <span class="kw">for</span> convex optimizations, it may not be ideal |
| <span class="kw">for</span> non-convex problems. `RMSProp` deals <span class="kw">with</span> <span class="kw">this</span> heuristically by allowing the |
| learning rates to rebound as the denominator decays over time. |
| |
| Define the Root Mean Square (RMS) error criterion of the gradient as |
| :math:`RMS[g]_t = \sqrt{E[g^<span class="num">2</span>]_t + \epsilon}`, where :math:`g` represents |
| gradient and :math:`E[g^<span class="num">2</span>]_t` is the decaying average over past squared gradient. |
| |
| The :math:`E[g^<span class="num">2</span>]_t` is given by: |
| |
| .. math:: |
| E[g^<span class="num">2</span>]_t = \gamma * E[g^<span class="num">2</span>]_{t-<span class="num">1</span>} + (<span class="num">1</span>-\gamma) * g_t^<span class="num">2</span> |
| |
| The update step is |
| |
| .. math:: |
| \theta_{t+<span class="num">1</span>} = \theta_t - \frac{\eta}{RMS[g]_t} g_t |
| |
| The RMSProp code follows the version in |
| http:<span class="cmt">//www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slides_lec6.pdf</span> |
| Tieleman & Hinton, <span class="num">2012.</span> |
| |
| Hinton suggests the momentum term :math:`\gamma` to be <span class="num">0.9</span> and the learning rate |
| :math:`\eta` to be <span class="num">0.001</span>. |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L796</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rmsprop_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rmsprop_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rmsprop_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rmsprop_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rmsprop_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> `RMSProp` optimizer. |
| |
| `RMSprop` is a variant of stochastic gradient descent where the gradients are |
| divided by a cache which grows <span class="kw">with</span> the sum of squares of recent gradients? |
| |
| `RMSProp` is similar to `AdaGrad`, a popular variant of `SGD` which adaptively |
| tunes the learning rate of each parameter. `AdaGrad` lowers the learning rate <span class="kw">for</span> |
| each parameter monotonically over the course of training. |
| While <span class="kw">this</span> is analytically motivated <span class="kw">for</span> convex optimizations, it may not be ideal |
| <span class="kw">for</span> non-convex problems. `RMSProp` deals <span class="kw">with</span> <span class="kw">this</span> heuristically by allowing the |
| learning rates to rebound as the denominator decays over time. |
| |
| Define the Root Mean Square (RMS) error criterion of the gradient as |
| :math:`RMS[g]_t = \sqrt{E[g^<span class="num">2</span>]_t + \epsilon}`, where :math:`g` represents |
| gradient and :math:`E[g^<span class="num">2</span>]_t` is the decaying average over past squared gradient. |
| |
| The :math:`E[g^<span class="num">2</span>]_t` is given by: |
| |
| .. math:: |
| E[g^<span class="num">2</span>]_t = \gamma * E[g^<span class="num">2</span>]_{t-<span class="num">1</span>} + (<span class="num">1</span>-\gamma) * g_t^<span class="num">2</span> |
| |
| The update step is |
| |
| .. math:: |
| \theta_{t+<span class="num">1</span>} = \theta_t - \frac{\eta}{RMS[g]_t} g_t |
| |
| The RMSProp code follows the version in |
| http:<span class="cmt">//www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slides_lec6.pdf</span> |
| Tieleman & Hinton, <span class="num">2012.</span> |
| |
| Hinton suggests the momentum term :math:`\gamma` to be <span class="num">0.9</span> and the learning rate |
| :math:`\eta` to be <span class="num">0.001</span>. |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L796</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rmspropalex_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rmspropalex_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rmspropalex_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rmspropalex_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rmspropalex_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> RMSPropAlex optimizer. |
| |
| `RMSPropAlex` is non-centered version of `RMSProp`. |
| |
| Define :math:`E[g^<span class="num">2</span>]_t` is the decaying average over past squared gradient and |
| :math:`E[g]_t` is the decaying average over past gradient. |
| |
| .. math:: |
| E[g^<span class="num">2</span>]_t = \gamma_1 * E[g^<span class="num">2</span>]_{t-<span class="num">1</span>} + (<span class="num">1</span> - \gamma_1) * g_t^<span class="num">2</span>\\ |
| E[g]_t = \gamma_1 * E[g]_{t-<span class="num">1</span>} + (<span class="num">1</span> - \gamma_1) * g_t\\ |
| \Delta_t = \gamma_2 * \Delta_{t-<span class="num">1</span>} - \frac{\eta}{\sqrt{E[g^<span class="num">2</span>]_t - E[g]_t^<span class="num">2</span> + \epsilon}} g_t\\ |
| |
| The update step is |
| |
| .. math:: |
| \theta_{t+<span class="num">1</span>} = \theta_t + \Delta_t |
| |
| The RMSPropAlex code follows the version in |
| http:<span class="cmt">//arxiv.org/pdf/1308.0850v5.pdf Eq(38) - Eq(45) by Alex Graves, 2013.</span> |
| |
| Graves suggests the momentum term :math:`\gamma_1` to be <span class="num">0.95</span>, :math:`\gamma_2` |
| to be <span class="num">0.9</span> and the learning rate :math:`\eta` to be <span class="num">0.0001</span>. |
| |
| |
| Defined in src/operator/optimizer_op.cc:L835</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rmspropalex_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rmspropalex_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rmspropalex_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rmspropalex_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rmspropalex_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> RMSPropAlex optimizer. |
| |
| `RMSPropAlex` is non-centered version of `RMSProp`. |
| |
| Define :math:`E[g^<span class="num">2</span>]_t` is the decaying average over past squared gradient and |
| :math:`E[g]_t` is the decaying average over past gradient. |
| |
| .. math:: |
| E[g^<span class="num">2</span>]_t = \gamma_1 * E[g^<span class="num">2</span>]_{t-<span class="num">1</span>} + (<span class="num">1</span> - \gamma_1) * g_t^<span class="num">2</span>\\ |
| E[g]_t = \gamma_1 * E[g]_{t-<span class="num">1</span>} + (<span class="num">1</span> - \gamma_1) * g_t\\ |
| \Delta_t = \gamma_2 * \Delta_{t-<span class="num">1</span>} - \frac{\eta}{\sqrt{E[g^<span class="num">2</span>]_t - E[g]_t^<span class="num">2</span> + \epsilon}} g_t\\ |
| |
| The update step is |
| |
| .. math:: |
| \theta_{t+<span class="num">1</span>} = \theta_t + \Delta_t |
| |
| The RMSPropAlex code follows the version in |
| http:<span class="cmt">//arxiv.org/pdf/1308.0850v5.pdf Eq(38) - Eq(45) by Alex Graves, 2013.</span> |
| |
| Graves suggests the momentum term :math:`\gamma_1` to be <span class="num">0.95</span>, :math:`\gamma_2` |
| to be <span class="num">0.9</span> and the learning rate :math:`\eta` to be <span class="num">0.0001</span>. |
| |
| |
| Defined in src/operator/optimizer_op.cc:L835</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#round" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="round(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="round(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">round</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@round(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise rounded value to the nearest integer of the input. |
| |
| Example:: |
| |
| round([-<span class="num">1.5</span>, <span class="num">1.5</span>, -<span class="num">1.9</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, <span class="num">2.</span>, -<span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``round`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - round(default) = default |
| - round(row_sparse) = row_sparse |
| - round(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L777</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#round" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="round(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="round(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">round</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@round(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise rounded value to the nearest integer of the input. |
| |
| Example:: |
| |
| round([-<span class="num">1.5</span>, <span class="num">1.5</span>, -<span class="num">1.9</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, <span class="num">2.</span>, -<span class="num">2.</span>, <span class="num">2.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``round`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - round(default) = default |
| - round(row_sparse) = row_sparse |
| - round(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L777</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rsqrt" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rsqrt(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rsqrt(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rsqrt</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rsqrt(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse square-root value of the input. |
| |
| .. math:: |
| rsqrt(x) = <span class="num">1</span>/\sqrt{x} |
| |
| Example:: |
| |
| rsqrt([<span class="num">4</span>,<span class="num">9</span>,<span class="num">16</span>]) = [<span class="num">0.5</span>, <span class="num">0.33333334</span>, <span class="num">0.25</span>] |
| |
| The storage <span class="kw">type</span> of ``rsqrt`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L221</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#rsqrt" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="rsqrt(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="rsqrt(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">rsqrt</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@rsqrt(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise inverse square-root value of the input. |
| |
| .. math:: |
| rsqrt(x) = <span class="num">1</span>/\sqrt{x} |
| |
| Example:: |
| |
| rsqrt([<span class="num">4</span>,<span class="num">9</span>,<span class="num">16</span>]) = [<span class="num">0.5</span>, <span class="num">0.33333334</span>, <span class="num">0.25</span>] |
| |
| The storage <span class="kw">type</span> of ``rsqrt`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L221</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_exponential" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_exponential(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_exponential(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_exponential</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_exponential(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| exponential distributions <span class="kw">with</span> parameters lambda (rate). |
| |
| The parameters of the distributions are provided as an input array. |
| Let *[s]* be the shape of the input array, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input array, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input value at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input array. |
| |
| Examples:: |
| |
| lam = [ <span class="num">1.0</span>, <span class="num">8.5</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_exponential(lam) = [ <span class="num">0.51837951</span>, <span class="num">0.09994757</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_exponential(lam, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.51837951</span>, <span class="num">0.19866663</span>], |
| [ <span class="num">0.09994757</span>, <span class="num">0.50447971</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L283</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_exponential" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_exponential(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_exponential(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_exponential</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_exponential(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| exponential distributions <span class="kw">with</span> parameters lambda (rate). |
| |
| The parameters of the distributions are provided as an input array. |
| Let *[s]* be the shape of the input array, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input array, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input value at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input array. |
| |
| Examples:: |
| |
| lam = [ <span class="num">1.0</span>, <span class="num">8.5</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_exponential(lam) = [ <span class="num">0.51837951</span>, <span class="num">0.09994757</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_exponential(lam, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.51837951</span>, <span class="num">0.19866663</span>], |
| [ <span class="num">0.09994757</span>, <span class="num">0.50447971</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L283</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_gamma" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_gamma(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_gamma(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_gamma</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_gamma(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| gamma distributions <span class="kw">with</span> parameters *alpha* (shape) and *beta* (scale). |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Examples:: |
| |
| alpha = [ <span class="num">0.0</span>, <span class="num">2.5</span> ] |
| beta = [ <span class="num">1.0</span>, <span class="num">0.7</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_gamma(alpha, beta) = [ <span class="num">0.</span> , <span class="num">2.25797319</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_gamma(alpha, beta, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.</span> , <span class="num">0.</span> ], |
| [ <span class="num">2.25797319</span>, <span class="num">1.70734084</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L281</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_gamma" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_gamma(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_gamma(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_gamma</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_gamma(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| gamma distributions <span class="kw">with</span> parameters *alpha* (shape) and *beta* (scale). |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Examples:: |
| |
| alpha = [ <span class="num">0.0</span>, <span class="num">2.5</span> ] |
| beta = [ <span class="num">1.0</span>, <span class="num">0.7</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_gamma(alpha, beta) = [ <span class="num">0.</span> , <span class="num">2.25797319</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_gamma(alpha, beta, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.</span> , <span class="num">0.</span> ], |
| [ <span class="num">2.25797319</span>, <span class="num">1.70734084</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L281</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_generalized_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_generalized_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_generalized_negative_binomial(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_generalized_negative_binomial</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_generalized_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| generalized negative binomial distributions <span class="kw">with</span> parameters *mu* (mean) and *alpha* (dispersion). |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Examples:: |
| |
| mu = [ <span class="num">2.0</span>, <span class="num">2.5</span> ] |
| alpha = [ <span class="num">1.0</span>, <span class="num">0.1</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_generalized_negative_binomial(mu, alpha) = [ <span class="num">0.</span>, <span class="num">3.</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_generalized_negative_binomial(mu, alpha, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.</span>, <span class="num">3.</span>], |
| [ <span class="num">3.</span>, <span class="num">1.</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L292</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_generalized_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_generalized_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_generalized_negative_binomial(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_generalized_negative_binomial</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_generalized_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| generalized negative binomial distributions <span class="kw">with</span> parameters *mu* (mean) and *alpha* (dispersion). |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Examples:: |
| |
| mu = [ <span class="num">2.0</span>, <span class="num">2.5</span> ] |
| alpha = [ <span class="num">1.0</span>, <span class="num">0.1</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_generalized_negative_binomial(mu, alpha) = [ <span class="num">0.</span>, <span class="num">3.</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_generalized_negative_binomial(mu, alpha, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.</span>, <span class="num">3.</span>], |
| [ <span class="num">3.</span>, <span class="num">1.</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L292</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_multinomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_multinomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_multinomial(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_multinomial</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_multinomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple multinomial distributions. |
| |
| *data* is an *n* dimensional array whose last dimension has length *k*, where |
| *k* is the number of possible outcomes of each multinomial distribution. This |
| operator will draw *shape* samples from each distribution. If shape is empty |
| one sample will be drawn from each distribution. |
| |
| If *get_prob* is <span class="kw">true</span>, a second array containing log likelihood of the drawn |
| samples will also be returned. This is usually used <span class="kw">for</span> reinforcement learning |
| where you can provide reward as head gradient <span class="kw">for</span> <span class="kw">this</span> array to estimate |
| gradient. |
| |
| Note that the input distribution must be normalized, i.e. *data* must sum to |
| <span class="num">1</span> along its last axis. |
| |
| Examples:: |
| |
| probs = `[ [<span class="num">0</span>, <span class="num">0.1</span>, <span class="num">0.2</span>, <span class="num">0.3</span>, <span class="num">0.4</span>], [<span class="num">0.4</span>, <span class="num">0.3</span>, <span class="num">0.2</span>, <span class="num">0.1</span>, <span class="num">0</span>] ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_multinomial(probs) = [<span class="num">3</span>, <span class="num">0</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_multinomial(probs, shape=(<span class="num">2</span>)) = `[ [<span class="num">4</span>, <span class="num">2</span>], |
| [<span class="num">0</span>, <span class="num">0</span>] ] |
| |
| <span class="cmt">// requests log likelihood</span> |
| sample_multinomial(probs, get_prob=True) = [<span class="num">2</span>, <span class="num">1</span>], [<span class="num">0.2</span>, <span class="num">0.3</span>]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_multinomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_multinomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_multinomial(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_multinomial</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_multinomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple multinomial distributions. |
| |
| *data* is an *n* dimensional array whose last dimension has length *k*, where |
| *k* is the number of possible outcomes of each multinomial distribution. This |
| operator will draw *shape* samples from each distribution. If shape is empty |
| one sample will be drawn from each distribution. |
| |
| If *get_prob* is <span class="kw">true</span>, a second array containing log likelihood of the drawn |
| samples will also be returned. This is usually used <span class="kw">for</span> reinforcement learning |
| where you can provide reward as head gradient <span class="kw">for</span> <span class="kw">this</span> array to estimate |
| gradient. |
| |
| Note that the input distribution must be normalized, i.e. *data* must sum to |
| <span class="num">1</span> along its last axis. |
| |
| Examples:: |
| |
| probs = `[ [<span class="num">0</span>, <span class="num">0.1</span>, <span class="num">0.2</span>, <span class="num">0.3</span>, <span class="num">0.4</span>], [<span class="num">0.4</span>, <span class="num">0.3</span>, <span class="num">0.2</span>, <span class="num">0.1</span>, <span class="num">0</span>] ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_multinomial(probs) = [<span class="num">3</span>, <span class="num">0</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_multinomial(probs, shape=(<span class="num">2</span>)) = `[ [<span class="num">4</span>, <span class="num">2</span>], |
| [<span class="num">0</span>, <span class="num">0</span>] ] |
| |
| <span class="cmt">// requests log likelihood</span> |
| sample_multinomial(probs, get_prob=True) = [<span class="num">2</span>, <span class="num">1</span>], [<span class="num">0.2</span>, <span class="num">0.3</span>]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_negative_binomial(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_negative_binomial</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_negative_binomial(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| negative binomial distributions <span class="kw">with</span> parameters *k* (failure limit) and *p* (failure probability). |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Examples:: |
| |
| k = [ <span class="num">20</span>, <span class="num">49</span> ] |
| p = [ <span class="num">0.4</span> , <span class="num">0.77</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_negative_binomial(k, p) = [ <span class="num">15.</span>, <span class="num">16.</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_negative_binomial(k, p, shape=(<span class="num">2</span>)) = `[ [ <span class="num">15.</span>, <span class="num">50.</span>], |
| [ <span class="num">16.</span>, <span class="num">12.</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L288</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_negative_binomial" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_negative_binomial(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_negative_binomial</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_negative_binomial(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| negative binomial distributions <span class="kw">with</span> parameters *k* (failure limit) and *p* (failure probability). |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Examples:: |
| |
| k = [ <span class="num">20</span>, <span class="num">49</span> ] |
| p = [ <span class="num">0.4</span> , <span class="num">0.77</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_negative_binomial(k, p) = [ <span class="num">15.</span>, <span class="num">16.</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_negative_binomial(k, p, shape=(<span class="num">2</span>)) = `[ [ <span class="num">15.</span>, <span class="num">50.</span>], |
| [ <span class="num">16.</span>, <span class="num">12.</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L288</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_normal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_normal(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_normal(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_normal</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_normal(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| normal distributions <span class="kw">with</span> parameters *mu* (mean) and *sigma* (standard deviation). |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Examples:: |
| |
| mu = [ <span class="num">0.0</span>, <span class="num">2.5</span> ] |
| sigma = [ <span class="num">1.0</span>, <span class="num">3.7</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_normal(mu, sigma) = [-<span class="num">0.56410581</span>, <span class="num">0.95934606</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_normal(mu, sigma, shape=(<span class="num">2</span>)) = `[ [-<span class="num">0.56410581</span>, <span class="num">0.2928229</span> ], |
| [ <span class="num">0.95934606</span>, <span class="num">4.48287058</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L278</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_normal" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_normal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_normal(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_normal</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_normal(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| normal distributions <span class="kw">with</span> parameters *mu* (mean) and *sigma* (standard deviation). |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Examples:: |
| |
| mu = [ <span class="num">0.0</span>, <span class="num">2.5</span> ] |
| sigma = [ <span class="num">1.0</span>, <span class="num">3.7</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_normal(mu, sigma) = [-<span class="num">0.56410581</span>, <span class="num">0.95934606</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_normal(mu, sigma, shape=(<span class="num">2</span>)) = `[ [-<span class="num">0.56410581</span>, <span class="num">0.2928229</span> ], |
| [ <span class="num">0.95934606</span>, <span class="num">4.48287058</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L278</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_poisson" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_poisson(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_poisson(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_poisson</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_poisson(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| Poisson distributions <span class="kw">with</span> parameters lambda (rate). |
| |
| The parameters of the distributions are provided as an input array. |
| Let *[s]* be the shape of the input array, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input array, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input value at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input array. |
| |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Examples:: |
| |
| lam = [ <span class="num">1.0</span>, <span class="num">8.5</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_poisson(lam) = [ <span class="num">0.</span>, <span class="num">13.</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_poisson(lam, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.</span>, <span class="num">4.</span>], |
| [ <span class="num">13.</span>, <span class="num">8.</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L285</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_poisson" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_poisson(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_poisson(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_poisson</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_poisson(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| Poisson distributions <span class="kw">with</span> parameters lambda (rate). |
| |
| The parameters of the distributions are provided as an input array. |
| Let *[s]* be the shape of the input array, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input array, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input value at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input array. |
| |
| Samples will always be returned as a floating point data <span class="kw">type</span>. |
| |
| Examples:: |
| |
| lam = [ <span class="num">1.0</span>, <span class="num">8.5</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_poisson(lam) = [ <span class="num">0.</span>, <span class="num">13.</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_poisson(lam, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.</span>, <span class="num">4.</span>], |
| [ <span class="num">13.</span>, <span class="num">8.</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L285</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_uniform" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_uniform(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_uniform(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_uniform</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_uniform(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| uniform distributions on the intervals given by *[low,high)*. |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Examples:: |
| |
| low = [ <span class="num">0.0</span>, <span class="num">2.5</span> ] |
| high = [ <span class="num">1.0</span>, <span class="num">3.7</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_uniform(low, high) = [ <span class="num">0.40451524</span>, <span class="num">3.18687344</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_uniform(low, high, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.40451524</span>, <span class="num">0.18017688</span>], |
| [ <span class="num">3.18687344</span>, <span class="num">3.68352246</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L276</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sample_uniform" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sample_uniform(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sample_uniform(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sample_uniform</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sample_uniform(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Concurrent sampling from multiple |
| uniform distributions on the intervals given by *[low,high)*. |
| |
| The parameters of the distributions are provided as input arrays. |
| Let *[s]* be the shape of the input arrays, *n* be the dimension of *[s]*, *[t]* |
| be the shape specified as the parameter of the operator, and *m* be the dimension |
| of *[t]*. Then the output will be a *(n+m)*-dimensional array <span class="kw">with</span> shape *[s]x[t]*. |
| |
| For any valid *n*-dimensional index *i* <span class="kw">with</span> respect to the input arrays, *output[i]* |
| will be an *m*-dimensional array that holds randomly drawn samples from the distribution |
| which is parameterized by the input values at index *i*. If the shape parameter of the |
| operator is not set, then one sample will be drawn per distribution and the output array |
| has the same shape as the input arrays. |
| |
| Examples:: |
| |
| low = [ <span class="num">0.0</span>, <span class="num">2.5</span> ] |
| high = [ <span class="num">1.0</span>, <span class="num">3.7</span> ] |
| |
| <span class="cmt">// Draw a single sample for each distribution</span> |
| sample_uniform(low, high) = [ <span class="num">0.40451524</span>, <span class="num">3.18687344</span>] |
| |
| <span class="cmt">// Draw a vector containing two samples for each distribution</span> |
| sample_uniform(low, high, shape=(<span class="num">2</span>)) = `[ [ <span class="num">0.40451524</span>, <span class="num">0.18017688</span>], |
| [ <span class="num">3.18687344</span>, <span class="num">3.68352246</span>] ] |
| |
| |
| Defined in src/operator/random/multisample_op.cc:L276</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#scatter_nd" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="scatter_nd(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="scatter_nd(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">scatter_nd</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@scatter_nd(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Scatters data into a <span class="kw">new</span> tensor according to indices. |
| |
| Given `data` <span class="kw">with</span> shape `(Y_0, ..., Y_{K-<span class="num">1</span>}, X_M, ..., X_{N-<span class="num">1</span>})` and indices <span class="kw">with</span> shape |
| `(M, Y_0, ..., Y_{K-<span class="num">1</span>})`, the output will have shape `(X_0, X_1, ..., X_{N-<span class="num">1</span>})`, |
| where `M <= N`. If `M == N`, data shape should simply be `(Y_0, ..., Y_{K-<span class="num">1</span>})`. |
| |
| The elements in output is defined as follows:: |
| |
| output[indices[<span class="num">0</span>, y_0, ..., y_{K-<span class="num">1</span>}], |
| ..., |
| indices[M-<span class="num">1</span>, y_0, ..., y_{K-<span class="num">1</span>}], |
| x_M, ..., x_{N-<span class="num">1</span>}] = data[y_0, ..., y_{K-<span class="num">1</span>}, x_M, ..., x_{N-<span class="num">1</span>}] |
| |
| all other entries in output are <span class="num">0.</span> |
| |
| .. warning:: |
| |
| If the indices have duplicates, the result will be non-deterministic and |
| the gradient of `scatter_nd` will not be correct!! |
| |
| |
| Examples:: |
| |
| data = [<span class="num">2</span>, <span class="num">3</span>, <span class="num">0</span>] |
| indices = `[ [<span class="num">1</span>, <span class="num">1</span>, <span class="num">0</span>], [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>] ] |
| shape = (<span class="num">2</span>, <span class="num">2</span>) |
| scatter_nd(data, indices, shape) = `[ [<span class="num">0</span>, <span class="num">0</span>], [<span class="num">2</span>, <span class="num">3</span>] ] |
| |
| data = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">4</span>] ], `[ [<span class="num">5</span>, <span class="num">6</span>], [<span class="num">7</span>, <span class="num">8</span>] ] ] |
| indices = `[ [<span class="num">0</span>, <span class="num">1</span>], [<span class="num">1</span>, <span class="num">1</span>] ] |
| shape = (<span class="num">2</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">2</span>) |
| scatter_nd(data, indices, shape) = `[ [`[ [<span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>] ], |
| |
| `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ] ], |
| |
| `[ `[ [<span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>] ], |
| |
| `[ [<span class="num">5</span>, <span class="num">6</span>], |
| [<span class="num">7</span>, <span class="num">8</span>] ] ] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#scatter_nd" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="scatter_nd(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="scatter_nd(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">scatter_nd</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@scatter_nd(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Scatters data into a <span class="kw">new</span> tensor according to indices. |
| |
| Given `data` <span class="kw">with</span> shape `(Y_0, ..., Y_{K-<span class="num">1</span>}, X_M, ..., X_{N-<span class="num">1</span>})` and indices <span class="kw">with</span> shape |
| `(M, Y_0, ..., Y_{K-<span class="num">1</span>})`, the output will have shape `(X_0, X_1, ..., X_{N-<span class="num">1</span>})`, |
| where `M <= N`. If `M == N`, data shape should simply be `(Y_0, ..., Y_{K-<span class="num">1</span>})`. |
| |
| The elements in output is defined as follows:: |
| |
| output[indices[<span class="num">0</span>, y_0, ..., y_{K-<span class="num">1</span>}], |
| ..., |
| indices[M-<span class="num">1</span>, y_0, ..., y_{K-<span class="num">1</span>}], |
| x_M, ..., x_{N-<span class="num">1</span>}] = data[y_0, ..., y_{K-<span class="num">1</span>}, x_M, ..., x_{N-<span class="num">1</span>}] |
| |
| all other entries in output are <span class="num">0.</span> |
| |
| .. warning:: |
| |
| If the indices have duplicates, the result will be non-deterministic and |
| the gradient of `scatter_nd` will not be correct!! |
| |
| |
| Examples:: |
| |
| data = [<span class="num">2</span>, <span class="num">3</span>, <span class="num">0</span>] |
| indices = `[ [<span class="num">1</span>, <span class="num">1</span>, <span class="num">0</span>], [<span class="num">0</span>, <span class="num">1</span>, <span class="num">0</span>] ] |
| shape = (<span class="num">2</span>, <span class="num">2</span>) |
| scatter_nd(data, indices, shape) = `[ [<span class="num">0</span>, <span class="num">0</span>], [<span class="num">2</span>, <span class="num">3</span>] ] |
| |
| data = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">4</span>] ], `[ [<span class="num">5</span>, <span class="num">6</span>], [<span class="num">7</span>, <span class="num">8</span>] ] ] |
| indices = `[ [<span class="num">0</span>, <span class="num">1</span>], [<span class="num">1</span>, <span class="num">1</span>] ] |
| shape = (<span class="num">2</span>, <span class="num">2</span>, <span class="num">2</span>, <span class="num">2</span>) |
| scatter_nd(data, indices, shape) = `[ [`[ [<span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>] ], |
| |
| `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ] ], |
| |
| `[ `[ [<span class="num">0</span>, <span class="num">0</span>], |
| [<span class="num">0</span>, <span class="num">0</span>] ], |
| |
| `[ [<span class="num">5</span>, <span class="num">6</span>], |
| [<span class="num">7</span>, <span class="num">8</span>] ] ] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sgd_mom_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sgd_mom_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sgd_mom_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| However, <span class="kw">if</span> grad's storage <span class="kw">type</span> is ``row_sparse``, ``lazy_update`` is True and weight's storage |
| <span class="kw">type</span> is the same as momentum's storage <span class="kw">type</span>, |
| only the row slices whose indices appear in grad.indices are updated (<span class="kw">for</span> both weight and momentum):: |
| |
| <span class="kw">for</span> row in gradient.indices: |
| v[row] = momentum[row] * v[row] - learning_rate * gradient[row] |
| weight[row] += v[row] |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L564</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sgd_mom_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sgd_mom_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sgd_mom_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sgd_mom_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Momentum update function <span class="kw">for</span> Stochastic Gradient Descent (SGD) optimizer. |
| |
| Momentum update has better convergence rates on neural networks. Mathematically it looks |
| like below: |
| |
| .. math:: |
| |
| v_1 = \alpha * \nabla J(W_0)\\ |
| v_t = \gamma v_{t-<span class="num">1</span>} - \alpha * \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} + v_t |
| |
| It updates the weights using:: |
| |
| v = momentum * v - learning_rate * gradient |
| weight += v |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| However, <span class="kw">if</span> grad's storage <span class="kw">type</span> is ``row_sparse``, ``lazy_update`` is True and weight's storage |
| <span class="kw">type</span> is the same as momentum's storage <span class="kw">type</span>, |
| only the row slices whose indices appear in grad.indices are updated (<span class="kw">for</span> both weight and momentum):: |
| |
| <span class="kw">for</span> row in gradient.indices: |
| v[row] = momentum[row] * v[row] - learning_rate * gradient[row] |
| weight[row] += v[row] |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L564</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sgd_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sgd_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Stochastic Gradient Descent (SGD) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| However, <span class="kw">if</span> gradient is of ``row_sparse`` storage <span class="kw">type</span> and ``lazy_update`` is True, |
| only the row slices whose indices appear in grad.indices are updated:: |
| |
| <span class="kw">for</span> row in gradient.indices: |
| weight[row] = weight[row] - learning_rate * (gradient[row] + wd * weight[row]) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L523</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sgd_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sgd_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> Stochastic Gradient Descent (SGD) optimizer. |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * (gradient + wd * weight) |
| |
| However, <span class="kw">if</span> gradient is of ``row_sparse`` storage <span class="kw">type</span> and ``lazy_update`` is True, |
| only the row slices whose indices appear in grad.indices are updated:: |
| |
| <span class="kw">for</span> row in gradient.indices: |
| weight[row] = weight[row] - learning_rate * (gradient[row] + wd * weight[row]) |
| |
| |
| |
| Defined in src/operator/optimizer_op.cc:L523</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#shape_array" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="shape_array(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="shape_array(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">shape_array</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@shape_array(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a <span class="num">1</span>D int64 array containing the shape of data. |
| |
| Example:: |
| |
| shape_array(`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], [<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>,<span class="num">8</span>] ]) = [<span class="num">2</span>,<span class="num">4</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L573</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#shape_array" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="shape_array(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="shape_array(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">shape_array</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@shape_array(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a <span class="num">1</span>D int64 array containing the shape of data. |
| |
| Example:: |
| |
| shape_array(`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], [<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>,<span class="num">8</span>] ]) = [<span class="num">2</span>,<span class="num">4</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L573</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#shuffle" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="shuffle(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="shuffle(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">shuffle</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@shuffle(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Randomly shuffle the elements. |
| |
| This shuffles the array along the first axis. |
| The order of the elements in each subarray does not change. |
| For example, <span class="kw">if</span> a <span class="num">2</span>D array is given, the order of the rows randomly changes, |
| but the order of the elements in each row does not change.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#shuffle" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="shuffle(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="shuffle(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">shuffle</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@shuffle(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Randomly shuffle the elements. |
| |
| This shuffles the array along the first axis. |
| The order of the elements in each subarray does not change. |
| For example, <span class="kw">if</span> a <span class="num">2</span>D array is given, the order of the rows randomly changes, |
| but the order of the elements in each row does not change.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sigmoid" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sigmoid(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sigmoid(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sigmoid</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sigmoid(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes sigmoid of x element-wise. |
| |
| .. math:: |
| y = <span class="num">1</span> / (<span class="num">1</span> + exp(-x)) |
| |
| The storage <span class="kw">type</span> of ``sigmoid`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L119</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sigmoid" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sigmoid(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sigmoid(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sigmoid</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sigmoid(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes sigmoid of x element-wise. |
| |
| .. math:: |
| y = <span class="num">1</span> / (<span class="num">1</span> + exp(-x)) |
| |
| The storage <span class="kw">type</span> of ``sigmoid`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L119</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sign" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sign(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sign(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sign</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sign(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise sign of the input. |
| |
| Example:: |
| |
| sign([-<span class="num">2</span>, <span class="num">0</span>, <span class="num">3</span>]) = [-<span class="num">1</span>, <span class="num">0</span>, <span class="num">1</span>] |
| |
| The storage <span class="kw">type</span> of ``sign`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - sign(default) = default |
| - sign(row_sparse) = row_sparse |
| - sign(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L758</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sign" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sign(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sign(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sign</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sign(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise sign of the input. |
| |
| Example:: |
| |
| sign([-<span class="num">2</span>, <span class="num">0</span>, <span class="num">3</span>]) = [-<span class="num">1</span>, <span class="num">0</span>, <span class="num">1</span>] |
| |
| The storage <span class="kw">type</span> of ``sign`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - sign(default) = default |
| - sign(row_sparse) = row_sparse |
| - sign(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L758</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#signsgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="signsgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="signsgd_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">signsgd_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@signsgd_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> SignSGD optimizer. |
| |
| .. math:: |
| |
| g_t = \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} - \eta_t \text{sign}(g_t) |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * sign(gradient) |
| |
| .. note:: |
| - sparse ndarray not supported <span class="kw">for</span> <span class="kw">this</span> optimizer yet. |
| |
| |
| Defined in src/operator/optimizer_op.cc:L62</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#signsgd_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="signsgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="signsgd_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">signsgd_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@signsgd_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Update function <span class="kw">for</span> SignSGD optimizer. |
| |
| .. math:: |
| |
| g_t = \nabla J(W_{t-<span class="num">1</span>})\\ |
| W_t = W_{t-<span class="num">1</span>} - \eta_t \text{sign}(g_t) |
| |
| It updates the weights using:: |
| |
| weight = weight - learning_rate * sign(gradient) |
| |
| .. note:: |
| - sparse ndarray not supported <span class="kw">for</span> <span class="kw">this</span> optimizer yet. |
| |
| |
| Defined in src/operator/optimizer_op.cc:L62</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#signum_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="signum_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="signum_update(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">signum_update</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@signum_update(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>SIGN momentUM (Signum) optimizer. |
| |
| .. math:: |
| |
| g_t = \nabla J(W_{t-<span class="num">1</span>})\\ |
| m_t = \beta m_{t-<span class="num">1</span>} + (<span class="num">1</span> - \beta) g_t\\ |
| W_t = W_{t-<span class="num">1</span>} - \eta_t \text{sign}(m_t) |
| |
| It updates the weights using:: |
| state = momentum * state + (<span class="num">1</span>-momentum) * gradient |
| weight = weight - learning_rate * sign(state) |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| .. note:: |
| - sparse ndarray not supported <span class="kw">for</span> <span class="kw">this</span> optimizer yet. |
| |
| |
| Defined in src/operator/optimizer_op.cc:L91</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#signum_update" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="signum_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="signum_update(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">signum_update</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@signum_update(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>SIGN momentUM (Signum) optimizer. |
| |
| .. math:: |
| |
| g_t = \nabla J(W_{t-<span class="num">1</span>})\\ |
| m_t = \beta m_{t-<span class="num">1</span>} + (<span class="num">1</span> - \beta) g_t\\ |
| W_t = W_{t-<span class="num">1</span>} - \eta_t \text{sign}(m_t) |
| |
| It updates the weights using:: |
| state = momentum * state + (<span class="num">1</span>-momentum) * gradient |
| weight = weight - learning_rate * sign(state) |
| |
| Where the parameter ``momentum`` is the decay rate of momentum estimates at each epoch. |
| |
| .. note:: |
| - sparse ndarray not supported <span class="kw">for</span> <span class="kw">this</span> optimizer yet. |
| |
| |
| Defined in src/operator/optimizer_op.cc:L91</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sin" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sin(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sin(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sin</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sin(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the element-wise sine of the input array. |
| |
| The input should be in radians (:math:`<span class="num">2</span>\pi` rad equals <span class="num">360</span> degrees). |
| |
| .. math:: |
| sin([<span class="num">0</span>, \pi/<span class="num">4</span>, \pi/<span class="num">2</span>]) = [<span class="num">0</span>, <span class="num">0.707</span>, <span class="num">1</span>] |
| |
| The storage <span class="kw">type</span> of ``sin`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - sin(default) = default |
| - sin(row_sparse) = row_sparse |
| - sin(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L47</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sin" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sin(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sin(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sin</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sin(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the element-wise sine of the input array. |
| |
| The input should be in radians (:math:`<span class="num">2</span>\pi` rad equals <span class="num">360</span> degrees). |
| |
| .. math:: |
| sin([<span class="num">0</span>, \pi/<span class="num">4</span>, \pi/<span class="num">2</span>]) = [<span class="num">0</span>, <span class="num">0.707</span>, <span class="num">1</span>] |
| |
| The storage <span class="kw">type</span> of ``sin`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - sin(default) = default |
| - sin(row_sparse) = row_sparse |
| - sin(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L47</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sinh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sinh(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sinh(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sinh</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sinh(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the hyperbolic sine of the input array, computed element-wise. |
| |
| .. math:: |
| sinh(x) = <span class="num">0.5</span>\times(exp(x) - exp(-x)) |
| |
| The storage <span class="kw">type</span> of ``sinh`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - sinh(default) = default |
| - sinh(row_sparse) = row_sparse |
| - sinh(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L371</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sinh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sinh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sinh(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sinh</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sinh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the hyperbolic sine of the input array, computed element-wise. |
| |
| .. math:: |
| sinh(x) = <span class="num">0.5</span>\times(exp(x) - exp(-x)) |
| |
| The storage <span class="kw">type</span> of ``sinh`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - sinh(default) = default |
| - sinh(row_sparse) = row_sparse |
| - sinh(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L371</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#size_array" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="size_array(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="size_array(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">size_array</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@size_array(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a <span class="num">1</span>D int64 array containing the size of data. |
| |
| Example:: |
| |
| size_array(`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], [<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>,<span class="num">8</span>] ]) = [<span class="num">8</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L624</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#size_array" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="size_array(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="size_array(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">size_array</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@size_array(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a <span class="num">1</span>D int64 array containing the size of data. |
| |
| Example:: |
| |
| size_array(`[ [<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>], [<span class="num">5</span>,<span class="num">6</span>,<span class="num">7</span>,<span class="num">8</span>] ]) = [<span class="num">8</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L624</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#slice" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="slice(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="slice(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">slice</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@slice(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Slices a region of the array. |
| .. note:: ``crop`` is deprecated. Use ``slice`` instead. |
| This function returns a sliced array between the indices given |
| by `begin` and `end` <span class="kw">with</span> the corresponding `step`. |
| For an input array of ``shape=(d_0, d_1, ..., d_n-<span class="num">1</span>)``, |
| slice operation <span class="kw">with</span> ``begin=(b_0, b_1...b_m-<span class="num">1</span>)``, |
| ``end=(e_0, e_1, ..., e_m-<span class="num">1</span>)``, and ``step=(s_0, s_1, ..., s_m-<span class="num">1</span>)``, |
| where m <= n, results in an array <span class="kw">with</span> the shape |
| ``(|e_0-b_0|/|s_0|, ..., |e_m-<span class="num">1</span>-b_m-<span class="num">1</span>|/|s_m-<span class="num">1</span>|, d_m, ..., d_n-<span class="num">1</span>)``. |
| The resulting array's *k*-th dimension contains elements |
| from the *k*-th dimension of the input array starting |
| from index ``b_k`` (inclusive) <span class="kw">with</span> step ``s_k`` |
| until reaching ``e_k`` (exclusive). |
| If the *k*-th elements are `<span class="std">None</span>` in the sequence of `begin`, `end`, |
| and `step`, the following rule will be used to set default values. |
| If `s_k` is `<span class="std">None</span>`, set `s_k=<span class="num">1</span>`. If `s_k > <span class="num">0</span>`, set `b_k=<span class="num">0</span>`, `e_k=d_k`; |
| <span class="kw">else</span>, set `b_k=d_k-<span class="num">1</span>`, `e_k=-<span class="num">1</span>`. |
| The storage <span class="kw">type</span> of ``slice`` output depends on storage types of inputs |
| - slice(csr) = csr |
| - otherwise, ``slice`` generates output <span class="kw">with</span> default storage |
| .. note:: When input data storage <span class="kw">type</span> is csr, it only supports |
| step=(), or step=(<span class="std">None</span>,), or step=(<span class="num">1</span>,) to generate a csr output. |
| For other step parameter values, it falls back to slicing |
| a dense tensor. |
| Example:: |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| slice(x, begin=(<span class="num">0</span>,<span class="num">1</span>), end=(<span class="num">2</span>,<span class="num">4</span>)) = `[ [ <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>] ] |
| slice(x, begin=(<span class="std">None</span>, <span class="num">0</span>), end=(<span class="std">None</span>, <span class="num">3</span>), step=(-<span class="num">1</span>, <span class="num">2</span>)) = `[ [<span class="num">9.</span>, <span class="num">11.</span>], |
| [<span class="num">5.</span>, <span class="num">7.</span>], |
| [<span class="num">1.</span>, <span class="num">3.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L481</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#slice" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="slice(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="slice(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">slice</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@slice(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Slices a region of the array. |
| .. note:: ``crop`` is deprecated. Use ``slice`` instead. |
| This function returns a sliced array between the indices given |
| by `begin` and `end` <span class="kw">with</span> the corresponding `step`. |
| For an input array of ``shape=(d_0, d_1, ..., d_n-<span class="num">1</span>)``, |
| slice operation <span class="kw">with</span> ``begin=(b_0, b_1...b_m-<span class="num">1</span>)``, |
| ``end=(e_0, e_1, ..., e_m-<span class="num">1</span>)``, and ``step=(s_0, s_1, ..., s_m-<span class="num">1</span>)``, |
| where m <= n, results in an array <span class="kw">with</span> the shape |
| ``(|e_0-b_0|/|s_0|, ..., |e_m-<span class="num">1</span>-b_m-<span class="num">1</span>|/|s_m-<span class="num">1</span>|, d_m, ..., d_n-<span class="num">1</span>)``. |
| The resulting array's *k*-th dimension contains elements |
| from the *k*-th dimension of the input array starting |
| from index ``b_k`` (inclusive) <span class="kw">with</span> step ``s_k`` |
| until reaching ``e_k`` (exclusive). |
| If the *k*-th elements are `<span class="std">None</span>` in the sequence of `begin`, `end`, |
| and `step`, the following rule will be used to set default values. |
| If `s_k` is `<span class="std">None</span>`, set `s_k=<span class="num">1</span>`. If `s_k > <span class="num">0</span>`, set `b_k=<span class="num">0</span>`, `e_k=d_k`; |
| <span class="kw">else</span>, set `b_k=d_k-<span class="num">1</span>`, `e_k=-<span class="num">1</span>`. |
| The storage <span class="kw">type</span> of ``slice`` output depends on storage types of inputs |
| - slice(csr) = csr |
| - otherwise, ``slice`` generates output <span class="kw">with</span> default storage |
| .. note:: When input data storage <span class="kw">type</span> is csr, it only supports |
| step=(), or step=(<span class="std">None</span>,), or step=(<span class="num">1</span>,) to generate a csr output. |
| For other step parameter values, it falls back to slicing |
| a dense tensor. |
| Example:: |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| slice(x, begin=(<span class="num">0</span>,<span class="num">1</span>), end=(<span class="num">2</span>,<span class="num">4</span>)) = `[ [ <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>] ] |
| slice(x, begin=(<span class="std">None</span>, <span class="num">0</span>), end=(<span class="std">None</span>, <span class="num">3</span>), step=(-<span class="num">1</span>, <span class="num">2</span>)) = `[ [<span class="num">9.</span>, <span class="num">11.</span>], |
| [<span class="num">5.</span>, <span class="num">7.</span>], |
| [<span class="num">1.</span>, <span class="num">3.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L481</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#slice_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="slice_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="slice_axis(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">slice_axis</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@slice_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Slices along a given axis. |
| Returns an array slice along a given `axis` starting from the `begin` index |
| to the `end` index. |
| Examples:: |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| slice_axis(x, axis=<span class="num">0</span>, begin=<span class="num">1</span>, end=<span class="num">3</span>) = `[ [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| slice_axis(x, axis=<span class="num">1</span>, begin=<span class="num">0</span>, end=<span class="num">2</span>) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>] ] |
| slice_axis(x, axis=<span class="num">1</span>, begin=-<span class="num">3</span>, end=-<span class="num">1</span>) = `[ [ <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">6.</span>, <span class="num">7.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L570</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#slice_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="slice_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="slice_axis(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">slice_axis</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@slice_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Slices along a given axis. |
| Returns an array slice along a given `axis` starting from the `begin` index |
| to the `end` index. |
| Examples:: |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| slice_axis(x, axis=<span class="num">0</span>, begin=<span class="num">1</span>, end=<span class="num">3</span>) = `[ [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| slice_axis(x, axis=<span class="num">1</span>, begin=<span class="num">0</span>, end=<span class="num">2</span>) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>] ] |
| slice_axis(x, axis=<span class="num">1</span>, begin=-<span class="num">3</span>, end=-<span class="num">1</span>) = `[ [ <span class="num">2.</span>, <span class="num">3.</span>], |
| [ <span class="num">6.</span>, <span class="num">7.</span>], |
| [ <span class="num">10.</span>, <span class="num">11.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L570</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#slice_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="slice_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="slice_like(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">slice_like</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@slice_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Slices a region of the array like the shape of another array. |
| This function is similar to ``slice``, however, the `begin` are always `<span class="num">0</span>`s |
| and `end` of specific axes are inferred from the second input `shape_like`. |
| Given the second `shape_like` input of ``shape=(d_0, d_1, ..., d_n-<span class="num">1</span>)``, |
| a ``slice_like`` operator <span class="kw">with</span> default empty `axes`, it performs the |
| following operation: |
| `` out = slice(input, begin=(<span class="num">0</span>, <span class="num">0</span>, ..., <span class="num">0</span>), end=(d_0, d_1, ..., d_n-<span class="num">1</span>))``. |
| When `axes` is not empty, it is used to speficy which axes are being sliced. |
| Given a <span class="num">4</span>-d input data, ``slice_like`` operator <span class="kw">with</span> ``axes=(<span class="num">0</span>, <span class="num">2</span>, -<span class="num">1</span>)`` |
| will perform the following operation: |
| `` out = slice(input, begin=(<span class="num">0</span>, <span class="num">0</span>, <span class="num">0</span>, <span class="num">0</span>), end=(d_0, <span class="std">None</span>, d_2, d_3))``. |
| Note that it is allowed to have first and second input <span class="kw">with</span> different dimensions, |
| however, you have to make sure the `axes` are specified and not exceeding the |
| dimension limits. |
| For example, given `input_1` <span class="kw">with</span> ``shape=(<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>)`` and `input_2` <span class="kw">with</span> |
| ``shape=(<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>)``, it is not allowed to use: |
| `` out = slice_like(a, b)`` because ndim of `input_1` is <span class="num">4</span>, and ndim of `input_2` |
| is <span class="num">3.</span> |
| The following is allowed in <span class="kw">this</span> situation: |
| `` out = slice_like(a, b, axes=(<span class="num">0</span>, <span class="num">2</span>))`` |
| Example:: |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| y = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| slice_like(x, y) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>] ] |
| slice_like(x, y, axes=(<span class="num">0</span>, <span class="num">1</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>] ] |
| slice_like(x, y, axes=(<span class="num">0</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>] |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>] ] |
| slice_like(x, y, axes=(-<span class="num">1</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>] |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L624</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#slice_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="slice_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="slice_like(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">slice_like</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@slice_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Slices a region of the array like the shape of another array. |
| This function is similar to ``slice``, however, the `begin` are always `<span class="num">0</span>`s |
| and `end` of specific axes are inferred from the second input `shape_like`. |
| Given the second `shape_like` input of ``shape=(d_0, d_1, ..., d_n-<span class="num">1</span>)``, |
| a ``slice_like`` operator <span class="kw">with</span> default empty `axes`, it performs the |
| following operation: |
| `` out = slice(input, begin=(<span class="num">0</span>, <span class="num">0</span>, ..., <span class="num">0</span>), end=(d_0, d_1, ..., d_n-<span class="num">1</span>))``. |
| When `axes` is not empty, it is used to speficy which axes are being sliced. |
| Given a <span class="num">4</span>-d input data, ``slice_like`` operator <span class="kw">with</span> ``axes=(<span class="num">0</span>, <span class="num">2</span>, -<span class="num">1</span>)`` |
| will perform the following operation: |
| `` out = slice(input, begin=(<span class="num">0</span>, <span class="num">0</span>, <span class="num">0</span>, <span class="num">0</span>), end=(d_0, <span class="std">None</span>, d_2, d_3))``. |
| Note that it is allowed to have first and second input <span class="kw">with</span> different dimensions, |
| however, you have to make sure the `axes` are specified and not exceeding the |
| dimension limits. |
| For example, given `input_1` <span class="kw">with</span> ``shape=(<span class="num">2</span>,<span class="num">3</span>,<span class="num">4</span>,<span class="num">5</span>)`` and `input_2` <span class="kw">with</span> |
| ``shape=(<span class="num">1</span>,<span class="num">2</span>,<span class="num">3</span>)``, it is not allowed to use: |
| `` out = slice_like(a, b)`` because ndim of `input_1` is <span class="num">4</span>, and ndim of `input_2` |
| is <span class="num">3.</span> |
| The following is allowed in <span class="kw">this</span> situation: |
| `` out = slice_like(a, b, axes=(<span class="num">0</span>, <span class="num">2</span>))`` |
| Example:: |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>], |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>, <span class="num">12.</span>] ] |
| y = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ] |
| slice_like(x, y) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>] ] |
| slice_like(x, y, axes=(<span class="num">0</span>, <span class="num">1</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>] ] |
| slice_like(x, y, axes=(<span class="num">0</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>, <span class="num">4.</span>] |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>, <span class="num">8.</span>] ] |
| slice_like(x, y, axes=(-<span class="num">1</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">3.</span>] |
| [ <span class="num">5.</span>, <span class="num">6.</span>, <span class="num">7.</span>] |
| [ <span class="num">9.</span>, <span class="num">10.</span>, <span class="num">11.</span>] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L624</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#smooth_l1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="smooth_l1(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="smooth_l1(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">smooth_l1</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@smooth_l1(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Calculate Smooth L1 Loss(lhs, scalar) by summing |
| |
| .. math:: |
| |
| f(x) = |
| \begin{cases} |
| (\sigma x)^<span class="num">2</span>/<span class="num">2</span>,& \text{<span class="kw">if</span> }x < <span class="num">1</span>/\sigma^<span class="num">2</span>\\ |
| |x|-<span class="num">0.5</span>/\sigma^<span class="num">2</span>,& \text{otherwise} |
| \end{cases} |
| |
| where :math:`x` is an element of the tensor *lhs* and :math:`\sigma` is the scalar. |
| |
| Example:: |
| |
| smooth_l1([<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>]) = [<span class="num">0.5</span>, <span class="num">1.5</span>, <span class="num">2.5</span>, <span class="num">3.5</span>] |
| smooth_l1([<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>], scalar=<span class="num">1</span>) = [<span class="num">0.5</span>, <span class="num">1.5</span>, <span class="num">2.5</span>, <span class="num">3.5</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_scalar_op_extended.cc:L108</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#smooth_l1" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="smooth_l1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="smooth_l1(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">smooth_l1</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@smooth_l1(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Calculate Smooth L1 Loss(lhs, scalar) by summing |
| |
| .. math:: |
| |
| f(x) = |
| \begin{cases} |
| (\sigma x)^<span class="num">2</span>/<span class="num">2</span>,& \text{<span class="kw">if</span> }x < <span class="num">1</span>/\sigma^<span class="num">2</span>\\ |
| |x|-<span class="num">0.5</span>/\sigma^<span class="num">2</span>,& \text{otherwise} |
| \end{cases} |
| |
| where :math:`x` is an element of the tensor *lhs* and :math:`\sigma` is the scalar. |
| |
| Example:: |
| |
| smooth_l1([<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>]) = [<span class="num">0.5</span>, <span class="num">1.5</span>, <span class="num">2.5</span>, <span class="num">3.5</span>] |
| smooth_l1([<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>], scalar=<span class="num">1</span>) = [<span class="num">0.5</span>, <span class="num">1.5</span>, <span class="num">2.5</span>, <span class="num">3.5</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_binary_scalar_op_extended.cc:L108</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#softmax" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="softmax(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="softmax(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">softmax</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@softmax(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies the softmax function. |
| |
| The resulting array contains elements in the range (<span class="num">0</span>,<span class="num">1</span>) and the elements along the given axis sum up to <span class="num">1.</span> |
| |
| .. math:: |
| softmax(\mathbf{z/t})_j = \frac{e^{z_j/t}}{\sum_{k=<span class="num">1</span>}^K e^{z_k/t}} |
| |
| <span class="kw">for</span> :math:`j = <span class="num">1</span>, ..., K` |
| |
| t is the temperature parameter in softmax function. By default, t equals <span class="num">1.0</span> |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] |
| |
| softmax(x,axis=<span class="num">0</span>) = `[ [ <span class="num">0.5</span> <span class="num">0.5</span> <span class="num">0.5</span>] |
| [ <span class="num">0.5</span> <span class="num">0.5</span> <span class="num">0.5</span>] ] |
| |
| softmax(x,axis=<span class="num">1</span>) = `[ [ <span class="num">0.33333334</span>, <span class="num">0.33333334</span>, <span class="num">0.33333334</span>], |
| [ <span class="num">0.33333334</span>, <span class="num">0.33333334</span>, <span class="num">0.33333334</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/softmax.cc:L135</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#softmax" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="softmax(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="softmax(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">softmax</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@softmax(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies the softmax function. |
| |
| The resulting array contains elements in the range (<span class="num">0</span>,<span class="num">1</span>) and the elements along the given axis sum up to <span class="num">1.</span> |
| |
| .. math:: |
| softmax(\mathbf{z/t})_j = \frac{e^{z_j/t}}{\sum_{k=<span class="num">1</span>}^K e^{z_k/t}} |
| |
| <span class="kw">for</span> :math:`j = <span class="num">1</span>, ..., K` |
| |
| t is the temperature parameter in softmax function. By default, t equals <span class="num">1.0</span> |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span> <span class="num">1.</span>] ] |
| |
| softmax(x,axis=<span class="num">0</span>) = `[ [ <span class="num">0.5</span> <span class="num">0.5</span> <span class="num">0.5</span>] |
| [ <span class="num">0.5</span> <span class="num">0.5</span> <span class="num">0.5</span>] ] |
| |
| softmax(x,axis=<span class="num">1</span>) = `[ [ <span class="num">0.33333334</span>, <span class="num">0.33333334</span>, <span class="num">0.33333334</span>], |
| [ <span class="num">0.33333334</span>, <span class="num">0.33333334</span>, <span class="num">0.33333334</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/softmax.cc:L135</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#softmax_cross_entropy" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="softmax_cross_entropy(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="softmax_cross_entropy(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">softmax_cross_entropy</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@softmax_cross_entropy(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Calculate cross entropy of softmax output and one-hot label. |
| |
| - This operator computes the cross entropy in two steps: |
| - Applies softmax function on the input array. |
| - Computes and returns the cross entropy loss between the softmax output and the labels. |
| |
| - The softmax function and cross entropy loss is given by: |
| |
| - Softmax <span class="std">Function</span>: |
| |
| .. math:: \text{softmax}(x)_i = \frac{exp(x_i)}{\sum_j exp(x_j)} |
| |
| - Cross Entropy <span class="std">Function</span>: |
| |
| .. math:: \text{CE(label, output)} = - \sum_i \text{label}_i \log(\text{output}_i) |
| |
| Example:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], |
| [<span class="num">11</span>, <span class="num">7</span>, <span class="num">5</span>] ] |
| |
| label = [<span class="num">2</span>, <span class="num">0</span>] |
| |
| softmax(x) = `[ [<span class="num">0.09003057</span>, <span class="num">0.24472848</span>, <span class="num">0.66524094</span>], |
| [<span class="num">0.97962922</span>, <span class="num">0.01794253</span>, <span class="num">0.00242826</span>] ] |
| |
| softmax_cross_entropy(data, label) = - log(<span class="num">0.66524084</span>) - log(<span class="num">0.97962922</span>) = <span class="num">0.4281871</span> |
| |
| |
| |
| Defined in src/operator/loss_binary_op.cc:L58</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#softmax_cross_entropy" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="softmax_cross_entropy(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="softmax_cross_entropy(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">softmax_cross_entropy</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@softmax_cross_entropy(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Calculate cross entropy of softmax output and one-hot label. |
| |
| - This operator computes the cross entropy in two steps: |
| - Applies softmax function on the input array. |
| - Computes and returns the cross entropy loss between the softmax output and the labels. |
| |
| - The softmax function and cross entropy loss is given by: |
| |
| - Softmax <span class="std">Function</span>: |
| |
| .. math:: \text{softmax}(x)_i = \frac{exp(x_i)}{\sum_j exp(x_j)} |
| |
| - Cross Entropy <span class="std">Function</span>: |
| |
| .. math:: \text{CE(label, output)} = - \sum_i \text{label}_i \log(\text{output}_i) |
| |
| Example:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>], |
| [<span class="num">11</span>, <span class="num">7</span>, <span class="num">5</span>] ] |
| |
| label = [<span class="num">2</span>, <span class="num">0</span>] |
| |
| softmax(x) = `[ [<span class="num">0.09003057</span>, <span class="num">0.24472848</span>, <span class="num">0.66524094</span>], |
| [<span class="num">0.97962922</span>, <span class="num">0.01794253</span>, <span class="num">0.00242826</span>] ] |
| |
| softmax_cross_entropy(data, label) = - log(<span class="num">0.66524084</span>) - log(<span class="num">0.97962922</span>) = <span class="num">0.4281871</span> |
| |
| |
| |
| Defined in src/operator/loss_binary_op.cc:L58</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#softmin" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="softmin(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="softmin(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">softmin</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@softmin(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies the softmin function. |
| |
| The resulting array contains elements in the range (<span class="num">0</span>,<span class="num">1</span>) and the elements along the given axis sum |
| up to <span class="num">1.</span> |
| |
| .. math:: |
| softmin(\mathbf{z/t})_j = \frac{e^{-z_j/t}}{\sum_{k=<span class="num">1</span>}^K e^{-z_k/t}} |
| |
| <span class="kw">for</span> :math:`j = <span class="num">1</span>, ..., K` |
| |
| t is the temperature parameter in softmax function. By default, t equals <span class="num">1.0</span> |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span>] |
| [ <span class="num">3.</span> <span class="num">2.</span> <span class="num">1.</span>] ] |
| |
| softmin(x,axis=<span class="num">0</span>) = `[ [ <span class="num">0.88079703</span>, <span class="num">0.5</span>, <span class="num">0.11920292</span>], |
| [ <span class="num">0.11920292</span>, <span class="num">0.5</span>, <span class="num">0.88079703</span>] ] |
| |
| softmin(x,axis=<span class="num">1</span>) = `[ [ <span class="num">0.66524094</span>, <span class="num">0.24472848</span>, <span class="num">0.09003057</span>], |
| [ <span class="num">0.09003057</span>, <span class="num">0.24472848</span>, <span class="num">0.66524094</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/softmin.cc:L56</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#softmin" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="softmin(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="softmin(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">softmin</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@softmin(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Applies the softmin function. |
| |
| The resulting array contains elements in the range (<span class="num">0</span>,<span class="num">1</span>) and the elements along the given axis sum |
| up to <span class="num">1.</span> |
| |
| .. math:: |
| softmin(\mathbf{z/t})_j = \frac{e^{-z_j/t}}{\sum_{k=<span class="num">1</span>}^K e^{-z_k/t}} |
| |
| <span class="kw">for</span> :math:`j = <span class="num">1</span>, ..., K` |
| |
| t is the temperature parameter in softmax function. By default, t equals <span class="num">1.0</span> |
| |
| Example:: |
| |
| x = `[ [ <span class="num">1.</span> <span class="num">2.</span> <span class="num">3.</span>] |
| [ <span class="num">3.</span> <span class="num">2.</span> <span class="num">1.</span>] ] |
| |
| softmin(x,axis=<span class="num">0</span>) = `[ [ <span class="num">0.88079703</span>, <span class="num">0.5</span>, <span class="num">0.11920292</span>], |
| [ <span class="num">0.11920292</span>, <span class="num">0.5</span>, <span class="num">0.88079703</span>] ] |
| |
| softmin(x,axis=<span class="num">1</span>) = `[ [ <span class="num">0.66524094</span>, <span class="num">0.24472848</span>, <span class="num">0.09003057</span>], |
| [ <span class="num">0.09003057</span>, <span class="num">0.24472848</span>, <span class="num">0.66524094</span>] ] |
| |
| |
| |
| Defined in src/operator/nn/softmin.cc:L56</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#softsign" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="softsign(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="softsign(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">softsign</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@softsign(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes softsign of x element-wise. |
| |
| .. math:: |
| y = x / (<span class="num">1</span> + abs(x)) |
| |
| The storage <span class="kw">type</span> of ``softsign`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L191</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#softsign" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="softsign(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="softsign(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">softsign</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@softsign(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes softsign of x element-wise. |
| |
| .. math:: |
| y = x / (<span class="num">1</span> + abs(x)) |
| |
| The storage <span class="kw">type</span> of ``softsign`` output is always dense |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L191</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sort" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sort(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sort(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sort</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sort(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a sorted copy of an input array along the given axis. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1</span>, <span class="num">4</span>], |
| [ <span class="num">3</span>, <span class="num">1</span>] ] |
| |
| <span class="cmt">// sorts along the last axis</span> |
| sort(x) = `[ [ <span class="num">1.</span>, <span class="num">4.</span>], |
| [ <span class="num">1.</span>, <span class="num">3.</span>] ] |
| |
| <span class="cmt">// flattens and then sorts</span> |
| sort(x, axis=<span class="std">None</span>) = [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">3.</span>, <span class="num">4.</span>] |
| |
| <span class="cmt">// sorts along the first axis</span> |
| sort(x, axis=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>] ] |
| |
| <span class="cmt">// in a descend order</span> |
| sort(x, is_ascend=<span class="num">0</span>) = `[ [ <span class="num">4.</span>, <span class="num">1.</span>], |
| [ <span class="num">3.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/ordering_op.cc:L132</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sort" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sort(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sort(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sort</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sort(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns a sorted copy of an input array along the given axis. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1</span>, <span class="num">4</span>], |
| [ <span class="num">3</span>, <span class="num">1</span>] ] |
| |
| <span class="cmt">// sorts along the last axis</span> |
| sort(x) = `[ [ <span class="num">1.</span>, <span class="num">4.</span>], |
| [ <span class="num">1.</span>, <span class="num">3.</span>] ] |
| |
| <span class="cmt">// flattens and then sorts</span> |
| sort(x, axis=<span class="std">None</span>) = [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">3.</span>, <span class="num">4.</span>] |
| |
| <span class="cmt">// sorts along the first axis</span> |
| sort(x, axis=<span class="num">0</span>) = `[ [ <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>] ] |
| |
| <span class="cmt">// in a descend order</span> |
| sort(x, is_ascend=<span class="num">0</span>) = `[ [ <span class="num">4.</span>, <span class="num">1.</span>], |
| [ <span class="num">3.</span>, <span class="num">1.</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/ordering_op.cc:L132</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#space_to_depth" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="space_to_depth(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="space_to_depth(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">space_to_depth</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@space_to_depth(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Rearranges(permutes) blocks of spatial data into depth. |
| Similar to ONNX SpaceToDepth operator: |
| https:<span class="cmt">//github.com/onnx/onnx/blob/master/docs/Operators.md#SpaceToDepth</span> |
| The output is a <span class="kw">new</span> tensor where the values from height and width dimension are |
| moved to the depth dimension. The reverse of <span class="kw">this</span> operation is ``depth_to_space``. |
| .. math:: |
| \begin{gather*} |
| x \prime = reshape(x, [N, C, H / block\_size, block\_size, W / block\_size, block\_size]) \\ |
| x \prime \prime = transpose(x \prime, [<span class="num">0</span>, <span class="num">3</span>, <span class="num">5</span>, <span class="num">1</span>, <span class="num">2</span>, <span class="num">4</span>]) \\ |
| y = reshape(x \prime \prime, [N, C * (block\_size ^ <span class="num">2</span>), H / block\_size, W / block\_size]) |
| \end{gather*} |
| where :math:`x` is an input tensor <span class="kw">with</span> default layout as :math:`[N, C, H, W]`: [batch, channels, height, width] |
| and :math:`y` is the output tensor of layout :math:`[N, C * (block\_size ^ <span class="num">2</span>), H / block\_size, W / block\_size]` |
| Example:: |
| x = `[ [`[ [<span class="num">0</span>, <span class="num">6</span>, <span class="num">1</span>, <span class="num">7</span>, <span class="num">2</span>, <span class="num">8</span>], |
| [<span class="num">12</span>, <span class="num">18</span>, <span class="num">13</span>, <span class="num">19</span>, <span class="num">14</span>, <span class="num">20</span>], |
| [<span class="num">3</span>, <span class="num">9</span>, <span class="num">4</span>, <span class="num">10</span>, <span class="num">5</span>, <span class="num">11</span>], |
| [<span class="num">15</span>, <span class="num">21</span>, <span class="num">16</span>, <span class="num">22</span>, <span class="num">17</span>, <span class="num">23</span>] ] ] ] |
| space_to_depth(x, <span class="num">2</span>) = `[ [`[ [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>] ], |
| `[ [<span class="num">6</span>, <span class="num">7</span>, <span class="num">8</span>], |
| [<span class="num">9</span>, <span class="num">10</span>, <span class="num">11</span>] ], |
| `[ [<span class="num">12</span>, <span class="num">13</span>, <span class="num">14</span>], |
| [<span class="num">15</span>, <span class="num">16</span>, <span class="num">17</span>] ], |
| `[ [<span class="num">18</span>, <span class="num">19</span>, <span class="num">20</span>], |
| [<span class="num">21</span>, <span class="num">22</span>, <span class="num">23</span>] ] ] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L1018</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#space_to_depth" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="space_to_depth(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="space_to_depth(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">space_to_depth</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@space_to_depth(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Rearranges(permutes) blocks of spatial data into depth. |
| Similar to ONNX SpaceToDepth operator: |
| https:<span class="cmt">//github.com/onnx/onnx/blob/master/docs/Operators.md#SpaceToDepth</span> |
| The output is a <span class="kw">new</span> tensor where the values from height and width dimension are |
| moved to the depth dimension. The reverse of <span class="kw">this</span> operation is ``depth_to_space``. |
| .. math:: |
| \begin{gather*} |
| x \prime = reshape(x, [N, C, H / block\_size, block\_size, W / block\_size, block\_size]) \\ |
| x \prime \prime = transpose(x \prime, [<span class="num">0</span>, <span class="num">3</span>, <span class="num">5</span>, <span class="num">1</span>, <span class="num">2</span>, <span class="num">4</span>]) \\ |
| y = reshape(x \prime \prime, [N, C * (block\_size ^ <span class="num">2</span>), H / block\_size, W / block\_size]) |
| \end{gather*} |
| where :math:`x` is an input tensor <span class="kw">with</span> default layout as :math:`[N, C, H, W]`: [batch, channels, height, width] |
| and :math:`y` is the output tensor of layout :math:`[N, C * (block\_size ^ <span class="num">2</span>), H / block\_size, W / block\_size]` |
| Example:: |
| x = `[ [`[ [<span class="num">0</span>, <span class="num">6</span>, <span class="num">1</span>, <span class="num">7</span>, <span class="num">2</span>, <span class="num">8</span>], |
| [<span class="num">12</span>, <span class="num">18</span>, <span class="num">13</span>, <span class="num">19</span>, <span class="num">14</span>, <span class="num">20</span>], |
| [<span class="num">3</span>, <span class="num">9</span>, <span class="num">4</span>, <span class="num">10</span>, <span class="num">5</span>, <span class="num">11</span>], |
| [<span class="num">15</span>, <span class="num">21</span>, <span class="num">16</span>, <span class="num">22</span>, <span class="num">17</span>, <span class="num">23</span>] ] ] ] |
| space_to_depth(x, <span class="num">2</span>) = `[ [`[ [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>, <span class="num">5</span>] ], |
| `[ [<span class="num">6</span>, <span class="num">7</span>, <span class="num">8</span>], |
| [<span class="num">9</span>, <span class="num">10</span>, <span class="num">11</span>] ], |
| `[ [<span class="num">12</span>, <span class="num">13</span>, <span class="num">14</span>], |
| [<span class="num">15</span>, <span class="num">16</span>, <span class="num">17</span>] ], |
| `[ [<span class="num">18</span>, <span class="num">19</span>, <span class="num">20</span>], |
| [<span class="num">21</span>, <span class="num">22</span>, <span class="num">23</span>] ] ] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L1018</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#split" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="split(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="split(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">split</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@split(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Splits an array along a particular axis into multiple sub-arrays. |
| |
| .. note:: ``SliceChannel`` is deprecated. Use ``split`` instead. |
| |
| **Note** that `num_outputs` should evenly divide the length of the axis |
| along which to split the array. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] |
| `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] |
| `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] ] |
| x.shape = (<span class="num">3</span>, <span class="num">2</span>, <span class="num">1</span>) |
| |
| y = split(x, axis=<span class="num">1</span>, num_outputs=<span class="num">2</span>) <span class="cmt">// a list of 2 arrays with shape (3, 1, 1)</span> |
| y = `[ `[ [ <span class="num">1.</span>] ] |
| `[ [ <span class="num">3.</span>] ] |
| `[ [ <span class="num">5.</span>] ] ] |
| |
| `[ `[ [ <span class="num">2.</span>] ] |
| `[ [ <span class="num">4.</span>] ] |
| `[ [ <span class="num">6.</span>] ] ] |
| |
| y[<span class="num">0</span>].shape = (<span class="num">3</span>, <span class="num">1</span>, <span class="num">1</span>) |
| |
| z = split(x, axis=<span class="num">0</span>, num_outputs=<span class="num">3</span>) <span class="cmt">// a list of 3 arrays with shape (1, 2, 1)</span> |
| z = `[ `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] ] |
| |
| `[ `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] ] |
| |
| `[ `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] ] |
| |
| z[<span class="num">0</span>].shape = (<span class="num">1</span>, <span class="num">2</span>, <span class="num">1</span>) |
| |
| `squeeze_axis=<span class="num">1</span>` removes the axis <span class="kw">with</span> length <span class="num">1</span> from the shapes of the output arrays. |
| **Note** that setting `squeeze_axis` to ``<span class="num">1</span>`` removes axis <span class="kw">with</span> length <span class="num">1</span> only |
| along the `axis` which it is split. |
| Also `squeeze_axis` can be set to <span class="kw">true</span> only <span class="kw">if</span> ``input.shape[axis] == num_outputs``. |
| |
| Example:: |
| |
| z = split(x, axis=<span class="num">0</span>, num_outputs=<span class="num">3</span>, squeeze_axis=<span class="num">1</span>) <span class="cmt">// a list of 3 arrays with shape (2, 1)</span> |
| z = `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] |
| |
| `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] |
| |
| `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] |
| z[<span class="num">0</span>].shape = (<span class="num">2</span> ,<span class="num">1</span> ) |
| |
| |
| |
| Defined in src/operator/slice_channel.cc:L106</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#split" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="split(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="split(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">split</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@split(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Splits an array along a particular axis into multiple sub-arrays. |
| |
| .. note:: ``SliceChannel`` is deprecated. Use ``split`` instead. |
| |
| **Note** that `num_outputs` should evenly divide the length of the axis |
| along which to split the array. |
| |
| Example:: |
| |
| x = `[ `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] |
| `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] |
| `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] ] |
| x.shape = (<span class="num">3</span>, <span class="num">2</span>, <span class="num">1</span>) |
| |
| y = split(x, axis=<span class="num">1</span>, num_outputs=<span class="num">2</span>) <span class="cmt">// a list of 2 arrays with shape (3, 1, 1)</span> |
| y = `[ `[ [ <span class="num">1.</span>] ] |
| `[ [ <span class="num">3.</span>] ] |
| `[ [ <span class="num">5.</span>] ] ] |
| |
| `[ `[ [ <span class="num">2.</span>] ] |
| `[ [ <span class="num">4.</span>] ] |
| `[ [ <span class="num">6.</span>] ] ] |
| |
| y[<span class="num">0</span>].shape = (<span class="num">3</span>, <span class="num">1</span>, <span class="num">1</span>) |
| |
| z = split(x, axis=<span class="num">0</span>, num_outputs=<span class="num">3</span>) <span class="cmt">// a list of 3 arrays with shape (1, 2, 1)</span> |
| z = `[ `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] ] |
| |
| `[ `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] ] |
| |
| `[ `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] ] |
| |
| z[<span class="num">0</span>].shape = (<span class="num">1</span>, <span class="num">2</span>, <span class="num">1</span>) |
| |
| `squeeze_axis=<span class="num">1</span>` removes the axis <span class="kw">with</span> length <span class="num">1</span> from the shapes of the output arrays. |
| **Note** that setting `squeeze_axis` to ``<span class="num">1</span>`` removes axis <span class="kw">with</span> length <span class="num">1</span> only |
| along the `axis` which it is split. |
| Also `squeeze_axis` can be set to <span class="kw">true</span> only <span class="kw">if</span> ``input.shape[axis] == num_outputs``. |
| |
| Example:: |
| |
| z = split(x, axis=<span class="num">0</span>, num_outputs=<span class="num">3</span>, squeeze_axis=<span class="num">1</span>) <span class="cmt">// a list of 3 arrays with shape (2, 1)</span> |
| z = `[ [ <span class="num">1.</span>] |
| [ <span class="num">2.</span>] ] |
| |
| `[ [ <span class="num">3.</span>] |
| [ <span class="num">4.</span>] ] |
| |
| `[ [ <span class="num">5.</span>] |
| [ <span class="num">6.</span>] ] |
| z[<span class="num">0</span>].shape = (<span class="num">2</span> ,<span class="num">1</span> ) |
| |
| |
| |
| Defined in src/operator/slice_channel.cc:L106</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sqrt" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sqrt(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sqrt(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sqrt</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sqrt(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise square-root value of the input. |
| |
| .. math:: |
| \textrm{sqrt}(x) = \sqrt{x} |
| |
| Example:: |
| |
| sqrt([<span class="num">4</span>, <span class="num">9</span>, <span class="num">16</span>]) = [<span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>] |
| |
| The storage <span class="kw">type</span> of ``sqrt`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - sqrt(default) = default |
| - sqrt(row_sparse) = row_sparse |
| - sqrt(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L170</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sqrt" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sqrt(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sqrt(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sqrt</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sqrt(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise square-root value of the input. |
| |
| .. math:: |
| \textrm{sqrt}(x) = \sqrt{x} |
| |
| Example:: |
| |
| sqrt([<span class="num">4</span>, <span class="num">9</span>, <span class="num">16</span>]) = [<span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>] |
| |
| The storage <span class="kw">type</span> of ``sqrt`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - sqrt(default) = default |
| - sqrt(row_sparse) = row_sparse |
| - sqrt(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L170</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#square" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="square(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="square(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">square</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@square(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise squared value of the input. |
| |
| .. math:: |
| square(x) = x^<span class="num">2</span> |
| |
| Example:: |
| |
| square([<span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>]) = [<span class="num">4</span>, <span class="num">9</span>, <span class="num">16</span>] |
| |
| The storage <span class="kw">type</span> of ``square`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - square(default) = default |
| - square(row_sparse) = row_sparse |
| - square(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L119</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#square" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="square(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="square(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">square</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@square(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns element-wise squared value of the input. |
| |
| .. math:: |
| square(x) = x^<span class="num">2</span> |
| |
| Example:: |
| |
| square([<span class="num">2</span>, <span class="num">3</span>, <span class="num">4</span>]) = [<span class="num">4</span>, <span class="num">9</span>, <span class="num">16</span>] |
| |
| The storage <span class="kw">type</span> of ``square`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - square(default) = default |
| - square(row_sparse) = row_sparse |
| - square(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_pow.cc:L119</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#squeeze" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="squeeze(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="squeeze(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">squeeze</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@squeeze(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Remove single-dimensional entries from the shape of an array. |
| Same behavior of defining the output tensor shape as numpy.squeeze <span class="kw">for</span> the most of cases. |
| See the following note <span class="kw">for</span> exception. |
| Examples:: |
| data = `[ `[ [<span class="num">0</span>], [<span class="num">1</span>], [<span class="num">2</span>] ] ] |
| squeeze(data) = [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>] |
| squeeze(data, axis=<span class="num">0</span>) = `[ [<span class="num">0</span>], [<span class="num">1</span>], [<span class="num">2</span>] ] |
| squeeze(data, axis=<span class="num">2</span>) = `[ [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>] ] |
| squeeze(data, axis=(<span class="num">0</span>, <span class="num">2</span>)) = [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>] |
| .. Note:: |
| The output of <span class="kw">this</span> operator will keep at least one dimension not removed. For example, |
| squeeze(`[ `[ [<span class="num">4</span>] ] ]) = [<span class="num">4</span>], <span class="kw">while</span> in numpy.squeeze, the output will become a scalar.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#squeeze" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="squeeze(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="squeeze(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">squeeze</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@squeeze(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Remove single-dimensional entries from the shape of an array. |
| Same behavior of defining the output tensor shape as numpy.squeeze <span class="kw">for</span> the most of cases. |
| See the following note <span class="kw">for</span> exception. |
| Examples:: |
| data = `[ `[ [<span class="num">0</span>], [<span class="num">1</span>], [<span class="num">2</span>] ] ] |
| squeeze(data) = [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>] |
| squeeze(data, axis=<span class="num">0</span>) = `[ [<span class="num">0</span>], [<span class="num">1</span>], [<span class="num">2</span>] ] |
| squeeze(data, axis=<span class="num">2</span>) = `[ [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>] ] |
| squeeze(data, axis=(<span class="num">0</span>, <span class="num">2</span>)) = [<span class="num">0</span>, <span class="num">1</span>, <span class="num">2</span>] |
| .. Note:: |
| The output of <span class="kw">this</span> operator will keep at least one dimension not removed. For example, |
| squeeze(`[ `[ [<span class="num">4</span>] ] ]) = [<span class="num">4</span>], <span class="kw">while</span> in numpy.squeeze, the output will become a scalar.</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#stack" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="stack(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="stack(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">stack</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@stack(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Join a sequence of arrays along a <span class="kw">new</span> axis. |
| The axis parameter specifies the index of the <span class="kw">new</span> axis in the dimensions of the |
| result. For example, <span class="kw">if</span> axis=<span class="num">0</span> it will be the first dimension and <span class="kw">if</span> axis=-<span class="num">1</span> it |
| will be the last dimension. |
| Examples:: |
| x = [<span class="num">1</span>, <span class="num">2</span>] |
| y = [<span class="num">3</span>, <span class="num">4</span>] |
| stack(x, y) = `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ] |
| stack(x, y, axis=<span class="num">1</span>) = `[ [<span class="num">1</span>, <span class="num">3</span>], |
| [<span class="num">2</span>, <span class="num">4</span>] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#stack" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="stack(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="stack(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">stack</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@stack(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Join a sequence of arrays along a <span class="kw">new</span> axis. |
| The axis parameter specifies the index of the <span class="kw">new</span> axis in the dimensions of the |
| result. For example, <span class="kw">if</span> axis=<span class="num">0</span> it will be the first dimension and <span class="kw">if</span> axis=-<span class="num">1</span> it |
| will be the last dimension. |
| Examples:: |
| x = [<span class="num">1</span>, <span class="num">2</span>] |
| y = [<span class="num">3</span>, <span class="num">4</span>] |
| stack(x, y) = `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ] |
| stack(x, y, axis=<span class="num">1</span>) = `[ [<span class="num">1</span>, <span class="num">3</span>], |
| [<span class="num">2</span>, <span class="num">4</span>] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#stop_gradient" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="stop_gradient(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="stop_gradient(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">stop_gradient</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@stop_gradient(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Stops gradient computation. |
| |
| Stops the accumulated gradient of the inputs from flowing through <span class="kw">this</span> operator |
| in the backward direction. In other words, <span class="kw">this</span> operator prevents the contribution |
| of its inputs to be taken into account <span class="kw">for</span> computing gradients. |
| |
| Example:: |
| |
| v1 = [<span class="num">1</span>, <span class="num">2</span>] |
| v2 = [<span class="num">0</span>, <span class="num">1</span>] |
| a = Variable(<span class="lit">'a'</span>) |
| b = Variable(<span class="lit">'b'</span>) |
| b_stop_grad = stop_gradient(<span class="num">3</span> * b) |
| loss = MakeLoss(b_stop_grad + a) |
| |
| executor = loss.simple_bind(ctx=cpu(), a=(<span class="num">1</span>,<span class="num">2</span>), b=(<span class="num">1</span>,<span class="num">2</span>)) |
| executor.forward(is_train=True, a=v1, b=v2) |
| executor.outputs |
| [ <span class="num">1.</span> <span class="num">5.</span>] |
| |
| executor.backward() |
| executor.grad_arrays |
| [ <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L325</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#stop_gradient" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="stop_gradient(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="stop_gradient(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">stop_gradient</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@stop_gradient(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Stops gradient computation. |
| |
| Stops the accumulated gradient of the inputs from flowing through <span class="kw">this</span> operator |
| in the backward direction. In other words, <span class="kw">this</span> operator prevents the contribution |
| of its inputs to be taken into account <span class="kw">for</span> computing gradients. |
| |
| Example:: |
| |
| v1 = [<span class="num">1</span>, <span class="num">2</span>] |
| v2 = [<span class="num">0</span>, <span class="num">1</span>] |
| a = Variable(<span class="lit">'a'</span>) |
| b = Variable(<span class="lit">'b'</span>) |
| b_stop_grad = stop_gradient(<span class="num">3</span> * b) |
| loss = MakeLoss(b_stop_grad + a) |
| |
| executor = loss.simple_bind(ctx=cpu(), a=(<span class="num">1</span>,<span class="num">2</span>), b=(<span class="num">1</span>,<span class="num">2</span>)) |
| executor.forward(is_train=True, a=v1, b=v2) |
| executor.outputs |
| [ <span class="num">1.</span> <span class="num">5.</span>] |
| |
| executor.backward() |
| executor.grad_arrays |
| [ <span class="num">0.</span> <span class="num">0.</span>] |
| [ <span class="num">1.</span> <span class="num">1.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L325</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sum(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sum(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sum</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sum(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the sum of array elements over given axes. |
| |
| .. Note:: |
| |
| `sum` and `sum_axis` are equivalent. |
| For ndarray of csr storage <span class="kw">type</span> summation along axis <span class="num">0</span> and axis <span class="num">1</span> is supported. |
| Setting keepdims or exclude to True will cause a fallback to dense operator. |
| |
| Example:: |
| |
| data = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">2</span>, <span class="num">3</span>], [<span class="num">1</span>, <span class="num">3</span>] ], |
| `[ [<span class="num">1</span>, <span class="num">4</span>], [<span class="num">4</span>, <span class="num">3</span>], [<span class="num">5</span>, <span class="num">2</span>] ], |
| `[ [<span class="num">7</span>, <span class="num">1</span>], [<span class="num">7</span>, <span class="num">2</span>], [<span class="num">7</span>, <span class="num">3</span>] ] ] |
| |
| sum(data, axis=<span class="num">1</span>) |
| `[ [ <span class="num">4.</span> <span class="num">8.</span>] |
| [ <span class="num">10.</span> <span class="num">9.</span>] |
| [ <span class="num">21.</span> <span class="num">6.</span>] ] |
| |
| sum(data, axis=[<span class="num">1</span>,<span class="num">2</span>]) |
| [ <span class="num">12.</span> <span class="num">19.</span> <span class="num">27.</span>] |
| |
| data = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">0</span>], |
| [<span class="num">3</span>, <span class="num">0</span>, <span class="num">1</span>], |
| [<span class="num">4</span>, <span class="num">1</span>, <span class="num">0</span>] ] |
| |
| csr = cast_storage(data, <span class="lit">'csr'</span>) |
| |
| sum(csr, axis=<span class="num">0</span>) |
| [ <span class="num">8.</span> <span class="num">3.</span> <span class="num">1.</span>] |
| |
| sum(csr, axis=<span class="num">1</span>) |
| [ <span class="num">3.</span> <span class="num">4.</span> <span class="num">5.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_sum_value.cc:L66</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sum" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sum(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sum</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sum(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the sum of array elements over given axes. |
| |
| .. Note:: |
| |
| `sum` and `sum_axis` are equivalent. |
| For ndarray of csr storage <span class="kw">type</span> summation along axis <span class="num">0</span> and axis <span class="num">1</span> is supported. |
| Setting keepdims or exclude to True will cause a fallback to dense operator. |
| |
| Example:: |
| |
| data = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">2</span>, <span class="num">3</span>], [<span class="num">1</span>, <span class="num">3</span>] ], |
| `[ [<span class="num">1</span>, <span class="num">4</span>], [<span class="num">4</span>, <span class="num">3</span>], [<span class="num">5</span>, <span class="num">2</span>] ], |
| `[ [<span class="num">7</span>, <span class="num">1</span>], [<span class="num">7</span>, <span class="num">2</span>], [<span class="num">7</span>, <span class="num">3</span>] ] ] |
| |
| sum(data, axis=<span class="num">1</span>) |
| `[ [ <span class="num">4.</span> <span class="num">8.</span>] |
| [ <span class="num">10.</span> <span class="num">9.</span>] |
| [ <span class="num">21.</span> <span class="num">6.</span>] ] |
| |
| sum(data, axis=[<span class="num">1</span>,<span class="num">2</span>]) |
| [ <span class="num">12.</span> <span class="num">19.</span> <span class="num">27.</span>] |
| |
| data = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">0</span>], |
| [<span class="num">3</span>, <span class="num">0</span>, <span class="num">1</span>], |
| [<span class="num">4</span>, <span class="num">1</span>, <span class="num">0</span>] ] |
| |
| csr = cast_storage(data, <span class="lit">'csr'</span>) |
| |
| sum(csr, axis=<span class="num">0</span>) |
| [ <span class="num">8.</span> <span class="num">3.</span> <span class="num">1.</span>] |
| |
| sum(csr, axis=<span class="num">1</span>) |
| [ <span class="num">3.</span> <span class="num">4.</span> <span class="num">5.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_sum_value.cc:L66</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sum_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sum_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sum_axis(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sum_axis</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sum_axis(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the sum of array elements over given axes. |
| |
| .. Note:: |
| |
| `sum` and `sum_axis` are equivalent. |
| For ndarray of csr storage <span class="kw">type</span> summation along axis <span class="num">0</span> and axis <span class="num">1</span> is supported. |
| Setting keepdims or exclude to True will cause a fallback to dense operator. |
| |
| Example:: |
| |
| data = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">2</span>, <span class="num">3</span>], [<span class="num">1</span>, <span class="num">3</span>] ], |
| `[ [<span class="num">1</span>, <span class="num">4</span>], [<span class="num">4</span>, <span class="num">3</span>], [<span class="num">5</span>, <span class="num">2</span>] ], |
| `[ [<span class="num">7</span>, <span class="num">1</span>], [<span class="num">7</span>, <span class="num">2</span>], [<span class="num">7</span>, <span class="num">3</span>] ] ] |
| |
| sum(data, axis=<span class="num">1</span>) |
| `[ [ <span class="num">4.</span> <span class="num">8.</span>] |
| [ <span class="num">10.</span> <span class="num">9.</span>] |
| [ <span class="num">21.</span> <span class="num">6.</span>] ] |
| |
| sum(data, axis=[<span class="num">1</span>,<span class="num">2</span>]) |
| [ <span class="num">12.</span> <span class="num">19.</span> <span class="num">27.</span>] |
| |
| data = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">0</span>], |
| [<span class="num">3</span>, <span class="num">0</span>, <span class="num">1</span>], |
| [<span class="num">4</span>, <span class="num">1</span>, <span class="num">0</span>] ] |
| |
| csr = cast_storage(data, <span class="lit">'csr'</span>) |
| |
| sum(csr, axis=<span class="num">0</span>) |
| [ <span class="num">8.</span> <span class="num">3.</span> <span class="num">1.</span>] |
| |
| sum(csr, axis=<span class="num">1</span>) |
| [ <span class="num">3.</span> <span class="num">4.</span> <span class="num">5.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_sum_value.cc:L66</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#sum_axis" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="sum_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="sum_axis(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">sum_axis</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@sum_axis(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the sum of array elements over given axes. |
| |
| .. Note:: |
| |
| `sum` and `sum_axis` are equivalent. |
| For ndarray of csr storage <span class="kw">type</span> summation along axis <span class="num">0</span> and axis <span class="num">1</span> is supported. |
| Setting keepdims or exclude to True will cause a fallback to dense operator. |
| |
| Example:: |
| |
| data = `[ `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">2</span>, <span class="num">3</span>], [<span class="num">1</span>, <span class="num">3</span>] ], |
| `[ [<span class="num">1</span>, <span class="num">4</span>], [<span class="num">4</span>, <span class="num">3</span>], [<span class="num">5</span>, <span class="num">2</span>] ], |
| `[ [<span class="num">7</span>, <span class="num">1</span>], [<span class="num">7</span>, <span class="num">2</span>], [<span class="num">7</span>, <span class="num">3</span>] ] ] |
| |
| sum(data, axis=<span class="num">1</span>) |
| `[ [ <span class="num">4.</span> <span class="num">8.</span>] |
| [ <span class="num">10.</span> <span class="num">9.</span>] |
| [ <span class="num">21.</span> <span class="num">6.</span>] ] |
| |
| sum(data, axis=[<span class="num">1</span>,<span class="num">2</span>]) |
| [ <span class="num">12.</span> <span class="num">19.</span> <span class="num">27.</span>] |
| |
| data = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">0</span>], |
| [<span class="num">3</span>, <span class="num">0</span>, <span class="num">1</span>], |
| [<span class="num">4</span>, <span class="num">1</span>, <span class="num">0</span>] ] |
| |
| csr = cast_storage(data, <span class="lit">'csr'</span>) |
| |
| sum(csr, axis=<span class="num">0</span>) |
| [ <span class="num">8.</span> <span class="num">3.</span> <span class="num">1.</span>] |
| |
| sum(csr, axis=<span class="num">1</span>) |
| [ <span class="num">3.</span> <span class="num">4.</span> <span class="num">5.</span>] |
| |
| |
| |
| Defined in src/operator/tensor/broadcast_reduce_sum_value.cc:L66</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#swapaxes" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="swapaxes(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="swapaxes(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">swapaxes</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@swapaxes(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Interchanges two axes of an array. |
| |
| Examples:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ]) |
| swapaxes(x, <span class="num">0</span>, <span class="num">1</span>) = `[ [ <span class="num">1</span>], |
| [ <span class="num">2</span>], |
| [ <span class="num">3</span>] ] |
| |
| x = `[ `[ [ <span class="num">0</span>, <span class="num">1</span>], |
| [ <span class="num">2</span>, <span class="num">3</span>] ], |
| `[ [ <span class="num">4</span>, <span class="num">5</span>], |
| [ <span class="num">6</span>, <span class="num">7</span>] ] ] <span class="cmt">// (2,2,2) array</span> |
| |
| swapaxes(x, <span class="num">0</span>, <span class="num">2</span>) = `[ `[ [ <span class="num">0</span>, <span class="num">4</span>], |
| [ <span class="num">2</span>, <span class="num">6</span>] ], |
| `[ [ <span class="num">1</span>, <span class="num">5</span>], |
| [ <span class="num">3</span>, <span class="num">7</span>] ] ] |
| |
| |
| Defined in src/operator/swapaxis.cc:L69</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#swapaxes" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="swapaxes(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="swapaxes(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">swapaxes</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@swapaxes(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Interchanges two axes of an array. |
| |
| Examples:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>, <span class="num">3</span>] ]) |
| swapaxes(x, <span class="num">0</span>, <span class="num">1</span>) = `[ [ <span class="num">1</span>], |
| [ <span class="num">2</span>], |
| [ <span class="num">3</span>] ] |
| |
| x = `[ `[ [ <span class="num">0</span>, <span class="num">1</span>], |
| [ <span class="num">2</span>, <span class="num">3</span>] ], |
| `[ [ <span class="num">4</span>, <span class="num">5</span>], |
| [ <span class="num">6</span>, <span class="num">7</span>] ] ] <span class="cmt">// (2,2,2) array</span> |
| |
| swapaxes(x, <span class="num">0</span>, <span class="num">2</span>) = `[ `[ [ <span class="num">0</span>, <span class="num">4</span>], |
| [ <span class="num">2</span>, <span class="num">6</span>] ], |
| `[ [ <span class="num">1</span>, <span class="num">5</span>], |
| [ <span class="num">3</span>, <span class="num">7</span>] ] ] |
| |
| |
| Defined in src/operator/swapaxis.cc:L69</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#take" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="take(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="take(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">take</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@take(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Takes elements from an input array along the given axis. |
| |
| This function slices the input array along a particular axis <span class="kw">with</span> the provided indices. |
| |
| Given data tensor of rank r >= <span class="num">1</span>, and indices tensor of rank q, gather entries of the axis |
| dimension of data (by default outer-most one as axis=<span class="num">0</span>) indexed by indices, and concatenates them |
| in an output tensor of rank q + (r - <span class="num">1</span>). |
| |
| Examples:: |
| |
| x = [<span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span>] |
| |
| <span class="cmt">// Trivial case, take the second element along the first axis.</span> |
| |
| take(x, [<span class="num">1</span>]) = [ <span class="num">5.</span> ] |
| |
| <span class="cmt">// The other trivial case, axis=-1, take the third element along the first axis</span> |
| |
| take(x, [<span class="num">3</span>], axis=-<span class="num">1</span>, mode=<span class="lit">'clip'</span>) = [ <span class="num">6.</span> ] |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] |
| |
| <span class="cmt">// In this case we will get rows 0 and 1, then 1 and 2. Along axis 0</span> |
| |
| take(x, `[ [<span class="num">0</span>,<span class="num">1</span>],[<span class="num">1</span>,<span class="num">2</span>] ]) = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>] ], |
| |
| `[ [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] ] |
| |
| <span class="cmt">// In this case we will get rows 0 and 1, then 1 and 2 (calculated by wrapping around).</span> |
| <span class="cmt">// Along axis 1</span> |
| |
| take(x, `[ [<span class="num">0</span>, <span class="num">3</span>], [-<span class="num">1</span>, -<span class="num">2</span>] ], axis=<span class="num">1</span>, mode=<span class="lit">'wrap'</span>) = `[ `[ [ <span class="num">1.</span> <span class="num">2.</span>] |
| [ <span class="num">2.</span> <span class="num">1.</span>] ] |
| |
| `[ [ <span class="num">3.</span> <span class="num">4.</span>] |
| [ <span class="num">4.</span> <span class="num">3.</span>] ] |
| |
| `[ [ <span class="num">5.</span> <span class="num">6.</span>] |
| [ <span class="num">6.</span> <span class="num">5.</span>] ] ] |
| |
| The storage <span class="kw">type</span> of ``take`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - take(default, default) = default |
| - take(csr, default, axis=<span class="num">0</span>) = csr |
| |
| |
| |
| Defined in src/operator/tensor/indexing_op.cc:L776</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#take" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="take(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="take(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">take</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@take(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Takes elements from an input array along the given axis. |
| |
| This function slices the input array along a particular axis <span class="kw">with</span> the provided indices. |
| |
| Given data tensor of rank r >= <span class="num">1</span>, and indices tensor of rank q, gather entries of the axis |
| dimension of data (by default outer-most one as axis=<span class="num">0</span>) indexed by indices, and concatenates them |
| in an output tensor of rank q + (r - <span class="num">1</span>). |
| |
| Examples:: |
| |
| x = [<span class="num">4.</span> <span class="num">5.</span> <span class="num">6.</span>] |
| |
| <span class="cmt">// Trivial case, take the second element along the first axis.</span> |
| |
| take(x, [<span class="num">1</span>]) = [ <span class="num">5.</span> ] |
| |
| <span class="cmt">// The other trivial case, axis=-1, take the third element along the first axis</span> |
| |
| take(x, [<span class="num">3</span>], axis=-<span class="num">1</span>, mode=<span class="lit">'clip'</span>) = [ <span class="num">6.</span> ] |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] |
| |
| <span class="cmt">// In this case we will get rows 0 and 1, then 1 and 2. Along axis 0</span> |
| |
| take(x, `[ [<span class="num">0</span>,<span class="num">1</span>],[<span class="num">1</span>,<span class="num">2</span>] ]) = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>] ], |
| |
| `[ [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ] ] |
| |
| <span class="cmt">// In this case we will get rows 0 and 1, then 1 and 2 (calculated by wrapping around).</span> |
| <span class="cmt">// Along axis 1</span> |
| |
| take(x, `[ [<span class="num">0</span>, <span class="num">3</span>], [-<span class="num">1</span>, -<span class="num">2</span>] ], axis=<span class="num">1</span>, mode=<span class="lit">'wrap'</span>) = `[ `[ [ <span class="num">1.</span> <span class="num">2.</span>] |
| [ <span class="num">2.</span> <span class="num">1.</span>] ] |
| |
| `[ [ <span class="num">3.</span> <span class="num">4.</span>] |
| [ <span class="num">4.</span> <span class="num">3.</span>] ] |
| |
| `[ [ <span class="num">5.</span> <span class="num">6.</span>] |
| [ <span class="num">6.</span> <span class="num">5.</span>] ] ] |
| |
| The storage <span class="kw">type</span> of ``take`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - take(default, default) = default |
| - take(csr, default, axis=<span class="num">0</span>) = csr |
| |
| |
| |
| Defined in src/operator/tensor/indexing_op.cc:L776</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#tan" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="tan(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="tan(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">tan</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@tan(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the element-wise tangent of the input array. |
| |
| The input should be in radians (:math:`<span class="num">2</span>\pi` rad equals <span class="num">360</span> degrees). |
| |
| .. math:: |
| tan([<span class="num">0</span>, \pi/<span class="num">4</span>, \pi/<span class="num">2</span>]) = [<span class="num">0</span>, <span class="num">1</span>, -inf] |
| |
| The storage <span class="kw">type</span> of ``tan`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - tan(default) = default |
| - tan(row_sparse) = row_sparse |
| - tan(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L140</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#tan" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="tan(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="tan(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">tan</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@tan(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Computes the element-wise tangent of the input array. |
| |
| The input should be in radians (:math:`<span class="num">2</span>\pi` rad equals <span class="num">360</span> degrees). |
| |
| .. math:: |
| tan([<span class="num">0</span>, \pi/<span class="num">4</span>, \pi/<span class="num">2</span>]) = [<span class="num">0</span>, <span class="num">1</span>, -inf] |
| |
| The storage <span class="kw">type</span> of ``tan`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - tan(default) = default |
| - tan(row_sparse) = row_sparse |
| - tan(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L140</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#tanh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="tanh(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="tanh(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">tanh</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@tanh(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the hyperbolic tangent of the input array, computed element-wise. |
| |
| .. math:: |
| tanh(x) = sinh(x) / cosh(x) |
| |
| The storage <span class="kw">type</span> of ``tanh`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - tanh(default) = default |
| - tanh(row_sparse) = row_sparse |
| - tanh(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L451</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#tanh" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="tanh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="tanh(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">tanh</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@tanh(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the hyperbolic tangent of the input array, computed element-wise. |
| |
| .. math:: |
| tanh(x) = sinh(x) / cosh(x) |
| |
| The storage <span class="kw">type</span> of ``tanh`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - tanh(default) = default |
| - tanh(row_sparse) = row_sparse |
| - tanh(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_trig.cc:L451</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#tile" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="tile(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="tile(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">tile</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@tile(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Repeats the whole array multiple times. |
| If ``reps`` has length *d*, and input array has dimension of *n*. There are |
| three cases: |
| - **n=d**. Repeat *i*-th dimension of the input by ``reps[i]`` times:: |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ] |
| tile(x, reps=(<span class="num">2</span>,<span class="num">3</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] |
| - **n>d**. ``reps`` is promoted to length *n* by pre-pending <span class="num">1</span>'s to it. Thus <span class="kw">for</span> |
| an input shape ``(<span class="num">2</span>,<span class="num">3</span>)``, ``repos=(<span class="num">2</span>,)`` is treated as ``(<span class="num">1</span>,<span class="num">2</span>)``:: |
| tile(x, reps=(<span class="num">2</span>,)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] |
| - **n<d**. The input is promoted to be d-dimensional by prepending <span class="kw">new</span> axes. So a |
| shape ``(<span class="num">2</span>,<span class="num">2</span>)`` array is promoted to ``(<span class="num">1</span>,<span class="num">2</span>,<span class="num">2</span>)`` <span class="kw">for</span> <span class="num">3</span>-D replication:: |
| tile(x, reps=(<span class="num">2</span>,<span class="num">2</span>,<span class="num">3</span>)) = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ], |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L795</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#tile" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="tile(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="tile(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">tile</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@tile(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Repeats the whole array multiple times. |
| If ``reps`` has length *d*, and input array has dimension of *n*. There are |
| three cases: |
| - **n=d**. Repeat *i*-th dimension of the input by ``reps[i]`` times:: |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>], |
| [<span class="num">3</span>, <span class="num">4</span>] ] |
| tile(x, reps=(<span class="num">2</span>,<span class="num">3</span>)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] |
| - **n>d**. ``reps`` is promoted to length *n* by pre-pending <span class="num">1</span>'s to it. Thus <span class="kw">for</span> |
| an input shape ``(<span class="num">2</span>,<span class="num">3</span>)``, ``repos=(<span class="num">2</span>,)`` is treated as ``(<span class="num">1</span>,<span class="num">2</span>)``:: |
| tile(x, reps=(<span class="num">2</span>,)) = `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] |
| - **n<d**. The input is promoted to be d-dimensional by prepending <span class="kw">new</span> axes. So a |
| shape ``(<span class="num">2</span>,<span class="num">2</span>)`` array is promoted to ``(<span class="num">1</span>,<span class="num">2</span>,<span class="num">2</span>)`` <span class="kw">for</span> <span class="num">3</span>-D replication:: |
| tile(x, reps=(<span class="num">2</span>,<span class="num">2</span>,<span class="num">3</span>)) = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ], |
| `[ [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>, <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>, <span class="num">3.</span>, <span class="num">4.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L795</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#topk" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="topk(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="topk(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">topk</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@topk(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the indices of the top *k* elements in an input array along the given |
| axis (by default). |
| If ret_type is set to <span class="lit">'value'</span> returns the value of top *k* elements (instead of indices). |
| In <span class="kw">case</span> of ret_type = <span class="lit">'both'</span>, both value and index would be returned. |
| The returned elements will be sorted. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.3</span>, <span class="num">0.2</span>, <span class="num">0.4</span>], |
| [ <span class="num">0.1</span>, <span class="num">0.3</span>, <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// returns an index of the largest element on last axis</span> |
| topk(x) = `[ [ <span class="num">2.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| <span class="cmt">// returns the value of top-2 largest elements on last axis</span> |
| topk(x, ret_typ=<span class="lit">'value'</span>, k=<span class="num">2</span>) = `[ [ <span class="num">0.4</span>, <span class="num">0.3</span>], |
| [ <span class="num">0.3</span>, <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// returns the value of top-2 smallest elements on last axis</span> |
| topk(x, ret_typ=<span class="lit">'value'</span>, k=<span class="num">2</span>, is_ascend=<span class="num">1</span>) = `[ [ <span class="num">0.2</span> , <span class="num">0.3</span>], |
| [ <span class="num">0.1</span> , <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// returns the value of top-2 largest elements on axis 0</span> |
| topk(x, axis=<span class="num">0</span>, ret_typ=<span class="lit">'value'</span>, k=<span class="num">2</span>) = `[ [ <span class="num">0.3</span>, <span class="num">0.3</span>, <span class="num">0.4</span>], |
| [ <span class="num">0.1</span>, <span class="num">0.2</span>, <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// flattens and then returns list of both values and indices</span> |
| topk(x, ret_typ=<span class="lit">'both'</span>, k=<span class="num">2</span>) = `[ `[ [ <span class="num">0.4</span>, <span class="num">0.3</span>], [ <span class="num">0.3</span>, <span class="num">0.2</span>] ] , `[ [ <span class="num">2.</span>, <span class="num">0.</span>], [ <span class="num">1.</span>, <span class="num">2.</span>] ] ] |
| |
| |
| |
| Defined in src/operator/tensor/ordering_op.cc:L67</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#topk" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="topk(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="topk(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">topk</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@topk(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Returns the indices of the top *k* elements in an input array along the given |
| axis (by default). |
| If ret_type is set to <span class="lit">'value'</span> returns the value of top *k* elements (instead of indices). |
| In <span class="kw">case</span> of ret_type = <span class="lit">'both'</span>, both value and index would be returned. |
| The returned elements will be sorted. |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">0.3</span>, <span class="num">0.2</span>, <span class="num">0.4</span>], |
| [ <span class="num">0.1</span>, <span class="num">0.3</span>, <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// returns an index of the largest element on last axis</span> |
| topk(x) = `[ [ <span class="num">2.</span>], |
| [ <span class="num">1.</span>] ] |
| |
| <span class="cmt">// returns the value of top-2 largest elements on last axis</span> |
| topk(x, ret_typ=<span class="lit">'value'</span>, k=<span class="num">2</span>) = `[ [ <span class="num">0.4</span>, <span class="num">0.3</span>], |
| [ <span class="num">0.3</span>, <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// returns the value of top-2 smallest elements on last axis</span> |
| topk(x, ret_typ=<span class="lit">'value'</span>, k=<span class="num">2</span>, is_ascend=<span class="num">1</span>) = `[ [ <span class="num">0.2</span> , <span class="num">0.3</span>], |
| [ <span class="num">0.1</span> , <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// returns the value of top-2 largest elements on axis 0</span> |
| topk(x, axis=<span class="num">0</span>, ret_typ=<span class="lit">'value'</span>, k=<span class="num">2</span>) = `[ [ <span class="num">0.3</span>, <span class="num">0.3</span>, <span class="num">0.4</span>], |
| [ <span class="num">0.1</span>, <span class="num">0.2</span>, <span class="num">0.2</span>] ] |
| |
| <span class="cmt">// flattens and then returns list of both values and indices</span> |
| topk(x, ret_typ=<span class="lit">'both'</span>, k=<span class="num">2</span>) = `[ `[ [ <span class="num">0.4</span>, <span class="num">0.3</span>], [ <span class="num">0.3</span>, <span class="num">0.2</span>] ] , `[ [ <span class="num">2.</span>, <span class="num">0.</span>], [ <span class="num">1.</span>, <span class="num">2.</span>] ] ] |
| |
| |
| |
| Defined in src/operator/tensor/ordering_op.cc:L67</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#transpose" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="transpose(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="transpose(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">transpose</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@transpose(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Permutes the dimensions of an array. |
| Examples:: |
| x = `[ [ <span class="num">1</span>, <span class="num">2</span>], |
| [ <span class="num">3</span>, <span class="num">4</span>] ] |
| transpose(x) = `[ [ <span class="num">1.</span>, <span class="num">3.</span>], |
| [ <span class="num">2.</span>, <span class="num">4.</span>] ] |
| x = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>] ], |
| `[ [ <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>] ] ] |
| transpose(x) = `[ `[ [ <span class="num">1.</span>, <span class="num">5.</span>], |
| [ <span class="num">3.</span>, <span class="num">7.</span>] ], |
| `[ [ <span class="num">2.</span>, <span class="num">6.</span>], |
| [ <span class="num">4.</span>, <span class="num">8.</span>] ] ] |
| transpose(x, axes=(<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>)) = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ], |
| `[ [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L327</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#transpose" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="transpose(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="transpose(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">transpose</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@transpose(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Permutes the dimensions of an array. |
| Examples:: |
| x = `[ [ <span class="num">1</span>, <span class="num">2</span>], |
| [ <span class="num">3</span>, <span class="num">4</span>] ] |
| transpose(x) = `[ [ <span class="num">1.</span>, <span class="num">3.</span>], |
| [ <span class="num">2.</span>, <span class="num">4.</span>] ] |
| x = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">3.</span>, <span class="num">4.</span>] ], |
| `[ [ <span class="num">5.</span>, <span class="num">6.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>] ] ] |
| transpose(x) = `[ `[ [ <span class="num">1.</span>, <span class="num">5.</span>], |
| [ <span class="num">3.</span>, <span class="num">7.</span>] ], |
| `[ [ <span class="num">2.</span>, <span class="num">6.</span>], |
| [ <span class="num">4.</span>, <span class="num">8.</span>] ] ] |
| transpose(x, axes=(<span class="num">1</span>,<span class="num">0</span>,<span class="num">2</span>)) = `[ `[ [ <span class="num">1.</span>, <span class="num">2.</span>], |
| [ <span class="num">5.</span>, <span class="num">6.</span>] ], |
| `[ [ <span class="num">3.</span>, <span class="num">4.</span>], |
| [ <span class="num">7.</span>, <span class="num">8.</span>] ] ] |
| |
| |
| Defined in src/operator/tensor/matrix_op.cc:L327</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#trunc" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="trunc(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="trunc(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">trunc</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@trunc(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return the element-wise truncated value of the input. |
| |
| The truncated value of the scalar x is the nearest integer i which is closer to |
| zero than x is. In short, the fractional part of the signed number x is discarded. |
| |
| Example:: |
| |
| trunc([-<span class="num">2.1</span>, -<span class="num">1.9</span>, <span class="num">1.5</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, -<span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``trunc`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - trunc(default) = default |
| - trunc(row_sparse) = row_sparse |
| - trunc(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L856</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#trunc" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="trunc(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="trunc(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">trunc</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@trunc(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return the element-wise truncated value of the input. |
| |
| The truncated value of the scalar x is the nearest integer i which is closer to |
| zero than x is. In short, the fractional part of the signed number x is discarded. |
| |
| Example:: |
| |
| trunc([-<span class="num">2.1</span>, -<span class="num">1.9</span>, <span class="num">1.5</span>, <span class="num">1.9</span>, <span class="num">2.1</span>]) = [-<span class="num">2.</span>, -<span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">2.</span>] |
| |
| The storage <span class="kw">type</span> of ``trunc`` output depends upon the input storage <span class="kw">type</span>: |
| |
| - trunc(default) = default |
| - trunc(row_sparse) = row_sparse |
| - trunc(csr) = csr |
| |
| |
| |
| Defined in src/operator/tensor/elemwise_unary_op_basic.cc:L856</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#uniform" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="uniform(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="uniform(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">uniform</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@uniform(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a uniform distribution. |
| |
| .. note:: The existing alias ``uniform`` is deprecated. |
| |
| Samples are uniformly distributed over the half-open interval *[low, high)* |
| (includes *low*, but excludes *high*). |
| |
| Example:: |
| |
| uniform(low=<span class="num">0</span>, high=<span class="num">1</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">0.60276335</span>, <span class="num">0.85794562</span>], |
| [ <span class="num">0.54488319</span>, <span class="num">0.84725171</span>] ] |
| |
| |
| |
| Defined in src/operator/random/sample_op.cc:L95</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#uniform" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="uniform(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="uniform(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">uniform</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@uniform(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Draw random samples from a uniform distribution. |
| |
| .. note:: The existing alias ``uniform`` is deprecated. |
| |
| Samples are uniformly distributed over the half-open interval *[low, high)* |
| (includes *low*, but excludes *high*). |
| |
| Example:: |
| |
| uniform(low=<span class="num">0</span>, high=<span class="num">1</span>, shape=(<span class="num">2</span>,<span class="num">2</span>)) = `[ [ <span class="num">0.60276335</span>, <span class="num">0.85794562</span>], |
| [ <span class="num">0.54488319</span>, <span class="num">0.84725171</span>] ] |
| |
| |
| |
| Defined in src/operator/random/sample_op.cc:L95</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#unravel_index" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="unravel_index(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="unravel_index(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">unravel_index</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@unravel_index(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Converts an array of flat indices into a batch of index arrays. The operator follows numpy conventions so a single multi index is given by a column of the output matrix. The leading dimension may be left unspecified by using -<span class="num">1</span> as placeholder. |
| |
| Examples:: |
| |
| A = [<span class="num">22</span>,<span class="num">41</span>,<span class="num">37</span>] |
| unravel(A, shape=(<span class="num">7</span>,<span class="num">6</span>)) = `[ [<span class="num">3</span>,<span class="num">6</span>,<span class="num">6</span>],[<span class="num">4</span>,<span class="num">5</span>,<span class="num">1</span>] ] |
| unravel(A, shape=(-<span class="num">1</span>,<span class="num">6</span>)) = `[ [<span class="num">3</span>,<span class="num">6</span>,<span class="num">6</span>],[<span class="num">4</span>,<span class="num">5</span>,<span class="num">1</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/ravel.cc:L67</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#unravel_index" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="unravel_index(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="unravel_index(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">unravel_index</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@unravel_index(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Converts an array of flat indices into a batch of index arrays. The operator follows numpy conventions so a single multi index is given by a column of the output matrix. The leading dimension may be left unspecified by using -<span class="num">1</span> as placeholder. |
| |
| Examples:: |
| |
| A = [<span class="num">22</span>,<span class="num">41</span>,<span class="num">37</span>] |
| unravel(A, shape=(<span class="num">7</span>,<span class="num">6</span>)) = `[ [<span class="num">3</span>,<span class="num">6</span>,<span class="num">6</span>],[<span class="num">4</span>,<span class="num">5</span>,<span class="num">1</span>] ] |
| unravel(A, shape=(-<span class="num">1</span>,<span class="num">6</span>)) = `[ [<span class="num">3</span>,<span class="num">6</span>,<span class="num">6</span>],[<span class="num">4</span>,<span class="num">5</span>,<span class="num">1</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/ravel.cc:L67</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#where" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="where(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="where(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">where</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@where(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return the elements, either from x or y, depending on the condition. |
| |
| Given three ndarrays, condition, x, and y, <span class="kw">return</span> an ndarray <span class="kw">with</span> the elements from x or y, |
| depending on the elements from condition are <span class="kw">true</span> or <span class="kw">false</span>. x and y must have the same shape. |
| If condition has the same shape as x, each element in the output array is from x <span class="kw">if</span> the |
| corresponding element in the condition is <span class="kw">true</span>, and from y <span class="kw">if</span> <span class="kw">false</span>. |
| |
| If condition does not have the same shape as x, it must be a <span class="num">1</span>D array whose size is |
| the same as x's first dimension size. Each row of the output array is from x's row |
| <span class="kw">if</span> the corresponding element from condition is <span class="kw">true</span>, and from y's row <span class="kw">if</span> <span class="kw">false</span>. |
| |
| Note that all non-zero values are interpreted as ``True`` in condition. |
| |
| Examples:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">4</span>] ] |
| y = `[ [<span class="num">5</span>, <span class="num">6</span>], [<span class="num">7</span>, <span class="num">8</span>] ] |
| cond = `[ [<span class="num">0</span>, <span class="num">1</span>], [-<span class="num">1</span>, <span class="num">0</span>] ] |
| |
| where(cond, x, y) = `[ [<span class="num">5</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">8</span>] ] |
| |
| csr_cond = cast_storage(cond, <span class="lit">'csr'</span>) |
| |
| where(csr_cond, x, y) = `[ [<span class="num">5</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">8</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/control_flow_op.cc:L56</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#where" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="where(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="where(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">where</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@where(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return the elements, either from x or y, depending on the condition. |
| |
| Given three ndarrays, condition, x, and y, <span class="kw">return</span> an ndarray <span class="kw">with</span> the elements from x or y, |
| depending on the elements from condition are <span class="kw">true</span> or <span class="kw">false</span>. x and y must have the same shape. |
| If condition has the same shape as x, each element in the output array is from x <span class="kw">if</span> the |
| corresponding element in the condition is <span class="kw">true</span>, and from y <span class="kw">if</span> <span class="kw">false</span>. |
| |
| If condition does not have the same shape as x, it must be a <span class="num">1</span>D array whose size is |
| the same as x's first dimension size. Each row of the output array is from x's row |
| <span class="kw">if</span> the corresponding element from condition is <span class="kw">true</span>, and from y's row <span class="kw">if</span> <span class="kw">false</span>. |
| |
| Note that all non-zero values are interpreted as ``True`` in condition. |
| |
| Examples:: |
| |
| x = `[ [<span class="num">1</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">4</span>] ] |
| y = `[ [<span class="num">5</span>, <span class="num">6</span>], [<span class="num">7</span>, <span class="num">8</span>] ] |
| cond = `[ [<span class="num">0</span>, <span class="num">1</span>], [-<span class="num">1</span>, <span class="num">0</span>] ] |
| |
| where(cond, x, y) = `[ [<span class="num">5</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">8</span>] ] |
| |
| csr_cond = cast_storage(cond, <span class="lit">'csr'</span>) |
| |
| where(csr_cond, x, y) = `[ [<span class="num">5</span>, <span class="num">2</span>], [<span class="num">3</span>, <span class="num">8</span>] ] |
| |
| |
| |
| Defined in src/operator/tensor/control_flow_op.cc:L56</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#zeros_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="zeros_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="zeros_like(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">zeros_like</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@zeros_like(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return an array of zeros <span class="kw">with</span> the same shape, <span class="kw">type</span> and storage <span class="kw">type</span> |
| as the input array. |
| |
| The storage <span class="kw">type</span> of ``zeros_like`` output depends on the storage <span class="kw">type</span> of the input |
| |
| - zeros_like(row_sparse) = row_sparse |
| - zeros_like(csr) = csr |
| - zeros_like(default) = default |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| zeros_like(x) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li><li name="org.apache.mxnet.NDArrayBase#zeros_like" visbl="pub" data-isabs="true" fullComment="yes" group="Ungrouped"> |
| <a id="zeros_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn"></a> |
| <a id="zeros_like(Map[String,Any])(Any*):NDArrayFuncReturn"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">abstract </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">zeros_like</span><span class="params">(<span name="kwargs">kwargs: <span class="extype" name="scala.Predef.Map">Map</span>[<span class="extype" name="scala.Predef.String">String</span>, <span class="extype" name="scala.Any">Any</span>] = <span class="symbol">null</span></span>)</span><span class="params">(<span name="args">args: <span class="extype" name="scala.Any">Any</span>*</span>)</span><span class="result">: <span class="extype" name="org.apache.mxnet.NDArrayFuncReturn">NDArrayFuncReturn</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@zeros_like(kwargs:Map[String,Any])(args:Any*):org.apache.mxnet.NDArrayFuncReturn" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <p class="shortcomment cmt"></p><div class="fullcomment"><div class="comment cmt"><p></p><pre>Return an array of zeros <span class="kw">with</span> the same shape, <span class="kw">type</span> and storage <span class="kw">type</span> |
| as the input array. |
| |
| The storage <span class="kw">type</span> of ``zeros_like`` output depends on the storage <span class="kw">type</span> of the input |
| |
| - zeros_like(row_sparse) = row_sparse |
| - zeros_like(csr) = csr |
| - zeros_like(default) = default |
| |
| Examples:: |
| |
| x = `[ [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>], |
| [ <span class="num">1.</span>, <span class="num">1.</span>, <span class="num">1.</span>] ] |
| |
| zeros_like(x) = `[ [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>], |
| [ <span class="num">0.</span>, <span class="num">0.</span>, <span class="num">0.</span>] ]</pre></div><dl class="paramcmts block"><dt>returns</dt><dd class="cmt"><p>org.apache.mxnet.NDArrayFuncReturn</p></dd></dl></div> |
| </li></ol> |
| </div> |
| |
| <div id="values" class="values members"> |
| <h3>Concrete Value Members</h3> |
| <ol><li name="scala.AnyRef#!=" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="!=(x$1:Any):Boolean"></a> |
| <a id="!=(Any):Boolean"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span title="gt4s: $bang$eq" class="name">!=</span><span class="params">(<span name="arg0">arg0: <span class="extype" name="scala.Any">Any</span></span>)</span><span class="result">: <span class="extype" name="scala.Boolean">Boolean</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@!=(x$1:Any):Boolean" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef → Any</dd></dl></div> |
| </li><li name="scala.AnyRef###" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="##():Int"></a> |
| <a id="##():Int"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span title="gt4s: $hash$hash" class="name">##</span><span class="params">()</span><span class="result">: <span class="extype" name="scala.Int">Int</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@##():Int" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef → Any</dd></dl></div> |
| </li><li name="scala.AnyRef#==" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="==(x$1:Any):Boolean"></a> |
| <a id="==(Any):Boolean"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span title="gt4s: $eq$eq" class="name">==</span><span class="params">(<span name="arg0">arg0: <span class="extype" name="scala.Any">Any</span></span>)</span><span class="result">: <span class="extype" name="scala.Boolean">Boolean</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@==(x$1:Any):Boolean" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef → Any</dd></dl></div> |
| </li><li name="scala.Any#asInstanceOf" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="asInstanceOf[T0]:T0"></a> |
| <a id="asInstanceOf[T0]:T0"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">asInstanceOf</span><span class="tparams">[<span name="T0">T0</span>]</span><span class="result">: <span class="extype" name="scala.Any.asInstanceOf.T0">T0</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@asInstanceOf[T0]:T0" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>Any</dd></dl></div> |
| </li><li name="scala.AnyRef#clone" visbl="prt" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="clone():Object"></a> |
| <a id="clone():AnyRef"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier"></span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">clone</span><span class="params">()</span><span class="result">: <span class="extype" name="scala.AnyRef">AnyRef</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@clone():Object" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Attributes</dt><dd>protected[<a href="../../../java$lang.html" class="extype" name="java.lang">java.lang</a>] </dd><dt>Definition Classes</dt><dd>AnyRef</dd><dt>Annotations</dt><dd> |
| <span class="name">@throws</span><span class="args">(<span> |
| |
| <span class="defval" name="classOf[java.lang.CloneNotSupportedException]">...</span> |
| </span>)</span> |
| |
| </dd></dl></div> |
| </li><li name="scala.AnyRef#eq" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="eq(x$1:AnyRef):Boolean"></a> |
| <a id="eq(AnyRef):Boolean"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">eq</span><span class="params">(<span name="arg0">arg0: <span class="extype" name="scala.AnyRef">AnyRef</span></span>)</span><span class="result">: <span class="extype" name="scala.Boolean">Boolean</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@eq(x$1:AnyRef):Boolean" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef</dd></dl></div> |
| </li><li name="scala.AnyRef#equals" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="equals(x$1:Any):Boolean"></a> |
| <a id="equals(Any):Boolean"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier"></span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">equals</span><span class="params">(<span name="arg0">arg0: <span class="extype" name="scala.Any">Any</span></span>)</span><span class="result">: <span class="extype" name="scala.Boolean">Boolean</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@equals(x$1:Any):Boolean" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef → Any</dd></dl></div> |
| </li><li name="scala.AnyRef#finalize" visbl="prt" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="finalize():Unit"></a> |
| <a id="finalize():Unit"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier"></span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">finalize</span><span class="params">()</span><span class="result">: <span class="extype" name="scala.Unit">Unit</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@finalize():Unit" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Attributes</dt><dd>protected[<a href="../../../java$lang.html" class="extype" name="java.lang">java.lang</a>] </dd><dt>Definition Classes</dt><dd>AnyRef</dd><dt>Annotations</dt><dd> |
| <span class="name">@throws</span><span class="args">(<span> |
| |
| <span class="symbol">classOf[java.lang.Throwable]</span> |
| </span>)</span> |
| |
| </dd></dl></div> |
| </li><li name="scala.AnyRef#getClass" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="getClass():Class[_]"></a> |
| <a id="getClass():Class[_]"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">getClass</span><span class="params">()</span><span class="result">: <span class="extype" name="java.lang.Class">Class</span>[_]</span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@getClass():Class[_]" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef → Any</dd></dl></div> |
| </li><li name="scala.AnyRef#hashCode" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="hashCode():Int"></a> |
| <a id="hashCode():Int"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier"></span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">hashCode</span><span class="params">()</span><span class="result">: <span class="extype" name="scala.Int">Int</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@hashCode():Int" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef → Any</dd></dl></div> |
| </li><li name="scala.Any#isInstanceOf" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="isInstanceOf[T0]:Boolean"></a> |
| <a id="isInstanceOf[T0]:Boolean"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">isInstanceOf</span><span class="tparams">[<span name="T0">T0</span>]</span><span class="result">: <span class="extype" name="scala.Boolean">Boolean</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@isInstanceOf[T0]:Boolean" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>Any</dd></dl></div> |
| </li><li name="scala.AnyRef#ne" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="ne(x$1:AnyRef):Boolean"></a> |
| <a id="ne(AnyRef):Boolean"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">ne</span><span class="params">(<span name="arg0">arg0: <span class="extype" name="scala.AnyRef">AnyRef</span></span>)</span><span class="result">: <span class="extype" name="scala.Boolean">Boolean</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@ne(x$1:AnyRef):Boolean" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef</dd></dl></div> |
| </li><li name="scala.AnyRef#notify" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="notify():Unit"></a> |
| <a id="notify():Unit"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">notify</span><span class="params">()</span><span class="result">: <span class="extype" name="scala.Unit">Unit</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@notify():Unit" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef</dd></dl></div> |
| </li><li name="scala.AnyRef#notifyAll" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="notifyAll():Unit"></a> |
| <a id="notifyAll():Unit"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">notifyAll</span><span class="params">()</span><span class="result">: <span class="extype" name="scala.Unit">Unit</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@notifyAll():Unit" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef</dd></dl></div> |
| </li><li name="scala.AnyRef#synchronized" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="synchronized[T0](x$1:=>T0):T0"></a> |
| <a id="synchronized[T0](⇒T0):T0"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">synchronized</span><span class="tparams">[<span name="T0">T0</span>]</span><span class="params">(<span name="arg0">arg0: ⇒ <span class="extype" name="java.lang.AnyRef.synchronized.T0">T0</span></span>)</span><span class="result">: <span class="extype" name="java.lang.AnyRef.synchronized.T0">T0</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@synchronized[T0](x$1:=>T0):T0" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef</dd></dl></div> |
| </li><li name="scala.AnyRef#toString" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="toString():String"></a> |
| <a id="toString():String"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier"></span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">toString</span><span class="params">()</span><span class="result">: <span class="extype" name="java.lang.String">String</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@toString():String" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef → Any</dd></dl></div> |
| </li><li name="scala.AnyRef#wait" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="wait():Unit"></a> |
| <a id="wait():Unit"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">wait</span><span class="params">()</span><span class="result">: <span class="extype" name="scala.Unit">Unit</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@wait():Unit" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef</dd><dt>Annotations</dt><dd> |
| <span class="name">@throws</span><span class="args">(<span> |
| |
| <span class="defval" name="classOf[java.lang.InterruptedException]">...</span> |
| </span>)</span> |
| |
| </dd></dl></div> |
| </li><li name="scala.AnyRef#wait" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="wait(x$1:Long,x$2:Int):Unit"></a> |
| <a id="wait(Long,Int):Unit"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">wait</span><span class="params">(<span name="arg0">arg0: <span class="extype" name="scala.Long">Long</span></span>, <span name="arg1">arg1: <span class="extype" name="scala.Int">Int</span></span>)</span><span class="result">: <span class="extype" name="scala.Unit">Unit</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@wait(x$1:Long,x$2:Int):Unit" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef</dd><dt>Annotations</dt><dd> |
| <span class="name">@throws</span><span class="args">(<span> |
| |
| <span class="defval" name="classOf[java.lang.InterruptedException]">...</span> |
| </span>)</span> |
| |
| </dd></dl></div> |
| </li><li name="scala.AnyRef#wait" visbl="pub" data-isabs="false" fullComment="yes" group="Ungrouped"> |
| <a id="wait(x$1:Long):Unit"></a> |
| <a id="wait(Long):Unit"></a> |
| <h4 class="signature"> |
| <span class="modifier_kind"> |
| <span class="modifier">final </span> |
| <span class="kind">def</span> |
| </span> |
| <span class="symbol"> |
| <span class="name">wait</span><span class="params">(<span name="arg0">arg0: <span class="extype" name="scala.Long">Long</span></span>)</span><span class="result">: <span class="extype" name="scala.Unit">Unit</span></span> |
| </span> |
| </h4><span class="permalink"> |
| <a href="../../../index.html#org.apache.mxnet.NDArrayBase@wait(x$1:Long):Unit" title="Permalink" target="_top"> |
| <img src="../../../lib/permalink.png" alt="Permalink" /> |
| </a> |
| </span> |
| <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef</dd><dt>Annotations</dt><dd> |
| <span class="name">@throws</span><span class="args">(<span> |
| |
| <span class="defval" name="classOf[java.lang.InterruptedException]">...</span> |
| </span>)</span> |
| |
| </dd></dl></div> |
| </li></ol> |
| </div> |
| |
| |
| |
| |
| </div> |
| |
| <div id="inheritedMembers"> |
| <div class="parent" name="scala.AnyRef"> |
| <h3>Inherited from <span class="extype" name="scala.AnyRef">AnyRef</span></h3> |
| </div><div class="parent" name="scala.Any"> |
| <h3>Inherited from <span class="extype" name="scala.Any">Any</span></h3> |
| </div> |
| |
| </div> |
| |
| <div id="groupedMembers"> |
| <div class="group" name="Ungrouped"> |
| <h3>Ungrouped</h3> |
| |
| </div> |
| </div> |
| |
| </div> |
| |
| <div id="tooltip"></div> |
| |
| <div id="footer"> </div> |
| |
| |
| </body> |
| </html> |