From 4beb73d34f359c05e96582ebeefe094ff49eea22 Mon Sep 17 00:00:00 2001
From: sebffischer <sebf.fischer@gmail.com>
Date: Mon, 2 Dec 2024 20:33:51 +0000
Subject: [PATCH] =?UTF-8?q?Deploying=20to=20gh-pages=20from=20@=20mlr-org/?=
 =?UTF-8?q?mlr3torch@87c85f0bdbb9a7bb29892b4b6fe6b3f9bcebbe75=20?=
 =?UTF-8?q?=F0=9F=9A=80?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 dev/articles/callbacks.html                   |   2 +-
 dev/articles/get_started.html                 |   2 +-
 dev/articles/internals_pipeop_torch.html      |  58 ++--
 .../figure-html/unnamed-chunk-37-1.png        | Bin 123684 -> 123031 bytes
 .../figure-html/unnamed-chunk-42-1.png        | Bin 123488 -> 123410 bytes
 .../figure-html/unnamed-chunk-46-1.png        | Bin 123783 -> 123349 bytes
 .../figure-html/unnamed-chunk-48-1.png        | Bin 123270 -> 123185 bytes
 .../figure-html/unnamed-chunk-50-1.png        | Bin 123536 -> 123160 bytes
 dev/articles/lazy_tensor.html                 |   8 +-
 dev/pkgdown.yml                               |   2 +-
 dev/reference/DataDescriptor.html             |   4 +-
 dev/reference/TorchCallback.html              |   2 +-
 dev/reference/as_data_descriptor.html         |   4 +-
 dev/reference/materialize.html                | 260 +++++++++---------
 .../mlr_pipeops_torch_ingress_ltnsr.html      |  46 ++--
 .../mlr_pipeops_torch_model_classif.html      |  28 +-
 .../mlr_pipeops_torch_model_regr.html         |  17 +-
 dev/reference/nn_graph.html                   |  32 +--
 dev/search.json                               |   2 +-
 19 files changed, 232 insertions(+), 235 deletions(-)

diff --git a/dev/articles/callbacks.html b/dev/articles/callbacks.html
index 9fd157cb..b5bbb3ef 100644
--- a/dev/articles/callbacks.html
+++ b/dev/articles/callbacks.html
@@ -225,7 +225,7 @@ <h3 id="writing-a-custom-logger">Writing a Custom Logger<a class="anchor" aria-l
 <span><span class="co">##     load_state_dict: function (state_dict) </span></span>
 <span><span class="co">##     on_before_valid: function () </span></span>
 <span><span class="co">##     on_batch_end: function () </span></span>
-<span><span class="co">##   Parent env: &lt;environment: 0x5603df3b07c8&gt;</span></span>
+<span><span class="co">##   Parent env: &lt;environment: 0x55fef5065170&gt;</span></span>
 <span><span class="co">##   Locked objects: FALSE</span></span>
 <span><span class="co">##   Locked class: FALSE</span></span>
 <span><span class="co">##   Portable: TRUE</span></span></code></pre>
diff --git a/dev/articles/get_started.html b/dev/articles/get_started.html
index a745d517..132aeddb 100644
--- a/dev/articles/get_started.html
+++ b/dev/articles/get_started.html
@@ -236,7 +236,7 @@ <h3 id="loss">Loss<a class="anchor" aria-label="anchor" href="#loss"></a>
 <span><span class="co">#&gt;     clone: function (deep = FALSE, ..., replace_values = TRUE) </span></span>
 <span><span class="co">#&gt;   Private:</span></span>
 <span><span class="co">#&gt;     .__clone_r6__: function (deep = FALSE) </span></span>
-<span><span class="co">#&gt;   Parent env: &lt;environment: 0x55a43f05f830&gt;</span></span>
+<span><span class="co">#&gt;   Parent env: &lt;environment: 0x55ffc865fba8&gt;</span></span>
 <span><span class="co">#&gt;   Locked objects: FALSE</span></span>
 <span><span class="co">#&gt;   Locked class: FALSE</span></span>
 <span><span class="co">#&gt;   Portable: TRUE</span></span></code></pre></div>
diff --git a/dev/articles/internals_pipeop_torch.html b/dev/articles/internals_pipeop_torch.html
index a6f0cb20..9bbb6894 100644
--- a/dev/articles/internals_pipeop_torch.html
+++ b/dev/articles/internals_pipeop_torch.html
@@ -104,8 +104,8 @@ <h2 id="a-torch-primer">A <code>torch</code> Primer<a class="anchor" aria-label=
 <code class="sourceCode R"><span><span class="va">input</span> <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/pkg/torch/man/torch_randn.html" class="external-link">torch_randn</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">3</span><span class="op">)</span></span>
 <span><span class="va">input</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt; -0.6687 -2.4004 -1.4578</span></span>
-<span><span class="co">#&gt;  0.9376 -0.9994 -0.3993</span></span>
+<span><span class="co">#&gt;  0.4430 -1.7493  0.3801</span></span>
+<span><span class="co">#&gt;  2.5443  0.4085 -0.4808</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,3} ]</span></span></code></pre></div>
 <p>A <code>nn_module</code> is constructed from a
 <code>nn_module_generator</code>. <code>nn_linear</code> is one of the
@@ -117,8 +117,8 @@ <h2 id="a-torch-primer">A <code>torch</code> Primer<a class="anchor" aria-label=
 <code class="sourceCode R"><span><span class="va">output</span> <span class="op">=</span> <span class="fu">module_1</span><span class="op">(</span><span class="va">input</span><span class="op">)</span></span>
 <span><span class="va">output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.0980  0.3119 -1.4607 -0.4350</span></span>
-<span><span class="co">#&gt;  0.2679  0.8104  0.1816  0.0485</span></span>
+<span><span class="co">#&gt;  0.6725  0.3022 -0.3800  0.3413</span></span>
+<span><span class="co">#&gt; -0.8161  0.3485  0.3559 -1.3795</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,4} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span></code></pre></div>
 <p>A neural network with one (4-unit) hidden layer and two outputs needs
 the following ingredients</p>
@@ -134,8 +134,8 @@ <h2 id="a-torch-primer">A <code>torch</code> Primer<a class="anchor" aria-label=
 <span><span class="va">output</span> <span class="op">=</span> <span class="fu">softmax</span><span class="op">(</span><span class="va">output</span><span class="op">)</span></span>
 <span><span class="va">output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.2918  0.1502  0.5579</span></span>
-<span><span class="co">#&gt;  0.2890  0.1217  0.5893</span></span>
+<span><span class="co">#&gt;  0.2155  0.3400  0.4445</span></span>
+<span><span class="co">#&gt;  0.1626  0.3668  0.4706</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,3} ][ grad_fn = &lt;SoftmaxBackward0&gt; ]</span></span></code></pre></div>
 <p>We will now continue with showing how such a neural network can be
 represented in <code>mlr3torch</code>.</p>
@@ -170,8 +170,8 @@ <h2 id="neural-networks-as-graphs">Neural Networks as Graphs<a class="anchor" ar
 <code class="sourceCode R"><span><span class="va">output</span> <span class="op">=</span> <span class="va">po_module_1</span><span class="op">$</span><span class="fu">train</span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="va">input</span><span class="op">)</span><span class="op">)</span><span class="op">[[</span><span class="fl">1</span><span class="op">]</span><span class="op">]</span></span>
 <span><span class="va">output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.0980  0.3119 -1.4607 -0.4350</span></span>
-<span><span class="co">#&gt;  0.2679  0.8104  0.1816  0.0485</span></span>
+<span><span class="co">#&gt;  0.6725  0.3022 -0.3800  0.3413</span></span>
+<span><span class="co">#&gt; -0.8161  0.3485  0.3559 -1.3795</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,4} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span></code></pre></div>
 <p>Note we only use the <code>$train()</code>, since torch modules do
 not have anything that maps to the <code>state</code> (it is filled by
@@ -196,8 +196,8 @@ <h2 id="neural-networks-as-graphs">Neural Networks as Graphs<a class="anchor" ar
 <code class="sourceCode R"><span><span class="va">output</span> <span class="op">=</span> <span class="va">module_graph</span><span class="op">$</span><span class="fu">train</span><span class="op">(</span><span class="va">input</span><span class="op">)</span><span class="op">[[</span><span class="fl">1</span><span class="op">]</span><span class="op">]</span></span>
 <span><span class="va">output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.2918  0.1502  0.5579</span></span>
-<span><span class="co">#&gt;  0.2890  0.1217  0.5893</span></span>
+<span><span class="co">#&gt;  0.2155  0.3400  0.4445</span></span>
+<span><span class="co">#&gt;  0.1626  0.3668  0.4706</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,3} ][ grad_fn = &lt;SoftmaxBackward0&gt; ]</span></span></code></pre></div>
 <p>While this object allows to easily perform a forward pass, it does
 not inherit from <code>nn_module</code>, which is useful for various
@@ -245,8 +245,8 @@ <h2 id="neural-networks-as-graphs">Neural Networks as Graphs<a class="anchor" ar
 <div class="sourceCode" id="cb12"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu">graph_module</span><span class="op">(</span><span class="va">input</span><span class="op">)</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.2918  0.1502  0.5579</span></span>
-<span><span class="co">#&gt;  0.2890  0.1217  0.5893</span></span>
+<span><span class="co">#&gt;  0.2155  0.3400  0.4445</span></span>
+<span><span class="co">#&gt;  0.1626  0.3668  0.4706</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,3} ][ grad_fn = &lt;SoftmaxBackward0&gt; ]</span></span></code></pre></div>
 </div>
 <div class="section level2">
@@ -363,8 +363,8 @@ <h3 id="modeldescriptor">
 <span></span>
 <span><span class="fu">small_module</span><span class="op">(</span><span class="va">input</span><span class="op">)</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  1.2548 -1.3432 -1.5458  0.7297</span></span>
-<span><span class="co">#&gt;  0.4231 -0.7686 -0.7230  0.6464</span></span>
+<span><span class="co">#&gt; -0.1274  0.1809 -1.4397 -0.5290</span></span>
+<span><span class="co">#&gt;  0.6121 -0.7253 -0.5453 -0.2486</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,4} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span></code></pre></div>
 </div>
 <div class="section level3">
@@ -429,9 +429,9 @@ <h3 id="using-modeldescriptor-to-get-data">Using <code>ModelDescriptor</code> to
 <span></span>
 <span><span class="fu">small_module</span><span class="op">(</span><span class="va">batch</span><span class="op">$</span><span class="va">x</span><span class="op">[[</span><span class="fl">1</span><span class="op">]</span><span class="op">]</span><span class="op">)</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt; -0.9548  2.4764  1.7547 -0.4855</span></span>
-<span><span class="co">#&gt; -1.1006  2.1895  1.5184 -0.5936</span></span>
-<span><span class="co">#&gt; -0.8588  2.2540  1.5650 -0.4339</span></span>
+<span><span class="co">#&gt;  0.3122 -4.8047  0.3374  4.3038</span></span>
+<span><span class="co">#&gt;  0.4304 -4.4378  0.4278  3.9666</span></span>
+<span><span class="co">#&gt;  0.3240 -4.4393  0.2842  3.9762</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{3,4} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span></code></pre></div>
 </div>
 <div class="section level3">
@@ -457,8 +457,8 @@ <h3 id="building-sequential-nns">Building sequential NNs<a class="anchor" aria-l
 <span></span>
 <span><span class="fu">graph_module</span><span class="op">(</span><span class="va">input</span><span class="op">)</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.2554  0.3238  0.4209</span></span>
-<span><span class="co">#&gt;  0.2483  0.3749  0.3768</span></span>
+<span><span class="co">#&gt;  0.2329  0.2873  0.4798</span></span>
+<span><span class="co">#&gt;  0.1835  0.2948  0.5217</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,3} ][ grad_fn = &lt;SoftmaxBackward0&gt; ]</span></span></code></pre></div>
 </div>
 <div class="section level3">
@@ -673,14 +673,14 @@ <h3 id="building-more-interesting-nns">Building more interesting NNs<a class="an
 <span><span class="op">)</span></span>
 <span><span class="co">#&gt; $lin_out.output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.5751</span></span>
-<span><span class="co">#&gt;  0.4885</span></span>
+<span><span class="co">#&gt;  0.6367</span></span>
+<span><span class="co">#&gt;  0.6419</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,1} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span>
 <span><span class="co">#&gt; </span></span>
 <span><span class="co">#&gt; $nn_softmax.output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.1594  0.4310  0.4096</span></span>
-<span><span class="co">#&gt;  0.1743  0.4296  0.3961</span></span>
+<span><span class="co">#&gt;  0.2731  0.3325  0.3944</span></span>
+<span><span class="co">#&gt;  0.2820  0.3441  0.3739</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,3} ][ grad_fn = &lt;SoftmaxBackward0&gt; ]</span></span></code></pre></div>
 <p>The first linear layer that takes “Sepal” input
 (<code>"linear1"</code>) creates a 2x4 tensor (batch size 2, 4 units),
@@ -689,14 +689,14 @@ <h3 id="building-more-interesting-nns">Building more interesting NNs<a class="an
 <code class="sourceCode R"><span><span class="va">iris_module</span><span class="op">$</span><span class="va">graph</span><span class="op">$</span><span class="va">pipeops</span><span class="op">$</span><span class="va">linear1</span><span class="op">$</span><span class="va">.result</span></span>
 <span><span class="co">#&gt; $output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  2.6239  0.8791 -1.1404  0.1583</span></span>
-<span><span class="co">#&gt;  2.4387  1.0496 -1.3065  0.2425</span></span>
+<span><span class="co">#&gt; -3.2734  1.7562  1.3484  2.9473</span></span>
+<span><span class="co">#&gt; -2.9881  1.8728  1.3176  2.7498</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,4} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span>
 <span><span class="va">iris_module</span><span class="op">$</span><span class="va">graph</span><span class="op">$</span><span class="va">pipeops</span><span class="op">$</span><span class="va">linear3</span><span class="op">$</span><span class="va">.result</span></span>
 <span><span class="co">#&gt; $output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt; -0.0769  0.4376  0.2229 -0.0915 -0.3279</span></span>
-<span><span class="co">#&gt; -0.0769  0.4376  0.2229 -0.0915 -0.3279</span></span>
+<span><span class="co">#&gt;  0.1362  0.5117 -0.3741  0.8576  0.3130</span></span>
+<span><span class="co">#&gt;  0.1362  0.5117 -0.3741  0.8576  0.3130</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,5} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span></code></pre></div>
 <p>We observe that the <code>po("nn_merge_cat")</code> concatenates
 these, as expected:</p>
@@ -704,8 +704,8 @@ <h3 id="building-more-interesting-nns">Building more interesting NNs<a class="an
 <code class="sourceCode R"><span><span class="va">iris_module</span><span class="op">$</span><span class="va">graph</span><span class="op">$</span><span class="va">pipeops</span><span class="op">$</span><span class="va">nn_merge_cat</span><span class="op">$</span><span class="va">.result</span></span>
 <span><span class="co">#&gt; $output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  2.6239  0.8791 -1.1404  0.1583 -0.0769  0.4376  0.2229 -0.0915 -0.3279</span></span>
-<span><span class="co">#&gt;  2.4387  1.0496 -1.3065  0.2425 -0.0769  0.4376  0.2229 -0.0915 -0.3279</span></span>
+<span><span class="co">#&gt; -3.2734  1.7562  1.3484  2.9473  0.1362  0.5117 -0.3741  0.8576  0.3130</span></span>
+<span><span class="co">#&gt; -2.9881  1.8728  1.3176  2.7498  0.1362  0.5117 -0.3741  0.8576  0.3130</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,9} ][ grad_fn = &lt;CatBackward0&gt; ]</span></span></code></pre></div>
 </li>
 </ol>
diff --git a/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-37-1.png b/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-37-1.png
index f8d1e32f1505a66ba8b1e2a305471ee5619e0a65..bc316f9a2fdc508437d38443d2ea349924a39476 100644
GIT binary patch
literal 123031
zcmd3Og<n)(^esqBD$*s=(%m2-2uMq}NOyNgBRzDBAV@b1ISfdb(%s!XH1Fb9f1mgM
zfcHMnkMo(!aA)p4=j^lg+H0>NTuDJ1^C|IDI5;>=nRk*ZaBxWOaBv7lsK~%8dQV36
zfhXjTa?+B(Gw|<9qR9_<LbHFT<pc*u7W(*rk6Ax9frEPqCnG8L!98Pd-pyTiJvHRm
zORA+mY3=na*w7fGjv?PoygVmZsrt=38hBUUru3&WJ^7)SG}L*|#l<TnP|4oc%2%@2
zIUKh1Ceg^Lr`P$B&f?5bN3AJ0_)R88#HTvxL!<@LP6l`v31q3CqfyI%{`}8W;2Xz$
z_unx{yW3ym@1Xz#g8tuAcJeUXY5qGdjOSQN{O_%K3Vr|APJS&sj_J%3E*Vge<@tKt
z<_9LtbF^>2=`AcQf;xORerI-mWM4VvgWKNT4ie9K;g<K|i&SoqIOWL5h|2sI74vQ<
z#5a7T7oJ-iz*&`=1DmMNqNHS~pTm9n^eGQa@?D$y`Eg5-tB5ol*@<V*@bKo@sG_0^
z31~)!zH&7&JdGyq9h>bO_(MJApHYu71tMd9yuCaww3Sd1bU?IYocF!i*Y(`~4!mIV
zl02r`q#ys}^t89HPu4|HHfLr=o0x<Io}P)xSyF+w-ep()B^ZCq=k2Sw%t>)|b^NQV
zD~5x2g=14w?`&+Cg+0%<?5V12{M6Lc_DKshR8&AKwh5MriHU1#Yq9b1@jL1X;w~;O
zW0R8z&R|%Q=*{%7`(X=AWj-X&^;6rdEHDdXJMcQC1To#c0|V5IjQxuiyaeHruD<ge
zzz_cQ>9fb~ORM5Ae5_F51jHi`aX2n+?zqH6leXDG6I0Uz40j*h*h3mbsIwh~kM7pH
z$w1ZjGJ4tN@BiPf`v1Ln`v2sU|Ae(V5Z02f;&yn(kX>)d?=p9Co5^beC&geF$BU3j
z3D|}e@_zJWKwCG`-LCQH$-un3LI3*1I-A+9jlp!)d8y(%^L?s9Td|#~iTp-*8I{m=
zry!BOU9lbf`{G5l!G-#|x}=$z8P~mh6JlcGu+EQvqLyRH=KJ2B_0y+MRe+B$F*V6=
z49230dxvB3B{nNgfp4#bW?qYnn1(z8AvP{fjGk21L0a~^IoH!So{fr$^IEU6UOo?p
z!%;x@5_WbvcF!=UmX$I^&WRm<Hm{Xe7)};=>aAD$P8D~#hU?ERlvVjCugt@uD=`@I
z`Wxo66IpZyqPg^2{kMg-r4mz<HxFH1QfX;va6P@f+e+m`92^|p<%}*4WSh?l;$;*S
z=@=Oq$8Bn~J<}l9WX}=d?SSpPIWpL}dYt)qd~n7(l9-{%MY#Sau2(yLvxq3t_k~j2
z2C*&e553Ak6@0EwLGU7{P*qgAB)uh|7h7a_WzuZ4rqn3;<|dlBCq*uLGmKI36?3&y
zma)#`4QVpfaW3tfdx&Xkr%}9@)zgLfI43<qU17Rqk=5F?;C*`u1cEmg`z)j*;;S9c
zr{J6_w2kp9+acMBL9O0dh+kZUuQx(qG;!8U7mZ94+kZe=NK!bh*-6EFPeHO#9<qV~
z&L<sW3fznFi4uipDhUo%SLJr$^lLpZ%Xql6BJ!F>W7yUK)p#kizk0+T72anyVMq1D
z^CUq{x#-JKc6*~ky$(Hp9JW^!2Lh<A0x5iGS=ZMROujxBzK6gY&5)b_0JsY->%%9Z
zUDI;^K=dU)?0|2kzLl`%g`;c<$~ePY8%R!BRlW~NA~kn9eEA<dpKnV_#Cw0!UmP2c
zyePny;#N<iz!>c0^;=VQI340y=aZH|U7PYhC)Yq)6@>6Z<!;1fOn1NN^1@ZMzxO-C
z{Dv7<XQKofd;RY=5YXv{P<e)!tbF3Ki0nVH$6eDVwQ`r+5XV!~bpDq=@>t3$>#_#>
zdEP**x8f6ZD?1|KGzG0ZPNHN$Ak#L+Kl4DKp`%P@olxs&84@$2pIi6s^_4a#FOTR*
z_Hnplot?(VhE$}ot(x>d;=<@a<uo#~X@754PghLKp+fn0f_LShSk!IRw>7K({ezSB
zQY4fiSrv`GW^ZOKtLvWp@>=uf*?JF3VEUT@8yl6vmJ6cL>I|mmt_5n9-M!G4&$vRe
z?CodIXk}!-{LiE;XA3Z-h{Aqziz&?#`{|&&uuB;@V`yFR+suBYE>!!QMR$-rDLxEs
zMMQrtRlfZF(R}-r#c0veu}hwriT#NNi*+_FjSTa@aq~J{@RsLoe$_uxz`H&CXdcOX
z*j8WP2E!M71-)8?@T3fDfW{~Z|4xJaCxZ2pZ4^7SCo_}09R#BO42cy5&}KaOqloWE
znMg_$t7Ee)re~eq2a{PeGG579JIH?)f8<`31B}fpe@TmLdxP1(kcREX6qH58Rz`UJ
zNkW~ljte@couY$)*7Le&IpENs8@frJx^qDzLt*@Pd3fVBRN89&8Z(2-hd!RzWtoki
z(Hx0)vfQjE<8fTCtn%GEnZ>InBnK$a$lQGV`zE&+PuXv4*fB(HFlKosD=e=+TJoyn
z7TQrmO2R-;pDnMRm*UGxMS_Xc`@#R_SXW;L9wpO3vuHM!x;Eh7t8-Nhrzv-5O&`jq
z-{-N5`)WPA!y@|$E>@u76crrbmjPY0|BW?U`h^$k4TtxMM1I-gB@B)Icc)o?G2x<_
zZ?<Dwb#LQIw%oShZy+)<BO?D|Yp}%EkHN3?-b`{QJ%9vxqb;uU7cY8MM+N_rQ_JN$
zOCDyeZadbNh~ETc{hgu>l(kWEAXlS*6T<;(u!w}5^HQ&%5oV$AVh8*VZdz=hed?!!
zfIGpt+gm?X^NEaC_*fJN>u@n2Sd=xYs{b~mWM@0~d+<K=<(zOm!JJP|7O?LFoG<rY
zUM|0*e`mIj(of`fV+S+n0Ok4oWo3ow?w27=;oU3tj%eO4sCl7jZc8cRU?;@9;FQyW
z4@vT=ldVy*ptQk%vthUq#4u+4BY>R9{sza~w61BC63Jf84=(FETeQx@50p2-{tw1d
zZqqjdqZDDG88s7l4JE8=1*$D|pdz*OnKFJEmEwQ1HQu>utzRcXSnDM#v%9;N2dxx_
zbKk{gt(CU}_Upeu?Nlo(Ld91niACrn>ctbRf%89iQm+v+1Lb}C_d%U%Oxgt238u6L
zCDcDEORH%AZ20d5K+b#I@-=4R?g5Q`#HhhN`*Di&b0g!wUkY2@{Fpa?brteTEUM;A
zeQ1?Y#!~ek60zw_s-4{9#%o>nz2_Owc`JmE@-hQIfB47e#{dk~^L>KU@+DoZ92RzB
zXe?oTGH8Qgxt92X)<6!F!1Wgi1j5XXgC-B|Q4WQ;;xtvcz?E!dveocm8eEmG6OI!@
z+CzCLS?ejo8PHyX2LGDK7D&tRp_-(pslac{sES7TzG|JtHLr0v4F3D2iAf-WiT-`C
z|I8Ol^1F7QlGFwajt+YdgG)&w(pYpsQ<Haaa`jNT^ssv;gm`lIq{LBt$Ey|t>he6p
z^7ETW;-5RP<hE0i<^0d3R*uG0%^FBKIMRAv+PjwTKr;yuC4VL>;iLR4vHK+_1hsws
z+0{{K<<Ybl+a=X93*xNA>C$Q763p08lbo@Dc_WVGkXB`}X3}r9#7_4a>BE|*=3-wV
zu`lH5H?+QP3|danaOE~o9?!px|K;v99lO$DnV3)&RaB6$j|1kTbjWnsNl&QO*2vH3
zaS;vKbAjC(hbW1dqH=XEM*|>ar4_Ge=dZ!@>}e?Y>>ypowS>BG{%>R?h!^Lx8V!xA
z=1}9datozo<up{aA7w5pWk3heR5f|MI`!Qk8;bRDda(#{yGx=Q?fES+*GK^tl;nEo
zFeZeh5<7hv<p<CPj9$i4<L}yk|Fwpag~?OKh>icjs_;(^_LmW~+!r&o<>8HfK-Th@
zt}rK=<H9A^LKu(JbBm@4KpMBr#DwUewXySNZO=7<#-QMyv_JcjVdTSP4@ky|k{5G`
z-pVJZx^D@J$K-3w^ScXmTZ_nRuJ)6?oo)IV?!EnZ%ru)F-mL@Ptpy&&aHxnD^v4qO
zvlIh#Ze_TUY7X#H*)HGNH)HL^T<Su-e-rS^%9M!)KfL|7Fr{$KA;JYUBZe`0c(ZEf
z47NJkQIC<=yfCL|`mpf1O_D-ZH~QiOV|r2<UFBDN{IUCfU;eiqkpi%HL-!krL7Y)I
z)P+<hzqKCsOWupZX(c@CvwnrvsSDy-lHxR<p4e6Cli<x(IU!=R&L;F9npM$VX`Y@~
zJDvVip~_QD(kDJLpqa$KAp1COLhH3W8Ut&us4g2!W)ZFoa?>LRDkAT>p0%4;>U|BC
zD%wBE`pa6<3q%r~G!Uyg9#cY+hIWcKvy4i?KTQ0RK>AB?O0$Cb>f@<Z{p4Qa|M}Z9
z^55C*BEkWkgFHNSt-ofFHMON<KATX4bb<10p?{I<16uIo)CKH^PS&&zWosa?U4Juv
z`NA@=i&`f0U$$Uyo--SMRs0e~e+u<by|%e6ftk$#m%r14)Tx8WW)aBA5}g0R8$+<;
zmlM9jp<+t-Tf=7qGc&h=1sWYO_z3n3ureA((Sk?%<D&U5o~9@XNC#yM1Sobs9vPf2
z#c}__Sv-Y5uQb#SDIX5KCM;NRWk<>O`FU@p2^urX`-fSG1#_OykbJ&)97IcpJviVA
zHIOT8p&6s+aB)r*enT_04wgyJ5s=lEBHp}P((cmt^tMb-rv4uJcLm6PKGVn=fR~P0
z8qXH(o!eXtEZ{|axB)ALDR0KCj_I)G13cjdq72jo|E3y61<32k^QRt`j73^!J4_RU
zdNMX|xhcihV6VR&{`gGlzC(B-3)BQf)P)HD7%M^_E8|K9qrzIn2i{jQJHP(&7M0{t
zfD0z6;aSh>N|@T>2m5#`jg`{{i5J?+15oq#CU>5%{3hh=Qd|!7YaZ-_;R9^!e*OO2
zJq_tucqK6JsbW)%GJ5|1&OTFD(D|wQa4P*e*$%^&zR!Qz;};dej&Vi)=bi8A9-E4|
zsTY%z{0I5feCvz^&t=gI>KtTbLI2)9w&3UeeAs(;_@}W0fIyQ;9h68o_fyUo>1j)a
z=RfaTvv?2vLR_jf?5BmZ={R{AxufJ%NREtO|AECZ3+5EQct77&$`$IRB<Rs<ppo-(
z*)q!;3)(q09A9qYJ}1rFF6*q@n8kgL#6W5n+P?u%s9H3w)VcU@S0DSX%?~>e-BZG8
zLWyxWq)A$A(Ru@NCSv!n{N43X4Kx$+?^YA^A)5?ob=+;8K1*w$Br@F;tVJTM92~oa
z2Mo+{m;KQ^>RQgf>0oSoOvBE&3pXABnKA}Dy*SDSds8Azo!u06U&mP<dYPN>e;@0N
z4-pa6{IGwS1vQ_W5GcU#N5VC$Q8okB`2EeKc?xEO1K*)t&))Xp3D8qGj%o@UTd*aU
zQ@e2d8_wu|*f1npro@N%J@seWH!kZ^#a4Einc;tnUw|9&Ga%|r>7=u6GK}Z#Bg4MY
zdLBRaM>?u;OHV2>hW2x69v_I;*-nY)m4!Ehp2&urEw`y+nN0PDf>P7hP~W@AUA5FH
z4puZ^icb$d6Spur@x-SF670p_fvcp?);J5}`f+3-u^u+sj|0J0l2w_Iikly+n$aC@
zAQa;;nIWqkVy8Ur-Ga48VLh<$wENuLCujUzi7dz3-A~-Y_qUeH=4y4pBUTs64|<rs
zez-go#jy!{BSV{^hrIeF$B*Yz0W=Af=|?axkE&=Ph`{fN@YEn<XdUJo*Zf>ssmLQF
zn<s4(0@HBMeQrxQbzMlTKi}N4kixU40J@+MO?!k>>e^JIRXWc5iA5Zn$TsfvqwY%k
zJ2*Zt+!v0Qj2J#uOKZ<BC+siB%?Z=G`z#`^Iub?mT0M1&6vZ+n9XuE$qDjz5NR}I-
zN*Ptif{%AmzXuaJ^*TJ(3Hx;atx6tGrDvB)*8N5{C27DRM(A!|;2Sr)khOMJDJb|O
zuw6dgYWT{(%9;|zEFKly7iB@?A=gEd+tqfpXMz{i#&BWBYPv;~MTXhlAk$cg{@vU(
z%+C~$_T28pg@^-(u>)2;c^;Pb*t}2Q5T<mD03`Z!plN^dM`MC-D0y3Ny)GOk`(>B6
z=Gefqr6`BX<&P=qZ?}K7W*pb{&(tf+SW5rB)jVc*59Hm=vLGW5d*$pTf*zSbRx-G@
zVz&o`=g})_rx|DK=mFb!JiJgFr={WI9Fe-4Ox(7y*I*N#=GN5qR^f(|NdwyAmu8R)
zCK;RFw-bLdtxWpq-k9NPpYN}}+A_WpRH?OU6EURSjgq`W%D|ZtQ5U@&`>+V#OCs6O
zi}msBhs?`tjSk%1rJh%WS=hoHUs;f?NS5Y{mLP3k9XcBJlC!DGYuGx?Udr_TCABnk
zafZF=PA`QKSHHv;yXazTD)obSHwll5uEZ&$Ay?K|A~&e|H|wecUdzJY?9|)S6PxyC
zHkUqU90fb^UzT)Gj?E}iQ5UZMo80e}b)Q2@U6itL-`B=QLL1;dlunN!XBIYXvsXka
z-k(VgJuK8R`I(6%GQOUwvtUg1(0Xz9+Vt0ND*6Kh&<+nxq5R)X3sTWgBe8x_2h|za
z=iaFs18C^Hx|9MfN*0H}xXrr}-{n_Uf6{v1+KUY&lG@*E5#xvv#_KQEX@LEdXWRB)
zo=iyCYbWJ6tupeV;_8ZLtgFKE|K=+^`;}7$)F|>F*r+&)Q-+X=K2KX=YY}MBwzGjZ
z>^VlM7c&M?yZHQDUo!8ZIB1u#73J@~S58)ju9yJ$|MELq;@Oo4<&_Qmy;x1)Fh6J{
z$Hzzkv-!jClP6IQU^jj^x<yC$9InMOS<o1%i1D9#9fYrd?lk09_UZ)|biUtkV`77X
zRK86yxoyQJA%?6yMJ~n;B1I#!^YJhG*nB*yL~-cyGr;=IcMB78cNG44!EVqI(q*_f
z$*cOSDTVRi;gyl}A#e%n?Lyl(k25poLV9Xv7Mqm_JzczGJibZpP$gLkSg`-Vel`?P
zgEQ7`=*!$(Nu*!mZyvCKlIXOKU_c8gp&u`J2~d^4C90*Ww7$zzA8ewZat}9r06z{e
zU=B5ar+mptI#A!F2dqc;oHsI^_dO|xW0lHy6~H}`{sZR1!E*}XlwqNh?2@$Bpexp4
z!$*cSt@Kt^kLv2~DFgpkUX(1Jw7?<0jAl^4hk8cfq9*^<jFuERI0(^%UH()d-y3m(
zg<>)*O9lp%XUjq!^(;aX&;Vy;sKds1#XnM@H(L;dYi5{BPy;~$(r$b9NBtQa<eC^e
zg^Dg3E&og{xijQ<(XoW@Vd%^KSM2D!b2SK`sLolOmS)tYGz`m|0NbEYyE<G`HKAs}
z`OEeBr!BNxzRL_MBCdlssKF}pmZ8rhBvWDS&iKtjfr+XF%1wo7l0stYzC7P0t{s;w
za{15NbGO;QGL4tYf-e3>Ut_qW5LD35GnMzyE982|1{!5SN#7r&6n5Pw0Mw%0K-KQP
ze?Z57P;KUPJ7|WhpeSdv>ZM$1&dlolV8@inao9m}I-0|kJ~cs~*w3rd;7z4U;hTsM
z%^z0XddvXGC|Qjmg0UW~3Y+T`B2XY(Krt!5N+W*?wkWawI9Os*l3E+ddvHY;>bFC!
z&~z6M$RqhRW-=pX|HV}yiFrxs=HbJ=Abu+cC+dj;YjCwEloBbn)#rm16FgAeuRkTu
zE?*iq5bCX}-*CDT&Za5=uSRWT!d9FgS;4bHyMR_Ldo5AEI+O%cU-0M#t-BSvP-XPG
z-SDPc@`HbbpEm+Zw$rB_V#P{HrlEdOqG26~7dv?Q=oR>@+tFgizLD<}O?8xVJKFo0
z^Y>nB+(`<l{fUs);1Nn6mGMSu3*L;tKP{O?usc^*t^rsS)$PWl<1Vz_4>Pbbm?ESJ
zCvrnL56+hC33hY{f}_^`h4;N<$2^&>{>0nz!IfiHQJkPAfEOKhAyckP0}=x0Js*Vk
z))hln)Zbhph_3a-3g9cgmc)l}`j^mLF>rb|)*>0aC@ESu?3QgVcuB&avf;=MVM)<=
zdFwIltSM^TrIhZzR-I^~w)k6Njdix%t{yY_d5K^Q1$$wA&hfYrF@J+~yGYLm_OGQV
zSxP+=n!AtU83i;(Kn#HzKC3OqM1Qhy<}*;5P99Kv)A4|TummiG4iD>W1x4e&rs>@s
z^McP79{qc4Ur6`a6in<bk^LqrUv;ESx|RD`1PnK(a~7f{qf-4}-K0WL=P~WM;9>r1
zEgvbD8(o`C)j)yR+l%k4=TYM*$*gUjzqSq!7fpXNT89j7rttn&V}g>c{WSYPF|QSP
zoorDA(^V<%T1x-=7#I3*wQ=<Fr(KynJ>j(Qt89QkvaN>qK613e&cCsPo_^NmME%hL
zTt#s<Wa;raiwKIUwG?%eWxINyZ96)0KH%f3T|qciTdK>nUv#eo>b)gcG+S-wwwX0V
z;km1nv7W8LQ+SP*ziD20hU0w<2e;7tux*G~OU<zU_Yx6bq>Y)}J*033^kRRuYns6V
z=D3pHr%9;l8P9T0Jw<mN$n;WR?Xg|gnJKYnV`X(NZgAg(unA9rsl%;|K8ZvQ6nNUu
z014AN9Vm-r{>yrKES7i8Y5>1O$5-f=n}xx9fye`>zJzhV)w;U})mc^GCADyiPMHtQ
zReCu&v%~efjqR=gV!nE7J?fVcqfcUvozYmn7l24OiJS=tsee&RZ*1aRCm`wsf1tLJ
z?#}7H*7S?FyTWuY-cI&i`FSHa?Th8}W5l!m5T6q1dd2nmT{AivvdsU$(nUFn`y498
zk?ZC*rJrkLrFT*bsyNMXj29}{<3bHp@fEr051kZ8$Df{!-{&TF9ZR78{((fON)EvJ
z$9&7=;e<tkUy@(<k<nt?;<2>x-owG-^F*{rZM{vnC};-lqg$xGb#19;$2!#LXtz<C
zMnnq;?hjdqK6@K`S_$Xk6?HqJ&NMV00u)0(rZth&ejCDrCyK(wg~ar~44e7&GER#y
zcN~-jC(Y;^Z=M6{GJ}B;aCrs`9d)&%nfV99Ax>35cWaub<77>D(8z8-=FOIHRX*@N
zP}<4z#9%__!MC|N>MBANszT11B@AkAiYE`?>|2^~q#wkv)gc|DCs&$?COgytD&3Oo
z`qx{VKUO9PW_(42RlG%R<n)K1a5z<)Yim*5E>wdq#SbsC^@cysuRczqUNWXTukBn<
z^{`0)SfT1P*n!H##DvuE_Bg|#i|A};obgcsrFJ1YygPTvx;Zcimo|>g#_B!`l0Xks
zqEGY3jACm^%ePQ_e?NHI{BTLl(zJO>YJV2?p`OLXg9hd7Tem+)9aE{lOH{=-EMG9;
z&PTB}H*rOPk$IXtp`02+<*aWPR1<@{?V|<Tykq+Y3nX^@a=PWW1<xQ{U_?jJ*G!P#
zvDUZtWT7_^j?}#`$7W2c8($Wk^X#WGblBEX?WoBv%Kb?`qG*#59oa|-nJh@VxF4w*
zXp|kVcAx%|z`pC>FV-mE_$^Jb4m?cSgV6E{+!|DYtT=N>d;du&-tSZ>FL-I^-rfT;
zKOEuVKruCi>Zkd3{-?Z<_uX*A+Gh6pl2gpwgS_0l4E;}_dhbk!i6V&VV>np_6cP^Y
zVooNasB(DEd9(y>qXQ}s?gY~3qZ}k0MYf%0hoBLariA`5#mM!m!@I;Dx%8!?`(J{^
z>u-6uq3C#kY|plWYdqW9i?tJr9)tB{L3*O?SXifiYKAf0NRJ8ui$WF<q<9XML1T`3
zEbHF4$K4*6%b`G*YjZ-!&SVbk@X5>!H%dyY$(31Abf}4_^q&{t&bde@fm&wgxv3C4
z1FGBU=yLQjn<s@6`$b>C12J;w3!H~_j;!06Lk4msS;&hR<b=jXV>_XExBD{=c_+(&
zLV{Kv425ZFHjP-UY!XCc^yQm~r9k2ljuF1&h|T@%eAugcX%T6^4czzF<JCZqz<yE;
z8v1?^PZ6=3yOV#X+6-i%KIS{<%7)^z!gdX`%>YSnIlLAc>d0JKh(@jp-&R=fS4AcO
zw1%WCMLlhd(|K%GmP09!m+o&4H}m2ZUeVEoD$z?yNg)F*^c6XSYf!Em&{PI}&`}<A
z0%%!~#8D{%jO@U7%%cU6FeOQ-Zda`!bd{m}UNp5)X9E}`uFWXdJ4-&HO%yK9F)wkD
z+qW$~ZPa|Mu@>Z8cG^_pDPetI9@3PIJPPHK=7NXlBC8hD6U~L<LhNmCcxIqM?UX||
z3$@UIWQZObE-&}PQXi_f6})vg+eH;_hN(12AeII1NaX%<6;9W&m*#AHl&*6gsA*mF
z?B;H`;;!=zcY-^#*{}DjeA$ZpzjW@VoX6D9wu;FJ$^*%Ir??)<F1%EtULx)IQIRh<
z@|_Q^s_%_>I&6|S)ni)sYFDRcUki@)Lz~}VAl|u`sJV&o?XZx$D-|F@0l#I9SBtIo
zBR+U@P38$laJRMSk8^9zJ_iv>0U0YkZxhs6>ZfP}*<c8rycudb-MJepr7q-eeBYM-
zjyi?U;T4=&hJ)0XrZ^x(8b8L-QwW{A9;qEF!T`Yl2YQ>`1}LVF;(SvIyOCpsgOA5<
zgziloWKxn-y1%ZP9HAd<(ayW&qr~rBfoGeA<i>>hWD%(r`ppp?5S;PsH^ndam{J-N
zf#RXT;+7YfRl)IsSdP#PzZco}gUp}S%{t<Dm#>MSJsS)<h8AQcHvZGxN0$PZeS3y$
z2l=#EEl1{MR~<G`R6{_K`fWm<SX^9eVQKl~t75hYlP);6o5C+6mD3EAoZOwlZVX-u
zKqnM%Hva<jdt`Rz0A1+4u?<2bW5_pZeT8U-pl{CV?{y?QCIqU`oo35qWZ*uOmws9e
z;Ql2_60%m?s+7${7&knDq%&1v$lZcX{L59IHV5p7#w&8~vfy)xM=N~9gsEDGNV#tN
z32)85eRF8}$F2e$6`hl)+syGJDtRwA9G+aWW`FFu?4{0HfK#-a+Hd_XapO3%pS|{#
zvrGI3H)Ay*h^i)|xfd4|!O=4_o2>i{pYu9w5j?aqH(yHB&|iLhfaF+~IkT?@^sNRn
z1(T)8JwBD+aB*=dsj9|1FBXnWjiv113q<cD-~g7EN=>HP7W(fi$Jm?xk&;xB#$H!N
zX^qY9LheFG$5mc(-k%b*necPjm(SrUjFyWGjX(K12Del4(P#e@RI}Es3K7Xgrd`w(
zoN*B5;^wKfUH5Pyqwuia+8V~)#V^C)k=mg1n<vk}Mz_?eH{|Y=&}JW(@Ivc&HksL;
z!WeDVDTuBqU_q`ksjgyB8@2aM_mTP^9a9lG@1b&3afZj6k1p!GdL|_8eOCL)EN+}L
znez|~TD9Ykp(!6WJ@=CPF|R3~m-8t0cw%i2R;Xfq2WoTFU<od%G?jVBzP;T$Kd$R#
zuh5W?kW7SM0s-%+M;OW#Wd<iro-A~AEul45Re^vBOpah}P3dQaWZd3*1Kt^ryYtD0
z!`7P=HlruTU6IQb9ayu*rBNijC%31=hv$nCQn`W;gG2QA+&#BoFnaSA)uV?ZCV7VP
znIcYvmoVJe2Ww0o&pIe7m|QF>ElMpR%6#1!`zDnFbvIYw`c!hiYsWAb^FiN=SKX!O
z6zh8^|I%d~sr_w_@WH6QkK4s$^z_??$i^djDLwZw+ryad)n5&Y>=w)(4;0~&GnMg4
z5uG0&PI7AV$_n}ipy^3T$cTzU(iA>iymR&lNJvP48L6tent!*X+BJIa?(WbpyB+9!
zf&}W%_96{)D=`=L?aJe__9wqL<w*uKGS_}!((DNFC@w7hLHmsNlX<i|lfvNj>xp5l
z-u%v8^B-kaYMEyxb+=xa?aB}OKw+%eIDop)-rk-$OQ7$*YzyN*A5*<rrFiK1g2xyp
z`3mqk?YTlauthUM$vw^MVSs(A407-3=;Xw}!lKZ6{GzvPK-LAFOzPD$Z{l>o@1d=o
zd^I3ZiWBb5Coc^cJ1mUKLk~(UZh9gHgoQMIc!hPYN~Q%^Q8F_}MMkomjxaC?-y|cG
z>=J`m>qLGitLRI)x_0~eJdfxM2d-{n!jH5r@_}WsZ4A;p)Shgn)KYh1aL~V>I{djH
zdgF*uA@5-d%CN)69~BiPXl7wPy2}G7F>4+Jq@uwzb^}-&jrU=IjzH*pt=XmL^c1Nj
z9j2dW=Al*bo7S%!irSTRCns>bxO@L-q0;sg*q%ipE3}#cOdKI0$i&0}km|WPT@Fsc
zAr#8g)Ko!~PMMvw#(}1UZ{NOU&I)^7%<YSKcu=ss{xvpM)9qhhQ4t%6E#h7e+rQM7
z;3K6KZaKZbY&W8qb9-l<9mM+r>&j_+LwuRg`e8z!@5Q67OdRp_hX)zpmb&A~9mb^~
zc3=Ap?W!qSdukKaerp4C4V47ItU31G3i7Lh`Y8FqX#8l0wr_XR-M6J<P8SvzURJI$
z?jksS)xgh#Pk%MWJli#JteV2_`xGV9gK%@+1JmPsuvh8qa*((n#mku;haAUFsjT}h
z2|zj$vzPEh4cSCeW#sAvR2@KfH5&zOdcYKM|A^$<N$9AD#{S|EWt7;sc$rTd6JHff
z8AaQo=?t~Gb86TSqWOMxcf_Ctdq;2WT%6yF-o`;G0bd>2YgSZ2`?otM0z*^^psK2>
ztwo5a)kH~7QPG$CJGab^(z3DxgA0Ihsq~){DoK?;yPF0ZVh&6_HP#^-5^j3-94#bU
zq^={hX{H|7Z@2p!UYfjSG;8_Lb)YB5L4P+zUzbMTTt`H%)$v#d?V~V)0nIOexCg$_
zq8UN+V#1povaSK;!dRcSm**#)MQ{|sCoHq@H{-#%>qcGxA6lpnYxb?fyt7IIGK41-
zhKCHd!gEs|R&I_?X!+_BBXZ>zo{T?3CKC?G_}1)m$O<(5(^<~hhiH;@pR|oCinh$&
zEP9{H!-ZgyOyr2j#{)^m8qoMM3|Dy>U(S5tSL5Y{=2`*JMe4PhYBQU0(UjjPMc#1y
z5Mb~20shvPNq63F(w^Y!OEQ)%Og~cef`keeJsQ7;W!t*4;yIT^TSK6%dK`CvJq*-(
zl+JFiz5Uorc(4Y0;M|&WVGI~;uQdoAj~g61eJ)=W@p7#=-W(TeTTnS?zr5@FIx%Qg
zf9nqDR%MMJ+PqGB>Eg4qhu3=JAGu;-c{xe|9+QF3xf;3mkr<pUcsb-l9RqOkHksqS
z@)?CNA--MUgAOutuRF=AeYf}&{%TLLLXPvhY}Z}Yn{FpJk@k?QVk2kDTX*U+>#o~%
zEu+h9gylVbpa@Ouw)6Np%iM-vXtz?yIwzf%gI2~EI*k1^-*hVZe%c;JPZ-+)$f8FZ
zK1~!$RT5WIl#v2{x<}AH7(wlxM5_A}-zvljE;fM8V+cBYIE?}F_s!enr|z7boLAe5
zezLZ<EHT0tgxS|)1>mE0^oFw$xvNfAU-&leN!<?0P*Tv!f7qzR_Y=_Ivw&RUZbUU#
zOHa67&q_kPT*qJWHR+^Mf^QYVYWH9|UN$kT(+N<Uuk>WCE-ydTEHO)=UiKLRLVH?j
zf6C2Xjq7eTl_xAG91veIr13h&*O>+RM9B_F;kPMiMK6W8Nnj#$7gvi-NmU3p(3|n2
z+6E<}X&6MAB-oDGUncPdjCDC)v<XdvFaeNuw^iU--#~otE;}gh+gv8~y1$AAqK0-P
zBqZ2xq?$P{`oJ7I@ig)p-vR<1OHPHIV%gQ|{)_WnG&}2%;Q@cT$&RFXL}@GHfO7Xm
zE?j&&<F;oVkC9I+TuekIJf72XK1&b~Ros{J0}nC`=m4%S;o>R0TQLQ9r&cQY?CuUa
zU$V7$;FN0DmkW`w>`By6wvUAx%<q-~wj$lTrgCH_vf;ANB&Ka0#evv*BDSiyLUs6M
z+3^3-V5@s~K|tx#`JG2CV`-=7_-ZC2gOsIdj{|sAeW(XKW@F#aMSf(Y(|X5m9(V@<
zQeYsUl$IxT_><E2VzfSOt#mL8OGt|A`pui!Sy*rIURx^Jn_;H(XD&N_%Y_)uNS4~f
zwRk9=*do)!u%68MOy813AdBt!jogn}6%r;DKKtQ_y#3Lu7Nd57N9l0X_F+xmjd^c2
z-p}g{0^p(I(IXxn|FU#9=LXb)s{K<MJKKP9OsFgUPr#(Nx@V`~Fc~}t;Ev^0kv359
z%k(lbEL9o!M%Pkgbl$|se_UqdEY#9MV%&8*P!*_QrBa~ih||cPVQz0a`1bKwfVge~
znj+paKlsgX5bsGeQ;46f;rSWO{0gxaw~tJP!zTyF;J)e{H+;N^i4jdz$nZ-ZD!K@#
zLvWh0h!gTCxkprR`9K~3z?!oaPT4>s^U*FMt?Ti;B?P8eCJM|T;GYF@`xxybiNC`V
z_CIa);yxdr{ZTq4%O@Nap#S#MhDkxAaBX1bY3oI=>|_7^CQYuc_{Vew&6Ig@_I!wx
zC679{SDwu<f~UjAWckuSxq+Cmyf8Q3IQ5Fd)4nggECUW@dH{cT_D8d?EmaE3Sn+oJ
zl389|9KmCoZO31YnRGFwPc`x@!Mjx9tGu79mY}cepeCP<TPVRAA5tKoO#f%zt3d(8
z;Kx>QQj2i*PHnoyk3u>~O)NP=AKy`e1d;b%k3y}I?yD@r^T_W^-X^rbtf&B5@7aL!
zi5DK{Qbi7is1JrPGM69A{Ow4)p|R6$K&PKF(O3^Fqim}>U93Uloe<;fftkjyVkFTf
z$pD~a(qwZ$-WAallYh%i{d_GM@3+?)oK%Jgt!C_C9H3Xbxc})gjC7CL6s`~6WSP!T
zl|PI4dB5I@Ir|{E&srV_c*}s^UpAG9k4m>b7fYNW^k-`GNj2IBCjI(Zy$un;D1rme
zgg7(up~8oI8QIXYX|rh%r4NAR)Enb(0t{K^C(=|A5Kp<?P`-E6Wp8E>^VMtYqT573
zs|2bd9D8@Qew9c(x_2J+*^~Os!;AYX)|TO2i1(LA+_|$#<}jZ-nt;&(>|YPX7w@rq
zyzf_R#<X6^x*%=SIk8lH2#9mWUvh*{PDZ{8JdLn48o{q?sSbYZB5nwsvS`e~-k!Kd
zn(w~>l(ej?i;e+8QGT$Iq)3L!lh7A$hTn(%G?JitUmOyio&spg_^mt<=m(9_zX>Rs
zCT9Ro2L_)vJ<xs#(b5U*oX|FQ!W}511WNaHW90fWDLT%*pAtX^QKq>XiFj48K+C+6
zArVK;eyKO~x$LZTlq6t(ZEgFRt>7>D<l3ae;=8>j<=Kn~!BIYKN&}w$^v@k4xvNn%
z(d8i4jQ+xU{rB{H=79HK;MJdCvuMIQWLaDGyFseVr_a!TnoWi%EXso<Aw!GN`zV*M
z*c_dHe{6Cr>Rlk%davw~y<UrwHoiRv7=VC>tv$H`*%Ls(b;+}VSOby~kXpJYpWmDk
zXWb4jsv78Z|DMclo%A!_GoeF-Kg?E#4olVAzBtU*kPw0b`+wH$-Sk^Pw&ysB>`1w}
z^*VyN*qb~g5roq&Esw)Rwwzu19Ne#9Q*7=pLS78qc%8ySdzQ~)?%F-gpg*1f+H7+(
z+JV7Po`J9&;6!G*UN+095(DL7{qSn`eC!fk<X8@xlHZKniGjUF-(1dA=ylA;kR+i%
z2$8xkP@^AC%EP3My+ln#J&*k;u-oUw`0WpK?DuTj=VSVwb^v|fyJxY_HaaPU`T$GN
z#kE~z>*bW8KqycM|2!!0n<@spnocCWeZw0;!`Y)JtAdrPwoQDfH5B>?=$1PAOP|*(
z%Hk-yqdoyX8d&3Q<Z*nUJ+CO4ugFuDy%sC*jq);1h8FUJ8|vA>CqTw-z6(v(S;z&j
z-0jbvd44qVAR^^6&IA6+tNXkBib}b<{pcLQNRzyHL1*A5M$vLoB;5*d3VHzPP$?E`
zQ5aR0Gu`h9{1NQJ2KeX9Aa?{G<F?4%FbjekF6nqdNvT*6J$Rt<U2Sbj(MEb`B7ANS
z)tii0%~-rOZt@435`1@}ET8lFyuHLuB^D5my48-xkO8%i@1bn>qS4FeoJSwL``hqr
zzaA@S*<;YoQ3-2~pOmA$`Q#K*?Q)m2K>_%AFVO!)mWvnAGXAuHJ{rhxVMji8@&q6y
zmTYc&yD%ZV$R`SwllO-d&@MCNH;W}n1bn#|(exkti@soQ9?l2)=3+0!i#EjB^I-sz
zkErI0f!!uLo!p~cyraRtHrEY6@U!-a(Q9+y&FWp&$U@Y4Wq+gV{t1)oC(OI8O5pNZ
z2>2njK}*-+fq;R9b$%CeJYi>$SD8;f#DJc^VqOS1<$6T`Nvhdnk&{_eAL#irZa&(D
zRO)$ewhN(M-+i<pQ{!RYgc|Sn$6hBvs{M25qAi<{b$+xKCDGUoeg1~QR{-FQ-Z;fV
zvFU~A;yU0sU)U`}$@Zf5FbNiN=tr>jvf*sj0xUqwM<2HmiM-eSoNxg(CSP4bBJykx
zg9W_n3CJj06R@f2VDXHHc`JINv4~_pfzLlFG>}yFNcAvK2;Rs@;au99V@L@N7k)-O
zBh>6F@I9yT#9Zv2YY9iJez;PoxZ74AIBn{{yED8M*bw4PbUFZkvFA-?)B@!V{nx9T
zUs&N4LFI`)Ox4X<^=cn`jSpcL{0@IOB6IrgC$9bb<AqJF`DA4zr38TX;R8PU&jZbS
zEe9gzbM~#bBYN+z_R-(_T&fW-6o<fy+@@KD83TW33q;4;HN*xibe0~L7o!IAUF25-
zB^{yuU8g8O+?`i&d7(e$oA%N)w0j%?dI7dlpb@sBaFRqwPqO)HIk;iiy0uVB*xMpu
zBoN7x2Uu+ZjF!4GlF$KK+|;%IeL=~^OGn){AcJeG9tP)>9kTI6skZ1FW}{_GUm{6h
zR1R~Bm5W<DB_T*|<<7DQkU3lpAwJ|Fwb+irdc6U_;n4P_FamI!6LFIihj)Jjy7Bpw
zVU8iByr9So`~kE%`2>uE4im^NDPX<<loA#^xd~+*WqB!ATEHzzXutLu;NB@p;H#Xw
zq{P|s&QIk?sAY$+;^`{O_{ZJ`S=gTrM`JC}YJ2BHbtd!Cc@?V@8rBDNu|VaCL5<9_
z1iP;nV5q}BGY$%g+zGWy751LQfic2i&~EYsZ|wQ^u12=$gEG1r4R_j1wt@A&g5`&$
zy^{cT*>wGByU%NTstZ^UZU3z61pIa+Ol3)W3F+RY)Sn2}5J_e)DyC+eiBnp+Qzp4N
zUqtGPe5cm@bg=50p=x9~Y+(mvBjC#mP@r{VUSjjIye2j)&{J6e%o_ws`SuNc>dMP1
zepm2Eoaz{I?q?bqmGnO?Bo!F`Q{}08=_9p^qIZtD`WoDDN>c^WB;1n%!9MlFG|ep(
zaDXwcng;N!b9K52Z+*st3KeJXpd{O6f1!jgDZ;kQzszto>$n|%@42Lf-y#r^-`J&h
zz{<jEa`v{{elN-lR7-VsU@Fp#zZ1_|*CHi`(u@z(YeWALt$j?9i?1)2d<vcpS@udI
zQk)JY?Vn54DVvp6J81_2h`Wgl<s?~ZUB{6_tcUeu;VUur`oDWc*n8v4Z5`Df$hchT
zaulHqz27C-_W1?p=(Sd`=PklAL%95??{CM*2S`5Q=y){0o1r){XuLIC<|#_DUIAKr
zraK+wjt7|<oIqEg(AEdC3y7ycb{-opLz6PztUqBkq{bR8mKp9dY%`LWsgscbCA|0x
zbyULw0oyUHXnqDLW=UdJt>ci&XZnlpJzbGI6a?}Hy{yAmPE`I$>^i5*fNv$5Dq{KC
zkuVFg%P)m{)Ma`(`dV`mH9He66c3*5q@?5G5N+?Rj~a8#+?m|OfG&O{Uu%(Q#w<xy
zN7F%b9sGS&$aOL!q2I3Eqs5%gZ@v9)(fMicXnQ2zDjU#IK)bJmFFROl6d$$?x+d%x
zB9xa>AQ{kL21w3;tv(XKab)KEf8k=~KWsrOYOu1j7f+s`1u$TJLyNu|AR_xH;ExM&
z_N_R!#RF9Q?+8m}frdYA+x7g<a5B`NVn>GHdLrk#6{y7}W_K!2X0?jyEan0&?A7{P
z0aQ2XwT^P>4jLafiUV;rl`GNEb~ouaw#4O&r-}<Vv%5F-+gM9M!<iC%%os7{nBs>?
zGh>L@<?teo013{+>i84*PlNe()3cw7SD>i3o$YK;q`xFQ7bpY){+35WL9I?Y^KZMc
zHtfK;a;%VuYc9J_jzB}$ijNd`pqf}gPVwrSgWzNMB|(HKkT=_7q{VvyPvdN6u2)+o
z)l24|PHsZ&`QJ-Xj|D5UP|j9TNQN*rjP5U<{D|gySh8UUJ5hd+=DSs;8*v+JyRiBI
z0MdV71VK!o;7I@g_BCO{b=1|-nr9f@h3~IA!dquFp{kI`z4s#LZOOeXdxS<?L4FT5
zI3hq3dh^V38cMkY2^EXM0D21Sz}Et1ay#*->l#8Buv>8=x$ZBHUWVC*vj^MT@E>bt
zX!u>gE-b1L1osz3dw=0dLHj%}PpmbSnhpb##-Ys6)^+1h`-^&!`Fp_apNIdS$+b>!
zr7xcU!E5q4Eh3CHba?RT2KoHl*|Gwfp$FNm38wCYNb?6V;JXu(!nYJ>MEoz+N*^N3
zRTI<>nbGo1FYDDW64H@1k`a!&S`_Pz+)6o54^_>JtT5Vl?x#x`9W4yD;-Jk01*+9K
zMK&aROYIg$W#9+&NTJ;$L**77(+(H`5;{6hm*^b(^H5mPa(Be?X|<t?u!JIMryAt$
zHQ=2t6a>8eHPZjth<$znNn7$T4#9o7Od4oSKL>kR*h{+-o{Z+ae$Mm`8Rsec^@=h7
zohWP0R}F`qwEGkog;%`ft7nng`|&>mGz${j?)s;ziUQnK6OJCL-4>UX6Q>cnA*!QZ
zc0sW?+Rr@`X}>bzzj$k+JlS}m2HTvOtlB9nD{HcvRC6A8!K<V!d5Zcj@fA%?EY5RT
z&qEi}gTol!7c?du@k3Wa*fG4EuvXX{h|g(bz+-S8c0p-#Zd+^b8@n0^-ZHX;V;ZEg
zvEj%ic)y#n<cSuaj(%#;8by{}Vm#k(pxV-^DPMxJWQ^#n^!ta%PdHtlxm3rS;P+r(
ze1X6*WHibk9N|Q2_h1}xRLY>s;IL)(Iq;<zho8{9zP?P-tK*k=9^biMjIVC&el~!-
zT*;^9*lTA;><o5u1Do~CWqiYzy5!@HEqI3P`SrVcTrCXsPO6)qB7=%6D#J%wO((=%
zakGnf8&mGIXWp&G_~WjLvnQD`z5?C$NI2=Gsi1|`MQk+;RV{ek+~2H%HAmw5nJ8zt
z5kcfWJF|2Ua8tiIWR9K!Hwhe(W=5yTNQ9+wA;LwBi|fggk?WIh?WL8L+<d_2s--}!
za(Z4hftX5(_hrPXkr}FEpceYQ?Af=uy>$9A?$1Tv8j8M>;clOcnYs~34P+05UB$!y
zO1qZdJrllJzz2r^GnE?3_;@ACpv<$Db*mk^jgV27H&ta8sHIU?8v^mJA2-S3y!PO`
zM{#97Y5=Egwfud@Z+XWHQc9CftjCG-P+)h#s|DY9tW2`TPBEZiWO9c1r^Fs$joRWh
z(3IS5s$khoDau>m@kHP`8?q4}qruFMuBcLdVY-oJLu!fI#7re#S5_+OdtZeL5Uq0Y
zxG-CO`=^(qRK|z0-yJkuN+G%aiaQs{g36Z?ec8#RC{dDUBX~iwajS{tc<$J-^|4)L
z>?>$;Xzf*c^yV7>^T!cwt^d2#&ZnbJn&)%RM<(|L&%f%U93a0&@BhI>zdEJ3eS0*b
z!u@%Yf~AA=$0ltqlMAM|>oA0`#&x(*Ru0Y$L6fTl6JItOzv5~i&-xAqi)^yR-xTIs
z%<(C`pCz#NaQiO1lVvy&s2o$;CF`P$<kfl|*CJPsBWMEnUK{YLn~LM5r*0%;C?t1a
zO-61=OkwukUo9C)i=+xhpNR<NuRgJ6&5Zs<sr+ew$p6feaBKNev$u#_V(OPK*^7+F
zrICbz5X=7@5Xfp(#+Tn$gmpC{qdI(+6E_NA4QeW)E`J+H#a&T|Xk#JLYhb-`6yl?M
z-lZq$Y!*<8^t~J1!)$?bENSvbQD5q3HY%Lw&0|5D#7e(W1^bqpo|H5(ULI%SkuFqZ
zr;B<`9a!qi1qQ>)yn)V&JgzZ~M@NPs;XZ;0ZSA1|$zHL4ZKwGn0jSlN;vwS}j+ezF
zrOmcPWFHHNZ?ka&k+*VXN_fT57b|)i80whWKi))c9Qf?wh&q8~kIG&<3qfZM1J7yT
zFobvu5pr?=ERsk|UU<r%(H$mNeWa5+@RXo&D-REH3+=dN3v*WO)q~5$S`k`mq`x#o
z%pj0dGeSBvIDOJ^6XS9Se=9C8Lu03ZlqY*2)dba%IWi=TJXh=0QwrChh?l(<5q*(G
zgrP6>RHfQ|wVe)XYUL%(Lt>|QOP4rz-ns1n{4Ih1{w@GpTQ)KjT|$jxdeV7k_qxoJ
za1|Za-9Fo4<S*!wSbMt*Dj04=tBo|xy?)bH1ik4dBgq>=e&~H-cv9*lz$VZg3BDzb
zETh%T!uG(b%hF6k>BRh*ogSnJ_UlHb|Ff0l{{<t`Q|$GlaSHrG3i9Z{Mah7|sN|1X
zQkN1;x+LM>%i=ogTV!pt)HYG93fx$3s%6!&PXm&le}#C>l0=;bYv4i`CATZZMegQx
zY-~?DDe#pu;tD_w<<#wxkUFo53#ULojKJju%^ebtn<}$C!5<=PRRnhGn-(UNB`lVU
zUBpM?0x!6icGkOo6z982h3&5pt#$k)8P-rcP%IE@A)~;`ICw+ToE$v#)vU`W*+n7t
zag}i9-#l;rbD(E9?j-kt<80;~Up-m8-?q|t>*$joJ&YzDB#X}4k^7bV9oz{f1QK6~
zvcfd(T=(%dz7^SH3SgDZ$oyWO!`EOI*6(RupB?vbe?|Uk_P65XAY4qRL<M@{m5H^d
zSMc5J1$;1($7^NuqZK1fWy8gga~^jQGM*pm$Z1$Rj8=klJ+5h<n{Oqfz648*AN7dQ
z+tt{0WC9NH4;51>cRh)S*pA;=EC0k3HWixwBaDVABxPpETwnVXZ5{+1U0{Dju#G?_
z5S|s)m5`S)3GJTJ-__SH!nX}mNvz%H?!je2MaFFP-CATWyEvD4K$Z`jh8C%STbiTn
zRbRl77#J+v?|W6AmourN69oHXy0atPCm>7fyqmCW+fV^Jl^0!Na&;hz?l4+T-)GnA
z;kg!Lm{Mu9x}xMrd^5B|nrkFzWl&OU8Dy~ZPv%tP0Qa@$rJHdc%n<33)cm<`x_Ft1
zKbR~q{rS-KN3chZO}&Y^j6abn;>Wu>e+kh&lS`9$0hTt`*FysuK5lpKi?GXb%TPm2
zt_o-stfVduFlz0;<HqO7V+XlKD;sD^-$Pe}Frpp3OZ2}96$I@Gr{e59fkb>Rgg=_2
zoci$Xz7t#=<l#?5Y*SRQKjcVnRTL(F5+aRUlwF?ys``i$ZwLA6g<TXb{gs!O$v1Z0
z?>4r@z%JO&v`t5PJjiRgtSGIy<k?6O<@e`YY+hiF&0_<%-NUFqgE5Qyi(Pd6kExUP
zlAre%qUdFvf22H0Be!y`@v&zu+YoGQ%|05{<+C|&iL_PkOrqFY54@<Cz=I)v5^{v>
z%-cr(Rthq<1?;5F>ecvj-J`Yf=$LW2PxtOK!UYIY?j1K&%f&*MWpEr;W028pe?(R=
zhi)b-+3RqM74d_6<FO>Q<&E9xPU#V$N3NjXoLV_KWVlMQXz#x2TS6i?q&dvl6Z*4Z
zm47qIVEN$i4HEdz&9lDz7eVS-su)saLAAMl)=t-j)MjsOEl?!a!6Q`iExI?;?1kDo
z*A39$+9#^fnQ=>lg7)v3B?aT=gndNc0GoBQdk3k{DGii)ry%IY;t77Od)Hs2R!S^O
z49!zbYz`DslZzPLPV4F^FL+g6(@pKJ7)HwOk7PIhRzhBQx{3400Bde*5q``YyW8AA
zG$PXAp;YWY{D9IE{x4N<rj^ev`-a>c`C4lOL-Bm1w61yP3Y?*&qyHUWXqLTFTdsMG
zUF}6e=JAL1@NOZ&B^s+yLYH-Jo02Ksu}rYEwt!ol?S;nM?B+wyg9gr~CLB(ScW$Bo
zu@L5e(3R>eFn1?Z$Wr<weM46+&~coiwh3Z+aBx_P-^V<Jt)Aa?goe$xS{NV0O$Hk_
zZ_CLiFYb&oto#hoQoF%|A)m5(l4T3I?U%oKgn>Bb!i4n~%Bt#7ieVJK%3>HPLN9nJ
zVq9Y;@e~%Q1~Z#31RM@-XWS0B(xoo4-UD<ar|K(4`gn*&EH*TdDx6#pCs1bL%OgCA
zJ0p6i(fq;LUT-RdLYA^1<Jw<~;dRLJ!m2)0L+?Xezqf(eG<#Epb!X)eIMlvdNU97F
zDN17;mg93+jx`E=g(4elc5F5%&HcIsmwEEBqLxJ?`i&N{`}?OT2^x>S)2fZYEN7u(
zU3af8-oo%iR**NmK4hhFem9&39^|dEIBCLakmBAt6Yy@o+n>n;zH|@hGN!vX#f2<I
z=B{Mbp^Q}JDFX?!tXmh@On<hwr{KTEs{$y~CpxqC4VBI3&FLQu#=(VXWG_;UyO=V%
zx+Wh`W@Xjd&y&ZiE(BB%8qNvz-UGW?Xw~9j3+Md-3m5*lgeUjbXVLFs`Yo~-Smy!1
zMR4}MUH9PgwzbMmX<_xB7@1Jc3Bu(g{m}gEFgaw-S&}yLS_A?!D4^|*a8`?te%lxj
z`mB$bvW3{jiowDP7F*|vjcX^6oeDFEURYXc*fK;IB6=!7X3c<9jEs!J*sc0HrzY+P
zy93z0ZhnN)OOaOYz&r3uvr6p)A}isza}<UEy`_H(9@k!x;VuMXHVKN-J-Nli;P*=~
zK7Qoi;;8?IUS~uQZFE=Mztk_3Lh#fU>#S_B2%rdH_r~XfE}Gpa(^w0XzV1jjN?J--
z{in-SY~9G$d?f#itE-HPs{8shNQZQ(hzKGn-AYR5&>`L3-6bI1ozmT%(%s$NUBi1p
zpZ|yVgKOcM1vB^DefIv<x#O`@cKW%PL30gV26XJl5}Fr-e94TrkJNHF?Na`LD7POF
zR<YWzdSlF9_{q--=?JV2AjoP08Apj%#~P1Nx=I2@R%I#~li-B@Fy^@xci0m?93C2F
zU_@~iJFTN$bffNwAG{BPpoJ4766@Z29leu+buhy#K5Z}qt39vEWNXWiy7>5E3Jyu^
z%scU(X?goP--4-bUP$Q+b-#XZvbhDA48*YNcTN;6{al}G)YFXLkzelmU1?ga1ttwi
ziBo*DZRI0?FC8eMO$Z|0?msn9QF6>5^tct_a1VGT)nB2NRw2866wIBr>_uX!uAS=|
zR`(xD^M8z{s=H4VcF?ofj~?L`to^K9w5=y4U)XveRBXbO)jj8gvwW5R{Yc`>AB3r?
ziUc62ND>+p`F(+Us~vwXm-~qxQ)0*1c(-G6qt(TAshO`F!Bd5_enq-BKKG1geUwGT
z3GOGX8VmQd(F;xF1SxnEckj)>A8yo;IWf>^oD{&DvNA2pi(tIJ6zU(9s&6m6r<H9I
zY@zg!w30KIbxsmK`T#=b{OI+HWngcD^^oZTl=(_Dz|ZItXH0&xlX*!L)lL6f-YT&Z
z<O?%Pd3y~sPUyzT+xmAz#epVlt9PQ>uB)S-Y0$~)xpO#`YyPpZzug)Fmri;-uRzK|
zwe1z(T6;P}!=}TV-BObw?TY_|m;F?xlXopWaiEH-)^4xbKy+B8+p!sNJTv-9lY;}A
zEM#te=y*r4PXM6FI8wbQCiWiEqcB*_-EYUzpd+1O&F9h#sQC5Y6?xdl@c@&X1FiyN
zi3@Hxw?MtPsJGbLUshA#El^X`4(BvtW;AM8rNm~659mkW6pKcjqnUzyD-}^yl`l6`
zXh-vCRSFq`J@ofsNXm;nZO8VqN>Xt2M;4LLSn&Xy6a6}BLa?u)vb1IdkWQ0(&8j_$
z2MZ#+`PmBQPS_&_tgfQldEB(w#I|EqYm%obBTvg(f}^>DJFJ?VQ8-bnmGswnHT37r
zjsV!4*!$odYnuIri4e}Yy*=Oxo{b4Y4_+Fi-jP$&V*GUV0^ju6{@ZxrZrX5GkM5+c
zjDH9faNBGfR-?sc)PZ1OL3v)*ksqc54Y8SoLUh9^#=^rb*^OZJ5>Isj{-EHI`Xe>?
z(ObU!uFU)7{;~868iSD>?Lvegt0F9^Zo#_aBZA@=xKx%1La?)Y;{D9!u3*(kqVNa@
ztMtqyhUQr0!)BNw;pf5P^6UMg{H9icx9A3TCOnakZ>PK)qt-VLtkbZCDCpvTQmfja
zb<hwL%zfdEf^ug~bE43n;`J2}m#|NhshY2)$6w<1BO%7+a{au$z@9tkdZ0XJ6H**j
z-ePqfTO0zD9J-il{<Y%r=f>YzZvHm1@TK+Gq(3wM`U8=vminK>q$M5I)TA++EDgo8
z9nm5!$%6nG-j|MU$Ype{x0(*hy#<?VlwK|ul#%4g^|s^GIyErfj~3v$&1Nz8R8KHP
z+-e6IMbiyUpT2Pwqiav}JJcmwbPIrt^h+x1ORwrRT3vwXxz2>nx$Rgr8k;(J%URa4
zu5J4!GBLg%IFSK2NYpmh;x4e8@^wabbnVVk7Ax-5zaN2k%&yBu%@L)?E-RI@2NlWA
zQ!7R87&yasjPcH|hm||#8J~123I2!52gq83*(9lT`iOBOs`b$D6Var?Rlch^GW~6R
z3>HT3EXOe|PWQmCCg-xI8TUPJ#agaq-0qIi4jzq=bY}&=Ofjd8?*i}*{R>^vC{hl+
z$&kUw{55&`*t##kCgj&fHa2jFF&j0V@db9-20u6;p>F0?BcG;790`%L1&>)L5$0y@
z+Pqi&5}MvZFc@xXm{lZ^^O4KHuifgW9+`SuGk@q$a<?rDuk}Oo3aO}gf4!<UI?|p`
zg_6Q2UarxvN%~^`K}8N>uV<+NJtNTp?v`#ZpYmtD(RX|QvA4kA!75Je^4aImdsmy^
zp5s;qMy(cORr4UoEMFNQsholfPFVOf({+jiO>tAu`-ci?e#drR{6Y{K`U8p~bMIjQ
zYLlS2wj6e~s@mTfJoNq_!mIGjk=BqGvb7|qNhz+qcTTsi-Ydt#9}uUN|4cj*Sv3lm
z_x^%*7sUw>a}v6yAn!+&7bG=wz7+Ue1+LqDAIF1(A5PXA&!haE)trwS#!IWpsmeEP
zeQf^HQbMil8qZ!o??6@+Ljr#m63ydj6%mb^*o3kCNJ)G#!Hk|RMxCpVe#41AqmvUM
zo+-VbYJVS_xiy`8Glf`M4U7YjgA=OpHx-q?6JaguC~AfwPs-pyF6|lOsWj59?GW13
z7e7#G*kU4ZA4~7g#fem_g0<(-9!)U0?A>YMBqZOTM!Z6fV^xf3dj1$Y75`y6A?%#(
zlfiXTOB?3|C*BfoVU`o7uOZmRW?cA*sK3f~Aqdr<NVmBhV1d0j-^7>N&BA~VAkqAB
zs`LdYZiz;lN0BPb03!VPgK=pxZ4sq9EW2oc{`6c|WFBld=zF)$(5Z5&{ayyWs#0Kj
zn?m+QLM>C;($2Kpeix_N9c)d7G{M%h2qAs6*P<ITu^|`!(4`mAG9&i%%DZkw5`@M2
zW}@uoFJ7Yi*SjwG42>gh*RIXSrDRJY>XFDay=Wa$uFi%t+Q8p$7-EA4vt7XV?%_==
zk>@0FVM8k$S$f30xziXt+N=%Six8g>SNbe^;MnC*b|-HHVbF<|AMm{OLZn_TJz=w^
ztQ1wX&;5W{(|^y#EJ&+OhaZ@j@&~**zk2_!mbJ6Y^GbnNg1F85-AM?7fvtSvWOrW+
zwcb3u@dR|_<La0c-WIIic)rc2&nMcHE=hyopW&5kZd2<XQfn|-KDU+qotaK~NA8<6
zFIUAr5nlA$#i?hXs;+?!YDvON$&*L&2nx@ZZh<g`2*#ZRIYqF#u_0)Z)@_!{lSRAz
zvO+F(Mbav*vg{GsptU6witco4_OqL~`wp!33F@VHQFiHkn@BUxHm?#HaL<D6<65+}
zFTP@#vub>HiGr1nnF5)G3V7FqLDKE!%$gQA@6AfeTTS;P45N7VtiSEO-v;!P;Z#x>
zb2R^g_TMG>>yh*7n>V2sS3Q4~3~!Nt+Y=4I**d*y=9(3AEB5pjt`qTrC0jo)ech3v
z?V%Hw!J*!t>}}mj?GOm?hW>o|P|&$5YdA786gUD5(_U<tP9FwfZ+|Y<v!SP7Z+6`C
zZy_>rUta#4FEg?8V{O5i)^xeLN!3Bd9g~Ito;>7aApJ@_C>PS4#B+k`tN#_5<<X4-
zX!)t-(8Sff>aKJxTZ^tQ)U8R7cwqv~EVx^*>})r=L{>#%^2+XI_BHwZo9lPKP)cZX
zs|yMA;H#eY&qmEo()Qfz_ew<Fn6>2X6CQV<fIlyL_?|>qI<T`E0BB5DINO?67MuAj
zN3EJhO&m&N<XCc;wS^dNEu>@lhe#Fzp5M8REqbk4R!E6WmhD{s_)2Pg(=@?5p!OG%
zNox;v&K-qw=Rwxk21Mny=#s@HBV%+?SF&ZEr}(?oRn|$TEUqozwx7KD!JlStwrURN
z6<+!ueB{55NGh$OJ}CL5244(UTb8QrEq388%^gMZX`{7<gPoZdK7L`~;qGu5BL*D>
zo@#Y#BSwXJVt&Q{M=klS&YHz4hke0@wu%klPM7`a6#c*}4ZYBW29AeRs?sku9@mtF
z<&1-()k_<<d1lsJMiKSJ8*97Wlfg%CS-HPf)z(_FxWp8E=(*2U{!~cilNg~T@HT~S
z>3KjmhkF(sX)m9dMr}v6vTDd%E~&GxY`Lvu=+FBb=am?3)xRj&7hf5A6J97aX?U|4
zDWC%D!FZ<jo_cQiwGLT5gUKz>Rznb|ly>^*8uA%n+PT$}94=~~EYahwK)%(Z+fA?K
zlV4=6vJ4=>9VT~gN+0Z8-9`RDcCKC<e$okkI2kU<d&+@mru47?1z5-zK>~Cp^G@e%
ztV9DHm+PGH#|PA-DDC6VF7KdfPsjalu6~D!fX{>`?{5yZ(YI0gXfz>$2;Ua%=@9zQ
z^mf$V%<juQtutp88yOK=iR7*G!6j%#@Jv++5G^@)0Q2qeM)9Slxc*mD0h#mT6JBJK
zA-RN`NKT8tPn{5e8IP3O>~}gntOM{s5tU@I%~b|gm;7k2*+t(OYkztGYj=v{PpxGc
z6^-rUxi5-gZingvt%SAKiUQVf_Lr=^5w>^A&Z)O>tpg_+O{&k5!A`)pgUbPuV%4*0
z!Z?c#M*uT20gO$`H4h3kxn2+$UY!6lz3?(T{2oNQ=(aP_=`jQlBjaG!`p@Ej(ZUYa
z|9ZdNR`hRQ3(z*4)ueYz;9drz3661?s5)G?pXI9Tw8d&_aSmrqRDQb^HA)CT!q%{k
zmi0>1oa^o#86H#W#9H)en8>4}^l+t>WIWyZ(VMlvaa>89!>)>dR)lH5{k8_oAwdh6
zR7icLN~RPg($2?}qH1Ke*i7hKP<j*%1HKIqQqdY)3#|WtEddTwwK4}ePF+io&~4J%
zRgfI1WQ(1Qnrh(=&}$A?5CCi9To+CX%v%Q+QHW9H;8J~o!OV>TdBYFG$|TXz>)Zg%
zZDK%>nVz$BTlRN+K>Rm8Fi1SbHPmPgeLhk|mS>{?e7e@*YJh4oqoeMj)_`280aX?d
zCG@IR_YNgiJfynSooUKsG3z<_M?Du@I7jd=49Ew?3a8b@BhMXS*k<?mUN)=$um3hH
z2KxAyTREd>YuE1;M(K0kTUY3MoM!AY3f1skY@gl%W<t`bDqx0?3n2-sG{ECZg7kp-
z3b<K^ejf#d#&>OD_~B)v!r>h+(f8JFG~o*%5Ozq%_(uQ2*AmnlaLV&u^4WO>my(!=
zGllED)ycarn$Rt3#hr(G@pztkdL%!bzfqivL4)Aev$h%y<G}a|%!%hl47pl70h&B*
z>EYg3{FKr)rMY`FA^^|o$Wu&f=YozhlU5y;I$Ck`f?5!S{%iDM2$DOJKs(El?_h?f
z_?2JXzFn6#9?oj-(gU~S^K3{AnwLnus=w7b0S<-OLx`RF^QF+W5_jLVqezz;us$q9
zSg`!6;_wCyIloTqnp*UKlN4pM3sY749@nTS6s7z(41*B@K9R_uv_GLp5--b1_`f$F
z*|i<GX#wU#ETaVQaHT}J2S#YB&&a7P;c2xes9~cR^Gh0oDcLHmX|mW2HT(sLOEGvh
zCk2`gY>-|9KwmmZB?K?oLIz7<Ft7$`;fDC02|BP<Jj(BNp`7etK<F*7l=M`QVmz;?
zMb!UBMB8?Vlt^hioWn(0g2f<-0*X2h)PB&jaWw^f&cP(%th7g<csWhJ=ihKw5TxdO
zY3-d0?;V^%rXs`EsHKtH`LTWoAmr_zW^e&|)e`D8I7Dqe8o2Ba7qKO-vr^ePgd+EH
zp-BH-DDCKO+<okRAOSYg?xe-Gy5h8;<kA3v!$;VdEua*;<;|CTg*-s9z6%MPVUO^D
zO9m*Pu&81|Os&K?f)J$#qi7tg?4y&M-Y&I!gB+HpHu6RMBy0#m#wts^&{H^VYR+#n
z{Q%`e_2QyRrs>)D3*gR<KCXVLtAM3H8+u!F*T>qkq&%9(W1mm9zX7U9(mEP6SL<+=
z)1lsUzIszbAKx>g!)bKH1xR4SJ-Wnql_r(Xs)VmXw>E$gwnju{C7MT--^w`B(6NT4
zM$1{o!{Q457;Nquy_S4y(NBW6iL(CiW&xWMIiE{JIeM=yQ69cEpnrI^J-4_`Gp2_3
zrLt8vCjaWkq&hZC^vAx0BK0;<#^5Vh!1rmRm>n$MKRWK(nBSt)TeShWJxyq(E?_iH
z$WM8dyz*m{mOjV_VU-qr(wo08kbjaNEr3Nu*qqXpEPdoLp1GN-nK=6BPrNx;OqenO
zh+Jxd16gm?h_YV>)bxMbSR;EJgToOIa3H0V;(O9buVkDGf9<V(2hi*(>SgAWPaA(o
z=r{8n$K-}`JN-OgR<se0+vqz7G)0iuYC6$2G!~cF=wxRnAnt?jl!)$k(S@pWq~kTd
zyUG)jcrM)t!82MY{|L($#(lYB<cJmQ=O0*+0APfPjy1QaRVms#3RSf-4@s-!Dz9rL
zFMTebl5ssoh;ori7YD|Iv$y~F{2S<tU;oj-3?Zv)bEAGKZELquf=>`MB*llz3gaKy
zl!Zm?-cDvm-}u6=EYso*zaiaT-}?ac;n$?Sx%s6~-n|~c=;gE2+6nPHLKT1hn_nv<
z|7`{Su-EMGzkk-5X6I21O)%)Ec-ML)4&3FLhap7N;spT#^<Dg7f8@}2mPf4J>ee(L
zSa7$s4|EbHy8)kPrb>rQ$erCs@6mB>p{qZ{a~w0?Vf@yIoZntH>gB^{{AVb?b;>%?
zdK@7{nF*cx>c-{68pl8BQKSoj-13lA4XQCLbT?%qTz-Hv_7>p6JBAE5Tha0;7i5(a
z2i%PfQqG~6YlY{b#b7Xn(6hD=bqil{vj-N~ARUlFyTQ{6K)Xq7&ERw6y1f51TjN4_
zA7c&xNCz61J%WKyIZmV(nnpwZmklC}GVWgbpa#AJ=r*9@nAw)*+c@KIOlVm#jhfr2
z78EG&5!@c6)Cj9YBI|LfVxyk7%e}j>A^4>PO{yzhmG0A3)6)I(evAIL8i^>l3@f`x
zWb54dNdP!<iG1d>pQ)L39gj6VsjMZB;S$xgR8kWWYn9}Cyx<M^h1W2DdBMThd5-9_
zM-TAeUf*F=Bz8?h5+d6)`ESW6OWcU9amIR;&BZxo7pvEDRxAtH>XJT*cggV!u-M}e
z_72qQkJh6VTV1rXbq?R|wGa4bof(wQ8d&75u35rQte}2gu`$*mt>}5wy*(MSz}x)>
zR4h06u>XtPw5{vP%e%Il{i{fHE(O4h?i3@+m_y)nL;p^AxIB~5hO}!@i32jXA8L3i
z2Vi)a18`=|-2V;ZgDc*Jb%i;P?8EAATiW#LU?iz@T#B~)&*VHn6DPNCPM;HdU?wOI
zlndOgcEDd9_SULp0l<#S^#w&TNx{otvq<`1?b%&sIG-Y)c`i8B8q;PeR5eK(l5s{r
z5q=*byjUo`c{@_`FjkYElR~O~r~G)b)4Ek)r^47ySRCA`S5|&?tWjk?FHwH^IB8^6
z8|X&IQ|L-aXnWG*aKcc2A3XRvOV|FN!P##3zpG7H%G<7f;tB-Bzt_yIi}x#aHp0bH
z^7Rx3t>FYW*Os+rEC<D{2FzB2@J~Wgs{5~o9{X#g)#TFiE&ZL?aCUUH&^5PK^3|e9
zXl+b>qGrr~scvo|uwk#QOLdKIXxY&AfyGEO1b~~Z|6fM+mfcUR{KH?6!f6k%zsFT2
z?)%62ESDN+Q*n$!N*Szl*o3w$?=``^@}z|i4pQ%ta2G@D12huXbBiSukXjJ!IE8V!
z1`s^R>#~YwJ8t&%*tugO%dd6veQdpO`DbU1GT64^+c|_pR3d-sjgK%*H@qk;Uu{19
z14RK#d?or|(Hfa>jlRAXm%i{p_}`QzUP-c%9`UM!J`|Vwb6nQDJZ8TZzil53%E=dZ
zG$jXHW}#8{l5?<)*I6b-Rpc5zWx6C0{^@}!BazDn0RohlB)`Xb+u!km`oFzZ;NX8l
z|CFsF309St<I^_fa>IZux|%f&95l+z&%=BW%YQe6-$p}~<HfMuZ7HMcHV#f)v;1C)
zlm(JiQl$AO*pLYS*YrT=$Tr1$3Kg@w<E<l*hwwj*0F!9%K~bV~Ho*zYeuG}gWGL*#
z9D2+Cudr657NC0+ZwDwg4!DA{n=z7-ymSA{0*H_E!aI3GmY{0@rz2^w%u^08+zLZ*
zo&zj%2|8!8<v?)coBqf;_JOFwk1wZ;Lg<Wc=<ME+*l`JhQ%A@EnHel2-j&;k{ZUl2
zg1;d?(u%M)EdFDqP9$!TBaNXvtZafM0w6(vhCwOZpOaro)WjDC3ZTX4fCqvStfSiN
z&Z?v!@7@3){!*3Aq&eOie-lIiae{nxn6^V?&9JMsu=1cO-m`&ypwr+lA@eI?uuYzW
zt^RWeV8HAZ?42N-bn~zWc#dP?e=&wJYff1t&Uu+^x&N$KA{lw9A!nqN%nOIncu6D8
z3Ps9c8ofyk9_>M%=CwxuV$Fpr#<><UaqZ)B)56F5SisLuJ+*PPqX&?2#i2J7Opg%V
zIk+F6m21?K%&Sn%U+Hpsn6;Vg@#l6vVt+4(MG_EzSpf$01PASpfywbvjzp;=Xu!og
z!%e(d%`W-<=M&fZZTZ2E&|}U4plxF?7m`+x#7P2EcoE&O|0|q;m5wo0mRuVhZ4dmM
zEF`am3qQQTR@|?)a3d{ASXKO^8{R4f;ez(AaqLu!r_fi$-mi5x%I<E5CB%Q(CjfNi
z42^_Bw@YF`(!ISiaY%p6TwmKN5>4-Ui;Z6YQx^|os-0MZj~$2)`K#6kc;nX4PE%Hl
zeVhn&rrW9+ZgL=u`M8$39_fMmRJYZ%Jl{kLaF>1(*Hpm!mPh+}CVl_?%nza>zna>}
zBYCs5{Fj&atjoiEY`nv{iQeNX!+n7xQ^@CI*WvsY8omm{8$Xn{ZheFN?+3aW>EN11
z#nc?RZ+8>_EHuYua<s5Ia-M6{V}nTT;fH=Kk4VW5kKHem>uyS10<1a?1L2kWiaA^X
zobP36CXvGHh=wOTQuS;5oF!ZOaW{L*u-{uDsr0Jk4%4sMQdX)AON(zd92O~^OgQz>
zKm>O3yoQ;3v<Cqp<eH-a$q60+i=26rY_XD<db8JNUpB$?^%p)3g6C;7`VnHW)5jWK
zWhN8q94%XneRZYoHu(m{lP$F|A;srW<Qf{4^fHpvfY{d`ikg<3h=y-XnxhuAJb<_F
z*&;&jvWSen4T(gyvbbvD`WgbC5bb11et5Qu6i~|6<~3w{7kL>ufm{wD)HW<%K=2-t
z(NuxYkigp$87NL1G!j8K7kyX_p0nn?0waIh5^C+uiqwqh#CV56Le;z6CS;DMrxO%X
zztMwGb@R?eTCcb=5W{-taQyuYs!NC8vtId}4nkqNL`@xD8ko~uZDDzoCAa^kx<cBq
z65mGEMz~-TqI!Y83m5CtZX5^%hH42EpY|A<owp*7xP2l42Hvxw=Vy-eX1Z;^t{x$D
z3X`7GZv+dwa<NmBnPKhBI&BmP{#cp(8m}g9(o}{Q+)m=sc>JS(-IyKGFwNiIK(?}n
zaLLlMm<EQsjhAb`ZdO$M0)^Ss%M;g?@hOFi2{QR*H<G^n+sI~$(yKkg%x>Tf@`4s|
z{dj$h_X*=KsYdTN-i++CHlNcKXxuaoy1G6+=o2l~k^(V<f<U^AkibeCar&kR2=KF5
zI#7)^U0Ih3$Zcgu(YknQoHvue<L)^hnQiW-K4C*=2w9GjsUYC9`aXCoxje2o+fhW`
zN0_IA<Ft9QzZJqaCXCo^7NE*|K911=gYmJ0=>Tk^9osqg965-fNk+gyNfD8i-#N5w
z#8KyfYOT9gN31_85lgY^@r~nB)l%X%p)q`p7Fm7Rn8E@uGd&i({r8`T(Ll_xpX8A@
zc^TlKEa7|Ke5QQrS~dJz>pbPKV=8V!r%RcY>p%Q;-Dv)0EoKgRS{yMFCv&JaNdll{
zIn=)cV%K~I<TKnXM418-M4V9_Q||6)BybK!zw(i*L0O+T@zDf{3!biECk3!S7aRRH
z-e}0>fz#X*C&eV<7(H5yTWE~8+{Sgdxn6Iz=U{C-^?0Pef@E_JEj^FCT+x=9{*y~Q
z{jr;vBn&<|G<x`RdT~b1Ih3h?#DGhcMIHf(r0;i*NDu#lRlshOzP35?)J}e_b?*S<
zKz}{EKY=!!Q(%X5Rcdr{hoLAP0HkDaQ}Hqz3<;LGF6jY_cws@(HBFl~6ZMD?!kcrl
zslnmfb{FMKFY5ZI#55=x6aM)=B*pVUlXI9~x21}SXh{NyC$vY-4PBE{P4!|`9*TJB
z%5;^}nv%72KHYP0%q<leU6Xd7XL$wGDWyATKR8V^+8uXKGX5$9HA1waJ*m!tS)I#B
z)y>$K_TR~)O<WNip=`s-0Dj=qHOSt?0+sbjmOWkt^=p8Ve8@4CKgU&mO5<m2{b^-+
zByZ>IZ(81aJcZ+ByG7=Pg{C`WH!2J^LwG~Z7R~v>Zd54Xl=M4=8XXPt2R9z-!H&Tm
zZy>MYYoDhv4Z9F@fpXtFmkjy1xOz5ng{C>xu}BZ`4BweW`$-vI;K0R)e{f4}s<~2*
zydJYjD<D_fd&lEb2PA8wn?^-};vNA2R*;L~;`Vtlw|xpG+)!O=GP|x`-M8Gn^2kCW
zv2tK}>>23!WB>tRTgqX>?dcWg-}Pq%@?{66J{qtb+#_B&#utZ@K#*cxqT50MrZEWu
za2Z=jQbhe>IaazlPDD_the0BX6tb&*#3y_)M*Bd+;^e{`It8LMem@5(7oB-?pSyXn
z%OR?2)8a}4RIMKF!zx6hJyG+J|FE@u*HK9CiTnMkCN_}ms=e(O3KS07JvxzOKN$=(
zew}l7!2Tl*{^wbGeV&&#5Aj~@$_vJ;4=-zge(3?n7Vq%;-cuCZcHepNO|Nz8^w<%}
zL{cQLA1g+myz<lwI8Wnjid!SQh3S0YR9XP0!fEnutrIN97fr8@hSA*%ew3Ct=;apv
z?(WB*jaEVH^w*dG$;?nPr5gf{voA`Ao1yQ12cSLGdH)IG>0#}gc}nedw<@|(N%4tG
z2hM|HkFat;{w@5-X-$UPv`6V;y7Bw_d7xWqnNJ7d6}`dMNil~9GcvdP`_c9F*^4Rr
z1F#DglQY$X<>OZi7ntXaNS5e_G{%cU<zF3y*9A)r>&Xv<Uq;dFVyqoH-0mcI@SA*J
zQUXc;wY%&q{?rNZ(GbdpmWL+e#+0nQRH%OG@Nf#vFpb$<Sfcek<m5h6Q$qQ{8Y5<7
z^li2QGs^Ob){e~(dHKUTOo)#h7|x`i7_}^%q$XU6pW*UCN_Ljl<PJ>YY8+@C!}H=l
zPy8^xr7vVeqzj-(=|b{g`P6;&nvN&^`XxE?>>vd+c(}FRRtQg{zvX)VXDL!d%ZUca
zV&e@6HhQ@R60t$F<fR><RwywCNB2E2T(*VOM4D?Q6iX`&&7R&<0XHz9S@JLPEbZ@A
z+D;ZPhHyxHQ<Jg)ErgGkbWxFO)L$T3v`$TnZEUJ^i7%P#Ffhs3y1Q8G08?3;+Pt%2
zY4@keVvRtajVH#N`VxowGGyg=>;aKm@Sfsg4+vW0B=bpb{84--)NN<~(T;-+LqllE
zI$OWz_=OwYasLJEu8OX@pM`32rBl?zOa)Mmno|Jf6P=5Oq{11gw-5(UOpwXh)wk=m
zm5Klof#kJgx2zdA36Dw-Wi*uni$)mZ_4tLNn*6!Ei+q<5C2M1RLOD@kh;6N-{~958
zR9c+n&V^rUk&1@e!kXJ%Mr6<ILQ@)~fBxGFxbK9mY2FYO2{*-cug<9?RLS7)+3g`R
zgE2CLUy1rJ%rClgngXE)Hs>%YhmVQ*?TE-9pw-_#o4;Kw45bYn#xVu<vQ-X3>SQey
z>(}xxpd$5!?gemv1o-w;Q{zvdlvp@q8{16YyE$QwAH|DWxcI1P*E=?hOR2#N8EIBZ
z&J$K<btA%}K-~BwD`D?qjmk+oXHA{DCRy;yKi>MJDrN6uulY-qWKwTKB<Ow19nUmm
zW~gy(Uph|7-Egp~_*U-#$8eIgFOqPMP9r|qqRW==USRHXKH1vT_S;y_dasS(lS>TJ
z^2Rq2q~2RMGYciW0aTel4OT$yxm?(_x$%LQN{Y{&qk~D9U&<k;II2V7_~h0g1z7fp
zOJE>+V4785TRv;+1uyXIvb`T7Ue0tn*$aUKitINvHHdV<f$fxeE3barrF|%L8T2ea
zT)6)2iFfG6-oDS%IXSfugyuk?W%7o<;kS>lFD!#`DOgSrIHizPIM?Mvs7DXj_&f8$
zTgdT@exHwveMm)D&**x*7u4Ytin<Wj@}w>C)Wj9-qP{f^o(4_@5v)itaj$}5x_y#P
zD*+1vc>P(H^2JxNxDL@gBm_y>=rkz5UBK?X;SKl{XG>*R*Aj8(+I66E$|5uU6x=&a
zoAH#_&sKeJ3;~zKI8mnP93OY;Jqifo^L?!uZAVr&`&u8(VTkD_kT9F%HjCxGADn;y
zT946cD*lNs4JSqkTz@|KN18?z3PtFEWCl5M2X|`U;NeD%gX(O#5Ee&8c!4SUX9hI}
z|0J#^z{H-MGIH}9rYG6EK@v~F7+PLH=+9D%m)FEMm&?`$6W1r_Vu#Bs68TX=Aa38l
zvz#?_>L_u62rk>$aCi|hTs{AlRz`Y#XqcH(;5x@xTMtZ2pbuM{<^)qkIZDiChtMn#
z<Hz>(x2oL>;h?XtT)4K|YKkr0Y>GB2{~bkzx7G?8VbZijg(#E7N>J<C(E9*hSS7l-
zqPSrvEw-SGfA~E&Ji?~UJ3YW=;B+)2=|6MYl^L~pkXu}i4kXuW9yNW!zAJ!bq@f=u
zg1gNT74N&Tk?~r7^|d3vi})cSUwvg+%zf0E!WeaNdB0-ruVoruU(E=$TK^HjIQ_B5
z^Si@6-a(1a@}akuIs|aEqJH1YybXB6nX5K}8_t@|<-}^D65o*QT8BFNm87@Cq*>n2
zj&by`8)9p#wVCDyxqEvDPyVd!xy*E#YwzR6(elvzIe6gN;N$%5jw?7{o^IWp2Xh>u
z8i$YI1YbOwZZsnuD&&DXsts@F^6O<K7t^*~j=w)|SJZJ60*$5X(C<gxj;(}&`G{rN
z@1vDU@+~yPJ>1Nm4-x<gVe~IWU9NZw1N<LQq<9)5e&-v{8yk{jfilPUYY)1Xfi&Ah
z1C5Nx?2{g!gxdXeq%v&(IelGxyv_Lm4v5fp^EI$h{)j&Ek=xTE_{Cyr>Hsk|DUIb8
zdk9W81XUCrlSf(zpfk)rW1PsIh8*E46@Y?}1R1l{A?24}ccohhu*#`uIpfo)oBeDt
zQ<#QX3<45Qn2d%2^~>BtW={|}DC$En;@UufpRTT|q{miY1UOs)tXWXoL3dpZt=1XJ
zadq4RECTVD5G2y2&R^EYS=EI>l4j1)L5*WHKWlX@W&JVOes~ypp9-ko!y8ooZc%Fm
z81Gl}FO_AUf9LyOOq+o=Q|G|-yC70t#dF(lX8GGi=N(w@%H@&_z%sulCrfTmxNV%c
zvOpb3{_-;Y>MEkEgP9OwjfTu{*8RQFD>n!qRCz?nAyb7->vD~)%lB6Tzmyt0dm;`n
zPBpdpqmw?Mywp1_!8Z$xt>C^v5w{Zqq8(oI5iItbfmiEy<*jieve?VlU2{*-j!hEs
z!Wx)}1P(XOfT=L$=vm!yAXL)6VdRGf92g6KB%J#a%_>&>#z2#G-Hoy0+{?rt5WQo#
zZ%81<&A<WR%<cKA<R#4??d>2uhoQ@7Ab${k&A}*kziacbSEd03CgjL(axm{}stitk
zz!4R`<D^fJM#`~MPkN~%HL;iJ$xl12s4`p+l0#D@357nkv(rTUQm`kgdC1djWG7RN
zHDIP!?0Bp$zgXXlN5VQ{=WCjY0!~B;%k>f`Gt!~1AgOH}EY;&;)wrzWQbuZ-N9x5c
zGEajf6rBJ)i4q7Wu!NsZSEU$mG?OJKaS2VZ-8_240Ma1C5*iIkmtIxb(d6dgLaqHg
zz$$<VO!rC3Rjs+nEZEP01+eFS)$2+2!0&hfD^K${FP`u8P_kRPgEvtn{abwoA}qVT
z2DbLrsdS~Nfv%2?UPJ|c?xy<*c#xR($;Esi2~&wH;l7Pn<CS<Lls0Rl7LQOh(LoXb
zy*PjK;6B~5Uz&R7j>$1Jr-v9vfzZ%$TNB<rTQ+Bx9Q1KnvM=oL4)yPpZ&IDx<C4Qx
zXnpUFx52BS>LJbjYeO7x1sL2-xWcg*4H1x*YO#{l;qR$db95&CZH+0TnDY6Ji?A7t
zRw7Mm-{ON^iM|A|=_9bX-2GB}3*?uct(#Z4VIYypd$hdwgZW=S=7F#sLztsA!5Z8P
z4U*(8`P5>KyDl`2S9dth$0?=Smg71$)BNn*@$+A+OwVEy%O<Ka$7*rM3z_X`)Jukr
zZ9ls&w#fU*&L2JKs6F%Hw-rK`lZ5v9A`rOSt(Oh*nDj*iMumG|nwr*7tI3GfyT#I`
zpFphI?BPP|AtNd7p#WNm@bkf)2^L%?5*dx#LOMONAaG#K70QE{;)B@%B;c5BoVhP$
zQ2pT<@FKJ;Knou$m|P}7xvViz!GGX_Hd{dH#Rf9b8{On+>#i9_8ZQz>*sri+0DA_g
z)7XlaD_dj*oxu?23$&#m0-0emYrET<c(U4i|7{2NRir3g?6$tIm&cuoEjtUEbolX?
zD-X<{jNE&7jMjkTGe1@;VWu9|KlRzaF6dN7x_<=(TAmT)$^{StoGN5SEn3XP43T}D
z$DSSryoO5Jm_c$KGm$dm<tDYOB{Df&(ANwWAJ$Q0yGGht!ap-fU$sqL!**3T47M!H
zl<rso>dHp>$u9r>n!nOsuf~GCATAd07{Jut8V!$N{b2(pNDG_$w*>k70b-31EU8ld
zJ%J>yY>C{pz8ALlrAmcdIp&9@NIjT9)^?KCg>Nn)2Q3`zs>*d6Y^-C&$aq>kn`GU8
zKw=&C*Fk>&Uvcpn$*xh3_SbnvG$wXen0ryC(zze0-&3<WGb-K2k^|zX7w~Y;9K(Ft
zB6UEbB+bN?c<z@(VWPQY&v7!~lpip$02f3I+(r$!{D3A8B=*o#W>RnBJMif&5!k(5
zqidsKb%6sI9lU5KHfQA05x6Ea>2H|s3x!W`hz`=6hqC<NrYPD+$y!x<{Axp0fVPKv
z0=KE2?j63)*|xCQqzN@)6LvJC1l%eIILCEdZ2pBx7`S1AS)<hLqlnauh~)Isq00oO
zItNTVE<dl1M9N{xwi1ouJSjMYASC@Z75i`?%nT1Y+}fFvqdF4>0!<x^BZN<Gz|#>V
zbw&y?YVX-@4&j~wSb#ck$_X{Us3^Ozu&|-SLQ$;=nla+qG6V%h@gn93^_vaZpk2Pd
z#y4$fc2Va<%b@0sPQb7Nt1I9{{SIt%5gS5zt;x=&?i@PL!~GNXskvd&mVJ*?5D55P
zM74e9)-X4O;CPC{-Top^G|%x?%@}F|FT1X*X)VQ4!O*Nsd8gGHlq4S@DK;!ZlCpxm
zIXsi7$8ePsZ@^5L0LR{hx@dUF<Fv*^J1mR!tL|Qal`E#foE4&L7Pi)xHGbk4hMS)6
zXUisM_p+6G@V?Cr)0jjw!gc39ZGB9RXLcDzLDeS7RmQdP{$tBw<uxV~6Vt@hl%1Wh
zT|r67ladQSY=nTzAKSITL*`olWE475H5z$RICgQTnmgU^$kuPjzSzrR!k#Q~<X^E6
zSaX!@N(ux+0r_wZm?L^EGch|TPNhbBVA~Js1$I`A6{}#L+b7F{sk@*n1A-1sI#!n>
z6P!<%^0u{3b+c-|3u?&S60G)uJ;Q3Tqx++U%opsg_Y@oZKkjmz|LEC*JB$;>Lq-XU
z!>ii90_R|gmyYr!9VplyhmLo*`E9t9C6+{!Q0APTN|LYI!qiKIN{-o{Xpcq85Z<!;
zzjq+s*xL(=1q~Nymc|Av-GwEmrAw>G29Fwa8vN1-tWW~(VN4E7*@w*EAol>WB#?#%
zn$5$qC=kkeb_Q)HX+j^PBupKd(wrvXkP)<2F5Uw|7_XFS?V`35?0vplURMbE2ETx=
zmNx-y=W*f9laD*UY?T3@HxSrN=2)h4Z+MP_K5vb_pky60iS+VPqAw`Pd|~De0F3kx
zw-{x;3&G!UaUmXHDg3Q(37^986p%*ZzN}P;Lr{nkr#7(T<lr~W!U^HA7ojZ?Ca;z;
z$93a>-Ds>+F~@pxC`qJR7o0gSGr9a#&TGCzgAO{EB7)VPQ|r|zasBdK{ahp|;&r9<
zyLxp~QxrIF$jH%N^RWPuqN3}k=NmsU0YRm)vNmQ7W{wm4AaaH{+9Sd&3bf6gclR--
zoA=v1z)>VQF3pMGukuwd6@v*Wg$HxqFeQpV-G#26f3RBeKDWF!zwhWgS*72|Ch||-
zq?->csXdm9jP<3svUmBZ-U)27V>)-L&?sIJvKx-m?uX7g(V#NpOF#)(YslqRm`0If
zGVs9XL^i<2`5uOQHZH_-0{o*}NlJg%JJrBqZ9NSvXP@g`Uywgq0Q7!Vsleb^LLhaJ
zj3$<3vqrE<bd!RC%X4(XY8&kvMKtfL{|jwNOF~R=<`wE=seiQlZPeO;3&e#J83|ZA
zPPHko;A0vqW7ygNXLVyu(A({>A6KL$dzL+Ko<*5ajK4~p4Fe!k2l3GA-TMQ9&h<x4
zGX!Md+>+tsYbD##W<2hG^4s7+qQ!(huNpU@le8&k1JnezB!H;-W5WW+)3r5-r#+(c
zDGt+=$cXULaDi0U7i-q}rwSYeoK`4!=jg4ejub>7L=-?kJiWCCybyDoatkwP<n4e~
z%Zc252X@J*i4dGog-YFe(Xv&q^l9#U%maC9^u<<iIO&L%|M2#W1+Q2zbvm&!kIelj
zHRu!<Z`F*5%_gRNM+pZAwuoaa(kZb)k~?qwyG5Q(p}!1vw^~QCS8eh&7!W)w`t_g~
zNOX1sf^Ygu-)>6J>FSaa#ESBF5QOaA+M2Q5$pF_O^iX)aZ`!P8|N3u6g(aN873h!|
zjDV=geX9YSo@H*g{<gBqN0qJ*?t1bgQ`6FdaiN3Cxsi=g8@LCx@a~?rE7)Ujkb44i
zeg^c4Cx#toJOAsS!nf}z{qn-0i3l@s37ij+qbrq@pmjqINqM1hh@lI*YnKODXRj;C
zIk?q(YTWN=6C}Kyf2>6FvtB>VHXNUPD7f-mUoQNx=sZ3q(!Z1_a}`1PO-k-HcH1kO
zi!PA5D{(C^Y|@)G$=UL2_42u|qLiOs*QYs$*?hQt#K=1t%h~mU+O}3x1<fS^PbQ9_
z$ZoPuxo2HaUx0lyS_>a4ZG;xH%Pq|FVPcF$w|`fE#l{R5=tDA$JhlvGJH7JD>XtEH
zxi*HdNoCGkwZejU6tq}OB?yqnCD(c6s4+#<D;#U5erM-`i0N9Gvo=z5w^pobDDi|l
z&G0bj{pI+}QS};)hp#tok^g|!14XRi=(S@|`PprYA<~l)8!dv@8w9Tt4|G^{K6Ds9
z1O9DhA_|*Jtx|>78q>5$+rdcc4Z0H8pEfzHB>}((OrtDu1HRf$tS_ymXTRm<4DtQ6
zmC$v8(G3l%2j8(+-m;+SH9kBkF1z?1k^sTY38vgcIYKo#cBqclk)}YHK0G0j6BlP_
z^3yGr4(1IjI*DbmV^x(Z)UZ=)!P3qb<*Al4Xjnr(`_pC>c1grBb2!`>_$T`!yT~f`
zn+pb=Xe~Wsz6w;dVShw>*=s}SGMj4h(*-@612x5hI%RP-c=tGdLoJ-cp3B11?_Ld?
zV{Hhb)Js8M4ch{0fbUq-{Sm@bvGLv7iRYG#SkRr<Q3T&uT~*AGZ+v!99k_gSnDl0T
zPtfyE1R_e(ukM>~M4!4^9^1rqA9P;P^&`zGMvz!1R~U!k6Oqv6K`*9~Aoya_Q6T<6
z4B?P);BLf*%~rl^VoumvRK`nPhyL-$8OzdeH(dkr@kroLL35nCh`-z9@t8xF6+Tu=
zV&GW&M5_a_Gq;krLZ&Zn;x{2O#K#(fT~(zMo{Phnxk1zSLhcLM=8SIzS}^S$Gny?_
zu&Us;G`wW%-ay)#y`3IXH@$En6-1OJkBk}M*UMPBm}5a$zS784IE%s2t+-Ca$l@n$
z{q4)q@x--BqocxT>g&gqS_qjoT+#AlxiRuIk{pvg^bTS;RJ&JnXsffFD<mo#sE0Se
zmr7C~Ha?feh;aK9kwCIqS_@1|k^CX_A9A{JB#%tdn#_W92j-hakxKVUT%FjqeOygT
zqO4d2F4j7j=6XbP4IeMWdM+jppeGpvE5WOkuX85J;zaDf-C@02N-5t0IrB7@PRh6q
zMJ>u-EJx=yisWENzc9`SmO~cN0;Sba{<JDq$ryI<sQYQXAI20Y-$)U={P7ppDDr?#
zDLcM;sLyy^+V3^h&y(%NHjn4d3_?$yJqQ#dZC7aE<D(Bx=$7WkgFfc!yy*;YO~&s{
z?kjP7rSC@!k2zcVMUeuzFSrPw785q|Tv_kx=r?mgbo3=%{<0w-jr0*_Pj_qKu6hc_
zZ|2~C3GZCZObhEGN>DG0I$AREkitH16!as;A1p{OTju?!bH@cscl`ucKnq)CS=Jkd
z*nJF3I9lD<zO0|OCImD^oNPCjf_>FlZ82X?=)Je}G%3eRBtk&nyufEV(b4$c+Y(t~
zCB@4BWdYVZxow9}AvHID^j8=!>-IeG1k5-<#dLQ_z8aasOp3m9>z&m_msgxEm>wOD
zF4GX}sS-aYSl3vvR&dw;i8NuGNhQjLY{y?i87&ImzRO8oYi3En`(5sx?w*W=O#6f$
zx(OQ6k5wy-f#9j_xjD=O@e5O0hVPf3vid9UAY2bLtHZ$ihr2qe^{C39g}dgwyU~GK
zA1+Sp?`DY&C$WxL)Ibx|-3^E|mmDxz7>SCNiNlsuKCrCB)`h6VpG+3wc4<H9yJ(=k
z>%PyMYz6HveD5A>BOFH64nR{Io?OQzO%W%5Sw)AmK&X=Ediw$;uH@_1l~Y9o_1P#z
zQ<z01fcgzs$7ZnUJ2_@O57#edvIff^vm_aIe!rZVWS8}?ajuqVDoI9tfj%jyYpT%l
zK!W~3?|Zi!m>SrWoq@W}wJsmM#!epQyM?nbIqPr6Eq{>Ch3uvu^BMd2$kN{tE}HJg
z>VODmqP{Qm?l>#2aMgfo!GbJ=t{fPWvWg2hp*n4$HUWR8OU9a_e<OPxXf}STEPvR(
zciS@14i4aP#V;X&v_GWhG)aQyiYvQaPFi*fs@M+=$0HD3tDB#9{rFkz6zS;u@a{JY
z<9%CmOr|$SI><8dP&EpE_iEEwN(~sD{>2nvoyw0Vh%Y^u_rD$t13Zi1E(Q@e;Lb9Z
zAJ_ms+nH01F*h^s1$r*yPLWWrNZ)%}w%|2RCcSDL#fEw9bu*3OEE3h?A}+^^{p}G2
z1K|dn<b!>DgYTb8t;*VFTkvzHwyHwoQxl*ODOuwtXzSRMkV4+FP=0goc#ILfCf@;_
z!Oicp6g4OyD{bB>kl7)+IwJ~T!&=&7cjAP*-PV6RVP*&v``%E-Wyn%ZLGEZ&zBc|*
zcv!$JqjhYS0A2&66O?ni%_ktj)>Y@L$m;MGGzI2z`1FaweWm+e=B2Z)LXLcK_r?OY
zj?cZSp;jQ+FTN1mHlSP6Q-~4>qarC}!bQBF+Vf42Dt7htN@!IqG=!Be5taZkVkJ9q
zwr|&6HSFc*D^Lnj*CnzXPjpYo(F1cVpGWwoY4c7geZo5Ww2?P*Q+l;E*Ui(0ctv7G
zPq(>Jlj*6qPjvK?Js&?|0%_;x{pvs$@L$bY`x6QT5I!pJZwmhGi$g10Tlo3|O(=rE
zHN>k@HH(2VMq;YA`m&ui&(_-%PF`9@Vxc$qy<cZUY=XFeAsg^2Yt2b(i@OSSvdK@h
z58;|-<^o!JDQ@TVN?iGkb=VM!Gt*$zKS82oSHkaSU+uou-OL2Ot*);N2Pho}o3n35
zIt{NNkc5mMLz4VBT1>f!_DDqyWcntuRoU!3Hv$|G$E-&c3@(o|J@jrajf3NhP7jU6
ziUz}l$)h%MCpk*i9&8{64n7;>QNB!Z3-|w35KtH&Kg&gZ!1-1txa?k8OoI)nd5vDu
zJ}4o)y^<l$e~2(z)12oe_BOZUW(lT+eJ>SC=X3N%0ULOnMxun1*U|E_;QryRTB{X!
z_O#)qp`2>Nmc(nljW%qi%DkS<&De_qRa36ypM)jjmYWW*;air3&u)FF{E5H{%VU|>
zKUZyBGk6#x6WqCMP%qvX>F2r0R^5$A<7X-fE#GxvCj878a0#WJ6`YyNH0YzsCn5{~
zZopQ__^2P&wKQMX9fukJS_Mt09Bb@jU!7l_X$AXsY&<&D&T{PYPI-fX1|_|M{ai_c
zZ^FNy9M1dY5x=!y2Q9}|(p0kFa1f5*Vv`DL8Hbi@gGmiOefc<ecyhps6N}RpH_!Af
zcTy@Uncr%HJEFY?v;K~5yC&(!0%+X0$x%kefZA9CoC16}*X0@Y>XBJE)hNw(z6T7o
zQV>1X0bH0O2jRkA8pPtuv4Zcy3Dl$aaTokLfAw2X7r4PLX|oLZx|tdiDN8F4RbRVP
z{8)eW$dVBMl~m=V27-_+kIAA4Mq*Ryb;@vx<dl`&p)|?;>Lvk_hFa>*aGOBln=dtP
zq?(<tzv|%woh&oR<?FcptrU#=AlP@T{Re<r8XH1;iYDI-9CC*DZRA-OM>_2y1nRS3
zR5AMOaOvZ-AJf-!kD|S2Y6_dPa@usmYzJpNci)&a38Ak$8W1<|(+}L3z^cLU)sp_V
z2O{jWopz8K-BtbHaO9;UC}z9H|56Up$8|3xZ4VhmrCI14&7KF|&Ia!h($uff^Q-N)
z?j+bXB|D2Q(prUQV-0+)$Y}j4%!M_2NY*QaU19)iG1c_IHHV0DVehd^vXa$kt?3RQ
zH%e5U+i~_D)xE(tsS_D!(?8qR<o1Q?tqex>Dpkfb8wnMsS3#vS?^wgV<CDp`o0JKF
zb(9%89&i3Y`32`dOjFN#$JkvTZRM!d0aTmsmwl?kpLNooNqT-kYQ9FaBW>yuU4_p)
zNc2a%&*zfzM28q83wA?+SoXi@2!<c5yX=`cTlzk-i*Uda8LGXNS>AZw0kxbWIp%uy
z4uFq;>WWCq&w1P<?74yg97~7Sau|1%q+XYL-oQ=qX2((ILr&GusqzL5!}_r1@!?P7
z#I357zu<Y~HU6@R_;Zi*Q8pZHAs%GE;cX;PAfM*uRi=DMLTL41Au(M!NN5far(mtO
z;rNg<g`8Xw8=AQdQd(#d(Ou;Yqg)<y-(t_Q!m=B9imYv+opMOJh}(Z)x~Fipw2#3Z
zxGbozb&mXssI9h=IJIJsF?@bc3>T)gzbf1#!gnWMv8*~-^C2%p=Cg|{wnsB<2)`B(
zuwzLM1?;&IT>HlXEfK;;M5}M9v^Q~WUNLmtz7Z}=z_X48lhriovPV<J#+MZ%3u(Eh
zr<RaeBZK8=bO&<Bjd(N+WH2Q(e#|#MeWQxTcxX?(NUE>wTRRfB%4Rzh8q%IggqKQ~
zKf9)FU~4#vYtyc+Isc$31wX+y&DL}@sJyMH9M{#$qYjn!lw`?N+=%<P^fdk}J)A%v
zA#Nv3-#!>OQ7_`%?%~0Y-q+f1N|J^my-yF;rX3ua`SuQ=fF;2S<p17+JH3KAGQ3l#
zo=h=fym!4*smwgA#wYOGGo;mqy}J3DaI;JIPCidt4H@^V`KBoxDV&E)U%EMSB!+Qq
zz_R<5qRGPHr72Ex*mTV@onRRp*ctyv@X5iN&^Dzj3ToJredr(4+_q5|Fd@PfTu@t0
zgvbT#QS;RXxqgLxo0E4(lM~F!d@fUz;pMi`*D}i<NY;30q)XU+w~~;y&I~Q)9;_o2
zaYtB{tzzDS(UJAjt{a=g9~Y|nPuv%&btHju-Nk62U1xbhI-4aOF;nUQ?%Jpm(<j(u
z%i~Odzg>MM`NpS5<oS(#%*cQu)B5$xVj1lH$40`fH5t2IFNuK0h@&=%TV;iUP2`z5
zdvfL&09s#U{H67j;xc_^N)`H88(E%#3FW1~vgBk?w0(c=hyy&_<X38h_)dPuWN0vO
zod7z$!!DHFAJJ=_Cdy$38lt7as-E|^L6gsMfZm)}D5fP#)rgBSZS*ln5WDpN$$|<V
zn9m<*QhddvU7g{R6?!~NnD<&7CJ7H-S2=THuGpHF{=zET<qr|qr-*3^D)J}OiEjH8
zeuqZ?1rE{72!&eG2kl4NSD}8b<lXbNfVs_&U)c1giODD^p=%VpUm>JO@81@CLL0EH
z==O>B0h^?ut&I8TN=%snt4@Y*Cg+;%rsI83IaWbx+PLLoyIHl{LdeC;dTr&I&H?VY
z@xk?{zXOm5;~$OH0Tz<sa(XsSIk98N#^H|0#S#`(cb6qT2=q-cr5SWQfN-;exe+z4
z^_7(5X_Lttwh2C5B9r{bxX61M{6nE8jcr!qo(Fi3lf=ZI^66I4<-Z4iYSi7gbk}=i
zC&fK(f9wSj679WS8M?T9YG0ZccuL^Gg*?~R1l7Ai*ZBMZa3ikAa*)?FZL2)fU_+JW
z-Q6`^O-Zu=EU%RgXFN#eg0RlEqG*1->{ka2$VQnM7(xu33RIifTu+lFbhM*8Je_eQ
zsQr;EV#Uoh$D8k1Q&5(>+hp~%s+tWJ9(O&(9*(-|fn}szb$>*8I8P}Ez6yIl`YqOW
zTa-IuR0-F^d_3i4Sg_PRHXCy@a#`p;ocfJNe6=rdS)|3vYheQkAciXs#ly2#hLTs*
zM1p+X@=);hI&=9o^z@7)UIKq18~269W7o&$wY?6oALi9Ny!(~8Y)-K;++lF7x8wXv
zp)ETb9(u{;bxYb5+9p-S#vhmoKp2LB(bhXu-e$c@NiMZGdB;SR0^lj?WK**S4O#WJ
zXl-0xO5bAUNq%}S8N@}?k)!^Ny|RzWAv-@<Y0dT3XCAn7K}Xv&VH}i?9Pr?w-&Tdu
z!Ji-13B?;Mn|}2ffmOk0ZRdA{d`uS$MIyh53x(H!|0BmS4y{Q2R_&{lKSr;<`gqT+
zA|zr(U)YRVP(Wh7zS=qiJG?eK*+TLPQQVik!DAoeG;X$jePcXx$WksQ6!^@ML{(u~
zrHTpmrdk$_Dk`*kM{&c;EeBf@CHDawP5;N!SB6E^eQ$#ZC?V3)2uMmxmw<GKbV!$U
zw@L{}cMC{KH%N@qAss{4&^2_&yLtTm-|ONVb>^JC*Sc%%liD!$KFj57VelkNC{d%G
zUnE{c?H`^z_#cOvd$s;d1H;=N?bd)yd^hHpCcRM}#C$+mN*2(8Ey2F{-@7$D=JqZO
zf3B^mhUjpzZvc6V^E6Uei2m<4Iws-0$Vc{~sdtS933;v3x)jXJMuxQ7hgOo9n#rld
z-dzrW27PUaBW(1#p}J(4j$bWiudeRMPeJngV|%c<YN2j)o?qf|>n!zMLZ7@-BsJwr
zqdbLD0q3^Q?i+@$ig8~WoZiH8LMUjCtr-D;r{Q2=wBbyxJ<sp17I%17BF==7-P;$Y
zgqopv?^n7p#4MFXP9wEPEx=OR^@#OTU*9)G=dKZpE{8JGa4Nvjk*)aR2@(P14bpd}
zySc9IEG<o%QvTNlQB(hMnl}CXD(>8Y!u=Ye_d}1?k$$o<Q6Q+<BzbjnG-2n`Q{NB5
zOcyRPZJDpfH2?(vax$l*5Lb+Mac&dh4{<vk`m9TCVe1XQV}1I;My#{eSQcGQStpfP
zXo9B@vsKh~+udimRV9Af0S--LXIxB^<Y3nvak8Y-35yu}uY7t)|L@T}Q3wkyY85e;
zuu^F;Nl3zb^qmb-y|F$3Q_d+mcch=}W@A^Iu<}w=MPZ0}3LGG#J-aAx0sGhKB{pPI
zDwoL^=WJZEp(r#sc=@!onLkU1qm$AVj3mib8!_bQY)bS#Qfwz|DMaKquXNeONBw;l
zoEL|wG00u(7lCJfb)5Thp&XxQp<JOw%FfD&J>`(0uo_)OJ7sxyFSu`J*P!6l!wzc`
zm&g0ZP%KsL2)NNX%EkpWz_vtqow2`LX|v^%`zcjb2M&wY@uJQ@Z1FiRv#306S(HX=
z67!UqCOa>Ft#F8L*WVMB+0oI~&{mn(9|5CF`UGhlX`IV`hd#ZQJgZHa(Elxy;QU-d
z3`)A(7lR&8K(LsZ7Kh1<Qd2NHhU9SKT<mQpHoxVfCrsMg=-T~N$m{db%IuIYlcvBU
zl)8fJv4DzWCl1EdfpZNLh#|Z-Y}9G6OsFr2%mQ4;7scYHJwGV-fE8B>r+(&U$Y|RI
zKKvhI#}xhlA5Ku<`t!TC+8+H)7af||U#H#7tb=rWM=J4&`MjtNi;j?@2;@jVJJSx@
z_F5cEjrxgtg!IXJb4$C`E~B&sox#fiIuJsXphCZ1*J^rfY-}zrV?@gbrJ`r(p>&$H
z%5e+Ax#^?-VlJd=GPkaO^HkUOv6$1#sn-1{y4P4W{JB@(A3Zd6aPJzt1kn&u^g{m$
z86(zx41Z-QuXT8{i?ML)&YqCbof&OH$geFrD~C6oa5$$p+>xJJupm-$vUm2D^O#ST
zS#p`-nRdlFGk@W{caM&KAF_4zP6VYz2(fY0p$SXMQ1kCcW@%7wIb5I7n%B9?DHp(W
zRuw<d)*~+F1z_)9EMbAk4n$>bexunEW=3?>zHz?Qd_9_19G~NR9^c)WVTmtfNMB2N
z&%;Xo)tco0fhVw_0ZPuyxRu6eloy+lZ03(DS4)@cPRyQ&hb@+?)fOuwp|}qdderB)
z$(R3nMXL6MJxHr*dU<rTeM#9?8(AdsQrcfW^fAOhY0SKbWa?La<KC!8hKBf60&bxF
zeezCB6K{h9F0>s6AVEJ0>>BV-brx}NUXpz8FA8Mdhjt_OlLTyq;t}HI8BKkugh|Tr
z2~l~uSY%Q|-PL2fczv@xUQ7(q8%LFfp;MPcfQ-eGVh^om8I081JtuyDQ}kkT@G3L*
zhN?vW(I&5#FgckdHBVzp=_h`k6!1B)8+XtF5&bT*;pai*j9P4MB4?_qgJC(;h}j*=
zcY|vi0#Pd<jiahliO=|9ue73deMS?pb`d2oeLo>Nymes76YCb$9G<HbH7nvVy4BN4
zAiBEG+UT^6#SJ-T>Wz7dk;bF_ai8|Tou{ud#GL!z4Sr5+04Hikd~&;^-zSG|BLv}P
zO*q)22zVz3Ck%csP&ONfbRuM4pD%9gSPT1^H}WUy>HVsn`2FZw9APwe+PA*Z7;JkA
z+RbVFDQ+9@ZoyFX;A(%0;O5r2rH%Wy`N~dPZ+Ugs@2uHci)#5Y3LfuVZ?Gu+yoEi0
zNh8lGFWyn>cZ31r)v&Qk=c{@k7>rSE@^$&*1=$bL!RhzE_)VD1F2mXGU@y{d1EgD$
zJAQ6J^5o|cm^lKi^9+9tZW9uy#TvdV7A<E#YnN_rQfSR>K>uJ4lMIQ0A8$$#Iv#BI
zogVONKpK+4)(z<1JiSq-QGJRnAR?aO+%n0h5nuFFyPOr{+|SXYbb}``S_G@|z8@0J
zpnj5887x|E*V$VS&CtY`rkV1rDS$geICmp%>6v%=j;|4)vCOzA9T7ZK-sNTLf9f&i
z8@@Zxp7zF4Z)*BU_0RBhY=cIenfk^_wXW#%g0x-Ir%*S3n@>d6I@IJ3g;ReQL%)yw
zEC`DVc;@TKQ-x4Ku<w$JXX<bP0B5X^FgdQ84S9+;hJ}AS%>HQWjyJ$?KJpzSY%Uo0
z9bCx9!3XeiV{4OVO`}FOsg+TZtd%baHoc3Ur3xUX<XnZf)DSb<7#hgR_+gi#LKNWz
zeqyxcjcZ@}MnjG4ciUf4Cf97*?A{Dz)%L5frRJA>sjW^_&1cDO?>XNR_QY1T8X!DK
znf}GOve(Y*<qq@mAoV6J;o)8Q1AQRB3~C!y{=HK=whj4X*1TssoM0<=6ni6lmMkHL
zkf^fKR_?S<NSiqYH`n39%i_MJZV*ppfz=|KU9~!OF3{k!=|gS7_#cp^{J*;=f%6pB
z>h~<o{25K4ttmVQJD$LNg13}jAchS6D%Nl9v7K$pi@Upc;*FCO_H=^V*)1MgZ!Vmb
zajp%`oS-2S0OUcJC%?XX6_6J5KK{J}&`=S+<d=8!O*bsm@if+vN*7d8AoABi?)_1^
zQ)hKK91rDz^5sK_3VF4qNercVgJl3gkib9jvZH4XapB+<^l5h-H7SFw&W5w#LJA#E
zk4gb=y&BSs>TCOa5J4v$UIw!^`STsDoS2O0NZn#lmlvE64QDgjd<+jOYBV(_-}DQY
z09Z(~K&UyHPtP|jqic&6hMsT5hcC>wNzZ(F!WuYjDxGRgZg@DYUCFufZ?o(l7?s)0
zBfUpoY`CCZrTD)2%t#*=H+Dv>Y9eBC{*#2nUj37V9%svNRM#%`F0L;v?Z`Tf8%#nr
zq%O{);IB!yTpo%mFwH(*Xd-Ao%?1*jS>I4OmFC1;fCg3E(Pj?e$*sAX?1yzvEf;*_
z@RMQ}`vm=GX{tsud{bGWdDWOoWyt}h1-XArNbk%!Q;*Q}9$hU@`>8W$J#Co`urRUP
z5LM9-){K-SCf%73`V=Z>zBQ|K<49)y?)RNM{pg&qa&Ev^pJe~13oXDl0)F1t&-TJP
zCc-@EllpYNM&mz+@AIl0zRy$NKgx*2?LVLNYE?$c*w2Dmt|Yn6lnnpvir^Msd{;2y
zv|aCu(%qMUMtgsPHqcZ@DZr2(ajfHHzL&O2YltwvXaA$E<S~_-U)L{p-$pk-x-)}8
z!OL0FzkAF!873yuQH_HDg;%&7<gX#L`<55OCOjW0whHw&q|1NAglj%XM9%}StZNh|
z-w<PlJR>unt+Ndb3W|x1Rc%zOd6~p&Xjc;dAH-@Hn)2f3@V#x?z5Ph`K<Pfh1lza=
z+uK_C%(?eBU5LY~;*>XNktjK1#&#}^PW*i_IB;S91MKW-99F%rfvM6y@@kfTKIR#S
z$m$>t(te<tAaswDW8A1Aoe5p05VRW)Rwi^0L3}=Ze#u|19M7PBapn{IbuY`oYYovS
zVCDrW1zUyC7!=YSiFHTGzRL0XD{{)?IrXan=s2;jsBS6il@9+b+eOAxU`@1i=~b<$
z_BN-6Md;e3`*=@m9U@$uXG->5)i?MY2<k*=K(@6kpEIb)tQM><ZH|!%M4h=tKGfhr
zyDJ{{>vp0$I0MR%A*U{S{vCH(`=agI>QT$P{mZZ(Awwhcy|Bg%RGcDGR|AfqltTny
zJn(;m4@95SW>@{HraNfCslF#g@M|e}nUFTl{Pmq!c<WFc*u&uO9>Sqzh@AgOe}jH5
zo(HS`uXo@Tb<#32R^JZ(gHVF+;tBG<-N4mOJlvjucIp6H$@Jk<Jo8uTk&b8%bRRyT
z@+<9!Clu>vL;^W=^K?#`UvX1mbDLUkhGMYxfVbz`B7lL3P#1JjrI5oGN1F{Zm)&Ml
zU+<^&lj7U)vkLpAAvl${xT+vkMhqWs%dqlqmDn9YPhWHjn-bq(P-Of<a~tdgFmJg5
z^Hb8%&}T4jlG+ubrJ3nso};2g7(09QpFd3kTb13tkxK1<ptS~vf&O~v;mE&R-rpKf
zbJ}R`3eLo3;3@i9D@MRsQ^W>8@xeB)-ua7ZlC6(^b^s+Zd+>S#{s@Q-4;NX`{pA6X
zYWhVi&uIHnY+q$Af=k?n1?d;LD?Z#4<%EuocQ-dT2v43q&DIw<{Jhfs^ACEnr>CdO
z_IPMcjz72N=3FCZhJd?dQHduE^WVDu$$QbjwmL6^;pvWp|7s1-bd@SWw$h=dY{Ggz
zHF-vBc%u{h#MANdTAmp=9yN!d^mpF`0s77_eX+8lVM@)^+$x7cO*?|K0YrM{*F{FF
z%rrPaDgRS$Pcpnk&p@9~UyP6!gyt+DN8_r+r5l}DJk<TCc+&YkDahA()FetRIulr{
zn1z%=Se6Y?da!f9d=tzSc?+!U?8P0E&GV}lZ`SYu3CW(Y1Xnd#|IDflZPYG(@yy^f
z4DjkO_54B~=KnTmlgA;|4z%0?zR&{R$E~WT_bnm<v$eJL;nOd)c#q{uVF?Ne#k`4W
z-H(td7*jDYFgSj@{2980yzojcQJuiu*T;v~?}nF!g$2|1#ea8)+}J7MkuFxduRX~Z
zb(Z$gX-R*12{M1%QEO3D4$oVW%|r(Et7zBx8W}24Gcyf%|1y^u{*@2$0sH&-FcC#!
zRf&|l%Xsc38<%6(@>?Ii$^Em_L6}`pgQp+U_T`)mP0y;Vqa8YM{62m8@`zR;wWCbG
zX}^J$iz{J#Ts@LPWYB%l7xj0R|6Ol91DAar(-W*P6xM?>?dtrav#Zk`3CI83`M{7g
z^)PkXyc&8E1&J+NiIJ6?H$cvh^>GxD&<>5q-^i(Q_&qv;dd}T$Icd&|rdI6Vwb<}?
zzT;-W0N=G5eo&e;51kn3Hr-NpW`%iYQwk-?UAKn{5Gfcqalg8>eob0hdFs%?ss-Vg
z=_t7?mjYpcN-3&PAyOGY61R87)HU+88fTX0siO&yW=SUCm-%TX;Ur63svFxTTD)%t
zpoQ7mygG57uaI5s&>paVq?&>1TTiQ(ZUY&_YyzOXhA3%k6V3l}rRL|qGAI~bePh2c
zS9n_%h-w%>_^|!eJ)3N2>oCdw<r0=#bE`PSGt{KLb{By+UGRf40Lbq&>jzPmT<&H2
zaxE@_{oR`1*=lq4hU+bbOrJU7a*nc4ApM@%TPO>5IT))~m5vfJDCY$si7ib<Km=3u
zuuIseqr?~OmUn-dY~ERPx=c=}<0y7I*xj<LYpkE7@TJ=+J$qaBrYz*Jlu@>iuhE#b
z+kZyFO4Jdby3E1on&`*z{f{vd(Z>)DGyu^VyXN=J?PqJIU^PTn=B3BKWYjd|&BdO8
zbi$;a>yEuI#PP)W)t%`ZP#`KDZQgJ2@+?wB`sJK{rF<ghdvnl^mC&glB=(2U(C5!o
zrHP}nvp7pq?%0-c@x+h3JZx<2KXtaV%0=quyf&f@O~lQ=XG^p>jZIF|_<KyNf}i~p
z;~f?cw~3_3BcHD*mP^_hcYS^ZQ{Kn}yLODql7C<(qNA(!6|FfpfKjmt9+4f@X<yN`
z<$%B#imu)NJBb_HUD7{KeU@Ele+zcFxf;?p^KCp?yzp3+t-4GXk%`Zm!fad5$E&`r
zqj~G*Wun6s&W^zyxvSz!fNO5NgoC86W3>3{e|L<h#(u%5!fBU&GAk{%iE_rQr39`C
z;IO>DMDng4xk2-R@Nk)y<iK}SolQG5sqP@mW+4T1D6n*6mQSna|1S%WgrlI^!CiLK
z5(@y8qoezaxWvJ>+_>^TQwu_mRFYGpeT=a3B&O*Hb~MrzFc!n|OOgr1`ME61ck)X*
zKb@stLIU)JT2cvov&x;*)8J47q!}J{YfwC~^2#B1zmD82EjP)+pMtkZ6A%z+X=yF(
z?DRJ{!uH+5zBmj`_-RHE%O|sCAGwK&ii(H)Pr-+uc)PyA2?Yx-u^M*V3hvIq-r-|h
z;TEwW&;>^Jv0sz<0%O2vZSE1kxljC-mB?@D(ULy-t?CI3B}RewdtAQY2uf#(I&Uh7
zl46V-T^?Dm<iZcMzpYSYcj0FSEFbkdEKnFT#h^t`h`bZ7<Qu_X5KSOD#C|wdeByB<
zFsF;v2@(Z)jbxK7h(Ko-dq_K-iD2_yolsu-2=>$Zs5T<%D*$q>&Ap5Z`pWX|Vu&M!
z{`bXiTpsRe-)iL$WSs?GlQmmSy73BSX|)9om90Q`tUxYGo9|R^MTxjo{(YJ1u(`TB
zW8e6@-HRH<p&$+2KaI@Ch?{Jc?it2&QO#bDKfw0i%AO#q=3q@SNE%qbsigChvNDog
z8dk(i&jyN4du~q7m%zZs2)|smRLhw<hcbo2Lqng^($eZd{K!6SZd$JYO$kU!B3xfz
zKiM1!u!r2%IoUIq6i!&*UBIxqq9~$@i<$EC^B<$2aM(<~9!%#maY<iUF`g{d?T%+q
zJ2*NLD^g!sU++3tZr5)1a5NjNmz9?<@sS6+B_h?r)02;c$Ii4bp5fx+;=4LyWnEpq
zNq1y3cw_I!`CqjfLutbmU>^y9mqPc=Ho3#?Hvd75C+B3q(w`H!_&#AhtPwxrIoExX
zaVL<kF^ICE%~$_nGhOHbHKphn^6uVfJwKwYh^-p>iwzPFp~5%Rp&NDMUU!WH$Gx&o
zI%{j`N+n&a0x@h@>DU8K%kREw<Yr&T+i%{U4AC=V)VEhS+6)}jV=YgZQC$}Cm1H#H
zDKst*^f1cyYXe^qBq1(b{L=wD>z;WvyhNIiM=d&ZBMS{R1VOY9^Zg>x-a`FL=}VfD
zFEtq(D%@W(5T28e>Lx_H+Q+3iVGptRU7(5xmO@R@_-;48AunwnS;gc1fLi-Wq@=cs
z2t_IuXP=|Dv(<9je|rN20CF_dEfF)OyedB-xGO)|+S*!tQc~`O_2uP8meq9CD|A8@
znwKvTYjp;S)CxTIYM~C!&ak#1bgr(sI$M&+$VmAN{shR~RXFHnva+&YLqnIZ;gAB|
zdb@ArLa8e&KWEr(P5Kj=d3kx+tVU@x$_*66#2%TMnf>_jqrwKtV&J)kpUP$9CWUx9
z**%`@ng8R*($-dQZj^A}_C#@LczA9}$+w#upTP`)?je!;zndeuMEow65ko1%yx>9Z
zEHD4bS4czfSPmlW>FKfD6#vI2=iGrriActLC9Kqh4gd1acwDeZUbMWxuk97(i|-mF
zwp_FT(MtN>;BC5)-d#GzQq>Xhs1(3%29eWe&#-CD8M}P$UgH!w4oR5I9`_Q<9!=Dc
z3#Mrx8FG1L%9s6spCtUphu@KA7e~Gz^<O_n`k^P?(OXXl9#)vTo$=gf8k`vn-?+qE
z!yJ$!^hKW}A+-bXrmy#cpB0Oje~%tY5_-DwGh6WZ{|;(-&_+40#w@-G8&creO1b7v
z9I)V-(N$S2vxPa85pk8vXSCmSn)JvDhFcJdcnS0Th~Wfj0FMfpJXL;|O@!Zvux=T7
z`H#^Ja!D*}L4*c{XLsPH@w@Fx%gUanC9oK_rmN&iW@{8<lH2S|mST|dJ}=eKUAvV_
zd~NhMnVqYvsQ7BVXbi+?EuR+^6-C5m4YO_QEh)&z6rCB?+s$8Gz@mBw20F*a;yS|#
zoex(eV&md42RTo=u>X-FL$eNA+*&5Qgdn7OT0iX_;}rG$@6b3As+d0ByJI_9bj<Qc
zjk>_`X#34Wk(X2R)ps#sef6g%W|`Yxg~p!NPLn6oY=n)fe=h7ZyNiOZ_5U__^<p1(
z>|%A#)e&WM`|9mnas-8CD%~Ocj(Ig(NMGvpO2N~SsM~MqDbCZD%&U=$m3`)4@>-vY
z5;#=^#HGU(Ie6<Hq8<ZYfmJJlZmgtzDX#V^I`R5@7wI<GyZUd^0i1XTLV7DJgq6qE
zIwVW;_&85Z2;|$V!zr{wEvAoDELh0&gEe#qwm)V>UJPS?yevcqfjWl+SF@EB?H;^y
zzuFd1^KSRsvF@!<*ETk~=Ng@#qN5jUik?&t35(OlReqf2aB_AYv966zOvJ&#QJqG~
zX>Lxp$n?zbyM&e-T3nrMAxuq89n9K8V$##~8u=R<8j^>Hm9Ng>SieJY>9oMTCv#j`
zS@F1BOJahM^VpFtFE8T}5`rySR9HyK#1xK%`i#x<XjMmKFt~>kfXc!`dW4xeTRCaz
zC&<XimbEi|FAfh6Z|<()H8#+)Nj@}8@tm<O7uOU^V`Ehq&cUxj6Mpvy#e!V$-R9e~
zS^3mATLuMc78D(KPUUKr94?#m8iXb}Z$yFk5h(Jvudlzb!BM+ZwFr+yLU>hHPVV!S
zAsX%6;u~UxwMI(K9(I-S))%M_7qiTycXmu<-G0(QiG%V1Gft6q|1UAEk3fJ%dzz#!
z(kI|PPhTy__JzyEFsL`wk_O7XWi#itv*cRCGpCCMZQiP~TeP6PZz57g&U|HZa>t;+
zzzS}$!&~_gYo7c%6$lVezQI;|saU+qjBbBVOvn<q`y53zoYcmgCEK&W5hM9jGb0MO
z@|aVMr&%s5h57|s!TNydV5vw~<fXr=icaKQ2L_;m4rlOyln9d4BUQBOl}7T$v~R_9
zh!v+UtCe}I%Xq_Y&<@T4%=kife7^*MBSNxv)5?e+ULE<KvNo@(PP~Fap9Fy7s?kKK
zn>_DH&evRkuz$|sy3E4JqCu@muiS%w&X1psecKJ^3pg*H*zWt1o0n%6cSYv(=L>SW
zux~0R3HRKN`u4}O-RajlHJg#}reD7}JPtmfV3Gg*gop%cZZWr{mevcv>&kBUsi<=L
zGFw5ZcG()ut17THHl`vcC$Eq21n+X!sx(HftE<y%@xmub`^lp{VO@(sA)GGcefIqM
z^S3T1dqO7^-n;CRW%}rwJaK74O>U3?8XB6VwKe3o9v&BD$dh0dadQ*+`}-r5={0_d
zic%_d!rOnQGVm`)V`&OV0ixxC(Z`mRH*nrWpj>?wa>|y$_n7^*8?1Sb5Xmw>*of&L
zhgVtJmxu}c!4nU2w)srbPi-}Jokv|x$6QCdS`cVYHB1W>U0iRPs=i0|YmVE#EZ?YI
z+M7DaBj5b|Jqt`~pR1bno|V#fAB81WrWI&1=Uuva<>*Mfghb+j^cce=Q7LOqWid{t
z^x$YeW1e|8*Zaaa@<x44tKQ83;57_CUtJLw<_F$LJPSIO=TR@6=lQ(*bhU-j+PRvl
zs#c@ZD<;jd$CUK+U)TDRU>h42Yzv%2f@*ugwrD|L%=!~YYHc(Tz%x;6L?bvsUP|k}
zUI}OY$(;J7wX*|qe`^oEB`z<IAt51=oVMd#os^iULEtXnviW<b9AYpwrp$zyP!JM=
z%8`N*A=$n)Rv_`YKJ=&a=ktq07-mYye<EHR64SX3Elv>z$@wH8yE`MT90qD8%a#KN
zSmYK5bD<sJWuvfBS2T{J=w34|-sz=<1(I{O1%>>@eOniPr*g9#Wo_fFCH-<`H?C$$
z9*DoQp1!s~{aaei$!$Flg*03B&oZ8BX6VRhmX`d87PRHrBfqM}KRsNMi&a*d#xnrj
zBtPC<?T=$x{7%oARotE1^Hm9ipTG9aV4y0|HG6|Yx(D_R3P9fO;O)+M5iS5t`4tt>
zphf{~);^LeH4X9iUt2ekiX!h^T+9M1l;h)6g|&&$>9`sf`40HmidnQ~1_p)%Uu%oO
z^xop~RxjPfnY4O)gY2>qOGbGA-@hfV(`o7GH2wU+4V5OoX#W%3r&gqX1c1ab=sTQd
zeR!=`+f4b28M0C2LI=mk?F$PB!8L93DjSdB14<nKb)e^ukzDx4`GV*N$fBjU>mj9m
zqu2<pZldC*lk}^y$bP=FOVIPdiDHHK5?08-q1NIbhmd{lYdf&{G04!&KA7!?8Y>Na
zDb;vq%u>m4b@Q9XSDFk5aOkspzZ@k4+Y|w<5jT0PFuAZ1t5g5yX(=hdq5}V!oCVN`
z$Ec{dq@;h->{~+>3#hrd6NAtRk8aNwk=UG8r8zh_Dr##>If4rc3jUZ#=cxky=4(X6
zT3-TF{k;3moeJ2=wbq6@8X6j&0?f>%M{XZJd>>)|HlghKNIr!l=ouOR0YGEbPHVDa
zVq%Q&>iT*MaH9mz=bU*31fU|0*^2MNQAD4ozRr-8b25F@D3em?jHaxeX0m4<0z6$)
z+LdmoxI_W<trLf~qZOXndM+E+pHXcHm9mwi{~7OVX8LuVi}z?_$Mmc$?Q?(7Pz?!5
zef3FBMh!r^2l!#VO1(L}^-t7v(Lbj0Elo}qYanNbsMx8}V(Ti!;wMRhnF5ePJJ_4R
zSzIW6s}sJV2$T{lNh=ldO_x(I-~KQs1ZC}s*6%;Ywt<23PSKcOLK_)`HMJV2Ru&ct
zdTZs_5&_%!|M5>Y_lh=<;CCCasb@&mV~{eZDp1eRg{R9-@EB0<{4~r>E7#X{WHGNh
zWui~|>CDEi8G2p?cGH1!oHWhF^>sDi9&_{aDLnS%#1X-@Hc$?eu5Z>GFL3J_>&sk9
zqjb-|TW?M@4Lf))j8~b-`Q4ss`dzLO#>U2e4GS}dmg&C+aLMy{o!x#xP@_ca&zO?P
z{P%ABZ(~dNv6sgi;Dw$Pz9%ZV1K5w3|0HC^4vfIvpMyl=%3O}XJ`cOvsPPBXtJaez
z&sxjQDaDuV7{LAQIBE&Bze#J}(RQC>P3S+^Cz1xP+u=630~SX~O9Sr6&FvV=isHN6
z(v>6@5_j*JD7ZqD1<FLRM+SaU7YHPVPP<2rwyg;4Yig$f$KT>=(Ce{{yQB8|SEY&6
za_$aC(3bes#YWEKUc)m=wVs1*@}7qBd=qmED|Ia!9sxiIRiZud<)jw*3z7g$Kw#2@
zu_9|)5IXF3XS}l$0@nTl+4-WL<E+K{SK8Vw?OGT9obwDA(0XZtIyK62niB$|(8=hK
zM_jG_VOKq}G+IqF!0k|UG+@l1WN}qwEf0qFTy>p0tFaivG|m&rB&KE5;|Z2GRc_5W
z9+b<2iWF{J;^XxJq$MDAOb~eB;^H2jm9;X*X%w#=t@R_gpQO7wU2O!cwufMi)!Q%f
zuX*iQLz9z}Q6Q?V*+(~qpp3GzvNkt2zhF0h98SQz0;=qKY5junOpVpa-Yj&<ndwa=
zP|o~7p-<><jbNUfG>ZhZpE4s7lC-zH)TRJ}w{IE2iv^ZUP$tpnOFjh}_q82|&UuBQ
z&nk`i&QJm$TeQT$3qb<T*FWiB`h!=~pRd~Yc#!wXrF(oPC~TApjiW4<v^I%Jiq&Cl
z6Zs)ObYd~KT~l>vM?c79d^`(zusace`qfZR{w*&6izAPyvys76Pe?{%Mf)KN3gRdq
z#tX)tywGYA&ft$oL^8Brg`0T~Th&Of(SlzJi}1FIH?NB)FpKfs6SPWt<_uD^RNR^0
z;r-MPL9&omxH+ifO!==1Z$8La8Aa(9|Iy<2pjIoSAC!B^e8$TM9dursLT=r@q(xeI
zAqPxc=nKGfnZOE&@vvhmMTPnzMf^A@MYuS#G8?}D`Tfdkhz-43(x-P{;_O*>Tlh_^
zx88Zm2zwSKNw+pcZMIm1E_f#b<(t#^4!7sMg0pie$To2|-{FS7yb@_pPgH#Ptc||T
zrX0BAjlDlQ$*GU6@w~wqI7N$R$D7ch<>^jH*VICbPAph2K-$rHOtio-AKgJx9-{>R
zX-J*5z|2>_%qz33d$=u_J=opo$>MQtt-bN$zh(0sz}&*-QjNt+qgw_4n?uw;fBsA?
zP+()Lcs2oMUq25m(XQ^BE!HUg>AXBXf1AbR*?PH7lIgJ$ilW~;tL!=XjxmA5Vvuyf
zYlj(u+jgd7ce<KnO-t;wgKH3ovQNQ7P=Md`fsBRd%b@PL4>=@S?+;CWY9Z;44wtqL
z_O5x$gtvCBQW%IC{cOGtMUPmyP->dG??gfH@s`hnCEtJ8^4PE|1uCOsfhO)RNAeNx
zp;$-0lp-c`o<{tlMkvR_;90Mtba&gIa7B~*aQFJ|Xjw3mb3A6k*K~<gmU$Hi#>^i4
znv-<ljzXAve0=p;US0=z=)3HyAA6hbmRtw#@qP+urX&_Qf1SI<kMV}AgVp>)!Vor6
zo+ZwhG;w{n5XC|vJd{c#H*Ebzi|=|22pDhRC-gX58ADQbeG-FqURi*_?B1`GlMKxt
zqrmEkCsS}i)NAF=-fG&PK-F_km#WY??jbG!l!Lq5p9`mrv5MXMAhvDYDGU;VpKmQj
z?qd9n%$+x^l@vUmCPguC>`vZ0*a35tRevLNt=!1kL}iDRL`}y9$6xRo69^eo4YfJP
zxmf@#n<n^1IG0I0zk))f!=}3WiVWGW#tN)>$_oBmFo(F(x=a{>Nb*;IxW!MOJsZ_%
z$ArLPOnUi*34-O@t2&7?j+gr<#QYJrPn-qXg%FB!CgPoq7IR2E-G3U@Jm4rnL1CaD
z_02v2WVg8op&*^l`RA^$u8PVZ=b_tUBY5=7`g@i#)uQA#Rycs^tc6hcEgg1K(9zQu
zPBt<o$kHmN_xy6*0lvy|H--OdC!t}9Yx?#Z^-Lopqt%TK(BAcI5w9kVy{xURGn@&w
z%Vb`AHZ^)_W?QBfOMo##pX=5eIm0>EH=iX+jec27bm}9hrL-b6*O0y8d9@D@|LAwE
z>1!{y-0JSv&v52niP#uQbZb%ekSGA5XyoWVya$)B^0@s$paUVe5<f?Fv+dEQbl20q
zFj=kBHg{9HAFvTS8WWKx-SyqL?+M%yOVV$i_}yV$4QK_(XBz`%)KGCNKY%*AHFZ_K
z2(tU3K=sRA#LB*<zW-U(fPRJFk<oRV6a&xv4a`bFlBbThX08OsuQh<eZ(}e{=nf9R
zGyMx0S^;B&waCIG&_odA&HQemNtuxHyz%r+Nb9um!HfO9mP1Ty1(4kZA&O$(Sbnks
z=wDHwd_#+FZ9ZZ;8ebYHmQIC?&P8TBpA8>}zIwP+Q#5?Ag4ovsod2g-?igHR+e);b
z*%n8E;R6k!d$I@R>(nc31Rk<7ZRAp$HaX30)BRnue$;te^=;dFBhyy_Y}<nu!a_)g
zo4ZnWPIxNUtwpYp{27UA#RlTV3<Vn}FDL4>R6v8gEkd+pl;x|vI_1Az#BbbSIvQhv
zO;AU#OO_};#5Dr1wa@qGFOT|J`Mgf(0XU71k5@k{pV)H2!^3k}|4SqpN!klmkshD@
z?o=fTtj>Nh^L?>~iuw<Tmt5|M!`+Ri*V(R&r6oNApx-|LR&f7&dviR*WPiRln*<mJ
zXl~+GTO$4;mT!oG0Kn(Gj)6|_dI^BvU(|yIS$;a6-arTyJnX>LYxbzDQp_LQ>KPi^
z0K`B(OSpwY(}e3?=H=z(34AdNxJ6aAv)mk<oJ?8wqob-+R8&BvQC_<^+p}>;&KYLg
z_x#L5M^|uE2Z3B5zkg0HsTn2ZxRNygfh7dF^!MUiLui@oMHN0TQ|Ju7>{M!QL|WnS
z=22`l;NJGqG^!6l2F;B_$pKFGYNMt8G{>h;Xqm5wuL(-*SLGw>uj6V>^%T3n6b4Fj
zGj>o>g357olVZ;OH!jMU3bTDQv{4NEJG{y*8j$C0`#YIoWKY)7*BGSIU3qLdRKbJw
zy{2O0qz@^9nTDK}slYF^PAtG8WCUbuh#N;6vPAt_GV|dliwL~GOi2JZMd`-M6u@rm
z3#?8Pp#*9H40n8sxBgUoWw7KA$n}EJN{nYpAOzGp1f`B+BD%R!)I&mj;raqNt--C8
zgtO+hp&~>Bl@#v8Stiu8-~&Fm+GDn8qC+Jhi$ui5TCpharFcu8q0t`<_5WE#;rc2x
zF#FyM16Z>ur}lug4qQBT<rkZTh=We`QHXSl&$r^wa*G8<_so1|U`0frU8-g=l|2Zc
ziQRi)rG%!xgezDND-b<osj?aPeXriwx119AUY~Ui3}C9Msr^u8zzKa^q#l-#fDf!M
zlk2m+1m{fOGi{yPpRIWlpcgSPF~L9~>TYX$6#DLq9!r6KlUrb5Ai|YFHPBUU5{}b&
z><7c}nHV7kHI~?)KYuP%XQaWAF%#6))ddBs3usO<ii)k&k>%xZTjU9m`<pn8QXL~8
z?p$7!l$HY8iPfLPTE3?S{owg1FgO^zhLfI@jEpJEKS5RkTn?Y>HXcG^Vj>s8WKfq%
z`6K&}A76vGL+jpRY12t(N5`Dw{eV9^J3Fa>+iu<b>7Dy@6984hAQIm?begns{FCqv
z=YG_g>?E286zdZ(*Rq=%N{n#v_J4{me$3CQNe2OrAKJ62*QpD#jB3O8owfAxrBC#o
z19vY@B>P4^Ks21U{_X{^s2)Qqe$KpanmDu+2?N>=un)VI0YG^9lc9JCjBp`57CHA%
z)`r?0i^@q+TnXNyBV0pn&@&`;i!aZskJFT5lhPxA`sZ74Z?G3OybWv*d5|{%e%gi2
z*$_%Mf31VX?#2*;!)AAH`s@+K8~k9+v?Tlu$B&&ZgLWxVdS-2)%m1u&ZM$Oy?`YfJ
z6ujdkZE2d=KBSK)z({T+FOnZz6xjnKB+A8?5ighjzG5N!_E1S%3T{<*zL<x9-#FOf
zMs{^Wq)Xi6;+-rlC8WN;rr|FQzFmx12FY^}x5OM|VE(%@tfU6|_$%6VM%w!erw9_7
zpCx}8b<rh0-Um|mXXeOfm^eh%<$QaHd@Nlbejfw9(yF26gpF6~A*Mnu>AW$x28sv=
zuQ}j>Krkwtxn^c&eu|0N*`9lKZ<FPBWp%bU8&X<oo8gR{%xUoqpd;&=;+dT^3uWc!
zA~y#}=H|iY{EkjeD_{X~MdA`mwV*YZ>eS{)gcH=?j#=Md=CZM|9YkK+-=Sk*&;Ws?
zRJUFu(tj&2X*&*J>ho-yjj~wcgeR@*!$aA2qB`1QtO979M_;Gi>mD9`>gX5aL3`n#
z&?2rq$!6Vx7mRXrGZ>yofS_)sQe&nY?FE(be^V^Z1=g*y_W0gS>$LgHn{wQE`~-Er
zbuhvX%9Eo#CkAlPT=^p{Lf8QUr-M0h92T%nFm&B+_Tbzy4uG_y#mV;e{e<I*kEzou
z29?mOfxj4*T+QgXt==k-SoeOPyHvuSAR)uJEH@$$@)n5K&=qwX1<97XPp8hl4L&!T
zcUdjh&x(ItA@Sa|2DB7V${`{r<gIi>c-!~aR8@_-&9Kh)kG|H;Pr>Z(bAq=xEd36m
zfAZwZr`?uiC;08HW6+yvJ^a(=%#e*nr=iRA<J*PhHRqvo=WnG09zGL0Q{MA_p-Tf!
zI2>Pvh$Cb2IQOoG?8Xn1>n!DF7lxD#^kk>j<08Vc6$<VJAAefh#ERbG-p2}7go#K{
zB{-V9DrXpexQK-JPDv=eOpxQNzEj|Cfo6V1T2zU4JYDRTE~4bT!?56K9r#tR7b?(#
zqUJ4^sdtOQZTpio=+KZ^FvDYIs4;me71#SnX>!1MZ7$!v_m|tV1lelQZ#tzpl+=j_
zzKa!yEpF-aC(oSp`CyRD(DO~Z2!6r`?Ck(yc>Q{HwKwjyc2yv-e?}`!r2Y;J#K`7t
z?(B?Zi=&bQ8vJ0}IXO9LReV6?(B7*(0lsQ^^AXy{P?p7HiT2d&EOlJ(LnOt}@A4H;
zyvsF<KL7sxR#i=Hy3v^~O5~1re}BKnlmi13)6vn<sN*X(SkGUhqn{J7lr+jisiOqN
zG&Bf6A!N#xit3t~Ax@DGuC3()UDxz*rL&{6)5uhPs93+2S1Col%Xvsh3jFl_%`q0N
zE214#R8V+<pQu)#2-G}gnpdxu*ViWtFP|zvUvq08pPa02Wchc^&)*s-Njwj1D%o~A
z_&hlrUo4NR(|-6d=b7GboA`^|b3Z)iRUF1V9J)o7P2yA@{z%<(tMtVq#1T|6q7!qN
zdm%J<Gaz-|ecnsn#sx!&T8^|O1lQdQv$C7Yl4tb{Gl9`iYF{Xy$BUE(XU<EW4z|sZ
zgB3XXt9y?`QP5HypRuYuk2*%$agV!p#1Uga*Laa7`bW6HQ*GC1gCVGaSoZDk;8eqO
zwf!=`aikuV=^m-r<e(^HLavFWq_)Z#BxzTU=3$k9Y!O>!>9woMrJlncf1V0S<9nwa
z;fS8^YwyUO<Xx-PCXqD6#fPC8M+H(R?2PvLtLDOzwd(VOyb+2<q@;g64^b<axxY9d
zf+Fc+p`J18NzH{yE8gCs72Nsd&2&ps#f@&o2%q@7zLwOeg-YI)CkmG)H(u&L^Q%ok
zCkbe*fT^+UUf|K-oP2w&g17Phl{cau(hR5N0*#Ib|0K@)|BBaZNl8hUoyiEh`Co@X
zO||B#%gUkv565J>$}A!>ySVstjF9F3vH<-RM(r=SON=xr?~;Ix%xV7jL)SN=(|y>=
zit2l;dKQI@g4R}HwS0La;P^&tv(9NN$Q}d-2QQqxJ32fp(&_YGZVMuYaK%W|ydWkv
zHZuzfm1!5oK>u>EB#o&wJa=F_*VoACmwH!({Zt`6#-UtY*zIg@Ta&ehmX^Ehg>GXs
zzB?uu@#aT~P;%mu&^CI?7fO^o^UK6^Ywm~E*&fPs@R-5xPSAQ359$lqluq&vM%)!8
z;ilnj_{9*UeAUV8>YFYfVJZPHLm_!P+L=L<-umct>SiimK`#I92o2i4@^#KRviJ`+
z04mC{N#UUpMC*|n4d<u%w-tk0wROekf~qwqdKHd)k@C<_<6ph6^}x92kCKW&0$Z=#
zTupDQuwPT&&nUUBo`CCcrS(AEoe|)?J%`E&`X}_XY<v~)OSu^*l|^@AujFcP>ka0<
zZMG8mOMz~YCOOnU)%9%zJ@Y%%x&7;uYj83(VXsWj4|)MYUuSXoz1I$V)6a%pW;mP_
z`?bsbmz4rB=gA+B?Ol<J%{a8*e}Q&)7ceX%i+Vi4Go72_?aL39;Wgpglw*({vbp*Z
z=$&&^$ysH2)%QjmT?(zbOeHx6`lCZmO`SX{n;^2_D1H3ohDEwx0<2_v>*54QG*0qJ
zsEXt_bEK!X!jhOm?$3lv9S^J&mPk%#l`B|+c(`FU+w3wwI4}lw8G0JLlpyp)b9&0U
zaP;BdV0Hc}iqm~B)*z-W>Z#3#5{x39>N4HC_gZ&))MnDbYv89Za>D~v1Yb8vlX7qL
zn%op-UxJJOBvVs+U4PXT<#7M&P`wcTO0xFnujZDgFV(kcZ58kB7xO8U*7%AXg>}5i
zg!EW*ex6g53UAb;o!C?eILa$LRk`H1-t}x8Q7TQ~xwziBEY&EAk%r$K7xhNP{QHLv
zLsmk6Z0r{+GMd2uYar51i;=`JjStYqb;D%-`H2IcmU*Cthi|!zqjE&ht&>kiA@aQo
zb#SP{t<7a3e&X}Wo!6k&$UIdVE;)*mW~<}wT<AxRdTTg?bu0yyh8P@vvnN$jaT|*Q
z_f);>h80(rGprh&zE`Pzxa9C2i*SN{FraHj!MQBVDb*l&X>nt&MaaFYjwjQdPezN^
z3sf|nydS)*$U?XQD5VR`gA6o2-BvkR29(Skp*g#cKsvgM+Z-GkvVy%CzOkg(bGm)*
zXK(C{x<I(C(tlyl^Hqk{r#k`%6Usr~%OnPsrr`9J(>XusaCV^6akZUYefLqtTsX-s
zeZB7w`VT{ye0nB|lt{hf7p^Yt&fh7>18$<X-h`L(|I_gUy`E>yfr0IU-4^|?LIlq?
zj{mec3VeB#diTjUWX3(2FqPGtb-*ME$1|F{CDYm^{pREa7?iAVs}(OZbjOh?Rdv1^
z3Lw4o6X@g%dJlne){S<^4N!@2g%sXq@7^0z`x=sl-kaR+=xG;NeEgIb&O8k@X88Ln
zxbi(LD~%xr^>b?0;u)Eqbhtd#+L7qoFOkU$M;3}g*K||`bR{|(RfcI9qj&<-@Ajh?
zPUHzuG0r?#whHAcI$n*>>$V<)Kj3&`cOyx$RY|AJDC@Yb2N^g6MK|)kCS(b}iGh`I
z;zpH9ba4tLK~;3&^z~7h1$Ha!jX5(j(ZLmOvy^)=r?%CyX+UNuB9Y@udRg@05&N4~
zOok+`s2Kf_{E3dqIv~HDvFZ$U+P-{WoS(-*_OKn8dS6k~rhH2_71JQ=n1*1AFQllR
z`b|H_&qIs6s|bjL?2D`DlwF!ln5ZiP<sNthdNXcl=yW8H?s-0+i3`{qEB_k@2^4ro
z1BS913TB=itW9AF82O{f*Vx~@I`!s!bNI-&_F<Vi?aU(<=}C%hYV?Iwjc`_dR8OCR
z6(bBd7qgfJz51AXj6Q<O3iL!9sFdRlJDGQsUylnWLLogXEf?e0WEccAN9HEsyR-(}
zm~Xc$Mf{q_b!1E61lH@Xe3fXB7&zCqhb%g$494tddCf?h@>!cj%}Cr|FO$Kq@(BMa
z*1&aM>1y<rHjq+n#4FaQGc{0nb+(%2^5R4%cLi^aCWh1zA@rSCQumd$XCmg9+b7m0
zk;{;VX#dSOXSk;Pr@D@JhV0$`5bW~(TfDF;G_CU$4xGUb%9uXO*=r)bCWSoE5mg(_
zPs?*?vR3=ALxr!<WeDA?toB)f9cO$I67_Pd^b%GG(xNTpgv+x^56#}zTV$*dSS3*}
zg=k+;GvSi@=<qOhWkh?hB-XON&ZXz1@HMfS!So?YiiEM*PXloe&6-PsPlhJgmd^OT
zo+dqX1=@>F#OKf73iF=yP6rQp@~@8UQ71S#Nnh}KpsjW`cdEfnigC;bTf-n-&Z$E<
zC$fZ+&sb!~P)8Bkjmmbh#~^JsD&UItu#c9@w#JL%-1`1J732JI`WSw)H3o!KOJxec
zF;h~A?Ck73yuHUtb$R{{q;0G@f1$KY%C)!A@v5h~G!`W73H%%})qAIknBMnieDKlG
zXR=yhspb7)PVV3CK1ze-5A$zXqpn9QDBTaTv(F+_dX0l$sZ?krUtRrRsCv<P&*cp*
zx7~@qAF~EahMf%^l&fRJn>TWy{gC(wMYNv|w-GLQF9?aZVltEf0w%BIhSar_^#mfz
z-X+dZ>rl4+=z+p8%g5`$uqvvXf-`Zba`_Cm?UlF5{s#r{(POF4#;R_AqeS8Qy9uQn
zT39Y`eU(e9!`%0WY@ctxn1tL=6T10GrhOvXB<8+3?!Fv==c*>cmd1$n+q+(=*LKd0
zq;lW-G~nQMtpzN?xJ>N=Ou(qby6l1MfQ^%0l+J7-VZu>2EL<k4H8R0j>PqMwE1%iU
z|8Da|^2(@{AWKG994J8F@uQsl=Gptg`;!k&|LSNSxn+<`U!08v+TA{ZmPJuPZ7wFr
z0X~laV7vbW(LJxs*7t|ruPzED9|>@timjKBL_LFHuE_^48|u)<)N*v-LFd?JrM%m{
z4(8lhH0@7gLflSRWgcK!Kkse@gL?ejW%FCtyVmk0_DW7W#HnL_{y3#KbTKZDbW(nq
ziakFA;t7*Dud2HxBK|ww?0KNM{La$_E}O~k%JiwE7Aqa0KEMOd$uFlA+@!&Iw3z&x
zB8oK_4*MQ~-SA=@;$QbjxapB44%cZTjiF`hRf}M%oURgfE8J#x=B#tt-lgT5k*SZj
z<|9>DmDn#~@~=T^dr7l!-G)e;g4{|lU9veBaH_-palS801)gaRCA|GA;u(4y&J9*)
zFIQc?_52sbIB(}Ocsr?%aLUVac(D$%CIxiRmS*L*`zG=B*hXgzS6$pz4-uQiYZAlz
z6l#dP#YvL|Ta82EIq#*F<`aX{*9f1T>E5J0o|+%_OBPvIB<WH3yLzCcMszXCO*5Jr
zcXp5O?1QiQMUyH<`o;agUjg2+EA(`M%w*~$-li?GkMrKfI~w;1nPIU$&zRFbfz}5-
zZf9fTDXYDm^n}Ea`;vJN-C_7x>9)j{pJgJwxZNjmdyPQ}TH<bt-sa`sXyRD`SEqJo
zWhtujjE&85rD6DGZR$knDIU=Y(xJtq(MJ=>bE1IldAK5DFO%^=NuGLq;+W%KsZ{(s
zR~qF!RXzjpbz^gQ?`L>-)$P1Uh3$^AcOyz|YDD?{yR%xCbRWrK|1%WRqiqQNuaw+#
z2gw$awn${ez%mg@IeS8|8*G+T{yRX4@`-!+WO+T>lP8pvlnDO*_qu-N#l>>nkz`v_
zmDE>P^<{QjfO`4)`E73ATuZ@0njDa<mmD0bvub#cvs0|*+v5zXgT*Ux7QTJz4{~*X
zB+VGqh#jpR2`#j{;97EepY6$EWi*&9!TGk7cC(v75A(9g;1(49>GQoOAlYvMUlUuJ
zJFKl!ifUEtp$Hk~s`{`br%R&#Ol)W>>~~UKj?IUzqQ&z#Pp;a@vw+3acc6o+=02;w
zZ&Cs0vrgKiPsRB=d335Hbom{sXOLuhM(l<y1|rBFsZKW*vNrZ&pAypnirdt>tK{j(
zB9uw~&D*ft>(o11@UUCYl;0_qWq)O<>M3ZG+Y_u;JxZk4dsYZn5aGH(R{VVez+!@%
zIg#H>@LjhO4W1qdoo;K!$90skqDH4V5!D-zZK}wP;#2Owqrq+(GB(|kzu$mzS(=r`
z<l!P2&VaFl_GNgT6*pH&XV2vryL$!g-7d<r`;!;=Yn`rGIU2aFa1zD4Q!3TQZ=s7q
zhA_s9rajOxdzA*V6YhU_y>ho-U**KP56#L|SFhL@<9hGvU5x9x^oU)G#b_o6)7PS`
zOw*}8HZ+0+$zTfPqSuyI6ynz>VCi63-~p(f2TtNPgIT1N-=C4RbZuCwpMeKWMfC-U
zzz=cAM&}KB78Y7Y#xTFzE5r|Wc7q@=M$g1VEhyN`$XB9Fzqtqz0oj6?%zVlx1M>Hp
z-J_1GtOH?le9H&o&FgNiXiQ5(G#<-TGzm)!(D0Egp_eQazHe(N)`z0q^S_tS!w$lV
z9=7RReE&q<dyRN(-xCXN|0RXKHKeH;tKN7DgSXvp>nGU(z(5+e_K|81h%syD(c`GJ
zi#R{<EPQVr$ug^Aqf3S)M1FZFQKG-sK;9ZMK;Z%W$#Qz8zMr;PX{3L$DQKNmH?|7Z
z*D$>ARRmF?xhG%WwSA2$SjOIz6P0{DyY9YxvwqIN)K^yLX(h^E0B!j*7}P>P*cf@9
zypQ2yA~p{b{PdT!B>C#NpU3*Gy-|NiYe(h{ZbH@7`S;xp<uiYtu%-7yIU4Vmc(UnQ
zGAaC$EcBbo#;$P>=Ggk=j;78B9KQJb2e<vbu%6Y3$H#6xXR7nrl$)%r=5=#1Y2z81
z)OJ7ZIiE#1$g22mt=1M?;d0boz4#jMm09|Yt6`}vCbNRaNkI}@Z8as!Dn*^os?<%2
zYoYo?JhEDG?>CWw%<_=uRqI?ooETuO9>f;@=O{ymiU287twsg(xSVWpukA?QE>8F@
zF(prX#yYH{9ljjlsw?*S_DlZi=<EpV^ux|1()lMkh5=UE)u8~){eQuK$~iKe+C(OK
zrUcnv?(o4BP74r@MFlw=3EJX2RVkqEe)vEMQtTiJ?QpiMz3Rk=7WBy8-aaQM2Vttg
zks3%OW9x576iPM5^5h8b&LQ{go=yWvYuxS>9XCGZ1;ZZ;Rn%1VzLaIMIHgRvA7rW)
zKfQ!eqPEU`&H15My{sdE>%jGxlMj#aW%lE?BheyB5Nvl|SH6Pqql6N0e`3&#SIZuz
zK|eV;vG#%E>_={j8-4V?b`Rq?Mg)gsv~|^zpya!$#}fiED<QrgRVCg^!^i!_NKnji
zL9T_KcW6`?2eP4yCuRaTAgr8|`27&>TNRz(+nS{#?9uG!<C;A#*8cbOO!P*xvk=_M
z>(>bUUR_i5=p7wh_>E;RUtJdc=*}$E@wEA9(zS`K#O)zbs*|^V5Nk`S+lu=>2&YJ1
zF`T=Fz1}`5Gv-xZP|B(vMB3}G<G3}xlMvZe3x9B?5|-a4gr>JOv8#_EoW5^VaKi=j
zZ-qVK*Gg9=Q~s;?)%<7zTzS7#+HBec86sYLocXMe9s{tZS09g`y7?br<z1#uBmw+W
z>9+N)wTKxsoTJeknFzrjmu;d{*~iqrFRu2fYwdhcQX?h^&XWVOi^$?2^47||fob^W
z`X=*fr4<f!%Xzn1lb58g`~F&trDrvI=^f`qm1QF{&mtRtT>GLf6%{*Ys+t?68jX$q
zDJTvTMj#SlB2_Or0EaNf<Hsi@n0&DO^dqL}i@}^ztb((Dw!&`=uYrr&5$8}0oxxBM
zixIOcgy5F@P22MggQ1lszhkyqSBtiSkqAHE4cd_EEnn8%8NS2a$#W8+MN|bDtEbip
z>qGS7;_cpoB+CX&3=EQGxdD-S@o3)BzN-e;l3AUdop^*ccMvXic2T8rPd%<j5Ah{j
z9GtdbOp^Umb-?Y!YcdCBDvXdoAPjK{*sUd8qF^FTyVZv;Gcyy6T^<7Ojm>uEjf;zm
zR^3M;V0s|jfH>6v?H{HmZDcet!>5OD0r{o7=++N}3yK@PBY!vN)mYP-@-h=i;DPSH
z)+cVYN^tPEoLvk_dgde914nJB*y7c8OzLU$+e+jHPgwAg?^6#utrlF~UP72keEjk%
zXU<5kJ_4C9?yk_4qmU0PLsHoUUn?MDxJiOaE7LE&Xzs3CJ3M}xt!rmvv1c6uzsPj)
z6|5Lc=$QZq)SY;nz462Ka{73w$O7k6l3jhNN`*B)I@mK4|Nm&Z3aBW%ZjB<KA|L|N
zAkrWpAYB3y(%m85-Hnve-3`*+4bt7+-JL^o&-j0LEnTuuoSE~U*!!t-)qcNwx$s#`
z<#(|7vrp<OpSL%(B_8WTb$E_4SUEyXwwafvB#LvqRgaN_Zd~DSvj;iK))}i2%<k^2
z@3*<s7VHBGO$-2Xdlkc6@8|4@e8^X_?Kj>wUZ!DHk{s~73P&gvHCW>4IY6v_tnP?Q
zv<X;8R?s`-x{;QKS5c0;y9CE&t|r~<(VgaG`13S8R@>VjChf4+u_R|-6rcCUHWV?i
z9d4-Fs2|>1Ug39wS%HeF=EfFNo7ipC7R$1Abx2=kF21+P8esF9zB0Fow&Fiv17h@m
zcC6{~Aa6H6K3*|&b#(NkK?|AI^q>b%52oJ!Q04Lts5FeM{D9QL0iXzPoFJ(1EnL2p
zxg^$+43xc;-<Ahn%DNyPaMVsX?UtVW7-5>ooGx6wNMK4jxV9}{lcgN!5t{yrvi@Gb
z^1^s>g~ZT3AY1omUb0Z+wt?|>>*YxkfX*fwLS+pcpvMf@l}R+~1+`QN0>&7E3dPHe
z&2p;fXa?|$u-Trh{5-w{Pn^6slc}H_q-p!RfMp{@4Zuk*xx_rCZ;%{#z#Q`eBchen
z9VNrMY#xR5k0ae+)zB+e2P2!3oS6@|`hFmGC*)a4C}_8#xo)6+9ATX5{^JWr)1<bi
zR(7Z9#G)ajui}UwUv=_<d=~`Qut8C|Fk#E%bzc`ky1v&UUrI8AH~}OOWHwN0W#O;f
ze>*+dJWsVum*9N#h6XQeAA9opQT9?}I>SFS{F_%^G6Gz@75I?9U;IlBefh7WvyhS+
zX=&ZG`_<y)dpzhC9Q3Fz9~~2u9bNaz%hChP$C)adw&LXRrj&q@H%iszo#pc5k*QfX
zq%fc(a+}OgN%tkwCuSxbqaSI{5}hX-PN$WLj{=CU;qLfeNa$r6eKk|&KHATyXrn#S
ze!GpGy)Ulvb`?b7KgP33naxy~+2r(k5zHDxY=g}kL(T$m78Hj&Jg2z!G^9q7<dQjH
z4;v9MHc8TD^3CtoDk{@91aWDar#AzGr~U`?BoG&oJKEVk7I8$SriR8VA2zsSVm#bw
zJ#IC#Ffy54XTOCOK~nhqqJw)CV_rdI-BdKf(;bxzM%JHrYsbO31KqLzfe&~BXQQAm
zofz^`SmY`-8!zXLJ;hyg95UIoi%NAT)^F!N&WomcH>^oY`6UK))p8gjgRhsEX{?Y8
zLyNV}^V}R6iTiQ4i$BBiBmU|_kSU;~6c$<@D4i=%p;DYPUZ<z3JJnERbNC+sn(j2}
z?B(#+^I4&44YKE8BFlLPn&iZsJ0MbPE{f!2WMu3b0D-HL0u2>DDJ{1wfA)ccsIs@L
z+`OaIvfLc{=4B(=*VryD^|if;)BKw3XaGX#GWF86%eG`XJ9;a+B(sC8Ib_1PJd)N>
zdn$($7Vt!%h`zGft9d}*Mpu{@%%<)((NzS^8BT?BSCi_1MWd!JsgL&V#xz*ddEvQK
zeZWRraOI9@FsKtX+0iXrBoh&d?)Qz%Y$}WHf5T6puOfNE4VV)7S4*IBSUWzJ1eIpK
z87@BlV5B6s^5rp5F(@vEgoc*XFV<LSnn}t6_})=?7R;)iKA(%YA0mijqbx>hv%l6B
zbQ#tpw6sk^%;s;+#7JK{M2h5TFMCxb-9Es;_~~D<-h{N(NAuj0F-OI2M5em?b|B98
zhM?$y?v|~B{dcdZ!891+0&{{M+YJ)+x#04~7xBfsnOkwmwm;hj91l2Hb#&Dsub1|<
ztO}DKKv*`L@E?W)E`*g4t<PNww8FaVQ5#ud_M2D$q_Ngox1u#e(Yi*E-DRzF$lP)u
zuZ5hh4QTR`BrShS^d!f{CtqD46K`x@np`@@838eMd~3dkxlynGl<$>96^J??ScxDq
zSnkffS0{cD9RBCM1t*qGWy;DSsHgdVA?pF@2nG6UAViGgc4HsU7H^-O#Rhm2kl`07
zmgoarul&PvSXgXU%O8_1z`vnl=dw_1^{)!|Ith#D`C|)eoG^KWSMzmY#j?%Da!IAh
zh+m7c+kew)S1_c{N>foOo!`p<s)uj5$4g?JJIf5nqQuYaaPD5^N&kAKe2z0sm$aQ@
zb?&rQ!cuCyn9+3&Jr!Quz5_)NDnd)Q%MNaE!nW@r)Vf!0f={z5G9ob5T>3`O32w+;
zXNSQE6<`@9I*aCi13?#NZ8g2Ewr~zNzoaZ6Ap>ravmi(%S9huI9%+m7L7eaT{<^(w
zG^dpwJqym~_zB(<xxGPdzH6in3p7x_*{uC{GI`c#Efg}hT(s_TU>qmc>Hk>;%w&!h
ztw0s|qr8>)=n~!MSJYHZdaN*s>0IS*jP4<!$9-!U5x?PdF+Mo+-{5@NiP<m*w(ePA
zE0fu68HaYTK&Pd5I9CNEy(TURU^{5zF9fan=;-w3S?lrlP_EIHM<d0McXp0!^?tW6
zI9U#m2c+I6=tIb#-U7x>-<}}I8S*9NL4UH|YkVk^j(F)Gbq7LSPls)mr|Z%cmK6m-
zZ~#zEp`FPP<8pR`lNi;*9u}NawoZi}MP`h}Bo(XC7!>`~C#ki+|6(%BCoUBIF)@_b
zeuap2cB^5FG{|<<F2Tq0g2ge!!Ani^*HVR$r;7N=F{3n<2=4o@MOznhbpR|L(bB~<
z<uWc_pFfVKhe?{hpaokxQi{nzFHapoGlGqf$vqkTm)e&9j#URku2#gfl9FNIX`ijJ
zAcV@=%8~(5)?k0Q2Aq#rSPRZZ%V7ARYPLu+@p{tJe|do7dH{P^!9|mz#K2Qy4@vqH
zczKkVjMdkw2fjs(^~FMaQeW0<M6H7WaU1wUM_J0WPE1@rmsu4k^2CR5H>KVpA(W)+
zc4k>yo@T4e&)3$FtjubG?t12FF&Ccnt!yT;%|`4l%$uhZdfPX)CZ+`nd9!C*Pz%o~
z?oEn*l_eL8PXz?;ucz{UxUk;Ay(JUeZrx?ddh!NO6(OK=dVeFX_6(2jr*6m>)Rmqy
z0toZ($e0iBQ=*c|IXHZFZm(gW@WRs~MSxB5v`ffzJLK*CJR=B}5On24iVvBhyN3oU
zgI|>!A$k&kZ3a~E#y~tHRHKKD-5X9U#A?0q5(rqx=;%Vvw}%;Rcce&=W~!9`{Y?7)
z#iMP)SYpKemuk8l-iuL9se&k`YQ;RN-|XdH_=wLbXnOuWmATux%T4^oOu(*6fYmg8
zk#;w6OkZtnB<0{QCzzU}n4W3TR>uTVse5v(+(2KwLO&7;ScvMuW4`;@OD6LTcxd;$
zb|i3cWBdU_PvFAC>{E(2#}s?lX4AHCGgc@-6z{uSXVTW6fo?+h?q`V<2+#$#w*4bZ
z%pTiKu}3ZDZ2R{esp`YU0&czEUl&wKTA4HO@`ytj{P6wV&rZ9hVu$-n7p5nwkyeMZ
zAqyLp)k6^S)3$0A!pwj(ZN3_}fo2Q8%t8ts=)c%ke{jcvo~OWfer~DQ9Kb?P-`d^|
zi-Ho`K%WHt{)G5Pt&Z+-KoV|)4(Y|y<D1wkSASJVcc}#n3^Ah6gfSD~rAUMbNUl0k
zH*y4EN1m$6E>f882?#wuEYLDB_)q+P@d>ZVXjAiUZzy3Wab|=YTt#?r9aTz0zI!*8
zi!(2xD?>)(l(KmKv<kLkyj!hz3Pboti<i;|oVVJBOYZ<5BCYUzn)vFS%bFx$bi(>{
z-#qQm1H7xihC6*f^w|i&(1b~XRTEb@?#(`;(5$abMZR%6M%Ar{xHIMZ7qLlNI3_Yx
z#6;k|Z`%sJ2bFBFXwE56!=_tki#mO$b;^?_(U=iD$eX0-7yVquWYC8~H!_wZ-_#Y}
zr^k6a-f!x0j6!S5IrK=HS!T2tJ*7^VE|92!l~DCfSz1s_fpePfkO&YB=-Xt(<Kc{Z
z<K(B<)9XViFaIkIdXFA*RH0wj3W0<vH%$WZ{ggwtlTPP%!vfak=Pso;O~&&acMWLQ
zUEh+}Cyyf>cx&g2@x}_Dn;r7hjrK};PN`>GYIWy)BW#p1@xuJ_s$-RqTX6od8SmPf
z*eHCvSA<nl=MFHq03GF|;l$rh`jzn~sdhaaib{`Y344cTMx>Ns(99!|(3^6c5=qxy
z8WT{<h?@?^pF$Fol<_;lvp9rbF7om1#2C-_*>2dT(*FhkppK!|LYkFsy>-_o*LR^m
z-SN~K{ex!SGOY0#n;x^(62oc`{;EoJ^Zw85`Su7Y{9hO~f%DMg(ZKb#xD0XRZ*Iq*
z(8_w;8s@W^*3YI`L^Fizk+$}-r79k{RgIq7h&P@^F2iJ<WtJy=Ld!k_Gw6ex_m_@K
z4Uo&WA_s&Bx-K-k4^bDn`{?>xM3McUtLfWYA@V-c|E~oY#sG|Gq00QE3jrYEDKIc*
z>K%BgO6nvm+-9m@eE=UgzAO)Ph=>UrCN<;;V@4u`YZ3s_9}xh-5R`qiUzaD|%5os8
zxm33|r8;<{HPnO&pWscPhCXX1@zo%`U4dgLCXtPc4+WePVBtfO7fxPYw71O)x*+_^
z1NHi_VD%C7pI{Bayn6-ur&5#c7hgf51&p(^GsIfKYvhA({0$6N9Fx^%);F)8@%R9)
z296qjz|guc{&|Pzs=B^1UvIR&Eak-MC(cFi_pe-}674{biFF|dZ}->OYFJkzj=zu)
zkY)4TYVROK)&Ro64#ZZ_b!!P(D0@*NlLwVz9nRO-94)YcQd4TWJe@xac-R*5C|^O>
zc7=>|e%c3OOdwh4n##tC!zB(k_VBd?RKu=fL~m1yuWu$3vl)IEi#N`n;A{n16)8TD
z>G&=W9Zj{unfwExtu@!+dcIr~Gn;~5kc)DOLUDmjzgj{DgP)pYPH)-&Hr^NMY<0w`
zGntv0yVE6vKs@X4u@U+%7ivVT5!otkKTF^7e{aU}=y&sb-&G4a@3L@}RI!7Yj<Ifb
zW*yvGGAVk|$|T7;&OZn$WIH0C>GZgSF0u43fk&SNWPv;0m^}?CcKvP&MTg)CWSvhm
zE3tAl#>OVjB*03=VgHJ@=@#q;$17#8cq^zv;)j}C+~X=g!=`;;Ebc7>4o)U|izMRz
ztw~--DZ>Xe7Yz-9WOlm@;3U{{0B3175YVkLKtrEa)3X%v*lzYULOw0>0(nPT-NC}l
z|KKAhN<avanU23|V^03dl*UEO9Y2+;t2a`k%2w@4p8uK@hW$1h0Cs5xy1d{57O=U1
z@H(5TRAL`<)_*ZSrx!np2uhfvw=gPP9(jDs*z)oa5P|e7Fx;2exGEy!;9L1xG3f8*
zBp$0%X+tlGEstJo+@dUti02F)C22YY?Z;lNj2XFRuKxk}Mpl09cI;^)hmsCB;uYH1
z{*~WlX10a<vG%Yl6UwF$|IJcPB~YrW*V`J&TTf3;Mz5^=I5^mAiT8s(B2Ghe(oNHp
zJH%K>3jfFG)zNKb&b#5RvC-|A!GgX(n_7E!M$Brrw#k*3?gS7@ki3y0&k}lz^hf5j
zJIDfPNw8slA$y+^-14YY9(v1NS(M9yoe|){K8I?;bJ3XxfC&0CQXr;s)~FDwiVKpb
zZtiA}K{gWq@XYJubc?mA2YMiIS{f_<uucUSns<O$l6Y|U>^LvuEWZ~34e`<Fz;Q5O
z=E8(YJCc<HpE`VZa8$lfNldH~K0A5YSOKumk01d5<;D5|&o6)OGe&OWC3CqXVA$=I
zX!P{0Kw>r*FWLULO2L8{^p~4|1`Hyyb}kRMmmlk(-{+i|pkhi~9wBiEE>fkxZK}9%
zy%+kF>}v7-=s9?;5waif>)S$Ov?rt@X93rD@y9?2P&ln*qH+DRc0D$HA7|vaj_HlN
z8Oqh+bLh|XY&f$zqZnL2A&>9(YKHaM)-~RJnqPWUH|Z`y5{SRKKkb;-?al%4l9zvK
z=L*+F@O(rA43HPuf_ffpOGs;dE_kEVQGw?H0EFc(pU+*od-`51pFV?@=I;AK!2#j#
zoPPU^6vfRcuP~J-#u&`}yAA0Z8L25`8=OTfwJjp@yghSuFfRu1oagr?pp}jff~r05
zGiu0z&KWu4iOZvaibTBNJb7yRauSQvQw<NR1L>BAr1aYaC_T8dvy(r+w6rw$zepKK
zG>{8yo)L&!J4|roeDPmo7v8M5^|;`;7xeH1f?lR!I(}&x4U^Q00`K2*gaLBJ{b6hf
zhKqXEi;<DShU5E`@Xd`FApiTuuJzjyo8mBWL;Hj=bMqq_Nb#*aVu=5XS(9d><Zvj*
zkBH#cg!&D@u-}`KdWKLGmwkJg=^iw&gNwttQSVwdjeC}n_Eh7rB}YPKVX!_Ms8vwd
zXid%`KMCDsQ1ouEj9CHvc)F!RWe}MA!wArV%v2HX$u8?{(4dzkyhvB3q@><n=TTf|
zcY|kaxs3_#_Nqh2l+@QCP?VKnv!_fNe$?Qf#?;i?Q9VYrui9mL0%sNI@TW&d%l4|p
z$j0p*{~hQSpz*fW-9?5PVdzZFEgwcgGP(cu8oa2RW&}FDv@Jg>jJx+m2x)hj%Ub(W
znY`<)wpe_iMSI4?!F^r9>vkLk(C?1q-yH)(ks6E&AH%1f7=6WrQlC|@q*-XU`v!M^
z4;HTVW;AmL!Yiz&OR+Fv*0b?1Z4ftE`G#6%Y(>d(9bQS$neN;RYXYdC1+k(imG_)%
zb!9#wsP?Tr&7P>VgiNwqRDev#!{eLPyJp_yiQm2b`oz&j51IeNjRvK(1<BkrJiJ4n
zl*Emc*~XjZ%8c45A~an6+Un@S9_Bs`4aD``H**>3>GwU_Z0|mfnH0_`_8S}_NPYbZ
zQ(s>XQi&G#^InSoxs<F!BbO+wSC^7XJNy@+V!6zE2hr+#EowNwzw2DI&=d?57`=g@
z3&49alAO#odcrtUe7h+COw5j>)ey$u1e0I9C0eGNe9{WM6-u4BjeHeEF;6~lJ$6XA
z3!2Lv@IVM+r2SB-T42*aSb2EdJ~M3aAzBpL{?21wfb$+9Kj5kGg<=7ykJo4Va%Rdt
zc&;^~yLMgv*trY+QXm5lKwf|dlHxt60T!mc%YlonWfL}_+(L3GHg>~`)R7c-*`zyK
zr$t-WtZzb7qiZbd6t@z3j(S->3HSGA4V&1o5u%BrAh^(Sas{XQq3Fh`QWqj2g^7@0
z?e(0NmuZ{V4Eb;|2H&{3$fV1iO~h)RuZO5ZZr>gy1O&WBf8E)X;+F=evaD+c)oH8M
zzV!q?d}s`c843~fScmumNfM0XRqEmn7U))?IzBdS!dC`5`HPl^ZGP)8ju&plywvIz
zWRd%$+0(L6(P@q%!YbPz+05+)IpCrtNeS#6X(W|!&&0*=Zkm3>H#JB~w)0Lz0FD8_
z$m76h3Jk~Jay!K8mWF_fT-k#sa}xGjGVln1;w})VTkbEY(e+G>IrkHB#y~+2x?ar+
zks}3$a)#AYo^9)u{@|p?^S9~q++?=%-Qci;QVwE5p#btJN25!VF@BV?I$r<Bj#kw)
zc31q1OIkJ0m7<EjKFuM~*JUU!!fgr0@zTIojNR|k9XOoACR8=)vX~j77L6Tm?TpvZ
zz&{!*giGJ8Z1$S&?ezzU8tlnI#c)J;c={pHzcoBtKp`LIN>9Mg=mKJv643icU9ezY
z85GO8c<8H9!60k!S20Mae0yNFYmj|6+~uhf>^&_-64B)$H#*ALRcuBGCZ%v_eQ_uD
z5)YUQPUii0K~H<wa~J2lMYN<`SFEqLS)gJLa3Dho{--%54enY`+VNYK{ta`|x)fxf
zrD{WT;Yq&>;sp-;uvna4KqZs-n3Do#y1t9uF?4rm-D5~^kFMnRM)AL6_M5^X$6x>7
z{nu5ctJC4$`(j5+rj{m{&Db-sud^azV`+utL*FTaM%>ViQdC6bH9!nEfy;O^*<q=5
zb~c`_`PvJJqz48D!X{3IyqlVus&h!76UITESCAH_YpN(Q=`&-Qd69p7jOyV7(0GLb
zRiOD~hn3nh3OY&?nWStg;c2`a|6r3y9c;yvh6}X=D<kX1yYO#hohEWDje>Uk${M`9
z&a_|(?O^Zg0PHTszqKhv+gq7+7An<v<8Ja!VAdBvB^jR-wM|X@Kt=pc+T<S^M!+r-
z9U%EQVOn~3nC6bvgPU^ZJU1M;Oebi0YmArZ9m?3O=vkekOr9mv3ls;+AZCGxh$t#L
z8i=pEM$-A?>2=@%s#8#QoFy6yrFTH202wJM919BzG#yWruE{5BvqjT66$E2=dV1!c
zy+KAs7S4KdcLzpaG-AP^+}j$JB7LB8fcmu8yF)YOsCJi{I2a9vbnmZEYc9<C$^bs+
z<>jT+;?4!-&%wJ;QdLF6!s?+eY4j?wAnse@A^}yBot0Hx76`C1qNHoOkZHOr=wu-Q
zqud7jRXQWjj%wP(@nPmEVlkJufP$u^`KL;-SazV?SmmHNwaH?BR6bd!QGVY#Xv7zv
z@`I#r961j_3k8S<dWXv9Y1vU^py;7S3n~;lx)l3YIoEbe2nj^Sg@2dqj<Gc{n$E>~
zw{mHo>|}c`F-6E`KaJKXk`>YHcJ~es&$a)UPNT60WbNDlk#tXAUp=WZVuIQnF#Fu>
z6_l9YxH&qqf+QH%t62jOKCpg#tZBDASjUki<1drfNUD)_c4qDO2AdC%1T<v;uj~gv
z(kTS84aJUdmTqU8=)0kw*qBes)}nVHuS=m=trz4GC<7HguxK|NHokI5`{oUu%P)VJ
z>cD@sG8c)|iQX_0`HZESDDTn&g0Vs2=hRz7?MmGj`2FQ#TW^bo?o-6*fznJM>~iSp
z#BxcPtrez-z%I>U;5pNJ&!sKS7oI8^fwp)oWY@ZIJG#fe7NNA-myFBhhlk5`wVNcQ
zP69{)2BvzizkW}-8cGhYGK$psw;O{xZhEOAA=pc(zQ$Z8iMJ?eG`cr`ZBZP6?!>ID
zalq?>hKkAyvcIUJ1ODfb1Ev>=D#b23U|c9^1{dQV&fBbYyaEALAWd&$bF)ml^L1Z;
z|L%MZov*Ji49HP|t5Q*qJ`tCY5G+zqd<ep0E2p|bq-*>N{4)0T_DP&h?|k`q1q9&I
z)6*->7Zbo-`)s!-SA-`D(v*}d^pQYl5(RodzRpae6C+T#H(pMueiIThIgjG`40bk1
zOS}`bj&Q`nPQcYN)DqJ$CfV#{gO;}J3SE<EFiH*8x$d73iC>SMgU~|%Ovq=dzCiyO
z2ok0PT}C5G$s9K-LNovJ3=W7-UZ3?~W~d{;E=#Zq+K+!DUoZhU(82T)mdHpcx50<=
z{Scz@@bR0`=8F`r3=L|UqQx(p31d2vQzisVl2YQ%rkiyp3slI^_X8TB$FQ&%<gW-O
z1!SG{Q<JT;6FzZCYOh5^KGWB_Y!*tSawGi(?tBYM&9s~xApk#K1Jk6%zO0f`k4Z_B
zA8@q+4!3nB`z9eFfx%`=EJ}p5LVr;Ea&MCJ?w~>*q8Ou)o7cgoqJjgAX*QD!z<E}1
zbbfic2_U<fxw&$^{`cD2+LjIu7Gq_NS92yc4?`0ZQS>@p&~Px2j?@mEZ~2FJrQ^<C
zH6Yt+4jBZIfP<6J{{_T-$;XVnwdy?42<>yGv{n>`9)S3td%9>I8_(CmS8lguqjjVz
z|H>=kKxii9H}BH}#T(G=tsk)9oX6mgFqWrPehvf>DflU|wN8AWLZ@V)MtXPMrVo-7
z;r`lKcC&BGYsmFwv++PxJ22Hp&novgQ4^=sM_9~w1(b<ycIioc4?i4Tr~uLJ%GlLg
zGP>m8iWZlmvco2|&*jXdTx4+>NhCp^@J!4!BZIETL8RA@<4u{W43&Rl4=!fY!!%UL
zQNh_^VT;N7$-VJa!TnKP{6yhM6yyDSL4bzuPi})96FB~sT+atpBN|u|KWjaQ4ltN@
zH6)Jv)+>LX!4NoYgoD3}{zOCL1Y)9YZud)|<{G+|d+W%MEM{=H!2JMwz{gfqB<YNQ
zkNxOv%PH5h+R{TbIgH?fXSjCPLILM8mdE?@Iw03hG&4+?q$5eGv-(E8N4!p~{|Xix
zV#NxpZ}GQj8%?R^th=KN2@HeL{GAb~^n=R?eZc$iE;QMmIkOUa$5#WcDBESTF)Xs7
zqN81}$<urbh7YJg!5vVksOKT0S|bHB^$0(V_d~KG35(bz|4au(i>bX81?&vv5bcl2
z3p#%`>%6J!cX=<C%ebTp<bBNVYNfZ4R)7xZrw1Xze>1mL&q({tSNrcq_%!<p*zJ86
zM=>`zJno$A?Cg!#Yk_aD8GJ7Hrz7Lyev8PT6JSSE>unDv{wgjuui}aBo%;RzH~b*^
zr%!J@wc7nAXY^P8{)7QBN!7jGGHg(3)!P1kU|E^fBs~^3_OrLx*t2Omj!OSAunR;*
z^)72G+akaNO|-woT<c+O%?f0mNPmF^$HdeC$hRw6IE<qs&LTL0qEmaMpy6}Hc`j44
zZzc-=GOG`ORVk<kp8wSeBLol$lY>mb>{(aIA3>5I`viY;9ryt#thCPxuWx^mP!?ex
z>K>TS*XyUK5QY(}Vv_2)?U!Mo{q<rr6$qp?Oy>8#!P9IzUlR=A7w&$^lJvg&fAtt9
zCWkm%g($YNcz~>w#!ywG>!+aL<Gnrgk;31>fq^<;bsa1=#D0$iS&;{7@*d1wTo2XU
zd3g!LlT^{KefgvhB{$)XSuFl25GpGt*ch3qpTjprC;rjX*Jr*_;=bJ>2GMhW*E$11
z<XI+wR(iWnN^7ZKEk#7UdxwTr=X^<}SfbH{%MX<*wl(D_$V$r~3)i~-0Qw0YuSEMl
zMdCx3q6ZK9tQdL;NPQybfhJ}a<6X~S$Wy~@Jq}#{jNQs>Mf6_oFzJxm6SzHt$|Ozh
z`tOEpoQ~Ooid0dr*YttW$i&Q};_4^eI@bjfsPkqnN0+OrnLe}VZ*-6I7Ht6qn;Ocp
z>akUE3P><)(d46wq}LyPqg$_1BK+?T5+F-p^yV<?3o;8eQdu&xP<#G5I-vD2xstvJ
z-o_Fzc=__c8xA#m$Iz+=<m8ZpK*T@5Ojffw`RUUa7q>$7hO3z1vHZI_pmp%xl8J9D
z61z|!XHl#et%#mHtO~`o{xmu|>UzEU3z+6AjK;~v#>WeoRb&e$qheW(0(yRR;)|Z-
zPa02^XoiAZOkiFDp+4(dTYia&^E8Le4_B&NgNfaZPUm2RYh}lwaeF=uwAPvwM$+Gb
z6NzY=Y2$thWwu2Z-6@YrQ{hh`1u>A5A6yqA2+Y2kskH+vsbZq>foxbz|BYRD*gnqK
zxj(qf0O{^noobwD#XmJ%WE05P2(z4+NFpbVk=?4WJ&$BtgqCZ!EY3{Ej1FAIOx5ae
z@LH?F<{bVWelAA@m;HEWRC>BCzga)<tD&G9Dk^jPu0hqU$kn}?O2D5b!(jdRMXkQt
z>;&&WOTnEarX)&|nHj4Um+$)bUSN2XDNuG5kv0CIovJaHq~X20zYpTKx`7#$&Et_1
zgc9=R$uW35-mA{p9W7vc9?s*sPpYanHZi$)9xS)CbTK2K5WGm!=}v&0<_?#=fP;H>
zayhNZXn&w&YHAwRzbQ=e4yL@KBGH)@rIMrkPgZ4RRI4|F4M-eipMQ^uc@FeQu%E5g
zzTH1O9C!SSbzd!Rs60f6NhbgFw@^U!%7e$RqZFS+eZkkQ;TfwVQfJ2;MqLUKsS{LF
z_JQ7(y3>gg<~qa`vt^X9^b}w6-Ee1w3m2lzulW}2md5Q*paDCCHPA8CU(T&dn5!SJ
zpU><_x%Q^!)dJmuyK3S@zrSu*_K(d4RPy81U>qT;=}v<gPA^5S{!X);J+Z%6PFf^e
zG(|I4M^tV0G#l0<3_umEeLRW+5R(ZiQh`~263r#av8+YFWQ;J<|7e>YUR#<jkpAx$
zwYNt+GX6f09G))5z?Ri$w_mRkm{E!Gli@p+tR6pZ%Z#7e*yxE36F3-V2akjNuCEni
zm-^laDDcqv-uU!*S3^@+7+hD!1_R2YO2g5$>N{9i*aB@j5KdRk4AREIX~Y5(ghijR
zw65Ur$=5eQ{LI$aao|0jhT8EpN+w6S7bIMWAW!7z3M3eSKp*001s=K0NVvzrDd_gp
z_8?ItgJ$yuFJ28rs6<PN_4f18x+S}$l4o%^TrcV!Ta!I4LTs?X@iaf%?S5ne(&U+}
z)dfLU9LmtR(E+(|mUR+G$g>YyC!NOK0c?e{l>=F_gLcmQnUlefie7v4L%|kfxb1c8
zSYg7y^R$wl&akJwF(E_Uz`PI@@Dc0<;wSkAP!01H9rmDqi^j@aow)EHk$NuF9??4<
zj9&GBM~u&}k&B<OJDZa~1F_==W}0Ki0&b^G7O`=M7rceS`*lA#S-l@4b6qkkl;nAc
zf`*`iP}BT^1$iorSFc_{>mQhZyf!xoyM=cDFq)d0Ab3v{Xfh`sva+(&nqA{TP?$An
zT2yauZ*HEQfW)&*DGKw%(55CXAS{4Q0}MEj!^uaBfn=%-7swIlYoKNQe02@&g5ge?
z?cp4TC(Dwf_sGad{@_e`d3hWNz0!0&6=)h>Zx`ODb_4;uz+9;cvjbuP?jXlPRM8(J
zd-E45kGII7wXMxIvnv`7^wS!eLoH);RGnkJm~|=o-HJd$$=$xS%A`xXtLXK;!8!^w
z{h%nN|7`X%zKI+-Ip{w(T_##l8;C!YB>4mxclSw*m`eTF9IIs4iheM`+fK0W0t9ns
zPn-Y!#EYm+jLZCRldT_}al;0mFPemaAzDI3_Z{4>6X15mOb3AM==F5UVfa3aY4ks=
z8pvl>!oFt}x*gLvH{Y|XmZ<5CEaTqjb$p#uQJI*@5zE^s2(xi@y&y6EaZ^Ayn(5Zi
zz_Oe8O+euEsNs0Zv8AEGrOK(|>EYNzS68>x@l-ooEM7+E31l-Yo*k2?IC=rI`)MzE
z>IwM$A+k42BnwCr@{9$Jz|-uArdDJH!3ZE)zNYhFBpnun&Rd+7y?F6rBuA1k1dmO|
ztnL?RU(ebQ=_as_wQ_*P2Bb@?*xiAMk9@05b5fmk4~M;G+uexj9O8c^>o@6`Ema#g
zwE6j?NRiDMr;DPwnL=9l>;x=88!?5Dj*kinE&mCxn9NNwkyf}|M>m*u6+_bdVJs?b
z|6E`eE^cn}{^woGMG%r9&p_%``?v4p{}n>^)mvM5H@Bi)9@npOGD}8IV_V`DnrYU-
zzSG$Z$Lt4>PDxdpkl3A4aC&b|l+<zm{KD4udiPQ>jy-LeEAbAL*l`ScuW^}8p;3sS
z8x#{0I|0oiNTo0UzS^C=y`?G1kKreqv6^QKHp8a?sVH-Sn1dja&tJZDw6}W$tb$A|
zPD;L{Mg|%v0rK_1DY*gE$Qv6QAcR?ZxstvT_;kBN2@Swxgm&$Kl>nW>VuOPLh(Q4%
zcX07yQyV(Kri;gF>4`zDPzr)mL1=9dSO?zlNNZqo&jHU3RIrFZHAWps?M)V_`CX=c
z%)ah^nNn9*u0>lIT2ddy+((%f<Gm(cH+~H+?5V8I6!#z$JEXxy7aHz>ff4ww5xFhr
zPVCp5r&R^nPs9pS3=1yLcoVNy(!0JBFUJ#7Y<1Wz7nLmRev2V;J|4c0t7ACz4w0Pj
z&)7VVxg7p@@c9Iud=#Z{UDDT~OEIcGYQS}dxT%r@p<xx&MM(xsc=u7jZW};f`MGOh
z6i3t`ku19Hzj9xxs<NEK@O0a?!U|EdP`!J3Rr(&EKwjr=RlT8fWtV(+q;K;`yd3BD
z_7=Lg@?M%+Ol6DXzNb;?$PftJTg(CCZh6JEX=tLA`2uTKFs|ZpZ7|#WI<+BibjEYJ
z*jGW8lN{fhXI;LqaCLQ^G~;UWdk9xnP{`h#g-${~u3S01Yh%XkM>f9i8l#60j}uJz
zEkWCqxzJ}NL4NGT63+ndD)2EfB9*{fCF|SArU%u-+dj#L=Thq;ihw-@xeUFM4PL(d
z>n6inIRevc$X_*7zFZxCVY&b~K%J{4Wf}PskUd@{)YtpHXT1w<bJ<CZt1(6S`!w5d
zUP{q_^mBb#8S6^@>s!<p7~iaF796aotHjL8#0JHv=PMY-on2(9D+bEl%EV<EZKzom
zSQeyEc{|`h*`sR^vkgJ$E#L{)!@}!4uPZL@iofF?cG8FMW?LAp3+{y5&`ar@>Lp0d
zOrE`&irVldpS`(UPMWbscWb?TV$Q=v{nn4bK5DFFFrjj;IelP*Ir})Nvi8XK)%^BS
zzy<<Cmw;p}`HSAI%!a06ht}sU-%FxC{hV=ny$x6~tS_F?4GtB~5RG?3Sz$5Jz7<Wd
z%_#)7!G>L7A^5)yv2+h@eRJOjVmdA^EC`E>!~He<5>q_VBwgSK(tF3|k#LkCETrIn
zJmNICB0K3vOO;AeeCb(~iNGD)u(orA@>N5DA+MW7){-*NEVW4cI270MI<fb<2APtj
zxJyv<>;7Ab<6v@d8TnEx{@~E5=>QeX^J8<Ft@WP{zpU*&KK9ZV<jb!#Stq3-r|WSO
zOb|ukYZj5R>0_iXrhdFRF|PLD3AsNwhpBb^B}Jagbw$g)6KLIAXX#bxW0o{?O6<?!
zk?|e!s*;fYS5$2<)}`;4J$E+?8WXBM%9WBlo_7uGI55FODbB0mtS)frWo5)qV@|Y5
zGedQ8BtxAfW#{`ISZA$D_o~c=V^-*qZ8100vsX99LC&yL(KyE|$hEhA={A;`6fE9>
z2FMY1WPU}^4=I_rb9Za0vb3B>CZYP0$meZ_f{l&!Mz{u-(;r1ilbny=g3BC#)&SAn
z{b>X|PuKrz0lEple(#6V)Lgdi)@)j<ktOf<u<GsW(*|Kp6aik5k!WE5sl28Kk8mO7
z>Y>_q(&8aVn_*R>1iR>0bD&3u=f2A8I5GGLlp@__LfN_~>!@P_*Ho7!x2Tx*zNdxP
z!KQ<|O{aM3EYk<iEu6;)zy}QxRd={gfQxZUYA%|+ouyjK;Lb5h-5xL=Pot|9Yn!t8
znw9bWI7TnzRb#;`ds<`?LXGxjy%N+4{r42FkyMt23S5@knl#SAE2CrT+PI#_b`-jY
z){3WwJ8i*GiT4T`rn{LZKexE53%uvN(qVWvyuPv})bFzV@%l(;WbbjgPg!7<Zw>j1
zF172XHu(Nz<nB$%1M47sRmp_a8v*ASZU|}jG~u_*3Uckg{XP(<;(_Kawq1YdtN55L
zmudGw)L7j^noGDonBo4L+dglnbFXp6sp^lL??fytW=hlBPNAj4?jaCSseHi{kdEVW
zVTSj9mO1)+)UeXf6+EAs64{233g<aV3Nkb;E;6g*%7+(~jWq-AWyV!8o(lX}oMt@Y
z^<4hi45!n$Jl6Wl{??(R&W;9n`?Z!-w+=(yW}h5PmBswD?Vs4OJ!vE1Gh|}bXU>sO
zKKxN-w5ChhI!REL=SKef6T?C}E|K@#Wp15dMScq_SVFX${G&8F8hwF5#^~Ib+kMGJ
zQZ2f}({BdEP<9u^1c*yK5u^5r>4+bqKgajLWiJ;ux`aj9x(q{c>LKcl>C^KbVNS`W
zljb*WcQ3{3Lt2_^5z}N=@yGn#>56k6eusX>7W?F}t15BWY$aJE6VTR|hsf6a_+dz+
zb%BE^ff@1u@k^XybAcP*Ei3VNT_rTdU!1*MWUuzPbYL1*fX-dMm(ue};BEmRI6F8H
zx!vUg1)#1$>_rrHfk1l*J_j_22L_~_T<lHuXVih~2%<rx34?CW+rTaizL!=mjM!rL
z?Ct~s8E>cdv5(xZWhb7Bsc1q~3NqJ;Q_<!gDA6PRHM<7jB6t1P%?`|V&-XsVFw<a3
z*c+1ucs|LXqj7ko2;cUqPnh9+EznBt$6wx_L>QUdWi8ca&w<btTv)-pt9I)jwX-R+
ztIp|0+_Xq~JC`Hh+m%O!!Ih%ICw15NU_CXunEFU((KcA33p?dZ<CQ<zG8OUY?tqH9
zHoS^E=e;ys{B2vx9%aoWR}M_F5adkmZrv~C8?$s2eUuaqed#>*?<!o$k<H!TINFIC
zNv;K4`(1uK^R+WHJ(xLZWr#pYDV*T^H(Ep;$*pT}8bU%s?stbQX@9u*Kx8>IoXEMU
z*>(U7m?OxuE;j0ghNH_n$yT3jY;7?(9H}`&j>G{s@aetcOQMd$<5bA|Ogf6T3SN`B
z&^XzuUMdU+u50?EM^yA}%+uk6tWD%ltg)Y0p;vz>ARnopTv!Zr6cikPw=KM3c4hau
zE?Qn$SsnO2OfIkV=F)hhdnG=TS9<2+1{P0)5izn?>Ig3~J?_CnOm7~dBb_I|z~jJa
ze#zONTw&IVgD#Nn)S{S6C|$MaW_FOY%e0F!frA@j!TL?3Ym;iQ<kjvgN<6iCd{qR8
zdPya<Jie#N1Y3N`_6CBFJdD5aHJ2UjiqeILvqNKGN6hX^SZi<OcfqyCb0h8@LTMDH
zaZTvD2wMM*Rm&%h6*w$TM@yr9-=Ls3f}sR%;y-IZvW@HQZcdiRBVYw~=4(m>Y~nZ^
zXqlOrg9v$=b+f|3tUN987Oi<UvegXfGwqc^t?KfthO{93{VAYIg-3XE`;tBAP}J&?
zbjrgT#bbE4fA_&7jjr!u`GWuvHJjf<|FUkcNox|<1ttMfBlbgof9<W4997B5&qJ)+
zRn`e+<BV7dt((fN4vU)ruY(Dbu1!5s8`E*u7;&V{k$i_ywi$ZY$8UV~F)h5R?h3TO
z6c5^_T9bc2E0L$lc^dA_K?RFZNK>lgC+Yk^fYx`H_vYW|DZdbYhU3U#*Y7lByilLO
zp>gq;^XZkc=g6(~xUykm3cN-2&YiC<Q0_Kbu*0L}4rXg&NtvkCt<e78IwUeC474b>
z?*wsryFjxI3UhXzhYQ(w9nctXFL^w<;V>7@9gSmv=xQF9s{<{-D`g#igsx!Ei)$P7
zZH(0wjZkOt!Irj6jSBafhHEKHw&1|;Mvm||3zmf<nc2fp2qzX81Gmbe!v5+)A_wep
zR801?ONMsH=(;)D6hn0Q%lf)3&dp!YT@<^iHYe|w8ctrH>K$^lSdp=Z9P)%9{GA$w
zRTV;cZNo@0)FQ{uT`S5OZV{9wVR@67@jU4_dyuw_;m8)Z?NzaS-@b7$JI_#=___bG
z<?kUh_&EHWIdtFaVEQx})Ui|gigwn?SAWbS3I22xuZ!6}2L*u!%ht6Y|JUBSMA4$D
zcMGei$e&UL@BO?dkP0m9KmBh;eLIxafz$8=+*pO>Djzg66)4JB9ggq>Gf@c%37<cI
zURqOAb4`zkh#1dg@*bKF25x84?*zq--g2{>9(VS=%lAL6Khkn@dk#2@27i2dDqrO2
zjh>#+WzhPX63Y?4o3D{i7WoRU?nON_`>jLo^0B*CQ{Mu5VL2GC=5#K$x@|@6r;zM+
zv~Y|eqX`Z#zr}=i5XiGN7Q8jni&tl~+39iUw#e1j%;sN(AzMDn<H(`#q_`IG=McMO
zGI!4M6DNto5b~qr2NE+LL3<H00jGFFTNPmJdU-S!Rr+XlVu7!=yG}w#wfnG#In~g|
zras7+#>9@`AcLIBS2De9d0v|klb9kdq4FN=7c&hDwO=){-NjuZ-!+76#F5<3`>H`4
z`&MPYe&<;<#wRaiL5PuDR~w+b`BdXYor(tU&0GAoVO)KZOu+p;-2I(8-x6Y#ioo&{
z5<Gu=jRl-wX4!>H$=eM3{L$yf80^U*pictEfPU8npf0KO<p*Bus*tK5t<PU0B07N!
zOa?+wF)_PUEt=$y;@aByK&sXNw_g<oru?x}aqyd4)qQ>?rRhmCy>Exsr*b$jAxD#%
z4?8>3Ps?$b`4<r5Ol}&(i5xf>vdmYROP;~WM+x*X*DiA<@^9UXyhI+;D0UD7%F3qI
zWoL@}tY~xeo;`6`ROHFAI$uXdCkb5ne!^Qw&c_@UQ{JY1z9P4HzUndj-QIZXayTy|
zEYkiJu1l<Ok!$lUD;p+luR$HgZr~;(-+k!)r`Ir<t-nv<&78j-N<|+(!NF}>G&)!8
zUf!C#uXsshoSG=JQuJG+q(+#Yl$JIm7?*_sC?M3-3m|YJFNwY#{=pJ`v?`E^Z81{T
zQ}eeX&1RdxVKNRbUi&$jC8M=2)ifwp;!0oIW0P~s;@9h3@I4BLm9SIk>IPJLDtdL0
zq&%Esw*Q~f)}rW+G`uC#=U}CerSW<jTg<#iNAKwHN9%VLlST)I7aLo4e84A9vWGCu
z&d8a?p3I5ywx+S9jT%Z7nCLF^S|xr?2+9&b3s&$}vBPa9MEBkB{!U`b+7Z9=!g`E0
zeVDGTECi9@=h6!Y6T!%(BFmG~p@3f1pF&^1?|U~E?per0d+g%D93eGQKSA;{EbTQv
zp5h`~gFnEyX;DygjrL;<e<0+mXUeonjgdcUa$&4>Y=9*lPFA#gPWk#Kb_jr@R;Skd
zS!F}R35g}-KV~6bfuXAL$8h9yisi%V9a2f53I>lYgl82ZA?pX~S61H!8REsTf(0v?
zRpf*N{W#)V)94PzKJWyBIi5L{JF2vT>9;l|KMHv&z$~7fo68toU_VAEQlMV8^Ud8f
z7XJx})pRG~;7Ej8ej3PCK*x>6s4o^D`u%6k6(!AyF_R%v)s_%n9PcP~7Mw<rpo{23
zW+4MuMB0l3%WO^Lj|q27Dyp<((WWnAN_Riw>eH+rb-Ro^@p+gY6x3v#+fURAAy&IJ
zCrqp-iaJvy+{LF;B!vb(Py0UnKKj<>^5F5?)Zl&c2CV;6gKQRG=BU-FE)}L^_PdtV
zx48Ornd@lbwpjC+MTbS_kSYsD7#Pqfq0Q*W&05hW^9X85u-mooWmZ%yxr{7Mqaiqu
zGYk`CP=FL9A(RT8EP~MIf9W_&2OxZlrMJ4N4Z?e;gV!v_nuqJx%t&Fw*3_S|kig&p
z;P}s9mtTxnNLg-<n+DQYZ=Ap2ATg^r(M{-MxIGOI57W_{?AfxLnpq@!7?@P4?BdgX
zqfs=;S4Kd_gQ2k5Jd73<v9YlMw-ZsofPDB-B$wxipgz%mvQSmxdcYJDBnAi7t3&Q*
zPg}5-#b`b%I{a3vdJGX)VdxJW<kcR-Q^N|VE%bijh-9lO!{c+i%W{uJI2lz{MC4rI
z7k?HVnHuK{S&hi6BV(Q5ab576#bNsh3n>n6-MRMlA49ZdEYA_iE9WC8Or}tB*nj3u
zsstd?S`k@c=44>nXm9f|yHzG#BmKg9y!n>Yv^dZ=I)(!=Y^8Evn`kDj#AU_}9%>wv
zL34PJdskwBD4MOPM-`kP$n_CqeIR^y!9)f1XT=)&mfFg~s7Gm84e7}dnO=y|LIoUX
zI!LUhCATmgF?=H%JZ@Mte4L=2*GnREw1j5ZqGYePHNh&fD@|>NpIXY?1VaCf%5W&I
z(%q0nE|bb!==JYX4arOe&AaJ_p(Tu#Y_}7WxnJ*LH4OG^Lc?)td@Y0q*8F*v{`T$<
z?Wq#X)nEuc&Ov*t@#9V;d5z8j3bHF27ktm`hj|57_jxZS)1-3*<Dw8`6c4_2+V$f7
zyL`aqwm5m*Cqh7PWMX>bjJ~aXb@I?N8>D*0eq}yoQLuWjE*Pmn!`i(1whVWBv~ihQ
zJ|sNVEoJ7CFmuY-@aF3_PEWb<9Oq0hf4{w@;ikvS-4hCx`TUwL*6V2w&@ERIjIR4p
z7bjj>wBOE(LQgi(lE8)D%F~&X3p@>VG9u*SDx^Bo;3zmXL&v7$aH9xKQ<cM)DiM`o
zXpZtXhg^|S8??LyHx5W|c>w1M!e4MqTyQHlQTyZ~NgM{OA3I>CPwnm8EvU5JaXP45
zL7r^#(S$b;_g@;bcqLVSdM|l@5Her7yY$5E*Cp70Q!5UdzmmRll!r@XqgSe<JC$!-
z4dm){&@!}VShvB&@Cdgpjs9edDTS7@beO4K7an?ym#Pc=Q?s@)oOf^m7%apMDv(}p
z-Yo4WLW>A@YsKiTgg2evXs|-kSyRcHRM+eg3ke_=vtc96MLjK1ZPmx!Q5x3j$~lmy
z*0SK#44cx-{Xe`ccn15rlCl0kRrC8+`$gwXY`$$$Kc+<KdCn`C3+EemIpewkng2Sd
zj45=-ew(c6po=6_z6r8`(T;=~4*_~gTh|R6*LG)l=VH(g^{+3l&ZaT-B1l?W_Fmx(
zTufzFjng|`t*6P99fz{~R%WcQ_T>^-9qD_r{LyJEl?N?VE9-GFCxV?}P4w?yf8FkG
zY3Hcb$+DT4nHe>^r>nhTA~iC=D=WyP=Bgn}UQB%%;MJ=%P#=n5&2285wS|GHTcP#%
zcI;jCGs{!mIU3nYg6k^^wFge&xrVr+Wv&ouO{F}ioD2_x^Jx^=&)}`FPFYbp+^dIP
zef(p>SS}lS4oPax!p)W^0Y!BS&w;qK#(M{zrC%$wtI-MM|LezV85dCDG>Ck1gSDrR
zh`EU1j#omy>|HoUZ=2X&Ej*0mz?j}#&_!{WVpvr09`4>xct(6Y9-F?t%WPKTY<lur
zXOs}AxN2WI(0nAzm9uqmek*=gMG3*8_c+*)v4(dSzk54j?ig)TS!TCZ9B%XM_wB|o
zRS9j?m&nl$fY&@Tbe!f#-_$#UByk_yUMjz)&533cqQ_>dyTg1;8jaX!H?cf53i{>o
z0&^|WU)5tpKih*pBxQ*6db5Jnsm@xv8|D5o)gAyJZtdF~EbHj1Ep9>Lh<-YsXw!~O
z+Fu2ls;^qf;V?nmE?zv+;6_*((|I&61^@^|ivc~BO+(W%9@e9|v+v3N17yu80ohRW
zSaEQ9LE^ue!7*7+&<OhEoZsrQ<rx)DEwKd){8z65wsievk58pd0l-kJ%TzNFluKvM
z+&nKz{OoRQ`XHbvHV`nyUARwJ!<|DDa_~S$KVyt!BdiG*-41=0RyO*1<Dtw4#Z*!2
zex~XUW?)a=#aDBj+Y!EUO?&JaZAZ?^;-8e+vi$vnN3@M>gLTjYn@ipYSs0=K2qRtx
zcMG*<m(dl+u9f<Kpc<l0ZBj~49kB`Jz;}cXX`4;9*ypg6W&1OgKUlk`4u}gaI(<hI
zTN}P}rCjS!iIRD=&bkRHK~BHoEk5K;e|kqg{Phd7{f0s6cANj<AM(Rz0<BAv;VuRt
z*#KRN<_<{?gbjVXcxSTo7!4a(Q+~ev{TrQ2Fo5N1Cy@lUcO<}R7;k`L^*93`<o9s`
z2Wc@CS1dLMUc?0czedsyz=Up_(TbnSjp8**O6-HqfWNXryE6cX%MajXash3<hutHF
zW~Ye9szp-IXpTu~w5Z8zx|gg#x3J}Xoc+Bd=Kf|5(qX~ckrkyxo7JJgBB7I+^*uFa
z;ciK12k(F?$b=z*wQ8T0`HFVJY^+FyZOX!l4s#f9W#dYz;Ju@yIkl6<UQI2gXP0`H
zG>`UM`hsO*E6DI4g3ynbjuT8aa>huYJusv{2_9qqc$sj?UsOh9HGer=!Nhh~dwN|`
zVtJ4XdvI+`vyDuO+H#q=-|f^eYOO@h9un9wHQU_RJvujMIouFBrmw45?OGqvQ6$6$
zyybT5wjuUk8fu8uQwe+kK2$OE{lB?&CtvWF!{KNFhi>RKkx#9Lv^kML9bD}cl0bn?
zuEHmlmTVZ^99@-<Rr3);y&(u&(WXD5YcDCtnF-0L*<fKpZ#BN`q4G*~mr`{O^w$PS
z1EGwD&IC@P@#glH%O+F3QW5K<-6q$X#<2(2k3;o+>uGPoarX|qvB)b+HkzMa{YEo0
z;9_VEl;bdo8FptiwC=7UY@tvX>oK%ZOmrQcl7_r(Y1-(3(w-?Ffod`qu0+N=iLmI1
zpJg+i6`{Dz;(M|zh0U`CTP0i5oMGXboZ?m*L<U#MD`z*KUej%_uB-+ndf@NXAPedc
zW2(OT-%RKuU&iryyT&XRmW%*^XqPG-b;@Q8wY~!T!nqHiA8l1{qoksG{&#9>s*qJN
zMKalYQzO%rx<zm^IZ~vkm7^9GK+#kf{ZpgfNf+N<m!F##JwKLDYE&~C%hB6);Wfz=
zFwnVCvkS5Ju{F~_tp%qZ*ynz8@K?QV+}+)c>Ej3GKgLW0+jFTn)d!4Z-#Sui`NPHJ
zeEfa;oH%rukfe`HS8rlvXuU*y6H5@9#c@~W3wfh&Q|B6D8wCZNCOv`!9^HcR3h!|y
zd)VrF@BZ5<i1IU4;4~B!6-76}3ue4F#Xl^yvHQS6{K`!Wnc03wzq?O_`;Dh6m!xQx
z!#w2>o%XcD{_jV&fQ5?#hHkKvjk(K79^ot^;@$-)Ua~var-2%b<C0a79~Gy_iTdX=
zhT%n=lJEu({Im@&Svt<K16FXR_4Km`X67)eH0)KCKo9tTA2oSy?iKj*v$u$VWvr!`
zAlG(CVUa>EBf`@Q5f!ure*o!kk1tugWr9eaF<74<?|uv91-^CTXZS)iX`N|#jtCAU
zEb9!o+xM+GO*z)~%l@y|^O=ls?BBFYMRnik;0{VkNweClEKm3J#hz+9g{nCX$58{c
zUrf+@a#-kpvzy2mcDI!pV7GZE&kkAZt?B4Z^vBW$A9Cg7pOP0rmGCEZAs8$N*LK{A
zuQ4^%#MF?|`z#B3?BEyG*n)|7%k7;U8@E6CmsYJ+I7dY_*DoIR?oWCU79gvvV3$@k
zzt4u9d)Gqg7dLslx?Gs5jBAq|tlXOwmFdWmCzSjlm1?-=k25MS0FJ_9N`nF405)v3
zO{{=k;emvBm@kK#kLUlLfNwLj%e?@sXD%!5T-CwC=^2;<u`ylGkxwb18_=4j8-^Ki
zJ8P2Pqf^CAT6FN*r1J>O`1(tTJF{E1C02dmAvRD_$C<mzr-3lxm$uyX`lbb&1IUQg
z`O>gW%jvxr5mOrNbGR(e*1dNB$^Pkz6NW1VS6kudOawTV7E_uqJ(GDP+oa?!#g{{h
zQ&q>-h(7-AT1-@B7x>?bDb}Kx096Q)LemWzPsnZ9@LfwM3*;%bGm={e0@<&N0>S3!
zzPZelC$EV*%tW#jMm=uZbE|N!J~{tnM2JyzE(UeepeA<@7@HS%FSBOTu9)I>i@KFe
zf6NAI>M&r{nRE#G*n8w#qsddXIdh9QAS`t&6&lQ*Yidp4nzO1ENi+WFYR{c8w!gnX
z-wt_{-Nn1~Wi!1u#R5qGmm^eOP{UGz0;UQep74h_k<obVTO|ZtN@`NObvaiwq_+^o
zksZ^Wrzw5~Iwiz^b9Am>_Y3NS4LTttt}WwLYWZ=^VA33h(5*i@I8RX_!_T1LZ|a3(
zupXT&jjx6kFbe*h{_$DuJQb|vzH?Cg@jhYuYBQ9tR@?!k+1VGarf{RZFl92jg1*2K
z<wSdiTry)!>0x~(JK&Kt0I+VS7E7t6DUym#?vAP3r=Zj?tk^x!tgKon<HhImchz}8
z#!tV`P4E>=4lE>r&7nF+XCCUO`Jd-jxTgT19bsEHkJru>$=v!WNxrhB`3tNj-OB@W
zH4!$IBRu~9WA81)s$7G%VJuWa3_?&qrIb{<!2;>-66u!iPy_@7l$7o+Dd|u|x;qx#
z-M!$O1@7lN-v959_s7GrIkw{Bjw|MxInQ%u%z2Mazc;0{d&NJdT(!5wb-4xIG0s6$
z^ZF+x!OaqheD8L)V;h@Guxv|fn}4laV=4~eyBJ<sWAnJ`^sCpC%ymWgSIxp7Gj8pb
zo>qK9Lg?u>b$qZ@G&#`69XtH3*KwVbojtC9N#r!^u@7Fstrwq%C7O@8G__Kv+MkSB
zzOTcYEOrr_YSY><<3-b3duwbQ{kTD#g2w;{V0elrhqjWqi^bP9_fsV_*Jia>WJ(rt
zaCKC0ock+&rwhN1JQ}5-upvGoFiA#`@iY;So^-~{-#1a(nUz>pIV7r2#P3LJRy>Z}
zsVx<M2|oNLLuz15Nl?=Kn23VbRckCQ81Q#X+8R6ejlQ~jF=^pTXm0#zf5Tf6He>es
z=D_ZSi^C%l(Q&%~YsN9Opn}Wk@4bJ_kP*NR4h$@ySiIklR8)>|&z#a%baHkav1<Y?
zkEXP#FLTz2{OTCk7<%5Oy#CVzpML!_-*K?$#@gJ0$#19C*saSG?n6pc2ZkL>JN-TV
zrh8f{TTZ5e@(i#aKBZLwX1soVN3Zs3GdnOVS5F-x-CZlyW)&S31UdOw+yJ6~rDZwi
zjHMkdMumKZkz@UG1501WCW;e#XmS}(R9W@W;FEB$BEei$Gsx(5Y_l3qoKTB*TPSq%
z?}wT0ru$3t3orVGE*We|+V&LBFB$@3LUMGVV{=;a!6Y{Kp~2dtz+L*XiTxi{*$W3<
zl3)@>|Im7z#beO`JJULzWCl)}8Y$RiJ{vG<iq+2jKt;s=gRF(~SFB7`Em_0Wv30k|
zpu(H@Z+FagL0lTvhxOu`<^}7I&V(BW<XsX;@12m(#n-&4=6+N7SYn!QTGWwsc>U#h
zbZwyz3wY_G?xG0H4N#6TuS!?)Q{!3RvptLMKLul8R-p05pKx#QXm?Epa#T$`0Gjx#
z<;3m7kMH0}-N4$6H*d%PWQAS-cl4*j0pc=Ibg2K-BLk*<8ma{Uo1}OzXVaNC8R-)(
z^K0wsqtRFHbtl=1EM7iR=2AN|UQ*4(YL#JkRQ%nlc`B5=Cs%tN`aWRR-L2xOjacAg
z(%lGFNfA@lF&-(&lP4aIn6u={clZEh0y>ADz%iXYi8tw`@R>B43hRdrb#c<jwg?)E
zuHXtqRzJ)6lSLaV>FV?XndyA2;*lN11Iqvr8kT?NXna<FA@w=pNUw1v_0q3Aj5-93
z=h5%KSU?1kYi?;0SqJ>k_nB#IZmvZjh`G49pwCUGVKe<;wnlZ#g-hs9p<4-H(~(_A
zWMpK6kpGb?5&O4)Z>SCr`ZAGIQU=Dvw1Ez1f7~5MvB(gwv9S@_fqAM|I$V%cRc-lr
z1rw!L7ZMhJL`(YxT0VV;5{dcmCw_OG2Hbr%8L`jCx!YUc+T5%KGL&Ob1%gtjqsrM)
z-FZh2fN_Oc`%qh6h5I}X!KjoR*OP-(E7j8zB&wCD_vez`qVA;~&>Q7xH{M1y76a9A
znoR1AN{5|yo70Vj`sPN)h9)^GTV)b30|}n01|ph_qb0FH3Ev-=L<wlNUBRuhnleBA
zTXIK5?fj@Vr{Qqj#tKhm;J=d^xQjI_nfX?`@sWRGW&+)JwK~ww6LQ-<<>24|o_or3
zYPK!fS3Oq`<YWCK8tU!J?!VmBbLXI?0p-)DUr~~@q$Kif2Ohxl-9|=KL9X+i+ny05
zBousnv7n~e0PRu-$V-4}^Hk5R*DgCb>h$w>o<a4Hrg-qc4Z0ecLo1}stu1d5pShl%
zRP2p9>ddN+9()(2M76MizTJO8n92@%ACL(`cZH2Xz1Yr#*O*;hT{+=fkhszXm4m8|
zH$|ZJF{-^0X#P>vXA6x71EJ<nsD0M32d*7#zY!@wq|h2#*{xfV_DYTSkC#m3E4o8T
zV+&`B2^Ke(ws?>CljvNLsl-<ms$QNX2RDXkE&ZnjFq)lK=f)1+P6QGawbB*az3Bvn
zO|7hQ)#G+e<D`%bqhj0J-4d6cf}oU<wA$%_5!#{rhBPail?<1iktCs@yF3oe(6FLj
zn=$?|28_&=x6l*&KA%ezbUF1W;ZatOLR2|(L5c^wK>hL-5tj`uq#aSw(P0NQzIy)r
z0*bUmL$OR@^l~dc4FIDunT>J-(9CzTPc}F>2;=4B>+2m5fRa<C_4bEyyF+pkbn}5r
zCVMO{FB^w4X!s{4lI2$J5<#<O-Ppt;p${l}len%}>5{A=-dunQzmzE8Os^QCe|PoY
zGaqj+F0peUg$S$UNN{|W^n+g|CShs<#}6kW5p^8RvHbpLL_U(D0@D5_{Mnu3_eN`m
zQYp?X`XAyIX$yfyKTrQ&%;}^nq$-<CmK_ARa(sEz^U&WFvR!D^E51W@qa_jzNZmpU
zqLQe}aZ@6!bljT-CN|j=QI?jo$Pg76_<ODT)Zfn!8?r(Y47yXG)m5P~9Wj@UHi*Dx
zV{zAEPj14Bm{?iy6Hfl!*5)rr>KyDv;&3fac%z`8z#h8EqYV`3;Uo(9{jRIK5+dvY
z7xN=xnXKF!t!-{zim-<=fZU)St*@^KxY`#kQdj2zLS9IKAVPz6)od7@%JSbp{X%#j
zX1|1;^|w)tFK-q}f+BF;biXk?(>{Di5)t|9>w=fC8IH@+g9+!Dt3lr2t>*0_V^G?c
z*QEs4_4}`JDO{zA4LNt-ZGRqTjRfAy!Pk%7HTzQ0GO}2oaWPxI8JL6%itxhL0o5Rr
zMdRKb(9*2~Hu{jo13?2u;Bs)*y0vxX$>JZ#WrNg>s|EhoDKf2^QQe%;#G}}y78F5B
zc=au$8{jWR4uh?UTb?_%(cCTPg|5r5rD(%3(IBr2Ki~;Wmg>stGiUZY(KWlIv47GS
zlEmL~If$!%OcboyahId!UM!A7e&%3a*`dA1R~-wwhY~woJ~ZOLyQPhTXU^TVhbMeo
zVXg#f(?l7hw5dr9*_F+$8V%1tCk^(l7SyKgP*qXG!4*;vK~aq=E&-ia3f{39_N28O
z;gIrT@k9M8Q}$wwYM14a((bIO!-J5@Uq$7PLR@~(#a6g}sJeREaiF(dOsDSe-_4cc
zv3yemC#e6Q(9@$oUZJ^9K%g-5y%?havw}|LXN6|L)MA=<LLLmUc~QqGo%!~AN(Lsc
zGO@Bw+f5>7&LfODj^69qHJK_0bBS*NY@3*#5l%F1HGA>*UqrQ8a_A5|!Y+emf@r~!
z)&5xc2txV+&Os&0QtE4%n6>@=VIV5BocMjEqO$S=YKx&k{#V9eo;9ah_@17w?!{A@
zlF%7plGzS@W}Q)uKEp|$q}0?)<5N>I$@ZxJB9Ij^<6oLTF|zXJClVhXv|fle_?5cu
z4d_+6ZD%Ny>{G4)JTItEg2A8q8W|(d^sRfJ1t+Ui+vDbMjFjl@A{&h2!jHdS1iLC6
zv9rI4jNwvqd42&X`2qczCF^Y6(T_srcdl4tXf+>){Gh>H&HOtzaAMpMd2O{5x8*Uv
ze3)?ZBuFi1JqXm_HU?6NMzP;|EhRk6^u54hXK_HMRC<8Ute}9#L(T{~*j&yCm2h(6
z?oO4chYbdm7(>LPIVV<E{RI-ACriioFPSMGSE+qRUbi~lH&-%&FmUo`;mTGoS|JJd
zC7{^lV$9~2#hM*c@&HLPR{0Szu|=i)U%c}Qug8q|V^A1o=~af8FG!dfwF+3S?DJf4
z`{9yJ7x@+s(&PdO)a^c?3?8WI;UM`7{c)iwIVR-BA&*2}N3b!JQk#Q0NOm?qt+rn3
zQ#vl0)kl;~TOVw-^U)xD*{P{9+yZzHKR<lwsk*lxg4a^a1d9a}(xEK?4}2<lvVG+>
zxha>AUo_K0;@06Pv^c4`XUl4v7*s#+*GZ-c8tG~MI``l{-IR;(OIz6N;qR+J(zNhh
zA!WW|f~PQU+TupW(PN+(22pD7o0yP^r6rL_R+9lJs|6};r&Ht}TE%vk2#ZC<Dx&<-
zI*4AGpmt?cR8)@rG4wtHolcQ}znYx$Ves;YO{i7+etF>8@87?vqk<w;@|9xz#0JL(
zd5_m$HdE+R7#dMMfPFh63c!>mE$euHA)8_LLD&R{U)Hf#yyJg-4<<{c4U0RsF~4cE
z1#OZ|yU{+keki5-CF?=c9xz=aNcB89VXW|~lu?!nqa);VT|hZE-}zj5*Q+N}k`IoK
zn8236d9pB6L_;MJ^GXU=l=2Q}yHla635$iQWl3r2#_q7$rH&jmE=XEYE)$CE)Sre1
z02NzpT>XXb^~b7sBsZf-xb1#e$rSk#Ffi&g1$XTd6f>dX)=wpgs^YB|YwF4m{d!a@
z+%Gf;Rhs8gpZSmv|Fp*EbuvzF39&nH*77Z^&zd*%gz+^d)ig2N&D_O}oJTexnU{OE
zXUSFlKU}(c5hWQ8jE!wilS=$ioO?X_P_78H21KmJ4e+nmZrp%m%fwE7{q7WOdU|^9
ztrjM<wB9}jimHvj_rBa?^RKAjhLW+Ka4<qbQntJZ)H4QEJPqCCMXSXgZ)l7d7!vXt
z4s&70wZ3#2FUZ3T!S68B>vNTZN%$O@#$J65EjwH03K@;A))E%Kl)7e-)<5toqrJD?
zz1}tQMYN~2<{E|FG8!$rTK)@#Q?pKq3-q|ZGMMcapE?D8YD(%@YyD<*f4D#6Kx1Nh
z`VVv@5<cZNf7iNt1Zh*E&@7MCY5js&1oQL6&W;Lu%afzAQ;YTS=T6QdtU2ZRjo~cm
zy@sgv<q3(2_QxCbkcBIl*m?8j&6*PPsey7^23}s?<S9rQ0QvRXljDO=US5gvndu6-
zB%ltXq@?tuigoJp(bFPiGl`U_R1f&|HtoHt*a<7jAim$Wx)3*Rp$y2K_k?17vhbGh
zAzD4aMAw1dKj)f5II)+y(i|_xJm<x`dh6VZ1~-y*MCOU8>upt34;*z1d}(g0&EKNH
z;WUWvxaN6UWR?a}*|B~vnQF0}+|7cdG1t?|E-g0G5xd%Y@Z_lRYWw;^2{fF(!S_a4
zeU5QH{>__`3t9?RBamjC_*kd7^YNNf4r*G*T#igUrqIAIfN~KPn(@a1_9OI$Sv8rD
zF<=i)f>{0?WSoFZT-G55HieW7H<8_?uipxbhw=a$MH?raS4-dU7(6gOdWQ|tpg)p|
zs<Yo+Ps;C}%KzGSSl}fpG*ex<e-^=gD!DVrIo!RsGcm%Gf|F;IAoQ726oMj)ZiYll
zi<-fDKlGKzn?vOkfHFLntHT~#eE|C(B>&f=#RcEL|54O^ODFByMKf{n8{q2bLgEC=
zEp-^=r;!pt4kbO-R8u|bpiunVSnxr~+@eb&(0=!U>4~*4{b@$cAkv#QW}DESFsxJG
z<ht<@+DYr-#B%RVzqSt_m$#CrRwsK-lF3QmjC*;Fr6ST|UR6E&y?BrN%wN4`UT%@T
z)P@Myn^>(k(sj=tdc?yMjcPycu&oa&oDRbxTzvf7czAU$ad<A*IIdNCfw06PC4m;7
zVV&CN1%xaH=EWe*K_#4gA>k=2{IvmU@z4JLx@}P$z--4ju5v)5NZkcl)UH@cmwkn`
zQ1kOjCP(2WF(=P%k&@<(6P8PEpMTjZ><^!ON~@)f*(VgJ{NQQJ^J-YL;JwvewW%t9
z^RGgZL(<!}$d;Jb?FF_I0`$?XZ~uH0lAiwOclX>4YW%~sR#6MzGP6^c(6Qb^6vW0s
zBc+6Y`m+QQeQj1pt_Lqm7ei*RfFF^yn1}2B`b0|%PvoCJua=jWQ7z;ksDr%CKV1u`
zq3K^>r@zfbaCNfSS!8pfS1!hNb{s=|S@m;e<Q-z_flm*~Hn>G%k8UT9@*mwI?KByY
z(80IOl}u;(WaSj>n6g?|m-_6w)U&KerGrtG^QC`HN)GnheNL;-dNKB%@D%a#nH$^N
z+VdUpkjWPcSszV{2$k=3m=)l<Yb~tc@b1Tr5L7>RM|Xml59Xl@8pcCng7JdY*<cQo
zay|t_F7ELw=VyUEQ&J%@j=;S9bdqDV9UsQgT2gJBjXTqlK53&N2ksfA%k8SHjH)E+
z(%Ym)X-JVshc;s23@iWtdB9t@ZlP}Wr@mf`nW(X$;VpDVdnF){5J*y2%nSl^R7p_>
zE-3Qf@=U=ji529gXdoau0L2S!v_EPSrRcg3b?+=Vb`Z?21DAJS_27nQLd8A*eu5bb
zphHhi9^ZDBqDK)!-6tl^Jb%4`9=Q|kW`vMb#<mBQRGFPu$LvWsn(^^_rI5LQp5&)A
zB<-mC9^@G$7e_}$37cuCEhyk7Z?{AB&g4Dl7pi(_50h46P$p4DKtReS-@fm0sX5gg
z(A&#Dt8m6TV%STLE@9Y*H%0{vH!L`dJZWj=&+bLhG(g?VVNBK2j$Ppg_**Ko^{Bz5
z#~#Z|{=_RO+8j=!-^n@UO6D-&XB(zR92Pz#`ARD)+98mTp%caHOymacXxHRgug4gq
zJ1cP#qatoq)baXD&2Nxdk0Z}$?-oIkTS?lLejM3!ea<*Ko>)>J!8ltFbG)G6-mP9z
zM#k6l%>xDx(tDLJDq~L!c-lt|IDP6PEBj6?a=uYPtfaMZ(F|0o1PtnzLHBkI@QS#k
zfclV&X1cn%1|N#!;^M4wl<BxW!Jw5K0KCL#JbKxpy8uj4%RyplX;j1&w@jciCN3~9
zl6=r24im1YEeRHxm~k3GMrvtk9N6n@U>LSPd$1*|Tr~ObuAkDJLHoG^^-!Tu2KqQ8
z-1cND;J884hYcvSec^}wmM$#hD>WzJam_)e;?eWn6EX|VbLLgf?!^ynxOjLM-WnMh
z$!Cpx3G{opqUMwl-5+vS;T@d!=cB5i=+tNo>AJ#ZpvMgxC1EvITv)k&hw=whN`RcT
zW~enPyyiOTO?9@$WQEOLZV~v$PW8002kudUo}aibNtRpc^Sus^P}N@v3@zC<w9cAs
z)z4aKSyS~aS0!(QzxI$~7T2D@G9x<Z-%Ng1aj#QYw^{5--`=$ybFLr%<$hs(cXzgx
zZEPA!p*-Z{tKL4L`CVH}`0U-M<Zb|3?@5bd07=hLV#Os?KKYdJOF#E2yylcT-Tc!O
zoS5g<*}=&rY8;NGABA6wJ3R7v_ad3`)HWlGf&Ak5q|5<!AFyEsM_$<XRmuY}H|BC2
zE5QBweYG$djPm^b7MlD2t@a<h=cyb`Bih)go`aD2wGO#Y-@?Ka98SMK(?aFh5&8W^
z4fU&VK<o~-de`3g@mz9i`{l>!H#{*u>l<I<Hu8m>QG87cBtY!zN327<1Ae?RB>YB3
zS~D$mD=!keC&%(1HeS7rLb$Cj4o489s2eoYnU=nB72iw;k(G?APJfVPy>a7LE5m&i
zL0b$<hCu7pr|EgAaI@N+wZ3k%#W%mN7BO-0{Wp_+utph1N;Kkylixwz;V*<to{$Xz
zXxNv4fPg_HHbg1+>)GzV15zQL;c<faTNP#Om=8dCr?K?on_*gZd5>-4dG)vHLUzU*
z4ksUA6WtS23qlwPiX%IC>4bQ{Hstwj%`?tkEqJF*QU8^wny{Xee!tjQaaFNS|6S{F
z^=+gGE_FzRKr~6JQgxX57WR?rB|OF;Ezo5ICGk|}Z67cQ`EiNl5_o+S){&1n-_?p;
zU%s06i9i25>{;iyF25LO$O7>K;86TN<>1RvAMigCiBWQ^iu)|@uJ?4_53$ci`d-bP
zpr&om_w$EbrtN6O#mhgZM&ilwC87fI2HtwW5T~^^m=2V4T7dkiOJX%+)2U8dd(W)h
zJ)z^E@g$9T#?{#DSf8V_`jzIKtd|ReS;E%Tn7YQu70LAVw6Tw?Q?AMF)4c-&<v)<V
zEX(_|vD9;q|LvFGWw0lsd7U}UstzdOTi34*mS+Za8;$lxQv0(->_XCwTXbvT^66D+
zAco&|rNC%M&*muS`-V+QUw(biXQ-xcFovWZ=fZ2vspF%DW^S2-MxWoKV;ZwgBLZMK
zx@Lcw&fMV3JGeGnt>_Nn4^m@L6!BM$#`^B*EOKFQ(g9=wS+SMD1xI&iViW)EsB4)r
zPki2WnoL@A5<66qxHWmQ>iHKv_IUx2g&mOocX^#kKq?ct5i8dxJ`oxpC;3&wt*s9~
zd;3F9@2L7drrGH><_8ncPOaaQ)EYi9VfLXqGB2gVeY}ygm>(Apdn8jP<t1fhFOQV0
zNa{TC`TZS-_ua&y-#~d^@|=taFqpEy!AOyCxj6XziN-WHx6+=YBJKJ!2I>?)$~)PV
z++xcMOjfQIx1O(LoE%s=Z4oRhvVZWO5F0$6pq1^)xPX?}sd%W~F{Ix)IG+{u?FXfM
z=NogWTe~-Q)4s-j<5GIA`Ov5FvYWfRdq}~pzEBNGHvo-Thlm#39-H0*y;-v}H^7_6
z$4+Qb(a~n<nIU?u5p^*<j*4Yg=|wW2DY}_isE>*c#LS*PH9NrQF{Cr`YGuxBPfHYZ
z{(V&CHu`pa3y;R=fr!+-bXn>uyAvgWy%n+NHq@EAHk6CL-;zkizBF2ylm}E~Y5V6K
zeWu905>#UOuqq}dEAk-L!nYi4X^-^5?lxfs(Wsc@@YR;40;#aPk^|4YDSg_`jqAE~
z|IL;87UFjGyNY~={=Wh?sRS>2z7Xd@?Z4K#Tm#am%8#x7-Dz8QZ)6|ixUP=WWl{ZY
zpca&R!kHbKtNvZoD@9Wj>GBYidEW|cZ?XOqXLmr4>*bLA8ZE3YZ?O=isRs3se%l>$
z^H6^|?QcV%F|RWAEsJ%*>pqMYxKlYTC%-)Mx}4Il)raiglAfGJA^tQpncFsi$Yp)^
zzL~%}Y)yN^&0)eh6R*Bh{3Zw(--rr5V4@<BAkY&5w-yZw?&KlK2s<n9@nxyB`w*88
zxG;x`JjqMe`slR8IG?loWf2k7=l6n|*wJ>x9PKz_H(lO;=Qo+R`X+U`KXI=7{r}7s
zn*_)LsKld6a`bXK^=T`e0k7y;%rEB$@DxfWppqLGoSl#4bJszr&t`G*juK5g20Mee
zxs%MT+2DGTuCDM01=;YRO5&hZ;y+1_R{_dZDlMHRz?)c^^=xYyb)I9;O^{AEw_>Oc
zdFfGVkB;^=B3vn({h#0U&@{#n&klkT<stQWS}5})=fh^um9&@y!^4)u7QD?VXOvm}
zJaBgGZXNo|D`XyzV*!t>9VEn=TSO+r4ZP(~?u(O_%Pp!>-db~1VOTTy_huz(Bex^7
zJUl(~O@>)tyr6dF$hi4J)+4}+3+*6G8pIC5U&>DZ;{G05bSTkcTzt|}4r-C)H4_PY
zvww%tpOpk<8ao^a*wuVJKYzXk;Xah%BnF*qFo#02^u+kM8x*E!>s2i!|8x`2fCTp%
zzw)vCi*i2T&VcC7ZzGQP_r>I0mA-tZ*i-id_Gcbv&%*s|W__NXWp)q-Ubto^iIk`O
zpMmf<EY{yYWhN3c_`vq)Q3B@#92xH%icLMlGbs)r<iO1D94wE(o0HwXjT`jsf^E^u
zY-uHSk%uiz>p8o$qZJ`RjxO2qANHnWG%l3Pt(Mb<MErN)6X|Z?V*`Z)^}alUgBLZ>
z=|0^IoisLU|GcX9gubNZThUr1Uag{P&99I(w5LBck|?_GBq*=0*s;aX(?|K7L__xA
zcRaEkJ6w=d8H`W8%<ZG2i;Dm6@)7SgF6QRunk~2ja7Sam)alh}Py{%zFSzD%ky($8
zHTKEv!}SR56dXBO%1<!RA7<4Jj>e?p1YCZ*6^0|F`c*CYz=MNe;ILH-t;H^DdhXSk
z6ZUziMw>Cua!wq|@Q4X~x%(W-=@2sO-<04zAcwj}OyS|-ed1D|4a=X5jbmV7G&D5u
z=k{Of0IVR=ToGJBfw%x1Qi+Hf8B*6K&x4l>2R*k@Vn+6MZN?vS#dAU<rC-ZPNSt^N
z79zi9-;~?x;kl$#z!u?}uvcJe>gphVc*&6<7GrKkCq@%l;n>*9UHc;^+P*|D0QK>+
zd1Dcg!i8#HzkAYUfM;N^Yyn9(@YY(tKV27%=EP%Q$c>Cjz8R{yfmflRdhpz^FN+XV
z@Zi<lynkQx{rh#6DP9P8AVdCl@DrJ*cd=MPAe2B`|KsLT!8v)sXNzHDPw?Ba((xm}
z%FWnMK%15oGw7jpkFO3f7(T-^`lr2nic**M)r>w1k&DX;o-!K-WGjlxS0EzpaJle5
zBW%8J#<=M1t6bw#Jp9#u_oi9|+f*b$>`RKu<Z}YngdPgEUnQ|~o2d1=LrNN;P+P9Q
zS963CrS2iA{2j->q#OnLUR&6|OIe7Dj{qnsfE{I&s-B68j=C4vf!`sHu5Fbhqp^de
z6;yFAm15KE_r@+RyucV-d^_+%vk>BBO6l7#K1;9p+kg4$;KF^haMAyS$qQ7769j#j
zptyvitS*k>0q7S&tV6LhatrMd5qh5X&^_kuY@?ZFkCJD$R8$GPQFWh6xAt|kK;=j*
zw%zrlC*ba<NDUcdVxGq>tq*7po<c${{0Z&tam`6f^nOyI?)gL&R|2Z#us#0^pb~SC
z@77CUru7#x99Mf_LFN|-G}%qoVuj#>KulbMa*^5xn=@E<pVkNu_R~XR+_vEf7l&75
zyg*f>9~3DoEE!G|#7U8e6%Qccc^WR-@3+#s<QgU@`1&=97crONCK>c<_-SIdHwVFm
z7B1wznX~HlT2B!j^iPFAF}PmP`}_UH2*;3Ox$*Po@&2L+_p>-sCj67%v)^k$6>wN)
zPE-^DSNOg$Kvrln(f}VlGi;hlJ_%2CTkh_zB79-0=8-KL0~}#vHor!T=+*u@NkpTY
z3_xlTH-MMRqm_|RpdvE3=w`3p#>F)#akmff?}d2)Fo6saJOcrnXxCXyM@qJBWMy%o
z8ed?t49pVafxW}i(?OvYdqZv&9MXyT^zI~W)jy;1Ue#d)YL6m1|BQ_d8DMCJ(zizD
zx$L@@zF)W<dE@W<*%zEj92|CKWjE)iUIv8TvSa5`;SxJH$h5P-)V+h|@$B}OJC4Hf
z@BlNwhm6ZjhUA+Cx|G11GQ&gV{t{ioy2oZ(XoF#k<-0eI1VmNDtUa9RYEjwoOLh*9
zkWpTX>Cb0D3C7^iKu5VXO-z{{ni<dLgIw<Sd;x;H-zIa?@G1}^cRMR-7;vg%PyLBo
zeQQ^zPUxePN&tw5eQSsZNw%EU)M&Om+t+Q{M&<lF4I73bMH?&!I<QIF3%Jc8VPRh4
zskL==$s;@&s479QtAYR3;dPmAxJ4-*^`_PnJFmDn70^#$E#0AL<h{RGn3yk9yT3(8
zw=Q(2+Cv(p{=koikbs1Jj}`4LGm#STB2W*x7CHbxv<Jj6$~9yMRw&$@X3^DgdtM4f
z@4k5X5_MfDk0lzx?EU@wcgA`ez%oK;P?nE`gruab>^5{<2&52cQk7LEoc)v4vSnSM
z<aA}NxA$OueZ6a#ZK&|Iqg-|M_O_KsD%qzJKfHTb%RYW&Pgp{>myNkk0xG$1eDg<6
z*-r@3J&7;iNt%z@nzM!Vhz96H=o*yAw75_;FGdv_=k+o(70|Xfb~WYq^t2dvGlGVA
z>=xVQD@xj=-^0g9uNi#2$i$9SrO`iw_pE)ZvU=+&R*RS8>hPEb%M0pEnV`R$JDU~*
z?5vSu<&OdkBzny2qWBmXf`nLJyzEIAy5oU<h12*x8hr6D03@3W)mjtd*6}ks^Ig3S
zO9**-CV!eMA03cIKpPI&x)bB{m~%@@nIKQlhyL=))h?BQugQ;_12`>XWAjtM|Gxd;
zoC#|rY81tyId}5g*M^HFfClM?<gR20Yw${#4aLR%p+pf2DE@)+T+AP)#rK8(Od*uW
zRw-W1A4WX{?EiTPggb#%!MpEt;5qG)?Fy+UJ3wa}0Z|p^%u7uz46IcLM0vRP?q$L*
zJ05P)Er5SZl*^P?vuT6ceML?O7DXn*^3a?nG$5cIZwfCZi6z&%GV0gFg|&^T`YpcQ
z?QPn=QY9s&8>{GUG(KOvJHfVXBZhW_8yiX{ZD62I_4@@F#$0oi`?6&w%>$*onXM<@
zIp<BXy{?a~p0siQ6$go31}!|c3;zDZ7|54+sKKl$bvaz15|VE1yOcX1na!S2yvX4G
zi>`WfXB(O(p@!~oJ~03T%Z@>{Bo6w)NS&OV820BNbc!LrR}E5I<)F0*vr!*ykzube
zz)ZqILY<()VzHQhgMb1=oRGSyvzD@_UH0xjEkJb-Fb&Mc15d?(X$2qdAoqjo!3+Tm
zs5l_>oxeRW<#N?|O3TZejaiol^0Yv~AfKnsAQ8i@1PoeOqjpzntV;&WYs1)>2Iwp}
zU5-lt33jgYK?9e#D?=_5Aixv~k<c`;NLI$XSVOQ_&~MyNgBvVX9j$i5Uho<5s${_|
zS8p3%cVm0{R5C9wFVI3}se-V~gRb{2YvdMRj#5EN;_!v>7g?;4lukSncV^pU{OcYi
zmZ$!X97`Er$KTw}CAV}43~XEO4|i@Z(ng@uDHT6I?O$URL-xN3AMxF@`e{`wz_Yn}
zBrGcQ4GUc^r(5cwHCJQ#3p}NKreGq{<azi#&C?6m4m)Cqk%ZFOs;Bar&z~n-tzAG(
z3<Vw`;S-?@*#bW`>xyIQ#@u&eb$?zxAtQSUMZo>_bKh{<uF^6vFwiOf;4jc?56#V8
z(}BbEr)c1IkBo#`XhCvH%3`=)j8vTuUJ)$kW(Vj@+L0K*OYK{i;EG(*ID`^}x!;kI
zsZaqf0v0je^BE{YL@sm+<VX+j9>?vD4h|XSQ-32Eb%ejIX<1n0LopL7p`i-<0_C5^
z;jWQhUS5k}g+Z>PZqo+*;?6>9?CjocmqrU8V3K?KvQ_YI-%bDlf;y@u3P^5p2EE7(
z@oiA6+6V<HJvp69poR0`%UD}97grfE8vP?KD#1PR?@9f$Eg@fGeCAIxR-M<POq~}6
zi%N|WB6dy!l*bM{9eAtW?1}rzcsY)Rg=xG-nhi=UB1e3$v~AP1lw%SYb!3`RKcjfP
zW9p!2E5V)O!eA_U^iZbvm{hB0(SKu+@t6s*)%FCj`DYI!t4lG{c1_BQDyH)!h)QIG
z8a4K4!jn+MHBdI8$o902=D~D+M89u&c{z#evAIxbf$r%l?DucvJjd&`Va_K3d?#Of
z4SOxyc@*>?IQOgB3I(DVz4?70wGBUL>V#0sk*9YQnfsENZIyvI1DXGR&v3J^h!`sK
z={z@%figROpkE2};ocg$_1~(r?TCARKuVyaIRTevKb?wy>BWf}pLZ=SO3K@FoCk6l
z$?#gTnJ2jowOnyET7D$zh8E5GJEYZWct)@ES4N#3=2xEN4;lC0*ZZ}|w=+UjFk`wY
zAK>bonR9xq>JkEhR2I|(CCWa`JZ&D9O+{Av_y_d6ZR9Q;d*C{*o{dAzuZtHgH28G&
z)PI!A+x+!tHdL3ek4&9`$P8$UMCj6xVY4!%&_dZ7sd;p~)!wYlh_}zckc5D8W${}5
zxf<0q?KnitZ{Q$u09bO9?Ex2%&fcu7ps6TNs=hdv(0~~VCRb(t8LW;#a&Y`s%6e&+
z9XQ-;FO35I*t9e`D+nB$JmlG~@?W9t`~isSaa}A(bF7-_irG;^GCtR;vNN&A*U-(T
zc)H)dhJj(Gp0wrUZj{zut<X?yL5%#N8N{n@vs2wuM^3K)r)Q7C{W*d(GFRCIoxtFJ
zVxtOz1S!m-j4dfEnkC@6J@zR2Dryas@Mb9F(hCXmIu7J;hDOmQ$M8Lboa-y->ZfAG
z;B{nC%Vhw4CKlHEAcoJiihQATE~b1ksv=oBTi)p8udeH9Sn)xBFY5)SaM3F+yzK0X
zQc_Y5>%Xp|Y(2tx50N32=;`F_%mN<OyH=|<$KuoEdueLT2V13%v$7eo=^a{|taiQu
z0oA7et&0Bseg!D~8$?b)NtprqxkHd|(GRnL2e{(^-Zq0upyp_Wecfm<w%by{kTDV4
zk)=?V7>+aTX;a-Wb8B~QLK$D^QPkF<!q{ULvVsn+vE#*i$$GjOjdY1}l{3V3MD=gq
z%M=UFHgG+7wFNy~cgX~Ot$9(S5w{J6`CbNEIqU`4^m_9YDs#+Q?JN&=KQXIjyNgA{
zr2B>301k^l^2pXuzQp3V@Z9GFdp=OB_=7R)FlgA>%_lRo2vU@b452QsKO-Y!p;rA3
z$iNf?8Ot}2;Ut0GUIc6Zy37?yO9jgo>Zol^U&*$QH+~)%7)YP#%3~NBGDDaoea()-
z09yMUh%f;8Vg%tr*p&)ch<zyG;OS|s)227no=_voMteNPy37?<DOE3*F0aySsJ30K
zubLs9ZIjWf`(rXNuuvO$criezy{Rh-ueupAofwK#r|x7Y8NFa=e&idN6wY^|nm~kl
zb+mdY<Qn<W-5sxTPk|!(i4);Rqru`y9--1TN!!p03Q-{VX@xl7kGnyY+@2z%{)>81
zKMjJ)^HfJEjgvt&DFS)<T~h&kIiS1Jhhslqk=6lf^Z8o2z=bJ<fVy@5R&PAYy{A@X
zzqE(14c+MilK?|V?X^mgz7Fp@-SCamelv;0Zv4ZG0PyU-J|ZJydH3gYm*FLT!oAwR
zVu49z#H3%B_m>n9;-&2@&8(*M>$@RbHcx7e<?XhrHaBUG>F)MRry?aA-*yB<{%nx{
z#Nv8(%p^m2YXGil{$SP+^V*ZaEDwg)lHY^4!l)hA7+_99S3yH1J)f@+3o-$`-FXmK
zae`Bg%F-6<Dro0b8Q-uEq6FDsB1F_ny(vl6y5gj~s)fnVOe=Tfg}Q(<NQL%axC>Dg
z6l$oC!C|Ft?g*zzidZD&6Wc9TJ=AMr+9i+dNT76xIyyR#(jp91^?Db3Gp&K9PfkrW
z0?A^AOsbt(J-TGo@|PCpdxDC!t2VujT&@b*zsxYrd)U=a9%5t+$v(S&oii-hSb+{T
zb8b`I2h%}@9dRGpmQTf0s))((w4}u+7ZA!@2CIMK;o15Znm^CdbwAZ<!bPpdz!MC?
zQ+jX-z^e%)FESpK#>A(U6mHUpbO&M5A~gAqU^WQsBiPL8&kTR6`4Qxunc$VsP*Dj=
zuR2u2O6Y0$MwUCI^1QO@<Zvl43GAZ*JiD-jL|AY+?>d~7+kfwwss8-*63&H9s;|HW
zsRLyc+G`lns@n}_z9&yhP)AQMQzDiZp8NT?!^eIRZPPA1=+DAMQS)Kgn_<o4d1?H`
zyE}I%gm|C#E@x<M<~+cW-~;Q1Z+A3UVSaJto?xKU$KN4$D;4stU&p{MJMmq;gc^^W
z>8Ytikm6=QwWc9h!aAz#nc_GtU%@Mxpc*vhta8rlb72)73{Q7BIhbS7?}%IKw|ck<
zz<CshOyD+XT$+Z5%R8o^u6t%l347(yYB>?3_M7<nY{k5UflA<P-!N3l?n`??ZD|@S
z8}sMtP^&1K+g>`><uDx`hu&^$t74qJ9pn-TWP<)E>uETcKZ94#25KJH*2x4&XKtJ9
zAgq9DtY9fZ;2eclmR|tC#Ge)??tYU-@Gsq>+^?;anXQkCUxeB>s2q>zuCYs>s5p;|
z2=9p9Je%})hEOL7s->r|ufHcA-Y&xT%gj{c6PL7Dj&JsFZWHa?F!9~q3XVCPs&)^F
z4$lZvcM!TmDf(V|E^KRStDN@O2{a(N4i&%Nb^y>~wZE?EN5~`qT@OGx5L{hdJ@nHg
zb1ueFJd({!9wZ39h-T27!dhX(hJ_FNKeSU{_U+p^hn>Y_4$E1P@;Gnq?9f^8K=FKM
zFw;Y2)|s9-q&e_(gnk0OX;LqtUVaDoUbsX=4uc%(>guRy1<G`00sPPmCLeyXI9geb
zT5CX@Hnz9tD(+=68*=)-)Lma4DV2fA?1hz+Q<4q3vcuU(vmTHv3SwP}i%OS%rs$a9
zPeV5mTygwTSe1%~?m51tXU3;b1a^*!?^kmY46g5<PCUsKP!~fD9sgiwnYED#qD)Tf
z@0U&m4d>*QvC1U)CdXMrv%mW32^G_8LV~tg2{hE60wgMdYqg}dF<N6492>e_4v^dT
zgC1*(O`paGCRlaH20T;*ln$y>_hmH;%Ku$x&@BK4fc|VSu<x_~P`o#;mgyDPQ_;6$
zfnJ^(3k&H5kn2}G%LA=I#AQ9-@f=Fc=Gi_*n~^N{H5!cKST9$z-fS}){&6B#@118E
zRrse{*q<jY;^D&#+?#LD_dWl@HEitOC=Sa)y>=3)W+6_eT==y--+E$MQ%Ci)oWA~N
zsL+`Jl#l>;A2gN|^pf{&u2c{b5{6NXfc5~y!&wkgWP?mw9pasXVr)XjSD*vl#kv}I
zzbsMas!&(wjz+M4SU0y|pEIS<Vt?MaEvA)Gr|8M(C})9dY1eIuFuSOR9m~!7XSn(I
zFQdjb&R8ExSB^R?(LrMc5d#AQVR7-$L1TGQ(JxRvI29(a34ts#mbwy5RDN+3HYqh7
zome0)2}uqVvY!XpW*;;LU;zo=YeB(G7v#vGm2N@lE1BjC!Dc3l{WA4`{erD@tNpW%
zDqgX?yOATiJ*pu~M*a%PWlg(&R{{rzr*Dl_$)aT_Z+hmoUr@M(xT8xrEt~uE)5Wpe
zS1hO}L~GN0dKdTRP5wAvi~1ZlTd-RPq^{7tAVYFuQ|&quiDXa_sg<*dx%3`+7YkjC
z=$UF@;X-7mFndv^;#?|~fYA=so=D*6-Q75D9&QXCo8_;+FgEx4zgfo;aEbm@&flp&
zUH!&ErWCNfk|@xNslG&TcH41>e425@PZO1s`e*=J*1v<%oN=vyaNW`KlBBq`L7!|c
zw^g-8%fjIfLfwa_BIV5-g}_>`gqf|JfJGL`Dnev$XW6B9${mT^*(u4hV7gnl>U)cG
zzaFbmw|tge$>6&3#)*j!crA2`J$=FT0=VgXXLok%(p9K&Jlxh;Qas$>$F=*CnwmN?
zV&TPMyBft)d*N}tdWBgU$=aAic>2bnj<7`b>E>y$!^x`;^|~p442BcTm09B}r6&`a
z$5Z?xo;k>V>-$q#B=$*{=jh7_0&D20ZE&c>k=JAbE5Y4u<rYHDLfw?mFJ>uBN3Jqc
zL@b5VMH>Km<R6_}JbN1Uk7~aD1N7Hl<=|(5CP*n8-v6RxTPY#d;oF|mbo&)$qN7G*
zVr69BjqDReaPuEOuJEWhrfH6fNlRk7ife0}Xv~@kL%-ei5@+7y$g?5)?!6eEfy~s!
zPaX>swxhip&dqq0W~Il}eX?&5k<)4qMMIadlK+tQ57uUeXtjMM$6go}J9S>Xeg${4
zp7!ja(DdHQ8KCJjB^D!JUg8={N{9IBWRcH$a{5LzU+Gl7)(iKGEI+oyh`6Rrd#|Od
zCYW=w86_YKc>AdMkO>T-rztCf{_4-P2Kt*tY^|z<v{p++R_?LTZaAO6^kcfq6=+RS
z!gv?A`^ajUCt`7y-dh!l|A~=NwaEX7M9NiKP_U%S6(};<U}{;|_NSd7nKNAZqB&yY
zoa+OYL{%#)czEP^m|xx6S!`<7&Og0z=GUMF-}bsv(ayakdZJwJ8WtU`mNZ4afUq62
zmK8x$jr#RgvUdGF3l<K;PF{vQmV7cA-;DO$n<h2B5vU1ySaSaAyYCfixZr5=x-VFR
zdHy4`*E0LE%K!e%AFkVvH3|oz;LiC?(a;`K!sf-mqh`;*`R_x|@?M=CzVP>w#~0QA
z{rJ5Uy!ZLnc)~Tif4^Vz`qujSxS+ZDJ{rIB?-K;C|L6YnO^88kv>LxZx&=?@Ih69Q
zd*>@>V`SdV`dY&eE2Yg8`|Nj09>ASSZtTW#r=KF%gIorF2&GO=(wEcGK7LbdYn)qe
zx|Z3uv&519^6cZ|KJW-}xXIeu^L(dAuVPQjUdTs+^-(PHhDfU6?bF9}8-?!@_YC{m
zt5M&JF_Gnlhj#)p-?>!>ah=6WXgO|f1ZcvI-j;Jcr!)P9Yx~#3oi6rPtXXE*cGqsE
zOU~p^G0IF@8E>%-A*U|WDI!yU-seq7&aF=P%aXyBZh8J7)C1bsXu{7elp{<XH7j<B
z`QoYl9OiS1{nK8L&E;ORrD$xf&W{*jRJIjkPaw)he@sm&@2qh*x4>ISeEtj{L@T)s
z@Ba8ofLLVC%T)19<!;_ZWJ~VmW~NoEC3nT89?PD0y#}ti;`Mb4C*DCi<t16?^TYof
z2BGE`>oQ-Qhf!Y>Ra!(DAFG6^K)|C|qo*b5%|DbfZuY&&v{~r`7drS|>e&O=^umm}
zxr-gxd?SC7aGJ?@a2fx|38G4qwTQ<;4s*jEH&)K4B$xqae4L+9NtbS{Ol_5tw0C8n
z-+M{ctXE?c?{}f**o8U@Dy|>r6CxRN6E1sch0{iqP`BVWkxf`<ZRQ_qEOa(GqahWT
zo5&mYE#7*aJwhC}xvkBmoF-#d8Xg|Rph@s|6`P+0Dvz0rn-G7R0)xmRdl#oo$3^9v
zH8f7|`@ZhG%}pX6%|j)$nX6jA>T*oo8%iGS?<>PPR?r{4Gl_1idUP*xC(!RkHhyM)
z-PD9Y=$<8%9#TOkj2gj?TIdp!Jm9z>l&?(Nq-!~1LAOMdd&=NU|1`1{0-I!n^tYMm
z=^yq?l5MxvLnQ%0g>12Cgn+mG^fY_92V?!aqNW!$z9<}3c#}qHsBZ3xqwd<Q`QCac
zd9tqd<H$9o)QKUofL`68korsHGt<G*1S{`(UBvlW@8wdMq7C-jAzcmJ3|S*}o(4zr
zT~?;Wzo;G;MK(@%7M)+c%2n=nyTHzo3+`1*8Ohkiq|F#UZOY79PR#y|V>l~nd4#|q
zhBD|2g&nsn4)!;49rF4Cgv4PT5dh6IwR_9fVId(DLYshxTTs2H^sNg#ReW#D)-<hY
zVIlZF1%_`ni`L%J@tOM`df2z{aMI9qCZYvG!p-&duU=dXxy?*dEb6Clj?7gz_k*h@
zio5}d^iSB><mWFpF;&%GsQEx7j}t~7%~htJzm+>Z=bsZs4%G+loG<@1)Drygn?_DL
z)4@9AHfi$y9TgE(DvH(blL2N@J53=y6*hEUJLfyDi2xS*`yZN>jzWPy83;8diMXm<
zUk7f#UsKmHMADyKQSv8hA-VH74D-n{G0S#3x_17OJEhw_kW!hYv6xc8xN)t6$rhW|
z<J@lY+dl_Wb7OX4BA(&ziOQ-#BC88B$v@{bp<ym&B=40G-#vOjk_S)imGMA6ohNht
z*o$Z8!wODnrdUEc_tp$PJ)G{Lc263*4XgNiJvc0<*b;Y+1TVoU)4@eXDpl>jlB-pJ
z_RBSRrCj9~&>rWKl<nZlUBA)ie1~eOp{(87xVt~2t0XLgb%A1Rd@&G9Si(ns894vZ
zqfGDxHa9lA^xAn7%gZ1g>gfDRSKwXwpOnE7e8ZVlW{p-c4Q2eq*BX&~wxg6m0=sLe
z1Z(#)I^y`LtT4`gMfKJN2DIS(x2~>YPTD2v&hr<t2P+kZ{qM`Q9Bcu6mH{3#-BOQy
zyL|Taih+eq=J}24;6^pplQ2pu!>{K1SZ}qNp4;l+JI49wXP-esqxS)Wc>bk|3%5D{
z`oD1o{O^VTJC6T%jQ@Yn;D1-e|GTRKW#R;(w^<it#DKtom6-iKzqhw{hGhK52sX1E
zNKQ-xs-?~J);}w8KV{@{1p&Qk2P70&)y+*!P8M3s2xqI7N<(YuexPTxnTeu19{&3@
zyL{<&J8X1!;DoYBcpTCo{}TfXD>yPT5{Sm1rihFuGAa6i9K)?*0MK*MEB<e!Y-rew
z)(?-4mH@q$fnZoJ@7-%aB?z(5?ScJ)Y$b<<<S=p}K*7G&y)(Bc{&&N-)fRWw)kPq$
zhg=94TWByXzWSfpTehveTp^#ya;haEAwgyH&z(i{mon`euMC9_Cc=Rr7oJ&JaqiV>
z)h&>gk-@Bh|IeK6b#)B%4rp82VCs1G9`^SKFIi|6sBO*^d#?rEQ#;w>32?tRl&s+#
z=~rLtN55wO%dn_%#WS4pgVBWjg+Ow3kzmJu2c%i~bloeZW=#4OZ2kkQmzOJa3zTG1
zM8A;(%W@b|52S>8JMYo5T!4^>h|G%@A736?qPK5j=@m=PYDcy<v2vAT2IIjQuUG8G
z_{e=uL{+=7QA$AEzhIc3@M}o;!ZN<3=ep&^z~JcUY{&}MEB+uIEcd1V)I;p9WoT9w
z?e3$CO{Ermdo_kdi`m7a((m3Std>mOU@`1z7^M5>=DWQNULvf3frAI@gvNKhrIvbr
z&`sdYoLrjZb6_gzx8bM7&HBtuAOMicR<|{+$X#cY&y+)FV<U`0e8z0ZQ!L!5d6+#|
zuAn#BzVt%C^TowWa>c^m4+{;lIpuywA>Mf_nfyGqCK8_7_!<}!KW4Msvzz$MM@%%h
z*HBLJ{a59loa|APN4V2-bK|!(ahB@pKSqAg^Mrv`p~Xfg!C_;MB#-QET9YbSwuQfr
z7ya>@P|iK2!xuSO!-DJ<Q64tPVr#5fd&*xbZ+b~RvNa`LCe;wY4a2djN|0mAOj8vC
zirpH;QP5EaKMsOYwbZg{;|-y?G`?YBXAWxq6EdEcKo4245k(`!!$S;Mm#X=WlNz_G
z<y0*!KGc6X(Ed5ZW|(%pmQ*@Go-WtYYnMqcG9)PaXxAvn%j?Sx2kuw(vHg*j)tz(A
z3Xi!`$BP9B5hKh=j%<~O9hc~d1H-742*MWLYPH>qtb5n!N?THXPiONfbVp}L5xSQ0
zq1=_+IuDvk<8`!kXLAh}>ehPMDz?91p$O>O%8{xHPx5~b9XQu&fdHV&q_}u7hl}b^
zzegXpo7tq2R4*xa*fB)W_fvoUrW&z+)J}AF5^XQu<Fs2W#0qri$;zp+HcD5QjJqgG
z!WkJ8Pn<VbIQ>3}m6SV4CDu`@_GlynC)P<dhMv0{aSK0JEp=k8BDJtZ^dM5SEK^8r
zCrwGM_tLS>S1U>uu}q1WJ{&>+g{6?p!uPGp&dd(Mo_nRegW2aX41wh8#Zi%31Ppv7
zG+4V4>2Y=Bv(0_Ko@^}*NonPXvVN@o;21<tmPH+ne|e@RPvvNKP!*5eD=meS^mZ4u
z_=fLyk<1g;_Ti>2gl0^k^xg9TCsaZX`6e`eOFiPhMitsNOREjY)Jp2I_E=5&r~OoR
z=6m+6IeFDVd$6`+N_b`=O1@jLWcTV8pdxM*jNc0>Iew&kG4Ba}e9_7C_)iN^JNSnj
z+i3=yA~|`X@n`n-6pt<b-PKuhrn}5Z=9jWv^*<|rlI@B>IPT=`@wK>OPWfpc^sP8J
z{H@<)CSq|NH*rYS>%Ulw;J!c^q`ojD)x1jhz%@uHw>3QTAhS=k-v4aBxzSLwT`+@L
zf|PU?vr$zbW~nFW-=^yEoyew6O|ihO#8wC8{yxglQfnQP5J4Y{f6Axfd@wAGtk5xf
zVypZ!Nl(Y%PmclkgNJK;4u<^5ve&%Df~|{(lm5JnZyb)@aYDWZN;~+S+qy6X(@=fL
z?dpqQDJ?*N_$m|f;Rzg;uaY{&!B}YI_MwFpvMrWW9cYD6J2)CS#FEsF3a4bm#-{9)
zDZLCkP9o5Ajjvi68hy%De6r_7YCY@p@vuKP@hF0odiS8XH#=FAE@H6&akSm`(IaBN
zVO@Iu_SvupU!BoKUrRP-jikjFi9a>{Q|x=GZl`z6^-W-m0cpJ3?VNb7xaVp`*V2=C
zr0R=~6obizX}V{#_SFKnEy8Y4<ZQf~+hu%nxLC}mJ;QN+M&Krlw2Wip;PiJTi$o*P
zB~6bD4SdNBBx@Zgax}j(|Ip-F;U&n`iDb>xL2%w!%vt6%UyVY=?T5cU7^;^_doO)D
zI!br$eQLsaCsXZPJ+4&ld^7ji<GS4wERFhwS;GtQtXE<p-ttN^=LfvyjNW*|@8k;&
z3rzm{*)3caoXUYcIl|u&MVmQ@3160T8+zjJMo@!yTShuWEAMejfBmX*Jb}?4O3ER%
zX1pjd(LnO6YFyBMM)bu;MA6yJ_zzyyV2)$(#>50B6)xW|H%bmKpb)B3kFn{uxWcP$
z-%mfhs%$W*D-;-8T>O9j&S(tfm4kyPu=@x(c0DTU1L@xcX2@sFf9`9lFeF%;%S+NO
zH+nszQU12v=zPU(db<(ue<))$qb#UP=%tbvmYeI>s;3Y_C^K8~E9hu6mswBGtt9hF
zMCTdd=ho?VZ>9w?mX)T14C==68H<4yOa*WO1`1gWOgc3glbIySOE*xqV3&>|Az4JE
z6eTLtbOYAcQ>@0NGdr!{7ITRrr(k~kR&;xVWdCd{%C6V^7t_6{!Hna7S!OHa_qFM8
zX26^-fKw`gh#`A0?n~5t<dXle+{C1(Cfmonm*{}sDpxCQED+Z5o_-w6QaBj_6vJg6
z@qeHg7}GjXAo{bba3IOagaWb98|uD;Q$&@D#tKhd6CU;aLHu5@fUl1HtI9Hm`92lg
zu$cIVA}-Rp{>cq<x+>X+bNMqq!oqzfE5qpuOP{5eanFqgCxo*eWplGRL_9yLZ!;BG
zZH<`i)O*YHb5qqD_R<)!&WDf5LlK%Zk0}k=WQ+^_@x%E#hj&h;VK@s{w+Ru4t&d)(
zCaG^l?QdftHdgkOZJ333GUPkwprsTs=Xr#d^92@;oS4h@Ex*MviR*;ni(IbWp7aMo
zMiQ@S>J2NGOOmo^TM@RBw+iMKSNv3F)>g7#=T;=OwMMX~Th0Ms+ZAv}UFJvtyNF^o
zE5dm*SzWzA-RsWnTy|FKCrW17V+oY8Eu9&6QhN*K#gN<^VpooixZ?l8Tf!sJ2sF3+
zR)l`fQR0cj3K2_3i3DGFVC^^5)Z%fAWyr;UJmwMe5s^^PerMq_KU`$b)Owdk*w0AT
z)?%*>4<9!vWKFbZzrDK0YwR@6mesVI>t1m>7?0nVj<wH@h*bKdH-TSFO^NIsH7b$w
zeDDS<BeUlVQ{=(;6-jcOkuDh&nA;w9nzGh!))DKtYuH~}ay(3fzQ6Qycg0Ghon*1M
zECdVP{Z2U@SA{607|FqB0ffC5FLHr^4)H_i$Un9tyr(+USYI+b;?UYSSI~4R(Q{%#
zd>Q|IE{R|+U(YVhBugQW3}jDT@)fgvM8x|%3avoNM_p6zCAmj!==o{hPZMVWzMs%=
zsQ`7GuWXWW`#!CiL{P!-`GJNC@c6y&B1UlG&40u_&?}y_fMqXB!`w4ya&AQ|P{3o(
zC2&B91<7`5J;jC`ra!;CeOsqUS&o8TF?+EO1vm7@o0FA3!#u<w=<41%nOrNHZ!hxU
zbyZ&oR|4+?Z+t1y2AgQU@dL6qQ3q46Sc$TZN&~oZ)}`VVF6+q=4vOR?YW5n9q8BfF
z%T0`SOvdD&ySTKOp5Zocufp3t)>+zJ?YE3=>-&4@YHo?iUfyG5nRVo7%xN}A@)EsI
z@28BKsBpzM=hsICCDxA~@oVI&QZ(0LA8%^w=+V$I$%wES_jF<31ISya)nV6Q$GEpk
z2oXx>4gEg1TQ55ABXOFV+NvgZiyZeBGRG8}08(!su}(d)+*31NK7)kL!eL==ppqjs
z+PjS`XN3ca<gogj#qMRmag}jvoWVV5zK1!R!(Mr8nJ+2&TfNn=@6^tA92bkhpZBFu
zV7l^>xYgt`hqc|N0Sv%b@|_$t8C#Z&R2#Y9xU)-%h0DgTDbzZ0rRsNPV%$0L2vT|5
ztCL*ESR~E(4kf5;*HlsJ&GAgL`RP(9(^zh#=u5e%kn0~Iiy<EumU^3!v50N8B9w$9
z#SG_uU!>*S!f>zA4#J2b^SK?1Uhx@d{+JRqniLYP`-<Rz1O<4Nr;J(@kQk3D{;j6_
zb#rnl@pDHL5rX~&CJsqwQ1fX=VVwwW(AT)!7*flZrm6(+#}u*?iE!NlqRUqof`%Q^
zU-tfJ_P@TnE4_o}YMm)1HpN+l@?2Q#r{pWk-{;+ykzv^i?-XAgE&WN#kwMbVn+x6-
zVlXZzWp^<p!&T+u!+~azosSo-_1w#d#**D*R8PF*Uq#E#rJf}6rHV-<E>4IqlAu<j
z<ZaV7Q#7=uL<yI}*YNZ*ug%P|%T7(PjXI}&-tS4H3)!+|<QCzKCcby&Yt<+c2%2?v
zL=w@HWyxm_1vn_fLvwm5^hR7=t9mqo<l+nRQeW#6()j<F`s%Q#y0vdCP!Ld3TBW-|
zKtM&hySqCE7#amZY3VNM2I*G1ySs-TdT74IbH4MO_qzN8FPJ?ud)@0^zuIeIZqjG`
zO&j!5FP;{>b4%uB;v1l(G{4WefwD2=s~YEqE;tL<TtO@89-DP9-4YTEYk{&MJnz|&
zkddsHA^B%A5q|{}g_uF$!lz>(i4t7!0(zz#^iu+W830c4=Er3h_V?XZ3rd`>cc&G0
zD|TIt2ed$}AXmi?$&>HE`H7nw-o@NU^qsA}6m8?klQC~oI+*RwSq!vfAiiMq@USk&
z{!!yZQDh)5V%Cghvr&h;KsGsn7t2!wXmM)!|FL|CgYG|t)pkoeh^`I^YH}&ZZnx`>
zRv3#J^N%xq+DIanj^iuA;blk!_hdoyB$>jzhjAWfcTahjFs#}peM0|DPgl<fXqynn
zAmfR{a#PM7m@e*;`G|XC9WKvYzaht~6MZ--8$pPh_aWlKlI))^-y;C3%I-Dj5&#;C
zMRYh@X|C|fR)_uUHXB>1+=)((T+Kb}sUYLFmCv3o%#NnpK69&-RBG+l>uTOdQz@}G
zayOm<e`s8$!THOWUHRmUNj_xz!EIfMxsY*OzebgLxi8;8PHz-Yq>=2>d~;-Wz<y(a
zPm2n{pHUWLZ*Jw;j2F=IFiIz4k~-^sH=f*0jg^c5$WC#WAUC$I(iz>gjtGFg7W8td
z))LJs@t#KKvsZG>#S~kyjGBA<&+aSC&hsB<fF!J6o4s0C%(&iiy=b1uiECnCqxiLL
z<MKwkT6xXuCQ{OT{8u#K+Zvi05D#m_$0BOV5i{$88wR$8m8tiO!Jp`R=i*VjZWoXw
zB4iII!*1#&1Zr(7!zAliCxoPBBqJ%5>6Wn`#rD)Yi~bs%0?4XOx8^l7G{b}c{ce6F
zbbg>5`5##mY}07Ujr7O<z9_H4Giq(6tJRc1>+QFgRM;z(`yPrN8&z0Ke>b!|niCxt
zk$=z=JzlK;2z`4-ecz@S0VM~S)amBpp~1xJO(m@c4=Ji~w37TB6m+HInaV(A`mRCQ
zh^Q!Y5Netl<JfyglrvSqp>ECPcX637ESB{!79!Tu*Hg3wS1lyfie7Bj<EEwW-PdrP
zEcSn7qEPSiL030fOj6M#R9I)f18cIo-sUTJmm$Xe-RybvQ0Gtkv=tZQB_uhUQRkO5
z3TC%SxwgGjWCAD0(OJ{|A{sv@d24;wezbV9Vye4o_32kJu`obuw7+z;`(5|dIiHrO
ztk@UF8nTvcU$yk|I#X=~UX|P2J@72MhJv2zwiJ_cdA>z$#G%9Yu0V$e!MKkyNWJDg
zU{P*<d)t~&<{4YnhI2iEz<U$T@mm?fKldx(J(K_1_W^1hmGXuHmAr_v)OUHlFzj6o
z$qB168a7(-h_O<)nv5Cx(fsd96TotFq<Qm^aqns6F=;h-&;R*qe#!5kG6)$LO;B<X
z6n+;M`KJF_2cX=(1DV-dgNnm5-Yc0YDTieO=lkycB-|={gj_Te?_kVYayi3y6J>F8
zm3p51?;RzJw)63EXuM@32ePs(-FJDc1+WW>zK7**{kR(yyc^ae6THc6g_mbM_lEI%
zp3cHA{@Q@vvMt9crSH|$)7L$Gn2C0I(?&~aWVC&}PJeRRBY4tTZ(AUg;jdaWTj<QF
zTrBsI=$B#C&+xo&eNz0TC-{5sLEt!l1D-?5Sx_MICG|PmrlUmuL{i|yKIG8N;DYQ@
zZ%4*+{Ep1x1##x0Y*JKgcczRqVeEc84Y|KFJR5hmYA2^_m2kGqI$)=e-%})S!%8{-
zh2A6?hnD!@j~I=5XsTD>dmGh1SF6J`;v<YNi-^bn3H_x-&A%$~`08M`vrJyMnx%>x
zn^Xvf=L9_Wo~K$w2qFf>@0`I+2+=M7*i0grw}0m<qZ@n^h78?N$h5~u;mtbBNKtW9
ziTUK=dEx-WaLaWybj4#y<m9DPj^wf$F4Vr_09~t+8YJBo6W>;|Raptm%FCL=hQ-83
zmPAfDn<P;<pJLRlybJ2jZvP&-9nDGP$Uv*Eo#LgEdV6KgKX(=L$&v_29i_d}2^Xt_
zg!M<VjC_Py0K8TfVdZnzIVm6##!nF?Kcz=D*vs6^Pu*@@=``kHUIp%n918h7ng9S&
zkjcuTEL{EGxz*u6ovPTHc}0>DGR(TIRwNOy-B9gXPmyO^AMw4dc%zsp`SdbYcW0Hv
zoukgW0IR4y@n?J<^BM8a#J_2`$-FV2V(+KZml3%t1$-o72RbI5M;*=CHETv8X5&(p
zWDTN?mXi!DbJ3ZgeP1r8JaDTQ#W#knZKhY}n5};Gog}8_WLM-4xeXgB|2>wcfoo0V
zl}c(oUiimOwI?~5w-JhYV~0ReJ~$cM@cqtd)Iw3eUyy<B<>}*08WIIsxM?+FAeB+`
z1z6SUiN+^4^bz+o{r9vLA*)lu*=LgG?N^51pU2b>=hj|f($dmPX3SQBBf(J$L?0{{
zQxH%%<q~GEBQQqjwfX4y`9#a1@JNbA$2+;J{Zcf0C}h-=zg{vT<-)p>4Z_e~d^o=>
zR`i3H7BfSBG}U9S3_V6)+H!m>jVT#7F(I3K(wgb)f=isaP$OrpkOBQp6i#HaXK;Um
zXfT0OhHqFt&2gQa$D0_J37EW@8(6Gx^{v%>GjnEy%^U`c*<fpCn~9N8;mQXhdajqv
zC;3d7IR#*WWUM#+LU-kYpVz5rp*%TKf7RQ%nGK&Po<!_$<7_pIR5`;1Je!%SW106i
zGu#t(JZt^ldr@I}dYbB<<|!%y#Zfzz05ZO6>Hm#}qSOZn1u0gu5|QrCj3XNmWO0Ct
zM!^l}F)7dUDJe-9E0oV9?F}Nx5bYkEmE;@fFkc&Nrk%B5JgU_+7}yRc1FKg`qiN}N
zLxV&?KigK3s6XwWPEok^LNW8%F7re!nbmaqAQY~;j*5HI<BN!u@%^Vr6oAKA6Q^5k
zGn%hmuPpdZ4!$_k_=g}}l|5fta$cvTN~W`OCnHM`$89bpcl8ni7jUhq-Ad%U!n<Nz
zB+2*Rsw_^X5D)V{kY}xDAo0~reVv%u`8L5)&$q6fcM<@PV=qsD7qZ5JpEq9Zt$Yg8
z29bG?7^$BgZ`)4}nui@H&uncFE(2&ZTBsX?f90XNVa&D_Mr+%<xsm+@1*2fm2acN@
z;WiEd&wn({xqoPE(#bs8C7z%AT?^t(OHSK0ySb{BflY+mA~wjrVE<ZZV})BIn}b8K
zn+%v*vPEiHuv;~HOlU?E$(;{Km^A$#2xiC19&-y=b(Uo3r&x&^Lf=eo+ommpB4bdk
zPe>8hg@X>U6kygkJ*1Dyo$S=CR>Bq3O_r9T8;N+4wPj@xu@cGL)jO}vyfD`EnbW7?
z{W14l0*Cj<^v?cs{dkJ4N*j_O4hRB%d&^lj{CTXPV3A4|bVeZ{9jZjeLr&m}6K3;U
z>^;BNm}d1NfvBjQy0?(Sq{}pvA$XI`5G3q5Q#l+cj-BL=lC%lfV8z<$;Su6&B1(F8
zY&VkmbU%BK*HJ(|CTMX7FHzckb-;4PpV5Nj7|pZ6FpLf`$k+hee97``nWld@$(I#_
zNX|cg_Zr&zfc6o)>t^VslY~}|eY;*I2HNo16l4b5Ned*inpWaYuQ0?S_AO^}+g4p_
zi8yQIJ*5}x6i4xxha2!yne7?fL*jWPBgX2T(f!UZqiO~kr3tq$1SjZdDV6bpggcu3
zcAwGGYE(m;=`LJ$CT2J#-^<j~rRS3o=iX4uY^zBXZYw>bw`Wuxt8`Tq9=MQtr0c1z
zi9tFMyjRa*c*P`M`_bLaS;lLMUL5E|Zk|)VZf--#1K{kFnj%j=>r1<Yl)Dplz<=eL
zi*<7FRZAj=_2ix~%TgMfbh~wvIr}@pv#Uu7vOySe!1AAUM9$!G!xuCj$L$KXeiOi@
z%0HJwna3MNYN|R@Mnf~gRqw45W!kMRa9?tz<2=)45roSLO-?z^WI?|RLL0eAyU+{k
zC!bdr$mEuU0TwlJd;Zw^o!vnoL701nOqW6B_Qv+O5rffa{jY5uaxGD;hzfSLBY!j(
zN?HyYxt+z-d_amhmv=5S%H6|27olZkmjs8R0TC%56tMr<W-+S}a;#aEqwDzVh6KFk
z%ZPim7~tFrk6DalUZ~77+C0=}&z1U1<xb5<HK0y=h1r!7x8I-sGn0=2gk)L+R(V9G
z#jTg~WKY4%$e%9l*2DOndD}2*Za$%&274^UlBQ4#b4dy5$h8{FYC(aIAiPrdb}2+$
z%x?2tEoaF!;k?+}?eX!JL*{q&=&?W{EWZCQQ7fr~GyQ)$MvEi$5--qS{fb>Rar}ey
z3#2Ja!l_F>eY_}2{M*00It8Sg%920rSLU7TgV0JsZ*b@Ui;RvmpM_eH%2kQBe5XTO
zJEISAf%3$TV`@+0&MD_VVy%{7-_jF_VW2}Zsx^v$;QYibHnkHMt2`AZCdeWvfLP`D
zT{Z@xogy`OsOzjqjq)87dbNpjW!kzi-c{gw@$Ms=7;j_*NF<v~g6Hb(eXaw!Gn@`*
zoD!;B0-Q<tv!9o%b0sp~SziDJXYxmr=ZRYLY@inN(4^`!d<Dmuuam_>tNyBUcz8Gz
z;=G=3+VxkJ@?6#@b6I!*`$^!FJytD|2T!K_hBz2Su9>iO4bEzkByx<PA<*Sxv(Id^
zD{s2wJ?l4Q85gUBs0ULN4J`5>-gqd3=x23R=+Kgl_836g^_|l$OKDVn$vLN2m^75+
z+A0Oo$l~Cj(|laQ{J-;co9ih1j7Ny!5TntrzV{Cp6-_mPtF_;g6(Dn5NOH{07X@jI
z7wqhE{UanfFO9Rvp@N#0PN@E%i0_qiQpf<Fjbfp{aG{?yP*&Pi7j91(=k_jCNT$<w
z+evNZsL_nIoUm5o*V3VioJb9o{|Ajz>VAig2Pj5lPCWU>XJcb?gg6}bJ8`rQ;xBM^
zF1)+!Y`51KXPn8l1N(IhdD=D>joKexq_Pv-)4`0jJ_bO%X6!i>z-f;|y{X=LVX0mU
zkLu?f`|FYuH~@52E8?}6kgYIR|IJ5E%f%^;GC<w+I!|$xo;%G_CnpAIWE!4y8hD9C
zZ94}uCzF*yGzLhBg<o)v+hgWveJ+^yz@kwE&>vPt*&W~TYxh>$Odb`@jsvXfiP-MV
zC&AO$T?kDH7@mqRuW)SbxE_oT76BNWj+I=oK1|pgC_wk}!ey0h0aPSZE(6fqt8*o-
zB;=+ee`jIR-qikG6Be4WNztuo9tqY&CkkE-?B@6q1Rmo=FJdtV=D(o5Yu}S`LoI93
z*I9nDGhuy^dh9TGLkJ+W89<7A?+Np&VcrFNe9)0KQL@@MlVHaT0K$%dAZAy}4#Rew
zw6wM8O>aH2gN%0l*r>~q_jJlt5hlgw{wguRwo|=I&FMa=bK`X#<qsaZqdr?mrwt4F
zd8N!Hdwh7JR5DRiVYO=F=}n;^L>O!3yw-GnJY&x#10*oidA&Tza;i)7V-?R@gkHd-
z1t26d^13KRoSdY6oqT`m3vZe%C-bH&=+C}0g)_Bg)Kwve-kGlKr@Rn-^jORe3S}PB
zOyAFv7_O7we#od`7QV8N@iFql!^bQ8#gXBNR*HrXRFOtlnZ|c1Ft06j&Cqp$!f8`$
zQFcY0eVi^cZ?XiS9pe;yY4iEy+}?KB>St$gj23r2Ox?g{^ITVuPt$Tyx6doE@_VS=
zYk}~S6|AKGen;M%%5Zt4DYdwaoMoNqxWv%{Yb4QD=5lVTl$Tx$oCe#gS{A&lu1_is
zkJ@_0-LX{il-Mk8-<~C_QXl(hCAky}Y5JeLEbCc3{PHR5EdzNFqx|r-#}?a5`@;&a
zZS~uD!PuADK0Ev+p7=~`H<5$1G_>vGb4jtjGweN!D7r*Dng|-E<}b;^_8#7Eg85-g
zb}JH(y_=5lGU}Z25bsc!{a8(I-_p_1vkp(fYD4#4Ew+S(4un=mZ15Nx?*>UP#SYOY
zNlKgU2PzXkMK`|7A$#NbjazHhZToqc_aP_c`nrpT`jJA;S@6kmNAJw`1b^T_zv)%K
znpe%MjtNE6cZAOIPGk+ee?mz}PfnhE{Lh*9FTNqWIo>^Cc#8O-W6%0%!7cV!p8Kh@
z{YKAjya#tKGe7^F4JGtJVes?#p_VZ1$j;x^D`r+9As70y7%7^Up2?hUy2SjbU)rgK
zXkzIJ-;3h1AU4gNbPYuo^9kIm2UEPT0zS8pEr-wb_1hl2LlV?_O5K?G7++J9w|Xok
z_|^$#vv6%pAQ2r1<E>hqVJ~+nu%y;dqJ(~IM4cJ!iNNQrqTbExR2wRw&t01BpLIE&
zG}gF+{x0hno(V*)^}&=)9^LYA=<T^U+I@nw(I2N{ITvIo^00KNjeyQ67q>tuFV0=c
zGfan&J%8Z*4ZPoQ@gnHe)>fIv`3IT4q`u%UD%BD7NN!lql9Na-a%G{#f>-ZaJsWeQ
zdI_raSgFa$cQ{Lg()EK=s(<)IDCofF!n;gOj@)3O=gQzidWLxv7P(N;{G^v1T-z)D
zh!Y#l?lmmBD*$cMSxsuT_k9V0DaxyNzlNJ=d7s$}?9KY>;pgzFe4dSuV`-F@4DF9=
z>j@WP0Yh#!Wqdi{nxnINjc+Z8-I9qZn4Z~?gs4Y>Tx=0E5aaxSDm&nhA-WbP61eLf
zDICribTCjUnlrzJlQFEBov)+tc>Z@xAAdx7`<MQy^ye%E*5mS$n)ryG7i6*+Qbb{~
z(Z<Pppo|?HCX9&trPOQhUN`T~6>)m$b$&p0d`dx8@&zA3Vsl`kBRQyTU~Bb0B{cNC
zVE!3lRS|#UaE_<ungr!&)n2xm%5<8&3{_U?X67aKZK{r^yyb3mjf>9hA->X|98dRB
zM?JE>=g%vuSjR?J8*Tnx^SIF<Em|*hgtD7fR!vu!kzy?&4xUcv*2}{o51g!L31$%y
zx;kG2(%&$Lk!V})PUYA2yj16Da613&+wtK1e2fRYTN-urQQOat8ix3H7x<zrE%!5J
z>Ey8QshKh?+X^F_%8FPpf_kR%BY5`ZWZo4gCdR@#GA_P-EKEQsUFqm%$;)G=8imSf
ztbuc4($F4mbIoP-8f=k7km@BarI9efwa!NHXwi+#PoevSi*cGwA4kR{TVq!2OI}eT
z<h(2(j)>5X_lOYAq?=yZ?>!u{hGXJ+iwiD*OODLmY^oYu;qekl&mwIw_Qt);NY8c+
z^+_Z+fMadfs>u5Be+TE=_aLzY)KA#c1P-B%Fy<vo!}e`?AtEKR$<CR;kX4b_(PDAc
z2Ar<#m9~$m!o}1MCx~jMW`1)W@ec0ui2426k8Yq_3i^rG;Ju3@f{a|4iXiRq_t%)U
z(yc(Ugh1Z@c6e3gg|^jBdCHsGSaujZTY+7x&>hv(NHDH^m2(m1dHF;lD<b|ukxLPz
z(=9~a+Cb7<U~~8hF3VM-ZDmJCPS&Et55dYEq=I6tqRCf&Yy^?jyDN6e(prk-LV>+K
zmLYdLU93h9JXitWbeXg_Dt$Y*D<f3%6v{R~KP5{Q_^$N!Be>b^bs?1<u`#L#TXpD2
z3`7E@AG0`4D6pD{=3@8vi=%pt#!aM)|C<GPe|oA=ba<?ipV|xKt1@)f`_S0=Aa1-0
zw*<N`rf*IQxl_yHJe$s1R8-KBDGkCF#M}$>dL+T<Xy>4}-t`8FVZNbqAoJ9_zrI&h
zPD^D#Q=5;qNV*>#snRvw(A@+D`KP8*TZ^*g{_ykvH9q%`6fRHF13YTZY>@<ZX0hqX
z_W~59>uPw}3#{pnBqh$hcASgKik+)=&+Q%zaB&qjIfN9`PnNBkFrcEZlPLX}9&_|B
zob9__HD%#qSt~&-zR9iiPKK~?a=l%vw`F1H+VIdoPFA^Y$HZ@?EsC-X4r8WSi|0on
zKn!988nL+{SLpA#+gg&4bMX2XJiD*11NCN7r(%KewW8wWgNol2^=UuHQ83Jd<@1z0
zDnvZPG-jq>tJ1kT55A$Yo)-SB657GV=I7J=wgmg6Q+GEsK=iHrVouU8gP6Sq%^eS&
zi)B+JQrlTnO>9H2TE}?Og2{4bt_Q}9h|SLms5cY0tvRLxhBjt)^Z}RR$`4f|xxKVS
zvb1CbWihF$3a`8t`ajSacYj$xvf}gfKbqw-t=D$6AoL~?2uu!m7UTxC1pS=sWHUH2
zp7xGduzS4ia|o+LR)mvi=z~%tJvgswss-D~U}Cy}P8Tk!PH`P(m=C@Ep_L%EPeCmY
z911T~BnxtfcFs)w0AKQDYWQ#)JI=q;WFNPzJK;6QSD*$m%?EdYxe<OVz>7H6utQPN
zY3hlI%!x0_>Y)FSmdgA_(4#e2wt3CyGeu<8Zp%R1H?Nd^7c~vN*U=FQN__#DI9O;w
zNM`#R9W#i<Y~NJdM%Wr`<5SCw@`=ZubTKt0J8Z~)gOpFW*@X0bMor(`Ow7cbAa;cP
zX@A$l4|L%NEx9i}u2vvUXU#^Vct|G572rKGL(TewPflE#4EYxL9U>N;_w*;y9&j2I
zQ&PT+PWr?r>OOE|OUS&(*9Tt?->o6&4NWYoC@oJfC_o!9$&Tt?r@+CnfA=QfSAJ$r
zcnT@CiA?sV^t8!TqA7=MjuP2dnubYi@xeY3SYXl!Y|Z*Y>z}Ygu5<=+j9E5P_ueJM
zkCepo?Z<40{rst*)pyv2qYyY&P&GR>dkEW12j8NMYK51QjR1JuF1$PwuPjZK#RFA+
zA34?8Abpj=I5|F968nVJ44>`{BDA=*;qyd}mS~UM(==4?%!kj!5xipID`g~0=IsT@
z!EC-zJtR&LllW7*(nx>5Ff8Qe!wu<PNQkNJcs3-=hmm@mGck4Rg`DYD@XXXJqQmB@
zB6T-`sJ)!Qqn?_X-k(o>UjP4pn3&;9xZkMkD5M%$hgo!)lU34GEcH!;sw;6-S5~^|
zX#V`|k+5v2Z{!zkx(rT!15>MWrKcu;djY%bC$uNI_HZ?6W4Yyad5z3V;a%9@|3mE-
zjVLsEY0tPhG9tNkVd|XF{?g6O(R|ux9Nsq^_j7FRe4QTYIpWmjpuvW_Wik;OmbB)e
z)$jU%xDLm;^PS-Ht-(3cWP&O$dU+n#>RTwc5I=v44cVWFuiMeZ649tAir52MjUwW%
zJSH6-&^GUCS0beOIqfn|u8xVRxK(!Qavr(_GG>45h=}a6TU*ea1|7|M!SZ2iO~KVw
zPoQbrb^sXR^;ri!w2eMmg6-%CtoAv!>|w}Sc?tekZuS&!0=#9C#^RRvi3xRZCu`g}
z<|c!}WRlm1vRD>Uxw1MAPesGHyWjD@*D7P-pvWnG5w8`R9{Oz1J5op=DzI2bxD4wb
zIZh`=MaO=o^Y1d_9dLKs?Dk|?K&(fhM9W&=ax~YA;sZp}a^(n3th=X}{V&>fLIVJw
z`byPIYk0UX@kk^sY%m=Buv9fQXIxxYOl&7}U6m8(a_}RsPTn@xd;=fDo}AKeuH|h4
z4=>hOOX|yH-*B!p6S@x{1L}%x(ds+2`N)XqGqIkn|3(ezj4U4`5}o%7)9UMg=5B$E
zv{NEd#?AEg{Q6VhbM%Aqs*UKS4<3c^1{8&W$KIMmzatdI`@7p0`%gnx(^*+LJHnB$
zHO^Vn{m20R{bp(ssiNhh08SX03r2G11AyM5uhvzWj9Wp$f>^Sv@6!{2ez|fN%BjXz
zGhq`_c0!~4i{Qtx-vO!?p>;|khN({;<>8Q7(zmde|B#T#E~-||xV+*P8&J|6-H>(+
zYZI~jw@U9w|A9|wX%@$<^oAKC2xG<Z?^Pia7KWDkAMY^si1@$PsU;>J?shFpkHkt9
z7sD)#$7(EdsKfP)jl^_nHy+zMy{)Zs=g4wG%dRM1>`)p<<kk(&uH)c6)Uz(Ejrs~2
z=nel)^~yxur8*5=zI`FDevxD&tX?@))|#kOIW@<aypxTc<AW|;(djf36aAw<b>V{#
zukR`)`}*7{uo^BXDPQgEaPm&;dd{P6*QWpa?*#KXSPFPtav%p*3@u5C6#yzD%6|hY
z^7royDj7aLGKy2%bhdrL=sxaMestAzmN93jaa7dNyq_FXmT9=MM&MVaz@S=!R*y#4
z25rmdaE-$R^<Vp~!!wr)vRWtrI9TThonLLTf~(Rn(Rcps35+2nZFp~k!^lsl!t`hA
z&kOaeq}`F<gxx(+zYH)ssIgLY#Z=S@syaGAM;;m+B@s;ID+!ZIe_hh>uu((h#jK|K
z!vwheN55d9!vL+pZt{%G$Jm~pplCMt_dijaQfEf+q1h@eW^4=C!Yp!h7?ZQ~ll=4d
zTg6);aM;Gf56<eVo!y0mof55d26BRnw9`CmQPHbLCi%zjF4l3oqI)z_tpP3qk)kdO
zA4m@!<C8SMdUxL$WhE5{hHF$%Y80j>Z^FgwAf-7Osbr0_PxOrRhBj;G(J)kEKIg>9
zF%-1T$su@D1R)72*8iPekC&IhFti8(s9AQq$ts{zAG#ynUD<I8A4<U_OBIh*VT7;l
zZf_`7zEN^9nJBt*=H-dbfs#>hHK&Pqn2F%k8?LyX{*k|QcP+J8d$B7OET6BW)!idP
z7rwN-X}7qcOU724gx`D>&WVDlm?t;(&TIDw*K59#$emuY$@Nw}`*cUK#A~mJftG=h
z@;y76E|Jr(YZn7lG+BQ;UY^PcXKRw(i!@VHk*lk>4~mWQ1mWBz#u;3;uvxe~J(9r@
zs+uF)H$Y*P=%NW{c|b9{*`cKD4Ph91@*AN(nS?m7va$SWpQsY?C~<yJ$nW^AH$LR9
zjlaQSv<}0IlY1XhRw+yu{!>owvzM@NhDxJ~V0FKR>xM^C^0Tl2@PUH<E~M~=L250C
znN6ufIvbOdYl(?E*Mk|TDh@_nlMK`;hh3kDQDAv!afCTl@A`Za(5(Kt?tV5=7ct$^
z_HDZL`_aHy_msNE=KE&%zq9S}AiBb2YBD@OU%8K5nd`NYzli^T27?+5#-rHq*uk|h
zfn}vOWu%A1yRY_lq(Q%92np_DVkGAvj|AMC7y}(<XZCI|jPP#+%w(rRf2xrsP3W~k
zz_)qU!!8`ulxfJ1Q`HQ)2OIv#&yNp()z{MxY?y#wV4<S&yh`or&Q!wapj^3Fg)@)|
zJe8$K`z_JCNic*gsW`7H!Bu(8ZNqX<grKEWrEYqdO#EtN)i1S(6VKn@C??7wZ=2K@
z4~OlzocY4}SS8f^*f^O;3ms0}hdJw9dSmh&&`wwTMg0j89h9{AtS96*9%egc-$AQ*
zf{UxwJ75fzdawHXcl>W-Zr_H)GPg0gsztfUro9mfNW9;hzsX2b_Uti(SBRCIpydYd
z6C8&uEK727{|mR__HhDJ2z$?(!wMQI&x_cu_%)>k9)MX|ic9V1#3F(MX(GB`P4I}V
zS`1Jlqpw}qHc<n8<D1W?_fA^R;#GD20tiL%=V~;^JowPT;Vl<`GXL<Ql%-<x5f)T2
zy@?7*7K5OW)kMsWx+;mczlx|VMB>1vgwWK??q|oUEGAh?)u`!|7ki9fWBpcx{_<_j
zV^CtG7#OGd&l-1jdG1Nn6Ur>W%2~n6C(u7<U(SLuM&AvN*MQIn1W+bc9ds_-6uQET
zGzM2czQP&{c!V!B?aW>!|7^jUdmKJIl#dYp|8mCTk1pr@#M7+H;kdCWAMbeJKck|h
zzxa~qp239$Lr%qK5BmQI+#`9+(9{hWN&OEA8I#WMXk>x1*?SC(ug3Doub8wl_l`IL
zJp08Q&6Xoc4LHBNVe(jYJ@U)w$WC`r+24FrDb&vU^~%nA5{nCnD;$f;on6-s=Ia-J
z3gXnJqD{Mt1rv`)M!urnJ`FRv4Pe)*vB!oJ+^uf(EV7O(B%b$+uU^<+$#Ch^caQIC
zfvR4$vgye2bYrP?f}Burj(U(G_Wmx!xqvMUeID5dIO;a0?B>fJ!K4z>?1`c7?rx7O
zHb9Mw*iLmUf1|iZp@4kqJVPBjSHE(f_X+6gSxzRWeutC#8O}*`P@zO^Wmzxk=H^y2
zGaC+266_8Bxw+V##-Mg0C5%ZSPszaA#9x`w_@xOt>7SNyOCY8AX_P2ypS5>j$AtT(
ze-&7Dcb<I3UZ;t#bG3y(@mXuNasO|PDy0C?+s2x2xUuA{W0R&<2pd8}f5+_~fWuEZ
zI81x7`?KCQp_7W{DI8vaQ?sOsI_Zvdp?k|&aPLo93Wr8lTz&9i;Z=XIMH}>WWlfwg
z^}z4#9%Ye1&9=dkrZ&Y^e-4*WM9X;!zYFvN)eAFBPbOC8U7UqZcd_)Rk8iw1Mf5F<
z+PaFh_y02dA6A#HKRb{12^$D$hPyuP6voEv077({N)=Pq7GyI=AA$Exis$s3IIFtv
z1%7ZZppwjYs!!S+uCgOJMYI7r!mrrzyHr)RC{eGnv|NAmn>@qzJUZ@+<k+ytN={jf
z@rH;auOP`my_J<Pd75Y-KeR7*Qdww|nC0elLP%!qEjDnjtqQxldUHB~220Y~`uA(o
z*FCvhC9Q$J?8tKRH|5zZMbbeU*fCgoaV7XUCer%3VfgkJPjI^F4k1qI-g<k9eV*s%
z3`iK%6hLuo`Mb61+U#CBk&yKO!+AH3!ym8#MTUYk3ulb}B;xp)>a(@w7~$ieNc7CQ
z>t>(!C`BzGto)I-gWumg@Jc;)E3yZh#rxC-JX0-)T(GTu^6l(qWQK+Kc0O=%?c3B)
zt>@LyJnordRieRZsDZNtA~!*RA{UEEoK>MPaGW8sWMZ4cDnl22G_e+d)W_ra$*^&)
z&cj_G>Y$H73W$LR0qPotm^!HI>u~QDKpIbT-&}KKHg~MqaH>KeUL;&!fI^Z%4VV;E
z(1C)X;V+~CA7Drp_j9>G<LHLpcTdr83TObc)nC-7i<ZzyW!X=s|M=!#-Y_OZ;o^s4
zGQ3^+xr2IWax^(Q!(Y!ruKCG&ib@Q)AIQ|~iYj}8T?0QM2$$?ruwK4?aFVIO6-@Gx
z8ME50G7<^yh;9JNC-+Osh{E%ell^t29qTm|w0oIvf2ZBH6<JSd(Dh<gqug-B%qlpf
z1`B_XJF&_v1X^8=GKO%z${f~+w5l%W;8=r~wLZ<yFSk(a>=`+dAV`z>Y@5Oop%;5O
zP7p0Y!LU0XZOT_6`~+-lol&x`!2yM$1U$}?Xdkn#uZ>DkFR@1|hOM!gAIpFiC`R#^
z6H(%+qd>dY&=`({Cj0Hcw5`%DEK*ngZxmy_?f4JUNulHX84I{n@@EK?`>dVnVCHK^
zmt~&12e+R@yXJS<QlLgG49N6(TIEliW0EEY*8B2{`w2@;HAK!i^q<;ss(bGtL&B0p
z9$p`e9B*3E4$3L$3}EXr0pM~t`qN9wX1rlxRmyVI2ab#o2D(jK?4q2?n*!*pyZPU~
zPP2{8KQTnC4bU};9yh9$bqPvlxL{BEIfVx2vEG1Df5FoKX5RNmple}%HZZ{xv|bFz
z`zk;$`n)5J)QQa&W}-iU>3j8n1=^nPM$NXLMcsUfjC-JA_qs80oVnJ$UbwM-xMLv{
z5MGea<G6WDWOmms`I(Z^(A=o&XxaBC*TJ#3ttEZTbGZ@+?HV`69t|v?_Vx8^&ovsP
zUVrqq;uz@>3&UR}gZbe?L$MOIcH{{}-L#b*0$2Uu^sq_(nuYH9S<mTa{Dlb>pb8$H
zcJx<v^xwMCf^(|?-#hB1->g0(^FnV_*b%Ww(^<b%U2}4~sSO`n!8*xOP@=;v7r4ux
z;96!aOKNh%MI#pi9H$&gwgzTUcS-J$8>y+n$mR`9M1~@3Nh4Y+Rkf4d)Yzm0OvzGK
zZc}x&$A-7lGEzGI=#Opjb>+IwrTNYe4c@7kWLs%-kZGdr(F<iM(89dW@f69)2ePwA
zA*aYP6ysj94^WNS?$w1(=&leLuif02eS{nzEECD9_Fgk`hn4kv#=jhruq>pKt*~qr
z6SR#eFTXJF96+va#@{VsnEq`?ibqAoz)#kb$yhJOz%VkiVPYdd&TV7ZzAm=Qz&NEf
zuJiA@d5(DTOlSG)`Bj%Z5cdY1%~avz2~eUfxDZ&<im50Z-fW9t50>gX1O?W{C;0?4
z2|(&Ls}3VlKDRWaY}duoBS_48wzk$*fpq9Xnfs`Us)P5GHOmR2F*SW(bHxxCXNO0#
zF`t8<k5>vQFJ;FA59`9XmqS=UdH8O5H}NX+XSle63f@GANW&bhT#o(o)04Pa{#Rd8
zq5JuqHk0b7wfm_8UpK0_?%R~x63y4<lI09VQ=a4`2?PC?pz4#PrC~rBeOz*$NDU=P
z6QgSFZ%I0q&E;0%!zEC+bnsZe^+oF2TZk{h7S%U4>XMCZF+bg*TX^S^bYuVQGi*Il
zN9&WC2wrdN+gD}!^!Ex`$lUvTx%@U@i=(_^NlfkCv-lZR!fHDx>;x)lwzWI}=EBTf
zlRp?C!`0Ko&nB>+@(Mj*Rp$~GK5i?%sn9;7zB0f*fI`;CswK@<m7CFLhzpB;Xw*J8
zB_aa0)vFVP2~ZJRY|~YN-1N4n=G$h22Sytu8_=XQZtDWGU_#+y6n)78(PTPqI+l0$
zYO_1yQE8g$q+7pu<aSlT8Oc~}SoXvMI1p^nD<mw4z5dtgQ?)Z{*?rzW8@Xo1>uVA6
z3KB-n7&|5NdK+hrk8hj&GoE!<oLP0m!<+stN+Zx`{*~Rnp$J+1rh5Nk{8?!LqvhRo
z0~40sCrNLnY}4QDYsAT#Rc61YIp@M=ozos7ySp1?7z>rvJwH8c8T<1kxZ(X#kYwou
z>%}|{(?T$^o&x!PxYApF12k^|0ngA>=&nJOe@Nm4nd0P^<T(k1{n%b&`O%RwhBoS`
zi^tYR5mNBM>4XNPnknPjll3Pl)f;l`d!^$x*;|BpAvn9@*F$D{$-0_%_P`Kf;B}2*
z7Gio4srpAWXW(#{w%=Izeo{VqbnYyti^R*r=KvPKqMSjNCetv&!}@{ztN5uAGHx^o
zlfcERAX)!XppOY^qr!3|6Lwnb)Z>7Dk8kq0;*;!KeJhA=Q~isM;-8QP7PriaV+0gJ
zxptKWMCRxcuG`j2OKvr9>0?X=AP6<E>oeF&5r<{vZG@~?tQuHKUh5XPtbb2EH-tY3
zrKEx|l%2VC2n-BdAD-=;SdC^6G}&fnVkIYt08Co6YemFSGF$0Xn&X#@u8NHvAT8M`
z`dxV$@yT+~B<b!@vuU9wJ}a^M2(+Wyz=mShiIK4=o=ED0?HblNf}Jy=(ErhXy?fgK
zK}%AsO-?opLk1&cJ0~$QsIIv!GdnJ3fN4&S8tezn$@JH-utZeUR1~-=FC1=og?xEA
z+fN9k@{|;qBJ~ZeK128tMNY>nw@JaXG#2kulA=9DlLl66@@HXq=C>65@2T+4?@UEh
zRYd$N>`Tc3O8|<l;}N2M)#Wr>bAC=Mm&u|m%@ki?bJ~xL`P-r7rm;d$U^wkqfdhJ`
zLQ<d9y{fPPX5cLSm9rCM*}#p8y7=>i*g;IMQCoJ_s=oPGP31j(C>q3zi0kJQ{`z~Y
zjV(@_i5Dx!5V>q*5||po33$@h`f}j1HRreby4r(zKbB(Zaa*Bx5NtyM7c07wV50se
zG<+UA)~Ij6QoqsA%E~L!;(Z_a@;Vub$>oULh@}G<I7-iaIL=;v;o8YGLBcQM;LO!-
zy#8u)N8i{&q<VtU)%yA!Z*`$J`uuh?eu(GBhnqR!^)-8b+O2f)1E0Fgj+MLqwUB_-
zY4_WQC`^74aq+@7;1raY7#AKs)*3}oAhh6668t-@Zr4BisMf#%DD9?a>S!JL_18%o
zzG0QEgBzC;JBgz>s_teZ@MU0~$_A$3B*Pu4NUG+;%|!kP$&{U)5c>Vkw3es96G}?L
zuvq2sASX7<Db&IGc*0FXAXTSJ9s8@QKY5~jY<|t}z{URP_YVoF{P3la!LG4)p`-^A
za<5Ask2k;n=0+@*FX6MuL@sz28aG=DN}iA9%k<X=r-YqZ7R6h&=<Z<*($eu`_4tu)
z2n)G-e_&*+0IK2&89ArgD!GLFWm`_l)#=n<wa?52=iOYgRO+Up@Z<iEg_FJnD@YUj
z(4?p=U6vM82`BIg#X;jSZ>Kz;RvVogKo&wLDvPGaL3%|*xXpnN=#t4tf$pvNsadsE
z6I<QY$g4}1;>)$>PsKy#q}2ri)DRNYaopJG$e6?+t#lwq;L)=yQlLv^YN}E;<-<l^
ztt(HqrN!*zt2*ftx$i0E!Zc@V2?<^)_2-<yiko_Hw*?qDD6;y4gA=Th0`u*qf-M9V
zJCOaMZ5$$}S|LR8bt_(Sc}i<Ic;Vr(-}A~?nX0&W8?QQ)2Ysq6RcRu_Rg`=_|B{)|
zoM0Qc)8(7LMSk|LNchPR%s9J6aZ0QtYiqIofpj$$*8)-t8Fe1!J;A>4Z$<!B?3<B)
zY*etZ*ui7ipGUvR&(s6N1*sT4HzQ9eZ_43_V<(i@5I~)%sMtl`Z0hI{_B8ZW_tT|-
zH@+;K`MUht-0plxD+L>)uf}0x9pzp`3J)lZOpbO!=Xam-Xl7{nefQ-juXEhk(ieJq
za$Kb9?Hm&$MiG&(@Gy{10A`!JeOy*qI^;4p*Y<5B@Ls~OE%F}|7O{WE=C`8q7Gc9w
zRrRgCjcb@+A@=LjgQI-=zD4auc`e0YY+cUKFi5}b9Z6vkz}w_yX-~5x#&gFH#$l@#
zra2HY#K`iHoeJ5zwtWc9GHU6m&2ZB2eBIG*mwu&yA+$hX@cfz&F0Uj*+S^J#ML;5|
z<@J}O6L{sXwqX6cf2z)BH5M=7b=T-v|4xx_*-8-c%-3@$^P1HnPlzSN9xqis>NGoR
zb*TE>{u6k^A!V$8nPi?H9Hlj8Jo3ME13jU1ZH|(-`6jwjnFx6=Ot~7qknPmi!B6Jo
zJHXu36BQ<4+?zK_vld+eb=3P3ivnJ*Mo4G}gfyL-LwpSZ)<hs_PZlh0U*@8$U=g}d
z?nkL;wXFA90q)?Fa5YY#3#@x!DGqH1n4TN73TjD8{Q|-a|D5*LXkRh_B7j{U_YTZI
zZak-B7rY8g`qa;90XMNV7Wez~H-MtQ60&3IJ<hp>n^vm@j4Dhd@Ge9=p>>|tPI98o
zdvg@40CQl+>V66T`T&4<aLadw(Vy8du&z&R8*9@j=W7m*R@`z5mc&X>I9iWKTnC~u
zWg2h3$GE=Z-(JX8&bTzU7|t_)?geG?T}<vx5c*EbK${-YNOHr^zqzs<oDok8%dlaw
zUAMR<^9}?H2{>P8dzU-cPY`${;0}LESCFTOJZx!dbv!p@T-1F^&eV(qByu;`n&IA7
zbnp`ZDChg1Ufm<D^&3%|7-gUqIJ|rFOvuxdm(zBR>g&JL!6%Kr%tzJbwFng6(y>Z?
z&dA^+E1L>*B(yK~{hqFrHII_DlZWfAg$>NJ?$MTqvk1x6N%5m32^r(r5H$*{XXJH{
zCwVqzghX9765+YIg_#-U<ZwqcgP*a~*~oWygfvQYpA<&U_dQ<C+H@RD<>=ZSc1Q~(
z-egYTMgW0AR97}EXe(YzVi3GALuPV4tT!B~;>gNhNk}|!)_4QPu0PzFRO^Ng)a8YJ
z!nVSyqJ#3XCUPbqVW3ai9tn3nxfO7T`e`{ib+-IagWEE@^6qak@9(prmoeY9^}m2q
z9$=A*Djfh)ApwFaW7r3Xap_$PrHXcKkEXUS%Y^N@g+6X61H;2F-JWii!LA0IdOEku
zrhpXK@l(gf@D|rx^J{sA@s9l_oEaFwqShoDfgGzrIRo!13l_sLp|d2|vLfQNc-h$%
zCqo`LqEL8G_YbicKjEQi<!baMg`l8m66`P_uRP)%YZf?Ybf#cP>L&A&0KpxB36+LL
z{y1Qfsx4OK>pb~pYIJD1iF^Oil5}GsXBs%(eXiPu>OmG<8yhD!_<z9c^+IcGIVyen
zwj5H*#<mv1lpMEE_k=VCT8u2iz+fQV<`k1s-}bxoDY)quy7gQ*N?mOt#{Z6Y%Ogq^
zK=;~!JagdSup+1RR-Zs!jfkiWa`ft!lrPD~h6zdmQaoQLrZ$?AQaJs>u5wj-Z(ov%
z%cv8KUY`D)U(nfc;TuR=_ts@{QbRBQZx$e3feBXb#m~b7VKcn;gp2Jz+!Uucv4a}l
z)GS-2auCv!lgC)iv;a51;*2?3f?zg-2nkNi6Yz_);3+%?_`MP3u}Er1b5=!V#<*<v
zYEpc`<mTpX*Lsd=X$`cRuv=Sh64Mi8b_xdj45FGb?Y9V?dm<hwKpZMV0VI&kppO2%
zF7iAl!b!8e{(MaxC=?YI8(&isgF52mwd<Gktc4xqSr@gRj+-ZiOxoB9Fp?MwJz9w8
z6*mk{ZxVRK=kx(^wS{3tj%$pMQaTR=b)Jhpy!Vd*<#e|qL?TP!mHg<pA-z~Z8`*oY
zG-ZyYL(KnsqK5kW^Nx<Wq{{HgXGNFcnduuOd_uKa_wtLLi52k5r5Qm^VE7m>Tve;D
z_v6=Z{HL$jA8O2p&pmP4iupi|;zc@>s#u=AWTQU-CZNmqj_tNqH=pUsW2Ac>v1^z3
zk^@2wVnYjCgi7SKXhVG?J|JC$ebL7N=DR&5J+MYokd}(Xu@g4h{$=(5i4whk0g@&a
zC6^qFEki%a!=#PUV$X21(ry`U`k6>B4BPZRBrg#x7JQ%JdKMqwcYQ4hC^NkX#zy55
z(0EmOr)DFCMyd7m&AO<)OH3!HAsVZ(%@>5=nn_|3`U_>F=n}7ElvLQvdzmEY=%1aO
z0tL*y=+O4IE*Op5W@LIi-glJBOiFCNe`M(Aso{6_B55=*T=AEOrW%oY7!xY*xzfmx
zMyvPU)aKSaFn#PTtOJ`Iq8FpYQfX!vwcM3uZvOFD&GGFwp+j*ELpWY|bnCG@=A(n#
ziaZVOZq2jH0$@a(wY42@Z5^W%=A?b<XzS+-r^VrbIA%tA?@@B3knj+LrC1>_FuvB4
z*&*xq!YYk2UCN^Sj~$*`wfOk5^FxBl4voY_R@)B8C^}q_n0Z^CS@98eqPd9+^!(0+
zkqBtEjE&r^dsa<h(jl5)rWBRo$C6@?>1NeNiw{scho!Z??i$Uw^5v;#O^bCxme*-A
zBz6^5zZBAqfHufS9rkyjVqduigq(I0-ur)C=+9N6v9GF3*Cl}${L2Df&q)I3^8Jgf
zb=+Nn0-4eqR)ZHsgy*%_6<mxid5Z2zf82Xj%|f(G9nb;VZNQuRN?Vk`ZI*<D!skUK
z@Zw9g_Z3O)ud{^wjgF4%ay=#x@c|0);n6L5S4EquBprjnAKk>7sFVP|yl3B=FQqI!
zj65FpI>XyLrhcFsexFqHD-2P3y9|5t$ZSV%aI34pO2Ada^7@Edr+sidIv8w|aaa2l
zG-;>M^oKDinCn~fC=cQUA#O$M&+|ierjyluQGUtgy{a}x@m3yJEAIV!{u%kYG9aQy
z!mZM}6&Mg^A&V~8kAZ5#AMg!C4qdJe0^BHd4Uf?_tDVV#>I|r|5iZWp@Ys^d{zFP#
zgWQrq;5LX@&&k-R*G{;6c^e;Wu8An!-$n<zPN>(7YjPH^4k=$?K%;wvlG^7IPNLd~
zXl>X#pnl5Aq{5;fi4ISW8m?X=$_hHe9wS=06RUxt&`;w{wy>uH-HN_<&dThs#Q^mT
zy3>->vELJvAg=fD>$`sVzKB7c-|Z=~lNM~2V0-I6mI1~$G9|Gy??(Z8LeBpNje`<|
zB1nI@7N{uF#*PAY^FHZvdiZgW3G`EYcFO33DYX#R?&;}5evii>%d@bvpq^vj!G*}m
z`A#I?O$!n=zc`)M>|O{qoV-`a`?jwCvf|E@(eiDvJWatxyqALh%OII=L$?UdZ^lOY
z`W8L$@EARFi+rbGV6CJn<blZL3t)1g3y=ZxPhhJ+=k$8I4qe0WUX$|D3e1Vb&TFSU
zxrIC$g!4-oIm?9L_TZvE0S4yQDGV9NQ}+VSafax@N99_dy6f8w%i#S9-1tv_BkM6(
zJZ@9~4W1PHL#Ip!u*<<z6@TztX^2lCD(}UW{Zf!@<ynbR-nm696~!Nf(51e$Q5zJ$
zW%ImKx3JmS*FfDkgbV26_jVS^omwx@RW4oK4wSCAqEwU)M$eKpI*%P4jPCHs42|^u
zgny%$VbJW4fW&eX|L$j`hs8?9J_AAsfD3wQIh`FB5c8cq<w;|qx8Vn!Ns|PV8GO&4
z_&I7;rc7dxlG9O_bnw$f_5x$c-=<)q{M_hiwune!;uTZljKu;H+$*wsongM>iDzEa
z;N@B7zS9|}6NsOge7Mhwsa3h#7TN1souhp5<F7X`S95P%m{B7lC`(gSMH0|$mj51)
z`tQ>DY4k;6;WNxO5UoLSp(-ts5hs?TR?Dui7$`|sJvSLuQE56A35f&g7r;%VEaxB6
zGbS>blRrp#_I$}m0d-b3)!0LJkMa%ZN=a1$3&Gl@aW&TuQx-g5_%?nwhna-|10M`f
zfS>T^^|g76(eA}mAzjZvR}xM+TnS1(p#flQ9eu9n;!Vc<^8s$+qOOibNFt=5!bOTv
zImuGR!MVQCV=Q%X&R&(m+Zxy#URXmM@IC8E6PF4=&k^bx9npPO5_IF@3fMwIULNbh
zCg_AiX|C2f=MPd#X1iJhCjmkXXniIQqClfP3wS8l*hh3d0X@LlrxVYoj%IUjWASp?
zy5PMM@a7LZ0jI6i)E^nubDf;otzI;K=LwH$S>-#^&&&}8Au1plfK^hQhURUZMRYL;
zO2M7cHhf8w1G?$;T-6@W4ty&Wle~ERCB=H?MdKy(a?mw4kv8wXULav9ze0RT_VMt<
zD1K@vwL}+MMJcbd&5`U*-%WP{UA_R}#LmfPaMOS|w^*eTxUeK?)~f4cj%8XkHVz1@
z#03s;W^5Ezc(k;67V#IE{u}kPOZtlA!(zlc#LOYsaf*i47_??^TR<#L>U&jTokR3q
zj*yG{Q5#_=nxrHT7G7vb*-LEAi=FoaQ7Zket4sD6e*W?bI=4e(EO`GCA5zADdUm&R
zYB=Ueas7TVmv1Bg1?LHBj^)MYy35fm;G~BTb&Ag48htiuT&b_N5#H!%zHnPL@tP?M
zYU}GGTt^3%hS|<u`{W?MP)XIP<b((%Wu&6oXDCz^m>JtPa%j)CA(xEcl5gL}&y<8s
zxa2AIbhyH?BV#(sC7WTpl{68kL1J_abs4gvmX=hAt#;MHev#$}sSS(!$$613_PaDX
zm$%f;$^4=^@~k=VnYDgk(t|kj(T-)wQl`EkkF<=^eEqijmZ@R?t@$c&ng49lPfp<{
z{<)|K=9}TDuGG&=<z{xa2ut*<y=<PIzK0(_YFtnyBud#vn*wLrCt@tUuRDNK99MlL
z`zs;>xR`I=(!*4OmEQhc?2krdOPkv%T%QRnq`VKmLnrN$orV5Bb$klq_igw%3xTX)
zD5VguNr{HS?D^}+^F$f9ueK(xwnuB4(MXbrOW9eMcTZ`I99%={`T!}Cqg5SRoUH$t
zMX+y@n<FYH3$0oVUT+CWJTafFl0W|6=sec`aZmHRqa|=8D9^H~cs&U{`D`su>6#m`
z`YX8<9ex+AFc5KuR))Xi>JZ7?SOX1pJNcD6WW;45i6g}!KDf8+5D>?<zNt7QTiJyA
z+HrN=Qybdp?aZ`l<<$Y$nScTWvNPW?+3x3?>izJ;ClRXH)s?QE-p<AO<g=y)r<mAW
zW{BacCjD<nP~fwmfHz!>Bv2IhnJbr7gW1xAek60A?c<ZO+q>v=tg-GpP%R+Sk%=Gs
zQ%3=)^beRevZvz5<{Tw`E^8$j7~WP|kXKy38f_#ptb+>WDPhI`6#MM)5Y^9zODCn`
z^l*abjTbkl0la~Pzb)$QUGz*;jMTBeMii2)tnDhxN9&%=`^H2l-gxXZx#5p>+t^x4
zI*jymb@i5%eY%-&H=3`pw}F#$Qs;z<{VNQ8tl=hI@ZLhZQ_ou8?$4G=d_Anx_%}|g
z*p<C}<R~!zov%ok@h{sCI<8L~{2uxu56n*u{TyIS9IZ8O<#MgpG#5U-;ApQ@)Q#M>
z9Ft3t2?dcxE;mxAvPza5fD|uLs%(`6tBbW65d$CS+FM5+hPhdBsd(P>(h+*?Wh&5G
zH#Tl7>sFUp_4SvrayP?PIij;YnxymhA?CvU{d@VHtmc7;TsLBpMe_fBW-|t`D(~}2
zDk=hQN<f*F`vFGS@TW+la3h;jRpa5i*%_ptAd*(^w;#ZiB_V8LP7BmObeVlFN?5+b
z<1Iq2ZHwIC2rQ>m<;l(Ffvt&3dUHqJ@xf$6{$8P{r>j<qH%(ma$J%b9sR<q8%r@8M
zayLu@bZ~iVP-kycOUhx_t~L}o5%H;Ez_+zIb3U;l`cpdaaY|$CW;Nhn@GxZePxAm$
zG<fvjWlFekTBPs5&9}_V2TNPIu9W7In3&G~N%xPo0PzFlacKEL8eANL2~EB-L-n$;
z0l;@4D?39e?<dzUe+w}QUL;1s()h_{b?`zYU_~7(T_Gi3(MELt=<4}9GNxpBBsZiT
zl#kc}S5(y8X`12BDGUaVMUqt^XAJtqojsccv1M*~-gfiH{&}+9MSm1wkdRv5w8tyd
z5CR$dj<`j{ZmXPbs;MXjtkFEuz5-7$_+OuwBP*NslIAJ8l6JlxM5$uHGxv+M@}4Qi
zZV9W23rw$uQATzAn0T17M*fB)COPYgwq_+oMrIT{Z9*jNDM+t%fS?K(<gJN1W$b~2
zWzKiG<ZcSIzz+dyh||8Qs^fbQmrnni3?wbf76Q~(UX%x=dnV7LC9(!?tQueXDr6aa
zjvXM}F|Izb>zthq0QUB;<0B_hn41g}u+Gi*H~gCu3kGhkHF2Y(S?z~A`_6YI4Ms*b
zH#S=EZ`|!&E+4-&8KhV{%qt{e4+a_gq1ZnN&FazrSKfC9MY%QWjwm8YHVT4@U;sf-
z5CK6XnZP0EoO6&IC4(Cj5RjZC=d46wNP>Xm43cvia?Z@@(Qof_{@tomb?g4PmR5~0
zGw-|B+x_&@y?U+ZG>XF9G>!hh$4UH~@s;(Y<E&5DTR;zy4O++T4bHrD542Lu`y{^O
zXs|KwA<&?p5T{SKV&FrjTRsvNKA_uvbpFs$bY(y&oiDn3`W+t5mj&Gmj5m(<x3}NI
z!a1Pd8myrWDJwqAQlYSFZr(rW){8y-#1!Uibx4IMJWmWubolPjUVE1N9JV*wcl~wj
zVTL5b{&FjB!mbl-(CbLK9cmYc%^b53+$##ju?H9U2q9J2cG|j85|96Vc*u8Dron3F
zt53xxveyt^pw9u{#!B-!A%8eA&Cg;3Y6C$5rUC*2%KdhyWt7Fnv-p2$PpzBx?*-y`
zJ5&7i-K3O);6W5gk|f7{xhN&+X1}A0XV<49r6~KQDHNcTLlPf}TLW2I&}fZrfokRq
zqzVkVp<5$ANsGfhzr|D2up`_!6$0qf)_dyL&zJH;Zx%~p1S>S%R~pHzekY<)_rq{r
zhP_Ac={qjPMfFYWZUw|o3Yt-SRPhnvqism^8Jv%2rV-8-$L!GgOKipu3vLhup%3aq
z+!<42e<#ypMXH<}MOZ(-o<II(oy$$bU)x%i9d-i1HZz-d!<ZP*ACKF&!q;N`YGZ9Z
z+1uK*y|>*t)X>x#s#k}%x#K=2+e|Kcv~Qgh9J1E)@!y>KbQl%`95hQFL<d5>z)5Es
z+tANQ<_ky8FfnVj@(&}K85^Kyq$A$-xyf<L7;jW(HXg9Kj|T=$m7MZfk_?fR^EaWE
zXawtzwMv0~*`t;E3HcqqrDHYVS)~`_Cptcu?3g+6uzEF?tbhY<oQQOc4>>GLNlB`D
z3a2KCG8>HWKP~$lrIv5^dozi%_3EWd2}McI=lgm*?Vh!rymi`qNx-E0@zmfc^a3vR
ziGvrfzr4^=d;3di(aie@jm~T4TH3V*oa(QDO}DPTTmBf3y+>6#`f;+1_x-2WwD;Y%
z<XS)aFOBY2Yc4Iay?J7LO@c|}^eR7&O2}bs^*M*HGEQR)BSFxQ0maw>r>9c?K=0o|
ze`D?BV>%pOPto>5f>BTG2gmu-Vjt_w#o`7gImUjq%4el=OR!v8Tig3LA-{{CY}N@Y
z6uysH=WWZ2-h8j$7A?<J@9;L^lr&c2aDL1YduoE2T>HuJ+4Rk?0z*D|eSPkD6Pcx<
zEV`C42Xc3?!_DQm7b!+aRf=Za`+Qex*NIi`Wn@eMK+!Z^Ity}WXA-hw2s7*Jx7AQV
z&{iYOoPW~j6<L9k_GYjiy8)zspl-b18=4N@@YqeciX8=QO<pdf952L!3}aipFbH<a
z2&T>AB|yF4<B0)YeC!i|bym#%NR|A)-8(3|Iln;(sk;$u&=@kt=j=IBu&F$4-Q4W3
z)oBTr5QBO$v@5c)bG=TtJ}!kKY^%lrJwgO;_dD!F`0fH-C*;r{aS=Lj_Gdq&#U=d{
zA8PLiB}sFv5AO?o7EZm*#T8jP?i$k+LeIajxFGt~ZQxy8?SuKeam@+n3RRXmBS4{G
z;C=LV+FZxK_$z4?ogO5N1$Vf^VW0?S)z44%>Pi)32KX11<!?=wnO7UEB;G;5$_{LB
ze7_rT5jG`2o&Gy4ncFq~7#C*8<JdFk@g%V*EH#-bG`j{n93QPrcVHL56gA?%kF_nV
zEqXv+ao$l{+E17(F0?i-l54Y>>0#%<^ihQR1T1P$Z}+aDdOxd1i^Y4zcs&`2cln&(
zuCBX*Z+k-rc7qBY9CWs|U8QHk_BO)m#vn}uB!RlHN6qZY+#4!R&+TC82J@lsKl5S-
z%MDN8@#SLf!``mrFt^QvLx04#$Qah4@%Qa{?Oah&($Y#uRo7dvR?ih4`4Jk;8|$Om
z2>EW-3}-*nG39afOs`TFHhP8opT5xi^rvu*4B1ZrqFq09p?D-Ct>hE+m6d7CF?NI3
ziT%Wff2KA`=Yr#g95jM*ZSBh%n`tTVyZtS+3HA0?YQE5T4VO&h{I6az0$g%A3?{`q
zAu!Qn@@!nt`pqVC0$p<Q$jIoMT?fQ}of0-Tw@AsEUNAN`K6E#@!(*7Lr$<RisVONk
z$TeC0%zXVqZlj2Z$h`*-aBFI7aqiqnK4Q&vJOEWj5VN6*d2H&~DMNa05|-!2&R@bG
zb4pgdJAGj8HnDc6@so*Crh%Dut!GrvTbEj{*hX%$YYliz^45H4!CY3@{Rf>zj1V^*
z7C3LU+`B*Jw{snqO!HKbo2W%(s_5cyj8~!l#PpPNyurSOB5Ws=aH|Z32B_KLXD(-e
zGiZJHVOtIe$S$7o1csb%w_4K@i^Iyvp71=eW(TFmKzGTi<V{BLX<;WjZ1b6W7nVKf
zHYuIUY8QT`HHtkbqTyrhkJqI$D#CUbX>w0u<c>&ghqQ1sP+p~6ul|1aK(m2e^^K|R
zwfXVYo@!3)O2?ILbTjlo+OE=P7k$*{zu(w47Pnu_6!XS)&l6y!6t+ktM*n#+CiD(o
zc&<OZj3J@rA^S*7tOT3w?mKO9^xxoX9v2D>ijI9@3BBuAuIzu~+8W6q#U=+`I>ZZq
z{dB+R@gpVm>)P=v?VcWFOZ)f!?qL0Z0HTG?xQnh;uW2y>Zr=u#l-CcI>6N%l&I{Gn
zJ#jlZpIKH0TcK1}-r%oqYQtLe`CPkG7GI<oZbI@F#;u%YLbEY#<4zOf10C3U2~jD9
zJOZ{81{RiG<sDI7oSgLiZmK5${ho%V<}C+BMBP@_G`{|Or!CxvM_-So>r-au3g$;}
z&C*gng@?Qcd||dL0oo`~+Mj#v<i*FB${YOEhnF;Xw#dL2FFCS#&)nY;bfyjnuvy>I
z(`>vLIZ!yZ>?XJ4)cd_DQR;*%GJ3+y=$W+LV_-Wl<Lvd@|DCiVvAk0js%ZK|%N-v*
zPSy?-tC*RgVS<)XHZ&|;%TRZ_GrzyD;m6NUld<ovjw$WQ*<$pR4$zy_7q7_=!goJp
zOOUHPvU<o-)|dWzeT)VcBYB?Ktn<<|LTU*)99y2yo=Wy(nzTL&5@q>kw-OF`jTwYw
zE<=$12^)n*$Evzc{#}mc$%2@gQCdK7m2_hRvas7B3RBRHK}cc`F&tpzT%vfm9DZgk
zfl~4jW2oF%SNcsJ#;o{*q`PkedtG=$uPgprpE<q`3%io(A*4a~4-5$LnAMX+?dJK>
zbeHheluITSCtpOdCXB<*nw<6Vxw!=OTMlfT`BEnt?!bZ(sw>p(ZF}b?)h+EM`L{T&
zwicEa^Q`%2t#FZE^@`Ea+FIJ(vOSmonMMa%<kCe_YPqskuS6BMNnOUB->(=v|CA9*
zw6URUr-^`kxnuu|E5Oqws}*jnYhCQ9e5q*D`Yx{IreQDt7hZ&u_Vo2W320h^9iKL4
zgHu8Ijf^~whsEa6dke6`Yu4(?3C=Lb_ndz&`4_dRHg0}`6zt@?Gv7|n%|UE0LO8d#
zP&*^v#@b(P?Ug5;IwfnxNGSfd1vjgYdVY0%^xl}h=w2R);hblJl6wR_JZDv5k!g4n
ztc2R4=TU@-h>t5>x%blPFJK5<hDz&)T0gpc#fR!sS8cmsx+`?koUi#sJsM->_%9m_
zq=n_ej9dBNB4J4b8e$qQ@S8L9K8aN_Rq2B@AE@19S0xM&&t2JB6?~ieaWUSxy=+N}
zM03tl@+r_dC<Y_e`zJ4lwvZFAPTN>9F5OQ40xeo)L3J^XzMkuyY>yslwquIz=ewl(
z@UAz;KMfI;$GLT@z!$gnHtMFIwtG^>&osF!q%S9njhArDv-@*%<q22Qag-erJumn(
zgcV#Mp!2<d%Udk;!Ow7kU--zoExL16ZklO7?hBN}G&%Fn&wRF=vv=ESZR2X?DceHb
zSq*Pw->EjK3Lj?AcL;D>8O?_i2r~4N1+Vq7<Pk_?&m3`gOPqMcTJ2)*y_=S{xEnsO
zF#U8*^6xs!CQ~A$#2z{?^Tb>_t|iq;-BE6^tW1vP$~T<7(PI&{@wDQSl=#j&aI&@3
zlbz`@BLCxM>!{$<&6<<Xn#8V48&BUemLPR<`5Y8G=yC?aWGhqx<9jhV`_jSN4f*tp
z58YWVVGr*RcqO%-^kJs*h%A1+IFEp^5AU6M*yzJN!=uJ>`>s#Ft_VMMQC@Q`De{%Z
znbj#B+-~yg$do!cRNg<K9zAH1j<uRowL<mbkz{wgr&&tr$@D25&w_u-#6}*`aO%u!
zKg_o0?dLCc#JlBrzq|>9?QSh24WUJbdwio66Eb?UsO{rOscYRRgPn7~&#O=l@T7P}
z7GB`C;8KY`uXgxzG@gz2gl_fx)!;N1B}R!SRaJztsdu}il(!Bw`qGSKTr4L(4YhyH
z=69`>dMf^z(TX;yM=n9{9v=I0h?~iJqlnF2dWGt%+=^?byQEhY_TQ5vlo;9DvgNw2
z$Z_}w^z*y@%C=*hH*RjAUE#GZ`T1?@;NZ|k!uS0BeOVn$rSu?k#3`5A$5Sr5kFT1r
zQ<kthi3(b0v_KVWTL<I3dF*dCUNW*(<l`&bao4}--b!Tf-fGW|sj`B$aKe-_r)d(g
zo0ANS^LnB_Ge%X0(<M7Irb^CG(;a48M*NwFxc<{h5ia&XpRVl74JNcpUO4Il^(EbD
zM&?T-m_3p<jY4@h<x(W}pDoqJ`oFyx*v3;s2F2HC#D~;IMf3UDpOvbOk9Lp1tsx|b
zCD&`0CN?f}heTz9+F|(8!l)my;h1Plcfg_wfqJi`ZC&kl={Qdg{o1uWTnw_aK+I>z
znpO_4UGL{ZB906d1sCq3fzIoW=(uk&WMq_vwz>%oMwC}G-Hu%gT=}QW9;n77%=#dO
zkx#Qu4A^wph&TcQq~Ub+Yj!VKh^xPr;w;xo^#4AbGJ}#p-Qfn9Pq2;c%UFVrKK9$H
z!J(&x`Aaw)Vt#_eJMiEgU0WGuQJc1WY$^dY2rn=1X7Q`Is^KOv8o3dTfjup-m5W_W
zZ+`9`?ax~%|F8t?7!$Ij)@xf<HePp=$1CR2+STp$d|oB!z?Dt!oVb~N2f^3&kjU%(
zIh8->oF1RYWvRIKsxJZ*rS#<-@``c@t=V%ls%xze3bw@4!_7o{3&dh}P7+^myYA*t
zo$)XxlhddiF7AUl$aITnFR<+XBc6l^JqzFD+Vv%|_JwEIXV>TA3LekTwqUmR;V3BD
zOlv<4S$Up#=?OcNq2ScRrCNtrzWBvTzD|7^kH%yjgLcD~Dl#$5-XmMH!RPrW9q+Dn
z7W;Ym{60VjG2zGHFgXjF?75?;#U9(GV%;@YFsn4xtrUHIqtPJtzI+*>%6wCWrh6t;
zY*A6yPMT$N2g}90Q=Enx&~bLTwmLn6Su~e!mo_H%w%*U>vrx3QbIyNo<kp+p!IMw1
zhEnJl=ZsE%`{!~NgIq`Wz77uSH<8^UOP9muna2;ySX>H`xYU7nNBBC|+9Tt_Lb7Ax
z9pz0*%zBEucEY$^i}<J<W4s$mBxPTJAVQvb9Fv!MJ<&&v+wkOD&|%|*2LT~bugDg~
zLS|H9uIuj;p%94mTVxdsNid7$sVWLLxFQnA^OTvk2ud;YBu*S_C^cUqk^z}JYnj0x
zIR<}}j&%jlkSsK$`G;AI^P<TvtIavQfKp1_4xAK66Nr5IOx9$WVX_KEew8Z4{)$`L
zfyt>1uhxHNaErk~sud&VDouRn&dy}Jn&YuwvDI6ih~iPtHT$K+4uzJ7L~Txo_HfDk
zsT7Wu=-lD$rh)Mcqn9^aoab-8;-ZSEG;KWJb!hl&B{*dq<5Kd-+A53jF4p1@2xC$L
zb+R5>ZebC1N2xAS(&%E-^eMh$MH7-6_jU}&Xgc&3PE)hDCpKB+CN>O*bOO=4)T7xP
z)?(}PBowcM)fb)KB|U$h=WwL)*gP$0MR`eA(0fd!yrN+G&+R8%nW8mekSgu}I>f$W
z>VMoqMz=PaK0M7YvsF%7eh{5)YdgwKP%vzJpGjp&CJ;T0JwToq-QPEoeyVU5rD0%s
zh>oyB7d>{yyo-L-+r?|D)56<<*<DWCS998daV}(%+AC;3|1+9iL3k{E913cw{%%ep
zH1y87ocu~Tx%ZAZoQ~SHr=tahOllKUc4pM-<kG2x^7_wX220d}*)wK23a$#4%`Z}3
z&ebQiEz;BQ6_4+bm3<l8?EB||yDNY2Ed4xUR;tPz(Z|}{W`kmJTIRKMv5`=vNsZ%4
zZFAaXJZZf;Tc&DEMxJV?z$ufip?YUJUAo5wcdMVwOGF})I<PtBRPxX(IWpfrAWQPf
z3u#owA-5Hu%=<tZg~~6T1a0q`bR17nfB7`B)6qEdC2vN~dCk9Lsme>-bZEs?F-KB|
z%)g6gbK_?57+#S&z<l-_s;O8g_87o@9c<+2z}#yUC*pisy}zf?1ESxJGBdV>0xK`R
ztB5~&@&xMEU$_O!aA#-LHXak-MlS>`p~z8u{QT9AZY$;8#vtNbx!@2|3M$!2s+||i
ztwFAHABWhDR3{H4`SO%ED+KFj!=`sG6^9J3%+)=mMIKZzPuxC!dIrg8L7pLiE8i$U
z-i-3>z=)?85`?Ih9mD>0=B=4Ug!5%w^Cw>3-dG`b`9=m#35zd%q{bI}WYyK<aL%69
zD;#d&cqOWvLc1s`Zss-p*-AY!S<0r!bDvi(8z-VgW$YX`Rly(^Nh&|n{e!bNF8YPk
z2vY|(9;tpAULm+ZRK!J^Gjp)`Atxc>W+slvqwjmOv$IJ48^n!A95b`C_iolOqhbRG
zij{uS@nPoY<`_BmY7NRdJ0+hOcRIa(Uw6yt3I7t#I`J`~tT-GSdFD(N>S%;aL48ib
zkDc;y%<}9AOSR!I$>1Uw3xR%66~C>rw^qi19P4cRyG_esBJ#W*EjFUv>UgI8%!<jk
z6Z6c6+{s$gw%^CeT`T+aOPe%<&R6SGdhZ?7EaA{LJ#5A(;Z%Kh_340xTq^C-VG~)6
z8V@Fcl}600039XZCQ)6d?+M;aUd@nTQ)9rAqI^l7d5?0?<YZXnUw=%zm=u>q5<ky3
zC%r0`8x3#Z8!AJMY*kd<QwnBJS`!N+|0BeRD5iurB2Rf%`H<a&8s?o}bC#LLT$P&7
zJQyt1@J~iBB|bIFUv~KOtr=bn4j#f*I`2(vF*Px^Jj0nN3Kmq-<vdfq=)8><d1Ua|
z+wB1YVuGx!$p3ue74{SCCtrp7vei`R(&Iacx2B3&R}`bB33;ebT$b$1^(M8kG^jSR
zt312Izc2L9q4|iayt&W)q#3xLH!ixI4n%OA7IPNs*4^E~EL<i&er4@%NRfR#`mucS
z`or0>QSQ${nJjRK`pBAV=NRlYVI!*CSn_e4YS)tI|9uWM0<r%WfqFu4A3lCp5=DGZ
z1<Sxf$JX4E-`V%u^&T%$1FAE1@Wf*G-5rZeW&H2`K_$uSlIu8fHTsm-_}BIQ(G=-@
zDkVmXX)jMK7QO_;^JXX12yo<R{2MJ*-Y{MD*0R<8GxGhLa6JOtr&QNQvfrQILu=Xo
zQFcJ|J}0YiZRT6Qn0=?0k~&c6twH-wsq}#5XV2stJ-kMgl$u-G!f0^^a+#bo6Y705
zQ$4rl>5z$}q*v%Y!g!b?HXl8rMG9JcS)QD{Lhl(yu(=r;5taGna7VAuGEduWj@wDz
ztW+gfmY#cPiv5f7>0%GT1>!JX+8<~J1cJZ*sm!1IL##`YRs84WL^8v_s;fd$|EjJA
zCjG0rO7rUbKQdW-zMlU``8ku}{v+kL&p_~xoa_JhUz~x+KD(cVmDDz9|NNEvJfnwE
zbTcStNlJcNwnOz<Y9<2mk&}_V4)%QaPOor-^PyXKL<BvOfPkQY3qNOO!vW;Vn2-b)
zE_|w+{DK89J5*RiMEa*se_;)bN|Aw+Qvoa|7*x`YEw3p)pr;q|@bDnGFfm=sZ(XKp
zRi+v&JI$v8g6KWy3y3UA3{Kb!UELJWl1>HHbfp`(VL#Z#_%fcsccd8!OZz*%IB!D6
zN8w<)Zk}f3C_+{DaKDbkp#y=eFc6TcFmRXMw*9VO*!OMzMWKFMP5>f-;Hv&4(_e`F
zGJecKhjO`1CaqEX=O^gi%#eB?graFn_yfm<#P_twy7wdlTOIu7mbVcc-@j<pg@%XU
zMUCeD`0+A(8j*1ZD=fY@<|Glk-7Ugm)S2+?3O&w*1Q*diTEG8m`+6kW!=4L@FC5Q|
zjkDV0c$tvKg9WM2U%YrAey+xNaB%Q&qmhOV2^#GRA$P^@i=Rul`>W6(aC2)*kJ6jl
zW<k@I3c*q1bD7@ec;6B!-IFFpe)XymjN<g~|6%+h?Emo}_y4n}$h!BPLDVoV_2>4e
zm{-cpMS=`nhFHwwzR(K-V%AeX@Ud#|NF<Wma_aAE?1l-~%`2ozN38s`wY75s@Di?l
zQ)LEKw5fuIG1~z(5M_TuPLCrg6PcEol0xfpFdH$1vd=FlEF`~tIT;l1uybkLn-SDU
zdS)LV9MiRhU94<$RV3idgqvtjZ+g6I*4wQ)@-jXS4!Mnu4d1}PfvI{w^1o_S?$yw!
zqEDVD_UEc7SYASKc(u%S_ek*PhK7dT9!O71V}wtD3Vl}{34hMBcjuKt|8nP_l)*<-
z@BO?JxUf5HUdnym{hBW5rAPC-RpH>^Y|n&9pbwYw6)Y^ifPvd=FGxL5DQQK2Kx82`
zH4{!x*4!c-k6^0VLC^iFcOr-a%IiEr$SZ%dNR*bA1{2RRh<A|((KHzW0r&QJ{(QZb
z@OylG>TA`QDrgVk=6Lg~6a5Wg{G@*ZcVh3&V|9Wv#{(e0#_o3NjO_k+m5q#o;*aYj
z)&O9G=vDPWn>m*wg*)xNTeqGC+YFnQHceV6UOz8Ttn^jei`Hm!rU`4xSmr_LjqcCQ
zX4PrPa5>ryPJzj0(!p83T!cN)mO`n>Feru4#K`Igv3mhLih4f^kxS#}agIM2MO$&l
z2B(BSPzmQsGw1-BrJS4`;}+Z+x8LY`4DOe!R45p1)AFEH@$-2Bu~LH$>L;<{uU@^%
zga;$vct%o;s7`Wkb(9W?cae(0Qq;<-fHw2R`-^?ygzlIrlo#T7EK4qX`(Qqv37*If
zx(};ZXuH+MnGJ7xL{x`%Kx9b?3VMS{15>{KxOc45so1dVw&{44nwFLpdZet-o9L=E
zx{Die+!9PDLW{H?ci~*Sir{C{sy#<XPrtNMI^IvnQjzfMm*U`x9o(GD>B0O^l?zI_
z*mzbg>kJ|*Sj)=S0pW^X%CGLn+ADhH##6gq8qB^>f6v+DpP=-$vp*fk#LcaoEa=}|
zWYi-}Wj~;1Z*;uB(Rcl#fd7;YH%I}(S@S|PE-Bo$XvNPL1Z<x=?dHt2MkRwXS=P@T
zxYcq%2oA%}n{X3;-DlS6sX!&m9LOTNwE85s+<q<3aR=S>kO*#&4kQ<bii{3dQES_C
zG3ICzxBYj0QfFjCDk^x8-BD}?>>yxHOv9(TH)0(Kx{Fv$h06%{q)TAUAFDiQyE4=_
z^#hSpVAOLLDLs&<cF+C{e^_*M7LD8S*Izq;9KuMXl)QX6n1#u+M~G8zPYU;OMpe!w
zZS1TL<j2G8q{4V)uXCEoIy!RGE8`r$oRoSDP|NGI+xw1$Z=v!57nfq`m}55h>&|yP
z8bg3i$GuVeTwj73ZSclmW~wfd8cdFIG4{?BL`6TUJFRdOzZsN1bLI@#mGJA=o#U8O
zd()9`Qfvn8vXzcIIRS(T;nTDx2uaU0u^30fiBXij3fPsc&xa4|>eSpehCVunduuF6
zz_w4S%nHRbIFf}apfvaEtBw^H)n@_x^%rGjWzl>t<+Mg%`^Lj1xqr;Rwzk%$@rM0^
z2OhPgnOPn@868Vy#Hl;Lps<v~rBt{9#n0yiY|HjdLIF-$EJYO+BloN(@dpw65pvl;
z3Bk;252M{_u$QLIM26h;_4Q@66piO|p;hBr!%OGGk?aHdkw6B9{)x6q8h1Ykcshqs
z_sd`#h+vl*xnkRIPoLSRC>_oP*GC4Qzer_hc?}`YaC|;ci!wFfUPasVhOEe4#Hl~_
zu0F-Z<+a6diUbdpn3}-*`Z6te+iG#<=K<2_k;Z*lVK_Kkv1caCOU%dfJUwx+SMtBc
zQNVo>K}Z%Ce}{vkxR?I|RvR<8x8m*ABR@?<1`$xDV`*?FtYw01CnF0>hm#A;$78dR
z4Ei{S=fO6kwgZ_UTPk0FPSc~x<%p4oM|Ei+f8h9FD=Z}B&SGsBx5+?WUyf2CZRSt>
zUx@uc3H}q=V0P^v7yLpHUxaOJij<2CZy_BI7t@Q~Fs@~zW=IM?31@)zXkMohT6HFb
zEJ>X%D<#5pVWiCZ8mF0Rln#tAJ3_-XDqr)g>%u};ax~9dtDsp*X1j5hy$uiI^%qyK
zUWHhZLyN`BTeogOc-0I8A=AaLjnp!Nf`XYkhr27;u>NU%j>Vl(2<up>>FFjQ-8fY7
zw(ueqSHj4MD&VW+4Q?Ar%;|~kO7Xy^KGC`k<ci8w&Sh1HXr>!atX>8M1%W@rcoSbw
z({GDbx<TTO8_ukjrBtY&DG|@-h|WyU2{>D$%^V?ujekSMCSk$Bx6^wvBO@tomw!o>
zj@rd2mztT{zC^fs{*U--^8bK|N{YWHp$V^>1m@KlaN}uC!1Nx7#-dMLz+04VU=MN6
zxPs-)38E;wmF^an>YQ)ixcc%nG+=tM5%+L=F}ab(O%?nYYw)5yMcl^4Y+^DncIJre
z5o=w2eMgZd@B%q)?r|*YuT>2e8g$S#Y8)S}H_qRI3mnT}xdK9<Ei@`Blk|?D;cRnQ
zy_txFgo2}Ed4Ik}J|MM1aFOPB9FTD>FW2cO@WTB%yi&r2Y?D^}bwwRpN0y|*GINHz
z{QMfcj@zB9qo~|~--moNKot*uF?X^zz88)H*64V$Gr-}zZ^2<Uq6ql%NK8gNU8XUR
zW&lQC^k(o8E*_rB=ileIACvs~C@8sZs*8(@%Qptx@Eb9s^ady;r>4$>Xoqfpx|Xy0
z0d4?+ungva3K$zMF76-CciLO6Zyxov;Ep|9NDkc2);JLcx08PT`UBRCv5*P$N6!pp
zyClDcU)tT>O^Za?ul3I~h1j-n3^3fkFA6}{J$|~&sb^rI@kj)`Q9eg85v+H~`?{GV
zIMDOw&wI;lO~fJ}KEuby=PO+`(OVlYDq`mayf+>#FUI~JFuoEN)j=J2L~oW=Oa|U~
z?TJ-NQgX6fshJw_4Q^R@a$`H*=9N40PpB^XQ$6WMjc!ag&RTHwR=GH>*@71B-W=2M
z+tspp=BUJ^Ky0d*vB-rrTQaS}lfAX-XV0F6Jzgu&YhkyVdHn>yYhZ1#P{?U_xyxy-
zGX2&IZ-zhKSYJ}T|F;SR1|ZelplvLxD+&U+mo8u4sU>7Ftn(&*MNsQEJ3pTdRS~)|
zKr84kEtRmIYnAGV<K3yJ&|tWIJBf#+Oj&O3ltWCWC4yyO;`=+lkAq)z-rFtzN>tEr
zksEhC$_0N@Fph&5c!E}_?%tej@l&m~S(JXDRxS(KKp%MRj8z%@m6eo~)K%3Ae-5`q
z9hm;pHiG+}+N<A+Q4@s!3~;bFRfGcBeYmq^yT3k3(DB5eor?JCqZGx1eQS{j=05i`
zXMM@v@^k7meAJ#@t;g~;V037RGqcSq46LlMcCwSbA+#yKdm8)H-t4RPwl>%s{E$h$
z`3e{x44X-QRs9VB#kEJeAFQmcTUw>n$z6zGn%(mgunB2>1*8jOExF3Y-E18^$-pL6
z3k2>Xm6fCKibmc<W+~*y&9%p7Ckq8O&ssH2Z!kc5)YR039<Q#ZH3Dn{oRx~Q7LR)L
z$_}LlzSm!7J&(<~fDM>epY)?r{szu_mV`$VP4&W?sH^&PhXo6AfO%tgUb3;Vg+)eY
z#F&*MyS;8(0r65e{+2Bww>&-{@1{;lN(x+hZ&f3rCDnAe<lf`Q^1(Jh?oa0VG50Kw
z{{H(M+_iAGSID;7u~TzL#{jr;I+i+RE9Cb#_+R%kb?cK;fSSJx=45YdsS0!I6qTkl
zW6&IWPi~hI>s9eu;Krd7P4w~XbmT`yfH+UgM!2?sj<ff4cQZff*x#5IL+V4e_jVVA
z%Nik$SWnb=q|p{avQcj)Aui4wtOJfq*B%ldZ*GU~ZF;)8K3NxmNpK5Kw}%a*v|B)`
z87*6Zy;bfN0_(LJ?Ks$)ue9db{iUN&;Uc%J@xD$gEj>Mgs{{O^a+=y@yX#UT*=lLQ
zuHW#=W02>rm=+#t1bEyWtAdy(O`{^mgT<s81&ec7?CBadug;yB(Iuwh`toxJ?xhF9
zEW6#ZA~q6RT3Q0bx@UP2W)PS#GFaX1WJh|SGK~2?@y3)Ni*WfDvo}ig!cG$&c*48~
zDwCwiUR9v(D?NYD52XHd0(tKumCLO3%H}eR<%-?XMWd3k^UAY+3_~K-i?TOIc9)nA
z`;NxN#;))(+HWs(0o#$ud<$Vq&e+>G>!%KOlAZB?BV8Y)TIs-H?y<39apE^p+AK(V
zd3n8U%uy*7|N8YSU3dZvU@LF$5YitkMnM=DiwFJ!um?%S&q)ggX6AO(3gjcgZ2+;~
zMCO=_fW=5R#`3(C#-=dXpn2tiG6OB`kC^&Ni-oGA)esx4%Kqcy<40>f1TY3Sbu>m5
zsA%OL##$(ldIvV>*u9C``q9w9$j)BW$~r3-)QG+7{a~m&fR{wG0QZ0gCpTW@vV$5w
z5l=H>q@#NUwp}vjw8o3<29&VfoI$=46m%>_edGC4F^F)9*CWpY%&?6b9(wb**tky^
zq|Aq^-Q0Y1)Ljqs)<|<B>cF2SEmU{+(UNjUoPa*9rtjXpqeGgNqJEiDP&5MVVzH*K
zd)S-!0?%};qO8?&R_Z_Zb}0bsx(cliiRv2~j6o9nutnqa_TqD?UC6{Kt}4jQMrl!!
zUt<GYW@5BAVPx7fHxlj1DLE?8#@3Y*(h{+sCx7I6?65WqrIoawrmU>2V=WpkJ7J<d
z70dCze*KbO+X7S8)8BO987ei8r_FrjcctEpNmYdt2se%My2nLoo{ZL3QSgbkJk;ji
zkgsX<^L-JPTN(ESjpAMPs@^qZu-)p265SK*gt|S$7QGl#5)x{h1{ZCvl;s-rq>Z%@
zQZVNf9UVguYp!sCr$5%>Ws-XZ@|Vy@Y;1JMt@(}th_3l_Z81#nARfo${tSwj`)VVg
zmz8y*Lh65=b5~vci{`T^oxQbjjk%^VX8CXZ58-m|N!Ti!fL8^?r?HmNa$6=V6-b%|
zKT-lM1Ux8)#9p|p16Y+2HZdHNmy}F^yx9?RG*3TV?hrdvYpW<?HS?3S@?b88TvWL@
z%NQg7(!t>yBuUK3!}U6n9e{~YuJ-ATg~|6c%+`5+5;b;HKYW@Au?v>Hw}11R$T(gn
zv_I!qXSwZ?&GIkS*||B#0eOfoQBhIC!opbN;V|jHzh(!u9hS!fCi#_fl~j;n1En#7
zxYXRzvZcOehq7G-nNY~bz`Y{ma<Fp>)3!~025pDh0U_*6fEr<O@o-vV%7+QRe5#=2
zF;VFZgEWD0L#=FsuLN=*6M!&CMOMgTi*nepvgyBh8N=2O{aHD=qte^zE<18-vk+g5
zzkA}2jfXT&cR`xVVKE_)t6o{&`Zg&ksUjj>JT?oPCW&dnAZy-Z6L5F!$L3=>kVEbH
z`uWK~NFcv-X-ciCM!)dh?c2|>gbAW#<uK3w#Jlp{Arrs0zKXZjF6-a%#xSn)vG%<s
zrYb;-U_}k;74Gf`aHK47Ovkw>11N8FJ=zkBW=9?z9AGn7AV#aP$ujBUhfo`#Lqf#C
z>^JiJnl4#zRR{&d@wq4i;t>!N@9>o@EEtM?=8^*Bh-NpEeru;e3NN9S7a@10`e+q3
z_Gn+4-z!TSBj4M`Ii7QRawHNA**%ATE9Kf+lUZpt0K9UczW>zJRBxey1Twfy{@BzA
zvN3MEWd<xJrowdSQhM9d@Bkl@JLH1tKxhd3Hu2>sgSf8NC^nJGlhdQM0d!wB)Vq}K
z?Qt4)(^hV^KTRD|PZz%dRe@}<KDx2feiie#es#g;@bFBAOH2=ES=^K+a}=_^5LNYd
z)H6C^<U_S6mA`7g`Z~+X+z+%VSGBC$N!Q`!%a=aW%zV!)Bq#YSP@`d&u_0|;>x?{>
zS93@}S`CJRt!QCWTP(l!IE!a<di)lM7&iTl<+klkc=nFnVO_H?TVXBO`Xc2+fyKo|
z?5Ix<I^2wbJaAh~xO;kf_9i?#YrD|-3=4zRt|vfRA|A~@B6=agv+*}J=Ye7Xy0C+<
z2mA+1Xn;s7lkjeT!UhBME@1djxotL}atgeU5l9O>2$+)yZxWh3{L6QNcRfD>qBrGY
z5Y{|W9d--gg-Ic=H=4tQU8jKvTiJo9j~*_~2b@dpN)~D{D|M3iKym%L$`lW_3INZJ
z14<ECE&x@V?9fy;7{B{a?L$J67<iTOz2wc{6tB;4&>f-PA|n~U%ZW*LYh$@kzeNTE
z%T9Z``mJv2CH6CGT0Yp&DvN?d7DVO8j>dr`7cgVl(nQ_wqQ2y!e7<@vkeK{KaQYb=
zZoTL<(H>ct>r0m|^?ub6qPu%nI2ap+GBOzaE<*r8RQYgVmjfgh3knLhmkT@4g9h;r
ztUr&M4Qwy=WZxr)w@4oSj^&8^q)>2#s!s;Ezq}BZ3p-yC!lrEr-B>Oi!)e|R)w{G}
zlL2FhVBBt}71%hOE<pp-Ul<Z;z_3qHqJhuF>qn*;Dc|^=q`-eIW)+jLKgRy~UzdUZ
dudQDkx(Rf%N@_*)rep69`AYm{-g8aQ{{m)^Z(RTY

literal 123684
zcmc$`Wmr^S_%1x^4^To3LO?*JTS__<0S9RqI#jwrx-kGHrKOS1LAn`2x^twvyPJ8}
zDE{AbKE5B`bGV#q_Jw<9ueH~+p67n<``&z&mzBiBCdGz8Ab8MMFBKsW3?~TW#@m~i
z;5XXW2Xw$2=DXLDFTp$bIT5S!1aG&jU#Z(dAmqO2*R|k<9RmpD5d`}3xw2E-`h=rX
z*uiD<jzEe@?B;~x;pFQ`FE4ZXJJw8$t2SP9<qFEmzvh~|@WsWy&ApKKX{RVWnsg1n
z&SJ4*IY5mNxf6u_BDPp&pjH=Y4{tO;Ov!B(%Y`j=571#|46G3R83qJWxAs7Y{OTAo
zO^S^kC;Dewk)ZnT5o9`z-s#%k<GR-h+t>dddPOBqV_coVCSH$ob*SU_k@)JcngE0C
z>d=Yd|H~J1xbl0I+3?55u<Q9`=p{Yii<c!NAZXR+XfvMY4)*J`EbExDeihnjScHrD
z$@hMJ7&bPx{?iBg9MOX07!-D!>@PwCB&b@%FqW)&^~=Vb1H8Zq^LG3+P+?(V3fUWS
zMI+{dQFSLRH=zc*PB(FJmCGzA1j%oR-eTwNxiI|r0$kg4qOv4I@9*#lq)JSOxa?Pz
z<ys)}FL*y8#pMKdKEn~0E^;lR?AE*wrDSAaZ{8%ixw{LIV^M@hM`u-63qlpGt)Cjr
zD_gIutb}$`;AEDYKN~yIt4AVLKYaK=A^V?Pdq;<op<#MMLqq#`v^u@EiHXTl&!GR8
zFWc6nP8w2De&-h#3tL-TEA-n*>+9wV^YgD1qg^deHmkNwFHR2<g03a~r?6ewYW^%L
zOTfImd&au;T|p}n8MS4CVN^8IjjBk?&u7E1!f@QY^g8qQ>oVqQe^>D8Uhw^I)C)^X
z9pKE5Wgm;wn3r&LHi_$OZ@n^NJ33rC^DeX#wEoJFbUw$UbhUv0H^brow>bMh{Kf6-
zE0D?d6+7_Hg&)(ezG91NO6i=W<wo}mZ|3jI%Znwa%-_fH*N7VnAYacg&JSw6=B|mF
zq)gi@sHjBBCJJw?JYbEj8dzPqX%GMl6;lk#_*9n?!0*cZO0k|2CDvrW`8X7O?u0#T
z?wShJIpOxT;Ju3hM1`&wiW1dRAfIqx{!M$iBsw~}y|a_Ko=ZF|HkPSsVsqsIMCuMk
zzW@38IROE|Jh)1cnsh>teEK{H67y#r#`D=UgzC?qe|UJ?m%6Q`MH2H<edE_@C<cZR
zjz>|nZ@XV7n3Z7>^#9#<%JTA${F`P$<UWn<a=2U<48r<t{w!LWHs6!=ZPCI?YpsP8
zR7&1MC+J0k9)WHiyx|FxGrvak9Rj<RrCw)e=k}hSDC(25v#6{o5;r%u>dWK*tW6A-
zC|WwccNw0Xp6YnKC4u=+m<eu<vn>hQ_%_b)>TZ(;N<>Rn-%$*S7wGr^0)C^%sA`+{
zgDu<ncWGLk|8l269_U8hZo9Z{{w!8iL}GoaLPtc#<uzF(Dm`vJ07qD-$m6yB2G2_&
ztVuRX|07{6hp7CmGz7w->cXvFbNKW+MVZ%bj{xO|Kik5q<9!k5!ALB=NgrDnHI%j}
zpOJh0M^a`Chh}cOa9C#QjU2l-NKEqICC@o&gZonQ-O?mj1YCLfAk$k{DK%46F*L|*
zFvZ=@X^ZFtLo&x15t2RN@ujo&ny2=%9plM1zs0e0>b=sg;n!kSX3Kkt*_tR<!TS#(
z)(Tpm|KwjU>?a~@k|^h7#O<w#<lipw+}48*+~ub~a5}W191jT8Yf4;%4`WBC*kxs*
ziPfiU7HT`1SxcCZbIz-z%=iULDU@>|fMlv1j#xc8et3c*L1e01(~X}wH=*0MVz)YR
zw~SGUJL_}UbqKcq-vsZ4J^icTB1Xs{*UjyQ{Xs9K0SEC$V4Z16mA1XWM8<k2=c!GD
zrxZO6Bq{UXWFd>v<aAXvqJb{LkIjr}q6~}6HwmLzE@K2XI2V={2y9<KyqNyy<YDIM
z$I$GFw_iFqb06FIS&eHtSB#7n8a`*IxdAEvbTy3ktO88hEy3#w^2#h2;=|TvDruM#
zG*DN^T$UhX2!!XVV99r<(b7qUUT?@L7g`$+u;si4Nw}(7qI^p~$7rCKDQy>TeTrZo
z2WG52i6C_rR|PQK_HBqwXzKc)o@GxvRNyiy8UpFLN*zgYI5)9j`sT2Bz5>k=Pl95&
z)Od|rsB(>E(Ysfe5NqPA3b+tg|3-dI_JSIUrQ$ia8+X30Og_Q!nzkr7rff}JYX<_c
zO!`|*lddoGZF`X5Vp^PR;4K(sRB-8a<Xg2yhQV3gu{ucaXm#GY27&+2$(cGCS^P&N
zxz(Kmrx$zm@A=ooM-zX#2bFd?O9~T-Ot_&sXzt&G$m?HCM)}T>@WF@K?7`ta7KCxz
z@tE?rMD0v>NuHj;<N}`76+7eKIEf65rKY#&Fdzg(SHmZK`NsNtf~RNX6TO_(=}}59
zy+K0$!W|>N_#l6Hj}%|_!u_p*^Wgk27B;0ae6l=N_bvK`L5SCDc31wt1%zH`XJ~aL
z9%^>5xmRYoHtMah-6pCoHcK#7UUogjwAH+TajElG{3t>(<1qhQoUHl{F|PHKp1~h2
zG6_NWkYlocq5A#Xws8%IL+vn2e45GqsliPareepF&p5;062Deg%{B*<9yby`c)0l?
z72?(L<KI2ycz|m<k(H?7xSGMqkWp;bft^N8uBu~M+VJEZp>9or=@8-tCY7q=e_n+F
z|H7M<T2RBH3U}KM7vhEUFDR!=bJdg5Y-?ftUmM$dx)hA{<4k$X(<XUvBe);^9smm&
zCHfZ&iDD?rf6MQgbbIOV8BMS0F^NH`TQMyub9v0yq?(`aiQrHuAuBx;m4ehQ$o~aZ
z^7ZSJl~QbYYw+qgW~idE){Q2f&Ak65pIqeFxh=Vo3zRF#O%17NpW+T9BxQE3-$UxA
z{)PF6-0wW%TY{ecakqV+<di9{kxeO>*K`wW$3{gYvmk2W8+((;SzW|%fmqf4qH2!*
zd>|5%)HzK-f?g;u?Ag{!3(~y?i}(Bwv&UA1wfEL{J((S8+ISvm<ypCD)+EN!O?g>w
zj1{PC)mUpX&|*TgZvTstUzxtH9hPkJfx|pGjpT{##-#+Ny5=>ktB2cmBr?Jp&yHLz
z)Qgh-_bl?JFZU~deE1{Ow3_{c-3tAu1SX!+@j<2zU3;Oa!IBe7_wHVUu>McLe}A?k
z163?ClW1+>=}jX+XA*yIX((OLvVKGTYycL-`tJYixyG5^okmycPtUnW$JcisQX(F~
zEGFebx#nWon+WkC)0kI5)m{;H>ukaLHzBE}uE+B}%PzmH0ca>l-9EElz$U)`?+eHK
zZU|6AhkrVi8L5jgK|{M!X)A<qAk)G{88`kG1kXh~UOQSAC?*NPLsEuM1<OmO>TVY_
z(A!mA8-ll8YpL7JcWqoja&iBcT-e*_f=%o9*-9K%q{&B%)E!(uP`BxG<Ve&Ip(D0Z
z*D=2#g^|!yr({%y8XDc^JKh~>fBJ=*fqJ~=wGag2hjleNgU>7?s#5`ES-Q1n@}=j3
z&`_&4pI<aEcC6Sj3LJbt+)&GL7Z^5O{fr#iDs8drxM1s8*Q=Nx|JllWqUILieEPFr
zZd1WpT04r9);y63YC2r%+rbHeH2Gd_2#!bGd<s^A-14rad68``a)GWwX``x>SwD}P
zL+h;_LI$$J^<qn`ny5E4xKxzrs^R2hQKrY5N^GtY7G0+-C9WM(8|tUt8WY>;uStp&
zgXrI$X@NX<P;>G&$`DM1*s2iSD?z+`7xaym7kphzdrF7=vil^Eh*C*(?7fYMJ<lJH
z0P8sei^=Kg(&IATXBAgA%ZiVUGyx?vJSx_`t;eS4{uY$OxaA{$M2K5&THIPj;J3zv
z{K~+wcG5fR{kRa(ovY+s*Vr6Xt1EeErKEL{Gqe;<gMBmq{fN+qkf9;RV?}Y~X3~wS
zXUhXTKHBfyCz>r_Wj2~478M%DZK5N)9RA@p$XT0F#43X0L7>a7$P}Ro8zgD|Y9~)W
zuPV*$Ev%SoBh`SaitvilQX*b#ackzT9`<sbV2GQ6pdUrpUeF*B91aP&A0A8bRjMOu
zhQ!sgT3AugI9ZE}4Vk?2t&!^|lXO$6%338ajQ)d!GhYR)pfOlR8Ym0gRlN4YUGd2+
zVuHNI3&~4^VG(!<=4#$ovw17L{*}hHg8~1*V&`Gw1w(4>pTjfs@(qU-F$XTLd9kOt
z)LE!$I{U7!U;;iWF!F6$vp+HH^vcp9{aIcN6D8`+Q0=a4)Y9OFb?~r^vj&;UIjyD#
z+>IIKMr7?)Qh9n_rsr~5E6ZPORP^lpJo%Y4Wmc<c9svPN%+Mb;VkAp{GEKT*koGFn
zjoHIx*?VVo-0^1z$u$VgRqcl0j}%#yV4I(056^3mbP+`@eP_P(%b9{nmYpYqV9k2I
z7Zjysg?Dq<b;`VdacFP9Ng$YEX4{g6QX^+x%nVh?`jl6T+;u$FHSto^kZ1QAMCxBM
zqIl2G@H~{@q4wB@`g70DS5SVPWudOo!(Lyd(hddrBnz_0<>h+v_HMzH@@)ITBy~yg
zmn~KCI1u^5t6abRZG2rC8al$(yP+=Z@jNsYAjBtI+%a9Iqv^3Rr7MdLw;`e{vogbf
zZrk=ZP?(^Ko6%utlprEN#U<Fhp$>s;J^xz{ynL0jHB>`c%MbIILh<1t)J?FwxkF}%
z*U?pKs}su}vP}N6+%CmmE}_mKsg)4sVVbBIWT|4lpt1B?4<g!hRav}zAK+rzZwjb|
z7Mm>6M=hDh?D$b<*-q7iD%TW>2bm`SH-n@Pj>0TvZ@wHdpDDJKUTVx=NIsK6Uc$l%
z3GV&4EE7?<4#0!~$yElSAb9W)j+H>p#22rzJCes=;0pb5L@~v#iwG8R#oYcx)!YW~
zhax-p$iL2>VmRx_qH;x-jZuV7+qT&bxu~83`xbQ&yk_AXJ<2<0(B+OOGk1=zbpLH!
zvgII#t{35ihH_Yc?ir*%4?jrJ$*#H?aq<%#K@QBvuZHp;q?Vs*8{p+Iw`Jvq>q(85
z$onGYV3vl3ox(&y#|=*QYo28pb2lJ@pZ{g1WMmT7;ISX;so^?~M!~e?hK8VQ6;W{2
zn|J`bS`x~ZOMm1u!lf+msWHIi|E1o?Xh^cclR*#i+FImA@~0{uMsNrqGZ9IK#(`!w
zti_D*9>vB=Cd7-w>)*BLnK@*IFwZ*%*dY*uhgY|b-m#FgXZdBrwRk8Wji*`ClVm0)
zj?!#EPh_NA6D|1HXfAxLsZ00fMD&MF(VDfS8-OC~dE%t3#a*(&A$nou`9h%pCKP@T
z<ie}(-F6Xp#yZrJKbiB3%9RG{@m#?#cT(!Ip1p9vkp5y&n1*~JSAEJOiBGA;q>fx}
zEfPy7vd<We2(0n{eTC^>hUX6(l(Rf2Ewih-AW`Tjf&$BNo>$!^C6um>BhnXyL22Ah
zoKj-7#4ohhOmL!pLn!<qWZY=q?(aH5RP#>UeL%L9lZC3@rpTF;W0gmPitwsXf;%ee
zs#>T@xT{Ini+5`#bABT(p7Z3yTJ6^$)wlkQK`KLzX8D2`6BMKD0mNm_skV=lx6}QS
z>F<@5iz5rze%V<={+`1DR+CKL#SnRps{r;=J@q(eYH#b7Z?p`~mRO7RD@O7>eWC3u
z6ITCbFq}<-iBqWLv*LBpwtsh6hmKQ79LGrdTd?QPm?N^4{(x=9F6FAmI|c@+CNrJ4
zn&R$2r2du6ZGO=B(Sm+tNCP(<RLc(ARUP|$YLl)|k|shth!o+!;q#+Q9(5DX>dJ6j
zCaJ~39C519m<ni&h$Oin^suXds#{B?CXb@VkWjWY3#vC(6jd`A+1(3|2vK?12T4-?
z$DHLW$$w*ma>TG$_VDwh)I~0VEI@DZZ%>5lw%pzvjG-AQv1q)P7AGZdDI58sA%##k
zFG0h;T7qTlC?t`)cYuAyyWz<kHFU(JWg;OW-a$(u1zP`B!LsMNSHHyHj6?5m%h$oA
zU$V!vD|Cg}ZC$_N+Qvwwl)s^jviz&hmgPsz2Q@p7qus|v?%yYr9MTA+fzll@!-f`-
z6@p8hUnknTUMYg#8gRw<F;PImuc$%M0yK<s-*2C2@97BPE)HFFs8-#PIi=^VECo#~
zNPSuJpUmuxCkV(#pUT{C7W@Ij$uI#h;LlDv-oBRyuT)7yTx8yc$0jpjUbHN{=6<NP
z7f+pKpW)F&f)A0x`+L`uelve(e@+=(Dar9*uzKnh_X?+$q5lD!*~cel^z?;3rp2|O
zSe7myv2FHu%XI9IBluVI@>Lj~@{WtG%*ecB9Fi}3Hc{f-<pxkk#j37QZY$?|&ae`T
z$p<8Xw4ljPJr=tz$}oMEznk9m48u+lD8Wac`cf?$)3ciF^R?tRFSXSc(m{wCWRIr?
z-wR$J66l#^&S+GS`Z}|LDRN#(ZL$&ELx1o)^2vrhFZTY~aed)JYR;LLD=f=&BJ|5m
z2u;x6#4-IWqw$B_)B{@A4>A^(6e-y(m;^5QBRrzo6vsM0Z19S~Kw09?v~SUhJ1bVb
zWdfbsYsS`&{_&%;<^7oh?@`rniFs2k_$m#DeA!wlA2wa~^X|!FLujtx*>vg9>-4p{
z%|O4*VMV-3VXgTKP_#@P$0S2jty%)Z2UeIjztCq`ULs$i2@Vi92OVVOB__>GAeCsj
z6%1tQX?hOiB1b`gw?#;zBu>@b`HlN&s-4&kQTAEVzg%z`OX5)tb(Tq@k-^h)2UZ2G
zPnKwaMt2+D_d%OXIj5emiRA7=c)jspBNal~?cfxhkkMtpB6crmmh4DT|6q6*CrIJB
zrBA=y#y~+Hf)OM!sA5WF{Qe-B2tBBojb#wXxY*waw)P)b8ElTQ)#uVg_lcTC+0PKN
zP%NbJTst2@H9gm-`lRQMc&cZXJj+!*I>YT4O7_MVBd81QE*BQN3n3~y1LT^)(+7gz
zi2f&|*0J!Zue7(nQhH@BbaISQ!OfKW{ywxn>zCLjmYDZ_&5B40vV#(zpkw8+VHU;N
zK6k0td<>pu-v9>BQt0URgDX!AuwP43yV}g-p*D-f)Sx+P*if(i5MpGMPE#d>Ll~8x
z2vY@R4b+Gv<$u7c?7#oKp`p7)N?^lAMLdCG<KA`ztpJ4Z+TVb$W4ZSH$8H%W1Jy~=
z`U#(9my|%QSSSIZW(jZdHHiG?ze~P*`d~2E!Vy4Gso(59dFr5d52xlf8%z1$PaCd1
z3k*%oE1C+8lZ`Tr8%4aJJ7R^sv@;ko8Bgb(1K()*?;FvlqGL*SLla)ZL`&q15E7Bh
zsdauCYPWa6)@IB_=m=zscACHF(`!tcwrJCHQI8}*BJcS08qKDL@^VekM-aZ&WwMl$
zNtH6AS!RgGG{giDulGKG!(W~`rg$^^ac~Wl<n!9aTBMQyksG~l%5<nJM;N|Ix2ur7
z^~Mo)uphF>N-nHTpEd0l1mYw@!Tm>fXi7&}1b4jK8CbFs$sU@z)T|;3gk7+T!78A-
z7!oQ`ea4Y2JYpa)mU}B1MyuRDK6Hr99x};ZYUz|$(~rxr0+NM}_tijhU8bHYmyl}x
z$e;9-LRZ(-Es6o~3d6;Wc01Sa>ES~RBYUzh#eB?qzZoqoQBKzng0ay3oxEs4&Om0M
zglFl?s<PbP)g}?Mu~Ql;8DV(by0mw*ib9&H{m^Q**$})#)cdofD?cl^@)cnw{{+B+
zRi3q^(=L7|ejpSr5M6aVgmC~CT-U74&?6wr7|0rrV+M+coa7GwZC~vZXTB)in-F?V
z-cNsY6~^yqX3hPOM`P~{cVntRcEzMEG_`KpdMJ15MPB}DhOE1#s<m;dZy3=eNQ=d8
zllY^>rf^vricl62IbkAQ@n#j+o`cZPkF!A2-^pOs0HQ=PV3fM$Y4O6$BBhZ*fv+Q6
zPCo<wAop6mU%n0?XDe0W&&J{3k7E*oup!4-f9sw6b6)b>UFVxOWKP_QIR=!dp%N+T
z*JfbUB;oF~3Yx>ry^-8F`(5QBq;^P7Ql}+Oj%as5vh#aE4}Zhirktf?k)0x_Ddyp)
z7``eBmWw7d6eDPqb1;UVm<i3vvF{1n%5jNLr;yfi$kpJL(RJT0mZg38jhX~k0uZGg
ze;K2ViCPKQviv^$ImxV+x-+}8pGbgW7iR$Zgh=_r{&-<O0tDd}f>+JVGm(J=7ohB)
zt9I3z?Dp``KD*L7kha5%J<|^J&NW=4+_O+Y?$6O<AFz^D0A<qGs@_|LB`*C)w^GTO
zPaO?ZnuQXTBNi_R@2H%d$E+96@fVu-l=ir*Y9d+sw;ylpZ>iE|yC0DWeIGA+c)N<$
zomXps4#5P?9{gFg>PV6<73cz}i|zF%8A@5VH0|GP9F0~?Q*~YLpo*Oj+=Nh*xPTto
zs^13_nERV{<-134%w6W%u4$Z|Nn?l)_1jbNWDX!|RkSYPOC8!P>m$`*GK#(eA_H{_
zV+2G~YUrUb1@E)kA`P?=>%=wO=a>r%`dQAXD6Zw9_-xOEp9lcm^zrgiPqirq>r@>x
zo4gc2J?|UqnM?&h>`?xF&)M!o3%C1jRiT>L&_xKz>cKBTm>J);=K>t80Cvu}vMR2d
z3{Q2fE-XeT<yVU9!MS^UgxhZBQo{`h>s6<+FrJF<YqnM?4$gmhp`(@pz!+N<9d&6f
z+}YAQ*-Sg@LST3Wbl4Fu`WyIc%n_61HaT&!8T=ChP?l2!bIQp3HX(_y;27PrWd12s
zAoWTOG1l5~h)~+yGHLbw4(f~D3&~x)fZh@+m29aZkEn?`VuCbFd^y$l+hTUK2jP=W
zrb+9E9~EEWK<IJ*?qK?sd$+fSbGL-U0Yt165frPvHsd|H1|S{0F}+QMR|3b#BMmg4
zk#G7NV$a)d0x}{Kx|IQ2YZve<E;EeP9NLe$+uqUEkE9gRd~vHI9c)*b(`gcD{}BFV
z(zo9Yx~BDSxUw$D6B)B5$Ton&0N^Nn)^~{4lfNL<>%BSX-T~b-RNPa99gwRte{uP^
zz;qI20PV+5L}*mcXnnIbj*9F#dk%oU>EssR$+@Z)odM-!GL)QXr;jRqC#&Yrg*vfK
zON&!JP~l;Odmx`|@x-v0x~^e`rZVUTME)uF5D4*KK0=hwfbA!MwJRJkZptWABwK-e
z*__bs4*qaD%r+u6UOnKJ|K@m!?V3ZSoZ%EC%J4gj%Fx<w5;gjQek_{mWu|>xpiEN)
zR|!&O8w!XInP;Tlm;A$!`2mLP#@`ASu7q`DuA(9l^KMm(#^3Wxhi!nN|18<zH;{R%
z%HYh+pT)}zHeNs#BhrzT+!<0~X4Ftk)OXcM0F#UL1~`E1z?f~QS!M_`aX2(X7;Jp}
zU-`Fir3^XpFlz5#!}hK!)}13HfH|uW>L72jkNt@H?P;mw92FBs4^URp@+ilH@to@~
zE_sP$U60iukhm+Re?i${kpvnVOlGu|V7rMso!iM(W>hhEZn2BwU=|ciBZ`><RVPWP
z5lW*0IU{tOKkO{dwq>AV3j<PD)R}M_bVFXdR7_MdWUAeLGwzz<F>Wsmi-G|XS1yAJ
zaMvdLyD`@$1l?4oe>~F{nmWF(MC_LPNJR6i05SbZFTX(|G(1lB#P=6tvMCPCbWlmS
z_JgN-VwjDCgt3!!8y4d8x1YvnV%kz7RL#qK;GY-ZT<o28LKwgd8I7AeV@et?8j`7@
z6(qzIIO6t}%@uQif>i&LldZ?vKHoSs9Dl`<cvb`A-XFeGhMCn9*vPdFp8<G-Y2mYI
z69E7@)GPZTs2UO#%Vq|P7;X*MNDo$5k8bfJK*FFoAMPy_P)ZnIhbjh{3kq=78s5sN
zE&knJVrSh$8!ue=JR)}A-nn~4#8OV$pl33n|K2?c77??g{(FwuV77+-LksFlwGK-`
zY!EM_h_R1HgIP+a2T0u#hYgK*etRhe1%=h#AD>+Imwzl9>lpk8fzY$srUwlimr+L4
z)Y1BLD-XpUD49>l7Wy=Iye(r`+kZ>5{ypRKhAo%HY;26k1&+|gA_4k#XQbR%(9w$E
z_xqZk6(-5UM7848A7q6L?ZN7Z3}X#z(jC&ctqAlWf;$SZ(B~srE#c*^1wj@GX+p0>
zke5R!SWA_#YMPIb-MgMGZ0xV{8xna!F0mR8%PA2!8LQhqiQ|k;_*UO(p_R#>);O!b
z_8BYsJJb}vpHZeUkIge>!K&bDS^2rv+MEd3xUnOK?k`}^MO1x6uGpEb4rax0ne^*k
zo)ze|>|Y#Bt(|Qlki{kg`p=@OVnHm)uhw2JtINb5rw7m$p;H{`k&Be@C9w{1(epCN
z@QC*>*D{3iFLG$t7TyZ&A1|Y_#e!o6zmaD0f}+iTP%EQU`yg*h1R<7jI~T;&(Av3|
zhSrha`6sq&!RJ7{qi16hkM(qx8C$B!pqaB$m7d6%xBER*u++stDnn6DgBnQRvp+rG
zkSwuUpysYR!~mTu({4cSpdK1S0nW~J%s}*3DswFQT1>>ULuhiRqzMu=X}(dnY84)0
zCwdTD3dqSm#396MR&{ey-=aO70ZbZg%}8)rWpGDb9K{H`SEkFxX9o+PzN&FK*tG*t
z((C=(`T#a=u$buTvi9}Fl>N=1c*le-kmmqDP^)q#(A>89bL~^qW`8aDniM;`Ji9i}
zG4w?JlbtCFn<UWCz5=CLg3udBf5`1d!_Z>7Bf3ff;g!F_C<_op6V^R@Y2VLhKZ;SI
z`R;l8R^;wiRA{5kPDjJ}{ftI9kgUQcB6G~FoRxpvh3mE;$BF?7!@X8ExI#r%V$oeC
zKA9NdB5jr-$O5gr_<{5jpn%kgFJw>4$jT1Sw}sI_^;&{p8M5*7TU%_O>K{>3Qo?dn
zWblRd?-*?Xz*)`!xTLu$mzN-?X!~Af0$~=Xrc7rEYjE|H$DpHU$d_%3#<x-DBlOy_
znw!D1vsa;Li55r|ipw>irsm{q3Y{TCXv2cf&u>Uk>nY%o?&f9>nrF@lESfd_I$QWT
zhF#M0w#=ky{)g1>XPRg-%g0qB+`7N8=?$0Z@)x;<xw#yBC0{dCcANKPEWl)YBI_A(
z`(dlYC<@l18xVt66^u?9($P<#q+G_|-sv%Q=lld2E^WiNM;8XPF;WLtWgEOoLTAq~
z9XW<d2P|@-pJM)=Czb!*yUuHL7>vs3+$VsU49nT`Pb$q{4CR;Wl#ESfRgW`YaII|I
zd!=S~Is7Q_0jC%sCeRVJSUsj~(o&}TUQT0Hwu@vy$<>20Kk3`eKqo67K!K-)m#=3|
z{EVP*cK(PP>VZ1Hop*Yu0QlqcJ<JRx$+FJ=&c3{e_jX~VrR7rif&!(3yVoJC`Qvfz
zt9JY)Z|bi@6@w{+1ml2+kaW)YbHS|53Kh|3t}tMxC3x2LM!sWBr*GY{3QfJQj#|Ac
zF#K;@1k`#Os43r@@D9$HBTw}+73uzD_51c?y{e#Z^kax$d&h+AAG?oS7Ln7*Xms^J
z_ONloA`CB-OPcJl{jHd%yl#R4(r+9)kdJU_OFom08d#B!h{5B2*NNMt0$*5(on}`(
zsDusI-Ki0nRoT$m4mr#}*Wex-Pd;w0AtX>2qxFaJ0j7MVj*(pJ>{BRyUIexMs%;g8
z%iC&DODJkFAYLL&JQ!QioJQe^iEL26z`&sdAy+Pgb^?pZYV~W^u9*(!MxP$6WoVQz
zZhqp0KtyeRSC~Y<FYE{H5hW<8m4bAoI>g|>w52%>-TI2k8?&eX_Xvs>&9-avu!FKv
zw%fb367GZHC~%L`uS}DYPgaQ!tyK!=@QSNu!^>(l(sU6hH4%@4(~&2R1N;6!W$;-(
zecZDeqe&5wpI8ze7Ad4^x11B4ZjN+Jtj)#Xq%%g#BPCAb;z$*EpJ*Ars|*lR>F+NH
zL#FfFW$I%iBWa<20Re*_F>v;GS|}4yYL=F8B37+TZyaLN-lXTG9Qv&^MjO8i#HQT#
zx@eH|i;Ij@RE-<s389HSOz<V%cv-08Ur&nfKwa47`9;}?dEr<Et-iR!rW5bc>|@4s
zktsIbhqr@#MIKHfWI_pGMe&-pxDsK>rlmtb{s#QNcD6dgz~&bdyJxL#(sbtU?u$$M
zh>Mo5jc1qpd#j5!WUw_U8gi`^v9r=F#RJTyyZKHJq*q-#5JT?M7o(+?A<yy?asUzv
zx1kmPQ2*A?FE2aY!X*>B24C`@m_4rreO$gf>s<oUyfdHRA(^9^Ts$U^A$V6uM5vt>
zXzj3X`K!1cxKt)La%T@EMc_)Ln3pzauayh7x3*Tzx8Pu39vOO8R8$zO*47&qIWO$o
zO4>%>yeP-R?q@VK;*_4}1(>)GlE|#86@aM>7|I4hX<qNQ(Ipy>7Mq|WL!;PuC%8y9
zo<D7*$Vhj04;bm<P;rx)Np<tUMikUIGyNQY$wqKGP6SmL<QI76je^PuKZX|i`~uiY
zWPbgOHcplwo?Q}KDU{E(ef`m8Zmx)IDm9H1<76T-e16Vg*MUq%*mSfsBc^oYB-l)f
zv8+r);=liV*Rm%+*~7N#i7(10X1-M)9}@Z|3!>P}#LFivQ5`r+koLs)d$eY>10xOn
zGuD|FW2Kc#39i6@L=BRuh=|C+Dkx~tRk|K0X;{E^0&(J@)b#`_?om2cguUC{J>8Sm
zK?l0!{bGIckusR?p}C^z`{X#nYHCJiK<NUkH~=^)yY=bY2I(1<TrIO}bXiUwIvYzK
zI_@XV7?31p1|FkrOf0NKlq2f$f~|6KH79>qRo6YO(HFnNun72^mXg;rO3kyNVYJ%&
zda@ZxjuZA#xhgp?#A=TCZ9F^{(#ITQ2Ue7IbP|9=sT(2h<+sukW$;stnS=C%FRQit
zLOs6e-H!D=r~X8?jyD|9mphKWezdgk7-|LpRK|)pLh+?S-!Z|$5)%{0_M}25ETg&%
z!+>Qk%D6mWPV&(BFuVZW=F`wnTz0g4Y)l@^p5tC+-xeOSSf4Zy7NlWg+d3byZgL*y
zkC(O35t73(lI`^&EQ$z=xQq%`g-?p+2^^_T#Zo9bDZBkJZ2qMhrCcqN->0*^rxCNe
zy9*4H0E@(CAj`k>_P$2RevggK86Vezh7OssC?Ui&n28eBt}kd+IRA%z>lPnsYkPa!
zAAalRO<vR%u!?!Qg%Uk{nCpW_rn9fN<PpM28mMytk+nJyS*p#z{Pf{P{>QVuF#8~^
zH~|XY+l-|mC_btq9Mr&1aY<-@fkn~Ka3G&^bVh`mb>J1jjcahe=DOLFA1FbEu_e;l
z(KY-FKVdj7E3tis_bk~2_#rSpt48$ctEiZMRJ>nmI@6h7Y`v&z>Ubb3v^|buK+}h9
z{>^Wmo-a5*w|OO4DW@knm0__)oEUFbN?+*01fX}DWru=9|8sJIVx^Yb8-ux4BE5A2
ziNi8Sf#GeydPZHcS{4r2J30rp$cad%uwb*M+<~{MfOG|z$+S0D=*ZVokBjsqCfwF=
z4;aHdFE67?8WQ#H)1%tEg@R&%1(cAIGXL}E|I*UZ7=$kyREvv>icp0Vps<P)-`7<d
zGUY*c4iV*MKW0}Q?M=Ej)8-cJ_y^OKP#n6Ov%A0!wpP+|wo(Vj1O1swk&axJtZx;X
zY+zg4EmX7vG})-2S+NSsDfG-Cci_vBVWz~yB8AkZvnG;tFt%qrDK|hFqh%1b=RpO)
zn*Ht$-g!smewb^CYg@)>#kStX_DRlYjeSIdv#lSfx~Qk>g~z8kO%XNS9(k&=iZ!xF
zRpC3^&57Z$A*s!(b_#E{o8&S94VRz*_Xwi^K!Mt1I9Hwc;<!KUH4U`H49TjLpAP=t
zDHd?CQPg?yyTuPN#62^%0W~sux<&@%@k-0>uqJZ+G-e(FrTojP{M+lJphqLOyI7tu
z_1N%7%SiKNci=!2T0KRr!vdd`(VxsK1iF-$h00MJh`#R;C^rhzaji})<WL;SY?&kF
zHH`&6vx|@KRYO1Ot!5s`Y#G-vX%dnmMPpTa1_XL0n5!=OtsXwx?lz)^#u`>9SLcR5
zF)Cg#Fuf?g8+v(Egb(P<p!cwChs|+ykT*4I-70A15_03>9zu=sb1IDco&-{g=!Dw!
z2pu_b&;r1Utjb2_WdrIDO>h#1`AUg9Ic|PDH)lC^^4hiS&OE{s13n@Tz}}lYKpj${
zCGmG__9|N$M6Dv!hlQ@4L^+k?M}cU4sb5icbz_D{rJUzY2MBR8x!cMBY@w0+=(wPS
z#K1!S0r^8ENu*%ei^l~AB{mBWWlrAh(;$%+{j%h)lgZ$|YH)3)KdGUgv-@4L?_U+}
z@jhQeYvgrT{yJ>A2B06i@SKmxcLmGwj*}rnrbGB5=Si_vwdxmVM@Z0x<pY;EFu^!&
zzu~qV|J?`$YgZi$D0@z;i%igeZfGB-5xKh*0o68YA_~|-Jls=K04{+|&~Dfr>1~C-
z&<Icc<oXP#Y5DWS53+J7<DP_7+>Cm``M_>rLyu4=t>VyJGjCzflfEL!pqFjRJyYqc
zuQH%&xe<p=Pc};a@JWD)U^z9J)s%wrd$b*hC8CsJrq2{-yD?%eEO{2<_-sacLK5S%
zohfy48p+ZbDi!U84p;TzB6tu{5@k5H4m4XiE0k9Hdy>FbwbD>FEUeF1`Q^*cE{E$`
za>6Gsc_!?_GN?e+2*(LwhwbnFcJ>R%%NAQpH(s?ZDbcI<1v-^;{U~}{c&yr<t~6ot
zyX&W*$yvY|e+zCZ;c<I0va^RYdVOk$tbK$?%KXc(xX)>ZrY6$X!g5F2IY5;`Gkc%v
zGlzL<s4C>+Fu}I_ltZZqT&*6lyKOW`jT;}icT=X6Q!H)c$E&2367zFpI3*c4PkJ%T
zij?jb+aGG4?o^EagYOH$ON_i+g{{CMA=<_iMVx$abcDp?8BZd!EUzC>04b!sWYs=q
z6>Xy%(Zn}vI%m+h*g%w=R<#fS>(b?qi235N@aXsRZeR5e3ykaH<<<^=q!=6)I6V^_
zHv|mOMWL=f*GVQwDB1~1S<f8-nK<}if-J&r_Dk}+93s8;LX)G&2r0SOXlfBV&}@1)
z4owgqI&YbD&8`3u-z-Xt3}{2BlXuay?#BU((ih5P8Hm)t0bND3{qnPF2oT&sb_Q0W
z5I_@RqNS!I;gd_6sa8CYB2+4v(N7Y{yMTm$uumppJQZXggW6ltsIp5}yywWsVzYF@
z9bv{LA1iw?bk_WqFu6L1K8RvGD>s0vX?X1ku%sZdNO1kE247bZp%NXe)D<2Fg=lN4
zyb7Hqt8$bX19~!G)ZkQDrAz{Qod*7DA}Gj&$oc()ssr{v6f1C9(NIie1f*UpAzwDZ
zl-6=ivycb)nWikPR#*^GbxE}H<n>MgC~!lTzoqbbQF75M3{pbwS3D2!)$G~{hI<W)
zz|Tkt)EhU{KX`VQkC*6o_=hXq3iVquBL*lM*tEh!?W%i?Mw?-IDOHIN{=5KjFC`-z
z-Z+C6C9Z#bQ8A2G2mYEEn%dtyrKhKU@%-`KgrJZ2A`^iFi{qr+lf|H@3ANr?{jki)
z`P1zAXtdk-*(fkamj4GFXF7kqipN+$1JD3GuZ4D%JOK6dOnU(@8ub9rlMY|*Rh1dD
z+so$hv*mv{T4ND0d?~CXiEplIx_b-Qz6dE`(1RGQtY`6OOtiIzKbE|GT3&?~_E1;G
zZ;S`@eo;eT_NR!Yk&WBN`^)cliAYqJg#Z$g#NX^4rQs|Fh5=x!1%`G;<|iuLtIt%|
zx-VcL+PmigV}|pWeDn-Uh|C3?<nTHrN)Ds++#h~o(XIy6k+o)5UHSay_~9!&fwuhd
zfZfKGxzzt0MDg3EkGo^(B8ja<5bmIjWwH%OechT%gQaHGe7aA4nW+~{pk#t4KD4`?
zOsC$1l#SH78&?&dahNBMu3v(5B1-{^!4@#!S_P4YQ(`m~R1&j4V{M^bwwz3<TXSsW
zel*gNJ(e23={7IJZ93An;0E&M{9i%Lv(RXbvF376G%a5EaZyutrRu~FMwxt=wYiY`
zxfr@G4)`M106Ao5*TL+5d6=T{Y)Jsm(`fXBQ={XarbmLXyIv+y=l4fYVAlB4>n-z>
z8{mBn12(kIc`>Qq!jfmIvcuWl1H~RIY@~;?PWCIVEi^`0n7%n41Bauadz0Nf(JnZs
zEDvWF*eI9X)J~u7eKlk;KAc7}nGQ-d(B#qACY(0{GDs!Y65yW}tO+s=d}{$piPBoL
z<3AWkE#o?wTLMqL5E8&%myAxp%K;7cKnKS0AUxhh`fC;OT3t5E6o}p~Gnd_!GYNMq
zH|jyxTd6aB1L76>SAji$d*(yqvhYKvaAa3GXW~ss?T*EX@A*OKwc%bV?EHRd(D|nf
z`0^N^q@JNowRnh}_{q5PZ0U&V!|So=iVgZBI@C9{62zPm%wkyXbS!ZKOMFNX2*+r`
z8!Zbz>v9sO45(I$Hd3_P$V1i6FD(OpVq?5mn#{vP!9DTh;^T;S0nn*|x4H6kw2D<z
zF_IYcT4>alr?#Eq(1{Y2ITKc$Cjk<qMGT)*=wRN|M3gpa=uFO}*|Vi}el7##=M<Ot
z&8WWlJrF5F1lm=fVh){E&Qv_|hNgtGDG}$vXxuduqWiuL&<E|0bUvc3_0;kF9<Qu~
z-D}KC1xqStWb7*)`!N<ar_i3m+Qz7C)4fan2qy<bO`Qt=2b{MJ;DURr&e(Z_@<8jz
z$?iJSE%XAOKhsI8RG0HJ;^@xW)7=`nVw0iq*B(Kv$nhV?s|WV)w>T_f&n!bYDJBG!
zs-l9TJVk%P5mn1H89Y@M%Q;{{N4adnK>=pjz2u$}`4NzM!*n>k3K-BF+DdQ?rF#QV
zK_U6v((b-&=;BkEG=9MeEDZeQKFgzbnLyYLG(_u~Eb?p-0-GO%-uVwv?A4tx@f}Fl
zz!o7b<07K8%&E{tlmjG|yeHYEy`Vqk_B<t~i7U1Sw$Ti<p%jQDtCy|P7@uLR`;&mW
z7`p84nYZKR+C2Dyaw2~R<%sTdF(0HpIX*T8h6r@;?|mr<ENAI7i4(MZ@o4uS;8*GQ
zCwCv!y0}m5Y_AoLz2`<XdU&%s$IJRYJKz84z<T-;q(9Qr2BRpg#OQ=%u>Dy<an13S
zRnR|B@$P=0krh%Yd|o7Ns4H_h?;I~|pWEvSO6AHQ2c4+5<C`N1w1_$UjTZ#!2B_HJ
z-g_=(gr;r@yNZ{dSI-8i=}!fbF#yJA8U|_~^X}#7#+I^MN&<hhwJ46EZ2A7EPu{_n
zLYO{BO!50&d6vs{6$4tz)0P9!sFW=F(t#118FX|EZZt~d05aAuu0XJSGbm%ww)X<<
ze9@gMmqIhH7|!aba^ZvDDqXIK;Ag1a7(n51pKqA>Whu{vCeAJ8PbmN)VyI$Or-{3K
zKf0FqJ}f1W)B#myeo_&$d$Rmw3mD{R{LRp|@T3M!kY1pQjX-=kH&hK7(qyW&u|m_&
zps_*Yc0bN{A^LHAd4V~Lt!(sC_W7q#9kww#d)4<!dxrW)rm}%Kj<uo`43~iPZcdr+
zROTnM+JPB+5k~lPRha2-F`(J%KaGh9bI_tYWbOl<8F28?_~#?g92EaS#l%9uWcQn=
z%k@jm3~(uO_lv%w(d_}8QQ;+Rn59c#vRPRaH?=Ir$w<4jaH`&z;o5wyLFhmhkPG(5
zt3+~4^O~dLwg5wC*IgExYUnOm9#;h2GAef8e87^3`m{-QSs+cDAaxxgud<zq3yLqc
zRiob&H54%Sc4o3AkvAfKT9iJ6$8Gd|bOgH#?Mv_9|LYe0HMj|&89h2z@QsiF>jW_B
z0xJpEVTRD@$(rN-8a^cxFd5UBBeqFv!P5{251bM$Yqx07PS3df>ZcEINv&Iy&^GJZ
z69jRIOP|~*!nE)(KgO+ow0HXfUiqHMLCwnJ6SZ?zfwSY834Z(YZ(x7OztOjk;ylHa
zpikE^qS@3+avwux>kKeyCxE1JW!+<}4!ealKS!21y*z5!bt}LBe7#x^%a1CuL-yeq
zVDU0X8}j%7GYmY<j5bTxY2ta4v$@H%9fG-^zAVFYhw)7DD5A{?lN)B?p>g!SHKO1l
zlx&rZteBD&R$UQfcQ|X$Y|hZYfYmu@__!=IxK)8~OnG(X{_eqns+uwlxA~-kJ!wfu
zz1lI;M*J&YS7x6oXYiod%ZNp(@d{D8`?t~6S>ESQjZS*e(bJ&-oMs=Nu(^{DG;HJh
zMN+4wCA@cy6zB(x654nO+-i=o{Yqhe#Xb6m3Sq<AL#g(vWt_@fz_rrkATw(_RW{2@
zbpElUc?Uqf%fF9-62=5K$r)1mgBGb=h)IMFkZ=zO77v*n)ozGUuTWYk^?!iz*Xq>*
zCZuhBE@S%pX!hi)p}N={tD5WPGmQVdBemY}0Cf=k7!}y>(j)3C*m%JTS`nk!%6SQB
z@4}q(V8vE!9&gardZCRQC|!s@Dyyis;nN2M0IJdjZKwj3>*T_wIh(1QdNY(-{FgE-
z+==yTK-Lo|zwJBZ2`Vx%7RUjcfD<gzFI%`2aXlg-96I2Ba~@6NGb!-p&2s|6^%~@F
zZd2)kRAMDAaXqf)U*D72*$YzDli8alHQy1FeZ(MpC-!a9)wf~`YY-Ldg?-~@%!MYC
znfTVo*}AKdb>JB1Qur6Xf>ZK<u8o})5%2V18OBQ)@slYbK9wz@2-@{kzGuv+k71Gy
z-+&%FTN!4iU5WRgv6p>o!o&rI9QR;LZT>uNG=DMSAp$!()4cik^Boj5o(_0>adRhe
zX&{}hAO!i9x9k&Ud;(HH?pFml<hXPNrk=L4a9Dctu2I>5!=NB5<tYgYcE6X95b|b5
zo-zY7p1tJ#k4*Z+&R<X8v2IZDaE8>`hw-71s<_9~H>DIFlYHgLoEyS@ZFaU=o#sj3
z+GEf*ertWAnm}Bk*Ele1sF*TR7O$67^II&1+`He<*(V>>s_cjakRClbl;@&LwQ?$s
zEBJAeH*`B$EWRRKK7$9JaAgiOziKOgm)vg>GR6xuS*AMM!Uoge{>tU#4JPyWm#2<z
z0@d##lUu|`ZoWn>MhaQ&9$$tm)Z2I}Y~Hi@NYECq|IWwGW)`4BRt}mChW36NSXY1g
zt#!}y8g@yHphQQD(c7&)(#sP25AaN{pJ8jfH78UY_g;($rrctkYzkbxE##xxs2$Eq
z#qza>_V#?c_x%UaV}0h+cBBsOBsZvvLiO&$?-yGTA%pVgV@>4lk~n<L&^ei}C)N(+
zBE2zcPuX0XhqzPi`M3+qJykV<$>TMBYe%Z~(c1^H;cmlV3ZFBD7f|zz=(TzLiz?mu
ztd`5TKQg@4_VdHzALq9Syzl2sJy-qSGi@{pzdLp<z$L3ZgxJ}Gn6V`0u!lKyYpm*L
z{y_RmTVm0?KE?bAPTY%mW(pSH6edRM)rP6>-_<#$+sY{HPSO||ncr-N)0MxU`R}&<
z7{_{;E!WDPLspkIy-wEnf;Zv5hZ&QG9?mdxLfaIXr$-%36>+!6-Ah8R&Kt29O5vF|
zzB9t0@SnvQ!SU3Y6gsH5eg9_}5~R{*n~rM5QrI88=@NK}wZ-82Guhgu1GhoW>eu8*
zd`nA>uDy4pSd9eB*2Aa6HtTXuM{Y%f&6^05z|IwR5Au>&JWSD)zBiWYoVSNYf1bbY
zP89SCjKvR@8THL7F^irxJ-eNJq^YT3hO5T0B$M30SWCIl_^YYrnwSLQ^>a(iGoss%
zd8mGL7M$>zte;)lyyv{Pym#CI7QGOgi4c9Gg;VR9{@mvOdV9-mu*s=^hG%{g+QYh2
zZFS8LzMQ}gKKW4RQqZW((<h{WVPMgszVih_NS0}=Y(O%F9{XTaI8Zezs+}-LF>B?p
zf9ty%e^_)vLp?bHrVwEe75T6yeQGgz*m#A{<Aw&+-uos2o^{#1dP*f_<<>0`H#KqC
zuvW^{aopm@StzOPSBu1w$1@Uj?s|$(dWtUg9H@OTW)}t$qUhM1AQ9*&Pp<wq;9o4k
zc%BhE10E6>acbNlU5Q+CepB}fl~PbVuM}Q=trYD&RL@P+NHl9vo;Nf7=LPU+YM%6#
z^Bk4a&R-l-IYUJyCcbVR(aRhp)wqq)%73VF|8)#r!tJ^k8cwDn!J3D^FPSPvqMhvB
zNnU4Qx-1y?A8AW7$8S#|LQ;=2rQkme2aIvQb2w6wy97PrWQf~HESsy7OZfuwNb-4{
zz?>iU-j~X9GK9h0Z^0rGlY%Idz8QAm7{vX&A2!^RAJFl=DREo<K6)wc%%WBV&IwnY
zxaXprx2#XbZGzx#7?2xOoVYT1ZC$TKRlnmCSZ(yW*Qqtva+7KL7FrN*KLW`FSG;eg
z|0f;Mq5VO(5@LcAeLKoC$q0xR_1?Tu`wEMc^O{nD2nq=2INrV?rnFk@@$vN;vCVF)
z0|nWosX@e%{L^Rbdb{f-=QTS<Dr}dZ!Ir-%Pys)VsSwIko;RfZm<)MiXwrC}&8-#e
zIElPrWuq1Jh|585iCz?tLD_U<#&#A8Q{wQ)gHay><8a316t$?q-cVB;nNiN6U=cXT
zuuY`P($2TBkJ4<Y@T!57J00c9Q-~>%tqW;{SW|iB5MST6oUGtzXW|i8DEC+)02^<8
ztr#xPs=B44HH!GM79|T)w8tlo(2zld%&iE&VU|3C)AU-AAwSB6#t`@?)BolxS8U1Q
zv#7kC4lZWa8A<hLm;(QXsem0J<!v>lb6mQCA^P_1V?Mr-@3ETWeAT(>QnFvJ)pe#j
zc-uygme&VDo>*yRmttDUs_+AyNsW)^MOyZ6;IepT2}z}HR+ccqXVFu#b-{D~pd+RO
z5$#*?n#KHyUQP<u{^>tE>rQe2SB&q%ob8V4I#bGBcL877J{WIQEkWAu){=v)f;4Fr
zIQ>`S+S`-Qe=^-pYX#B6)*W{;UgGlL&%a~!4N~8kSm>S~pn@)zf1lmbnyAUH>g0GG
z;6L3%De|NhWd+~_JtOnWwCjN%h+H#9?UwoCK%V_iOm6_Y3y2~eF-@F&wVpS=vhy%7
zcS<S{aA9gN=bD>L!rA0%`0r<EoNfu7{VXbQy!MERrm&&Ha>+VpO?yV`l!LqMg_*Im
zimCF><xfK6{mEL&G0mP=e`5449ZLF7#@O)^|BT6L)&QXmHtT5CU9x65re4W{W|kOI
z(=}T&EZ_7ze;Ust=~869xW|2&Gk;Ps;<nR$M~6h&_r{nf4Gnu_a>MmvTB1z600!n(
zx@Wumu`1mLc8+Eq@gnM56&{*g5q%T6zsuh}W9P|Nl>g>CT6Z(=D~-4}Mk;NSjrEdM
z0Oev^?yVcGUER6A1l*mtEGEdkzJI$x*dCS`Ch_h!gEAvGUQBV`=l1;+#jMP+pL`g6
zE7TK$+CQ|)5}^}5EQ~2!=-B;YO2_oKa8e}Ao$);NZ06m4U&LOvS67A996TidoR&hz
zSc0joaE&2kslwTiGwL?;J5(uQobVyB5@IAkb--ar+v8S&0y0{#6c4annaZZCWt3ty
z4n5QJLl}7>eDv5A(O*IOpJ|^l`1N_+XwOR5cJbJ<R>P*cGsk&v&o=t-1s&X%fafTV
z%%L%-<Mp6vKe3)ImV6d{XUCJx(Yr6?a%WF>rg+`WiP`WnyW$92vS1ksFkv6krGmTp
z1#XkBzjTCOsl5Nsu=I4J$(70)&kuelL&Kh(iwi!_*pXp@e5}Q{)l;I!RIq29rW%y|
zQL8}~ej~{~>Uo5E0y=76mV^*iqxZ@k{MMZdvg6f@kog6}0k>^jA5V|yboZ&o_KL^g
zE<j2o{Np8GYv7)Rhv@puj_jx_hP^vk9-{c{qinwL&fxXyuST<BwfgB9Z#G6h|458W
zTYHU|SsXx=SKuTmzBQi_1-GTws$ck2di+C3G|LRKn|+5t*RJwxYb|?V#p!5eQ460h
z@W)ren#u7OC@5Culm~7zQIKOT-0F?KL>~Av0H7-4p6y?MW?8FmkiOf!OvLR$XlKC5
z#g+a%taRBAo43foM~4r4ukhyRDFcKmAyy-LVI^&8dg&<_$K&~4IeqHMUM9)yT5tUf
ziGj+FM-SsBm?~cUiG<b1e*xhatZBn#@x`78SfXXBm#Iy?(Z;v|n<RS1rZQ1WQ3GzR
zE%QqY+f7wsZLeOt+*ychJDVE~NsJ9lF`~ni5%Q7AOFs&<*(0f&Uit7{t>O-ub9{x8
z#3Ya5WdQq2Z{6nK4I%XdH>|CM=WAVmYE(IKTCXaAmCty@7S*OV7JgDoLAd31h?S|V
zB~<xw?Gh2irl!Dq-eC&{{MOJ?Jh4R=9!lEL2nyNblqqolgFa9uCOs9Y(%RWU2HU#q
z*$tZF=}jId`(+cLm(*$ds$}j!a*r()ciJZ&Q!3?e3>e@mXBo2}>#dfOh%35LuX$#X
z1ZI(&5#6J_?~`TE;!;y^H}9)Kv>waD)xNc^uFOHqINtK;%%9(?f&v38yJrxpU|+2q
zG)pkYw|gwd4Y$T6%pvs**JKr65X7o@z!rZ?g1^J_YUhJ~UC08P^+79}fge+QPrcO$
zZ-Z<emU5C}SbVgKcSPhwW|VMph?Qh$gIvK<)#bb)bvOenzZ0A568=jtJDjcV#>RmR
zzrTRR$@*$G3Qce=-uh@g&$pJkSX0+r+=jH%k~hMp-n}ty=q~YpCC8wtl*5>Jd7=l=
zz(Z9##eRPOK65P6M_X4yfq<yn?Gipm<bL|ZFS$2qZ;#i-f$>OCV$q?Pf{)D0zC)PJ
z@x0NzYc?9fwJJ;@-e;rV<iC!3%bV3#pBAJ_K+f~2^ryWF6Z{MnlqJJr%0u>2MFJbZ
zcGRQS2&TLrca>@<Ujc;iOJ_%~gRZN5z3hSZTaT={BBf;B%QGhjPh@cQI)P7MY^N99
zluI292OWOOEPVPh=q;G+M?O_X;niG2JL#zDQ^j85kfu}Br^Qzt4%(Yu+ilwjwYMiA
zkREdiVcdUB2a-*SV?z1`?fv)Ll5pM1v&jb4?la<Gck{x@y+MmTYC5FcV<x_jKbf8n
z?wG<I;lAQ8)@9&2xOEM+tmG1&a`D4dJDMCm+XJyTMQU!5@leJI3cP){;Rk;d+?o0`
znROFjND6x4U+^)eoYsz4k7n1O1#L~$wGyYp_OuC$O6_O;4J40u{8e|x4zOBOYNe(}
zq2<puKY_#v;MDv6xmVZ%P`6AGdhGv)sH+UCYH8aW5fB8CMp8s+>25(lTDn`hyGulr
z7Nko+q}+6OcXu~zx;wvtbI$wy^Sa!uoOzzRW){UC+!|Kbccr&8_xCBJi1Tk_RxGrK
zyMkoRT72|B9I;}Lr45L&W7oAVWzTW4R!7C&%XGi8;q+9U7;l{_Ma~|?tMr&l`O3d@
zXUlgz#f5TsRN!#j(p8x{cLZuLz{Iat7G$^QDq>am&)@B7q6bayag`hS31&29$jOR&
zxQ0kx;o7ZsFawMkiRpSv%0v@Ikcp_ZzrhhVRo80!C%P|o+VwesU3_rKYr4PNIm=EG
z03a<VsnxUCj3oTkxkY$)Ta7GGTiC)$=#E!)>i%`HU#4-WEtP=gq4Q_C{R;uI^pOiW
z<jhM{xkvIpl9=OJLn4-5(WKMWw$9a{_>`J2qri%s4k*|zU+JlQmSA-e7VcA=;#DqB
z*0NAE@wewg#Wb*fq`PxiZ)~h<5-58|eSVxGq`qw^pYGQw9`w#q_e*vuE#NKspKkLJ
z^CbMEf;i&D7hkuEjKw1ES!(m|#Ial7!&l8po6o<$P=QQSVVPr7Wf`R-A**mkwARK0
zbTf%W%zx+sIP%#3J;B()J==mAIUnhTSqLG_LlLy;thI)}@%pVz%m6kh-=n7H&uJ-L
z`LZsN(B4tvlH7Cr69wjm|H%k@K66kjIVsA%nK?hO?r59YY`2(&ovppxGJlg^_+enH
zfACXz_y<b1A`KmMM_g7~Xb?xqMAC02c&T~xZ_Y*TS~;x0SL&}#-t^hZ`jXsId^2>u
zyO)Y9EN%S*ih#6GmcV58;5y0?Q=wC1TPboMT6Ov$K@4{ozs_)25X$OxCwxo$m&hN{
zpOey`_LXi9*Lacw?%bWU_pYF%42fogtxM`$tPdmsxWNoqFtd5ti986pMq#0IW4bSc
zw7Ns9pt<Ve)cuno`|>Z9%Si-s!cfGrcu3_JP~6>M;@^3ZeA3O6Byyw#tWF4;eWicD
zPs~+)B~ocGixyw@*0%CpBzpt&N*NpUaGH+QA>X~q3U?B4<r#n27@3=kn>pjrZl5U7
z;z1&;dlkBybAvOe6FCdMmr5GjE=zpOHUK3YKxSCnL2;Hw^D`@ZJLuEvy}FYODsC`p
z)8%suot+pIq^x9cd6XG+uW~MaQ?y5Mx3!FXo@>_rv~o23$y)QFrJ+Rp_j1kW7MboV
zW-m(u?Z^U}K=BJ7T;fY5WmC5e4t?IO7*ClMe=TgBm1xzs8p_micpV>}v<`GUMwgw(
z9M>}AXy|AO{`>PogjKR=##q7K)gtJlfReRe65Ch~;o0+`j}Qn0^7Xym=}XoE-xs3m
zALi{|8Os$Z+0$;=KVVJ}M%@vOGbk+|+YjqMe2%bR%#xH0pm2VPip2X$p$v+CxWf@E
z&{^ycZ4SJJ+iTh<%8k+oM}ot~d#6s`WyoXLIuUYfk(rfxIllRF85v0u;C$2<8?QF!
zd_5Wj*AvJ4`S*DTPe55E`))`al*cRt4zLeUfXUZ}_{(lCh}}6D44NouP3G+@v`>)l
zAI_8is!wH~XY!|=2wCL0a!_?C)7Z+WoI{T*Xul+B27Hfu^OjrL+HbtC5SRUw9)SPn
zcKej^$F3UryvEL6aDvOYTD#nT)DpBGpIffBY?h0IQL5*~4wQ)WUM%~)A-WVPE2CF1
zCl8tZIfV1K#N8o2Y=Begn=bn(K&#IFQj?uhkk$?*Rjd=LR=W%vjf_{Y?dWXYc8d7B
zYVsem7?*kUM9<fgG1Gm?*0B~c`1N+&J_|*;EELF7t&={rnmkpH^642iqF^(jOXFGg
z^OSo$><!kmI<hZB-n}E)?F|@mTS=5ZyEp-|Sz?Lm=Q2-;jz-eBquVFR-p(Y#$*yAg
z1P{W|?bgZ%wTDqdqg<_Ra9`MFf*(H`m!Wa40YXqZt&Yeb4%7AIS$9c9QBEzZD;(sV
zh5#Agt$O7$dEd^0XoH!e85@>yUd>hohY_7HbnLnF36=lFWC=cZWF40Coc)`U!b(2N
zl<Vj@tg>{rvX@IUP7rMu<2gR_uHMe2Zs!0@JWu#D+-7`MJ_tt&oHHMhtby+PWo<0{
zYduJ`jCLY-k>4A8%Abq7q`{dwSJ*xZgqc60h~#$90c8cG8{Xa6`4Y9-*^hI(OKnZs
zJ^u4&-l^crcx`KoDj1m+g>s(*{t-N%&ipsuROD`0tz<*$+toa|`;sK|Z)sZc#lIS}
zptB?bO*_`U+6ZrtS_82v?NKPyBWj`>>0tj@Z(#Tm8V$7|b`SYo=S+KOKQvLT%cw^E
z!!T6($q8&=DHDRvx%u|*PU0^2SX0;Tm?J`pSEMPwOPdy9-rd^TaLR6e9%2*<ps~fL
zp1WLLy-Ttcwr`#o=jR=}IHCcSgt^yXW5lO@?t;+3-u-rNAImuOM(IyQ&KQtrJOk>5
ztVAEV<2@sztUbQ5k;HPpcp2dBiW?=-nyXMJX6|xk$>~t_sxE;kx@N}W9;pNPPJ3j&
zGI(fPRkKeEd1}NjYvd*oXa0Z=pSS)G9Zv9!8g;1<4w!{-mg>=Xb+`RU!0_&mJ+8>z
zzCC5iBhd=Y2!U6CSvF3zY{Myec<a#de9kE4Gc6>uNEnU(X3w)bSuuu#V`Dlm<kYtz
z`p$}Cu9y2@>G$3)ml$}^uPNUfEUyXYF6u;Ai-Nbyx_f~eDAPWbQM}7dPt6|zpl8}3
z!SKndz(+#-Xmb+#D{cq;D{@Hv+xS5FHbgHach^Iq_>yf2yZ6th`OSqL_UGI~8EATz
zuXXG%+_gzjj_mP|+VDztnSpHO<$WW5UV6kiVlH;%JKU{hkTfPc@N4+!Q$}Wi>8i?V
zDBgX2KhOWS_qD(eRfyp6YuG(8=P?GTuJ00k)BgHw{5x-bP;d+3ig9KHiE~te!WLJ_
zC)NZbRtkz0$nHtP;b=akja}{8Z6SJmYATa}Rb^<9J3K%<U_?CUF>+oTx)u5TKKWi3
z-Y&t%6q0>j;)$EUZx(!)gS;MiXgcnRpYrygdXLDRodMxjYj_;cp=viR*iE<z;K`p^
z>4@vS^U5}OwsYzC`I?u&ku)qWBtiRV#FXpN^i*%9?VEc^WFsoQ=!FT|g-Ihl_nBwl
z-<2n~$>RZ9_SFCmQr<(Hm^eib&e;=##K}u!!c2~(mCG|ZxaG{7o_!d91voVa1N^Vm
z0yRVJ$}8))U%fs{WEkCztx|C(X>&bvLBQR^*&fOQQb(>SPo0KiG@GP=;n~*|WUg~^
zG9+(0$G_jR5t`eijXKty%}2Us@A^l2B1E=?&pii`EHv0s&bQjX)q8E{E%(w{xE$14
zZ!oP-Uaj@@cQ0x-@Rx|LRa`t9s1`MNilkZNjtOL3y#DjSzm|KL1~Me{bD8B<)0xM~
zNKt`RmZCq1Fy1dLTDq`XLA~=NK7F$2s!q8RpSBV{Uk=y+B>4UIs<1Qhe42qkTiwMQ
z$J+RL4re1bl$g(<*aFL<d?$H>J!8aT{h=*dd`+VZf;I7F!#pwft-i%O<lkmb!bqzj
zrtRvDAzQOgMu~0>vuM!9Tz=IdCEM851_vbblI@X{XyPjaGb^Rp)wvjTsL{-qe3#8q
z*$ek(S)!=}!ULwom9<Z)%Grzn{nUCXT!y+th1MnuLg}YtI`q?ps5|;w0TN|J6vfFI
z>?~&a0dvN9L7-{o$MNd;l^MM7_Lkyb>aCGqERY|mjW}HjlZrGrS|slAp|zrRi^pV-
zZpy&Vr6lxG5x<H~a90csWWrlkqczsv&Y~Awm9Ou2uRDlvj_qZgp=?b9D39fQ@&qFl
zFXJDfSMJ$=;FJQivt!q5!%N;<JI5G#YE-n47uUzrsCmEr7UZ!E;22AtvTf+G%vB%}
zSOyZBv>pBAfN-#AI4kA^?&uo1g}Ism;VIh4(miHsWlT&$ezjJ#JVmPuI<EClP=EgS
zL~qHlMaU>=lS)=xq<Y>K7C8wjwT+gHHVSHZ45-QJZCtHI2jnrLL!f${Q4Iv}@SurQ
zr3LmYDB7I4l8BV=6*Ft6r@7aN8m6v0{LY{VkBkeAO2PtMBI|etOZEO-OM0*VW$t3J
zX;hCd_S&q+NfY_Yf{sU8c*apnzOtH5hCJg77&aIv#wUnVG{|@}$&H=A-Xn#Y3y>z4
z#g3Mwra~^8>NNL{2y98vz7fCvx$KX0FeeF!IN}Yo66o*p2(j4it|Q=N;7GX9h*|cw
z!D=*qcDMYTPm&Nx-H|}B9L`|ha#%9H_JMPM68{J3E@9wc5cKpHx7#0-kV1fhLJRwf
zufZuFo**MCL}e_yy}13I@2|!wSAQ*EOrIyMO^50m0QQ1Pp;+RJnH{}SEfX`^dzx~M
z)o>>3)2Sm{Xr4R^_1@%6F{g6Xle#ysN*8mIlp&bwNX6($TT4$wg}E3oQ{J0@%F}rz
zLbkay4`DsrQT-Dx%;Rmm+C&YCF#IpJ!j`c~LTDin#11~a)`T^j>Tng@sCcNKEsUaj
z{ZdX6y&R>ypZjut|GN$B_3{u`Qhs488PG-1OxgE^0eA~e%|1zD0^M;bkA%!+&%+AQ
zauFCl_9yx;`(>_C91+laa!`RO!>)x?Y<wi2M{)sL;oephCG|BnxIJ43sQoy6%ajvg
z;N5MACVz?Q9-{R-ZHiR5T!_iA(jCP?bBmz)hg=zW&LVfV@i6Fbxv=q*E-d%fu;X)c
z7njS0b#+*>K~Nxy8vmY_(_ghGK4RUAth;P*?TASqbMMZ^6Kei`0vEM9)5pN+eF%w6
zv26d%pfX%!;9YC>(EhTr&36Bo-@mfwLJW8^#Gm?B?kDe%#el<CPdDF7;<np#AYF}_
z*qV;`X<J+U0p1cLYpKx|=@F=IC1yD_zu)!Ogh)WwuJa~Wez!;QXv;Y~{+sl;p@1oJ
z_mb_V&r?(nwl3OIxdWuX@!YW^9^*)0_5o{8Z%<Jpp9noY)nE?crO6c?*xI0rIG>j0
zrOoWhlBNOWtn+o?9z~%+uTXH0h3ob#-?4`$Jm5Pw;DU5N8A<xd^dFpO;@!KUiGLmH
zT%4)MjKe@{eJIuOQUw3#d12vT+SR#vzZMbkcfh~ldmGS8Gk8t|Xv_`%d7~VvG~Hy$
zA?-D~g&SmaF2%i;;+W*}m8%*xP&*!rAXKgf3Y6Ls6vm*-cc+til6YU_Eth<43wAwv
z3VBBF72orF->w%6AaekPea8l3jobQZRJNP@ifsw2cu!fmObdz5H^d!85Dd|ku^%bd
z3+t=CbR;bMrExcSM&$MBjG!z(KI}&}8hj>6bE)m#l}zfGi7B5x`#nYk^01LCfsClS
z{Nz7SC2R3NsPaUue47PuQ<H&>3p;PExv5enE$wSGA03$NZM(11=mxOe(yf|tTXcB3
zbac*MQNBH`yC0_4cP^1f)oZ%?W{pTX)=h%uF8#4qH|+zkgP2a~Fyr=Znb%*F=oCCv
z!mgWcNkx6q7#AZ~D@IP~6lT&_N=*e8mB$m-7x=bs%Sc`0&a8jo;;h{(Lz>+Ce7MR^
zQgPLb$r^sxE3m_@&r6ovRkeqdV6Jp48AQ%uWzT!Dno=KFsNNb@>Oa73Gv<H!2k$H9
z;m8Z~W?6jvxIwcoUTAb@Yr-~`GYJzNTSU{<c@eZDVfWr!{j`kdo^p$hKwDc_4VQDw
zTSAnh;!2EycxjgSIa4B;lzRB`s=2cyu?!&cT?MVP>!Jb*WO0|qcQ4AQZ?9q+;Amlh
zcX=?*QOi}dB=XUrhI#dh!12SZ-zPJp&R~RUoBa8ovEersv2)>1tLw=)SPEMvKG#U)
z5o>h**_Z=*BoV+z{OSc*pbwrUHt*#ty>6Rp={1G`x<l*zg9ZmG*3;N9@FmF{??+OU
z_ASxF%wh{CDx$crCkNj+Jst3XgwJZiK>(zy+9Q`U7r*s|+C<J0|Gcw{u4_O2t%WX_
z<hAPkgtpdh%0^S*0t2wjs+Y0+NHCr46%Si{OF^V_nXSoC5uvqb(RE!&^IgB7c}Ghe
z?ZL5CZ;C+gLeacF{ngvHrc{3AK5hJX)yPN;S&5cOo?c1g^2?grMSYK{<G&H^-m07x
zE{yIMOwoNw4Zb_|m=#W#l;gEwQ1_=j7Ut>7gxK|>i<E;nxcSel=js@&`&7RsrB8bv
zArZ}>`Z)J<05mc32Nw`541$8dmv^%c+idcCn*ZFf|9@x#1ArO0kGZCID66)XGe}YD
z!uzK6T~xV4_%9L;r%F=rkB{G%<!cN6npch>|NJ5JNfk@m#7JV9mgs=-ge2?U{)85~
zQZ*hx_rEeJhf42EK$-gns^v_Lmv8WwE$6nZQG)ogO&JgwxR(Y^P)nlh!q9oJ6rRpq
zx^VfWxCRjC;@KSTI{|g|2e#Lyw?Pk`>7)TZFi%T0r4NFPuL0U#8p;y$H*+}JkTq$b
zp-<-iHyFW-`QL>5_8~CrQNvK7m;WIcq<Z?RQf7A8;61YCxo&_tG5+Va<36yUulbnz
zI}?}ydRq&De#KXwXH#!!CC%NFy&NfI>$xM3e?DH)ZxWXjZyC&Bf7Z7f8>_C^#T%B!
zjppIHoDRed;_tQ?phW?9*agN$v*%7Dzz_fyWY7_9rVbLQWU&Wn2#|JuAH?Y*e`rq%
z|B*I~A2E>#*c>Lb66T+@6InP273_0SV7VwXQ7A~p&OqYN^`HlRTI#JuM}L2NOUwKG
zsKz@Hno<HJJai&WUYehj<tTI0Kq|xOd-HH)tp6QZ&r+pk&q_v}PPs3H<|i%XTU=-=
z6NOQ9l$?4-5Xw`%dp8Gowai12AQqx$q$dsK4p|r2L{oPfSbk0~XP~A!m8NXoAsz&^
z$=lZ8ve-_fO_aRvH)#~%|7bHU<V~aLWDsPIM>bXyT6?^}ChqJPPnUt8Cn%RcIdt^~
zPK2aXODX2~kOJ2C?-gyG&C^V5#h+$c@?9*>b4NdBWLbb)!8=2I@9!V}!v+57RS{|8
znKd$p3}j(*QAus}LGDnbb#%mgd4G3gG-6wQ@NmH6eTEG-JJrh^!-H)xXUT8o(9z=6
z^pD_iV`AB&0cW<Rs!l5GU<;p@&nJ}>Y0!8{#sf*Fd1x?l`>%u7<q{jeOC7ww|HY_Q
zRo3G>7w;`exhFiXTQ{?nMz!@1HSYM$f4J;RrnC9XrduSwQ^dC;o|CR0DUofhvqvAZ
z6&-dssz8aGcP0%n9PX;53KSZy+dF94{>UAoA(f;|UYI|O2<VOt7@fK80<eRA`oWFK
z43m(z6+i_Z4`35iQB|ESe@T;d<x>dJJijtWv12*z!tgS}0*2kb$wH7A{CpRk?876U
zO489x_%93azgL_ppZntG5(Xi<+hGIbtt&=Qj<f6#=ddVZgWrazbc-$TGflVk)g}Vb
z)CmtwDI8y0er9i0J`-zc-Ihz~6a>gG%<Iiqv6EvIK+H+0nLLA2rppl`&Np({y~3`1
zflPFFb7Ge--!Bu=b{ED0hkw<E$7Xefao?$*c>WJ!4dq=iK2W%yA5MEn1)bET;+7n$
zgsxSj7ep7-x07aFotjjXG)*U0U`9t<&XJIHMmG#c^$;rizokJOwewS7fvN$MC0RH=
zsL#ofI`FC4-{rL+)~wvU<AK|vNL>%?!alq6CUHY066=<M&#91qfC@-r|3QW4pvok$
zL=Se9lLK@%$)3x`@o0R`zSKZs#nq{Qt2^T+h_Sf9XsYI|>LfRRL(blAJA-z5K>;L2
zl+SeHB+9ORddy0+*mpR{-c$M3MXq>aqMGp};VlT<*x8=(-rF7{>p{-M`BdCO@XCcd
z#q6)e5;MfR^Mdd8jn!)JZRKxbf$#n|N@GCRWAUEp2HScxr6$$e@53m$idPMI)%in{
z#Av1Hh`}_W<lChI1nRk`DJd0-m1Aw(X)mtk458xL+Zl|7avYyEVSYru1iKJ-*_D|*
zuGeiM;qll3rAZ2g(NrXi{cX^=t1aYv>@?WuK6zcpkQd^<yv71~Dkck%QKxtqIn(y*
z#Y~PLB+dvv+QP{&fREpsn<(}=KTIUK%Kr-F6J}Osy``Qg2U5o9d0go?tShWACoi>4
zRZw+Ac?F_w$=lP}SiKse=CnYOoQyLCBw%S}QhbjWQPim7;ut2nK;ZN@xqqFbE~|w`
z#4$@L6JtIPMISb`Hwi3oIXq`9M_cD?13toXz3l{@rRlP9_&oUs?$h|cG9r;7KUm|b
z6JNe3sd#QELp&>PvEm4HYfvm&`a024#`sJ|{$deG_g<;GtR@OToOa)S%x1&sGK38V
z68f4~xZdpQ!0^!VgvqhKh9tEtWKv=hIedAV`7DBnRK!V{d#p?bfT7<?nQ&}oB31W=
z<h@dsYs$*QRDYk5yE+)IPJ)(O9c=ggdkvybp<(M8JbNMCTr}Am>b$1P_q}OM153kI
zNc>QDMQ7@VMmZ|<@8C-lEh@5w{Cdfe-c*TVHw?$fR!-&oH;t-ARFj;ZJNS6yHK^h?
zPPt(rwUYi@IrM5Tvft0w2uaA(sly9f5NT#>jT46WmQ_jqlcibxFWV<aK6+q=DCjmb
ziUN}eW6(z1Uf7mfi*2cR!k(WPbmuGmx>YP%UFW`i-hJl)K2NSqh>~<B?b5zZ?nJfW
z3qbPHhjDI}tFkL`pxkJ>>#=89d$T$5|9Gp<SN|<=Pr!N1M^FGC1|$3vg};H7%IVrR
zu)OZ2@B7AlMw&NV=(It{{`Wirh&d3DY-t<nd@77Z3qo1I*$|nCD*mFy)jcLqXLwzb
zVlr)+<#oab(NpO^m<=xcqQi&e1GLj;Cp@}ltDp0zQi59;m-0!M_vdptug_?zPDy*s
z6}($oO0H>>A6`d`-~V4(_NdvW8yigB=4Im*?ts$>huiA+nt~v}p#T|Pn5{YDYhMF~
z_Q?n(^6~UnQS)G6HL>pUVJPH!TeknA`u#qM(qx(rn6|L)v#(5RAq@Bj*NU}fe<ggq
zrq9=!GdsIGn^LVb)KL@sbDi@sk;ZXWjBIw!CHkxq{O9uoDr~vFh2d_uG)~sW(>&;I
z&ka-ab{{>oqAwK*Aj5?i6$nhZ9|WF-GY9^zYLD|0P#b*YtY_@O2%t;6cmz%aJibpI
z!NVmaOx2%2urH&jtv7~bd3?~|+6@&|`r_uwbbGSo_Y9kIMlK$FEYB+Pp<R&)lLLj{
zp9_Ut@p;T%EeK$|b9{3P2WpKVd-Mv3Eh*AOA7q?b&jfKF2BXq{l`yR>XFLm`&eAQ7
zw##_0iR_-(aVe`PHZ1BSzhUFXXL)K4B3Q8GWoK*ko*%QhOvSba4L_TKejj6+%ncPC
zaBpH)G(!{UclG-=u)sX78f9l+6wBiJ7y-ex#8scZ7pWf-wOMB?YE*XY>5fMq;%0Mn
zi4g6{cL1mOPaN;R#`#1p-Lqza!D2h9^TQLvX=BR|s84IGrDWWQs_S>eCbH2?SP?xc
zx2eodW(6+SjEsveGsEE4Z>Q9y{w8)SrAwwKmk?c21Eh9+&>XND2&Dmj(MxK6Ou#Tq
zrp{{ZT65UctLn>e$yqc`tu-2ZxA1bZBp&*boZNpwCbsyg2A@6Ul=iu7cC}6ni>n%9
zFUj`jSauLCxyCz08Wi9ban_o0Z)Lh+Vil1B`<umkLUP9~WRI(TaW+?6Iy7A-_Xkj<
zeyJo$U)J~m(~|O;6$L$wVm?un$l%6Aof$t(RQwn@?5FAKZvBgd>17^Wn!vOI&y&wI
zSr0`{ND=X0b6jY|C3@`iy5?=L>u~bg#O;aZ*`43kt@p3NcwMudKvDDePG+nZliB63
z6oEi+IQThff@`eG`c|+W3UP6CmD?j=`${oleuj#7dmArUZ;9$wby>z<;l|`Mvcj22
z;LzAr`D4x%gykd`F3;|#1G7@VCQ0tEe`?%GxU9*;P+3a-rJNhb>W(H^Hvd~;?>^?1
zO55{>e32VO1b|$u|LK!Zr1I?fHVodQ{L&&Wz8a#!fS4wqZEH1AE5ugCzUxV=Bj|sq
zII?qLJ)X`0#Lw*+4bYn;^)=rc{(N9dWHfPq|IL!2<M6brOU2i&_2nx>E_L+w0sy_s
zcQ#p@kd`I{Dm-JMgJ?Qi-bYKLvRw!O&?;22BD2YNPJ957f<VT(sXV5Ejm<%p4gWvy
z_P@SGMq)PRB;yk~cMvEOkVcioigRc3cg)0h>5r}B&_*ESqkz!a>xX$NLHQ4oQ|vhD
z&OBe;a;|iu4*y)ki!Bl$ks$;f)7?@Hf|IT3EU=BF;g!{U$8Q{ZCk>=pP}3ysgjmf&
z5H*lgcxyF=|4cszLX#i}5t@*IoS)rXijT0ylV{vQ;8zG2m8#QR%i8tGP#*L&N0={u
zLcGd(G6@^_k3{aOpvZu746hwt>hFansHg>{&aFU?o^l`*ip142pW)9#R2K$*eTPI4
z@TpRB8r-RP-k?A<kjA@cWEsYs`XHY&l|9}Z?6?;dI%nJD<va(6qI_nt)U+WCJoAX8
zsL;|REPAWtl1*Npve~ctdWI^qh7b8h#TWllbq3%TOSxx1wUy(5tdfpni=<{{g@1{)
z@zP?96Y#nYht1!{!^zeu?Yd%%Ebi-gAGV_pe-A^ZwdEtDNbH7%76UbBW{bmc<_lNJ
zCaYk+dzHs7C+(Un#me>|<~Vps6}A_s0ZzB~PD{G><u@+?8?wLPM4<DCWE$L^;{3hb
zFo;+pm(&akdq%!8!&G*A@%DtEbY&7*XK}hrg-y}TnIxHClhZN$%SUe_KAw#v%F`|~
z2skD5@6fS)rY{2o@c+f(J<r-B>=nZZd@oEa7D=K++PYa`!$81hRzFiep5cDEcPA+w
z_y(A0An4&!xbgmLsB4&>0i}V3zB1ZF4i}_VE(pKhLx8Tgz-%n0<O@w3PL(&m&uall
z+E6D$82pxqs;8fV@0I}IGrNlc{h|o#qEEZ;A6=Inbe>gr^U8W@Kb!YHW%jDRg#@Ap
z=YT)$)1C#gvxuC-r;PWQ#5FPI?aH7sBJ(G3hJ({HokAxKwR>}^P^NFMqj8QX)7Kpq
ztRCGEr<SsD68ie0$fGRiBN9oxy7}-(5+?H!L_;I%H2o?wLbUr2<iNs!&THA^R_{Ae
z$czW#N01BWba!lQ=|afJ3h!A-xKHS9{?K`X#?{-Ne`Yk&&Em4T8fM;OHV5QO`^AxW
zc`4@#dx-gIv`eB3@jMqf3c&hhVk&(J3}(O@7z;<Np9bwK5Dwz34HvaDvLa4N<jH%B
zo&7azqy($6WBlSm2(@RmQ_K1V>4^E57@6;FA)sY^NttSJyqi}mcA)@e(P)Z8;}fag
z!2p**X-Spj^9DxkoLx*Hapx%prcs@dQdlWq`?Fh4?cZMQv6ziyZPJTbC&(%}Bd~$_
zyRx#PYhv=v*Y`T1-CNDei{IbB_3A0Xix;VbireOMON-GlF-X|;cr5}+0+lIY?Pnkg
zMYQGoyZb3^frFXN_yJJ-?irMfbsSuv5czzl$Ax4kPSKmCK$HuMCBe_v&IV=*L3s>$
zEg5qC)kst*dYK$H4y}?au-}6K<J531<0;aNdcxFe&#^#e9$B3&_{;|`^jDqOOetNg
zCz{pT!eB$pWOwK-Ww`}d4Y9mD?rsmFqHCv)>mWRSW)Vsg7&?IEZMR)M4ziE>4Otwn
z7dQUs1xwYE(xJ<xWG7TqRJ!NJWU8Daw9cVmbl9E87$2BTg-&*d#cb57L`8e?{Q!c;
zKvw$}P&+*<mxEA_CXVF{!19HQ<95&UlFIh}d3Xm9AMHtDdgim{XS^_d2haMI!kIYb
z$MdcqtNlP-MhCH;wLH|8zCvYxGf-?rGKb-G&CJwOHJ27+A|w5jbD>4`k{|rajbb-k
zfn@FE#6CGW2~ky59m(nsw_ocad->7_jgb8w<bI=(XNi`20P3mXk&&I0`DY?>*qbE;
z;`N<Zdy_K=U>%4lF#KxgCW0lPWMEE%<m*YDJyes?A81MdQn0wVEz~AdLuhco#QL;<
z43UCrPSjf>@QlyYyRLidD%jeOtHRsY>V*&A|18<4;1@M(>C5X)E|79dzFVq~W4Lj9
z1S~LJjcP7G0o01UE|n+=>6=pneAW|A0714jtG{__c)JRY&LsffVHPP=7(gB06j8aK
zDGGV#B@@rD?>w-4fZ<*;1x|49ivjZmEXD9mFlK)UC*DrGfCp?kS7#k!x~u(^zZ1P!
zTQoy{7+^DX)FjvbHF_U^VLlRfXLPV$KpHq#pkB!6y}O0ad9ZKNHvi?ohU`Zt{J_1M
zs4V}r6+}CT|I^bk(EW(4#M6!h0y2dna<{XDbR5DAMV6z^K-e4Hf<j}}qi$>{L72p?
zt`NCYwA*@KmYj`iH_zbc$WE2pS?GY_&qzQUu3<8g0-DFnbD|K=GHbqfw}$g>OL8`E
zIm2fO@eDvNb+5ck-(vCOnkvY@xdrQl0T=YbIkKbc1GGt^n|^H#2XY@zG+xvNsz5h(
z3=u}l{Lv+_ao_~kS6P&l4O{^wK3oengb=AS$L|jlc4KD_<1E#iJyHpyZ=V|5kO|%V
zLxm!_R(BuDfQt2{%H(1ghAI)_4_#d7*nnW7d86+=fc<w}hD*Ue%<>y^XMg|Vp+6P`
zBo~@JReGb^s$mJ>`GClI?AhRYb*M)5iUM}hvw2@q{f%|CU00&b(saFY6@bb0kGd+I
ziOZe4i<X2Nr=|rdr|tUvqTp&YJ@!X(QjC5I+>(`T0=Q_xbgTdb&lsM$3^}kdSSUFy
zpZ`wonu?R_suP%;ut#SGRxn>6F#yx4@MJg|ATD^3!6*WS@EGxdGTF%kL&~BEa+@RK
zZ{VSU4eUX3SYUyNN~?WNc7MC25_`?K_$~m0=;q{2k-P-S7iW!`(dO;M**LHWLC?yx
zyTrO&AGUr9p{=X_wd`%bS|b3$Gef`oc~(QmO+q#p*N`A%^hqLedhZNlsP3_d+}lsh
za`!mqXJ-!h%GiM8>1LpC?Wx=9$Lx{wWA_xlpp-Qrfk+TXro;&UnrBB1?%ChH_&RrT
zd7RP&G>rQJ&@Ypyv8l2awyUObEg)OD$%n4Bvk;S@S&J_S!jXV#p7fuD1J9SN!<z%d
zLLZPLFaM3e!sZ|j(BwlYbuKCnQAU~vYquaKQ`XGMJMDHZ7e`-G=Dy7jF+I-nL<b26
zRc?1Z6MbnbQC=XYe1A4442(syQn<IIcfm&Re+uu~e}Lb2fDBK`4)<^My^S)^;}*A-
zQ$$kg8lsxJE8H9qWYD;m*C&w`%vBr^TtB_`$MCe}0Qw+Y*4AhuwLQmzV(zs&hDubr
zdyuzw0zdl;r5ETQuQ_<altBO!BqT^a`-LtHSBU|&<DVz@->9mbMRm01FA;JzyEqoh
zV1LQMQ#`T-YD75bQ)1oiOpGIU`>$9CIktJ#Bc6N3uP3Nvtq$&!j@uWrR63keshT!c
z6KSwqCj>6xG$iOFAI?js=!M-<PXi0>eI{)aa0JXkl=pan{ZZ&}xCGMF5liVb9{Wln
zGjsaQwUf_FD5CG}EGDum<SF)`3R*AVE!4aZ_D!WenGQ?62oKx)%j#^+{{_q>WbRw%
z5YzofC3^>)(i)Bi%Yj_m(`AVh-2vZfJ+fLj&n~53usdPSA_l*M8e^5_Bp%F%NM!se
z+iL)lJ2(U|Y#tNeY5b!zxV-*X^)rl!OMc|z2>pZ8WniN3Uzx=70y*_QqO!H$KUxed
ze8@zFvIVF*>eBa#E+HwG?}*kQ*&HZ?<k%CGP=X_uS%HyclY1(r<spUt@!Es7>yNU+
z&~je*b3$!PG;I;NZ(+}x2U<2sr==@cH@^cW22x*`?ibG|sejxP1dEL?pe7~~I#3rX
zb?z8r>cG!GU5hRYPGAgRvi52meG2z}<_X0cCPe7i69AnFE}gREXr1orfh`uXoX%Km
z+uSjeCsb*6X#N1uQ_=r{sA*CtR!Q8}nvH8HK0o4!!A;Fnc>H<q$H7v2uA3XkZfn&K
zAO0g+aY@~13JPCBN4vN@042)kJ))|**|rjx53IQ1HvX+VUkv>IjF$p<PXmU^HyPQ>
z8IF9l&)oV)vCT@qeyt<EH>0a0sf(wP4wOz(o<49Qbn<^LA*^7#hm}70l$?w`5o~b?
zAi0eLR2X1%CU!Ok6?*7cc++QII~LkG?9OhQzQw_Y@)kDx?xp6wLmFF>r@0Z*WcIll
z0XZ3X(IiAMNR2PT6(JYg1*4y<ky)9MPU@qt=ep#kfNL<_zk7OgbB8S=c2989`XVE7
zgCs0{?lZ9`Q&DcZb35acO;wl4Nn-=}V7*mTJ<JS6By$;LK0zEpTs#;gJ<scLYVsU&
zt83)r)KuY>?qHwqA#F?ZhlkP9LcXrPFMhuw3UuO(WbAx4)-)xUFYnE!)XY}Qj``_Y
zH2caY<8T5;Uv=rrO)ZUgGTd?E9B>8BIa*6q=|M;k>Os8Y<mIxTSwWBt5)UKT$EQ*1
zA(+YVD6P}1R1LoGRSD5jCo?EZhlF(8U5du;l<&U6u>0-7s7gvdf{s&ddgsf5gW@@P
zH{;a5`IOoElfZaqNjj2$Z(;Jnb!2nNIEKLUiY=$kI8#2`#^K;4%{4PfqQy>DY$R=+
zy7v_#*5QG@#V3Hpdun;(U$Sy8-wtRoqr>*=suNPJ^wWh=e3L>)8?YQF9$4kTXg=wS
z!t^{7z<VHlB%=Rl03>7x7p`;wMO1UpBQ|u5AcjTX6y%LQCRZS|D~|Z^omY0mzy5wv
z92OEg(|MdTK8M~jg5FvIT!(eG2&i|@^4cVnb>k{;j!jg1!at3+WuoQp$W}<aW79y#
zm#m;_>aKq3!gf<N!;JS4*j^|g#iw~!dPnb5n9T(uWEdW6jGA2#9^vzC8clls>t0~&
zSxH~r=l(;+VnMKKDWCYw?k2`hN2UT6xH%|NlI5byK{P_|^vPiESb+5K7~jWSw)e^N
z-$9BIxSJq^DG4HUS&BfC_!g;x*Ynpr71-`z`5q`0zuPpI*1J?HDxCR2P7opp;?`#4
z=9*BXzi_P9^D$Zb-R~b;X58lx`hf4fmE^%D4zg`@u@}5*8`Qt#eZ$_c7E)=)X=h>Q
zs|v(PfD9Vl`MDafx%;nlp!ssNZW<0Q!JZm1;E#EdiQ54|du0NS@-TtI9{oFkz9nv>
z@_XhA07D;u62YKUmV?b}5DjX^GzdbmL+09te>ZJhcoOFN{)_gfO^$>?0)lzAyLAqJ
zgDuke`x=9sdJCd&I=cioBJCGi_7IcFLho#Y;9zOfS;Ob;r+LdDk!*T7@!PH+-sUzx
zAF@7jg{+t}BrCTyarJ!(aqsZSlWyOVN1*x-8Lg;*`lUAW<SJ(SCsm(CLMIrBj9Vzj
zB>}O#1gxf4DQb=8R;I!r%gAZJ%@BHh$P4ltAF+u~ruc-r^@<B@sBV7cfe09~CP&=4
zx!B~bD-U3%k6QDE9l_a<fuGsoKYO&lc_^PPz;PCdODJ@{Asrj90SK)#j$(ES7(WIV
z<$E}HOH4@kDuhJ4cHXcdy-sN%VHp-z3v(m7>!<x~mL!^mX9I<2SliQ8iBG-zGrX<_
zvyN5u60yAk!2l~Pa{O%@M1R*l=NTX-=ZQviXC=8;8v#l6pA*{q@PnQv?Q<GcAVT9I
zyh!V3l)%eCpGRrGehndP)Ru=2kUf{c9CBw!*sk+N{r>7V<?{uP<o5PA?y#P?$;nS=
z+q3q$WRee73XiO3Io5+NzWM?Uu`cP76;FnwVK5WNPvA+K-dRl0TX|ClBi3D(@9z}C
z^k8$}7KAlouQr14L)!Gu^z`**ZYbaYfV@OOaGhN3J8J6Vdi@m%q`l*gJBRe!LtGZ&
zK@^Pg?)K;LdH3p^?Dc5+m+f0P1$)V_p<dM!w&-ZYLA8-QiBbWId?(d)4Gq)*Hr69s
zWvyC}r|z=cA<44&D&uKvrC*YeA;Cmva+OClF*qc9tK^b$rX>!pR!p<ns_GWDcnuj=
zNF-jy=+alU(%Mcm<FeLmO<mExvQiZarg+`;hIsJ^$FKNqnKRkFw0C#V5Zs>Ku#=<@
zHtZWy;_(RI$nccdN6F8yDK%f6EcRBl;yJ^`Pd0Fb=oQd{1%gs3oIriwa;9IzEZH6i
zyOf@eBy#;)JYE0x>=ej6_jeBo3-crmQC1nHeo$X#aO59kuw8hK@gCUvoTDnwc*b4}
z)Nq%~Y<Wp+UVOTsz|r`u4TP91UB549_D+YYL&=!vzSRaLH-@LEtPzpR-1PnspnOF0
zmG_2%AaW^4s&tcGZ=i#tqg%%CYP<1MX4_9!0f4I}E^a=eBDf{HT_|i5X$BFS>zjq9
z2jS`dwz|q(XWeXP?w|QPs{drR{RR$vR50YU>}lI<Kd&bNcJKM_^|q|xZC8y1G_^OK
z#dPFIso@*6)_(Ajy7TmrwO3Gf975yPAKQz4Ys0xd6RL#)>*eot1yA?5LtgB+Y>mbm
zPG(TFGPzITuaZXI{E~ceHD;!?FJZa9e4q5Q=YOH$DIV<yR+Rgf*9+&eklD2Id_jOx
zAC-N5#CPTG6?#r9mbEqimA_g0Z=2MESh-6_{jfUstO9&o<2E+G=g;=(60;PvtDL#U
zzm_s7<qE;a@aWH+lU$k5+M2S*`Hj)X><Zk9&%MQNKXCEX^{+o)Ysy%~_JvLgGZLVf
zfZZGWYxs*tYD8sY#gClOe_%R>o}`q$q4?z7Nr)c;uEyx*7*ZskzQLr+3NIGT%W~Y*
z3I8RT=W9kS9td4BIE`7^)+C^{)Q@lBqa3Q$?MK>Ky60;YuSg!da*gaT23GB)(4U>c
z$$I!D1qcjcq!LY+*AXz&)vG%uFfJ~?$x@^_r6&?u$ms+IXmo|Nk^JN7I>L@t&!=En
z%Sx^Fn4X!DZ2&X<WLJs+IS(x{Vx>@~9lgY#M-(H7915S&Ef*aY^2|C5Yp$=SnY}h2
z1&hM-z|deVRbf;Eu%8vuZcR#M9<u?oV$phgD}6un$9|1Q)guHv%xCF02={wMERo<<
zSo495!-yFdHf#zGCvb7&?zt6O6$UXzSGEC<B+)e8FAfDGs7Qt~mb<AyZYI0_UBAIM
z&Jzx<4O{#W#N)~ruSF79M%wz-ZnWV)k4d$funzJ5;?SYvR{fL6O7M2i>jqEjicj3=
z(2?D9(hoX+{N{K8lR2`9340v`N8f;HN9OLeve(a3WWQ?zwzR}QJWi;&Tujq<-@~QR
z49>gwbv78U_?K$jV?hxtuid_MXL7Avwbrm@V8%z8D06}XQoFJ+f8IybQgzAPR0%nc
ztsK%^=30r30ej9!V`yv7OyF>WQNiY{TcYY22J-l$LE}u{!4EQ`kA4`J+pU+!sGV5x
zz3R7U8_tB&x%POj(!ZpkI3s>>$?J4o`ozJ^CC7s0(mGY4naXe)nrq%?Fk;YPGIza=
zE+y6TM4KZ-X>Sy^@^$0Bvh|C%o3MsQlC+-;6LJ-}(B;mX3rclZd2F&R2Z<nA_e?#w
zMHSv1ukIsyEquEZ+J>hFv!}93yx$~5W}Rd)u!;f<+{dG;syqb88+&xTL$Vnh$k{gu
zVxC<$Wx&_ZJjrM?VVz;s&ShDQL)Kv~NYv5dU}A%X$a<wKK?@%5{g(x3zokp!i0UZT
z_|98hN@aXI%ra3eW036Xi3pTvDz>u5kAe%gXj!A`X63FmnyxnqaZy+#Ay>~yYNChR
zouW<W&#DHB+doJLHbcEs{w8|R61C?}LL7|761mI#61RH@hRfN1ZRyr&6SU;d%PtP|
zDLu|l5Z$jbZGeuK_q7C*^<RfA_SXjaXY^`yohuvrozWCK!*4gFV%8e*@HB9!aCzxs
zAzWO7<CNK_V4R<`t&Z<e+FR_ofid*Pq*rlati5AoCW1^>R&Qe6*{p}{!9tLW`rUAE
zD9tSAX>lX?Xx!vQ1ul>MB82FrW_GjAb&e*Gj%kq{1JUXBQd+1VDXqV|(S-Gvd^P|~
zARIMx_QMvHlqI9d6|C2);=5)gb{7!cPwfQ^IN3Z#CL-~9h&r@!t~06X*7C_ZcdQXy
zE%-yjzyxek5jV3}q5M#$!Yoa6)$+^PC+OOes_Pp_X$=}Ky_L#`@xvz4w%OCZ_mBMh
zlh$&tqnr=<5*6b)?K~_hm|)jVF{k{+k;0_cEu!IZ$)9V}3G<R2O-8c?cB;_d7WN;_
z+U+cfuZj*!;3oC5eVmWjwl@U}zn0gO$`3o0F!s1_YpzbX^R#$7f6SV5*?Vw&Klt&<
z_&11R1}2n~LV5V41k1@LCb(vI93F~gP6)OOEfo(>!q1;StCK|iS65<Zcd*TfYPDs(
zyl!V@gl9zByn7#yY@Mv1w4RaSoYgmNkx^!@%!RVO1FX<2<QWSsOrjMm8~?b}Di*a>
zQM<{ydb3v^&G{#%&!E`J({wEFe6rQm)$Nx>|LjDT4TNrok2Wc$HN+kd56f0olO+m=
z6bsJBUu!g}IyTF^V$|&8hKo|~WcoAs6pu(>**fs?gccLlH0Al3)xQ+{YsiJ;QG_x|
z1Lk9LaixTWIA4<#R!NU?)|7L><{x)qWa)P*+p#%b!M8W@sTBOTOMT1)A)}YtuiB;Z
z3gds3oZTTd3(OOBbalPrsa`9w?CR~Ur4}G-xe-ID!TBwpJz&zsE}TVixkng@q_+6f
z!LTeYF$O8FQMY0PEkP3<z3lVnN53xf;XSpVRdT7Rs!zvEF_du<gy5#9F0}Pb2HsgR
z<F4Sfx<AkJhAO+h_(YX;O=y=+R!?Ol(y#TdE*X3}1g!t-;(dByMSUIbwP>p%x_B{e
zCT(`@sB-Qslp(Kle&LpB!pfn&hKDanqKey<FI>W{72+Lye5q+y#9UBRCZ-fpiF+FY
zajktxby@pSYg|bB&hrTDXLWPuJuf!$dmfFA8jtM#N%$jcgt)jkJxfc^^5HB-EH%!G
zZ-XYUd3Xq+${NQ;I=r!9FXur1$*~QGv*Q5Hv_m_4dpO4IC7syI8lqFH+obgBal73X
zFOWBdWQajCW#QGGNd(UV#1^&M>wB1p6M4b3C#QXbB8jwz0*j9O-P7e@bDDLyocCRG
z#vCp@4qJOe)gRW^Ia=KV>oZ}rv6y#zwxik#+QkQ%Z*@#Oihjk_=uqy{bg*Z|zV&3r
z(1jA4F^A!g92zd?&aO)PgDj=x{Oap#4{)7~OibAa2zbWjg!NCC>}W~pw7)MdDIbxf
zfBx(N7K}75YMFBqmIRFa<`ks1sdgRBaLf$Yn7X6hWH-~wxa#ffq^4OLDY1XGTeIoK
zpl7D9{?=CH;^*nD04;6M-@kN%*?6(>V~Oq3&gdS3lwY!lvW@PuW6m9Oz1^yZvv(hY
z73#R5W9tQWha%?s@02B2K_NkY%GV3Fnpo%8W$l+a$0WQOnx-Ey$7~o4qWs2PpX8R*
zrpQ^JWj_tsKeE<X#6QF25#vbItQG&pE39nBid(je6N=-yJTq#pS!VA{??U>t-Xkd&
zPZrXZ_HoULBsO;XN5T(5ixnJpXl(Z%9DCEbT{3N#D;{1?LS@eS9OZml2t*E6ra}|S
zPnlHF9Z*nE)Y@FFt@NA1ipnGSiG+V<Ee#2k5_+ndNPkTH%K0rC-K86*+#+5Ssfpr(
z#Uoq-6#3)#U;UrIz-_8BoFAxzk@McIm9y>QKXBlj&)E0dLwqw+{utg5zHTsCaqSLv
z{6+h%7j{s=vF$5*CM$bfG7+SSrr9_NEuJ_z)9*vvS)9!uUtAt9X*q`j*>}ei5Bi<E
zI5*-Hf0Q0NRsu9l?(-nb{Ht&~Yi|bmY{ilt_E_m|*mJi9ym&s%LjQ%K<^1RFw&=?2
z&sZ(yvPTXQqWkq0Rj?TGQEUz^g_rpw_sxdx{`k)Uw1td*8^n@eY<JkfK1V{_@Oz`%
zWY|#G-wy1X=t!`<DVH#@uN(17W?^AAO~^Kn=!RI>iv-I3OgJ}EEnF#Edl>?Nt~eJV
z(s*%Wca6Sp*&Cn157YGUxuvV`T=7FhZ#DNZy!-AxT)bl;#9u>%T11H~qO`6x-M&`E
z=v_?8<?_}ijjY#H_yfy+kWOmfqudA2-?M>wLl=8MqoI8FEjx;QR_B@+{Vu?YXRNK1
zzZa=SES)6ZM)yE9k65R51h4vNG{4>E<gdauzq-SE5SU`G7<S07SGdhesl0vJ0RW1~
zPx%9`m4J2U<L=I$5<H1_R4*;e)@>Md)7h1T#XRN^fsp`NlkJ7@JAU|Nmz|Wl%+`sb
zO{BGc51dj_v35ColYx`gY21~!7PF=kH*Yqby+RiphdE0R?D@UzsfO1qGEAqFSGx-d
z0Lnz<yFh$eAKy@jV?AuVgsYfyh;w&5X_}~-iM@^(;*rih<nb$e#0-)5fLjt)hxD|0
zu^I8>T))@Q;bIsSU})e1YSj<v#0z$srf)sY?w|&H*C)Dve9Jcl@!F?J6LP6hode+V
z_M6${Iob{yY586hdoWvt&~QX@OtTWEzcU@w)zg#9R$ZElv5fFn&J=b|7pMeJ*EVxB
zIBfBLFkX%Hod^^go`l7!@vyAdlDFbI-}SaMh3+!E%>+EN<AdbV5l28oem|{KjcB^z
z@fHhYqTU46)HZdsi-~&abwEtrqXrf%qp|b3G^Nhrax%}ab%KEE@xb^4$?DIKJa}Nv
zy^HD#C}42BxE}e9*{;oN{{?|XM;JOd1PhU7U1!f5nH)2NB11_pf}hFuH|tXAuHBK4
zGtn1maw*5m3F`od|GA?T*kjCFKF>1b>!VXTmx#YVsX+cUYre#j&G15(kkjtbGx}s3
z#uv(BG1ki-<Q{GqL&wMX=Th18uPYSjcez{V_Qq*h9P3X5@>;RM9ZI2LZ8lIb9CNu|
zUkshC{25Oi-Fax=G?M?0y3Rfc<Ie3)$gzv-o3TE}?iwhNf@eGSY@p+D1p~tt8tzwd
z7XFW_(DW8MkXMY}6q|~wEaj8D9rj9N*w#N%;p^=0)?cXzGnvWJh_kQJ;pMT#E!`@A
zc?uR0dkn`U27ejiU?$+IHAM{Y_9hNdzDFzidZr3ZHsk`{eBVJ8ekAJQ=bo;;+EoZ%
zL?FtocVCKfo@H+LrcPahE|p-0JXVbrLA#1{qX`w2Baf#O0nZvL`O*j?*sjKu`umqe
zeh(yOul?9Kjl=r5+=2MmrZBw%I?Uh<YcV%|sago-P?l0Tvcddq-jw~@r@~%v8Uj}e
z7wsC3jr~R2gt;HxFq|D;veA4O7O;dy6FoefUBrKg;OM96c!b|?+7&2uUt672N4vX&
znNdwyhoHNfwt<iK6|-I7s!atve^j;XL5u4&EHFwWjO5QP+yAus7v06cCYf_ND&xda
z@ec^X_ykI4xaQD%dfvM3O`iZf3V0xg`P>=oI&^Z~e#rWTIV-`px`H4yma69!X3P{6
zMl~0cd#q-Il`Wo$Ty+T`Rj3aejQu*RllL(h`c=BUWhQ!O3wQLJTFq9p7n)$uyS;Rl
z^2L!q45|WrK3ka_`iv{{%=`ijK|xnu`n^aY-#IHu2=N5K^oL_SY=$w+E^d<&og|bj
z@ATibwDb^pefQz$qk5YE!b)MIm)_^3aKB0Xi9~C4sDMKK4WWHdgi2_=LZ2nd>kV}x
z)^gvzXRGMH9%&!KK?8+@%2NCLpojlGy^V*JbBdA8)nkx_`Rb)wT4GQ2W%c~0g)yhL
z8z8t*-<3CL;=0qEl0E$EjU-HZ^*WC$ah?qcX>V<!5-b$l5Cr6@^GBxmcWxw~j8fwM
zx&xfr0A8I%qCKQgdTo!w^_NG+HBEVFlAPC9TSQt1xb%)Z`a3H@qQ;`2#oET8|A-_(
z@^L2irtm2FQWc^eLvXj>HwyM9xuBsb{qu9M+!&l=HZ+~9-hkcrYHZl~e3mv)yYjh|
zj1yw?vr-eLbzzU`P~E*Ixz}Hbji>L>8b*fT@akn$CtP)@1|s9IWwtYJu-EM&`=%#P
zl9JT~efa~zYXva9F2Ej<aE@qdEn8k#1C~!%aEX_u(M2(wZds-H(aX5^%O8nh4G_Mh
z!0R&-tUG)sKkg;VY}HkRY5Zs26t4`Cmg8$>RDFsK8Hd8UZkU%2rpmL#%y)Z%Z=e5T
zM@4?{z26Ta3-Pi>lRlCA8z5COrSLR3-MN#^E!tD<8x2|+r!-30il{_fzVx1AhvU}8
zoE4r3g&36Lft;?xoU$p14sPqVh<scaw1ig@tIqWxpuBVvhRb5J^7!`B64&>N<2hDQ
zLTtEqTh@r#wT4sb_yP<d73@uBT=RaSc{CsAVz{WT!Rn3X#CfKKVc+UqGhgTSOA!hF
zUJ|fKo99dvHQw!~xi9UL$!ed8aK876n%=|g+zPZc8t9p}DdH6)JF3U_?dosEQPI~d
z21vlh!nR^z%kP#n>bZ0no*M0~8CvLEVZmg#+ij>W{xAW{pk#M4g-o{IKgBzt`lsnZ
z9!VWU9jBxySMwLdeIll%v>ERo()R7@Kj;x2f$uk{+c1?uPouPI^k%AJ1pXyc-L_75
zxucC9DDL<tQ+O?++GP9`iD1tXP-1PIjv8B=%jLOd`!`uFT3cje<iqnRw-43y)0+lh
zs|)K(zOu7zs*^wOau>hbjt`nh7nYpmC|a@Sapd_|&!myDAJXjimm*8duTNQueXQPj
za#*Aw9)e89!$wdl8c+%4aU!`jDfv_0Qe+Di(|2K9jOqX2TOOZ1N&vnI8Muoc_KN&%
zT-JUNaMaRSm|7b<;q#|L1~#R3o#M-*Qqn1T-e74;M~Mv6G!LWtl*F#cBmN`x5&m=Q
zihh6~0Qwwt=T`i^5|f;V96B?dqewGTv$?dUbc!(dMJRm~;45s7y-SEmsk)R@m(Ddn
z9AHZ*J+qA`Ie}jJnsP3DhQ`r%>RA#XD-2ohvG(>%h@D>2Y>V6SZ~qys`L%WDeN<sl
zVWwB4lQT~Ub9cA=l)J84Auc6H;Ya-yLkaYI#Ybp{^xsQ@=Rs0|6b<Xs64UZXZ(Nwc
z*3~yu&#I0`-hT!mvZ;(nkR(o3Hn*`k{gvlU2tI<RVHosob=AoJazdsriBnL>;xQQJ
z+etI~C*~4h{RQ|Tl|3EUkS@uO79`s6Q7^Jaal9{-Chp4jFKNQS!+T6T@!sJ*Rd%PX
zs_GPYXCjg8d?ph$<OI@_M@7zdNI7el^t==t`P@;VZ_J_-|Dwte&2?qiw|@u?OZ)Yb
z)Sc29C@JZH0TdS6)7aD5Ye*`Z;k147T*Z=<(^usj3vOM0fVltbo&|;WO*2DZM`E9@
z+p`SBsVi|Rg9UpH@O)1>xL^G=JemVbMOQ(mm<}D>lv-yqK_%C|F35tYX@hOg{%Z63
zp?UXF04I6E_ASr<(e#x8RX@+yC?#Fer6M5RT>=8q-QC?S4Fb|C-6-8%(%s$NU6*cn
zmhbQXyzrGbxck|iIp@sTnX8RGRsKwwfH>pOgX1GHVD1xNqqI9KcC(%MZn@Qf6D~?d
z8D+ImTYqAItF?LrKa?I+9Do$X%aiyN6)9%wU`ZJGd{#UY9<ZSL6VIh_upT%19Ri2J
zr0x74hw?8G;F|8hSzvmZ*OAbLdu=*-p+ez{ekB|6{5DGCQZgccJg@xPg@TF}74xh$
zAG7!)J@tq+JZXGr&kIWZo+tSKA1dg=P#F2hS!T^OBWs=EvVegdYRLK)o#X)*3Q)4r
z{<kOi3)8f#J@}<^KTF{)>71PWO{Q0^5aGx=cZB2!ZiA!eD(=Lttezr5(rL*v7JfP$
z?Eu+x(9MIBO1wO;nqty|31)=CtK6j^;w7A0fN+kAj?);l8jy^@UzbR!lhVCmxFVKK
zUr%WU{t_<S#;CH5LTcqz6eNw}&VvWLe8Wq3uEAvfNNvJ6D7}MesPwoFy9?`x#A>9w
z6xMMAdQ5u@=&|<=H@64kR=N#8sVbk^mljGpXCFxkM=IP&(cN8W?oynbOt6y?@O>^n
zsrUT?=wfY8@!K!iNza~ZTfJ!yS!(MqgRG-N7#ef_VN_;H682Y41l<9BM2pS-Anh?;
z4cYEp&$=mPS23O*OvhwOy0kuh#M@UXhlyzQ`v|q~Y^hO>_g@W*E&TuJ(s=Rn1kr3w
zO0<tpsO^E6eEz@+5A(%nX%?kc7bqd5EO`~YA4sy^1JfzAShcB;BhLKAmf9ZJcgjnV
zvbH<M4>pS;)&Ph-8;^0nAbwIvymE9p;WlYL{{(ycZ678)YbaA!Z%^(Z&ffX8=(tNK
zt<cp&@9EUZS<i!cr&i100Vy4QIp<XU$Q$qHHR$1!B;k!_TLTm2tI`}?31p-X51rh=
zRc%f(+R$0<4tf)EKE0<;;q?tH=6MmIZIgR!v>~PVFSVMiAJIg&hOaT+d$@TxL8U*{
zA+6jEg+<sJZ}=AGGL-*P<!ZRJ)IfeJTkL_hGk6s9WhT>D1k;5yJ5yx44Oj{Ee-@sl
zR8K5%m(^l5TKP_m;dwU<KO~l7;$n5Pl#b9cc~87Qig|r<_G$Ozbq=<g6&)k25VKv;
zQK&~W{E2J9%Z@#kjn(p3d9jgonNJwM1A`+exz3O&IlNVlk617}-e}#TLdmu$JRO}l
zH*h=^ymh6khKP`5t2s5S+q~DL3%?G$J_t}5dcZGr*x!v*xW>ZymQ=dp@!Q{%?@08C
z6cm2GEr3@>Ouv2BDP^(ie@}bVG~0Bls`99Hwz09V9=t5v^LIpiXZ1Hwv1@Hv<Wn9R
zX;EoOkXpg!?U@|>J6kTA`Rhc$I}M2vZFP4dRPY2|q2cI<W>~}X`sz|jv#m|8{D^9w
zl&r~JRvQ4J@d*<&C(kgg`O>;S@o<z%HeS8qtNRXc-X8#^E(6m0o0{^}0g=`}zOcue
zYJDr<R(y37)(|*23CW@yynv^nvyAubyK0eWM&(VO$`L%l>8D-2)!&k3rXw}ch0W;6
z($AZW)x3~<S+l=6KVQG}xRRF!?2Mu1o-D@%2d4DVH_QV!#(iNh>rIUh$Ip}tw>y63
zNKra&Rl^)MSq%r?^A<N)C2F@3w4{X?uMK)s#x~r5KfZIM5rXOQiidKUW4+)3|4vR7
z4{IY`Upa6#PB+#&nTlC1EfupgXsaOQf4AOW4^Z{(8D*W@4JJ-i5}8k<-<X_S&LfDg
zvyUBbFgpUwC;zf|k>aba-(-3`8LY&ufK0spXNE&hU$q8yZPJ47NX$DJY{#!Z5KK>9
zPhvQm&67#0%vOS5d!a6vqEV3VTP0pWquV?1!Q-?*wf4sIcehs%MV02+pIy2SKJ9WL
z!`?y)7%o2bAYW2ULj=9#3e_N62Ou%Wb2gq#QZm+7e!na)qg61km%}l?h3LQO+#D~H
zm3+9bq4tRU-<eMNzcan5S&DD#c}564AkdOJo2~1(z2Z?c-q4w#*^_F;(?rcjv9v#c
zM~8deH}J&Up_K2(GxQEi1pCpCc9%T7y8HK;hX>PuY_{<%*dUmA=^QtHn51jD5T6dt
zOYV4y)2%k6D+7&@0GalF?GdwQewOR>c;*1~?b-mO!s@X`AfM?yxeb=<?K?vLlP{h5
zE%xUH1J5QULWMNm_Eec|t+tNKuo)eB?g<NVm9qH~BDrJdOx0zkGeP9pIK?Z^HdsJI
z?v)5Jdc1J?ewZ57_EY!J4KOgk#`N#6g&E&&j{TX*d=b1oxk3p{?{x(*2;D{Ys(D(g
zA6{w9wxX+oJ0)u6X5T`Dk9djQJ%`<@a=35EU-LJ;v9HzbMUg{n^`nB!<8vlQCi{R^
zDz~1A0=*b7OIU~d^*=Q6H>vuBDcsz#hWXql%HDoOsD$JZI(@q1)M6ZsW{2BMnzmOa
z;{9>|`x4lakGG(I<u?@XNc-rWIJMCRQha3C!{r;AJY`6XY8Q0GRS+4~;d0q`LD^>e
zh4&3lI>?^~+DPMO+cQ2!d~5378)D(AF(n3>Sxp01_vh8AW&dE8*p1b?-iHxuLqMQT
zZP?LIwgbu>ZGVIW@_|+mAGHO<)kX;ngj}FU>y$Z`6YDVv8-AOL2Bo2G-c6lz-qy4=
z?D--RM}-6uN8C$&Fgv;j-G^gxyK!~x6bq*YwXxeL<G<NGb$X-H<M-@Iv*a&b*)Q(9
zcC`;f3;U&ociR0Ow%*z=2e*SL{3W_tDJdWinky=w#N;dA@<-umhwL0gr<L)?H6_QN
zJ|wUF9eexC-bI<gKl1d1>eI{{eBmW-_X(ey=d&-LTLQJU_U}tOCSmoTM?!E=D5MsB
zqorv*o$XcY?*Adl$<O~+H1^H0_F(qd;!=5HO&u!qes{5IAcRKWd^9_MsmZuMHDW&Y
z$2PyPe<ez21eW?yH+jxi7=e#ha>Ep%Bu5JYfr5!1Qw(<``FMC{$9^560!nIbY<s$E
zOW;{SEZBVqg1xso*NOoPImPVpm@F76$`-V`+2B!mJcC`Md;)1kx0d)|cfAW7%)j@(
zW8^rEh_J^iP`)51(y$4S)edFC6nmw&J`zgMLSE-pYw&B<MIBk<o(<I6Mw9p?DiR9p
zfNG&F&+=3yd#Ut&U^iWZ3a5a)BZ#4ndx!FN%zmQN*l0$D0VeE5aO93_n}P(j)VFsK
zv;BbqLV@+wb~>V4W1hREkTRar8be~shH=Anxrh7exWt$K!{6%|#Q)=9b;2)F&;6Av
zFViuZKdQ~T7K9)8%zB6V+QHR9z_Y6Mn9*0h4UzV?+x+P*BeX^$&`_rTiD+g7TVDS~
zOzU5EYcd|GIOs?RxN8OzQ|V-h)_&E_*&p2y#1^%jQxI`^&kl{?B2!xT40*WRop=zK
zFpRP&4J}W?!^EsLoDXF?wuuLnX6fEy=Nzuo%QPSoi@n#Jfdl$4CbkPk=9pT}!HdP?
zf`mIRLpXV^$`f@4OveE&X{tp7VypQsld24N^EggVEmiUtzYJkw+aqL-<?-s$+*=m@
z!G_7k#P%+qe9hpS7y-*iiIRC<_oTJ#6KN?X^fgMFf`bfRQjCzuQ!8(#iHbzp%UJNg
zjz^0#|AUE(`YP;KPDj@tL1osxRx$q<=#LXi7@OFn*(+#n`|CHJ;-Po{QtlNyqG3$U
z=SiQ$grtdo=TI81C2Bl@&(yrb5~R@*f%V0NAb=T}=IHAiYi<K8<E~gC>LYcnkNwD3
zD$nM5)syq;j}&nxtosY0q7`;YFL|Ft?tk0(ON4=iP5xpplM=t0%PT08?b=~D`bAl9
z8<T9;%#prNE&|3FSni25JN{oK!okJr;BOJoWv&hkt|`=fIg7|3?STidwOsQ~SJqRC
z6BMxX*ffYz_<&{XTZU=Q4D0P(<KLdIg)D7tbI|anBMt(hCa$};8qu33M$vY3Nkv)6
zgI2V5#YcXL*0<@D|78Vbk^eNe?|JgzgJKBfxnj|9QXL%!7Y-@ujvcaHqtooNXCL2E
z@^goCrAKbHhd^&`o2)pr1jf=8gVu)Gu0EaZ<2h1HNDFdVUE1Aaxz34>x1|q@QWc7Z
zW#z7-nRNGqgNkCVWzIvMU_(;yupH&wgRe4MFh&OoD>10?VGm0yrGQ$_?kYvi^CebW
zaLvzT(4XOzOv&<#q%J5c)!L-e#N0*uZ#%Q_`u1Dx*3LeHPIFWxVH5_26Mkl!+r;Cs
zv?t_bCW&3vxo5J#fll{7M%4d5JrVpi0=ARy64~u|39m(|muFZCp@y>z^^L=#bSE^A
zYigW3fu4k+I{XXA1A}np4ttpE?1BR7VPKTw?dmf>-cXu!3ssZkD=q1^1tMCupK)la
z6}=MD#w&3GK*k^k?CEDuTDpdX)-#Uw?7wgPzf9Q`8?5`nnqoihKe&Vk`7<+o$05ZU
z1~B7A-`|4gX1mc0$V5d9M^6?TkNs3r^z^IP5_i<2_nenwFWJY$C)5Jh@sWAO=YhyH
z;EGy+QCZl~G8<c>8N<L)ybiN1XUlQ6;tc*yfIZEWV(^Qpfr^0^Am(K?jvpSiJ{WBV
zD8sJk;Q1L+LEY3IDwN|*ZLXInT0PiNnU9J0qa8ZQExfj>>Sqcz#qy#E-(}G!(y9yO
zS0u^3B_lqu8GqRpZW;g6bkNyCBbrzuG@Q0UAel{v(I%qCF>@r^Bs8CJ|6E=vx8O`F
z(HZLqWU_b{Sx?PBOCPm(M1i4+vtjD5%OehX<j}~7;dnG%NS%WWm6pIJa@4<o(ypOE
zs{o8)<JOa3Z;2%J=Nhl;owh%ncvqHm-W2l)go_q#SgbnC^WMRlOw(g?=bZEGK%tLX
zf6X^!NaG(gwMBi;Muoa4tsleTQ~~ceGz6`9QASR(kuGnQCAn?>|Gfa3mA;^s?*?Q|
z&VO#oAU(O5MxJbk)dv4cjC-xK%{Y(5Jxe39>?1iCFm#Hi@}+!jWKN6QBZf-%8E|>a
z^71JQ6j1dkH(VMidygur1g1d!GWs!6Xy@^J-(j^!`Y@pyE?m^i6pf58<JaZTh$@*K
z8<NVGUrj82El<}h@}}*(9G3{{i~ltPcK!Yz_Ie|~@e)3>TN^$R&Cf1t(9%l(CW#aI
zqmYkxHTWD#%&ymIS)_YKhwCzuDOd|eSaVR%*dPaaMm}-nf7%(pi<iNl5T`nx-(T&x
z8|}Qif}Wz@YTiT9H+?<$uf5#sAo20kvGJ<yufc}q9Ooz~`&(%}@%5;-NamX2#2vUn
zeUK@k++SorY)UQnWDU^MK4tn{(p3(L*xSr95%j@zn^H||^5Ajd&c2zM!RK%8onoPM
zvsHlae&E8#LYi>cap%~?-HeV;4z3U6DnOK_YpGnP7p*W`-`O<YYG>qd8-r)<+B`OU
zE$dDlH715z@3#F$apky`MRG;dQp2+_5^lq?^4u1?TOczd9*Z+EhM`Xjnid&8V~fW6
zO%2_%vnKKD_#=}Z)xubz8%^)gq0=j9<<G=OH|eAoj()QA4u><T2^xuGj^kxKi533~
ztX++-|7S1pcu78CS6=U!;Z1Z@nzwaT>fvr~K4k6f@ol<$9<b54<fUd3cJxRAGg(jM
zTtHvn*H8D};i6z|LuMuK^MM`@oOhb*kZi|g@(a(5@u0QjZs2-*F3iR9oQImEb}6|}
zBx{(7<9s+RKQXjE0G1^Id;ROi3KA6H5Z;+<uzL&DL@5v|s`{V>CG6+7RXVI3+7MOO
zLFgRHaQu{<ifVEPy73w?Xqirk4fMxmeroV<^o)y2FMR8oe)c`yE9~V}(FmSfXedQL
zc->>ISybY98%ubK#atBUv89lxR{SmD$UL3#+bd74XMBU2B-3DhaywI^^v-))Xm%{W
zb*uxPHnj+6!Ul(ga=&A}Iet7bm7MpoP%!_OfKY%dAmL_DL5tU-!|$WpiT`+V4Xb@U
z;~s5_WG)cAx|<)h^OFI14rJGc`;CL!zcR7#%d_diU&`Bt!o!DP=$5#$ysE~uzo?p=
zUKCtJ>zH3Lp?Y@fL2Hyaw)_?rR|6>;@7G&4Wr*dRoNfzrH7WU$IR^Yjr+}$J1^l^i
z#05eIp}@%(P^Tj9xyqJLS0RrSsNaXOfB#FObP3lYwp6u31@>eHUXz$|hUmkU&=)Rt
z9AoS{D-Lme`_X<JQIB|ALq#^JLNpu#^5N<_%;4+9Gfo}4{njoiw+-aZ&A8^Ma8EY@
zn5&Nl8^^7XXF~YwLH`y-mqD$-Kdn%PmV%WPbPj5rJwy~!uk`K*WATvs<stuF9nXXM
zUx@{DF(`Lj&#Yg76fC{*S6G|FDx*D{ZvTbxW<;fV{c%yeE=Ih=gxLLg0VN?Jt~sem
zqj+s-ujNHs-#b;kIoOCJ(qfM6EPI!IvfTn)iFV#SRw_-JQ3Js~9c^>bQQI~hLxnaW
z_ct3o=9>rSF>_u=M*|+zZ=>o0DiQL<%VCJ;wbn-|iW|;{;|oCxh0V^DbjVEbP@U%u
z2T*Wr2}~I`v4Cfa*fFU5GPxNPz$T98mM^{?{8E_Ux|mSe>JF>!XQ#Pnv8t93P|W-V
zN+Gd69`ekYr@B$UWV>)^oLK#Oco}i<I4j&dLWHYxxYH7z3HYrKI$~+vo<0tu1(aEB
zaoyRL*F(hKhV#GbQLwvkHlAD;SL2SeSMgvALkihEd#YSDK@?u3#>omJyV1n*p$Z-K
zGA&>id{J9}Jm=MrG{V9Hrqh(%^8A#Xk@4<tX6Cax)jybcdhLRen3w!)AN9ig!gW#4
z0ZJ}^EKO&8!hF^8#kVkD!IK&9J-@6v5h5s^F`He$h&Ep^aYt8NJF``m8Jxuz-X7Gd
z(?0Jh3SA9T8l89!7NwRuTnJ{e00T!7aMg+|yiC!+T~5@_f&GJeRCkJf{X~yVQ}rq#
z?@uku14&=EKufN2us1m>8t^+Mwh(RiFmyWt(}D+XB)A1_aI_v=AA#gPXAWd8!Sqx(
z-+)TW5vhcb%O6DwisFp~!tJqm#yNj|$U_~xQuGBm<6TnQ<@7!m|Eh;qCco2yF=y%<
z9hKNa)0BVI*0YCs&m%Mb9Po>LxU9OYc73l4YmUW(FBsRF%*U9uC>#!+m@}E06z1bL
z6|BzZY?U!jH_E%ldrlMf*xvHcz(dYIi@O^|gY2bmu>0$I;$?axqbCn^JPYyuTj~CS
zxYs9LVYqwCO*~^`V^bJ9FH#8_I+7$|cMAuoP)HjbB*6j?X4)=9Qbgnf6lqHixTuuS
z+|jePsWDvocO?V#&EdYY22)TXiwk#N<T?6OE+Lt3o@01oxD4|y#9vC=3|~wi+p`Ev
z92)3VHHC6EBJ;EDC%?4n(Uo%6fD3nR;?{%H-V6zJK;61|iU9WecWm3Qp$Lj2w_i1x
z&luu0!9}%oPNwKd=&tU6gJ5ro8b|d=AfXiZ@+ZrTSF5G$$LY}G8awiRQmP-DuW|4r
za+2%uuc8sp>iTH7aV)oP7a=1#BHm$kuY?x)+#DRnBa7+CFQk40fxFijo;;aLQ6c-c
zdblHB|4`6&xk^a7snm6jT}R?nAL9q2-<PiM!ND6lJMf?N^xh36^}&at6;Lxka&mIU
z3zTV?Sy;lnr~d)<v(_def3K-r=)c@zK_a^n6rRW##)M`1D?VBJ<^Fp$9FJJ7!_YKc
zM0GJ4$R+wW2~?X~+vvDhe*6sZVz{zaR+98K*pw}w(M&9|=>(1IKJ_G~ChBazNLD~j
z&ou7w26hZE!W;Bc=x85@BlIwxw2-c!_!=cCZ&Rn-=I+WCSh<(glP%#_i<sTBWY#0E
z95oh8JHDYPNe98HaH-C`Clua&>PAPrxHam`Hu%J-w`CPJyqI;yrJQ*&OE^<&Nj{Pq
zfMBNl%=?0ZmF92Hchu`0$QNp?HwF_~j7&`pO-;p?{yCdZ6=0`wJOA~%znUyiMp8!E
zJJ3l5g8sLr<_o{|X4Ng&WiH3<x7e-+lldKf(C~JSg?cB{|82SjMkcnNsY;pZ6ERT0
zSv^GXRnaVumxc=v+}kRo@p*rVL$OcDghX9c4?y?$JgVeD|Bw`IZ$r6RatY!x=U$9q
z|41wcr81SkN%!rfvfx4(;f(g!hK>wDDvqXXPydd6@Fp%vt~hzAEp~=tbh>8wIWR)5
zlkV;9@jl&bSS@=VaKkV$F?Du!+O4*}G8sujX{|6Gz~ge*OqEgm(U8-R)adhcYdUrm
zOQR}b|6jLyG}da32#w6PpJ`~{I)-3tjrfRUf>xD&3<L~%mW}x}P*qhcK2_eR>T`<k
zv00hexCckAo2U$PG+(7F!HUcg_81*0^r^g_`p8Eg7{(*xa0!7`$o$<aKG5DjhvQ*u
z`tXmft<(}rik4bHlAVDic(lp%&5~o#<`1`u-^`4aOg-I8e)C6!3N5PVKEodPZzH6O
z{CjpqqbzB1lWy^p*=S5PG3mu<c7_WuZR0DD-Zem3rhDYp(Sc!P@;W*suU@^<s<T_O
z$yM06xw(lkZ=$h(E<F41=!e;GSkG4#bo@7e0yJAGIs(1fd>!#e7vU5xA(|hpegr-|
zH-8!lnrNMN%dWexgGY9SS7b|(*I?mnxL-Uqyy2j1AuN779M`1WX}v|b<_Tv}3<t_A
z{gf(!%<GDCk4xPCs)s#lz&6<2NyGG=7-6lm|1n`sWArvV7y)kzt2n=8@xE_l%<(4P
znkb<qaq;4CA-ryI4y8v<R#T!7LwoOP{vB<ql|+Q*<ieE|H?K;_G;TdyT};UE;#<wA
z(CSb1Go1JJjz<vKt;O2}RbIKZ^C?-4C~$I(A0QoQB3;_4GClFS-hHEggij%m&xKRe
zba@pOKO!QwCi4{^9<Nst5)+rUiUR8@3mRJwwzfpf@+$HIL=|CmMgHp$TGzv7JrNDP
z5Z_S(lP$+X_a*VsA}7e3esq|`A~7HqPT_pbfr3aXh%W>p`TgwK93|!%dWppQu7o7Y
zl}2FDimsO{F^4Ev_!cSIm)VhvjC4ijDM4*Ig~LE!jG<zE%H4?E^pnU{@M1<Pv0Gw5
zwkXx8ZvS`azRg2PgZuE<lm)kNjB_0@_~P)00c@L_^X!vYkzOV0-R(E2M6DyQt(<?V
zLt)I>apHGe!2l4v{Sddo%YTjCs5wQ)<Z^U{GM1Q}A&~ZZ?^aK{E&ee$`u&WrZhE(w
zRDrSAHbJiU$TrcvHSVE|Z!pbiTSvr><zpJ?u)qSC9HZCc<%~#e6sdssfRPX=RK~_)
ztEi|<7Hh`qcLXi%y$!58D9E6i86xqz82h`hw-=i-^>WDH*puSjj*CgWvUTK@;x2qE
z;=$dv9tZhmo$a=3Q}A}JaWF<`<*k6)tXATw2OE+O7mDfln1wSa@{DWqkyMLHhlz0u
z`eArC=#;+`jag?KJsTTVj@KgB=CX^_6&q0qZ2-AZT+*!zcYc-%DNfxoR>&5OGC>1v
zn<GF}<x48+My{9u&AGn@AgvD<yQUKMwIOV@R&Xa%0mqRNo|fS9evJ!6!&+_pHHiZS
z3sWk&tWW+RBc0#0lvuRjQ4D^a&li<Lqqlf#s#~Ylr)s&OU3KucHMYK~xnwC{1ES5b
zkuZOuEMPHKGiU0rs3<3ApbPPSK07~0Kth7fkWOZgy~}a4vty*7ph!qaK*!I2ux_+l
zdmT+7-99seD=8^CUu!$!y5gl;RimVCWNcjSw5Je+MjF}Bz_Y!*&1yCK^K_%f@o=`B
zxu<V<*rY2IJDi9MQ&(44N?ICh;S8VWC(q}{YZ0zzO*c2UF(wd-3pCDwHMvN=GUykX
zm_n`$F?eZtb#=D!KwRU^dRUQeOS9IFv{ErhU6G^$&AV4|beaX+jW77VldlK(zrds{
z_yBoxNycu%rycmK>_N{<=714GK~mFRIUI}q^IlFVspBWBnPCQj@ch`sag%j9hg)dc
z5MRl}I$#-i?yxT4Z$tcw8fCh5AfWe&vDHmCK@HwYBC&ek5LNe9SXFG?;J5nInHJ&&
zIYU8gdg0DU{LW8|^n$l{rgs6ZRK-wM&JY!Cw>D7LctxZ0=oi!5iOKPD__M7$3-F9<
zoZRW&1VgMWs!ua@YT<J)lSFf~d$I^CTbdh@skROcQ!IT)gxmD8Cc++>YMUDPDg)8i
zb#hdm6>FH;inT^OGRJ=Cl!}$k7&79imFc~TAYfZhwrze-OB)VsuaQJt?cWiwAP~sv
z;ap{Q_rv{3i@l4>_Fitf$>shONEV=VgYeg{Z_c-e$Utw0wgC9(RIdhhsb{rX8)6w5
znU~!1aN0xqJuvWBe7u;G6KgVu?RNEyPRGE&LEP*8mZ*I0^YF;X9F<}X-m_k^m)B}F
zeT!{u<Q*R$=d@pkLBgTcgRHh)94*vNI#Lvf#p1d(H8rU>xX?i{G~awUJv~)!`Y%lX
z@Q92~j(6D5$8~GRQZinYEA;2s|ICtoQjlsmq}Dp!z8oqzkuGnt?oEOy)~~k7_S#3Z
zZTzbgtrR}<IhWFYMOyfN^<||%pZGwK*rL7}*Knl_%E@FPoXVH9fBmN`*MO$Zmhyz~
z6QVv%ON^;#c5kTuxtJh6lyI@2STHT3wXcgv=0o$jh&gL(@4+3=0pB25(y?y_LLJ`2
zEuJyl9dy=S2GqlStzVs%+x6>@o3Yj%;hr~)FhVT(Q{7RhhWARge?gQ;2StM_v$5B{
z(D10}BqTcI_B+F=17Qr!YKM>J=H^vabH8_YcMW=IHL89H2EhG_ja6Q6o-WmuE!AlP
z!-s2|e1=oFknx$bCONT{=XfJT$(lT_!J*61I0FNr+hLbq-u;a9=8fLn#a>ocmRz9<
zUGwvk$8frUtdf#a$U^Kz=l`lRsv=rkJ#9xcvFZs7cQH+MnqR@9S_EixfiaMM{Y9%Y
z-GvEYi|<}ell@eLBK_XF2%wS+@`y^Pu0UWW+Z5BovrcDr^A7Pq?MM0C5{!zCkDSS$
zSc(P)zMx=P03mjGWpiiwiFdd1eD=A6VvE2n(`61YZuV*TL+6x*w_gtfe_#gJ;4O#H
zZ@_<g^vu=rh1HjkU0&Df^LHKTu$=)h95Q<tZx1O}M_OINpvD$9;gu|A_eM`b*Z{gY
zXSQ|M5uTryxCBjXhEmG@zw_Dajj>&l@a8U1^i1UkyYugYx3{-c8gF7oMur6TmzJA#
zB;Z39c)Y-(qN2h>h6AQ|1Y~4d3+O*H2ly-|gVwcC;o+ONJL&MP!^4q)CSYM>E3@)&
zf-4dmJy8>xVrWpP#N3^s*fcgFj+gt&I$oDke|^vh*~6nLWq+rpMzra@6A(xTtyW!*
z7yGbedF78cib8n}4GcJ4kBOi_lq6?l{CRQVgbXJvA8t8Qa(YnOG7ch`!_KDy(Li1M
z7p%e!OZ1MCGI&PEqj=CB+>z8_{jZbhVkWAUZb@S|{(cVayfTK)_Gp*iAz#Q#6XD1v
z(!%+VG<g$sPqW{+E1s6Rc4xX<yE)Cw3kNw}CaGY{>v%cSUYky|y<tB_<>lWo=1sa=
zKDr1?S#ry~UbJ&He%}}hsr>HZVni&yHot}rz*)U<Z!Gjqugs7Vj^cTB-mLfFB;z14
zPN1WU;XB~jqyKuH!1>n0mXcjYQ}ij+Obr*Q_$X$%C6arT@I5>3YsFPhIH;@Y6nv&@
zY%PMx@YfGZ87v%t6m!gRT2r;rTy@N-#x#!|9sP+UpS6fJaIdRCCt<4ht;m}%10}7=
ze=#+LAWruijb;coo^}&@-XGQOUVbwfOlS+oXMwVFb=~<BOtKFC{WTp)0~sNP%f`+Q
zBF*O|EhS~P^pCA4iZmCjcM7+2D4k{vItGSc?e&n4Wv)yrqHj|ZpK6I#+wk$Wh|195
zpcz2+3$?bnAjMcM{fk8tc=(AyCc4OKbGF$hQrq%$XGtNG0z`9y2sn6n_-_}x@=D+7
zq6&J?SFGPpmuO3ZRAdvfwy_Zq6@?}%;N^<@+WV)ndl=mRU_mADyu=a`AJO3EMMFag
z2A*2{pn!&Uv+e=o%>Ti2N647Zl)Xc4exv*?fj-igWlxQJe)QmIt+8S&nm~CG_RW%q
zX#LpO9|u1hE8VxHvyVz=8oNbSr2lrV?r%$RqFlD8jwa06{vv6_!l6>MJuqfBb*sk5
zc1;eO>Kz$~tz3&j&bp&Wc3*rBA#sA~d>sh_l!^_)!e56~`8#i(sHIw9R<r}Fnb4!d
z#8`CGv^f1Jz1>7YL7hy#tNC`D5Ts6-bbgN7=8(uplfK{N;RI}mva+)OT#h8n%*^sn
zyTkDn5B;sVQzO&8@A7w7G;3{|_Pm>PMt+k^LfxDVFxXNhJtb=tZwH|9L%sX>aWluj
z+{T7}Af5p}LnI96<>6*;Ym+F)=GNA75JWHwuI-LxDko*;abu^od91ej6%772M{B(q
zo6s41vLJM1mR3n@j-P(t0^3Nrx+ukZ?|Chnm-UvZgc#xdlflQm(|ofkf+DR$h?NfE
zh)K1jRLfE_UPP9bER-WkowCiV`1YPpTi@EPPw4O5mn7&Lx6Dzwoy+g*i3vG8;Jmq$
zycNX&U2$I*&g}akDPs7yYrYG$G|t)1_E3?9G_zr_L0Mar&OLhYJ8KCMr`SU^-hYgx
zx-~xzsha1`@Ibx~4IUuEm^|e|6|1?5!0>Qnu!zIuhG4prEh&%7;V0^^#|yPl-!Avd
z%_k`CE)P_#ipDa<;{TNj2?`2|SfhTXqQ<?-_qyGRcL!+Hc3)G|tpP%~4-_JiYx0jD
z_001d;eLdMQs3QnhG5An+qc{o^iNDq#_0My#!{wx{swr6*%I`k^1d?%uq!G(ois8s
zQaPzMF98wP4KQ2HSoas%LufG`5Lk061ifN9pOdBMim6vWle8Rh%cqj|AqVp?EhV)z
zlzI7Qj^|^hJIc*<NP4-vqUz#%wM{%U>gImF(&QeM1yGuFTnOrTh{$<t#?J<608Ta?
z59Q5h-)+XmXGfz*2Uk(@Py&CyyRel$SQ(p_l?el&8c^y?uvZ!{r!@h^7ZMh}-A=Ym
z;IJiu0-5n6KR=1b{eq9-3w9ts7A{r+a)g+f*{660-R8Y!pHkB7nS+7J<|;XzWq~l4
z?(S}!C7anY{Y1kUW#ugdMu*L>8dYX(fWWI)7$Nxi`H5U}adR`<uYc)|Bn~SqwA{kN
z#r~lE9Ei>q2W~m;)_-3ATk<>Zw=bxv%Ggxr&2U1cieoo|VdBV{V_!RCE5TJ6;75JA
z&m2jgHfcGIiX@rbkg5|1=G8i!4q|{tFaH&d$FLbzMBS!rl(N&8cMi_$K-U=1RmCWa
zT;KZON1>SY{#~tL=CAly<RKA_{b@szA&3oMbZpp%kNH_`XTorf?lM&3n-X#q^2@3!
z$jDgK)qmtOquB;kaK^<--?l8<vf|Z!JB4AIVTc9|ed}XUfRTFAb{ti)(xdWoY;20)
zWWFU3L-cwXTF3j~;_uQpZ(?la8H@9Az}fc?_aBFk-|l{GVp3BpkZZ!u;hm-qB@?@-
zD#{NEuv^r}FCKzx!dMBBrf-)CJg)R<X=zfa+_B)+TO5P#aGm`|=u!@?z3t8li#E;1
zilv<L<)L=oKVBp@i^(79K2M2(NVu377;W{AJF|`MEMYiw2K$rwFQ1_6eH&<+=AI*Q
z4I0MuugrnKT)g|?ZZlf8z?u-;zoU82?jIY*zp;9Sos~<m?9ejzT<N+ZFRyh3-P1Ba
z-=v!V!9NtG+*Q7u!ea7^bU1}Y5(P!~Ran05zkCYY@aFD>Sp!?M@bcQ(1~sZs|8Wn~
z_@&23rXK|Q>Mpn<P3exDrnBI$N_<T_vg}?GL45QN#b8#Fbd_-Q=GB)^iwuFcTvl4_
zVSLHB(Q!_XX_pRxT<?K_!f6mWsruL(v`TBO-)K;8xv?B3UWw4~I+3~}S-#xm(nE$$
zPq$J!;|9~M*^u~mcAF!>EV<>%<>|IW8NPIrm#;Y9vW0k)un|NARk!%G<roXd{fu`W
zxs&$s3>MOtsbuGKyRXdV<UWZ1>U3Qs>Fq@E5j&tZcI{O+?yFT5vFj`i*Zv_1!==Ba
zvn|eCbNGGNMI5zdf0c8+d0CFV{FADfcx&!O3b)RNS%_}E;1O~S2E^o-`-+<>>4IIe
zkMc-9(GC?JD^hff4FnDz<p0cnjmuY)FEz*J04cb!_v+Cs?ZNeH0S&b$V&sb2n7+oq
zqQeNHf7g78>ip%iK*yq%hIIGzV4<VSnG1klk&uAr>p_8i4yk*5vR!HRaypnE%1}EY
zIwHWvHa?@=lE&U?>DzMHpDGjsyU1-lgz^T!{zNuQoQYgn)4>EL@CNCfdTfiJKCDKt
zFN|-?Y`vnZgRIEjv}cMHPF2|NxzeUzzo#}vU}oXcY2&lILTT$CMlsfBpA*Lw|JN7{
z2GN|4T*V@L9$V|U9au79o?<u|jlgYtKlLAd^nuoVQv;2Q^hrfn0byT}D}}!?Vj{S9
ztodGV`C&xJ>WC(tcU7=|WN0%ts>jarIei?-1<~vXQ72B1Dp-nKtDlnXihyRvXHd7$
z$7EfsWvpZ&!{SIRQ@+w_`2Gv+BWi>w3L=Zy#z*qshuA+g5$yxtele*eGkE3kq|pwE
z7uZZJKY3pWZ1-aF?Necm&l0r@*C_$7HMsn9vDJFBI#&XDB{mSb9F2H3?!KKq5b82t
zbh#EBHi+Eobc&?=k*PCI4nW>T8O2`CXqImCa6EO}uCz$o>o>DH)P?Y|R>k(ed7HOp
zf>EGYPqakXHr9mk9gRTpo?-rP8S{rW^?);l3iazN&vn_a&vt7@^xB!gH}X?32$O#E
zmp5A?v2O9&?z%Q|>Ls)D0&8ByiLXpVHq6Nw?XGSUMyo4TgXkp>N4}l&<8TC5<#EPj
zNg+nxT*V`d6DFw>qW`-$?R44MweWIpad1@TZRRSB0rOM-$Ir}M;aLwzw`=kF<t3f2
zPPsu~TZkfQ`a*Q#({;=97QoXSdtY@FyVS8+SXgqT65B8KCSLvc@k0*)EVc@UKMo(r
zKD-es&KS$P-P>R~TsVVw2qYi6ZxKFko+z(ssMTvI_?SM>u!CN2OpzQnw5~CzzWnG-
zrZO!mqeeDKaDeMGG(N|YW!t{A3M)W`ac7&YF`N7<43qO8hyV9N+*`NdD-DhhY~)Ol
zvdg=|3PMFOKL`%s^-JEBuA*oeFwxcvE+0EJCvhj>QhE(2X6EcylNyU-=t`6J?+I(R
ziK2Eq>ix)NwV!O+if^C)C^17`{@_FEKu7VLDxE>NBri3jDamDj6{k7j>@9&a)s33?
zOkP+RE7N>S7jD}+4C}K|h0_&sNl%LL+0KCT)e6%C{)aekc#J;AUPTE}71bb2m)uOo
z+6?J<N@w;di)z{jeMBGnXL;@*KWDuxv^tH&1+{!XYxA-TxCVB*O@&39wC6|bqSV6z
zQbQDp`jplR`)6_iI<dP}11;^{RW19zu)@GHM!&_(L|Xhy;sBG#_eNjEQXe^vXOg06
zuG;QC?jGozR+maBAQ1I-F3$gMrN^fHsO~fQ|GfZi0sq*<^o?cqmXY1YaCwnGaVZtV
zuEP*qTPlkf3W5OP=IMFbg}XIStXW%9wN%>j)E`5oNL}QmsiDz7z|f-g<dx3vb*y1)
zY}^G|>rij{#s)M5!_D4bPDk_A^W_>E8q8L+jMGKx9U)khH(T+#cmm#@)%W?~`RlyS
z`)}*(>tA+Ol5aChVp@vl36`PIP$4$LvSz01&=Bc}4>hcyZ+4|=ne71SEcZPoKhtZ~
z=l*BBWX<`p#%G>MOP27|N_Y%+E8I!}l8&lUm6@+cAJ(4cpRbvz>Mf~)pOGf{+NIwI
z=-7<5A((XkcH>QA$U$B&TtPK3R^<7u2xd@vo?d;Xs$W>fUq<=;yEOK+#vfmBbPWH<
zRaNa`gY|4*Wa6Tanq+R=Vvg~MW0;s5v3VL|P6nBhM{JdOW%!VhO!Gh)D$Fw@x1TE5
zf|ky5_?gr|9l?As@$NlAwFWoo#OpSR9<%L#(-3{(*<Wx1TE@@Q*yu5`*W%ownyx{u
zCDQe0Bd}`jSfaNo%QYV~^p>4Jjg6XE;i>LhPtH9+Y!~LrJtJ5x?mL3e3o;jj0~2GQ
zjt^g_fBEHh%x$^+5Gaqd(k=C8L#HVUaokb;wBHK+ngwDC*su1gIn5}Z({G-n1X->L
z2v@YfH|4UtBBn`i&GPQid(eRYWyg%e{;^|Xipabec3USvfsVN6x^28<{R~MZlD*33
zeM-Q2ZToXchZstsZoIoS<|1v5PO#p?C0;7+8xSMVDzA<fSis{2$WvBM4t;odSWQFY
zLv-J$MfK*+j^pi_G5_r**3;cV>Ddt+8c9**3J#q{nc3I}I23~K#l>_M)k$Hv49qr*
zoHA*=2%p{DIG>)bQAu9lp&HmpPyje?9M!IvJlvd47OJ92O73-NR%=KD(I1pJKn0aB
zP^fJ(-xX~!fy<G?a=Pf19+0^n&PN11A#L!W8M(fVO{7X8IRitK<IeCY5WUVPWCips
z3!hsl6gv}>lO>}__+A(!sm}uwF}G9e)D(82l&H8kg1o$Z<K2FtYpUwnlgH|afr;@_
ziS0~=pIZR+o2bl6F<12R3m)1c;yn)Q_~Fgt)WfBN37<QXL#;=yDoZ*Z_6LoQUx;jU
zbX;la0v`BA%r|*{Pf-jDRIAaV=<BVQ314xZFMQYt&DdyD*aAV^%IfJAu*u`8_sC^b
zb!&Omt2faQ)+^aJJ;%$u{$_`r7?uO+rX=JTeAE6yD~EKeU|TtPL&&n1O3M++dr48R
z;2b3s;?!)%5*qLk%f}EOL8!y@%Ea)xw@X8&$k7s{wDqs<kBb>}sOTWURS{mql>eHa
zz7KEr#beIi4rwrQ1TX|RQ@WmP8ldX31wj#(5Z4d-={`*hbQ4_dc!c#y5BrnYD7@BD
zb-0#+K4{f^Wi~}H-Ncvv`{8f@Lch_4mk*4CBi91~L#)MUJdT<jaiq!G;SMNFIK_|d
zK4|z5d1<ZbQP4{KteKf<lTyQrS`-pYzYdaYVSoieW(~vTaF!`YHeIIHW(jh2)jdB?
z0MdsMGroz52`4A#Xo+@1!|#TT?ubnB=+N|>R|&lC%o5oz+{eJwG+2s4_>UYqCT2T`
zFi`o^hwXt#G>gkCE9ro9#`-*OL<wLL5rqRD!kk#tC}3)8I$v$s1EhtoQc{h%>8-7w
zfoOJrb^`c@4A8L4%E~?o@QaCw{d3xDKWTZ^SyD8u;Cux(XkV)zw7sh<gTOPW5&&V(
zKT|B~wVti5uCal39=<5Ko1vj$NJvQi!NIWl-QT}|kJWt+XsD?}0m)fpYgldxZbI^h
z>`qv)(msh7=+4<+Jlo|>lxZwz&NX0N?IS{|3_mCQUhhAkMW|ls*L@0*P)F`)%&=y&
z+@m&v2tt?1mw7#U{$4k<py;)z!*C5rP4<q*96e#rtNR#0Pn@Ekhg<N0{NBX2pej*A
zy+4hVEX#Hoq2LKdUMG%GB#GBx#`R#VfNg5w>Aj2128IJBOB=9txlRqmI=co^@_sHf
zel%JlWp-!1(1&nrWZ5?V@XO}-tlU*kFV|5-12PgKjbmCWM|uA^LPN7Qr8Mu=Z&KER
z?XNDD^{nfv^c$lO#cxfvm2K|{>>k}%$2>k{pHV38t3DH4Ui*^Rfd8_;sU_x4{>D+-
z<2#}Y6@kMa8Z;H6zy+i2Ah|YxrxyJ^uo!#fiMf1ytf5woYZh|Y$iY}!<BQF)maLX<
z6-uA2k$;psmJuk((z80J_S|;`;493UJ(FQK^7cr2Y30ynU-T_NpfuXHwzeaE(Uhm>
zg%7qzKyd9HPUQi5H#?__FO_0WnL$@5fk(m8EN6quQGcWRd0Bb+*xYbOM*zUBs&g{n
zS;m@~jc19wy9b6UDGhvy=U8^@NU#IR@Qp~SfRNUtxz>Xaf*DVH20!m>9);R1y>(g1
zXc*@Mj`2=;#B0z774^R6;Z+Ty+w?-2Ov>OIzAUSxO_^7Rr%uX6v3WpIj`e|$VBuIR
zBdHxLME3YA=?zb8<pJ8n&eEFN>@p8C$2Zxmm@KOU7`h9h;#*r{?RT@0RRbYfJUGSY
z@>gPgPh82_fjgcTq}q?RnO^#01md_G@;Eq#`ii%^dlZigkU=Yaj0vAir#+O}G`Il1
zY@_Dd*qBBo%QcPnM1%ejeLq!0QyZW=b{Zp9S(YxdE7vEon9)h%6pD3)Z!`pb&|;Dm
ze}r=i1>4_A@2V(9K9`GW;B6tr!9_~r9l+l+bC!ED=wrdZi&kBh4f~V&v(M<^_Q_xp
zX<c8kz$s#$Cm=W5DMF%JA$0{hv5X2Iaf;Ke{^fDn*=DSqlybLuH{SF(sYoXXFPAuh
zN#OzeU#ruht+LR*SAA9}XZ<tZZS<8|P~UtwqqKX}>N8KraNs+TZ+MMB4T{U>i5Ccd
zI{<8G{c{SbtK$U9Zh2YRJD|4y{E43CwvGUYjOQDzI0Y>N0th(JA3zNRRQj<nFa#|u
zsQ>)=!+X8x_!8Y`7Zz0KWI@vD>+3t5uMPthafjW}_v`EHWmQ$3OG~LHLrEAoIHK3b
zO9&_^#Z~F1>0WU!noG6i%nSMbBODC+%e2qUC612D-CduIfD&w>a$1{`egBq&j>i$3
zj_0WeyUk*6fif7i;>y6q#Ralh6p)n}4LaX|H!w()@_(-W3ug<8X`nvx?$f7T!Qc;X
z+Pz!=TY-XcCW??Uxe<mHBo@Q+;r8FS=Lms05#FEtQ=B)IJ=!KsaB7if=c?uqV0h2Y
z2Z1>$zE8H}=-??={j5njdCy$uu=*k5ewpWwg1b7b;T0M5xB8AP&%dsz?zEVVNeU>5
zGm6=UC;FZBj@7<#k#@AnhLSE5#*N3Ah|+h}-F%~AfI_@?d3L|q`gcJd?^jH8CSfa3
z!4N%WU`F{(hS<MJZ0flvW(7&VBuIMGMY4|2O6I-AOd@>Js&b)3eZ_O56YbIk?VJLJ
ztfZNR>{`8+O-e*ZsdC2f_IqLO^|9XIm(M?j80Xg<FIr;jkR>jL$BnpT_ACUHI~U)P
zw(+<JPj1BiK%3>G@dt<FueLW<+)B)`xJMwXzowT5jQ!)V^e&$vp0j_UdWZT8mnAW%
zsi~>5T}gX^B|rd*z9J?d=wGaNOyqGbY4Za8%ST&$e89rnD$!nDU4;|#!~vylzbT@n
zML?t2yE-jAVxqr46ez-s2H*9eE59chEBRjoW$AkH>-+cbK_|0~ARs}D(3iTghD%X&
zba%t4s;V~eV`IRqjX7POQ9<zrc;HP%soITHJjoCeM?>^FMPjuZ<is|(&nmui$9?P;
zQ&iCUkv$dQ{w1&MU3^TkYPNhLNnE${^f<<q@lQ{t=5`p{vo1IDwpmA2l|r^#1%LZT
z-j53OlwY%v{jE?)^}QtXVpbyrNT9z-c^@vfCjV-YUN<o!imy#w-){@sq%f6_QT$Qd
zfF;vh0u#uVEB-uP2e&JL$&vq{!)W*Mnl9p~b^P$E>C)F|<!kQ87QCc6$t|*QjD4sb
z2k~Y+U1NXT)=F`0QHC_3_Ps)_W$(~B;rL$Z;lG@vZ#2;;Qyv%1N1M2$?`qrXOB&kr
z=jwTw&9=jw+=~TppZNt3pB#<4aut6j2FGR+qox7-(PYAEb1?*^7$_qz(HiItV)0|-
z1W(fR@$r;y!a_pfdphXfdp@gt!-T3C?gj>+GJ7F?bIs*+2&7MnhGzT3M&Hxa2Ir@O
z5O_&ivofWFg-~366nwAb5MZx55;d$9qjSAtfBOxFWDDH{qLQj+D=arn{4I@E>(3Do
zS}8yJLiS_{NuhuGSTge)p75~X*Ci<aRthipN;MVg1=UZ9>%?&;jLCRN&}Vd0jR*#*
z#TQ7*$9*cVDHl~aGHCr*KY5-p@SSp(S@8Oe-G9FUM(~CsI^L|#`|)3X!&?AfC|ITr
zS2T(*$a$;!Phb_bx;3L@2<?5$*$6gFa!Y@0<QwQx%#UpRr8wq_)<zPZ>SV&72o$9W
zG}?8i5+G58G4#Z}j#kMOF;O|cjvkx&`m#_U+pJ2ZtzMp;F+0bD;Mbm_1%_<=_8){P
zIz;j5R926X&aXg68wsKam#IrnS?YiC&WgudQU7;#%zr7T_H*D*swHQ$EtV>M`Uf5|
zfI(BV_R(kS*pAxa#`0$vi>xnbXq@idVx~}^NsM<ohnv<hk*`Q1L7o<jL(~e#>q^d<
ztWOt<(A1k_lVEA%Q*}8$B!$#ho_|{K`KOSSHy}qqXH-yDRa-K>yVPFfN&n!Svb$s|
zUhf&;Kgc_{X);^SvredyF1kkd=6C`zFx&e&|1q`4u&9FnBH#GoT~-%`!&+$tCd$mj
zr+%u|_Gv^h2He-&uvm+^*q8HctlSuyKhp+oTOzS4?`iXyLf#r}e&A63I+JJouB}PP
ze3IvFy?b#&f$;y<YBa!YhvH2;H1=c`ifN=3<=&Z%*5gSrT|sH5WCPS2tHpVxKM2?A
zx79qUNM3x%C!fmk6eJ<}xZ#1>&ap$)1Sp`7&tL&;`mgv@Z;!x$*Q39GeG3SmBn0-r
zU*wghjO(#cyU&9My@O7&h-QWdzVW)~;t1tG6<#}syLCQD(L?D-KhkI&O--i93~;PU
zp<b{WRFq9{abKV3<zpQN4@w2Gm!LZ+t(tvtcnEho-mkGW`n!v~u{4<8ixQmeHc#-Y
zw=n@t0!63zq5*fZKmqmfUj!gue`!C(>BPK>+|HA%-dr4fw_^f_`@|iYS&PJ%i~Q#{
z<4T!=CiQW#ys9>5(}DkOener1ri+gUd$FY(Dcd^+^5oQ&qOHaB9-Am7m9ON?F|ZQj
z23X=@(wnotmnN+ze49_AS_{=LeMBY0lpoGpS~x50;0yj%e0QH?qo$ks?M0S)NVu2s
zeEw<J`}q6w$^`2k<x|d!JOnl*@)v(q^zZNm)-!%ZkB^cDXbgYM)2Vum{zaR_hVa#?
z?UycmRt%zC6#GMb|3Nh&lRt%gNY2)FH9e2m+}HbBNMDd%z?(C4<Ni4YlbCNYBcAX(
zr?2r`liFXu%K9?0h(#~+hr6NVFC+_HS+em;TuHUyo2)dRpbw>jcf62!(oz&SlOh0-
zc%;9cZiY9Tt%Nn5CjNe89z^L@-72%U$cAjIuSSZ7&k~4Va=HK0n+@M3yF(5-T}VPz
zqE9$6yk;agk-gW#0=!XEKoh}(c~wN9jPcNQzc@WJh_5hB_Rl-AC^oDeOrH-b1xpJn
z!vD~c=OsfVe1+Ij`u&RYH;D#}I;vLcTv$UX`MpW%MYaSEOc1#Q^@r-pZaEK!e%`Tv
zF<{c_?U|k;N&gAeahLZBA-AOLE~5Tz){X=zBl6Yv;o((VvPYTm4@<gY6?($Vr<4Xa
zUN=&3o=>mof?UQu1l+0Sijp8KAv+}j?Z|M`zr0sq7nYhuZ?nhQ=6_2}=d(^Xb49r@
zTx5-SkTFIRx{)G)ha*JyV&1I!Ta*s@!R+quda*ke6BnmqNth@Vl$l8eY9y?-%PAm9
zZfq*?kfvk7v{ezXTFS~rvv=z7woQ=@ygB6jH}93z`5-K`X`;pFX};oRhw_DL5skeD
z^@0fUf#(BG>sc}+=~d=e7UeR2{QCA3Z`r7sxhkdX&4E7XdOHgNH_bM9ly|*X%RzA_
zqr>8N1luU-VvbI>Ff6&eT|VB+L0a;+c_Ymu;il*P>7zyB-zoW8`+PC}4%mDyPl_XJ
z09f|)(<&Ta(#Gb7#8o~rmEc6bFY~Ksqt8ySoc+B{ZSO((eDCQ>vB>KHcQ5T=1Jwnt
zkxe57DqF)-=yh3HIkVT#snyK`(SYOkore+MBd8b$eY2cBF{~sz#)m_3gt_;v3TQu>
z>>;g06L=}!-MW{)x>2!1wb~r*61n4!M$+xJ`!@6{CXk%>v4e(N?5u{6n4aA2x$ZqV
zc`@==_NO!`4LWPF>qZ7=>&stGPu!Ba4{75Du@{r=Fg|;{OX5MST#K#TkniwCvZ77z
z(jrc1VMb~CG*c9N*LPaT&%aT#)&#f``kKPFNFKf&9a6sIjziJyaj@#Mi3yZ|^SbrV
z8>4v<S^VX~5_#JiET*WHsAI<-dCvz~9xf(i|K^mw<6Q>`m!a{Nja2&04y1y^?$S0Q
z+O;e5Esa7<^x*RC$ScU=jqK&OkGgqHqK$V<_ciol_59)lA`4^c&k@Q5H>?DqlNTOW
zDEl{)GjpC}Q4?94HeQ4`H6h+)L4@<|BSZT0BEzY(9lz6;^$%mzW7)4u7Je=Z>*z21
zbBDMidrsY4nQ?g$^?v_hL~BL6$8P^QduVeT5un5NJ1V5rWHl;Lj>m%>he|x}aOiiQ
z^Apt+uUrIJ_g3I2Tbzw3vpNRVZi;@fA}Z_hjO%F<Feirp5olI$^ivVOsru9T<VbSV
z4r68kFMUx+i1L-pS|9>0B949|y>-4%#`Ij81xLJZ2?rITUgq65BL|-_LA1<264(;C
zF*TSj!*E?(Rj*t4+25mf^DJ@#VomEgDktH&e=zcPHr)@9AL8z-lI|e3+)uf_%2?Z9
z`<8G$*W6bMqQvQ=F|VA9R|Uqbp7an%P`sStEk`ZX1-4XQR6+!Pi#OZfv@%^}7#r}V
znp+dA;UPB1vdLgbp5u@I;v#_WMrS+4+^)w`!0ux{ku#)SstJ&1g|VbYja5kO)DjsF
zuv1He3g`BAtr&Z;b_0%&j}N_Wv(-L2tYDsTQteS6%f_;>a!SwiNr!mDw=bnf4vK$o
z?-!2Wnk;2$w*0yv;><CD#<oZ)3`UFQ*ZIg7_aK6NC|+uTC9Xr|iT?j+y2`LByLJg8
zAgxkL3P^`^mvkfDjdXW|q|!)tOLuokcXxMp*ID>}=Z8OVZ}!Dr&x(8Io|y%Wc@7S|
zgi%z(Ju^JdO-^vMQH3bqSIS)WePxX@*DF{Vj(*<UVp{af>1uq$y~L{a>QTQT6LUdS
z#c$|)AKJQNsg-ZDR$%}&_wLTl&=~E!GRUe)q~%cnv-Kh<TuZDw_U<j(l*9}7^o1sw
zM3T&&v*CgNRg;;hpNbOX4&8U!ArF6PG-ENm5VxKCr8i`gt@kokgla8F<I!VHZ?y!p
zWB=GqMA$+9<Zz1*Qj_)$Aa<K6bs;icq|Qnu@(Xt7F((8hx8XTtYFJlqH~LerZ#<ps
zZdAU}k4w>$bEW|_o0MB)zU<7E;~J>vc|Hr97)DjaRKSkSadh<C%M?tQ-Rn*s);FR5
zoU@OadE9R67Q?D5khZzTu%x^RW6YaBI%fHnB&Qtm>4km*Md^L-2HX;t>s6`gG|j?7
z7M52shvU{-XAl}DrtRJYxl}T{ppFg!U<K`tmvA9SNlB6Mxg`Kkh)JVPCXHMmPXVMN
zKV*%)<0HWnoD>Ru4%U=z5|ELiS8G1%8jZG;i;7g&Q3H)F!>5NR$<%5RQno<F<h%2(
z0EF;g-NN{A$u3BG3sLytd6lx>=jR`GBTfTz8BZK38nNJVY)=Z7qa0hpTF3f?wz;Ps
zuKeUZO|!zHlJaS!#XqcuC}!#m9&7EGnS3|zcdcMvmHVPBR@sCFH!qt#Luu3WhC9f=
zL@y$*>f(8a&HKKxkewtxn3M-sF!KaMwrTX&cGeiAbM)t^3Wbw>`#3kqv1mrme5j^B
zJQ|lCcwa-gi-XlqFu=Kn)8{=+nEl$-nD|!(CG?37+Jh7NgeqU^kwEOJ^ulsl!H9(&
z3VX{GKD=1==i|bHF?~;T@~PJP&%3pH@F(UUd&YU1RlhGzKJq-1QwN_@CPtMeyfW9}
z#g!Kn3w>zjH<ZK1bWORxq{iV)&Q@()C(`kR$)&6nQvF1Mztr^lyqHKSj&5T<Hg}#R
z_l|DLD<kPuDD@)sq%hvGR@252H9u$DaKa5L9bic`j4RGeFHE2}@h?XI;@#ECC8_n4
z@i<_=&3&m{-gsN_^i+}*(-tX$P@QxrQmLdgz3$K#-neHIy(N9Zh|xD7pu+K3laG&&
z)%ol*yx%L(DF=OxUx9(Iv>)#-fhg~9ln$0e!Q>cFSUep+KZUJ!1n}%wA!1mceHH%Z
z3L={V0|Nn-$eaH59GP20wJ3(Vz@Xw{+mW1{=5;(`*d4C*DFSGACjV}yDvzXzcwdDQ
z@&~O>y=?{a4}(p+VO;!2x(b(%*|I8{{0n1a<zvel?T^-Lse=y>v;7<Te?-1aR78-x
zwxNrCFDu$9OFdjZ8l!O+tlfTpURGv5GFu|1GO%OCBE~3K`X@x^lJo1_@i=K&S-Ih9
zs65(My*MG-);8^SMJ%AJi!QvYa8%j7a-z}r|G?91Mu#<4**a>E43tgc$rMZ)&9dA}
zBM;1b|B<KWnpzyddnn3|=&Q?qtMmr*>}!Fbj<M8rO=`xtxI5lGRw5u3lv&rX;puI%
zG0iqd#@_8Ii}lumZflPZb|*w!@8AxT-Ur;S3l}e$eT<LRaYd;`X{<E|{J&wU$6_Pv
z<U>~tW``rhpy)pqr5*~kprlN)oqiNSEpT79reLMi7Bubn9rcMgSva+;u9(B}2++^2
zrJuNBk>oI~E=P!Wp*j36Y~{OUv62(geb1f80&u)4QaSjWJ>iyPe!##u)yUeD_q`{c
zVxH<j<lM2v4lC7}vmBYm{e$&xdJFA_JhVE~b?iBF`$#3(oxvxyemGocjRukF0ZZeH
zx6;ts>`a&vi@Uy4V_00q8et{Y?lKuR_98w7d{{HMrxI;zD{eox{|n~>H=K*g_Fa2B
zA3Q>K4(`F>VOeD%u5<{58<6<ma5x0p?@w{<e0%*hmw;s00Cd(uQAwpBPCz?tM<N`^
z_312^IK6#*egy~XZ}vsMsIY8pZMTM!#Q|3ZL8(}n%LJ$S^{q*$1j%}@;c|xU@oJ&k
z1v;!_A<1PQfous*eL-@v55r*t%GPe#@#>}$hmbzppXopQ5L<$O%bho+4`v(V{+f<j
z9Uck$4pJ8y%^r&ShsJ-+e7hL=b@9FJDl4%dv@Dkxe`LcaVk5N~IVbu6%2+AII!|m4
z;Xj^D{{4n1<sPTgNo?g%lj()7b9Ix~S)C(Nm97Lhw)-X%qtrwUk)!TopTOSm_lL7u
zZRH`FprP%m*^$Pf*&>e7SPPPd=6;wvE?pV9SCA*}AH3TXqL$;8)-RrNaq*;+U!O4j
zve|xUB%V!7>y_E$F8)dmZ^Jd7$%GETFk4?p74H4j>lNqSFsD4DJeFqpMnlupo4u{s
z<<j-hC`ukHt*K)57k~Ks3P@)C&g_WfIPz$nk+!M9A2LBCsQ1Irp%2x`rox#c54Oil
zXJ!Y&{i_;4R97JHk%>VrEmcz@6mlA9XXT%|_LIY@tbNkD=4!mH$5ehAi<4TkIf_?D
zk(+s_exDxUztjFYwRicHMy<Ngjg6JHw(sie*P2L%{upY#iCh_oN}&6ExCH!v94#Ms
zU>u}M(wYf(MC&BOYpQ0S^L|bD3`%tmwSBB%_n1Jj@Ke`pc+$Z%@sAF{9S(L4Gwf;-
zmUTn?bvmm5{j=V8^Pq}Emamd?@+zUL%VToyv!Jz5C3?0Ekw|TOM4HmCN!qB{&zA}%
z(J;}iHpfKE$AI#c=5xm}VliY&?TzaPd`wUdcLC~s<(ITjgd0Ih(uPD}1wyGqc}MSb
zG$3WeCiP7wM|IRK*@O2bJxq|dnemAqRq3+@DUu!*-W<&M)|+E<?z}DN>>Kq@kqa2H
z?b9C9jqDK<$^1lwBF;g1u+ItrZA;&~hUb__`51Yo&f$m>^uOr86QdLWuLKbvoMtij
zW^XF^cO&8ZVJ%V~YS^lLG!94M1)oOi3kN#xVyxeZ>Q`-{sS(>lgRcq)EUvGQAEUI0
z3gD;rCH>J%ZQ$y9RS4VuR~940bMluv9xu%&@1NG*Clk3HFHB8MRlBhf6L)?tgY!PI
ziinSYH$ofb<)4nij+Tybl0OeoQr|^71%K8@l2p#nmfBj*eP&B3Rl1}5-CTTJaj~z3
z=diZ*#)(3k1Q4lNcuNGACRxwI@yP*1hnk2vSpxYQ1++@qGJ@oxs*)rfO+Q$UkQzgd
z@>y%l7@Lye#Ac%;ZIqMLz6o}?s?4dOj+E^|LRX32)>SaI)m6x)NqqU5Ikh0iP{&%5
z+ZaNA1?Bjc=HGerX@Y`8`XnK7xtJp=F)=7J^GKxy0p6k|zQuN)gN=2kIEU=zeRno?
z4zU<E@?DNG)R(I}x(PBh-rpkO>CbA*pE=`CjR8fXYit<elf!Vy&e)#$!?DVQcD{a2
z9D(G4t=zt7>7J`dl^@W03#_(&jZtQd*(S!4$4T1%uLXE*31zwGXyNTAd6?(!M5-3c
zF_fJ64*!GE4>NSo4@Q^zsl+|TJnuWbqj>P&<mW5WJ$r3AoWiA2VSohj?%lf=GY0#-
z<x=BqvAHEVh}73}0bM&uoNRFDp48^A%W0|pEP66(b5)PH%U#FScGL3y#{ZtEq-;Cj
zO)*X*+2ml6mpB(ny_D1M@ErC}1)o%IETufY<438}*$CuZIpDiQ8$tj~hNDp#nHl6D
z8yBsmgI(w64NBF~$D*&Sd$Y%nYa81n)avBEylRUf_M&9oveZM(7U;ze#ikjiS)=Wf
zeEK#Osv38E|A?biYlE}b$2X_z{iuy4jc4${E|v~zcJ%5HAb|xErp;v_3f|xBpDxy*
z)Ivr;V6cw^U%g$AB29)Ylm>4@T_jQ{o9!d(x;+ryY_G2!g?N6m9fu)*Ly{(AIzwm6
zB#e3dThodzmSn-i2v!Q7>|4j6E^C71R<t~Gb4566|24ha(7u9Hqf&`G5r9q;-4*Eq
zP%$u4%1%`fvSKGDs+eju`NAhl!NGc$`?kH}y)X<Pec5?UNsY^y>k#Rqt+-JxPtAsk
zdH&uGb~~-{5E-*I7U-N#dV#gZT4weIXTX>H$luZKTYecfn&Rnk>Roc;Cx*y$t_0|8
z)o&k^ZSQrtH*$05iwQ8Qg{IEY|6^%Qe1T+(%jF_XtHlEdrO`q8Q(-Xpfy;$4Qz#_E
zVzItQpZ@*(_sY#~YzUY%RLeA+oJpF%@8Q`32JdW>fZsokT>_NWt90KDNOHuF-#t{B
zFc(Vro<8~8SJJ<$9Lc_Zqi+C1Qu8|9KUX2U5Q`@Ol<C=bd3uj&EyA6m&2%1b;}On_
zZO_ecT99v(Fk>aVnbM@PAJxE?Pu2*HS(D-`*-JanTHoAn`D(<Dw}AqSZ(zx`&_j>L
z<U6^YjeqPTWeV>BxpdDLzsX;@N~EXT_u@VE5mLkCRqK!Xk5ZDQ*43FOA{gq_zIT&}
zWbeHQnF^(0(sohJDa~}(&4d9&8RG&4IAY<0%bvVLNW!jl$jjKYMNP~<yNnRXCpHd8
zqOiYw!?V{V5+r&Z{z!;8tl<DSyeb8Trx+Ry)m>(d28YdIp69;@p`PJt)n>gwD+g&b
z{wE}_1MPKsOm5z<Y{YbVa&m>}SK7wL&eNa6+cUFAKYLbNmh)DTf2^<LjWn5x=1hf6
zlmm=LR-2kf)gCv-YcEhWMZI?gLKJ7BlVT9%`4j442F;^Z?5WsJvH6POOOZP!Wk6qK
zm`)^&wedZ9Wxs_Ok<Vjc0p>=V#0@Db*mCE%m-!9HE8_&FBtaJJQL`*&$|;Nnl<XlI
zZ81%yjkk0r+er&GI5yti*=_H2Dnp?uZrQm%idMUq!F$xgFS9@fnYf8$L3`98vr!}<
z8wCjyrHUmwq+UeX;ZuouRZFd?4A%c`7+V!!mTI=}<vw2-PX4d<{#J(d(FVA}IsyMQ
zTc!v5PgtY@(6Hz8Bg_5OVaMFvkth7!)nR|BPR9pkvl)$FS}!1tHYT}&Xmva#22btj
zGuB$ZOy-%`kIE2L@#Bvj7ShbbZ>s)ALt(Pk>Eg2{bPTNIM}j7DT-yb&(Y}C?I(<XS
za#}^5YdA}?)**_Y$qg<1Z}AHhdV0)e5EPDz&l6T>zZn@~%%lod<=gK%aGb|PAsACN
zbG*4wZZi-A`ewIIAD#1N@4A(J=7JAGr9~hFHan)v(k6Cl(lL~+QLD3b*L9_jpIwCv
zZI8oUDqYh_H=^4EVhJ?zJi|%Lb*2>b4d_!I?9gm%Bf}8h{kO^_vbyMV0O<i%j`iAK
zIMB$fbGxO#zdkm6etLZAZYCw+071TFi^pTQg5H$DzhUrm$aEbs5m4%pZOcL^Wl~6x
z$fY--2GG0a3vXy3^RQ#irEP{#<F|gJs)IX+rb^DT6#2wz_5uA9&mY=E+Y|YZphdH5
zwzPhjV>CgD_C82wjqQe1PK-}cDEr6cuahBSGR1MdGi`trC+fn)=cR$p2<!E1JA%g~
z2Oek?(*BJc(?!2Gv@EWFbpKTIjrPH_WPN3qD!C#LGusBKUY|_I0Fg?J7qo9r*uX0a
zmsy7HOSn>xQX#7@5q2S=`n%0S#8@}$Fpc=o6i6>GX-@}kkN?|_L6H=2e<sriY#YGA
zmLZcVbgk$3a@Krf`iEIVngUSo4W8jDirW)Dzj2HddaP3s5e>H^PU`j3Ft*D4;Dr|n
zJ8s{Oio)#iHJA5n6Z^APZ~z4M&X$QXY3LU14C>%nqJRY_-%KJDr8z}pBAwZlQ?*C4
zYy#^+g%SC?`3?n&`xO&{kSJ;DACYWhM9+D|FQDbRR|GtE&F`#+Q+VMRnkSQ<8%Hf%
z+fbf=tSrXg-MQaT*#yDzU`;M{{ED4w>duF=6$x&HK_yJjWUzDi!K-Hcs>fSage>NF
zx$9WEfKO8Z5z!ONJ1|jP4!b1?w(GQ(b<YFD*&3eSx}iab9AueJOlZ~%QBY?^iDWuB
zX7l^Mud7S?q17nT6CTRsg7@F_dA-hc<5=@DIwxK|1Ny#AOh3J(Bxf!URo_c#M3^Wj
zW>_SCYR}>E(JE-Y|B(G@=Zk@KR^{6*6DRtMsjGg=_f7YOL$AMx!2JdEb&5ZiYUzH1
zeU^GNWEpgHwIyitapfRqOO;wi>*NMT{l%?EWVSI{mp|+v+fD=XfGo_w`cN6|hlWY_
zPkC1q(yn5s0C1d@dkz@{B7bj7FNB}(8{DkKolm4eO=<0#EmpmcP$P5NE1+ZCL+=9w
zOgLWj=jec30jt!QizNEjHM}K2ef!@maH4(bEr1T#!K69r-<jy+aJO5S(T3sQplV$}
zBkx`N2J&hRo)`~4@+Av3f9ivWt5Q~#$f*3b_zUW}C<RHmU$o`-Yt<snYxi%;46hoP
ziBA`pG8~c#nD$6+ciDQ-Z4MT2PvMHB0R>VE<Ftn>q{xS5qCn1Q``Ryp$5rxh)d*u(
z>9?t*H)hs`={;SO5mN@Xb-<T*71ubkaJy*wTRNNCJ}-3^2NOxI>$wT3!!cpxyM+eg
z5SQweuJ{6hV{UO(3Budmx*lq@nFfxvd5UnM)pXWrdzMAW|N19yn4~^Cz0gmPyd)hJ
zW%wx}*;uK>#ACiPF)@>cIuvhEzInJ5SFY$<s0-8Qp-c|yg1Sd}W{*WC5p<KJgCN{g
zUw^21rYGS@;kfIO?x$-m!P2%c|Jsp44O$_BAxFZXhApvw>rv!mV#o7ez1IFHrw6B5
zr8>CO<T?v<`@kYaNKOt{q&OXn{2O?A6W~Q(9)L+SSmDmbv*5AM)UjzF>IV!AToCz+
z2SsOfeg5sv0A;AfvVeV;mJ7`!g7jxnHc3gqLlqf>MI3P9TuSkB@1ezfpH|4DYpy8g
z#?x<7_4B*cDEp9Ih~k$G_q+4IKLya*6FnSPJ3B`8j>l56va*tPUtVfll%k9SHz_J`
z!OalE=adGAX?AQw4)1yvQadBkJ^MGzCOm!Zx=&9|Y;>-}GI92nMUrYtu->i#W7i$q
zcnSKJCKHlU#dFdZM-`x7R%@DU>O~sTt~TPL|6#}@?09|i3h%qUNu+pI?+{}41n(bs
z@aY>{C`V7FCF^9=pI;iNffK^FZ14YDV!pz@y~}0W8yh2_p#l$YAzeKBUm6BAp6yGi
zgr}w&5<}EtY51Hp7`1^Z{-r+7Cl*~f;p?Bk_M2CPEI1@K9Xg<<ZjVe2p(j(RyKYHj
zrL{yzkgdA#4#9aSY?DH+%eKml^A1**Dqa|lw-4B1vwwrDwWrsf$yBtK#0U1K-#sD1
zSuUjz&5A1qkXWCJe1mKCk*P(tdNPtL-J3bUktn2dc@r{mVT${AmkD-7fZ5KPt}&CL
zE^UhYr28h>ZK2fG@iPxIY`MaU{$D1p6eKS#Q##R%|BmAmBRbF_0!Gu#-bG+fvOQV(
z%E@(^D)FzcipK;LW$(s4*2k{ov`KtOIkm${U(bkCNm=%O@$*qBJ%CX{(bR7X|MG#^
z@YUrU%b2Aw7Ls`PlgL84_}Fz3u1+lrqhqbBH$^mz)OLGcV(##E*?IvJz#q{%72Syk
zagxMDH1X&qLnP;_6s%<0zAP7ZP)b1!*d)=d)$)w{k!<x&m~c5IKs)pcw3E5xTJghS
z7Wu>Ddq<9jePj2%lYdxVK8Zo5$v6#8`4j~XNC;(~d;=p1>zUJT%k6J(M~+{Nv8Fza
zd(Hdhr7Hh#?xDoR=W(c4nJC<z0V$xM&A+Xt5FGx5e`1Lp_m$fhwoV+}bDv8#XheEz
zn~zRh$q=uyK%)Tt%d<e==If@bs=bhB(OwO41|ZfPFhiuqJ(fCp6_@yxgY>4ET%rgC
z{$p%)%Es_DM&eeQWELQ(vcEiiB6ghTXv_HaVawgp2D7#Qeb-IfF*Q(j{p~TO^4x97
z4U{-BdglK-HJO6J&moSFLDi(AObU1tD-0l|V8k9-s)WQ!>!{W2*)S-jct;;KdCXhn
zcRLNa^B~3(9^X7v_3w#RDH&+uA=Dgd-xxB$;qtsYewS&)y7j$Y+C3l)gL+e5;=l)4
z=4~*5J(hNMcDFWw)|6aqdihmOc|gM^i_iD{Zm38U{k51+rhnqX_2DomU%5B`)b<<t
zu9R7_dUI@UvSF(_j24SO?R5S%5!9VaYfeb3MAvQ=Y;EIV6nB8ad8O)<^lkJ)mc*_y
zw{h08k^GB%6^^lj^j7wj8n>qAdD#^GA}P{ulH~btMAxx`8f!5)HyTN&dE32kRhg}J
zrq)R2VZX2GSlprhzfB}+|3G_$1Ryv&F<ubZDaT?3$-V4SzH`~R0=+q&R5?#OWsUUv
zErmQ*;pf)VBE!9_CbIOfLs~2S=mkbBr{;G}_XH72>v(oO-rBU0q)zVc?)|ZwHg$Oo
zm@l6bAOtv_`)#wBK6<yY+`CJD;afge(sD}!E3@9>zIkKDgBydQ$_E!NBVDbg$4cwp
zLM<k%CC?8sP3EQ|7bzlZ$XYQbu+j)oL+Q;_11mg8ZKFv*uoMB*JX^1d`~aQ(@veh%
zOk}d8xb^Nc`KU^?_$}@3tBqWvN3#zE()PKh%9oR_U9@0B=aqlC17+7GctgQNdrCn$
z)i*j5sbjl2G=5>j(*0kvBG>Qz*w`-fclJ~o4JqqizefkLDpq%YEX3W=-;OVRvORk(
z6^z@vt&^O)z0*1*m$Q4ar5|%Ja+-4gZlccCkcx^*L{!vyB>jyL{cXd-i&e{w?aP<p
ziNzN}tDS#+BhY=!*W_{G61(o+l75BBlQz<NXWG1mRPsJnum1`t5`MTak1ThTYNuYk
zMiB?>gGndZ63{A$A$++NB__@9AH_+D-K)ZTurGika%E*ac-OvC(hU><ZH@c1yPP<K
z{$rlsU8-6q`EH)z8{Hx4qkR!~MOwmAiWIUD(PFkZ@B0&P{HFqTUVU)Hh*|KOfETku
z%hgfCYP7?s&eq-EYVCoJv^r%uG5ohj`38uIDKamTNUz%+ptIjOS%c)XhdY5Ucd%0(
zgI!n*@rc;A(1Qya@l(xA5nsaA&P%1D@+X7uk6=A+_m5%S-&{)?8<SH+?|ls*i1X?t
zRdEF*?d@Jc3B*jbIW|Cb43{)7hrw$4+p6>6O4^>EE*H&HZ})gd^KjOryA)Yh=Vt8#
zviwN7@>#|%rM3)Y30-=TLcc7FTP-X4cbjE-2Xp?b#||Crz_#3cMElHM<~p3N7){8+
zayy@*N(~$v>aX@nZ+(cfG@Ie@HDR#pF1~t2vZ2Ma59$Jbp+l8zW&~8^0JAkI5<6O!
z*;j4*@%$;<Z@9#H)<a3)v1r9-$&y*Jr^-J2{85ZtM<b>O8PIGm-Ltm-pZEXdL|~TZ
zsSYxvdVJ+mW#`9&DL2yAqAGBzU*MxLRK1__b*iPaTQndUj2D)BuD3Iuy5JV1e$&Np
z`$m(8hlh}m5W?wlQb|-?{C}g7K1E5b_<{^-<M_QGGaZj7hTyzF^98+FDoi{EV~+h}
zO=|g}9L*j}o|>d6>qn6FFVEf{$C+7$!-qrG`-OyQ?@#ZVYaDpc+f^a>baqqyj5~PV
zeC9Bi=PTbxKsnDytU36wOXo~$d6*(PFT-W>9klNNFYLaI--bQUli4w|1cGt+`Q9b<
z_yhT$hF5eE(_rdeD}twl-iBqQz`tGnEgq4$!@=4~r?~Hw_4YNSd%nccQl7LI56=~R
zrhL&>5!?BQ(Pf@BK#xAX0p*{GiAReyIGRCtz>?>)``%Phf6cPT3+)a}wLSSaIc@#@
z{SyQ)8jE)K*w`3qYA4l?v01NADx>ZsdW)^>jmyO5gmJ~!uhzdq1B1JF_tJ4@<3#!T
z1^t=R%2l2CU>Abt`n~I$-VFmr?3ok;sQFCAnBTDpl3#`JVjL7_Tv>IDEg#eP3z-dh
z9Pb)glDLnNGe0q;K2(}bh%q_fg@Rg=5VgwTpK87=bIR!E0L9pQ@%ic~v?}9{yd;({
zvFv6+FjOmqXl;eS-(lBmkG3F7vEU_V0}wSEGi7?<0?4-?K8OHJ`GLcMDg>LQ*lzxO
zdziuTNWI!(QA}3$J<x|Bf}8`EWj9pJ*m-;xkdKoD1k|yyh9{?|Z372a_c=(|%qPIo
z8RJ<pX>tpK4PMN=0m!&~MT+C;{IHtNo9O<?xGvMxT{uBv*}BI|O@$f_F`&O{b-pFJ
zu(0q31;r5PtAU2TJ<axxmDbsNdU^S|_zt(Tv$GMnKU^$J&{M--PU|oT1)15;-xNti
z7*)iGP~<zKC8)9|z1s&)&JZ{6BXlQonYu3Zvd-2Sd+Q?a$Xf8i*y{XVk!rCdxh>+^
zk^Lcv)Y&%)D>7vw#X$SFiX&_}DH41u?bvT!#+T+x-y73wi``FgVwU+;-J>jAzb;WN
zA_0P7A6S&7%R68seMxgz_PD41`0=A7vFX>oY+z^&0d(i$<sI<Ag50TBz}ZL)i-MO0
zUfA3}8d%JmG?_5D)OcLXkf+$4uleG7ebiKN402B@-0vOcDvdju-S11w>hALd`9r@0
z*X%$F*M0F72tc7yuMG#ENpHI)0jNjoVP{m6Gy$J~hr0Im5gE09+^4cQ@=KlVE#*}c
z20t8mEHbhtSHWK}0B#LipU9v|G($Z9@~)d+abhNWj_w)!y|h6NJfY|*ibVuZ0#psn
zYQqV?p7lm;mrEZupU;4?gRSQDjUXHJ<lgYm<X6=m+BRR5cv<{v3*@eA;{S@J3DtRN
zF;r-jH}JIisY{o8x&*PY*N!}gGi5$tjKD)grO|BsDVqnfEWfF%4}c(zcy@aeKj|*O
z3MYbb8U^ooX_m{#yaSd}KxhD|ewb|DXE4SO4{i{1)#gEp1=P|vAlD>g(xkryh_}Bj
zU1M7|wgHTCdT{}qWJKV~Ix%r^5V?j3vc8g7t-XQqBVj}N)vH&^RVL^_y`2lLu+sxB
z!`7*(7!W)Z3s8kz%9*f#ukgte2g5&H5k`bfs%DmSNG3t%^&nfyuqJ&(yc-gy_Q;NY
z+&>%BIDwO=p+-tx*1!#?NXqx4zTNz%2#hc^=Qb47zDQ)D#sE~S)UbZ8V&-Oh%=*DC
zfp_6epy^SnrY!_O=2OG=Is5z4drV_v)TM>$<Jp`|Yq^}fN34!Q5_Y8j8%UQ^@;nGe
zdt0V_d=3QLJ@y&Y#Ar!PnRi!H*Ck1lAQA}@tfD22O#~`xYU71kx@Gt4MXq_^qhfGA
z(^uEfU;`;SQ;vW4S5{gnOB%gZiwuC-iAt+ENlDAS%I&de3_dzq8Jz7pZQjtEWw&>C
zF(sy>8PK4+-3qJ?V4=Uq#uoqfEnKFX-uZ09RnykWY9%ixWes>-{s4FWKq^m5yE(oi
zXhc+-&reNg{BwbJ2$K-yy4nIH<L{#~GQC*pA&+Iu2c+tS&VG_Fz-AUH0QBjd)I>Aw
z)`PNsBWb9oWis5%dnE(8I6?8Zz+GjGfd*ub@zqG7u`6DH^%eZHf$XYd0$SjZow8|U
z0<XTFmx#lPIe?Jqb4=?f6;Q6TR_kMKv*7F9Bu(edlZY5DgOi}l-ZpK84=;<dZ7>6i
z^Z<GEHeA(PiQ##kDF1&N4q;h?l#90JpsMIm_c~>@uNe9pTT9bG7Vn%#NJO+dS5=@n
z57LidLxegwjvH-XtW458kBq1!658W`Bpr#J3d(`{10u!!Y8IG{YHDk1Po~r?PdiX}
zO27AGEV=AcXNC&KIvmbHmfG(tMywQhk&XUaoJ^kqHZ|6nMIvJ=?RwAx&TB%lLX~XM
zJ_}1^(Cc@|>iBCbYf}?VMD^xJ7#5$UqmdCMu_yNIS00p`!gmi}UMI}(n?Xwa^yOCp
zG>)B1cIr?#!1`2Jw?$D4m(|`S*c~X6I=Qh?JKzUiHS$#s>bDp*fYt|E-7@_P$G=BB
zB^93W@xd5A9AZ^|&$T);&QQu!=4u|eFI(P=<~k;*2^S(>uv6&Tz5qoj;9hOZbY~<z
zE-~>%CjY{<0vseNBEp-Q=N1~QM)V!Hw}1Hg3HoW({4T=$qFJcO84(cyadEudGR6$N
z-mg}G)kX-vzvu%9UJ(NB|I&&IN*)+BH8lvJV^S~Db5j5Ly5wbWCc<5Cy#dpL(o0PH
z>~hIbtNFU68*0RMZ$vDS?Y?e(ndA+lYToA#pp+A&y-}o_O9Ph0`42AjUyJ2iYGjUM
zW?1JHnnM)(yLMY5-RLyMwuUTVs0%*e?(5PJyJ(dF<J(}pXiNCJ*h1ULp_0pvu_df0
zDZfmW`umceEi~4MkjOB~A|bnir>p<4Ups&2vK)n>12=U?uE$xFSn>A8nwg=amhiDR
zBOx;lO(;m`c-g4^k+6k;{jh(37-V5CRY=;DUo@*`j)JSBDL`b-kGW*jyj>z9e)d=k
zi)iT)SJU5>AnPYHY%@}t`=x>O5@04EBqaOooAw1ACyj(=MmnLOpkP)`PVDfA5;%PC
z%hbLGjE5rN*}bXq$DkI(5L8xHO$%R+#fJm50;Vc;wxF0X`_fPSQl(Zht=;>4bxL29
zj5Ost-v-v2Z-%zlFx=|E1#p<GqwD2y50Q6uRLPr%#-~s7qF7^Ysa$7_txol3k2$d>
z+Kszlnn4|liYgS{IF?qVH5qMUs<tR|DA#Mv$y8x+<{k+4L&4@V<?5$8kP2X%rj1NI
zonToHSNpkq_}`#~)pHlAp7s9zZWh|qH*r;jcx33R&PfsC&uxx)-0^{%o103d+!vT^
zUM!z4u}LpJVX!TAzEn?kb{87XOu%;cl-hKU9F{+a#rlvKA0=w}nV|CRdsRL-g+T9#
zpJ`DD?U#Y@%&+fB12tV6-oBMVdsp}gL*VUed6#!t*AV2D;l3A3IqsfGpWlln$A5l1
zb<gR{8b528Y;WK%Vemb*3R1CHy%(gM6()!qpH^3ns6`A6NI*0;79YB+C)xt(&RPh|
z639n91DS~9zz4#~$vI-g2+kHec3cMsgKCQgxQSUr&B)%rz^}72k%tOuzDkqHKwuDj
z1qIa~au4DbU&0MQBaE+PTIu~ue0l9@(xO{%RuAjRxOWp)Om~~9;i}EoO^)5*Qj&Km
zp<Ww=JGldY7P&novl})(f48~FS?r@rF_$}EvmnPt=bVONq_}dgi*GL!3+yF9o!@@W
zFyncbMm46WCa3D1%C&u+v^Dl4U#S)8<7Ci$>PMto7^mAp`b4&xz-9C?`_dK8flx@I
zU>8IBr>6NyZsGr8h&pyB)47V=8gD6|;9b|>4+QEk#yfeojG1UZc&Qb|YZe(-se%xe
zcy4!g;V``KtL=UuA8W~ZJK1I($k-rT0S#1?!oNM#hxG~ctn%yzz<B_%GgWkE&uL;q
z143;eRzSdyHt3Ezxw-<+qmNj&Tdk<Hbj^hwRuwz>2?hoRz10dYi9{T)fPg@>-ud}?
zJcA*!?bg65s6Xu^LP0Fw=qIVidaYTfTy(b?(fG@H9#lMT<cguze9BT%T;%WVK#X(p
zN`f0cRZ|{=yrG88Z3HgCxXx_@C;k-udGlcaKY(rrk+xmfI6ynpK=s3hptUB!D2J3h
z>*v#f@3O-hKR%J}99k?{o5|D4#1p%fU!3N(dwtCOfXa;?QUv26;&mTN0lg5?YxN2B
z>U(R6=!pRurxhAAiQe&AYm;~?p`BaDlPt|k*dxJSm}>S~Rqr$8%I+XXFJ)a|BXbN6
zS8v2$eLkpp+IlT5(;TA+klzN`Cox@sgwKuR1p*SRj~41wJMA2mJaUEy2a8UgKua{L
zQ%X#%e(MPY@?lUZL$rd*#?);ga44G*R&73zXSfmAP?<tK8S*1v9cPI~&NR8Qq;VQ^
zI35WU72kZz{(^DSp!(${%4bJH2$imZC<^0E(?YaIO{$|`UQtvvjBt5WbEBmYd4=3|
ziVu=#c)DDd=iEm;v%E&RYXzEyz**<=cD>F6DwWCuMd|+EG_yP|!bqWV^D6C1``eHd
zKLrJa7XM!hkhj75JOOTYOqsbr*66<E*9ac-)@gHg+N5g4fJ6Awnqp@rk7ePk{YKit
zIX3M_GL^``i>UY(ev+Wz4iqKxvBv-Ii@HoCo0x9Y+V29{<{tAFW{g-E+^4K2(Znq$
z3EPW~%ilrv<KNECXpKz^Myy{TvUF4!A6grP`)pnTqpji9!7K!b5r_uHyx)<Lat@wA
zgK0RFqyzr{V6mZQp&6L@Al%LeS-zVqlx%HpcY}%vmKUInmr_Y6^Ic@GfWH1`_q&U)
zIpZ%zKG2LUNh&z3yEt0J_5xGV24ql{cg}!>!Z@(5fknLNblr)e+(oXwBLJns;ZPL<
zRHLmRO7a_!!7(PlEA=(pL6_=vBd2&iIh*Eu2G3%m;o$-&@Ekt6y|{WKEa}Q5ZUT|y
z=BofQa0pcDW)Bw!P*>W0Gw5=8STC6}k4D#H(3PX$RCF=3sk1gf$lqd({oNnJkk~-&
zG#{B9_aj|dD-*BIVzB80mm$1kBl|Dg1|ku5Y+)%I42mjvOiPvodt=E&kG2`3!K5~-
zf78>#39Bk0x};g4B_~Bs<HFhBy(Qs`5OQ{Mj&Sxp+DF6z9xLA(hvSadR32NRcQB5H
zEm<Y+KYr{5HKgqBBM6~FpBmJ3T7?D_yc0OI;#sYAv+sd|qi*3Eh4Tbrb#=Az@oN6%
zpAjS0gb5R{*^?(mG=LU2$St33@#FzL?j{lg5FS~eSX6$a{$hmmf`po>GF9?B#C!4o
z127|F+?bd=X?WrRa<0C=z|9i7EemqP?G~jyZu`Bcs_~bcLM#DI)n2W^MM^SO6Zw%&
z1>?38eTV@>-vG6*ao1Iyiv_wHe+-m7s5qEdji|)yH&~sG(eX)nngVU}FUz7bEHquD
zn4tPA934J62}r|mariepUqd)Hx$@NTFeEtmH2g2_m~UnMXs8%vYqUZ`4Ovq(YaWv_
zu3QIoKV#UIm<6S2R76pRPB>CF?s57nJGI78V#BqLp=p0PX6p9pI~}B$dk}V}+~UC{
z_eX-@(<g`!As>)y${&FI9fYKS?QUhm?ob~gTC5ZJia~Cd?$$uO-O<8F@V1~J1b~1y
zIx)32fRP*#kF$2cqphtC%;eaH<7R9W6cj`lDE4IJ$X<D7A&`y(^rwm?nuEaTYHee~
zS1@S<a`7s3JK;gVm}YUc;ZYqpx&UJD1CCTaF!~@%VQVn)9Bj=tpqGs2dO%0XJr)uM
z20eyZ2_`#GOH846!T~$6MNULC&#5aMY;7`H<t16DO^>h1w@bXl{2Y1Z=8*Ti3$zaB
z@MjuW$*`T7TBr(TiRmhfL(!Lt?5)$ta(GqJ2B*AR7GrR_m$KK^XV_v><cJZHv(fn7
z2=w<_zanWmyj{sftE6O$slKqTN8SyxQ)fdJ7jmPx)8#5|KDwzg{UwL0RDjp5&lvGO
zk{5*gz}uc#Us!cD38HeUIrBrQWBt23*E-X5YqK}>db?EnQgy5N;ny({;Vm~|`PixB
z)lug&o=bT6)YE0GK6IrjHIZ;=p#sl*L6!M@%*lyWdq;<9qcjLC>M#CV5)5z>Smwa%
zS>0tl$cBlHjSYeZhCnP0W#jW+dj;9>9`q6jp@=b`pP!#LS<c8Hq%<UYS=nBrJyW+W
zRvQ5tX!@HPuFw0FR#uJ~-Ay$NH@gk#2?cTxNlh9u#uIVx;oFQ}RYCjJ-@JlQPgHa>
zjyIJ<DStl}y=Jl>7*xT;A*EB*4f(0~ZG9d2yr7VB>>D-NRlcVx4z_tRF5hEg7;yd9
zainZ;Qw0VGY0>TNR79=b_U-;T&(B%(AH^$k%aRF>aG&h>JnKVR(5pgjEB*k#_Dscv
z<B$bApG99)#UCzJ=W%0s%R4FD(5XT1ZH1S_w>6TistUcSEI0{ld(wTNDOi8<5$Glj
zkBaya<U{14nS!`8_KWWEZY*1pmp_NnSFliX#op18l#cElgiqTyX8xLTMI^w-e4l@M
zLb{KA4?dD5b;f7Z?Yc);tVR7B2Jij>{E|c8HF>8MVIr{nJ*ijz@aeBUOZ}4aNANn0
zV%@~d+-Q>Pr#NipiW$x8hPc#?CkY7EYh!S^Skb-R%2|YMA1k<r=$--M&CYy!djr&+
zgZA;*@+vRP!X^SNz0MPx$QPCH^mq(*R!V9dcqFklRzg_joYojL;6)#QSXE|5!-o4X
zL)$b+Ar~drT2ex_e$mC>hjeAKvx_Ww7>)74xbtlU|HChYzp1sYTB(-01C0Sj@8<p}
zcGG>7Jh;>6^7N>O^chMU7(;T9aiQff{iKGneb&R8+I}R6&3bdl?BBu}G2(ey;#2YO
z77U|>h2-Q~%O33`Z2H=}Hok>4FlK<*`SKDy5*3305=a$7TU@#<pP#0lSAIuCwDF^G
z`S$9+Sj0E`W7QsCt`N152=JWg?3!-r)Mq638zrZX794t-m2K6Q4NTS2__HGmsrAT_
z)Uf34HFB!ZVrl<ozZlI)i8MiLC1k+wCysie%SfTzeqb0wBS!<Kx!0vAZE>EnIo27{
zi7R}rd?h}X>T38@O{MVup2+GCsvANKMG`N#+DX(S)AUNcn{9zjPAD@8?_<=J>CddO
zD9n9n$Q{KRoayO66;igAjv~tCC}>PgR|kz~oPp+CV(?08yOkwYlqY6@<=UTt-Hjqy
z=(2d!`a+Q1!@~eBLM=3!s(xDCd5B`qH0j$Lp3mQIgyHuM+$C-FIXkoxKAB_xJEVNy
zZYsM+R9UZgMVnhL)>qA)rn+CL%wB;qr!XO?SD%oR6Q@ZttWCRByOa-K0L-<J_D=T*
zGq<9wTGyH{RC|S$?{jmt3WcB_IakMNXvZ)kNB@>QKl;yf&?(?$-gDJZJgPcolUFF3
zid%XMVnWuMcZtYHS2~{{7z>WHN`=e6<{_DQVuk2m`y?*B-Qco5+SkxdfW?7*$MqhD
zB+>>)jj3?~L(>%2oPzA=6|1o}AA|B3^c*{R=P2r|Q{}~<aSbNQz^SUWM!sfGyL8!w
z!&MmZhSp^6X$;4&Y~61$<4_UDALw>M9?l*x)fcp!rEBO(Q1`V|iqRYsjaS!)Xsf<&
z22C-Ix7(U)JHT31>ntlJ&Q@HY1P%RLt)L>lQEqAl#kF#!Do7NRljagsRE)7ngI8sI
zIUxYHf?8Q!Z3`#hFHGn^T5R|QGPOVjt{6fImU@@L1orLIXjl}<qh3NgD&+SHu>$=}
z@0?dFG}%4(ixbe=jEd|FQ*9|aUqY3dok`)>;7wAp;BIzaCG}<-TTz0IrK)eWLo>-M
zWC{bInABY+9^_X1-qJhxD;i`(m*PD?RCdJuKU7KiuQbhJYg{J}K1d#DYg(z`FKuZ+
zQSE7&N_sZl6ZOiRj!dF2DF)_E%0jL`tn5Mr!w?mu+)1eH6vKB<KXm@mdXhDG)AD?+
zCsSl$L7PJ~lQ+w5$z<I<G;PN754{llm2BIC)&v$CkDFCkUgNO0XlQ={740LHLy7(3
z$w+ZK>$Rc&2wn|P{nDoxy|ADTs_9mcz3`FQEN*fQBpcR(mfy0N9dzg0jn`q5<<AZ9
zPK`{J)>*7AKP6Xrl-L^k<C-s;Jxj-wF7tktbJr;JjdkotoVe15rqD<{*a=m2emkxP
zYDd>V*$*BZs_{CexdzcHO>oW4h{vhI*@~wn&|zy!jeFjQsX?A{@axx%K57dEJh26A
zMk^s#+L?ql`*dfrgK?8)-^oYeaPFPMWGCFSH3s1yl^VujVOR)wxB{9|mAD%ewX*c?
zVOz+S8*#GGoJ})}BF9>0^mR>dhWn^Y{u>V8G-UWj2#_>GW7-s+kkAK^mvV9IJixCn
zsmJBV;L?~rP$Cp1ba+D}lvY)}l+X$WST8xk#G#K5IpwF{3If&W<-3KT<<uapF1j@)
zkx^Wo3ElCrkn3EWnBE&C|KQsFW~H$I3Dt^Qi@cWG3hnCq%I<N8&+_2kqIkoQG5C+O
zL1o80!Ci`(7d`G5ugxT$T;Em7=5?0R!2QAVIbJKXt=vwL{)5VxCKr8r&S1k<XOSw4
zm^{%a#HNv1hjDcARq5NnsDft#W=9}0)$*UpT~H-245%k)i;0iqf{oT8_%3k%Yx1I<
z@4;Wh+D?9zzr#30ZfSlo;Or3L#5UAK{BJM>GW9M6r^LAKb~1xM(0PMv58^Fb9;@5w
zfXG-6Ag)SdxrWJ{0^5zAcgQ%bt)?YSic>UT&-H~6`4yPMS|Vcnab$7ufwyU)ts^EQ
zm~V{c^D6p^+U~b#p}|rWRHISqaB5q{2<y+#q59+%Ke)`IHA{Bc$>?K3j=7CL`K00p
z`6S8}jqSNzQ#--lp5D*%ymckR!iZ8II@Zl}`@SE|XZx#lp6|{5u|UFW#U?Te-ht3a
zF}e#y>Fk=~x^_2hjrTHYVYd5~9D`wc<dMH}&AX$wb(A~Z15E!;10qz`>2uHw|7hjW
z_-X<{gjyISwhji4aHcoG(#hcWjB310<L;QP?)NX>=q(>3mc8N%w9#xiF$mb*1}Qh6
z2nc@M+&F<Mvbnh#Fc<K!0a#Zl1%AMX0;*7ckgc>cTalwe_Z}6M&F#j@?>b(L<`;<7
zM?*)CPd4Nq$G(_EsTYmn&H>Pg>#TNRAXw>A7Q|V`g^Q<QXa3|7^JsRoEpU=4O^}65
zw<zU`9{0q_I407WJQ{N_!kj}y+&aJ){x#>4)E}VRBi`So&#1F)6sDB(=k?-Je0YP)
z>h3XA-1T|A=)J*0&4UWfQP>*zh1$|>L>0K*54P~5GqWpGbFr=S-n*<Bu+;UiQ#Y*E
z-Qt0rF=$<1jYC$=_}hDfcI@iKFvAK=>&R6Ap+bT1*IB-NL`Blx(T7OoZ*nbfkmlFv
zg=DL<jPlXJ_U__DN;jxeQkp(jXDBlF)RBFV_c=3ol+R3l<|)J5YY68|&1f>|{xrRR
ziEg0<3UL0}ZA}7!b@?3VC0Pdthf0gZgsCDmkjVw`rwT&A+L~@qXqD5cE<6H)^HJUQ
zK&bWAfr_T*<0VBmh)M(~Rqu)pOVv8!dzh_W*0hF}WqXE>#-gGSBaJa;WE9e|>J@L1
zh-h$SP;b>;-JFB6(j4+Wyw|Aq-?VYP0_xN-CJ!l=eGJ|T3&=he64ZKcT{GN+mHc*j
z^U%m-`7xfEv2LmCzqFb?q?Bw>!YGFQYw_en+>*1UYo-L0bdPi06Z*37y`fnB2Ha3L
zWc%nFpABEW`WSk@tM5(XgBBVDOcNwOvAK)D`nSa4XK3guW#t+?Fs3$<D&WkOZbEY^
z$Q<7-l$Bs}y`lo~!FnJaw6;!#m0t+lIFhMPAN62k8lsX(OxLHIfXbICR8+?Iql}#b
zy*f08+CcTHtu-(Vx640KQa9h{oKB&<G`Uy5e9t-<qjBqhbwCL4XRcJifJ;jN=m~0R
zY3W3$hJO8;7NB(aV}^JMx4Ye(nc3M@(83wD#`;}U<QJR4Oh_WJWA6g;+)eD03pvOv
z-oj+27aC=hjixCsdX>ON{;Ji({JG?k1lA6i6dC<r+Vg%<rWCoq!I}AW#ow^HGfT@r
zQUaS&6g35#rL5DXknp~vV>xGwM@f=|2A^AaVoFb)+bt9rVlJRNS)CierXUWghDvM<
z)%~T0%eKjG{#lW#lQIT!I(aOn;-#ZpMx=Ak!Np<mmyvIqoNj8L_O7VGRyj2mOa1PO
zFjTWq=6ITS!<u0MkRMo(%+<`fRC}j{7IrR8#f9QN;7I_aS0xl8WWK6^wN|gVxcyu}
zT0A>HpE;_|QvwT6%55Q~0MWD&d?V-s;i{;p0866U0_q@$HOrIBQFKa0bHYw7e^n9!
zgDshP`-NaYdn~>WzL6O#aeptiJVAujIbyKFb8YNL?wR9EK6zufz?|RbjG9#a{iM`g
zhVLJ0q_BS!`Fv+!w-h*pvELk)W@+y=%N{X0QuBvnO+n(lZ__Znu)PFDx>?F2s<rU&
zqkATNsf?TxcSPm+F9CC|p{bM9-RmCQ3wI+{=Bw%9Vbhjy#lEgzX0?Ok54>g2;n^KE
zdg2F_NY_{wC+F>>9q|w6&f%F#80id3y{73ZBKz=K2$CGt^6-tM<<Zc0RGgt&R+P?P
zEFoNc@O;|5ii(QRoGfize3}bUO2BxzG=V@?pQ4aP46KSh?&t08?O1fD!-Ip;nida_
z;p(!oh^13LMw^g73HBhi-3#7uDMnK^Au+KZaE_2S02Wgm!U4vFC4q?NKzALRGU}B*
z0(J12-_p^DJOUQ_2KFa<Oo#0B`KXuT|L2@Tl4wwBYf+sNe~wm(s-3A%NE7v)FCS|n
z0`6eaAWj5Suiy;(%qr$W{$3fdl{GK5Z_V=P&?mGmLkitd_;G82E1CPoJL1)6;)O<I
z<f}bT5MH+a;hb^n)3UTrMuK3}ze6pzA2B-BM&B~X)cm11g>kIXb*Rc2%Qfc6Md(&K
z_QN%{iY;;eL2qxxil7GT`VTvW1195r-DaH4(H4f2Wb4&fcX5z2-M`gk@it+V2%8<6
zz~V#5pH{U#?cDk@cl&Dug1;eKt>r}zPgf{n5mpI2!4~;uT7#lci;Gbuo-6?;wx5%O
z*ZY4}IO`8mvbrv-yZ3V-9(Uzwp|r-9y~R_%S7{`Pa{saNj;jFoX?tyXhS?KoCU>=W
zine`w&ZBUMS!}f=mP5h>*;e-_kM)ldQor+)y;y52%bRK}n&zWQ{ITesqHY~?!iDj|
ztFM5z-hwy5MoLGusPgz-HL<=775(bi@sxG|yV3xyNa-(sS;WZwodWwhyORJ)wQos=
zwrl=bZ57=*MVf~3+eqQ@P-^3)D#^p^|8BSZ(TI2M%%Ahk)DnhkVTAl=qI{-td%1So
ztCxLybsDfEue`wKJ!?+ys`O^`gZhGF4A0V|3^`8oG-TJUhT6g-sg`Tp_&XSvh{&kk
zzTF^;dg#O>k7G#2YOGD0Yy34+{FQlD)@YxuA^!+V>0%qBi%?WC>uF7-b9M>_#{xC8
z0zZr;6n~&}RhQD4luIRN@GOC<;o(vx|LN(j5^t@J?Sk5f3N0u!hl14k$R5YFL|BQ0
z>+dM%veWD-MZPrMm13uy_%k_^3(6ok-!87H0;)z5Kku(6DVEJ;6^9fbftwSg7Xibw
zUmq4i^-E#*!j<_6vG~*wW;YpRZ3n;2priX3`~VusVynlq5Ic%NDz(fMPTX(T5Poqu
zr05k~jkvFw+s7W*cL|wx@0{cndRUNy+cM=T;${l(;u}d;?wyqJU@1P_KXuH^C_E{e
zjO^YX4jxBarl9z0$0m>(=lI8ZujI+8-|SQ6u;3Bu%a<e^YYEfflaqVzNuKnnI2GaA
z*%x_qb3!#@BIVdXh;XF->@}*VfsZCyP8f)mi|v_<|HFAC4NLZ{XsCoWpm0YXhMXEG
z7nzSApmv1ce#;|`TCtA?6keK^u#yM)JVUo`PF(M8E-&KggF!`QVx!9UMhCql`Py^`
zt=RPz8hU1KPS?~{&aJW88JfxN6(F<fgcy9leP$NSIi5TcVs_rLj5tSf_YhYmP8m!P
z4ct%Y<C4rlHRHKU3wJjn*}3=MsT%EIbk8W32`ErBIg19=71~-!j|gu<@f7gZ|0cBb
z1dA)E>LD`fRp<M_;8`NIhK5bWQ16U?1DK|PTcyN}q4rlf?QD>t_NZXbh(dWJTy=JP
zC8`yzTzv9ESHwoH(~t$5s%4#0XK%ls=ND%&V})`y=oD6HSbECT!9@8iG>7G1{)9vw
zTJ-^;I*SEdp<<Z^3$S4$gq0y<sQ0IlC#Jt;sY*neA!9n*#&bU9{aoR21sT$(TdDsp
zF7<tzA@iGu4xZpI(%yw-Kvan9gUK>~wR|apoejagJPas%Ml#gWCeIq)A$m9<E}18k
zJ4{V~2-9!7i0pf1ly1>CAch*Q`ntQEpc(=qq{yGn7N1aG|FXhZ`q@KymNrMpJnzWj
z?S_D=#QR<uxSL8#3lAy;4es~+7Gg2<dX9?WXU7>qC*p*K;@(3vBTFAS&7#*fqWM_k
zpjC)Fmz(iu8QzBKw{eUn-5%271}JHLC_>>rOg!&NP*T3UFjVKp$sCZf(1qn%OwZ$*
z$tP%1R#E>#k14UTvh{bd0zd0<WCRNz&=@_B%iP13mhQu&(G5DQ&^LsNY*OOXwKKeb
z@P@YAM$JY(%9u_dIvFt5aB-j2$DRI1i~H6#E6cCAulivYmK()ODfJfOQyK#~#Qs!K
z!MGkMW}4{ZSzER%W$pG>EmR7dQ>-Wg&RwOJ?VvTihBEI*JVeM$weAvfi@NF5DZIp?
zr-R35=0-e}$sC`LArEk>CcBuhg4Gu=;X<)?WPLDT6n6I^>70|6ZR@`xG;Tk0K00!z
zQDHa+1&nn&spXH2-#9tkCQndgzjd=bZ<+sjVFQbezWo%O+p=NW+Buy}zTDCpFDX3R
zaN)QXrscI9S%SO0F12&>bm7X1=!zMjir?L}6w|QOl!_M#Q@X!?nKY04Lq}<jY5gFn
z(yp<911jZ%pRl-P#gOPfZ~*8875Np7QthwFn1+6D`P2)@YQi%X_>6YbDpgB;TV&Jv
z#6)WN3|L4w45?8uq@Y$7j^5}xq<ki#L5CQQS*paH@bmJ(HAQFm!i4Q3g#sxhHmoOJ
zN#Pz^O#Nyqr+iw(w7bRfqS9}-NGE|OY0&S6uV;e%kT_xH>(3y1%1aYutdp{wPXs6$
zU+JomnRfWqwB#w+{q1w~lpSs4%71d<@C)%#`0H|k1LCI2QWR|J(anPxCR24xk=(A{
zfM(G}=?-@%_Gv%h#NpDgvK;k|S%ey$>FZSd5B>bi@67?%uGB)J>qlP%J{qnL;1JWn
z(--XUxT3-}Qu~5UgN%aWRqN8yQX%X2u$+p^l*|VhkI;Nh(ke0ITbXgc4KXc873XQC
zAJ-H^OBR**7^+vQrDauZLsOtKXETPw8tL7$21EBUo-D9I-1dL6vXm)Nz)w{xD@R#o
z`f95bfxkZ>;|;}h*SEz7lK_?H=TkeXcsS!U+0~=%Us|O+9?1`;lg|@vnHOwhYAMR1
zh~3~%c;Y6dY&;UXJ7cf!-<SIDaS>4I$o!Zd92``AxvnD@+Le@Al|>jH6(OxPP5sQG
z7~vM8mgTx!)`2Bu?Sb&hy(*rcvN|&Hx5nxRLrV}BiZ?)wg4{Nhey7qTrCB^PGoKMg
z`8ru<H`c6pe}nDJXjEXf_FEbiyg{)#R~Fj(#-?}3)YSj6_tsHWb<w-<K|~Y;EI<%J
zlx~qul~B4tT17xQ1P%>?fP#pU(%s$NqS6i0-Q5j$9`wDxamToSeE)y<9pdl>&fa^i
zz1Cbap7}gHjsH%QA5u+qKYdZh$_H_gaTD?w&e~eZ=GQB&h`_#!obeEa@7vBdFTcnM
z9_}GY)L*V_omfE3ERD`yTYhS4NH8>;JuwxVZGK=bS!mHq%x-dM{oyG(NqWg!jR#aO
zv%8Ksa^5pfF+2Dk>%OX<MGt++!oGj-Z`{H<_b{~JJwEXm9*`K=Cx1eFA#BQ)Bcin-
zK9SRz{8YJlMbyIZ1T9#`Sv$1QA)MN=7hmA=drbAZ=h`z%wZS>}E>>OkQ0H$HYnu)~
znzgW>iZz_V92-y=*cTW*o_p(ZNrSE?+38)_#r-g%A3TsrugMIhj%bprYI((0>vsil
z(Qn?MVfkNlGC51(lW_$sYTC(H+2f~b7L@l1`l#x@ClzALzI63tMwlOUBm2{;jdftK
zSF2l&Z^SoVl(|%+*?UVy{q01rO%BubsW^#@&;tx8;=LVZkLl9n8L`HOOfAKV;(cRr
zPnPIG%FuMLh7iQ<?UviJ?91;Gx>AVtdqy(CcM3NLRXeT}6dtQ@EDD!=mDyrKD=4pg
z7dcIP{&>Q)PO)F!IY{5kpQ3<h5*QGFwsn$*B(3i689qCTjV%c&x_@)>66GUX(1yE|
zD4%|N<EM-4(A&WK<D+UC9NKR_?8QyUqoFzRb{HtGl}T*#yrJpKKRgromBLxWcDCqK
zo~T$epR^9>q@Vc`59jH$g{3m(-pz%dJxP?S_adAyS8fepV5w1;`zrgq{E6mU!33k$
zy-Xp^U)v4UD!g3Lc=_=`{qj;P(q4(VZ<md4ZyZ+QXyu%z&ae4~biSPK@V@KF^_D<T
z`sjPT;PUijRer(`ifb;D7xX#r&iEl+rsJP5=N^`}H-45dnGf_}QFkY_-LZEnWrj`1
z-Vbh5{Ag%~#_p-!Ee5~zs&N}~vM3+dlTh1&O}*o1ezX|B!+*!F*?h&1z66DFHxm=z
zE%#MZaLXos`<17Y5O)-#zhaiTSb6=_LD?WRHhq@cyU{&ax;EFiVu(X$`PeHv{B;T(
z!gas8166Sy9*rM#o2SMsMtXl<-_f<@#ge32wcm4w609ck0ZKsS>Bn%HCwfd{0{dup
z#$G(pGUOPZM0<ojv6Z#{PG-O8P799b73Xoi=2Y#yeB0?e!X~_<9jTmjuU@@^2-gCr
zJ;Hgs4m-vfiAnC>?Od?Z9SfJ7<)FO<msOarKGaDs)LF^M?qK$FAM>BUC7WNtbyhQ%
z74pPwo2`pp$f3uP?8tj|ZO`$T3wN~*#|djED}#nNF_D$NxJZ%mh_rDMM!jG?7DMN?
zSLMQDBfV>EYt8M+tn*69f0Ja#N+IgOk^kJh74EV)bGYx_j_e0hg7#M(^f4d#XYjV?
z8e{&41wds*HMxC)x(Tl%m-+bYFI=*vbNW2Q>+Y4wiq0K-8TU<%8SPiC&zXH+9&VAw
z>AcwU6u>}D&*jeN_<D0&y+61UqGUeweOY(BSMZKb@OQ%{$K}{``r-ZyRlf22$@Ir8
zD>qNs&#pfe-7)+=eTOJgS|w{1nLjH_MqE(?`RzJO@&^tRL$Mg@bD30@%@6lEqz}?u
z-NSDd3pi6(u91EIxjXA{=XXiOQ*p@)1y-*2k{6LFlDztFlPDLKby1bLhF@vd4zo}9
zJ@)ghN85ia^I?o>C|A*qPtTxGA@de%D=FFC1HEt#@a4RwItnqFX>YhgI~;n6x0~+n
zn1wC!lAs_#HsX<<<he%SWjS=J_)op<Odw-FGBUnnGai#_PmPJgH;MONnuXf4Fu$tq
zQ4Jvk#5(~6*OWWPqn80mLz_tiW5Ei$c=&Lmn+Ub?OJ{C};0k{Wh!C?dC5M~a7VRdE
z#4ugNS^f|iP!!v3!UB$0lc+vn^MzDbX(rfq;#N-!F|-h4xQr!xTghSo4VdoJ`9b;_
z2&HbqQH#0;#(tNnxr*DR4=<NqnyEVPJl4?{s`jd%e`tVs$SZxlaO&bhBDM;bk;TqV
zO?7dexk$j{YES8NC18l3uiwGD0iJk)xh%o+tgI4;!dj_&rMFK)a*ua^VkKooSdSEl
zmCKn76=lXMlo+ttzTX3WU7Sca$VbwRK51RYHT;>`y=pwKm?cq48`|U<5;;>V{yMNB
zgB+fDeWjyrb4r_fIr8phqFxCm-J9p9E6%SsuYb7}HtpWbtW70o7?I!GuPsuX{1QoR
zKPNUgI@*v%s|+OmP$9>IYA;9~|M{b+kvEj5i;Wz$qXs>`S;&5t%$_PFOXq^Pfb-c2
zGE1ZMDJWV4xO2RbI{HU<{-hY~U0$1Mcvw(M{2ChC2=#dYnxI=)SRC&TnE|8IMfcho
zKz{8*>j#(lI0Ddm7cX71{B;+}qT6(f$7b19YB(o{rO0foEm0z5eXHE2?wPRgMW|yD
zg>#tD-FXC@&Zw$;nN{g-E@!Bu@t>UcfQCkuEFb{{>8*)x&q!Ylmsq6q7FHj2Mto<t
zt5`%CIB+pO9xC<)Kh+j#N8bFJs-U+}HlRG3Tukw^_0QrzZ)d47=;MnnW=eXar&h>E
z%pcV-0>WySZ{O!2N$bCm9T0F6Dg=7HX-cSC5gPg>MJ`i70nc0UprmUpi^C<UP>>aX
zUTd%Je*AbD`Z;XZD);{6>*E0=p@J$2q(hUta{*j{!fLDvK>DTj<x3D`1H5eVuoJaR
z)HDD-Tc8J^prF7A>vzyjHROd+RD^(8&lz3;8nl<4aUv)oOAthKg!ZacdL||mjvnSb
za7Z8UwZ?-5pQ893^4`r!gSE@%*3xb9RUMh?y_VJf(-{US41`z~DjfdaO=I8GGPW6`
zEN-k0<K4uo=l2R)p_Z=mwtY^|R4tQtqkdqhMf=`AzfAdh{!(CGXM26#<I0gI@rS%x
z{Pv(jgxH7X1iA`A5Oai1h4IPBr`cLF*-`2x=H!4qq2}WH4y+zNAa(_sXpmY%ffpgO
zdkL51xv8mX;;-n3I+ySS_V1yT1K+Z}W(~b9;K&7nr=X_Ie-RV&-Ql)@w~x<5VqZx4
zTEhRJ2*B`^7za@~x_tAAr_0OBFN%zb;L0F(2EPf|!`GmI(%%vh30H<1Uko5bmxqe+
zBBi~wvgaT{b@AdwbD$nyt=t<Wo_p#0J&0cm=SdO?xMwA)`_`LKb9%)Sme|tMOZ3W)
z1jsU9$FqVJt&mXTO1LGY7`3@0#P@l}y|ML4yj#3#3cKO))8jkKj{lsRuS6kf#s#@A
z$#oI>ch(E31<5frpU72a`F27T2{shHO4c-RPyJ|2JuU?$T~gQNXy-BwG&WAo5MVqf
zU}k0pNaqK4ccB5NtgI}Qyer@ZNnMUDs*YC6jr7ifiFvGOpbKJ#y<re~0xsm(ZcJKc
z-UiyMg+I9upuaGEgIiBmccV*!84V>$4tj!WHqjve0aj$j<?Kki)XCr9e;@eQnk@}F
zC)aF{DB27t=z&CL4*c%x*RL;QVWn~48U4su-Y~q(iVE|8NHsu8F?huNo)D(G!pm8M
z{oFSbg;$(&O0Y8)Q;!{XcDV#upn|O*^`0^HYi6=iw$0lSLuVos?gh@Ky2H&!;(AGC
zs;`)I#6DMSH|dx6Vf@90;MM>l-nsPz@L(9@?=eYfkibeY1aBKfZZqRuu5#30eTq?2
zX1zqme+rEXa<6jp3S3q}e8SM2KnZ1)5CBMFP#Ml_f)o_mcQyn0(x=9>Afy;&zetM!
zLiiVE59u8$h}{OV4?%H2hgGSwx<e8O#+Ch0j-Ezld(W#Mq~4wwgn_E%E0GhuMdJ#!
z+H*>e?>zfHi>DJ_7cOVcgZX+BF@MyNb10?t*ORuu$uw$A@X@H8skICgNW;zA5&Sj<
zqM5Hxj}Kty7H80413|uQAJ2GnzL<pmpSCb|pED3`!iPXMC>>uWT0nCl7bO@tv4X0m
zfut<o#*dpBKu-3I)b4}2X{iUmi~8+f<-_0*fxN|wfds+NKHrvG_zpbyo3)r^=KJdt
z_CBfuu%}y?);&EvmH*HT$Xm;?p&Y=hzE^K)q@$Ro6XzZis@N8z|LGqmys7Q5fA69e
z4|=kt`L0d;N==(}bk5)R6DMx!Oo{E-K=4=wxrL5TM4oz?mzr_6O>K!n1sjaYkM?%I
zlcT+Wh=|W1#fcC+-G6}65uRi}qM)L>Ab!2L;H!FQ8Ub2<et!D84a_~0MoEnXbLEfC
zndX54gPRCo<b&QvPgvkZ4DE|>CLjXUA7}D_E-!-3pev#4VIp=#Vp39{dP@^&w&y*5
zl$jp3n*5?I>s=mIv~E})hL4mONS`RQS^skdBHHI%)}Qnu)%#Vmo(_N3;$7?SpjCWs
z#v-es_`y9Moi59-c|*9KhKAvX;wD-dN4hom5h=WAC%kz}0XsR&LI&A-Sr1fW9^K%g
zxlKI2=>PyS;GIX6m{2V)uu6WtVGSzOPkB3=?&X!2Gu#wJAIkE|P2^XWBy2EMur?CG
z3=?3kQ3n8=Ppl5R#wf-(Ma5e{X_UHqTT4r;X3S$WjE1owm_V1_KQvU8$=T^gh1TFB
zR}6mJ2M-=_BUdAOt5IHU+3s92C4b<mnSquc6H>iIW}$m~ZZd4g_Fv_(CldA@yfF`*
zfoqZaTGwu_2jmGn`^8(P&RnV%#%6E>g&IUeaKG~e(_%hkQmn)^RRB=qnlYc8Udl{m
z31EAHA<cpMC!8PG;M|3V42AfUa-<4rKZq;z?SGHtrB{&(-d)T5{SSPs?%rVhT(5l`
z=knzoJyO(xQQ3ii%VM?tvUyL9x!D%Fhh2Twi`O=hPqkC(MU9LqDgGT1g>ii;r7IiX
zQ0dY2Mv?D?S2+qJUX?Q9OV4y9NGZ6jEh|ym^?d;0OyD~jhf=!mvu9g7{o3g3V31U}
zoMBZX7}SeCgGjG$O-;?Acn}Kfdkxfr<2UL(X^p<4RLLLdw{?J&fZ-ky>zm6l5!$x4
z`|K-s^YZdufEITQ!>IC2>+AXCq5LSz^`Yc?16j3e><he+VZ$@*vEJpLYm44hvd`*L
zeUzJJ{_f^iQL6!$I@n5r@@O{Qwq9m5@W>=vj=OKg$Xom}Iup$__~SnS3k20NL>le5
zd1W1Gu^l};G7`$~z<~hssSDA?^z?M#i1v#YzX9s{9lj5ZXqnbbYgI=~z#({#i3zt-
zOZ7+FTQ=wmYL534P}4QVpE-n<7qpIa6=vtt1AUO&KuZ){hvj4o*u&;R^YyK*DKPrz
zF+jTpeOMPjbR4idtgQY5PKPnw3zxC6<Kf=yP4P^xIVcutQP7km|H$vL#%%d@ZdWKD
zg60gPALZX;Dd*$QKF?3lZR1-OYhR0LF<7mDpr&c?PSC@!mpZHrWf$!q2x*hZgrC;#
zgji9E>-#dP=3R6JRhOypaoL4^LM_|(fq}PBZJHq2`(mM_HJr1bC*=3<=g@r^B)*0R
zCx(@cg~;jA*cobobI~qsK=VaUqM*jbRQP|QSM?nKqF2)_aINw4V}D<Tgv2$@AV{?6
z*4JX0y{CNWy@~Zd5B^>KkxbGtiw~7aSY558?KllXRYFTgFD>?)*BAVKUJBjdbG{)^
z=Gydj@{0;}=UVIpPM8a=zW$tLL;FH^vI@6WqCAS>5fp+@vQ<DBB^uv}@=?IMumi0%
zZg4)~7|pU~f-U_D*hVEoSv_1`5m5eIU0uC;mnRVL<-oP}9;zC?Q_yV)R8vi0gTJ%t
zVlk=}JdW=~kyVg7@4vz&<+GJ)zL=@p483$HX*Pg0|7vUmzK>>5)oKH6YcA(wE>P_%
z<jF4kI3T-daw!{NHsL=tgVIikQ4!pWrp9Kaf!`gfq!|>R3OD=JR+6^S1X^e6AHJd1
zjbWzw**N42)6liM%LH+UGMVt-)hkEd-Y39Uu&8UpbR6V9W?BzGqKvAz;RR!Kk4qhA
zBUQ0>AqdZEM?6M{i|n#L@sB`=-mp>zPM<4yc+BcCnN)x+POs2T?`5@|{;jqb$EcZ*
z!HU|S;HP9*@BxDfI88*nHW|H^EYy0|ekwvK`4_Sq-@JySnnANH5WK#?r4zZ4bI($9
zV|!1IBd$e5b<gmd_|@z_0d``o??LU+(I-FO&!~42Dr>wP-5wQJv%-G!duwWa&FN&%
z{;#Q<w0{E)0qY@av2Z3woUUAh9s_^pho6Q6jh7?9W3~QOjdb9-ibEM=hAFF>Bj9xn
zCqo*m0+e1*?+J(>dqiwe-iC42O8C@ZkNfiPtIz2!Xgh3O(4=*HXIE1EHcsPI$+!7g
zTdw8L`<wFlwaPf?6+ng0*6XH>W5B}|j$YpViP!#%+3BMYtpLoMyJMels%qT-WhLqQ
zlSok&5&;FnqRZc}ao$gNdCe9o35t3__(+G{<;FbSX4&*JfgV4>)8Ak&I^VJ(P+mLA
zWC4)-1@B#WxH!HbZzi2Xx*__>R=+J15`?DcUg~c42@ExFxvEs$0%?A8pX!#D0ro2a
zKz9xD3JKzUAH=Gf$9jZb+u-TmT)Z(=Aj3lZd3Nq^o(ff%$hUuw1kSu;Fr=uqv1q5G
z-G|V1nD?5ghJSBZ2s)~P99sxb&VZ$W*y{K9hdLRNsKyWrUuQh8SZnJKg8mba^;Xc2
zgV5CFQNUX>b|036sPek@c81!RM>T9Nwhu3d{ZXoNSHq~U0DxV`Qwb={8AoXzjjEr!
zc<`4gt}^pFik#EfVYI?-2CMvRdTp)!8JUwXbP_=Gz#Dyi_47HHB!+D3s^gYm(_-nG
zNd1Od#TiMf+rYz@Zr{vCyecK|ecZz&GhB4J8|*w-Qn^2GxrY_)ZsvO-d0g{xj;cqW
z9&cc?x6A!q{otF{1E_eU92sKb_2Gje*X*K4kGM)IJp;pZdUjQw4G-%0^5F4ZY;~nx
zb#p{-;fXIWqREKEfBGH`(03C0u{AX-?VqV^h1==;uIKZ*0j{B=3yt)x+z|zU2`7{h
z#ttsl{|cZmXPxt^0ImcSP>2aHFK>uRde)-MTh`E~s=duky7Ci%Nlm~8E3y$kzptsw
zp{a(`$iV(A@^*FURim2|$}*R&Ps)9tia#W2_7v>CC*e0JTAzYkdth~(+51y)qgg<A
zO0e~b&TLx6tI?eMU$9*wDS02F!{VE5Kmqqr2ECI;wX$YudKzYSg1z|Y;J{*`ixL4R
zNV9jf3Ze7f%8>qw70S7wvyNs%oT(4yI-N2{oSv{`$jDcniY4CPZ;OI56$79)(tM3r
z!#)n!-7r?{_~jOd0)6sNga|`*BJSwNP|JS@$hEl6T>JN&G&COz3fK@pw+yO-2&;vz
z#EhEos3<MvaYh;%njw|UBKu#cM~qX`LX#q@HYWz;knkE8#M2UQsPz!zvyiBNP7hfI
zjAW`T_AjbmLnU6>2wdcH)0sJ$H%h47j0$zmb*sz&yX7D^7uJhi0ajTDINA^iiHV&R
zcJC0YB~!T0Cp#(#o7Le)2$j*eK$d!-s?uSv=<nP1<#s6VG^xw93OA>R%mM<mr^4~+
zISb!RaNwyo-=zN}N1*U;^?aatwL_aP<aw2LNWf}dc3>C9PFVzvmOmln*oWpdB1l6G
ztoZS5enj<6d?coAW4_PL8s!6TW%V~V>uUuTBB|ar8ez)ih(P)mG%PPj8OCI9bho^I
zCr+^Oi}@O)cdk$~`G*AS*h&9;d*oTz%F(reDm=?u$*wx|h9*e#@u{hLSguV&X4O)F
zY#vxi1Ul^RTo!TXcNK?%hqM{rkKE`z^H*c4pAR=Zxb)3b263(=uJydK#b7guVQBDe
zY8b*%q<c<t;pWR(Vk3lxoc&PwVP8dv8BFm}^*~eT7+wGgzDw(gl)L(MOMMx8Dbml8
z@pc|M=W7c7!U1@`781FKMW~8y3!Od&y=nOEd|OxT?Cf-lWk7El8s_6VU;K%O_9H`@
z1!EE;?PKpB-MLLTZ5$nKI2Fq=zb}9Cm<Cop!10ruEXdfOI^3(*ldAbSiWM^C6Yd68
zeS%n%kG%G{aa%Ew0SC3XZ!V0Fm23KXl4+KPw@#$fv^1%hzD@Gz!);pPQ^zCR6yv)q
zF}d&pRfFNxI^ZOaw-1eu{%>_$(%HGdtT>u2n<&2l{eq9I7Zm{%M$5$Xv%kN8`P37t
z*k1q)p?b1>1MsQhohzro9V0l#rw<!gB{1&<Jj{ghXYr62pm)Tg`9B4ONYqm>nR4Cr
z^q|8SE&RmtsH5WE+!l&dQA)!AknBydRPNu#Mh~Ne##UtY@VxBmKj?KdO^~6#$39pW
z;<MxJ+CLbeq=>3)`FeIIeVO{)^%6#Rcr-uZDgET*9nnY>;7d8(U{H|>N$|9%x+H$?
z#m;o{ZL3Qq)fmUgA~PXv<pXuq;q~Tbwhz4brwCtgpWIy4KIJ1WzjUxcYs$h;h4ZFe
zzgi)O5<~Rh>9e^(43@B$8zL;VnSQ>H6I08qPmA!J<-E&Hi`LcD=a<zs9^-fXysl-6
z4Muz9?XKhQ9o$uEC~{`laG$5tFUw&%a}ReA^^ATe#ff}hqKYzYpugu1?W95QRVcj&
zYkrWfjsDHdXaC7nHkXoY@5$BP3V<WCT;#Sm_|08ZRbAHh4&T?k(g)o=>TZ0?cc({K
z{yv6P`Bz@t!2XrDf%R70&jl~T(zc~dVfkR0)4Yc!yT8f1{H&+>?yS{CUUv05msvnQ
z{f|W#cF#Y&Rj@be04W>H8rTt<`bS{HO-^<|#YiD<oI~dJl1h2xO7FU!!*~17=*>l)
z4x2BB)!9<Rg+<(LR~!%eh2_4<e=aQW{hHcNk%U`Wx5+p(o_g7~knw|J_iTK}WwJ;w
z#?%rX?!2WS&bMS^aa|La;=h?Du0^*Op}Ba8|58q=e8jhkN2YaeGt-*E&IOmwpcesY
zIvNMYKYs>a>Kpaf5dACX3ERPD2Isr~LHuaVsjfO?g}l^yjb~Q{?S^Wrk69LRNRDmF
zd*9vi4G>S9%-4^etep@due=@Q>xztr#dTY91~ZwUevy8?M(&TlcXSAKxPG<75kU|k
z60L}YyJ<lpR~-8zn$|3(y%%1A<x8_42Dl<3;e<l}ESPX*IaAp4;I`4i!l2#-k89Qs
zcFnJPs2()EHf$}J#Vo7Gtn#6TJiccomli?74ffaiI`v(1QAuXz&$JHgiD-GdU+NQc
zy{-8r9ZDXRrGQmI`4bO7Qr)D9p&|N(8!jhC)p-_Pes(Suh#5b!`pSu|7kwv%ep)L-
zp+wLBdj==N^$TXAqM`#f;8xyYxk)apL*TMpk=I#2lJ{6@CAKtpTcQy<dnU%Unnm(|
zNJ+qs?fA$-H7fAcjKl2J$f(ouCoZXpM7O+$XwFhZ))HpW*(kYJ^o0FO<(Y9pUgt<_
z9)&|${ao%c?@a?$lK(8$&d<0DW|ERusT2;HC2#;xY6CrphSm+5_aNCwOG^tm!jKUT
zz~4%A!9o_Fq0pzhQ>8o~y`Y1);Q@F}vG*!HVfpUCL`c8aKS$zOLM4I6z+YSOgI_tX
z4aaOHqqdlYir%Wd`u^dr+^&qIOfdt>Z$jPvqrHyYdrqJs&!~JV!UJbVy6q#qmK+N3
z@=^kysBHeCzC&L?S}Yd?d8SZBI{YXmCfhQ_{CPEY?;ZQZ2Dd~;A@Ng}TT9!)|NB^%
zGt2T3>4R(4M7x}d$;!4eKf{OZXTHl+YQ>*fviP1|?nZsXue#<5j~xG`DAE|tCq*6}
z1OELs3e0%)Z~n!4*Wv_Tu(|Jn&)#3~cD^eS8xY2>Q@{M;JKxd`t0c+;+Vf*b2q6<e
zo5qc&9~S{?l4T4Hl;w`Cc!B9&J{Vl~Tu`z$jGX?EBb(fJ8c}r%J2@}bWydE^X;F~_
zL4JCw+$8zmi`UFA5vdn6U<=E;U<T-ThAYrBt9<j1;BDVb9M2P?ulk5&@3VcyOwEOt
zY@If;go=jP%wB|b;-aHn;X%B}`dzkWtcsy-L`ix(<NtA4dmm4U5T<WmzdnBY)U8km
zGB^VOzWDHY5#!*}eq&3t4D@zg$9!CTrC9YLFFC_IWVL{8X=z;++sDR7%)d8`x%@q<
zv+;z=ywo_K{I^=Hf5s72;BRbg<=Z?b|B&N76iq7i;}HF}v!tu3X-6ReBY|X%*Bp7`
zcN_Pzj<n?MtAzb{gqWCGrN|rY&Se)*3DL={(?%F}B=5HHW7dS@4SPlU_PX%@T@Z0k
zzJC^tlU|uykMgDXA^I&e_(NYO5p(qH2dg`e_f8*}+Z2w|agk@C?`&SliivUU=~0AU
zcmyh8<X4YHF3i0Oji?)gl^HJ9+!aI^*+oiNmAwzYzuce%s{nA<2V<ccdyQCf5`JUs
z{$seLue)O(tdHfd+w-pH02plgzAfVof1lfe<I1OxY-LBN!6czP(7OQhJubAyrzfa<
z`qWs_Lx&?%%gq<q0?)#J^5a#yrF$Xy^=Us?zo--uqW%BYj$aJYN3h66_r%cW&;DH2
z%`FINtJ4p;SMD9nf1P_><6?%?kJ{ExC`f0+ugcRK1aY3wjI3O-ePLwiQtblCcjp@%
z$*!G>8@Tp2Gi2O|lE)ndimg|F5T5N_r3=~){#(Q5EUI(ko2}w{Uz)#4PR@a~#!=j1
zRHq?YkrQCC7BRD`f@g80BY2}{4NZ0+XTr<Rb~EZ)8_7|AtO6$}0Nn;WxkjmFq8v#%
zdHbBGBK0F%Y;0^IC@qvD*RpJkot*`sMIP-Xh+S_(B9BQPwgRhA6q%Qb5gEs+pu^tX
z(Kn@M7RLceVtW^x3~A$+RXG&DQgS`H*1*WMe%TFMu_wNy=KalyV~ufPRa{}Y$HA`4
zm3v7A9cUva{$pf`_{Q&x#KfXEoi+PQ+jO@0_D-b2lBT<P$Bt%bn?mh04re~xx=X{3
z_4O&Ee3B5)H}XvBVB7>Te>%>~{U$fPXeivO=i`>KHun1CkfS{=qh#5%*aS00n)}CJ
zdzgg1x4&KBOPnS#a0x4*d${+aYy3`D_mL235<5Bqfw;>2fM4m%sU&^ZSe^0Ao4t1<
z>P%rrlt0mh@lDFL0rA4Y;iSIeidgnX(Y;YK+Kbg^3rTWo#|g?0`^!@g{hJ|4aW?H}
z_kAdG?-v-&$YDreM3Jom8;cTDzLX)<)GZ#uju--u`*;d;pRcc9C}B|Bt`@iE;q3JH
z(=*y?rw=S-6;lgU9IwDCq@&kcFEs2cj^}Gh+>OPo(RHk>*zSGwV7p_de{m)|`Rzth
zt7Xe4dh@7(dLe$b;;!bsk<7lG#j;p=7)(te39grb{fzU<K#^LIVX=bWtLopp<C8vC
zj&5lw-6z^9uiwyT?wgj3+O6VNp3I0VWe<$AD&cUo&!DXV)5;Dg%}DrzLR1E{bv4)5
z3+d&TcDNP;4fMe;AXDl3c>RHf<_*Lxx_b;SI4dlRvz#Vu?rKy@P}`7oF8rZTtzg7T
zQ+vWb)TiK~rZlWcC3`<IjHC0o_$Lh$tr#|^aaU}KEoBsUJl2!<X{x1pWkqk3I$ROR
zkSp-)B`u59&YgL7-P43>Lj0bsWp7FxvOGt;Xc-s2bF|tCMOW;kXhm^V-=ep!#K0y9
zOfMY~i5#!~4+~)D^7;2nb0?KlSfI=CYWZ4I{nVtw{?XAKnb|;IL+=cthB7oWD(!Cf
z^UT%7IO%#b#iQEFo;a7Q6HHAWIpZ&J=DSoMn3ap@kls#7KHTeOjC_3=iH1PPW^s|=
zN`>ybQGiG!SrhNQj6XWUW5?bX>f;IwdZPy`^kUN`DpJ5gsb(r;T)L!X<(Gz<Jj238
zDA91OEfFrM4&}TOf}c`SQ!kCzx)Vb;m|=ewV|s6z;Mwt`vkIVHaDt{W!>)u+$H&J#
zz~c(lHixE`mL4bqP{FSOuAbob+0%E?ylarlWrHJHL{zk?y<G-!n!Tu+Q0<XwAZZ0~
zCG|p@xHskr`oaEw3M@gWf2DAk3@VNfFE1NI9#RHR@Hhvk-l>5)4=%BIdbHO6)AtTD
zmzP|LBgR_k6$hxWZ22)482HR_pPWukO=)W4BGB`7=jMMOI*IdG%BNn@pT0Sr9Ag;k
z{Z{oE0U2|@)U-K`&C#odRlbaF3$e}ZJemYGk>0mNl-1tvv%&$Z0dq))yD}2z?!ZA$
zQ4wm$wx${PkgtV+z1%r4*m=V>xaB}uN*!r!5kv@73CWPLeG9!EYE_P<un;xo?O=da
zdHMP5w(H|tRnEClg^<K*YHRBQ3YT=qE=D24TlE4c5)%?qK)q%Vm?Te4K;5JlwVV&;
zqSIBfwcxjzw!04q!+6a`A?0kiK3;q9POLO~QoG>oH@H_k(0SwANs8qSt=J!L{9<=L
z-ruUKayfIBir|t11sp|V;|w~LtO%DO=zdfhKZd4ygU#v2Fh0AypI`2#qe?X$2cK(T
z54}gn-1Mug@FuKzqo-F^LW~d%w58FKpla#c73`y>XOFL0iP9nv(Hs$uB&!rtB8DGF
zeeO0mJlZN9WO*0b7D4mJMTIKhew<PIlu&<~^WH2**e1$MA(}i3o1FtFM2eP|dK`*B
z5>)E4+D-c*xlt`=VGngYfz(9yZH)sIeNPUTvav2-mWATKH~%SAE7^hb#P=?*GO$$$
zm_3|@Ks_6ByJ=7@2;+X2wAi1`>{TDDl&+i#$#eU?5u}7;HL2}58b8!iiTUjd=#<hD
zfUpKw<fPz-(NPYo+^8R6;o>rF+L~4F1po9SV}@1QE^s6XXac#%k6kS-Ev=xQDiy|-
z`uN#<wL(J!ppz>DWoxC_rx>BFdXe1kM1l3MM-w@c2y>sRUF|$q=+Ay51hWE&2fTfK
zg949KTr2R)2n~C`1_cGVkPs0qELtLNf2?2a(7vf})c7o4w|Ick$E%0qMcN8IhFU@(
z$%1T|WgnH%ui>4+hsa8`u>p}vt=sqV6K$FGdFRYJGW4ru?Jd*Z`1f>Z8{WRBl=D2)
zhL%G?Ej*)sY1eS^!A-?{O{X1)HSz-$EV@WuJk+Gq<o<c!Kq-d*K%JZYOR<Z{fyzh4
z$;OkWCdrWA=3n?T_+pZhpMeve-DHpr1}g^Hh`A8~sBnpO>G<*6GzO=5EOf?ufl*Q%
zSJ?ga<!(2$_v_}qC;&hM2FA%&RWZnWf%sl1F&oW6(mUj^g%{KezMqW&IJ69Gz|Jo+
zm$yxUw5VS(Et&AV?;8{m5m7yf1VDu$Bo0qa2z*3f)%St}i><Nu<EG55eYKY_W1vPF
z53F=4P$@nHzyX`-aLMFoC2Q>rzb&uT{8twZ=n0?8+V}nR$-<f&akVZ=r@Q5sKwli5
zUgIk5LX^3IJ^e!ArLGh9gj1!&_t(38b}Xv}5|VDsr?M*$EG#*UF<K?Ac^sb%kcA!o
z{xEF57bf1N(%FS6m2f63UaTCjiW?%@^DU_$P%DeMcfIp{5bus^bw7Uf+zSK#i>Qfp
zy&*3BF(Bc_&gco}T20Q@&k7Abetz~hzpshW6$q|C9Lp$H1_dE;wo|CZVVlI3nZGiv
zo<MDP!qr(=FQ;~P+7^(oedpm^2^C`h1b~a&vV&vm9cOj<gy)?$Y-D67)QrVE1ZPDj
zHeHKhi5>i!<1noH3Mpr<vweWpAW>s~z52&saQ8?pH7{iQRlu-of4|0lqcYyYEM3=+
zB>BK-+{}SYvgkue-0~^^E#njG!r4l0q0P;&r|L4qs?^A~y6+U$m%mZ51(hlHEc_96
z65M}~NSzpX{N(~_oEmNY%*eCA<<H&HenF8@yq<nzGh+diW?KanES&9N%nM%CUxJq@
zAtxtCNn8&4B_7UBP&f5YfCSf}8<h4|3Ms=B`~m)>Jl_wh)D&V+2o@nV0tPVu1gL!{
z{;@V5Dw292`dJn}HnX+{JCQEEH>_IT9+q1JX?}*eUCmIjnYNvSgM_~)9$6L?--bV7
zsq$1g9&8?yl0tQVakR2Ly&3H#0>GesY1stN(W1K4eG}BGM9NuOB<$m36?@fr&xTV9
z#~7>T7ic#c!`K&nhR1gg%H)$f;##L>$ocP3T3?l$oUlFF-3*M=JU99!40OtkBhi!x
zKbVq#=FYsKo9TICXLr=Z95e6qN;C0Qy=R(QVH!|lrIwVqC|`q=nUeo@%cJ%{TjTUz
zwfX*=H*Qdz+$<rWV#t`Lk8ho=BQ(P=3fq_8EW?N>f3GH2alF}VI9!qo{7`oxFc3>k
zO+C0%I%9Hbvr1(yI<-k!0lV=KPB?4G&e1=72s124%xOAo$2-OdRB+I|7AXL>^ElfH
z$YmNv194c&2=|n%kylLhkz75NElZ_m8@48?h7YwdFYqVM#U;E_FE~6I2<wW=K6a|W
z=C_{r5~ShfmsnI6F?5;G*6?5zv*!^fE_tG^pWQcYv?kwNMtPJ@i&_OWg<%D&X+!}(
z=YJ<@_vS2h-0-_u{w~)tud64n+m{mSUZgH-C>tbZ`46T8P{!QzD$#wf`ft^7#T4`^
z?RU#r>}pNr<;lbSqLl(UGD3tru<$AHv!menHvkFU&E4HTAeT)3{{6PchY!DvIhx<F
z((>{ixlI(Kj`i%oA)yD^J9SW@gp;6V>%u4y>TlJJ0LP#r>>37Nc|@Rvi`hsP;}PET
zo#TXb?~=IPvBe{a$fx+xff>=k%*@+?1uc?J+-L2>y(LJ`5Mxs9d@Frxu`cTBHrj#P
zs39Ukj7UaQPv6J#l+EB#-W^h&v?ywGkn!)SGSXHFQZ}|@lpF)1La{DDtXLoKTUaml
zh~d*KTOByUxe`Rds9v<}HrEo)**>!sEDS1Tlw#l;z5AA}3MVwh0{BDSt(;>S;z4wc
z-QAA_*W1)<jo&2RSQqaChsqiNN+ZCqM97y6yL_-5nN8}rUfV^7{+MX5**mBdbA)~-
z%bziIHHbniUnPTXdlqAflkUMI3Bx-K7~C5Z9!&TY0c;L4hTJ3}s|VPq(a<<W5u_iE
z?iG;WmMc4-eJ9zSbQwPF@KlB2Aw85bQ``DSm9Ufuq+svDANsp|wz3c`^$}`2l8X^P
zd@Y>X(w`_nDAt8Z<`@$g=;qJ{s45xg##Kpe(bVYIc;`#80JgmcX2ZuyYSbz31%t=v
zX&diaNfH+)U7@RrKOo_@p+-gj=#c{GuJm^$N}#%Xf!$OH3;^of1{YE@j1oaJ#j!jR
z;?ObNIJ={tJ@%b*)MM6QcABr2fVN8|_sSJ-32_@!2Gp2bn}=CmeoyO7N!A&fHQL$!
zCjXFNhJth;fiilANuFf;2XEesks(nlAU5?+0747{VDa=ny~3%<$j7IFI@p2s0TmiR
z#jJ6{{I_W_$_SM$rlpJPj{0B5#q9&)ru$-sAR?sK77<!425wn*tvlupz!szW>tX>@
z<!?#W4Fyp6sDwW#mnNbdry5Ti(P`MM9=U3nX{z~2hJ6`!+)3y!@8y2_REmiov5cJ3
z4|Z(faenWo9Pv9mIS%`9Q87;(v+Gv)*O(^qkm>~i)Zk+7F`H@g?~O9>J8AQpMD9g(
z$wbR~l~Xq@-n5$VA;G=-V$VHTnmA>-<<`!f{ft5667s;i%Jq(E$rJx=0AfD(2Sf`g
zxXposj@m9Kg=z(FV|SMN)8R4aMwF|aop#p7v>`DR;`J;w3!l*+s`gsIE8GP`NIE?^
zBD8OVg}CLX^ZGT4%bZa)Pq?$Q6QeB>pfH}ejCNzQc{Pb+K@brd0vs`n-6)uVQ7r+g
z-sP6Ih#6UDl}q|VceyAA?ryT+pXLRRNsg`#v-wk~n`Y`x5Z*xzaGb37!O~LinR*P?
ztBJv4OK)h&i}$Ur%9{QYITslcX(5W)*VYqZF;cMZauO678mCTX`vi$+HzE}j-3}ba
z73-G2zmg~NXoGtO$dx2`!KuzChpgePhLBZ3Il=H|QVe^Cr6#vHxZhC&q}2&Jp4Ia8
zd@tb;Cjfg9_-(x@azcPVi~;(mOaS`KMVqmPZo^a<!csxmVqh**thBerA>l%KpIyLE
zy8P2&ikL~Ywj9p65$B^7lz;wy2-a>U3hv>;&4!3bH}sG-8FhK=-QHy4A}g<NnIKu7
z=1Ev3rykTXtoZ~#T6L3q^M`(<Al3PcZb=!5N&M-QFREQ7DDN}p7;i0D6$4-Qd}LxA
zFQn5@Pkn}W|F!;G`GpiUUwh*HeXha<$AEBp!7*^Yd>mJSj)KW{2$)KI*gQi*LQvQ@
z{+%8L4AbVJH<_Zo#3E5g@>{#}ol|ZOLmJDc(@@z70uRoraf<?sj{?-cN?*#FZu8qj
zXgPBLTtej8GeY83@lxOKLxo0wP&{lLNJRt+nJo4;^S@YoBRG`bU!0}AcVjEU#xyEG
zwQIsk#M2}7bxgpCBgB^oMB;s}f;>`(8Tv{=>wHrpHgWbPj)2Cf%x-EM+7qh`x2N@1
zO6_R&hAqfL^TR$?!qs3(l&Punz}78P!!9s;JTTsaH3}r3J1aw+XVst~urnX;hjQ;B
z1PXz;AO=7{&;5ZhE(UPOjbmeDrQT?P^qkgjllzk$nOl8A4poBqBtxoFSl+$6lQQ<F
zYF!4~%Ct}TR6synH9YW6S8xI0T7}DOquCMac@Z=LhCQYk<z8PpM}&U;o-n83U-?{7
z;iSu<9c+?%^?`tZ7HpJ$pb}Gx7sau&x2FR-QdIUJ5;;mMMhw`97QOs%dXZNVYPGyZ
zeGC9*6K@)^SOXu{9*%1Ew;lIT(F(w_lt#)*(-#1&<Xu=mQWF%bG(uBo_UAFr_b9S1
zf+5fDu0@L8%h(EQxGUKF4b87RT}&&qnk=4c2j&Rc$``S%l<-MX@pm2l^!GiVXmjj<
zMhflf$8{r5|E$<lLO7D9ko<*#z(<lQz|qmsYQ(yq7W{;MZI?<EUK$zl3EP}*9AXu{
z(+wz{;@`45&H4aD&vQF<fC}(@)Qv{t1s@%?HJ4r8_~71ERWONT&MIO5pg6ld9ewRv
z$V{D+V>D@pHaXg9=@Uzf`e_oc?Lkvo_Gfq+{Sr9$>wJ*>=M$={2%sq7AQFWwC<*R-
zke8PionlH1@NqBhjk)wg(oMBe-IE)CJxjBk<E#YYPmd^oE{7ja0a$|T(Op-|#WX74
zovun`<2QqVh0v9*+55YV{9mc~OO?cb_c`%9%(t0pGTGa|ee9a_*<O^CDqwq6U?JqD
zyxd_*bmKRzKjbDnnR<mr2E+Au1exa+6>&8VSh5eHRtUg1rU7G)NIpAx!Q*vz`<+D*
zpE}mIRM$NqiahuQoE{!gQ!_J^cz(kMlarGXp}3E^caI~Bjh?UFnR(t8ENU7e@P|3_
zH^cGtw6Q@Kd8nt!x7Uw)lXV=VNCXO=SDR-U;Agf!<0F^B!!tN`UMOAZY74A$AC)ZM
zR*%^3pfb?RBkudyA;VlvBWPwew|ah>AzXd9daZs;>{DMaeUbB;BOKCfk)c?Rx#wQ7
zF4V`K$#tu-?`$UrSk3m|?8>H2pb<<B&c3xw8<{KS)s?tH>t9dq(-1eeTB*=J!hH{^
zWO{w1J3ib{B{lA7)cF}^U)0w=Dv6-V1W`2kk0MQZbDquN(sP~_{+Fy4GS2vSQ!X&C
z|N8qN2xLq;n=9)~>{7PU*!M5?$e(y&WwxJiS3d>pCPX2nUp&t_Y_RI-MmJUR%@L;z
zqfQcgU~+V<<8(N*c4)ckm^oCLT;u&p+stgo+aUE@@km?|&SJ>E!1<a*1QJ}(Lg-32
z#60%3dl{oueI~HNW3Py*Gj%IEA_J0VB8q7#qU;ZR3SaL%&4`L39y}7oS#gU05-Hx%
zhmL8*aGKy)QE^L-Te4HmdfKFr$HnV>$sxo)HD4H|j<UU9-M?C`tFKSEz$a!u7m3tR
zx`A+qJa`%zY1ZMyM52I7Ay#bK?fLA8WfHN{lllSjTjv)Dv}VG#9RrYCNW`<6bDs&m
zE%I(6vEc9Z5w5JQsrVJZ9-ozVp1^=_%SuXPQ~<H#|NhQbijm=_5q!kYCq9?{eT;<$
z#{2ib|Kp#@_Um;eTEi;6Bb0$exZHtJ!A$8PA$+ZR$dG@RzePlE`*+R|H8Gd}um7B{
z#YTjDtDlf{w9I>b^9C##rNpNxb#Cr6uUJdV+0JkK^6|wAggKEhuhsLjy=Jqpojt6(
zp9JkjO@f4;TIn1W+3$NU_2iibhSQx#vItRO7=Gfsq<vY`VYSRo1KHLaB$s9N?S%&(
zb~B{WEnT0`KOH)@k)giIRu-j+&*rZwUA}6r-Fbey8Wwoem_U-xJ_>7RRT-I>xZ<Z#
zT_(K=Q9%i^%?H#}`V!bOVhUdj2Pdlw?E6hy?Pn<2Gp<iW)t!&k^;J$IF(S}yx<^s-
zY}luC!gTAgfhG3Bt|3OfrWjV<b550SRd30yuXdhJ<(93Mo-GQ$?wVmHIluR9EHq~L
ztk)P>cBGfz^Gx@|T%5$1`;IT``WY$}wHMXoo=wUuX62eSe(Uf1JGC*_V0Ok|yO<4S
z0kBsbKeOwdoPToaJYhHmzpA?TdS{$(`1Vp$Z=y2w`7Kxk;B}hkT6F8rk`CXD8geKu
zso+b%v^-j?;&<}IrvLlyW^mv^N@VTzf@iZdL+*6hWWU&a&+|b%mf)n)x_mMqKvU)3
z`Q>O-V76*JI97*}w_7rXeS0&cD}IDg6k-LBOmsA@@M<cP{rx>{@%4dM&*i~o1rpU^
zP0>$wZ}iK{Kd{!v4ohVib8}n%SuxGJwECv_Wb33oci;_k?bgKlr;=g!w9z&~#Rds~
zD+&Rvf#Vh}o8A07!yA9J2bSlIt=<+71TjY8spM6RVGj%pu%_NV>6KfJD8E8iO?^+y
zCR}R3_=|d$uGZwqS!p%~v{~d07$<7Vf18<}&arzW+jf0DSXNs43o9!t>XB3}Er&Iq
zhU(-G={^{5P~`}hqCB8hH5^@X>%}lTZqGtUi^-^P2<DsC>-h26cnR4=;e8DRw%O5w
z`jdTrG;WfLkdg~;v{5%hGlY>!Cgp8;Jq~Xbt-%g_cd>$UdTA-WqeB)MBKMkuj0_nf
ztH(7hYp+(8p`8AA-*7fHwaH7{tcJ5Orz_3g;3K<F=QTqbVOeV!^?^Z5-nS`ern!At
zHzYKaR%~-~)AAjqIdynz@R`Q?He0h{ZdvL55`{0VNf|AY2m7}9ZhKfEQQy9i1+ViG
z8!Rk9NRwx~xm(8f2%gYIO-=1I+41D$q`9Y>yt1NVdg6(fbPV4VH3>o^#aY;PwYKPy
zf4&RFSTKYTUpdcRJw#3Qwe<_G*d8hwDFkxX34bIFRKy!Gd7@}nB;|w?y$#jUlhT|1
zt)Q!@74)Yjir*5iibsC)Nfi%+SXs8dA8T**U+Waec)oJ+G+i?9(|d`f2P9}q-*Rzn
zNKO!xp<`otEHm8Izk_j}wQ;p{cO^t<w17(yA|n`7ZZP%%y{lqBPP>BdNs@&FDp+4g
zrM@*ir($S*{sF|)gK-R3d2^+JpNG?*CN|BV+dE{lnPywDF4YlI=u6zhdmfKEn)`>`
z$vS*jhP1LQ|L)!=c4g(@P-U-HCV|iEx!9I#MrUDKQ81p+!%DN~5>#O+!xKbl!pE2?
z=FhsaxwG3feO+cTSo+~ABkB1<zkLC<h2xCWM2zO9Wp;K8hz`e?t7&zW{0rz(YwQHH
zlyMeE6xzq^%IUl=mY)}i5Qvw;Fcq>*Pth;1?Y@6P?6fwbmS^D8w>la>1rFk>c?x*s
z?Ae+UBWeX-{X%+268{d;og4<)`~zl%Ju}W)D*Y`o<nIrfu5hUSXXD-Pk7Jxa83OVD
z{`q$z{@*<Ozj^ro&1TrNLZW4Xlv4uex)`nu77kW9l>@II{;gXH5U`{{HO}x)PHefv
zU*`$(hP0e$YtVgQ4Q-zU`}U5G6+oI90|nj`NEi{b8wNtKc;}z1g+TPvaiRg5ULMLW
z(5#onW4%bj#r1-el(ea-sSDC7Q0d^xR4Wt*$Fy>_Y>5dO@}FxeP0z`1i+<q(D#eVt
zbee3e8(bRVw+ir&Z{CcDzKtjTY*5I7P={0HzmK)i!{a4_9@JC+X>%^C*%l&BmH$j#
zTlwn?TIC~AF3U4BGe5=uS-l$eCMmp82~d#sqM(J$8E8-t@^jMDO8~^mV|3P>YxC@E
zxZ}@T63Y^lKVo_|=g>4|srxQsYX$i+I)GmDbcw&+oqn)%%*o4ru0Duq&AQeU`FqM+
zmFm63gf*H!#i&G}Lx}y^8rNK_uxc}QTPDuq11ph>72t0`f;)o306K$0_*@P)XCM>#
zL!@0yN=gZUK@yAaFtZD+zz>*xqpwTL+dQHalkvN$#R_?a=0TTT<qy@O2$ICP>5}rR
zfoP#bM5|0_x|dOY4R|_GLA5~=(TIdL*q*t>V(6?wn^vQi@MCV3)qYY4rhSPXiGKR@
zsca!Bya&q9$dLbDFOyuB$~q4TF02M=V7O6$7l%sD-q$3A1px&1TI;(I@HRru`gLwe
zT2FJEdpmyAFVo`Y=?U_Ha_on$B;iu@MQc6&jOtd))3lM&GAb|3a~2gd<Fqo>(rkL^
zR+(~tEaN_Nwzx9T_#YM^+Ox}6AVN%{|9Iz4foZ!|0~J-1X>q#v+mB(rX^Byparl#K
zYm-D1SC0nf+`K-ic*3MVW00Z(&yLHB!fT|hI^KM=n#fkbb?=6Xr$psh@qqEcnI21c
ztL|c9Ex;nEPI1r!jb=E;t+_~06;o8(4sd)cquMJsvDK`s3V>cd0Q<{OlGGi@DFWR4
zcMzqvGKZPj|9D}LcR2UL6!di8=q|WaA14(jNvk+oTE@<OKDP27<VeZx#~ew|Vhq!s
z#fl6GicE}H9`@5#XCw&K+;-%$-wf0`+U0e6h>Xg6hjsQg5bz-AWmbLj4r?4%&$Xi7
zM45Eg{)|xJ_`f@Bc6JVp!h!EZB-(%fUNcx|txnIj1*dIudpocpf4HjBlVXc<^ej0s
zk(QBB${##Q^`uiAkmqPt9KTw#(==liQ(=zWm~1cv&Sc1$HTV#Xf6#pPC2ee8rEqHf
zdMJgp@8=F{YePCZiWl#deWVi-jhu`loCgb4(ruP|*xUKCpHyabKCAcq-CRRIqB-G}
z2r_#sdx2r!ViD;|#cXNuS>LXEU%vWPtSdKAkx{@(xO#as8DKpIKm8NKSR7L>?Uel-
zN&LGN(P@xqYY_Utyra6wLo(~ovxJxSmV4P{1d|hb6s<p9?JqHn>`jpOxG*;&FDRaE
zwHmX)YqPX~;bO2!wpFmVZ*j7FNWWR66-qx!)fSdHar0EhpDn{0+jp@;G*$pZ<p3%l
z@1>FLT@Lj93z_&v+a2X(^S>Jd4%al18ms<)y}8=Jblpr`FQK)HtuF*1P6U0aOTem6
zF^myme*dAEleWDQl&OgOpvYIbGW)cCgbxxim)J6rXZCPq^K-mWUKr`Yc>J*+i<a1I
zmTK}ZYG(%Kd{u=bt?c4*YdW^!)|+n&7Zasd*pG??uyWLPHdV$97vm^Ha$lcYpO?Zj
zRO7AK$PfnAIIy;cP#pSD?iwKKcpUmAf-e;fXK5kRH1ZA8YqF(9w9~`XGpV%6wdCwM
z=8lM7Bk|I%JW9q#3V2mVRm-?ur$`gdomBPY-~Bf5DnmGT{KdMy;t9&vg+5ADp*g=D
zONC{3+yoS8)$%Big+s8NdjP199d$I}MMii3{-kBlrQ2P>Cz#tKbs}lwQ}G*>WdsiN
zp@yfY_SL_7%TAKVwvQ8-tu@2@R9f0H&yHqP?(zo3j9MElwf)5Lb~@RyHZ**h|1+$@
zAbm87Lr910wj!3&xy$Lg88e(Q&L2`g@wHX2eEclyho296!%k~7Hl>(Mfv@t*D-QF!
zsVdJ}>*x%(PSFPo%&O^|>AS1K_L3<Us#OZ<H*UB8DhLwm;@l!^YBM4@A5Mv@Wxg9H
zm~0Tzu|yI*F5MVv>h(`>-ia6I^AwXGJCqx|g_O@%$&%mVDb!Td9I$3#l2*qfQ5@-b
z$)2g0`(x?@=6<?KA)T5XgYuodRRt@>&LO_@OS{rPE7hpQuHl9-t{iyWQv8YyF)_K>
zmCFs4D8$PBNZyGU=#H*Fct$B^lBR57pHcD4w142?-<bQ<*Ol-w+9!SaK%M+l@96Aq
z9VG!?QRCNf-8w|IpMFq?S$68oQH$|JeI4I^aeliPNASXBU9&?%C}`Pve=X$7=jTz1
zIge^d`{`b+8nyG&Eb7)t)YBL9KX!7AApI90$v{+qwZ4%@_hU@^mx6py)Re~Wa^{5s
zic~cZtSp+$mUQP`=T)dD*=L*__-Uqe>-=5zo1ftMVtLq2;uylEeDLmC7)A*0-u+5p
zZQAz(Ln+9qb*U$~+OPPO$FU_<se;2x=-nQfKbWh##OzkDJpMi~AH){2DFSzgB>Yd!
zMrLpLT@Uw9i?&$_3TcRo%2X>$D`0Jfz{A>PkmdpPGi{Smtw{^L)jydE3k!=q1)rWC
zpJ=`yKfe@>45SEd89okaOS`TtR=vg>Q+k9K{Pr_p@NwpL1PQ-oHx3q0`Kv!c;|pjf
zlwvx;*<FnV+po1<c<3dVPYR#<Z!fPDmlbfFUre?NB2b#Z;@&h)(h6;Vt1)Jmx5%gR
zxmjmpWm1QF%{EF{SY7$WW6H#o!K#p9vsz;x!H*8kLK8em`b?t!ExWa7$og!#bPx)u
zOG8C+e!-oIFV1Vi2o@g5j1XE4^+vK)n3+-Py;)o|OE#LQtq-Z7f|oBLaoTrti?Dgp
zpzkwbv`RxvO|z`Y0PAVzvz`iR6GmQ3$$m#pg=g&v8k=DO1`A~dhwWjN`qgnzk0g{4
zv?HSX8-<BNGGIJz)P5cfl@+|v^~h?ZC)(;rrrKa(BEBhisGXOY_L0Jv=}7Fn*w|t#
zzR@k%r=iT1G2f|thZRd54(uG3`unw&hc4|D9`0XZKiD=pK{i^hh(BjAUp-BD<EvVw
z<|}X_?t@DzHx>3R?9nc}$EAxUv6ko4_mU!Nl?m~9VAg<R$sKaC;g@#U7V{eq4f?fC
z3XHPdDl*cDZY3a?(<i&T3OFBf3vRE#%cR03OY>yEKde#dpmgO*r(5x_#Rlf8zGs4G
zam6Yby+ibUX<^mXb?7I${#Fc>N?tP2-zY-)`Pd*aw>>MgM_=>nS7pQBlR@YnEGr5m
zaxJOq;oY5v;e$iFgRw)SEdrCzKJC?&PhYxbnGJ~|ZI;qV8P$KkVJ*!a%&Q+V>{W+2
zy(8SLOTYRgOVuU22U4ll3^UQeI6G-7{Sd$}zF%@kb+#?Jlm2g-g=(;4ejis<vKQ?6
zqH?eU%Qg)F_p>q#gSeZ%qd!D;@BaPzy6wdphaLC|3XRO1x^g9N=x!P@F&)6C%A5Z_
z@FfwrkZ=nrY1)k&adhP5t4s~%6$}Ku85#p4y61E1YUv8e2_|&FZ0$JgE{0UA7`CpC
zz~-kZAIVCKjWv`@E;2P9Q(htF_==exgUuBgoP_7RulndrEiFD-nwIf_LX=+qg%ii#
zBrSfNr6G0AaL#wgP~EP^>Fz`88M;>O;wP9>lYv%LBC+q$P%usVBGc5R!6N04)z#(A
z=Z+oMdw2x#fjq{c?H2ujg4ai1E)<&#7Hhj?sHSr!ORIo=WZyk~zc^5qcqfg)A<xos
zd)<M|u*pp%FLI=YSw_#MNd6Hk+tM07D!mFR>;y>vXPGjFm8TBoQ9h&;F>~aUB0D+u
zVa?CS9<H)#R(9H&yMS7YpVUqJ6pBadembKj<E1Prks%JCSFP~PO0CmHbuoE&tk|-A
z3n%bDWPNK8p%ize#CrXOo^6j5MGod5qqAI2MfgY8^eQ*T@7aa5u2!O(TCEF54B|Lw
zPzT6J_}lk+_b}qb^&}bMD{X8k8QK)>Qb+cY9pTL{wenhV320SDIX#gX?NTulu&NGE
zM`rsv?4_>Ft60-8X^leu{ZGa>)6Z#{qWD7i%PWZ=wy&Ce5=>Mr-!nMe)wEw3T4Q(G
zkU!bqbWviiC@XJ1N6Or~iOS*l|Dtlv_>`~L+&`I$EZ5F-2;&Gm&}O_{ckGC-S?cdu
z{HgEr-Ewy0)u)wb^)m3jW6YxN+KHHIAo<9j_>vv_vF7C|GP4OsYmEu5{4c7$Ix4EJ
zjTa15Bm@McRZ3dAK|wmD8>vCMYbZqlk&+lXq+#e7I;16s?(Xi+yLrF+z2CiS;aV>K
zIGnTNd46^FfvfWFCMZ(afRlo<9fCKAhg0%rw2H-T@-@<2AXmRzBA<I6wAga1?&@ti
z?|vM`LDF+ligJ(Urn$H~sv*1)pB^8dsHsR6Sr@2Fu;>lTH?J@c)n5BQKO9rO*u1tg
ze}=P@x$#f+r=FkRxrg#w;NqxameX#;Qgv^?Kefhs_zxr){d>(Sn^ZO8y)IMTDwpr0
zGYbG!po^F5(S>ttO#qN@_qD^qU>Pxc=34g1{D)1`-iH-hN7AcUaRaM-v2rwWH=YGF
zJ?D#Tsyfas)YLh%Yom7nF>A77#D!a5|K-$mMcn(Z7dhTGm?741@KDE|>-f}JnR=N&
z|GBq<cR-qFc4j*-OS)NdxVvE8IUncKT#Y4*G!Y-p!zhyF#q}KZhS=2mwDOUL874>>
zea~F5FF{qxUiY+7EQ(eSE8JEWmbP);r1DR1zc~%CzZtHms3o}`uHdsjNVm4oP+)P{
z`JGyD%@r>TEo>j&P~}&u2_+w#9E3!gC;qczNADytLU-T`@kJyRT*h;csN%HlnoS7g
zg_B$0<>w`k*4=EMW+)nr8&^yo&va8zOD$HrsAY$`6t}+^D2dfld!h`5WQu!ojSqu1
z^GH6SZ`I*O@prW9?(v$=k}Y@Wxbc95U`?kvsInAZy}O<mcYpWI<v&_E*c_S6TCb&`
z)t;NUp@*)MpEY|R*F}z1B@XsFs(P^}oABYzznw~pKk*$!Rq}Cz=MC7Fg?9!CSnN#P
zOeU~pqPrj!n~f`SDvj+~AB;XcKwIl=JBB5}GR7Iyt13nA{?t1&(7sU!C+yCCV70Z<
zRohd8g%|0X@QH=RDI~i9m=pOE#a+JmQMwT;-7u=-0_~bH21TTkTt4x^0mtT8&Uc1P
zxn_#oQRz<4t%Uw8Yci>U-P}<gUA()i&mDj&jzY-yEPj%`LQ{h}tG#n`y5{yn9>72E
z^%~|gTh}yoUI+(@eQa#B=OLYN*{WoaW;i@Za_JRvK1WuKG31`I-(Bbh(Va-#*D+XC
zC&O?-(BQq*pnBAEo$cu{0PDx*<mFTiS1JC^gw_i?9orUycCQx)%`L4D9)K5bkcI%&
z)V6Ow7yQRq`tw#`S{KL>0W#79)zj|w!g1E6Ip&|U<A&flTB`Ilj!9?f#-25wQk=<n
zQ(M^S&<j<htTslK8{}CJj^57923vB)(_FJEIH={1%%4RZAhhuD_2fcNH*+%)<^&db
z&%C{*l$6BPQYzXr)6K5~Tw60UJAp9!GdVfVSCz3WQB+ABvJ-=P(+Y3QzCxnaNX5p2
zgyHGfux+EXIpyP%!@<TO6_cg0rEN3jeIB<^?LGX1gu52=<p8vgw>+SG#CkIgUJ75O
z!}y&$=WU<!WPLQo#H^p^yZfgpD05?UyF7Ic^*V^=3;UJrQz{9;#uk$p?pH?!Dc0?6
zP$#Y0kGlw{jS8OXZ}U6X{<+pHx2ma6xXc2qc6z!lUdZd}V{A_}R_|;2Wv>>`_MGom
zh2vyca?|B9c%xQB8;ic)M|GMSVwl$bDyN(lR9)l0JXUjB4%--a;Xv@WdW3D(p7S6J
z+5s82oQRL4l7ds<?7$B?%YGC|^SvO<N__IfeCI=U;%RN2Qj){n^eU1Au!iOgK)Flp
zO{UuR7@gD|y|E6%f#bY*sem03l99eq*erqAo_i%Iv~T2jwuroN9vU=PZqO@U;dQzw
z<HPV)x(r!8@i{Kl7^;JOPZjaOpEFU_F^?N*Z8IA#*2o`6Xy4V^ordx(Auo@ZuHDjH
z4CtkxC;Kh(>JwO;Vu)NdG1aT<$gqgC`%+oQtBuXllE^+nhgGvqA78Dr-gxuT-#23I
zR!X^<^ba4d@20iGekG-R55kxh+;p+x!C{etlg||b@(agVV~LREv!@Z0!J-(>tc-v<
z?ud9+LcZ0nd**NW`mG_q`2#~2E}G_-arr3{r^01D&&^_Vy`t{Xk>lkEHM!|}n2~fx
zry8)1vu=JHP)(=CTdYUrUs&y;X5E|7Ki!I^atgJchY#G19ZJSlr|as|NKp&t?GZVa
zAT{8Sf|WL8q3)s*8ppY0z=qG`qHCF;r*lxTLcx>HtRR@ilI$gY&9AST?&eG?lvgfF
zp{@nV1nug0HpaQS!c5c+za;K>E%tfc{;9cVy<M(Q@}WQ1?rdSkZg7MIk;P;mbj@g@
z$h43u$NO6~bRX^vCnQ#y6Nqn4p>W>mzXQPa>BhRpXOBHkmKK4RbNV#I>SRxGHkWTB
zLVtDbj=Y~FT}yX%t8Bty$L6b88@NO%-roMC;c8@o!%qXFH^F2NFHf#lr+vJiVQi9X
zBP($Zb*tK&iPm*FUdz1mJ@F+%En52*&GqZd1KQkdhWVrhoUEkDaY=fZ)LYkJg;{Hh
z$MKxCPJy_`3vz$|_C?26bTu0*fyjwW%0G(VJF_?1IZwa5<|Xx3^3?Ot-0$ds7K!2A
zb-d#|e9h053{p(F-`ArjFV=br)am$XB~I&k<$<JCxg<bGM6|(kMVDa$fr}o)wfNWM
z1&30lQsxUHYxtZzkOddZBSPp02Lo`j8jiVR?u6>Xf{|O;D8c}#2Wq%neTwOg`FK-|
z9L@B^?*cW<T2ar95#Vn3ecP=s?776&aSX|NbQGO6o@Z0Bt)c|U^sBH=5py|K9~EiR
zm12`JSI_^sGSjcX;$&V}wnZv8Q3evYO}Z8QF8-J;gUzA<7K657cw-3a$WOtx(3%tT
znH$lPm>yzb%O4g`8JmemgH#%F4W9$5HX4;re{zf^g=EsoT<Vs^tS(>0AFTN&8rpgP
zp5C8NpbeqbT^Gnx(U^TI6(^o8&CL^A!?XxL(un_DH=G{<Frwq7h0Pnm(!Jn62S;88
zW%5}WIc^%%J@xsa^ye4PbJ}(|DLFPrW=diNj}`}(?~uG$bwL@^bcw1kOIQAKS3Gn*
zq>a#~;#d2ud4FVlp>p9w80ZHZM=gH?lPHJ?GK=P3@{(d<T1<=BIBlw=_pHi>i#vTz
zi{?WBAwlt1ax$^l$Qz~L<<^#Y(dFI6{A1b$PS;)!lgVwddgtSPJm}nLRd_m1#c^~)
zg^L+XljbgxaS!CJX5da2_J`$$x|j1C#}|f!Tz9lk=5$Jx?>~m3^g~pZPjBzgu^0(=
zj`imvp~_5|6b2C%$miUZ>R0fnJ9XwssaMsUTl^_SpT2f^=fZr-u62FE7{$3Ym?piQ
zAs|jKH9u+1D{(yXUUuT85+t2g@!TAy8$m!Kr1Yo8WKx7?q@@%_`3Jo#=c)kO&os)i
z=dK~1q|O|35}Hg2E-1jc>%^Wnms5uNHpVk*xdb8Z%Al&tP1ehB%U^2Vk;x<Cn&y<k
z3QP4+I(Ar8Vp?uI_b|vZRmtK<lYJb9Z^VD31X+pod0y`g-_|ZzQiMalg;Gh$@mQ@G
z6f&=*DbN#lv)}+%dv(O7`VR`ReRwzfG`$yKZ4Ibyvy9Qt8_I_3)l@Hz7el&0-+|7Q
zCM)z>N%f@$07(RvJ@SevJ{R7Q3|0z$4)q5!zOb0|h>on~?e&~h{&J@N7q4dY&f~r7
zSe+&wBID1xz`GV*iIaq8%R3F%m+p=k0U_|TbDLmT#c*n?e@J#et^Rt)xlGyCUsjU|
z`$;YyDl!?V{v|nWr{U4TqduEMm|1C#dc0cPY1`&MWWaNh>tY>7h!1$M9m6O%%v*mm
z5w=-&sFG@F>OEqG<zL>CJTTUvp4&*PzX6Ja>Z`Z!8bUqg)H%(zR^0!orEi#lABsT*
zw6v#OmXQs^sqq^lpbKL%e21IB-sT4e53XE48p?CdzqiUbTdf=JXu60-ozy-#n0JG@
zVSW{d3BC1CKDs_cM!7PY245{XNGTz;IG?YqYA@LBd{)s%GV8$8+37&IfbEvSSb@hu
zysO+&SxlcoSDB%t<*+PAOVdg8+g+R6<|ogcP|b~WWuyc$N%4snr>3VwL+UWUv)Mv*
za9pKzTKd#)K_-4$;Vl@IJ_}17+aCEHJ92^PKy+rygVnr_>XWW5OCZ;EB1jg!%vC~`
zh;d`3Ulq~eY-}dNtJwT^nsdw}G<r71Zcnj)IIMIK02seOpyUlDgN&esSk&j^ovP}g
z$D`F_Jo&Sw74QC7R7^V2Fv;Dh1!}3<F(0ifQR*y?TJ~jnUM3Q^YciCtqK)XjZcZf5
zDoHLRR>c@Dk9cz74mc(QdBv9SoA~L1TuOMC$5*-CpKzIA+iiKt0lzx;!b*^u24vgf
z{L!xYby6MFvr~m>yi7A|Pze?j242^tVgU#dEa8&^T@L$qx48LH%!M1i>Z90c`h6FI
z?%)gUCUM}4u~_Y`bn~)5K`01cGou8GFwfEEYt6Rn3BP^ykMua_^`Q34S5kf_>2MXn
zKeOw_2W>n_zcl8`(@a-$R(4siX+=A0-EfteXCx41)EVj0_R~fc+r~a>*0Y)~=LCUB
zH{fR9&UhwR9sU8pDIo4RxDfe0CsUOG`q7;}Fs+Mzw~Bs87EL{=4BEnKf9O6#;Ja=)
zwxQV3{x^bK#Na&NzoU8ikcXk&+Qs@y>FHpAzwm5p+u5r7l8@U#o1-B{&GgQdk@`@T
z&i)+1`;ckoOOtAa<(}mW@_%>{GlkmTy^OUFhT)k^da2_KYG#v-;I-PAH@U5;@RV2P
zvqdRT7t1UmGXB3Cnq?HN@RX{WNeaQmEAz2h)`K<y@1$pdS6_Oc?C{lgY+B8oaZ63>
zewOhiU{d_uIz8?AC$ji@h2)U+Q$bMGA3vMQX4~jge66ybot*Tr)6nwTcM)ZBT`>tr
zr|o2+(=dr}-6>wA=g;MA&N%V>g?ys+8Mpz1Rz(tD&p`g<lhA-P*%WYg*pwGjgPa3P
z(hJrgw={{78ex7JnS-A{rOcWvM)T!zwc4k)woxVN=&eiarpPR2DRm>wIVtkBk$#J`
zKNOI^m2y7Zm!!tc>+V^fDd)85Nf6(44xW|WJjj2+mB`caRmOnI1QMpN<dX=a?$OnG
zF;ImUad|?MuG%FYiU-~1qbSfu#^#Q`VP+7^A9;xy6cn^)=VW3(JHr^mnKqN0b{p<6
zT2Q*YWQL`OMM=jHFH%M7&X7AvO`LhOE%(bi@R6Y0`)3dS^OJ(p(%|=SUg(Ybn=|8U
z@hHm~aWk=piU{<<j-~`@Sln5V>;gO6`QjSu)ztLaTCqgTRE108b@{<TOLxn6cDnUQ
z2GtlMB0c|84MR4RL<E&fx)R+80l{-vlsz3~#4qt7Z$T|VdBwmBD|DBY4Ni#H5gpw)
zZ^bR-b;QCa>-Pq34^7Y?8n6}%9ByMXYE=0p>wAPesdFD0w+St``XPgy1J^}5uYVrv
zKp~>GFIZlAg$gapsYBIu7p5>)qHB*({%FR5Hab7NcRO03EQn{TZ#CXl!!W9tsW^2L
zj|g}|&?deV8TC;`tMhAMG_UPG+t+#_OYJfgeeO{-Bohr&;;7d-41C(n=C<bpO><9Y
z_*IGh>&1nDpup9x9&Mt2&KP8}htG%}W}#<7LSoN%-(5T$5*K&QQkB<@!vj78($wGu
zx(X%b^RrM)4}=NBQ}Vcu&A}`V*97mhTq&vAi^IaL2G{kI+3q0t&PP2>qIuTh@9ixI
z>wa=UyTjj*D?&HPd54UTeO~_avuE=Xx7MexWx)P0rtIuK^HgO(u}Wt|;siUw31wPf
zinU<o0lY}0G37gzSKyBYTM@QF;j8baX>EJ0wX|;sO5NgobLhVDKPu)w65zLPWy=pX
z$fZnkuz%9k-PS%Vwu&ZGcYm_v^h#{xgDy8uqjsiB$ublJg+)F=(e#RY()sP}Dp>u*
zI*d%N1v1Z!?q7Xg*FdCd+^51PD5m8~+GQ4hrmpyR2wQ03{K?{({fthXr#IKB-3Xab
zXXAG&Ht8jB8V#|d{AQ{R$|oBuU)ip@l8Ta4h0Av#VjoIaxW0b8sYe)YEP8066`%W-
z9qzZat9s+A<*Tu;Ot)?|D8jm*I=d|A$Zr#^tbQx?<wc<n_S@^WET8@{-`*AWp_R7p
zWZ%uw>(olz>1obs?TLPEOLwSHhN7M{I|D<~)1ALDD8gD}lhgGQ`wxA$LIPGN1Lv>Y
zZeG(Cekfi!Rwo)v)2?ghb>vO3+hxv7Ex~o5q)ghF`WldbwZ64t&u^Bbt=4u^`N0dl
zrs=#4V@9v|=Qjfb(zCPs>9_Nkqvgp$UCqg!0kyTHHrgaf+XG?!H}z*@<1_UVtZomt
zYVq4$qMOFOE*RUJC^c_;NicS%*551W|97*!|NQ%Qr{N3OZhu+m4El0yZ0FnLM5d-I
z*<840G1e7qSVAW|bWyaG@n{eSrpTfFtr!7<N9^qNlf{8qQC`}n!84|9A*ge8$|+#m
zVJ4-axn7RN!t2|D>vY*q^pXYkU*HCOpM9NL1yk>vvqPz`dGX#)C#2@veO>KM1@YTj
zcb3AdG7N-xvvMN8?RE!m?L^(ve0W3+*zXgQBYw9|JCbLM*u_N(E`5JSwO#f)kzGe8
zyXH3jfr(7`()hc{LN<NpOEB2sc~u|^s5Z)`Q|yc4_lx$HH0d$3@pDH$yI#5u3qqZz
zV)z#Q>#Dw76EYjnM-+Q3eUk4@>dgDYRS{8x!WQtI*_zp-uaI_hRj~UnfshJ4OjhHJ
zJS#PAle4hdU_MB<C`QDMj>TzP_9&v%Eyu$2Di_?OQ|Q%}rJE4X4+e$q6~s<sWF(B;
zZvWAipm&PiqE4k-#7CP388H)TQIB{B&zWYAsRqJVI|VNJKb$$ZO(8WSq{*DVurz-I
zYvY#UJ;@8Ccb(qi70I>$NoVA#u!x>suePA*4-Z2E<ikhv+KE%XaQEkPpV}Fw`S1i?
zu)r@rzKmwSNE9Ot{NeFyknsN&(D%<U{@Jh}z|oVX!skBPIc6A#hbwbI0cT?qL*uhr
zotFR20$9f+b7!r{wzdxVz#d`k5AM`K7azG`87gMIG(3Fj2DuGGvspnLU74J;QqOHK
zGz}U#juu&MPA7PgO{S^j#ugi~Q2*>i3wEIAg|%kE_2zHqeNB(r^a2#-s&>6a?%zGI
z%%juhfiVpm61><TZujp+kS$x3`tj;-x~9l=t5TEwdsKf4aXwmBI+xC@-cv<bnQl)>
zymx!JLPI0K4L(ygwm<9yODl13JO0~&7JPqie*G{(T3(C>cCc3T3TiZ~eaJ7-O!=0Z
zZK1C_pc~y%c{S6rmte&*jweonzf3g0|Ls-qLs0uStRf;|JxAC=?=#d6xBF1<Zf(VB
zsFssGd|z*t{EE%+Epf1)CWW`WCA4~mYZ?9FH+|2iFs+_$=lpRlGmL^L0_ddVmvt*`
z`=yi)jo23sG#A7+jQS}RJq`Umkt!=57Z1S&@YDPbtao=}Hryyt{xpsKKt4J#=$Na2
zqe`Dyf6Ge8%L|?SZ(V))*qh~bBg11kW0y6z6WvP5r|#Nnp+qykFDkv9%Pr&HqUU^Q
z^1;vTVCk>JiTUu_pf#6eacAyN_1b3l<>51Yc(wY0`4^-J%fZP0U<@%ysX=3WOl-`W
zRjI?i$W~t7kIr6`7C$TF$iJITLYE7;F-0ca(ug>->r1xGnj7^)ii<P$A#={s`;&Wy
zqrJz(X>Cg*sI`N+B*dR~)b%%m<s^}niO;+}knh!WtF#JPe?CXgw4Q7Z+lZ?q$fN;F
zeKY9xdd=>+G&R`b$;=W53n`1oS{-1^X?U<jVj~kI=)-<1<(^w!(<JU^EaQ9BdQqZu
ziDOwS(qYDaXyO*V92_hq0s|fDU8Q1k<pWbiE)OR1zvX^YG_#1(Z8qx{UG1EiT^3p4
zt)0Q`JTqVu79EYZu+tgZ{K#<Vi_+6XokUGu=nE8?3hP34vum5fwgY@{TZiVS%-x^W
z-S58xUkI(Dboh3!gMk^+4@aDu`TgIP_VRJl^?&vVka`}WBp>Yt=jYtC<Zw;LRk3Dy
zlAtfy*H70g+gP$oJaVpUc*gDSuBKllehS+hcVfVTMIB%f&*^BPHILXVPve=gvUAO!
zR_M`In!HPND0$0HeH33<#KFEO0;SPayXtj`e*Gq2#XMNy4dX(GFgB~fA@fgsbm*gU
zqv`m<buvmunyB0WDX^jrJHF^S6$lNl4Daafl_Oqd75x}~gl!QF&j0i%D&HVSIG0J}
zsIl)Nn`Ot#{j;}p@~xiJ<2Oj?)1Fn#j+4ZH`Jm*KY#%72QSOr5doFN09f~3_Rga0a
z0=aVJ4cE)%E)r(OL4>$Fvhl)DH@Ob?$De|~IxC;*rIPq(cx5>``eTj?+n>q6j{ME~
z<WsgNX-&3l78L&8gp8|@p7EXM5mCLn5+U=St$P>h1&IZ){_xjTFPNJ(uHa==#zm%c
zisz`6-mKZ7L}Bb`uWl$?iC<o_O(7p;NVhr+pUyYTNqgMmNBRwPcmAw^I&)@+o?e`^
z^ma#rhm80YJKn<n{ND<;Ls;Aq_mbmlOCBTh?9WDIZ&W|37uxBep(L%ExSq*m4ts5G
zRvW{+U*1ne8ZLi=ZQGlZ{W;Te@>^OJ@BLcA{gUD~4SH%s<@bn=94dVL`+Wn=-<)jf
z#A$B{j*}u9{P_bm)%q`6)hm+{ny#8@T`$kf&^S(0ee0OGXlRUKQFgqP5k=jWDCxLQ
zZb{gG7ACvFdvBsG$I<?5{o&$?iyy_wPt1{R1M79KtzY@7WL04WgGhy8Ma!OzTS;ks
zZlUE|0<yPtLwQAv(^H^rw^fZV!64yY&F#?MGvLFP5#G{d7(zv(tdn@e0qfx5b*IOt
z$M+nuG@e{%ujMfF$hYf!j*h-;Z;bnd;&``}L{KEA&Vhn(fSstqjg~oz0ORvcKvpn5
zE>(I|r0r`u*w6x8b#gm_RFmDAhiS`7lf;RvO60vEgkYwC8PPVOs~f)H`|IVm0+M&X
za@p_K*73x}bQgN>tP(MCdb=d>28yT67l&9C=NJuFFh{kubz_<%A7vqnua7h*xNP(o
zAU%!E*+b?~VCvQuQ>b``fQC?12pz1t^|8u=n{x?|(-rj&5LJ)G#~w&tmNV8pYwKlS
zb@lSuVWaxro}2ZT+(Q?^;~~~Vu=f8T5JJm;m*3ynkE+kFUC=`NOIPDZX@35Us{$=;
zh}Zs;8c}4a<EFE`m{{rgMZT6@;+6K^QZ|kNo8#LyG%D|{JjMX68$6QG<RyejLwH<7
zSAFGoEjw{GqG_!P!{g0E1Pv8}7su`4moO#|5`jYgVDtGd5(xzyB7Ju~6n_@~S<YUU
zX0!w6!h)c&qC}Fa^NQP==m6O^JhJ<mY?bC2xmszB<4p;r>V>aZSlS1=w+hZ|%6CwH
z`<t?pc6ZypWnXOB%CW4&X+Dj9{a8+JRO^<zWg14ykGSjX72LQa3zG@k+(zfN(@-zB
z^Yfa8VRO#d^Ya)IqW*>|_4@@XR+QRjwj8XnJ+umDa(m=Y0<%LbrUZ${oBXlmlNa}v
zZD-<G^GEhE+vPbrc<O5746F=AtUuzQXW-)9$4k1U@^9Yts>kH)oQ=*PY~mp$U<NS#
zKal<9<J)JqH{QqSP5Li+Ys{C;86I7VA5)@g@r~i=D@vL>%O^h3%Z7LGXG(Cl9j%G$
z*|E;E+NCcqr>f+bt<1gSr?3%7ASK}}Bv!sWHZX4xdLH?A+;)qHTUoh-fj+Y;@M&ZI
zy%)or56teHXWo0>nY(dOZf`k-I4rFYqYqOo$nJv*P}i6*X7lfc)xhxQLy`ABH4l?)
z1u9Ub{C+HhXLPl_<tjw=Sh%KU_p8e5SAQClwjwzI!Fas)*?DxoXlzVtQSysa{#EQy
zXYy1D+v^g0NtUp*k<YdjB~*PKINS_fS`OBl)9pP&Lt5>)r6v6+Gv!>E@Y6iG9v&h$
z1m0AiBhsaZCx?$~U%D%w9@al*+NeR_3hKy@JWFe;tiGAitdG~%c<IKfx`R=2j-f5$
z7XKSN(tq0W?;_|St2X<f_^FBhw10i@h@)LmIpkATNN|#<E?9lB+)(3)qYOp0;X?$c
z)f6H^ifSw_?w5XnV54?~<cL#fCc>(<to(Q~2EIZ*eB_ojmHHZGtGg_k%@&})-!9i`
zQxn4iPg`3NA;VO8m^r5&g>i@d2s@n~w+Rk=Mlec2=}L)6XDLl06z5<Mg1Os5T_v>>
z@7ImmFQbhhR2GgU4J+07BMzcs|B~nVM!Y9_^EP5>{kBZJu}1S@>Bg&yZ=x41O#_o_
z_(i|*FG6N3hs{m{m5-$-69!7O+S!&fJRDeqx`@V&qoWKNVGGtK9X|-_1*TG%XioE7
zd&1R^=M7C=ZfUJi!1>)F>?v1vTRe2*<Dk(lT3X0hu&^Y>zx4xTl!5RQ#L_2KNlM;K
zGvC<6-0q1#Zn)%|H?M2*FaTqywag=$+YV^Bo*tD*Q3})7`zqUxZb++5^6pZQK2On>
z=T*RF)wyx!GT0yU)+SO43NF=^Rj*-Fx~^&e89qpeYHmg;B4|*N)3egtZRc8W3iU6Y
z#TAd{@Zy-crWl4Q(#2e!xyod@QcLTFfII+^(g>~9QY%vvF(oywaXDN0@r(?VvV$W_
z#hRa;(jgK$uNb+aBJ(V%aG#L($~&H%X4&|^lMN0g5v;gu93SzE;oznIDV-;ak!;k%
zp0zzr@Na**N&4WUex|Kx!C}ecoPa%=zY4Yg*<1NFk?-K9xpaLb^H`s!rutmf<}rR4
z%=6O5RB6D~aQD>$I=V%xzX_z2o9inf7P-;yP{Les?B@QdI|c;XA_d(>kGa=-{AyD)
zMdhHId$#nX4+k7Eq}3iOp*O!O=x$enn9t2xWi@)&06pF`wh;YjMu($0URG2n9hodC
zZPK5k*qffdZaF$L9^e9XF|mC2l<E&`3qE?8?`S!+vgy1kXny~lrPr%K1>cfwaSWI_
zV#L{6EvokUK!95!-zNx_Xo`}ZAUC$Mk<k<xsd?+^`m?{)%@P6K=2r|mIK%NHP6vf`
zcFBOiDFu87Jq(?mo*1!#SgEoqSd2Y%aG-q66#}-H-vyLJK;HP6N65NBEA+Cu3WtbL
z!5G37p>+;^ACMXzb0q>s-D_u}y;@z~04F(b_V0o4>AFsbrImH>R9Lf{=Y<NXRUT&2
z?DgE{j<~EAUzq0j(UGN^%Y8EQq^dY%;D0Nf&)z95Zf(>3=ve(fcOrn}kSfZ(L^7iB
zrsyav=dqZ$aAhrV4G2Mv(E=|nL1S3a<0NwSpVSIHoS&PEE)4+H0i*72gr}ITB5=NH
zz@=076|IPk74kh9TTn259%2$8lJUB@=!fVNDL_0H6+LigW^X$W!bAZ?qp2win)$k2
zRQ%Q>!&te-_K@a|G`P8`E-?HZj>Yj3I^!R^tJ+bukzK>5Z*q%%j3;~aj%rspEaW0L
zr{>I!N1xqM>BI`dyZbdgp)k-razcrl)L@^O4LxBF0Z2nx<6|QtLExl~#+I7j-)!=P
zt)+rDw;jWDwFf>?F+K5_rSvUh`csq|>(>4CKSEcwbUqa!i2#ZjnVNTkChKUi;ujC2
zHj$=&;wwfPq_AKqO<YJJ*vc}mKE`3dLD$;)oZ)%U8gZTz=W3cL+ToG<EN)V)oS^~E
z3|hhWZkED1V|#ksx%^o6vbM9zYTW3s#$?tf0AM#9;GPaBI?17slW?h6bQk?%w&eHp
zBja8qhRscD3ij^(`jp?E{DL=#%(S=UP>nS3;^eJsbJP9(fHx}iwF!xvs(4Uj)nWJf
z(Npy0jF?X2Aj^5thLt2&4Q*6ahK^9`3{VteskO`ZC+qL$c^;}JuI|6Yc)HWnC}Xfy
zi9eeCU^hftH92j5IGON&1#owFH;fGLrW9y|L_D`n;iV3b9i5jBzk0s-jSw>9uP2@g
z9ab;c?+swTs<glcSZ(b1CbglJVK9Vi1Y><5_L`8T-|Jip_i5nUaOv3D{HK8P=ON@b
zET+Fzg~cP|+vSVF=9O^SZP0c8q;*^GeN4AjwmS8Yh8ao$(~=|PCKRF=lQ6w!w*Cpf
zLF~ifPDiOut=rRC%}`s5<wA|M9J#B`$N-N92k!WS=>>Cn@-R(aAW$9HHR#D&P2$F;
zJIC!d-_(r}#{aR9P_Tb16p)n5{Y&{;`sRu*qAX-_ed$BtpYnu|>dvuXexC4{Ab>)(
z>uL@?Y=4c6PRwrMh^RO-gZb^4|89IyF#V%KPPGZ-5W+i*IUMY%IBBC5L|G1+i@x$9
zb(5)ASw7ya6SIK)kdns^eprV68%`cxJE^@Es?pbWPrjQ%j0`PXT4n?hZ9&`p!Gzvt
z0f>*IuBE1rlZFG3et+MDmA+;p#-pTcJKaj*glI(f#>B_iGI59AEoj2N)W$_1+zAP~
z<9t~^))2cnOcHiGnWX6Z%povROtyQnk?GW2Lny)*rl!=i5V5KO0(M4(j4SPgNz3~<
z&BOxjmj;HrdbPzK4;#ckur_EDCNPy18jsR~-6))D^U1)xb8TmOr<w8Lf2aTZWs1S%
z^ivT$dS(WrtEN-Q-d+PG25<T9I5=gMcB>@)a&gEu55Fm~RU#=5C<eFn=j6fShlIUa
z6{qA$4^^n8w!rI}4il~f$Lj?yczJ3eCzi1seyyJFzflbjp4uWx7mMd7)Rk_A!Saf|
zn(+si81k@u{;z1naRC-bScfaz;gLKt$*-Y~417%V2)d77XXMJuKPa4iAIOG(^GHEY
zd&Xx<%szn_h%Jglq0~CFh4ymzU7kM!xU4u8_ZG5)fiY0mF?ETJM4gi~WCL0TX&i4@
z1=MdXUz2!k>u&+wqHCJOW6+oiR+a{)hJDwsooKNunDo9p&>(Vy83k0!gscIjJFt-D
zJ0+@D;%#_uWd%gGxpOKZm)FgM99rDCxCDznJV?Pq5#YAY>m4p&c6JsJ_96ug1i2;b
z1OJ-L3)+#xqtK69IZNxTX?wQzL&UT&X*Z`0W0VyNp&W#sy>5N|Gw}~<Tm%qE?d=+A
zU9ms2|Nc~k@FzxC3AFpI*9$I13;POjAc1<3v;F|8$m<Bnvsvz02M&SMMu%F8xEdaI
zmnB>2=P&{9xi^^CYAW#IE1O#!`eAduIF5N6bsv#`HOf9YDp%h}ko^s(542PRvy%}p
zBR>NJorZ_Y0Q7Cev0<spJXIs(Y1!#JJax-FR?bm+g$jgtlJ?Nl&}fOe38^omHILZA
z^5J*~+GQKt_;rGLU9gwwE0{DjfZ%&wKi<cTIOw5bIfIK4xU(w*GO{-uWd&UnBuo4P
zcdv2i)Tomx=O2%W*=s+;d7wV3i6}HOF>}kfmo5FlCXplV+3CeF5lotjabqyiEpZ|Z
z75%5o+W`kAEhyCIgO664!9_@&PVZm%C=bf0!(mK`6E`{xoeI}S8*MEqT1EPizCP8T
z)QFV4fNHf}JJ3>N%B^;{iTm-OX8Nkyi^uEApV!wLfe5T8;EItE64!dFWX4eoU7`Xt
ze|~eNa}rx_l;pNoba3C%`E^xxjFh%ZgnB_8c>xzm!SvGhqUkS`S;0#VacC#|->xpV
zqeE)2Ex{Ch$+6NJaZe!H*4ppsy8&yN{Q8Qalm7X!wYxA=!u7Kr-Gp&x_9*VqBoSTY
zhgzllm8$#gJPE&K<5%>$li{BDwf|2Orn+mwVL)TU=65kNux#o4D9TzOk=V`erlGw*
z-G_=&TW6?RhevceXr%Aqi{{Eo^`pLLGkU6S5Qqt%SC=GXMH*{zN4TyZkgWa5=$ew0
zZ)_X-zD()=(rRw=w%)m%kw^v6=>#fNwiDF@na-{|3kW4D_(f1!h}?i9D>s|r@S4F3
zsSKT3qGlk-B^DpnUl4rZcJ~|NuX4`<LR{l$Z&d1lQvFn!s_E1eOLS`g=eJ=UZ{8Gs
z`cvI=Md~cJDsHWa=sZ4M+dR@N4JgCq6<4AInjUYmS4yee2TM}Z0=?FWp$9;go2v77
zJ@0!0rs^oEeSFCAqF}tGV@V}Wm3EMDXuo@0ORL0c|K3)9o(2JVYh@3AKu=-a*SSPo
z^uU-(6T(^jpMORMe6|B;TN|}*I~8+OI(>KR+nR<V9n6T1^y<PQI;wZ+-i3)Y0=~Mx
zS8xrcYbBVdHR08_qw;g>cIJV@1ESfIho$!>*Mf&v`mwfOZ12iQ%VP;|O-#?7oPLYQ
z$DAZ!`ZF<=0GMG+i50e#SSxgLs*~Z<Q6^d5kD#{xwz&M^?f=zHZ>+F6eDWW{A>;sf
zvK1=(LkmZ0N7qn<LA|o&WDN-9w&vQ$AfYL&{H{%x10XtUL%abAO9vG$!BGL+y~_T!
zF*70Qy^u}?`Ipsi{PUlzA&L}(6Xof!G(qrg{7FK?Z_brTEB=pg6^y+Twl}O;eN8h_
znqTvQwpH7(gPqSB5UWXLYHDN(>2%Qb!60t`K(n}=^a|yw(mk}dQg9XCd;F(cs?4z{
zOC)2#_gnkp3nwOu1fllWNRDveb28Jf06i|*SC7iLl|rC=^(An?$Vb$4T+7Uj5QF{w
zd=ZNb*rfO7wv%+nh1@X8Gik1Dd40Hq3m29}X?dO{OUino0yD3Py;PH&yfr&`lBmd<
zog0ku%4(0Vgz3OkTU^>a4i#iw-)yRLFoJ!SF$ED#>yH<{ezvJaK7&Y>8@0hnjF8E!
zjRZy3vI?g;%^j=HYn=F`of>zpKWI>&B_&ben(MPedFgUnr%C1bBpxuWehHVeSzdE_
zJ*g4VfN-r-+k+;<^CGzgC{mxV59y=y9*e#X+dJuA_c+}egkFw*=Exg!LHXkASv_hY
zKQUCMi@5#O=*fWyVY=eEK8`-Cswt+WN%~`A;cAI|Uq16=Udx(1_CP7s!G7F!o2Jxd
zhjMKcK`JAglJDZs>vf2aO73UVNAs^BE%3rHf^+_Bw$v9n>-@{(!|FQ@e!ns@tQC}2
zN2_Ug0A$p5c6ZDd?Zw7sa9t8+>&BhF_teOU?eLK9O_-7Cr-K?goof91e<H=FD8W8i
zQbYA{T@eu8nP`5&M8a*wqo;b`*_@=gJ;!`D92Ta|>%ptDxu2n3asMscw{Rw|#)$~8
z@1D^*4laAC;u~1Zi|K1l=(-w420lJ8FZz0>VSi~M#b-mjLfCbZfX09&0CXxuG=7*g
zRW~O3(&npc)U}0`ri>uQ@?_qwmYF*9)?HrpVI+{Y0j{)LFZOrcRe58PR&dz7{MZW|
zxi6IGkHFtO*=Yk-&|fOtwhGHeZo=kqJZREwFj@ZT)qR0DoQrLxDC_B7!g#LI`3X5L
zoxz*?ff=zxg$l&Tt_WH~qZH3H9$i3}B+oqYk>02G8@BoX?x#Nwp<OR9-CA8~vnj0m
zu<N|Nv303tPRwey{(z5vpshdU{b9ATyD6;}?}P&<T|ng##ubJJzL#*>v30NSZo^8K
zoVI4Q6F<;*S3yP6GO}W1j1hLiWs!a1pAl;yy!7w1J;sg#x*B#1qJ>J<n2oQBj~kv$
zPQIghO-;?+T881g4^j2Wt&v#76@ZAd@zqq>0=K4bbCigDDuXU_^piOv&ffPPXnR_G
z6Ws8^9Y9oQIOR6?BexQ27BI-OG*vHa4Yv!^C0g20Z*RL_bdC&tckT@V;Lw3u=qoeo
zjSE@fxFh($AwQ=W69E`W{<@m|>sDJY^gyco$_Ku1bJc=km7lp853yZ7caaFz*qfTn
z?vU^3bpBmff~{#inX&iv0Z$JUg8l3Q!7VeZRu)RM(jpg^JI({NYEX!&(H{Z>=4^73
zwZxaj6G&n95}5Pg8bG>8*z7o-eAe4J734x5*L#T9R&u*QjirZ9$;uv>kop>sNVA=-
z?qAG_L<CbAABB}bhq&6iD)rc3((#%?_#d~2M1)6(N@?=8iL+TUF|PNQa{}x6-RQ`1
zVT$&lCs2;C;7K(=9@J5gznP?RO@XJkxrcVIEO+3IyUf3S#NrQ{o*Yon<hi)cDy!Ja
zWK8|gTnMO%{wEg9--!iXsw!&XAVLux?DTE3dZw$V#9Z~0xR;l*Y|kemw}&Wb(eafL
z?#H~-(o&wEp3Ize%c4v%GMP=0-Q&)MAJRp3A;?*B3kWyO$BXcjPD|c$HMO;Tf9-pM
zSy=M)Y3}OanfnsN?1`}hC0&)85BDo}nN1$Hx%0V~N!SJ6Za?65FRS>Tcx83__|zEg
zEE!m}PW^tjg7$F4g7kEw3bF*gXm4y}&=^_o88c9!@NI%K(l?f_P&W22gE9@5lsD%r
z;Y5M1C=qn~hTa(*rb5vdx_<vca4qM5i|{l6EJANyBT7nuy_TgjdSh_KsQY=V(xd<G
zS5%~BqzTe!aTzkG=b*f0XJ!4&Opl1mYr3bzR$^OH(cIjTGs19sY@l}C>Fujz(kHN^
z;vlbvRsd(9W1tme!OoEO1(HK~6oDjhjATWwyvJ~N25V<-VL)-nCVg&wVQ`%@F?=O=
zQ%5d1ICnMevMRUGUR9N1L@RtCDl=1#OJ3}lXC)JnY{PC8QDaD)1}T69ztP=S2UKK<
z06!u&In+0nXb0;A&GV!Op8U{rv_LZDWt~RI2PIfuML%ortv-Kq{kMN-^g&ATKT|42
zjXOoUCDqlD@!_Le4Y|3sCNGklr>Vc?IeI6<jyYwDdB)UHdA@6*4R#Pb!Y9C2428qu
z9hn?}*X7El8s0}dkC=7E_4K?hV`S+wTt4yBoUFH|HT!)s&}x8Ilgo#oP^loZSj~HL
z5?!vj((g%jcP^~zx{u>S3)HI6Sxjlgoi;o1DE)AjW>fX9c_+qRhDpL@o@DPp6B|;@
zd)pf)v^MP+92}$>=p`8$tLumdp2bqzKc^qw{;*RA3fAM$V2f0D6;-s?eeGEz#VAO4
z@Istn!uJikt10t_i(McahRKR68L*lMXMW6`x9H7^O-pO^oKlczdadEP_tqV$pRMXc
znXW3+cCyx6+fV5zWONGWMSl0G+3gRW@`#f7P|Q!0bmr4k6LJoY1e%zQu6n=y#I~sB
z+C}vL8&A4nJzToS-d~ffCR0(dXlYRQ78(~hZmoGIty`27^}g1{5&`EN<6u8EEs&i?
zoNYjQiILZvD7w%aqI>N3{?P(_;?#1qTyIgY(k`L^S^K99TL=QYl<Y{DqoW<<SLmJb
zJ{I%BLEr$L#p4z)pCYl#&xm$o*hbEC=-Yk(W-YFZ3`-bZEbN;sJ#}S<cj7)i40cC7
zt;jvYkzCd*T1sp^R(b2o<3V0Mf`HvC*gInhZ_mi=v9r9rj9~f8G@zGuq9<R@rIM|Z
zAUmSRUr~>xq8}mjb)xIxgY!#N+qXhFVT&q=QPoFIj;953Z1MnefJg{%ZWlLKv^LMa
z>k0lsX^&8{Db(Zou9a4GEzt8q<MJ4q^Fid4E1xbqd(IG=*|_-Ub~Y_(dx8U=oJjx-
zxjyb0$Qp@F>B%0>&n5c_US77o$w2F#Dcj<*xu)QF5MYY?;!k0^KC=CpBaZJ9(;ELv
zm~7j8i?IqZW@~H-n)83N0O${IC%wPRH{>sZfJzJD^955=844Ec3s<)}@IRgRuV!^H
zXaRf>aG84p$R{9bz`rswH2x4B6+B6$_%B;~hC!`%*Nw)KrpFu?s2aXY?C-}u1>NHQ
z+Z2<hN!aTjMj_UJcQnE@0K0tW@j5vn|EH%R^1XakdhsHJ6nM!Wr$shM9j52g1kQ)K
zRaDdl#(R86-5;>Gw39qtyuE#q%|R@pJb1n$!UN(&2Yv0ycIzulLa~=IGx*_n?x!u1
zG>b~&1kvmzYg;RZeNN|0Y{Xx<W_rTR>7?wk@_g>mFx~tDuFhTR0;oE%gY}p<C(bUd
zCcxuxy(R!4aw#;e{c)XhFxf6ky4B$d^pavCgMAZ=!o=+iAMM<g$Wf@u!~{4$_y5R5
z(0mSdgLc2c(y8r+GpM^E+}}dPG4+}9P@{Y$P~Rf-_48Rnu_!tnX0QJOf>5x!(m(MB
zKe|C&WjalsPfe76Vr*{XlY+RO&}QMxTo~5z<^H@LP$U0FMjKigytiQ49`&|F?j-#D
z3DI<Yt{YEO-l(o6b*ERp<z@xq<;Ccu`q>aaiMPksp&cxAcTOpg3f}Vl!LH#~8@BlJ
zqubI8tELIuTI}tieXnj!qs^iIfYwN=UmXnJlmCG|6{C-SQ$dk{z|XY=9ZNYm*D7s7
zNV%17bkB6|*NC4#fs>e+vC2})DDbmy2!njRFQf_GcPv}>-}!d^-GZ(+|21Z;hES-W
zfHXwmwE#N_KfAKC0$Dr7qQpRR^L4=zvx?Pk&U%xlz1^{G{?~A|t{VazIcrTKcC@4B
z!49ci*P7m>#V}+vph5>7UuF{G1YvflOgqv$_W&k?x^Wpbe-GPAU?b30R*te9Z%b9R
zI6h&4NfXWHK0*OvmoJ<=_wg=FOQ$K^AsSWVU`VH~*aRQFjaj36Ja-f<1Cb@XQc#6M
zwK{aO;o!76+h^*@3C$pxbte<vwDG4eudI_T2GS<Kqk)xdLm0D^F|=DRXL1}O-_V;4
zPCfO;cFAWos&;7}?~RIFz^$)(0qNrb0^X7c6b(YtRx9NBbL0ik!O&dR!PcGcG0HMK
zBl8)%5~*)^{`JAxMQoO18<Xp2G!zh6k}&W+k()iDQ>J%YJtIO%BI4IFoib|nGqf>j
z1=)`8s<)B@icdq5Gj#q@UqF(8Jv%g$fB>&+>|r37<N2z&mrWGjKQzBufdLXxU5|RH
zZt6uU^lQ)|a~^=FOgvh*tFFQgoeqBQ11c0ferPzD+6&H8D(DWW-Y&TgL)FthKxn`^
zd3L=-3O4wiu*@Bj_+W0BE+THF<zHI(pb-T)VFJ%G+u9uSyf|kkTE3@Y%R!1lZ}2k|
zkF_p95b1C-JLR3KkEvm@9(0llV#0bkhy<K2?}cC_p(x*QuHL$$z6};K;XBAZQd}2*
z6<qDz4A#Ht3<PD*0bsj_?MRL{U*8nw8F!Wo{$*fzVm{yg=XVHnQUBbHh|p#x)q<TH
zt0yL%^OX||2_7C-5Jg9jbG~F(kI{vH@|v6v@8tMF-)~fu?x}rRx>jmT9?CPp3qE0g
z-S_liMHk2CP$O^G9U`hf3lS-#@{%JY?u8&Xku>A-;0w^iMr)*IN}~1N0gh#Ad5<Ew
zs_(ZQI>U3>nQ9VWX)y;v!s-WW%qgrkN0Pt|m11HxIO+9#GrR!}1x}<Zbk+duW&;be
z4tH$i0Fi~MNr#Un`$r27+39kBMmMu>z3b6M40(a|ArZltg$l_NzspZ$0L;WAtWy9|
zN0ZR>xyPqu=<t`FrH$^KapnE8a>K>_2#*@i=$W%ux_rQOlgW_ou@$BOJIOynY&zFA
zk*Rx}-H8;W$$-YSxbg}4g-@vPU)$2hRyg26;oiH*DD|RK9aW$0ijSh72mg#No0eI$
zmg@4QqR!~li`Z;6!7S+L7*<<<$;(;{mb_mCY~-NpZzD?WNmt<;)oKg4R%~ps+x(;?
z6>d~QL$|}DkkmdreM+D!y>GCYw-+ZE0g~kT0$$yI|C0d5$r@}x9k@a$G%fH;gn%hT
z;g6YGSBhtvT3B*CL{zumrSFA$3+Vv`ZNQ~RoY};9nAZ<Y1Nth6_JsW2+%$NgDE4gD
z6c-SnAPbUCGiZv^z0%V)s$vHA;fte70ub&yGMf#lfXEM$G_1Esfm9n?!1q&m--C&;
zh-2ycHvH#opQdT<a=2lVf0-Gn*c;y2#N){bwYU8J0nCP=@v?~868pM+6rc&Pm)~|=
z<Z`oyOY@#AXE{Y}xLRMmpgG-^v;WBIIbyva4V>|RG;2PGudCSP&5|@mgEcnu8le<V
zZqW6uro}<iH@bHP1lZ{e^_?Y%Nk?Wt8sMI_6%ev>>T2&u9&$Hda4K=Fm6e8|^NTho
zomKwWzs7k){G<vN3~UDnykt&@_lN%rx_r<4M>i|}G;B@}lfH0wXXE>W{4F%IK8*=#
zsC7mr4h~@!Ck-{Nzmr4=h$pwj3)T>G^qxgAt!*zi21lwmCt(hg@V};t*-i2!&QWPP
z*n7iwQyl_i`Zr2LUUI@<+FEHv?_>%fZWNan&24eZ6crcs@wbIDO#=8^J0bW2*y(7=
zp|N2>p%?1;&>N7tt}9l`i9DQTX`LLtqhFq*mP_lN5<hx7@)06cA{mA@miItdWhpo<
z>rTuUbbh|863vj^RajX2cZ_`cBe2qd21Chhx$3GUo&O*Ylte=)4fgg~L_dj(qQ=Mf
z(J0<nUMIyfH^ao3`j%5Z56+y&P_|^{&Q}b161^`IF-NVHQnT}hY3hIf0&Z63q0MGx
zTtqvH4Ga11SjfOnbg!{0?Oj(8ZXrF}5(SnmZ(NLQjY+Pi{Y}Y;YPp-q@?GQ!2q9y3
z3<O58L2|qGa_;)c)yyOSU(BvFv;eRr#mJc24#%u@UE|@lKeU)m1F0Oc3d8kc^ZUDr
z=&eAFJGN<fpL{aG{Re!iAW}NYdfYEerIYsBUHO#nasu#dcK#yFhxWW(Cp&{gr{irs
zD2yA^WF&ag$55wYc1{z6<Q)``YJ&(=cKm;t!}rWrT$*^~Ken@w_x6kx-M@wND8fF)
zj$~4Unl+!xG^u&M(S<N{_D_;X>wlYO#b9}CS)CsGz?JRIFXfJ$Edhs+fV$)8?=jp^
zs_knJqJ}-eg{SClz*`PFB}@4hh_|B_1pp|BJULKu-#W)LqoR`DEEwf&WPgP#<&Kj<
zu_$vQK`@ei%>xDPlB>Oa#IGnu1Gr>+_9ME@V{J`7+6>XG+sz7NAT!MHI!zOYC^$Np
z6GV!NSy^A@mIX0r^8RJW4NnL8Ob_sZrfwp(?DvNg)oWxzx(xwh79)04tg(~<9UutQ
z-`0U@>fms|&<Tt)1&C9)3Jr|SW$1Ce`^h@rWxee+8qxXr(Sc=~5!polXxF!_Q?hwr
zHDd2hgyzVKKO&^nVPbltOb@1sCKC!P06rH;LAhqBtAl9jok|x{n!-4uEQI>orN+4>
z?*{huaCQ#pD>h{OJ(KP7gT3Q83KkYrxRMM}4%GaQoSdpK6ijKcJPuwl2?}m_PBY5N
zjTg<y)sL0=hI;8E6{0E6*X^l<%DK+Q#|=#kW&LU+v$sg48DJ=mW<-BVqLZ?|CEL`s
zSIx?Q&`VL&>jZ@v#JY*6kRS<Py3wiXob+TFkrQD-#zN1u{Mq%v2_#e28t>B{&)`Ja
z(w)jkN)S?8du5~Ctnn{)I<vt6agDE%IbOjYMK;q;F2_V0n>|Pab)K|E{Pj5XEdL8A
zw>4=4{3I(ibrXNK)QY5p>$)*IzWkuMK#<DBWSHLO6JtdG%&0$0GBjLhdowz(E-V40
zgjj)hZD0e@0(Lo-O5z|3xFi{V&k^eNQ5*PVCKr=y;MkGl3e$Cz!>IKy9Ti#okNdiw
z)nntV2z`orS94}@OxNXWdC<0+*caRPwkm{+R;Sz!0%dp{kTf<UD=SBvV@cq}2y%5M
zk}flcl8c$)DH(5feyr}8zl~=fKv?oFv(B7uIy0P^@l7!8+S4slMnWJjAAYm)S;2wC
zB6R`pO=MRwVyNNg#B}3Vi-zl_5O!0|B702kaAlBOrKaaUbq>0V2Bi_%^u~<LrJp5w
z)^a(qg<_C(1BapXTK3Jy{U=n<38q7e-*TGhAU}~4b9{MH2YsC*!L-pZu#L>sU#uqt
zyL1uc6{|M6ft-lgVABWmP&(u1R}cwPiJRA>VIu-*!ErZ+kU(ym{zB19@gD7eXXhEl
zE8uD9%jc-2DdpOgyyadSUgJNXP|ZK+inb_3!~$=g%ycl@4%Dq;Zgy%WZo*)3<=vJs
zAekHhk2jLx`G+=YTPC_EGVJX4JnO)rG9OJnhut$)RHz?%T9LR2xtxSWMXT>}F-=YL
z9<58DP(dI#L=+0nUZ|)^NsgBD+NN#+7bR@$HFdA~f%3jLrm--<zUSzinwHG6gN2zp
zj`1r=+bPM*`L=;gMMU+$JG+n@X3hzOY_=u)?%*K1ijqJ5wI`X{xIn>5Pc<BXbBpAL
z+6lV9EilCf(ck_6sSngtsY*GYR~O~nh4sJTEu7jHmm2jaXfnpJ2VRBa4mcdXVP<L_
zZcKuNgh07d-20RqERt-R)ptcyOD0Eq{icrC*bbkX=PNc`^Ojgl@!UO9ytYL8zsAtZ
zDGKjd@Aap*S{ci#@p76R?^AOi-&ix0E+G&Q3Lds=3nuTH`Wtg~fF4PY--ek`6C3bE
zSGywJV$;tGg!2Y|#vPjdv`}}FGcWik>Q+-dx4kj2<~G#YIPvrAdorRIL&YDjr|O;g
zb`pWe2uhXS@t9@a%0;v)$JK;XhSJ34#ZYd&wb8h?rh$!#(`>sLz1+D+y!!0wz|V>J
z$>AWQndixAadsG2q)+~ZQX)ooNGiItxCmgwUvw;D_t8e@9$JqYB1*FB25|rMG?ITg
z!sSgG@uRICBF5Cr$lw_t<KU^9jGTgZ;TV4&&(2L@TRwmb<relgByMh7(Aqfab0QOE
z1iM5ZA1^}m58bSOuGSnC2qA(aJ`S9jrmU2n>7SFXrz+)wC7dYCqB94PCf0oTP$M>e
zaehg1_T(#wXas$8nifv_pX7%5Pw>US3?dFPVmPtm532Y`rW8J#^P$WzPpF5NpBxy1
zXnAjz$+LwW^F1EGh!x?(Wk&}bm7$q)yYpW^yq*Vdy_g51nNgZ&`(;8<`rxSXZr2fr
zlY@JA`Q}g;435fVLth_&nJpO^%V@A4HY4CS9*GKFpP<^!>NTrTFq{beFS&JzBPFkH
zh2<gnMwm{_`5XM%8LQu}tz8wt;E;qSewv3PS~9aNg3phU*D)`M?hy4%0xLct|9<n2
z|E?4fW%IbyP@lecoEhlAx?uf3?R|Ar)Lqy12m-eVa-$*$0tV8dq#$J=4Ba59pv2JK
zVSoxEQX<`5BErxiAkq!eB{eW~=eq~*=UMOf$G5)c`Tuhbi&^-K6MJ7f&OYZVCL{fl
zQhK;pgAc_Ai_~~db(b@ZVv|aJAoKCV_FI0cTU6h(CK<t~Opyp>`hk?^(X#UOqfLHv
z7DK!fB*&GIgnb0T924UU?v$=^&#gL!c*P7Vx-Nnn5!q*qnPE6dsy}Gw!@4u!cEO<$
zJ1W3eX`~dD6FxH7d_iA8<NDXH?;5#oIY%#)xahPtCZ)U~Iz`t?vNRi)S)Bd8(O>LY
zA5+|gs91S2GAB`*UvF;kFp={)@h6Ze7>^W_x12M76?7Vk$U{CTh}yaFi2SQW;dFxO
zTb!3(E@CHyfQ7)TW#3xjMbFjUv=!yRhX|66R#eqYw%yUfWMZieR=3VEG0^bJ!@5%3
z_v^R$gKv9?i@n-2ktQRfC8?fNNiZQ<oSNdHOjpfvG3oF=mJRNz8H~IwQW^xih62NH
zkYXNWsIPZMtRrxSnk9`15i2iGg1Pt*%JbN+Cz4{4U^y+Mfb!JHw$$~YTWNcC8%G&-
zzH*nG^GdOc$KwTx%Ia<28hXaF1B&ig^MSh-!jklnE$cP+pcV^P<!~>zO%^rX{x#@T
z$=2^$*njcB@#(c!`d%SWM<X()@^38N>HiyN>E`+qL}0fY1bPu=+kdml)hdTQUQ>=8
z@H*{n{3D>|p`@<PjfnT#^fCC=-TUW4x~rdkg;nQw6+|pKs%t%SN<uRC4#Mc@xZffI
zjiRNbr{BnslUjN)Uwa3mlMf~Lt(Uu(YGLw6-JbD-r?;&TbK^Q&Sq+vl`c89ex%*44
zo!jii%^sf<RwfL|@K>M;X!vKVmG%Da8wId}EpKycHqM!+@hsK9o3be?d|q1^WnSu;
zRWo&9vq`9JRX1*A)eb8R{Kbc25l}JR*&hBk9gQRi0MvXmiCb}1Sb)!>$!~j0{~QL)
zZ2fyOvFS(BP5Qis|C;{SUIMt4*3Y!C_6zZehs_z-%W;c>Rjpz$CHyC@xErEas2EwF
zzYvHCi0eWNgzRlx&q)7l6Xn$BW|%@8=2zlHH~GZ<d`^Y%W8T>W*TS-~_W|!US24|?
zS*w_!kdrEcBBo$v7n252P{1+K{S9nD!x0;?%x@E3>ZVX>M)Og*XX^45gQh3jM|yXg
zrM$tXex;#sc8>r}G7fGGOQzy2$8~v#aC@_2pn##XQ4tba6VD85o!So-tjY^Z1e^6d
zPrZBiJ;aR+a=ewJcEdJopVE$m-@{IoT{f;Krbg<^MaP6g?a|L659>TM_{C+C`AW-z
z)6x0y2u@m;5;J2zX&IM^{kY}jhyPAF9NmD#y7QBAUEN=$&ZoFDLLrc$I@rGiAU5}2
z!;Ck($~A;8tM9_6h=4*LK`2OqnCIA#JxFTD;z-8cY{XCI^5YH7iWdR>mWh1@s2xBt
z{C<m>z&RRNaxJh}E&HXVMeUv|mjmd*?zT&Sm{}63D;FaR$z`(9O;%hP%}{(~IV}HF
zf&jdZy*T#UD@&P$2>Weorkv#0M+^L`n?ARi!b-h#-nMc*8!QqjcbjB~RW-aTizN}k
zO{r5%$C;+bv$&EN#;qj@D5CS|x13CC4r1OoK-h~21g=Iotv2HJT}0tR>`g5Tu~Ih;
z(Ww19P@kt#v@}zAgQ<~?t^aZy=r7wFg<BRwdaSCfpM?3sC$7~H#!QKYf!vLRnISZV
zis?Vh%#bb_D#b1B3=Y`N*-|lGPl|N)`?nQ47Cax2`eOOVaf2yjOz+=!#uoN}hEsuG
z#eS$!CE606V;Ocl+LI%!+G2$DRRk0Y{*dz0$6LgH;o-R{qxZ-eiY5&vp`)j(+eZS{
z$2<aJ)DA;(A+Z*_x2FyJD~Y`U0U#Wno11*^pCU*wO)=ckY;0xz9sAJEx*1W#?oLF;
z)W9gZkZJ?1;}Jaw1(%}b{d+3RQ1pvYJ?8;UL-vqhb@thZwya)1euO|#aDLExtGGjY
zh4)lj@o7ilGR>_?yDfsarCZ$oIh&>P%Q4xRhKmNKHB??=upMCnWr~bYM#4Z7O<sT?
z@TGFjkJ>>$xeJeiu5mq|{G|JSX*3K9Ej&8XHuF-Hvs|2}sYOS=oIyCT$^0PP=`4((
zfrUlDqjy})23|VXA^#3Vy|{&Rhl#tV_x5;AvgMfM*LEmnSaQ;<o9aG8QRl|?R$c8N
zaR8w*_`{z<zsKdDfd>Hm7{U&%+xbp;sbZ^DB`cpbU&gt-fZSZ0sBp}n_OIS{EAmcU
zjnmABD{X;X!b8b+<`DGz`C*R5E?=g3YGA(RZOkAnB^^dAAwe*6X-10V==5L5gwLm^
zO)>>e7k>HTTwOp;_pR$I_xpf=+kHy>@;go+NoiBIolQ+yQlPY1Q(M~*h1%1Q&C(M+
z+)!zY6KjWKN9i_HPv0w6c=!_i;^E8fBiL8%ytlBtGc{)u>Z*=;91&Guy(J4pS2g4L
z=id0NxB2zz6x*m6zeoF@wG!iv5qv;+h!wcj%)vTSKpM4U(;qv8TytwoR!-eqZcY>|
zSg~{3_cz|z{<2nmEq2G21&+6y#Vy0*wx3Nj7Ayw5>wj!p)7BG8S?V}h!g&w29H}HI
zcP>UfPrrHI7H!e<`1r92b+}diN?TL!m15_|#_y8?_}0^8X9jExZwc_R?|y>49YhO9
zPkLWGU=Eeexee(^+$NCamdF=nuU_4oI_UN|Xn1rfTc;J0YAywmZ)*;rmc8?!Y})yb
zV9*!K=OL}jo;dR(582F~f4=OG;WmI>J&1XCjDN=KtPvw)w)5umPZev|7Xq(L1sXQs
zR`rshbluDh_L)1QTA247OvdK&2_1+}wnuh%4|q+Sx4C>ddG`0@6T=6PEhKcnNWuOh
zDTd?sUd4ign(On!4Vt{hmn3goFSb5pG<9)-ZGe%2LVwRq)qWZ0B-O$mSY3?zL?~{I
zKK61D>lg^)xF8v#tM#L@__XEknGn6}4c~wLCBP=*>cX28uEFzunHk@JHKcRvc(81F
zB|Y;Xa4PODk}-l;>e_j~VsYj!f_Xy4hZ2#b5B-=A4l>MM={65K_V!ZjC&hCfJmGjt
zqokhl7xByMF^KEZXS>dy*SeZ)%x&l8^~C5Mi}+ycv(dcS#&w;UShpEFE5_Our?%(|
zv>5>zlA)2YdTf_Qvzj^m%`)oS#j<)f>^%yf0)bC5^JX2=N9g568j6gabJ*U6uA;D`
ztEo2jp`$J<q80Z24>^X)CA7B|)z<QP&aUzWyi*GQ#KF7rm6KSceb;CqGqEa(laeo8
zRgCl{E$Td7>Vt+rwy}BDC*tC2btmU##65gLS?E{QnwpQF*YoRzhWfro;oE@T3OE}%
zG-EhgX8<4H?9gOn?G7H6dic<mbV+Yj$-!YDU&T0e+eER6If}66T!7iwR{1{a@5Pl3
z<O{S>9dJ9hnGvb&AHEj6QktM(HPVrWHtsR9p!RrLXm~m8X4sLFeS*PDS|NKOl?HC^
z9tNS(u0!0}5Z-x-)&A?BqV(gj``ZgE@9Zfq3Q^Ep&gheiO21d(sxFD&#&Upu*)xCe
zQhM8=wEA(jLbb+4kr|$>-^+8d^o6FUWN?R9tUS8tY$Gpbe*8GkV~^e=r1)_OLH$?{
zjc)>-g_c>@xEbG8<|#c9yh{$+Q-sEk{#rtcJ<jlyMIxCLkipavy>0c|-k7fBnuvyj
zrHv5GST@pUzmd&;zGI%<hlq%+Wg<P(8TX#?5c63}JpHiPh>WP%-Lk#oc_oh$#~70C
z9J5G)?qhq)JTXUXGL1I6N8(zI7TPe4v0~4+bl4-Vh!I;T#IKmicGoGG!DDmbNP4qi
z^XKi9=g+giY>JA}v$?l>%d4$BeVeSlHcfE}B|iH>?c>S89PTtXoRu<@Ycr_V>bj+c
zqB9vMp1PJQkOi0zN#8rFmz*n=Q^^V;Tgc1mGi|MTP-?JN8@V1X$y@HE;2(^iV&;Y9
z!Oias9Ev9|nk_0vt(G1qAYc*15HQ!B#X5CLFF`-;Tqgbb#FQ$!j@%1Oy+W*s_YccU
zOtTnFAI6QUT|S{Ec0UiT_}-Y!PU?oS!nI;S`{-WOa=F~p-ESO9!(j|jPv`tn0}2m{
zqBdmRxuylr+7F#oDSSH1c{UjTIS<K$z%hR5;Efe@R;b2Y>9}Yg1ASkW*Jw)m%B>NR
zXVz+xr%@Hv-iSk&->tS$sYl-q+)8BEU+(X;FI8ySjDOUs`+?~@TJ8OK9ybl{X9@H|
zt;e*j%$#ifa?NbsiIz;b`V3)bg@1h9Flo1p6dEU<z8h0IJwO>66RvUi^u^g*j^kma
z<qm8X@^>@cK5<sXPxj=car0u<^PWlX?bD(Qs;S-k@C8Uzbfz9@6&=UNYL;kGDo>8s
z{{DPqyebf!ta0QNjCZhmogCKy)k_3=7ek<s?!3BZp8x%qo1-7iej8I!+8-Q_2$`Aa
z*ri0@NkEJ0E6TCF@LMey046{MW`1sE`r~pL0ryhz20Pn!m*KC6M7`U!3ij+wk|DWe
zkKiYjunVtH#rjG7%rTY)ByS-gK-<3IdrH$!F6M12m9Fv_H(u){<0e*XOyg}d$|R4D
z@50HUSC~7*<;<z+%=_Vb;aPg)C6Y@=vDfC8S1X=NMIP1Tm9VXB>c$K4efZF|7Ji)2
zCl@)l;6c=5^kaW-VcV<_JGW|ZgWN9KN9*%9g<7#g<p-j>_vcGo_NHJ%_(G-IGo5iF
zdU4(Et;tlat**1E^Rzi$$6vKAjLY*V`z?DwW$FiGIGH<)%YyBDhue9Vnr}8a3T^hU
z8gBI47_v*B!;P5(*;v49$%N*VdK&(wF{AD|M+??rg@yFLHchT%Ik03%4_8Qrqb=2Z
zSY9-l^(N={mG5T8x-LIx)!Yb6%^gb4Sm0;5Q>{<w^O_m`MX}+lov_LsUjKKUXD=|B
zTwM3TqYJMu>?t+PTi`n0=4yRhdEZOBP#*X5)UnLq{*@+nYmBDjhuK#lMU(=)%}#PX
zh2c&GyaKqU_3&MEp2$D`sgz=Dl<Ah_OF>=XR+JUV*IOE@Y5@SHm8mE7cJyFs#C7fm
zdE`D;nnMSzS9b2E*;oe2W#0tCO-zQ>z9V&C*&EAuRrg<<I;Ad2h>G!Dy}2v<uUY_d
z^1c3Am5r63QJbrNVOrljlY*7cnJqVx43InsNKCpomhj4Vdntq5!?kfN@3+CpNuJ(K
z3Dsu<doD0BASvIxQ7g@3oJa^x{x;^JOPOvF*}&fuVrbb5KWZ-u#M5{Ro|CO;NLhb5
zV19IJcDl=ma-rnR)aLN(Rc<+z0he9d(}UB-hKwgCsi#{)e<y6qZqC1?RA&o_d%B;a
zppmkVhf=myjPE5!<QY-<-lx5%-hdk%MBKd>lyvz@z|#7=C0lwsEZ8A6&^IV3Y^gaa
zrmeB_W#K}h^$6ji`?xn8E5=gh<!z93y*JB(hIC;>1z*T1I6wk8hj9NCPr++f-}sVJ
zVE$4z7*v|RbC%+Y;_91*Z6Xy0WK867Lgu7P*!^&s?n=#gFZby93u!zeXrf={toCJL
zk+He0Zo#~V%|4i(#1g%|xVsK&;cb@~{}8!(3nW&bB`d3{%AN01i_6i<;c{H=opa1#
z>rt>Z`lL=Pt@KRuv#ai4k)KMR=X^v2_2Gdk$Jkdj-(>B<y7^B<IX!8}?CZ4WN`@_O
z8XK7oRKAL=z#3bO?q}t?EFwijw)jN^9eI;nPR=tzSdVHRo~|DjdG>{2exP?DY&rAS
zK}q}37kM8`G^3ut&slZZ_NLbEYDv2OH4beDQ~TsT1*#mfH@aC*ZCWX&t$%*}ZB;NI
z>^c$IQ&OQGNT1?GdO7){H$vpDaV}lN?PiS1p8@lxOsMuZAhcgf&MI^4tW>#NYBf(B
zLFs%@<u#B)<~f+~wnI^b8}kX(s2L^aiY2w!cZl2LHBAmPv{GvLJZC#Z6Z}YjP>Zc$
zH2ZD5c*X5wyVl1?7Ok)U93t)ApH%IC{>ZB70g8$|fegLefv=zUu^TIzH#KuquHxiT
zDC_%l?v><m_S=pQl)P3MI}BKy=kcq^*Kg)SXb9PcEyb1%@AAIwldR1PVGucJf85m4
znv?taPvg@cFs~%O{?VzlH#v|-K}(k#p^aucs#O3Bm%uH#e&ovf0L`VDd6&{7={{P!
zd#OsJ-(gujDZ=cPv1l|K(zR0GN<#1orTalxYa~i_NL}btP=eDX_Dsx)B}O39Uuixg
zO~DMZ$}gl=<SKc2DZMM(o-2Lb-kYsgOsHp@&zVb^DlR6LV)We1VeP4fv<OgrDdxqF
zdi_?Q?!j{VclVcvPBDF(nbq&%6_!PRdzWP`=7s50I|ED&-OFAbH*Q$vet-Q>XZ)cv
zvf(=RL4B3Gk)n@xEA$TSD~lMgz1j5WaFc!pA*~BU`blpruOxgSey@{rwsK@IiNVy`
zU*;Lld*bcxYm8cQwYA?|6oTcW+f-GZHPs|ePPRsf(>{EGeyPJ59xb2Ndz-bzzND=T
z%dzU=Amgrolm2clX_+G%hluX3e40U0Vh;c?X6KE}vjg=WdtLFWLW$0v3#mgneyL&~
z9gQxor*tZXH5o{uz;}F*#^bQ^lf0MgOs=X|@vYSTstZBIaw==z8+bB5n8vyYJ5Q}s
zXMPwuSYCfQ&+j6<b7vjxa7Kp{&3a_RiNRaXo+XT>yoyYnpUGEjC~03kI$j*}vdE$=
z9$7MI*G6FIh|4KLqoT6jzbDBuRMmXqc|<_peN|YtUw^c)4og5lphn(k8@5Zfqxpjc
zseBk;Ej|0zK6?HJ^d*NzAk*F}oywTjMJH)(jcvvKU1{UA?dl9|a57#jrS+wwIBy0D
zp|fh~_p=P+si*a;_m88aqfxT96U4;CzeU^YyH7bY!TT>>xstuNxu8v(^nLi=J$1=+
z5m}Yp-Cg3p{`%P3o_Alj?%_kJV92+#(#|@!MTi<CB~kob&Ui#26cals=RLw172v?^
z@mMKHtlAQN=S=o*$1;&X0x>bMU82*P(SNDi9`j#kyLC$uPI{JI@q`C=StU(4SNI+r
zxbN=oH?{|&`yC>tt}Gv~udRiKJIz$+SX+NtES%#@9<dD9Cm376TRC??rg{&E_KuE5
z-F9^@<_forF`C{o=igOZDLqJd`T+RJ4VhRYB=&Z9&VV5?qO4cFL&o~!Igcu@!kgAx
z#YdvKE9gnmKzw?qzW&RkH0@_AbLSRv_{E|~7)@79eV;nhkqL8&2a_aCF5eudJ!r$t
zQ_VGCqA$$UtFESqUz*Chl8&e0Fwb6l>znhkXbJ0Ww2}da9F$^VtuN;yIklG0nLnVW
zNU1bVkXbE3zS)~+NQAo}3451qxV?Gd{7vMIeO)EBKMaxx#W%PMf5Ix3ba3pRX8@2r
zaP92)Aon-wV=GV>DKg4(t8z1^{ya7Q900RwkSLa`7yF73BjxzPGPY{v<t3dV>pOB6
zE^&r2RM;#Zv);M&rPSc$ygNkvIk0-F3G>@ySx~yjd>fnY>h&?ckR@Zfh|9igvz<e1
z9uqjrmQ|gR_TznDfX9B1qShZJbuJq$$iAUSbb0{0hoNsb{VtP-@#kCM9Zyo=n<H!G
z{b~{IDqx!YtK;wKG997MK^azCv_1P`hlwNeOyHtctW||<jFZ;pi2lJxgflsDxj4mH
zyF0wg^Gq8x0?nH(M0OjSmH2zEV=uUe^g9zOQsqk@j0}CttqJ`HC(bgj;K5<!RFSt#
z;w`cCdC#NJsG8Evz3J+Ew1O_m;2WT*E@$Xj44G&y$6~1am3^OT(q8^^v))@gF2#VC
zi;kr81G*1f9nnH30RRGlxk_9v+GN$S2{O(%Y>8g$?tV^oHbnnPe6P)uYZ~gG*4Eb;
zkoPt*ja*l&Cm()Mo|x|rP99&e_7AC+woFsmy?i;C;RdsmXF|dyrk7#F>+4~05g8wU
zN)4E2t1UK1GbuDp+aiO%Cnj;o<7ZJs-??@?{$yA1in`<}#16Aw#Gi>lc!$GW{%I;g
z&kv95pRdj=#O42d9Z!Mo|M~v^FGl>Aqj8XPZal|xXM!jtElryiU#*%~ZhB^h6%YvY
zS-s(f51^l_rY18Ieh0VyFn&_WE+(cQe36RE#OAzaBi50Pi%TK+_wq7V*@&ghbxpuj
z7PtTif^afMbg8MiIb;a?iGhHKNIRNPy}7kj4anIK4qe=^&0u}t!_Zs0{b>kog%-l|
zM*TPzr}&$kJc0JEqe;UDT6M{L6u0F<1W+I08K#lJs#$s;Yp5jx31Vs7qf8@XuYY(-
z<-UTi$0L3|@3P{RyvtW9y!5^dm=`Xsc6>8c$-03^f8_gkgXA<qbSc6(wi9U48IT3W
z?dtGs4*Ap7s3-xeK3%7gn@C_0dmZb(hV;L1DxK7qj2`ISE&}?R9`0?bH=I&D+Y-qq
zgQGkG-^)0zcf95m?)LCoB>!kFqWmJkzbJiIlrR39m{$f+>mM=Yl6XP+&)5IsVqeq9
z)ZS}|cz!8)`M$;wrkh9(t#68e>h<psnjClUYF8cYWihaDb1MVCSkWDF@4+Yixw@>(
z%padpp~fmJdTkIrct9lH8+1dG0Vx;BtMu^UYq)pk{QuIB&xwDK5%HfiblS=Rf%wvt
zsg|`dXqFc?Gz0ujsz7XJ#tbi?!BHWVJLmIT4yPwco=~u3y3+!4EB8?Kb#<DwWa6aO
zLNU~Ww!d8pD^`Ed@L3KQqvTJ=7u}Lmp%Jj`5g2n%8Yy#PgO1(4eH$}am_<lNm;LLP
zX4Vb76>=E49rkYCY1@*v0YdNP+j1&#9$OKjJL9hm&`eBBY}c;cJ9Fktq?lJFh!6|G
z&P%%rJ?1WGDM}tMz6f6Iy3r{a78<I$xb^$@vzx$9;?GstG2%3qQ}c>QJM?dFt_j^U
zh&ZQq5%Ag;atnkAy|tc=mb*mq7=Ji8JlyPfKsLYD%%!2I_>uVRS%;mKryM%P9hkQW
zloW_<<-uAr_cpzVt7-_NT<}vWOjn8wDI@-qlt8f~O(YUYeC}M|RG?UHk;PEc*qDBr
zLL834KJ4Sic&9NBjc{GZ2x@+F69*lH!xg&~ycHLI8U-K3(ZbH^6z=^D^H?C+0dhb%
zMw+Bn5#9%6f2m?etAX!F39cfwN^DaY=5Z4;Vu|J_@un_tUNt~=g|evh<Qqy7oId^Y
z7D;cGhJxXb_ey1sSOz4F_BU=u1pPfVNWqP>ug|oG*c)-mJsL(Z2bxCW#*ksv;%fDq
z&#4*FP#3A}e~CldqB}{F3<;FN>H*Vu<7Fqndhml+jzhmvC)j?oR|Aw-+yV_`=|K0(
z%*ra=8p&56bk3XFfXcVjVSy1jh_=$tumCc|n@H>sIxk-b;Z8Zmae93H6Q^!wUge&2
zV`F2cepTgQp;_7se4<Q`{Vi^iT5zvXijN-o0?05RfgrwTX{3zLYE<}eXDX2R%o(LL
z`50dNxejj01;h%6zWb^V!~9H?)eIm0xFMr_bb4Lg!-}K51y@W)+C@rACh(@9b#qyH
zm0d6~>5w~jO+_KRG0#4;lap0TdK@2ayH1B`G!2g>JRmyU%=3CKDpLJsajZ(8OnAjR
zF)7JmAqR~b{AQX`RECq_r{7;kk`l$ML(@BR(4(x#y}6{oi<d4L*<45PVShJ<M2ff-
zRqij98KB?1emz$-WZ4Y>F+DqLtR#e}d`(`Ca90O?UMgD_1s*`T&4vE(r&LYwupa?5
zd_cx{xY4cP3S$dLZXeWR(m~cdJu4YlL_~DEQ{}qflHagz=0G#^@~Q$L?ZH!cJ@)Kd
z<HYo1gdCK^Id$3PNvc^vJwq5A=DU0*ImeTNLqawe@~S=+i-4AaTmo^Va)s+AyA)w{
zzv4K-c!A$ctB|&o)Z4x7<vbW3%&6OPFU&~Tr%ylb%XkMeQ$L?_Eb9eP7;%}60WO$(
z_ey6X%|c^ha<)g^6aYZ`w5=+htEnO`$w*0kv^RPvh8QaVen7c6QC-k#*ZGw2o<xbW
z;IXnXF-fv%<o1Xi?<(ZzmZ|d?wI~9n=6pCC|G_Yb55X(}sw7M>9&^`83q(uq3>n$B
zIT;Oqb`L*3A<0Ju+B??b@@WN+Y)_@@Dl038Mn<xhjkMuOdy|1>>4p^7O{7Yiyjm2$
z#l^q>zFK?hB2WrJ*Y%u+ZQcN#Gq2m0xqmm_3IPC1Jc)!qGtSIKpB9%B+&2pi3JOXo
z0ytSUr{wraSdC`aEmaK$Ia8eJKkeQD(#e1X$W>QWRn4&2+rqNVx0KSedz_9h1yJ|q
zUK~Oz1n)>wj6cA`$L}q+F+_G{sH90-goK2sWQ!oiCT&Z;Sx?+YcE#U&#V(IuO)N1s
zew!qzg<oQS*wQKB8Q%C1sFyekaCd)uT1$vjhCaY`@r)F$X^&N_cPrGg)RDnjMdmDC
zUY~dn?yWdk{{H<w0IFtwem+>GCod8=1!1n655%1}F~H(x)EXJ4{THEkruYj=Z-pi}
z%71a7pwR2s6J*iNvK<BtCV6xG6`6D}XW1e*AFAhP=4S+ZN1nczkFPHZo`J(5TRXd(
zF_&`+4C;wnqXkuh{Wz)#tDW4p28@G~!NzDspY~P*CFw19u|swx1&U@?xNYNHIiYpW
z$tbJJ&g{np468b);*F>nLuwE!_dt2O0OFc0K$>YpwukN9320XCy&<FT1MKpXYp^cL
zH*8q*C0)69`SRrxa!(e3`Eze-Cm$lRZ81G8BVNac&>7|exhl^?-sephAFbe}JK{`5
z_P1}mIH{uHH}9zQ*bfa4PxB?E!GQ|(ttZ}`6&uQW?@x)c3kYZcxOD*tr#*Ocj^+CG
z`?_V0y(<&7y*MXf>eQENa140uW@I<9LmGj?Yj44(YPQVVqk1#dwDJvnf`fxYLqd{1
zf2MMn>$umi=a!>Yh<fhu5Fz^Ve>tQx|4&o@U-<?9*~Q+(Le4Aki;iXR0kdP|a3v1?
zy!lf5@{OA}>oIr0;Sdjq$2rhVg@z61aHH=R7^v~=Suz+I14B3_&Prcor5ayaN(v|N
z&%hN?GK)5iv=xeh55>|A9-eNM&}d;7TZ+_mPcSQMI#DooF$ISMRb=|^!r*sT3+!ss
z!h3q;BOt{kXq)_r{jsuvzLr)bPO@D$Q5Bng+QG>scC$FDMn2jCd1ltX%XB#XLwYzR
z2!;|Za8A+96F=$Z#fuly-L{toOYG8mm2iJYOrJDf+8GE)Me0Yqy%7~)gK^Fp(B?9D
zd3i;g7%y5u+lNC%7TsWNfrR&&4GMSbiU@89#n0qbY(*O1y?t8=R)Uy|#%_^-tg`}|
zf}7PrDln|rO@87V82}m~R_^-v_y}2#doyHcqb+axm4yYt#0QG*NePVDhrobMTi@vc
zF&=SSt{4QJFKMtG_*!2ut}ZR@$6z|!78A+++>f*5_j#Y!uhWg&V_nx<1;S!tJlfbn
zwsCG#s%%8!XoVY_kdPMm$RI~ub?+_|+D=R7MQs;m=2^jWVx1??6ry^i;r{{QyVl#q
z@&MvuH0Zi{aTf+q^GSDhb^^<AEXaE5>({5--D~@G>;ganfTeU=aOr2zjjuykRDW*F
zY>^{&&}EI{P<XZO!fcD;zzK&8>Mke{!rQ>S=-Bg!hKfq_`5~uHv0~Dr0A+c3`TQX#
z>|hE^@_dZbXlF;l1Mpw2zNuB3+$o?Owk;cp6tqhP+TmOSi@rSlJTPVp8ELUNUSdCr
zzuMIm72_d05?*T69xF0fU>s=F9-Bia;yMSItnTdWEFoX=NNswPPfi7=kRPu-@GBfB
zvWNo7O~r8&0|Y9=JT|w4#Kq-BS(YN<DptpR(9Um9YVg3kL1l7wQ1^Uj_q^)Ur%&@8
zB_8_~k64yKeh3|7rc~M#i(73RC#9+ZiTBGNFvkp+6fl?quz{95N!gDEL+yKxRx5F5
z$L;JT_6{`-UA4rI+iq4>hmP+imj>kHL~}TGOPTrj3h%r4XX{t#9qw=EZQgLn+S{u<
zKH9%YoMhbG*r;@TbSQK2xj%SL6SZ%`nU`WMnm=(+3)!cG$!Tg;Rtgirk)Bm+gRuqK
z7T#Yhh{fXUY<>e>HxQ<_KNe?~gP8{3g&B$ShDN=PJ!nPUHRaP(EYaE}B_-0bvOV6X
z@EMS1{kZ_N>rL2oaKc05;^F|~Q$QSWkr%{SFkox4EED!Z4h^(Up#u#tU7K;o2a<2j
z(JEV7<{G~?F)~WU%@bR0@%^rd`@g?FZ|2-7H<NbI8DolbSDu)d0M%d7(NF8n$k9f#
za&pQC+i%P%dF*4+DRfmJZX93~yC7NAb|eC-HJ29WsT?Qj@y&dI8-TZ{Zk1BkK1jB^
zr(FyfAhEM?UJ$0$d+NEb)}!v8jDZ<0Gn?W(m+(dwdD9a#sC9L!A&TB}Cm|v@*tX4+
zQTFo_Y3>`>ueXFSDSR4eo3@R3MM%nhGXk2!nJ;$xcE08Zpt24dRJeEV1rSb4DwMq<
zB8xpn9&;+>=~c)gX95I9HBnJ8QEp4K4fI~w9f#4P9)sNthudRbH4PdKn}tJIBP}hh
zjKeE}wsqZ0;OuY|%5^%hEteMSaPo+Xj&|pMGb;ypY(IjyGt|Yh*9NzE<c)rE8qy8S
zy&$`};9`<e^QQm8W@KdSfd$2paN{iT>guYrv@~lW84Mtq=#F^s280jcg3CGfMyR(i
zSXx~XE{4t8%-9cimzeEV(TU_q5|!Yqa__SQwi<Oq8@Z5LU!Ng;_R=kT9X!CdmC&{C
zlt@|ffA>xa`p19-9OB3qyL)$tht&xH6!R<JFv7*CE&8#MZ&PNAAT$&)`Wk13n7f~c
z>!zm2gtPMSsNj6AX0EF2r?tK;3=DPMMqtYP{QcPq|B8%^#6&D_#=0#jfd>d?V3rA@
z{TDA@0A|1k%)|B-w}YKkZo2?_v7_h`fAi7uk`_mZn7hlCE3-i0vs~6^fO966-LXv9
zu;D|Py-{V~%a<>0m%E9f3tsv?#}%MN&P&JJ=jcTj2j~PWbxIEF>g!qV+$kRBr%;PH
zL80<@RlNYb;#f**>P*kWy$t;-&%O1yV+j8sfJ^Sz_tHHHxJ`e4ye8zh`1$(v>%t8&
zy25rRrfJqAZkj@st8!=3Hn4EO95wJY1+4Tpr{z)QAz`p`vE(_kwe$8si5=&(AE~I@
z51hB<!-lV3zGOghhG*4P7MyK>*o6&EUwf;k(!<RH2-}ApDU|+c+nF0+I;sg^3E9TR
z#ub!dbGzzT7Zhj!r#0ES4ixPqwdtx+AY#Wa{mKf%!urwdsiu9uJRW!AQdEjVC#oOD
z{q+0k_AN5814VOvpanPHhan6xjhsEaKuM`F<3T6v{AE}cRNifCF^B;op7OfN#1aQ+
z<pxNpDZorqOG{|;bnp;%>)B^9uOk;-{4CY}&7Ajhb0+OuC7uew-@w^HHR3GT4lyp^
zk5`7kRb)s|P_TZB&$O$q;Bd8pUSq(+69VJur7-|o<1&-n-s=s4^csK-45n?-f~ALB
zL%rFLdB89>85K~_>^`PdAyQ4jlvajyrYqfO-x@BlYqltZ!L_AyyQ|%)Q;U;2za*Sr
z!8*M=8@Fu^Hly%YWqkbAfGdK^!5a{z;G8El+c+NMww|__yD}phnwsgz5Js1y!~OcA
zw%D{H6S}^xD?;{C5EY_shN$P^UPlhHU^!BX3>H0Dd2k4-lu~3@Hc|?)RWo*Bk>J+{
zIuR|0YtUP7p+7R80GSV5F!v4B%bJp!%F4yn&V$Z0jf8RQ!G%}Pe*O(kY`5Jogi+EP
zn_V(oE=5yQQ+zIK&!ELJ5K+Yr>2ik4#|Wi>uaCm?WX!5*VtX<|M2~h1+y7DwZ%&Ex
z+(Sb2q-<c2$S|LomCKsnp#cyjJcJ(eEK|?XN*Ny3X)lSnV{;!km2qA?r%C76uit<w
z9rVMotYX-wl{K?8*Xnus%J;LeSWpd<rQwp;G02Pc;o@AAj;kEnMcKoiiHV8AL%3+h
z)zx+LG;WQtoC=p!Qv42veq*JUj7?Nj56pjX?Zk?#{*!`%i*$6k@C38YM2T3#rSf%o
zU5LDH?xFg!BS3sa+N;K#r$`G;yB|RS$4vqmxu1}gASWlsj8>G<xh>uXlT$qG4!Elh
z*lJmZ<p%+9-JAQBpTuu5$bej1U+=Cu+RQuOFo<<*|JiwgNuC*6P*YPAYv=#rLtip7
z1X6?Pkh9YGy+)w|S;Dk#7rpvP^r&mI3W;dHLJ+37GK|l9T%45ZmUyr#_y)HPe>i(8
zKxz>Moa&#r^mBK1oHyq>9|S)u2y=f-4Z+(vS|O=ma3K@Z)0r>(PLvQT2(LaYjK$fK
zaZ+49!du#HF~5Nq(j45|W1jn1h>#)x8!~?%F)GAn3yeA^wH$(P02Wjg0uGsxV@R2}
z5Wh(|19Vk=Iob{&Eb2lFKCG;SaSOBNY91bJ&9_K6^va9tHmHoj+{VDmCof;7qodP9
zl|dC6B}*wGAO;E#=UV8sjA7tc{8nRAQ`Z)DsL*EW5iS6^>$r#uK<xK6Ob4)zg(jW%
z!IJbq762!Jy?Oe_(H?pZy=u}jGTm#_&6~?!$J*e@biyr=NlhmtrA`Mm#`NdK;UaQS
zm+&D-5xB0@5W&4-t)Fi{i1Rus0pZVt|1$$C0RC4QWFXXMR>4cuE;8=~9|4AhBnMm;
z@L(|m1!<4oQ+dz0_#hB?e<L5>Q%!hxfr%A(zrnq(7&kKx%Pg-{yT;5cCG&~B6W64z
zAuF@pwQ0yJD&nFnqZX<F&^4H;Sc28b$w_H!8fe;J8B|5eMd6GL3LHE&ZU>x{Dr!&}
zbwVvB!duH6vP`iO!!U`-jB=5%pcPhVH!EKVx)6*DhIPFTH!m^alJ(_eD!jOX{6TT)
z66*WwGqGOViv!)Ix8QHylqbOA&58ym<8X@pxPz<eAQ4hj5veR-8^Xd;!q_x2A&}Fm
z@~mJm-C3FN^Yi<9=Qi9r9`pGXF3GCK$;KuF@p3Vp*TJgo5H69sw=vHbid))9cRTS8
z5i)gkbtd3!qK1p}{{$O(^?%`VxT+;!G)-%LN%s!c+Y2EI#{SBG0J%JM0B}eHWI^gf
z?SU_e(i2z-QNjUlp3&!2Q!Uzm)*ukp6aQ7V_5YF<{?Ecccsm3w|6RiQm<*{Ui9q<@
QKQY5d3HgV)_n*A{AHQ;Y3jhEB

diff --git a/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-42-1.png b/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-42-1.png
index a63a8719caf22c6feab34c1748b5482c1133e80b..ce5d91917ed2dbf57d9d2007679ac0e14c39859d 100644
GIT binary patch
delta 107523
zcmZU4Wmr|u*EWcVf`oJ<qJX4Gw~|VifJg{PH;1l`A|N1LBHi8H-7VeS-3{-6{{A1H
z_Y+=TaL(Q{Yu37JW}A?Lnvn9kxglEulA$Bk&zBJq5|g~-&))^WJ=)G`j(?N>)H?ni
z4z;_SiWs7pxX)mKK>DmHy;NSn(|Uot01VrcFWVR7bvGuD0whALZid%Y+#QekE~X>$
zJh~%b+g_aI%OpR$xAsV1h3fiCGMq~apUv=cDsJ)OP#uojv*nzAlNYrO3J!%VdabEO
zDy+yj!ZfvKDkl!FbIT9Z-+7_Aw6%Fw9=K$zpOj(<JS4w<Pgn(he8-sQ@2xap)Ou!$
z;o?Gt%Eg~+8?$@B4ZmwJQCTH6&>3y=q@i~!Zff;RmGhF>C6g+FwhO7QA)ZePwc@kk
zl7OZrDZ!(7A&qHiDUo~d8>x`_Ty8G@X}xSn!@WOPrgEv8FuLrm%>A%ya(+h7<O9kA
zv(DGp!fz8<QxprYmAP2o7#Y8nQXzg6pZWt8{mWFr*rv(}Z?4NOg4#Ln$Dw2;F8>F7
zzdns)+HO(Xe|(Amd%YC1GVf`4nz&;>A>lyVr{ftM@E#dh2D`}MZ3LhnH&uSsA_cYw
zatvEnXuVRs338#A7@6F!gH0_C2Du*lO6U)zLx>6=Ev0g-(?`m1$m{h+E6p*U;9cLt
zHe_}Fsp#L^GM$KeX|!=qO_MJ^br`1GDVo&knWv*GuUh3zI7H&;m-elaTHV3$S*#>s
z(~u7p)9cd2vp52l_2bvj0}<U5Ki}H7v&Unq{Q2HyDg44ht2!|bOTitC703oG4~lk}
zqOK-f$XDy0sXV@Zm&>E%93B)K`Syaek%_JR<Rc1rjfC}(>)_1Wefo~UD&d26rQcUX
z(*9)S5#Ooc8WuG&)ku|F&QpK#-YPZv5(NA!IavzE`36%3Lsp0+=A%$;sjap~>K1Re
z8eFl6c=d7btIc4TT@m5eoIKYyA4pqEx4n*PCS!Bo=Yw)UZ&kXq`O*A);`I{kQ>qn%
z@slmB$V!<ax$+GZ+O>mWwxSl?EVb7@wd_}_?H`n?1ACoaMYQXhYnKCYZL0_OKDGzm
zH@(3=ow*PLj|*W8d9v8K-#&U16eE^A-+WRdn8;KIZE*`BZqeK4C7A1ONHp{qH%yO3
zx3<=5Sh$A?RJUYrlfD`@n3T3Vawr^a*{(J4?ONm1<SmXwhkujU|Jn2-FWS&}*}OPI
z6xR>mNOph8Y5(+MBd>YGi^nOwpC)9C9xHR5NInznglfpP5}TT<;RRc(#9gGgF7Qxa
zQ1R9@<=c;?Z{A#6SpHd9yEyF}G@8t~M|)d2-}#d2X+hyUB){L+mP)}Wt@abFQZ>p0
z%p!=~J7@PR6ZlKwk9#~THclRa`7Z}n%$mCdYI&tPCrX)H;;502#(5<ha~MNu6qDr!
zuAt-@rhF?(3i=2^ZYVR3Mx255rC)b#w_tYvAEO45B9Y}}Z(GwrH(Wu!l)OCG8R#_c
zTSit*OVu7HJ=Y<{i}TkL{6^xJk~f|$&qY-~x2|lxe~20vZ;PZJQ;>&AbNj?X2JhXJ
zZYx9J#{?<S9t~E?%}->GatF*$d<H$xhneIr$wF6V0SN)b_JMUIg=4R>`9}P2Xx0$-
z;NYIZy%&0`;O$$D5_@gsT7mjW;p5hghPpah{fb|roRL23JgnMMUy{y%@5=)?qwpS@
z-u**M^VZzOFFJjB6P@6^_R)$Cxa??Yt$m;QD*3!cI1J`K6bb5K&^~f?)aW@x;H9~F
zTE&Y55lWt(MPo1P%#bK42Z%2w^PV!0^@e@ye(zf@(CO#D*Z)mTAJsQ=7wvZOE%+_l
ze*dla8|m67xiX+{xJ2BbDfPt8j%Fyqy?XV!v2XBq)S;afqWzoxVKm+ce%h0tV0qTm
zV>a-inloWb)ZS>}BGph{UT&ss%=uRA+~WHCCXwl+xToj{2#uCm1y+~NrSP+3Yagp&
z`1rr&QoFHK`SN~&wz)g{L#Z>fO7vU47lHh8*20b?f1z$W)%O?`>DBZj;gPPbx5>|_
zDB!UX;h4pVV(jJwJ$Me-42x4}#PM$H9O2!6`vDn1qglKf4E!zbt_Y_jlTBx?2+vzY
z0z!$s+(SvEir4!y4&W*dorA`Yek)j)pK$EdqQl|ks_S%U>l?aNiyF<Fox`>g_|@qv
zLPYJ8Q>c;I8Dd3@jK4ZZv*~m!tq(<ZBH_Xo_SY;kw)+$0GS~PG?}3$A7_G6se`>no
z(DxFO!_=0+M&w3PT4ynSJ7Frg`sbwGuyE<e?c#1r|7-dJ*n-J&a|1c-d|M6(GBPr6
zY;16smvL>;88dT&<`dyx6{f2#tERA-nh(VOhz!rns*jA05K%W}Q|$4TP46h_j7s+<
zf@^=&za%?O(K$L8H7%$SchZLkCj(h=#iha1-4>^)&%Z|SotoTtS@!Hy^$_P`UtegZ
zJzm~Qrze?F_~p$-d9XZoU`q~oy1i8g1-|&ZCO*h9;&zWCAEH=6IL0D_`E(s(gCXKm
zb3K#JAAQdAHPf*%gQ2hqQ>*dppw<83;YDOoh!8RgvP}9LNIR~1I-QNg(RVgoiLbtI
zv`FG$acL>frkzfTc=VmXAw4qkK3E{%ued0*pHzx=TIF+l8q&kA9eerKjW^Zc+9dp_
zVL9BKlj{zloZ&SR-PwJFguxM*AB*^n2?(ogw_A^onK8j)ezzNJ=O4IxUWJ0)fTz0B
z#BCva(AMbThH4H{^fUSwPl|_sCVkguKq6i8C)3?sre?(LcBmPLLU|aed>E$*I%wTC
z%)#M$PSqr%9N<6Pbg%d;6XKbDlhEYngADthYnt7;C@HJFBW6CSOevK(k%~R!=fkY~
z2QXP8{CL*<q`arGAx+9?Tj#!y8hImlEuM0Igb0@GQvLmOO&s7bNSyK{?Dq@g-#_H_
z6-!QeJ7BXU6XBvjgJVrp{FI;Dx9O8!N61DGbNWP@j(rH_?KzZP`hu8yTPi{?xGX+X
zpy>Khu~VVIxY4T|YK2W=>y^1266B2faF&CKExvxzD>Xlx)j~7YXpdNcn6@J6S&X6K
znvBj`QHe#=l%Gc8A}X(GLVur=w8-UNF>|nPdJne|W4lNFyr1yx*z%~&GQkZ!Z>cnw
z(@_guBb!HdeyGaco>KyfXJ-HA^V_k<w_+e~LwwM199<iRB$H0v*8TkWTvk9TXSu~|
zh6Cor&WM`TJsC=s_94|mn9sv-B|i|#&DXkiY*OVE(v{=TEjiz6c4Yoc@0|mes$$K=
z!osjZBc7^O8n0SC+j_K+nRlR{t>o8mNh~0pjT9$2Qor;u=I-r719ASsqtL~suf9%4
z*VLA<0)3`0loWim0);L+noE-pLsbx5RT7(f5^?A~p^h|0!QIV6I_RYrlhdK0w7|E{
z63hImWXe{iwfAJN{{-dRm~`Yx_w;Wi&zK*O$_!2-%A~(4F^^Nf&|BrV=Xa*sYHI5W
z%iCCqI1EAOnrb|;9$K@iy0Ahqy&qfO*W^8IW~x)S=r*|X3RfGYwNhh*{nhJ4#evq<
z^9C^w$l(bQJK1YyIXg&CqSLSzxuS{yprbegZY%e*s3|V_Tw=uXi2J4&*sg)SlC+?y
z|1oNV;S)w+IGs`fn6@5Dp?~wNKrTUl!1@(h&-RjP{M|AP1w8my4P`i%UWm-C2z^(f
z{05xHxP%CaNG<rkRR)gEaEkEHne5BQoc6X~{ek4MXcH$f?I9<}eaiy#!_BT)gVy}?
zq0M+E(`f{P*A#GAw0Hn*zA{@+^K7B}#TnL?M~nZ7<MkI0{%v{>J}Y!ZBRs|jlH4A!
zMAF5T<yShY6PPnfl6yURkgHg8u{Le_+==7bHyn40S{_kFvFEzWiiggZOUUW@Ea3R!
zLtM*hU_=mK?ma@@FA&2avt78aSJL$+d-&C{FC$Hw*xlUhph8u@`-fQGer3Bbt*xhv
zM751{IFC$n!cu^L`h*ROm&Fs~(_{yh6q_71Ayq5tl;Psls}{;4%CHxH#pY?EJM5)#
zazy6)eYnpA-FCRyDEA+TUaPx7;GmrlMdn%2oW;XbC}0kfqplIz++KTu#_C5OytIwJ
zwdKtwAhf$C(yGsn_iKn@EBy-h-Pzy!9y1M>_9mH<mZJC`Eq!@EqqC59tkiRlXx$xK
zgtqtojIyc<eSF9C_}>BW!<&xw_e`}#RHC^%wD$iXjBI@lq|Vbmwg73GY9Pgab#NL|
zOHP;yp3(ZO-;iVN^In|cqkQJG?6Wg%I2HfXD*edsa+YZmej-!qBC^<^IbGK*;6mrm
zyc7E3IuAG;Y#5GsM3!xe2w0I^k2?7otuGUew}L*xv1~KJ3-?=CE%^|iYTkb-`?lFn
z37e$!UTJ{|WMgD{dgYfcNpHwbCZX%_x^mcABO-Ywe7NsNx`tVTX(Miv>XFU{Rd@M(
z<$blc>iG+WK@^;O7x#~yMq($kzIgb@Hr#^K#f&=;NfSm~@V7J7OKr};y$PNUCas`v
zy!xe533>{C4(Z31^hEA8*2QwmtURk{jZjWdVAxNuWUM3ZpF`Zzh1c>A`3@=A8D=1}
zxlFuE?-o8b;T`++66YX!qLiQrY^l|uH(*l6+vbQ0J9W|LsO`5l)G&~=I!Q;fCGM(m
zh2=3vKH@GUAYwO*;^EfT+VW=L5C7RD^Neu|;7T-VuBMy|+RS!If2zw#8!%y@wlusH
zr9CyA{1WSVpC!?=Js+jLsVk&brE0OdiuN+LHJ{`}Ditj8f&s1fk9mbve~?ilQ>Nfn
zvFK=qUQgb^ItA>1WJV$H;Q9hSZOX|!Vt>$-?|Y_bm_-ky5z3ZvY^;6{dn?s(#hk-g
zd32<Cy!hoy+3+&UddOd7n&jGock|qS|G+Ze>aeM!JaGCTO~DXA<!7gT(wk#N-<9}I
z-|ndTdvot_mM*)!O>RiGU=(q&kM6v{vp8C{45DBNAhwtzW1<NqOnsoXx6SW9&L?b<
z64vu;!X5+D$>VTPh%Re0W8P+4(^#v8yAaarTV)^1n@72K1s!A9=nn8MfU~$dd_y3@
z!;K?O<VlmJaQkKn^Z$@*!nfwuNrCKstu>ii3HvOyD!6w)Q5pRU4sl@{gmA+-k{wZ<
z<IO%HY2x42QL9qxi7=f?CYSRot~Rxv33fzn$Od4*Pyjx?;pIjUl&?aPXm+eqkkify
zz=UvwYz$cp5hx_G)hyB`&?pWdR`U-uS*lfS?;cYyu_cVN*zM@APJLZFC=Cn_tq1Jw
z{ZIer+{~y(pQxgY9OX*i!W0b0LZxn>mo@Sqf}sgWp#G9XrTs-0iRMFuCBg48xPX}7
z3Z%(v^t{>nE`stKuU9Hy(q0yU(21(Jw?9p({Z9#><@R^t46q#0w;~Sx&V*AdV;?!$
zY4$Q!D|;EL&-u8u<zY_aM>*@SZ_hA!#`phymsoPcpYa#~X|UBaBT87>-ZOHkc4v6k
z8oKCRNB#VM_f>ujq$jGdvMWc<wUeGiYA(#i9t^U1QAJ1$yW$&<M3!2fg^4$4q54HF
zU_$kd9BNYtmYJ_VwdmJON~P3ZS^WmfJ}+AB!-I)W$gN~P4jVpx!&xmtuy@n;QO;Yc
zwOIUG6$Sa}OFm>ZO?b9r20NDBc!N$;e0|hT)yUiD{r=x*aEDoDiyO)jzw8eOeIAn7
z;Q>Sy-rOo%UMh60HCr8vqby=^_drKHiQ0B2{cWMmiK=N|5xx0E4Gy9L4e|l3%xTw=
zVJkvfQA{D<0QG0r?j}dfvLIi>U<aaW(!{~wzP@t9A8aYo$*u~W!g71oyiq2@Vh_b=
zq}{IXV}secNTr<R!-&_mGv{*HBvAVkJel;NQSQocYDy~Z;AkV<0_~%FazR=?B-&br
zZ)4{eU$3976|mnzPI4iF{}p8M>M6bwW6a|47`=|+u1ECLy*HhHSDl$G)rVx9Ur{^x
zspjWR!P9Pkr2iz>9*FneAHBY>?LNA6tYc2)ps8!17ePe(ljqPhmz!AjT-^!9JL$4V
zG&p`o-1Fl_WXm?BX{;WAm_4z*q7HoP_y{y5K+42Y_xR;YyY-*-?A-r6yIM-BqNE~r
zZ$<)I-+rqTCWV?l<@8EX3YCc^{}~l;48Us-`Agm8MZK?y*p7Kys#%eIQyiSCOs8uY
zX+phsVH5pp5VB0>Q6?>4lmBLsa+Qiz{b1LA{6enUzTS37AfRiLyhYu`@5#E1r(}&D
zw~6^ZU(R6c4?iiq>fISF)#1eA|M=cho#HS<`mPiGBQ^;Q)@v8fH^N-R`mWts)|ek<
z=L@I9lA6_D91c22bkP}%r5h4e+4vRb<wLgQoaDb}1^N4nb|xH_I7>HB;C>fvI?OP2
zPaL<**b)KhLQ-p#01W4}p)7ur*k87>x@XvyuQ54l4`APp>6>iFI84uTUjV)(PJ|R?
zIOP0Ua*wI>bTyQ;No1-CbcK?{+Lr7wSFICi2D-MTj=5OQ$|LQP>ub~}@%0kWg#$Ao
zf=POepm1J+byKQ|S<l(d!g+ou%jS8Ix_!?Mor7=1q2qqM_LXz#-qB$H;>ybWv=+XJ
zNH@Jrti}$!e7lq{x%oxveX{ew@*V%hGusvB!wB2M4B6WinXi@d3%QdG(moe&u*QG7
z(F3!Y>hsj1T<Zq%0cQ!c{sCK!oFc}#R*3c`?YOu+?mdjMEv!HDIpJJ<mzPrD)!rs6
zG8+tMT=+R^FIx^8qUMkI$p(ZAcWM@{vKx#bC<zcnP7G|fo2@ZhAgO4hH1zJk&AQtz
z^XH{RZA#QQ+2ZHRvRjyOM%6Q4S{#tt%{1<@JP=`7;|(MyJJ*KR1_K5}Z40cXn=mJx
zm+a1qrQtcM4Y(rS&n5ez^A_KVd-=9DsXMz!>^}r^W3;aiQ`>ucleboL*REQYkx1ut
zwU#7W^ZENWjwAt1&eOWCJSCe}Stq<3$Ny_m&1I#_pA8o3aAsaa&G4q`OOIg!RseH?
zhB5*8{_dH|?iM?2R%~p)G3bbWRIL&m9^FynE_lnXER3Cgka>plAlAk8n`xS?Sk2_)
zD%zCh(%PNG3iyyS?@X9qv9s^MAKF=?rYWQi8veFjy{OwW3U_Mg&iu;3zA4hHG?Kvj
znz5e;91*+C=vpSW(v=-wev9K%4d{h;roOKe*B`Sc+Qp>SLbHCEf&uRt^imEcB{Yl0
za6Q>{g(|Ki$0CoD84tl!+QrI<g|_bYgr!|+gW;WO%vurBY|_*Xl-hfF=EDbHl3s+D
ziQFQ^hq6NdxN~S7;Ft2X<3ALQn$~;=cGdV|q7KWDTEKD8?k$D<Ew)CH=}d@!u|c#g
zl)9k`dyUTdm(-xYCbirzbhaS^`|0#~K}G3T@6wgeC~GPA87F}q>rR(XaINDk@wI1Z
z5H^`R(73Wioj$G;PE{fONgDY?%|)^i)?Kqx)>QTo0MhRwMiQmrH*ZWnD`(huZEf^f
zO!~uJme>)BL63+|akiFT`sj~m5WS!KYJ8JLFAKCqjv64EqCnjXJ7>->+d6eOi8T6B
zHfE7pPOp!sG)EX`Wpa7xA9Pb&`&y6lymjm84UP!wTK=FB%BB9V?*DuJr1zaUP9=x0
zhHZK#cQ2w*$p%-Nc5<;SG}+f4vA08=V&wG4gz>5D=5}lNY`w+;R=aN}H`o_4HKgYk
z&u6c+X(Qv~nfOhszxp^L01z}*YW|9Q#cFQ<Yx8r_C@Lz7uUUio^MX>CW!h|!{D$9V
z*1cAd5A<iWIb!LLoL7k*OX(2emij)X4H8zb8}r#cAVi=*+fo*O>!O~ojkW__yIoeP
z5}y2Wjqr_BKNvJ%#he(YFSBP2APSe>2u-Xa&__MfpECtLz`HBS{5obPeCogf!)cZ3
z7WG61DqC?(`7T6F9Y2cHWmz8dIFW@1`dZhI6+XLF;)@bGq`0Bou@~y)7iRRH&fmn6
zm+MM!sJWumIq!z}btbeJz<OIRKg0&}K04NXa`r3k<kRVGiM{Rk^`0(tHd$Wjh>fk%
znybHTGlx2+Zh8m1f|`lGojLCz8EaI#mJzjjRa#uuEOgTu+wE1Gt#O`?c`o;Kb}v`m
zxB&UE?CkMirHD2Pl4_O16R_yvJ?DR_)M6c5qPXOWq|Kly8|gJ(binZ^{3u`Y1h`U6
z>>2+NxlvteYa0!*<-ZBGneyCkMQ8p|Zu#pMt>+GRjZW(IPsaoCbvEO@zDWXuc2zVU
zmBXJzCx0qMm_OMP^Kw6N+@c=&c29pat9rWp?|T~k$2i)vj7jX5u9|uF8a$)83fv&r
z^~Jj%s8kg61hk-it8V-OikFHtp=)~ihY#TwGKR&!_poZ`(?|H0ZcnV;idmDeXJBx&
zmr8QO=pgaQdG{*&{s*ojmwUw(Uo@Wics&Nb0aAENbss){Xu1ehFN2WksHD(4nS9jg
z-2m}4)W`X`@+B!*@@@=pLf`L)LNBH5<<k|coHq@)j!2uEnLQuvPDu&9!vJCxK0cx5
z#;h2{j$hFRANI-l34h)8OhQ9W+yKtJa&7Q5ZPt$Y{-(fYapHGR)u4aO_vL>x(0g>W
zvhQRs{8m2xpe9!SiiJhDZNQ8<DGKeI`1>F5b%W!{UF8sfUhwb4e$=d}xTrIL)%-P9
zkti2xg2X(4)Fn@%|G0?7Y$SD*vXn0ft8<67;$Sn#takerPhli{`M0QhWbxF<V-W^P
zl+Bu2re2}N2pO0;DcQ&m#uMCnkWsv6*h^Wg5X<58QFqiHWih>aldPCXOLxKJes&TI
zrBYVFVwd&^arkCgX&wouQn0>=+5E|pwHft_srXIuN=)6|m-%D<;~0aVy%R82OOa-#
zAoJx@(?4>U+v?1Hgnx=Y*m4~czU{5=)zbeW=N>xUIUKJqtS2I%9|4<c*EHI3=w&Td
zaVyToYWHHoUewY2te<hi;{}zjSd`hu{9f%jpz9rJYpGex{0f5=aO9aFx_Gi-Ssl0W
zqxBaSS;Sedd<7;IZfr?)^4P8Z{)15bUx#5;uzQEW-y2bv%v=Kpj(JuTk*4w7I~OU!
zwLS7E?(%RL)k%TN*$U6eF;B%IZ6mq%QdfI1!*IzhTZSkBaL|T1R>OoJY{AC`O1Em|
zvhT0`0Als6Nwu{sco@6#XUo3ICpI<3B6GvtE8<1|h(9o3z7Dsk)(UwGEGE~&>u(OE
z3z2);UiD-M%fI*bzxT#S%PYGO4~*l2f%$j-we|HRwYR#sUFlkTCfVhiY%brU%v;O^
zVI&X+|0^9-?AglbGcrCwDVqmCIMwG{6$GUCTYUP_6jMcmlU6Rt3;mgUrii9Xk!#Yh
zEjhSesdHp&0`x8#beiJdB~u=Gww<qYNUjifu6H&<shBodyfpOyu!hYbG>u!6d4P9P
z<*X9gd0^&Z`hfg86JCEp61t(>OKf^$^lqaEfwMvnmEA8l!J>EMr~?Z&6kb%eVQ1&C
zu&T21RT~8bMVaYF4Gfd}uA-Sw0DgGV`XE|1^X6nEa<c0=c&dfV9L8J9ARlAWyT~}w
zmoWfWZHRqaJ!v@^J1ItYbtJ%uCrzfc9r9%Hi(a48?qdjOQGw?qtyhWVC~?F376y~S
zK)H&anW$u9Db}4JI{<y|kS60&-7VIhC{(4}eG1kl<wI{>m9V<2!A{^hr%g3;Vsx%{
z?$^@O?+d`n4>Z0wrEEN#ZgMJmKZdKAhUxSYbxu;@#+DkCH?w|Yi#jjDxzv?@N!Ojf
z6fSRpBwbcdLuVv)SH^|$BgXbeD;?L4&PAe#tZzhNcy&Nxdk`JFuqVwp$NQNEoUvQc
z>RT@ft+nsn>kLKwV_}8?gGn4U4iL7xJE9fw0dKZPZ%yjnhbWcqbLLeyZT#QL1@=Qq
zK;l<@G{hI>Se<f6aQv)1q8$Q?$aD8I&SY<XfVu$YknK!-;!Fme)hrZwwET2FK!HP}
zxBvaKDNn8&%f*y?#m4#TaNjRI$}WZvn~6Pz?>|-W9kYz3GZ9~|y$9mm3v@;7%aJNU
z0ev{2+9qANGNBW@SoNydWU0FbrlY#0khB)RkM)#_AJ#WuwnW>4VlKlPzBw&RkiK;=
z)M_{SyI;%ywQIL5jcbCnbT5|GOQ-Z&2D7d4?>UUu59)x<e7R7Uk#RWhI3rssD?Xn?
zwve0a7#vM_(1#CfP#37G9c&d7Q}L`@F1vOdtJ7zR6L{_0UbvXloDY-I!xWGXO{+12
zM4`Dlbc(mKzfRkcRSg}(sB<XoP0ppuhK`=m=rcOjY)PB8G!mt`BbF1ZojnUjcku_#
zi2RC))9momGkea<iah;DzwO*huTWT`+|X!I@AGwAd_hA?Vj?A{j~U8$vvpnKzg8h6
zqJjyTTs}geU{>)T%>B9qO>0r-NS<S0S-34+42Q$q&zWAc>9p8#r|k^=2D5eSFc5EL
zr%*@mtvW9I)kvPbZegWwBLRxdyyD14JkUvK)a@arDI~uaw46aC8dkHX8A`yhvg}nB
zR(jVJ+IV>+N#>X1^aPtESK_Khg+xv%LZX{4qTHN)`zZ1Wi4)}N%S>ot(aXO0<@qEt
z8$Y|-Mypc3y$9ZiUv}MiPlP}2c0+sdHp0s8l@aSBrc$5<zsuIW64yu)6OIgoF}-5Z
z9I0sjqLUKJ=Xu9aS)Y4mJ-U7CK@2^=trQ^nKiF#hthJV8OPU*FhJjP$AM9EBXNz1T
zBA2cNEeCR{84Slcn9=DCOOBf6@EwRu(cTy)1`0ll3Iz~&-#z5Dwk7gR$*wpUrlKVs
z=eWFiT;w6~yBJwIm+kUwKqBXqVa|I8la6>@dub9w5X-1wVBsAazip}g+NNUc2*2t#
z+X;0X%N>7YBs}!xERLQ6do0ef1*Ln#lFNhVCs4k%nOB5~$)KTI!U**y5F-+X=76)g
z&>hj453_RAcexS#1h>0>lrqiwdwj5=y4hUdQVDCd)KjWjbS+x^ZV*=dr!AmbtYtC6
z75_%1mM0ydbe2tLy&t_t?_bd23!K<o$s4bFX=$z2-|oQtq5xngX}1ol)cz^iZ7^c}
z+}X2M$j8rr>O`g(34s2`E)JMe4}NNbzHmx1+LTQM7<{dT?N62L_>UW;RvR~ut37x;
zTi$$PI5|yY26dk8uk>m$!gjP=mps*P^XXmpO}$}n=?YTh^*jKUq4*GXt<^AI@mF#N
zU&pX?5kOhKpF+9If7TlR%-$KbzEA(j%!JRF7%<_)ThD>f_xX#5S19w#_kq~Bi%ONt
z$)D61)gzB7ZX*wm^5>pY2?fW7`((S(0vR5C%BVC6IA8YXLmzpr*6IEPEqM8&QLu8t
zi-wPLa8v?8LzdxSP{^9_bhkycQ^tx56Vhq~^C}@PZ+6>AVf;h^cy<wMFO@k41ru2r
z<V|o(A-*8YgpLr=|Fh76k<pK^S@k=`jJ~LcFjJw(!!lp5WIm(`MMl!IC61b+bz%T?
zT!P-Xt+TVSsYxg|wEjkzrkD^E0kJAiM+jBwz3B<gj|7lT&x>1#C`<AmZt}=0p&hqn
z=VpCCyV9c55gJYCHr$7g#oW11ojob+@t_bGK;@R#_ty*n&_`h^%u7*>Zh4G<^&p#T
z-Qr<*S_vJUiEJk8m(-RJPNm^3Vm6ze|NlG8kM!@qxs-&WAflr_ZfV0V9N>GU57ecO
zXDd`&*82nN;t7e%Rg96r%SS?<8Y_c=B)@)teC_f_iHB`|h$c-gggsQn7cdea=D(LN
z^Ky4?NAI~PJ&mM-2Oq7hzI1m46<%6@z<YX!fICDZ8QkC=@)na~3-9m%x`Vy6!BwWZ
z4$&}X;@0#Q;c<cQBN3D*BD8Ugh`}0W2L47PA6n8*lG^Fx-+n*?{up6!dDv{^wCf}a
zEXj-tD_OMbz3bud1Pq7u<BLl1BM0a<3L>!jKV0-ZRf!1Ll7Scw-n%H3haG`{Qz!Et
zfa12=;7^I79h|WxgM77tbx7@fYf|ksMPYFHYY0RSC+6i2;7*Kp1HItQMl1iyUN6JI
z*_e*LFFlZq{YO06-zdq*43$3ZNNcULfK$D^FxrRxIqG;~W5W!@ygmI{H&KMxeY!hQ
zD52w>aXrG>e_4XR)^~yj5Z!*-U%?<asmo)+F@yYM@i=N?jU(jpi1&y1EHSr+_nXk<
zQtfU5w4G~`&~euoEVe$I!x7*i&Dd6{n7jcQir1%&!N10aDfyOhW2?q;F==se;-wmw
zt=3MkIoI~0POpySa^-r(yc%}=+2Qd|TJVNUFP_}4(L?9|bUf&R0icQ?J0z@YdEXK5
z@pD?%3Ly(FHFywRvSj{Da@@1c!@ZY@oKy4#_*17xmF;j=#D%}OvPHu<hJm=mXJ*n`
z*Y_PI70q(FNdxCYO4@VADsHk*NJ_W1#?5&<bwvYa^VLnPXN+6RK*m8Ga?W3Ad;f2}
z1N`^kR$}0Uar!!Q1>)OyY!yzJGk=~!P~xTpfsWow7mzx@XGSqf>7DTo&DMUQ=nV0@
z>=7a&o|Um|3n!Z9hHy=n*uZ|H|CSBeDE!mvv)<e??QSq*+}zX^B<{FexKeqYB&1fc
zV<+7IWWM(IR&4Q$mzTtnpRBbTJo%x{b+x-`LH*!^n(OTZ`A)4N(vs43Zb8QHbd7wA
zY2#yK7dMCf_w<iAY)G>~T;(|<w{Ov?5nxvD2bZqmsKKz-I%of6VPOV}!aTfohYe2U
z@fH4|aR=)fv)4i8N)QH-l$aLXu#MEx$Aq*mF_Po2c1D74u@nTH-TWnOGS$e#`fkUk
zxPCDJDg{O3iVLj11gRfN`ePcnYEQmi0GfXc_oaF7yt!*cPXF!SgV!vtHUDINIK=1h
zo57W)cz*hL#n$=brZVV0B7FS5@QIXhtSmW;VmplFY*AT16@JnaTkLOU*^&YXFj<8J
zfwiXR9-;f(MWhj;9hyKhm@)9nObMyLXzIjRh10W5-0??MQR-+uc;?sJ-Hdw%Bb7m!
ze^C<g-?sfdEgpE}U}hx>d%6gDsLdVbs{u>?DeFr#d1mp=sL5T^f#8;I?!<nDT7P$?
z-V6B8E7O>vD)K3+Hl$E?IYD07C-&_dYnMHJCh#(+W$lmc9WdX16)m?r$9QiM2!^|+
z$;Nka6KNG&Qsr#Lq`PCT@O}N{{hih#5K^f^>YKLIpO(={9lH8s0}lUFm^1%ptj+`z
zf*>sI&M0vOMnVMoX%mO}sT92r;m__G?u$E&2Z0^Wcmb$495}(9QMCM2sBrQyAGuZ<
zZvlx@Ypq0q_u}uH>c>#^O3wkI4Jpsb2#Sai7|hOmPSp~}`l_j=K6Km~ux<U!3YNR0
zB`fxCSulp=qCL9d-}zK4E~^??2xopJiQ*L_4#ptr3Hxxc%Uu@HRQ3bX+SmZT&jX&l
z2SEMO$;nSxvvMasGh)3cVW3*wd}3A60I9OGmL>rH_n7E{6i9<gvY~g_JE7<wN+r)o
zcJTd_$kZ2vY3=p1=~dU3*mTqtk}&!Fr4*b4s{Qs{qcztsePnk_s76q1IEpJs$ADx~
z6L86+phG*%Mky?ZBjR|a-ux|pvxrw$RDdj{XgG&H1xd-8NKU^zqZRm9Y^AO7E}|0I
zszA03jKL6_^j#o6b7Tb`u4E~^ww=ki|F2(RF3p8^M<`A8e`^q=lAQufOjUH`r}{Sn
z<G$A-xjawsY=d!aNdzgdsvs3B``OW|N#mNB<LB@-6aD(Wx0y{!XG%@OquecgZ1=@B
zl>4T!lB;W|PG{5REj4?H6tS@>O1?NR>ukAaoK9KF$09feO+-8A3N4P-oqOx8&p6j^
zqlB;ty@kpPbW`<5i;3=)=H@oJ2_1K$tkC<k%nG=>%_4`k^%@Ri-zLxRUpIp;L1XRo
z)|3+DGT~mC9}=MxXHp~k=vKdR-Lat~;U(ZYcx)olQ(g+Fnw;jsvS9C{4UoD+c;Jz8
zd3|MdY^n}qU)n#w3hIhHzYiX(h?vc%e&8x?sA%RkTbP}XsLnexyEpc>%<hPiXAu*`
zAt50f#j%q&xOY2>Bl-`gCyZx|l)d0Hag3+!(w<3(|B~7q#-Ru#&_Iot7cXcYf_4bp
zjFYbmC&uax{k%r63N5r}2Pu>*?LzfqKyUt|H*ce0h|1sV?6w;ahdH7gXj#A<!L9TB
z?5w=VS&Nk3qw9X?dc1{*E>v>p=*SKLndJt^ra`>vm>G9+J;wPzLFaqgFrWG<9@}xx
zZKm_8u;lso2Rtm4-1K3*GsjeihaxP52sNrMkBiT*$(&&Yuayr5THy}uOh_mddy+Hy
zfm5o2dJzIB*7fOJqg=CM2ymE!@bSJn`oyzroTx+TT0?~Nyy~Ak10(2>N^%#6eSL5g
zlK(bh+JBnM4?{_l&vmv&44-!tB7q870!@?7=xBSP@%_kobps*bfFkjU8sl+89pC}~
z=?c8_kNlg7Hr15qQ44J`tVz$C<x-ERi8pKqGka2;XX*#XLxVj_Fo*~eASJ*3gjK6e
z<!ZQ4G08|-n0S%p$;@qXn1BOSSavjDR})d;<zEQn_Rm@;&)^j|a&(y{?YIOw#AV$B
zFzV`2Yx(>1C0`ZW2q&^WqkZ9Bl@&7agocO5`BZ<7VUyU1mj!YKbAiytS$2nz)q`@-
z=;krr%{BBJZf5JTOT~H#TbWIJY)F!F6rojfM|u9VJr6gpc%8=y(Bu4GOM0%uXU6@l
z-k|_#{kw9R4~$dHjNs9u7q^{EfW`JJE04v~wvqE*KL8PfMV;i&U%m(o3{RbHuDCgs
zE~|R|u7n$5K#Y@1w=f+{lwhr3TZ3^!u~5W*$WZjz&R`-VP)Ro@KLrKh3kn$9{YyBJ
zcSK~}q4Vz;{@C2DDiSe;$J@E`nA+{_?REXh+XyFP)v*l?_vC`Lz+@D;L(RG++i=gD
zMMGaof1HZAk>b_S)ZLH*CRPwuf`$1WHH`0XqpSh_3%XlcQUfp=U4y6hyrn!7sMA<<
zZuGN;^}i;hSbX@ih72z=66bgqA;W0?rxkT9wuBB`?jKdE))K#QF=g;h00F&tMWa0H
ztmZnZ=`W|J?K&~3Px<`|sjew!+iV1KV<><(TP)%Rm_m%YiT;E$HJAqpijYwUi+=(&
z$TxZ3%w^w4Bz^Ic&-=29IBP4)B6lW)P3^%zO6OBRkvg5EStVW6(Ng`lmInj3zN^)~
zu@rbrh=5?crDgy^j3$-n$-wX(G9^<&$R-B{&6kvxxK(7?&^g|aP~u0f*=|Pmp{aOA
z+<I^Yave26r#-x5V4gsnO2Q8-UA_h*vBUPdG%>X)+Bzf}$9o>(v4e3(@UC*ZF)yzm
zbN%(*-qZa2E4RdlC!JEK)DDJ@wne(``x|*D$S5nnak9d%Wr%7Z<9BI0FoU{5Dum{U
z$2-A%F_-^r!IJ<8$I~!gKCYj%E7Ezet!Kbx(7uG4%8RynWO5xV0HvVClOR8x%y-Or
zK_vt`==g}Xb53%LBWe$yT!~7)kw>F_PG$|v?yn@YAFl>lZ|zYSF9S`;>_O|;@k%KO
z347~NxwuPLPJfVDyRx@)fbAWU&kJ6HSW}j5%&$yOc=OzuED)0ykAA=h2Yj!ew$RPV
zQBN?4a|?{V1kR|1Im3kw`qInSEWbXWq|QSZu2naTFPJ>~GFLc2zOE%D7#}#f9{!Oa
zd-@Sa;cuKLf6?su?SY_Xg=S2;A}%>V6^UiXyLjdg8t`IXaG5ZH+}Kk_A0X!e$krU%
zDBdPf(fr3gc=u!q>Y<w1dAY}Po5#8t>pb1QMxf?;l>z=17-xRz#kAfD6wI98(&C&|
z`D=ncu}kfbCB1!2#vWl|*8OtNP#v_0>y^?h)mMnZsR*`M__Gg9vg*ekK2`{bLJ$8l
zXa3AUbm1eL*)uTOA}*Iyy68;~9sF3|N+~`-;j_oBMSvpVZY2$nIiH=(qoq`2pms<6
z@_g!6`x)`O7I?Qyz~@qpo~RMiF3$+5nkXKWY_Wa!Nkz9<Sd4XGx5H5*XGa*G?p6u~
zrm+WbN`N;V9F6PamF)6K`Fr&AZTJb$Z-5QLV4>&SCAK6e{Puoe+;$mH>W^OG7($Nd
z4S)Kv02l<8G7vk2N7sCBZK`5%T?>%aN`VSL96!MoK-bF-`(XX(gDD3i_yO|A8Z^SP
zA)e@1wtzZKl$hAxT%8c`I#_)OPb{TR&8+ygE+~s#P*?~egiUpCuDcg%=@=RPzkN%%
zLiGLlvroU{nAL1gzq7j=LiC<SnU4iI^}1H9a}W;)0f7D^_fUXO%63jc(vac$idC<q
z=e}@!zBE<M<<c}x!ATPi?k?Dd`{!HrgDJRE>Z<PI3+gB6GS~vW#*3pJ;ccB#$s++P
zcCcl}G=&V$o??(r29bP}BEw&^@CYPtuUyMEB3`X<O?I?EiC@uSVPqH5reguI;hn9y
zCv%pwvbPrDRq_nkDDeku5^}1W#Z$9f*?>sU@XYr2FMMe7=P$(HdoLBoAGP)N>NV-r
z1HpjS+7C!qx0dN`JAf%hQTIFMKBK>Izy$C*324e|9LpDSt$4VL1!swrV3_W%`Csfo
zrWdZUwPIHKBW57nN9u|zs5{R*i@@SRc|sCPO4iVi(+iYnJR(QD<!t1-o_xg!!?yPJ
zx2Xg0pA8LFQdH;r!=s~p6*IH@q40nQj9vRq=5Fk?v_4r`bZ~TZbQwc>ZRN0)`JJ8K
z7GI*w3T$#A#B@^eyM)-@ySAPlXf}A{0@x?kD*g#KAqeQk+xI{OhJ}pBZQ9KEZSpXN
z(=|seeavxRxgvd(SlkRL2p{3UrkJe<cqY^vKn*59)Ke50h(~LJ)Qrfcc@DYD%G1h?
zvcB4@zZzY;{WWQ3<<ZB#KVsNy@%ZTS_Ck9zUE$8Wft_DP^RuU*Z@`7;w_3^sIY4me
z9>>_+qhS7Ill`gd+N4%Nt{7d{WJAdR?Lt2=%Xd@zK|Vg5hL4pJzm|XI{_NTaU-teo
zLmd{Q(+^Pl^+?bxh@&K<C}K-pvVA=~#Za+YRWkWqIpQX@*SHxjn?86LmG?EQQI|Kb
z9Od(zjDPDiG$js#jKNVD-BIoS@&U0*?{WUp3!cB&BcVyRdKxb}2ZQ@0%*xRMhtMr3
zSIWYSDa1y)2MMH(bE&&3A-5&PwihM2*Jy%Rv<|h!sPdUXWZPdrvOwL;SR}iDf8u1|
z{d@1~om-q6)*Y#Q=#zdF?<MTDQm+6Zkg?wrrSRLCc@wA6UK*43C<h5KPj3Jd@U&HM
znFRb0BQ{!Cmh62*B$Gavb!z$==u<e#2JON#707S2Mh%V`2n9@a<-F#D8W7GLUHAcy
zfqz(TS<+xghf4cNbULb>*tJo-6~xTsz2>bd?p-QMFOKKqC{J)rb;}Mx*ut7?Ri&Ve
zle^Q(LBA|$$~Qz1tl+spL9?E);H7r0jtp1)oO}m3LjOrC)PYr1#yP4ykKTab1CP~^
zszsGUwO6#GICiw(8JTbSPMNhhzi|KBF2Nyrm|skCOl!QlBcU}EI=LxPlJ(nm#saV#
zMlS3uH&n=~g#L&}nV_yy7)d(&+CX=0flM<KI>&bnYvgP0DzQmlP$N8PDjcwTJrFdO
z@9@?>T*)vT!QYZ^77m1j0OEb(Y|lTzXVMG+m<+aBKGRt_0k~fi60KoF$^4R-J@zfg
z^08${37SeNzL<Ht$8}`?O^>QO5zGSw{-2A`Xf9)5gwxJjJKH98?uHyrwKAvAOv_;<
z5;rzPHh!j|%6nIcF{hCr3z-KL8vq_(T33NI0eEq^Jkv2_WR(5RN}*dk^Mghf#{TFQ
zrU;d*Txs`{R`j4&&dKDusdFmOuufmitn>w?a>C|wwOSXKbg^bDNkG%Bf*h*jZL6md
zQY_Evv_KWq6$UTkT!fx!JC5L7^~&cs^a`?bpDy8!Fr0y8KMD{RuVB2-KtB4>;e=aY
zISFcIO`0v~4m_Fp)~J}@&ZpksTqji*#Pzw+nj<)t4SvL4v;~fWdHdTNM7Cqf_lB!I
zBTwGYMH6Z1Ye`Bu5JSKs`++<z4&;5qV)$&*o1e=Fsj%5r+3(R~O`JE*MH_5IVmoVo
za|@IUK+kTDt*INL=q$G}egQ7D&906S$QvnHos}n1d0<)_pr<n_!KF>)Ap(D*DP3{x
z<XH}N>oLw%6`P|mat0=8+^HE5P%KRTgr~(-1FY3tDGBxN)7>S762$%qpf4;_3D&2z
zltF}ZpVU>%<c*Y~Nd5>T66p&@b|Qa8f9=m{H*><eY^J03MbG#)AIp~J$aN<6^XII5
z(J?=pWXk*VP3dKc?kPKtFs&<lEAVPPYZ{HNUrqqyO(2!F38*ZFzKT0Mz+ud@GKAW)
z1H^By<($oC7L*k>dpI7%gZn(U9vBGQW0X0w<tsUZ*KOuy0yGb#$-{63`D-fwOjb9`
zfHi%#+cYzjQjgLXzp}t$G!-pLyHWsMAZz6iQ!aPRIkuFKn6>O$1h4U#(;4zcrr|7Q
z&5(j>D0Y4j`qCdzAR!2G>jn6X|095hyaP7@;D(ETgWrnx!dbT22J<2S<Xo~F`IUNG
zggxOwBBsP1&e)KZf@$;dUccrEiyhVig{&1<nzs6261@+Xu<z>WMDd7~MSuL8{sVg2
zNR@7&M2I;_D2KAz;Rus6&|nK<WkQfK``n*0JU{>gg9G0H!{*Bi(SY#KZ~4hsxM)Y_
zrhP{-*<#x~!DWW0>aR+RbS|69Z%w%zwL_{)-P@3WezC|XSJ|JoRPADw{EHhL+{IvQ
zJg*2}A`pbu>G%+>L2(E6o&Ve2An0Pye&IEn65P+O&3FC~`rLA8A$WS`Ki+61G)7n}
zJv?y0a63`$d{^!48m7bPL|hS_SY~gSz}>sub3pIIoeOgi8{(}5ARQ!j^USGTGSReQ
z7no$LM{y^r)_?@)FAfBX@R4LzgkV;}Ff@d3%I`GjJ?g)dR@O~NC&_Z{C*vNI<`*dG
zg|?=SXQC0w?~+ZA{B-+tOM_w$?8TL!VA<o8cc!|sXy8KSi7Im`9oHsZDe+2vUPbB4
zFQ(QPGQ~yT=H0Ym<kTdX2jtR7B`W6ZUt5;ru1A7qCOEw#L9pOpbDL-8Mq1Vi813Yo
z`}crNnKt+UM6u^(a-H*$mZ$8aA^NP$?J)yXUvN89>C`c&0+<pxh=>@ZmR^hB@qyKJ
z^QnGAr8J|HHY~!ORb~%jxYhpzc)b5=KZ9?;Y|97m24JC=$4PeC@lUcVAJ4BT>CLIv
ze$gmoFD)YHri~71^8vR)q<6;m3kTZXRUx`o%Wpa6a97yDK>@_pvOApH@BFIn%WT%`
z*`O5zWtv)ZEKa6K((N9xgsZhIM$iW*Ia>O|LQg8AE3+drWt)DUQls28;%Kv4k-!^A
zU+}4AWh+EE=L|1^d_2klZY2CZc1G%D=uK3_<E;!y<Ia^G;^ht%b5Ha^3e;2rskXYV
z7Ren|+i3{qH?&(F5O^1!%7-QxL5Kr~yQOg;QgeZFz`5?1VKGu}%bQz+r~Q%aqAsQ8
z7u`(ToW-<@|8G|pSC$GkH@rY#`#v0nJQ0t|QEP>w{7(s!wswi<ob&er@opM+>L_o6
zu>pG1nxK;PS<s<t<-YtA>XIv_*iS1Z{Y5jC4kP#$7VO{#j5byw$U>M-69AnG#Mk3r
zF<!**N{Q0|I?$9i01~qc=ebw5@PHfc8BRyU6^hPUa*Yx+=#9*Z_0O4s(_>*YOa{B)
z0(k06vIv&#6+TEVCS>`0sO{~V@~&#3P<fZoLHhkQx$4DyhQEu~yi~Hu=NlM+i!Fpi
zRtIvO;PNF1o9DYwMG@LGcU9nm5X5WD#@lu+cBtc*?cgqcfeUhnzHmt`s_{VW_~1}S
z_y*j5{qeX`z;2d}@We@Nhi7i3WU5;!8iW!%8b02})E?8|*nJX^_els{LllGhy^rKy
z0ybOM;2*2gx{g?NGvr}doKULM+B#Zmf;hKgGdpHbHErL3P&(C3b<5UevHqUrH1dMN
zKWB&8@b_c6J@VZV=smj=G{f=0U8`y4l2`nmvu^wtAe1(7{Y%=e?L~@nA!2s?p7Vjf
z{Q`5z)gC?Y`mxC7Ay2_-wB#IFSu531UOZRD=m*%`Eg*TtriKJ_nsHztyNtQLzkS35
zhj)FB?AqfT^Cp?*8oOA^IfmMBH|L>epYglyR&b}tMAZ3~1ql1*LG5CGi<`$ERgct=
z3v&e+9e?yBgS%We5ja)LVQefrec^?zH}k8Ey;$?LgphE*EVye`M94>k;yI7L?p<l|
z&GktBN2SY8lGxJL{z5F7LoS8eulo#~P}NE1SPwM~n&XvaO*!pvw-+7cVYDEXp6*I>
zRJi4TmUVhuouI+947Xt9*66&w<3vbTEeuE;xZKs)B@E4tC7CSYT)=@#dG*&Ov}LC!
z5D&2H-NgKHTD&w<y{d|k0AvGHP6htMHJRfH3F(JoR`XMK`sx$0dylF;r!$|5$zup1
z9(kvY!sV(fDA1*SzEYrbrKZNMh&!AXfBGHs;rF2LR(~8*oFU>L&z?uHJj|P3*|TWd
zEbQUZ>v_&>beu(hyNMPH>1H<f1Rg>lox)ex-n246I;XH`I_cJcf4;2w`_Lt&>shFX
zp5OtzJpSvgrwlccUZa&UxigWjW5x_HGDXD_7Q_B0uGKy<&B}7hYdq2>5t0t7_dyQl
z<qiW4Y;u)qMkrbC$=QnIVoG%Oq065hvZoFcrOFfI%ubV?I)0T#r*^8mPUO&Qyn-W^
z^|R-08dE-I11=-<+xJGx&j(B<iO+dUF|&(#jmz)X{YstAv5F}zplx-uxp}4l1jFk>
zS9NIyn!+-aS)4!aj_PAlos}9XNv|S(T8A}PX1ivQG^%6`!qY!akQS|v`cD=W6004&
zO)h08INH3jf2LdL^~7J29_9g^`<8&zP;!}3wg)?5y*RlJo-MozXT4sxW*U3>7(v%B
zo{+;qA9jX0%c3nyc3%~MD(5(;*qqhZwtPLbju{+qU1b+$-XsnUS$W@qWOjG#+xy%<
zCh9zJU>tecIDRqYkE5#U=SdT!)Z-sV6+GeA(<sS4^O_>U3+IMh&e<bnGFg*aNp)K<
zMVSc1=z5s0S7SGCoL|RjjZ!ei=hIZSDKI7>3bsfjfcKNsKudnbJvM)}m=vYisnp0s
zaKv-j<ycl`AI>@4F^+4S3s)!cZpCwr?U>s~&ve*``RBCs)dLzFL*L^Hl$NQUOE<#s
z9-woKp5Q_{>QM4<pUWa0XxC`37B<&ER=5s#A2*4|YMW?_Hf5|uO`Q}rPee$H9AEPU
zCdq$#WZW0!NahreTUexOBSNDg1*x934$7=*YR;%+BgbT;Bk;6~N$Vb0E*LSki^-%b
z@SN@A>02v)XS8=8ZKS1}9{w^|xZ!ydU}MTp-aXDd3)MU-{*`_!_cr;yK)k(VHjk^M
z@T=Dh7BUu6O?mmPSa3<vqZD?OdI7oIpA9!lQL~jTp*1!kvD?VBj+3(>cY3tJw<aoO
zC{ex#HssmO@A_p%Pfa!ggUJkL|C(2s4OK1G96rO*+s|m3yoYwIpK>~UIwShk-9)P1
zvrYh!ylj(7=&fGB=SjJYh);cHi-l#4;Wz?U;TH++Hcu7YFCmmDbYcAbnoyZ%XG+Y(
zG4oUYF2&&s;qO)i;_p^OX{w!Dk@y>%a|zz>F%eIeSKz+Rs&XKwx7&+$2k97_E+X;M
zf^>ZEgU%^Reu@}hjmOJ}J@y=2l?Uf~nRz8pe&t(&aI-b^9eL#d6v3+aPcQIZ$ip)Z
zBas<BJI+~AJ{?wR&o2;&wEjF_RQ{`*vc!q`8%5@HOBy4MZRDi*%Ey~ExSp*WL?v_$
z^(#9>6}6>L=64Wb@Ppi)Zfj~W$E_3`8Y59%(zus3lP|_}89C)Hy`cWQxz+p-nhmpT
zEm<FpRpS5XC6-OO)Wr}cZPbxNMoI;$5PK;_i!HJbZzb>vS$jV#S>EMQ3$Z#AYT=l~
zhTqkh`8BDJ<sbc`s^c}X!>SANeP+gj<3v8=Q{$&hj<`XwPdwYxTJ@<Z*|+HWCu$dZ
z(JKm;HlBoFe_`4T47c0PKGQjdrmN>Ww5+l{tG*6r7z&pUyQRh#7ie$9qENrgahmY{
z6xjad%R)yvAaM;I)$+;shQ(odb7HbijRiY~ud5B${@C7^2aoQXjnF#W^JUBkm9t>o
zghB10sKze@U1nkX@OK>fU#t1~8w*AVgUp>suHX8e?RxUsU4&01%!|B+R>emzTb=sS
z=QKNY4>bf+J@PwBh{8sqThl;M*sWd7<P{W<YYkb2RZstDtJMu4#rs-AjaN7(Bx_28
z=bx>4@@<a}s%C2MpLl3E>W#j#Q_f?gak~NY$K*bDNu<&;H|~%zfAVOu90!b0?*J+E
zVB+d%H*UobWfe|PYm8t-;rYFRm`!1^dBmR0>p<p8sESz3cg(oLuu4&{Ig!z&X;#d|
zda7eQ?ETKQM1>{($bVr1<&n2^=emTCsMrF`wFC`%E;IGaKf5jSoHs~}Ti2F0j8X;%
z?y5lm6Q|F(C?xNnQxPGdU;aja%Aob470onXe+DY8_=Sbk-t(?GRaTmq=UXN*RoqbZ
zsxdLAgE_OX`<ZnLF!*Bx&#d}2UC!4(9GiX=0c}Aj>o=OJqw_|VG5+!D4gOy2NM50B
z#d`vc4WTA9onuo`Jo8ME+_7!b-7$N#Ep?y%xLjsuqe^sqNf~`{mvciJB-^azU2g+o
zb%+VI?3k&bq*!m1BG+=JOliG~_3!<~;H0|Soa^K~Llh^nt8`+_koUqs86I3dQ!5?e
z&KyZwX=cw8Eb9O=K4-6A!E>wP-k>>x69;aHn5thtuDW?C!gFmHdLWmsa9FYOq~aKI
z{ZYNaQKrp`*CP~U|MC$?<7Rq7jGzq-<gdu6SZ{i&Wam?xil{s*A6qL!DsJq^ap3*z
zOnCGRtQPpz?H~7cLXCPVe~(0o82Ku_!^4y^e#*(`F)Na7^sJh2?o%##ZkLMKWjXb+
z*{Aw`w@+mun9?q$xH;Q=b{FhVvnHN`gtfY9>`LTP!`izgkxoP0XcLIFuPvq_jTl6g
zb{{=Vcd=xBp7)kn{lW?dPfD}^nh^6N8y9g<x#w=rSytw_fQ2vi|7f}jpenkz4<eu-
zp>&&obVzqgE7Eo8?(U_L25Ctt>5%U3ZZ6#+aOu3nxA^`)W|(2dz2Vt&o?kq%4OW+<
z<ugnvn!Hv{_5Tp3N*t|GNl{hxv$FE^kX%-ONQi}%6Pja$c4?0e)589zPFaKd9vQ(V
z?MQ8i6NQ2;yupqS7cXvkzMP>KsEz%up4HLaR-tZdp>-SUVj3LB4f95!cU>>U*o?@}
zKb<X@5L9=had)zn3??tzuS@<xi0a)`2vnW^Y4Ct4#H?Efo9EVp-ygBeKEKXBA;kqT
z-7UP3tTfm4hE3;4gpv8`R{~S=P~wo$<ln-El&P+sTZMkH%|Ho^v!9NDiqS#t`|h>1
zuXjC-x!05CgJrf4`m?nUaa_5%6j?yCQC?_dOKtg}(W&Kwc4n`5$!W;aL$VA2*k@#9
zc(=FLPgLjE)?O9d`eWBnd^}Q)mdX*%Dz^yyqIGu2LFxR5mT@!=YGkJkGoXywE#R=Q
z<6D4+iNE^-pYdtV{=NoeKG`SJqTMIZzz-J>g&H*S+E@Ptk_W{_OdEMtZ)VKm5`-xA
z&zGZ;Wdp|(mH3>+{nC7INx|39r3H9YXkP!kk6SN^pxuTPRu!DiEXK9q(8Lw=kIayW
zW!2rXESoR@86RRHK_-s%+z*7i=UupX8>=fjAqDvMO(z^rHh^>AjI+!30}uMi^JP~K
zVX?!9hdcgE1}!Ydj|6W-52Bt}<3VR<S9W$>Dkkz+Xw^7tW1$w{C=~2sWsMsn$U>47
z_#*(F-gT%s3*yYY2?-5FEGt-rB;NdZ>r@1fLz-89<!C_Q%0=Woo;59y$9zz~EPd$S
z{c1KNsf&3L;Jxh~{+#og_N9AGE6sL%`me!$htspd<MoozMvpzjsPvaA2dXN=sX@*e
zUE9MJEVee=W?-hfd(1DnuQAra)hQ1}e))BK{gUh@BG!13Wu)1*$^@@xtXP5B#)2Wu
z_V7Tdc4`ak1q#xK4<A(3)TXZWf_{%|E+gl6O<~370-Z-E`R<X?V<tKyxgYlD>wR&8
zYGP8O(WsHF(!H_g92>yGA#%!W(+1*2>#twGnlk?`NyV{UUwnS}p{TgX+6j$A3tTgF
zG&E{zYJ^xfpDM>iF5`NQ6(sIAqUQ_EfDaf$zH=M()I(ISx*k5uD>ZW$Pw&_qJrg%K
zln_z|UV$h3dpK8j2r}!g=OHJ#`|-OJO<Jpl(K0cK;dFlCwLaNT`xh>8r_y0_&jd{>
zFqI-~X>DyU+d6onq@VYI?4f-aTGL<tiXpuraEPAA`Rhxbq{}h%6amktU+@F|I39QO
zPY0|uN@u@M(LeZ6s@AY!uI^kPN(H%p@~}b)C=Hk9vRN#>^zO`)oeIff?Ei4Suf0#2
za*S@FijyBaP{J!tIvgnRq=1}S1B)3sXC$G1W3;3t@nIMD)w&$5z-V6U1ls;5!iux^
zo#J3^$xs1=Kg)g@?z)yUvbVte9~F4|@!zzkszH-aOb;=AMQXQr2tLIq6WJqxs7Lad
zP!_y_%1y4{bU27{wKGtD_GbHu9O=@qd3woP>Po}9W1b7r{=iLp&Wb%$FoFG|fT?2C
zg5B0=BvPtzse`8JuK06~!q~5wFVhTZvV$-RQ%CtJeBS|C9*O?l15VDmv24d0w4PEe
z-egybmhXKAo$qS-hEQeRWOQW!h4^l?y>f3E*03T<lf^df!y9=^-hDH=6^4`h6~^kz
zqFaoTTDRO}cZx;51fJHkvMYty!&*xdUwrU)Y{Dv*GZLE44jkfHTicSI4pyvHeTG=A
zgkE|4DsjkRhq+c+gIqBXYO!@$XMGA!sfnYgwDW~~ydG3j!-a*!>dnRhh|)hE8#$>W
z_YM9^(9*f3{hLs-AoU;mOunsv4lX2(lzUw}tMb%NxVM*Hu{G?~>!&u7jNL5Fn9gmh
zIWZEB1c(eIvj5_scyPeBen2sVWyvP^T%od2t`5qry;b>ID!YkZ`ACUej|nl7owIER
z7~+=hl^qZBfNj;9Gm?nnRLLxId7bqgr8)#7*42SrJr?5H%eL(Yn7ex5`9QPbaMAV`
za|Z`9p5$&mD$SbL*4z8iF-!6jd2rynDqCunIsb}ZLLjT>1&X@Ujz%4+xsOziUav?@
z;^lEDspbg_5@o4GPJ!mJJS4FrT1m>ssy9)xDte0^pl^58uG07Cn~tMTFY$J@JyO=s
z+cy^ALb;+xRHPB^MxFdD$EG{~P^{OJGNqH{^T6&+4x15Rw>BqSMZa|Gs3GXpj_FSx
zx!64J`gUxp-jviBFddL6D;0{u!_#2a^{S>>xbR6OH(C9_UEiZFA}O}=c)|FH?77uN
z_khp&KubsWSj?N!L&l0)xV*9x!<m;Gv#gtdthO8qMdK$@zvk@0W8(05TgJhsIy1LD
zeUjn%iV;M|(fb-_bO@2noQ1=e_{Ea<ZL)++w~|t570{FO&oY63Mo!OD7PynqSEX%i
zPmOjuzV(y2-&}cZ4ZD$7i4S;7nwu`{<pXHO8iiHLl?@M`^}Ht-AYW2@aQ8l&^q1OH
z)wsa|UR9}!=uJ|-RcX=4w;ZB;H^0p<Dx~~Yj)DA>8MH3`Ph*9nTM<RX5<?jkPCS*8
zmA)U*5)TF!iu&>la%uN{RgA%h)FroAGDxwq(O<8*7jE%i7w)J?*cxXMx=}J1X@J$s
z-z}M(HLxP9$2EEORqH?2ZYHTK#DLzCf+m=59KW_hq<35mPKJ@wZ*FbSb=K35D7Q7d
zX59umj=EQv85K06`sR}OBzb_=%)9rgbN%&^a6RgeI)mS=CwPkzy114Y$dhlMpOaL*
z*&LLc2H2xA8>R8A4cP51fCHI}yPGG(#Stn-^j*VC`6;VLURMri;#ocsSppq3rY%yN
zv)PvuImi#o+cKq-)m4Vqkra9?L|`%vWP!uK3`_OciD(z(z<sp^y?hCrF@pX!m)fo<
z#k?~TOY`ZDMKKYZ`&(jKdJh8hmiOLuYX`k-t(D(w-lLX^-@gFd>}86#&7Q}!^#)!7
zD)-0B*w3nJJoNFT7}rXy45*CCF5wteKeGNQ#G%*!6k=w)g(h=C%HStr2U0(f7{)DW
z7mtnYwv8aFaswzvG5HI}k+L_5{C*TUuX%#mD4f6&s_ov9?+violu1ITyQ`;r0=x3I
zUMQClUO*aZTB7nUsWmEl^+n_TwVAK5#t+h_a6Icz>l<5)V+5PptkV|k(GF&`z|g)B
zDCJbuphx5Imhnp2OS?<k+fjBovPHj8i2b8yBIO#m$bm$wVa#6q6GCNlT%vyV5y3LB
z(Pwx%ILlnX;5;lsleZcrrq5=`MMM(ux3d>862g&;Ls3UeRr-ba+~pP1=>e+z^2+`p
z>3#7Y>&Zi!u(3X}KzB#>dVjll?w+C>dUzQ(a$1;w2h=6-hQ(<3R24d}$#X&ZU7bX(
zJ3!Io;OXKb4TT^wFWFOdJojyDQHZMMI@f<e?Zx=N>f^i*@ahtT47CS>0#cxZFOO^q
z!#&E*h{{-wl=`x8A3SQ?%QrqK;q!lXl1(3vP>hE0I{qv%!LpI+e8bmR=iotqeS;6;
zh&qQm?fNnNxesMj1u5$Xt&6nYpa9Yx9yEsD%m_d3E7bT~g3_oGPPeC7W6&>AaiV{r
zO`xcJf<2o-#Ms;v6$RuT(1f*E57;3*S`c7SXxq+SMYZ$0>^)97tNR~Esf9$GYg>3?
zBdH-7c}!UDHExweC-`k-;uZQ!s^j`t<;M@OuDK<Q(ow6~x+`6lzkMU$-+>}F?B*?k
zZ#(Swe3m`)-_Y+%$7Hs~k@_r2UuL4k8m&0O;WeuQ(z$=ufR^PLHx|U)w5P^opX?Ew
zX-AV^*AdEX&=eDaxp;2dLmn*sgD+iOP_wptjF<i4a1WYg`71OE^PRO2FJqK#r}S*w
zhXQ(R>VJnf;j_=<q`5rgn;&|Q+alZ#mHeDXnO0e_(ET%4MU4Kc2JOC5fkHkZ@0OU`
z3(^h>Kmce@la>A)9Qpno|4D%@C{K5tW~c*sKM|o)&6sUEH<XpzuLRT(Fu4qd-X5P)
z`5)edZtPgTmw4b4&rzNzvc$T<xWj5XBnzEu*E={CnMZhL;b>iD_rk&zi;4s2D?cjc
zsA_F+J@@#Ku6*e&6%cCuEl`H1HQ?xj=wYHXAa&-n_D%&-6+S%Y)9UojgkJ|u`;C?~
z0Mjm|-?e?2!r7P_e*j+?U9XLC;ds80FR-*R`}a<i4l1_ri*V4LwUk)@Ku5%+G~U5g
z55bGx&sr{Xle6Uz*F*_>OQA`SOiR}oh2SFf$CH19k*v`F7v6-F-B^rbo=X6w*op;z
zhol@%%$^Cl{oI5MI6Ix{Euga!!Eb{=%Jhz&E21&w1*mq8&YrgqFxIMbEe;cZdai%s
z9?C>~)|tj91Qu3<Tz8@A_8FRKgx|N#7UHxtx0m75#O^cRW}90Zp4aNNee}c%uZEET
zNOHzmc*(NvCve=&qg%KSY?z=`#DLH`%MD*!n|TpFee+XQZV5v&GW0cU-}d9tG)tVg
zY~iYw30ot&7!)}h^7#!18+vgZri8UIudv71&fl$hB$4Q}_9Bm?YGr)cLQEiMQ&i5*
z2pQ>L$(9v{<0)RsqvIM%+dG0NX#DM>T_gU_A0<89txZKYKm*K*$O}H>M~r7YqQnK`
zaK>o-)l>E-f`wChhK5SJoJJK%px`JvSQJv4I*g+&(UcFkj%&d;QxMW-N8p~{gDfjo
z54)&y+TZmvZyKnV)u!e#vkyC6)sz4g6J@HBKI2PB@PaDNx{%aYS??ik|Cr#~#s5Mn
zfDbx{NM90T6b@7axTHo@2%Hbidg`Y(jegs^T&gpq1iybr{}PnS$-gNcm2&lfpv3{H
zxq-lkt^iV=+JL-w4Rx`vBAc@bRj??0MKs1s;7zrzH)^EP;J`ePlPSBQQN|!v)&FH|
zDLH#+vh!3D*IO{fJk*v}@OS}D$_HklJgYD~!O#m&lks|5;_yw^I=8DQZ1$cIF3NOo
zl&Odhf8&4n>J7uVh1SN<H*$@1SrTo|?7ACl5q>pk=xzViv8B^6c2_jR+mi7-#~D!2
zo2{D}Jp5P0@jY1<?QtxbD7}r{aqIL_@czcPozT-!H4W|FtD3V}8*d)K4c`j<CM_PP
zC=DF+-4-lb&SA>NVl8hXS>4(-a*%ub`$vZKu6>hk&*2HU1X~GQC6+S{t{c!x)6v>u
zF<Vb05l{w+iZ8t6^_^Gw$l)zbK;ujgYTdV&)*IWUkZlXat8>($q=<@Sf=dDRIH7k`
znHBZb;q5KR92Q+^vVipGNInBwnBn{0oB?Lys(pnf)y(1Eme`I6C}8JuEEo16a4+I(
zzS1YM*$Doapmpt*H7q(sVzv^WG=t9E_MBHodAAlf3cusFt2Lrjru6?K{BQnu%zQvg
zuV`;^f2lSxRN^7iD`-q9M4z1~IP?n1yR5{|j0Nb@=<dd|zqZ<{AjEL9nG{yRGTD5V
zHm$=VqRvb@+*ZaI0}?cSczt@(E$l7GH<a`5QbHDn1`{Nr;-Y=k&BdD7JC3ZGu_YpJ
zOdQy84XC2sj5x|)(te+<i1^O8=VH_DARZ1N#ns=YW$0?5zQik!nH3o*P+AG+Yt#&$
z0q)Y!dj3j>Tp4wLv$!*VgD}h={@l!+0@ij=W_u<Ki4~DMF<pzi7gIP>e&i+KOuzLh
z`TJH5rpkp`Kk~$0#^pxI_F#;iP`fdxq9vul+%*!e0i}-qo;UC%i>3?PaUjW3LXF>*
zq03baq_5gA?5!8t(-~hsc0u*I5-$M*>?-z@ZzPzkHgkRsF15}9^fJQkRU57CLuIr`
zfpA57A=WY;B=5i`b!Ga%M;0PV_U?Zy<c#?2Spx?H<;lq9Gb@f)arGxn@ANP3MPWIo
zpUmfx<d-wld;eHzoSlD)A5Q;;8`zD~`WaHzH0(;cgA7Gcb%@T4>jONF#%<EP&Ph-r
zT;?+GNN3AK)JnrTO-3g&)(Ph#Z?Qqdn+Y=?wS7S*4hWxJ9CB9PvUw%5rf!59yRIG}
z=k<S-bEIJU32E_+0bzQ=V?}zF(?!(-vjp5%G7Hg4;Gom-di@8xo_Jh>lj+U|PP{r2
zFag2Me?`=GKv4@gwY4aZdbHxQti6Li%!hVLSZ{CIlDQr>6hGxK(=zfvrQ&G3^OKb1
z90|rqA2-_3h~p>sepk4R5=%6;HtlS%%b?s;jXh{YxK_+-tmQC+xqpZ_^0edRy^H|F
z;9$NbDm6;(;US!k!)rh{HTKEjnkeB0^0bnw1wAi%=BW)(dKhjU%5&&k@VK^C@>80=
z%8y&v5Nu0g`d_vs+I*?>?80@o+@H}C*;^v5`1Ycv-qPGlA*A6#{i-Zqfv#TL_7csQ
zUgiau!|_obN=X{AJ!{=UyTl=}50i%g&$5K%e=|Qwlu{9B)zrENh?~oTmOj(fRMRZb
zfKp~TBu@wbb=%W3--37-6e$6T9Y_n1*JX#a42i0}_<+2x%|!l~<Y}GjE*UHb<8kSd
ztM9ky23ssN{a%t)uqfnaZVe_-i(GBABAsk9cgb~P&zi(+NX?aGy18|>PqK_;<8TlJ
zrg%ywj}bce)q1=1&BshCc`vsv0g${-(wC#koo-jA$2t>~2DYD`mqHer+Ql)jt~-j1
zRx%7n8qYzzRf6%~YE1pMePHjWTj45?lV|BcfTW>jjKspf<QLZL{5OG6>&^c*C+LKx
zxgPKoS?+&Vy{7!_JXqy2e;ek)?#Y~(x$p^P=>b=E-l4P$kl9NEne>*&RnbY@!HmeS
zs+diZOB+7|Kjx$`kL!zd&{ypzHDgasx`Vldpr4wup3(n2cmLI18Vp$_9{q98SOgj=
zJ&eoElDBnvc0B@TgAFd%$T;pUTyP-SA8Jr(B5<?DTkXm7^Q`vS&+6UbR(Lv%^R3ST
zP-cHu`ndi!Gw6WswjR{i2K4ra^_ySU-XUp9Md-H?l*WS7HsJh~ySxZl&ELVy*9Y2T
ze=#`X_ejkl7_$5mrmbD)vWn8qb=tqASxc=2`2u<@KUeB>cD@A@{iP>e9TLw_E{82b
z`lJUKoUd9<vGn%zunNDeBs|U8^!;G>;z;=u@Sb48Kv3f+#)}GG32@(mUXk4YFaynM
zHUu=>!H!qm%l7B{+DGrj>hcY(b=d7!M-`M?W_$+y*DLo&6OI`H+x1r^@8a*S-hge|
zOta{74T}7&f{Z&G?K!b7)AV=MgSx2I4{b>}s5TfTV##l;mEAb3d@M~^H)Mf9N}LuC
zxOkm@swS_?!GzK*PrZ%ZX@bR-2R5{CY||>n%?S(e91Xw90@iS_+aHYRMG9%%Z9jrJ
z+mQ+gQpUinb(v#{LApX*h}HU*j^%d$bPK++YNgkDXccHXXS}BMnPTOB*`$w2IdM)f
z^<<P_%f<v+XmMjkNU8r^3(S8Sbgq{wGe~^hek6599zVy;FPB0HWE#Xts0g>ho6G!D
zArDHhF9d+I_)yySX?0Xkj=2ARQ&qc<F-UDmlqP^lWp`|<9Jxxtr+bXb_0N*7d9wR&
zO7!HPVNaO;4S40vGE&MCSL7+G*SoyMXHV1@!u^bqDwP2zTqCL&H!DVh_5{!!p+>k1
zybt=F>!oESbm(NjLIkbeqWV&}vg%!cN=)z!UNL(yBMg$`-FlJg`#J>gYLR7{ngAi!
zpBL_F`g=yUHgK8~`-(DWx`x+exN|?$&@WE!H-5brwKV8T9(r!Ume@KiVsmWi8<$Ee
zR%dFU83w$+I70i8Rf9OnTSNfb_aQOT*R1dK<pS-GZzVxU$C*7@q9KcnOGO6_tJR}^
zj@fdnmss*CU$<1N)Q4sUGLtSGDCEeYU0}qpwK$*yv-ts_B>kTjk&b-FY=*Rr#qf($
zfJ~7FMd0eK{O#o_%_jBV_})$TSdw+1;Ml{nmqPisjm5baikv%kRqp<_{+Awc-itlj
z+xtNaJ7=TTC7B%<3Yk{hh`Q<CkNlts*M47%-zNwrb}AZy!0l6Ir)i~;Z;w$k6Z1cw
zZHx4{F&PuY%g^9MMH(0Mp!bAbVch1}Gi^N?SU%z@n9P7QpjEr)1mNM|m7D>Vn@_Da
z&yI>d=kd8{*fVhZo=rXl8Fgv>d<NrcZ!?BT@%}Rt=DhxfF?Hz6(ZAZmu)E-QaQ4xI
zlCPT)mxVLm<?+e3N|&2;ppg`>oHwx6B5rLL%VFAxVXilpZp%EF;kj!Gma7e59<1#^
zT-iKO&jub%s(p<DtTx=!)U<-LFU(wlB{{HAvnMjWMaF7QEu!Sw8??1*VJ_AWH@pIO
z5<hDW)HZKL7r#bDOSg}qujnGnk`<Eos-YIAP;;lq-mVh*a(|+6u7wQfq^X^<yKfa6
zKSU;vT1y96?#__yX)h;({Apqz=EP$EZyb2d2hbG->mRT=zK14rtV8a7QPbU$JKU#V
zfgEM9QZytK&F1*1?a?xKe?eKRyT|jV%w0@)f3q4{mb8H2Gx8@#?^wMbYa^Q9R%GR+
zGivh?N~0J|iYyANXiHI2xf(91K!;Z2)6!gb1e_wdA2e$0_E+%VVp*&y*ntfPNt_LC
zYrv*TI!MGo?pr7put@t#Kkm+*eMd28y|x|Nzzj0`_uXmT7VH`hd>0h_Y4;sH_AZCq
zpv^kebxPGYG6c?qTv^$X>cCGW<iGd;dKXHx9A9fdk%C;I=Gv@$Iirk&%9$eDeI1L2
z-iW|1=X928N3obr+zHkCsyRy`YXwMAg2ZUHfoBzeL#$(o{@9#|4!z*9dwZ>#^1pmM
zioK0p&h{eK*eNW?%8ABN!O1PeE{nUuxzy&{oQ6RTRYG05)Mb8C1T8ZQ;|C=(9$Iq^
zF#DL`v{*PTx4K+HYFrg5nbSTE4cZ6A)~S7&y-l^tL>YrK&h#b>_<I0ZU^!vjq~_2x
zhh6`4{*W6`v~Ov`%@mgV$L8ONbJ&;v$gF6po~3!|0yT@0fa<T82qg37#z+uo<S)*a
zxdzauiL0ZKeovLMsX3|3G9Ghuf|03v8d2+ZrOYiCPX5i79Bj=uCuxwfHvhCEDIM{S
zmXD}PJTw{jO#$vw=d;1H`tCfAsMJIuZFNjyT8s#722R%o#rgoBVVGUy5wS<}^f`H5
zmurA#eUzw>I61hKE;E-M-E}7;IR9Yd=i7g)B|Q1jD$T1uyfgDJsx(bK_m<;DOpIIg
zTti=zL&@M>><9Xf@zBF?u%Uhtmw-Uq9%xDRV?C(HSlgXx?igO`%S=?6`?8Up*2R~X
z+EsU({EIWaecvEOlW)v~K0V3xtksCUD6wOeZ;yJg#SK3fRApptFIJaOA+O7XDrmMT
z3)(EU1~1l0)&7Jgrc#X(;6og1VoFKE;Z?a4S$DSR7k{X;vt~BMkYmhX*XqO-046i;
zUa)Fo430VGE8vRGDCW8V?y2!~3_K^5KrP2K=9KyMe+{#0|9>0CD;+t&j^C{#I;L;u
z&1uNuoN1xp|3c^&eQDkN7&NsNW$3#Aff$%iNNJ>nU8}|W>zgB*;2S~PbssLbIIBzV
zh$^n_qI>6wpI8EI+DKd|oGOAsPV78058a>_Ld3U9Pwt|p9$=eXwc7@!9X@k5N8FJ0
zL-vb#gQ+;c(hhp<XY`;i06mPLvU?EUMCE4tWdGEejBqGQLAbk4bei!Y0)uvY9GJzq
zHSRZ_)l4d;2dS(kabfQL=l!+;&&=Yf(S@(30eVa#fD|*=HVfy~sd788C>IpuKJgG2
zZBjhIc;p^<1*%IyAw}{HqSjv_){KL#ZFC5K9ZgiL2&*>Ern8D%&0lg!67^$gI(%8i
zOsVW=?!o)vz#E^vUcR4HlZIFtk$HZnWo<S7h={dW|5-!Wjv+CHR3P=DqS2$;@An4C
zt-y2z1KtaxIp>y|LAAV8H!P}qO<bxPFWKLy0;S_t3|gF@K~dv!GnwU;#l9KAMktkd
z@<n59MC@b0UIVXgHsjNIEv;Vmq&GBJUb3w&OrelwS3lMstfM*O!v65(5;&hpB{!Ja
z$-V2vDz~k)z~5>~U0@;cj%W%@T>>6t|G$_(iSy#D+_Ujvoaqm%LfZ^+H;OQaO4*bq
zAColJ-UQP_S}!ORgSl*@W&|MN4Jv|hcDV5kWdc9Da9VPye<M7x(A7k6uvH3ae2CS4
zvNz7eC<&@4`uCz~+ilkA9ZOmJL>n~WKgB&Y><yq+SG0KI8vLPm01ce=P4UwQ<#XGk
z%EoPb%lW%Sb#~y7u-fO67}~I7zuX8F0mxA6^%K<}pI76hZ;x9jKJ$_fEKZQeb}c@d
zZnh%-tt`UZlx?l^!!^Cl!hz{&#|DCpgM#zUYEiyAoUI-q_igjg0KC4@4#wvh3S(!`
zz!~0Vl}VCAAP`1CD`yE0>~2bH@2WA)anffY;yi-mB5?Cg_`PFTPT=ucx?}y!>x`J<
zc%=N!WYy))$1KinIgGMw^Oc8PgddN}=>_T{;mzv0;bALLd-vJOioC|7N~`B{yIW_F
z_t-FdmBqeByc2xa%lp>Bu&1{+s=rZ^mAaq7TSyx?(kx?fzjV`dkUF-`plY=L2~Oxg
zx`}1ry@yY7FWaA^`Hcxm+CHy#s?ICmt-9@&;!AC{#%a#Z&LD)H`aBoP`1sD*nbu_%
z;u^NZFDty^>g9ezORjZhSI6qwz|R%ci{}pBx@UL1tsMn+lK9Cw+N21JY2=D^mgZlt
z80Uf6+he;N?EB>}JDXm2J~SCVWVejCunS*jd9dGxUX^Ed)Vbf2gT?SM$SwTpcR^<O
zx9F#z_NRQKU_)<yJu^yGYTmNW*TZSFC$@W0>6yW^B-@|%=I_d`%%w4`&460>G(qHJ
ze}lob*Ps#REq_2PYr~P*hWu<E`6ec?eA0*ZDpKpq9Fxa%<Ng|~<^0_HlGsnZE85VV
zvx+%$VU;hPEUY*b2{g+`@-3HDU6?|OvSaoWIq<#jze!|{#qkle8hk3hsw<6KgWWYV
z<qJW(76H4lwj!g^{@~be1ZwaL#<>2ks>7;!I>()lS9RPU$=Ezzby5cSMkWjCqrQ4O
zQenF_ei9UPHEt!tn4{Hpw$o2vYsAvgWBjHhYDF&GbiCzO?TAC_q|ar#>9o+U+33^9
zhFZ9Faq3FUmd<b884h#KiUZXH%6~nHKL2U}pM53BS6sN<b<upntK*E)=q0r)s}Vh^
zmuUAS(+))1&I}HqTd`L1@;Ay@o;<Ccg8r4gI28+{vSt)B?smmG+%{$5_5-`zMcDn8
z+T`3HV<Hl%7yEZ!O(&5m!f#r0*YvM{R}UX0D&*bog{KdYnH|D<ksy5G>RGynBRIjJ
z?xXzHwau{Fi8Adm8~&r+qCSG+^8}*qW<;cBb@JrS)0`Oq7IRKBL#tNk3o7S|2ynMZ
za^XRO!vk{K0jRjR1&u#fM{R4rrA2YzZg{I)R%cz5I_wCHZErJVkvLPaVj@KBZ0pt$
zb6($yf3lG8>6;0E`EZGw_qTg+XQ(1xm*uC92cL-{^lea%uZ6A@5(4{>dI(fC*!SO4
zl|=D?{|}2lQVWsv31r&dbrN72cj!yRR-->KkS6)m3UAUbJp}u+UOx>CU^0S*nppNj
zUZcg5XI=vQKHYcoQOQmb0!yZkl0BN-U3ItN8Y=+W_qcEa+UC7bPl5HFgZ^3M7~A4)
zH5h;O!!iZoxwxSd#CWDZ4Xq2fYhE!G^K!#9b|z<ZHnI8L&C3KQa;Tmry*|~tcPYd(
zzPyu>)V)wzm!U)gi@cz`aDH|V_NcX68E(Gmbgze|mUf2=uNI*?=3i-s6~YfgYK3R4
zY{^g_K!3A_V3RfZ;g}Fv$+BD+Jfy8y6=R+E$jSX=1y$3>D=FL0{{o}WhVfc)qM4}h
z_0Z$CcLZ{9JFJ75V9!Vg+xcdb%h7Xn4y{&;F4W_RQ(K{0RINS4f!B)kpRt@xwm#1#
ziR#g~H2+KPD5?~nudoq^Tg+F73q0J$bOd3=#KxXoN{wxD+Hc8N=DdFtw;EyQeDZ0k
zH#8&yn!=qwv=RXzEQn&gZC?E&f6`!-awZ)nvPn^;gPR?KCnbi6zTx!%DJ_pVDExYt
zu1RLO;Eb8hgKyk=NG(YDY;CZFvnqbQrVR$GYjfs5kr8%h)R6x%5z33%!=mLtaaAfF
zMbXI#yiYqae}al{^(I>))GmkvHuIAzzPbNr4gO1ll1f1Pi+LrdTcF<FhM1%-W?jPY
zFUCQ{*0K4xhI<ZQFax8MD0Opfz>q4fY(&FeDE`&Tn`$t8eQ-J}F^qDOILZCXw;!2n
zeUiiel$cV$Yyt5GGWuHZ5@}xsM^7wI$Mo=&j|9Gq5j>#~$^pC!%#ken^nAf##kAN1
zJ4!nf*9Aa~Qy1HNxkz|4M{SW(5Ur09y5u2R$K3iY$x<7SCf1%HPWo)?Gx3F$lK%C-
zp+>hF#kel=>HM;XOHGneQU&2h|M1-p43#WbO|}mABM6ue&3+RaZ{tcsf8X-O%oLUj
zfc)#%q932_iDTio7wA=M^<ZTm^jeYL0G6n~IV9|uPk2E+nxmiMp=ml<!)7V&&{3*X
z)>U4CGZ9B}pji;Bk$*7^E@R(UTYQX5vyBn`mPVPuVH{G8s`Nz5o?bbAF{hJ{2$Sj<
z81?KqecnuSVLL`+VaDE}FSlb!rufKF=|9<K9Emaz5h%n$=Q71Lc}cStOjM2mRP0T=
z2_YmNyJ5qqm=v|uQ|YXUaWLeAUMN<O*W&Olef0bEFcp0M#=)vIs_>g?N(%=pUXi<k
z1PFaCq3yL_%1@W(Cj?mO#X1=v8h>T8wYBvR419u3PA*rdeB!f~TF{N&2!%rZ0|KJc
z(;+t%P6u;iC0aac!I~g10<e{~##)2GfTGu0n8k#<+b269T&%plva6X7t;Z{Ow9SA|
zT#nxDx{nehL`Azi*Dpt;f~b(D6tZhfnK1c=tD8!HA=B<R&tGeNJ5cjA|Jec5cQlQ@
z)gqU=ChGP~vxv;1%-LR4DL}2FE?=!5Bc0gAa@!Jv%C1j#DE^5xfYC48v=EV%OP*zq
zype0Daf=v|RfFC{g^_ictxI^0a-vYdOeD+c!ktx7Rp2F8Ma{8!a(<vSWWQ~MtUk3o
zLI48el+($EWITgbKiFf-<#o#$5D-vXTZ{ebm9VNR;bOgGmi_kdWT6U{64~yacJiPR
zt6Nj^n$KFb$~rQz1g$Vmz;`D3WYP*-Y&bs7*E^*7Cn)wt=XPu`=vodfWxtrSak|k~
zTvgW7G+VGF#gb>&6PpO=<a9DAIgU}rp^;EEm;q!<s&=kp>{=L&w)Gy`FS<^)vMGtA
z7}7_G8(~rumr~<9m7MSf6@f^SPMb~6;sRICLWEReK*V-x7#vXly=8N2i^Xh`db(J{
zdB_tS?(m+9s>1!!s`>Ur8o}#&lb|~+-=$pZjTHf#_R8Je<+8LjqgK6QNW;I~Uy(nR
z#|D;Qql_k(Fw`3U)rC-s8i<%Z-yms~L~WiPZ4<j5kvqM-dXY0z7I!yvr>tnYyc$MK
z-4AH<Q%aPYRlmyn0N$Py@Zes1{_d3<%m5iBs=To#65sQKoWZc_KZW<}Uw=SE2P4)y
zEd=C@JcyBrTxU6<Ya4swcf+zKlciN+cWI^6WXak{kz<?WGnnL#r_2SWm}$eB+{xJ~
zJV9oSS~Q=8f4%I<(D+>o_CPp(m6Uu=2Jkp-Oy+;%;ECg=qNeW6JTK`yC<Qm{Z~1hF
zro}&o13xkU)n(WYl-vHvdd-TJd6H3uW3acn{q%|!`Nf0RSM6s#-WaA*w$8B>VMoa3
zY@{Q|t<b|Q{4ic&#u^3AXpJU86xNhMk>aTp{0*hMPO~rC>g@B)syTG9%Tgj>Vt4yT
z@FK-fbWTH=ZFSb;X1BF~R8GgAGu=~xc|}hFD}&1vN8D|EZR1Wgxm!o<^D{C3aF}H2
zc972Ds-o_~_);~M-KvjSWivH<TapKeU7T>G;+FH`Zq~^JbhzBdZHC_Jzdj5C5I&Kd
zbMV1diaj;UUK;8P-~)3!fU_})7Zm)Y0s;b+>hzDfM&Bn23JRQVXVk#`H5EHMI|rBf
z8>=E>#v_!cPqA^b=>BGmna~ja*F*wGhmHH^v8|S!|4_(FB`+TIs`}cos6E<;_ZRBx
zhw=x*hMnTVp#8$(idp;m;jFv=0qnl!UY=Zk{sO(L8yb<>3y^MdF3><`pT;H)xT5$r
zuyyn|!x~?P^4a|SFyfhoyLs@*CIQRA&IduGY=#>7hV{CPCnhtM$l_;RY#tjb(cO@J
z4m}@TbZTPTp*)a7C!DJw#OS=Uh4Ra3rO3|%WE1VpGBU{wFYgD~<w^uP^Gee{Fwtc%
z%%1ab2pO9P0QX<;H;7xyLm1(ghQj5+qliQ1l{1c7(|OkBnD>o%)mpyp1;=e2o7?7D
z&Bf-_te})*^}N~rol}EF#_cd<APnYj{rr$HF){u9{cV;#FBf-k18R58GCs~siARvC
zx$$t+{H)-9B+YNEkeVQb1!wD;2`Jy>sKv&b$;*HY`o=GikK{M%v|_agBsrs(X4He`
z#e-LT5QGtOzx6s32jV$_Nq8=Kq<U8F(A{FINnkpoGW~a=92bmAGHBD*kHTF0hO8^o
z1uAe4MiJ0ippT%4xDyz*u}Bmv>>R8kP7t@JD34D1bn2IT2%LC^5={0>_OHOB4{5s=
z@qq}YQoP0`Y_|s5W3sCLd+yrS({o*2lG=e7FnHqgpG32CRFbNbkkJP^C>X8W$A77c
z(f!mR-@iV;<zuB3YKx8rTOb6zVa^{&(OgRCuOQEo8FaylfEXWLwqy_y7KFgk&+_te
zgi7n>X7y#<$nW1<`}@&oX=z`*ifsu8e#@om^~ZjEii)alZ7sh&oO-<$$|yAx|3_3v
z2r(3&Ua#}}%d49kJ#%yE7)sehPTRMrSmc$Qa=N;2q~qyy2NRheUutO?o}NzOhk~(I
zSlUc8be+Dowk8$1yEB$m_4BuOsc}~bUKkN;APNSt4tQxr1*_$JwPCye^CCdIsa|1Q
zcJgt9%cXFZpIzLX@0oKqKHjcQ`kr^43Eu2)bXaIDpOa(xKLq%iBeOz~ELQw-nObp9
zVgr88g~NiYF$c5v?+5un20(2|d10i?-`eiN853!J>h`JOx91R*DsqyoDV^u4hT+SG
z;5Mc@L{24}5<iR8b)!*WA~j`024CQmP2BT3)Ql+P{L%D18UGU}{7__sttv!|U`%$)
z@igs4dmhK1jXLW-sn3kgG1?<oU+2W>KvE>FGQGh0oi<w={)PCW-d?@b)It45rj8fB
zmsJs^)W_f5MS1laMaHOJme9Yr0{k-IFeB$@RX2x!;;J0MnTbouLq=zAd~QwTF@2p2
zb@rxnl^@eRFNnZfhKq|^20VCadmM}1-rjQBu89UPKir<m^YZbzZbVC{HhXY^IDw0Y
z_mYfkh-2A9q$h%mh}(fWAYjJ4<E%!hP<eHAwJq;!*407H^7ruYoa}7>?Ckd-!t|Vu
z2*?Wxwz44ulK}ZQ%dMo?SkWf8^V)}d_%(PjCveXTmJ4reY}jn{zT6s49T@#3Ld0pK
z6R<5a8Sxm4P*DEZZAu|f$|MS^tE<Z%KB5rzTKxLPdci_%4?6joiqV<6aG|dxMn>GY
za}^7^<rF243Toh>i(*nRT<FDqO<AuwikCd#FZNs&s>W(D4SdO`2R%Xa4tUGA*l*7+
z;q6CKbht&|I%waF&ieRgZF8ONMc4&V+?t|GQmthqWhqonsk_|L#Cp|>G<_T#Hl7&F
z*~D}{vE59op-|>mb=j`m6w_ucJ9<ke-wwvQUlhD0l~J{Rm89d*Mt*Z5X3ULFXbeD{
zYPD_};!;RpqQ~DVtek->#H!>x@Yrlknz32VRU+zK!WKKbx~4ual(yXee2zmI>k7!p
zVS@Yj!^e;LgUqMfBgVZ^Z@_yM+fX`Hs4`Is5s;yra6C-sb7$`e#3*EvQ&QrP2Rn8i
z?(Zb!<VxahMThe2d^jFo{lE=4?e~Hw11xjM)NXR~IG<8U;dTAd+1Y7VLwgwf?w{5#
zlnRx6Z?^cxX1d+~^+0HYiir|aK7XFlep*AsTH}ea&jR%j%v8SRVF^o2+=Z36x%tbz
zMNh48&egdmsk%AYyL;;=+Tpj>s?D2{yly@_L(wayIfG>-bhK=ZfGUMIS;a%=7mvHs
z3`RQLnHuvra~%aw@M9i6@juga6fMS}+EnzY6PXX&5swVMq`C8e@FCml$E7tIa;YGc
zruwHdW>!?K`7oxp(brnhdpnEdDZWU)COn@D5|-hY;YHa~h^xk%(MNGIqApg%G+uwX
zHh=_?70qh6t~dw)u-B9?hj|WTmoPCf4F68%lk$7isw0nK3EX0bi+5Zu1M<qs_>?ih
zYioM#9Uap1f;;DK<iPFmo>h8OR4a}=FQRv7C^|Ym-D_rW{*{vr7?*-D$udEKfUv*V
zknlk|UK3~g_WG3fa=||5q4}^G{yi)#8?4`t>GN;Xc@6OKrP6CuAFcRdDc^9Kza81w
zFv!fzL@<H=`2e1tjjb&`mLw^C%~~6ubg!FW<zjUuZi%m~1-BvZsHyYXKdvB%iHY%o
zrxL%1EqTCY>`~=`rA8s@IG9GPFD{kv4+=sMG<H_Sk1rg+QJ%jB8H$kV`PiNr1J24V
z)d+V$zkzBDnq_JnQ4fqe=XJdLlb}B<q6~>`ppgGsk=|Bncr+x}G8xR4;64(5q=esI
z81lt^XG6H{o{nKZl(JGK)I2a{o=i;FsrgT$dYRVLg??5b+(Xxklxo_4eKGsi*CD;H
z!xrOAXXp&VL?`L1^l@}pZ1--j1gVKgA><9fi!Qf?xS~E1GX@a|Ravk9>fQ1vzr(bS
zUYX}IzSZkvxH?N2=cuos{ZyvHU)IM)O`*9z;J=|KsHS+(9GTuaSMFO)0v>-l7K2`F
zO4*;C+vxz<`eM8_m^7Heg@XWrKw$NI43EDz7YCu~>FHgz@7}&G`MYawXeb6WI2~|0
z>|h_v*KC7mQ3Pp8ySj3Df>e61!DBi12|-g!i^~~C%D|w)_{U6NUj!ud$?TSTqZvXL
zvt>RtOKzcKSrQO7D{$xJ)Ck<2eWR35^O(9L4oBhO;J7*4{ig0;9QL#KG~M!za*3uS
zh)^ZLFL7~2zkI=t<URSE1K8uTWp!(OdJJE}-wV6;IZGCH{;D?#E-o<i6xLELgiq?(
zNTe;_@{^J8`pIQ|n`0Xc55UroCC^1)TXkW$fXSV?yFMK$0czy%oLAE$qczT-I5wRS
zl9lQROf}-y)6Vx;wA_2$7Z3Hqcxy~ibhF$uluUKxQ3ivR=FB|+4`nr(0|Rm}-|@o3
z^V3L?D?EY$rv0mYoMq!q!u9qtfAVnrDGNq_5FsQrh{$*)^t6nH=tt<WIo_bisrCkF
z?Z&Nv&xV>T8{vnCi3DLX>$FZ4TmSg+L$%iCjebu=*Ya|D8lU?s9Gq{ujXe}V;~?_9
z12@?+@Zh=I!eIf-+DzTKHKhfh6FfN@lc3Q0_UVoa%&aC`H$NA2a5z)y8V64vKBM*v
zL2%s@Kba~7xT=WDZW0j^eizS@`NC<lA{f3!Z6+6Id<0u+YM?xQL~j3DOGB;3r)El1
z#unORt*}EE$1uO&tbR*w-y3rITLbRKvQy<=bI_B71Im7W3Fu(0?fP<oSL$w!J6?qv
z*Y9fLqx_{O40vFPr2tm`R=dW1b;dW9zv?z8wuVRU>$UE4gu{nClIk+AT=SuzU&85D
zhdo-2*gWL-cT3+P1{rj!oK~ebApNTgFzRuJ_urzTgTgDw$j=+|maB+{y{2w$z&Wm>
zb?xmyyK0LUpR21YDYwISKUAzg_S<so)=Qe|Q6LsmVUVyJ9L!ZMfUMH6o0!8&GXP${
zcC^}&V~HvFkp{B+4L-;^a@3B_s5w=0(SJa@{Dp%!Jm*rY-r>2s`zD*99{Akh5i0q#
zu!2DN;*4V2^z`)Y*@Qgr*{H}P9g^q22K*(+UHyksd2on{LmC>~!?)-Qa{G~&Fv++l
z4aonYY7G^r3I=&?CY75nVJ%#H2-Lm3#a`M~qUOc#Dvi;F-L>b_GD=u*Nk$3c6<ffD
zY=wJ$#i;yB!PfkrO=?Qs74ezU$p74$ftJMIRbyji&a}iA@%*lFhKH+wC|DNsk&dR*
zLSYGe9;$TeM*^ONxV4oP!BXv}ovA{Chr2_7Ie7?#q$8+t`Y}i}5SUz!bjHWWZ5*5K
zIgI{-pa<gz(el@?U*JVH7yHmc7<~KYt6S4j=0%Q5Z0v25i80tDwR-?Y2UXGz1-3Y<
z)<kA~lZjl63czScZ?!$ZKPV_+X&VL`-sp=V67b?(T3VWt*HjF4xFZmrR0F#L67y^n
za7JQc9G{~~sPKCH9V>H{a8PDDqSRbq0R{c>wp<i8$z!$SG`mx+`}IVb7}frMs+E(Q
zl0u3`Z&$DVMvhSyBVF54$MZ;R3SJQpbB8LotR+D}cmvt)$N0y={+{V6!~_JaTXhtb
zg4YgI{UI%#V^i?Y0_ncZc~8NA5UywckgdV`MP+RvUvZXm?rS0>gRWCCHHbrAjdY+p
z{{;C%MV}04PqVBm^vREO&9$LWQK;N%a0nWkD}`rUrGfYGRiyT7uzEG7?R!%Y4hYG%
z=s>|jrjEKfJy@UL&8y$#M{DPb$V#%kCm@~^snz_Fx1g?+7fJbo{X`QSQEDpzJ}Tyu
z#cUKvMTh|a$dyeF95XREUhCpLpZsQMr7yX#cfBn$^Y?0-Sdop-oxqG?g!U)9Wj~0f
zt&Ogy-y<WbXlOzl_okVFdk^r{-Iajp#MN;Rg)1mS%^h0WSwM3kmEY5CSL_Vda5z23
z4))Q{fJHR%!$3M;a(q;Bs@KpRyA!6vK36PQxGyevm&@@+#5N0jpvq}araTmETpItO
zS6T<jni?w*fxrL@8v42O=HuteZP!f?BTJqVVzyOSqz??}bWMJWFNMSaQNeN!a47Q6
zyx^aNvU&xSxwBr(=dv7AxR7w><}s7Za`R8!OYQ`Ch+lD|mUuKkzcf?)FYImJr9I~M
zi~Up{=l+sC))8KgK)+Lt7@M6Z>fsj=Gh~e^&pnHHw7M@vbh7}Ao_&9Zi=n&Z*@NHI
zj5H1Fj_B^R>R8Wy&(LB4`iCFL3kx!){y4)ZF&P)c)!I^7St+#3@amFMj8c(^?hj5?
zxuDYp8t(E`QjEAFW<)OPI|9v}3W7hgi>jsU(4Y^e0x9!0{+WZmJlZYMw)6UHE8;2y
zH45Bh%=7a0S2LFHQ^QJoGaKWi?Vabs|4R0hdon2{PA&YF32X`iaJW=hUrQj%RjK}_
zVM}=jCjsQ?kReEaD~)J=$%J}(Q=`9r=KJ~?2QPfT0uLwjV7aBGAHiBr@2!`Ym(oSn
zlv$n2@!M9in2@@<#m4AH;}LSV^S?PWs@6kJ;o;$IH!&yO#M>=>n|43h%*88=h6s6G
zn84uA2mYxC;v?|*zGvVWp3|qD#L7Z-0ln+fkjle36!UX632c9Oq}*@)yC2r7+=rfu
z{dNYFcDzsED?4Pq0paNVKX^|RvLwTyT}wmK(3*4kK>lk=`Ih?+c6BbVSH0|AX4JR5
zZq|lg1L)?@DiNvdsV4-@xPHc5r*KyWd<&``t-E9bwBaMdMV5tWycuuewg@WjAMjt*
zoQ9pDl6-sS{1gkBdLt^!A6;yt%eZ2@r-ifHfJ(%eU%Ie&l-p9`_t+o%X?4tdYg+Db
zd>O;hHX*e4(4WFS#IK~NLupnfLp()``97CQitIA$9w9-TSy}nx#N@psO_$CkKkUo=
zYT~Qa0KDsD&~8&1MPt6S=iF^ySMMp_$dSoiCZJ5s)<7;_gW%D&s!6!|>)|!l_LG(6
zcX`S~^2G}y@$wI%Y6=X7n5a-;Y4iLr!E%(6xJ-UE!NKZpofbS|v-*ofx2xY0)d{gz
ziX;MdU5mWz`Gb3M9&A>1C{wd2(RS5Gl-sX`fj<nt%MU{1NUtqX-+_~kCot*F_UB7A
z%s;lezb8PMqDQ)7pBedj3X4n6+FI1GcRNSS*OEu_t(d7<pOIbI!F?)(?N2gk=GbK?
z*_X<(XMMFEDe(RK_q>zyyu7@SP0;LJy!csD<J@%unrUa4DQ>QdDT8N06kaR&ztdRE
z0Ls$xZZ#b6Cs!t^{cLvvQMJa3EQU(HB*F5z63)w)LfFK}#mXI#6#N5evMuj8lt!IK
zH7d!N$O48^lL87(-s(9GzcHJTxxdlJS7At7J+Pk9*EL=7X9K%{w`v&GSlk28q_fd^
z58fdQKgbsQEketvgWomGL_2W(#M-t1x~4iVVVJcju6=5^_xsKYM|r_dy5WNUti80?
zPZAVRLSsHcle$iV?Ok??dv<R;9(d-Jgtc<@x(e-sm)$XoHkkTR0~@4qB<?h#iIsCQ
z)l&OQ4yU8ZV+OygG3w~4OU|RVcJUdj-JbwCMP`iXyQJqfKPbiFHJu7q^v@T8W>P1Q
zrmdTD_wu=WKa@awlQw+S1eC~H>9*ebe&zOIx(y>c^|KO>>1{M_G7Wx6TeR4Gjwj*V
z({2|Z`wuyUvFfuwo=gAWYPvxnvhCwN-%D2Io}KgzJE!KO3>HX(`(zG!>;`t(%l-=0
z-~7PHkkH!x9xY;{%w?a0m*rFq5Ds)Mu6~$C>fkE3z$%gqKrD$%hyTou@3<!&q{CA}
zSw(WI>Y8kSaSpR^=_Xk?qvpz<mRx_Tj%5K6(HFvgD&%70aRD{P8(1~_(cJX^wFIlH
zO4abq1V|!_p;S(l^^_SKW8;x<*Wn#}`Ce<Gz1{a~mUy{d7upsul-vVi@eb&Mr}h+l
zp(f<AdrzxUvT=31etUa5B9qAUYxqvFOLd*s_2fM=GP0<ss3MVfGh1R>e0=U@BJv$u
z2*P3bz7}gmcM|>Eg>*S6BZ6R1gu<!9Al_DD2;C@3#le6Q>0BkUxrFsws^Si4&0ipF
z2$#K*><oX1I}9*@?@N5=uJ@DY4_!H?fUxH|<tyZf7XN@<(cq$OoS?FfPK^%VJNj2u
z5O}vgy=Tn0dBO3G=*-@S0h5U}Emq&QQJ@KRJz=DEp}=<1R2e^^d_M=OU{h1^u956D
zGNf?L6p{5wM&UaT`(R5*QRnbtG&|PJQE$9lPoOgp;|nw_#UI(y9+l+cdWV1H-9u_h
zSrsT3eXs|g-@}8_{E+R%<AyaL+i!5~&jr3~0;|j^YN(mnPxdPROO5twnJXB}Halro
zU(NHTWtZ%JrAFbqqeEz*G<>)teMYCqInI(c2Wd+yz>c1*M?g{E^t(ulRg7;#b?&%o
zM?t=lC1CXy*}H+KO`<$IrK!Jv#>;5Gor6TSwS}|L+mLZM-nRj(uARhdytZ(B0gc}j
znPhxyxT*c7hCp`@_8l@)K*eOb#D<-Igz!VIvjcoqa8X3kG`9{6@5|l}_b(vsgEHNI
z@y@Nod(<uK!1R)KRQf&b1Xm>*P1-EA1VIn;XdspD5ne9;fE1L~ZsRkzGfYNM5FsKW
zf*KbMtpDUrjmXS=3-Z{jovg?!_+bm}Gzq(fYSmp}VBjB<vA5)|8!wBBi^ohpiAqSE
zF4&LQu5~^~c>C_%@p%)g>jvls>Ve)L0*El7phFb{`X{4BYOJ_uyAN-O!vjJ7EzF?Z
z*b9C*GaK8Z{I5A@1+%4~RD0gq8H@w1v8Fg|x+F(LL`+Od`ZcjjO-Hw}53RVp*(<#T
zBb}<o41aka;W`~Pbq1*TK^A$s8bmRkBTbN!nhH)XiA_krAt4!%W^8hpx5a}sqohZ_
z;ozvzs}n;%MpnWm7fk@LL?4g*ApHRmr=>(y+oA@vp51u#Oly`C{;KKvDOJK@;dV6I
zIlaJ#AdJIXO60!+CTj4W1WTFKEM;sx2{##*`4&y37NlZ7`t;J*r+Oi0gx5jzg;=m9
zns%ZkOGK8dy#q1a+-;?KUb97Aa3??299&J#`0LkDV`ha60g!!CKS~fBHmV7)A^rh_
zg2@!|$Gm&1(8mkhdX?&*BkreGMDP2b{Y@Sr{20ftn)1ROR>Y|Nf$XTEGQtuE)vcMO
zc5cg*;K%I!6b%!r&j`}vhxhlrXg0y{Qr6@m?YK{W$iE}pg9bW#Wxb{+xT*b&c=gw}
zp@w+wsWxya0>bFc&~+?9obsYFrG9HeR&;XTYJzCKqKM_LH?qp=XNCy=5(m%wUuMb9
zfNPW@{6-LyPU^>WX)zL48Pm(Ux{r6uM23@+M95$``o+u*rY<Ia%l}p>)pl`n(=#;<
z4i0|KlJo+EnEn0zoyq(O%eKYsH1~*%i~?zOKsJq+C`X|Wj6r4P<wBG(x@KllpFcnA
ziKIZyR<A%OU?_6%1T6%7dUeHP_@f=9qw^XcAHUb&Pslwf8QJb+J`StV;Oh?`HY2=l
z@9wTI=B;VHAoUI(g@l9>lap2U)0>+3_}tD`W*|WE-Y?cUPDIdWZUg;HTSv!Y(C7r;
z49u6>fj%dKg{7tWiz4*%jtm5A4-Z}fM(xIf?P;|eJ$?N{m%DCx3ky2<<zb7ZQxovJ
z3EkJ`VZzkW;=zm!L96G^Un?fwQGL1n|9JY!u&TPQT|z)wMO0cry1PL|>5%S{?(SHW
zAkxC7JEa?>OH#T!q`SM$#OM9ab;%E@&6<0yImWnaj>EpiM8lU{cQ$>7(S!&HFYQg;
zwz-b3(!AP)C)!q#m{368uY{xJri|VvUl$jKZZXQR${4b2UB^7B;cgv+dL&pBa)QW$
z58f%}=@0BQ<(X&8=1ra6@@&RbBWwx0j|xD@7L63yxiAb-s5^(ZalF+%lDBOhlH_r8
z2!WkC`nOo7j+7c6F^+LKm~FO~6emY|Uha_6i6JXucGjvOmNT^c(3hm|b^bD+i7!D3
z?OeE4vK>wy3k>Fk8oZ!emFp<h=xDi2pzfwFn7KM#+$!bz>-b4*RVtgU_?jDk1=&M`
zs+)Rc@V-iMX0+Ub)!+S!&Di1KJz;@pM9&+AoOhTh+J3OySatP--vg1RiLZxR`@d7J
zHGGfD;-;%Z)^3+Dkk88)7qbuyMI_>5fsR#Y)vpzrLLa(ZJ}<-G*cJfR<y)}29O*X#
zG@3@cy0UT=Yy}q=3b$6fpwn+V+nuqow;y#YfRoRb)SW6bI6YISOIrrp0_E{qUt3R4
z-n65orDY<ILtcmDhg+r%mlF_G^8E-6mRk@l=Ah+qswm`7_PCXw>hi<@8cxTtIYMnK
zUN&yHvOeAyA8~9G<*N(@qrzk2>8Y-a=X?G~t#bWEf*9kr&aLuI(uBQ>HrVz-k=&*3
zjMxpQzid!cMJS_pBM>ae_?xu`Alpvo7D2p3Rl9F@R$;BPSN*-Szcgj#!pkxc{l;0;
zj~8lr-lt;5F;i=B<16n%89V2Z4R##j=rXQj0(Dml9WHHEl5!b1ve8kxmWmE%_JgTd
zd;;nnwVU@s4R7(P3|pC;sAKm{-XJlu^Zb%SK(W>Jz<-`sYE!8<ny7s6xTvww-ttSf
z0&=U=o3X30fNCTRTcH}}q@NA}n@1X_Q$J64C->_zzn0F+ZNBgbhR(hDrwFxc6TXQF
zg?H!gQQCjx1lvBr{4Hjm&&SEAH%S@&p^HV~{Sj)EW0!L&h)~x1KxAHQ%H-H{nj(7B
z^UJZfg8@dKDgB^DqneFC5;<Coppe@&5~SVq562RvkEBpgPeIOIjiu>z2A(NY;$g8S
zy*K}APAsS_5ZL-VOZ7S*Fu<1#8n55H@dgZNZ(rZx`8nm=w|*_YD69ZiBqTJIehxPD
z`oj(a(1I|5@fFY*BErJRy1Kf2mn)I=C&Lnx(@NUU&@-Q5nV6XDOqTkBQ6>^VIzPh0
zy*{KhkBz|=7j;NUNjK)|IKfXTo*n<jk%0~l+OW$_$7mi{5%6N&eSPgiLlGq<On|Eb
z)EC*7LUN1Ba@3#I`kSLae8;`44-n<*pO^=3{^<DFWV`?mz&*Mr8$%fQY@$(G7VkpR
z7mdLt^#Dn~ew~^YK>D~<({b76z^iCEy!!<D`8su*i@LdPIr#CxJTR4<Cr6s5I@(0A
zXr1D2q6I|;&e+tpAlBRP77sKMY^Fn$RoR0>@O{bV47*|3r(V323VBrdI$e(X?*x^<
zQIlGBZlaJj=PR0JO?R<mirk;Bnq%$mEEyAnqLSI^w<SUK2os}oY{3_FKT)O>_wI^U
z5q~lX2)}M6gzfy%hx|e}gcow#N*(eq1!-|7KKPY;;?&GN*+d{137oU3eUDg=4~v&@
z8xXr5&W|Qnr&%uY;;4#Bq<?)Wi6F3hn{PqPQcr7HSU-8zF8&OwU9{3+T`Tv5uzQ7~
z!Y%!!$2_5brCs0s(|VYZD65vnIWezM;c{<0;zqQVZ`MSn88keu(6cx`+xyybp8s2b
zVQ{9H4v`rbI_ODS`Fp3Q<H7wgQe$TXpd<a!#<tehSY9UvAm<pX!?v@tdw7ph-}_?V
z2Nf35FL#C+-CP`2YIRObfW&ae%47_p9B>?FhDey<iSg~Ta=$Bs&tB(X#&OaRee>px
zapmx`KV;OOm}~okG)~eV2E*;|@2{+-1#j3b^{zW9+@7}w##Sd;N%5$*eR_-DGkoFp
z8uW6`0y1nQHd$b-3(ZrY=0})=>j@I?mg<QrW0}K9JSa@!-|42H=P4bAD;^4+J5?(L
zwHaf3ofym{Zw*vXRJK!eJ%nPPFG}$_J~f4yFH6nPWjb#5F^fKtKPg{*8-x8(K0I2&
z#1nPA2#d>^7w5@Pbn5#huIn)Y9L~zWYJ;qi*Z$)isZ1sZ1EcjTbUi^muP2pgNOZq3
z@p7rZQ?vgP0<Q&ko|+weZGk41!+=wM`W6+1@iX|Hs=CzUmRMX@zlPJD)P0a0noDz`
z{Q0&ro67ZwxPigt)%T$BO}uC@>XzzF=gN74?nbaPOJp)%$nK3ey-YN&vA_H9;6C(=
zjJ^_VHwhXH#g7mC{rp}HH@tp~?bRx2@|Vg?q^&HXw~&xYQAIvB!YWu76*KkkYz_XK
z*V|vPHdmh;tPwnSX7(F8Umq8^Z-5F<_^8VVhnhvwi^3|Yj$52b3_7$^-nF(rUKbtE
zTgUwyp!|&_VuXa%9pl|lNcROs!=9x6qQ0<;sjBZhRc@@t;8=6Q+tdZ_AKeEg``S;R
zD4C5?{eh{dsgmZ~F)wa$v)g5K{%$#mRq;Un_3v-@d3}>GB*CW9;5s&`38DE|lSJZ~
zomnMGUp?K^<YYk7OYHdGIfJ?JA)nRQ)>V~%M+UacJbm58H8)F3BLD@v_Th1EC>x!Y
z%J|#cEPG_`+vHA?=wWmmFzitzDFyUq-W5~&4n+L-T7!$h#EYi~2TYDJKCfpx%{6w!
zDC7PnzNIri9jH)<UeiibieB6q`~58gXOBXCVtW;-tOs+SPQs;OAl@u`{<D^X_=zCp
zw;7xG(c?h5=oCr+kRT%Bmd1+j805yE1H>u0^Mj?@@2Ix<Zg@2WDR4dRf(jG($+^FV
z=Oa)emPN+(Q$c0YQ+4DupAzq{U)k^ugcC<f+DN~lu*EUg*geEDlg{t@jBs7usUTao
zJv!bOyPvVW{_{m;Fu?~yceM5UMy>QEQ;Mw1c2LoQ-o;P5`W^W09Mzm?WRx#K<!D^z
z=gbUyd_o=V&%c+WhbP2|um4OE^uBF1KK=a4wp^H55eoQkDL$kBO-OkC7_%p%kTK>g
zg_7NnvcJ)3{6%456gqF_-L!&WTm7K3%Oh<;ryMjLuY^0Itf=?G+zfql%#}a6zh#bW
zah+XDpWQi;fs5hPE{VJgq!ay<^_o5g6zi$GQ+!ei%(OOn57XWP`AHQZOrntb3)8}q
zf-sza-`K-{K0VzX>WDds)d#X!56oD)9R<(InVJPQ)RXsyq(U*D5iV&;IA3#ls_{5V
z@m8}8Z=5XS?MRGH3k^H-;6LdKW<bb#5q4Ht_kxXv9R-?d^}bAyeGwxDMZupL8jtcE
z6&plC^a;lhN!$ELV57YM$XP@`I7E^N^3af-RFfl?f0TvawsoM@)HpZ8Hr5p8`Z+ea
z^Q>6O#<6WEA+9G|E)c_@^hfW-bCzX!u3BeGDc%Yl(2ZMpYK?o>O13wq+^rq7TKp6*
zM(Hh1A*)mq(XxF3P!oVXJ`y2OCcWQh_CfY1uo8J_!vQGdN`_GH^P25vG~^*Pbw5J!
z7n?p?>YlnwpThMN2U_IJZR+~0hb8kX=*F$8<+noUnROAHP(xwz{YQ&Gk21~jDONva
z6(n5={&h@_Kp^taNKHZaoAmyBRjE?+l@+OkKJo|u)XniB3)u9TcDC0|d)F+cIHTIo
zt6R?KoO>jiZo8G+BKTI;vLUv88>8K8f|&=5#Jq1*usI*GN?XtoyX(5^_voW_L#eLX
zL}cDF9lvCmHwM&Mi)oU=ojIhzyj?t|;5YWA+@RK#E#Z+cP3*Itu3i6JyL`CQRJlzU
z0y~$prW<@`Gq+1Z3ZMUcL^agh?LL|K{%(kHT>6?sIrKL;p)m&vPf)(q;$(*5JbBuC
z8PciaR5e;HAlI|#{&c**%K7WX=7b2pD341ZZgGDWSN^-&7N%aJT3(-G*hyTbP;DdD
zdh0aXvoP(I30YRWqs<XeF_pG=!AMu?{udJCcEgpF(lN%_JZ2bNTjQpYlUO&FN3Nk{
zRXv^N<Kdr+xr6qDJ3bbzJG_0^I739Q&vu6b%RE~>dN$zaZI(OIl=8PLLM18M=I4n4
zraxU}Apul4HrB?8VFRc>IvVq7p3YSk!aY;v8Ui;ccbc7JHDM<~*P`F2za`a?dEZWt
zwNZVod#|e}w(-p}QLPjkRMwc&F48{PmUuQv$yGlnELUJKTQ>LdIP2c_RE5x4t=WzK
zb#v_x@3?oAD4Rx^l<xQ<0Jgm0*2U6$*%%mTGKTkcNk-UO{}9WbR(?^J(|V9h=p;;^
zYGlw-(yF(RW&pu*D2E<PtA8b_HyVL$_mm2!gpuIycavuuAy%Ey4%+v`W7mBms;5^2
z!|W5#0<QVVh3Sd<v!4!NwLw=~_l)?oy=CU>iw2=zzj)EcbJ|4teZ~W`O2~1(h9IT2
z)hW#=sA_cl$Y~-c+lxQBbICz+R<^|`SQ%^+yW-Y_OVM5#?YX6r=yBPpdzs(2vdhLG
zhjwt6Oov2HfAW~)F8&b@5mD!91FeIF8QCc`s~F99OuxkfI~i^LsJ+QGC+^S2Jv&Ij
zG>JK$nWh@vbyfWWVU(szJzBffO2&hpUutj~dEOF!59S^{8T*0JXmj0|*d1?RsGV0f
zn#yJ6IOkJ{)_)I~a4&chK{uEomy5Bt4v3U>K(Rh&_dr_mq>OyZeajrpcMH)6+&SYK
zEm#xs(GJKJNSb)!*R2bpih9RS^Zj6C(`wL${OkxqQ`I7A42B#X3T}I7=-mlXvhaEM
zhypWYPT7;HJxb-!_Bg72ihWIEm}iIa?`B?Pf3jP(#+YeRle%{jR-BrhDn{KEIbHc(
z#(%K+=_38K&%9=OG6fO%fS_u*L9^j-nuP8?$xU7dF0*Vpuj0R1%(bJ9mNXMUiS`JN
zG>3oO7(vHL2n-us7X=iI)aZsCZIp$XbfeSL(WvSyp!E*ExNE;6FedsaP4NDv(Qe}f
z#d`vR$7O}LJ>>YC9+_i9daXKiMCN%GcN4*%<g7VIu>?>HG!NK&A*{8Eubcxmcl&zI
zkFJMzzrrhWoSW>+EDG&@n{q^J3=KF}8@E7c$J{L8Bmk&eZ7}o`i*`I^_~uKrOd=VQ
zQ_7itPQBSYB|z#YIV-*Hf0rIMwEwC+$!pZ`mF|qEZW3*o>&H5HFTs@G&b_q)SMIBO
z<RV=>9cHAi%@e!{X~t+iz~tcH9%+W4)mVd@Q?1j!G?0l{Uf0#sa07-nL$l6NMP)X5
z>M4*-1C4>9AtxehzEWZ9#6(PN>?=KN*(ZL&9mQs^`6;l36_vWytjN0S-G3=<<7Rdq
z4ooyqcQ*(h4EFntE!ilh*_2uQZO+gRjUkq}+acY7dh+JJe>HVxe36cR`1p+6cKVxX
zg3)iy<&7Vjavjs1g<DEkt$SBI<F=tURhq8#P58G9#A1=}tiI8BVSmZ$Of}Q%JoO@b
z*ksKTFX)^_DO-uR84Q?Zg8cKdDrclMm?gvfN!c}1_GR9EwB8-B3*#%-k7r98Ev^5a
z@Tjs3qDC~-L<Ttpy6p1`ztm^dT$mS=hEm5{>q@QpiXCYUsk6-9ls>;Fw%AaU8pM6Q
z^ciC)LB=ZqVctr$x>rrvZqq}bMsFYt5{txF((B4uwmz`+_uytaPFLW`qTtbyqPaOo
zqsX=;zPI@TQ$28+xl7@Y4UeWZ_%lxN%Pqej3O)EB&qjl(E}-sNZfQmS#VPTP=RWsA
zSvZT68P04&%Lu|9nbcO*@l0jZYRsr^_<t>C>zw`x+<uW}-DIgA3Il^2=7T;Phlof@
zRTU3V>MMXLj*5zULO>t^i~#V~*4B#v(*b~=-QCUQ_r((NtOpGc6-+o5;#LK3Ug1DO
zn=xk@O4g-``Bm9VkvsRu;hP>rBS2zFTV2F931CqQ(nouJ(;>dR!8^Rz@E|q*jDlRx
zi?6^#-F&(EX*1%m=V!%%408Q*q?{ugi>2<36s`8&va6jj55Ze{q)#2rVlBC~w#v=M
zjP`0o*K2p07U1|*B$*rN1|Re=S?^{fWJ}0}@ZAiGqFX6#4h;BMLPFn@@QLZ%t3?q<
zbUR-xmx8ViwiP1HLkq*ADj-cw%<<7KYF#-F&&u4*(A?m>Esd|XBPlz#HSG17$g(9s
zv0GZDqHZoKBDHqReKjBUb)V5!SOic{8-VbdBeKWj3E37phhk9Pm#U#1_1Dr#ohiq$
z&=*L<&aS^O&+p#2>|Do~7Sz?AHxXohU3aIUATPjy%w7#YQq29))?}Gql?#icm$^IQ
zaoh`)qn}5W1y3j&ERl@piTWceZ7&z)h=h2io9%Uz={Ko+dIex{+X*$x<nzM#TB<Eu
z<_P@je8b40z)nWnRbvfUA5H&|1S|^r8S+|vV_#hLJ&2!L$nAXhzlNpmeKxD@SCbH7
zp%~z~1Tuc`Od2dSdUUPAptMob%8$arWnk$3@ZkfHLO#mQ&IS=157iQ&a&Ttj1-cgp
zOC5yS^FZ7}1VrWccl-CR*z>kJZ6Sg>2TFX`N0|Fjw+L|d8&vev?HanFxf;E^ld1MR
zp$t>qp~VxaH1a-@{_4j?T$g@2y367&Vhk?1FGa&<e3ddfJ~HIY1R-u5uVr)vGr<S{
zBPe`y(Y~hl3@|M1eVf5;2iR&P1qp(&A-UA!6@3b%Ou0XqYRMDj^fP?WY{Z+CP$6AC
zRg0!ATU>3Y#9;A!9~jX|{yOTgBMEP4>#&cBh@lu>>KGLxok1IKlNeqYT(qsR5z5|+
z;a9c_)@dcA$KxGip(2K%#dgAl()oG=JD58u)u@GPWc=eq=ID*la@n>IyeH^sHtYe^
zvT?hXVei&>j|fG+h1N}Z&_$q6HfN`1{b~RlNDKUc*0D;p<o@Pu47?60XWBw=guP3Z
zHezV4tL!#b>1Cur6s|^-*l?@bl?_v553@$yoq(LldkV6fKrIaS{XkPn0YsB8l~#4q
zNFwqoFN(CopJk>V|B~n8a9FL}O&N2>d|`xJ;<$ytKvMtu7Ha8Ga3x`v6`jhR66aE;
znp=S*^w4-9)}5=5C+xhc<^TW0He#1EP$bwpZF27mU&J)#Wn&>3K!UTuRmA8k;CvJs
zb6jL_thbLRN3qw+bUQFKCg*&Ju~+EfyjhHWLbC6XT~Y8nN$JWe3I<a|3==js&dMT>
zjfdiN4Qa%`&J75lIj2oeNQjG5QAZ>@mQ+@LN;4*u34wFJT#X^*cZ~u<7S5B;kKX^J
z6b>^i)M|(ahyvmPkjg3(0C6F+?tHz=BV^=fxVY^p&|Np!mY7*t_4a0Kz>SGeVm?(S
zjI)UQD2=#V-`<tyN9ubbt^=S9Sg?l@gC$V9f7u^Y8NN7EdUUoe85Zbkb2Mfr?2tBO
zHbv<-;ppB!wBTrMxcj~gXLzY+H|%i&P47V%M)vg|UsIxH&|@Mgn$#$S$t^50h{8{Q
z=IxPLiFDmF4Vh7%R6DxZL4BL_ZN2%XPeb5I$d%3>(XK{(9S%=AK~i#P0W%I%!@`1t
z2k0rklFr&vXr6+>oLg)&4uMPFTAkPIB`*L1Qak_nmROKda3EP9S+tI}E!YCt!p{9A
z3D7rW@pRpP$zy73G9n<pwdH%^7{XhE!%{#?f9+>$-SVQ}lIz1=yZJ8_cXa|gS|oa8
zKY$~&$Z<!g{c?S))Z??i1O0>+pJX-{q~O+|#;s(nh#I`g)7|~jIr%9vaKS0L?S7^Q
zMpkNkKIak&px{AE1)1rvJ=NvDn7TqzuDKd4_72*S_NShbub1B0XE9RFPD;N-9FUcl
zuvQLtd=vO%mb8E{zxdBT8Viir9>Vt>*PsbgphH1M9-kC=y$G!>+XBHT3J}8~rOM|>
z<3bBfP}5Tps>dJnjFwFYyv~zT?|C!C!lWn|xr2shNdaY6dTjkWBg85s3=91n1zZ*=
z``G|RSkdTyeJzgvDGn)Cou<U(OPG;$IS<Nvbh%;_-}?N4lR{h)39bmnvEhPUnk9I?
z!{fIKS*uWHVx73Gt3Y0vZGq;F^_d;#v=lmu)yoIu+*c4^V*2vq(TFY8AKH%L7GxAF
zqxeTxe4it8$3DfZ6Q}>N!<LPT7?=tX>8kzVU#sxdHw^tw)7N;q^pg;Fsq_Q82U<Fh
z91JobZkWxK1e1~Mq>9eW@23Mg3;3*#_w)-;>^_+lx+qoJYxe36I7T=YJopA5)D2uo
z7v0<wfu3HG3^~1J(=anH%Clr?|FHaI1$V8BSI#>cy)CVoJp(BTDGo1Sjap_bnWfn<
z08P<8%$5IUneZXR`Z}lm#;F>UzPW|?CAaNbpm4+ev$?q`j0Z;j^A~Y}80XNqz_02v
zlZ~{;Azt#VoQd&x#5V=b3UeKg?;mOH3#D`i0HXWC(cI+Xl_}VbT39pjlHc@gd8nTg
z5OBKj@Z*TN)F{-AkM=&dE3Y+D<iuyeE28HE71+dD*#K#x*oA;Bs;ow!TUYhH4JPJ(
zW=ZAmVzWL}q!2OS-{WX4h02*f#VLj=Wi%{z7^&*=yN~033mpHiA;c#Gl9#?-%W)za
zDyruB#w*pPpdc&HxF!cc1Wrxi8Cio%ZABdD*>pWqZR|PdKgtv+NVLG*mw2+JYUB7O
z<}8_vk~#A44#IPe=*b%1;Q|TTSvBYJT;%f85VQGj`($S8P_4(ph)KoUq*}?NO$jhJ
zrYV|bV$8otX4@zBt5ICe^>C&&%TW(a3WDdvuz#0~hVGUxo?@EYRnf=#m)rJPio^ew
zI|3}X$Mu=PY7Y#qK)pKDDA^@@`iy|nvfg5jKwQGAKX$&>!Hg_S?m-psnfvOOIQpT(
zzze9pE$i>LNA0WmnAf9`Q9o&iX;UZvPbjc+Wl0o+^-;pXj&^CZHzXN{gkoigp)$T=
zc&X!h<A*n1Qf-3AXdWM8DjoHDSNblYYDLnI*>v%lu0AP%<p<6e3;KA_UpJ@|TDog#
z(7;36TLj9Kxz9&=2peecZ*0s>Ga<Xov6j_m`nHk6npYnl+VDq9?+rn3=ZA_zgHX@d
zctHxBJgHI%9I&d8prB=-GAz8fg(++0sx$(RLA&k8kA9^A)ZmWeO9dN*&aSz{;Uy_G
zZQuI=yy9}$6M4VF1Vl@EGce)5o|RGyiMC)>wqB_w8YW5X%|xwMgQmTFsh^uIs^JlK
zlda1kZ!f5I9cI<xv9H7MDd)|`YjAB@<=HxqQZ>-g__x!w8T?Od`KH<0r?b{D=FMbL
z3!DuYfPhDb)R&3(u~RsOS>&JBBB<+VLxg_-lI@!~;~dp;k0(?o%YRCs&Hs1<xA#!c
zvCW%sV?PfPLCM62>foPR(0bIkoWItmV6hjmeP4(u&u<0}X?WlGJv&M5i=~AQzxKcR
z<WZW|EB@x092pX1EW)m}z69A^1z@jXwVdVZ2qlsj$(90x*jsY)Qu|H07YrIIrCR#>
zZ*Q+o^Dl0JBrs2f7>MWpZ>jWcai;lZBQk;@r`8l|rWWO<&^t+Zq6(K^8L<VD#`XZk
z`a{|K(23`IZaI^@T2XQd-&Z98Lx+oUjU$D(wm;Ov3Q2ZO6^p@c6V$1Vaq+}Y)^HA#
z@q<Upi}r`j;41KjbN&6TZO&Dd>Umy<$YIVAE+H-SP(E=RCHg*ELW<P)U4f&%4vL*!
zp}UecAzYwLa6)hG8S?+pR%2cBC|-T6AyN*rKnyXpZ!r|cQEFS!(j-E#iGM+f=&$4<
zkneCyh&1t+5cT2qXEv$2n;R7R$?9?+K;Q}<Nr0Yc#YuKdSDHOoUS_JTP;|O~kkhG_
zW%f9@p6zt|V-hei*Tmfp6+xQ3N(CzL@4&nl%-@SLPkoJq$q(*AKjaZGvPu`uhHVwI
zZLB_e?@&PNQuKr)ThB5p6;$3Wqi`P`F91JLBc0XLFO4WIstT<cd?RvSKyIv)915ko
zbH8ljYud7)+sVTc5UxLb{IV+f<9R1LHS1(&nWZmZIO4T!iA^Pxvn=nCXA$qh{B_Lg
zh7QR|+Z-*-?SsR)qPgT!P4?hCyg;bly-b4jkptOI>7E0D9zRnhQ^q<icKK@g>cO1B
z1;^vw?Qr-)M9)UFOPXnn&2Q?o#WV?Lp3AbEJStI-h=`gSLLCB~^ji-32<tT;yVWm+
zCH)U`<>pU;a`KY_4WeFcX#h?N1{w+FDF!;aU|=r{;CDOIF>$$u(2y}+HFrgj;c?rp
zSa|=t7n!0$^7$i1h$D&B@2IezZGFx-k}ItU748&b6`oHt7k`2uC~-Yt_hpK>gwr?9
zn;pf;iiy*E^KNj+d&iH>(X*~<daVdP*&Z^1o<tch-l~Yw7T><kKwJDH0f%vSK;281
zpKLCfGgf+6_!S^IOVo_i>c6VQj&&UmJkgmVN;lgCsd~Hefo?%IfVtri?a+6U^n!&r
zYI?fxCt|dwuM{4TWgZRQ^J`hy4TKxoyVf#3&s(~E)g?FX8QyuUGj0GLHsU*HI$MY}
zWBuzxy#oJ)<*Xv)<0OTfJZ~Q+3?OI%lI0SoeXX;-xj#Sy56)p-2X<5pLaryOM1~eG
z9yTkz4YnfeU6FG?4cvMgVh$tGNL0m78Q>2%Saa!89$Lzs_Y<Tx{f!<HEZqqP3DG~1
z@nV%G+pQmw@7J-FY^5QF1qsV9hLwHS^_NtIS6|<Y<4N+MU)yHB)>l+eQ;9&Z()41%
z^q{~dD}m}vt8H|*#pL?6yy488r*kZ?5eNH;{p&Iq<V&5-I^t&*kZkSbg>O&Lz(h98
zoPbXcGy5rB^~_Mq?~@ql`(LrI<L>a8CZ81C&tw<_;zolW2K5gu&2A0I*1YhP@psaD
zpOpo^cCfE{>5l*3qI_7kR<>-PnVufbQ>0a^G<D$QpnO;?_Q;2!)Mm1lyTs@WIo1JB
zDf>faP}J%A7Y3yFWl`3(aE|&PZ1F%+0d~gv?-op6&-Xt+s!6h#3<WM4AvGkPZvURg
z@UMX~G+je}o(2RuTNuNn3~X_P`;(j3WbUIVu&&=qqy0X!iS7~LN|1;<!dtZ_)plJG
z%pk<6K}8hA7Xt6R8DamX;&s;7oCdR$nrkZ=*A{<!prVQO-WT+qtzRmawU{Gx_B`dK
zOW_ItwT+1xt(ND56Xij5G=&$Pl8{{7b0}&KhP0|?KDUE+JMxxKD1OY87r+(CH>(aD
zjZJ>6W>~E#rGq`fJR%n6S$3Jj?_F!6D(lG5%V1r7D;T#scn=c2iRJ$usMoKGmYU|!
z!0LQJjEL*{=6w6~rWUDmd0X*qQvKVpyn+uSSnwaEA;J_uRM<rMe0Brg%uizIPy4S>
zgscjhPVfN*qd>5TMD4`|uxaBd$|$0R0t$W#yuU&D<_mzA1M<`Nk?;2L1JY;|e^@gi
z1Jtu6t-nsTrT_W#R0iLVKd_9_i*sW_yxb~FtYGc47o9PChWm*jQN+DpeQRJ<u!QbS
zuDds`l7zvq45_3lGKiCNb%-%JOR<RwJeJV~lM1<fA8`SRld&JQ^y|WH)P(nc_d(ke
z_y=5b$$O$g2yXe_J*Xm9AHe@_02hk1DHq;e6iD_!J?#*Dh*X+?-Ghc2qO#Wh+$60C
z2jxzkm9``*(t#S+zE(~19R6v`#ksjq&lK(NJ%~$6c!&La*-=D750EKz(iU(Rk_y?J
z*%hBM*+#Mq{TzK_j-8F%cs@PZ8leHSFzoSgjG`?Q{R3wiw;Nh-32U3%=Zv}~dj3(8
zytMzLB;gxo#<@$}2;zd91Ez=atf;zk7;&nPG2&jzQ}xiiv$XtPn9o1IdusyW;`!^z
zD=+V+{%=p*|LCRs(1yV2vw(hLab7L&P*saT>Q&a~(Nh21*KDcj1qdcB-MuBD`f?<B
zW(Ht;Vy-Tl`)C2iC@p}-f4#!1y4?=060AOvjf%5!);?|S9Zrn)5HgwKtk9|^i%eL=
zA@-7Se%D|r0$v9KAHUqT=$rG0j3*o1I6=eq0A6KgXQz)Sa%T9h{A~|9Wq<q702nKb
z8qlDz9s}u6sZD7cm<Qql4_W=j48&O&vcsL|2}G=3YN)&FPCqsc-;&D7C@6^n@=|>Y
zipkok2_86?CRRla!D8{kr6kS$j5ph3@$m_B9MB72jr}$ST|b@D|J}oEq^CzA{0t1C
z^9^p1@$qUTbN}(}`E14CmL<-g)ZMaC8x_x<#ltc%!7@MNNGIs4eS}{@WFUrj@BS+)
zsBS>+hnrEMl$jAP0tuR4^R>)8Ys-A_q*<a$W6ZB4qpqfCj~TOm&%Fnl5^|S(6e@td
zx4c?qIk7#O4U>pW+LXxsN{Fiz%F)$_zjsw)@r98-6GkwvT01Jz&iw3Q*9G<Y+Ha_-
zbbWn&q|w9edytiM(E~W^?EK5QR}p;OS!c6?2kX8T2GlchS_%}mL){pBbnkQ#xM{?n
z1qG_PefwD%U<^bBP=Fv_zH*7}IX3h#{Z^QUoS@KhU#R$m5TcESYY5vo!c*SO{@ENu
zc|gYjjB53}fxW2g<{1ZcMJc<DEpe&?Ur*oMPFV5Ikb4T%gis^AW9ie(=7oY4z|H-0
z+kInlH*tuYb79v@&cr9y!%Z0@NU1~Y7kW2XS`3rFplSK%DcCMF)=Z0u#D5T*DK#X2
zQ2kerX0-u3oWyD*{mHu8<#>&d%Now<d|D>^KYa`VK;axl?mE8~Qnq|V_MKaN>U2>4
z!!&9BbEZFH&4CCDkP4LRTo?r>vPKvtsgGF-snJVxKr*c@=)?O<XG}df)6;VT?;)5t
zdJfv+fSGYWZW)l8zfEcE?V7>~t{Uk0krduh7GoG`M~=87sarvL;s9@SJVTHj-4=~e
zVK&TlG=ud+TJvw9OC;!HtYW=6!vSRz9@jOv+voi<-X|nk`}9s4B~?-XcHFlxu=k^I
zeixJUO%2Ljw-;r@$@a#Fry2|`P-!l0(H(IIQRe^MeoQ!ePQTnQkG`)0k8<E*5%~PU
zJD{a2NbuzUHS-bPZl*P%FYU6)r|vb<GV-t1VqK=*qz5{n;ec$DS*&t;{74!4;Zui(
z<6Mo@YmT|QV}-qk5c@!9T~uL8T3}_fFqjLp08EaxUzz|Cvr&SVNWPRt+jDgE`lQP=
zeB8gt;1z(7eCn+&^k&c>lnpwMI~0X1Vdsi_;xAyn*!Z?L+~-5yedS=A4TrV*WqnWd
zAiTeWD3!RDZvs1R_EB`syQ;E1B4c}|dm6YoZNkSa%M&&Kx`njE9%?%MrD%)bD}4%G
z9Ev4w$d!&h#lw4Ee&Mh0CkZYW=Ek%Ds>jHzu3nrB3`JjRfR(soPwU~1-=hwI=6b%;
zPx#iA`(`y;N=X_Vh_k+056P44K0>wo3{xVR<SDi1ES-_@jl*a($9&q)ZY6?@VeK2H
z?j0B`pMQo^zN&BGn18ovUl4bMj+Dzi+~^LC<~dINWLP_pdh%%I2Y^hZNEiz5OF38o
zB=j-f-qHeprzdZwXLI!+mu<-cXn_bm#i1JsI`Qz~!IR@(Fk$s<GcCnqU64QRfBAQO
zIH)9of6F5AU&&x&$XXQ_xbvXy>kL_MR~0NVyoc{6*LXuKnDQsrWJk~`c_4-!${ESz
z#Jf1V{-`N!yzvIzXZ}GFaduXu68&fO|E?SlaQZ;#DogX{KsLX20-I<XGl{6HF1inp
zZh4kDV`aPfio-{=eXqN!EN3<sUmG-SE$-jnXjMCGi_I2M{Zk(K=3Gj2mQ?jhEeg_u
z-3S!+A1{(rAPCXI57Z32!?_^2?5&qEI#PMhHwwDHk~%N*t>iz3NPWUZ>PD|s%(m0;
z$PUp?L8BS#N?c|d(VaiqTJWcY0vFhYg!?>@aB7B{t{zIi&){Qn1H&P+9(w_;vZyj!
z@?{_O^0U&kw=%BY=_bWLSY1dvJHwVnzFD%BSL;F7-SYM>7A!&4ipNRU0;7O_0{S*F
z85t`8i6FCFRnI=yGQZT6lnhS(ZzEm?G~TeEDWaxX4Cr}S77etSB5V#~=s78+u46Kw
z9br`stiA%B@B~j^k6xnL=g*B^a(t(oM?B_~Kp(gkC8yl#na4=}u~kWYVNeb|6>+oz
zjfS7DcBqJ@DZ+74E5Z7rDDlsG+)ZG4lUSUQc<5whp8+8kqM&oSEQ|&ClMa6m6fY5t
z6aM|aRKr@rg~vie-6C7yO&@n~gE#rZyE2GK??t_LWO4F7=D*!~k9xDl<s_ajME)>l
zV{5e~ZoTv9&!Q*Bkrm`01S6+l!BaBKepJW9T4={5%NV>07Uw5Be)9Diujk!$ivRrA
zI^jYkK%?TjzVdnc%Hv0+vX;wNfDIsc0edzRP^Rly{lgCSP{G|VfgbkibJ+aMv(_Cd
z{y3SxffwSG8ocFKsTBt$>T@ibRtAQoNa-&r?aVhV5g?R`38)vDrk~;M<_dZ4xTP}X
zEie8e+Hvu-rfWKU7=BIMP}3Qk>6^B?zbg}dA^Y?cWNY~KB#273o5Ehccu_x>)bRAh
zr%GNpDxJjf1&7L<S}Fi?qaew`<?7r7C~+}6Jr};7o$FultM(jOrib@L>eFU<ou0mP
zd{?aF54rPT@nRzuk^Q$$c%~u#oq~74nZ0!W)h9+ngYZ<Q&b6}|3`T0})*ogyUX{9%
zS@p(=5G-V$Jhrrp5vG6a&+3X?92SPXJzj{om=&W%%E~%2neeYh3d?9_ldkUITC*Su
z@|QVTjdjC&{a$`a>%mR?MoMjCLub+vDk@GP)~pko+maX#bROKy9DeHm8;^b!sMlR&
z@$gC~f;k%90-aRsw6yNwA3R7w^KW8JC12M*!CGtHMCelZ`*cSFn8=+y)+KpDcSopv
z>o^}qje|X7t+^{y%??4oj4;}_u+~F9t|)Yyb`=%%;j<wIzT0cI2U*)wXcqUOnxl06
zwczD??%`cRQFOzT355Wy{ptDf>b2)na3g4?uwoY$aj+LXO0!{drI9c-rAgv(_**_8
zusr9qkYzHG?S6YY0g(1DR#ssJ{`YLo&d#sgL?0eT>j_-LH8J$P7yd}p@UysbDVr{^
z#sUU;u^HoZCUah5=mi{=1g*=?Edx#Ouf`N&<UQ<JK1Vzk=S4GT==^outlwf<5-RaD
zUWlBOt}t=?)K@BVx)KMwce-X%gF*$YKwNXx#QGhmc{90kzNPo_v58}HSe{MRPLttO
zIzEtBOWqknrk5*DCO&f{#9JX*CUZhs0|7|KGgHQ)n$nQYoVRcA#?ahkTS0_yhVEFN
zBAv#sa3D*R)6xoUY7#_5!O{o9#R|*0+8&1ipoaz4<EeTV*4*6Oi|gwLSB93`E_-64
z<@||&<IbzOI;Uc)fyLF;uYuit`!Z$|MXSK{P;{Z0Gja$7@~l?#d|1SMz@42Y5yMb_
z@z*#g_!aZtDhJG|moV{mK%%Nvn&JZ6{a=8cJj7(-;dK*O)xS43{;Ab@@?PfzuDG~(
z-{O?Fr>Dnxw{pU?p;3DF&C`eSgqSzH9r;f6)tkZ0a29L@14T)?tP$_rz`IpEnkur8
z$n6tiNQfCL4#Uhyy&!3utMtE)`#edn)aEVNF#BdA`;)I=qLJ|OJORfmwNeGsb@xRm
z&g}S@PoF*o-OplG6Nt@-E;kxDI-SstOiJ=lFL{xvGF@RZ63cEjZUIi~x#L?ONLdG@
z%zCpoqW#8T+TTibJtyGz4-vLghd3cBnD_&j0WQ;O+YLN0=5w_$V5Tks#>j#KOUP6<
z#UqZ#fd;%pW1$DIQ38j7J}@AIsDem5R>N7FSwB~*p?{4E!sM@M3G^x9$8>9=9+ST&
zm9NSIa|spHsFW7&KGeLxT5E-H7JXG>-HGpT(CvzwOZh(E(E_m%tw%R2@OPRx^MpkI
zUOs%-A809Wf2UJ6ec418ELlog;HtZ00#vEP)<DOUv|1NOs~<ME9LYF?qyIVie~lrV
zn!1cA4iYx%0qf6{fOz(HptRiBw*+)20SXr3YbvU58ym(^;^b<D;3ylG^o$H}O{Q5g
z9RpA@e}ZbcA&5BmohB4OMNj`Jf9yv}$^a16%hJGrbo#S`!lQQ}C}wEW9v6NgD(bmC
zk7oNo_!}XbD^HL<G6e>dy}dn<!tn|OE2$N(0lDnO<)zKdg-u5=o?ev4;qkHF<<V+p
zR#uu^F3p)Ph*wilS3d@}3!IlPjd~#=&jlt|<ozbH<A-PS;B}d8Nyh`^m-&TC&Nl@9
zenl(1@mmd>mE2_&VIg&ISIo47eK<m$nVjw?wC83tngI?bI#y_4%par}erq=L6>f5r
z<y9Wg{xxmBs%umivHY^iehwU%VY8Cb7m*E{mpyHjJr(wcO<F~pA9M?$4%t(<n$<}>
zrlJ2qV1pav*+y`K9qm)8p60!M#->IR?v}(~v$7`D<{NC4z)~RU0A{Zm#`u})Id}%+
ztr{=aLLI(;{~pWd!U7b*go2)gf7Cn7jEp*i7#qQJrTh1HAf0AyRE(Bu?%z{_EZjVr
z((vTuejg-s8mMD!cUSXJ&DfX<SXesy`a({(CvZqeK-!Mi`)d}LqZQseL-3n2!UTGN
zK3xcy>|+HX&#Y~M4bT<sPL^hr*8THzx`jy!zuT_@4%SCN;_sF3dHb_s*iYr%yN@&C
z34l5H9akBkW5waDCaPK+2n;m-#L|OfPu-!yW?`(arlM349rkQ5f$?lt^+1-^#N#!<
zmhc0z=os{NN0EN-agp7oh0vea`zUcN8Y5@ni&6)W)fk`Y+u0vt^;_`u{NyektPpxv
z>EZV1`pfPYk;BUmU2;A_E<{+9^nbs<&L9laSxCSe`_}|MZuOwf`CMDNuWzDj4QWe9
zzX(U4LPBz4Z+6qu%KQ0~3c#_d`3}H|`c)}^wXk}=mLAALI*Hwa192#ifGub4#bRN1
zC=oB5^WQ*9OZ2_@#g&!c+r>@&P75BLz@_dCc7Dw1m@WIQQB<$K1kNBhEAo$l$PeZ8
zC#5b~tr$A=)>mx+{}?iQ2Zet3`nE{p8`YpifX0Gn{>MkmC~MKu%}wPA0!f8mfJRzY
z<?|a014P8R#Ym+C`Nxl@KDaYW2K)HdbV=&lyV>Ye3du!Bl^gd?$XQQ}Ab=)&q0;*)
zv0N(n&rM;L*FGC`QAl*+l=U_zynVS5;O8}_*D-NjkBP<=^XP9Nw=&O8%tcO<x$npS
zb`d~HRwtTUz;#j6v;Hb3@Zg%6n`6-?6Ao_f?F|--qyTvxa3EFYlbqZ$HH$Gj{n@Ul
zBxXJ2wY4?a-R$EkVG)rD6FzY8DjcvE{OsWXU_as_u)`ql{|RB1PT&X`9Q<r<ZccO7
zGChs&6&nl1^6a+xet6{k&y1DKw*BmzvWlH3ZNR1S_I%5j+y(nnuSR(hg{j`5PTU17
z_|!2may)+y?*W9CI)8KDHKhXwvyE*8B)QvpC2Vj6^HBDdgFVl<%ss4P<0Ew}FYl}A
zv2dbopm1>2Ht|;~)IB8-j7$nw+NGJ(CjQns4m|_2A6d=VxprL0rZpcn3ZoHVa~01|
z|0nWr`(9k9(7&T$SoQw)ISovskFj~EW39+=-zVy60toc^WwSh4>p)8;5}FBWKJKXz
zD+|k>3MIYNl~&=HDWF5Zs>HtA9S<BeCF51+iDqC>lo@o>`%zI)K<9hDngN9gqyr|&
z{d%|h<EKxu8XD#Y0(?|^;$GDi6;UAkib1=v7wGl&J1G7w(9)*;PYKwtef{>Zj#!z(
z%PNkdsOCE6&lGb@4760aRW)Ylj3#O2WRgdSmr0R~3O&6a(Hi|lr%)~F8}ckXb!Vs<
z%=uBJo*(GamJ|wA=IBCuOv8~!=`44b#;TQ4^hD;bjt;pto%@0Va4sts`upi(?Z6y?
z8s>RQXls|DocTYj13&I=H;YA&U>P%r*FsA4W7gPklaQMpMt*;NeZ5M(Gjmo}7BDn@
z=eApW^6FLBOKo@WoSbzc8+H*95jenZ*vx!1f!%5}s6o}4kpwL>1m5#KE32#Ui#zT{
zyUVpZ5*iwWB%T*<fZ44;vo1O~IQV<uI8e%nXFa7TZVVe!_fpOoDFrqdA79_nva)or
z_Mr7x-rtAR>%h->gVM;P*MR|i=C}<Hd`xQtrQwXpuwG<K8^Uu6Ij?iySE4k9uH-Q2
z#dvlL!TOsFLN$r0iaNu>LR(nt2_IltC@u7C?Wj{VUS_kz4im1h0O%&9?1sjtn}-N|
zTVyolrC4Oe*qAfP!Ds8scWlaGN`i=%32%rtft2NQy0_DpO!KvL(?;5rpNrRj4nFZ2
z>jEj-!R~0;p5Fhd`?!e^A@!lrKA6yi9CvJDx-yAcQm9p!XcjW@{#aXVTRtl~I(lns
ztMe*Mnag48IcNfeGlmW}b5iEV8Tj}Jy+Q~$@jWmsX2k8h%oiFPdN_EU_8u+XPu1-o
z{OsBBUS5b3i#rj1_QC7PGc}(#3e;$FWa)Cx4K!t2{B`C?9@n0gLFn)iitnG(ej_D)
zj6fsx{dEJ1XolYtIkDmqPiXe-vk@Z-RKH`FWWL$ndXPwE<p0)gtzmt4zTwJ&Z_Ae9
zfyY_|;hshySI^naO;1nHS3Ik}rp9*ldsfa!0lTPo_0ON*-uogmd)_#MoGMd7_0ON1
z6S?hTS&gtn!iX)PZ}tyCMie$j>mEHF85T>BCQP;Xx@@sjk7=8&pcbR97{usLCL-VS
zqTqxzcdo-e^$;HfoY#GzrGJUt8R?L_CDSi+@A%laACAl;%JK1<an{!st*DikPwQ@-
zVM0G-wBPdIu{`2^PvUeOs+Zw&4O4BU>>*CU$4GRf)FXu`GfKyrzrS@p_$Jpp7mIdX
zu46@UlZar|(cC7YqepYbfW=p`qT0G|F)i5h?|{CK%d+_j^&E#OaD3G+D3?5%4SMdb
zD~|#q<x#Os2R80*3zE`5A!8F`P_b$jbA0*z7!D-8io3hJH+-qBqlHG+S63BX?t%92
zk6un@CZtd+$jh66g#H4K$#nE54@q^N9rd4<@v5zJR^TMoN?Km{hM(EYtK0er6eJ`n
znQ2gXKIi8PC`<(*-Q|rxoLe`+8#(k5fC^biAFQ0R7Iuy1bZphL3R{-Uvl#ain0t0c
zYYt1^Ah9^~T!dNqh)h#8<z7?AXtwWi$*w}Be;uP3)Et9y48&sgNa`A}I&=5wmhw-}
z;c$N|+9fuj6U>C=a5e~4+uX{=q*mytSijKXMlAl+7O<{!j%%)fZ|UhwHDK-%v-n<<
zUi9<7Tq=z)3M_(FdQBnji0J;z(*&yFi3aWMj*DZ-g(`G{5P$3xPqfOOAA-((#}J5Z
zDRaNH1$j?@ifus}(c8Cg4}i}Nj>-Mv0PXr$T^+02nK@jo!*)jxj25`v=$o3r9R&=#
z51$9wJ@UE>2M5-0jdxdO_KMGc-`?CfAGCZG^Y{Bd#Q+NY#z+n}Xl@v-;+_CGCQ#^s
zUJ`VcgQa@yw9xvN8-vHy3A_$y7lEj%!kxJ)I2gmQFTNd2HQP5_^WO^QqxQht1@@sA
zUfzf(AeQ0<NDsmf$?kz26?wvg2QO)JBuAZ>9K03jL%15<S#O&}sdr2z1Xn7Q+x(Q6
z=CRXlm-HF*qo_O_`OVWpd|6n>m#tFNoIWcBLw%B>`9~_s&*W%>S=FuSJ3(H$ZsOFe
zMaxVw2ztaIzv4OVayUN!a*`-Ed$;xIuhXXpTZ)L;QFD+A%`k6Sfo3LhefMV4xG%AV
z?jg(VO_3XZWjw67@#DSGeS7U>ImW#udrBC-u!u~U={YsWtJtP{Cac})AXJr`8`cvD
zeRJ4ZOq-mCCmsM8S<~0PzP|Yd2@My2gFwrCIPbP^bANlTTz3!1eOmW>U_QhF^Vf`3
zQy=&U5zcsDs05N(;*aoZK)U7|4vuIb_!ufwuNG$(5)uM~8?1eu1#}D#@&MqB%QrtS
zv0nNBdL$)_dV-`eqXWqEu>Ixg3`iv`c3iW>Gg@y_+7fU%%B&oD3a%yoqo?p@&f1>h
zb3%3+1KDzvo29sSu}>d`RepWE-_t&q|79g>T>+xwW#=-clc+u6y2Mi?$ypS~TvyoF
zqy98^Lt9gcQP4F%<sBZ;@am@Z)!yBH+wAF{(CpcbF%Q-zMLVSV5Ty3lJFJeEcOgZj
zIFyLXnmcYUh0$o(U#cL8ANVNUPlU;)+Q|vU#PrTmP&R7y@vQcygOUpAC5)ukOCz32
zMPKKnc=muC&^R`eXWri4#%%brb8~nc=C8m&OHM<BMw!zSNe@my0d`+93JQNvoUBKr
zIMarVt!*HiG+;b`nC5P;&w+C^9PRq6pWnmux7T=W$;ZdX^a^9co<}mfNy&fQv=W$x
z1Ay$06>KYsiHU=zm9vgB7IfN;i9jr-_;WYTje!RC9V~wtSmh5wLbCZ|pe!DDk%d`<
zfhgJiNQ_osxGxN>do;T;a@KckVMwuW)R7>eNlsKE-?bwgEX9zDEra%(v-9Q5qy`Kc
z+<9;MO(rf3<)*7j!4fB%&m2eFo~kO9;(ErKMEUjV+kfBLSf-V|ODIH2nTsawoX^@?
zp5%^6l4nOx%7Qg#k#-Gx`{LK{ydIRLHdeNyvzS`8#~gmS{mY%lyH{d!1CluTbyd4O
zQHm)|uTkhq2zD|xjc433tkga9%|mT?=i~GT=G&N3)+P^W;ctGQ7QtdDXBHCLe8$Y;
z>W)wvgTAlT{oF9^To}sXy!SYH`e6o%P3wmuSejsb_B^dy^WyO|Aruhs5EvX>a4Z1&
zQQD?wO+2QR@xaB);t1_b@^rGVYk;0K7L3_IZdZ78mybS>PMZJHc_2klyV1hRO1s!2
zC@82fF$DYxvM9o2>g)HvmvDJOh-b0;{q`RlYp3*e#A@c1n!2>;e0jLU22EL$(byl~
zn~f1Fk-pYSe1-c}Rd6-TG7(LzPpK*UJ%2Zkp?`dr>r*8$gyQv^bYm~u{3U^|CX2aP
zfE^Od$+k+?_3^?48Uo?nG-cr<HzRsSX|>~Hi`Ri714rvdO;4*?(KK~eo1|{q5Ku`3
zz4p%MeWp8ZR7a*_SSA}*l>5zjB(>C5RY!;|8a7>xNUo%ay~q1}))oJSo+4*fmg|lC
z9ZJ?5V~s=mGo2kCDBl{QK8hPsFkppVxVW$|70~Kkos$)K=*Z{L`h6<UT6A!7qU7er
zgY#|qEFe@}tAq*CR##`>;iay{0zvsEkZJW+Y4o^3%+h9{LQkDrL7R8%IX5A<<_iGl
z+}^2Fcq2+w9}NtoWH^K^;xC(;TNuxf#iFBpS%f5s{eH!!tTfWWO`q;6tj_ITIj(`A
zLW4wdq<N1xI3PdOSe^5}<taDnE-n$o**p4rtI8xuTolF7H7JgrlmvT$^$9skhY*II
zFrp)YKf8%8kFaQr74-_0aFnf~FHauCsSOm<t&Vp1_F$fwZmeR_@4{ZZ!XtkI`;)c2
z4*jyfA%r}u0-2AAvmBhSU5Pru;UIQPu6GyCh{#Fr<FW|;rl06Ij=u{f{(Icbm(Wr4
ziGb^%k3+cLU2y3f!`Q1F1f9L{GoGJKN^-W|ctT9blX=$iK_dsO+U`U#2~c{&M@6H9
z^egtzCfAOAFZgk2`A1w_H<+luck4r~Q&VvUf1*`yUp?&IG670E`per3IxO)~ts9vf
zJ)}mE6}1i%b^8D#qnvwMIOF}(Zq0oG$%_W|`}r-xAtCC6B597MW`ZOX*_vNfC#i}P
zj*28kAF&<jr{T8cbu=W7r`evrTAX0}oV&qBSKsqy*K*Hqd@LgUcIv<dp#YLZz_AXZ
z<eOA8={z_O?72jqj73-=ZyBI|q7BE<liJ@#%a}Z<P{wFbax-)vK@6#y&@-2^V_2j;
zcPTbRkX~(_cv)l11}i4{rf<KZUpqgQBi%Xla{J=&QYLWTw!(ghA6n(k&nvc(ds-lx
zzERCK-gotVbVDN8*`&V0;tBMycpze2=k(J!NW)idYyfF*YMszgH|6RK4uxciDFX;o
z?dlQ*nZwAw@V+f*9#23U#)iIrE?j<CIsA8TuCZi;A`A9T@%9Tl&Bi_|Nk9vgwxlqQ
z;Qh53pjpP8gNe`Q(of9h;?jE63K<(vUnHR{j+<nBRvcMtVT3B^TYPIjh%>vVDBLf@
z(MkE46LHRsp@db<D`E{zP7U54K4yxGTAs{XTE#k06EW8QWb?9F3~^b#B}aF#MT|O>
zc%UUCgi~`rPhGYm4Yu;~a1MurrhP8ivg9qs-a)R$EYd2Wkfx6GV@cHJRL-N3lL79g
z;;UrKRY{8RrZPy9dar7@RfMx;g=36SZ9*LV-W0o>*S8$gt!cJ*kGx6MzUaRt`nJ}&
zg)|Q~t@JUP+mmFEGu5*4au}lGP`d6wGJkK1APF2cHnu+&k?EKQXca#xC_LBe2rf0%
zc<2y0Gv>RTWZYBPHc3XFC1cU0D5{YW;PMeXhMcHB&0A42!ex|BFI8ckVWhYeANBDv
zhpVWZzm;l6Szoj#rWcwXzMWS$1t8m+vw1~T3Ek7XUvlTma5vFt`HXt|f14Mfv3lzd
z>q~R6ukwn``9w<`#kE8kUx|9rsrTEq7w8Am?<&VnX6hCq@$FD+Yg%W-nJblObP`GU
zpqJl<bo9r{Q>f7A#ug{z1-b7=QhG3!&hK35OvmOh*X-eW+11wzDGoQLYIiCBbvS;%
z-l#M*Zv)WfJ1#9;7XJrpYbI^zmYpZ5$h{Y2hdQst8PQbetCW8E)bP?EUrbAj7=#<>
zOqS?^<oa1t8*u+<+;iMIk=M39gC0BjP;O7%Ya;QDVNLVO$$jq4?jkWGB<9E_@gT-e
zKT-7H$g3Z)2q`-~aAmuR|DKV9eh%>!j(_>-6rdEG%r8cb(L|KrTf#nC)>AP5NL$nI
zygywo|Cdn<&adkUjgnivLq>7kn?@z;a<|=eTXq!8`XN+vg^yPJ;-0m5<xqSf9X&^2
zhTxvk_u69n0`;_eQ*iWX_>1Qn-39|jV*^EF;9<JGO^LG7!czKPe>&76wyAB-Qp$cL
zV(_B!p(cVg8H7X9@&duKG1Xe4GBWECBE-3$7m1j3;7I--RbK&BW%qmyBBCH*5DEw;
z(g@NiDj+R_bSmB5@Sv1}fRfTJASvA)0@B^x-3{MdeSiP8KG$Li%DwkICuZi%?AhCS
zNBVFdMSKq25OS@W)gLhjhK9Bvz@~cY3orbHyb(rDUJTmh8TZoPHh}w4lQBBS)@hrP
zmoON(Q}R3L_70Hm)`{{Qm5<%{*(X3bVf5jRne)NrmDakaRbIjsQ6H3P+qsemxqh<0
zA9uX!n<2FL?YPc-;QLYkeLd|iF%4C0aX##EwVJlh=wl+uM|K+hB-6xNbLQ_x)sk)@
zT-BA5oNuSHJh+`Y2t&-YdMW1Wv*S-&3aE{_DDy6`KMke)-gAkC2G41mTgbV0H5TRE
zBrPaMR^sTUbp5{ZbttuXU@$$55n*BctG5h6_~5Uj`-a-lc2ie?QNO8$vGNaaImyg^
zhwfD-+clM;!mRP7?tD(CgD1zx{`!1=zZ$J6N|)m~T4KFQ{KHBG{o5V2ttZdQN2_Z?
zrvBV#ZcCaVTu@$p9W2zkp*XES%KM4lQtq2Q2QHp2ZSDtNqhPfxd8^k(bQI(7YwmiA
zQ~0<=$dq@LEUrat;_XfM7s`J%dqPAMq}I0nF2QMbA^5pvYOL~$&j<UbNZ#e2_R_Mh
zMXQg4-Yw8ai<+uh3LaFI5WFiOpAMn6{Z7&NhVmxUupf1M?Z(ezy#CKwyjGS;3ubIr
zu-@zU7g{FW>s0oqCcREXCY<mwZfs*$5MG(29m+P>c;nyZU?Ihoo4`b|m$rn^PeUc*
zF*WsPh)gw~+}hYhc9Wo<&%x32YhQ&IsM(ox+sIJOhLGp|<NYl%@(*O?VgJiIh$O0I
zB^MP9tnfsHkHuy=ci_gaaKbHU;alUZkWSgp(4U=pb?zz`M#%(E<&L(?OO<<lKK2G<
zU$%7{@8;Yc7C7Q1mtJPL<LS}q^7^u_3-|LseeYFxkXynoLU@!3SGa{8%Ss9!9u!1e
zE>B~RM(1<2=g(5MGN^ZJQJXlaxYc%I^WCykDlp0l^S#M+5^mCwp~D+ihL=@^_7biA
zW4S0EZz|~j8VIQ2pvqv1G=9ZpP97yGzdB`(OV3CjF%m?l{ei>ZfQTzCjoieHBcg74
zUT>7Us0$etdE7=zI!f=fQc_yeZ?<?~f2cZkvDrlmc)N?AuL#+`jUYIzq>9}scP}Ff
z&mS5)U*0{UDa^vW<Q8}0oOLbU-k{@x>1}RqZYWO)Lz(Jl*?3b^YMI9i5s}Y=jY^OU
z1IF!>W%YAr?uM|Zx~Sx^A(@P<>`N1q%)v%P&)`vg2ya{KH#J-1(;Gw*%S#2!@%Bv|
zG4V!r{kd2MZ$;m0MM`D#xA)@JmUe2|v7>#t^x#qyxw3J+Q~IW5$=m&BO4j(Y6-;4p
zd^DCdu;dAUELlcI26dfPhw$0ZJo4OBK_5&F<8a_mCTbTC3uO+%rL7>!3Cxmp5tpDu
zlD@8_{4$z*>gu1PCSYO^a>8KxyyV8=oqK2Z&xtWz1x%}6OGJ=&^7PD~BAS}KM195=
z!Oqrb5#4*a-A#o#HUzhmGy<8nNY2^<<KgdY#v~>tL~UQ;_SQ)kb~OnZQF`B}r~hT^
z`<#16inL++YmIbYQICuSZJ)4^N~UaN>b=*mMV-L!Boj{Jy>v#p%wg#cWC@sAoT&2%
zoX6W15(IV`0#GZ2Ou>z7*EBXi8p9n+Jc=v7`B<FtCIxcBWMgB)!NsLy75PKU&~P#-
zrS;R5n+{RrzCvzFqj#6_yqDdp^{5!b^^90j-u6!}S8LthT;P0HNd90_1EDLXR$RdL
zHaD2`*5#&v-lC-L>@^Mtw$?9Ah97j5lXlVd<-VH-$JK3I-bg)ee;fGO$7|~IphFP+
z1(HC|)fGMl7LuxcsJMu~B-%JRxw}?Ad`rcdC&<?dhk`rs#)~BB>7`Xa(QcOW{*lwN
zzmu=xJeuZ2&*515%^9J7v6yRQT74M8x5vQumO_|?E}`>NwMJD#-=$zdUbVVi$*lfh
zTm7%+KF^c8<N8uMQ|QY?U#N5&IG!WBb*zq#jx6DZ^Sc`F-#=Kl_RyOPm-w&`SMx`Z
zRD!6<$1xuB;>9OW#-UjcykNR5qc)M>sga=1cK*m6P`xtt$cKKR_m&c;(sIOb5OVGJ
z@ImHPW)n}DRdD{zZ;sICE9;L3xaNhABuU!nWrW@>YYcvATb6FD9zB?z_BFdNOwAI0
zOj~=AtOY%``>TodQL)3O8Bc`n`Ln5S2bCtW=xhewZIprLY+>Is8V5)YADmKDM!ER=
zRW+Oq9onxFBSUh79MKqs4&fA3ao2LJ4{Y?^q6!w(+{j(jV9UyYAdJOs^L?v18fW@E
zY-9OZ!#J>Qq969{{BoYfcr<R}=+joVM7WN{HflW4Ff^pa93zb!i)CD|r9CJ1P|JFe
z(5YegAj@Mey<q<w;M231YH(CnR<oiibB&FR5)s?7Ypp7|sXBK!EiT43hZSmv2F=aR
zGTE#spn=%sse$7K9JqDPZFol+IlV7_6Kj&;!R;L9&z;xvQ3w<B(x*1JsM`x{Nxuv`
zFTE?c^S)*(*?fE7U}=Z+?ef~{uV00NF|rqRw_<(5a>EiV?&aT7#^OGCKGet4r}b`K
zUU>lcY_T9%y~t5TrJmF=Z<xg+eCWUtg~iQGH5SmezJC4^%;d=armTt+9o1ze)Wi=|
zv1ij_kQ*AFiT7kIik96QP5-WXn#y9pbro-VV2L-I7h`RUmGN~59-j95Weus)im0Dt
zUFNUY;|s~&wvCQl$hvY+zVZv_Dav0WaHqC@{`8%!O!D12>x4l9yOvv5S=kG+R$YJ<
zeQZU&Bv^f1Eso$y^2Mr!kmkMKNS2OTiNhl={+A7b7Wcm~uEn%Y8>&?g9ycs0kJ-Pp
zl{_n>sCec2`|DBr6l>fHj{BFDx62#+MCUaHhlW)`xb1mv-}r)$e{l^{c~p+AAy29)
z=IYEB4>&8ruQk0-Q*Wp)UmP|lIw<MZ*Mkw0yX7~Tsiz%b700`=hDs0XU8X>Mb_3Is
zqt9<864+IiU#u+`Fcj)^IDACKbmL6p+(1{w>&r@0$nUoVXVIBFCabMQ*m}B`>Snuh
z#7Iv{^Rv#+{<_8fZ~fDJhD!*+(rW8OV>lP4xtjr-gaQ$rLKP$%f1W4-9X*<)q~vU2
z(~9c7mzHKO>$hDT_Z^eW4r1D#v}IYC@2@+pYDyB{nh`V+C8AAcD48TODmmuIhlDa!
zXlYiF>^a)_!<r@)78|xN4vz19U%W4*koPX#Gx**w7*?!2{7yL|e*;mwyt4~5j3jB%
zr!T_}9dV0Ri#tUHPSSjuzf2!>L`+28qGIdPSKwIduUgscn_OMwi7)MXv0rYNWT#F6
z$?+kr6*Az0Tx~P(8sz3STkeq-J+XavPfERV`XOQUc^mo@s|Mbb^j{hv+J%Z0;}!MT
zFvr4!cEi6#KZW>ju1@R8R*UjDwMXUrkifpfpPGE>>YclnZNNr?ruNXNhP|4PQJ&2g
zzsPg~n#!`vv;j+U15e}r?5KoXrnz-I_I_aCEyDCgAbCENy;7zkjQrv~25YG<NnA$E
z7p@rdXFRp3zF5l@rWhr5MG)+;SJUb?JJ$5kjH6`<*ES(4Xv}St>Q4oirA;O3GJ+qk
z{C2h9eEp^opT|b0YJ~$$NLaEV)x*%Y*3?Chsv4v8`MY~%RzKE!dZXxF!;<ewNsWut
zUMXSCGdcdNwg;j~m)T!_0Kx9?1er?m`}Axg)FgfzsR)`QE6*sNYZS%lm&HpF^~1Yu
zd?Q;uR7leAW5Ek$RwPqfY2)`ODRiRW=w$xX=;*e+ZsFKY;&1NM8l&~-2W<w^r7$Mt
zUI7fLd<>0aEW;M!9eo-hw{KB6{j}p(_2cNo*|u5exbVYD3~ZRI3#`6dkl|@j=LQaQ
z2as5PeyZ%pqO`_pzGX2<5yO|3M>80CNh-$uRIwK!K9%v-G$~`gSvc_fQ|MZP*h_yU
z|AelT+xutJd3dH-J9REg7f&MN30l(LXt*YQ-WU;iMNK*5r*<{MKI`!DU-eK(KZJy<
z;;X@lm5w}0vJ#qicG+FOr%YLgSyJ-Vk~vVUvAwgz@oVo^Dfv}<&8uUp$~iWD^0)AH
zYmuIt%@_10O*(2MsW*~{YdgtaQC)oYK27t_i13v=y`_Bz$u7K!M&G;B3dHf*VOKn0
z@aXye(246Xf}-U)RA9b^hTx8>Xv$@8t;cU~$S|jieTYzr4SZ@hKvUn;B4~D(G)K&*
z@{~tI=o(p@r|mn=)0TTBUsB4S4Bdh^*+ve+<rQ{E9;@&WB>#~#(ZnhriA&;z+qUTE
zoFX+V^6ypEt)EME`JX3iPFnN%LE^%zBCL(k)Vc9naOp&?<%a!7i%2#lvtJ+Rcq3cA
z9&pA!jqgjjeQ=^@;%pF<r}N(hN%o!ehy8kD^}ObIZ7x;TT$l7j$TL7wB9bZK52wPn
zS!-%#O-8gcHMvy#*O5IO+qN}6>smrPbQ6-%P{|jyipO2@bw<3Nlh5y|P^Gg-zBhc4
zJUKaXI_~w$vZs%vU+;;9izy3-!QJ#Zv3<kdc8sYl&eCJqKq9;LGxMb^r)P?EYdZKz
z{+k6>NvgHu%4{ibMeiM;Bj=SvpLT4pNC`KQm#k`A))Q09ZHen@Gdn*XtSD~LhWbUh
zsmsv%Z<{|p8AIPe=q@-CdpS@vHjIUojyZpdzIiEUJ7KqO-Mc3HX+jqV_hWk1wWaE+
z<aKBVE>xWVEAIl^<BX1GgFzuXw%R8btN;yd8*MyC*5x4%DkOu5?s_aiN={&=!oA$(
zz^L4iO$tf$yDsts=G>BB!@lFW>XMp*XMtv)P>1?IQtj2T-7jJTyS4{z7`H?xGz(TW
z$exkt&mZ*rMi)HEzaxId^{KsC>g60OQOD0jHrJNo|A>}{yIpcxp%!lzmRI<g;ij4p
z^&&%-rsU;LtQhhVAH=IfI0%Gw-f(%|;AN;-JFqr*(ZXO{h3$N*Jlf-TC5A^r#O3-G
zuc_a~ks{KPe16dkMTI5;5#$1TMzil||BDk_XH?zbUhjpSE9O{#R&J)_*LBN7k##4g
z{KLgj648!~IM7eW&@eSMm5bbJ>>hMzmln9mFO0B82P!4youn*3x6sZiT{$sKt8ygN
zHD%|*#b-q52L^^TbPZ2cRza?N%J8oNeSTT<lFyj+kEA3;T73BM-HC~{M57snNgD62
zTX^JcA&n(%VP7Z%B%>c6%3e?enuIr#YxAGgU!l3EVl4HZ_f#jWaCRw2E9hTsq;x_2
z&bS`ag5u)w7FS8jvgF_EJc!e1N`L5YZEaPuhkr=Ha(qm$#4hX|fD_jdtFmR1X^Dwe
zbnK`|UAinKBEW>o6jjew28M>n%-({bABmIIwEADI?~B`0ESFTv92gjmm!i86CA}DG
zlxaA_xurMye2_yvg$`RM6XD@a44#;o{pj1*$JO+o39Z9zu4(Xvrww-(6@;0KWv~7~
z*Ya+)>0(gdMfP~e+aBl*v>2&phm`H8B<ZeToNcJYZtc4AyBB>8jcx@;5C2aKM_S%c
zC0|J5_)InSH=1x714;kqD+38T6}SddEe9>?4-lmRB+DKlx+N~*!uQ|gf&oHV?@XCE
z+Z{PJxrfw(&n(GWK0LLz%iC;VWT`SJR%MjT9B<G0yiT2KW4i0kSXzA?xABVf)@+&+
zZ`UOv_nP@h$Y_E(q|@mHkEAh=g5k+mZU<`8aaHDX&Y{pN#MqmQGg3a!)StR3cp-?w
zU6Tx+Nv+H+ZM=jl6;c=OYH2aj>im~!y2ehas3n0KPL!wTH1SYKi}aOjw%ZMP>FTb}
zF`TUMT!*7eOVv)bBi!IVaG>8nd`p7Tj`G^t&m_AcC7di9ck8BG#Ht#tU)muotf^Gi
z$S<`jU)F)_?5l<*t$`o}`#RsQ7_$_<4<p)>gF~iRgQ59XNzzXj21^y$muYA7b7N`s
zmJby-m$W$KU16|yOiNP*wg~>tR{WQWSHdT2GIhrM0u6@*Tiuq~mU)dyN!;k=+<#*R
z?Tp{<MUqWdQJz6EwTDKiVpFPzjU?%EQ8IaGD>3OLKk~~g?rRk&**NINc#7?iXi;_M
z9-Z^mrm$B9xeoc1YRf{{xA9W>Q3H<HlZBt^FC{XhEgK`wUnWV{p10M;t&}`g)N?I-
zcQ^HupIhw$OP=!RzVWlshMfnow!$3BXu0o$``@UpmRu0K(J?UiWNQ~IyiDV-805Wh
z#)RtMNcHUS;rN826aVZ3yo>=*-OzpIBR18wX&E_a<vu=k8jXLcvTg}+7Lt4CAd^hz
zJ{VqBmDayZ>jZl>r4$09;_ax<&D!fwu+OO*!ZDf0o=}(0X2gL6?#OX<peXv*C5w&W
zRW0>q*ReIx>Vk@fLjintX2u`p2gj<ccnI!Ff7qwcb*%F1XU;<Usc~tuoZc;7x>7Zr
zWM$lW?^aG4GsE7ch$iTq2+IU1;yE&F9#I#)R@j!T7dlwFUv>wB@}_P1@wJSaAy(0f
zzpFCUAoD%Ak=?(ny@5R>FeVy<-LdM+-Gg+}lL4HHMLf$F8f6qkgBAdIpd;T!E<tXG
z)q1H_^ySNoaF11~a60sCi(_K3_qhaE82~#T?o$R9OJI9kn`$6%X-6~4w@HuMeRQP#
zc|(k5WDUc4q*@9G&)|1Ybr{dVH1^6Lw3~U0xBoa(@OrqFTFS&cPoQONmNZEC)11Nh
zcCo@Mm-FP5!GWKns9_TFn;(&=KweTcRnJMEivwTdd4$VHg64l$>~L}g<8WsMwKw3c
zjC(lp+ud-~#d;cG!~wQfT#eHwScdqU-3}d(IoTx#8s{(EoNitj%ov=hxAt^v81>DP
zE-Gi-c|XT18v2P`!WsMDUJavZrk-dyp8x8?c6*4kX6Nq0f~@&h#0U3a;03$9T{f#z
z;HxJ5P(>!~d;5Qo$PR8tc8RUlFOtb=ueVF3E%mNSs7(-p@h$F<F5g@o3D4M;yIDR<
z<rJ9W^tKv;1m)-wH$Hq^f%lHnrzF42z6(JeG0)spBYDU+MN(K-FX_?jA3<<)R6hQ_
zcTb|x4?HeqTaO|Bb*mg#)YWAxeu5ROsnWOY+Ws*7T0J2*EvkIf&hGH)UUAhMlK+7T
z=tr9=eAwUQJ;1QhlI5;w`9yv-jE^Ym*;5^H6cPhf9;8h8+{;KO?On<{GO_D75u0lN
z7z>w&98Ey-$z#9#+={xV9mJI#1-^lK*W@~8koA?}JpBh5x81MRU@KibG(T_bh68Zl
zp+%Zu#s3~Vt4~1L{mAms4WoIn#%fJSGI&P?762zXb1IWabAtp!2<97fRaH2Cc>{?`
zIAv%RQOes@y6q^Cz(R(w2*))s69Xbhbfew_dnal7tI0)ta1GUCIVU6^8rl%gGk?Jj
z%bI6X1u+kGj?FWJ=v5}u>wEj%T=|-72<bnbgme8Xk$3(PH0F6W(iYe;B%pY=xL|6n
z*E7$x=f;lu!;KcwmasC5KGtr^Rfl}YC;pI}^z59$96LBT_#r6(euQv+Le37Wa0)V-
zEY>vriDh|o+O;ihTN+3bQ}g@2Z>IBo2R0_ghnrc~_uBB>#|D)dv~T|1Hxi9n?~Xs_
zH}7??R;7~Hk23EcLU?_{Kl+hP;}xzEBAs3WLjVl>dwF32xEQ1$G~UZtH<>`HbHu#y
zj(8-;w|4&XvaVC0;L_@h7eq1f!x;;UjlJ98PtpDqY3FfE*6j^^Q)0FF95~d0idjor
zn+Y)KY(@jm0B>D+vV{XOW%U|QvH@;}Qa9dPAD1#WX9B+V5trjt@o<*U3@Bk^((Bi&
z*DKH4peuW0JB?@Z0Vo%NV3s?IC&C{Z)gQtS4>dJ|DI7sK6T1CTC^^SHWTjf;Nk^fn
zI%ou%LjlZ7N%cI9qV=<4Xk*$hR(q5ybk`G~HKyKS-Em?^SLD(T$&4`Gb#@SYKk_Mq
z;ct%2y(mP*PjsT$Lsnbj6S>J8rgWOdbK38BgnK;Qe3d4wEiB#v-E+E9zBlyGKQWuA
zqG_a8$VQ8jYy>`O9_dLN7!bgw!vNRoF`N(||H<2XivRo&AE;7(05~D|-O{cDVDDN&
z0UMd8Pq_B<*40~AFQK@M0kuv$VhaiiD2jT(%QXG|og(`zGxHn-7G@6TW9Dn|`P0pg
zi!J9B^7O8QI_rHXBCD&9Zykfa7O1{x93y+HgnNT|`ZqzXEgn7$c(5^qg&Rd1)BK8=
z3?0kDYw`yNq)?M^mw|7sSxD5pp1VsdsG`1Gvg*$9#Muk8@q60Gi#M}RZfFI`jy!(*
zd2+>gj^gXzz4QEX+NWXrH-ii7nu@V}Yq8R&gjv^lPj|Y}prIkV`PkCJ0)eU!1_J}b
z&YJ4xQZeXGeLdWszj@<E4NxYDfHtF0!nc(bBR&x7gAw0i(YprHYrW(M%mdHX?+37x
zZbQ1RNeiVqYcy6C2E7ib7pKZ3Q&Ce_>#@9%m92vx4l2)&`DJ(Ybgu(=2DEHGMMd2~
zVMDmN5z>UtS0F46d`*ACS<|}y3<8w#m-DfWQh`wuw7UXS2@=tM<h-Il(G5zrGas<X
zQ!MfNJ}Hh)_Y!C|e?wK@z192C01ouy!C<qvYSi_RM>XMER7^B(Dn?bMc^)s6uGG!Y
zs?#!5EAW3`z$YOjYZA(ko+c*g(3*dM{M$jU>2+4x3FVkL)EFGYu35a8BE<{XKBU&h
zU(hiw#r7H~7F#@0ciP01lq77|1`GunN}iGsI~Hc64D{se?Cs4rCIraH$ePzJt*vK)
zSQ6548$B{oo1o4r7k;X&1M}<zP@iD$d}s(j(*O-HLM5hwVPv=))Gq*BBtLKYr>$)c
zx`i%ifbysnv_hQlDJv^;SuIo{=f~sN$BBuFC`Gs6`L2>(_0=^LBn$}9H#avmfe;ZS
zPp^}asTh*~Z)PrK{hgTvucyWi7hn9qEBI0UMP|c!vBq%vRKxsM`cl$~<L({@!v7?m
zz+N}flYa4QLbCC?pWf(_%WV`62F!WQ5jL}!?vD~hO@{aSUruOy+ii?5gTyr?p9x)|
z;+Wrc3e?`16R36;Gte42_XBk^(2*qcE|o6@$p|#iCQ{9BM5cY&f6~c??6&g)3i3`{
zypD2#5xPcCiZr1&Qp!3A_+d0Cb*H@kL_fW_*aSU9I8|T@Na=N~{pDqz9KS`DdZ@*J
zkQ@<*Q?s!2EHOa=^HbVx6i=kP+pp}2NM^&rz^7$H=uSTmtkf{`Y~88RpA<YQyN!`-
zagTw*)x{srbYuozgL<T3Jyo&Csxyo9_#tUK($;2(@yTa0f4(CslrAiAgUt#my23<n
z^`Rp%5=v~fHht-eC&%OGnar7;DJmIuAbDcoKMF-58dC~(6EeWd%mN%$2f8YF&JLzW
zoVHs5aMbUpnOp0Hk0HxZI21$}A`G(DtKT(iJvYye*Daw1M#$&y?)}5C;fbK=J*WoX
z;7YTga1@z+_6i)5(F`-{O&(9=?Fyd`^iSg7RCl!FCs5<UeQasIXOmDH8$rUS`PbXt
zd+9rJB1(K9O69J$Hit#S28Dcp%B)1?_labZgGv7gD$hV*`+${|kb;5&4HOz2(RA7(
z74>64H}x6yR0u+7ktWl9Ujss~w_|HC`(Uf=7L!613XE=pNxJYXt#=D_g|<O6F=$>f
zA}^1{6&oj|m-4{_3|ExK2lo%74~08BJO4dWq99yCKaomK{jwA_@9!p3TU98WGP&)@
zTGn@+$EZoHg>rGm!&6KawG<F6JpO=Sh&tA;wdT=vOd56EzCv>Cf`r*&xoFZmfFq)T
z$;N(7`647=H^pEs&C4q%djvp`Okv`zE~-VZPme>xhj3xKbAeq48wUrtP%cAvP$Fze
zsDeiTNZ%arkz7$Rp`pYZ@^OU>&Da^*f`dQ7zrFo5ev2t9c<-tPBJhD7uz#Ronuefm
zf4Hmu{$h?)ygdzN<f>~08?jx<i~jZ2%bWkpByhlSnqFDO7Ph?0HePm@aso@+*^i-c
zKU>XWmjzI&C@mdy4(B75P)7B+K4sGkiw1qbCw!<R(lmSp9vZ-!KP01?zdaB>05BmF
za_ef@b{n_#P^B>btT`|!8b|rtcS|(C4b(n$Ov^DIW<!H=<d_yyM8F_D$ps9Ij4&^l
zxlSr8kNUdv>zIBdw<e8UVlSI21fXJ*1s|&cJ>!hf#DBXzUag0OwUCe1TYc;Zk0{#9
z{+@<R`|Xh7E`Vdw4;4O~@k7RN19VlPW2iG%R|@*BUIII#y7Mj}8X!!wEKWdB0ZjqK
zS+dIaNWkIy2Gf8*!=<~Fls_b=z3}Dxo<Q#sZn9L&53uI>Yfr5fUO(fk{Qdj4=;|Fn
zXDDWCpvTgU$r0mT&!raexdije9wgS-rb*TL!nW|}yxNu(;bwI&jc4roX0=t;yY(;R
zdHy}Q(fvc-RsA$N@3INHTg$)Nx|>w&&goKQN6v=$UmJ!OXvy{cK=J#5AG>yt#&Ki|
zJy5Xdq^{7D_X|j<EG#Ws)){GNrogYGHUl{R9_~=+PYH!)g23<JKdEKWz0+kO5kX$P
zn)v)UI)$n5r-k!!Mj;wXZG!*o-~rT1T)QX^YCLetPWOfhl{jFIHbHgxLI<!t`BCtL
zsjJgs1|{$R4^Z34HQRX&wN_8<P)2H5TGB#aDpO8Q;Yl2ekcmae?5}_R<{DAwz>pcl
zy2ydi2=|PssGInm6ZNk-dqUP<C<ne{`!KkU{mu^C?{azeYWk6?Z;oOD$Qe_k-Fz%r
zS6v-9w25Njf>^C=f6Q?N$~3)x!)5lC<(8lQ{N_+hTzh8|6B7oy#PYM_=V$C!u3SNB
zj*U33m%sI+rX>qi%+b06g5@j2g@`}Q*jo|`5)!4c+@eg9`#Lh;ZkK^2GNF2eS>NpG
zM(BbW(m>V5n8OqDZ%@^{gM2|)-a8`LK<*fd8ynXZoCLz+(w@?|i9a)7GJ5qPSrTwF
zuWw7B@VZC*T=62?jMFPCDscy#)=NDkjcKwp;O~UR#jk=85Hj&2!sK~E=fu?1b5w5?
zrzIsw_MjBHciU#WrT9_K78PPvRXt-N!G%@6O3L;nNh}mt9X$dAWum52dksS96l=f}
zqR^)R{e%vIjsjz4a>o@WJTi{gsoEdq(zM|OrUOCb^#?-GtK<%CO$EaysJ1vVgxlt-
zuI{`;l>Cc2BN|^eH|VPfmYU8Kk%$pO8L1G9EsM8o%+?`Tw7!!)HTi0#)XuI_=9{m{
zIH|)HaKy~EY!U>IL#&GEuVy_PA9DL=*zWzD8Eo?smM{O~hmXuvu@65<ueh4VNNdYA
z{OlWKG4o9~@Nu;2TJ}4UAG&5YwN>(=X(Mjz82TeW!0Fk-99p{1##Yd9oUcTmAmsbg
zmT@RYu=!^<oY4wECUZ5)r1gHe!yN)>C?$F<%XfWycB~$kXg9*eDP7J_ZBY{f5jM@v
zS&A_~JVbn)-K7I6QYM|ubjN(XN1pQBF<iKe%?uO%QizM@c}~WH&rgOYrbq3fO1Aj#
z7Iw@p?iBAi^8U+)nbYUknrooOYP)47d1~=>VD)j41)oP*9@$Fd9rI1=$FWk6x$tow
zXRGVJR?Oz}x+ak>fDJtm#i*qMR4@BWI~YTaQAN~=x$&Zaq!nc%fm>F&{Ral(7fuM@
zT81)JDAGFxphr1L<(DoFcmqE*V~KZc+(MK(e%^b5rE8(tPk6(>ZhM!1zRPjy+Wj+U
zB#?3A?%$JCx7(bbgysAgH+em=g#H3$-R|$Db;v=165{pQ`cbuV2Nd^O&r&7cF0^Sp
zg`ODLMrd#);z3c+wuU3DeUiekXeqCf?i-2oYdao3+Mad_y!>onp(5X9h@$lLq8LSF
z8iB1>o@-X^WH8>fNsLjIxWK17`>svKliMaw9$Z#2;>b249hXuBC>IOA6g=$TP@^@x
z7pDvwQfA=OE^1V{qNDW9p@lAl*b*IT`Fto$u0P^0oYtFTxA_j7bq$DWrAPU}tEENG
z8}E0NaMhTmX7i_=p`ZP6r`P8w;_l1}5xmtk&A2_i9rLc^wuUI$pR1lkHofbya@A(0
z2Md&X8INp^&DlWSdY%4%<6yr8P+p+uRjS}M9V^537_G#)d>JemUo12dfS!(-g@wg`
zAN#ql^&9utNpg<E8uD&Zf!F1ntjTlGT;rp(DetFpF<c+FaPf3R`}U(lL8x0oCP`J<
z@o=z~JF0rn(^=hkvg?<nb-VO`XP^$Z765zT(t=IBJU!(&=axKrn&nby!72kJDbI!r
zX6Ypo43J5$4^rb$Oa5-Hq~HHq_64o`a|}8$_t3UyaUX`GDI|XXR%0c9U?NWdke@hT
z44#paGZDZHzBfi6*c)3qT4(%k!s7>X0rvzdV07aU&M=J2srJTnXfX#P>(FxIaxBGv
zT4&kLtGgNmzyFD5bZ=y|=m;sv-E1{wa&f-iM_EfBeQoVXp421n21hssqV=qRJm1%o
zd))c!hNEdJ>)B?^9u4XF4wi3N+Ln-A?bge-Q+rQnQUHdZ@5HE0gj)BPxmvpyvH}Bd
z0_r8$WPf{m8e9k(0Rce}F_Oh_0yoC~p+(g1vMw=c-bF{)ZD;lm`OTO5Uu5$M9v*QA
zdqc98X~yH1UBPxLYi5%kC<a~0mKIRQ@y;P3yt?5Y699@s0p$DR1}=2}TjEwf*lN|)
z)mmCwXs=aNzR0Cv-ld~UFRO9{?#ZnSrXZhmSYeRg%l+4;<!@mdHI-9H-UDC|f;!D)
z)*s+8&_+&pHgSx_yoPJc3?`N4zk4X!6U-7ki+#W$H|}2{$wgiHG%GSIe&b)Ch5NNB
zPo3Qhrf^w7$^N0;zI_{FF$^>4<l6?26-M^^Mm$Q%$jeoewO%dztKG@||9Tph{tOQL
zu1%Z99UmxDF(3?=sz5wCUBP}a38&D0Kndys0G9cCfKanZohB6qY?FG2dtl*P3ZW%v
zY-^QT=0~>4D3{CWt~`W^GA_rUp=z=Uja2cHqS)Rdz7=(c{Ow=uIHLc!vxp1Hta4*Z
z<8nWq#U{eW1v~yEsDUMSATp#i>HYjCWY|6+aeBU~{dYdk<Nr-(|HZU+6nya6;a;e^
zOB3kaqi<|(PWH;H&%$-?2dI;s>@kCUHp=t<H6K5<nZr5qH}47~Bdx{dx=)s9iKYMr
zzsSK7=@RHQO*oMAIftO1me`d2b?V*JTjH)^hUw^Lii^hU$ddBDT7@7d=E|`HZ8F09
z(%?^$=zif<eX)f*<qN##mu>v!2>t%MygX;&g`+$=I+~`Ug*ySt-ieTF;sF8Rnfdw9
zacCRS;yc-R!9s%LNPvCs<(;HR5v|ofj3Ru&u_F^k7<WdiTJXGmzpA`cGs5$dUzz!h
zk8~i*6mGZp40+i0<cdbrMy^4z9;&WVQ{U`9wR-tU&S4G5U_N^m<7Z9cqBYCFCY%{h
zRr;(rE|_SDlPYVQj5m_@E>wA%l5Y$g&+3>UvlNiK8a_Gs??gUlxeWu3J96=WwhUAU
zaZ})9Ag~gJ3x7pLh0Z=Z0pOo40Snr&M9e2BuMkf#N^XF`J5n`gDnplRoy|np-TV-R
zd<1OH)5$<xvn>n#@Va{I<n0Z!q^eT3+d3D+h-Zw@$=a(^J!ms(hx<8t{+091A?Mw^
zGI`nb);RVM*NO36vrIHOmHUX}CH{&H>Ly(V=AH8fn&T_T-!<<9;yai0RGDj2C7pKx
z)0NU|pi~gl66JBtmvQqIPg?ZhE$FYq&-X+MGS9l~*_98x_`hs;vcD^`S|J1I!%UoZ
zcjo|WlfRUC)FP+mwNn1W;2L^;vt&jq^GtQLgeycbST{$uW?spLhbzyUI-WmHDq7aT
z$+C7uB2?1IQ!Wf~*R<TxC}J-oZCR6YsjeRV3~9+8M&W-ulB7z?E1b*ih_pL5_31Ts
znl#3oy#`X#jit1#E%;(vfZ~y~IvC~jNI9jEs2gbbeXmE#!$Sdo{6XU}1G%5>vSS)4
zk9;T3Kq<$Q0}rO=FN5sw@8%b+aU&6~Qa7PrKn>bl4*dmK@%+KKjK-W?@FSVsRZ<0z
zheK&?B!fn>j@XR93J;n3KH2|mMow7U^X-`PjW+Ie)ECRYaUd=gpUd)$ih6H09QB&&
z!rgE1%1SRFX?)Y`c05JiuyB`?-+DPP-~xH_XoUBQbaDQ=td@JRdG--Av>S>aXes%g
z#z<Ux5HWue>$Lsc@v>QpjZrWcXRQpn2J^2+7KGo$cZb#gXKw?4Is7e{2?~{ea26{B
zRFZ!|#_*rf%IyL-$CD>ZxMT-&wrTTJ;ncz2sFLx9<E1LXJ7w1}*Bfd!$&e=}8DGQI
zE~GKmLcgOJ7XzP(_{ovF9cJH1AC=2Kb=SMVAfjs+<$qeXZ7B4XtLk5vt8kf*m7%&7
z{%&%TQP$3vZ<4#wH4JIfx^eE22-U=Kc<0~*)%Pgrx8hrs;yXEhUZ5Ag&9|+l_*Jy$
zb3>nJ>G|;!xJLdmJ|jW4Y_+QVu9|hY`bROpIPnG)LJ0QlL7z|ZYmN1d<$VWCD0FM#
z@%Q~nOezW&*mUhTK8;xroM%6Kit%5$5)!0P8)*IWhi)9zsCQH6zPnx<!yVo<qn-3}
z1-XQfy@{+@$c|#?ZkwBB&BrTFH6*?1Jgrc^Ql5+8H@b{S?1UsdB=dfO#AQ2u68>B6
zP8c3=Ik3h}v0M;q6jdMR9ieOa70T@tveAG37b%Q-`hFfC>@st6^Y3SU$?Y0Xf=r=c
zlOQKQidmz_5)cw11m5V-7&_;(AKrlgOAf%5Ig^IE_D?7rivay;P(WZOSdI*G4s}CN
z=O9IPf%2vYC*}e+#MBl7?+wSYW}(YZdoHukBg>Scv~ucyPg}&GAUlD3<qEZF*6Uzs
zs_Um61HKVD5%9o2wL5}c4z%8T^i8w<^k(vS5-RH97I72LM3TsZDQF>>2q&cAD)s5P
zX2&AE&jZx@q0WDNr^-#AbnCT|Fu3@+!Nuu^q~PUFaZbljwUTJ*H``SuRKAl_?<Z$B
zo-2IL^(S4B`1)Q`Q%-I8zdc|+`b+Q9I>66q^Yt#{;MD4|TrgW1U}7PW2tNKPO9QH>
zn_F8o_4V}uEbw7+d{HOCzJbMNp;?C?wvZjOifh97x{+ln8G(`4h_)6Nq?C+qME?1J
z9zBP}6^7wzn_U^q{NKg+a%TsfN#_R8r~D|ZA|)cy5rNP;J3JQso}Q8Mla&^CCk&3a
zr;{=nHfrbzpYhykU#HTSk*Qs#UF6{Sl7fphv46D%yR1@|YSYsu_l*vW|6l%-B)nId
zMWx8>19aR61O%ktQM!EZkIsDJqn);~*BY)Ae@+f%jofEHqghcPX@t2qYnY=vJ*F2X
zuW~=V%K%xBMvDnEwzV_0Jc^F`k}KH%PKLYQ6Qb6M4DXnjnEV&;Z*NETM)czQ)*CmR
zZ&=y7ne6KQ(zt+}*%Eb$F2L{#$`f^qifjfr^=}{SM4cS%VIP>1Id$Z_<o4uRJz*z@
zN;8FTL*EiY{_o^<J`@*v!1Q?ETyI|};KsY(&94NO7ruo0^T&;1b-k~86{$*Jgf02W
z@L58c8!a8(LtfrHQIc5CAR9(HoelQ#Lun=P19imFbvKK_js5*ZFP>;zA={;!bQNgI
zU65yc-@|>Aluz;AQffax)=jtvY?j7PjO(ImTAIqoeyQ~vN-T*)H<@#YH4~26oO1qc
z29_sKOo&mI|K-$4`|wenET>NQlGLwX`46U|sYmt~Xp0=|*eybyD|o?#2SFwaL6s#y
zG-YaG;T5p`Uctt`PeqkdQ*?g1jzB`(2l6X%{mMm0h1Z_<#6wo(s+~n*j>3?b?nEHY
zSDD)O=oFO$d4hC_2B-jD@rh{wo3VL=DBch9KKo_o7XflHA<H1niyqNkrT5bTtNcn|
za#dcSS%ItT*qz(F^VgBrBBuI(*UQ|QmgbM{+MmroQPvcYkhd$W3>%xCXWpv6^Tl!g
zs%@pi*Wu`s=!r9Yw`NaVTE2;}<L8{tJN`c;M~&dJH84164i{=2^OYYwG0W0!e<tW@
zXi#N=m7)B!zF&||`4jdO%N0@(p>Qc<eJKh{V6w1-nA_!VYfeXu$DeO9pG005RlQe-
z$zvQhaFH5?ooXXSn;0%;-KP)t7A8G>DX&tRXX3}Rwdw=65bDQ$t1?3jPSI%Wj?Y{%
ziUuHFZzHTa6W1nP03Tdf$tIxm^zUBzw5%qr$(O~)`MpR6%gm(H@9ZY=GzD{f*UL*t
zmTUCIy$XM|{XcUP{<)uqw^_3!Z`X*epT2BIw`Gyuo9kY*nw4*3_j=Dhd#K`1G7~id
zJm}3nrH}}l^JO2?*qErk<=Y_o9_Xr-D$1d9LA`-}k3{gGyzm-JHZ3>5Qv2t8N9w^)
z%;Ovy-)-z2?W&uFH`xmK4J6!DckUwYy#rhWOnc>xpaFB2o9`fO`#63wv`2JWmV{LR
zqts&9GC`lPFLU!I)P7-|YMN1)?{d~Kc#;hh<Y{g6;q45u)<7$=5haV|nJm5Q9f9TJ
zye*E5=WfO7>$R1t^Fqx^ms=Fes~i_P?)7D=;#2_<6R7aw803@q3L!Gu9}1y)c6J!3
z40Mln7W*Vcz>dyLO}C{KlF#hPQ<_aayGs4*^UOnC&9wab)^;BoXRaxTXj8ihYUluH
zvDJ}t{(%2M=r$64@7Us!hvFgNmZ$qlFrjI2dSb4pNB^Ha&}7B3IsLQNN^8Dl)RL{+
z*Q+#gitHh4++|HU=f?++B?8x)#?n?4W2u<6s6OEcjNp(R$XwjDi68U6iRJBW#!vGR
zCwTUmimTDJ+uG~pqH#%%kMslP{kyOWJPJf2ew3*<PHMi&T;{11>CQXc`LoZXynL1r
z9DF00zOp=3YRu&&qm-hoUvSAd@VBP*j6|35kNsXnS~^kP*pPGH9}DzvKTN)ihsRJ<
z26<Q8Pz*F;Etcb-`u=Q77Qz&}B$gQrcQ{{1dG6jF5e&X2OaIb1%ePY-UnS{GYBXhS
zA73^it>oI;1vIo+QFE@SBb{cPfJ8CV^VethSjBjSc5t((@^pd<(obcS`x(t`KhFn;
zg;689gXtX|9ouWt`Jm1o)~W;SY0mR1NJVr$<g!8rjpPgre&SK^rGtA(gVGptMC%dU
zu5kQvl`@VXf!v#^$_-~L6<Ec>In3kfh<qTsOapHUtrkQil|{wv(ZE_M37v4#kVt}S
zXkvx)@e<IEG48p9Bx0>Y{c)I8=g+NtgI-ZkZ8sRoOBM1ZNdgW(0<J!tTz#P0a@eTG
z>Wvq~L46FoNeQGJ^MU;I-TVfTiF&o8P0biIK74HPe8*<IxqF)I?(TjO=B~<T(72eZ
zVAJ)D-y@o*)T-#30@!J!v<1YL|HBfbUW$rnLI@`2D?Ud`*01yYJQx)YwxJQMaG@0J
z$;kAKN+6Ks?Qz~lL;Y#4|8_cTWj8J-*s~t%c_)jp`xc*2AeRcS94&vPR0|Rjx2Jks
zQrtWRqtu5Z8U9q)PcBO>14@X*_9IZ6VvK100ZI30%hN-LK49x<hKGa8BSOMz(8F0c
zxN0NvMo7rB1o~H?hDkh9pjms#cpzIK#(6Icn)j(9TIFVvw|yeGkuj){7|i_shG{8c
zA2S8Y<*8$K)40BOxC`puiw4wAcD(oJ{Q*QlQI@B0Wo6ZFCXfhY7KAe0PasdNHkRVq
zd>J?GPsSM!qzQ&c$j%}Oze^dkGY;OL-@asmZ-;+}OV(_#J<AIHOZ_`@OIEZp>8)=U
z>0@}*xz9fZvp!AkSVhf{>X>9XMX_YQGCS2a52wE3(Q+>45Q}YamlV-jC9RtPZhXX2
z4p!!yma{uJtg~3CIh4gCl{=*oG3c@XB(^?EVpXDc$qkXJwj*4GhPh54C1HU+iTfC1
zP~<t?YGK5_az$EPTs*arg4H0%6Q3e;bGi{E)|BSy8c=EHe9(46?HVYzyalwutQA+j
z&8m_ra=PmPxAbPGN>q-Ecc<e83A;%oP<!0-sgDP5wixZW)(L5tWG`}uSCf;-<RkKR
zxK#HgiUex0MF0{=7MGky0pxz0TU&df&c<Rql=u7&W;ly}JZ$MD7?(VJsul}LaCCI$
zBe9SLKkl^-(Dq@4+^MBflf%tItg^|;NpFAu(4c*J*Av2$`h0_Ir3}naQdWbRH48Mm
z>&dmLgfer6hr-*l*$DT;K>r?k6$26~kMP7VDvPfQEZ7qb`ak*(<!dCAjExCns22$e
z$t<e1*)R?z+PAx8uJnjE-A%R5?%{g*Qzw89k9zFB@vGj>fz$BA4lUM-#3WBir)Sly
zHO=#F7=O1e-PwSP$eUduRTYv({-TnPX{>K7rUVq}2bqsCk)njM8;hL@UMHugz?eq|
zXbA}cf%$^B5&THS!^KWvigFFHo_H1WdctrvW2HE$!N&S}eW+_OhlwlY>iE{woY{{%
z4r+($15Kep@6VNpwo#x&2xrz7t*$&hJ+<B+wO#h-KkI-Fzy#QUi{XZu)SU`IRftLJ
z(x0h{qye7_(TEOnSPv8sd@1-VI_%-s?c$b!K7rM4Lj#H*rc0xxg?xuIHzmG5iOaLi
zdSm8@KL&{`Vig9F7LQ!{lf8w`+R<6GjT?R{rE`2ozd~G|Na3lA*!G6%*nU$pd^9@z
znnX^RWbB#4FTt^0eD;v%&iQ8%Nk>a-`-r#7qH<*AkF&Tie2F$3+pP3^&bR5EJ0y=P
zg}NlHlem|~q}T#07U{}I-Wc49eSxg^^~7b-YO1T)^Ik;Ft}C;s1lJ=G?WOW#)3Z3%
z_S?Df2S-OsP?8QwynozX0mZd|M8~z_WbtW#QufT>L_ftsW_C%*QOADLqyh<{SB?(c
z5AK64DL=CdPCew|>TqFJ-Fteapu)pZ)_xUB$)aU%sFH}#95L+^!3jlqdb7h7q1O7t
zgK2*phH;1Gj5P)<*HDEgE+IB26#_qJA3F~)FizmI5IJWL3YWwcuq2fe5GFoY-e2;1
z{G)l%OwBBMwKBcCrVyEX<-*s6D8G_oZHdUGx+F6CAhBsRsiHTSC?<Vf$R%+l=u6JH
z^xr2P#=JDU8uPA2#L}Kl*=Fn(x&UG2W3jV2{X<QYilQP8I4L3$;Vja!va(?fCQu#w
z)^7^Mod7G)%OZAMdn>Q{_6O&qRCp9YHHz!k?}-3oBto;P0F|JCM4*)X)ng6tlNW%~
zN#wFwNdfhB@JC^-b034dN^MqYcE1OE3}cMld*S(ed|W+%1-wSmj~{aqvUOdXz{x?p
zvjK>f*N_ezRmDSN%(h*W%Q&_8MKDEMJ}P8zLm}s;;gN`GLCDg9;eJd+{^YRsJTK1!
z<lfuu2A62!5x#MiB|^m<`xB=`-Gs!J`$0}D0+kgf5>5XOk?g0X{Ovtm5#Hj)B=ePG
zBJD^%I{x!#m>xl8jOh>*jw(TIsdF4T&yYzzQxxE_oD&!6dfLt-Gxw`i$MReM<42D~
zjvkxQi?)VqtKT+eAu6{^OX;m6Mgqp^CsuHKM5ks$)#V@p-vzT}0QRgIuG0g*>5=Wi
zT=oy!Y_RNZA3yGZcNPf<2n-}7Bus)5+*D-LI?Bs{bB^f_m@HkdpxympgjN>QJ=qg;
zb8hZ*`qz&)QEw{n+>b=tXHR!61{3XbbO<XIr*lO3TSAOmm620N{Z0NNZ>_I8RzU<N
zU3-ViR_}%<yQ1(>OHReUTCx}X@Xqj-kJxnGy4nvF!;qoI9a<T*MH;v2*yLyj(Et|c
zSQQEAUZhj#R1OJF)Ad&kI?-T%l=P{!v7JhUw9or>h7zktQkpK*z)+?b49zaUk`_QV
zBWKVEvEU%H;EI}G!Jo?*A>R<S=VjSaW_(CVX#nzk^B2o_9d<f|DV$|D)**c^i*e~H
zmH!(wSqtI!`^!CR({Ft4rs`gi?>SR$(MXj~WR^-c?Eg4$AODxOa{i%LuYH=iX_k6{
z{RQ8y)!m08q!rxf2AZhndRDJbI;vb9_l*mwSjA7OTcz6O7$%ksdZ;>N`>9lvHvfd^
zhiT<$!9#tJ*5xvva(CEWN?PhoD|N1bcttD#ywKYfCMfn>9hYf$g511F7ixG@mOwQ{
zBZ3NqOo}Ei^`=O_fHWiK2ir4j9aA|5^sI-zWSnmRj?o!qJEpk81h_V&_w4l88Yq53
z<!__W6!Rh`&7~ccLSr<8IZNsfOv~z(y~_?K)VYHr-UpU=h-p@6$0SR!Hks*=&wDtY
zL97R%NlaE#`n`P~dgxEAtaOx>6kg;vLA1Jg?u_r18uaO>kI9cp<Gl?5)QoixfPL=p
zIVm{qul3HyAPgdi128?bXjkw#AH53TJD9-G)*=|QoP!Wu3EKCI|EMaVXo(&!0%)oO
zavfq;l4s6f`US_%V4s#GbKuhyy2=NTu;|5RRUC;D`3M3=ss}(4`9_1x01FaRR8$P-
zw#`KOj_5~dDKk^H<gKM0o%M6Ox^4dB_A|)Iww0NAv{_&e)$7|=Qk!|A)`k>NPlZ-k
zJQvdWb2!^$%7(71WKb&DKw5I^7rhc6tLe9?PmSZjO)G<J@rTAFl;YoGA+IZT^mxRo
z5Lkc=A^gfw%0M{{D8SxL?xyytjT;zH*ah+)l{tDJs5q5xHPZ$pf>3KK62)Dp5y)0L
znhAiJWyrAeCVRXnG#Pz3%liZHClF74eHg{(tklkTY!rvPTJ{a6Wk9seh%F6z*$qaD
z%mu}<s(iE_&KcLG7znaEGXA=oK%gPpklGVJFT!EzY)w4>Es;cL@pVYo%{PRy<x{qA
zF;NdkFgH(;GEdpojdbs;#INE!rr};qEE(q%oO8{dlCt2krrLANh-6b?6Z7=ML*zpD
zNR`Y5iKoGNetYN)D;AJgx>h0v0rM=|xV*Y^l4KN@{&a({L0|eHXCch}n_5rYZU|{+
zre!h>1;tQUz<!86Sy15(DO>vQ_t2QHXjUxjkC3D0DTGlyb75iOd81@Bk0QkBG=&bd
zAC)U<fVHc?D@lwM;F$D)u>`_6>-i0|pRpe)*E0*a%|tCYv^S_r+Fw-j;PwCb_HZz2
z-tP+P@lptuSPfF;&&S`rO;c3IIy)1xY+&o2Q&*EWI9apw%biVJ?(i%=GJejF&>xOJ
z8Kk>ivRFZEHD8<Mn6hPwiG>w&Qw>T9wX<C6uU>t!Ss94Ca+5LzXj+&_d<}qT4;AsE
zybajgZ#(Ujy3sS!(+Q{(xZI!B510|zN*CwbzCx@<1Fzw#QQ4QFoD4cOgk+{qZtcLf
zH`tnKG8it%08%(aHcO2cmx50PAYq+QlU%5FoF+F>tS;5>rW^&oIt1+Nal^2Ef1>Px
z1}6(zBz=+4cgZy_%0~*@Lli-89Wu4dsftYOer>-BxIw2K^FaOi7+%CZ*7j|A4b)t?
z3bPpZD&DM6%VHcuL|a^rm45M;#ST!YeWkuDi;l$Q2S)^EuH*$;aBa(>^%@U6uce5H
z{XW)jU$kY}8qHf;*T|q&Rx~gSibKlrOeZHNJHQ^X0GJUfE9EmvLs(9_%@N}8eQ(;#
zuesmr)#f4j55-&*C8O#Fhsf6MZZDWy6{FW`DtE?eW(mu{Y{^<#S$&(#4C?=c90rIn
zbV9_nup|o$Te3<zuR5S$W$1B(+tOte&(|mV@%fRjer9W+Ua+4IZMqsOrqIPwHX_uM
z#j3aq_jM&?Ii>FFn)LcUX!vQqjmOMkB+6YZJgC2k?USAwjbYMx)6^*qU)MCtYSKhI
z2c(ne6?0zd<QBg^z9fssfMiUxwzc&EpD^~>om;n3gvsq+W+>%%0gVXU2DFlr=Z&9R
zwa^zX0FVY7kMbJ&hJVN@w|;$)dZNdbu57M1O!PzydID+~yCJ7sMD~``CiL2PTcOr3
zMAcp?_o<cPY|rU;W+lAzxr7>VN)HK}aT>R9c^rJ*OYTlxLDH=b53!o{(!Uy~mI*z7
z{*h85JP|5}X#kglse3Ca=?6CbCv<eH2Cx8n5Ox*5ymx$2-bY&qaKK3i1r<)Ksi;Jb
zIV{ur-eP(V8le4fQR#(fRwSU;fWEiO0*%-U<~7yD>gl4Hjle5qgNUv~YijH?c?tR~
zWI=Z3o=|R9$(T)X&~gwT<9lR-AcBQf@)CvHyR{YN0w&bB(a3=;g5VZ`Xp!<oFlKP#
z-?oN>+>%y6wZ>~;3I;52A^_@Cop*a3b~=T5RVlqE-&8^Ef8=8e0IB2^28vIqUOa!^
zw&;N(n2`U0@tW}V)sm@2Qe>M!itpi_-!!9#pXC-?%PROz$twkfEI0X698bpYV!A`e
zc2g2=pk-8-p<|MGX}sxL(vpN>=$g~tanYp0xzYz-H4>oi3uiZF0F)m@`4TTYuXX|+
zGZVnPxekRlEU70(vo&8-&!VDsd$`nzSI+{^aVhg1#Fgf!X8+vM_QGyJ{XO1XnIWYb
zzGAPO#B-z20HY<29p?PXvw&@K@@qtiaqIkd-TVa{)?IhDMN%I+_jFq(5!;9;cgd!`
zb;fmCn+xdud(U(Y1&pH`sua08m@8~ks`~l!=OchVvYJgiyR9d!kXlTzLC$THK4GDu
zfoObd1Y}iH<73ls08<t--*Kf$@2~Bj%jbQSFCw)X#K%Dul}F%g_ERF<<|RqFgm+;|
zZ*XVHUT>*r&&@)`_5zZVxld*NNg`ND^ERTTPrdOR=<yc&Wx6nw69;5_KmANl<|n6@
zag8yw{d;K8sw7klxTdVm0zXAEHnAflz7PRrietulKRvj`vXC_VHci`^-pR0lt}-&n
zg_BwkQ$*rIl9@X7D8rqm5F4FWlF46m)OQ_LxON>D!pq(kKkCtZW$VG75=RkpBu-6w
zBMCS44LyIV#3usHfONkyGkY;5wcoU<tNBc*LC+AAy;j3V|GRk4Qaj%wB+rt%4*SLo
zShI7_!jHTP&T}NGk0r&9ncF!jQyfy1@E6j_@$ud$jQ(bB?anp~5~8<z`)i!O|Hr!g
za$P~KhuqbO3nv7uu^9}c_Q6ZmMsMaoA0GARwc2GnbXl$s*Q=M#dwiG9XI@9~S-qid
zl%%DlRy{*yV{WFBv?9wP2K6T9Unte<M|!20Y<FdjQVHtUIe+9GXoa1$VrMB!SP!!W
zWW_Y$?qB}<zOhb4^b^h}n~gXQt7j3dI<%eU0XwK*B<&&ESewS{e9HW{86v9|F*J!f
zMJM+!1RJ&l2lewu-g*A_sh?@j7r5>N7RR>5{9+jz^7lQ_UNhG-CjIry@UyG|L&k)}
zt5;_7cmDk^n(G@2>IqyN9LxXxi|fNhPlo7Bv<>n+z5jj(jZv@)=fCgOyj_X^`?f4d
zjMol6d1SU8!<t9+`90W=5GDo87;Xg*YV9Frf*Xk{rK)m1g6JR6nSW0#T1*HGa&lU5
zkp(0I=nP*TL#m~@Ed-8wQ)prTTcx0695mOdUmySP{Rz++|NAIA>I-WBeY^37<;n$D
z?-?sdeZ9DSh1!BE5<Bzyk642^=7Wxw>wkYMh#X#t$hx{an=A3ZPqjHOR@8qS&{hoo
zEhT(^H6^o9HIr+V#6IQu-(R2bh4qXjQ+#T<)42RJUw@5;^G;hCEh*+J*3&J9$4bHG
zR1&1R866FZv^{&_w^Vu8Y%txi95esT&jc&#)sczh{28Xgl3l61jY%5ni*qGNmn+Ge
z#6@+ip~OeYNv3{=$BfGag|lBxj;->3vZqal)%;yG*M5yaRx~QCkR7AeqW#k{v%>c&
zEaib#hNCr;nGZd}5=jljVr&B#d|q#LUUcSt7tegNo2J6--@xDg9XNQ-P*I+1jMliJ
z4TT{!q8lGRKTFs_-iW1|S|>%l`hUE=1yEdFw=GISfCNoKa7%(DK!D&5QQU)TaCg^D
zkR${P(zpiq;Lwc|T!S?3?$R`k^EQ0v|L?g~@6<i@PQ6PtRZY>o*4k^@m}Abp7ks)y
z6q3|`$o7wwy0L(JX;+?6dS1CyjZ|_jKhlm~wtRWog(-P$&cuCxFr6D#VcHerP(XM)
ziJR!{=RqPN!gZne%)Q=Kx@SNfTJnX?9GTMyE9JQTl)~u4{ky&tE34xbwry2)KdS%$
zSw=7-tn`Cx!))8fm_7?u^5&mUX5D@gq$Ph%Q)<n&TbPjO8sJ4j19~>>HpjEkZ3=D^
zG&E%v<CE>zbX^veOw7zZ!^Y8I%8YP&0QvgKJyCtaamPVmd~nlX)s7FFb#2md;M7$z
z>rZk&Orc{A>1S~+83`!K(;WzJ-N84V9ZT(KSX6HBn6!tBeZz;n=TT`OSF}F_SjVk+
z$drFvZRR>8nbz|tw~w(Z3w@hmrAS;-HIKomyB0mKY_h2AwyBPFm86eJO!VyEBKFPy
z=35;Cw!uQ8qLwx|x?l0e>V3HGTln3xkSin_u%Lkt)4_*aodTb`L=e!=F0B;ToXox~
z8WUav9uLF4+e%kfJ)i*y2fkzn<wJ5sb_d5(?>pDI+a)@=XmUB*_}{jT)hW^T^kGfs
z*X8KS1*1&<>4z8325qVd_wTAu-mxXaz{F(Y;1~k8MB|N8T8?xsSCRvhAhQ-jxkiIn
z13jyWYNASQ6;Q`(Zc?5aB4(J^@jX|IcCW>pH^5I4?#GUP_avpr?ABDlGsPC9SgF6<
zxs$@0{tz~!yRW^E<nFfO@T__6rc^w`TF~~(7Y$I|*Ih1ibii(j>^_+c`~iU^vqW{#
zMX~MC=W{G!%wC%s8xwhT^YZc4C|%-jtgXGEp%DO_?y8W@EuhVdjV8kTVZUj09FXD5
zf*gnKR;p8NUv72bjmnK@v3a5?_^xt_D|(S-C?%wfz^Uc=PQQX#vIKZ{$ecIEZ7S%?
z6bL$=>B8Yq;X_t0_rh&|FLy56!t_{4NG|Y?$H>(Nt)Zz8Iz6eifXtT!wv;kc<zsw&
zjdQX@35gz3XYQi#?0|nJPV0U;kAcR^nc(f*8gRaBcoe}v6G_YveX%?$ITUai*__sF
z8~Ri4K1KEII(-)(?caUFZ#G_8TkN@%QYQ?bn@A;@WY7#2eKyJ<%TE_NMXNzc{`#1q
za(%4SFaO5hg8p~5du$k^iBnfdp<|d{nOu$^k_=%fPfhg-6_bxeCn%dN?5<pqiUjS~
z$O(BmxFN>kRg`B#a~%^&LMIVSnANZEem%P1Be+_jGMU6sBp>k#ALH!Y{4DDFZ)m-T
z1eSK)d%K`EgNSI9_U$s$%AU_-_&PWl$C~Y=09Sdl@!aCq=cj7Mg|ZRw>?Z=OO#_9G
z#s~%x)pPlpj~1wNpWJ@UjTW@<m!=WVo{<A{_d{Xlax8OYvJ}U=!$HG2D|P}ZIhvUQ
z;7wFM$pVZ`1KS%)|2VGLXW(Smi>4eK-K>A1WzrQn(zpLWQ@NK|c;`Rjk8dovq}y50
z{;M$ZU$}WNg#Gg~p9RYGpSMlYpd9{r`+po5K&$)zY2^Rjz`*~9HUIZU{_hQZ6aW8n
zwEn-n8NmNkfd9kI_y_;|7trPZ!e#*f8)*5zzZw65|Nf75@;}_>TWkD(+huV6|H=Rx
zQdpakm6Vh!+z}=>r9d#9Q+9AD1J5HE0S~yu1Me*WE40T>LH}4f+B}a6!@R7t^f1T~
z>#U!FRZte)I`v=@HdT<$;sT$ib(6X(|AV2?&~{af@9KgSE<i2pk!w5G05V7MEv>C9
zOG~r`qxoQNX4w+(KojOt)NsBlI0N3vo5K1p#Hp=n?6LnfAYeV-G(`~{TN&hc!LiA3
zF;7oV^FWTsV<C`)P7-`9dE|@@%C7QX>w_{8?(0hA$R*QB%H@38*w~;h#{3ul)^`1S
zuMYoP#(Wvrt%Jq<AGV8z%H<GtV>1HlM3I>wv&a1O>3m&>W|69~k<k<PM38enxt>Tb
zt+wyp*m%r@_iZV)gfYCQt<N1EJIB}{&+39vf)XNNwJ-&c$#N+KB(RGyjBl-sslg`k
z_gDkQFeD<~05-)Ho%PE`L?JIBgoTX{#)GCZ{PEE~I0;y7dV5o4h{K_7^gvn&pDstO
z2{UkVev*`ww8<MFyeKOZ)_Ip@SgvVY?g7PCCRPn@8>UZ->|8}YHXR3%Pjcec28Y4}
zbq5C_ox4nreB|ClTJa=;ubiGY{AJOLmw-JE&@R+wQ2LwXkwTlwDR^t<xA(Sn2VX&k
z-+9J)EDubjQ&?{qHE<KBe@T_hI}0Qyyw>+KFf<&^5GR-c8-VEr*j~MoNfvZvJw66m
zPEZo@G7PQ|%>}O$?FD{H{BO7-A{=hehipGBt1bKbXEYnE6Bi%LGqVV4aksVtqlI$T
zc-uo7vRfmdE*Y=d3E-S@MRy>c2{UU2^VleDH(dUZmy2m#P}u3b?`$^o0Ui2*`8nTs
zA7=|u80P>64SESK3Cu5Cog7z)hSfn%Qtq4oT!vl$9{6XIXHal!;OzC(&avMd8Hlm6
zJgmV+=7=wI`o2s>-I#T9Ilw2dSev3+tbJ1!+Q`dEVLHIkbaJWy%Y9J>9ic%%iXgoK
zs{FKuLtxqYxhFJ0s+BRPth~&aBYxSE%LAGu#oS>rdhK(iVmpvj6fiRKJluVA*nAV7
zpQTG<Uw!7XsoqQk?YAgr8jC?4SeAFuWhAB$fOlSRRg%qZA^;2=;~ti?3`H>-nwg=T
z>h>A$9JO!DDnTnUiTFILbu|;<+SmL_N)%0^CR&M_qa1czG+2mFPtUvF<nhmdHt`fq
z^21x@Xh+4K{w~IK^NoSMQIh(?ZL8C>v+iXr<2l{;kq3A8uFK2Hq9G7^No-u)LgmKy
z34N<lM11IWKqJ#B;Y-nJohqyNo8musN6n!DX3gh;(j{_%U6maH4;f5JZwl_gG?P#w
zqDRFF<&rPT%PQHgqJ5^z^(<X?;E(~Kl1JzzBjOn}F{vaaPOjhf)QX3`yD~6wz7T(A
zQf^~4ECUXp;6M0~jIPbt$YLVG@XUShj(53R-s&qrFR#l`%=}Liv#^f9t)+tSnRQ-R
za5^+B3=VdB>{Y3_HMN;7HC#duH`byf<MTv9H!a^a&RVEtXr?dOSdnF^dAN>k>v$rM
z7KswBI^OMw+P`aTxcYI?N3!;#nh~*T*}HuGLXv!@GOFDrT+!&wf{KW1g>_cNm8$Fo
zAgF3vqEUi1d;X^5F(GnJ(57HL^49!)WAl`HG?V3WX1!55U_Xcqm7vQAc%YU<4NlFx
zk;9L#bw~M19Gu)x%NLgp2)vJ9VW(cWm@`ArK-D?#V1bds;uWo`jdTJLrjz?iZAQ&!
z>cTuLhNonq3H-|z^}{7<5n%aM0VvmCLqe4iw$z0W!M6bad(K$alBOU3EP*$l<BqLQ
z4GcUWqtbRx7i8FJr!lmTmODN<zhYOcuCPh3K1W8D26z{41>W20&>zclvHRJeS)txo
zq@~vnQDxO24&DX!y!n&lF8A_%$fxTxf(CY`a=$dx*ree`QeW)@wDegyv3lvZTi>*$
z5>JeGL%JWXxEIWTZ7ta#4N2s=4OU-rL5@#Z+1yuc(q;c=gHD<9=-s<q=AY-3Z0SGo
z@~Q9X+1J`!9;gY6?%?44*zowfJ$=@g3Mfu*(CN(8<=NOy<&UiC%@me?bu}y4(5l)>
z)^OyyYIp>!DA3=|or|5{i#=P;pDYJzyume$RFoAz8a@WFM$ie*vS_9oZ>I~^eB^Y_
z)zHx*pdN}r$C}a5%{pFb5v7;B-f@?V?{I4n=Ov@_t~U#!tTXv%{|uDliR!>$GFIn~
z5YoX^FupiL2}lkN#wZ$7w>~wO=y?fP^Fwk^ffAJt6WJMMofYSk5k_lKC-pws*3kwr
zBuDrsQ9AXRO*nkc{$R51+V@w6)a>f*;Aq-H`ODb4XvOw?$&-#3HP6v6M!AG~vr1X>
zw6r2CQ0hs7QUyoH)Pk&%Kk`0Jt&|?pz=TcP{~@PNctJ29NS9*@yd1Mp+ev#9+mY{V
z5^}O^6x(r#`6B8QU6*XQ&QH3H4@!|ehW}hlevmK8n`d8qd?GVtw2FHSjWw+!GoJ#Y
zH{bYjbsC=Bw^GVsttzQx_r~{s|IQsBlaKeXwmyhumpSU-xD#Y%N`epZD_9L1Vgl+t
zOn2A5L@UR32&nIme}oGC!$0IEfN=gvp4z!f+2oF<%W|Cb(uMu}Pr5wUZ!R{u`G;HM
zg?WDH7nO@<Nzw&wJ90V1ltRxl!o#7}<~!}!w+nuw@~#OdO~RiIrW|>6ZBQNK9&!>Q
zb6I2Bcp<}pw>~tXlciz-_%K9aw*7hef>%w?W=3Pxesy=UV0dC;bF?A(xu5@x`m=wu
zL+WNZ8!;mFqrXk?E`8>Ig7ax-v+Jr~#g;ibA_twvUT*h()hF@Ku_a@W^e{wf(wi8^
zo?UPisIV2O(@J_A!xN`9DgGHu^JcKFIV*S~3YcS-gs!cLR`4=rDX(3#xdb=NR(;}b
zX*t?1;*Av#f~YzvbIy1h1N275TQM|%id=F1pS4L!{24*u3x0Oy=B5oh^p~R(zbf|a
z*mYEQ*nACUyxiGB$sFn<9bZAc+O1gly^j3WmIWw1%un?^y#8{;dxa(-77_W>#DM0V
zQB~F`YIg~$MDEdiLwo4qRINZtihZ-=<MJ}+ne9|AVD9b5(V~oj)GK)m<^Pz444ehw
zasK6;YxB$!5}lm!mJ=8$61~cYh=((eE%2??JPH2DS+%EhR6*up+2;{>@1EJNG+VAX
zjE6dZy4>&L64|rGR-jb?#Av!MW=?vaL9Sq4mB)1SBm&2urSfOh>*cIPE|sJSXg*D~
zhaOFxj6ETHy!fU2A64W%Kswz)W8&4zP^1Yt6K$)oNH6`Av(+Jdj$BH)#71xy*sI6A
zyXckb>>j#W9`PHj1{EoR!z5ah*J^%YQ@Ft8LjZ9jr!Jcha0L-L+c2#!H}Rj<D1Ynh
z@2?`trH(0SPZlqb5W_0Tadu+1%<X>!M@KQyJ#8v)yKN~#D|rK|)`#8gIr;Y<+Svmn
zuZH9cu7`dUO;wyKziZwZefQqUxterlwxpzZeOeOJzuA9OmHNu0e3sfhRcyvCIs57i
zwt9)It^#NT3Tr|DO`eKeHTf~!Nm-PRuma2080zysxUp#p<iTE+J_UeK=#}Qp?*tV-
zeP%Wj3#WAD^A;j<Qi)toYOwG*QqZ~w2Y+MQOooFp>G&z-MuS3>eN!ifS8n(0Hm^>T
z2@w&=r+Sn=`RGv!YunaIJRoDCT@aJgbE`aPDrg`js*DB}hC<Jex7Hf3;Ai!o&d(V&
zKwRehoOU!%<CBi3Ij_zT6yl|hp(CNG8Y6OY^`cLwRH;KgRaG_YV}n@6JO8^x!@mN>
z2x@h`1zq-?;vq@~770G5o<+oMHbV=aa-t(s(mMwqau=&eqXSSEc$$q4(Id+34UXG|
zvVxq8uS1o^6uvO0a{_#OT4lY81q**O97%Bm@158L4ZnR(P#YcUa{LAY3FqcjBapjL
zn0uwmbRLg)aj0R%W4yX-HmhgJwO7Jzp^))a#Pg6Hxi#|lJ*eniWMm$QDUGxBoK)k|
z3GA5$z+uY8`*O_y<Me#%fWmf=kawoD-REXEZV837B)(+1`r|%VkJBEq$V}DI=gyu}
zF`HlnBmerf^XZ10Sau8zf3~gbHaz<h%L$2K+bm7zUGF;=?;JBb+NRo1YkeWV5&1Wq
zk)2yMk}FM;c4nMOirv|wz(paA!<vT*kdYjw!6MiwD1te0J$s9u>izb**8cmf$Gk3q
z+MY%tie~VqeaU8L-o4ZFGYhwGJW!teyh-J7<(kIS$TG(YmW7Jd9pSVILDV{k=gg1G
zAZH6J*ZChWJyKX{ZZ1PX3QQ^$<fxMmOS5|uOWs74%x(d-yB2er?Lg({T!8FJ?z$#7
z7XyQ)qsz|dI)?Po6-smybo&iH$9cgHOHe{dN$^?E*Nhz&D#4xiy*-db&`g^h$FLyW
zeKinMA#K#<4PmasQ<m_p9qujUZqgnkDI#5-nr10|Q<&LhPXy_Gr~!&({VQ?b(JWED
z3QCWK^!erG4Vt6S1h|HSh1vds^H%K7KSH`4Jt%Sm{4}|9_tH0S)xmgSY2z+h8q?X)
z*c`X<td|PT{lOtYo1uMtYwSRnDW?Q9PsOpQZp&np-~qfhs|UI2TFzWNFV(VOmphf)
z5p7tRFMbKt^cj;4Y?%RYcoL7K_IA%q_MejBVai^617ET)kHs-mwB3`WB>xWDmoS2;
zQC@FSWp38aIo%F5NB%8v69hJv;8~aSrbcF+(j_msIm`W+ZtJNv=%BsSXn^XSa<$9}
zj@VbH{}Y5>x{Hn(&s(@r?UqFITLo&3S7P7fQeJgPF{j2hB&FZ%C?#?J#QS2!8X#3K
zXn<;xL&_GpMW2%?evJxpxE`&p&Zj`F;8ID;jPw<0t*^|^s0>xkiTS@JEUD2(PvjYz
zU9$Hug(o=_-LmiVx}c2bP4@aubvGs^pw)(2NGpZ2lx!+X_+SS6-?G)!O(fYI<b5C{
z3ha;lQ!6SF$#Aq+)QNVj1cV>ht>JHK!?Md*h(nVFW@hMBHNnM>m^NOfD~Oge=>CQM
zS+EhwXif{qJc3M^jbP1PD-WKN-s;vVNVNG<fS9RI8>$_7bd;sjBh+s=^UIir9C_J2
zTDGzZEvS>)iLKnak%KkT0?Lpt8Q;FoPKr&9(&&kSl>^|v&mjfIu6#~Fhf6;3LxpU1
z@_?`F$5kv4>q?8OIm;RCXE$@D)!h~ZZW4lT^2th0MdTI_$(2^^TP`m~U~h%9OiV0x
ziiK?oe;+0sPu-OqLiQ!bcwlT!6k81`_%w*MCCfq0Cb=$3ZTvIqI{JzH6M3WPE*ors
z`X26EJ!8VW5#`_Xle&~zDhs%sd*Q`wJ(hesOelu8bB}kr;fZi?@M9wmMkXO`Fnz~b
zl&i_8kYc-bU)wZP+fNE^tnm=Ni4DHIv^SiwluXZ-j;8SMqf5i<B9G6r;yZ*&tYJR6
zdGYa)nVZH|>~>Wb9X5Lh0N)LK3Nkl8L~%298abFJU3Bbl-B!%dnk-K|pdeM@$VC6K
z)DyMh%{61AmfR^(X*r}WYcb}qCG>zXFQL>e=$KM4d-(`=v_$beJByZ?Sqb$a@}QxS
z0_>`mJvU7E(Spf|6Fw#~<6+MC(Glix`aa2PuAN^B;v0HEl93<&+%|lmbEG}$!H<F7
z=++J8owTPQ&aLls!Dr-;sbhcdC#~XNjV(gBovr?gH5_t1&h6AHWnue}`blPMe0+CT
zWzXE0`l|gQP4#~8GXHH2TIJ3iw)vGLxt!Hv^(<85wIZ-L2{veHtjF<$l3RcgmFqC!
zEysBba5;4qj1hcvam41t!Ke{v8UaSAA7K{RY^m-_ESqD@2qB`CrXV*wkKIOrhpxW{
zFEuI+a~jn5b3l50W}-1w#Ao{vSttp9Klw=K`S<hUVM&ZJFx}{zE*1Y|z-u`Z^h@+?
zHr>QIbbCY><zz-X9RJe`wXy8jUbwlbQc?_1eG4}m+n|f}=+fMA*^!rwpsTb}nA2HN
zvB0x{4}#IELXc6CO6PthBDI>N#}WTV7xVAl)olu<5^<KQuD^<b?@qEv=2KB-`h=3_
z>YIJ{^2Ytq(>d}ZqoZ<Z|Itm-<$cmBpTK@>-d16ynC~u^la+TKcfH;eRHyqJc*_uV
z#%q(7``s?^B;J{cQ@ZpZr-Fx7cMG+xGIf3}kJ9CSLnF#Fxr3U}5v$`=)MGm<e>%RS
zP!7Gp`IY|u(NT#7?-|X{5c`Vs#%t>c3>;+NBE9C}t`;;?r3Vo<RV-;ZdLL(4uS5PN
zqoBAXKR?uQc5l*&<87oe8y%p2lN+}VekL;-EIq<y5%tj4lB3^@0OP>Ine5rwS;5G2
zj&}gH$a64LoSCX%ox{qAtf`c@Wto0hpogLv8jf{gLGUq$w=uR_d_w+Kh`mZ%kUjh8
zI?`=Dh1|M&&x%pkxGnfna+7AJODF5r)Ve3y4*E%Q{os2O1;CarX->e1@=ab<qT>@^
zY&NYRg84q(&~Qn9qiqE+|C>&`iTJ~Lwz%<vtI&1eabNk?x}q}$GOp|u@hC3#2;iyx
z3pQ09qxYylBRt1+y{{3zQQkL&oUjGVwwnDl`R)C@m^7xo7H3e8W_zF0OSWoqkAWda
zF}U;lyN(GUS<buBGgVjfDi%L8Gk?uc-U7VCZCDunj6|^NZ-@AM58Z5epHo?RHr@2o
zb9|;Y1I00c$|H-t-Ob_LFISgD4pN#FBqby<MQ~8yqo(avGw^H(M<Ea1hJF0q%Zq8Z
zJ?0X@;`JP?xztSV4tcY_143wpZZbzGr@eI{ojIT&Gx6TQ*3oc)+GC}__UNF!h<v#6
zQ@qCk!^-k{aFO7O;p_FY-PoA-kQ>eOJyfVwB0-~f;CiO$uji%lATJNv_(||4{^6Mo
z<#yVsMGA}+=cR(_eD&FJGz+hqYa{ntOP5{fybu{Ed8=N+?lSd}dR;1OQhONn`Up@A
zey3XtY@d6hcj-0!Iq>w<9Q32m6(p|?v}BiO8!}dDKIao3Dkj@#@}}8h;uk3%=FCWb
zEgZj5_Qr`0Z;V{-)t3}u5hS~Y-6TICCg*h7R{hcTi(N<1e1ocWr20F!EZoQ&eZ{e%
z$dC2V$$zC)oNYb9D(L|%glXx(T6O=7b%<A^w!lOQOixypC#8kk2|SN27n*4~K*oaW
zOdmz89#m;PE%Dy-SiwsyTUV>HYGh52Ubxx`q69{+8jGfn!?4?^bZebGAZ`nFTN6%}
zk^2im1upRN;-v>E;7>RH#+lX6l*AU+)<Th$ya2?S(S!E4*I46gIdct-P4)7xt8+r?
z+;`EXlnt-j@woP~w~RT?*1tgtki(meQck+=TD8_NA5Dl_56Kwbh_&^=oXs=6-xHa(
zHl&$fWGT8fobtz(<dab+V%(PbCT?+{4-|9V$pncKroXF>n39IDj`b|v*zOG{rdKZ|
z0BdsYSBjx3ZPDx*PblY0LSGPa9aeuV<ENw-ah(Z@g6_p)L}tFQrOy5gTpu!(D|sw6
zEMVi|j{K=CJu}z}Q%XKxfQFuV_2A#wxU}nxqR2m`&~6OS0UKYbH@=nTwN*rP+Iia<
zr#@E;NEygwrE->4CcFU`%kF?EiRU!j|67H@luNvq*U=81ztm;Ko;1if^wdM8L5jxC
zfwABfO{=cEj+`WcSoPRX;yx+o;=i8<*T(RbjeHw9%F2<u+k3&nF1tTeeh4O4PmG?3
z2uta#mCqPDxwoE4_4R^ag6$>pbhZ~vc_Sa#cC9aEXn=rqQbTkDaPw$)9{?$gGm9xX
zor6AY;|T|6_l{!^=Mx<wW&4#!D%D{P)aTJ4wIzHqmFl5CVPU+snu1EgEszao)y_AK
zDU9wK%bj2o$$^eOXU;XawGu%PE9jbZ*XhwovLWj###JAz8U8$)T!~F4vusF`GmQW<
z#fhW3n0v{<WS+U3dc1{qjidPJ1N@<`<><$i0Wal23M4u%fdSCd69zP@IYJ*T?5X`e
z9ZTdiCGwn(PF<6Z)ZO(q4^rK|nrdtmBS~D(HPJjeb~{a!y&)?my$-Xk%FSw3se+^0
zQpT#985vn=-w#(?sST8vY<$rp=`yIc{47Jl;sqd_2WIaOK%rV8K3*#yZ}Q}pIf{|p
z;gX}>xzb5*0gYqMxh<ERPg9TOEHCu7MmD3^Zk$9uG5N2#j>hbiok@=q>YIy3lM0rF
zqp?LAmzo`9_hIuSI_V&?jj3~-D<zzZbw1sM73BOgXS2!7LH2^`y=c0b*(#NIVC?|K
zZJ}PUno!{rHG6#}?v1)AH>U1Uk@r4~r!#*O^);XD2^iaLEQgf_ST*vZk8nX$L3T}4
z_983Ca<qynpPDp{IgyJAAp~YMdG1CvE|em_&|fT$R><vegs+5pRN59?>dB3PS&Z3C
zv2em;m<AY^rMqz^%dKiL^nnsOjI+}X$b#zzm^iuRFH0?fjK<k{1`4@7vXOuqBNuM8
zf~4qGNrC8B|H?<ItF+S|PUB*;F|1`f&8rP&zDL@BZ-veQ0<XHx-HaQ|Z}BEd%viJD
zvq}VSi<##tnrC-P@8!HRAEnelw-RWqtd%-A;4|V1O906&Uh|$W{;Z)(`P_ntyIUZ;
z3*(Y2N_$h8N1aC%Bzb1JR{V~cx|&Mp6|6PAhhikEG){I;Dqp_!Y%($Wm+U!8&k1a7
zAN_HEY?EMixZV34dB)3G>D8l}Wo*pE-|^hP`Kf4>q@wfII=sRAXoRJ;yFwG1Z=u+j
z(m;dq1HhG)+z{4&cofb@BaHM}rnx3=HRPkkSzmMnw%T=tN2=^6t{>h8&V8X9-0>$_
z1jHL8kGBuiPYshKq*$G%4mc4bc=$2l!Ox7Zgu2Mq_A)eQl{h%)Yxhd0BH}U|U3XcZ
z?59XbfsoPNkobyRN9I>Xf_Z=NfI9%iEqOvUD)U#d?`U%$CaiNy@T4XEo_>(~20y2K
zJf)gmpO3IgK;@YIa3x$p*=Nge-~L>vxZ9aen_|BA(rICV&8~s{>Xf`zKjng5hqQIU
zuI6eS;>7AIR73xB%dS)LEh;ZRFSq2j>nd&Ww;#z%anQWDEGHcL^nh|zL42COUCT!C
zvJ4<&F_=}PwRP6R`ggtu+Ow@yX4@tZ;SQUkk$3gxwz?ucHB1qd55SFF9O4994s<lz
z5=G@d*lnm*_%u2|Zq(O$()G=RY5{2D%^inYyjE%FSa>7;W4bzv2QB`7(a|x*hBjW?
ztmZRigVvR_&*=ckj-jC$9{FU8z(%pQE@!@@_K-a;#|Z~2p+{_5HpNMipy$EBy0<>W
zQ*<=FQ=Mz7HO4#qV+JN)9$PYaBedmZMldmaDa0p%X9*vHBxi6Jy2^t{Z{bh~jJrcy
zoDIX9RDzI{WO}gl=1-@mr*}hqw0fXMfW5jR;Ln^mNfaocp=qLmiq|d~VM$U77ADys
zHpkKAnjV4W2z=y5IK|&uILW>Z=}Rnr)q>dzo0!=|gg=OZB;5;M7hqv_nz*iizL2B}
z5zVuWeWfxiP>F1>Zv~e!fD<-V_SX9}u}{-`JHWwV=Mcr>Ni&=EEnF&6faDRd3qB+H
zCEaTUF!)VA1eDlWj{qG9lN7l=sVc^4aN*A96t)}c?+luqO)|otqu+%j7%Enc=7ao@
z<wz#u+1{Go#!`ovrNlrFE1wODr^tE!LtPBDNWM7K>D5$n#FNyByyToHqn{)7&I@?!
zlf2fiK$bR!N|+Tktg$Lm$0~P;82=di+BcC8g;jrYs&;n0J>%>aH8$0BU>gaA4aQ)z
zXr{dhhQYQlw|Vtr89eVi1MEe{DuIOaPQ!_|WW;oiGHWA}SDKc#7Z<OSi7_iJGvJbh
zwe-4${-VBqoC9E;W$j0%XemoEfzL|9ni{w_ca|6-^EEh-zc$M_pySVDYQmms3SQ1a
z$ESLt=eR7wn|kMb?3_{d2oZ)?0;_Nfle`ap<+Wt=MA+CX4mJ!g0L1Cs)``uq$*RZ|
z>7Goo@}B)wa4~ppINTQaz1PyF1#q$tYG))3DZR`1^&s@eYL7Do=~M=VUZ2NVL?PnU
zeZQx-uYJ9zkL&yfJC({D<9^Q*_Uko%NmuSJ2ibtyEE=6@ZT7<HdrXbm{mnJ^u5g3r
z6||p!620_`eGZ5n^lh8I9xQcFh%aSF7sUL{c|tj5BgUBp@hPCCEiZc@Iq;`vBCj1U
z#^bEo1XYVu7*><*>Q}9t{&~3aC7LdUXfgbRrpn5&-?m2~Z<H@S@xUPmP3-)1a9X$g
zmpUHZGRbddff#y{mhW9XriU+554vmEy#zgD`WqS(kX(Qz0oB4A?Yrh5+iKL~wwz_a
z1Fq0NTbh57>8UKM{GFzz1%!PKEaw#nd(<#HYOuMeEk*vf_O#;Q@K2A1>tod{CHBn>
zjBUXZ$yE%f07Cm)74E@7%irE7Xjz)FW*!V$#o`A&S$5A2mdq}XtayBm?$AQ-jk@ce
zvuZDw4m8jKjWq!b#Kd)!<0*<hB3$K_wa%dih{FjoqMmS)cN31gcVDoI|2|8uCLJx3
zPVV4$V~l2}VM(j|b@vlGddk*#*a3^z#`>ZYvw4b+W_LrJn-5-XZ*4S}h<ZaL`m<--
z*B6AD;rm}Lwd9i*f=U{)Y6@}XEtz#UM^YjhX5A*#fjOv{BEP#;BUx)?aJrSS^Vf{@
zg$u+240n5c;jO&h!+UT4dFAGaqx72m><E^LiP-0hT={`Z%(um?+9}-rG?>V&oNrS_
z%dEbYrH=?TPs%J0)vIJFLz9CZNLTI5*S3kmH8W6ce>2e3-8kYn0?TF{mC<P^g^M;U
zf!=j3Zo|FfHct6x5fUeyuObuq-I3H-pTTF=4*dwWuPo1Sl%DBYpL5w=?bC||PXl!?
zHWVxt5sa#ay8QZR>_Ea*kS1JI<d$ye`7YHb$-N%xN8T`37s?B0<(C7dS`<lEg^X%1
zPftep>%(5;RD_fa=%RhzJ?gLs@wD%i0MJ&G`Q8UoR~o2TaMMr@YiGo^hvaknbb2~Y
zxL!SP=OF9(O0fJVq)P^6>&#q>^E66geR=WE3vZ~S^)Vr??dr$Ccg;2KIQ(Q}8aeJ%
z*bj`4rZJXDV)s4P!aj%y`i}{^{;B4u#YwPSHZ<&G{&Gj##{|>X6)!WlBIE!^L?1xa
zBAm(BE<!0nPUdO5nVCqJXQD03Q|A#4!uK7V8w@{I)aNmVSDJT~ZbYfGJKWF4#H7Rn
zpHP_Oy2;fK{5|GumN`4&{L$0x*68_gVL)bWr3tN$ISQ6mE9T?EQ#X-C8&Pg@=<pfM
zy{#?e!cIiIwS{K+2;)Db`VhD${QX|ZQ?AFAntdT`4GSg?*zdiB_+MKK>B3bjt0)$i
zy1L?{>^XQ1+phdCppMo3Cts-0Ne7?#u1|aEPP^93Cc3kTxCYg;+q((yA6(=nSdLd6
zTS=uJTPx-&%S=@zK6n3E-_9Q&*XtLRpmqZIAU`IT#u_fUo#yB%qNf6=o#w?a-miJN
zMkd-N7?$d7-WE~|Kk=pEu{G?7FBh1y(tJ(wU|(GQ-d6E!E~3;oyK3iX$Fqx>xnl%8
z<EbO8nBB$7SZC#Xuhp!8*Du{AG%t@$$`^!D5t3O#{6;_f#{&zl7bVO{Z}!d0N@(~}
zYZS_um=VuN(H_3s<pNlPP6OB5l@)`YUSNvO$cO_O#io$2e8E>klZx8;V$Q;CD~oIN
zYI`Ph?i=2RB2~^yL`jgG?C6w}SlYiD1}3s)SWV}(_A-?7+zMr-t7J$6%WLENc^-eZ
z*jlCZ@p?`y_BAQ2t{c`9vt@<QbE0L8kJt0@*DNwQ1->_Xjg&LD3(Vd}d(Fu;zjalw
z&h?uYJ1eq_29Q%yjp^t44-fnFeoN~|Iya2;h;3Yvj#&hWnTcgzzT*2W6`HY*_5{2o
zv8ND;IG7hv?MH6gdkhg0>Q``zn#qY$o<s6-6lJEi#<`R(_rA(L#;m9dw&DBt@IG$>
zLV%Vw*}94??ptUe#-sQNT&?QBieHNYR0FedeC^hgLGQ8*q~>NrHuKL%FFHi8u_01z
zxYw$6@LI#<_FRtAEE<~QbwBbr<P1u{a7k&{hW=0Ltg|m9wq(RS<Zf+JmzIltO&LRR
zeCeZ)S;pM$*UrnN4uN-rKD3m?wz%)imjVJs%PM)JulOGI=Hv`KgGhjT?}Z%rO1>oB
z-F<qIibQ)wCGuN0i_7To15w;vX;&0Obk~Z|<&Ht}fuow)ZqJ%vm7%IHklb-6Zla2)
z#C2EVt>gomi&`bq?d>*EC2du)p9P$axI5u*g{4@W>|Bi6I-jAvnyssZXCi?<7|;{*
zNm;X9OGl8sNEV5Q=IjnS5)+J*s88yh7ZA(czqtQ4l>pDTx7+VkcWjH+us%g-X*d(~
z$r{RKuWvG?x=ciZSc2~e7$%NSl`bv0*Njh=Z62yXi%-5$@fs>vFz%&c_Gf3ZFQ@i-
zIbGl~d=<MuUJ2d<cu)n-awZJz_pbM;Rv8$Q<YjY)4;$4T*h3<*1K}q%7k5);9DmwV
zxGKDi9UgAPvz+tPS|fd?_*XkfzNw}2C)*h^*BP0|W9z&VkW)#daf(qWU?*W2qFJ;W
z#hqX|GBisZDAl$t{4K7+m;{ut=2Ry{>gq?G*nBEbDkuQp2Yql%F2e~{hkDwEBHWF;
z*wBRjRuQ>mGhBEuCQ*Zr+hJp%R>2!v|6ssz3Ab5BaITfU=BR^wxGd!OQqV`y$ddIg
zJuNi{i%<*hBfMww5>jUN#ulwAS>*kMbz=DpD-%7W5uc`Qb|f039yaPp=}X}0)IKpC
z651i}K>rMQYZ>w`YcZ9!0}a%B*P;UXO0tcuJ_Q=KCsOomYP#RVn^(s9s!|_h2L)p6
zx$YIJ(cpNxcC`PAQ^CT26q_>J$@%18A?o2SQE6AgQ!c-g5M0)okQUG?B78(VPzX&0
zEY$!#{hqF2ugr{g<~P2!oPM!h9Y+I=tANo5x~Qt{7I1k!DZ6g!8U`;Rk=p8ig8mpi
z#>T#5w!7LsiJHsyw=EYK=o;&ToK!MzoK7ZcOUoGar!RTgZ9ln8@EGiclQYxOa!5(l
zM#soWcrc(T1qG0Zh0dRoq!v_Oe@v`Gd)?Ydbex@C`N7ikRn&o3hfAI3GF{`|B+%OX
z7CpaBTUKV-Ly2Q2iINZ12p;CX_ZkTJShi$>D`?X`*fJYaEURtL&YPi}{#2y#XX`=G
z6IBM`P-70c-Z+_BCCQS}tqF%st`d1tE$bwu+N6s`1JGPjkvq+>8NawvZ??5h!(`>)
z8Dy&_N0ysmFbKwDIyw>}d!c9X!1BAk{2d<rW6<kHn^eIIgKES|wdK4#FjM;FM%yPf
zG$I#A)6nYsDc<;&(_RV*^F2=Or13mH_!X1TI3<_*wn!DBd=6J1E!%`KtbYC4i*45P
zBB#F$IYOq%Ypuy*g8cISr3w?Ywsy0=l*TM20}K?~Mbl2ZioaJfvrP~{In{+FOR^%N
z7NOzQSQU5hz_;I*M;f|hYK~-&{}rU}Ei62{L6`T#0xWnK$l<-xB}BO@n&VRo)`oqc
zdT<~R_7zodvc28;rU=`H&!+qQ#J02DUEd{mbYU7S#s0RODOBts*Lc+5)2X{?C~zE~
z$6R)w8NP7y1U6{sOpVKV8HzQNzQ~Bh#;~X(`zebaNLq(|)5HAi#<}M)u;oIV;Evdh
zhzjUXO(khcmri#LDL0Y;MArYt>u*&#2n(X%fGcNYms@tdXfURQDl~t@g0fGB-*JVh
z&Dc0DF>d<Nb?d4%nv<igABc%u7#Jv1+2i}hGEO!C1JdTMJk70Z%vu42)we1kc=k$B
z=pRgXUgND$GZCG0g2u*!ZmI+DnU_b?+A?^!>r*2c)Ta5+g}%BQQM41jsAO;>_!gkj
z&b>{JXS>=4?!Lpi(#t*hM5P^svv@dfH=nRj!e?`%4=!)-fWDVjm@ws<;oG~tx<({-
z3ow@0w)URy$zvvPga3i@Pw!tpzP<#I*e%{a<MK3WCtLcf-d;-huDbC2SzU=ridV7p
zBjMt97kA!j$+gkxWPC-(#q>2g$?Ay|k?i)i+0sf-qQAHQs*15gs;<oLaB$r(12~m%
zd+4%}12(q#=JJQnIHF=s`0q5s6Tq`1Pumip4J_LfF1ub<MG;bT(Oo^2J}J?86MUP;
zmdh6UZ`X&1@2I=32zVMU1@$mDO?N)jqoB}zfZf!w+HVQQy#4ZWdQQf`=xxf2>4fP6
zcL^t_N_)Dd0OAH2vju4vb4GnKjZ9*q45b45%GVup&D10~2n$CW-HDFS<z>JkwawJ%
z7p72A(WVMmqt3S<a>##ClYX$fmv0|wIsS*4Uh5N6MmowbUHM0ND7QXYytcf|ucB%-
z|Fd&miSEIq;nh&5&_809|7-WCpEon#4}IF`fMB;IT*5>dYfXGB4uxA7n3_vo<L^)i
zeVZItkK^=Q-WS^dq>Bef5S2VO<F!$ebm|lo5n9>8%~}wN;tg_H^BF(JqOB@+l~L~e
zO@zw?1_s)AoEe+(XIl#?U;tWFQ<(=P!Nw`@R)zeA=gcBz^22hnLGmYTIRA@bJO!D-
zQwZG6mIM-W_w#U*=bfwTX?i0gk;e~a>lO`UaE<Wfj@5wV>I(sz+?W5#<DLq+fvXo%
z9<kt7BP8S}Czn$)AwF$u<C|1pTH5TMZwjj!=dp60tY1{99pWj-te7<2XrsTddln#V
z9&Z#A{rM$i(L%rfx5=!~-(qet*CMyRm++$nBSXi%7sXg0dZ-$1zVJRGm5h&n0*P5L
z1FU7G2?%tld4I%EBbiA)8<+*uy?d)FX<lh$?4bV$Yw7X#9_FXz6;A{S355z@ZLDpC
zmFy(1i5DwJYN>`5ekJ;Y4_Ea)wv5(eT~pA$J67f>$;R3~(uAgl;|PVphPv9w@|OEs
zC{)IB^9QE5OUDXh`bRkAyj743g)0pxXTTBTX+g!pl96(lVUaNryV>sEa-mh3bdMbm
z@33c5U75oZCY4ZB6$p<Wki{GEw6Gh;M57J1%+%}dfM}hM*d_kwx>+O#M}!1j4ZpzP
z-?7-)oeOpzF01|NT?}zc=3h!b7$em_>t|k57VPp{qpJR8WPba0k00m)3+ZV}1sbkL
zq^ZT1-z<Y9x!aL@gDK&uac^T)S@xE;E!IBOK2uf~TP<o#mXlhDpTP?1>M0?niA?F*
zUFXvkwu`>fr7oo8zvuahP}zbzmb3h)6rCF%|7`W?SfF<Ld>f6AL0@yL@=M65h$yj7
z{G7{9d-LRvy_D3x3Lo!HV5G#qk_ZI90meo{y{b_!MN@wEUPLt~$w}pv>S&2$iEK~A
z|Mltni5APrU4qgY`K_2(-Lo^6ON0id!LRtpdSkW}m2;Ifr)t&X5FQFxCaz}Rg1+F^
zDs@$EQ-<K-ff-x@@mxW&D41wJ6Uta9srja2qdC@Geb(+>?BW7Q&_7;8;B+yw*$(E9
z<)2rX<DrtmgK_l#vV$w)7@zvIXWzoyao}JNKdY?{##vx54ahO&NDnNTU3${R{JpzR
z{CULuQoeT~%y@-lzfE*Yqsr7alsTS7RyLm7>m^8Ogw#W_W)U!wIvt(wQj+vSv{Ade
zvKF=q{`sE8n}GYHf5QiApBvS;+o*My0Xrb%K5E}-n){mqI$CoZQ-4S)y9Rw$K&I($
zPNtaFXZn=*50*|9Y1NOWU)A&A!bmWk?!RG*_$?@a4_<n?Cw}Q}xATZ^V8pXlue$Z!
z*?S~7j>}T?2d5a6A=2|k2H04Jg!>m81V9WP4T#(5<LR@Gj%@uQ@mI|B6jP@Zyrd`a
z^Cz+9kM-{Khv6a)TE_Y@L0X98hjfjydnm^tPTWD;m{+4ab4TMj6pmN?)fL~MtBMND
z<NNkaX6tJ9UV}u&oxzdr$cP4!3g@y85a3)KIux6%fJ+SPiPw;sC`)`hD+_3FTfCgI
zHM5v=G0&>+ZTXodhqhfq)*5^1d4OnpJuB*g9VitTTFTy<;Qlu&2#1w;HJa(kpA{0j
zr#xRcB^m1+K^XH**QB*3f{ut->18ZPJ*B}%ljQNqI}vl25+DTvUbEX~?PHGW!X(0=
zE(u>0O=328d(T9CTm{!L@SzO^-ZS0O)3TwfNoYdcM!(^4&)z1FgLr1j<`37)W>~#L
zQu^C3nxcTo$wLrUwjL+4_wLQe?&0B}I@)Q~m=5fm3O~uN>zEF(BYJE**}$`z7!2Jw
z+0M9}JNkz14i?ghCj9)9Uh{$2da@<e6IH%iUu_{2hNUm@`2{|pjjKP^7l%pjRy{K!
zB>tW`mE=F64wRiyM}WAhm8QVuTHSjC6XL6|W3QP;lFJJC3+q-_^z+{!#1|3rNzilI
zUi46UEGUw4tx(5dZEM_>CPy{ly71gOw&<BiBN}>MG{26(0Z0mHZe=#iaR1zzsI7B7
zqJse$OKuKp4O+%n*+)CxjRfgpu>KU&i;Hl@d}SvG`$aGxw2*nqLY=!j@LNJe!2T-^
zug#+HB`~dhfMp_JLD<M|E+lF_*Y&l4w>A>;-c5`cl#thK4px*FuQ)?q%IbNAEy1x^
z$H8#~A?0`3wDWvtBbA0oQPZJ&MN2h;e86C9BX`d8;@2LRJop_i3_x(|Gf-Ujm^2Mp
zd5nT$Jc@|#6BL)Td&wFxYYAU6&ypO6JB1o?kjYYROZIoK_RQm42xS7H9mPGr97*yN
z@_w`Pc0j+D-RHz2_<WUL>E(U=E_5xQ=iFfNzU;uC&OzS;VVGqo*-Q~m$>jyJCGfoX
zP#5)dnwc;a8KJ7&d$=zVf{RN!Ri6l|L9&2>h0&kUZOW>-5R(wSa*uE@h@T$zVxVQi
z^@&bxVaNMJ;i4I;1@8IY6e62{KW=?>Ex~ktnw9$<<3)ej;USUOfKC%zNyDGv)~_R;
zVW^gq6k#{l35Tou!Jp+~<@gfJ0lltbX)m8UfVgf)?|58<4s6<A1N+~x9QE$93Putj
zrmU{nJ?@Sp+1qRI4o5FcK2`g@gIenwrj@*7Yd%g#>>{ddKFY`KIHDH;nnexa<;4>%
zFS5}S=z>Cn*gdmDjD1Rq1$dh+^lf{xA7rQx!YC}BGVv0aQcwUX<oVw8E0@HHG5vo6
zQ-3Mv#qG}hXbqPisTCLm`H<SM9X@VD?1n0y%(d0OOO?FNVG9fE5_7&TFGnp}ILkmX
z>cf67qm1S7ul9hfd!+E*;7^=6uSYt9$LqYXSBOq{?6ZJ3eTlu^7k^~XKnX9JnRGXI
zv?F{_Pm{T&F9E8Z33hYbJr|r`61vThLaPJ;LH0`9zOGgloYm)dz_e+>%&U)H+`@sK
z#eZ+$+w(hDpqmf12jQh(U>>7dnZt>|$-DZFE=K01Q2&lQZtlJ7<Z3X1iShmJarS6t
z$c-!N_7HJos!WI_pv=mA*Nn&zn_vB0!Y#)HIfMhtL1?}FPVX%1mnXpkJOX})$n?s{
zoSbyU3m6jUVlS-F0PPA)#-q8<Ks9TGFWW#sw)BF>3av*RSBxQIzlA{5o{p<<*0GKx
zL4cpb@<idsGErUG^?l@}Dj1gPe(${|Vj==FBh(-y?cZePLkot-buH-SpGN`B=Szzl
zTO0j96yLw*ES&Ol0O?T|8FGJ+Bxgp^7@6+ZdI}9g^AwmO&^Yp<k%r=fJ*`D@J9JHg
zI_UAkS%em8X_@!l$8YP(RrZVd@P%Us`hgMS#CseqWQ&hYRFDE3EVL+y9i+TKnw>cL
zVD}i`cXiSxl}pn!a>oTw6%)G<b|L)&aaOgq)X>mWeTKyqwBAQ3_IO#dS|9^VBv0$1
z9I*FVBX4x|A7Kjge{^(p{V4&mRx&eHiPGj|&g5A0nlAWJJbNFCmHj@v`|Q}5utpCV
z>Yt)`@cR10>jGNvAh4vmLx*gu3GV{2qZ;Z-wNX`f(@nYr7_1avV@Mgpr~Ih+zHbr6
zIBEfsXnr7HvF~EP>fe#G8FIaUkXIup@Kar(GoVkfi*Xoi<xF%|v)ykl9hUH=XA=(p
z5_X6SW;sHq$mbyb5GB~?SXS-Z66&wuVL8#a(~D*zlJ~i!%33)TsIHG)?Q=`YmJ740
zebnbEF83?n5O}Veu5@sCl|lJKI#cBn!Kw_r0bBRqf^y>vx>vIQ2IS=q5y#1KWaPgM
z{(Q#WQh0%Iv56smMK(>~OAu31==gj|*JCqrPXWCI17i?oJ`)M6)&i}}nsS>F(sOFI
zlF**|S$L<>Eodu9Qk{&V)avAECerJ2ZVG5m_=wI`puJY28Jzz$;IpCz89>92-87_+
zURt6bRgn-!PWszn`5|tm66^H$MKL9Z(9-h8^GCiYDAp){ttYJH9Mv+_p5PbxxMAjZ
zZF`Nz9IIH7sv%uegqo^cV<p8Y%d$f#MeP?8$d+Ej*m|W0I^)MSQi-hGQ415W7Y7Kn
zZuw=%=e>d*(=G;LA(I`EP718>C^g+;1YbX0&X6Y(Q;v>dIFln@q&mj4e{wnrG)x0a
z{pO*;`|VcKa!xi#ynD7Jv3gA>vVtcY2Bh#+6DytlM=mi?wXafHO8so7)SQNfAocXu
zanwVO4(=s;2YcCJR@fo(h0FGPAWN0gbL74WK6zSsU!=95L|>P&6~Y8$yR$e|6JJbK
z>SvblN@{3a)ES!APtD<gXxl1YltlW|C*o8B{h=5gx+hOItsFUDghx<Rl?n+x_GFgk
zQL7^PH>kQ2{@|zP_Vh}`;Ysbg=573)%{onl4eh1Z4UG;t0q(JCH~#J9EKE&md2Yi%
zZ)owkDT(#Wx{fYNB5vx_CzL~519~XGaJ?7wm<CPv!M1O5UXZT0jQLCw@p09YSV-yC
zw0jY|sLe#!yS&WQoNu~8_P!vG0bW25NP#U?0+yjc@&ixVISDBU4Rv>)q>b^3agfnC
z5!{`RO@>(jAydw(j)mtmiEkVQwYnivH;CF-6u#e;R}Ny}XX$A_|M<NFckcisPW@u8
z`#!<&K_+g=#Ju_R1vf=O@{f>0cK!x~>$hkk>$CAigyQ<vR(}R!Tg;GCQnP$+NtdMu
zA68bu&?W;yp`LfNq@_O0H$0k{9P16y4qPuAa#fZsgADZn8B~qD+Rk<Dh#=T(W1$~J
zER)xzUqLAuwpJ#7FfoqO)oy_lQu648FLvpio5P&A!KixRXxWmtNAkD5z=ooNe^TAi
zyG=>k${I`(Ob6!@$qomIeF$o0Ku#*}V;i^-dl0tqe+wHvtcJW2q4E0wJYVqd$SzyX
zT5~=wQ7zDDa1}*EcsLggCspKGz#Qh@)}NbM?IfPsZJyg6*gNa1DZjdsRzr*wqgypl
zw}GK~C|LS?Z@0<JC`IphLoB#+c3dJs@Dy2oM62sb?MCC<#Q&A>eEu#A55W4xbcd(Z
zxUVF2Ozf@<V7AfGR)On+nVR9)LmbKl1Ml{q!XIh+Y%6c&Z1qw#F<nF#H2cVuL}^_`
zb{-FgMvB(9L`5_iize<!DLNUeg$|H6KIjS42^lK0ABGwxkpGU-+7J!P_IA7pUV~W*
zMX7AFbws|8`s$qF(?xwR8%;>12nbVCzG$a6bprGKC4Yl?GhQ;Xb@WXBU@h(s5amq9
zcrkG3ypQqA5plXa!Y{&~RbTjB`MJKTujm}XCeE7|BrjOAjejGL$l6iciYz{38nU8<
zT4xKOQnckzn=W*HhJVmrT9ekl@RIcOR_B$*5!svw+_)<GDM=S%JsU#98c@c3Ldh0F
zod+OJE~gV7UMv~hFjl-2JY~C^cyOK3EA-&npa*+0q&*ab9fro)-WfFsAesLNH72F2
zamWem3OAJo(1p>DXb+5cv;~UJ`6Q=UPmWBY`j(&kQc^l_t;m7Y>(sA0!>YpYD}Gn6
zNWHtuo@aitUdnW#Yqm2HjS%%7sHWfn%H0sEARXKeI;V<u&G>kB)5CkZRJ@pugqJ>@
z|0y?m6){gl{k`De0cn>JNTXXRh|WxFb6?Wa@w8VQxh#BTXdWFc(tkirh)zbOlf)Mv
zC$nE^iHo;ALONW)3z7O5uN?+P!M+`9iV?N3Zp$Z-q+duxW-m}ExOhaJ6O`>Omwv1^
z)cVWN_qx&a$k^KI1BJNfyJ5~+N(%88qH}^GC6i7CiHXcAA6ZtB+b!ROVf({@3O;xb
zUS$2gU#kS(5~{$|2GYteCM#EGc9U!C%WTzn)j%=XEjk!g-m{I@mXom{r&f)rQGrvt
zR9DQS^wHzT$^m9r#u8sdoQ#wr`$q@<{#BEc$;sDs@x1jQXXd1Zw@_?T46+;1mO;)_
z5Bpn20tjeQyP227%J*76;%nnR6A|_=^t-!V^vZm5y%WUzWRJ)5-ij)L4*uezTiOe;
z{mN&eG#Ch@re$Of{x!sbi)W;s7(ZO(EDct!!}?jkp3(Q!o9DmMQIux8Ffc0snPo4t
z$8~9pd8PXRr<>uT+i^-`(--MyU<{1>r2gF>LZr*ukGne((foz#qa^0PvXuaUAC%B8
z?C=M+L-3FN-(`iaI%BZfI7?NG?3GoX_K~BQm?*3?*V>?|T2<=dzRYE;nE6myKpv~9
zZ93_z&NI8xK}vZa`%?ITH1k%QK;oEf;Mxq>hVYw%<JX&F@797|@QO~kF34-+645~t
zOrj9{zVyJLKp3tY{cB&fyIZ={?>O=4R2iCWV^d;X#d~+AU$Mjy%?}i!T?>RRY4VR%
zB%KM1tD2a|Y|aEr)_aZ?=y(J1@uY~=%!4|9bFYjeHKcd?&h%((!`iM-Y3XZIi$)Qd
zow2pI@*n?=2x`Q^cv7>6i--GDGAo+#33vMq4ba=oP?<}rr$v!4HD!=#Ok2f~^%#wi
zmbQKsMAV&~1GJ1R=vZ15)0C9?^$vO@43LtGSmtPtz9qm#rNbzFEguTXM#<4Z;UxsL
zGJR{CubC-9PKn)r)JC<*e-$1kk;3hCjlurW^fkFqlLl_8g3@&T6)%g1z0G`?yp2{R
zX{E{O@tVKwj3tyEBQ<B2)&^!y;xg}*`$jYGWHIkzRcgw*ynVJ&TN@W22Og$aFjj7>
z<j%G(0Jvf}@aF}#Nzf%2)vMqwU0l}=?+Hr$`*I(WgKSRVxg1l(yn(5-lq%21!yj5m
zg>va0H}h-#us<(AnBgN%@gM=>@JXxsv;Lk9sWJn`5&{R!OetI-0Yu#_FXRye5r*3!
z_*pC~(@O2iR02CG^u0=`|DUqHI;^U#`}!avR{^<-h_nhwcSwVQNOzZXgESmsBM2xG
zl7|NAZlpm%8V=o!aOm#**17k6-`^kKeei+jp4fY@z1Ey#jycvIpg;OIWYXr8m$k9-
z0bf*m*@Gl?J!aB%)cOsBH|U{wO_juxH<{)|<?ViS{O}ghC&+2_7e+(x`fzf66yd|K
z>7|k0`ypdz8+q%ln@r3VUN-s`k`iuwpy01meX~wgpZ%m-(Ty?8{iT|K!0R+ak~I?Z
zF40qWH<gPXq)wTFwqXzfq{bH#Ds`lGLvEeV{=M^k2kYAm)t6@69wU2PC6!~fq3&xR
z-#^K_Pi6ob<s*_GZ&#T0t=j0s3fZd%pNs>-lapnq&(cvy0AZ~EU^6I?hKKbg1nFUw
zkM1Jnd;dJ0n3z7V6`p(bzmm?UKPH#;>!Z&XDefDi+0dsqdv{@V&ht-X5KM;MM`rt7
zT_xP?&pJk8F?1=)UF=&sMBjnf@U2g$dF^6OU8`EZ|N4o{#kRHPnECX<m~H>rZ?)L{
zPLvg&6G=v@&#`O>?(Y86juB!;sy88idMhKzUKz^~W$G@6u8HRbxuNi<JPk>vi?Sdc
ze>OFGrV$}|*J@kK(!)&*YllTkHjg%WIq!(HI_~>>B}OTD)yp4pN-l9z6X7wjzl#K1
z2u>ICdQKNfOL5A_vg_jkb+;@@L;J8|tgj&NDhid3Xgs#Zz-SPLk$pk<Cu2}UhIczF
zRarOQWT?*7BxfEKUBl~@pZ~DTi~bXZaCpY8vtu`ZQ=|a2P9(!}hTkM5jCX4(*-mHh
z)g4yZuzvq8w=2$S%)-`*1KYH1+Rip!#?%@RddUg)F7|3E<c(Aula4pzW5x!(+4~CW
zQ6MSO9e0Vj4;3!A<NN^4f+Z?LLm5{5mvFbL1@M=&uCR#qUo`Wf+;w#~e~IniV^C(e
z-;R(@!61cNSGRS>KVIH$40yMDws!D2)ASZwp&O^}-E~pa3ziNWtuS@PA0HqqLA`{5
zR^hS@SXQ{=xHHuq&`SQo@~R@F`ndl=($zLLAZ1rq7h45pZV~@k{u;Y*BXUQwB5wYV
zvKKF9*U$I?!Rs?M?K-;%RuiEN2fsh3Ni$&DRmDKjO&R65{58C5PP*5)p|?|_OqJ3#
zB^9Sxs8~bq;s`JIMCE}}VApv})a0kNUr|9XxgF39<u^J<6#203o`j{{)~oOO8H#$G
zo7?_hUBEz%xl-a4K8V#70U^aqha8x7f`|M`n}doxdZD-GkM1s&w+V_^+`eGdh76XM
z8S6&4|0eJ~FM6zsDfNa57BihyIT}M7aDQc)rt#x*_E^gazVcy*-^-Nxpt1YIf299d
zRh1Xajp&GO5w*gDMrWssy#*VP(B1<QY_KONnqm@p1ICq)_(3Nq7x-Te*!mVlrapHT
zZ`TLJiuZk|B7Bb`^1FD3`%rswgqHGW(Baqtww>w8_&k*bxoVjDYq6J)?PEbdpAA)M
zoKO!>6hYM$R^*EjYT@9(D-Td$aW)2Mn`oHb4%3(b?Tfv42w6d3*g>J{Qg}y}j?PG?
zmHTmWZ7VpuPsGrvUM<fbMECD^#{XN6&6qT*yZ7*+RzS_{+B>hvTiyDx7CRpEq|9Oe
z$u)&Qf{jg~3$@FuvQMMqE4{oz4)>Ce2lAt@fKmvIW9HF@Z>_Iy-=1><U-wOe?pvV7
zA-aOw+qW6DopRSw{VUj6nEmXu<X`4FXqF!wtWj=`bD&l@Fwv;YU~Cm?14C1oT@Zi%
z^Exfiaro8;Ow$J(${gFGqq~WPTrf-vwKLMb5ddk8DTuBk;k79_l;@HkU}{N;opd6U
z21A3y=i344MyOS?rFhfiS1+PUm)~(^g&%7DuHF5k0qBTwi}g=M@p+>$3Ocp(W`;6~
z`VYV|7_dyhQ?vaj%#BAt^=1}|<(IGw3j2hCDc~fUp?WYopfr4Q@0FvCu!s)sn_C_?
ziw@jvy`CN<Fv?_pDX;2dnK?&#6}r>iadVh1nNVq4RhB<)+i0^-McCM1IhSGS4J#A^
zZ({#pTRzwrq^eN{GgC9f4LA^e*HUin{)F*a_=*>ZVP?Xnv1HbwVUVe~7x82=;Ev{P
z%$xj|XkE{zb}V!{1xy!j!)Ar;eLC{30HI22KPo-DMxxGKjf3WvPinCroNY>j<!#5u
z`OPIX2{vCKK|q4~^*_<X?N|6{QvL5XfB*6#5AM0ajxSW9az-hLZNiKcH|DEe7V<NP
z)(@gk#60m_UO*qu9x-vx>AofCKXr@WU#UpB@06>?KRy-lCRFAZ_%QNXx4p0o<=DpR
ze!vad3w&WQJ=2C+8&RT6h1nlKJ^DyZ`S1RofNP8`r?$qkDNwanNq_)CmcaAM^7Xn6
zpj+}u1rp4yHNS@^whYTP_Q&3KC@)!o_Astsu2xeN;rG6++r8aY`@`*pQym0A|A>Ht
zW83xR{<rtuShPNTj}Y0M$2LGnA=r4r1_Lx$u0up!Mp0wn+voanj%3Nn!g@t^``ydZ
z({%)(HLpZV%YQba{o3%jW1)&~JOW4yg0|d&G|GuW9Ngy1**DtVeNB_oP=cSFbP@Y#
zTXRD75G4rLoNnDLfbCy*l18sD>p$FN`T4G<>wn-C^=g5V-A_br3?-NCSB}=Z#;)8)
za_%@{n~KZ!#RS!6U~PmZ&bJo`IJV)h4`a}hS~5g3_@BF$fb~SyW1;7_0T{r-PD8ZN
z>|Vw_A_1-Qk^MU8|2WU)dTKN|rl6?Ww_O%@dn%dg&OmNbQMa7skPm{tLL9HI(yF>|
z+>LQm9w=ulUy;WQ=DO9lfszAtvVe)<dT}BmPiI+}W1(-*+~!946ey7#w;!_Zi1=uW
zdqjb1`p#v=S3q2;$Hm&8#<M!MiMRbOgB(wEtz?2UM4VkKR61$^6XHe8;SFjRpqpx5
ziO6ySH+)IOBcAJ7ebwjLyfFuM=Co1v;a`gf&jiaZp28@5UGv-Q@2IpLZT9svVJhT~
z1WlFuHZoW;jjeppzJ+-``PNPhtA6t5!c}|bYS0|WG{Cbc_HA=>L{pvT>02l<yolZH
z>`<%rQ{DvMd+LWkQy_ut9hz7<O)X;QU_)n~G#)D;Ae1&06#b#yS6<+Lit54xyikbO
zg>gX!zb=J9;MaQhaCQ0jMckxSpiREP3*Z-0HW?L#DGrj%`0l{P_}GHD&&5J*uhx4&
zB}XRX?>OhxA9ihNLrgsrY+lrW?(@qV3)fOU2z}Vz%o-7f2nC{^ym$Cb#nA^I0#va<
zBWVGPHs-H1Wmy1gKWv%wzd#Cb%|A@bN})E;B)-y<b9r}uwmQ-dDy0vG20frIY&I|`
zZ=*IgE;-(PuqiF$TI6n8-$p!DM=T^X4OJyY_5d*<zhM9t7MczSlcZt72~N49wA3K5
zi@N1@G<HaUb5G;r^O$2O{~4DMffOfa-(~S%jGUZ_hD?*?SU^CirGsf#!d7&$7xSd!
zG{4AX<bdcVe~g$hN|Ntn$UrbLQG%W!^JM)9y>vjj7*VA|Fh*^LgrtqFLq}*qj73Ep
zN<4JFrOO_x;%upR?8!_4qkTqrmNWT>)0zS@srR|j-asv;{ayyOXYKx^+KcpTdD(IQ
zVELC?^2VkfSWvy^GVi@90}fl-#$!O4J-CN#RChVL15mStHPF?%-K^5{Usp#nYVaK%
z9L}#DTZ6(Mh1wE;oI+4*o2}PH!oSD+zf<jfE!U_C6hpC?>lnu?X|@5i+MZrbob}5z
z_DJ-N#4bNN&fO|UOI98FHi9+=5>GF0)f8yATVjXW2?<nAax<vVjXQJ3*)Wck>>9C|
zw6$FxG}Rtj<auJ&UC0#57rPEU26G3<$RZ}kCe+5Drin1(>w*qO7L&E@dhRyQ3NI?y
zfxGkXuIv?fJ$DH^kBM$kA<U!N#+QPV@PF*u=eB1&8Uz$y3bOiOyH=uF_2IlJ62;Qy
zBMEb_Sq^403-Vnzf-Tque33s?^I}0>9_8oJ85FxENIH+~Y&H6sEZlvXR3r7e2cqCh
z_$fRalvw#4G!Bv`JpXQ~4=_jl@y_$(0}%5rmY+7VuYM#mBAGIv;f(B?rbP;a3Im{R
zeuE}Rr?d4TKZW<FS(&}kpU**>Y3*SWB(p5kw=y<q>m`(m-CrX@ON*Bm!kF+wTU)xq
z0Yj%wOdO`5*tHzQ6v)UW0Vyg`QF+0a#Q+m2U*u+;bn{H_y6n?g-k3eFa2QZLjmO3G
zcQxG?S5mU??ygEol!6<FE5C^1uxw?Et~>N7@Jw!WPAs3+29S9n$Ju&yYoYDH9p3?<
zCP1LIsJ}={jKBodJ-T^`B6_G&EjgzCVrj<amB)A*C+K8DV6uOhiv93-l`a|9UnWDE
zs(SV6hHuB7lh5xYG-=T{Ez4}lEVMR*=HcD+M~@x&IB`^rt8rb#v5%oW`retDb&uJe
zpkR%cCqfMq#b1%v=K*@Geg_*FAM5f9c1udyfK;!1@!@lhGP)djSbt76rIa4z=PW$y
zM1j&3pzwlXFBczzC_T_Ks=4EpwACLCFjw)FBb%*UlKR#GpqE2F`|Dd8^t7sa!9zJS
zR2<Oc;+tnb00k<<#U1p&g!@_i+k1eg-YPCz{=uxbEUL}x23#7*Wj$*ox?zQICwOtC
zX-nJQK)DF<1C6XxV3{6PvS14V_BuA%W%Kznsx(bE%r~(?KWBA4E~d*4Ewn0)wvJk<
zOL6Y4fj+ht!O|q@u&B>U4B!=ea=yIJJl;s-)Mbb&)!)8&o0ZAoz88wdi6G8D#JRV#
zI@H<b-{}e8>vbq&%Wh}fsdNXO2VF9OK!n$2Ea=&WX{ooa)$%}YWnA1Dau+`@k0%*<
zEhp4v;|Ze13fDEyCnt-cN!R<WZDpX>di=Z9L74k0gr|4Q3SHrSOPmB$?2ohfdS;md
zrmtUusT!-ESUSw)W$SEhR-L1EtOPv$s~J@tgxJ3Vi7*&`+utC!0HM(CY&rdrY~b8%
z`&0z&ek0U`dFFN@`!nhOi|0HcRhK?m>|o6Xtx<F=VV<)Win~xdWs7-=%)oU$NsjCL
zn<gH(D=GEq)`n>R$W}$1T^^At(?uTIuILmE;-aNstEja;(5Tj~vWpPjn&dH?{l7S!
zf{!L@1VCF|F{8PuN4ZEX%@ze&r)exBwDRjmP3)ajjT2OS+rXM{8*Pn=+_W0Da7rXF
zw3Q`6#z(X%y+9C95VVhczTG1uy?b(;ULCJhaHO&mosc1M53coc)755&iUz2_tmCR{
z?0377urGlB7UYTUnHId9o-v>6sU6*a=wz&KU~ehrEUW+E$G|{qI+VGvy(;QXX{)`p
zu2x3;lv~RoL73;Jkwe+%#hD1eU-zN{%To=2B%hs|t#zas*L`R`-bSK-+69D$bY+{*
zszio0E{4)4HrdJy&zA_keEBAxqdeHw9RZ>WoVImm#mKPbGIl6d#DTwBOf205C9kgN
zQ<lgqv$^GD^ZEozp)pmPnL96<#gZwVnraK&99A53ILIxKZ#*KK;Ic}Xx_IB&S4gAy
zf6T~saKMS)c{8Lc5o=OLU-s@jmx?u8ro7o)PkwrS9Y<erlVL2$6SPd|bxvz>lEX`O
zj#P*46-OJ5Mw!`dRYhGV!C<>>E>!wD*$4v4DkA`jN4z9)Yd~eJQuZeS1W7uKEO1b3
zso;#1+$%0^O}if-tde*$xPP83kfSWAWn08QGG<K&Dj_RortLG{;0*R1+o7%vZAD#A
zhuT~_Y0@RWToQ34>i?s2`nq~>zrKA9x4Uc6Uweo3W8PyZ&rjnhDjLS({1XPmYD+db
zP^wghYJ`HKk8hMpeMD_=XGSU6tGMXknsvtdD)k}LWxn$whX$wZK+<Wo5l@tI`~Kgr
zXR5|kg^GghAjI@Q)=#^^KoT!jWub4OT@K0w8IoeVQI?y0P!g#Gz0sd#eSM3-cweFV
zsdMyhA;>wp;>Pv)|72jp$>=EZ8F0{I85y55Odrl~X1j<0sUpmk-&{{(a<IiWu?fYC
z2nX`&ve_kV$!^gfiUZRvIy=H7zD4n*55InPpf}GRwn)dz{y~zEa7XkT+-l{>c66`>
z)4Pn`TySPqZ^sJ|24G2+aBvPmae|Ky$5Qvt_z~f4!=vX#&PWz!8g!3bi?qP_%{7#8
zXpuZJHu0#}Mc&~A`(kD{1$_A^rBGj8l;szo$0s>xPx5?1{DjT@&&Sz#9%5S3y_PTk
zZ9Tj;v`9z|^^uoT$+!nf@4;Gp?&6%%tDsL;6>YdXWVb7SLu+{!NnmRZcXqOWX=Q0T
zU`W!55FcM?xN~H%a7amlIwPpS4{JuV(xR1Exb6<N^IK5pF*1ZG`n}%qau)mjOpSq<
zWcV&-Z2kjNI@}GizkUI-0d%0@CuK+PxBw7*pu=xjfS4I<A-5$mZRX(k+SLmLW}u|k
z?(F(e4aJjs*~lDjqvmpp%Ok(y2pgXh=@BHxNxpok*04VJttC|Eslh0&PjXDxBzsHD
z!rVQ~yO&i?0$a(Rd!r`s|3#PAOMl<hUHEi5Uy=j6bYu_e30JH?AEzWhf;~Y%SmA2F
z`k{awJo_LuHjgnYLXj7gGh@6!!3SNEJPi5qDJbXae04<J5N@z53VWP@`>w#05|lJ@
zKtc}Upb&?Iq=1yvJEx7;KvvnPeQzueAG~6kg~VWU==W|*NP+=L30uk=Sf`&D6aKg1
zyikDNSo^eL77g#fjEIdw4z$xE*I#OU&^MkU4xO(D6~w#7mvj0TQ0#{~>--5JV<vHm
z<PNY;#?i&rr>FoAsGop}vwLs>v-*6K0Ce@lOuU3K=iEX)iCBbJ-~2BcoFf0Z?$x0}
z=IdEqPz17Bp1g`&KhI#`43u)Uo6`qXGUvTClJ!NB4kM4tKhm(IIrlI5@D~w49)R2w
zhD0U&d{FN~vP&MdULCfkV4PNSFSx$u5YfZUbGh)fyNezlbVo)e-X7oNbXW7Vkr-47
z!oqIR#X&vr%4GLT+?V$Dcg<E>zW<Uv`QYE2B`1XOE1+G@0l>l8Efg8Qtih{iE&SMn
zLya=Qc16LI-_k0$!KRiEpt*NM0ikvn$0j_4?h$Ee=@P%JncceS1vRjoSN)g%)h9qc
zZEw+LbnXscwOX>WX#>M?<}7N<@JKS#^PFZZB7A%3{}90S6uJ)h@RRxUFDgvT<F;0I
zcjU9Ks4o5rDP{(fp0+I@)B+lk4g^v#hto@OCrNsB#25PP^BT01by2}lpb2WS#n<s^
z{I({%iW6E{a<XtKA_i^n%Yv6|9BlIiEwt+7b`^crp|sQor$6h#I`r~bJJTW!6=K-`
z76@oT0J4Insv>rS^7$*2F!iwX=4%}g-(fV%%kZM4$BsBw+vudLD9>Vg*tP}g-y9_Y
zC`z;UlAwtRFZfqeS0D{*V+!Z&;BaJyZ5=4A7ba$~!HzgB6rW8kaj%2>eSo{wk5}@5
zvU<3B3u+o2GDYjfl0a{oOyripPz`zGZPyYd_u19yFo2wNuhuHv1r+|*DG~pSj%`68
z;>Q#9h=FDNVvzyDyqRpRjw~lx0{g&jw5wlLM&0}!p7b>%mY&G#nj%N}#G2_}u<vL-
z;q%R$*iow@<VejnWw6BJ)Jl@%YRu-(PR9m%*N1VFxO#m2hG%|o0f-_@o=Yk)bXOm2
z!FWFsR>ktYTLWx@TiB=!*VFBDA9^7YPIE_C6RN)B4O&g<j|c}`w(c~DP`-{$U7;ib
z9mv>HV<Jj&5nmvx6cNy}p{6NY82@^~y(Qk)F4t$|fusO<w%@(?j#vGo5bbt+Q6PbR
zQfr5;m%tb5mvfgC<td*va!gh0yppFOr<5ztW_}<$E9~KuFjz{mR3;_mcF=DJlp&A|
zkS(U7OxU^T<ckb1D;4ubJjxBYVc)Qin`Vd0XFYp%MhI<5E$;(Oj;rfiLpg>2F1>{?
zDuuj_Gn-2)*#cF9M~_;8zfSL;&Irmw$HUC&#*J5U3(*c^Hf+0FC!p5CPLU!Y^1pyD
znOgb+Z4~kE48ZgjQ!zmPU_TkmLtBTp@{hX}(|VF4Gc)2~`b4A@@VYP$YwIV=s5kDk
z;t4%6VDA?Sbk70JX+sl-8Glh?J{th2m6!VpFGGA0uVaqa5MYm@zQKoIA9gYiPY!>z
zbUl0%8yVSb5JLN+jFvxoy3`)D;Vw@HaX>?(!#C;N6A=z2z8PvaSp*&2iV3;h-idEa
zf<JN$Hv#EQr$-5v-@KKLY|iUfKq+=%xWxA``Sa_H=3cPTed*nIRqbXVy?8QF!Kp>$
zahQU?AtIHdxHk|u0*W<ESmEQ9pyPud4g^FAy}fjW*-v?jygk|XKrQ<^LlsD>n^)W@
z7k9QGm7)?JGLfdCIi}3?Ix-xpH~wd?17kzvpm{uO@t%cg(d_j@_tYiu4Zo&aKQG(X
z_W{C`<mMNp{SSqK+YP4-AW|l{88D1g7U#u_v^;mypVK$NqwE6ex8ZCDD=W#2@cphE
z5#lE#AB@Zny-leF-`~r4=;p?kX_EQ=&<nct`Nn6klw&xG)F&=BR%m|WFMLf2b9#S&
z|0kpJJhG>@-|;AZ92|s)Ka`Y|gcX;Uw~mjWdP(QWHeBvY0XMmJV0}z+sC@esx<OZ&
z^w#HOmM)^(?fQb|((})gd*qBqOwuy3FPGSLli6f<O78o8-e1^PNoe#+YkML`OA86R
zdN6Qe<GHTBeAdlE_b1PZ^whY$P)lOJ7BsAYHnJChM<76`Msj-a=8r?)O&fH2Ht$=X
zdCBX+0EUQlS;VbPGq7Z;{$Oa;lb0y43w8lB1>w@GF_`bSC@Md^JE1i?rhcPgMuxC?
zx3>_evf21Cr|k)ltE@=Jbx1*m@ANJ5_4b+h7~|zeuY0?$u)`Q1X&M5G@oaAXlLxOS
z2~aD;gAZ&^z=r8q&;<c1iDj@`5q=V(p(r^-)BVxm<5Qchz<2R;^<XP>^A-rg+0Xy|
z<75E6Ir^z!PwUO6&jCJP!ya_c!F@Aq8`?wyLibMw4AYGuW3UY^FfoHhxR6EgrNwQ}
z$qMUTbxi)-=xjQWld~xRg#o>+Vp(ZPS!rrc1OX<0c^25$LLCmZi%f(5^O^mqQcp5p
zB9*>=LZh>9WC9u3Kn1|{?#aHv`wv<00_SViKv@Zll(g=f#+scSGi2r-RVNd@v8^=O
zSeXN_1T#ZhPd^OEl;Ks}KLW(T-d09TAO{_oYzjHZSxaB}Blw7P=L+F;cwl^zg(K11
zsgJD2nU+RuR%4JrZbvi?@MCij9iII$;gX1r3qBQ0NLZbp3i?DY@Ij#+UUs9Rf>gjZ
zZNGE0WM&dr8%6U~LnBg6`Gb|&_>&=m{}LLm)M&kzJo|UaSm9!kA?)Kulp(i+)3V<H
zl|ou(dHN$N!=(VPib9smws!!yweI$v7v)~C@^$wNMV@>dF70X?=!pb8%K9`SEq1Ly
zSj^|-`eQ}eQqB+GoipC(%uF;@V<%KIr&)2vR?y(uk7%d?oA4EA9@f21jEe65`sEuH
zjR1Ca>MhW+)O4*{GPAx)$$IfuMQN_|d&0<WfVDr>ErtkVEhs2BA&Gdp9Mad8^e2D)
zuPj-Zdk^}xnu$SQj0W16$*NqxDxM|YWFy{Fqw|KTKfdv~x7W=&VR17QAD?F8H{Y8u
zeFr*9yOU~v05dbV`DoKX``1ZeZQ4pwV_1X~VDaiN=#}UsnCO~8%^;;T)E1ZYUsT&n
zC86p5WjDKvp2?`$M{f0EMZijgAn0v@!Ix0lvRjvKAO*Hn?*2ZznXKIbZ~J_;bP`J)
zA^DX0-RlQLe~HyR6=VCAn*F{EQh58fUvg0mkLy_JsEQiVCc>v?x0&I&&07tfz?7#7
z>Hht1pZ%RqL+$uFTq3q7VU99J!{FX`XO7q%RE@pPQjbGYHOh$0WIMr4*6hpZ&UA)^
zZCyd(iBNU{y?a3jUgUbH4!=}*!?hEXu`&Vw6T?yJFq~#hnyeQ}*kCG(f^yGp%tk)F
z>0)FW_GU@UeM*RaM8qji$h=sN;{7bkjv*xd^2eGu|FIzlPq>9q?dZLIR<F;6Q(P4<
zj*64?>UAx_V~n{YD2M7A1q&C{PD<(c<ZJFDUaED+Jh7IQ8;cumSsyC5^ILH<9eOM=
z_(ixrkAq^K@Z4!T`_7Lan$YzI22C<=22a=n!Z=i-j24Xq)w#A#)E@dYe$m?9W|icP
zN-WSvUaQat3^^&~iI^P-xtM`nd^BA;5~alch32Dg8F#8wSX+-&lgdP0>NsRbU1+qv
z1FvuWC-C(Uh4A$D2Cag+*JEPH5}u>t>m!HK=Pes+4bOb@_@B!|8(4R4uOg{zBc`~H
z8X2MQ-CV5j5NKyM4{UPGF_sQ`cM_WCABjGjc`Z6ug=XKKhOC(!^O}<7<@3#z_(wLC
z$u6XC1_|nGEGNA!BX=CvkBRgYWOa3q0<c(`6UVL<)GEvidWprIZuX;yRAjH<{7PvJ
zmztUBjWeD^w2-<l5`Ma(A80+O&Uj2YJapI6k)3cSnyUTqCfbWp0)7R(p#PQ3`Vjr(
z7Exn`-PD-$8TUuYyS^-Vcmn}%3K7Zr+QY~APCKbTN|+;^7eCKl*(-2o+-7J)nhc?k
zLIpAYct>l<&YH6WG5+lip&e&%80EQ8ks<bgHxV{aoo<9pUBjKhTNO(0KcqaqKGN_D
z_`JH+%3R~VXK%bh6j>WgZ{7qR<>h=$)Wz~~J+>r`yW9mCcNZ5uCQf8ntbz0l#y5i(
z3?ul0H(|Anrti<SrbjradrvLGrU_5aS>zMr-ilc)TS98_nKDPc4VV@J2UJ2gE&SFZ
z@H%Kb?8|-AACI#(hjx)SY0r{9Ga{<p&WDPh!B3hljJ5wz3VYXN&KUQWFOH}Ub1_KX
zoEuk8_FleM8l$l|=~VjNmPk+6Pt-|I!h3@ge1YUx=1qiqU8)O-*yM|Kq6aNyd!aUL
zm(H=e+|e-g(wZmOeC}uSLHXApdE^UrO(;3K5DciJO8lrw%){)uQJHwNEdkrr34D>!
z9vVH%$#a>wpI#BKZ(TZ>n`PVM;5ae0FK9IXaC0ozjpconrq23kw~FIu)TkUJ1)ViQ
z7PDO{=Ns!s=zqDUm5lsc>Ka(G<lT0z*B=MBOHJTgvtMf#a<`*b&`<vKu7Ui0h$t>X
z$woT!nU2h+vtNtvBw2nqFwq3(CG)5*J7Yb-e?3uTCQICw_{Z5CjoIEgo}_SOu4fjf
z@e_ZR96dcxf%389JookED$7St{{G-(lrn1m=)n(2m`5x$xK1XyP_jQ9@2jMQKz@A|
zhcPEEcSxYJc6$hkmlJF7>JX%L?L{Lu`mfz6**#s?^km*=5;aV9x)~^-*?J+-bviTV
z^l36vOqxoF)44o}SU?l2?^a@GSD4T*y&un>?WYCZV(B34WgU_tXxqrRyJQ@fX%FQ8
zWM+uu;HVUEk+75+h27?sWoM~Dv1IiEE-(MBN<N*aB%ExCXRn}7s^YM^IQ=L0aPF7#
z|MN7|Q(SNwt#5fEsWVKB)!>EYn`I7M7URW(4sK<f_X8dmOr`wZc)5<k#CH5MOh0jj
z`G{abzJ97=1dM{YD59^_Zjc3iSl~bh!(tV@?#~Wu_Mr?JI{&XWz&A3|P8m$P*S8q1
zp~KUK4{_8v4&OwSRRoI^SMANu)xI5`$SpZ7N=Kv4VevA4HkE^NNy95b^l^L1zt^Wr
z{dBLO6SEIvERSFObHd<ZovHJF0W))YV=g<Htg1x=YPBa8(ic30akSiwT6y>xIr{bR
zbx*Nc23d`Z#GbCf7(|x&a&!R)^$ax2qgq*MSs*N9ILyYY&QhtBqcNdbvXio6bpsyT
zd!4(H?C}mc;}(!q?SLzO@UI`yypP>6!6S#c?@i(9L}OO+4S5C|sD{%eIAP69oa@rl
zWC26CvrST==Oul?bg7EkTkc{1yM~8*#iU$JKmDHj`WnCAubgmnyZR^ttq;DkYzrL}
z3BXk}6ZO_jc&QhPOt3zB^X~|H6aAPMW5=v<9Cezhv6x$v7OL`y>iIT(t=G6YI-Tsy
z@SI5+JT1LmlVj8*V%9@z3Yy6hmLX+a=$jg$=g*zq^G|12D_3YvtfYTVCvVGosd%93
zrN*iM&vU3?^G9?L2&p$RZpy*WpC9gpB*IVQRqShKh{|U!lU>$ki_|*UaC<!$k%52V
z<gKJvbaydp9i8Ee<eIw*c{(3_@SEBqs<UpFN<Hk{C8d!Fy)yh+YVVN8_~PHQU;q1_
zZe%|-gQ8yjX+mmq3%BLQ{okIKv8v>}$}0nPjY)mOkj}U0jU<}6-2Sc6!vKo@V4<2Z
zvz5ttztDf4&E`@8MvE1GA;M#usj#KFp>C2nt5=?$yY%k`q}*Smwxm<394*llxnd4J
zpHimTnoT2JYE(WrU#L0W?{?I<g5NtlP=Y&$jHXkJW*$<7r%SDmjfscefZZ~UgNx81
zH|y5v-GZw#B5s_bEI_}ix+dLhO~~vCxNICydJQj8+2%EW{`Y#azkv*t+09z7yfN#h
zluUD(@={nYo0;C7+kGnJ;Zoc?G4sXwjiXg&D&?YE;)SDFbJjC2BuJ+B?I&>4nhzX<
zLKMq(;gfZ=Vzhlm^c;1%Dq_&T5Aqi!f7_(>iUlDhn>DolgwUi$n@I5ZxT>K>sIW*e
zlzgo1&rBDuo8p;lmH~KvH;7~FhXLKjm#P*uEC#casb*_;hqhG3hA2CD-5(tgDuhLK
z8N9b3nvf6ZGg_Muqe4GpH|Ol5o%9Jr^xQ5!^N%0WRGEB8@WKPyZ5!iOCa*l38^ap1
z-uKwGTd_sae5{_%Jxz8{Ac{fN@$A;XBsgSVBK`RLJd=<&sOxJ>Exmc?+HdtX7}@T0
zu8eemsEdt_Rc+5M{&jizZFVJm7hbcLYeQ4WN=KI+_&z$XN90#Khn!S6$!#oEe({$1
zG6$ms!5khL5;X`etQ=GpB3Pv0c1Gx0|90k&ef8<?eF;U3A{<`q8{jo_FVT-d`Ns&m
zeDFT&;z47}LUtQ9QBg{Y=r)d^8$WCZJ#f#SJ+tUf`82}L&_n*n$m#QB3s+Qfa*?+;
zY~u2gflJKs?PuyWBOBd}B{G+UM5-4IlAkV<ER=r?twHp*w>F(8tP%nS5rk2f7|yl7
zB|gwpPHnldp#4-up8i^VTO58-NmY6a6Ej1Ssq^*@A~~7$%9)F7XY4!p3xY(>&i3|d
zSa5`qsGGVv!R6(|D<-0c51+d;!|}~4r<T>4kTtM6%$r*(-+N6W8XHfxg^7Mg1)NPi
z$Mf%nUNqj@Mu>1(P)khOZN8-`doMyM@~nkz_;PUd9_rQaDl7lM(Y0dYI~Q~9gx9LG
zyMnuRD38Q};=P*rjlhOki-WSx4`FwbO`iCp&VFs&qV8m4xYoEe+q5H?e1jC?ZABuB
zRFAO8s%{N?Uh3Uf{`5(w7OZJz@Z_^X$$aZbDAMcg+5}%su8gJFXFnDF*Z<D+VX;Sb
zqJ&jksh`fB*0T>c$vjI9)%;Ly)}G<i=0=T{xeO%#S>kN>bW;NB*sf(I2Fpb9Rp8Ej
zVz!OE#>ownWop4xr0cWsHu>wW|5-o2bgrKm2D4fSYb^UK_`{d_tsxvDz2UoKJQRKX
zpMQ7LNZ5+(irl-4n_n37pO3sje}rr42ljBDjuwqE*~o~)QUwch{q!El3)dyfv^pni
zuZ@X#k2c5IYcgz;NhJT{UTUl6F1zv9FvctA+J@^3skq&B(sPR~INs?W2fXt)4#8GL
zd3o_XkgBwL_P!1ac~8$^<;DH=v}CV=?T9QRPTcxyk%+^N-j)Kd|CmkfK1Yv99uZkS
z7JzltbA`h=4yd(wPwYd0sf<G!GYzq2;vFmD<<?5R^#8^uc_R7UTO+?rp2V@OY6fpp
zW)a3$S`3`yoiG1-Mb76W`3VCT!(zDj2fWfUPhaf>A=iOHOqH&NkGD3`vQGDZgXS4I
zmlB^JB5zqWW3!wrNRRl|fkAo8;T<Bs1B2;dmTc#;J+R^s6Vphzqsqu@_quYu0|Sn6
z8`#MgAk4Zj+=)sI14bUelsT`la=pKN`D<&sS_$}Yl_So(Fr?I3>vnw}Giy>y#BrL2
z@H?#=**%A;PSv@UZP%aZf|B6@)nN}OQa#}>QEN7j(~lV$8J$R^94U`gG5>vU%@4<$
z6Y!g>6B-HDhSStL_!L&);|-r%<Y5s%FhvnF-57Y7!#PTbt!@l%>@@Mv^^Pzq1bq*N
z3{<3B@N&+hY?>W77OMc4dJDP--jge<18JHyjyVqr2%HXA6?E)7Y1W6bLJMoR`NhP=
zwaD>B@jmh=lJMF8!3wTlj@II`nXa;sy`A!tO-2cAkbFF4Tj73Uv;5}|9Y4S3OpVjf
ze3LIB0fC{@TbQtq50N*LLq>@Ou~BCT!A>cLxQ>1k>6vX$SN|@QlaW#J@Tgl{TB`WC
z@cQiHK=s}(7RtR*KHX^pZ0o%(YQ5mde*|31>1b#&#*cDxawKJBM$~|9{l#{zoG&*1
zPr7F?z2>ZkFyS3w(>xaZa@NcV7go{XgN4Hdtmn@zE;a_lDMwV)bTU#>o)SWjAFuB&
zv}o43mW({W82%2T)fGzQhW_9fjFQD*x&#OA8y#R>)?@_BKU~X7W}u-F1Az{{o|83P
zs3U;>x}I20^vA&|A2nf9u68jNDFfU4{1BX!5JNm&m@v|0DdX>75V?q#HZZr2psqie
zbr1aZP3hyuV%!!8ZMhB1t_wx3*5t_QaYVx-zkh!!QM2GrYE(H@fQOmV#-vlD9L{Cj
z^$0lBOJHDNfHOkd;Eb0TjsWjA2}J{g)F^h0)4w$F0G4)OQZE)AoT%X9Tf8IcCOtIV
z?1w}1)fN*&xRpx=ICx1Yc1JTykq?!b%1J0XgZ=Np%eUrh>CkqMUWRd*z&wjyqGr2h
zgGm-M<)@VFKW1G<M(N-TZJfwp+^uZ5PCK(Y;pM>QO|hMthDIWs-*Hvmuo`yC(xCO%
zFBG$0gSH-YRiUV)G?MZP6YX>~rt4k+-t@GEEk6W1<&Lq5NiIkn;ltSqR9mx9y%x@6
zrhZ^g^oWEc7w~wNz;D}_q5$((rN_~TT6p@DO>OWAu6JlGuU%?*c=+L{mhIYvX|jry
zRX%X>{PU6b@MuiWa_XnG#Yk?HY0e#3wyb|A{UXL`Hl^3;tD?qhcx<K1=C}e<<VJ8~
z4MW)I=K2bLx)4A}NSHB@CJtGiAFgGD)S@S`e)H!zJnR%zJ6<xh1v&Lvgq*NKJ^Y9X
z%<RlWekVzXm7al<IY@8wyuV)_ZDdCx=voZUf=Lql7vke>R#-CO!-uKZlpgXkwJvJD
zckqm!_yf~4u!^`n&mFH0hveW(!7mRHjhDaIvtbtRy}TMA>NUVs#|eAZybF5r18ohj
zudm;QM2>;u6cKZpm@PE>txs2%0Y5JbI|G=iMx$3fb&lm&VPvg$D0=FHE7V*(Pe=e;
zThr+T1hl}^sC-Hc(|gl|^UIX5Al|7Eoa@ofoL+%u#ULiWU<z1A^i)*71k4!1sjGAl
zDVSwRLhkEH?%R9Qj~_is-zaL-<>TW!-)+I?2LA5#U|p$-VSOk$J$02Z4PC{acaLV(
zD+H&8UfM+s=ee3ct>_^L0+Wk|ntGM;<}_fmKr1Vj(=_*$?;X{4Ectf3aNxHrHk2tN
zgtc3!Q;VKknV$m!+*$BRA)fqUcFa4t->J!YKZ?p`mA$ja$krJg9E?m#VyT_!yBi6k
zj6@)^!13OP1By2j%oK*FPeuLp>A1L*U(Ps2Y&TrkEH5vc_a*Q>&cd3<0dqM8SBzGH
z-)%qrw*+iB(xLI)M6rI0V*6)*e>e+K;&(i6Ta5;fYHD*Z)u8;MyQe9b*!SOu$8njb
zc64+owjXXysSQpXczs-_f{g?NckHfW+c{y0MMf2G{JXfg$`LPQs3o%Gl3ARBS{CSd
zD7<a$s<qJ}fZDu2<<;XXj1PXPw!2UK6;@(Wr>1ROf{kv?PoYoi5iv3GWR+bSb&mDa
zPw+nZh(Ks`h>78&m+jH|2(Uyc(CnN`9wLS{n1Bh*z{G?;)FX7*B!fMs;};RPu&{`Q
z{?;us38v0*J6KtM+ys*+1mAp`-qR!98_Owvak8TZW)<ojU?AY=C>tuc|Al%D9g=|G
zSV7?3U$?H#kCfEaV@C5e(!e*+LjX1oPNX@3LuY7|nMf#>5ENtVlIwZw`B9ty=?HsF
zi}4{UJDaY;Y~XeHA&9$XJG1rn)!N~GUP-4Q%yYXRTS?TZ!m^3|I}O-i9{p=WS=?Ti
zr}pjT)>Epi^@k7jn(lb?FSG=38+Sz-bh9UH^a<?z!6cpc<>2Jxwt=R3B@{ux_wn^*
zmc9Lp0SxQt8+;*fJv|)lhZ|#ZN}19Nf9b+TLeN>GtD~d8#%ZJ6Zt30DRHdAoo14XO
zb{GgQDHxb{Rs(@oN$->S7T^?JVbuB16XZki()Z`jpAQ#ky~Lvw*6xktF5{<w;V@~I
zzXzU@igOJwJdaM7Lbb@Ty?=v;i>rcc2z(OhL68S7Jp%&+>*FP9HdPCE5FtYt-yjM>
zy4o2qD+|HFwaD91))S?_4=&zINtN-Vy$s~3&~a3>+If3A2cZj>u(PuR75fixX(08b
zHF#dEF9cAkf{VRR#`|=Ls~|*kNqQ)9Xbt@2?~-X~Xe24x@BiskiRrtbp`~>{ma(_r
zcWY0+xY+%Tq5|KKh#AJfdlp6|ZZT16Ok>mC(INltoj>|cy3Bd1g>ymI9>K(%nTWPw
z3>pQn&hkV4sg&ukv4E>P-GVQ^@NQ%+NFea7iZL#O*1x!Irq!>nuR$sX7b_k@HgLQ(
zRpGj4fQ5ynk?;?+^uYi3qrETZ4iSPQ7*C*;c-z9YFaLld^s0vK3i$c?@qe86rp|Fc
z*{125>H5e^cK#<q&T*rllHtXRcTG)A7E?b9C|u_;XkNUCuHA}e))J(hmXwl;px&CS
zkOKF2U>TwJPn238k=m&{2(h=f?~AA5<W!PEu#7~~D>;Hhz{tQL3BC%ln5n6FA)Am5
zEJ=&BPAYOwY&A9GlDy8%D{N*o!5h%AQrGLuI8oU1+-E-~GY7men1owWK|vu`ud%@&
zk2DhmG9<iZOb@Exbgq3DOtL(&4bV{O)Rx6uLp`bn41)VU;gui=*Qj3k39y+01{-!J
zYXXQiumqa8zBrlt_3IZyyvyF=ZxU{cNV5ClB^wbD5m5;V`RL^ECg46A!GOCiWd&{D
zt(>;0H0h1YaNnt~9FM^x<uPmV$AceFm{vG%O&VY=RNuRIZ#d<ZucGZ#nQ171acO)G
zE7_*Vq;oLTk0>Z~!veI^a_w*8GIMC8qUFsxzCZijW;NHKGgWCj##mIBDihC>0}zVw
z3c!P71(mb8;BAj;y#`XnG|SDx&_0kAmSf-v`QRtFxVU({DwRnLoD+ykB^!M%h-YIM
z#1OaT=qf;HQArLUijwd<ntk2hI{|1k{Szj3G~K=z@YS4URM<{JW1aQMioC(0A;!qc
z4ltU-jnIuIL~X0GGZ!j+p8b{`&Jxh_<NE3XL1+983_n=-2s#&Jg>m;|JaT?zuws`@
za<aRG(VWpQEdb16-Vb#lk3BKuG6+FyowUirE4}eZt*6m1n3-h=Xk{a*xy|}TDLoIv
z?o$dgWew4Td^J6-O+!Ot{>vL991RQ2n|!dA`}myJm4RujIT%8I&+~F{iTei!73M=P
zkl_!3$qZN`F0l@{;2dtQPa2(8`=2eZtiVJ5?(OOtgusP3_HvKC=hu2$s9v&|O6ek#
z6LNI{05_eWp!VM0UeeWuBY%a{hU)LOHp%c13ZdMchKs`CErN#+%YmVzgsI%1%l1wV
zz%?RQhY<i$q!BWzngQTT^Rw}APzun2Ue{-)NGFk-%QER9rejWk<t*IM?6LP48wN1$
zmE2oR3ylkxylic4jah0BeJOH%YRJjVyj6D~9mCcXY4q;x+Zf<?%ixgEz084)M`9$&
zF;#A^w0NAJo^HVR6X2|ML9lR5z<UIj?o$Z5F5sWLxWP}httB~ZtCyn{+c)PL=Nv>p
zV#({jO5k&7<7@|k)^T7;tsT0qnA$RV_wHTJ0U4itE9dqnYzjIj=>acB+r}zw7ETtH
zKF-2&&pHQ`EOHB!EKCO_D(!~og;9qkj=qJ8^+%&xpf5WJCUT%oa#B)ZwFAmxy2=h9
z=&y^5i;>~sitA+HE;yk2qNwYSCoPQ&-5C>gNEWM6NfC~a^UV0ZY^3waj0?hTzcQ8T
z-8YYe-j~LGN)++EnE@2=_u=JM6Q%ch6=rfRZ)6ksRX}((&#45pNS@2iY<N0Ib<tHq
zw6yIG%G)vQhM>~P;yCDu<Id~BjZ%sIX9>F!0DniAd?r{jirolSHC5H<hYuefPFfb3
zk3+@!_wewJf;8&k6~GNwiuu*h(2&qZ{VcG=1uv)oq>iFxW0U{%=@SSO;0M6>11tB-
z(>)G{U`mney6SIMvvt)ycI_eLLjeAzfLWg>_U-Y#`}dEHN~X4I4*CRW5_O6yQc_Z;
zFUns#ZOzRg&Kt)nEY*-#H?s>+OMnofJ(yH&Td<_0<lf!8^e)n3Dl6>h5uR<GTL-^V
zj&X2slp4LRhKuw>uFjB9w0WGQgv5ZBZG8qh>V$I{b^HxnO{E_^c#yQ!$BYEoEI}59
zHD~JPw{DEkQQiK@6E6cWprWK?;PLj1)_M#W!%l~{8-5@8p+s;^M1l=3l*g{|@bOt)
zra);V5#GJbU^i9$+YVGM-HGi0`LFiBB7JGY<f1L6{GXHiOa(IDFELKsH{itKAWR%q
zHeSgLPNZ&@A%z)v3TwJJTi6)5SJFa-YqISt0Wo<K{hO<oq?VO&1#MDqL5jXBkUTxX
z3+Y|$>%Df$Z!?d{$+xO6uTJK=xq5M~zXk_09|+@92&&t>083xU{rG)22y6<AKT|85
ze0_WlK~A**n7*s43!HE`um?6$ull%vwS}1nuBIReU>D@_!BrBqIk-6^;h5l{yM&K`
z(|wBKJscb%1UU%b2)1bqY+6v=sZMx$?RU~H&w^y_j9gBwQ2j;*$~KUi!_xs;S_1{J
zB0_f>gB(g0X`BW=QrpZYMz$yhO-r;}im2Vc-;2&a$YtO`OjyfINB5peJS39ZbGHdw
zvmWjgHI2&GC@Tj5UeT73iD@`pB0})y>LS8Au~yc4fImNj4<MFV*QH>|7X`W_6%Zp8
zFk}YR1cKW0a<BayWJ*bLlyfbu)12o?l)zXu?FRw)q1lKZn!!Xfy<m#JMkWb?)=QBk
zz?6t*GZ>Eqi8<Z^w6kyZ(xCOrLx=kF^}M8R08Zir-AaKwG$84As%Mw&*ACZ)B&DT?
z%gu*avef}QQ6_wKsD!Ya!61NseZrJVSltMhV%gJf*eI+m2PW3&_(?JKm4KEDV89CN
zspC{xP1R_S)jDk|yKO2F?gnLnxrJ9+#6@-cfhc=myv+>|3Ua4&%h~c|dxnI|?7M2d
zy27?#SXfw6H>eJV09BGl%LcAw{`Wsf>ModV!-k<knNU>~FB+|Y1fR5w16;3xy*S#l
znpvZ?wX9d@WIBzpuqw4xyEp#j81T1UJ=>|A^U?w!TSZ>}J8sK&{U`e;Cu(oseuf{e
z{-90vPz=DQWO+FUDlhPc6oKukH=ycD;sdlxI!LDy;h?5FL_*hPtv{V>XB_^lLf|SN
z5fNp5|Nb}9@LOOY-~}R-)z!1X39A{M5CT9=mo?<}%6PAKr(uKM7kQ$T&VOuctmD2_
zLBjR{U?Y$i4BJDFE|b7F4`<4d7HC&1f;+H2QI<vBX~KB+lM7TcVEIN;quq(YfN<cp
zl;Hzhvu5E*1{MhDuB6O^1rv1)pq0f+Ppt8**O-or`R_o&z%O62!!P2&y`1&B*y6A)
zQY%UVh4L_%W{S2D2%|#3_04^Je9A%gy4nkc@*PLNsVLt`-a%=r)I*HjfQQtnDR13>
zf<ogTfX-I<I0OX+>xS)66wh?i%_yt13!&V9I-RcUP##!7-jlST&=hdqln2xUy4C}R
z;o!5L{0=KY08)W^FbBkF^WFI-5)Pv;;68tx<-@*DmK_v?i@xrg^RdQq=PgY%f(H?`
z=1;iP@Q*jxpm3GDx;Rlou;_Fpg0m9Z@1R;AD@r!$OUQv(8!r{bLn&0ii@-(w$qYbK
zxoSnau~XRC*p`6N&@BG&3H?TZud<7a<w4PoX3V(k7T@{>1XzNSg{XcfUoq;UlHoQM
zmN|gWn$`9hv+g@W=o12I6Mdkz5b`{B!?q4266x@|ZhU-P5w!;H4+xfd5U6S_&-Rxm
z0c{e+l!)dPhY9^P9LK@36J`GCc3^^(n7&WWpAITi&4j<A0ODc+|763<0i2$39FloV
zi-zKL_)=WpS4tb<<Hr`wpYJB!Y>(t2ob<u}gR#CtBTR3GN`v3`uC8{VF>uMx16rO9
zaRKlEm6nx7egB>kLM~td;A40?NHB~X9D|Fk;FI1i27qP)jlSy5O)H_JVI?#<)kTg%
zq0m+Husj2J4fKCO5e0)8%2CFFU)N}xarNh*fCng24HTLGmI|)SA?Qga8U<*@7#^GH
zZ1k|_sTK@!0%FE#rH2@uZ~^m!W=H&SiBo_563t6PUzlz$o6Jx-KRO7Z|C?Vdd5akV
eVa5_2-C!KMNn8=C4ShwwAR#9Ew(#`_pZ^E7WPjTL

delta 107601
zcmX_Iby!wO-v&`o5m8!Fr9nWXdnE-4X{5V5B@as114y^hNH-5HAl=>F-5uY6zTaQ$
zb*;OcGiT=h-8Fu1LYZwsDd^&YPKb!)OL;6WxtP{!95BQl-v6*#bE1B~>E3E_z=Jn0
zzxlq#!4eWOefdp^sz3E3Wl6{<e$+?gWT_|9VemS>d*=i9Kj>p%%wOjm4Lj93AM;*J
z!>d`$yW?f)mz!c084E4zEmp;q(Gt~<`Dof_P8Q{=PUJ#cX;(%I<^?2)#~{h~>XPt{
zZu6j39#f1I%9?A%fs4B3;i{15B=%9E4h|cxzqU?P9=`H?c0-1y4u0@I+)G*+DI|Hs
zIO5A&o{&D3)XO1t2;V|Db=cHUOIuqUE_>K#P~K<M@VWurRcdxU%kX0WVu7vcH=6<C
z=+9L$%F2S0Vq|+!$I>BT{Lt<W6MNkTrgx3w{j?I#&F~H8L!+RGAOW-Kem*Q*>;ls{
zxnuKdL?P*3yx7@k<A(~}5{%4+sV`~$!dt`0gnbIOea<$OPFQ+s&d_*}dE_S2W^B|Y
zt_ll^NY~n*`|rPd`)<ID+}=p^l@;!ZrlBrB6W<u52;L|2W9n~OPYArvtLuZxV}h1M
ziH?ej71Zh~9+;T?nLv(tn?%*`qp*#)?^x5+%3ra3(p%!FVoA0-uov`9(C?X_Yl@$F
zKvhvrC|-rKyf8b2cwK7uwX7t%HLT9Y>0m;aQl!Aq{G+OC2mX-S1uWOADpSc`pJqpj
zpnD0G@@-i`#?dYZGqiE^1i5*6sf9v=RCmKAk67c6K`{vh#Xbf)-^F$fLnV$8GqUYA
zleepi8=38mV#$lWGD$~zls73AZ+*I5n>DqSuRo)L&ysvBcNLd?`!=4>j>O~WJq*W-
zFxMDY9&+tU3}I3|rW~&JFH7|e4X%};EJKRkwf#)WuzV>A$r-b#5nuVg&);nNsnufL
zrG`-29MYUXqIXzjuxHGMV`lWI5cj%iaME83y~ZWKBma|;C?l!?Z!35SOVdF=#+}5J
zrFumg({27Sv86<*Jl*;ZmAA}=x?BB@m+-v{XH~l`f|EM7KEo#!1tzaPiz;RlCPfi{
zY#lyvubZKTZg<omQU5IV-IL!q(nR=YsNdWRYIgronva27nX&)dzhRAU8N<$r^eKHJ
z_wFbGi=rVmv-$LPK2md~#P!n|HwO99(JM9q9n^ElvjpT^k$XS-3=0Z=5%?IJDeXO9
zks_`iZVdiCaH6HZppaF8d`*hCyN@%7j}+;831Jk5p6`e_^I094wrBd~ORI)-)#eLJ
zha{-3jV51;AG=;5JQ!<x;xhXJpUpziBjWMx1LOYhdG8r6&qNsL)4x|yZb)SCd0uJ3
z=wEgSV-z}zAynVHZ%|$nT<G_r829QcJk8uWUZK2aMO;uC3*W)%eo~^8BDDgPCNprc
zrt!Aa`V}Q@g2x1u_mhT}fr-vff{#HkmoarF91Ra&MUHwp?K8=Epo>p`KZe`caxVU_
zqV3qzd6bg1cfNtAA5C5BNZUp{w(-M=Z4%8$Y$?d!;gQtQb2I7e@np)~ZV7qve~0&#
zIX?F>M|$TqF}9`l5i`VEHxb)`*wNbId*K|C9762;HN@=TGha6QgP~st2rm&t1>eg(
zQVhQG%=Mt%aV<|0<~1|UJ-*4$%ycg(ZuykOQbKU?Rg&Y0M6Mx^^-~0ssIKhpeOqhI
z2`Q|_rP+av7di29>)|O&KSsbiQdr&y%Y<oX!y76kQW!Ot8-gI)EjC#EAXS-bZw%6m
z#LM$Zd$EB18+=7&nGc(PrtZIi0=o{B!(yIEPULwwUUU%Q;l~Wl6WzYjy!~Pb%k9?y
z2RP}yPEy)WDHAu<Lz=l2n2qgFF#;Z5oQ|%(z<c|5<KJKZFg5YvZKPm2TY50cp<rI+
z?yR-5yhP;>;f&~C71$c4+9XmWkSsN=8<91uXQZY+L_$h2&M7kfcCQR!c52Hi!3Ga6
zrmvvULv_5)Gfbq6@yn*oko>gk@20`{7pTK~E6qEfix1UaOI=GYw#{_peP(wT6M8TC
znCdaYvO*d6>7q)Y%OUolv4mM9J$Phat`Ow>+b4$ruFenqVj`jAYd@R4x=%dX3ZK17
zrDzU`A`VC^<>}rIG8%EE;z8yW^k}9tu9~#&ez5hReyBW2-7&D!)N0pQYX9hEp6bJf
za*5z|s{D2=p}FxiS-e=K`Q8E#b@-&qg^ke-MxGj{yu{Fy1~}~k%~C&m3fc&#s$jG3
zxsWOvkEpQmJ^y6QomF9hBv#u^mhZ$!B)^zcV0Vjre2bE|Pb&C8jt2WW;e25xiw}XE
zoP2R*MN3QTY-CQ)PF~rX8)3N&?XX)<e*<yE!x!$BHH=?<2LFtOy{xm<PZ%Lqr?%W_
z%7gGh$}b<*zpX-{k8?<JM0O<t@jP-*9=fN8rg{4%CpG^6w~-q$pLC3@6eW!+ijbZw
zJk3oW;)=?{N;}_KQIM1t#;*d8)O5AscIHiU`*_kjs%U7-r?qJDsM8|W4QSTIOa(<&
zR2OGZ`72w3-IwNxyrpXTX=!A_2NUHMm>8#&pI>B!M}|W`WWF}nKSNnT1{*QbC&?9;
z6qBiDnVN=1695?DOx@S;Q-bhtte;em@sR}H<Zc-bI6gz?V?#G^m^w+UnHl4(xEB&f
zoZ^tD!oiJyww~}&Q6Fa*Y|@ap=PG4<ImAKJDknM)2QMDoz2DAazLD4MYHk;<`hHrL
zB7&L&TR02%GJ@gwPTz9^<p#Z%Q`4Gks23mBxiR@fHDBxxRa35CO0@+%-ap+PlIG@0
zeuYwYCFmq}FOB>UUaWF5>)W5H6~3O)CZ@`zA%3o=r)wQAID3u`zJC_qH0x@PK=rM&
z@|vILA4F~je_UEwMqJqVHHyP!XE@_@XA;C2_^mqLxO(71B>Sl=f-gGP55+`)dor)3
ze#&GSjTgzX$fXlRhzX_Wn^AT5ygq6+s<EDOq6msD&4u;fZ4BdtoWJN@;RD;5sgJTD
zNhx@Vj_S<da#1Iabn{$^!!bq1q&E`h7MutSSZ&toF^(<N+SYWcqK}|4eHUfBysiBX
zQQw;8z&4_`HZc)^iK!iA_zrGZ21$tk{@jy40ji+7N=#3pF6lDr#(Hx}Z2;R@4vT1^
zt%)rMP&6iGzoM-4*`_D$vxSQjZ;*4x`)DQhZ(>5CTJ(O3!OyI;Sw>(toohN=j!sHQ
zyy}jGC*EBq`;3DSher!fkH1iU*fY0~$r8iPE!&BJ6Q^BeEVD=8?7e;WE^C=4J6#qG
z240fZA)V}k6XSeYSpWPzaMYjv=aWeKtC;)t)A}lt(ti2-X+sw-*mm0p7O3B}rgXK!
z1Fv4jD|o24jHkVzDMbe3^_g?NeF@P4@dBTJJybQZpqomudu-$C_i(^sN!zYNTuOUf
z;CfLvO=n^Fd*4oC+Q<Ok!W>&k4jlD|jF$G=hlTFoce>}i$dVJVLO7AMVh`h%7F04`
zwX^YS$k=f!lvNd#D=0LJ<x1n+<L3Z+FIY?QHwoA0@Y=719NTm#0|#vpfsQ-(%80KQ
zp=I6O^`<gAH*8xwrg!gt8g)nly2f$PqMBeGj<t+Yw8VqZd}X0v7nu>}Mtb8DT1c(s
zQ%lOVR4K{9f8r{nE=)+|1gr1+Qyfc+1@^UP=Kp4S1&X<#>S6~x7_s+aY{rnwi5MSg
zfvXuWW~}$X54+^W(NiI>GZ*M|+^~pAc1H*vvEQtdKuy!y*jB|ws!!B7Tr;Gm!+q(^
zXf63ReK?P-NmXaAbI&2@%zB$lJEeiK&{lgf{w4QBVC4pJ4r^ZRlO#K=D1BEd`Zt=*
zi?E2={PX#t6X-#NKdp$OHQjF1uLrb`eV|Jc#}|dsDoRYdUk!{?n}0zCc9wdBO(VUL
zSXQ$WDAd&L*)lHur3>-X+$g2Wjj6bNIX`LsepVi%-MisGi9}l$tzKsBFv!KcefVJs
zNuFjL=HeqvL>Ur`)p$;_A*^G3F`oayKe(y0IpLbYys3BaSASt7ENJbfUti7`Ns@3<
z+*Ehetc9XU9a`8;GWnD>cnDdMDs)Hi*>?Aieshx@)-hiZHMV{er{F5HHhTV&l_qEN
zzjw2({*-X*#TKzY5jabj1(%GkrE*c^e#nGHg?;vjP?+}2m`R62Y%(d5g)0F+v2-aF
zY3S)~8Jz0~VSxHkYN&Wdm203W-T|i7&Ib=bgTees_ExL4{KnK`uQk{E>_&_8a!@TT
zc|agZM^m>QX7{mGEMHYNn<Dq@nh0S$`09sgoF|PnAb$EaiSMPcYGY5D%v$Iicc`(1
zei{W+TJCSFFF6L%ZLYm_Lqw=!xcSk(cT!yXaRsrrx(5qOs3SB&s_-VRP`AlTfn8m?
z1e$z!=0SHbVMDiz+i5yw)a!H~KDGu3{ZY-EtFk9SfiY)m!uV$;ujZ@~pz4X(UH3G@
zkNz~b6tZJaBfa1v;1$+#8q;#&izm*jOU|mLccZH7yTI^YY#QBj-9R+#gT~d$^(u&%
z2-xr)5?hxV`7HNa?azZQH6RuRC6+?zC;&SxgY7XCY7}en8ABm2Jc%wNO`^=}%6FFz
zzUt|q^z=F=3KNmcFP_INrVC*@a(H_4n%Q>S)qRkx_#F}OUWt=Ad7~-cXjF?Ev4i5d
zofD>X2AkH#oKLZXgSgpbFAnbaFnrO?KRn>8AnN&y9=X}Gc#X9QLP=BYdshT_H;bFf
zwb}1j?YmtnE+5Xgof0B)K}(~*RvxOe5tZX06<Y5%_YV4B<_OA=N`1$A{?PwhRt&jR
zkCJLOWd8S6{h9ybkMSt{hbscJArF5(H=1(r;{F^7J3&dg0)c!L8{LEU<AEAG_IUa?
zMEF_&U!cH>0M8`!)EJE@NHw_92{xb1pA%nQXm6B$z)ceoi+tVqi0hQ7ij970jIFMZ
zW8`Xnm}{la>3*hM1?x^rMR1<A(Tt)hZ&CbV^g(J%{EbMUSXwjONb1tN^T$`{b9_t$
zD$cPwP)rnDOxPbyF2*Yoti{E<9PO?do+RyJKpToKA3n^QCretqVe20L9ic8jthc$q
zR9Iw}<0W=?@Qf4x_hL$5un*!3<PgTKMzk~!;kpktt*%pjG<d|8hoHGI%)g@wa6q?7
zjcnig102~xa5Tm^uH}5W_@i%7FW1P`OY%8n1lBx=j|+*E>C(Hq2`FRe9cDwDFDf}7
z6LINRMtxRucGh*z!asHR6*Snxd9{nF!lAKvvIh9o<)^21TF1$BfYn-_H9V3D(`oDJ
zYH03$5rsU>Y&?!1p=_Uqc}tV7DE=~SU;I?6#~fp8cySXRVrh!>q&6xDMN9*~QkWC;
z3?jNLboFmZ19TD|i>0!ytZc1X+NEQvz$zuFP&p0XM&TIKJ27&JGcefd&*{%jzHzc2
zvpX4!iT*r0lLeVY-SJ*&+JAIO+5>Pi1=gp!da22VACJcSf4r>Pv}D;(zRo(!JwvsY
zIqbIENQ_slX;-tUKC8pljY<-cE=d(mpny1Dhg1i$myioZk5_1K3}XNY1xj0vJ(!+v
zKq@e3_OcC$?2kxjLJX`XjFx^|!f-(EC60o<e76IG(il6!(BX?EvKvwZJ!)k-MS5f_
za^~ogMfQ?3@=?x;`Jut2NHb?GO~X_Nv)kG5j32-H?k7w4Snl%}BT?NZNTXr6CZgN1
zU#A;!c5%2MmS!MGWC6!g6LYO^<?v%F6*MI~ppek6W<Fj-YI#tL4OpmH0BVKptO@<L
zx}dtSs+SR3VAtOEIbN<vZBR@pErJo-iq*+}mafrb%BQB<Q|Je9H;&W>M&nX%`zfFO
z)$lXsUpgq|K+4~{uC;*FsQn|s*3BF=CRF97#g+|8>NsWz7%Myf_^uy=mQcr7zs;Rd
zit*ZStcWe|7F;|N|F`Ik<F-VdKn7P>=@+Y`O=mL2AsSBDeXxs24|}zcUf_0y(xf^n
zYGm~uK2#JD0hHC|;={l|iM{<}28XHA_p*T!U2Tv=@GHQn<c!fS>WDwux7&TnvNZ0;
zE|2smC^+HIWqi#*3QNrv@kN^w82xPlciN<^<T>UE;uL?mJZX>ME%D_n&rVM-b$4vj
z=Y})0vwh45=wHpPO%E{tfo@HmW|ue|GmGa|Jx=NiQ#aV}K<zbLZywngPC!YP$@1le
zs<3m*?w2Z;SAHkG*VZG}R4#$u*_pe|dENf!C2t}lY%n#t?x~rL2_u3HmlV*<`~5ni
z6@>p?rn#D4ZJ{2Ut9K7NDE6n#C?%|8@?WyW7heA?`>6RuYQ+8#|3fPky9I*GJ4ogk
z)|nN^Q8_05d4V&aUC;bP;!SouDJ``3E|d~#0ro=KlyUiQlbuBv5SnY$djNo8kw_YS
z)IDQFB=>rg&3YzaMpK1+?-5nl3zxCMjFXiK_i@gCW$R)&XBNA}MAa%fH+hWyaH=ZW
zorJ=M$B^y;2er}|=ltgMDvX7ka&mUd;35Rifrvm+gH!SpD_B8*c9IW*m!yX53e4Cn
z2RocDMD_UkJ<%D?i0j3}Q&{5|yl=uqdC7IWV%n>oN37;2B{hYQxl$J8Dh!+^+mS->
zET^{1_x%G<_e_~$pDys_Ste=J{bXwE5r3(>;e0zof@knrF!SlO2UGIb<x*}}q6tbM
zq`s3>(vA4ckoo)kSHUtN3F5nzYx1y>NF1PDZK%d&nt88<(-dTeOEL`JAivb@sL~}@
z%e8g^Mw=?SvWHl%{&=-~Inl}d<WlAkK1M!aS}Ij=S;`8d;W&&d&y%&#@&_2<Oq%Pg
zV4?kZ%hB8^HPmcdVju6@oUNAjku|He6mQ`PcS4UE8ZAc6&|3!Z@CK4^<b-RnJtK<I
zGhn6$V+~mm6&KD6p(`^;vxCiJVD2e$@a`|}2qsOyJOU_@S|;J|2_4Vg1nBeeTUos#
zqI40v!DXA;d&VYff?Fwh?(N24+Em{fjvb&*y4p)3pC^5X0e8Fu$oItP%l&&$jw#V2
zbftE90&4qlc<y~OBgPg59ufZ2_tn)`czeF)*?QS3<<|>F^AoqS<d~jV)Z9rtO(%@>
z9!JTc7J|3bmlp)iD-*ENlCnBSwci%4PsRh|ohQ6E>Ksqe(t2!6{!8OeQy8f1sON%d
zR2{12X*~V?Cj-|tNhRQCp->vMTxf0HKC5`s)YR~mkIkheMyHAfTX7)v%t5br)PTBG
zQh!G=rM2hR7rse<GEb7l-QeI9nXE77%OiOBQ*H!LbIZaRmL<~*UIv%xOs5R}7@@s-
z;aKYIo+x#Fzkc3PDkLmIeGUI^kO{u{e+K!q8!h^*4fzQg6vR=gg@f2ZlJb&k4aWG{
zPjU3>Of!#A?Q29XX)(^#SHG+?o?oZiysB>ZX$erkv>N5CL+V>8n<=!)rK4N!ugxFb
zv$F6<wca(GO2|lutXep}J-vFbY)x23>{dc@Asu0IxJw`$;Kq<*SekR8b#2y?M)9Y>
zbxYCkiNxv=)R~`&7Wj)f*h@g5Qt}$dJt?Y_KT8*>E}e&IfmXR}LrB^7p~}q82i4VS
zE*JwH_2Olo*CpKWCArX<xYM2h=OLk0OSHRM-E+GmZdef4fc|Mb&B!P)N|?8<@54W@
znJOQNMl9UD|MkW{<g^MdF`NiHi6UxfhNittzSb37cTo`DEIrWf-XY^!;uLaxp-Zm*
zYJbq_DP3G^I-Cy3Qg?$e1(*GSNAXkJ)zD^*zOHUVIi%V+R7`#L*>6*g``Z?(lO=FE
zz#zHxmDTmR+n!TC9AizAuqc0Nc_deoCUk3~o<u#Ew1e@)bvzw5;xZ<tfl7rlI8}n9
zJ(}yIJth>SWmM(nq6N61w2I7-W!H2~^NpVa+-MU)jd|}0z}C6-Rydb3&5^x@t82bc
z0vHfCS7wIj@mssSTpt3W1RPiDPEk6p`0a7VZd1poAe&el>nnmxck1FggUy@i>J36?
zHU^X*Q_*J;>hKE-$V{XzKrBuFCK8)UIaiY7j>=tU|8ur{om(^H4QiEKBTM<JTei_=
zFzt6$8lJkT%qht-eQf4t^^nHY$d;vbxZz{6p^eNzUx*^ND6%_>B2(iM?l09TC$ca%
z>X6NcBxjxn&#*X1Yw@mpJJvEZ{pxe`Ho=6s&=zVMDkouFaj@PKN66(uqhecLoGWs5
zBCvB(T5<f8WjJVc&#3R!eUfQ^hk&P*E5C%BDfFw4e_PnD^Z0Aj#R%}9zQSlgpR#wE
zFc*lv>=$9)H?kR-<oLX%`j|uJI56&H04@v$M|^cv=$EC3iu@jXMnj)hRPR{84NLzl
zbm#k4wB&5Xe>n~!zctF@xvPVU-#d$n22xPc5}Bxo#82V(<z0Llw<GY6@v<|Ea#l1h
zjcrMHYiR7LYe^!gUFbbd67>&Ge#wVhD(2-E^B_?|nbC04#&+Z8rLo<Tn_%QKg7;R^
zh<RNXKN;T+XA2wb5CiXCR9w)|3qqS-^$h?=z|t^NLv|Q$gpQ#7n*XR>b_y^m9uz))
zfIV=_^Jl*iRs8%OlJk%gBa!NC5>6Q|k3>k;h|G;G9;1w2si?}aE1OYEN>r{@H=N^u
z|38^J2fuBT(7yvO_w|1i)(=cKB$6-QU(vZ`C(`6lAyQIIVcz}6q}gjPD+0ssBXpXo
z9(C&(FChP?Ean0?`6J|Y>japu_=%;>Jk{5Ch|Zt~!>PY=!2NODWaS5`+}fVi;o^^V
zXYkt;fZw-0^slkl1bHS?Ew1Q?#zcB%d0`K;!ZK;#v1P1yc#%Z53d%BjwQ%{LbkxxO
z`rgb-G0mW_h%}}6W?9z<;W#sHiRY`st_m#%a!x)yneVW2BRd0i_ic&_${d;pJ24_)
z&iC$q*`(C}s8BGqKR;mc_w29RfydI;{^4vR_Uq5*bQ4GECUu2}M0@PQ&FQ5|MvnF{
zJz7p1nb$|%cC<bT9x4#r#JCihfXkp?7L$s)y|KPk{KQelqI*H<?ja6xx_erOf9_|0
zMMd2ot<!t(6@~4&)ABCg)4t*&yHwW`twUngM>nvmw6WWf>g)9JAHzJ7vlXP8VTf+{
zGk}PYIf$R577MHsL96#TV{Omlc?x}sDUWwA_R*o!8Cuje#t*>)&A(!bRY*-Y<3)ow
zZ=G8ToTr>C{%BKZ|1sU<VXIpJqrd)&z^|?(7%;4L-kv{k2dDDfQmzYqt+=u_v>5rs
zKP%p%fDmE|gZgnH?#=xlaHw3&i*S4oHM=%e6aX@Uce=tT{8ZaeThvVfW2|k2y<iIh
z9QEQ+-BVY`?Jf&Ti&53`^E(JwfBSFi3<`j)_xW%*1#58ySdxLmjWfybH$s7pa>*jB
zqAWfiLF&1E^_}1q1UR5bg*9VzONQ$=<sPKsUw|c~>*R-Sj$I-t!Y+MhLfCBE!|?6=
z2QvrQ;Vji#h}sEGQ1GoM-+)dx1plp~8Z98$YmC5o++QQ=!pHcQIT<W4;aH8gNsB^X
z^$sfM(k=B<=$8kx+0xqNOG{x|8hMV=A^yaO+XHvSs=|_0KJ1M2De3H+>yfyINoI3T
z)453)ItR3mz(l;$c^Ew0n>X<(!T7IJl)b*44w>9R0Y4?KB+=r+3mxj#o}Wj)9JltA
z1eix6Hn{)gGB%YuYMj@V#N38dt)1R<b@cO5<i#dg`}@M&=Z|ShV-L7+Ly{ttmc9t}
zVU(D2BIr;^)NS<~IzNebuL(l|)ABh<nMQ_k&yLc6?-eirY6p7QASGB&1`9({PH65R
z@t|)1m~UMT(!1FigYczT?yI~7MK{KTEt9&2(Mj0v{H>oes05Zq!xtnVA=9zpinYHX
zL9cdre#iE@<{L+ZG3}2HBTW9jS;zmM3*B_`y<01gj_&8JJ;0{#25(jrnp8p4I*k)y
zd{+u-5B!pP$X&2TEa&Wy7^Lz}^H3IRIlcHM&MF)SSC0A`5HCPGnx?W}0N7YATX+&A
z9(~yFgFvWy{*WU}T0WNYd{d!&`}COrDJF5kyE?JL6RsT`qBV}w=le{U?7LF}z|W&4
z$)UQR>lhK<s)^5H@^R-U<_`O1>vpNB3Uo!@tX~GavnRl0*yi+7*7@LLg}Qp!yAmFW
zRjhfg1k5>FjXUBhNJRNh**YzrEbW9P#E1=SOUa$B?6aEFt)30`AY2|U1ww>M@^Ohy
zSE#qYsDCJB^Sp%bEas6zP+uet=IO@A_?a`tbp6Qcvo(q*PlC1~yaDG`GGyM}Cl5Z0
znrieWfK)f)*G_vV$=FR7`zNgQ+Nyga$kP%2*;0u~S46}?E`F#}joUpwpQ3F}kc~mC
zE26;rCk;tM$H=Uw(~Drcg5?8%dOkY(iS(3{;yNuLpT$_p2;T+)Yv+4RVv{#Ei9EH!
zEfGzo4@1ts%YOxVLc=1O^U!9rG)*NhtSVUwfUVe&`v(C-JI~5=XJ<(Dr+J`6XY|1z
z*4WcB%C&Ju*4nKO_=M@$Cz<e;qaQ4$u@me0&2#Y#<p2#%@(wcPXcAp@9@Y3ryk_8u
zG$XN}15Qq>_(KLZg}%t|&)j^~lF(*b+cW;_gZ4K&^EU=x*`0ikfn(d??lcvP(#|&?
z7i&o}(QjS{ev1#c)(Jqhsc<sIj$1ypURr%@x~cZXdG9Hhic?lFceQL$(TcA=GjGA3
z%5n;&!6s^U4g4BP%{5H-ih(VA{V$_ff61XMare@R?5?(yAUa^);(jhmYXBu2oF2P|
zR`k}yvmc_=Z10DiIsfVI3qp^4P@vssvN>HHC`+JkWxQaz$*ruTwd2*?8x8yd5D#3J
z1_^Z7*EpU{VLO-+UPFKqiN}>cwwQT(c|4`sfE0;AT!8hlg+i(7R8!Yjz|9_Gt<T`b
zSiSNK^PE?C<ormXM$qI1bk)!|6!3r+w%6E@ML2ql`FkVXX8(k(;J`Ehu7)&d>`~N1
zwN5tvy=Z98tn$fx8oZ;>VZXxU`iDmx;_lYT_imQhvC4%p{U^`gXFqy0DFNoST{rnH
zhDs~`QyR&M6<;V25Ka^Xj@lQ7wO`VddiM?>4oNGNGO|L93uj|9jPs#JrYgnJv+i`M
zLOvllMEfInm`tdgM}#f)j8X&X8?zI{zw9YNHvK$UJY}TU&s5_A`|Aap^}_grqOklc
z7UA6C-e0aHPH2wdw=Q{eeSz}!o|^$M4?1>oG#69yi(I?3C5fknjR&cLbiLwfp7ghO
zZF&0z+O&~3_SfLXbRVyOII`?|^sBGYf(pC~{L6aCU(z7W@c-}eLBjt*eS4Oly5p0;
zF2+*(SEO{2{om&~N2(7E7dsA$YnR|`b+5=7V-Lf>_y;^Lw6SgM`*V=zyrmEs5rh=X
z%@!<@7$>6O^v-WjS~opoc5xg+5W$ra?|PH1mw2A3B@3y$T%dq?AKqlx|9ydqj_z!O
z>DM}!yj--Dxjizlu>0_0SLNfDy7)+psI8oqp5LN(=KtGJQdr9?Aj?yBeSxnMzVxQG
zv9m#W$A#s6n#|nlykAOyK~RsL-iB{X8l8}kP;_kUdI|LIwC+Qi!ouphS4wu2y5nQx
z*(v(gjy|raJJ&$%<^Le3p}E#{X6mx@KA7Yi5wv_vhyd`!yK;Y}lo;a@yl7rd;v>JD
zv??l=>5GiiHaGHcZd`tnu={$ea^3P_RA%v;+|5tfEQ<yds|mC_`cGAyacgWqCI`;{
z;Ddi@ekSE)bu40t3(rGfg4uT8hR)J%6lihhBk+h{b1O(7dC;9t$&ajxj0oj-Ig|(A
z(;FPxV2KOQ-#SUzUNk~>vROkFPORjt?*c)SMTt}HdbS$+SnGBVfsG+)okr^J1S*K$
z8t(vq5?6b5Aoh*O8W9i~?-6h#*_w6>Cn{@1=CAm!zmc-Y#;N-aUQ0+C3naa_&N9eu
zw7?Nm)&ag3;ccw-q8gTyn(cBjcR)B=`TbT<)&FohUcxTT5D1=Y684U$nGu-orSnUE
z=D5FnF_mIoGh;XOVuQrE%9E2}h6)`?01|kzG3vGiKI1jlV5yq>?W%ib5>M9>?HVqA
z6k8Nc?%C<376I#k>%hRDCevrq+GL#f$w8x<Cq^P0Cz<MTE^<w`2kPt9>rol}GLm3{
zM%XBeVUW+vr*XbrMD%LNZ##H8`a{eAx&$}wKmEwTVmfMx%a)LnL#3{${8Fyo63DX+
zNl7xQA-S?JV5EHr=I~LJg{ouE6%8Ek7(%J=|CpU8FhAx|Nf0bAS5|;ym7{!52*9fW
zb*Ds1&u)uhZ-o$=+-JLq;Wx{@-5Q^=&{nzhcz`$I9~#qT){;Ghe0u2O|J)*>elicS
z?GeQG<s|L>NrARBCjNOzddm+VAo-V185rD?<*0dq`Btx?aZhtUfMgiHC345iZ2vVW
ze4SElk{A#14&TDp7&p(X@f>6JBK!x8KR#6DiP-qPVQIOyxR;eQXqgrvD;Lrtr6G>;
zW21?j*_1D@zwkD%`r=57EDWW`^EtW?v^WgO#mo~s?w1OUa#UJ@FyxzC5-gc6S8np=
z=AhT%RHtNvH^-ayuUc~We|iB1je2ZV{POkGhak?b;WMHN;_7W!Hqn4icd>KdOZqL1
zZ}2p<+x|&u!T?Xs`El*luA`Wkzm?f^FbHsRt)Xf)kKqzx-i$XR7xKZlb^$=26!Tsn
z0Y(e_)uwF*?~vyXdu0pQLLUJ&EUTRfEm<utF&zn@ws2+0cDU8`()q2g{<65EKu~r$
zsGaIyBjN7&NmOUKd*Q==(8Bel)EU7!|2SHiLpnT3NMRmZyW4_tyPx+H7B_@{_U@vX
z&+^Drd7^wb>W4cW0*D4--9F8wcyEJjxcDdlngtFDnr<d!7YkYiJC?83HzY~!&Y7eD
zBG9qyaA^gsN%T4`B~C{jU1K3Ph1b-flzyBwemPAY2%uwd9~8sryoH-fT~O}!(a`@Z
zgb<Kr4nWJf$)F8!S*?rnTq}d4d?Dp=i}D!fyKTSEPl)&)2#|(3tFCQ<+5pJd{QS}|
zyQin@x^*;7VPJ#jq2%jb#4^O3&if20rDdXRqs1MZtv|ERxrhy-#eL>%#a4*xpvXR4
zB_u*oclaE1CuSD@Wr=A2WA~6O7%Hr4$bIP@FF~rwVP}g1igm5x9~(L+A3GbMUveI%
zIc{tlGTZH#%q`h`MHqE16G|Iijp&%p<{9xA0TXBo>Sowx0QxF0hC3lNSe%r3@gZQ-
zWOUS#)st_4ps_NWmAKe$Tp#qey_1NspH6Dq{;EY|Xk~loi9^1AshV&^Q;X_@`X;fP
zD|d<9_A<Q>QWAq)QJE^t$lj-Rs{tB5wWVtQcc`5Eue{=!8J1&4gh)l-I?SaAq!ofb
zPNY0c9?Q#fHJhwi=zBz{e@XxfJ+TA`S%<3f#=plUyq^r}NX?+oG}QFWO(8(Tht852
z-{x%nZ4wA#|AR*UK~XeUWu%aZFG4%2pGEbuzY?ic(-4Jy$Lm^_M}5Ry;0J8!#~T|g
zi_+e_r>`l}mz_qG21>?dp6)1jP4h)!IxW9dHe2J%i~FoJP<!!KSjs7ZqISz9OYC+5
z9!ksqufJsn$t)~y_2J=tf9?<pYDjN3X=mcEP=sFHy6YexoSE`5xpmEW2^8D}t)yQe
zJVoQuT>5`^T{z+O{{H9J3`u$&`AJNA#}2?2Q7%8<{t60~r`p@P$$9Ze^-T?kEfi{u
z#r8rf%7XS{WTGQ5tJ9t#!fP9r8gEE)@$B=Awz0wycF*1ZoS**wSHOeC9W-VMav8T|
zVUF%~4OnBIV@lY+O%#9tyNwad1{PpC0gn&PY=;HzqnHZm7=D}SjVZx<i^jA8nYH|&
zN^MD`1f4<<60f9;w7gyWgZ|>A=}m%hjwq0nXPdbqRGmH0pF(NMaac@noW+eb$_O^D
zuFcUYO}BZ1RDHJwoS3*_BEVsGb^i0mh=2CNrxAO^y9ec;><%BB@euoy=K8VKw@PX0
z^m~KqHz*0QzTgqVv{g<rl?%&`=r^P??WM)8OnAoHTl4p4%k=JIUb#Met{In-Gt?W5
zZ=rV!QTYbN1_MPjBT8X;s!z1%I#H-Qd(yU69Rs6@d_j?8DYgp1n<Da-rlKtS%&$)*
z0mr7A&y4-d@qFj&n>$c8hW#^G1QUiT994j4FQB5_G@UOT3S{A$5^$A$AeZ>Sicg(%
zX25oCaUiU79*}`;szV;|T$DzA;^yX|a+w1F&M-vT@?;s4)9VFXKiI=u+aQ*#mwZvd
z27xIR6(KNm`c%;9_#rdJ(hX$ctsVu^)LgZyeqSnM%ES21=Jd?_2hXq8OB`@^8is#+
zf#ByBI>TFMtC|-L$cruxUI`|Sy9?{oLn0l)$qF;q*%t5BpPJM<qqy_-<TnnEr-`vt
zvhIjf_3ePhPG|m1wS3xW>@A4|^|@VBCVx;U$(P9iY7KO8wM@-hu5)SOR$KgN^3-?h
zarnQwpL9kR+LKf5E=R(6f{-|zu3>ae%-u$^`NpR|=;*GUe?WJYzk#^r0=G!EJ*^8y
z=6CwjY|ci#6@Q3U*sGn<UKsg-RUyuxP_z4042qtN#a0VVzq`nqg;&E<Dyfv|CKMpx
zO{F)U2Gf>WpUjBAA>1WlMHP}BEz`F`ll%KPzdQ(wDDEC?og28=wKMo_$z#EmrZ-Md
z-h=D*(s~WRE4B3(y#J~T(AqT@&fMKUUfKG?Rp4M<XL1dgIRY5mVk`J8!s*Q0msGgd
zm*Jql7WgIgxQrFp9=?Gkg7I<3P2*|sZb#7JmE1jNd-!zJ|J}!c^{3fC{u=OGAr^*z
zPg(5`h3uGGO**KfZ0g5_`y(agi4;2Omt1G~FRXrIE$L~Rnu>KC*Err>Ah{GSo2G$1
z2;xoP{973fi{#@^DL<)*Vfc{GeO}3Olj~%wr@iJ|W1#X*2J?exVBWuClpsqO_TRj7
zzE*o@vHS3W$<6r+7-o+jTU%P2*|xaaTR<5lC7zung5tTVAa(`R8lFIjVAKYvb>B$t
znZ-~FI&e>cVhBsb2nGas`s{8M1&{s*5*7Jlqg+VcoGrA<bpFXd`d4p;lyVvGx5qY@
zie}S2_)jf(TlAJ}^2>?;<xjmGrMt41r)wsvHCW<STF@*1WT7;fp9&@gHr58C9t@1j
zg=<bdvNEboF$)le%NN?vsGjeUgg(kCzq|X;fV#8Y<arGID(P$%r>+#L$5A~f7`ChE
z`~t9$NJ0@Hy;+axzpxsK`VfQ6M~590{hVfGU1e3}>a$X~I2^ywg1%B%;&=DvdIT;M
zHWx_QE?s9wy2NxMyL)TFSkurGlpf(Rk3a$@=Tt4=xK@Y&2$8&^BWGcu1L8{tft!e?
z^6j?a-|{=EEU(tr9VH1zo4f{mK1c)txr<8RY@o_hZ8W2Um*5oK+&^Ah*74VoVf=^q
zl0eo(U)f4w=;R}{^W{-P{`^KpkJ|R6FH624*STHbNLBAsHxfI++YhMx8|KvN_BPmT
zK$Yi>0}ltU|E^8TSnV+_cU;MQVHU~lWJ%bj2C75Bjl-j$SfoSkqm<u@)9H(+Tc;0b
zRYWMbuUJ9d^1Un?E!mEPze2~g8SUq*DXvtCKjBr0Sh%92a7f4WY|92zi(3|C`F_{Q
z-Y91t9s_%=!gBp(+w-sNjt(OnQVmN22w+X+LAyQ`Xfy=ixX-aIq9q();Vy?@*kJ&&
zMyPf;v;C9vkFa;4@t*iMbK;Pb8@*JDXCR(yIkBj3GQ|ho{_f~DrnihN`Tr|cL|zg=
zezz0-*BdHLRMd1nP{jn;XV|ZYfi@A`_)|+n{0$MPM3K_2(SVHLnVTUs+=xx7%n~0Y
zpN=ORQxS<7nQxQwXCQC)gt~irD8ejj?;hbMO-GMa{G?L09?<ax9BoMF7Mkwg7T=@W
z24XJvzlr591Co$i=i1#4F&O+XES)uaERgWS1FZtj<Ll%8fH`eeLNW4<5AXQi#EV(h
zE2KjuIH<HL35hzcpF*M<qjq>DXRm`0zMP_^SgnXy?XCG`mC}kCJ>oE81HZ5_f|L*A
zxqYmII$ph}!Sent)6QVbP&JWr&|*WDhbq@zTsNA!kV^Z}Y$>z^4YU9~bm-5pCHpND
z{s`Pk029MEE5ckwNibpD7u2SNJuW#6Zo<@A<THASD95;M9*{o}ik@(q9k}?6e~Q-9
z-bR3=WAoTr=7S%%=wKF_D(a;d=sQ`xofRRz-WJN4I@0KhvxpB)rn?BBV`5^eIk>BW
zt_q}4`CFerTJtzU(yOC|q!!hC$e0x3nOoMjWoL3`ik}>){Q2G7c!h<9@$gz_9517=
zv9YCPWXkG(6c#e#2f6Vb;GC2<H1Pk}RC{(~d%fs39rF%BNE*}!f<O(wsU5y%))Iw-
zYdO~=#c|c5`Uxxc!eJqWq$59s4p&jR;bOoW6QcqmU{Y~#;wgPRvV3Ju@`Fk+ak{=Q
zLP|C~NwhBXW+4sE2WkS_oFlPmJ4N<u+;8O?u^8}{@2_(;sdafhP@EIy1(z*87hJAF
z)*arUJ2dU6g%HIL)12!KnU=b-ZduMuHuPiPBWQ7RO9cD?6c+pprcc+26<nObWZ~w6
zelh;}^C!@i@5#rOX8CTbqfwftH>d2pxtztCfA+*`oXb-zJa9EfE?Ms9j*A@BGW6Qb
zkI3H!k-7sDbghyw&HY~^MuVmBX%=}Oo4DX&$L3(5AUJSuep)0mzN^_f1+y2H^W~do
z2ci<Y@<C@hhZ5(+s=R<3h-CiE>_Z0!Z{IRE_0|`ZeSFShn0PYaP@NbRC8DlQj1U<a
zX}8X`*g^t{fBN){fYmf}OUCmVORD*snA=0s<ob$_&&NRn4+Je>L?*ssI-V&69Zg2&
zTL&8y`j~*xRAAF`>e!Z3)m;jXCI3AKorU#|R*v5e=G*r&<-mUhC5GC3ds}R;cQYKZ
zspNdJtTp>pLa$2=6Yb5R*Yob%KldfBabBUf{d*?hl*ZGTm$*0Gmp)P%)}SM7k2d~x
zKlevfo{N0eAoJQgzvsm;9l!=aZEv)Esl{1tD`9MuW5D24=eE?jJ9Qs)>$9<vVx9@l
zS}-8)wjS(2YeSQ{^gJ4PFtGczP-^FzW4h!~6|i78+_$VH_R#RNL2Q-P3hCJSa8IV3
z@wwz@sfoUW?a|2x;7U*D?8tF%#Q^CdnL*L9S3WgeB+|m|fRcv(SymCOJ6>N6OVfaA
zNgI*P?!}XHv-3BIleTp-e2xIhjr@?2be~zvv~NZ0@Q?=Z2_<f#$~E=X&KeVP;$2ry
z<|#nm%kA2U6>OGrA{Dvl9(2@<$g9Rqir$8z{sYbQuMV+a!x&qRb1BIR_En??Q4|F{
z#%ipnDI#G&V55Q`5wOyu-ISg1Bk=SJE{qA2n}PT_CHwOhs>m-{)*&qCT6|G0bLF)8
z{$znuo4%L+s6~jr5BX$uSQNbAHRLJ0PK53P9TedHiB8&l$vDP6MX9CsKI~{NCB}Vy
zm~sZyrjrFrn$RCxyc<S<n%TCQ<MqD8QkT9uDUwYI0gH#@jVwxc9hjz?f2>SSOj??E
z4p$PP*AD7ub5LVZd=4(LXzh&Q2xaT_O<-DZnNJm+d@51C^r`cT2y=*7u`->E(XUBL
zdCLdxSKv(BPXdt>Mf7<14^HKTyd!ZA+H6dbM}`8#R=&P>QaChIgY@4--;S2~2&i9z
z*mHhri_yBcAGDe}*6H)x>AeW&RCnFpY}OP&R!8L5{k&z>BQTJ|CsU0-*6~NRc{p}X
zJi)gv39@bC>qhCUK~q(@d&9SCXBQM~6UGuGBm-Cm=$l;41TL&fNrlC!U%#wY0VNdu
zmT$!1g2a7U6C2{|Yw{e`^=9G$MRW}^3bK1DCkiEmr^XQrEZ!8h35RG1h#D3GDv*rw
z6!~4PcJT3-<=s`4h@$g}A|6{c7pJ+=iU3ItIT{u>f$7=Vei3$87ey}}|HlW90I>sI
z=GrUuLAW}IVBjUtBd`Zn_KZ+zFWWBrcsAOvI~vV7djuGqPxp`ka|?~PL;_Cdl+6^S
zbx}Vsih~NhczAh|pni_vnO-)ywJmwe@pZRAPpz!$gZ0_f=@~>r7G30NH42+Ae=r3T
zra#o9hdoq3;UsozEU@1D-8$&~ZIcWTGfdn{pYO@sgT_eXJiENPK3)1Z<qN=n2n7-L
z+MtPC?(_hiS`u+!ZU&}`el93uE~#=<VyvX|fTk=gCI})JpXV=<CH1SSi0kyUL{N1h
z*qt4bEin6Z7BR*`XWJr*050!+{q-G%>N>7Q(Rot&2^qaDI-8{WPpzoR!lOGn=*jz6
zarlbChn^gnGAZYCZ@URdXQ5HGc4Y0Ir-hJ$dl(Bt+lWsBZTYHOCx?+h1D#AHfnvKE
z(wnS_y;^Vr>LaR~_<@PQZf?WylbiHp(+9ih_W@urS#~C%Yh|I+$8-8BsQ81RVGw@`
z9lOZh$Q$&8*k03rkHre833Fyq_{>sXHo9xd^H@NZ_PnOx>uJH2r)h%J{1O7Fvq;l7
z*S7|O45m+ikPFIXJ^>mgbAaFx#BptqV6=OWnPd<rznhov$_@7w0qiy+@!S<yYUVhr
z7&yv$e;T~~O-#1jZPQ_^RS4_{v`e3jh?e3ms(Jns%dOpiF?+K5KVO1|925r-2u>87
zmO#YT+%O#mpbQYd0-MW~^|91bDg@K1!#I@s;TmixOFBao7ECiTD<J_B>NEcNQ|NP;
z0ts3NeqS4#N&e3FdD5>%E7=TS_kv0a1;?vj;9`q^atIc#Uuf15Us;@`*)%$souGa8
z>eb~l$2G=V@u0mFn5Ltn8hiYvT)r*W;JX~k%mZ*HkgFMC*ZF+^ZNfLmG=eB8IWU8u
zP{c?{)-NSysu>EK*U<QPnIcT4x8+CmqYdi$s^z>Xffa-6RDYwJ;eYN(gp8`HJyon<
zViz`uBAtuENGpil(p5d!D6=;;X<g&k%a;*byWh;TwzX9|x<|^hxDZvwq$E48r>OL7
z4<8J*<o0lIr5q=!0k#hP5qhA+sZcHN<L4!ihJIK%z@?~|$!yF$&iXsWENq-Ut^aXR
z`!RD~_l-uYR1dg=b36J3|M~LC#2p=Ni}Py;-!YS0L4&$~oS3rQ8<2TaOr`*urFYEC
z$*C0~-w3Y=6usbj9Kg<vs?yhzw;%ni#nP=~q*hA}>1w$id@9ZP?(cz#`ihxACgA4>
zRC{kg2)pS1yUHi8W3!QCmU6Y-7@Kmcy-7+o1nInl)21WHHi5PKw%Dq%&}!u|dS}%U
zhsyZibTJe2yWhR7sm(L+O8!2Yri5U!GqE=uZJ!OFuz<Ng+gJ98mH26?n!1;GPP;;z
z;tFJ-YiyF|K!0b9RsMIUkpy{cSq{leB5OZ_N`u%({|+@a=6C6?!E@aElib{Ka@z*S
zI0|KpD|%tioe@tWJ#6!G+4%zE2MG|^+1aQZrrL&3-482JIeV-=I&BKMKSt**cCZym
zQ%Dh9^57q1MSE=d^1<+*_k!wwL~|iDtq>&c#p{#xnas-~NAqFz84md5WuW%Wb)f%2
zaYexe6S(Y2%D`0Hzz&Lw%1>RecZj8aC`9p81y-meNpJCRMTmKYW#bk~&n>36-1e1h
z@>F)>Wr<vHAHO?buQbJdwGkW=7ppdhm!ZFH>umhiEroL=i|5M_GX9$cQ>axGmhzSj
z4Mf@rHfb|1D>-T#=9ubt`2@{hR2d`%C6vqEH930=)yM9IF?8x%-kJ;37&=TUkF8}z
zgE^!&z0~e+&$Rj>#o?W;W^=P1=`hD~L3FF*5Vwk(T2oN5iXq1;OM*!tyZ}>3<(r?;
zpX`;`@@*I}U?E_U$|{Fzm<JEEw}UP$%Q_ag<ymoMa-)U-5;zwX2g<`vt%`88W8vLv
z?1v0omtFmZkBD=<0mFP2^z#ug-O1E=#6Syzs;;FGru!wA@vy*SK~DvzWR3e%hVl+u
zF6W_8?(j1kH<a;yZIG-2`$P9}QP}&mhAdIGc*xaB+k62Af`8!t0W=AD4*8{ICoauT
zGpwC7AR~an9q>Yga)}F&d;eOreox2v0S{e`2f=$}-Tn~yoU_2`LKilc?($c$2W|?z
z{Y9pFPJS|F+E=>BqH~3&ZVMukmnBgsQQ}#$v`%q{rlPIhDS}B;-VWt0woo@FQPEPq
z%bhE6^tUGD>-3tBA?`$-SZs7GC0dEI-~N!>#fjT%ygqYs6x=*&c<Lh1T6Si7X7{}B
zId04m2tR|w&Euf?fSux{ksi^9e1g`3FJVDda&TdxK5Ms?7KD-w){+KB(A~0K<@l!K
z46XEk%{OBC;uHa`Al_cTMyN#p@+{)lyzF2l+LwZy-&L>iUmv_G#shQ0mhk2gpb!x(
z4cFu#%w|$qvEtBt(OLK8ituG^NID(Ym4U*pveH<8WygQrgcE;4Ns?Z^He~TnP6OHu
z6drg%YDgRSvC=*sR4YPkkEkYiR|CTw*nVu<WSUJu$lR^Rr5aXRC$GW1Ew#C%xT-es
zhotL=6rFE!_I8+mApaVam<@Iwz3Qthg77v^q0K1=TRHiW`nj2sLv>MfGTVpGejQTy
z$_P=@jJ|)4eMpC6nvZ=!scoT&f^|RD;UUX?jF-<Il0RCji87iY!Tsb(7AAfV@whm$
zLv(r420E#+@#)+gD@GSbySHNuW5|10D^79s;hhG|IjE{G(F+n>#SSQ?*b=r~n6;25
zRKIwtRkJOeHazGEn@qh*CI0M_l7aeaN`;j_{O%hnst=q^N8EzsdY9?Dy}5X2b{qmP
zgAZgW8i7B%>P508WQ=U>OGu8fQkFLxyP$kJ6#Lwzu#t2v*0}5`pP(QfuwfB2M}mX3
za%l2x;#*j|DA~>sK*=RT{i<->3X{a`Ik=9u**-$m&HH)fN;hfxV?$(5L8*i8<{+fV
zwY^-3KDWJuQuyanW`_KY&x^B7`pfaBa8qmC`rkb6ys~wgt(|&?W<*-DPZ7d0pgNfq
z_I$mS;FHBf5^1B7LuXAyB;xDB@kM%fsZoi^^Q2tc>ZKXE2wdyqUVOrM)fXIVZM4|-
zrQ>#MB7(uMzzsbz^PxDq>1o};;tih{(OneHGbiGu*z1bpeWi?I)*}jN3?<((8h=S`
z-xI8;N7>RUVOV;QcDdim?f-5+A5yJndAKu^Sl><8D#g<oibgW*=$!KNyBZ%a@_#oi
z<^s97&fk>o;E4MB{o5<ERzynWhoE?_0xNdFYc$9&uJFkWkYaC4QV#aWCQ(svl^Q%3
zVu3+t18ATbZ5^gPdfZhL{X`z8B^smrJ5OCVvRoj1#fX`8h%)9|BgFaEc>m3>cGR)A
z7R*atFK~7;s1EMO@e&8LWv*<CXf$>;mX8;*mK}dNzkg4^!D&h&i8@I|75-w<@d;AK
z%{>}y<Dlar)MmAwd;H28nWx*%Gb1>Vo_gAF?-s1wi0cc5X7kV#frvr5aNwXdI`7r1
zGW(~5O4siw9d0!D?De6rP&K~Jmsg|u-;_&M%0#7@wZkK6yXcmKZ``skcg5Kciwi7r
z6&?h$_-IXs>*5y9!yg3U>Z)bRP3|}7GIP@P1>6Y8bjk#<vAj6iJUvD%lG^%Rr$ZGQ
zcFv~#<L{1)B)h#B<ZGfT<VaoT@>5RsY4kB>nWUWxd+TqA$GsNa1*KjpsT;nTY`QWw
z!L`@l0!q{K;)mM9)oh;0IZgO0Z1^;OR|j^i-2@(Ex~X=84BM{rTgnTu=Zh1z>yAth
zjP`QA<Z``uq+&c>Z%RSE?4VFR?NQWVi#){|5gKNa_k;(BKuG(Nmv9Pcjpr!9INLrO
z6N~Y?y+9&VughGpnyt7|vQ=un*5J^qDF4iU01GU%?8=t2_&EaGjeTkcS&_wszp!t?
z_P0jP-NhRm;7N=g4`C%k&d36KMnMzd%V`DIDWQZj*Vx^8iQG58txdNG^FB4POzf>C
z$wesCQ)#Tqwg|P!)rWQcjG?0}Cz<Qq)(OJUQG?DzJm0M!Wr2Om>dKku(V_7>Ge6TR
zZa#h>#}&fMIKK1@E2qWejrl4}nXrD#wymy0{_ew|fa%(A-zJ}Ys|x&6<rwVL^}Kgm
zKem+kKPR!*Ru`;8?5xQxpAuayz^%{NPE4*-S4kCn7iT_dE7~lo?gmrVpZRZIF+k*z
z74jDQ7YDh1pE*&A$HX&;1spfx5^(;c_8B^EU-7L%9ctmrNqR?I9!e*_EioMEQ<)iF
zgh>bYeG;`g7R4xY>uzUrE?x`UlZ(K9yPTL2bJou`bUe+obE$2cAY9`<obDzHmn2>J
zQEH`m(EdP9ez?Z$Pid>}yCS!-s7&zpKfG?a%vZO`98!MWLL<6oSLZq%GWU5Y{r7_7
za*f{T7?+jdk*b?>9yv->9e;O>;o*{yCabLUcY*a#yUp`C>Dzym<umf^S$}MK?UbLz
ze@ApQd1NWm^<T-Qf5<jYMamy)@Mm~ax0hW&W5e!!pDDE`B?50O@s(AGP$HqaPK%!M
z*T+30Z4$ihwaIwJV}e3fI0opsuZ@2TTwc4Fyu&D)ouCLo!Ql(|$aIqJTF63O^1m{{
z!@CWRCdONd1O6$`>Y|K#C>Y>#-VGTIwieG01*sP8MHyX2heV?Z;dIK4VXcytHwQZU
zOQx+Sl*zRVf`gF5O1g{=YaBg({?M4^PT0=dBAA9+d10x+a5mqwG3WbyZk#@Yq#GZ2
zinjSXFH;OGKLs35UNosz9@pd!!zCz4`?#(JbBFt@Cp<a-968uas-KpV2=9N@BOEVp
zOLo=RzzE(<dCL%>pL@Qrh^@Tic#qUvz+gpuvNlLAJc<>HsV(I>4_>|^(E*W@!C2`U
z+tW8t3{r;%GJQ?09jkpH{G=`rn&luQRLhfTiV_=l;Ln=gnnNb8JeL|ALE>c>njaR3
ze$q@aP+oRrHnnF&RNw62wsM-Tq#fD)N`H-T_FZsj4V^0qp-_@hW8c_>dTz*F7=j;k
z3kwjyg~=F9D+Fc}JvFH|{ah4;knk_9$$RRx?_J6+vHbQfPLHoLM7i;UYJc4;9`h7S
z<If$Qox`l(oycIizBbq(iV{oHuFpN`bzV&%94+;Bo%lbVzA`GS_vso`N=iXMS_J9t
zMq0W-y1TnMq)1D5OL^0cba!`mcQ?ETfB)zG;94vfg7<ain%R5Lo;k`kTSC;g9J^rp
zU1)rOH=hn8coan+)rU8$GDOI4IA8`9hEt^7;4nW&9zZ9#3g;_$TqIuz{t}@zY?p9o
zQ<ZRpO<|^&CWD6_9vUMgpzdh)^909V+eYH){0dE8y0gN#Uw-Y%@-rDz0EmYcc#>Cf
zM|^vDiFY_`EW$j~O5z5xDf+^~l}S?%Sw{MT`fl{zj-T0}URCqf!mi;T4gVfCj&h|A
z(BDS|qAJOJvIaaTB`9sq>{^Lu0#AJKO?fJrP>!TkxHSIJ0uf?_B{xbsy1a~x6|+D&
zYYYr06)t~L(UNI>`j5p<AJ-@upNr6BOQy4@Rh9!4HJ*pGQ86%BuAMC2p4pAbnmf#J
zsK&dLIiIU+^-Dv>ZQ-hSdJqXqAhwc%sX+J7wDzDKRZmA-rC5T}gSzCzI3(|EckfAC
zjub9uRcwms4i^2Q1EepmJ1bK*+qA@oNQ1uW>O9(l=U3LPJ@siu8=5j;05c>H=1*cd
zbC`#AH?$4?<v?UQi^0{3qrv%3ER*w!os9+2UE6qn+gzu3uyL2;(8x`?zHW{I1BgX_
z`1p9pBIq|vQ_Nw+^zp(iRAhY;MVL8jZs|E$67aOoC@c&;JIfufE+{WQ%ewJ@H%%yZ
zG8GL5LbfT6wj3k^Xkw^~1c^TRq>-JLl>uq=ZWc+21}a<ix%C*%JtvsbWmV;VWLLNs
zJ?cTSM@<-#L$@i7_n&KeEtmm~gkzD~rz*8g!(Y87kAdz+(W`AKmpslQere@npYc)`
zsfz`g)R@)2tU5o{7TG>Mm{+k%J^kf+b_U%@llQk~6z1D1=0m9x*>zh!aqRIGV<s$N
zUJFx<%GmjokdC2F=-9ufdXM+2mV4Vm%)GI070s$N5U1<FPxpN8DFi6oB~QiZy%%6c
zdHGPOsrW-fQYF5v@~p_R1GbIy$%8TR@!6ARbPcmuY`K5_{6S2&*mwAAekX{h+t}A9
z4rPIJLn!H~1T#PNFt9|^|C6uQFY0qnpERU`=+kc&&HJUU7WKS$6IrR9_HHh>3JLMT
zs{A<N1pJrbMa(h<eZa-ZmG#NGRj|i{FEsg}M73M-`SwshThE%kF>B&XjVDY!>yrqW
zEH<`(r$bOCR}QfR435=zxGc~P|6WvZHvQOIlvH(r@DFJ=*&TU2_$5C$8q<Hw|MRD<
zt1EXDouis{>N42z{94h<KuL{IvYIh4u(@W^mW4TbROn&P1PCwU`Z~7sM%+FG`b+B}
z!GU1E5B@IuUotOz0UU?sO^i*fKQ7609Pd@MTwSqo!eCRUg1x4*dLf68fT}9Cp`oF7
z6kTbRCFe8ijsvII*<0JxXp6X)G8tf@evXdFcSl#JY77lt0(RGctSZ}#!cVEcI|RYo
zuD9!3htvRAQeN}iqR<{y)o7Z`X2Rt~5Rx43z^Yx<H7}j!be6z<YS6rP8@8pfJ-7G~
zZ~cy=AdrGevb%M%WXV1!jmw#4Uft<v4b@^B;+(c}aq8z?P9D>k%*T3Vr5LyIE72`@
zGK5d}a!qDu;@pVf>00!Se|3Sg2Xc|&Jpt73PU}*@jKEWvL6yvTb!8(MknR5gFO{-R
zII2<iUIG64%@mL&z(W$Zx^d_mk}lf*X8NnNMUeg?Zb%PLYQcj*_A6^#n-Lw#k!W{L
zzlCw^gC{2*%Rpg9kD(4yXQcmX6F=HZ3s{Jl2?yEZe91zY=I*0+#?Qz~Z06U^tbQZF
z&eL9yR*Y-P_-F`6`4JCmiAGCgLJ{}%lbt{t*AzPt@szE4eUFtgb4o+wzMSWwQW-vA
zgihw<l{dg<Bbi#$j9H*a`4{!$;GXcgSm<5&PxbwpcN(@=6pRtpoxPoYtf_YsoL-k%
zw;q>A(Fu|w4FdI}UNZ`BoIcu)3RoKg0lOrGH-7ErOQcT2ZcjW_a59Zg=!<t@g({)Q
zH2K2r9*@b<i5}h!F@7#@9E+#9^RAvkdu4?Uj%`gGrRv?T)zIIK;)&8>iAQ{BpS6Rp
zELj%Xw{Nk?-OP{<n>)*p9x$pNQg5AdrHV->*3i*l`stvA8NfwXf#$T)y5<lL6o(MG
zJ|mD0r1Q&8pFq%SUYL^*apXUUHTrYFh)H`WwRdiruY0OD+GpIWt9fw~>y_Q9rEuPS
zTHaws_fNHxcg9!GT`8uoU9aywZ(u9)od{E{aiqU~GnbdKAmec2XieQu%aB^A`}I%}
zXfoRbSH!iXxP4fnR3nwp3nYC>w*vu=+NJwDF5~pSQaY^YkMm3kp|F3K<^$Oyj?M6z
zp0|r0Zm&@DSvcKf5@q(E#%Ys(Ion)jgKW-Z@*4W3UcX?tu5ta^?fO~f<^og2Gbh(l
zGIK4S!KTh<qnU%Md_~9my}YJ|pCG`5&X{bcVPCmZ()XXu0FXppE5$0!aLr_A1(9$;
zGUXvV9FFkGOWJR>zqCsf71;L(!`^(sQ0&u?>Wqs}E<Iew(8==ws=U3@;`vy$)juA?
z)!eoj8<T6scA{Q;wmgQUWvb5K)l~em=9-xOMhF3qbo3`{(M?=cGZ3Cy!Iqu2tYB$(
zeQbfZ@DaG_`8<c@;SWmKOCE02u%1d)>nu1l3h4=mE#cNjdE``O(hOTU9p(4k-B7jH
zy*1YP!g{X+=IJdC+CfrSz8lR&kNt;lINsjj++#?Hu4}YEKnr!^VJ5dL&FGSfFMG-a
zo7J492u`x3YEG@FDs%tP+Gd<5U5_>K``L~02f!HIOLiSQ<|x27fx}{Yiu+UZ)GW1{
zB{|gM@r<_b?U;>?^XF3gOG2HWcl)l(?@~M$b&Po*(zI&Akd0gp**{?KFdvb6sKGHy
z|8jKjuuq9vsG4rMKg<AFdKa536xyR{bi_a11yg!?Zs=2nBoFOf{kxY_|B=a2;9FmR
zuqyj8HWVz+-#5m`^dU81EA1lX;e8X$eN>Fg&&z(ZVygoN9VB8hJ~2a8UdZ+p+5Lr9
z$W^qL1(W#_erV>O(B5(FOwSFd5d8+N1-5&hq?$d+t)RogM;&&IFC10Y2CVjAggHGw
zyfG{K!L`j3T%>@k?ZcL=dfin5B!r5%Nj^X|S35#}STLQue|Z>ZZSvI~l?~GKtZ|3w
zC9MX`;3^)D(g(w<=PPQX=(4Fp5QAH!@P)_J@`1gXn9X>*tCj+)cc)>Z#@7A*w@f25
zCadQpHxw8HRo0?q&|7$Ze!@4EZ*@9CqH%eYB@&QCrW#KnXYI{xpMZtYvoIy{#-Hx)
zvI%kB#U;}$;-UXeNX73z+9*=)wHyCo(|A0q+aRfGPkZ}t5(l4$%oDyP%37=2f^m}l
zsY&&i(PX1B1|zp(uZZ{H$(A(7wKpDLJtnp%{FuPt&|R>AdT_luD8hc*&wdYY{jz0f
zDRP@nga4ZgAPVC9(u)8myhV0YPvO%q`h!O3yWOnk+Y4So&sRtwNm%Dg5wWRBa(7Fc
zjs7Y=Ahz%@KW?;=eHqMTx3E7mQ*UR}8%`?MS$~iFg@4L*0&(X+^QZbFCY9E4>#QJz
zJ`pKClua?#*5iJf+Ov#@qKwMs6d_NkIS`r&Xu1H}F~Xm^-j5V;I6o{qI5@GMXucib
zETnEQ`YV}Rvj9)UxVrlNCq)O>LPW1NIN+MSn4@3fdoww@k=pAkO*>KpZs{uU78Upk
z(VJ%5AB|UGb@{*JyZE;i38(jHPy*V7?IOr+colmLV>#}zhr28lUG!?~kA8_6&~3)7
zMgZ)pt5qJgMWhd{?rh!SEpih^WJuTER<%!!`UOYfohhKo{lY}`&aS|{9(0S()F&u(
z4xN52lm2z-&WOy~i$1hAr+((~9Q{D^fzG4NZT$cvSj1RVm7fzVY;tlbFy79!eEQ3-
zP|YYrRYQQoA~e(-N?MS3QGp?Fu{q1>_HN3ZQj~3I9b^rErJLL)r*=<!6=RjwcHT>F
z%hA1`^ZY#Dlq6s^<~8-aY&pZ}wt~I{2ge3TitvCB+8}MlM@}31>$i==i@Ixx(7U&2
z)t=H<r;7B6Uumk~!ZI&MhPX>_u;ZTg&CA1c+n7XKmEaCjTE<~7NPB>Mf8tJYbJ&db
z{IrFu=6dsD$>z53RCfY~V%5cag!JhFOe;IlW}XNIIZ+Undm+C0*DPfI67}Ed`ZWEC
z(zoR*vp;KtxS~R~-iFr34xrC-d}yredPC<8Qj?7hvRyV0M&RLufQyZql<7-cL(EJ_
z7mR;p?(iTw@X>MFB1L1NDEQb-cYCE{riL(MyHRlV@{$uQ^~A)wh}Z}~m|TI<8)tK|
zc&>XihFmPlFhhji-#mRJ{O@a2G4y4-)j`WPL7p>V{U3xSKZ&dum(Nx(v2WR$t3#72
z;+aN6#AdF(BqFGb1!>xjvT~}E`Th;hdo0a6LD=^MR<F}qHG8CZTM`?|;14J7f(g3A
zz$Fat#-Cjx1*xuVbWTyCuF3S4Vu8_*>SSDPa1I-}aDl7K3>SNE)cnTlBYc!$BCV#R
zCff0<w{4BC8N~i=DO&sc3E^d1XV&`<8%8>;i8faVf;r{Il%t(_U!dw{3oxzM_g32{
z$bc=z&C{A5NgiP*5ymgVuC=wUtBcfc2%lvkdVkYyt5CPDt_j-==%ZY@V;d$cLStby
zlIm9H8Q-~2b6z}h=aCZ=f?Q4X8jOGYQ>Een$YNIJr!TlS`*Uw{tR2pe&`p;-KPznp
zd@v%B)k+9g%WhwqL<Kb+lgjG&fz7pfAZlUbwL$8)*f-Y13Z@f6p^ZU5%h=@PzKa|s
zN(NWF9huv)xpM5J8bA#zHf`deSKOhXBZtveHymX)e?v|~25N}r*yR6o8B3F-k`?eP
zQ0P)MeGkRSg$3CzFH_kfr=$3DUJcr;jeK;O5i31>*fM3!S#By(T399!0jMJBkg%cL
zBlN^U#kd&#{T-M;cU~nLwoh(<s3~g92+UKLn}?pxrfb3)sg^L6i_`gX%NwG8AV`1j
z7@G9)mL4o|`iKdkRG4#L`QD64hGi?el}{N<@Xz;G%lpmuWz@sujTJ!?R1B7(8jp%=
z<k<P-;P~`jU9n?lKpoH_RJ1h^^!==S92`$^ak)pID#q2c!U=Emv-dmbmd)J0ILamw
zC6iBx49t{|sED9yD)Keq$R{4+b-W3#6%PQ_=Nyi@1WDM5y}RZI&gJLn-ijkayp6OW
z65Dy*v#4_NL7KLyUV?a|BmXNiDOm${GsJl*D0bKSSIG9B=>Q+;^(Q*WHky=#lIj;S
zXFMXBBcx7Mm;@x!P1gy|UmzQ{wQ(+Yp><agRW!rc@@TyD!63J1cVQi9#`>>LB>%5Y
ztgLMXG-e|PV0VwG#AQ@eaYvrY)QOEku{mH^fsz=X=iMtgTufcoL|QHpNQG`s{(U0*
zA@Bv12==PdgR>a3n!dL$1tgEet7tD`Ej7RPOD7PmaB9UI+<(6&DCR`u+6#C3_E5QG
z2Fh-TeA?H&G{O)!VK5~$*!cXu{~MRXR=Hrc*PHM(+KWUl#KmI*|E}Dt_`oJh-=Erz
z>7Z60_e0ClgJXLO{=jZsYk4|(Eah|cAfPQakahSGf)tOREe|o_YM}Bnis-CxJtWjB
zU^ZLuYv`5e^G%^V`6J5m=-Mp-z3KYT(Q3={p9nY#4{lV9Y_ujad;C1q^(Y1g{60&>
ztO3bw3n7%y6~X8)A}yvVy2q&DZQdWF8F0ol7gxUU#wXP<(b(lwM7i)l`;o`g(-VjK
z^h3*m3h=wzo568zxNL@e>-ic++&);=+!FqX`-eA5yX}=(_)AjP1(SN#H_p<S4$DBX
z1*ZgPnb6#4ePX!+bF0R8ycw^1N6Eg;Jtr$6Ed5Scj=hC-Ds(K&lppTHY9h1RrG?T+
zt|I;2&(1AaueW{R*92t!%nDuq-SH^q)@C%-SoZUEh!g#3MYoxhhhww*yJWrxKT!P|
zQTdQv0nz?qaJAbQvze#2u4672CBy7VR^4%We?V72T>!;}jCA7vP35EvCQMvxh=8)K
zfhhtklX&V_?I#@3J6_i<YU&C8s5?DN0l2!N1fiD7%}|416#)V$8n}rkZnT$Js>S@@
zeIa=*vLFb`_#MR_VI#%P8S>MO&IW76#}q>xQuVZtCs&B3{^ZHkgnWtZAP7-G6I6J$
zio#dgJm89V^g00c=9C(vpV8%fw~YJjAP<S~`KR#xY`rCi_477LC@c?+n`G`hzV+&k
zt*Ikm6+#fb%X`Z6BBVYor|a*F_V>yR86C`DPbZ|90qV}x9<?di$=l;zH{&xLla2WN
zR`N)_aPjLatCAs0mgaco`6K6#p2yMUeDC2@_}nD#yhp5@wf?SXzj<l$xKflI+-{12
zAm?lnM83xnGVAlC`E$328u&Jd7r#FNUjm8um-5bc;`cnbq2N>+UVZ)kjB+`r#Y_Vi
zf1q<~kJxJQL$0aMeoD%Dw<S~S#Dx4IEqphyqR?xT?5%=5_T4g<SxWPILYla*vm8$7
z3G}=hbv33Y6v7u5*6Yra_ONT~KQElBSVoza|KE)Kp2I#{lrp}1WV&0z44`B^;d&pn
zK#~>z0O##a0J*=K{!0UCQi-Y$S95_BMN$9Nu)F3~p&5V=z*6&39>lUYMyqkwFS-{K
z=kB;4UJe+|$lKm{)^BO&(_*c2PZKehs3(%4KbzHL{|kp!V}Ui#r>V5g3*Mw%#!R}q
zU_B?=l<--PTUPW*2F2L-K-CHn)EyryTB;|qwhV}gpAMXVAINktkwnSS35_@0cSuCn
z1MBZ6f>vd$GY(f06+gtD^<P9#^ivfx<c?Y|z<piVtLc#nE*L+BB019N0U@89IZLSP
zc@w8KT~gfd#%@)PZTHQ^&WhC&m-Buy=V6c(XhNSjaoLQOpF@BuyzdR!U7l}*nDaoK
zB93u_CFY1)-UD?m{hEHKz_v<x>EGak1{KM_J*doRvErTC!=B+qsjuPU8hcJHwbdFd
zO|{<Q^5TdI+p@M>r%TTddEzpR<>ry|CDSdrNfrcq^5St$>dKXQ5u6y1lSW6nmMlax
z{mGikG^zj+Z%cTE3|%QwMVp1dgPXWxIXz!^&Bb^TgkfTdH**F6zfu#F*KhbMy4NsQ
zgyjd04{i-HjiN`h9(SlDX3l1X?4nWMEzqPCN8eAgKh6B`!_%t60t>R0*P0C8!UN4*
zgj!L~^hN~yp%YnvscT_+{qL^R%tkfzsTv*;PXGpYD=hcZA$UWezXfUXPxuBxf4tc+
zu}C?fUQ%Y}5oz?<9QFw8HStZ%q|Zme*uzg*?NrWCF58++r|6|Ew7mgucd4_zKStt-
z!k;m_Ry1K)7Yb-bN>ipn#KRw=Y6Snyj-!+P4~_&>hw6aRdWN$3Pc`<id;kVjK`uq<
z5!Q&w_=--q(|oa4hDWmd$j^0WnNYLNJNf5wlY7Wnx!%X!*NY2#ysb8bJG&y%MsLmI
zJd_Vh8`YmzIwBiu;Y3DUCqf*T!bD|_IV0TOSM(0Vp+4o?7W&nrNAso!rG~r)@vZ)1
zATjyCH?X}^7|nHiKFIsK11Rd?nMvm2OY;#?@eHH^|6T5N+$)>$7@0EXBK>sv*kG_@
z+A`IeT5DKuCZ)Fgc!^#M=c=Imt1o!kJ*xB&Q8Krm?hi}`9i85@^>F#>IzvLJQe|Tf
zl_z0cl@&uU>2}K+{zjewRil$NM3ioGh?c$!%hMHOp6|AzE7J-ZNRv$9dtFq7(j?(C
z9WJPRiTGJh#I*n9MweY)60Clqp4<E(d_DijJeVF)|6#hwOC09#`0bMQ!{?>m(?e}r
zw_`eg2YeZwKy{llE1#(@pR+cka_K{`hlh4H+yzZ<aPe9a*g~)5q*4xPhE0?dsV=-D
zn4qlXDg~a<rI3AjU@UI!iL#H99A}A=RcN6y`gh{ZEOcd~s|5+RN0Ms9W<E*?%V9gw
zqD+Qgf*l4wiL~|jUBYP0=CmBTYBF*?68=<?)vak*TsUXS;EG-@TnVrz2_`+LcVpQY
z&R!4Xpq$Rq73Fk;#6dO=R&f#8?)g1OZdMe)>Lp7u^lPn*obe!enQ2`5O_~v?U*jO=
zH3-@{eCEQE-X7Vj0a4_uLAA^7)<ME(IT<l4M$^-M_C{-l*c@;eXY_oD2q~SZK#L(b
z_si9FYF}7^GPm~eTW^d+JhigGReRFRDaB_Gq}|crd#@+dW?s(pwL{lZD#AOruK=p^
zNsZFn-;MoMm!R@b^JQpIa|;UrWyicf!~85C*Z)lqnE3v8`8A;N>rsuCqFmxU9y!gu
z?n66L5x*_et(5C0*X#|%BwCN%=C0;jwNO@c+?WKPRS>noaGUL{swUfro>Xlrwn5uf
z$e&&y)b3dMH3j?CPiI!y0-s9Tf4Uuun6H1;>99;$xa7%m>KH$ovSJCpwm9mq<kQkQ
zZVNlO@nGqNd;WC0vIzk+lj@jGdPjvq_cis~_8ZWXQK8?>U7|oJKM2+@@(+|_v6tP-
z4kBoMEmE>y9)A8}H0s3KP?fND5ZzJa77Z#@T22k>zp(VTdB56R196w(GR*jfuARTT
zQlaT^P4+dOgESo9tqKTB5>YS&`8!xEt=l|@XLA4D!*KGdisM<f{M!5lt9nDrsnFP7
z8+MRDqF(*4_~yUec*PR`(X!%qY^d4<ch{nlNl$MXxKz*q9j@D1Q$^*N)vd`;?_X0d
zC1@lr)v5>RIC3oz+n&R$ET~hfQl`u(LN`#3O0!#wdSh+nL>Y@=c;mn5-_^DBNAtFF
zr?XdcIoh3cJnNLx9rsR>kbl@c(9Z2zPFfg0hnyz@1d({$DYl3gpFO_9`F?Jv5kE+a
zcPD)fb&Scg?})cucUKMmBqa`lirK8+-#7fdgrtF;?%u<|+3QCmJ845%x$i&3v%5fC
z-y%J@d)Y$oG5=j|lOic)fl0NGWvolgbx1Wk487w+lrv|=6i6?@wBnbJBf7jx%w9*F
z#^jDhJ~WV5{`-}v3>LS5hY3e1@e{tFA^*QE<r>rf3nE=<+M|o+P1?I<yXi>)B}a^M
zfG3`$D%b6r8#Sj<nBPjI>b}4}eXgIKPt;bKKec*%Zr6l^-Z$<wabY-g26O#Rmdl=0
zPTNuyb^tblVw|GRzJ=>6?m}(2wmZjCtLL>Oxuk3j`&zbz^Qw)96wGql%Iw3FZ*zTZ
zrvE?|zt}%g@=hwas4*L`%WgAxG}sh@Iyj&!g5_-{6f`WDtRGb&=i#-)JJ_r8^b%*?
z4U8u5rR-p7iT3`v#eQ2m<w}s+IHq37A6`)YMv>Ct%JCT5lJO&jWPdhE{ktk<Kai%-
z4lZo&pGy6dvu3~kh>=+~x^l4oBcwz<m5?~Ky>34nI=w`yQPc`B6yxA99s|pHE7+=8
z-g=s;Q(kue_r8g&vj6`d4XkOlBQ9#Y>m8$_O*?12r|9D}y|fCpmxf+B*f%<54JfE9
zbg_b_U?3Y+QPi|+POacsea*>uHc2tVt{TVgWR-i+OukQ&Mx;p-zL|~`n-ha&w6e46
zuLc3iT<XWUBFdkq!OQ2A&=(a#$Y)H*riOeZ6yXlHj_<X*PE-t4k%+F2YHxFjf>uMw
zudU@ex=8CqP)RTrly;5MNj%Y=MUm<`q<OxJeC^RlZ;Y;&KVh#k>J(l$CkpCS1ElA|
z*Y}R_nTs@<yH3|%Em8~3Z$H0KRyma@;Gc{k6g^h0K&VKGv9<GxUg5bZ%}|70Bh)XN
zif9#|u5NyIGCBkE@EZq)jNF#3hKKp_t!nCoAOsq|%x1OJ@cHS7AaKxS!5dF<Ru%Bc
z-DcuzB(!ua$f1W;W8H(a(FeV|MaQkXEUSia$BMo%G+Pi~p5P=or^_C%ive7Jp(}b#
z<OmhL+bX%0xof-?<WBqO9L!9%k<vKnlZYyU*A!q5o(-uOLaIkMhnkQ8ueE?+{n4AV
z=nB=R0cVm!uHuI_W#q%uSY>ipCIe-82g(RfJ`2*j?IMR}_fLY)_Hb(RN$PYifw!wG
zM?$I|2r(a93ev3U23mM?Isq{Sb?@VwEq3L`3u572*FL%hhmi$njmka-viN8(_MnS)
zv4aOQn)bBRuX*$C+Kc}t`D_LMYY-46eZX?Io^v{qc<?zaB}Tn-ov_YgQPrUzFFLj~
zsf?GAzr<sYi%~emTYNwOHOucMiyJAkq9BW@y>lNaw6r(_?kWDyOduPkFQo*_OL$H>
z8)PcJc4uT=+m0%LPb+PZJn&BY6<?m-oJz|UYFs3ady-iCaG~uJNRC0QnG&ugRKvB@
zHZ)sKIRJ3}HRXjW3Un%VJec*%ySa>vpctfq<x@{Ud*iv#`a~Gp4CD4aIqY_tS<h=x
zS`AphCiOh69sA4A{0Gb*-<j9ZDTDhyjS%F*wn-g{M_nJ=L3E^8w>nJ^G;77`ws@0+
zvg0MH>><o%V2pOOE@!O_Qzv{M-aFvA{iT^xNp@Q7?jF^Jc6n>Y+Ek;c`mH#ROY$nP
zdI*b8d$47f@q3EMZ7$x^&2i0deCf3~4j|gVmhGj>t?~X}>b9N4w|U$%3$zL96W@gN
zS69pHy0ADTjRxEV*0i(rjtpLo8R%M`U7RRT<||fu78+Jfe?Ji<*zqb~or#k%<<a4^
z^0M5o#EtH4S<+5_l`(wrzUqqZ(P&KLOWpn-zj!r%xb1#aMM%uOmi6moyA_~O0dM(3
z|Cw=<;f4CO3RiO@gjkadRcHA+dFJ`v{-UntXt*e52IF^UNjFbVvQv!{#MWXdk=n~`
z=Q*F8+V4FtJ(FU=QX?K)YSaVn5%M#abl2(!FRvToqvy*+(q;QgwKzxaSnb@w>_3HQ
zT`Yd4L>UEJoR^h8^{->PcVdC{-(6{2@uIw2ON7sw`d_S%0|U|P9muw{$8~;UxlEQ>
z+vgxOiNj_L6BF-MTM`F9nIhsYKWd8kKU3ksx@dW2yfs;cdPWXZ_gI-+Ke+b-ja5z{
zj#-FBQhjl)vE`d-`TWNqcxd1pJ{4NKe=G8?k)hN+uTjiFcmka9f{`5RD(IRlUqSHq
zHV(y?MXfsx^7YmmMipZ=zBdt8BsMnl3U3ABg9`p9JAr1LKx;Q^3e&(La?6jrCfc5M
z*XgkCt>#7iMQbb8TpZO;U&Gq(K|<xuMiy@UoTRVLTXmxI0cKZp7IaKO{ou8S!_mB~
zf+Ech5@zC)1ET6{?pq5nCqX~jkQAJCjUPM~V??SU*t(gd5Ygq(kM$8|;+2c4cZO@p
z&iwaX{jf`xa7C%#!gvPn>Lknv2)0D0dtb*m!O)RSShli6V=wU_^ltiJSxDg`{^=_5
z56z}1yj>pdyLD@ny6jy9_BLJ}SHZy{bMb0+1_3mIIBc>n0SNzJdHk9CEk|$R9(5IS
z2S(2I<bffF&C;@*a<zc?opvbD{Z6ETzHUnDPMVM(!vFq*;g|BW%e0O3A|4|Bd}_n_
zX@wUl)`>P;^5LQk2aA|V5qswHQf)Kv*<UcjFeF!Vq!Z&v0c{d!IuQ&L#OFx`>dx0i
zWzCrn?sQ>6KKHX%@xD$1w4EqfUuOPk__c(JoOf8-gVWvpgej~n@Mp~YdUfURrTEmJ
zl>p_HKg3SEw2Tos6T14k1v}~0zv5$HB8%y(wyuccOO@LDi(2@z2kc29U8^tYOJ^I?
zrs0DT{*TZ;obK1+0yhC~W|DxAkaW29XKS}2r4yCDt{R}-fVow$^_`q&NO1q#vcNnw
zsM6KLF9~6e-(3EM(Zqjk8Te5Wj#@4T!&U{CfLne^q6KGJ)B2O)j!5ipv?@(&k6)=%
z$iI*?hW2Xn22B-j!DM~iC33qD95(^2EuFJ@H|v1rOq~s99U#UazC74WK{v3n)8+bb
zr6D1G^aO3J?VL-U5awR$9Ckl*2LL(~xUqXpT9*eL?)8K&^Rh|m9Uy8l$HSScU~wJ-
zKBP9%z^OF#5z-0U*j=N7eL*eeAp^!=k)^R==iOyTeyfUcTjS?VnMk+OhgCl+_V*q8
z!c1G^lmecBXo&Kzu^DT{e)6T`vev|a#nZh02l?jCw>*|B(!hWSx;v&tzHR=La`$gK
z0u{_HE+tsht~M;OUy8-tqR+GyXkKD|1#r6B&}o@XlBy>k!qYI7tJum5{S~E;*k27=
z=t@@HNsSGo&v8V1eh%wUODp-OT27f&5;TZpM*=kJ)uncu;(B^|-(+N-&2x5eQrK;?
z$^>2nsScJdkQ~CrbNAoXVA_L9{J?Tfi!;TFIfNu#(x|6c9y9P2Gpm{#FLx+^MHe`h
z`Cu3x=0}6RHLL?RgJyW2Wx!4de_J(Osr^+~FVmZ_q408Fhu{hst(ja9k%=@o!~?>J
z0{EJ^*$W?66`F-`y~A*RxU~-#fXP0n)C=DXZ^`W&8G*(bHAL$!a>FF&(pN3JbCwIw
z9gFZ!UkLdSM?g~pJQVo2te%ctI79bT&B^QXn0h#3GZiVTD?U@O4d6COZgp(2ONyYI
zg-sAJ$6VZe05$W>u;?+u>L!sJpvCtC8fzEhT4yhdGPtvw%XZjz)*`jWcX7F}yOab!
z?zdkcN`qxA>MJ`)%p1B4uszM(gxa2#3}Rml2G&ylFt}LU_J&xq^E=thl=>Sz@Q01r
z=?M9ClD*vNQ<<=+6&|k#_SbMIBs@HgVE5$ZT#-Abdeb6Hq;jT#*X=)V9dWzaIw`fW
zh94kN(|L|y`E<zL|NT^Z<2qr^M_Cp8atr(MJ!hW#S;MVEH!TiKL6S)qYHPYB4QWl?
z;ch~GdY_3oteFr8S=iYF6=K?*t+#x6QSZ|&Ms3oSp%<>lo5dEq`~~~3f7Ncj0~KoW
z0d`n#HAto~a9mx9jVZQ$>Ph{_1(w}?CC09OZryfF0GF&U;zP7EsY1aNCi0qvWR_jF
z`61!gi5@{X8a#5&SW!6<ucBePFzNo4yspv5BmRSDS1RyhMA<S6tKOX8UF%rCfXIyw
zFPFf3bu4XS@MkZ;aUh?1zzlVz6dGo{;6J`zptsjQF!1#|e0(V(e~%28Y_pR$*1Ebx
zk&%%xadBEVn$yLayK_~LLZN@J4RSj-8x7SRme!f6K!53l4AFiqQeNLURLe74oSUiV
z7`={E-@WfC+M$B<Nc6XUH*M)0#wVM;Q-a_ii<Bx*St1m!W;ewBp)M%(z>N()c>?f3
z_cio@5oK^ch*P?zdQn7CjTJBRlW2x0Wv!Uay{Da;s%?HJtQ87~ov>!C71|%N<LohT
zDv8)SWp)x~u;2e>u$lS@JA~sWXPrR;c^5tUVweT#*}lG+P-o(Hv9&hUaB3j3i4wr0
z4Ti+`!=2G=YSnW8%l&Eln-f5{ufN~c-kzA6I<%+f+ug;kV2M^EJPL})q|n`8HqwgO
zP_2>@&y3n578hUVr-xgtsCpW;%1@M(!A667v-RRHMRtv`(us4y&+N#3{@sKc@=%jF
zM!K@faJ&?0^q8OGG}IxSMGS}JrDmmAPTNpt>t0DaE@X&iF4x?f4Upa*3dxjTN?|4u
zoUA*a^`2e~=xz)Ljn&1_X;a;{0Uw75^TJ}MYu=6&FepwPH-N+2HxjnC4B)7?&R`Nt
z{6c(WB)^CVtel+OKr$Puce>YIP;a!kwuQ>iePKT`7vS-78nGsxTJ>|U=1ZAdn3*sG
zoesHruXm)d$QFRJ%Tg{vln(;R7%@04!f;;IBa2eE!~EydH{3bXrD5xnkIEAPP3^A8
zzhZctC5L#GtI<qIzy=N8^2h$&#cpztb7kctky)#N3M&F(hN8$k`}b!T(DVGO_K-eE
z9W^>VN=j7=!HM<?yiOwxD!6LMy9RR2+vl{QgO=Xiz>Y?QIax{W?MD`7@Q2t*=TdDC
z^#<2SY>JSxt>M`nlGFse%c1~{Z&{)d1bAFd2WEj|8~*<O-z%45YoEY}a(#2Uj%DS1
zFjJP9_pc*VyERkCE*q!)>h#M<FU;G^6G6&{M1&te6FFU=V9Q^NFLhk4gkASOMK~x}
zuYr|4z!6<UYjs<V0*|8F>a!nqcwEMm*$Mn<8E+}9Vw7f2TtpmI$o*rs9_O_pte86*
zd#wSSQ!&bg=Z3PIBzEZ3R5?KxV4OcvabJHa<XQHkZ|$r<l)X2S?>f`#R`fI3m2t~@
zw#|iIf#VWlqa|olN>#3G%4cp3l6SbO25Rr7OPVjMy1)iMQ+))*Hf=}K6MCw9lUty~
z{(B<-(k|n4PbZ=}$oQ_ycB=D7E+^`zrzevtHPD#^n%@kJjL@*Ky8B|O$|@@A`{WR9
z4Z$N}qo@DXA!H~A7bc?6Z}Rfgf9`AVNLl{W#z0QCj}bcTM-ab+q~che`T%K1?<`QI
zamblldgW&lQPu5N>pyHh)72<Hp*rA=@Bbl?0b6$9$5rgWr08CiyY3*(P^69j2!A+^
zg4>;YE<$a3Y>PSkpeWYIWwNqmG%}9->abNVn`B2OMnI7lWBz=G-WlN(a>Z~uo`|L9
z_C1T9CPJI=_PeUTWR5R%8ZgrRx(&ZY2`)$E8%5yaZ364YyT@_N@Z!Z`e;EQ|a>l~0
zy5ws!zo8}A$ER21tlsuK+E>Rzf#T?^?Q?7<CHuLBlPl5(_qt3Ly*}ziH!G{4>Hz)?
z@&|B6ej#BAcz#5zxjtS7aYm+iPFrs;0vQ=uTU*;lHmikL6uF-Oq!>EMPV*p~>!{YI
zzILYX56XY%)zU&HLiNZ#ZP2bD8uNQchwl3bOqxkC3F@OoEQv-&YSfN9D_+yyZ^ef+
zTEA|T)>S=bPUOWS92cbs(IUC=+bdoj=5|!$eYPO%FRsWAb=lVKFvd9`x%J??<N-sl
z4Gz=Er&|EXOinJ>Nvu#7yFfYf<)B#&fj8L~n#&|<YFbAQ!J42HA70<*T9S>b2Ze74
zyi3y;qzlC@=PAnj{Ug{2w!nbYRU>QAJgMq3%q0%KiULeTU>Pn@TzQI8YmU}sZolqZ
zpT8aAy-Z%m`7j4zv4T#sDTUVy*<%wE=db|hFQ$6KS=)(mS-*nqFQ1P{uu)M_P&-pa
z$x6!~_-t07SXo(TW@dgR{dutDb~^Yd!*lr|Gc%K&lXDEBLdSczlZ|LDp_&CcbP0in
z$kX-i_3iDz#zr2?xk}xxQ0#;Gn&CB9FvF@gxG+SM$&fNLe^6Ife|UVvX3%>B{*zN6
zAmN#ogX6aCWeo$1S&Ca@T8>0KqwAT0TBV6kQ4vjXak0bhSb*txZkb+ZFlQ>;XekbN
zE0=y(C}Is*o90Kx|6@1y4Q&tb2Y9fg6=QKLPoHkA*=wieZ*ffEtg8cbmt?c3Cng;A
zfJ&*1sPg0N8ohmkyvoFxSZ&}M0W-HHNd93C?nU)PCe-2CATzGm2pWzNJ}~lcJHT9#
z@2^8C3v{062V_9xqI$7I_VJR1?vG*q#QA>qIoO0yM0*5zONN>nbnV`Xw(f}yyhvkY
zClg=xCzhHnDOhe^aZ}t=^&j#ynkw3)f_5COi*2SHE8vUFaz{-8?2C!Iy%J|>0e<}Z
zk19zkgF^I)yG)jEu8PBv34enz{ApsZc}%&wdg`))EqQR~((S^Hhcr5fuK6LV+H(G^
z6^`&%OpFldg>}C=SnVZC?^s?=Cl(E#;M@kY!s*r3R+{_K>h&@}MojD%9Q+my4ejP)
zTxNGP8!lTkA{S_Qu(M2fD)+dxYiMXlpx1rX{BVXSCMKq+q!gK!Hk8C-?(%TjjevyI
zK;m<Et#3BKK0ZE<fP%7hHE+EgukE#Zu{#caubiS{N2?#)?O8t^Hjf)~dmw7bWlB)j
z=nBG8M#iT&8g<x~O4IS4p&_ds&6koXpkWUhIjGnGt4kj-XmR0^FO`%DldnF<G(j9y
zO&|;}<YMpAJyv{y4P7@PKunA2#2n`#0rc$C=r4Ie#m<dIkJf@NGAaz=O`Cec(4uyM
zR!)`&G9)#wNF3*+M?Pza?M>4$Tcj5W_8bS5n0iq_cZtAm9@ktv=x$>j(3D@d0WAp&
zz+K;B9~|N{mz8VOG!iv8VsoDU_Fl^l+p&t;w}r2q;d}h`?-xHo(9*bIIh5iozw3F&
zqkSriS+QDOxELwzLIzg`6x*9sB@Lx=y?%#*A#7)7=QnT<J_ma64wxO&QBr0}#MA!H
z%TwXnO15s&oyeDqiH)7CEy18trZ?yT-bs><8y=_g0lZ*H7dWQfm9t7$%npoX3ikH&
zn8e%&wdXtgvVqWb?%C~Pr**%iS(=|8y0x{{1AHuaJZ>?Sm6bR(kli3+@KE7xi~6J`
ziM<BBY$n?(-}Oh=Z<{L7Lpbu4ggonh^hG5>z81%T8isyPvsK_*;)Vn$AU&l3^x@ZO
zh<SOYhJq~T<~jL?48)rX(^_h5ZRl4i17?F|%-j9H<#O%OgU~aqz?UC={poOE3yF(Y
zy-4GAMUr~j!ef1)K#Qg*=;wZ6qwJOEH@p|gw#j3|*R%R3hEZeDHJo{UD|Oi??Wzt}
ztP32+a@+#RQ?Fc9`~mUKgu3hZ3p99W61rVP-=lLXi)!(hN_k{VbI=risw5CPdvU)G
zJK9s3M<nW~Is%^WmjmWkS68{u2belKI=;ug^V~1088Kn@0D;bj4<B?G6Y~oTNjN$0
z1aRRSwT2(T(4Re2P*89^X@!HfvA5T6^@aJJlcSlBLQ6+y3_L$QBH=K@du!G^mARa*
zNo;X=-Z3*74>g>e*V?ZC3=7L?Xuyl^^>5Yr#KsmcmCVXuI`*5B&{(h|5d>ZyVAC9*
z%L)8-No#9*;G+!S7)s%22Y;uGkdRQS{s-jjr4n`l0Rb}_Xt@0R{FT#g!WtK?in=5$
zsee2Wn9jdZ8r5NK|ITMPT6_mz8R$izXf?RGaYZcdsEa;gd<U=bnp^dB;nhVyqmB}$
zgRK%}IFF;a05#<86;?~69=#t(;L*4Ef`RT42>ug{sB-E)58INADtrlV8-8XUuA%!|
z_6jba6KH%|Y&%NIG%)-P8LDIh?+&}Ak=KETm1KkP!r-nNsX>I@mnJX4M(3Y+pskc0
zKh9V{g)Qg0BoF1nsHf0>=M}r!D{uEwHgrB$r1U8%VrYP6=fE<te8p6`@qri)^Vs`Z
zhQc%3y-pBObn|txp5Iw(=aENG{+k%#$*#uxPt`+Jm5X4I>PE_Ci!hi^QS&~W0D@pT
zv)>*vJX)xOy1BizJs)JV2mdg+oalgMf+p>{#pgBF<waFlS*H1P@xgL)y4B*}_aNq(
zt1<`E6cPE(T62~C)&O+N;$o^op7iRL)9Gi3larH>h)B6^JFH0P$JW1&`!2gV@d@m<
zYr^Yku3L$SHOs)Wr_pelCouX54i%uHqQWBMI1nMIV<EraUY-}AR<91lq*2pg6#e!M
zo`{HO!G4&>F<g#06xMq5)5}F8qK&IPEVaw7sdq%bJY{FoKIB;~{*6NHesa^rw&OMY
z$pI&Ita!Db1Y9__tNZs83<}oz&TiW;AY!|$s)V&*gZAN(;jGjO$XePa%x<qb=<YMm
z1h+J=P4ypdr*kfk#wX%xk=g0-RZ=1!4EWJS#a!VWrzv78aH;uLMBuikr}tz-wCXZn
zZn#BdsN1?0s5fFKEl<FnA@>D-BuxJ?Ez=^Jr>hEUptwf(zE4w|r}y_TLG%XLCu1%}
z<-PHKyuW6$m;pMAtjEU3RcmdC6mlhrK7I0jdc2QLNcg_*0R}Npg;>4oz}fV4w~xqI
zqF!^9b${u`gNQ}@W+l~msgA_-&KXR7SlS-%y+KGG5|tAh8%wQL3GWSNLnia7Lbdhc
zkRQQJ<mB*SA{4SS_0HCQUF0~@jZ`k#Sy?!YAk+f)AOGA}Zk@%Gh3sjUxoRN>HgB2h
zr8H<_pN?rRJPf3!fTml}%NbmOIBLQI_I_Thve54blO>xMLlOr2II5<<GVNp96dW@A
zJ-Mf+!diM;3YYp<dfHcl$yua{HCBB~atpMu8Ia{FL@E935(xoM&qq^jWt#W5DX)kM
zpek$dQ2~=z%2fJ&Ue;PKZ#4f4TPa3yfw>=PaOTo@EPAl{3XTMXeI_PWt#_gU-#q&g
zgr4_TWuEs=tX2yyL_~GqS8CN3y?c|8y3IKCKdI(>69rXA`wa;R35b>`U&ScXj}`8B
zvLntq&}B3xYOWp*&zEJ{aKf;Tfd1az)=@c-Ud6>ZtzK6c4~tB;JRadvN`pA5>G{e1
z=<hx)sZ82{`*CA8&-KDTc8UtVE@`}ie;xBu;rNa9_0b~k%Td0ZWb~}0suu#g`Tmmo
zb-IQlRLmxZI-v^j=-eJivwcpU;;YF_{vgG|UTi8&kvT&zm;qj)eP^+l@U`D(la*j<
z%?f%bKm{;!>ibhCcTp<%1}tI|4%xen$=K|8MEvfHtcEPS3Ncr~gB6`kI5zMa>~-|^
z20_1urPpbJdb%tDz$tPuNy!aARNj?w89o@Djh;vYW8?KBuUqSZH*j#%wWw8UQZ6p+
zK|w(&+^%O?!2MH`SBzFkO}^)@tCA8xGnbf{=(_5Z%xX!U)-I2XLDsU-8@;<w$7J37
z$O=Z@OJ;BLOiN1}NaINkByfY**4Ew!GsI`j=YV%oalHm447^E&mE#{UHdZzc!c{7C
z)T)nthH{~o{6owGXC<eCnC)bY<&zZrK}un)4di?tncAIuz_Bp2<uKNlkpqLIso^a+
zP}0zrw2HqqodKD|y@Zw5AC`?tlW^syxH^)qg>eJ@ua(jWnYeQN;wZLMPT>3vx~DLP
zm5-MHrIp0@bf=8s_Bw@vQhTS2hDu4*pS?+eHA1R*{X;oN2wy6CCA&sHfbA>L8QBHm
zZxj%Rym2t|8CvQJ&L~bRe?S8ZuWX8vcdt&zXg>_7AG!EXSjy2C6=8;jVd@8!UA0zm
z)OaVR?r#!iBSvTzyl*I6Pzi#@73FfJ1;KD_CODDv5{_i)9s*y#LV?M#)r7g+Wb|{c
zOuE(~Yutc42T<Zr)wMjGVFv8%X`0ls51Y?d(=2=4(`b8MVo*>}baZ!D7!Bey-EQG-
z4J5XwaM<4*)gRYhFS+A7G`BG#r>%J2ZpPDT2$eM5nPpoN``3DH8swI4B(u%b=`w)6
z()b1I?G+(#c&P5Y_1XBiDLbon4lW#shlyL_|MN5B)8~vCT*|BbAb`|Nk2$CyDM7Cn
zR{_}N$-OZPJq-g@*RZY`_v5V3pMu|xxf|~vN61ZlQ+IklanBKG4|_F~YTQ@6&*YqT
zPq5@wg!4&Y)rx|SdUM5EouhFJ>UR350aH?P^TrV&YhJ?7WQw8``tyYLTI+X|fkmOU
z)^ih<IW8^d6Fbs_QNYT906d2cM<+c_cjITNh3e=Stk}iVzX6RHMX*96=bM40*Ju$+
zN+V{oOzy;iTh=*+D?^EisDtyq3zE_m()>uzM;wJI7Uq*hjd|!f+aW|%NZd4g1k021
zUz2_pR*T!coxSeoOVI=pvIcy)w3->#s_#2m!-{FGavV7IfSS~`5o>iZ&PZ8?;4@{~
z!9`Q9KXCY6S6zy_%J%7u*jEJ78ydadco^)HdZ~XVY{Cz$A0EvpZg6b;C>0P4)ec?@
zsY3C;d&6J5nrNI>knr7Nxye(B)CE+rg45HrV;@1SfL5ax%BK6bo|P4wHOcW(V{#gg
z`xM2Lx@dhD(AkOoDCVl)d24M~mtr)SG+L<21dB>AfXesK5{k{_0@h-%U>fhtRv1@x
zE9Vo6$eegR-EL#FH<>3F7WH%bep&z=IPv!}J^R-<j<e<Ch?ycmyBx0Qu`IQPhHTtM
zWIT?b*@5(xg&H`YD1m*~fy;e4r_!^z=Hq`abpXUg7XsJL;^)O!H`cny{H!YmD26!S
zXhhvVr-!%?+u9y;_}*0+&cqK-m>R4v$|h+K{CyrE^XfL;MP<!>w+c`2R<?&o6k$80
zN215*S_ZIn7)ANvb&BL8r|V`Znm1am--UznV2#KbhAC+{vcxY=GMgmegM%hP9zwga
zaRxLBm2i9=78Bg%)jfHw?aahP6bmEc;dq`@!4%+hwPFRm#Yl1LT>MT-4*}~sb-kWe
zVVB8~eNhWfQGeNM(N~Ns{Y{{quD|G_A9oC*H*me8#==*pN0`i~8C?Eb1BV080+FUD
zhw)V>POWnAoe*@Hx(r(VCCO@plm1Ddj2>X7`_;FC03&z1Y%i=%gNO7n5D|`P7mr2m
z+q7FX+7L#RBSlz|cJa!|@jN!ex@%IkwH)}1HPyusmfGv+zcs&asj2e{PjZGl;ISM}
z1uH<VBo5s^{z_?V!f!5K6UqAo4?ZDq*(_C@g!oZ|D$t$|Jc$_?7{r!{ii)aG12vZO
zSJl%Ejg3@sphzgc-BA!fJ=iM=QghKXk2A6C@C_E_6cz>s2Hn0GiiX>*WT-;T`ncX`
zvg!V6|35*YKR>~T`6@$J^H9@uue<nuB~zL(?a<62xu6FOrV=A6Z3=fBB?w=1(cSD1
zV{0_oB-J#l2L|N$pijYvbv@Mq$um!bl)LfdST22{;{H#`YH;=Cy8LM1F-5r&1(YF@
zvi*Hu;M%MDa5``;f0iHgs_XviZ>Aoq+gI`Ou!zwVvyq+UK@~gKQ~D0B#m}jSl$o>&
zWe>5lBnKRb%61<&U$-yj1Z6)JeM6>F7%dbx+`cv!i;yzv2;F92DEfT}yk2QLVxo56
zZ63=vqeI-qb99a(6}hj0<ea#p{cYs1cs9Dc)V~&}dW1s_U0G=uEM-OhQ76&+&C`%d
zR}|BZiG#*XiFe-~CN+F8oo@TY*K&*($?}zcLrYzhJd!w##gM;nZ;Dz~dgzTo+QT6-
zHNM3~L>WO&H|gni1)qEjkVv;!>QYVN;8fVK(V2smkS^4^iia8g7^kbvEx2<+J9)Ki
zrt_vd{@}-C0ewr=nZ0)OgxyTMUt&&HABo4;Vb%n-gv_N_WEppQ=8GTy=Oyq{#}`K_
zaFkE^3h_BKGoc2903J1bnD@R=kK^;_tUjmar#mVT@D%`+S4C<?TesWkB{z-0^L%rN
z`uFejW)tLbblM=3vH!?qgv8@^F~X%BAkf+qNvv9?2h$ynzYa2g-8O#&w`&Cq{Z7}l
z4urRF-~J2@Eo6lDG-u?Yp(z?-Ky;iYM{8gcIZjGs85;fUYtzhW%2$#pm|uDciN#x}
z#WokU&H_k_I%jM4fY8AYtd-aon1gc>2J-{DLWHwJWa+|ZGct78iuH;G@>$Aa&K?+V
zVeD@G5_il@jGs8Ezc^!%o}_#4iPk3eWt1Q~IC;~d_3$PQjUhZ6K9uh!sI!D~4)(1v
z#x<}jh#xuG`PaPuy<x7zO{3!m<zVY?O;>)|j{(foagE(&;t<S^?S6kgI+nMao=4+F
zd2?(g4n335#>6`^j{QUzo&bdFuA6nYT&L&HhcC6;MKh|h>c+(^9PT2GKeT2gCFM%d
ztUkt-vY<CLcg$Xj+MHPmluQRI)dhnL_K8Q$9?WCk-H^N7RA#H?Idz6G$fw72J(_VO
z071M5OGwN@6-%rhat-8fSzS;J)X~-(!p=y$t+!NKj{#-p%O=bS&ufLR;u`3|4{~K6
zF#Y(u7BaT*mKVYQa@WLI+h}nfvDDGIqeNm7J%RZc&0j_cN~{}QM+%4g&-OQSeENIF
zd`E}Gk>)-$D$FZo2RF*#M|-x`VU-m>AntD3P%H86bf3u-2mG>Hs5Isq{&)WFs(I3>
zA35wuQaK$zfBFOk=86w*VHgvUQl&-Z<ldK+m3dr@iavnk8e^a5tcQ5=3KtWz%NGXm
zBdf*7si`RwW*o4i>2ft|%mFsvpupzD&E4Hn+0nE%FnG4v4-J+RvVRL_0aD2f)e0l>
z4`a`Q%s9va0RbAdHa;Mw)-f~dn<!AQ-ro+PYtrcq#(<JcVUKKdzdl=XKYsu2op)j)
zHfL%$xPSJAaD(8mONQ^!zsm6tjuo4D$k+Fk+x3x@kkA`KuN$INF6XY<S!~dE>gvj2
zYirwZvldc=ud;e%ega@)+u4J!E>y34r!5P!%I{zbAi-^0b}fvh`HWYqBri0^7J-ZN
z8tc$;jkWby@wYq58!zz+WS7qr?7lw_9m{7u`NdTM9<yB^c;EWVa*gRngFc78J7h)F
zWIk=_t$+`^3}>k7%IOP@O23&J=3g0{OwS{dOAbS-T`%SYQ-Irj`r3r{npV)-U_H6;
zwh0B6OCg+k!M4J`<F&!+jP>Xr7J2!{#TSEw_2jV9Uj*ob#44YIj*Goy8A6I!-tP%W
z^~Z%2nqTlw0xQEU(c#=0xNGOOOt+(Ft}eytnS6%*z|}`feZ5Fc7T|U;4c-s^P1;Wd
zo}rf(k6g_Pz^^#OM$O-EJREx$)sZb<mE2~Io2XuPY4nANL#(Kx8CDiYdj>a7_LT}0
z1ub{IT~$_WXv2An!DOplHCW%s{LO7l2SB0z7By8;^vFT~OlfDYaM#S#mmkge7EC$<
z`VJdC@0UEzrQEI#g7Wgnz@@ezEBX}OOV7a2Iy3}?hvRcC#VvS(p@M)?{P*`H2>0bE
zV?pQvi;IhEz=)ljo9qAMhuy&p!`DA7T)@+V`1JMVr6H)|U^G9Uw4=U(fzj2~^^qin
z?EF8Pz5=SMuIrjcKtV;iq)WO(knWU{5b5sDqohhpw@SyQI|Y>P?(Xi6fAc=yKL$F6
zc;Vi2&f0seHRoJ&4-5@OfnUe&bM36Es=B|~0ie)>rkkSWi+>?_*u=!4wY47q?x$vE
zMjE}`d~f!6e6XO@p=Bgq5TjzjnE%GcDwi+8{`VZj@={wcd`V>`jIES;6wCi*0~#6{
zh|Hy`bRDy;kdTo2l`|q?5J~{yKB$^HyMGHzkOOYg6$^}JCWbRE_1f=3^`hfEVRi;P
zva2C-aFb^v`nj^Pk`XF3UaogFm@c*GlAiBSQK35x?${PDra!X%M<!mKwJ9pXBHQFU
zh736x`_V54?W%JdsKi|O1zMkAsb$H!0*1QVlrSqPRk$$L*Msrh1KMxFSF93Af`*SN
zYq}BZE5{CIRQeCZaMR(^5gyI)9s-p5FNz)8UCkOZqp&Qz2=BYX(;hRtPb%!Wmb~Nt
zAi@NP7r4|$mIm!BPBdd${028X1)1ME(oVj-d)xCyt2^sOvm*iOy6tbO9OdS-otD^)
zqrV5Caox%C7DW$ygsUixdURd%tBVg+;0?+ZID)*dUzxiciU__@is~j;%X#-KUH>yq
zUV@I!;U!D7Wy<TJ*8cAu8x4MmS$uD5Q6PO<X_NQ)8RIfG!r`bSPj8k9W_7OAn!_LK
z7QxS;W9gr*bt<so0E~d|<x((%esyK#8le0xFV*hs_CSx>2T*W(XXiqvLO77n47z?$
zpPj4K|8h9r9P;OLTK{CCey2-oYikSI&9Zh^(L2^$aWQ*)dkK%gZZ(Y_D3iaq;rKiG
z(i1;>ELkC!ZxsmzyjKA$69&$y*Y+m|q9aui_Ghcw%T=jV#^$t?QuMRW)&jFqKDa$8
zTsy2Wo8xz2_3scIWRhRMEPH$EwZpc8l9#h0=PLKDDvI=zeh1{(;oc%lkn+(d^4>0@
zb@sYHs`f}ABa6xe74=EM8f4xF?JOgSa2vO0+h0k0S4NaAD4q9Czh)9#Qok7*djOtS
zRi-}(op4NK2-Y>!BhoNQN7Fh?yR~lCACy^SgP0Nzzt+d}UQ|T>h$ke`W&UzkTf+Pd
zeQdI3n$2>&$30;7>gyuRvDj4B<n`<;RoFt^Vby05TRNUh?8fuwsZcu1m($Wm;JQQ6
z!l1BPtuBkX{pQcUbvV<!#A+sa=c@VzHurBv_QcS$gZpsPdG*Nz^3)8)@WrCY{wQtQ
zv8%awL?}z-8;SrHEou4*dJ&W5#nl*9XK>Dgd$b9*nM%!D49ty|5*pP#GhxQ9ph!$<
zrf3>9op_YIwFHJ<eCT|QBjD+lR<#lNV~ui=@cTDKf7;ugp$sWL_rup41L?GEY*Ao4
z@9pb5y11a_<ct9{1BY?vb7EpkQaXz0E_Z<YehUpX1_0t?Wf~g#w6(3xWBD7c`_VtV
z(n?qp6|OWj@$m5QWRs5ofa2Exph!hS^IMtD-o=F)qM`}_yuat!wuHO8d+eRy{pE__
z>7ZDlZtagss|m-&Kky(wI5;~~2@0l&_@l5H{(aKe&4JITSgYGF$I|%sF#-X;s8wZ6
zQ17}&AuOE!*hJMjnMp>F_Wcy`7vCB~8Ktv4L_tPovzr%0KtLeme<kvKzl<otAJv%v
z3YOQn2u4S!3ypr+FryTLbnuqJVBkX@ilrHz#a}BuiQ3jy<K^n~M_I;g_)htOWbJ4p
zw$A&_#m}a0+;|uN1}*=jVl$}6sIwXiu>UM({EPdd`EC?3YH=*TQq{(oU^wH!Wz8DX
zW^YNFR0?<h=S{%2932u?=%29#f6!Mzk(M+MjuO{VJ()zrf!oOuyFJDMVVkn)d2OXG
z1y{n1I1{3oPIEMl_2k{i+?m+bcWVAL?pzF9co!+L=UHY<a#d>WdhjqCdcODXj<>|U
zr>`_;1fAXSW)T`Euh-CDVztXwx|rz~of3Dg(p2~~&^9i(BUx=EJy5zc4h)(?4JGGq
zU3;~9w#IITn3_;Rie)A;VR@W>I!;bV4kxxPr^!M;Xhb~6?f<Jw5-KaXKuz)+AieTQ
zoL#v-!NI{CUgy>T%hlTu0zpE+Y5v?F<rzC2E&$NlBgh88USZW3UR=z=ueWFUmVhuP
zFc29qU#6#<b<cWVLfi2oX!GE=%{h<<1_k{dlZ_xs6Y{C+miZ2HkA{W@uYkbBb`j`u
zE^;cIG~w;Bwr!M!Nkx1z?+uFxiK7Qu8T$2V8VyK1=zf0l+x_bdcImAbn<&a3)nsi|
zvpE??`u?CTOi#D%TMru+X(TOjO_$bxEa2Mn{$X%8OF0Y~MQ6SkhaYz#Gw>0siCJv)
zJVh8aJT{uZO$)tR*wGD9q;fq$W8RDS^_nSu`1WO?9*O;-E6wY&XD$8|_U8q6(zB!A
zt$u}5*4mkyAfiX3JjnOS_5W_jeJX05b^IdeNwq%*a^^o>7+VX*BI#_5L$eb3)+)7O
zb;|3bKOX4AkRv_gn}5@N@KR)&+NYLchd<AGxm{pRTHHnFU^BL;Yv%k4ir=V`ji<Gm
z<#he@e_Q8K1l**O+-xKUKfAgI^Aw-Bf{O96XuOGUp(eanV2;<Mau*X5d$#rQc>wA=
zgKSQRn92-;jHq51F{`@P`-CXF(2@WYWruy;ut+G<g)R8h@jLs;Gua0G*3BCytMrF?
z4bL>rvcaK0QooBjv}3M5HAJvu7b(47G7NB1>oD5D*$&psB~!5zvUGp?ZYZql@@eC~
zyz!#3q=)4P|9QH+1l_^0Kgm?e)lO8=OC$Zw6y$~Dxhiptr4?o3*=}r$+k9M3832TF
zPS=9&p4k8I(<6yo@?^xD*Ep_yFRg!tv?N5sF}}ONpL$9;TW_n2-p(6_6H!ex7^q=3
zxO!UVIhoF|!Zv-=$vZd8$RGrTdiD|UY~@s6PtuCj)f@@s?>OX5k{KZxx#PJ-D;9{E
zsy<9aw`@p1-X|mxb9;cIc)Gje*UOQBt-y$z`8+dP-a+vtZj}p}1@Nj6gS}~)D){yy
znz^4)uvfU_+k)@+Q76qT6<>A&Ze~~afXkQoIFjI5#VR^#5F~%UQ`siDCn+L}{x0<J
zZ@e$<M~Y~t`j=?<jOc)VI!Lv6sGho3BIV|W?npo}nlu@_6*-L~wzclw5vHBuCpCPL
zyBUN0+&LG^9n;>6p))775bgG7<Q(LIPLMv!kUe$u@U3AljH%?pzO~ruIe7I@p^=<F
zb}7&SrQF@n;kl;Jtcarjn?mmFlh1M*9Z&z+w1jkyA!R#Gm%R=BchS!+MZBG`k$#DI
zR2;1>H#udt@Bf*)=%4;w8SCg<KdQg8ZI*yX$Py8`v$1as|J1&~ON!nkQYBo@$xWWg
zQjub-f<OGe)c@_n+&dZDp+3CbOo#$w*pXhkhuU_AzRIUOsZ&PG{K5EuP->js`i;Y}
zOEk@qEK4vAuy{4}&4<gqzCJhKlk>D@Pw5U8LXpRHWpAF|o3%?qx_ZqK^-J!}7jK+N
zqDW$4YLfcwRD`fsdUNhUmoFRFf4)*eMSPjaCh>E0d8o*B!cQ`5{ZDc7l&~8%=q|S+
z*#FEw4bzsoue3fcHgVcN$e8rvA2RE<pMJWdW>hvj1a;s9!l>6uC6x2o3Na7Pc4it~
zM(Qq9u+T?N9&7aeLSaXz%(6T+3b7A`RN$uf2D}0tfXK)>6aL};W^#JGr*eGW<ju(f
zaZ)pm`qo*%)tOVK#!jbPKLrb4Ncj%d%I1pfC`3x_{wd(dRqemqoy$Yf!uUkg-*O6~
z1;<<1Q7tWsYG0)L$(yR?_`W=UU_QgcG3`AJsPAl+vI1bKIo6si$LP%XQF<;c@w2nd
zSY|_R4A=VysoU@o)!!M_7^f~@lm(P^n{j(8?d#yGOjXMSk|Cjwq5xJUy;`3llsc)}
z#cuy~rDLFjE4Zazeb=Sgwj{W^o(*vx_?qqA5Y6yaS_)ft6?*w;qFUGc?t&v?qQ2#!
z_4g%3y3jwXGHdzDFBBM)74svQoq!}Ql+!lS({hS7R3ZIMVCvLi1y;hu9>>}GP44>D
zE5HVQV6))XjG&gT^g^wbnJ}=@p;tju{{02}s)XMEmXfb9@m!dhEc_8;gD7~4uN!R*
z8E9X#2RIRhw-#CF9n?wF`#VW!3Ct`5r`T$tZIR(`A#WfPZV=JK&w`|8e<f31YRja{
z@MrGqC^HwztZbKJ+;VSYJeIcxly&2t0)bl|Z;~(dRr__}-(%xXi(fZjl+}8Ke;*y@
zn@$kf|0b`|#2*~4boL8+5XfMQnmFQr`M60+W&);pQ~ZPQl90(M(gEO@l?%o^VU*Eb
zCDqlB4GXu;^kCk5T4G|tjTI-q*ZA>n31n$DNdOLb8jpGR4T^X0yomBOV38RjDsgh7
z%C(#E{Hj_*^olzU<n8XRH7|oBF84gl3~48|68>k{@A;q?M)^t=7HXDpO_%yb#9^pP
z81BW$NLrdN6VRNUBYh=3TXT;7+P>`?Mk~h`cz7l+UqGsYe<V&7jAtuVO@U16h#1Tt
z`uvWApqsOTWCX*hpj!&QdUFm1mBs8>(O3>G$<>7pyM%{mImF`SYM5Pc+;uS&v3=Iw
zR_MKlx0M9#9H4W%Tw0!bz07rS54<+8Rj+?ZdiJknCJ?t#w6=tia6GH6lb;G!9M`%o
z|B4$o;Mswy&3h9Ef{?W9@X$-E;hT+A+(o@Ewvr!WZnRq)YVh8^U!ob$f4!mK7&_3;
zb)~ZFa8Ob8yNEERr?*fEHeRul`8hIVpqpN=I#I<9sSbQJ7#djsz@Us0ll*CH_z28c
zI=T&XyZ&>ZTnLjqG(TL7dw<rZ_UJj5yS(eN%gjO3p87Mx^r!L1z)Pa{n_F-0h`aoY
zU-FjOqjtZfZWDrJM>lc)+sgljrY-7$;*TM=d(oHFYnRRUdW%+NMDzK+8#@6=QQg0x
zoTin<^y0-`qA&htYZ<Q?-zTNF7u71kss7I9j#$MA(=46{a&WyfP%C(!rBb0sY}H&Q
zi$T^QmKSucB(b#l-bY~rRBDkw>ib}e$&1rr__h|$9&8Xt3~RjU;%xMbe}e$ax8C2K
z2yo5twD<0^y*x$mp4whFubrNZN3?=QSgTbBorWVBlDqokAJOZAoL>K8>`Z(;OMRYL
zE0aSp3OJP!QX5XRl6rD?WiMiHjVLflpilLO_RBEc==5|FS~5pemvQLjQs6a)8kWk4
zWMs+3o4PBR!D$pf8>?o=Z<3y57B0@|toU-M$(G3vTs~i7W}lV&EBRT*dqdBv{Xf_&
zhLt}iAX4%sXCL2pv3)+%SJ<P)OaSL?<v~0AD0fGrtecMMRaWXynQ?OI>shD;$)1FK
z;C28gn@=}U+G&wT`YJoU!NMrQJxkiN1E#;QEX}huLT7#rr7QJEeD5=sw%-S)eibRy
z1OkNU0SC)9A;Fo;VP8m|#m80Uzg@8=B=o*Pz1nsTNZg3_W~!U_4%XM#D{N-sKo@Cf
z;BeC!{`Kn<&_J+p)F&o<sCA@nYHIr1-cBeWF7lj^pqW|3Ih_|@{R4S&QwYV$@Gotc
zW+dr=@WSbDstZDLF@YCF9V_{wE5<W_zXUfE0=x3tDP(9eQN11qW9;xF@!wa=FP7y-
zD_)Fo4}5KIkCG6`dSOL|x~+-Xx_`|-?uhx&q~O`mOnA3IDii(M?kn?WtmVv(pH@aU
zX91Lloje0$yhk58Bf|rPrxghQ2{NG+#f;ku9Q`{qeflbI>sS{;pb)dHp7&YZk{HQE
z%L-ptpQXx$V#5enS6;$3<_M~;C<NC3V;T&BKL|ys#GMcGH}a^|VUwfR!_bsFI+#yl
z9;4EO2Q~Z4@-5yP`o+6AV8rLb6+bVP&F=be8qb+$z!l`pj|o=(WHO?@Zlr9^s5ok&
zV&KEa_P=WVM#2h(j<0U#stz{7US+PyW|&w-&o)Bpv{LR(^n|fNed$c~AF8WgI?O0V
zmu$Ro{~%52_V|jH5rliCq*0p=H!HdxmLW>%|0=R2TXs)Bw~hW(KxvWS;pL%x0uS!%
zw{O3I+z^{q(H@}Ava+&YLqak%i$8+14|wel7e#>Bkk4&jSHe*X92N`&2Qjh`WpB!x
zC!!hSH(^rd0%ZbOR~FW{`Lx0DO(GV)Yu`9bWJq(yp<aD5smOF(sdvUt!1~q}6!J`e
z0y=H@gK(5sMOk}aPZ3Y!26hIf+h}+e$+j+UT^)D!>kUm-V1#kHJ)iy%_k+TyA#*jR
zX%>fd8Y5MsV}1(KoEhy^ZGl#r8@_`ohJNVWkkh(9(y+KUR`rsKv`O||_7K~qKkbuk
z{g7cwb^<IP!MH!Uta<aBnH;KHpIaGTc6%^fo}tSNp?;8^7Mx9NvJ=uiaora?|JWaz
zc#))LyYh~?;T040<d`~DKQnIIjl*FZ_2A|csDg2h_GweS$i&Z;nt90C%i{T~f-I_o
zP`P1aD3*StHDAr5ndQ~W+<3Ya<X+ldSKxJD|54k0JjYN~wDD4>Es_RcRm9geR+S2{
z6l;&=i3yNuVl0mozXrXX`kA*%y!E+;!u>|Dx7KfVG+=2#P-^5H{<=d83oKc_GV@?H
zG7rCE->x~LpnCrZh*~p3Z}B+lk|9|=B?FI<{y~|jrD|{?w#@Hk_-$zh=9fpH5PZ4$
z%bTsDj}V3h0r`j%0Q~Blw_4{Ly`p*;A4K!UEWdsKZU`iwk7(fSrGvbJf=Zqm5ETa1
z)bPS-6o2eHTxts=Fii!56@KT<KhXUNnPgc0btyF4Zo79B6#nyQaLXACH2@kkJa5aN
z>tIpl#^V`D>z*@z_?zjLb32U;hWCE*K{_YQ00A;ISdsORVfM(|kcn6b)5vHSXU^Hd
zK-|1*Z|e+Yg%4HD5{LcxVSpa^yL*6nY5(g2LEp8TZ#)edg1ITR!jn~F8tP2do=oku
z2}nw($~PBLOo$}7wY_ZGcWjNJX^WL0Rs)%FdIFY4D?Q_LijOXbkxi0D9tH=K^e~{`
zS-8?8JbLnpKDf$t(U$yQWQN~KAP%LSjqV=)`9JqYsnP8Ue%ArrfhIm801yuw&5>6v
zTG5B8F)NE1qNOk*q+-4gBxsy>PbeBVgP=vbAFH}DX7`)L^!-h~eII&$!QOJJNO#o`
zQ>QQntg^$&uhN{R3jOFGIJNncTY%Si{yJIcgXo8rvt%l!!YndWjy&SF%AEtsSX^Uv
z3etnFKNqCMuFD*eG#WQWJ!<xxfz&Tt<W&7)oJrkYc4QYd)a{iM7PkP(Vq8A81Ij|&
z$P%7RkUhwtLAP+W1gs{mEDs+X=s1I*3*vd3yB+++Jus8d?MNM`5!bn0Gh#MQV%S&l
z&;qBKVEIm$Xm}|{oIiA@?I%`R&B|JX)zs~%RUPL`H(&R;fuUGc_a)AKn78{;G1fNe
zflqct;j2`QYdfH_R7aGe<g9>(hhNI~^|Er{WVP;_ks%A}@9h8;PfU!olECEj^lxR(
z@El3Frt59SaALmJd@!o=2+(YQ8!G=$paEN+EH#3=1+vyL8fkfX`2-F#Oc4<gaVe>C
z=PlLtaFP(fLjctd4Fd!EXKJ}P&LpJb*H6x*X);#cs~}-n4&3WQ_e-<wM#j2=@7n(%
zR&$k?R8!>I(bg#uheq3xn5p3rHpVrIdAQtc5MTKS<V+C17GNV`Ubz2SlGpn5A@_M*
zBWLb7T1pH@yh?1G+!@_Y(kwE)2$8AO)sZ)-Z1Nm#&~@ZUlli`6uAuFv*MF0{F;XV$
zY|KanTTZK_bJhy<UWF@MyePC+ToO=HFYe7X&-<o2uli`|-s-{SVAxuj?HNjeUeXz|
zcwVLO_ir|wVv|CG=cWMWT~(}4^R1YS;x9GSD}8w!)5{C<zBc`^j@BXvT31oBEzzEN
z1LB10M2;P-Xo#GM`PIUB`>d=;2>dVb%SzM~GjWTE!=Y%&7B|O2ujS{Ne+$!bYQH6}
zX01w_;O86c8Fow}N6Solg^NARw63)|m|x6;V=Z*923^&#ro;ACR{G-WVHA9GG7*F;
zv0J-OgsFej2xkw+W7}Ehlz-XC&7-bilom)Am$7}J5k59y@`w?EuYI8mXyxu?K$tPk
z-_H-Sv?hCQf8tOp{BRpxYSe)+2>$xsd;qlRUtgg64NgwgOmi$Yc5FTLn-8!2=mPKw
zXY~^dFRNWa;v@4tot2EX*1HyPSr^~@w}c#6`QkY`n^GiIOySq#(5;guR|tmX52(yV
zw$F?s-Oea|tp0Vp4BwEe<g}`n6^Dy>fY{O0E%@loih+VCBEOK>3Qr*AZZ^zyq!m=a
z<mp2H79)T&$v(rC1s{-W8Et*?_pjzq@bf0k?HWBDx|;HJ_N^k5Ah=4IvQH@NvB;HE
z=e?|1T%b<KCeh|}{_0s;QvG{ibm=tl_(>vK-YSiWYKq?1)^>y$nsCs-0t;Wmi-qH`
zE1~16&wpQ{t2$Z9A0oVtelf@q@Hve<1}_e*%U*9z1}=d3)S$Q<9&8!A>z-vUpR{)T
zG1@T1#HCU9y`Q};EYCeX(~c(pm!zR6qFp3?_}`|@DGH#B{prdb(~U~^*}L%<P1QwK
zHa2>tDQRi8F_S=+QVn&=8`o4Hw=~8oi3L+|A}IFF+FP2sAD1tyxDZj^V!Jm+)X57O
zo96wg7L#i;%Kh?f76~AO?m^rI7dCTt*0Z}?m(H)&)>QYj-*de0?K>q?pEK_?uBkfw
zTN#+sq8E#!+q`ZQs@%PPO%-f{{Ev&XA-Z+^WivAu7M-~~LpYRL`64_ujJ~_jv4U`q
zgh!|ceP^k>=$(88&12T_8^kAg+hS$vTE4Fgl-1|wV*}TEy51)}x3Hc0?*9IM-WD|Q
zGAu#HoSlq;WAz}nINCFVHESz(eLv*^(23U8Y`o#q<anHgGcfv=T2I->kGeeG`o5U3
zcbuolgA!_rC`>)OY04Qfm*MyLw)!5AUVcEVU|gEgUB`X6L5Fo+BP{29309bD%je3B
zGRry$@4@h97slX|pJkY7llEdDD*Cbz`@eQ06to+ABRN=%8ijD5_NFTZ?@mX?suoHN
z+H7{_m9;&c+D|oL+5~*EU?D7t@V+lggoEH-!^wz?sI6_TFfdVV3Tq!?;)N7nY#BaY
zfRgHOmhzLEO5Is9%CSCJz<UlJPkHaJgTVq5mcz2aL<bomDB+u<bGsQDYXoq>EF&1t
zdIhK)Vpeh0rzLG|WkEH1>S=?2Ub%Z=>riv#;ZTZn3oEeGP0dtQ-myO|x;=q9O7o)n
zqcl+CWh1*32wk=?{<la2P!or+8y-&_l|F1bY{jUsTSyy+>1a-K3kju}rFmp8UjUIo
zt>G2mi*b6uJYqZMd$73O-_?Rxnx2YVc9Iy*<qYQUcTa@{Oj20S5!+jRsQ;FSv}8MM
zSz&!=$QKMPFyFOwLGnrV1zFE{6GTf<ieg)q1Ra@8OZH(`#5xTjD&-%G9F`kRQxT$)
z6Myq1p{{8wkCrlDXBkSfbY!@^FVwRn1?(Ju)Edh`HtSm~dE@7{s4p;wjcptWR_T_k
z$&)fX0JZbEoG6_2@}V8<TM>2tG_epogY0()^h$(Luj|L5#*91Fu({8j+_6ow;Bm8K
z6V$cU{v7{eL1Jaw@k*rWdp3!Vp{>oQxBCCpy}v%<zipWTwJQoHac8zdvT~jpP}j2B
z&hmnMBR7(*0D3VhYHA=?H(Vb`*Dw93ruGWxQDK*N#m)EmTBJhWHKlE@K)jGDv)7`*
zAhPX0QGo)VEYRf-VLDQB@86-q$hKe8Fk%yugvs`mp_NNdI%(vm^UtHA)T*bgdbOke
z#sCt9qEyAziKX=8({#J#-HFWZ`##t!B+0-Rpf$LSMWNi$ns7Vfg44#@(h?J^DcO12
zqn{WinHN+Pe%N5QhW+*JE~bQ~R+^tF<iW7BpeJqXi0hNh8GaDmSC3+}W-iwCOSeW1
zp$~R?kg0}PFal@(wni*Vp^(_-@L*oM4&}b<^Dp)E0ZTn$De_8QsmF!-Nj`}Q;&FcB
z8GQle{t^S12lk$BWA^dF4{cyxVyeo92rdMd$#CemgnH3Ph7{I#v3|t2E~sya%x0#_
z1OyTs2(tM}1EE93=>e<F=Y-El_+|U#*fbN27<jIAL;*UJdlS%-qFahJrf)Ggg64Nt
z$mmh^oIBIVHtQ#l5^mE<i?II5R*IAT_SR1BlJmKs6}|_7C8}^m&;g4|#oLf9mb>^3
zji#_by|J#2&WT_L-ySr`8TuG4iE^X0jkkJFJelktR`8Z2ZCB4`q!)v}(DM$DzV=4+
zy?ZZqetof_5&8gkpeyE~gW!$t{MYDWA|MnSVQKgLWN9E75l?1#V%D>GBX_9Bny|x$
zppAO#h<vOLj9}oa-xV_?pL^aJZ1HPFB9RdP(Hp4yJ5#=>MT!Z{GY1HdA9g7IA26s(
zlAcqL23QcCDr<wCaTpj>P@7@~^cHk~6+Bn(W@O=U1EHfl#s35JJ_LM@t2W>M7nrOu
z;DZC4AeL0;o{se8S+0slsg|}>M#jQv6c26k%K*t`0P7O#ml5suEr$F|imvBLNdA+2
z4-K{ey#lAJcXjskI+$v8_HuY0W4?IDE6rH!6FG!@I3@vy0$oy>#;GfDXOC=a0Wc<p
zdhIYXo>RM~%749WZ}ATI6NS-AFa%skHE-~DRBqTB57E#3YbY)m;A5tzVY!HLy8dAV
zT6|N(2f`8RGqW5O1SIL(h=G>w!njabOeNN_7HMxyK(Bba7+HjbEsd=o#$!ai=F~Wg
z<BRPwKHZy%H|hviQC5CuZT$vJQW*@Sy-g8@(r$0P71IQ}0B0Dr#b#Ch_%rxo9S)5a
zJawq6`HN>xCG=uR^0^FzB-Xv3dHTU(K36)YSVmK!{Ah0M%wGkSYeU*;qw@I*1z6Yw
zw;g1Kt0`Jl!LtR4W84Rzmq0py797K#;EZy<E-llUpIcEPh&<1aWkTqw<Qwqw-lkx#
z{fdcJn$L*<p>T<w?Cu=(C+zK0Hz~br!-3vsT-D7O0AUTizB1XJCCM`4=8|NXP)@a`
z_I7~Br83B8H5x69he}U!J%-Bx`~2HuLc=~wjbAF*u|QaAP5Y9PU_J~nHtvy_ZkAd*
zvG`v;1cJUcy;>YQD=WEApI(3Y^2N;v;E+%JlsOVZgM~<gVabT~W<|jS@t-6Z#A~H9
z?sa>!sHEdwD&qV(mzQQn{CTk|waMmS-`39ZBvv{^0it1AXSn<zqO?)xVrYjFQInzP
z9k%Hb{MFmVJXXQP5nHl91+1xfS?pk+mn_tVUIKC^WGQ`FQO5*p_JaDJtHK$?*tKYg
z!h~Yrxm!Nn8142MZhb#rNmhGfr{vyp?c9Is^7YGsaQJM)l2pcY9pB+>zMB@e11|kg
zyn!Jn1jJspii>tk3UD6^gEHCa>FYT#x+^E^rZ}!QWFE_mF;@V93m)xQS-V3oax}>R
z>hgL}l3xmDljW2-EaF+qY3b&LYs?sx+(mFbH4=b%pjzDV-(D@ks@n$DKVpFaC=?t~
zFRrFpZ|hLX?~-GwQfuhO@(V>rpmTWf#w&cGz1suB7(Zn%JXc9MjHrGWrJ#8&=!k3C
zEv1ON1w$3V0btd`Wp4O0&3Y1mrJ#DHIn^Xl%zc{3??)pt&lU8GLH&pOYuv(Bpkh6k
z%l%}AZnLLC+P`)j#d}-N6aFHv$`&(J{}{Ie_rP$V@L3FLXCPv2Y*j3j+_lBpL^el@
z+1@=QS-jWTr{HnfWA<d1s6YH*JQ63KAvQQyRQaLmG?ASc55j7!qWmuoIG;u>K3*BA
zyTS}AtM8Uh=06(j<`qthjUv@ZX6g^SNxYA6R*%pxPP9`L6%A?gzw9iT(iqo@#Fut`
zg@05P$N#AfZb6nVbEx7$iwhZC^k&R}L4HNxFhSSdEoIBl(LP36e7Fv>%F^z}#~oIn
zs~gH4eQt=AjnZ_XJ=40RTgIHtm4w71-8G>&aGEu@nMOq_I1s33)+yKhn>Xh<h5#c7
ze2e)?9QelR&D6=pFnLMk<|?FfG|H<{YH(|5PDCMei4TC7QYdw)?5Ix_{-Ux37*4Kv
z%>P3j&J?NKk-@`+e@D~S7>l(J%Op%+TN@5gK@Uvhl1Kgpk7I%(jiG27<U$pFt*pl=
z_CYng)@6^zGCHyI5dEZ$v3H<kPv5NKt~M(I9OTn>laFc|w_dS}E!iVva8^f>2+bm;
ze}(g=K-hK>_}`}Uq;>AS9ook*$+jZB`pFUlL|Vm^*IZmUsYP`JBvg3Jy1B)XkJa8U
z6_xlry53$f_58o>|1jlHiHvF5$;b<t-~Rw)HE4r*J?j^zhcQMZmHjLFo%y@b<><S6
zrA!U|V97o(?8pm&_y5jL@}W_}`KdD@!m^LByxJ+7Qv6j<vTSrps*uO4eter3dGU6;
zM@L7e+hZZ4qodcGA0NNJ{~^mm0@O&!H8we5|GBD~0J-_e=2+0-3q6Z*zRsU}baLI7
z8*Tq|CtZSm%nDf9Xba4|7-dO;KM-~ppz`8aE^_dmB3*WaQfMBu&uxnNZfdasYxy@p
zn%IYQwxh>pu4p1pfia#&pnx$8*HKN2Z~8TlRn1OY)wjlqNDW(evn^FJoR0tFOrP&#
zOFp!xrKc+anFSql6JM<)Q1`uf@nV@x6I{g7m#mR(-qRUepq+W=p^|tiMaywm0vCw3
znhVj@YRjJ{`2(T1G!PUBo_0<0Q!~2VUE<xWTZ@A_94h(YA3;6Y`Rm=YE|XVgc=p~~
zq>3>9cR&O2h^bQU!yIDQ;!Czhvc!u8sK)$}53AttoCXmR*T7Vf!>5mg!MN5LZPATi
zlg{a~ak1v56~OaWESf~^IR07xehw1gSBE3i^A~PESX}c1IaC7eDl!KF6|sM7ND#!u
zU#f5a7Z!K_HqqvR8d9CgEt$uzA53C7xE0?S6F&a8eX~j|!v}#qC}qJBLEe~k63CGK
z(___kPJ<}8b_W85qiZXKgFM=VjaCc4aqhg*s()rt`G)WT@f(ENFV`$J*+!>iZsc}@
zq|U&U^5%2a#9#`O@lHIWG|)UnanR!1q=!)pu!T{x^iyp(qD&^Xnk#LtAoQ<S^T;y=
z_Ys(D<~Uql>`|pQbON0i(84aTe&LA({7qD1=ipw~myKjU1(N3NyHw0~pVaQg{bAn#
z-0(?_w<CTZ3WC@d(TeL_oSU97$1YBe;%GuxD=75;mu#N({+%RY)~PBq^jPbTrKn{D
z-xnNo)+Vu9Xh@s2`Tcnca0lyyxB6Xrkv|*|b(gB`-T$TnLOM?H0eh=drgD0`Ow%{+
zM?E^Wdo{B65c>4=LfR}cP(FrZ9_m>oOhd%-CMgTfDQ5?_wkfIHKL9L+`}VF8h<8`v
zFj@^VH&VOw;INPdR}h;6MIxVpr19Z)HX>Ix?9lw-?O)Jol)RYBOhyj*u`}Mag}|E-
z0$MzBF@6G`Km41w*80Yd*jjN2Q?hpggPWiBap3A2S)UuX7dtUMUZ3bn?0ZP7^R`ci
z;H(QTKq(<$W3#lvd)>oR!6yU@_s#O(z<?l3mFG`QPm%1ry}kcAg2AWIxyHvjWA;4&
zU;<B)i)r1=f~3FAB_(t(u9>p-6R9WuFu^wc#Q+m3b@4u82GTSH4g&8nbc$4~`S7Rs
zizsBMe&0f)882ScUy0keLIw9*&V^~mjL3O>!%f~GGKO-Q*AH!2115jC3VjPBV=j8o
zB1-&Nj1^6jsuPVOP<{%=Xo!VkX&Ux8T--D0oWSO+ctcC^OB|7`U96TAyJ+jPesOE9
z=&ZKq3!yTv#()2NmZB@F+C{4?!nh;ev+f@gjwlN*_BO1SO`x*lliHZ1_NMZUOYbgr
zenn6JFLupJr&)i4Fh4xs?%hQDXlHlu-K196c6a}APg+V-`JwYS?E_I}*;6v)ZGB<q
zLLO(w*9Y3%Qn<Fob*UOX<Hee?JuE!_pLFi3M&?Bb@ACbd9}ckY#$mKTTPGuk&p3ve
zw$4J*HqFlS1$wYTJ2!po(tx(_49}d>cw@iw>w$ixWWKM*3ymDvZ~aLrEPfLMpm_V*
zL5b#5<J8!pt#g+h@fX<sFMt2E&DWX#Px@mPuBooySf19T4x8n9PeNzd{g~$CMj0SU
zN6!(G_I`eKMIwZrQInlH6UT~_MD{pygFfjT9a!AaI$Gc23x(~<a(D?;rSi&bXwsJ3
zTX8Ma9*E7=xwOsBrktLh24PCDMh-Qv37t<G;q(9@bo>5)`!E3z|Hsj&5p8PMN7X;j
zAbvbX#wI}Zo%41Yu7LQCH<)Aotqt~~qtuQuO-jwRD2O<?8f4)AtB3Y!ESC7l@}Sj#
zCQy;f<bB+^*XJQ%EAtZTng1ntJm$WdO~bl@-r$Rix$)&%Kk*}peI8O%{S&(se`4xi
zZS|~&{vAZ>^oRgTj(9j`m|yB6Z^BP|`wSzRJmbep;(vsbDQE+!L45>T^^<I-^E<p(
z<42AElkv5_rGCp3krtifz`jHQ=`{%Q-py+HlFY`7-1pg+mi-l{Y3uHdd9G=e?ehKg
z^;M}i;S^4#WLx=y?*I$*Ffv$gKao3Tsm7o#7EF%+Wr+5J7nnAC^cH}hZ1CU!?gpIa
zG>@rkDeZanUQ7jR?`O}``MBq{Qd{4=i%6Anom(qM3NWQHuxht>hM_YJQyNtYPXl%-
zyaNjJclC80)<2hbMSm#j##rvTI^=svA?h0J<GgfD70FPg<l(rUNRwm&61J4vq>`n^
zK;yq3{V$AKWSQrl*t7b-<kDsIm$%?73{1soq!ktoruE{=U)_H;-d`nn*1=f?)hA^@
zyW%%e?6*;G!j5f!M?&4v#*A43XJ_KB?*^Vf(q{r#h((m(<AWqiU>O$FA#@*_+50A_
z`qZpt5T4G`y>3>Q*-Sg<iRhdzeyJs5@j#N8)k3y2mW5CJZ(O7{0w1K}hzKk&4voC{
z4^N+pgJWbe`M*e1(#xh;>ou`wM{MQ~oo`f^LWL`WNb8lidOm!!7#2lEZL4zfxjRjm
zR@m0i{}E36S&=q2oRw<8oIY4tpgW|y^J~3n#pOOVc-9^kbEu&SM}kbAfrj~PUlK%l
zBN21ZU%&q*^_2x&c&n8_)6cfa`7ricFaNyOP(YNe$1<f4dUy*H`{<JAy$WAzY0>@4
z(%uRVmFLECTh9~DW?P4fLsp9(H#fm|xVCfsEeR(9JPvEs<abu``}6tD=`6yhVdAi%
z!|6gkbC=3I2kIwb{zJiKnc3OHrAAUhH@h73J{NjP+&0}j3*KS?_c%X4@3g#B@&_`&
zQnCLDEb5C`Ih3^Rt6AU3paGTYwLCOXaBAX`C_62Q9cVCfR08+zgLaqEft!j!``O(o
zt39;1UfR>yhN$0mQ;!NniUCje#5)HS&0{RlIgaW2nZ*r^zJQ(wU8e`^)OJAoehQhr
zdSqxF6kkXrdYH!$0*De6mr^vAdyh3`E231!*l`!xFjBqn`A%RKly^^GywNIlc_;r=
ziiNt&^#NYe`M`9S)e0Q3ss$+KlZ}B5&}xH&{#FhS20*JxAnYp$)2_^SoCC}PJ5V<Q
zO^CRqB?EB&2_@$Hp{ehKLz|P#(-=YeRu)jA4BZ^=9v%jUjwd%GkA5ASzk(TjC$^j)
z$1{2pIiO%*e)e<7d=m`OvNRkbP)df@*yxF!OJu2*T1_x(6zO&WzOB}EkC2&}SzJQm
z=z6CZ8y8n2$LRVstEb<@#Kc9_H3={qYPjAmsQjo;yC3<O+AiRM6iMdio6tyeRcT5r
zH4}CDH(4WJ@?_?EIMl4AzN;4u5spWEVK|Q-I>mv0;RnkHRhBc2_j@Y7p?m!KR-X-X
zT|t?|56)*m{bn{No7wf5K9$1Va8FWWf68qFy8#e^K$?3F%zI`Q2EdR<B8S-}@X!Dz
zB27#mFjSd=!G=X|0^3SebGm?AI4}`70%J_&(1Y)EwVnRaG^|!2OwwO6qZZao1jWcw
zBceS01n9%+vqqC1+$!rS(VO$VsYWjz4UKByO09UGM1B{BN8<<B-B^KUA~585d@hf}
zC%_cUjhN>bsVwK?=k|jGK>4s})r*%Qe5zzXx)<AIZODKci~D`7x>>mh^jwDv;+L;6
zY0~{4Lj0%Vm&{4&9-!FHzSySKM26l|bfugj$1^Pi(B}1hmk};s{8s$o_BNsr+HmVH
z6cD%YE&X^m#XBQ|F(sLBNG`SAHvO;vu<3EPdrUP}Jq{MX7s=s$QO-t+KfLc<JN|yZ
zpAFNgE(6!?=kE{ZszvhDevetwGcr=r&;;eZ&(XLxzJ)*>B(fQy0N)T|LHEevVbuz4
zrmL%~-c$j+ELAWMNF^a50q49q2%oB`8Fi>2BO{VCqJI07fIai~0XTXw5QvtR_V@dL
zi3I~R670>2sr-TI>11FO%m_RJ$tgTxVPW9%BB`xS$Pk-vNH$)i7XqZZk3-v_vuK9|
zS1+DrL@eI9nf@=Sk&0_zj)Q_<`G?w53Cj|;#q`;eVLZ0@bp?inxD2j9f-BW}OiWcn
zHy4ICN?!~WvUb4S-LAqC8VgWbB^fhiKZi@!rV(?|CfVR7eVpm*YSi}Vy91&ia-L}H
zBXJhbhFfKrq;}n<KFfr2QwgNuphoh#npgmlrtUu^{W)P{!4?5rg{uR1E*X&%N7D4v
zkH};x0}J}9IJe@IF(5#P2V`Hr%^C)nd;+U+dB(1}(I^$@)4@1?Xn8pYxc0W=IQPBX
zUH7FxJm5g`IG_XuO-`0WSTq`ZGb%*Ysd0g~LL!gdM@YhYetsUO7^HOvKy1&>&T3WL
zl7N{KBqSvK3Kn3E)U;v;Oe+L$58G}5ado;o6<zDJp;5Q*-w%2MOcH*e6@3hWnu|_j
zB%sp;{Y;549!kwPj4W0o={d%euh#D^qSxpAzyS2?-7XX*42zikG1WX{pqScRegZXV
ze4-2%<)Vs{W+Uke<kS45gX{f@4u*wvW~XDM1$2ynsg;^N9~maoo)N{OeNq80#X&dr
zPL_|jU!q(fnG&Pg$+hi;WS?hZ_8wyX-VEhmJiudpAHWffd#T%d#`~%ID~n&KBXAwJ
z3(x)kG@Kk?;Bc4^tsI6wbmLf7@maT%nXaY)ZJy{vRYFP%Ahh6Uh;pY5MG%X2jc4OF
zU<Bzcu#|edk6=D=rwYkVp~`Ln^Uq3zu&<8e1q!h4`slUC^=OL#j2AksiW^M_o78du
za0h2ST_FR!Hpu^5qoQxo$cEIlFF6QCT?*MA+)$v^+$Ue3=CR96os9i@9o;hiCc#-!
zcE74T*yGosh}`A|IvCPjwfp((sk6Tl!qSHFld8Y~!vy?isE(k)*}QK3V=1N@7gPSJ
zmj_G;jvzEyA}>@2y&zTn4gPTp%L+K)gf1JIPTjJoav}p)ny3v1rYvvb6FZ-HWrnEf
zTtX46btF8W7qVO!`~4sDA86<n9Lp-<U*}ZxE;iYoE?K*|^>~NjvmtPBaL_0u{rwD&
z_?R1j#ff>P`aqO}QRXgiT4LkjDL&nhEaRe~kt}xqosj{L&uOk4Yjb_Rhw%s4Z#|oj
zCt_1f<_;?<VYRllW;kz|o+kV}ISFYso-c`j^Nk<hSy?H`1+$g0fiE+*ZI<!c%|7du
zZ#_C_cPaC&*Ed(b+ugtoZfE{@Dwh1mOk`27&D?)o<4}PSw1dd1ck_RIZAu+0@;pgW
z!(Hs|CAE9i=@SJwkfz63`NzBDWlx)CpTe};@+Tm20XEpQI-swg2f`p~pqmMHDmZ9n
z2W&n7r9bvp5Y4@RIQtKS2=Mw|T(8!@Yi(L3a`%cMR=11g&rth%QTRNvg}Ak$A;n62
zm{Ct0(_{7ngowsydS7Q}jZta^=ayHcMghoV>eo(22$edtss&-Y)>~S2HEM4~4#H%H
z%gV~0&c{bULf2mbxpXq;w(tU2T{<|Fw%i>4tYJi(2U{VqJgafudKu5GS4Wim$eRQN
z<&6AdR+k+)DpK&5=u*|2oUYMDIA|A5bPldB{Fw=0Y^H|HK*ey@F055I#U^EX4s7}>
zLWVF!d#f#Vo0AhAB}V}KoB#fdh1jAIVW|`W)wD?~f%@ulBZGR3yVAMDptEBobLC6|
zfq0<!&ehepr4V!!3htLXW>LvRjTEu}#mHdisU2-B!A#^Qt$7Q%Wpy=ern{MH4oNM~
zljCE<rPe?oXK_4Ul>s9s63WWY^?k3Xfis5N>KHKz3A|R5_mvHHmioEV+IPs;U(uGz
z&W_5HF*-gT^>g`2YGuUPRQ%<oJ22r9d-LXvR=Mdj0s;aYY#7kb#K=<0fAk?Ms75MQ
z%rOFOIpCi@nV3ib?g<Rx0F~-Pm8<amCFMwtk{QU3xntW5o{!sVfS1zl`1{NbYX4wl
zI{8eb1Ju$;HSIH528gLp#vgQ8l6)0EBuDmBO0&Dwf&7`b*08uDtgA08piD^Kgf)w=
z(*&rF7)`F!siIx@UHi2dnUs#BX)H!2+{q4=?ZIrOX1JO#qU{bj$tKX30==)BVb)Nu
z#F)2cr%DS}{9+H%U~8mO$I?t`);-(*0elxJm?F>@P8X-O#Y#0ys+`e0ESO@4TX2>t
zV#L|5ZSUc!seyqA@FW42YV5R1X~w|u0?e-p0cLe(T0&eL4v?!sj~*6n84TsX!@`Wl
zWSB_i3B0JZIa+{n7M*n1S6P)97YSq{KNslr=69>Pz|WoeufOAnM@c+KQbWpoGVBF?
z<^)cmv8Vid7aa-|Y(Azx-tneybhY*MdJVjFGnHmHoA>%yurUZrE<@=Afm%RP4m@Sx
zJWq3dNh0b}dcag8oclrJ#iRctNF>0L@24ugoo&&_M$4@&POE3Ci%bj*hKGwS@(HY8
zfX@_R2kxs^tYF?~y)U@~$Q>X;3KHND^IIn6+M&7ZNK5C^PSa$3=l+fREpM+7qE*qK
zv;uoN3||HRhqPcBsih2zZVD5u>ozaFUqEyswcJ&wdq%rq=F1l<V)j`@NLB#JsbaYG
zIg9LZ$k#cC-`(V0X3SBf5VBwD&*FNIj#8Bt#O^ms(Iyc}&rlSaLrS5j_k)v`pVVIp
zI;+ty+A0PjekftV_Lcv_W4IWVQNoAgg@;P2drrJ`!L>)W_CIH=-$%LX3?GwE-Qdz5
zTxhTO^%{-59Z#vGe!NEmvYG_=J6P&(iJ+$h2y`5J#axc7(r{p8T@hd?jfQpgjL^u&
z+M2q@19*1bMmd>Krdpk_fIzZD2rfJ;@KNI@tIl_#?`avY+P>fbBciscD+{57nSTTv
zmc?BU=AY3iO@o?PrJz7aZ7Kx$zHt1}WRQm+oLN1)PnU%(O6Hceu$MDuAghxFYl*VX
zVmTmU>)su`Ix3+@W_RhnjHqkLu!&h1y^HNBT6Rq<gpRcw1my6T>c>XPGUYBJ=2U`L
zM&05LD^c7(qc7+4_3H<Q?F?|}%xB+R!sp5<YqP%LM=eeZocO)ue^Ro+z|HWiT97@|
zth!0;aIpxN>VGE=lG}<$2Vt#^7yU|+moqzVXL_}@GhgiDbc<|iy$}ZAWz+$|Fm01A
zvW9iY1DJ<C<`YmKu4&d1g-HS*K=<P=T6lamL-9bL*}1ty0XJqiK>wMYZI7+4ugA$J
zlyd!lJAnm#VDAUY?*#A~uZw+Hi5*xjaF1RwbqwgszP`SXBXe+%BpT3V##GCH4G+)O
zZ}tTza#K)HY)+OgL$OY0Z{wSMZV*_V*5yGoYc%2O0M}a3-<SeZT^C<sJ4p>YS+~nW
z13=nh{Q-k>lXWhPmX(CS!gNj6W1<Ppb$7A_Oa&LF{s8t{L!0PD>#9HWGhew@W|28g
z7Z?~aCM=?=)Y>!PvCGB?)st2Ywu>YZ>>R}_!35|Sp>|W6Om&aK0<=7w_C%(dxD{P_
z7;C@ve$}FllA<F`&`Y0G&(*psFMwkh?<a6XUCyawrYKPHa1q672ZmG=!CI$>eBV9c
z*5)u{eT)UsuS98*Fw?V?c!I-O;mPM-;V%}7*2;(4wl=p^_JsbSAYxz<FZIpcu)@pf
zyT?KI<a&fGTUJ0fcCuYitl<TL5!6_h^7xOQ-zq9NADdcI_r))k7Y}C>`i$NWos<#J
z2?)A$>>B@pKY@}NA~20|V0?B=Pz%x^B{z2*u<rs>%onn3VA2Xe#$o?9*zFhJ0T;8s
zfK<r0z9<3efJeU{n}+1nF*E3YrY&nTz(A?cb={W0Y(rRcTgYWQrgG%*Xs0A`U&*bU
z^?mhM#9Ry}D&?qmOH=O>2#_MG1D_mp|C_TiT+P~0gWd>m@tVJpt2^brB2XvGS+sn%
zp>|+IFEMvpP+R%5uycOuOQcF8!SS-z{?!2G>|KA`?0qXfKjt1Y^zVP>!%x0M)C<(l
z>%2P%C*gIcP9#l-86RgeS$}=o5!>K>DoHu@kD6FP#%P}mb*s)A*=~0}I2Fd!W7fBL
zr1C-`?gls-wYG;5B?@}516~qr(jaaLguHos;#p{c0FO4OJDLfcVuFGD<~8VjK)?>e
zba;TR05inQe7ZlE3~rPInCJ|`B+)i#j*NVMzCYIkOtk=dfQsS|TA$#q(~a&yZCD(*
zvt_P(+LGbK-=d?j!8bthpBl@k352Id=#_7s1|FMdn;C9gmXGr0=5!!VAqXP%IaN_i
z;ca=Z^w#8`RNYnS_0##X_3oc)M?Fw(e&V#P(TN$lg@yV6{-e1&p<I7#!QXpVKQLDK
zE~{0bgqkEMU};;k`5g6}8M9l`ca8<4K*QoGtiUMN%}S4$uAfN_*Sx%Pv3zEDWwKt)
zhge{v%qidWS9!(D8eOCh{PxQJ@gzw649I(y;aN;xDMez}h9sK(RFNKhB#_THQcZog
zTmGt+yubqW2qX3+=@2=J4OtqAT4Y{N-j(?}<TRVcUuewF{&(V3%~0j>v2jQJiB^T{
z+Y`-)2)m5mSKJJB(?@RSdu<S@SInR*d|13O4k+D08DFF+JY(guIrs(i>%b>%r$GtW
zD)torEe_X)na<WYghE6-Mg0zae0X4{9~5xyK(?FT@G#w4p4+*NJP)EW#%^Y2X2NO#
zxF|B3Jz80gei`bl3KX_C9%Du?KVe*WD30Z=5g6KOnIX8oeXHCvMi19NUzE#fDK8`O
ziSx^}pXl70D4ed2hL1&}l^j-CbP>b9F;{gwvmMuz-`Y?F;yc3^r}powLc&VO397Q&
zuKc<M_-5$d1{~NS!(&4G(Hee@<+uX0u&XQX$iGG#mn$RsUuSRvi%O0cXzmTZ1SsK4
zN8H#-*cjaxZteCE?0z`sFsmjeXMr6TT0~UMJ2w`v+~tId@0hlnFo4hh!~uJFdWy|!
z87X_-g%%>ZI50c8_y7A*k%F278xK|Lwn32K)p%?v;x+cTP)XhhPAi25%VAx8eRN97
z)6XrR=+b}b=oJKk+Vptu^XKflOmW~_-$%Rj>CJ-RQPVIBUJUYJU9Der>amYDT1w|H
zR>t&&c<bfGW1HK%@(qhQj+*uwLLH*%^QY^}E%*>`GAcRqI&G~m#%I?%esvGGdf(Jq
zXR40`W-?#gAR4$YcLJIoZl50dclFLWh&u!7eU{$gz6lL3$iiT+7#1}%%L){W5~0VJ
zT5Bz~DB*gxe;k+xZ~q2WHWUfdHfi2;jJ}^~wL>%B$@7GK4aJM=)HOKnToxlV^ySy?
zwc~HliLQ`Cny}DZ>m+i-5PySDV>k%G$x3(1W~c(qzyD|N@G}P$PgVh|*HTzVSH@{_
zMi)O6h;KawDD^!)A9|s3|M~F0ZI{z-zh6<1)0z(J8IktQCvsb5aMu7tl<u-K4lmn-
zEG*ni*r6}5;<os1r`e&aw|BL)gS5>UVqe<b?fs6!Sm^uoF_UMv6}XmO>Fa|J5!1T4
z%H2(22HDVmVx|8#z{0HJT^h-9kJ~kOASin2!}Tz~O+56T{;)`ayZH}MDo8RlRi{7I
zCtTIZ%#4*>S*DPP-hc9zJDKKug}=DLshqdTDcI0WzGu4?Fg_Ob8*%FJ0|Hd2h=6Th
zKr1+^ZP9T!@vZv`Wika}fu<#b{<%IJclXbJN(Pp+LA5d#vy$V|haaSngas2@9@kT=
zU#~ohO%N0}S~cTqJm16?lZG10ufD3AAFEdE7{c4Rybo0RI`3KNJoFY?<15H7c2Ip$
zDEoUeo|C)x{>SKMPpG$tUxf|%qh%$CO0`dC660VhTED3gNTRemfX!3K+6+kA74iB{
z;PD~6bBXe+%Hiexw@oFP$-}Zg^L<MRalpx{5v`vpSiqSrWDi$3Xu&I%f|3I_cIwxa
z``yZ>u<`hLBspd{`<c_B+<c3ZMZD>_#TGsCEc_<9{09fPI9k!Fa9g*Fe;F{hSTRml
z?x*e)2=X**yj7O&S%?keFQzqBW|8fO{{%o9hcfui_Nw1AXdMsDR}?1dcQjtvm{VU#
zc;Bp|$~siX6TQLk7w}=ga}(j~P{vEZLwFC4$qc(?UEyHfEpE!u;fwwjR$Qd~y&hI^
zXuy*&oP8~^qiwN7wr5~#W?yBnfY)4RHpPOQTJYZu`F&OD`Na+FQ0A4g!i^hHnpgPl
zuJzhG^$K&~R>|_{bor;P%^x%XAGqQ{afFS7^A8BZVT9mV<-cKlxje^v(FcqQ*Bw;>
zW(YlIvE!YR$OFi&bFxtv$@C$A%nMy>VE%6@P3LmFPh6i;;f}wR!9KkA;Ek5*iL5%U
zIT$Alj!58wCxS5*N(U?yL*sY6aLt*m(~Oh#ysWEaTr4Sr9^zOgOKLc*Ih3>e>eWUS
zCH~0dYd-FG^{CpT&ErYF9vcFxM$BSvn(lJ++2gXvp%?xeh-`$Rg&zAxCco+liS;i~
zX?-un63K7((<3Bco;58E%%5^hXZ$LuNCh;j?;ba+$2EamfCss!rle?9TH^eykOF5Q
ztQQnFiRO&NMofWw0nDSU1LK|oNqqjrZ)>iuo*pA0w0lY>l=6a7Qi}=rihd<ph<hlM
z)YL*~Q9M+8d(&L-IcG-1OG*~~`_#)TpRl&J_Vb*BHnG^I`~ivBi&~gU73n?BGvS<N
z;Eak$rchi7;cw6V!ZBJ`)Ie{u`^Ugx{j9Q(y@01BK7Mh|DR-1%U+<J^|DFra8q&Xx
zR>f~_#+hCIc=P}Gdds-1w(bj*5K#~?2t~rATT0Rb4@gLNN`rJGYy|-o0i_$H8|iKV
z>F)0C?mHj#z5g%wch8r@@!<39wf0(b%{9lEV-%}4KgG<l!?;b<IqdU2+O$4hkN%aH
z_|(9@Q$Xxm8u_yPLS<j1081VgmV(vbvEgp6@opWDi_<#xY`B7J;5{>ESaL@s;bJC;
z*NJ+l5aYiYVSn<0L=&o;vl%2%#&ZC{BYHx6Y~W;iQo1Ofo#yB^gF;fgoRwmp!L@>d
z0=~l;|0c2fFqN@A+3Sb`Cxx+!%`1<gfn-fl6*M20!nD50Ygv3x&KrcEMN!PXs9kvk
znLVL;H_z`25%jNdGJVMqcjr+1iw7GET#qb_XO)ujp1s`H8Y%US%kHJ&w~XdP50^Pt
zy^ta1c-iE?53~&v;mre&hpf)03@;EYDZe>!ZQ+wJ9Y`X4PJPr*n<RRYZ=OU@QghuW
z^Ki86ro6Rgoo#}6!!P*>y}7On{=@Hu>#DPm+T^Sf@}5D)m%l~YJ|Zf70YWD&tWDU;
z#>Mt59y7V0=j#524t8F74V{b@e@TJHWT`m&*+7K@01DE`9=V*@0}WO3cyCw5xrFnj
z57{I@7-f4y#<Qz`T;>3FZ4Q_rp=ZQ&ED5Dd_T{x7kd9D#QGZBShT)t>p4ydsi`5Au
z!|6h{4KhZ`<{J+eDqYT>o*7-syQ%nJyl1DKB*<DiZcUq;Tcs)}kFFJCt!oOpxEMnq
z($<n%oIPRUY*O%1r)3YuDPauwga|wV_92N7)~8_$X<Co5rIejM4uy{dCjZ^l8@4cw
zbWA$5V9;f|mU~*X#rHM2WZgXSU<mfl7atwO-<&20&PN5(>*cFOIoGtRE!gR9&f#gO
z-P0RpRkk&J<HCRnqo4<y+o_4A{vLptpW1aeC}0XzMS6G026-eXk`5KT^>TYo?$qA<
z9OSaL=UO32qJRu^IQ?d0V<Q<bbeoRF5`teW?K*z(@@j3%O<BY{8p1|KA2dbD#C3k6
za8NbXt2cH`f%}ix?}r_&Qa#KGg5yp_a^*CNYANr{xegd3yDQ0J&u?7IU+dDczqyL-
zX}%X6_sfL0E#bMvjQ)%!Ti8#lyoNO6SFUF3!V9C$8&@{aTaFz@V`b<mlsE5@>@~n&
zxF0cY<*^c<cGJoW8X=I0zP8|);b#kMsU#Ur7kdx6KR3McstY2le`@GGY4W_Q5qH75
z#E8V#*ucSXa3)cQ7WbEfY;K2=+OD<R{lN!O+>*-z)q!ggJPuC>7qlkd^w$zqxHeNW
z{a)XqdcEfanrO{L+FYf^L-D`Qh0qh$AI{`=r1$QfX=L>)6g%@HEH|&rt~<G3ZIy8&
zwr?I79PG?g;{~N>h1pXH35i)oGHCc?;OXm1ctuM`H${1V;5o>Ol8dPX?OD<WLL#E?
z;o-)6tYOL`*}A?~R`LC68kHY~x=RW&z8a_lFi&%PY<J~@a$wiZTO`uUyH@$NpVq8z
zc9NsTReyS#)PTsVuu>YIM40y8|0d6m5ojhxW$pfOfM=kZ!#(AQjG8(j5BZtk8!7r(
zi%;W;tr3+EMO#qaBbm9`E4sB`vfNi@(gG=E{6%OV@ES3MA6}V=OIP?79`>-ZIs9AZ
z3zwIl1RG)4njN=UWV;hzvJ>n&oNFj5`nucP{veT*#uADMUFtkx7yVGk$rIT|R~G=+
z;~c01!9C6=Jly?}4lF(PS5(gzDGU2TIFw1__!1xAWd}PKdcvE=anf4L27}s(vu0P)
z4+gYMN(LV!UgV~f;lLdZ1#v8j27_v2i9boerOq-{5-fFNXooA?ZsP?ALa|I!1-PEw
zNTPV!mVkfbua}h0A0s2leSCa~$&+XYhm`GUjUH*&d{j?*8P^#j&-J^+ZR(;H3u%)V
z73pm}qh7oNN0!k4_-vyF-ZqGE6TUJPB`$H*j!-U|8f^27sa9KLFT?hU5`Xk&%W+~*
z-q86;7?GZ;E)6Rm#=E66t6rU{u1S0C0&-*;{%t9f+d&dEF|+ES5W`&UR@!Cp!oJ*T
zo%5msC4uh!(VB*~_+PILdx{)O<GR>e&pGS_nNgFErJ%7-r-9oxW~!sn_WFonSjD^|
z!MEH?vbW(HUb6waHp|VQ+4)a(rF=;&;<}e@X4Ob%&SPN?k~xgnjc~KNuJ-36Lu2O}
zycVrvV-((VM-Kc@XZ!7>b~mh8PiEruFZ4*o#@00zWq^*3E*8p`&~zMMHGC34V*i$X
zz#yy@xvX{3zfD$sBENG$RRIa;6@nde!n2R4@?DkL#8=@EnwR;_9sO*@jCzoB-RVf2
zqB=w7rPi|M;Mca>QVn;J(ZkcYR&%zOwCrKW^tBhr+R<aWJJxKDiX9jK<Ev~trTqMI
z`n*gEYap4HCs?e$SO+FeUu9<hiv1Km%EkQl@$&q9zJ`H{?(_@%kvT(`n2*nX_e~Bb
z;Moo(I^j7UI7}`~id7fC``t_L)e)Mulixv?rxuvy9;7xzgEfZSc-(~HOia<h9G`K*
zWQO7BM&VZdJ|iQ8(B90_(XFD4AnQbf(6JmJfYl7|GUBjYm@pDTLYM%6!G0E2x}3EY
zWM_XL+kD1u)tVS)v$G)eo=$%(NKHB{IJg1C4*pazQ&%rDK*AArAC-jf28w<t$tQ*j
zMa3G@A&pB)M0oQK4}q<2%&=em+QaZqgDw%DUqx59j;Ez`o5^N4<u%lg5j#s?i)eLY
z(reT6X>#w;HX&cxz%^9aR`mbutYJFaUwjTK)=%GKe%+J9ajw{FvI@xZeB|)+k=s<p
z;>q3^0V+(MQi|8sHMqLQ5ZRD*MT)<}D}97aY!=M5zZeb5{B?V1^vhsxv}jyn?Kc*m
ztWjKEwh2>>Z4BMT33JW@{Jp&8{*<P><es+Fj?1jGBdv@ZVeiMAC`wtjwf}6gkvr`7
zUvk;tJR4yN6aP^7wfeQyo8zH163CKTKg<pv9A?Ogem6Hl`)ZhCqR}jZ$v+y9Rf!xv
z*Wsh3zHxZ+p_ngwO)jF`EVE)Z^5xz{k1zJb_62~j&KcBiG*vwLojjJ0f9|{}X`)et
zZ_=}2!h3U8rkFA<*IW8vc^d!5sg^ofuw=x1^XKn;6?;TvBjFVpV1LUS>0v|dIBcO9
zePzEe75SFNqiBbf@9nX1xlmcK(GgkNEvy#(upi#cQgb^`k1#_)-Q)WGNdmlN+j^Vd
z$?7KFs?R6Q7MrH~#=fS#x7B^3D!^EpS6%6Jd;|{Qa4uS49x99^iAO#wFTZwTFOzh~
zRJt|sa&xSZ7w}x42WohcFkh@{Y)m=f3@xYsXn@bO!rRQe{w|3S|9tdT{V^iPw00Jl
z#ow}w;a2`zyit2M?;+QytN8xO@iq1W(pk%Wx!`5WiRBmEfnN1(9-U_5sTKw+J~u=y
z`qlApOK2Y18LOD2L1R1K2Nwz*b<sDiR+RLQkSxqXK|x{KPl$s^sD&z}<Bc+M%sOEn
zJc5Q(o)G>qo-%%z;@6G~70B%@!u91h?Q)9IUnArA-4e&;R#}|%u}#A{l|P@`GR!P4
z+eO6RJ6Lx>^Jpz23FII)x$*a6#g5yWSf%2hqyDJNO88P|M939k7sqXtgpK`oybDsk
zN=2M>A=fTcOF+uqkj>#l%Fw?2-Mi0+^}9^Kvssd0O*!dN949IyAT)~p`ctgwF=w*d
zrAy2ntP49QJN1Q4T#sDu{7P688A_R))x2araYCv8xcu<-=>FR$!9D@RPj{*8cTF55
z2b5=TPiS;P5%m36tHyruPRE^ae{<v6%e;>?pYjFtY-0MV-D2Uwe-GblZZxSpy`e%u
z0XjmN-xMz~M`s)(+BWkzntiu^cXyj@U2-+W%@}t1JHAihHR4j>75r&#VW-_w#7kWI
zi^qNECCVywy!Jdy&3;co?UF{lO<Smf6bkSw({8y$p)p5)C^}iP`_$;EloS}6MtAe1
zW%T1nXIlN62*rJDrO%6^#$T)TTI6jys=8Ov1~$)*kS#3durHmOI{j1LO<d1aXmI^o
zc`#owJxR<Aixm6GPo5a{-c6}#WoFMOzFu{1r7zAq$Voh+O!Z)xx}7#d6z~guiRusO
zOI~NRbu5DFp`TSubY`0OyC25tXEaC|_IyRwFDrW%DmZEjI$w7Ylln3%N6S*R^}_y)
zFrM&+C8BTU(O%Mia)apKTYc4<iJb)|JPZ=&vOICyE|o|7-xzO9VW_Er$6@2_4JfqS
z|7~<QddZ`Ln|OVXQAzOl)z_CfT0<hb=y~Q`23kpk2VGv1iuex*F=DcNS!~bVeDqn9
zF+S-Qq?G4uS(H;rOuBu6_xrWrXGUVz1(3*-R$tv80JwmAp2u1y3$Nm#T_e`P-CLaT
zPq;`HX0A{ah+1rkUL$MqSm~NFLkOTq`!Z*8Rk7{FI)v%IG4of>i_sEisMyiRcg`A&
z-$0w|uQKq|%+A1&pung|IhG-lBodnnQt7FvV&5)9d;3V=I2)Up$t?Yi?r>z>(t6kQ
zV1n;mPz+g;H?gY8omT|&TayO(3t!=wyjafSF&ss~m~rdh4JlAjcETB)HPUkoLM0Sh
z(<$i<M@u+vbD9`*yn!4!jxChCa6V7kdTG{FDg7+UoG#ePNn+d@5@*vlN1cX>+{+qg
zej#rgSXCK5dbtPIBZek#7ufGbIGt@@vh^8Ko(9)IrWLP?FY!M1zFTz3IC6rBGrt#M
z+J&)RJT7@2N<iaPH#+rWhjlC}Gtc<IxcOqfDD4Hog>(Je-v!n*<L;g9C)dBKz*l`Y
z8hWz&Dq_$wpuE{-ZfkRU!^&vZLSXIVIM+9H-<O)}Q{YBDfD8;<?(-6K8&BP_buL1I
z05@%QWl8nt;}e=k`ETvCsM+UjaxO>w8-8oF5Y`w#N*wkU@-ST8ItL-bd7`Jcu@ib-
zA7TVx$k}$u(d*d0$+7BVn{-P-`?c<K&vagd?M65HhRL+J=UuG}_zTimA%}~PZ(2KE
zvX(|1DO6(L4qC|;Um6H$(zw4$c3#%mO`MqT2E{f`0w$8DK@x_{oi+H+)B==J`Yk)&
z`|`Vs94_Fz`f2miIL|6MP*f=RQ<U}R1rc>+PN>uT787lF9otaXPhKhU)?hFJPw__<
zl4iOZa<KUb?quK1xqTw&LkVYRz?{U@X&E8TTfFtLUY@A{RiTDa_K1yfp!_nESxWO1
zRy1(+=V-j8Lgb#G4~GBO@%G1sQDRZ>B18igo?aa-!WWsbPe#PPX(c=TI7Lph*|&u0
zfx=!%b6aHvPFcVCo^4$Br;V+CSMLlw+k(FK-Zqgmh6<{Gld1P}K@HJYRu0SGf`deg
z0Dn!QN!W)dGnU4Ca$?fNA^mnlqrtt!9xOje>TiJo=yXJDE>YlNJJHV(qkbw$&es%m
zzkK#{V7<A?)ciubRZh)vW*&1^Txr?bzi+6yPE2G!Ix><T=n2IN$Dejxu{KumG{skO
z@R}a3&+%xWA@55og9CF)Hv<R>bJx8z9_MXwZFO0ul1;Cy3Xh9CFP`nx8sO;>Nb7HG
zCc5pUvh!Ie<`LbgU#m?d-%E-N6`I@gxPd~|ErCg%dDkH<{p=i4rhfeQ#`v*FL}{Q%
z3f77L>y%3+jD$N>+#$M3>E)sCyTT^)tT#2UP$Kw<JGob%e{fXV&R6FS$|wTa3W^=i
zt0Y@pxq}tv<dpbG;47%Ud8hRy8w$qY65g6n>=l~kVMTWe94mY({_YlXXgouJd(1)k
z8^bN7vHj)Wg#T_BChZ0eyhpm7Hi4%4)CUQd41A^^!;<pl96NBosK{7&yDr5W8&PCk
zRMYKSdmFVOWM|JzNbwdk&KZ>AhpS>=A25)a?gAj@l|`Xkwb<%u<V|U*Z>Bpv+&m}v
z^7zrD_FaL<6T@o@#(GzmHwbWlWdu@3G|5%8Xt7t@iJ*P_y7abg<==~#ntR}tei<Go
zwfm>}$+Kb0hji$D)U^Xa$S&?{gZBjOnC8dZ`#ouuCRXN)=465`CuDJr7uqk@jOG(c
zsu*iYyd3!U)uK_Tul3{#7s7l}iY;IIs=w4sYu~R`CjvB4Q?7!W?St)|pF8)^o~z%)
zO-*$6ktOADi!SxV%dM1E=k+RaEr^OH$`be3#yarC>e}2h#q&$Liu_six;sKVQ~MEA
zrVS!+W&9Q$<p#L`6XWRRh3@YXwY3(^#zIuT@I>|g%&2IBJQ$T##uaqZ2vwEu7MWBG
zTsrr0_j13mw=ULSdc>(3+kf}@=cxSieOlpi^u&uj>LZMC81;D{sfX!U)m1&7DxBUk
zGV%`N9Y5ej=)6hKoY3+0`}rH6VV2#(6cMr}yDprhHM(43hPfnAAc*eFQ&Wr)ayzXo
zXvt__PBG{`<UvyBjz1f5`~Ez9>>(TJS2E>dw{!f^Ry!HNnT^y))d9PpyrKt8(cJA&
z=sBE?T(?JlH>aOAB<Va%<g#gc7Lw73Tjp3^fz$-U&L}Y^AB!p-i;7p>I}^T={$G0e
zglz@d*)^==+zo@}R_ArW1j0LJ_dF{GpYF|GS(wfwG+8z-qujB+zb#GrvSL_mx})kW
z^0WOcddKa_X#Kz-(h}vUxT1IVYlLLY)uqqr>lxpjl+b#lvKvvo<CY7Z1xqQt2oEoJ
z4rE={<Rde4wF;{yn*}Nm)1|A$&Cvf@Hl_mZN)GQ%7o1F7oF78OH21=8x#B8WTDxC9
zxlF+#c!BERLpWoS%Sd}yXV#6}Q^-EY(xJCiQ(wOUkUP^QGScV%g$5uM`~pgvh#6bx
zaQy2_cTFWG_wj$p8#f#cYQ!KK-!BO!)Q}5w^pl_DO~iPJk;5%xJgX-23d_@Q_3DxH
z$!<WQxOQpTOSWQH05?b<Ig$=sb9xYde|$Bu{rPinOx~aTG4OaW*RT!eN;Gv<BH`@g
zrPV~fmwO5Aor};bmH&>@cH%wL$o`qk055OIcb!^`9c!(orNQ*50^}ngPZNzdSboba
zz8Pt}#b*+D>$gaq`JEdsqHbvq2aHy4mQJviyx1{WlKESfYn7FfK-b`7@!hd%>}b2D
zcj4*Fq%Ur&CRGg-C8agbGbWc`h@RQG%gK;ft=CXrDo>N_elXqwMI0QC%5V<P`i>ba
zb%;+vxa6bQ<84UBV1XuMr;2YiKokmaYltY?UF2On-gwPkz%`!rq2-lfeBDDh;4ayw
zI(sfzTYJbWOSh@t<6TV6IZ(u_SttLbVStwVA&>2U%z<IzcakG*G8?@QXVY*b)S0e?
zxsa4ol#%b+@RSY>4tu|pZ__yk%&Nlrkr3}AL{nclf4cPRN+Xo}HgMSA)aU&jrVUiH
z6pyFa5{ibq@y>d$k+E~E&%}J7#<=tQ@?2fE%N~jS#iJ>bp#lp(;u@T=lSpUiqlxEm
z1Ha#=UUvj34h(n<Eks_J@q+{l>rA4al%K-K+``*YtM3-MAAAL{BHOBq)20Lwr{bUz
z8C;mtR=Pvf(TnBx1^QQ0^Q&+thKo4}T#oo?^>2J0v;F$fEgL@M2R*2C4&!A#(Ex~B
zVk71?+(H?fv#uGenB3g3zfewvf6ugmR@*jW!f3FDs7+bbng7y48iQ)*z28{?`2E?g
zt*Sk9ZogJ}urkO4u>25X`ZaZ91Kj@@pPoBl2$om)(WdEZaJ}s^l^rfLGg~4wB^6hn
z@b>kEAEC^ww25!(Zf8ypHBEh)BullXSIyF#Wh)E>@=zMmBR{k?^){RD3!*&d(OO@5
zvY_TXBhC5eFzU1xqq1dC&N30~#*bK%8ez1mq%@JF-b4l-YuiP|*RE>Gi)T{-KNvKT
zV|tB>9e3$CIpu;^3QR|v!1${%Uc~=kCAXb!6j|z_4PsELX%{?iflYP^2M3Bom-&uY
za_t@E`Dww*lL@SZ{DXpo<>iS1me&LZ8Hg`$fod|V$pFihi<saNHVGu_38)zpk{~2E
z2nm0Y+l^GCOrN02mrw*9ekWixsaj@#O|Y`*hCpm=?7_jb?_iGZtHTiRk=yK%k-SDM
zf0)e`Dh4Fb&+wHwcI)c!$)`uV)1RrMGy$?t?%l0Cf=ii=Lm8=xKZ(6NM7es|v7l!f
z&)e;8d`pngTPzurA5TIG+3Q;emj32GpOzZ~lP_edi^=tUTv!C&ay|}KforkFi?=*I
z$yn+@Mg@n{S|Hq`kRp&APPH6?doC1|P(T2YDLfBg2X5}TB=SZ4U>BX7OmlX+jzCF|
z0;~ztXcN`5UM6*CgZDd%rZFi1g_deWQxiB`D}%X7HO|09n}Jzqi586h`t_?Q*38n;
zI8=kx?I*DW4kp}i3c)in3KWNa>NH;W^Y`D}-DRxIijV&Y`gK5+k<yArIm-jn9lop;
zxJ)2noo-#W6mmPapdpP`HQF16`+T}MH8q`S5>Kd5cE9|^q6)%y`e3A(l7jonKBas?
z^yAC)%+G~I>YSwmZ?=u;ouSkIJ-~l{DN;)c@J6Qy!dhC!C+&^iB<+0E&WBkwRp*sf
zCP+=~2|kIFQ&Z2P1&Feujwjp*6v1B({sFX+l8~5Ko1L_=t!)mVgdgz0pyv&oM1|GO
zBuON?EzVZ}d`lHrNWDYEsDOXb2$~j6;8%3%@@16OXh1-KCOavr_zVAWGJbwand+FA
z3(g}b%dx7qHlNw%VE@QS?@ohw_~?+3n<!yu!#Napb};)5Hq40*_#*)^32X;}z<nQ_
zl0o@@6CSbeF}5-CD}8O#2#0MJO#gm~*JT-_(dFZGcbe!XX0~0Tzpkam7H~80DHG>}
z+EC5ij0@UZfmw3FS`n0DTiD(>K6-I|$(y)%207a(P7O@e6M5~DoX#;<_W<;Ww^jX7
zuJNkY?seWJob|BUJB{+w)yu<R=a5k+eVNa(m=;DuqCkTI2dks3j8FaDrVu!?a$7H4
z#mC2=-7hXFLAl3ivfQ*kT1)~T9y7YuGsZ{n4$KA4PqszthTO0TFn)qnE>6bHmDN?)
zJh-smlD8l0>FaMp=M<(n5MN4QXShUzt1dZ@^@QO0p`^Pr7$u<mV4JNBM~~VOTRQ+b
z=Rr*4;Nak=%c<kk)D+-=nH&Vh|7OMw1$>Ivb=HP*^1I!Y!5;3?oHIPgw4Thr?DCWS
zO5cwBS-GuK8Rbo>FJ+I~tNN^ecXk&@zRWYVsk{P{idsKf&RC%dk7P#ES%S0Kh0l7I
z^h-G}{N;rp7pq`CR2R<<x92f8fCMF5bXqv>Qft7@25|xhh)AR0gNJ4^@bkyeG>RfF
zM@-**G#|xn5sR-cl?!bU7$|lh{9>5NZ~@-?&o`*$X?_K5?(EDI&qrZu0ubX(oLd4L
zAmlVv5R5*lWbjULTM0=?jdnxeogu&SkHl|1t@)v-XssTaKiqh0&KfP6P9RbNeU)4b
zigm`9&3;uas(xz5jyZpGpOnkt&xr26c&OapB|_rh*z7h-=PXOEdK<ad@Wz+FzQ!FF
zI!Rm{q=@g((x#NzRB7n@2etl;<agpcKbaRy`)bTir$CPq93OgPx6pz7Xc}$+&2pf}
z*n{a7syAGcZ}URoh$Sxc{_Ls`tU#X-8c|yYr>^@f1V1@my@QR(GO%SsK}wz4^85B2
zSh=`m;EyWDDOw==oX{H;>WpnDD17{ra1Rx%?zRw(EneBPIjH^BPceq;9^77)q8d>?
zy1-JTQcI0MSr)Z)lfQ>+EMjO}cJCM-J$X_G*4*`NB3|XWeLRi5&w(esvAz8U7(URJ
zH8nkrS!Flwj9rzkluNgYBCpU0jpRHtwJ}n@(9+hH*)+`0&;J-X<3q7)LA6K|_-kn3
zNrsAJu*U1Q_>AFrxcFKNJ%30C(a}NcfFNqr5Wu(*6csgQz?$Y%7Ur5~0FPzSEyCxb
zPP^tfR1@K2L%$Z3tx#)BUR@hhmS^H8p=MI_0>+A3Vm8;gf97$mRERZWkT+h8{2ue+
zZpVe-@ibXFy|#$URS>b!blG&2oIFzXDje-;aGNvK%SrtkWvBa*I<MYJBqe3^ePyK}
zkjj#N{`>~WT@G;Rc<k0%z=^`JH<bt!Dbu<S9UL6c5n{*X*z^u%`G5>j(293I@V>U~
zXxNTMOlTh$7x&rMx9|uO>Pwv+F6YU4o?W$XTb4U#O>qfmox@x-lnW^&|Hg?=9^Hxh
zEI?;MfKj8p-HNM?ah8<c-Z|?N)%Pkhib^8l(i3RWLDO%Gn1tXsVmtg+u8n)OH;rUt
z{+<b*NLuf83W_#ZG~A9mG?ec)&@Z}ufNfJMalF5h!k4++pOv-ET52Pc8OE?Yt`Z2k
zr<ffuEGn<TfELrvt2b{NwDQ4=|0Ik+N<($FPh;Lw^`F@`rjUJY&V?!4$Y&bywZg{6
z<q!T%@NR5ha`Eb7=gH1o;m-pMhD+$-j%`0jqVSU}4Ld6y-nH8^{KmIj)86i@UhZfC
zoKEdGZ~k;|hOoX74x9#v_;MAfIytQ7_d~{DlrDRVP3ICCc#I)7q=MQXb37WsQGwfP
z?uuH8b&5?iurJV?4uyn;<&D<AMMk#50zuQ=EGRMf6e2F7%nq--C`DDkh2r^V3df?}
zbAk%v!IZlrM9$xR;$=wEgQEYgiW)wCYW!H;|5S9PW`yRe6n$-?BxY-z=AR@?`ShQ8
zD0k;Q%tdlI@laXH=2Tr(bX{kqrJbEV_)4Hzua8ASQ<#5vxQYt`fG)fUE%@`5(bB@P
zVibmkvWxij>&1?v?ls=H&i|f0!;lm>)bTEs3tR@%=gUu`1)LAgj@M~vXwU-Asz{jC
zXEF-MQafs@VUAEHmhXWYYdQ4`x%^f5KA3wiEEpW^E_+HO-M@Fw71*+Uw7wpn;AKTs
zqwYc@pDg`eIa%iC)VN+wV72tmW$=CYci;Z}RNv=;4IWrY)8kagD^i`MiF4P0e6G|t
z?A3F!ik8=QgrUCayNCH4$f!mEt)!<j{*q`R?1tCjjOq135UAz-_v+O&v`Nf<z@|vH
z9f8wVd0iD6O&oT671GKDxw(sm2J7Sw=LOG2V3g4=N3{E+b{c_ins6F)Q4L@qpL4vw
zo&+r6z<N&)kDz^q%rtm*S)GeeNBtBoD)KZtXHwH$<3j@Lmu;{Ad-4z!dPyoID@@#L
zk8x6yn9q_hNL^NMVd-nrog3ejuY^laEcyd;MppQST}$Zi_YK0FhG)7Fk&*O-f=SR3
zGHo9yS_l49AP=hr{C(@?K0Lsvyu5IM$*BhdQDofa@1c3Ek2e@z%t!CrTquiq3{1>U
z1NNfQQq(buVvi%xQ3^P-jZcz%ni7aQd*YL8Z_PC3nT|r{^?bPCtpj8*;7g*J5On$f
zpCxJmkG5>1j(06Lg?;FXI&{?uc$ZRfIFq6d48^9#X1T%NJxUbm>>e_>{f?`NJLm$f
z3AKpDOuSAkGedRbus89lO4sBb3veWi3I6aXAT0ao^HNf%Y~%0>9(;GxFl+d^hm&3&
zJF;kkDRi;pe6XTgkm1|+tPx4Ls)`3#^q%ckp7%OML-!bviIpmzsHlOIai4vqe46gg
zLg((9kV?USN;cqqQa2xssL$IQA^Bkz)Ag*NG?q52DbX@Sk(Jm1GzS2LQ3hK;cp1nu
zlyM)U%wMI3VXzanqfzSKuWt#o$s)@5wP7?blQbzbh;?$dDM_Z7jpr=Yi#5l1t}ebz
z@$^bKYt;LSek6<f?}?fD5=gFJ-_5$fd0V|0jeo6l`hfy768|kY$SpGWK2ycog`L$G
z^OImBuzrOOw{QMx>}QVuK5&5tH#jp{!F9po5(;q&X)W{@%w**M{kiR9fc8c{7js9T
z_NVrSavR*)tt@3#p6l&Pi?4o5wxg?XT-kCJFa?BBnL}q`!h~t1LUgLrzT)Wk-H{ln
z=r2J^gDnJ$;_pT(k}3v{-}ub6op9Acp{kEX*OX4l;~I5<&K_G<gwke%<>I4!I{%~M
zc#Rj1mCwR)=o9WF>YAe^Ht8ybZ9|KwlZXhE+q_q^OtQ*MTq3e=pkKP(oz*9K{KhUq
z=kWP<;M7_N@rE3C*%11@cQK{()|Q9|It5Dt<GVg`y^9~b@URt3=1Wt&fXp>F|1b51
zOJ6Q9)}W6Sl}n>tA>;lA(r+kF@-1zqDroqXgo6S)F5!d?Vp0=@vl?1P(*Lr&>2ATT
zR9WHW!9yZ^onFBz%1|J3V%q^(DCg31#}lFo1ev^Xg$Gj%@4ZrWM@{Atc@N#qNjyqh
zw?1WUyNHSY;@$P~f05?u)STun#yUjt=0Xw%#n(T69xDJQUUq(V3@nUm6cn(HeExlH
zXUSvB`P+zseEX>#16fwXdUV_%fKXyAN3EVe?NYQXUjHHO#XP7)E>RuT+6yKd8#4<!
z*_S@z9{zdWoB^@DV(l@`--YwB$WXDosubeO)@nyb2VH||={My;IveO;g=~VPPXbl)
zOqPbq8{0x?>5nS3+;*B((l15^?oe4I`n<8FM8j@=jQi%&t=`i5H@{YW?oGaC$g_C=
zQn_ES5s_E6FJnX-9#>O2KmB*uJ!5wQ;?pWzMrVgX4RUrk)%fT}m{{W0dU;t{liB&C
zC#5|YqGtT*OOQ$!j;_Q8rZs2fCSH9OQ}vNfQSXf!wR4lWZ}J!omJ|C|e)mDG(7d0Q
z;F`;LMt;5EX03qk^!iy0tCidsa>-0F^!K4)o1cHgXJ9dJPoY*3DoHR1N>`t-v9yeZ
z=AN!q=(s>#h~K-qx^7La=ev)?8c^c>)8Qiwr`)}XAfJ<h4Qz??cQ-i`X~5)I`Wn<~
z0i(iiw-v~*=eQ;QYnAUZC*6YnQz`j4*AGz}&)F<^B|DMr55q10J4~*|4gxuni<>(H
zJeAOZrUW+@w2$uYfg~t%S!MMJMra!wSI8)EY(PC_ti1XM#^QCil$iwG8>rnJ%LJpi
z{60LI+U3XFMRTW{5{#<LWVtN#zOfM)m#i-|EFE|Di%`$n&Gq8S5k{e=8$peBqz8sd
z&Z3D1=Lq(l`}Z+{T939?G>z{9o}_5DYa^)g0~$AU{2aomxtDLt0~7Qj^08BC=bM_0
zoe*>Sy1F{Ff#KnBB6{X>>q$-mmF64t-cl4Y|C>x2-Cf$;S`R$>aRKXv*N1z~`=bgK
z{5DIn2&c&a>ITvR9M5Jm1WcsTBYgk`fp%!tl+NJdjRvLbzQ6oKzj-toq&mO6o2oS*
zuUh@=NA;E_mPc2J9&vbgPVu`FK8sR`@77ZohqsD@<{$ihfmf$WG<R>6K`TDP_!)Ev
zql`7t1Ox=2G263n%+U+=Du<C~5Vzi{!X!{cz0BmGMuVxHx#wiZE(G)Yk#)&gil40Z
zY!jw`njcBt`k<iMOw}YklB`-b`Xs?MUYiA7FEjgM6;VQAd79a%(;@LJ;m`B5w-JYj
zEGewP^rPc!@r*z0_7dCUp*|U=z)k!&`MlJ*gIZd4c3GC+5svp<WA<*Kg}+sU%lr-e
zgI$gTXl0<>K+s{pup$L*M>YYLoZokIEAAXxyIQ*C;#pf`GV{`KKC`^$ko}lUDc>vZ
z{5Wz9x!v+<@W#*V!kXE=sl=G<+#fdfPS<_wpPL9+TeArcd|*JsII>Oe-iS-H+cusd
z{gQ(la9i0B`4Q>1#nawddx(KTdEW7SaMn0!BF{SW?kGJ9iT;<#J^MuUW*385c0B8K
zg7hX7gC6NN5lYL*fNym*s1>Z7dB7#pXW?yw((x2EU}FKX!?kZ?l5P*2=zh^Hz~@ZT
zR*w7y)W`daGgwqECk#-Q6!$MGnzW`su4NM1_fyxl+}U=GpnrekF?YUsf{pz#sb|<<
zJRjdV_jxEk*(BZ#<)<p1l~*pbkyNEN&#eERI3^Lid${t@<8+)`SMcwq4I8*!?faJk
z0<C2#x54?mImog&_iF6sLke&G)?7-hj$0FxeH;6Iy&630dbkL(?fHFd!|_V6Z?yhy
zUL9YV_C?ZCs>4lMQx>Qu>a{lr$`t+K-ZQoB;%c&2cj0Qfszj5zi^^CnpRBH*#t_<#
znoQ>7ceLg&eV~gezhBmcE^J@Ro4s;)KutCh31!I_r<Q*@Jg&x~pO_b=JW8(2E)V0u
zw8+glx@U|)y2k^4;acFSiC5waD_=px<yu-&FtJ_I#j58Vgmzi8y4bkS?hJeejCycy
zZRldV96JFV<yRam(%IhZdEfkxG4r^MYQp~lcaWQc<}1h3x8~lp)3#CC@bDkvgdNJ|
z6-G|e1lCxxa0gNbNe}sheSY=xzx+F6$j7VVZA&s8s6ts>_ad78az-Jk2_<ZWEJ1<c
zXMkyr%w3*7St`seCmmb3Amd}9d1!kiLHfmMnHbMM*+AY=kUQkp^gMU=8UdgUEsK|a
z<mVmIu+<#TYZ~*wV~U;FJ9?i+i(kB7*>)Vg)#|l+x9Z1Gw%pRp6-n$W1ZmIZu*CDP
zbQVzt*z+UWM84JTN%$g_>}$nOP&Fywv~*X<B_8Ry3;}01T>w=K$$TkX{^7K_@Pjax
zZ+^yb%V_0Rft%AQ%Mu>h$q`rDJWUvF;Fr>IhY)CDsF5QtyNR{_tL8HqO9-8hIP-^m
z1yuVZ2RR8P7nSoa0aAuxY-&tcInQpGEh38h>8bkMtb->(p3OwpBsWuZnWj3bd*RQ%
z^<FDE_>^t3mzj}~F2VfK;3lBGG)+veiDirvI0QS$1vXlBV+$$X{7OI835rX;0hsIl
z$^5mUoh#bMH8wNrE$<_5O~jy6^6GjV$VEc`C=pvK7ji92;JufxQ^w|-0&)e;wGGzw
zTHMzHCpQ=IqN%Qy@7{z)7`==Cu;ELcAz@`z;h~{Q6{C<^Q?<MR-iw-mt;FDhoKtqX
zFym^qMsu%M3ZdTo^n`l5Q0}kRR|&BVr-j!X6*PR1O3#IihX1(9JkPYX4zB_ZX?^oL
zBD`;Ly17jHdcGsrH@Kq0sYEI0jhWp6`(*n$^S^L)H`57V-07()mQXr%bt9IWO@(mR
zhE~lr3@vt2kyo$ov$5rz6nww`QrQn;w=6vu$E(4O{9L{eUw&1ldt4kUIAyV8D`cyS
zj|^#?sqnD;Ki5QH0L_!zp%{tD=2-8yG&L$<v+5tN=`Zvhjo$n>;m<~{GSF>!eEb+t
zsE{lQHGpLzGw6K|iid8cUbY2M!FM^?&z3E2%hiQE3wf1LQ{8$Qn~U13=BDV5s2A2N
z6_0?kbC!jjpt^s>!AqS6nO;JDsJg;><++6UPli{t2)xI(v4r4E{eT|6pG=}}iVhU*
zAD3N~pIVu|=y+pBQ86U2zu@lzcs63r5LWH9GLS9GNLo`<0~J#=GC>zIRKsKiT_6rv
z_Co#899rEU;K2q8l%z&<r)8N&R*hX^`1n<6es$K~qGncDen9=v{T<JdKXLelB*aoP
zCh72l#4Oqb`HERW%m0SX&hDkceiA4gN6Wn7;qI<<UH#JCbSeYQOVD9-<@nLvaWn}z
z<Ulq}O`+N#7Wi_qM!3wLBsHQu_?Pb+bw52|tj)n#H@hWTrZt)o5c===KY8fw663*L
z_yH~L=lJ*s#-7B0zvd1;HoRP(MgYAuh~C*dI?kG=7if-RMXg^3FKdOO;MW~<54BrY
zgU!^x0fPPDwjIJeM6jYgdg1@g*KT_0ZOGTPx3>@7b-!9eXNtI7k-ANwdBAkpvFPc(
zW1b=l`Z+(g;X@R(Iy~0(NL5vEB_i*N;~W!$VwImz<BzQ?S1K(j@q|zkRIM_9efq|D
zSCHWEWEfnhiu)cSbNskW|B=XLRvq2=7#r*Q=7#7F3b%#F=7epT#Hv`-zieWp#;S>k
ziT$FYd?b=EV<1uSworukP&GR4C6&ehyZ$l#g!j#m?x(anef65I2_#y^JZ;FB=~ovg
z=~`;~7yP?t$5$cie&Aw9c(ToKl6K#Ips(VM5o;MSUP>Ln>EZ(+E_l)De<u^KMTc#(
zp@1V@b&C$3O3r66XYYoI$?B&fJGy<rJ-SSaG4F_kt9@E9<GNy0H=vsO1N5V~d3d@C
z-l{<E0d<&EmNlgE{FE8hle@PvxUfr{c?PWz`3sLNu3|=$CNKo;Z7Cfb+?(my$fc$u
z0X%rw-@@EVd#X#{eAB|L?4x_~{e0op@^${t8vSV8k+|Iw&?~0NE?4e0hH8;IqUBb2
z@+8^^tV?K;;|u>HdqvqqM01ajz6*2h718m?$D3ut<IkcmZ{OPOlBlnA4sx0!o7zjN
zxNl!|KPE@3J9Xb-@9cFO*>QP_43`QBK!Rq9YO!U45?Lv2`@C?^ok#XwU%sG>t4r<H
zGi^*j&>d>Nm=kbIi?f5i?(fh%9B`ZS4WIhzyMdE>u7%1<48g=F1PYfd&?Qx95YDH6
zxK+4VXa}~Yt*5UNCp(1&I*Iz3Gs5rtx@=#+t)bS|h<6-VfE)d=ijm^#1R4|Pw+zp1
zEb}`ugBKme$y&&>!~^=S{E1al9tyG{?GZ^HSM84Ov@s5`YIaGVJQtP^`#nL_;Z-HD
z-+eq*Om^hLY2j3<-PZG?>;zdO*devozoBa;02Kq5*>_glwlY!e-pNnL8pR{Hv&)Kj
z!TD4jIkkbtch^pXbaq>)5zh|u+OB-V$hsOU^>N&e{;69@&T_4HgleU)->K41!L%-J
zfG95tIACi{UySKv9}psVrs$pFMh+~)h^QOb+1sa-#Xki}T|_ewc8LV38|=ySz?bxp
z*IFeNl=^$#q@&i?WVp;HUgxXpdxb0~gOr^mJ$xb!)fo>9#Si_6(aQ(+SFOs9o5n-V
zdtFY3ZA|0S-G4^(;+A<5Q0f5`hjxXO{Sy@EW98GJ<|0RI56`SvQU0a@U#}%ZYc&8a
z5TZ<oz}i=r?YnFCHup#f>c3qN|4+h_xIkHD<fP1?zwka>Fv5An@3vg;NV-pubs)I)
z$P$&3=Va-;4hi&)^V~kqwHu!lhu-LhcJ{Yj<srJhIr>UX0Y4`6J~@5Bto#xq@a4-9
z4U_rA3wswjSKPqaC#vsFuHMpJFBkpMQSQfyG_L!0;S@X>j*qu<sPw6={Wh|1<LsAc
zXtmcgFRhr-lsaCVf>GmIrbdN^hBm=!9CVsaw5?A=ljD;~#yamhxfU9>d+SHb4ENik
zRXpddzBosRE?2hno^4_#RmC53dvm9(_k5XOhO_C4R?kA&TtPr!KKdRM)h4!}c*cjQ
z;iIX38=OkJs1}}f8p+@5JhNN!PEjzQd?e7Egy`Apgsb^Q6>%ke_W4I&OGF|m7n0@_
z3aLl1(9oW^Pgxf4w8oP>5D6aN{BiO;HOT&J2znMxo>3lW#+gibKhu`==lQ_z-|u?W
z#lp>bdrc}|vl>H*fp`=-Kh;-5Dx&g{)-VqLI&CRY(J%NEe96$*l?<M@If@KNph<-5
zlRLb61diLifW2a2VWj|Y<_MY;I=O@FP)vgl5|0)r?>LwbJ8p4+^O+rPXuTll4|XX_
zy=lQN=coHtW9?Hr6Hp>a17t4+8iDEV+<AgW%H9FiJ^6;+iO92txtSRQPy}!U5wo5m
zi#J$~8OovmBNM=PLJxci=r!a)naZk7i;zO2+DWfw3@XB|&52efLsQGglAx+|If@DU
z{fEbvmNq)40uL!4r+odzj`b0XHeER0sL+_E5-n|>%CeZS^7QAF9fkbTXPrakQ$ne@
z9AqiaOaArj>-JVmc)w^gH$3-DJFEG_iKNo<m-33UW#1fxoc0_Ibt<17o+=*=v*#~$
zG}p>Ssl0yu?Pz~J3;N3qK`mto#G}49ltV3~sj;yWBq-WS&ff7GTUl8J`uoShI1K@{
z(I1^N=YAlvxVX5~@xu+SwMyp02-Lly*vNx&LtpC8ing@0{BmM_Ry@}hZ3iKll+#49
zT2*>1=>Sg6bddEhf-+tmg^XW>L=pvZVj&*FGMT=-zdoKN0-`Bh@giS*eHHSJdT~sh
zK<WXP99ihND;~)s2X?o;z{TGI2%q?I^fi0F34YKHsMbwG@^#R^w(4!0bD({tS$ypp
zBoUFeiu`Jd`mVNMtC1ERn`Mwf;X4+y)M^y$7`4(PcDg&r_co;bnPCI&{&-+x{17W%
z&un#CZ@gsqGx0-o)FjT<mJn5n`YjdQUg&cw9C~XO&^RN*GFG><5a(+ImDw<E3+?ad
zW=2M-E{Di0iTlTo$sp+TkOF}@B4yM01pWR=M!l(#`8T&AigyE)-3vv?)+l}z@i6x9
z&9hywxNN4qxt5@4&A5A70<*pgi_{pv_f+YGFD9Jf5MX`)Y^c!AzaK&2a-ikz=}F>L
z2AA$v;6CxKPP-Fec7S~l_m9X&8ZIscXz}ZSTQ80cVX{VL1)%ep^qED^0zrOz&_o^x
zio}e{@x1G0G613Uflx)8(L>`06a@CQwY5Z)l+3j2S!jp!*6qEi)ux3cs+lT<nMuv)
zns=CzUNuv-_ICNsH7%Npx(G-onWsFBAzY$Pm>2cT2`FU};+;b>`h6UhdnbLOmiB8h
z*p35NW=-Z@@xr!O=9@$tzlr&UVZCa77BA}J*GgMA9B)I*LsqKLO&OK!7|hWbvri#w
z9{4k-cpEkC@1?y2a}@k!i%p#h>o*B9(<el20<A48R|*)E($S-6vJ^{fSLvX%(i+10
z9E{VYUUv+>MSADjBI!jOSF#(J2CnDRfbv^HqFAW(g3KOx-G@E4N<29^$usJub9X=8
zt|$a{hjO;Idl)C<X`&Sha01~f*$p%J&{3;){v(j>+G*t)H>mqURI*mG{Dngvy0@yo
zj02LA1_S}DLLiu=0E$V<WvcpPJIa3QGxDq-Oqf%V^RuJX!toxkaBzl&ODgE{PGnuZ
zcw&>HED{t*ic}+laP>0}!2To+HMLjUR6X?s_p*r*mX{J5B6j_LdIiDi#&n|!FG<NB
z;-xT_4{)y^y<8IcQK5X?s_kskUMiHi!yjBpb^c?KK(YnbE-Uq(^R4^N9f<T%Wl2{n
zM|`P~h$Kfqg&}qM$cxk)F__4D&&<;-dM!y+CH@P4r&m;j@2*I~Qu(pj*$=sP(rs;E
zVD#Si<+~>$w3p<3yju8B1aMiilzPfXdSc%hWl0#Dmbv)e9c<**jJ|`V`<Q;6K4m`2
zDf9O-6uOv2hH~{Kk?aQLpLfi*OP4cC*V4`V$-(N{Pj=^xDIu&cY&=p~^TQV5*7+?c
z{gS9wtQHLO!GYlx!e#1Fa9Saesuw6=5=ydhG0aO~VCp_zGLyK`Lm}+Nj-k<`$;L*@
z$7&=I`peA5?B}gYe!VY`%-$Hv)|Nf?()R0l{48uhbV&|TO4$@SG97r?Br3stN)E>g
z8k)VT=IgOD+Q)+_roMsxfn35*I_k##A0pd%CoEhVs;UuCGm8PVik^q3MD`7rQBU&k
zWiGQZ29c7fFZ68=YrP5@YN?MOrU-kJ3aL|GyLMNksilQsPcb_?+b=%cv?S%14Pf=+
zlKi^?Lr9)s+{A<?93@H&L9T&akR#A+hRnJ*V2*vj#BLfN*YHD~DhcuNGvczpx;B9*
zLcFu_>C>mz0I;Fv0&}3;K35(oevPvkIA+AuOTaMpSuUgNc56HhvU*|_*XLY1Q|6gO
z9dFRK(7x?DJbznc^u8CDZ|_S-PUWTR$dCBi%SXGjW?$?mP-CBnO;+VMLxKXPOGii>
zz8lLNrMTrBVbjsXA2@6ACqTIW^5sj{IC7qGgZfPADWCI^wMZ8*VPuxF8Hrzg@294s
z5<Q|eXB6#CP0+aYmXoC1J}tTT7m43E<3tNy&zI@hU=4Y&EZJ$}T{i$~6w>tjH_gCr
zMr6D25!Z)Dq`vzTcsMv+kSHXfV_u}8r4{My?1b%bqQpC%p;|0G?sA+V0E&68%Mcn%
zubVYptU6GY2tFed<Xxxq<@y#YdcbpaiJqT?(R(2F;f=ZE-Kw^n3&i~OewPL5shgSb
z(wWp;)+kkl*hCz8lu2`trz-QAQo+Z^w5boBYtd!-nAg0vY%kBsXV%^=ZB2Y%g%<j&
zK0G)leZ{0$V*MV)O~n>LF-){3*=GE5cH<^R^1&Uiuw7F1C5tILCaFw*nYX1H8mVYq
zYE=Fral+nRC{I2}F<$xdZGBQ@QmFni)|IZxZ?zITgI(I#C}e^NxF}l;{Zzk#3bwL2
zj<%~q39N7Gf5VGM4HY6*D$tukTlJu$adu`V;?W5wj&IgP;FhDoOrbir!&W@x#g_6_
z!}-Sju+1a|Pi5ia#porC7GUN1W`F5dTb%UtofH*Tg`P7Nf6Wx9cs7Xy!-lefdjyS*
zImXin0sXf8^q5qd_4b9Z1YRRizdmQXv)&riT6pi3BqT<F<te<h5F{8tR+!N<Z4!AN
z^<Z&J9g|;9`SWZci!+p~9*6)_uOH6lo^mHUxlGl;>1{ZHMg8)$>%Le~JVMUi`L?T&
zLg?~;00Lu@bprYTHv7dxsF1|Gh+KIMBcEAbR=H|FpQXu}!*<#Jp&_LBU}*RaLTDy@
zf<ecf)zMM~0=|tWN`dGx5yQLl9%Z=BslPMoUOT7Gco&+>eCJv}H!iizwThRRln%K~
zslK2j!WJ3TU^5w2`DZ9O^UM03RZh`C&9gn@<NQ>?xigf7f_YkqkB{l0nq1scN<*(*
z(9^&=&^t)jTOD4WsCFAFcPa#`k(Edn=-bFqZSSrO%J}V6V^h-bmv2Np0x{KY00^6O
z*8EDAQqvxY2%c=XfqJ8Qy4BXYKU9*x7ghpL9!1;c)x<Cn7!OUo%OEe{*Q(bII`{eS
z4lNm#a*!9W<6rCPB?4<J1GJ0+bT6Yd6(6j6ZZqXs<T{)#Zci6d@|^Q!j7Ct8NT%LS
z-uXzV8zE|C#_(rY4XSp_mwJ=C6)lE|4O&~I9x4a;TkH4-NwlfNbMJh%o6+Aiz1;~X
zj))&<U}OP12zKYlPM&hUQ3{+nIm5^*q;8%tVCvxJhe_Qj0|SbnE0Q5L$LPJWVP3w~
zMDG_5aJ&)ZVp!pMkIC@K`Rre3fpwq~BN>m)(gWIc*3q9@ZV^et3=p|RQIZKi@K(S6
zp60jr(#}?t%~I!w!*5+%47=({m*m{rU&*4Td!?7zUZLwHME^(Yh(vwt?pN4*q7g|)
zy!a%s>$)_d?BUdUDh_IRG%ZV%$@633aY^gtVk+~MN$EY)OYiz0C}GsJIr}vyU}ldS
z9|qiQ%1$uu_@-~;u>O=JYr(1rmdyi^=yzM6A1tnwuFF4p@+7o%AA)iUN&uwNr5ut<
z`@XkuAazbuBW@g8H5U#K0v44!wH*M}0!^R0;!%7`KrUn23IQSpkt&$Ir2)t42M{gk
zsHhS<vOzthXJay9eOQ&kekaWmT0R-o%M_uC+zE6*8(@D(;uFMR&>3v4nU-P(KS+8-
z6&-ubN7yoLZAL2MS8KBu^5^AQ{?Hpms2q3C8D$ev^Yhl%&Df2tpT9kswMqE3&bM3?
zvBFNS$fZ*0y`CX-emk8^rNp%29BXv<_p*BGi3teX>bnO+T=_4wjpX|ESQwlk5I>E&
z;y<_O<-!Y2K&8uo`F#1(H2jGR@qc8twYk}u6z!4<vBGGf;|CZ&1|jgfOFb!Z*cA3p
zphO~t%AI|H$ceiCaDX&izCq_Rz-&^LauXrY#6l9Y_`{Q8YpQ(nif{S^>eZW;u5zx9
z6z6ulMO|wkCsO%IMvB%(sQOT>!H(?v^%aLz0#SNKg`CGks`pPKw*T|HZjumvU&B?+
zk5GCS;X|-4qGrQh$SrY|{cJh^XzEl-d26`;%jqOm-Pwlg3C1&`tIBG{gN99_{7k|0
zW>~MidAvDhHaC|BrY|xpxD<fu74h)^&`zVa3xLc=Vtgg7Hx-1lq*08NAH!Q*N@<Y)
zXU|u3l%di))p^Agn~G!qEC`fg2%BybmaqQ<P%z-%`h9=?N>jU>r&Lk7s~ud)zI^^n
z%Qh0YMarsSJE(gEIml=v&J!%zIJxpN++{D*jMpYf?OB~A>Ydv&8kGXQ<qi@<3AIW%
zX-Xm=u2rvj|H&q3ZcKjf-5H?QSUOz{x9RTaOzlt37i%L+pcqd`??LR{h2-T5D+-qy
zv&WZeLsgp9id3Rj#>!RTko2q5rtPUfpLJ8Ax6jSVVFyB^D}sIT;!@{JN;x<@;XN`z
zkwnVS&>{TlsN-@5?8pFDx63ckPU1wO3zRt*Neo?F*||k|UV6FXh6VRODHmDi6dF^_
zlDVPedZMcl9D$H@1r?C2pOiU}?wj84jQw+gY0BrpQ(nlNd}GtUk!@M5+SDZG-z)f9
zZ(C?EsEGx1^w?p~;N4<-VYAfpJu53KCpR}A2xFrb4dj0Da8GSC;dCmNCalXKDuYcS
zB`oX_Uq-$U%*q`gaiUfp<i8O<6#S{e-Uw`n6<epzK|z>B>^%l@3i;ZkqS(1O&xWoK
zf1}F~kymULHL}zlzDK~S@~(Gsk3L>Cc*lYE*o}JrzeiI#TdXP5YUQF_f0j@cy%?20
zhTF5tc3u<KJiDMemkRrLYkS+jQAhg8llPg$bEIH{x;SK<)eBU`xm7j`e8d#149b5@
zMSK8;y1Pe?`m#NEFq5u+Ggq(GQlAHn=zC*4eRKuw`qnSZu8J1FNTiFY{Cj_Y(*6xh
zoq~vCY@7b`Qekwhrw&goMQUf@;f73UYk!2Ol%!d@Ej{F={JYarVLuKPS;Pr{z4-&u
z!p4S%mvHQLReeA(caPUAje%N~KE~KEf@9wg^#gMdLT(eH=!OmK+6+ZY8`OiyM&f(-
zY-ito7%t<eSDIHG;%rHvBYHka_i!$`HHOK^W?FZipYIOCe{N&QYbr|roi%~Q`Dtm3
zqvWJnQRbCzoC&GQ-O5Y*BaC%v1gNJ%b2At%LGn>*sBk#H!uNa*1ni#uaU{#-^r-QL
zSB+(B#`}GsUt$GgVq$vn)f?qEHBg&{LG&wnME)Kp{rG<Wd_Mnwjo*de1`7-@lNS;>
zn~VCe&wgXq_98Ts=}yzOK25VDNuxHcHD&BM_x|CK_3FmntM<3SwuvO<J3|p&S=2~8
zHkU{?ZR&OMgkb@SyHKZavLpWUbXCtsjCiIjPyIhdFgJ3Aw_5T<u8CpAxIQ`zv2J~3
zm#pls9Ckh@Ht6!0>C{s73%=hHw&-Mqbl7vAfQN4-gH@SA87r2~4)B)FHf$WyEWJ`7
z6U50<6B6_}5^G$R5M7pzH<C|596k|FQV&KtOl~Filzmf>_FCp$I#O5<;^2-+r@;90
zB=4`srLn)aOc;I3-;<}od&+T0@b*;K+iyR@Z%-$}Mhwa)9LvKJYLq3yS{(OYaY#nu
zcajRM%hkCQ%HFQfNO_}zPpr(kLZT^WJx;vDwqn$4B2(G3e;M`R2wF_~)eQ_9JB<6T
zz4|@kOWQqnB6-YJ8(&J$K4KL=(|U^I^+h&Nm_syAdtbH1rDDn6&djE)w6II!0liEA
zwS;==#`f{p4bni_I3YDC>hPKS{&{G$CXWV;G3qG$*qHvcN=O|Em$^k8K@*ct?mAlB
zqfprt7aO@$4Wzs!2`7b=$@oQ#rIqEw*<AvC&)0uG;PsdT!-O-Gu=p*>(3$^!CK_RT
zad6k)SwTZ<;wY5gpnCAYJ{AAp|9*10m<R=zIQ;KVp7^oHlo!&YeRU(o`}aGVuP>DS
z-;c=u{cO5BfrWN+-0L{DOzl}Q)Yu+~$fxx-2UdlX;lHf5s_t)BAQtorrbm7K``m0-
zl7F9a<;(m3&mSqu(9BsOc>O1);P@;DE(xw_A8C`Nlf?w6%oP3v474ks5cs=Ht{?d=
zlzHD~+nyCmr8Hb`FJUYaO(xwnio-9|_>mlBQ!ab=&v&uh!%#<~dA|46<aG1M8dU{i
zN}^wHDnb8$h<od(D8KK26jbzWfkl_n-2;*;DBayCDML35@u-MMOE-vg4n4ro-90oz
zcXz{m!0+dMfA^2?{jR&#@3$^XT(fk}bI#f4?7d(6wf8xXqKX-MA=ELITpV>u*ir1!
z%u!O7mz+mx=Y6Z*vr3KYZ|g_bttsin<W|N%J-62%R-hkI11eH~3?h7AtQ^5_+ogO+
zPq7Mwf&uFb#L5@TEBo&ZJpQ|~GdIm^*+vW>W3+!^W|kV*bbZ4*Vey<n#)5HDHQBn1
zQR0(F|CiPNC5u#+r;>bezv151>HeXa_%+ryZ9me1GMB&W;c+PyKY`Ec@K0njB(&al
z=c^C!wrF4Rd8V3CDw$OOI;~K+c3vUtUj-z<1DD--WM-sl>~LH&v+x6qMRON|du14!
za>exvdT+Tli*&5yvr;tbW+!9#2Z6E_SckU(^EzcnRo$)0kit{(qu~gu-1B`W9M!)f
zXCMNlu>E@iOh;;zQiJtx8%pK-w;ck*C@$<@JHo+}!UAe9Zw8$q|Bo|ft>-L)T+z}|
z7TpZR<I&ZZ2^sV5Sg>Dd{ozjbUuI01DfM`rjjGKO!3e$=WZWoB6D-6+NYU|Vw~sfr
z2e19=X#1fW$&o&wVa+=n>^fBA$H=>>jP{}fjvi(Kmf=*XBvMQRTN=7JzKa4?b)-Q(
zsa-D3qTaIJO1!ft*7O}$<-bBhDMKZD^+<68*Zha?ddR*FiNp4`9U(yo7wJr2Tt&mN
z9FaQM-pM&q45i>z_eHu03}t~km8oH|vDx7DYF3d7(DL$gic>4ubP-vTlap7Ewru%G
zY_sNpo_ktDE5lfR_t3|Qx|rr2oeTq8qj3`x9K=Y9>792%=^fOxJ@>@hyB3bH&OH;K
z|LUOj9-tSa$Gcbi+LmSe@~RU9Wr^r8WOxJoWB;1uPUTxHv05!j@Ul=}fB$_)!c@Fp
z1j`%G+8;-PZoD%jv57*DbLv?nhijltIj@Tq(9eXI-Dzy+)ABPzfqJW+9_LP>&*3-r
z+qc>`UAfs_{_bq%=7{a0kto#6)y=zup0|GeAmL$kwXw!IlWyKW=&&{GR?8y#m$iww
zm62sR>+9e0zb7WQcD9EkCMKS{xv(k(N3(8UzTxC7npUa?G3&yrlXy0Yft;Mr{uVCb
z6flD^{VtqXxd)lgEMt2VG(X<I*KDpn+4c2i54`>Gq4psL*$YiUVN~pJ@T^6D-hMxh
zeJ9O^wAyFolq^W*Liw#5%<|St`1+eehiQ2(_|*c6r|+r-;&_IBvncYePTc@8jCTC%
zt(Sj@IFNVllI)K$1pOmGP5!Zsex)b$!9O^POJ`@W>bEx?%YOo`@)vK=hY+z{dYnYL
zr}*4UIl15T#>MClmj~RCBl9X&#OU#-xxcz%ZF!?$tM;E$h?~)|h?diF*FE9!&_lb#
zUm=t-d@ln=sYl~h`A>}@o0E$aDi3cu@oygn#<8aR_h+l=tRm%WsUr(HiVO=h7tFX_
zktEK+f-QLyaoQ3gC}{ZT<dVpEEZ0@Z-3o(=Sjm{km&^k6B@a8!;J%QpC?)WouB>8i
z4G?25TcHNiFOy>{oWc*wUA+HF(Mt@JjnQ(}A^RBbfl`&ALB!?J9$Zo{o(ag6WwF=(
zg8GxhbBNk=nn;5_ozcyBanp?8DQEz~3R-_X2|xlF%@cN!5td_r9FRxLBbiFJ723X8
zFEY6wuVy=qmp;hXtkhw+X~jkWO1Vrt$HSu{%iTW@N<O!t_;(aobu?x-pD-}~bppWu
z^7;Rdk6`~dKmQKs|LY*C(JUC~otp*xuSeC&<baOs?~f<N|L_s`AHVT`_6vXK|9`fR
z8&Ui>T>$@+fc~3y|GS#q%>VyzAJ_B$@45i~hyC&2wfo=I?0WwHr~CMS>jL;!Q+5v=
zF%!b#;&Q;F3Q#a%I#%tD1aBAw4^O6n?|07xFINY1XVD0$n`8v|$e>KNn;hCEc+O3D
zn3js_Gx&B0%`)pGFbGZo(~%uu5Sk<me)0NeJLkdk?HFkqhmA3nD%gP{NQ>Z-k+G1I
z1LUI{j=4HW)w!0SVKA?D0Or(IKyG^9vHWjKjOo$`1D-roRPTEP_WZA(Eg8-L2DlxD
z9Bt5&EHFBQf$zehlHfG%{M9hCZCmo6X4PPiy5iso3Tg?KNGVfOQ{BA32nYjX!l~^}
zrEGf9bRi0W9_Um5rwJi$+Py6=u)awoL^)fP8SBoSxFdDsWFGjgJPIqja&4@uH=JDD
zM3LHPmZS-5zN?d%-7@VlZ4<W>o!PT_MYU~M&)#&1<D%Ns0tj3(I{+PPKAqii!w89K
zVP~B$R6a+~aM51qD`dAbrviHS8>iWAK%QOa876$va`<>>j#>hI(Ol*94x?Hgc*a=J
z0d2s+!4Vb}wH$m}Sr-$%0K$lP5~L2Vqm|m%-FnL+7v!C==Ngr|vs_}9d4;xiRXTUu
zF|`~^5mIk?>wE5tD}q1_Oz2jE|NJZW&el~=$;VDDb8ceSr<l0fUV5%zlny|^qKNCD
z{0Fh6JV<F&={)-{X*j?4(yd#wE$0Gn+iYrXzLE6w*L%JWx?oKh$fY`Uj~f^m7;(og
zEG;c%DzfsC3#mEdr4P`D1_eo@SsF2HNwIUss+FV=<_Z&SN|Igeil}XCh<{d|$HsW(
zJ7ou}NOh<z5Sjg3PDDe1yq}|}_~XUhrp?jp;}k01cv+PofBCi30~EsDpW>URpaA8>
zjKGEVHv3MybtdES&(yZYy_siAA@jC;CxQ=5F1l;?FhK;wrM-Mn(=bDK9A9V79hp6R
zmz<nD`k3@?MlXYTEjXKfkdcIf?_~!$w@{D>PyyEQi@`Iy*Ga!65(JnAe(=vf^56t%
zrK&SXE-!Bux-B)QlL^cmwhfyfI`S^4mQLnfU7Tm5V}z9M)6;(-<6Iq}UmYy=(Itlk
z!AdObf*I`-Z=geEkhojb)Bn^G=xLpw8y@!V%`CSm!7RoQr*3Gs1z;ZUqK~d9D9CbS
zRK#GjyrcJ(%Lje_$u!RZURDbb(N337ubD+O^~^5#PRo>Id#`iH=3J^q|FmT>)fgSc
z5@fzOsX7gbj1H#;DYTOl)e*$-FT?8Ev#ZWdX(MA}$q?|!Q3q=47oK0LHdZ%=09^R;
zPCJfMB$LJ;@RZJH0L<rCMqyfxILv3<4RBH6$0iS!P=r^MILI`>d;|Yc%V?5o;Vdc^
zy8P}jQKFXZv_7cPBLGjj$m<uQBnw8>GENH4oGmbN?Qa}~1;^(;@%z97vC7bOFw1uQ
z{8Kt)wx6?jzW3=xrE6lfMwxn11j}$^a08@xN(IeA3hp$}{t>+Odg$Eq$jLxoe;<4k
zm8&@}nKiv;jR`ytcFId_VwhnvsYx`2^ww0(R9-HNjd{NoP8JWbYq1opvW`x>WY=>I
zA{~IkYbbz@)N&A^#+4KC?p#TPh}wV7?u+MV;P}1GkjSsar2-{KqR;qMt#UQ9?@e92
z6_ZUE1c0r|3;oDPH+%k3nlQ7OL%H^KlwJ&;aR=)|L|LnkDP~xvf6Pg=5gqquOLn*R
z{=~L8=wQC>`^ahUBucuXM8hDra!Iu$Pg6sR$D*%++)a7=X|q&YWk~_dt-qx(rYlBh
zTeE5s=^8v^SuOAPfz_c#+s<kQ4vLBu0C;5v@L8*V;@*^hU{Gd(M%AMXX}_X)ox?F`
z_J1{jHL*X~66!ENFEKfmQ-EHb)LGO&EXX5#mP3AT>i25hyHjYL@I|Sy1)-W|wc(gK
zY<lu&!M=RICfhX|NdD0zgixCtJDvUE0jAvM8-l{3B2*o{{&ENn*#H|e(KQp8Dz&GI
zv=}Y?0P^laO9kW}$4SuPJof_0+mB2;E^nVY_UWrU_^0=~N0nq{xOBku$0}d}w_>A3
z@8CR))n29rJF>Go_v-8fzbRgFFd%?vIS@}q%I&h`@tUdl8eU9=X*N26U21NVT=TBV
zP2XR9f7j<s3j(1X3n2|Nua|(x*_NiSA*ZK3A3{=S`I_j8!vyWmcq}u9R2a?nE^m(&
zTh-8ZFb-G8U8Pg4)J(E49uV|6ZBT`#!10+vLWghiF&Ng^xW-uE0+(=^bdy*`Vk#5V
zTvmyOaAio?XST3O5RFi2#>&PGPSk+)01bD^vhsj)KIhr*rDU%&M1yCbDyg=aEt$B_
za%3S{Yg=W`786K0_P!>ooB9R}3>H#cTP#+Lqt`kt8+<V3V9ZPC6SBo2^+;3(i^P~a
zRrmAMa93myNjQ~*f7a%q`bp4a-hjbhQJ)iIvVIW`>F*C;v(A9Y{~k9`;TIHj0~ABX
z4G8OYQVb<_dksK8#upK_e5ii2AAjqEC`(@aH;O@`y|vB3HRq9QAQgS<)as$S=#%Bd
ztC58=K~>Q}kjE!^GX4K-0|pd7r$3TP^~kH{xxU(}o>!Jo43G>LP|z$_+uOgGh>V*(
zR6l!2h6KbBEq@1y`y3n~g@YAJ(a~E(3Y*l%Ebh%A=`X5-&y)SPRl)%aB5rx>FEtUQ
zfp%uN@P>u{$_sD9!ySdC#V!H$zx+&wILH{h$OOPtmc$jSqg`gpPr07nAmm~D>QZ|3
z#<)vXv27Rl(TvYg+0inudZMxl7@X@w7igsy77kXaNPW0-&oy88X7cjE*R))JuApm%
zEQ;aOeG0Q{9LQHTsLbDIHCk9)ogvOs8d-VVz=9<=#it&^<+uQi@9bpQImvWH>S~rs
zhU~9f@VFLp|9x1&^}|AHn*Q*RmC1$JZ}kA5o%C&ex~~AebsG07BV^Whua(hpOM1U&
z9l24<KPr~8gd|%#`*Yu1X02<B74&r3@HtFPxQr#n{W0Tr7!2RgZ_oc5R(-SwL(lG&
z2*wJt`C8)eFpB})0b^)iT?4)V?eNN=l+;!K^M?_wJ;ON~tU~biPf7SmW4lK6$vD?L
zP!q7Bl2vo(*Ag4(XLasw>tvc_+oSE8EYj<P{MkPlxGMiA{O|HmQch!8>skpq*Zi1r
zu6kBf=+L!K?l2xy13j5ZZg=48;bR#!9w^S`f5B-KgVruleT9h`bJ72Bb*!*FlHn-t
zB>-IPo_1XOsUZ^nyof}P)GHR9OhCs+!YgnRoK^wjaK=fx9(1RU5~={t^B_Kop%4+S
zl;^#XAloxND}0RKZhMf}gSzc2rIxe+2y9pRqRb3_?H4|&`mK6-(KkXBgpJb<X;NQ&
zub$K@8B7UoK;_6<u&Qn8peQMdw>NBHJA3f`KDjq*8!Bu2hgl9l)>D!k&7?XG316w6
z-kCuziICA2v4%XPsFjd&qVTO-m0ALS2XF@rYnyD<6468>C4NGQh_`QY&!sxJt@@gu
z-^W6zJ$T$#63@39h_13XLY8twOY7#8o^K=6_lNS&Aa1pfW6syMV)llo#$eOGXZx>A
zM|;bRWpn_*g?19Kcmv5)>F`7Sm2~neQ0vU!U0xD-%pF|++P;pL6qitdil$|K_|C}1
z_DlXqU=_TSUd?+V8T_Txk_`-`rfFZXr)S_$#8mY|bo@2ucR}vtw{j<#y(4>?wEFZZ
znx)Aa49m0Ba$^u3H<QA^1u(!EU;7S-P}h#uQ9Zy#ZVA$9O<n)G!0rzPC93q(Z5?oo
zIzu^`Zt7S%;!xu>q2o*1F5OaXF}Zj}<5HPw<04_tTc`pXg8%~OWLSym#s)`6YAAi?
z>B1ID^9l0ZJ-Ik}Tg?gz5fNp$CgfnFS9aKb^*3l9tRnAU`(F&}+c!J|u2$(vbemE!
z^ej*q0pb%qAV(*iC%s2G_Qzwc&FWXH<4N)OSxmrsc{yyk3g@7!zR5NYUCWxtOe;n>
zItmJxq3gsf9IBT2rzZiXT{T|1qgb*ar3kA)A3OdKI$bdtwY7aPS?$)L>VB~#jCFFh
z<`WCtU2coD(l>Z>Ee|$2aE<9GL1Dn%_u_RgZDXi1O;g1u`~LXAP`SB$Nv}90_+k>0
zQoRG0$1V6vK5Yr<oSpl@+<Gypc53BMV!F^(w{`8ekMHYn&)URGON3J$<m##MKmt@-
zGFG$5US#;kk!;AxqR7v1mb1I~?O(qF-d!ZPzq=|o{f&0p7P{W3#7CP0we0ZKUY|QB
zXQxo{wZ#gdimx%?Yv9UiRD3i>$3m<n?HsoTIyaE}W5rusc5Sv7+q(d&OjX7z@+PrE
z_v&UEK{BW6p^Kd=zZ2o;4Lh$*OO*5?Q1UT@=tV{pRE4Y1z~6Xd1jv25hqnSAJ{l}&
zeD8j-o*`vBS7G4e1Kp{;q93<Zn``~Y9kvl(kt2Fg6zKD4TWzEF$w7vaQZm0&zgG6p
zYOadeL3hqpcvwQ7Tmr!gSRKabctmypTh%Z!G%<TS_XBp+PL{O;$9oifc?sgV;V;dP
z5A+coT>5B=t|>qq&>?qK>_C&l(@M--GGW)N@YeL$8Hs_B@l)vHz)2>CQ!V_~t>WYE
zSV)S)o-kN%EFEHLIVRq*RsEqv_#4E1N3m*WveQq^ns>&Y4`q5h*Iqh~Dp7sRTy?ex
ze!T`u>wI4_sQWbS3c?X99dmjnAak-2<fm+NVne~J0eCi7runa%3%a}9bLXEPyKaGc
z_z)H3X*HN11(xs04cA8MDoBbpf5;-Z9^LeP(A-XcqE%8n*r!cwGv_cp{bDv)0bS5f
zBqn{DH_D>}$I#QOIz8awvp#nwRMot!++W}#m6BGj%9rmVagMLg9nw$a&yXlauTSc*
z01`nLThA9Rv#qQ?@GN$i^4adDSeg#0vmPzy%;sHi$4bMt+8GDo?{qeVh|&=zAU(`?
z!*?d!5Tl2ipzI^UQ_seV<q>1lTc%@aP+Y+wcG1HL?JA)SBu7TR8<;qQSsmhpr0lrE
z=s0R?8Vv;c^Ge24;8{$uGC(q|+=0E%!qC^bmV;3%fNl`l=c;x{=U5ioW6rUXTwP_z
z4qDu!H@N#kH{oj-a6SKbppJmJF`NpNo3~=no^v{ZMyIiRja#a%t(~BXZuw!yD&%@0
z{@7CM;Oo>}mk~>21)cd*O9^U#|9i0;4q7$XZy7)egmrW%f)WuCdoe$n_QG+<(H4p<
z@5!~S<yCJo0FER=e>`br>C!ZDQP_{^4_?oxX2S;aGQ<2I<)RBZJaHrRlssZ-B`OLO
zJfFee3J$0othKJ1sZL}zN~Os(kedxjw7afP^wlVcNc3-+4OQqozgfxIm(y4P1`gko
z#9>Z^1BwBx_M1XFUF|p{S~$L)do<)q_7;`Vz4P$!#m0#Q%WkX73=-&aCprLh`Pm4r
zz|t&hF1qLXNgoYwWJ0U72a8n2W^%b20kTP=jug(m&r2nfLnTFjP-WWEYdwmizbY7a
z=jz@l@Siq>W}UltJpgw1?9AfMTQ*(@I$y7AEt^}F4hBXKZ*W}B$fWD8P|T(^9cb2>
z*M^lYR#5_ca>BO0Y0TOwI%0uz^vpUXS;`76Gy#wdSGAK<f~_r4FpaBp83mtGh4;;T
z4}DOq{sMfb5*ay{eD3NnW1srdy1I~*d2`obL`%BK=ZhWSPLqjY>xykAO#YXqTCTPs
zzxTYk7CP!x9d%Y+#y2^zN{u^q++g92qk}+x$k;#ip5N@*C`Crgu}{uoqy%Lz=kPpy
z1eUI)@S6gZ=cDcgThqfrT-ena`N2_OJXk&86=<r92>hYr6mb2^RGoTNdZmxPCHUsx
zA*mc7tvI7W1;(HHe~4E*OcE65G$B%gJxzsM>Xmsen>AzjPC3078U+@}>Zqf1w=UyJ
zR60V)<6I%S3m>~1s@8V)^DI!w2WyyDhec+pR4P=T2ijHe=gld)0&4WwY+UlF)RLuU
z;MGaDl7vZsMjt9iTVouXtRHM+c@R><50H$-#xY4i6_F9??To`vwcP8i_Sr=L&sf=0
z=+YZ2*6<~g>)9#axl{S&E2)5uFrwx%O?7L8ULuQ(Bn1rynzEmNy17eV<A=fqv67AH
zSAs$-2JXiVaxvVerVi%)W$*XmBxO{Vg0K3j4P2@|dY+#Trt4aK5c|&92tfOM{H2?k
z6MI+7oO35jw3x|`-%L6==((RC{}VK2B1d?7=z0pPOT?<PT+0e~&<Fti2uaHNgf^+T
zEGqFH4kJ6GR$@;q+zOAS|Bn-T&`2gW&S@gzU@cQpJ?fGVCJnj_$<FQ}9XpTVA9MFZ
zQes}!t*RhOSHIuwpb8BHQc}%UW|GJ3XH&@H5#r=|lfU6sm6`#Rl>19{BD)Cn!E%?P
zLU*=ypK;h>@3}+4slyiLU!^ZP7mw@T;LRezTOA_j@=tM+CT5;0RH@k%oK~$p>qxi=
zNfBbV5L6+$*&O~lL|blOAAH`(o2;1eV!sA7;Cv2%F<c#cqbj~#_UH$s#KeFwE6at{
zLu?Yo9@@k{e7QH7b%GVRW(lO;IN1Z+E#am`nNa!Pm*WQ<lLJ}NIUjP21!E%ecrG=Q
zd$=u8CN&pHU1vvTC$4h~O5J1J_5_zv-0P!^5=0Zl!Mf{?(L`dRFZjA=Y*)_WUHAL@
zYl>L`-Vc*xfMn>^UfJfx_)?rfth*amz81dwZif{NHp6iZg+UGAd#^H(Ey$N`0!Fe&
zYOU0!4%(Hv>V=9O*yJibrkoD0dufu0rOzLSSsoN?3Oa9f!1>|px!&l+-zs83rl>xH
zib`YU%w`uWha9Svn7~vVpMb?QOPzyuKy7My?jROYC^D{iFe?Kt6n-qCqs6d4AfA=g
zli|kH8zs4=vS9cwSl3G&06Rw1DJjceOPNc~H?;;49+MOeIucN|sw|J{)Ee|UN`|+*
zyke<qN3+De1#3y7r&~2LSdL-9SZOInL6c`6TX3Y*D0n6>48i9hJf(XIoGe~G0IS7d
zEi99#r^J?%YI4Nq?Q3KCCU7R>qBPxM#K#hIg-)8#YoD9#J&|`AK-9iqyYkg(YIyPD
zUN?Ws00KJ#KAK}X?#KWJ0Ik>-CCA<14tozKrD-I<$TZIdfzvNF$*N{7o0n!j#vlQk
z=^mdzHr$<En@i>gU`Jt8f(0La3hhX#$>zLuTp_^KtCQ%i7`EA`J$UMYB^hdJo)r!%
zZm=J^S_Ku2OGqj~gtM2TIrrnLiY@8^<c(KmxOGdn_M<jS<d5o$yg*uNGY~e#^dcge
znwB4%h3u=h5LBU%NDw(zY?SbX88N~r^WAu=*+ese=xC@?iSHu1Ur*)MZC@3GL|6%A
zX_0ur<ggE?Lv1xZ9Kc<xEGz{<<3%wAJc8A7IelC!u4YoW`lDZBHKBFDDI|k2vLkg<
z&Yy2<$01u)1+~pquFVp#+^cGJu&JBVwIm%P>a>{=S)s=QelP#`O9)&mT^*wOB2*Mx
z4FE&hA#%Ci{3$%TmKCPMa<f3v;ftEiP%1%(j*nPpXOZA-vHc@w-gI1S@>{uqC~$j(
zFHhhM=L|2!mR{?)ZpB%5A^l%}Ndx*z&l7_UK4x^8L;yv@L7j@ApY~IjP{TN)&!O+&
zn#7Tb+uP#}8n3+i5@9jQ0K6e52XRl#GhqI#kO*~Pd(<70A|@24UeD0cJguf1Db+8+
zp39YQ)9@_bj5~V&<lq9lf*HP_!&Jq&cV<V%?>hRRDZYJt9je`vN1eAf+D+>4Azo9m
zt+m&rriXFoM0C-r;B((GkI~^ypZ^BZ{=z@eo{RaSatS0c+)IoAWZt<kc=p)V;X|w{
z%%OD>KXZ1MvO+YhnoLkqh$fWmbVIHx-!K(qmwQ}gD+I-Yy7Y2P@R%bL;ToKJZn6-D
z54tJ7e-Y8!+J^+4DZDk}A-b!MC%jtJaX}LOW@`4lf<xNn+IbW~u7^dLF3k7GN|img
z^A%C2;z(}5_W^eQPRhJ~$#*)bme3Ff4r-Ol>S|cERJvlo(-|w~k@K~H70zy#iIl`8
z$q?O58<zsr&CBtc8qvrYFi_Xr9(sMEb$-n;LayeeE==kab6-EBjirQJcuFa1!z$7&
z$FeN-x_UipSblZQ$kzmJK&}FYt6c9@E|1BQ9sqNUtPE5y-_Bj>reDXzPM3iw&{f&S
zUFtf4+&+l)=Ro6@mgtrgG;F1s+WX`}XcOXKHd^0tbZu5K<Q2^e?-Gcob)|U`b2E(0
zrV|rE=nh_YF?`6<Imq%2cmo<loD2ugQ?Edcat9>U$~7-07*qq-@Qmf||6T$R?_VCh
zSw$`-G9RdK5*@Uegrrm-tYtvuKW4U+SGhbFM1C$08R%cIer<TE)|wC%kp{+TnYJIl
z<Ynn02KVg!v(iB3YRqx_*CDB&nrFspDp|+R3IsH%N>3IA#xb9}0(D|uAQeEzEZP2W
zjML8cVZE*dXeAHYe<2T!f!~;DhVqHBQYEpAi@r0M<F4{`5SO4=DBQ!U@v5jTg$>9K
zDYCFl*zu0nay{$?RUyL?m4#-!aVzrBYp=X6;P!sh5qUZU*UCa6O;$D1BlPFwK3^nM
zdEL?li<*l&rNJm*u)4xENv?;lgzMABT`<0!%C(s&0idqn4Ms(l%!uYpycZdZ;glsC
zhJP-{s*`(tTbR#q+Gp%nea9e1LM>@dO{gMS<_%ftp&wPQtmR+SLcnm`Y~kJ(6zm9g
z5xb&UVOpff$r%=A0TvvFpnBv$4#i#zo<QWVHm#T&k(WU<oRzuNQu+J9^>Fo?v$mRq
z0i;4U!ak-eSwv)c6{ze?y)(y(Ojk9~>YcuMf3w1sU|9{@uQH`{DhY_lw@ETaz~2Dm
zOf30fumqyIx?LO2l+Hw2`ipaS<ql0-*bk~6LC0}fn$@ih{)bqVEzh0{^2wp2yuke$
z0`AwCgIPN?OT#Y^ZZXhQB<TXAwm}G~8~q7;8o`f=A3U(YXVNo1%NCEVrNeMRaIz~#
z$53;?e&Xy#=ClCT;61iJ^{Cj{#(zCUaOD;tz$SMYW$6!wUX#GOFMlL0*@S+p1K&cW
z_)+Gca&<k*>p9dmH9XM${fJafatI6nAWkT#1;4X6sBMNjz@Cz_;dd6RA!#1Ra_I}C
zeMA$&dCFHI*abOGE@KBk%r6N}3=#z8zF%|s9P{?&6Piu2Z{*%vx2YC;)VP7!tV=PA
zks>(_N96VSik972RA%nIyp?qCyqeOHIj?(I^z0BBF@fU|ki(Ej8qX4YFCG$@lICKb
zkRj<+1Bfn#tl2}?!9d!*O>yH59z1SByPOIn)*oR1)4R|WyIRg~se-`Uv`WK8x&|4u
z=%l%<4XyR=JO@KVHD~KztID}q*<($gt%EFAM^XyL_j1OnMc{I0vWy4|HBHsDIN3<)
z5I1*s6_ERQyxAvnRke1g50L9fUZSf-iXwsB-t&<IK6{FH^n`@I|6sFx?z6TFvV;;(
z_Kq9f@Ie!`$WOl?8IEtj=8%Liry-p7;!&!F-1;06g5M<GN@sIHX5c|ccjV^+qW3Dp
zBWFLZhk{Rnkl4@!;al-;E4$GW-~5)=sZgaK4p{_M2_wY>cwLr)_{ZPiW(7)EF3b2T
z`mP-MR$Ruonb2w0;J}AT4W<TzzlUNCYjg_vKKe&lvQ!2X@X*$1_RMpdm6}8#HHe%%
z>FZ>Lw)I4JSFGt&fits^vwBdel~M)&-qk2_ZJP#;_&PU5<=~XWjH_Av_A8!~)9~XF
zt6QDnCj=Zk&D?4EGvn4ykuO{^p933id1hB!N5L(He*8=2IKNbtJ4@C{sNNz}Q9QrR
z*5<sV85W%NIV4LoxA-cdmsb-G83wmD$GfyntQUwR7r<+TinuuHJA1$6geX3clzeTy
z!*P#Aw|Q#ZFZbc%%*L$tR{4k2l-(h#nxZ-L?Ts43Hp4RYs8B6UQD<qU8Frw+t;qCY
z1K+(})Vj&%6@{y@7N6*7|JbXipZeGlYMfHk&*Hke=rZNy=_2hId82zSHG2@T2)3=A
z7H8Mh6j;!nl|AIuS03OZ{sNzM@-o)v>mt!YSj00TT$#_xTzWZEts`WU$F@TQS8aSf
z-(LRn_s`W`iaN%B0+^@)e((Da-z?S##1crRc5e1Yv@u^S&o26YTA-0XS+K=!qTUu$
zzvDgkEc&!<sgi)*7rrFm`8BX>pwmBqaD@GfPTkU|2V)F<?!Fc&VRX2RIE+TOTNYxO
zKR8eWaqZSKlxcLRdS03O04o+QhPl0ZY*W$<)Hv|l0Xl(_xy^0BgnuL-7TZ{l_u^p;
zZWHw7tC}y9eFaQa1ftTqLZNRi!y;P(W@34Z><=#%a?&-Nj_K~rJrieSXl|<4h8t+&
zyWGaml6q*UnHQexg;h0nYRAMB=iX=IzkQI0r-Z+x)c>=)D?K2RR=2w%G@l{dSAm?x
z@Cr+YgS8hGZNgs+82t>w4>V50phH202RKJzl)j0q-p`*Bi>;+E<*zgx5%y7MbnlG;
zJI{jLPq9v#sJROAr&DVW{df10qI0`8%wEJs_T*)#dz2BhqxbN6U8ui7tZLdaEOk}|
zx|2GR{7<Z3E#Trt5fN;gw>bJ+%}9yir;khQt@kS)wlwYmQmwe%=_xyxzA7Y-?$6Yk
zkW*gZSe2D->}w<(?-KiVv~L%e&Nj(z5U#8RL<Ssy{e_3N1ZuG}RlR609Wdd=BkG<a
zuWYCdBqclC$9nhV&ds0iu+y!}hw4u@JtzK9+^syp#<!fHYDIZ578!3}{I1^Lw?@WK
zVC8dj1Ny=b^&I8MIu3o33pm+7tsKS+Am_a`&GxI=U@tKIWU-Xe<cC}Xq0<)zCS2|T
zgziY=^(rA?k)G<=ec<~CTSi(qdJf}yy)<pu^Bf)REi?HEFV^w|D|b3Ll%bFwNAs4F
zQBId4Gg?h$8mD(f55@+@e3$r3Y4=xJqNjjt#*7&x^SI00uo2&xQ)x+ETXT%9S4fa-
z;_#ZYtz^gX>cj_(MHwKcxeUMA|IWwDwj$+a1xD3W)xArbE(pw~GedLEyOKZX0-xm*
zQ)q$P@mtyMs80JURYWKfgTv>$EQ_q$L^}WQl0e<Iv92=R3;)=fpGv{O4l%<szvKbY
z9xZVXoBMEwP4W44!pfm?%IfeoonCE9a$+-G>_^k>oSe?tW-mXa8u{xTvx7T2W8gKu
zCEBp@j9gJyq3-2Yns(yeoy3EUcfYdtbQu3^$jG?l$VdgsC4vQ@B5V-0Y4y*q-IuC$
z(rqpS9OE7v_lqSeYO6N8PI~?W81n(FSa+v&q9?GrAiG6HGXX55K5>xH$XmBt8Ci_y
z?>&5thj+=A&iY?!`|aiOJ1S?aq~Ht`)Y$D3f(TI}n4N2@Go9NhEX^vJd_pcK+ZE>@
zT29D_Loyqw<7(fGvrbU^1sn7I(&AqICFKFS)2!{JW8KfEVW7=tNfZTWH<Xr`D4iBt
z`3a}$j~xoWXk%Q3bZ0gi;wAJ~)?W=UvM&2qJtJ<BdUN@W=s3p;C*sk%XL(XmH@7GU
z;g;h={jTLC47VC1e*T_>fl=eVdKd98Qw<dAz=m;r5EfU7Mf%o=n63tIny6?cm!o|^
zfdS(hgP5W5-ePd*0nt5RI&@%(H$2K8G_!55JnfrVH1}x5L>ky~+B{AFQ$7F+vDaWO
zdUGmkNM@9zOM3LR2bEl@nWto~@iPA&9UXn&D9tO%&`fe5P%(AZyYAw8oYH2QE@t{K
z4HUL4&$-QCLQ7&+65T)8L$kWMjcj>uPr2cwN;Vk)7Mg_)^>m<(wJ#0EP}1UAYMpyt
zz{W8zJr6-^R12HL0?DxwZJp<{W`j@&qv)f>h6Ek<p&e0-FVRVudIwugvtO$e=)$Eh
zmpMZU3++pv{#8KmZ!+c^8*8Yx?Q*D7CeEp4rW!<~2|q0p;vO<5;klLhzP=I{$ZliK
za-n4Nd0+3#qt}A!4s-|}jxIWSvd})Jwwv{p)suA>y)!^SOjPKBMn(}3u9z1Y%($5|
zG5u`qG#)w4O~J(-UU5EbYGk+EV|l@Vt8{v3tgJQ(Uwm^p8kqs!!hDRFMRk*|afyce
zFkc-17Q-K`&9SN(AS8~AapZpmT<&jr=2Y00$Qrz8U64js#8QSsY`Xjd=opx@*)MyN
zTY?#1h*LfDC@1O8_6{62rG|r2pK1Fj#A`>E`TiRy?~WRtAp$)`)ov5O3ghGG_SXZ3
z)K{V6#~#*t6-b^WX_loGtB7U)M1-Q9p3rasY0<6oH396}&eB}xTsuGvbehgrK`ML(
zOH3zUXA3#LX~&<qd<Gt^DnOGwizpezurWWjMOc*3<X%~TEAs76KL>h!%*WIs%L*f-
zV~=#4ld(22u3i`~N7FUYnC@RqnZUUb|9r9ofVu55m7q?%CL0GGNcDNhIbKQuFN-2C
z-RLC*#v9Rr@#I}CO+yhVwY)xug{k#QTu((QmwOOv7V3akfYnUHa}#!UuGyX=G);cF
zUalPnCp*pYnH4(^`;0J@FJAS$tzYU>m@-{)bISrgo9ouAW{WY2ByCx_EAAZEBj$L6
z@9#g@L4+_LQi=#|4JQww(#I-COP(hTsTwW-`1z;PpK{s3?HM!-%~Oza)dCmM?vTv5
ze`xZ(p|P0-r5eKt=Put@;Jhti`8CzEE{Ap&CjHR_cS!frLh75_+waF!N1|i4I~J^A
zIl4WUQO?Q`?c`9OwtX#xPfxdQZGG#u<FkW-_B(mp@T&^{2VUPi63C>@HO>wyBa;B-
z_F2p-%n^@8CL<^FZpZYDFy%`D+efZSN_dBxs#`V~Ld$)GCU%_M@voau-|ytIKM41&
z2_n1qai<ntk|%^RPJ#1?xL3;4g;y%OkX}B_Yiur<GiQ@wVseJNnpB;t4<h-wX_|9{
zoTJM2_eB$u{sTe2S=aoO83OJkJ#92Gv_4UN0jh=n{$3w4QO%PtpW~cCOD=v!gcPES
zdCM*KY306F8gPi4PC*e}?7(xFmbSlV-_p?jOe8aGzMevsh9*>A{-3vYJUl~HM|h-e
zu_zbWYRkMvxYmarwW(>C{xdtA#OaH<^p)X3Pvg=dbwa{m1;zG-V!&RwpA++!B$dD$
z=$)ymZS&j<R&*uq7sAWjd>%+HwI;towEq%c>0Eqr`mx24t}!-|JoBQaJ_S!mi0RUN
z?a7tQT+@Zvu^>}OJVJlOh4Ba8VX3{vtgl}=AqaYQPP|Bb<WllER!o^fZETwj+ruSc
zxudm1118QE^&`+ZAgq019v`0))R$r_9P|oaqd)0c*{_)HC>efh4xVH^C8vp8re7~#
zm1q2Xe!q<hAF67Bu*o+8tu9wliLphagYSav>;UJsy<-;lP+oh;3-}$ZWd}6ItvJ$^
zrIotA<M$p@$VB|&!ShLQjlpe}K89${icpz?u=8*ksseAdgoNA!Ez*1XD@+HV>}J?x
zqM3?)U0o_$0T*xgNH+X@4D=S#?5?o1<V>O5J)6^dW-I<NBQXSfHKn;4wy(aEnso2%
zmc?}Z`d9P*3BnGZ#{Dj8&81w#-j{Ez3(I|*+h32ywjwqIw?}YvVo-8PNw>4py?cRx
z#|1C)TThW;2e<5p_k}B`xlkK$`Pf|>*)DmUU;cnm0p?7dKI)ebi31bobB$_3<HEbD
zqspK*cR2`kG)q4ojN6saP?hjL^}zZv(^a`92<H#@yx7GrCij#~%Uu`4o7Ec)>F$4Z
z;Zh@J@*{VuIbeNa{A#j8oMtZi@RA82%g|WKQp~pD;#^oqnZ1Y%|759et9E{*ReHO(
ztLnZZ4b4b&l(l}dVd=!r*%ql!cWBe0xmU#VtDWoE7NDo^7$^35yldHB8r$bqxM}6#
zda;0<RbjwgDK7^xUoRq4Ot4$op&{1AhD#_|hE~p0W}EE(eehibBM>5Vrymd`mpHe*
zAdimaP90lX3yS|7W&?8uFPm)rx&H0e`~xge7>&S{1hzNY9auB_4(E|#`3~M>qYR>e
z|6sk#SC<!8tGM6qy<}G^vMuq7x9~g-#^~BG|Jl_+%gdQPbcgjbbT@Ey$V*|e;+K-T
z_-Exu;0+UWVGJ?gOQ`#rJD&60JG<MJ-99f=6*ZBL@A9E3=Z#}rY+Kv>ix>TaZ9cVl
z9{$TD-;2p0N|?KT5C?ay5uWvllMzkIaB;fONV5w8dZi+p!T8};RQ<Q!^z_^Lm#dEt
zo0^|Ts(mRBepSuQT}krdOHkM3@LPu9X%iDEW5oybqrAuEkn%NP@7MA&bWZ%seP*}l
z-ug%o7T0qWiZe5armD_CKLyr!3bNoWq82AA3xmKBp@jT;THil**3?2}ql>sz=28}v
z<*Y^KCZ|;9sD|@>)FeyOlto3hutYLnfB>$ZgQGk+>b;f~{O-Y#+^?`C-t;cw6(XO2
zq5pVk;K?QL`)_Jbh~9ZDEE){aG(X<#C@sX%A<jiLcK3%|l8*5>W0w>Zu$gqX&II&F
zYA0nbnORKa+tG%KgYcOGA?^AN!4RxH0Ey|Zz*5p2wlK)bGE+t^#TPz&fRWGMr`ukO
z9t^uoaL}?bQ*vN^fv0+w!|*9ru4>6ug;Y)kAZ}41O;@?y!okDQ_?97e2hGC3j8&WV
z_2}05Rk>7a>$x12^PaCCJNqxl>rp)i^zK?&(w~2A==QeH9r(7mLN-cUN_rkdniA0(
zpgEcQY!DEX&FFcXe^!4yA7y&vC$b<po)gj!O-#j)+`edf!jf(jqM`T=wEw=@Uc3!4
zE_?5hLj;pWvyf!<0q@X&u8YJnGGzwb@t!+-!v%9xM#hTDnm>|DAk2aS&ZZM|ny^8*
z_7|U|r@IfZG^kTiA>nOaA4q(qUb4WyboMK*OG|I(7p5ep^H-NTC6bpbN?2;}780pG
z;aAtf9D^%(l*cvL+}G~u2B@pb2$XK$)@JTp+1!+*#WzU_K|Zun)=AD1+8S)sEz;yu
z>Ue#ig7PkkI71J2;eN^K{}B|bz#CW+q0;!fye71$u#JVeVRS-z%VxK|T_Hmmf*$9_
z!=o}Z3>p2WB}_x{*4u|pN3aq`c*eBz2(Iv-WU!>{Jw@>J;__QDJwOkJ*lQW-&9AdV
zl9S}-{*<Io6W9If6?|!zQ?XP$IsR|RT%RmK<?JQ)4TL`z1#{QBo<~cAA{ed_aTS8;
z-__A9ar;t~9WS@=L;e$CG~&vUhThBMkc5JUwMNqo42xa=%sDSiv!A-U<U{A5tjoyA
zXQH&*fKFjhCuZi4YZvwKZN0wXS6>B}cn8WSDXX5HJ0IhEOjg~yXU16VZizi`dXTA9
zgFU4BA}s7lM6d*>mdu~?_re?xWwGpN=u?{=El5d^f18+yo<T%zX}1klqOBAfL@m%P
zPP)me7+?TUw5`mhW%XVmNzv3|N&i!~**nrGU?7Q<<}Sy)B=vBS9UC`SBdR-3(+%!;
znZxfEw(b9;W|mB~+*Ie@)GF~&`Pk^IaPgl#tW@yI%6Hki*_-2OkO?~>7MAZHTEG#t
zeYBFRYPk<5Y3XRvSX)Aa|K8I0EFq!oJ6<&`NQnyml7WZZ=l!>!?8Ac1@Nm7_CBO>!
zc!G9^<QbN7L5zmBlPU!rI1W`KU_DzN+l~&e7^uUy{iQ(^nOXJjan@#e*3yUn3UKed
zbKbw+`n{=3i|i+Ja=fHAxEO$#>kpp2wDS1Js?E)Y;bG5066(OvJn;K>8!9UQCDDM7
zJ4HqNV{Gdl1ih~?K>VGs47>)XUUWoee1Rg)xJGC{PMrHDtL}qFr!`!1cJJZq7m>A@
zjW4j9Xr@o*wwTC;{*g12;*gSDS(&=a$a$eA$zA%8$BJXS>xhnJfT|jFyM&8#Uhw$G
z5u&1oFCwkWcZrPV@!96~;K!Drk8?Xy$tBi)yhYO3zrjCh!wAHHPRr%f)zhI5y7zu(
zhEr3zyMJCY464ecKUrAN?r8rcXHTNWQ503^fa6&iTL>GJcbpg;tlsFCbdC0U^)4Rc
z8gsCo9?hmpeX4X?=XwzSfY`v`2W0P$)N_1orcQblx}P1a(;v-mnS?4jpDojSJglm3
zwxN_@U`ml^F#*(<-sWcWCX#UdTB^SI+Pg0e&Ut&#j@l4ZSlZ{I!5CYD+%jhVkwu*B
zrl#uwz6P4H72%O(kwjg*n9k7FkGbq3-}nj+_6bX;{7vP*If*U27V*4Q;&Iuz{mT4}
zT?8pYU%eL@izdb+`xzjXr#w5etezgzi|mOvS;W^11@yGP6-YOOk@>#qYG7<C)^807
z52jZ>4K_YZ=*n=g4D;g#(rU<=1gGIwtt?69i~2(?_Q#UjmNls^fj?A*SpOIO-0Eu$
zynm%pV7juOc*gA$t#Qmhwx@EXyFZ|rSp0Yjd%a<mark(;k<bHO9!o4Cv3~x!4xrp=
zFRluw_T+OgylRpv)~Sv-KEL%SMRCz_rO_;&ADSC_syh7E(r{^avp{LxCvdJ!=gcaS
zG^!tQ<SF!>XhX>9>&`<VVGuL5HTl3#kFwam790K!c~ehuvOTZCg0A#B+P|_fTiA+P
z$&fj3PT4sxGr%W?PtHPBv1bT10FVsa(?}$fwwxr1PJzX2g;ooW;FGp!iOJcraev*+
z9u8GGk-VIccwQVhw>kY5=V~Y}u&jOg@i_0Z#Gjz#Tm_N4U`k|Ycx-9|@BMpYZYy%Z
zwwbc`F!<z`^l+L_S_6=fHbX9(d|F<{Sk`Z>y92*2A!JAvzaG6m!m0H@^90EnQ9DH^
ziNF`E?BN|$xgR7Udu%<(hmuS1Ii<+$=A-pIh_O9_rSf!uH7B5~89C``4VQKs64O**
zXRN1W=V1G2eq^a>@WX!fXJ+uz;X=C+8^#!RkFHfgkl|bOE5gT&()3oDuh~R*_`Ny*
zu9>@-d1!H*FNU8>jIIKn4So&#LxGnd(2~WdMkr|Y>P^iqg&YpC;ubF&XU_F-I(~r>
zMAzTSb@*mJr@2!jS$0ma&T(CCa~c?kI&RpK8`J*_JST7t#y54J5Gi~zr8ih&rC{gu
z17QU(ySwe<Hl4?}Gs8gaJG0(NU+kDaCoWFV16)x8(2`Tk?h0eSgedIKAyi<bvL4S7
zfmKzpI&ml4tBN_6)?feAmSx7*j0u4_3v*{^#$k9Ej;6UL-*bF-F#6T602l)WgcO!m
z-%>{VoA8&;?fcUokB$9Gh!qP`o9gBL;hjwBetRrQIs1N3>z`1+=XIRvY`HRiM9(9h
zeN^WJSY><S>gCB<_4jt^!_=n78^bR#Ayynmea|^DUxQ1W98K>{M%)9^MfO8X2TsM{
z>yg<j2bh0SxL~Oj6`HQDHx5=qXmJoMao%^-2|dCMb|8@?nQf`XkE)SReD55)Z<;}x
zt9UFP;H9N$ceT%CPx~E^(WdA4597pROs4|u$!6LctA+**9R@A(dCI#WNkq4H(wR93
z)#Y>c4|#yq6F~Ge*0XBpb*6HAY$)w5wxex{+=CFay-D!CUk(m-PWG8(T9Z|OL3`fs
zksinQ!H5i%{9_y3W}2Z$l^AvP?X`T&90GQN==b$|Q&+8hz4@@79goke^Q}2EA^?4e
z+V<|(<iR5?&@b(rUs5V(C|7cSnxItsCj76h&xjV^et^?ek6Qcb@0Siw=VugqT-TXz
z`LJCuH+O{B^ilVsaSosLPmqn?+LK;O6V7~pt2ITD+n@9uD-MFMBoxES>9L_@q-@65
zZz0bf85N%KqWFkU>f9KB7Y{&}F^l+gn=Nmy@_gMKd>U6NO|!O7z#K^OJ2WH5`7vV5
zkOuvq<0VS@H3$iJhDSc#Q617K$u%-E)$k(erTT-ACXo1LHo59AN&O6Of3-ah+d|>C
zi`NH&<=6OKlauc#-leK?I-HS#pli^S<jdM9uk={q!*a#@Acgw1>jIh(sVvEFWkpr{
zEwnDjr%ZJTL^2)Gg$_L%M16Cx&w46a3mciCpEoy~!RTIB(D~7a#}L1*ROM`!y;ccF
zv~S&bQ}hveVixD6XG{9*=JH$pcdwq8u2ZB-y9#Ps4Cx`;z&l!BVn$RSlicNN=LCbC
zoq07sg-p~r7mz|8ss(}Xy&f1Vo<6H{SIANpI6)!8B0AP44n1Pt%-O{7@fIE!b8V3c
zdSwOrID1JpN!<&ydCd~6DkB)huW`oU{GUMN&P*TGv&wpr;9tRQa{P{;j6c`5KlNnK
z5sB~QFW1n_x3_+Sfz~U?>h|$JTgimie0W?*1iW~=mF6I|i8ckbtk@<$U72O6LB)(1
z_;z4F-EU+`Mf1tXym5EN^ViVc0RRS&csRW|wmSJR(k!B9-#!=_ojdFkV08R=+*tiX
zrXQ*g{aBj?(yqTG%bhuWLE;Y5+<kI*4+FhHro1%28DG6R^$>gr$T=Mw&qWZ;83XP`
zcS<w^f~7@4j`Cv2Vs`i!BgAt4Zj4GEAgY@_921c80@vVWICzt8^hwh&r1ZXD;7lhf
zTb+m)#6nReOHbhBeD8E!%Dd-Y$Hqx+L5O*yqohOI4NogIH+jKZNJ~M*1TzyI!juF>
zruX{FQ9Ms7m9#xD=Xkc}!Eh2#kBu#`nDHLbSk)zD2$w9+-{%KYE4+~*qCF_$*U{>G
z5G=Z19WG1D^a&&7<l&lKaSD`Cyq~{xb4ww%s35eF1zoZ1lQrSp>J|y+!#{$n(Nso8
zI|JQ12bRE_*XmjY4s`-dGBVeZx`dV!$dY6fb>+qhh+0l``)-S`1G!A2ile&!YTEJA
zll!_eNylr;>|pR|$5))a2kmkV*xyS6O_$o-^D^f-W6GX<58DTRW#M7|nO`yOkjsdj
z5NV1}x^gm;9pPs<+>87ByKNOUn9jWUw;j{KTc+4k9qcqm#;<Ab%{~0$=JR}qdk<1W
zKpF}LP&Z@7jB0XD)=*zlU_%$F5E>yt&&Ez^b7_B<Or<Zna|CUMl~aQMd`WcL$YZxW
z;_WIRb2h&P#<%op3@#v|zU86U(URU{1LE!P--Mq%d;kisIkFD_zUB_T+=h^NlJ+&O
zz)u)7&vE4AC~TfTZoln~;auZtQYu3Nc$E}9eWQb&jDlx7U-aI`Zft%Y9((QEl2)Rb
zZ~{Iy{;;?v(qehbAMNbS*M|o64Yr3QCw*!(S=Ke3sGp4IoMKH&_2vgtWTsbVn&2WO
z1`Jz01yf-#oBL`-wj*wozO+u{zc&gYpi^z>$-SRmqa{rYudwJ{{A|z#1cY~37nD%b
zuqM6(EIzePX`L|7KX|ngoq~r84y2`S<jx|Q(7O&jwSsx!f1de#KO&E9@5fIZ7_HIF
zHEycDjnTP<Dv`D8Ps`fKy71oLOfD#~T5MR^s;+B6c}*%59J6XQC?RnQ7errxu34Ow
z<rN2G?7^@lhe3-FC^?{`;FwC&m3v|RXliyT?dq<@{|%@bUU6$YC-cp^Gtw9VsqxSy
zzR#AFN*>wIF@o!s(9NvH)xW;F)w02_aNDy=umDw%q;B!!)EQKJahN3w$Rdm^N1t0h
zinRQI$(D&ZEY1+jR^UaR7{}Yw!|%i?+0iVs2Ih~UwX&9}SFd9ZyEH)W)etFC;!zEf
zK^UT3dqWy_2=3(bm``+q!fvp;wqCa=J&`q{tdqWfQYuMU%@`}n_?};O+T$hgwW!tM
z`*e3L%?&J_92XxJQYiX-*TB4V<H=yfh}VuQ3yY{L<uoCjDiSL>o?PO(kTzfO1~)}y
z0sxEqkp4}B*d{)a#d~MKXyf41Bop&-z6pNX8A6yQn61=>v$}?$eRFzG;0-LLd+x~2
zOn4hdO1;pCSkG#r@zjh*1CTkxVOpf2sTBOg1eYTT3XeVvn1#+lO~fV=vz=X5<jxDW
zUbZ|Bd7U6{Cul&9kALaB27#5DDw$0ofDomhk`hf*eF>_Ejt4Qr@nd{^z4hpb?ECi{
zQ_^OVTe7y#lE4HF%lR?sl=(z`2=n+c$OeP9X_&~&$hdi;?YXnj1VN|F{@E+Bf3?`T
z4CFMN>GGOrKT?JE)r}UsyRf&jL+wZY8X*|yA8(3{xud3r_(AO~4LmZmx3`&9*5{}j
zhI4Y|D;Y?`(O}#Urbs|9$b~>y$m{7a!WEQleTRBNIf^!Y2VCQ5Vh(w|#`Gj``~opA
zuM+wqK^?Ln*|;y$GcRdQ5`XTtby^7>$hEZ0Q^qZcnksrozQ0}{gfDVJO%)ebPcbjq
zm8A_5G$5;-0DK;fajI}y>~62tnba-^*Qt#$81<|Y?4Lc;)LC6)R#n4wWU~L;)Q`u8
z9zA*22;5MggZ@L)i?)T=v9U8|W)0iM=>6T#-}*=)-;ovNyBAIJd9*_oBCj7RJr{Cz
zV_*#;?z;4geC<sC{doUT;e!YHh{BZ2WXfyEFAjW6zk4D6>1FutKnLsY!MetIr5z7I
z&M&K|iU;QW2#8wqxjECd)*>qTJV@fThN^y8qK-a@Z1MI^IPnzPl_x|Uw^BL3Q&KeO
zyE5KBZQNN$S_gYZiio$iQl;lVxyQ&@sU1f+kx2=1F}@*>{#C9kYcM+@V7&m~;?CpY
z&U`Y`*s91ovSlC6c_7`<W#ON>!puyI^LESjxsz?KK$@0B8xC_Faclzi7r%1e$4KF6
z{UqLm^I84e#7Na4KL>93)Po+^XhNcoMuzXdYM=d3i#|CC-!nai*s}B714*!7>R|K8
zgd_(N!fo4+7_9e(ssz3plZXR3(%+`&1zET9qON^38<<N0^~QmyrNlPRl=r|jFsHLK
zY7mJZ3OmYD17GvS$-WpMG3R2?e=NEJUp*5ZFVALm@0s+}H~7>&=;0nU2JdM|@coyc
zvQGR72I;kgL<BEGCFzyup#C7<djc9qr~-R+DK{4aKs1SWh?*NoQe)7Fio&`+WGV&+
zCyFU*-o-BEF5`Wz?7C3RBF&(8!m37h5YrN$71$=+qv8NAP0Hf@Z+QlrlA?yOiANHG
z&sVs^ec)N;T<<j&c-eDiD%%<0W+FZDL|txXe5wW_4ecj`Zc`|gXvG2V)QA-DesP=R
z03}_3H!#!Y;B36HcR4TWM@CGx#BE8n|EsF6fQxGD-X6pxL<K}igKp`@Lb_w5rMr8G
zV-O<U(%s!%BHi63F?4r+d%W*^-~Z?D-V0ZTnSIXLd#z_ZD;~bPz!z<NeLG-d1K@4r
zaF0^H!EDN&MxWgl#-g<xrcaOvTRS3~u;ujHCVkc+@ZZaQ#ukbKsxb;L8k>8<Lvu`j
zEaI21C+bIp@BaDnlGsoBMSjjj9>*g#x}~%-^Two^X^1&P0;2^<0P9O3wF7;+?5M2A
zc5*AX<tj_Fo?f|Nh-m7;fko2^fPL0&&$@iqm83_z8%aM`beyuiAAVK3n=|RD0NZ`~
z^rFfJhxc2rA{~5wex>0%W2}?@*aMI{8O{`RrlTW&B&tV)wxpiFKMEh-9aB~@zMT3s
z6dDZeS%ZMzJl}Ij7$N)0I(5(f#jyrR<GX)oWgB202>r5iyq9$p8DFiRT+#A`D$kmj
zEMhJHig)6p&a;r9K+W%Bx#oSwCs_+d+7|f^WXFR+fy_!|d|h$-1X*Eoca8gF`lqW{
z3>FIrw+jK`X`nVUJ-YJ2p0Ts(#TY)mxGAJ@r5=Ae_K|g9G+G(W4Dd6RmkzFV-~P+b
zKDe3qN>TF>&0U<=$0ND2=w1uBFvr6=t0+eGo)Mz{#jr3P6YaNw*{tX_e1_YL--EvR
zh9+r0j<5F}bf_=Nvu;X2e-(8!Jz0`BSOIDvQuZ<PnU@xN#}H>&VnVkzV<F<Fb4ws~
z&#LsnlNhJ|7oFhM+fFKj!mBZFo<mhdvcpfNF8sZ}r{#QpfcWh#W!R&!sq&GNy&o8;
zT%?k8&vGYrbVYceh&-J*qJlF~THYaQay<BFWHVFs(Sws;bX_MGCPkPfqZzdk-%9b9
z#3-n6OGkThv(dwlDG~KYcdgOT>I7iKz7bsUXynksz4mfA)B3wqrP;dpw8Q)>M6JS$
zM3e0B#jpsQSM2KBR&&?+vd0{`ghSFgK4PMJJG|ShC$Y!M&L?+oH8-OGbbb0A^{9$D
zUGo+XBF@NiYj!tjzHZ<)WU_}#kBG*8;h;zQHtl+YdKZi(>$q9S{XUel;N_F}dwFT$
zrH+my6AmZBsbU^{2MTazHL)n>j!SE7=ctADFlZHM{eMSX#iJ2H%`02#V~&5w>pec~
zj7MU+ry1N}&U0ebxWD!qNy*r@^)cw9T;kkoQ@;54;okzq=g-nZ<i^}sZ-YGdcg8oR
z{*_xhvii_DKPZhs`JYYmFIanfzq@v>$<R|(BrUCJXEpEYo#z^oUT=sqPS-Rei{l0=
z-&(u$o^#2kfZq9MFwysK6t<qU>8go!&RMJ6CQ*+$;9-K?g>UqA6w}BtP|$3J21T^G
zd&FBFt+7jBW1*@zK<4+?=*V(jMM;u2hgy9zK#8XS>3+>u*Etv7H<~WV?76z*u+~=9
zB&<w|@IccbMUY6_8_}qmzen4daJN#?i6dhFicY`;6|Z$D&TJ5V_kME;$ISksm-rAS
zu{Ee_(iQUV&i=qaJct6i4mXHZ)++z(9W!^cugNqS8-jBfvuP^Y*}k-$&3U3(U3Gb4
zna|&El|m5l$w>c$IL;;#vOUg=`SD#>+3;p;^h$XnTuk;MV2+vDS+rk!RgK=MoAvkf
ze02&X_qH+6{%UQBI<C6&YM}Q{G_TTd_LqVK2{-J@Uw}!8JYTo&QLk<!v(X$>8A<wq
zDvp{wwE;Bf?AyEoKDMuIIi@*&evY-`RHP=;H*m76;pkL&ctsq4c*J`&^})wCwi{jK
zs}t>YE@>yT{4c~yd_?Ey;RR{C^pZMq^V)E6pg2Q-)yLVLupmDYDNfGO>fq{x@z*A%
z`}#4F+#b?rhDx1R4<Czy`v@rP#_7g4NJmgCvH)o^q`gg?+jryVX4hfSxrmw$o4OS!
zN`7Ez@$1a0Y5bKD+ek|biUqs{y8SjZX~S5`ggId|0tgH}h(zh(3|(cx_huv`dcn!K
zGLMdU=aO$+d0;U-zrVlPd&eNHIK!db7?ko1;ka@$f6gv8VCm@f8Y{8&tlLKe=_j7n
z6T}fKU<t}*e3xy5-Lf4CVtIsvYQBQFcE`?Un2<Hggn-yA2N0!y&V8OE{EBxcBpFlE
zw6t~GCLwt$BR>@N^(p;+#*pz0zkSr1=5tVLh6FL4$meYh$Fi0RrvPd6x#U4)8`Egr
zYBVEA2a^#CqBmeB3y(ZG9tvz6Q~6lHxX1H6OfYz8v}VBo_(w<cp}Sp@5(jI^g&yyP
zSP?UCt$ltt(lCGSOD8j^g0fM!mS+0V9JJBa`|jKZv5Ysxz<@f3($KzHE}aunDhAd7
zxWh-l#m1*?z3+)?pHxTBrf@)uQc>R1IDYoRpf@GLiwH2vj>s`l*h~ytqSbx1e}UtJ
zgJ-A+Udz-_<h%pE&*10Fb$0UpHeyvn79(nOV0L97Nbm#JZ8qp$+-@ym#S}L_sL_|H
z?rpA8vG(F84eT|UbVMqM2DmTAZqobto*d*KZU{JN7+maeYBZ1a1%k4KE?&1ZjeEcU
z{J@|zo0hZ^x`o#5vO?AeUPDyRxbUZ2)z<vauS6r<63Ub<Jw4suXN66XBWbt56O*`H
zw22jE7A7(wg~<AFKDU-_Q)CuokP6=fa2N>!cM*z5O7;A3gAHk>+RryXI<ou(G%}CQ
zHhxfKx6QdfY-gq<lZUfi^=N_)l(ZauXx#;->yVJDUajXhLv<is7@M7SQix_Z1>3SO
zv48Y8evs1w>XbMhVv0yBR4v;kFX~X|baz9_2fq&pb<F6yA~`Q#fggpGZ&4;!q-eAr
zsJ7t)F2A+0<41YAR33ctJT;@l7f{4W3XBGB`FSqG_3&8Z*AABUi$lKc6{;!)^CzNi
zV`cIN7YR}%mPLhtywc7HAgHDcXl%?mQmg>Qi5qczo^NI(Gi9<Y3yV`V8G5s!|3Qu^
zX%v#gqtIL0;7%7UxO}DkAb0Jw>Y{{$!Ld-@d8nbf6TadbTW`$(KvNCj&=;EAs8JJQ
zkSlQ`5_s^r-0hsiqi8uQ>oFyzY1Fj<2r-+HO#1_0lH2Tl!$s+-bsOTND`UdHq(;&R
z*Ij{$Ke}psozT%n)5!0i*g9TtRgWAWg%|V4=jev(D-Drt*{0lfAHV}2)od&m#U1bw
zU#Znc2SKAT$vbTN$_Kw#o$sUV%=@+)(;LX>sr7CF=J`RYN_Ry!?`d&Si?tajIc1*f
z?Z$&E^kth7%{ZI>=sEGDk^%%wkSIAl4}O=bu+#`db4nNzAzg1rR2d%o?fr}Atb*|e
zQ+%;zQuJ3CFVWFK7|&|Fa39nU+P0wMbu39tmPOR(&wvc;(io@akV@7<YGLxvv1%&s
zTOJ20)j@QT%BB9va@nT`UAHs@9I=3p^dqbDawTyicYIJcT@?1ai~hsu>5f$Sgn0f`
zdo%Pa#&1FRMVQFP5`B_NA@*x(E4~Yir%tCJGhK){cz|WX-Rr;--2RajLA@3JIAsaU
z>iJ__yZk?{`mP@;o#>@KlQj%#5)Te-4N~F*G+bT!iZKBRJtZdfoSyQ7i3A8k*8Pc_
zdxqQ{svSm5VgTFIKiK345s9{Ma1rIO+JBUJCa&pnKU3k?Uf<n`>pK6^hojVia7a?H
zFd@)`>Klrb0EMlEY^)0#$3BHDtb^Pw;E;igam_gU0|Lx_1X+Oh7hmafV<mp*>iBkS
zp<HI@z93`c;CLiJQi%NV25@u$WLG!yu&$It)gXZa7N`LfSR^WCJDDHc70956%S6x4
zI)(wL_X2P-#>a0w^JiOj-BXP3eFThE^N=m+Y+XbUQCWzlh7^^c=6$!*$et?zb~P==
zd~cp;i3pnd`kh{BQ<GeugaB?rjyykOgpz|P9&~Z0Es(8OM1{Q*l*WlOpY;shvv!*(
z6o-(GaXnJRCwsWQ*x1~J@gmWB*~%_gMI(x3+^KpO*4)I`dM0CTbcFBzK1gX$0Wcv&
z1opPza$QRW>9lt*Q^I7{VGUx13q*0*>dA5*UpLE>R?Cd}LCLF*zG>I&cKy5jvI))x
z=RQD$iE~&qgT1p~$DipM>uN^=#gIhK&;F?SqaD6E4%eBAAmTtPNHBos_!{Sps^WqS
z-16}nMgze*=h`dzf1T$WLt^a^_W8sy`!jl2G5W+>*l#q+uX<wH%NK;!9u}MyI!n>{
zaG!B}7t&Id?vfFPf|0`2V8%zE>^}>pKMEPQS|(m_USquX$^crloH%o2wZD1+@&-=&
zz0a|ZNfKj2(DNQIhth?S<KbDq<UO{JmU43=zB*&2H_rT$rH~0|oBJRMLN!_1zwQXT
zjo@tVocIx#Vfcu(c*ioQ(z7ru@@cseqfQ4^d@-m$R&p0EdcpSyhfW+2bUGle{2)pL
zXDOo2>mwJ1B!O|?T_ONX<m=kyQoY7IF1yDc9_J{x2A#b3|4pS0KsntVUSdhfb_S*e
zfcIC!MCBqCI&vO1eLVYPQ+nh&e}+9^ZL>N*<DNcInpo~d1dQvi_TRcFmQyFv*Iw#A
zi0);xWFEw8iYP46s)N6CsJgZLCZGqE$%X)e*);g3oRREu<8?2G*>{yo-b-X%Q|=<(
z46l9EaX@UEUo!tzc(@38pxno<ur7LquqBo$5)!VRQz|wGhwObPR#p!E%NmDM#Dc!S
z<-P@~z>z7U3DnU7R7D69!2>_UL5XEj%3HM8>C0~P&YOA@r;ejeG3T4_Pw=5VAMqz^
z3qqi%v-Ui4I*}nePff2i^3zNdU*QDu76k<{Ha5yYdV~41qz2#t*SM|R^m_M}hLO%$
z*Z2wM|9AjYktiRYXjpy_&Ou)7Rvshpo~P`K7zW0ULRD+StBo^p*hH<ugI5agbz}cj
zmzxW<wA7zK`k?uFqB7du*VlLn8%vL60^z*dU0l36TxQ-rLF6r}OMo{3(%68*Ddo)g
za<lu@<zs7)!G}Q-o1?-8H43zP3pK~zr6l!wg2}iXHg5wnNd6daW;i{>4?E=i;Lo~&
zftE}Fd}jOLXh^h%*&dJSY6lnOY^ll%zOx>i?m_xS#$xeDW~Yb45mqBbg)tmdbouJT
zr3a;@^9?1t#J+w8hhZjw@I<~fK2tXXArT7*@CbNHTe}C|?sGb;0yy#r6xq93y&74r
zg3DG<j9YC6%%aK=_m%SIKWIOC7&#8gA7}l8#K=LR(EPNK{8}tteQ`iJrSu~74Z5>8
z=&x?di!1Tol-TO@R=EzUA7PL`u=UP?j#|K$$eN7L*_*1%nlrWyXH*#}s2Gf?c0;dx
zE1S!>4+5i18S2&9u4;|cr(1e}BI_ND1&Qhk$Nk5y6QyQ-HUAb@-F!p#m#6`0^;9vB
zZ;_cA`5$K6C!r6;9ZcmPopZYvH*z|-vTyt8=$QKEdcUF>AI}(G<DqJq`N<Z?N$Hm^
z!AQkN>2GrH-p}ZtbK9GuHk{=u>)A7O+~Og42W94CyHn%BVp3AElWaD2OML|KHy*^t
zy>T?9kF0b$sjMEXi*(3cqFC$r^?UK?$41Pr7jKi3y+tLVrs@7KM*=ey6T|uXtgnsB
zmbqL_uOPjQo05A_K{*vZExwUZa|%9zj`64O*&a`i(uO&wgA*#ZA7NmKW=`GBWo)tC
zZuz-2Et9G6zJ=<+LxKRg%oNZjVdh$MDD{1|v%SUh7HfL(r)66JKt;NH5(`pp8t|${
zjZLmkxo>Gf;yrm)^@p0Zm{Gdr<$n0%nW-trWkuCQw{z|yqcJXrqoCkeb@7z{15?NX
zw(1W&pxj`VZhd)uzd*yRiBA%VeRYw_qRLKsF<ReRS=gD$%X5f5<h-@`k>=XE?~&o%
zt5Zd8(KXZADK%@3gbPzfJ)*<hSxnN90To|tEFoXO1fVkd1QYl#(I+@F;wO=Lo+bv=
zVSlvryoE_Ph_ZREQ<n|}9e_(Ko6)e#%Ab@ed;A#S?x|xdLUEm$*|f0H%9NV3WqZYi
zpBybV3he%Iy+Qw5^nd6aU>AYwIQa8uD8T_+W6qvI3BoL1qhml*+b~4~wQ4QXa@X%8
z#BwQ<Q*Pd&%SKQ+5QT{AaerlFAhH{VBdiM3dOK_mERZheQ}~<}`B~>Znf^=9-K(q$
za}gRu3s;Pl+0teUDfsLc8X1Vk*L1zIlA*Ntjqv>9<6pPTDN_T3BD_2@Y0U2*9hk58
z7W8}>gOo?E7QFvms0{SiiU1B8FD(2~c%10x;^di2P^x%iZZ*Y~!bltk$NpA4(nXB$
zO<oToz=oRv-9t>s2FIcH+z_kI+#H}`(1emGmmS2B;Aw1WO$cLHD~uc!R0h8Y>q$-7
zwRMJS&uKzCz1zB4I=UoGpgoyBO`i_rI(SQY)%E%b{_2v@zbSIFl5p(PYpa25S0zoP
z9NX-h?rE|uho+_^x$JSvjhp-MTnZj`MU$>==VKk+PYp3aO->FIx-q7|abEsw6#X)-
zeh6T^<yQ=&9kYHZwk<H!u-tLPdi!XUkhR9<!PLWaC_l@|e+4`bH^PpYj^Mmpm0%6E
zU*o!)?V(j|^fXI=A_foI)?_;Cp~1-tA@SHl;C!|+xWw==YR|_>@V(SBbq!^xuyl()
z#3VB3*gS(xS_sJ}%B@cKf<jIZ(S5L8%_-He7sqj^QIEOT*+x#wDZOo#XMnV1r}J7i
zhs=Rd0;S4O)~^9MO2!NK)p!W>7SK*lsht)<O#XZvU%o4|-gVQaU2nnC@r3BG$ak8`
zNJ`=_e{FUj{7Y`Rw>(FMTf0{}@@OZ!a!(>DQW1w%&nV;eX^fsceX_II*QlaXRc<82
z3WxWQr&@9`%H~@ENPPlPkQ2W{e=2`|*q<HHa`1cZ3yhu*-&dtqK!Dfg3n<YBcJREH
zoMw}r;puQL=6GuAg$uV@z5u8#kB{TVY3O=`PH2O;F}F2`Y7QYMcwG2`UM^n%R3{4v
zIzP-fB@`upq^~3(Ocu+NKCAmhD9p^@0?Y>mwfASWyBxed(Cq~?o$fdq9c~aFf;O3i
zYG+&*Gqbyv_PU^-C>)!tt#n+PdAc4S(9l>e=junAH3tPa{{o?T&37p9n<QObcH-<T
zt0vo+!|x@NwH}exMa?+xaVZRQ)as)4HO}xcc^0zUAHs^}Xg;e|v=%9$l%c>^jC=-B
z;A6y*a3{6MJ-`{j{dRi}IpPk=qc3b}a%MJhK+~GI2p1|N!)w;z`g1<n1d#L3f<#VM
zhPXk=!uqt^jCb)=S35;QB+k1VywhG!G2;0SZ^Fo8xSBwDFrqx(`Q#>WiCF8zBA|#L
z)KBoZ(ff~#<kZ`TfaKn=a|~3UH}k2}k6O*t%*knL1yO!YC7(SEGUwWFlXX^>f$HGX
zkk#$g#A3hV%BP*4IE&F}wZP>>9<>upiHIRyvp)Xs9yoEpMI~Iv>vrDxVquLD;I_up
zdV$R5`VeDj#?M#L_JIl?-9ZhBCf|Na&xvTQGfN^I=MjW~@wC`TIF?h17ET*kdi>yy
z4zA(qe9za&fYHq9OMSyn4XAdKwuaA~0WqKluau{pKPa$kVe%Lp^3(Zdq`3v(k!Fyl
zg0D($nnL2IsWf$W1Kr`wSi%#QlNYmb^}aW$X_!Wx4@(kMKiLQZSQGTsnmd_ua}HrI
z6+4j$4H-X<!9vkQ_m!0BU00oAcf-@t=$$X|L0eeyF|jsW!88_GajC@CyT$MO{)N^l
zQT``$Z8>Dw_3g=?7`^pXR#c>9ba8h>V%`4Y(VU0Wrc<rW-)4}dR?Y40_y2m}tCy%;
zmLV|rV?~yvESaGoz<C0lAoYR88h(-wF6VqrvMTp1I`BY(zFVU5WY9q*waP*VFpYrx
zN*3>Y3>d<M$+bh%3VYlzf5j%G+`vk`IW8nL(*T-2axn#ePfV$8>zJ9UtCs4Hujj*<
zO?sa%Z}3ZG`Tw^2qSlin--tC^;CHcSJrm1uDyXbI*Q5xUpslPeive=Zb<}LJ8uNK5
zjq#P$a~{rk8dZn=dGF$%Pug~hi<k7}Scn4RU1SEYdQVr{BYn5Wzk%{HP@B~Tux8Nu
z3EB$6{=N?9=Hh-aK>;(n*5JkQddY;F?r4t-_qq|VR+MJr%^=_nmRC{IaRCgP1+hS5
zXs$Cv4X0JU*r4~keQV9m_gVquD{$a>pi!nlU@SUnq&fgpF#Q21G_mt}%-sOP4H}I9
z8rKV?$j}<1j_8^!T`Bt>2F9P<@)_;D#daKL9)vvU*kwVw<>cu)=hKss>d?>#^ghrp
z)8tyaTU^@M^82UnA2|?9jy=VP-r1T9zy$IbV>Hi5I3C+cSA~anIvrJ$eFP1fFp$|n
zwMW*5yp&o_0~EatD^&!bIm>aoLAA>m-!=La@dnXp22m?mp2YbgCV&{sub4j-ILMzl
zBj$kbPLvQ6(|Sb(LLMR#aZrfpd7H(}an~vCDaWFg^yL@vcBm(DRfpB8xMc1uhU(;O
zisEtX(9kP;7EYbed>mkeEjc*T6%wBb)PF`V+y!};w~2D|`bLeQ?oot}QR`wk)^}qf
z(hO73Wz6d^KUE>~1T^#3eXM^7gJar!I6v*H^u`1wMSz2N5DjktMK}Ib<lx}!TG?rJ
zUf}E;A3<58ABm)67FY8lWaKpbJ%XM$xzwzzuc9gqjM;%bwE{Jawf$D$9l*9nwRY(M
z!qg9%3#jPJFubAZ=m9ej?h$wuwx&8Z+)btx#8-cPx7ZVB2?^IQ9pdKTNR<<HTDp$M
z!op@|VMDaBJ|d+AxTvOvSe{j$93cpy{i=Usr2KTP1N%*PR~PA?sk>KhN(tV(cki%F
zsEMd9c-w9V^qo41irxbgj$t2FRJ2V^FP^0-Qfa#E-a_`q8vvzl9B10E&!(teG^zp$
z$r%c5U*bESmIH%(h&}wn5TIVHD}MJ$zT}hi<BuS8IsMdZ%s5q{tmB&=f(va63WP;|
zyUn*39jHfux3_fW2NI{po8+gpx^gtA9<;`;)su=UzPYCYgi54bzgcJLwWPet{wCxB
zwC@sh3g?$CWDi#>c*U_C)<fc$DGz)6!4m|pY^QEdX}#Sya*2#{uIbtQJIj+!J}nvF
zSF2Ie@dcHdHdK;}&7f?JnSF%mG#qq;w60%i4}W>rhL4f^s>Rdu!eZyOz)t~eI4^H)
z%W3-gMm~0zYGu@EZPlK6`Lo*dc6yrB)qy3DTYOM%u9w`s?S`s&kpcip;o(2uoX-Lh
z^y@Rej^Ec`+=ah-2R)>laf#KlsOgv$_TL57@F}{Gp1gv}kJwZSa0K{xjc=l&hMkY-
zJU5vW-og4aGdJojy?>Ul$rTqB78fODhTx*}l%#*NU?mTRS_G#ya_LXCN95^$a>piX
zX{lAXg@ncs<B3s7Zft^do3_ZObVRQG73U8rQM82Q&g=TB?QO%}<=vNH>3W(H#DKBj
z5(TKUs}DwiX@jU>PxKPT#oDGu*QO8YpJ?>Q%v=S@QNBk++n3RH2m5-xR6Ziz?ONNX
zK2u^5ag}I<klT__UCbq$7o*n$Yr3cuiS7?Bq*7A0J7z7QkGLR1r*<65&O&B<E7tjT
z20CM!r#pgw=g@c1(SizbQ{UBl{~8mjzwHOF0>vT)3Iqr59Mx~@u!A_}wEv}Sa#~3W
z-b<YYVYkve#<Z3XSfJ*4r{@fyUrz4MuEEe_%b}u<mj13#&}y|d6_Fgdx+x&^^|;Zp
zw0I#i@!R$}7w{;)>;17-{8h6WrKi-nb&g{ZmQ`rap?WcZ9(d>&B5)rP2#eX@w-&ku
zQkEq<maY90XIj#d4~+E`BLiuvBLLXG;nb||it!5)8s^s=qwWn#3lKI!7ymC3afd)7
z0KTyu?Tg^%=rZe_0X?xLhwEgG7eRfgdYExf1wvQR&n}bU4H(zv=KfLmU(2nniJ+kX
z!tWj3=n3AJ$(5&Z&!(8R{G*(&^raYG`KJ%J(j^kf3lr`FMHo<bMP!2bd$TSyosUUK
z6k`jlE-FKG<?gxv7NP!gBT<AP<?ii|`1nQ~Kkg>&oMmO_6__Zms}O7mzfpJ8!hgt_
zlq2>KkNn5bgDC2oUrXHGr9@b0ppeM(t1pohX)q(skW;swcJ76)?W^U;Xse%2)<_35
zm_WPz*7F@jD$tMGX~UFttw}lvHyLebp03IhvbmIwxPK;J-cwa6NvBRsm}JLmZ~SN7
zB6GbK$56b3G+PHf^-L!MjN-|Mrs0`RWyAtmz8l1chLbJGQ5}!%r+t->@s*%T<#oay
z1}1xgcsi}f-$D9yelHzmvqImk5?(B_igAbLrOh86NW{kx_UDy;z3?GRT<4I?mEun^
zfaeZIq2I9^P7!81wg@mD3#+9UyC0g@l-fPaLir(jV6c?uFz}4cOfAuAW~M?pn5{;7
zJInWuqla58Z`K>bjBx0(O0&IHI!jXby~nbX^TK1%fgJ(3lHT8GsXL;?jXsa0^bJf_
z1|fJyqV4q98qS36@h0JDp}My<JlNDTF*}OA|Hq_7zl~LG7?QqklSz8mGq(LP`JE2^
zC!TQ?t|{cJLEB0+`F(bcHaVjFDbRIHPTLNm%3&uV%m3~Wg!qDncDDXqb%{grQm+}u
zI&%3|?{2)hbBXo!n^Irz{){WD;lRJc<XtdtF^M_O#X6MIu}gB6)F{Ny*e3tlJhCy@
z5kcg@9VAIC!}emOrjIT<JK8(5uL>G&RV+jauKnQ1Eu~rbzNK3B=W9o%rnH{%B7tUS
z=)oU#2kWm8`$>+MIbqfqrC+JbEE<1gfeL}#uDO7V!gcYYDwsQT%_|FPpMvh92B6UT
zd?jcm-<-%jaVN92GW;&^#3l9uYvF4dRNzd2?QND=0;Vta5MXB0P{_P;<A@e;UKVX;
zGtH28lNynXXBc`Vyv1FvTDSfA%<Ul4pNBZgu?eyR|FL5ae4G*c_6B<C+_p3wANhL=
z3iA<lpKu!1`h+Pn3DJFrtr7Np0l57ad1=3=sn2VjM<hO>7L>6-?I<fDOLJa61J7`q
zs|=XB<bOBdV+HRBWV`J2u!hgASHHG1euSTipMUkvEQ({8)A=}i(--G(1IMIvv#n4Y
z5`MuG<Dn8-1aA^J7yBl7DMw|!e~_N|?XU;^xV@r5&;@2ESuGxAbV|pRZKM!-b=k>S
z8oisCmLoZohfC$-$Ez-r&>JFedsx%Uew3ZsVa0Q}QSl04zwyK;7&X84^>mZN%X!%q
zYom70_otJlgwu6q9!MkHr-^Zr+9iw-gSu$7{W^F?*~}C&PPx$>{ZZwbql08FDNlXs
zM~3W(*zKl2c-%Z$|I;V4^!V|bp?_1c`@8)W!&*MAnh!-9)N@u&rgXW{M`>3_(DlF+
zKHcFlm<bI2Rp>Q`+Zq-WVZ`u6E$m#(r%Y3HAFT1?+Fs&MGj#veF|)&lZZYv4kq9K=
zDl4z`SIBz!&+tzUgGngrDpeeEJ4xYLWQ1Ae*S7hjbxsw_t`P}B>mzihpR$hPIzqLc
zRAsoif<d?%@A5Y?C;#NheR}<QIOvN?y6k)RtHU;3ewIH<u0<?wd`78gDK-O-V!A@R
znTNP&<*a}bD>$m)Fj_#T=V5%xRH%oUU`_2fALMpXLB1Nig9wYuHgfz)PP5>kIeN=D
z`td8jgQ1_Y2;(kld@f8|9y;M8806vU-WI@u!NnB8UjLrSmn@y2B_PBo3haKkWrDgJ
zhtgNn=$X%J9~v=nA|@F`30$;G2TnkD$+E2?f-!;8<K`_$g#>NK*i(y|o~v!`pUJ4K
zU>Of_$EObzKm2!SfJ|^`RG9i|<v>wySAT|9lUVOkw-hh2N`-yx<J`%F_~Y%-`4B3-
zV;mU?-FJuPqe|e8q@t<wbhe0s5wNH~GH7@;J16gwKVV?qmy>IG$Q|Q>aWCU!5L)m;
zHwsSaVYre}+^rEy>`XP%WE>I{@E+YARE}FmH|4;unidnZdK+z@Zg>Xw8LHI7tlblI
zSYOLir<eF=;6BXaJ-EQ?`ue=}8Oiv3m0Tu=&Hj}mGg>xjW2G`B5()Vx#Np6+k-cI}
z(GhP@SZY6$<nWJfRb7#)T*xfP++<Y6HL4{;%d$Ytyd1!cQkQMC?wO-vj?GiWQU0sY
zxXJ~4e!Esf|K)#FP8ysRHmInVXBxkei}vPg<h#I|M|xeaTJ30UtJ&v0zX&+A9BfSS
zsqe4gsuVi;F!~s094-j2Cnt`Y7@A@V-M-TqxM)$lI&d~0H3p@AQwku;gs@37`aZ+>
z1s&o@V_$Wy{Ui7UgE75r^wGWz_!E<AyWtu6NW`acu}Ff7%X(-$m0NIA=KaT?q$Ext
z*2?LotlJGQyR3q{MsX<pA}P<5w^gWIz!ptrGD&s^vya}(B+RJncr53g<mvzVXF_$u
zy6g^=<bEA}Q9vm4ekFO6_n?oHZppT~*QDA>aC?YflPN;$N!0$BCjP_6O5LI+5%|&N
zE5&4P3WGa{vxBrZ80C|_&XfKGLaPf}oFc)br$~A*Q(Xxz@y|0KdX85^ohYi1Qs;8@
zRLH~2WI25E$N7qU?i5temFxuZhh_3Y1dM~zKJ|);letDXGx<aa+x-5n(ncGy?hx|Y
zp=~j6w={Q-Ak!I%my@L<{aq5Z<(5{8t^Z8MeogXQbVMf0JXfdGd?K3g^bZ~dFwR7O
z>SgxN=Bt(m#BD_{5f}&i@(6o>rF7DPmV>6?6rZ)xQLlh2A@piZbE`8bD#CuFV&vk&
zC(2`V(9BGSj;~h}VK0B-uYR`SMRvIIT{WBe`9IgQ_5wpAbalXoX#`9i>r1~UxN_M0
z*wChfw0oke_IE{^tZWqX3J<@_zU6RsLr}I|qk}3KAb-2GMuDR#V~39<IeCb_CXQoM
zsvqp_A6WTt#fNC9!P($++>zZ;3Iq}cgR3PxA3P!%N1Xy{8S?VYeb%#wl3PojJ4LWa
z#mLkRX8vpwDXmV4<hh)eds-hBm52GWybLsu{<5uB>_0eX?ICV<QC0wR_MQY!2yqNG
z4l46lo)i$edny<Us683Hob?cyN~3X_-+!vu%GS$ARs-$YVJo(pT^-vs+D7w{xx}1K
zI^9!aET_Ffstw&*R?f$+VSF7dG1zUB=x-Ch0A@nfUr&q>R7`Ja=c~fjNHJzGIzWRv
zP`)01MN7{tc>T4!lDa0&ye3K2dmR%4BkM)*674?DPJfqtaLVmBU&4C&nNPZipHfL>
zNQSh8#X&gu3pp=)O%ycTZfP*da)ObI0yRK4r(yj@&#!viz?8QJi5QB8c329@7nmm(
z+m>8eui8E6sHp?b`K06SH}HZRCnF;AosiA&lbtVy&7E7yB#B4Nl0Z#{?%A`~9D?;2
zk`GB9J{RQ)iiWRpwGtaFfA{_&_pCmhP;tc;(E&xns?|A&kckI+LKM*FX>+SbAKN$W
zo#Qaw+}!qX?iO%8%vrqJ{GQX$@H+F6I7{O>cxUl?671G<_C+MJb9PqR&CR7Iuk&l~
z?x`RXleCkQfk3B+yL+34&&<1tP5V>Donl7Vc0>JZe4vH7xV>Fl-793Zq;*v$@=dF_
z9m@ZDzJ%ik<a5i&_%+(;2YOrmJDIWA>$$3aImx_Yj806l<1zJYygIh>s1M`h!*Q8X
z)-+%<tzkT=O3E*vxG-h;n9`X3)N|fRV5bTu%HsP23jge|!`q6yF2=dPyT6~U=!$kH
z4`xSp9f7Vm2h67K6VGrkO%+Zt!XYLG3FaJZ5Ucm6mA?4(E9?L6^aKnw;{Ml{Ja!7;
z^a#w<7600=sbUljW=on?RK)*}45;0cTTC)SaYPPY<g2S>Bh>u7$gky^08MCa%_O%3
zVcYfkaVpsTGkPC3^aiiyHU8@0Jj*oq4k1a8)yl{%$cJ&o-Gu1Ap5aB3UT{kJ*0rX`
z&D;N<33-hlbu|ZuCg*{4hyO|k_pVBGxW?AW+j7<Gsf&=jOWiL;m@x=Mm*MnOxiog_
zKQ{Vvc{YE#-OaVn&#KimuNQ9lK}&Y7mHM%_T`nSTUnN$4DDfE%lO&>pDCreW86)oN
z>aR}L-Z_|;t~VU@4k>!-sc}B*eYP*zl;idn`T^~!7pT=|6A<T|_VQ(N&*t^n*etI_
z*Pb7C&s^&H*JtV6k&JbWK$Mutf1~+3i4P`2AhUC0xIzC3q2Lao-x^u^QgoT&eAVHZ
z-ZdJDfRh}6zQV>s1&N4qlk7zJkzDzh5(`$-TH8P7X6G&VKz0l;O1lD%gS|D}v_}2U
z4>G<{JGzF>HW20d%a~M5n=i$4J&zD*LnFK4suH2uhDQI0O1fjoS~Pe~NJu&6wh|qe
zn_78n8yW)bCQy#gi6H7gb0z??T*D7x3anSRvOa(Res>e7RRIb6f{5G>H1UtU{Kqkm
z8P%voqxhZ6fLNBU)eD%?WQkD-Q0XEC5yE`aAyAf5j=B5VjM;RIIW;x4y|-6-eKfzE
z=l*AvPa`?Xi0#^A;2j08Yi?r`qJF*3y-5ZZl8F8~BCrz;8-1}oBqE^`jr|Ugs#*c>
z7ctd|CJE)Lq;k$UCQrSRPm_?1<}fp+e)xG}8E6-)6zO_D!o{^)Y<s9~+b^^>m>!T<
zd!!;HEUZd`^BUWdCzkM)_HQS@+SBD|AVoE1EOj%Xj9F6N+uQr>Xwm|uunn|xsd;!*
zfF#*q!kce+xVSoY?_mO8zY=`zWs#I;j98D+fnX=3gKUSV$as^U81CjgX-P>%7nhnp
z3k#*68gHDQ?<-;K+(IJiWm4?cM}}i=zBU5`6A#C90%#~GQpXN6Gc&~`C5Iz4-7=lG
zYvJE+V;50BhiU#we*_cQt~=iv#O9p(WQPSSZM!(#XLs0AJ3T*NpZ6jg!H26QCL}z=
zhXB6qOMItNVi+|15N#;%^Jkdr;V@ip#tmFIN+H}+D+4H%vH!&S942tOh?|;{LJbtn
zr1^H;fsz&#1VOO^t|~xrTn?;y2D)}tcRO7&9?A$Bwn7`S%4x)GK#rT%;N3XM2MSe2
zK&(1A#bMSJZe?YKasPg}))NS&S_R0g!*sz*?_>uwpz;3b5xM9nkQd&tuP)Vs4gI7L
z4;Klhl#a`hishjJ1{SOc$V?Nm>ibTVTQG@=Nl28iP{PRGWP=l68GgEt6z3@prxc!o
zQ3n`BsT6$rLcVMEtikJN8jV-eb5Y~*BE3u<#L)(gXh%;^9#p6Xb0GlIIIso_iNT>C
z_;ldIHJo#sjbKCe?d|Q>T*N>PRROFp*B?E>^NDBT)^xRa@X^+^>e|ng6mn5Hulpp%
z^aQZkW(Q2PlN!Cw@YVJFdL?7yEZ`p!jy5OPqfGK-!Ha+m9NvWuC3JO3;N#(m1QW6w
zr>Iv$_C`Q1exk}Q`|Z1T!<9A$KpZ`#E7=Zq63$eAAs1}Bo?E^bhqm(ymr_DTPA+)H
zWnYs2@?h|XpI;czNhQATRd;n}Vm4i+2&9Ad=!apj;dW+jAwC${VH1wvO6jNu-Tsk3
z3GdptLA)M6e}mV?WVtjB|3P3<QWBf-5H%zj&5=7~#X0Lz+!8>hSv(ILq6LHClRx+c
zle=E*wr0We<C?u;LnR?p(#A&{<7Zdr+dk@KgoIg>7WL|2zoFDxg;rrFkNoFu%@x0W
z`}Uv#9Vf&2c+0pi;R{w1P_~oN)=o;3j82PpI}^>XzpA=ADz8^2!GDbn4sZlS9?NC*
z7Ra{^f)i;sX;w}{PcLzLdW!lScGojohK~iWV|g8vM)TAeq_7%j*GF<g$blALm^HzU
zEZF^LhohR}YVjc<$)TZmdIkpBU~4=kN1rpHc0@bLNwO>Mc?{Nib$Onxn5|TGeRW=H
zzp0|9t81A=(IAXRq6aP5R}U8(NKd;SPsdGu{rc5*+HRDFib|NwW#@~2Z(Od!j8kh?
z28{kO85z@R*TYMgOK*{0CmUEAumtO2a1GhcaE7JrZ6>d#M0fX_m;3S8L*Qhv>$G68
z8}>D<RlwG`q>66CQVl0cGSgu*l{V?9y%oB%W7)fu_zv<4=h1KYj24`7d4AlvYC9sr
z2rCs=QHdHD9MmhH<o@z#?%Uqp9_sT%2Zl1`hs(@Vfx2HCGnR*^=UNMq#gMwgbZSiv
zUw0h;<ew6l3mVwVp@}QxVpLKw1zh(&M_C_@QWJ5Qeq8(6*C*4vl!RM^w!;9{96}DZ
znmcsJjz(4jUAu|Og-}Z3J$<UY_Os6QT#UzN21m;EW<r(h&W+bk!FHz;bxFn8c@ETp
zjR#V^$ydQzQb8XnadQj$4g=71045-q9GjS{?++U+N9F*Vw28TcSl%{zax*VT0_H-5
zTGoq~FW-Ts1d^AUO^$%q8O!xUh$x|3J=<XOiUkj4Dbk&u+wh<_jGBTC?cw2p`Y3H}
z?Hqsfh8sdiq==*>!#>WhY<PY+IlIM7&D4u0u&f*4o#R6zBQwW~4NPY1TswMukJaKM
zBGQ2cUh9Rfipf{^I+@@LEq=K_?ls)FJUf&JDtSuz8o8A=NLq-K6Fv}`Y^Ri`mKn)v
zASPFcTY$Dh5)WNfk{kX0{_rss+NZFL4C+$D{<pyg!0sCL#_<Dbg5VyPxD#OJ*-b|1
zMQfB`8BaaiwV7cqeVOv<Z01uH)~zLGlS)iAJ2x~NZ@T~s>&51@x2WCuDrVE4cWVxv
zXzm*eD=WJ_wDD0CWjeop`^F%3vw=2UDt6!-jzC6NR~Jj`!TP8)ICsl;HDJU3D6i4c
z*476$SE*Lh?ak?GxRa9;%GC1Mt|XwL-(D%|WL)TeGS_5nV^iw5tBq1%=W*Ogc=6)J
zP_8N`HW7zJcNBXu4=D_jUadlg)u2aVwvN*Ea3wiFh2+lXW}w>x<T6vk=-{*z6hgoV
z^!N9#jTI*6)n6eZ-VCDo;1P2@tDa^AvXW+#Wy&P2NVA!m@rb!WG(2cyERjIN@eMHR
z#$yGE;9~G`ag)J5T1)gIMP1&}_AJvNTdu##g2Z6x;41hhXNOo>SvgM5dLTtq@n_FD
z1r?R^k)*Zt9(QZ}<@ruC5;gzVC8T~BSoPPNp<!XE4_|&TTpdWIFz;+@llkz$6ZHte
z=@brTv$#FO=SRw472!05Mj_i1%bV-}H{7JYgEH*Y^Owi7$D!3|XlMxF+SpxAEZNPc
zK3-j20VfPD(+h`eq|9ux#IXOxty{O0WB!hZcES1QSN%9=b|SWi(4IhP(dJpJ-~YlP
zt6+8j&UnEFov4TE$uohqqv)9Ku;d~>J6Q;@>traWrlkA;B&&@l%W}8suiYppDZ{Hb
zm%F3+siwrl#Y4z9rz+*alO0Ko(8vPE@^Ecf+@k)n3`kz~L{hM_%8Q57_XE6Qw-g4a
zrKJ@EZv_Em(^5)_p9#Ro<f|T+W*u7qD-q|iCt7MgtpZ-KHdR@yfmo%81>mV>?`K*j
zc%dILx16l3Y?fwyo#%a$tdkSRUPRNVx<f4Rtsne}SI4)2)h(LbG@h-iMOFc4<-CV~
z$#=RKg6xtRZ$(|}kUeoDCJgqY5E{QaKb~!9XrPUD=uZ+N<hD*BzAs$39ug7~2BcSy
zXPl+qd1A@_vz0+((9yl}Df4puo|sJMt;+PVpNH$CCgAKL&Q{Y)^}D0AZ_STmU|<XZ
zTqI{PS?gSGzde%+PH?tCZ`|2(v;`h1sYZ~OT5^{4byQkrvkw7evNPA%++sReCJB^L
zLAPH$O71mFz1m)H890(rhVrQ-@G^pGxBrN_R@rZ6fz#QxY76wOvI+_c#wwENg}|Do
z%pD5Xd+g+@I{xa?G!R+dMD3Cm{o+jV2*%));cWPN!x!|hI2+*a2w8MRy!UpGA4804
z8QJ-DRqun2odL2k_qIMZO_z>GIi|3<IE};_p^<#eeBfwC&t5uZBd*;}=DjF>^cQmz
z-#^@*t$$CW`kPG>oWY#;mCj||(VQl2fzLZSJLx#Zg%poop>FjI0ERGF=nhIrlCd0U
zC<X{#RdAS1*!J8gWHW+{jG~yezLICY`F8tkrwPYyrSCb=8m$Cae0iCchDH?qHdY`x
zN{@CQ*~{e6LnEKo1^{xge)j+{NsIQ?050T%JBA1k1<<u>J+HbWMhC3n44BAdAe~4H
zIzMdO4V9F_QP+{3S0y$O`NA6LjS^f`n26bJK(py_2(kJ3`5yrR#~YepHadf988ruo
zQm`#Riokn1q?*SE1||RiB+6qGTvrT9gVO|_cKlGfG#P_tof?3V4274?1e){!@b32t
zI_I4O*sB+%Ul<P-t1zj9x(PL&_W(EX!;8#qEL7nS6(wa0F{gRVX}>T;773Jakue4r
zC%b9ntS?^t;&a;5ja&fA4cwM<ujPZs0mgV#PDMe1-&qLklN^$Unwpu{Zf)_Y!(79c
zeW<U#bFkXMLO$4Ozne3HM<2L~xbyKUI}?vB<iYA-95<=UP9x?_9nTTawB=-FVG)Z(
zg3}Rabvz3}nL5yqBUPvK<|XS$VC3SgQa2Y!c<n)1OQDW%WI!$tDT*GNW8nJx2-mtf
zXB7ZgyM3Sd<Hp&+fo*prt1!U7iYOOLK#(398Y+hbPa%rcz@L0|7_P}+Gp)MXHSho>
z6r94k-V*MRR8qZFmE9Fmk5BSSWii^K4x+J%+Z!|G;6IuB7MzkQ9-(CDzow*q?$8lR
zQ+4vkN4%B+gc`NZ$Km8a3^VojCO0>CT(Y%r0BJ73&0&!S%UzLr`WZPLd?+!dhG-Bb
z!alR0(7H;QX=G7RQ7CwLa7|amcV71u4Gl94y0R3r@ra02jqgRp#ce4=+lL^YkOw~j
z)}3Dv-~bFPtk7xmh-q`Unp%v(B)Gn?NRz1wD`c0tE&z0T{=A5T3{*&f@;nSGB{G(l
z1ynvjFPOw(>iKFHA{XqBrI~d6ER*Z2x}odZJoReC*7Y&ac3rIG0|Ava(%a8ZQA9+f
zQinE+1qwg~xI2_rN_W85Qcd8)&konq0o+^7hc2@v1{K7_exT5CEigsFDUILmtgrMZ
z|M>oW$JD{zY1x)%$(E=0T%eOqfvIva*t)!Q+|+nDD@;yKt^}AIxz?JUUzl+(ReHUE
zcbYaA+}>qRtU9uAbabqoynXAIXs`^J5$`?O!{{H)X`!*j2Z9iAqb3Tp-c>HpCb9fy
zaer+k<4>(80G<z5Z`WlLXu8M_L?&$vN((Y@<W+UBHZjw(F9T#=lMo?3iAJOk#G{YZ
zu6A3=e5N0hkZe|70AJO~*1dlPuAO0D0Ed`M#+<TzGIzS#K|B~lX|jLHl1lBqef@gS
zc!$go1$R0+f&q&+xp}1fvEWmq(<XYh!su15FTgClOE#Br<VNR5n-N`;Xrxr_RPhSq
zu5JL;FU{1MO>zJ*SgPbh%xf<JJaKRefJ19Qzgi*U;S?GP6wR4GRp<=Dq+?wR13DEG
z6jA@$3a<tv4dhU6rgzDFx(dIVftp(6<p<wD@>+ncRZ5K&!={j3k*uXQ$R0ThI(qt{
z-`=?V*O%uZX0f$WX8k-lsoVgPXh4^TVwAZW!=(`2O91%6)kToI9`p;JEk$UGks$4>
zAv+Mwhx3k(Q+?vKUKAPjK#6y+J;2Za*5NIHV22YX`9{N8iel2za1D((12e@zwlDw^
ze*b=tP0IHXfSUtTPM4FNM>e&49dvP>;Lt{~83hLeZxtWET{#1UXaIr*vEvZ<(wqGe
zIOJkfeB1jl0f1S-OXvW!OmF3)BD?``K-K@e*CbH44MOh4I?QB(P=W@S%l$EbGQm9M
zV*N~DZwPD$ii%21My2el)$OZmr*NYc73?-=I~-4!BKRY!firU2Uv_enHZNBN|1$s_
zOsUK19@yi)z6WY?Ja$=9AlT;rg4*OvAbA2vTCuSLt#71Giw~Qdn=`EzT0p>jM6da&
zcpUh~JW_sy;qIU2II}I_i`4DM^(dB&XKE@1PUg^iq58uSRTT69;S7j6K@e77Uynk8
z$-+4NS0@VqFUnM&Qc6apeEoW3+U<N30~2$<t3Oko%517aHW+Y>=|FOGU6Z1$baFH9
zrEmba%8Cjuz%V+JswyaiQcA^S{QC7)GKwwpMRop9@GW<2&pxfstP_$%`l(A^{EOj2
zhz=F^_J2l0Cl>|h92Ko2k5TUo9GYSC5BwiKKJwr(QiyV>wwI?SCMKegCb<#7cOcvb
zc`hi<b!Ra|T4{X&+z?m|Dyo5Hl(dpTYCtZXdjE)byGFh<f(icq{rj+~*%Z-`gArAW
zFehgq(GIL@3P2cPh;t%~nt7W0;<$KtDZp8+1V9%$sl^%@Nh~L7f(F@Zm#WI1T88bx
z6cpjDT&m!hlmhYM-XP%P=2Ibm-t)V=FNDq;maG*eKY=wg_?)Vubf1R=#>dA|67zr@
zLIsm38Vdlx;nl%(5U>pw>51E|4T*xtcwnya+c3a4Ha$)NcG7{^hfXP{um1YdB3HTi
z1NariPoEMPt5#kBSN;YSmRw(54g<q(jby=k2SHc+@x%F=g55t~y#qx*9{$AKNsQ`M
zhCubY3K{3Dkgrk8QjT-~{s=&o?9i19>UDq(gKKFeA`aJ5=bOAte!1Vnz`+@f<9B9s
zoK;SxJ+@H1i-9pPkSboP-3$t8x2jbNwLMTM41A8?Q`Qg3!U5qRoJYDl7D9!EsXc=r
zFy0(57Dk2VAm}XtYo=A^L3z3iI;`TA0^q7YtC)hH#)VXn@Hxr=E}@j`A*~6rCl*LB
z$C4WXpjfrrLIffk@H>uV;fCPLYd@4(0Gjdq*VEacS^5|FJD7$q%5lOJ%>m#zYZbgo
zgf!hwK@4(7svhVQp-G^+h2Nzab+Ac-7@zL>^Ebfx4gvd5A=>1LEw8Ja((H{oIorPp
zfq{Z$Sc6A|_~T7=@Ecb;l(z@qpuu$M{6q!tGmsp9h<Z)Idt2KQ0Q=LxIrvC|VsLj<
z{+jR4w4C642%3!*yg}V}RA2}`Yz07sAO8NJuiKR!&?YDw16-hzFzOd;qO_vYjO^}$
oZ&805JaoKohrB|MBv`vf%b0o}8yUIyH|HWMB=tV;?WeE*4@ZPVl>h($

diff --git a/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-46-1.png b/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-46-1.png
index 0b65a405a2d18a8f493708eb292fc786ad7a7aac..4dd846c7518a5e94a6531c3c7c532210021bfd50 100644
GIT binary patch
delta 92060
zcmXtfWmr{R*Dgwl2q*|hi*yLmogzr5AYIbk%~BDN?oMf>K{__w-QC^YaVF1uzJFZ0
zFIao5ImWna`gS0XwICONbA@>Snwl*hS{V+zS&<`z4+>lTktiB&^Lg%{<}7c*E+<F-
z8S4=16{3jX5pBK10g><~UnF_Pg%)Mp)jek@op5*;`46}q4`LFGIFEDHYzh{$b5-@b
zV)&QB&qzh4wj8({5)Plo$dI&7RibzZp#E|6eRv(0F&wKVS2^?bT8a}2_A(-pA)fCJ
z(S~_z6*1Zp7#c>8=HntXE*Tuig~n)Zb29G5OX^*UE19zC!_<_&)#A?IzKD`&+X~W-
zAchJ?XZ&C`fjw@X&9lSt_aFJ`OEo9%TdeJmevNE}{>?P3<=Eo8|B&p!y}(h8s<Zah
zoBiYZ$kkkVY^-`qAp#tPg!Yg&YQx{Xzk5)g-Xmus<-C1H@ei_74eQc`<9JZ!y&KZn
zzPJpn+f}kmn)?|Q#8<CPc$V7CprEsoZTxaH?)1_u@ybmicVYtDnCtfD*{LfLyf%T#
z>a#QCy~DyU61Ec>Xjo>=H}{{YoSxuMHtp67P9&X<l2e9kN$Sl)1<1J8O!$@CLVefo
z#e;b}u80<*T<?C~YM?W^A%EZSl@`(0Q=2D!PWtPmFz@T|MQl)5V)&Wy-i^tIv&ow4
z5szMWFSFm%)rf;0yN>z0d{ZYER)n(+b(7~2>wAI)<DqYGnj1b9rS%N3aJK(uA>g{6
zcq8};rZO+4uWTO<<^F~Eu3z9wYdO;Lte;Zwu(dDlWwdD-Okms%VH>+#323h@u4;W#
zXR(t^oU41ncmG~Ht8Xy{jksF1TwZ|t#RHZS?I?c4jm(=1&b(#yJnL4cnF&v(%P^NH
zwU1>+*^1dCwcm-|#3q_i{t;Jj%s-&pq0-3|rnHum!3gS-AuW~0r3JB<9wr?8o(@}Q
z%bm{yhpltPZF=Kam99Ff;GDY~Ysx8UowK>^?I<Xie5$q4@~YlS3S}xBt@|$M>zRrk
z`iDG2Rn<SNsd@Sy+0!HGa!2e4!(6oTiPaaK3{b6KFEOIYoxNWQ{}$7baioO$RYVml
zpdE}}3kePus4hw$d%B_F&3A!y>8gVabEmIr(}!0W>VlcehKJ89dUvJNG>l8`a91~M
zT}Qq-_HX%WsR$Mzy?1d~J;hwXXEx<VQ@5v}*nUdTU@6nP#@AMqI)Q~#8jlv?)hQ8_
zrC&sPO00C-IVhQpf$6%$IUYR!kC++ODPQ~P$Yd~~5$Y!!!d@+NUsB0byO6Im+<NiP
zh3m2sKM?9sk*N0^3@EcXd``~&E^_6DLrL&grpg$MIK4r`Um{a)p}7g62w%+8(y(Ob
z8|A0syPJ89DcDPOHzWR}dSW}&Jml9lN6PotXj7_NgB?WEk+urV*dP9KJIC4`A1e1F
z9TbMXk%Z{p7ITqN(hC+%l$gy6$`m~Od}<+6vXenItFx1DsUdKU9L$g$_~3{}nU%@d
z^mq!N$*YS0Q~`B$Zbdv;5R}k+@rB@`que!A-a`px*I%sQTtw`b?fPY*-VFP*?dHok
zGx^-HRxQ59Z4?L2!yUPI{pRFUeCYLBzU&-cP{!{O&_2)e1PVTN#-nVWx#A-^=x(Qd
zL~6?ziaqSgclft+gc;XWImw$&He={SbvmfP9MKGpCQau61>Ze>%*B0gJ59mV*^1zC
zslw((_f6f+RoXkVgLxYx<{BRlN<BS84TVD81C-cU)e^TA%wt<g?TTgjj1lZ&kG-dm
z5-pqbwTFokiD@~C^fjIMkwu;-!@yy8#+U3LfeF<(sFDQzmdWjZWlgP3?{LXyUx+3o
zA-VE#;h~5mFSIW$JhXuSdg)?``o-Oh>1^U-l?qLbNSjw=QJ?-s%_p6|J~3znLlA}{
z+Y9CKxr&<O#Pr~LpTm0&NYBu)sp|}i_<Cj{En^ig1xDOIuC2{8ZrkCNDu;6`bOyZk
zZCSHdb}kqVZ1^7>hW^Z?pMz5j+Mk5ED<1rC^qZ$a7VmCz;n-k>M?Uw^>WG;s=EBx}
z;TqE@2E%kB=8C|Qd8zNbRdjB3yQxisuj<lK#q`v5PIFIMs?XVB^o45-bUfq$T3Zx4
z<B?Xo_%>Dd__ah=uPnToPKrcKDJbFK-olBC2r0!kf&K)Co;AK=?~C37x8{>3WcEMG
z33ByQCz<{`DfH$Vhks{+4kOIC+}vY?W;{o|lLcf<6$pvhW^`8+`Zhi52u_<vo1<Ae
zoWdiLR86+Wk;->^pnL<;+w&G@^aHiFuCoZvYEMk-gEI?LLm`-ofZ^g!ijINBMd-VR
ziK~d4$e~&x^hG5fwK9t&opUr9!dWSyhKk|Q0*jeFM0fb@lNpQ^&IT)ZQZvezrG`(f
z54e~s|79)Et>ArFO_#6>LU}yBkyPS`kHa*iDnVrcZV{3%(~P@5${8d6UOZTaRPJPR
zv6Ou_fl_assBKg)yV<4#`VqVoC0cCLawMusT6@Tl6W%wNgPS2r;?qsG_8*6s4Nezl
z)Q$^_#r3ffaw?H@WizmTS_O*ng!Zn7-{!EGmba&RMl7d)mX!u;kj7BiAn(bJd>P*o
z)dVJl<9hD*`Eb5^eYVHB&qS)%=EUj~D1WQSFNFfbmg8qd8qUfEO=2S|m^OD$#V+!f
zdXF^8TSY_Tl7FU`A@6^=H6G5Qde42zSkZsdVtdkUV(`ZkFQPvhKgUG++?#b-zxn9!
zoYBpCt+j(ORMxOyV?AnwaIP%cj_N%04T(79D2k8XzwdmBF!--;1fl%EH_5A~1^uzT
zU#2yG{yFh^YHBJYBZGJMLtu1-{!#JA9Q%K2@2{?|FnC@Zz1(_W!j#J!OB=({^i|@$
z-`Bw4;%b|4%zqJ%nI>l@RZ?v8UQr+#T}uKwlFyvBvVvVp-Ri2BSak1jvKtR2A-#wE
zi8T{rvJqd@I_al3KPh(1S6Hvgw<M{piv2CN98Z&X8T$+_pRoI2vQ1>JHQls;`9fu0
zqyBdDZusWZbIaq@mCv=Fxcqvb)+p*Ik*alG0Y{_z&v<?rvndBV?F;0X(v*|qb@_6f
zJ0Eu`7%2(vR94-<kiY_qQu_%bF?~Ix#h5~o3`axLwy?uJ_mE83+1>q)r}j9~qPwrJ
zo{pcibwv!bPE;XR$(YqA=;hzQ`@5!9amNj`8s=%K;GPJn?2);%AJT=sRO&)PPvnj!
zey}t8&L-05TM$X+<Ep<7YDcIqTs-eoU3p;NLOn*kal<Txu&~xFLJL!ONjlG<n%idn
zAU;RLE45PQk>U!656-WTTWmL&?s#rAah!r-etxQ%u>JkpB)MEA(S;%me^I*|3(p@7
zmp(pno<w?jrfORQVaf@TGRjEo6~Q^QAI5ngGQoMqXU`$V&hkTNx~-E3n1Gm4*3KX1
zws*gB-h?l*g?oD=R#a4gH<%8^b+tBmgDQg892*<^+H(?DHmk0#uGy+WT3VW9i_u18
zYHfY}sg80#>_fQ=*Y3=*cJQ8RkGCRn=V!^Mk@)3c?hXC3VC3fIdqV;f^t!RPJ$lDR
z6PzDOouA|$82p|^94~5Xb^FWyjV4~Ti!$O{7Mi+Vk2x+!hJAh{5{Rc>rsL?2fV0jH
z+bH;eoQw4Od{YjsFu8%%WX`K%PK&(wNYNm;sIJ<TIzlO`r$kaYI4MD8Oahu)+t`SR
ziSdn#Q_Yk;tZ|*a*L)eHe%`+v#XrkSyTC<D-MKLsj7>OI?{h-xCNAv{ABj-uTPYwt
z!z*2z%($YRZBlK?0Z~z9e~43@3ifk8TrF#jXd17$bmOC=3&_u>7wp1~O-K;<UHn4t
z`kA{B)h}v-D^&7F9QO96;<jn{K!B1-#>HN(fT^uL6Q9yh`CFz47LIE)7$o_s%t7d2
z&QIsyO>#wRIQ|i>e`~41^=Y(?Xw;Zxbn`Ne!5sPR+u~|y$b(mcFp<yN=7N;;m{2u$
z=<Adl{+!C{p1w+Yp>D^1bJrbxB2zelip-Pl3^~wBv8vpm31Ncfs}Af%mtlQ$o*B9U
zg>_d;7fQLxGn<#)>OnO^a3Y$M@>!l>pR{W71Om*e#n0AF#%eQ@WGC{Owr<(NT`X7a
zSs!hkHbfhNh0I=JJpMJUxrka|FsE_sMLz563}e3+wNl?E4aOgs5~7ommx3-}dG<GN
z|6BjW<V|$o2c_blKizTq2wN6<Ah^bCAuCfS!XHL^!NyV#Dp-8)&!2PrRI&D)k#e-s
zXub-qcV{-^Yj|=NuqTkCsa;W!JTL|iQ%Hx-%ye2ih3mCC#y$o_-E*JPon42#1Ecli
z++(WFo#78M?#QCa3gu795SYwvRQ=>6mwc^m0_KdB7JrQQLa=Z)U1O#i%eEsyfKz?4
zlk{nH6u$IjmqX$VbJB=d<;ha$+jg0IFGJZgcG~vi^%WIdr=smvpA>SM<<~O&L4nZr
z#&yX90LSU#J5OR`yQw>mmC`L=qRZ@OB!+mYPc2%QXo$^V2^M8vDb~~dTNN?^5anL&
z5E1Ho5WW(E=XtaEM0z($d};r8pEtaC@w<4kT4lSVQP1|iulj=?w2uWlqRu5e7zRK6
z?ugVXGHUwgbmex15j(><;;{>{*96Xfce+b<qHub$J8L#vJdw@Iv&KCZk5108O^G$X
z4c_#bUgc2E7pK3M<L}P5J)32VIB5F^KT3)>I7H(+c;$S7Mag^c`n&QyFwmCAmnCF}
zMG&i8qo2C|R0;a15-MASyKtjDH}cM0^9<`G8-_~DR)G8<k|G<jgv$zT7OpaUkE*9C
z=c(&s(bi6D8d>S;moVzSxdVG#WlKLFjyB4lpD_q&@zNh-Xh%}OCV-tMNO30$K{G`t
zy@!qe3HX~@jKK)?#5g`)OqtFj$4dj$O0-eKDilW>x8i{R2*YEUHU<hBuh=`;;x^6^
zBDpp%iyr<X$Q3_whwf!2n!`^1gXtlt|69%2#uGl;Q#IE{^yZpdCtLD$Piv#Ag5#p#
ziRQIr_Ry^&!lGLvFK!}=`shmQMfICQ!$#mZi;byGZ2n~WGtn%wibdDMH|C{hsa#K|
zt>sFTPuuT2P`XUz?p`0M+FyMl*|@2$a%14|QdwMtR%I4qI&fyryQ|abWABMggHYx~
zt+JGFmMDTNloMiyPUEbM&$@Z-Yo2C7w}gH+VVV9VY4%$Wl&0lAIjB6G^-FmQ1$h(L
z#ZFC<p#OC@7KH(mqoeMoQa=z%x3<lD>Km@>V85l@KhZLw@>W{1W4<W*lDx?avkoHb
z!G+v&$?c2~g?Wsqw^X9N@kOtibgYhTORo&g1xlF}NIuv7kHP%>AA@-wo)pzT%(Viy
zd6^Og*dIY%p+sg@Tk3&L(a_`EG1smwo;(-YyBTsDq_!;!pW-bbI^qb%`<JvSSvz%X
zL3Mmm`iAuub(`J9$~rJ4U;=9I&2a4#ZPl6^7@wN!>iSIWLQ$U#&XCP2jBF@2_IT(t
z>Z|Gu_K3znQCYi!T&V0IYS~f#NVO8v`2L}zZu3LYRfI&j<tNp8XIcOXX`|vJcBpLh
z#+*-Ueqv6H7-pB)?ZW}cFPy0m5gp#SRwhO)*eXkmDrv@_-0&O=VzImAdU5}o)tubQ
zm0zj~A$vGyTRTLT#8Z2z1V8Pqc2C+=CG7K6BgxZzG;LLHg@CGZEEJ}t{2LEL?^f#n
z+UvvT|L920+nEXax9_=#VW~pSqvk5xXJhz^YIT=HaCErRqD6G~OqUa)&}03>qwxC{
z6lbizP1Qo*D7_adEz~rqk#S&6<8hjueNt%?s0T;01XvP8D=Ur7!|}4PneSGD9(oK1
zx_zyJ4Eu>*>{NHj6|VKck40tOqXB)^(%bv<0(QmdfM5RJkzhk+cWF58Bd7CNCJ;Q{
zU>ij|iC`-GXVh{1DN674!L>mdobjm9yR2wOC~=B63UI|o#BX17>o&cyVPIqMz?grP
z&+sanLuUg=urGTa;q;buZ4a67E+XBq9><+?jj@-WUe^X8kUv|hE~{2Hi=!5fy|p<Z
zo0J?IEMZFb%2$$fGsq;D9)z<cKQ|Ba1_#lnxuY83u@5U4V{^V%n&WUy#jO#izl6Z?
zYHfKNk^cyeu~$3y+29ZynHn53%d8Y<&3HfUj@@bSLHUznQ*8KH^EbCDx29}{fF)?m
z^2G$yZY-~kyhI)l>@7&@?E3tc>zJNK9v2Rvn8os;h2hKv0@033PHV>#LqO<yQ_j(A
zIo_{zLSnY&ZOBzumF?r1_z=D#YN<k+U#(+Rv2sy}_S^J^ul`U{7R5RAZNe08+ujv7
zN=;FC=Z2-yu^D#u5>)@=;XRr?$>m{q+5}?~ITa7QS@vJn?GvuCcvxeDx0m<}o~HAM
z9q;OLEXyYr#kPmSDybAKG>MaW$kc=DaEmjmc++(?-=A8b$6VY=L6_xroA%tJQQt6&
zI;7=pJf+m_V-NTVq$b<SKJ34nOb|{*<4pLWjLcMR%@2PZ#TY()=Ipn$vVqORd?&=?
zdgDY3&nv;uBljA^&Uo$f{ylj=;f#>n4VAuv_#@rmo&W#K63#-4A*P?Oxw#!NW?RT$
zyQj8k1auuVFts>GF;4p>SLu|P)jzuiLvQYYh3lU_V#%v)aeqz^m0H?{a^sORm+LxK
zBm$le=C+Hf;^I=>Rb0tUKI7Lro+;voKcrhV=0EnXJ(L5cjlk~d-|wNF{jsPi2Tf``
znWqF9E(NK+`*phPU3PVq9lDX1lfWp#K|fkRq%%Imi$;r$p6sy6vh)TFL~ECYr+SPB
z6w9YRwc|`|l}*`Of{(j30orL!;Zd=8WU2kWyP2M4!5H=e$>d|E?BB*osHQ5{FdXym
zyaTe)Q+X2IRR|=_c;X$m#bGM^0OU-4^#5lJ{;HSX^LWn4*(y$<;9>#4vZh*!COYW9
zwfKM2Ej0X%msWPmcppjBU-NX#y;)PZ&^K0>`d8;NmM`$Qf|?GbDcFy>kC>LeWl(@o
z`!mm%(Nb%y`}N35Z;>NbyL)}$Pxtq_6a82O*goux$VMED?sq$S`YP9AEm$j-f0&p$
zriW^!N~Wq88gPhh{d^{&PJ+gVLTR~eQ5CEEr(+2)K_G<XGR1Vs1c?G%H$0DpTqMX!
zxqfkKu%YGEdKmFbn}-Tz{&|+N+2-M?4nV!Jf&ab%>=c^#GAK8WONK$qS*6I!FL$a1
z{iNa8woF`B8o|lVVEA8Yj@}bNz1;%a-%^j{YI!LFq3ESB^i2K<0RC_v%BgQw`V6Bb
zNT%>iGPP7P6_%FvLVPQiu<B5}+go2Re7|<S97SksZAD-y8>WH*#I71=606SR)jhe<
zjqOyJz1usijk-odM}`fj*mI-3U>JQ~h0kew*Uz6xLGYnEiJF7mF?ck8Vp-Q$w>UBX
zg+bGF6($wrRHAU**<VxHU-2j{w5ET?eB3x9h}PD&$ZvVxSM+)qWYe%|6D7{|%ub{;
z@Uex3dL>l5oUw`@$>)sWCu-Cm0<mpHd9KHTIj3&cyq|cX4Uhxz-$8;2F#+iT$YW2p
zM*u$nAAGSd3BK0is*HplJ+~&mY59>A4B8)ksJHWo5>>jy7Dl(E1WMQFDrbKF9+(<g
zT4lkbmXr8F?T1x_n!G=bs~Zk>BYZ2h<8Cmc9tlr&u=&ci_X>H3n7mwQNO}Mj5K10t
z>?Bv!o7tG4dMw&<e2>zK-5;6M9+N!PQT2?EE%{qQ@c)0uVmoW7X?Qw#G*;5sj^zVe
zB~wjNYq8FED%aPR_9E0!Z$f7iQP0qh{e{J&(73_>(PgP%02M1VGb%2*d7ENn@^pgG
zwu~jfe4@$$L(64@08B^r9BtVYxNxjA@#g~Gl58tx36!(+bLFWMJ6s<2Jg|pLnh6Gk
z1=!H7IC!KZsMl;|^J@2*RNwyl^h2vLQ`SX5TF8BpK5(ADdrH3KNW!*LNvp1%HvC!K
z-{7jbXtKnSbh>W#rUJD3zirmg7%}%Cm9cn=^IT=K!<=O?@Z999KP2%a#5LsiEo2Yi
zR&E|&?5`Y={~JV6YSIzoe4JGtT$TOiP@3xrZs4|95#8I{Rc^TWfBUM*r2NbkMXHSh
zBlCaWYNJ+A4o)NDPyWo~+B|)kf=X(#N+NE!vz;<kyXMsHUY<W1L8Zy<5o$t$PEJ9C
zQ|^cNh2Sh*qJDp}MoCQHR7lyB)kXya)6o`<3d3S)_0Rjd%W?8#682Y7HJ9<J(vaH4
zDxf&OlD>Y)La_SUX_^VEkd2vCPKar#>uQ;*H=%($CoVs<;Dy^Tl`biE_?QqOs@97Q
z0L8oOV@uB@wrC<VTxQR8Ze9<6S@$z{O9ZI*Cb08lW^g@2-&A+DbL*0zqf;WdwRzZw
zwt<4`N@UGxxslS)^EzBXv{e`$+DrQ5$5$2SC7MZYG?4X2AhP!l5gli#8(Pc!Loo3`
zkl+_#ln6?TWW$k^g_a-Wn+LiY-(44VFyW-pszM&&2PrY)|5SO}&o{|M*48(+GFG?o
zrqU^&G8sJfItN=|Vae8BjqS2a+gs{S<drRZJQEwcKSe~f4(eDrL*rQV2X=(DwL+%+
z^tGS*r{84NEu>U-45Z&QR-#S!mw4>N-4B($UNt*H#TNm3y564BS+b+2keBg&vlPR@
zT^v@Pz|z1~#P4c{pfGKE8g{O+iW$_~i|m0afN>paVIkf@OCebcVQ+ARFLTc(xoU<#
z<VcoSoY<CKxr?Yk3`zXI0sI$Y?9Wmw+t^P!EP67!#~ux2S;9NJ*RNtOjsb;gxOd?I
zqn^GLBH*E#%~K*{KStW9>`Rkh7?7+NfXilge?m=`)|PIyas8?H&h!nCWUi~|F7my$
z>d|Ve8X@~)Zv44}%2<Ir&1Jo2XCR&H-B>N07*I4pwY<+=YkpL}iYP$LZd`j&_jkM%
z<~TB3#Q0cSd6X!d=W35l`#N(<u|8K7M$r|v!FhMVk*Os>#uT_1Me^jZn<rEILB4Y1
zU?Fm~upC9IAe}s5KW4VRWgbCP$;*$eNu=-A<iK`)vLCbZk)ErjG>8Z_MB{kfhn!nM
z47FCa7wAELTtAY;hTU#;^NE0<d0~}*U4D78qSY3{aOeA{Pkdu=cxm(Of{{f&^0tm1
zNV7<dCoPnEmbzcQsn=@_U}!fqWsV@*oLdN|_%$h6-8a{F%|X`3CZ8`ZoaBH5@-2&s
z#l>8EID91_Xe-GN24`(}qJ-HZECH1Z!2R)!w%yq^WCZ~0ZZ}4uh><o1C>B>X$JVE&
zI__|FDtGr?b3F8;hYHxS#Lw6%^EJ7>h<5y8=%%A+3Zm;&t4+7X!^m;b7<QI$hsQh-
zl7ue*RqrS#)QRoCVJ@Gmbh{Ce!c<wjygEHlRciI4X~z&)1}itv{AxzW$X(phsL5!>
zQ?deg@Q85&XGC(jiz9Hx#zvRjvRq9$rknPNy;pkF({+jO61LA=*FD(+XLqIW&+ZZ%
z`^lm#@EeC}`87o?8(Vw*1w5|<fA8S>TRaRy4b6Rd_`0NdT-O`Mr!`N#D$@J<bOvM@
zL*ICYa|cZi-^D>oZ509)PDD4A_qb+aRyD!1zoNW!v6^$oDE|?j@{Pswi3v+_N;W*?
zOrD=5IAqB8SiUy4n`Fvnq><l1n?@IG-9jVOIHrX2n!m@S+fhi=mfu=6JBdAM+u3}6
zyfkMzSn`@|bZ4sBh)t+KuPHDL2(-8}w?NKl8J|aCZ-tsk!X&y7KW418Y{ZaK#zYe#
z5B?nyY|(%H-_TtB|3fp?6iD#bc?<G+Y=j!uILI+|;xeupLv)*Gv%DqYXxbc9T*riG
z2~K*Gz6%0X-|76m1Y`rMhF(9H8uxLFXLBFvJ>loyK{`F%DgNe;<_Fw`$pSv<PVo59
z*;W2%g3YmsNIHMYpNXx@m2AJZigG*ar3lrX5&)!^`?<_-3aJ1j3DBh4pC)COHu=~Z
z7v$`Gb8>Y0B{mj^uR2mId*z}!1#Zyvke7Z>QZ-vCGp#T5DpsAV)NN!KGh!SIwh-us
zEl{fPZt6odV$4OKSauI%X+7ZaSWsSj*@~|xsFm^9l`0ZCk#^0SPnl)<bBwK`S}LY{
z=eCR<$U(cQDLgJ*s`+18t0%Jn;JS2fH<jw6*Rerx_ds0U7YHLrCI9ZFxL9vz&7ZVg
zj?c0kkV!Fq^XOIna*q6rrUhC!?PvKe=XY_tN~XJ&jjB?Jlq2}g&k7p(;XJYPBPfsL
zQ9sLWf9Z7h0pQSZ_njATNjDAshc*H@34?BgTlR>%S2rATd5Z>CD?Iqi@M4OSWhOKp
zTKFCdl(ifTPf~3b1_Ive$|Z6qEv)QDZxJrkCikAc&bJ7V`Fw*kYIGf*?#`0M7=~L>
zwq90q$+{UN+rrNY=5Kp%cUa=7bGT&^CH*%vV4%RsRf@au$2VZ|_%4fy`e?ma9}fzs
zQZn<Az0(ri3jsQWLv0Ynw>^<_3)S`>GJlz>+%Z#64Tm@sE2q`vTzEn??nFSK73;4~
z01fqx^1=niB%a?A(B#^KR_Fl1M2PaAXAE8r!f0TDXhr3#XRAZrDEQo*WzrUtMti8=
zFM{LwF`edGUt*4O@5*q=Gs}H2CF3s4niFxFTR4==IgZXiGrZiI;$vLso&pjakpGQN
zV_>?v2Kw%ak2QZ875YED06Eq3M5b5O<wq1u%CP5e$}_%yzkqVIKT`on{|is2=TMNE
z1N2TT&Rgr5@-~dU7(QFi)H#~G5r4L?+VhGW=pCH3U#fT3+%1R!+G=U-eSYA|7c?K}
z7l%EM=4!KdXqCua8pN)d!dOucU`Ve9PIHz3%xG*m_g9LV*k!K6g~iV9dS7ZKEcIg%
zX&YtT{xkk!l<#0YvQ>??-YG$5=k%mzRG98MiX|voaOHj|3-q%C!R4r`w{yN39oj3s
z3yPd47m+s=PDKhy7nJX_ELR*wJE&+w{;fKLCPHE`l<OG34BnOg^Mfdkv`d5A<9Cg;
zKaBtB(}bft?TxQDGcVY0kBF+>Fm+{@J!V|L6m<2FHDrC+FiPoM$OZTtoY33<?`H}U
zCeGrWFJ07e-l7VlTdV_M0<dLRLWH+h0P>^Gy_-9)ls9zfCnGH4GsmDVR~wkVt0YK7
z@Vk$KdwH7=Wxn<Dv#njJ^@){tdvB>!g6)Y}S9uHU6}D$v>iIg3GVjSWrz-}2+tf-e
zr-wN*3ezj-TLzW>!KQzLiUYu2;(I;?XcD{guduVs7siUb<!@|!>)qb>((m~_ul=c2
z#IfnRI}E+iyv86~J;Y&WOz9~g!0nQf6)G6g(ck$T<obS7;L^h#9zT+XUA|i6lg6i_
z3Azh2Rmftj;zy_wped2Di-?c>eC<_)%X+y@A_U%^U6b2q{}6II7BplkCtmlb#?$L2
zrU>jYC}WR1_~`3@MCIAh_o+#);ld6{V8{-)pl2#bNC$qw>1KAlK&Q&A`y`<!6*fcT
zkTiW$H6kXCC32S#S4|;3bDn0RzS2!|JCk%rf5M-u^mW=T6yGjxx0f~Fd>2it91_nX
zW;Gd0H=n@`35?{q$kT6|FKR3KSDFd@S7w_>o0By8>;C|jNT!~pNdf)pX7{|R;}bZX
z=!;;ahO(sF*<NNV(S!*3RmT^vuv~(6aD_ih1h)5$!~k&=IC1n20&6w0)p$8dw7u{1
zCF(7~^1qC=R>0H;4i#i=d)7SFcjHGKr!@r@$Mb<Ozg#?-LM_f}WtuatQfI*nGlTG`
zI8h~MYM@tso#9&EQ-=pU+?Oc&zh~3=zh_%s-Oi*JS&OVM+f1yzsT*({UM`GsVc9~N
z2$+?TUruj{y*c~9ki`Aj((;_INPTK@<e#+cJaoTm1lZ1t?GqV1|9@3xoHy)joY7Fv
zKp!~|8xoJ03M-{gNH4FQITeU}P^}K)l(LNMyY_XB1+2@S{h0_LJ32yT<!0CAt=E$6
z{L{Dl^#YX&!-|6AW}oq?#;k|&d;_D07NNJ%7(UXOs&k*1%I@8DvB`<;WP>9qa>ZR6
zl*!^dxU(-~d^2GpvC}^=yi)xNX`$%9A+q+#U;a@RN>j)kDGp-x|G<fOYOqP22iKpe
zZ=fGyC3xqiRZ;HL4va+*g2`yZSLUgt4m09_e`zc@C-?o~#{5~!jq=9n5D^f8=LqAu
z1l^2`%`wiiz8PecLOUrhwpM!~f9(eLoP|bFS(>B78NJ!ySsVu&WLM1}O;~^0I)1Ti
zC0V-jr=cQOGE(s{!aucGqM<V0dv-rgP9aj_a5fO=TEZ_N;~9{+TN=WQ;}ZZtuwM($
z;aQFr;Z%9Nh9>)Hy+^Pb*UuDel#*BGz%p=3DQIcNBO6+eO8j?LsD%uvBjZ+uKEb<c
zN6tYG&W3wAc}cT!lMB~$!@;a4D2*cpg*=KHl_S&_CgAB!F$xVeYMp;Bc)l5!@wl5T
z*F9=MSACqgvTyMIb47kjwDk`~0dQlxwCc$jnn9i#SVDn*e8y@vlh-Mq14cE~@--sE
z{(Tq@qo>g*WQb`=Om*MRq`~)Bq24fEZ)SaCgXui8T|@(xD)o763_JPkm#MOotX$>@
z9ZK1QsDd@GH;AgNZddEOGrM_|+;Ldk0OY7>iNKgWRi+3i5{qdH!+t!fAP}=CD(Y>|
zpDolagp6iz9<us2@ej7YqGxmBkQ!<pcRqzal|UYSPlX3aHbxHE7kAe!lmeTrfq1FA
z?F-*2CHcfk^38GqxkKLr+c+5$Mn!A7`m44q*t+%EEmQk|ny*)cTbM`Q82-=cGu@4A
zZCypeI-S+cne?ugDMt1#;r^wrM@pV4H-5}v8mL-<+E@RVB@lbFis)c|(4c+R2nEr9
zef8<O5MkY&DbfsA_QS5%Lqylye}hM`lLsACR#=I9ppIQe_3LY`uQ6px;r|ZBUIS0!
zNs*_<w=ylpI^Sw<?3nXjGKy=xT#odSHM`xS5^ts$C7Au~AUimD^RErZy)o|oT*Ja^
z;Y_hd%e=eFbv58(|8g}z@Q_`Dd;R3S#ctx}eXsSg9Sc{<b%eqJ#v`r-3p*IVrfGYR
zhRKSdC6y~%C0UB<nGrM`_FoJP7nDv<2&JM{7t(y>2BLvKcQlrxwffEA4R-biDWED!
z5s#>g@1q}nM?_2YI;jU~G9wn}YJXc|vnlJfS^2EMJ(cCpp#QxDlrZC<PeQ_4TVB0m
zP0iMDk0|W=jIJ{}>g00I_h_?E>Cgo2@^3UV^WDNl-t)t<pCRx%u5TyfjxUZ$9swf;
znfYU#cz^s4t^StC7T-x`HJw6xXQy+`Mos;xz(<3~-34*@pZ(Sa63`f^OLM)-kU@+=
z42ntV=;!hP!v6<%9#_7Odwj{=F?gl3A~BGn^S5X)->3Se^#aKZt+z0b+!}4D3$dXP
z(G<IO{Goe@&n&>={<hd}a+=2uNY)zDECYk}Pu@~ozq`RfM_MeGXZgUyMptSfB*Z_$
zkDo)1o^teS$WzC(QC$*QkpgABF}OGnGWz;qrCy<p`bBnMk3lywd}mj;5@zK?E)qUG
zrMw-JmlZr}L|p*vrg(CFer#t@e{n{_&qq8)l4B`e24g|X>(G3gZBJ(~=LmVJJxY;2
zmPvt2E5eL!?ahm!L3Bf;pHV(h0S!y7O)xrkxyK?s`@b4}&QxGp-+>IOT)y~vi63rJ
zd}c{)Fw$TM<io%n1_846Vhk9gP1KK%9V1}vXS8{j+SsCoTP3}>0UJl^{Eqoo0MCYw
z#ZD&(I0YmH(lavpmcs^=?vLc*oYf=i7#~I_iM13Z@Op37_w*8`fz|`86k}FO!g~({
zstHdY8I|Lb0A2_d1k3(M%E%K1vWpo8*+%{rFtZlDX5j%^GW>;n*Mpq)<=7k@J?UZ&
zgNt^G**e>r*L0{KEmXNeVt>mS>tTF!m#8eQ#-|)ZphHtv?kD5EH7)7$4UQ8BF4U3)
zWgtILt|_d@$Ve!)eCqwGs%{ROB4HAQh!3gsclY}C5aQK`d|FoOwG5Mn=nvx<-zgq0
zUU4VAk|li{YQmBKQ)ZvZ(@b$y9kzwc9D$iBtguwBcqq17??=Y>GP!_v`R>}JDqegK
zsUYrhj<UD+z2t1LTEWs?*8@e4hrsYkQQ3Y^`y(oDEq*z;z-P<J!{!5OSG*m8uL1EK
z^T-r_xm@np1C$20Hb?+U7HMFY#haTLF8Svxd=yAk6(1J=TC#I#nMyuy;vYOo-&gPI
zOtx%x&$gNY0_(lbgH?m<`7xR79;)?YEysSM#tL9$^+{w6NT%c6?nh#_s5(*U-YI`+
z5%s=Ls7HBmEL=2E<~fxQGzR}nXkzEVCfHpv<~yuq>P7Mi@RM~~QXW*y6%|=fL5}W6
zjr4TW8NKrgYcGaZ@va+{WMH4SK)OyR4QKLC0C?S*tdP%T+Q$MJtaSevaXEY=aE|>b
zfU&~u(vA1j!mr9RIl{M_ee2I7RuB=VH4iWU`Sa=d?o>;A5CIAb3Ms_RtrnM<il9=2
zBfu^A-y%WBu!6GQ^EWRa@auHcXb&<5e{XxWXjw>}V(<EUK#|h-sZ?+87R$HSkCa!D
z#m=izU)P|e3NKIf;zla^;g}5&yp7|AtFMzadMmpD?V~^QBaOjZz%CpC%y$0PI9Lcv
zm}tIQL3MM>5Yk9PYm`FM)L4Rr-QG=LJYuoQbr4))_BeXwnauqYmFl9*YDs6*6w7?>
zY4JZhGDe=D(_cHrWg~Fbwgt_=WgP*SkirKO&oQnuAto}+{|8cZd{v5NVe#9<Mfe12
z>Q1rruq&viC#Z)?Jg^qA+#>S2xP5yIN%9?MgA>GO<;j^n`KrPZok1$THVKB~-c++-
z^hXm9jV`^Z>$OH-6ysCOlSl-{RE4~)Q77t;9lg!d<4u=udZtS1cOtzl3%iZ&^mF8b
z=e+HM_ta%aD(bu6^FN;5-%wtt(~wG`#-ESA=ptfbDk2w;u*Uf(@H3f~N3KrYT<B+@
zar(u}d$jJA<&p=2t{noolywl=@>sOin3cSuzP147_G4q(!6v3ET=_fB3pTc^(}84y
zTVVqPYL{Am`+c6s?~Ut77fCO%{?N0&*T>9@<GbYu--!`^VX6hvn%V;kJ-IFFQ?&+o
zROq+Tro9vj`hc!7F=gt?;i^IQgn}ClAJOoSURbZ~zB2fef1|JewA|veK+0$P%P$81
zDyVURZKQw`KOIv*Mx}UI_HM})!Ox~rq7EhnLVPZ*-&}clep=)~>z55^vsJbdahHee
z-~`22?tGlMw_0Ao{aSHJ+qlN!8|8gmrTqisP@|aMf%+`pH@CE)K+m4Rz2)SLMMCpW
zb^~rzH(ygie7r;irF3U+Fa53RNS3t8^go*Qb)S%oSC8ep%%&YPWwGy-@&PkO%aXkd
zWeC^+Ai*NRrdCN_ao+!ekUJR+&4`$Uj~Y@4({+T2bANS=4!$X+Tp_9z%h4I2bT>;+
z4ea*Xd#T9hLQ1Tbb6)KQZa^RJ>i$f@#)VF2#<NSP36z#hbTS%#3DrYt-<5q!^o<d5
zncvX}R$R0fcJ_P$tG|rBV1hP|dCh=ZejA+84ln8><06d*nH1m^FjZ2w{9Z;*J)EqL
zIhy2(`{C!mihy_e?$^t;wdJ;ee2wbxy_rh8{aJ{sw5-g=+PZOJA%#*pelSa#K&`@D
zeC+QRI%&-d|BQ*9A=7dSj@Kdewz7x)Nj#Mws&_UvI-0zYj}AYSIz~PAb*o2gocgj>
zTTydmIulb{SNU@|O3L<ni{UMMiTBzg!&C|fx>*f#*)|dExk|Jk8P<KvS*PsgcKXkR
z3v#z+{&G&k$e6wB;U_#@$Fq?`m^EAFF^7GwDnXfLSmk;h%@>TEad^#6w{lkSRBFuD
z0+m#R>$9xv#??lOQBSnnPJBme>q{)G7VuDUEJo;X(Crbj7ZRH5pziad!AQ5C7e%mp
zS`Dpg+e~t&#{lV(h+^JY4jaDx)>SGrHkV{AIo=%hjWh1KjZ0`8hGW*E33?w<<U{^_
z*c%xIQ#AFjY66S(RSD@QdlydPYajGh&-w8bdM<SVX<WW+I7>vEmv)!pjN78&<aS#G
zk}0Ky>u|;w4ElWY?z*Jie0v$<X<@)kT4!ry)FgPv+0m<<FXdpWRN#mzLVxv@oP#lY
zmI&txZ8Rr0^!taVywo`uXgm6|UXa9;qQ0$#qe|&^0bjZ`{T(A?NNVa^2F;q&;*YuZ
z?j9a!@dpyT_u$aFt#wB=uU((*#5n!e<u5}B3Dm`g6)+ctky!tEDZRfXMoG{keOR#S
zE?c56ey=6I&wqdj&j&SWplw7Bzrb)~&2T!KaZnr}(4t*#I6kvxf)DbeN#=iIfRRr~
zK_zkmvMjq_Z+iI1vdFBomS_F=b~Rwn%y`6-JuCqEa>r0)2~e=2V&q6bgWNlSH`Lkv
zxxTUqkF$7>ab4EUw-A1Q)w`0vqp7RD)zjm4hZc2XWT~1NB7bA&V9kbQCuyd^`xk#<
zfUI?fF%)>CQTiKZUp2RHYCK@qN=5DtKHe{v<&Q^X1+XcdKr%0mNgwI%=Axsg2Neqo
zD-+?1GEPZJNg|h>PY?lzp04~gtV@`;8x&?|Z%^9Tm>LdWnv2)}e|d@Dc?<bi$WHXW
zb0`u}n&Ys6!D*g1zH5~bs#uOG3OoMiK3}sqYihi6%I4)~YY!2oG!Mwr0(adX=TfH8
z)b%YkKCf@RHV{aZ3fkHHM*D45vFMWueSvuoM<+8UNokA;2^u*cg}ivF)7f_%^hFDl
zaeN9Phz$#!GNM|*k75C8{Xx`&_pql12E!YuE(vgP0EV-rgtHWc0L5BCc%rOh^*j#o
zWf+=RIL%~OBf!C&b!ONr4VoVT#t7J}1@3dK`qwbGcZkC$P2-)FcVgkBDq<@#b3Ntb
z4hG}5x%`0Y;hvK>j##!k+243tj2FsseWC;;Elah`5St-|(`Grj=@)I(a{`Efz}(#Y
z@Z!R#@%s}DZV^)A8e8gahf1Rqj|yupzWTs)?nlrFu3Ug^D=fZf;pbbj5tU&~h5L2*
zji|CpxNq{h1ZMw?u=r^Z#&>VDs8=wK(#51QYX%}LEX5iO_nSZaA)@-i8}mg%KZreH
z1{zeZk*jQjnF~8>IO?iN5Pwb5!2QT4@%?jqugsvOB-6oOp}6ASaS`!TNBbWrY&xFs
zIr=T-s7>y-YCR9qgI|9BntaCZS&$h%TVRBWO3;|img-eD3@h*}tzSn!)N=Sb{j8=Y
zJpOe@>k?7atE{#UER`9WAu*A!svuE(rw`9h4-;%XYflS!H*(U(tf28%90R7-v}iKf
zcoX6x?7N>Efg!4#luR>m33h~dcuI=w*ROBj`r_H>Mk1D}WRqKadtb1yu)HISf(6Ae
zXzJ}wmMG;ZqnVkRfx+2zu>fgvRz{MRmX?04Fvo*OMsFP%!IG4eG@U5y`~8ZH)AP2f
z>F%46QJG%bb0jo;NTl3!>;ne}ILHCXd~U}r7y_@`Z)KO3mb`E5*L&lh2Z32tYS@Pl
zKKm;rg=BqweR5`Iq{>Er-gW0ayZNNt-xbPkm1-(yvA{R-Rdo(K<H!42k2q>MOfgf<
z!gS0Gg}YuMS|<m?oe(aVBGCP;;yaB)Y69K2RrtX@;6KFT$v;o6Lvz2wqA%EsWXAH1
zkX&dQepVIlr@cq{Z|m6o3fcHqtolSO|2~&JK+9?Pfk*JQn4%5WYR#2D2K(b$tJJ0L
zbl|S0>-1EH*4I9)-Q5F|fzI!K>&K#^Fb$lW;$<Vr{3(TYLdR)HG6lo7VM$)y;rNEc
z)bB4$P`B~cu-Z%Xr72**L<66Z;qE$w8cUHyi}<MgI0SFvZDyoHpq;}9ir4n8$u4L5
z=DUgA0-phl5zZNVkW`OLh}jdztk()r&ie#%*dtqCi*=qjY>(;!jQPvkJ1i=yZ75v~
zkInS8mX=nc0MyXR&tW<HJ1>u}sP6cAo(fb%NlEFGm`KQ=-OyX-a`No?^E9i42JnJ#
zX9mDHwXv~T0()j-K;VI#o<6XyjyEkmeX7&|^`QQOcz%BVf?V#QS{576$ljjyOP=yt
zoKf-;`0Ofxx2tPwVbRg;xr*73&*X7)p6Pb6PeIHg&-6H!R#sZ3%gvrZ&N~H_ze7S8
z52PPonlP`e25()IclO+38*M7H^(QRsz;iDoM>tNUaKFM`LJeo5=BEba`;l^-raIv&
z7`NDnJ?~12OASYHCFOE^?%wChKYGIXr%KN4Sl2PjoC$xI$e<j9ma&r#5SQD~OR&1$
zK`zvO{zFim!o*B(5o${n&dr{gND5@lWEHVVoU6m*-xpg8@MbK+PhK5^m$YiyHqPH8
zbUrBil%wRMRUG1*?OG?Y!!RJE!QgI-g6S_ijoaPnJHYtSm=c7)G9MN(XdjuTb1!fO
z@Yzfe8o^pd!N6dQ@dSHW44m46ciOU&lIb=p?EsWQxKoBhX`fEFMtt-0^Ytprbz6|Q
zo%RD|vTF}^>^Ax{wPBtV=lc1j?YR>>Tf>>COiWDJm?_jGqOgSE;9!Ph7+8puvdK~s
z63-v&?@Qvj2mp@Ar%%|mS~eWP|KXE#S|B;0?|sb;Y<BONI9%N)qNIJJNi<43P7fu?
zC})L*1jUpA4d3XnR`G2%9DnT#fq8n*fBMc9;kVZmgn(o4U#>I$y`FH>Il`ejZMFeG
z8wg?D&vu5H6<cz);pnw5v4J;u!`{=^r8icj7-^<s@6Zxa`@q_ZODO%AX5uJHWT-F!
zNY~~3OqP^|q+R<bqOOGWJd~y}@Kctd`JnPc<-g?>lf1)xwe(Xa{$+6C;=tpc4wTe;
zQ0k2hPDLw?0o}eB#;&Y_MDDMQWzBkY5?O6;5VV00QU~){OxyFyV98cc5N@SCh>V;(
zLkeXG9KEz5<IA-e?a$)kLfIo>2?>(c*4Das&kB@#2LF6KsP%-teIvK3J3@j(M@Oeu
zE&Y{~lhZuZ1j6u%oHjp$iFlq`Umofz=YO@46##v+IOnwR{`TVg6L?OKYgX_n5FI-{
zIT4qV`ZiIhS$DY{kU2ka{F?=|6e=zQoE*&4<ENLGxU8&L_6YX5&B?m7=gl4mKC7|Z
z_M1;vA6&!PAZ_ghOyBSDtE;Pr^KJ)tvqf6ktqGL>@fWdCkUojJ;Jo9D6t4|jvOtmL
z3ajS`Q}nH*O-H5sVWdUio0rXK2lI@WEoVA(B4n*l5hdhFphYRE;^ho`x-lYz0W$XP
zy!vcUe)JJLUV5XO-$<&y+02#tWfTZ!X1pXgJp?&}i(pwmF1lcSjHlq4h<AT89m@}#
z3HHWEa&AtuM&JJ`Qyz~7l%C_?Mc^vQ%0})%s^%-S9sNJKUJ5r8Q?LKIg6Q+&>#Eo%
z-$A?{Y-um6??>h6Bn~*y<L*Z*=#S`!Xvg!1>O@q{(0ih$0sLL*$ZsxH%&&-<sv$U5
zs|nvB?R%f7pu3Or@St&cADaOTtrJk}KVznG>=yWdg}^tCj6@V_)()4Osc<`N|B8u;
zX}U}1bvZ7KZ!k18WVM)%#G;ZJ%vWXUPvmMDFHraU{kySaW1kVCl8S}<o}NxNQKa3`
z)U;GM@t&UEcsK)v$8k5+2pUK>pQ#7}BP$VC%Xz8L5u6@6jbt`69R2pd>K%t5QLl&y
zEFvC9da34UQJ2=Mfzp>|6NPO90|&|c0DvByovlTw5bujL%9eH?@h1M)Z~D6+=h7`~
z)=}y(Dw?~7XPXU^SkB$|x-_w%=1YN)?SeAFjWnS*QfOZw(^v+-?{-_9oHZ<$^|`mj
z+X|Ty`^^Fa@_0L$6%6-;-jNgKgUYG`tBZzS@lN|U+$ZB&nLFP(m|6(YDVlCSWY)D;
zPDFN%G8itdZny2|8=c%8+sWF4Oao}!`M9c3ngDbKm<yFs*Hw4S8-dyHT`EgA1jXz~
zdbTLf%=@9%g9pRgUWf2mY~kOw>~81gGKri*m6mh+*DheBOU5u594<B~<tlhP?oGeK
z$5+#X0Gobq+sv-0sOWjM7Sjw_%~b3aPL!*?2Omb%X%ED^AeVj+FHk<2GD>-3eRaH6
z8=hHRT}`juK=@rW2%pLIbQ5!hA?VnDq@x2F3KGRjwFVf%Hb_2_Bwg%~nAf?Ocjd8+
zznsn6MCahjyDwiQI9<~0bbOkl_RZ;z0m`M7$0__rF{r0{k?)r#kC?46D<x$XXN|K$
znHDMtMf_zwb9@j7*4IX-0z9!nLStrzozkTpv{P!4?N1T=H>wJ46_}F0Wj9zLpf|7V
z9%(+x?TKV8b|9t(^7PpE<q1mNcBjQ-mV@z(8+#zx6g?tVfrb4_!qk=*+_Q1Mz8rd+
zgIY-Zg~T!T(IYbh2THxtl7fuP3mil1+e@4CgM|S=7*VmpZ!Qk9!5255uZtt*bA3HE
z5lZ?|Vs@N}l$6vPGC&s;r(MWZyh(Yu^$jBvoBvXI-hZ4#HH!ND`EzZ_&z~Eo)m$)M
zVq<@~JYmvrk94cQ0hODbO<eH235wBj^%L?!VzOJ8(sEhDeq^p-0QUFyui;mI2p&S2
zU++tZF(P~{=F1QYU=gWFJe`b>Y&hSCO(i@Hjn{C2s7lz_%2I81h*>Q)PksM(izdW;
zw3Jbvil~J{Z%ax47l~ld)W2D|^c-8qO4v~NIU?c}xCwP(WN0Z4q!#1p3Y956=cinp
z)g$*?DxF<jG`BkD24C74Eg#iG|DYf=z_YW#@^?RMLIzhSgg<_a<8xy_S?_~Sbvl@D
zg*rM`Hk}`+4F4*>znA1HSCa-`;Lo2w8nuq#P~CGuk;-{&Iqq9TE-nuV|5(`BE3B)*
zNkqEP%{lhR(iuz@Vl^H8y)}|uX1}EXW-Y3=oui{+Z!FUb0`?b9P8X%f41o&mPSGGs
zWCJq%!i)%5xDH<6TbY{^*&O9aK;X7aY<yIUuk}&q;Le$dhJY-%^#Z&>kd<kkDtOMe
zUEbI|y|3$Q6qz5WRG7#kq2T_ynoE_Fvvx(Z03Arukvo+aSJS+}ixvee7z|J#hp+5^
zA$n*u%#?V>=ch+zRqWKmoYq_z9+MCR1~#}aoFyM%nIp;6ET0J}=vu4@iW-(Ozfx3{
zvGQT&iVo7;A~~4Yu?J#~Kfs-zrTtrm+tH0UtW3wvepI)m$E8Y0ioXS0ID9#sx^rTA
zaaZWegRAp3LgZf(AX-ZG^$r=n1~FUmN~#ZKCG`(nr4{ve&fg55R7KjT;~{=ko9eEo
z2l&`8OCdK0l21nxxW6R(@>fZ>`4`IT*RPA+uWWrVNVHbDeNNp39tgM8a2^u9{?>^d
z2w%0ih$V45T%Jv6t5#W)zIydaqsscnd-aN9<DqwyQn8J{P_Vw%IWrYCT$@%~s=UdX
zT|kqGXU#o)cfFlgL~Bk2Zqw0`vj2Qk<Uz_YBb#y_I;ruFLIZl^F$mlr7HVyfick|j
zmU{esLfd|j<nIvl+6`dp(V)dUEyqo?%NLpWB9o-PdW@L@lFjMnj*EzR(f{pW{Y#Kc
zy>|LQkA8P#UGQ^Qv%j~to&5p#qhIPWH{BmLo`m%HH>}T2*<5)-2$EzvKnPD2f|;|&
zZFODLHQZlV0c@2CGN*y}+Z?+@b1$|h#?oo=RKnkKBiGRfVtn8wkfWqwQBt$z&BJ*R
zaJwMV9=SrczO}PDPc9Am{fg&TE4Aqu?UzsLpN32eDh$nTsmuS0mP>p=&OD#bGd*~j
zhJu30^YOeNTr7kMpp!<~s60!CE_NKWW%fP#?HV=`-6$^Hl21Gu);8zI$f;U+&uaW~
zxzOSrQ#%^{3<?=6bw^Fvj6%%yZ~k6sRNFs2!>YvRUsmmkah&7NQY_Mqjg5A%GhoK+
zFI@9$up`2>YZ?-{fFnsFt85;Szqh!pEK5!E-k)(v{l3eaNN0Ah^%6k{dig6x^WpyP
zgbIhp^VT^GWqzj8YGrRv^P2$h*ubWH!poK$8_=@?dk7}v5)Qz5Z|{HBHB<-)8WK!#
zUe4w|CM@h#6z!KZu(#~DNBs~{aXvuz?(M-ulHk{x0Bl>*p-fGL1k|%cE*5CAGOJ5X
zbNm&;(ptK5-c$u-%a9i%^-d+sOUm+BtrZ<%-L{_PUf?^HxYf-MX+^}#pcI5NuxN&Z
zkko~ZvbN34n8B@&>R)+123N?TFLa0HNMeee{a%}d#|#-S-AzfFvd>>#x^nQ}xK6DM
z*%(N<n!Xc|*&3)$=gFYT4{BOKneA)reTIo`O_5Z!!=Vb*+CK+oR_kFG3~MkgPkBZ@
z_xNkM@jG>u{!f|R@cLY=|6(x3+1o?^{>qVJ77phsKI1^&Qxhr*p&HU}{H4;czn;|j
zffC*H=#tOK>;>9xV;`ZEfx<E)E1H24C9dp&k&c@a{~Pv*eG*S_ZTjr#3?{bCD;X*w
zC#N|t-lS8g80Lu2_TV;iaLaFnFcr5o+_@CbF4rePOBNc2d{Nq2AKD$)qRwUKOlMJp
zz8F!2qidQawp{X>qIFLHzV!kHm7kJN*Xk)6Vbj=)K!ecK?}M!Sb9LRvi29S&>&T)q
z+MoKa`y_X3Ed|wSn%pQ+xJ{$07?n^qa=<^JTWUj;<z|H)zbr8-6**E1_C22ceEY0f
zQZ(0{_xn2*ma<OAJmrEkgP!P54aAt3;V=!D=iSpH7Z;Z`#v;`+(T&#J>};d^V+?`Y
zMr0Cqox!j=AOA9P+v-I5bWtY9-LG&JmUG>mVPuj9>cT@Jq9QN<A5B*o7G>A1kr0)X
zZdAHEq(kYJ?nXem^AQjb>F$ye>6Y#;X^<}I?mipe@A$_{W@Kje>{#or{e@LHr*E~J
z=I(9z4O2+SAoOGGh;F-*G&u2W3}M#6VYK#o_>GA3upyY^a}CZxM<>+=)|ne$I7AZ5
zu7+m{wYlZ4Geg-<>6j=)9;EXF+Mk!+qMD)FtjihWeZbjcav)POIuK`ZnkG~_?wLS-
z8WR{Nr?1>YYI{*((5W+eAG9$$GJkm({I+4jtWs=N5n3}o+MC{oPoG1I6Z>7j*RV$^
zOLV5UB{hu~YX5<~07~l^w0s?s)bxVPh+mAcsf96RdN4%MCbRE@Z(glqn65(%rPxi$
ztj6I6v2r(tMy)MTO|g#Q+UZ2+`C_&#T6J*>l9o|ML2^vUxT-JgeNgvwM@#_4xSUGC
z1pSRS7R0N{6mgyzOm}(?Pc@cDNT*zS@T9Pzzjd$wm)jky9@AWKsLo6h^QvhFMblLS
zndH&PC_=Cgc1D@#G6mBcHiH@BmEk#q(WOb`^jqE{9$djhsc6oFQkyZ1X(E)per)f3
zX^d33ZcLVn(zcQ#LVr&4-<4#!ra087R4swJOVD@G-q$&r-!MOpUK8(4;lc}p4PiV4
zpY1sQ?(Q!5)5T6nGpoh;4-v~n+E^)p`^(ka^hMXKr2|I?2M71F?M{FZXp|d2Z}hmx
zbvub-qI`%?OvLB+I5%z&K!Ha<IO=C=W_CT)FS?=mmbubky9}G3pI`ZDro0FKG`jaR
z0ZPa_z{8Z%OT_zm^6|7dsJGxMM+qHX9gL&6^MgMS&!u*!XDstB;v~%>Ftb$T%5|4)
z@=!J8*ZjYj;0hK9#Gk$Oi17b#jr+rZyx~{az6J$_%V3DU6369FspVdsh$v&LR0B!k
zkH*7yvo8nmCR7ND)WwWVa#@U^cepgQgqtwU<NlLF{!}i)bzYD*c7i%~Res1iQkSjg
zLmfs9*r_zwy&_`f&dNWhZ%aIHY>TRY75ZvlkHq^2O0j9mtp{9x(-eK=ZFx|wi(g_^
zy=3n^j@+Sx&er003{PzO?2EtZq|9-4jt2yj^jCG$R>}p$5_gn)uPG6s$%Qci369b=
zoV6dD;|_S$sOx}r_nPpeskJmhUYz_q+<+v%51Y`J1AiD*r_i26Xdpar^^?j?YLIG$
z+Nw9GOosF*H)hUC4Haj?hw?qS%M)6yw#x0pL@wInOY?0Z#`K%KX{&Lo@u!RKgG;J7
zpGKC0g9y|;F=H$iw%O{S9t^t4YSf0Sg^h>;-*g+v9dk<~_PPFCxpq~8P&u5(`GBjT
zum1{Q5$n@sMgX#iPEAdHn3pe7jRxpWt;2@v$spGiC<#ucc0k)zb~jm|7S3Tg84(%z
zM1sSTNv}x;uu0s|Js*tKpUO<b&OF3$&x?7Nt9}N}N_}XpC-O^RU^Mvi;Hj%i(%ajc
z!s`+?Q(?wtv(S)toycLSdVdTG=F25Nv|sk?Qr>AY;FXh-`hqa2E;qBnkBuf`Ua!c=
z%H{z0<FOV@<#oXzB~6EJ9qjF&J%65QHQViiHBq1@Ux0_u?z;bcAv?SKZ11OP(KQj-
zOL%D6t=91`4A|~F$J@eCjZjyV%QoqX<jZN)62pwp4EB^GqY=66_4`XGO2}i((G0Cm
z#wx>wDgAm?IGVj?fLvcSu&#Qp4yP?$&|9O<$HHLdR&g!}IoA3tT}hTeqaY|wj=7As
zM}L_MPVy_A0FA5bg%p^e2t&EdHI%OAkorbdIfbVh<|o<dD(mG{42+V9-t_9Ys6<(!
zY$NO9<#h4AzSj6rV_ib;^sYvrzu`O(G%QxW*6-BeK6dmvdXw+dob3cCudSnz4#Q;e
z_u4T6aFq55TS(pnllIQ+kx{J+aoVUn9l7g^79>VATI_p97%TfCbY^&~WE43N$4Do(
zz=c>&pV9u8iyr!Hk~1>a#E@_?rk3drEbm5!Sc~5=1ebd_DUxjy74|_{`}G&3RP*Mx
zZ(hbNjxhFmfWPOvrpH5`Wc`ir%ccj%hp$3h3cp8OnNj!X_xDJAe0-&)r9zDHWYN75
z{p-_z{$QY^w}3|ecnY7}_7wzX3Vr!7XnejaP;buHb4^Z8_87B<gofs8)i4SR3x8I&
zbOv~%f3aSf;rloyy|jlDUj(GXv$M{*Iif_FxW#loKe#7Po`eu`g@Hy(a3VZBT;Au)
z_Vz?UzD5OQZf@??jhUHQw`ECV7b$d`xtGbNo&vKmoSp~T)fhTRjq%|!IXU@5i7r1V
zP~bj?hROy7Ib}L-#x6B_-|+xghwjkTKCbET@GwsoI(>$QwsCp1ieNf*b5QCcjIQf#
z5&KOLirH4<+`3__E}p2=T%5}#Vi44TYx$G&p{JLzZYWCOyy$M-(jI%}_ZTsOxrX$V
z`qPJk7ja#?iw#iB=IoRdl?20FXkFXl>_Y^`@6`evev2CFP~wbw-fz1ylXpB-J*Kta
zh*Tf=B(IkXX#dEguUkLO$__$Ty=T|`Ie+tw8=R#i9dl0+F}J48lj%I>pYzmas1{s#
zuiSKU7A<{9X3D*%xno`>C~Tk$U7?o~BN=FZLU&+jVLAkTahsK*Pc5}y<~Cm`3-1$k
zva~uEBRJyEC&25SE??nQiYED3<4yCd(KaT-h&Sh$GK<y{3gf90qB8>zmF|_~i4XDR
zOAQ7fCmtgG7-xP9JGmI&_Zo{z<vCkPYrw|=hN8s7yK@XGiOTXav$h7S1P%4JB#*;+
zr~FS+b}^7e{vOxMjDTUvyD%4y=9Q_Jk2{Xq?_YMS>UEChWmaNgM7`h(|K>U0)fI5E
zG5n#)b0j*^`*Jz(_U>Ylo11%+h_2;_VwQS^scioIM6pfEU^16kZxn^PrY6qpP-mzA
z($W&f?DHJa5_lWHV&tn9Y5D|i2PKAhd3mL~Lw6(XQ&aJvPp&ZPUX2t1r`8x15%^w(
zmLJk6Y5DNFkG);F+|}{~v^|p)GQGJ#(*=sL0v|H0v{;}o58J&*DhyfXN1Nv<<MJV~
z{-Nx)yQp+a>NMQeG}tT=RGqf3pJz046KH~ix_290I8Z5)3{Si$={aCUL!&F-4vi!|
zOmw{TEULxCm)g7{t2;Y_)qJ>V@6}nn825BMgU-%iG4Z+D{G{p8Mk}r4yM`8Cg|6wi
z;Xc*%l^xwXzsll18Mh@9CD@u}!aL`0S?CS5@xDh)C+J*UMt8V)3d&(gC!PAbt!hEt
zzrRQP0qu@@pqsBn`n1xl{f&7h6pov)b(y#G`S?|Qnak=hg>K8o`5eJ?`8!M_Bh1S^
zPESwCpQ%KtCznJt6X=mopt;DQkx2Lyoqo2}sVwCm+R|AOv?>MNr+TQX>O*j-U82pw
zqaPP`@JBshviIhP<F@y_9l;jS+R+iB$fi2>eHF2e7)cm(9<Hs~d3g@d^{F`yqqfAs
zLSw#Ov%vZJdC$82z~ErMN&(J6^Me4LS}{2dO%R~%42+d;7oEV~#ME@@yR@VqfQ;{3
zU(YKhCU$vq!{)NDy+2={r<m`<Y4zuk3<v+(nkpd$eQj$`&oe+k(W(|oYz(Dd?bjc0
zdz{lRdTX7|D;_|A#sE5;+oNt$$Gtz;V5bNP39)r{&b6AY9!Ta2|M7#mu+TBtMJ$~=
zG$iEmbWC1SLPCG<PgTR+`7D@yuj2djwQE^H`C~d;Z8g{EflEqCn!18{szynvaRR`>
znnmv{saV=!UMyksv@-D;KQ;4F+(elROF5OK({*PEBykbX%LSu0qF|>TBI|J1#y9X$
zusg?zHEjQseXDL4Um;$7vW)A1jGKXa+)rF-H`$0uX!NX9+>P8PIPU_NCo@X8nhfqw
zoL|9K3V&THPpIb3!W5E7j<Qr$q=`h${Z3ym*7j`cjInc(Ru3;+7w&9>c`}Ks+m!y5
zFN+d1xTa?<iAIL6jxU_VDi}!}-RpWch_=P{DL3XBz3{z9W=^y<W-b3{7rt|s{dtdb
zoj0@8h4*RpfQs$%nl5vkC=!ZvHCERkvc*h<`_+?|4{5D6)?<drHQRIH-&D3sJzMee
zoex%$N3$Xv$m0n|s&A-HN%&BB#c_+;ekxT!FHi1yhV_CekB5iWf2b|`O<ZJOWQUfq
z`|=(fJzH`2+1nH}WQ_eeIbr>P0H)0{f~kqgdr*(~`1l}`3e07A-CSLjxgHtfFzFUL
zkb>>rad$cr0QcM2!2o=Mdy1?Nh+Tq`58(6&-S|GjB?s!(1+NR81uwN49}N*`Ym~0l
z<7lNbZ$ScZhl8Y@9)M+eQ<4M@A$kS|JA3=Kt}d~3W-woU-s?8Pd``BZYrM9*u(~U4
z)^9YzNDKdQ7`QPsEY}<c8q0~0n;Np7-l2|x+pD=ZF^Q2<yNwd&qwMi-D1*k7y?D8!
zgw`gOL@*nk@;akgR*CuRLE|OqxbN1gq(`6a(5lvjmwI=|OC1IvkkYZVA}M6rGf+PK
z3bX{uvb$kZ#<aIWB6B*UAEHYg&RaSgF;J46vdw4HTvZE;?eM}Ek$A$^yt4y15_*G>
zp{9qtu7&V7zz%FNi^Xq{k$d}T*`8KW%}-k95LUTZH$6Z2f%5hkqLF*3NQ|gJ3<|H8
zHg$jDJd_3mkzJ4KjOx9h<oNw8mEF`iJb^2hg)PA_Ovq-bt(U)JA()(E>xpXh5O=++
zB3CJctfo_*W@;CYjW}4?k89g)>&z;*)OP}`&a3AObxxg|u)^ry;;~3kKF9jS>)s`3
z{xSMw8oQVd9j-`v9+oCRZ$I4eI&k_=6M+r(OsZj%toQc~-20|xOwze{F&v&vjhX`O
zqgl6^xlrV2oD>w>FcRW=_`NWl3C$<UMc6`%!^muziLC<S_W0aM5!k8M+DtKg3#9x^
zQ9}1DlH)6P4AH%ZMYHUJuM3rF6{)e*0jce2&vpU5P}yR!h2?)L$NaJRv3LRPZ~yQT
z1%dT^oyjI6g0O24{w`9vCIkN$T-AGzR`k51pxIfceUThkuZ8qsdkFdbEvpO{)3@^T
zG`6ow(^QRhu9t5|&wlX<hzC6TE9ezxXA%1%QM+V|m3<*hU_;`619Xp^w&djE!UZ|T
z`@U-E0AswQ`3yhEg}H_%&q458p>fPN8SS~B_{==Z8uqr{mp5br;txYB*a$b{LZ)K0
zR1XU+wPdrkc4x4?&sknY8vqN5JPHJ>>6`;5gu1$gs_N#^J#{7O0}WcZq~gu$=B}K@
zD4awef;hj}k-+c!UyVI{jg}Hf5E}Afnup_VO4>o0gXgi=Mp5?<tylsKhqB7*F^f}c
z0C!WJm#of}5zjT?h)5X{iFI`W-?Cz(oDo6Qrd0IgJPQlQ)DOXS@36k}n;X!DccT@q
zwgvHb%ry!F>;)%|le<1P&pL|}=)6)jQ8c$xzskymtY0=&!4uE+7D0p$QBhcdWo0yB
zL;K1lvtK^P**xMmA+pHK<vRZxEXuwPa!W8wbw9^H^>mKi1T|fw$zc<2&JEwuAwH2x
zG0>&1KOWYxrLCN@{{9_y-%L~+H7(YYg`<MHtX4C}@ZvgM{bH6KSQ|;3^(~QqNPexC
z1`BD3WTp%H=D+Id^4pl_98Ytlhew)DYp!1{lJe-=;1e}|YnJAO7eXeBZGvP19JjS@
z4DgJ(LYDlLNJw&#*K~%wOv!yJJFnyW72lhPNHI=tMG`er6W43rGoWAr+k(ZV!^f!?
zG-+bzbuL(Os4uVuoH#s~s2#9BX&e!2-QR{44zO=QlKA>&zEX+Zew$i+Uz<9Q?sn9-
zRDA?Q&3=QUl7f<Mo{!GR+)DiK1ba{4SP7~Pz86FVpr6jR67Fi3v(;LL5Yo7wo}MKz
z@-A=MK@ahoV~5LZ>BoW7wN747uVMQE+jP^l2i8pDXf)ceVzH9s?j5z^fjDSJEwuUR
zps|CK8Fs6&c3WQB$I0*eDkf>P`9K;SJ`*oA$>qt?bw2iw3EI?~TJIJ4{AL^-WQ_R3
z-760B@W`^~DN<3UKc*&p5I>t9_z#3An}%+6**G-6VxS<=eA|w!H~kWZkWea|_zCko
zt|m`p(&=Zt=9`ciy`~fHAe9jD1DpFWh^fiIPASHg5+7(DZeeG!|70&`?d?JtH_qcN
zi)i6G{l#?`C$N!(gsZxlNtFkPXWZgPFKFDTqesfR75qB$f&;=AE&?%xGH)L8!v(}q
zI4+5D$h%nFBnUyh98|dLxVZ`IV9{1Aq<(;wOxQLl##~DPKaEN1kc<$0Q|<~~zp}^f
z3{T@^<;)Zw3(HV$U51~(*zo53l54cFDFSwvx(g$HHBmX%x^O`dzb9g4iA)+7nB$aV
zR5M9hjo9+@DQLA_HPCT9L+s#^kB7t7HLa*#?~`CYt<qm5WmBN80N~C*iOM07mGv`F
z^`q>5ZN%+ODRDoZO&RX(hsIVPj2^d4YEeK7f7UD#D^guu&sij5_Eng~^MnUl3@=66
zsvA)S%4h>Kv%v6hG@$Ny&CHDM;lVGK`poyR?T03B1T6=lJwN@|p0R>M^HH@O0mBmF
zRk<Fy6zzs*o{d^G?`Zgt;#8|Ey-!3@Ki#3IF8aySSu9E#GJb->T6?CypdKJIR_C*-
zwGKq;fJz^WZy={b47(tYFsb)-i>~|W>F2GNDQ_EMHL?~~cGAx!CAZ#G9<U^&S_1jb
zvVf!W6Axmsuf=spGj5ezN6-JpO#+zdQqxR(j_#$)GNrbJ&r|0=HXG`&?~E17EOaa;
zY%=eznSLy!d$&U=M~&GRbA;<Hnt_6$_YH7T94eLwSZ-ly*lK7tw|YARSF82<TJ)#m
znMAF7^wSqGNOW47PWx(~#=!ESL^E&SBYcobPk*k`S`cvzFucR0tkYi!FW5?sE~*zF
zldQ%=I6A}D_HGx<h!&f1?(67uzOYOgc(#$c2o>difJ&B!vx|!;K{(s+l<1#~p5C~9
zO*|z<9y$4_*RrTno!C6JEr0Q1n&(ZI$DhDd?vFNi>;d9}Wov9Ir~bKxdi&+iCDk5c
zgvqYoZIbROvH>$wGhh+UXy79!pC<npq5HS06aw)1TI;RndF&k=&i3=f!%1#0SHr*O
zmO!r>){I&4?*x(u*Irw3S6Sf6Vc(Tq{K!AL)}Ub_(!CadIY7G?%hEdbFW;G$&$Tr^
zu)s>r$}DDJxBKn6u8r-9`9#Cb(bWd8I4w5VstJ@^G__hYN(5FPvsz0!EL0Ca5aVHX
zt!Ic9ULMgs4N1~7b_jnl@$usP)wz`>>=;ziTEmg8Rl?!K_p-W6-`e~=K(?`G1^`Nl
z7BIMU^WX)uIR81_!mk47jP1ar-Ip5X+UwjZQSviClI-kR#~GE4;lcaX!X4%!o}Mah
zaYT}qCPYv>i4AJ&xcgXS9vo@_Y&F<ohWLjG3I|64A&QV`+UZ-R{4ExJK2B#jsFU*n
zpj6;J598!sS;TpsRrtSm(tk(YeNochJ(X^K*M&1PS7ZGFR_~w-0O1BhPM3NdG)MXQ
z#AUYZPduo&>dr~>JHO>nfn*Ymf5@A+?xBxb?~O5eG{!mh!1B&B@3bo#?1FKcYHdHw
zBGdZ^cj?}FArEt`s-UME5OoU)&SXGjCehqIqKXA)wrKe4KYQ8#3IX540lZs6?F!Su
zkCTP?D1@B!hYk7pG-o?gdER%|3CXxbGs_1D;Xsu}Nf|EAX*KsN@*6gt{n@sPXdv}7
z5&RE%ik9xkDMHrH1Rfhz%ff5sp_9^~q58uqQ@t{E;-Kc|Ab@(wW31#D(>z&BY90G;
zLo972+Gqcc#T+)<K9g?G3lQ`Q9Hy<l0+YPl<4%6`Yn9DU%eXnOj`L+6)h?h%g}boE
zdGPr8@U>R?rz)C=l7T&Uwxw%)f2%9a@-t5sCzBvdQ4X620YDyneuoh9+)qm*$%^U?
zTvvc_T~y6=2oz@>2zAP6*-IoTY-`)!zlROo4fDH-gz;o{d}^I0kb{@=msT~uy=7!k
z+609%KvbdJl&*^|`=V`+{+6J*Rm-azYxefI7UFiZCHs}onr-nXw2;B<#iBBt5xymq
zN-Jxj9+HU}zp|qEJdC2P&@v)R`06=F>01Ktx*%|bKzc(9hjZfUg5^iyiR2ygv10ue
zQFC5O8E~H3PfUt^1^<S7YS+}DZ-iM>U9RaKvd4?<KR%(Oov5k!DJwq2kR;@85rrfC
z;=kf;II0!1ArO^#WvSJVn^!uzf!%U4q_mV7sJv?d_>9Y_{fdn(`tZ<J7rJwT+1-Qo
z-a2h#y?y)kOF+P1szA^jm{0t~e&PSb6kwm3GCTaP1XhdJUkhXKoQJJj6RXI2q-buD
zngKr*%-dve!*!qC@ntN-T-@$vQ~I$MM<zPsFW0V$SxWJitutw0uBDyEL`?r8Wcy;B
zZp!esp$JIDaq2@+nujBz-Zb*(Kx;A8a|25{Xydflg!c)GpFV(iY|4SfFf$50nvrwK
zXS>1ChD-xR^a(gmDoV)u{e<-xv+W9Mq>&{lqmQ<Jr9b~TqVkiA=T<M@QAqJDB-&wl
zN{g8lYX!S<cRbKLeVEaXmnqABA3O21-8z4+tC#$e{hd@PU9d#Te9vdeWvKKO#_@RN
zKF?>Dry)?aaD2vxl4ZvOK~&Zlr6?EgZ1x#FgL0Kp8qGfkmiUkL*$CT|A?U>Y!=0MF
z$rzC7A8eIG7_QV?lZ%1(t<*|oglzfAC*nR9J0In5f$h(kx#7bog>+wUNT7VlYrmue
zKzCOcWyqx|0>`cHm_Qe#6p@zHQ+y`qlTY+iZJe?8rf1e#LHB1bq5HB#<lyn}2$g>f
z^!uGGp=%ZqzIpHHdzQ!35|k%0QTf63yCr@=l1RX1Eu)}-E}O)@3?OPeeC&I~9I2e|
zvdNr80-gzVPCNGZw{8GZ>>V7;kxS(Pcr#q7+llFdcl{c~R5R?WS(FR-Do>zhIo+Pv
zaSP0<DB4JOfj0@%ETW{NLjp>dkWk=Ne)k(Z)AXXwzwp+a59Fu4ie6T#eIB;HV`!}t
zmNGydl~saH=nWq$6%-lCGJhSKa}i<Qv+h1ncpo8>*;De`+SF8a?Lnul;-%v!DnZ$s
zgqRL>Xe&r6ZpjiQj8aL2mPhO7`ju_vx}{A+e?*Wpa6vOvFKWN%^qS?4&g%0;L2)YB
zcs}a9_%=MJ76l^t8ZGu2hm4Ki|NWRioERUhI-anXmMvSnp~Qeco%t1+&0gx<<?fDP
zC5d8zYQFl-jxr}h^7}MawYsAEY82(l2r=3`i1sxtrpTCZQCG)!%7S0>a#53!?o}}P
zh)Xfceb(RbMF&+eg+)~D!E9sg@310cOV*q)_GDz&;R(|57vg-~KZ0EVpzWP}f{$-b
zGm5EFlpWX@q#U-q&<<+N7l|@mg!6*yStCfyEDZN;;-G(k`%|Ck%pAaRUq?YBEX4y|
zpmYuP2X<-|$6k`i$c31+K3v;w?JkaDPIirM^i0SQq$Xv2hQ_jki9Q#YnTW~V1aCVf
zZ{q0-OXXnri}?8a`wykZlhr>Mwyrh(ZXybqG#bAcQom&RCf69iw%PA5-$>;0K$-4*
z7ALmCcrtcJG~eed9Di8+xBGUYwXPewiRv9163dBV9M{mqffN;8uS`r#3=9q5G)XtC
z)nV(rMEpE3Aj8VW_N26|taU&pP#8YP7>H{hA#O0i_zx67?MCtR0j1J>jMCTF7v@oJ
z_)Rtfz*uqg8onhZOhD#Ss8b&&;B~EUXxIkCndW1;C@_|jg*0r>bs=GNsc;#}tP_v)
zu|gdtPKJj#OvFh{msN5*E;*c!W>z~wOFQ$fW-aibFL9F_d683OaKfcge#XN)b>EYj
z3fVuH!p{xR|9<@bt{Hg78XverF}bHih9je6Ur%I>bJQX!`BtZ)ByN28Ww-7bG+R?y
zZM3_KqTT4|I|vOye^<8P*3vw$<IIMonjQ;8p5_a;AhJj#vgm%dhp~K6@ubW4o7!Ec
z4`%B~eU{Eg_%p*>_XtJSU62Mw2j@-By5ZNxm-HRqv3$}y<4=woe#5clql|bZ2>eO;
zi`o2oGCE;7PBUp(=Vx+W=9hnxKk2*;%<@n>S^C593yM?B2nBXN6s^Wy08V|#x6lc_
zzPq?JFvs%D@Y@d(jlif_i0gPaza)IhkdG|NvP-OU<v!xgnD(c;y7}+X!BzVjvBx!$
zo7zd1PEEM^#d!t>Qxl)%>kU*XBp=OjJSG_<#7mZKyYl)Ud+d`ZL@K2)Tx{RpFh(U2
zg%>qeKqRTp|2sKyT9=+LJW6PpZmzH6rkkW<yL%JCzen=qNnv1NVUGd9bNlNF85VqI
zVTFtRDwx!G-1Y)oC4uN*g5~#uYyh66UCp$<CKRa#Cw14i30j(CMoaLI6~O79<y<qd
zu#yNe2!Y_{_E5+%6)lR+I4X02vJsfzWO~#VV2AP~s@OT!v`X+X|JqRT()kgp4Mk%?
z=_$K1;4>fnP@mVwWzgntz~`}}Ne|PKQFdiMVsVj>mhI>odX5_|<*HV6Atjre8(K^n
zlD&gSCiyM=?{rhuTs^ihxb>hoZDAbcqioSNLz>H1O%N=TF<hDm&Uoml%lEdgjf9d?
zN<^ST02;cL)g8T+uk#PAn;Sj-MBiCKv#e$4i<=$nbC&F!;XQ_GoSVmYOY}ZExu9ia
zhPb4UgJwfj_~Dn7<b4@k$wrw2CxS!Pn}-KC&#(a`FC8P+Y;_}hiLdo^7!Cgov`IuO
z8+jlasrOv%(;Dap-`lo}UZ<kM!sB=&BO?@y<X6v_a1ylv8z9(Fa0S-a7uJ?W5tTVQ
z<%#<Akjg+zISs-kwWW3e$3)MbxwdbbC{luoAB~M=Vv|CDC{-xt+w7REF73)~TnJE6
ziT~8G{PZa#-+YDu<tIv_jF^94)enW*xCJl%dm{Xdm{2HmKb;Ju6>hUz6hX+t3io9d
ztSw~~oNssqQXBZzKi~?mjgiK!GcP?jU6mpI?*n`TH3pE)U-bhndwOODdscvK^~lOs
zzW9b*Yj$Oa=3gz6yGzt3i9}O-?Gi^1?D3o5Aw6G(J|&FRU$z!ZEG{+?QPiPJPLTcC
zcR;nqlxRyAB$b;iJ7g=Zza+8$b{O%2ylYWBf1tCqH}r)-RX*--9N-bUGonMNmM7sg
z<7dSC;j^X<%w#Vb3n-v$tMcU`Mep7J-k?aL;UT;YU}I-W4d7saEO8y!D(I{49_5II
zHxjCotN3G<<39|Rs1d;a^F=Kd?5yU??qWwM9Co7H?Gx=<4^8lyG1_Rn-9I}YR!0-z
zusCXpJ^X*35^+byrCxtzH#fQ8*c3B8LXrONJr_v1jt>KLRx@R$Pk1M9g*5FwGp)el
z|G04PwQytnVttNo8#l|RmfvmHv+$dZb_%aXhVrUX$AjQgUPzpl<DKc7c*5ANQQZID
zeF>`{?jj+dtL^=9hOpl=6cmPXt9S3<f!k^L!EN(*=2btJ&BB15^-Ot=un`uh1$1>5
zChoE$eKWVNLqAC?v`jb>^}8vQNfEarP(OPb6`!|u;YNlx%k&xG<yqkDkz-{3&MK4t
z)W|EN2BNwomEHLB|6&65mtHpnk9wg?pFuEZ)eIM_?)2g*nRw%o-_p+eg$#V2CZfIJ
zY*`qo>i7#bS=PcpCZG{{r)dFuInySNtuLaIpo|ir;u|NoMnHFE)@>iq^dS{a3mW5Z
zQc`G^L^NOS7>jN<UthkyALPNlwuc&9oRw1d*d#OP`N~QnOC2S8a{?AE)z$xOE`<Nr
zg(z4c3;5Q6oyc+Xw+|p1qJWAW5D;6lHB4UDr<o%DFF?PwKUdohC{K;*Pa5~`k32iz
zFRlaudcY4|?bR%DIwpX@ZKIQ7<XWk6J}9@G7g6^heY(R9)!FQMDYRSeEnr8-J~s(_
z!~jNYUV7S>C<{6XxdHH`%Rk0oQ;!6+7QGB_?)lrNy+ae4PMJxte`3!xgPr>cL!JHc
zlGro}96!G!l4DhjjP-q&2L&fTWXQD`wil)p2`5XuwEE3o{?&OLVK$w?omb3GmIj{}
zy4^+AC8WQJJa5Qe>h5DzR^~#v=g9ZYRK#SD@*DgsIyBC1nBRpuB_hd>lFF$|lr8>J
zD%84|6d|VZEQj%?H3#y?<~ksz2AVoS6}A+!ar9nz{j7xA4<d{0H3lM+9HDY&U<|CZ
zSxf^~jHR6&X&@_oLGF^wWitdgf(XGQ2%#4Aa)o-$X`rneD>V=m4Wt3Tw4rSQpPA%?
zgES2bc3iF(;zMrhkUF3;w-%t3WjHK&{yWYKB$~1flP7``aZw)r6fUIsmPkkz!h~i3
zzfOExZW)AW(Ri_?ytUHx(?h8i#wu0fi%}3zjgFN~l^P~5gdv;kU+OB^LoNykmv!#T
zz4PLQp$7U0m(@?z2`fM^1g`KY+&mwS$WKJzN0f`{``}6X`~`!!P=Sx0wXhUpN#l`j
znG5<U$!%yQP(XF8eI(s&b$uq)HPN2Tse@?I-1NT|Rwkw|f(*1Zk8)%mI7HxdV=7Mt
zEvv`*?lKT&K>!HHaxDS^-P7HfXyCxYk+}W0eKi5!?$*Uj&8SuX2eMaE!342ftl3V-
z<EIaXVF~)O)DDbk3&Tya(k(;qrdPh=EJ44i68mz>Q9^fXUFcGpab&5h-LCpLw(ghJ
zA`%g!+p45r1!TAF62f>MPrMlz2qx+L32IVqC`WK65&zI9k6lxJ>}u)ulYx>-M87(Z
z2VM}`z9rBrZ2vZT1`r`3SDt~UVz($Yl)_rX`qfe=>h3^2Pqj}c0QnlZnjd~8`#ic>
zq(S85m0y&BH#uu<ihTD2eeBwQ2r_zKwo-pepE2v7YAb2bDS&-jyU{&1OFSI1KiyO;
z)~*!+{+ZU+R<Oi`3pA1+nWk2Bo-+cu`Wi%GB}UJeFFaPWd%t{($_tpcV}66IWWP1-
zK)tt!ukGj|?1~C>nRGY;L@ELe7{8Ltr+l=Z>Ro@%d3jTFD>X1r2qYW)#sp6o7JYAK
zm~ZvC<ESZK2S*k7o?R>#Ky(qQ;s>wCKF<BTRMO4^p;PYa=*nfj-s@Gfq@S;y5332k
zrO;14T!)_T#KQuNv}<tQH##`paX>6ka-HB?bjD{5z~m8|-@Q$?b7r`5MaFv@?*7DT
zSvAfvOpXd03bR)}oYG0aL#?HK8j>ImR7X{_!vD2U%ZaS65XwT^^z^hsqr1~jwUT>!
zPO8TxMG@6&jEvt5-)&D)sKtZ*SG+CY<HuF{&}#?I*r-=WF95)XNZK<0Nl41#43GQh
zYX9>qtd+l2((n8c{c0#GZAXhRc`-Nt23VZ&1A;u*MewSwC)d9AC200r4o<{M&DsqU
za9&dwYQ2RdO-NH@aw3oI-r5ShuUi*nM68Wu@$0&i2qY1eLf%-a7R>hd)=dhZyDPci
zHxQ>6KQCU%ViD9g{AsdMkHsB#vQ#fP%xvk%RSa$}dHwo#$CISDzTw!@$wC1b`^M6e
zGonWkS#6mqyqJHjvZMyZ2&}pR;2DJETy8w6e?J425m5#L(Zd$eM0w<pVljcCL$+c)
z;_7Q*pRynt(T5AVBAyGndAQgAV7ynGdx_+(kFz6_KB{-U;Q~;r2$d$C;d*u)YuU;L
z<<CG@n+0^LM@6%z_+m8Al8|_+KWKj3%JyJA9l0(zu-l&wGsJ9oMS(7OW$_Xvxf$y1
zOYd@ETfxvCrY>)UD&$V_1%jnicH;2FK-C7SSl^tN?3GjQT*os6qF%h^!{jgBm!Elu
zF8*$0g0rXJXa!KgBX9rd(u>EO3UkR=Dbp%cR%UD51MdJi<uj;oRc8T1V|2MD`0V2q
z4uDHq<`3zV+`hhHw?5+}MTSy+;$2XRs>%NQe0-#DS|1V1hGUD%#;$?IaEFsW6o8->
zUYkH7?~r8c=B;&I`Ouy&Hdz0N<SMF4P>=1u3y?(OusYrO0Yn#gc3j&pEUGlKF1-A5
ziSrkK@lUl4+3jySTkAB-D0w1-<{xZ=ABvn`g=#u?R(yy4x;S<&1iwox`25GYQywEP
zNR!i+GN@irtxpCNookly3uLZYH&kCz_<OD}>w_9`B>|5Gvcv4`kG@;$YuzS0jq;S;
zkb}|=^hqDbzazbjmv`9TM70(Vh)O~CD+|K^M=VXne8$nSJ2_@qBKaRulq382O+x7K
zPiU@|bq3K6EWBaX)+8{+5)=6!v}zHAKLBQJ)Hf^0o47Yyz<%EZ<T-}<=a@b>O4OFC
z`r`vMd_NY?Nu^lF@A*c=z0*xXK;T)ZremgJFk?;~aqZkQAh<g$nkrPT`TxP?&52tY
z?C)PaTyEc;tBnR0F?5KSI8;+-E(A3SXsu&aB_2y4g<NX;a$v5fq~gW&5LW|^&xl?W
z!H4`eywS6fnH}#J+dA8mK~2ca!NGOuoQ4fd6M(q|%Fv$B5L;X4Defm%<!o+_y4bXL
zydKkmLA@vSJqDUDZcg4bUp&s;$LY(Q6SWlj_3IJ@gnLvxJaz1K|D>O|s=y&~!^yBb
zuw0~#hD3Xq6xzs!lI-+dCH{Wgd|Ikx&)6*Yq<_~yo4YJ|eC;w%x5mLS$F9H{nW%Be
zU($8H&O7lgIo^JM8kOYTBKYrSouIOwi3J2h*F?p!!?(jL2l|cI3>|9{(WzTfnZ8hm
z_NTp67ETrp8CxFeQo-wQgEZ55|D9d08pr`lDk~|O2|O;w#(Zho?Td?~5CqvJ7l3CR
z_9+!8f70SeKWmmHqa#ysA02_88*G(94XVwams|Q>j7{{X$`>dI@K_Ck<a+x!q?=5~
z$`jYtX>zFOLo*r>)Cq-uLKyp|(NTE)O|s1(x|j7DIk{KMON`Lg28E8FBWQ8MyET#@
z;%_1Ztx$@rFalrC@<NgAeo;bu*aXrhI067X+crM1QrmUMcR6X<wK(e?m|^TXsCW^^
zm@ATYUAx$Iu%i1M(aa^PSX|Veae7Db=dw0yi^C$-e_sS7%*^gb<%-Lo8IwL&1e7FH
zaw#o?gM+`5{|TXRA9n#H%8T_=!DL}ZKRL$V6DSP~IhuLl*>ubAFbeDXbJ!5^#%f;c
z4bI3|Jxp=@2?IC7Z69(0L9Bv%+ajvFn-KA=FQ)SP&sW^olMA9(HU#~+A+K+b&6`ph
zTlH_CPt+II{%l6?aK9gD!)Y10PQhvN`i((iv0@u7P29(v-K7qLXaFpTdVW%~ytI4o
z2@A;)G)>}z4(<Bm8vxQ4I^BM-QKN@>!cb4Z%(K_KW)G|BXtMZMPhXy>;UHg2Ty69<
zD<&5cX+wi09;d{AA6wXU2w8X2X1)$`y6Pw7vc~Gw0S_$4=Bnmj8rTcc)b3b34c=+_
z+NW?O_^98t+v{&7_5M8=3NB8}^%@CEei_gHXL7e}K4+~sk%&^&UxOo~Nv9QDWfDXL
zyKA7!fTx2WVv@r1D-vL1oC9)%6847l9|EeQqw0OmOXtws-T6}I5NyHybNB1Z;qFV8
zgJ-1<GB*vFfVYva@d}!GTlf2V@P4$@H`c{zXHAJVMXIg0Gx+b`BG*UHy3ONOi8XsN
zbr(_$tUoRhc>y^}lHoNsUO_VFi)d;8CNEP(M<sjQ9<NY|dBq_iA5pl=jY82wcZ0w+
z9VL@qufOF~y31GJ3@!+K`DAb29LMjB=NoPA|C6}*ms?GE%H35+O6@ZiGEg{68`Aq_
z#%7e@e%qx76f2k#z(Kw{Tlob@gM5OJO{YG`$x+ydKdp4}vH9tz4JIKJk1=8b$VAio
zj)jiyg~)Rg3#UP`ES}`Hxg>?zfFECj=~MX&PyHB-paoJsY|vOvW55Nb6*p@aP7@_C
z%DT>)T@J7+{%xx;boz6Fhn(ofOaW_OnP4l6a_degGWG*#3OQ|1=&QK@?I+Y)<@Q*f
zCsP)*7sM?YSAwNim(TA*>m)s;UPO+t5Iu0kBw9|sWcGw7?vj5)&bsXuNdJt7&DD3R
zzaOH5W&nNtd;YPvZbKsDc$)^6b95i8Oi{7d-ERS1^pEm$cgee&H%5_zz!TgL;nCeO
zS7MA;j#%@e?B$G<Sg5mFmOAgmMFDwYt>->zzVjg-=@y<vpE9bR8XPa?qwKEdaA_2&
z64~DV$nZ3TV^m_o#-`=(N173hD2Q(B@I7_*+Gx{FUghp|-nh}k_~(}n`Kr9=ZDR(G
z)E$=0QEHjKSCN3RpwsertiB=@^9PvdAJ0n>e1{ETJ_fPS_2k%s-XX=`HO?s24P*ZG
z?ShnE1I!rWsINZ?+2q!IlA_?%x^F_4SsfNDUukLGAiids08Uk)T84r`L%kL<pSiQT
z`qlti@T%Y3+&p0a(VBy<i+98I@J9B>WM+yfnahLpQ%}v8Ypm6U38HDNHNQ<gIO*T;
zvJum2h=}9P`+9YwXb(t8x+_qM9a-NyR97A<y5+BZsos@Yg3H(4i4$yQ(>G=c2pC*k
ztDxy;;pFt8s96ezpk*0W>WAnp6}ZSm=7Xmp&-~vZfQn~zwM;NBU+F1j?q8XN=2QzO
z0?~I*ksILY_C>UZ#zNVS^%vnTH!*gD+d<f2ui$QLqi3uc>92oNe^~pI5-WpP*z3z{
zlqA<WhpjAu+dugX#9p+^pUi!<DceUo68@~G;9+jc_jE&E2@B;keGPn1hwjq!ocG)o
z>JJB~W2NjJkRHNbH>iUHfSAYe<Na}y*A4P~<Kqi7Fb@$e&@wnqWRWz+BV5iz0Rkab
zn(Bp>{l{*5^UR81XEL31<2PUlw4{XwJMFX!0np_c+FP25KUOP0JjD0<eacc!sGPF~
zCNiC&9i(EF4etSU_b))L2B3okM{IX^$)Gu!mxm^Jr?oJCqTr)~gwD&g{`BmQQpwj&
zVd-o@y_Wl6|CO;(Tt7ReUC$*`zBHyM)fT%6HE=mT(<~yUG;A;vKxG3uZgNt{|6rrT
z(GP#uM^=;Z=iJbTXD2+TZx&`Rxu+rQ`mp@SmqxI6Gk>y>y6oTT(QCFHL89r39lrW_
zz%}(Kc`nk*VX92`JbsQw!pD3{`R@Q9_^@Vd$4~!^NVY{QaHL-TMJeJd`b+gpro~2V
zMh5osDH~q2sqh!gKlB9M>+HASK}|+WoH_hT35o27rKqYHZB4Nrr@AM^6a~IqHpq#Y
zEU%>Pp=v0BTJ{w^iX3HIr?P8C3m_m#kyD4FtRfwEISOS$=kAy=KR7%MyH0zbTs&HJ
zAps;L?Wy$7qyj!jqDT6~=I}G{o*DdM3QV|NB>^H)!GqF&JA|v4DiL>2ekt8$ZQN=&
z*J6uI@MxxutC}$+JR3)Sa(wAIf}w07YoYyIWJdo`#CT+61Rfrq*{BPdQZ7Zp{{?me
z2sQ$+M(fZ}aR(F1{{DUds^sGuNvqAi{Ak`fsNGCX@i`eE26T<M*Y?(O)s-k}Y<JbG
z)~G_xssY}2n>1S^-gP^*f(lF1SL26M&CR$G+Sc!aA=F0mkxVWiuf|wC0EIwcPAo&%
zI$voeqqnlO#@e<<A`rHZL2G<K>>Pnb8mg4v(mF#7&hs^U7PK$_rlZBNGHvc%6~SlH
zKaFAjW#=>{;kOni8|%zN@5Rc_Iwb3wCkK8T#8HFrB9P5f{%MAFD2-oQP7YN-K)?|K
zn9dWprwUnOp>L*upc&Et{>jIj3?3c=pd@ua-|YoaGyz$wtGufKz7q#z^l?9%tGoM~
zctH}lI+K1JkQR~hPyzOOAXR@%Fo>p-kBW~M2R?Ru_K)cE^)5`M#R4|~H0ob&51jdB
zNA}2$E;M>@^YQ{)Jw(`axq<|cqr5Q|FKj%%R;@C_PS`VEz?WX_Rt&T2>M;~YJnnT&
zJ^+X|Z=HkT>Q9jDrR~fTcn~d8^(oW7I7YsjjWkrrKR(?Nmk;1~KKAu$S^@B%?b#tf
zBW~OR4bfQ>k-?JLC2}g{XL$T&ZPKgSuJA4-9b?dVzn>7=yi|7YXGNd;hGr)&!f+Dk
z@ACyf>pkYl1YsgbSLif(CicfOv6&4KVPIgqAh#6m%L&D23%EMjz;zX1XOH>)TfuZF
z1w5JR<8-OPy`}u${o~`~p8kHi--~1~{C&^&W*sgMmtnmQ7Ms%nkoXAn06U{^jtSIS
zkK;5G58?(YY?s?$_W)uxSZ>n)m|X*g+HmfdojfsDCYBWY<J<OyIhR9w_#wkAv^Vwc
zeXGDjd(Skyy-GwT{iR6Q3pvVGc}3>8L}CF_PExpwUmSWf6tF*~lix*}Sy&XSxYDYS
zs4<e5t;?@o?e~+&5RLnqr@KIgBbDKgUbxW@m{r@uL%k}R63i`M(1Mgm&%>`D(apz$
zK<Z{#4`ykQ_FAXMS}h+o7pN-l_@xn1iIzHoFvkluq5;E=^#TDnHg|VFwY9b3@w+=_
zx^yvV|B3)(^Z`iY0qyx-RTT$3BPIqbj#1~ENM;7G^<_hOd2qjfD3Xm^IC-~S+~3{&
znApDUPPS=ET)ZeXXy^94WVW`pP7&}*nw!&(k)pt5((Ra?O$MPO?ic%35ne~<=ce;@
zP8}T`5x_;!j0V8w&c42HN2}eKFJBt;h~AGfUp`xSF#Y?lvQaTM+V}!G3Ue2XS31up
z-EaG##y5Ab|7zsBN6t!p3y=H5Q67&`pPU6i!X9cv_Wen`Qnr8#{><P4_FheOf($oS
z<sv1!+ZLWjaaTe;*9)r?AoBt3L(Uzq-_~2(?_y*O{T_l1Dmzk1Vs==>ufoH1I@`ii
z{~b)ehxWgS2@!=D%%~c#%Q!hQeTz__spu%l{=T-Zwl?%-^R>BJ2N<x3vW2c-LB<jA
zy|sd$q-#>~^QQuf%F5c>JAfYH$ajFWqkaHRgBfPj6Y;uGw=qdx;EMd=Ogsns8*Ldn
zAz{Bw^IaTia(uknz2({V1Q|8;GhAHUYOC2UfHUSR<$Tk<VYOfD$%E1=s;btDY8DKK
z)A%btj`-$C$K%xP7YqS13S<W0*GWIl0M}&z@e*+EbY$w#Exp62WXjVLucb8?p>Ju{
z$P-B(NBY#g3?BIi>zW{)<!ECjNkORV@P!gpVFaTv6hLE&iApkN(fYj%m%Jpos_~PE
z1U2sU9i`ynAxQnOSpew9b=<$lp|HDr*`8d|ht+Vi6+|2&&*Y2hDXtQ<|FlTZ1=<Ug
z=BaOcv)xmqpeKzIU45nd1Y8+HR7&+O8^|qzKGMz>hvWX=ZjY{O8e~4|?<J13w=YH)
zlI8!FTv)_nn@#pmASs@focs&0TUb^Ofm}ufC|}w0Y(fnos$SlGBh>|u)%Y1OATvcQ
zG4z7moF}dzW(eT5q;`&uYbV2k8#f2dW*7VOp7&Q9?t9X9fR=-qu6HSsIK&Z&`S`e0
zz9^0Ul!nX=CUg|G)s4K>h{*qrW|()sk!EkW={1h%T@vC$Zl<SjA)IctaWb&yfIh0B
z{?w4{UEH4h-HMQ_5}3XtB1`xy?TM(!k}(E=5v}i<=sB~2Z31|FX((Ly2MnB1E}L*1
zUE(nsN&YSwQBZhxU$&%sABw#NM+CFEt);L0K0`H>s6Dh}a}HNUa9>w2YX^7vL9^y8
z>*T+A_^P3j$g1#(#6aI9R9_<Q#Xa+9dp@4C&gCF1DCl{6AlhSK!Y3dn;&s{2P|E+N
zy)hy&1M-3lEG#rPR~5@S=;>t&Tt!4gU^ao?A<_z<nn*2Rp`R_pi?f7*jS(0#4XmtE
zo6%HNRbc?qsZ!wOs`3@S<niTWqQH&iiU5G=SZf*8g8sPJ(jnddLJ>+|2KN~K{ktU!
z{KCRHaZ;>~cK-NHODb(2{rcjV^PW~Z2Po%z+pNaf7+D@@uhCjmC6VTON+`RQ#ti7F
z6K>C9Hxr9v(v0v{Vz~8b{|!0lY}8urEFQgUFDb38_UE^=rOdy)N2*}M@r|;xD|uIM
z?N_O%2k@7F6FgJdF}g40?&g(niT5bh3+tTh6@9Ba8Sw*O_BWigwBJDaR}MXRgY!@l
zR0GF5!(E-7{8e%?GRTkjQ_juJiI9d<DtOqx;gBUKm*^gysogw445yR~4iBI5(94xe
zl`7J#O5m~~0c^anvGHajTFE!FsHH)uuWtpMtMA0bF!1rO=s6$3|7XLgK=?sL^&u(D
z#S$5KrY8I+ItFDDT(Qe@=;nY3rIWJ+aEO)Yn=tK-t(#lDxGzlJ#7C7l<%~REFBk`M
ztwd2RZXnpV<?y4KXcY^#ke{E+$V1e?H^04}H|ylADswRG`J$J-%CQ6Sq{xqW@;-F5
ze@_PR&6Vf2Jvchq_0;}1F0UI$G6Z0FOF*vSA5y1e?bvTz1f0z8)HP2|PmMdjVH=y6
zyd)wT`2Ic^SmM@aZJImV+gI}&l7J}=W@oB|iCIz8AY7}alF;#(+U<sVL@Qy<c+2Bk
z^w?Ncc-2h&SLo}rT$x1Q+S<2T)t_E7GBQHnIjS^6MGQXo7U&rQybAb~A7>)S>M}An
ze;oV6qodaD?mVG-QtwM2FezEA=eT2}OkOrV^5|{B8vr?)t8>PGL}C~zT&eW@c{Sdb
z#1!~4uz*IZV_gVkOSM8Jm=dQ}N@R}z9$Gl(O~7yM=~EKXAnanrQ^5~0-c>*2-^%w7
zPXMo7czp$pqO~{XNdBqh$l&{xY_~P=M1+Zmrg{sgWV&zsntcZN5^Pln22*qg7GDI_
z0zFexW0(wrDAm14(Nl)1aK-<P<TCqlksLz2eM%kfU~m0})3#`iD%$}RqHK{N9=x&<
zN^J$Q<u0$TSR6MMK#<5|zy<KR%cKi53h@HW3C4MU&ioK>dpsYmWf6MNF;0f=uMj^2
z`!nv0ShUw$a8Sa4pYwymFiX4E9*>ZaELu{}8UqvaY(6z!GoReGqN2ilrknzJqsiIX
zE2pay6B7s0ApT_Fb#(v+l+kDZ%oAzBMjmR?NTe#CNiC$abce~?yW{oeu&41*s4!kr
z<D>F@CjFja!+h)cAt}4ChRRcqkZ+)tz&`5N4Hwe<&=$m*IiMocZLjneUI!!*(@}vv
zo*VL1%ajG02YZ0c0&W4uW4_vm-p_^q3HLR$d5M)Z)g685xt)%eoY3~?<~J2yQzL5q
z>e9pRyo=`t6Wo>~y{LWuj2tX#lNbt*9`pzeY`t7$FJT0eXWxHhWGumAX?&-CLrwh^
z?j<~en+9w4uB=b>04#B(|9<S*y`6NXGnF~qx~E{e_pd>({aPw>sif$Qv<>|?yUhp?
z)X|2xd^pE4?FdJO%a!Y09ckSto)DMyF97J*!4DPM1>$o(7~T20VD_Twah5nD-p-K(
zLiPfkV1d0^4W9QP_hl48!L)9n$ZR+brck}i7r?49w7G$S$6Q_(78X7?^-eo+guI-b
zoQwC_+2lC%>M%kjx{Xqv3%>jPdl1nlD<p5hN3$_TH-sK5>Ei+p^xc^XL<R4UfGiX#
zV!TwYNvF5X(|o*BY3?B8DO~sYoxHiv??0kjE;q6t@DxSOEp6E)xB72{f3fm94Ixb9
zA=Om&U6v5EpG?H6Z9DgwsELc-(;jMzTQA;!jdzpWAcBk!a$#La=RKe)Hz{7}OY1#;
zVFdAenoGWi$(*SQ`fG*|MtOliO0A+weRbQHxoLhb?fT0U(;k?-yKcu@YF8HXZkm-$
zp8utEKT~8u*;$AcC$Z?r|K%F+Q0p(F(|c28U1ysZ#CxuvHBx%G9<NX^aOF3}4vJ`F
zkZud|br>#aUHqU{A&`Iux)0b4fIhf|9UUI(PZnyxJmzTu({yxHR8+knP)t^s4YLBL
zE;w_JiDm(G0n_ww?@@_>)p2)wi^t<g{pjspUh9pnWxxdW=f~M<2F;>4;L-z%sDU~r
z@Zb~u(QHW~L2o`6mg+BG;6Y^F07xs9f^NA5zRLWnF8?SVMr3ezeFpOwA_B<kOx1#s
zkL(D=F)kOF`oOm*lfpeRO7nOf#X2JmOIXM$R!-?o#L4TtMuGrke;;r3B-p$UuU>0y
zRqU7}O%TT>mZd#hkJlBK8-5z7@)e96Zsa#h>D|V1Njq`tN*rj%p`DGkT_u42>=t62
z^%U*mw$L3w-NDfU8<$NFb~Lx}H7i+)1rx5S$60dilk8r<c%9GJ>(K-Lo0u*i$1POl
zlSOXSbj?IN-n-vc#@To{QT5<2Ex-JKEkS?9Z=>>4_rIikeSse5M({ZDc=GiDD1A&@
z^&pv%1<3eCE~X%mQP)y80}5_QZEYeDi(~}iw){z&fftQ$QSf5UVIYk^8N?g0d0jiQ
z6QF=Y7d)~))T2}9NCWbBh`4P(=gKDk4W0otJow&>&b_vNXTpJy9z#$jAyj)@9``fh
zb6IO=9+5cWFdZ^#Rnxn8+}~dL56v|Wc6L%gE;k@<Ely4xL`0Q-%T9h<(B_V18x$#)
zLvImRQL0g9Z%Do{0BI=i*Y1ul{a~S(@Z6BBi-q26t)KtmTOHtyN15w+^OgnGCf$NU
zV4!y<^n4x>fN4S}4*y099Xl*M5V2k`Rr=u~&w-lH_wQn{p+ftsYlMl+s&uAUD3wTZ
zZ2MP2LdOH6?%!F=Ol?`Y)`iKqoRa1~HXYI`&3)-OdIK;fx977qkEBUYkA$|4&Z9_o
zSm1`<4-u_vXozvq4io_(ih$Mq^do3)`apg8*5NO*oSfV^2qOSiE*9Vdobzt1;rKKi
z5fOoII+)Z6xb9z0I}i@It1jzksP8uj0Q~ahOR4vrJFpTCfChPF=d99dHV)MNtRTRG
zJ?Zv9RLVmSV343deKZ);YgT>*%VK3?1FRf=*SoVR^Xbw|EZ1QDBiMtzzkM0XM>Zd@
zS*zx|{>r@oIhy|sGIi|bbOF(o8QBzLI$UfLWbs6}(lZ-3*0Ug<HY_^0zC2R(xWDa1
zEN1&m&Qep&5aWJ!H#rSb<xJ9u7tID=ZS6yFw1b>5Pf>*b&`SBTJ0-bAt0KyRGx}yW
zR;9FCv4;;QZ*;H$)vuRc2v*LNy0HaH71l<44W10{Bw4L36Vhg1!yBbErQTSR#2)4o
zKoYSQRY-R?lK;{-v5J4z;PP&GwfR=<BvxFjdA~qe&HG&{(k(cW%R$<F$QNqyG)6B9
zTn=iRp;n@svu);41LU==AU4i)I8A582$Tu&9u(e>K}BFKXFY~jhFSJ`foNg_knt8m
z%v;@Z=YpH|UKh+o-6qesF3{{k)1@@yL0oLC$?_+#!+u@CM}Dfx;2y|e84m)=4z~-7
z_ZQ#=IRf&;A@$t}8kn>N6@>y0yzi{7`tVi%$JKYobKQP%8@H5Q${wjGTS&GdL?U}{
zvbStLc1kFbY(iv@WbY(<kL<no-k$UAzJJg2dOiN=rHIexy3TdZxz2f?_nSPRe1#m%
z<%C4Aw|^v^!tVYDX}jXDAITS*+d@=_T+vEDU1@i~DN?^QM1(mqJ*D7w?x$rLK~wLS
zDe65FZ#Z3u*f}~ibT{|!dbB=0{C0L)ZO<loV8y_En|0#6JY={lr}J|80gcbEU%P+W
z?7CIeN(^{ToU-Zf(=kk(0}*=f{X;Xii-T8Mqc}C$FaECK?1G+2(<~h(`OYlysp}+K
zECJQ6X}9YICDv2Ri~flvr13F=PbT))Ocj{BhL;c=eUe~O7a>8xpf~d4sf&BZqvYiQ
zRph|o)-gtKIe+~wJ+QaY2qrf<aqaZqjr1x@V(}gr!e?60V!IxNfA^!5l(b%Z3D=+9
z9nr6z=cIeFKiD~QLvpBLHuUE1qBU<64u1PRBbulGB-v%LVXb>*D3f-?<H`SuR-6>F
z#q~xQaw<qlE9_Qsvo-Hi5{&y^9M;!8`HY}j&$Ik~(T~1dagUXXb5FXCmmBNHueu++
zE<L2b{;AzE`pwgo>RGo?@CiCHB<)m<<2hN8l&k$+Etr`@%&Z*O3hB55g1Y>k_|q7j
zwUo&m+R{G`8Tqs<DB<VRKOtk%z8%(OR`eM7&k|_+sq4fsoB2!MOSCoOo*u8^@#N_I
z-W(K8)7+Rcw#`FPmeVou5R#e%I<2&HSsx5cB#5eoH{w<T3d#`jE$<%ty3WqL-&@X|
zhihHQerdf^KtxixS&vH=BB7<Fr3TFBjhzrVrq6e4igkDPvZFXC-bw2_y1X(*J`^D(
z*Sjbqrx&)fQgx-~*FI2_pL8gcACSu%o&L~1X2u#XzW<ffr9pv?8w;FVpIKLCE!|rA
zzDavw^ri{RZk~ZgY@nl%WvE!1LibmPM2y#hecSSJ+ConSKL0ZzQNMhT10&3(uFG6n
zAP|vLlI?Bys&>eTa~>3cyQG(b;JKAPYi8eKvH<Hk+>C50b6U#o(J)YZu&5%%xINx}
zIZn!FBk!8$so=)D<u-^>l1-E8&m3vsY&)dmO-<(R@dd9RF^d%W9#cpf1JPIN8G9(Z
zFxBJPBSpcBVL`Kzn{cYYGJiL`lkmh^nE=V%eW-^8Q(r*qcLA{`B<KA3r3_XIPH>1U
z$VLyQ3~DA3eu1NiGu1a$b96itH*IaBtJbGxM*rePjamrKuzQ{A^Q0(t*h_IPN&Edn
z5%Gi+_X*NSubc|v2=%PAG_LwDt!Q{2c1)R~xq;u5Yp_^{MF<MiH=ijjbz<gEQ`|9I
zIvnCs`lf<Z)YAQM+TV5Gf6hS>xigeGwN}1qP1MJfREj&9?c<}YFGZZOl6_rLjzO`p
zh9q8E?K-Ok`4VH>=gm}wQL80_XR)=!!biasNux~?xtmgAWW?UG>Be;xo%SGWY|v@Q
zO`*<h7@;tlvy7(89eA}LSND5vCT67$?#q5`Qc51=^za*}pT^&ldzo(CU<<<rD6YW2
zp-o1r>d-MVO7Y2~+enLvihdSBgwYZb5*%Sl{7V^Rh_9jQc)Y)6(OnA5yKZwSJ~TS-
zOwz1pq6DXs>a#%B_oqBeF7!m_Dc4;^s4jRf`6YKv5rlRKKGYB0Zhv>V^ppJdoe-pU
zA!^c+$!9O7x5~M@C?GMG16#lyTg?^!oUt^_Hu>Bcb3kHdcKR2h8zVp4^_q4qtL>&T
zdav)K{6=_uN;q|!)#s7dCF^2@vHdjz;h);wLw@6?Cx4@qQGNTZQ33tzY_k6mo*T6T
z1K};92j6~+IazDfbv$Q|B(L9D7bIy!%-0DLxzPk3h<Mx&*i<}u=Jc1s6wZndgt5G$
zA}*1OQ?8`%k{OEW1gQIsMh=g&%+sfrXPUw~d&8#|5)rn&FP7p`&z~FD`C>V4jK>4V
zr5Z~vn9s>XAyyu9ut~pNV0YT;MJ|R;=9S{O-)&FlPC6%8><?4q@KTV$$*uR<0ji=7
zC!-{cQrH^-b@P;^t|~8EDkp+j0|zqtgSIw2)31a3zsF^{*O<+j#Z~XDsy{BSWd49y
z<QOY4^FgrSY{?ugbHLVP<thEJ>5bMgst221LizG`wg;E4+5Gyne#lAWdOu<fVNrcs
zuIMMNvRE@hEZkLmt+@o52A3|I-0=NRbhZY!^3nVnzx0O^?q&Xl&uH%?ZZnfq(=%Vj
zMbYtIrn6VCti|n0mrEQSz}?!~61+IwamA^G_S9rBd-d-#FCY#85MA#VJ{GlnG)+xS
zsNd07EyI7EE%0oxH=YTgz$!7i7aD>i(TAotQl8dGK)Z$fO5Y_vzLdJGr$HtojO%VA
z`b#^<?_F)9W7^~~T#FFgn&nrY%Zx}Pv-<AIYlQ7mwORBcwZ8^rmp%wA-?1#WGr9VG
zWV%RSjO86Kfrt+13q&_N4D?i<8H)dslYM70IIk>y-!TYzW8cy=fu68Q$9y|vZ6zw2
zq`{ZxY#R%a@Iwz&!g3~#Vj$C@=4LuS-g~yoZ{;Jlzoz!AKEz@(KfN$0k&))W$DF{Q
zW~_ZKi1=&cUw;C+T)f5#lUQWHE&k!d2Rv&2?_Kwyw!TmwfQw>6CJ{P-syd&`=1rh~
zYOx|kts&YI9bUkyWcF|-1yn5gXt;YL>8wj;&000rbY@=^Yc%W-&u`DKJ~%uc^i7f4
ze9XHmgeMky4{K~3Tfa@}aEpIg=JL8o%%2I`Bn>@3g|z;qt&8pEZBK5?@vFHnYkh-T
z_%2ZeE`*6)G;FrXA5LHUY1Cb*&EZfjqO@3V?krH~{80hV^XIdr9E(gnS~P<7aYWAz
z50`@?glhQgH79e0q;W6i(Xg1}Lr{+3Q+%pBFIKW)kx#zRoskr4f10d4;JNy*HzhhC
zAWqMWkb&B+dYL^vkg>AL%gX~j9yM^W!_dbA^S;{d0{^N7X*#$)L2(#W^X$!pDb*0t
zyFd?pTQc-|>Bc@i@_|1-G&`H+<IZ=U_>OlE=Wj&R={-vCRTse57k55TctD0{f5j2&
zM&a2*<CHS@x?tbP2prRS0;H+P=7i!{Skp1<lJ`LTaaZNo%<9Az<!h&lx<AAh?_Ib^
zQXU8UrQn|9FLxJ;vC~}fq5w-mlQ~AdNv!><tT7cxmUJq#L7Hshc=J&?>7pWML2+@q
zLdc_-68x{)2R?l^J=H%Ku84R(`oodVK_G2`M@q>S$M=p$prWgBre!$i(A4^uwkk=`
z4tEi)Ev_?f%#)u>e%z)5_b1kay2M%}9JpvJ=B+oYuF2^XwM?uHUQsVq9X)&Kd&-9M
z{@@y+Z+7;V3uL3_WC6(&BHeYXsw}zs#h1AkW5deH$qCifhyfbr(uqHR_(}rAoSc5B
zFK9q>49X^z=f^7yO}x(QR8St)`CtgsIZ{$m_m8LsKE<yYtzBQ|I~Wem&(EzGX$g8u
zf4uB!oMO-Do<wrzuEcIXZkpSfd9N{=RXH9M0+7Mv!Iy>$?+-4F9drrNePbMNX1wXy
z7&qkx(Z|3lKZk??@ie~9c1rn{z_PG-={LzW5wfBlc#u7}1)F>+{hZYjh?R)D^)PqA
zT_!~0HryKPqzuQ~OweeJt9)a(<VjpZFZcd7WM60K#75uNnST5}JX~kGfSf!{O|68=
z6e%9QMS6)ubo8PntyYsNN5$fUK?~eupstGzExd-Yu{Lr}RQ_|PN_YYP^Dy*@9SmlF
zar;OZ`bjA6vp0c=1-2w<kpUSVAD1f2;ShPqfP0ZV$-&)+FRLBMYX^K6K9HxP;y*ZK
zs)Hv!gV(-mutE{w&dr;(t5_ol1|}xCfRm-q33xkQ&%GP4^t^myHAQ+>R_&G7i!ts~
zgaFAo>gvg?&A_sb?^Y#z@X?BrGWNvYM<S*B>EY5Dz^r_VnhtEa*7>8C%u3ylCRJ0|
zSosC(8p8G)+Z)~qCmrzBm={Ne@Zz?WO205TRL)bLzEFkLT98Ff8|t(vd7TZ|AI{`{
zjF0wtUBa%6xjc9NPHVKXt+TVfl1WvWXS&^E^_?_cv4)>&8dFC9fElGyb4huje%*qv
z^IRsDFsUH3G9BA7juCn7_`AWFxTf)<_d7v_#vEk8>3gtd;h#?xFB|1vRs^P(qY?~L
z_9-8DTyCh_Z6FsL38gP>Y!EpiS`Y%6t;!|3p%uWm2}Flscj)Qw!_B7XVv^wUWxbz4
zSFzT!c(HUWOl&twD{3ta74^O7{ja=}FtqkjP3az0L?R-89&x=svaWj~8n)h2(V+5N
z!|0a$hbh!`k69S`INn()aj|fBW^&Oo-Y6$Sx#_MtQe?>xC#`}L^0nHC%!jxcCd`K_
zp4e=5#^X&@U=d#tDBI`=VXBpQJ{<WZSNF<H*E)Xo;py{ZA%Qp5AV6;6yum`Ek>%xJ
zJY&S^$en^l9gQxff#DmYZ(~z+vM{4ctr{ksswLxnow_t;i{kx`R>oM8xKC$5ZPH6=
zp97;pr0`K<^ujmgyy)wZ+2D7Q2UoxgL}F}UuCIH2+OD|g$WINWvu&L(z7F<G$Ie3V
zQBrFGOaxU`p026Nx3=q2iEiiyxMJ};t&%<R;j*zIwNy}4L~$>nC3+p3B@6k0a*bD!
z@!GZOb0extRRo!QwI8dyeqm<vdP`s-OKz8+1&m!WvU)Q!Xk%h)+DTg}WFr)LhLIib
zRhBte%IdTqY9wh~Tc%{HXE5U1b-naMK$Tx_e7y-BSS)h--Ft&X=BFvgBk$Y0+sZhe
z)%z`-_nu>$DC(<XkdI6sZDpB(Z^!<)kXym!FwmkQw4GX5a2H)CTmM+5Z*2>K1>dE`
zHGWA^tRc1%DpxoYwxMT&aoRNa{m~l@-f;h_6s{1>SB$n#S7a|x*>;QP!{S<tf6hBx
zSNSy#3W!lh>z+kSLREIWFVR5=4v1jqlq!c9_Z!kZ5ii<W=sbG4o+bUkRt9bN+ZW4;
zc6ChF(a5~?GyQ!mwnY9qWR{el!?}@8koidct}C<INE&;^!;61fZ|OTV^mI-?_(FyI
zeM2~ZXsU+w;<Iw_I6CEncZB{;p4$hWuL|UGkQ^bsAQ2Dy1h{Hdo*78aXY&SSRX8(#
z{4Quiiwrw1x7UqP>UrYfaVBqiDB`Ys=^h&D$#JezZH-8uJWq+lKsLe#lO>I9Ig(AQ
zmidh+7x@>DL`}3xHiX~bNURZWnb>NXQEBzh@^-pgd&G_57jt#Dt07I~UxgLn?eQa<
z?0C^{o-Y=}Ez)n&ytwi8N-MCnK@(ogM;8<s=uu?kWhKq4+OQh7%bOF>>UfZ5dnoU@
zxDi!TdSZiA?{7+BF0Qt>Qw~Ny_*SDH>^~Ov>-pIwN=<<|`-8;1DB)Vd!AVD)A3yfr
zxw_kZ5DPmkl4o2Ww$QW2G@iG%3wBd>@OH&d&beDK$j)i9ebab;LrCxUBH^Pn-;G=*
zuFz*<CjR78?~Rk5F9;th<)}08eUf-F^W*zkkK9tS0J7n$P{|%r^-$jvk3g0U&&^IF
zxgjoY%w;*iIpUD(*3IIG8VC94%|G;Lfgy<j(#MOa5+o`aR4OH}#3~!JkZ|h&_k7Lo
zac9q;naKJHi{hj*_h!N$@^wFlZG}EUPW`7qpHJvGgr&l0<GvGtY~}X3b6esu9!pK;
zbru9$mDP(O@9N<ae|F_1rW?dBPrbFT3faFJw*1sVljp%zK*H_FDjshBSpi#zul<{t
zz)*$N14siEwr5&@mwlwu<Mt)^>_#w-@B7udk@oY~*I~C{wb0~M>p3X!4xS|2w?#C@
ziMab8YJX^%IX6$*wy}e)jKf7kJ%)=5a${_Is+leih)zjD11tj>7bdH2W*DiSy0U$3
zaMypInR39D2RW3j)cPl)FBrV@UKi~;?nK5zX@-Ut!8swjhPl6wfhjnE=zwjdChMB$
zW&QB?Ru-jh_mvna_IDXQ(cB+YY7u+ue)0;c+te%0Um`o@mJ4x_WL#zo_GdIfraUyJ
zu>I5PG)*F+vzW*@x3!0z-7|kXoL1Rk705?L43Nb?H!YaV@QE_V*zBD?uqWLuAgR)6
z%A4qTQN5%grT%#}rzn4sbF_!V<9*#_foVt-c_T8~WOL_!M((8Z=I<QXw^rR-*AbS{
z-c0YwG_s(o^q?SMAQ6sz$n%U5S?vh&X2ERGjg_p(#e6Fi2Evsg(IQv-N^^RMPtri<
zGMlrg?B<f2njO9le!vh5|1R0jqPz~CvrwFW4S^{Y+u%|jWW!JBM!b}HE*(Wyzh1S%
z*PfzNy(X)BdwTcr@&hxIg41>E3s194{TRzdn+;bN!x-v@6k9M%k@nHZ=tzm1!nmP#
z#f&bY0l|GGzbAdzE}53)sprVxJ<6}UnN>Lf0)2K2%O1QRNM(s%q!%m|v^lOnxeUoh
zBzeWhm3tR#<#;tswSj7;t2ldFwY@5gvJxis$#<T%FrNR|@s}!|GQHLobkpdBV%TZT
zruf96#G5}Px@eln0ZGT3AAM8RvN|j6^tX27mqTIwk~w%KAkFtGgkZ9Zan51Kb;ao&
z-U>I4nUxxKes@7l<l+*Ej<3z_$Q>nTG*!x{eRYl8yr1wiFwH~o{hB;}*IwNiCFV-_
zTe<NkIv^>+skEs+`YKMs+G~y*y=O=V9SB39S~PY;K;ieE9E9T;>>o_x>7Quku`Cv_
z(iM5_^j0ucC$Sb!f4<OgUT-I*t*%WTqwLad_GSjL=08KjTeb_;m8a_^yu4-CDOK|Z
z?ZCTCrus0@o9e%aa7wTh4k5<=p;27xK3zK?aO7R6HrkLNO*=!qB`r5Y-0VuIkQE`i
z$%ULd4+nkUN29#abYR0+GyAM|q$x41`cK1o&EWRQTR#6pVYFDAS;__FGA8W{t~jqw
z+4|<R>pAlFL&JO=PRj~z`?T%LHo~(*)fdSl@$YToQa&<U`JHgN%vQiHhI2{yKYzg_
z42ytE!g6nCIvY~4Q_i07=BLjBk>yb7EG&nAcrl1biOqc4*5zb7tSgV+aNa&XJZFW(
z;bV^mdd9SFunKizs#L6X+S};QM8UGL_&y!dorVs5r4PL%pmG!u5+mzuy;~Si@IYS%
zE$&&sO7!~-<>6;ag?<vbE!lk)eCD-H2@x6pKs&I&*>+kdS>c=M;48E>1W3hIJh)Bd
z{NeY%;Ss>illJPvGZ)o+!4<r_m)Dzr!=`iMk%Sx0Lg)s)@IKQw#uD3|jCa6Wvlsha
zl?3zpf2rlIr@@!#{7VVb?o250oNM!PiOar=NIm7cO#7sLct2HrBmp|r=z2CbHVScl
zBO<q>-tKToGcF4w-;||W!@}9_L^&iXe6$%?#2T+~RoR}Il02CgCWVYFzo^KguD`oE
za(<1r$VKblGxyW4F55XCRa0*fyG-|lv9x@FWhm0pC+nWQL8+?}r^tY3hbf2c`Kmr!
zAoXpwV<u=1NGoT`S9sx?589|_o~Ju|N$~Fwx4uF2*AhR((kD9edAd5#-u?H-`x7zT
z&#GNn*Ez0sfcI-#Se$)Z6nod~{+sv7s%)+g`_mlJm`<?N?p2(~y>w%u^g4~G@zQ=`
zAS&Zy2!VX-XHpcZNoQy4$j0-vB8m{iO-=KReFu_upYimsCCcAeIe+JIiWwR5j2vnC
zqW4w0|FISic&MR20vZf_|2>1tqh^IR(aUzU3g48tZ7A{hZ&OKgv#b4PMT;qM=U~4|
zC{!LBb=*If-nbF&!lklAe)3MFk|M)HXoe79yDXBZJyU#xBNrh8Id4{;TnR(5-q#gk
zUui<`k_uG6khuB8ETpTxB?{SkXhR9(G!<q#Nq%pV9`;P5dv^S1vi|f^KB`dKQiO1m
zZFx$Sb{+k>-7WHtwCNSy;Nd+@HI|Z-#)ONrpB3z{6ko16?fT9=pX!y^^mw&6m*pX(
zOHIP<YZF^{TW9oLu)*>o+v-?(`OqP?8=eA>SGo0Z=VC3)|A#+*6#q*JKf}a;s5Rr{
z1tpX6H}rJH2zEuXcbm_yqpejCe%EvQ*T^n!>4l3AKxs<2|Er=6ut?Q4I!Hp?-px@c
zM<aKG>pq%X*m;uDnK84k!`=PuKkl9S9LxG1H&2^N7BM01{C;zY$j__$21(@s3uKs=
z-4>@1Q9@KivZ#)4-s{!RpS$$WNd#N75;aRTMlBcnbICM?$aNQ7#jFQ-<a@%p9=xKi
z>ntvpsico8)hRKfdcKWh;Iq;$Df7JfHyV5pi~2TKKh705;riCUwIH?euHk}Fn8gi3
z3=M4gJa;a^%x1Sc)93qoRk5T{Uwx8>sBj(UutQeHBwqVoL5nBR*c$@4^ra9@=d>9^
zxhU-~2RE&pNOscCNvmXW4Kd!buaAGlI9BqJf6!%Ya{nUU1j^{R)VbY59B$jU`poe?
zoXA7(L<%<J33-qa$VY!)KD?77u*Y^(V4TtVv9l*b{BG%If}&dy3kp(eWmlddw+jl2
zO!_o+)vEMZ|0Xs}Y<}alQ_2EI*{w!i<U3fBUgM2(XA40Bw2gZoqj*c^$Sz`!a1))Q
z>xTzx=(aQ}ixo-gja1k-mmSLmbz!HZ*)r@jb=*!Y<_Zr8>-Fb~r*W;LdZzYX@V-Yi
zo1(cbiWR&)ye)NhZrF7rs!7GK74lhg&b_+o<A(rDQEqMaL5#9wy(Q?OFjhSJIA8kr
zzAlm-F8;(zd2xX=%}ROwL4z-|%T3-Un->Q@6jAq+>wlDn?EK`irB7M(E5zFzQ~pdt
zAaQ%<D_a`0Zs$Jc|E*WKOd8$!y4|poSh#ei3>(wXkihYc2l42=?qz``!u^^LUq6Yq
zbj1ER43lMjPk2f|R!v|Jhn|<0Oh7;YMDJ)9sapK{HMx2D_&}A)q2?jQb1MZ9)Fk|B
zUG>^<WAbM`VZvSaxd(?UmjvfAtpsa3(W&}vpW!YjOeMzV3!uMyQDcTZ9<xIU{j7m%
z+&Sx&4L<Qd$W(u%dRbap3*Wx!HBB$wmKncaH~1={Dhi5F`G+(27e9&D|5BSX>1)wG
zBLA~+`mQ6jc0IPNs37+H(yvtYEIWHAF1qHznR)eS_11k(3k9TJZtb=F-Q<r^Jx!$v
zxGuv?T<=o~Cns4CnF*|+)G_(npHQ_YcLmL20*lTcfwhr;Y|M~e<gt18r|vtS7@#ZX
z`G;?Tde$6Mc)zO;KTySwHuuAs1-D4pQ8@aJWY3~h^gF#PPzRQ4-+0`fET~ph%j{27
z?*U#yroH^R+EL?mf^`i-5iP}PoXP3=M#bCXU*Ua*hK3^@z7CMOpWRYrPw-g|ebH2Q
zh>743&tfQ3mr{B8mVSQ9r20*Mb`#c9vEuPF6}Gsx?)Np5Cs{43=%ZC0WV_SczS!>`
zr<6U^%@krjx<9iLZ+Uv;FyL`;->l-Ao~mf_%!t!kXO|uAl+9eg^j~X!(rC<0rhYQ4
zkXAZUPd$!{(PWUT;@DUF8!clEFEP#pIh&nx!8)EE%vh}12q=wGJJU9VnqJxiKO#}J
zN5!?5WTFoQ;CWG3Kd65;wwK=H>`~m@%oXiler49bC8go$LM3@u)gs;Efht>^3J*o>
z5m|=!Yb`pak6EN*!K|AYec2=aW{Eb__CKXyT{@!;-AiwfimSZjr-{s*L}*`xp+dWB
zMU$;9cgmrbtmWG3(#h%_Hv%&5rAvzc0~54&SH<$4c1nD?efw!8c`zgyI>P6IeP6h&
z_z#jb^e`>bVGrQ&F_m^bf6ZHS!mZ?>MDmm^x758PH*!*fr`-6%>E1b$M0Nt+!PD69
zC;c$~o*iG0i8L%@jRRsvO#qA`<3;A6`O`fOd~yztx5bKOcgNm#E>+E1`TS|+yKis!
z=>=?WhI+}HqlFqtUcQR0r=4_yJq|&|ihq@%i3pieKE&e=Fpsbg{$oNsKu$i3Eihf?
zbb9M*g{=HR_x4Gw5GyOboSYmL6;-M04g(JlRur;|!o(8G3DyDH_z-&hJ8L2!tp_?Y
z1^|WtwE@KnC#y>YRKJ0AeQSH0iQ^?G8<zr;7X<%13=EvW`^;lK`n%X-@DP+L76)=D
z-Ks#nZ2+ZWL_j0>6fn>5wfTZ38bbhpbDO?LL?vx)IgpF9hoB5z*NJqpPn<&|r~3TZ
z;)ms*aDjPXf;QP*?B+HLzbJS}5rP80H3pLsU%Te3la~&_ly_8o;eeA)QHk>Uc3;up
z)9e<L4R}A4|DS9A#gLw##r$QV;_A~rrXt5NE8LCVSmN6czy5VWMD<Kx#BP_l&IJ}S
zJi&GD_03R5I8vF~_~jOVHOy&Dz5?E9xkTY?^`bxkDS*nI(qt*<=_K`=gMg<}!U$B%
zKwQiS;2W;ig!uUMfP4=VJ`Ms-h@MR4EMR-%rv}swFh$(A)*9av-*pp$#Q=0;5h!g5
zo&W^A5zr>6NeW=RKEnAOuvhZ{A;XjT%;6J&OFCb6)|2)O1!@ovCb>#X?3b5kvtA8&
zasb2qDRVGMFogA~|HP_k3fK!kM%9DX*TfNyp?TXh4ZFoXPcu+q^e+u_RnE<{{P}Wj
z>}q1pE-Q^QP698QUs!F@?Uw!4biMvpPk(o~x8c(Vnxvg3O5oBVf33Eq^Jt7OXK2_x
zhBPAt+f>PUIx;0n*Nfbnz2>yA_6$TJU>mNgs;X*~bW{+HLWdz5x^ckBXEagm32;t5
z4zj1x(gpy^0O8VJcvz@F;R;;<iM$ISh2MmR{?g;n<EB{pQ*;*q>c&7e&J92isDi#n
z;^RZ=^iZ5|@VCuhlr~9B;sU_{$%AnK(bd&ewE_bys4bJ>U4n@SknuFxo^3vQ*%1nA
z10Xa3nnyQ)SOF!zn%--7dT0mLR1^}hQ*cM-jYIyEwt85a_vDpuFlp`;Q+|JvreY}g
zDYdPypF(f$ff_01ALNqp`z(B=j_63rDEuk2mch=!<V0G$9qbUCQ2ONVp%+<RxLISg
z|K^1M+|dbE<uQ*g`xFq2e>nV+CeoMZHm;_m`_4z^WCCh_V7jmUmjndx1FjpbiOBht
z+N`|lm~^xlP|N@cgus%Lhd{matGoNYb7E{P<!wYV^hw^}PR}#;t(h>9`PegeO#lv|
zXhf<F)&QW526|q02`YH|`}^FcUEWZc{&XnJ0EJQGw>`1|%kcxC7`Jm`P7W&ymjpke
z!$HQ<l4ZhU=eac--11t-2EUsr63`021Ljfyq%NEu?eu{a4hgbcEy{0dMJS9=m**ht
zx3QA+6x?fQCExRf2pHmhbiYKzW1i{R5v%n|A04c(oqOp$V%(PH#~qobQ<rR*r?#`>
za`@!^-??I0DzSh0wOmt=%~1B;_t=0=Cstf2TLR#O#2fSg7Dhy1b_6ZKeZ5Su<O0TQ
zU_Vhxtb&cmSOq>A2bwq7(sj3Un7^s1HJDpFF9V@tmb$8YfUF)tTLd6n!nC#g51R>s
zT96{yzHVT%v%NhH4PtaXn4y^?LA?oujMqRSpTnh;f}ryN1{fXq&}fKq3eGEM6N7vX
zmV_*&scpkK3mCo2zH;b^9a-KQvK$Nu+Oe_^Vf~a0PMjQOQfJr0mWQtYZuS#l0&6}`
z1;tnM*w5sTGx*CRD-*NQP^gBsL=pH!^Rpzz=wH5kvDd^tb>~@HBmV)DKlAa<<BS}+
zbxB$|E})()+Nljvju<vD_99}~6gmggWG~^`jdEci6ibF67l;D8r4vwKcncon@Rs1V
zhKZcJBpU(}y(2x*s|B!6Evzr~!=fI1)|#{ee3?&3c)K!*xFI~^v^9jDy7nT@=91~8
zH<WGJg!}Y|4L=8IgsuKj?5JP5E?AY-^`(K>IPu^7i|EVKbuT_#5Am0em<f9^jmdBF
z;0l-^*sI!9EFZNP!Me~yL_`Rvcxz!w3P8r_li&f+Ct)<A#aRY$-bvT(Y)I{K75i@K
zIXPjm)2z+SHNvpOXHH`r5D749{ltm}eEb~x_5bL&uN9991{0H#V!t;*FP^@0=MLHn
zn7I&b8r7L(6&9`@H6Yn{QTN-OydjEQ-l~1W>-mUr-R)2P_oVO}&NfA=4Hb_$j_;IQ
zGbpv)zx;|~UiXd4Xkus?@^Qf(G!#`n6@E?#$-#H}anM0)_vLhHm%p&6(9Y5RNn@2A
zUqXH;-U;5jOU4G$W-2Qr_18CaA{w<et~8eHURNO*N=C1!IVIkoP#DgNUN*gEis1LH
z_Yh^<9#z5IHXtNXtu2)Q2lW&KZVHDHuIt=8{k@82<Rxx<<`4%>2g-0~Ha0qeXjBbk
zmjD<ETDShN4k&ac?6{bV3y{e6GwoA>ICE{hsEyPp+xKy{7Lq$4ZYOuT9psw#-2{!x
zZ8I<+z;3D@3kgP0w>@Jk0OUl3m}JEcq_j4W1TuNXcvVn8`xh3W|Da#9uP&{Sl1eTx
zh`ZO9%ttqJkB>s|WmQ1LYdms{S^z_!qcjy@Y7f}=3u?*JK@;$wb(VUSPY~)+<f$m*
zgO(PHRw+@HPAAAqNjrK3leEl~Bf!3MEwKo2<hAkVRak$Z?S%#hwnRh5g46Yqjh<c-
z{SNAkw}`)#pnwt$`d$3pY)iMK$(DTX@D#e(=8l%yj_S!qyuWb-UPLZOzIQ84=X7sM
zZgbA~^9YfpG+%?gq+pzEvO*3Mx5t-P(X;B;yu7?Xgev^DBoaz=^=h5-geP__j8frJ
zO%$2Pp@H6SUByN<ie;&lS-ISC3FDEAi_4o|2aDa9-r(o#_E*(WBEGBl0eyrSaq<wz
zbUfCm0V|7Ax=vylSGi_<wKqm4p$?UljfUM4CH&If;XA*ccHy^S?PP8`FX5zeX`z$q
zhZf=sFNb>ekKf69-Z0@?+zY<NUV1rDA$!M>A4|CEGSiK)suOPWkXO%MLLVJN4~U*n
zM3uGi(~C8(E-oCf?PxJPRxRN7Ax<@qB}PX#L-+#SBoY{PNb4{LqsZ1MMoIZ4_c8_#
z0lMzi?c1n);zQYuh!q6<z~x6k6^3GodMYP}9kFi;iIM*C?vkqp?|*Q3wBhe&4b30#
z-Y3HYoY4P|@0E^_RR8&%VF~rJ(>~$0Gckqo{TR3mvp-A8oyShq{(He9Q%ftd(|+FS
zB@w^xtaflSvguP9K*sOAtGgU)UHgSOiL2`@Cq1J-S^tP$J7}axB8o*FqZZOaTSu1D
z%({mY9v@3fd7+Ko1Yrl4x&$^tVc{AH$uEc}xMRBMvhkuOq@=Up0qHn7!=c_7n4aL4
zS$n?(x2c3xyF4N&=rVYhN8qq`LTvgnuA&$i?GOUv8@(F;#UN>bCKOJY`b5>nkpZ1$
zBi?`2tqAOcOdY%nSI$UtWgoFO09RM@5RXpIdT<Yr8X+uz_hygx?PO2pBy+VKCK|5k
z>dicwn4g{wPe0^yS{=I8l%>Qp-x2=|&L-^P7ch2HZi;_VtjG<Usyz~HLYh?AAlNNP
zVNOAB9i%PdYHJ@s+m(if1_i3Q!K^|=LGiYz$l=Wl2#cdYt}xC?F6_YzK~^lK8wN=6
zbv(AjAq$9rQWK1S{6{+pFxTnT)pzg@kSg8-HNR=F^U~7Nh-I%xaTyDQv4p%5h3TPi
zmVbN9joK@YsEVktLc3=cUiook>*Ljpcx}vFw^7wp6w>R?Xesay{Jp#A776FjEQ$A1
z9GufX`gfN{vntJxH|t|c{uFN#5}f5@o(-F-QRb{V7Wl{rEv|pJ*|fR#w6g?TzRH-I
z*;p%I7eUU7`<nW4G^4<17udQ~t^ul9eBJp!fa$upyZeX2dem8kX~jt11fXgKQ0eZ<
zV5b4!=W9i!5O-U$Yc04bj$kxQ9YE1QCehY=l2SSYhbpLWTD2;-@T&E6OSv<yqJdHs
zbrIRO<`T7&F&3Wl!uPA$?+}b8mc$GGy_oNYL#?#E8R8~84*%}@G`i8oGU{2oot_YO
zoWA-T6rWz|%J-?S)M!;J&USC_X3}$^WoH`+y^7-`PD4)uy6kKjoJvMUM$~=*F!__X
z3!LnFBlK4kGi!<={`nR}u~^mnU$NeH_~M!!s^I^ud3lByyiOD#ug`vSCRI+3`@?*3
z7C&b!`7I`I2Hj&Q9q4sFua{jnT+WbmHRx)6=Q5+nO<@*|nnZw&VjRtM>2Y!)9427@
zb4w5W49mFRpegq?DylfmQ7g#OnN_kM<y1RlLL)9f>LWvPw<+3Z4t)$b?I_sfUpzA6
z{~K_NTc&yXEf2HW)Tq|5jSGHs$l-}Ax22y+?pqrSFS1G7l~~XpR+G!*;MNI4MPRw?
zk-Hq3JZ7~O&PQ{<Ec{p%q^`+kS3Hlsy79xm2^l1G9|53Q;k@w@GITT)a~(i~xGabA
zM-A9&VRG^_hyVi_b1hO==L6TP-lWHvC@kHQBWl|`jtOG@i?Xm!H!e}huy}fVmgs3x
zLD>Y5QW77;^0fI|M;ZOP&5EGZU8#)C^yh|`w+5CFQv+#8atx;A`JdkhiO-olL;rI;
zbTpR%ZJeuCaLd^pyr(Np#RUmSkO6W{_T@{20W_g*^YQ)ZSwi{Yqta4WPAH@q?}68z
ze8Q-Mgwoe#@96p8SF8pCkgBd^XjSA1QvisG32*E5xz~%H0UfV>juhe2q0Oc++{+@K
zNwa+ab%y+1^wcVAD_|~VbPcdzH+bjQ1NyA<6hI-dd3&yvH9nLG@(=l>&^kFT)7gsm
zxv!f|BAxJ~1M3NYcQv$FaLiE|ExNk3y+(C_e>+f39EfyKjbQduPnDUc$rS>MQ4Q0Z
zw%Vg&x9LBGnt#S7OSO3`<f{%P+FDxnB^&E<;A$Ld4)*sus!-$Q3;n9P#g5XjG*5gI
zXwJ})@yIiE&J*L^3EhCb`<p?e+zu7J5_D{op?4cZ9@0q#|N8YQ>MPq<&5f&#zb4Ae
z(h9e9k*?3BOg6U0YOg{RUpu{rDT;AnhU;(t>-LB$sgWest*owUO|C(hEgM-{S}GfH
zCiObln6UYGe{b!rioEUg;uAlyM&urUAvu2+Kz@~jPU$Fy9G%ioo_=Mq%a90H_4TKW
zQVVV}-a=>Mp^T)Q^rb82N?moh-YM=EoI<d6e*<R_PV+RE3$UjEL=yu|T^*E}dC|I3
zrHFyl4WkxP2?vvdmqtdjG+8Q|mE$a6Ue{Yil4w2(A4UH3CnG>EiarD{gkD2t_|X7^
zkLnO;2>0L;vQJ!Vl97@IDT}KZ@wMMJy>{sp<Xqd|l~d^JQ2h<q@W+i%6bN~wWT6(_
znC>%7wVa-vp}VXWGy!--1Z}9;f)tW%%*n;>l(MEDX1%}uwK4knDeZuDPaSVz_V<xU
zB$jtoRh3i$;1roC6MoF!{X~5GzVNUA8#d{ft_RCSuK}QWeIGo&LrJ3B^0(17kOGIu
zc!|emTm}x9gAFa8eQH|oRneep%`$TI+oz1O)Wfnv%~e0OD!jX2{Pfinz1~oFKb_Ds
zaQswnNg7YmMD-(qoJ<fxce;xXjo<mxXt6tEmHJ+?|2sA7S6~}Q;0idpHrE=}i(<j`
z@bLJ|V?_a<!$7pq!Tt!4D*l9E{MVANx0rI|JCB(Euu<#y)_|l8n(eF}K%FeMqiP9m
zKwE?vCTT2<XDvnHw18PMR3afmCmvKoFmLh`qvf@n(+l5WrT-v;9KU<NW)W0r%Rh0h
z0if9Arke?5yPVLP$9nSLj$gJD$#?M?Y-sRnYiq+e@C0G;z6^Q4E5r=%KYjY-eiC%`
zwi*oYQG`Sb;RFzU@n@5ao837`Tu>OgKX-|g^0Ga)jlgftM=}g=uHLsq)GPe0YLI7}
ziiTvu`fSjt8F;QW`Vfk6SIkF0<$pZ6c?;hcN5}vB69NW|%@?g5^c2;52R4W2I}fFv
zRi}*tMBbdAsdLAOT1$GtVCX*dn+zMm61S&+cDkf45cTf-1&F|t>puV|Bsz75qs(-J
zz54LhZHaf2zd|SxC|Jy_tXSA1Vi0*70j=@*qNCTyIF(tv#3V}$*E#=a@OtYVR?Bi9
z{a4&b&UrYl3Bk_(DL%lg)b85ZO}HL*S4;aid*aDtT4}rwUb5vRCuG<4r6^&w&!|f>
zSz6sf71}eDh_uKrsRU4jx?)rKJX_k_c@^$7t~Wk{$T`5QlUS3dzqE0O66#(D`_D4t
z$Iq?-6~T2%KzOo-LeK>XvjN^>qG@zeCwp3laf%>lL3XP6U8pdh_P9E0x(w60sRSN0
z09w#no%m#HYMOdzxEWjVi%Z>^WYK<w+|w}<Y`J4mKZPHn9Ax@i#vGymgi28%MLg45
zGsP7nr*kV<D>X?eDsBkA!0NO2Em;u_YvI@tw5W*<$ciQPOpI45r~22!+{3J)EpWP&
zHdvBY%mqcqVD`fO^OFrOC`7<0l*-p8@NH(ssp*@|v6vCl>z~ocG@jTUSdTk#l+g9r
z&5t${c^z)Aads*z;mNWg>w-Hyj>f-Mv8QZrXU$?$`p|!{6(qm#b`#w1`+Vy)E&fq(
z@kBeu-H>e$)@%T*<OofEJl0Ep*;rq~>&%(@|D1h3e{kpI87Pu=nMVe^g*u+jOL!8d
zK*RNkh><cU&l(@it?sUF+~cFjbv^e79xn(bs1`aAUE!4Z;K;2Em2Z~UZXEb$v!sz(
zSW_l5u@Cj!sieL;xaOxS+wlFS>8!@?<LFC=r?)JFo5uS;R6l!dU%pKkK=5!GpGD`;
zXS>~^_-1WER{>V$0tP#l-s$SK;A)T1DUqsg-(866rN^t(S}syr-^k_xwKT5eT}h>v
zIdS*aoe}crXfpguXmN3I*>NHHKX%{YSz1{Y6c@B$6{0-BoC#Dme^&iY>UGd-bn42O
z44|NoEjzJ^T+{QWy%e7yCUR=EMJ0{}IYeS~XOHD0?9y)mgQEH`|A~{iRCRx2=+IJ)
z62Ft$&uXjmxRh)4WN?dsr97uQpvojDI7mSQdE~gw!`j*8kra4aH7su7_*?7$#TQ?e
zUtfO{MrPAFUJM`h45C4AnYMqsa(?WIZ0yVA|4g%ZpB6B|`XiyeGdbx-pHl8!g==}0
zh*#(`t9g;(Ie(86eQoYPwte1V@tk6(UsV`4=S>e?n0?a;e)Tt{3>{95$RYPA!;U`h
zPhs7o_1ewGn|N~V#~SbXbByPBcm5s)f1)pf>Q%MwkTZ?0!KjnQePTc|Ih`Gf8>bj=
zH5T4dF2F{xt#!&CODIr;(xho|KNBofVu~SLewuQ3#CE!*%HK*rcJ{;0Vy@9F!B)0T
zbh&m+*LPikQyQFmc8e>>UHgOy*3tIXf=Fx+V%LQO@xM{+GTl$1&TNbg80x=Iv7@7t
zIk@6b(okRjA`5pNL!W~zC^*;<Na)K-_5g4m2T9YH7mN?DMdIrU?D!oGHSEbKSl@x>
zZA47`N1>e8uhRDNbM?5xKg5Od+@>_KQRV*ewTdNu^r^y2WL<@4=V;B)`EJ7h0_usc
zI@Wj~KIQxm)q782itE`88=yg%PrdC=6wqVgAnZ|C@My)3QhEHd=x6D)`tC7uNoSLy
z`o*GG(!p2`nt`*YUW>*2JJbKYi2uEZ2S0CF0PO{2T<6J9g&vR<hn`cJQb(Di5JFe6
z2uF^%p}(oEbt)jAh?Lg3i=6IHTYa`*$$uHebhY6C-Z%^#ELe1-B0JtCOf;a$dz8|B
z-6b2Ipt<=9;0oYA)h-2SfVXpsow|Q`<1sF_9tge2Bw#a{oL%xQDv31?C$5(fetiB@
z{7K!C(c+}BTT*W`f~J%^W0TzW-)4(OJSfuSK8=+~X=dHzG3)smw+XKMez5@e6GIEX
z^IF!Yy?zdLw8A$m4xV$6eH%%;!c-g`)qtx{_gFiIsC<j)G+s}~=V0R>dYzK5BrP)k
zt-%xZBA4JmC{q#LxFN2j6p<zd30w1O9JML3Yh>|supy7({362GTX_)8R5(8ctBAN>
zaIgGfLvt_IU))`<Z`{+s6e*WOcf~9*%vJ_yJN^yRU)nxD@GH3qY1StKMt>rbt;H<Q
z%V#?`k9EH!l?iaOiU(ue-uxpT)XI?YjM)5pc*4~5xc<WGqx7gmqn+KvshJ&Q^GGZ$
zHZ36|fR7Z-#Zi=&`v=Lz&ZB>yI}0FW0D!6eP!${W7VqwTIuh)=q?MkhW_^{!fbR_r
zjwM&EyJAOpFUr}LoSgh;Q&UR6IeG>(AY9{AF4JvqZXV2UIVVnanlVrL*?v}0W!xo^
zwDE6L)zcJYiyT{eV5o-txy$^7vG~}A^p;b?bbZ1Lmj$Xy(ti?uG}5}<#vRA^qg3bi
zsJ!v~_&e)~(SQEm5OQ#UM&+A#D}A}ufJ=qKmRee_U8dm(e9SJ>*j{AMXTs=jtX3)B
zU<Ka_-Js1bF~TVzZwaAzkfB|{10A6buJxi$VZ&`?wpk`YkUYMlt-l761v1t^$1e?w
zO_7Z~U*n(y^2q+`#ex6V+Lh#d;Y(oVfh=C+s<dI-&(wmhDOmL39Xek3BE+hVzE0+j
z^W&FKo!TZk|Ls$9+!@)1Mn~xRcQ1TgcuZfF_V>wup0UC`n$HH`sAKl=6Y)UBvVjKV
zI}PvarioGU3-X0Up*4=M(DMNPHJA*=r$o-<w)`mU;LbfDJ7xsr#6UHFm6}?cuEVPR
z#6g7VBY7oY;Zd~BKq+utfCo5gV9wwlZ>V~_1(DlGameyNWR3hhcNyH(@5L7Ra>7ah
zQ)lzmU0pXSE$Lb>7ew3ML$0u;V@_(;Q23FAG!~V6d|_VKCRmphyqJkS<$Hyd&O`NL
zn9YOo3Zc8G_wCZ*N8to)X$^InEj(TJPfcjtorAu={MQr@s@mRhe+^l$2L!b3H>HpJ
z4J0!Y`6ktG@W#<uuUHBNsbIdeJ1l9ABRonlGc4yc9^d%By6KNZ7Cwwu$!f2%uf7#G
z5(CO>N(>vbpzRv2Il9)9hKtgGg<0o1J&sw*ksQsE(KTqhNCPZIo-Iab?{cw#(9d=D
z^Vz=RGb7rk{+Mr$ez{^F_vrT-axV?i6n|}Fb0exZu%upFe7GU5fw%NCxAS6h+lS+D
zMWWh{DI{hXiKm(FRy}!9EY-d(r}VGc<8o1icx+8w?nvZ&GN(b<>v>AcTS_7%)zn>u
zLoMkcndbdz^<wG9eAO8@OqO^f?Z<Nw#YFQ(E<Ojqs101=G~VFl3amzWU<{Tb<ZCU4
z&5IW=*7u<k7tL$;B~9^~Xb?u(u}jxVLoy=B^XjrHK>#tOUC8X`CzjNfb_Y9OU&h|7
z#LQ9?cupsCt&2a%M78>{hA2+dLHun-exG+gx#I-2ZSz##)H?0iIXeGviuvmdMAh1o
zg^+yb74x_%^DfWRK!e~4XNyf_e_f!mxzKawF#7BBgMV7`jY(Pp3;!Yay%*$1zm5dp
zBEB~Xl=@WdrXR^)9aj~%Wp4E;MN==7{NuhM`s_d^cJ@Hh3`;Y~N}(s+L!@a+|7pTP
zbzZ93a9owcn9yrxuA|R9XXjsqT~Fkz1kJqoRF>TU=9Z!4P_c7P_MtT@<#{w`+UB<V
zM%7)yupPOpL_{xKT%lT@%Z81{_8YO>QabB+-oM|mzRW*;=@nIvW)Ai%hEegfNnQFE
z?bfDw%rxaHLXu;tvV{98y%r<a)-lk~k`(NCsg9j`aKlqWojvX))@~|<UG2TNffXQW
z*1KHu-E~b=HOfT?d{Utr3l0tr{|1dRXh%mi8gye43fd0rEv%@IKqOITK}gb4Qy1Y5
zLm})@x(tLV!Hr*P&X2P;d9(U62}Mp{zzs60rltl4o6~{RiuvwcNocctf*%BWFK^!=
zxn-Sk2T<9jxNqPO+#t7sDP(W4TMPj~VZIF$DYs1+K39fZ{{;+l%&n|s0bhs(prfIn
zs`$YWXxWhmHsgXqr<)C^h7nvJmD6RZ$Dlhs482Rf1rD63yh$N0eH&=KOd0p`@mOu`
zo~HXB9=fPTVb<1Kmbvm5dL@R3-MM%B{`2a(bDZ28Uw`x9&(5u^mA|1PXtWZ^E{%U3
z>Bgy^qDcMYe^`QqaUGl-W#!_LC(Sj8`TLc5bTs&g2o`fMS)5F0#|bh0Hk@<($qbLM
zC`v=;p~f>2<k2&)3klZYJzKlTh53z*JgC~6L19^=+M^7TB9&D;IH%R1ZTG<F_xqw;
z5#7#V;BT9opI-)IgKoH}c9gfjERqWE>Fw>$)fQMAEi2mm0rgw6F{J#}G4RawRXE$g
zx09SKojSB@b8J8hgd2WTeOc<M+!njW_op_MTY(Z`G}_bm>6@)fEM|Xyh=Ep*`XF?+
zbfKmSos*B(NlDY83Qjt{Kl@RRgzvm7SuBSCRVHx%WI{J(`Hx8=tCR3;Ti06m$y%TK
z*75N)Ts?cl-d;-!AMLg@|Mfh2;sw)_#Dc<?Mx<q9y(7=$Re4lPByE&J7(N&$+!c7B
z$SfS2g<tfEaqUU>c?D<mb#g^rRYAv`FYZ5>dLn31i}>Pxw&s59oo}+S=(1JL!IopH
z3?C(@Tu6IU!h|!UxlGKV`r3F2ROG%L!}RS9GBP9sW^tq3PW2lC&7hr1A>fn^JnUWr
z25nH!)CK-0B`_R_U@YR2lC4v=rhrXZ9x2HihvURKk%vv=@+eGjJLuOt@)Yy_Oh&Ly
zxSS7G9gJr!B)H3>z66XN(t_h7HZ?hLLH=z}U<wUvu=7OdR7XZdr6ZWw1g0$!jN<$_
zK$Rc|#qnNfYkp%=&irl3!_3P19IzpMP*VQ}vwRk?ZYIUy53WUFS3-69ZE*0%hznJ3
z-y7u%=KXrb8o2C(-xQnr=g>HaUXLZ$+->g;51apOUGz+cM&UKlx9Y^}v-D{TQm+dm
zO4!5&=5q%_%T|#7_0aX@v-%uP_sI1*%Y{??n6uS|7U|~qGT|}*JbjLl#-yVVs{OY0
z!$8`i?W)dmccU-MIy5;xGqehiiYK?rc2LXi%^56Es2Z+REUTIRYwJozrZ>~8h=_dM
zfjAcRN>6Qhoi~o_qw2u77$fXi4tz(qpzJe}*&fX$iG=SC8HQ=+)0^AnRv(+AeAVyt
z)i7#-ujE`Wf!^&IByCEzn<5+Td8N}h(AJU(p^WnEErSKs2D@WAEeQ(?TQ6InIfJVG
zG2mB<f!|RZIoqSBV5X-}vu)e}lY|m|R{cn2Wn~}CMT7te!f#hppY_9vuP@@;m!~Uo
z2KU$Hft*<3qgi6;e&&96&J(qDB^-w^azlqrsx^R1^%VGr{-D<OvU|ewBAOGmVBycK
z73rxJw^PZj(grD>9SnZD)E{w4QktEorN+xtoAI}zMU8f9-nwhvP-OK&p}}@y%>M76
zr?v+%<~>pOd$8NnF0yVTV_MJ3nMK_)eRi!Vm*z~r{m3K89{$#hsjGULU^ur&gj!y-
zmV5MDsdpm39jw<J{S%P*V7_sBY3Uo(hNKk=&derU7pSFr!00r;ZZ*9lSG2_K55V0y
z!6&5TS<SXBdF1H0tM006CIoCIjXD$Xx>7yRm5>%3I6XQ-Vkc0mWfD-)0n#phsnfI$
zDR4vuY$t2$TTOV(g`;4lN;kKyO2+1w?a|17ju!M5uZ6u*cxAKHFn@%emsob_$4aAc
zLD7>+q7YF-kTaG;n;6IPdd#lQ^~ZU+#LjpOqpSPV3$r^R=I=#J>}10E1Eh&Hp9*C%
zJz$fXKnRbXsy8ng*QES?(WsrB&Iu9vhXVae1@=YTO`Cf)hwg>fEfZ9T=1F*+RzLGt
z4kZD*8Y2@^ri|f3sAmtYfEo>pRQYxbbEoTi@t}_8S3df0kbgDm&{9w^Nwu`Kv23XU
z0Y-RgtabUfzE_qbKc348?L}x5o2PkuBWUU@B<d`m>LYsuoL2AceZCRUia8F9pud}%
zB*VE4eqM(7G%Kgn*0TjwIMV%1(8Ey%sz7aR0yNB_D$^nrr0fySZp64XQ~wV0pEpzE
zJ+aSIV}yr~1|@P%thZbV<-+8;SGyAzuuD20MwACeTGH4MF8#T;^!5ckb?6{s!;WYc
z{yju!q|nMaIw?wx8=C9m{ZUDkmdJn4hF0VV`%@)*b{4l;&l7M&ePGLu+p{9fLFvz9
z8Km6iqfJ@_e>55}?jQgB`9$jCpWMbPgzP6RrbT6kjhc-)4At*_t<g=alT@yuz&tS0
z`4zgYs<Av)3Lp+Xg-BRq#n!}^U5V>1t;EiQNQ)vC+S{1Wg>L5Cw{LqO4t5M5Ss!iB
z9j{kx>~>4i-q1OKN_Lc+B2UEF@nrKlTXMba?_?I>i5VwxjnO`5^})3rqCU15%|1om
zWwZCQ%&k0npT4UV!3`~Exz9PDF!`IORZq!+@1A}cMOd=9lLGs;Mny{B8i(&XMsU0M
z{pe6n`8GKIjR(JVUsaJztnOFhsb%^1KbAl07ktyB(phvGFBI^=UJlI1b6>xdbrqB3
zIm=`a3XF7(D~%71G-+2_$$c2X=<WEa*{CN&K4teOv79M{OA3C+3@QT2skh^W*;+yl
ztJBy2K`NS?8!d&sG^B>}b?}rkl|CqyyE&~5r$K=iWdakP2l2r0nhg-O==PV<G7JzU
z6Sz)!P>Dh0w!7)|?>bMPKJ9`8rW3~n!i9%Q>B&-hBE%4#Ie8p!oOSS-@uRLv=Ckat
z7^mJ2Bf$jk%L5r_lwzFPLRk0TBcm)?Jm;l?$b)t2tH!ZSoh)N^(RY?U>LTnTWd9xs
z3zFXF=#TaJ%je|iX-8D%?S6DnvP&%w8G2_)GZ*`kd<Ck@A9u`7_Sg79Ag$bKEfY+n
z+@qR=RpV(m;KiIe454Yeu<hFV1qAcLQ=tk__HzM%W>3Spy^vEE)MkT+{Ft6@qCcsf
z@B8O>3;kqdDxFv$cm%QXmE4LB*S1<8%wp#a9ihWXzd7Yg-Z<8Sx|G}OyV;*=q<>_)
zskGAC)O|Iu8Lu&@*e-NPBcM~rCuT0LQ2ij&6Q^u3YpYedz>a#Gq*GTnb?(XW^f}w1
z;^mX3StLK|d}jYOPuoB@D{`;$rS_|e{SNz#LL>8m>~8@6S%#Gzu5vFCzBqM|>QR&v
zQfAwEwK1+8zFUV)drzopF@XzSycfj?$L0_<@L*8uw$w2|2#bNi@mjp1Z6bL6w^e6j
z&J(Ym+gU4FqvzcbmylQnP?mzMZ0P<5GLZ-14Rhds8!j+>4TH?6Sl^$gTO;@k?L{3V
zaB)a-<j3|ExkTEMv#z5Bv!z)<IvoXSkw$ukWh48y4u*e|H?8?&&DU8fYZClFtf~jv
z$mk;t<Yjw4JlEjlLq1G<hSe~$2fJ$)!~It}oEr%Hp2zXZ_lQh=%|gVVJjsBLYN19B
zDgZ$;a`$xuUBy`L(2&|IAT-dLXAK4{VimZz@KpFJG|klmV&N?CPz$=e$g4hHR%fN+
zcU-*5cyWI$U*B)lsp8HOH00g(|Cj_@nwmTzO`FPSG?0gWqcT8|g6~tKVM@vIQirS9
z6;6cIs3Y4Cn9q#1k}cd$msYaMlk#LA1UueTw0rX=lg4c;b{!M7Yj3yrw|X6BOjefM
z*#`$)XCforMB~?E?gZOxmdV-kI^NvJW{u}mE8|<D;5}4-RcSui$b3wdTXXIv)pIw|
z<CawU$$AB3-<sAto<KHgK2}~lkf)n>37;Ykd8O&-_#Hk=MWsg#DO!VXKm7)#(LAx&
z1ZI#bp<o^bO?c#-shjoyw&~B&e8{R^_yoW-;-T?ynG3{ccAuDkTbrhfQWG-o{vsYD
z<o*LpKNV;^E9_m+B46l#bT%2G=DqWWn$z|A#6q$yO@%}6({Rh9VikW{#1i#_-iVgU
zU5&hR8=p*nNz;i;n+Fb#m{zBIuhM?539gimu5weWJ$zOZRw0dh@qcK03#h8TZ*3R@
z5D@r@0#c$<($WnI(%lUrCEXnxML?vayE~6acXvs5_o2JtUEu$J?|aAizHfZ@9rry0
z9LK{xYwZ<t&SySz?lrkYz&I<<zTn!9{u!NCI2#9tp!TBDgHxT=3JwsFkc=lw%|Joa
zIFPcg&l)?NB8XPA*Gez})&^{8%JoFa-l>kJCIRqlZuUrlCX@@NMjFK28Do2LG^<Kq
zxbD8sS1l3(Aw+ua89l~zSl#si$QZait{$YDto`ny8w4Uz0*YPUdM)PFYz6Zn*wqBA
zrhkBm&H&CiPDMWO4sMk-YNWt}`sn*Ph#Sp;By!a(byx3I)0ag#>=B2k+y%|vl?+0K
zP(UB$1tH=>en=X=PI7M$;_Y-;W}gR4ex4dNIcm8`Ej~SgVP7nw{kYmgbUH)oV9WB}
z|2-H6wg86kHENPt34zgWyOr_mmJpk=dnrI|VW(ptX*TYl6r6n>`S~%R26@hwLY?<J
zDEy|Bes!pUtR8Xh6R(w;b9inaR8iJ*wk`byh`KR=!MNN~A>2GXEFfTQgS0F(tJ}7|
zYzPvO<v|8o;e0{Z?DsQppxUT3-__El0s>XC8ZCT9TV+R&Z*p*T)L6Cb)CsjfI}tv`
zjiR(}83F(1lWY7l^vd;4Mwefc=2gCMSr)J1Gb=K$v?zwNno>fcJj&S=_LS5tQVlxu
z@PQ2_p0iembIaBdRNQSPQV>8N0oktlg(^RaE#9@v4WZ6}!y-V&V$$v`VEvqdJsd8#
z(!CuL$P89cQ;Q*8A5pg$1{s#m{`qH4qXnFfx|WcCbdsi>?zA_M6f?k~Mx*g(>Jkaw
zuUtZPyaH7crReXS1lws^yNjrgXjLEVuT~`M`al!Bw=01_l{Rq{yI!DUu?BwI=Pln*
zGfB?yx--Voca|!6KOrWD0lE1MJeS+e17<EfK0Y7h53dJuZT}<%P*=&IXuKypg<UFt
zE*=O5BO@c}@=m$K<`_fG(TIBG4pe)=Y`>{O#N(_0QZU0vRr5#WXlaoxjk1PQ0$+c^
z7eOvU6_}8l{7IEBs<f-G_{`02;}a%>I`{DQF?Bk(&!H3Eg(t<T*kDjslimP{o5h=d
znXR(7(GvRyn6Ia3y{FAbfxgWFLVgSA$wNhJYir?ynL!|5%NClp8(jz4;WiG@?8uq-
zXH<&}u(c9^+xXS9ZzlZh+qb)UTF*ghu&n`z4+nRo<tElr-t0F&7^v+Yh)?_&KGVIk
zVKcmVrC%C%fo!*Wvq3WZ@#gYnMX`=`f`GGGSy*r;^7*0CV*B@jc+A@geTS$BYHfj9
z;j%zMC^tdg=*#o-$-v+GT<>x39K?;8L8hwGj?zad#A*<9pst<*Yxyb22t;}3mjQ9f
z$wcHKmSbP#4Zl{=rfj3*Dbk$KXymOxW=y9j74xkyX4yq8S#vpa!aPAg9<4ti4(Vtj
zttpc}U;d^L!V5QY8bISplgu_&txIU-x%>!q!0%LS{Y!A#diE{h0aCj|dJN?G`M<iK
z&LbafPsieTVaLhe+X2a-sPq?SXV@<OWo2dD-20gvPWx$0gZHwUD=})`Z|%Ter7wx4
zmWmT^MoKL`gDk}gn3sP7(Df`bwRN<4&0BdVLTvXz=KDwk_{oaf8BF}?u{hLc7HOY>
zOMJL8mOd*bhG`b^Nq#8JKF!_vOz`aDl-<GbPIeG9nzJ)VFxvQh+s{H=i^gfHrnkF+
z0CI%-h`MQ23Jj-`Fz@8!b)Fr4JKR8_?9|R#rSNhTGh$tri74jLv}V^Ub<EjlCm22o
z-Z)TRj*#iP7yk@PRgw=1IG5<yz)G{D1Rh1UQPMLwnd;Wr{C>vJ<xk~-{OPLox*(pM
zB%4bHwA<A~s3%&ZLQmp1vQcMa*`5}nCcnwfuPcz<Pmmu-@GlD7gW;!<*Gyy`G`T&~
zpw5#x7&6CA*Rj!7JZ%m*|FrGevOOcO*YYbdD9G8tJXN-A3QEHWNbPb>ylWhx6YU<o
z2WlHV^SfKdD?u6^oHk1l=DHNd_;hgT=_YJ0Mki<Fmr}OUW`f^%mweOtOp=b9#h*fX
z-wEeAkH8j+_R&f8;Nd3hr^D}fas1G>!C{o(_3q44wf+JmCEy-UT$O>tlCg*=M0aX0
z3UbAu`yXw|Lvx05lc|o3!!8s`a=v97%XsWtACNAeU>f?!+%+K6UqLhH1`?+=tqJel
z0V(itSa(I_03hS;UR2DPO1rIXZB;oK|N7(|B?tB{5;F3CeDW@2s4l;X8o}Qa2mN2y
ze3XG6R{zKQ|Lv{R(VUl6O*t&1N<8Z=p*7Gt;y4x-PBA^(fHk~(y<t$={a5E6@1OiX
zzU}wKMWB1_)0BJ~_6)!!z#HWXg}?Rk0Y{<fSvwfrH3QXqJaW7@_j=5J?dSDLmqkp?
zr}V&Pia0y<qdzj!e+oTc)Yb+~F#B7}d)-}M{~cVOM3}Cxzus}<meY3=M^V)j4CxbT
zQ%-T+SbtTdz#v^On!On^th>opFWAgM3BphePkMKP{ddfdJw0#{Iy)LdDTYt@=uI}D
zFP?H9nYQAz*~%=G-5Z+^<gL@aGSe)AjL#S9N(@HLd)$}FSh9)M6B_RX!9(X)2R=fW
zXPEzE&66p=vw8E}mqx}W>!oUfO`4*p$06sVKY1-^HJ{{MwJi><R@|(8klSyeC%n7)
z=cr)u!aRLu6O-b7f<(6X(+VNUv^gx-kBxmaHP2{|vT{iFP%3RUC#|0q+ueNy;SCb_
zPWjjmvq=`44!P_cL1#L#edh2>z0foYxOi}_TCr8AyJkto6#U`c7g~ku(GciZ&0F&>
z(Hw-APG)Fnfxr3X^15oLJ8HeL^X{T|`U1CCg@x-v!&7#JXRplqa`cI_<fERgq+_d>
zQS*|u*dgAnu7{Wxm@tjci?wc@orNnFc?MPsWFw^6X5Tk4e3Z;)*MFb?ZE;!7yAMMk
zBu^<9faqyWp=hV2mCCAbs7=Y2l6j_1PS#^v<^@RGdSzD!M}0-(dh|^P^Si#G&>61r
zIlHzDx`pUK@TueSK95WPxU_|3n=WKjlvvoGJZ+#{;K-}diRw<bw6X%~NFFK+r`xQM
zXI;j5kdB+NF8}5*Pmi$qa&lu>i=&dS1iZw3y6k39eqyW|FKg+MV;3m=Aw51mzD<|f
zUA@0QrOEH5TQsRpGS_0n8Y>CupM}$;&e=6Ad}X`NR7S`&)TfNxQ7xM<6uwv0^opww
zS){Wom)&Aw-W_yEh%qCsao?JL_xh$AVc58&Q!j@ddIZq;!|Hcr!u9Gx02831q1ia#
z=^Gpz**TzKdkbLb>l<`@Dr_G#wRhmKBrz#abgDSB7^tSjiSF8Ug}&##NPy<U(3B2H
z)yPCka+M>AXNzz(fpbFIF*!Ne6J8&0%BntZ(=a$l;}_Y|gWV9nZ_27_W4*Mo(S!<a
zN6Fq^N2Rgvvz~!|2QaT%3Dq?P-QBp3Hy{&P^^|hf!@~phh%4>RWUT!`Q7~}zdEkL2
zC@i$3Jr+R@aa;)Ro2`eWwIb^N73WK!0};w)J`Cp=6Aixu#Gie7q+8M=$3U;Gpi*uQ
z&Jnyahx=~LWjg-Zz3<_NzHMiB?L{9H=Jt~wQ#@pQ%g)XVK6?iw5vXwPYav(``V>?t
z+2#0%n}nomONF_6v}C5$VLiinmemGX;Ln}kzZ0!@>F;kh<;vD_I(>pWBQ#j+i=5`P
zN5<#mlp?Ws46273b%q8}kImNom%YB6Bz|CUzO?sb-%`Zc5XNi{3;W1(f<jp_koI68
zO4E92?5OuAvaLWH25zp#-F<(QN!X?X&??uZU>(Yi(b1{Y-BcHgA)u!f`8n}$^+vge
z>n2HF$w)V+m{qm2u&b|4T|Uv7pUrM-b7tYM7I+0l{*I$E!_41a>PlfpA~E~sdOvnR
znL<1_)Q@m=!>r%U&1L<J%Uu5G&p&JmsCWJPf4|i|swy9oT6rxKTc>OyBTS)*6uYXE
zwbN|*-&}+_s$>%R@X<lVeD$u;$%SpXznwLJ&gumhz`U4Dw64XJGHlcRom~Yi)89yU
zO(P&634+P{`}*&^h5kDi|6jNky1V873l0DOY!~`Jo3{V48@?O=|5$6cgZ}R|2K^uH
zn*UxO|6g6)j{kqIwf}x&(Eo|h{14tR82|rZYoM{^_3ir_lpGwYAQ68E#0HDqF6_z3
z$SB#_RcO_UWx%7C!yxeL5c%kKuOJ{m2z2sB_h*KN6sB$ZcmxYTG-VVVAXWUa;l3TH
z#SA!=h;dVUfxqAj0bx!poP3TYS^^Fth|Y&=$-t+3LrVGqJR|`KEVU~8e2`L^3Al?5
z5Iq|OVXO6R>pL9=fV)w62J-FmL<vqSc(+a8a;@X05TIy7fNi?vSaG!+--j@RgVq7N
z#iso4h3U;@zPIC5Y(A|PvSi5~77?+sV*K~s5$c^HCF)JeCei3!TwJOOvHtb3Wq{bA
z=LtV}^5d20kMG}q!eBd2K_zPWV$#y|$E-e0vZpsp0=lVI?ep7M5Y^+Vt3()OP*-}7
zRcqwzt6FI~J1^J?Hzu8tq5BaLMgnGJ<=0Wfs_9oj;s((V>AsC3ii9}Q%dTs#UDYl6
zvpvP8rDrc<MF<}xd$OyRuLQzJQnzt~?`SXTz%$T1M?><fJme%K0=~Y!blH7G)l5aV
z5G2&3glZJNO$Euws@PahY2K&$PNPL{6+=ret?J~oP4@u%os@lPmH0cv#y%7gIdk&@
z5Z#^DMCz24qMW%&G--&cT3TBAh5JONL-D)>*}h8iHbQ&rQ7;Ct<$xz#w&_s*H60o$
za<Dl&cEVdy>XWYSFnB!omc}LHqz8Kf0;K^sR<}TW1Q1XFG+8TV<HjQ=Qgi!Fn@4su
z36MMBenN0!ja4$A%_dVXT~leKOgGS{<R<-qZagBhQ);GAXMpYysbXO7lln_qz=@l9
zfU$@fx<YE%9?vaLBOTweAAW?U*yVgQlxybC|KPp`H%Xih-<8Jh+r3W9Y-;H8dwNIn
z!93RL&!u0JXFRBeSH1NI_o^eynKi%!_ZU!L#y6O<Drb!%VPaxNpAsQu4N;lZZ&<P)
zf*7n3U`XL0Yz4CW6&G-eqoSU>0h(!*une%j`$tELmX^P$#cQ)nd3pa8rO4~10YybR
z?ywlPQt5CGbZfU~5?BTHC@EWBFu?~w!E<&s*sMd4bFsNyFtu$0%V<vZ)Mb#idUds)
zegDDy=SIq!;~s}|*KH7B$G+)*Q{>}gzBMDMyIs-s=arLwh)ty=GJ%cO#QzIy+ghZ3
zD%(p&OU+gFD>EI!tW(zBXDs@?w7yPB-zRo)j34qZ@Rr@(Eg>3`n5gY_PBieTP$8=x
z&tpTeGw~gw0ly*jzmo~(w-;LcZiD&(FL+{bnd(uh0ttuf8DPe3^CXO-a9IjDH~b+*
zh0ou=(r9r8N2ecIQ5+2x)Tq_yW)?WDI6>2+7x-?>u36RS)dSPm?kbfMCpAo3c;n-!
z#VWS)gGje61i$~nD9Ra;klf-ExE48mak%wWFs4E;n>_G&#)Y%?;`YE_pmI;4(&MDt
zl8u~PsE_Zax=U&PUS!IV%P~AAGg01J$h&gX_fL6lWwuiZm0(~MC8rq77TQPy$6s!&
zBSI$d#Dn6KQ>&}vbL#y<jTq_~@}7j$Cr@uEMDb?|N`xGYVQ*(sa$UDgM^v(T^6~4a
zlq_Hxxjr)&?}V|i4imX7>c+_KxD*7>Y=tg0A`G7yw-ZfTt!foedT0O~5iC-U#y<Z+
z?l)5PVNM7WBXhN;$ra=*e)Fkm`u+>@P;;{Zk*=KzMcSrZ<Bkkk=Oxb|RK7+>%IGRq
zB@zlZ8jB*;K_rb#hXS~oX}S8#_D}udnq0=cv{%nQ7p^VS*Yy_*iqm~+xs$s3?^K{J
z<UiYVfU5>-_fHgq?jwhL6g)?JdPJfkfAS^q*SB&jJ2e_1-jPT|q_xcVY&k+6b+s{d
z@9JQx>b>DgG|?O!&bu`y`VQ=evnDXPI&X9_SMwmw;PZvn8Aeh3^Lb%0332K={=ZW{
zuNZ@a4(cEbhmf7+PJ@#IQb5h-_=B-qMhTn_jGN0`|Fd+HFKGvdP7YgJTXJnWbx6Hq
z2o-wt=R#RKd(!<}vQPSq-)*9-O3&W>YKOiKK9>FA$V^4sf!Ra@egHF(QT9}49UPKW
zKk7Y<#=od{Tb3ca8*A2MsvguCCJsUj;CtCzJ5ai75WdNW-Kf10b*>ih|JpewP^O)V
zWQ?b!7<)qeC&TSRQ&r3Bc)7ZcRP+W~w|0tdbF8&NjwW^Efs!|?4gL<uCX1Q+wlXg>
zI04il@}2^IUA`KLDE@OI#T5pL5Tp+v(Wv8Lk19~>_(Kk|cG#o9r7t<Ef6ww5P1es?
zv5Q;k!$%L!p`2vJwuNG=9p07?;nh8Vw@10HLML`LLY%Gxdoi@JChK)}p5-p`OI}w~
zE7zOU=p<OT;7xwxq}x&!X|`JrzEqIn?${P55BcAftpjAoo7a@Inm{WWpI&hAC>;&z
zHj|NY@I*X`6A_Cb#pSWT4oO?|^_BX|rXx^M04GTB)X7CWgwgsS?ge#I47KR~$)3G4
z4GFY+K<T(?9sh!wWRyx2b@#~5ndYt^8IIe28b>En{&*k|N%zSpCD^>$sa;YK1&`4x
zYRIa2IgGDgRM2y@tQQ{R^4jbCm)=_76DbvWtg$%6oC`zj&^FggQTWL33bdS@Q#F<>
z7=kwOmD_2Qe6Ay8hT*fa%|{B97$=L_LL=(k?7!3JuSZK#xhhzDBYLjAxxyIqBr+j6
z(S9quEHcu$#B4EVfQs3Ip8m*aR%_hJ>D>C@m@$aSq|H3Ovr}#qrw1GVpSl0qRGIxw
z#>D*aVO#Bkn5vtP(XEa;9jN-t%g}l-0Ze&!9lePTQF@kBB;>Gle^{)&o9gH+TPauT
z+$EiKeSFf~-pK53R+6k=%^}Lr5c>&Q@H*C!uFq&+4{bLu$;IEyj!B+7ciAhVVT=@&
zP0PI?Q&M4&MYru^;lWOfPFIJ{CC8mgu@YlV$~LY!Y)bez1V`;Y!*Zd?<}ls#zbm^2
zsHM0;T4np;@Auro!{Sq^>!OprkTJD)<G8Cd>1kDy+g9l#A~RK)v|Y6nJw568OR&4X
z2@}iaA{~`%wlioROw$lair39?phT-yTDA7Y6dBf~7!MDvO7mGgp+datMM$U~ZLWI8
zeeE9lhKnZ_RP=k;a6-CrD|5E^a94JDe9Mq1GA=aLqI6rPJIvy6q_YI|@Mt(MQcGra
zMRHo9!UeFhYonvuN<OXqYBrbOVxaTz>mt1hdoprt4tvP@$Bas?kR;!goFB#u*`i9j
z!;7)eXE@K-ygUEORj^}a6#@cwZ02|#ruS7MG{I(`)*qS)&(w6ewF{`nrCH{dzwX!2
z{^;u-nn??5{8MJ)w-UqrDJo_t(AMT?TakoW0c{2^7nkowjQHx(Z$`C3Y9~)4Q|;cp
zACFLfiVhSWJ|>cq3?+@`;bvYs=XKslzQA7F*!bUF=QqJs&RXgB;$eTFRh!j~)n)3#
zlNH(tN5<OSW2*OGDhmW$rmYk)HK#L5B*;Sva`V|;X>e+ImcX1rspL1uJ_azA^n7Ft
zxXL*nEdjo!eYE-7*n3QkQT6^OPt5ah?q>Qq`aOCNtb>!i{$I6hpH*fF^ol026GKvl
z{3R4I(Pe)o5+`pvK|?5*qq&|t{oq$t%Df<ug1pn*A<IA|%C&}eo9-@TfbjeoXo-=*
z#W73hbC&Tp@5-uYO?$`ta?|i(0Lkt;%X)Z6V_LEks?{07+*J$3z|p`Xd(C$M=rNl7
z2GWropdc!%MRlEip;mq$>XvZZZR(4Gj!xu=c<jWY$2s<yW3iWZGLHlBjZQsk;r%)3
zRb6>MvE<~AHgbkzh^yWsOD8B;t5r6By{}m2Y<FZlD&)plEdTkPM^)Dy_s9O_mR}hE
zO*?tmy+o;2;U*kB$+|TO;~Om)R39|RQuI8>ce(ipzSxScsKh-|&CSCNT?@EXeGE3u
zWZR|w85*1GziHZ|RVY`aoTs2yVtl@t9bB1XN6n_FUPzjrxzV$ev5}exfr&B?o{Tgf
z&jDKgB}7cj<1P>JI2}luZr-uOcOK1h^StF+iACLLm$o~JsIdl_gafmm5-?ynM9<!Y
zrlqdqsadamhiWHF<q3|r2krXmvVtF6Z1f=Sgl3L!&T!2MAxp;%EjI~My&_W#8Oqfn
znjT^o+cFuEL5KMCmsX4+>zwN;=$*FNOmfrS9{g$(&3Pf?-tBgq)~InYo2rt2&{~}%
zK(V;IxSriW8w>Z&IyY8hk$A;BBL*rzMx4sW)VM&22Bx#=qqS#ROgQwzLG(Am0k7$7
z9L{bR)kE&}9HNKy{#aTNi_tORnUfQ|CP5j#jU%adt&HYv3GuZkz%rg}=`!22nWHCG
z$?V+hhJp<Q-!K%`3?UdLl9o^?LcWn6SIWir@`4X;6pSpF7{}R3M8Seyd)^<;Ki0>G
zrn5a~v?}hPjDz8IQre-d+y3bzHpLyS!m5+3Hu*j(5-#;C0+v56$`11RP|3$f2Ya#r
z&YiOvC*f9U6wKS;8qZt{z%R!_&o8d5j0y4PGvIfku0KBya&?m!&kxJ^n8E!k?i~%M
za9NDZMM(O6t1@I1t*o#TfM{zja0DMBeV#}mb3VAh!-dVEHUgr)=~r-PsoF3{N<_qz
zn@zCg$?Sa^E$4J~3b#pEP&avK#x;Yoy!*|;YefyKI0>l{fH*0^jY>P1X6~G8E0a!L
zl~p#pKR1%mM)x&J2BSiET*pq=)6+=%ad&eC?*J=67x5Dc$;sq9lQ0Ms9Xvmg?b^*D
z<7J~_{s7{$?`snhoR2mv>_ClO>%8Ixr}ZuiTW$tfJ5Nt$-TEWcy<@t)iU+Bwha-xb
z%RdjtAU@fdy7OZtd)?YgA%F;N#Y2UhE7U2%ZMVwC^#)FJVG29vcEuNK3^hz{1{EWB
z?vW)#dq+^77+fb6=g~~h=0sP1h^Wmp%H1_~&U5q6b8>RNirwi-Zv1hZ_IjB|OeVVM
zY%d=Llg@5?K9P|=G4x1TNh(lCIsp#HqRx9jA+-jg4+=la1Q%i3N(+`{9-wdG8t3*;
ztQ(%h?sS`Ow~0##S*Ji<)O!(PKt36?YOh6|D|M@Y1^q4HPh(I7K)6PQlT!f|jQ_&!
z2wz`4-&m>SM?p}Ias~spL`Gwq+xY@n0S)-8&bsyZrajMCPDRcsNEr5SPkJ}qPot))
zl)I9rwzcrrm@ofj@PR+T>A0WvDMwjePv&EE8_M3XQdv9H6&4blJj`WZ@vHHNN#~9Y
znW!7V2>lW25R;}nthAb^htr;FFT`kYcQwW?WS^MJ@X%!JUC3cuOo`O!oz^L5zZ(ih
z^YTfz8J<?pdLb$(P0aUX^-x7f=8r|WAf+=7Zrf~^xa_M}*6f564x8oFV!6Ah2@buy
z_Fhn5MY(wH3DfPQJjc6Zp81X4cLs9Y(|3q3EEKQy`HYDZU5yR}Bs<!Fk}A^b`h8)-
z=C-d2dSL_LlZc39cu=P5wU(Atfc$QGd<ybibnLlzW)&L394AF+D0Y>L^tJGy8S606
z!UxWFiZCI&$lj-kaVZu!G(!atiEvn0M8S7prFkoZk?gl<q%jw-rrQmQj5-q!4lm#=
zoP>!l`TPWBpOi3L&moj#DVHhedLca&2?;66Z<%=kaM2OP0(NBZtE#4C{OfA~b4$kT
zF*?RM&UBHG!g7#OQQ3($$yb-yd6XJ-l_wFYvjbi>&uSwxF^0t(hC)G^72S)^xXoWh
zw%go=?x-Hz*Q%c*#PBh=s*$+WC`wynlz_BYipsDvHfcX{yEUr69V4zLul*_xjxLAj
zZC^8p3K|A?zj*P(*Eoqj(Mb&GFzN$t+^*V1EBL__RJ=2`_xi+eymSl1K(&FLu~Y8L
zJ=|*-)@OyZ+1dT_B5t0X5M^Sa!HT#DV@k~;$y>~aVisGV!y0`@MRcQS{*zKuD%jww
zvtD^o!vNKBO{J2YQX=j&@SS)#nMuibF&l?!$TK3m`J8>K;`UwR+nr`lPM#C~G}*r4
z;8@P7_g3oP=>w^MJ_eAw-NlB5(NSQ02$Y6@@47NaD$N)fv2(B~O?Msd{9;$ujLT63
z_*{9I(533Ezq{I2DnYn$f_!#1oOuCdiz{IkZ3OHno(&-?pe(-3;KX;gE0|7+=JVoY
zmQ>m2j(gr@Uck1}gM=|0&fM(7FJ|yyu+<yI=bNREVL4qU!!iK~kpAlO`ny{r!s)2u
zt2%z`&;wJLj8l6NL1#heHe(e#nwx9f(4b~@SdLZnJ3J~rzsley8J%K*OceNRS;Fxm
z!x#YYkQ0fDkA^e}m#pi3eGy+JLtHSO1iR4671BvQ-<W%Kra}WBDi%HCeU9C1HeF>j
zQ>k)Qdg76=Jyy`YNCWX$ORSG*D57In!!hxNtQ94wX{UcH)@||3Db!zVBwU50B{1Y`
zD&pS?NK%|8{*LPxvCmrt)0KN(3_qkrZ&jxARwuQs-Sou%+vp#J1OyIp^mMApXUG`Y
z7omM?ewm3wN#>ej%zc4PKEpa=BKj*!Z>DCYZ`NHawvGCs4>K)!20Sql1#HTB{e}c(
z@Y~?&%GE`Q?IpXD!mm>FEpMI-beChX5F{4!tXqo%#w8e7#6ftjv-+Z@SkLR2EHkUC
zJ(Ky&H@LD)xz8rh?0EN&2c7a7NMalvM7th|<GBb2kk+0hcVpN*eyBz&?fMGaT13r_
zMv5My(JH9JBvw4JFjzaz2naXt8TRN^Ztt6%^g5|ss62q@9`0T|E#6<V%I9|0AfnGz
z&9G{P1%{Iz8EGH?9Q$kzA5ew49_90D)v1z9l+el3&opnEx)nNAaZ8XNZ4DI?=G<wi
zFrOdcl?ZQwoAyxPCktI*=K4MJV@=kf>9Q>48qOB)s*s@!p>p?%^u!r>zaUV*m8<zy
z07JDQu8Q{cj$o4{C>Tfua%w8cZ){fyV3wF`nU+&8!=&Q><odW*R3v__sVQ-8qnkcI
zkCmEMRwg*?+?e=f2m|UoLu)qIvc%L1SWR>DDVbd4*#dWT0+7qI)^IWdjjmiF_j0uj
zmSKbdNN!QOrC<%rNwG=3paX_n(5FUCQDCU6mC$=?+>*556SC12PBFiYa!J>{xGZe8
zF$lhA`uJ>@=YXugy)(b?cth%==X~1}xlR|D9bKj)-{u80)juMR<e3wpZw{zht>ZbH
zjl<;<cYYpDrRa-wbU^)XqsNtc&bYU(@n8A$Z<_z(wo{$?RI~$s>UTEzA^ep%igQy8
zDbEP6kcIZJA^?ul#}l!gos{k*zwWzfKTic{wgeTH!BKW7&%!s0H9S^io@OvRrL1Lu
zc~x#(;jUGi@?kZ+EM2`6zN&v#;%cR9F+H<+xrKuYJyfYaG`r!2Yt{t<oma`r3$Jb)
zMZr{d6!``I?_Fk+Uj~D#rSz;ygxpd<k(KsgP4hCp>gr-fvNcoisr^JryeaEPg*QM^
z`1YNHaFepg(<zoV>9sG~t70JumHI#LySjGfcB8HbAR|$?_en1*x552-CdKTweN09+
zpaaQC1zL6$`e`@AvuAgt_#51Z)gBF(754EsW>ab2I<L;>Vzyff9`PjVn5nOyXE}3*
z{7mvKf;d*KZuyK^_0rC+CV|<b9eH?=t2Uqzw6-TiLzoYneADxOyFPYaw&Nh>Aspm?
zfY0(~xB?)Z#MwXUI<>rFeVqeU5R*ImYGKi}R)%ioK3f(3WQ|SshHI(3i>4|UKP4XC
z`DRc>I`3>d@PB9tPa=ShkEwNwG>^J23d~4PJRt-|2Uco2Iy~DeXJDRCU^>hX!>GNu
zMP(p>n++2V3i0Vk^MP)}>4Bi-RaYM*gaQ%MJ-}Iy_7Y!SHiAa{aTDe9sdR9&Ki#Sr
z4{8u>L}D^JhJ~a>R@!nK=6*Y15WUqfQb8=j7(S_EiwRez#_r-@M|u?bNXsQkFw6$U
zSRP#3y{4@AK>Mn5&^JwwEG@0GZ4f*L+@CC=`|aEj*ZlfurFean+kAlDhE$Z^VM8S)
zyz1t72i6az11s@mrH)657l)pf*Y<-X-<h0rw|0uYa3Ez{#B6E9t92FU$_(>~96D{u
zQGvQ1<-nednXO6VnOIoW2}XIp9nr)_rm~ho5)9Uwua>!z=$`JN++w_%O*$?G!6peY
z%zXLM3#mu4_9(R17LQ^T$==97{%(?*F>E#mJb&YlVK}jQ_mD!OLsJn6SgGn4(g^?t
zI$)Os$vBb$mMyLEKqDd#{dTW;-9|H8sTpD`T-2(c(AGAX_dbBYxN!pA`Nk*l@amv1
z!imN(;CzWmR1W4`g1x<?ykcc;fuC=POQ#4tH&fQw?i<wcr-K8<rd_OxM+d_!&=EY$
zDdCtUH&;jH`>jcJYg+`AdH`dP{s&EU=D`{IOyl|}<zj<2I>{f?S)zpmWXs4~KnZy1
zz(~)WTsFcR988w#ueYfx8q0y(oPWON#>3^5103Y+Rb2YP1HZ@)Q%1_MHg$kVekbnv
z#H&|13=&{IR*K)X1W7S><A1{hAjdoV7rZH^xr#OLVah@HG?N{SD0!?r6u7e%!5*XL
zWz@7x^7nA)OY-M6)&^_rzmexnCrMr$bFl;#DVg_S=2eaQS{lc}uibBbtVt}al1~?b
zuc=>@ZSuW|oi3mzp$|PHjGYNrIGK1d5g(O@?sB4KeGdka0$;p;A9H5#oX6$pylk*o
zZ_ps!z~BufmoUf_oYB(yk)tSIcC=O?TZTb3XuRH;BipQhuG5Wn=Vt@D+GaF=bUu_K
zJSw7K_vb&u*|M57lcoKVnq0Z6)4y+C-6b{5xdV~@Ldaz&7GkUV(PK~-+uw?tA-mm7
z7?SRjr+)eFg_L&~;sI%TVqK82lekph&!x&Xb9Md{kl~_AI>(a`wKm|i-$rxJQW@6#
zf#pv(ZX(F>@O}o;ya35`2%mGA(!7iDC&m4mLMaEiF>M?!GZEj`ZO7m;<85zb-tt?W
z<@<CrEg!H`_nAAGdkmfo`uf!UIfGhHS520_I-$1OetHa&>MnJa2P!o7<|5uDOw(Nm
zR!WEWw4CGLxlD8~K*rTF$S3-pL9dm<L9J{u*&x6f97gIF8ENe8m7^O<4tHr;hC2vS
zsB()R=c_Q$fa=1jM!hg^+UKbsyNgy)!T~!|{CJ!QBU5P)r~4_sOeKV!&qt$~aRQx+
z5*%;JM+{Tl#>WaT$xBF%J|@mMpL%bIach0AtHY8Xv~BhZnqHPn0cpH|0}8PvW$DpB
z7|N47G_j>~vagysU2&0Aul{fm)JDlhFO4AzIQ9FkZo`E}x;zeT`S5}6v!3El4m<t+
zWRS^r<q@atLRttkFo=yc2b-9mugsA2N3z>UIqnaLOE6qMCc1UJN?a%HL9Q2|WZFqP
z1_$Nsa+OB`nmJ6YGN<1==}*@iJ|>DeI$`Yb(}V#}KOFcne>^-IanRWuHd01p(Wf-;
zSlfD(Y67Ir_aNTGR+m@+oV~2>I6~30+_$;V`5D!y!RU4Q(+Oe+{#ja$v~w1dwSWs`
z6PYqPTtYTZMt+dU7eUNXu*n_k%H1{N<IvyG6q3f;@{J2bJy8!2<c$q7Mq1)!Tqw@#
z=(u^LH;~9CpP8SPZdEP@l-ZK<Chv0kjSJ}1x{G!6%0>+5fOv@U*f-JhzE$n@#bBt3
zC8-M=m#|EJkbJ^{TMDptBdKQU;aj~TKg9d~9LxKHAkI+Q8j>w7d^CK#Nx%=2IJ{^6
zjSaDu&$ZL$C$Z=C0f|ScT9=p9`l|U}qljuW`v?09{00$n;8_L-v%HmSJ{Jd~nWZCE
z%gflbZMx!gem|Tye-^g4EAL%MK8NHDB=7`QTBFcR+dmR)Z)+9v`;AO%4z^zit7#<}
zO=JkN_O4bQV^#!lM`+gABwT2ef2%9QRmbUVn{26=vGEfVUCy6%H@t|RQ@=4F4W0j_
zBV}%cH_uS`t*v?gsI}+u24Q<-AFqIT(D^8@RCl-D)K^Rtzrc!;7*(p^IaA17NYv{q
ze??quM~CbA31{tiar9XbU+bhi#js1<O$s#%O7+>s&SsV6-bh{a^UnpcQbv@?vi;wL
z1P{(C(fzmj1LzHSu0%$ha2~$@=kDi}{vQ;r_n-AB$2lZa4xg;TX6t3Bmd7#7qNfIe
z=*$bAE1BO9*QQ(jsBs_q1_@fep$pC?hFZm+q@EfdGd*7$<L>Dmi~2(JU<}VS(2OlL
zdO)mhI5=2!{9Bsg!t$oVB9b`Im3q;|6ymjid4|K+GKJ)igC%bew21f$MoxE1O?9_n
zU9#E3?9s7%)|m)G7iWK_w2D$x&`FnYe^GNr5Yv5o5*pdJ6vV63TMjX}^0-9w*49NG
zF`Hl#&A-#QZzR}Kt`fCjCt*SYd+^c3=!-;G$?uY1Ny_R_ptn~M8)x8?+L_@`@C6Mq
z{Qiow;=TRmE>#CCrF1z4xToMI7ff^ov6AyZ!>9D|-d<yW_cMfaHObG;<mv^&dk5*3
zuRkq*zB)8zb$=W5MDr^I=Z3lA((R8?B!+5h@IOUE)50+Ad&5T@Gd?j@DOBimw(0+4
z`2sC@lf&W5(@2ujgDp0^;e7G<HV#M1Fa~0pl<JiG@(&&)Y>Wl&(Ymg$E!a{UC1|R3
z)<!wHqf~TPgfXw=tF(Oi`Oh(Q{$xPdv8g0Kvv;={S6f~~$vd+I!lt(~?i10rs5Z^B
z&`iyW<1U@`pZD=gHs!JNPEDS_Jh<k1yg4@iURwL_>Vki>GKEj*KfW(Vw)Xq+e0PN_
zuFl=_r7|vEUrd98=lQ^3)0UDoZvDjAcNALJaTX_<N%Fwf-sW$_rCW0q>9eD1sV6P{
zsR$~rj8TkU(C--=g$D<L2_uIr28a1Eg$@^+(V0I*NH3q@bo9hkI|n`3ZSs0M@Waol
zq$t%!e7a*z+hKbpKszE?F4w6*m#cu6E0nFuK1ER98kadYi6=iFoZsuhpXeTkSX22G
zzjnGhfX!VqqX;roCx|o6qf~x#wmZa86G)L+>`&jP1tEOAK5R4gcd_agLBNS)7w~#f
z@=4K{<si_gpkG{9Xc(VwY5HuF3woj(?Z4cFZ2hTKFXN)DGJFd0rzokdwbeo@?{o$0
z1&TM@K?;BJ$EmL^asK{!gVh*b-x!^*U+?>gJLcO6zfO#AFp1|{P>-HM>|xn2EHse%
zh!B|=-2ka;KA@zc)xAs!8gturjc?_E7LiC++}=3}B6?QGTALj|b)6_feqYe*koUcm
zjb*uzm6n6m4U1YBqhYAt1FX<t{Dpmd>lyuoJ9(5Su{*z&D&CGA=IhJo^=iWB9uBwD
zGTPqP>&;eaOKLqLp@s4IC9FIuc!z%Jv&yL_6j%H5(bvB}Nyx!-*x!Uwlyal@#+9ce
zq;$RqQ6Ogov~^A7jMxjfa&vIlFiNc$<V`_8we7b@@c4e|B0S)H_)mOnpWg+>s0$M;
zeU+||jT}Rof+B%f(FT2TBtde0u$C4z{D;w`y*w!_Ym}YhX{f5L+yUy1G^{#{2|Bhm
zC8WPOMKhiotmRSR{DryV?aev)yhE>glJ+~FFec($RnO#@)k<D3h^F-j-@kl;(@>eq
z26v@N$~NP2yU^P2>||hJTv#qlHSs4>S5@YC%@oUqnFxA9AWr0L>uA=~lcSWzz)NGY
z-%Et7NSPl4alznaO<2FqH=c&VdiuM=&=-aG^{QC2hK#*ynG(N-V8xqZF)~J<_Zz-Z
zq^c#J3X;A!t}l*=DI-T>#5vh+wP-@Rou%vb$Pbwra8r5mNT%JM6PX+Z?vV^Kb#nj4
z@hJ&I0+hEUb~YG%yid?b1f+zumgASHZA;JSD}%$jyZh?+^OcG|HA5eja@UgoolL;H
zHjb@JWbqIo3J-r*|3-@jZ;Zf8u44(4kBpL9+>qd38up)u^YgE<s=xjbu}4bO-8n;J
zE-EoN%1S=U3EI%r?eVrE`6)))tz8&1l@h~tKU9P?&+))cV{21|z$swd1w*bN&y5U*
zfBO3L{tiUM<yDycYGb9&&o?-@mvs`Cw!XFT>-TO`@BB%AHCe0pKtl@&!bR4t56N;2
z=@u43cEeZ3*NchnU%-LQ-Vn9c)>NUCu0r(KK+uVc?R)g@^A+{@LaS_}GOPvZ9nSik
zE@e2g&=-x~CE6->4b`Wk-DBTV27co&-f&t!uHc8he0V({DL!BAysHwgw4XC%j1&K8
zWpQv7jNo?ePYneE_Q>>?UC;ZR5hBouRbX=XjK1G?kWSL8T%L9=StndOXZN0-QL~&Z
z9}rD4P}kS>E+TZBPW~9>bVy4|`{o4Y4d8LPte-rK41cl&@9q=0(3J_2LnDeKQsuCI
z1R;pCS>&@AU1=Zm?qfd-kJvL9m69-#cYE+oQ)_K3LG|AC(leajmcubF%mV*8{aSY>
zj+D6Iiom1q{J42S#bH4qmQGS5YP=NNsinoW&aSF=jQrSs^VISlU1@1vt!sDz^^Hzj
z&+{;~POhvkMU+O5&kUJ}AwoV+-{dDy%?az5o-5<{&}d(VhFLey3SX2}`N8$#1$y`L
z=fFhso49JY-|O4vXfF@odK1>QzCP<+vQwLG<Nq`3b3rr3M;*O(!z<^Z%5#NSpbi@c
zCw}Bz-d7@KPh%0vS32m&?_n`Evhd?i|8$%nATjaV53F#$*^(GU1_^4B>;~115fS-(
zNU4w-_Y$O;tSM_f>pq(|L`e8NiOMDa)V<=FRM7Co81B=c9B&iAcSgILu<fr$AE-RX
zn-NQKmy50+UOCc^luefT(>LhpOKVb>;J_3nrplPahA=6Wn7moqGj5I>9@44p6ZFZ#
zxHDyD8qR}Jf58KBE}ELRwk}ds0H>ZsC|Pc=FIX1=?E9*b;misD!y|*NAY9b+kVtA+
zQ|cVh6_b4r2b&weo>V%F<Oa3WRO#w|>V`k?`7SPPv4n^EFHQFH(aDc7J;{IY-@6JS
zMVbrE1E~<<Bd)n{V##_y-30r!ykx}tM|KhWB~)LA*TCo6bSPc7r1ya|v~;s^W+==(
zVcStf_0ya15Dk@P-)225RMdN)w^v#xPG(X(Es8k%I!1fKPfK#w&fo+Z;u4>>lNY<W
z?0>jR@_aTh2?!5<f=8CFAtp{i%JUQ<`$HpIs?hG1U*GV=O^BpALe%^$s*C)fT1p91
zL<i(ptIIXRVI4ASzooO1bZ_`m0Xexf4`sBy6h_>|EVI^xL)2{3>v1b4Zz49`&pH<3
zqvQqUQL(Z)GXErMx6hw`K&1{%Xl2BsFp$X5$W9wg+!(ipGs7gGtDD8iRm5E`eA2^0
zNj%(Y;F|s#Rp>?!b`@Tc6&FzBhbHjADscWL7AYw%wiWlQ=q3K1oGn(2<4X?bH$A5L
zs!y!~K2^k>9zN!dkIxAxDf&&?LjC@pb3B^|f}!xald~l$Z}9BN=As}i3se@*>#3H%
zl6{<Cn4>@JAlZ$M9*zAcB1kG;e<Pcz*nFZ`tveAJBUvGPZDl3rq}4siDe$<pxezL|
z9;^^<hzN~|>FKlIg_YO}Nt#OfpJ`8y(C`p99uFEuGZm~blK#iSHZ}>6IwB@NAw(HA
zm0-?`j?|)-$u0g>)->4~(^Y7tx3aqU*({%FX1d`EsOQ_-J7+^PT^%0!B-pgtzDP|2
zdDeiN>(lca0`7kVctZGCA)J*DG&f#HZBn}<zh^~zI}I!(G)mYXMEldxs{cCQs!)3x
zGC_!0_qcx7VCE=cbW4c&U?rY}0wpM)m)=7ll7Hv<s?V#d=+RWBrV+DcXZ!8^n{9x@
z_at1Ph{r>Dwu%6V#zb4MbSnPn>U_(_P(gAH3>w1P%FIKqr%yK_2?=&yUB+0L@ZMs}
z+pnSWB_lWJ)^;lZE<5E(cBvEiqVJn`ju7;_Rv#0)ySVid?SWcuhf^97=#SSXhf-e{
z=2Xu2RI_6xwzoc|C?R*m^!cXP%Cd=%Zn4rLoY`x=CiiAXGFEQuz)m=bt&^RHJLaR2
z-Us{l_pH6{lkhx+5^l(S_;;%7j#wH;W7Ag!*q>oh0n_qtP;i5Fb5-YzIL|B6J(Q<m
zreipn<WBm|hKu0S!M<p3cMeMfT1-rJYNefyLKz*<lvw!K0%B6EsOdk*?d@-&aZf|-
z`-2?D3Tb-slXXEYS%!_k7mGsHj2wtw%y!nVp<iCaB%H{v@`Za8JT5;;YR=*uqjNd%
zmX3&!VRbGotMN&94w(|#U9eb}{Cat~zK)PrWgTBrxYxXl{n1?SJ1hX*=Ij04!iocY
z@P*4w(+&x-S7cnsd}ZVgt}n2NG|dQ6-X?p|KUF&o(Nq&t5yHt9Yyoy&gSLCm;a6J=
z)L1<c7pr`1MZ!@rVYr7Eopn#zB-eS^WhBEmt~d^kV*a`MqRKdK37>UOtYajs^YdEG
z$2Z{%y1J4oLgb@QLqBh94JTb+qlYjuuNFIzXp$86$NYQp8tZ&Sc2W6B+F?_!-4o5_
z4Q}H~$H^wCB{5^FY6Eow7f5|jGAk?2O7{znYT5G;YUU%?`~U*`Z6LBEtwKu?M$)@V
zWFaP%Lh@>5AH_B*%_J1<0&>o)HNpBJu#!)R<Bi3}49pgpQ=6ORY<o2}8Ye1}<sD_8
zxR&KCW*aS*=PVW*&#9~kc|R9PXPKE4OixD`W8T&BxZ!g^!HW+1@D*Z%l|O&_K-Dw!
zN^zC8E|7rgKXoW(0=Roj>=Lz};pPv4-rk{M(Sq()41ug8r{$H5?7v9pO3~Hh?P2T+
z1;ss7l=;W1&Q>V1nvdq?)X1aB$exR(m#alF<wdhe7n=n1Yf@`5{a|rpA0L~#@ZETy
zE?2l^`xb(Td4%K)$OY5&+11ns2bw%1+X7P3g|9jhqn_ixg30_;lvATsjRtDCq5R>(
zK59D+ZA)+SLriM%bEnEh6PMMqVCwLth;Wiz9mGP<66~Za{N>0lj%SheC7T)petrxr
zZ|jE}<F>a809F^G0_QFX4YTCH59}p3S#r~vN+5rEZACYBlZqX;X^ngBRA#5_c;SmG
zCaQvuV=lNjsAD$=7~^_+4qs5f@(8`e6c=mEbameI3z0_$SW19FA{1LXz?7@@f1n0>
zX_Aw2#-#a6kYb^j2V&SB9;G{QLH_9I1l(mV4gC{8vcf+7Si)9$zRuT{<03U#Lkoa6
z^cQ?{$9?#!?Dz(3%n>4<);5!yd9_b+0ajOB#j8ye27l5OyDr$PCe9ShA7MuXJ8jHL
zQKav7Z+kW`Ei3zBiuUO_*b>4B*~K7wOr-$AuQl1=SjL#HXjLE*+;N0Elb0L(1v0w}
z4t&2?=5(mSI*k*F(dxQkRrZU}N3#~TzCHuBI$k^YXv4%00Dc=88I25bMZ<`tMGWRU
zK7zi6#tQONAMBiLc7Luj5b@}Wjv2k6H6F|AnOr_L8AA0~n{vg8e{9fo^oB}-Dt}9Z
zrrU`&>p>A$aZ~d<Z;k9FHG7O!Z}0MnaYetK;oig&pTK_-`gfJ5ny4VQ{Eazvw?$rM
zF0M2zl)W>rrSM3$&%-9eZj;5{F1>9nUcc6!gFC|8+{!*-pATPjtY9~}nwo6!+Kp0a
z9At|A)2O^-1?Kg`@(M|?$gb5e@5uD;eYDff#3P#7qO#WT?ud+xZhYRbvP`MgnNGQ#
zvuU)M0=f_BX;=xtqL9N0c?nzA`(YmCqw0S<LN6a-@&Q-qMfu?rH#u~OlKHv-VQ%Mh
zA20#TWkn60V^Mh_a{L2KV|dtI^FQBW_<Y;D%z8rOWFDbK;6057T%woZBDa7F41Ju(
zN5x^8)40B|{ihDEmaT)uS3fTf>-3P*%ljCuWRN8zgO_I^isSVuSC<baz-=MjA#%>=
zWeuGYH{l@xTbBU@k6jOVesnSW!hqTJK69KmG4xCa5Tk`mBd1yj?+mf~bW<Gox#5em
zuBsbI%1vCzdrhT8rgeBd;D)QqcisXb<KCa4&XC|*-eSAoZ2;h0?pqfaEdS!-gaW}f
zkr@9S`+i1JmB;waq{V0BS$m`On(oG*Nm2-#<#<i2S1x;lEpKPIoKd~RyaEatnqr*#
z(|ytCa8|;oo;5GyTE1lc!X!#r?Y2Z(=cQ%PR31!<jU)Fak}1z7X=FG~8Uw%mVAfR{
zP#}9}h>q|IEXK0DxTn4yA|59%px6wS)~@mBsAA^h4=oJ&lGB~yQ<4EIaR}TjdcQ`Z
zUcHYW19)cAyqWn;LsgA+T=M%bYJ8Bv=|83MyiLck3^be5k~`>VCvB}H%2a!nwt4cB
zYnq0A3<n3Wc6z!76>?oiCi-(`ySAQz`+xFC`C@)HCzpSSFr8>5b~!0pfu`9^xC4=O
zL_ShrXgt=L?1@4mb1a)4;uaR5PQ?9Bz(0Rtsn_;RSjkk4^J=wL@{<A1b>Ok<cpdIl
z@d$-k6WfD}U+4RZGkoqbi4L0!&T?2>yc(x6)r@8q`W;D9MB@(gugxN1uXyVa9%P4A
zXmakUtp=)Xq}4*hEI&8KE2>=%Nvl#99j#Yu)%B4x4!7LuFq4I<dJ_yU8!=^b6mE%z
zuIn0OiI=oQSxZ5`M0nY)ywTXq7kI9qDUCBO15V?obqq$_eCD$q-nncQA>jg!LO5WB
zTqiP-L)F=Y$+HtyF3K%1kA>RyMuMLbewR$TfZORQLbV}$4g(MK7t6WW|1{OHfUW-4
z`HB^zjp7qc=$=+zK{v|dKj&%}RnuR;w|B2>kx;Ndixk;m+1M){AT<S+i80!r9OIbO
zQ2)Sb7ttMmpsrPSRKKIbPa#WTEGkx4Q$Y1wv?uZ}%OkZ~ZkOa%8^u^R2CHuwhM)94
z*PkX#LnW>L+aq4e(76}O6J@6|mw&dyef!QusHqGD9Z7|dko-Yvze)E`DWY2^-`E~w
z(n5%-6{cF#1{gk9_m%qy@*BT?T#jnGIIdqL%H$4n57(j-AFid?HB~8p<@_n9(46wr
zP}S}XHFcGKVO3dcVO>W6&eP_i<J$M4y2uqmtn(3kTP99H*i#JC8ztkd^Qa?%=^B`J
zw-;z`H_#yddkcJXNp_!{M&l=;qM_B_{MtflV88%yMO&tH!K;c~otNSJ3Ky(#(HveQ
z81{r1D2rY@?jQ1%X4_OpP4}J--jfVzZf%a@k#GI|OIbNS78C+6(;xgMO9R2D(OLcp
zkM$`EK$k`zt_|6D_mdPiTv!MKt1fY3(K6pg0oNe<3=pM8n|)ji1ERJ%!GO)d+8&&M
zvu)SXHqGK4SC}!&Jf14pBd;A-o@JC6$?7)B-G6Md(vlr5{~W$3XsX5QwU6o=TRy;-
zv?95}>-vct!DsQQ6*JQ^G^e}h1atY>8<P$;^kx*Zw?ieOEGcYGOo>%8K0a!qY-MRa
zLhv9M0k}<;%(W{aBrFwC5e;Rl0>KD%eQx=!<tG`E4UI!#)T$h=c&^Dai%VhQd*=;{
z2-VyMwgI`C@usFxO!kkL#mMNlx%_`8^u>wguPi?hrp%Vvz0881lyi)k1cZvES93VM
zMzFz*|D#4<!8)6O`HF_*Dbv%Zysp+6y<?t{{ElZAiEn{R$z%M84}BfRQlXyi>xEc{
zK~VqWiD$4VY4z@O5FrYB?g`<JSB6yJt;T=tglKb2oLoEm0}xQS$3z+Y=?f(&i!Um%
zp8Xr(LXY@+ZcXn;oO~^DJq>O*o@rGM<3|SGcIq<QMrJL6qPmu>Y`VkOo|}((9fAzi
z*s#UOu-=h}|90Pe<m^;FoE`trvC8D7O_eP_`m?aGk8eV!yRt_>2@YJDWSO_s(YST!
z1RhpItILLlO@X4Rg)9s!z>A$Vok#&b%FI3JkH+LysL+PQ@9Eyo_V0CV7$m=b_vV<<
zQj2#{EqRbXsc~u)`Kqy@WT-$D{0r<I1BcIk`0dm*jqlnjK5uVhVBMPf>`8kYO4?MZ
zMFfqmUY&uJ`1WwIpCXS;x0Mg7)8r<cTuAQWBs^N_`<`nDfJ}89rMmrW&J&&jpJPa|
zL~$m8(|2XteACZ&b@+j>?2t^+skvB53?R`y0=}YJGcNk%R3UzEhmRNAK^Pee4K@0g
z69tF<#KydF(KZ}?g@`NQ;pu$(D#IVygyix(e~LPduV9TcUtL8_<v7K+MUP}4&Z2Mo
z$)zuJ(AOmwYiO;Oo(z1&3mP2ifP!h+Qk+iOMHTEMnyD42v}T|%0W05PV|c?<$AoeN
z+P@d-5hlg0N&f!LU&4K&_9>IV1Es*{{yIc(y_w#gfv?0!Cl5Z({7|%$(wh>Ob~{2h
zo*Pa=aAwO_w?47=b9P7#@Z?Fi<{-{f8$5lQmwAOtZ^graLahstL|_tsD^7sjxJ;@P
z>3JB0Z!`{@$PWMfhlYxJafd|so(ik!%VQSP5T&PpH@rnxUlcmM2xO;3#0nVfkJ)A)
z?3$k>aZ)%~oB+bkaZ*KIMy_#`fOYga+r~zE*08a^9wVSA*URil>bx}ytVr1Y<XK-)
z@8j28PrVb9g&2_JD2+>Eq7xxS-B<v1q_iu(rWZbu^Zlk;g}D#p2;iX&?m(uF?89rD
z2NYPLMsm>-iP~K}hGyoyKrjnSetlH^{7{$GoPx9L9O-MBUl;wy%M7ARM~25HHcX{-
z=LlTl)6}5I4dp6yio>L<+nbTrZ*$}<MLOatc@IKevc9DsDC{={(v_H8h8`JU;3pOJ
zU{2cqj0x=;Tj`0~o)t69W0fv7X#!z_kl(*gjoNyh%j$3m^YX`jTk-;Qb?d38(N>AP
zc{4P=Vj%T`gw0eJXzydC2-5;n`u{*xmoDEQ!gk+Xx-Gr2_WC+bG%@T{0<BC7ouB>b
zXz``qL(wRAPYpIy;p6ZUoRSc01^jk}=HaBHD0%SPy3CC2<xV$bd>d!yGsi7Hgl?2g
zkN9j~EjJ0H={TI@Qb6w&ZcI7lGt63y2l8j9C1(0)`C54xl#T!`>I<lerV_{`z*oBO
zB?Hq_g8xO;S3p(OJ#Sw{#X=OM%K(uUkuDRwba#kIBi(f{K#`K}?w0OQ1f-Sj?(U9n
zuJ8YSf9qQd7I!V~y=R|&X3sp&Gqd;LYiq3BjP2zTLLQQAmSLY++iX*3uV3cL;O;s0
z6^|{AVgi4KLDV&?MtgHz%7B5tN=dzZ{p7ZuYzIa-eZp`ulctUhN}PFq!PrqixOcb2
z3!$D+a?R4yTd2)4TiwN<(n@4x>zzptRF2EO^x{(LA;x^trS^#g1o+|6jT^r&3T2Lp
zAk~FrrG5d;*Qq4i1ow%>vlzMYNQIo-SAy!9hK9M1hw)S#N-QBowo{!m?m=<bXSRF~
zQelZiA!alTpTv6;>@0d+2AB^6oLSfRO4-M|ibeG&<%@9!$lqp`R#}jM_6Q__ZEm#{
z<i;!2Zff_a`2)J*U&mW)a$S$88A6^l80;_72S?*Rx#Ra>H8<^Otz(ntP%)!Bv#*@W
z6>=y&1`-2uh3bW?u_MlpigsF^xVD#jfdGDEpeH=lK{mRwv5c!I>$;c7J3K`{w>0Sv
zcaydBRfV{Y_L@Bs@PG^e_vSF?Ug>=GlZbzb@%yz`pT&DVV#m`XXE<_oT61E0h0coU
z&o|T+Tm?xx!Lpgq`%(G9q50sou9{b-lE&7?_Z1@I1T;Vw_m7Sc1sQ?E;a)XaZC!*k
zO3bi^-Cu-S%y_(^A*FqN(?{K_BhS3EOgo(PTY&1kpii>1rGmTW`0pgCtQeNz3If*~
ztrgZiLs_l&t921#5nTM3=yz&yHNVE%a{hR7CfuiEfsDr&=IpkQf^1PQXxle**3QRU
zZtl-@aSQMyRc3ycyQ8JxDmZ#=;mTuLB3kBTop0M)B=x&X(o8PBaHbBv>cP0K!S}ip
zO1<^QS3S?wCR0SUj;zp*&Txp?70~$IAoT#*Bd@n3UdSr*$2^8TauzSjw9UQaq44WR
z$1=etBTb5A9C_eed(^v6uslHxHkBl9Uc_(E)=73wC<zBDbz`@ee&%QmA6eqB$^A&}
zL(`?!-UoC^aKyno`t3m1z;5f*&3GA^b=!iJpi1@18SC;QpKG9$O^ZA|%bI4eH<wF)
zq=D!!_XlnYx-^5!%D387IQddhuj$uYP_Gge#%jB9PJ@K=f+ha3%dZ>%rgyJXGX8bM
z8S9rIOJikbV=N^&Jfy~cOhe64U$A8}?tbrkcXyW7B}#lO5(@Q5uCP$CwL+t-I7?l`
zof(`#A}_+!e6l=PTbJLGbtxjfO-46^BICC=6I(KQC$=b4Jd&Ly;&v1}YU8zCPpkE|
zblw_2Clhvj+R0W*P9{t%ILa%KGhmew5y2q;f{Acz>Bk{|`P!tLl*^?{v`ODCmdfPQ
z_e#;%^$&T`4iwHrtwfbp=9wyTD#Bwjo3MCOc+NCTO+w6&lu{wMM7{^1v?(qaE7rnN
z{+-VN2%O6=R7^yg^@Rm@+l=nszyL&W&PveCtW<HvvrV#{Y;Ok|c~}o#Z~M{pahEEl
zDbhRvgYB`I-LJ<)KT&vc2Atj_YpXX>DLz*(Ep^3G(OpfHaS7kV)8yy%#888ngoc+*
zP4TG(SD#^#^ynfpRz}1AiVePC^JTrivkt2C=lHS14_6#)n0xW?slv~(=!+OKo{V1A
zh~bK^oMkZ^leq}e4M=ShuXNl(JlQ_*TE8)SB`od#FH$`>t|>un=XRJ~5&wQD{tjOz
z<Dz=+wh8l0xcobV(o%0$_R6H`sWWrxCKEQ<NKk_n8{Pgp8awUI&NF7LoGDt07u6Ua
zH9$5xb9=L+$sxCGW?OaKo)aQ&rkB_KuSTP5^lXCF76LfcLd-KWk3j^(%b>BF>|iVK
zxr;S(?!p7wZ!6QcA5A`ft%G-DGjDHi`*R2rq9K*ia(OMMlPf#OQoK2B)bfpvZatDX
zQW@G+u2?zH9RjHjQ!&zZvVeL{5k!8JXrN8qdERklg^28mW%ABqZI_EP<(AxkcKFfp
zIp#Loe%GMF<V3=rN^6VxRlNU|X)NRUWxcR4uC$U#NQL7ygDm}-2dob}jF?4&zIYw+
z`FA`t`&Rh9^?Td^!>%==C9~gzls--C<ug&V79(`XPG_KJy30<B?Iz=q5k=86qx*6p
z>}7G9n3H^K((NYtb`LF_S$PF}bux#QF8vwR&RkM5QiZmZnZxYtroeCk#zN|(R&Q^z
zw~-r7=3?F%qoeoSg=~L5dfu%j4)s10F2GUV5ZweX<{9R3DI2jVa^OF_7qt66+fwmX
zNPvox*YkLEk81aWw$m8i<Ejt;-R`-#H~^|8tDurF+Y}|f(SdYoi99ZQd4X5R4%N&{
zCp>$dY1;z7SY<lSLhCD#rgVRzBl7?lr|Z4#)PyvdjDCxaN2lc6^(USo_n8^Wx5LEf
zlaLgItbTXld%&{`TDoe*4rM0Sc2Ga%+fr_)`=oj64Ra)({oMX*te2v^|AteMO6g@>
zVQe9*`4+ZX_v>SWnG(nSs>-+hUeb$Ry1!}{Rb7?t)6mu-vp7qboE#V!f*+%4X2hN-
z(Bb(7WV&L!Qd`Ty{JW2MlKry%lCOTO4pBr3+mD9W$?Dkaxy*wm_e$iD?a@E=q@cfl
zJbEf;X#GK<R*^yy-_f!1up|V8mfVAy+cq)0npx+Br95YVItKOM{nTyeJDjz^HMy};
zA=I01sSGLNwp1lhs8{z|8tTLSszzEa>wCsTwPqo`v%D|=D`dD%DvNB>_`NryvN9be
zWjDJoaK=>;mpow}x4GYfKfEH(!dP7FqJ_1tYY6so6glg|uYNLeE&i4vW(PZpJqjko
z1%g7IRgMuo)}*^f+hkTQ>jcS8TXd=tQ9@~{v6TULu`2lB<MmCyzG{z$*iScT>hDi6
zg2Jbf;*znDK>67_2bm2kku{WN1^)IcQRZyKVGpUMhV*ydX2stNje7g9kA<@DQFnlL
zL%AY8*Vz>{V^ajobG^B3_*~WAWj2AsFH2{a*$r1`UcG%$D%0rke7Vqa(GZ#Do}bRC
zRD0~2gNwp`jZex2K~UUUO4w$3=>8-qWj7*1d?D>S!i=<a3sChYkl~FfJ>AIaZa$Do
z>n>hNF*wes)?f*GIX93_V?RBu_ENmpk$msesH;*OHd}V~L0lm@YU8_5&F0Jg>6)yR
zniv_j<AI~YZ_tGxwS=h%*=JCk-2K@J*%wfd)s3p(F&#bl+Q@Y?=Ce#zvJ~^IoAamn
zwFX3`s`|PJ5L5w<o~;eDgpf~U@CqQy=CGU8Sr^UP=0jg8>YP|$Ze~t*Ko77o9Qh`G
z_CH#F?uq8>^8hNnRz}=#8YI_L_0Ppb3t+S+)MPlPwuBTm52yJPo53$X{V$(C)SiTL
zntMB8>eyUCfj^4KSYIO4ctTc}(yd!SlU!Tk9F^9Pej-dnJF-If>lTY-%hl6UR;l#v
zbF?WN!~0T2!(KD%lGS4q-Vsh5i*!*8)D62!KJUcp0k$wF>HPjvhtApD5VoU6W${Ja
z>6Y=IupzOxhaxCey==8QfV0JA7ar!nWpH}?cIC4uNF2w7NQBc3r+3x9<TQH=@<q-i
zu07%r3rmPuiccoL)yL_$5R$R!xZd?QA;Ee34mVMd*Gpla*_XcChm7N8Fl&ytX1@z1
z?m`N`ktz|Ao-yKMgF`rOq0rBt$kj{<R2J6raa%oLIbCA|;!qkK^-PpsQSF2>xm1Y4
zPzeXLDK{iPiphYx^vH6W=hR5Cn2Ktph?mD_YXXn4wMBnE?)b=eWdxHwW4z)W3pjvd
zPlS&CHK@(WLPA&Pq;?N#b_e`^Ke&hN(Urb<vDc@xsj+(t0R1b=!E^`q>%q~DIk!VF
z>Wd|!-qxSdyRW@sq|&S+2dpI)#PMsNkdN!CzjT5>30)5h@-EE1kz+gmQ#(s_Ov`%K
zaO~szr+}Wc=SxN5^0&CSy-y2`p1G8&8Oz=Kk5uqU{qy`+`>|N`^sd$2yB;3x;qAf7
zOY1wnnp-<RGvgo(B`{?A-hj3`*;3Zbsag&D1%KRWWM<M)sYV1q_?ja(Z3xRhT5YD0
zyMNQECx@0+PfXWKGoRPL90kP&u=JETO}GYWz3T`1LTMt3zL|BCMjeYL$<*)6U8orp
zmWa=+rT<Y*LqwQ6wM2so=HnFAGd3uV{If}94v9aWGOimZEmq)>6_`q|gdfCNp2aHJ
zSP~Q3-@lGT1h@UR9igS+6`m#5<UhRnFf9-mF?UenDE&wsZMzKEzPv21)36!{WeWr6
zg~^QkFMIB-JpHUOX<Xv)VKdDjl$c~=rwgs3?RE7kabFr1=mECA(eWpW7cro9{WUS(
z*dS3%2CK=*WG8BGtxI{MxzO|=Sv2kG_rnH60&8tz&X2vGpZ(2)B0|DU?+T%p)4tT(
z?5BMsO*bz`Q?Z}5#K*5Z&^Nv<{l_c1`ZGWZObyZ2d_&mE=z`A_v%X02IWFFf&EF@H
z;oRqKMXzv>3x4m^L}V95MoRH%QEQ+#KG)UhOiAPQ%h%no>8&wrRCi5Q2(IYm|BU?7
z)9Yxi^#&EX!gA{z%2Xo!Hy1YZe~<T~15{d!zk9&(DA9R3V?S6(h}_=k4pyaA&i(VP
zM)Rgt^S`L>Ucd@J*S@F0Z&l>c{_9o!tR@uaD6?8T$hlyhDqGp{eI%OP!T2nE)@fsx
z$0mwDh0SH3Ea$bc{x1*Y=bITSUiF-FugzbahwOiRN&I_PU4boJ{T#LvZ_N<%@p^Ia
zAmd{NmsId&Z1W#Jd_CMZE@ssEK5&UV+2ts#jXE^Jo81?84!cHsaObO>lAQ0|_tkfD
z){TDYK&Z|bjm}`(NeR(>ewLBvMhi&KQMtCBprr0semY}kI)KPUMNBa{E)zh~$T5C$
zZ)<Dc!8<^cS#RR6h!f7k{iiW!29Gpf_*z5j_von~cVNFPlyRhQa=~++-KCs>+)B^P
z45Q0y2k%@=Dlz7S82;0!YU3xZCX=Imz%=y`^#Ac!I(1Of*_*LkDI=#+zvo6JLm!eK
ztB5F8EsbVKu*dRp@mg5azZqg;QbLx_jhYU}vAIa4T;u9jl8uvN?P~Ch3TPO6rEo~E
zCd?*lBTiX+Yc*N!zI;hve@9O6to;9C>I%(Ydf=aDVH49$-FHIF`dMMyN!>*yg0P}1
zjo)x=7|NjPnUzZ2<2xQ|AUF#(feNjQ>^{{A{M9&A7i(0jkA0phTZ+qH!N=!d4*m4}
zj#gu-Xhp^8MGJ~`c~GB7)_U)Ow|!-x+ccgd&`Cl?Le!o&><G%5?s4CEx5DkkJ3%k^
z?0=yKczDcsHZDg<M)#0ZA8jkHz3uo2p3UL3G}_aEq|hhsLU1&KJ03QSeP-`fy*7I=
zTk6AXJzL#tln|nHhd@J-SJp1&z@Ee-hD&~7pr7k5!dd0#c%gUXK1mBW?iV#4Va(N;
ztXkTIKqv3U=9JL^9JSUqVA5a1Vk8|*qT4tcLYzGO-kpKSLsF>4%blddGkNCnv_<mU
z;aWjOlUKKpAd9WVb|a?l^^HDhv8CaFpYNYKWcAlfw`8jPmAEYq_$;rH*FC<cU5mY{
zrTK~K(#j`O4%6GLY@X6C0u`NEI~@$gmbnk~YWMsg-A8y@8VprmNFW!XOPE)M6JJ)A
z62HSm_rl28L6XKdn9`1wV8X=dDGT}uk;)mm|I0sO)|)pP;57E@lk{%zf?L~GdWkW<
z8qO7?`BymZ)k8JlZN_^-Mq)Awghsej-`l!A+Lqh)7)9o`M*Qil^<G3b8VllQve_s}
z-4d+W;*&{(;B}+m)S1cXPPWXgoRe@D5KlZM&UK#bvSZ|b1Ks+t)!<i%j&8E07x>gz
zg$(<Aqg#-&*cQMr$C$0|OotOl;_+x>q$?M-FDdzKbT8^a1><jD<rjOSU{grY1S$=X
zwz=(D2N1FFAXB}WAiU<HQ~G}3=%hXa;2{BkBH8k551fV-d}yfUMQfnKj5#rRbFYZs
zORd`b2-&dU+26Mq^OUsS$&ty}H$1*aU9<b_Kug=W#g^#l{9c1*X1T+MJN7RMIiMJ~
z!glK>#i>$rP$oGPz^qvvasbNR?A(dsY1p)~d3{Pmr43%iZ}0g4qiQPv#e(d-Ryw9h
zC~(_jIShbeX4b&QzP0as0}hVeHUm)RiX5@-ve$3VNjSNj8CJ-e?EcVD5*?+wOd3Rg
z)3}^OyjXq_3T25wwmS*iy;eD3IEq?z31GO%I9zZZGD%Q-mbhcR(`{{L^9G9XHokZ&
z7de`Gn`>r{Uaa{Mt)BZZy;Q?g+}O)o`F92bDkf(zwn6NF;3>`=`K!ach8i6dAK7n(
z9O`TOc#N_t?6bXJo1O9v7#K#KDpgvbiUyR-<U*5og+AgTR(`1V3wxFiS;Ll7+8<V>
ztoUCC_mq<5%c)Joc#nf6M$d-or`f9305|uTp_W9c4EpJl+kneW3+Jb18{%>V>f(b*
zQ-O^|eb+-+zlzxBsgqUuV84pBm95Cky6Exc9xK}w+=g%?V%4xPbIa+6$xcTks*+J!
zpeZu^xV!!>FEK2OT|r3bO<<q}_g&A4%i{iBUB5n4%~t$sd4jdTp3{0|baoTr__o$w
zNbVy!1<0}J-n{~gZ?%8A0*5wWk<a>Tgi!~flUfi`Rxc_IhpL{ccFl4L6%%Iup@=oP
za*nMXvq^B#+<u`g_`Ew>i~)bz8qerivWXB}p4hzNk*=0x1o<(E;#(uDf5IUB4G8Aj
zve*Mpt3XNATz@i-bOw@gI%HcC7Wu<bSdw*hj&6W!VA90E<M~-Sw>q`9mZ$Q{%(cUR
zkax#1&SI9f#XK>b@Q{02k&qk?<B5Rk6HHa)kp&$&ixGp9ac1s=htydmA3wJ+iqPLg
zSsAW0<n_+{ZfhuSn;bG~sAU1V{zHn2CA?o{akxF7Vjyo}gP<tp+Duj>eulFKNYa4P
zpsp&cuSs&%Jh{NIZ4uko@;-GSGLe_}DcL>uf6AQx4NSm&neOE66l=R|ZEFpgcMOW-
zR<C8~kk668-Ptkj;}Qtpp}l)&@#biyHN!)WqgIfQHamUPso%VjV=#pz$e*4;&zu;?
zV2!tn!I_oOp-^#jdw4kFmC<W8#}A}0f0K8b!dDX9KpKJ7Th#-W;%Gs9=ep!{S&pOg
z6>|cVFly^Kg)>io-vcNfs&X(Pys&LIeQaYY8THdsTX?-=S1@;RiQD3eCzo^YJyVdO
zw+u#uH8rC?<Oh3C7jR44FQ;cG<z(R;rg+ajHs;G<(bUlt<AeIQRsGJPwqRg|6@6(4
zzezA*IFetSlomvDBiDS2h)DZwwj)6yaDT<qJ@2<E?GKRY7&S?;J5X2rxn)U}RmS5S
zR2Nn7z7?qqm?+*|@0s#AGF^KNS&4wGln)-Cz3(ztk12SC<+5s*VQ=iYx_B-Ap+Hj@
zN{>nwM=Qp-JF|dj_7jru`Mue95ykuI-8uqrU>m_@n&Sxj-i$}J!;mt7>X!Ut-K#_<
zLy3~*ML6Nb|2?2vb5H>1b|?`i1I2V^57@dFyXOY&2lW;9rwNU63d16!<Vp0pW6fdB
zLrA<w!%GJAYxZX!$SRxliR*!l)37tdoc;D0fOSWe2*1I~jDnGKcR7uof&crF9vt52
z!NZr0jcJw<y4l&}M@RUmiB7hQzTPx!oVbFvwga4-*@YF))s;g-hgAHcWR;%lTRj3f
zv4ofB_ecYoV);S0@r9SIJ>)6At*1FxgJRJqU@?hLlkQ}th89e`hPpNL!Uwe@1qaEh
zc#n1wb?^;zL1b~=b)hT6#GRF7^g&;K?VY<bF(FMK1O;S?S-!<1m)6U<`}@9`%~%n(
zMM{`R-$?7{z8M@iW~?U)(m+(iZ+`v8e)-+vlMLG@0ZThYxp#OA1Gk1M3>msi`Gq-f
z$Yys0Wb*QjJZ}Ccr+jpr96uJ-ME2V0={k)$bb2m&B&Xxl-@LpW5f&5}q*zbFVaF(o
z^~|EVN&0>eagXK7Js}}0lm5syrp4mM!C{f644)L%wy2p%xu}GL4J#<Q?N9r@nd%*t
zbp9A?pi?l``NKh}^@?GMjk^pTh>dn#TKT!@o0=*br<SxqA_j}g@+k*p2#$`C*1IgX
zI)18@+($FVZ&mE0M4K{9O%&OE47B-#<tOL%$uIqv1bh0-<6Mr}sB%~sh%PT#n4jP{
z^p2f3>3&`KAfu`Sucoez&=u7EA*O9ftF$h|haw_o-}2IJEs9x7<$~|N?97u^bWS27
zGCIiQ6F#+WD{e#7%|9Hbm#s!FC{aF5B_kS4pDkovS@q{V=;Dx2k*-~JOV?YMw)ZiN
zmrJ`?W#P>~+irQ6&dsZWk5NqA-11wBtt#SfeeGvk-VT-I|0eeU|1WCiWQ|KL2>iAs
zb~ch?dxe&Zn*b$AyUc)Ab+G%!wnsiI(}93K(&HE@YCv)yx&}LzEDj56uUT6@yW?lZ
zOoJ1o;$3itUG76bCrV%1x3%J=RmH&$WyV3{oicd1I4l{bCh=UCLqS(pCE(Nh6BRiD
zP58>t0hQD4mHwn8&^gA9q1w*)*QsxS#&Z4vPelCwHN%yYSJh;cH-l7K4-y&><!^>K
zf6~%av|UE?aqnVb>H3nlm3Q8od~?@@_&?2CU(8!}OZxgkTDsi#nMG-b-M9-J@<lPM
z->Q2;$D=-s6>&Msw^}|Ai3M#$3KU?gx844LG~1|+B$K;d=%VR>eq$)F>Y{=-SM8_X
zu^Jyxdu(3RNh9e}bHnY~sa#VcG|xWpXx$26pSbF#6@_)C&xS4ha*)WXOVP+@AxO=|
zi|S=<=cziBOfFA|LG2-=k{o<3nRq`z;{B1WWo>(w$H?)KDnW4YL-RikVy8=@@v^c&
zf18Mo9nJT0FFkJwf3k)ztIVrQ%yy)zb?e<~u}R7?lFu?SASU@AeAg$x^PPe3`X*(>
z<r-h2+Fy)x-5I{etm~1Y=-OX3uC$-7^^kT}iZg_lmRa!7xk0KH=!%Bg50=~G#H7DM
z2XX*1K1!y&fP3!MpT<{R4m<1f^P5l$!Y2?rdN&}kAiv-Z6pDhJu>o75Di5TPR5>S$
z19B7_HjaMF{3X}&4_Ht>^%d&<zI=OGd2nBmet)UueC&)7wcpE7<$ZEzyNyhjfb)ck
zMN#V)N3KBK;Fja$+oG6D^-%viEL(3eweA5WE3vRS092NX0E?bp(S&^izq<8eUX>vX
z4>vbS_IAoK``MmQ!LowppBU`>198XHIzErFdAq*5gCT*CeT|pvWswh)R{H^)TYx2p
zN)vZW5%oU4H(5)ocXsV)<k_2Uwy;d8G&|khGZ2;eI4z!gC*4{}p}l}p)1Gyy0pEXr
z{<Tu&AeEQsSjbni7I85<+krsG2Vejbr>&3g_^pjMBKaS6-t?3oA5H3&_yveQW}|XH
zXH$mFc2mG*;Svf{PXyIL+a`j$zs)h!r-%D%^vPS;u4H|rM4pwUY`cAdEPFU}mkZG>
zZ3D|J=^J>cH#Fx^e99>cwJ`hBcuy0wq2jtJjU-@xbv`cNJHM2q*&TOk@l0i!)cvp>
zWPMgmQlw0h?SI?%O#iOk{-QZ75ES~z@Eex+Kf!ZY>|!oV4P_-O4v<s&_$lFTjNoQs
z+a9{o@nM`IE)>W^Rk>pnrvkkqeOTe_Ef5{`=f~TAg!%TS4Oi;wzpDAdso4M)1XTrN
zm^uU#N2oY&C`OiFBq?kji?)vr*eyx>Y07!2WF^CfzM=Z4k9u#Qx`GswZ5@p)|9B4a
za=nh*1+l8HEColx`TX8na3^q@%&OQ(L=Awig8~O#NQ7TgP&iuKxWY5ZBrdKDM_In4
z#)}?y{u(8nYaKmV6!C|zc=kG;$H!^Yiua`+=qP0HgvglR*_|mxViL+d>eugnW6K&`
z7j-4#Eu7tMqoBJ_Hy%WVR2JhC{OjH9o@3J3HFi9_x%PjYU~>^LoQG>U;;=+AST*2o
z+JnffRJU<&&xa{N<@b1ssh^1TXgyq1aHdKqB$%lrH26tz_o5QAeSsdK$Z&6mggmk=
zbmCLt<C(>zO6crbSXw&LFwj?4WZ0P3t9j;C&&r3E5zEhV6dsFbOC#%OKaf<pFzioH
z0|?o;tNeQGgS-)HZed15MEZtvugTcp=IuZ_!xQt{Je)C5>aqTPoh_93*IIu5w2;h0
zGM}gt%}$por@bk$`dwt{2{4b!UgI%<o4>4&P*&hka>;Dh`eAcNX!tkt?-e{*MO6KT
zY?>_&;WBRT(Gj*AJ>&ROfx+FxF|f~&?2`|hPP0i<kO;_Lo=~0VKrvYz1o`$4LCD!}
zR<Ed*#Kjkf8-8f0uNURcj3r(7`6T>k=jOj01i4PhjaOI}y;?6g(PhZ~<jIo?=!*j_
zsrR2q3Y#_^ms4)tG~u>gzJN9A5aPdKfhv1su2qu5=~Bee7fKu;ttO)HaDg}e_2;%$
z?3#vdH%mb$PwyAQ0oh5@7SV@0jy;q-XI%jVCLB5GuTFgqD5Zle$<VJx9b2oRtml@m
z5KBQ0COaxl>_Z`eg?jTfnTwqQCy}6WiGi55G#q#mP5E~xNPxeLQp@_!H#^T8{8bfv
z;@)|!YC62rbH}m&EnB@RlV_=a&?h#;$W#Jb0jqc%{px5RS$jpbv=UD*KI7yhf)Zjn
z`TL*SSzbXAfbc|sRH|HwvdTBDA>?Am7g6GB1MIuyHoA-PxRhI3$9i<~^a4uzw7}L-
zt{HCJ3CB)zLqE?0`v48%b1TCufUymD{-D%j((v}*5BD4$yNxn2^7odaC>^A)|EV<v
z4v(pNEF|>@j0|;;(Q{nz1+WJ}t1cpkwMx&oUT_Whq^2+T=p9SdQ`h{vdeNaG+e}Lf
zGf-ft(PtZ~T>EKblYxbOft)NtCM~gI<5cp3`vqbWQvNZLOSt||)1GT+_@sr=;Z??<
z#;f4~udCM6?y<VMy0~^6KRzhPsC<cwyY0>%=#76u_U41|+)+hh;zxgfJQO1n6DtQt
z6;)?C5&$RFMq19c<r;J*WAhU{xxRA4Z!8OHeJVLIySq+zuulXmuDbaIIY3H@F?5|6
zBZ2QKeR}%T9;q{I)2x=(_DTuB-=vIb0`5n7ApR@w%H{pn!xvmeau*f+k4!*NrE_-7
z2khviroOUymFgP9F4a_6PRPUZXR1{QMHi7P?1h4@pS3HgeWaqyeZ*v&BD2>jhUieR
zZx8G`+^P6#(DSD|!Q1RT2q9@hQiOCxwFUbji&{D>6cqcH2He}|2$c>cT?}y&ZEKoI
zC`9rNPoGR(cIzS5e)jBy)9%))WQ==8RFuZ#(V+9euER`@TI+iab^Bt~ugJ*po8Z8X
ztxwY>++5d-ZFCY^S23H*#%&-D%VDS(yg1ayY&W}3sXKA{_{X#FKYx8DFs*9K<J6dr
zsTJOb6V+t$@?aBbU0Vkpl_F6z8T)&I<pB=6%rg{ciGSaA1y;2aRsB*-g)~jMzfW2D
zPgkSI8H$0<@5Ll|NMrRRE~g`ds6fp2%`18mvoDp1XLE*c(O?pwXnaTBqsj_myuLJz
zjHpx3cv%_L+}y@#&dMS0K)%7z%ZH*O{SY9wm((EV%F6brrtU{1rHT05_@8sNgH5i)
z&2I{9zD+tA@}zNd<JQ$Xp6XB>JM|y0zm~DI=(Aio%Nc+Kpj+GOesrBvbmVm?-cA98
zgmZGc9;D6~O?wZPXy<S!{kvF7A}*7xdshILmX>FaL;F?3HKnfHktIaF+p_C>f9{iG
zuJEu46CLzx1;S|p0%9^!j{2>C-Su@!GzQYhK^{ICcd32r-SO(aQmdM}z1G&tu2h7M
ziLUNf-(H+%R)%L<?hPoOX06CS$yhjPez^H)nxDKdGY{${b=;8*p51AEU55-wFik<&
zOlfC0H|8kdlaxj`>0qU&=DSj?@K!)9gN3r7<>>6=LrXbL3EgCB!khkKF}~#=o_#pP
z|93=oasc9`^uR;inLs99Hnu&<POa%~^C6^MflUyCb7_~JW@Xaj%bcIyxUee^yLA&J
zpV-OC4R5-6Z;%@t*oAU@`;W)$apGOlh*JPAuAFCn1)EtTUe0BfY8!$L>|a`IS&I{+
z`n{@Ic~)W1)Y7sm^Lj;C0PNuDdMsx`i4tF;YC~Fg&y8!C+jb3Ya#~o-FV1J%3hG>@
z@i_lNDJ)FOz;HmR^MB)}MjPSYpp@r!bHi-xcOJEiSk`;*mnskY+}}V!42z9n`yOEx
zS46}Or)|=9r^>)<?Um2<<+eUV7v0kD7K}YUEwr5-7Z190*svN<G%pfYpnx>pxh8c#
z?@Hc9LSh1f{voC3*WF@rcoObbOg_v`@VQo^d3mAcO2TI@0v#)*bLZ3U7|})Czxl<O
ziHSGqZCPUD{_(q3sdCh?M5mKrZ};MtM$70SZtvTrjvM6Y4#n!FoDM!!UvgPpiEM}Z
z!p;3{rE$j#HfhnPz6umviAV8w5S*ofQqsSTK3q*FFKPtRZJW8f^KEC!;OeE?+hOq!
z7KaEGnP%KN^%tHi6PGpWxVBi@y;!(2c3YioU%K_MS)0_w+SJn@OJ~u}UbN81<IjxO
z)hsk`W>ss1f?;$q=UYE&@43Mg(0w)HkPr#Oo22Q})#Or59<qi+B+^Kl{m4jxy#J&4
z)$VSRzsH?A;R&zB2g-7DlkT)=203s&<1F5f;uv*Iskb%MOdkrOj?kJ5Cu}X)XdB#e
za-y2%y<$Dv+*x3*_~m@#yA2ZeINDWm8nnaKV0ROXh58n`QRlTqO~Iu54=diSP(3Uu
zyL;AJ<JDu=6TS_bQgKA~w0y&{RBSVTc5ip3QpI}gt4hO9{Vmr$(wBcvXsmL3Pn2}D
z9%Wm|%cRa(Sfv&-Vv+x2d1uo^`=vvLEv`73o%iKPiQ)Ah%}uTCIs4ygUX(1BjV`9!
zye;J3Or=z++K6vT)0EIN=pxc=4BRSqT$wp%w5E(Z`cN#lH5@UlJNe*1Z|=>UJTYuB
z6?!t-VM>k8Ic}#uzqk3~4(3RtRQk&rfnO^eeev+Bdz0dBAH4d7v%%)&IdBq3sm}hW
zD5yg+@;9$0V|rZR<LSMA<AbICl$)I;xdQuiFN&RtUN%tJ&kmcwWfpe-r0~p`>34{8
zVt%`xs>GTgmoHuI`cW;A0B6@!1{J*6#T?W0{>SBFRvM<hLgC)lQ#Eck34VT?0{5MH
zdER%JlVD<_6UJU#s!LkZJWxw=e>W>=v0Ho<jd)h8xF?cmGW(a~X^x&Zl*vzhndP*-
zYS>#*S~X((*qlwU%W^cQb0?uel-y&^jL^V*K>!CiT@obSm=ECz3&|0@N)~T>vghjM
zc~pxnHCj-e>84PzkryufVqQ~lRpBs<%5b5KdRf3e{bvcAo7({&jeKw0soLa8zF4V5
zN;H|ELc2Gak+j9%NkjHbsQJFAF*6N^QX>)K+B`qy9inagZY*>&3N3cB1;Tu08}=3<
zq_St_dyDMZvYZdC`8U4{hnG$<h^tnkJ9|uJUb<E@L%U@SX+NK79zId70xzGQW_()x
zn~QE|DHZFB{v6DkF0h>LO17$&iXiHh8kk?4C7TzYE-;k;PSY;E)=nicpxN^@XM|Il
zQ=_}KV+b=gsZpu5AKIsW2|Qs@NLI9Xgxtd1u%-H7o?B0fo<9D0mW5RJxHyMH+}%Y#
zRbl;;Vh@!Hmk%=6k_<-Y8Nt3qNjc=IeJCC+EohY8Js~}R;hYi$NvdV|^XGFhz7l^D
z?wa<}X5baRdFC(EiEFf)#@K)LnRvn-g3+`9cg<ndy9Yad)td7G)|jJ-5LCMRdm>5_
zT8Wqb%vVgMrc1*=eM0*`L^iF}W3vqiI-g7WQe>qx_$b8XOX=lV>fo*^>c9MabU(rM
zip7iO=B<g3>kj5W?81`MhK|mj#~zs0)26xq_X!x>tXMNE9rm~SL#x2x8KZS*Euk1R
zd)83)rBrP94OEUUvHOqr$cIdyzo!cce1KuF`gzl!4f?s|qZ#;Tm&nCQIkPv^MwYy4
z<6bOs=&v7mAI#UBq2RU_u{lN6J?Bt}*)b=zu{H%C_JTuV;GYAoT}5KY+!|KHA4w~j
zef4MaZ5n5kdHHxchH{ijEQH$HV7q+b%d9+DhWM$4-wASpelHQ(xT8k(DnISlvO;}h
zz(579JljdW(%Sx)(SM5x)K!W*7*1C;)CaRN+&umc7&Yd7-)>xMcefXZzie@{+h}Wg
zrINHWPCaP`-Q4zP;n}?5Q0%({wLewivOUhU`T6WKNd_vVIvH2dwm<6@e0JV;7u$Ow
zq}bTa(>4rDyuu7S6kU<af(Dx-npU;=l8gQ2D_xzo-sAcjnCV2J)E>0CyvG9<rsTi%
zBOHg`Oy1;ph{#1L=T5OIzA5RHrJh)OV{@;yFMnDP9CY_3wSM4%mX`pjO@ez}k<1&j
zZ&%SL91+Z>#Mx-*8_>EZeK7m=333_d;_oY)GmE)l`B^T1&!0n~*6!S_xUco9m$|IU
zk#m_hDoK!8lV<Zy@o8^+Is0xy)R&7_I<*s4QvJNF`x7}wI(9U7(o8p#MJg7rS^M~;
zn^oL4Xdze*wCi1l6S=!ouT=l82pa8?Q2OUS?D)Z-Z6LCg(gB|@BF+WcDU>G;L6IiA
z=ihRkrl^;#W)M8Q`plB*H2)54=;SgO&a-G%aNaq_HKZJQ95eDzY24=Uyy@V)|9xk^
zqkHI%gwxQhrHYF1FaI<gadr3z9{Z8!m#z>NzKqm0v(|yKXX{kX>GIguf*c>MvZUl&
z?Z!`+**0?0Ta+X0$8p%!vh9?G;ib%WCkgv-?A(E7)~~TK&ykZ|CRKg8!y=yGprl<w
zrnD8ie{LZ46IUZ?Q&;8N%pOTZHsORrz7bkzC4Xe;%F`;zE<bvx-t+C<+2f-?zkm#O
zgHMHYlb@;Q{oQf4$F)y0p73WEa5*oF3c5TsLJpJ>rXOT;J|%r3?VHt=Wei%gCYU>Q
zFrHfO4!!=*<+g8}XKAwDPYN*{9yFcn{JFY)pm(K^#yUsWbWqw*Z8P=0IJwgY%2Ml(
zq;eC#x(~(%vj(ZrA5=A&Bedikq9y)F8{Qas9I>0Y@H5Q#g*#_z_UC`^<Gvlj&^T?;
zIB%swk<cL)viQDX<i@C^?ay7=f!xV~&^*;JTFLa9c{#B+i5Nc?h1iyQX^M*XzGzWZ
zyvf3^RW_V%e1H|E=li<q4t<`8@>TTmU9!dBVa61EAO<~smi~x}*=PP_^(nPUXJEzJ
zzSn|wxLyqI$X=QI!X(G(VHxtOOgtJ_pzUb8;v92`YS5`A^8(gpWbXwHqFd<X`I%Cw
z_R{vv{nMdQM1Aqfkcq8FFB57T_q-|1^z<|x14HuyaS)-Bl-Zv_HN26>oSNO0UE6MM
zZrwK;E#&vH_mp}J&@u_JHT1Lk<W4p|Ezou`Te*|I{Zo^9PrxjDdNvY^mlKj2qsB}`
zHF6tGKV^Mpc5D<H-uUIed_kYbbHPyR`wOQNTP||;(5VRmLc%8JceIk&V^R0*80et!
z5H!@z;}J?K{qR21N6I!d`SRt<H?Lp66%nJEl>FJ}^H=JqmgXFb$Vs1k%b(9*(0v_+
zqYf>2_@~8PhnXY1H>D1cya|878k1tKOPNzP<^1jE1qB87vCk<*UR1C+b$G4-4U6R5
zXu`&h;hPgRecdOIN#WJh>ME_EpnyiN-uK!4qnMbOoTH=snCYiapT5a?wt9Qnru>M0
zUzy0&c1$x?B(L9iV|RCCcs9Od;{l?c`2Ab|lw#xYM)^H)YBa*@6P5a!a6V;{_6@PS
ziUz6rJYOn#NUPL34+Tf<tJA}hc+F<AXl#L%$Q(9O4h<qF_4QuWw=Ntcj2&a0S8Lb!
zsO;2TsY!&U1x0Ik_o8MuPfM{s?d|V(r&YOPDY_o`o;JTTG<#c|4QpC-cO@X>vJn}L
z9=WuTL^i}<1>}a7*8gYeQaH@s`iUn|ymWVI6{%MH<mc@*)MFUYr2VJi3t#>@?B3PE
z>pfBGy>{PfQU9%$U85ks0LMRrEE2;hEW1ncoxkeKR%as8c`5%f^M%`&HDbfJmhIeD
zD6OJ=rg*Flf7B6K|0wkS#Kw+jty`cwRB=yypPt1sqW^nZ8w#~>13p69YB%>}+k`4*
zswJV!!Q_2UlAat_)vx6q!J?zoMM34%$fqgMB)me!JBzBK2LBfmmKWL#DBd~T`eo9q
zh#=DV8usd%+3Gp(sF36*)kbN4%i7k~o5^1;U6Fs+ob*AtA34w+ed4?~**@UB5oqil
z*L$Mpe`3{amO=IR;6N0r;^KK0saDc+*hb&;B@I?M8Mcpx><=79&p21EJ}#ZineUSB
zyf>T>ZN6;C`cFR;)B+Za{zBi?XZRP?Q~qF06u#b6U{a}mB1pxRdN)Oa?!m3c5Ly2T
zo`|)nweQ=zwGT3V{vHgDLa{vmex)n+R=N5!huyBPBm7xD|BS@d;q&`m;eDL4{)LOy
z?t{w;-x~w+IQ4H;*F4Ucta+LpNlq{|E9(EAlp|2{v1^^jtE6}~J!<sr@o_@hcR@oE
zDVOu|sL?`FUypwLV3at=KtOOa^%gIqYMeJ&v+kr?T~&+yLZBLR2zA;MxxhMuSA{?D
zX{V_0?o^Mx%1A(=oNlg)1;k=~2x!7m{=Z))<S6UfSQe&DaTK3@#{zNcg=7PI=(hRp
z-Mfv=&3v9~jB-j!sqyi5&@xg|UyV6VF3dE;>Y+x%8~zT;q!`{ugFZPq$s1fIq<tR*
z*T*bja&kt@j9^`vo}F!8wh{G*mBlE5m4Wd&w{AT?If2tCbR<#`S&0-`|AisFW(r;#
zC5khTaVBBCeg6FUcUu&1q2;VZMnY_OXmTSq`>$o2=11b_bXn71ue>K_&*tD%wLTJ;
z^w1{0fb5-FCAsCXd-Xp2eif-=?lUgp*>%$Y6AxYt)pe_u(FIzeQtcjXwi`ivdH+Oi
z%b8RVj0kDZK7Uo%8o{|(b+{5h?M=-36xxy()Onz~&SP3jJ2;db9UT>D)n3S$M5SWs
zwnt~}jyOsJcDTZVY1j>)9~0%edOn~P>#_brvCc<X*Uq4<pFj(7i6|Z#wBc}B;orMH
zS7TUOT6!=YqDhU0=3)x@x~=f|Vwqo}(b8I4aaU`l8hj|Se6V7#aXD<I`jB(qK^u-#
zl-{{>2i{f_BK2Qu-jLTG7XnzZ=#h<L9u!JRXM3UB0h)!;le1Yb$U$f8>|nezsYX7O
z{OL)eA@p=~XiPgHnh>@4D(Az6gT)+0E31O;byU~cl+HBhMXm9>Pt~+iezM*>WBfU4
z;d1ArpuoUH)2e+TXqv8IVv-%7kkDP}Xp2M1r^L$2svvxB?CyQLGkc6k=c8$h(l*`;
zuB)u#vN6a>=}cRc>cl}o!6O=K>if5EKe<B099S)AnuO9ctJ=L%H&FCewHA%)B7mpu
zD|k{OBP%Pa784wt44-h`pAIQIJvn5zTYKe+cXo}&iT=#kCZ*ljnPIC=afJqy)F(!z
zZ;zpcd_Z+|HKqO34Ryr2TON!|Mq1kM8kIn1ohOkh#f`^Imo8rXjht-8SeOdJJl+!z
zr>=d_m8Dv&B6guhB*tmO11?K|hZj#9<0P{TtQ@|sEs{F{KF$C`UadZ@%C6WDVK?ab
zxln>y;O-eaKelPip>g@lQSbU>1H0)6AG)J_wJKu|*|R$jp!31i^XJZ`>NE!*)KZ<)
z{3%CKqoGBde<GP_g|toi@W2C<VQWW{C=FURSxkDiEs6l*8jR?h)-z07cM2Y>uhh`j
zPniVY)i<^qJS|rP^zEaLQWp)lYl4#d`AD2etwhRykK_8-GqFxMGx;lZp~-Mr3u`6f
zu<cJhJvG&(!i=i&d`|>j{S=c845yod{1X>*sz)Xsp%%&l18+h@rx!v(=`ccthBH#4
zU}(OS{M6{fon;0~F$;@4>NM>JpTgiPuB#K}ZYZabjp^p?ZbeNxCMIzR%QBy?QCLB9
zcg&j}UaeiQLm5gdC<H@yT!aQK9{%Xh%5ZrhbQyd*SR@`77q`%n<mrBmLS^L-56Y=I
zMAL;1U3R=PXy1<73-!RKrH6)wU|PwT=7Dwf^-b;VlGi>x({$brDtbGZ%<*=>F4z#|
z8vZ6U@(!h|@4?C8>h^$9PLQKxibT}bkY#&rH09W{oz>BtPgE|5lC-omDZfKOvSf_L
zObNz_@$ZSH4x6DCZbGx_KMBI3{sJd^+U)wjYbKtf3<seBJf<nU^X{-MO*SqLPB%0*
zO#A);yScr+J>ClCbV+#a?3(iJ+qVxUsZLWV`5jWF6NDh#!U)l5bhv<Xnak0-aOKIt
z;z6_K$!o+~4po>K8~fPbYGVp59wVTBSIFx(h3!z>w{H&!Zr+>&E9vwiVJk2je`-D7
z@nX2lI-iRbm654gt>F(A$=Re{Bb=0!G>@4jYiC$#V3|Py0SWFnWGY{;y5_`z>E(av
zZRJta*H680l{gmWXYM;T6{a0zU|1N^=n;&fl6DI8h9HuoQE9bWx$9tZ<?`hY9L}NG
zp&>QrweQ&BY<fN;_Olcj@n^p&K`;8meBmvuHU8!hmR=azO$v%EMGgBGusnic9l$}|
zc=-6+zR1zD#h$FrmN2&NRO!1JlV?+B0jZ+`<LozQqR-m(W~&pXoRLCQf7I6UhkU{z
zQ<w>ThlMTTe~p~$R!XNbLn*3@X?}jb>DRBX!zQ!mL|k#<L1V5*jqXeqmsL}XJbU&m
zz@(?oo>_0UaxO+Yu1LGMRL!)6X`8E{s$9XAo1~rak^*QYRU*$le#^aU6bLSp*-{K9
zV%B2)Qa@@hcp7OA5%mXKYdQ#b#pN*@bEQsG&e#3o<;7?zR_$_39mT(jog#2nsuR3b
zEIF8a0Cknj`Ms|-HILbidc^##4z}hWHgJ*q;7y*{BMfOy0v90a51laIn3JLu?w?%n
z)gqqup{r_Lw&@kQhpO_%4C`xHSQ-pP#D4-LR|sXX90WNj%4%arJqz6#G-&9;=vOV6
z!iI8c1%sB8lS|ZX4X3+zuYQ3G#VF<E;!{viU~zh@LbsI^Ez=y##K8FeOe_!h_BFJ!
zvNGmjVT}9z`?Kxa`}_MU#U@cFTPcEQ|GzE?Hshs-q8NWyBM3lq1Ee*aJ)F}t8oE8g
zy{hPSr^|=48wFDbah)MUjjr#{#)$e0`H&l!<DnEhzI>6%P{^W2WBl&U$}mGnGtK<c
ztOv%;=j*ix|1>)9I-nKGcmb;hJ4(P=WwOCH$4m?!l!nv#i>miHd$AA-jhX`KXo9u5
zdev=*>##YH8f+aM9fP3~bN|GTAJ2g=RE@moDlZWukoAeu;U(P9w9ou#b#<~>*pC|(
zC`0n|EU1_um_O^q9s>wB1I4B>02?rEOs$vtq_MECbrC44eUD(R`2_gDpwXY^V5NN3
zYVP;#7tf!kI_|EJ3OJRh6zFSN^Pvcp3iO})!Q3Xf93RZ*Hc$dcL3wgVNn~kMs=$rP
z7#gO7HDf|onk?<G-K4kSU{O~0jWc@}FaMtmfLfEu45-1<1`u=|c}zpWXD0>SH{oZn
z*tD3(r+qL_$8&bqFFM-9njGA6tO6ziyIx)%DwSoa+KZo<oa`<!*M_@hfHsN82iu*s
zxPqB54-{NxkCP2lC<=SRlIc=9;YWF%J1;+<KWS>JF(89AYF8AEUSEm`3=G`#5!~G!
z_iPu&oZh{656OVT`(4ZEczIRt3j1{B=`@QB2g7_p|Hv%me4*8fO&@B5lcRl7HeIo|
z4Vgq=uDlLu)-m@lODg%sRE2?^v5zo!CnW-S!eM7AO}*Sknc}aSBX#EQKboMO>nfcs
z*4eXL3mG|vKR@3v94JUTI!2Cl!joaX0CmJeOf*`bsDsFw$HflciQ1$5=d%#d#!Dk3
zBQpWl1eFXwxC%x{$H*uK&ZCE^ct~dO#$W6{L2U$)i1W*rFIHnUXV5YbY(#>}<;S7Z
z=XwLxV$q<9teBYB`wt%oh=@|Kshp(1w;}{4<vPuT?U#x>Iy-65<u=P5fpoG~fC(5L
zJ(8<*-0c}Ot<D-68rsxtY-;L&aLTUJbW`1VyBjVe1KmLZT4Z3t3oGlN#UAIZmE(x$
z=<Jke2N4KMnC_uvR`c;88V=|b$(X8GK0vputl-<dW#r@nNjZ#D5v^Kxy4$zE=Z~y`
z@4Qt$oDKiAFXFYB`n+UQ&SNz<R+|0n)3s=)IzdV31Z&SGx6O)~8fyv&87Nmg<D%1(
zW6lQ;z75Q%SJ>r38&euMetY5oqHj`S;#}qKNaydkXF*@T#*U4R$qdFg{$MH|glDQd
z515mIRTrss#~Jvh^#KzWv((Fro<Fh6{_*1}yJ1&q%izvZe*(Ybjy|9i@G&7(CkIAu
zpFSzh-0n?Z7^;Sd4=aWA;hUEptYEplJIjOH8_g_6;Ajm;tDH;i;XzsrzAeVACz@={
z0)0P1CY3}$5C+Qwg%&(f@Iz0J(9^SKbBVGR55l<Ap<PE*V0OAM=5elFApZ^>d_}(X
z8!rqLW>6b|)8PS(7Bm|P&I%Lh)18j8UU(ADL!4%kpE~Y=sHd7f#NQjnriTe~I#9_;
zZ<>ok#-U&1ddB+X<;cQFHI>VrR@&<=HHfTO7p~ryN16!P#AU(7u*k1O8-ZUW<*`Zy
z-@wo!qb&co(~@g(>=;;bQpMVjE78s1X8+(z^2Gsak12l4C<++;o<!t*G|ZQuz|q<>
z3IQi20GTGms`mEwA`;Nn>Kzsh8WYn1ThT{aE_9{Aw|6e}=UX3a8I%qqCJ?XnValRu
zXd)$2R!V21EoO;f3@68{$Ra$trsn2N^4sB2QCV;!bKyqWepW@F)pB3gy?`biGWdtP
z^C>Y3c6KFrczAq=z4KOvaIG)i-a*t9{0?%XE{AW!25IlzGu+>p_HT<$Tpb#fg(ZqX
z6Why!iVO-!=5L7UB5#rTq>!dZ;>TN^lF^GU2lKRH$;uOYFk1E<VvN1zXaR=|9QPBZ
zWTgo@0GP?z;7eudvh_z;*TLbyem3lZ)_b}}ug{+!NJjC<uW*5TxJLTuSKyl>paosa
zjg5^o7RIAhl`Zyw-&%oOIo5;UTl;}fZJBOt&385gz8P^?%+m4cuSTk8T_A)hq#XiJ
z^3wAGmp&I?oa1o}eVSWaT3Q~?a>MG-8~T7NpfnRYajfvi#k^Ea*U70uPfhvI4uEd-
zpzX?#@``qbM+0W;c2|5bPf*7JB7P8f5$J0gR3ZQkm4SmH;7TXEGL_lt8X6eb*$Z2^
z;VA%{U_d(n6m3@DRp2Rq^fSL*fMI^AESDha{xU!Yy4@+$I$<xuL*O*cQ7LygBE)g|
zx4u#v^yldWg(oC;cb!D>Sf?{rZgo5~_4M&kh!Jq6MynQ^Fj+eS*qTipV%4t4Fu>^<
zB#hlip_C*&o&F!Azjw0&oT(Z>qoiOK&FrNmW14&S7624?f77bHNlLmvlJBr>2mm%%
z`MvI&H+DUa&SUSe;UDQ^W3SpJL2NpUg{4zG2tNc=RIB|vd+U?aGo*qpNA{LPgBA9c
zrIS7syve(xh$e<&W*`b#z?GL8Ul$KrLv%&+lxrAiX=%Cb8n}fIR97=?BN;I6=)==2
zc)KIBFBxa{4}0hV4Ln9>JhUmjNgFf66&}^yux1|*6A=<7pBx`%KvU#{&6rb0pmgEk
z>71iZvH%gIsSSR8z0G5@6bjUmZ!k(VnmY=S;Erm$Qq}(Da(6RxdFGF|Z#g0c3Jemu
zy5yFJ%ZpoBF%fU}bI+A=D;~Hf7CWc6scFY#l$09dMNn$vPaV=Mzfff9IT_5Ps=x)f
zX`wGSfx~1-X}qK<_5TKkLDwi(Psmolz_%f?FPI{S5sCRP`cq;M7nRp);z4@uT5<+0
z0PXFhr%lv#rcaoB4Z;g;G!<GfO~kv{|2`Fwdt#wurAH8L1522Acm^!@!Q%uNgcBX%
zlaZ;eI8*Z3efuE>OW$#4$&b3wusgl$ZJp8Lh4bg<!B&66yaYu>F#-N59$aqxW;Xug
z$^bT^kd}6=N&tK!x_S1HAKd{@$B-J+196N2&?D+`&`kmYduCq1ip9VHhb;F4Bn8rn
z4v+RGeHkjCN%eAcbDdEZA+O6(350qYH1O2UVcXFxm=AggksJ|OKmB6ZbFCg*r(-T8
z9>KA#7A=UlxM}aM46BYKjU#RH6+I8&g0BAmT(DxRLo~X1xZIY-ToptY0CIF_htuQT
zLWk`I!_sz$9R`5Ibw(5bW1<7@*`FL1(+=|3Mh#3@D>YjHXu|Lmc`=oIL&H;r7fMQ@
z(B}O%`d}`>cWbj<Foe5p*=7;wRgPJ<rDP3avo-f;;R;6h@4EYUmIE~^9p(#rH98^W
z1o`>J9UVEf52Qoj1aJZY8{*ApKws1SCV(N5W*nx26!}6XCnwqMH{L+ROO1`Kv8-&K
z?wIdP!KevPlcO$`JCGzwgz4)YefP0CbI4!sH{U4?P6@h)8Rb+Kb$L`VmrmiePh&zz
zhGLF_hDP*kZWNdKQ-G)nYHC9-q6#wkFq|EFnh%@gto_stl9xR!rq^)U*J+=|#56!h
z(W3!_%RRtKf%o0YQp%10VSnfDUH*_1@rYy~6g^_E;4W)Cr7_kh12&o}Oy!Kot*@_R
zpb)?k{$`>+V-AS08G7x}DtT`_^dcNrD-{*h%wMFHsuUTg;R@{4Y>t{%ZU;|HOx(PA
z6Z8e(GDpY91UGKPMe{o@c+f`w9<Tkx(839`0b2J$8}BLzXrX#O;Bhg5*`FHJ>0`{w
zEX*1<*(K*j943BV!uURF>Ytl^@$vDG3_7R@2nY-|rW%9#ooT@OAh)0)rF;ZmW!E&e
zd3-a&<5*Q5d_~lSYvlLQU}Rl=x!U~IAZH7L1|-3tm@Nms7>tDmo!2i?&4dF9SRm2V
zXo#+|Z{8$=<h=9&5K_7j6*goqQDZp&*d71+PyBFhOWo4qQ6S1Hxv$;8O{7ns+>?yX
zEFN*l7#bvdq#rvq^`;P%wU)tAyGa~LbEV+e*jsOh%jBz1k5@Ayw_g*4e4jn!R<vFn
zO~>ffA<RsLeD$ROq+&sau=lwNJ>Xd^hf0M%6ETcnGlw7<4reiqt2A0h7vN9q3K+#7
zSc8C(6ESQ7q6vmA00O!T*sLc@H3Qr>DJ*pNt0eJ&wxVp?z<?CO!opNaEs{>Qa}W%=
z!F&Y#uyrF%NGhlo3&2G%EIM1gT!AY@i};{3<^(#J9U9HFAYlg+o-9^dbJEt!1M=GS
zUVq@;Q-BHvmOBG_#I)f)e}zyV2r34p0p{!~w>8Bmh4{1*auD``Q91rOY@_ckMKYej
zYK3K=9K@iw04|`rDTqNH)Kkobmilr5yL*A0$VdCr)7zWLG8&ed!RiQ7iBT6Yp$34S
zTyqBrOu}Plt<btp;_#|VD}vK>C-Ubq01=ENSmz-n4-i=f+zt)elZd6Km2=eC{0_Xc
z=J4Bd%sd*ZgD8gall!Y9m5!=J7`cD#HD+O>Qf(L%GJlVrp4c9P5(<qRABxK+BL?)e
zw8DTV0BQtq828;mW4IP98zu=1t+1m76%0mH*bI*a<M}{4%ll#O;^Gp{sx8Engb}tE
zsQ!j`R~q14h0cdIK(SjEFbQne-ieDC$|YA0AREK`FjHM%*iHZc2I@(rpPLg19F8>o
z_4V~~;DF6k@v6H(r3XR#&a;8F6c-QA7Aq~F$b+jqmrZ<EI|G?upeN^T!Z3Lab2+g0
brX==X-uzByyDj4cU67=wtDnm{r-UW|CiwyS

delta 93188
zcmXtAWmr{Rw<QD=B%~V^0qO1r>Fx&Ul<wLh-6h=(($d}C-QC??cjNnA{=jn{;IQYK
zF~*o<G1UR*-vd`>;0iSMotv-TvWnd-H&j*{#x(UWYgE_D^2@yuGOXj1v-v0}r1L?J
z?+v<Q^<fihj^PNy{CLxiXi_8Kb4@0&kPQu=JmTPj)BBXX!$nNOU1w*x<t@$k_LC0^
zM&6Xkt!IC*CWpscku;<Cw_kkfQ&zr~i{>z}WO<GRUZ&YXaozx~K7@1KGCE-bg;TM$
z$Y^MGW~^_aR(t=lk`Ip1vyDhJ>_nk|2ofkTSjQR=$G<r=PfM5m%GPuy!GKzV)0pOi
z*HhPcV=OsDjl=ZVgxYh3&4lVk=ERl)X=-NX<R0o79lokvy1*UW0D*a(T8YP#W@%O2
z(L0IiNCrh`e`yTB?%dH)ax9zo#Ae9ijIG6x*t`Bv3+&Q@z<OcfI(z29z4i=$OlGo7
zSqk!vp?*rmb7~o<k^4S@#00-DEZojKY>{etdYvf1Eb@%)tz4+vI}k0NoZ;C*FxRB6
zvdFB|-cz+k@Dc^Sf2hT3!#!PVOh~0ct`s(({A2M$oHan6SiL88<X0>f$kB6`vVzy>
zq03Jw65I5ym64cN*w8?8jvOA0rT`a;{3?dP+fRg$3p2wzx6fCqoDUP`?T$@u0rRzQ
zM$Oles4SM|8t$Bsj#TCLZwoFQGZp`c!XxXxWX6r6Zyu!RZzhD|Ke$1qe_stgNoOFh
z>s9P%fdroX!^n8aLR0KuBcd@2Z#>60O8!3d@SDI-HJ*0I&9c|VvQ2kR;)`I>@&yA2
zSSGB8SWU*3(<TfN!&xs-zEC;+yQv7>-2Gf0`{*OLmM`a}u7II-hT_<If!gH3J!X2>
z^m3guWHH>)lv8uh<lwK-D}_48OXQ}S2{U;1bD#%XkLIXtp_IoX4Nt>YZJ(B!J(XQS
zWcQhl9D-wau4Y;k(rVf#bvT=}i{U0XQYkJQ5*;3<wCM&~xSRA%se9UKM3t*mA-SU6
zX^o0#t>^ofR?JI$u0QaMuhH_UZOLw>d=1~Gt=->S^JYsH!#ueG`YHwsLuUjxnR%e)
zFZR!^i`+#!x$4i`+sf|u8_Kd@YjGN*idP(oxp|5vZ+&50l9!o6wd`Osr_kl|<jl2k
zISU6H{lb%kHk7$VLz`RuXf;sXT&(u+OsahRi*!3GKmMK2+Z}ged-99rVZ0pQ6ZFwy
z^U7($`r$u9Ycp*r)`)jST(+Rf<O>57sowhXFYp{r(>&oic+x!o@LOBC7xJz_c0J}i
z<bnzee817@g{Tyfv1WG5`h9UnLhdX2p$=Qf>QwaC-T?}Sq@qnx$H&odN;pit9|yf6
zE5%@t*Le&z>y=tOgrpr|FP(fHVQIv3A-bpDuqp_ITWnaIrc(5c$wGQj<1hk-km|*L
zOBa6DW>3cGf9xcdNQuv27mRK;@5{GlIqMkjlB%$A@RolGzSeYReH2tFion~m2n3!o
zugPpxGbREbyP<U3uG9C2UK%OVo!|UywmXmIdsyK2hc~TeGGr+N8(brp74~gTYd(BS
zofEYb=29<)b<ig5i_BxMZ~-{XdHx9Y&Xb!|spG;q7<XrOxNc?3<D9J>0=u&hE`Q}M
z%eM|_EM7{mbp7L^V(`6kWu?Fh!Fe}R+jxS-B5gHroaPsimeSvfwX#+$gB3IDRwFv5
zXCL|ighYzn1=)r^NVWBL{^G`qi%2teUI_A!6jtaBb3l;&`kM*w7I3N4`x+z6gqKYj
zWTo>YVVJA(=|?@1?s&}s9b?V<VQfOta~WpdG$6b_^nH?jFMTR1DM57d?g;;G15p`J
z4bHdIIJ+$92w6ee*jKX+4tglyOBpEvFulrAwMVUlU$jDE=p<{7u9>HHa2@tIlbxFH
z5zx>W1!iOnuSBZYfhJR2u@4#fc8sHD%$r7?b_a(>NYpreiQ)SwZZ5PuRk-x}V`~c)
z*S=ubKM@lA?M?g_f6j6p)`zi{D}*iNOxe|yeHcC3la0>jE5i~RqDdQw3)V%0e)E)o
z;O4$5!QeVqld~gP@k;d}f{oA+DQDBgnM=mg!UNLWZ6M@j9XYY;sN=7TLI6igO8&({
zu-!H<w|CRi{)I8|yAcS8j}T(t_!Y-z6PT=4Y6fR<r9QR6v1TaZ%2jkR!fstz87WIU
zPo9W<+vONeoh_beCUVr>k%@@K57#&fQ)R!4<qT8FYQ|1dv)k^Z0gLdD)XIE%7<7CS
zmCp*03=Bpt6~5ErjRD(&SHp>wJ{luz^ChI9_8kg>L=xQZv!!AG<C&pl#J|w~@Rx}{
zP9Rp6RyQ-K(J}K2!C!REoz_K`{;B1KEfDR273HrZ{3r7~2qzt41v?4YX10X3j+IiD
zoA}q*eKY+3+--2VdJkb)#i{%?Oan-fhVb!EnKEyTE@b5`ZIPb!MSEm<AZVw7%^DsN
zDx3nRjVy%{Tdni>LEqSW9QwLN7jkrX4(_whKN@(Yb|Z_EXIs3IRAi!~u)XCA+L!wX
zzy|cj0iQ1`-STS;MbkPd*|+G(Sf4BzJ4q^vY3S)<N3zcXGam>YugkxlVggHcB^Zr6
zio&_Iw>6*kA073GJ5Gz2Y8l20(T|zU&Yp{uzG=6#mN%KR+v%|e+C&-{X6v2o`W5C8
z4b5L=DDCj)T)TM6T57WwyN#Cp5;C^Kn5_P?Bf#;l{6nqFoeEc-`1%o6uYbji-9^7?
z*pXIGU9_8+c)({eMgmR|l|`Tt+GP3o-7gNr&hFto{=C;-`Bu&_yH!J)ot-^9H<xFZ
zsh^s4=8!*6_9%bK3?3ey&>nXka{SqxG+n(|xR|C6xxo1nV+pygF6YLv2q_9JOYTny
zeT8j_EWRBlvR0#yiFeq-K<Ou<m)fRD`WOd7ysCYnEM<}rn&A-@XyCrdaU+u$qO9_!
z$XXk-W<<-~VpGs&L{U+|;tK?(V~o?WUY6Cma+b93h7yZ0GVeY23(`|609ULWtXwEm
z&{#2`=FqquXj5a4oJ^51Q<P;nZK-L`<|vKPvK=#2EKC54^zzfIgZHbAq@hk0&|;h)
zHlwqsqh74+8Rz^y8_11)R-Z9>z8Ug5JhZU2{UfuG+t}F1d5$K)<@y>0cweX(@fI9`
zsWAYccS*)*;-48}iM}{zn=oN=dTb}X>DVYPD?`b@f8a|=lhC$giJiVSybh)g6gTuy
zR&aLPwzzxwF^-9e-s6UU3=k79Eqs*H?E*f{n863>$KC*o7?}Be!{R;T^U^13y~yHY
zGnc_(4WEOCTOuRHs#;Eho>OGMqHlRNEz?p=9o5|V&X2?G3*Ky?5bJN40il!oT(%XA
z<$tS1<b3NepYR~hJ^5uH?f8#<MKsh}US4i)0`~2B74tNGP?QAW8@CLMjQpyquDqqQ
z79|bVR`me&;@IS5{E(24AYcwFJn!-Gab2c5G&GcGAKB(x*Vfh+<ahZ#Ub?Bfy2F)w
zvYwlgjrIa>gIqG`dcvwy9tV@<X4;fAY`+?v^ksU2WG+#OFfvYNTnV@n+nv~$qNJF;
z)kC?x9xZ~p+8KzCl<b-9<w3(K(S<e3JJN+?*6TnkY}ldtPy$+ms8~oHH>aX+MFaBC
z{B^F!$j)1f`Hc2E@YGqCJ}br89?RSB40mM-gFykauy<snr?WGA9?Lmp$?}CNV5HEw
z?2XoYteSnNP<R9cwnjS=k(;6|_g<dys0~OhuZ5w3wC$tRbp14})6$tn`J1GqBn{0$
zpulqH+ahq@(7DkzUw=8{>gdQaH#hgQMEkp$nJe%3c<`CvDP#&trrxH`M{okJ^USq4
zuuA~9KeL^eRPtgd^ox?+=^4sDax(j!mQ{eQ#PBF8!Ty*tIj+YeG}L462bLN0+x6}k
zg>O2GV#SWc{V6(^z620&-c(aLh4=>oDOq^ZCCgugQ;0dKVG%TLY<q(0J&w($W=<Bh
z-g39Lm-g6h^po?|eEnUQ0!O;-xM@g=isg_(O$q1?^rx9D{(*bNLaEIT11UO&g-X0{
zf9@i1Qlg8bB=J915v!`st*>bnGRULVU}zBwjIRonp2LYN2&y`s9prGZyI2Eg!A3G)
zIM#!XX%*qESs)QGd<)Dn+iB4-JQJ=4TCyBw$Xu>cP70UkU+2GoA}smoD;!lSDl7Xb
zv?BfeB8Uj`)N(SsUaNamy9R+`<K^m`On5_k=xR*z$}_ba3(XpM1U$)2glRwha=MZC
z*?u0GZ`)n1T1}cgF#OgBj53r9*Y?-C>YY>Ia2uaGL^EVZ|4u+u7sX4=W^^&RX1XJ^
z8Gg|uJMQNxvcS(@sMSi?s40pcXt(Ol7(-v*T=;d*O}0LjK0<)dWSG~5^5GtNr+Rq(
zvYeoil`9}{B6cEEu*G&ub*}Mge`8-Rx;8*IidbfCbtzlZriADPK$GI<8Jb7x?o~D7
zSI$+x@-DNpCb@FwTI`3Uy*P)Nk9@06P0e{(I35e<H)>Y(*tE*{ocfY*Op80$`kmFp
zs89pFL{yUyU(zpz9)GFHFE|g0ZttMVwMqKFiB0++Z2uFXTxi%fSTkF_jU`!(zsI#N
z!v*EGXa^L9n*YGPc?xjVTHYLuid^KXOC-<YxEMMiv%b2dv>}q3zigk#Ly}WLMzL|_
zW=_P`PDkWy{QX<QxIl?FfCF&>8w}c?6Gp3YhbgXM%LK6_q*6VjBRy4!;#d$Pj-q3Z
z-qh8+%-mGISx|l%d{nc_P{f&VoYma7$9EX{1vL5=b2@OgNFQC7qSSj*7_-P2-*bA$
zb2xD#YB(}e^-zEVaQEfW8Pk(742^OnYscruS9Cv&7Z3|2tCv!aUJ-FLQ;fS8k^dII
z=V6UC<n?%iK)gD^@qO2rcMs-Ng$dsGSMx|{aTM3UM(Le`TKTFSokClK=V;s6a?iFI
z057<hQA+vf9$h(f=6-L4Y(&K&IW)iZAS=L?D)0j<_TD{Aj77M+8++wghK5ke8~ZL@
z0YxAHRVh4t4UBMr$mB6%OZkr>+ITh2iI*&W5=^I;q0aWO;nLD{%bl4%Nf{&XdA`c-
z#Ib4TFzDURCP3h8GX1)|rSrCx4C4ZZ#tf~XPRDE2rkZO^49_o%B}Vv5Bh#Y0kJC~G
z6lBj`n&X1qwrv@mNgR<%-aG5+73h~Hb9Vnyv&vG$8MAxD+_%To+O{yyiF`Md_87!M
zJ1m$*ZZHie6OB$}C>y4f^N>CR0*G%qvweGrI@p!P$_l;BFeA|^xS^pbXjK4Sbx9@_
zk=2u5SVvnI8d6?sB6bH0&1id^%114!I8WM`JE>m}B)6@?S~fG9Fcr`l`tQ1y07iv&
zHeT#)Ax}#r9VvWKN27U<K9$!BUH6RmhtTDOZ2vN<au}mRXf3PyR#mhtFHoS1t0O)^
z&Zr-~L*^U^bI2JXgx?34B0L3qBI8opV-pC(YHc)2pE@<?JEpDb<s;IDM#s|RTGYsz
z?A#R_xz;qyrY!{)r=Tq!O0en@0u8OQr)(VuB*UKBF=Ha>Pe#;zeRHJ9JE|U8+&=sI
z!o1;rAUM9N$N8<lG+c^>_4ty?JC**Uus^YjanuA}?(1)#lC*xHZz_7s41QoyIwg9z
zq|Q5#tKyxnfS9WpYTO866EY63=ingq-S=s;5J}0kIBQixnEAgo#90uzn!gnGCQq$e
zdv84R&o!^)RMU;doN9PQU1o@&sIFF2y-KibG~iJ6iEnqptqQ-0ymqeuIDYuXS15=v
zSMCV8NB-^>20-j8b`<fhQTY>MdbM~1QKZg<SS+lte;d~W2!i2{Ou_8f#yMiEKcxN}
zMSRT8!p#%n9js%xP_xuGAf9_tVRdm?sPKOHOzyjpWuM53RVxTb%j0hTUT^Tx94C5n
zux^x;gA6NP!flqR0=lh`Ojl2@n^WgWtcmnUVyOoJ;aHvhHk~8kRl&ktsO{FA#VM2I
z;E%aFPr0i2t_IIJ3y8$a7i@hyykU!kHa}DTMUDoX(A?FMGnl(MP4-htbKU}(%{|1~
z+BdQd9p*n{)ACw8(<%V>Nz2ioo7|Vszyx-xFrTv>dxTQcW4XN{avwyS(|OakVjT;#
z5#+#%Rp(;+{WF;-vcF@=7t(bKrfRQiBtJM=h|-6jk-vc0AyQVd!f`wFS2pd8k`4FD
z|Kx<ij@1zOzTceRl-OUIhM5;5b!u_9PF?g!JW(`=8+I<FJ^SgT<WeQE`o30=&4?y`
zBcA>aVL(9+G?+Ts-%?1J@1Y=Krvb-Cu4lY@WcR8?b{=?qt;YEBUAhb7=Dt)Z3HJLP
z)v!``EUcGmRnEIi1{Pv)$ljWZ^bQ5UwSs@VO&WDqi}nS>2GLRz>`zy%6@d)J82bBV
zNos;qE6(KLhHVu0@6b(tT(z0IR}vYU+Xi}jvfiXmyj?Y06|WpR`!pIjylrWYSLawo
z9YsRmL}v0e1CKcXq^trCG!Q&Pso6^s2wxR<Y+YVxziD#%3up1GNnxJ3PQG!#?Hv>n
zxqXh-_0i!%h^F~fO#epRE9+f8GfEQN4r2!24ZnQP@(82rd852>?uu29<%^_YS<>2*
zPjGF50G``9(igk!TWR1I2RC7``0GqvA0vg)20v%Q=U-gZf;j;aE4Hcke`Kjjg=J8j
zt{{xPZ+_1eDuM&jYL|`U61Sfha35Gpp$rV>lCs_0<Lq|1pUlt-HwN4Fn<9%t%HJzl
z?St(><!RIzE2B1#7d<s^h!(#cZZ&mkKd)f>%8P+27GRHe3ZL?TBv@cve!bY&1mVoO
zuaSY@gvZm?rDey4*SO<_>Fo4FHG3d$3@u&u`;K<_vqG31*bd`vXg?mG^L&=Es@2ES
zSD<W6Le2nXJ&?vwKfCqdN{I!%eMMim)w{Mj;TtyV8V&RnasGdS16#+EhZt~7X0k`w
z_5j#+5_pMDM;#!2C;3>`1>Ik106yo>NT|d)B`VnnZhX;gi&Mh?P{(@Y$NxnLRD&|7
zOv)V3L^ASci@xE-sY84r*?p;U*MP(F3s31Qt1(6aQIgN>TZW0~@^ZWOxgj_#<Q4d>
z%dZ!O3nA}xXtK)qq;9J&fKDNQkmz-dbwLxo+^f!EKTn-z-)iKCNS0UL{E@X}DXBT}
zg<H)0M6(Zj+Tw<0?~!F*?H-#*okjwxW|gx8o7r9Iia7d~2Vp59A1R0@8uaFBr$3W_
z{vBYx1M-Ki-{Ke`J{H~~^NUPK6|nEbkVn<0W1;VA>w_*$yrKgbSX@R;mlZs<k#J1K
zR?YrsFcp918iEiFYEXKY@b0y1aXJ4FLhv~%b3k+(VZ69j1lG|AOB>a<flVELvGLL7
zVjS>d&B$V4=my)5<#H7y9RovqXXoj9oN^%vCnbS`goy0#8@DYEp&JG1*tHu0!~S7O
z=+@asVELvdu#S4zby$;@Bli-W8?y#2rn|p8bskZjdza}lHYugR?YmktCYe*&XjgAv
zbHhmWE1T}@5dB{(1@RKhM>6h`_Hx^C2X37sd@V}+1KqgD*l)cV)QgR-iJLwqU=Wao
z32))5(@m4KW2&(g!N)b0Ioxh900h5u3uO+EbS(EFuKe@^B?c@wW8@}%kYgX>Yl=w)
zfzy<)fa+zTH1oMd=qBz{CUFUhtqB5sq;7Ujr@pSKcUCYKqPl&C<wkYB;{A@{$tIHG
z@ROFkq@}heMi|MTxWz+BX@aHUDFiZ6v4GsNG6+x}b~j_rbZH<fE4Ks2Ib&mrq2Y5-
ze+1R92j=YS&2rQ#a>f2r44D5v-DYRmbuCQ=)elJ~T@tX9)MGf6<(=tT^|=vPNf|=6
zE_k*_hcQf;6Wzk-8CWtjUOsSf?oX^kWoP_lkXNfPAWeEs71Wki4%|qrJ{k(P^7I9J
zqidJL4bn$|gx+s6K!NQpwxz~&U8>_4rdWAkPoML4BQLzx@zd?H_qI&ebek`kPU^R`
zyzcK7woZ7yFXzQm0UewwQRtrXCyNrI5uFS4W~>$p<*J$UqEcb{UG229b(Wx-(C8K>
z_cf++kHZJ6+dWVubl$LAkwU4~<^BU24s77vs-hR5jH+T^`opR8=XuZTGyCKV|0kF#
zLV^vVqPmH_QXi&f*do(tz<2;dq>%i}=Sw08P%u(RAm?FnEYfPFWn9TF<3pSl9-V%N
ztL9xnZc-T$0rzPS9bZa!AyH?x<_ekAy?k)BE1D*Sr*rVeFCr1&f4tWRR|<B0E*-#{
zmS&d2Ndo7r;Nz5_kADft<*Bj&vej0doPvwdX3y8(IX7>9Ojjdg2_U0*+B*2cH?Dm$
zw?CWbmD^SY1=EY%P@|FD9EjQaskXem@Aw```bI~;I9O9IXDLCw@!Q}Q_bVRZoS`;2
zuX$NL)^c4d9GTEp<vBsi9JUexQj)yN)$WX>$rjn99mUa&ww0ale`{@`O|A5KC*59W
z?Ku4SHmEq-_aSlaS1@@#dj0z0oPDLg!v6@S;U?$G#KRu_*M1Y-@uDR(i~9pOw)pqn
zi~Vuwm~6k_^VxB0i@e4a1sVE()G1pb4eK?kd90p+R=`D`>{rAS0vyLpCk9k2l-Ww?
z;bR`92*a!D!#r{jjYz3r<8H&q)Op`=59`xniBkSFtm)Ol8n|?VJ-gp@xx_uM*M4Ac
z2DmGhfuMFMW>0=W)3eF<R0L*=%)_)MU#ZicVh0{P*ESQ8@64YS^{lk?4~-?nsimt<
z?x=ug;PD;S&0vmasS{nr_)FoQP*l5M09zFyo|`c!$I?1jfR&mSw%3+3;a^tfK(<?V
za{A-!atYy?`IUD?;kzX21EIv{=RC`Jgq?9Mhix^E8>kx^b8M^odlH(Hh{INCQqhoU
zxb9pKc`WX*yGGxF{Z~q^g5w{$L@U*)F9D~iCZnT}6T;c0cgv$7_+<hX^`47r0>Tk@
za>^^jLstq$5bB{$f=qTdpmMiVEvVTrXpGn8vRIxSlJJ_F{;N~PHWHIpx+NEdNpLUM
zpldFlIc(#4p#p1_mVhpO`sW7H`An*#2ADlsdTc1EH8*|3xW5-9bVG3rV3srgqHjL6
z`9UYo0Bma{vo{wGKE2ZTYmpHJ)%4t{(U+_V^MPZvMe0lj1`<%Uzn9|q1#0<jtS`g+
zE8;CYC8?hFL`<j-mazVDVZwz?PDz1L=$poNCViq1sW^A^xki{IVN)P=yuB;@-!km-
zg)JT&Ax#anzymGPa_HXGu=KuzM{rRHdae{cpR2e)0ejQkSt&&F9D{fEDT;R(+{eRH
zd2k8!3j|jhX^&>7*O{*DM1RB6azlj+-;jcMi1|coky<a;XNOx?`J*aMx#ix>;bY$A
zL%Q<Q$ECr?PWN~sOXsLP7`h!;2--&W;=zO)FfRcxF%eu;h+nz=2Va3UJfi1c1qm=2
z2Rv*2&oHi5Niq0+!$kLyNs)Y|WFiWsVh<F9mG@#YV&9)bSot%4hC`*7hd$#uS^u<Y
zHevD|yt8<-|G0?iZ&szg?MWZim`s)U{`PmPrM|<%q<52wV14+~&&WWS0pQuE_W(<r
z0C;)C(^*|#lW^|E4SQlHEY6iJeagRJn0vUW_#)&ZUn=J2%FSLsdDe6D)Zz65&@(*|
z3NKG8E1R(5XUiE6m=*+DYq}=BbPJbl`rk+ptdt31B~~7bWpKp3uC#DHv`e2LzNId=
zW{{LZ(p_GSl$SBB-UpfGfhv9Sh&Df<pbBAiCx8WQl+Rk`Cu8O<-}bAtCn^eCVgFQR
z<&Mx?18x=tZXnB^8tXDr8zN2HZ<b5l2}d%iMDTJ%?zOHn6QloQT)yt+%zBSlK)T`|
zwn}?+|JX)hOJVmua47_m`XEG|>Mtbw_#mk^+;rnzZ1E`!oC-FM0HRYMpET2bQ`R6|
z>-Z=+Q8zx%)$i}1`V*|&AghcJ^(+^Dkh%x-PA%~plSZ>&$V`}VV_zO^3iN**Cf{0L
zyHXV&)z+FLEwr3~<7rF2Ca}dZUTR|DgNqTdrS`KsojJ2`Hwi6G4!DffKlX>H#9=|U
z(|~^XO@R(BLx~gs1h;Dob#<e$`wgOMux3RZ&Cv03tJcq-?@X9rNWxThvh8fI3>=@?
zP^5Hue$;b!MSO@x9FSAw<b1h{sm(vA*KY98LPQ?zgoUk-N=q8*7n%?!*k4~8fYT{$
zo9%p$De_8XpO+EpAnT#5ttlEkkcN$AE2I^=J*q1QJ%N`0nj_j?+pF}fSU#-<%jj_I
zXykFqq=H1X*?rb>YRdDb6t=0{_PPVy<0MzAu9p(68J$vq<p7z!@KXhwdoNEO!8QDI
zXl_JP5&)M2rF}{Pq^~VoQ^hnT#g4z+o!#L7HQqplNJP4=oWz99DVRC>F>IOA73gsQ
z^!1B&jn{Afs+8dqG@@qjRzZVBanOYWSyi}GmZvl8ad?WGmL%0--;n6XyJ^#Q!D(>S
zM+<S4<gosvR#;X}>0x9r+x6XYsWu?m3ZW72ntJ9eUUgQkW#^l~w8!9c@qn|gWFHJ`
zj(nd?AuZ$96^$mBLqvy!h@ImAm03054qG!%P+oSyr;ty|Nd9XVVWG}OkHkeyEM0c=
zq|T=C{`h9AUrv!$sFc`GW%cU5k<21M|7Cis@qYLwUy<Y`Q641d&FaoS_fC*L-|?a2
zi}_w(qZl9)=GIr!kek&;6=*~p$&*49b0F(I?zHF|=_);Q{gbBtd;$j&0A}zFe@CoV
zLd8B*p=V0>^b{|eqw_m(*|unBF*GlQeLZcTSh;tlI?RC3fKqhuGT#e}FTS^?77HO^
z&AfHLB^$M#Wo>`fxUmQ2z>^nd+0)wzRSN6T--JhuhR}$6wKX5P!tP+tdF0Bau<sm>
zTXpBC_BQGH-V02B6dPcl=(F^b-AH*DjrMzwCq|AN0&TOlLebMUDTrGZK>S1*?Ph&p
zog9p%B=}kmrZePkVPW^GQ;64zes*&b8EGBLn3QK{baCkx=5$rmx1fNf&|=zSyF>OT
z{POpkl2l$KX;Rsd`bRH|(2Ph#wNK<=6OO+h5^v#!g3<s^J63j7ZyshWRG+zB+@il*
zjdS<7PUa6G=r?Ce)#MQlq!qkosPm<m*Q>x%D+9t(oV#UYF4lWS0VJz6fn7*>)Xctt
zq3+>6M42>>%H2`q74=Y}gU{a-Z;z^B94Ofaq#-9Oe${g6eJ)*AT9pEqFSM+9puHep
za*z_Jq{M2O*5o|r9Q&YhiECwM)D}%ElL_**WbX2$i5{mYY&<a|VfgCn;aca(Y_$cR
z$sdc#rR|oHyVgt3DNpJ#2W&yJ_0w&gi?O?>l;9S%>A2_ETX@`KR8%0$!f+s{Zn54b
zvM5CHHwA|B^!23KXb^wLavNE?oUbj~lmuYEsZ^<6e0-%oMx^0T{P9cZZUChr7-g`p
zIO`5FrMkPIf`_`Jd8!?Sme~s~Du&xISWh4lYutXRdr=$mGykI~<NtxiMPp@_Yx~0|
zE&?~+mvn`K*zu@!dt?d*x0(R&k|rTDd5gdY`dXPIZuG9N=3e<P4D<kVp36fsla)@h
zBm9e!eKH0QsV|hB2J5&UZFF=@m~Tw+auSRAM3T`DR~3tw^^8dMOV_H>`5AN5LFyvY
z35NixWYSk<rzaNO`5yu(j!)FsC)_lnAf(DefndC7FowoD%KJ!L*xn$wV-#CU<k2at
zwQ=u@5*4jyZF2);faCEnWf3t<p`hn?T%bTk>*Qej>JJ2j$uFJ45sB!*@`gU%6>wmB
zhF=X{?&tEQ=(vZ5gh*LiU$?f+O{|(5IHqZrPKklsZ0v(>HVN5xQ~mHb8ETM#6WOc(
zOP!GaQU_%!$h*S-nBU&A%e!lH8RvrVK`b0-QC?dz_@Yf~!%HF-@Xq)+j=0W#AH~Im
z-HF>u*HGfZ(Zl(vL#M2)a*WbQ4ln5P)|~C0X@)cIg`#)hGZ{4|2-FmX??(o??A)<H
zX9(Hmsj}_g@e}9awmI)ug8PY^x^m|Sm{?9$IZtra6VahDdpA-FdTxZZhf|2&umN?h
z6QJCF&&!H*gIjeOq~<=-qj+?1#jmil^H+aK0Lk(tNx~Q@Yidz#^tyB=xq2xz^JYT-
z(zEHGD@760RG8#7{$@Ga8;<pAK5Yf*#eC_90&H<!pk0Os67j@!et!CcRSO^J*NG&|
z(>Ym;`j)Z;#g|U#Aqe?=0z|Kc5XweR^f>MZHI(Fzwj*({3n*xmF_UsKhC(kq>p824
zC_mrlXxl(1*^h47S(@J=ZC39FbAqDbmG)cePb|L*tgbgF=$Bq0HTzuxiU}eCetI$Y
ztwvNb_E0r*3*B*@-y_ox^g4EoG()OOEVlz(qVVIvxQ%;T`x{)~oQYlb<(k<{AK4QV
zq@sTfyEK<w{}3Xy5C(I!9A*})47sW<u($b2_ELQXeyOj1qk!;|;{44kjq#SJERx>a
zyyE-x0@%_~ibKV3IF2^o4*h7G^D)PgIe!R`cJcU0)WAyRIZBQImdQFjG2{P|bas{#
z#8PfFJ^<lQ@4k8{MHh=ue<*6|`5S5*OtZpa?|?>6uV6RN%qhE-ICIuOR+$Lv4YpxR
z=6tmVu>ySr`16EYcZJWTI-}I0Qyn;|ws1PZoLKqeCJrzayjfqlBKyx>i0rNwZFn9n
zJqpf}Y5*uuqGkQ1GP+)X6R+0r`yig?I2D0U7Rd853~REz{;Jc&7*V<DY|ZMcyF5Bf
zIH=@=B&pO<{iYHSsn9=yY9CHp%{pOS-Cb8I!S{V0PFWlXxq*$2Nrl(mrdyOynPh$%
z(G1UapH2K}WXJ^a-40L;OyGJT0^6lTNU`D{NWzk$|MwES-Jea<Ism+-&Q1XfcH<R7
z2``^V+ZwV03Ukbn)$lr^eat9xLu5hj^wg`AQWa`zVmxKx=yJ|T_O#H|`L5^?_nQ|3
z8`js(vxxPdIP$IPVsjwc9%fnS!)W{Uv7WA4)KnQ4E{JGX<9NQ>sIC~V7<HVV)KxkW
zI&g$`4>zUqH;kFUvjNXHo$JX1`#Kyh-#9^ueHxpm7!k4Pgj=>?l$*7f=1|(ls?=`1
z6?WVG8Qu!`o8Nh>%z||3YRi<Q^4QzQOf-I+;|oiG%m+%apvLb7?&bdty4m=ELU*2$
z@Dl6U9+>5Gm<?2Tb;s`dbsbvc+ediT;gt1Y8;^vu1K`sj)gPykDSx%jDm%%f(eTAX
zD(pkpEO$Poc8`4%C|c+Jk}+eITH9l6g=O;&iGk&g-pWm70*CHNgBxL(`IfAVBs8`J
z>{x?TSWIDv(aRxBpLtx*6=mPmEw{PH4F;`Gk4q3sAxmy@5T2N){1BIpUbH1;E?wc-
zoZ5K@oj}_Sqv5|6p1Vo>?+ZmvaQ&Jgc94n^gHTbzp8?r~d(L!vhasvm;jvnHJ8h=8
zF3Tyx>K76plKU|YbVWrZCFH+&R;q`+a7We_!WFEo{3P@ij^%KlTSR`>fc-v};<m)I
zuWRXLfB%inG-S@Nr2}ANcG)2Z2{76OdQ=xUY`0oVHRm~HR<a;JGH-u=z(Rsv-Llcq
zH_)`rC;=CZ2`9*77*1*S{%4*fSm1hR7~2R&fEcg6G*!24#tQFu)@T|BZ#W2mrZJp!
zPJLM2C1{SHlcL|bqrG(~3gGbYCSX*F;aOZs*%sv7Q)iNFvY7p4j!W%shxi$=J^i8N
zU%w=g1_QKoYKIsc>;U`2GnMtPZ&X;0=&nITrHG$glThe3qWmNCA6OOrhXhcjA_OwN
zx5o`wpXI;vk6O!;h$s;OMsB&{oe60SR*7bRMV>s#Zj%ZMOXdpsduI3em)odQ%3H>0
zMUeLn=)IJ7)PcAi0PO;j&yHFJ!I4{F$2|_WasZrpAsPq!d13)`GX&ZLb)3P6{VRF@
zd5+ps&<&EWD2#un{~jc0{8{7=i$W<Y!`pjDy(41!qo+3}OILmJJtxa$@!&2RMCPSa
zwHK5JhV=5SJymVf@t_CF8gzj`vtZuO*P9WRN9v9OgP4A9c-7?vsJYXy2pDYxH&O14
z&68`~WW3%gXuP#iFdHNM7TR+Uy;MKLX*-&~SS%7?p95JLA!?^L10`yOW9Vkz4G>KX
zRx?d`w&})@f2LF@XHYYxuTq26`Bo0;|I5=rf2Y%!_o%hdi2JRWcbrVyd42<eh&u4`
zv&ap`iHdZ`6xBY9*+9>VReR@T<KPqMomoz}CgV?mjf%8R0eGzX*K)GN;E^GZQW*X1
z#`Ru<SIAq`Rh0#-t{(Qi{_3HDmPwDts{uv$$-DYFkK!XlZVV2Ydhd#NrhPU(_~afR
zs$hpWFP>)#>6vRbC%b8HcP0Q~P0o9`J9bm2kO>p`fTqT9Ejzf1rv~j!!@N<DIlOl>
zcTL!PR8TA<R1$1a%Xl#~e<R2ZE`p16SQbu?gbFIvk6KD1!*yY9KZlW?R>I8-Aoj!_
zQaR*D7SN9j0eES#Rco#+jjp)c{({cKOTi@FTWvaHbdTuCqFrM%F@UZFsoq}xPSG_5
zCcqwK4maYRME>5MQYQi_48=sMeafUADY<a*%|FEcoYn7oiiwH&*Xcx?#d1+pRP-%9
zJ$-%GyMF7d{i#3VV*hj|99CMuh=!+s{ba>(hqonD3;$g(aod!liu<_9r%yUHZ0O$&
z^u%A*Hn8vScL409`28j#Mg&Syk0w9h`8F1##j|hg>7uR?;4zvWR%O3W|8MaU;$z;m
z@G2$FET?77{E`)MTzU?2BZ@o6n1_)3J-USR9HMuz!aH6hbt3N%=YZc@)FpCf*o{r&
zL-_rrT^BUjsW!a%Zn+`gDb0ZsI3u@{4GkU~&?m8BG**6dVVH{wS`PA);_MkQpebd4
zAMCwH`23cqHn6=NN|8d1F!S$XExx$?BXz!c&H=K9wuk0-u`h8>DY+wLf)XjZ=DJp0
z!4jlpR>QDiH1qszf3Q_Cppm*)E&e=w{79~&vI>JP4ioBMnYH6T^4&<m3!idsODoSF
z8F1w|FiMK^hgzy@h-d>}DwwL<RHpw|=NL*PCz++1bpkuXDS?531wp6(!8ujcdAjYa
zyZ7$8pBH9mRzZ7a=QmAi8dS*JH%!@vJ&>2U47g+}&@DlpNuX~TEX!S%Q;h1;gpsqS
zgF4q^>^huxUHagv%Ve&c7g&M~`S`IcXSp0rpV$F8aUq7XMvvNJZRBf-4*Zz8GvW)Z
zJ3By>7iOWqGHsa5Rhv*bE0Nu$4aH|@Ai0r#T)EE>ZwW2D%e;ZJ+CP_74~M}`NJ*S;
zF4TSZMelRTzL9ybxXs?6?vnR*H3q`w_xGNd**Y?$+$=D7PX6-#g+NH?E{6>rwa#vP
zAMcHg;f<*tt?)i3PvcOtA7fPmOYU3+fi==;zN)9vs`Bm;vMt5N75pN6opLuHI5jfe
zKY#xG_U&7Uk8Eu5!l9ecT|G)USk4pU<0jKZIBIHYUBFDEi(MkCRaim6Go$HiOZfCW
zG*LrCOzJ)m@NOjvNQ)B`*LyFX0}sv=Z=-|QuKDT6PsF5*2Au}y7q)@nq2E%{g0q6j
zSgLcdYQ*Jb^I;itN~rmTYInmJ$tVR+q(C>Y)WUVH>B2Hde*Q8}h)M+Zw2#ArxcIkt
zOq=N!vTp~3wMy+SPRza+IN4viMZIcJmF}7AQ>BCj)Dr-h<<HY4j`oDS5yKmB^7%9$
z8F>(;GGHl<HA03Yv>Qt~dIKb;xUGv<>+I+tiD7Wg3NU2E-+p=nfkmrrF!e|2aJJmr
zZf7_$I@&uWh2Z)5naO-Q#PMi;tlq)gY%<>~Gx5@)R9w9nvr1j-n(wT)@Dl1|vCb|y
zxWUuY6Q2XHf6H29dtHM1np(ARXbN_k+)sbpc`1nX@sqj1H4eqf0%@pl*1?v!h4ZH?
z8q)p+dW3<hI*_M~&m&131Pw_l7Fvy@vXuDfB!@Wk{7y^LI8)7x=V;l*RcepIt2t?C
zcjw0BXoQO?|4^M8hB30iL;jURIvRd?1-?~P4B*$k8%_cN{7YYN?`nc^s_wyb!_23E
z&`=T<mRLc5Bzogve27?Do#D|8!4s!M0WP{2?bWA;la^qjQuQkFpoV`nQA8di4ie17
z5GYymW2)Mm{52DJpIm--9s`V@q4c#$0ur1{0OkwDd$I#p#H0j`<(;hB=%$UM8^EbA
zhE9L;ntpF$;Fe!hI$#?#u_hxf3kq1gQNt;i<*Z)=MX3r$50mL!RwGPhjAa?n)12X7
zgKF!HP+n6KotpSFCx$@Ua+%3uEGCxV0na`xs8mj#L^ErFyQDaoV)l6LlkS#<#CjF>
zQeDNL)=~*}-w1_(s*w(bMNEMJVE-#GZ(4f#M;fjAOOUZTy1Tnys-K8+-BUERx^4Bx
ztoI<|u@oKtD}5AX0!nkYaRszB0a9>H&cyuq0VZiZeYM4^-)@zvjGJF&(W4vMuhzQ&
zb$Cymk*(oj@kalawKUPI9&YE#3d2SQM!hyZu)i1E>Eve+nR1{!DxiN0s`B7Q^r{4B
zLXMTz=6F-vu~^3^CzM;Ji7TObq$#vKFrn?CZ)89h#ILW*q`3rBCuBBgxSE%t#{{c!
z{Zn&r+oS;E?17MjBORTaRd-k}Ew6Q#RM-+DT?@;gVv#%b-&_{S(2I30mr=)yce^uj
z_o+eJrI1q>9RNs7oK+tmFGjk$wDk4$o1gE^VG(hC{QcpCg@x-e73bzUySi+)2XVlG
z6OfRw_CktqMh7NwX<QtZuC6XbzGBR&(zjOx8Q`zkfvLq_X$NyCHi==Dr#(I7$*dq?
zg=J!h%skj84r2;qO^pCLT$;bA;uk!;Omft}Km#eM&SzePGE3b*$}7vs4($INv*Cvf
zV0~pmWI!+({R6j+IUSQ^0F#V7*E4wI9}$_DP5Jl6+FRLi)etF`Ld)yf3=?1c(*YlY
zgr{?wH>V2SU6KP^z4Lxk8o2z;mBH@~rOe~Y_$eJWIt&R#)jEy+7yc6iTv)DbEn{SH
z;Wl7naAQX5Q4w1oIv<PMpc_mRpQsk$#DKv7-&&t+^$?!Rm+JR3^-3;5vG0CxO8TfI
zl>MxSa5m|!Ozm&p(W4o2@D)mOMlgz8L#f;iw`YCPWG9J|`LbWj{{C(6?}rBf02Z0h
zy+r|i-K)?iSJTOLzaTG&$k;&9TG<~30L!P7`1|fb+8(^K2HN*^7cIHULNqE<czyHz
zAphgs`p?O62_>Vf>=2kN(?hpSk%(5QzIHg+I5-pLy$8sb&WaSo^;d2<@KvWh#pf_%
znUvblOK{Edrgch_uY1JJ(Bt}bOCl`QdGzH=Urfh(n1P~Dd*5XmX}TQDK`Y?RRS~#n
zj7QeCd}`?$>W{$2>Vu=L0f980YDc^9sT}q0VScdHkL(2<2~Ot*QmP+N+{t1RVLKaR
zTA({1Z}Jv2k%Nj)qb2OIj@%X}j<5QymXk*k*HVc)J)uBz@xk|<y|;PeYxb^pzrJm^
z4u6v@9?V}TRNEr`>Vpy<3H<s6)!xwoX)=~6<l9V6h>IH(82FBql=Pd>-t&(}`#n@N
zGU?8l87v723H2JwA!Oc1*~aSMJbx=H)T%9T1pVH(EiDn)+1c$+7Z1eK>$BOc`z+ho
z+S;nt*^+?ms|);51qB5zF0S2)JY?SITYa117t9YI<lH5Vhm#A&0j?=CX7ia6!`;!0
z+q*kmGqa%LVrr=*7Q@MWIg8n{w)6{93Kejs^@QLkJUTB_|A-;beuY`<85?^VaU6*1
z-ol0E5_Ii&JZ~W$7>t@-6WT6aM<;L?91xhV+-;diw@hGU=lmMaKW)dd)U%+8$iGJ}
zV?r%4T2raPIjRSE74F`}3LX}l4<=qRCALFhZy=yOw&sSY74mGKN`17q|3dzZ$+%1)
z>BZ49DK&XlCkYs%kO$J&3%nHZozi4s{4Az#JU}kJlT9Bso_>(|K;%&|DV0iaRZ#My
zb6<7I!c>qt;Aa57S>@Yw<-DG}tHl1<02(x_k>~q++-zK6wuemydZL3bD`zR6Cw_rv
zHgyiX-$AE9@uE`Az|34MO9eWuUya_%Sz{)uW#1t|q`A>>(oD?7l?*zj!w5LLYi%}8
zPfu4L?=B3x7i(>Zhlht7Z`Xs1)GGXDXH{+O?CRX_7%sMlh`)XffP@C`8AU6YC0`xP
zY;10J0*oVE9Xi01kh3$}-Q8VhXJ?tif%3BF!w=AfeN2)5{1E?#VI)2t8=R$W(K0-J
z;P-Zg;<1AEUovj)6oowL^6za>u<x<?O0}E%xr{T-6kWY$78a5m4yGaL%qG90qM}l`
zVZ9D1uzQoS(k^+EX|A%%$nMr<L_&2x0Q|&w5Qq4MJy6vwj?#_nhF+GW$Cq?$xyF$y
z2#*Aa1DSLcWd@7J60%<@pQugiUU$iJIHsrBobMS{x$r30Jb--tE>P>3tKN>2_qV!#
z5CMUk7oWM-*tF<K62DhT+A)L*!-PZeC-IHl^{!np-aQ`G+s8l8LfX`2mzY3NVB6sI
zFUSdzU88hfVccuv&t)<Q0VCeXwETK|LJxWgFsFNlZwE4kgCXWa4%Mmx;lX|(U7$Fg
zErtexM8NS4{K)Bi+RIDBW;Rm{9LdYeo2WEZ(9_cs$FDFM`+R?MBID@jSnx1krVG7P
z?;w~pws>(@XSc&(z1qsK0sK}Gx|lSpvOif)nVp@b=#x<lvu#NV3k)PbYXQe%u2hn+
znHdGM+2m>u5wN-575q&^B(7e|lotQLp0Bidg@C_YC(eps_dEQ#Yw5EvJhl$nZ}@4N
z6d^}t_7)cHzWFO>KV2&+|3!h@O@>GNip#mM5A5~pGo>G}lYsd1V--N5qovJbpoU2o
zOWNA~#nqZ+Y^aRocOWWwW$8O;W{}<4S+-LJz4e7aQb_|`_snEV?SmDuN9%Gt=z_4l
z=--p~cf_Onm8X68_AmVRa}rgQl7l{tsDruFHKE%L%<gScF76#fe0R!G|7+bGSvu?O
zfk~4O5V~q~z&4dd=uA~{Ef%%)N3|Jok}rs4fK%$T&3lTCo-C_!a3z&9!u!&AYs1F}
z5lt%f`OBAV0hloyMx%83g4?SZ9Wf~>p{%ipxHu7OYik?4w^iyx0UwBu>pT<{(My}}
zDImb$Q>p#++1=e;pKAl}R3vk`M1f!f!g{m!gK`O=>7tU1$7*>xtzm`n;&Qs$mdxXk
z2s-kaa1mi(V7`!$l)K$nrn()odxuvroQH&lZngU&KrmU%My#%WXJB9$HDT(aw_I-Q
zJ85~5&JqrNKP>lpf*Yr2XXJM9Y<!lMG*&H7)yp<j^Qg;A{~3z1k&ZMo+`<`qrD^~Z
z6&L3OK{z<uYZ9CNUY2=Bpx!o`Di8FV_<&A)yB7D*Mu+k;cEZH@eItBM+yi*MvpZ`B
zjA7%Sn8}(=cdG}?pq1TubPtQO_Q`6=`_pl~?`c1AE2{cjq$ibEMVp(+boGdqpfu3}
zzme--OL}IV;Pay%$#>pMx^i#)9f6Y88ppk0=O(&s3DC%NYuKHXuhQLm&t_tockpyz
z^4^TPbBo`aLUZs<E67Mjo-avh94>X3j^*?%Ec?qyPRd;N83a@aedcByJhC(vywgZ#
zarc6@K7fBlz!0>3@E^qIZ6(R#0h5N;e5HMQxG{dbJ#&9~x&d#q7!3o-eS-;%kk(F4
zc_QJ2m%C%$C0Y&ZYik)!i=(5XoUZqY@$sV|g^^C>5(~zpX0zLV56%E$2GFZ<!pZxX
z!e9swaeaNQ+2TpS$=Se}%F4)?rB-2dxL6m><8fc+@!-hwbg7W)aXrV`EtSYraC>$*
zSCK%c+fFc2s7Qg258&^P=SbLmm_~_-+-X@5DAK4w;^l2=(7V08UESE|2&Uzk_5Dd?
zJhky){jZ6ImyLF4I3Ul@t5C^u!a)cGtt+7sxFT-tIOtxwY)eThZPz;@BDCuY$Cvo;
zMTvqsd`c5#$8u_0cp<G);X2LZ11`HFmF&WJshp_c5(04Ku;7nBpE+}C39YTWTwi#J
z{fJ8%-@axLPgG)FdN#zoq-Tc3Xk&f78WXs3*1bJKc(88yAtvy{Q;Tch2S6s%D5*X|
zKK%h@qX&=fh~m*2db&%%f{-xO-Nkc_E%o6}iu@S<(>Icb<AxJ9m-8>&Zp(A`;C%PX
zk%(ikm<8y#xViUc%ZWiOY1oZNulFqwp&ADd@8Y1e1@pz_;rhslt5msAt=^s-0&M+P
zT0#?m0!57bc^?^sX0!WUn4YP&HX)^Q5k$IJG-(JCZ}VUyqMVwGtD9Sv{2u97d|uB+
zc{}u&yhTQ5WC9Mw1ee!IAi(u^J=5x^yxcQ@lO*(LztAJHlJ8SQ5eMGJ$Xb4h{IJoo
z25!9COH+tR=wXS8^}!8M!BRR#)jph3l4rw9?v$CsOs^)4C5A?8PLa9d<IOXMrt`oA
zen?Dx<JBhE=i|e-8PAtqHVy-q!24V24Z@_kY;10_dAfLy2P}dail}DwWqNnqdO%~+
z9izn_rvXW>-P0%xs%3XnN^)RAF^n{`X8s%tibgoRC)-DmX)6vTGV6n>2VUT@J>L|Q
zPG)QKf`*%|GFP$TKX3Jdp8TT}Os!Ep+Te78s*gy-(>UmX@$TI_a!Wgi`b=B7w0YKL
zXYX?Tp5?Kc>Rp>L)=2~egy1v@KvJ@ET>ggeBL_$Q?FpG={Ljp&hb0x|)ZE<M$LEI=
zu9wG)_XHfj!5$KerQ;qsU95qIg7PyfOFw6ux<79bW;vExL)nP%zuf#%tSsSb_E{AV
zXW8wh#Ud0^RHC3ccosJ)eys9`Rbn;uUEH62;eYp}9J|0P(T+KQPS)k*w<YK#_-i92
zh-zUl4)Ut1KgDwZE|3c+)|^OOTzgQ)cIfwx_xbrjYeYvUc`Ii<_|CRx{vuig&yp@6
zCI%yg!yyxd0*zpjfg(VAd;2YzZSkyDgb+{nSHzz_LApJ-0s6VZp}0B6f(44GC@6fY
zs@OANsRJ8zKq*m<a^aPkj`|ng7a-WNyQ>R>w?w0+WSRa_NzDc7_3=ri66wsRsbXog
z#KAUyW)~I~CK*rX6-g|zzQ3=2)+tmcIs<(AR+Te`jtZjb+g-fuJW@TwAK}p&h-OpB
z?X%gap6{@b5yALuUR`|i)gfgjM^a!!zz4Lr3Z~LAFk@MeIht(yh9x)p>-0g*B04T9
zG$o5tyZP$$k6NibZxc8K?XDheQ1ys=C1tw7m6}R;B50_j%l~-E78G3H)?7RgfYc%m
z+O<Ic7?b5xC~2{4CewG!Kd(wbf7oCV`jUVoMqOh<eOGyr#rqf67Uit9?ifI!bGcqv
z-Ii-xnFXXxat-{Hl0$^UP)gu2g3~x68MJ+P;ey%>c>B0@iL*7Gt((*c)C@C+si@qn
z)hubh8yR|>m>ZaSJRC%-`K~Iqc^G2qS?W@kzAh8OnSBNLi`A=cL3na{d1=u9n;adK
zFZH_Mg~JK8)>-a5fu=ZM#J)HHDCJk0t8mIVKc|u})#mnmIAX=5(G)c>_<}{F`TbzJ
zm`c4efLtM$&afXHlU93Bg0@Mq-~r>l;L^z(JbCKoz-AC~QY;9+J=+xz2{xF@EPZw8
zfu34$k7}eVB-;bY^|X)PXg2KzffMvV0&W*UYrD}1&`t8C(`V(XvA;K7n7ZK{Tq2o_
zHeNi*l;MN7S%(ry7}?$QSbbkcmOcDgm7b9C0b_>~AqfrmO1?S66E%bUkLe#jeoX)R
zHK-`9R`y$0pr;VZ2T1;irf&wcbHA00G}1@M`ELltq931YW%=frq~yO^YmZFq)JY9<
z-`I)F1DnxyydrKdlGoZwg%ekEA1*yjf2s8y+{TM4)82+;J+OeL8i$pOYw428qkArB
z@LsG;8box>MeW$0eUaKNds3{2jHTnf$B5Pah6d=fMND*BOTm$_alB&qqa$--w)riW
zUjS0Eik`bQ-kmQodr(zvudhh!k>sI<hK!gQIBVWfQf6WLu@H?V>><0izaIOobjHqN
zMVN&;sjG_-ZfwP&aP3F)l();DlR_@;Qz=DdLn5mLNk0(`09B7xFcEf9F&EDH0pBIg
zJ&dcVaLE+wyKu_10W-SES-$CG)yKELKE4fa_&_X-E<@$G+Ug~b)(DE$AiK(%^t9&y
z2;}7CAa0~6wznL1o>rPY9(HHTv&71mx#yWbe{NrQ>erwcev$u*u2`y_vOiULaXBG#
z(SgkST8+7q3GE=(di{YmeM-gm!DP-YU=pnBu5^={OR0&V`1FKpml|)3TdDD!5Hsxy
z7}&z_v}svinfru#!|gY|XF#O^BCJi3!BFh198vCa-xBz>JjGx)U1D?mHRiVC#q%KR
z(-|QAMbJ6~4__G}8*ptvnulbyF}e1ma(^+dhDY1p*j|$abSiX*r-iKEKDT7U;r|{T
zBycZkQwRS8MQyy`_?CH5{L4F_^ghsL!ZXXThi7X(OghhSE*W{Y>{0`;=b-+&HM%Gy
z{)L_5Q{mgpi(bT$?>&Anhj(u8pmm-Cu!Ng+$j!5OM9WoZPuq6GoJ!cqNSGX8t?<gC
z!~$C`=Fo+9Z0>s<h5wJHtBk6u>$)IFh=72AG)TA7T}pRLcXu~OK|nydTLh#-8tG0&
zy1To(zKzfOjo}Z5+;O@0oW1v&YpyxxIXhg+;cWi)(0>p=<+h?W_a&SUReU(JIbutg
zSh9I+L0OD_q3(#&_hz5y*|B4GRQnlWW_2T^B1Wo?;&5dyXX@a5fDpbvCuE)x)o{K~
zSX4y60k4=Z=`e7u{{~m?j-G$<yWbC6g--lf775k%E9>PF9m6B^n5o;1hs47vmTmSe
z1=EEvP1hLA2Jw;11c%@o%s7W|{Hooq64}sfSE7NuDNie2@B-FGbm159!tMhvvF9K>
zp{Rreb<28_{`jGM<y@r<baZrPUEAU8Uj9MQ5MnbwTy11fQOo)G9LShnH$U7(B<Onf
zekb=}#5U7db>s&0z~iw31A6sO$XHY=J=LFwgTH<uMT+y2%U$c17b~;tC!vWQXWKnv
zLiLhqonE^4+e>aEWZFxAcNw1qS!;5HS}B^jAJm6UP9}33geh1Q-Q3Z~|4FC2!`y~#
zzsq^Q31M#;*wHbBuWinwcIY;?P&-wt|H8=UT**qhH+7@+Q0c@MyM9OQrVx9Gz|;g2
zKTGQ6_DoN1yT2dch`P*}uhZeb-efT6o^Fh07VGDJ{B-|!P%D{hyV)@W(wBWAYK8UT
z{zP0m^qJYRS-%3!?^kt~NW32QC>iqJXld0*-Zt4zqw})KulAY^kC!e!I6wUO>3ptI
zBA#Z@X4a2-zCd;Ql!M_8hl4_d#EuqELpR@YGR;rPMD&EkGI#sT6URp|syY6Fbno2(
z-J1^F+2Vb>qVd*h;}4$xI#3FZ!``umKzbv#_(mVQL044|tM|1l^RO~zHZy~ms{eQi
zWkN%3GGeHqBsYG)xX~{b+i5#zkFG;wIvbG;vzcj+GA3A0_jvF9{%8dx#No2)D(Y;P
z3(OR7q@d7SX^na*FPR7)vSh9*MCpcLnU9=Z9hWlyI~4LkV+A$1*8#koQ+)$^sU$83
z!{92r(2x-KvoZBh=wSYC{$Y!QgX1|NSEOp8+FQGMzQE^rB0=>wC_DczCO|WRl5X|J
z$H$+)dh>CtP#p^gr;EAywhfR$eRJ{_4A>ihk?Za2djtqMRNoHdeo#}*L*0oAM|7fo
z7tr~pPTY3ytpY956NWVYI3KP1mie)WzYYZU=rng3Kl`RX1vbv-c%H!Srp^T&8r3J>
z^&(M0bp;}pxX9?^_fWiKua4EzTfJ{f7`9S)&L<zonckeR9Cedlc^i^6-R_ZcxT2vm
znZq5QlwaMZ%_}+9#L+k4Qc&tiD~}=kK!Ah^)K3?54ZucnxH!)8xxhuW6)&A&-KZE+
za-0~%5h#Z9MX;51Q)okiHaF2o#O=*DEVbtvJzV-CsP?zA=-=)?)05ezr*z}})xIiJ
z_V!J3T{5G$3V|y{`-S>c^O=0Zo#k8ifm0>*M*s`lq>JG2DtnC<cpW!g{oWJOpLd}d
z9r8A&E>T%9FS}Y=sID-&va}dI7q24g5UN9`uZxAT-9Q*2uXQB*YC%$Iw)x#Hk{(sN
z5M9Am3a44__G?kBvwI&_*F@wpldLRnq9Wq`uVCAKc2kHa`V@a@)Z9q_NsO)MZOT&&
zuVt1Z`Oh2u+{2pa1C0oxo|<9Mmel((RmvKtKpbYb<;C0L!xNW8wfb9^qzMWkZP+m_
z#H@Y7R$2{XZr&H4#k(sDO=r&zKd!EAWGUq<(9(ByTV;NaEp4dRe}ox^EgxYkEyqFt
zZ=tQD^Rv2Ij4CE`VyCyacL_9EG8qDC%@6nW?v0?z;P2KrtaLHxLJf(aPt&)vi%=_k
z?5w}yvxk5q1?B}lzLZeQoxk`aA+A;&eLvy0I~nCR?-BIYs7s;I)1|05huN-5lM$H3
z-;0VqHtDfalrdCuy!J-x14%`?jrbn{+CBzd`SDT^3Fga}pQUm{N+)8{(tb}>m}%77
zlL0c4{2V&(Lbon|p9*GW&RfF)roWSqdZ{!1g5lBZ++0+zF@Q^$7#OWqrH!tG4%oyE
zLB7nUp4VyDu2q&}o?oJn{0_@fxqtbiN;h}*jVGnY0BQ36{O_R(s>UC?t*!hY-_?(o
z8Qqm)6Xq)nMPnIVT67hX#<W&yb}cmRe*3)T0)30Zn#gLqy6b)@oD}Le7%B?x(B)^I
z+kTL=Y44q@cX13txBT=R+{}>CY%c%b7q%LGV9|dH!uHl0w0FF)W#3!e@m$PaK8UM0
zh>djjlksY#pMH&6=@1uQm$|w>mwx>DAx2=Y_}&J=VFHzUO040eSmj#37j@DN@;Oq1
zwmu1FyAT%E{#^>oMdgrhc-8->bmF5FnjSoX9QY=(PxZzF@Ljce9fRSTZ;iaaV(*q-
zK-BbaiGI0DAiKW<AMvc@iQR-y>DbTnw5ylvs|oXJR9W4K%-<Dg7bjljtN8ysZSmW&
zXCqRF<fj`9Yd?$CPk1fu?w}!#=sOfal}$g<ezbmn^-@jhSF@nPwHNKbvkBb0@V5YK
z^0@9wfGOY4swz>Y1oD{P$o@4hZtkISlYX!0dO){)*KPF7Ip;QLgCDhIe`KRuTVVk>
zQp(R(%o6(?5CE%GJS7^66IbgC_@hu<W~TeQAQE3NE`3EnKw!+G0XaY=B_);CQ_-OK
zVUmzQdV-%mee%ATbL(to)@w@jJli%fF;Q)R0I6DOw~!uyf*XJFSA#+BBUq)N*|P)f
zb7(|!SQsiGMb|fnvS7G)c;b^j;L-(ZAkX(F03$!P%i!Q##i{7~_p=wbdl_H8AfXX*
z*$?h|J`kNVK}c;Y@}(n3HxWvAeH`t+UHMXbj>UE>o07Q`_keu6m8W5#Y5Wtda8-P{
zX2iDr<kcb^W1oh!I;}st)u*)Ry`A-#7g(xqh2z`9HbZx3D;1=Kn$-(eqn`&>x(3_}
z<e?mYxU$;HddI#k0(9$9r~UQ9YZ;_{-S|{eSY!-$&=!Bb>j9SGLcE}a+*erjIOpBI
z6#AWAG<h$ubFc3GTM<&!P^?=nx0<D!y@{Vn+kITWM{e7XDh$w2*WW^o<)B0OO%PNZ
zx-_&^-=q<$n;WEln<>XFUOq#YjnVuJFKFG4l9{pIdq@4_PB(e*EEZE7XVQm?+l{;g
zT<0TFh?y~^y?O7=WTj>MHs8&Z{ogGEO^TwBgg&xY(W>xz?_0Owi*{{Zo%VAT5j~%H
zQ|(Z0^tXXiaZ0l_!{PAEB;PuOOw==fFE{F2kN~UrqK>*u8P($odSG|iR#_j+uq*rh
z85!#moB>ik^Fu>X;I=Iu=v_ZdC+PVmrSU>;pn<e>EP?_5zY6MMQCa=;@30Cmxfn>{
zMCaw@_1^<TI2U-Yq`l7}25NpaV8M-+=-i}}%;~V=AtE9!Jl<5m1@bs=w&+1~c(kYM
zDR#+N;Q)1i`{E@2C7dfD1wvxieAV={kxloLP2PAmGq64r&egZJVLXG-rF7+}ZmMr#
z&!t|ej8)2Ms~q}q@}$M@?VvX2d%dVGzIoaxM@?CjCfi`+^o4nnZnEOa?f67Wt~!8d
zQbx+~YwSxq>s9mL50)zkt!Uuv7`(!|k=a?nsNAm6im69CXZLkJgHTW4;=a;j<YxB{
zJU5@Qfl+SKWDHx?@q~lK(mquU@11vMaaDQNM$?C#68S!>rd(Ov>Nuo~IkeEPZF2UM
zVNt*5XC8CeTd|qGznl_cW*+qj_-w~zs7Zq3d>g?(+@>H{#&J}<BLPMBBWX3^HjdjX
zH99<0pWb=Ce3nCq)6r6BmOb5z9N_jxF^c%LlYv2zbNG4A^6CoI^<_=@u*hU~p_H!m
zzTvj}C1InszWTH0f{bD7jj8lHaHKZOO+p1VH+>c#UWN`!i9Q^NkCe&NkAw$2wl;i*
zVAjOtxQbL^HfY)tNycU|43CS83xFwrEu*D+8Lj?^70z2q!y_X}+)h-_0+udZSdzXT
zt#mtWkJ5I9Lqwqw5$GTz;qm0(t69Lr#WhpO|Md5-0wxyLRJjRCdk{J`4GmI&$R{8#
zfQ5zSI~id0zB`*B<aPZDU{3j{FBsRp2NXRDE>pIAhJd@fJ4kE*^O$d1HMWoE$Rl50
z$mgH4tU~j$;r@CDWcoZd;MZq6L4}3%kYinHGP@OvanB1d;Ozm22-5TCY_tNc-f`b0
z6zuKapx`i&(n!3l!uU-6dvB)IfBN}5D%LyR+;1H1U%0D2t=w=9{oHsV9YD<|M2&7E
zTFB@(PCWcYT{*xLs1FEdYJ!20-lz^nRzZ88bhQ0cjxg@;aKLw;{8WMR#t!5-h@qSa
z@9%FcrZs!(2P$9|w)E7CK+;i+QX$X?blz#spDq3t%u3k52<nT6@2VF<N)4(HY%{)#
z#%;TCzUeXZL=H*7^Ihyn9cSt&nwX(lNb8r}HZ|pw{ugyOUn~Fi&Q%7_+M=9&{LL9W
z>5@F2^*n(fPeM}`L;XA><~%10ilWq#TDCzwBkJW%=?r5Xy5wZPUCL5y=5QZ7GU-kz
z3>NT~EGXu^&<x&?_z^i`tRViYVFcI9t=lm`wMx-N%<eLBu7bnfbWZrRe?{LT)mfH@
zE(c?FnwR{GRdr#vvNECqWo;^Z>Qm72EMb`1*~u5G6`PM0V6SwARhTb8WMpImX}k$w
zAd@^B1gNY=&$I99rMjk_%bS}$JyGO%+>Y5pJMi%E-zEsucY0^JY^<%_k2-NZjzmWg
zOp&qsi{KBoMsg48Bk{rfhcDy;(Ahb@R<C1v%PK0`!E98#`?wJg2Y%_u;cQqEroq2H
zaV$dKenRUmjfIzL30nFRUhA?ilKOk6ryIZE0~QIEB#2iV=Rn`U$!wTqu<g=`qmn%-
z=P0e9`cuXnSC}9dm$+Cp`LkZNkez3_(&LWrb%|J7l-LKvsT6^>>*^P@tzJu`%_dat
zumxj9gOwx?z=!%ad{?)7D~ne`_Ls1$8r^w1Q|)CP)@VN28RW^{-WYg9?tsPAF&om|
z5j9&g<^p5I9b9wcCBb5wt&pe;gG!iljIoa7;8&qRq5FlFwH0Mirxd^4H$!4KzfgPH
zH+u6OZm*O~E-5dG7c;K*9Y4@;k)3(6cUL9n_8M@z^Ng6S_g6xVxVZPoE64X0+1=TZ
zDe+-p6>xI%P|2vZR>C5&>=8=7#jz{=V`B<<g9MEXQz^O+#R@bbSPfzL>@2xn!Y|Tc
zz_|%CS)a^_-`n@+X-<`o1Jss9&3L8u7Lunw;XNtyO9{x%rC8j%u(!pMsyV$E%Aau4
zbZlTN|2sDDOD8=L&^C1rXMUB72A`aqZzgl%51`4hf`w?!GcA@~#x1p4<v%C`&&W4S
zyZDb1WtiYNu>+8wnjeoT6;B3{*1JOkx6$w+5t4F_KaO^Kj4wG+8RU>~EkYG=3w~Br
zh^p-ywglFxUVOis%(W04Kcw&kb?EWmp&kzNkp11Yi?kapQk#*CX+CHSJh#N%jj<s;
zlz6VN8oM-(mFwXh=k?!m52P2}hNa~F&L??!))T3sDlGy@V*zr^0mj=7wDTm=@J7D4
z=+U~l;#8%VJH5M36#xDNDW5Zr{NeTG<%eI=!Y_!8nD*yiV9VKRGH`_dSXeIjGx$v+
zEhxdw21SgTPojjl<{rv^nY^G80$tAwgheK#bY`X%ayQvwsWYWfkh5bLvuMv2y!cRq
z$H;q@X;T9`IgO{Tw)y8iub=4N$CfB&_`Lgfdr;A-vBbp36ud0=n0{!M*rbVlTMDHY
zr*_?zg!dxIYnQOWHf~Eo@f|8_U(Dud<d4?rp=#|cx07Kfu7?YU*KQe0>rf8C>JJlg
z^&W>q=Pu@KDi;v~tl36dDXWzd<?i~6Zta@uH$UQ!+9SQW(@&rUx*>c1q}u^)0152R
z9kRq`^!@5Nb3FT}O{oo?l*_p;E*!qQW{}a+rO%Esyrj!<4o`;1oMrsXBOmZhK{X_w
z=f4F8PyWTt78DN^VdyN`39vVqOnY7Y-|~C!I`n3r3hB?8#y^vFZwVQk%V*eJGVv5Y
zdK#(n#qLbMsK}<@x@2N&vV5CbjVV#sRje<KIyLrV`e-&mHr3qXWK$SJ5`P@7xO~<G
ziXCYTLHf(Tf98=%9!M=POH%?AI=^ZS$eY0`u-2|lNPs`14TU}n8cF&)_O-W7PO_#1
zF}^&lTlTkRSXBSo=Jyj)Ww<}46y5UD-6g&iayTJ>*i;GNUo_p9y_82`#Z1C=6e@1d
zwfLy`yDIrL+TE`(|C5^Ue;6Z)V`2Z$Rf>ADtSI$<m0r7D=yoXC7MlpZGoQ>=P3n~b
zNW6&1hPELV?h4X8**={A)zrA7&e|C9EYgS7KTWT6X;-g%|Er3c6eY`Oon<t0cMQ1v
zM$)dz#U6~F-iNORP`~mryV2AAJv<nMoQ#Li*sf|Y*K7VLDn_M$C7?B)l~|h96L&bf
zpg4U<&()Pb5>$QM*YHb<q?iql*PW5))bw{7bFDYzj(k=x|DK99m5u)Qnt!wOoy)FG
z@8M@enrvD0mr~f%?OVgo&%&whL#Lyjzj9+-T{Asu7k|gy^uSCz1VSTzc?gkHmo>RC
z;%qO<;Ot6JS?RGf``+O!b-TX3AE7h}U$+^`<V=Z&(0P>?DfLFqsVk7eO}~|wQ)Q1S
zQ*u~9^O>CW9E^T@y+5ek&szRyb(biUj9j~S^u$Sd7mQKuPg73DJ-L$wzV`3&(LpSD
zjk@mb<-+mGKz206V+5pD0SR}leU;7mHG1Mz_3``}BDR(sh26g?sdmd%>T41boQcNW
zVlm@jJT$qIkhNlZ_~N^+dz&5h%;DvsV@&8-;jrG0V~*c9IaIFZE3RGHxN8nkZ`jnq
z*7-S%1u=G+ovbGv=?!fKoEy<8(Ud-|e<uj0>qI`LopMS>IQtMu2Au_Wd=~v_zbKP_
zXU!FMQ^W4bJH^TRf%}|#c~BQIhtN|!>VEzy?<}`Cx9BkUboBe{UgPlw_K5iQ`r$k1
z4mXNkrXl?C?qp(GOjpu+Dvry$%V_Ym@>);62?+`5s|uH)YM+`)YH1O|=dhLl<<s6F
zKec8M*-u}i3cvDoo~{#<9@5HolnHz_p4z^Et=OHk^`+@-<E{Bvo!!&-w0uae=W@<I
z=O(V!A7J`d7SE0&6v$=ab}$ubRH0LE-><DdMDf}6WHx_a#oS~Tpn88m@!LJHGiMBn
zXBr;-Sa7I8A%0=2zuvKNd6{bi0tq6ncaPd$ag<(S%x$~7ccEy@ydwLxi}2DM*=LE}
zdY4ejdbr_^)%{9^dTW)PR(NEz?sn#i6Z^Fja^BdNYgwLn=uUhzazbz|<pM&Gn$UKW
zw|}!#Jaq6=Av;&?E<Ptc3$e3v&|fo+0HcbNkYoNcEEPqu(XoLR%QzYb+x|qCzOsug
ztaK10u4J5gdhOASb2&A;@K+9YZ!SAyMB`BOWv&{DP?~zr$wCANS9Hx<nF4L=O*bkn
zt%Z?Dxq|`2=2_*d@v~2(RqT*hTip(W&lPXTGqzzJ6eGCo6=zPo;Z^&_!j=);GA4{Y
zzgHYh^++vw6n4U<#Ldm8s1_43-%8*b)KMCwYqMH}k%{JVGK}yJ$4tT*r^0yTWiyxq
zL-8B`D?M?EomdU@>Lu^9h7#unQJ5QAA_VVgEU%xn8Oc$_yuB@kN)tIaV%z)YJNUoh
zF{$q0HTwFRzV_NcN6<e^ln?twsh8KF>%E6{Zle?KspH2MQXN)!L~rGMXhE@RwZ*+D
zTW~qfHt7xMvz%31S>DSZ=?>(9Rx|DLs0PPR{{-$C^hO-8I4&0{rKlIHrQ}|RnF7Z0
zP~)Yy#>F3VZX-fS!XGtwa&n6Ga4bT~=7);L(Qrz}$z2e0>xBAZF2bUL_ngY|v&A5D
zn84;UlcY6jVrrWVe4)4uOM6x(d{f;(JeO?6ye@D!%tQmaM&{Pgqw}dIT4v0R)7660
zCT^#Q4`QB~_MEv8zQv6N#pY=x$q8FEQy|uO$`7Xa=z1tDO<%0cp?Bg1b>hg2dG7A3
z=7^OM(vQ`p=QLGy(1?&<hGkn0QSM?&|Dk~m?U6-Mvu)eG3L#&bIe+=8;vu@GuPGCK
z_n)Zf`u4j;%I5M5NwOHs^D*ni6$uWg7QS@4DX^b?^Q<W4vM_emX6?y{rGMWSr_tf3
zrko#<H$eXEI*cj-!e4KRjpN<#|APxK8n;;5^{#v76D35yQ#hM99O&ukli4gpfB>fK
zXPuj|c!lY2BxGdaoL{Wv7GN+9;HmxAFr{=54U!1uNSUa*?yEC_WIyX_UtFo<F><lu
z1Q*t5sA)?^*OQRQ3oX;?00WyKPhn}c5%R`+=r!D`ULDNRmzUBglqe5fY)7h%lm*4<
zIRWR0{z)SJ#$&a!=$OuR=~G8*+ZyLNm%7(H65KlXg#&C7<Cqm^;2}dI4%r-n%~u^-
zr8bOLf<6{WSQ##xZxjCtkdw)x#xS>>d9PGsC7~!Y3CUp$qNcqcsc!tLb?_EqF`6fi
zrK9IcSG=YCq*zVST2u(raN4!ueuJQ<AKP_Oup%9Cq4)VUk@pK&1H7){#P9JX0g2vE
zW+rmOK4sGjoi0ghPkKISC|b&zUL(-hv7l@Wk7YT(s|`+FmGS?bacBLCscHYv4u9Zl
zv{Y`aD0-3f$BYs(bvpN@?w=o~pd>|<89%h!hJd0DXN$_Bw}8;9atu%}%p5bHI=|}U
zb@7sZ@pI>S!bJw09)ZPb33$Ow!E{UaaJFrIvzENDsV=wQ%l}XXF$I`_UF=S^gSqzM
zVWcHYwbev8XbTk-6j}fW&JENT9v&V*FZQ?2$qYtYTbs-K5_%I55RfVeHF0ql4VeQi
z3W{abQ&Tw{_)H}N7E^?|bgzKhf%nzkV^LZ3^JIdoKRc?y;7sn}j5bBUM$FU@9%%2E
ze@EbuV%0<l;7q#`wB!XZkdJKUz;Bo-4l}M$4s9-GP`EP^%9$+dRR|pT)WBL1@cPeu
zxSQEfx+^Awnv@(18XAh1Itw;kc+{A0XZ4i{w7ljss*o49Ta~`=RoPF*Q*`P^?_9kh
z$87rbRR$&N!H%Q2{*<1TB1xueM;ZD7W@<`}Lk=6+!?0616tS#qhdVBb_QXn~&Yqq$
z?F;t>1ABbod_GSuHzBQTM}nZ?r89flR$UdT<y|R=$@?X&p!U<^^b#Mj<J+DKV(Y&q
z=G;Xnd%6ct>qM2J-Nscd%A2EKZtC0kRW2yjU9uyuUXke1SL|=!XWU2Qh+Qot$Cjl&
z*;|^T14x}vqqbdNnU_->;o+J?ZoWi`asKU6nILv0Why!Uizv7QozET<iZpAdH|y91
z?C5k5^c9;_kTt<b+D-Z_svAC*xz0;V(C)B?l8Pb3s{6IHjhc73E`8>$|J!YD5jXi~
z<;Ps--L_ayu>W_@;gjauuzvs|x>&1*ih<!Pht2dMAf3M}#NGxf<&I$B<M+)sdRdI-
zVNlDZ3gxSOA2K(eEF;6fz<>c-qj1RguJ7LdhII4HQ?t^%_2F*e;Vlpb1`&n#a-_OX
zVyF;%Hm2X+dKVQHDbRH`_@l>wiE4k}6vN#X|1*?hA*%L5Q-axB0)v%BNdsLTJY?KD
zyr1Z4C4jJyTv$?m#o~(KSck`2xbR)AFx2XcmX=n+b&a(9o3{J80eMzkN{Dsu$ElDD
zuZ0LHXIyNg6iuY1$2`8xRMSo9*E@XyP`D9p#_~{J;cN?MGLfipau%mkOU(G(!@F{X
z-!_p;;=@`B0Rqd(w4pCk$0{4q8vAbta)V9wt6?_#xApF;#i8|_c0LVdv*ea%JVf-q
zX@90gS%+{evS)q~aEW{@hS-$2AE&O1-e{X=c}3Q*Sk{%(2{^8TAEGJ9I6^LTx9m5P
zE7XOs%<>EwDOZL^?F9&IU&4^8)4V$RQ*%mu8{TQiwbsyL=mQG~sey4kNGZ}~M><;b
zd^jRhimv;G+m1{2&?gzO3SpwHt0hu@tc*`=Kd0Sns6G{9-dmeOn$eiq*(YVnwQ47;
z#c3G{8XN4Kl%Z!aIQD0MMYmuPb53~*3>m}ZWL4Yithc1mTE-S)d>0!1^sxHBmM{(D
zWHPB3S($|nbfgK?A0h2fKOpe)8ARhb*fb#vij5e4Q1_H2@Lm|fbI8A)^eAA%bW{AY
zwXx|b@KwOyO$XA2X4rn8x|6V8J-ym=8SbXx<!#rad$ykBT6LVT+uuJZo}-kBnL$Zx
z6|z70%3FY$Iii7Y%N9{wT)e25zFUVJM$hLAyE~l7Xm7d-o%f~i3%W0z4{*gg_3>q8
zZ(w|{)^PoiG3&JdG<cjKp`lR$(J(f>+8gIh1r<7MV4I<Y*w|iv6nwa?7^ZN4DxAj!
zxG^(?M%nsYgmTxT1j5;$A=&gn62BZb(_PVk1Og1*H^iGFBA~TbQIsEaPtXQg?AlEL
z5{u$bfvE2_DYWY^a17$&P-S@p-aTL_=a2Lp8Y|R@rU}X!_?D<d43DldNgz*mg|DXd
ztoC(cnLxG+e_E4h^@`Vrll&hCzj7o~pVk$wefGQy5rG~f!Nc_Lj7j)=reGN+v%{T{
zt!XcKE8O3!J^L=k8Il348W)vL4uv8!ws4z*hIIS7O>6UiR(iUdaYAdT%PpoZigfs5
zK%1J6b_{{lWj_1^aV3|z>rbYm{Ii)nI|56{&0-+)=?vG+gd$6p5*gh^I)|F;_~`b~
zm-A9V**JZ{tm{T!HW511zkOq;9b^*0Q~0jg;?yx18B}owZ(8QtgEK>^?XMP22(N5-
zC&I<P-CW!|t?Eo<P{Hy3J?%uJQHi5pQc6v=RSm~qp|se0CMbsGb0`ulFTOA@XT>@Q
zDY9!FR#P1rGI(rpPu2;{2E#uxRe%Gz(YPf#0CSFz1b=`3eruAK&eYr$)>ABjXhhG1
z`r{cxNd)Z|KRzXY37=W+<Z!VbCn@;GvW=lc2aTHc9$y2GmW!)yS+O%Cxm8w?hHe7U
zHPaBps2#TDU3eIb-cWSB#{g85CuI4Wqx!toQ-v~5_r#G}SlPBq3dI*EF>-JLN6t7p
z*|~jF)sLWN<xV4ore|;s+_LQ44&@J#L7`gM1qh=*Clhc~3&t12(AQ6{N9*2;r@p3q
z8^2q`Mwm8tRG^uLUxhEI7?U3-o{>QHcOwTGurr4T)z}?65!q#=5p+%Lkojmi{aK2$
z_as?-Xjlc24)s~qPDc}ko*XGZN7C%5brF%ZebKVi7y_(G_~c#bz-KK<Rq2NCTWeCD
zh@#J@ELaUNJ;v}Z)<O#12ro@Hu%LY&?%pJGc2O>Oo)tV5?2Q83^=?(+a!fm)pDT9j
zKTpb@5L~6*-`A%tI~gl4LnmS}WSz+$B-9LlNgO`=hniJHa0VlH=jt!yHpFJXJ6<3y
z)K65^)G+Hdd7S(ZHN=8n8hRbmpRdUK8Nccsu`7)T<)U&ij=-2@(ww~Qc`<S>A;Tf_
zp&ALd1|{`}=Rbc|<@%0<$iD>nuYTKAq<Olb&lbj4y2jdrz)pi?4Kh{g)$&ygO-yQE
zisw+pENK+drB@Xs&#4NJ7Z07H;n0{kSw*+|70t5X&ks%y#pB!b5i2Mpy(HoFLLuZu
zq}+}ban_vkVUIy6KM9@rV90B)zWX2g7s)X^f-jZIq0{}-*VhInrTRCsv%Gfe2f7!w
za&mz?w9ot?06BlA>hR|T$b?;`nMn9>)q5l@B#^3<foor~D=4lh!%~;8@qu@8d38_f
zY~1CrK(%V%y2h=-%`Fy(;gYA^jd|gQzfnybaDlU2J^{Nac$*=W!$|X*s`*nT#A>c_
zx#vrByL0<IuEeI^2Hy}_*e1DV;<(6kk0fTEqd@6d$6|09|1Tz(`HAws#w7$aE;{wD
zCcy6SPg?@UHQ?g|G`vu;Z*z6!W-3z(B!U51kJu|+Mb+C%3MHXmQ1ehgJzJnarP#n%
z&rd;kxxMt~#to`Ax4Upbiq?~wZC&SAH_yY;t-^{v8!th1mkZGK_tlUja-K7{>o)~_
zaNqvW1NCScFo(u8oTg&V&=ke8xmQ!BE$sVOMBNF{Fx?Xmx5e=xRdUWwMoxOu46kf&
zC@znbj4?QeJA2k92g#6?j^;4`hYiw=0VNlX$NGb-mDlss9Jny$fE-fVng9>47wD!!
zAZR5F;tWYB)u~r&n9WzFmJWIxe)45njv}{C{SF;C=^f5vONc%ord2M$XIfia*NFT0
zEifKc=>>yO?$}SGb-M8TCTY(WSxVVj)frgr*vP_w+9zDe!qagSplSS4w&sP$1$6g?
zuhRKSJPUbfRHQ1qS-&YDj#G_0H6^2Pq8vf_dZ=Cz<>H530pb<gVRgm}12A>ZMN+>d
zg|ca+h#**4awUYFyCay56D`lK8}Z#8y-0q(tFP&eO`2x}*A~AQ;BWDFZsar^O^V&s
zpC|72BG<IuvuBm%6ljnn(50}<H0vYblJE<vio)U=lg_;Vce*_@ID)rJVA6p$rz)ty
zqy%XA|9%sO3q&R80!GJtwPJiwiGX|M(fRmYG&E!a1r;?1ysop^Zn`qxy-u&$Hy%y-
zQN<+EyY_5a@a#2Inu<!lY#{T|f~fclB7Y^cE&5g?Nnr~d1x19cLH##_l4r`|*YB81
zqhq|uZ^`&j)^(`&AsM#j^P8^;39_w?VsT10HRTmQ<jjqjtodP6FbzHiyZ9t@!zw&X
z#hb8w&TBN&I9u1Q+<tAo^Q;shmhaP~(|n{deBts?iPT~#dVHiWc|ZvkefLdovdxac
z#<tLWlrxa-u=jH+Qovkken>zPy33U=DYXH=(!aSLW<QjNFi)B8yz!lP2-)8(6Cyt@
zeoBj)hr6<OG@POrX2h;YZCHKF$hrV(gNf1BTs<eR+djPh)^HAk&acW*m;a`YcLBF0
z=siH+ZUAQGFl=`7v%?0DXPJB%sMg)#Va=KIu}DO7-*mZmq|*BXe{=o>UvqrsKGhM@
zRe1a98G4#pVIymAvQ%)~_E4g+&CXYBCBs3Wekp`g?SJJl)r0JthrXakX{P>+UiTqb
zeY_zkOnIxfTMnq=9d0B~`<Jt$TJNjZP>%KSZ&L#Ob6FX6w++iI998(Nx*uLsx{0g%
z;}-a3!seq%y;4O@{?(EjMR@6fVhaoDe&$;MRy8%4oUjlyJHGbnLG`u^^`&}GebsR0
z;L{f72{PsO&@|rTm;!EX#=5N$INi*%d*m`M9L?H~ZrnSV2U4HPo)>tj8@N>9mi@9k
zia?7fy7hZ<%0mRKA5;`u|1CI%Jdr@csmaMlM}X_@q%dfprEG99F`+g9e%gT`rL3y)
z;+=2=G5@PK`Y><^&ugiV=pUK&b;q+qgG((vsTxN4h|imjqj?R{^u<|YC!?dVZd56l
zpQV=jKAogqI+1V{>YLh6f%dRy^$q0<&Du!(w4VKgaRTdoJe4@t(lb9kM$qd72XrCg
zr9aivfnD5^*A;<Gv^<x!OGhe>=0x*Oo-I-8SyYz&977OEUCh!XP58MMV`5Bg@`}Xh
zsii<RYg3q|(0ep*7rarP6}3?~sH-)ef-g1g@_0-!utvEaXC-|HbxQPbh*Cn4fW-I%
zLb--|RKM|n&yH+S?CZLJ-^WiuL-oVD3RTQdu@+}r8`Rd**S8GRIlBwZ8Ni(3bhIqW
z!ph3+vi*|Z>-@c1xJJt31#HDkC*}$V^i7~M>Q_N@t(ffI&r(@q?eP@u8+^u==J!fw
z$*4H&%K=hirDhYHJK1XvV>#26Yh($n4^REl{9@=YVu)j`%RCa?(zE1hUVLHx=wP<A
zz(xxpYv6~Wsdcv3<7Ns;Cm{PE@6UHEaxmEnO5RRcfgH8=yZ#zM7}?a!3{KIf_~(Q3
z+O9W<LIyd;i-UHHt9x`r6_U8wj;Mjz1qP-s8NxfKr8=iANR|&Yj^VA`Zq}>Rk!&!Z
zOq}F6IaACXa?_2rl5j5iBc0+CEi~NI&80w`j0<T0m3<`iQMfC*M~wiF%eF$UTpHp}
z?nfv_6C*=qwCi<Ze^S)fcQs-#uKFPjCX;wLPY|7+7CHKR;D|;UJsxOE{+fvB@qG2U
zUfTai8cTXKjEc$JA%x1ixoN9O{fBOIC(PD^aETz|?^f}A4o4jA>hnbYK&UvQIFbr2
z1%sB-u)bE>gk-ijJL*_6!2Y%JuQkE^7j+_XgIn*+1+txzmQr<{gfw66o89)zJ>c)6
zKy$0RKgdsS&0jn#3=+g28|n{GA8#7$`|0!ry&1u%Ju{5H(#c=Q%i>AT?CUR#k<sN|
zO5YTMBI*ft1h2lVS6Sgz5ye4Dy+80sPP!tTE`4s~5^-Q%2(z7378H5J@%d*NK9@b;
zBFrgJq>UTMIOJ&m<x)HeNA};mDRFP9x&E;U{v7n4DuH#gO5S|7oIr4pNO-rGUa7cG
zNyUoz{bnpKU|ud^K$Tt*-9t{m;i*eWf8N-5f6N7@KW`HkpF)xJ*nD@r(xdysi8>1<
zYwI}#_S`k2=T$tM**n?yD#+*btj}eek~iNaw8+8!)&a40ovE>ZJMhkioJ)v8Ny~hE
zMsCVU#hG!%?yr~(`Y+0%Kc>g%IqjLJm@}HRaScPw=*ozw9qm*3+A}}Z@;|(B;PaWT
zT=y4kcA|Z)MhBrQhBWO>zcK2t-&SKa=&1Z31=kV`UkZLa>?kA{PLHCWrv$eXd}OjC
z@8=qvITWGF*y`T+Ao1ObykPVQPEk!FuOm8ZIVa#)j_^`%Y_CO4xGxU=?@rjSeSaQy
z(z16xfQu(LoUn*~z1$<&`#w`;adGRborEm3{h6byF|cOHAiJ9?YwV>koS8!M)pX$j
zlK6;ZUGxVF%m&gpD|7bm1hRU&5~oX{@Avqn#jKx&2S2TyoLSje5;h-s;aO#_yWg=t
zXUu{(SY`9OqFBrEbeWQaguCl%=PkvLT%%&8rImH56i=6}xgXe{1O#wq-~DSc;4`7`
z|9w$XfTG<e&bsSM*bIxy6q4WcP5(+|mwCEG)+gB;jWesV?m`iB{kH=fAf%4I`P$%V
zV@JHEFssFx+CYz)=1#*KCv_44yw@*OVO6TtM)#)^r*#P6-t1X+QfOd1a@*`oWZB$#
zVh=a}?*cEXe(+N-9E1L4so2dA0D??XLPDsNlP$}if{w1>y9nmv;*!eA6ME<Bh>+~f
zjz%mu`HPD*M$7lJz2c|j>^ygtWhIBRla`8V<~}I}|50O4{-;w{s}yDcojaU5a`3=j
z)hcT)@RVAnU8T)!uNaIMs%qx%2X`{mTq|g5k!mm?{lmXNt1J=FU=;E7Yc^<!DJUpZ
z-d{Wx;=`W4!5%W}D}mRp=~VZ-RbJ^8PA;T*PVYpnYjQky^ed{f{?IRG$j4(<U4ZuO
zVq&_|&F%GN{tHiUw#z8Vb1Kme2g5br$7RFHbe6s`Yvry?UHSHRpmW>U!U>t^<sr8x
zG*_UU@_|Oy)s(IZZJ2Gcj^~>Kf5T4u&yb4B$PebeW}G<!ArVpkT{KcYDu(-Tepc2P
zU<-{-@iKD^1G1H&A`MoDl)`_jb$I6#FkoqSP8`aK%B1|8Kc0{fl*PYG4s1x&Zkr&X
z^1d^$1gZ#D*1+W0qS>2fS%@&P=BZAUnkU#@_tlHvJ{G;MRTQ$d#GzK(-8h|o=bSbg
z!BOg5vpBD*2Idc1f?C3tUF1sqA7BIt4$GbpMQTe}I=>FWp>Q-llGAJ$P5RqAm{sX<
zEM|GliLyMp4fi4*Ohsf4Hl1<I&n`Jp%C+sO?2pb`T1S#n`;#IwAy{12KU+u7VQy1`
zB{Ba84!c7E_;*?#pq1MnFVcwdM?_!Vv$-|K|94YrhgJ)x-du5l5`pZ9$G7*?ry(;i
zO=ah_I`uVIB3i$#{n_GEoYR|BSf?(fMpdE}j_i@2&a3F}Nq*7X%@5!qs09SdUC1fq
z47KJne1I^5&b2vso3ww2ji}?j@%q|%KBc6Tg>`jOHI=O+{e+sb;RE(B&yV$Eji^}9
z474zRxdap5`QZM{EqP-;&jscHByx-BlEr_*B703XFLGN8$68uDYCNF@_-_D5qKK*c
zvebR$$s|kt@Du%B(yz7Yhl}%D#FF(T&EHsqLT{l@iT~RW?onF5+T-ZeahOXrD#R$<
zz_s{HovfQ{)*h?XJ4ZMEriQFR8O!pXx(-E&r6h=&;rXYg`Bk&^<BdW=#}lO@#Tm`3
z^+6z7)<h#2^-FTe3_fY>)q#m4)cn=rc4nX?tI$_ldj=0RKGd7IPK!7ymw&iEzJMMD
z(I^aPpI}uF<oBOkerQL6K1{yyBLX#~>sb9hh-%?U&AFh+?i`WDY;)SFN$W7d4;M!&
zkR!1I8bW_FEe#GKtV-fB)Fcr!`V}%&oqM!O&qRpmqidFz-@vg;$-jG=*DzdOp;2jr
z2xm4xVLZT9(8LuxO|F}B0%f};RlCztXE6!xzIfVd4LT-}dx68uX+8OOc<!+oY$IRm
zY1`ADb)~Q4t^X7|$r!4HV5bsf;A{SD6aaff*`}qA@!&PD;7JdB6_(2aCTB-xFJ6{!
zLAc(Uh^s7e#oR49Hq>EL#UZ&96}HQz#+2ETH!O|e4q1x)MT^jk=iip7j?kahSH>Ql
z-I~45#0%od&bB_UcO_{T`Z!URSS{LSN@A{6Ls)S4*&8ebYOz)=mSQAm7`i`Oc>;wi
zS4Nn+jh+ZXWoa4Lf8SO67-h`+y4%qbji7f^<H2a|lYeSxlbDn#l{ywCxBWh!-fdlU
z{K|ahh8GEh;mdw)5X68)RpIV8F*sQ2u92zm;TKT5fY!!%HD{v?)U$ks_EVm<QWBm~
zbj`X*jD)_Ql=V_}xg$SUcG!oH3C1$hjjpOkW<~Mu3WAyi@3Xvj)FONMm)?^!QH-Ib
z6Ved9Gt%f?TA5xfS|0fI4rqAtYlY~^sO{9RAxKaYKmU3S`$GaLB~QQOTg0x6AQ{x(
z2rl1;oQmoB)QtGqU0}dD4`8PZ4?3;b*HzUWeSm-xJ_51^Zlmdww!u+N)8ob@RT25w
zOY5A<6(FC$l>1HLcNv0<#)G$=?N_h3)Ly%24g&@Q@}NVF+-em!SHq27<w{?xC(@AX
zEBC>>SC(MXrKYAf3usD%px24L%Lk5n?n?p7Yarr4+U-BP!dJlU#89d1Y+1|#02B>d
ziuh5xD^ss#E{SbVFr0BFl&9uJ!Hh67^N63%JV8TIuk7!Vms#%Ue-`K%8<v^lQfZjY
zup21GtoSc?)KJ9E&GY7RFd(01{9qB<Add-yTW+poSkZ^x)9p`%9*1+-0NG~eTa1>X
z^5>!!sN_749j8OMJI3)PV<%(51+qy>PSX6(n1pqjlkv|>Q$N{API_^@HEwmEeIny+
z^A98)O4KE#3wI;OaJyVRv+?%fI_XpJ_4!56_@r>fz${dos$CAs$^Mkf8>7Weemj=w
zJ)X}fs>ztubeFDof4^aH&<3ZhqVicR%VW|22}l1(a(v!>1-0q+SQ@6<W~S>SEF424
z^1qcr@qU+&k@56x7I9++<7ciYh$SWwD?vKdes5s<-EbSL?i#_CCuf|t)O+T-)}u8q
z<o%)1)5PDgDNwTPKfE@_Vf({vv>j_*Z{v8r0`$x@Dmc)Laz6X^1VRnI>g4SbW;9c0
zt2jVmWIh<ciPZ4=3;jo93$~)L7$P02)0|G}G<sFX3l1+g@xX_-NdU{OT#|VGy>o!O
zuO$mRE-_?`qXx_!@{U&U?zeDM_wxz^uB$``i0EI>d)PxWZ0WHyG8il?m#McWgmba~
z$pz0XL5udKjP2N0F<s(Qz51jhr*%caZ}@uzqmYmyir*E3fZxQzLIwy3f;pVFDve4_
zwce6_yz(MB>MGyJ?swVkm4#8U#17_*b5LO38Yof<q+zDfFjm^Xwd1WBZ}w=dg(T<b
zq;`Y*zfDNzkPq~-PSXKFegOAV3egS?h9>b~JqcOe1+K9B3f&yF4KbQ3$J$8Lpy5Sl
z9E!ptix)zmVe#;~kkNd;r$if5Uqf50dV0AiIzKDmKTe(jsyvqA2^j+%TJAezCXVxJ
zHt!Pl_J_oQ$c3)Z)sn|O!v_h+?@o3uq2MCtFZZ^Y)h8s)Hfmkq!`}Wa#sW*hZZIiG
z&Q2zbUO11CK_HfrjO-Ke`yK;J)=4WOi9)d^d(+iAfylde4z8|ZYF$AHBqSu#LH{IS
z=-RH1ft8CkM!M0MEi;<iC#GV-^0?UM*)}RaIqY}^IAHQlq-&wQJJ>$CUsV|bru*o#
z+^&*%moM~vp$O6>5jp?)V)MWC*qSaFQ<8YAKTM7G^!>9EP{zZn(Rn)@oz5<@Z*JhW
z?&o0~M;v=3{|FV1FE6fr&B|qEQM6Kj<gnAZG}FTIAPKD9`-A*HlXU$V$DWkgu+15b
zEk}hz)?Dsi?xDRI4W0oOiz-b}>DvL*!{xgB3XDYI5fFwe%;d$w2_->jhP2KF$n1L5
zX@V(RbY!HEzCJ1F(HDzr=G%Vn{YjYbjv&beY+>gbp0|%rRR2K(q}lAtKgf3?b94X(
zXux)o;&HO>u=$q^JoEs(hbc%JI6YW+NQYd0PvM4%wSyTQutA}Eo$rz{F`)qu9Jr?7
z;a7Y>kjt@^D1gk6zKDv1RGaH7E-Vad*LW5`w6oL6N%2^pjD%t!D@+;!6pG}coG8R(
zKP9U8;T>UG&PKj>N3Pi)Tr|@>?h^Ce1sNT8vjsv1yRqa9aymK(;cx{bf*_(0lOE&A
zdS1f=-9N!k{}mATUBZYj6dc?E)xN40s$-Ry7A*RPg{IHv%L3SM(D@GGfgwK!xTy&F
z+z->c2h;icfurRLsLmy1WgDqAP?VT#j#qmJlG%5_Z!-i3uTD0umP47j+z&|sf6DB;
zzdsB39`J0{`H=<4#<D7znBTL^?O@*PdcMhrcc#WJ7hJ2|nHqH7danabHWNMIzU~0A
z2?rJ=SXhtu=tHGB2CLa0<L%M>3nI?UCsJ(xjyo8%^gqKQ1y3<UEtE*mi2qOuFsjoD
zW%<v3_sXCdZOwGPq{aUs+Vc72&;cIUkGfff?lPeFw!coP<SmNn)ZIOc!xPK;cduQE
z3cK$HmRt2IZpR*uCT^OYTvsrTS)KaFGW_dfDkCjFA(ISe3+(cz4jf27t}}Syyvpi~
zNUCSP9C?8ncYLai_-fBY9U$~!5RdctwI4HcfVl`HPB`-rfmjdiTKi8Rt{~3c47z>H
zgJ8G9d3?#?Qv2%0M%Kty&%l83{p}TaJ)i>K%gLd^Fltr`l}_MNd#|jl4DFnC3Jwa~
zSUW7(HC<78UF>&)z#DS$@NYxrG<0;!r&}YyUh5AAaKJ2Q80C|%n1ui|I23Q+VuXn)
zH=q-6Shvj2r(eu_Mf5;D64J!S-GQCrhxeSU{|cS_YhGNhU2USC(BA>6ukGKg=<SPc
zg(ZsrrlYj$A4}iOmdV#a!KM9K&jhR1`L3k8!(}oO>5tT_U6jT`iI^gt9)y<G9^*XS
z+>P7#mE9{}xUIL86SIO<*D%lG{7n6MN0!3OzT#VKVYnA+k3gN)S<5v1LV%t(m2sy2
zcOOYrJl}H0z~mh)*{TVmsnHd1b31bL`|sX*6WWkya~H03N5H+Yn{V*y!~LM8rCejH
z({y(}b2AQnUIuNSk!$Pgfd?>fPv?k-oBJ&wzQL9QZXT1X<F$tKX{*k`1*T#J&ES9(
zPP=k_sMT+I5<K)YB*SX9_ALmZaQE=|>98Wkq}L>)pb)8X!vfy&hS<r`ak*|IeI*0B
z&jjx^kRjOI;gazvd95}dY8%#hWN_ND#Hq+ZcLw^LK>w}3?IV@GY&;es45WF=?U*T}
z*6I<gJLv8ADtVUVVZ&j)2~r%DZvGHQ$qRi%v}Zj`ew%!>Ka95C9PsSQeA;r!KKrz|
zFWdKAI!&wPS`y{Uq>Gs6w<dCDdw(9Z#R5NX#tz)JA?L~)CGKit+zk#Aqs3Itym7uB
z{RIKkCuJA*uP?kgvh!No3kVT+nWDIXd{o~5BQJTuza0(~&Ptlqr=3kHP;uClPu<kC
z5Kp$L3IAXCgFTOYkjJ(4H`8Bn2P{p*G`Vh&_T(iGtX{yL!uNO~M}Ej&P`LM(TTL**
z0P}$}_#Fo)C+A~mQ`5t}SGDz2+l*ba&3MzR;c^oI6>R6|FdHaeP~tysl}x|Sr*`y=
z4E3<0RO5*c5HIpD8@!99f@O61idLgAl|K>foBbjnRoZ@zg5O#>eHcqUTAdKx>_Yf0
z7k3~JPLR{jGu)mVw=1wc(;098HsTj-ucCQMBRk(U0TGm{&0^uNLm!{b19D2w!fByz
zjj@Mmgh<VtpZ3#Nb4)6XCu_9lp+EW$peyR~nUR310v>7&rXAoebu?=(`TkpfND+*j
z;)pQA$0!5)JFX=Aa_H3q>JRsKiuo#E2{^+ak1NyP_=ZMCLt9cA<SXYAQ^33h2573w
zgL}q1z@pWMEf4anV1SmNu1dDXb`It-mTc<Ak+%!b3?sD;MudcfAs4Ol^Tg0c=>(=Y
zE0Xk=uUr4|quAM?v-f^yDPl-A(5!opR<hCF_6i1g8uTn#fo}FmGU6vW<~ZX`ou>lm
z2+b^F1f!MycH1KA=snTf7vmlU$gU-5Ii*<>6%KtOfd7-|ot;AcCW)72G^e8U{UU4|
z49ALy9#Ha!f)Ug$E<FD!ahEbExRPyaAEK?BkPYx^FMo}kc80I%>HT-Mxo5+>ORo+5
zq0CJlldsnNeKx?LO+-k%xv3K=)*&V?&I)vz)piSi036IyE7o+lo6644z8RO4iJe#9
zapqyemkk!Ip;oB(dd+T>{R2GdxIOC>+0#xwIzBqu5IS#q^kstC!pWT50tPX$oULui
z)XiabnR{?xG&l`oqNA1g-su668A*rfQ(>BR+Ks2^yq~2+ixri#qoTEe@t5XhNDyJv
zVb918ZJ)aI$dRoCyg23avlw6I{wlu9H&32r+ts3!A3x^+&jap1TT%Smhut;y#+HC|
zY(5Plq=^$j8f+`9E^S@MbxU1dK92zwcY%EJy_lcZp6|+&+s2M=S?}ZnnVOor#%-#Z
z(D-h)-i*s7qO&r$i&tePMVa8Z=C-6-^<X=u-vGN2{fbel>)1y60?_0DAOS<8DpjOH
z6(9P77zF^T?KTnXJ?kjEyVqEcyI~0u-kh&Q9!X#@I$0}}u^by+^{p1~{x-C=IX`<U
zFtD_<(;rPKohS$ifML}#7){+?@05C9ttEJ${N{WwCN>|V?&RX~R3FS;p1(SL*qw88
zS6;(yIkHO*cODYOt*_iwe0&-R1}=@Cf5t~YBwJCT&H$QNvJlb`toIlg@DN5E<-5N5
zm*)M%IbH`}Z*ErNrE*v$waf=MJg)6i8ZKs?Mq;U`5BLw>6s|iqx_dSL)k<7lHHK+u
zu-*63j&}<F@?|m2ZC*Hc<Y*;Aa2SyKAZrigBb&3~pL+72pC4;B(5!Lzt1G?)YeN+1
z20%F9qiGTVEX-H0K0x}jAXjW4mR30tL@6QoXMQmUQ8FES8>c_WJ3i~dLPT7XE#>rw
z;}*54s;6?ESVh5sxNW2yh{^dC77yt)KYDbA4WqelPvp*CF|#xHI9%{(p*GCEsKh__
zZVcXyeI3#p(zbBot6GzRs(@=jUpVB_y70-VUe3`$?}&b}8mi4DZ`5rdVF^3Dws7Ha
z@_bv0jd}LwfpCBij;eO!E^72(w04*K-@^{R%*b{Bkj>jz`MqB_fz*g;qV_!I`y02?
zY0Je-;<@4^-J~sNp6N=9c3}7e(z-|MREs&=F-XzDeKpbK;{`Kbs%Ljbs9B{n0W$?M
zrJ{_gZK1hmLNnE(r7McbPE*x31TZ8%rx*ZZ!OYLkPfkyNAIh_{1x_$V{xrH`?Pm>O
zhJLy`6=TpIh=GHnwmUg6AOje<P7t*RJZKUg+5ev76oS>fr^=aLgh*1OV%E4X8<-p3
z7OPygjlV1P8LqdC7EM3^ogBU040lwt4<ar`C8_$X$-_j#BYiy9Cde&#HX;1gll=Ly
z;tW}0Ry>-g7kK!S>}s`909vqsOVVY<1ZfW*;L+U(wNRg{n<0}l-(SAW>hADzCpCVq
z$;(*u#jhK^@aZh-&oXAbbBb2Gw}-ayd=9TTI+ho=d$Od{bAP{!yZNsDFNaA5A|wb4
ztPvAHXhBJ-j)`)^_R#Qlxa11@Ze)Qq8F_QCX}R~3aTxeqh0?Y>vp{+vOc`*~{YmH7
z`jNQ5zioAWdU`5yZ~#D~?`~pZV#>a}q@)3WPB71j`D#bI8F#fyHuTNZFK?27!z!M~
z*F}Zt={yJ{DRbJ86S~W^ffkyG2vXY6UG93HJgDt}nC&k!X@*K%6|6EgpZ*T%?77{T
zy!o1+Tg}cD`~7^)hf^dijpv5pmKaXb)8*EzSYJiwjM+u~th|tQCIZ*9#J5%W|Fr}~
ztTQJ#j63@l8p9qp;<ogi4h(fXcWCH;TasRfpXe0_Ij>EHK?hiTr)buWidt_51QG3k
zMX1f7)nhX5OR<>r2KDvv>^dI&nYD$m-@kyRWmmRxJ(6EF)~!di?qa1264jZfcbc53
zjV*;*Zia<p<%?5hWkV|7ygxudU77clV~KWMG!$3gYy3E31GNk7<Ne(Y?qv7eTv`&V
z*^?I_e{7&Hj$Tq)y1!YFmp28V`nOK&GNAcD*`4_zO90aZB9(5d3pkH2E;f$3Nf5t1
z;(lJ&=P|vXztez}*pVD*T(HmK5Ye(7R{Yfe#GPpa$O_+*EgAE;BJT~YK?s#h5=#JR
z5fZK)io%QL{Vx_<{fl(!aXx}V3YB@Bm|9y~gA*g5B-U*OOCMxhHC%5Oj>|o!ndjLj
z1zW6|8W#rW=*$PDqH(<Wj`RE^sEmy(3IVy!GD8QQH*e?v7<<d8th(<D6j4A)m2MG1
zX^?J30qO3N?(RGw0s<0(v@}SU_|Tmq-QC?KJT!Md`u_eO?ihEBJI4Fv9RuN<efC~!
z?YZWfGq>;a$ytBSSl9@@VKj3)zDHr3R2fP~0#okCW$n*QtbJB{E(^P_ESw&9Xurwx
zSg9F3UAoc8thSuG-v{ksQm_5NA2LuHm@u<DamiHOM&+o><gMUiH)Tbbc7XkmGS(3o
zGyL75^O&Wa`H1Wq((YCBzZ{$x&iv)=@L6nam%q)EAD)I3s^=A6+?%ALJ;Z$<2WP#P
zd!F&=CM_*ZWZR^q{*=Vt-X3v-Tx<;@tiA;RI=a_w8!jaG@bTjoP_ZWndGi64XfPFy
zfSiS?UAv-aKzv8ECqDT5+lEYLh+=>UYJ6Djv^^GvC_z9eCNeF^s9hBqjQgB`xTL21
z0&}jU@rDoboU3ye?VG9ZYhRs@sge7}qrdBfcoWGlUn1do)t*!+;zg*M<MLXp0uUim
zH1cAe2kXG)+R`2t+t9L7s8DP@7LR5X>-#zTmygkfWlap&)|J-`^F(JqxBYyKO0cr9
z^zQgI`Ns9o-5K~I?hOAj^vC1ShXnp0cBvSl@cpr=b2gcrvN!$Bve-@|fy{JCEc*pK
ztT0rd6ZQualNOhbx0)<G*l=y{8v3dI<Fl*lUh`aEcy_ejYToO+3`}sAhO-YqAe0>x
z@5so=U~hn$ez-(W7@@xfguD9sxaBH<@F5s=3<Bfq;ClELOobqRtfpot{S6weTuKi@
zNoY7n&hHvQdoeS60VZK7DJUq2!o<LK4ar!E0QyNE0|TKS9w!z+KGe6ooSiQ@93-YP
zaBM({r<qn^x>pUd7^0j7cQ^CF3>3sjZ?R4inC!RTnP5xtxiDuC8d8ZEH9gR^EylFK
zzvYv8q%>`9eJex7Tzo+M1}w!}QHlE4cPsOqpZHy|I7sU5UYU*5a&lgEkKHsik1EVU
zzg}D$>x=#lO)pqf-k9Z66L%Q;F@iF+!+5sCwa<fH8~`lfRxfz*%LX*GnYDWZjRbk7
zeB%V~@#LZGJ`K;Vz2TYt@WauO*W;8d&cNVT->6debZ9IV^1%sf`K{d}>5!JZ(6}lx
zLU9KVRL)t8fu{b5S5V_qJ>kDnke{H8(W4`y=^2e4z53Wm257Cb)XoPJgT>)$Hkt2e
zF*?(1FwLi(M&mxX2Y_SSJ<@*)RMa?u4Ccqk$VkQgVVLFk6~kdDD27`jsooI1G<uh}
zoao40QBi>qNCS#IcpkT50j3I(qZoBS%Inv>i=Ewz;RYatsoE!o@E?4E5*^HR?6W!$
zLT;38G~;zU-EHsW`u#eye&e{sTfSwK4vBe=bHL16KI2DomTwqJKuv)L(#QeB)zA+T
zABwEEH2LkLr~!ngQm)^29!K15im@`={%*4#|6Yf~Dm>^ofqQ?&py2eOPZhz4>y>M2
zl{T5S(**x^Pk-gN9YM%BvASUFVBOE($i#Hs3F@Q`BNi+b&O|%y>>Im^QzSGrOejAh
zRK3NOlvW&)fph<XuYZ{nbU$oHJ9uHF3QiT+?k^Ji=2GaW1KDm;YBd^NSx8U!o*!A@
zu}Mf&@$oJ0uS!IWF}W{<7yq;&lAqi}*MC$%=r%h$8yp^v3V{k?t$%<iU*(OBjm?eA
z^+n93L@+q3TkAtN?|g_ZgQr*sM^FYdN0DqjPcEypd15719Z+u>FOsjNDi0>qSmbkb
z<jd?~E)^q|gl+BDkL>u!6KfMwSLTe6NSw?SH)gSFI989+%eQ!>(Ca)$pBlaz{~j<$
zJ^-Ujm%8}~c}v)pu<S?$x|>q5*}e|c<TKmKsIdFRm5;Jq$}xe5mbAP;p+}@g8J;Cl
z9PFDCGdx}Rg4bL_;H4pf@V-1@NP9CXUzLBV{}eJc6=BQJ1SLiD&B6V(V$I&6aou#s
z@)pXHjCi~Nc--x;AmQS-wyeHlnL@%_d>=vy6tV2MARwM!%(e{VL~8c9*MtOP74BZd
zv}*i~yvo*l%zXlLl{gXbU8SS<R5nVgEC)bZKZetC49TJ!1%hr3I_G(oD*-QSOg1Ph
z>sgA|kp!M-b9;LLkdxB5CITPFMh;DF=Y8w+d}cfzjFqpZxaQX+q=im|4?8>WHFkiQ
zO%8jhsXz>SFeRXTHDC1o?sOi?-00vMp<C-P6=N9KVj~ulX$Q-gW}SVajjb0%L|dl1
zNrBSTSHx%}STYwdI;VI3eKG^1c27O^(7l`=sEYX^JHbD8Zs+b2-S8{$w%#W-Sm=Cl
z#)R5u@}Y<?*Zek#P%f8|P7SC&0sot})cWrgu|yhZ7h}qyb=-=+(dGxU#(fmkVkK%h
z2fYO>Y#FZoisaAvx#Prz@9WR*37U~vEu>Fv^%s^lbSO-gJ`MTWxa4NB=Oz2Cv1!L%
zPI4L*k+(qXxudFQWDSI^&!@F2tmy2o)>EOBMnQwN`vq$mFVMl{IT0p=`dN%xEEE6Y
z2`F)MA2aVfFm2UxZ=b#N(pO-B#k|tLHh_^-wc-j{V)=V3NTWB_ACkvmr_h}e?NaQ8
zJNA;4Qol}yX{w^45a=K+^acA5;tj4fH;M{=@TmA0Pj*I^zted~-2+AO&(E4lT#A!A
z4Dt+E<!X4@-PZ~u%M+7O({|WGQO7_jP!IAgCkPz<z@qtv+7!m)!H6GGGxxKuGTOab
zQ^Q7#OE_FBjlgxfwmE9duCl+kT}QsZEP$cM0(*HolbeK?)01L4f}G`$CnST_=O(9c
zx*pA9Q(nj##c+$D-b{yy99#-axI~u?y01nXUtX^3G4gAGWkTq$zl||GLEZ<^YhZK4
z#{+?K(9zLt`JtN+0u`X&pe2NakW;XoZ&2+u238<I)#b41E{-}U$aSBlle*q7b=s%q
z%3M$$_?zB6GL0eVR=wMg)>Lot1s=q9+-yYeP?el{_xibTmd~-zJ?$=f?<n+xE_a@o
zaJ^#>=%;tCXIbl|97SMJA9k#cAM{uXBJ*uD&_CfM(<a}%b~wa%&apeMn*2Kd&Uem0
zUEOScF&WSJVSz>*+ua2uD=V0Bcloy~&gJI@)WsrpSn+epyQJEWu~<hab?-O5=XiN_
zoseaOH`I8J+RbIVddZ|UO5Eaj>&+-RGKOWW3bAvXa@8H4xb6r#*En7NO$zs^@=-v9
zN={DJC^JAws}R4Bj=lsypf@Gs>k$vIuvP&Q#xM}2-yVW*toyZyXLEBC0Ai^=*B670
z5LD2PFLK}ha{6ph@{dOK!xc=qxU$<?RJZSUIGmnxvj>GywfE;*UB6P}yDo{Cg49mV
zORt0~+a=kWrWX%P8D&$7F>kF5g4^Zy#b8y?ni&iz#3}-V7e@UU4S(c$-N5Q~P^!7v
z-SEs<66%#+n;ueH=wB^r4oudsI97kjB$zWtt{`ctB%`99`>fHcN<c{5{zcCPR4?45
zam52rX03OcoXnDvV*KUt)>A0Rczx~V?sOfOLJ3-)=-dypmCr95guo_REwD94(f{ux
z34!&vvj<@Ot1*nj!$Y7FmH)~Q5qG>UHgnGBS%APZxSObR1_7+w(A2aONTO${C4>Ee
z9Da2aQFCZi-=g*MYn$p_!>v1ZXF23F*pz1%b~Qr05YiA9D-+rGZy$7mBc0!DKSAM%
z%OlWE<-zofIw;(4&84JWRP8QTHir?-e4%no6|u}w@mvY#M<q45)w#D@bEd6Mc-Jtz
z2}iBW)nQ>V0QdgZ;E##31c^;Dy-IPW!0=q1{BJgjT4=79Y5l)f5Ava&P<_$MSZkGO
zr_Z@4Hoz%JhP8&-jFa)oy1vFtlWJ@SebvHC!0A+;@Q}Up31x1k;rbyi8NcgX9U`0}
z%1HXot<XXpZkrh%fc~`sq250DJfyb~pZ+H3C2*4kL%L?8hLn6xpS&+m4}q@rYGo)Q
z93d?|Esje1l39Uiwr=8+7QVNjFD?ryay?i#JU~BSa>y04d~-|Jnmf!D*;}sq#pwk$
zc6M&LpWx1jZb`7mFSgYPTkeeK-n!%;GLsw=hAf=*OzZrw&K)n%naMwWJYOH-*R~!l
zQTEp~>%(<3#XxOf`<Cd}&Ju!dXEJRMBbu63+AXpy_7f^)4*bL<zA1ELbG)n$Nh*Ar
z@KC<y{likXZ<lw*_sR}S5j(q-Vdq)>2d#OImC9Mq(J(&(GVuaHcL0HTY<E)1F3tJB
z`U#O=R#g{Ja<wX%0k*MRP}1;}o<5{Nt;B$X<nZJq8x)j0e6c^^+QdetrtCnpa;m~w
zDhQjrC*%=1d2h2ngoaVSe?jvF)!z?@ewQB?TA{WU4Z;sp4PGX&;;|j)=f5N>7(b6~
zb9%#~0xn%2w2^i21FmD>x1L&+ErcixOjrXQSU2J8D|%=fd@k>UYdN;Y69Wa494|^*
z-&*nhUcGDj>)kt1Ho6IZ-i6Vjh&0_IGkaTcf$dAAwd<5I$VxG^Gp@cZp#IUfo9&0|
zuE|en=pQL*a%p;M&5)cub~ANM#Ile}cJa#=>Wj&AZ#R%aUrwv&uq&h4Iz##6;r%7F
zGp6!S6&^rgMp2doLew`3e@<05r{kR_ho^aV6L8OZKY6M+Yi7A<I9`G3g@)Eb&H8Jp
z48PAvk<!t>8G51iNyU-7LgLH3=pFZcnw4t#Y+^~bTWHy{OKaCVd(iCbsjP^y{i^0d
zeIICAfE@>T+WsQ@t}ZTpwu_NTNtT0UKoC~i#f1wYhcn40n<~@`;Afqkc+k5a8XD>m
z5;N}Zb<W?xv*E3rqFfA#OfsC@*HUpY)FM`ZLC>^5DR9?yPuzvEy0blIZj0T7=PBI2
zCpp1EZ8GA}A0aIr#_#DQX-tJpp~@Fn`(A`)fP>WDH%MXgLON!ZCy&kj+MmgLhr<GM
z_*`1iAlmN8u8c7yA*z@*msHh@SvxF(8ZK;$gfB`(Pxa{A&)8ls)B}ZahEJ03>Na^Y
zprs8auX_4?f^!`_Br*RxhMZo)3dt*97q1~zRbqm%A|#Lch9)Bds<)#`R48*%U{0?9
zPYLmkK#3XXB|f#WA?S;I1f14r=sM~ME63tF6A7t>zw=W4<eD%ml7G(_EasmaZ5*1g
z*2q@6WEhlpWD@;i)lR76X$l6dXc@F+zao?Pw~BqZRHK#AWQhg(l>j}EZ}jmFCA7l0
zfB!w`sRYdByg*O&J|zZP=RCZ#4dw55cNpvNVCV{XDE`g9ss-&G-?hic>x{dO)1G8@
zeWEMj$%Ks>WL9IP;)JxsP_GBiIDC~tdnyqsA(oop59$Utt^=x4Qz<C%k7a1I>1t1W
z9lqO`k{0Dd(*R<(2NswS7ed*$-imn4lmb$1vDDf^Zm~Znj4(B-c|2<u@fW1z`(ruw
z=D+vUyE2GiQQ|mL&$V>QDj1sm<Qh3}5zj38HqYq_u{gJe^ao6V_S%K^0sPN`fpp=N
z>16YV<6h65YeB))+XG1QJveA2VG~o68D+Vhp;fb+Sg5~hXEb~W19VF|J33H{RHqhH
zjDdLm?L3ctZ*%ivtcMUl=2UCpH6qH~D&LzffgEjQRFufOcc0n?2e-yc`hoN$dVR?j
zo*M(dzgA}$imui){V{_A{ZU}n6VZ|tLN{wzMde87kD*U|kdL1XJG8dc@aco=#AKwm
zP~gX7wzoIl-<ZmM(z`cqoG69w^NV+zIL(EVz_Hg-`?;D9m0<&Bd#RWxJc;P9qJ^5X
zFT88KHW-skZhB%`O>gTJ?oLcadNU_}bR0wkfY?aAD#INeCvOL&3hc_h{X8htW2>V~
zxoUX$>Qx-{<B4d5ZF$YaHvK;D#^QcJbPz_u`pYN1Y0M&~#dYf<(_(%BLWCbkxs);L
zIu%DzE*J%5lhOi@AU7ITqA+ls-r3p7YSQW@w%IVHYYY{!Wle*@+KUtG<8o~z=?_dy
zOh6^QySr;|SLLh=<U*$;_uZ)StLTpjH$b}+2@2?|tM?%`^zh&V>Unn-G9!fOe>^V*
z#-29VMH9_|uCWCuTYP*~orHYV{;AkMNq%vNxbf}0d)3Gmo*e4)++bTljM0ana!NNQ
z*(V@>xPqzz+l46O$0BGvusd=&s*|*>y<azrfQ~)dBUJECy=oJ{dW*5q$<IGRz041Q
zg!Ir(Ouy>;Ds7eu`hDv?yt*DNK`dKqs!Y>p=Z6nz_?&y>J^Nlb{J0Bw9&7z?XCIez
zY~xzB@BEg{grzI&kCEw+qfb9x(A7IerrBP6k1oqc4pIjp-%1K|vU;UwgvZ_*fgAq9
z{#l({a&PG0g2Fy&0GmjAlJ34eG@bhrWp4X3O`*2t&^0KWY9}k5Fd$jIbb<$+<3X3#
zBfF%$mg&h!7bowm@AFFa95Q$bdeIt=*9C{Vo@{B&Ila=7y{f=zXt+n!+btBT_Ij4k
zOeptFaO<tc6`Iz1(H$*Px8ggPN7dDo1))WnXWJXt)E2FU?An4ZDYoJtp@WgFaeJqd
zs|u3hhCPl=Qt7kU%u+5Aiei=1C&3=-j##8xOxA(Ul|E6%BR5}CEd1a>1q1V>P9}Io
zae4vTxS4y~L46GeKEH8JHu}DnG*BJxyPeounsunW;(xp#joR#owK?)rViNw>gQY60
z&U>x<w`_XP9F|A^52c6DsB(;X%S-6S3&b<Ejd6gq+_3|py-~(wADt@y1BR#U@%ZP|
zk(Na?8U5VbUSK17<m{`_3z~uEY+Z%lN?$Z`iaOiRqsgVghV$cUOR6+ak>>UMaLvtj
zt%(J^jmq9oSCT7*)&l4-CxOqgGON8}Yva9*<HpkKp;aDLNbt6r?-))nupMwY`r)iH
zZj`vwSYLkO`u;Dr!7M$}8C$=}W#=i5tWDnv3!K?%)6pew4S=J=26_0m+?EF)iDDNl
zuC#4OV5%)QQa;!;);1TUC++QAzjJ2S^TKN7rPO~FWC}gEe<ka<XYafI1c=<F_FCux
z0}DDjq;aY-!9QVGP(;x9I3q;<GwqGrzBudvu88!Lsg+S1G&4x<tCd$ftdKbqq%Ht>
ztea(0OFmRhcdbmLd@-pzZLsM;lcs+~#h0?A<p)EWALVm7gN>LPQS5$BA~PZD?2Fz4
zzkwOpp9fRZO(Rd-ZtktSrheb7G$ooV5Z{DDEYDM|1+s*&QBmduuxvR47f5LK<Mlt{
zJY9TCf;4q-Z%$y_RSJ*3Q}`S`Iijv@b=Q7t`9{R|!dAlNSz3ZO0m|9hXxzcci-waH
zN1U0?@BTsO+fA?IJxJzkFgY<&54D=ArQP_+Xl9#2-%!4P737l5&-UMvz~qfhpaWnI
z;p}+S@82dWmM7YJN5~jw;r-oyE~NJtaA5-ul@g7*R!_Lm)~7`a1^BiU>j~bst`{Cb
zpD^RS6qBp-{VFLuzoyUD;(LBJPO7k@;uEtD8yx;TnIEnRZX8Jaj9Mv;X>!JFL}5K=
z`T-X*sfk-HM$vN385ogSEMCu}A!My({VTSK)WSYln~k>MV?uS(n%WizmhgyL{>FMZ
zmrn`>=$NDk-nonlP3a|#zFXeka{1*2FRe+0XMWo{vM&IKSC#ZK9m~I4$nBVhx7xB<
zl-02&h=EIQFNXfZ=+;sS*O$k6gKMd||I%!RWYlH|IO$1z=bd*T=VkrdAMnWf;%bJU
zPODqNZl^)w1KAo%1*@+7@j@2Qdbh^(duGmwu(O*!E^i^5S<j3l#GpS+y(_M`s7Qe1
zQc>XrpJEW0IqrMBe0$W-?A-cI7d8%a^nYUdv|pQMr$tl~m4j6{HozrSK-9?~iqFth
zGV)?S1LMGwqcGGDQY{{6LlM{+Sz3WvO{5~zxwp?3Q?SxK1b^@H%heA@!Z$Uu3g+vM
znXz}UCTiacMR>7>4wOc3?W`o84m6Wrku~+9K6V^!KWQ0$eFsO4u)$+hbZ>hqd!*Ce
z)!szv7g}BEhGWCUGK;z$MMF&%MNTocwl&ox()!B<JNgY72>3P3UCl4o-E16a8FZBY
zSBZ=nDEPr)<GSU)GmJgFAC>zQX2~Y=0l|V|?Fg`Y@Ff*^Dz~#Voh<aOj2CIPwjMkm
znsC~%gE1T+3rtmV!<}|}l#Yc=uJ`I!bbCI8YH){@SC+5d6zixT{SK|#aH8qpBnq?Q
zcff?c{T!&lW5Ns>9=+RVRJu9vMUv^jM#yPp-s_bPV{v)VRHvI|-PG`znMHH56^8}Y
z+2GV|#iPv)><SVtasg%gt~in1L|v;?{!q?Zk%Z%8i_>K6AYnsWlJ))7u8lj_?#2`H
zGvAWCy~vX51ydW($d>mSLN~i1qh-H0D|_#tD)VolnbdDQXPc8E1V!FJ!sUH`ZCpV`
zccP)ZkSq-OR1=zlzSIokbHSu<f2=^iqe8Rvt+zN5b7zMJRr$0M*7N784ZOf;5e?_C
zv*E7xr*PCXeI^Lib?<E|SF>>lar9T|;$|eQ<k6JQ*xWccdq{2dg!&iejurAo$Zxoj
z+heNvBPwfy&^Otjg2~82wP~r5pFsr0TZJQ}ycW%!ejjr=UOGD1IoRcvI=%=N%hWf6
z+hFd_g;@p%?_l${X9JTj2gk#bP0?*%qz`vVHM2{g%Z{+KliA95-M^o5>^zHGUXqZM
z>YSbpLlV}j;j#^^FVeO`_{QId0D78IXNf>k<W(hgOD1j!`g*<)Qd4olKKhNe`kw9*
z4K$Dgv>1seUc*aTQi%|^wLMWdlcU>E<4IYS<B!2yCSl3iNbNaZn(x`OiYRlW=bj!8
zx8yoUsTdV56%*4j1`0o4IygHSy$?E7jbQkP*yd6&dW;hjs(C}Eo)4(r8ap`k*D?x>
zqN^#awC8TFnYb5Df0Iz)A$ruc$ZlFqMj9%XnI2xry-R{JSgatQaFOaU_RF$6(xOzL
ztd5A&Qhj2pSwxrfD5P@CZEp2s^~9lQ#__MEC*sADMNxFQ=}-aClx=4>mhR~Il%Egd
z$LB4Y{WU%)^}#DaN?QhF*Gq~1-zGO(dgnWp^T865^iH1}bDrbw)dsBx%>+5pc^q6+
z)uxwMQeR4PZOB+O8ae9R5}#khQ?{ZbUmaaErSvNjqOhyg+*xusw0CVCY`$PO_-h&r
zykq#vO6@bjlSrlyJ_I*o))%G1lRwncEnXiNLlz#vXMM}h&0cyi?lWGe#IG2^Om{M2
zivdU{+k;aBGclGD!P1v5bTXZvsX2Cxg6*wgnimz;7?WZ=2CpwMQ0~jz%l*@RhUR3J
zW3waslX0NIh?PFZN9Z3`a9;ucuUoc_64TyZu1(=5#ZcchP`Bm`KRAGAT4z3i3RYce
z)eqq=7jiytl3|`X&%JOlDPg_}&ibV#7o7RbJ+^oBV1T@%E%h#|-%{z^y%PjFob5@S
z*|Wp^bK%T-2|Kn>5Sz>tb1D*E7iJnpcZ{4I2I87;#qmm`U*VxNy%DU@V&ngpb3W##
z7=7UcPOCDUo71Sb1Se1xfy;w2vDUjj#u$;tALaV*9`ya{{5_&qL4O~MzblvbnDqs%
z8#B`036oC{A7A_!(U_>Ay3?2EB5Mlsok6lPNqE)WpL7M!7lKkh@z~>Yb+k0StWbZ)
z-Lw>nR0nTcHlsf`&4QIJ?tY0l9GN-aay0dk8i+Y8Um6{>6N(OpWJnxcbHZ1n|4R<;
zn3EoTdwg99)}0bn58sG+(Vxm<S6x&qH5wX%tCGKGa$-#K_lE>ui0cq3WN5;S=ue;F
z#DvA@)Awe?%t`%h&1hf1ip*JmAReqWJ*V-&_&vDo?zKjSiAw#sM9%>bB(UrnDL)s8
z7wh&LvO}iPmM=UWacULC2K39kvNXI)spsoYmo4|Q`~JS@D`hT!e%GZr=47cxTV9gM
z5**hGiwEm3^-Z!awCRhIgG*N)p8|g~x6=ek7WIGa()?~C4*U1#Tta(-0jV9(?`Hhz
z>e6)%d((_6j~#&=Zi=-MV&OxH{k%e-bEzn(q)pA`w@7@^5Jnwfe>BB`Z8}si{GAG3
zC#+b%b3J=s-6ebF!}Phd8!zT5?Xp6D627gp0!HNMucZ^xDm+pQ^4ZZ3)*4tRK`u7m
z8VP#iNN*Tj%=WpOd&q?eFlpVEe@-dB(QE17kVX}Nki)Zv>U;!aW!z6<=>jdRt<)i2
zYkI6JuhE;{l9Z338twaE4L{hjwEnp-kJsR~$;-!Kqs;c;--}s1XW&iV7x!)hmlB}A
zfu1b*Uh#<+F=nHa@BVbl&KheE<;)Jx;j}IRj!)31B5?-(6MAD{crr0P?FiISERx@d
z=KdkTl9iR+IJvmm$8VpZ4G@9uD+t2eOG#b$JjZ0uQCk7kMVxVF^=|S^vmV@CJNbWP
z0<>_h$gH5@%>%ioJMWNCwkmLz7Ufc#8Af40<kDNbVT&xzhE6)psx}(tN2h{cTjHRr
z_K%l)bS`J*@cB@XVfwwU=q^fFn)lhZ`g8r0F4!G@T@oitL;mE!=jCOWKe0^ONtJRJ
zSB%7>Ry(JA!;&d`*V5ra@AvbDS@($k94KEBaogX);de%LP{!Vo`XIdYgpE)f4Xwd*
z*ErB+Uv)AExKx2dpP#!kBHchN6nf@3IMDP!7wcn39|~Ia)Q5<>_r`rYx0~Q<zpl?o
z7&=&1n%xb}A^V7*kIPZ(b@kHL)N~wk?*n&XfFJ8u6ZrElP5;E!S}UaWI(~F^6LO()
zU!2>tM%a-%`DPrgD48is)$T{k&J4^<R=4J!Gdb0->J{>kp$Qa=O&*s$fjEO@z|_&%
zIRFi%!3KqgN{tt>!f;C8edYqafCCrT<=#$;#m#2OzjV~1x4(=GWQ`xrr$LNo&gATB
zSJFqP_a;*0SI($r_ZF_%ZZjp4`S9bj|5hitvUt%gTXC_u?Xxa62QG$ojEg5fGYLAz
zf?k=%I|ss)8k*Ug2tjP2r!}W4OamJmcu~fTVS=|p;k<iId^7oWsqgKhhV%P`a8W3d
zAW7QBgNu321|TBY_WNUi?h#yHUi@RY2-qQxh<2M9FZLK?4GiID{NQyzGzdn~^;Nxt
zrG9(EPbLic!O}g?*548-Ja}NVy0Wmn_&U1AnP{e(`Y!*^P7T-X+s|Y>J8sSs{+~$&
zJ_(<azi(~0eybbyl#7m-@VU5!r#m#5=`026wE&ZoEmfDAzfn6Q1Y^21m7QPq6}h;x
z-%3<SE%%lZu7M+)mv^x6oJL=KjM4&ud9+LI6tF2U&XKnUmYYC!tUM<91F%b^n|lpA
zz|I$iAG)f!&{t+z2D^t2x=(6RFtT*O+Ol5mj?t&xDMBVu_qoz~iv>M@9thuRKF}4U
z=g0pWqZM@M(5Kf*0Ab4-sn10F0lZY$Ds27t7st0Y!m^pVMMa4Q_Q3ZLv}9i841bA_
z|5RkXNK*@l<&L1m&ki^z4S+q8v?px8bPpKzErE8<$`s5J(9K=~o+83X5h&(kgiV5>
zMZhAGg0!{C0B;z<B$$V8lL1Zo0PxWUDsA;ZUks26mXd56^noqW%F4<^BBEb9uCwhx
zWZ!(Oa8kw|4V&y2_(#3%p@fHUc;n^`U|b8`UO56G?Ra4m08<KF{~7*i<9%#c@&X<7
z4)}roKP}KZ${c(L$b3ZnE{t!2Anf*z&oJYjV2#Aif(#W8KQ(}{=e`3$bt^DEK4l*r
zlw+owHIw+WIovFP-j7%s{uM9}@s~9xXdbh4<XejTbYy4w6ox7WrtVDdUtYfoAVBnP
zs&8*DHDDI_>FMczHWbcQf<bE{b9|ea+eX`O8eepAR87x#t`2fXvBAN>pa#leI6`N>
zV1{83!1zf9B$%jxCjtBeOqbe%QTEtL5KxcD=}s5WmP4O$y*jfGrxfl39~}(MR^|C>
zE8l8GceRDGH2nI2f`LBcb93cRz@*s*$oikc!#l>uxo;b0X0%bkih;=?^qq1*u?EZZ
z4!l441R(v(8ml+9&BW6P>c*sdFeNdAqGhVU`C3$5TDF;g$#Jxy9XFzN)>HrNwcA-*
z&GnAGx)1(i2aG-3f{p{0rD(hVZj|jj<;x#y?1O@g{Qz8~dTd*V{%5DWAGy?Su|;HG
zPDW-GbSgSWM`OatF_`aF0_Rw0tv5*=c+*bSdvXI0j(BNWMMXtmBE|Bnw>R6gKP4+C
zhtT~B5NuuoFIn*Jfn7!-3V47p(gYYhz<rMpY`pcAuIQDuHGtUv>g!tr|5JDLo5Ae&
zKp%o7fr3E*+(r5U{YpYYqS@%>m!u@Aw{Pz$eV48_U)yPef^iYFKi~zAAY0e#Ml`=)
z0IJ^zLB)lpI~3l31{K%ehsBG6g?as*47BHRRi`-7Tf{P(op?l(^9s0|e=wBNk~qk0
zJyFv-=(AjW7>DtQcxybQfVGWMYz(ys6+7pB!RSzh!u#yxh<~>$<fsxbR*uD_{*Jf_
zYA5g_|Mgk&O=^sPa;E=kSBxUyD-Qr~F~A%HWFivN^$+Vme6Br^biq6vGRnZ8KTf(H
zn=}Yh0=oqy;DdJ#9|gk(z%K)IJlxk3Oc7QGHi_4sI*mTpXMn6~OFy<Z-$>~TP6?q|
z`@+J)E1-Q5&u%K4*Uv^n0|fx;G;*o!ybxeH9q*TymuCWcjTiVOD5t@}!HeB2jXnL@
zeOMUW!0NyeED7*uSFd$p0_M$BFJ9!f2^2mm&+ROD`3qAQ?1529nM#tG!3^GKoBC2@
z(tHnuMe|&-kAvEa^i|MK=1D7qF8m+Be(`VO0cMEu-C3;)sHwF*Te1J&mFv?MGdrEF
zNq!Rk+RL@}U+)+_DlFYgVK8HWfFoUjp(pNHVbI`0N^x5XMg}4&@OiLb0NF@8t;#bK
zjneC(FNpN`?pMGjCTGB4_2i}+D5Ej7Yv9>}9rl-z_{zSL2lV3ayl?yT2nD@t+#T}p
zI0EMDcS`}K6FpR8WLbr{$7d{6!2Pi(-FGNG0f_SHpTvW(+!m(mVR!%b+kOA@^-+yl
z8SVF))CY7c!|qrE0}ndJgQ$D}&GMVHKRYo}-PJ&3h|FJ?YEUEM>V(w^<!z#N`)B8Q
zE0bG(ZAdME37oG)1A~c(NvPa(ozVDBMn=Yw7TT37*X9Xv1=yR5Vcc^sIEuTnbc@`;
zEYH)3S>V?uVhfsjFK}t40n0Ge7>q@ICg+dsOXgPsQlT$dtX#l!lP~hRYSebX>T7>g
zeBSRsB**K|pg5QlDs3?!IB`=!Eyt*{1{%LV6@Fy@&NO30^xtPT8j6z-U0kV|c~LaS
zeTij2?}$a(i+mxI00H*AYwaKQe@$+yoAv5%_wl|$+051PSbNo16!Mu~98j*slkFrs
z^HgmVMr#vuSdL)=!}Zg>St%<kMj;^~q$^+;ha{aMaQ@vpBQx_3?2~7Gq}JCZf;;6v
zQR|NjBqt}=2WFH;@{|yM;=qab9!M*PAOIW<NwKW1YeZF5@%SQR<i<P$Hj=);DuIA^
z?E={rG=cF6{_ebYIJyef_0F>OZLeNaF|cEJqM=#ArX>3$mA`9yGo#jDn3wqb-_>*%
zQr5Tx_v>~YIsW*uDyR0@Z9wBFf;tZVuw~#z^WeD$o`%^b%h~+9wB?2)*Kitkww#nc
zm#{7|DSktj#{FobmRGrD6;2Ws2!@9akGfgPEk@~S5uTkd&UcB?lg_sm)0r779abfQ
zj*-b=S*T*Ii#6cP{RYmR&A=o8^zMxTt+3}CZ#1A|;y!xxx)ln<k`f4vivv0t>6IMC
zgqkIU#u(AxzG*-45pAnenPT@v5p-*9ZRPbmEgWqGx#!c->0(ICR>$pD;&^8FEH*`n
z5^L<gxRNn6Vomrl)D?>-3jGW58j6B%>gWk?;_PUjM_0@sP@F;nnz45lz)}No*)ab7
zy7B>HC}(%N675FkW0AoB7kN6);?`w%q|+JH(Ah#JASe4mmv90T;0cNJ(|WpM$Uv30
z-1GqT-aU1891sICM2w6~%QqMF?M7OVY>)3^{RyOrI|G>H4+jT(rz&UWe$XvL{9H}h
z;FKM-&hCG{Tu-ULh!Q_cX4TEp>R>7G?OmqH8{x$H*)r`TMl7(OrYddwfsse>=g%vk
zwTQ9+awn716&QFxQU}JgU>uJCnD*eu(ARsyQJ*}0nxVXW3E#(Q0=oq`x8Q-<1qS<6
zKuAR1IuH2xab{$WiiRctq~p6=OOSQ80tFjG^d!3*`0$^m2LXA?;g5&ey6P#R7x#}!
zKQ7K+zq?l78)cn5#{2u+BJtU?pD4~aOLCY)1a8UVf0y<R2XmZlkePAJoQzRVm@!lu
zahVT2LjpEbtpHB<gF-WJSNHb%K#g+;XfT?r{*KAY&R(5Q1y)pfkWx-O@Ff97f^na5
z$7f<j;5h_}p!ck~)>tyMxS&ok0G96UYpUwS`}J@NByxV26(EI1#m<f^e|cB0&g~1>
zSA=GZ3;X;1fd7yE3h34#Mkr~cABZXd7Hkd}Zvj0^koeyL!wabxard@D=^=km<-CPC
zta|uESDLEotW+!=qh}eHT>sHe=mW?4Vi!|oR=U&oqs!Q^qy8L7rq!H^0sfKIrvOpH
z=Jy%%D?w%xMegl1Iv=;o#enT21`?Rn3<kcmUzYVecR81qmg4zcnSpx>i^B{^)xW<<
zfr%V2E@$!uj)sDQ&s$r?YFl@R3L(ZSK$2NW52B!V>N#mSbe4)`q<Z->3Vh4QWMp2$
zXP}NLwHVD;T$k5nV+id*Tu&5=vP_K&NW)dnaN`N7nHai{)c<{(`=QmNM@D}_=titI
z&@Q7FgIi0_GWg!8*(##Zy|TgvwDEqjAi2IwHo6&3>wB&Oyil6rabsV5d-khCN04RO
z!^i)B(OMH65fhc_V8%34Jc9Pg6Qm;5(ArvlAn#@fq`3e?X>n&qd&><FRuSIgfSt_>
zsI*{Q(jQE20!o6w$%qmVT)O=p3q1g=SKv0ITB6$v_A_X`h~*3q7O1fxgwF;u#Q|xV
zZrD9Wyyv%~KD*nAhxZM~-Vz?%98rbp(^0x|J5x*8x|h}wHSFyfPcc#RC-nw}mu|l$
z0#XCNdb*nXTm5OBx9tGL!BMO<eKhm0dd>YVKzeqtl0QBPH7szSLXG&#Tb894i5@wj
zOh%;52ey}iQ$=@)4lY_ptSOZxhza)qb&dI0=pXa+8RS<n^ocQj^_y&tA#hAL-V^oo
z{_-OO;g1DMmYR`6Kqv=H)cc^LSf+`+*2cJbn2!l)8Y*}Jh)G-;sn~<d{YJu#YlR`m
z2Yk4<?&iw;C2l18Yrku_(bWv^jTWoqFG_4~e!Tj_jxly4<|1p8Zm4^GB~`;=`5Vnp
z3~_1dcKml9Ge{iloQ;~cafXehA=w*y-(LW_seAawL!17TS+?q;MyXK=H}TZ+4d7j-
zmA&`IrlQxV#&a7U8rm2yp@e|LHABP!V)smHa^=eHCURDjP~fofE^NQoO|#^+<MK!K
zY2YZ*%4u7yl}lM_hMPQ0`ikkoHPDK|SL2t8?Zg$C7Bp;L%Kk+#M|RkB2#<Eo700nN
zqFuFslSTYrU7TMHSd31@v#HFPQkD1PA68Nqz!8f945T42^u!OGqd<LHR(1}T7ry?3
z3sk;ElHh?S?Sy(pAa;S_l8&iOM94$|bZD{D?ING@tNe4nhDPM3WpB?+y2areaJ5Y9
z&&so~uauKHE5@puypMVTm^5Zf$gXZp@g4^%tkURjmo2(moVSTD`Ywq+iya^TT|Nlx
zBZX{@`ledLPHdO`IjmQp{G2E^#R222JuGX;0#^sEA3uJyJKfO$qqE=w^&N~!3mi1v
z9g`IYk5tvg&`QzP+fIBF65!osj8n@0fU9#+qctXug0VquI8Q3heZ^m60}B-}SZUE-
z%Uou-mJYsxY`@BBy(T9oJ>7*&kD`X)+5cg+_Z8~Z!J_~tzN=kc$2GhcJ2z_JOo3ua
z4wUvFjBo+>Bu##Duog?S6wJ(wUM|Ql(h?p$5br*KuiK#~CYZkR6YZC2D~caA0C%i}
z2w@sQ1`M&0!1NpWsWC9lci^MM!Cc{0XTk)EfjA*?fPeo>o}+<)APuzjdgg5gGepx3
zZM5I*UeG&(fgi-A*T1)YG4g3nFyVB9`h|vAyo03mc%(JGM-U@5V=c0hT&5AR`#KbH
zEFZDgPGPv*{N~E>DdrYm3oZ?OvtELU{I}8xlG}1pHVr1hf3Zq|T*Djs0t1)4b=%si
zsukP$iBx%ouUGr^b#+I!x4@9oeKCNrFgXhGjdCk*;Y2VCkHeXkepWA6;xVx|u5xBd
z%ky05_SJUJ@NXnO;t9G`XcE+)F=nF0*1c<mC*7)YQxsrnva79<Z(wzo9KgS1Y6nN;
z$o3|#wf{@P2>go%eEsd69l+o@m8rbEyhKRAAps4z<IqNK0v`Y@82+1HFbJ!^WcK)J
z)1@`fyd0NY``-KW#R(-ShJrk1x;!(*hH_ns9xo^21z=sm{fkg*+~QNyRPW9TERLJq
z%s`Pjs8}4r|5Yk~Icz9%z$X%V{P;1_udXgIzjg<>E^c2T+Zq@c`1o-H#d}DPl(fM0
zGvfP5j%3M_`*=9g4@yW_P1L5gn)W$x3Q_p-&CUJ9LSNaFB|ki|V{z{JIHIgL#Qvn|
z!MBh-alM**qm2os#bO!+h`oyVJ^!et<gPQ#VBr7GyLa!PG(c!ZnsFZ2xjLP;MY5W2
zNCu|j$aa9Wd*#jVyk+$7-4&0I58n6a+(kNhBZ&lVLgfbSv@`{z6fUvPV+pDG*t-5N
z0a4}S704xkqRRH>6rAs)N=lE7^s8N9I96^m%Xf2iEJxwC;CD3*ZYGxCcSFEgeyIEh
zul?dvtq3xjKPz;7I-wm-tsf#qzQ&YidZ1uvh;{oL=d)=ill4vs^3er*=d>a^yx4Ew
zK2G{(<u3c*W}>orpqkn|4YUoec0rmA1Ob*338+DUL(>7qTe^CBdK#~5KrUtVI<vFP
z0fTCyFX_;QA=~+`N>c8!-3_)sfgb{RgmG~8cBc{uDZPAYf$5!M?2^ms{wC--y$=3;
z2Zhx!j1#FiuYT8Zwv^ReB5IiZY-JpV=u+wvGS4|dNolVF0S%$;a<6^WgH7|lMQ^k2
z+(OZHb%l!%YTR`LC{dnT2?ZE2c}h%744qd-5VHq@TYB24=0v7E_|%(c=NO+;*dwS-
z`|`M}CPWOm4}JuZ;l4G+B~NIJ!h()YQc2w}QfeO%^P?u$d#}<m+mK?rxH#ENeFP!+
zjS(t`3zzVOkaLH1{}_`L=T(95jlogeeB`&3;yHOKG%T{wb@%)`eWng8JsRQAH{QR5
z<lBTC;<<vL^Yi~oAkcCHWOtFlYjjQ$EIGjwJj0I9R9sxK07L-_6TN%3FFrn8JF|xC
zQX~<Jn7D*%I$i4Nn60B$+$I<#WCUTA#jENX4%VEIefqRA?wYf>ta}kq@%F?8G{#1e
z_ovL_@yv819TV&zMtn*fS3Yh_%^ms7t~Lw|!g9B3)B&^iY226vC?_|=Q?J3JwuBV4
zSxYUH;k-dWtNRf*;Mmb9h5fs#l!m+drvQ50$13ZuQht@sK+$n_ey-Ay9`JB<CPgX7
z{p=cNkQDQvk?%k~IfEtHwcKCaA`|l<(Df<s0Ok9JHlN(Q7Kvk5THey4k@1_9`;1vc
z^+tMVjR&xP605&5`AYb;6Tb56Y7I(<(KR^Vfb$Os9X2ips?=sGnFoF(Y3aCibN4Si
zGZP+l6{h-hw2lU-1qrHGko|kW7if(Xg*%wP_xw&{lKpav3mbk-f!jyxPxIk;h|5~D
zdJpVI<Cxf(_BZ>z@}afM+pWv(c&@f9#yF0NZz9<ddAHZLfdQ^9I=Zm7)8i@kS=qFw
zf%HxpZ}z;#orbD=;)>?lIb3o$6v{gX0AzkYCMTR}Ka$@+o)J=eJ32^itcsvnOJ8qk
zfiAqt#edHQz*0V_pnu_d!5?Q}`veILLf6lnK8khU{stN6Jb4>%X*Nm%?uHApazu(9
z^LNll-6|h^=55KbgbZ7};g1*AwS<C0=#8%gUCZ~KcmTaSV7(Q<0%nVvSF)3|XDeh@
z;W>fjxR!Its#dqA;PYBaHK4w7ucjJtWpuDgPpnE4fRreOg!)%l8uc${G=}nD|950;
zaKT{)z0%63_03Js8;kE2p{5k7)g;v3u3F>ZTx4<+om<U%6s*6<rv>6A>nwu(xp%HX
zLqHeZpZJR%BmQT}=H|od(hCyA;AiJDXoY`IU<WNa*|N?FuMO7cEV_zXul^joWPt{2
zz0y*s)AgV8K73@Zi}}BFh+nMead0z!GFj^~G^}`EK~_XA<teYAj*fIuY~R%@*K@#l
z)~NNwDenmXq7j~<ykl*xadd>?WIgwpyHQ7S1_9qAg=-#^shNIyzBP($n-b*VWIA_d
zKXP=$$aZ5}3(qHYJ&%V&izvC2K43Glu$0-rDfqa|txTT$-$Ud9@E*W+=hL9@%x{za
z_)ujMZ0OOj)4f=fBS+sUzCPW0F#}CLT~5iW5#3J$AQ?SJ!!mL!N%5*QXWtfqBY4D<
zgylz__s%Zu<<@0C?~Dhzgnf`C_uSvKbIwdof{Z0x?1I1_UstX&w*Rjd5K$%yMXv)8
z*J}lIU0n+flH*En1uvsBXt#~-dJ7P1FuHMGiABWP6b0t`GTx0pemsgbpK%!<n;3Cd
zyO)-aO#!_wqeq*}S-NhY0#Oi#WJG>Vku{OU7TVjJm`;S7o@<I^@Hvi8LoMr-5jBvL
zLB{{<uLE4O-?l_P9>nihy1BMsTqX&S#6q&TiAJ!r7=BVzs+9Qk*Hb9nXE)~{5zx^&
zbb{^pf6(@SZC&=>ygRaPK0>=*pz_}Ve7NLdFU9y{?5DZX&G|azhYykvZ~jA(uK^`F
zP*Ba6NUf`;p<khku7ckm#`N?L-^E1F(o~PHznM9WTCez59*N8_GbT2BuKfCvq{;Eq
zW|kzn>s}MMpA3Sz(Y=B}Gdl-|mhNs0H4IZ`(9f69MB$L<md$FeZ&*W<<ZcRJtede|
z*rpO?1<bU`G$FA6KH%mJ6s?FtlOaw|@qGzSwzcTk|M%TjdzT2XZnH-%jxp*$LaTUI
z_fSM#rqRW^tP)1!YWvuTsJZ$8Cspz%^&@F?e}3n-j-l?wfYPYuWd1UTOW8C2<{!xW
z^AC__rnAdqUgG|*+?p$#%UE1$4+VbFkMQx0cL%@~3gS`|7Y7}XI|*WElRdYo4$`Gp
z=ty#4;4NhGfquf}-o}SGEv$(zNTx4)FdR7Wv9smp$Af@ks5<xS|9jky4C1-~(1NIR
zc}MlJGz3wpF;winx>dsa>FmEif(TMVLTL8Yn0qt9HcYk+A6|d@y|lIQD=8uzTFCBi
z@WttGdd}3%B_*hV1JLq=J=_RGTg#I;w0}>ojNWeabg4}jdvEI1NcR0GT!R7L>V;LM
zBm!&Cf#T^W+7{||_}?B@`l7p|#l^&kdU*+W)KmBZ3`(_4Kj2}8FKR=B*Gqc=L<=&}
zm0JEgry@o@4I6#w*;6Yhyc6Ji{km=ySJC(xpHr$`S)m=_rjENo3t_{cxpq*>;NCew
z=Egu~;OLes?n<#eTnm2k?}KH0qYG)-V=7Q@*pSI!7RBQ;S>9qPg_WBw_%FMkb#dn{
zF6uOEpzD*X+78+!XS^uXgoOb0s-=~cw?LT#_?$jtWra$cs(>X`Je|^aR$%vxiHf0i
zOCa*;@n=wL(&A7CM%no0<t!eDMr!ul3hJ95I1m+vA@YmfymzRk?$Ks()yaefj4}0s
z1aL!|yEx+Q$hg}5B<Q@*{E@c}-!DDjiR(%((mWbvV9EG0H}v<pm$>4RW14rmTH^&*
zA${=V3$vx{<UzeA1D3a@`wgv``D0wz!qF16^$(w<-xJwqa>{y#dk;VBeh5%)5Pvj6
zo#BhZK=2B0(Tl4zT~Xzg!mApm$T`pF59lIwO&Wwnp2pvc>iO7(-FR8k=UkCcKki}r
zQu?LbSe<2zM<Yy6P<JxpRIW!6HcjzG8Y;dx>5Sm4)6?0VZWx=)CR86U)-f(qciNk&
zHejn{-;a2N{~LV{%+@(?4vAWh6;>}+gNber00upi0P`IxKnw;W^BK~wqtncYT+MR_
z5}^}FGSO95)9p3-arD-iq@RuA51E?CXV&2{;m_YiktpXCR+ogU#QZS+rY#r=)jk#y
z(&Rl9yf!_;^KSGq7Ml)Wk&|SSe3zaQRZ@VLL`yCe$t7w^CPm_*#x3Ks^+1yy2??p`
zAnNh*$!}b8C%1i*^RM|?(-b$tF(t#I7vmy<QRy<u)f&Q_yT`>9J)ROl<K9JTH^T~<
z&u<E8i)#7vE<Q-f$n;*Uc-hBKLkaNCPw1=3GXTs_#oq%M_G^GsDS-(MBba7+@-*m9
z6fOf*%*O}FRdlGX^>1{NxCW=$p#j%!wHH2ypMSE^PP>o3`E9A(*x*s+C|;d-JUEPJ
z_n5`i)%$qqo#5QHZ2iqHTodLPXBFC7o^0x3udx1I=FbH+PM&$#mMoNu2VTdHF{$^v
z+?<<o5Zp#}lb&hB{Y=FDOt*3OLk>g?^ZvsSgKnLHCH#4yjSXkRzl1yAD_fI0&!Wi(
z8^;aYm-F>}&PAiZXfi*OR)sV(GxN)&US8BXPK%LWe7e;R*#H?+>6X0vKQ6)MMikN$
z@l{Dwi>*(Pe&U}M`JzdfmS=eCz~rycBd5w2Sx)9KYk4`&HzMD3Xyu;(?cCuX@OcF{
z>CdWG^Lf#Wk3t%}G-$4`mFalY$w@fn(@ILpyYGSbATv2m47jk|v~ovNRbF4jup!#h
zEpo|Ol>Ee7kQ_QcYPv66G@ir6&?ZT43*?yEEo>muIZA2=H=H>DBbt)SC2~kGGmodX
zbap=9c7yNDhBKof+5DRAl}RQI+U)YI*@gvC4CG#3nAwf<-*;7DyoYix7rJ!uNY|sR
zXE%yW*`L<(o~odoK|w-#Rc25=`RDp)9E}7MninI^e41CpBbR;BBaHm-$w@aO#Y_xj
zI!lmqduM02LD{eV{{0$HWyJasgo*;-sCsRCd|N{c0O%Tw1nnIiTOb9J2zgcIE9X58
zZO;Iz?d;oiU%%pcUmbSltCghG)z^=L{N@chSCX=_V*qid0zU@83S&~?%IcC7M1!1!
z&EO-DR#@HLEy5<}+W@`De(1z!EgNk=MbMK2%w+-<7pat#lzheP5D-1TC-b|S@N;u=
zssIpm6jx3l9a#MSnx57IXsrk6tH6K)fc82odQiziJ)b_Xqp%sXw@Ip~(jDDrdHS?v
za21KQ^y^O!>xo*u&c0N&w^oJ7-rg!~xj!{W^O$ULaDsXb%c04i%fA$M`CPQunRUMg
zwP%DhzqU}x3I8B<kH?Avd{U%Wi#+Z|m9k`W*^jBPl*{xZki!-YT$KFxGGV6$RGRf}
z-?zAT7Z&Ip*THLn_pktwL$lH*88qGFmffSD%c)ve<OgDtd-RT%jF)pUG0EiR<t6eu
zeg|*p(hFIG>A|Qh&}8AXo>E41)N8B(vJaRA1qhu<0Q#$dALsSq1B=MMp3DstGO)=L
zIpe+_(%pb#xdo(W20=sqSm66JjUrXS%boJXm&U(2^@Lq^5T(m^^WpECSLes**w}qF
zyP$*D?{Whqm^ckPFni<K-Yj=Q5#v%U^<S60#1;;G;b(`-8l8K4d*?M=yu6dmqe!e7
z)a!kly9vS;H&f0bQqnyxwL$GNu3VP+Y7}0tunJV^4g}cj4-+|D8yYlgq)n+=OlpA)
zcAoLBCSp&{4*w7>hH{Y+>&P=c*60)cEP!h-#`+EHDt90^NSZbdDm=9Lo|EdM36!5l
zL1)Af(~U)^CQA>xR%vKxumD;R4V{!+R`Fo&DGS`;8F+Zk65&-5LKnH%WISn_Fbi{_
z^^mVo{@scz30Qw+0bvNUp)3iM20gHc2czYL^6Zy8@&KTP*py%kp8TSYFzbnD^Em2y
zj{h(c4z;TQx3Grq(g}!N4KBqZKz9FYNl6KqoN@wPewkDuK`n7Vpb-LMzeWwvSyTC8
zB$3c-q60zzgKFV`KaNl?sKW|?e&6`dD9*=0m1b2Jqz=w6Q?&{LzkdB%xD3!`@wmH9
zuL`-6=DwZ#Z(*NDG-zDpEDc3vuib%$u2-|t(}#SI6q8=jBotkomz-D}jocR$G)`13
zK0b_@VN`E?J&J}k$S>S8G#N^mJtd^L|6^zwKiiK=F>dGwN{#vTocEUC9+NW#r`URC
zCdQ&Jh;&FuJI*(M9!V(X4bZxTEiWob3emj5CwG^n>zX7a!~YJZ4Z2%f-;_h9{i>j)
zTro|^r1?z31N5(LCLVJ>2Pba~pcX1%T`XH>fw#Lw-#1<dN5^5XX_a$i$-x3t*VJrP
zOxp}2KGgv~&88n3#Q?Zb;^E;vi|%>`6dbY0c;tadWMr0C7$qrb5^Tv8<Vp_^A6D1b
zpMxI(l4&X>kgg7pvs2p~8b8(Z*{fa$mhr+UXlUJ|qbjt~5`Y%4HCdhm=Gz!0E&!2X
zrzW*;<-oU6r9f4q)7`w<?j&R$ya|5|84~K3#Mtj>b3SS%5WCW5G1uneLvvZ&%_7aA
zO#JKy<LuY?S!$L>Ue|e&QK<ztwu81?M@#BDeUN^(tqpBoa>?BCoZ_EVp3U%~A#VL4
z@sq6EkCO11XbSW%VyZC>wv6?<O8FakD*Uc_?Zf>R`(4oi@F$36MnZqaMXw!&KH>?V
z7r6RROnhu&@9aFPBqvBC!F!s6xZ6IHh!8k>=hs0As3+m6(mG{NGv^OL!+T|aKNZ?N
z1zj@TX!z<ln3{69@&Iy@G!j8LlzqqQ-Lb}2KCa_Bu6dw2j#9j70q!JmUyluK@IcxF
zs)xzWUe1|oGl&sfjZP3u8#BA_yno2zb(+0abS{PN61ThTL42nqj(0tpiFv5rILrJb
z+hR_$)-wXNf%~;;(r(~!52OndWAPEvp>-wPj3F7qP|04<p6ZH!2BW8vDPnPck<I>v
zaP)YT+v#v<^~1aZ;)Q)%-Q6!Gq7<W<Ck0h43_6HX)H>=CUUJKIV>+@Y3wo*o>`e_c
z*;I-&VaQ=RpwuL5Yyet$5<|O{^C{afzz#6=Tl<>ju=;!T3_5vCBl#{0ppZubV}Ke1
zimHo0U9RNd^+o{?WVBTO&H35cFOZ<Xe<1Wn{e*6BydpG|c?1@AM+@QvimiMQnFC}f
z6X~z5Lx+HYV0?1$LLuWL(BWCxa6lqzT9VQF)5U^eeB>N6S3<3F`g=Oy_1;h|Ec*wA
zCRF8<DWv0cO(C(Kk@06Dt1C+4M_3Uv46A2aV2^o7YBXzFa$}<Un1t;Fa<9ckyuY2&
zS9+z_QFxw<X*$bUIW!D=OAo3&-MXgXYqm9jKx7@mq@@ZP>B??r!VI5GdlP9SJb+<6
zhCoHTLD%87f`UTS%>!-6$9T#IF@unaV_#?``j37YlS<zLZu#!eaXm-}yu>KTw=f{m
z1k$sq^-Aod!ReN2x!Cf7?>JZm;hVF}XLNjgnjm2ggMZ-N+vf_h#xR&DD>3gn0FB5s
zPbDs-4pZ19hsT9l)U)Ywm8>7<UgQUkF&#!%Cr4rX4Qkp3ip4ol!^td;<U7eeU`Y8A
z4%8)IN|30?pxX@>#uUJI<>!P&{s!+C+9!!Fk<XIn&I@#wzX_L0n$tbVpD)Z*;+o=0
zkoEvD`7;R+!DF6HT2&POKg7LtRFqNoKME=aC@P8~3aFH{bfcnxLw9#~=MWDnh=_m;
z-CaY+&?(*B4Bg#G-vhqi_jlL2>#qCX{kqVl<2=u)efIwB&)#Pr{7NjPQuzKbXUkoC
z9c^vvN;U>FMT_sCj+=0+2Jfx=#QorS_}uh#e5~6s1e>0o78Vl=9xx8>&@b1A-X>Z*
z*3jtfbvK(<K6v&l9b{6dJ=y{_6qRE^af?9+0v_}Vh}LKYU5opnQf&}xPr^+<uoW^o
zXC<&+J+K!rdXxG!h5Wl5mC#wLJ1s58@R8VO8CJJE9-9&~*fa`SAoxz~tZExCR}zCm
z6+-9Q?`-4rhohl43u_bC{0j`0y%0$uL~u@Ip-Y>o)-fnwCW1IvA4>(}O9%0GYB^+3
z-C~ADbyd=;g?$mI#^u$e@&MGS4Iw)NDG5$hmjPx((*1X*lAyw9j&LY_G^!_JuM1nB
zlqSes0Rikc5}<+=5O8hXkAeUXz6t~#zLwX7opBdKVEoU$=ghcV!JUfaDw5Gm_RXq=
zS~7m@3>VH`2!u@10w$u`q;}CrLz)TZ!)6{rcb^x+yc~*W*FZ;ty8kakhyFk#No+tz
z#D)J@?0D6C1N)2YxcuP$jm}#*@vRk9Dn*&#D%*rdl;VJ6{|JF(!5lIeh&Is%9f2@E
z1MGMy(f{$&r=jZyAhc=>;$NOj4sK|xCL%-<ST7ie*R}Prh99<cbkH@8fjc#q!e*k<
zy059jkdpCVK_OBVa!Sh8;XEDXa{DY$uqhc_YD4qt)d!FPBoj<6hh1<3jPZa}h#8k~
zbfn@OYTf5%YHBl!xoO||Xo|uUzU(3!^hZZ~{W;F%sHWaJ?2}|$<C`q4P3VpCEMifS
z_*f@MZ1%4(Fg>Bdbu&*Gr|mQ&jTlL}tb7{5Wrs|H{)h?T+_@tK6lpL5kpRLA%K#wd
zkC^5wXMw}#k)S}G%$waOpWeTB2i3fxdoXE`j|Hx5s>l=@?d|Kco2kE335xj+6<Qij
zRFo$DYO)1G2G1grK=f@TIaIy1C_eh(NZv2HP1+|uKpK*SgQ?NKmN5vLkCv8xeM4}v
zJ9jZ#6wDb7ma6h8TfMT0k+$z7g(cJVZr_OZ|0!+h9XV-Bayh!+(dCd=XroKX`$S@+
zcwUirvNPcR+5K)7C7v^z>Sq1EZL?jJ{qf6Wil+3#xX}K5oy|xETl$T%RiI1_VqGa<
zSLSqiS;=tM^<=%+xIZo6{QSfMXx{yZF#w*@LDGeBYY>&S6;(ef!HLO!O);9sNm|2Y
z+w=HfJ#v=wBjknK!DWsvLjky&I*_42LrY6LRtk!o0nJ;7oi!0z8$THgY7H9#p;pe-
z{34&N3S2^QPHRVu21$8^o6ciI!(>Wk){OI~`5^<M6ULJ)h@`z=%07SYvt16c=3T4T
zV1MLAu`2mDep}nw!wBv&r2Xk}k#-kJ$wEaxl+?ymWf${AVkr&(Vl|_Gjswj;r^mb3
zKPHYmRR#%6Lmr`k{+EsKTwGnv067W9V4W}VUb}t)IBJ{&jET^ty&1{ZBR?x>31==X
zDOuX>dN%}!4i?8PO^|QWY|56Q-SG2rhZTZiYXDssL|ZV0?}3oI@e0(Rj=VforK&1h
z<>GXYkmWOoLG}Y*W4hjXmH=0&wj5kog@!K9xGbmM7mWe$CxOK6?<*5d77W!a)0X#_
zp2N@2EryV!WN2TKZbs|(xe6(j@S0z0AMgWIw>X&ul2o}VE@vXHIZZkvkHV!`Ktsaq
zL&g&*)$CnmCItSVWtVv8bQ~<D+^a{fUiLDQP6tb}^d~FJ(#Rd-!3|<c1-C2^jGa3g
zpKl9YgpYyz)E$=Z<>lou(76-u#p&*ddE2;d78s&a%!%Nx{l+-wP^DS61H{B?)cXn4
z0IQyMLd>nKAX8)D)I)LOOU`S<m8Yg4PXZeU2Wn#+4`w1{EH>=n2R>avr3~hq?-2uO
z0ajbZ(PySPs*{EK`_J2LTB@90%OuHoW8cG;rh3PHa&%=Dum#?0=EgY9J{OBYiMZbM
zmrq`~r<<R$((Cz6xWZQb{L2U9zU1l{Xz`XSkAXx0j)M>4qZh6bD4de7NINph9%l&U
znqYy5V*-4CC3t^QU0odx&<ucgbVNLkl9yr##!L<aI%EjsPtMQI>S#1z{0w$hMs<%(
zUc906{e2t~tSZW0^@dV}i7;&v&9o?iQ)nV7!50(rbBr+XVegZ|r<=XdLW}Xq<sx9b
zK~T%5XLx<DOa#A4BRim!dXF~i!!5*!i%d~6(=Sj?O)Q*A;M>i6(x8{~VAcTjlcicJ
zeTihrd`aG^>lci)1iz-V&ze#L`(b8wmRxVG!~w+TD}lNv1gU<F{v<hI$E3mce+}4z
zN;*f%%D7X5-AQm(RHm8;S3nqk@50!DNbdrH++A!)+Fzc|hNl6z_!p(+<w5#^Rn(f$
zR+y~O0cUBLC_KLjjrIHr&rdxDfLMLP&7+(%S2+s~g5q*0W4Y4O(iUB&mcBk2kdG*1
zYg<_6u$gTT<J35!_xh<lRs~SIZCo*oI^R#va?ruuLqmrV>IA@&p$xb%8h*FvLcQhw
z_c}RZ;QU+>zpF998apr_A6IZLX_c1GPc%8mUQ|&}uLeFOTVxj%FMwGPA!hnml2jB1
zkPi^mt5=riO#(<=1->}$JrX`uph7}@e0+pMXwyNS%`!lxLze~1D^)y3@V9`TVO=c&
ztIHYF2eK+MXX#X+qG4ttcWtCtWf|OT>RUc}YTMonpOFIb$sDo7@AGn@pTYyAYFsAo
zn+}F}dAo-cCjH%{=qUnOmeUHiz6OQzt$=M8dMLupkp3aKHDaq?$0_6f)k=Re?2h}w
zcm7xBQj&m^{qxtae&7iYpiC5$A3w1IbCW6&+X-<U?Ci8|zi@DHC@s+yeMC%b-ersN
zGcWJ@#*hr}MhG3799%1QuZMA&-@22sDMh$XqLsy^*p^R;*TN*-i1hpX@~uuHu|IqB
z&5Ls@Gq1%sDSSsLiGH+kE{PQhS9^;1Zw3`LAnhBlkB{~h=GUjSCZhp41AbXPD39<P
zhD<NWUJv2O19g@upTP@u--4<rg)7h6JETS_LeP~@EflJJc?SQ0T%eOXKwMKzZd>kc
z`DCOKrCR_@u$azG-0m>UGiLeug`T=-ZlrKH6MXKQH&@mE*Fsi#-~#{#l6*Yrv58<w
z#jdnqGfg_6i;gFmCE6H5-|1swK$l(NxUQCr`IySzp<1R=c`^kcIg>%;0GAbqD%T6B
zgYyIVwf^#bGs1xjKDoQpYH?1=Lm7#c8g9rg=tNC}K`F(|3)et}aczD&K0E9As;8T4
ze}#O<wMoj$cVoE!;jh7oYo?mtOm+Bvw|C5+FX5X(=TiNVr9gcnH3XIvUpQ1cS~&hl
zIWO|LPe#V~{SuWr_;^;q)5OBXzgE${KY~)r%c*bNqRLd3b5f)|n0)o`#r!k5?uVcj
zQ`-va!~(D+PFxZQ3IIsb5EukeExOVjzaI?Ty+gHE+8TUUmg*fza&ii+b%~@CH~#AR
zxHKEZ8i?CzN7pM#sOAc8fX|%*Po$Nv{vS|30s2GO6F?;h4O~e&__&#}{F*w*i2Uy-
zpAj2oGi%rwvvC4f+Z23<;4vn{f8Ox_`B&)w<r_g61vnV)D&8IXGJv=06cT7NMj=I8
z-eY0iXl~H8wl}(JpvL{SBjyyvu=7sRxy7`Fa<1DqaiLG@HY%3hcxB~!HoR5}?5x@9
z9WeYWb&Wy#dJ4;(-<Fe6s_SBM@p#2Q8Ok3ueWr1DKj`Z?vmdF9?X+WImFlV2<?mzk
zguG0Qn=tvAOSA&ZReNvmUQ@?-Lq_?hZeTWe2(|e4PD_*Aih}rxv@*7B`9b}#tWkr&
z@(FW9rLiv?2~QH1MMcUeJ`G2z<<;Yw1g7~nF+5+g@u{PmDk2_ku<b#C`YebIhEU$|
zU^0k(d4j1|!S_Ed*9zMLzS<AJ(j(7*6Z`SiW#_l*=cu?WR|@-C-V<u5>~z)LqNr%^
zHDOK0O}6EFEq&h)(kN;#x^mPoFgRaMVPV`-sdk86$X2Hb&i|}~gk`5xXqY1QCaI;d
zB(lb#Jeg|qaaBXRe-kuIdig-)TLbmR+PYWIDjjy)&;o>r9_a8Jrwx(fv#VjeF$c3=
z^93(Zrciga=-2Nene?ys`>7?dZTpUSPaZ#fzFKnaN?zXQ0nx|6Eu3AFBBw~zHO<t|
zn<XD4ONY4wE{JV|dn)(%YX4!p-<Edo!u^N<XD@<&|DIZ7!fq#L9T*{h{ESV{R695(
zgd%caWzKj?bZ`%<x-eQ^&VMp>WVyoc*!V%|{s|8CEIo{i=wZC<dB=q9AkFjf?cvGM
zOa(JrL$P9NIUW*oz3_KWrDbIFr1_3B5{FCSk7V7%-y<@YmxUocn{jQrqnn!-Ic{{H
zVK6B;9Ih1_WME`uWIA#8vCWMNte}vP6kr4mLZvj#jf^_!<C|f{%E96_J~RL>%l+19
znPJv=N{b7yv>ukFs&<0(5&J3&&q1>;YD9Pb2m@whB(NyP9pxYU=yyk>v$x9om3tEx
z8bop6rOmn!H5YVSR6>Gs3hO85wTWvxJ3G>vnsEhfZ*pxVBn(0&b%%9jwD}VN)K=}#
zHwe8>DGU>nQZ{akWO48j0QqqwKr!fnik3CTF=>xs(hXCZPTLj-DI((HA!NXya+l(9
z$bc~o+SCpV(E3C+wBXl<_xAP$Bcf;}B<{;YTN5*Qfr$J+cy5BEuNb935lU;D`E@%h
z{i@oN3y^u2DRFfC_wVEM$yaDJdVYrAOFWKaR)YhBL$U+=)w5g93?c{T0WKC%Wa;t?
zVOQdTadEZkkEM}8gmM|w(GnrJjrcBcHZ8qhION`_3GdS=Sj#$HeKjr&0z!>(>v?~&
zO=B&4Dh^*Wdn%xdY?oD&jvcOK$UVW+{VB$awe1ZDY8H;qqc2G5Dt=#J%BuyP|B{W1
z&52p<Yum=Ul3sW1Z<k+f2UtSGAI_)KSC!tBocz$K+EFb?^pRh)E|bymZdc3XSb6u#
zWwZhVqfrO~0_qtDtncw!=|k&0*zwd4mgygl!x1M~dFj1{v%Mw2Cadh%cE1ApWH(Zd
z{_456ZxjrkVqi23)Npg9n`*}0`~^w$h`dgk50KWm_;nF%-^H=4{j+cuQAjQ*T-4fD
zr^YOJt^gtv_U7AHFQ|C{wou#ybPq%(dQ0CW`@VkHz{QjEloBz<-jp!YMW-?qmJ*?-
zzC@PQmrQ)T^wm?EpKerONHzl&pVh5d)A#w`-}0N|Sy^7VtK$tmK#@_aFiGCakd<!|
z^qpHax3W+F*Oum$lC*NL=>vw^Rw(eTLFm8T{__qSr1U>p`y0@WSKr6L`2X>*m;3+!
z$L9VYtwH|_WB$K)0*wEEbNE>Q^zZ*(C!qh`;s3*b|9iv7fc{rkALBpc|G#&_+Rh%+
z@A0EYiQuS*;bkbU*di93oSNU<%mKG_WS9(PBh|`_3Pk>CPz($&DmKi`WRXx7aA46?
z!~e&R?w$EI6eyz2YB8Y-D726fQ+s~T|7eye#RiHg)^Jf{z_@^`!{d+;un;d_z6UXD
zVHm9ElC;qJa|0X^gI!X*#4cMO3_|}0ncCB^@jAM?y6%IM4@huB6tz-*5^x(tDFVmQ
z=mBsPx$<<U4Uq&+Hqr3%qFV<3={OAcv!7x$&ZHoN3jx?!HW0TOX&LxWcihh)^Kkbe
zW0XD-;1js>@c#3;<uC2NZ-;^ho0AMM>^pY?Qd1Z0y2JoF06d9b@g>K5QO(7|Mi^;&
z2?uhOFOFh%O2$V`6<kluW+q#*=vTpYtF?6NE1Ojvjl0wGp+{w!sL?s=Hrl`%->otX
z7~UrN&&ubO7sZg;zJ5a5JeI=H%{a$0@@o?yq%0Qp_5%Wt`ahn>RD+yd-xr)aWsCJq
zO*GWht*`buzd7?K6o|aBmX?tYikMlWmXu8*qvS5X@%W*4h5F3Ty+RgH35B!#?qkdr
z7_5^XQ{eVH%9k+U9{`%!&i>h(Pc27|7qMes84Y0K;*dVW>$otQ1C#r}{hDxEBoYu3
zS_P-1r9ki^O6`B<gX2@4daX&p$kpPE6;;>H0si<U2JF)ZSuv0xHK^my8<M(udUiGd
zm~B;yNkB7#l^Gu&e<8Obw)kfeJ4XwptY>W%&@$pl@Lm~S#<s>CH&E`dY6{H>s0oY8
zO1^hThR`KuixFlBk1%!8=hbAgsmRBE|KrG~D*m}Yzq;6DY?_pdgorL<0Lk=)yg$xb
z&&~)h1tgE)#ymX8Rd=E<33Jpd8PIdB@70|-vh*o{<tt^K5__7s8B85vOo)sVn(v*z
z<K^XDb+}nDqMD@s9IU184CFu%r$|$TnbV=)2ktB=v|m@glv{?1E&dpd?mvJ2?18h6
z^IYbDBb{4brn>ZeLR@5$ogF#KJzm#=`_~`g%`YraC=`s0%s*CwQBfUF{+Zl6&*c31
z!?$BQ>M37r_5#XyWxe+LK)h64q-y+`w6si6NW}KGiBDu?J1zw0i}J2WE7m`c%L$>C
zGgj^-@ntG6xlC#xWZ!!A(QB>&wiC=oWZLdzA|fHt1*b15i_o(w;3$2~Ap<5-8WA|Z
zF-`wSs*{e6k%}rKHThPx!sHGf-Xp->P_qwdpBxY(+S$Q^|Nb54;c)|o2n|&TI40`V
z&U>6*^UNKZ#czv!A3nW-l*{M`%rLlFAmS-=4el+8SgB1b3*(FQJ|@`81LY5Bq`Wi=
z4|AuWWSO5o;hc{oJN7<tPs-UX+a@0!){<4CW8J?p*6drt4nkqo2MqVlr#?1NlD7<L
zOT}`@4ZJG+O8xh!m2BBmwL(6cC(F~zZ{=j329#*rR~92;6#Z5{F}>FN5Ab0=E_Xj%
zv1F&E6$J?ncEi*>Q}TO<)1+lO{R9p5O&P9+KPYTXQ{l-X`rJ7Z9meoPHJXkQ=yPI7
zPS3A*AuG+LHplDh>~mVBtO^xkD!HFYQ_f?_j2!y}3DL;#YrF+HJ5IUgbZRR>3HR9{
z^S+bkMN8W-l4V=nBz32)FOh0YS6bk0m^LbNK)=8U9DoOL_Fz!fHcdgpHQ4fLZk^{6
z!9ILl4D`kXy!IG1S;_gmJRA@g`#EYKS8pCDe2bG-uCytRIb90O^nW`c9AY|>Vvy=e
z?NFdp0Jj^uMKrcRu%W+Gpa|_LkA*1r=cKyOJ*hnJ5baPdJ*JSqd)0$qX~Q=vz|yP?
zWXOSPoFK&F2O95dmZ<JtZvK4P)TV<ejScaMV(8(1PnYYa)H4zb*K*U+Cl(XZ&B0IT
zK==o4+N-N%#&)*Z>9?X(X&jTY4%Sj*vQLfzCgFPOodD5m&{y#PoA_z`$S3ze)@jem
zz_ScAIE<48=<@~m1}KP-iH%P^p4p4>anyCK#7!|znw*!HHQB_=#e!U>vq>nBdoue3
zIv3MPg|f^$A|{x)i-A9$!{y8<%!^d$aW6ydxVj!9gIUr(sC5jN3X!zOWNd%D`IJBS
zAO48rc*92Hm;<QAAdvpDdSTw+?{ocVA010W2J2$IAE<%pymXK$C3o+uu+dQ7?5NXT
z7-<_#?`0YR*>W%&9X3zn@L-1?Vh^2<pT86F57SvonaXc{_wX>R{fgB|fBg-c+gOPb
zqQ6`pb<JN3(VwBux_;Lm>QC^E?9VmEza^(&8-4r!P|Pf)EbrU*`y!##TO6;d_>9%F
zf3d~z<`4<_QD-VojPYD8jk}6ao@y;l4WE3Z*sZQ8skl@x<c+^$tnxnAdSoy!@mRZ0
zMgkM@eFkK(j@FPo=gjt|jj3S9t9iya_jP~zo~<WzHYg+p%4V8SxA(_)(fhUA(_`<i
zcjIY-4q3|w|L9_}WmVZAw9db^neo1SMtx%gk@UBprM&5<1T2`$$+={VH{_wClP;fq
zI(SKJ#B12j_Fkbg?Nuiwc^+<EYi{gACjd3c_4H{@J<AFIL?UA-8EjrS#1`~W{RpBt
z+%*)<MC%Z>!Pl;>6aO=|k0pSkVclQk#+#g!YBJE$XYae))H91nvcI>&8HY^B(Hfuz
z4RuGz7_2KdM4A^Ewd4g}sx7(bQQXW2SFJy$yn4rOQ3xXGVyq~mMkmVOXQ=Cz|7ALt
zS_F%|tdT%OITRUL#%hw1RLJ65V50G*DZKdNH$4Rf65Q2JJv}>HgMBsp_g28Ad?|D`
zRxHyH*L!@nWJSkmpYUF%q5@ut*Vnr@$Lr8?(fVDh_^Z^}i7;Y9sTdY;$NQ-2Tw3_2
zx=O(Z#9r^&A!p%LL<U2$;sKbJ0FdkCK;jI^BKDg9yt92nO`$P#VKFA|TI?Jvd~&jy
zd~$*>=q(_xh(Z!N3Q-D>ks&X-s;Q{xc)NyD0RL>;gGc1!EqA!k(O+t%_(~r8-<I_Q
z%anb|9FfBJC2ghi&0{%*XvCw+rdf^TGt^Zo)w}3a$k`PN7OJ@9lTXjgA%2uVTTp7O
zY#mZ=Z)mxoY}U((%Z78~`_Lx}vXvOlYQvss9i%)svp~2%87WD1)yi8YKtm1fpU-||
z$!tRZ1UFL@E4^DikA?b1WekG!QR+78qSpEP8zVWn%gp-|-|2St_xdY_)a6&<c2TmW
zHcrltOGCvP3cJdz5b9_gYvANmr7|BLjVmZ%z<sDFnh7RQ$&T+rkVYa(9xWWN%yUkM
zb$rYfZ++FgCRuhTJ;wW<1!K0=M6C;{0%cZ~W1nZoM^g{aT(*9kqUn%xn+zHbKCkic
zY{|ITX!guhs$)`o39h)XQurdZOoB%U^j1B`Z6C_wJJO5i8_ty@F&#!)^YLMC?*_nn
zM@k%{%WRbpy7pE*-Q{J=l}#kUIjN~OKDVxV@clgy2MOYpeyK~o78uSxhSfonj%@Vk
zL*f8CXV^OpJBta?>s<Ddi;cdzOyN<AW<Aqdc}r*UA6eEWoc9_mKoI%Zchu$-^5|?l
z>&uT9JleS=4DT&Z$X%vAD$@O%XhJa2P7CYP@mjnagP@Qj4cb@^@R9Tlly6Etxt~^4
z+_uXhWqI@20+%q2{nkBMYhz@mfIy2e>XH@ULDO~8|0ClsSx)zBj9`a|dgEOVzd(fX
zg#$<cg0R8Sp?|bm906eK+#_rB3zj$QgbFRs1kE^YKZCRkJ(RK>+gnLrq<jyWE;2)4
zUD;X--<B;D4{xvMRD;E$Lpi$CQR)t?#hNv8esMinpVipVF^71iJNHXL8!h6s4#V=0
zZ)BnAm_1N%3hnb8SjR_5Y0KO)7PCyqydx6n;qU)1e-mT)3t&*NM>NV0R~i^nOhC1u
z5%V#iZO@1E<u*Ga62Hr37@j76>rXb)p_EnHOG-*!_P+ma@UTsR`mFV)TR1{S20u0^
z32T+zu=e;KQnIYD>cvqe7n|`?lCe+7QVB=pNu8<dP8&aYYKnfj>-q-7d%V##zjLtF
zZR5HzuzR?#rStA`>fgS(Vp`cE#VlMe64fvCQvku#^8Q|$iVP7yO<F>1jwqtDzQJU_
zrSDj46VvRSwJf@+3Tfaxnc+b4oEG4*QimFEan;)jKGR*)Oa}<Ebdh<<{d74oppT`V
zSuixDrmQHolS5}#Yb8}^lt$|7*Zup+q5G43m3tGj{#-yzN3G`4;J(f>ly#(Y(<{gi
z_g0oRjOAogW#g{#jN=D~C7~ZaMNM@}b&NPjT6COB*~n)NNaR!}5Wbp^R^z)I#mm@B
zn&&qU^SaC>5K%YJj_6;u=T)sPYQHe9eK5D44CPMs686xF+8mPfZjsZlIGuNKJ;hJn
zREJN8jF)Tx3$Vg>Kti6JlqRTo$91BzKskuUri1fK3^8(QF~c6e&BQii0S?tYsjSxW
zxy<7NbL@IBew$KmQ*{iNNZu=efr^g`b~u=Zgn?}4@oMF_H{gzG%H9}gl|-bv8PvyL
z7n<4Kt3x?(9KMLs-?;%ITQf;3S85k)dS>k!MjXfSY4IehS?%lKy}ka{Tbq4xH|_F0
z%N3YmWjXS8S*|6q{4TC6CiBLXC$mwJ3YHH6*dK5iE)W)mNv;02RzBNu9r=TXcJoS>
zeeJXMT)WJfB1I#%Pw>PvPN60-WXaNPfA34QdD+uMWwwF)hxT1hHqKSZp?7EF#a`dP
z4=C8GM!_o&zH86UT1YosPA!Je6}Rp4t1?L{)|INDB}wTJWTnPQJ5?2d(RHla6oXSK
z`wI?}tA9OlJQa<4c1PKn?dYk6gHO*`Qis&AEgiacQBb9ce4jJXSS!VH_=RgqA+*0v
zx6$azER#w~I*?<3uW}+gc&F@O%#>-luCqA9Bv_J&%SImH7nQVsl=%yvOU}`>cl@X<
zDlARt`Jh}R_R*%%=-afy7^5dCApf?KcejZ@E!Pm_P#jxMNFb8#M2}K+_mNKO@C=kd
zUkF3*NNZc?3qp)_J>*Z6niy2OLwzoO$Ce9_*qetgTCxX@==M^oaW!iLZF^L$yU%y?
za`=y3fq9*J4>m}Fe0B(#uklD>%`DFtkLZ&ErQM-y8Dr&$#8#SKH+12pSAtYQCT$?)
zOW*61go!HopRCDc5Gkn#utCr|SF%bLJX4DoEG}&~0sg5AYn-_lc}kLDPoo%a_LVJ{
zRJ<bm4`R8p9!*bU<IrbeyH#NkQ`Vr>)~SOOOkFD2*557FVEbe*`-FZ*z-4m~oA_x8
z<%!k;rR5QJ2ipB~eJRua9V+B{JoX1mN|i{m<--n@wd^}r{DY+pFv#xLiQdKbFc*K2
zG%9auT|+ys{c$R9V*1+dR=2;~Q@}wvVQ!9m^57*%lutD1;!G%f&9Liau&EQeAH@xs
z?7Lk}3h7n&;%7xrPkCXG!ud(lxe8UzFaJ@Y`f<WDNgWWYh9zSUYfl-6`=jHT4c3X<
zpxdt*)DT`7O5pYOzJW(2YDu@Qf>#9JFC$P>9Mg=9qxMUdjY55Z><3zSpISkBo~af$
zni6DLF^A@K*Y>62yK;)ZN*2jb>L6S}7ok+uejPd_f<j4R7dq)rEJibB9tP~>E6UO(
zMa8V6MF9WZI;2SM;z9;vgh5oC6K;I;>X62d?xVks0ICOXAIJkS<}9-L0Tb>d=o_Nc
zU?m9d3f5$&;2hb=V<IX4iK_@RyMZ||!S#?NC#lfZsU!l2M|@%SiKP$E;-zfFae%#X
zw-fj-6A{&yDrL0f!rhcjrO0N*L%tu_Q==s{w5}E1-*+caPAF~P@H357e%ZqOuGise
zs^FeZWR;UulCeh+!xj&FTaV4+fLS{Y7yLlA=wx?j!u1&55(_NNzGSN3@!m+P9@``b
zM}{QVEa}DxgP@*khjD??;Ka9xA~P?x*tVm@%}da3n0O3N`ao2wcBkdO>aQyO_L*##
zbm>v&87S#M<5zx0sHrf#%&)%wGs0N@ww0`tfU;*?FV{oj?CVle#)d*^(MJ`PaxpfS
zgOhv#q~)glYaZ466=yL(@Ls~GXJ@ng;_+pBx)ieMDWZLSh5#@vJgkTDa<9v{p7t?B
zaKeKk1~LP3oaoidlH^IxGZGh>;I;7bQYc*}85=48`hC^L6VE0}!?7bvGcIR!6<lIz
zbkljF(B@i#L@d$9*2&tNZ}0v}PSTw)(+rzEkQJtD4Zd`@$di+o9oxyzoMwU&@B;S+
z5Xr}GN7z|me`A#Ukxpo<Rk`e>-t%*1Xdw#x|Ks4GB(Eo@+3QbV0}Sr!+W!gyj`8tP
zGg+XxhF{+b=ruEsv;>1vynnIjkOS|AtZxn8@oqf~gndC)zWp>Gfk6G)`k2x3pxN<y
zv(>*{pULYgZ$*_(J-=x%M8fhbNk{Qx7P?ZXnuya}q2^?DS_hFNuI2kej_^u0exnID
zUA|}=w$*+tTbzfi08t?a<PE-hFxFoxEdY`(Mmo{<e5yIw>N$E%63DY%hLu&Rt_+KE
zHm3-Ki*y-dzBMIsGD@1}F_D}?^Y}0OCT+Z41DOW&P0Q&0$o+Xg-ec=+l2N<0ut7JD
z5A*_=W*_^sx(rXINfAj^DCFt@8=Wj^?x&*NUNg21=$)lvPL_I3B4eKM?%HS9?MZ|E
z&!OU_erDjHOb%P|a)HKEll~o6W!t}D+K9k`Lety{(;?yYk2Fh@ldmF^#0<Jt=jB!l
z7y6%}0Mx$UpPpP2B)f0y8d))pCUV^L$-c$y(Elhf)vzwyX4G;aN;!l<CC-X(yig3$
zpQTg_ebi1|?n^;U{5^1XJwBxOH5^TQU+I{LK<w>jyIM1b_b1A=@uyu8VUh$t<yWq~
zln@vAgj~$Xp?S?F@6U=%McqG?32luB3;?3PMD+KQgVppyu0koGfG^XGL5+#(X&x&W
ztjw%_SNOBIe8!F*%5DpTcOp&QZ#o+a%Hy&V4#ib=)Q~tTFE{Bs|F&{*P;QS_5L!KH
z92<4}`-4`$K!1FI+yrGR-cAv?ks+rRd6PFTW+0;#AjLdbAZm4Zd$p4h32Br@bVQUM
zx`TOQO)w(Oa>>o%ntQMzjDSf4x3qBX%LF{jOoTis#pt+`7m*Z7QZZQtahtx(rl%+D
z&Z{I_F4fBj0zT;`qZ^mwClUYXpvaL-s=8Qf(=P&wzztYIr=U8U8dkHa+2&8+kx7g+
z7t*HI`8Ov==0ld%t~iSZOiTvNx4<ZUdl#03)5A#EALo~1yZlmw7+x1a#@k0L3Mh*V
zU=-08gXO}aolX#Gt|+fMajM9jJH`aE%09>YvPKMDJOjh&(pYpQIbP%K^v6c)Txa>_
zt})p(rX~)XJq(au|IR+I+Wvd=YOzZ5oQ_rBAu#<z)bSw6(fDK>XNdP5faD)A&%wV5
zCkg{5Hiy|HBD=!Le9Z?$b5LefksTC}g~h;~tguvw(8mX4v9vVxeY{k`-3@I)#;Ap(
z(T+uW!{!_jkQsWiAMEcX1TLQ<yF6h8AphQUirF!IXu?J9WP71{f1tCHdVMozh}bRH
z68nl2e~f)iy+-pAGSfJ$$?n%+xY`Pxyio8^@5-gPXQzN_<sy2Po{sZm0X~rp=`^AA
z4~a`lOP0Kx%jZu>CxK$=Sv6CEA!3dljrlTYU${;>s<6Q%JO&hU2<3L|vH%=VaBz?V
zl*0_(BUY=<Q`N5M;su82%CNkVfxib?yK0LfSrN+zKVSXy^&bkYW@tIw7hKTFGi>lU
zy!5BmTyNPOu4&jVSdZJ~KzXC4$M%POcIR#}c+!2g&UMFdo9;UH^qGW2<<k6P^F17}
z3HIt|-tF%o?Sw@|$~Rbnu7T?>3s;tCXN{|NMC`5*1k*fo45Yp?!5r5vC7|!gDK!k9
zCDx{@Layo$>s2Y9wAafN01LtybJ!co%%qIRf=s2xJcqE5`{|AM<OZ8m0P(TYkC*Tk
z<UB~}u8}*Hlo_v?b>OfF-noC6*-<?a&B{Hrn?k1wj%ie?wee()pA|!pzd5Vc0MdBm
zWOiQ5st*~MDz{i*u{Wugu@9VSVeu}*G0i|*HhOVp`wgi`3xn0|$q=f`W8{&nJ-rT-
z_8Sn9Wx_^B<nkiB`lW+OVLFb4$3hw2&x@C98Vr*)j*4RdyWvnT;Hy!sgyK6<jy<mw
z0U06tCx46kQ`FTcC$HkC@7?lR2sK{<!cDo#)ysv}yet(Se9d2`TsIEjyWXPRX`l~F
zz}d&B^(Z^%0~q3Ff0o>o6OSs?v7(YumSQ@R-LeY}`s-*=Zz&$YrwbU!VQSP0vcHQJ
zzBD<-iHlVhYG-NFH&OBOsx3gCYf_;Mt2t<Axz;ur2U%RYEuN0nn;CFT_k$RY^u)!^
zOA;N6xb4U2U`__<WV>XJm4)H%`ISvhF&leDMQSMVS*#7)nY;yRL#tWbg)4R)R1GPS
zKYfOdWbbQ~pjb_2(7*KH2+LO(eAV_tf#Nyk9mfw0qFt4PhL=H=l8I+bpW{D#=%l%_
zDw3KvD%`)k;ILB$U~@&9v+#X}L(X88_K%|I`$Lmn{<=7hS@Ie5lw{3KGbt2$YdEf>
z3U_;g=*A$_?h<2pW8Tnr*JoesPP!kIH{LlokdBuAGWz^k^8U$zfy#Et$?zQjQp+m&
zox*KK_ULmipN&r=08Zar2#zB;6>!A|U0Gld3132Sb!_qgx5KhDe?>wdLXRnYZet<E
zc)oSz9Cv-6|Dp;@nNCf0;5Jpx(X9F)=on;pK3WaZSSGpfiL=w%tgx|fOB#l2+L`q#
zf^a<`Kd>%7UTx}{BY=Ugb2@J)kfm4T_@j07{H*AyxyH;4vfD2yP<s3CLE*cfn$)WO
zjkSEdeM09-n#?V9gFu?gZ@MYor}*>TPE3vM=e%XX9&#B-a0J%M{@M4(5I|dhka7c&
z-)?f1UmgX;c>hYP^pWKwC9@(`=ZOW+W#M6CncAl2AWnYqqb#46hUwj5nhs->g*(k*
zpP$#ew*AJxi_UZIPE1XOI(IV$j4yOThe}^;Pc3|i7@%IgC~yevky2B;-gLqI$l)pC
z!K+&G_g~A)oxe~gyQYQ0Au7pa!E~HLM^HZO5exs<M&FWEUHy%dS9chGw~&S3NQ;_O
zK0}oItBO43UfkII-MKuupoJy{SJ(ghnVx<~<A^nlDcsSiRk!_U4V$ZLpu|zSpn#Ey
zrGB6@gdP-pL8=%CI;{Q-_^_@npisVpUJu+^ttKqW_iyFfgov7+Jo_D`@k0=?g8Er(
z{X)g3EXxpWy}}|=WJ*fL6rr7r2SaSQgiI#n-Rina)V}tx&KMo*Tgs{AbVSBBg<2g?
zPzu%o_swFEEzGkIb7UXLZA#h665fl=R%<$*rTkGE%^xo8d81j6iB?o>xi_D4Q#?}C
zUw(g5baf}|GsG6d(eg#4wmvl_Kuyx!*OSvXI*MR8p7#-a|7EUHAKd?u?K)R%rdm^Y
z#!xc@J%3`rZ_JxFZ#r^vGsi33Ul5B`UOil@o*PH4n({eCd<*^NEg?c#ocllp$A_Do
zkm6wFOSMIy5U<#OXDpt$zWO_uu{ToGN`>aobsu)K<g~)aZ!|oYHT8wVl|PCKQ!`Sf
zpW5va@MY7x9QQvGT^dMibc=K<(b3ke53e5!O6w0JSkmx1R8+l&K~zZQFlWsfqOU)P
zk8edq9$W+Ohj^74{h+{6Gi-;271k8jV@~hoXAW-qNxyXrj0MeMR1L!9wE<G?N9UB#
zLY87jPEJMhFfrPAt=OdP$)`#2Y4BnO5Mr5Z!JxwORBdbLziDeGWp}*{B(A-gv-YDT
ziw_~IT-_2ry+E00wA~nIk1XD){W<WMCK$A0r0WqLkNHfX;j%+hGkUnH6{02<3`O)|
zUd28((Th*QT_V|`lV~D?ZyTpj`{($YhCfkjGkAXzbZ-0jO|Ocn5R6i}GsL=nUm&*2
zagF*N(>Z)Z@{kbU)qu4?m7;EB{8s`K{*f?wF#_>-Eoa}QzzJuUII>YihGNhu3QEk_
z-mchQ-CV17T<+`=z1zAeC2z&}DnQEPli>EzvG1?tiC_kO-m}=uUE14Up_jM6f}bsm
zmv|LlyK4~41q9VwI`rcu-tnrP1?yvP(@ROB(W;D@!#AjlWn1Eh9nWvyTeLL2Cc~#x
zk26hrWPHF(ur$om(LNl8Npw1bZyU2-oDeq7Up)};@LBkc)RfuzHQDu>BD~7Qx?M*;
znYh*24quBEleKsEKVzsrgV>82|85qU8*JpXb$`YaU(l}@?unIh<6}WFXh_jwZkIZ(
z`Td;s@`?YnILEX+E8%3cW>^IZR~t-fw8JA%UjEi^*<yme@~hIe?{lT19<pQPq6*)~
z&#R9pD|Q!%nfQ5r&tsqtnXa0U(>9)UvBs7fDJ|UDsc%zqz>`nxf!^+X^IWTzjA_TQ
zv|A`WFe6x-^-K(>!Z3!tCS#4xVt;*fWKA&5m{;fmXN%->^q-EtZo->$_LPBfy|1f<
zjy55{6N}+yp-xileKAwBjq${=C{=rb#;vo42M00L%`G(3nQp811nt*VSFLc>aJBYs
z;ZV`h)b-c6FLD*?sY1*eTqPW6ScJiorv>l0{4~DiSn_dt9){%1X<EL$#dkt*1Uc5B
z%u>5{{yx=^QI!6gioyr7*>qdOqQW$Db3Cy*^M(ER&jl}$SEk&qAv&+LGMz8XZrU1N
zq@^-kt?lIs_k#>Zjhe4p7`jg%cVGAxms-wJe95L)4To62ZQeYH_wpKxo}+h1<vpB@
zJj!?OB`A_0@Qn7p^Uid##YUtmCr@xQN{STo@0RQXa%#Ki#W$FnMBFkHLrYQ)#*Gse
zJ<JlX0$cu)vV2uh;<O{aH3ily>G6{pV*{bog|fh&>W$I5)oNM+QW(!rJl#-Cbjx!!
zuHMYor8v-hMjUTe)?W=Tdk>y7f?a2O%sW-+{OBP_4G9jXsyvS}&7_h`KU&&zs_dl9
zDDhzEQoEr!v1c^U*zt__n)?xDmvo9ymZC4&_Qdv%^WLAO0M^F)ZU)3B461(=<26Te
zF{7dwy_d68#$g-R-sm!Mm2qraZJMjbo^jMgI7&e&(v0@!UQs`PqzzQy-tu~DJi%{`
z<Ku{ux%E0Gx;7)4Yf2?x+;hiyPDiIh1oK?gN&lvW{hheD5)Uh!clupNI4Hu!KOK>G
zq0Q^UZLI`L=&;oSlEnf(^c`EfDg%MS%5<teW!A;Nd)}D@j07^*ulsp=^cZZYfX5~c
zn?|6Q`5s&y|C3BW;kHNRNlsodM)y0{AZ1tE{oMq9i`tuNAvb%riiy(e4-52nz=xck
zp1&L+%qIRi;`u$S|7up=A2)JdvOItx!t$1A%*n}_hnoVw$uv#tnaSyt*Yd?gMR4As
zv|*Ni!BNjpez5e5q`Nj2I$uBYZtc|TJSVlAXg6s&YFqR1Yv^n7UKEnos4Ge~(1915
zd=pO8S~=i-{HAS%r8C`w>`4sug~vr|=;G3^O1wvHKJU)I-f$UA<rcZQ*ErrSY4DuX
z^Q%MI&3t(O1B|M%{OD+du4U5#r)eJDn$7s>%xe3B!NqIyK&Ur|At7-XRf?cs1&6*Y
zDo=-ch1x6CV5c!4(2nrdRAApCXYg+ie)`3h+Us_WZY9SOe&0Q`jWuiHjb_6p$^?8P
z^Eo5J^R32j$C!>``;sxvtvdm&yKMR>Hl#Di|7rZ7;a@jN4ZhVrqCP_7i3ox>e{&TS
z+@5n`fJr(Z3H^kIs<H~iyro;**(U_DdfsNq*%bMaco&q0cd{{Q3b&=aa5BC2MDTl*
zcS++|Tga;Eu7xc#qjz{L<VgmjCE@#VA8uQ#*z^<=L;w3tmysMwpIdwU$@0w0iCv%D
zo4ZE@{|6_0`0M=ej`|0~m*rnO3pW1<L<}nz7|;FYf(DxQ^YZj(ho^yoUZB$uu>Y|C
z{0lhJ5mLQPm~C&L>08pdR_XOdZm0sGaINZgS#x=lgpR3U^Rd&WrnYWUL5H8`ZeCvQ
zOCH|9_|;;Bw4`Fj?#NT3w&9L)Nh21~jXs|MHCrg+V%R_<uE#K}rwl8{c11FSky^}f
zqv-?<ag$;eR2|71n69k{ao^cabs*852S+~%NhQ*Cdk4`k($l<6^i4l~u!5nsA!1%X
z>12M2!0opAx9Xg>74kGrtcdzQfgfC~7kI40@1g|jp02XVKi80zZZ%kC!^pKAcIOK2
zm2B2^VNbh{9)B*CAiJ|SBx2jxr6P5%$~PGg(G{Gk48dQxhnp#H8K>fc)!UvRp|NZ1
zRNvtRTj53-*Y56bH-i57etu^IA^BefcWV%IW5%M&7h@Hbrkq#}<P|;1q%taUXbj2F
z3p>Lm+KWNm`Naaw`G{LAAOE{IrawHmntqb3GaPpTSGcBGCbtz*Y!MQqZ#V7x8uf?2
zYO?5~fhnmox4jz+$x{^-d9_l(Q2f<lCmKRj$PO-^B|WES?P}3ht#?5f0pa#yxy8fo
z{+ve6@Nj}hduaBs$S})kdupQ4i<bML{8ZF3M!x<X9kU~qUIgVvl)zd81E{^uZL#u#
zl!NJKhSjzr!L$4{3+RwgA=Y?TFx1acStTGKvI+Uk3)T1;0WvFq;kI#B0Z6XkkXV0a
z5bcp_U~HBsn{mWE!-WZ*w#1|<>gMm91chtV^zxn{qq{`jo!+gPHB;6!F#f|3TDMeX
z%zKhJ+bwJ3FHS+$<FHP_$L)-XLt4iT>zAOxD;cdgSe1raWvP$h@VD_!9f!Ma->o@)
zP7beWo`e?!jiWfn(60k6b{7M5h9e{La}|cT{I74h9g1^AM&>QoQ)M?cnfOW>v1D?C
z<!iVOA1FDNjeJ-&ad&$gY=K?v2G_p~B|ts+6?|aH&z(Y>NAIo7)~Ta&85#8lYU}D#
zQrNccLjP4?9=8Nc^C@oMdr42{+gAK~8|(^?SpSK$3cTSMd@Y3$_{v?9rL=AAsm(Eu
zIFc%@V*?FqYM!s&Dh@gp#=;sB4m_E(j6rR%R)#5Ee!Oj`BMcTh8=duXi{6Q1Sk~g)
z>Vok2Xz-rnyj?aW8Pnm`oe6HYh$_gJ*HGKUH#p#-kwAm0IE<Wwl?bC$wDCd^cX5of
zYhd_drb7kua@!oFq_BQ6G02qMa`7svaNT=cy*|;_{^cyA9$d{dC8&*O7X5<bqFz>g
z81k6>NDAa8BDy%iNZ&#vR4}_FEK({{`eA5fS3t>)9U4STBQ_SLfoOs(l$El$a?h04
z^}9loe>#(Ro;1u{>lJqRFNqV|#r5nj?ZbKgIK?v0DvfRdY_w;#=2sq!#iUDy|Jd?1
zpLf#fj@($*i!~))<J)4USxS%GqWO~%CH*Ws(b33XsEh=%hPE%~a5Z)e$Uq1zvn31t
zm&wBo1HX`n>angU0R$NoV?2bHK`8I4&@|rtTj|{F@83PnqG&#mwnpEK`b<7Y0+irK
z1$Ko>_JeObhlU!qv?tVT(I*VUaAbx4Zo#_BkRn$iR}n=cPfrTD2>p$M*>+BKLKW9%
zc>ibXZGI$k#EQp!agS;;%+T-=L#2G2ys^H8SFtav0;H`C1DDzNj4B&clz+P}RIOZ}
z9k)w6lJ6@s{m`_`Ekjt<d6I&M+vSbKT|)PG;uezUI5vMm&;ER$rF61c-Xzo0&1c{$
z5{~-9|6y^VLveplHR}${Ox!~aQ&aKLCRf+0d0nLA?m>@?M}uti)0*kFSBJlL>!#c)
zhAJO@9I}9B1@#Z2FE$L4B)3G%2Wg2Vt9s=~VH^?R)(0!B*doGj*%?bI&&%n?Fqc*)
zZgW3+v?eOb!Kj0$V*Y7oHd*p3!rUb5;`!1_`z2V;lH28e?tO31yuBz=$|c{PsJil?
zmv<YK#p_=)6}@+Nsc$$DR{@FC606`K46R8COoa+59Q|*2$=45T65lhKX=qCAvifU%
zds94Gb^h<M4ke*()CRF{E(_CZW`_I8!kz&!ce)@%-K#^s$z#He)eZzs?F(GoOVjcn
zOXC6kfG(=xSktdQe`e4q9c}L{Qa@{UPqAmKpPNGj#YpLVecEcW<F_^%9-zh@2nEMV
zeAZ=YQ3_A%CnOvh>1l0y#D41p-jUIp1ZwT}ms4_Ld1k(|kBWrBbm!;$hFiSd_1*4n
zKK@aG%PNCujnlSz%vO*yiQ3IW3DO5jCD0DQc^E5QX4A?y12R0ZUP(yY(mWe@zUJdi
z%0_wLhV-M0nW-aHHwj6k^X3;-3aGBDGi0w`7dReUHYT=R+f!+2)U(a*+Z`^?Cfh^X
zXSDhIm`c<nfq!g?dYg*+#!LTXL2hBZFu@=6NY1YzIhhnQW>DS^d6+3S&q1Xm_h`J^
zvddwf12$kf6C9<aPMuiQ^)+7A&)=6d-e`IJ^XDYrVui}%6Uxt`G}S4P*z+Y~Z{NZH
zxbq(#oO6Eqs_=mDy?Mie&LBnO{zh9-!hN;pS#nLCy$HAUu>ogX*1Wv81_wX1?)8RB
z%c*w_jh46#iN6ZeGBf>T?|pE?Lhp5HnKMhN9y3L8Cc05J_%JQMt6Xnk9#c~%p)%+T
zK!j&JCdD&@DLYEdP*;w1CX-f{MCIa%T#SU>^Cr2NR5{)zCU&O!_w+djBlK_RZ`P#5
zbX^0!rjLZkg#?+S4K@xVqhvq#h=vo|+N3l-;hzt8>BcR2-^kf2ehtF|B@sEE;0bEO
zZ<cy`a*Qg@&aSk^x#N}$4_erbew&fZ>t#z!D{0Mx`v1JHDko^453=p``S~`e%?kC_
z!0?d;vBel(;)#4k3Vz7pw64zX6lL};Q0el3`NOH*8i)7t^5e&$Q-)?@HuvwVzu-my
zPnG7WrnaF+zNmnplvS{<p0pdLm-ei8M+<3X2?5wPKszt5F8k%IR}KlK%Ll4S|4fL$
zug8T3<=qdQgH(`nmZm&I%mpwy+WP()3<VsUveNS2o`z?oYZHH-|C>VAd<G4==xn{H
zGUQsFQk%`@>H_wAL94SSSO!6pavQbw8>lLfiK&H_zIJ3>bkoXOSQwb60lqP|Cr|2A
zlLIF(7K7AfZ@g!AIOIan(9z5~F5yF86B9m*iU?5zPA@FIzonr8N16HN#cNeA9=zmG
zR(Tk%`uR)#T^!Q0Ma83bBj=+3W3H+x1E%oLpRge6k9C@kbl16nS&@G}l;g_H2Y0y{
zt-=ERD98wxX!tG(#LmIZo`Nyl;NZn;F3MtI7fWL<=LBu1qbWw{*vTo^x?XclObocc
z3cbv~+jJc{IQa^XhQ3QDYj--l%U|ez*xJ8B-Q@43Ha&!Pw#+S>iCi>bMp<ks@2_f_
z38+gw1TC=S(Mbfw{*chFO@G-*Mxij4?VEUg^KhyDO*KWcJI(TZ?4gQmfmCF18~QK}
z3j=t;v^@CrgV6JUAR`v#u~NmbU1LpW{bCCr?=RPql{@y^X@A|jXN4-rz?&>EGZ`f2
z6IN6{duMe&Yq&7G(qh2(M0I@)bhdteAy7tHIY%xhxXWzTIEItt=~Mp1ggas)2EU6_
zCuTk5e$CF$WbT&~``~ExFnT1XFjs|xYCFK0_|e`e0D+00MY?95=*=zU8KQ57tfLjy
zh4n>bWc2Sn{{tuuEepN$!&BlL779-R7<m*DBF=s&$jY83IM1wbRP|p6_cjE*uBG9z
zP9NbEJ^m{r?xlehhf3rHE?!@M=S><N9j2c@?}&@(DCpi5`udR*??vk_saL_@Qb-AI
zMsV<Q&tlVGl|n~CmR1-QVF2(JP+6bdJp*2Sbgfugn7I8E$0o>_)vDC##ZTWwS$-xj
z?V0V<ZHJJ1&wr$%$d?5g_6*V3G|XoT!fb^&>=sN@z7ABhR1z*n+}uq4O*k6d&uM*s
zeWszEv2NfMZtKu#Okg;&Z(|d8fp+zLpR3+mecXJ1%preP{xRq=v%$r~K|$49ae2Yy
zq%N+8v%3C3XOt!<tF|?3gDwIo39GQXf4%ZKa3LPpkJIM3j+J^IFyVZ=&V4O+SDq#I
zaJ5Aul;7!a*a*N!w#LT0W-|hdg7hVdVKM-PL`F8%hB<apQ0xx&Kr*LKtUY~ognoy`
zN>oX=bc<suE7<UHT4HoTyS%T`m%;i@r!NXOuOXxXihOl_BAQLk<X30`a}O2On(r??
z<BZYZ(303sN358R##=A57(plU_xGL^P5#PO`060k!13|g^+IAue^|39mb1-IrH6KD
zKyLFM&i(rOc-ahS+tM~uPHYLi*vY)Ro6}53d+D_+K4O5g+c$OeUVEbfbED-8sLQ)U
zJ5dpQd4pRAw}q@TvHzbbcfS`EkJ}GadiI64%&X<jiSMO5YW!D^dsJ6f3)zeQ?fMCW
zB~OkQfBxKcDG(Y5t1ykMeVEF2hx<fknKk5lC=oTi7+qc(5Z2u3|Kya8FTAKYUstmU
z7Z7OPY;A)NLL<GtsCJHb5;*}<Aj?jAat*J>x7~Dt)BY!~R31;M$whSj^}v%o_;9LZ
z7J8BXt$D3b8Y|%?(V$lyGeGe><gVndkl_uID4w>4$l|S-^BcwTr+(pnqxCrw(2FL2
zZu`9LQMzn>-G?C|Uwk|{WA`^YlS0I#nvp3-e9L{&u2OjwPJ3jxbY_-|jCr5o$If*{
zbEvO!Y~bSBTA&eq(+bZ}THHc2cyHz3I2H?mBE(xm$4V#MYS-^bbj@WYQgXhCe4oSm
ze5;UJVp!h6|1ma}?~659YxCT09OQG#Qlf!c%x;_64SmyRn9IoCR2Gqcn;F~+UyDS!
zIBjU7hL98|i8^;+VolB6?pC~IM!wGmP|P`}dpIV4a<;{b-;dMa#o7f(OE&N1-u`v>
zE<s;b#$Z@25x1tXZau1<7SL1Lg&Kie{)qu(eD9nex>KPP1<qGDDl4|mAPL~>$gC@G
ze>m(^X0LGO%;jyt!lLDI@(WhTW&&i4o7)jq3^7+pr;W>lMo`-vc;UOwh8#BB4)r~v
zC&Kn>GuFF*X=ZNfH#YuS>OI12m;SHI?f77&2wnM-CGK=ayRrS^x5wVyM)uwL`aReH
znM2#N=6RlhYzW`LKs+V+8xhb7N0SckUc-7nM%nH2xwvpi#*pO!S7mdKk#=cBV39(x
zW6f-Y$Bhif*-qfv>jE3gqP*$!SECBje5Qb}7X7>RsGsm&pGxsmv(6*iY+M%UH;&mK
zRm8i(e&uA{1A568nzxJAapd6YZK|dN^dc?YH$Qqv{6<Gqd%t`KQcSVcJs#jfUK^&5
zE)_poTF5n&zFUwg93LrEw~Pz<ciou!ta{B)KjlRPo1a5D)=H+*=x9$&^dq5N`=R&P
zL3*kk%}IF$k7SNzV|m8QEQy&k1%?4YlGCtG;tG}S{d>TFn`!E>G)MmV;YNRpVpxPV
zE2$5}@Qog@#+e8_b#B{h?oWmMk@l#JK;L@|zR$$cl{=!&gd~Tyqx(MNF%a_zN`?Pi
z6vA!ls8u(KQl($?g+0OubN-t&2ybCMI^cf`%urZ(*j7&<vt1)`*^73c$qNraBO?R8
z9T{1^%W6Rj`_h67CJmvG`y?Lx6tmFe8PweKyPK<P`-9=yTyv(@$2XE8+r3{B`}d0z
zgcGNs6MxB+o~@O?d7w_}<4dCY>W6<HU^GYFC)vs(pG;r4@}QL%t>5Z@N2y3|7@0M6
zdM8@3vFpr#`3Ryh7w!hi+<L~WOz&TX7HG9w!r>(Y$mqL{86p@!gg#w^c7FbREBl6c
zPS(V%{$!U*CY@lg-e>arK~qV{?oL60*8j!YTSsNJb#KGBAfN)GfOLp-D=DoYAl+Tk
z4bsiV0HhnFyIV>crKG#NySv|Yd!Fa~zHfYgeD6EP>(DdMv+upvz1LiGUh|sQT63*M
zZafapj*=?cn!|bof6r<U7CARXj3v4u`D3Q1^Ut@&HZ;nVwB9^Rbb1Mi&d4_|FI>CM
zpWd$-E$YnyA?0l!Q@7O|^20wJ%uR3FD@-<bAG6>Si0O5n@_}?S9~yZL1F+}&PeWYs
z3IYQbgt$v30p{w2%GudJUP#nmREa$cN0*i(qgh<Gl6ctD+8xIl@+t>GU>qEt3A!1u
zsk7b3%QG{>nUi4@tFu|A;IS57+F9&&3y~VYzq{2}_~>N+oQxUtp?3^ZN~<R+;sABi
z<@zP#(0c-P`@F~ycPA@$bmabWOu)c<Oab)Wt_&P0bG}|+k>^&jq`;zOJTK5&GdtC*
zJVICdoCDjWgDvXuTX{rfQgaPoXlD-WFMiS|3YJO{*2h?t9&C7|{qOoacf)+By%6?I
zMq(|Sxb`J0+lSWuw|Jvz(yKGm42<WxvO2I?NTcD$s9zw;Ms9J#TLkTg$Jr`Nlax+-
zm4K&H=NeZiP(1wA$&N1Yi<|r71`8)9W+@AY-bBf#KU?P)jUa&xxQZ#Z)haP#j5EXh
zj6_KZFB`!md>=P8#7Pb2=>w{;^XZpYp<UI^tl{x~Vk2$6x2tZj*$o`kt>xXBhHYpJ
z%bL&}7{R&gQncBUkw-n)BJ@(8<_|4x=H8^#HGzBPWYBbqd(>9))3J?eSMm-dZ@fP%
zz-e{&HRe@HnJBUu7iqT+jz0jTK>?qy^G^RmTX6}8r1pu7sblr}u!x6ZY<cgq`tlcI
z%Q!fVw|={ftE-l<Q6vn+J3fihf*cTk)wZs#zOp)!_N>G9uc*eK!twM<%-oGiwXIG-
z6>txTEq%2XuCQKehDeQ-*O(1+5vQ~!<!fEyIRpOGD1@jNkPr@)7rIx{R>p>`S*BQ7
zRaU92%JI}wm3m)`IsNeQugmRJ5~l<nJRViH(0Yjhz*V&7130DBvIebeS=iCj`#vSC
zD&u89aHE%~R4m8+H;VqvhMXyGxn0pR#ap*eEzRrIk37k*-PhL_TR@?cY4+zt6Kg?y
z!8L*vC6z9C-P667mUS7S9pA^&>vmWCZ%_zrU0vbu7^eNmBqTA{PuoM%Y9i9;v=W$I
z$rW^sRQPJndghaOVODXrueP6!7ggn`r1cDuSopcgj*hziqWsYN_cHBmE#Gl@F1Of&
z!vuf1%^=)6cNyC?TXIaM7Z^hvY6BoITD~GK27{#e?u^~t{gTKSGs3U0iyD}|T?eiE
zQ~R){73g4i_scSU)K4|_j|&}#K_lR-3`jAaNXKUW32e_{tEKWGc_|0SVs=+*`MCpS
zF){bAPv2hyPye2lh5{6W49OPI=1PJZ%sFjX>Ei*RuIX$`5LC#(MSv-;8MhnN)-yfB
zbu^<F=l0X?HOD8%#^Qy8#H$JC8yrvx%fP|00sgj!^M#A;*}X{yt)AQU*Voe!fz9Pq
zYx!Z`46<b~$LO?bDK_99)9h^V&&P0(i>ErRf2ll>mRfUsOBOc4MHJ{~yt{zKcIN=+
zTF4_hoUJFZJ)IX^5^p5JTMj)?VBhD;BnnqEqztE`t}r%cM1{=9T6~A*JcB~=QESQd
zP3rC0Wp?=gL=%RAvgV7566o8vK*>>^+KxcttNl9<^!7ze(~XJmWpf^1lH+d)%OC#h
zbVGc`4WyhY=&(Lu#i`s~xJVf^ov(Uoii_#1^t*7gTaUge%daJ^+Ixq9?Bb@EuP;;@
z2bjR$-Tk3}`>=iVsC0n!V&)0cjQ{-)dVITKz%>f83h+Z*c1g)jyG9w&LGh!c%*FS3
za=mYtBCD)B{Y&zR8i>+P&H>#E!<%i<Fz%U&3p_UeoO=DhCfAfki7pUr_lDhdLzzN2
zGt*g3V_;$=7F12j_J_F|&}!Z*Ix>+a^Y^u9SHwIE>oVv*wC<@wyn^))wE9WVnp*JK
zMFS|K0{P8(6xAg^?78_^Rz5$d5OT1P`+W4P&KfX+YkM^}4^>VHRQ~mWSWrPq3GKie
zDt|bqg{JG&?{dp&_O<uu%+CB>|Cxm94?{v7%juR&b51o#>WWmgTT0T|{2EU?{^etO
z)*ZrDrS_5D;Bf(!!$AyzgZ73mL?JTE&h95CP@spV=J`^O%EmyRQjo`-xHuKd1xF$A
zj;XG-O18^PA;{tY?x-dpl{wzQc0X2qcKf{i*rMB;v~guCNCW;{NJMk3@7tETeqnd_
zcOFw-<`CkbYClit&(nXGF$p8cRZ0BJZ4U5gvCEwt3seNOW;DC50%RbQZ>AeNi#><;
zRdA~=DV-(Ju<bwEEGi0Oz8x!OrAW~ik}U3JBS5DVrvxM)TSZl&ijEfDuAG7VkI^Mi
zRX=Ol;FC^2JXO1x6cED*@<+bCiyepob;LrV9e^0u#3dPsZ`}8|w5vCi0gXr(px(VV
z5PRlkzuF!^c8y5{iP(DH6L+}C(-f*EGPYv{yhC_Zt;6DOXM@(XW_4tETL=A9CCobT
z{DCTk{gNLog=wcHwV6nb=)_IeBwFKG&$>FkQiP*EqLrlsVZV87)CaI@8yf=;(re8g
zIN8{()CQ1g%H+)We8kD3s=FJ<)+XTb2lZQ<efgI*xsa*B@(Rc?#q>o*ksY_2Kvr>b
zN$w2*zYKGIlS*IjBFGKYoB-v4j}Iu14<;#1lidkLjPXaEKTGY+w66$iq8gbuue#q}
z*g282#&11*2*uxLj*J;T-}l3><Ip9#V$ximU-_^&G}zJE*Z!;X5i_navcq;0U}MrK
zzWk^D{`|pjfrR(_k1(`M)$k{fkepT3MyrdqZZx~QRFJX^j5%9|kE6q*5L2A0{c5X~
z6NU;~GDM^2gT$nenp&q5E@#L-meL3$PyZGyY!9H{Y~_w0v-DoO=|)2@VPSb8lk3r(
ze>^?ye=&SVQvE$1abx<MOiuWGkdzcO#2t*_Oxu(kgJ7mZlIBkIKIi@CIYxHswWRO0
z)oSW5yBgWy6&f?5G5_i;gOMG+A1ti;dc{eS+aqPtJDZ)|E#JqegO^VHmIyK$JLhmG
zi?p;h+;&KC08g8ByweleP)~ZI_^%~?MR`Zz?Csmz-vsm4l=c&clq@Ab{Rjk3$O;#G
z(cD}}!A&Eu+@eJ}O&Jkh-x|{d5X7{aiy|vy;e+{?_%A|HGP~x@8H*8j@=@x=F4KjF
z2@J=YN-DCNzjq246wTN{(M&Zyo-AMg#Z1<V10}AQOPbuDR6Zk`D#EQEg?kkdG|AQu
zYi1TEs;`nalcCy4I=PrL-CRmqODl`%4@;5Cf%;~eCl?&A*lMV=`03Wn<U<CZVSJ$w
zGgw`Th)WEL^?O+4Sm|1SMfI%jbCG#WwxTUh<CE9|UNvi_p4WHK&hZs7#sBaUCzP0V
zzlEq{IzE<`QF4(-AQ(<0y5_`!ven1=2ck<5W(O^8eRfdbVU(?cItQqaXuZmL|EO(c
z^<IC@V+S@+UK>>~R&_dljl!grRg$sX`wG0bkkLZCu`(W+$0XlkZFV*{YPiM`pJ=>F
zzRcmxR!*vz3Vj5xqUbp7>KgY~)2-0)5)tFgR$~GEP#{p_sGk^H8+uDp<<n56Kn-hl
zw>5dZq&yN5h3$-zQnYl8luVp^y=Lz{GH|fd6!Z5eZ1>hcA^*LiJ>pFIA<L7(`no)#
zBRl3fnj*3`ze2qB_RP)fI}p6^Xs}z>R@1{WTnl}Qd6Pv>PhHOro)bTOYzWFd+gwnp
z_`~2PBPozPqLzuG6`tbP>a0L6ZL8C6Ayx)PWejWyT4ApMOFe7X-QE@lDoMa+e@Ra4
zNle~4;(7lstb9TsKSCMydH;#M`2^;}*F?A&P4p-sL+8GR7WGesK{>gSiVBnshvvGz
zI7$cO;nTclqA@tQ=z8sOMQ={efyVtrYqminKe~e8>~|0pZy$HW5+QIKSeV_Hq8v5Q
z+YxrFO#(izZ)ou6poG5edWo40vc=g7Nz74dj5?9SXlofc5TD)x5n0z1AIO0KWFN)T
zQCEf+h=&$Cj4$+>sft>u`^P)OcXxx<XT}q=94jd@jQ6{W`0jVR#Nr1uykM1a3!1#*
zQq_Ru@|5jZ-n#8fPjbuJNV!&yYBM#>jiGpbd3A6?lwIfH;2<k%zcKuL>7CKqlZc3b
z(p0ipre3Aske%6Q_#b)sH+nY>JxnMzdXLyDbo^JY-uzcfFam_(EcjaLY}DE1kpWc-
zCnPAhcO(wxBWc<<Y~_EJr_fxl_>iIc8xr%!q-4b<p~{a>kB^R1?8tEwx_f>Cfyj|^
z!SB6;Y&8&L!YM|^t>3T;pkY6H+)kMiO>r0{ep_|-VCPP?>kL;2WUH-U%92Q|j*mY!
zgP+zDGy$vp`EJkL`JhP-B&UF6B=VDJ`qShUEVWf7mh#_J4*L9jU7`RUoaAP>%MMz^
zJW(I6m9gSJ5R*drU@=5hhdrJo9i%BcUGbtcN3s>*F`Puk5D%lXx6;&9Wj97_EU|%b
zNe+?S&O$oxRgY6-bkxAFR==tm<d^+jSgg%=5c+WVJJ~_YhT12;Pc`ItPbY9SYg__^
z%aR7G1#KbC5G!D}0LCoYiW3lPCLGr=L0^So{lCjz*)men?nXyPG@pAXe+1PaP{466
zlYp^Kl{bxzt|t8isv{1~##Y?7xEEXNuRa%P>0gN+nNIvH?Y%_-4t}y(R1I)zz?xhW
zWsGcIfr=3VJe5_~Pa@??K*VJ(JM2_~X4f&j{NB!}ph`3ihJJ)mo$O|^uXd1qaX#=(
z=86q%0X%HnR`ufHS^uHia6{Zk|EaR_>#2$G@Ti`ed`{;y%bnZvkBxwc0svFj)JVh*
zV`mIftYS)d6!O9J{p&=6jjCTNwr}qNzWEh9OB!@FIfmld4`dW9K%?-738HBN(*@`R
zmfw9S@wLh$5^qe*xEqmmbb$(>`^Eu(t!QRGq`KIQ(Oj-$^!I0NOAR}U*O7Cdi%S?N
z5`fz-2nup8v=hy$u<-F6sn7qkbaV(WLsG6Wmw_ZFB?CLjWXgY9K65-r^4}HgoYC?K
zB!+ln={%5jZ!b}>A2kI#F(H<|qZC1pj=s;pNY6>qGE<~=3Oh(=db<<4sq4E`x^ABm
zyan%3;p(()=GWUbXP&;KB|m0dg)I5f9BT~ep6@LtjZ|k8PWP}`@MmTSOFg$=I!Y3G
zSCqaA;IuXbX(abmOxFs~zdiG+3QiK#3Q+ob21#@>e>kY_f-!hnMfZ*`%2TGVGA?hA
zm;)5_Vmo%jeH`q0149|Z@Bl$FzpvK-2)zK^gXVJO0cuwE!3QH{eJP<`6RxcOLw=Wq
zi#Gs%0(&nv+5S`<4`{G3E$$@~eJMT9djL0%nkNK}S=lZ)KB5xwsbe=x8-?)55fPE?
zEkZ=Xo!7=qH$8oy)A7WRGJC&3Y)CxPtgr#fzDG@>*?nOec^*TYTK-COvjU#TUR>!S
ziiBsw^KaqVB*!B6uV)ZNZoW%<pM0A9!S$a7AV*jq5fwrW4%8d2^9HJ~eV`j6lM!e;
z9%bAW+ZA-#zyT#;PZDS_WQdh|$;$Ls8y2<8<(9RXW(SYcZ)eynuWLA%UO>ZA>y1jz
zZeq$(R-?jw8n6}2gKbGbq}WjMeW?mbygwi0w?w$w@^5M6{#y(<7mwNI<5Ql0&Nesr
z=UaM+)CaGa-DjDyw`!MLi9Gx5`)}X4d^XmE`gXaeS~4?VEgdn3QTVNpBOi!=xKbaX
zCtlv^STJ2GWn)<!Y8)-Qns!M%=+cIe^o4`~5B+NIJeHgC32Im8_n!`&-iC(foCm4N
z5KrPk@m)<MnC^kxFjU{H;IX*Snq*Mi-)cwGH_JDoFgAL%5+TeT@d1P@-qA*87&KJ5
zNBOET<+1U7IF25hrw2M5<$87%<82w{-nYmMy!b5(<M$Iw&rkuEdf_1ob9UOqPK;k>
z^f)mzV8M4w*Spy6_T~<X(8x3R9j!$(Vn)EjH84<oiAey0TyxK_q>O6>qwR5D@$}r2
z*92mV7aqRVNwMA))m)(STs<_Y3V;%j(fWvle=m37H1!sC;Le6GFEFW(m6d^!$R|$n
z(*T`IpvL}FAY26!xrHuIK653+n))*GY$M6vGlQSe+hq=KDigVvP#i5nB7)cr&Cmgf
zIL>a$_SlHM__qTatKnJ30NSyYWUTs^fJ8<Qb7!5u^9B7~lnZbFGBq@Gw1$1vhxeyQ
zn^6YSf|I*sEJp%yxhsVbX4NV+jkm9#nbXiPEkre)N?Hw+OD~l>makRmBEp=Hs^;W!
z^NVxMboG8`UF-oZpsM>V3p2@Ue|b^&%&&Fsmv_~%U9_gEpsc=}FAVm|=`E?&7?V~~
zL2VXztOcEP23pH3_yQQaGK8Wiz3(&)yK-5%mtyKqSsH=}!qipY6^{`9j0{&^!ow_C
z7JS*a&2_&Cg*8h_qc>#`D0jbenE=72-u7zZwVIrMH5I5lZrrUIJ?iRQpU8R!1kVMm
zo=bia@5AA&W3$5BI+taG!tmk&u4?F5t2WF$Rt+4EH)v-k*3&yMvD&i8{EFq}{2NIk
za?ai7`9I$o^S%M4kzq?WHK!9QWLvKGhTFO}d1B^Z{En~MeDD8sWv|5rX^Qpb*bo67
z;zlfLo&7!@@*^Zd0cYGR_3N;C-E*TZWz0cVyp6TyDJ^rRdRtCVQQ`<D`~+!&0gae<
zFy-8aV750Gg+@mN0EK712zpzI&z3tUP|}U0`^;OJz7<#I$ZjqkHBI_ABmdkV3g<y7
zRtq(SW5tC@Wt>FLFV;ueulTt2_4T?7GYkaN(<|C2j0{Rn|Hy4iN;zk1P}jHNgB+HR
zclu}8_m<!?=q<OtFEa9eGio1sOqVJSrgZJx*bk~}0+-Ib<PGZOm9l+<Cp(@xLhsGE
zfyO%F#_E-_kr8*hYMl#xIK_cT3=9Zwx7_)ukRS2yC^T%k4Ah`=>#x{ytwEZfv&&ck
zT?SaNPitqyWG~3eIv=e7mIef?W6k<cRUt`&;Sfa<DSN^hedn{`<+CvzUBxjbu!NE(
zJwkQ2s%&`qOT`hC$;CA55<B&$&RF!C9c?8fHrCg=elhb8O*D#xpPMxs88n@Jxzf|O
z=yt*tTHkL^1_DC>v(=(FUU5~|xoxu%->P<v1Tp~q%f5ZBGImt-;ZU#^cl(jO<6;wp
zoBRQ&=>~q@2OA?=K@Fs$=tmD7(?U#_Zn<_%9MCcYc6j{!ms|P4Y2L;UxXM+3H$QM^
zHt6RxZ%vVAAH8p|9XV@lCm%Rhdfb!VzleV92v7*!fkA3#vs1?w=^lu6zEZjJr3OdJ
zXv?-ZCmU~jwzl)l_O|~7jxzp%Xc-JPk~~J*JW6%3V5mS%*qL+bT0c_yio?jWj|t}9
z;F(Us`s{!5Cm%8nH7pW8AP7d-FMtT^!(YoLeTbb-qXjX0A--oy^`OmMNzV}}Q8b()
ztJ1=!B3LTMKI~rKWz{N3>Unw-^ffVND9MaM@I(knt=k|_7XgapqzStR%^spATgzVg
z-dvex>_9&?{inNOC@Vdg)VJ{xEr7r*C6Ao|+_(7+kYiRB9&?j=-J1<M_)+-+=-EJN
zg_kJ5Sl7bPsIBmsMC^6P3m_uu0`;J1U*UOOLoqB3^MXK70i|0NS4Z|XToIF3ASd^9
zH-#RvqMQ~F)m1V$C-;q68vNR<05}m|=>RCc{9{S;z;NXDASTEUBGkhul>V%A6Yxip
zYCAWyH85*7z7==m0FnMcaZ364@Z%D2&#wVw0?ep>fo%V1$2~wOfPe`qM1BRTndU&c
z1u7-8d$WEy+nUkdU(0SRo<OJ$kutFLfxKs7`JUOJPFNik*<IT%eKcD-RBl_|J~N0J
zixz%_S{<JBDJxCWr#2lC1;;~B!Kq4Jhm%!bf3(ldx0u6fbx2?gE{gQ5Ci01?c*f5$
z3gms^@~ZEgU6F+i>0sO(pfP<u^T23(?dTG(3H9%yPz}|VC#1{}5Q1*a^a1fu2|$C6
z2@e;0#IW*+`snEvP2ST0%0<#-A^Yrgga+G6g%I^?A|gRuv5vVcpVI#V=+FCeFF=M#
z9;SMLe*xQz+uGe^`7;%v^QHJ>R-8-*y+9(#@{U@gjeF-*@}rO(CHJam#`jY|OKv^=
zTZYF80!mrs%(_phf^P|}nHxNJ)m$rcwSJt62)>AQxv;mh{#xi~(}SKzmVTY!iQ4dc
zWB&oj>I+%5nPSRpW^3QgH1up!&e`owqGKK&iv8(o%$A`hBm^k7L#40hcgo}By1AV~
zTC7-BBMN5sJFMEiL~5Q!FV-{5LnBi&@jXY#^86$!=Huv((gY4_`dzJ+`s0d<g{EN}
zKy~VmM44}g1TX#G8Frgwt0*%BY_=VDWpmG9+O_d%C0peW3SO#cz<Ct(I25NjR+KGG
zk~Ws!bgRMlbOe2U_j`yH6hP~+$YVNywy1?ifqIOK*3*g9S~{)?>kINe@B$VP57s+p
zNruvD7biWiX7Ohjz1lC&3@y&A{+P)HT<X>wHCztwox!!IK%&9rbb~A&l_##Xippsv
zpCvz3?B|py&N|)qJ~gE{n}^3r*zRDJ(Y4A_q1XlWfBZB*+)b}55}=y}A)~M=)bj^5
z#8*jO&^2xgLOT^45iG29iv^qP!9k_3z>EWs@<L3$5KF&UhNyiDs~u=&F>#2yD$bn&
zsc5pZBc{MvMX;4W%S@ZjOS^8V)p%508?szbpk8IVxY?Q7CxG;vi9M*~JKkWK-M*bu
zwiebNL6^jEw6|YJ;<E|03#6~_gr9owAUTPmOoGh?OJ*unaXqjq)~>#x_SHKTpH3Q?
zTa3fg49Rq|%#_f`--00T0ji+p+49XS4VKR2+VeX1aC#2G{GfrE8kQOj6+jO@wK@3O
z-j=GIZ@RJ~2K-7r*5aAnk5b)V-5U(wxe~WOBUl+5h3P1MElXq!+TL-n5<JVHG*srF
zgD#6_y$=nls)C-yRt13goBKoXrq|6tq0we{mz;~W%;)`om2LOmI={SaxEsX1t?(D`
z<cw!A_j;$Wf~-#$b=uSIFD_2Mh_=NSo%@Mwf-*B_*bHs!wG$ROIbf#6vsWFamUldb
zGo#jqo{SY;$0=&3*BshA{vMiua%Ycn{1P*fZmy>W`Wm(BFRSf#@~WMH5|%_c+Id*7
zth94WM^pXGoI0k<bZJK^|FY)9&|LRZp?51HX?7~x@{y&wO>%&EzLk{<SFov2ooLk3
z`UT1V@aBRZa7|XfT5K!`9`6|$1QF`Z_Ys`TO4YwBX_-O~=k<(Hn|3}6PBWg!*qc)L
zm8ZQRDk#`Yn6`Y+^lasUe7VK2l1i?(_lk$IW;(>P#xqt_GV7tl?<8HpfI&7K9ALR}
zHTk5^X@JkejF37r<+e|7rS@FMB_uC>)Z{^pYhpbv(?`}<vE>OwU;~(qT}ie|{3zNE
zwpXHI55x=o4@tZ!4VXF6-_tE$KLgdB!q9<}<8W<qO2!n-ncGxHY;wB!i`LsUuH*#4
zl>D;UB~ZN-${gq)Zr0l4C-__(PY5#cqYhdViia+IW;w`|nXvwX$MG&V5$tFCuz5|}
zM>19Bg-!xM3B9lps~Q_?Dn;_>(i~Dg@Jy7E))kkyQK<!cV6b8nzmydQjs76J6OcBG
z(syuRs=juTRX7v9s$Gx+B}&5J#1FTf?7;C9vag=iUjd}{0ci2+SRF8>ITkU<|6O(H
zxw`)Rj|7R8qUpwM$>sO>DL&W<U*RKO0{Oxy?UP;qA(a(gU&!e(Aj3}#0RtR^(*x9J
zix%Oh2I7_(xX<iZ$R28-qSjPe40y3#UlCyU-}Bb&-Tk8qLQ4JB1TlVqe4)|%529$;
z{>TZbFUu8RuqkU{xvYWpnLZ%@G@<ZYNaQgsjG38kCbq^k791M%>+T(JOW2B41Eu=3
z+;~Dyt03rjcP*K&@#pu4r~<l|Y0Q2#9`Qw1MeYN|)S_8I)b68bpGlH`PYwvh0tqrQ
z%em*&<4HcAwO%9Ff4`_;Gmy3~guzQeFi=7ica<yE_8|He4BzY|f{e_|We__0-KtQ%
zzhxwf)vp88YWvL-fb7rn;p66h=*=&{(4K6jlkxGJz+k8{NMsum?Z(DT@VdL|UBg{0
zMP_0qC3&qQRoTz)$;`eRBOsGm7-C(!_noT*5|WmIO%oHX&&?)-G$-0u1SFk6HIP<@
z`)PA8Op;39sMYOwAlL3v3g^5eZ71+<zD!nDQ*dr<POp99c^x=^X@b=SO_Vgg3GA(b
zM;joOxa|!Nv|jBR)J6Pxy1fKSfuE13HIeI19$ztQyAEgnPv9#ig9pBWT1!H8H#a#|
zWqAEnR>DC-DeD=|E1;o*tp);GiQT~l9s3c#t1T6*Z;wY(ggByuF31N+@iG9PWCxM^
zsAxP|)OB_L09{WL*Z@&=GL!dc^$pnR27f_dH8^ESd%}PLTmnKLqzU}o4#EB9$J|xP
zPKykjWj_wA7n?!V0ONztZ)zF7$4yt%y&nM}?A=)-;eRWr%`e1*)SvE<Z?>e3&xtH*
zZVoTDwQW^x^7iVeSeHO6A{1crW*s;VE^+EELM*%`LWhZpf-VjiK|<0iKTn{2nSu-M
z?Z;sayK@07QE|D+?!7@RExv5}|MiStAx?OGCw2!_0M4*^43u`yx_y_gs8QYltU8P9
z&|}Oa)cue4%SXmZ!}tQFDj02N+tTjZ{OO#~PR>c2H~@TQrzDFXBzb%Irqx~^m8w^`
z^`oJqsou7k<Kg4&gZ(&Q<DT+XCG}#2v&9uZ5IvqT^Fqc;rFZ_*p9Lwz(!hqk?JTXY
znTEr8iTB~<R_k6J5oCb1H|t&1`x#Sm;ZgUEc5b*_yyHhc)_mn1aF5RsZy^TshoC6p
zPpW;L-0KXlOgEge08e6kMx>`=AX&Kf^F|rFZ-*}s*wVxnv%Zy{{8r^1@>Fkv;noH#
z*fUXTI{k55(%k8m<%4v)W8G*jqKmLFAZ?2=9Mk??B6m;Wgak`cue62QYYec8<Mz$b
z&h1#&A|Qplpt1LuTSc=OTb&MIyQ03SQD<v%gD}(B4PKW4%kB3oy2N<SC3$fIEtNv!
z=8AX|`vL;~m;}Mbig>v02uz-Z*7?2B)m7JIbD^!lG%g7pH+(NOx5uyC_H?bb-fOPW
zM!7rFebXhZ4AW5W1SsT&S^~|z#1<U}DqN5l6HQ9rWRSZX*>_B(6G-zL*o<{JI^+Iv
zkpbjX*V`7=vEN(TmZ(Wb%kS*qR=nhDPtC*`qHWG`2mog!_btqxk2T;`yyK7cZ6x5~
z$Jy@c)e9?02QsnXJ0$g@{QWc>4_3n>44%a(5+<u@c5nI0ymZ+j5g#$b5r8-Y$Ufb3
zO<WR)q7ZX0(PM|fH*w7lm-l)A)e{oXMu1I=tM62ug=FOZ%ik>sh*F6;TyxwCNQYA(
zAR@YeUNo_6ri1B8!L!3Cjslys(>Ax4Qpj}|k3nS?W^c9p`=yTcwS^_v`9W1!@;!8z
z%~3=ID=HDyD+J0%U4zbX2m>amwE`rfc+Ymu3i#ZyVLBRwB9|Bay&X@f?G%8Jw>mbe
zCn;x~$8Xgu%hMw1j^FC>DZ2Y8`cnVOhb+%SE>JW3j)(X1PwlH4UOsr4P?n6Et1%So
zq1bkiKUr-Cwh$uHT;sk#5<d62oXkzCL~nQ6L3!*?2xtL3PM6#rb}>nl7>f0`4XoPU
z>yXyeRL@&jNtcQXYg&<@=q&=Z{x|XTUp<4LgY<nP<W*aK_VD0?Hw?5I`#CaOIueBg
z`(Ohw$kxbVML4U0hexRjfJg7?p<$pWr=ku4LLE@?eMc^K{<EgU@b;?w^fYfsl~5)D
zTHYv*mQ_^u|Bu#mX(Kp36Dj<@*aZ?%fQ*5)A_E||0GU^E9i)OkmhRS^d_Jf7-T?ca
zop#&YL|9pfOzI|%V+CAacsB)oRcrRyKk8y;TPGrnQ!mUaU%Zn+_CdzOC*WJee;5)j
zUMQlb?w=b=%ThlFTbN7vnE~C;)f19sWi1osC0vlBr(<gC?|;&4*4g&(?y>*dW(N{x
zI=Zx+oCh!o35md<AkwySvlY-@YYlYLv6!v9kdqO9^0%b7^`63+G!Qd0e68{3Q)4iR
zXr_K<l6Nlck+I~`$btR|$=1h@55S>_t=@#3`9TNvwu|-cSugEAfV~PgR0bf`a4+>F
z7ucKjpceBK01I@Pzh3pF9XHqkj^W}1VbDJ&T8u4@d0Fd0*g!mv;<bM)xth$|mGtMk
znFP}~9^sK~_9@snOo~b_1k9SiqY67nQe<@$ePk5EA!u9C2&-$<dxpZKUBuBg+ZkSs
zMZFI;ihd{9ih<lEfDUr00xCUhB8#Mp8pQ(yV-g;l*IBkq>qxJ3vCZuU|CKo+sD<c(
ze7tUmU3SkCDCJGBn^~rbh`wpnru|wydk%J9euVpAI)L@yKipWux0^k=?QOaH^tBNQ
z;86H}J_#Mj+d(w=036{;VQe5Ts~MOGWm1Cg$jfQ_x_+uM_Y8Dlyc%k(7m%y6F1!0p
znWiBBGcn9BG3puEmA^jMK)mA-|2*P`^>)F?+OT7-EJ+UlY$9~@mER}vci?C$Us0tM
zi0E!wjHQEuEL4!<Nehx-7}I-BuNQQK7SHA1Ev(6C5pJuGR^}HKjguldETBN<Z>PWg
zpc9sTg1vG7iJW?9bZ9>U8DB)~)fFPEgyg+JE2<+;KAEI7ma}QrWH(lPqXxTGB4TM7
zmDH}xwy>vWPzPmAmc|$=!Hy>G4xhVfxV>8Xou@;iNAr6R$E7F4T^%FgI%no%ei7n5
zoys4*8VS68a%3XfHiybf5qjewWQzVinHUja6aCGnnOWIm{@!PsKe`sx%?^(G=-BzR
zbS%d+_-JXx`E2A<gP*s)6QOAkayzLhx6BmyW{UYN*NLP~hIxivu8YHZAezrD%6+?A
z&|NA$eI(Nm6}_$?K(NoQVWK8)eJ`g=7@dAZrul=fUi0Qg|G`qH6ZbPb=!xvJzbz@6
zsO9E`3m2`+Ot+ip*&XMjMm%*G_4MrbO2}9Y={TuXPMez41=Oo}gWuF5J2h&4sa0`T
z)Y5X?9**7MBZy|{%+f6pTAHll+iM^|=Hc~_pD)CJ_4H}Y(d63^Pk|Gn$20xeRaWam
zQ><oX96=2wlcj+};o^4?PmS!rFVGD;F;^cS-#^Gd&;IX`!1MVu`-^oetGFgtw>RAG
zFSL-nrCZ;<%W)V|WiQZ~+cat7^1e|H<xbR*-Dr@;eI0P}@Z3MG-Q6hanmBPIorCei
zL-=7Dv)>SydpCS{1hy(<?$WI>Iq9oX&f{I1Uk;~BnM)D47eZ~?>`nQiO=(u<gZteN
ze(`Nk(~pGE(rzr%ipekinXEo1k=4>8OmgYolNQLm?a`y3meEd?@Rc^EWq7T6LKK|m
z|My5z{tm}os(q!6*w~tl*v9<3WRj5>^RbSXGc{Q*tT!@ir<eL;(lw<H5?a`p1r9up
zW`p_GIbbUSWNJRBov)9WS@Xk=T5M3Iakp4eKqG`mL0!4Brpk2Z=&1Df8QQucJ0uEx
zfV{SDDzg2qDpD@|%QaQK`iz??&Yh4)Cc|5p(VXj#o}Zn41zLwz7Gf{!xPL90Vyoc3
zD7mdMOB^NQ&7h|eDByxz-Zi_OmD6y^)||$$DL?_=@hqjKKzE(yO{}S(f;V;5rU4YD
z?~=8Z@G<ru-Ecd|ehGcTcyr?AnD{z(G;HD7=30-G2BjGJ<#AD`Gkh5H5qQ4>Fr|$Q
zO~(}=g~0_yt+_*tN%u8Ng1J;nzr`Yo2{a)UHp{e?s+kPqMK{_$wUjCC?|JAvPHLe<
z5KX);p6hj`TzH0WqgG~};26?bDHRy=%@LEZ&&be>5VDHHyOdE>v<LmkqO%NDrkgTd
zjWJzp%s)f!wrBI7p*+qNmJGDoA=N>rg&$hkR3$qv{P3*)q|%W(d%XrHW<7$iNag#C
zXRkh*Sjg#iRxV<ZY{O8lMv}+d2^SY=gcRo|Odmx%w@agHaJ4+W#(3ipl};KQKP;5a
ze<z-m(Ku(rgIe&*nD1lr%T{*N0>dkeO~0so2Qtc8M$zC62nQk;b6tU`L3xZB6F!1B
zhF~SV@~b_g(Oo&Kp`lRSL$4O`^s#x{J~vlb=F`Tm(&>lgx^;v!5{a1u;R(cp9@{rx
z^<KO<KIp5pZ_Y-w=cPVNXa{HG$a)A|ziYR(W!tDRhyO8q&M3=+go12#JC`J{`mtno
z{HL%JgJxdf^=I3VSCl<d)uM=No>>J5*-1op`97zca;%@=BT5NQZwk=!_x}NHI8<+<
zB)Ki`?%{{ulFVuNKp7i44576MI&(;Q3K2Q)8KDbOGh)L0;xl5|c>KMs;<@6H%JlYt
zpm*IiIA*$o{OEVb66hDpsF`c48zId;>xEalVK$#g-C+(=Qvva26B5ic9CYG$w}w@d
zz0Nw!MKd`w8`J?6JX*3TeU6kHdDY3L(=if}JNBE!OYQ5$&{CUfBoP=RY?Gvc@L`Bx
zj6?vZMPy@7*1*oB;?HDYahn4aB3ttEQ9s>6n5<{=9w{J1jnP&5jn-Y{CAsbi^{5_4
zWfe_j<Zf}$zHZXN_5MmzRU+H|!;V+sHCsUNC@L1U3HFgU0!d73#c+*j@fP#l8FP<X
zgCL|)Aq~O*bF%Fy?E5-WZGO0Po2U<k+LGBSFJ>?9Q(9B(f5ikj=FEFWrdIHA=9ts$
z|JJ_?OBe6!Ry=D9qGNMCvru6-&vT&7oqYERw?gqI^hy0id+kYo5INh~RFwKpt9nOb
z@zDkad=@kz^excC?{tT)zzJgp6Y_=-5D^N3(T2H$em9>K`)02%gmG>yJ)^HMxe=kv
zBpRKK!2f~oLm(<l*C2#~zBi)))?Ij`PDd;E-xFG+gg)Vt<Wft&4`+Iuud>mg*DOdW
zX4-gF!z4S!7Gjz~-9DWM|AS!sN!r+`i)6kybpFGITp&VkS7r~>s2vv6wpIOb$uKVK
z$I;=+gYvGvHc46dx0Ia$-?DkdQDP8@xj>mfp!rlIxk&sBX*gp(<xM_0OuOeLnm707
zg5bZy;QxZa%&Yf;oI<Q4RxN6mo^nf)H0Z?(x|tYGp2UcBB=a%zg&OqBi>7<TTrTpC
zcQ6pga#W@`2u7jWY~au+2m>7d-Uyb00G_*q5kx>c{4}GbJiXHfhwgRCB1Xs_M%Mdq
zZuUDNrPd=lZ>E?mkJ2}H=da&VOB=gdVU$cLt~=6$a2bC_VR{*ZI+81FSAQ(vw3)}0
z$SC>Gar!W8FgqlU5Tph6XYgxOsOtd#0S)Y^tK=jvcQX`nN2YyqJsI8_tW-+)aVpW*
zJ-oh_ZFUh|$U-qQNF`kbLN0&lEB}wLmu+jK5@ItAuO6G7Rwh@t>=oCnmQtS_k53&h
zj}HW{{&W0%lZ*Zd7Pj44=&$e$W0hi+hJ3xFHU6t+?q@YKan3W(lQmxP(Q`sFh@>u>
zT~!t%nQ~9n1$4{#i5v@kTG>@Hn#qL1!7`B8^fi@VJ5}fNjJ#zx2o*#?INnaEb~R{Y
z$d~!&mBF;XA~5^o_g6<0?M@0`NgJ*=WYQaG&mX`SA^*9l?{6WYlRr5cd6fE=92*s!
z<vU{@z}KtNx6h^pLj!Uvm7Eko4doTe!S!kjZ$G)}->$`#T~Eb5{7>VRV$n~yl-F~S
zzQZ)Unxs~9sVEKFWZ6?5+U-3zF4?=6X0d3^{U{-~Y742UBMIe>Y7K_@{5Q8Kt-yU#
zbD}8&-#gpxXo8m+;-f<iI`$@)MaIT!X#H~-J_M8Xg*$Aiv==9>Wis@143ex>Rw~!N
z$tJCzgbagMZ%!0_taM5gotQ2~ysfITW}xDtt=U2m)-mKh_tz@n$L@`@{`Rk#&d=%N
z*++6J3+)-mh4vU!6-WN*v;~tzDg8qrK2o8SFe>Gym417hh6xT4ba<GYuXghi9jaT_
z)6#5ALb$n(jEgE9>F}+fJ^4nFt9x_ldKV^ODqGdXAgyyGADrADmkh}bmAKYqN@;pu
zT_}VUiI10RBzxX{A;EXCdq(w7Kb$vOz{m`1meT^tPkY9WJ39AHHnWc^C{2rvZhqFk
zs~yjwks@*rBB`<Yx~cSG*$mR!-x(#P6#Jr~`66ByIK0OLa%Kd|GQf^ES~es>2+Ys>
z249fC@qD4#R}Sr32_M=h=N9pfWg2s-s;6uyw({G{Uk1kr>W*1uW@gjRZ750_WK3#L
zD-z#*^b4Hxk1Il5ud1YpbNy{!t2A(oGT0la*T6>iLtYI{%n~9ew_mr<dQ8HVhD`P=
zY7+yu&Hm2m1!k40xc0fT$jGZ6)({kHcWt2&oA^5q?UyrDAt<EAx0Otvx>l511$u?G
zJ|vw?Hyt5Rbg<3v;%vS?<C?uozPT#4JVB_M(%N=GeKRBUsp_+>on2H|n65+MBUG?)
zU|yd97{^JB6C-5f_V(>trFTCx4Q~<oqiw{)ho94w40637GQNB-A)~TI6+}NnO4p_)
zN-~yHcm0T0C(X*XTC>i@mm=3|4JpC<4)nJ1?iGBN*T0FEBXWVee=9&uolOv-`SbH<
zgz7%K1AEx4AA&U!XnULwdZ{b7=e%^14IO!EhB2h*>Dd4wARrhV9p%g74lFEy&dt+x
z;_X~Xc3C0cb7+tb-SXtTrL(i0u|8CF8R-*?>nJNOW0XuWN+Zx$&sg>8|LW_LxxIG}
zhx&8WbYsw;M`&n~bgw9}>U6-JkEcvtYZsFV2<Y72dLAAv?Ko(yzHYy|tE=N(SxIzZ
z#w8+RaASmA(M_wSR#bkS)xhcy?+}$jdyFC)8_%@)alPp=j;h26(Bh>$pVcuzCd2Q;
z%Wua0#Err{`8@@FbaN(z2cGqQ@wb^kd)70g%ZgUi7(#zt=-j;2)Sh_aAky4+^tqV+
z(lx)QkLrsUfqR@TUQ#<^A8IH?bI!TjLuZsmySh6_!0qF=<Yb+~JlvNbk#eri=~42U
zl(e|JUZOG1*2)x#cUwk&cYT+uHc?aOK=1AAqpU0Vk2wwtV>^GM`zYs(c(OJA7F=*?
za@G6j?C7xOo!H<XPZdR_e|l!ZhIQc=9(JFXhG%3q=%f!9Y@_mf0x028YFrEvh>ENi
zZ58Kl{>762SUG}HhRl_O!&S;B?(oF)_%-Ib6r?c@BI%*8-V&5z*3(ywj~=5H6#b{&
zV2%*YK7x|T<;_{1@f3%+Bv0NfDTRrpxLNfM>WtNmU6)MF<<dnb-N((`SLqJ=bNuX)
zIRAWw*`ia3Ve7&0y|<3F4VUe9ar=Y$mwG+i=;!MS=$AsQK^BDBv!y}Lun5xJV<b>m
zSx$NyzPR3H>y-1{WLWs5s@yt-lho<{9LfIYelTXiPq;aHAld10yL{-e-($;n^oz)&
ze$T66)+wd?RK?z;I5%Qf_Kc&@{3Yez&p@zEghzF6*h}zzsL%*<G_KmsCsqs9ORVZW
zxjxE@N+&6Pc;iJN{O+G<S({zcRuQR-r!4yStWB5~ZdwznV*2x1Eg@IOvCyQM!r*_p
zUU_kULnsb8B($#wmX2OERd{`DKe06uq4ObS#O4E;igNVf(IJJ{+Y@N3i}`ue;=2<0
zg^i)`jQNt60TE5&R@n*{`1s)z<TUR+Q&ZvQ{pjc@DmgYc^xK08ivp$Xt^{hC&J|m+
z@J(T1wFG2k`qwssx4(WJ+Py9SBiNU#WBDft!eGq8pFpR?zrX6Rk+lEw<Gwt$^WW$N
zlD#Ggx%vOitN-s`4E=W?|MzeGzjf!o$u4*)|Gtp_<8fSQ5I1Su>^EbG%g9u5Bj<UU
z#&!1eD1j!T*Voqt!HB5Ou(5MWN@T;av9T%QgZi4>)qr=RpndtWf03dD3ZL3h*49p<
z=$M<+ajcyImJr^xA`h-0dhCnVT*CdBAu=Y0x@wA1?9t=Lm9NnYqT}Q9y}V!-S4XFl
zF<gn#;I~+w!yaRm^_-Q%eb`9g-6y)|yu6w`j@$jg#5@JTpXv1CsH}&dgL@3IQRXVw
zKmWq+PlC$JdCoD!2Js0KHNQNcN2xRNqR?hdd9BU*hVHiRL)nDc%u=f9P?oYNOg0Fu
zY#sjr%zc`*k2f6&73Y1pQD0IaxVd`51FsPd>)C?Z-5Lpa%d`i47t0Yi)CjVKXrLK4
zEiLWg(a{iSTUQv2n1b`2-8>rJ;!JF?tdBqc*B2Cxga?opvo790y<1=~XWB<~|1r?{
zOyE@h`PnN5EVBRnv$j9Ne_I>>zb-}_Vw1`jg1xZ}O-Lw+pq3F2S1&h7ouB^@fOW4#
z-O;hq^X5cFO#a=wuW@XqdNwa$6~BtKY87Q=q33{$Xz*~$UxMWz1(WdQk&6bu57*Pz
ze?~%L1+HrV*zP~icBah&6{>%J{r{{gxDf%nfWZoWn2pfYpH7;uRJt6r;({)DTLj|w
zQ!2S!k8^OCwF`nELe9<gfpoL+lGMxNjr9ED2K9k~0p)f?S;PeCY>{9h7i~-Hx&sl=
ze5xo8>s}6{<j`<i5aCG==bXve?o4NIuQI&py5mkBc%SrCRBfBzVD9(9I_K|Vld9G%
zva7r|?*=|&oa~W>!GE4IzSGgsA^!fotf1iM)xKnA&`?cV{9ePnW-+dRx!FiwTm&qK
z5j>g1<5-T8lJZ$C9~YMz_{b$_`e%PO<7B=+{|krm{=9^<3PRJ`j9njJ&2G&WNJCO^
zVmYmoR=Z=#!7cfY`-N&iH$Y(e8NYuE67##OqT(`@bHDH+z<lx~g}L$iNB50r2+2qW
zBo<+S=Tl?9A^)WVrn<2dBA_fU9}XJ2pPVm+<XcWv9-f_zZ46~sd|VKuq@Y-zahe-3
zjM2&x2_jG<#(hKFPE-eTPwHt?AoIH8_PqWHAAp$a)lStjQc`B$X0qm&*x-RiJ#iZ$
z&^39w)4liM%0-&AVl`k+41r&U&=FJG7_{qD!`Tcwu^qO@<?h|P2hzlot+BE1uk?qD
zw0Oh@!3!@8eApx^3-4R_x~tCRh$1moA_AQC*r^|v1%HiV?t~Ba$+Ar8sHsK6he5}A
z`Kjae0ip27s3;}<YS=l`U&C&EC8f|6!yH4Q<#95ClYLKiY6qIEkl^Ct0?pwSL5n-!
zE>e8(d{RN9da>}+ohcQw_?DJ;V)9p~JLU|yu(?(z)O+W3pxNUlFcf)P+aCmMCMk|P
zlcw`cpUmbO>fsK9#ET#@EG!kEFEROak)U$7Gtgp&Q$_&va}*R59L`msMknSj?;hZW
zxp(2}K<;E>H5aS#E<LdG?|xX+bgy2?6*u12B)M+Vwzjr*0DuLCcL|<%4CW@!uq!$#
zmjIT+jZY=fg19d?2i?_cyR1a9oH?!czlCnLbYf*N_Q|f!_Qnt~cvEt6Xl7iu3gb8|
z)pK{OfEk0<im67;_i~hMZEfwh3aiASaQ@46@{xR%zQuPiv&oJKYIe6%3lCsvtyUNh
z9}ZYn&13<~?&|C7OS6Z$KO}d)cU}P2HLwGZ&C$Ym@B+XJj*N&%2P?#EvfT7?BRlyR
z40n?mx(FDNyEhzqeeEtL@9J8UB^96blw4#zTEj}t#3ToJrjr>bXwDR@yEp-;f!kp-
zgGsAKH382{X|zy-T5NML6SY;VNtBYBdTUGvVzhAsek42`KBt+$rD=qsc@?KNp*9&m
z>xp-$3=Ms9vNxy0s8Kmwe>R(6VLnF3$Qa}u4I`Cf>ulR`%m?OPbJWYjclG)6XZs%Z
zF&Z8ojlFv4ED7#Qt7cuh@^WAd(V?NwH@d%h!WxV>1~cRMJ!(Mf&yzt3=3L<b9IzBr
zq@{y_{pLcqN6G57j)g|u-&2v%2~__sg%{>9S^x2IRhfIvYrTnZ`%(Z66nUqbebDJE
zdc5u-Qi!2ECJG(eC<xPpnZ<xkmzh~mR;N`}nHT2#7tr~d&EwKuY!H~pNR|{K(*65?
z0LSq4>-yzZ@ffkwbSgm3JxP31i{&u)d%(lc2o1at-`Lo|p#~4?y(KVw_wHS;C9pBx
zDPhzf_x2ejVwj>RfU!k_vpGoZ=bP?{$wyF0VdLYgnZ<+7cH-P+yC{$Z>Af5)07g>K
zfRRT~G{_^Jhc-36VK*Io7jC~foGU-Iyu6HUj?q;0Q<I5N^QUY$n@RukfFW3k-~gvH
zS~E?V)Br!objV|2=Nw>_$@-U<vssRpybT`)Ujkg0BPM)giu3HFr}aplQg{YvaoRbT
zyn+VPIf5HB1R}Fk9Gje^WZ;MmOf|*CBN#l7Mc|2cy=w(Uu3DMV3eM+yDRi>)Nb}%q
z9~L%XojY@lIu{ogNmECMhwz4#eCP9D!ZW&<Y9)aUR?<<yP>;99WKvU8)xkZ0LrR~X
z))X@q*(y*i6=XH;jm*$PExET(oa}KLO97cK{|N(c@Iw?>EAY;j;m7OgA!d8C^&ra%
z@2yKdUyZkBGaqFT<5hv>;QMqKyo9;;=PG2gd0wB}PL-PuiS+TE5$XIwcJE(k_GLHh
zjMN)?m0-C&URtM%=q@WF;sd-FwHR>uq0I{=ub;y56F|d*vpJ{{e49=SXsB36Kq{8S
z`)M64j~E^|=ouNMK$F=KA9TWU>pA|Fm6cq%bP>3BK}N$%kzlTm#MSr0=<4Yi0dDRX
ztiz4|RG~*07%F2$TI}xUAK5J@he0yIQ1amyLQevhg2&bA2zdyvo&ymK$BsJ@-pWzA
zKs6O~bYEFsrh&M*)rq<7`zyw@^OXzL)rg;Zg{1sH!adhNF!lc!mHqERS~&ju{{rBY
z&yPuXX=-PfYG;Z-3{oR*U$dO}(Ry?RI@FhPKL($XC5Ph^FWa5*QXwz_35!ch&z?Rd
z^KDN3{vGcGcy0{deJ@PVe+`&@Y?i6RFoAJg7Kp0}(#GCbf4aMmu_>u#k~F=cjk-C2
zM0m%H!0TJHfKK<CaY;kK1NNnEmEo0*3I!vhEPw-YDJ7yK$77<Svw(dUYS*h<*PldF
znC?zj6Y#m(OSKnZ`9Er5is@RUsF0BRRcS&9KKn)tj`H6;L_$K^oUN~Q+?_Uc_~(ps
zuD_$lob`atv}9~JK?h)XOcn{mO9j@_1)%6Zp-eQGXrR(^svHDga1XVFzh}+6W5Cl0
z+27{H^{W%ol9GzPkBR+vIMIZF#XmwvR|>#kv~Nb|9JYT4h5++YH9&{KuBCN#Qy&w(
zel7-_wuaxLl_6MVz|i@jK`G9e4Dimlh#pq}V?LX819H}B)!kB!N{a)ApHboAFT^IR
z?Xt5ZW78y<s^5{3k!?<tWzV`FNf{a&C$mZb3x^{qT7|3}r@dK!fB*321zBKAY&MJU
z3N@<-w^lt0H@D)$!_%mfoCQH!UHJdcmQFmrKI+>VhH@3wqT1P{z?JZ5DX$0`evK6u
z>o)t)l9Bylo!e(0(<BF00G5DDFpur>w{6GT$qMte>O!qr$0T=Q0U6$d7A{uBQ#gn2
zH5m++5iA8Sey1_{4xCc1tkwu{&v1@hI*3pT+o#|&reL03*AkuB?AK)hpn?7@e2jd3
zyNy&eZB=E|n~)2TbiQglfFI^JmnW{9x!L6ayyzkQE`z5LfI!LIohma1b(D#8&}bPR
zR4##geujgi9Hvktr{v-xH3WwrYzBWq^(?{Efp!Cx8{W9u{UG`cgTJ#JEw8O@WyM<!
zf~)*z8X&Mr8C6T(Xw}#c#Ic(%+jFnjb1SK;supO;MZd7B(y*$k9k*Olk1YD}x3!To
zq6pmX^(;8zxt(?y#NaLttQFM_eq#{EQiXh@Y4~qvKO>>qk0_}`(JbH0PFXi<rTuiP
z9Lko-2N3}@K`xY?eX%uGta5jIV-DTCpUrD%*lWD=^!vR&0bs#!BXox;Jn<sS)a-nJ
zfzrR$c<P5?fX`#%yt=wN+cHH(!#;TCx{vdZZTInC6&Tk7T;jT2+FlOYQrqD?k%$1Z
z4Zw_eI2<1VU#H`XK^i@@5eafFE2|<*EG+g9O)#-RFp&`OxplTz?f{&xtgU4Of4>3Z
zpKa|-XCLS@DHYGFApSa0dd8-m*zoifn6jHSfx7|lH6!EWxA4j0Vq<f`RBidw7lqnq
zJ*l>s_<_p57bj0VcGf|itf;E`9k>AAv9S$%GtkM~KOi82f|$=$anAF~GPk`Ln4AOz
zVx+xMSVI{k16CLuybNc{kOCAT34Ac4&9yMD7R3<T36MOvSzTSMB^$;Wl_ulrxR*}$
z{(TDsRA8CYlDMvuDaiXJgLPzgeSWa&OX_Kom64H9mi!Ql>U((i{{F|K<&Fr$O1o;!
z)s(kiB*l!tcqF!EBq2$LX#J^5%gj>+S=kSc+v5X6yYs)ktU-f~T|@aQ%nIQyOF^8`
zlSW{&lCCZe+DYl{1+rz5PmgWxZq9V}4@PV}b{uPCqobA01c19ZD6a)eSi(}lex|dQ
z3HYPHkdU)-`W=G3@zUlps6Ck2FiFE90p7G@o)PaCD)7JJC%}T~0q*>Qj85n1W@eVh
ze?E^0La8)7FuS|g0n8=7)4XB7`2~YNlmgym+_2Y;J~5VH5xlNwd!yO$LmctirC^>;
z&Feyw7LzbIfbI6|)GwQ`2ZNLdk7iJlPvy_kaL}3quTBnE%U&+`3IHnrJ0WJVLEw=j
zw)=W}PXjx_?38mmZmSw4#j+Yl$EN<Evx9qkFHQjTV6aa63Bwpu0L(hOyN5fY=uZy2
zn9avabU};qgK{+x*cKNT#Th?mWo5apy49~GxnW~szVK~cxOj+$rU(cD;5)!?5GwfE
zNd!FWx;2;Dp%4M%-52qk)=XtHrTXp1yS008vp1Wqs|ExBb-;|(*m$WvWnye`MM_G_
z^i{ba_!4utCr~k1OG#IE_07H*Zk6ppgeq17Ha0eZBVekM5>{IMThg;$v~~cam%Z}#
z@yXL^^aP$Q`SyBE0HlT>x7#=w(;0(!_Ug#R2v`ihfpk6#*h!y_F>c@AK(*#FXY7JB
z62RQ>a~f1G;?*lHazWJInW?QB=77gM<9AyZAGs5MOzo`AG*BsyouH$l(w~i|q&yb`
z90pzK-u7=BkkoYUw7*2Fu%6Q<<glQr(h^qu&)|MP0AU3B_zSIBR#P)!7OaVzn>{GT
zK1=o#EFNrg6|Gj*=9hFMt2bwJjVX&(mWC@Qi~*)`?2bCh$8bVmnB%Vt$dvs4@>khS
zGll^66fQCG&crQvi3V}2E_W@htx+z?C>SIRoDowyqe*;jpdD~LCv$1BiPa1kCIIS?
z+hikvLaeMSw$;mywS4#>n}f8+n&?IxJm|?m!}l_eqPYkPO8|e=|8`$(_zWimaw0)m
z+vg}PEnVtST3VX46C!Z?vvc<dL^L|5JK$#l2DU52;Q|>cxHqMk-D>y1PSsoy@a42@
zY>FK5h2wwvHotSHKy*rDCxw9^>bg||(5_)MRqu8dT{}KC;lD}%CmCgl{zhBcK7qSB
z<67VpT2FzafTPCn43Gn<fJpbhD!cZWwyH2*w>T3RCIW7B6A0TFvW1t72@A!+0BuZp
z2nFShP*{;i9F{?W7PkO_ZL%~LCXB++0VzU}m6wjkhCnEYSfJzOa9}~o#=AvffWHHO
znM@O#oAf64wD;U|zw`Yb=eD0%uzA}du^jOJk*XE_!nEmGU7WOP;AkA{-BxS!XTZ?7
z>S6!WUrD*UryCgX;>;@;g0(07Y-3eHD9^mU7y|?{7aSiNZPlIx@ha9PPjC#G?4?;6
zcWvOb<bD~m%Fo=+j+v5@B5Gzb27N(J0N0f!7qTvb>58U7A-u~Z1tHv?dxplw66iT}
zIT*=wePZEKGYGwT+(eVNPlX3f@-PM{%wI!FNp;WoKFD%wZO{{?whNG+P3iu@9A-{t
zD~f*qG1|JN8Lx+rf<o`y{%1et=mDtmA@vfSujLEx0^ksV*p!yBe%|D^O#|<iVlU6h
z-4bW})z3vhR#HqG;K?4XV<|OyRO;NQ=p08Z9~|UYnOWm9f9u>ML}z11N|8K2TwY#|
znP~bBQ!sjCA4(q9)>?BG9|wclth|v3^j1NVS^VOU=w5rAx|p)P)pazXl2LS4OG`@-
zZUFyadwZL3Vpuflk%+vxr6(QAnIgxOIyCrfJgyNzFf{lsvLPR&=OkTFQc|+5mhgr4
ziulG<_b)`SbI}FNX3gzDP?D33LgJ$^M2^kJ3M99%>rmXkdg@S{S`jGvc&V9}>)Wla
z{`T|zuf()xyd?Mr;{OKKwre#NBG(6*zCP`*OP$@_gf0k96jlG;D~DX=K(Pz>(#RfN
zH+OfL^}(O*lBaUa#A-*32PNS=v5kGlBJybERNJU8n|Czh36JVnz4}H*A*SnZr>sP_
zEO#AcElB(&BHXcID3K4=+u*8*9)`0aaiyGjzhF26!hY_Z+}sDV8xDcY&`BAF=GtK^
zOB-6(DA2dUl>dHyvU^r`W~Oj#;puBrZn-sB*b*1EqYD{_R~ygay)wP7u1=c1nM5Lo
zC=+BVqD+dN(gTuy2(}22dSM{#l9$4CsBgLI964hjU4rlt%H?t?u6t{9a!2trJ;;ed
z*_TLnb@oN{kdME7;D9*1gP%o|4di^q_SR+<;zg-p8DtryEja_%J#C}<24+k6HDBXz
z>2M58TMD<!61XAhn>KtkyCHV843vZ2b)6c}#%s~i)*ikW?Rmq%#zoJ}gTTLL;n&m$
zkNQu6m{qxWdEEpOl_`4*=8+2&E6QG`7vHOtx3$6UyJbVZ9FDQ0_7r><;h1Nh5lwn|
zF$6)yK&l@`@)Ad?$Vz%?jdfoceyj&=Ff#K~5$#$N1k`)TLWj&;SkES!eJ~@0Cx}#r
zYthn~gNBsF4t=n2O68}SnVE&w;a6rSX{j2i<v~*ab1(KOr82@gcK+;sXRX(&?M)b2
z<!sp*28+dFD`i5}z@4iKNvIY5h*MhAvE#=rJV;lP&1T5b!Fge?OeTk%u0g5#`x8k#
z9uL+9SLZHqa8*WxF;d!f;cJdH#1#<(c2|Y8e}c!Si2h>614GAZ!>>B+JWs5KNG8@b
zJipfs4Xfc3s1$(-!#7PX=o=VZ<oKd}dlj{uUFesM-Q8D^{r@s%{~-;z|Ft;sB5AmL
h#uu%m{Cne}o&EH<h$!8-u-$bpx|@e<jZ4^-zX93U-wgl&

diff --git a/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-48-1.png b/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-48-1.png
index ab8c7f0b233cebdb92483ded057ca8cea6a212ec..bda90f0e9785e51779eeeba9e0c95fd15e5357f0 100644
GIT binary patch
delta 74336
zcmXt9bzBww(*;2RNkO`#ySq#2?(Ptf?py%@>F$;i>6Y&9?(Xicck%hXe)xxb*UR47
z`Ocg<=S)BcOnEy@8D0Z$gReJn2V?u_U~`>b5%%)~&yP+@iz#ZAFIr#5EQ)JPEgvbX
z|A<ygs=(x4FBe*zp!|Y^L*)^<9v&KMY7sypinCLLO}{yC(vv;*8g^*jN%bJR@kqb$
zYTnvio+72_<h`m7D&Rs)cFyc+4KgXAw>FrxU^9#OX7upqPP@(#cyonTi^IRJ(U;)7
z#Gj{BYLJ*1@7ZX=j6Gbk71Zp2kLu7i_W1^aFpZ?PeMI5e>an&m^Oyhef?MYIL*7&x
z%V%8_Xd#pz2|6#&PZtyzJl>D4QE<x(wKf|^M>@s=)N#R<GbNC`Yz2uS0~;+b;dora
zAsBRy!B>4_C2MPIz`GWuWqG^jgs7+wanvfm6mumV_a}qi!ecN3&z{}k_)_5KnvebZ
zQr@FViT_USbU2lZ1(>bP>(kVk{s6tUm91JhS7T3Km>^B87Ud^x{iDH#2Cs47(E2Ep
zm?dd*m+@CA!8#ZUyoHu{`jd#vm7U72-Q0xHJrON+UawJLgA`DqIjw#trqny^w~Kd9
zpM&CQLD-gm{o&S~ga5&OZez^OMB3egSU`4fq%NB$hoUH~Z3S*|sC5t)72TFB?e77H
zMuX$~hUu5i)4mq8(O-qC3kn72>eXL~wbX{c<n}`uaj{k<ph~cJNB;ToTbfxonyc)J
z1L;Ups5F8C7&B`9q1Jr7oz}}pj%0Cm%V%bZfOyzCgePsHxX!?eXrfGxDR*L~@8QCa
z#U67^=&fS7a`k2b72WQGEQOGZ%aR{&+GSabGY;E}$HK`YzYX+U(%yLQS~9!*kR&xp
zWE9LzMQ>ANe^Q4QmxC*fRXx&bOeuzeMV{Dt`A4!&V9DsY^9Bx{pPbLY<_#iV+Y~~m
zNo3~tQP$(7#{ToSMzeL>Fru=0Uxx07qz)Ub1+}Q!Jjmg=E#m(&D8udi246t3)+4XB
zEa~}Sja;k5ky(7~1?&3>wp~iJz?1hoF$+s&uXCYl@h_8s_%F=_sHo8`+AZEsZ%W<V
z+_o4>0gX!0oz8;%e6#0sWP!(47(!3|(Wu6N;7U^ax>SX1Q6}d@b%+}4rT*S15-Afc
z;jz!6pHY4=3+482^iVx<RXi|+h7SFnIHTFGA`D5ons~e5;4ofyKSC_*JYopu_+q1T
z(Ah<efpyl-4>sYX^1G3_B5h9D`>ZgQOBx0;K=?(fAfN-T>=DTf$!1H|7)uG`h{1tG
z(da~s$!Qi>@w|5m;azNCysW-*FQM&Cr9qd@_;b+C!r0R7X$V*2lv$PNg2JZp+0pDV
zbjBilyy$2lZ{rcU4BnOAp5!c6n7tBfA&}lVWJwg7-29%zh)<Nhxs^V3b|_TACaYfw
z@XfDt4A*stB^P}tUC=n)!B_4<)u^+@t1Zzn+`OFXx?ahVL9QuDh1W95EKG?FomBOG
z^BmMO+Zh``HYux8I7R#5{gFqNA@Vvagy!-Znqnd!mqxkl1hS}cpzUbjuiF!|9>Zct
zn9f`><A!M`S@T^ZiNx91_}dU4^vrS*V2zA{n8jcYe|==pV0?8NA>)y!m>W|dNiv50
zq|9albruitXaL>&SPC@_st1j!vaG%I4A-CCe6)%r&lH0aiJ~=7clDP@A5pGmFxuDg
zd;Fsr%zp)8G8CKTJmKl-DdgMDL1_!K#pG{c%N6Q4N&e^CjmL}?*X-33M+d;c!ToB#
z>;CFMqrw=j$>S!^?IN0i{3RhN37gO3+PEVC5&G@hvjK({M%PpQ;s>hWto26QwKoL?
z1y$eXDte(WWBM)=6Z227&?NPeu>M?py(|gpE4<59M1fWZ!zkwb;veUSnbwn$xyG}C
zk-Yepk*P9Iwx@KLo1(Fx76O3y1R?QRNcrx~H9<8jmq(<((mm#H1JcI7;l~<eWG+LY
z`ig9~e<auXbRwgTt<nr6@V^?5JS}_}#F|pUDOMLXHpyc$`ix0chr0{WGU-1(>`&n$
zRPP07qbI7PR~LkC!FStwzSN`Ef`v+h)hjY~@v7o`#=baQQ+ss7r!e5tu^xf<Z^RPQ
z)JG4ff#w<7sK<(+Iv2mB>;{RxB@Ee92c5l@U#Onwv^h81>V%B(c8(`F)3kT>vsTK5
z_~K9GM?~Z>)5{b5;%sG`7@NO$CmpbAkhX%Yn!Tnxsp>3^5LTwYk2J!|9luHJ&xJmX
zu9xpf#x)Qcy!%7pCOHI9tW#R`1(iz^p5?_ZTB#vnO!<(1A$5VI)@rZXKTYDGKEE~J
z7o^X4$e*>Ew3>Xk;y$#hit%l1EhGp>-4iX=VtJpXz895dx(2E7Zh0rN&^N<I;=tU}
zh;?xwPqsr9CrlROH6JiF^z}bE><n+smK%K*6N^bpOM6+8D+W|!;^VvP9ClnUhB)r7
zX7ny*4yKE>%AckS)gss|rz4}HAjR1%8T6W^W5{JVfoEO_tNE%d!>)Xo2+x}(m%9Nv
z%_{xP-l!jefid8}7f)SXg1)}KR34Y`xk@t@o8`v*`y@6?)#vj_B9YrwKjgpmTaw=C
z(#5JV$;tge02;;ZZg#}E(Nrway0nZ;u1cXQtHtEcG#(dJLc$E-(ZSvx77i}UYN5yF
z<5Z!VTp`xm4%cJ2<(!<JtD`@v#rJq5U>kC)bKHgi%YD~mdpMF2(z<f_E={pq1(jM-
zxDm3!k#bZF42Qk`Kq*-%X{<T2q4mW?RfG_^U!MxV*5WlckCmqySYNYPkI|kX;H^>b
zV_`7&sIVA>5NCarp(ul+Q5Y02%UDj`tG~tpCGnGnpUTzsMiNZWx5IgiwdAhm>O0jH
zWS(kh-(+U1t=BeCQA;EHGHT+ZlVtF6jI2x6GQ^Jh+Y-i&b#c8jx|_5<bg`cZ7?vpC
z>vsV<oac^SXT$|QEjdnjI=p*lW1WU6Vqdgl`Ju?|6Zhb~ai;B^S)-!cmgBXNxI6Q<
zRxAjNsI*v*j8NB)g=x&NHb}^F*~Yri`ENvX`;Cr&-1O4s5L}TkCxu3cGPKTie)Mjl
zi?bMw#koDgNR?=xs<aQvK5n=nq*yYyCH?@!uZ+?6c@TeyU+SKn428_zPUz2}1^b7u
ze0(aOM|`<a&-kw|@VvadWo2c8^a&&}eUSrOv-9()C@8HZIt}MD*qpYXz(k>~7>Obg
z`z24gyVSrjJw4rP%p4jPR-jc&FC--NUD?ukuENB>M6cZNOFV;K`pbpyTlmwft1e)1
z5idzPekH@t4+;_zG8C61{Nd(AU@9UaLeA&M{{B>9fkq{HUS8hagPECGk7a37HzDwt
zb(F=To(i!ul94Z!Kp#t^M*nh~l9HlSs>=uB6R7WDVKPBMPFaq-ajQ+<Puw#lIy9%Q
z_VLZ9r>FV4z$Gj)^3Ltq##_^w2jHa4MF>UL+afMl0EpdJW8ZsVsVSMN(p*{0!=n?>
zfNGu3RqE}duOE(<zb<~-vb0B^8=b(%G1riqQGfSR_&&b-aHTPJcVR}7LY!_ftiF9^
z;UyAvbfXZ1&!Uzx3_r7hC-_i$`iZ-`*R(DePxXaY;(o1=dR`7?%lch*4iJQ*`pl~P
zXXzoB6YQk~or~{a(Dr7{Q)oPv;JE8DRSWOD*B`pri&vFW81kN}o)|Z9iW+Ic)@fx$
z2?kprX-*6+OoxBCElAR)mD#Uxny;5f^ouxIT3w6cobeUlVD-&btn(<w5PYrmrh;v<
zjm<RT$vr2}rnUsax$6aK05kBtvZK;`v0>f<$)Ny*q*M6clZ;$%rdJaBi9S-Oz_FCJ
z1$-T(D^5Cnx<;iCuc|0FYj3nl)KG6v_BdT~D)=U87i&>)#PK0BV1)cB+=Z=Wedfb!
z#ZepmYqz0Z?`U3bB^r*`2R`s%&!z6}fQy|GrDo5um?ZDpwLswU>1KtKlXDl3ruDZ%
zwtA(hOu^DriB0NI3Wr%=G?}`lCdR^WSC{|l>MH62T&_qdw9Qa5TY+k^rcdC0P*SLu
zmsf`S(^$vMOv3h1N~KZHMwBqvv?eI<z{e`M_L5Fc&5PA@?(NFqu9nZQ?U^i}<;@Yj
zo(+PrLLZWkKziKZ<;U>DXZWJfH9q7e?s5(v0_)$(Zig$1kAzMm?af0i;z2d($NKq3
za}R+g7)VEtF-3z_!YR=BD-xcAR#a5FavfvIFH@Z#Jd5kluqAizNb0Z7-e|r&boA-0
z+)R2pUIABEZ_uzg+Wn+xk;kj7<hqBKKZR}TxM9B21$<@3kFM{sIWHz{Nknk=78tOu
z`C6Cz!fd>sVKQ*KR@P7)Zr*`#Si(uCp?<GMKzDm|%>QuE5$Qy?K#TBQm01U|c~%5w
z;@)lk!T0k|3FR&uBV@X*Uzc(PGUT4njEvB3kJvpuCH|!0rCr?OQB9#lK>~|W!(&m<
zsX7BJz{X6r@^5Xa>_}>r!k$Y#qz(09D5P$YmXPtU%Lmxw9uPT4OC$07N8XNLiD>KW
z3{_xJo%po@Q;!cX1Wq2VZ8`b*4)>Sl81&lWC(BI*dM*6d*VnyU_Jc!11uBIYCoM1h
zG-@TJR8&D4m8J&9%8x5fU~OV(zV%&Q)eis=uzedEctk}-Zyz35T#mJmmm2aF3VhhD
z=26hlI>6udW=aWF{wTNg_QFa@Nl~j7iSG=j-5ocaaC%(Ru6S!*E-9Qe-4MXR!9AY!
z5IP>sql1MaG&I!K**VW@p=L0JBjWdO%Az926c^D9&alwX+sg?#32|}#qxtH_C*XPk
z%)U<v0|nZ3%pm+RU8u2IZ1TV)BqU5*N4r!bC)c>Ry4tE;@y?cvqaNY;D1?$;E>`QO
zW<HLYBz<EktCD=V<$S6-jK3rsg4BeFo_>m;!&#Tm$V<lR9533q|3`*P-7cY0tma}3
z(*Xf96X|>azshd92@Tf>woJ?oAoU5!zrp0riWaINftru^E8I)vt54$&(>z$7ffvqI
zmaL945wCqd=<oZuztA>k>|CtX%LCPoxzK2yLg4B)qkreiq&T#xXDxwDf~}4%l*}v;
zMH$oQ`ZR>R$MP*N7M51%*&{1A#u}}TZ@e4Zx!eA_*SX%CQS!$7GG|Z)uwC2KWsDbr
zN0h4h*gb?`F&F862l?S8y{*=I!Z4+Fe=#ChWxvd`4XeQUWIbg(JJNwP0e7tCf#Q;Y
z7m-H{v$*|_V&#X6XYLWb5c2bp;jQ0lD}GZqIX5|B<*dFuCugwh?mkDm0*3T)f2OCb
zU*3XQv-Z~1#N-QzM|^w$9|S`F#Z0e<ySsANGeZmp-697<u-rQy&PIjf^X#LCgoPnO
zy+hD%{SJv!ssy%=u$^CH9I_4$4$EFQI?G;awLTicd*d{19%t)a`OD%?2h&4@T^@Vm
zxx}T(U=yOHqqDQOZ}09F&0qx6)#r<D6U6rv8=9tjyBn*gGN47j$p|$);`eFb&hUtA
zOE@@KPK7?ylJxcscMd+@EfU8jMM)kuiJOnJCJ>VcO(=Wua7GJmPOS=~HNN9<Mz*XL
z_1Bv$&A|M;SuHgVdqAyPA5rGrDJOXv@RpE<sSRE}%bt$><!_*6mKoR6h$($Km!R;X
z&bU%cnZtE!7tn-?nCz5ezL4&!T2x|(6|n-(9lq(E6Tp_(7lZ&bzvOo>M-bDHpj#|_
z<TJ?3d;GR$PpzQlCnbIQM!7^cqafsk{PCPh_N6i@vJxgJqCv{k{k`*W`d>Gd?iLM-
z<eH<Pl!W7KmBX|}EdD!}<voFLG{9!Hy^pVRIfRsK4}esy8Rl$oRp2P2lhJg_*G%i?
zwh?=!6u@A)?Oj>rmHAGAr}GYOx!$R3_l*!raKcAI#P4x_3A#^-n)62AOygEEPS+(o
zPs<VkE~O_P2X_BiJg~rCNjC11^o>41eQ9n%BV0@n#o*r6s4dh!TX36O3`2;)NJX>_
zCjjspppU|JrZgdyi_ryFMi5xClG^yi?6EnMBhl0DwHac0mkIe8q6MFsBqrCN=wkX#
zix*f0h>Db{6(~QdtE*#q+OyoeSJ`8-eZ%)omho%LD|P|L-+_@;G92rrdXruHw?eK#
z*oW{Hnsj_WFjb#D+EDV(f)*B>j)imI0A9-(BlZBoH5aoq2SaehbvnyW#aW7`de>X7
z@vFbQ{9*yH+X7zkb{28(leA0sm|2&@`FF(s7ocZ^^i?Mp7Y@e+-%pL5V2oEaVX=c;
z7;CBW9Rz}ljAMhPwHN<jGjgwK*xP#FK9KN>y$r9TzkQe#G!><$cv)@*>PQyq?5^JQ
z!7+V`F*w-`mO}*IXgcSD38AiTp{lxj_DorY^g@LkA))ZFv3n?MF%Bi(|5l7o^o)P>
z`EOG%Z<D1sJTk&*x`*RoYWhjJgXg)|PI1p~tvDQ1hw`eL35!c>3wvIQ>(Y(Ia{R?c
z3}Hz_Jkjp%S)MwDCRrl@r+Qa1W_pQ<iEZY$K!<mD|MkNIIED8h7j3i$@pUdX34z=T
zMm#%bLtMUfHan~JQko{B=6>2w8QIXS+vaL${Dr<^+?VL+kAdaoRN=$N%B2fGzQ^0V
z(l?<p2#ggv|BDrQe<!IW7^eC^lizx~CLV&C?~!HE0o>iiktZ4$NXb+KUCM^@5gl9V
zsu}BFzup|1iD)CG$9XcbRZ^DMY33T<+-IoYEU=D8ZxFWVTO$0H_}L%@o}^)dx$a>9
z4Of>@WA1A#&GlYxDH^TC0ktUdv)~~}H9l*mmc;jhCM)d%B>ZfTb#8Rf^m&4oeB|)(
zvQhUmfFTb<O25hhQNn=27ZYJg`q{lGycSCQ2F+(W#E)s#o))(bUuWJ^rHfwIyL^mC
zdXLWU#OA?3>45%C;|yQx`7yj`kabT2Ti?uAGO5RJSBv*&bJy9^f%=}R55I`nDA*~<
z$Z3{%X^hOR#I~ncd;2F!k!-NN0E*5fqptQ+fV0NsYNL)WlrX-xw|5ncybEy$I7AS!
z9dMYf{yuTK*U9hgGwe8FnQgxJ_&Ap|9)mohP@*Vt_(W-VA_fkl7TSC?6DJpQtX2~p
zwmj6Y$?y9qHhH||L<$8u3o9(y1#<Pi0R8tAb=pIn_qtp`3x*B?YQpK^9h-SX6i^PA
zDj9A1duGZ9=DX>M|6r)HY1m%3jYG>PIx+%Ht_}pf*$)u7xRMe0kc>;1n%q&zm*4eT
z9zy5znlCtmR6@m0Y@Wjznhop}V{OT?(Ue#o-z=cd=d5KP9RisT&XcVx$PqdNCG|HK
zZ=!H<H*~X-t4?69IK|H1Q@K&bi~;36@_t?UApsG~H-V^vSr0D-5&U9^Y`1v1q}@zz
z;<zAS4k|iy+}(ZSV9{P8sD6T+g4;eV%2<a3J&Q)@kn%R>p~Ch4lRbJ@L^?Y&dzR2d
zc&2jO8uZf5jyL;{JfodmVX(N=-x%quiO4e7M+k`B6bV~dA`r$0<~rpX0kzYFH88Dz
z-hrdGs|E^&XQ&-i%K1q6mZlX2(J=wqyK4OnLKb<-N|4<dELAzhv$BQ-Q9t70&nC>i
z)Kd5Jh18M0fr*V5qt}W_DT2fL^}A-VXtC<XR_+QOqp$olmM65$O0$;&b@hXYJb8?P
znOR^&1hUiqBoQMciU;7qCz=N9d)oe6lP8jz4cDHJ_Gj-z;i>t!+5x{|DgK6RuWYJz
zBdlkW7S(4eUif&`nkw%L5u|TVh^i}oax@kz5{C5O!rOXhe;}P8(AN|&tF;YA>3~T8
zBky2trzm=1K5lX!QLC=|<>mLi52;*@aT?jn>jxRv5)ymFRVM&bVwxrJxq6M?(HYVM
zU-V~5J^Y+o)!rH0cKkF*Gu>*MX)n@za9JbQ7WaAQJa4n3{^ql>e7S{=#gt9f(>=rQ
z<qYqR)U&3Xn?>BMR?R?xu;)f7Nj4SBw;$cY)6vzCZ65Uw2Jbc+^tEU&C$sR{j%a6Z
zAmC}VG@XvszD)q&@gYYx?>NGJk<7?|Q)w%VJO@d<)8y>SzlrZzO3!X;R{Dk|%_c+H
zy23Y)9+%DVR$4Hg>uGd;FwGcvwiCJt7K5z$+DJ}GF}bxi*T{^>d}eQ7{IMpMq5`+9
zTr`mk(o7dRcU|i?R&2|X>DnYDMAh+lOXpEeJ~`9g1Td4^0B51zar<XgwHN>GboXex
zgnOz?z}(CncnIe-un|%&Q+|)p{5w|i0oc544c1HC_6`nL$N6Fr1dq2H5x??EKQ(R|
zGh;vTrwnZpS#efdV9BCCmEZg>IJ?)NV#3qC=Z82!eiqHvI`^+QSdz=LH9oQUn3A1U
zLdOc&je2frqkE!3YPdPN+F+HW$AwrmgK&$gPHRpP$Lec#TWP0->ZuYw7FzdKrbyB4
z8O^)UWIbbti1$-pZ>~RGTWP+TC~d1{%h4)jbK?C_)2(l9{sm;V(dPz1k`g&!X!YU6
z3u0yIdxnKyCB_xYiAjeqCB(hgwN;YDcRm1KX5p&yio(Y5<V#!80b?<DZ#AbF3_)u%
z3<#b?hqQIveJrw0PBlPsHN;{L|F;Pu8(SeRqM&K|C6{8s9+N&VyR&Q;`wK`?fww$@
zk#}bi?|D_}|HVoFGiA?BX-CgYhV@f7#@u49wbC2ClWvd*HyC!h)nlVND=5GR%57O8
zJt#QpuL%mef^#YO*rGK4F>l_wr#@<Z55}Z1sMqL&YX`92>33A<g_Be@+J2hFrq3_#
zQhiH;9_AmbgWl~>wh~}W2SjDzEk47jSg_}aL=ep%<@^HzzNdp&kGR^ErbAz+i?9)K
z*=bK33ks;N4rcPbpY9WZ6imFiwUd*Gj0{3@@(3|@tHr-j!RR#hSNkd=ft0Yq*h={d
zmhK3tg4WJB9y`@*LYwAc(^6q!`Xi|`ebV(}An0eq=_8G`lBG}gWHPC99C!@1v=#4I
z*q(?zZLx)w>c|fe@CqEEuDJtqyu#y2Zv1Dp%^%D7MX%25HEuw)`%wJ>>c$%5#pBP*
z&pPEls>s5M2KJmemaYi{ZLU;nu%1j#CP8Q-Y&I|apnCBA^V?83KP`=9D+*$$?f{|s
z=-T_xc<Vr*OIpibJXwBU+y42>o3O)hzq=?1Pe#Xg*6IAYSh?HOss$abV=Gc7h~xnx
z@)f2u-84Bj?MDFZBTh@3mRHZmg~#(cfYZ&E<ZohIj>UXf5uMrl6=f(RY)c@GTE;>>
zGz%?ZeO&=AoUFabGBR7}4vwvi3&*=YD5@MzZ+Q81QA}N+;w&PGv~wv=big8d(MvHC
z%2ONCq{LTXJHk`Dwiab4+?wKcQ}>iLLG<|bg3b;+pt<B7Gd4h%Ea+|#jUn{@KlC;d
z-G<f}h=jks+UCc}BNfxgYB?QRR>oLSSy=~?&zSVupIBI8PET!hfhQ-3!=ocEr~Qvy
zTwFf_0=6jv!P&rk>JR#Re@Ii1`^=I);B&>XTDkvO6pQ6NV%?ThP0}k#^$6br>Qf;A
zPqV=T$8%2SkBLlkF}sIdsaGtHfOo}Lp<Nxjn(8f6Z_>zEM?H%Mlkr2)_Wc&kjNxNr
zabl7dMnf24%SdE2aT8zeW^B#k;HnPtBsDtjbE3jGB?ovK^1+pGGcrA@v1<v~!;lz5
zhQVUmL=4EvQj!5bAwBvWyTV#21POpV=4|h82He*%l|LNZk9q};f(lomF%D}pT8zve
z*U_sECj-4RM(FK$7_#k;(UUGatP2*q`$%tDKTDR;govju^?sLFEBl0cK3R3l{oUnV
zC{QDmkO@{q#}ff0=2*pO7w;U_IX#04l`<;L`4da*S9~_cGGho%;{Fj%K#MmC9Rlr(
zt)ej9oqAhJ$)>SMnU%^I$r|K0{C*}oALZb{4md_m=x}mDU7{Uv#2@(`w=^Kr-Q7(d
zdTaWY?NN7vzgsafJ-N5!O2DT8@15EteceOvg0;NvpFUjoHSwsS^N}$M|5#x3vkb0l
zHXgQl-}o1%*KP^I6KNp2Af(~2)UUCT7l*@IT3#MSCYf~&<fyT*(Vt;*C3Antq_E@h
zdnVR99oRoVx`CWx-_TI5Y#KL6I75}WU6?L=H*7N0j`(U8=YevS=iOGq<^I%xTVQr&
z@lJ*dv`L_5F*ywlypobqXc(ey_2*IWO4FNq|Ds28UamPH=~K*_VjW!g)`6k5PI&4d
zX>@ie3a&SFoMcc`6w}guSnf@vdGD6{V9|4=a8_?Ak+rF*>gJ11d*uhmZxjMD4~emz
z>U%+w@vD}I;pB?K)ZAKsw(e}Jwk&NL2O@){%Bw2qYF4zruzM|V#$@;VA|g5!?!b*Z
zuW&`=)*%7_!Os}c?-(R3eEwf1_~S)+->72=d1=|QBp8Yh`qP--k=PuiE#4j;2viX$
z6si`eKO88t)1`b#XI85(Zm2<2u8I_;&Zj1#MiZV8D(>!_OkMVCSu1Wf(!C2I9djvR
zdd}XCSaDDlm0v;99?CJ+{`E$9V%3@*!k&cWJ|a;HNVpN>?fD(z0utKZDHqt-=2YWo
zD#ba0O+m`xYs(!V)_k8N-Hp2>u$4Uq&&Wjg+%5(V5K!|CS<Z|BY|r&%R6>&6H;COs
z1A$#yC2_X|(z2l@Z7=t>dxtCIXw%)}JH1oVIBCh5-+_so5WMe&W+tMt4<Y+bDZ5zu
zLXto(s=u(0zrX)*MgmF0i(%Vl^Jp_(=(N$~&9M3{Q?P7P0L$)xzg!cZ%L{pi^Hsd)
zI{n4O6W&t4uTa8i$#&1<R9k(|Lv-KpuxM^H{iKE_2B4tedS_x{Vqj=U+$`0&S&y#s
z0p|PQpfocJ3uIY&dE20Lpb&Jf@%7<cDj(nx4<;Dji9872$lkpmR+&$b`}+Dq+-@c4
z1<OQ&JXSodhHq&p0}&BXkxoNAzt_FKp<(-aSBUvU9wLP0bP*K`oURaxWP~(%_Jv2r
zM3D{yJKak>8q74B%Lb_(hb+cdGphr^)q^EhvsP$uE^(6`d!JiuaKWKbamCF8%slsI
zr2+OYrqGLnw4<*N@Scu)qUnV*42^R}cqA$&j%X@-lC2J2(YGcYF=<EXuichs&_Zoh
zjnUyDqIQ#`?+`G&y}k#vneKTXZ#E*?{E82`TW;KfNTT3LVg}q^CJI2rlOZ!;YImP8
zl%pg0U8>;i{2WjHD-{XU-w+1Igyc^zxM9~Nl=h!LF@4j!;!BAiA!gh2QHHq_kd_`Q
zSjpkjlhz5(b(%~6czq@7Wqv!K@=b>;Fxx}zVs)PGZ@fZgSY0j~qE=HM$W18~Sm=b^
zKixbUn1A%l^g9j`iA1ejj_>@uv?_E-SAZbGbcnBW=RW35pFR)t)U<4$o!oV7;(OfV
zxv5=b>(oY=-&|*+GBopA673*K!ux1WVlhY?!Q3+K+f_9DK4P7|z*8xM;9&VhOdp+$
z7g5|)NstEjU*pJaTZLN&C8RBehx_~Z*=EVOp1vgT_py9ALWnnS-kgK_&f{N5l8?|?
zMU^i08`;1ZJQk-tKSyaG3YcI8eP`+m+^dJ0>HW=!l8sL4u3U*)niEE=(60#K^2l<r
z1z$)}7-1Vjm6Ov$KG#&FI41L~+$HBN1C(^H+VY#>eDP{lwoR>4Y_x3~3LY9iT(#kt
zk3dH1p)~lN&wi^f>0{Dq^EF~~+fikNYe_2uu8e0)F5*%$o!!H5m=Th$YQ;B_GI@Dn
zC4`|l2QVZO!4ccD&DDzy=t8p2<R$6LlZansitp*tU4Cko*EnU4lqJ13d6!@9KbvqR
zBo*<%0fDpLb$#PNZ~f=e3-j(y?*QH}W^h*4GW5mF3Gul~anAIfKr+tD=Y7Qc5(A_Z
zwvNr=m-e$!?WhX9{FRh=E~hEoDR*Fh>yS<S^5W(hJ_zrnW5k@JZe%b1vw;S+@xMr$
zM7FVz1|k!BFV?-Q{mskOe$(evTvT+PU}R*3h??>VmH{J48&m@X8Vm2h<Mo5NwMkfI
zkw$r{AtSUZ5KUGCcU5h*gCB5A^6Xt~57tDK6j1nTY%Cp@9QIqWQlY@+z+_|fP<H2r
zpMpZ{kB;THZ=nU|b2x~95R;@u{rjtb%h$y(d+|NvVQ0pMg&k**NLljivsx4b2qW^(
z<(15><(2HjJp5^myjx0`{45iM@mq|mFHU#m@c+C2U=U)ElLO>;1E7{YJ2!{Ez)!Mq
zX5}ka@<6J!uzo=G4~yg-;`K|wQ`OzO#M53G?Ckym^nB&}6;U_-+FCF$x!6QTQ--ZN
z*<bwvrPkO|ZK=ZK3X?U59L22<1opwFkxHc9D`EwMU2T0~@A<0>Fh?;Qq%5C|Xx`Sy
z5%8Gt(PII>eKxg=tnlAA6_NotRuyZ*3f_nR-Jx)j;VHBYsK(Bf89+g3)mm?LAPdme
zJPoJuz!yD;t4?oVPgqX=Hdv(uIn^I(abRUNUvn2dLu9iP+3%R@(0XZx&WhDW=IQy<
z^|CRZ1oe@vw#38#&m|sbRD9a~S5|YA=N({I)btEd>X-LoAo&(HL<P&lT)8PE&-A09
zroCsD6?pi+E<bxMKN!E?TBO;>%=W3{b361b3bxTs<<ZDg-cam(5qQTVM$PuwlsEx5
zZf_j(znd>*_QPDk<#n}vKF<{LgGEH7tFZd~8T#qrX5__fcQos6fWu~aP|q5etH>2H
z`UrvnU7h8rr<^F?tiAiNZ&LEDQ;v88Zt~@l_#JO4Vc$h3<Zs@%kzg$_d<XgRY_Rvp
z(lh>ImM(Z_<ds=VaZ4z(GY|1UFM;|8uLqo0yU?xA5SX%RhAY)idI{tVJP8O~)JuM$
zgWqTIs2|u{m&d9*fo-tJG8Y9hfF_a8nig+v=i0^4^@UXul@S9}eB)&|acJ($dh7$5
zmC~TJfC;{4Mfny981uExiI@)at+o5-A#U`0`=(ZB#k3<9iA-AF^3td>NAbSgfE7!1
z_4(c9i2tS#c?$%7-&(K|Iqr`7fI35TN{SRHK<ohvwG3YOms!I8@4<QNc(JYnlsq+R
zzG*zWzv}FufN>`P(g%FN-BIldyJI34-F6yDdXDvK=aUM{C1G_B!gmLp^)^Rd@*S2(
z%jhw2a3<le9Ke{(2T%J_WdSEaH;_B&_K!8#)gu5Wi#|HyXTFYE@36$?OGW~8;1Bv-
z3s|}#>FVvzS4C$Dp!oP4;T@}ErLCX4J;>Pk07H(I@S||WC@2~H)s0|(xlh;exH&Wi
zPhPPH*&4iFD0Vm5xB9yXONOjv?mkxK<u2q$j=Y~ug-wpggP}jsAhY*Cj27vX3a8A@
z^K8uABJS~(Q6SaFrV27lW;;v*&6^ISufugfY%MTzfh1%pYUAj=Oaxm>=?9R+^%;XE
zlU%_H=dJ#@Dw~yb95##9g9E7s=cD(eE-4%~!=OeGDRB0-4xHtR^jgxvv2LQwKtv>v
z3jEB*_GN5F0wo72Dkk*!JTI7+ytZL=P|4g|22$6cK0@<7vEKmjROLG+kOEWj(H{O}
zE`$Y^@bDHwxMm=Moy1jP8H8rhbhD?tx8D87L$MCRDoy-{Q4l9*!$kQ^nPJLuID*OX
zt*)ZIi~QT$diS-yC9$F~1O2zRHSg4ME5TU^Z?U;(_j)9<An7e1jBISbk|!bDdpa?}
zLLWVAAxZkurZYgd+y&*5;4!QUbU<}(ekD9?bA^@cp6W<p*MYHUY5v~~D;?V(NdnF_
z*C#6@K2R{HH#ZCwY2e7}aecT38nZwi{G)5J*TePY;ap5Qp9hBc<G<?FjF-B!cQaQz
zZZ)7p@<}oTCys+T$LW0X^2IPbQD27AfgVU-9%+`5Y8{3)z4H}g3L2$I>d&n}3_GlI
zp-FAQkfE$`yX$A$dtOtEOoEANua^20kkfvM3*mV_MLakdLeTXG#H5_L0&_{&r#|_t
zn(7mGtB>Cd6jdSz)VV#df{>3bLA%2Ks1Yp42no7!4>p&$MXMne)xk7ulsQrM0D*d*
zYTqtE=4<S3Y2=d(Tug~@qwvKizi0z*Qs%l;xt<r=xXu44WR(6K#evj*W9Io9D=BbN
z0PD7PlY3mYScLuMu0n}+ov@_j`?j_=@Q4W&YNWg>O>HRLS2(f_wJ<`8^qwEydu-(N
z{RocE4_LBeBt}rQ-<t&-D34aKwH-Z#T#-OyCJhEZo(g{>>fdDZ86WL;diURRKRnbv
zN)9gJf}RaNV;o4j72k(B#z#HQcuF#&kmy3+tDCh#n#eS<6QYT)i+^qvweyjzw%yfH
zl*@g+x2hKizu!Bb*5C%G(oVnJhg}~eya74V?x7{$n2-d=K~W%3VhblYCiA-nsN~_B
zKXIklIn&*_B4BYvxI<d4sm42o%Tk~N;r7a>GdhV_NOjcjLKDS6*HQI?(0?=3S`xD>
z;&S`!?5uo~yVD=F(q~$Biq{iG7TsrzfZYP!V^2_|#f`o#)*kTn>jrJuy#sq(^ry4;
zAioAf2(&JJ6O^>L!s0x;JO2LZ<N9{B)MtMfzgn^?+wo#F9<<%<0E;U=P#_O+5xA@G
z&9krj08aa@Cl{h+W*vr!7(`@-T3iw)gsIZGQRjACw)|h}w*=^6>Y|wZx}U@Y2}C3j
zc2;Wya{Rq@lOq<NN^kfK#AwB?OV+cQ1oRF604D1VA35VMRvQFH7%d$+N&?gTh=@iz
zA(Od$BhY741Orfyjin^!M9w0!+p|)6(6+6zC5J?Dta?DkGYH<f!gyN$d9ECfJP<TJ
z?BPvSL=7vH;21jOC^W!q5DEE|2T_T<+|U$r-_R^U5&dKFKJ6YQk-ENSM;2`~(6`li
z14^rJt4z8g^z7I+bCe4!V8N+28#Jn(6)#+3i&DWR!*katX@1?y@nF6jyDvPkJDv?U
zL~Hy+h9Yoh@c}WVrLRAu+ks^rReOZ8q6w&!y~O4ZkyJT|#}WlW8wg^3b3d?F&A4-%
z&*6x8@l=eEesEud^$uGZZ2}me>}hw}K(63bxc_$R#cfW3wrZ@HWfdkPy*KHBb%K!k
z9YnaQ3!p+{bh{}4`}Gb3q)S?tPH7a~eiE}<U$GM+r1>PcAQsn<{P+EM35nZY3CqTF
zi`%B|!IcPyi+M6opcdMkf+y{iVCmthb6r>J$PgWBkOx2}S3y;ra^m=NfCN07)#br&
z&~$-i$Fcw3qFOWi#>+1czhLDr-%R_k-SMuowN8t)q9+14{DVbMso42bn5J`A<*#iQ
z$FAj&&q;;f=bgLcQ1gQ{*=@;#8Why}r9sfSX_>H0;+p+H@gtS5_YSQgs0mX7l(CAw
z9opFgXTPlVwH{NP#<|NMNI_@^4oY7qf5Cr9kaIZRMY0wPh)zZED-Xi{r!39He#g+U
zyEtcBCHNmxlq>V!qlBcoThkdrKC$qI*xVFH6HQ9ud(o<UE2IPpYvaDzLEiX%IsEp=
zCedke;r_X%a7IbmveolG%0MCh@0DvpN#@CC-Z3%n43p5LI<hcL$1H^q`rLZ_x<y#f
zaCcffQ>0w`|Iy{ijb9xa7}z*n>o{Dji-E_aN5RJr)6`iEMT!Q!)^V!huMkK+kJ7$E
zJUD`!f(Ok*OzqoYYGfaxz)Jx!Ps~D8R_B+M_O6bU_xwOLBO4pXt#dkhwWbUx-GbK8
z-mp+xTjv>0$h!&_H%DD`>L(tL*}$N_i-ukU&G!!%#4R_k$@`kWjJeUPVLyMag1$Wp
zZti;4`hV6>OjS+EHxKM|YlCaW+Q{(Kr^#VWEQl#izf=;QC(UOiOOK4r@-7Ar4YWDS
zft1P3+kD+x2gh8yLTd!Prd59l*QI*zq^Fbw`{P+8g3l}9|2OLbq4iuGs4#R-Rh~Np
zpWZpp?hw&+Zi>gG?MY_&c4~h+N@HSYVw1MzrYsY<4<4eL&Ht};eQG2HMN%0Fi7X+%
zcgDuN>DnDDD}(?5*rgPLR{*E|iiOJGwAeDPT4YFQNL1X%$DkL7+N6<!>T;H3S4T_G
z@#d?2K??yEvq6w-Uq72vv*|=d(&iRbE(L8^X5(9RT%kXx$EGpSSOd*6Ew;7~%=ZW>
zeOg|kxVAQkG<<A9D?8q8QG5VjGcGs^rOF7s<?UTtF19@`PV5Mu0^oON9VkJ5wtaGN
zy{`L-_jcN{du72pFw@v|NZ}@&K2JFPzHX)aWL*~y#>^$UL`=k<e)d4&&zd%KtHTP#
ze-A>4k=2jF6_ZXgHe<0EG?GxrrnU|Z4UMM!GeTj$RsneOo2^TM6d`&)S^Ci_#74SY
z&HRWQ8er`+YEk__E(;9SL@klt(44f@%M9CmxMZ&3{wW7&h?RG5UqNzr6U1NeMN{7T
z^NACEdRgSo24?^>l<4u?yg9Y0P5<E=<&CvJ3uuG=b)pTWW#l>orOD$r0f@$l?zA>@
zp0oFqISe6#TtPJ4MeW+^;j`x(Ko+NY8aqiF2*3YI46?LEPLE19YP1lLbPYI++(&(z
z_HQ&DO;)z`^yTmxPYSfe)W(UJ(Kr|g8yhXL*v0?%wOxl1ba!o*>YeTea5=0$_UVAH
zuf^r57Tg-x3sBV^SiB24X#LqQe<$#2-?cj$Xe0C<9SXagB+c_03rhKrzzWPyAC@oW
zZk8nBk*oS^uw^#uv~ALV55jwMXrN1nrGp)6lFI!z3gpJv2W4@^?G0&_0%~HS8+@+I
z7LncEdD9l(+H&|kZ>>eRuUZaWl{rX1G@^mBja;o)&>UC&=>5?1c$aUSi_^iTB6X@{
zdtX<`_R$K*SI_$0^EUBKdw@jUg%I`Qyh~($K(3-hMD2rDko5Hma{9mQWvJ||VvRrI
z5iB*YJjLb3i*R|6FMjE1<iDpOXVB~Ox17mv`RSX<0X}>ZV&e0JgUJG;-Q(BOx#nMC
zHRUP$R4E~OOkYGt<}77M>z5UmS&A8aNDC^gU`m*l7alGERX;#)kWUbT>CE?dSuz{3
zcU3MvHh=uI!6XC{P{&L_HPQ6BbGfr;IqKTP!f8k}n>(d_F<E{g;P=lE+BCkROFufJ
zWkOzbaIl<3g$hhBY0)m4#Y?1@aa}OG9b{J6Zm%?Sn!mw9NOEH&dt+afXe)#G=#DEq
z@f)-hvI91VwAGyd>ItP*h5bj*ix~^*8~oPHJAtyhTe#=2dI?X-_fca^crP5WNtV+e
z7(Jo!yXA;UnfKiSX<@lpTzzK-255jeP`w^qdY!G?;YrxI(!tXl)6XnjT;laK%CC#^
zRc`Sq<xum%C~8QV8hrr3qIqJhLY=G{1FU%vk8;OKE!3GU%bXA5qd|XSo#!!Of%7RA
z;U1PnzcP}Z8Wa!OtL?7$bafoQ3c=p~%<wXlZCrfH#-?@qE7h1rG|k@W7s{N?@#cs8
zs>9j*Nu#OB?;jirRC!R^Ckz}ZJ1rTb)v|o=qCmldM$6+BeMKo03^LGy_2&~P@i^=V
z@zRNgZKcE&_6;kH);c5BG*0+8bO?}p4KkvNArXBQw8^XeCP~Jl_1ugiy)hzMvEJIY
zgMZI5RiwcJx|V~&!n~HVVBMKreQQBg@NU4|+&tj;-)4?3)&s{2vCQx3tW;AHmlvsb
zo|+#vnQMv?MbepTM@<0_c3NT{7JO<AVKK}lU#}iS?Llz~cX@KrGwWxEnyOO;w}Q<d
zHHXrxPzAaN@d7O@`o;_a0YfXBl~e;v?Cd^dwW}e(nlv-zOU#}MR8$h<$-7Wk|Icqh
z#IvzcE)ZX!_>MepTRO2N&B6&s<nud(MrfL2VeR3GFqU)u6)50#7xgftV|XmCPvEev
z$urK3@bACWAJP7%$V?|1PIP;ZnCx2bu$RsMIA1`A??t`#&D=+uykopGaegZm3vE}f
zx5q1Sxq_;{k@wy3Q@Wn>k=t^^=^$mCq`d?DOE^)ZI@kd4xgEbgpEr9wAS^Y#etsk4
zFy1mXog*-nP0*C^_I55BR1o5%DBfGyf9<h1&#DY@Ceg_-CI*i{YkF9S(?N$INV>ox
zb5TvrTdn{05IY+1DNntiaLyi_%5pw{FHza?9#nV#0h-l7<{;4#-5pvYXp!o}sR`C)
z9fY3<_*ILFj&3&xGIBc0B#4~CGgv_LTHXs_|C7E+Og|^KL(e5kt}M1U%@(~GDR3<z
z%PcasEPN;n<jMwh++-yY{_#eq<4W6GV;d>hi*6_*3sau656cU;oU`Z+;RR72jNUxW
z&1b`R+jHsBYPO#NA{h#uzWP|;($uTvxmYWgp(?}k91fX)m+_MPU;iFd0yd59_-JAA
zWLmWX#~S3`m%)6Zy;aX*SZPA1r(>;{vEf0UiFn`gTTj5f-hK}n#AMX?StFkm;Yog5
zil~ZG*Out9t9vp;gAbPlaAG9MFKvIR9!{i``9zB-OWxk4?3&pM3XmW~9FDe%ay(=!
zk`7yZVn9=Jco%-3{v`z{8Lz$(2Q4HWX|%AC0ZI}mQ9g0Gd~`fl2J;MoiH{oupi5NX
zr0idXa1~X><LoUcqq(h%--zH?X_XEc&$4k<GZtEir@T17^&CT0wve&Vh7+FCKNU6}
z8yka$hGsPCMj)3>759ISp2%W`1hN`!!^0(=42Z|a#{o!y#4D0XugkgpYTh=a-9k$C
zJp~&A92+t3?X6{Nsu0y!o@zF%kp!Jp1H7MhsrJUa>ksM#<X30!CQoNtS}-HEtv`br
ztoc|L2WYQG-8cb(KwxehUHBGnSrxsvvXsW=zD5%0Yo7o%c|osp<VV6V#e&whIef64
zgY};Y`NzL}v^ZC$&U>oH`A#_B6z*SsO;sB4XmPQ##W?&$wEU`5qP}H%aMU24l7S&2
z440$g+Z^+7I-iuRED}FI|0|~h3H44sTQrP#2J|xn8jY?@9yiBi+}t>zk<|VAu+J>R
zD<FGggJ%O|@5Dec`h0-J)!m&q0T3X7sy7+H;C4AqeW?U%J?K@Z)u{LpLm?NPkRZ0v
z9fr;N6=kWxg`up3{{du;2G%+P=l<G}ysD$iO&**)JQH~`xXrig@E~!NKf&aMj>X%i
zRc_ez=86ZDOYaUVM_6_B=t?4AX9ETwkcc*KTZG^kNR;TK?#ku|UPOvjeaf|O&Jpew
zq6}3E&Myzd<O2Ad&wag`*FpNv_UaU5BOcrW4N;hr5Wq8Yi04$qNB8<^?NS@suFx(7
zofDG-euBtLk~w|f6?~o>Tbwv>BM5+PxMfh))2#W{7lig!s#2%fGie}!fyHbX4;2;l
zJ*lNoe{L8yz!GqGv4iQ#&&nD*Ix25EoC;n|@pZP$;Mr1c`}q9)yccXQqbnru{e7>G
z794I**WP%Yth8i+OyVok1MV~W=9;uMU^vYsjAe^e+OD;~IRXi@p$d}$^_p+kV5kij
z|Jupnb7Xy#WPQDBUx;g2<d@&ljDklqpT7WY{4X87v(WY`ky*62!r|{_$=l=<7`gC7
z10<a!F;{*#^kvGUf6E|!iZZjXC{b~xRv}QMCotQR+qgR(AdtqJ^fk|LF&wLkcwL1X
zS3;}a9~tgb(G+KF{ec{$NO&D_|4MFxKf00+BYF_4L)7;=z1C_4Z}LD><r5zu1%rgQ
z+8Km4S)>sID%>C6zXew|4-db!x3^>QxjSaLbTerGjRa$)1X|-k+4GC4Dh7B(Z0yH)
zdYxe5tW0pP542!Ijs8|3nY3{7ZohecdiXuH|Jaja)10(&Q)bY?>3Pd&ZEc;(@0Gl`
zs2wXwhRLAYxv-Funfb{bxH-0p^g6q~HU&{!XJ=<5xJJ=}43f=V{r$mb8$D<rJ{a_h
zJdZQp!Y;p<ZvTTeD#k`TKfsZ~+y(WM4&1c+V}BFz6VbLtfqT?~WN<|MJX=KqYC}pk
z$PxBZ8nPZwW0kQ4-0)?E6tebds^et3F)J4<+C8>%gWfJ6D6xU#z10P1^8x3F+$SEt
zJucf{q9k+!9s-Rj2a?I6b|3LSMMUUywMV4=YfQeU_J8qlVFc;SD4Ooe+1avuixFpH
zq9q3U+q>J_QP#{i7wa4#z%!I1c=v{dg$3MuYXe`&(4^qwO9NL_);BjlgX9s0T&Ja_
z<p2P((_n@f^+pmE={6<H@!yfYT#4nP2UC}$;NlM0v^>QVrX(b&JzHMwPmxek!eU}#
z)>ticgG6J2Vs5bR1GD{RZ+?1Zb@f(p?XuxWI$zb-G2dLN1dRIQ!eLO15)H$_u9tev
z0M|8;;l*J;=*-fgS^bP!#gMNhRtHd<3)8l?Xygm0Ou~QbSp(nt`o=X;D%;V<Oo9wo
z*Wm{_lKdEIQ5Z;#DI_UMo5kq&(cSV8<f$f1!{F4qH*}U&Fi)tTw(x_taUJ(x@rbN0
zKlZ0r_1|c?*$Ti6lV<To_m)%(*neBW=?13@#Fm-hqlKOsLU1OH7TNd&=t2(XY2uYp
ze9i&wmY_k>&K861`Cqk1(KQV+ANThXgFiYJr3uaU=OPsnw%BKpIpt4Ipe7~#0q$E^
zRt;x~1j-jEdtObcy^U^=bKgmG!D2Rs1s9MRB3J49+WcT3T|s|7NY@hDIXZ4$j0o&J
zoV1wT94~o3-|e^qM^bj6E(bB&;8H4niXj~P_4Ta$P#pgw1(+L5>B#S^8+ohYk#3J?
zn)eV(u{J*R8AtZ53i2Yf(2}{}&bHY&8Q8OZRYUrtA<?(8KRw#^Hcurm<3Lyj_is9+
zi14a0D##*QKQ+^`=YU%i;EFF5nG4^bfivQ5Glru}0)W~?u)S(TM&{Xb+nV8hD#``+
z2u5>TOJBKTx*8x!d-%ZS8mb!SxxR404(e7(v-T?c;=ern)L4aQRRk$M*gxG+D(1yG
zH^0A>z+Ufi5*`!;*Aa-E5Aw0!KsOPO%W<Y+L9q7DnE0ICMz?{5h34*tLIoQwt#qL)
zAS^5lu?zYgqO3qx6TTHZ=vT`LVoc%KECJwRrh%1JS_`tOswxD?bgC43xvKnxE`9y;
zvC-U!S_&Wm)4AC?t~KvsO9TJ>7f~={jV0vU_V%6#xGpS|8!yT1Xy=dJysFav)vrIE
zG5=kabAWPzx6MYpjgjSv_9nFjP@PPe=P9o2S{6H~qmH}3fZl>HhDJ5UQ-$i*ue}|5
z($%E3)>Sh8)KOYiRpZZRXG>mi`wU;ng5eu&XIJ{U!P>7%Pw(}afeD_a>=@G@`t<Ng
zsMLG><9qAeoOOMxCuy-0UshsvYU*GR{#5`k-e5nJ0MWqt!AN%((8X6RD=m!x0oJs|
z#YJJl2y%H3dtx>jQd04r@wvL)6PO5ciI9kh84tZY*)+*w&FVxB8v;;`H#RojZ9*;$
zHj7>z>hE6%`|4*=QB-W~J6iTv@*j324K#jGP$(sbyI3NCD^pYcQ=LQ7iLU4sxipKQ
z38jm@6qFEwDwJJ_j;6NVy*|wMCd3KRrB1nHa9f3wpkFIVM2i!2?Ax;WQBAdphFHig
zEoSD!XkeQ^-Y=PTu~(Nn820|q%h_N%;7*kr^W=Hy?D&-evNw00`}SbxWYtsqUtEZq
z#xnUKcuEfvT;>U#QncfOIdIq+KdWnATmqNIUBT$aCMF;7@CJW<2?2NFwiaw!x;i>G
z3L2Bar8tO#nNkKu1x<qpt=cMF$8$=z2g)(6#7*NpkKmYziRy^zxrCoTVe_PueCz7C
zv}(Q)(bLldzu2lZ!-NgK_Z8|Hf^-$Qru>?T^VOC{=H{=lKRr8Z>*>iCYyb$oZ+*b5
zWU^l5jFmL`(DW*#w}<S2)X`$SGxjSJLr>;Pq36$|@uf7j(3g%0oU}T(1QGXCD^)_s
zG3q3R7x|u-uXz%&+j{#Ig*9-ynXy!`LyZsBulV)~{38;<zg={H2aF?kH04UnSFvUC
z{!Ovl8-x^Qz@w_!13;Th&x2o!&meE2tqRUis_x**`=C0|(A3-%F8x-7;#s)(9bI*V
z!v8|K#(G{X`!>NowH|Y*ui?gNUnEzR<s?+uB2z46eJ704YI}S8_U?|!aaZ2g&rjp8
zEfKiqE}g;OB*+7DP7uz=i{__T`;!Gwtt-Hbj&TYAJm0~52iKo5=R{+?xWG;c0lvWp
zHp6V~I(sZ!T#^_GJ!@1nw5z4G1kD0c*UHLD^SKH#a5tKim9=WNCMhXtFr6<2+;w#T
z7f{Aw|LsnshZuRNNx>7Ud?&P!%GMnr?dVC+Uqqk9LZm<?qQpkx{Z9BR(}wZUQwd1U
zDXOLL)WhW+ti!R7{`<fIzcjoDuxAab2=>@3azX33CW_EdfHj^I@Kno^0X+})AU6w&
z3lJVlHAb|4E_{%j_q(?rvuAo@Zaw!iuu>A+=O0EXu$r5Y8a7s64wqb9l}sL5&upT@
z9cyE2VpZPf3Jy>z1!NBTW+IpX;Pxi5FO-*%LfZ^vb8oa-EGD?GiTCppWiK)$b_GXO
zD3Yj;;#c@3nuD7jbNJb`9K_fM2c0HP(vi9muV}$!o{_qYC#tfWcaP5K{6Bro!dn9@
zIUDL(wptGM?s_RN>^};_68%7N!LOF=@*v58Uac0WjVYK=;q3lJLcy5|*oKvfB4W?h
z#Z)U*iGy7M(j@Po`2KPasue_VKf}U`<)!81wp{MPjV{BJ6EMsMhlOgT7<B?{Y;kTE
z+mNFzz>7U$?Oz9~-nQlD+YuIr?LjJ+M-PSEkFH{&SZ!ckfZUQ;X!hc<a5@X5b}erm
zvS>mPhPA6@!4AWl11<nb_#fBlk}V6Rnr*-DCAIuI3MBQGk$=U`6;v7;JgNN6QtEyw
zeaovQ=azq*qwVbE`hQe?1yogA8!d4GX+=6k1f-E}FaQOl8>CCR8#W=KlyrlXNOyO4
zNq0$iH@tPO_aE;)$501{!(s2W*ZOkK`CU+SknhO0>|IP2Gj849Veiez?#>4|&D$n@
z8^mc;GG>u&Noxuc9u#&sc;G<y<b8VqKlGzVfq{u|J=u>gd%1y~^?`J^HWV=w!x_5e
zysx71=Sr5;di3u?y-}2*aywhPUr$SVFSSy$$Q_=+0yRxwmTbgVJJ323W1eVxqE_@q
zXgd$EdtBQq=c~4WXalXtmYX=Fy`$rmQ9mJ^>&3putWZiyN`KrdL@z*<ECHPB-=m}G
z2SYtQ%VRkT;B{Y+Q!G|mL4ZL6x9EApRI7)!a({m>a5km@2g_$xyTS-6rFiVtG06BG
z-(`<JAt#r%wk{Uk_?`9^1&-!pLOTez(FycH3S@Nrr>8%GGjA?zwV^xMj6u?AbC`0X
z@DsX7mF4_Yy$ic(1#yU&!mfl%RU;fMcmm;WL2?9m2y$zXyaEoucdMY)&D^r^l8v*b
zb77nXblV5M5BMXHHB=ZYIA$kr)aiP+s<G4A`>g~u_S=J4<Kg+OjS)Ki4$rHq5cb4u
z!AqV9_dT52`Z?RP?z1E8rSXd<-P6n=p4yEE9s-A8k`F4fdE?IGr?f9K1q=3f{CxGt
znZJNt`Q{Nsxlc-5kr1u0`12&ss+jqju>n%0L`FpYADCcZ^3LsvW?Npnhb<=Z`o3^~
z0RM#^0=~4fpETt_F7uW~60e;<ox;y3XGm67w#1=~d(l0zz5U&Ke|#J05J=8F3<*Ef
z?;Q*YoM3?5=TAnM&<gnbCSwF}d)3MyqmRqMa~(khPszvzfPXX@&s7SKOG``3SFh+@
zu>&+VbYP40yF$r2D%~zuf>;QJJ>9<^?K>#595RC#*V+wW2B-(;n3lG;)vL)sw7zE#
zyRXFe4Jl3Z`Npqe2|p|LrW;+MCq~l=i}jwR$BN>S_r1oHCbpAjvQ2`<NJLf2l`LiI
zEs_koN`1|0GP7Q+>}gcZ)$!I<kJe1e@?8*{x?uggH`8a1&Q1-wvdf5$@5M<2zx_O$
zWFW-H>0erUHgj{cyt@1Dr}WTtl;Zutu1lj3eosfaJ?Y+_+HI$qz{u+B#I;6~k>~!{
z6oV{Hx2<3w1O^4QK?4IJj(amlfRoAYerZigO8RX#TRtaF(2WaFM4tx{uo!_HI7z?-
z2e3L5I6l7t5w^?eh&cMq89vYj&;=F)wwZxkfbD80CI&Jh&FDaMT7^1OozsC409Y`B
z<mpqLqb@5U2(qYH>JJe>WG0?Lcg>rbzL74U15K40q60x3y%ry&Bt8c;;3UB586dH$
zfBpV_Ag;5d?iqsBYdb^P+)E@yYAK7Vvby*;dV1*<qffw+(7)K)7R9=2NRix-@s;>E
z@1--du$Y?Qc}9=hH9h`zx|{LAP?x$lY<Wf9eNlrz1-leSJ{oCemua&x3QB67duV~C
zshgf9tF^u{{_0}Qiz08ol!<JG^2lBpZ3a(ZB3UFhwr!fdtJ-l5p|+mX^m6Ry>s5T`
ziNmsO8!0g=f}v#C1+a}OSYPCSwb~Og9k)<iIgt7Luna01%!(}rY5O%AV$>L*^?+{n
z+s7_H+mEmvOopHoC}tQLUI~gdcvY^>1hVNOS>ag?PRqHtcy?npU_@;XMspOhTN#cd
z)0Ojx1e{rYQLvevH0=gE1_wj`Mhw+@2Tc`+jF!e{2nQf=MXw?w>udaTtXxx5qdX`E
z%>K*75fg(BAE)VZ*C#}uJBkLk*OxzD>g&7Qc3}6_^C30&L3Q}170CMTScD&cRQ`~4
z;w6EzYFaU_b-kIOenlxwSIA{uxx~ZkcO=1w1s7}k!@*ORYx2<!j*cBpQXhS7WM1;`
zx6Y~-kF0BJWc0|TnfG~k>*7<bsK4@+#A?S5tska@{4p}duW7O|L2<<*6)h4MODkcA
zxa?M<(eu@~A|XKymB@%B*+2K{4kRNW{ULWb#?+jbY|NL1xve+2Sq|1Z$AM|VWQUKN
z4raUjJx~H0Nr%QsxUKDMI~VJa=heB4WjCbD%?V9*?EQ)Z$I<J7;eGMqC51c&^4n#M
z`3ND5jkPsAOO$`6&J%IKXI`XqQ}M^$&%kAAGrxq;e^gkbq~7-V;hZcv6L018<X^k;
zk$5HzjSb!Z_?$PG)mc&I^g~UR;5qhz*C^3rz1Ng>+Guc0GY%~0`|EqPcGs7dhDp>&
z3m)8XCew}LB~Dc)i@uzFMy>ZGhs6}TGTt>M)`HM${~2|Y?c8$f?pH67R=e#l*`ILH
zw!#TRgd}j{qtAxk_YHPXbSamz$*Zw?VAS1f=FBbq@^?&;?UVcL9iD^|zV%LgpO(*7
z*erI)s5PJu)pDfdP1h@;ko?<iDrKfS!Ukq+Mf$Zfxgk_d&hI}1rlfENJC9E$N2iKy
zg@zjL=>#cU3ULP2VT+pXdSdUeoRcExoVx)fpVehQo-hEXEL%Dm(El-ilxxD^Ee^K@
z?4EV@TY7u;Kt=!=7*PS5NK_RS6;MmHJktlCQ_X+3xXp59+bJ0rOq0cWttn_2UIta-
zvc^TNZ2S%~`1F1?nV^<!f4<|Cm?wAalB^C}4y``#X+O(hN{?5!eZ9vH&KQ*cspKu+
z8DVQK_2}|N9&2Wdiv@8hM)9jB(GsEoD9HHynJ%*^HZ~_O>>zq~G8Bf@mVf575mR#W
z%JnZd-mOU?npdL<x3^iTP;#N~M!mjpV0oa{ie@_MMG|tapokqzmRFYy&PNwCqZeJ!
zi1&)zI@}X2Bn?@fe}#RxeX|G+bKV}=l8J+JzFQ?;4BgD!ie7BL98$CK2wAmT>T+g#
zU<n@*CKc7}M@{kY2rah&#vQgX=Q@4FJUU3&4BjuF$1rL1l0E>$GQMRP;j7l%nm8#M
zP)Yl$Rn(=!s>7<9hMoTSO`z#=drQ;hvG7&Ly!froYxJyW5t}<mkCT2=6Y+i$=lVvM
zjaZHK4oRBqu5L{Kl$b+(YqFL~Extls&h;rH4|enLh=OdiHK$bDPYaXL8Vbz;ztj~5
zPf{<#*k?-OlZ`U6d6pS^0aN@uY?oWyEe7(T`bGXu#BFXZ!^Lqc%w9pr{uBP9Olc)S
z@}z<(Vg-C4hZuy?O$_@0ziZ|Tyo60v0ScofD)tpV)4q>wXqx5WheyGfB1Mz04jQt*
z@{||!uYw!Hmm<l)buu$sKEY?r)#?eF7@ue*!PFH~u&&UbzeuHL|K<00iB|A|kHjJu
zAS_TuZ_X=flrS~Twf&(d6t%z5kO&lS5SG@~L?Pf$3KIH2l=w9`*kJg_YvmAzFOOID
zjmCE3@2}e=BJW>UEUh)Ss%2%BDzxNz9~R61xjT&ZBn;`q2{236bT(>bTw@tcUzT^g
zisXP#P4-nF`iJZD$0BiwRHobhFKRqIrn*<umvxsyt<={|X2sWNgN6G7SHH{U0~Rip
zo<WAlIcRaj7}&0{2ZWqtI*#vkDHb{h4$6K(1)NOrCww*~PlkKRgyg^F7|O%Hy<M&l
z2}TwyqrnEf@`|xfl?B?Jm6(0eM6Gs~LimQJd+@0oh$x(k9`{^wwC6Ea!mzFKh_YR<
zWgAHgl&(B)kHkejhdsu{Z3mJDLDJa}18JGtg{79uTJ)?s8;s6|rxjJeK}19Z6Hd@-
za2=_0vdZhg0|E1e2HOFq;ZgoG0_PH+yG5mDNz5N<?7Y)N?l;qkSy<3vTKFl&JzD10
z(s*~PP*s-1cw5s@bEjFg+tuNDl6D`X|AZA`FT=yV7##+LThL$38CB8?<uClRsQHE0
zST1K=>h_Y>-HiLCL#%epy*OPkb{^)1ON<9|mF@55n~%-X9rt8A?`u3O^lnsqs>{O7
zuPk!p$C7b08{a$>kRSVzbAG4`?@ll*lOX-s*QeXzL)WXlfGYp39&cm{sS{a;Enb!D
zfl2MP>c8jbkz6-e5-$j#>F_JH-b+ZVOdF>*6g1CfNfHP>+@7>Hv1rxD0l2>^X8_1E
zbgit0;$G=zsw9JxG;6L_r8w=}=y}9M)T@{6BX<TrRc~Yy-zI*yqkt@Q#?n39Hpb3+
zg}YmeHC<JX#RhL&7Ui?UF0ONcdAo!3%45XQ-E^EqMg8IeA{yMG@CZn5@cm8Y)NeCo
zy|@0{1S#TJXI_cW0k^R!cw*YGr%ZXY)^1ULH9PA?iN`8h79n4{bd4>2dMG~mF*P@T
zB$HWe96{SRuG>>ynv5r6W60R-u{6$w<R*%@M<)b?r?q+kyVJD^0Ee?A#A5hF*zWYi
zuT~v6lqGlQ^wo*p-vyz}O_rcLL;$j3{?*kyDX!Z&I9(V-Y+t6PRK>!`EyndJ01XEO
z3hbYMXSO=R!k>RcG|m6vRU?w<^Wq|3M=?D>)j9`yTsBv46m#B=n_>=0#|<R1t1PQz
zQ<6&iVW;a1zb3SKPq!&lO3_IkaKMr==nWlGvap@ZdW>P3E2ov$p1DQH#OzaMI;d2~
zVV(@qcR8Ov6L7=zC;dW;)wZ$mj52m2xi$W5&l`t8xo$k|6G@cb<{f9$LhU-*1(MD~
zsKa_!F+yZdyEj=AU0JEyO&mq{&=d8V$oQ1j9-M17ikCc-*65)ba<63!9}s6PZl)mA
zzb9m$)JR&1E)|sAO5Cq(8}F~pR#sPQ+^^V>ec^?Lg|{b*!Xg!T-G-DzK+6Z##VgRL
z`uVADMTLY#^359r(2b!#Z}gFTe&#9&q93D;&t|R-NEH2fbQvRoT18b3#I638ds(I&
z#A`U4rluD$+EF1B<Q*(%kg^v2vwp;T3B92hB@EZvQ)NuJy*kWHhiuQ?`D=6($J@>v
zeJIlQj-Dn1t00oWA%Qc^BKCfc;-2UJ%m=Rj8p`p^z^`x8K@Sh8EHU0;A)URT{h-34
z_kO5lVIe%>4Ds{)t0cgp`CV?fVM{?vua{b?`$3m9YJ#xPu{}iq3S?#}YVTO3pAE@m
z(3GjqXMDV2s4Kk0MQ+U#y%~7Nfxe^uXMeLBu~*CtouY1H|F8f@W}z%stL5ek;~azK
z9og%9CbaKEI2S&1bH5)Bn$E78Mzzq9eP0XBK9U<yGWD+V0#z(bVh&JfNA6fSSXq&u
zYNcdh8A{h@0<k%iltjAVyoy<-4{J|A4>yZhp+G*2Qc!SySAR-fE2XSJp?Y=hvsG(r
zrKJS}Rc<T;fqoQbD%!aSleA|r_=4f|emw0a!+!JrB^<IZxi2#|%HG}si^jF*qsMcl
z=q*gqo)<KcU+0pfNGq7DDwuP-Z3?x6RJa&Db`>Qsptc&uh)5g|KYvyeV5g?hROHIP
zRDWOVSktX<$`)T8ZjY4288hIHe5f?RP@RHn6+Icq8YXbQqI~T7_*p9TKGN2f+8TLc
z@+WUUhdeLoYz7Bi9m@9kA_=s5!;vj}WxAiR8w0zA4JdO82rwp5Qh^9aosIBV*ej=C
z7)lis7SoktH{ozJ;fQ{4%$Ls@=BrW1Z8Palt<me~8Z;=K4+!byNo|c24K@77je%U;
z=sViqJ$pR=1WC(mc*`Etx01cQ;DQLu(J?R#K)b7B!I&-na`E9l0IdDspwj`Ps+$`+
zi)ewE6wOoD2J#Rw-x;x3)F6Ies72@Ht<mh@;C`Xzj}Zg5CwL{Xh_jP6+ndsbv|2Q@
z3{S&|8!soVq!O92Z5DlftuM-dU~hgJY_g;>q#8kEFN}yLygwt3xp`0m$8Jj>DuZtL
zq$y@Q&f5F_tQo9s?TPPs(fAF8=ec3_os^Xl3sppiyg;YmkHmRfFd>@(88MBS&X!~5
zZmGyj7q6JL%w{{56d@HQRJ^aZg|`h^GHG%9c&~q;)N1o?jg2^j5W~OjT=0!*m{t)>
z{hS)ud&$y(;U|_6&gDF=rk46(m0Z2Px#=w4K)$=X`=_k%@wjFM-J$2h-9_flpCGT1
z0%!@WM~`&Ie#vz#sCYnV^_7E5gEnT~lY@mbw~>#HSmUMqkhDU??8NWnmC0YxVjmCG
ze|<;3;CVxKZpyYQ9~c(2GQ7B>c0QdbaNNFw;o``qPD?QR!tCu{EtbxB#Fq8*k8~Uf
zH!->hp;_?5jXP9p*Ip2a(i`Ks!`Zto=IZ5;rgD1fzqhWT8}mTL`ec1VZjmiAGJ3<^
zU#Ixzc!T`7%xBc6c~X!+F})P&GJKmG>b1m5N@n5Y*?50ueUks-dfMRqq;De90sK{w
zpV+_*?v45*3@in-`mWI^q-#b2={6c&dBsxr2Cc9bo#urFEwuWJYK<GU-X_ts$(@q2
zvdAHv@uV`lt%nJyth5vjy&ek-i-(U-SyX-PxwaIV+?`UQ(+?`~P&i!vlgb}Dk>TMj
zfQbtaQ%;iHMP&r#yu^=r1V?MMtFMSpWmV7-Z-4mGj{Wu#LsCQLiG=5m=$Vh|9@3Us
zZ5~{7HXc-WzeO}p_KY_Dolb=nD*dO!xNQr=#K>sYryH`bjtR<`xH4KU`v;uTMbEe{
z?h_h4He$+V)ax5MKOtf9HQwH_wj7M96shy|7mW{#idegsAw6Z2b4k2r`g{8nKg@Tf
z{723PQoq(xqk5gslUK?ct;^+cmLKqW6*PGu&PFWj3ZEckZ7#{uhe|V-n@l!LQSQ|V
z_TTTBMMHMnRn2Ay_r1_=?aX<}Q&@2Q%RE8o&LY|8u~Uet!V;70L~dv4cFE)PTD6B&
zV|UIQ&aC%5TyV$}@|9L=>E2xAy+f-GI!25eGNk&)C!gXPcHy)*?<^wgL+p2j0?9}q
zFKQcB_RN(s=Z`OM5R%<6nd~lEBp=$oAszT6L&RFBUpyN{ZyIZ7==_?_Q&`in+;QI8
z|M6TPnvYiNJ6~O2{?ziKU}&}qi>y|#6@)-K76=&Z8HetlP|ON{l$2#}P<3cCDMYsh
z)c?YDZfe44@DMQ*8*%M0>}LP1KUGFOSiL`I_7ZHepwp&^sIG!EA{F?(_x^EG2)=EP
zAWKi}jtc_VZC3}~@~<;!WXtMO{1Z~?fVM)eWRXw|3j=|s65h>82LWhb1M}^i*uQ}H
zMiGS;>%L>sPFn>nfP&kP@?JEk?%~hIqf`!Gl=>RaJvMe4%o3CE;<}MbI&EF!gCCmv
z_p58)P%igz-eu}r9du7Q;m7cuLf|MxuM}n=8*wWb?Z;I7Qu)NxkLn$&rP<&fqZS!J
z#||C9QEA=6DbioAr?ahAKR>03!$+zS;}c+gyf}kNPu=zHk>yBu=sXI8p#4TWeaFqW
zDF?C%nja31hEtn4)L3?oHX9`yi%KgpLH@yL_lZG6-VZ5=I)510Wj{~|LVr!^yGB!P
z-zpxY2BXZ4{*U4U`;Atu^g*akr?IAac=&WwkQk<e_*GJvu^Xi6YUe(qQ{i?Qu$Ed6
zqDU+socRyWWZd%4=IH2ARVTe2``X}svT;&Qv`}#z(J<=VOvRq6dqxtZyCs^X!&~n-
z&F=EC89Npv1+p#8HPFzLt&>k*+c5bAwfhHIu0NVTqO@Ea8hMRrh0>DuB#Eau8m~_4
zxxL8E@qzXSEHNv@NYvBjsL?|$)^b}Gw=O9}cT&tJTX5}Kb8Wg+9lv_%9aX%UAGUgP
zC}%&+r{oLl7a`F@F^1x7{ojz5mgA{-2`)8q+IB*1C^vQN^c<mg&9m9di!HLFX}haQ
z13plyH-;ct>JFv{i|s!y6jewKuVu6>$)yLSfAw1~G3EN2cbQ9xt&ybFxqQOXC7Z@l
zo0^{0x1zm^%PV``-7iJio-Fwj0vX4r?LVDwUx;-vh_(qe;$eegLM#bf@#@JUE75*1
zQMyvrl^53BgFFOf{i=I`wIGD_DxUh_aWtw`vZ7I8?vi&b1x=>atbCY_`vAE-Gc{eo
zD{|J14xdP5EmOGW%ApB*KP3`2;^5uO6{j^8K~sA4$OZ>A{G<SqS{z5VmWdaHZlqY5
zFUulH`0SXr*|NDQa!ThUS6B8rSA(=V#=8*jrM|>c=|GS~Pnk>6I=$cg=aKJbnM1Q}
zCin|K`&WN)nEJb4BRI4X(+iep^T9WLQZnKqpzVX}V{S2fV$B)*Jz!>Jr=Y7}Snj_a
zRPkaI&+qXHv%_(iZ*M2-!?%m?OpjNP2j<@gRky6IZv2|94Ie+QBRO^Ucr1izR^oWf
zO^BThwN^jvMKIAN@9md(`aKFA)MA)P;6fVwUzerOmP6BJpJglsGR@;g^D0G1upik{
z@~+cQrmq}X$ameU;@3L!_y@Nu8g}mgdA2r$w6Q$-kE#=%diCq4Du<;i-y&6Pmqh3@
z*9jngmzGR1YaYCxqITP1Z#i}p9Cgd-FS&BsAat{z#oHAuHO?=^;*&K;;b&alFo_qd
zw4DhU+BC2e=^)Tz6OhLrF@ALQi@xDecQQ`7w|ow{Um}Vt?|-@4SlZ!P<2fr|GD4wq
z<w-;p|1Dztu0U?oqE)1}aqJ;9?g%)Ge~n9-s^XFd(3mYKHHnv2Y0al#fN6v7KobRd
zW1yPzE}AW#r*0=x-rjWQvs%71=d<U@Wp&aEb|6Q?N96yVX-sIgc<=v<sf$&!+aG;e
zwDa@sp2=_iVDbAC46MNpmP`kRjF#@J2IFQmM5exx)*l?n_0MP`!Dq~rH@lr5L~@_N
z8<tex)JF+*YD7k*bawr=T>MY~$)mN%jMYqwWj4@G<pkdGmu^R}pl}Pzq}(ijl(yh&
z-R~f&ao;61uUEEDDx%d5b4WYcv?iTTqIW0d7k491x#zGDx7Y95XnI=5j?)(U=lLqB
zW;usjHRik)xtEF{McLqo|MJZ{HN~Z5ZY!-U0%K62z>Q7}e`_fQVS)aJI+N;i5jxqd
zPVK5UMozRSDok%T!jPUh8ozioD`UMXj|0cKeKlO|CW*k$ohRGhpUHv2C-_h8OTjQE
zxuzvvdi8(uCH#uQk*iJ<^L4RTiJPb3c#`2`?VgEdm&g9vf=jdP%^%a?{oFhmzI!QU
zKkYKbADw}1p(K}M0^u+C8!cOOzn7xkm&M>-iDdn=Wpi@U-t<Axlrt~;)S^;$4<D28
zXsz#$;re5zm;U7p+}$7g+H3qtQXI_Wi#}3uoZFT}|E;*B+aKq=enh$dW;*ApLGMA0
zhh5w=csEM^dT8M-u>)q7+@vf}&Ve^<5wyb<N-3Oh3oV7rZ%LzFrA?!MA7|8~_u$p<
z%d;QkBOEYkdI*5Wn}1lLV{8@R+i+&*U_F8}R!}p(`d{?BkBRTGEhQyYu1pR#7=3fk
zsOa?jvQ{PBt5-?&noMC1vFt=2+r^V;mkmhij(UE`FK=lhK75C}O+-=r^^g;((qflc
zB><J&5)N{ChFYdX;%PZudAKACt~t>D-LsElihnilEHCR82!Iz9y)uIE3j)}AQK!FK
z$TGWdPu4KP3G6s)(ACw+Xel$h#oqU{;6dTxamu@dDiqj!!MiOUL+9-aIlF2N%nSQd
zyfYb1QHXpQzgea=ox1J*C{_I|wSky5)NuPXiCo|Jw{^^7@)tBXMEo0sfHH$ylwSx5
z*eX+BBt$a`(|&Nbc;oiz?KQ%`EI{2CCBS-j<Wl4`#bw1r^kGY4Xo_~b+046l=&8fA
z_K@IA#md1?Z5s_gt&e1-;%&x>IWhUUeMu2XnHo(rnQ3&Ye&1Avhp?<dEbbw@CB@n&
z_n>ZS?kNO2tKebEk-hg;0Z&ps4ZA{x%QINzo7Xx|401O6VeW=6Wa2e$Tc`2vwlQ^$
zYU5>@W8H*TR>9J!taBy#>4k!4N&(8ro8=wW>L%o^N%_n3oav-F>*D@PO6&>+F3)Q7
zd%N-jWX<l*st;tQ5b||AQ9IvuIYxViBWB=x70-TMjux3CxG$$>#vQx=J&2Ws7eV=`
zz^UYZ7-GGByJJVqhBZOc{6ifN{kw(N&=CCLl3AGIuO@Q`aJv{y(&|1V20khX%gb!v
zvK;`!hH!#fN7LGyGqSNWBb3`g9Vw<-@+9Cuf5QjDd?Eg#`KQ_AeT|YA?5D+t2ypYX
z7zoi8sZ`No1BgXJt+hRiFN>0023~8l9DGxuZLVPi7yII}1wT3>`|q|#I_2flb4q?;
zG>aNBq)p9Ehlye=K4<EJGb|KG)6jazXd+6}{z<~<5DnacM3(Z%!5yvK?B}TGOZ#Y$
zCReK2m#~cGH(cJy3BTP%0(vD+u?)IL%dWMxb;n3_%tfX<<`L+1beISgw7(u*>tWPJ
z5XdQ*jR~B#L;^2$dEmkmrGEdS4D(PIw|(RBTm$t-*Gl?_#^_(;KcxRD$M!E%W2I7e
zVTzs{+$rY43h^boyTAJ~fIji>qlVJ2ffps!ED3pVOzoG^7v2~So8@h~;Q0FLNK?o&
zbTl#a5E_M6%J!|wu0x96*F+iYMES8M4xrb&Hr!WIfZejD!ktfv_%lS2+3n#OT8uV*
z-I60S`a<9r4?5@Zi{+9bEk(5N7vFsst*Fjy*~%{z(JJvc+bL$8tylG)Ljll?+V*g>
z^hZ?4m}fKOaQ|~wmVaUJ#!H%~S|MH;pD#HI2bO9w4&wBFNn{vpZs<p=VbxhRw?|r2
z)}1EXh8zxvUcN+=qJdfB3JtDb5$^w%f7=^q`Kf&mQtqNK4~m|FN3wt7^C4s@Px(-i
z|FV#?AI&@Meo0rR9kRMnoqF&{uOu&X*dCW~cIE-85I}ycr3|LgPt}Y1mX)#dB^MqZ
z&_vdJL<9bP0r%X5V7CG*Ke-<F?pX4$i*s`aolKxb8(rG8I?Z>K=<vc}Djb96*Cbb<
zZfet)$n3lVnY3O!e8bed?Ec<@mrD1I0SdggqtY|N;|*ibGFi&Lw{H;DPBwE_@e8a8
zZKae9NAg~~KXDJF$x;Diljz3!D2|q$`l$_p8F`UJAC$>Ah%JNFdE=v3@vXNu)d=lK
zYsY1;3At^MSlkSeZQYr7o8X}rt!Dz@Dlro71O{4Xr>=4TJxT94qa$NFelMqfx95T<
z=4opegC6hN-aO;xKXmw_ynm?p354e^!rIGSCEACZa|Dmbh8MVcRKN=CS0*erB)nBB
zc7fT2FKcgZs=A6kf2!=)8i*edU#^qvm3fsUpjrKp6|i||kIiv@u`t^l67=;&XZu@0
z+DEo@ZUGVh)V;A_QjyVq0c2xJFIDMU$@hF9$OKNq=;2NO`J>D8^kW4K0xRy?5DJl5
zTJm=u0fU8(XGcr%c#%_8=I9X1oXFt0!BmRo%B7LV%&`87PCep4+$*NPx5f*)K*~*x
zs&xC6M_v$$Z{*L@<Rp&6H*g%b-ZFZs{4R8y2zQ6r1=^e+&~=fx9t^aBTKH#Z;MteQ
z%CcqQ!fg-SDeZC2gXTiLes^-2+LDwkR4BlXn{pgd0oUt8ZW8JnG%>L?(_it+*)YQ|
z(a>8_#Ev^~i1HO*V?P3mm-c9_z|pvwrl7}Hc8yRRr)LSLW^|jvt|C$Y!xEI9Sf3p=
z^ACqDkrzDL#{bq3nE@GIDZzs?nJgYK>iCXZuTu1J#U9TjI)0(`_{#o<s0fPT^0$PL
zTZ`nve~rzr_2=BA&y2+X8FP@wf6ts(?UqN*xvWlpO>J{~bdM%hkmTM~{-I98V!n2K
zf^Lr8Tx%&LB&5Ot4cPR8%YVi@X@7BDgn4vN<jjS$;V48wAkXOYUF&)otS0UGac5<+
zyl}0n=`Bw!-xIzc4Vt|)|1Pe1mx11FeMRqiA7*nF0<cGIx4LFHv7sr{t~aQ}!c)I<
z>0)Tu*`Knouz+UrWRZ5)VBuxNJOGHgY^DVe^4h-820~roH-IFiTC4*PxMF{LqZwd2
z{}6zW9``a70qp!yq<jLWnV7jb<JYfWGcz+;-OkM?inNCSyalu+gg2+lXpw!rfrLVd
z{gyl+!pm4%GKUazrUSIMz{Q*$kO<3vKL<3y?yq-Pm^G`tnxGL>itzAom?V_>OnHvA
z6qtMBeAGbz@LQYuR-hfw<O%fe2$&53h=33Sna^>z8=a-T8O$k=&13L9%iw+0cjBzX
znMmLaOiEy`;#ABVZHXbLr{B{RjJ8~-V&}5Cg*zLjw*UIfK>nWSZk|Hz>Yw?jBb{ya
z&rI=xe_^ug1!_XY66o~+m4`vf^Bs2ipx*-2tdw`FL0<t%8{HygH|n>jN|d4jG>L8i
zRko@VWMhly?tU+Hf2j|^5u!lu;9L9H&e^6t9}MpUr0cEfhG3mNLgyuKl&K1{NsEhQ
z9vh#hw9>3Vi9aGTG6?s(<Thl+K^)JjivR@IoA%DZ?1vJa76g1oRS{6I_Vo1hyI5OW
zg8;5)5MVJZXN>`m04&p2@C!>kZXA=6lY>Z$UaWUA?i`i+@t2jP!oq?NbMAuCA$c`b
zi{Vk(r%dD!0{IL|8+J?aTd+51bvg{+57v34nn_@&P_RSZf1zJ*R7Lwk9~owp#Tt6^
zIs1q?uxsJDf@cB`gR?;28ibjcLI9k4b$wmz=DbQ<kQ)phS|3c7eE<FtkT;;@=O@V+
zd2eb;4^SG<0s8`l+}R6&!EnNWl8<wW$GI9H{`-uWv>UL&M1JmXPh?<A27Z`UCglYS
z3pV&+P_hSK{Ly~6CdfrAFN@=~qXBEu)zac^2=&E;CnS6uF#&Id>4ekBq-Ywf?}8<s
zuCySjFdf$eqUK=TPhAB8v5T6O^*ew;9JfD}iQ-2wYiHa#Dn*L7eE~X#DJzQtaM9|1
zTkmt(6jY4Ati|elMEFA#{o2xTDRQxu@)dSETXZS~OPdZvsD{Ep&lB-YJ)=_?;%GM7
z02N`_9`cHk{Lm}*79C$6*}4KzSiE#)bi{B&d25F7Q?32$yzx?i;)1DB1@-N190L$F
z;^B;0(Y!L;01%D^4B##R=?51F@a>AusX$X526h1lCRWS!D-dCC?A~2OXmCB#1ETt(
z-zos>08VGQ)eoww(6hF;&ji}+0LEB1w-E=XasU7-W;37>02Mu@We*_d&H-Q{54Rh_
zReN?o7qRhrJFiu9d!k`_zYy$aQUT}3+S=Lx@FfX=yZ}9DxUL%#M=hvT`HUTQa^@70
z8Z>hzX&zZ_>PSA9;zJc)W2xsFqrk8HqO=D^30uvRuV9NsJwM1)DCR}|eL7}E5j>G`
z8w>wmYbKv%N9?{|5u2gkus<x^rQy$XVKkr+!7wQB#o(Xk7efi0=w8MtF6))E58&_z
zi&0rMoK0AC%xef<Z)U*dYg84sgV|U2Hy!v4=+_(M!k+b{#&We_!~h}$!w=4WtIGo8
zDbV4l#Dc2}9&7vnp#6GqlW^u!#c92IF{jScKGQ|lfRm%g7X7<vSPI~?l*8!2Q=7|O
z{k70e7VPh`5v5l5QB&7c|Bz{Y)&*~Pn5YxEc}e~7HExnrK%5jDpWtGhY2?3!Hjw<K
zISi%Rho^!3L$&s|_;^plHm$7y=sNZ%^T&cm0ip$X(0@YLK)JQ!`rLvs+Y?E90wnM8
zNhVDc3zY?2jt#zcS%N};NfkEe(1ArbI__gB1InaY&dW&PjY0t8s<7-gfJGsMVLe+D
zo%N%5u1LEHrweG~1<45CMZFEcQc~-WAct5qSnHZe@bQ(+GYlk^#71D<(Q2hpE!}D`
zkl@6XuiQRh>tjm^X;a)KZ)<ki;W=I4`>ydW-%jlEU$<ZWjkUf!&N@CE(q3X<CYf<b
zpm;g?NX5>^dSqoYw?f($Zx~P{Z}0A=tE?yhw+YN?An;1{yhin^T5!$l7J`6ozn6SF
z)5||GndPF*F#K0T16Y38{20)oPF(AYEd!ElaKNoRrICVdbC&)xn1g_;V)~&R^XHEN
zoascq+De=c(n3T;L@uMzhp<?t_d&=YFuh2U>>8x&m?u7G<*arGUMtkpnT4=^w+AG<
zLMw09sOi5vhF}NpWo$F^RfF!uLNVj;vf?n&=P~Mk%&o5)015kd<_9<c#Y2;&d5d3e
z{L*qgFE+Nccn)~4SMgF=_nOtR#~ZP~XE4EX2LL8ccyjUp;F!QHo5@T$9Wd-afHeYx
zR4O$$Hum-~=NM08#9yaX{CKI66crUUy8@|*_VZ^6BC5(R2?DHTNlA(F!@tO~*~N#x
zgS1n7!4p?|mzg#NiPWD>1TjTjk$-cJ^y4wJr(g2I_(?zdZc3f_#CfGgpYbM~S?D+X
zp*Lv{d;-^>#QW4sAzRvJtrxG>aZi0Wcof)q1a1P|JSp{p7{E2^3?b=)z}3bHLqfn%
z071KGvj;AVs4fQpoLdtz3c~MIibT3|3n~NN3QoPs(xNP`(!epuzqwh&Y#o*-Dx!XJ
zWG;ckeaeupFXJd)&Y|FUJt?C?@Cd~Mne17_e-1z~P(9=4j^y3dY8%B;&<QL5KfCli
zZMJ5h5)3Jn=`y_`H`S(di3w<G5(a`tYr|=8!Gt1ED_VqOwOMJ`>$=z)Mezcwd$izo
zFaoB8<Vr+?!4Qa#@$i%ij~d(<%;NIZ%KX8-j)4E!F1A3V@FQulKb}3P-0A(tkA@5N
zE?%Y18ZL@rQW_EXnBbA`)5V4*w|&nJfF{WC9{diJu}mM?iKk{z39ZzRAODwt$7K%$
zT8^RR6fu4iIAA1XC>~UJ$LH>fuSkF0c)?>DVj*dSWI=2+lrlh|?G8W37<wO>olOg#
zIouCm7V|Z0B2G_FTW`PwBOV}63SQR%fd<~7AwW0htCskH$u(({5JS)uQHa>e$_j`A
zq;hKTVs`>4JRgbNobLe_somY%ix73+3<OV?0BE$fz|R+~k&NIOE*v1#E&(q^7y@hQ
z84o6<0gNc{Q>|@ns90FcD&)~mMq~O@S^${k&0NTDoMSiJ@yb>o)T>v&pUN|aQUCQg
z#eq=gLg?}X$&kw1pW!uhPb@c=v_qmQxFz&BTKvt5mk>5XPF;6i5j&#h)MiS`0)-YZ
z3J@-sloyAVHXEm_>1sXk1q}@w;Pe4pr`Aff@WX9cYt8=$iTzp@6a=$fI0OV;D%u_$
z`eA~~yKz3PHr_DQEA;=sUXOrT^%D&0%VX1<`PRP#!<@+U^z|xnCFLLD$5Bvmvv(b<
z{tdPs1-XnBw%7?v|1@e-w|<6{!i#{;@{--&LCo+wYIO|yPEo5Q?NXWDuV9*MX|mTF
z8#hxN4+w>Yg;jXMT_GeY+s@Gv9lO(IU!iZ`U}qeVwh%y5G%=9`Z2eUjlM9SJJl!10
zXsuyXFArRV9+Edw+(N;|=OD%e_&%5!1sn1S5z*J6Ajz85oN-eyx}qPjom}>7HhzK1
zCUJ;AU=^^to|<5h^2lZy`Y7iag2xXdnE_A<=%aua)>=>?3K;)`l;06UMP(8#S;eru
zJeoeSC2w>KE!!scXSvnCv4Lcba~%T-w|xvsU)ZwzXv3R2Iu9=yTTtuxtYY<JXt1rL
z%QpejY~->f#u?>0n(Q4gYOruEpBU5K#dDnBJoSMDrE&I~?irq{hiJ{{IF>EOkEZFy
zG-;?$<hq_c*ww<<Jj&+C6Hv0y4o0hO^Z|TKty)CG1sud!*Cyq>Ea>4LZ(mfS_`TW(
zKve3BpqA_<lPOSU2UFpYvB}LVNc+F4O{^NCM=LLJ#!KA-L(K^AGa#;8oVdxzzXO*O
z*eg%ftTG+QW@k9Bku0a@%zel-2^}VlG2VmOgXb)ZK0?`-2q!<oROG81s9dE<--rb8
zH%<O$@SL8DbJM~uH8jK34wTs6S@E}=ou~j?-qtcs1PZnYd>m<?i&zo(Igz4~cI+PM
z>PO2jeUy>ORw>fjU2JM%Yeff(Vsv6Ak8+`|m6==vuVU9S1dY^-sMu0+Jsqm$jP7gd
zwLY?DWawgnRSb5=hX+Vz+(ej|Ovb<G`;KlEpYX+owjb|BmYE2bqAVX^Rb3%q|J{#s
z1#@ODBS$q0_U?u5L0+gf+@;sFVv#-XoSmI@ZK?nQnNgz>qvCo!&GUMT7NF<9h-JX*
zw0=Q_V3rGv&r;<l16Q}=NRP*a;viqCqRAyZ=oAZQ!dt8#zj!VVZniYr2>3FP?9H;q
z{U&VnNQ>s@u<;wzV$BtDpGTwNMBwJfVqEN%(Qj%tVNFQ=uQml>e*@b{3jo7kY&=8`
zS8Ko33RuHLytW3quLyLeiXmO#Cn5pF2RK-e3c=z)3}TM%nJw536168M3!RuJx<4S^
zmC14T66DsgdoG>AY)`NM#6NS2;g(JJ334(#2y`@Fg9g$cvN8J2J;c;1uUhr=Dzc3s
zBD{l!|11qCoag|QFyM5cRT&SZcxdQeqypaGB?S1%-2m^w$QWE+&S91f+wQn&XV&X!
z*eQz2Lw7vHp-M}48}I4+1L2>ZJ{;1K&P%9~jv1_6(Dn87^?N^Y&<PZds;Y=>qLMO0
zD|FnwF=7X%odF8{&rneA69E_?0LhnG^TP-`ud0|cm5U`lft&ys;0nl*54`bRNdoq!
zNL4I_KO>@)P*%BFH(eIqn@ac?ZJ#Qcx0T)~v=}ckwd>@=o|LgiRQry|4xy&xXdJyT
zGt;&AfU6u-`*d{oUnulO(M?&~U=Iky`pp7+<2ReDas4zw>St?{o0}WK!u?KB(=iC!
z)%dBw0;j7&*p26p_B~U~FPGec7qz}rpjtzikL;SJ#X!ao`m<%R=X<WWMYeA9Y4#(x
z-FvoGiU)_H(jx=1;kX)lt7`20>h>DJ><^OF-fm1RCo6x?J!oOa<ZCTC*w|14Gu^!h
zcpaM6*6?2W`T4CY2L*nMK!8yF-wl@zDO!lxQ!O>>+$Y>H&~lUeEEdrVKV-^3IhN&j
ztrv;|2Ba_8j4dH>+7n?{ad^ZhJzoszlWB7iYP)&Zj{%qhFV24}Ew`uI&`g4h`_9&u
z4Gw^yOe)mj0Q*}_6*hy-Ro9!FpTEvpyJBXW4EB%%*<4U2R!X2NeONf=I*VHPPU{)d
zGBO<bJtwHr$gHAhtN=?l_o_^R4Ky5lM2xMOYbZQwIOG=Ci&9yB9{$%@0-2W)01tzi
zP_PIS4q%5sD2Cz%s1(N_je#!)sVta2hT05h^XM0Q!)d;Mk^8*^AcmcIdjMsh4c`Do
zX{2VLR*<hs^x+D<%4J$3Mb$I7HST{~=$%lWO#B|;9xa~(3y)q!;kH}Sx(7z%m4b+y
z%W|$8V4Kjapk+UT{N&KbZ=xhYJ7989bNx1DRFjXiFQodNl$gRaKbjl@3c(s0ekmrf
z<7|JrILfAypt#;O{f)rsC$9li&wwUdBjrDP_a>0vXWaPE5imfkVbhF2@!&BWNV5P5
zvR9nK;~n6HF7F||dHa?~$h{hNPk}h`!8FpAJl!MOjy9GDpM$Sw(QOKa_W!xS$Qb>#
z;EXa+Di^b2xwgIvPmkBW;e*O*7s3e#6~1iPo4@N_?Bk|o3i4RATpsIus~J?Q2>|hw
zXQ&>($o9m~fm%X+Ysufh^Ub5#GFR`((b2Lg(Fr)A%OymBjT`!c`n0>hpRwKpz-sza
zg(*ThTgm-^1Cj>iKD|8+$qzh4*>g*#C@thmdvG;k@4G3f_(Ix*_m`mVA!u~}2xU{y
z<)ve&a`l|n+*1xL$1zN^?kuX3$H}NI%<Cw#sR$w>0ZX=+!lkyf(4eNMv}ug(TMH%*
zYk94W&#pYNqOCnVIMSUZPFty`$;Uith%<?)uzCeJVt@)$;_6Kxv9Iz=J_j)BIe<vR
z3LrxMSzBA)*f541p|_M@Si9nxx>WD3pM&0*#K_*3>u@LH?)|F{HYF62LS1#DA@76x
z=lAU{Kwcb;dBL*sgXWw=!Z{3aWA;=>p6bL0bk0{lSC83eTdKCn2P9yK-CrJ}V0kuu
zj29%`M|^Ac)*~Cj_-D|tB`mgQE~cNQh~@e6si^C{8B|>Q--U9WzaBQx$bQ!TDMKD$
z$K>+H$8xD-`OztJ7r;v7yRJcm_Q=k)%mQjve<U{qcbAX&hPzM}!d8w;?D9!3kYrd=
z13O!29z4F)NzFdNF5c-2F|d;U0`EE8U!eydOrz>dE@k}?bSL{+!YaAxY*ZV-E&>Fh
z`bk3dY64+dBo2E3)_!Y65l22y$npT%X?}wlZ45CPf0r?OY4UC5;ExsLiV6qh6)nRv
z(TN4<!2`A%!FgkF0F8iOzv}x*l|zSE_CqnfvL)#D7kX#mjRHizqZ43vl^PCiOAIL~
zVuxBv+e04T+4RqSUslIxK=;WXo6~uC68xQXm}Z3TMRMBoE5*UZ3pcA})N4u7@82u9
z$y{a<@%)+=I3Zvr4`uqlrrmrHxVA9i`#6xgyaK`^67%Aq6KFa#nF#useubPm#ZZT%
zqhcG%$7Z>Kdqa&6HBpMZAHJDp&p*W;wza0MPC<c+#6%2Sr}@_ib(Jyl8+|eopHelF
zYMVY7N49jQrrze?uATWmUQ%3{(&M!8i$!QQ&^zf@(qU6F{|`Cx<uRB38A1_=vcCqq
zBHqMUz=xov6lpic12oJkT-de#7zxkeyteX?oT36f9<~gWP%K{8>_ApejyjsL><keT
z-pjrJ*y}s4vJ_7R>L1+N*kYbH$P3<CRh-}E9U|H{owJbMaa)}n@9v<VC09GY^zWk`
zOeA>wpQ(QrT}PGhe%qc8j6pDTaHV{=7RZnL795b$ICYDHEAQ*;){9u)va6R3i>^}m
znK*0dB~`284+3u+xI(7*oFL5j*+^YCHXhI%ejQKg`6(OM`NZ$kO#Cuxx%$dAhDg}7
z@0DRjQIU(bme=u`I0eNeeJFckn0&V3k9cp`@q(Q(MQTtawVJDztDXn3BZ<-TFTVzv
zDFpM(%bDs_m2=fYWySl>&?md42ikkee1eR=y2VO}?&y9(C4+{&^_zBA`qwks$-cdY
zDC?{PLDQVr$?8L1OY)%=x$ORMY}i^;<E*o4j6T22CH?+-6T!d78vNrV5QJ9$>eZ{b
z`;lx=55=`i<+VeusJ&!<4Df(}g9i&@oi$R0JQzQKjUf@`$s{^`Xjd!{S45>bNO`B8
zi|Y)D`tkDc|MoP%o(bE9r+GU4S#1-P>5m-W>RT<$+Fn;E8m<?H`4w4d2P7W-?{??#
z80ioFzkI>1RCy^4%BBpsR=1Z&F`2-xQL}9xH@TOVz1gwhAK!cB#~Lf)7a4#qP0oz(
z_v+79G(LJvB}UJ~h3@bcQxJInzx9X5>AiDLg7P^nW}-lq3Z~1Ukqwscr>4g-%O{yZ
z1`i#VKV8-*6^FLYl=<!BlZdA@@5_o=8Hz0aTZ!r&RSw?^kgkZTbAYkTI2j|4nt^*7
zHsR>l&y&vqqYyi1XGyrZxg~$YJnt1(PKNor%(gbqbqJ+*xb4auROoDu9IK>l->7zT
zS7Uu0IeHw9{K*ug^v^NjQh)F%_e%Y5zZsTQi?#`W=16Eou<DCpVnSsB4V2Mel8=bp
z_%UuQMSj9-etS61k>W4(9eKC9&m@<_R8Ok)hyitfA)@;XWvR%Z53L@eW;k-KER{F<
zd;ee;(I<{;aDXVq`|=W~m#-F(0jf~{`cgWzy@*>kTk{Z@_!4FI1rLP8g5`O5Wlvk#
z03v01cMXj{f|6#m{#n=WC2RT=hx*S`S@8}}|0_H^lKae<(4s*KSbb&z;I%yG!wT<r
z_l*>7GDpKRi0Vo|pbj6@%LBFd{jeU-c2#7`^2*@yQhL^GAL7*INZ4U2;bp!{hfz-U
zrj9Zr9JVxJVCTYi6Zzl8(0xji5&kRC11J=FUsk`Ak0?raaO?m`y|jY5k)M*nn)6K6
zXwrfFBGK+XeS?EEbabf6(kLQeJZauaG01L7hWMa`Q?~<D&(vB{K%UMCiGE|Hh5<}9
zJQwZq_Uh}0&r}S;YLxW9<|F67E;#y3C5M9s>7^5^^K@*(YK5lr0fuHDb*47II@Yf0
zRnK2!gj^4(2)I?|>j4HtdzUC4ztQNn=+rd-<NHlje?eg+Cze$DLHS2J#jZjA2w?um
zS5T}Nsduph6@S?DPFPC?4AQ6yxjBFy`u_Ap#tXQxxi<LaLFkn4h-QK}#bOC==z_8u
zJct|X;XY^GE<TONvU=A#^AG6%5tbYpf)+s@vo2-thXT2i&1`*eh~a2Qb>~n4<t+H;
ziqml765v@0Aq*Nl0sV2vJ5>Okev9dE6K~&=cRIe`)$ApV;3C9~mg;{o;<?Fq-B6QE
zIVye9J?1NLxgztAgD0}0WB$VtyA~0CZ%zg@%jZz$7Z9+zwA38Uppx5BV&Nj>lpzt{
zk{B8J799Vn8av~7eSmbm{1Yhy=e%3eWdcf(hMAyl5BY&a)5i%>o6G%B@LIdD*-9m?
zg8xhPYfQIp&X#e|u|eA?gF7@+kJUx*Tvn249Kmy91}Kmr&j*?&8MV%@4mVMi=-OQ^
z9$Jr={;_oYo*Q7NR{Y>!idjx}1n03xLRmocui0q-#6rs-nZx2pyZ#4W(~RrKNw?n^
z>K3G6@Z8r)L<3<KJ7YeBo#(R80D$uP#!3X8-3E)w_m01Zlyav6Sb4o|BOmE(%!VCz
zZj_8IC{({>+;==!NCs+Csh}99&p{j%{5M1eRldw-hA+Qxbi?LlGAswCkFE}qRO@2v
zJ?M&AvbuQ=M1*UDw<P;olI6_kp}lkz%KXlS9`6=Au~DOU+?CKd$;L&pw{RgIQTFd&
zpT`P#u^YL*&oGfFdZhog_Nwtc1}i7n;!o_3;e;>`GA0?C(i|8(B0?j@nQB9?{S)9V
zrf=Hv`dv0`j6CYX_%IWc<tE;!!kjfIJC}f3Bpm2}b%O4oTpl!T!jYhdUJqoO{bs@J
zRsEIruM^S$GWBh3edogic53=T)Dumf#~tHfG7oj8PEkcy+hQUzna>eczTpmsAFHIu
zjZTWzJqugpC~wmSM~+HFk91G_bRS+=Ehp;k7%qUXX=KAn_29^klTKZc0B++~$&l+)
zSHn;79rKU|=gt+H8r>1;gFbSQQDZ*}I+u#x&xtC4c{Oz~*GL(f@7^9A0c{ugYC$lb
zc5M4M_VWKrwN95JM&Sqbv9!HN?B8eO@b;unrKj{iWmYrfB7t7^8<g_w6LgO)-Yq%4
z5G>kED-F8+e~)RPX@>#Z(H-PN2WRiT!d`CKh!;BIIvh}agKXpefG<sYyYnzzcW_A4
zcYn8PoIo>gU);MOX{7cI10Ii@^jV;rH)7MbA41YeJT^amJsOe`wN7JGYDelt4wv)a
z6_iZNa{EctB>aY93V()%x}opJft1|;Ph-(eB*(|ZG8LiNDoKGnFcSq*D^OO-kzdPq
zQ@p#sIVe>xs0D?t!RmW+K{JgCQ|i3&+_5o5G683pxS2elVjEVdpLJ0{xTMBq)2A&c
z6u-du6f{&+Igo&H>bCx~w!UuK8$|~`C*hU8SlUrg;?1ec>dFcSD1>pE42uBamhI8(
zFweWQNxcK(53bN5n9->OB)pY@FTumb%>+ehGcZnm1{kC?{gdP4a_i+c^U%Yc=|X)C
z(8?M6@fva7<6?fM{If!d9iiW(%3(@MG0+*PT`J9q`>1hy_do}JN6YLQ94o7M$|rHI
zU4`E8cZM`s3@-)VEmveH7U@4#^FgTR4)4pl+zWv4nof#;OJ{ZqH>Y>cj4L+jEBGnq
ztu(m6!TzkR3@<S8Rc}~{PFQ4L@1ewe#t<$O-76%{z)PLaFl>@inH}_~*EBo=SH@v7
z{A?gaP$N&dK)y(;;hVsI!^MJvmR5>bFriK7nemRAS=DTGY;5+{Xg2H)jpx>_d0s#t
z0M%|AR40HkiJJ8+C>d6OWZC0-D|;K%tbqL4w_AHp8@r7meo$i#NjRKr_#c>n@Kw4F
zbI%>!J+)-aCMdT}0(zQ+z|R-F$<l(y-BbxYHY-GvKq*Zdw5=w98ONibC<6U;{JHZf
z{V+Vfdr)Oh;Ie=cett~+2^vUSTiL11wn9{guFY;s&E7tJ@bFpaIsmTMpos}bB_R~`
z!!UB)?8d63)NmcQb{;I(&)5pNqJTmr`#e{}6k1v)NyxKSgk&RM@%pIs&dHE1M@vfo
zlj|ZhwsL&#mvD>-TTpucGX^N;1?MKuRK|tn86O4|)$>P%b>ZrC_$~HlFp&pP{N=;p
zt<=>MV1+U;7bpi*X|l*6uwcYo=5b(mWP_VXDItWFlqUdfbG+10av+{voJqYL45t4L
zzR6*|B+_uRUuW2;4PXIb<jza*A!Pi@<9RA7mX?JeCd{eZuayLP48h=A5WMKScTvvu
zpc^F#+UHT@rsGj`3b1Dhes_;$)lRcFVgjFoX<8@(aE!kJInI9CFgC|@JeMtTKte`l
z5)^tIW-Y2d0~G>dHiK@W0)vqD9eAsE@4kR2hPvhf?R(Rks%m9SuEc^RTVG_%`cGJJ
zh&H^R)P-CcV?1~uZRr4o(#sE@s4_cS<ne-{bYrm$`x=Fm`0>q$Ly<_vt|k_?ALc%?
zX-HhyR-RmhBwU;t_eu{$NkyppNffkZ=Hk;$(sgV|>?k<!I2IWRuztW~u}q_Uy02hz
zc6eNK0$X)2wjkkBl?cUx*R)}ss2Bx5hjTHblB%^U=gc8BM4x&&t*`b63!l8by%Tut
zeuA7X_xpGD+1!CtVeN?1!#3>HGHO$~!s->YkI8%$!`ZvA^9*nXYK-^=S`F&U%MbMj
z4X3{)D?op`!d&hA<A!C;8ve%=ftsa7@R_wIgZyWp7PAFPUlX8l<^ZI^VombL#>RZ`
znjnDMkOR#QbjvbTM$pNN2TxdKvmR79%7-1m*zXZgS~=}wY5E1KxYqy%!s&T?`Fo)O
ztpyBqs{?H*>zWr#1k&0^vkh)eR<Vf9&ypMi)&*|EW9o%vq@fQ^W6?$q5n2{>^PjYd
z)g_6QWn9y0Zmv;CB$RF|_IIeSq2ocPAWfj+A)I~tlL%YNHlR*LIkqt{e8Gz9P8)H{
zf9Cg|LsnPfdu^vEEmO*I?BCu^j;6ieC!>mK3+u@dV9P5)ltq*LJ(KNh;jXHgo^z1!
z#Qo^vV$v5|=;a5|b07n@S4I87Y2HC4zLSFA(L9_<k$}bVf`pvc_74%XY}ts<q@EOh
z5k0{^B%tu;4D6AJ7r>Kex9i{C*SVe40qXK~u=V6qP(+SaJk__v)xvfTz#x*hzB&~f
z=(yU<2t#9DbUztbU`FwJ%}B_tFzfk=WgO$XkjW~Rv+sZ?h51=zVNeWBRdj;?=5X!&
z^?nJRE$^p|B;y?!GpOQlU|QZQ_0-*N>nqJZZ;)QZ7?XU$nDVy5$InMS*N*K0z0y4|
zZ^)2fck2AfI?(MQsy>M$tgE-rG3*Jh?fr#8&!Pa6aQ^_H7iR-$%v=yWnga72-)RXN
zv*D9~d9@gk*~eU$KV|oDlKK$!S^7gRxR>xraWOFoD5A8qR6;=6+&urq*7Z&iivt;m
z0!9Jerx3_JGH`GxD(5LF=T+!roq|yx5`spe%V%Crhf6JJ7#LA6LdXTx!4!!7uisSp
z8o`<LAtwibpW55wz6RJan`-!DRdA&jj=Rp%;1E9cvf6-l;?eho|HIu|xJA`<@8cNg
zBZ#OdD5=sQpdcVIpdc}{bPcF<OT!QcQ9xQr>COR0I))VK?(Xg!I=&m8_kBOt?_c<F
zffq7o=A6CvihJE_?X?YQ`|x_(1-Xn)xh0_vq=TqX=Dh!5E5)e5h}809r98*0C&_71
z65*muho9YDU6)pZVs2iH{xw^s2xi{^4y$ZQvLxTr)Qjd%cdP#3@h!BREvvY|xK9Gq
zi*$oZrxYUkGqtz;pz$fZTR|44G@qNTZ1Qt}o|YDVME{vZ!k<)6`>`1pNreLnO7Hu-
z5H!Iuxl7P1F`W{wB@Yh#!M5{l81BrsH3Ld*u#pAYDemF^!a<6@Sq@y=(cXv!EgbNi
zV3IQ~b}Dn&RRT{P3X<lwqLQAVV_{KmRNj>r0^vO=g7qDi=QBU+<+J50?<j7C$6B6a
zt(N{e=IK4GQk&`a7G#3hM&`9no#0!7B|f@AySfUDe-g8I29^iIzNYX|U5r}JeAs54
z8qS+{2#kmKM5&^kM%b|3l13^{w*p--T$X?98_k8~4x$NfQTm-&{G3@umI}jgSYs|B
zaW5(`PCke$Ye5)PlF)|c#DaA<S{=*<i`Q#E;gUu~56OUzs<JX^3asHPey25l%_~>X
zv+D^zCy-BHzI+cNB|hrOwebp!z{zGUC>Gpqq6c)NwqwZEcy~T>j0(LSWoV?AI834>
zac@HXn*=N=#E~7rNBo!AQCzD?I;v7*PRrlPWGt_Z$TZA6n;NcuzBUB*t6dDNJ+A*{
z1@u6jRzj@!BsHYOOpnrF<<G64WXYr3d&)*BVH1CLK_)PW#%!Vz0UGcqfK~xY+3K9=
zj)5XG4R|*&Wxg<dP(z~?SNt3lk1@cHro5;moZCTej0O&>6*_GP5NV(*Ys<t$)sxM3
z?HW_vx%-<E-QVv~gAPki$6Sxs>L5xE-Nw6XV;GhFaVHDM&@1uoBiIJfeUBCtMx`mN
zZ?11_hGkfDw(a<Ql+{TlwwF@~Z=da$zXBzZs<2biKDI-0xkzUxEXo9xEAUcHnN#dq
ztrd^tx#h7i*}~Fdn6^@qvM~0j!)#7Yuc<dvNiv4lwj1!+g`f`21=<@j2ucQku3U3|
zWeihHvjo&tMO1VOW&rdr4U}Jmw~VIP&V{N-1XUid^MSdjc(*V+LGN29KfAV3z7t^Y
zvy)(j??9UYngtwzDwNQuC=}QwZUWMp>SZ<=cB_L4Sy|67Tr@BatR44p6hCi`5MtO0
z_J2E`W<WbL#Z<h<{A>Kt^nadl{-X?(kzF>`iHrDRSk)#ul|~`P0~R?YW>qF@S@aTx
z4D^SM`Ych?TFI|Zy6kmxkBa;aU86ift5Oq?#|=8W#a)pN&>%R6`G+Z26xh$50@^S(
z%~>@q*)VZ&DZ`Mv4B?I~&z?OK0$qsmXrK#x5y@d7Q$uHV$4mGoA!RQ!y4*?>-zx`!
zk+2d>1+|JHJxd`pG5n75pm4?}fnd$A@vELGtwEuBA80_!X*#&+EOoQ`4g38uvX`00
zU2Wr%&;LHhjR>`TR=DVRUk0JwR_jQUQKrQEBW3rSqK6{iIVSVcZ{Xqoa~taCAevul
z!m#%Ddk}@h*lCWL$2~WZbJ(IU)#Sa)r^1|iM?uTn)AP2IdiXAd)tL`#eV>Xp3;e<t
z0;e)+=><+JY&NR|gOLX%0iB)%WWCAoU=DMWu0#*dI~19qn{IOlT64fBtTDes00jat
zBY@9IK|AiB4StVaLWz@BAT15R5gUs}krW`)JOW}~83loyiJCb!mQd?7lN^Rrlni?g
zD1w~=-KP5#0?va`wb#<W;YF4^4X;<zHZ27;p%(=Rad%$BqC|scKgaOUT`XR_P<l_i
zht?tFt5_(CSUX-ZX`}3U7v+kS&e-DXc+JX2kGR<nb#$@K92;+)dor<c3v-X|M4LmK
zoK#A#shypj9DiDFQaS(DT)@Z2zbIaL1uFgu4ZDQ&^VL9|NFTrxITsEGG&l=78=7qm
zKRG$sZK97e1Rb+TK~`kBKbtv!C<|cf6;L?R2S|S<&<@yrMw%}8>|llv95f2{|7aA4
z#l^*ea;c<`wVp=6=$3&xF&77m@>*tQ2X6DxbkIgu4YcrlZleLBA1o+JdGOrphFgz^
zGU31$g;?_SD9q_wwUm(*%`_nsu?KvCG>lJ+SFL9yKX3egtUU7^!o6`Z83*rX!YDY;
zP_{=W74!wW;phj)W7a;W>kyBcdnQA=M8PW0QnL&nYj{h=)pWbldD6W`2iqmv>_BG0
z9*`oIa9#z?A$=;$z*^CgQa^d}CJ?swixT9ago@`Z@{~bk>-`Rw97i+&X{A1DtF}eQ
z!Dvtr1EA|NPhc<&Ai!V;lV9s@%+>XMd_akwM3j6^`1I+M4|Lm})}DIam#tD26ggC+
zDO^eaW?XExSou3)u^*4tZuw+*jg%NBdxj<J=8vN9qN8Y6c8JNu3%HFneSR^pH>Kad
zXJ&Y5p9yWW9cT;?QYCq|4L8lUDhzKSnq_=d925z#i}t~kPN!rpFHB54@a>s{K<haZ
z&@2%Ii5}1}cl@VKLtG0xyCP6Wpt-p*Ir#wwq5)%Y?iVPXN`VK{F~Y_wZgoj>k0pap
zhUrzvo9oqr>YC$kTj<YeSfqY<_6PYG<$jjIA0_D3*frChZ6m-G{otgLE5b6a|0v3;
za3CpLFd8qec;|b$6zQGI+~(OMVwE}I2LdymvYKUsd;La>sQ}xE{9{oeA$L%Sn-0Vw
z1Dq<hy}z#_BNG4!fJ6<83Z)=W3i&#sO`l4|C()t6-5!({0WTdgJ8sFw$|_?~d6)_c
zVRA0wDZs{F;8~g`{rpg4BOW(=;WcnvD|f25%?r$vWR^mIqwe7c9?(EWCb5bQ!}i_$
zKJ6HbQmH&la!2zZ6+f+P<YW5aI5@(5L!s#9#Z2P}d}NP!KrZj`x%eO8Fwy1PJ|J3p
z=3DnHxG5QM#srUEyaBMFY&jEk0osqtq#o2Jx!Jcz^WcW3fgo>lljHckr-w&O7rqYg
z*RfJJ<JsAv2%Ea#0Sucg?~}C_rm;hBt|bC{?YPQ7E&BYz1FGJ!a|<{3Kf<GLt&D1Q
zF_K8z;*ce|R)63KEp3+CFJV!<d!|}`Dwo-CGC+PApnza|XGa0}drnXqata__Bq%RK
z*)zWgygClbK}a;$`?EEEHrRli=Hzx6WV^y2apw)KQk69a*3}_CWF;hanDrF#N1*g(
zUh=tlXwY`AYh#Kr-{y2aCFKLk#HSl$7t?~YkK3KG%%E}U?OE%#=1Xof${fneITEzi
zX<EMY?$u9#zNZli9J~S5(?PXUa-eA&s47N*veuS&zJKuAlHB&e$!QR7_}8yrefjBt
zO1ZkO8v#CLs>ue2DHoW&8FQx}k@T=!1QM#U;V)kNs<4xs;g$SeBl$qHu$*0VpxjE#
z8#&Fj)TRD){L#8ABC<pqsgEh(NLq~N9$-9m7AmvIe-M?qEN!whA1K;%|H0%?%q5P-
z*}NYZ8Bn5mP6ItNklb$5&ySW#qX)V-E9FW@;IO6KXZl_Re<C_%G%i3=kRm+NQ{3M@
z<sXP~>?uWaSdoD^Yj~?Ngms=($2w0M?`MwQH|8#hdbX@IurKSJ8G`lhK2prkce8o_
z>w(cvQyKe`a-KGdZ3$x+X%fUvGX>s11WahJOPPXj;(HgF2;zm79IV+2S0~;>GKbO{
zkUA;^MxO$UFd!$A&-PKruB5N;6UsTPrL(swR9P$dYi{~{szRjtTdOPqr0Ai1Tf_Zo
zZ-KT6f_GXlv#0WX#0Jsj$G>rWI*BVcbzRJJd7D0GU@nlbtLa)psMTz8UQqYLN0f=s
zruzHL@hrOx<yGPa(=C;7zN5?*3YSpVXChme00tHTwXip+XT~Z@YyAYXR3Enfz%E@U
z4g-}Iz5|uZZkM|j<~$5E=uVuWS&FeA-%^I-__gcG_lb684l|UErs1LPCAc+y7}{A4
z8a=HYdivn<xfE@{+b5hhVB%<Uv1<Iwzv1A}DxAox|L6BN9`jvz<KMnrx9Gd_pFc6s
z0ucrq4F2~Q@t~ZQ*%dsTnRt<F;V;#$NI)S~XnY(ue4mz-*5uoIG|7!~HjD+x5}{1x
zpG_UJE6tam<aXuW=GAJ)AHV5kR#MrzKqSJrrrQ0u*h3ke2sXB$c38qQl%sA4WZu2J
z{L#HDY&ap*8u|gQPG+85?P`qSJ#&?LZ+|FKSG(|ZrTSen8h&A2%GIiw1NE!ecPZ&u
zD9OD`-|)G#b~v7I!T9EAu#0dPj!%g)tGrAeN0QwAVMj{giO<h@U&yLE%O4%37o8O-
zTSgHshRKR%s0XW8GLrv$^BYlFSKjb6N_X_{VkWG;Fl#9B9$s<o0ao&L3HcI{GFJ7Y
ztgkjk$w7qlizsM{C?w*S-5LI+VZiYey14aZgF%(tbfrUv^|x0!p%1j&38k(_rI2Wd
zyJUiK=gspUdCyvfLb;dstN#6u!<FimI}Pi7McV!FzN~HzX<A63NtO$KWTBIx_2W*~
zcml$gJ6t2?-|G)9uRn|4jynEB0ojgK*h?~3%a*0Z(irv8S0QuRFE5iDy1RMRjbzAr
zyTIU(*k0Vpu`rLYmqo%SZ}Ic0mL6@26Z*}+7iL^uSZ5(f1JJgU4kzj+TP__u=Gm-O
znT_s@0J`trp)rNCJKH@9(#;<V2@RcFVQ+p8TG|Xiw2F^dOSyBc1}Us~hB}vp=52WK
zSw|@LIL;9AjF$9cB*wT%Ejb%E8b++@{fT^mNt@EDGyXl2qLYO$v#$P=ho3*ToR|Ix
z52p{7=+dSL?J$ap-!8*}Tk%FWTW@Kmx3*)?&UXi{Mzzs-)xIC64gED{9`4%~^+m$p
zt=MlBvcp(eSvey_BZk&awzn)~(u8FU>JFdYp0MI2x60h@@EToJK;tX3m2U|aUVY8Y
z9JQfFHJvL4Vns%$$7h>gn;SkBAD3NMd~y=))>7x>jbBqbejY~|KWxesA@y*9@XH;q
zuWlsiv;R1>FuZ=P`u_d<O4g)S_aJu{gpkKn5F|OIHk7BDz0JMDV)-<GZb-Moy2_|n
z{hBUUlq5`4cWrqcpT1&KpRGyxg}&U~iTni(K*`ZCK6z5z5Lz8hen+gvx@L8Q4OO>1
z<1;B~=Y=TJjJo>OUA4yPOG|qbt7=_KhoP>4LDR&ipLG+{EAnyyU_u0$nNc4?Ees6}
zA&(;p_Df4{vv{3}qq6o&wQnY8fSvYW7^~xv*tXf@^_NlaoIoYb#aDXPh3xodz7u0Q
zN<5Nx&M9C{TD?Zm5^ATPE|JXuZ18{+?xneLlkjjO!70zG#g2`URKAwqdt3Q>l@gSw
z$m1?g{A<yfh_7nlK%03o-23wHVlshmUuu|1q57}{UkZGOEAlOU1a?}k{p8;pzq%A!
zV8RjHXOxE#r^)b}$sV_81swsta7F8~8+At2KD<k;Xm0zjwVR=y#C5Zh13nUQe1uqs
z^-qki+%Y3EtBV5;N?PEF{}}BYZ=XdgL|{B#eCn8a89d=U3I*24&7%A_It(qD{zB7=
zk&juU-wo$5s;!dsrhslsep9SLI1TRg>y4{RI|;Q|^#fy_G2V4)S~l??#I)e#-LZ2e
z!p~IJ-E?LW!Tl-Xr82lTZx{~GocCd27|CP)a^X$!i}l;LzCywZ<Z%<y+$=t5G0r_I
zc?PzBP5ofuR(V_VmfiboktA<5n!Gex1}+bp2nV9Rp!M_&T0Z4APIru~NMvVRSk37z
zqJMCjyFz0=6<80`Bc0|=Pd6IkLUZXv)Pv<0OX6`*o`kDu1{M`+oYd6La*UT#`QjOv
zb_Rp@A3u`XcYGH0uh_9;87UW8vES&*4W^&7DOE%nhYWL@Cs$JNGBCJ>ez|;CybV}@
zN&Z{ERc-Xr9wCMp>jzRUUvQIfm+r?2kCngjD!g3&3i1Ds|G|e&To7*DDapx(poBra
z%C(}%Vj>4@r~sK%6bP>_@(=2T#&aw8|KtCvO_-#sqr$>c`8LY)JF#HnhI)bFPe5KK
zgHYKR@ZW$D7fAnCwQz9CQQg<e0Y3z=z6waef}oTLFzt;U9kN_e(tzXYIe@li3iaDa
zFX+*-wW1Mc1ncDz0G2VUdv}rmgnWQIO97vq`7hhf9wCMI&p&b?0>X}3cff-Hlh5c?
zJ2hkc-{-RqrJIC>-=W|Y57J}}2mP*HK<eV(FNbyaTMjWY1~qe_PZpK(rB*@UrSbPh
z$j#UmKrPb1LB>x>POei8bksL(Zfs=O<^qJ)=6R2>PLgF4m<FcXX*+Rf#=sVjHrwIw
z3EE0oRvTROK0hI=VH$)*u*MRSl${LVruEQP;p<0SZ}%9$6%TS?d%qO3a$2=gz)kd!
zh3$0(7fxso?!8T5P$7RRL9HN=7+bWatTOInf`Opa00y>ED!}1zt&9$mt9f}BW;NI#
zpRr?pg-P!Hpp~@?;wsa3Q!O@2pIUR3qOtWAcDXCF^Roy4vDq|WOCRHfC(?^2p9`Bl
zg)H&wooTtb!z;eKN#t~+VV`+pkRbOK8XcVlnV>6_LD0Wz3;)k7O+E}o;-Wzo1#B#M
zNJfSNdF6}9nWf8dYHDg398!^6+)!WNY{~^R*8$bL(;nFIFt%pNrg?UOJ{XPv5(iV8
z&^KalJQT*_Bgvmpuxi@({kvSSde`zqMiS|O&z`Nf-n>mI-qEX&hMlJnj9H_~VlaIW
z=JYTw`gvMU#$vj7p-Hd!DQ_JE`>Wm-jz2{03LUtDMCjWjW$uBhky89rte;uotZBg$
zNuTRCZu3)B0~b{bTofO9fcwPbR()#^q}k@qPO$gHM>mP=?i=N-;gTDb$2nlzWCD<+
zWpDaehAn(`DJegJGw5LEQ^Q(!iZ5VL9#B%Qw*_@Di+zayKT}EUD{`ivk8a8MZDq=J
zVT-W+ergip(k$D*=mjD&b67MM3#aTeWxe(mN`*RxwoUUEN(Eib3RQLYx_71wr{NX6
zAm}ankg&Zf0=;TpT}X>U?spIJS#aBkYemdLvBo_$_0c3;7nT$ByygD?8qO<UuCU*m
z)9lQ%w!%ZI-`IA-;<7P=1L)4q@(xsniwAz_3N##_18PlAPY)gh4m)7(-S`2Rav|Jy
zoN~1M+4W{Z`*2o7CdiSQ0Oc(Ib%k=!b~fG9&qajq4%JV%JQQ~W`zjtT%<1nZ_Yfu0
zke^rh^z2*ZM4^g_-D+>BEDSYuithDFqEvMIqEhBmJNl5B^JMK-b8FN4UwxS?Y{rQK
z1j`v(p)Vq6>hyGtmlCLtEcXRmDh$2#*)$U1Nn*wh=gck$-b>+n;U+<{8B4id|M%}A
zXtaqPX5xn?g$h1)Xdw<~IV74|6*q3!diQa>`fqxZv(kdCFhi@b({iBLFf-iAwkL}M
zgPl#Bi`*L#T7%h130CdpUCkz$_bG1LnK#)@KBtG1_LY9Ew^>O{kN<;mHa8iQHwqtl
zHHB8KC`87rJMFtrz@>V3ipQ)fB`=^c(El3>8DWKGnyeKdMRh?|4i{UXNwgfkd?wkf
z-kN<t;JWoNMOWWgapBI<QvHh0ZTp?W2&u9RCF|_69lAo)d+8P~i>_-Mkm}~Wdi*AL
z+gzmXwiw=vC{CSCB_zh8<5}&Pvr?9F#o*5UwkpdXf55N1bvA9Pd8_=&w=e7u4vt?s
zu#<;V>!32e7tqQeQyB*C-Z?`h-SzYDyRqKP((&aHh57q?xj96{nHyeOsKw5^E{Dg@
zPV?*#JSytBx^wf(<MtE5J#*dXXFV!kBRLtjN?k~Mdlm-v!r|Rtshrc3cXzj${|$T|
zX{8u-eFn#6vuHr`T5Cn%=rIBc-hx3LK~TvHSilYE&>N4s!%lDRwoTJ!l0HcGK0OXW
zGDi$2Bp^3@T5NXq1y_bL;!NIzx0AHCXP+N$(bvUDbyw^A+nTI|r3jp7(h+cGAxxx4
z%9SiSqO&dcZ?kzZU0UcjdU8g}o3ZVHi=F!X0JJ{#IIy&gFQ`1y$)#E9%6XY7Ur$-~
zHIn;1hO}>Y|LoduxkVLs3#zkH^}vf}>5!5n+|kbfV;>6dR;gkO3m&+npK#>-B=X%a
z^8b<C7P06S947y`b9H7oX?LPgM8gX>>1bjGSDbk7NmYNkOK`nbX3jLP!}PZz%Gc?T
z{`-5z8Wrjh4H;y_nK^|?SQX83uC=kWU#YYfmw8hh99HsBD_l05j@Jm*HD09R`?p_=
z$?vUqsgeu?gwd=C++XnTW-1>O3y=sZ>-E`l$U12-AINz5uR}WXcO!Z8v}x&a9v`{S
zom=45$h7+JvNC%P&#7O>Mn8k#TF6x)8hArc5I^O=A*iCv^>Q4%TX+8XL8CW61fyi`
z38RPO-MYOA-ty0#R%msbpOTrQgAR^IfVb`Np9$ar(If*6Lv?0jryNB~qhirm{ttU?
ziAb6ZyS~hEhU`Bm>t)w<NjTOa>l8N-5z?S@`Fw6rvp0CapU5IJ>Zsa;6t4h>SM3(Z
zcf^B`g(fo|nM{>?(xl9<801<A8ypLK2BxfxYR-h&j}J3OM6%0OpI%CDx)cCX4~#lf
z!o`Emv{xteD5(Xg6(%-@Q8>OLr@Xei%0>^>-RcsrAj%lP)Q~_HY&);5?N<AnR!t$S
z>4@mhgaoaue=&n19T>cijlzajrgl=9>zMr;{5`7YJ&!tAjunRNFq9>G#vXCQFII&%
zXQhQ{;b^pa(%-CA`M5a!{H3}%6fInK!uhL;$tC-P6N-O@&FvYkdNy)q90Q2*(}O5G
z`jn96lk+<Beie-(6w(1IiLE@W-)6}D<6|`Z&R)y0^nPBRZO^_SbHh5rR;i;*+Ino&
zn_~reTcrWkvP2yz9%a0LRW<<3TdsL|R+=awW%_KiCW6Z)jF^hwx!rnICQVCoFOBll
zCFA*-ew*u*ni4+RiZU}>l}=#SZJ66AD^zUSW{%t1<`oQ@iC&=5*IOA@6Er-FJvCOx
zgb6rpS1lySU*6?|E|`w?<?Q42uEs9|42*nY*YusyQB~(^%y+0l$CmhKN<v08r6<aS
zUn-=nwDqk#Abya{!zPfnZpOO(ey;N2{YU9>T263;Mp5UmPD9^(FdOzHPiZ(bM00WQ
zlraMdx%8<ccAH4ltv0`_bDPF%*H>hw|9(NGc<0|y{+$@W#YrDLY16L|+&=G3(X;lC
zV3)pF+YS3QB?mVpg4)`L#2=J|l*9Ju(uluR!bXa^X1l2SGF@R*DjOPIKYyz4?!B(q
z911_$^rFJ<En|^RSY+QF0;Msg1a>&&)6&>GRbT@T8tRax&Y_Iw-oR*leEYVNh4S(N
z9$f`Xc)#>?yOMPDRNkd%(*E@KvMkHFlP_0{I73N=ap(&~U!kx0+K8Bn&FYijrrBC!
z=72XSB@6!&F={kMS;6&8Y3J-jQiG6OQL|o<g)DL)BguqyDbp!u8BJ^ptsMRSm+krR
z3QV=cbD9@&)Hk(~eqKZWtX)zRGS2K~&sXGN7gBQndds#W@vT28Qg1Nk;-|T-r6)HT
zl-jWN`S{4ox%Caq{j2j|p8>jyPG3+^L`;;4m%V!+<&g{!f3U;rsO<LTsN`+|f}z@R
z(yA1?@P4NU3)<?%PQ?MVv@J_EnaYYGyswP1E*{cXqcDUD6^{=Nw(CB&hB~9?m~D1v
z;d*nk%@d}|p+7#@8jWiv-5^wq+go!+=zE0xicF0=Q%{=E;a;BfCX#@6$SH=t;LK7}
zIOVnPb<2r)_?UEgh5hn5DW8CN60$0?YT<%$>SZchOuSz#YCGOr0cCI4c|SGdiR9e+
zF_;v1=kD9khk<S81E)j+oddpMjh4pOs+>z50qfn=RW7JvG9HIgLL+Tmjt;1oM-nOm
zfv5*@u<zD$b6}?k1(~?n(4A|{MW%f!&~th55Al>0XS^`GTCC~n%}f7xL*#-Y`a5T}
z)Q7I>lhl?9rzy-LP0`ROmILrau}b2XR8WAvuJI5+7y2VIW}_Hc71*&m<zuUu@e=zI
zAe<B9SEzU$PQI~HV!z*4=%In1l5%!pS)wu}vkgVOt}}D6$sQ|o?~a^ZJKEl$;Enzf
ze<mxKcCIA|salNN7YCyDv!$K(jW!2WY!=_SUp+nL%o-CZOEuC#D6&Z0`*KC1xxFjw
zTX1BtQ=GopM2bopgi@1KS;U0Ls_c~6IwJ%wmNh;E*unT{e*Qyh=~!;<8e*uz=}@mM
z$&5MTyw5#BRQ&fRW95NG)|ZhLI>n+as#*K<t`J%W4nJC6qGxUNUjgf%fvq@CBaf19
zOL~TFQc_u*9_|A4>%#}>NTol$C|3&qrG6!gY^}6nWpoc;{sI#ZwJ?<+@32_kvGVFV
zY<cIP=VDE4+oBf7n2{NKRq8L^;oitJ(*2fWGX9Qjw?s9G_<8@IQ?K6BsfL1+iY>@k
z4V@F1dC6ymc*I?V;LVG{aZul^<Bk$1gSezQ=56gTLXlw<76b1lV2LuSqu|aoS54P{
zu9nJO8Do2fm-7Vh)@1?C?|8VP$_BEt_3TIG7)DQv<xnDQ`^%~5n*=|Gbh`_*Neq|S
zs$$Z7m;;Cy8A;jql=S9j7ojFe-j%489Hs~BXBxGeOh>MZZ~VU-p=SMV1K_2XqOTvD
z=5r~o<8OeFXoUd*rx!d^)DMPp<dDVNab|`djHdp)7ioRUiMPE!z^v_(eXSudTqosW
zhp)EO08miQ+vr5KuW03WZ{B!!P;@6|jIWL;u)+z?H*|*1p}ykcJhJ0Dn+sUzpB=f9
zqBW-2T{r<gpD3q9>XeMTb3Mrt@DQt3*_UkgBm$U;NxhBsSk&&S>DX9hA+nd*MYd;;
zGjnB%^=01B8q(~jcuwwP-K-5S=f`&YpyaLaeR)N>4;c*(zjgR*{c;jkM>T;JZ5yFg
zR)uZ`Pz|J2ouo2dQI;-ARPhF25!{<<Z-p8)l(oNbCf9lV=t8*Uwt$CsBPrEn1;|k<
zXm^%2riH(q8lj`h8O3Q|$22#zHqL|=IyP!kpBpqaMr3E^m=C8*=^R8z;RYVJrnHo+
zX?bvg3wfet#zrdrGjHIA8u=k1r9U<%E5yx3L$A)!oLV`TURX}+f|~NRBNs^gN-x8Z
zs&Ctey01KM`RwY|UX&IESQO*70EAZ7@9D^+maFg!;3rve6>Foym5IiI@~36!5HHL~
zT{SRTLC3i}f8+I~Y2Z~khy_dOi;=kP0QL~4eDO^=#=sl`I1fOaD*KJInJvmdz3Ydk
zo+Dw22A1!Io)dHWf4?Z_t1QvEKrk)PxZ}~_L;i(nwyg1vi%+`rj*}wZM(g7#eeRMI
zf_oF0xEMj>xF1#@cZH?fT0>}*n00)=^vns`VFK$N`NQtfq^2--PGsHntxK}prYa7G
zb|){XyIRn(>k3Z!!aE6{Pxc#@(5ta2x$Ja?LZe<e|3nj{ZS1!YMQQB<RZIZ+dn=^p
zvNAe{l_k<dykmWPpJRN&pZtzj>RFLoRhAm6_LetZk$PW&xfzUI_xCWDv1(@SGhqN$
z7<r1GSp;@8_Il3T^66*|>6w$afyBQL_oXZA@_B_sYwrL0n2?fLlP)BSi(xD+r3?%`
zAMD%zBVkPa{hwQ1HLX%}gA{nX+~JPVCk-sqR~_3g6yuE^P>e^>V4i?;z>be@U<U$X
ztV%UK0!q{0y7-qM-9$BEZ<NpTpb@BY(jLaQW=$b0p**EVlOGrN35w+EB9VpV%-gPn
zIL7YiP2@TdILte)`SK|*o@I>|A>!B$gh2poUR913K+ZmmXj{0`zN};q{dH+*N_Jm~
zW)0Uj9Z6}*jVjnv?V@tAP)UjuwJ}6th7mL^89$sCnr1obvsRZCQK@D(35^<>XabQ<
z!{m3?q~SklK%-x1Kx0nH5NoVBP(wstz`+P#gZ)#u6IV>qY<-D9n66#bs;<5h`#M2G
z%Si#mq#-mSUY_X|Uy+&Gv{GlTbQe+Q;Z)MF-R2L<h?AcJ2kSSbsdkdf3JkR}9c5Be
zGDa59jVE5(vkl($6$$6gN(06M5C8Y_JN>XvGqQ$|v}Sq#+HiLN{#o2Q@Ch!vjST1s
z1B=IJd)HmC$5gmA=0HMiEZ2Y=@w`B<UqW7;T5kRh%Cp`ba`!kwc70-TU64kM=GCb+
znDwM5WO{k6bR;!|gv2BzN0W~UY^a#=$3B5)g1GG>Qg#iMi_q=<xqF+1&si}_mIhi*
z!1cw}l=&YSPeQ%?O!5#pe=>kCW30wXm4u?Wmew0u>Lmiof*Y`SdbTCac}rz-Zv{^l
z*}gw0huxb5)JtSNS{JUh!->%0!RA?kK^4u?_OCnROqZ;eTsy0uedQb40`mcL#PO&q
zZE|WR!rvH)&6BhhsXf}}fFkfH$jBfew&%wRRaCE&#{*$09dXW5!cJI*E`GofC=dJ$
zWMJl2;S_SmI`8}18jL98F?`}&>Ds+rv9+M?Qf7n_IGy7@t$?bOr5>?W;q@0K>@<26
zT6So`yF0_$c7+0c`}mpiIxX*nOIq8ncloxQ^;U)`VjX5Jr4KhC>N3>CTwQ?K-cHXX
zW%%+;oNHIGZ<q<d!V*nwN(<_oSP8WG@>qQA+37<oxH=IDdLb;Qd%ISw&*|v-nALkp
zy}|8QCC4dvcmCO&u0WTVV-nHpkVT?06)d$+EO04^x3KwD`!~+qpyVrvX3G0yZCF^@
z${br&zZltB|02{*(cJ6H%}6g6cV2;Tr3=1che;;SynbO`LcDxJ%WxQe0)-~s{UZrT
zpS}u;VZ#dtsGW2}#1j+L`?Ac_wcPT@_-c=^qa9}5*}5FSD9d(7b}K8JrNxGj&?uC-
zt+CPRriQJy7uUsGoNSl>y1MMn5PW?=c{iRD;<H;y9T?wStVE7<hHLlK!<_VhL0?;T
z#YO4)e&UL-5pK~}=U62sFz$Z`c&z=(<JhxtXfn&1-?2vykob|6W@++P)|_S5H(5%}
z(vKe8TbBrech?mr6;6gdMdmKhHaF*+EYD{LB`=s`hVuDLpPae6mK9o9z`K*G*dV2V
zvG>a_D@O`d@b{^L>P-9pz9&Q$SL0>b*4`lPPQpxcuw$2-Ia35CXqO^rnL{I^P*dki
z_BFe#7xal+vzr^qi$|b*V#=ZlHPggM+Q0rxdUZgo|NgEgCmdNK>36|gaA6SKRY#tA
z*1qHaJfEmusApnfmRvJY95{i<Ze}1q%Xu4@mw$w{VLrw`*B5jt{uJt}iZ$?VXSTQy
zZ9?X;LT6VqXsr8ab&Yu2L!IU5sO-w}M9CMx@FE3{{;CC9$R&9izx&oQXJrI<8YIf1
z<Gz`ILC~vI4U!Aq1$a#luWyTqH?We5I(jc1cf~z!;2PI%UH69$=IrNqVrT{I*K%jd
z7SeD*9h?PuW@mNvHdDS3@-+5i=6&+4TPB~dhZ7?bw(B=dxEJ%=MigQEnr!2i8ix=5
zxsVhFlI!JeH77;p6=9~sS%6YNo;Qz+<opAR`WZ94vYhH~L-X=|@cgOjswK1`7(b^`
z)I7bAW|2cC2Q3FWYishUvtd{1;K<Fd@RY1P=i4raoTW5lR6WXHqqrISN&z5<L-c3^
z;-R^*y`L|Cp4e#>82g9z-4#Q9CF5g~h=rc9vc<e3+tA2}3|R;qn7rfvywkXEdduR1
zaf}$TUy6(nTr@=1t94kGX&za%*deyC<a%>MkgeKr5r}3e7BZ%Nl8n~WkBnQ1wmBkf
z&e@>F(iBvPS#caT`nE<arB=y1Sr)CQzEx`ZS`MOxC{Y0A{s+t+waz*U!h^I{E#bM@
z;LtCcrM?zc7-zq8sb<czK&PV~qxG=s6vOG3?;IkGU4H90_2Ch8ZOw+<hYG(j$TkEk
zTm&v}Kz)*v(Ez*uy~mH_ejg5yap)5}jm}anNCs&w+rqt-*Uv_(szga5f2B(A9&5}z
zFf23#gk#y_4BPI+M7Yu1&)0Ljqh@59LDTH)ugrFou?JSE>&J)obG+os6Dh&Om0OV(
z8*YRb0b?a}i0griYXqHWuUtmO?b^A+j_hgy3KJ{9P-)zkF&>RG>D!*^z;=07DHTfm
za;4bML>3sW%{X1GI1U2~m)f=4G{7?Ya+%Ag_a&M{akB2b&dwCg(6J$?tB>oPI3`R-
z%!qmfs1@2`XqNi>Sz7pI<Ezw!HMn~I+LB$m>NO}|Hr(;KENp8lEhyR5e;Fkz{yCzd
zbS{w?iQK|VNj7Ml`rLSe{sCizA()Fz6$&SxAcoT4Ml3AJqf#S9L<~o2goo{)Yx$dq
zfX|^`9x5u_bA2iOwV^Vrkk8J@YL!kpV2q}TqQGEo2>*{39GJ<@mfH1BYT=w6rKX4V
zkP-s3h$fuq=}|~E0%06>wL27Pw<HlEP3Lj*1_Ld<N_)@QZ)?EET<C%^P_zG(6ofp!
z!uH__A0gHIfVNauf6@>m?D`UN0F4?vAR7K{V3FW90E#VP%#sPpzPut%#7Db1+A2f?
zX1((+$1u_c1ynKS+R8Wqw2XWu)Ar-TMVvKqKX`XxkHSNl-+Ty?yOX96-iM;Lei=0$
zs<?4vTgb;KLbktNF?m%Bp85E3g3?A6ioj`axe#q&I5b}s8Wx%XSeOVYI$W@a0MIDr
z>pdE>qpu32P$|qP0B;P}2Ny1a?KVWNJbu4L=M)irNJuP{Kzi?<@m++%DW7dGNNA#L
zYxB;ZUZ#d7oq$q*`5VK}c;UU5w^R>x$R#CMZAQ8jc6#p+gvNP2)jQ*FdTg+fq&_as
z&Q9yNkvkIhE4iv@<N1S4N=b?BW&ZV-_>ey<%}ePh!#(_tgR<Mg7c>+2?_%gj`=hc4
z7yL8M8=jA)M4u^n5ILWm#>2iXDGn~n4>dG~3ktnI%#Lg7SlMxCUbjJVRc9P8)zoTF
zn(CbFaBlzMKh#0K@a$RIJ;`NW(j0gAELpv7)9e_#uApaKSt)uUq;5N256j>1@Ww_b
zMo5Mmoh@@-;G_V0dXQyES5{1BHG=viuCZ09eq~*wLqScI4^2c$?TZB?*AI;cPhM^}
z@LR`<17<(Ppn1s(@PFJ;Y-4A4wqXqBQ_6VMaRt}<P?~9!G;26&Qs*#xd7!7_+R@6H
zNe<JN!@Z|ZB_t#V?y(8jkrW#PW?`{W9Nzu2zy{r>fHIy!zqh=g$>N~1Z2jkm)YFFP
zP=n$Y+NKU&c|*A$tflEGv87l#tl`?BzIrZZ->^S#PW`v**E<w~2_Q5S`R17pWUeev
z$Io+bIRPMXC;pD*HNM^9U5i5F5xW#n9k686MAI}ep-o`Y-`lAL+dWw;4mmhp6xtrx
zqlnCRP9d&0?@S<qZb`JYD(V^<zV-DaLt^E;Ql*^{bQg11eXX4AM)O-B-5|`(us49x
zvPL!G>q@`++`on_d^UzoEZ4vW1q))f_cQ8U$kX+Yq{FBNo(8k`=$xzB%GtbDQapBE
zS5I5Uz3iK6YJ-vkM85HXj+=P+ca+tY!}=SjR7ePqpej}YPFXKH+K~t5eMM_!SS+_K
z&C}dTUeu86`?oa`ZB0r=ZK*)cjc5NhA<y#JuwB$iuSyzH7RZYM1zWJ8S{_EupXJ3{
z$Tkx5^FiFtj5YdT-JsrVr742NVJ>q90P@2q1(>{Al;#D>Sx<Kc>q1_*MZH+yyP)Kn
zi=ZLA9Q=IshoI`}lX6jaUS7cmL>6(DuTqpK0d{O~p^u?VKf`2%$X-g;^(15dw6yHv
z(Xv4Lf!1KbaHsP0f}kIdfT>(7v+EzBURP>Crfj%4zkd4mtG2U9)ZRL?YTK#-dxpOF
z@4W3*iQ=Jx<Hh&>$NHXhtr4gN3k;+rJLW8$v5?_cM8TU(FL0z-m93%u84+J~YyvQJ
zwm%s8Ol$?xy8bz0pdxC#Y$FywuY&^z=RB?gb31T(nDq5ooHxcvh@9Z=lbW%yLR=(s
zNSS<!;TS`Tmf=q!jrbCba3{@=Cr{Vcc^^9pO8vptWrx$fRo~liTn~>egyvM&Ya?Cy
ziNWFZ-_Ec8oP79=E-W&*<(mkV<(cS%8}f`Dc7`ppVP`g$);00QMCtB$k8jQp9m4qC
zg}2~Rrw1HbK9A~D^5foWx6B)*ii+M|ebm_g!`c+{xHd$&EIYujwf*TUYJ4{O26b-L
z<;R_oV+*XCVt;L8h|2Y>c%q;jG0?&Dz|yv!Z6wd1rz-1BhWwPFesfFA)yHca!q$9P
zpHB~esLD0w{CZtkbyvpJ^-8ed(KnTP|EZOs{DQsHx~w!c`^A55P7}e|SZjtxrP8d`
zNgS`?kl|bJz8YmFv$jSS5=<H%=<U`2<@nQq5tVX@mXxNaL%VTl#X9gnMzVjHLM24o
zwEtG_EbD6h@FnW7HEg*{#YQeAjU$(?_)SkYcTV(`Afim_ohE;FTx^@*&H{U)<mZZo
z!i_&7l*hBycoGtpoZJg@@q2#T`!xggNBuCL7Pb!tHslkCe{?!Z@uN%b*HCSg*QJ{y
zmSOERk{|sds=oU)OGxcvA;%mYzWU<Lq@?_r>|{q7wa3Swhz@vm@fPHJA~05*H-&_U
z&X>f*sip`9ieh<diL{-i;f_49%3<Zb88GV8@~yn0GTXpJ>zyH2j<5|j&s=Qp;{Z>@
z{08B;-iL-egvcXsmi@oqLf2qr=U0ipIXyv_{JLUM(Q-)k9=eWsZLQaTJB(miQkJk-
zqX~h-Of5`Tu&FaJn`;{Nnp8tar*fj`>q3~Tih=()iCEy(4_i%lo9BFPbiBuHg(20)
z27YxF8`qDo<r~BbJ^Q)o<Uqr1y+g)-r+S~{51fs$rmjD%ou!aI2^uPm;)b*W<Wr_g
z1J+9RR;9+5haqobr6z<N+I_DeZ`LDV?UvJ#CGqYu9@Bx{4%hgJUTObXjA@o-K~bVm
z&Sb5^f(>unXr5Iizta^r#~vya=X%{s>MN?Br*DVy_WFRWa;>n`N-;M7$4Uk#KlbNy
zH%>(Sc6U`@IzOS;m(oYRL0t?G=KW2kJ;1p)jc>pJ60G^G&BiomRnMZ@+SJ&HY+I2t
ze3Z1nk{F*HGQej~vpw{<e7uy}_2MZfsHsetwQk|fHgj+|U`+1T>CO3R@7}|7#GnUW
zFC-;CdTB54Ff{b`ws}2vNOsd`f)dQ`Snjjo_P^g6h6`o~ci~y2JlBc95vHDq)QxES
zWlpFa>e}OYDs{9;B+=aBSmpF0JT9WW<?R;-6<YhZE`cXz*^c>=V$TfPX6NGHY)eW!
zI;;^2@YHtw(L(}M_TO)tC-=+UJKrfXJ>(EL_K9N5xV^PCTV1f_pr+P8Wd1#np58#F
zzN<Av>6mbPr+rL)=BC|k-8WuuK5gSc4we&BH!S80cSmXW>g#8g78`exjL{!QhVvi0
z3>S;!D6)gT$Ub_Z1f&bP@_4>7nSxX2TYV&j^;+J(3v*PiqPWh(#Wy8y?$xw{hL~i+
zFxdDCaVu4l*VW&qJ6v|rJgCC9RFRz@I8`2Ezt7YaD>`s{3(e9jCL~=Qap^oE&vJe{
zo9tX2)H0Kp?%Zbo6cGXG&B+;sh1U@w?y-u^!NNC;{1GP&apbq~8{TQheGT+C0PD!h
zt!Mp$^W)6HPyV4C0*O$cgpKy3-vx>$QGZ*Xr@HPw#!Cs>xUoOeQwe*Kku;y3TS@8b
zWIr?T@!I<gMXB>$$xq$g1x1YEP}8G>{$qv^2}+ZmAQmFgDP8@{LjUW(hRb-osN%kt
zCR|CQMsKbAgsP96kciNxz1Z17%Vj8Lk-z+P2*Z3zp7X`QQMl$(xjNxXd(DYjh2hu9
zR2q$efr%OQf3(j13BTq$ca^)57iMu423^r6WA67O9gaUL5$IQjOv$;gX1x1T9%|Ca
zoZ~|M!u#Ey=Hl^cpft8W2O=sKo;>s^Q`ydRV-sTCJj+#h^!U+l*KwyuWTeJs`7T{o
zv)bCc8=F549<d~og!isEgEizZ?~$36pj9~9)4xXBwg2B*0!5XzxOu^|B@%{E3H|E@
z(}nF6E1aTGp*PFvVZ0cLVs(~9WgjV7T$n5K)3e>2Ygv_6g<3ifnsued^L%TSI4!09
zZ!@woqdCtys2hGX&56_8cdfLE7`Q`;8j#RRk*G@~dg|<aL@w`r-9x?!BssG<D+|iq
zWzWZPs<)@Ekts?D%Rc)!G+Ak5Z=a$`H?xEeH};Q%iX~1p1txMS@~>W7hTW-dEy{MN
zuolCUl#Dn&HqOq?4|f#<_O6Ci=Hyj@;1K~sVP{R#7bR*hoZmJ({k*~b=bRfks&Gp+
z9t+#GzXe>J+&I1Sz>Y*beDucr{XKjYbcGnlga2vDI_{;1+ny?`!a2yYkj;Yp3g~4x
zm!_9n!JQk~7^g7VBZc*4!$Wy9wYaZG27k{tbC%^CMNx3;yf<Xnh`-vNn#wV|(bg*H
zdPC=$@R{>Dg?r@j01v%_4z*C2%h!n9mkT}Yiv`=Vl~s>qj8ZgpoSol3?O5k&EP?Fg
zXo9;q;t4=Gk;clSgszvG(&EtH(N4FHBW$`gHiHff;agfL?GUa<7SCsK9Y2Yb>7>GC
zr{3=N8>RA1t1z!$1$rd+)Obio8Xw@D`IXCbZa{f_bd=ujT;i;JeAFtPEV6eT=Z@oW
z91_i#O{M(~54-h{S!x}XgV#d}N0W%)jJq29gmJMvfu#vG*)+@aO3*D$>G1<So=_{E
zQ-0X5!MycjNA7DGA+*<!*M6_<K5n)ciC(ko$}l={JYcY+qkFzQmR&Q0<J_6VA)}vq
zVtqoRLPW$_-xw-!CSY~j4d}o?>5*daPuKC=*e#(PxlOA+7l#L$-o6IM5qCA)+JJ>K
zh&fUn#qjmli-J)?-w*u!=wcK8>!aL1qrXe{Ur)G2qpNSVr2h)2h~MshS83dUJKyi+
zZKa}U{x)Y8POU3fPn~Zao{^<ON3`_R&mzFYfT!Etv1zF7z6zc{B=M6BP09~d)$8DR
z0`C5bb1sS#=M@;eAUDbV{_NBR6ALRCV+4;E6u_}R$auuLQ$0|RCn+q<F4<BN68qLr
zlo&_e`x}X<|Ktcs94h~uHmV5c1$F#Ga1+~iCi=-~4LcMgTkAiAPs&UPIc^f9-?L_6
z^d@U!EU+$E7(Gwj>aMIkA*O5!h{|0pc`S*~M^;u=HfgFa5uqs~^!A<?Q4m9>1d?<K
zTC)69ugmlR4|$sVg(Xd9=jdB5XFX$Z69WU|TiIt*!IdzFVvQFZMH_R@>Nj-Gh60iz
z7*8WhXUvTPuHRy;t0R|{dYU3$VT|#wpn&`+ebD|7O{jAeo(2E>@zXmkSjRutPB}E$
zf%COo+RBZ)i?eS8WeEdlXtqXTUCPrF=An(E9s^*|`1>k3!)&vTa{ME{T*<#WvfY)D
zZwj;Z>~D+g`+hi6{q%tZ?bxWM&G;Z;??%`7pu?WB8Sg&Bpp`r#|8VHe72B}(bbtTc
zd*>!2ktqwpL#{{9(SE3k+ur|M*v-wmx3u0GUKRV(OGZSngY7oLH7+EE?1~&TG%RUZ
zW3$|^l4oTV)MG|QtY=XjgO9BOQlg5j*q;SKITKm;25LIO4a9EGKwocuiWlB)7#{We
z{!lP~YVg*$))^V&5YMW=Kl)x8I!!D*j74<^^oh6ymk!XbyjuNPd*epOiXF9ch2wrB
zoVB4Znxr~G@}s@q&gqXcsCBI?;^RuO6*c`v|M|LpdRha=+BH>nxr%Pk5oGFv@t+73
zs5@Y5XdOHVIM9nKzS7w?_7H#~Q-6OR0Uh=zi+4M<b)SXgyI=L66WCc!58W|y%#v==
zAoC^wFrYvEk$~1dcU9$9GsTufN|0YUKLCA8BdtkSR5DYiDzzYTxcGD&5j`iJ1EH6v
zN*(7qCBZ&+I!VER$A-~wKTSHr()qM&Mw(rQkWl6MZGtAo=kMskb<%bI>`ctc4aRg7
z=ZGFS=I=T)cnQ^BxS9W$x+(4Hs&ll~+Y-W{yk2G%jltnpnN+trNBbj7POi%5<*i6<
zx|h*yq(kP2M}67&r6MBVL>hu3qrPe;{O#)x1K7A^XoS;PtB-Dfscu%FqLGdD<7t=9
zyOiV~8mjL&#Khdy4>%I7Uv%H-mBM9Uct3zO`#4ZyvY?~mq4+F8Q#oPF!72<nlRmQX
z)3=o8qpa-B52t-(tNxNt1gh?G2wywN%66oEeU}Pa;oJD4_^f84-!p4gQQq2W%ig+T
z#=%fZ%HNWv`WvyUl;M;ZK*duLL&*=r;}tVLzvQN_9=~H?olk0QuGJ@|C?Vmmb1OnE
z9Y>bA<wl=Ve=;(9cQo>rxnbD?1{ocq&b1}c_CCDXa5ewKlhA1$GZ6=UV(y?Iqm{M3
z^s_UFAc&)Hpumx6j5@a~^1s6Rr^-uw6HkbUA${1nwRteT%VkIrn_!kyI6SDX+F2!h
zEZUL4+CQ(~O370WKMHE<!#ZdCk=n2A(%jGTyWPY*5IvbKLo;9(m|9?76Vi_z`IUJ@
zskHBiiT)#mljYAD&i<R@WkQv~sWLRGt|EvpBElJ<M4x4wvF}{Gypv%zQ?k5MU^=U+
z?|EZUGNn*pq)tx{N0ob9&F#F(&O1Ua@SQt%Rq}NmmkLq(V-#BBEsR_7|C}5pKX7uC
zOiZCZCU7`80Y=J7nG2Ve_O+_U)Xems$@qZRBd^SDhbum57j{Ejn1ZbOT)!asFvLG|
z@<?KE#``1jl#*XjFIxwq=!XmIH*+4MBWRZe$g{IIPXF4*D>4n)@qVicprqzu_MOhN
z0!nMRG*#0wBihxS^l|R0AHnUPO3h8IEU2GZQjARVADo|U9eU`7hJG|L_47)>J>?(U
z$}inH0xk@kc>;B!^l)=4LTTd<DIY2}zJ3tbh>hlFmymeTHPBEu>bJhmYGY+5#8jHU
zDT;+<uJ738*e&;UtC7|lr7sIazvAC_HD|@q4d--?Zq2mJdX@F;26iVVC`MRDjR!a=
z-oRjOtuLAxS=gZ>E^hjl9KKm}uhX)+<#E(YFx=?~5lD}=;Sc&QDUq<1KJk6;>)bb;
zc#jUdFbEvc8#nx8M=B;i=)rA_Z`I{Vv=Z3H>0mk1mWD2Z<$u2^WuDbA^FIh-dbkou
zIWuDmS;H7h`%BI!hIgijI2OAz=Uk!@W6qge-|#wG%ditN^l96z>tFxQlk|QkY-{{v
z4@vQ(DPK!KY=p{z<nA%vF9DQgRvQwhz0Xfhu|9ILB1{*)Go44J>a_VH6i>u|rh!#^
zDjb2yEzX;bS}@?q)&|FQn^@?<KN>MkMW`JNrxVRJ5GklSuz%}m$|x){S^3a$6UN6e
zJVA2nGp*#QW0fNdS)n7}glnMsmAk<zP<3MPQ`6FW(kb#8xiR^rkI^^nKiGNMJz80L
zs=!bZ8|?e@%N5q5BF<^2Li`V737}`7=?VS%lu}f`?Ix<!GR`-f1vtuY?$5TFnMezl
z3y(_8lv_9!W>%#fq@IH(y1lp8>lodlG~>X-XT-n0w%R&3_?p!~_ho`dT>H!esD%ln
zIo>?`IXRQ)n1#_Jd9Xb+L8h8(V3Exn!6GH~i_@85Ln_8ECdH$D2SZYhi1wC|o&jq$
zHilR+m-AgV$T@Pk_kZi*da)j#A3GjOoh=f{h-l(Isjdl_svLh+Tv*@T_<gmTU5JBy
z=rKjZROjbAUY<M9N&3C#j^<%9WS(RA*Lbf#3l$%vAPSU9m>hz6sY2KrOw|COeH73F
zsI_ir^mc^MTcNj40RzGl5z<^!@{S)m@PAl)bIM5Xt+LT$*tOvgMw!PFO{sPQOg>*M
zDrDiMyju?!XRBXOv7MM^=e~GFlb;+SskG0KAOGWVoJ$i_41C;46x#}D4C}D);Ud&t
z<JNh!zS8D<d?%lEn-{0;Im@G^1z`;_zEvqHhV=AFbvss!&U9%|Xj=#i=JD*9A7_$h
z9!vspTqMcdrbp~~-Tvx_hBj4!jR{{WHqV{2E(L(jvi$uX&98`%s2D`IwJ`*@c(XCg
zzR`SU00k*HE;Pbj2{3=^EvezGOo`%jI;sm#{82n{<gb1#R2>HV_IrLG4{xj2l2%vC
z5tlwl{s+#6AeXd`UwAN4?$Xk#cYuL<)KO@7NfaKI^NNFU;|Kqj!tQ*zbG=4IB6VC$
zHJA$H=!BZLq=AAQ&i{<a#C<$&QzhP8Pu=nKIEH5K|6qLJ@cMNxif%V2f8p2Rp~{1h
zkWT<$fw=tWm32egA96V~jbP-j)`SBZkY4(;;k2K>S5a&X3%vPJT#o*Z=54)^wn0%^
zhvG5CDluWyILUvt^ZL}({%75WZ(_=t$Q!2%<-+pHTvzVcIS*BloWAO(H>(9lpB!Hx
ze1X@lk#Tmb)1p@Fd;*#Wa?b2#21YgUG89AhQ_sv)vS<_H@494!@E~}%kP88%E6J5(
zoZrcK9Lw!@X~0@IS}jy+=-o`+U300DO;E<P#~PgcA$-pe3QX5*%UOl7^b;0}^vV5A
z@%gydwI1nF4zlM|+$yJ@8DY@yUke%14#E-pyscsB9Q_3=9u6`CYYu_6p07>WQAzX}
z5%b3-7KDJk+S&U_ME+ee`8X1*tKt$QXwZG#f2L$)q=!15GwoRpJF=GR4n56BFmuN+
z%(jGFe0|68#0s(0JLNkY<UBQC_K@f?x3m=Aj~XK|urZ{zZ#*CbS|t}CEUgX{J{mx-
z4?pLz|0$7j--LrB%b3ykh4m&5agJI;xB*))GJ83XjVl(&!fk@x#}oOO`d?jRWt+o+
zg85w2q#L$`^%YE8LwA?zXtT=6y%Q=0bTIZkk2sPQ#H4N$l!gVj>_xrw+uAqVd<_g-
zM~5cP2V<D>Pf(O$LdfQ_PYrxk+Hmnl*FaWM?s6y2l4}`tb<cS%!}?UfxW=EkBC)iw
zlw8R**5xGi`>Qn=X60=ShwIj5Hh*dqcO>l0fPfw{Z{Gm?qu=WA9A&Hge{s;;5)#s)
zv5F_J5)$*+>seK$fCvz&A6Pmc&8i1sjE^9BM%hA(+K31RLo<sOU=*}e_pUL8E(^4m
z+4;u(x|1GE^Z%Io3aBcV_iK<)L_m>{5R`7D8w3Odq`Rf1rJEN;K&7O+I}a(6hmh{>
zlI|8Hzd7Fjy}!?;Yq8{<H)fuBVn2IkgF<B|o;}9Dq1JK^wa`BqE0HGHNNL|T&v&EV
zD+%N=mU(N@QD-{9#vT(w!D9I)9xoZOMTC{Mw_XA`BxQRtRD~mF(cD;0QeRjIusF7#
zkYQt=U|9!>W&eU4%(#=2o?^uF`CHg%k0z8xA+>rn@Nzlc7x%9J+{2poTK!9EA?7uT
z)q=@hE3Mg`_q@+B?y5Y-7UoVltGjDetVO)|wj)&97*QpFAO$RFu(vBb-`h%o%vOsN
z2Ln+wh0=`bJ7jSaj&S_>?ezqAv6>GyL40;oNgtO8#l7beh-DRL;dD!yJ#v`mbtCnS
z?FcS$t}4u|;(`S~755iw2CgXc-J(FQYQEc_CM6lhRY4I;RNLP!#LHRA^mcM7_nMN#
zX*tJbOM0r;#W4nYTSh12a8i@<xqdHZE9&3uNj=C?29QS!urjMRd+Lq=ER8ip(d09F
z;Q`@d0P4jDO2Z|rl;wuO-k}d#t%1r+g)7n5QtvhlBtYGZd5Nu(uX=Rx$sR`!<NkkZ
zJKY80>R$wn7g4ece`rkED<bAXRq?xEraP-FH|fY?<Bg9m_qY{x2}r8z;3!U&@xOzS
zMjKsjxO30L{l{2gGOAmy)k6r_kQf?XOpH-lt}AP<wjHI2=F4o|e&Bq3o^m5)di3`!
zj3+#elmr?7&*&d+jZ>8pR`qqsLTQcFGqfS&2_|*P+3C-rkf)N0YRPJe{OU0=8r?|F
z6duQ)!cpr+Vf}fk6o!VdlHR>R0{m|Vs=<vhRztIMPpNQ0(x6fe?XD?%m^HK`gHYR(
z)h7^6kM}J6;pi;<x4U$9<?(OHJ4}L)TF4(wYzfKxNT8{F-)U5%^w7oWE3_<Z;Bdz6
zo12>?Ol-2c8%hDoHancdW5Bn2D%=dpPB#GowzI>0?PU=+FNQtr-&FVnY=+134n15G
ztz#zFRdj1)0wJi-`Azov(1dbIdRBcBEnN2_KxaFiU&qZBxHojfkLRHEpr1*ks~)Yy
zE-0!$FLa&pv}g|Quvpt@iphI2FW`V{D)ji!Av+@WjC@tm3%5AI{e^k?n%cpCpmR&B
zG?knGT?&wTkyr2k9DSLpMu~2$7{(u8?O5?*9lC73MvAL*aBg@^bT+JF;O%$cjhXk`
zeBV~=Y{Oi}{M<WZJE{0L?N?(~i9R2(DsO*L+fq>*TbOiyUGn?JeQ*rfSuw^uP=q<?
z{V;plt&RJ!ZFTj^d4&$CClKprNsQdErEvyqU@KgFFjl7KA-<fdgA72G;D4}v&uZ^Z
z2tYNSb*z_XEi=<!TZWYz=YN3gP0N5)B3s!oFINgcl-J|et*q%9mUEq0qlLP+0Ca(g
za7SqCer#0z3Z#+N)(ys{X#{CvMl@hXh9;lEWRLIp)!QHN{=vuB9$0=OL)>sX+4&_4
zPq#2)^LWc`t*Ezmviuzx_u3j0PS9JZS7hc(S7FVE#ob?+_le*ZcSUp`baeMd6$h;u
zTcvxO=BYBQUY#u%S(^Nu8g=q0nQHoyHN8*&GW|_^os|61&rokVH3qXV;^tmal*;da
zd}gqR9gd&O%*Jja7gx4h9+}i}2cI0%@ChW6yY5)Ozw~UL?~Y)`TsJf{Y8gIGhu%z=
z2mNfJD7LX(TjOw8=2q7a{n<bKu}^Im?cb&L3Bo}X6rbUtq!Ki-9ChM#9<cHHnb#0G
zKc9dD*Yj{po{R*y0IWzI)!x1pJW$}qk4RL9#DZ#Q?NoK5LAPrErHZ@)EjjFx#8=}Z
z|641gA{BBKZJu2JWC)9Tb#;f|Ro}SRRp&dvPdC;#(^ZT}0UYyn6B(SFtm^e+k2Qo1
zgD9TL)YQ<%`X}fNITLp%DumU{`5~WMte0pnyCkcnZr^m!fburrzW+xJ98&6o>MAd%
zEvDQ^6=^0Wiu}^g75ST9j~qTMMni|AB`A1ZCO@>|9B-td({{^T3Q8-vP}nr|(6?qz
zpu+hV9Vp`d94hk``)Ku%+zqP`Vfg@D{cw=%y`D}Z2~AW=ZtLuYTIh3kpD_T}*HgWM
z&0+`uK27^v<x*yuU~ex;|NEyHvOUtT3K`dSR_;a}!rc0BX%iQ0BI<12zJAk@kasDw
zRjJ>5ZlSjZ)^~1fOOh^JFl_3#sznq1s_=Q#nLDQ1117KTpkG(TKZop=CI%TlnVE#E
z>xwz)3*Y^P8}GQm%M5Atff)oi?3yrl^uB?IzSS*=qtTvcD5Yr{6CQoikz3yj&z_Ff
zgyoFSPXwwg_=JQ7anxlj-_yLgtFX)8v_>)n>GpI{XJ$op?vRcf1R1seZa}^H9+QdD
zRr%YxJ>GO-aX;r0-R6I*+3z{G)?>2ITd0#iW2zt@eL{>E>DeM6vP=QIrk`1db7+Tn
zFVIFn5!tUnsBEY_T~jaM!p+d~?Mq30vOjIqn%i+#w)Y9`BvD@EKRSEO0ENFO^A{mc
z74A?FaAA{aZI{~)%eQKhx6Quf-TK|9<@|>skj39Zu+iyQ9ZB-s!>YJOCsygE@yTnP
zNC0rXVd+r3?H9=e7gW(!`VbpUt*4vx<Eqz*AX%^{_Br9_F2==7mG$$DsI_tmP^V~p
zK=u;IV?_cXw<a#7YkUMUf3i>^6>_d;K6d@adNVB1pN}xrBdLv6T$98;nVi-Cc$vEI
z)NqceH-4c4a|A^@oi_Z%_Zrq7mQQ7jE2ON!WcjC8*K@iUe{AVxdF)mTLvBHr18*B<
zar2?AcQVeECU`aE%W@g#SiO?2jjpxLms>|JnAa^Od|#-k?57D7N<FSWFxfE`zp&tw
zoMJOQI(60;rJA8Ka_@a<W^}TR-<osXckJJlr4>t-MigZho97#)3|IQ^cBg`mMZ9_@
zNO&s1&;T&FKQ(-Ut6e)Gf!+KFRsSguI*0L`Pv`$v8ZrG{#8wzzo@{LnHFu>eT{}5C
zLP}?`aie%6!&Clr&L0a4b(V1j+b=wvoLp8wyXW}?jDs^zMdb~wIZB3lm7R#YU6Ql9
z!a~(RuW(PvNv5EExl3!x_2;HwXko3lZTOnPaI3M=hg-<Z$<);3HOXCp7WVQVxP%I{
z5L_Ts!>)F$ug`xgg8<XRWyvo^?|exNdLk02Pg9lmBwnn6Vd5o2TUT!sKpocR_L6HF
z%I!3G2w8H>9e&A}grI_Rfh4W9`N?7keyEL+xm3Sc2i+X4!3su@f%An&iCNlXw6PO`
zqW^qKlVf%wQJ<{j$-KJwkHSPuFkALQ`935q@rh54o0nDGO=sWz*89=<AqYyRGBJtJ
z)BK#Ajn+p>`eZbpQv4%7*Ot_MknfWlxPO7AB)0+6JUkmR25lV`UG>lpjb!RMW=>Xf
z5fklUnT|ad6NAM!o<o!7*rNk;AxVNIwA4bJ82XB_qncN6qIx!aK~JCy*p#h^%TdwE
z%+NAFjLT|Hi|)GB37{A#gUauJs8bR^ozC<mOsU=5+T1j?Ie(OF&n5MZ7Ty^-K1wjM
z<d+@jyE|Aj`*T;fpnucMp&Lc^pxx923E;|-%X<%b6AK4&)3RPIAC?z*l*yZ6=`6YP
zfaVkv78ZS!#ry$7VOpA{a0pzg-^}bK0S|7ZC6ybr%)!nTnt8V#7q631^pq7Ro$PKz
zLBwKpPv-!fa?Qn-V6FH5gSYB$UpBPc=O<Ws*tX77%BrZ)obI({%{LH_mZZN432}2A
zcAHP)Y?QtzyzxxmoaTxunb`9w<K{88gbTmfomdxDkoaFlE(LJPANvnZPKP38Ao|!I
zs}TAaHQe}Qacz4HAMSBrV9koHlGdyD(U95$0{Htt{t=<|O~Ju~<~R2^fF&o+R~m(b
z3x)r{zA-u_bnF%kbm?}eIfHI&*6JFC2HpFzs<!T4JZY52V6(^<*j5dvr27~bTi$Bz
zmfGAzyS@Pa)}L{g5&-b19YguoHH9z-|5_YOENKf4%1j!P+zd<7$C(KO)ny-+p0b4~
z327`#MwR>^2klipe+iS7RYObb;Gp6V`Ta)wv5K_aY5Z=JEd1za;dzm#F;&6Gw-fT!
zaMAx<=Ep%zkf_eOg1(r!4vBJwBVCNzd_iD>ih957B@X0B^*sLS=&Qi=YeiU8;&d$<
zXj6*A8R+8_74P|4p=`ZlP-8$eJr<-w%Rur2n#6WbM<Qi5i3S4>8$1qHzKO;WkHJz(
z5*wZ(9jYB$1aCQ7E^EQD@{w@aIsTY-*t&d5Dpo3KX>Cq<q{n~UTlyUJ;?H=Yh=8mt
z6|S@zM33C|H1(KA_4E&bHs1-_+8T*jNT6Y$`B$rpZ&76aK#do(XAJk`A=J@nF)~Wr
zAj|K0iR$J)#=Ev(GEY}R-~I(g+|g09mV1jmuD-Fo8Lp}LNjckc)bf1m*uJV_q$bRB
z^Zh$sD{NAuvzlWrnfIvG@A@}v_2PvbG>Zo&p}B$q4&mE=PbW&+F#M%QJHerpEbzrp
z<G+ocV>9))I-mt^ckQZogO2gvn2M<s)KDou-1FNj*2%nsO-*&@VEThG=cNytj_uEH
z0PzbcV*@S22M0%K4Vfqb7amyzv85}s)THohe`hrk^xZJE>izi~=Z1Dj2%`Mj&)V1>
z6((%jlRrYY5?$%)^wu}_F8IDCG4cDIBYgeqC39_IzMxa7u!+Kjt&2PmfQj9x_oCEw
zTUL9mPqLIj3vi6cKuN0$6sE&lUFLs`T19ka-(<vRqY2IGZI1LNuYKnR`O7irzN0pr
zGa%aZNn(9mmiv$y!0tea^@Xah%5mEY8spl;Ppi>=nU;!!1a3pZ7h3CW<E$v<vO7z)
z{AzvmajbzaK;-88<+(z`wui7V-ZQbSj&GUj(Ww>a42)FXaEZWTWvHBN#v-o1-e<`&
z-4MjSx)s`*&r!>~u&@M;yazt6EPe~gJ=5ZW^i1?nI_Mi+3khxf+6o4prFFYkY<6SB
zSE+6Zp!t%aoWHfhljkpT;o(6l^I9xjS&1t0u%)rp>D>D%;I@F7LIR8kP)-f?v|@pu
zjE*riU7iOwEsP%>-B9B7FyyO({Wjk|0X!f%fyXxT<-aK<s_0;}CMe$?LYZ~0hRcVK
zuZRhnb8`nC$AqIb{a9<P$DRGqU(&xZ_no4wQn~-;2dc0@W~jH>1hPDJZaFgp0cewd
zW@&Z~t=SuVlyYrG1ZWqCn3ZfR`rzp3=jh_C0gAtj#A_Mip8mkHU9$&a&;=S{rtQ&G
zQEp|)H`0X)0q{<`Zw8<hWe=LC$xoeYS>OhtsA<vW99hIN%CjsSBB>Zb!fU-_R&g8K
z$1W4k%)Z<Br-qz9wFel7^$(v&L_>ZeJoA%oUHf+&!ogZCj^b;az0q*zz|c~Po%<;D
zf`F@%RL;)5@ELva6l0mQ9Z@x3vg#V0Mi<tPaosb%4uBCr>c~B<6i9Kixz+7)w1L=b
ztNYNeSIYM%933q3m%sKbw2YnQi8B|b?=mslgR=>Qc*5YY`9)}V>ukAo&ZXmLq>}Pi
z|9pC8Ax?wap}9Ilziid;k(0}qjUR0Eml61cvC+FuU+hyDju-4H2B!H@U}}VhZg5x?
z{=O@Y8I%Rb8Luub&_G@y(z?R+@JApn>#_Jr4K98Q>-8mNMG{;H4plYg;_erR-hasi
zU`4|N^bJ8S96rUTNknuc87nV1UW-N+<643)&%|V8^V2ydrS9iZ={?XI`(seYji=iA
zQ_}Rm72JYCt$uJH=O=WYB6Q&3@Jmsb!jMQylZ2$q@#x77u|N{gHFu?Q6>a!1KDpMI
zl)~p9cA2Nqb3;ZZ4Z1-i7pKj0^ZMIv0{uy=kO51&GRt~}j}RZ9F}v}V_x|g>I~&qe
zC+}TNuYW9Ar*aZAQc@z2!<I3ts^6^9GDI_(BuD@g`;tKh72eyAQC97ZsE9?<Jep-H
zE;GevmmJXf3EF-e8#|q=<~bEVT&sxz`)2~jo0{%h;XuDh6dcEXljwo>8474vRfQgo
zRSY1I1!{%>m9E<2UwU3gOaO|}22JIGJ_+OAyb`_lrQa!P^5RSh#hi)Fm#gUdbF+=D
z-!%!SK(rRYY@-wAZ*#Fh9XBuF@ik!KIU;Kf3|e~aYiHn9GXBFoOaTROpUmm=2J1Hy
zxdCmCoKyg-DJWBFocL_?Tk~XGA_6H+R}+cg9IZDtr;F`z_Zhv{8yw=im6`7^+3^BE
zXh&az-1+FfmFAKo=raNlb!jYfxEm`_lY;oAqvaajEOYOa+ZCz;bS;<6=)V3Y38`sD
zO8y~;%Bg01ulf_Z1oZqYsa%m0w6qU=&fZd_CQGy8@RXRC9`6l3Mylts5F1JQ@oy&c
z@!7^pMAhHY+2s3MEp<)j39-Y<;g<&|!;Kkuqp8{OsccGk*pml&k0HTj&(pzan&%vm
zM5)jhP*JaTHrEr(rt6=|-1YQy*eJu$d_sKGWw1Tv{5xs<9N(@n0Qf(ELdUu8IQijz
zrjCwAl&}%qGIzYaH^?XTt*#Q`OzUE!a{P7tj_;>YLoo8|w1|nzeRkgGR9T2x%Co45
z+`O^E?QNd^+Hn^;L0K%b@qEp~VB1903{vPn)F<~rp}@7@J4AfB+`~$NOpttUNzp4}
ze6mgg5R|UwTaoHTYzoHZj$+_947guD39%h|d0cq;wbSQ)W3xY&H@Oco$)}qu2A&up
zQ<Q<rW_D6CN^GEm56D#Go}_>&bL@#@FGNeo$iYDXiLf<{Yg##StjR$Sz^_)A>pGd}
zZ3g+Si9P%Q1E{RJawR!Gl4SI^J|N#9U}eA!&>SpeH2A%*1|uGPnncPAtbUJR&6?Nc
z<No)(L1K|SmAjx>m#0ccw6)yU-(_6NQ|yUf;+Kz6M-yiF&6R@zUna@MW<#UnlHnfo
z--<ko0LdrLD+_nDx`Ben)?j*G6{7pFh!o3V>Tivb;UCNc5|lAvk+v)#+rZFV-^vuO
zVG+}U_KHCu4(j5r0VH0M*N{_CQ^t176c_Z0!mz7Gw!Z%M9y{~lSj7FHCkN{^EZOEp
zhL$Z$FyRqma-!6xqdCa4jE&o<b#jSI{kauo^y_nrPNZr6Nk{%8hD3#wabozd5RF4e
zXGDoQ4_Pl7d6*>|=lKGLfxFq6DAL-1ThX`!`dItvPJeGRic+3x@JywjNTDUu#46}J
zqTI~Bk~XDonEt+W-I4PyWC0jBL-c*X{{!`%+vUL*4{FH3_!x7e+Lm&{!6E!Ll?guH
zxA0h4Ro@}Cfl(f800bUMn^}2Vt(tK+UyAr(3!SrrvoX4pY~@g&Yxm@Aetj~5uU{HG
zBUCkC!Kz3=Q$}wcfp9Cxr>;k%NDWPl7%yL)l@)e`p`Nw!BvyZmiteL>$me&}T{oj=
z_!1MLGW)C7S-2QMx0+J=_PZi7q;+2Yy;pYxi<z`IJno@8>#olbZEYTQY^49tiHl50
z?8yIIh?S80zm>*G{kN`jFPu;WXCh9FB!}M!^Pe7y4zK`kYR^F(eK4sR!qzo4RV3hr
zs`J=*etTXk*S?+mEIT)lga`D0CVgmLS?!cAK5K+F-LsT27+Wab-OlijOX_ASsvmP)
zXSGm(C6Eq)42+)6mX^^dj-Hse7M2Ch!GIK#rkw9qX!Wpv%?wMJ+}K-4c(YLd^a`}_
zD%)kUmHV$t@f2ujm4jqAt~=0E7ChrRcan%1qZsk-gTOSh9fk$fd#+Fc)M8yKY|*d4
z0nPc(Zr{X{xnf3k|BtyYPv4$j6qCxnDGB!l+)mxsJ&T7no=<8FNzm}|osQ0yn+Q`a
zcMbBD561%JUi<C`oU~4Nel{g+**L#UErJtBlogdrCIZ64yokrwP&jPa(uD=EoZM=?
zNn{uq&h0Pdi=kzAe!lYCpfEP}7mKgOaBX5>2)iZLm|Ba8K0eOwL1nu2v*9s*F^HJS
zn#$AP1>5D%!(X3!_iqPp4<&#4C{|mnKVWq?%$|}K)-J>l*)!$B5nx+5QE_f^5A-9h
zgwFIls>bS(=dWoZd$ejTFfWwpQ%`rL=eBpOwicn4reI(7mDs3Kud{jW<;KYa93)In
zm$u>3-yRKBW!&sWqOcMquSTCtY!3YY+MJ(q{()hNbNi7Ftl#P+NARY}m%gzWOEM6H
z9rWcC7kO-l5VHx0_opl8$AP%4uFb7CEO|$e<VjOk7T)(%;hyu9X<4yeGKeqfyYPIl
zUTGW{rj#k~hkgx#7VGcAf@+BH@1uS3mzeGBsvFCRgo^|=Uf><`Tc$BY$pA=Q!hm75
z%f-ftpWOGueBVKNITa^=Kvt2pASob^T3c9l_o?;7A03(_I7HF0lv?h5xHz(yUN%Lg
zoS+3-8O#_z6{T7bnFY}V1Y8|-2lbM2rimU;*B_t=-+EGF-fDI%eWF6Nvy83ZFna2c
zzIoGGNVhI2Pe!CBtt^xur9~NVcr80HF?IH5o9FdFPi9~pqe7LHxE&4;!ps2EO|Z<E
zT++h>@Uc1orodj41bWmNh3cEco12HW)o+D2hT`Mu??CFGBuaPR+zYCVQb5|6%E214
z^2xkSa;=Gv=)ROvU_0BOpXQxhvNinh>Q24{wQg6pq?Q^>>s)inb*lFpN7rf(10EGh
z%JBiuFc`P(uSk*GwRPE81n;iCi0QEGdkg3upF}8Al$X;NN1+Vn*NF5OK>jYm^QUC%
z`T{<Gp!;al*?=n7GSM!x?vt`Rh-i70Upw8iQ%KQVb^~2~sg@Rmi!@?{k~`vu3;66v
z1^d$nUw}f8Jb}3VKlj~|De)I)ui4@_`Y`W2F*jybT1^Z8q`&0EocdY=;uo5nsjy!@
z&zngN{#>Ntbv9nyIeShB8Q9JCh6%WY$HfDw$qye=$ZQ%K&M#`rH^;vFbWM{;cdg(N
z49(5qlNe{#l8lOdyh!rFYG}#-{5t3Nb|#lTLv*?Q{N>NAY&OqLS47eXyajOfdmAG|
z(+OSP9*1L&6@zw7wENYb=37+yqyk@K8!yB4%m>0XRC_n;xFPq77i{SYm*20BrWtuH
zCN<?FAFG{}T<Bj+X;whfZrDGm5e?ekZ8YzNnU_Pk`)q7LH@q(>d3xE*u;j+hg`2aY
z@)-q7%Et)YqTWhahD%x1>hwOm9B>!ciZA@Y3%oK}eHb3?$SBDWzPP{E?KzUe$xH9!
zNe)K1^A<wKS?SmnnDXHN^!bMyfZ*-g;)sdZay%b~;cRsLAe##x%_KOzcG~tnwU}M-
z$Ck&QjE{T&`vS|v!9Aq57t~8i^csT;6Lq55L<shnl!$`s_l4k9kbhiMTFvCSWBxZo
zA6fq59)OyAKO5Im<#DobNMkZoUF8vNp#Nn}JCUAmO3gkdgt%k6^B##YB}gp+FYi=7
zxPi|?5cG%@YTmT3PP&y)(-@gqzwxc6?VFq3DuQ`bM_D^756%?WuRD5@=hxJn+AeSZ
zYH1!2L4Nj}GP1|s-eh%kd4Esd(_o*!M|L!}rtGj7f2u~J?Ni&&hlx;3Xq}(Xu6MnC
znT=+jnQtZOsr!ZI?QORr#@5R1TRR1?IRXIRuK|z$)dCWsqsq=3OWRs%(d$`i0)lYu
zsz=)1!JG3TY;$$#stjM=yr!ZMM)nsIC1F*VzHr-4dI(PhZA46SdGn>-%@usY$e2i3
zB@24_@7zD8r!3|^?LnflUHLwZ&IS#bTPD@jRn&YIrT}7@&`kH)7AgnFj8~t9Cq$}9
zW&JPU&642k!%zHgi-sLe%~W#azP&4$fBS}IqCQ^B!d*fwNi?&AeTe57to}}o*auOR
zbzAFquh(PxHfPHX6YJ-SD>E*<5`InR!=z8SEa@NxS}DO~QWBxC+DYxW5s^}%Ym^zD
zf>fP}k{C~OKo~Q$_#h?3O;Xr+V~URjcTvud`aZ9JS^%5T1RWfZc`qE>C&;J>FVLu+
z)l_Dzt+|XX$8olKstSdM?uY;Hp*dp^&|Y$OK0U}70ozJ7%@Ip~QJoo5cN|GzWJU%F
z8PaP;mgYIJvsbao*%TF(c%HI7=(Mhh5sqqnADysBoVipbrN7h;6j+@YtUGN_M|0M~
zw$}U0QdXaTy0HJA?_-eo>kmsqrh?pZ79r1#R-O+bdGuKxHz2})E1Z~eyb)CD$~ReC
zH?}qU2*9F-8T1!Ljt`;l1kSf^6#sG6IS~*f&L==)29&c;<|dCeR?-|rp3_?(x$#@-
zOHU8C`XxVK=Z#H}FAa+9U)GfwcvYf0Jlm?bE}ZZ&HIW3oon!tcC}pxj^3sVTBCuZu
zp<Az=+KmslqWFRyjNqK`%XOcxChQdD4yBx2V^EwRp3_9e)-``@LjUM^EzD(ro{^HD
zm7JEuD#TR$21*Jh?jYW>F1WB$bUeFzS-m8xRG@0j>yd<3IDs1(lf$-FEp+(;R8YS=
zm+E<J*h9yE1wQ&0X84pd$|*sPmj0nY7T>z@QL!UA!&?=#OCpp@t6-vez9+Up(1_Zp
zpp+Aat*S_fskw#dggSE<9sTN?uWKeU-X_s8_7wDJp#H6`vK{-x;pK*fz4aMvN+{WI
z>t1n4zMR5;3aCPF-@T_jIXy{uAJ9_TFNy@VK8XW1m7|DGrjcTSYVgLemT|)lXa|9g
z?4y);2`BwVFmMB1s~{WcgXU%rTph=1QlUaXP@rMCh}%r{mVTp649K?U84xSB2=|B2
zS{(P>jTpAu9~eu%WIUlhm>b`%gnqZ_ByRo>5|a~0xPPNPV-@7{B_m#~Y*YTzGVo<E
z8&9^jk5>kpiH&QkE&O3Yn4l@fhnJc)!V9`l&=si%5sJ4E!kep*%WZyXMnTqeVn>Y*
z^#(<aV6+T1@PlWaRP;8?ObDVVpQ-7U(`LE6H{cWr^Lb<Y#`~djuAZPa&CRDEKfNPO
z&tPWimL>14N{8p=vUW%DQYhjzRd10;n6RL`dUe;Gpu*{XH`G*rKpebi_=7zI4SZU(
zv1!|*H|9G!e+Bl|WYyI%0uUILpYKJwKY33e&dPL!!V94^d@VGz2VUz!0Jc9*Q}}m+
zQI|yizITQ~-FH+sPbFoouA;fSN_oUF1-(r4t8Lza$Zqz%bPufZg9>4}9f1n{sEdCZ
z0y}NrvTK#70xBMa5~belZoBeXTUK;dLE!<hFFcVE5a`YijokcZ$b%JD)(3myh)9m-
zE9uYP2H?QjZWBMeqF8E;GzUZn%lDV=9Rm-e6Q$B%IG}GZ=SOs(T;voLiJVlVFC?iR
zy*d>%>ZJW89&D4o8t~<=YH#57VZv5d2gzw$`!)}WAk=T5VORP`lCz(iyPn<F$#3R1
z5rAeXR}L74Y-ON9vospqb#itxINc7inAO(1N2k8?s@?c+F3%lpsc08+iT(>PSJhuo
zVFyjtI-KO<-rk6fAf5xy8SuV(RLL^37=k`CV0X2*Hh&mMbyv2!%9<C7A1u5JffO(J
zkGW?9k9vV3U0l`rGk`i;Y%cY;!~?K%uE8t~Hh|cVqG*s>jh(s}_Sx#x)>DwpzoU9r
z0_5FuvI7g#ZaYd!in<?VvNPVWn9qs`)Fj30!CV*?gT7&UIqu=&rd4K$t&fd-bP?^l
zcK1Abb)V5B_{LlSKwR5nkbJ&!*}<!O00b)qrS4^{!6306XiXpS=u3djWIKC)01+-c
zBdwkbUaAoa{Pvov$n97SKpWe0zu|AUH#w~NoQV|}Wp9W??&w5bB*RhN%2Yh(H)g{@
zJ0AeIv7Q1-{~F6DsZR*!7MvSNnMl|tuE1bw^AqAizp42G7D-h*qZ-3aBe6vF81y(H
z9BA%g5c*sZ<f~p9GyRk9J6TM94BCu$(y2(nn#fSJ)Z+5u=1RWVzFC5Gk|4H<m=1Y?
z=Dw5gM9z4XnrsaGkO8T}h1}$Kig=cVv@T({36n5U%K?@SJs|NlSVWR{;i4Qv8Zfjp
z(4hi+rxU&EaJzv(Wfs_Vhd)S7bF8P@u9nSTj8C)zc68nf9yghQJv`^S^Upz7eDMC1
zAE-S5qpNcumW6WU2xagzNw@u`!f&8IsbltM97NDtJYfk^ErDaOMUp<UC%)31){b0Q
z)K<&A_z*1PdiM=pmG96_j#&EJF2{-|#H3jxd^bA4_|{+-D8AV!mv7CDh8Kkies6uP
zuEM{|>hEyn@JdBN>EA}qQaGhjKJmHD6{UQ(@*`5xULbGR2Q(JRPn?``CgIzn3QIAL
z6PZuWe|tu$t0E1yjm0nd%%9I_8rxXEGLt4UB#Php4EmomFeD^$Li2v_Tjue#Q-#_X
zwNAIanSD*Ls}l)$0<fnt9f0_uW;83Y8epk)?8Tkyd|69dyRp_jKn5%%Ij?cKWk58_
z@xI~%eKWvZ<(RA^0+{weX=T$(1sO{vjE+d<$;XeS0DTet!o%xq7*mkLy3;S^!aEj9
zh?K49F#Ba@*pJw<H#QZ-B%EGiyaTqN(ut7Q*6=Nxm`VQKv+?|L8J9QWTd;2a&%TzV
zkY&kl`pVtsasp4LsyK8A`=G(x)6{H5hIsfEj<5mz5y+J|Brh)qr^bkAIaFjA7{Bze
znt)z^2W}(?VCO%0E8Ls!qPjtzT<?lPIus60G-T<E^Zt))dUEQ7;5|m)SN{5y4=^87
z;@rR$*f_Y&R&`(;jpkuFB8P#xnobWySsP{2B2mm!;$-@mced66#ZZ6JH4bp7jz$Vu
zxi(1?6U5TfGPZ70Pq0qdIA3@*ap00sfxM_F<IZE1$5i|f9JvJ&1?5>m0hXwooE-Uq
z^{2bVl|>IwP*5IN(v#3$@oZp`rJuT4SkPx@XCrB9YQ8lwaT3W<V+DKH`GDZ5F*rCI
z$AjVHvsp@t=9!li&7O8qJtl&;yJvug^TdIf@w8X&jpmDGHvQBW^6O>T{sD*0-CRlV
zM`~8VxtS29BYA2!;I5Ee$aOm04>~uxmh_n}+g6Xbvh`4=M{HsPnIjmkg>@ofQuSSt
zP7VWBQ<uQ+f7|$AFBB86eZbt1eN~KZDL4DCvHquSH7d{60?<JK`oW?Ts!Z|RyeB|=
zTLe>XdARSLux3@jH6hCwu`%FxCP_bbhb(&O-UK_zHfEcw9}LyPH5+c9vDfGt2Lq7t
zmPGIyPL4ur+qinJ{mriMwH3;-da)rW$m=JVHP?O0WqRKIzP>-Zv-`!Mx}a;#mGv_!
z?7aYhV+w#LM*QH-3I6BVJsgh1+m|ik&69)QqU{=*k9Z80_gkg@Y;UkBL!ckX!q__A
zatADd6cVnF!n@Pl9z`IO5)D587FWsnI~o4k+JoO>if>X|d$@Nf9Qo<=Nb>;Umz9M>
z93BB#OySg1Jdj|ozjotYhe?GeBnIiGH;whjAhEIrpgLO)TCU$_#Eq&V@$=u_=3WB&
zt2MDmMA0xi2oR2xv`|>(FWfg+C-&StnSqIn;8+*7AH@$oho_Jg@7nOC#)xX$?unNj
z&O=&KkcbOZA6Gf6`#Y(wsV@p|*x~2}4_kX%f6TlM%EmYlu!TO`S5z<OY8d<qGEZFt
zOH$0P{aaG@{O5tzp>S-U#s@o-$j2VziPx|KpD=!2KCmSP0$j#fX6D!3{W%%sxRfY<
zIdLifX3Z^>2<lki*GARulMmPN!lk!D<UJE@{)CcYul&E353@fA2`0N3tDTN}Ld0J0
z_@Y%{yRl#o&ctY-H{UAS%0L3H4R$&fxEf&R&{SXGVa#8Gxs1sTz;0Y$){KjLQ6l9E
zwV!VISdFc<R$lxP?O~up<%CONt-iAqQ#O1AG7dnm<*60<UjT^)0pwI&{W&FtaLQ<s
zuE~)MpmbO)Ert3;kNKq}JR9vQXLzor`P|CMz`wj(-~#WHl*B2A!6?;!cx_mq1&Zu{
zmx3sPooqQlpkG6MYrWLeCL7hYcO}*EiXvUMhd_|Si*DrZ$)yI&7yo8vmUNU6)oyK#
ztAHH;VR9^vZ{b{P&7YKKvy6M#Tjs(HJYj(T$bY87qVY{tGZ@qu1T-{qTtffdODdvr
z@8ObCw2taeu&;mUtj`r-eyCq$y1@P$1v51C1EbvE!FEf5k@57;CrXZcCuBDdW@W*K
zAi)g_%#V_wB0?#p<!)5E2Ako-^>xM8W0=YQcoel{<{Dh)xQBu5WJ)ZtBdn!*k9%eQ
ziw;9Hpus~dk(<Nju;&H%Ebpr;-vWN`8{08b<7)wp4bayHy*#LV@zuTD$B`SS7dl1W
z`!?-P`xm}uEcX)7+}wYGIySqK6MSm8bxtkU2kcNmduttLE5|1!dakiC(#Sk+X0~$)
z|Gg(zpU-Z1d+r^<tw$d<-pzomNHB$_g@^gz>sM=ACW_B30~*QpA6#|YZ6u#QeGkVQ
z{Xi)B$ey8lqZGP#Us+;YnOlhtZ%SL+MP|lHI`a4uB{AIvxi&fHk55VA1b*bYV2vZO
zxWm~BADmS2=!=Ra3BFdk`}}+*P^$QqP*}+R_pH`)>d-|AjNI}HuAM13!|MB5Tia^^
zo5nNwkp0odS2c{%+^vIEeTkN<$<>>Geh7Cv;&1fFLb(>pB_@~=D5e|_#1-cFWFk%`
zs|%wwHX1kh#&gKV;XX3zs;`9RL`>eRNugRcq&2;~wpiIcip#liT|IM1T$30$+@JHT
z+G=`UWjU*vdD)>RE`vC}<A4>fo^B~@zrJuT-=a2uZ1v(KCJ}$>eUdAQ`0(Qf&v(<l
zR?+(o&}Jy1zPh}={%If@ON-vbwZ2A`O=cgF)M?XEG@+XOwFAFOX5+a`P8NL5k%v~(
zZ^7d{#_+4n^ee{S<f9`PR!(gA`!SRqC$}*^l+qr%5DoncU)H~B2`^2R4l?4h<~;FI
z4;omoFvsqS+kTsS!D4wC=;1UrI2j`&U@r#kdzIxSDBr<1uAVxXa#Sd^Zw<Iu$u*ff
znxc5>x|ZTvhkuDvurHOm6*rPE%`=w<{wbnZe3+)^q<f8ddr?FoCVr~^i(sz8t%m&=
zvwN;Rs`BjwnI6xrqbcE!litZ8;^|zCKGeU5b@7LOBy7WrE{FCUqqpuR-*(O|K<rf4
z%zKtZ5oTUBuT$=Oc@yTz%VlO(9*w~ZZ7R$zJQk3v691&%99{F<h<MabhJTjVNGHhM
z895lZ$$XyblOI{*e$iU`++#fqr}Iarh!>Kx++|OKRU0E~p_WtS&FQgiRqCJgoD$j5
zy~y0JAp$CjX#<feobPYeiO*6^yCM6Ulh{M!(^z|BW|_Yy{yF0;gnZISpDE5Jd5r(I
z!s1P!3w}zjjcS$$mrc>B#iq=3gA|w}6@U5<-Bu$-`Qwe(wJFz&lzqBK(K*EnSr*$o
z^qenV#s>J)SC=YwI3&$elWKzl=U2Zx=KOjph74mg*0H)i;W=Huz06&MigmN$CHg=_
zP#@-j9ca`d7Nxd0Tt=FEhC{o{YE`1KpO{Ai+cLDUMHp3{6`k8s)orIza&Nct08I#k
z)5%^9`uBGw;JKz>H3w-I_-Mw>-?UzPoQyw8xV8C609!~~5zn7?<yQ$X+=#e19ji9J
zi&a+}yZ5<HUG&2Tbw~=`GFdV_{9@f$<z%vR=CrE-m3iu0UQ}i4v?qfKcdKIL23?!;
zR@Rcn+4yIx3%Vx#Ba^?XN50+Zi~brHUmgQpT_F19QDXLJDIzO)!Ap-Y&qkZuHp{#{
zbezLjs<25djYEsi0TrE&iBr|3n{M%=?$9sx242H@{iGL=evSnLEvvzEQS2$3BGkWU
zVUC4-B-FPJlZs%u_-UD4M)DotEgBvyX*B-mceS1(A6Nt!(g{*cZv8!%bdd)P$rx`z
zIu$(^W8JW0wrOf|rc!G_bV^TbvORy*yE$}beYldyQ{FVnT=}j?5A)iUu*Pz|1;8j&
zqdP&n6eP5}{|MZ?L{mIy*>(*frWd5rlqQl|b;4)hEd8YrwvUq|9Z_3(rKi?*Oz&@7
zodeaphHk0!Y5PswVK|jjsvC!N*DKOC^N|@uw^~Ik#oAlZx2`aBmu*W!BgDJ`=|zhB
z`fhx8?l_5u?Ok}C5_{Kp!rGnQ=|Ji3%0TnrLNS4Oyf9Z-qZ0Njs%oLDh$+O~B0=4D
zia~t%>8_Ep)FpRqj+Qak1B~#QWhG1X@9H}Od3-a_@LB_WBY-w_$kBB!-+Yg_b*TQI
z`~7&EbdB0JD`HYgwdv1A61874c~Qf9RjTh#7DhAA$+HkK`Ik!oy%JUJ-0zP|Nlwh&
zCWB<a8DC@gLs(UYEl<2TcO<cbZ1VHIJtT^_W6iCnFCrFM>oS^fn0Yer+z*f7Ngm-^
z-^4P9xz6?f39c-P)43`2`j+T9d7s%Lj1JOmlv$Ioy@jQ^TJ41i;1iLRbV$u9F%wx;
zNZZwC@4}&-5w5-jx}_3<Q9hd5*~5S=HjPTP7QJdlcHY-^%vWN^>OMLghJO!wLs~or
zmv6b1b5{+apgbK4P4+n7R&#I{B&?jfN_E+6gsFGFc+ltlOX2GS9N1b`RZlmw&gs;d
zNS@X!A#abKw-UUhg!MMLvKgm`2Z>bDVb{jrp>hYuLdL%bE+HXpJ-n|UHOR>?W6*ez
zltD?sX|;{r=6#Z{{*+gBZKxie)IT~G6tg`~y^uG!JaiIBHYz9NIr?sGdht)#8sd;R
z*QyLLF&2-@h3a=DO1m=~nr1okI+a-k%l|xo8ul+LD>}6^?lNtWYq^k%8C7a%cRn-m
z7hEyu;!*A1QIC_>wa5MPndp?{Os@RED*EfsRGVPco-Qrl7K#d$RPuWF$qM+0SwR<C
zeUw|9p)O_kDeI7@v&L2b+{Ml%`N?_bOyP^?|JcA5u(_zsBU#q5+UkYDT#SY7(*XiY
z*>3&5`R|!Of0zvoCGbM$=t3Ub_r?mF!V8_7t97cSBhagBb-7#f4g`~pGR6cmQn>e}
zhrV!Vef^rJq4wyX7g4;cbfEriW;dp1wVYqlKi>8XEv$jE1h)5iyKvOFql(SFnq59c
zwXr9=T}jCj(sFdZnOkM|Vdk16WV5qTsyz9$d6<{6iDQxk34Q(Zbt6y4l#Qw?%rZ~h
zyxi%a^AXoQ%;Hh6AG|s1aLjPGH!4TX>l_847AfAs=fAdz3qm`t_P0N7vbhVN^Bqqd
z<NlZ<Q@a|d$;0e*)F*M%^QDT<30xO@y3RMSe(MO`Ru!`zxCyrDS0jxI;nm|u0d3&k
z-ZzEDs+x*E(2azhqcjNcUTSB!6NGqpsuEP-s4DVB<M>=^X{?Wa49S^>ol|^>*O|$Y
zV-O-|XjhX;naHcZx?iuKVe3#g=j9eimG3)&mgEP~K9lw@5yLMS+<KBH`G@fE&%&mY
zx^YC1hN9xL0kjmsr@nR>i^XY!gNj`52Aj5?;IBa#*x&iZb`2A{2L?3WzD=`s@$}F6
z{3RG@{r2`1IwdXlS2%KphN?e==Mt-JIJoyDz6=!SCRtBEF_)l*(pnAeaW%V^58YBR
zUF!6ZKYSpd{TM9>Ckw~N!=q9&lBPLtyt?w*oXA{qg9m`0&`>8$tB((BSuhb}Q(rF#
zt*$0tGZ7LKGe|Le;B3C<HGwUf>mfCMfIcvEnpMz0YjSl>P5jl}bLAY4-f9-v@bPjL
zcu4687g7PGE~a6H4h690s(x_hw74s>*(RoEIP+TvG~b*#`om**1aYor`3%fCt?OCN
z!MUi_(>vTmOv8&44_Xg|$q^Gp1WC4u$Tsmz4jlivo%Ivnaq%qP!sLCAmUnp$C(XNq
zQs*3P_(Z$bHFY*xbA}GAhv5oGC$$-F#0qcu^pBnA1<|>EV0dp~f^xb&E4?|zcj4Ai
za)01>>w-#pxYK7sT@47jApVRR;?=*zclWu{i^wvE8-N3}lc1n^ANy`$?;3Mh!Yi&(
zZ=MC!vBQ5pjWF^%YAXq$8XJ3}6PvF?RjOHay_9HmXWV$j9}}>~QZ<||-qLZ><<%Pf
z6F}0i(-+YS27?31RQG@HMn&1HY7B%-)K;E2bP#$7UUbBeJ73dcM(ZsOR~UC}j5kh<
zDnkD!^>|)8?Wir4hJDY4_hf9rVDyuQ{zR<RgjG{Vs>R^G_B)eHO})9IFT{a?^b5P=
z&Am+C75izzSBK`i!tOF4v9-G8kPe$=pWXULQ4=6KV$GPUWggF;Yah&xBwY5%2#L6+
zG|s!eG}sqVTsyY{#|>6WpXEX3b>F$-aQ}H@XJj;1MJ(%pbPT_ek&PRB?K1a5g5m|h
zg(MvfSINc016GYiQlN0@LhRyf=Nx9k-1?#@^dCt9B=T3_hM8{z#l<}UDQaG*IP{-E
z@#X*gaYAA0?wqyrC+*hdL%Q1)X}+zoNFln;E#6X-Zvp~3f85<8vKB8+O>rJn@;)k3
zbz2LjqUWS&`9)P4(Xp!RW0phJbq_m6NO1Pot5Rh{pT?t#a=h`<bZhQ@wVeT75Y~5H
zWFxjDq;!(d)fsv1#gNm}%L6}YK1D-X%yl!rA6wRmJQzFBi5w2$zja13IBmrd*#JnH
zddExr7jh~!H8uC|-;bEHt>(vOQdLw;YiSYtD-mbSeS^L>0F-2@rA0+W31~k=3zCH$
z9UW0gBO@a_@RA}H+pKLh_y9gGE(KyNP%we^6<Ck2BgDkTndMOmSJ&1UUc695M)s=z
zZO#%4$$~R<BP{34d!7ooXg)=L$+c{dcK>s5@F-A7s&ZVHFU(B)8kf^`m#cI6u#;0x
zx<Jhi?B~r`zXQX%>Eo2M`(peA9XdzeCwlhf>;YZq*@JIi;VhHZ-?h)Y-p$w7j6Fh9
zrEF#L0E&SkXIIAqK!95!B8$&)EzfbipMdtxtGA#6;@Jiw>XB3-PuRo;1doH;z95aF
zw+7;9Xlp0zxYTi5Prh{~MXGArnW{Ji!du`}RktNHEHc5#U$2m^Fapz&VwV<?31jt^
zka1(5b|DkK3$RGY8GDYD#{F7OZY*CVmrA-=yGr@pyF74ZO=enJDfrb<5LG_VGPazq
z%>R2kAQ8+!NB7O}2@v$k9J5Cny(1MyHWp4PMJ3&rz@=hhk_A3?{`kKBKZc2Ppu{wU
zgqTJK2w5G71r!$PH)z|{A4=93=g21)09o#FMGB;<zM)L1DyL0lY8sj?AbniB5+Uo<
z$2GA51b{Vw+!k2bqe$SaU~nJ(!AXRYGEmblTziFNi8;a<9T%4eq?0A<T=&(1)-uqg
z9&PY+1@Em1<h{T1Qy__mHe)u~LGd~C)id7nY)DsO76}(%jJaI*W{oCtb8;Ajh4l{)
z54C}+^_@FzjA-Y6#|;K_!#Q$^mhyMfzQ3>0_4xILHYy@QdFNtuR5d#%XRC)5J_@w%
zVa<Mbb)p{niY%#MAi>X?P;fSK5^c-*(S5Z|nm?i;RE6%`xpT6S8MXx#6qX?53-4oM
zI-egHGV9k*Zhc0wd<|m9&2j2&l=~$Z11^hUDx!dNAV@U!mK{m`VkJ@m+%^Nm9ih`v
z70ZRj25>p2wO%6dbxrWd9=F$L&a2TH1wcGmy#r0LqYRxD32t6-sC;h`9!f4!;(oj)
z9Wn2Hrq>3wwb4^iy#gsXFl%P*%0cp*gS(;gXwRf&UcdfEHRrY*0#?R&0J5TS42vFv
zbOO8SSAN&MqNb*o;E8|?y4B8P+16mX9~C_t+fa8L(iLhkC=Q$W2$(dVZSH6%klLf9
z3FcT>F17ea{H*?fBx0110|XwZ^3@8}vqLwbvBFp?dI5pDb_e7G13W~!3ul*8Vbmyz
zqSAw|PYH676=$fy38UcjR|W<d@PwuO*uKO@PXd6(-E(Omu4RyK91#&gB@I?!QJF5%
z72yIi>f%`pt3e5@@j@m)Ki_gN?E}BlM%s2sojmwZl;IQg4irT{b`B08%?CY`wj9k1
zciEpWM#Cg{H}l;F%!CTiIOBBM7-SC{M_x2&ygHJV9^S29==Vf#8hDe$s|2+4JdXQ#
zPIoE>7`VCBL8u1W*#_RH(ZFzXfD*N5tR<4~mPuR^1(|Q^#m(jZ>9iFr+TAl(KKUdT
ze&v69f+SK7-TDY!9YC#LLqo@badUZJoG5q9d2AOayf75&3qVmU<V}_jArsP$I63Sf
zFVt`F01IAmL_V268#sWCy8T8*7jI3?xV`#gd4yTJ)Xil$ffX%vaOXfLJw5$mnfx}7
zOvoco)DNBgfjD@=FJHbS2z%H1Tpda^++1v50Q<6roK=yys6<2zBAkv^q~%hD^(ZBS
z2cNpmrUOyp)%EplaG#N2B96-UOJdHO!xD3~&K10DV8I%|^}(^AIy~KaO$rFpk2Ub2
zxAz7sO<7qvDml3rn1;@`2flC+LKmyQj`#6vfM>lp+lPVi*>1i|K3WST0>Z<g<}Wcw
zB1%qCAsx#~F#QIn{f09SJ4|fsaV(zMw4@|b?}O&M2^{93^X}_U3p4JNYI=FqXH`g@
zpd5(w#IjC+N5gykI8#I4T>{~yUcU4N#)XTEdlCj+i6bN-d*eCA(nL`Tm+#~)^L7y<
zMP&dr@%cM$6D4{G+k#XmC@AV<1F=Zdx8jgd&|gPjlkz?TB8`+aNTVzk78WuJ3bCzA
zi>NZb<lQ|zh@qg81`xdpPyok!@`SGS1s(xGPNH3-*4eBhBA9TR_mc*V+&Q=lY*{fN
zBa$vz(0%w96_RCljX|dMWT|1kaZ5`}eQc0?Cwdc{2{LHZjR3Z$I6i@THTVJLE|s)g
zBDb2Vs%pOR#qV9l%^i0U-qT0ETawL~NT!q3cBxc{(F{#tsAy>U79%+kStGd$vHb0u
zP6Q-4P4FLcwg?vx!TE0RS<nblsn`&Y+k;j*$2UVZWMR3I?8paph{yUG9Q?p$_J>l$
zXbGez06v%}EwODufX@rB@;J4WRtDpS|FCKE0TAD?nc>DKASeL0vYh=<23)&l+Qt<r
zSXHJsqHe&qM`Qu9&L|yJl$UC0V5Vcis^S78u)6>^`19w_{PywDQKju1ztB-9WkePu
ztX^3zfwPQ(1_|?IdqM#yKtkFTW+7{RiMg}jG+gtrajh!LXtTZqYUwulCiEuA>*mjT
zL>6%H+DWLsY8jcCW96n`F+NrhOmFZ!>jrrO|E2pcF#JOxyip%&Bd3`?p$Te6IuOrQ
zNF4`~lRZyIO7sL8&VL1Yp^+Sj2jY;AN71W%W$t)iQnyzdgacWFdF|~T{b9dMB@JBX
zeGH5;-bYA<Ny6URI6^1i!29%&j=&5=<l@==Dcz7&1wat<0R;HJ#?>A|Qf0Rw43dO=
z;A#=U5B<TDG{%x2A9;*@$$R?x#v-U?Pv%?~HV4zi!KkYOp;0cc^Y>hK^RruSNKB>g
zUJX%zQIz$*{5`fDCSnUBHGyM|u6Bb*m3pyuc2pD|U$T7L&!2fJx$?D#?IdG4awLV6
zcMcx^^^}Ncx|pfy7mKxoY=j8j4<X<C{B8$nKt2Tg{cg7aE(os%dVKr)l+4V1TfE>4
zFGIjYAt9|!*Er_tdmT!&wY4E)Ie2j;5I)C%Lj`WB^=nwz1h7UEkP0YSC0Qf@1rYET
zKZTN!u}4QnK1HzR{>Yz3pzONeC@n89ynpTpk>8wqK|txi7rI}H5NOOxE-x?VbXa)}
z^gxS#+7J;D0R_~-1TL$Seu4eh5o-gfFgD}%QKw%1^t@-ldlAs)>wUdHlhK2CEtw3&
zLsH}6;pLeRq)b;?#f~W<{vT=e@6%J)-`-r#JMe&@1-yKLWbl)88o30zXV9~kITqcU
zPFtPfl+%sgbp)*XMT?7XgH#X-Zoap`8G(z|gMpmy0Zx2tG+(7`2CG%k&Fx1^YwK8n
z8snTp4>Q-(M~FK;lP?F6%qO$7RPio0={KNuK9u>YMu!M^QgZ03%(iZqo6B}q^ZE1V
zmJ=n(WL^iN)HD!HOx<n_vrawjtc<K|B-QS8m8?q6o1xWMgObwH7pHr3I(G2u??6zy
zKam=!vdYG?i~)OeIvvws<mcB$Tx%pJghoCo52(ba>YP;-oY{dqF2(1<qRL`K68wG(
zh>PlbY*Hr+dtZD6xOG1GU<fgn%+SzKfdL$948S7EIX`#X_h{vreeVaXKr)1Qb#4!g
z2Z-Z{Kz&vV)NI|3@vjB1Pe#`J6sI~6FFWc&3LFg5*RrXb%d>^Fv^0i9*TY|J#9Wq9
zN%H1p+mVrx2z}SHc~1rLKx{SOgvG?fWUQe*jB1cXgB6>7xmO39gyOJBxZkw~V0rwW
zGOu#koi;*SLS*BkX`+70b~6>`VZV2#G7$-^22B3~<Z<|q9w~(f>Sh)=+$QAYQ_IG@
z>-_e-t=)Q|QE#T&ZUQ7!MZh2CgY{v$)-^CNVN*Sq13raU=L1q<gjO&JBRN%86Wu@o
za@8RpDlILY{E@*V0Uk<_OrdPM-&MJ`TNnXXHxa%^Wd#TAx0_YuTTPTiWN}%IZ!9+Z
zMW;CSC-D*UJH7jIbXbj$9tL(FLwDq}4c0BQ(q76o2q9bJ#fjjI7sjaSPB6fuaOhSW
zqOw!lU%(N$Dt*P~!2*<KZ>}$6Av)6_Fegi`fw6S116BVhVXC^jyRVw%=l$PvXeb`s
zLGr~#q~ZIkh-lbFn>q?y;DFAv7#e0~xyQ8fQB(j>cuDSao{mi}%oN*09oCTv93(Y0
zwMCOJazqx86}AA0<Z3^k^Oow=u1jBvkPdKoF&)sdl8;X#x6Sm5o^lHiKuhu5qjwBV
z7Z2Q8xV@<c6Q2bxv;QbVOG|6I)X=Y_;kUN5c~B=lqq^?NatZxd8I<b2Mg&Ba>4b&n
z>mWUr*zdM;1(TDS(!(GRJq1ZnOs%(=*hi3Fv25;0U-VxK0cc{QgB%*=n$XR}hZxo0
zffRxIJN#r=)-)l<=VFs~-&y4PB=XG&({B#vtx-!)M1ko$@B%<jGKU}qPKc1X;<TNe
zBo*<g|1%^-9uI=<da`-TCyX2_4tDloLGX(4%b+gf13^tCVhZF~xYe}O)c8GRz$%p2
z>FMZR2)OR;Jas)-Y=$tQL3kI4iR3B~?%T!HG5lt?H`nu*;9ENWd+T5(=lRGyHqwLU
z8~K03T9;mcJIExjgVix_k0X5kHsAQ`Zyoo`wq2J*dwctN`|RE2WC2$XyM-D@^Z2l0
zelmm~S!Cq>NC@;?w)z0PpsXB$jzz4xb8&Lw)RzniXaFaw0uls#Lc-jrs3_%iP$wh^
zxRwA33k?Hi$2pzNo#BTALT5=?72EAGt{Ih{do{Vek#PK{Y}zZ^FgS~2%F(1Xx6sL;
zctn<IPmJ2T&@FYnOIP5lYQdPwHZarBNP=83oXTddw(93DA0OX*rb9<CQ6?xkW1)ES
z@ggl#v%I2i!M{cy``6gm_1ssZwHv)^fJct{_AO1rz*`M0QWcPrx%O_26#`*q9ZQUa
zl$1Ta{nHP3$rQma!1}Hi0d+E>mTq18*1DvqtDEFHv;6;Ry7q9W(l9=4PwlpqvLfB|
z6w*?3*;WzVrkz9$#h4NrQV1>4V5+m#nqn3i4Qi!aD$I_a8cfq9+sgK&nj%DQUEC5T
zG);)v-=}B)o!dFz`QG<;o0;=s8ru1TUq1JBJn`H&(Eh|P!S+aZS~I0RHg&%J9tv3R
zx@F55YeCuTXP(?9(|6s`k_i(gv^8JT&p)UgEw0u|U}K_GS{W+4yJT%h<E^lc*!Xxi
ze}8|uuX+B8L=s7UFdegnv?6aHGnC}D<0C`wnA8xe2sbE<Yd$)#tprdc%vrFB--#L0
zpw^Ee+>oetDooxQCb#@-|H#7CO3{5YOoqYRfw)jmrG;Eo^Nt61AO)I@cUxh&3zl{3
zRo|-2+n)~6>2!I+#BaZK6gGx%V;hu96Fdt2U3%B@Ik~<M5?<a4^_9Oh&-(x5#-P&j
z0>>~eKE4J5SzJ2Ob4aK(@Vi6-sBeVUxMj@ZyW}lL1HaxZQWS|q!?Gn;18VDz?OF~9
zqB%E=?aJ2k%Qz=B{eTeX?&7pGLsc8bN^id!;PrtS6clt`6NQ2fTN$<1sHa}a5LRQF
z?+-G4iDb^)xqqQn*AHt{?krXcIs;&5UG`7RWGjDaKKlxm<40#OcvhEG8WRWXPHxIv
z(sHF^%qiNzW_Ol!hua96P$?@_ZA|;YoQmE7s1Hlcr=A8G&uXTmXm=~LQ$<CE&;<Dc
zliPAk2COBPpgySBhjE<splY039y!oY0af>TQKMf5+Y&I9{SYNL<Jr1b?k>XtX>r+v
zZbki7j#g1txcDV?!T0Xno4qD1$K)Mc@(UE6xYhmuxWiL%L=^48;SdsU9Z|Jp1if#J
z2o%(EmOcS?eI$N=MSwcLV8mk<nHpZ1jp4o7iS|`b*J<~3tp+#_t4d27<x1QRx&b&P
zwlpDE`Se$yjaGMEdEv#|CU(EhijKYwjeh5#7-WMn*9aB@!8gz(DO`oF@h|7n-izG5
z`n(bZg69a72;(7TRjET{1YTJ_K3Q=8M>doP!XSPGY)2;2A{0$S@MkrE9KbV$yj^E5
z`wiVYbx&BVrXLg{!31OrYwH)ob=%WdXVyy(h_Tw45mlItaHz#g=kz9>)7g!Bf4aFw
z^t4n&4z{J1q$<E^g99BrifNR{K!8yT_AK-%k3*-##r1n-{2Y7%3T&6%3DPU`i8G*W
zt-BSPZN#IxmlHkbkh}<buFukYue5%Ku5M<A9WU^q(6l+Yw?yOR8CtGy+8$V$6=RS2
z%qpr>UF$&<490!?8S<x;Im1LPG&Zi09$?xHwkF#^ugV7(F>y!PlixH&bzJ;@+BA)_
z5`Y-|vd}@ePu8G*N{veI7weoZ<58`hv3h1RfPDxjRK5au+Z=~vS2jEMVX~R~K8-#g
z2mhyx*x=V^tGe@c>8YohojEqBBVa|u>}7j@Gs(Mrc|BTAJ(_{=>Sb}TSY+XDZfUu_
zBw8&rsSh%3N#Wzigpk8y==dWt0X`>EbMHBGou?#mJehy;VFD4ym#Z?4_B%XDUn70o
zfVK*euy1TkN=kAIqms>-wInQR?@@Qv>;!%zkMZj2LHxv*tzNPXVGQ9s@TEqijJ^?9
z@oHcsT?`%Ohrr{CV`>W9=<S|)HsR0Ek9Qb+2AzJ+rnk}fzF<brcp25;x$5h#7$BKM
z4G!<a&C4Go%-ibZL>zA+GC}kh^MEov35Y6YNvd%niGAP1pW<?A;#8$=$KNN$i;|Ku
zvF<K!?NDlMfu|Tm2b$YLYRU%v3^dH6w4+^+#qz^AD~3@NePl=tuS=~|m(TR>gi^(m
z;(x&kkY!A))8g>(*|XDw>u+wJY_Q}Rc(905Hjy|x^yTd3KED&mdFIS_JxvNtmgb@x
z<QOXk!WnSxge<0-fE-RJADcrC1jt>r|FI#IhiX!vkFT%e#cR-WFBM$s`c>wog0@Gg
z2cKrxx!nvW9OUe7xWAG^rd<}*LFSa0h1!O9o<+~?4_)Yr)hVj)9^*Mwy(t$&AXM>4
zCnK;v*dZVUX3a)wkUL3lQKS}a&|Dis3jU}erf1V=`HgY<79TaGTF-UbFH*REr|0Ja
zs0$TJB?|(@2H$Z*if$3|r_C?O&0Q?eBg}m_c2y)k!ny6HREp$``7ZBThB-IADTL$*
zfs6I=5@*erOr$N*MXZ3W62tnk7-PpZK(vCws=((u&9M%tp6Puw6Dd{zezj+Fxcv_j
zF)oIEBAE@@O=st`H7?|hm-eS!4cI;JYkp3n{o2RQ8kef`dK`yXwgpqBv!1WOtL8c2
hyW3QIB(MfU+Wh-x*u^>HBzT<Wuz5SZVB_AC{{e2N3?2Xg

delta 74322
zcmXtfWmr{R*Dey$B_&9Tba#V*(%o!Yy1Qc$0xI2I($WpmA>G~G-5qD~yx+Mlet^Q>
zYtK36n0Jke4)~x>_;SKVz?B$R^-9R@QDCsG{Dc$%0g0CO<8fm_y=iVdBrmQ6Qd{p`
z<z4lUUhC0ZEk3u<ywrS)G5LWJ3Ho;d^+OBW!AV3>eZ?SiKxX>P#VJ4D_M+p6jtQaA
zjib<xql+y>%tmeBx=ahAIZ?VmO6<LT<e`pN_tB#{U9@rfVrHBNNabn3dqRyM(A~ho
zka;l}VKJH<(154{O~$PF$Ec4Il2X1_9QBZCFW;HrDD{6k1&=Jx(iE$<$`?mQ1(_-#
zo>-k_aKad{!u2x(PnQ#0SRqGqRr%vkC%MuIW{dUqxlg-Jyu;}N$aV`k-_p}(T3?Rl
zYeICJJ*Lcxxl9^??tlN1o+RBJ!!Fbu8d80Ip`f9m-?JJatax3ZI&Kdd++XhTgRgtJ
zKULIkNH>XA1@9&V{(N3-D;iM4p(&2>-7`YB%=9UmNwaoUXYvoL5w1g1z{BeLhhH_#
ze1F!p9hARG{9NigmL0gY9U$Sqy*=Q++oz9ec0%VZcVqxAOa%df-jzu-|D*OPB&nK6
zrJ!pK;+`AZ?c3)er(=Oo^<_OL0lw07n!%CnM)dUU(__?el0%%)Wz?p)%vjM{q+#sn
zbUQyGpGV(T^=Fcp<%cA>W@1(gKHjwW+>2$CrD&$Hz3R+zU$kK*(mvrgcLWmjdDOHC
zB?q{b7bn2KP>*9BVxpI6Z#`GGJuehnp_PMsj-3Ku7r9kqR5ezbQ#>7C%8EoR_}bIK
z&$j7qcgkY?S7J5Zb$g2uUAI_vr0`;HGhUJK4yD&KE1BC1(=US~dd+g?_F1t!i{o#S
z=Om3`e^uM%!d_K)p^24l=x_3VBqkU^wjJ!0G8X|>D{zf-^$dm=BCs4Eh4#>g<41>a
z2&je00_btR;Mr}7(6{=qg@p~Y8lMm5R$DDp&9{j1UBy5*!8_`g8~8mSsXB&>@Ld@`
z=!N9C2*@Um9__n;KNq=^_>&^|+eM6-l~O5S<Kd~#+w6_!eh&^-zxPm4QMq$isI|8B
z7y@Qy>b%_V&bK(wQtxkXW0`=5qdNYxE-aIwO^zAvB0~d%_E;L_L_u#}sDtSe!}HC)
zno5Q&hmRjWB8W6+PMF^v2jtIPjhyu2NI}qj0;XZ|v#VR1^&9^dN6r#m{J7s?dxr1^
zBb&;fZ>t|v)Uw!j5M(WDTrAMX52lely#v<J<nJZ@Mzzc@F_Pd~DeajZh$MFmZ}pes
zJrbgl<5PQsnb0Hs#ZhCKm(e?xKI^vQ)o&_kcqRE0C3y)tzn5?3ObPCh<1pj<Q~6FX
zL(qTYfzz*qQ0%VVj>&wzKnur0Ull2NiiZB-S%)z->|6g}ZZ8g{5##c|@S~?-`8<F>
zRIca^iRB*Qm0cXASmRXgyITgtC18#|i7z0;RG!s+yOQ?dgy4+&Al?h{KxvJPiRQg|
zF+I*<?-v8}ufY};`TJKMxZVPRr~@rg*K6%>nY-Z^i*0O+$J@72G#rwS0ETsu4w~hR
zMnw4hdk%x{z}pzF%YA(ZR^^-zxFZ0&KO-YfLKC+hN=S4se_~f7);;T9bH?7q=c$-N
zr!cqgLq&LP$=L6A70(*V<fQNq`^3x*#WfP%;BcDHSTn0heNc(gJdI^?o6g*f6k`g*
z+MU0_+?c_G6$Ve{Bvpo&Qp;kC_k+{|cxdHil2~&)=IaQ;f_Bn9OCN9EzyLPSTubxw
z$%@%x6%Jccxbzxri7bY@b5-{#eRWE-aiN4f(LyheG34Umr=93R%my8>*$;)G4c+dC
zvn=Nq7l`QSW$aMihRgySSJzbC-{+|&Rx~a<QHqO;`>X6xkM8H;zL}f5Ymx2Ti<Qkx
zR2Ygs-?YB4SWFiD)o|&)fC8XMp(MWus}ReDHD{EUmj)}??v5!dY5T{GBL((Cm#}G>
zmVA5Z7@BRkeaB)>Wb+*NV(29+-0faF<o#oD(+q-UF_OK~j6k_~rc$VCHaSkDm#g$s
zTUS^?rocJ>nYVYo5Lwv#%?mOseP@XROa?{Unl=IK1FO$JlwvE{q6t7-P3^73j6aRa
z*W5?`b2iptN>NihH#P~R7By3KTrYRDekcC(A>?Y0%w+M}OOI)}ha_cy2v+aIZ2`|M
zfdbO17fTZfyt%QHM<r9hviD*+QDgAtCu3&Yw#Hp{$nMtL3F!8DU#smB{L?G<(cL|c
zJz<YlHRsx^anqKyMF1E(&0z|?DyclOzHG}kHBiUZ>Ao>-(^3WC=Uy&wYki*cV{4vl
z*bvw<7sYD{1VaPZ0%4sHO%8w0R47bFjI{qA^ztDPjCq_S+&y_y#_^U8W_T=!JrQ_=
z&)UtHEVEInDx?2<rfT<$NrE>Sis`<ST+9S=;CHjWsxMbF(+OP970o?tQ0^Whop7+V
zFkX=Koeoey|AoEST&JOYTA-Sgku*p2NV^1xSo)F`l$1g{#@`J*TvF~mt#@Yb{PAk8
zudnCt-CAAMKX;XIOy;n}4MrpScC|m{ED%eux7ff2R!)xN&M+(-+}irOzf$2B%9yN~
z0~&hzPn?|b=z!2;Wb4awqgV6pt{J4(O7r=l!m!5%=^mn~d3I3NdOj+OZeVH}tX4!p
zNf~6C?w;rMuHwfJgj(wbmxV+|-6o08pFiK7l(jzdVo}PFl9FaVE?U+=kdTnLJ+JBc
zTn;~HDKI<l5B>bGGn__Fy?ej-#;`kdEzNbAjG7ujS{jhP^bml9gG=Rg{-x)07kJY0
zfX3H&@Kcs6d}FgO=6g_3vUh4!RMh3(MDX>|!bLkg$!M-L-s0k7WS>!NTpT7IUJo);
zQ(Jog{2LlRkHJJ1oK(py&Z)bqL>``*g9Q%j`D!$AuvhBvv~mSKA#)iTZ**mvFQ6hF
zLvy>a8vv)5X|s)YSXFz}J!+5Y1LQasTDlILAC~WBk<^-I22Xnuj&M#591)y9F<i4Q
zn3Vik@*^;~$fwI)wc(?>%}g?A*NJ5HTCqW5y*(JxqmdXWsdS(G3!`K=vxL-QWv*d|
zm-&;yYwR5%4k4r|;FgFARqKSf((N!o$*!&_#0eNgMLhP$Hhavt4D`Bb3Nxa}Q(f}v
zmcX8lFTbtCZsUq2zSV*BWcjfP=n*IxUre$|T0MNj(SWq&w+F$w3jg7Ia7Q+?8VH1`
zIJ{fa>+I1O;B2}|!CpMtV1!$JeeKwWM+?JC2vL#VOH^e?bpJxX5sH^5yMr{(qmv{p
zFr`l|P=5N6aIpQ~NI>a`nvO{JLr8%AxqU$M+^J=BbxYQ@N#}3r-V$u3Og&gv5bAvQ
zT+9*;e!Y>JJ+6KmbOz1N&gvT&_{mUqc630Ci;MF*?>i~m-6WZ&H_+637WKtYWh)iF
zXEQ;$zP|1?V#P*+%gV~yYS(Gd9{vq{4WZRuSy{=@1D<LgZ){ucg1}NPE-8uXGunD;
zsIMO_RHiZ+OcZu-IJbU+ef#!&N?kARrNwHF6|B~uadE15&pbRlW7*>9K6hJqhUjT?
z-H3LVyNbkKr^5XF{QVKFH8rnAK!bkF%gxPfHOrjcLz6h117#~B0&O}1q&M&FbZg$P
zudmy9KF!N(YZH22A2yh~G;2*P!wJ{l>^XbXdXp_l9`hc15`JK3v+%ZrNAJ+nK>NJu
z3S==BuE@VZPz`x>5lCj`2)zsn7n>ee?OjdF)2G|n{k3GICsCFrOh9&_jYOq(6a_tM
z%UM$YNmTW3eH()9Bn7PtF!WS3ZUTu5N?R$8@o^{by+!=p6vzBEng82Eg5{{CZASIJ
zuCv9{oAc<Z_J2yeF5TVz9*yl7@}mcuY<KyzDIS~`2D5l060KeixbfO=ojH2`N+3H9
zw?_|97c;n9wSTfoJu1>RemI;`GFhwg8Ex9Ht*eNV#&31NfACrS0@%Cw*&odKhXil@
zT*0S^<4PLux^-lq#k#LoPSt69&ZtwHt2Q`*3L5r$=j7qWM65<By@l~7x-6=)=tZWz
ze>l0vC-M7u%RWK(ba*X)?W!@ob+P^(&h1BGk9S$L1kIDz*>0nqq8S2V>c}kZ$si^o
zU9{*9#eclrh)}If0{GooL8vR6TgJ!D{km_DkF+m!*M<4%-#~x__O?m4VkqT283Pa^
zX+Wljm7(&&t4v4I={3qT=c&~A1q66}9_!sIbHfua`8@{W83qPeRC0bpC{^9>wi0CV
z!$}cbvj(V!V+b4()mxbfTnTlu1^Hnb#|w^_8Hx17fyLU5@BFiKgW1@}-z)v!$(0B}
zxj48L8w~N<*M1jo9%E`59K&hcd+E;JCd!ElXI!Cxn-}v4P)2>}Biev@$#ZlV@_F0D
zgGAT3z~;E&Dkdr1A3i+V-DWkU<?I}&@5%ZtrooQ7WnX;mgsgGLIu@7shK0ZJUTwq6
z!165(3y`3Ilkm-0%l+DEpq<FgsWjAi0uZa(WVPz^HPBUJZq2H}ZkpU*6Ja?V^1CWj
z$ZTKkj8eSN?cmGVzaQsGV9jC0NS7*{ww1kwbn45ie3<wFarI>!dC!@pWzF;WAZKn+
zT&D6jm$Lhzy_7zjKID_vxPsAn|ART(?+>g=p#ZAwOOVSN#VwT%uXvhHn>0-td6+J(
zZ6(!*XQ{w%4B^YAubaxBOmKZSMLHXi?~^h7>*odaY&s&m<672RrK6qh+Hq7efk-7<
z*@9X@9Ho8*Y6kb+i+n`<`9hN(5s(>S9J2-n2Br#nCwB&;R{<|R!>K%?-@bjkySvi>
zG&M2bzn3sGqZSQ9G1wi=;^tG&p#B)s=XO*xgMf~{(?bFnE&X-Q$<N2a#qHjR5E=&C
zy>CE(XwB1?FJB^)ll#GI;ST)w**aU2pFe+6$4P<59sv=NjGsSsx!Du^D>ZYW2!ZSO
z==`TpMQSDAz*jq5sEbHVCDhi|1`v>u4Nq6wL*KKk=bfW;GaILZGz?GAD=I(T<JZ-}
zOx3r4NFT|``vwNcNlASfb^bDYT$;T_C)TzUw#Wke->R9N9Wyc>V^iVr4^o7iV9SUA
zcxR$u8MTXGB>%tRAW>M5{{?lTLu+tj?gSxnvQJ>uwWRHU!PTSmdHBf-pr3Z1)bl>r
zeD@sUvsd^mZd%<~$z4YkgbTBvTAmh@XDj~p^1^pkWW6q<o(Z~%H!S=xWN*CXhGlKG
zK~q?ZiaT~7>mT1POV<87BnJ4J9`ezYl8hJ0j^l~b&aZpoj4?KL*k+|dl}s^XCSSPj
z<%q%qOs`%vX<BhRmMl1<fCUcB<8qd;`Xj1+_O>EbOYXXd>T0HD6#6+OOeTtIhs;|Q
z!325A^%=(U{_U*J#`U?y;MX@;f7VC@gzpPi3Hvs2tG$cyymd)}(zLE#Fq;%RN1@X7
zZqWF)i#Iltn|Bj*{EXFeEXRe(jO{GTOfk6jmZP89#KuWydkKKy|7a3kRaI3sh*F%6
z7n3AoD3fYV%FCIzM>0kp?=CDCgX7|G0^re2z<r_&``6y?cec^P=JV(lFmrZu!{zo&
zr`%(1Z9TL*mW?m)*W{<B$Pp_|o9BAh&jrSu(NP5_-aY%y7%F)+Udbr%lA4~z<>KNB
zIQ|9FGh09^tOeFr-Hs&Y{tWGy?h4PbZ<7VlIs0)jkSA$6s_q>Db)S@ssLt<bV!u)t
z<3I`!TS{mqrBljPj!h%F7)KE}7Ir|D&4n*bK}rfuqh$_u7^84gJ{fDk&rR`n-X}AI
zPqKHa$*g~F7?FpHe{o@{{Zs0b%LU8uNf?<B2M`?H^zD#8)&AO<oVtmF0lu!Mq+%_p
zo@Q%C1eL9!%-jH7Dunk%W~ez;QqzFuJGaCZxeVquLjJjASc8hSFJ4=bF?shBXJF38
z&d()>&@OM&UiqP{>P%QZc~k92ezW;T@I_tQa-+n<f{jnsy1-8SH3APF>*56SolC|?
zU`L7a>l~a0-`vLTmPm{Z0|N0Z+DbVk$1>PC|J#}q)`l;gC7TZJ?!GtEa}VHgwo#2F
zlO;6X`3B*Pjw@k9$|Zu^bKia+uK?dak`{~)U3sh)sU~J1mM-asr<~{w*c@I=Tjy>`
zB?U543cho8E)K-M>3N_gU)zU$2OOJli*xY12xf(yAzI6HU5d+)z$my`vkEGg@gd{a
zLzLo`Ao*NU{Zx3D%$+iu96aKz_&!I_127!~C2v)ekUnUAFcQOqDNVNYr4pb-%j)U+
z!`AknAcFzBre?x4W_Py!)zOs(p+i8%q;#(J%41Kw-czR$d(fU>fW#Ky%0f<C&gt7T
zt|=Fa3-7TyA-`~PtfPK;WI5n4UCzuHvoOq%*;}7~D9G!{^#TT2-cK~v$6WFXu!hXH
zrppM5zb+Cff=C*hr|-<d!os2slD{2#m8<K=umNmiDuW}13mIuuNk#%y78^m-L2lkP
zIfv7Fh%RZ>;N>u@Dxm&Q++woF0ajMmqi5{x5tYQxmQTiaeMYm5eCtH(@)tiis|gpg
zIwXf_zv;CI(Pjq&DMdXwO(R;|ZoG=AI#4_ob{u3OT%xL5F2@|le(J4NMu}_51NAOL
zQ%4n5H%)7QEv-6Gx_JeRiWGTg-^Vp;r7BBYt_1&Fv5&5v1;B^I(<iHITkb>zx1Bx0
zO!U;(g*rA!Cm$_++lxd;K>Y0_Ybi25V&9h+@0u^2|4^jPf0b8Ob@wgAbualN5|4)a
z$_>e&b>SLsJKmpkrY_k>Usm;D8*h=<sYvV;Sd+Ftb(k#+p)?pbC;kzAy$)c7nSF1!
zDZw>6w*ihu8U)62x{lk1UaseV2T2K5>Fk%OPO%LlUnRKqPC2^rlf3nqZpz+Nl3?m=
z{)vZlupc1dk(S=sdZK^xVx7aCe|nD`J}wk>SID!Ozl~9M1iN5TMh4z;y!>PN^Mky(
z&8Vf%>EoYt&5cLfzKajaZ^Fdstt^*3N`^F}Pyy$eWr#6>L?cBGd^gE&+kJZJjdrm@
z$mCJwC}Df#C3;E~ceYIkbcWbb&ZsPoe9ZZYWD$(IvHhijUHLjo;YuK#Eg76uH$Aci
zzoc@7`k|<#s|f_$)hl}&qrZZyY;UnEwL;Tr{-m&D5fI1dLrhI%MIFNM{auN<vs;<&
z3ZPp;Fx3<RQHYEmsAW>Kx9H3|Na{x_{1Ws>;;WblE4c2jyg`g8nf3H&Y(|JFER#m{
zV&w*K)E`ruq)><1Rv(EH%ci<<plnEB<7r+a1>fV#;|V56%xznUaQ<6v42v*?QzU<d
z{NyX^&h(QPGJ}_I5Y$9}852by+vnZ&GvyWDSa)hC)^TQAdP%uL?-)^Ulu3bW!J3ew
z78}&w{2w_@14PUGXPL+F7^TeY2);|lU<Gjn-DcbYCp%V>S`&pPx)4AJA_v|IK|?-y
z^H$3y-V1i%BhJPn%HK{}0q5*?%y<oP{*(eXW!``_AC=lkAOk)JI&ix0XRRKw<b5t!
zsi3inJL)@lQCKd3B{q3VlOq#{%^%yhd;6u{9o{?#e8uv;4IzC~T>dS1HI4_F=4Zh_
zK7xV2i=m%W*)ycY-{s6V2MFXIl*a{N7r;_acU17t#>y>cS-pw|$nYEQ&)czjOK5>I
zE$U6!z+jB#Gv+UViKO?;7YCXzc;9xmedWzIlCKnzqr=k-w+1d~`Uewim{BhLj~Q2O
z>AWfTqDsqPc*;5Bq{WBN2MR{s^Ik~Khwhr3x*&}^XtJEWNC-M(V&k6b!|Yu!&OYDs
z&`wJ1>}EJQgd3Od$HBcS6fnyKVS!CSGf}TiW0L%69;gvz>c2<4p$2u`p0E$WV%aPf
zL`gD1F)>(vety+9OUc7&d>e<_LL?bju<cD=3+#Cy8jqGwTYvP+g>@k^bs{l0eIci@
zAP?7NguSuULI!~>ELii_Von>k{V)1+IBpehHFM@HG-7@QipV{%sKU=zTc83>r+_v;
zAv%CsvX+*IVWi&PMxPl|b;(vqHW*6fa$x_Q$Q)4-qT3^)=q_gP(9_9N@5Xm86tnh5
z-mX+QXm@--;ZN>S;<Cn^S}NRPwY+7j_hPR=JO!ND{8Xk)8=#4y_a#7I?~#v_&F866
z<z$A$WU5F=VrCHvy}YblJ$ohkb9N6^H4C6t8x(h1z%bl$%zi}FKi5mGk@1$?5~I3B
z1v-UIVP0#(#BGK)<<BoE&GSw5oF2WXWg%M*xD}*uP-XO}mlC2nUz?~Rd~rnov(8F<
zcYjCA=FS%u5((YZsu%AcJ<Wl@`5X=FAwbk~)lBYejT+@}|4};q9!a(k;$dO%yA^Os
zwfQ8POn3|mhh13HIlpz>eqPzqLo-wNY9*@=+R6{cr&F34pxbi7l_3dAgSz*C>^)B`
za@9NTYB%^l-*C{q?{f;q=>EGa6(W)Z+8RN?IB8MS_r!m{o9%V8<oY|WjCAq9i1lNh
zP)9UX>$jKj1p5oy2*=J}RoHZZQgnjUW<&&xQf7Cg_<7)c(M=J$y+^0;q@_|0{d_57
zP+9SGyZD*H@7&H9%=%kZoeB4Y9YCsmiL#ctG1maUbLepO1=3mlo{{i}@d92E{^@jq
zR7xX<&F4l%0iM5Jc!T`N8FlsP45mK>=E@my=(H%BDN#$RlNObPR|-1;O*UgPuPD;6
zv8B_$h1xWMw(s8jNW7M4M)X}md<713wj2$TNvE37UW?lOA{>*`D6^>`)i><o8kt%C
z9EQPS4U71aCe})g2$>U=hVN8ooR&HvPfI7p3ew-O(lfqjHF>0kb6w!>=m7-_y-q>{
zT}!to_A!8@o3AUw1{gstu5Qq4{OtcwM*)^yVj9#5JDUykKUDK_K4edH@xP%bwq3t!
z@;{aRz`@#)Y+s`CUPD~jzma5&fxZqaUVC!mY^i!bVj`#t(q)hnZ&A_IFU;XDTfU8d
zo0=^*%*+K#Dem}fikP7GI=?e<x%=?<{{b8Lq$l?x;CQKVqS@1x*J;;oVx^?G*leMe
z!Fr+g-~9ddU^qxEK>iSvyLHtz_(!$q^yY?&lXH4e6dN`x-F%TFkr*pDlaWDxLLJj6
zRY-r-<Ieo-k*|^KACEeMdpVnd7=J14q&ymGs=M2?+g`<{{BXyxOs5+CXd9nBhO)2g
z4<SroLQG|jU*H)jIQ}q}bz0EEMw<trr|4j|kLPxQbMNe#7RAl+6RN*A87;|jNDOfR
zXV-Pn8*%;FL(`(PTY4iist)2kMq?KF5Wf(LAZw}?guI^5SwW>EXr$EqUFBEDWbzD2
zl(N*$+;xwXe!|a^!8|)}cG-+IYPfKYu}8K6K@=}bb~C3_7ec~54y%7MaZU`{v}EPt
zbbr+_o!}%ENu&q!vR4bgyQbhLXdW0)G)G{wx_I9tuooKO1Pv1gS4DhyOhC;@i<8P%
z5l1hU^m}Ra4HOy9UXd+kM}C%DxOfr*uVsn3${<)*+sQe&bj$B*CB_$2bfq&#C7&08
z>zGf6f~bPiO53Gm?Ms#fkj{tk92>qCh_tzG1!e&D|1PlOPhfv^Jznescf{T`!2-(V
zWH~J{@x7F^G>^;S`<`&3?E3n|`g(4Un`2bpsw%dJo0I(F;?UjQ-SgFQNJ*Z2hG2n;
zNg|IX#&?gBk%LY%zDjrs_rgbjtq2eP#Vswm=@*PC=$6S@-Tr75w2;@eAQWCB<G<Wf
zUOyZ2`~q_G_qW$(3J1ARJA+l@<Qq7O2dgViZe&Q|J8NUrWZA6%zI4i;D)pC#XImO-
z)^2emIs5@>#BA@;gUd&*mlN6)$UL?CVX(TL?K-F+=>G5@+8dvxa{ww}1;c|i5qqmW
z-nQOwx5J=N3w(?@YZEs#Cms{~!^OK7tHPU^8SU#(O5pPsAtswl`RhS&<EZUD`IPMr
z@#f}6cVb|f)E4LUR$GXolCO@k<m#a|X441wl~nkW^m6gPV}>1(e0;kWP6RW-$4qEJ
zAH%RW(;Rh^56eF0@@fESACGS4K6)*@J@4gWQqgjC20>LMZu%KrM(ilHp9^#KoBek<
z@IG@8V`VLMc;c5mRl;WN97q$^Z+%x>GdgZ@57JVMWqufB#*gY6%wE@3?E4r-kMAvo
zC3P{unvUEjmp|oVtoIzUV>+!`8;`2e=El}DuR3?U6oK0!K>^0b-@J8GNY)$7T5+A!
z&-@>n?non@NNMP(Rx1KsR7OG`hwr62jpr-C3t58tW1v$01STe?u#wRxP)8&b@W2if
z@dLNm`Cz*1-@iBICPM^4=p;772GfbUW%-6Z;g2W4i{WHJKgmGJ$+EnL#*AnI*X1q^
z4P(2MIS?qqxJ^{7>>&U?tp39xM}f2|Y0TPO4qd5Fmd1jof?BGmO+Wtlx@*>iMS0}F
z-u2_>`=)1RRyG-^Y^9>nf?WMXrNZLND(QN*byu;9wUU*bzF;V5t6W4sZH8Q56{$j%
zfeVXT<05!oR;O+5*lSMJ>mnLuw#E&kNnVaeU@pt#ieGsCm|R?8UW4sum5>Ie8p#-r
z!piM{i)I6}O$^PNsC@uJAX-Z>j2&ybJnzYUK5DuVHe4`uaCw;A=qG^uESx_k72zDr
z?b{j_TbNqr*Y)iOBWyEq;CS?MnUGAKDg5FMmt-YVUWM5u9=5Sg^Y-E9=rke$b*{cP
zP>b{%dDVdi3fBSoe(1eqLjJ?lT0s$Rjg3%>pst+x^Hj<w*`pj1Q|Us&&bXoz@)|TF
z)_I}n5aAPkwXH5wQrY5b)tSQ8-@aQ%mbf{vG`sxA+>BxTZpI5#Q@R0PlkSSs{zhIr
z3`+pz>Jlf}Z>4DXIHck7k#kZAeM2(ffzDPB+?85C4#L$Tb_nVpRvgY+&R&c9iTr62
z{iEs(gG_iC4g>8E6RVA1c*73{R&R(~kG2-+G`(g#gJOR+QWtOU?0!S!2i}$8`ZS%+
zcyf@O2;}8d!#2y^+s2lZSQABv6<AVG{G+C4s8IQ#WN=f79*|yiezm^-?N>-{K0riB
zNT@)2(4$NaRb?`?KAI(3tlRvzG#IoD7+5^nEkP#*tUp*^5}(_weyE>;fpEyU^rYO}
zi3J7J&$k=s-Qh$bxw#Y<J0nydKSKL%4J6E@K$Vdq)JRZfPsCt?+|939=u!GoNL)gw
zF5=Sh!N$Vm$Vl3*b@(!o8&(~7As8-Tob(*J|9;+oEzfhg;38MLB(C<2K)e19(*WGU
zwjUUOTe)Xfq2$gBkqe~zb39@>P(jNe(|SW<Zj4_^POsSIK??o@aT=drPB!A<lRehX
zO?T#8f}<AS4W({1{#cltyqHSLPi@Skz|iF~fiJV<OS4t0B*RvMVtqiL7k3h%%*W>{
z4naHq3@QT~qn~B0kXH9sd<)@Q*4vo0<}Ny0u}$$LoFi?s6d><rt8maq2qJ0>pU@7!
zqa8?930lwA*|Ftw5`H?d!cArPTNoBs5}gMlx_7pYQJDJU4I^=AOyw$H<0QWwf|7o1
z1q07d{7p(zY2){Q^ml;c-dR1Wm{UD=c%S8tzySN*=f)>O_nOn}qk{D64@*_^DNL>q
zce}q;m>k`7<Nx4Vq(9l)$^<e|=`P6OVT~w~e>hc&_MrT^o+@Ic^QUxx{y!u}y6A55
z(%tB^H+~LUW2+Vm@NhbS_NxOQO2mijE0@L__u@U0E?@!>W|1OXr8Ng*!K}E4zOfCX
zWH@F8nG{VQP1k%Mnz$`qlh&R3!*02qZF!~YxZyg6hYX>}b)ZuvV~$c{R770ZyN9=O
z>?nxLya>q|4jk_{wVd<HgdeR|&C*q*Cq`B0;m>B#V-yl%qrC1v2AaDUNDq`dx^|UR
zB+S`BxF3MOY);!TD0g=Bg<}c5(nERdbqF6!JyX4x%rFjJ^<?u*LK#Oc^9R_TXk<n<
zCc6WSbS1tj|L&d}Noy-*FrTb2*Fhhno|db<Oyf;$Q4WfViWHGkdizxqj{2y?-fUH*
z6XV5#&BZp(boeH!jN=n-Oh&XaC>uc~g=XrK!CnDKS;d)OJV=Z;vs5k1Z|pkA3&`Y9
zb!17@8E9UbPT9_88?^jtzUF2K%?KS7(|IlFv*uLf#Ybf%lS4QCuUTJ3&cQT+7TpY=
zy2{E^7=8;^simo@dP{12yh(foqQ9MMlxFX##aggBa_e8>Ukd2;Z{5}j^gT|i9u=Jd
zPqs%ax0;A#89^iS?5<Wrwuv4EROALYR8bhS_XG+D3x8*3J$c9W3K9DIbLMN1q~X-!
zyG>PRKCt2w%YIlNtIq1viQCp|-aZ;q=-=M|6r6kab&9<yZM;3GTc=2#x<I}^A}tL+
zJgpg8)dZin<hb@jMlM{<sb&HbB?Jo=c$FgxA_j?9YpR=cU8#kHNT~Avm_9Mm{(Uxn
z?sm4_1bxx;c0}-_-Uw*T6wR5sTTJi`#NC%WJ2E#j2^ScHOP#vloqQ;b4{9FS)phM-
z1A)GbLuXd=qXh(mE;FU>Y%LLoi=huK`7`?xyguom8=)8(A0n3>$Lf{>aOOzf#B0n=
zXUqq4RXR3<s&0y1=V$cIKNT*OC8ve)=Q0c0Z0Y3cXPQ+1;}|3$$5^Pd)q8%tkcXNs
z(dto_7f>~R2>ZabTz$+zK`~=0V>|XcFiD&S{Qbo;Py@^fgx#ouvvn-pwa7TmyP-g+
z`NXpi&DW~i)6fjCC@n1_fjctpxDNAOaPeQJBwborbHs2`;#gtpa<K3gK9s6GK0L70
z-L6-sn5wK{@ByUpky&ZGPy3?W__HTz>t06}>lzjkC{X{GCD3MpomX9|r<J709NSpe
zI@^hmCTB|{Pa_`2BahLVJQ&COU*X)aRlx&pR5FPQ#H-%VcRPEdSuiD#nm^^%US9n2
zLQhV2Crr(kN_Xdh$awt;ot>RS0y5xstqE>Tb)fTc!T<V=oLkJkLon=?mg_$&Ldc(X
zW@R$lrQ&Z#wO7qs>r)#Y&}yCVylO!sTR$~OoQk866G)e9$DDBJxK|7qVv5IqWd$sj
zrKjYY@J1VW9HimqI~x@8V2n`@erGWGSF<*0adZqf-=i$kDgG?>G8cq1Z_p<TU=}w!
zp_iSeOc1U`lEcdsNFC_oc;Ie6!w_925nWIeu;grkU`*0|CzmESu;oNKEpQk2%q3_P
zUMr#VY5+I;RbfKk_7y|_MKHYxgwuTI=LqBF=<FZv@@V<r!ibbGd-~RakQIf1{fBG{
zXO>zKA|j#zm?i_eSG-sl0a;9+gp|~`xk_VtNOgf%Gib=)Umd7FKY%W(Y74*Dt(_yf
zGFaI4@~`5&M<#hN?uoBZw1Gb`#CuPUe>uAE7yhl1HEVec_`^{QGX3vM`MLLn8&CS>
zrvHRn@}Ed)NpQ6{`}@AUW*hS3<$6l&1Ee16zz)d&8<iy`yhE>$xiZMvJ3;47=G3U%
z6#Mt8sC<pPbFXSzR3}HjokzO#&xZy~)RS%fM%8r6v!~bJ92iZ<U8SCG1?O_Kq^)!g
zX#^u-c0z!xpCxB$v1SIcv`{5MPQ~RvOF{(v1R#<#ZLNE!`dd3yQtBg$Ki=#}yefx-
zl8sv@e~JCaHerYh4#*I0XsYBznPU1h8eNWL+vh2y<C*PNKSQd`gh5m0>EWgo<bqYE
z@|;$)zJdR))~c-+;^XMlE5V=*$oKS*4Uqf+@LX5Ca#eFf3b?#)NTDaIPH(JaI{Vcm
zZ}8w#0AGLo-?AhxWN_DB)~hcfIjg#=_^Dapr?PZE?DCn(BZPDxTP>(2k$@wzz3uLC
zI;93E!=s^~m&%G8yWbmF-hZr;?KzXC%AQE!Tr+H8Pq@ms>pjDoq0`M_GwPr`;Q-H;
zmtd@-8v>zjNM?C$&W<#?#1>79AUbWDJ<+Zo38P4QvYIF#=(^;O=QZnjYCvu|5S0c(
zNe-`NxT8a=%J>W-T)2F^U)vaXG{V8c!D|EaL=0NB<<19c^I)SeZv?%R=cgMzm!tW>
zu&^Q7GE_qD$S87gR)@`AAO*VhwdnL|J<+K?<_UZhsC_ix|3sjeeXbo<9j+$-f=^y1
z&q;xB)dlLbggPzAsq|}A`etUUhF5C1%mNs^yfJZ!{o!TPYIBR=-La(i9_1gzjcd3b
z>YIw-ATNAdBzk(JO}&ap*+kWI*w=zz1FMK6G+MKxId{zBZgl{SM(^!TKrVL94uN1d
zWYR#!)=T7rs5aT46f=Ks5q!L!A243w?40@XLJTG=Kxc_pU=9z7{nUou=no3XW|VfK
zBQc0;-~|an7RL|KpI<vQ_+_GG8Ks;gBi8Ek#O<(&09uV*-QAU*H#Q(~97<v{?Hw4{
zxINoI0SMR;k}#_{kzfCe=anP5WP6H)6X=En$?lCCiNuZVuVpy72@sr5I=TJkIwHa2
zSP%3ND>DVD!uWX(R3T>x3%${4>*w+~N2`A}4vfYsb4?=$=o3wP?>{AU<S3u&)nb1y
z4u!ze#jJ&~fuh;SrU5LvnbsEPuoaKkIZ%j<VFWLKk6$MQWu2y9kC&{((YRw%-10<3
zTBpRk|BzsJ=@1n@H~GF$TU51D<#bLHEC3J|zIwo8jrWoxp!xiZatYG*@7*3DQE%Zr
zVcg#8%>c;f3!Fje0NYQ1E?xfrWcLD+<K7ZCczF0^E?c>a?V*#dZQ!+RvOHMv(Y@Hw
z;N?wShL)flhdVQ9FxD82pBl6zg(GEfz!2aOzFh;huV75@nB#inU!qo06;h6C)<Twq
z%nF7}haycXxp?M5<fTfwFC{{sZ|Tt0=2a-`^fwm~t&;D8+3jb0&Xh=NBT`hZ+lQq3
zW969wvfanYU-pavjb4WLmuv5NL-%x!?Bt5)$r<?fEZw$`ONej$P|Z{C51_zQ7fOTb
zl4&-t8}G&8ew_x3J#n8Rp8hv%CuhIoQVcUHSt!jMzZ@kR>P9xKekafuVYz>hWJf_|
zHsixZ;)!^<r2Vj>(*btr)n;xp8>X~uHMDUar311OnSNk?DIHOhb~E0A_;Hg)B-piN
zYU6+U5NTrRaqAKiSQjEhAb>#tuE+>nE&IT{mISYh+#glVpJGJ*KN6p*K{x7Tl^U+y
z&J=Y2k=JZE(1UcffH|CX*VJOxHe@9z`(AkY(PH@zwLn91_3S{Tf{OV=9+)q2-A5n*
zU?0@@P@I4M>hiA8rwc|K7CRB+5r?aG7I9_{Mh;#_I8+p3RU+a~!*6{=1?_tyFe4Ou
zIN;~agHs!))xXExA_Tf#?bH933y#2y4*%ob1p5hIH+UYS@e@AA1s8ld^#!pZE8)#&
z+w5BM|AFI4RLs>^XW-Z~4B&-jd5++c2y1@T_H)GaJ_^YqRdRFMwS(%MiMOcv=px=1
zO|mTF0qzufhHCZ-PA6Z6+1@_>1Za<D;TdlQHv~TctthX3X!`~ENMhH(jzRuSB%W(x
zx!dm?S?*sxEE0I)^M|J9pn#(DcWzYy$y723Pg^@7>qzc&3KSC{(LOUQf0MnmtNZC#
zsB^AYd6UvuDOSg~CYu3NwAuHj{WF(K!q5(vzfh9b16s*9)}32eJFn(dDCjboY?Ie1
zt*)+pw{85KEdG6MytmVfkNW>UCGY<}B`j1D3P=v!Y9cMX;^QQ-#AxEgrkQcR^Qt5!
zF#!zqkmTtonSD|3Tx<~NsJR5mPEuAyu1(~z2wD@Q)IJpr)(UZ5Dkesmdg#@(jd%Y7
zeTdOw+Z4!BxQvaT-UI35Jp9zY%j-D;vr&Bc)6h+jlK9?U2~L~EUNB21At@<bR%Rgv
z-#1rfYH2C)+GrHY%%rBUjW&NX0*VGF$3McV1Xz(TOc5-lRC6+9DQ~238C#Cs*e`??
z(JdlD8AIM&7%>pvy_cTe?$n8u8_K;*nYt`pJ#P4Nxb#{v;mQI>EuJgv-zW741?nLT
zdffke7>^vamWrO9o{2_RTO}nWi)_!=hpe}Z+r)J+nqyqAO(O;FCIeG+tDGyS3S2Ea
z#GJ@8rOPE36mb&HPv8f*Ke+!|PhGIIo^RlJw<r|qcOvLB1M4}lQEKP(&jzgTPbX5-
zvSke%*T2D?o{vE6`FHZ{-L%+#VUadJ$!cC%m8`T!suc!a=?(*b8wCjoxFwMbjq_aN
zDX$cLLSLWq6@RZ08(iBSy1TnqTh4_4{(TT*S#7(L4k~>iJ=(8%vxj)y{2h>p&I;I)
z8wci&ljtzCFA&7yMwaB=Dq$*G-eA?Z2+97f+w_a*nWomjzdO=9)I3G!4O6=hC?R~b
zYSTr#BdFJwQdAQZf0-TnIrXs1=gh5>L8RZ^3~-c<H+|zAdJHJIg3s|NhiaK#wSnU$
z%?>T@Qn_Z3*=j3W01YTx@M`q$xPKm)ytE|gIup8Si_+d*gIt8mZ3d|6?L?gMJ=7bX
ztbH*MA>RDcKKI?xgBuU3_4-@%m;cHFc5V6FHPlIy`$d01qiiy}!O3#7n>)T>QV@tg
zK$LC?M!}|KAClO#!u<kKg4{Ypbw_8_{y2NcX1%@MZW|v+UHQ=2v6jK?u>1v@my63v
zl$S`l8xUP@VH-a=O4C;nH9K;<fK|01A{+@8D$NwoX`3E-**yPw5t_=^ZuMZrsQI&y
z9xz~>!y9i`e3Sh_M9<Z(>TiRA6o79fCtgPx2O?#vy;~k)EoMYd^!&g0ZbatueAr2W
zC{M>$Zl##BGp9w%E9Ow_5d}*g<x^lkG$Y3B9usnD2T@wR{17j%GQv)Xz4@k}cxIhD
z{44c%{c;}cSR^XdsQB&WKJ`cSWBsd5vN!{p=LdOusK=ZN9go2DsRx#sgR~u9j{f3j
z^jiv)Aa})R^{(@s7IFs8zD(;9w>wXo`l4#zx|9x6hRl+Qp>`c8Y2@8a6ulb%LfOw`
z8=yY@DwbUwRwrQ_Z&z&=mVwz!Q`1`x6fw>wfqLZClJEIXs|lPKR`L#$aPa~U4Eyic
z_~!w<3P;CGVCcv3_>ezEg5`8ypVw?gKz0H$nJ8b1t1rjsR+3NeCT{sh{7RK>*UYKT
z2)sAD&yS>$Nv*>*|9!kGbr<0`A)L0xlRX)joCfbu#Upa(CPv0Q_dhh<9J~uplLbL4
zl7An_F2f=snDlIx{?;@!{ud2z?Dx~FwKKT@rdE#^F=)!Qhyqb`k$ml2K>ZN}<mk?l
zQ+_4c10R1bxIohzGZ(>qZ&Yn1!e<`E60|_WWeqQz7d_lI3_4Uhn)uK3I9IOB^}Z`_
zFqkViRy#G8MzANnl6m*yRbG7E0~xvWIU+M3Z>T9OXPc;xo9NfiV_&ld-JgT6KoW9Y
zx2}^LKjE($FfbX5&$5<Ge7pDV3*PG7U~qYqfQ5DHe|tH}L}5%Tc<+rb+a0!Z$Fovt
zGNLO-rAx}j6mdaQc*sLEls}+8YCe8|;&Gw#dzfnaDiSWM8~o%~S62*r@dI1i$GWc_
z@|fTsPvBvCog(c37^~d?4uZq40*8k;b^F#w*N@`rCiZ6U|IwP?<wBY5=`XtQa!b#Z
zdE7oct%PaS#dn>AezXUQRruT>nNBFE#>Q_>xL~NH<x1zoEe3PvKbQ$@?rwGbD9cD-
zRTUC~2Q9SErq09I9hPH}`FT%<YaqeRo$j$H0Mke7V1w_%=>R6pdf7>1)N=!Pb_`^Z
zKp`bG`Q;t1LltoKpLd=`EY;teGZVY#Rc(ZyYfRf6Sgn6Vj_%+-?uaetnRM8P#lz}@
z!q(Y{xw_d~YuK}Q-fx6!Te8k~BT(RZbF87M85JL|HY18DV{~oX5FvPPcIO~0?Dzkx
zpc@$5YRo~Ldh7l^zQL0Y|Ma(zM(^y+#P&f<-qQN$xP;1AV+?)d0FYG9iZ`_>ODd`f
zH}|eDfr)YRHBF(y&F>W;9!D(HJy_1*OtLp}4L0!>!RNH(cL;=|uSY0L3kqn+Y@iq~
zf*;1~RNq;>GV4{t<ZYr0SoSQ|jbl}y1{hvViBnWd=(De<gOVCv_~Ufc?TR^pL-?nM
zlC!DdX^rHDxGh0N0P4zz{|aGIb=<*WkHejy2Wxn2swX98Bz!BM?CCw>d*8+qR9sg4
z+IS=p{>8-U1_8y9Fk=%pE_MjRh}Rv5(BqS>&PfMR_L!MM5g1QLX-63p1p4NsnVP$H
zG61x;?V%KW9*3L;F0$9&;Tl+PO_dWEEK?wUVhLjq9!of|F+TPkiAtNV!6T#k)HmNg
zsg$d-Iud#;gvrrO^HV#_!j#t_C2c<~>E0|);k1qVOtLHcD2z{&gvcfh2tOR8tAL}z
z84%?CtFeaxZNtRNS%3~?60`leAKnbdD-`?D<wlYjcS88)$4$?fc?HKBb>Upgz&^3s
zw~}A+G{8al>`nD|g{ku-A(#H-tyCeuWV{rTh90UB{1uYH?w~!lM~WTMe-y-SX4I^P
z|9*_>ld%zNlO}1aAUB3FhkSxo-<Hh5V!DF7t61@JuO)DSJKEohDIfw_%2Gu3-y2MC
zKiJvn`RjZD<+c`x*W266d_(zqQD!Yq#Qu>Kbbh1FkY$fr33_RL_`YBeLG9MT<lCR>
z(XqMeBW)cahVF97B@5~CG1$?A+;PmjE_~}vjC~0(I<{x8KFHb9_|sMcOIQ1an0%XG
zd}28ZpsvQnP=<P_T>BzuRB(!RmJjliroS6(yx~X6OaYBE)Uolead9TwyF|u~hmG}}
zH<A=~pYX1CK6_zKyKq+f)FaD33x=@>wNm;wz7~ibiDr{C--F>W#&lbNUJzOVXT4|k
zo3N+5JUwoLl6dtUhR^w2Io)xjljtV$#(<=xWR`NioY#3DRnyb;0=vcJ4-h9|VPg7$
z5r|5Q=ZV#$BkN<b*EQ`RQjd(HV{p?<0gZ-JB`NQ`Ew$UqJQW7(Y`^j9BR4IWOxnWL
ziW|NH(x-hNkCb{LyQkUQo%OKRy{Dx>BL`#=;zOC<fV8nc>>%A#idFR;Uj_(3(i4lu
z=`8RiyW!TId9_6G)o}ef7l>`L`CHYz;>M|?y%RC1pyoe||AweUuBWF}8tc$!p_Pjh
zX+u<ZBq;E7q3}IghP=_zvUlk2*wSKWalBfQF?i#F>dX((U^GQFH8FjC(tx+9-yUzz
zs?ElJk2{0v&JY-G1m>#Ev1n*$z+^i}$0c;#w{Wy|%2qA~y>6M88eN%~MpidA0zmQN
zMBZ|ybjuH(nEFMhZ06!-1*p<*hXz3sIGXWllJUPhUi=qgb(WpOtAWUZ?$~SC6`Z=z
zZg7Be1@mYhKfY#a%_j0>!@~ET+79370>BH9vhNxjAAjCYSN6WCMR$=5^R-96*^e%Y
z`K*~jHkuR5iawbeCqth*6VUN{dkeq6jyy~?e=2D@;%EQO1bvU(e)gl+;?AZT5~x{&
zDaNvcti)iMptKYi7aK7-`v0(ahro`VHoqi(lGBB?#7+<gT@)ouo|DbU#=$dSFm){8
z%HLKrWdV*dnS!{8>*;B-Q2BKp1gwLYxHxC!a@-t^`D*itL^ji5Ss)@joRXI}8Po@3
zn>}yF!BK?s8Jo~0^BPDs<Vf~O=XKVA+l~|zq`9t|jR$aRt>#3)k(k*UOT5}zE}<HY
z81E!rXIe09+FbaQ3rHtuRGVRj;4-XNjR=CHDfo3nTv@nM?EfJcD3HgDi^StYhs-IF
zTp!Efv{+?D20-AyCa368PqeoX-nM>mV?v`8?|iUlet{+&hMQvWTO{0*P8TsZpF9Pp
zE)v2&m4kibk!R1-$NLdEUf_lSoGl4cK0sH9RyR@sfNPuZ*LR@w>W#k#jRAI0Q+~ZB
z!Sg=vpps1OW~N{mbH7J2ia^)U)YQsn4h{hxE;S~^GibfT!U~Ye70I7agg~TZWKh6U
zRC3yP+u*pvYC22^PDk`TKRtk@C7eGIm5|UAA@ux1D)%FpitgwTf#P*J{GnUMgbv(N
zQc-2k91H+KXhe|Zn*pX)8L)J8b#)23?VxHPVPe^9T|cqdZ59*3K_X_8!S_{%(i`1j
znib#R3JVJhruNUhHR^1LTwGk5?zU69dU{eWB3IbJHs)z*^uG&Vymr`S_V286stRl%
zK{?l2#v<D|9?K0SD(!N72=JO6RcZ_$R?>HsMJkpF`e+?^0LD~Ou}|yK`DZ^T8!_ZT
zL3zRiWPzE#T>!1>;=p-R`%nG$GYG;->tZLK&z3K}_TB`rGAzNA#ar?!>yL8Nl8|x{
z@?lZ_#d=*yVT}D|Y4bg(Y6n-gLK06pX#yVJ$jBIJVM#<ZPQbM^b!x3M_>D%ne!J`z
za;y{u1EhKtT!tp=G;pCP<<mn`Qa*@>6Si*$o$>PWo<ClUWKUTnaoc|crwuj&@pP$y
zCy04qzZ88=Zf=F$TE}_C-Ti%h>vW#%hj2n3|NMNagPAfaGBUr+OwyQZV{n4A?rCUj
zEGn49ClNe>PYK{K@EI6jSF1mM8a{lp98ScS+$jC};Pnr^MzB{lvXk>q<y`P@Jtm7u
zYUGkJkP(n<bqr!Cm$V!lj(ld*;806)>6fG_l=^~*OGl^R9<AGT(cr)_+xTuLYIt1j
z@775m)H3%bb^E+v5u8g)YkV{r*kvt5^OO|_z+5^7ZeOo{L!^L6Fa`whn7JC;o1MAH
z+46T4xHF9vh$_T+dj{Vc?i@=fI=wOa)3((A1I8ubRs@V=@^!s<i2V-=``gL)Ge+7R
zHqL9LEG86%acq88^LZvhJ=IM~prxh#l`h~(P(2K0*~a#qn$EgEaKX9SUiOkl0L=F5
zU!Ly|%2J@4rsi6_`Jlk4{;VvR1ku*fale{UzXSj84ul78;P<lg34fgSClP%?P5dX6
zE!>x%Z(o<GfWm<MuOzZGQ$Su2$M%a{uoVZG-H3Zj<Jo*p+G|wKfkl}X?caEms&7Eu
zG=p;Q`I(K6=5}F`_^o>f&@;65pkTHKh4_`31?hT=v?_A!=uTJ_O;T06yUfr;x_v2V
z8ie0oOGsjF(tgW9)1cl2=amoFVQX#R3;%Bpj-L320<lqEeXey-L@dQXCwIQ1Rj+i;
z9_CgSQuoOJUmpQQW_opaPpC?1DcR!#2#HO&Ywh674@~lMpt{;<JVz1*RAg0))S_Zz
zzssrL5#rCV8k-I#`hnw3fkHF>B`ho~VqsLYv`D_-oSXW5o?fdD6c~J%-1KhX>^7gs
z8v|?Y@aSmZrhRxA&3ATIBl~*IhwOg_oq{wtO}K$I4XZS;+7r_4YF7d4(*XN<rI4J~
zdV3w%@dJWMNB450QP4Sk<^9|1lC1(>qh)ZH&cR-VKDHOn))mebWnM(h=~oi~8OL@s
z_F&ovXzQZBaOav>=w>Fk%n38ir+IX6a!sv43KY$=oXY)xi8Xz>sV8+KLjuCSe{(%y
zpsI@B+VUT3^0!kzyNT{kN*nHFEd(;4<9>y>-__49tc)FkqjcY}u(0sCYzVb#tqeAM
zqd+`R(DZEO;c;($kdr61Qsm4_WKr|CBf}D_V2L{_nXIxh$8hgh_YSsshEj51aFE?y
zPw??n6jEdHPhHRZtBMNt*RNkGPdcC0!(rW|rKCXLWF5SBQ-y$MIV&ZMSN(H8T{j}{
z3x#Y?BO^|{cKm(^!MX@b6A^OmIMMO=<2r4MC;L?tE^lL7PB&?*_jw?aCInQi6lnO`
ztCHm7hDNLehLae2GX35OnxUbIsdy4>!;ybbQq$;fC1EV9ECj_JQ6T~<qOy^9peekz
zFc*b*!3ajKSJ#-*g#jDm>t8CbxCzu51=upvR7|W4{mt$38Ga@x95|uujt&!cvL;7F
zpoicw<>lmj-{^^8vt8!*4+yveJCXTp`I@yo2Ob_Clv=5_`@|P>oG0xjGwt!dn$@G4
z&$eG3939`-IW{(5SFk+1BOw`qC-L$H2jW01K_MW}?DZSqXx5F8G7jo1Ix(ohU7b3(
zDu&ejij2gtvbHv#EWiXqs30_1o;+T#Z8ZjOM{vLrCQM9O()RU{?QVW9dH7e?53E|F
zN+Hv<J?PSZic@KchQM1Yuhp1IFa_W$aUKoxSkmd_k}@Q00MAs`h7MRXnN}!)D0%E}
z8bV=!R;yh0vS|HE-H~CQgPbkNuv{=x%dw2!6!)rm{1a6j@m;Vm6ZdF+X2!N$E771y
zX(<17f;4AwyUp*F4#oSP*OBl!Eg4@(?7vnIyH+~EQqL2{FE9&G1bQdLSWX7ZmF*M-
zC>8pNdtNc`oQ9fMTgz%_ya%ntcRdjRiRtlTgINv4WH{Abxe>%uFy;0eVpS1Y&X(qA
zkzf)~b>0MwmqArqYu)h^2OUa-w<8p2vvG6=p@Nnq446HVp?un*<l|eITx!UllJc#q
zu8uuDwG#^`lyq`(in9(02}uISa$nOW;7wY?Lik_L!pi3&jKxg{tYKGzBs9ywVXB50
z<_(l$DZl195MS4twAKA&KIZlUgp+fuLeag5pixUC?}6)MKK?(fo~yU6CQ6jeQQt90
zzkiR{M^ODWPWF-ZtI9MEj>?zvV{}*@&$)lTy?;o5;e5fwgnfhahQ|Mrm^AF2hy+Em
z+{2T7jD$=^(kF~zV8{L3j_qREaU=~tdNAoG{l<=8GvUwk=Hq2qF!L@7bHBKGuzt82
z<o)_rNgvFLg$Q+?Pj{DJ$2XWn%dsx)xlXt_Y3t>aD7=0jo)xyY;kz<(Iv^4JKdQb0
zs;ch^mk>}|x)Fme=|)jOLAo2HyE_h@0xDgCw9?&O(%lWx-3@PF{r%rt&$U>}MeaRk
zpV@n6zWKfxb&R}^J}BX|NVAeZi-gh63|pk>4qVX%>>qGte%fDdx|K+wb{84BlZBZ5
zutgqHkwmCvE=Pa3f3O(8;P|C*WQ@OqMoh|@hd25l_3ZQJ!oW10n*Q%ER9<W&r=}nf
zd7uMopCH(nV92ViJ>vEN@k;bL7AaSrfTOs$c;+>b6RB2NQ^BQ+MHC#{<nqOqkeOTL
zdI}%!)$Wh;-i777My<o8k<y=l3N*crGIl465zA|~-bB}{EEby9gDbMBXD60y+YOOZ
zdvfBad^P7ejgeKRI88#<uXJ_PK#|@nl}w05D*r@2`}6$<wR2V2QDRMs?UdqdIIbcE
zgQeJJnn9(rT8#bXUGdDeb<$NoPJ<eRCG%Q?v-7gH$mZ7B_E_DXbb%6bcqEP%;BT~)
z$$lWt;ckuTsNF1@xwIA-a5x%M`G{h4%;$kWY=zoRjfW3@%nox#;hJoIoG*%h>M?g+
zZW$sx&}&g!G-Q0S*hM3g%XsNKAz;;b%f+CS_vGcv;v0IW^r81a=?a2FB%QXP$M|P9
zPEIhBhqK=v4^zyAl)fe=s(V7RvX8T+Qw+fmc0l`*mWHLJB{+BBayNd_23=oX9yTFj
z!3}0hQ-)ItYt6G!fdCvdHAoxT3<?TL)$0sV?jJkX0If<8oAOjkd4NLh@{>6+F>!A^
zd!Wnl+SsK{-DA~yx;YT6!-7RPL=;SY5IyCLnL->SlF<ySnzsm(!1;p!zsylwq?EGS
z@PA~2hg5IV8LGZXeDM4#H#rvL`P-y0AUJ=KLO7_~KeGCZY}(1xX^}q{W_PbjM5Bl4
z+-udBT<lc&-FliI-siLSn>Uw}t&5oY+ZftaB3ZfFeq(<bwXr?=Y@%-=^yU7?HK-z%
z2B(oBB#J%wPnIEI^jRM|rYoxOPoHBoi<+&W8&bjS(`IfYXDrY466s3#mSw#2yFpu}
z*@gLuHJ_`iLalUhbCb_)po`vw*2kuAMk#69?_)b8uIPt`hU!j+CC26{$;dPe=5|07
zL&L-r=Ct4ud%ag%IDHFi^#P$S2%ikTc`GV<w3cWE`%3!@kK}3AR=A7q4QEOOhlLp|
zG<bl=l`q!|)ND$HS{8@Z=tQhu-iRnAAdR7)5rpNL#O#I&r2C>qr^IzB(4EWWnGX`$
zujFd=j;YJS=o;zo0w%lm;9i2Lzoy7u{j7GwM-mGYSA|A*JRxU#Y(u2xbPKAMT3eg*
zUqmjOyknF-QW%&dJ>QrGr94H)I_aw9T6xE=Rm5ev@5L>d5E)9G9G~bbumlKj$JE|k
zV>>n6KJiT<f7AJ6E~AR!@18!eaWYzWJKh{CMnqwz8M!UR|2SV2e_Cu*ofI1}Z~VNr
z%I_H+WYT+k45t~(ynd?bR$N+cFtrmez(=3fWqbc8QhKsT7ZC{es99OD!$f^+Yx#kI
z)ClbSx9{F9{ZVi!@fXkhba!*v*4D;qHqK~Svq{ZCg2Bwp47%lng6t8dHHD=$x{bi0
zo$XAuYC6sN2L_@b`@$zn1$LcGbPqzHRwL+k1|yAv13MUC4&Vh4WWi*gAP`sjQDL-K
zoufcsVL8{G_Ac}x2FAO5rIr2t0nm)27swD@ZVo&1FWz$tE2T86k82P|(R&gra>~zG
znBR$0G39R&zX5yqtzK8PyZ_`uM5N;xTaJP=N<~KN7gmkit%*CRX}<mS)bGM;1KlzD
zupdAicVw9#ir8N5-;H2VLE!GF?n>qs`zu0kUBgjojL|+u2_ZJ*i3AF!=|PF_E3wQ}
z&A#F?gNPB%_u6?YN1G4WSTfUDhE}$=!?ioPc*?EIw`@L&;+oie_@mLEL0btYhGjIX
zUfEQd&&|=C=`HdQ0`@&otEBba96Bc_C{XRJQ(QQ<P~cb)eW%QScP3PSx&I6ccbYl%
z@LQe_05wiFhEBJP+{!V@`LKzJKj`c0C!cRl6lz~aUyRA-j(sNO4!2q97~?JAUH_z~
z*9zJ;#0z=2MxSeIIuqcl);GC<$8}`k=63D6tFED8z6zPcY?6<aWU6LS46|_dgRZ0J
z>b16}w{I=3;67#VpwXU>T7eY2^_k<lOj$i~dW7^xqXc+c%=ie4(<NIHJsRq+Uy|+L
z(G^tW*6^rTuHcQilELAGrq(8wXqx+-b``xK@5fh|@0t?bq;j`ncxZiLmQseZlUzQ@
zYtlDR33)%K5bQ}wxf7r5r?vNwzUMl`LVJGm9o|008qJjS*htbyyIbdcSevKxlGVi|
z+DOu^>@;he++%~=;JF$%+IMBx0TD#{WxPL}7k6KGn!K`9cR`42m;cpH=5skVb{iY(
zwt=2VZ(U8|nMO+8UBpB_O)ADurwE0>oVx)35{?Kx;E1?ptlX3qB<l;94~?g##ld^d
zI^$rr9rfw$Gu+y~R)yQ8Rz9eZZ;)p2Ug5<vTs`hw4kJ^kh=x4R(DU6MOb{vSnZ22z
zXx8AtMl?El($;?S{M-P3u)r#te&2GUlARE%lVOFC?qILy!IL8Gb^3kGor!os2?+AO
zjknHb*%|ObI<(HTe;uHtwy~2}W=+dRi?ouXalwR9k<m=vz{&&{zV<5kYtw7Vk(oR`
zV4gQ8!)#I(d^xi^g#lW7gTq=T-d|-AWUDp!Ji{eA2M*V>6N&lnAFBmc7EZuSke|?E
zF(QKN(;#t_Y7G^$IF_Li#-9x;NGPSZ)MQ7k?kl3UmENDw^T(d&UdTLVe^+&_Z*D3W
zB>-d9Tlb9_C1aaI_ZHH;=j~Oo<F48(lj_Cwb>)?dT%#LIBAr`q=J@z{furBg;yzL*
zB`2#kxL1!wfOlEGvNv9p;;O&mq)z`p_gTs^ozeE}JO%+dQtswv`6<*+h<dq9`E~a4
z?2Vsv&gHq%dJu)Ux07VaCDQzcKu~vkc_vDJO-1e%&oj3ovqRCzF9<8@2@_dAn&28U
zyv5&TFD)rznmF0flof>I?G6MC;OHpg#CfR;EP0QM^9M9aEA(?WW#L_LJ@ORyzCb9F
zMRuT&bvg9Q`sP9Biyr~qM4Vjoh?@m}$k$at%lz%eiEB+y*iiF3I{l}+1N2<)Y4+f0
zxR3*1c!NqO)_|wTNlbcX=-ws6fU2TLSY(Kf0nWR|tYXjpSWvtF^at(y{R}c}?ZBaf
zq@=BH0r1Av8TCrbe1{vLq}ON=1nG9}+9$U7pwFM*7Eb6vtVFCtd4njj;1^1*J*_wr
zwdk1k=2y`|t>n}mDU7iwrTg4{KaeH*eAcedaX8Z0_;g7|Bdl}7W}ivCaXgv0J8$f2
zI8o=&1<o@0C5a-Z1;vfX430n!Lk3=}(}z=G%%+mJzsGGxl06QLw>4L>!;&#>DMR5R
zKk$a<lW6TgaE11MXNq`E@JSrj(hL{R(fN4j+J{znoy!ihwv!zgLb%Gl-8^kbzVeov
z-Q7f%XkBzn>2BAQ>3=zjH2=uk1fJ+b)HCUlc>P>)VI)7Pt1TclhM|2<N$ui*I&L`Y
z>fN@Q$&&W2uBDN6h3-CKW3k(2AwhC3Y=!)=(=MT6S8pdktTQeb7p~7v@l4p+F<*V9
zSY|v~YJ>%t-f&H=t-hegi7IETX0796eEjywV!eD4WmrJgm&1LuJ9-{XeXFP566w8K
zDsHBq?vstUB$fI5x-ZIKheL^fcD(uf<5C*hg*tr1V1Ebc;RkgPrqucnTL#k{&8*xC
zbnv)zA4D2xJ-AB|3=Uwq3V#-@v7IqvqF*Zaq0!h=3~8$JC?eI(K1OyM+a+S!(eZrm
zLaB9c=}mB2Se~W5a2GxrCo$NEKd%t-vs?~-<#sXt)o0{qxhZ!lsv8B!o#S_1hm`jU
zh7TLBC-h`<_bd;j2#e%P9qveQBEIKCGoF7}0vusJV`z0!Qc`&u)i^TK&mN$l9D`OX
zKSnmYJ3&Yo2-+e15|Ic1^to|2*@}cetE&L(5&=SsegI?S5&T^ct7Ay{2Lq$G4vll&
z_uB=<CfwcB`e!eY+RejnesnD9>Rk`5H~$dV9?MSezMdVYRmUoXL|a*Y^0?MmFNm?5
zG6lFg8Mnt(M@F|CBo=q0tG{8&?cI_*SbgVw&PZP#QcEb+KG(SBvr8%Asf4<6Ug48C
zXC@q*%WJ$?E~S(H)OB0-XrwQ0%-q1F_ygIkZl&C)$4|XE)LZ4+Ih25k)J1V-H=%q_
z$9%At_V;gd+>bv&MRI@gBQT}J?>|Kb=YJKZEzS8GZu?>TAk!Wod?seksyh{Nb>+hn
zzOnX4*!O}*ya0{3dK$HB?CC%vQf7(?nm@FSkH`F!&!%S&gE<2^G3esWlyio~7!$Fn
zUb+vR1w7BuT)$eNSe-NFEd}u-7guLqlC`9OIptL7;Me$T>P2&JIctxNhdX_3>`j4z
zjq(2K3<G?b_<S|oEAt+BVfmt)U-?gY%*Aib9+$Cgz9<qx$|Cc(V%A+Z9;gq^!WAeo
z3p+$(JLW0&ER;h}-q6lYqCcEy*%jJAmEbA4!aDw>GIX~(6U9Z2tL~i`FM<2f5|IQS
zrQwboN^jV5lriF>q^9A;wLFs{IArgnr8Lf8ZSf}hHt;-w@C|IAh&&wHQSta%&IZy5
zgV7w>x{Kvd_uJD6#c*O_PXWLrOZT|G$`{g1{I9_WC}Tt)IX(uK_289S?+fvx(C8R=
z!P#D{(HU|9bGm-Urz;w$THXy~n2@5$=;t^)to&M(;??9aeF`bx)~RlD9W3?CiEVsx
zJkp-d^EjTMAMBk=i)#vnWtKJDhIPKT7pr~I&&fXf`)9@;*G|%E^SWjZYS3?t%RIgF
zpk(qf)xyJnVtM^2lgKqRdsux!<FQbYkbhi$XO+S9?pkMYRM*?y3cc=Q)--6Vq19p8
z3~Nq<jxR~FrCDYAbXUXV{mHmqx2PXd=TMGL9Jjl}*SLH!lc~(G%i_OBs~>iss4u=V
zJSa-E9<BCV?vxI|U?mW&6QLu!dw7@t4Iw6`!;YXb766~>j{EuTfm{v1yz%(YczNSN
zMM%*l<>$$__w`zkw2e*ilnWPU{cG*(RMNLocBUL8L(d|5!g`$*SrpTUf*w^rsNoH{
zDnD6wbE^DBh?L8KkG+T7m~FQ;;d|o7m^QA-hmMtgn~L3k<LU>yMQ7IVo6lsx<vNC+
zIXhT<I9N=4SYOKh9bU-GMYvGi-n1RDCuQZPfe8FU9AsB39hq#8q-Xh<gr9VF`Vz<9
z?4hRUX@)~YDjy}yp^8cI<hK?c+WePE7Vr6ND3Ya=cb9E2m{^z+XQLPmym=x_3HWkz
zDJ-pd65CeRjpz8QhUY%e^fQpoF}d&5)V3&FZ(g`uYR*6C_LSZ@b*&cJ5p_~b#Jg;y
zO+BgitAhv<%BSb=w@<I=D+`g3eSh3|6k?~jm<}RX5%cr&PnVmC0dDWFx`nQ;pLPCk
zK70s%*Qr*b{|Kz>&9?SE8-E8NbHcD;GctJvg}0WL1yh~IrZ3uK1pE8GYB^cl;t)t~
zZ&k3RxpeWQq?tQUl%kj_itFhoN#{=tjA1^5YKC;*a>IcG6FHee-J;z+@6fJB@%)vV
z1C~O=%lJY37@ch$6lMIZExV+osNuWj7$Hac7XgPF@KZnId0cta2?jpK$FfCXFf<U9
zL>4M|ew3un{IZi%Y`(_s6;Yz~+R`-UhS^G_2IrET>h9@1A(Gc?t0o!AM2aDS(e+yh
z($y6t7`J%9mt4@IPeAqdc!U;dej2n4&Bc=kZE{*cpEb}A5ZTg_p#KZlPeY9H=$X*0
zn%=mv$>*cEHNU|eS+hJDoxCt{|NL}=(D<0fmqp8ZfyJAKk#Qu8?jOe(&5hXN^IXZ&
zkqUYL3k;5JCGni|L0RbVYzDHz#m3e-^sxXy-^Uc8`tIwG)VP!$Cyzlvhlh_(%|0nh
z-^^?&Gp8@;5w2@81xUlx=ALY0hHU#a8`H8R7@KqCC{8SGH|sw=+~ECGN%3q+hq0PY
zbptiT#&jvfh@a92*&%b}V2{U@tM7A{nYWR8<|(|1Qk+fnkG6w{2hh*!0maDR(7@&3
zDc30QSC?d;h_;GPWF;v$A0A+BiFZC;I^C<EA=2;=hzheuqvVgod6Ox-vb7s7KEiQ7
zF?(D0ckwkm=5pPcc-(5gdeFVuEH$@kzKjt%V-`}R6ysn7+o)Nmsbd^D`kkOw+o@FH
zM6{F1m-|qWY-r@ic+OA`Q=Rx5jS*9~d+4-{)6LC|Bf)I_R7X$m*}j9n@mhkU_j#gK
z0~J4iBH;Cfx}_mKc))7V9Ywz%iC}dsueDe{q?2yKb^pST_#>aZi}!@6o}J{atP&yE
z2*)$C`_)ClS<B=VrXl^AUu>8`A}C#SnoPOD@A9A$`bg@Z?`y!$S13jvXukp?<mO<q
z%t7P5LUdefMieipH??D%>s#%uhc{g3Za5Kb_YUc(MY8htT=9Va*k&%+G5(u!<9^cO
z#7+M>*V#n>fLY^UMR&&`S=HFc&8tWLU-DZkB}jM5B+p4s4vt%fTH}36FTdm}6Z~LT
z-Zc9GDH_+(YojsMmdT>lb=pNCT`>xTYrj{A?bs7!L69k{tE<x;%rBlGdw{S5Y}*r8
za*{A)by6254x{q=dlp(+T9$+9sVQA6X!j=B0<2m6%HK8P-2VOb8JuE&9aC)v2MbFC
z$P3{Q=H2!cXK+;>QWbCqD(~r!KxPb42o7dUvrA|mvp+HkIOu6vU)fXa^uDU(4bu$e
z#wpz;!i^OOPHa?`&tAw;bFzvEt7p2(c*+xrqS>3&M;49tvwf0pa*1%bVy&1O*R653
z#mcU+Q=j9t7vkqLMO2T5vu?Yn!oU%iqG!I#Oq(}%^AWIRtEZIgZ141ed{Xfp$cDqh
z+%am|7KbX<2Hwsk6ts~}_k~Q2rHH3zfkZt#e+qGieNjsL7{Oq3QHHAm^Q>%p=4Uq*
zd7$Gn*B;r`**>O&1h0i2idxp=Pk;TT1`WqEF>VJ`5e;r&Je8lDW$g5>r}qd&g0}na
z4Ad`y`x5_Ztp@iQGWlt)-S+`ea5kzCL=E}^bTl}ZHx1PS6=yG!>aFq`R+GgX*Z)Xg
zFC9M!LjQTssD<0z@alJtRypRqUe|!RdJ%#7=z#)1(W-nkWo~Jh+{N}TChqN0G7mGk
zaAz#;GZ*{5P@XDo4|NPeZ6`jAZBgLbikOM_EEMmbDBBl7XKbdV7)$4~pKhOT|6CVF
zrbC*yVs3tDHJBh<re*U;{Ij3%VB<9L1Xa3g$qQRf|Hf>F2z5910)**!c$>?zonhpA
z%iZn9Mp!`I=J02%N;;9x-_=d<Vkk@M^=xW_ByJ3l@05V7jlPk9N$r7GWTgD11)H1l
z!y&Sy1ve-FM6gnHWOv!@pU-nOW^ehm-JG-0(2t&SzX(+vAzZLV<C;572^^d>jp~_G
z*7^l`72+=GGXrNO+Q|E};|D;b8QL}7ID7ZgdS~-3iu1>T`ryHp>lc@oVPn!zUhhpk
zGP<YXXF2TqqcpaaDxni<X9nbuFnbb;@M}(1$gR%7*^Mt^r{HAJgIkMkexh@a0jv+d
zm>Y6`Rr<8<SwH41pGO(aeZI=?k%@ht5AO~d?cmT89k}Vu7$>R!#RJ9!u%}FIP;Z#A
z{uw)s_a*V?o12bG@umHE$H97vQfmtZR`!QJ$?pYRW72b-xK+MSJSEL12gw~o>)*}7
z+|r#<gMapsmR&HG#s{^d{+gD{aZDRsZ*cSS^3*EY$`JW=ki=YMnS3i+GxlOrweWzo
z=}2?U{)`th1v8Tm2$6^5kG>a`cXK!84q3AidV#ZH;^UC(32CIY+3F*HHXToasL?~j
zh-6|`ZcMM2U*)Y$m(*4c5-A}v(SxRfFDegNl73!)axg;EIWze^GU9af(8m%Z|NkWi
zEf}(k_##1T8>1|N??WZ!LIR$7Y*998VMtdX>Z}I}>((?I9y->iV^s>gRHHlc{0_&G
z6i{`u<rLHvIF%h0pG!}4@+aAEcD%Ghk4*cTMm2a--0bsfL#S8DVG!gAaFcg;HAhGu
z2_Zu|`s-;J-P~`~Y0)D4gfKz}-eeUdB!4VoBF`cZL{#nG`EwSY(V?;28}lqI#!WJ*
zPUE7JMw%gC+*87rM!MxLZ!7eUuAP-dD?V7=Tpy&w!!Odr@;hZgpb7221b~FH>nC9w
z1W$3<kX1#9juj-R&TG#z@NjM2SONyV>)ZvS{&SQGnT8X-AhEH;%l2eW9&#zvzIFQ@
z^T@`#fcZ~S@1NmAyRtnB{qmavls74)p-$H^d5Om_I$Eb#Z~aQ()T&6!isVt<H`}c_
zP;gM;fdLco?>`n1J1^k8`>BmdN4j?Mi)jRE#qde~9V9`;N&P5A>Ta#GWU#%}=1-=<
z8*b$|+`OCG3xS(B^lkY?YC=GHrKk|MIkJ6B_B&u_c-cxAD-QobU@0hyN!_~#F+Ve7
z`OJk3-UN1bC<!0GB*bc`D~(j2e9huyY=Tu`S|yjpSU{MJLTznrw_H$-HDcoI3?)aT
zJ2?DlE;KqiCS9=AA7j2)By*2RCCd&Y&@g`@fn{{O^d58La}K8|vJiiOdI}L!KJkA+
zaI}iTU)^4QiE+G?a$g=LgUMVCCm=gm4eG&Qd$xF<+UqDe`>jn&)jZGd)@-q*wbFCe
zNLpTr9VeM1Mb_i>2hfX`>JhqCG7kw+8Nx;;rw2{m@;Kd|Q|*wImv?ZId<I28jBt{N
zxtGJYYNy9uRqp+JLSq?7^QBPI6$7OVL1^vNn@znsCQXxxgYy^+(`dPe4GZ^$Tu<xg
znnmB;P|AkIx|3I3)UaR9tn=1iK6Hz@Ysu7GUyL2?UU8p&lmfAO-9?D%>@UmguS`zP
z*sjlD0VU)sN6#wn3r$%fM?(b?qxFo;o@7Cyi$CN5l;GM<`d<!^@(yVpoNZjrHcH@j
z4ccAa;jQDpCPm2{^Lc>~g`ns-jL)F&=uu)fZaaW70vEJ^{|=Og^r^ED7pFQBG2gU2
z+*yD!wO$^N*s*S$U~goc2voeCDOu04eO&Z1x>0nR_s$_bo9Uw!$_T>ZM+yykAI9Wi
ztu%jfzZ%+7xx%$ncuF_a$fKdZG3tczx(q{f2=imX4V)q0x-xSd%=WMGQQ<${MuZmf
zdnQj0mPn8miD=&|ahK~YcS_5B(|+6Xg9ie#Jh`aIi1ci<y?ph-EHP{n_3BCSe@UM#
zyUWBAfDiv@*%{U<hFFr~(>X+Sl`w9Nte(8C3;?MDZF_q=+~mwmB7XMGp?$@?ox)(-
zfNa=$h7ui<@K((ijj3&B{`+bA1FiO{nJ}qmEj`+3K;S<)r6#P*Pm^DA0VVuHVnpJa
z)0~n`OiaeRLB{!3kf(~Dj+9raDkO~~z&EgTwYb2JIz2cea{Yr)^eZJf<Q#>nYnNq0
z953w&FV|CHhXY!-@RJxCC6d<S;FR>n{w~zQ+Tn(}@V%oN+<ySaLkVE64t92>N)+IQ
zt>*a;q2qmDp|n3Dv>ddtb&1;=@Nj!~T1wq}SMqW)z4I)dD~lluT@pP%Z9kV*l6p{M
ztgZzH>Tb(i=V{TfD4XwC4ABsH_vf5Qa7}O?k-3yv{xDBsP#{2JSN={hEN*>d3t%0y
z<Ujve%QCWo?usZlSR_R<pMpenFmdr!j6pGUrjH<Q=0iH<kNDD^=jL4ni-wq%Yr!!4
z87Rw8%UsqhP>N{$mKZRc@hJyCf5A)<lFM13mY(Ht&n%<wm99t_w%zt%N7i)#ne))#
zX0`CEFw@RNoiG}D7PT<tzjMsb7&D^@)MO@p4$inSkTOI@Jx%2eUu}vP04#Xt6+#l7
z^4S7H8txsV%M7Pz)!Ite7Q(s3dKu>`3=lyzP*@_Oy;&-?u4<2*9c65A)e*|(BG2RW
z{JmQr6=uZ}e^0jC<A>&dVSkkUF10ZJbzZA5BX*?{*zd}hY=J`$7GA((v;B(2Fa7U9
z1d&EEmH@{&@l&d`1v*5t=vmEqpc0F&r#{Z|>Ib}#07}!M(cUj*XliY7`3H|@(kZWK
z<fHGXe*9*1-z&yjddPJ>*gDz?Rrt5rXK(5c_6byWnKk$8;z79h)rO6jD_g`*`u1Dz
zNb5H=ICK|B#+-yRUOAIdbk{59@qtS@mtDpk)M-~wr8;Wjeh7b&lHh%3GdJh^@Bi#1
zYKU+%Xu%@{bj1A4lOp2jiV5=Qsxo_;R3|Y$46M%jJp6d<3A*@Z*ox9ctF`A$+yzgI
zrsKh%kFJzletLF8L%@uOGg-*4xxX=D9jxVx{=#+qW+9vC6giGS#?7YkGb+&tPF98D
zCxJ>l$hY9?-RLL@XJzA^v;qm|&AtFxcY)hCe`M!p$lpRXc%FZjYONXbSU?`NFc(H}
zb@D4kV&ztHVHz7NU?}2%2XJb-R()d^+W6nAPrGCz`Anu!oP3WF-|93X^k=Y}97I_G
zWmV5%@nC$P*v0AYzBBY2E$?z78&Gy2ZijA>xcn}(I@kJ!mRNvZ)tHgbf9Ic1$j~^u
zb46#A&Z}Q^EP%ViN4hQn2yUO>Ii)h{7w#{<3+6P}a>$@Kj#_aF_!_QNUqQsxy0H7p
z##pDZ|1E&JHo^|}*~KL=G?@luHG~8_aKhIX=YHU?5D2*T;k{8L;XHTefh6Nsi)e5S
zzx$y1bM7mc`?0XJB)-&Pvw<DN^m&@;fBqE{Ed~5lUEFUE+{|4qkS;fQtxHjQQGKLq
zo3ZK_xA$0C9_$?=a;r<-l|{Tr*M4R=Q=BDLuNoW_)YRV}-V@8RFv!ahKy5X&xUmr=
zmCbNJug_Nb;Rh(PKyfYm6khq`;SP2UuNnm&drJAwxTi+&Uzjw%#)#(wSq7xxolLCb
z!zJijecnoAZS7>aC2wrM`fjSrB1*OTG?RYr+?p$}apv7VO28;j-GJ|#V!INf70`zm
z&{Ig|qt<g`6OYKIVui%AZFSe9kESNQZ);hXZaa&e99f@1T>&c_n=Mg_b}F16FN4Jf
zT1Hn`2jw^QX1j<5%%h2ae`qff#`To)Wq2`Zx=REqDm8CUjJM?fotk|LvQqHcv*(zb
zMv*^~pY@{qgiGP0R8M+E2-@5Eqo}n`1V_0;@zEm$Hs?N_59`yNMxHP9ZrInLayF83
z26hbofsnMoArXT{K1aIHpFhP1VGOB<w@d_gE8~;RCRMEl&uktq%5YS0&lh_4hMHuw
zzW@&Iq2n))z%6@wP%M(`*Lcpq8B%hgf&p7dZDTm~_4`+|;gJVp?^bhBrL>sQlWb5x
zCGYbUNwVhd&ThvW2E$^jDWyA=lHfFQVyNg@`t+fu=$Q^+&pRU}ywp?s>w7>+kXSO$
z#quXW{g;pn9Uq`b_VELpw;1lBt&h{P3impi0HOytJZ6BTOczP=uD=100*4b`Kcvb_
zefNd(+=Igx7=2pht`_~!|F{HXt1Wj#w(_O4T?{Q#GZ+Jo4fYTjj_5&Z0VoD0^3734
zl0;^h&;I$O`UD^FD8Ge(NcUHH6WZcTHRm~Wg-&OL6Fl!4<%d`dxB>v^ZU%2s_)41(
za~A~BGUntvOg)8<jd#p9?u{P^(qKNfy#Da1f^uDass+DaO~XU*;YK>6#?aH>g+E8|
z|8Cbl4QMnAv29o2Q9gMT(l;}PJuM4kwD(M}?_Cgt8!LS1a@RE2Ertt}K`(5xEj&vi
zNiCR?$Z87@)GkyHF)jU_tQjoh9o^)!Z6&)UdgQq3b{MggNYVui$uAu(!>rcFV<_)Q
zboX~&ZIIWm%5nX52t)QCsezx#HW!}4gp#&oXPN8@Da|El4mVUkIVGjGF0$rksYKq~
zxPwN@gQn#${y1skF9}CL$y+*`u`&XDMfbBQqv$<hm!s!^mbMnF?YRVSP%vV^W<c=#
zU{qduX;bwyBMNj&#-^l{R#PKURaNax5l#V3D!#?V{T>j$?=8}8WhFw-gS6@sSQ;*T
zfLLE%L*t{240h$5BPpo)X8}%o^s?yNw+Gkzjdn{dNT4ye^hHoe$TFbH8Uo09RSz^m
zXb0a&si>@Wf(=~^cv~@ysx5$O3f}o-cP6%z!tFDt(Wfg-+;4#%aO10orsiM;VQWt!
z@1l<CfguPjNWO}ZiqG6ORQE`kA131dl#BMyYZknSO~<LSrBYPIO$$tPc|A%WH;KB|
z@wog@C!N4KIVlO&(LPtX&=9nd0rbtHZW~Ytc$+=)(qU^9wKDM|HDHu=0S*TJECoJ3
zjE8uCIRH)8yK&dMp8mp<FPump*;FZpJ|VFi_U2fp0DMY^mDPCg2@t4jE;RshlJ#=i
zBOr+p6BQNBM6h+ZwF(n`HXwA(NC`UI`_Bv{^LH+{hpN^&zwGMjx&Vyh1!Xxok+*Lj
zfaV~DMFZeBr|MkT!H0bU^fdtGDe-AJ^oFU<tc>9m^mV|hK5r3i4j~(5F1Zkf`L9}O
zCrsaM5G)r^&dGh;u;^x_zkA`wK1t757WyfUZe+!N?}hi*W<^xq&Gtuji$YjE2&XnE
zVOxdHVH;gEDGzjp(%vb~r$k1!c7&1I*x3AV+7D44@Hl8j0nOUI00RTou>=(j?IlT=
zzLC)}_!ThD0%}Qp?F+>Cxn~z(9Tt}YDc67{5rv5Y7!s6M{!}h_>YeY+`T6--Z;jFc
zZWk*_7#=>p4fy76|Nah3QE*xnMxFt4A#{w4I!g?Ce)sox7;^x*s|eu4g9%DMIqVz(
zanXFCAsLY7OPuz#?D@#S04P8W!yzJqegm>c_5I!Tc9JjWC#7>|PNaC-=b&sHusp#D
z7p>;M`M!cpUfIZRHMoe2Sp0MM6GGAhW{agSN(e76yC=FOSo^W4FMUOk#7bn%Mlbx4
zIi@cE1<zDZ7+z{l8Cx}rbi?&?Et*NcmSe7ISr*kK)ob#C2aH3de6{v<C<O|SlyA42
z0!$aU0fFO>{s_ieS`=U#k-<3l0?ybdD1k@UFf#N`aVM~A*lZy8<38?a!}}OwA$J}a
z*C+GSF5m*esg&q{@K=IOSpa>H{4bB!CoahV5zYnZh~Oq0Az{$?oS|_5l&S!E<6z<b
zuChgl4x*!TzuL@RtuC_!RkB*3P_#K%eC_V;4&!NohX~4m6A6L;TNzl)16F2Z>0=^A
z^Ov8#`6jH}c<#zI2h)O{MbFWd&gJgvUES&^iDX0SN~Kz3?0*l9R5*#1SvoYLzyG%^
zx*kaJne6g&d91O;9HMtn^p;UA*8@^gQqZCgvx%kgMsD*CP{!a7z~B+Zs#~CQ=``kG
z12Y*=+vjrUJkICf^0XTprdJxzrp*)y7<#VODTF=iFpXqOLCXs`(42h({Ko|Ixf*=0
zU|bp{+P0X(laoK2*(u06Fy#wDX-#J?UEJ+2VTUa|YIzLk7OfADgv9hT4j6aa*p(FV
zYfuDIah8fNpXnmj+Jh5PbmuEXnC5ROgCm;2(Tebvuf~qWdetZ<%3*o*Z$ku$mlb42
zG9R79p!6tS<?vN;-MtMLqj2#9t-@HX=If5q7+(Q<blLW59H3@O)a3{AjJw-?NSS!T
z7*Z%u5^&g3DDT_@t;eD(BPfNxZ;oUi1M(BV%v>qlc3xUGTpHfo-fq_yO~Hf;^hxR#
z-Kah3qG7)rdk>acUH0qGdF(f2(%bX)4oy)F37r8gHP<}K{AU0y3G6IW-WCubqmd>&
zD*Pi7$51y%yt91}7+fe=M~KRj1O-G^1JHa(cTf@|rRZN&ktS;CnTh890#Ag+=@=28
zlq!>mpN8qIUv2xjGL~kIee=LTgva&S*Mx)+z)OZ(@Hi)j12Y*4%A{>=*;4UrdQhv!
zewsXH<>#-Zw>k%_KU-~AchG{ysNIm%m&lEnI%etu{D;e4^?EmSoF@tV!z+KJQPaq;
zU*TT?CC)E$`HCiWS6A2nZfQ6Tc=a|6n!`!n%&Kh@G!~~kbr40@lW1)|q^JN`Nu>Re
zNb3vwf0<TJWD|U5z3#c88o`D&*T&}u7!SRvt8TP1<z51g7#yI_VbDfW^-QxXa2}7f
zLT^-6EP+xh?qe27BeO=%-K~V=E;3kd)e3WBFtq^Id+=Q-!BDlG(OQ491dvpLxi;p>
z{R2HcXcxJI>VyXV<0Wb4iX|l_^let3<0i!2kc(2ANXf}1-QDZP8o@OM1)Oz?WeP@7
znCodpFgvL`zuFu*>HJo6>WuS`e<i|^#3krsbLQxu&WoLBrG<@qP+uY>7CYmr!cy&;
z0F)+hK+OaP8=wF}S-b80N+sTvGB5<ANjtQN^Lf6KgtIfxNVfDkz}wxyaSMu&^RC<Z
ze7g(jfk#yZ>XilkopqbktG~KR?&=k%DlAChfV9l2Bu=tx2O_`#m7nomSee%tucqbG
z@_i737{=!lE@2rzZbr!`mg^r*-tn22Rg6o>{oBE^ET&-z00`<EuyRKW?QgxoF8aBS
z-ea)|03L!}bsSr_edMlE=9`-vFvf}oNUxrPK|^3;<9LBmof5AnVD-Z>s#kng)^hPr
zg;G;H#G|R+1i)eA;>u?qJ`<%iiv#4x7;rx$K-Z>rPRwOal$xXsm}Z%0m{U_zj2czg
zG6Yt+p#ectQ5o-i!2*+Z14xH8YU4+(`tAK<Z9+0hizdk8-t)Q4n_fQp=gxaa^p;<#
z1(Ycgywh?f_<X9ezNyb#=L8Wb&|A8#Qs!_<s<Sbejj%#@?l748Mv?56AE@TQLq~g6
zX*s7-Zi-)BUA=lv2%OK=i3xbU2Lu{h$vuD=-xq?<@c!e+w}r|C#q1<u&!0aBBacQ&
z>~iTRxaXY_y}%fXW55@bk&%I;z4d0%sA>Y$B-J1Oz9}gXC2TAOoX6G45@<*B&+h@^
zD2QGNlcmIfHz((J%#)|Ucw9;sF9-u8{i-NdpU0m4F*Wc<!WGv8jr3G0B>p>EABW{*
zzye?C&c=6lYaG8vo9~h6%&ZhTjeP(O`vPGj3_V5A`wUTpgk{q-s*mLMPb4G_7sp|O
zgNcbb77l?PTpuJPR_2@*RvWJOavf&@lUQkABP~7MWivDOSG_O~Pu28PyW%AnAtnh(
z#VMXwbYSXJj_pU)rK;ZlpQ}22F*rDQ^u66-4EtqbNG~ktvUx*CN2|m|mUjR$S5X_!
zudBs=(7z*xz|L&7>rb?u@leAHvfn;|;Z5Jry3sY0Wr_6><U)1zPz7ud=7j^hjh6vm
zcliG6SIGp3*vM!dwS54bmJ>X%-3^}Hr%#{a0M(v*&K?)ksDLWM^>nKpuveyl_XAb;
zd?)4<_oGLoT;@7Jg#~mCX+w~)HQ+;E94x^s60~$oN=n)ylg?KvwcnJ(q!9QGW>CRE
zO)#+yRAP@d(j#mao4m!DaLd$~fKSrxOW*_(+Hh!SXuu1w5f$wr_+<Q!=-|>^la=4J
znbU`H`6_3RL`$y}e2`2Lg*~Vk+ap^=E;ff46o#-RXtbqVtCr5gOWNj8hK!dkegjRy
z`p7Z;-F1vT5m|t8s@3l6{}3d_BlGpvXSv$?ohZUKi`y1N;~D{+vC;g=7OB7Ct{SJg
zF9MO+j@sQ!M%b%@*$SqC>(t)&0FAC*fxUByVBzqXlvI7?7Vulaw4{kzLW`-V24m1B
z*rTu*B6;QKm?;R{I;(6}@?v~>Uz)kqC5Ql)gCsRB7!bw}!f^wj`@?sIMk`XlLmQG2
zKy;4^BVtz`uU9L>Y0XWRP7WY6AUx>o6fT7>l}_F%62@`g8{q?H$4m1yXM`(_>-z8i
zZl#J~s+NZpT<Ru2Wbh!{Q|SH98LkFziinmPrel9NfNOCFvvHFH191V0lEbKy@+OFS
z{0Y+h?5qZia0FzHq)+R>HRy6JNZnOcoIIi7!<7qZo|vzD6&p$=M0q~W6xGvcYIA5U
zkZZ_DEi8y)KE^tT=34t&g2{O7%MV}n>Y{DbV6o-PEhy!9L(qiw@qI&K7oqb%V_bG!
zogVgWh!h)ivYjJ<$6a>SKXv*{9SPusprd8WLallg;2ca(wF9q!oB(EUz}UQuQaobH
z%_d3!H_<Hn8Msk}47tZ#6)7e4%ik_-8l)aPG5@%rCcGOl=jmgB)hi!T`$oM<8Cp7!
zLkT+Ywk#OzBEE8KimEtJ<p!`mGZ&dkL)d;6Q;RWcLQ3|(3lj-4>;nwQWPYb3pol|Y
z)oFq|+nejj_z>A0S`BwT@5TcB1W-i)eWEJhNB~!@yE9olE+q!cE+^wY{jsqo`O<{K
zWQyWF9AwxM!$CF+C2uJAJVoC;jJ;f%l7<OBnBWXi-k{ZPImzVe_;I^C<Ccx%)ZX$m
z_T@wSO#jYOKWuK~TteqqCUvC-LFn!V)qk!L?cu{k5UpkB<OBini><9~UUE3NYS~}_
zvd(DVXP3C{L7$dX1U>J`x&o?gbbN-a1ib1cAdFDogpcOVCcG~$-?0u+c3i*KRCl|a
zYE@5eo(k~Lwp<!6wsRN&^u)iW(}vLs7635uqOAf00+e{yw){uCm3}hN(=UgzG!!}V
z!P1)On)#tXZ|V6*NXT(^Y9<ph3U3(vL=YLJ{RsPC{qSxc2&d9lC?tt^T*(ju`S}(P
zcED4ZYTiFrH(460fHQY<lDs`9{1@5*-jgceJ%J=Q#tP6;Vy$uuwxo^O;=3EqkB;aY
zZvm(B514aS3`7gC4@MTYBP?jxGk3iu{j9C&=g?&xO5<01HzrDz{~&y=iIVtBzL-V&
zQg|ih6ZaOT@($rUXhq9lCFL&lrJdsZLo1c9zd>#9OJq{8nFQAH@_#q&XB?BKwdi1L
zo0gN4Gra5R>8Vj=4e#aS<8!pUKjyyxWYB8-<Y0l)L_sg8=ZkCMHw}woQn$7bxiV<A
zp=`Owu*#MYH|Bt&5Gbm^bS*saT=W}euVs~>bRYQd2YPHa{8Ga!4eEUnFaA4i(xw12
zlADJo?(=7KxRB7$A=7d=KY#ydLgzw_YB{Y%CPqfa1?Bt-*G||~rYUK8GsPb?a~utq
z1Oo%_ktO{7q5nZs-vBP4Bfw5Fd9(?>PR8EublPk<1oGzME6)>$;SZNkvdgMFzQ57`
zH7pKbSfF|L0Fal!f$j%Q3k#H9GnTbsz}GzT0?v@tWcWE8CZ&)@gH?)<$A3m<)IV=&
z-?&atvtWnze=8CUEcNzFTv+NoxC~*k6!QODOj?6ex>6wCVptGzKG1<K&-@<Yc2{Og
zCU$@pieP4#HW9z0`Q<K<p~jVuJO<o!;IybYuW;7y1>P(#IkVvo17xavpY|>1PEr(u
z3N42UMP8E($w==fJ2kG@@Z@J8V)_=ILL>Za;^=>+A`uPTJRx5vkoT?5XuBUG0i8K~
zQ)g#jtTyBs&%NLt344|^Jz(0QEr_O7VGlDWrptJFG|Lyz7`fNazBlI-ss!1cIcBaD
zSzk#Kdn@boP<P8|6l&DXy~^~-_{0itSWt}IWRO`|pKCSec1f^`SG1p-pLDm14Vtvk
z&n*DL>5t0uwDXJ~le`JaJOuyQH#Bqtvn?m5LffyLUK83-GR)5bLff~L6m8FIFnac4
z(HCpbm~lsz)(_hu%9i@8LuuoaO9dZW-%<>Fdqfs~S}qaW$1cUZWLnjc&lc`q7eeTt
z+LGLgiYr@I&z1vHi%Wli5<8%-yfGKDw`UWcykmqzxU$b1Ie3=q2py{bF%2m}iy~Ds
z`H}2<#5XY1T8;2c`o$M_dhdGen+^+>Ti-!`*TfW<(}7t@5VhVZ0*k^zp-iUSA#liS
z2Hkja^74Q?-#jrfv2zyy{6hFfey7UuEDfMPMy1X8TBHV&PTp9A7*t>a`I%aDn|C2Z
zv)SfC?Rf*?0f<sEvZ`v&p5k!<p7X?u*1RXjK1oboE4)BHk?G)kO%j-ZF6!yIqK@fV
zsu?dxwuEtB=2L+Qkv>df@QRN8z3{ymYrsR~C_WVZy3wZlKRzu-@q{m>1pw-*EIHnL
zS8|z_C&VzV+8R;4U6ByL8Q97eYDr~p@`du3Qu1>f?dSUl?v}8g)i0G3F_ZIm{@M(6
zw!Zwx;V5~uBClcODn-_|7XkgSmLLH{8O@|^!X{&%7wgRrDg}g4qqEk=FK6bjW$E^6
zrdk-t7pW(Hq5SeX3qSaTLbQZo)md`Al44`+lm;BYY0tltGzj@%<wsV3>K)W^X1;~q
z7S_ELnpY1}>bjew9@rN~O$4^cs4|fJ<r+p2IAtR>A#&n56|Nbe9KU%o3HVwoQ&)Ql
z$ODQTSTL@#i)9|0-q|V9d!~Zxe(J@@2RsSB$~qFRcX~2mtt*a6`sba(zUiVa^-9XK
zx3%+o|2W7mj!+<Gits|W@c*olDj^AwkYPZXGstpaUnKURh?P?)EIjj3OO0!G>(R>1
zR%Izj{q`T6nVzxCh$?+@ZGBFckgRo%_0(KLxJnK|I>j$%cG`)at)-4f#g5pG5QW-P
zAid*mfZF%#!u+My<y!R__M*)0B)x-e07Qe(r2k3ck11>XR^?w3%l}NQ5L-ke^s_7C
zb&Lg47%GK{dS!eqv~Gdobger|%5!kKuQVXbAy1DFQ$8&@VR<p|IWwcqhx@Pa>RhN(
zan?To;TecyW9&8DXX8%B8V^wOs`F)U_T)3#sp*)WAiKuLu3HE}gX7+S^;TuST<7xg
zzoS=#qsA;PI72EMv1wdMDVtD)<Fnrz8!)GdHuPuta02B97v;qo4#4TzKAk;K{USaZ
z+%X;i6Jc4nJ}9<mRy%UiZwg1Z9`uSxW_xoJB1Lx)gGxLW8Wg+N+TK||egP`(bflG+
zapy}&v=F5ym|7`~8#Hf=cVG019r3?($<O#BKuY_1dl?sCzyz-0Q$OQ=W^%z?^OL-K
z6{Yl;v&F*IZhr;T(qkLz!u2%O<&I3bq1$=&M}Py;={YG0bPCVOLf$|>irj<5NUiwL
zAEnP!d0seR<k)RUbc$;)2+lJ06vsmR?g<AI_k{mD|0e`OQR@Ki6n8hCJ)pxHWhEvi
zmae2chty)s@hv7s6c)-_z<AfsVP~WgU{{tbnHjmTQo2<E_NQqDp7Xo<Dc<Q3w_zDq
zX%bXwTz_69a0-BG>}x%<`NGxNQFDW}f=d5FE6Uo&<wsP|zcsI72E}2FM~~>N3T2`}
zmx2j~c|iSU`~V`SUj~EGo)`Lh{^KiLzd~O6v&M@1`zGRIT1pZ411(`R0a{E2I^FHP
z?%*a<C@k*#>ko%Ca6|JPz%)DIy9-~?fjmp<-NT`eyl$~tHQ#0Bj+<&|!<LrymR6wy
z5tZQ52+|G7FJRQSlGe<BJHckI*cf>q1nLas#$ZHYt1$<HHJJDeQ%hd|QlKXzCqFzo
zY5~6P8wbq$3MJn}hk-$p*@c(hM<><|*g)q6h%xz*N!SEOn;X(<<~T{c-tgboQ262z
zy_|{~quKsdV<8a94>DQ~9XOGMWz{v25)c@EaHjGrN=SvLGB&qh-RbdjN90kwU6Rjt
z)PheB2CM~{pG<XRS`QgEL2WW83}x{gy$7|f(TMgiz@b#m@P9j*Bw<XQ1H>ub85z&N
zefyS24(d9+hU*zGPQvG9biDRK^BYj@!-s<5fa6wTFr@J_fK^wAT8SMEUuiVByaom8
zCG*+adSl_#PzT)qu;+(EI*bX;>iELK!e)WswOpc`UZOOvoz4SI9D65U%TwcG9fUh4
zf?;u3<+P$c!>Fpt5fpcb0eR!*XifaWimYy-+_QAF)Si`v!mD`z3I}#^Dvb5-xf$b7
zr+<x$yyhu$!~O8+rM^K%s)Kz;PiM;D9^&Rp9lZVm0u3aZHqKPp`DT0qf^Xlyhl5?3
z3c#>18x>%WO>+yOvW2q)0kN^I+b<7nQ8EYBNqry1Xm~Mg0+j9LRsIKzlsxRy(QS~@
zfH3Jo6Osir(gRbmgQ9$o&e2cKFG%D<AA!wU^wNV=NcG>YAc^bJUmnQMRV}`F@eSwG
zFS1gXKAOqN>Blzb9D5#rUT6r`k@8AV<+W}xk#qz36M(oRCkixe?CmpvejQY09^>ID
z=+55XoeQt^#<wQ%+6`_qY22zmq%s|X04jnkASABJxa-e;aESE#=aP}M0`m<6>iCP5
z#kID!7t+hiEA!5e&}78`>{Gjt7@|D^#}kb<BHd(VqQjwB?>qm1R8FVk1CRQvSNq|g
zUDkDFK3JLjy>c552Krv3`5m`7?sYA2<p-7!d;NZVXU@74>s>zk?CH^K2wC`!y!n<A
zAsT~JxnV&5Jizv9gHkfNW(xExDk?2q0o8!M<w;2wp^0?l_-1tlg*Je4O_iG|=j)-N
zqB^f9JHf@V>LP$ThiOn0*b1{n@hgX>>#bQ6jj!)p4KC=AYHCG-(4415iXX{^EWYAO
zVu5R_v@_e&6d;e_I)Z`=svJ$`$XqTG*QXAEq%ChDyG-+f5#I953l7D)*xOtXc=_A-
zZmvPQ<tLVI25Du6|F-x<eOimVmZ$qen{3R**);N78m4Ap08#~`YrzXX+VNni-}~Ov
zObI5^tuttS=ZYenzy10h(b6hyr!O0bY6_FAFgq;){#i}I`J2BQVAU~%PJ!seEoO!F
z9bc&4fI3HX2PROoySrVV?aBfCgd^nOQ6)A<5d**QY*Vm<8{JeasL9iZ9iyHYG>y(Y
zB%Pov&uV6N#cnYLrAVUly>u{P;tToyecPqS(1D$4rVV{qLaMhWG1xL{{^e#jZ-LSU
z0;g3GTVz*Tpw&mJy;1nOnT6hiY)_GGx#Q>LD3vj@e5(>_4^H&=E2WS3Lhp6vm9G!1
zCaS-d5VY7Tu^dqF*$08SEAX(k+n~{sk6ONH`jCR^A`sN;U7mq%3h}&lS>^ILB0==k
z7gg+6VJKC*nZZD$_(t6PDQ}nRBxGt~b9|hpnG)`dRAIiw;m;}otVp2h=KlA<@70?W
zqu&iluXUA#!YTTz$}L|5W&uwXFQ-Gozhg$F&s$Tt__|V&$ECvHLVnwfX-_M^rzo5D
zxA^kKfZgTuqYqlzb(!$&<qcpbaOUc9uU=|I<Oz}Q9uEk!ENB(!K=7sxeSy}Fj?4=}
zE=L)EBYriL9dl#hCw|}W$$Imp7r>aT$9~d8Puewy7<pW(v0cTiu7n*)TMg$V8^xYI
ztj^EIbNZ_pdRq!~br*(*5^yK=_a0@;LZ{(F9y_E}cfxMx<n#Ifgu3}Sl97n1H09RC
zD1%3gKp%vL^m-O;6eYG_2%ZQ!o6p5lt1(JAyna^Z!!*c6x2ZW_xN^Yy{M<t`CBtMo
zss5<ilw+(z5i%V~@SkTs%mb*iez}@rrg(Jep6n?|Mt_@@Zvp6zfA^~NG|yt}S9G3I
z?(p!i{bCb*7|}}+P=AdXGaUnXkE(N0pWz5-L6pN~F-`mY`P*cE$Y~s4KT)x<*<b=|
zmua~vkhi?+5FVM$?kA<BECdb1f4%laMt$^17If)41+ZuGp=V46%08&7ae&rk<>pgz
zuV245ovkVcAWZ@IC<0zCVPS1B8D#<yOCS{#BpJZxV32SO)}PPkVvuqr*<hw0T%pd(
z<DP*z2856nSF-L10q9944#3nCM_xGe-mg{@OAc<)t~?)9yNFRyv1FZhX~(uF-^QfE
z?Mwc#rPH0Eda=WkakBFB2&;&;#F2Ase^P;J?3c3@1swSEs+y<Ds1!60&iGxktI&9R
zYCzIDU@lvm3)DR8=O$f+@?#ZtRxHXchE!Bk5DBY}7bx2P#Ai^-0qA-=%3GV@|4|1h
z&hM`G8Nls;%<<pqme1^V4-8CzUS*u_7Y8Gjc06R?GJ1PI0;jJE=HAoRS?o@itJON@
z3tjK9gF(X5pa*%oCm7{wN7%~`DiL4-@~oM=Zt3noePJ_<dmaK>j&yN>tmnMRQi5Q1
z9Eq^!jmgCROr<m^29JW)A<<vHq$U8gY5ok%&L-r&TWt1CU^58G`WT<~Qm<nJ;1G0<
zLeCCdn`)u^yOth!_)IjNqobob4Gg$v;=FM=FL2jPXZCGM3|0xM=Y*TmvSQ0+2LcP_
z?Q>lXzSD3+g6fb5m-My~lX6FK+O3lT+fS{JdYV@$xueTRXYUe@5MfJ7A5h){eSOZk
z!84eCZh6Y50ov~uTGt(Kg3|3c==Z>>+v4{LasK4Q{ucD70p(tKb93{bos>fERRQS4
zunWyG>`TaI>Zk*FRk>a#Cd_U?Zxp#uh=A#jZFQOH7`-NRd+G|xTHWrA0ALdm1Su$q
zpljJL<-&rxIswTPA<gqyyNK@|XR5Ftg#-oB9y$Z4x}qb3k|lTS7cgHFkZOa^j-0cc
z0B9kn%h8{w<VYl(rWv4F(q>5yvpgvI7;q&<fBpIcFgUlu`S%8MAlY;d<H6JswJwt2
z3B-9YQ{lz)=kIOr@l$pD_u|z!C(SEbdVu*Jz$~+E)!t;Yty=A9NQ0MLs%x3XA@~ak
zii9GgT$UzZ+_URx4vUmZshRbB`SlsMp9{y8zs>39dZxmK$IbfLQy6HNB-lQf;?tL-
zM*dRE5Xg@DLX!_kj{K++OgLM7z*$%2j~sR#A(9&yJCKY5TlOf`mL<V&8KxL~E*w%i
z#+?327O7Dg9wn=*ze%wFgdK@z0Ff^$I{GM3+=JWI_xDShSF!-@JicXsO5~p;CfQKC
zc6?=HfP23aGN1(e8T5<z{i)b^wL2OW>wev1KNE`8`US=p<r?&qSn)Z{SX7#Tel`C0
znBYx8MIe6xy4M7g+QVX4v<u7%s<YneWoBmL7hGtTn?}RN3^*sYf06|clEbh!?&&&c
z;-&YlAXhFk1R0$u74*wd1YH(JX;R$NKuheHvpvO!TKmm2055Ind0|p*y%n-x3C-2G
z)RMs=FO~6gY=4a?-|*$Ny^^TOE#OFxJW-R_oWB}gR2%-{Ivm<JHz7PK^J}?~bAfSK
z*{M-^M|^?((rkDiws=leZUzpUJw%(nT*LRAUo3wy&8aH8rUjo!ifyhE$lX5Ky>|<w
z#&FdPu3hQbrh+XlT$HTZkqz{LUVeWLLd%m|DSJ6U2tkk_=*;m=!5o@)ns*WVW)F(;
zQ6@JaCMTQ`aaa;@YFq;K*c5_20?^R0-NZ%PMDTj2v|URWz_K5P!CXNJfpJQxio3CP
zEx}9*%}H=5Q-bhAyYQOm(y5ozVCrlCSZRw<cWO%L#%5_%-3qE7n^}cp4RX=TGTXK?
z%S?bD4z!qsw<@Z>@Xhxe#uyrItB@PPCf6vHSP&Naz?{9EbZma<Qoh#6DUW&(sb$P8
zIMpU-Wb^S`ji@;8w0M0kogme_1&r(Wyg%a&8q@#YX(1Y>&3?+al*GM5l}p~emp&r>
z_um<j60`CBplYP$m6dW(b>=i3eGk}FOe2$^zQJ&I1^)fv8*}YW3Hx`qF-7HQ6R8is
z00vt;0k|pR-%3hLB?Oc-G?Jcgc92>zI*<cP9K)3s_yOEM0|$qKQm&#>E`U-_!1oe@
zhVNi^4~i(jSfieYkpX6=iOF7Yh$<h{@Tondg@OWL*fjuwsMo}VL!2x(j#Tofeua7*
zJKyl$6i=h%|1kF!Zc)8Y*f<6tpzu)vrBzBmKtMVbiKT1lknZk0h$1Qi($d|rbccX|
z<kFqewd8{I?*ZTM`~C|*uWP-4-LvP3d1mgpXU@z+%iEH+6=4>b7VSx7Wrvfkrh)<-
zY_g8x-Y9&VBuhH<XU1`X8$JUZvKi9Ukq*xl8|%(wJ9@8mSNx;l)aldGbc)Hfeja?!
zM0R;{1(A0}Z<?iM9*t;0pk`PGc&%+WUXJW8T;@MX<g$*XSp&HPEc;_%3S%4L;o-FB
z537i?#Dhu@-?-ue_3GeL7JnQWR*CJ_u`FnJ_qtf~03PxHZ9mIFI+C%mF_F!pYSHmz
zPTWtDAfQ!18Vtx`a1Be3h#=&&o=^d7V5>R*Y=L_58;Dx`8&3gk3n&nywk|u4@i?h#
zZ6L|ewEA+_%rHllK@U6kuQame(xck#jC{NVS*Nk^XIGd(xP4jo+_NJFvLouES<A^(
z`Y5KgTsYF-gDvGMbhYQnQ)O#YT47FZ3rNbK@FK(F%?hjXpe^N0i~(i3aG1Q%aXO7T
za15X@n&>WgRchr6CN+7AS{}?M8w(4zxR4~3Fs}R+QJ^ttBMTEj@WBEU0l88&Ai#tG
z<;bqAqFk0TOY}zBdDFMfjT?wXRElDs$5RFdnCPdwyv%c3BL#6nz@jxn5Xd!I2MJ2<
z5jT_l_z5c|+Fi**V*@0orK?h*!&%LbQi7=w&d)?AS6YcYC4GWr$r|+Lb%Q)Tr%QSs
zT1D9)s9~C?)8lG**ZbL<G{PoGq)96!!F5?U^tZs6(zNT$J}h0@gWy2bJVUjw`CzMZ
zHi`ydKUs+oN_YsRph|%zCjvQIVXXxNt--pGZrz|5T)mL3hSMfArWker^2wt9#siBC
z3j@KhEHay_*TZ>g(T$Ce{+v#c-}vuN+PTPdP!M1ZSl=4(WH0yzkii6^@s!h`mg5o}
z%NN&XC23>&5eovPORJ?Ct>zk1`OdJId;8QfZnOGWDf$k>RO9CUK`vvAT<-#(D^l-j
zlSDu%*h+QIoy4p~2VV4%Ysl}V7w7F;1cgay)Nzn<kmkMN{apUel+LgmcwQ8r^Ljd{
zp=1WUyc1w$X<_mLyKk$zNAuKJVFRGVTXJ`Is%I~#6RHNncFfV$(i@<@YXpt=<@wPB
znEIDqhAMgDzzk4x&-+Mcn{Y*RiFii}QUXrY0eI%0ft6Jr_#A^EnX1ff*T8zVK?lU$
z0rH3<<+jmIq{Yaddr(2)pl7zNzkd?GLT|<IcIZCg&LjU(%;LZN?q9Er6MpYly||i|
z@2Iw8*PY?QS?yh|%f59I`6vrTxz70@P4<p@N&8j9?M?%<HCewO*FD2_D0mo9@igZT
zZ9KF<suCwqqvKXw=*@9uP)IwnYbk7SO$BI~8j!)ovL7Y(_U*^3YPm2aE315fDBxAF
zxR&`okOy_Jo`*;rX-)xC51R^E1FZg%6CW*%UM{|m11X?vB+5^{_l`uXL!&^I(`rls
z;4lHI!5c5XL)|n-;@@&Ly-JTB;IT=+HZi7@XYxt>6+WuAZ$}BswTV15Wuj)2eLCEG
zS(tl0yVRREKcF|Ocs<q=1Ivvsc$U(slt=mLJ4p$BmczAKwnE|xjKc$-3$rSJ_#}`f
zeR7u%5=(009Mz_Voq!A<8kp;}b2Q-oJFOcK4bX`c1)ozNDE&MHB6@>1mY|Cl76D~G
z$YAV3BD6=pM*Vg~3TA;Qvjxb&Bo%P8D{j9FK;(`C$WIsoZH|~VO5TD>kz&CKFZKNZ
z2G_MgClctM=YkxTw;-)%066}j_V6=slizVIx3WLt#?^UFpnlSJtc7(T{t7_^{`)+F
zapGZ%b_qgsSMy7LlRcR1uX_&Rsg#TBY=N+A5Bcz0G9Moua~ys-eimVF(bLi10fgbH
z+59CI6PPCV^idtkYt@+w?o$Ke|AWn40hAuu+ADZ?c!L1rbGmG*mz(v!F)7smXL6X#
zwk`!AV~|v%8Fn!aEp|tPq>%%lZ!Lh`A?0(*ULPr7Djm-UTetz>_vrd?9vmQy<^@^0
z@T+7(*&i=o7l5+xNl8iTLpf<6i^L3ss6FQ1V|%&T*0nG(xq-?=2rb)s_!S^`&;oCw
zfkGf=3BdAI_E#ksE#3;qcpF!H{#wI$ECUkAdSfXOZ%+wRGSD4gdH#KMMxP_UP2&O=
zbT-4l*!j}N%x|$dYI)FlDsQoi&|^ZKt`Hra@0Z+$D=I{qjXQEgg!{;_M0;W?Y@PIQ
zF7_9H-EvO=>Tw!SiBevhAQ!aPZ4xvxm3k#EBH~L4%)TJ#ASq_UG-gx)o{xlR6Hm4G
zb?sY;91`@4N5@g}??InJbAZ@Z1mwa%N)j5R#3isj7gnYapWnru0SSIgbaa_xc4`jX
z1`VJh0~+L*xbYGBN@crDenR1Ge6=l5FrY^wrVI6V_X)KdS*A<f!>k5U$o)ria_4?4
zAlq(UO)Mhtb9Su9VkE?hz9^Hev0r`82|A|JnKlfBY^lh3_%)|lWP8qP9sDJmU2h@L
zV$`nE3f-J2mj>ys5yKYD0}VHbK1eXRIXXDJvYY$a3}ItqZvbaN6%=?YcU+OX+9VPZ
zl<a4l>Lua?9{V>bDYu_tT=f6h!`I~7H8uB1eN`3Z^obY`T6u+>7O59YI&7VpDw4+2
z5WQ)v6}OIW-Zc^E+TfYE>^&h}W_C4VU)%zpg^?P^zhqZ0hidK4j@~On-o3nE!W^kj
zLq}%;>fVsD8+~3MMX&-Ox3s)$0kTtTK!I^u7zn<Fj$0ktXFLxE!(Qo*Nz2cF0Ygzh
zM*#kyN3?9sSNl>(IZVER>b<Ezv%tcvxU_PjqHa~a9I}}daW;&V#CE_Holw$BVkwY=
zyCvjW*oy@i5U^>T4<A=_(bP(x)O|Z|Qr^#4+hsqFA@nreE(yAEHSQgp@Q(Dk?~EU7
zTkZG~JV4cg;dIzTjTgi*GIW@j5M~bW+oWKyexQR30BZj=YW(M}0Ir)U&~9$-oe>PU
z1j6-H#_H;0=JhXuqq^4NguzH14pS3wPCMQEcs?+*IgpTK6h7m0gkIYpyT`7DJo`{H
zd5Kn2aQn<J$zuJ($`yv#3M(Q1T3sEG%NfiveS9@63>5`V%aI}I(>=0R2EXorpJhOd
zITv)|!@?)ep7lAE1y$yZ(KeXNia$XyhKc4b0RET{e~gWdHFsH#<nJ2sVMMB&_x22J
zgl=RAWE~uEnJeR*E~1c87qwziy=tegqjR^=kCM<k$vgku^SBfS%aVR{C?!l0E;S$_
z*uo;%s59{cDvV{`$ek_aj?TZvEf{xG!F^6ZwUsL|;w8tYrKL3oRouXyCdrgN&+h|E
zLjVH7ZoD8ZY0U8E&6|rWD~Zb>jVe(FJNd!JZeKa70^9s}Abc{N;1Jfn_f?pOSufU0
zfwWjd>bG(ilcCdAo@-<_y-hf(fOD<~N<ocg?4G8ZR>Jt%0|`9#;|EyVGcK(n{tT$p
zn=?<J?B3DvyD3mxdc9^Cx_=Z0??-PWy<N0|HH=$D4Xzy+F=9l<cOcU*Sf{~U)-5iy
zI3kYr+16GJ0dCJxEFzp|HRehGz&lVO)3FpbzK);_(aoYH&|T1t)Oxc8VY$HH)~}z;
zcOWBbxL}%bbi;SQv*3n5V-D+C!;j|9tQFOT6(E9!%MmjfzJ1Ai9(VqYL@KCo0E00i
zxbje?%ZCCznm${;v@$F;?fqsV!FhrW&>`9I+2Hj<5y*cum~l`5&Cfl&<BniN)zWGT
zd-@042q_q~7STe_MPyGaMlxCol@bnvI=zzqeS0U~4OU-VdLDL_Xkqs>7Rcbc+@^5m
zPz!xTLVU{4o&Wqtzbc_k%B9N<j%2!tq1KsUiDFl@V%x3??WREf!AC-kn+0F9b>QV%
z+}|PJR=lHsqfYj6W1jN{InB79TNheD-oAiD=yZU6N402gyyOdq!O6t(EYemTk{I*F
zKi+)y!5CR7?Qle38lZ4JE`@$~a6JGt^$z4WSpN+bBceMddf-~Afj=<53Q^smqoec4
z!~Z{Ti!ito?tBUe@ciFjM8cdz>Gjkw9%D0;Vnm0k-;jdNymc@zv=y{jjA>07IoRMV
zS<)%PEksJ?UuAsg^whAfzv}L1&_T8L!>{RGO%CT!m$T~EYZBq|wajH6M4TJbx(T+m
zGUXGmb`<@uKmGIb9k(0F0J1xvPd5h3d`29q<5`X(@A;*Z-avK1xQtQqqcv;N<u)fu
z<xm5is#-{c&g=Ur4NRohmp{e=^NR>CqozODD%$01*#Atp&N{8FwXUV}wd8E$r<eIW
z&RQJx#NNI|tL+qGQivkOLe}Cq)9(7}caN<~urQXMuu@AXZ+p*Dpehj0WY?ZdYu1`o
zF~R1E*FKr8?c)z*gi7&7WFrEgU}aUMlVeu$f49cHy7h;o#8V}h=fm{g`vX5zmE3TI
zUf#Eg(zSSsF3dbgV~9nMWrcOYFIxXCe9N;rObpg5Y_A3BP{T{5kX(^xvb4hEG;57P
z86K}E8Xam~Sf<o%od11(f%@w6r@(th;kgb&&POxM1sK9>4IW|XL`fIRt)15+5iu71
ziHD(UgzOnbWyDQC-C7#%6jB&{&d`}EYVQd|alZcMwT>0Zcysl^h@!1zfVNRV!}*kT
zzhY^ezk=S~1M_IldR40HFU;2N+!@!G4C^sbc5$gdbUiNx54`9_Lkjv;i!;AE^<952
z8HwiM#SE^!sIzh%lRtHZWwa&P>17kGy1(wf5td$-*5QRb@f}>cIN2Z58^cY+4m;)a
zTWRqxm|J<{<I^C+C<ZyJ2xVFMb#C0G=~%O4oNaYXVdoz`y;Ro2v*L;)$Cq%ARxxm?
z)nVbX#bV^VIm2Eu32k-u^q`$dj7=){yV@Ih6YzNxh*kn#dUdVY{mvN>rrL9mYhY2Y
z-)XyC#GR|#ByTYF{@hN9sV$eaY5cc^fRQR-8ny92;r))|O_oR7v@Ew*JT4&%UyIUl
zBX)Sa@NGVF{F}PTVbpWXX>)V$KpCCmns4lQQuv3?I7FYMS)e7OmEQ*CowMDQzx=zS
zY{La-&oiej?tLG(VnII+EzhfEjo)dGt2fe`|MB{@7c$Wn^z8~u_w~h~QoE_=z(ekj
z#4K%W%qPwZVe(@gr*IkHv%<ZLYk;|p$wJSaWk<v+oE91$&I|>L5kr9p>DRn`U_mxF
zH=E%Npqo1pwC?&FeA)4Ff6Bw`?!Wlh7_7M?JQLpfmY){as5&Z5YTB(HY|V>Y?_qMu
zwXEenpvYMbT$jdiAL#8aA#i#nKM|V4Se~j<DyFkd^iSO|65Z6r*eSv1c5&4EyJt6!
zeZ2SGGyiq86j3(x=$}BSgH!qw>f)Q8P`>Y+E0w35krW*1<J(G~E*wrsct5r92#xl6
z|L?{FPp*WPL+jkRW$bj(^e~)mGkTdm>THpyV_-}$tRww~?LD}Nx<d6Z&h-V~LV&{P
zd<|@*(=^#p)sZ@68p2;DTas}6vCPZ-kB|9*90M|^A<YI-+85eFDwkhh^)vfQ8+k3X
z)|Y$fyz!q8lja@gEy23|d)DgEnRn-@R)g%JS1k?0lP4l3@<{{~HbZBlq#K89whYF)
zsbx=K>yJ?v2oK-W9}l_AMQh6F;GK?)JTFC*6aOM?Hv45KHs&e(BmhH9Q=NoYWyeb%
z+`Iz$HMZ)h1`f9qHAQdY2Uxkg<Kp6n!44!Ezb`E=cP=?eg@MxJ-rD!PnEB}?h_7~W
zMpw;3_DXBUlL)`ZNG8`AXAtMl^vZS3L=Ij0oQTB$77(5z7L12mWU-3&Mo(>3=9(CN
zBw)n6o}~C&a9)>QFuss6M3Lo+^}EjI(2{XmZWT`RdHnep&k}6`4^ld93mVE2#m2-e
zV*l^@stA0bnd0eh5C;}+5~eZnxBjgmynaIgR@6Zn{Z8oyWwegcji>*={0|R`Xhh)c
zWDX6fSdSIG2F2KiLDp47SXe4Zoh-MWR6RdID+8~{OXu13oG>ulWHWHw1_uY70eAGM
zT8NEJ5!5E^10KZ?;1q*MIi*ZW$FYh2kLB}qg~$4i8nNd9B5r-UwlemWc`9JDLHwb^
zq}2S%wSquWIdHl%wi{06T&L^<5MBTF0s|ui9!puLudnX`h?-&W-S7fUIPlg0PjutP
z4d9N&C4yXG*Z^Rem+W}Fw3+`mzCuvga0N7D!G?u(Eh&kMe+tRFR>$Cdq|VD5h$}Tp
z{@;IdEmBukK;ZS0pHY@z#u4;<sHMpP&mH07;__`2bH&MmI-adF%MR7JKyQKaQY_Wi
zvhQ@U&$(n@T~sZ1yjT0{sKdW-2c5Xmgt;|@Xy|I)pMRsaQ#0pOv*_CNBK&9dP7Ma^
z!8R>py<`2QYr>vMYU&bZ)(BRzL(P+$x^X>FCSakZ!a&7e_ud7yeDxIiM7KZr?MBAN
z47|MKOz7h`A}+xQ>1?|_Yp$h9FC9!qTrJ&1gtQ1|Xv+qzoWs~wXYr)B4i=+t-ITUo
zv@h<*q=(<rfdN@sSvjq_8<AG0RCg&n=*Onl(bZ+h`Ybst@4%8EUM_3KZQ))<dOEaP
zvr(uF93*d@2UqI>{<(=|=r(OQRKEf)GDz#V+PfO?SemqxuA5sG=+@%aHo5TUj}cd#
zaj|+GeBFJeeFay3!HLg!Z$G?a)c>*X^n72ZHGi4WzDX;6f`W%!u*KWA*{}jL^u~~4
zkfqpZ%`bMx8f|}X($0lr=zC~Fn(#=0+7wDqN)i?dm=!0L^a&6jX-F_2Bq3d2BALfo
z=~(eg`l9v;PiJ1@g)Chn!6Tk%n&`ofZAG3fSFA*Wx%yxJ^V-{&6!;SJ+jmV$tKZaf
z{RJ}iYx|{0hN}53-RQ0i5aW>t(^QUT(K!HwN7P?TW;#0YBnY&7B54l_3W6&rd~a@P
zsn^(ZU%GSm;GYe;HbKSzlpK+2VNd4e`9a`JBA4hFbMscp{l%>C;S6KEdZ7awG^+1H
zKy8)xS`i+#BMOKEyLCN;$YFdbqnSQ=$dw+ohWfuD;xXrEPSS`(QtzXLPxu{7;ZR+;
zHQ(5e{pMvm9^53niwhGCW>qS++j=GZ<RGNN@Z_ZkU1~g@rkst9zI>O6529htcxCbT
z?5rN~rPS#mX;}A)+htNhLIJ223RAMPv-3Zp7_cZ;&HhQ^dj#)H62Y8Dfrm4P!Ba!v
zaDIKn@$(ix0abr;Kf>ZPrG*^^txqQ>o^YFIXNKO-H7H+IcL^S$B_Ac=KQbFWq^Q|j
zTX|JB8+l^(OBDCzj+CqIk9w=>PYoP$8r9XfrJ4?}O`QK1mvmkblO3vkEjXO_fG#}!
zZ}$skAvbm5t%*Ea;uizmv|X?C7wztCxn>}wufs?fSPE-5XOy9q=^sBH3xX(`u9iQx
z_UDnUnKY&(57itEjrHN=!nVV;9IoL!VqR1mu`F5&Sr}XF9i6CIq@wJTgb<VnS&tso
zQ%puTYwvQ|_i6{l?_AQrsQFy`vI~p9;xg8Bt^|)TNBUIDE?fi;PBM+I3U7NSx4>W?
z^TCt7t{p(w!TJCBj&Zm|6^O_c-3{`qRRQCO21(oxP&X-gqQY|Dv)1Ph<-Z78oq8=i
z>m^z|Z*xO999CIoQAp%Eo9e=uYu?e;<FWH01XsA(?P*LUom`0AnfeWr(#(5xn%f6!
zidI!x$zO+aqdl}!tGNjjet;hv<8#deqvG4YLf5*<_dd{*l~1O0N{qe$S~L$tb{>Me
zd;D2|@x@1^YiV+hPI`O%)@~LWac+KtZkgL4q##X2T&O4OsUa9vq?n>M>jb7RXHFez
zv`=Q?-mj81!XG3}g5B`8*Ab4P5j>cekm~*Yyz$My+s(sciSL0<Lj7S7Eo#^%j@xzw
zkjO~4&8sjxcv2++F@PaDQD@uOpi?CY|KRtkL2p~hmR6bnmCClBeT>cd{3Ai`V?1=o
zmRHl})JbEKkLuYMBc0j&VjJ7(lA#s7>GYIOWtIgXxo<ULje~Q!A!SK{_FB7S-(~Cp
zf%a-`7GUK$E+E&ZKHB8Z3lNvu&tV?<(%*6;iUUx-%d$g-;vgzfsYiB%+Bb10nuy-N
zjG(FSJO9HTva4Om!VWt*3GsK>i~VOVdXoq8=Kuj$2%-lI8|=PP3(<{g(qUNTW}3@v
zGGSO1X1PJ_;jfOi;TE&$#nHKWRrut>h0Lb}P_~B3#|c%0>azX%C;wdOEwgiqq~v>^
zhm4u#9g_mrL-xg!$k1-W`^g~}EzKW7uBf7rOe*ow^VLXec^WOO7Qfvytqd2W2T+Hl
zk7@Q(hQN2mwgh;uc}t9+jrVtMK7U3_t?UvY+W)i(yt}{n7>9v@GQ{|HJ9z4twnI8Z
z=QGWknL|mt;lJJK{Nnpa<bKl)0PXX=YTFHs$M<MqPv76Tx}@nnW{6G$_}kJX4J_sL
zYs!#+3&2PZP^H*yOm!j>(%iHdjm~0E;yskCY&6Yqq2@4I2HeeAZU$MOV@?@Tj@T&<
z+;h8T#W5fS-m3wshE!M^Ir5bKozo9#JNDl8fhi$ftm_out)mSvLw_txi?tND5Y=5i
zZI+%%R|OMcVdwhn$VW`KuNNp12RP;i_o@yazy>Y@Csmf&9<Y}lwDbz%7O4+aDofZ#
zs-|G7zgVXt%hRsa(X*KVt8HREoAHpB7h3PH2zn+eTCe`}pLoas>`S;2r7H+;6n8$a
zaBW=P;#bIIOVrzocGz?gberbH6|UAF^PkJn!jcY~Q7@K)jq_#*2~V30N=vIrg(X&_
z=@29ToMXmEU{P9e%QVr;=SNausN*%)Eh#zX;bSn!Yy`KhGI|HH6Lj3@SyD9cCmJmK
zf@;`vpfy=pjiT~VESy&S<G7z7H_h*wc$ks7w$4&qlWHi#tBQL)&~L$@y*<gamj*!)
zxm&%G!@kq_5Q(PSTpus8m_Nb8yhG1pGo-cTZye`->xv3<U8j3?g#pRs&W9@jjG_u2
zjhyWr2>k|iJ0^xlB&2PaoMl(n9_5&CX`Y<rTqXuc{gZ`Iusb|?m6wi5pY))DFh_@B
zXo8>y9iUgX*>`_Dp?5G?x~FS!h<TET!##IKvszsL_3K#l!2bHk@-r$pO6%$S_b12b
z=qoc-AmkxqEp#r$`a&Zmp@Tszcse>3R9#Wc9}!FP$q~cG&vS0UI&to{{ECaW|9a_q
zmOmx0aKG=@o>{9A?3PRUV&*u@p>&25{T0gdBFB*CTlp4`DGQ4*@*b3ymebP~ygb+f
z`$-o}UZ@HlVW{$${Tlpj$j`NHwH4VoAR&B_?y6ci%+EBM>tdF@u8I>0gOt2XqVO|J
zCBwRXd`Mknp?Y+ZClN8>;^d!y3-I(mAfBQ8tlMetr&)$q_AmNEm@WneWaG-`-3;cp
z;CUafccxz6R;R!CgA83bL0$pR%>vbW!A+rijq`l6CrjKt#~ZfR<B-U1%w?ICnQm5a
zVSRcas<tLca|kNfXtDTrr~a!u6@ULJa<B11^XkPLFGri1!63+_#Ys)ALd)|*;pheF
zP<D!1z*NE2tM1H;LRJpYZAt~%&oMF8y>+freK1}=2V!SzaJaARWR#vtP3^7ob^f;c
z3k&tqOhni9vSLJHTsaNcEYXnkhRb%pJak^bS?dCnz*hjxEX2feM8n%^4#QY#s^HSc
z#ke(}4>qm5U!F?zXx7L`$i*wT*d(6yn80yxRIT=xN!fNh&^EKZqsFR|F!QyO<8Al}
zwN3q%h8y}OBL>G&Zi(d_u5^?@Zx3jMrld5mJiHpk+zl{H=(%YZx9#v?IGyEE#g>@O
zWR+`@k6x*&t!!M%p)9FocfIHkxROdmV|NT3zn3Uc@6van<n-J-Oy8rE5cyNuJ;#<A
z%V*_E=v=i?Q+a`0A&QC9?1`_mv>eK|KKwdV=YPZZ_Vxp(8m#t-z0X%}`rpP{FDA3?
zTi~Al-R5(JzcJn^+^)k=kke*~);(S3uRmK-fhl$6D%R9H<DqR7<k5K_J80G<O`@X6
z1@dhxQGF+Zp3Z1T*7g3^+^apO0iD1J-J;k@xz!%L@#lUBBLlNe*ifb7t?99n90N-+
zCO#T<jpO5q=-9&9qpeZ3xQY!0gVYnj{kjX!1xVyd;Jmf#ZLP;j)ow{N=6+~@NVK|*
zc)G<hP%cZhk!qp%_0C-d2f0En`oz9?P~XvTQm&p0&EKP5Co|i4<`+9!u&}XrEkYUB
z4&-9!7b?p?@zuB$e>dv8SO;~iwK~oa#CJL0_z@|W)`+*YnJpjckh0mP%p{#dAA;%V
zH#STY>z;5I^2f&Kj~rblIaqUXhI}2$0!7aQEhUm%WPrt42iu4|aekF+`KZ3$A!tUo
zT70`Vz96Do#(Qr}k_x`9Wnb%(ck~eXoQYeF`;~3#QQ>ON#HB^ke?*V3z+Fv>`@{ON
zbAzBCl?wYu6e^Vxl%oSj?NF|9eT`7=wcsic`!uiswlYl~I+lI7(hVO@e!^U{KzGmF
zRWa1=NQDl@U2c_W>^hxaqn{F6VfJO_%%<&_smc$85Z2kx6qhEf)>SJjqNbdc!N+Ic
z6cE=Bl;$;%^y!&|!<2pEvN{?sd=eQ8d4jVtE&5hRBO>GSpe&ZPd|TV98p*0t+j}i}
zs{u`;mg+KE8Ac`+bJB?C%uE{s@PAP+EFi8sCFx~Og6IxY_BfeX7g4A@PqE<F3Rlp#
zzI-wS{Ci`g?)eGXmat1cIa7g~?ukP#Slv8*H=>@$Zb8p(-jwn}B?BC(nndXU^H_wE
zeqTU{hWg0qRo1L^JEbrin_+F%qs`o~!gH=T8B*6)<Mq+1_bwD)M)?O#qbw7yWEErk
zn(6VhUA8h_TwI+wEEulWUGt2sKq%^j{A&AcQn`W7nT-_nCkh{28>-Tnl@4puol#bN
z{XT<oAtiWgh^^+6!`|L2ivdF0Xmx!E3|1!T{G>MbwmK*3Ya%VaE8Y)TI`5&%%6yYj
zp&#dl`rC$Aw-srXxMCAQUfL_`eqd*nqn%jxc|u&igbrJcY#K8~*j9Sx)!}Zf!1E$4
zOLLsdb89lq)7XRNP1%WrP=k+Hcdt+s_~4=)pEGpIl}k#flF7CX!Epw+3bdZLn+#4h
zs<~IcxxSbC(<@SG!e_PUN{Rf49fl*M)MMJe+_|HG^l*-dip+pdx_T5hf3@60?9;$T
zxh=KViL$e5N)){@|HctP&yM}f2X{`_%R$&XfgbCX<ullcHJUWU@|qWjrvk@)PVDsI
zl%OLtIh+j=fH1+6gA~#`ZZD{xuX0aS5DEqcJ~SxKHh=pOD22l|Dzvyck#5s|?_hi3
z4sW!bKHBHn!c$)}wo+2AM=S-z+9wyqm#WNT=ZdGlhVas;EHqW4XmT`3wC={|&J|BO
z)66Use|y3psPq29#9#ddWsvetRRWZZeNEXv{sedsj5)z}Wo4pD%gWQVHT-}pJuI8{
z$%0U4FmInGGt+$C&ws-GZ;=u^JG|1vxwXQq@U>t~$1Ktf;s$(U?AfW;d5uLmJ}n)?
zm<4>WrX0PR{N7?RTR*4m>)m^iG#SeiU+V>*a!79k(=t9~k_K?HAA0)KLM=_Rc#B-g
zaevD2lYdk$RvNA#$!6yAlDmLvlcCxy9#c-xlV>z<u}exKEM1@X&1x^(<5E_B##`jQ
z_M{?j++clY#}X}F8pSetS_<{eP6Eo<frxiG6${O|8GCVMRyEGMF03zG>x2)cB3PYX
z@QR6<h60QG6*8ZAt0#2+5oj^xW%u6bdWnZam*bDhxG&5?PfNWvxV+i6<&WcfYWDYp
zPX6`>7;`1@^W>>bpWi>)q7TDE07pKy!a0ks&AM+rtra?`*9o#99Z4xbnY)I*pcmbv
zpjVP3D=o=WE{Lt5<jDrt2(Hsw0;EaL2@e}e)rE9yBh^xd^AvTkgQ-S?7mdn~o(!?L
z&FDTy@?MEz_FaGnZ}HomRoQr+GZ#!Wh=yIcK0^1_b9^FS9-qb#CdFiA1O%IHy#L_g
z^r!t~=ah=mc}V%Ib$^}8e!Qf-+GyzI@Jt#?x9BNxAd9WKyeu^}A@RH6-F&N7;OIiR
ze{UCzS4$UcUkXoq!u7n*R_`EYO>Id|dU?Fh+mrm8Y-&k|o$R;F!~#A&9|cLALr&-J
zbKe0HLm$^2pO`egVfH#=)|o14)o8aK%udM-2-rYSOGL(p=PgcYPxG#D7YIi^hvhyb
z&Yig)v_g#0!2CBEW?rYaWFhp>+Ppg+dOj9?a&nS4B*3TyfmIU8gX^$bz#9m}G3_l8
ziLs4ba*dng2%sv#)<;nUt7S6l4L9ex9bn(1UVAvG+r?$u&+2*TNw{-S=b1#aw)X>n
zhVkCj@+tJPd~<COp)E66VbM7{GXr=g2kel?R*AH;Sj)*C2Nd&_oGjW-o0a8Q8rl@j
z)(sIZU`_PU{^04t+AlCkK%o>pAHvAbDi5p|`b}k^v+JZP?sF5)jnHp0^~a+Y-WB?R
zLZ@g~;Ht@YwFO1$VOmc1x&NisZ1fDFF}Le7@hCoRwV))@SmX)l${6Q&*~p-U%xx`V
zYlKfe(w!eciqnHfdmO>mhx>y+QPJz+Nzkw|UX%q2<t^3IrM#laiMBu%0xBT!;+)6<
zrb(2Pyz&UqHy|o8WVwEK+C@>?cw&#UXfgvA|IW^zdPEhfDm8@C4XRV6<jswg2wo54
zEAM>OTLlNnSQo{{1Yd%d+fdU7oochI3(j$`N+8DocJv?F-k#w{aTOg%qXD!ID?<R|
zwK-o(G00%SkXlqQ2*axTTk>FVzzGWA(sUW^S)a^)T|8e%<NU2KCCP>|xzfW`tzo*b
zc9t|Ra<U@HF|m6}TZeDY-gspx$1*`vlmEcwJWDjrphc&np3m89cIIR)^9C-#oxWUX
zi{H6x!`8z|uikox31NAT5-f@BiFJ^jZ5}RZWs8%k$uZ_AAa#hlQlfyPIGy#>i#yyB
z-XNEsW>>!eB!kU#%21s^(B{<ZK-A;>nbwcBVMbwB0^3t-{att6{!7XJR&1SpAkEq_
zzYTg5pX_sEQj*Q~pmf;57*!w?Y3Ul_@on1`)qm!VZOj(O>GMl;hPmjb<#f?|Ea&aa
zkKHfoE4<)2NLQGLOTR<L(vtZq+?(~sOzX9LkpR$)wE&FLLBf&K3tG77giiXdw`Z58
z-%Jr53>d`KEg?ap(n;qHOIt>I#NVKwxZ0OzW*MfYdEP<mv7XRA2Alw&5fsPe&d=iI
z)?p1Cm!D<TN}*CagQxCyPL}j&9V|GwGV5uN%2$>PdDi;hOi>MNs7`4-;k0=I7KAcH
z(0ycq48U^I`|}_HgS5Z;!whOL?QI+~h9+p8G!M`&R|~iE{Xd<dWS=h<`7f^AlfGn;
z0{*CMeW>vw#Vf1y*GeJqj*iX+<iier^?V>AeXOnv(6sFo_glE%2TmJ!e8X5-xyp)9
zl}CcmY!!UCr3F2p;~F9V{%RF`9bW&$d)&QkIbzZODaeADD21SK6TbPiuasmar@qj9
zLNXvc?3I46ys3cAV8gj)jeFn=AD@%MPZt@`#qT3PBcL5Z3d7O^gyI!C8N1r&)O4ez
z2X3)ha~8RWD<xQR#T*=$pA!-hW_QPv*Yg`5l7u9bJ5~RYui~yH9n90p+3?D=%xB5d
zQ#04lIN7Xp&k&mZ4SXPG^3y1w<YODGvkDU~(t?eQ64B&YT^R^2+d-n)ZY7rn9Voio
zHV-&6&R4iBcBC)^AtA^B=q-Y?ldMF&x;MNory9tPvZamC<G_Q1X5%Y|cA?c2UTZPr
zp8NP}@^klY*ZDTd#P65VtYx60<BQ?V<3DKXUB)x$+hdtP9_OY#(u=EgIxJa~j5GKS
zt8xK>vlT)!<tblOm}V*z_21H(x>Ro1z`zyiUSZi%%4B<X>lQBerHA9h0S6`bZWhQK
zkM`76N{y_&Jlsa~`_pBbKRn+(e2l5dtm(B=#B#DIeSQ=|tLej|>yGa_i4_glM;Y1i
zDzWLt*uu-Ry?{M(e01Pe`mt>TYLb6WRL%9vFae@2yD0fd5OnTYFyVC)Idcxu$mG(0
z+wXZ!&pVhq{=o@4wk+55+4C3Dt%O~9nU!pa9M8BBp@wUt>pNf&&iPZMWPh*h-8q)O
zCUZ@p#sYiDV6H_3pZKVPcC-Bzd9_8#&FIkml^Jj~m6bLc%kDSNdq}5BV%iWIh$GX>
z<n2VrGqc;!eZ50oD4Ygut)d6iY?nZ<^`r7W4!mdvPHJWR0MTX0f3t&xM6<y?FBcPg
ztyBD7r+~awLF|>F6@dPG+x@pW3yE}=i@#+{OpZ+KFA?4&KmV#~JDc;xp6nHy%?BFS
zzQhDu*1Mjx3_MRn5O=GkL-1>w)M(Wnt3hBbW~u5CrgXCX@r|nM)vaCVsXt2qUgHwn
zRj{y0jL$mLUjcLsy3Jd3C|Ug8EYY=LdCqZD_os*pIbXhZoUAt)=kvk!X=QY-o|-*Q
zd%4X~;<Kwji;B2Q1@Ho#zaG9gSc5kfUQIj5^T_>GaF(i#Blbu==@5iAi;sQURFkb4
zQ9(z$fv}zo%y*lr+x8`w%VNn>lUU}k@=#N%UmqPVTvRETEnV@$bEC}xK|58ioy04$
zc(gGl6GZ741o@>Is@>URWUe@;G8g9QiQ4GII%B&SRcSTVo{f?I&3>fiG)gcQ5ZKEF
zI$5G1U6_|z8RC5QZ0Jc%=85xqHkqfB^2R6QOVMjq>*vuG^7tZb|0Ya@{>9KjF>V`A
zY`oH91vULj4h1QrFo9^NpSuz}az>)OF2lqGwl`x~$;qGr*-k3b%n8QEZw|Xp84{n9
z>PQ;Ia`ktEq$6jNPlSMFy81Ow6m{s#aLqsaIy5`b&j|qP)^yWd>p^C?ex9^3RhmfC
zoz|sFA39+Ka)F_CaBUoeeme`B07@wh+C`53#@i?Sw$}7mE^BbVy)fO;eW8i@M(rTO
zihB)c1buu$dTqgasZ@7vF0F#|mC34`TFs<p4qy^EchB-#Q#BPsYtrEmlMb}jN@{1W
zrg1KPSfI|-Szmh|dKanILi|nf<?`D8+fEt=A!>kCyVBrwnEP)(McIgLY#Mi=FE**l
ze2#RN#;kzW^{48xg*PUmwrAxDv^|}_v<Ke7QFQS5gzvdqTw4VXd{!wY0TT4_vrOHr
z6E)$1K0&>|P)_SXInb>6#2?xT*H^EPU23(%&SdBS*|-)KsJT8qX{&fNl4zTGg~v@^
zXym$Ej;_rWFGzKCtGqP>`E=PSO#E$Tc1RS!dQ%JTsy35_qgg&#L|T;$)bv_%S<-Z{
zi;IgZ)&=Wmv+xnBuHe_K#G|mrO6$@5j4n=Ln*@AHc}N9#34njZ(|R>ER9egT%KFa+
z`}NvU&P~5mzz42){qeL@J@>vRFy=4^L~}kddAe7C^67(MbG>FZ?^?T1($TLYWJmb@
zbBWVPUYkG<XCgDu^Lsw6vz(uBRqs%!E)btQjK=u|-PG>>zOuSq(EuRJxaNNt2(Dg2
zH9Ly}6tNLj%2i<CYlqx31M|Pqn>X@JgvBs1nV7lddAZ`Q<a~aCe%&9CXzd}veE=vW
zI)Xyk<pVQmIGRp#-ziZ;4OXt=N^eY}DiS24U`nH^V%*7=lT|4|{cC!yekNy8i#gq?
zIY1gdQf|hunabZr|5?#vXg7hF>R7rh;6<%^fJT)<_KwtX+Q)(rhBYbY&{3M6LfUF<
zMxnLqnc|hx9lJl;c{Q}sR8i50u<$%^md!y|Fb~&dt1D`0`);is^G0OD(QeAGxA(l(
z;DA!8wBsyI>PQDiytB8Gsl{ZlkXqu^@}xN_w`GNH)q)Eo=)EDEk(k9k;jxpPc=ng$
zyurkJ-ue9n3ZdvA7d)~(lZLK6S}ic=a1yO7Jh|MQHg;1P9r?L?DKj<w@Q;}<1`Y~U
z$lRxuG8D-uwjs`deiKMIakF)r!jJC#Z5(^o(T?Az`VVxUgu~cz21+2LW8*<SbGrl8
z$Y$x`qtZmkhh{?n?_bw*zb9vpKHo+>c~v|VuTA&j<numv>dWs^?iHe1bd?8LfpK_P
z@Hi17mESy%lYVc{-zWbnZ?N8;w5#;%c|+{^-4jhNc@9o?gLmyBE}!BJCwm@emJL`t
z8yUIar;up4@$!z;G$|o{M>d6A-wQu0+=!oJ4)21-odijJN83X3IDE#Vth<kmF{L7L
zB{-CIQG~zEAgb8O95uo`-(P-jTv~tap8MMw?Bo47>!+_{x4El_Ow1_XW;GG)sS%an
z`7cS*!yc!1*@Q1)u;w=z#+dq#w5)mE@gH%6dDa*bEv%*7bcy$UW;hMj7ql(-*xWJ8
zcK3ckT`<oFsyQm`78lPB6V&MU{j}rV3PrU({dJpvXK~XPl!$*p9mvd`db~DZ{Or#k
z^Ya6%ix}bP`#OFbf(Y+NEPG@0Ny-XJU-PPYS+UDaBN_=BxJD{birPLm_|q2tqTW3A
z?A4k36?x~vZG~Cfc|&R29$y#Nu=@-YqE$6Wg%C(bib=Kxe=3Ru3R~J9$Ii~NM{z%q
zDQY~?vAt7~dA$FTc3#K}zd5;MTUCJ`B1!!F$5N`ZYqzXqF+*;PWNjrVx_r36lQ1C;
zc^Fsh1+jfA$m+FwwaE@aJ1n>OUEKNxAKm`fkp1<4q^teskBXj5-Ewbjz3aB2(>n`&
zem2$`!&GWgWTl|z;ZZd$g}SLuLc*TnPgP47Vm$e)e~grD|Ga`ipz@%W&+?#w$Ab_f
zQ2vtwi`px<V8(Z-{=xNzwr>q^jzgOy!F|F9Up~&ZP*=yq&Cc0lG&KGv=K;fEsi<(W
z$q^(u@9LedJKlGQU?f61TeX*)n(nDWqjF{0QYb$UqOz<=T2?13`LV%}(quI?g{I25
z%H3twuJXfeBg3Gyav~xJOsYh_tij{OsL1`LaXgIfxC`|Hd)8nQoS%LVO|yqmRP&WG
z3{Y%nK@aopY`%qnI_?nf6;^}vtoE<-w|LnM*#~cFUQKWnM&sn*vX~|`T@4AH;(R*w
zP-KYauxT=o6Bi$`)D3vk@>tOJJ|=BO-ApxgD6)4B-Al2X2S#~9d91FrE2Z#6EitBP
zQs={S7ADiP)Q~Z6lvjNlFBMVi=j#2O?0lo|t{jC(A3yNy#*?wuwP2!t7dkqDkxbh5
zOnHOh-%o$6Za9J+@?NP^h7k4_ffY$O`U3%=2(NDM>GX%{vu-C{eB58PjMaGSo{P4Y
zJ#jjG4|px<Sg|p~^pu)^6A(MRI@|jY-|*zl|2eqHw?gNLKGLNs<^0$kSDTW78GjC;
z$Ic9D@9He+brJOv;N^8>4>rlnpMbt6xuQl$g^LWI+~T|aARJQr8F-4%?Z$z~SYavV
z=Ax9BkVvG0d)}KIN$go2Y+!cC2s58_QGStU)yG30uB7gybdGZwf@sL*Je1NFV7=Vo
zV4NFl5LD+YA~3<i(wuy}zP77d`R?MeAfc~uTxIb>w?=_A7jtaD?Z3<de(2HEb#IpB
zLTi5SQ{BBTrk4=gtHq__G|M0gEw~!*Gp2w_{3NqJ9t^4^zV?oE!h>|BL~?dX%gsN+
z7mJeWJai`~<X%<+U>}F88R9<V;E=V4EOxk9E99!&^>(H7&&bf6q%kpJL#H7Xo=19d
zS5YM$ZoP+5aM}i;(dtvRWR1Z81Ffo0vh*H=M`f1FLe(;DKPzV>QdIWT<D#r$UgFYo
zC>L1M{2UtT;XV8Nw)@^a9t!s$SGD}1iRH%UjJ*CaNzhfZ>!r_4-P_|t)XaLfF@BJ)
z?i*j~z91PRtA5k8NFYqh%dcfg_P<kidw%}egF3&yw^wR$`xx)<gnF^bP#@NCJGvOE
zFj%O%FG>4gx!~c=5s+;51OxW=wTiZ=uKgC-dl=0F)dn;cI6(P<b-kN{*f|PA^<iq5
zUiTf|AB7|||NiNV-k)h1>r=3d+X^IpXKf?*Ru#Wx#P(@@Rl4p<DK}T5eVofPHO82M
zO7658i_2G+yYh`9K;2O%6Cs>Y>IjG}SJr9`i7i-@eAC2xzhGqSZ`pws&Tf_~#XEFl
z9-1^wfzEI*N{75K_lu>4k)*wbd%-S5$@@u$MdDfL&VEK$oy(4Tz3Lt#oRK);&iumQ
z%6BQ6=Mhp3e?1Kgb~BcUyiSg=Fro;k#*&ptJs#k6VI5`E)-o-zlgLB$Dwfb0NTFHT
zI}YCZLG)N*+aAND!n$=MfnP~ovvX=z%`zyV{(v9oI~*7LKUpc~{k9@G{`hkE`J}wJ
zeKX=JM`Fq@5bmz6QeV3|-SaHzGs|<TZzOX8J%Y^tiQy3VRlUk%OE5O&|DF^()@utF
zN-%nk`P0<f6EZMP7qriXJyJ6TcNX)3;mlf37jlja_eN1rvHTow`^2eM*zY_%yIDmN
z_(s$no5+rfGvO&@YjUy6@tExK-pc9re?wD^y}U7`WcN!p$-2J<@g_yR&WeQg&RgxI
zn3hNv7VvQ0U68Na&PzW$i}{v#!&?1lW>T~LPEhc?1@z|G@+T~>5rjlSo;>H(6a)%h
zbg(=(>Ul3pYN*0UL{$KvO^b5j`W-1{vd#M|ow@e&@6P@)@l1ss(?w@LpUd!Rwp~e@
z_0tGtmF~T1Ps;=()9mH#pAW6I^eqj#k<l!JhCLpKt#rxxOOy$cQYo`F(BFe@cOcj{
zw2Hz|j?|pRUK?vyPDIz*%39N}DCb#ZEs12t3z9xEU+XNSg3si#Rb`-@>;keyqL_nC
zUvb;!9wqkN2g(F{c8;Z_1^VjPAK7f;g(`bTp@taX4Ak}xXT^Ry*f+@zdF0^l7oCS{
zNt*+@ViNgr&cX(0IR1^H_CGVz#avtlJ<xm6U#6g)@m<XSotBTDVpb(|cVwSD<p`JN
zPTMW6Fi39<={>I_o_Y_DMwBcqnlD@C?Bf2nx+4AiX!hm#AS*A){SL4&MI3A?Nm=}q
zQN)iWGtvo(9G@}x|7w1Q5nEebj5@Den_JP^<~^>i{$6CmDX8Zez5(m#hIpOgtJ~%J
z5i!^eT&n7%q+uGGZ5ao5-*A**070&MX>ohD;jebWIqP^CrZ35j0&tTYYE2GmLBYiB
zv0&Gb_{C8j&mX*IB0)baVX!Glxs-%7!IcZ{py)%2<;opNm#J^Y?LF$6W8;0p!<eQV
z%AJc24{*Qo#D9OQ2p<@PB;`XMa?mNKDD-WF+_U0!jvb}{R=_3~BP0GAh6Bz^(!o|!
z?X+J^u{Z0=Ap7HeqzieE%-U!-g{!n9jie+4%d(pRE`!)QeA0Sz>#yN~f|qr^I+=aK
zgP3n0zsU$BkXwE$FU_TC&1-tkr^D?lLXUmo?qNss^vmqw=7QKgh!d(z5c(jM@h9i7
zpg7-f*wwoG?%f|mV^JY;3C0^a9A!2m`QqJF*!U?5jBCrwsrzj{$vz>6ZEdBM7K7Ci
zO;O?Nu{}taJK3d9;?`E*g8u4Fj=T`0>O34YbvM@DGM4<m;NxRJfaDJDn}<9PGG~66
z{Xj&^F)8Fh<wd2flM`Q=?01cb2<P4zXK>C;&m>aodAId`cYHBj7a4v(^vv<xB&Z7C
zr$K<<;PEAyz<;+W!h}bbP2T4Jjk(t@w{y5RE7{SjZ)@ch#N#a}SW{-z6zjTdCL?L#
zaJEN;0qi@sn{Fl9{6<f$gq#C!;c^`Yz7B4!Vf=OgbYN-|(tB5{F_Mhez4F49TNG@p
zIn~dw_v?{`0EuENKkU`67l*m?nzXgK{$ic1yXR|zDg>a&(9)`Vmn3}Li#nEy<<rBk
zu*~8M+oHgSNEfp%Lu<{M2}vse11iK@cXn$UkI_MDsLyJQkw%M4`l$%{Srf{8RZ5)q
zjv(|O$u|!!<9Tw}nml=4)iD;@ADLAl9#JLmb)`t_*9luo1ublZP}+vsd$?5@=$lE(
z=|5$!CKsi6n0QGS(2VkGaOP}A#$~LC@)DBL0W<SIRd@N|@*G0rHhX`M$k45xXdcN_
zJeTxct{C~X^yqt7q>icIqajXrNWyueIon3(H#6h27kp7eQ{;=dw1pKFSJ=2`rciM|
zm#bQymF{Rw%*P%FjDwny)_<z`Su8K#gQ4c6`w1UHxspv7MP7fYDYV6_ZPVeWB4bK_
zfU{JB3-XOhhsY$ReL}RmdvH8G_%JpQGw*D*Z|)xZ{~#fqGqvtvYI+a-RiAYy09Ra8
za2e5Ng~f&kHpXpGw2K-VSbU?tL07h8_y$kdp|kT2vYW1;>UM^de_%DqjYAinxVYgT
zj{CVnsY*%~<kG*p-(b3FYH(%Gl+q0^q_2#~*oh7b&VnC@O1`LB1lOpD8zy1i$|0n(
zWNPLptqP5h(XvH|qoE_)@9{YenLTl<lwX2I{;#lJ2)^}G`!81H_3*NxYMXcvsw*>>
zz?!UJxNW4_?}B22+XLUZn2X!+@b|pzK(tI@;fkg4P?cpCeds%5BT3!r<=eK-&#Egt
zUu8JsWtJ7rwy6vsW!E4lmk5b(zO&A+j(DH&1?OCF1FHNpQgfp5oH5rvpHII)x_+e=
zo+NKg)(%g~GNx{4=XfQe$5?<KXJKKw^|K}vdGqYDT)Mm4gXTfqF)i)mjSWt|aXs&8
z-1X{|)c@_g-_G2Z+lU<n#T*rFX{kSEW5!{nB@E9QnwlnXsUAJT!gl^Oo@t*qh$1@8
zg(54)GmJTdjUysTHs3st9w`0I$lNNz*xEYvO}}AApudpwlWT!n&kMz)c{6M8ZTiAn
z8A=@$gHO+#4<2`DXb?t6F1#}qlO`b0rR4q^MR||u(L1YmpYxUET+B6sjEtoe*{~NL
zenGO*<mKPhGBgx;!MwV%7M#?RFe;I$@LdhkeXq;R^m(h`#f_e$?Bq9XDSzJ<t}KQs
zD*1hB5Mo(2adLG{2`_!9)RExNox82*oK)#pu8fH49e61x3O|}ppRYc0biDDT7CfA5
zq#9(Hp-;q?owOaXNe*NN7^q?Q__ICSMJeL~WJP&3@0S=2BhDsX<H)F+ea+MLueTsZ
zMkZcjAOF4|kHa`hzf00o@s17lBSU7kZoXuHX?8G$!o+5hPF0qE9tAhB$v!2|OX8J9
zR{X!EwtIl1JD-xQlA{hCJf=dN#sRBR-iF@pOC(78@j6<Sd3#}VeObBusfwG)NWrBG
zA8(|o3l$AlW3srHnK&_2Yxu_<{ZsM8)4j-J@xM)(Z;H7p+8v$ZjG=S$t9G-?dQ|My
z@sFa<e@=V|$WeK2K`!iajN7<55mX;ZJ1Za@XMTP%l&9y(B#kRAjf3oN9u-e`E}ZzI
zxjFo4yqsJ?xsDdLr_kznO#elLq%V`kI`Xu~ay+|d^7znm2-o)qRCkj&;jVsHkQrm7
zRM1v@Pmh~m)vZ$gvKDncf|xf==y>gKckkL2mlxch&C|iyz^9mZFj)|`CqPV;6@Jj`
z2|CrbvJRD^p&=v;)l=-xh9icN)}@+>yi}&9&4vbGL<U8bS-lI|g{19W<FEyLTT@e(
z_BMN7>OdUkGbc(YOI;P_q<iDRm)lg2dJLzTYm>htFJ^vz74)>!>9Eo_XwIJ(bvIo%
z+I+$6lsH>IxAjNH$$~VRqjVL8(sccxsUqw+ydC)e_G9NE*pK&Jzq?HLB$DhH|1e_Q
zz<0TOgwCuk;EWO%H;RtB9~;`yyUK$f@X3+q6%@ZB5%jLZvD9he<)-fAzNi1?ylqDP
za5KWiM51_qo#b&V+yDUW?*48nK&^mX_)UoW02h}RFd!Z2D)?=*e~w#h^B)V{k!Qjy
zk?L9{>BEv%o>r6MD&JRByz8~$E!8>v1p%z?r(?HyQ_CR#l&3MY8QQru5;2(m-G4VB
z=>}lq7G}LPR<P7s__42CZL3C-7CSv#-6_`}&Y`0>%uLLiB!#Ia_$KsC4c@#K!yGMf
z9H)4QjLf&nv>NYuvTc%2!WprpxIR&-qiN1k`jV#VAqKxq+n)QS?wK1~{832an?e->
zYolK;^;m1F_!g{e#ACGOyk0-W{__Ch4Ug~Smyp}D93$ciCWARB-WOvhY2m{*2Z>FK
z$?pX&dt5&n8-*$;e9j7Hdb;1q#1k6fciZi<*qqd`E<XOVth}kblfdA_5Tfqav`RJt
z=clqLmgP5+2UA~|v*b#d87mZVX13r9ckkg3xShq!i;lUR+?}*-D=Bj2PT+n4O;JVr
z%B1{kX-d)MOmR&RZ0|-krY5Ho0)PeVc7Ik3zYX!SuLtdM|1Tr_R{CP2L1LfT+NARB
zdiaLBz#UOPKKnU?=aV*VT#QY%TbU<<q?+i4|M|sF?hrHvQ1jCXl>8asSQTEO2rDVU
zu`62Wx~N~xDi$&)Yk)|`0sz9_ZQ30?D>Hn=9Tq;mJKr4eC)}K08V}2P{dWW2BlB7V
zW*kiaRq2fGwE>U|<2s~HkKQVLJ@va|p+1fNYA?e1kER*<UuM{DpL*k#PdFb6EIji0
z{3a(oyp&q+a2sxU(|k<?voSz|{qh$H%iHN+o+p=8)$80cQ27v@kK=%w&salv(5Q_A
zCx3KT%s4P^yXHxhGB(rXg0P-fv+drVMS#7k42ho;(S9~PnonQ59)89fPG?5m4$u<i
zjRIaG`^bK>3VWWXIAG;ww#OsY{BCmkj9dU;{8@>E+sbbouzLNTTUY|j!#I-@sL;ej
z%EChzDBWmf$Ne{^5BWgjh~ub9M|GeIW2Aaqj6Pgi<73PR>>|4!z6Jr$lo3mKw0H3?
zkOh6SWaQt(Y0kE9s1g5BpR<_q`=f3M&zC6p7vv{NRu=uGCCR&JG|6OrT-%SQ348_y
zX1~khwJT=7>3?gq{m>2+oR`n?O$_CY;;&~=w2r6SS}hf@RRY5e_x3%V`m5%r^O`6m
znORKwzm(=TI7IgE6Ymb@vF;ueJ-wH>Q2(;Qlwd}uIBd3=`p00#L;TEGwGzv41%=;w
zq1-P4|7bMIa`Z0h|KgzL`o5wq6lMwzQDh{WOOpW*ftbGLkcC0|(Xc5vG#D-Ap6gB4
z+a3SL(o+j4h5hCdDuo=CBm0qZt+QWFA_+8CjtVlO_{HY%|Hsr@KvlJM@52WakrJh(
zxk|T4w}5nagMxs7fONAF1w;g-ySuv^q`SLIy1Tx0y#L>QzcIwY;GDD1-fPV@*PPFM
z=6Y1OsWcdOK25T{$MJV@6hV5sv9|KTqEA_by)dnA0oYhn)rBFB6eYkXtS=7zwA7Ia
z0UV;!;Sd+Yf+R>}{`%->^Kd#=tK;DF=On9lyf}0aE^uX=t2+{%y|gmsBPrwG(jVfl
zZ?|PfE8}mihyoN*v_$GgNB0kz#x<2KnE(s6XDf;%S5(aBgW|)w1%3_lzF`<LuY(L>
zE65UPBUFI}`5Maw%Is>p9)0dfahC7|Ns|p*@>vLZd397mSLH_QFIZ!rbNzVLd8TYL
zH2eEo?o_oKflYGr%4$U9aeMO$td7svE<ab7BtPHTXa}DT2Nk#?e9;FB-04znJ;;Gd
zTC^LNisYY5D@X9D!=;?fkKV%8iFuNX^=xz-_{pT4jPWws+ubf!W%7%WnZ1vW90d*j
z2`@1oeG05h-_WKq27slF=tNE3&%Q!Kp%PB-r!a|8(5w!K_q~k+pY#$6gc)d37VZcN
zi`>(7hfC>AIz~yQhk(S$^^cUo|JHW-UstbLW=s-%h;njjXEK)cB8qU-N~jt@iM&(+
z&0W1=W7n7^tgEviroUC8!;ygTXNO@JiVya*=IiZJH1X;}U*R@HmsC`S^MJqJm3z#E
z=&~7+Zp{WbpTq4U1pF~uy}OF>ohUXC@sk()ZaePFVANWf+ucDpt)CMxEgPqp*=4}M
zxHR8+<(0FQH$7ThQdS34QL%?&see*TrhC|nrDLI>x!M~!_*P_j-lQ0ul<WAf@9q5)
z41$IzsK$aba+r0S{YewpYN;&d{bZ71e&3ST@5%?H>C)pXIkDa?`8~b+UZu8geMC<e
zRbq6WPdK#V@=;G!dpABi^yA)Md&A^IJ>Jha&W!B-`xy$^PTL14Ank0MnewX+k{=_y
zZ(dQhb}XXKF8YoV>lmV4P)%AuwOuI6az1GaQ8b-1^9MCV9HCRLt?^y1VF7ej@X2P!
zDkvKA_m^=nYH(CBqo#*YH{SS^pla67%uAyg^U1M_BUij_JD3j+rP+@D9!VcwS9ebX
z+~&uz-ml!D?1f(c1D#3bUN+P7Jq4lHO_e)!#-pn|LKgh5E~8y_@v+_z_X6e~auWc$
zx8pmjA``B*h*X-)o4=e`sx_|e?c$y8i4nQ-c7J%iUZ^*7g8eD!_BpdPIW2)SxriD%
z*|)MUiTAGB+QlC{UBb6p*tZqNL!L4)F>e3{5`u=nr+*yEKZU3fVY|Dl1rd8SsC0X0
zeM|{aHVcTq?Q?|HP4HJi^IdZ@?o*A==f~A>A|j4X+xyDf`>tGQ>B3P#DhMFTKb3oA
z9w_xn+hUeUdTUnzbaA;Y4jx^`I!RXo_zbJyea<+?5=(Jhg9|Pixd=#QN*%~n1HN{2
z6=n^(keVnB|GDiDlcPvq)C!kLRPSIlF*0tC5Irtp;WTm9fF2s&P07jV^<g};dakd~
zc_GVtv1GudW}NX!myJvtZI@moU5R$1@g7f^oV-6HW5l-aksQUChdctR(d&7Bn-FS;
z9g%D$7T3|o#!m>7%KOe|9DYT_c1*4}$Y6d<XIBfQo+*EymdM#PHFZ;EX!3Pt0qsmJ
z;*%y9CK@(&L;{3qG#sHzfrn>mVav*y7)(@Id}Kr65P|opz>N97rFP>AP_u(KH}<PE
zAR3(_nI3a3U!N1Zx*qJB3eH9G)2Vtk_M7!n)NG1@B~|9Tl)`~HrW%X7H0t12g_$*b
zYdP-7F*YtdmripFusW<!1Z`}?_^YVl8g52t=MxPrv&lsDF`;3@ObJP~u#liUW<1sk
z48Sqq3H$&u(P|Hw{*>(Oq@bloGdCm$m7%N=E*}?{-d$4kCV5`RF~Z92yepk^L5?r$
z6&xxwICU<IG+!F?*HV>DxZ~rb4BkJJ^cIbimEG|Fh4!%U#y``6BuRp*wXY@a7?RK7
zRO7`Z&@{Jk8@u}o>)oA?&c_yiZ}g!vD6+-$w8w0o%PH>G8U?ueP(2-C4KY0WmY4X&
z@9A&HZ$(rrN#Wr7{i-2bY1tycr%|G()d?#<J8qnmP?nv1-cWGE+J%7b0q<Gy;g#&m
zQrk`%+BGu=rdZ3j$EyXmWxCKE^0fi>furr(Xr=VW71c(JY2v*6s$w(leh!Y`>(5{P
zDeObPbVY0Q(R;y0?q9Zym2!N<{WP*OPNc*1vd1hU0aNs4@T>FUY`6R2W36lTY?Zv(
zS~5?h`kHZbxfqp7-TM8ces{L=vj%~<t21*qowzAn67i-ZK_{~b1P)ODS=W+dBA`rv
zYrIi(r-j2>u}^tWSxzh~kx_2K+Rt$P2;!XM{D}QuVoZu&+!b^1MuobGsv__9__#*G
z6NT{>zOw+l#?Epo#m{)+v*4$hFR1uCyM6%?7;RgxM?27zA3Ok?UnetDvsTBdqjHrK
zE1{=kK$SS8<oC^w@#ga4gxh1f*k6&5!%_>9(S6`xLqFG{SZURt)1%I9M=P^(2+&@{
zS1l9T=6!QDKabAa-nc{ofa`+|x6hU19YMVlYs~Nu)Ve%|&E@6aYbF_`e^xac(5v6b
z=pg_sr9(WEEfN>xDZQxZi!(y*2!B?0I@?ZUDxd^`){6vwv)}tXxqBuC($FkJZ}r9K
z3WESa%k=l{8z<U0GG@~P%x&ZM{Wl4^QKW;Y{C?kx``k4hUi{R3vElUoL?VA<e}7%7
z(@|Qp#+hE<;>H(LIW!>IV@w8s8P&1AXRwkX4oc5n4_G!^F))|DNz8R#yCwr&MhJnF
z5S1-M@tI4zK!5+u-P!iZr;&{Q2CIxTtoRr)&#xr;JD-Ty3TMjgi!#0AG1RfK{bZ#(
z#C=tl&V|hfjN|T?v>WEDqQTG{0l#$jv#x+;e0VgBJCyohtBNIxVEarY@PDb%J2)!v
zMn_uWw}N~S0MP@sGGn1w|C+|%X&F%w{Y-DOv2}Fk2qPC+#wSu!OUXgG=W(@1e6mn{
zjDZVAh6){RWLFUjM@n_PZ0gmPmVHo)vJyL~e1CTXx!i`e&e*Q4KN}Q6N{WqkbI+Pe
zPfso~%{BQfBDgKQ%8BRldd|AeWhCFyl1b%xxyE2A8rjKC3nuiu<%0X<b+%|*C(f2?
z^&=%(IpYjC>ag#y-@6;`URzWBnT0<TTm7Xug;S5wYlD&D!=kdL{G79*T5oPp&FS<?
z(1l0ferY&L{vb>5`uv);ZJ<47sbT))=yTga1yCTH)6oF4wJmu%^R$2VcN8XRVmJMY
zSm}O?=o3f>o$>{F=uV3;NH9OHFE%B~kq}^wW-AIRfq4JsqFdVSbcX?MnvDSb;^vKA
zF0nyBIW2n4o~peaQODsy9Irjz=wo@h;NLD_o8+-)xw>l2PrI2cdEix(*1mk-w;15J
zGUa-Qh|J_25*00Ksrc+HBEpHAm!(Qs{1Gx_k3&iD-{QJk1E^EgKENcp=$YDEhf|vE
zb@$glW$58XB}7i(D355mSPFRduTfcUDA~(#hANY4cyTh3KY*FAvJyJ_Hcjcc3m{Qa
z5(Y)8j``caHlIWtsEsE4V`%zOB3b6xM_m&MN%h$DhTn#R^+8EhaFhx;)J&8ZoNZ0a
zzZjnMMh_(ia(;Zu^-CrEVWX~?#~levrETde6Sa4+Qp5*a?19o_IwQ73XQY}jF|V^7
znW;|q#*!Q`jLbg$jty-;UQaue9o9cxKm%u<C)?FKocoto6DVJwzBg+Wz;vPeUqsFi
z%3&?;CP{@s36P3ZwT_;-OmSXF=MM4k59x!_;*9Ee3?D)c1A28D*eedUQBx)mfAo5}
zg^aKkR~EPT%{R1DpP#<hIqibHb&^N;EY;oR)_DqY=$XK2uAI-Sg}?acGlCsdf-p*=
zvOvmBbs(LkJGa13h;c%-aI1miY&u(Ept8c2{q?IXC`NvFK-WE-e$-v#ql8o}0Rj6I
z7q03`6#kf7LjVT){&&-#Th)Y|DRynQNlMwBv<xrFnK6|O4FT+7=W-s9n}B37QJ_FY
zo}cy5?`?cuWu8&oeWOFoEe6(qv-+(ajuAD4N>OmlO#h0B7b(s=QsW<uE3wg=hM*{N
zl&;lC_1PgYFPHS$(Q!t2nes|iviX~rFieL?L0maosTdSJ?gVUpg5XC{NzY?DDnUhR
zz?0+W{@7P)U-XvIERR#Wxt(3u_re%?LzP5jqJb+538Y`YqPv)R9ZJbpKLyy}l0G^6
z`DZxV#0TTW;6SC`?Ubh0%U}s8M+${z?;@@-IcK0_yhV7$+7=sIik?n_hNiBZQ0X~A
z8)YZX4{g8M+Um(^EIjO)x91YNk6t$Q<hHXwj;CVQvMh3LV_X9L?1HzBMHRyO@<G47
zUWU=BrmD57CkExl$M?I>O8bhmYT7P7tsRIjRP-XR615d`-0t29ZZ&g<+V|Al6AmvO
z=TiXypZ7TEorxW>^(B%c@iDyclgL?$IJ0bJ?!Mi)eZ(p8^uIBUQR<@%_WDd_XEjHN
z`<dp&GtY@9#Hmx~^piH^J^=9pl?YHWH1Lc>-X$S<H9A3FUiFP~gK6nvmpLzY`qf7$
zug3kHBYzuW-@7DIwDUMr&2T%R)s~AFdr_EHpZefI9p7UI>z3g`Ps7N<gDp(Kk)tln
zJVBQ_y<4>jasH=~gH80-nP59+GBHl24#yguE$qMe(Jci}Ox5HCdY%*AEom8G^#QnZ
zX-_KAc#o~J`~@w2EU3Oy{b^H=24(tH9J2yN$dwadcOb=*qcAOq*VezzEs!jP^{pWy
z-nzlKo<>e~T959U=YHiWG2fsfpypvtthzx2h}?mLLyonZPLQR9r=cHhr5f|LaJJH%
zOG3tJ;iYrE%FAT)+=x|mmCr92MK-J#@AnnK!xE0kD}zFw;X$n}!Ofn|I4O_Jdq@%m
z)Q$e^Zwi45D5xxPL*zf4bH^9&e~*CT3vRfSs!5{|<IT{-$cln67HaqpP{DkDMv~FG
zzYXWMfSJ0`&>re(4bUUC1iw9t)o&_6d;jpRA|mpfEE!;X04-knMgj1E&CxpR*oUb9
zc>%A2->7;^&=>r<dNiV-#KnoC#{&HB2}M5{VD0SZYvGGe>>TXH_A7Ax(!*`eIw%jO
z2<&%QYoC0k;|{k_ES8|a60<`ou<f<?4LsQ4?8(Z?O9Ok>z8#vuuSk$(+}{(@sL%;3
z4Qn_E{IuXSbh;T~uEMtu9WdrkxA$HJ00#V-3?LUOkbZ6c*~uEQzr3}ffaUYWz8F&o
zS$2rNyp+*6VKFJtYkjMqSXujsrPPA{=2PDOcWHoew4TPoi|~zs#FVSk>=o`+p5uKn
zrn}JR%)mTeyKlJf)$teCk^xsGEWG$qj15*OkDtl14zlJ+XbCH~P-iW_KnQ=E(+$+W
zP&G*L5Sms6>e<`*xPcf3bdOCPqEwX-DlJW-u5*W1&>&7?rio!;V(J%~kFi<peB*xl
z=-yntk&JAwf|n}cXNVGmjI8Nq^5s>~-aNcs#(=5!_15G6xCH)5rMZ?wZs~Z>OL?hC
zu|-54*=0fQi>|vMUJZo=Z^G)94czkbof=(Vk>9!(HCU+Dqdp$!T#g*1hodjtjd_bZ
z<wQ3E8}<Q4j5%;VRjn=ztmsSj6I2ijfA8&Y?(Y1>S!1_cJkEGfRA<Zn8AROX<OQWC
zvy>6OqEACY`tPU9laqw0IOzU=fh0|_m%0=5Py6}Km&j0T-(|oJ=xmvu{2m|LaWFBt
zWM^wzO%50^7-j7o=W8duv~IKF^;&1=2ND3;j~Xw;+&eqd>X@-!XSlz%K?%3vPCVW2
z>LLgwSlM?Kw|V|X{$;XnE&X-}hRJo>5TG={kwd%!z1(srTcvxFbPfhEF-aWgR72FM
zGAzX)BCh^nQ^<y~X#ivHFBX^im+-QmxQRHEAZvVEeeIaUo>z^>A9|rLI_?fssW<N5
zr3c>qsU`ar6Pj!PvoOK^5kM9g6<Ok>#w~F7mqSee6f>L?jqG8`L}(~xzw7n29Jfof
zjiL5w&IlJ#p{TV0BZ;BNgLx>LN7C?Ta)ZDFUi&xMN{919%4|1Jgh13PsDmz!=yttq
z3-VD{H$oWTK1mCCugfl!(T4V1E>4=l^oaRO#;Y=Y(A(Q*;{Z~Ol7AJPmb^~FD^jh~
zf1E1z&iGTFGMOwR8;=iw(1O)<dV7SveB;06g43XaXOZkTBi=yN$R32#iOQ^*i<a;E
z+gGTauC}eVbh>y=mLJ0A1#G;)u6)I2OKmeQGrviw0w!dzci8K^78(0;N5;ju%9@fs
z)-fmnwwN!89kc%XzbgJiA>_YVzu2QkWrbI43oV1c$b;lRBt@R#rYGGo-1ae%T)O{$
z_XdhzBv{Dx9p^xaj_$p<kp>wxz4(*$^2H~hpO$BGCR8Lywh4S#{!VM4pmd|Gz0m2H
z9r!<hLbuczW)g+HVaU$=08-d66*a6s+amW(XId)@H%8pgQ!Iz4#-hKDP=Pw|T;K5E
zuTw)`)_hqsC0m>5t3wm5(05P9)%YNRBr;f##f*rJqV<FR#~_66XHHfJmH#|vlfdyw
zGJTI{C`EtXJ_gW6S^GRAASi8|lRjF!HB~OR45_?1RS%Y+*zxn*7#Z0hVv`f{{^=L%
z_)?kX<N6;7>icQwkG-Va#GJ0YN<V(UEBH3)CG#sZs&okpuHU!0t2Lsdq2dd_&GL4w
z*jN^;D?{)SmDQA}v83tO{>=%n{`~6#>(eRsW#VU(D$b>>HMs$-ED<f!?xm-DSppTh
zjCt-Yt~LRyx2}9;(prq|R%!1$ejyakv-r_MG+nq$fsENfj`VB2>0It-5*gV^c4zdj
zb+L-s3d^6&MH+eL)uI2RT|5-3iUh%@(?3!@^poyt@$atPfB&P!eqeiPYN2(YlzEor
zd9Ic>N?7}DObbr7VXS(fti^6M$OXzPH#~0tz^v}Ch-^f6*;V@j0xnIObtte<r2BQ3
z*KWmsnfet*sA&B56y&s~<;O--#(n~wMvq@~UObF|_UHWrik}g+KPfCm)g%JN7kQCK
zgX7H!Bf;uZZbxO~7r0h+KN554L5g~EXZYax2<tYn2ojuOz*>um_zG$_ERSH#;wUh2
zG=4(*T{zufHiP6wQ`NAV3Dg}44mPNy=;$}cDlqm1IYiBGfPn*S7x4e!`Q41K{3t|!
zNf@jDrqQ0LIwI8n?aLRrNuEFJ-102~nJ=+K+}**(^U7?+!KSCXy!`=A*9V!I8sEl3
zv1I0Oo0IK<ZHBSKqe(@Xow?AomX|L1bDiONJ3N08ZksE<LSn8T4GE+~8;wD_{2mll
zWDQew-f4NC#4fK7EGuiI3upVMpxK`RE8;K1B{Xh2c#eLU-zzRpuF-#3f4*pH)D<%%
z@+xbEXQ3bDYl32!ZF4PgC$oHVqB<6XeE(Z%$+Um-x?b&iAF1lNhJ?MJC+9=iL6|NM
z&A^++(q@0_XgB*Q+23zGw?sY}gMX}d-p>=jT2al+&p!)Q1;@3VvA)nU^hH8Gyp<jx
z50}z-h9`8rFkZ~DOz2$_lekHZi&qSm001(i0sMiNkrAg^@laY>Ip2^0kYd86+O&1Q
ze&*mbKez<-=tD{U)0_=Z!b@UTqYJm$^G$c+;SSc)5e5@*a5LWuaE4KhWSwM%BCvJo
zLC~kgIdW<^`Y-76kW3|<KC8#ds-gXlrEV}S4qy2;taZX_O~b$@;3434E(>ct3bLAM
z?ZlX&$;wI!Z6rf(xKJ^_mkR$;$ni-S(Ty{WY+OtXI9aidh5HY;<upHx!r7+*;o;zJ
zNZIbjokYFV7S3`S*_~0)Qfv;IafHlB21dGqZ}r`N27f^i#|rsa>V^UAcDSmWqcSfp
zky)4O-6h$lT7s`%;4?Xpo?of18;Rlfwq@T|&qOC^4-`I!ElyqZc%6QN0G54EF<Pg;
zB9_HYyFQ_fS_IT1miMdohja7SF%I@P!E3{^+Jli&mw8t5TY4t?$`CR&vhl;RmyK!P
z)@skk3D4VnAehEvDSd=b`NVd9KzZ;pG#7iM#p6>V`>5>ya`}JQF#f_YpZM*8-tB{#
zt+3w+_FhI(x$8GrF5Xrg7Di**R-@mi2yC8!`6(M`EE+D^4B^09oN9u}QKc8%o@ln8
zkLqC=K!Ctdc6w;M<JATqrry1;_&fYEyyC8aw2`GiBf-G1EG9Rf&k=uf%Y8DI88xB!
zKs8wfK<W%vO0@D7=BXo>F!U2kpU7~^<RKOcBcqii1y>A1U>%kPTK!=`f`))YlvhXG
z)!h^9q@lHFZs?s3N}%aRI$gaPX-e@B1v_AX`l{=FSWiO0h?!XTCOf!XAkw_(;CVt#
z0WW6}85(L$%`Cq&5Y`jr=CnEL%#sS(l1%R<Wcj(T@6yoUFreQk<f2E)KdbAUqf6xC
zT6GA_EKWjd$&cjv)$m9v;9~(yA*^c+Ow}{pa;XVGxpJ&UIu;7*|Gf&~7QJNfvM>?W
zNK1D^d^5md@m=qj=ys#7r!P!pUeM)7wYdCHdo$=tN29=!V5D1;Qsd~(u_4%da<+57
zV;ECgJ9b?YO-NAeG4JYDZjGIIbIB>FUjygetr!{k1(oB1>Tz0R7*=q&tZ6#2r(EBy
z=G_&2+3{h)IoJ90A=INzLqlr>sB!@B<T*e7<XZs`x^^=~oLrB76Lz3UOLJr41p*g|
zt$n%8*B_?V=e)9L{lwb`B!YK?qp$G(ONmGs|L(Y5s)@mvCk3^81sq@N`7bF?P&3Uh
z-_~4>Y&Op)099%6xw{W<Kh!VRR$7X#w?mB=E-S_!v&H_udwU7jFwjp<K5p-}O$_i$
z7gn4~kIzd;QDqi=2R#Op_(WrKNqAVJ8z<&}aOFhbzAo1%%Rf}g4huM*r>0^In38T>
z0f8a~E?%&ThMjm~;P|-A-iUu;LaevP0|Dp8IbLPz!{tfEx|6w}9Uj-fx)cMs1*p6$
z*4q!oeY$cjazd|JvbEhNCN8VSaqN5;fH=1!?Y93geE|j9VrGsn$EF01Wu}(qtrCJ(
zdOM?}oM^VEP-$s$bUM;A>qDHduy;lo?zrV3`JeE(4+q@ES5{+Wv)}>1uKhhPPMbx^
z?;*sgJ|D}G-Yj`Ia}7rLEp*4}^pBwBLf81ni?_equ=N$3=-Wp-zq8DJb#i+n*nava
zFpac4Yd=$aKUaR`j>+z`O;i8xZ>1Af|Ah9Kj~WSpd`YU>>#kuQgRbv0oIQ5sTvK{$
z%m9x+GIeLf`n3O-wA#IZ;~v!Nbc!_Lld&;>PE1dr-Maj-+?yjQwuxz(uC9U~rJqK*
zfDZ$s9UK!1czHRF20ML{E6$)s+->w4P$z5|S)Q>P8Y0&5o9?ZvnYfSDrhmu{^_o`;
z;7y#lmgwutcO=q({``r@9SRJEAX+ETpPZhV403dwzIK_=)JtjT4G6Hx@9bx!qhg@)
zfhNoBW=VET_g4b^En3Yu9;B^h4UBYL-K+aZDNx@dySgf6tAqgH8}Rt4t{zoI#Tu9>
z*^V%xYyrA>(f84ufWa+8nRCmZoAVgyN|GOhUp_lTp}|3&c5)sb@6g`(b}>hspRdVf
z%Y$YjM_pJIRTii7LrU7Zq`Y~oUv8TUvcuGpk~=-lzuwYSpsuaYXY7)sCnslaY)3uS
z-Ve54X|Jvi8jXwNdHKH$Z)x7i={}wL#w0u<moxUYh1Xl<(?^Cj)$4t|Ab81HMknNt
z@M}N-k_4T&sOW@Tj)o#4Je8m1HX<-)R(CDaamO}Ba3<v3WpfG@ArWUD?&ZzZ$v@7c
zZR0I*cVBil$tD{VZ)a;9%4c~2VGL6mjc==XO5^mDIrrm5d8kx++38u1y8B#v2@7XO
z`|M3Z(1=RR(IrQA%|JRXJ7;AsX-69EeCn39om<d<i}PV}@n2|<K~@$-j32_nQV`?K
z4-hLMv*MW8aUD~X#ZTrB-e_5)?BUf=*MJ%v=<1);*5cr*o(YS~?S7VKKADx9{8~zC
z)Ll(C^`0kFd!it1$X!*?G4PZD^TlXTsqE?XG|qtICW*TlA$5J#;ehaI?L}6dcsVe(
zZ#wn;3CA0LQgW26f@A>UnFTrj9Ea!oc<5_UdDe9Y_J3hu^3vZZ5YU)undr&+$o&C#
zCH@*kQu0C5h~lx-a0Dso4mx*7^{AbfZk!$IhLQ^vXc1T^-QzhQ!nv891r(|{X<6}X
zr56@&F~)v{*}=5f-mw-!rQm}wcnxOr41wodnF6G#;a}6Uq3rh6<fK|W)%^)@$&89|
z_i|S`O@blkORgyma|WWL`;`x1K`59gk3nB!Y6T}7F))9CE|1os6$Xv^h^R8Q0G7tx
zDPNF5m5M*vs5cF@F2fW2`CsLVEj@$wismtfif+!k<A9Ix*9r)+Inx99llq3e2mQ^T
zA?>(=oDJ?2>CdTGrVN1dQBz?iDJDs4qQpk6yhWZ{_7hu@b`s>g`S!U~aZ9b{pB2PG
zQ-XtNhcZ><_Uu5xA2?M=<STD-Vdlg)RvaJuCy4Jr>l0v8#a~Ym;(i79`ZltkaNDTz
z<Z}n8$R6}EfC`R1>4u9O6NpA2u!wycV{`j&Z?xeA1?!;>nfo1SgW=&n>+3DgZ(Xe~
zic#M7#t3y{D{FEZ<Fg`aB{P57udvL#EoGjcmXZI@yL=@4>;7%YVo{qgpSO0=F^*&h
z-4>&O+2~NvPo(yJGLAn|FH7&HtHTu^*PVPp5#QcL1JSN+4EEsv*=7OWJpSsVlsr1=
z<EE6*7g=^YD<G-KR-totd($XRC24B@z8OvwrKMlVl#aCD5&?|wu1#B5FELvZZ6-Ag
zq~qv+o=O11AAKvi{Z^`Q@?t6LbI<Ah;I}W|8;`l3EOnLE$l__}j5)NEUoZgT;K+WT
zBQ)tj_0Y<k-a($AC%A+92=vy>&FLPN7E=v&eE|$H4FR`*CKBAeHWBhW>1f_9^6nS!
zx3nR<{|`sHC;0o^?$hOzO*8%>-BQN80^fzlid2k@C25)Wc+s2u4aMGTNbBkVYOsKt
z4Hax)pZ~4HgU~Kuw<eGSSyGu&zb(1=s21w7hl{}=>}F1f8xZJ%p0^Us-o0~sDd?-$
z-a8p}o*`R}%nHs*zgs{DxeZwYpR|*)+Si!@UJswFSw2K}cYLhTP10V4i;HY(*u~<7
zDJTW7SKWd_og^KDfyTx^E|1Ex+4ss14Xukx3_x(oq1QMCEn45zxP0y7OtG#gUJAEP
z1>d<K1N5|4Bc&hv`=!UmqGcM(WHe@i6d6P@3}gbo3XxATlnUNWcO=@9b6*PdY5boF
zEH%nfIC&!*&!k-9Z*9YyKc-xtC?+e6=IR2+0<SaMD6)_~dE$208>+QJeY)56O%LJ+
zA)Z06;LLV?v68oNa~?l3s4hxKV>;sXyVdFK?y)}r#dFYu;&XLXXMgb!daIM`IJ4qM
z%hzDR4C3y6J(s)jcEg|_Udvb@cQ>-(Ub&m<TI=0$DOX0&|E9ATYa96smX3?dQ<#8P
zhGw=mFcV6xr18S=L~y1$AygYGbO2ZJrWOVu*cXuM!cUYMcY>Fu8xQF~Gnt7a*#{Fl
zo%)TcU1RitUG8tdnA&0s4U9kO>{r|)Y0D?R@5vqk%-^Gdi)WKDqx(0MP5$xI$6_0S
z@N)hZX5;LP-+#u!_Y&^h-T_wpRNY5X7bRMAh`fO1U_cful0tSpfWMrN(bVMM#SZ>t
z0GL|U!Ahk>`ZSs0zu|tSbPt}6%s|>q%;~2pJa2GB*HUTr=866}CL}r!ZxvzmUpF)|
z&N!~-C@;v@D!B<zM~EuF{-^Yu_@%7$at-LDIqZ7sV>KxSNPPRUu{RPC_)t`%3nL6v
zr~uz7W4|WA%~c@G#(7te>Gvw`E2AE!buz)P#SNr@9sQOvv*>Wr@0xvAyx&*8|AX5A
zDLnUZ-4*j_up@|$AoB#g+1F=@ByJAteCt&0H>K|^>(yq0cBZ32i=-?_J<|Gkov#m2
zKrU+@FaVZupGZeERAR)5i;5(j5ZgAm_)mQQ;eu4U5(_j&24f(ZMCQc~eRN)2q^@3<
zpU>q)8@3Wj9^>-kKL|I7*iUM#$!dknnlAhYIKzD2<Jy8sKH(-QGnKXNP8ZV!wKA@j
zw5_RfA+;<8(UYm5AN~BeeW9HYRPkR+cR%jw`~ZpDF@RP~*hLkW?cTLjF-IXZT%IS~
zZoA;o0lGS2pd4fjdMW{k|Cx>TMPr>8SZWR(zZay~klD{oM!YZq3(0)B)RGVojb87r
z48ZGV8v2RN^L+r*{^FaC^vFd?$zO^qA*>}KitDJG{?*{bN^PXxV)T0v(v#7Bs223P
z#_)4sgJ2TgvRBxz_u&x5p2QHepo(2y4)*mCP_wfB?b$$`B8zMa{`O&2iV{!~KM`|!
zorEa0TTM527Q|20e7yTAkblqMt<-g+3LRWt1;9Y8#4^3$+h5Sn6&<alq$#Q6@dDKP
zxgw%|Dy6q{v_<!!3H$>!&EyIhTJU*zWj1O_nTjI)KeB0ICY<+}rSeNjr57+C&wl?_
zR?_r~E@q)$aEU+UcV)Zv6I^XcF?{TDsEQa*%ys-*#mw@(x*(t!9<?;I!Z}oldhXeh
z8<V3rs7Sb~_xEX5NkPe9mL4PoAN<pWi~0c<H!3RTA86<j<LMn*cXyAXs>Ne7GqWpC
z6LP+f=|)EQ`1tP>q^G!+8ilM59_M#ROH0!-G9op#v>=j`r(Uz<yIcanD{jX<;#dv~
zmqF!c50^-vq0B$RLYa3>)uCFsv}y!2oQ_}Sn%O3n1}$C$L5El0ta*a$Y1n>Zsue8V
z1Y~I$@<Ajd(bpF$fV*-qXeAt>tnvH#Hs{eupl)Tzs^dtg6ufZ_M2;ODZI=QA2TMg}
z_d*GmF0_E4yaq4hgI=ij&3>#qfq}W@t6m^kE_B#y>WN2W;bcn^PHwWFvI)aN;nwCD
z)w<XGNEm4;cHEr9zSH<QdmJC^_3JV*P^UE59SqX6j~FR=eS^kt+qQ0zgS*xU4FSk_
zv9S2X?Ug$r=)dC$IJP?9)F+CnG{XQ1`C7eYGtYUbsurjst1lW`$=}qd%5pTi3IPW^
z1K=11xF-g`?5BdiAPG4|`Q0O9k)pQy_~f~5V?;Dw5<+AG-TLA|P%x+mT5A$i;NgKK
za6Fs!?8J1UV#pBV&(luNCtrT2r5Eqth#HKHNcQwMj}M0!2L_`58B74ZFcK8^flw8w
z;vHP(Wom!`TfC)}8{+>lI;*paid5iG@zBYcg(6G&s~Ye9^R~1M#KECUaK;nn=OV+A
zi{KC85un>Mfe7MuPQ#6SXO<LHT1~(|tFyVZ^rP9r=*>@%2U=M5T$Z5^+t`x*2spp%
z;kmjlzFuidTU#&Lm_@S8Nc$BL>Rau{!sJkKVoCI-ay*~#R^(@FJ&9ob6_!5VH5*ux
zST3wSs|=@NZlr;Mb_d5s?>e4DO_a{>K&q#%qs?8(iqI|KGA>%03Mn~ofcOWW0H3Yx
z<iDtyR1&_os{p1|2IkJd$(P&xON|C5rn*-y6m%+K;SujYemEH{93N}{-T7-N{wu+}
z=DDi<y^-eS))DCwLbWMy?N{VUs`b}4uh^@{G9?b?$-rFlvD+Q+lXbNG{$|Y&)vy+b
zpKIrGkY$WL$JA02dXaR=Z$=}oj2(jN4<e4Hx%q~nTa^EEd3-rw1Iel^K_z&0c$l`n
z`Z82~8BjXVFdd2W@_;{im@F5eijv)es+oFC7p6Ktl=%-IT~6X<3YBxvs5_qwk;Z>}
z(G3yYPM}P9z2?#c4Vg(VMaE%6aElTPMOUk05iq5r1(FsTU6Aq)7S`93^j<izauzDy
zW_kXpbA##ZlJ@tVKU~tYc2p2NR;+dg^v8+nM4M`_v=#%9V-&w)G0oZh-))Da+50{Y
z%9%}h3GeG~uxe-pn4g4-ln->_^qT&a=X2h26E`%$xLz5QYKM4}20ba?4L#5hv@RsC
z{giEsY?9QIzxXg<iyAb;pDbg6Z0fLc!Vs5~|8|<Q&(#jJlK~|ak1xA1@6`Tps2&G#
zfCA4T4G()ju<ZtRe2h<RIuhXbzPHI;$KPJ47FN$zvZjN$!;z?kx#FYV7EUY}?7CX9
zX1Gouv)Kv_GBovIgFKcg9jIk<hm-CV@SK){4i%W8tDSlGCQ8eFx#Px?Y&gnw$9`zp
zrc`p7nwqs|NhechipDzyl@GRt&xCM18n%9-)vdg?WZnAXd7Ym0r0&&^@5+i|zM=}h
z=t?3={Dh(8&hg<A<>?IRn79S9N%V5G!Ckh#b6ef|f+)U?uJ{rBkXq`2LzLSZJEgBh
zuNd#wJVWH;lxDIzQTvY1GV(*xo=xcJh9AOD0%~Hme#k7v6&|1ee&9}()_26-c6-6}
zxZcssE%S`BTUWA#z^A}sUWI<YPIjQ<R52m<jU&{+Crji=sr`$XCiXqP%T~W)NLojO
zZhoEcx#&PK$7#{ssd7om%dO437hiLEUMg%OAwAlXC%201U^{K3Z;YH_V|EDyAbPGU
z@{?@Um3G28!2&`h_%VZ?Q1YTE3got-yDKL<BAd@Nky$yL_t<x(ZU#Rj;m@4N$RnLx
zHz-otvF@M*oXD6X$ng2uUd`TlO^yYTML$;#as)Ud$Yv{YB%1KFP5m|H?QTLXcQfJ5
zcxxrP@~06ht>XKn1UZdV|7@p9V(yIl$`c~VVe@-d$u9_Q!GE<r=={B&w^4w@J1)NM
z+nZ`+3E<m&-SM$joDm#N&Gl)oP?`3Fuw6F%MaROhvhZvBG6oV{wE~DZMyUlUw<8Ad
zHmg|A>t^yNAq98AE0vBoIoX=cN8rbYPy%gK4xIk1bunRr`;0KB607o0LR(LGw7$X~
zM@AY_?b)B2%S8IxrgW3`!0|zWdNZn5@EIDl#;K#p$O^l)P0Uph{Bt9|2i--y*tD`%
z@A<{H(j$7LvbbIQ#@>vOBt5+FPS+xqePrVLN!DC1=4s^qgz5FrN?YJtyBgGTE~7*r
z=!^{G@<ckmw{L{Z2V;mAa(mM*dbji-%Y0gYT!&uIS+g%C^|Ub2G559?ligl7#UYVT
z_0w#$7ovMYT*{YY!HKGS21q%uC%y3}S#IM`B6CF>_)TWHURYApq0U({&P8RxVr30^
z%5o3*EK*;N<(eAi%S)>d@9~2<ilvi5QPGPvmMuTqojBb=CNo_RDP*&_rIOWdluZ7D
zP{YO`Q|;}|6R~o_v^4mVN{m@PE=e)(IjWN`!Hm{+W@EPC{emFGZnZ{tYk%qj{~5F8
z!okK%$wQ0`T;r;TxJ)5U+j^IUw{uRan9{azq0cwEBF$)px8Pj44{8xE9HuJCsg48e
zMWVLgz9y(p^+uIWTU$=f$o3{uD+32_tgy?RZXZ#SNsTp(M88HCi5x8~Y#`zNI5}2t
zCq7$_ii!@E@w`^pX8ZATRFEt-r9*7KKwgdVW0%j{^fZ$3dF5!ZWMbZm(-tUGV>H%O
zis_7qBb<d$Z~r}^NaYC<bKS6f+%F1UJjKFcTXz4)krYNZ>)SPNEJjMK$@y6w$`S~8
zwU7!B<GZ?p$zQ#BpPXExEQauW^#@+$W*f@N=`!@~@U^TmxAP?`icP1bAtlY<8L-5u
zJFs!(3$YubR-+G?E4t;$tR1#RV5eo{>fCzlDbI>7vh<=^tpomMSfdaal6FN5O37@^
zsqh4P#_Jv8078zu6P<>gpCj3#+Z-PcuL7<%n-EC2?WF=e->?UoWeM!gn1WGYhmy>n
zPjQ+4D|`c^^YnVI*JwT}0j@69%|(})tC0K!7jkifUde`xrmyYi&?U&z+SRh(_K03x
zQy=nJQ}wvIC-r3r2Z?V|%bOi1;PhP#7}%SRDHYx{y0RB(tD|B66=;PuJy$$v>3Q}v
ziO9unVk39rPtTkydxi)!gW4Aw6P`cjJ3LwhPPwX1amUTfyOlFGGVPTfDYNNOBM5nw
zIM(uiWP-hq&vSXFD~H(o2y@>{e4ix^QYtMeH&_tro4O0-*uNF!<h%TtP`fS1Z##h>
zxc=|q>`h23JJdAmz=cstGUrVtxjD9t54bP+p3<E_PHcx6LLQFoCp)uzy5|?lTGd9m
z1>Y^RFtFSIq`v1g)A%GU>iw<r7tXuV&7-SlF{9<F<3GvLp(g}sMNMz&Us*MfCUVbg
z`Tm^>7>rD~+27!i-EPWut>XwVx~r8V)z{26*U)X9wk<T9-FfhtUtlZHtV<(S)@6m!
z0=hR!EHE8^i{dz!+|M$!7RX5vP@R&GwO6AX+$`i`zVtk%@@OUo1?}JKCyF=wW8GPG
zr<&mqWCuCyjJE0*hV-mdC+)7^mr0+ymkay6|G0+@DSdEqVyDj6RkNJR8ECi{>H{8p
zMEena%Tm>%1+Okvevt11g`~?48B`h`9;>a(^7qIkn1ZjuLCUv`A;qgB?+qN1XUUq{
zjx>9z^5t7i$*+UfZ`bTH_Vmhh&Y3R7yh(#H1W%GIx296XSSKD0Ky3OUCrXXExVTt0
zLo@dcqP78M3u2&w^mIOjQk4Yr-}|?JO06l9T~p~dcC~3`HoA}j(V3o;&f(m9YmEjC
z+DGbV^lVKi|C%c}*&)uC<j<4c6sM4cuUbs9dwq;8>-4rwbG_+9Y_02&0W(rXdB-yU
z`~$xGncUiCaIA>Q!hKSQ_`IGs%Kv*DU~?%lk@_y#<Mp>b0<sr+g;0fpw$ho|Sp^AY
zJG+12IP|-j^UsjBc$OYgpiVwo!(5GQ2wT<IS_y%e_r@j;cK-Ua(=uoIA98VBN-A&b
zbAJ3QBIU{R-K)}T*0K^Dk2}Uf*K1eH?6gB-?QSowqn}R7TM*Dy8*Y7+cBIS19{5A)
zx+f<0c%Lcq&^&vy;k&cBe@Z&FOprLPEc6ewvZL6m{#IO-&i_TWyNU=2PiXpA@xUx0
zjgRT;t!*s)_qyMI8^(lS3M4Ap%N*|<H$T^(P_XA)hxDJz1a%!Z4@5U6L^kc7Wi3Pf
zqPp)Aj;EX8B4cd)e9yPI2Al=%6R)oeE!@9!tk9m-ASLZ3FXkuB&gvK$8*7oIws=8D
z$ILFV(!e~{b88~28j_Nd=w50TyuOE#_3$xDnmr{}*(Amg>F*B)k2B3-p|p9az*4|-
zR42B!roMQJWgj}TyWYx6A&FGSv`Boziy%p7hnu6|t=92|y}R-Iz`a~%DPgEiotGD)
zGt<H9Cd>^R$qLEa$0tYV>m#*}Ea;W|=c2J^W|$)?Kq(FJ!2_dGAQdhu^C695^Mj>c
z`U<gQLVT*VU76Pf|5%oAWRLbd5H2g3;=)W9q^G-lK=Q;tv`LCIsN-6vXl57nF?t!-
zv%J;I)9-mnN&jSHNxyDB2TDBBjieYI*X%$QlK!72rrP+R346^>Fg6=&&>AiC{*FoY
z;GhDx*^sM*Z|e(JB*AUO1B@Fq6(OKOgn-acdp(PzuD4$3&xX$tyMsH+4EK;BWh;|@
zQpUj^JRV@pWwoWV;P!w0VLgM6n?LfN75%I+k^ZK<ZpnFZqyLg8_5F%{^8&7W{g`oO
z*Sp|HjmG#s=a=w_g1TPO-Dx5=G<zfr+;5N)8&uAitLis>p?@>ialiHk7k@H87rk|G
zxc+*z;P<qc<feES7?0fP3PrHOajE{*e|V$&wi!+3i!?b=mi~quI7MuzgV#aTDM5XL
zS~5iJthOR2<=+L7^dxl=DW~&zWj}<1|N2%d-2~duIXUxJKHKc!cO9e1N#3iy%*YA+
zd&D9PZsgCiUQXE62|Hre$cfv;)wnM$`|~VKx3?0+6ggE>Y~~~1i-S054EOJe_Ar?J
zV<ct;Z)3ibB$rC7%(S1WH7=HQ<LT*<ybXgZLxL4IS=)k|Sv!c42_(E?m1p%Ckk9{9
zX>6JpOBmrnw0~37np}8ymqJx{RMVhStM@?QPUfq}gfRI`$JUmy>?jwX%%x@J6Jz7e
zSa;;pr657`^DbquOY2=Hf+NKvmed6Q9TV8aQ&c@5`+|lt<h7A7K5M16by<CUa@z3V
z%4Olp>kj>y?9>d|8c3XfraZqn>|ghkg+}A_#q1p`x-7n3yx4ir$d5kl@{hNVZrv7>
z0^@knlbYVJn@uW!YT-yxu4}IACvyUy_x2zD{qV&<;0p<nc!>#zYcqXL_Dt>n)ggn8
z3gOgwbPN13&LtluBn=Y3q?YeiEKJxD5>o3h`r<u1HPt=P6H0*u6lz`HCAImzOK)K{
z0qP>pp5ak{3Fl*qJYr>~lob11w6nB?N9`4kdVCz77@l6_gJD!^U0?vM^VmySR%(Z+
zF!M|=vlVGI`(dL##t7%9_`L<f3BfjA<;cWNguqDGPJ|Npe=ndcSnh+195Rm!A!_-A
zGM+~{o~AKCZ4j=(wBt}aX8kT1tXEt-yuT7}Mx}?GuOJ|;%q$6%xk|WD@>UO>!XhK7
zB~c$g{?)ov1YdCuwc&}0g+S9DJ|M&zG(b2UJZT6HA)%Qyp#~!-r}~#KUr-+}u9OSf
zRzo`A$ifhnZ2|4Q$>P?n$FLlNdqmYYA3mfx?9FBaU3o1MB-P(+vI1E^Uqw8_R7xq|
z1vK^I@IQC5IIzkvb&)1+eFY>Aod#f%btZ4D>r6aVZtR<lOUG(9M2^?{@&jQ>gapPz
z^iN?I)f<VB9?-lFq3DZd6$jU@;Cl8JmTNIzrwJ6)5-Ci7e}Cq(9&N%P_b`bAU%;(u
zczF0jCFogdeU!Tb%R(|9NXUn4SB0FkpfP_uSZv0lMqHHVLPv@m8b-j6{56a~^YOLJ
z2wVOW1X#I{nOP1{YNnJl8_Nk777<~Td)Q>g&CPwa7O77m3G~HPMlxi<ACovF<mA+Z
zgubA+Ds%*r<Oe=T!sBu|$qFRqeko}-o)`U+oE-kyM1_ODC$@uZD#*a^5quiyAk-or
z7)-<HbZg9IC4jS=gw=ja1xUqJ_+TMqSp<^u=cG$V(g9sKpmRS)8L5+EUQjJIp%2v?
z7#QRQA}2kn(qRf9=05UjpLf~LhwHerQzI&9M`dLH{{7bdxQ}zX-nG)k#s&u)J9Tfa
z)*L*t*=$wu*fWF+6jVK!2h_tRTRpdT9Eh+8$&8GG;Vr1P-N36AU}sl7JUsjn6g1J|
zjfVTQNt?EbEN=f+IGDp?TE!L%R_@)iHjFLLmmC%rW-yVNkwFJOq3>~-zSnr~fkDbQ
zneq*GgOVr(yQoJdb$Y`&XBRrg0vmk^*SD*4KE`?aG)bJyT?JCtwGRi%gQ^-DS*^a9
z=7-B|YT^$SM8Mv0ns|Sc**g;i#y~QZs_PZH{q}hNqfmsRVd_M;l&~-yu<vYTXC^2C
zg(+NcIaZVYkKp6*TfW|JETk{2QvxDgB|J=?Jb~;^2p9aZ0g=M6d;6OU>%Le{U;=x!
z`*aW!liZ6JFF-Z{EE;G@d*2=S3VuFef=Ma~3wu#)wi8)}fknZA?nzi`x2YsK97d}=
z3go|#9zE)Qg+5lOuW#59sOtNKghJBK&rhxW;Ue`%cHlw!i{CXPN}5azuYM<jt+B`|
z!&S8@8ZaJSJR4#At6QJ1RZeFw0~PAkQ%FutO|eIydETc0p#a!LxM_@5wM{z3=GAs#
zB^QS0G!r+srd+17qM{=F4#%uvqfc93ji!oYN)9);UrUjMhJ{f`ZcP>&2CN}2HeOi1
zfHhXSUD~(<A!;dES=FzPFh+rbpDqxk2X9J6Z!O{>^&||7j?VV-f)FK@QpEzwT+b~T
zHHzNMR9Gl$XhhZ6ZKfoP_(nLFJ%QaH0dtoMv246`m}iA~7%%_!lai7ej<ssYHc50>
zD=scBnc21BiND&)EiAK`;V8ERhsCNqfT>WouHqcRZbr`+pu!0~*ZkfYUi+=I#+x<f
zyiDYa;M?nSASVuysoB|;i3O47Z;oU}^6UYb?y+wu*k&BWo?302c4jAkhC<SU1n)G!
z0x9Ks<{9I#HA)L#`wQ;py2abAEiHB%@s4J|tifI2r9S|R+3UII0xAZ^7}@=W`01Y^
z5e!1MU+?k39ZJc_i~@nf)0H5>t<wqp31D9)P;M$>){D!_%aF8Ckn&ueY!b2<eA}y9
zdE)KsOD`y>556bfxHl@<8|`_HbR-j8DzVKELBj}0IfkSGc|*lLL=U*>H{Kjc`)vcU
zbhz9jkdFhVvEcOqc7uV$3~N*c=N?tJw6u&iLJ$}qElh!nr15<1%%49Ybh)yJ7k(+J
zpZG|FfL2(#>wf)&tM=#5MV3`dkJf-D##c;%!8mT)lYZWP4P0E$HXvqB8PA_W=6SE#
z;8qR4H6Zy`4{8x`F9jFNw&Mx^^(zUwSs~M^w2TLO;@OWd2sKEazai<69ET<00HMfZ
zJh<R@39v0F2J6|grFy<klSN7$==iUIa`IvmLNz#GqYEj?K~h>;*~O)X_z5z#0^*H=
zo?Zg@z88`)92POt<>r}9O>YQ%l8cI7&AA=-)odiVf_)^D+jd<e4$o6CX%WfuHBS87
z&DE|@JZc05<^EejqgEdbX%cKay!Cn=J7Eskoe^|-v=)*EY^|h$YSH%zG1otmX47Ta
zDJdyZBqNnp`jRFC*Ui~MFcnVVt!P?-xl{NRsvvk!(9lT8#&U#H#0$9BjIw~ZbkYfS
z*S6~b_aGr@ARN_<d*q4xqoLUlUo6`2hvl&oAA1OjCj>$!Po)nZK73qh4+u$ffOe<B
z8{f#%X~~ynHC+Ps3=ld4<CgCUlQV%`!~##8T~wq1z6!u$0l>94A&PEnNXW`E9%Hru
zjr3rO@{dC)voG;sIf(F?hOa6x0w3@1Zo3Bt&h-+(hycH}87xj3Orh(%7H&42=BLVn
z0JP>BJy0<*p@|rFb2h7aJ|JZ;#p7DmpQD<~W<Jg!N#}qF#%JYtKroU5IQZdZ3RoT%
z2}yngqc)J`a#}{&ovT#`al-J_0QQ30aRAua82u(tVeGB1G|?Y<lJU4-m9+<v>F-4W
zW%tdt=lZMkli;sz7eyH|(Mq4VeQf9KCv@GQcMon9jEyte1Bj?4f$LP4koex_2h&&s
z3&rJ6vM+cZ&Gjgc+bdI`Q%~HCmNcz+fOw!y01nUw#>HVdo8^9eMhoHt;A)k?=Ewf7
z38rh^hFeGQ;6}sJvLJ#%GX<_j3Opnj^|Ttt{eq=d-(2V8b+wKM@4kq>^~W<E&(qwR
zsVpRgZmhv3T@(d2maVbea<efyAgS58X-5a7CF6PQhXXk0v%v17LGsKq5m?v1aT<)Q
zjWAN76Tucz8EpQF4Ae_kKXaA2UpwTkT+}maS2#grrlzK99q^FZ>>-#U7|ld~JTF|5
zkPi5M_%03}8El++z_zxQe96?@-0I+V2!en>treM(u%0y_eVTJXtf{FvuVizy+NB!R
ze@;nF?RqY2V{_!(0aCJqjtz3S_R@{A=NR~+H@jis;pxHT;$~ANMwFI4ot+9mITbu+
zs^h$~cnBMG=N3S~nvr2Rjv|UwJhNNDCAq7lGKr@Pmtj%B`g@bfNl9<`UC%}fbZheh
z|Aq}T$bVm0x;<YwkEjLPl?(9i>;}IdvzrdTML>WM0Q3ZQ3Ahm9+3rla)1d(pGO~8u
zU$OBU%zr-%hVVd7PD)BCQ!f6;oF%>w5b4AvB`pA=v6~Uvmf4%%|BkA42-w38h`a&d
z_j%EiA{zuBl-+aPK$+<3VkO9a#-cLYdZir|1w|3Kk(0%*m>@!dif1h<KTR$3FR5v1
z#ZZ_|6zImhHu}NTcw+;ulkK=a?{d9c86FyHJI?^5V6|%OvVkS0<Q6t)4Gj&!uLUZ9
zY}U(fDJdu}W@`#n>!Fe1;ggdQ`CPF2;IG5QMu(-DH-TwB59?r)1#kQ~nx*(7Fc2=R
z3$cHieAU*`;W|Oou;fDotbomIlxE)nj7M<mk^(Tmx(WL~gUzvp;K-=VL;)+GZUOjs
zCsnLG)&s--v+r!7@%|C$X`HG9x|QQuimAYr`s*gf#pSK8ey9gwBO#An7I43~SXfF1
z*S7LiYllwowK@l+F5B86;6jCnI6e)Q4hZCyn@`};KOO)fcoJ~FVJ<TX{H|pt1Elbs
ztKQ{w7({US-@kwV!M94N8h$;&Cd;T<@>awLotDEGL^IPsJ6F%+(#T@EOmcUol9^Q=
zWYd<lyBy$MX_KcQ{vA#c^&idG=FHcvRRdZz;GZ|-?cLUnYec(z=maAAAG(eWp~I?Y
zEb#ek@FaIafW2V>39xRhLpVjgW~mr5CTZSDob7t=&|tENbXZU$@MAAAh`Ea12A^w#
z6>I5dbFO33AE#JpITx0YpqE@H%JFG3>F)M=;)>ik*X7>hV#$~C$B->|yxi&jLgQPA
zRvQHQ7-Rx<Z>wDkz{>mhXE3?Dr-z<XTJ-<!!^;amlVJxS_I6?h*VU32xJ=`)ns=K=
zEM^0%G=|GMg(4gH?yG}VO#6$2_dwUunU55>Mmkzr$p??ne^CHcN;wch(m3{KD`$3F
zR+^vfOu=0pA1M@cbbMTHIzqJ{f(6C{XrRhR+oF^3sNCP3H-d?O0WRYb^t7+Ld)Nnq
z$ZofMLW)QGZGlGdatfKFY<A%7bua4x&C8d<AncvL1&Y08_S<T0?d?(_@F?7>@wQ@g
z7aI$!3<Oq^Ci26Md;8g7i3;9bq!v`K6Xt6`#lj#yr7ylskJO_ZVLKg>7j!PX2A1|A
z#<(~U+^yQoU+F0~9~5A1KS1z18m6+G3D(j2#o<a{kib=^jI=cGJ|8xf%qz<1;NakO
z_#+Z;IMnumA&Cy@1zOS(Rs6KH*&2bsDk6F0LCP@&V$uDougI8s(8bp~*#rUitEiA(
zZL9k7QI_lT{qy+&@F_YwK`@ht{fC1hn7SutW5p236JPKRid%eObsPq|9yB9MlPH%h
zbk8tbR?htN_dgD#w;<b}tsZf`2F|j_?bF!dUwI1s=4W+a$*5R0TxNuXgs^=aOaj4w
zW~TYxECl*mE#p9FkW*E~1sr!6k1ufKv(>hgl7$UdX3^X8QYMi;PJ7iG6!E&86?NF@
zbG8HgMhQ3hf4&dB)Q4{EqmRQnj7v9@eMv(ofc@kN+p1J<O*blM%8h|B)!MRPVPV-C
z$plI`V4x{22fDfrNyB-Kd-3h}W~+N4zEa~pI#v-Q)%Aak{@Y(;+w7(<96eg=2?rke
z+qZ}HAWS9X_ykTg0}VfFg?jmulaor)gC{3;`!0o;d<UN31sp7_<y#a~R3(r=6F>0<
z|CX1R>t;v=J?eUORyu+Nt@d65kyj?$S)ew3a(BINesKWKUY#mQ7Zw(-(**C|tJ!7(
z!M`~p4Gqn?K)!<R^?7CbofQZK`dsfD;g-F>1|~k7SN_wdJdj!TkSHsLJHv(9Jeg78
z2_Y=&7I<b5Lkga5Psl@8BfyF>l$4ZC{=b^81g5Doj_W2v2I2%S5F`Q)25Pz7u!Srl
z#Go>it3r{fh(m}V3gxKy7+VwpHCCZh90Wu~DAWN3N=s4Xv}5Whh#bXv(5(YHD{^(%
z{XWf_CVgr1@_p~U|M$E4^8TX*@hU*kCEF!tF5ckVK#*#(_=L=rN6FUnWJ&j?LnDgQ
zo<{NV#FgEL1977AphwXUET29K5iq&-aoRehoEF9%dKUx+y+X>y+WNupm+tO+xp4hE
zhuComL%^2fN@t^OmJGHWx<F>hT=xIu{vc3!>@kckCa!=4pC6B>E+y94YbU85|D1Cb
zRhnO&a(A4Y#R8_44^>w3hnl0D{c74LWZ9`;#*Bf}<f;TcfX0{bC_u=;`nJ~A6dDEl
z?&9|*lR6j~DU&=zLFX5Ay?7n{u+WoR3eA(>P#Jx67<UDwqgDr+spjhHYD^diEHPWT
zDLiGXmb&YJ!*R5s(+4=SC!6hj2Y!j%E__Mw?_WK_er1<iTD;Uzw|YXy>JE2bO{Gzq
zF1uiMkri}@s6Ku~I?NG8SN2+WT(Dw>pxN0Gg(S66J0b2Z@iBxMkR7gB!O+yu^Nq`P
zkb$nllqzIzln-giV~Qq@jPiUsD9Y|Lf6w6-9|q(_q;+vboau?SI74IOzRrs_>wVSD
zIw=Zo-}>n-;EwSDVTF9~Q$s_LxUHJfO&ww>s{&cI#<53?E$OAi>2rN6m`UR!4;C+7
z%7huV^|eZ$ym;!}tTLbj;5ba_=3o7mG+{vX1E<20){(PJ|4{pnDwwpUse#zArZnO2
zJw16mo^l_n*Uc{q{m>C9fNvi2Lu(|1=T<wHw1x{Ln(XQ0;o;YC(-mPnB&{4lvUr%l
ztBI*8`Ulq<I)sQ}4FTJ&oJD1W<9%>uWq=%ez%#{lJNE3ZiQQb|OYTkyk5EWuL{peZ
zBvN#+pSK!x7kaXAwBoW$Xm|(Fh#qfvzbfqKmz2lqLfU1t=CP(Ba9VfK*CSevCdGq9
z7huoQ6^<d78U8^*m!~zgM|u*Wz*Gigt;{&V6D-)>f*O6>A4y&K0(#Exs@opux#Zc8
z3c7D*Wi5U?_B3#3{)X3aC)7>Ts%*z9duJy2bMl_PRf<kqS+59yf>~~1vEjbcq1C5h
zpO%}M-3D2!y6lSNP^z--uTnL5mhAFB;lF~=vOW1c05KN957$;aTt&TmQ?$M=$=yO>
zLv8oJ=3Wf!LoAVI#e%E--0x~BO3ts)Lu4%$0CK$Gs|+1-MpBiQ@n$EbMb<!K$G_1s
zz9(bh#|Yi^^voG-{`52wtKPb!tjcgvKte(Sv@p-M>#*Z&of~c=gb;p@j=PWsxES3R
zFD*PZ_g7511?jz=L3=NY^QO>}JqEp-qZFsGtYX`;S7hhr^10dqkJM@s6S>3-n1Vv!
zw;V}KKs<%uNz!)U2x1r^IBy<&DM;+g2@d+CNUaQq4s(<_8ExFxSlg;utyDcK{sHT8
z)YHYo+PXxZpj^8yqBPp7VVIgI|8P~%zZQQIW(?Rd7%T>RZ1=%~CB*9@tcw|urqfT1
z<^mzZgfTbBiG*=CcR7@0?iPJau|?n5qAWMph#a?W?n!ekdprz82b$YTYD$jvOwbU)
z0Tbd<_A_SXsp+&)JrNYa<x;n4(itx()nNR%8LU8LSvgY>ab@t#k3UAETYcD_{ofNi
z49#B#PX|!!%UWhbB&UzJceGm3w5mzX56IKjJ1o-IkA-w6DGUKQG9g?XGEn<@%L?bn
z_RhF^j(^swHEUQZ52EJ)0MzF6xx|g%O2au*W4F0CTv4DIjV(?&#s%0aYlF-g^E^8U
zrz_`&y5^V`SHqXxfKcDbT}g}JBTjkx+?(v9K5&*mh`fQ6&QDJFv?G{k-6m4-&+gkp
z$1T|XDulncxLG%^o|{z`Axw(8dUIC~6~t!OnV8^dBY}2DAw{`vCuffN`1pigBh3Ae
zAO9Zyb<A+!p}YO9!n7U!o^OLhFOUzz(eK%In~dCehRDPP`_Cd!I65NT2iYy8lm5)l
zY7na6GdQmQ&Vhl3gzM6o=Z46yb}*7B<X2q3BjKXdE?rB~8gex=Gc)<;$&NE__uc$@
zRG(p!yfSh+OqTHWbS~Uvq?$23i*JsT$J6tu&V{k7<T`zYW!yHJMo9Kt;Op~^ca_)v
G%>MuqMq(cT

diff --git a/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-50-1.png b/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-50-1.png
index efc5bd675b8052f10186039d3710a00875cde6b2..2ec0f04a9a926f1125b615a4f9b2e05e25695629 100644
GIT binary patch
delta 107886
zcmZ5{cQ}{d|G!d*knEiZk!0@?A$yN($=-Xrvo|4olbsc^M|QUCz4v(8dwdV|{`{`%
z_gAhfUa$K;=Xsvb$2wnH5o4PWi+i{r2Ylk;BNpw;xHyR^9x{seNDtmyI?$m#6nmJ9
zf}O+v@~gm0Y-&0>IWIYC?5Tp*M>tO@RFH6<F6<<CyYW3ZZJ@gs5g>44QBbH@X#gvQ
zB|S13OXOHBjK$2K{WUg4l5#C?cs}-RVOHVm&2*&h(Tu!kQ(>}h^L`(P(<j8LaVX9~
z!@;^hyyBiN9_LCOmn7^aeiG(06#E4+*tK{Pmc8K)p^$k$SxEWzi`M&z020~tf$PK}
z1?xU1#;SdP>fnL%Y5;o$Jlrg69Zeo{-P8;Y*}|6nL}+~9vLN4lgYo_9^!sqHh&C37
zQkk-pUSnY|N5<g^cW1gS=~F(fG=~^hZ_Xp`krKUqNkc)thhSG5uVZAUl~i$v{_w2o
zRT7&=T>>O5i^Z&}Ys|9kmnjcl)`qi`)gFmwA5rQ5Y~-3dA<3j+aTvWm+h&$3zvhr;
zb)XT~MqF*7h=h!f3tLeBO6emm_6mlZ6}qaRF2Q-nbq#olcaqOhHNk5MKuCKjqiL&1
z$EK>Ctvs^}<9i<z^8S#1v{O^smo!gQ4@wJvM8|cU?@c4#_q+V{5kWZx^AfWQ4jRS&
zrPgPG0?M#~{k`!qmoi_-Oe2++=GoQ~r<BFDf3UX<YoDQKkb_l5wP4u%uqEz`uJ_^#
zg8}eW*7Fa9SL~Sy*^_)Vw4k=ej;)=NN28A>SQAXzmG_Uzvi%>|w8$5^svij{lZUi@
zwx9MZvDl=40MmMR@RoEvYBg|H4tbx^Xw(g_jnaY7<n}cbQhq?z*x<#$PGJnzx@i$G
z8;Of8aR{KfY8^iaYzi2LrP6jA>@r!5`ftMt*C|YO1aUW@M5#iO7X~;3C)*lhqeeE$
zHn{u+19C;vOpmWlUy;!HWzf=4Y&6Zr#(rR&`(2S&efbm{8|RHhC#HIp<$b?uGyLN{
zj`%~TnwFK<SVamu`%TEs4dvJcB@a!ayBA;~e2P}X!xk+zq^SdVmh(kU-`cfkZCvfe
zu-;Dc-n%}je=2|kk-h67`@0ojC$i;o9rx-i1`Hn~sT`mU{BjW^{Z5@Y!|1g;NLb=v
zzq)4yQ`St=lPQwz;zA6zBf21;;hB9^QgUITN^avEtR=Fo?{%-=kU3UuBv`PV_0^5%
z0TsIT;!3k_<gt$NwrXDGM7`JA)ris%0ep0?TVzf#v=0-xcw=Q{(fDNmCif-bSG|y6
zN6&j|J5zTlf{S{W#Hrv1=iJwfVB^2Hi1~m6kcMuwi+sE0NcWfzw;Q|Ualk5l>hrw!
z0H)@jPFrmu0shi2hOcjYJLTgi)6TLpwKt{x7GZ;d$32Gw)PVwZKcyFkmoph#^AwD$
zp}IpZ%IMxtS=pAB?G{lAdh3}l+Dq3foN`!HokM3ok95>Z?DeA747%=01O_6XU0tBl
z=61%L91^AsfNhr3eO<CW$VxGV<fttnw&e5^8VPMF=C5^~IM`qEQIyHOJ+Yl)1hsdi
zAuzQ+wyTWKw@szc7@&wberd%76`s<);9@z4ys&(}=Ab+<OX)e!J*u+bVjNTohJ$+s
zC;INKg0ktB#_3Y~)zFc#rdjz7baqzdoyg-WR*<&5+jrQR#!~Wad~O}?`v=ba=Uszc
z9@`BO2}uFUqcJ76#;MAoly0-FEq@-}ToH-}hULv6o3MhaetE$92b?>IF|uj=`U4l*
zC3~R14HtE1Ulmhf&YQ5Dpqf23*3h2P_;4JTI2>Y<#f*ns0|W%hi^O+DuDSGu<!%$m
zu2brpwbDYBncU-U&|><CSJyg~;oiJqZ0;Y52b*W>bH=|nf;Np9thc3RQ*1{)ZGTTu
z2ezU|9f%g=%LA1Z7v(!7pKmA6x8w!12u-IJzafQ(Z$|$9!}R{;&--(rA=KI6;l;~k
zA7QFbG{i=V{5$}QNSjyd=UEbfAG!K{85DYn^Y3==Bz2Lo-*&B|-P><7pR+MxU_FFm
z5+_V{oEJpkJ76^~&*Buv1^dwA_TEeR_9X=Tp*_fWB9uOwc-eK%{Sw`DNqx>eV&VC3
zseq=YPPA;*Bg64rzN8Co(QW6D{F<qnM3ga<&p{4-*2@gc<-cY!uphDAuXHAZgRLad
zrq?pcWBF{xNTsp-Y^;~sv|hG8+G)+a3aWBnS9qv+Vkw(Dd`QTVyaV56Z3|PqfA!Of
z?vw(`6)m9-^i1-i1)X;=Yg3!^#OX(yDaZ5M$8(89D7q~m?<X&0aurU0Zu?kFYQkaR
z;6y}5hE!FxCN*~%=Ph~T!F8J&$8g2uh7e96J=<@~G&b2?`IH%%65|zRL#~jwzs+c1
zyokUr_EAFU>v2CLD@*q17u+?LrH61>P!bxoIo#cZng7?4k;ueEGn`a-d2ylE`sV}#
z6*g;F8dWyy;<7Z??&hmlt=~#+ClPloNwjp%-P467n*8!0nyv7#@wj;JUni<0OcOnF
z+^Tg08w(4qOKBXQ&aduzRo+sGbbUEb?s<7b$-@$m-Xx$v(+}YpmN?`|`Xx6zZA67J
zFtgH)a)a{z{(&)Q@q+COo=s#^lPi|cTiw*wrq1Y<?eqPi{8G!j%$u1~bS^>{T(}={
z&%2T~?a?Z#XO??2>E&%`%Y55DWeQ`%(A7ePVcQoabw|Os-GYq7e!s8m?s*koJ!g&`
z$3%GJqz<XEUUSBn{Gt+-mJ}#pA?IN56@M#ox_Gf(%kd{cs5H4~PwPcrU(d>cgYRx<
zDkss{HSNoQ5-e<-RF)EcGi?c-x4b3cqvK`PBpEeo2b-S1o%@a^7jlDl^jq7oZTzd8
zqC7phhY*AfC11UydG(CtWvCYyq<i^BT3;HE-`pP3e>k(=oME)N1J~kBM6H~BnaKRK
zSt?B~*Qo!S*+Yda(kL;ZY2Tig9imk0E9=(F)mIW5tfsOUnU8KOSb^)qUsxdP6AGJc
z^u%B6=;~>k>V=1CTF6d1Jc$lX&KV-G6?*vXUGnNmT#q_uaqQKklI8iBmc*4alcppz
zspxR#5D-$`yHsM<AtC)XQ%=6&3U8y_1%ayQwP8s2TiK2t2?|a}bY(Ppa>`G93D89#
zQ@g{#a^Pa7W=}#&N>>>!rxnk)%Bt0=VnR&~&2**6db|X#&tgJ4nmAUYp--GYVvmO1
zw5{vbI2!SrtrW6L6>x^o954*AhlMyoA<GA|<j4DOe?*YKDIK<;pCR-xk+Jn*AI=L(
zOS5{Y{DIZ125r962`VI33rG^7t+U3$UV3V8O$=91^Ej_#|88sZLcY)4P2Zg7?I#v<
zT_++DF$t0fznJi@$Mxcm-Oy=QeZe}w4H3}Ulh7G4Y+(w!V+0zs8;C-Zfy)`nE>8&L
z>Z5|N>F8c02S}es#3g%{n_&97yG9q%U->3T-z==W{w(HG;sbv)!6Vopda=f2g($55
z(_4$Hy(QaqV4bA4C|@djaXLG}hG&V{ms07_*#_l`ia6E%f&ovBedesX-Xk2pJf2%P
z23P6LRwT+;<yk~jTqhTlp*D1?*pJ=Q-ye%0Igw6FGuIuB@4QUi-SymXlL{`_(cD-l
zIy)xWE2X)=!~2UMUqra0U=h);3!LyI0?EzSUY3^jMRUlvlaVaXkJ<6>4~;$jL9b#N
z&_0_NU5W;!mG<M|O|!Luj{LfhKKtxT+|ugu?pclC@Ar^+UZA!3@|D$gM_*6dKq;nI
z7D;oK(ja16<=JEH-9vtf@9O>`wY{}3$w~1_wcGI0o9J~-=R4-s?^Uvq?l<zN6lm1q
zvma4|87coIrVCV&n44Bupabs}yaG)jiN-Ixo7pa$@6<%s%N#%CGu|*&r1ve-D~pBb
z>FC0mzCZ)8blUFKMs#VJ*c6Sd)6>m%jyfk!<8ogH3s+51#k)zb!1Hh9D|lefO3`Rm
zaebmZn&8zq=Z^}+@iLIB=%J2SApdY9-eV}Sty1cL8v90~z+_-EBm|r~ablBsnJ`TY
zbKy+Zxt3o&?|giwY)nmu?9guN2%;b{$s#t^kLwEOf~#f1D~Ci4Bur`!5l_>{%J#eD
z+If-Fn*;;VDd0n2K_2-c?k^S9W!=%-yV@}9jd@b<N5yY-mx+BZI~1-v#0YWaiokLY
z<e5FDVk1veumbRDyhi(3b(5>6qXREVGH;an9UOQ!9R9bhzk5+*4b49=2)**DawB#{
zT-#H0hzk5E?e67~_K0irH6>4`tz<!G*?D#xTp7XjhdtIudP_1Dbf`7;!Rgem;*TzQ
z+Uc&Bgs_Ct-mb1%9d0SCI0bia8}eAA!EHvnDXpCPjmN(zg)<y-;|@$;><GjizqzuM
z%kUB>Dx&5zM=;ab+=7(UPi~0i6I6%HS^d8UW7!v871D?|P!*(M$k$K2*C<+P(cIJR
zYVWDss(7HG;~ubBDxc1Nec@;iO@`HOVflt6bbqFr&0EO3D?FMQq92sEM^g58C`=8V
ze^C>Yw34p5E>^V$z*@sXqU*Rw5&M3dT^no_o$+@*S`Zl%&qGXcs*MWL42R^bD@si3
zUBp7I>XVr!<=zY2FgMd5GY3Ov2NYDJ(qyz8p2f6phxUvM9SZ$~1hym#H6J%N)xRbd
zhD4G^$oJGZeRl^Fkof8z5wejca&zf?+zu>!7S{dkP5SE{GB$D&YTlN@N!t95uJ_Lt
zwpedFCe(qOrFfu$UHa2*;XqI1bpAkjZ}JGSU8!DdZbGz@b;m0+gGv=s-1^Vrdt@$A
z*Bm}-WXdd@T4pDm^6RNP6rq6gUv>Vlxpx7VGi@qUwGWmvmevniF>RbNHdEN1DSJRY
zBYko_Z+Tu^ToeSFi(Cf1^;s<+gz7eNv*>(UZiBV7s<XZp5M?7+w<?=EG;~b-zT+3^
zc%9XwBG=uznJN;Tidh@;9t00DAMuRj;L`<ym}9zy&Tu`@+NoabU-Erapka|ht)3gS
zU@UhFT9lwmVyHNH2G|n5k*JEwxMKVERO?tvu!H_hZ=<Zm5BIax2WK#d^762)rfqzz
zroVia=`m!F9L<V0{1>;rTmFYX-n|g_uM1GkX#JDsUbU}M;|Tu+3s)9HZ=LRVMG@}1
z1|a|+<w0z@1GtZBp4uvVu`w@6`1I)7ZH~7J`$Th%Ha}xHri2@LedA(};G1=NnBpmY
z^YUA(CM63e6q{Iv2gV~PkOffM!Yen%b9o9XRgRHySW<H8pF)gwXI7R6lZ_540H*yN
zg+ldS%>OwREwPCjRicTDT*X@^1>^B3soR|hWBqR<o(su#bie=A(T%2%XT8kGTzc;r
zIbGFru|YLq=e<kctdZqjyNl$Ag^V4jIy&)*hRy+A98O<kT(aDbI=WYP@WKUib$e;4
zwYkcM;($Z4hH#IYeOuNFKwyf~dVe<^<$vE5CFYY^&a<8?N?o09ZD#Z(LDsc0UXcYF
zzC6CHiVL{=Dk2T;o$3|K%@b>;on#SNTMC~PO0uDpts%$|0OVMe8tui22dlo*1BM~u
zi6SXQ>+M25*4In-7Id3_zZd`2JMeH-^~h+Z9(^VHr(Ulz_N#(z4`k;MM$GH*(lV^1
z(0p9@N&=$y)&k7c=>mW)>>AFAN8)HO^b<tau-*a%q<<vOUu<<rHKuII5il1W`TSgH
z#!2pZ-@Umt)of3i!7#PDHcM^Hml&&^G4Ew}rsTb{%wW}1siK5)BW|?FISOF*TCX=c
zs88{LXJ30dNPrhNKdwX6g8+HwMxns{u2{)ko>%PrgC#{Y)-sV|V6!MD6QDa}j7rCK
zv>LZpnyc1~W^Q9$fN}||Kgy)$l!Ko4Wys|NZb`2wVJ?Nkr9}VfBJbfh%JwMsww58!
znP$yTO9z9WEYxpb!rwaLzs_kgGtW?{&i<>1DWPk44wbJI|1%EpxuSxR^E%r{?#&oC
z;Uymp@U%1Ksz1x*(qDfGe-a3!1d`->pts~s!;xb$0puuJ4LK1+sWxB3d?Ujh!N4Ng
zbZ9IW)#~2nmN9w!TPlR|dIBTe^W@2GQzE{I3SD$hH{X<fmppl|@w2Q2PjTVl(7XEN
z<Dn$4&R0pb;OIY5lG4PBR|!uD=l|0)_%3m}_jhya4Jz>giSBH?m#{n&uO^$xC?WuL
zE#qcuwmwkK@cD4mlMQJpXQG)NE?r~&v&KDTdW1I)`$%A>!H6bSCMwzVf9~Y4AbiSx
zhvCv17At-+cD!1#HC-G0h_ooxt#96WnCfne|4Br$>wrW=l#BC`InD|YoFw~?jHP#T
z{8=n~0$>xRestNIN*$^uSG(%iSPk?g0#6Y18KXa|W_A0C(`OuTA!@c&tFiiK&KPqQ
zQ9#C`e0j<2b2dI?JQ2n!-CZM6>dFYpI&pFj|CbBy!pVV(`GfCdfNqHsA|@IS*I7tD
zDq%ZYi`r=snr)nDO<>KaZz?dld3qxKnx=bkKP`p5RF6{4{U*Mn_(_&KRQ@Q$<EKPl
z8l%cM=fY$bt^xe9_4XQaZFTK8<zySKK&^~kw-nd8F}Q{TR~9|2hYy_q?hxAyw-KLO
z9nHSMvi|e&(Zum1GDJvw=Yrhvdiw0K(+Y8&G1e{8KP`$R*ZVt|J^%{0J^{nGW_pUG
z&XwD}2XwK}LRSb|NV#R#V3vMZth4j@&yo+^1GB5x!&4Kz&@lE2FT#8IfVN2_mhUGo
z&cv2HE>{(t8qUC4F&O-~_Uc?}%4s&Lzis8X%x$00S$ZsVWX#jwX7ZlneA3xT!Amx(
zAo>|%G4(^#DYeJC{fBT<t_V{HmfRs87$Gi_+Pi-DR5&3gtBnq{=-vdGz?vcg>4xJ+
zaM`R<<g^^=-FrGzqZn4^h{wG<hxqn42v%cJrQIpV%wP%1@Q<+TX2xu11*aswC$7I)
zadED%$AA~ajP8SCJ`1XKs%P9DdG9a(j!P8Y#R$<WGa@j=y=`h%uwXCMkZ<3=z!i!b
z>q6YyJ=EzRJTHhYFeG*f`Bv{hyKnPntj6-qhnL}*p^`{PtNjt#aCqz+bl<lVcQ{6x
zI(J6w9;UcQ8u2=|@emf?tgH;C>=YFuL?K71N|qkG?QpsGM1)mYy+AT3Dm-+wLsca|
zkn7Q1ANt%rb4I9Z<8nQ-OAc_4Vjfi4InrMN@8;JaU@m;$1s9c!s&47ZQv9jQ*)u=L
z2`5UdHDA77(8LjD#%^2vygrE`x^~J`_3?#wmfI}tk;&4)zEySYE-f620gtr4vskXl
z-Nrv@{qHP1+)6TLTJAzMEEi6kc!S-bx7g9+-(Vr(Jf&z49pFLAVXxc&w0YznRA`%i
zR1(7G8(ZxP)iyGdRy9SM)UTjim<qjMEwmGQs6J8QpJY=b`z0a3p;W4teSPwx7sJvJ
z<@cij0OtFRPjU4M2YL+PLR{*7(g#b9SI19?Vz5Ymhq46q8-}c(?~~@BKJH=;15l4M
z`0C#0SsZDhmwFI<!;d;{Ck#nd`xhL~{moK-l5sl1+}eR;YIR-Ybjj0z?sLWw?6hd3
z>VR)Mi`UR_O>sHlM6^+!an`u5S6CvoNoCl-?)2l2{4HW-Y`>-Q0+*wre<hCeKxXp+
zAL$s9ritFT&FhBJYp-*nn=8BLeX!?1?L8ZZa=*1}yx1CYMx!I{$-)l_c*OD9<toed
zZ1Yi~I;;UQRb*mjUh$(F1ri-U=us}XXdl}XdvBGPuN^i(Lykzo5foT^<oUBte^4Z3
zPmATAhAi)PHTvRUraO9~Zr@7EJwt|1I5g$LMVtw!@u6*v{$7*Haw=c67Gec!uA$Ut
zd||@V?YjDlS)<%`s-cZP?QP7}A1$0`$~!OT7$T?&@~Tz_fgbQ2pudLI6P?l89dDw`
z@UI;QOL~;pq*4x7cm_Y+yW7@7*sHGH6dF&B_!us$&`H?aXL`Ee3^`2gbCq|l)q>JU
zGvso|8|*j&rLvXL?#9Ut4F~i|@^+r5BRjE}wO8c#{#aR+9;=(D5<t7=2W?S|$mFB(
zuVSl6e8%%>f^zGV8zz?0PkZN2+_f(6O)AknyFO=FC?c&{b!0*z)O3XrIzM(k_NY2I
zux7_CMlT5GFt3L{EF2hXx`zu(ewRGk+REh4Y%+OK$UudT3ma6cfVSnG`@8_fXU%vY
ziu8PUmzA5aVuVlmAtNK}@^s(zH?s<|%e^bxkMyL^kN4ADLM;YLNOfl~J*zc)NMpH}
zczl3d5tw3?k?cHbsJ|JSVIZ<lNHTA1`Ak!Cay?1>_T?5Dg#QVUirt+f;mR{w-I(;Q
z9lx6=IS8TR7$D0q4$2q*uxbPXJDAhidhOS-@oI_=Rp{T8^6}Z`#gm=4F>g<-pZ%q<
zs|0%$$A=t6nqbKutD@Pj_2inFx<ynn`Qj%pevIYFAn)5)84Y2Ec*s^6+wN~kES+h*
z^F@dmn+rjz^PgyZa>2Q5(#grP_L%>Zp=H%Rq`(YjVN82+mTpnFI=h=ce}ftL{m2MB
zK!PxSq7^b9uJcct3tARIF!-%&I?hW*P7RFEBd(3QrgK)RzTOrM9)WjjLjTa%SC9V~
za6m{MiW><O#~aq5KAow6Bq*qhz<0&!MZ5N_a;QZGPNG1gGpVZun!x_N)Dl}$h`qEL
z4rqT|0T+$pnf)84x;}0ek2Zes(w}wWrq-L}FLD<&Gba09FOO`#z8kh;BL5ck1|u;2
zOJtOksx>#|O0>!cu2h<3<B1LJR65?Z?>WUwZZ=&4%CU%_M|O9j8VTUx6^vP`)A#%2
z6)a&+B#=-Atn|5NM*2mqaAw+??YB(IFOhMfQAwD9W#k-AVXEBipw2&S&R--pof^RQ
zHj7cxMBbl4heIt~qmRpcnT^In(0Kd?b^A;42;vlUGaX9ps5J-T;_#Gq)GKVpCTeEv
zBnw8g%{JeO$-M&894M+AeA0gkASa@o<W*VQ3n!%|lTW%@=fjOoRpVQ;)kHU52}ry)
zr=)&sA12uQ^9JFnQSrPi5;Xs8-6}0D<l_HcKP<<;U)w#NDLE+n3Qi`MMTLW{WvnUO
zbIO+SsQ`z|M|1<}EuHBfeU3ZZPPRL0p@MBRv5_Vk-_4TVBl8JC=K9aU0CipsM4mzA
z7o0_Vp=q%;c@JpS0aayT6Fmr<e0{BI2d7`T`!d=J3fTaBY0=;kc&kYX9EYo?FOg+&
zZxB+$E!yRLh$hasXmKUUILHX#3IbmI?42<Tq3aM0qavT@MErTv31snw!D_@J_g$Xy
z4dy>Zyth37@+Xk^??qBU?RDSm;#=D=mROs=FZCdms~OjRELi{x9>B(a;&6ln0FNb;
zA&Qo?OFTf_PWB5iIbBZLN<uQZ!)5E7u{;UT%0{@>?)pi+rc*?dbU2nffygN52{-?D
zvx>A+*lfqLp!lKrF3ytPQ-WgT&#W3pki}Cz39x}>8_b`{Ym{xS#>5DY95sr#OO!9)
zhc)>6Wuu1{+U;=t%*puzlrp{ccp*5)G9oEJtSwFZAC;d&P~Gn?u@(J(rQYq!L}0hP
zIUBc^IRD53T(i-9;k!N;pSNOOeMig54z1^h-mXZINSRAf_8>?aKx0(6IkP%jeDO7K
z#vwQ{!aQq7EBfd4btr=JCO@wkQf6_b^Rqf(tZQ)<00|||viKz)Rx-BJ28MnvkA%Zx
z)&IlOkTqM?N4-Tf?ayA36#M1MdDz)~mB@N<aeeCKB^bY(QHwgvw%s)VTBS(3+l$35
zf^qtwFhM=hWDEL+7^=0mO=182gLilwU07X>UviN}W0g(?#O!b{xZBjIf1QKmX6ZTa
zRW>+ey{j;^7fV|MjHA@Ihfp^1-<(^(h=?ZYi-CoG1M7sA(U7rRJnq0!FD937#E0N+
zi6w|B&RTH!^aeg5FLeUZrfAfAMo^1_p4JDtnhi_Mc~&8@cMH3K4r%Pf{0#_Hvo?t%
z{jm%d?r4p_UPuEns+r~Vyw`kjtG5rhAP3#2<9dN^V@|B*f}nR%Qqe`j^YI20=A<6}
z3KVGEaCrY1<(qjkj_oGjd3|%(mS*#@BFY=aa-F#$U7#skHNFH&WPwO`+BOt9E2E&&
zm?+u|cpi;g$SrDn!jkx}i;<849Y;)9^(iVRkJd_2ms$;LR##BNy<MO)->?_wan_yA
zR1BGod7hexoNeZ7E`4j|eIu-_@cD++!rFaQT|$EHbS_^K&kz+xI9PClck858t?zBc
z4hFOoM`q^1HlT2=v70L-87xZITB4YV<jg>queGe(B2ydAn8dYy)3b-<^asCChtiw4
zL%%zQizC)IH+6;r^JvP!<csH}y1Rx-t)xhW@^wpRl}?<o=^vNpN0VWcKu1OGYM_??
zehWk2NmBlG82v@w={qw*W~UhN<V>m|{Y4(j(2OfmVTcPRp;MBs%c!EI&p87ltKNCV
zWG-wv5Qu?er{Qd?J0*I8I_|7&_mxVGga^4on%7b6=afb-JPvXcXf&N5hG~RgGOiba
z+bI>RW1Y&PL3OU9DniQdww47ikL)bKgJWP37fGBI3KPpIM@jU&iYu|i-8o8liunUd
zK^-Kt);SpkP-ceZ=QH|9lXj&aM%8S$xDa5bFr1y1-{m`QtGgE_*a0PMRtNa}1zA^}
zn%~5PHZg1%-Z1N|PPBe9$c*9fyL(qu3cv4n(meNbDbi^nS)F^R6CrIae?(#2GdwfT
zI34M!(9F^{;p-W-6itcMp)=E5(yrBpuALrnj&dCuP=(DS1^6rWB!*?epR$F4&%WQw
ziU#1}VnJ98U$NEc4Zb029J|Sj<e7rAgLZuNW3?u|h8HUAL+zd8D1w+K){G0H0qic>
zPU~ArDWiPqp3(`)QPwbY6hvI~l~;UieFv15+80F+3X@uX+rc2+&6)d5lM^|^2(3l3
zr=qq#t@+If6|k-Zrf8kjwd2Oidx&iCbJGWsbtcT^xjrrjI^GE1pLalPfGG@d#r)mh
z|HK867JzTy08?yGcTy)jd7C-%#C4+Rd1u}m;kK;XvO?hD3E$4geY#c=wdDUSg4`3@
z31$F(lvxbMA^!2x9xiVCKEdG~kxV&;we<koZN7M76P)=5h@7oR>lyj{c@hIqhO-K)
z2Z301NOVa=0B`0qwb<MXq=E|l*T7$g73&@eNRiA)$(_Cg9r(&3IPL|V%j3n8`J(9D
zm+pJPq{i@q>I6sB$|2^wSC3F3mDW~yzrSR`U&QmjbB7_iCkv;-IYO^Y0!kHdEK$kO
z^qk_<HWvEn4d;W2AyUU9FPLa>(6bi|zMhxdU?;6C?V&l%noK<-v%bNi-i38#gYe`C
zesu|Uqn82f)kAH+#)&cIgOKA^WfihL=q+3inF6ImppWVg0E587#-4Es={!&F?CT<<
zCC5hbb2cn%Th8lQ80izS{%BjZf7=%XhOqwp^*J2hEiHe#*NQr^|A!zBceT5xZD=ZQ
z>Ws!qEVBuG52C?|SW4HQwtMm@g!=~v3=9kdCVIP2xL85kuU2?O+%W6q-v!<$_P731
zmCaqb4n6qaN2^R?!)g*Dae|_3GOU+7MT3M)41)vW^n|`D%>&nN>~d0Bo!22T_>jU@
zNGSG$ihRBl`FWVLS2Xb<ySMm|3QqLToM8e$%)63kFO<*-xmK-LLDYcEJLl0qR<`GF
zd1Q#h-xLA|B@)_?a!IRh=>f`_GuT+sR-jPlI+OB{sqHDI)pnuDdTv0=JN3n-`R&;F
z4~*suG2E<|!^Sys05__J1i8S22kLlV>*3?Jj^2Aw;Y~;dXL+=nm-EjTimI+gxh+xe
zC)gQws~&K`BSs8NcStUl-wZmi!&e}K>=eI7q0Aa=pb!THRQjgnqH1brWwoE$Jx(TR
z?2mhU7*~lo>oHhtdq}}Wn?%Y7g`g+H_;qyL?P6ZjH1f^Np37l1NzFJz4uv0J7PotZ
z<o3UrG`~Jg0DMUb-z18KOQ|~RrPXg%Rwu3sG}sy!A%%0uM<?5mi0-X-7?|E)26N$N
zp!zDc7%h`M{C;b*?AK3wrYYC+!HvB>$IC%+Tjyf~kiAjoh4pPV6Qt6<`o5#~WO>M~
zQ3dYx<rVhBD&QI+@s$PuTIR^S+nV8FAryTl3(1-pI?`7CmnHb$4gky&?0y9?!G+!&
z6L73s>4~RO6RDjcJAL?VOUfQPaYUdj5T#e)b|iqdeNG%T@tI_CbWndt882RS)}~g~
z?6YzVF8<;Cqi$1^$R{q5VY6FmKG;E>GVS^JNf*3AfX%l9Y|@=SR4Sjbce5x7*H*vh
z6f)`l?yi!x{#6^~=v}tS8+^y_UPSPqAbd%}k<VjKDtvOIh*^nIiQ{76kesxu_d;2f
z^Q>z8@_d^Djx$(OB1wr{jM!v*K@wK+Q%Bpd=`0eF-R-idlSW=49Cs!utiX$Q^a`TV
zo~FRkkj-7VzUdPT^H?+-H{ySxOmhbv?Cw?q5<uERlFe19<<Q0dYv9D2m8#j0TW-54
zb7j3tHaNMg2V@tI3B@F<XVx>SKo?ud&4SPMn!v~BVI}j1XmerFSB@F5Rq?2anu(i(
zKiY%hTia%Qa<wdewRA@W)vT6YOuBy))~a1~6dJ%FC=R^k_#6wbKug!V^n>_*Z)n_>
z4Y%Cmyc)Q>0+KBW4wj5NYSyg3*}_KKzD)As<M|y&$Fyg7#%@ni@TSC;f{!ZcurePI
z%D0D^ZmHCj!$<lPqiU8`7BbTnmLi{LmEhvh;~qJ)E(7_mqe{c(wo|dZL^I*a4tyo6
zz|hxL5O=cqh!=8zsBdj5?m^tqwmLA|KKo=)`BFwNeQOT~0WV;tybr6n&tF8=cd&uT
zm&(GNx59OIoN`qD>lP323F6*ZjfhpF<8@gCWj`gbj<?}zptky+83Iq9oFaA;zIYoO
zvC2G&#>0JUuTRj16Uy#n|A#9K7^3F9&{Y*ir7v+l?wj+FKH6e2Sz_a1qZYP&&#|gw
zF4n?W!J*dacN$X~3o3nWW_TAXxVIX7>pxo%^&jiwP6tP-XYwS|>7pKZN-IMHNiD*f
zRCF99@=W4eiBr2q(_X?ooM{6?wP8NWy=U;+t26hbROJiR?IDr^P-VUeqp&MG_dQDn
zW=5T{G115F&d-0n6{)n`c=mfw7EE_tvrTVtpb6O#$YdyeyVF<{eB)0<HSda9p;DCu
zHFtTSxVnovGPc764o@u_;?u_LS}?<xi&*GP5+8a+08htT;4(K@En*jQ8}2}orMoRa
z=yo}cDQF21;gGGFI+OCBu@S?`+XOMT(*z2bwy-RuH@}yb1U((qpyC|pwO?xrHzK0K
z#SVLDzfrWNv%G0(`xP}|S2uuXm|w&Ew|s>E?adPwIpQLHI3V!QITRkAE(|xK^^#<U
zf?Ci%=yW<I9eXF|3VjMVvfDm+e3jw|uZvk~J*qMeWHz*$^1wQ^XbjhtH=G&+e1!?|
zY6E})`i44%ozE2H<%<3NBBlTLMRmdj{dk6=A@%R|<=9v3Gc9_rz0#jJnIppSAo88e
z&2ti)K*>G6*jh4QQV>78C5@<XI_VgG1pH6<zUX{5QjK+VF|EP)c{^Kh$}zcF4eD@F
zJ<SvQBcj<}Jn_?E8#ug#$wiq4co1P+3*7(x{xLaFUVwue*bEh=o7@c4vK}EyNOaXN
zeE7Q~bpJ}jpFp%gNq#h+W7HG=(M-jcpBeQdw9>#E09iO-s6T)}bO*-8%17^4l*KEW
zI?D=zj;`m@X;!d+Sk)^NE(9U67aQDfC-pIb;j2~477T3ZUjNnX4Mkl&<ip9fQBqz>
zO(`5SyU9Nv$4u4hT&C?THQt&djD{91ZT%cS-9GmsjPW`(oR=a>oX*d+{=0Al8ym0r
z?;aHLce%}j$>dg+Sug+tt&2!2Hb!W!`ZwHTyrVWZPz6otALyT~RiKKCNc(oPO9Ifj
zW4?1gM*)bAe{?T7LFz8dBkg4a#1_Jn5FK6B)@gh65$+(1`z?4ZI^W)p{mhUt%eu>L
zw$Q}X!y$~#(bfLV>zO)#q`Ho5Ob9U|QsbKE%biI6Zuy@x?VVGy!D8%Tk&p})d%VUF
z3CxGg&|z$B%hMPG(Gaw@?xy$>i=H~z>2yN0<);(lZqq5eH_LBEUe9F4XG`T{93~f|
z0$)YiX7<^0=E@-!rL3vMSAcfDefyTgEIC#LL^Jp^owXCgp*T9X*%{zyl1$aQd+2FN
ze)lxKdlv8df0K&LxGWCkrm1Zv4rDrix=r}D6gMKafeux2k>Ku~H1Szjw-L81G+k--
zBE!Pmo-!Xwe8hh;I`Z-3vU(PQ2Eoy{<$h^unpbmSFuqsvrh4Ng&Sau{;lM=@7_5gu
zQ7@GW2L4=?m*d5a|9~cu%{3cj{>2Xb_y5^V-Z=w92nen<=iJ9q<-f&3wI>a>cQ2eC
zGms(nV`E=hq^5s)YUeSSkX7oQN5w#P0U92|=0)95;pl0-4be<p{Lm8nMj1hX?iHHz
z@c6erF5q@rFU;vLqzgbtowBmPz}wEpNd=i79&7VlrJd`9k}95h_^fnl<Vj`v@7^?j
zcSq=p;K*&~AbS^L06?bnWK%5Pwo&M|R-fGsU**mXB0<5UtdYWh#33cgnO4BZFPol}
zZYm&2w*YYk{csZbc4{9L%IX>l)Y<o0&cmQ0Bv8ev+4&YGUO>`#FB#gL`xf_mBH?;c
zd+B27gF+|;s3O$zb~!tx$f|$O7W$d@Uon9M*|<Dj)MljE_Ck{Lh{VSyinzK-o$+a5
zBmuF6yFHDtFeS6maELU-*;tX^7oAsyVtMrO8BG)ca5}pbGfWm|Dxm#`UumVCrju{%
zdCnPsS~m99Jp_?BiZ&sIcSd)>E#31s@?BIIX-}ra=6mZm4wIcrPLQLTnvx*W*qCuh
znRa<eLqkEwfrcq=z^uAxJ*=<4iN<d&08sPS(d{#fQF#(zpxs)oFBlMR!S5>OUq}cf
zEHMAUkz7dAH(PzczD``UW((4IV)EBvZ9om7HYZS~N-9KxJ0iE0{`)YW{<ge#Rq73{
zfm-#>F(xthUv_i}%Gv@d3VhC=q2+kHEeVea*>a~2uyB@aR!(Ohkjw}J?}mcvUB-xE
zv7Ke>U=52|HR`WyJM<yQSh8wG7R!fr_39COna^A*IrugCFDXIgNlbvD_E#S~Q*|Km
zFUfVPg00Jf|LDris9|?<B%B}Z1tMusTO*vBi*jxJ&kV^5GG?QJ^0w_o<5ZRwK+?aL
zsbFbf<XPAH{g6+G1K(Savw!%}Z1pj}z44jy)7g3|LYElGdNuDc_$-@+M#|TTyzk`H
zDaJpnt4wmtFRX~>_re=ano;8@*$-=E-4$NuzXov_k@%#iET;Mit{uG0wUwkS76`|R
z`afD7^@%)N^h0;L-Y&x1T~gBnW;Yhv%@Gyk%9nH+TU=8})#7b)Hf43I_W++E7)p2(
zqm+cOW+x4zqXw;>zd^9od9?g%3!b?;Z*y8wgRaqHV@?cj`fSIZA&s`rQnS>+`dEY8
z!+iHdQ90KojnT!7W+V=W-~z~9uuhoH$-XJ(79yZV%_IQr3|ESnf^3LXq)!UO<m;@F
z;7-mZX+cbW$Y>6QAIJlz(4mVKjQgQe9EER6OH21t#MOO^6Pa+pDpaa0k?&QYbzI!Y
zMSgq)<`c*xYMXL~r|Hh<j*b1sONdnz%)Bd#DDi9IySyeN%I;)mqG(vgnk`71fk=xW
zfT_hrg9y4Ph{y+}qG1ifOJ_;yyk*|sml8m>J{+zAGIPD0-jVAAVa;ZTl=|S=LcY%n
z@EPE5y-JnYfJiNiMGRcHcZjg%i4ttkBR!?5&^`f(66|Iom8*OJ=EG0E1|6_))SSeX
z3H*?Vp&4qguy(5F*ZY;R_xvK>^F=fQ$`2+;h^W^0UsHdCi&jsmJ9>IQ{c7)9Xn-Jc
zICEb3oY5Wp#JXzjkISBN3h(vzCDXl5ERQ<qy?qn)S2pYLkM<uYJ3HxzeCU}38P(T`
zoE=A)(8lzfqOH&rlDZ9*mYwlbIxUc_Thu2+fK5aheQu_M;ciS0(od?#c)o4?YQGdW
zU~9?cuP#qH_7AMzlv*|CPtp!S$N2srVS^y7f9oaPVorHN(8{=O1MDEIC2`>#v_<|3
zbUlBOP%&Ao<zLM$W`HbZNfORLH*Ufq1d7wkYi0+z8-=MWGFU!|iA0Z4Q5oKRfeudT
zyZZXFi;Eqhx8&sH%<=nfJ!;%j`}_N!7fw7+vk$0Fm~31vAS|I@CGH~CwZ-?vvk^{+
zv=qK{`7|^v97f}(QZw&wci=Wm)d?IIR;CN^Cb55cb~7D2F_gX;L<>xggPD7Q0%Pl_
zL>hZf&p3>3^$3gJ(}z)FM~ziV^k=aw;`$y+sEa(>pUHDzm=1I(E9zbe9DtnZ_5R4X
z`AtWNyng`2o;h2*en|Xlsvd1x^Zcs5M;5RI=LPA|t78Hmrw;%HP|`Ov<RH`5SAZ#^
zNhlZQ>3QE5BJlJPtYciQv^RZmYm*jbZO4{gI_A!}M6ItDQT_lT2#}GJfD#G<|FSwH
z3B1M<!UEJOm&NEQtS4t@`2vUz*KHs!3c(rjIwI6x@$u^#TKDbYKT0F|aD6L27neBb
z%yjZ^5t&5g`%X=%+j0`aWo2b8cfKHT=iuNlELT0NX(2aP$-mI*>+PlF;v$4YK?x!_
z!nfX;s(SF?L9tG=@5XB(Bw118wurxi7?!m)<Jsv^J;>mLbFt3mbmNu))xZ%HFaRYw
zdz3IiHkZXak*f{R2wuy=D3t6cOAUl3xE2V~=(P}lFPZ65`Qd<<e+E&Li;4p6)p$fU
z^ec2z8uJ|j6YXTS$JW*0*BS>O<FNW<;dpEDw5NaQJJ7mhJ4k7WZ_G?39~&|6okNEw
zsoYom8*@dFgm^-M><q$o`m-Zic{vjoAPAOz(P*_6h|IpbNFBIu(}|UWL+E*vVsU@>
z&?H5{#RhjS0GGD~C{ADCXKfC;SAIOCMQ(=(rNR0F4gy|g8pMgu+>g6*6pR5nlLqn2
zoyn<(#g;kky0D!z&B<(0(O{k4Z|yH%breDAfp^25Cgrd6Bx_u%3G<Rt_D*s^sjjbl
z8uE5sIr|o?Oj51F7aJ|dU@gzq{IX^`d_Jk<7L*nMVgM6?xr*!P4nJ)7(^m--WK7B~
zkEfaCA{Tc!ZeeZ8swDMw>(}%CvGkz>?L3b-)L#Ssx)&puOIqgzR3y+t0E}-#W^FiV
zhwdh{;GfLUa;t;1_R}Nq{Lv6s!1v&x+lb>wbIp6tUS49k;Gj~{(^e^1{l<6;>}h%+
zJivKt^2!^rMEgV)kX!VsAT*|BW<27nD^alG)F8Xgr?fm;vj%?8&QgwfV0y9bj2af!
z(tz`lI>1Db?1oCY2U$|e%|mdXr3#0rBUw_MwnXk0C;R8U_^TU#6?1n-i=j`1OKpyY
zkyaWPQh;kGEdVJ5O0{2x3D4S0O=PB|uGyZVT|N;P>0|6$lqn)ct*H|O5I8}(w`R2x
zkQr=J<dGQ!&hcm`lJAe_mKVITm9E$`pshXh5=Fk=2dS<7(r-!&G*%pe2eP}x9{T+d
z>t~Uw%)oNgI4GS&l&P?)Bt!skL5FJ}kyf>yzCp(lQvN6btSE4uMyDM&sY5KZvBv@6
zoA?mgl`k;~8m$g8KS2I{Zam<Cy1P&R=7gO6d{g{W=9Jr$88a(o^imrs$l&$1R&Z4&
z-C+HAU{Zz}wvwyG+xD9k5O_8<X2}#yap)u}8Xy?d={|w`e&xg}<ns_)BmHg`mw%D;
zyd#n{`R>#+f}`X4v}i{Yn^A(9a*Km7H?-GI@q_I*G*%MsnrI&GlebqgYE}uIM>hu{
z<d|dnm9hFGZZ615Ge9ejuiC(D?}mFh?}Xd7+)K;c8c+z3;sKK8DBq~ybKo+A#)lU-
zIX}ECRCjQCc=I`5MA_05#+mG@r{9X5<Cozwdau81_7nGH*z)fFD%gCChLrXvmB&0F
z^P;`J<&ze`QiZ>Fi@`vY9QcnwZ!UAal#ho{W81Dtg`c`RlY`{XpP!pwjO(%|eM6GG
zOV!+Pp#A$GcD}eVfX&Xb7Mx)~1qR{(_ELB_mhfTgG)NF4q!udMWOYw|mM{1L6H@;W
zx<hyEWOhFGNg<P)e{Xh+TXlY-5*<h!NNAj%(wijMAO2grzsVReXxW}=sP3%adO3G_
zkQAs~Q(N6ni?onA1;DD2(2=$`%G@*u0eVw>U??hZYLsrFnHDY545|0^^e+7w&(O|%
z2xBzu*|nevREDz%d;x@UH{ZH%P@!Lok38bQdY~l`Ra7&ExxzrO6GX%+@twoJo#t4R
zx+;=XRE$|d7OW3|S;?a{%P-O4?)kHaawm`!cn=M<ZH@h3&$rwAT|2tDXnUp;(8(?j
zg*`)WG;|4S;kI44r3RIz$0|R*zW7N;b-Ygo0s_@0i2zbPCgch*hi^+ADGgBl(0d9K
z2CZ(=LAz@Iwu{*Al7Nq(`+KaopMLuV-CpJJm-{UteXp@vY4G-9OG(pIZxY~I5t!*E
zcBiWw<EOEfaHi^AZ0MgQ3$pH#i<Fmz2o`bzSb<WAcHKb<Wy`YJl^98u`Ntv}t`c}M
zm#$r1SxWLCarWg2p{5zCrjV?E7+Q0G%LeJBBoNI4K!6pDdyw&Y0*${6Uoy!4Xu9t0
zBgm6TXj^5HA2HcSQIAlM+f8SN&!+QQcjhuNANRh0C5D99$H%G#e^UY5K#1A8hQN8C
zXcy7ZU-S*H(x|c?UK)LWM^Z@t;q+XrDNehY%ETUQIC3T~;7C_lopo!V&U7CxcolR`
zwm?-p=={Q}e>6rJ2hE97^*-Dq^)p|lHcA#5*fwyuIE!xI9_1h}KTp<p>;4+th*V+!
zTC3<+QstJ5epo&0QkW%^&jPNY-8$F98e+NAy#7T$Zq#qjr?(^CUm@7hThAcNYOjp+
zQOP{GXQr*A=a*qZ9ACB;2v9u~XKhsO3y4B^!iVdu_nJypXkiE-qsUb8ig)W((f!qu
z;_8~0pUmwxWy%kmBD@y&pRvP7AV1SW{Hbcm<{`Y2dOaF3tHvP~e{158PrKLOq@=F_
zXI)Zk$O%aBvbm^l%I#azG;5xsch_8Ts;MZ?mJSjSWUw$79-gR6LLhUXj*3R{{O~!r
z)&h*en+ng-a&x;F<>x^1Ox^$g-{pY@3h(cY7o?a=4K*#13+2ATWgFaIvbB<YT(17O
z^~_BD@lD)%ESK?GCSO#m&V&|RTVYzKgaf=8R=u;faF1W?@L}C*CEbgS=llVPDdP~w
zkU*wVhWF}cfjw4KWXoq5Bf^4)_P<37%KAr1lD{7{3^&Ykv>r_5UCFMlxo)xqp3j;n
ze2MJ!nbxvtwRkyh#PFi+LyhAgVAYQH?j+^ar{kHzv+f~zu<=}w=hWc#M{bZ%5~I*(
zHFsXEzDl<F?g8`ZKEwicaD<y01d-8k)VDBzbpB&vaBi+dMss)^SrAl#p#!Eg7<BmL
zOloyDneS#V`X1g5CBN{yN$oH4{;fL+AH?D@O&9~G{mIx%q|04p)YvmT7q#kx=xr!_
zqe<)5^ACVez`N|`3Ql-~8?tguio$tF@Oy-B#pYm3DgF?0Q-%?c({Bk3V1ESpvxOs|
zQ|cIg5&*I%*7E99dApI3UgkS+uK*bc;p#W#H?KdJ<Z{gPaNcTj>YZ?0f~SWx>aTe(
zdz#XC%_T2%%(W0hMq;<h@D_@QU+ipPe6?dJX1X_1fS6=9@3<HJuxIjUAb-S9a&UwP
zl<qqyc^25#pMh(Cdc_?L?(50qT7I9!0C#_4Ka2_}EARt3546wsAyq8{QAL(ku=^ls
zc<*!BSQk(?Dpk={;0<T$>1F3*0+1%eGDwOxlc*N9M!3$5BR~vMg?;ljn8u%ZpH$rN
zqiii>oy<$k5_iU^U6a)h><1sfe1#e>v<i!4X`R(<9fRG}Vlb`jxsr+7*OREg^_L{q
zJAM^*|F4)A<_}7X*D0Tm%pmNe={!FGR=XRQ&yC@mYua$Yr6B7$m)jWHCh0){(6)w(
z*g%oj8(a@qxOR^4)sulH`BdP_DrdC5n}sGY^q^+te*X_h_vUFk9)7}~>aJ&@+Zyee
z@UsJlC8B!Ch;x@(TKD^NPcU;7UZkXtl$>C;qRuR1Jc))kdN}4(_N}Lb9&mutN{nOr
z@t3lR<E%ln?&3Pf*lR}U3{K33e~$S|Jx9TD(BHnWTp80MI$L%UfLBVR<^*@Ee50H7
zb?0UF-)ca6qhXkp=8Suzp+Bxl?b)5}9|UUoZQOp{7P!fRQnPw{SM@1mPC`M)>J0?>
zH!KxR6=pE8963I<o7ld6iy7MMR$a6=g{<sTi`TuG&3_VZN!$L|NVm!%ZgR+^f=0>7
z!0pFPRwtrBeHrc-138B(I*H(_3b^D6mHe>#W)<{0N+$CTR19D69q({)=NX&K?OKRA
z!;-uy4oRFXkpH}udH3)r5Ec8$(S%G%s;Bay{Jn?o-oDKght*+<t-bh4@$E!2%MBv(
ze@+p{j8r_cy0`ab+#tnHm9bPmrF__d^Um8)Nd9fOln@|o&H54D+(YN_B{YY0R8yz$
zF+xmNM>0bb6YtI5lU`m__G$}wCSEs^$lnSJtgq<IawaQL;Vy7$gVU%RrmT{uIg^Ig
zGIEBDDP4c0w%9kPT)^EXsyC#6OVz9FY*T3Qc2sTiY3u2>&&gUukL~>Wl>v-54}a5+
zA=!#9n?uNelo!=`hOr6-(%xJUX_qZsJL^O)Q?f5+dH5NF-c@0T9i~5ea<bs@Fo#){
z^TvQgBSlT-qcv_L(>ETsn;I`l^g2;lcS>dH@9iTq0oP$6H5xiGIXXeT%w~g;TT4Mi
zS{F`Lqih7WTNh@JDI>$vq#HCAkd7wJW=Ar`tVhLAVWU2yf?`rT(0yh*?;^rq{PaJV
z24gQL_~jxCXX-zT52RlS4w?N4JsgWnoVb((pS6ZEUNNT!M?!qXOHd%B=yw-@u@ZUj
zd3%?&646whH^_@id0`Gi05-}RuOT=z!1Q$nlT~MPsrk6P;fKLsy4<96d=);3FdGkH
z*!Q|2XViCWL3p~MMJtUdMfYeBc+NNpx*{Sa$^{6Rv|Ahl_nF2or}_7pFuP}@lolTT
zbmzm%c-xLi?-p2^uN+`TS72&+aEK`3UFPB$N)9o*Ea?}lc&!~bk*3O}jTl;R=uFjh
z(-&**`NO27Z$2H9AJCNhDm7>QNo}Y7{D=xSY~fMa4pe5Kb%Q`_q83`1*wT8{5LX8(
zM0?gk*PpK@<nx^Hp^K?0i?C6BcNs-}!eQ`!ScPC(X`G!&IgMi#>*4-0au$#El;Fbp
zXI<dt;@(ukY<hVyh3g&^!1-fp%Ps_gxCXjL{~d$&<WDhnoiDdmNci22vMJ4oJ0pDu
zvIax_l0h6_a{TDbM`b+5Lf=shmyh#m6&DVtH_@?6biA;%q-3=jHtHqT8nxT{{U(p<
zgJ^i>u|+@b(-M6~wR%GD--fB0gZA7k_Qr8mAp~qu`evjQ<U7pyyk?S3Roj_G+Kd=>
zpYZc3E6C`Xl6S6z2cUQB<br*|TV(stA=#Gt<@y_ArGN~3@u&S|e5fv=t%m)jhP_nF
zswd_ZuGcYuiy|5EMSQWBj&`|X4RZ@AwGdp;ea=y4qR8@2#r<}>b{5Zo{%GHeFM9f?
zVgPp~x?0Ly^>IS|cHfzXjisgaVqKB)M?Eu_gy*lPBU1Rt`REA|9N)HEOF?2;_uGaM
zENDY_CL05G&4!=qI{C1g_QVF1GQ<U&CUJb#rR8f%rKPq@+WOjM8aomsj)=3cu+z+S
zqZT@ne))p^mV%HO(}><(%6F*R)yVI7FPW+o^viL`Tt1EWl7QFWp{}r{{hjq?|A^cq
ztJwh@%(WcYZ;}f>)32a}vCq?de1hG3!&slFbQtjmuvfm|VE)=GPbuB1b0+%0ixK(z
zT`4R2ZtGN?e(SdZLnns`EnBBGE9=3cA3I$;rkB-h;V!F@2{<7Gsm-bM6kKP|SM4u5
zHu|`V?b>7l_ZjgHmo{tn=ck^Uz%(7)7_0T+XGBaslzdY!y~Z4XR40Cz9dQRDcWM1x
zF#p6%Fjn@F6)?Esx1`EB=;f;7W7;oX)?2NIDdKBpmcbalRy3%3l!x$8jY#UO*VwWD
z+$WbS^v>w&yB`_gnTYjK+U(P;1%1ikG-pe)dYG0rPyMMAy{7w%n4HR_!|LZhFIjM3
z4efe6k)K<9O!E!zgsKT+h7)au>Sn1mZ?@21(G_r1TR#<_%@=Qp|NTvvR7m+t$nQ_z
zo^b`_Q;qyEm@$m5Qx5aCnSAtp@GJewP93YU)}4SoLl$`#J94NJ<cb;yBil~SaipTb
zCNoQUHa~@iwypkQWozH8`thfcJX#6YLfQZ;<;vn!+}F_7*|Ms#u>|EJ>UKF!G&jqR
z_<791fSoukVH>nhs6%{StiAFMycWgrP}@T6nQ{;7w4cg<VS+huLP6OhgVZo>sYS*B
z|LG1(&akR+g>#v@)m3D99anvbGlrY(Ly&o5F`0Cz>5%v9kPvUGF~{AS_M({zgP=io
z6%{Va%YGaX95#`do?z8LU^M%M;eK^Zd~BxCIS>`e4i?1n;K)qJs3$dh=HnVP(>l_-
zhbH7&Dlg(dli@JQ?q%}AxX0ewr-<Ihv5)hyvqng++GOsi<$b|~-OUpV8^cwa&{ruP
z;mepF34P&FA;ezEu6|@0*J`?j5W`=kf#T7v$FT6)VY$f)u6ENtSAlQlsQUgBY^SF0
z$2*$$E{pe{hG(Pi9h8j*aASrR=D<b%_7}?(C>WXBSQml$X=$%9!4%kEgE2H(Fa2lU
zLj;U49zXlSg-#!&Kk^H*`7D){7q?vqY{ECAgk+GADLt1IheFA@n6N7vqg5u93K70L
zOc%Z*)p)D06maqUm&;ZBLEaKe_JHa<{-5MAc|uYD0E!&*(mbRhvD2ZvW+?kL*~%Xy
zG*hkKJwY`9SIOA8;Bnb3YA-z>ZI1c<oJNRIT|JYkvgQ@|U?}(dTWJWw#zZ*88MNeW
zSbHT%L#mGb2tqRdKbo#GEULEaf)WDKA`L1C(%mH}-QC@t(r}PcN$GBo?#`i+?(Xhp
z==csk@Atz$Ugw$_?$~>+wb$O~Z{xU%!=@)3Vom7^XtM!(uOic;0xFYmmtZrh4rb!m
zwGmf$4!m<a{_C)2k6bBEzrXinW@bKe!oyLK)AP#xl4lR<gYM0bGC$rQy&D%0d9mlj
z?bHyt|2xQI8M>d5o_;YG#wMsCL1YmMKXpkGv)9u-tGdg;RRU<;9g;5iUqP1DERqVi
z!=&dg|CrB5isItp?82NMEq2zK!GHmhnmF#T+AuB}!%<a>B_<{r<E2lr<0NsTe;tkM
zbb;#bV?p8%TgC@cAmuR}64q<fZp56SvaGgQM3H(P-C5Y@>wz<o_)SsdkfK|7^+4yw
zQD?Hra?)@l52(ds-7ChLSAufgZhHhKx;}Ot8Z#$sRCap3Bg?w*STwu8>;2x;#Cvvb
zDKq=0y}g}q4O?izgS<^skU9oAw7UsYh;0-i5Wg^70ufXdN$Rz)Ps0!GCgd;ch1vlu
zc&|T0brHc{8)THTGDkbAN*yb4RsuSTtkt={vq4YR05kqY*TW+EG(@Ga36+_E`sw?E
z(FE?5`lzAl$UfN2-#*kfdm#VUr4QP3QF`oRtR?WMQEVIvIVyQd0f$21wCLJ`v|?`F
z(=Y!ba7)gc7)*e9r`uw=G}-SM5Iqoal_v4_p%Bb>$GC7^5Zn!oy4o1%&C_xknZ+1X
zE($Qf6jDMLKT-KzUCoAuhUUT@f?OH>%P%-ViUM?wjapQ9kzRH^<mct(T_;(C3UA}B
zB$2#ZKtKSDn#w7Il(sx9&-Bw!+aV37mlX$>y19=+))Ye-2^bNwG?ydb;5b+tWmR2t
zna3ml_W$e{VoPr?-~2<B#jMTeS?CF|1U|^7X~1m!5ms#|-R~J%q-i}cG-W0La~sZG
z4ejQ223>Y;r!FCJL0bpq?=m#>(fnycR;$tau~TI>bX()d0%|Vj-M;eki(=*VQl4fV
zS$TQP*w|P${<i&9uR|9>1^rDPNtN4^ixRi(zj`B;F6Rc9wwgovWISLTdJzo=)Yby8
zK&y0)WE$lw^N|XPT1`1Qx#WR-^PaB@-e>A-k1H#3x6P?>aiTgpB<UL{MMXt6qG~4P
zbto<A*g{HS_wdiwHk+uFeO%_J`C2h)1(FmMOWHCQbRkP~NgWYh`3TCNLA-5eZi+io
zwr^y^@32JUD*EXs%ohUtx5EKU7Za9m-295G*K0DiFWyz@?;~H?bJ#ndcQ4ZTM9)NR
z7<U`)Zz!&|7?ct>a)%D^IOm2R84Sf4)_XqUrAr|juO9^Ggt7#1QIU7+7%Z;g7iseP
zfdFKqo&(awkmnt+UTE=+pc?kw(Z!YoG>hgh#qJuv)nUpBYMvra#xKCKB#kW0s6FZt
z=lGK%VuV}Qm87(T>lMM4Owwnq3%5-~0vuO0-(rL>fnhUwj-4VYhMVoB=F{u?sa_9B
z1hwe@7@D#m*7Jm$uWT%M7XuG_Y*X~NBtAZW9`cvHUh_j6v&yd<Dcn`o+~|KWY)y2R
z&eXV*<c1xrESl2*NUyCGES{b-(|U!v*31LA7`&)Bc<N@%`1NI1xH%9TzrN<jMX?-(
z(T4rcQ*?<k8>zZxbF~K8ZN`+kA5QPl4v4k$mx~M&4yWl7hmxN{JIq*DNG5_rnkiP6
z>xNOWn>UTSvAr%BWJ=N`I=@RdaTTCqQ!r#JZ$IJ$l>-Iva2)OZwd}t}Ma6@;0xdtE
zBSRf`SB~+Z!v5dze5Zs&1XXglL+4fw_Jq6h$?Fo)AK7vC%@mR2rK*k=AR3Y+(LZ{P
z)`n+EPq&N0qe($H|9%N$D_aWLMcd&0fs9|*DcDUB!qVe>dKuS0*dY5)?mwbnTh~qm
zkrQAa&YU1Mka~t}iR%>}(@fY*X|?gayzs<!Y&&p+-X}{{I}z#;uAip+A#YPIa;=2%
z^u`9<Y3gr;V6mYaylKC1oiT~aD}?Mtj#6ZAC$T5YCj#B2=T!5;8|5LQx`H_>js}Zw
zkSn74mcRFY1>qqR!NrBg!6alqHCC?!7$c2mgW|~L?=qetA~d=DUSTJ*5j%P@!+j$M
znGfc@-&F_aXZ*Ntvf>B47{N472ZT(12JN{kr>8JP8ut_wiZ_%tqnoDoRymRBgp3BB
zG2N7(b5zEU%ggnc+mg-l#Hf~N`R*})NaQHo7QBCqXA>7c(y-JsAil16@8kqDGu5N4
z7{h<@T8Y~lnT{H$GUE5VX<Lgp|97yikpIJ#lV<w`X|rVqqD92KcqdD+o`n2rTimfV
zLrxy{m>_`^lTjf^gYe@um*jO`5`@Oe8|K24c^t7AeH4zR>8^@@b;;n0?~~gtU3^+C
zTo|b<=jD#-9hP}71=I}T1<LE5*|)^7%>Qxx=*ct$4i)e;5X!e6PUCGo$S=*DGTUYI
zCWbb&$ZeTfR!>F7q5~p#l%@u(1o5|1WZ4v5;5Y1@wXSS~dP--?l2m7bY+rQ=$YOd9
zdV0Z_H#(crAU&t77=2ayNY=f&w~Od8uCMNSqcWAWG7g;o&dKt4%I))vTyaKPu1*xL
z6i|t*A14sP<@mI(8e2`!4c{RK1R<Z@Shc@9b6~cW{BM?2tsnXGtdE}-k!f**_s2^J
z>35Qnb|<fxJhWpZq~|f0*OYUy!Po=hS(*I9c_-qpPnheU1|&@vqv=6@Y&PY*Hf&?~
z06e*;GxQmDqL7Y$Dk_|1TB1qyKBTdjww757(6+j6;DG0%GA=b@wg=D9<rDwg&?(<1
z8ExO<6ho>(>3d%57YKf->>ZHqT)|<7E;aDPbRj(nm`YPmH%uDOGU)2D63FFG5NUJ$
z>X2M+Gh&UdPo)<KrW^l(;pmA9!t5d-L}@5x1|a;<@}xFDb2>GVV@6lJ((_3T&vHp@
z$%;>k7p12n4`(cX)Ye<`uspFNeVXZN-%>Zb3G+pU^Hyn*P`>a&TTHaS+2ieDp~Wxx
ze`r|dH{A2!Bzb9ex-@AOcjh*0krnM)$@*zkOIowL84eiQ1RngT@{67oAdTcnpy3J#
zY8r5z@*a(NK+$>6L3bN#T^add_MyMh>Us|r*>QY_<6fD>&Do&dunG71cP?=hzItsf
z1hP9?0P^0n#`8zXySqO%Ak)9DoL6JXVOf=2?j9RUoEErCK^WwTX$Xz_QR{elt&{9U
zC=Z#J#0g@t`fEgMoea!0IoO2abu~J)v4WmWS5g#0(Xi4xU*)KIJ7}jyh{%KVAyn8c
zc^Qh`FJSgns*{?C?@Rvt4OeR?dK$zvDGLHjSTDqK2s~j+S`uD=MCeXaqR?Sk`Y@hk
z>!bYZL<^3nm;7Dbdu)h$XlQZAA^QE>ODi2I&kK07tyd{p{lM~fWLGAPD(XiK#!%n;
z1?uqsUqSI_uhCBn&=chVL%=V0Iu~Q}o$j<?JfRb%5i~~X<%d^TvTEo&oS6FVSL?M6
z`~oGg6z~qM6O6nzO$$_;tylORCFFj9#Jna08ZK7`#@C;EUizgL+l5@c;E)Ac;v3`C
z6#*TS`NW%pCoft=T*!A*F+`#TzY#UHs5ng7u`aK!YWJ7Egv>`-H~y?<>LyQjwe&C3
z1;bUrHZ$X@7B(6lRUz(>wSm~<SIgBwOJ9z)6fVG;3^++cZ(AdH>m@&V9@Orc6kAN_
zvp?tx6dh=tzCQd}T*)Bf7q(DkhQ)9Ibf?q6ykk7j=jLZ4^egdk<N?TU*Th3a$o2@B
ze@!?4k)9TmN@xtX<;N4BNWeb1#J8*P3A|h}LE<JOf76~%G(w|G{tcv~I<BTB`DEj^
zwJ)c=bD29E4=P@NJgxD)q<?cHv3}h)Shed8NKsQguc=4U|E?}LuMpTcIBiy&Z(P@D
zl+L+5vz?0rmzNW>rFcLihdL$*3>c&JSFgacO>=3ZBV`~3XU>GlQFV?|B-(>WYTUYB
zs(-wrkbixIGgWgL@?b{tAR4k>y4>=?^NIj#$eMC>Vc8n&H$>HR`m^2oG*Y}QHeo_x
zDNC}1_WAR;cT5lykS?w}WOWmUy*htn=Ihd5P%ptNedj9?P$~NoL5lY;sCXpLb0u{;
z^8Eaz_N{~at;!`_6_Es})SbNEmAiMW)u0m#<?jRWl-XFpo%Le)ow(ehiIk>mGzf?)
z+`43wxoQa4TP_IaW@fm?Dnv;^l?fpORdX>zOw##-Cm~>_jIpKVygpWC|6SWW`PDMT
zKO!cyfKC2yu$7Wu$ON2}2pL-H`#hx>!kC>Mbwr;DqjS=J7YC@(#0*u4Clm$CkRc)&
zLmJEhVM9l*u83>9gi87GN5*Z?4%fqRc)SblO#0LhO!UYgTD3`SfeU19Un;*356q>|
z286S!+lRU!3gukx+Kq3I?0yAy&h)n~Pg8vQmStH>B6B#^LDe@*&az5$lW>~H_I8Hz
zX60ZWBXnx-PYY4=-fsf0=$&B7k8PTRp!;j_HY^)Ny&Fb3*tlM)7tyY?w8eJcuDT%3
zFRR|hx^NF@Qd195f`qTBC|Z1+x?`9C@R{f=RK`en7EwA=DF6NOnFJe~Ca`+?+BZE(
zbk@=QXO0zp%_U%G$GiZXEpSfhij540;noofSHubU&xMl)=IW0dalJ+#7&uU9csS&g
zFX|n-p!-w;wfECP%6<L4aINiIQ`5CMq1YrQVOo$DS@m5LDI3aykmf7^<r`sb^SFL|
z@jC`l^|C^>Ur~fJ)Lt9gi(00d|6-}lf7CdWfV0sZ_IFy0+HT8PWprZckd02v(Y{Sw
zN<v=*-+rCLd*7-JPH+uMNLu_^t%QRwn+vU&^D^5#Wr*hbb`3qm#6tvCm3{^{atLNf
zgLCyr!3RyQfE<N<J7?$zGB(UC`EU*G(^}o>{U;qrOCK5iEwAWh5g9*4<iG{F)gh7n
z`9-dM;n9Tc6l`eseVx{Vmrp@`OX+b`Ent^(hs7o-sDsvXX}i(B`z4#QN1t&AlziWp
z$tY?#9|UJrHqLs@^3-Q46f))PQ<JI%j{S%#G~K)d{%cHu{oiSODQfCsbwJcB#Ti_}
zh=@XShkO-7fP-eI8{Qro{>x}};A*zg4U;14G}ap2&<moI*Bxn3Wt_0;I`y;ov`HKR
zY=F0@zPHV<Lk%a_pOm_fAuH~51AEhy@1nueFqRp<N}_Z+_lHWg&xI~tisAzrV6|3@
ze{6s1bX`n1nCKr>NQn-1DSHKmwzZ#aF(W^{Rm!FE;9#I&+bf$bkHdIJHGP{mpY64I
zi^V1)qVXYk%d>q8#}Ra)E>(m|6#wxG&){X+QuMT(q^p#z*6^#<Wjtky_^kkU6!Yn$
zy-t0t*ZmiorT=xReE?4z0QSBUfI@mZH*aUEba0)!Cu+4Oda0ZMMb`J7EC3c`DnnK|
z&~<vI%9C9{#tq+32N6wUX9Z1~xU~a0ghGtopG<w{Dz{R9HvB;CiLlhGB7Gr>G9`4T
zBbwt!WJK&w4NGxQ6+W?CAVj>HGovl)?GK^cZ70bLqu<_Hf)W9Xjj27UX`NYF%c42-
z%@k&8KlqdR8-nvXBxzz+Lb)rXc1snPndkR{kmI>-;A@H#(0!s@_+DtfZllZlIE9wF
zgIw`ff>U<jl&v#uzI%9_qRg#}Xz6B;!SjWLtS8jX=rxL|_veHBQ1HR>vXYs<yBqz8
zy^`TwbKNA6)(RlXGjofl$pcll`KnGqq*LRi_Tsw%B;MlBSqhPrlogDJ(x%5-T?flG
z$)lw}$CvOBL1hu7sIo9G_4A(EG9FX}4^`UF*kzQcKCvbXqW?gv%gg_P)@;i(%GpYD
zOw2-ok9nJnN_%J1VId1wG_zE3<};ZrB>dke0Sx288olnXu>&bZ<bfi$9gL)9^<DM{
zhe!$TYIZTcI7pt4w|Y9BrWDOTT0qc-cBIui@q`A#Aj}05xZ*&s?<tnWNptYE_=5YZ
zy7o3b;Y2!@^uR#k*Y?R^Fd^vV_vXtq2cF-dliHcgI}t#=i7h?ONu?q5Y<sUpdsKk6
zoi#yFUX7ls$zx1I@8SJB`;!j@|07%FrAQXgzk}G1YylvZde&sW3v^>T+92A7S#;zI
z*pPur6xLm>i&IxW!%r|~CeLw(ZoL$sy?TTp(u88Bhx~l&0->YXW`2ko1&=!3O3p&%
z&|3};&sJg`=JpEL5Qw(%)C8ak&U0GJFISX2gn+9R4R@T$xwDR>f)j5=$NO?<b49a|
z{yh|Ir||lx(-GuFDvmrW_?hIz+*i3aQO?;tspJNSrkgqflx#sKAD`e?dgJ;bkBc^U
zbGk7bB`I%P%+kq&ZDJOhhs03~eODeD(v?{(`)G&+>T(7SzXHOkdUGD(h9moCJZn!o
zw+p_S6fULXX2oVRj<gTUvp-UT1?Hz8`^M5F1q5`w+jtpWSsQ$<E{6cvsNE@J3W-b*
zJr6gH!tiq6dKDE}OIlenCtN?Sn{pABBi6PuA~cacrx072qenl>&n)hcmIMSo5MuJA
z^kO4K#-svT7E%n5Cv2FWX)i&Gi3o>7^v@ehH(Gq<XrUtgG^F=RM^NEG2XID=iIg{N
z6oD({8=YdW{f(n<Qx>iZ=zrMZJm_e)fZi>L1O#o5)gmioK$gl^U49dtSrUC&8)4;F
z9AzBW*}p!OHJkUx7XLt9COGGJcagNbq6XO92}0+DIM&SWZ~f^iyP1CkW%%*CRa~=?
zCPg+8jdP<MN|ycLIKdp>E1`^ptQ#5^=Ky;_yikOeyB&ORu;fQ}wgf4uCnIn~P(3T@
z^?R^<p<H8*fcr$j%jk@wA*uQH@dTDBP9|>d@idOVZJgt09$7!<Nsp_Ph@blWD1ft7
z_V8mBRJwBn&t)#WuLBh3Nf%CW8`<x}z@<kI)AlQ_bi;1XpVXKOy^c*+3I835G1z4L
zq-k+S0rr75X*0ZTc)<>E{0^y8G_gi}M`3{DvBihj`V&{6d#Z$JTXlldT^n=R>w+`{
zfv*V*Zn2+9sJZ;5ryTACwF$rl#9l;SD<lcr-Ewep*0&dYRl_Tc)?yTmCE@2qL<#_V
z)BG>_0uxF9&C~gc`u_<Ue$u?}TGsnZ@l=`V<M~0t6%qWb_(!nNW50QVrQ$W02-fXb
z!LB2awRqgUD0v_Icm{vtHaoJ6t-%!s@moqAnMwdX&QlD)O-82)qRV|3=oYd+wUs!B
zC5`rADS;g{=U;gV;vN*h6*5r+!kQ)?>yOT8L&>ve#DtLDFDC8v3e(~Zb#ZKIF63L+
z7NDN%YPqLS_4cDIy)iSvjrDCKOcf~?Hes&A6qR!R4yVOil^S|0O%}*NHuP?}q@$l9
zz9~da1(3GsuTQCDB!GUbnFd+B>n@Mgqc?`g7tBKsG(%9+j%E;%U$=VPrvyP3Rkib-
zTMWc+D~$1PaHcQs+u;-~3q{%GcT2Yp_-|vUP9$G@x|2lO@I?@&llyeAtV~APQ76qo
zH5fqzZ#;GPPe86q68mRA%1o>V7Z(s8=U|2wi*iCN(c;h_1@-zft~tD4d4%g##nx<?
zQv7x@Csl2a`|S<o1Urcyf8Z{q%3BB>>pWL#S}s-P3RjSxf?QasVbF}qyaqiHc=;b=
zrE7{)5kCiiCymu;PiIASZ|VC5RphEN{0Qdg8LsH$-)OI`+b7%*uMeuf1D&Z$ooDC-
zm|*1%b`W9ZK#^2Z6BkrV#2X0&Cn7e<ljqSsXO)%QyX;{RrAU(E<GWEL7sd@gD@;BV
z57boxpa{Xb$M&6!X32E>7M_>s&xU`JMzKlvn4^?uIge@Bp~>+*E@!z7b+x8Ikjo<d
z4ELP@>&{>CP>Q1q5qf}BLrUL<I#1~Wup%06Ih77Dc*E8Yz56L2Q>ROM<){4XKnwEw
zxkeiLXCs^{w7ubMHd)*Av0HD<xvRb9_0Gl%YpvF4399|b62Eh<K7T7yM;Oqw&$uq|
zh-0aF7u54iIbO-X6#uVIjG9<=)aD{7Q@N4bTcmsY5&&ywnDs78bn{as^T`pVo-*f`
z&By)nCK0(EF}J1C#BJ5l1gXYI)h`)n=u>zkEEH&MQoT8vil%Ywt)4`{O}gedQd}K0
zqq9D>oeK}X^WL0!40ZB68C96Sn=0s6MF}h03b$3i?Z^E}dKbK72rO0<C*=x|>D>Rc
zc@T*K5>FcFL9znltAWPSIMG=Q2Pz&BILFv69@6<pIxy8L{icr2<zP<tzS)wlh+=T?
zJlqDW>o;`s9ygO)lj4w(JmEOx9!d+)uxSJu(_bCMj<ja15p$`3Rupvz&vs-tU<hrO
zfba_u)~wA7$X2XPGpJXe5F*#+wqH(Y0yWTL<v7Z2Pl>zKj-&qN{O1f_clE#3l1|oc
z#(UVR<-KedC7Nvv8y9tnNF@^07bF=vTDpdx$^u11xJh$oUCQ?cStT)adAhyw&eZAO
z`N-d5kvgvafwQ%6pbda`a}BH#H8fYumv%#_U0-X_{T0k_tKtDxy-5J?TkG3!yq;fQ
zG=Fi{B}Dt<IaG|>JDj$T*Rq4fH#vwpVkK7NbeAyI{M9QP&5Vb-q25B0Saz9C;glXD
z=7ev2{9r~xe@!wN;kB<`YsRh0koH^#suuBU_CxtbP-}ehcN)}KHS%Y^4r1Xt!bbIG
zFO<eSR4Xr&gc)Js0$wcxa11^vek`&f2?HInkiqf?LN5H4l&TH2GlxnLnz!@3i;g9}
zQ^?h~OpyCGEa{5%KYUZ0(42*$P2!$RkIv_v`z1euSrI!Xpf6;pc}SVcHX<TXZq2z*
zJNNQ!$(8&RocxoCx$&2NnFkR28*ILLJiPT*aZdx3sI8KlK)fyLZS%I~QpTi|VCwOG
z`<#GM{b6vUNH_FuGROURdTeHfd!(gDE`jsEUqJWYFUS-ruP9f1iyRVjOuG!@Q$0AC
z^}W<qR5xC8cW?*UL1H0{3(J7PVfO-aOkw@>0x4T2iybV(MK`8VFXpxrDN*A9DpNI^
zI@9UX&jeD$;sx0Cql&J(!258#AF<08s-vpp83rQ#I``|fy;apCM<cJKqv=F)29Dsm
z9Wk(|QSRu}CQycU8Ak8IP5TT{7JB}p5<dyLCOcHsf70Z6nx4XQT)G7x8hz9*x-h9w
zq>rgfi3DAXOb_FuKr&y=kM{QaB%Ru7mrUax{-*1r@R>QT#K~<a^BZ+>TzeENb7$J8
zKMVsO1z2E3anV^d&P7c~1nL4p`%UeQj@{23D{Wi{Cp+S%QV>T#)fFtc6xms2u1Qf~
zi(Zx(p6d$OJ1rw1pWEJlXa!3OQ`(5kE9LfJ{m@V)t!o<a^n@;zx+H=gFlm~tGLQAN
zWWE;MsMBtLXEDE6ryZ`4A0+sOmk<B-Oe<&O#XK`ri)e&FZtAR89W1R%zrxO5+c6Vr
zdQgJp)`g1-zbHc3HdN4vBc_t+R_zE*ax)-?K^@E^L_PsXL)HSmSO4NHQx`vSR~Mu)
z-2wIx=#pD|eVdY{oUO0@<oOoV?#FXYpz~#JrXN$I_GxO!raMz+Lwd~^Jw8-{g-eR&
zg<ue-wt{5kFGJt=K(vur9PYZ$KZ(t$?5)%;$<|3M+e3PX6@TmA$alAUBB?T*J)$gL
zJ821A+G>krFp&~y^N1d5znh7D&cAd(;t)*-0poUm`A$%u%-rp?jgMa?Px8=rBCoEO
ztgt_rVa#SS9lP%G5mp;jSWLvUO=P`dKr!FYVdB%3xhjX*E*wd$7O!KxE(G|2i-=J9
zZIZ!-+vwj(Fg5z8LcX_n18y4u>v899R->LvVx&*OPpEi0$7q2f1Pc(&s5xYhkC)8l
zs<cM1CcXP3lq6p7@Dn5{rza)ChYf?0)`io!Osl70YYlg?lo&J}@%8!lck_gldCZB?
z{<FvBcxc9Dc3XE^oNm8eAUf(6Phjm0R5ex{u0Lp?-o3Cy97aAIPbeOTH5>mXmjKGz
zGsC8(=pR_<kUWd@03#SDCs@M4P*fJDviJDoY4%nc{7GP$-t{<1xz6oY^^=Hx3#eV5
zcO`-EW$1$e2=O*c$-fmpvQ3dyznY>n)FSFh6o9O^=leCW95R6MQsO|fcxvQNKTJ*0
zJ^mDV+X-@53*T?FUSS-1Sq1IbhckGHG@zOwR#J=-G9M0L@BY7;;3NP6jbo{lV#R@k
zHjK=}5*Y+-uW^ToOfCqYD5Dymdq~!5e#0Gm?d6Rp^AHP#>bHMwG3Jj|i8bqIQmn7L
zdb1levL8Jvs-mJ|#+t}Hg>hBU-Y)dH{%z;sw3QT!j6f~QmdF>H+@z=gWAX8}b~U+t
zGvLG_1?%Z(lHf3?OGt9aCQZD#gkj}iJY6e^i2zD;@Av|ySa)xLX**5AT3~y%w^O^}
zh0NO%eQmP?yZu0mn6tT!ThKJt($JZqU3k579`^cvEg_O)1x}xg+i_>Q_?e9nr%^vG
z-^Kq_2CD=9l0Jye9S)op@(!fts*&*ki#IcTg4;+Bwhq#ZU@R<+p|=uerL~Bjo~iwm
z;ipUfm%QiO^J;TT#8(MR>A87juLv~>l@^Gh{!v3c%vaUJRDCpH!Rf^FotebX#X+pp
z6UCuX(QPi^3<i7IYFH6v>8qu(Lw=aMp+)7lYGiPKN5+X4==yPuGBCHkz8nG8HdWZR
zyezKU4~rhWzem0T$dvHQJ-`>K%)DJf4*0>l8g*WzeZi-06Su>p|M&2<`ojC0ST{2|
z+Okig;)M2B_yapU#J(6p(PAwMZ^6*;HB>Rz(ol$-&3EQSygTOW&AdrDGym}6FR`g_
z!=2g-y?e>N9`YscTqx*<+<vG5JUM=YzM1n{LuUJvVI*>$nNU-0fp!wQhw`&xT{_C6
zPMosw*6C2tD)*>~?s8p4lBZ0h*$`Km6c#(z)ZhP5ZeO!~Uy{bz==CRoC+qJ|fsP?A
zGFaD@Lwz$^iO02ZGscRY7<>tbD}7QtQY{CEtF&Y~J&P}9dtz&|g++IOnSQ;&n=O_%
zyXveafsqSd?dUql)Liv<vPZEekU+WN)R~|CR*4na7hCW7tB6LGsYV%hwQeZ1SQ1wN
z-1ppR={3;2r_Z}tE4Fd|EKEgD*Djl`(j~g!7cD1pwX24<s&pJBSk{+4e1#(X3!>lu
ziNCMpvGbK)C$Iwqrhv;#6vA;z04>X&2L(>MzoH#C;WEYd3pe0vLZOp;2Ku)~j*m1{
zk&{vLk>1q@{vj=laVY|!&$4%!bCu!sK%Bg#B#Chm!&v2UE^42^BO&tRl<{%DK((E3
zZp>F`;uSzsou88QGTig&gGFed7bJ8QBz=qo{rt}5KrVa5WpLP=P17-bCtHFGWSVNP
z&M`jnlwf6xLw?m@2UN4k0+2MX7zuWsd6B1vQuXw;YptTxM!k)J)0bhSPWPnI%-2`Y
zE}Q)Hpe0US?N;hS(gks2Odbw6rF@@Jc3{1+v-;pzMusCwsv7vXjFMh}PnK0Hr5!Y?
z=RaBvq?>b5N~2VLM{0wyN^O9P${<4eH%8?r@_#iPa5#b{u7}7_l8?tvbfRAW+;A#m
z+{YA7X4ui2KC-2!Up^@HaQ*>|2A7;wQ^o@Hg=2Yc*Gb#?SDzh0w0Jt*LpQ_o1h0@4
z=Cj)!(Q%W2^vs|nmy0_<v*GH>CeZ%Zt#nnPKWoDJ;2Hv9+<D!%)8b6O$@eDchgWMy
z<iZ3k{Smji#W=Xnu12?#YBOr~R#3`&sJ<fo(Nk#V{Q%({eyBUv(4k)c#rtP7T0s47
zmDeq-d&{X^!s@}Dc~-@VC*8S7z++OBO1`NDbGhx&pR-a12~*~Wbs~5mmH`$>N&3jU
z7Tl}1(b_ba!#=5@elk;}ie#mYOTXDxiOEQVoOx!OO3PG^@%YnEzGp=a!A4LgB$)%2
zhl545TD3CUu^bdbk(-M3e!FyW?;aqa5{tE9h#KKv6+S2Hug1!nZSsC?t(S-M>do$K
zqG3eu`T2DclG9&<{x7Y(lHbn^!9LDDAN)4SCH-;SV6FyNjz78vD{*)y6$+8EBG}}(
zR8e1G2_)&P)Aoh0PQemQ;uCw%*i?{cZn*yPLg+zaOQ0i<slClMEm)x~b3yqE+Id0q
z!)z~Z@U-~|PY6FLe^~Pm#e|2nS-Y?5JL~G@UkQD32uOevpJhWXPJrfp3Srz_lg_XS
zZ{r5>ly{xaj$O`$FF13&jqDAfsBF+#<zJb4e|iBF7(@nK5>Lxdp@>y<EYTDFNZDEv
zaw631MPC(4;t&VIw%Tjf@lZu32=G1Rl-Fc?e#iVqgC54*mf=kHU~hHV*+{~9@#D$w
zz0w5FRVt8tMZnXd+4WeTG{W%f$H&tZ3s@4BK3v#*VF_PcxqM|-!np{YWB#CO!;c0<
z2HQ16if`TBK4lGpz4Z<UGNz8;$9-#HE3UYHUc~p0#Tr;m+rw$Onzc5aJw2+TSKxp8
z*M+Oi&W}8e%U9jWGKAgNN1bbuMfY@JKaunT`|88t?tDq(XXW&qw{<FJJk-@eXjvZP
z25O>cPvzi~AL<C~6MPRg*4V?A=qC<GL$<!~tj)9jJ!oNBSBMFS9l{AJR->W3=8N?o
zA2ngSD%$TEM_DF!n5L2`;{CjFiETuY{E=mq7RI7Da0LxDZ<*%#6B60h@K^RmC4oL6
z<EtJW6zE~EK#cTU%va;8nmjDk@UPS;bp4MnGL=f+VpvH|<*-&@jrBJ{Sl?>6OW_Ia
zQ#LqWUlAuS69~2^;2E!VSZZ0g$<>Oov<JF$lTFk*oGSYjik5msDB@hvfPKVW%mmM(
z9S$}zv9y`lCz#dMRaR?i-A_xUKml)CQbGbapdPOte4~wk<!r@AI?bBsgul-|33>ga
zwxtXAZF%?R^G~-gh-Z88^Cm40_>%kBG-_$thc2Dm%NbI14uVU{epc1Jc^%+R<fT>S
zQZQqt?6(9nWsyTGXzWUToP+F$JfJI@^Px)8P^gg=0kmo!cseLnmRHmQ-9wYG%J3qC
zCX~jXP?nTWmWKL~WsIXzqN$LUsH0Pyiit#(Q?ex8j3`YLAK-)h;#d~D(KF|-1@WBR
z9U_^T2tsB&IQ8`n)o<bp<ek^5_vZ^Edxj>lstO>-xZvL$<GeM1A4{v={zp2A)9sA>
z=H`ZzhX)%AO9=dviFp9$Z0$zpi2`L5Wz_tAofHtYbElLX1g@p3ZXzx1j_2Z1y7>9|
z%~qPCFD>B|X@C58S5Y)9QM6u2lzM35OlY8@>GXZ#FBXPhX2%beKvf8){v@E6f1zKr
zlPyh+GZj`7<4v?A=VU#`9!!0`;&fQIgR<Ent~`dT&6BDOxGv(PT{(=Ync1H?7J3rZ
z{(8Li)=)XSRe^nLaS?#SWAl8fV^dNFv%>lNZlU|O?sV_pfA{x?yF8ffa9Q?3Y;6xf
z!Dq@m=yeXLQ&n>Fy&4vHqSeu<F#1LA^1O%kfMTJ%<pf?OG@WHB-cEA5?a_%d5lLM5
zcOjxE3UE^HJ2lh?)FeXYbE&EW=?;$$6y>|V{TMHJN9AFiM#M{_E|H#SL}_g2EeRKQ
zmG+T^KGe5}7d!)YNGI>A3YnSb#%j*Cvp&`O{7s?!wO}845pQ%h<^ZO~w$YxLxx+oh
zT2q$FJw9F41@>y>q0wrB>B}Rwb0TFn4|GZefWDiiXQ%~>H0})bw{7eDlOu5(6mVjp
zl9JNbBz9{~kIRD;w483F*RNSETWIW$3(w$=Z_c-a<C0j-6($$|O#)i@Rzz&wD>?jq
zoNHtNllj|;CT)>8+yi7edb<Nj&#(eS6%)9Dw>~ihGA1uqHatwRMUFB+$tJn4L<<mg
z$4>4QES_8;*ipY}Es&FOpqPhf=Sv-3m12(P$6@{CwhJ?qbT&LRlFY@u8#6INuTt5X
zWPGIb`X2TGu`(Dgg1#}?Ue4?imMP})cOQEJUcrq0!k1-^F0_@)YjZWZ6{`^v(Jh}?
z+XJ*{t%~f<iq*XK8$(6&&{Z$flz}kqbm=ah24{%7zr-#kIHN6u)cZPvv))o)zb#JJ
zyLY9{?-U9Z2CIzB34i<h$(WcT)_cM$-Jupm#l^>V6G96*%Bd^Rt1C)og@g&gtkLMv
ze`7+{mT)Ip_;aTNJ8Is1l;_z^X1<-&Kt*>G(!|X(etAjoccP<9YbFUm<55_R!`pm{
z+M{|xOnhzOB6>ynj`ocur6%%+is%PBI=}#1P6R9#A??jzNH!*(3)1ebxV?oiD~k8c
zTdWA$n(e=SWY^1;$GX6E(3j<gZycN9HAc#vCNa1kPz{oFk0%39jDYUFPJ+Kwj&F(0
z$S97GbzAQX2ULtG^gCdj$wubA=c@CN73V+}iw?y*+U$uI%i6;5fY6T(9r(YC5q}`J
z94@D!1&9cFuoaAHOtYykk_>WQ%(Ks=zVM<s?dqt$tF{R@QGNp!aTn6SYp&Add3~~)
zgn(V3M1630SYBI8Nlxzn{!8G(5k^4WdO`Z9>Ciw_LY1cE6m6hap7=RMU|jjw7$Q<b
zIj1Q*{1qh>s($bcSi%}w()4~IrTK`oZCgx!yMz?LH5iQvi927}#0DvMXh5J!xJzT!
zUDi~d`J<gEWBf{OA<llMe`FJa#_N5hZ*9tBIefV{-zHQg9BRWU2W(*(5JA#y+|G<x
z9JzCNgoeEjv=;ptX-%!emcM5`V5y{>=nIlv&(%+(>Rs$cW6Y&$Qbla2r+VNqi#e=z
z4pg=5xZgTSN374E9RSWGjs4B3s;1tul8mU}f4>v8>Ou=_gM(0S*WQp3(@gbff6CM5
zOMX~bQG5H<NVzB~-g73|H>Q~l>hVL_+6Rmbq@())^sBHi>_~xfp#&e5_r;DpEgfBX
zRTVkA%FVp%<$*edRQzW-ISglK=PJwTFk8q&jjIin<HqJ@xnA3gKs4f@@Njf?cJ_%<
z-83e{Zo8e4_my|O!^5VhYh59DOup&q??A)m2^5+mo5t_?^l;V1^rYzI<aEkUEa2Uw
zu|5IpI??L3Bx90^U=R_7#Khpp$jDTgjrwk<0jbJ`Dlz+YzY3JWNDO}X8wT5%vW$Yp
z=Xc2Y*X`H+xL#T2g$tp`p>1P<JyH2o7tM4wxlF)s3)9T|PBbAefAfUi%(7LiPlf^7
z)wiTU#Zu1Dd`~<O{vD<MnzcZ<qoufjA817u3zsqJogs}DKj?;Qvus;Y>n=yD`u@b`
zPj6%NF}rdn2OrNVN46k`Vd8XKa=cf3ND1pfkx=L}ehy4B?Mf(yD?NtaQ0tni_d7E7
z>IvnB7VghfgypP)_;lJjm02J~(I^(mABR3tHUB9e8S`qd9@W*=@egsBVxnhl2eu>h
zSVvE9_-8n0mA!t0o02A1_#~sLuC83N)!b_&lsld=zRO4uI*`hjgn&wL0-e^y#=#K=
z&9I1QM0yK#_Sd(!4to>%%y(KKwj?1T;l0_2qSdSk0aFArGBT&j5fO{&P_}du`)eZ3
z-}m>LG5WXjeK9olE-u@nqNKoPxWMB(YU-fO%nxhp>*^41zS5Rk>c_`Nj*oIr^|F)r
zrk<X>)fSU$t3OD$z-L#^Hv7S8nGrEDUFC*7@q8ZaFA>qQU5=NGNm%6hKr}&!fzcX-
zLj&Xbc(W}oBU7^XPcal)_(s&M9Uah;dJ><;oH<RTffIrr_(uX*oII1NQckhbKEP+Z
zp)w{v+K#Ym9Uw17x@JLp+i75K*_EFEB$&f~rKeOHn)V<rsKU>C@Y1iJNxm$A;F0hJ
z^xzktcR~l`n6~;RL$?+Spe7*T{v@thoN!kvl%LBp5_~_P#V)I^VBsigi+k&R;eXcE
z-m^U@dzR>PqpA<6^G?asyuQE3&++f{%I@`HXFV84jznHvto*HDv=qeUetkmpbxWa8
z`O~53Rl(Bd0E{(~T$Muer6w*IFgd;F=5CMx$D$e{fN9tBSE;G7@o1(n3b>S#RxB<@
zS_rT4KI`b{<j-qVTj20{Ty~nVE}cO^T)|>ClFodW59IZAn=>c4u6#!n6BEmez5Y6x
zXXneo&)?GdG*@NbvC$jV`}c3UO0mZC?Mdcx46LiGQ<$bX?8f|e4KzQBcH6r@=-Z98
zpMT$?1WTs{dY;yDg)=N3d;wfJ;V<g72`;ULhdD?BU%d3RLo?B$@~Sd)YKgqRr6MSf
zfd{@k9lP=F-!9g;N@Ba*BX8Gy_4$#*+}_YcI|=qQ!tmslis-uTuc!Aj7+j(6M_YoQ
zeje5M#LA7KMa$VtpWIo=l*4Y#UqzPEbD_9*)8hEnM3jonP11o1eTnZSM7yiKJ5sd8
z18F)vkqvgqIzxfUtZ#h60tY#wo^cDV2#{6iQ&|W?nu+-nt-uxe3nZE0o&pG=T`8p2
z4E`7{jo^{1bLC=XjjmnxP9hTxQ=)!<+l8x}K5BNoo16Cc@84E4Rs}URF;`btd;C`^
zGCeHAypH`)LqkKi%UM%cJ7;H;r6#w|t}dK#w4veQ?UU6Gr{g7JAFsR1a@W%}AnCT=
z^M<*!=}h7j^5ZqW(ATeUAVkCq6}fQd%g@iZT5O2vA@L|$xKqcmrywV{yS*@nNn|w_
za&_gLot?!C?Ox+A(W-ARg*-}(|B>z|&dMXe@$SLI!&BSBu=4?D#prlJl`HSq5+q#5
z$^RpI+|BGnU_t&MOKMITz<gvMnT=^D&evM_CV{W5_R`5S16{`3rpVoeqyY!1j2%gX
zlZq^Ckv7_2O#zc=$VYKj?*&NZNm3rMTBLoCoUXfMiknqDZ1TdHF?EFf=L9uP&(MtQ
zjGwrK8rfU&d%+CUJ3OFFp@>pZJmXw6$=HdwCE{pd&8Y*GlobOfP+Bw6!yXl-RAD@<
zxvC|+qDzUHgV?-HXAv}b?{C_gKm-GF3r5fGe%Gl&e57lhw)XCzh;tp&!)$mW?IyxM
z+6sFjwT|EQcs5azteVfdx*naxKEXwcPX^u}s}~XVeQE@=Ydh^rgx7la*QcK6gB&|f
z+<&AJ+M}ryVNQV4)73Ua(hcxmwrDsB7|aqEk4LjaZS!_5&CSL8W9hg&uNjLqs@JYh
z*R5vCzQ@Poy?9}BugPP-_7z4;ORLUl7i(yUL#D#Q+FBZf^hq4Hf52R0wNMv<E^z;Q
zwd2R~$kAc#GQCbJyEQ>+^W_|BsU<6cmbUipQnQCWfK|g=@L>pu4NxmLfNyJS<A8#D
zC#I+OPL}`1sNV36a#o)t>oY(Q`bj8!#G6m+WS6fcN~~Q9RdtrGo6om@VsC2LWYG|q
zB}NGiP$<!Ne&dOmKf#!S`;xM<+|e<6JL!v?TL&c@bv?=xuV_ikeqIOE(WxTjQ0Yu#
zCmSdttzVy-zp35o;6Cl!>h(uG4pGVZ-fYfDBSBSq3JLBrvM_b9J8}AR<<a404#uEw
z4p-cvIkB*5d|fO+i}c|o^u2be=daTT<9JWlmgd(qSno10mj4Ly@|zNcKb%xNvw+MX
zuLtNf5x-}=PO}ZI7d9#B@cqp>K9>VI;O*`0v^y36UL@u_kF5k#iR7}u_IBSi@7w(A
zWkL>{rqyrYr2?OZy*}$_u$4%BjM4xNMd<u6D+~3G9ORNHC@AVxW|%O8NgSr&1{-&4
zlPKeue*O$Y5X$^{6@;(K1X+39I3#V2ro*WPAzz-?axFE+o(+DYASfL4-_p-q9zR;(
zfo$N`^O_UA<>8eoS~A&<C+%!+f>{;80bk0gr3xYBPqn2~^CDrqR3B6IsE^Agg&oyC
z5zZ5Db|kj#&8$yAgVwZfRxQed0CCqBY*9P6C;UjV^<MG5uZH<uZR7Y)Z`-a}k&6<x
zchFE?q3CbiPxngqu8jey<K(*H?-=M1LBzXY>88;5<x8?B&CR9{xZw(5n@zcX2ZDox
z11ES2f1nW?OqXa|%~s@azOT2OrU%n*N5i*$*PV<X5(kzOdGf|K`wd`>dK1S5^D*DE
zS>{lZp}BLW+|V{p)5GiL7JCoDFXOs8I{MAQLG%@jy}f-GAOIXzPs=IWH{TTmjra8h
z=>iWiJ*2=hB08B=)ji#ak{v4n-?;GQi@l>`;aHt;ml9{&Jov5uzm!Wx%n@;^FVW`C
zrk5T#C{bX}#wut?RoO;0)G_9Am@>3&=w;;;cTu1_XTj8I3-EUI4@8?eG#@eSij1sP
z0GyTVoOd75v)X=Y7(V*Bfy8#t$x<T4MuD=l>0p7_oH>v92L2{Kqz}<Rb6&#Ro(%N<
zXo0ADcdnWSWX_wR980Z6%$WH2z0<nD;6MTk?9MVoptrZTu5RJ>R|1&!Vj57ZHJ)XM
zh=>TSCfC!z=|?UgS|*pJ&~x`2AS?{X&Lzaht2JpsyuB*#7*&<KkQfa*UnO$b4o&1M
zCI|rhpj^>a3d$3wbC#t{?iXfFO-<CKb$J;1w+!H<Oeq91drDAw7JtvcM`%gOlkBN|
zYN<gOnsa~KP#P~quKkk?>hPhSN*&Y=*H&^uUR6~<cu4)|28OmRI)RU%h%mZ|tG|8p
z)8i)x$$NffZp~)FTuSyeWcAbh4bu&Sb%XFq2oD%*`(iK}NR2zQ?K~&F6&NTpX+E{j
zE~orPLw&V7gE6tm=?t=cw3?0WEh=m#<*kdYPf>>IJh1{9M$Cy~oY-I7ov=XMt~RT=
zzH{iSTd7BDMN5MTOr%7`rG8mBr2#3n-UJ~?8#^mYNK3wNCcTle%n=3;?{C9TjrQs#
z(t0L`PWz5Zo<|O-D>}W_n>*!Im9ZY+ywy%G=^O?5Tp41Mf%xxfX{6YcGM##iWoDxv
zbX->hu%?-Emo;y;#iskNw~3UL`P^{}=!f<RSj}2N$N~(2nBwQJU#l0p<3+lV)OyEl
zW-zi+$Rz(|fZVq$%JM1a%;Rk3pS|wQ7LN*I$OOY;_S_Vv&_ZLy+V~8cBVFz^F@*rx
z_Md4w>*BomN-k}kqaPkWhlzj6Ezk$d-W((~{`$G;(HkP&y!)+M<qH_EcCUp<ednec
zUt-2C1Djr^vpkl3aaSpAW{gW04V7txgMrT3i!A6$)NU#6ToKwVQ+77z@}Pud<Gwi8
z8;j7aIc6*+kbJ*UUSJ?X#$#G!{4|{O>2HIZ$<~7-duI8_$<x82i<UC<sr7Vvv*Ie2
z>O)@MtGe2}x=IB}y793Z;|}QH-reCedu34zU_5(mqF!&Rany{1i~J(<(>o8_qx%Q@
zH*EMW@AJ~O*V{T?UODn_J9Y=`-!%>2jc0ni$`Sw&G*uB(8or&`=d5|+8!7pQSFf&1
z=EG!*_c<x2>b8GB=5VxGmzQu1vh{dy@3v0MCri}QULY{9?mXrkj7Y1Ea%axx4n8wE
z0_d7|I7=9`DbZRN`Z6E+jhU01w0j|us5cUJllm!h6OLiYB~QFLFJ6Sy3C3jkNTABJ
zWzzb*d*%C1i^J_~WA9fN8WE?Qr4Pr)j{&W%tr~ZuW~^FucCafo;$6IawhLlQjV=ri
zw@?<dA`a6=5W&RXi@Pa?o~La%aM5X00qMfQZ=abOxZ9ENC~p}(F0CkJQ>7r@4|}{d
zLLH25K3D7E0wpqcQ*3IMf4P0%E%;Swpgq_UHbT_GH8O-uD0+B|VCcQ|K@)w}Cx&cB
zvapcr#pHR{=ov`ZsMpy$gol0kh-N4Bpk?QvP}?U7;nnIFPsIoCI!UzXG86z*d}TA4
zxAyh`E$&qh7IPgLx)<|o`am~NMrK|J>XU}lS_#V-`oUmEC5o%&_ZXrx9J`k2_H5qE
z$1B;er<!2IZJ9sPN_9H(qXldS8f9T$p5_BpieVk`o$O;@h@)-E75Q#BP#XquRXYeU
z?W2v_&iP%q*>`&By0B1E`RM{AX)f!Dj;udqWLAt(W0|Q=a5a6o>yv{K4Q`sYr4WXQ
zq>s0MOoIcQj-8ECWdq3I8M}WHDvBb*yxJR=$=%9TA=caf(JvpZDxT1BNR(ckq9l?o
zg{n#;mqJV7eMhxcoAzNXVkyM+3TdD?^pzcZdSsE%RMX>2s?jl^&aMXdJtWUUv#T+z
z_}L1nK8|_zv6njRu_1~HB{jz_P6c{NJX2WwW`$1!rUTj!)6!M#I8os;xSeNwT$8Nh
zC}%M}{d47@`Vj8Y!qS!07d#er;ovm&$^OmF+T|k1(KESYX0Di;HZ(9Oh~sfSaP7AQ
zGyY^h2&`B*?e_Ko4@2g0knsgkrn<IgCp7&bF!dT37>EZ6O{M8Df*yEsq|*4C1#3P|
z$ndVyz7mQ`SGH7p>NXg%;q+Zt2tw-s3s*Z<MF+OSv`Q*5SP;wiOu+Ew-{JXQP{=43
zW-!z@^qu=96lHhvgYF<1B63U7N1S>Q_afsdd#pW`7r6cybA6cmxBmA~r|ut)`jCnC
zDlPA{`+zsR7AEkD_wUD0H`qLVBuo82nnxe^8Fx(ty!Kv~cjsr>xO8TdQ>IqGIjbc<
z<Uy=)z}rU8YR>+Z^_Z^&U&ta?+N*bkyAS>f!vSAYCYI8SL$8(grOG1LRYI!5xoHh<
zQ7^5WU%==iRn7}<Y2ikJw3PE{9UkZ5$yuOLKZWalTtzJn3EfS+P0VYPN>7#E0fnph
zcA9tJkqvxNA8F~v4g*iNmunfm)Cqhofdtbh)-pCkZ^`wGL_Z_SKc}6;rH1QPrB8T-
zaw?wG0Ph1%4wvv4YH)7S7xcJ9NgXZfA6@xxNP$7q-p*lHvm>K+k*lK0r<x9KM}t`%
zwItXtd=~sy(B3@Xmp5SInk4W?KbnShv_w(zI9s6-IAH|~o)R>n*k(LMihQB)f*5eC
zkEbA^S7vYl2S83a3p`$vv6u|xm0uQ1<g9^{V2U5T|Nhn#2-eipWU*P`R4LKw%@hs>
z;u-Z{fLv(OqX;b9hY<6`f~pH9G4bHlGB6y>(30mOP|Y;SDVxeGZg0;7X7Fl+bX&AP
zrY0r?ye4gS(jPVt-7iZ5gvSTo4x2iY|G$?26E0)*@Y;&bl!K^ks4E<CiB~L6ls})T
z4h0;)CHrI5_GE;I(*Dw17C6f|x4ND-7+HFZFR{Dgy4#kgs<Sjkcrr(|g_|Z?fTt2;
z|4aP#n@`EmhW3!bP4KeVY>>2nW>KW^FhB?I(%9H~rO!ChytP6v7JfX86j-bIJtyQa
zt?0vL={nR_tk8IU9wO1EEkJnv$M4NZAyplU0>Y^ga83OtQ600E3rMtfR3&K1aw+$n
zAW~_nlS4kbbo4uJ&ju_}PS|QAN(N^I_2}0Zx(yF*_J+A@R|!eP<d!dgtHbw~?g@y|
zmF*xaneTAh>ow=RM|*10Mn2+>@1V3^jcHBvYlMk4QGjuk$E&KQd^o^SX+Gm4h%w)<
z!Ox<tjW-_l0)IXt;L+NSHc}b8LjPG(^Ew^13F{{U9pTbFIB56un@yJSFvdL(Vn+<J
znrSN_Rn=Mx8i8}RS-x1%l^cMH^c7{+=lt)-C1U%#mfsMxinE2oF<v<Rm3a#%h~eL;
zes#2Xb5J=v0|5l^@$r?w4W2HYv;gsEwB7x+y+{~QZ#W?EsMfrcC!e)<z`@I#Y}_9!
zuR_7soTF06<#9<13Kkokzp!CIjSv$9!*0F-qy^?s2QWFr^Lue4At5O#FJ1#bHH!a~
zmHh@~X%^chUjC>1^ZDBZ9v2(v8HnH9)_zfP0*{{GlBEhK6|}Sl+WZjRK%P<1Q$V8_
zQVG<7Q*vJ-j8t2yvRlu6wXv~*o|YCAe4wBR6!b$-u0=<mK1TZ5#!A4~`rH`ayigEx
z?L+<QDoYY^ofEvGU<)4}yCtO!Z1u8vbAq3!WZ>{pXtO4ay{3$vL)Rk$N&d)4(AfFx
z^;GE?x#4lu)WfN<hofsLsy|mGPFi;u;8sZJobEnzdn(~TpDh_40h!n6E1DF}PI|Ku
z)0SrYC?u|;pE77TMY?*BCfc@$#czq|%{}F2ZH|&w^HS6?jBfT@7N{H<fwwqY&qNk?
zH(qo9`qZop@ty|H#|wocA}(3k<($hfXADttOz-%REypqk-aBr}-fT;1IZgzanB_h9
zq~_t*Mty5?V*lzZ=Uxr5Dhd@Iky-yH*(uSrcjb|xl^)^pZOq!h3%nJyO5g1CDtZzt
zgi?e|FcP2NqASIyOf0ET81BrkM`-?C?Rk^0<X2MFDyzC;221})a^Y!Wmon%9VbpRT
zThZgck1WvreWk}wD<f!QKJ*)??M`q~5COx7pjl8DF;DlwY~}8BDJefc|IEyco{34I
z6h+6tz-v)aQIK<VgX~1Pd1zc#K)TVnI<6`aJO#)BzDU><dbYOV;FAjr$z^4XnORw|
z`BMjLUBQ-G0`K1W=z8De+}=GSjNiX8z_LX|uhHZ~eO=v1u_nN=HIVSd-Tl`30S+F1
z5lc?H!HMc{uDWAvtit^%IXQV}JQrQy>Ate%$B@xF9H;#n96JB?J5cAG>{*_hOFTKT
z3nSq}_VM%k()|5fhCUw7inN5p8dwvn^<3nY)D;qHw=UIvk`NPHY5o4PcA(^HxXMKk
z<H$=RG?q9`A{<z|O1>Zt8`RjZZmIS|j2LEksW%KfeO1q3yul0LF&$PVlvAeCFXC|e
zHf3w<h=1RRGt<XaZa{hyS4n8{+x>5IeB16MVm8c0QnAJGT(LAdF5Y@*g;3j4spT++
zGL6!~qeJt>AYt3BEXJlaJWF;jnX`9|O2MKBXqj1kB@FD|tQfo}=J7H#epuA=r!=cB
zN_jL)k>=C9W8NFj&9UWgfPv(Qt0#Y>Qta>zKG7ujKcc=eDvR#@8bmq;Nhw9TyF);b
z5RmSWZlrqz1XN1tZlt@rq(i#9yG!yt`27CsT?;;lI@~ko%oY3Ed!LO7Jb@}R&BiPo
z=RrM+_W~Ee!WF~i{k|dEy>iigJZJ5Ay{9r~TFtQ9yvgrY4hV7^dV-^`Q!<v5Sz3}#
zZm}^QY(0|Rd+}B1uLa8cLK3KluiJVZbBRTDzN;J^!>dR6WF>Kic?lOTAY&JjSQkg(
zm3=W*TONI#YbC9tteLxi^DWgsa6_&$zYv<=1$8J=o0r*OBJb@{JEl`r9M}<Hxs3~p
z`FVKkR(>Pj-dR3q-Ha*81GE?e;5BHZ^ew@J?CMok8h(P~Hqai=&BcBn$V4TTg&svX
z^78U>Nqj1{{9q|n{Hj5RIc~nnnVTCerK1ZjYZyR*)TTqTe@4l+s%Uo}MqaY(>UVvF
zwj*^Kn(>aeGWiQhHj-CK{WnS4Ib;<|+{Agir+aHhv0(;DJX$Ro3uZBypFVa7Av7Q7
zm?P$9<Mcrn(mqwC3rj{zDRan8CA<mjmYD2o1E17q=9A+RYWO8k6rG&{>`2NvA9Azl
zzf48oqztYbetUW`c4&JwZ?V|+exz-+sm@gGVf3}|RI-?aHp=%CX$EV{Kgl-S4X9z>
zjD~l64{6to=vC`2G|aTIhlY&s%;TCh+8Bta`eIa2WKPMYozYxd%n?T8-LuvN+{`t*
zPf&sK<oQeH_KU7RL;)V@Q;~1`uPM%4>kh2O0-SFz=SSMag>xN)UjDjjUl2%-Ie2M5
z^Ulagwx3Yw7d~5DFetOXUuQqR;@9GnoY!gTljf78D^%XR6cM_OLteK4K)jpPaGz_%
z$o?gyR*nTFUwd)U|6<BCGG!uW{1O;1Rha^3o%KRglRQH)Ucp+7%1GJ3#<*21MV89Z
zh`T-1|FQj;GnST-;RWM`t}wDkR5ei;N?uzt`W_VYI6tDH`3eGoL@+^X_fLvX&PBo>
z#Hk`4`}aqr0046>V3gehe5;{XuY4cj4lwT_MGUNy`RJqE_L7}FB8lIn%=5231__@y
zaIFTJ3*ft0L`2d?Mz8N~E~#I=N}up1_1I%2bzOp0E!G_Xm*aKZ!h(l~2lzPFvuCZ~
z6A`kTpn<`aI|;v1uo6?uX&Wl@(c?x0h>mK3dTX+v2MrxW2euyG+19Y%@oE>l=at<D
z_sVCS-=ro)MMWP0W+V&@o4qtpRtM@p!~!*&uZ`H+$>@*{c%@K|4g@HkG`QqudTnMB
z<p|DY1*?g@b!!Rdz;;QtIfJoXDY7g-l&F2N@Xe3c(Mh6_J1hus#f<;#0y!$LkB*?6
zgpj2h4qB0snsd8uLOQZgCd%b1L)#vP#A-D}C8>E-yGtqSw@yFKQ$C+T-wl@h!96Ri
zK{+mafB*fAGDeD<M_QSibULO7iEK2;hh1`YN#g-h%`5hQ$K6h4U(eJ_!e#uk<kT?(
z!F6037ymLT;vy?0lv2xL#Tb5{ghp6Pb|%i~7UG8=pBAukloy!2D!YqNQQTj!WtP$7
zOr(gL3l0rCH*DF>C~?St5ccIhMYe2p*nJo%g5TvWzaXE<QZv|A*^YN)iD9c5=Vtq=
zbz#%B#A|4dx1#2e{a-6mT3(*vad`j$_ayVtUvU+OU%w(X-Cbxse*4{&R8)XM>}VN{
zOlU5rl!u3h#klu5K!(##`CwsT$ItC{4fl8o3kwN_y#-dgUPQ-Iaxi2(X~>pOY1$V>
zK}8kI_({qw;IM&a?=~>a!WM@HO53@)ImbHsS4eG*wCr5)FvUM$CC*n#*c5h{nL4&u
zkwY@B=ic%VPX4UM<tTr}%!>*c`-{3it$PNNsKSqU3S(pU2{x(3B0j#>1pyFZEYObP
z_oj66$K&0E#%sI}umL8d7~oWrv`jcL`Su;`LgXsGNMKKxpypqeWvKWQCso{4A$6@I
zODM?Jbl-t7j$P;UpOB?)jEWZvgw-PR?4Q{4is9kdYl|%{F;z`MpnLvG7RH#o5osi+
z9IrMJowqz~(;vZa-a$Wl;#z4rn5k#_!+i()cvddlb4#Ej4qiE0S$}goIH(!1@Y*GH
z{)nlD<C4Wy{eo0Zfa8+(x`8FJW?&d}L^5xrnT-EZs`F2;Zmo;<=~^Qb&LB5`jYhDQ
z1j!zJ)wmGqq(jR92yk=y%+0{iAA11*rUo-|?H!KfbS*(lh>`hjaBUO`<@DQM1kpoC
zd4G}uk)M<uME_8EYcBWmgAcx8tS4l1aGEmPOBcz8suS$_dS^ym#^E6B@`Kz+CGzXJ
z_jaLZ0?thaK34(o*qk@dR=Z##byI@>OzjTJA{5~%8!JOxJ0amaJ@A6T4N^}6k|^TP
z9^~GP!x}_s`ln0#gcPE)H3q>>4pjP6d;JPc{WDg_-tyS%JhnS2vbyQZV3PUkKoRSU
zM{K`xIb>Nm5@UB+=S>^C!l5d>U?YxCRQMrcw2w$1vLFfTS&;m-y<zj=?xrhqGSS@6
z+uiaj<i++YX!nppEKKJ7KK3I^r^_d5b52bo6Ld(B&@1%cgTneEVR`~nTwPtATmSs&
zGO5YofjFLd<a+kUeSOzoS?TFmQR&3oizZaWNp{q{uG;j>KeUWJO>4hfdQIq+pVj4!
zpsaAEkbl}VXZxs>$sinu=HVjQAO>fKgDVHsFY8z&t`$&SjNfDod0bW3I?LwW+#({%
zQ-d^~DuP@jVRXDfvokl?*eSR*V;W72qwT;uSzOyG=zNjPt>fY&(XU@HEZT%kkNBdo
z%%u-i`rF+u&$s%)AGj$J8s5kz_UwSGa};DIm5^g68sOMACtbZe);bz_m|?m3E)de}
z40%aSn>5v26lGQuF81~VVYKrEk6S(V6#bVUFO$rM3oD0xrpo3%4&FH>qJqnMy-X@=
z__&6*F~Yk;Al{vdG=ml@Vl({u*Uo?23d@mPs216GD5rt{sok*%1L1MK9coQuRZh{!
zpTo#&OTA09Q_<0N68HVNB)R}7Ejhp?QWtqzrKb9<ZLaQ~JX|$n{pzrO>T-ap7ggG}
zVTtDM8^a(&Do^=mkld$PA;zDh5+m#1V9!k^G^S^lh?Z#IuSBW94+OjZzI9|(xH;dS
z7y4Xq#lfTV6fPu|VNdx((roP{2kNn;36)raHpW|NJGNIm-kN;Q1NGGq>+lxZDh`oI
ze_i{CYaP+E?qGU&e}qH1IvrGttnXsNHtS26S_+XW(!!Cja(Fo^n%|2yJ~(oQ0d#A1
zZMC<BFk|~j;nLkN3Y6OFyi*FedMLOSt2;-!^-Ya4+}sUe2nyjU-J$Yo4$iBkFXDPP
z6u+Upjq^>+bhHm2;jMy<@CWKEK7!ia_O@xvFHf=O{_uXKdu5-7z6id*IEf-kTE;+=
z$=|#dKx)tNA<XpqbEN$MC$9NWl*brLj)X4p)D`}XnQ|dwfWuOKvlL&$mkTiy@`yOX
zG5L{Jy{!B6w)eaO%NQs-v!v_}yzvz9BWULs2>0J+lHuM>LaJVQ()8-wr_k#IJ_>j~
zgqn787bJ(L`IeI(qYwt4w_gzVUT&2By)%AI7HT?no4q@*VJ*m#ND)-s9;)ryHd%5<
ztJpd$x)!}rvFG5if3rKw+qgk))yFE(k`M2!&F_5KbJN{FhgOX@(A^k`%<o_j>?8kj
zoi1}JO;%!(8hZ7f|D9kZDZyYgW_0_3fNL)q1Nu#p`>L1ngC~ly8)ZTK1taaS6v#`<
z$s<lv3vW=ApG4*_8$Gk?`L(4kv}<w~)+dQWaD0H%&1eY~C&cw3hoVtBkA9kppgy<$
zdM5pqv14!mcCkdH{n`vBmzufkWZjmwk1^VYCHTa50b;_^r@xaMd8MN#iK4GVMLUdt
z<u<1+;eGXb+0cEvAql{Bs<L+hXXR=wb7-ViRCi6NrOls0rBz0FulW;vBX|qEji1dY
z6L<fp?lh$0KH|fE?7gWnm9pi`-f9RhlNz7E90KSZc*VIFQbhFF+1tMY#StiMh8jIw
z5&`#kW7WZzJT_X-m0XBK^N05ZLrm!H8-+ho?w3EE41{x86YS3HUm7WNArNHQQ3E0?
z7AW$vRSyEQ)Ry7l%HEwum2lyJg-xO!BQaB=S7CteUlb9`KP9uga-m16$~VwA85+>I
zCfbD~qrz^!d)}v~-Tk|x*w94RL&9Ja>Z0JSGoqeR**^MA&GX{N%WFv@<-Bd4G@z^G
zG=;4u8iF;|<3^F+kb;euLx@D4N<sW)4BjP*R7=ad$!3{z=*~d3`MS|@0R0%o15SwO
zE*^!B5k!3TnH|-(j5MB`+R;3S0RQ55O8^6brN|Ip)|09YePi4AS6j(WTq;nc%E@d*
zN<@;Yiu*&uILfUB4?>mw)+(J>5>r4Jvz^5_?DKUMfs+mgCzBKBgy)p-u8@3$Z}<el
z(-)8&d|`A^1j544x~jTv9vA<`<ZEJZvWUBdfyHOXL>nrxc;bUr6P7ld36u1DT7Ps~
z281EnZY>Sv!D9LTm}Dw5OH9bN>Paz?cg$3Rm@xPDuKLS<nU$$<9hzQTfwSX?bcy1B
z@0ci#30OtIYEhtzlDItG$_P7Ie7IK|JH!3rkt8cIwkLVD-}8s}wyUoO-m!B!dxN1P
zUirCk*=ArDEUiLDAH`qM(ea1zK%VMjn%+u6MLcVH{CPE1i={hAJV#-4y@y8ocDHD|
z;jsX-tIlHVx6Rh-w6o;D`UThjef8w0tFvcaz)RbZ{K{G_eRjT(-M(+e9?R{hM^dvd
zrtRx^eqiphmsV7!FIN1l<ciRWWw+Zfn;5)87J47vxmDybuUw?s7SjVGKH#1S%t#Kt
zIJeg??KKCqjcG%%2~Wu?ewHaXRSF_jZIDOw5S*?%pmQq-iVqNpjodq!;b)6_c$jUb
z-t?|pY<p!6k8E~4o8p<BoWA|T+x)I|-cIt0T)cY8@(TsS9!zlx1;E{00}+OHr_(CA
zE)SoJcQ1whwiSjyjaXVNJWmk$WQe!d*S^sHlT-;|xAY^_o=Zb#2vA)`yGtoy*S6w&
zukgOi&t49eSIy#FcOJ;eKyA-7nL1wP+!+R*O+9<VC=r7rkK}yf@~%QGX&-35kG8Qy
zVvz67r{DgsGQeQ_d#c*#arq0-lE83#Jvv28Q*)%$P?FDi?`!3GVH^o)FQ{I<dV(4R
zz3_<}&sW32!V>ZHY{(6yK@!EmnPUGUq$HVIz+lj~`C(0)FMZU#L*tuO@?y$~Z~hmd
zsU##pQZ?Losz8%L0Rh|ZXX+=C`1&qIHuTXsON+=@YQ8RJFdxc00kp$^OxNW-%V=EC
zxG2$Vj}<yOQ5xZJHc2wFdgnxNvY{Ri`Cq+}Wo?Kc7epuwN}7BU+LH74&lvx)##XnD
zxWI^3@5f|ZL^V3cS-Fpd{h@w}xPmE!qIO0(HoNm;$=BA=9BE_6Z9551`Tp|1z|(%B
z-xB=cd_lQJ36%3}XNAY9BQ3MYSUNeg-)m8}HQ(iu=g51wPdsyke<wsNKNpR&1ewD>
zm>24q6b5FU39-eLUN%G05kwOPYk?7V7fyb6e_1YlmHD!SH1(|!JRJx@oC&*ku_)9y
za+&cty#K{hF47x4PFdqoe!z5O<uE4AsU@uWDZ|DV9osqirTHT7QDronixcvJ?xrF7
z1NHlzs@+p3WxS|e)_4D{6|^kJ=JweIAeDP_wKy&=xd{4)3yt4`f-=g=!Xv2V?N8RE
zdi(mM-n~nO9`0cPb=%?U$RIB-?<=s=!oZ-9k_HT`&!PstaS<R1jvzD$eGfIfQTPzq
zvv_*_$?@;5^7C<G9;HZpd-v6w+rQqX-`Q}GLebXK!cwrq!WJl?8T7-pq3SL_MS6H1
z;RbR+HeH-71TNF|Q&zh?q%c*X;D-Za2CJPx-T8-kJw*j0-{R+qBQY_!!Iy9Ihsz#j
zn|G`%3=BIA7n^-OWSD7(E<Dk!2BJH4Bj|CD-n{CU5&bp6cAM0NOqedV{VXmRaUta$
zD8zCXwkpaMw?BtNnooNDrLT@rWW)Tm;YmmjT2rUP#+s8nenb?%!0~ohp{xFs1Ec%(
z=aaoW%_pY#eM`c~N9&Xdmfu`6@x643Kba!xgxN$O4BhxIEBp%gZ9rv`intrkRD`o+
z%e-q+2+->yC;fZ!*L}C>;Kn4Q;-~^T&-~(!T;}(4KtYv;2Rb>u;7gdP<2$+vD_+@J
ze>J}1#fosT?o$`wkO_}1y*zLVAZ~jHc8Nbwz=n?RIhi3`4h`Mvlx=<HbvHA@7XDnL
zO$Gkd><?@&pS~_^OKVlH(B$H@TR)~5-WDuOx}ZMZiyJ051iHss1)v7VY=z(G0TgWE
zezx@n_z!XbjM?bn{;qqS6G|O3;{Z@|sp%h5VD$Qk^=)iy&_qe|Ic|xFiavp{oGda1
zt@<zGU#%6IjF0RCy@zZ3)z0yIbsMEiNJ^OV0xKSkQIoEhzNDh;w>HxTiIuV@BX|Ql
z2LD=qP^A;0YA-jI*S>2nsk%(fYxvw244oQL9=y#^HK2+a-K|YSbG<V#qKbu)NK;N#
zlx}6Ar_a}fT##KP0FcJjbe_-g7fS?H%UqNkK(yphA85x<8ltVI?|q7bZ|$&EFSwj8
z^79ngns*%|i6{$pKB^1K|ERio$DztK(a%>W(hE~4rXmFoz+9DH*QX^v#Uv%6qq6AE
zAH`ozDaYfgFGxcLI9$Fj0FU^P=te!;wU+ee;heD%!}&A}EPV=NI{bX{TH=o~Z}L0l
zdVTMUg=LFP1LiKdrA~^CN?mWWBiD86H}=+shLPhnq*`k{;eSVqXi~y}764&@P$sO$
zkLc5wQ;-G}%SgtMQFb-+K3o-Y9*HUUR}Nj?sl5ED2**a4M7G;i0tu^b{x&6lUvS~}
zCvNE@Mjzh`yv(PxTLRn9nFa)1jk_ezb^VNMI|>s1;>2Ji#dz!6GBt*QUpk%ZhQg!v
zYJLqy`&N+P?k_ty$eYdXbG2fw3QgYgsi_%qcDw99Hau;08V`PHq%5P**H6~Hz7JHR
zP9q`#&8Jt^%=o$ktIc<t@8W8LDA0r@)##RL*<CBoH~$JRvU)aL#(oC}1rTjFf2gXe
z3K1YmD^*dBG_*1B@vo)9b!PTpzMAIKV`gT~ZqRn00dy`{`utzVV@NFxZl^rXdjysf
z1?G=%&X9;K>40WA=<<e#hhZM6l}-rnz}!<EFqvMyeEIwL?-GDM0T#qpe}DKz?;FMs
zA3neUR&7DgzhYa2jc`6WI2|<K#JAq;rI2KV*D=;KKIZ&LKVT^Qcr(Av?)0<>7GW@l
ziNmcB@yl`VU>NVkuH0Y&zqqbFo_5puGXn>DR9|rZ==NQkT@oD6wkdN)K~DJ22O=Uc
z^AdS+@NWFuE4D$MTo-VHJD0%qc~B-mb3rRo<H&+Bp5{9{Q(JxZ%2SrPPLNQ=mdgIU
zkqO&%^99zku(K7gYj~T_U*np7UzR;)+<#>804UN+mmpM(owbzqGw{JNA^n?3A#A(Z
z0$nE1{$c*PL;m1b8wLNyYxeZ0fmWTLYrjQqDkr&u6l*XorOMuU^}~0jwO0Q!tr_(h
z&0z~O0&umLQ8>e`Q5ndJc%Cl!?QntQ@FsB9wC@JfSOoRx+7%G_T_maQ*^)BvN{Rn2
zaxsy;7b$<Za9A6kNC{oN9x?Y^bhmAqgUrWUw-4OM2#TdJ@vm&zdJtDFJbC|KJG&FO
zg~rc*j=_!FRH9|3>6pkmsO$f)<7rDWJ&|vKZjU9;?Ez`Fao^9+R)Faez=$l5q(CsI
zP`x_^tmIg`FTUg)P5>D}ySp?*bF&oCKsJB=YhCe!!uQGu5+LVtt%fjjE{6@47xc|7
z-vXA#7Faic?Dy0sN}0>CMsul7aXYk8SNJcy8CnUpY$b25WA}1ev4x}lo~apMS2V&K
z+trj3P?rAv%t4K2fKF6H!#}B>!p29Js@{BgZ*#C4GsWn)2M%Drq|vaa^tQ2CCGLN^
zaLCa3`rySG7cb-v2?)lL%47cGsQ(Hiwc#GTI4b_wK$7a~UX|fCsTv|E-UtvJHc-zY
z3=+JRQ$&Ez<&9jA0tIx47(2eFDBRA#)Lr%~rKYgkf|s?4E5W|*a1AisMo%wMDQom-
z00cgALz9hEsp#Jrh(p>c#Phd?NVo_=@$RN=vD<Ld211@Nsl30hJ+2?D>zs6=5+kxS
z!38MOs}}ly0ek3aoIuMXbMHk`!sO!J0AUk%v8Ac0T4Q2VlyP_|vag-X&$yl=i`_6a
zt%pw1??X>(s*ZkF`X9Ic*^wOIe}6=Mq)8SaD`aF~_*P_cqBAUw6oezsHq9lRiS*WL
zC#bme(lLEj>4UP0RHfwrV}EJIkZ^uo&ySuXvvog7da{iAJccIcf!T!A$bm{J4EmtH
z9dlQH0Cqg_m`G6=ic}^8YA#}a4~OTqDrRHxvo@Zp(Ht#@bst%W<NvGOm5cy*<KGcd
zC)Ux?!NWT)xa*=Z9fEnvUF^Xw2}w$hqRBm?VTaRox_v_QibRVVX=8zf34X_3$f9uc
zee}yKO%7|>YYNe7!BLIqp}I^JTf+!{RE6Bf^{eW=2NI6Fj5dEQZVDD^itXgeWF|Ge
zlTv+26s6g_WW6rdx3FJQVa->GX=}3SK3eL)y%OMqN23gew_2-8jN@@ol|L0=g|4_H
z{r?Yp2qsk;fG!TRJ6XJ$o1L6jeg{-7HX4_8-@m_L|5a^0JMlZ?IcPHv)Bxx(A{u@_
z?OR=|?p!GEvU-F41tFZ&A#Pr;U~{FFC)dUXU0cg_#AOHWi{PACwiJC*adQ|GQQUBD
z<8YwdIw1t})(uw@nr<D|b5_t*&@_Beo5h<(Q9lg26}yqrMgESz*M~tt>sUr=E~|B!
zs-LBUg7gNOA?^5tEn@QuyIkV9xdN$}rvI>?niQ~GZdQVrRS{iNJPvs49JfWmc*oKz
zAvU%fP}%dFq3x`gyclM^ChbOILPGN$$lwvixTGzvx}U1$Rlw7tGtT0D!LwSGH!%Ev
zWCEw;Co3|ajvPD7D{XL;klT_nKUhu)xll?%^R#y@<D&6)`IUk^31V?Xpv@T;Jz+K}
zUk>RIXP$4b@n;_AKh_O8u>PSHn<1o&q(W4}R>E|h#nzqIR`2DoHwW$hS!yJT=<khb
z%6grT)o+BgjFgm=)iblk)~(i3i#^tb3czlxO)e6{-w*FoS8JLALW*G@->&O0K|2<*
z09S&A;WxRRHg-mVM0z&U^(1WUgfkiNsUyC%KAhF7(rNFcPHhTHE7;4D|LY}vhsb3~
zmq=LjVcG-G4Ix0p0Tm9QcgAs;V*}=1rrYU8w-KA7l9C3Pga+X=7wIwsdB)=~9C`p#
z5>@6fQo)(A_J8ZI#8ykVPqx(y1_k<Vd<_Kp>XLn<Z&1OH=&0y%;)yF2JHF%NhgTdX
zDhoS(7e>1*;O*v~Cj0o^U>LV3S!$ze7$Q3vBP$0cC#seY(N(Gy-%zM+`hfNVr;|v1
z(q&r(W+rS;RbC97Rlz{7caK*o3g4)`H2pcA8p~q}Ra&KljpV!+jjO5gi*JQC!k`&Z
z3bodSMPOl&;JPHrk|J+K#dvcB3TMH<)6~1KMVwrvPNmeK1=%K=N<_z;@D|#I9QTR+
zckKQFt*wz`v-cKzFSp(ct@(74`r4(i{pC0V^DL*`iFxdTJHX;5;B%z0o~@z;dM>~t
zSj3*&pRHyFEal(*d!2TmyXyvcutu#tC72?!p0fa?_{)uY3cjSifgvsYuNCRD!zFEG
ztKK#soL*o?<GzD|hg#oh38^!|U))+PB`DC=sHq{C#!`4rlcvwn4$yx5i@G?*pQgHI
zi7Ca+eP2Ttg*r6Z>N%AX2BN7>3W0*@Xd>!%f!n?ZZOx6g!&NgEC!$MnId@=o0w963
zE+oQ%kh#r?_*gZwR9B@Q%3!`*r){cviW1^YV&-#LECT%uzo4jJneMEf_;sh}p{{vF
ziOZIMMz%Mo{gvtW^q-`!k{-LwP5**}ubpJ9KntqAwe>)i_3TuW7e8oYiz;=1WC5_=
zO+eoSM)q?jGec=&g#6A7hFzhJlt=%<cxTzA-7AX;iV?%ULrT#yQcgPNXsO+tO0;B%
z{i4%I$#+OxTL9}3Q~<)fjeyT#p<&}%@vVF0TRXD4_EhykdTOQ)Vftk5!otM8<3}k^
z@1A^R?O-{41w*mmC0e(vh*~Jo7r>2|K~wBY?U`j|pcAn{LA)$n*dp7xczFLt$~g_N
z7({fa&bqi6u&|{Kbi)tc=rWe<(LuLKA%*#Heh=qnNdsPOnnU|NJ__l+55L8VL@(b+
z;J5k8boUzx9Nmkr;ZZ?H^8dvdmY)bgiM>W_KoBMF=*R)$IuLu8+b#+LiG3H4hl2CS
zR{-|}?FAQF9BDZh>SNtN>zKgtS#I@4)CXHP$e(d~#7MOAgC(?lXyq!1_dU|lEnPT?
z44-qwpGqR%c=NV$<fH;m(OT0hwr5H2jXwS6+IO$#nQLRyo-MnWqb3SZGu7HE%1eax
zP(&vra($w<k2>9Ql-c#KX#mJx;P7IzI_dS@o-tCDANAe&EZOH!IhGkzm8@_tJ=?tI
zI}-A04%0ao=4A-9TqNajpA>5Y>eguW58%Q2gUYaz(Gywn@S0li#mNSb=7j~S8eOJ?
zV`l1sgL0P7;^@-;Qw0lvAMp5IPY9$0{I@%XIOcw89fLC@D$GX%0GVd)+W<1SanU;?
zHd0pz8zXse@6Mfqq2_5m8+orz@3lj9q%S#Ci^8LW#_vF0i2WIMRC3s5=W8cPGx-=f
zyHH4f&2yv*e{Y)rrxhdEjD{hGM(O?l3h|~B{lfRg_UxVBcdxXBw3*%$U%W|w{*`2j
zD$2{QJd$m_Jby|2=UF%8j0;kIKqI#g6VCM{f8uC9<~vmR>+NJ<?h$s=*2yq<@Ad^U
zcyL3jHC{E%;B)0W?T@ckXLWkH%^G81fp+HTvrvR(MExE{<CpK4G%Za%KYktuk1HSR
zeZBeBm&ze?ev=J*(uZReuDea{!2-djhp$Cmcx)3_y8IaYsfenfqBwCdsNJnZUj5^A
zNmBk9SDmRJDdK;)L#u1Z!?r5cQ|JvA(YJnCiM4#pXH)qk{NV8B^R>6Rs>Y1*F340C
z7a3PBD0O}y+s^Aukx(#-*AbGUH8$Ok<O8@|$@DXRCd|rLmz31=&vt;;5dfwR@96R?
zz4Gr#2zVUO21IXnNCL`4-<+hmmPCo+n>M;+#Aon2I*NX(_zaq5Kp}-V*c3?wlH$tt
z(7_zv0PqQviRnIz!E;tcX?1V&8MA<~SIb0)Zt;FM<AB>$L{A0b^VXt2lRp!Fyo?IW
zdE5OX6fKdJ=#t#Fv5Mlq7Mla(*%Iz!b>J~j|CHVM?+c^O?4LEeXmO>q&rPM?;B}cW
zJpy#@`UC?rSC9`4{c8>bUi^_Pe`kq_02>2TE_-`slh3O8qfTVRgwlZQmYLEh+cr3A
zM6I30QH@kvOAMqMo)nrRRCL7EBFssFAAZF9Y3BMWWY<xD8-446$fGH8(9utYvjk6j
zH3JyHd?AODC(EbWQx4GpaqK;j0ow46;XeQ9#u;pA4!AM;%xiTUeL*{Tm*BGu>L+nP
z`CZd~m(5X40E!U|Qn@Q9l=Sz?PeJP`TkZE>4<PD?1FQSElrW%D{8xSEcABb0aTr8K
zMJ02}!lylp+Q5@REyC+`dL41Cpj#;ZLDxSk$p)vgr{Cx*0CD97soC5`F)K8EKQ|@`
z#+k=noG_>_WHd3+vd>$N5avT=Z33W8eZaos-g5rm7?3}^^+|(8Wjuf^O3X*8<NzsB
zLIOe4T8EwQ9g}v=kb>BM;Qw2d+|1Y6SjPTOz@QIQ{6FL3gv8WUa>;K*`TW&R?T#I~
zGAfxj-lrnuLIBC2lcA#-pcMsNE5GcBLD+7Oq}uVy)4U;q#py~RqH^8B+N=&nL)P9(
zq5eWyEQUvNANT?@hBrc`)KpX>;1GFm7!Xmq$K%gyELEOIV>dc7BlF8ecwd|_7Ejkr
z;mkF|Ju`G4PM7@pVHR+e%l^)1Jub(IFEnduJyXkz&Bo)K_TY9~gnVH_t-b#HyBFBs
zV#DJNTDlIXXW-%UBvFg6tUwP!x(7>cg>u#=Q5@58Y0rte)jsZi44X?oF2W2nksA>;
zi8@X)mhw_2qxkPVds8DBdyC@%PK8P0b4oAR>s%uON$Z!x=HLpe8u%yBa#Pm0890%i
z>w<AkT*W-=OiT^+s0x9ey<1H&Sy2Lv<$f3N%Yri6b;<7=0zCbt=Q%YZ-d%`9R<yT1
zJtFj?Bu1<hWrU8lkA4F<<N#9iS3rgsJ&k)wHK#S2$3SP=7ufd=n4aVzqAr=DA4I3Z
z+UIr446Y5y#}WTEh7fg9w&ceOBw4DP9@seV_z@kz$l)u~7onm67son+u^9;{cyW4?
z1nha~i0`Qi$b%AzqpSWfS1x!Fum3lKyTpq?bj4@VRwtRW2L2g{n54h^WG;I+9v5m^
zH={tI#Czg_r+NQ%Y!9G#39P~UR4EpY?n<lJl|V4}?^Z`zdAy14o*ClHWxc)Rs<VEV
zv0P272j=|^XI`e%1e<Uq#I{MP(z!lL9VNGPI}Vgq4@S7=+03jLyZG;`{`?A$N$e@x
zY^Hj!=i-TPcO@g(!mx?1=IYP!Eu1!+FtI<IOv+5I>oyGfWpKDW#sg>jy6`&iC3{5L
zd<=y7hYQgIPI%`o3SJ~3C!i2fi!3fKnCru{MD-?ofP=0!1voi2J!gtq9hrj*BexyJ
zQ2Z|kV-dy0<??Hnetdg=vNN9l)=uql_Z5TT`NW?lyg6$V_(kW>lUkyE&{LN;D4po%
zpHnFGBk^(H8Z&rYG#hnf?^G1xGD8qGyh~yK(!MH7HKU_r9p@uM7y*F$rxz^cl(Bt0
z&FfDAUN+9bZ~Y1YPPZ3h`2VH=x7e_TUN)EqPh7QH0f~f{SaeS+thRr|v`JLMz*q9-
z@RhR-#+R^d#(9P-#(@$!Fv(ngLfoj0L^3^J0mmT{mW}u_BBAv=<k+pWwdkXXDy$8g
zdg;(EVsgzl`7gw9==`OmWLT(6`b5@Gxzb^N6`b;Jx986YwDF?hw9huN;m6pN*z>0u
zpyL7Eje6r;ni!&z3-e<Pc8LPm%IfWeW$S#gSXPUTSHJO*B5LefV3}eosK?U_a=_Vm
zAGb-dof9DhRKk8s)Souzwvq^F=U*;oKgR_t<zr$spJy`Fj~`WNf;rC5m0(YEy#>n*
zEv~$JO3w)lQIc+fl-|WdChMzGE>=)CzKeIX(V^NK9JgIBpu19tKXP8`AAX<S@2{1L
zuR|L4#h$Yv0ZGMOw&oYSx2(?;wt&_RHfxo<UkIf1FO~ongLLanGjrB@CapmA68!`z
zPj~HF-x?n7Ct8XKxXq7$U8hm%-g_h&*xm=m%JhT?8ognO?SX6fCnm73og15fEV9~k
zG8$4N0qa%Dm=Jti=x*qn_t1P?TfQ5yR$2LlGZFf%unr?*^VpLa-mU5+ra?vMA|LPm
zx_R+jtJ>%FLFmiBb`8q>uBtyA(VUFU5gsOq5*c_v{}uO%7M**DsvBetRy_e~<2-Ns
zXA>sxd<L%*nf)g}f+lOT>lLy`jig_gikjbMa0j=*W0`5fu(tD~Cx58LSEhsu4k-LO
zy>75g>?5GF#{~H_4Oto*=*e0Po^E)URiw**E{kH_S0_pO+Iprua?m~Qc3bINGMf)3
zg~xDP&rcF+`Qg>5ZH8&xF4F@K1yF8365RkoTpIX_pl2<`p3LiPcxOF49Q*&aK?qLY
zg)~kc`som(!)1z53o5KW)#^}~maT031!mX%lwW!QgrYgsW+&f{aPt-4Rr5o%+Q{oN
zSIqm8x0<e<dUU*L%+#o^qIr|<M_r6m#MAR>=bu?51`>es0u4{cXL1ls-WBdBM)KjA
zqS`m{m(SI^)v&ngsM{CZ3Z$agSgOmGN)Po~nbRp>N4Jie>PrzBO)B_i$f9dGEXA>x
zaK7M&07!RxMg*|RChmWEr6%79_xXtJHG>?bPG)v7w`p=*cNhf%t3VF?K^0eQkI|H7
zEY8kyXTd?zH#g^3WD*zP{OHp&Q@x`*y@VGY_y+`0(*BCMR(l3WRQe}U6T0r(ltO>k
zaA2xz=6it@@_i%nV`iwKf4#@U$S43Ai3_?5zxZmGxnlH;mO9)b(Qdy#UTmZdF8SKH
zZlS+T7Db#{YV~SF@C(|$(N%1Sa<rQ0PEU|x;e0ZP&>UlM-Zi2l$>ljwbf7_am>SLW
z4>|Z|sn7m6JDfgNTNXanr;o9D%A+Mxi&CU28XAdixo_6@c{^k4TKs_u7XrTIg9AR)
zP)f((p9y=wtu*tN_gfszWH>ZYRdfp%@PA&?gx8I6fZyAVNxi&f{oCpF<15~#BL;4J
z>xS%Sd_>>Y#GBibC53_=x4Jn+CR5eR!7pxg>?Q+Lx>_|~FJK3@ac1r@LNu|0O;6Bt
zAqbi%*28Q${Ua<dCkKV*a=(1M;QQn}MXBrQKHz`lgJ5*HP>{e51IP#f4CUrk(%jmy
znsY8HOu@~b&9p(MJQv3{t5n=ZGF%DfvPW{r%&Q3B`;t%fVsZXASZii0*T6^pkjeu~
z|F{X3+LOSOWk=REnj;K!{g$&24Rw}GKyPLi-a5bA4;3;HTwl2AE?mG0{zB{QScFrg
zC$Gc8z!&*QIN{9Y=vsp)e=ih-{NVBasmni=iBOr*jT}Z-5_NhUqHq|oqOIMfiP#gk
zD|?04bd~XH_)MZ1<5`_i<Q(W!^2U$e1;C2S^K4kM;qGFV!*(GNv}eG;!apJcr_B9+
z6k6Kfx88aA*r>FTsD&%hFuH2Iz|VYoR2F{SPgihk4CeVW672M=RjbzMpVQL=x)u&o
z!}lz2wyoFhTS5$DbiQ)Kc3!&RhxZHs;xB(1rn6+B!iR93cN)6LsuvTWT`*kNADt2#
zO8|^Hu?3>8AKYDv(R$M6axX3Xi98^AYGQ%JAKx_V!6)#wTl`?j;dxZg(R!ffcVh^S
zr|jO1`>NnA_@e7rrl&yI^((8FdxCrHxnll!fo@}bM-aZl-V`esM<Su1e4HvVkd0&W
z2fg`3-2o6%UR})EPL~;@0__FhwzE3#sq4C}<7(^J&YcN)-|~O~brBQvXK8mgC~i9B
zkNoaL;i?ZJY0kZF{`d(H)0P4OC!mRZ0qP2%j2F0^a{#>QjcBKkY5Xk39>88z2ZkT@
zE{7W+`_|OggNNE-S==b$>RR)r>&#pK1jbfGq=oUEpN^j1Z9PiOQCE-t;K$=uyM!_B
zsFX@Ac#maP28FX>Mn+Z2GGvYW*2XQb6XsP@gs5ap5KM;s%)1<Y-6gk^aqT&`3HUvL
z%CWk)xiBThiXMl?&yBb@cb2HG@GpIM3?6?GS9Gx7Uf&dRA#0ZgJ{pAc{HIsf%^Hn|
zEhLyndE*wZC@6SckCqsPg@LdH0LsSqw}02xp>=@lz|%%^SxzwEF{s=1{Q=Gm-$Fvx
zf#`4ph{wVBKI^(N!O+E`lBiPCO9q;xeU1WrXL`Z?+iZKxR#{ukR)sL>)Rnj#>RMUb
zdRYwr@QUMgpaM#TbH;roMMd=nS5^R{nE+K3P$zFztx|D_d_#DQL0Byd?U0maH~SlX
zR=u6}h56jhn1TNF)G?8`oe9ut+U3ABNd>rI>$@}D>OJxFAPmL#T_N+Cd$KK-TpQ)D
z+di7dnJG?PdRkrySV|<#0{65sDR)5Rc-K3PE(xZjqxr%f+>+)m|2)MJT9fgbs^uqI
zjQL-)Ze-2>^8IrGh>d3`#~Y#AEHq17cYZOc!t2HvgH+&Mz4QLppdhg{1z>cMQ&8}l
zi77OW-E`v0==M=&?F)};Hd#axh)MP<J^e|Dgczu=g>R4GfEJ^>`yJqA!)^h2wq{`!
z-%>(ya;R#4;XM(ry^+%*&|t6u*OJ!0z967ePgSOafLbt7$Sc03MFgle;#drPr7kF0
zSbpHrtNw_K`=*#%un>5@J85WY+Ld72B=zxQIxXg3!S*%6hKJz(e<}a2dtw>`So!rT
z^`$DqUr5~c2r_$NIoM7YY*^P$fj9AW<?apxyz`_<ORpo1(qs+BE_-VaG3lj0VuDlz
zRL~IxqhD;uqMSdBeRKKA87fWojf}TENYQkHsr64B47<z*sM%Je`8KIzq4wxs?4-^Q
zH!!u^Q-rL;|6%67Lu*+^#&AOm^8-2FdyI@`2U{YFZ(dU3;_^(*)!M59embH)FlDKZ
zWG*QyO9$|wCvXO-E>zOg>^+0FM}e-Mf{%|VA)#}3YpdJgk6<9Mj|b9z2A#S{0Jt^*
zYIzo}1%2rcV0j}ie5WkPB`n;$GuhVO{w*}Lb96KU{RP*L$VhQJyOIw#Y+l!Av)&uv
zLhYEs{Tl-bk763IG-4sU3x@#r1=5!MM;7U$J}XRC!VY@C0R0Ub&;sNH8r;!vj(kxU
zd4!S^D)#0>*>^G8n*@$ZqsttX=rY66y|mM^QJ+rvHWEuK|IlU%*;a>TsC&Y1zud?*
zaxDU&fO8C)52`(Nal(N39en0@fiX0RJ2T}M1h1U)vJoW|FLfcItCW(Tw|bjk6|YJB
zl&KYRy4WWn`V59v#-)@$@?l(L1Tm)GW6Qj(M>!n&{DN@&F)V`EEJ)idURQc(VQB5B
zSHv};0OQut;lk14;ia1g967Ep_VY*iKrE8T7(VKJVLOLC*EmfPCeQ{zxW6FL*&-iJ
zm~IXx!3bPT8CE}aDRyvjS_ke1>p>86)6!w{!v^>Rr-k0mNTC)tOqJF2t0xVdSd>Kn
ziZmbb;B89A+n)$@aF+cKQ9v2(OCe^7u6W_G{^m!Iz6X;mKRvt)KX0qN!I7LL!5(bk
ze&&+x3lD+~WU>mM!?x5UxYtpwV+1_Pp8?QQycsag`x*>Anq^~|;awoF0s~#Gt8~HO
z<LD=iLK6UQg7-9qZ!IP<_=>pxmy#jwj8N)kaD2KM{NJmbWJT-c#ZfeV8<=M4UDcHT
zfl15B>pS|GBHnMahP`wS+Xs#ZqyG{C@c|wlo@@eFNSat+8SrZZkp8flMGD2L`>#D9
zx(Dp{g5zDty@ZpANv`-9jGSRufNz=REE*99(q}LUAZTQs?BW4&5$rVJVOG3=$j{FY
z1Du7<X992hk&YglPf<AYVTz}VU!#E_QCNGp*$=Qz98h53Oha=k<3PfL)$x6k5hcwg
zFoh<B0^96fhORl&c3Z_VMDjiBy3rcJ1^PYF(RATdp~=;M9|r(G+SnT>8pi@^{)ur>
zl(VL>Cn3@U@smx|%)I4;D(=lJXkhXNuzN|=or7kogfdVn0D2$*c(Na^|F=_gC|hLc
zOr4pi|KxZ}KSwEgj=TEN7LqJ?dDHv#y_}rMNT$?d$Odwt5ipd8cCO2Tk4|-AL>#SG
z(S#WXO;j&YbCG<i^QpK+U3XtZXjZ+)KF*nL%4<e$?#V4Lx5LIuv&AOC6zHCpm6a91
zJ%o*zmqGm4ONBW(<eZ#QGBHe_K+@nZm3oZvW}cdic_{uIB!}0sSB&rTQ1gTO){mS8
zl%-v2z$|eWvN{;<Rny!$cryCrjrYwODrXr>gm1BdqO-55<sx4~%A)u8(^)xDc6!%p
zbe^pb`{t4O!RKk9;2Jn&%rDOkDHo(zOnFUGTq|Xl{_b*ruh-u{hBYGL_$XVRqMF0)
zE#1FN^EB-n;fNF+dlfD4#Tczxdmc%IE>`Er)&Z5AZS3y$g;UCa6MmP#>%yR2^8={l
z9!q9VPfweme4vwqDY0Mg)uNKl-`?%Ty$hEsLbNSLbhiYyM@XL&X7?+@&ZlF3{}x~H
zy3_&cI*n=@67WdJaSF{VHJ_v4$m)7wT9&{RtL7V@?~H#52>57eDRa6pfQ*9jiUy+D
zigwoXI>&X`=r;f=t{mk&O?RAs;Yt4XViXvl`BZbN$4S91xU9UaOuvYVBvOw-F76O@
z#;u4dGk0NDSKFnd)bLN#tkj$7Jd0Ko(IDGkiyg*ZV+*})Na{`3^zG4Ptj8~iiOz<X
zj;PqS{4VUO^=Vt5SS@s15Y7g=`Kh!0aNVv@6i~gAWwkNFCq%zSzuYHtB!z7T6lxJj
zTN9cOtp5XAT#|}jTALkgw=U-s5=+Z)=U|4Y_H9aLDZ+(a>>BniZw~-TAkdgRHXT4l
z)&X*52ym+bp|~MS2gD!PsS5Lmp01%ZKMurhpmwydhRT7D3k;ugffp&J&V})xWI-6u
z;|}~_B2E-cOw9g)fuB8@4|8jqn=k75FDnp`kAM^j8%UTyCj+XK!g6ajH(u~84S~lL
zNTYQxfj$kq2O%qSof#u->;udB3becDW-kRi5H=D6#|gp2Fdfyv1H&^4Ekdr<B_<1!
z>ZQ7B!%tf>d=I)BaJJCj!M}eCsTJtQS>}}J0`-x|ED4qZfZQ3?<D-RDxSLSaTo56a
z72<p%`fO+|TBnRgn`tpw3g4NB+v^M=FQQt1;^^%5-%5%wmxoq7sz`v;Y>3wogCw8i
z+gx61Ut-V}R*ZdYw(3oT{}^1~Kv?%Fjr?IjdAejg_i=;-0bV#*AD(r<vxvWSpmm`v
zw>Lj^E+6TY)jE_fo+z~0Z45A3kS@0xw_+XTsAO{)opVv<!WTDsHa$7*4H3Wo=4ArY
zv3ciye8O@n_{VLqAE}>x@znE0m}b^W6f%UkG`(2$A$4++zq2ba^b-qb67&^gpK!jo
zxKOF07nFOj#=myQfocr41-udOWMvD~WHB+<ZLe={Z|M%n_#DmL_9`kWq%Swu)>we3
z7)Bz5NM&<X)rm>XzNEVN`mUe<ut90_iaQ;F_9&E0IGKds8KyS^RODP^J9(H+>m5%Z
z)m`9<`gJ#H+K9?PG8h_pk`@y<a(EG8!;)J#GO?r|IudhP$z$xilPInpgP`GVd>hE!
zN?{$jIOY}ASA<<fl7FI(Jll)MSl`iCm)Um_wx}7nI_466G=A;oi-MiUJE-R$w%@1p
zVj*+3#ydymgC<MAFgkORuiD4uJNA-OHfK6?ZefAQZ0@GFM=em1>T8-uTd<c!k~jhA
z%h5#!#C<ite^<Vs;Tmhy+r9o>F1y$O#~%`Knc|LJ*&8qH%6K}VWm{@B%@!UJ(N}EG
z9Vh#ml9JW!#0Vw;lT6ou4D4HzKgsAK)i(3$O`g1NZf@(B`1tq^fJq5{K2W4nPq+8t
z*)!M{U@!U&YWnV5)7&ilD5`}49m&SlaK<_#&PPK-;8BO^-k6+|L&N8k=en-#^(PC^
zX<>jhgkO6gW(;#@bX;6dfkt^OyQy3fS=Wo70|_G}Ja(S}n--2mzXgWNW)29$K|!Eg
ztGF=E{JBu88Wcy4TG@{-XdHEIeQuF>&*>q59i8dGL<|dpMU;c{z*qsEWW=~E=@cD#
z53|*r{^0Z5zU&E-C|twxaJD>wiS^7glvM^i?Y>WgS!x8Pte750pUAm6Q4Mp`<OJ>K
z;Tj0<F2rp-+yuVye|*OuC9CU>L04&nH?{SDTmlp|G8aw%>I|=Dg*`{SL4D}aGye<L
zq+p2Vz)6f=>LUAt{`*_=0H{1%UJ^YcBF41xCXyhef7av2q|06>8iV_R`E@`5hwg~l
zW0#T@qnmdDjwoNE+ckN(-4q7>?oyJ=!t(viVX^ZAAdVHbT>`Zu3t*Ww9CyBW{5Int
z8nfWcS1qJV`C0D@9e)HF{n1`{zRoHCIXeSGFets%y_kT?(X{0=tkq&uW3MwNDjtJA
zumc?LtRJ5CL^w1}@ltIPv|VVhFg?&+UOCy6o(ENKwG$t>8Sx03Zibt&xs30=84_6x
z78fEg0whn_`C1IEU-o9(f4a;g_#0!NFC49k9!j(eg){NXy)hZ@qHWM_=N-RMm;O+_
zWUtEzM{jTU>9wQLYrhn-cAW>WSpG%#uW$Z@{2KjXqd52PAStXXch{1;`C4)3o$jWf
zT*n7TLYztK$EU}%fv?XJ*G)^iuAqrf{xBsu*&*X*-DOj8alsJBo75#C!KraT2Hmn+
zPC`Y-E5E;i_Bd_W?C9u7AN0@+8f9NVO_`WPFJA>z6a7j2F*P+jz}u+xJ@@<<iHz3q
z@hH&cm<%TN00%chHlug`cYr@JU1^E$GVg?+oSgihd!NZ^3cc8_BcJpBOb4I^5VC%J
zIyg88%;&(=_;;CcACLXoJAY^<IcpOf#|9Du4tU7q3=C+%tO~f7g$bM>fwp`78+CHY
zGR6Gu6nN}W8@;K@$HfXN?3MF_2Y;TkvE?PR{@c#cb3rlR7VRU0NAcfewNd9#k{|}I
zZs;{%AYv4S_DZ~&XGYCe)j&&1P>OQd)h40sp;yE+E3aNDhi1!Gd1_R>hzt9o9P&)k
z%7Yba6ws#K?Uenh@q7~_eVsYGMs(Z~i{1oa3*yUD{d8K!4p4-%@g}`;vG8+1<xU{&
zvSN$)`9t)egW>xr%BK^@)(j`qhM8*-9Brb1k>yjP<@$cdgv)}-eeZa`#=YR;@82iy
zWvwb#Z4c+6pa-FAB!m4YFPMAiz_WP_sKwq@YP(aV6kyYeOG-|<r8IfnMC*OiBdOFJ
z0C^8^3#^h>HSL*zwpKQYFDh3lTh;Ah+~4%W&jHj1(AhICI@{UhOg2$cQgRJSfY&Pr
zF5=t!Q}4B<>RH-e+8ob^=9uNJW!XQ72&~ZBbbV@cz7mYK=bVE1Sgg|bRwIN;T>m=M
zw=gx81=k~N)zQgoncAw~^Y?*m)x<teM<prZ=e3Vp2SrvfvHI`ntQwx0qR=0%eat??
zYAVKk%6DHYFF}jJ#8)xM*LP}7-V-Q)+C|+QULuI#^<lmF{gXDouQk=e^M|$Yl}YPR
z`7Z}1qcVqM=ENDhAGg81<-Y`HuZPY<7tWG6o+Q=d{`u$O{~P^X_Bfm6ZKG`Ms`JFb
zbo;g7&DZ81qME9BMWJ$69v-aZ4fKZB=adaz3Q9mHu41LTbbfRM>QP*njA3;lvFbWi
z^sKC`^;sXE_cv1C9^C}xN19xlAn#kjEIlR~)YSPk1R>4!n_xwrc$p8*u+7ZcT}_k;
zS2x*zM(QN(LOyX`I9KdH>#z0tWJgCPs>3L6FV6+Df9+Uwa^UP$XLqlk&NdkT<S>~5
z#`oGe_v9e8NoT6^3V$jA-sQS=!74TeDW(t4$Sli`iAfqY@MDG|OK6YZBE+U|h%mQ|
zsPtsnjwv!8iD-(WaUf4AT<y0!Gvo5_>$_fb42!R4pQACPQG@#xzxJ%Q4YO<WxCaOe
zO}E6>hK=y0do_j{a>Y$I=`%J%H5eHSIsGH(PAcv!^=9OgFyG*fk_Pp=+MM=oEZ8Y@
z8&ba3-*C5R{PXSRuq9=J1_gK=kuWhx6mRrj7AbKt9{~zS4Dg_VmF|Owpa-HZJ;Bv#
z2%2sF)YIF$0t{|CthiQJR@z3gWURf>9z%CF+@g)?U2}7r6&t<(@Hgwow}baHiY;cU
z-&0ajve&eqr93dkgYXsm)x+Yxahx(QP;p5o&ce-yn~UrtbYf?@EtOxipC>f<m2^+N
z24mt^YfW@?r8zL{W+qHk9ClL|j}n`ipCe34!0GUCSQ>q;p8UjQ36I&eq;e#5xfmx3
zHMm3wy-CU^WrVyGYcKMRPJo)OKw9pacDK=V#>1-O%G<2U-(MzJD*P4K8siChej@56
zZvsfR$&)-D@Q)+5+$9hBNV*i#Hv>0?sGL`q%idl-iQFv6YwD!!II>3RXf`P^nZVEN
z($)WsdMQ1N0PgxR(8F`-8Ce4d;PqMHl%D5wD@o?1+$89I>zc6fkt^X5oX}?h2f(uu
zSq~wCQ^m!>X<{x6yLz<r2Mg<nsJ4ol$jajovucK~x9*VvPw+{*72D}(bJOc>5z&yY
ztxt=L(#;e5alj4bnl08~Sp&u<i3>%A)0*_QeFj^ggg!H>#fA1<7^5ZJsC(F9G^z2}
zYFB9;oXi~R9VY2BT|nHWi_Yx)MUC&x4c+&8m_eeNYFtlZ43ea%m91fIAW#`PAFP9L
z-@ec^f5W;c>#aw7aN!u(v20gwyS~Civq`XlVjI{NDH4o;BP{n2S?-Z{xUD*9WwI|L
z6SZPPPvU8ZBWs~_m+n;PcocTQQjc0QReE!}I8}S4mOrTSnAJs*hHs}DT7d19x+7m!
z$^x*46_b+-wNItf<k*|33<6X;2zF_Ey9Y3)tmFGYLjauc3_yj$6+#b8Dz8o0Vy{e2
zJftJ;Ss9meQ%^LqILnk;+27b@E3VzSk9e}g<&oGNZnYf6UD7Nz5F+-xb&?-JYj8o3
z+Zbi5U<4s1t8=;CbRj5dkTjaA{V-<NoA|TdsKNSMVr>$_+pu_iIcSTmp}x#655=_>
z2koS2mQ~Ztz{bptY~i!b8`bmT$}GLb>(H|r)S&_N1%>aqm5Q*dx69MyvJyD-$$#Hz
zY!x9$s@}B@lRYSyAT^(E8h*$>I5=`fG8?ywuMb~q&)%&{dE6`DZUXdApA?suvp=%g
zQ!6BavB5?kb2B7fJQ)-Am^%Tq<>()c2L9T*5#Kmp%G+D$bn}lXNXU=+e3MV0YSw<>
zx^>~bPFF;74xrwde$aszn8ujl|DdVf7O+KTK};(2i@A}EDE&;ui|SXyAL|f%w4+1y
zD@qb-X0?L{JI%rlg_O`j0zr>-MFMXM-8ciw0?0T@!;i3u2lFsdB1`qt<!z7a(=qLm
zJFnqC(gwFps&?g)`<wQhlIRUHgnP=rZTMtcDN#*0XVoP<Q39c<+~-buJId9f592nM
z{x2D68WOvT<nvN=ENO<z*XPNiY<1C8zc-u@4{L1eS#g!qULzXgRsQ!tM2BGDbw5ME
zYc{4@_t8ZbWc+vjPi%n}<?oUoW^c0|0FEgnf=kRXz;ozYSV-=Ppr&UJ0o&>;TOf3$
zGwoVoJ<dpIVVLbJWev-7Hh-ZRDB0*L>_tWb`xn8L7tbW=OHxeQR^G%LlO@MAMLVx6
zti8cD`k@%freq%v`eME;TrMaN-*C|lEu>Lm`zQZjngaO;hi4V^Oi^v|9I=iBO--LC
zdVFs1WGP)mCW%(@P1YT(25hnj_=HmBcvY}d{bs#}@uTd`Iu4^dO41r-xaqC=I;)?!
zpvF+eP9bp6U3gW(N2L8f(%v$x%IygoCq$$TLXa@%RJsur5s*eYM5J3f9}ut*P*Umc
z?oI^-5$W#k?)uNx^Ls!1zq}vba~-dvp3UCRT6@;4nYr(|XUS0^jhB)Z!epyG71GV3
zZ%_10a;8UW`IY495G~TbimY%{RtDTZm#~2N-!G8taDgQFa49P<lM*EshSKOc-rLpk
zsN@B51E5ni)myDBv#YmJXUywkIbJAfC?gxDYq=J6dIYxfkaF;|r*mJqt9?!~sWQ9L
z$wilSHJ#S$o6FQbIo!o0oMSBV7Fi9`&5PJY6kagXj2oQXfAf^aKdH*DrmJ!>H)<1S
zZ$5&rmd+Fr{>4!hJ>DyK*viA`H^)h3MQ~?HZ@)Y7VY`1K2CaoiD)B?>O(oZ5==mY*
zOTJRJc&Y4Z#po7)A7|A2VfRNa)Nj;Aoyz4-DC47Ue8;wXGDIU^C91<r@P5?((crtn
zDTJSb1b^3?qvKlH;$60fL+aEr|9K7zZgp?6c+gEp)q&w~fNB;okbJy4O}=$>1VKtU
zwV>WIX1A*VnpVTsb+XXH_Wt9?V#dd2X3QYXp1vh&H%RySqK`KK!v5P9W*3tZ_Qd(#
zv2<koeA(zF%+ANz{j5d>k)^+8G;?)fj&ZKPFm8QxG)tE_J6GktuaHH>xA(Bp{mpbu
zr$v6&=snDb0=vt%rg*6)PtGpja@yUC`ehPBxv@4BJR#0^#nG6f8p8v<%~Q8*VdMjL
zglNa0h;8)I?oDIISXzcz&eZA`+?qlo!-t)deF_gA?<G3~5ju#<AzPmdUlj28*#h;u
zeBkq|6;{`p+M(~$6q=ro30+JFTQO{7ds*#O)kH=npA-I+?D$iuW|LB-92oPAw^HkH
zz`@CfW3sWcXsfCkNBdAj|L*k}+=3gU&n%T@&J(=~_pfegn)I2uom=0%o;gxW3VPXF
z`!x^V&1Hz=RjZklXd*);=8^z*WLA$44pvnAJv24Vc-Eo`Z8$*cJej6};qJ{4U>@w(
zo$E}0nVNdBwY9}6`dFETtzga8DiHt3d^sZDL{qIS>`J@L+#t7MQ(CNuLY5Ngoox%(
z8qZuu%CUgJrYiL3oRN2z2nLp!=M4Pgx^pVAJ2TxrZ3PS>A8XTz-6YwxpMA;fuzl6t
zdW5ZYJdhE3J(bIECfRs$X7PoNFrJ~#4;5aE1I-^&HQI;wc%s4{6oxcce(gZ3ZFR+R
zSsX;Ieo<!D$@d`SM2{HlvCkC)9!u9p+X`{}=MoaT>gr`z7Iu$tLo|FZx63CHJDVr<
zFIy@-6>wNWh*8vCNbSn2v;-bPu>7d#C2#ntt+~HOlHt8p1I?=sk{;(2h<g|CGL!7d
z^|2Xgo;BCM38G3(jgoy~<+XiJttdZc(4utF_Ni+1+~2?eKBSFDaiRpPMz=%Dx<=yp
z^Y~u8Qjo^E=djenyv`NX?tU=&7B(N`wUBH<IpxB#vMr?^qi-L{ea5^VCyD1+eR9D5
z?$Bt$8BQk4?MTs+xxBIaoHxY%0Dm(o@Kdo^Y<024jvpO=nN3qx3<uRk$CT~-HGUVK
z_K!c!z16f*cG1aszu1OleA~d^$vN(LR`hin`|A5)`!Dbx7}V0Npz@(8NN7yl^8!M&
zBAqsQWp}M+`1**uNLYZ3s;Hw8-5cqo=JpoMuvF8tJeLDEmK(h8Y~_b8Du<djo3;$|
z4^q2~2OB<Yv$?J4cw~tNnyAx<;3Rdvi>aYv&T_4W>P|yE$9mVzb22~OJjoozBt<Qk
zus7wYy?bZpOYhBrBWuFl|9iRN_isdQY#F!n7f%$%Iur|`K1~Q&kLvy-Jbe6@P^o@4
z#(*_8msVYGp4K4OZE+@Z3Gd79+!jV&O!HdgQBpR$Qeez%+2FXHm9_SFp_N|ZIH^0^
zPboIF0t*&_h0kTnU-p#zeNB=BmtdY`%tfc^eWL!R0_~JPCX0FaTsE|1h}Z9jmA#Ae
zONT!1Pk4{rf{EzPT#|6}m6hEE!+v#Ve6tj6OkJQ^*gvq`@vy((NZ9|o!~8tIw6(cY
zyS%q;>psl!NowIpCeYOBuDK|3S_27S!iGKi40));0_(FqO%sy7+}8dFZJ&6hz4uLW
z3-8B;wQLeuvedOjc{P~SBPQH0tgSh)#%@G6m@uw4vA~u!#7ic1KDTDb^P9^q-9HBr
z>{%?`994gxrjl)^(wLZ-oj|h0ujahH*$>|mqN@mO@3qpUfM`CbN0NHMNTRZ#*m@Xb
zQ6q4$>eyFT`IJeatgnH}^UpA*=F<J7*AMP|n`_#8x^shbpuh&9vW*J5>)EMXj_u6n
za1@Z~u)jATFktY6>5cN?<rk9{b6<p?a5Ri5`XeU8ku8z!D`zk6)U}^c7yZQd4!qkc
z&$M$Wz}TVajV?FWRQgn!SNBC*mDp9d2lhSqN!0XMa9{KGnA-65ZIZ0~@NF5=6usuq
zO-DEO$%^+^?wJY&!ULJQ-OE|_fk+&uBMnkUEh6k!`A+unI&r_B5iW8k-zd02#4>MX
z>fMiTQL#1P_uqXvC73{dFEc~=@>FG~1#Y$4^#C5e(!5nqm`pOZqEq-@Z|p#o>8c!#
zZzo&WYL<X`{M)Ln-7>Uya<r@JoN=RKQ7vjUzWyBnNN-tn+q*T&b5joY@C%&@yK&#m
z9}kjUq9WH=-WVr4{Wa&3RWC;%U*_|<Hte{da)8ONPM7+wDB76Z1Y~0F_e3LGRBPWQ
zNp_d@+U#z}Z~w7^APFN&d9qe@m*|xSQ5Jh*Y4>H@b+%IZ!dTZc%Xp=U(<2Q0ee@VC
ziDg4F=(&CL&F3^T#+FL3kl{v>nnLkx@K$+A*#vX^>eo*S_rY*3*qiH}pF;cxManzp
zatFy7%IRhOuW8Jeh`$kJ4sz?GG`gX)GGoiU;WZYjM>}Giz6fOJ+ZzT)(++lX)<YX7
zk;;1h^VaX9bz62`We=an^TFl98dr&IywS>E1cl>|&ALh2YiFx|N7x6<<bm@EokRDQ
z0&sW;1IArdnAQZ-^{w7<+Le`lFV$S0h*4OWmg+miR~6p=%(=Z!ad%;`Q(FAQf4lYL
z^wHfflkNPB_aJe*K;2)zve|cKb&)f%s_Wr?%%c=%JsKg7rX`hPIYRs|j*=xv6Qwp~
zg~Jjf-C5M;C+qZaTcr<^h%wrA!D%a%QuX|Q!%Hr8vT{dZSj*<_M1GFtg6r#>1^4Fm
z@;IYko$;{vzU!~X66J;+!pkd9w<fzObqk6bwyH@77mGgz#wHUpi-{49r}pIJ8$bTK
zSiw!Ma(6GLCBIu0x%W}g8kv~yuT`x5I)?s0XE`=mm4M^I!}vSKXZ+va?;bDLT4M?#
z`PrAu?l#G!wYoL-W!7Ih5A>UML3<f1S7Y>{FG)Af%Z-I&ARNc;T0(Ikjk86E?;s8u
ze6Ft@UKz#P@smFaud(g@W^OxAs*I9l3F%zol7CR^Eqd*3_hesw<toDP@5(5@HF6nC
z_m;0opxntTua5`BDJJ}lWoPJ=JsT=QXr;);wYeJ>q}KgHu6Au9NujsbjiojkrJgXR
zd%i8l_~P>5YenX3Z~0NF1xpcg;pWa~$fp>0Dbs*^e_jt}22wZ|QlV*$Yw6U+%PZd_
zm$lH#S3NX(f@i2-h<s&$)bdA{T2Tl&uik4WNqxOiPN9mLYm2sI{3V+TNwd1wjIGtq
z_BOesvx3JtCo-n<`Kpc*5lx(5QYuDM;a{$AHhTCz+D+f(XAX$8T^E&GezzwrzrqmI
zAa5#NP|fW3SUniOgZgdp)d$u=F9&|Pv5G1Fmp~WwtRSu%xVOFqOnvJx<Pu!p&O?X>
zSsv&lWU*AV6qVcT;OA61<!gBM``qdXYDn{tQSCUw=x(6=;`&qrjjfSN(<BUIJmo5X
zG@g_Mn#l^!$FJLr`bn<66zAVKutN!CTzhppo2XvL{CxW4jtD!}_;2*^mcp(>9~{t?
zzt6|_g(Mp}soG#Y`GRZSH9@bou!0f0CvCg+;@(P|cXZ*DwVwQ!PISV#`l!w5lS8yy
zcg#3G@?L~@5w2aLQx!hGGEYRySQMgVa;pndQ_!o1f<1!SfBE~I4>{#%OojR09eZk0
z#rf%lk!a_Iti_X)BR)#aTFfyq%s6_c7Skk5HK88}Z~Y<$e<+^oHM4+7^p1eJJ_boC
z^Z!*C=u}CZ&ksA0p_iSUHMNl%V!U0I4sg;ny5n0Q4SH<IYv($Tld{m$-BhXWQ8VF8
zo#qr<K8ah6#uu>ma6aKmw&x@sM63<YxtD&EJflb}C`;PjxU_b(-eeMEU((D&j>txj
zQsK!_T*^D_I@R3W_4UP&TNmz3#v1Fci#y<=Q6TlN8dP31s6Gtt%AL$(yV1j(a{1(b
zeVpYZR~N6C58uWOC^!xU!Sy|#;ya(2dGl~Z^8}A9Fxp$MizR4#_vTdW*^V14F1A{W
zD^1JqhFDY`J9Zj7|J0|1cI{mQQZ7=|#h+6Y_({m^?nH*rhQ+i&rvPf!F?hR3u`Jwb
zNdBxhrm|tVt<qcmOg}l^@CHg}VVPfG#>$KiZR7p1a+K`QX}f-`jP_*Y%EQJ|gO_xA
z81-2JL~yfQ>)}Y&y@$s>`=aN@%^w<zE*1f$YOzVJ67|h0XB4LMJIH*1?%v%Js;4-<
z%7-=y*9&*0-QJUGUb1^;l%m_Nh*?WWoI)JW!TfzpaoA*DfhFw2r*vvC_nTsO-X>$>
zyiCz<R5z>d?Z7<$ZD~T@b<yHae_gNa>h<$ZWq?;g>~DIH_-hc)lTOUcevn)yEne(I
z52(?$_&dfB3bS94YI9iGvVx+VM&3Xj>2Fp=XyU4Rb5_LniLmSq<O+ACrKQnqfZM_-
zcR0QHqG~=p|9xB%MTb<5lBQ>m!>tOIysGHe=2`V_BxE#Bf<!@#2;0Ema6(C$N3N!E
zOyNP~f9bY*eWp_kP~l2jA>n)^!tct@a-)J5DgThj`X(vOf43ml&M4$-=(Mqv$md6n
z()8&n&IfLjG!*ZU{(9G4F^8S%ElToi`E1rV%Gbmsk=PIdT_-~iml)w0hyQ%o<-YvP
z`sKAXt1Myo`g(kPiU?z>u6@stP<W}r?o!tC^#PhNMzv+==x$nZzVB$<dGnbeqV=QW
zN6xE{{y{=D4EU3Cl5bwDs~>0U8e8+NP_EpgCG#0e-jONo;n|{eDYZ(lh-OkE#VS&%
zMdy^Z-E;gdXKWoc^Iy<Cb<xmp`Y~9J?1PlVrX2m}cbjT0mS3#4Vubcx;7hbSo;r9w
z;BOIA7kMJ%BHb~7jWIEJ6PaE1M!V!dcdkSjHJ>x(qt9V`cK<<+Z7&7S>apMbf-8h-
zI^4O}xbkHdJ%$tBb4+s}0Vn!~b#pa?&$;;y5C0{rw;tKuM0xLWIls{{nZW~I2__89
zsVD5wtt~|XL}&Wmyltzlw(hIC;d0KBUrXGw$lS*W#t3{Qhmd+kvDxL%ym!U$MFYIF
zs(;9I%(Y?9zj|?craAp~j5@=@-L7G4-b9`>nx(PK4vp>g2`m1K8|ahb<z-v1r(Iej
z=$;xKuw!}s`TUvpm)w7kWA(J~2Ym0^X@W!n^<vp<WCp>{9hg6wYSf`gJU*DfNAroT
z$<jSFjPgMsBRH^aHvaO<$XW2xT6T)_kR~>V(`%lLFQl;sy&vUmi8kotwG>LK>}$4-
z99gyH=QdRaJ`pO`?#ghEqxt_vzk5(|FWziu{z-}=*T&Z>b*^Q+*%GcgmW8E5jm=gS
ze&r|koiE2+9PPR3f|`K}fvI1Dy8P7dYfwFSJckT&<Fd~tI=recNx}B65&Ro9sA+!2
zv5g5PuZ9=LfON%2_ucjMRc`4kHfkd5Q5^Rgkf}-SE8-qiD4c*GVu9y}y4myDbC1{1
z-ofNuLgtjGfS#n)I8!S9ah}YeN3HH%q-N6GUjJF3bkdvi7(Yw_2l>Jum-1S~rIC9v
z7vFZfn3Pqn4|`PxEXq93le4M3Y}CS%mr^-oOHRsw3WT=CiMU-F{?2uLz^zB0^!~K1
zbgQnR=)kHKoASW%hOtxJ?K`*p@FY8I%$ja_dMpc#-8jKuIp4$?{237`Xn2dYZw5=(
zr$J#7svD0+O9D+gsI27_*xe9>jWC?PU)M%R`O^72Z;?Ug(%s3}{a6LuPkf(F^{naU
zAJFe?Dd3jw^8BW8Ee~-J_3;^BO}8~s7-xS3wXW+=H}aIFWe7ajIuswE3)@i=&QEn0
zkQmI1IJOf%d`7{u?|E0^>uf#?UTU0|odnOi%8S<=zngB@KIgTY*COg<Myhp^W}w&#
z-quQ3V(`#?_eWJW;y{iS@oLftb^D1fUKD%Os5YuE$2hV~w6+J4GLiq|8tIX_JF+pZ
zMOT4JteQ0oQqz3)gaHXZwYd2O1QPm}+v@nD*esO*0OLyq$iX15yU?|7{yJ1)I~HxC
zICl2X5Gwx6TQUey+ScTLTfWG(Uw=}K@9ukS&<0Yo%X;7j<W@5Z6$Egyl+~&j9#j9C
zoNy9a*qxZ5c)Raua!Jz|yRW927@So;yR1v#?a{FL9|<c0f}gR6IU+im375zkMb<^U
zef*QKq<ATz{7KIi_kAk^PfFCCf}X|?U00fB`ZxQ8X&&<Hz8sTxjmC!EMQ!%)x@m^L
zMkI0GFLm~$-Tb>Lkr`7X4@B$EE!kt_{2b-()?nJd%H{vjOXyAo6)oWbidyN^onwD@
zAwzVh4!7uhAy*#Rt@e+)d3g%kRf={d__DeafY$2Q<SW)}EO$l9+NoWw+Pv!?mhJwm
zp>y;rq7_>mGe$%pZh%amQv48%6My^wx7gPA`TUSmx*P+UNy?qo9+y{2>@5C4y*H=U
z&JNyU6`HOQ(OL)*>cy(nW=^cEwtAU5{n_>wPtWTd>}3~S6Rx(n=oW4<wJ)*EG!~9C
z%$Y?+!<mTvFg57A(2v*T-GA%R4|==@UmDFK+x}3@hr#kr3^qMI)geh2sfm0Ee{VQq
zWEpz|s(qq4)3?0AV0z0U{^$5$nQ8Z9O)l^;1jP#z9j7#McfjZe0^SOg@dZ}zSV(XC
z+iu#nWW~_HnYsoDg{HBNa@z3Au4$#vfGF)8ykv~Z$_LW-$TyivNS)D7DvRg55&2lU
zRTa*HD8i6pj#zk;9f_3kgYs#r8OgL;qnD^Jzo|*`aX93`_zP=DblN?wt=#w@ncxKN
zYAH2?)v_YHo<@12#M`Ji2Jf0MKNd~<4eKTQzX{VLF0vv|8l2&vhvt%o-ZMiVs?=_D
zVRXsda}S3J5$py?*Oo^A)2QO*Uf_yG3Hsf5975)N?b@{{K`aQyRo!1}O<o)Mpp`l@
z){~(fvc**xMqr(@+tB(Vv?cpa9ed8^a@Ww&wB8-aVsNUGBD7oS|CvqT;IF>JtgzLz
zA#Xbo_|TDgq9~5f?g!kP!-1W)3gxoO*;3--=&?v_ETF#ZdbJ;8bb9Hk0&eiVI$kZ~
z(r>&Vdb-msy-y?N8SRHh2BW{~Y&NGPK#8foW`C>=6yO6Q)OtgdKqeVzI<$B0_^hm0
z%>9mO1ND_Ba&mJ(oHbhblK@CON@^MY^c%2xCMG7}LK2QR4qHBr=Kl;lH)+H+DKT*X
zNb<nWY8tQG@(lv$U-W3<4zI<4f~e8VkI%xJ->&(-Rr6eFH{t?{ya6zI0lUFV>U<I>
z>fwB{*+R}|MQP@6n&(+eYotS+*Nk#?;OG<Yg_Wh;?fJ7kivMysOrvsbuXlfu&ENJ;
zY309&WJ>eZJ$mGQ+k96T&n+ljlRnCMy_xge5#!$S9-o<a|KQ+P00)X*e8)vG2)xiS
zGb#Cia|Ck%L6<DorfSf_=xAs0R}^1f<0>!_fxx*DcYg8~+8co&rooij@ce#We-ZF>
zP$Ow4MWH;^Y4Q?O=t1bLsL`V(kZKyhbObo*Z$WrS8f#|hXgu;MeJsdkj3MWzxMguL
z>=CxU&}Q-hP)9enw^{1)ii<;mkOQqWn)-1-JI)6wYv9!nmRjC`xv61+YE&`?eS6M4
ze&$65DBP#>zP5)*Xh3oB2PbXjXGf&+LKn-{y^o3NJ+MPp(*hd`@sLj%nznKFbZ`8F
z{E++qF34LJxAl*04U^*19K2oKydLiwS3lkX;cWE$=MAr!*E`zV4S}puURl|drS?5*
z9-h_%j8_sSe1RqpwuFSlm#6eFGVS1CI*5sl%@CFh{5dtZ6Oe%3{Bb7;Wk)b(+o<R(
z%s4|fpmtthH!lYa4gnDS0*oAf`h&ke>JR7p$obn$>Qz>trn`VfQ>R%8iX0Lh4dWB9
z!7E{^<2}emq`8dKWTTq<a<ojL`x3=M&g(b0dz~Lfwuy_2uZ-8lVN*ND{}vdyZR8N7
zru>OX5_Q`!p6OsJW+2bL5X~}jIqcG_N8)EGCG;fYzbe6uD<mZ(Pr0lX@9yD--h9;L
zi}ojdLV?gMfZ{yAo?eLNUwUxiC4-(4?~qo(Cr&SassjOqFTrnX(~}kZn^7as1Y-hc
z+ak7M^2+3j_NI1Ut`1&(QHQ8j0$OBTTnCV%gUZVIj*<QKmd?(C6Hz`sK5FE|_T%F_
zFky=px)&T$2J6&7!Qf+u=yA4Ef#Z0s2O6tpary(6i`dxMaPGdqh>e(zdkhS3$#k93
zy@0$0?CG+tW7KdqU~inCt|KffA>rZ83~}yCu`<hjNNE)upc}{Vk^TJ<z|#u(^a;5S
zv3-uq|F#Ad^1#-}#j+PKUcuLJ9SB9|lB;(z#LF99Ch~98mq^}P?;V~#Y4BB=iLt%P
ztNU6T)6&i1!MyH%o7Z2@<9oSO?aMP>uHMm$cI0g@h_!P)OU)nV7Zn)*r5^1B97)((
z9G5X68qlI_{DzS-i^11FU%mi@i3HqK)$v*tx_Xh3Y)K2U+pCtvs;avf=LVW@g3%J~
zz$m*^52H$$iJi;~q5m4KIfzCt;j)`wL<Cuf@HRe5FUPw0ORtEoY{Qh=>HBL8ccaV~
z^h^>!i9V`@7*>|sXWV5Iyw}on!qu8R`kQ^@6(YHjGu7=v@@EVu^+Mx7_=YBB^ba3j
z474Lr!D`!(zq5ufKTt`uV~Mi=Y@IKv9@{0^+nKY;@Jje-AqftDkbvD>D48L;l9E!Z
z7WRn)?_wBvX^~O;6iCylh1tr|&zgM)GPjD`3c$;%F307z(_aJ~nB_%wiNSOyk<AvG
z`MD@-`kOc7$MB-4gq<(PH&*n{N$b0S@9PaD=e4-LUVF?Ah?2?wmWE;d=+UFFfZ&|o
z%Mu1)ic@AfZzLvGW~Ccug)cezlvR2?q&mC1Y3JqCG~pLwBTR%!%m}M9Ca?V?yFKXn
zF<NKgotJ$3-42p3tnePEOZt!CvW-pWAbo}FB^;cF@$oq5asLFZBYAloXyYFHRA?|n
zIJ;OLu&sDv+dSe|S<2nFpF@L5>u^nVH9u&ub5w6&0QK4sL~YREMnSo;Smf+*OV~W*
z^JjDgee{x{mqtbye?V^EtVLU6VigD+0i;DpWlI**6Ms~}fiHUIj&bNzZiGnnNQTw-
zQ5L8@Ni%E|GeD{vIz3Op{=c$K%!OT!Wr8Z9w#xh1cK59lAD%0iylz{6#!LbOkI<ff
z|3Ko!NxmiR_9oM)*4336uZ>Nwma=G2d%C~EuD1@!$QZp31TUFz)`awQ<Q-f{s9Dw^
zRn!QxPfU9=NkCvexBJk=%?%SFbzPp4PS}-JEM*2tHOUm;<6q_}BADum69EAMbocIc
z&gp?!J$iV2+Ehp_O;l2lEgA7vL^yrN^vFYT)K&PuLIQ<c6m<iH6%CWx?>O>V-;p2a
zYQ)`{DWxt&WO8G{jEU}pj0qZpeC!NFvLmj`InRFcuMQS5Y|P)bAdpJyz5){sfE<n*
zaRA*CqK}h4^NqU~$sO-6BnanbspaKv-?OfisxaG<J|D}p7^^1tS{tq6)A*J8($@bC
z4B^(AL%A$&Tn|N@_jN5^Ar~JXs8chxVv`~apYu=o-@FiU6m%P=WBPZ5)5%I6x97+Q
zPR;k%d1D#tTQLsw+P#^bld%yFB3;$G>gtwv@n;eH$?54LJR_LqO;OQUj@knZFCdMo
z33z^ccLfdQsMT~`{Gfx4TQw7(TKKvbOlA-apb^#9XnNV3`7B!41;)W3pEUQ$LDVNc
zS`Ws8r|c052nhU&6N9>sb(a5|cY~qH5<XC%xY6o&$M|kIMSPXp>G?!eq$<G`tLq<j
zlPSf_i%$OYJK>A>Jki)54m11T?hqqLb<N(k7ls+c1k5C5HCuc=k*RE7bxgbO!cb;t
zG($Vlpgd(EB0~OeT-l|VHo(xn178rrn&#&Uqy=36BiznV*kv*I18hX)rCZ~=2`qC3
zVB|BNdUwPKwcN0-R!s2o`TtRC;cok7#8psG0OJ<1YjovR5EWt&M#xmACb&|>q#^91
z2GMOf^=xYR?u?nC57xET7c(hFS~UNG()uI;HB|cue}H^*ja0)M{2yIS=MD;eW9flE
z@t3>xCk5_)xz4*yJ+x0`S)H6^YuL8{qImfAzBt!x?CrB1PsGj-<v}XKc|OLe{QT@V
z3*HN^-(I1JfdO^k%%_;G6A=>TIy%}uJKN1eK8J>WgS(?qWPe0?_6v*OGz%$*?o${C
z4jp6((A62%{MOGeY@anZ3%)3omlyL;UzD_|{`<m76CXWq!+X-WA98~JD>dauLY<o4
zQ^+3-sIk_<`QS3EVA`)gNN-fAxY5m;3w<dKAuSU%T<K4ZQ2gCfYO2oUL%qqVDW!x1
zz)5>3TC!A_A;6T;)~1AGK9L$_^&qKpVrdCA5(JzRij;s)&Y3(`;SMojwoX+z`2VZK
z#3;gOQ<IN=)0={qKd-)c@d747=h_Jal@U(bAV_D3fp<;@g(=|W4wYKA83c(m1K*Lk
z<msP9dlX41j75zT#*s$H#B?s&LvPMIn2}{W^WzH4vRQJ_g|97xdk9~*MpsWXpo*Ty
zg93HX)Y}G!We0zBY0*Hr3JUSAtEoHaq78?bI5Vk+`i#uq2U}ubAG&N65Pm>GOR%qi
z7lEHs`3Z1pP_FZ5%k}ebwRyY-+AddAbye2v3w&POxqfCbe)%%>w|8XgNwOUi^sVJ)
z;}Ck)?4xdbKrLtSihCuXXCz=BNUta!=elmO8Yxqj95gkc#(*tR%IBzZ0T-${kjlJ>
zlJ&jD@N_RiOp#r0&Krg5Q$=-2LSRSYiuC+t4RtC2qaZARA5X2XMl<_rQ&o(0%p0<K
z;;myzu_C8Mdi?Lt$iDEM7z;p={Q1I;?S;MCJvWhmF3*<jm5sy8Z~pt8aO29IRL33U
zw3NH^2Aq7<E@qc2KXSP|P{jq;7IL|#3@<%fJ?`N<`o5c3W7XBdZwPKQL$Yw=b~Qw>
z`uFO7qnDTL(DL!(bxGHEG!)(^let-d7SzWU+E!;~d-nWxTN+#PgcAEemcAF?&a1cK
zf`2k@DMMQSyu}dE)UDa^`9J23mlMw%kPwcIe_8dn-PW8G1`m3|G^FEOB^w6)x7{Ce
z>(?Q}0G-d-@1sC34rkSPt$8}1P<y;4gqo}RI0=2CVjfCTna{E{OIrIDQC}EIh6-p*
zWLa>4{91HgF#08$(X>AK<LNSLgP%Tjp;K$0NH+{GRQFs-bWe!7Vejbe8GAQ_&3%Z#
z$R15iXpe357viK$3fu(N4&}cFfWrTcwlUU$1bGM@uxO}25&~MIMqixS@7i7v9R)2c
zm|VSf4QvsD!FSLp)cld$VA&pw=biMC2&v!ib{tRHw>5*iL>rQ%7QgB{o4(eTur}H-
zZQd3W-oN*_<_g3)zn+^U<alD2;J!hQ>#_O7&8arEg0_ZNj`qi{{<FkOwg&kw-cCP*
z>1KyPtR+Wb;4}+*M919LMIbnT|L)yQ=*IfD31JH(G3BQvbu3x)XAIPVJ=6Og36x)A
z(2TC{9{KQ>_vQ1a5B?g82v^cnz@Nr^9xc{WQW8;zSv`oQy)uAoe}BwQDWJSh`8QpL
zWsKYf?(8UtbssvatE-0@<mx_IZDn=<I*9w(;e4r_By3c=5&%(LP_9q-bE!y<d6nXW
zxrv{`OW%~A8;wkioSX|4oxXt|vp28fKpVgi4Ur|~<YeEjd|?<NwK-n>oRexFf}qaZ
zrCSJ)C1TsjG=w0tJ1q70mpLq5$~^NmDylefv$kH}8b-;C#(L~&{Oem#G5u5IG&wil
zaMWMhyff>f9)nxvEbcoTXGzQ{4Gh|oSR8?iJ&qJ>&dcJr*D6x_4kVG09E%eXG;Y$|
z)*G{!ZNUQDEljVJi-2}&szHhR{J$IC9t_-Rpr;d)k*QUVLi`1rjt<v}y6#C$<wlj;
zzLi!Qj9z}wliqTGeqn6nJsa?;0c3h^8M(Eq#!Qk>CdoN-Sm0c;ox_uhYHb<5m#-{N
z!}6h!f@Ej^HI1^=fMYcE{<xV7>%WD=x&z@O5zxp*#bpXgO0VV8FmDPAS6A1&9d6BT
zVOYaiIIJ}*X%|9qxG$G%5t{uJS(m>B;TY;LU$+Y!>0w11Ijz|sGM8x)p9D5Ymff$!
z755Kzc(;DwL}#u`>M_KGjem{kP*rsPFZ8@*`>@ExyG*an6M`*luV2Worx+L*XyvP=
z!>DmfXakVSUZa1CHUZ5*v&VaDqr?A<%wcY7T5jLXI^npc44xGO2EO@76l`&Xv5m}`
zT6?@eab^C?UVSecT8J7tLiJ&G7;+ts@!I)Vx7M;W7&#xwV}39jp?96j=;`4q%tkHG
zLAtLzcqhoj#(GO+55xGJUt+7_QDHFW-KbwB>x_?hihs9~uo9E!5UHB5`kU$(2;6y#
zC;9smgiK759ZH|R8hgn7cV2Gg;_${b4A&v&TdOcvIT}g_n9$na28EMf5DSRw>+2t`
zs`>i)*+L>aEJy`L96tDR4grca<(WQ%v1450%XIa3cAK_I-G43}$x#G4DYVhIDrx6w
zdyI&uybis37n3x%;9|W*N?A>m-l*Fl={#xSe4A|a$t)UxrOcz_KsnVflk5U#S1cD!
z68OIZ#Vry=<}4yCoT>eiV4|d7HwikRQECFj#KfWFh^T27Eh>ZOreJX+#=Z53DkVnR
zX!lnKpfJfJ6ECo1MjEv6=;)uY(Tj;j_>5c4|Jkp*c39?(ALqc~NXT10qC0XKh-I3R
z#SDiEx4)6yXE1K_S#7^~Gv6sqkU*^19lK5<FP=Y5p&}uSa7chTZ(AEPnu!(^EHW5^
z2TqYg>yf>#@k>y<nmzn4TJ|;J1RnT<-BVl$>4QY}u7}Xe-WCyw>FQE~J}}lbYpCjm
zzZ<pp4eLPBw^W!aMS3o}mivtArK-;@=z9px%`ySq9FM8xH|JSW8sIdRk>9>Mq_u5a
zN~y0BBr&qjXZ^AGfIrgT4iDjjC(pA@A2dM)yZYz-T494tilR2wEGEkb5#IG*9~|sq
zU}<q3P96qHi8#b2ob#y;%K!D%@96h5Hw<KBUSIAHfOfOK$2qFqg(=!=tiC4hKi|Tn
zsGYahcON^a^PJQ;>_+^q8g3fLc<xJ7P0<3eth%TJIuTn;Ao6&aR!sD}L+bp2_sYov
z*#};udMeM$Q%ZGc^jxfhct+*3??0z=^OXG_KUoWsw(TFK)?K{BSMf|l0jt>QHJMPO
z4UIEPYz)om$E3Ghr_}zLdV&8elG}@hh}+@fE?=s#vhhtT6(EbRqFS!O{^#JA)gKS2
zk<}v7cpo%GbF#~Y=q^)dCzQMKShM{i@Oy6jo95}uZXuQ%!fKtu{N;k22Z#4sKMwX!
z-K<Go6N>T<3YoBB9}{O=1XJK4ln3VmkAN>nF}tg&W4oW{>QX;^EB;y0y(4URqinbT
zo(Uk;XmC0xNZEP2ozG>v9pSrS59O5fJTcMH`#?DyS;8_nT`nu~q0n8%Cp6n*AAIV#
zmGVYPiCgG15k9TEg<iQfZU^PtFV#a%%)AZJ7t~#@V5><sBiuR7XB|dK@5jf^bPwka
z_bxA}uMRLfIC?W+Hdq)L?BL#wS}v>m=WHLmqc;FDGyqio2uzQMS~W%hF14jr+1Dy$
zHp#E*^d9*~Fd+YSJL2|G>F+VtB9Fd}o=z50#|C+qO--7xUq4rRQlo;)!qM3D_W3lO
zw#Vq#jajh`!pw>sML?EbeoJFyyLr)BD}3Iv_D|nuz_`k7R<aH6{pakV{mne_eX?#I
zc0DqttC~jkfcqnWxy19t%>EfAT^iI2ym|*m7{EN@a-S0^ycw2$4dQ;M(|!G+orQW}
z1Tg~PW0<o6;9*U}HwT6_N)v3?1UZA#Bi34a`;+<(Y@-4hZ<HNs?oDDb(s+K{@RkpL
zH|qSwYvU#!=KnclPXQWeM_OB%$Jd|`dClmyugPzAO3@sPo$MNQ)ueF3%kk4&74{!q
z;o9t^&nIe3HK!<bp4MtXotaP4b3kgPxrvtQNxe;Oy8Aoh8#IE1glWSAwFgcF*>($f
zlC8SqA|p9%_qs4erH;1G{r|V~=CRMP#(aG*`~u|#V{QthtqcS|6tl9{F!VXU{`m3Z
zrl4TuNhwqY)k6XyAZFnZ(Hh+7)};B>kIqgU5)pd`OIOialL6aR_WFZ7oQL%msf=Zp
zJrtRKkL8-IupO5~N5m%m8rdz_+Fi{$<%<2^$1cBykNxxK&(<F9Gw2SNshI+1Z+oOY
z+`oyMk39z8$~*F!h*+g%URS;(lcB<2DKx#XX7TL0L58?#!E?*DPle|+k82t#`Tytc
zDcX4oerQIp>l~c!kK@3Uz$F#9QbYaa2UC{6rtlDD`xjo%tyR?}<nRqBKdAI+z^_mx
zHJUA9`_BgYsj`eVs6`R?awY#aFvS!qnrM=Nl;Ws<$vo+l#-BfNtgWrJb#!v1053P2
z9TvDh2)&WOQKM!YM>F=ffkv;v;db7iIQn$sa_C@k^_$uY>xM4z(-mk37qAByx&C+2
z6SOap5+Es^CarkYlMo2`LRBkj@sAiW7!}eiqM`8o!hEKfduH!m+jGutTN{4nlCX#-
zJpDT;7^TRH#^PsjZhaM<@_WAGr~uAu|Ls8MAB(ZX$&m`X<mBW}!ND(<|Dd9EU{U}E
zC7y7lwrpcw`8`fJ9p<0YF)hn4F_-%mg;_Be<jDqNad5E0u>^wgu<bUGOB<;k=a+}+
z|24tFx6z=^<5R|bc6L@u2;~UgdqC~_mObjfSw+0CMmW0E`Jw^;WS#qEXqTxCzvZ12
z<MMJxlNbkl=Osc<muam7YmDmt63?QZqTRa;SG=HO@v(vRwiwaBT{pRE6!N}&+KoTo
zPJ~Iy_N8$ul3@3$VVKV=e+_e?6QO0dQq{&{_{I~jIMzTyLIOepbWra{1C?Rxk4<np
zVo{@yeRQ8i@2jmQ*7f&^5MuTl8@k?}zd&hG?F$@&q_OBf8$WRs<v$}dPfz*`fMVi}
z#JonFFyxPioSt%ctG&2Y;|o~1Z_rvdyzpQA*I5rm0mMSHq0o=xb1<>m>=yzSvF6{p
zGCiZp<?EdhHrqR<Gpv631W{!s<&DtC0GLFj<LU>{P=dQW1u9tf5bZRUqe{NNB6<8Z
zij4H{s#hG9LA;m<@!}5hBDqj~2rq3S$?lU@Q%CF<;R8whY`O29qsZ4y|JX)w8k<*1
z69aoFc%15l$j@`VZEh2#Qu~Z^*1ybk#r7eGrl{g+o61l1x-r*J_wK|8k9GYugQ+7`
zZI$}yult<FeICiQ-zU4zj9f>B-hMtGX)foJx9x8`Cg1tybYb7^x{71=@t#O*Eyu@@
zw}|yp4;v{8x<&qCx9w8{vM!?iNf*C1XaSx6-IY7wktA~r@bK_l0FFX0hY0T7$B!28
zLcu^r{wz*!Sk1SgLhy|yi328r$|~`67U#i-pt0~lUym@F?LvK?jQN@tKS>Ag*8ALY
z$HkEt=b2@sa4&_9QmaVPv}(ij>c|R?ci;qKhAnWBU-^<-QK)4++`%Jzt<I+H|FXnf
zx9<{I)hFMcUzJ;FN|6vcl2w0nf8m|<HS%gqx2rwNqPN#$+s|l>KIbpX=*DvTl8qgH
zjy<)uPF!J20|YuR+M2M%s&%`9n&*d51&kMsAajE5iC<ZHwWwD2dtaZ1H4`FKg2nD`
zP?lE$n3ns)J2$I@Ca+wJ^XF_Ws_Uut^1LfQbGd&D{Rlb>vA*MOA=VRlv1&p5QdEIc
z4=2zF7og-dJa+uP$nC}C{Eqar3Iz|<MSH_PMM<V4W4W8x{x#}0HiLc9RExg%y>`4|
zif}c>p5erpw};qMjv`vGmXkS5t_*~Xjte&lahwGR=&jv~*NJ%Cs(ER75ORTudeneb
z0_a7?94NQF13F9rkChF*%V5CS>@ZY;3;_fym6N)3D|2RPxgouuW~MZv+0IIj(sc9Z
zOsbJVUQx^Fc$8y}&?Hq)VMaG`<87c>Z2svwPrAt!RD>YWx94BbOAKQ1yHrk$Rl*30
zs)>cQksi~3(Y!v_vE#|N-{axaE4|_R@_890<!&XP66BS?#58us8E@O}C|8%-7+@Ni
zxdD;hO7-7{54sBu{jWz_xMwMn2vtX_^?;%0f3v8PE|XQ0ak0=?ga-MR(=j3*_fzrF
zNVndHszeQ&Jx98gK;0d;6W%w!%cE+8B(At>$uq6}eVlhj(0#bQ+ZchiuwC!ApzX0t
zm2~{2-*%f|ZedK8a=7k$A5&j?d^{SkrK6&>0?bvQKkrJ8YFko}sOIEeq-)<BU*Gs%
zv!6eIqCBVY=?>CeY&}j5uK<v+wIwV(d<#<SPMC%c^(JU1nHhB&Pf1Sp1x)Ayya>qE
zYD}IEEwxAGfTap=|KhrE@#56V%E|f4sFQPu;rZDnmu!fYpt@mDq0{HYAD0!xwJ%NS
zmTzS7Z`dY^ez=A$=Q}^&_`X%;3K`>#M~}%1k^R)-6t`6JI<W^V1EfRrbTz~K%lyLl
zKRxvN8B8O2jq}{u39jG0Uw3w7p0QuVD&ov%J(i74<E{WoM#>5bpEOG?G9cG6pKcEL
z9>~VX_~+zs+o3~bKOOE@TpS!lfI{qBsGLjy7fAW7Ppqrg;P$F<4Afr|+ftQyEs@Rr
z^EBV&WLoP8%=&stEJrHMGD*|7!C^7M8SxvJ{rEmAg{<8xyny9weLF%1M~jGA|5)6R
zHua}ied3Qd4n4(gEPg#1vo$VgsDHwy>epn4>Xxk~c-InBb@H<RT=z8cuhbS)Z4<3h
zmht?W`CH_ob46>HpRcb8xXefa#4Z&8>IBczqcG<Y5RXWD{Td^Hf@djoS-!uh&#Ejd
zKK?#fdN0T#$XB#a+B^v^EG+Cvk|eT(#{7lS67zmD*odQ<5P5FrjprngQXMRwbj`6x
zI{{0Qg2zmyQ#<%r`NW@;t&1hEIuX_#?($UXm0%3$1ATO8H9+S6M<#$_GtRBMFfF$i
z1|{%-oCAp5r^9nel!<Kpxbr<w%x1z1FnMK<qs24|G0*B)_pP5-=eaQQu&ZIj0!t<3
zWrbQadNHx_wqCTmi9MK7nw+M&rDvoZT8_E}!>5?6W)9*EyH)475@ps(lP+7FuXg4c
zwvl4bzE_6SEMtfoeMscD(w;*ny~I*7N4CCD8q|&|@MP<p713sB?CVq4IGJ~?N`HOo
zDKQ-*yFZT9K?{C{R0w_8^$nw!FBj6{JVB8LdUU2*Mb2Qqc%Z*NfJRIg<io@a@6Ay;
zjU>Q5ngwH<APo<PS}RLLTM7cLzKg9ac|A~Me>QpS0F~(_I03x?lc|(h>CDd0n}dI;
zgz^9ifnJ3F^$ftTuPtt;yZt;c?L{L;)dT9i6oU5YFf!_8*>DM{z{`H+Oi`EgCS}t!
zfzF!BiPv~4RgRWl+l>s3jLaq*{VEqzqqw0EN<bruoP??tKAzh3;=6tO0S*pM@Cbs%
zsGXptnRPL$UDdt<uF^4fY%lNfXm++3VKtetIF_^~CXJKi*egnH&oja$S_@^cUm5B}
z<yvm{nGRPwEZG_$1fFw(U)g5e1cuL!UE5u@xMow0W$2|=AExZ-D_Q!7u7_UTN-0?s
zyk4h;bg#uE&&`+iZj`O7nv8C=R&1lT-3wQSzgR0+C2dfEI^f!MP056kC?+14Jd4AM
z!iyJNlBziuN2|&(Ffn;RQUjJNaqCuE>9Z&p2SB@+@RnN8K&B^H+rCpGlG9MSKz@)$
z;MF52nt%HKq7#%5NT|;`j>iCj)@zGEIMk{fNyhEh-rTVI1BT}h%H`M7QC~*g>0VSV
zoO%Gyv?FvJHNG}W%gb|UZdRG~GCP3HFYx+*PXx@}{N!Ns`)@bsDDu})%Ti?{pb}K6
zJK6FF*|apkcom?U!?-@0b+vlTr%WwRS0bXV>(%58Ngd$TdyWDK8h)dxlS$u08k5c{
zyP2H`dl>`MW`Qo5)nuVa={pop?nFK5D7Do?^z1Ui>-yYlp6HDoslI3NklCJZsWTm_
z2^pDO`|gm!@gl`ZNb0=*X=Q4Hy`8~g=k>+qwXSW;lvMGd{eDLe;S)o;RY4^UB6QUD
z#=q`TzDk$zpvDO~XUkWuQEnF20yN73kWMM3_ot@g&eH@^U6y5|stKDd0XNiM;Pv|9
zjh?;%K?o1&a;<cTy3%GmW_Q6P!<<i{dzelNdf1q^)$7jJ#!l7>+t`KZMs^=T;hiq*
z<jW1ee$27Srnh7|q7{mErlJQEK8bAr8`YGxp(D>PZ#T>LhwJrA?_wtJ3szu`J-7Rb
zt#z85m6-A~CPnp@V7^gWt=o)2MRYFL@1n!B-pKQzr%d(oN7gqM3$E`+)7_I=B4LtU
zVkJmV`sl0TA++Q%n`-#%BoG6&(rR}8Z#9UtxbB-41Cp7_mP@z-$8LqhR<Wwr=ST%#
z%^>eUj5H5UN;(`w5{fDSLg;kQ;c&|YvzJf2R_dsk36^2zPNU_q7|8#&eCN&`Iw^qY
zz+;Gt>IT~bMjufX$fi}g0G;{I`g3(iC@7e-df{KqVkK8lXPl)j%G_|R7qS?*i*@m$
zbTGC1hImR^S``Dc{^9`))I>hGIzM{+cnM~+?>Gve;Xj{GtSeda$5Q%Ix8sVPAz?Ep
zmrZuFW=PlUa{4M_@n^`IG;VJrJK4US{P?R-xkqJ54_~Im_Jso}`O@IuK`KiPw5KO0
z{f0tBsMXu3_Plpo?+Es^cq=*0zjD|rSgVKb-GGb3;-;r=Z3-?Nq2M-&hs#ao{rmIm
zo3Xr>k+h2-`H{Xcpqf?kZ5i|<X-}_}hbQaSe8~IA8zrM#{jAGuii>H5)Mo3Vnc?Qi
zc1IQr+Hi$xMv3k8V?$$OGRqm&^26buQT5~j@V!HY-UC9|;N+ejl@haFY0xb@WD*+B
zQ7?EvBn<jLOl;&Z4So2Pb?e3Y18up8voCksPJ`~TrCgkKr%IpY5Scw&o}GGc_p*)2
zqp11&HAF=`x4q_Qzq$*t3Hw32xT&h-D-kuH+t(?3+~y>TT6KJv{nEYr!S*K8^3(w#
z`5px;rIZeVH`Ei*x|Hcdq@beqh1L`Z7s0d<ZOF6Y#IoBgYVy=ow0;d%nwmIKL2(M<
zyxGPN;f&LT<)I{$Oh9A+7azX|!jCl28G~Ey$D)1f?5mhUu`h})A`@XFBHXxI9JCA!
z=0I9VyHE{hyYG0tPH$&DNc8B5&qadFk4k7Nw%F57EVjq`8M5zs(mt}7o#$AlBRzc5
zgDqSomb&im-xALZ6;D=?to4@j@#YQ)HyAB8KMqbV8IScWxr<uZ#LD9JVfloTM0@9M
zYiXgvl*Va`*z;1_#cK4X>Svl2B&Byf;9j7Ug0Qp~f|A}cYrWnK<rbJZxi+E^T<m_h
zr4$6sO11s-Iqv`<mkIVnz;AKJ6@k>MR2M*@Ia#@t(r^aC>go{137zuSgFl$*Cc=^x
zlO8X^RHn+qxhNSCBbZq~RO42*R=J$p%8YrRq-;89cMXrM&mk{3CmnCOvPCPu;9J!G
zS;e!xYK#nHU*a{rO_!to@TUxOwlNj+9Dl#0twXJB3P)sk`c*~r(em)ynOEWG(|g!`
z9i5Z1b(y+8b4yM?Wy2H<2f0oD+@FuO3*1{D&jWJV5|D^OXhnd162_?q&FXYgzz@kS
zEc~!8Py~Y$*>oz4mFC#<{5fjXgLfc&G==2CP%v#nRD(7Asx<<JTB96I^h~p!MM^f3
z8|efUsEm^n_tnuVHJC-z4X9`d$YssW%$Ps~V+OdOVK`d+2blraKOPBLZ`+sUqvP^-
z>lf!2rzVB3kY;G)7Y28Qc?fqG6`+=drfWCoK$NcE%_~?q>y#6_sa`=kn@l~FPDfKk
z#x!=e5ml(}G5okVU$zNQX>L=bi&pB~H)^b}+|~efbMQvVo*wOXgSzX?tBM(WK@u98
zJW-D$mDXw!;6Nk*(@8uSFp@O5mq{$U=4|+jk#o<QgDL;CVw;7*qBPolgG$d+SAiHg
zsV+#lb|Ek$9scwCov?bds-tSwv-T)HHPCD8Xtqbgr)3$eL9Qg3MU1S}{Ce<9>63=W
z`$f8r5c{bkLF(Ukq#Sa(YAFrrQ&^IJ4~{=0L9I?f`~LI%SXZ||>bg~JOR&9HT!9(=
z?IDH)xE!an5i-Y9#^!^epP7GjbZvZ=U#j+Q>d@^>DPJjUqm`oZJV>Bj4A-(!R8UYz
zY2Sw;Y^D}cW~EtX6$7}mTtyCCD-`!$_(CMeC{`z}!=gzKedoo5l>G64Rj7j&0psQu
zutX9au-!7${Yobj&dAGKI#2)=@-7IuSaQaS8nX1lK%mPUc#@G^#)074lYOv>T<x6E
zdKDByoaBe8%4r33I1j@{<`Yb)ku>A(C&84t$}JaA>r=rAh%D*#KezTa>o?C9K|FH4
z8W<E$rbUFs=AVX6<8E=5Sg*34a_YyEo>}|iZWY<L$8;>r9qaBX&ZglK5QxxKSA#h+
zJJbT_3T7cl!0oR9*MK2!RBwfyaSf;jXx4ewfYZIX7$^JN{U_4VArHFgpTX9W;QZ+T
z0X2NkLbW?p)}dd=$rJ<&GXX2=+yuV`NZea`*~nzzQMJNPL6qJFLIo(VI|`z$qSbt3
zlVK#)zya&u?bpkLP?M}G2eA4hU{`$ze~4NC^TG0)<$=<NO2O@;XYq-gt4g)X2?)U*
z=@`l2&``Zi0bIHYrvdydr^q@COw?w$CmL3*MQ-#`c%1z<&^&2V`*6NS{vFx>$SPh!
ztiqzXSN(L>h%%&YRu{b|-}FWRC6~DEY3_#YY+B?!gU4Tkbn>5zXOyi1-Serzg$oxJ
zK1)(}-U$*`S}ARm?Th5K)SRp6K)BZ>$aHAcnEY$xoDkd?yWav3v$c{3h0uafI2>OO
zxCl~u^@*wTCSt&IfiRm4z7w6(ojF*u8sPsmY0Bv7ZzvqC)@t)}XZmmr9j8v-=tM>c
z%TMkua|%Z>4AN8mdgh!ogd2#zYr%?I*Zq`t?=th=>K;5;e~KJysf~-Owre?k{;3iF
z(Lg9P>T~=?y6Y0xf}qG&E1#=%^BgnTnkt=;(^^lXAD&@}rhkd`{oaXA+_8}nZQ!70
zFN_0*SP68J-dr7_)q#SCV2nyesczR2mvgJ+K$lF-gwgww7gAi49~CZlCKVY+8(?xX
ztn0<($GHmRMtT|5K*_hLsC^3-^C$5Z=G`N=EG67Fe>-(_8$DNhPfCD#WQ7zS%RxoH
zAWsR@FZKwT3W2Y<m+->}-Bn9dI6p%EC>zbM2$IGsU^%nv>v_N<VACvq!Z+reL~9BW
zDrZ!?Xf83`rU}+MXs?FG#IXIWnI8Lm2a((vGS6ECrSATN&1naiEnEWR+TfQ(gRLdS
zl=s{J<lAxa1O|NGn<w=vo-Gx9s2P*G@wlXSFVTvQ`bFtcRto{y6UT@h4Xk6IG?FLX
zK|f}Ul}6eznRzdwp1YpmDSgQ6hh|sS!s{39m%fwfsXTr9k(JaR_1CXobH7A>Tuw<%
zUFwqI=zT~eGPBI7h#Q14hVpJYXFJ{UB-EZ}Vy*f4`O2>PzoG@YLl)6gB<VxgT~EyR
zXQhslMFeJr#0KeWme!WJHYT%PmWH4CRPLV>C@MI=6MDjK{l&sbieojjkm^HCY2H%t
zNA~eU)MKHkC_+}&4@?VWryi3i9d5j>?QLba@gdB+dwR5-7b)b#p5>Hv7aj(3TvDAi
zpo_k50m3bbj&6}r07Ex=dTQjJNzNIO4%4hq3PY(h5>X((R!g}k7L^i;;XvhCE2!tP
zW=ceaOVqjHv=;uegSMtb@@VsmO1S9KSI5P#mR-&Gz1B#-x=&nXsm1Q!xcvEL7~IzB
zwqBD4QOA!-8%6;E4Y<_$pj>8{D4`!VrwCP-KM9F#U~zQRFtq!onAkq|`}gl-pJK;T
zEW3@*u9adhVs#m4r8w<Nw`|FX^4Hvte0AUTc}VD|JV&yYr7bdRCT01NFQQaDJH5lB
zd5FxleQ9~iuu|hN%AGl(l(=gtReaK$i<!fj-nr)!=o+svgsY1F?irfylw$Xlh~>CP
zGxk!ogBIc4i#sow(R!nQb|l~G7)?L2ykUN3&J}vzeQz<%Kq>2ThD1K`B!{73GG;|r
zAM4NA;K8N5T7z7LaW@XTyjHd(_9j;*BvDGo<9vWYS}Te*M>b*!`w3NJ@)Q4;c9l)D
z0px3+)RbD<rtJ^?%NTERrx#NHjp@<6UL|F5doiaR@1FKp**KjZZdwnWbepBD3bL=Z
ziq3xdBv{hpso=C*Yy7V9#^Yxuv8f`OraRSSB<HJFbK+_qNb;?6l)f0aN*u3>_y{hY
zC`P0!_LGv-%Kuwun4cHf{jh74<e7LhRhnd?g4!+nRLVmM)2&iJRh~P;dp^prT7EOD
zS)fHx^j*)k3}0o04KwWPmsDS>x1pauG0Vh1yFy@|rQc$-MU?M6N&WY`MN=Wn!>UHH
zeiidZ5f(;9#%|l-9R$3yYZfiqV`I(?M}L!Et1CK-%6`563in^;uV6V!t({G-T*m$T
zT$zIK^Xt{yfog$wCb0fJs#hVC2NT=B-=oD#Z<}CQD=7t+E&ca5XrUpJ=+>Z!|9}35
zR!S{qdKnjOI$q+hk4NB@*NM^phGhQ_zYVR+!$b3;_&%%TC>;&~(aft#Vog5nTP5Z{
zXN<QT4F5h)z1JmY^a<7{Lk^G3Q`x)r#CRE}5tCM?GQYl@ls?SLOz<V?;%P>rnV0{*
z{`wVoE|uF_j<KF@E|Gm5n(PtXv$cg9pHw*NJp=|*vpvW)`8+cEWc4xwBfc+o+43<n
z-tKYww=lG(n-f@G(!rhw<+bEIitmL8WfIQ1N(fTK9a0$E>Ihjsky1z0iXz$VVmW=E
z1CH&RB#F}Ho??y@6aM{hd>`39wD_wt#&c#%RO{tUqq!x?@>Np(e5}-K(n{qLmCTw)
z*-`eM$x1}@i&?|ts*bOk2RSJJJvc7v!C%wFojv&Jb~Ty$O4q@jk~6vJ-76O1vF0qO
zL^UZRa(6>A81j$9@BfUt={9_+970eN(el)|(_0`dUlM#GYbZ8IyC}w%LN7j4_A1Je
zbwu6N;os}0QAWUVygfZWLS|*UvR0#|i|*MKBTFmAB6`&8ANJSiyq4a0k)c)ph*gW%
zk>RU&IE1&X;vTQ7s1z!JN=6vykZw1uzwhT|23C(&&8WYOKDdoLawAJa#Q8mslpw<c
z_B5Hdee0W=9)H&s@0w9>{Pva<1gek+*Rm%y;^N}E>pTdX40gN5@^<uAL<m8S8E0Cz
zhW+4Le%#i2p|sGe5sw7%_<1Y|>?grpL8%)e%b0Qz$IBfCdPN9{C=u<x`Gt<R{~v8%
z85LF3wv8Alf+&gz2&j~j(j5Y#G|~*EGz{G}90UPr>F&;<Vd#+V7=}ir85)L;@1W1~
zzJI>IFALXV33K*2`>yM{?tS)-FWOlVefRFG4E+bdL0Op2!m9_3P2Ig_o!{JUm&V6G
zdUh3|d*7TZ;rW9PhB3-Uj&<71Tvm7TIkD$(%ceJbdVBXBNQjK8_j@~=c#|IRCaJ9W
zy>;(hv+l~<6Q<m=ldor1Z`kR$`icANJW;FB81}uB!bb~`%h@10ZX?L2^qbv66Ds)Z
z*kLmhZ_}80u{T@^^NLhXnzMnbDJ3@f1DqQy3jL3ER29INfz#8|-L)lrZfD&(tNljb
zMiXOeXKPprS+ZKzOs6d;nlNCoZZfTg+%(`y=oOSOXsg}2-`TKZ#L}+HVWjYIzT8;T
z!=sV<&5IX6b7xd@4ERXHb}bv*ECucB3jx2R9sRVc+!L_9z0_MY+m@D>o)-OCGUl+g
zvs<((?fPT2>+mL5+}QOhB_aacHkg}(qeAN90m<KLz-Nz*B$@w>_3OO@M52zxygO3h
zg6bri@hFycnBu`-jEj}efUSeIlMWdU3)ljNl%=|291T?5Q!`FR4jGYV53|5T?gkh6
zYi)(1hdbDwF%EUL)n~25e)K|J#5lA1_`<sRq8D|4|3U;Q+I$~N$8i92@nkr+s)A4U
zbXWZZN8MfPluF~>qnGOrD=)T*@BaPUeEjQy*~*c+YO*y-9!A2ZBjEZQ55-A&ov>H3
z6HIY3q^DDxxOea8SMdFF-`z($*djD0QyG=T&P!ZJ6lkjrN587>suB5*EI(+sBRV#J
z>?M(M9FkU&OkfyQ?*APosAnE^8%x9i>2Q34TSQzusV}6sRZngDu53(j=0bNC>*!sl
z#fUGSscd{CupLR$m7fNmPkRuPxG??a->@&irxu|JSR4IxbhTR+4eS(LD{oaQrX%?k
zdYvaYkSA{Yx7WE(Z9(nb;qd;yWb7*rPDB*rYXNa$L66EiK_vOGf-3ZC_C=43Oib^Z
zvU<Z^q}9AQlUrpBklK#Q`WtotmiJ3APnG{P$EeAqgn&a1qq6=!g{PSpn1TL%HK&4^
z^Z(>W;Q#8v|NrZr!?dy77bXSq|78qmgc<L(pxZTXe(1ov^HTGx>>+mDbzF!TB@=K$
zu7l*r?gnQ}+wa$|%*UGNhL!3ttw-+NTh?9CDe%-^FX9aLzz?R))C(XI<s=os&<rj3
z3SU}W?$HAyWpEm8Rupwznnp^!s()+VK;5X)x}kCdxb_fd@BoZ>@HM;C>m~*$5=hC)
zs(_`E&uOa=JX^N_c6Qjv{OCUhsL!6dRS&+rUyB*n4Af4EjLf`!|EUSM+siu0xxw|s
z4CRv>-njoYQC+@lGn6fEtj@{mx_1P=92fBY`wwt8PB4326TN%DV_Fk%A?+y$yB$Vl
zxld}f{xi;wdgWRR+0&=lbji-5;43J~AS4C3qaj<|lZOve!BxTNIOF6nYS|Pa?FW8s
zoydQs0Lya)9MO#yT)Z4XjnHBSW&YC|(*IskEBKyslAQ7IlBXvYrRsmY4hw7UlZER}
zayPh<8?_|JI#m@EhD<}jbk_r}h}q?e=~X$niV)M!u<2ex6XxUP?5Durgi7}>^R4x6
z6VB6dY7e6Y2fax*7vYV)?fYrj4DQ<2eA;^jti9CMwDrs);XmE{&9-H>ue9kJ;&RPg
zTDmK5<R+VmH$%%OIRNneO34v#0wL#)1vRk!1n%5<P|;~%Y|KDI)6Rf75fzauhD+z#
z;+k_2O}J~{&d1gCXoO3P5QX)kY2>UYHhU_kVcK`{J<sa}bYnc}zok6>kqLILKxOW@
z7Ed|-+_wIT#iAt^4!(F-PW(urU*)V4M=nPRUcVOrzDwu;v=gc2f{mC)=6nA&a5@D3
z%K@Pk51s%6Zb*#;O*acNa6>*80RgSqIwuy);qkFE*h0}@d}<ZtI)1W?XjS4#l8ZY(
z2R-oRozG4@6}D(crpXG)58<H<RNQKqSFfIozJPcSt`$e)r^B@Me0a5s(}nP(cHO&x
zzHa0;dYr)nWl58n-Lqy-&quGQ4XGGPwbLUaAzfc0`F(H2f5j`sU(GkZc$F%e_VJ(}
z=>SdllKr0B6S<S5uPZm5Sw68Ru3a>Np>dkjV$P+Ohq@yyEF5=CekXT4-=Z09i#Px`
zN(WyFG`x0F3b&0?wx-nfj*j&<0zjD2VFuh(8r&1@jex*R0!@laNVAA;jDlg!qN1f-
zF-XwP`;ByClnj?9%}?h;h0z#O_Q=R&oj<IhFpl?v)0N2H>iPT|v$e@fxPrzECEYr|
zE{`jbdSR2&gMMw@otTj^tAok<6Z(>v^WoR!DJjM1dHDs)hRLCqE)L-UkS>cuW;bsB
zz%OGJ@oU>E8mWh|dA~zb*Y*-LG&c54gI7Iy!?yc-0j`icdu|TFF%bz2pdO%cUa&b0
z{m7Q``vSP$$kf)BS`wr}D(g-kLB%}_+Zbz3jS0Xe@~s2c@~eUeg<p5Mbo#IeHhApS
z+d3O^EpQ0BsZB+Ia#~WtQLUm^RLZ?(!=;H@MbaO~j`$s+$g0iSWB(k7vtNOj%bQ1U
znwuGB8@?QtE!AEGS4Snq<h(6VQFHn}>}e!BmZ=&m6CPq<2;KcEc51Ui0c-diX2h(K
zElChEzUj?!=&<(Jx0<o?wVT~+YskpR3cwcO6trN+6NopNEuO7C-pz@OPAa<WCxx~)
zdw~0RI+iXJTTV-{9H6sErXAOJZ%bp=tPBj}K&bu#OH%xu@_yyn>ZgnIz5BQo<9$R{
z5i5O2?}*D3>b0*Fvm8}wiRoSe@I6GHKEI)s$>=oDj>bem?T<9K*A;BbF5O1)jq41q
zZwkQr{1$1ZQP8_I0-kD=2Oi^907VtMg2w{gbGkwBTv~RtWJO0A;#ftoIxH)!?WW@Y
z+w!Gwge0ocDubw6Cx;ocZqnT$TYEHn6e!fH9?_@y7|t3Ytl&9WXxhe%B)fSQ{cP)L
zzG3A$qH)D^t5S@}sB-;;CEIK&oyNFS4;&x-`Yev5{l$dm6vL>IK9Ix@1x5C6b)J3{
z7r*q$rR2sPHm6rHUtxyK`jlviFKMS9KZyb!;19<%u?(>kpp<R1RHWrR8n;~0<h<I<
z7u(@qNvHU-`&J&9jEkq}vZ~Wc9yJ^_DzdRNz!k1X!}{!u%zobnwxKEw`~kPhX>h>-
zhu+WcilH#@M3u2j$%t|l($60g4!1_R?-S=aD*T{*K%uZzWI*ccyPg!tJcOCQD$L1o
zFMeqwXx2ASbal2%N6wp^&Y)R67EC@Jb#^YT9}hK%2|2Z2Flu0hWp~WGxX~#8%P;xf
zX60T-2B2;Odi>ax8StmG{p;T8dZ<r$rk`6!zUpna)dv=jGr)OW-EKImuo={`idM^l
zv0f2KjsRzgn!j0-w`P@SC3P<v736gO@o|wi`Te+a9l>N!@Z6F{=9{@(9+%KcYE$QF
zRU(QuYE$)T)v4InLIhBxh>DHY$uC}?N{P+wKRYaN0F=M+p^B7#+mQPC<Qun{TsTm_
zBXivTWL{8K{cj}!7Uzdbt(N=E>~b;E{0}Hql@(GU65rf7#u4}#G?M%T<hv54Dsuy3
zel(Ic`00Wa1v{nFHhMCDXWA10sfmJ&nNjq{1HW`JFYvdQ{;Y(GN`e%<cSEwWCBuOj
zin@oABg2tTLOcJJFy!MKES`E^H?gTDCvQ83cJ@>eeGC`Hz9|5H#Oo%=!?D5tV=j>l
z7xnfMp!kRLl;v;LV+Gv9Eum-QQ2tO}76m<GT%3)Fekp0DURT8}Rj@~zRFmR1G9pL;
zwHnqZUfNavdw;2}0LZlz(i@ZbyO*s^!0@FA-k(l}g=czb^SPwu=fj~g(mr=?o2|7_
zQ55}gsL_hAKJx7I@K{HtH4roW1JTHP;0+Vv3tkCGj$8#^)wSwV8yKR`$m@tcmzoUm
z&Niq(qMm|3e$-0b+i-;K9*yIMvO}lk$OrpX=7E8vCt#Xy|A!bv96$)2DvrWz=VW8n
z*HMCJ99G+xxWT)9;lfeSYV;ffv(m9Yew&$BC!Fl1q^n^&Jq>su-Pt)_g6trZ_8Upp
zy;EpCgzWFb<?4XQEAsV_8M44>_N8R)X6JkbJ57<+$`y-8M&ne4ibGo)23W+PR3-dX
zRLu*X&ARUTNcD6h9YFsUsLSYEdCHSitG^|%DJzZ25?rwVbTlDV_~h@kpOJ&ltf=kw
zPFIj;Ln>=|3GJ1~USWOJ!k(<}#D|#)7sl*t2O92jLzm0t3*?-O$`ve-Y`d3kNY3o6
zXiC2Hz^1sYwm1!AK!A>yPED=v_??~8rb>Y?{=wwJ74QA1f4)Ekx{oyk4mR5)bX^dN
zw)-3gQ%ALzXDO8%zrIW+4Jd_3_YR`Q`lp}p8|Le1A|$@Ndl!$HVPBuk^A4pyD*DkZ
za&fS?zdl)9q)I86Ek_B<yJFTzBY%_E9x|H{U96@G6sr1#HV&m<_>GmiuwSrU(!BTa
zIZ3ts`;?!lFB_S$M3Bo?T<F>1tDG7~3y#ADdCWz~?5Rhmr4Noe62t^wZm7~MEt)u)
zxqe!muC}9-^{}p(;<os+unuG3V^h38_Y7_<e5Sa4wUaTYj!?BK)GF{zOEc4AP^Ojq
z0F;$EyrohKH#poM&F%GFt*qNtP*?b26^Qfq4tzyGe~__kv`+QK2<fKupJ@vh9xF*-
z&1K^IOTMp06%%DkyLGH}e7y-;$C52hMyRlA{cZ&WbDGP~d795~ushgfHjo0C?l)fW
zg=6G<eYt{3*Y@^Z*9W9=3M>{n0HMOC|163RT9obnJ-o411{M_fFQ#oxS<$tYt;|kY
zm`$|O%ln>#e1=AYtE<go(e=L`me)gI*zzVZbAAs{RPaQimj0n$@9U6yX(TNC{(}TP
zAKM5j>zSNqd-^g4eO8P2cPs^bkX1En{IW$<0G24ri{lktNV#23NOmR&fb!odN7lAi
zj4iET#t@s+-JvQ&?FNgBtFx2BCgUT9eD3EpwtIU^R$K+DA4~pBRSe&3D<1PK)^^u6
zii%83PD{CjWX&(2pnzWb2wV9R6$Y7pP~c6anB~kgCV21bds_Q9H+kZK(UBSsFwIxI
zmv<L<&rog2;5gkO=!Th&uXMGT!gB~V&D(Vf8UsHGQ!qs8mrpwkPH(TFZwD^&lz!Ds
zJNLpiTv;+5NMYby8H_ShtS%OEu@AcAh+fYyD~KK|WNbL?w{N&=3k4_Ywc}ZtYyt*J
zdH)IFvs^!Tqg1Gt9Vgk@jqCd7`R*^ZwLxgPuA!|k=fp}hlAC59);dBgI8!wqa~2WN
zMr5{WOhRmb{b-yl(?&aO4`;~>7qGU$N5IHt2FfR^otpg)DW%5{1VNu{U45xd6#|WF
zYqc5LIlrsge=I2jK>|0ESkvZg4$IYCwzs4t>G>l2=^(Q00GOI}I<9e@pDJ++^$m9W
zghLqdWOW-Jrp>pyr^<v(lZ$+XeKRsVAxBoSlhb&GqJN@Rj79cBmHjEpsMFw;nguEM
zp2J@y-P0DdyQkFN&?2})(}31(&2!AS5AkaBz^37)+I?Twmn{KBx@rU+?PkkWNv@yg
zL>t)8+{gi0$ze5lAyq2wV*dE>^bn_uls$8<O8<bd4>3BKc4QJMwlZMFi66sNYRV1<
zfsK4aqiwK`E<(a;Ah8IXW!UkKw^06E^&h9L5zh}P%8ok8#b%E)s+rewe6P$}9uhsv
zRSkPon=c!bylGcV5nHXWe?+0M?ME>@;zXOP4McPZ<x^aHrNSc*Fxav3GeSKsRX{kU
zvYGmVsdkMtcw;0@y~d3#A`?1QIbNn&tg=v|Agfs+4)@?+*dWq+4`P4Jb3q9gt^`+Z
z%UqgjEhWa8i)w{}_t)SJ?)M;U(qJT~X<#%`*Mnc0NR8az)Ra_ciw1O_wcUNdbv2=Y
znd#})rRM$)1cWwGq(svic^Vp|hcox31_y+siT<jy^<S?wIYwSVu~6XHzmGN|Tbol{
zdmZ^^9|2-l4)l62(c(6oxakyf7D8U|Q>q|&r7<=?<INj03LR;7U+wJgo}O$e{Ph+L
z4O}H~t#FFddB3u&0xQ7EfqN@3k1sN@5IS8wyno@ER|swDr;V6vT*bz>oRr}wpd2O}
zHC8oC`Ww32ppmgy+@Q<mhRe(AA;sj#9LW>6c*i0&HG=l`9;0gXQr^6(sE<m-=-SwH
zd1C~%Y)wf@x)*k@U<YirF4xUoaA|o3y3v{9TKAJ4IuDfUpDxDCD>F3bSF^0EEn8Rr
z0H^EB`%ae`m~$tEpJ0UkwvUuyV1@s6-R~NG$`nQ|=%Et|R?%$Z!a}2RBNojx?7CUa
zB^6WQB1n-yf-E_DY}~L2qHKaE9FezblhWwy&)r)v0{-DkztOz9T1Umr)KuKy;Y?K^
z)^cM)6~UFTd|BDkOZ2zw(BB6|Tj${_Z@&s*)=#AvbiQO!32x})M-NjyWJ%u$@@p1U
zZSLs#sGuTsmnFlC&1cJI7!>3kZ344#cev4AyOE@$*}toiP#Od!zdW7=IhEw81BNsd
z`uFOI<bR*!jBrNKu5Jt{K<}#ifU86(q<5Sdsa|hz&mf))21*m^mFB95f74=)IegB$
z9z2p?-@1~N6RJII$DC$_@%TF@*VPe=MQXtMh`H#o*5zgCl`7rD#iz5(iuhpyPVhFt
zRdZfUcjQ!$CpHD#e=d{;z2K!1`+8~Qqs~YXtaMyMo`VDY)%&0MUaR)M1VPsfm`C(p
z4BPvO;ispM<-ND7{9e3`bO|3B;dE#b-Q=<Vm=~vErd3`1XR2y=r?Fva+qk}ojQ9Bi
z*%OOK4~<F)PDD77d1b!Jf-fn%dn{kIiB56pD={$|NmG|9LM!>Mp7C0JNvW=Z=+Pxb
z`;Xs&+OYy9J$}E6diSH)FKJ8pExDrdx9Ut!PAEcEjwd)Vtx{%G-|yZVzT7Zhnk2_P
z;8t@^Z0%%9_Z3*VcF0sdP~68GSh6)8A6arWK0G>wPCFvBiuF3}O~#con90&V#o&h_
zoBqrnrVO5G%LAu~kgUx7W5)2wLHW`mJ>a1IacI8mf&@|Wiut4+$2+m)Q#d%~kc~^m
z`7#yoGh&xU<<|vR=$Tzp-!eOnvFv)S@Sny_XQ^Qa5E748Z450Q<3nH-i=mB9d1@ar
zY8}QcwE4PhUs2C{2w);v3+f~s)=hdZZ%#Q1v}fp)arueWje?6kED|^lRRI?jNQA3H
zCFp3=N{VAhZ-u%Y7fq#PyX%KjAP4K@ohtBm-1X=2sut7ClGnti8Kj&Qs<G!6GJN4t
zMOiomQ1IfU>?b!Hj%Eqs`3`N`D=n0YT|$fva#7w_!^s(2Iwdck1~Qw^iG#V~$<xm&
zdX6D~9urosfAj*zHzx9~v+h=n!4IKl`<|%n^6Ht(j_K3&BjEtA*=rs$<2wuv`zd+0
z7-mXxs(<w^YjJt4++`nw@$!KjFGKmNwWmizra@;x;uD}!{6gtzxaHrtxYW7~z##ym
zWcb)?M&+CFoS#M^X>8q@VsRP}+no|FT3Ym_3veGZ7mqnz?~xv(rB?(tgi}QdR2ujy
z;4?TG0X2<)aKA1YJjH2ybYTUPyTjA_fc#T*k%&Vp)a3vNpNiz!v`R0R#rR;|k;{{v
z%L*u*daZTBa*qCAtGwH+gI_n56jH9AQ|xS%7(OzJ%+eumux~l3&2>>aiIZs$0fOZ}
zj+IMaj%}Yk$i1SHU)}Gd9RBVB>){xz^2R~h9XwlFTq)4`SaLairZqc-Xuy_I?|0>4
zmxf;=ASvb(h4LOJvH-ZM)6h!(hb|FxdPx634AKOxrZyF?8+WZ5M3_A$(%9hf>&=>+
zc3MNw0A})#+S4ExWy&k#AWJg9{!5;I!o@j9;o!;XPLA%uu~$}n9TYyYw~x$t2Wyys
z$DDFQ(*eLoD_oUaa)>=p&19{mZC%*G*K&18PNNDN9}Q~+*;2jpc0q9hGik2dTwHFH
zU#QvA?+iZpCvV)~{lk1`$&nv50e(Lv>>N_&Tz~h%F)g)Cn9re>M8JA`LU>Hh;ME*G
z$m-P&`QU{4e`$@fZ+=uQ*6Rpw$!^lqjeRL8d6f1npXnqM$_xY-G`@A=V}iBG9`%m7
zI%Gp1Zadal!YsU0EoU7&T*~(L>@U4TU(*TeHjr`-O-);k&Cc>gJ=>DnLgqcj!5<1m
z;i))_c!Ca{2B393kQry|7F0UU8fd4@Xr%dgP7FRhgh`XIy!&s-%@5-?F+o2A)y!GU
zIRB~JcH`*4?y&>aq7@f(J$h3LWB}fRwWh<eR2-?_tkuG0!Urs#6}fQZq3mFTN^)Pd
zmAEXUsRivuG0C%EZtfoKj|Sg7lMbi2gq7}X03o)ayR#v)@l8jbj7H_OLe?ox_X}J3
zxU8~&!(16&GH9|e>8fn)YlkO8Dzw&4PHHd0-*IL;v0lfgGS01f&z2hS#x6TF`J&^q
zLD*#SC-p{IbQ*V|<cOF}sVnTW?&OA}_SHy1Pd|$bRf(vY5W5@+FYh`8Uc9WsuM9YN
zd-o_W`1B{33~;LT=pBwa^{TG=5dqO_^*vSQqed4cGupi?YIx)L<nr^DX$rZ>bU2(e
zy@e^Ed&xF$yDc>7@Yo2mo~I)Y-bc)}N&_d@(9G^%kqPfO3>XEc09=-%==||J9|shY
zO<EAxO%S4LqL4aHE<KW~>j9{J9o1O2Qv=x%KM+qR>3T5r2p1o(u#l}FQ*G;c9;Btk
z)lH>d=*mqRF#xl|@WHBV0@r>0&<IV-{6y@(h(F59llX1y{CFWU9iI%zM=!dqGt=~w
zC@qKmG_kn0Q04x$h?B`H9PK(Y?{;Lmmfc@=e6p5M;2U{H1<bE-1&!~++$6)d+^iIN
z3ksiKdjhlM(ND+m5-!#yK9o7zT6r%H1O00S4Ftl~u46+nG^x@a5%S#0%8|1X*tSKT
z7{Gb7hbl@(%F8j`@ckZGW7us-5rpSAXB-$=F!Td$p$i=Hfj+)XUNB?pT2AE$HYR5!
zV%IUR0#h0+A@xg)Dq0Q|W)mX1!7v#Q+K8=_B?7^6XCK?m!_njjLxv%-ed0BLpI!ry
z&~}u|=Y84CEm!{Q0?lL~jqr{WSfP^~v4agTRI0SD#Zq6V7`l#la9O{~9~*g(y}UY_
zchz!cB#(5`Alv9U6?{=Kdq&P88K##2{b=bJ2ThLg>!EpF`IPx`QGM|{cb-^WV%96x
zvM9JuIuYTSD`!b+*->>@huiQ^vX7_fC7POqbiS$vKBeTzd_d`uc6Ht;-b+CKsj+Rg
zs)c6lqiVF!Y>fu2F|oJ~xYRy*;CH7pDdPm*xG~NDODyIvX!M9hpr9o6h3qjDDB^=1
zWX+v3=uC&^^w|7yoBeHTsx)E~d}UWT1v^FvsUKCFp>Rjx3%ZIaa^x^~f@{aeT56|-
zdVXD8T!FTn;^S6A?BAoKi70Ylo39;0i*t^@iC;80rPNlKf7*y8N~o3!2SHYG>;A3N
z1kQAAsH|4G&UW9W&b-it0pQcQH>OYN@k)|(EIn`C=N4Gr$@nUe5%*)>FtLLonb<Pm
zv+*V_4$^2ExPA*~?a$MGdI5PY?^82!qBN#60+YevTb1SywSSie<yx(TOkK3+kaBI^
zE}E+ni&)?`u|2;dxs@uI)2#F4P)gBS#8c;)OJE0HQ;!;r8VNWRz^LwaQwE*zrLx_n
zR$dM<3u-hlrq`V9jlf$GQ<oFl<O9DX^y$$kgHwodxhuw9jpO?9Q(Hl+-<<T=`2ms@
ze6!&~(E=y=@g`IKb~<V^P|t10*2x*o>q~s@17YH$n<^j50Y*FGYm6GWwkFZTJjoCQ
zl3e}BYJ4ICuvz!JRUnr1+CoxpI7PFXNOQNwblZbBQBHd(GpsMm+*M8WB7tjsAd;hK
zzM-z;c76CY9cP$>NuwP%$9r|<N5g{w1}ZwuJeL!Y|KpaZkL7>4sV$t^L{Voc?xQNL
zcFoa7Hf#p3h35DMbHs9yc8<5dzf^ahV#9afE@xmdyk*56g@m~k=<f$yz5I((f4PCF
z?dx-153TF!J+@Ga+sP9j4%w(`qgyt8MoKSCAlB3BqF63B|Dm!h%g|7IZ%)6OhW9;S
zF{{J)OE5nC57K2u|LGMek{U!=9Q%_w?%KXD+p7XVW${@9&z`Do%-?Z<Txv!q?15J~
zy((sd(`9dy?rLi1A3yWK{qXut6Ndf71x6u1nIt!OhsQ=((D^tp(J}~Vl+cOIv6dm|
zu;xoj7IrjD>LnwvB$-H(;~gA<<y#Jnqy9FTtq)7Qh;zj|6-PN>yM+DP2r11&%AwLR
zWeujMpky)TvN?_J0rm%xL;G&s>$+7|?RP$H&2ADHyEc+mSqHVU$|un&f)Wi|`I0?C
z0Mf3>BX3}GqFOPaQ>At7zF@WbRNPN{-5)DGKa@>aEq>|kFt`*G?Na~!$HOcGgY_T8
zly!E~)l80>_6C{MYYGzCsAW@yx?@x>ha=d6lDSYF1)j1D&t=@BFPUBnUo>WW!}0p#
zii1f`j%B?ZjL)?(PW601JHFH)@7icLI|-xWqq4paJ>)D>lP7ixMS?)(392f%bH8w=
zR+dHnLY1M)T*b&TBvCF#CLDZ?16l6@3tmNixN11@WdkUY)S!8?HKb#p9_M`IA7vC>
z9f2wkoCiWUv_|2fwUq(84sqB1jYY6y`7HMqNNII09Bb#|Pm*l+_Gu7mU&ik}Q2g4K
z*Kn`1g8O%Dqejg~CNgn!6u*R-c$1Lv^J<HT4%`Yz<)5KXf5AH3nR4#^8G%%IZOp^1
z<CBv&j6CXK@q;8-%h*sH0NT!aK(ZyP;q(*B_UNM8-6)j(>o<s1zJf{lPw!f!27A?t
z`?DSD9H9q4yHB@`b=SRF2~5;e_6K<^;SV+|S4_N~ZVKm~4X!~-`KjK3wOTVYr;9pI
z?d%WPkb~u<7N|in2|24jh$(1=IcuM;lm;^CV;vn9ZXawI>(Zl)7;2e;J~Ne)cHM4X
zL{C@ua6RX4d^N335`+Wgv~XGQbQsgCJzY%o0dwq^Vgf2XfyL$V{nuF7%rY4a<$_?h
z!}b2S@@r+Ofpwj^5~_GP*9kHpjtvBH<(H%6bYve`O~43e$Z0N?NstA}^!Tw=_S&XK
zIc7e=a_lh*&{f)U$<JWId9YeaAT(?R9e=7#yt;i!3Sw-ti{ja9D5;acC8mbi%0g{7
zmP;JETysCqbx%*?oieIUC40<1iz()i94x7(%I3!4?N$z0jEO`)P7ujjnLapcycu8i
zCE{|zNdeV#p1(D3??XFd7jrIO?g9a`Av3T(2X2I1JVL<lgswd#(j-Ulp45QYKym*R
zBQ^UuPd?|tsSCL77h$UDb8~LyYH&{5_k)84{`0HBpR<B5L+;%D@oQ^KC$(C1Ryr!(
zAtTZ??LDr=*FRei-bILD-}<vv91(WZMqMvL8(~10X?T^q;<2Wo8L}i)IB9HwNR$Oo
z88@6Q&TIG<Vk{;*5_MHmmGi7Q*q4^%z6Kp0r6H1B-<RTkWPEL)Ci8q=_bwe*lV>+2
ztd(J*V+T`SUU_)YzLBHuw0alk&tpjz=9bABnH)O^G0APuU>{yr(ygdq1aHIC9czwr
zHaGo0B)?a*sj`Izva%mXsELrV$k!(V6?9*sq-ZjhS@h;z9N1X8+WlAfDh$7e5<keM
z#d@tw-^x}tK}gguAH)3k5wScvCwVF5#ExW`re<z`%IWv3QQ?zBuW&`J{(cwUT)~WU
z*iL0>qeJMQz55d+<eg5$trK=z*mT9>JDvPitofQTuAz22iqG_yS<=(X2p_{>Kr(tF
zWS8lDb#=p;!8BP(wWD$3mdAsJR>YY7a<Nwb*CB;N;PlbR2j{lRqMU)9V`5z;MV*kG
zC>j<Av%!zTR`rENma`eFGwxOMK}%<WU(5v^d^5hpoSb<(kxZ1fc_%l%xITyd`g=$u
zbgRkp_HT+V24<sX&23wu^*cZd^MF>IV1%Lf>Bwl?rm8J*b5GZgdrapOoGy&hv>_ei
z6S3RXo4*iQ^GJ=1qxRtpEOmFbZ?=f^7CYq|cA?3tN1R3nMe(IhXB%sDVIQf_DJZ)8
zXU^}0-rTwHu^y3@vM#O6vXkEIUe$HnTJBGY$xzO7F4pHS7Gw{su5|>`->cXXj}@c}
z7BRY{>ASTiy9V8EsP)2aoUIS$1~iH*0Xj11^P-~M9}LGkW4uWNbk&84EQ2~&pXY+s
zDmxl@jHt2h%O$2M`aw#iq5=?|V@Io5tQqS;lzVdmhe=J}!;&c&d<hnR_I3B&HF01-
z+)t$o8|t163q7@8IV}f5*=JCJQiw#G`%mrI*vfSF`JRw=thxlge*I<%GeGX|ZbS=r
z+WM}0eV0dFol^m7Uf77e1DUxkQgIs%-UVXV&n>$ZH0Qtx8TT_*Cn2J!D7-z(4vaD7
z%4nnDT}PW*{r43~{`W?wD;6h$BnwG-Z-HhH0TsFJM5lF5U;r3CWPd4hy7O47r`M&)
z?QKj-a!c>Kug+>TPVZo$=N3gS<<cKs84vzhN)_3YmU?#5A{yeW>+xgs-?8=J$@LHI
zUl7)N4T&p*+kOh41th%5!`t0mY_8aK*3cNAwEP}QM`tYCG29oaaz?PXKQIefz;VPg
z5b+by+rtH%ZRUU9v0AP?n4vlB=vY`=?b^>UgMXQxDkp(WRlhG*X0w>K3^0tsCtWd6
z#0`|o7yf;<izKe>(D4skS)%lixa~ewjU$b7*wWW50wog@6r4jp?W4@_!!9x3=W<Ns
zEvW3ze!u?%tI6j>$>mhAYs&px==8uuW6b#;t_&W4@8Ytt3eoRWMXH2!_53bsqa)>b
z(A=3bV8iz2wqRZ)oiE-9Y29(CFfZZL5dTIYoSaZ&36TOmZbQ4boL?4>!vuzl@j=6<
zZ4YFWR479ee)chDdmND9W`%FxIa&a1@HQ`Fxu~>}BGAog0rdsk=~%h%_&{23WTc|%
zO$^Ze?0Ed_b)-~wwP|<sy$owMcI1=<wv5GWVq2)}&zT|5(J+*=Q@d1=#bux+b<SJ#
zJ{&xXyYM+{?1^BVUglCD7M9XYSmVZgAk9rTDz(?K=f88Pjg6reRR0aT7Xn)1UTP2H
zy9R#PZ3qH4bjcXTgGi@RPiut6RRQzo+}L^It@TmRE{0+l)mwk@)}HFQ=JxSuhcJ|j
z<KR2`Cl2sMl^eR>w0>Ra=e%Wm`}v~&=q(q=*p$&(`fc@ugvVh~Z6z4Zk1Wi2sFRK0
zVB?wWq^>A%GyKi|+~O(WeqdviMS(u556`3>n#4a24KjPIdAcu+5Csd#3*hf!f_I_J
zxt^2r?ggA&LQEH`qUU+qIO?z^vxbh>k=?QNhQxjaW`E*INffnIb)e@dA3|P6^j^oh
z`x+aZTtF2bRS#O=J)P+B@d0lp;XQEfSuW?)q3)(Gw^J?~Zg?)W;c(?f`X-CFmbJjv
zwzi}YG95~~EJ#vez$)q#;ohjvZ-jGa%<Z|ShK1&?>q^Lz=xDa1BP2vWMf;STW@+hG
z9nbFQ(B&$3>X;}F9gb@7bL{4keoeFt<O)?^r$^InxKxj<d^|ceHiM;*K+IQV?D!cA
z@%}xdR>L7hzruBp{Qa&W{$~n5^d<JM@<zc@fNdn12bmdIz+OI8^{OfJ+jf;DiWF^T
z+#5nb{L5UwEuI0jTDxxLI1}_fEv?=gmnR{9;71cT`_(Z?5I)|?{)N+Bvn}0~M48~=
zyBdf1MjaiA^1Z2}J;EoKdOTTCw!yo*w|xZ4S6(b!PQDA1kr?#kDs<bnDANBX(P_>3
z5_n`Mc2fwA${0ZA#9u!`vTC#e@sqOU=9j^&^ict!<N4d4r1{AfrDz7Xcz<Rx)qU^k
z8J>{HR@gCKbnx=HikM~H{Ybrj7OAA%w>RqsO&VmFzq-}DV4?X1s@FFd!#t|r?|Rry
zn_9R;k@Qh2ZN3&bKIn7l!^gLz`6QeNNc}R~>tN~4i|hqoWNY~u?Jz%BNAfX?kz|O>
z(;{b3!DD*aygE=$c44sXLn%A!mrvW4_!Kz5yeX1w*w~j+7U&T1vAL`g#maj}Vft8%
zBolSlPZD&nZ<HHKZ0q9{Q?S*R@k<Ik<#_PhFEh)5vsr%hSm&R)*-|bpz5cI&*iT~v
zj$ORl|1r?RZ*K3+%V_V)zIe$VA<dn!TUw=;*%Uf(QU7%2t4a*KY-!PS8Jf3?`*3wd
zy6b5E?ZPMvFUg~BaAHc>S<_N-_$i{FKKq!J`Gm;s)f=Cmt*@}+>ky?|7u9RORy4PH
zPY{S7C6=6mx^59@$?iU0z&;VtDK`v{!)fGFRVO74*MM|n9pH(wmtp;oqON~yc5A-z
zSS#s*WwPS7XWWe<aF9GI2s@RaVDk1vh*N0dB1+pWmeWKeI2I}?DNd6!G%%cOca_i~
zefD5q)0vvzvZHLlh>>Z-e06=@t32+Fp)K#d+wY&teN4}#Uo*`FaytFvBJvbouVOmy
zR3-BDqaSySKg*G!%}C9>CFuW6{`BzItYu5zOk>QxDM`?YF@K0r_5pM`q%Mn>kUH}T
zv7G>`Pvk31{wCL<<2kGAYw$4&gKOmcWh6u;Xp`kaI6I~^R6sWNOC+h7Oe8hp=7oSD
zPxz0YR&Iy8>Avwm)3jX5yDY_BgR^n<@zV7=E~7hl`W~3~e#jXp!;FE;>Y=Js%b_>@
z6kbQ!+u04f=nE?}eQ?uXnAEIXo5}dxE9ua`Y|M9SOoK0ji>pK89Bu)PP9`usTPv?;
zD?VhS{txzeaksTijN18j+P&xZl&oYp`cTq2M7c-^FfsYZ!lFH@#<>g^*Se%^gml}|
zYA_5gq;K5C8*Mq))BXM9CBmazLJLP&H<^Sr=n9|W`_~8Elu$eNSE}cIzWQ{|ZYNfm
z1}i1pT#XWmjmZ?+kgCu9{B|6))SSE?C%9N`um&;#_;m0NW_gy14)I*gP6x}6*=Lq@
z;7IHgV6XmQ>{~Q1r7SmBbT!+rARn4DjZ<Qf>Y7!cnAO@c_KjT){s}}}aizQDyS-di
zH&YYp7^*rtCP>Nb+?NfwKD^&=Yqu6v9N`$h!gsX!6ZgO(ZZ3PNDEQ|c7jAuY@G~Ku
zBkrcg-5&B?sjTpzdI6aO%-VEc#tpUn?B9(#z;nrue^MUPand>y`1z`}b8b=KA7TDv
zC>itGzX>|Zk<GJlzZhPSWs@TyP-Dcy?|#E5P8*}2tKYgmzo;;gG+13MaqLol;QrcA
zwCy^#{U_7?*8S6T1=o33Ll9mz#-OG-F~yNjd2eWDJig`}Td}MREwkfQOS_#5SZY5*
z7TpI^<+t~p;qeLHT4_I#<I!9?$u*PHoMt*m+M~7hMWKc+7UnM(!}>EHC0|yTd^ZXS
zxmCGiEWYwN6F)Z&S3x1Y7(LxuY=au;?7+<v7ey3`F6=HcVmdxm&r=cG*|jFe-)KH9
z__Pr${X(eeA-mYk^THw*8lH!g0GfaMtMaRs`ElREMP)@h+g&HSh6QI68JS=k>gI1x
zJ!DLNe~{B4`5iZzNfdKU4p_OVn&%!E+m(~rS?VA=gCiqJI|<p!6>fl;6F>i`P!hwl
zhTwcWGh{#LgyDHpY5pfFK7%=Gc*3|vguWdFa65gB>EYp~=uKqq<t2b0&W=P?xICVv
zDjkmhFH~-A^82`hEBg3#?v#69&qVGpY*HDMW|2`jH33l{Y7#q>7))axUp5lpw{?y<
zpou6zpFl`Bx2D!4sntnH!qtj-TQrkC7H?9>n$CK?DMDAXs$c~3xAq-*eSEOg$)L=J
zpN;iIfx*)vpr*3e<7q&E+4BN9b+qF4i6wGHX}ZG(xnermRoaJ#KNKScg;fRi_F$D*
zE|z0m0h@&E<rMz6tUM&|b)32e%N?reR(rWx=Dz-4yy5(+g+V|WgoCK?a2`^}xWz(q
zTq!B*lB(9ucAu7a<Th2NGwT&@?TuzxfQ6%}MzB5!O%`L_!=9`I#xf~s)*n(ftlavs
zaPH7*Xz*`s1|4?kF&Ga-bN<X9Et02PwrPJCAQzOdRecK9T^@Cf;gL4)n9ktztPgEH
zx1W(cdsPq_p$$=`(uh-3?rS*~|FeI;o1UR<eBv2_^!-zp+67~~6-?NhCyP;PK&o0V
zuAUo%w8*|7eeMsN*Fs@vUKfMN$j|BU&_D#q?5y_c;>I^+TjSrs>u$=kE1O!IV+C2|
z96_TBNEnZ)^VY?T7FHKC^qeP7WkKCk{XO3L3O^0MUt*bUVxNoy1zl4;HbPMYU2b7v
zDW_-jR72KUP9rM|tHRpJN}(<Q=#dKX(S6B5>$}*;NKJUp^1-3~!G4BQ9aaz?9;slp
zTT4q%`05xTh=60s)`dFPjpatmnf`aG6Tz5se*Id7ew<RUqZ56(?5b#6WOKn^vC`IQ
zJd&l-d&W~&d!lS{mnb>T0o{}&C%3dVi{szVlKY4<KCFEUh;`Nxw2r*gp(v!_y~Y1n
zKa!qEt*k6eirSfnZ>IVL<tM|u(rzeRa`@{DJy8H@XQ*~#D@&cDzLjA&_n>fZg$HaM
z32%s^@yW{iW^D8unjsfSDKGQoclSd%x}5Q;;v#QB%7cwQ6F`@`If@apa<N=#KDNB+
zekN#kTSP_&xDhDlTS7x0FO{XJ&u?tUu0ycZTuwgqr4a5XllIG1j#=W)otlZM@$cJ6
zW={vd5NqD@-IN+XH46x;1-{$?W3H<DNJB`78_)&MH5toCt*=yRFbyV#mb3q6Ed}q0
zbz)j+#fDiecwuQHbGN3toIOT=p(Zh`9gJ<hP$ht|fSM1E<uC3P8n|4#6|Jq%Weiw-
zZclEgEH<cm%lhSTC#{~{fz2`cDc|6>&W@uEEgxraZ=8m`lT%UMJEr^;#QM?u|7N{+
z&d2_@;5*HRi%aY$8S*?tHbkCSIQ+h6uA?*Kq|#JWWzxNO`@0N`?VtOJwedbD9RT(-
zBuW+$lYAD>99i&IAAVPU^~Fg>fX&Ct`0%;j1@g7BCnD`en|uB0)>c3{f&~1<H!sl#
z9dx#AZ=$?y{GBg%$Uu78I@;q^IKt+7r=n#<`6wz}cJn-#49^`6lj=eG!$s(Dd{3Nz
z8$*MA(q6{VWNGEseu%s6(|5ih4S+m1YGK}8W95!o8$ZteMHRVu6G@x3u9eP=Q!GdU
zZkQOEwtN((oZ_4M0_Kc&?{A~Z>?X+{4@MQ5XPZy<zt}b^B;kzQ<Q}WEs;d~$GWzn-
znhpzut4sTr4ad8BJlm9lK3aJ{r@Fp>Y1FZ|ap;(w31UMe9)nNulg1XH<&AO*0Z~i|
z1IR?e7t>z_7_j*|JQ%5o-}ira)8#T@`)+U7tF+XI^5klw%Ta39uio6gw{v-5N~&L<
znD|{*-dNsV0L0z!`k%APx$t|<5n^|sK_rCWzRWpt<(m@b8HDxI1sx=$&og(AN89{L
zHV;TZslxEfwk4mHdJQ1Vd?QZutDm=6gPG_ypIeAV3h@4&<QoZBSG*%kx;(qn3}fF7
zjMQF+fA{kMdA;szrt?N8Vd%VD!T7%OL*H8tf12ABs%C>Ml)ap$Wif6X73;7^|AqSx
zDLe>1ILbI1v}YL)0R#NhhK7%SeeR68a2uA_hmCN#vjZgr&|<9x?}eNb(@y5}@QBIX
z->rUsB24+E@85A)?`phHY+9$sbnmv$s&rQ0+K4}j-)T&p4zpSOPJC%`cWyXhBgB8j
zKk3Xeiad6t`shN)2eQfMmb`mkSUgR61k@Q=&W`dqzgF0%dd;^{#Q}fth%IM+x^-OD
zAlA8MDuFRt54#a(kBP>JV3egTCx1+D+$2vz!meowg^cCwb3w^|j`>cQTM0L9a&>6p
zFA=R*v#3J4y7h>2-UwQg=bf{2$rLw=c!_MHhRLdIcwXKEq3PWAWE3b<aC(ei>QsMz
zRc7nF(sl1aJKeC`<{beh<h|*st+3=|isD1ZOu%4e$EEAG$Mcb<aXXYzHzH7(K1w|y
z?u&}F#y1UboD%DPzD5DJv~j3P46Jnbo1wuMlhncE<XO&+psjHql*f5o<s+0c-0kg_
zivOT5NJ76XEq%npU`VDL5;|gECh_U%nRHbqYJoQY64Yen-T5hB+1+Cq8%qgY8gL#1
zXuCPD)lq_^Twt;hf8Wi1xaRk4K@0guCT62y|MJ2{+o<9Fr@NyCEW2YRFY%KX8s0V<
z6U=FshR?TB#g1k@evlonRtAkwQ0USP`xHw`!deE6aeFKMUpjH?w~on#<B`|Uw}76W
zXLX-hB{e2NNYS}V5&@Om%3lMdv#tj=Hj+)NY+wSDg3Fglf?VbNGamF5lZX4ARJm|<
zGO|+R9`&b(>-i3k3!gf2na95kc;8T$hMm5@XH!<S+?^J$rZ{J1vqmDLb<PTsUy>U|
zJm(YQMFltUF$dRTmt23MS4uD^iOb1>^QLT(zB3Qc5O)bVZd3pOn%myik=kJVNp_{_
zCupl8Hr_tw&sgCbJk0JZ<|fZq8Z;io%m)O8um}VJWy<a5-`}X8%gFNZzy!9P2*x4h
z9L+5u<PjY^merrZBBj+okQNs~rF1r)EHG1MLP7W$xm566Hn(B%qXl!e@!saiTk2*j
zI~_RZ)`h9kTW)~$x0xfeuCJEK4i92!sorLX2a6??L>q0U_5ljjA>sGM_3<J;B9|C$
zc9P3e+}-<{5vkha;xTr2W@D41Lo!QoE?(e~C$FEJNnwmujMbnYVRD>LUMVZ35UVtF
zqeVKc{(LqRBxGQc`bOz@fQp?dS_Ww#OWU0i+n=8g0BS-V^~0MVT3g8kI^ELZGH#ap
z5oD>Wb~BuX4h=O0*k*+jAChM)<}G6GEl|`FKKUO$A-Z{8dOtrc()eKDjtvu>)!;O{
zn|%kPsjy{g*c~FN4@x<Pm`m>sJYJ_zV~B7R?rlO5d3wXPiAM)})&*+(_jcq^FMwtH
zwQsnu8JXHe<hOO?D(56TN<2Y15QOIEeCAjZn8cW-B2IqVU=Z<tXRz6IFE<SwcXr};
z6M<P(%aayyc02M*i-}#5@&~2n^A&J86f!cwBX6vWA1vRay62duoAe1(cUpC{U8=-1
zl|?*0v7SPT-`hDk{rISmuPieQR3%8S#bHQsu1jqM*S2`Jf~UBsg5A~B3Qg3=z%<#t
zk{$88fVft{5zh$`k<4V-nQHK=jUWd+zvysaOp?RNNnh6-8&1v>P#+!N?xd8fT<>sv
z3#6TYHaC9GbY`0cU;Qf0tKpxBP*(r>t)IV}38h7#7J)%ozCD^SrB-b|Vi62&jNZco
zRThEWB|;E?2tQcsTvY8x3l3CpF{yLauz)U2*Ldz*hO5yqJtuwNNJxVH15VY<_umav
zbf3oAAF-&>ySQ9^y3R29aX5CgP@S02$Qw@qn=?5X`~wY6y|;HTxe+e>-1qGoDQ5^g
z5e#Z5{xBOB!4o8Rt|!Xt=GttbNB8VB)B0>MNMj?r4f=E&s@CWxM9x@rUeUiHk){$;
z#iRUI@ipbjO;fYfz|%Df-O7np7!QBQ$jq_=r}}V1UC8`}W2JexDj6uRs)j!IYE%Y#
ze8`4pKFZ~B9!^>{-u%_D)q;`Rn%nBVY}l!)N-wF1gmYUKb6Osvq{002Y<(?ESLa=m
z!_|)TY7~Uq_9liNB}@e$MTZ7wK8;CbiAq<@i^ZI%RaV^#^<O$b7N|cXc@_0Bw&CoB
z)pG$kkM};~`#cfh65GODwqTJ&<pTz%^?b>^4D9SxjPZV8C!)9Q)wBIUN4-5W%NNdd
zU<K~5nra~+ZYu>n6`g)E0Wr}fp^uW{y}GCdGnow+XHqhh%lcK`p&PEk>xkX0un2)h
zeM+gQ0OJqApT|>gK&IWBEO1N_E<qMsy+NL;5PeRUbZs(;Q<sRur2bHV-;t~4qTw?L
z6{CIjOF%HScA`LQG5KEaOV~IUx7|Jo!iJ22<%6EFY0Gl{;r~R|KlVfLQ};v2tB79E
zaqC>+<Mq=Y0ZKkzep}9$VfR0A$2<GtM`$Yd^)EgF32ofy`|9AB18Tx$j54Z0t8!aJ
zNlu;)poE{1d8&EweTM24Dp4vyxO04x@IgP@;B(T^?{m_u0`=0>m9;b_z30uSjsTJy
zOrRayYqnT|S?`g=fyMfZh566kBy0!LPLP))X1#A5o30c?795jQQqY%ck4-G0Vm_iR
z1vnvu$I}oYK8NVh?mX(JPs2c0pB%FX8t)=Rc6yqeZBaKsR09GD-m7n}`jtvyVU@I<
zYMa7$n`gF?Ci!fhs7m;rJHJU(Z7N(2_U@ddj1jTd5n5}f0n1(KD@LzJWK|<pv%)i7
z{raGozLF9v$1Z(8<eN{#k33kG&3z{)ITlS(X#%I~0JyBKyfzXUe4ZL$<M#^+S1dMK
zDSJa7Pv^Lqv$-ccL7d>(l)#YFs4YqcKeZTI{%rAnEPsh-dg5V%ZgXlA7yIHi(>bZD
z&rlHKNPzVBf!-Cc=G#3W&skzA&~O!c0t&=4Yf2}VSu2%?aK$`zgp)AVeOHGfR3b;7
zWu5E%yQDJ{vrQO`2Lo)Pj%#?z8Tdjxmw=tc;+y6n0m~3MW{~WO8?At1Jid|Ty!0<C
zC~y1^8R(!<BB)<pTGm<((f%*Ld8QY^`TJ|GoTse0#@Wh>Lm1X$$f+HxU_{c(mDL$=
zSEK&o-yRcHH;LX(Og|=LA`3tog1TS!ojV4LOF)>emRDs+_jb1Rt`P@T1gT9ZC=Nf?
zJl@}3vpcWQC(ht7pQ*Tvt-Z6cs9GDEVL6`&Dtxx9im@FF(%e?JKYxQ)kH^<JkSuKn
zem!^#x<uEaEgok3oZS_?UKSA%*xwjwZ};z&AN)MrUG7d|`!%*68+3z^P&rd(hF1U>
zoaE1%tG!RLuNyvrYjNg+FyCE4x3TZ@4yv<*Kcg>eG-B;`yF0I2S|E%$*;V8b))~>F
z9Pe*U1m*p=RH;O+g!l9l&dN4KpVf&D?6$6kK?djN`=fqrN0-^}t2Ry@6QJR<pD#H3
zg-c>d$o2=mmDaP`2t;Sca|Zwg!GT3gyFl@B)Hbkq)PTbi9otv>a|%sz#>2mVHVq#w
zmrdY1Ym5t*vNt2W@xZa7tlVO#-V!z5k_^HOEm;|AE+%qHa#f%C`$d8YHkW=@C$0-d
z*Q!`b80v(5ueL*7T1jbHCb_v_q2ser;WPp7{|2!)4M14mdeesK<s+~u!^gwRk)L8|
z$-I6^7!j`r;j6P2O9Ib~8n3w(3id|GlgN4FB1O9}XgvJ&u_vR@w^@pk2pQkt(ntN9
z=A=k9RKDAjr11P*XIs@navQ$Tg?!5t&CHXCgPmCQUsow4+8QJzBONUvB|(hEKifyU
z=eI)^z!R-39qjDVYIEDYlC_Uhjfp*K9(D|-#Yi&Ig90|ZhZ7VUKoNP@vB;&`&DavY
zF^%l^0nL<HL#RR3qRAZbC?Pt;FYC6YN~PZ{Q{Rchgvir_jh?MoXF2Mu9}q3kQW@Wf
zxHleL;~-baj`3sgtwuiFT3>E0yZ=Q?J|CF(wO*#}aNn6w)*<1JhY>T|{?dC5x{(3#
zQ!wYfW5>a{hWI~H<$_PcM0M$xI@NUk6c`)rkA`TU!w#NxPfk9GiPBGn`$grE%tERc
zD|Z(_pTFd^FSyd-IE=fAu_KO(=)`1Z`q8%BUk@h@yyA}FrVTOPKDqzA=7|&l$_=2f
zHlYdmv$XN^mmE3C^RZ8n2LD4=k7FL+)A`+4kXuL%uK}fzz&j|1hJ^AsSC!AMIyy=+
zvc_wO6D+=%gqkF2B=4+@oXh8Xb2UMDSc(08CNt#N#&(#KCy|jR0y?THEaTmUUl;_o
zjm%O*v(jGru=SJa0ER?Ql>2jbkGdxN@VZ{nv*t;7aSKr?E%3NDGrap+PpEvg_i)|j
zMIQY<kGi1qbACO`6{9Z$8_ihP%^2wOn0!R`3fG2U83(7K;<k>=`fS0wP12sRiSOKW
z!7Kfe3kjJ`>rKS*#oA(PuA5uLpd_1ph;AG6fKePMgA@$-N5j6kiBmiKoDGQRQiA<F
zTSWbMteyr<*d#9~=>0O*m$l<3!{apnsamXn_1WAccU0?%6iwXi6Y3meJuCg-ga%l!
z{1&>cW<BBx7a^MCGE;n-*yo_7qNb%P2eD+f^}#0A)GX|i3FCMhW~m)@oS!VMAE~DO
z*c$>A7w8~i78{jHX~|X=D`p0t6*yC-QtD3W)pE~DhJHkHv9p<#+0f^Au(IZ#xRa4R
zn|n0+3~9${DZUnP28!ZgO6uYgIuzL*9iRfTday;NCU}3USUXKd0`ln|p+BF7l}0ZQ
zsL0u$kx~Co%Dfv)1I8VUF6tv-T~+y#liyEIW2wkBmlteNRb0LOKp5W50<XykzJ-xO
z<P~X_)Hrb;6oh+$3=^w{D>WorF-xzvd-NQ(_zVO;>6!KC<E-GcCEu`kHeY}vk4N$?
z@6*NlTJ7ZYjfc9NDIK{&Ib6w@=^wb)4HI3*4WkcTlcY=Z0?#){fZ;55c>*%g)D)|R
z)AI5P@mFEWQG^OJyS#vyN=j&vk59+KGA(VSs`3v71!pK{CNo=r^p7`BstFNJcBPdx
zjI>q$TVb_$;+`>T!TOh5>%o8G^%9TsZQK)~9;e%%Eqp+=-T(Ic*DCoEMZRz$N;Q7|
zq_dis2Y$ZgVE@VyNa5Gbxb@-x)%Mm=Rjyt8=)%GP*%G24pn}pMQqm$K(%mi6-OYmq
zN=kQkgLI<^C@GC}cXxB<((m2h`Q!ZY{?0gOkIi5WeAW~9J?}ZMYtDH;mxcE%P{j~^
zq3c;Yw<}e592@7x9Q_lmHkEFWer=zB0=*KePiA#Y8=#v(53{6UtnaW7>7OzG{GV*>
z+q*0&k)ro6PVml;R<P2HOFg@f0>JyCa7__1R~}GDHu3SjVrTivJq&Y&gyd0AKY4N4
z`v-Ff!^HjyV|LcyBLgJ6!gY@=XNh(7Af4F8Z9_iQWt&=6BId=tyyWVn^XIF$p;G^?
z>zBJuSl;>~4qx}?r{CFpPE{Hw3jIPn|AhUjqqR&cn#5($*Gr-d!R;B&Yv1>37D4k+
z1xfCDLgLd2l~=fJboBERt0HlFt1P9(_oxxts5tx8hw}}J5eL5&r}2$)rJdO!)k%YY
zRf57WH%3+SC@WVu57jnv?%atB>d8N013xT#R*bXWaegtLdw83YlQ9mm^^k~Mx3hy$
zSMxV}>8|sMx>SL2=~;%%h*NCpPTqlIwcPy#<E+qWMLCOm^a691<7aEq@T^3=h@>nO
z?sV2Gt#2Q%s`Zg_ifId9F<oAnUt8cORMfxICdNT#`_3_|Z~SaumTshqAVfr7zm}i=
zxU;`|O6nOd(9`=7K4|(?zqYi$`s|kF!Nq+g)y|fBUky?-_;dph3%J?~rzN3bdt?bp
z3P1;c0zU=%ztB@<i9}tv`fogXoBRE$#>AUjBh}Z4{o}^_^Dp!AxW+=bKsB=N&4i3!
zx3RN*u~tIx32LUhO%K+8-?m-uuDotR!x3W2{tAlCKJ&P0bKy(ZNhTu9L#3sag@yEm
zMi*{=zso2ohJAX*DwQq=Z%;2+cDXfV*0?^9rfGp}G>Y>^oh0`W1Wq~I3293DLu^qK
z=}1m0Op@rI*=q?v6>XY$$O#C04e8F#F>7}=MVC8a1Al<#j_=teIqQW5HRKG{W$&^3
z%{K*qTs-Bp3jS=|Kd?J-4(ej5#b;k*IKQm^x}or1UE4j(Q(`U%`gdm24Nw|!k0f+j
z|Kt|WzCBRTEN@m-fA^@*eBVf$ARn|p+<&?{8T&ZCn%B}eJ*9+TX2C-l7Rs4hc1*A4
z>d)Epg=nz73zt%rxXXHL|Hugc%viCOjBK@p>tQSaN;)xd!-a%{L;uy$kf0j@bU5De
zg;{3AE-5z&3%MSW;t2>`vPnl=#}Vh1mr759_>P0CCSSQ^=Qu!j3Hga@evPSo_0`KA
zKPJO|(V`P99JSfw%f(b~aa$FQUvBoAw6-mbFu2zzOrT*Xf%Xjmi(0y`e-Nv4GF97N
z<EC5VA{{sT&wB{eu%nffz7Kv}qT{juG^2e3=3uMV&Xg+s^C!f&`QJp`drXn+aRSnh
zfWF3OroShYJ%`4sd?*3WO*0*{+C3Q=R$G0!`qt@uPC=V12aoYpvOVd%&=Vf+!00cJ
zQ+Jl$RaRH`Z9qFo%UO-uvlO*<{SzlEtJ&9$`$wYz!xw)Ay@Zw5d}vw!o7a#<;l&HO
z-$fsg4_WX|dQF;;$0b7Ss8J3pTtOa%8>93o>ERFb88OglUlpAen%6$Dr5~Zn*V*|y
z01YKA*UVT0E8}y?4n?HyD#k@LP$6WKUCg%*S22@kr;SLavh-tLKq8^V_m;$ir4ldu
zjh}`1Ei{I`t>^jJ34NK8D(t>~9rT(`L|%oeU+y)GhQyu)FFlEl!1>V#PGKFNgc{!S
zT((akn=;xd%W{03UrYmoPYp}xoXgRqb#Y^xvk4K3ES6K0|HG{KSP^wl%fa}7Axh&8
z&zHV3N>8lD#^Gp43}4AP3`kCxCE*e%S8InXww9g=kSuX}7ETYym<1@RwIa|)^qyzu
z7V|lk5f~%eg-}RDyAch>$rh%Ln`yT<PEk56E}1W?{p8Z0ynq_zwR-zw`OYDWHKo-q
z-RJ_59C`V*U#(ql_bM^*4qEA)9Lp>p&_>z&$D+n_gz+M0zyDjbzrc@)E;;=JwSuC&
zLzb~`zdn=uizBb&!*&T$;!o*L+89U<T)%5QjTyf^mgU*Ub|oUB_4rs6N>sE$*Jq2z
zukU0kat;gy33O9xX&d~YL~``|e_N|eDOMd2!AYZEifE4(bYqL$4V6=PY}mpeLPL9H
zdnW<Wa?aT>Km8JYjoqy4cW=bI?uQ0qFGt$Cetq82@kE$&#IHd~G2(DTUnNNEdY=L}
zTRa)6yL9DVd3qU!{srWP#vH5xs9*F|F$)<l`!m*lyr5)r|GmJ5u!_#wO=zsya>h9i
z8;dej*ng^?9%j9yz}@}+eQRN13!5k{{#oPqHZG`Ru~^$A2O%pDSYk;y?yx&^$#{81
zl)Ew_Rw-Xbr=T_zJLddxiKAR8BBEe_<92bQN>mh+d7}-aT7pWK?-wQ)@y~4=&lB_s
zYBY^@1=^=dm`M#er<7$!Ll!w%?4b^SUakvBpjR6`wYRVupKkqGrW#&2F>s}3s6xQ_
zI9`IpqNpTMCgC+OFD^>|ucJk)@@3Gzh%85Pd+?98rOQ!SmJ}ywlB{L@Z%7>${WD`J
zGX%@LSrYjtys!SrA=)oLhLn#(hODz(oinY1k*|8b5k0!M2AVIZb0XOdqM@4NM#B4#
zdns~)U7olbnK|3ocwDH(fcpsoF2ly=Dfy40?~0-{bg$~-kf^fY*iRoau6WiSiW$2=
zmafHqtpW0A*gBs5N*}U)<?c_U-8*lvq}8$TBg=%>QN{S!i(R9kqc;S+oYPss8}Qe*
zehKHGp#WVq!wBcIcFLK-%Jg0O9Rs=PM5&BhrG}V@La$P83zQiz9hZL!Y!-1UHmyJE
zPBZXD#vG0>(1cutHi`yH0f(dYPfiasb+)d}mzt6HTUkL&c};--IUp>rxNYNs=BxC)
zj?=OWFHDCe%#_Rf9NZ!H8LIBB@AVQx>i^Ek;TET;Z56j2KA!U)MMnNd4P$nz2eg&#
zmN1QR(YzhBn0|?QX-$m>YdVVlWLH)_%@H>9d(^a1(x<#tLpiHmw58T;>^jh~2KILI
ztiZ2oau5lAVs~&g;^3l#it-n|?SHTq%i2qsd%CCmI|VFDA5rqK`^(-?BxaHPH=WII
zAo6G4wiNObWEj%&(zw8*+bmAKi&Y12I;kf@Gf_}Lq!U<JsBDkyUoLsrSk+%$C@}>Y
zB2h^k_QXKh0k?<GjhFMs@75p-j>$4tXusVzI2+>Zd!d#jrLe_(lKk!d&6vvxC+5ai
zT7G={))UO!5~^irlxgepAgUx@Cc`1i_@DM!J#1bw9akRSg)EJo<KJV%MU7BEY(L4z
zm!zbW@fMvwX}T1Pj5&w@SdJ#54W`AvYS;7~GHKXV&eIe$AzpJr{^Rd5tuERgthDvt
z8p)xBx*DZDiPrvM&K*&)+yHm^OgXieE#n2I&bP|7i7yfm*lg}i)e**lKL@|r>mmNq
zea&OL9;#Zpgx}`(kK(dx?cTjo*7aorkWz(g<C7qoNK@~!+M%P<y4(>PfZX+|-58ao
zEgP%XXM|)rueIs|-ct<HY1Ri?hq2^-X``o|n-dMYqI8N0_(3ZkJ(JeZBp~0fD{wzx
z{V2Nsc7GBVpmZD6>e&A4q~tV|d5t_YK`p=Sm_5w0`V048Urv4T;+SVo-Kn^@3S;!-
zArjQ0Tw!?aP=-2gXGM5qY1w#je7dgHU12&bINxP&RAs7WU>yf_&DpM@D}UFiy10Pt
z@he#CD3=<yr0dKC{~J70@3AT4k$!{*VAQ&pmTFAklX!`IF~i<y&1ZS*ZK*lgPD{Q-
zOk4u(36P@ftGKA3!lvH!0y)QxcdPDaGJZ)MZ_9q{py)H`S9oITT6G)hATUhakh#3%
zjMLrTNlIwV45$?Mn)k=%pOw9-LA89IHHY|HoJR39!6J4Z(SpE^KiQeuaFV&d8Esqt
z@`VDzOmemf_Z{dJH8C`5Zj*0`*xb-ZcOQp@kt;iL;NplP*IZ1q^I2FZ3X6PbBYp?V
zWd<cb^zb-dZ}x5_Vu|P98mwAB4?D}qubnA10c^ZZNujM{B1Y@1dPhv`M-~+&4W++N
zCPhZ9T=Vjc`hCcFEiE4%?@)DZZK8vH(8}Pip?crBJ<{3dGFScA20WqFcje9sq!d!b
z7$_6sk&pA4@)b}?tQw*Z$8oZG_%P2M;EV|u-naYoGk7Sc5<`LQ$gz38Id!wt<jvDg
z*;XDWFShG;WvWVd1D+a5Ux7-zv_eRWg?+yMnLb(f@JP;Re_PLwk2_nVQ(3-I`_mn@
z2M1lcZZ$Ty8EHiUVm)GK0a?i4kJ%<7v@;0G#%Q%;vtKhI#^1iWsl4A$v*sDpHU(U(
zpi8`~JWd}v&38hwO!!E?4Fys5J<(87X%5J@05%go(5yLUnObDjPK@bqO~v99xVh<>
z>QV5o|6Vt^0l9-WFLMulT;lCVZ0(D0FR%P$Oi+BH-a^^c8z+IYyNj3%KjB^6r6j+<
zL@-uuOZ%Apw4tAveQU~3okCBqkIEdZTu9ZNvBqunwo`GA<`*^^>WJuX_bVzZU+I|0
z*}qeHDHYe92j?WX@e^2mQKM*z3|qT%TzC;nEc9jS=Ltfl^|_38Cy6VkiYjxtBx8+}
z8QUTD{`866nph0--CIX!qsyM}%hEE(rk59oYyU&@n-3pu8;*r)YB*n{#3pVrv`SnY
zl^+x3BR1ZZB<^@qcA-zL9Nh!oo~Aq8nOdHMO@314SRm^R+1uW;wssG_(3{V{H^@rK
zybX!imG6P}1GS?Q#gZ6a&^<u@2?-W^u5kT^GDqpFWL)IRsab9nyQ#>#pvls`jos-_
z$L8x#-G*y|a?%7oe)OiSiC<Un$q2ZiQ-QfL;O^$L)JY=DV<bH$Svsm1*Lb2nKR)da
zt=sx}LeTGO+@s42bV;QA?P<*@HcguaJEYa1nWgYVj}A(@@JiSc-^kK`C+P0{GX$OX
zkjRxwl%gqMXYE|*o*!~l%~Cv=2{FhmjtC4=C(!?$Zkg@tlh4F<pOq35=-1l!fvBLy
zt=Z&_;Fi~Fe!g=mukJ%j=>-SpuHm+v_QOnj(Vq&G7prrK?0HM~0wSYS*Vt+p815V&
z<DqaLKQ6yRD5Rjld%C(>w6~Ls6jnWVjUXl-QwfMsl@n7E!zwUbfJ#vJXcIBvkVmh{
zrI#(zacTZLXL(p5@e-$CGRZO-6u@`3198Q+v7yzfp*?vvRTa0VL;aQZjSP9>gx;p`
z)sb7>)x^-u_T&BY`;&N5RQN?3o5?x;rR`TXDmheC3oZJvZnj7AYao_Ph8*SiA)%A6
zjo)R(#Po^JlNJR>FV3m%^;X79*g+{LFbJ@^hMO<JV(BRPQlqc6Gg$@)hwbzK<&EE}
zH`Y%@HTOGc=oQV`bO$edN?pWmoVIF<iU<i+lxw)n;zWn%dLCY<|Ctv57K_Ddqlbr$
z+3)CfdSayH=g^1<ZDvOrb#BaIq?~4w$EJ<d_wJtsom=VNIUhJq&JT6-<+~;wmfNlv
zmD^vEp@!yBhaR26g2g3Gh0b+j$`BEB2*^9E@YU5Z^Ybk~;2}%PV42x6m9dQzr8P1$
zQ{?dzmaP)%otob#z49;FeWn9#C>F?=a%ANFfR{h0s0A-__K#mQ>wU$_o>Sh1Tlc%~
zk~^dmhUsl2ZK&mG9*c?=lopv-Sd~(9D29-~?9Z206wnP0PCv@7cyVScQJSkRoOG1w
zwt8ntne<sYnSRvbUNLp-S{K)04~vA8<y72<Mf(kDM}MOvDVi(gPBA=l9oFOsM@Z|j
zORKn;MPX@@{SXS;I==NKu98XqZ(#92XCU7^VrEM%%l29Kp*t6I$7Yq1;sG{FQmBnq
zSk=iPpLzPuoJ=>QSGpXdMHeEVvpmwdY-w0XcHP<<BBKQZD@lsThoT#-aso147zUm0
zZB?gjDoze43j_*SX@i*a!;;YV49hY!HI*pH^X;mNT(u^gGGb6=uXmM0SNHkc%+zb0
zoC1>{XZ}IbQ%V9c;__$MMysb<XKdV5Q0~GO@-NidQGu3ymd4`}eKL>3Lv$ot6OZyq
zD&E(A{-2C$0yV36rx3qE{G!aosH8;1c2(e_Nv08>wHD&b1l=|Drj`|%v2!08zWOev
z6)Z!xW-#*sR70YLJEO%i{GovdsJNDnnM7${PNangP9F5ibq0z2CLocFP|LOEt4y2%
zw2y_o`o1Ps1Fv1`Rv*V@cT;wqbU7N^Rek971Sm*eVbiC?SefF<HHx%LKm*i`n6SL9
zC;iq|<&iO>*6i%ZBWkv*b7Pt5**>4VdG#9gHF1MNU$C<i*36H2*LPF?!{q$%p1-8-
zNVqn%?0=}a04P6Zx1-9zn3?Ba7?@!A5&E2-w#M)hxRWt>1j(zU%90)F;5`Y8^aFY+
ziA9pll!`b<DePC2S#WZM0HNXZ4U(2=hU`1juYMh)T3*hb&R|k9?tk0`DFvh!2Aa~+
zI%G5CHo*`2iv@P|eS0fCQt=t`8L;zdsmD_-V>+mN4u$S?hSDBtcxC1GuX~?^6JIHK
z2rc#qK^0D4p8-`oMh@kv?99FS`XGH<B#JJ-EOVt*XkD3#fi`U{YAydn?DA-z!gB26
zP~eBtA@NTk)=WYU82Xa&RQ3y8_MhhRY&`+O8XzxKS7$$}Gxot7=Mev2gFR(N7ya)Z
z78#4Wml8e*327a5{pUlbc1Ck+pbT_O-Et>v^7)n@T2XfB<b=5Ff2TvC$=C9whe0zU
z4<y%kpBkn_&%N+VFUJ3Vvb`UVv0qtk1s&96j2`S|C%_?K>y=ML0TY;(6%MG|s2tZu
z0O#Z7b_`u_&AwC`XMc>vZnbrJdq?}FRGRZ6fr%+fGU$e%O8eCO)7avO!_yC4XU}iv
z-^MaPWQ~W7r<x|UbQ=zh*#{g>^xiccEE(a0X<C*POoUh_5f7h^i9z%A_b}1R-k(yU
zqOPUsNu<`M0YjG1OQj>@{hWunohp;7HP{q~7ZU9fr$_gnJoEb4Lv(qk(6rM9(rv$q
zi8~6NuS+LL^<=55SYFXN{VidINHdXMzwUCnji)k2xwHE7y#~)qI=WY_d_zms1u`&Y
z%RHDI|Af7S%H_>=r)-UtRo3%RJFnTf1Pct^MSyNQupgTGM9p1d$_VRlO~dK5uTjU}
z>+K{iP<Lo$hfNJXXIht?ie>+7P>He*`;3CZ9A{qa&7x}vpdY|BGD`P!w)uw2V}n%l
z?BFaPD4!{j-h5M?@?fm?_-&uVH9Wk~+6@|DkotCH(3+=KJy*7(ht;vd0r5q<Au8{c
zo{8F8$=O9!$vEwnjBemcn;rW%1|oabjvv%{#|&=0__JYTa}lR@c<9r=p@SQ_|EEKQ
zJLb;sDg#m{8ERwlT&DF7@n});JXQ^$5kT#1R8QCwZYwK=M_NoYnFojz^EJYajLlU^
zgoV<Vg~tk)vM&9$Sh*xZpXJe!n%a$w@EsG}{!r+Pt$MSqskv2b&f3&SUA6f20dd~B
zN6vfvu_M+WS?E74J&W)`fOX!g(WV9V0w`pYT=|vO>0&X;#@6~?aanlH5*u{mNHxzC
zpOyU+q*xbBUm1=ZV?j^v{?MQWImZXzCW+xZLq02BmsdBZoOkag#D5+&X5ZXOY57rf
zjIx1@VZqt*cp>u~6Fbvv+(ql9;z769{p~JHN7T*JRu4@_k)K1JlAl#!%Z>|lP@9u-
zw!<}GtAWzyu!+Yd%mo|2GX>yc2U({mt%W+?frxOHGXD?N73hY1XxzC|d#AHU_TRPV
zF#S?4^O76!*+*~_7ceom;FZP9j+@h%JR~vHyK7FrWXooTccAI<h)0B$1~osqh_qVp
znID^G!%h!Ka`@!?h}k`F!VOYHx%+!F&()mWG~PO-REGo{_uh(qHkZu>6rjKt2S>j~
zJ28Zq@itp%%L1aBR>SvIKh4I^o|>qvW0>vQiM#ZPPmQNYrMez-I5srgT3TWrsJwl`
z!?T`WRQFU;1Ymnpkjwxco)Dtpc=6JqlgnP$HEtf0AMauhUBe=SL<<)REeapO-ar4X
z-rwGIya~BF%LaPQRD5I=<s4J5^<G4%Nj&4fcQl0y`y_mzbrOofzEdwgYsfV3i*w!j
zUCgXX*oOQ{+lfn&qKT+o#NpMfy@3J-_2TI0*YDojl+3b_p8C|nh`ff*9C%>pcth9W
z9lv4%j$<{*_HB|}?@<(`&PlI>{bky`8rwTIO>N)qgj|Kh5l3WrYizg!oPmj%O>agu
zkGV$QKb3F8Q%%VRi=VMZ1YD&)X`DzK{qo}6-ug?80sHXZ^`4v1A;-FKCo1w;LVV)l
zS==`EHWR|~$fb@Mmz1LX*_-Udb9l0N6r8mf-w=MFXPKGU{Nm!`Wa}pHF>{KtvCf}A
zk7YznOnJ;ck4>7e<6vk=odR9bN=iyvuV35nrOGpcTpPz+dyEdq%HTNByuCY2e*f#x
z)B9hY%>k-%pRMd2+fLuIqNm*fz@hZWnln!lxs|r11ltn!wPuLgg+|g@8@oFL%KgQ4
zp>+l4<l$?M4}vZQZsdF`1GconZtPIyqIPvgqemCD=(Q~@vg=v9WYY_IVVfqNB`#d)
zj3~hMbWbcjMmjBvL)G%x9#8hRlQ@*i-ktThBu9l`#-QpabN`-xi-*b*-!<{6E~L$X
zSE<<Jgpi?^NC$dcqJ{As4O^+id~8H(QlRg#v!Yh1`t0uGay1>FfHxRt#N3~8^u+2L
z`V`VEPgi(PjL&Z=<Y)sSuj;2)nsfUi(qgcCauxdCK53R_{+zXEgcxkW6);YaiAH<y
z!IcI6a`!S0^TyfJZvtPDfnT3P%&Wg{vTIJQ)e9Xg%`?dqT0<zT@B4O+j8H$8nDs61
z${?FlHU`N#bT|cG1v^$>nmECGb~px9RJvL_129MiyT5i2ewrTJz+s1`RNlA~ifj69
z<1Zba%;k;UWo0LX+0BZf$I#Hgy0X|4-`jpB$dr+RvavZu&LsEF-&r1DaidU9(S@CX
zXhIFB6MC{N+1QgjqiA`!)tJh1u+V#Cb5=|_;zIdV%vJ>oHbxO;A%xEksR}O|R@;rP
ztZNI#@3Pu_#Lm>%&>lK!dVapILOu!kXjxb~m8PgU>8&3-2}|;sdE@8kl+lkJ^_>CN
z-D(i5<pGQ>r@p9@yi7O$KOG2Un6h}o^z<*A+f!lxGx_;<so}Bzp-kV%g@G`6dez7^
z`@A`cV9Gz^A#3Sx_(1r=|M>^6FI|tZe0@j{dn*RRZ{2=eW&Nn0YpEu4wU&-nQ+>3a
zzn+E|&aHQ;DAQgOa-XG#_*an8BbdxkO3&*xb5ly;;g52K>>GFH-Hdx@>x&L2`CDkH
zF0j-JVb2&CJ(JPCp?vXT+0YPs1>b*A)_b84*afnqpm2};K5uMKwA$-U5j7K2!-)yM
zcOkv+Vk7VN{QN?5PhRPsv5n@3FU?$s`0s@TcbGMMDAoch{5=uF^Z)jtzDLNxP!??-
z-w%9nW-?LNAR7F0^;oF-`TYdZmYdu>E>z5ywvQAcL2)`Xe5Z%@g8A$$YbF`NKQUoA
z?pY)C<p;5MhiO-@FY2GZFe8O-N88mjwcNWI`M010gicH>(#Yvw_Tf)qICAe>5)D$C
z{_%=8Z{pft;Ds9!c#k)>^*YvggT%`1v2QrwsGjAj&%p}onkWyu?Ct0HtNZ2!-$Tq2
zOf5+hm!v~T0Nw{S*iNs?y}6_GK}fQ8SA2_<MEGT*8bg|<R^`L+kbfsHNgV1sEM8D=
zoE|=So|ZacW{7+}CRGm~&g5u3w9owo8NFx`3Ei)sU}M6=&ZV7FrK8aypM}vKj5f7@
zSgx&aeuudhH8g~SQamfOBu-#M%GJ67(ve6jr?=~QztUV)v)fvsed?*41S#yNvsu*t
z{oxU+ad&<+xEt?32+@J!6G7E8YzPcKu}JCX`5qS79%a4obD`B}DD#CC;ZvG`Ur@h<
zIzxWCJ$o~)uGrbMb-TL16N;Je{h%-f?r0qEMfcL{aq-7zgOywJ>5K1tr#fPUAJV@0
zd`r3EoOXcvXF;Z?(sHl6X}m8duB#N=zA2So>Ql(NxF?H2UnRx={zb5n@`Vq$f}zMi
z^$)MUxvP3ko01}eE-2#Y^i2eAA?ml;!>yC(AnV46v6D7t_Nhl()9q_FD;S;qunmQ_
zJJw|J38W=)uy3pN-gu+JNcBjnme?_jJ2m4>eXDpl)$zCWv2h1)4v&LU4iS6b{HbN{
zXI%6vJC3hiWE9z7;bJbPq*pI3u23OEMl37u-|_P{(|PHXX^$gLY>~97&M<N^N!$Fh
zCF?Z5PPc3?RdcbT#I)$aR2iv8@$8r846+VAqn@<5mf-EMqqV6E#_Ov7BTQli;!y@W
zZv>p!w5u7?CP$o>-=FF?gasz|BRs-yFkAxuyl{oa`z>Ju#oF(z{r7pcx=QKDD4CV_
zvWKPe5Z_=@mPccW++l{S(_D8-Z!m89t;)7ke@zDkUt+eN#5Gx)rY2F_t_bJO6$N;y
zQ4N_(GM+@*!x1}g_z}fUzG#&83tq>0(7ok#8M}8#Sk?CIi+pn{Bb7a|Y}e$7vu*b`
z$zb*IE^8ejV}m#iR=is<A_!{n$yE(}S-IxZgbv$GGMf!@d!+sIc39bx?#sxH-!hIg
z>pSZ4u%E=QE3*GIv!pF@!TSWW*Q}5GWMj#9(DL|o0r`!X+-LNnBY6{KXQ~3chwyrd
zt?ejnf$es4vQWziWFNH{xvR+9zo@7wT{v%WvgOHR#Ah6vY9`hdi*)OI`En^r#Y@V_
zuv$z|678*pG+UN$H*M(bG+FA<i(E&~n=Q_m?|a3Bp&Wv!AbhPtLorV~g4i@uxnw6+
z(`@ktp^z&5yj<g<)T^@RLi$scWli^oh@DPPa9E5F<|0cLRK<t3bDo_)Z=<UeZguUN
z?fs?&^=}D+Vm=xX$P@k~B}V-93bvABHgQB0*{?+mzD5=wc2|g_e=_GRs2<WiSH8#)
z^TX2V%%bZwbJl6)>!*sH3ewQoKd%|7@OVq8f$2xmzmQG0W73uWOlVERK+Ah}FxX9Q
zOlCbXG?VORYVFDFJs!)s%k;;RZ$nK*<eDc~R`{Z*>ybw`f+vTILj~_V)W<~an?};(
z>z-zjpD&h3`tkB)YEO3iD8SyZbMOMaq?nA+;f)opAnGXXbYYv|Nlx3Y^(ANH@7S*R
zT8xfgB+vu(bVh3Z+tPf7Y+!QM{Y81FT)i6;fyq8`uIhy8rYdY^$#ac<eP94tb@K7n
z${bb~kV{l_EOK_uRO5ka-IFxcEZS8Xk&iS|4QVJDHSh6X|81Im;g4WTuP;2MnZHmV
z!_jx6lmg<CxLF%TCx-ZkA1+}klVGeHdT3K|Qw%JkjS4l+PUSGs`y?}gf|{PTvdcO}
zKQ1mwy51uswk0ganzPtOCo$A?FnbM$6zM5o)}Ow7S#B!bW3js?21Y^ke1%<rZ;;*J
zHzmiPgH!EzzwpH;#m%1fnX;<5N<gHvk=E4j+4VqJUF9?}bMhegS*dC_pK4|;gI9ls
zweslFc~dq5CvHxs!=jR9@m?Q<Q8rA0erNHFJO0f`g-ZCJn?#{%KE6c=HT>LBBV9?A
zyvn5-aVx*mm`DZ{3}TD2O>(%Rb3Nd#aUHBHSx<CV>^W&gbu?mASyfAaezvi%R2mtn
z@ijB(Q^FQ_HofZ*92%0-dxT@S>bF*`JLX?8voN>)t$J_1+_}ku<$naj7;fwH@iT$j
zq7!zqbWPkGAy7s5_8QFDGJZ{UejmDYFmk)Q(0a?JJE+5;t<SXLVMPKirB5ejecb2b
zp8qrxnDyolgnC`Q5^#^YPOrOnhnywDSSW{#Pvq>@<CofR-Z)u)ea6FK?AQBOyc&h#
zeGe`NYkp9W-F_|gM4x$QNZwdiV|MrKSlfRVK}tmfCMKM$dPaH!3)rHzXNLwirrbt;
zWq-m2k*pY<#1<0smqga$st<;p!|uO(x4j!ES*+ED7(M)>4hg!Kq-dX7OD&l${W_U<
zpj_17wLpQve=e{@b^pYSXaB(HwnxH&<cS1arZ>9CPPx!Fiud%8nAJNy&3_fSm3`TB
zDzJ3LMzDzYj$0JP!X6Trls!DW==E0&z?~Cr{{aiQAAQD}`1y~Q+eS-dEy4E{W8#E4
zvaI>*Guw`J$vbHsgwbp}u$*)X*{F7qvPPD(?QdaTdmzlXSE(U=?_OB-38N}YfBLSe
z&3$IsH3oH!-VL;+!n>uVFf9XFgEO6@luRQ}LtmWeW(Mvp_nL0RWeT2L9<kYxugG-c
zx%hj9GcqP=2cpkPr%qrp?P7m_73G=YJ0~?`%frYpx6NQSm$P?~LOtChE3as|JI;0&
zbNAJqw>SF7=6?p*%WFSFUk-)p5yY>e>h4w4$nZs+Wyz2SE;;PH8i_d>M=bhB?YL8s
zf<nfV<Eq==W60(5&I#rAT|E0;q8`Xn*4z8#_KOd53;1|Bvp05%bj>p*Vp?S6o>ymj
z{nd+d=luA#^zE=&DZJybmx-p;rjpTddBoDy)?Q6BCFV&+8?|$=<zW$oH&#C{nWL=8
ztS!b|By>YIW1+^!(bzjNg;LC4h(H{9G<7W9+W%Zgp2quLhO3+a32#`6K$Bkvzmh@9
zZ2b%q&zbti?>dozIJ|MvwqjeyTVDw^`=xBTXAw<8G5?=iUppfBV#4ZHcGD-3c0qM5
zjrGAgv^W_H4>#fBlMX@NGtK>@JX6;<O;gm!ZpFp;)r9e!c+dw0B_0)$q;GFGw-}wH
zXzwpB>@HP=mDR+@$J5+b&V77_iu=J$D{S=79l3=&*aGbx0q{73Y^E~Mvd@&uxmopM
zW=3&*2iG=eXl1sZi$eHf6~p+sV=fF~8XJO4sSgTGPg$F5eLBu0(h7-!)M`Ki#}Q%J
zm|j4+;9WGo_`$;?lmFc%g{JgJk{0>BH;iz)<v>#D-1+mmg+sI~f@0|Rli8SQq)bKI
z#Kt9A+7!Lr=jly<7S^IVdV%xKm)}otQut_T4$hNY_YSHPA@y%MQp+D&0pZc2CpWUE
z4;DUV$H(8zz!iP`b#DghsC*>YO-BeTi0;tdtz)Q+@$a`){sDtCH#0-cxOS{x{_B?%
zhf$~VOV9dy&KLQX&aGn~Uz6hpDLD*`n#!YLd_|2}NpF^0Y*EVu!;G~CJud@`w!UYG
z=mpg9If{5_r#FN!5wySavK+!9$m!N*#@eloqdm^3ns6MSW7y|R(w?&YI))IqR`=?a
z{?rP%RI7K(WAEV4Og*AGRX$hq)h(dojF9xstI{E3dCj^vv`<Hx&Sstx+#1Nq!m9t}
z<$5%e`+e}mPa*#W13N#6zb{~k%ZLB-*U)WmTx68GnCoK_?-s*_K7e<)yz<*tRSC6n
zAj_9AUwH!b3uw{WCbqZWNWoL?Qw`y+Elu-Zoy*ZJqi)KJZmuQvagonid)}3fDg0fo
z=+!E{eolVc!>@n7hM#>w$xqgvUfE8WQf<hELI5u`Cy=80<VPL>5x$%{)87ap<tYW<
zFlbEjPq#m^6KQ{vyENBitaI3D)kNroa33D^t=ViPK4@mJSU4!qvC$tRHlm{c`wTAX
zD;zoGQkWOG?i?QHF&ci$GM_KRj$#?F_h+3aq-Xi!nw@Lj_uRL>K6uS;NlGuYW*B+7
zRddAeED;yb_GtSsk8)vy%R}|#f5v27fXd+wjRGUZb8mokTeWIQ2w#V0JmW4|58)9`
zD934|1PQBY*9nL*m|~~?mCkso{`BiLx${xl2H)bNSSpZc1@qsjDN(pP^<M=z?Ph)6
zo;%;0t}U=*I3`h_yOy(BkfVFcqc`C1vyZ=n3mV}sTNu1tizVu+OzV~Y`()^0JfuE~
zM}~6zq>LT7NoV4C>4mDDzF=NdCc(UE1>#bCjFN%)YPB`XZ{FObd>eALBPNLA!;9xi
z(H9gp%hZ}XT7xNWfBpK!(^ED2vI0$OhoYN8=r<y^;I_tcuak(S*4yPHX2LSn^9J7C
zbw(G{B&hGQ$nAdqCo2hXIR2X0-y1|wbwV(qe{ag&KXLr8Ivff`CiwNQZV3c1ueXGM
z4?wx#U;K~w=04R`jsN-Ux-@>(|9l)zqWV|+?Y}uO^8eeJ|GN{T?f?Jup8vZu|2grw
z?CaxMH3zfAF}$vWi?`Cx0kC6dSLEmC2N6hxrZ*oD(4&-7Qlb*Rdi{E6TjN;Wu`>8P
z3Ti5<ACnaMflX7=8XB>GB<4UtZSr}3AbbrDc=qhsYp^m)tpt!xcH|}%mN76$Z!pv<
z1}}c`(j}U^?*9F!tut2e{#{|2oFkKR>p6)ma#Xq=crc8BfPleb<1o9GL0ww}U+zXX
z%R&4lI7vh*wQ%tw_U&xxZME^@ueA}W3-p#)_2W2v<2>gnHexXv;74Yo|6){|!cr^_
zh_jb+o^v;P9WI#n6g56~dOXKxywH^rm<(6Tw4Q6fNqJ7Im|nGH{wpR~Foo|ere_^I
zd(jP_P(7DG@?6o$sbbj<sl>&@%keuOe={89&qrPN)P=je8Eeu((+<qqop0jNUojUH
z>i@NoH9r#XF5<iZS2Z*=<g}jkp~#TWQULk$;wLvyNY)JKb~KS8!4x2-EF~Vr27h%Q
zy!+sRAjpk9#u5N&<08)M?keLTIL&Fd98Qq|a?f5qJ`ht-$g}_38hP*^3A_JdLDAHK
z=B-u;2stAQZR}%o_N_e`vb`W~M<j9~zJ$Z=xJ&_b=0GQ#gwIV)QO)v$S=q#;5tpqt
zuTP%}I7!`!K$e8XYWjyLk@H-Py%A_p^+sQQYN{QX8W<S34N~<+9kD#kKYvPte07EW
zS}s~4&~mC#LFmFbC9KbV`fipPyGlmROYX=Slelc0%jT!qjyTQb!=fUVhNdQ3US18+
zP&(<SDDxB)Vo|VnO>nTtP_|7N)pHZn4J3JXf<c!Klm$aWQ{fw~pv2`gpWw;ju&(8e
zdv2Z5g$84MkJN2nd&F+YRG|qa<xQtirF88koz&B(?y(M&SJa^!1Poq5TN_F6yTg$P
z!tcr?H=bf&!NTgPJv&)@Y|aN`0P^t^^^baTR7;7(F4c*6-=vo@GRk}=Ah6TQ+>s87
zhgUCO{sp#HWb*ql-1SoR-e`Amzed&zUm|4^!lxunL{+Hc*tSN|d+ukPj+@Ch>6GpY
zM{-!FPEJnZ<Ko`0e}sPvxs`sCUS*!W_5fRG?c8UI2NV=91I?;;6hJP(xE__xM5~bf
znL<3`v0x0By@HljJXXEb$>9zwD7D$n|KbNJ`ry~PVkoz_3Lt?oeh!4(LT%jW#r?`K
zq)5A2ieLZ46`mY=>P`lu42ph%J(==ZGELiyy`q7jYObzRj-d7#(SFoDE<81Z)wmBX
zUUQ)5-Hl3Lk&%&^w&5foAV`5bR3gEAip_*2z~i+2nPO$Ewzh}q=g*&k$$lhUeM$_d
zQ}4B#=hmIG6mq)e;*eaJ6jp;@Pf0j!eu=Y>pbu((gq1agk<K3x5kW&m^_hazsOP2g
z?s6wG+2CF5dSv(E!-ovfFa{z{8-oa;CDb}&oa=4`#X@f7>Lkzkas7u9(dpCEud6{&
z(&_YIj)0h$wsM34-XGk-ccwN}YOV;1l+~iFAY=MiOyUMkouuo@PQMfUv{Y5grb!%5
zk)8c;a&jl*uo!m-Z9gVKAtB@K`OZjQ#MLQdYPD)hLOlFYf?hpFx_q=m+6#YDdf~(M
zdK~#Y4IXbijC4c&_UO*xa_gO5MNNWAUOGCly<uzrOE{!-l_O@-oT$_H<Y3h~0N74;
zOKaspO0OrL|NL!_dvJ%aiqHa-W3GEKqYhKpClfeGJ|~SkQD9sgqcA7pdRUYy8JBy5
z{Kfog<(dLWYJ#>p`Z|V<KDSlAy#7>Svyf;uQlT7u%Uzz$bdW-Lu`h>shG9IdtE+2!
z%AYT^jF*y9C?g}IxxHNhOoRUArMfXdJnO20OM2>5aIYsvd+71m`9bb)FxBYW{Nu;?
zb5Zv=D$eG{DP$vEF;By3{0(MR=|xP;rBGEfY0wHBMb2E?Vz?C10^0J@7)o@oBD)UA
z4>9)Rz;~f<4Lm_-R$fh}LXHYrfE0}D%9SgJM`LHXAdhJTx}8S9;@|XIV)R%>zs5$1
zbY?3vn`2Zt@0p=(i~7cm8>21t9@w;qZL2%Rc?w}n9Ma&Hnalt+lu_9Bwhal7liYRB
zcX(77Zd8nnlFWLoun3~q#6jRw_U|H4QVKV#J+T`>&Q3tgxEm&iUgBCE89eyY+g=ih
zHiz4W-a&9BLO0(p+VvhOepjwt+tOg*yhn^W`(o5fBi!{XK|om(!CMzVUz>u}bWl!O
zS~|sVVW>1Od28-uyhjL?Nepl!So`esB=EapCW;RiA3qtC@I(UJV|i2;Z-H#H2{-C&
z(yp=?T!3&_JpVg7iSu<=1;@sju##Gt1WO&cV~KuVsH+8t^OXAW<HtpgTYADWm3Av|
zT>L|VoHRsPZv1$-!gh&582pUFQ&bu=>bPxpWk_C0DIG4AiFTV!7+P-<x2%&>QbG<e
zCw7+lMFUle45ZC&V4>vSyd`wkY;JAsff2ABufqUl;hfh2BLY%)JNr}qqJg>9yDF@P
zUEedqQFOoZH2FSwdK!bBsN`$Cv9d<2AH?vf=)<^1@;H}J5O86@_-y>7jC&dQ!taiu
zB`!+glb@eLODO#VVYJ_cJEJTeqF*?k`@~p#EMq;}8t~otqI;xvINhc5bvQ9R&I(ga
z{*3lw@QmACqKq2^CUrO@99GX|Q^lxD?$m!+t=`k%&4+1aW|jqGM1m;L4xfVJ`{Z*}
zw$<*kGSA_B5<Cx`wOFm&35#wsz8^?st)Qbuj*kzP>2R5J6uWuf>B&(@SXdgk+6<j0
zKcYuk?z!c&M58_!ZghZeTEZCn6*XN`Ak5P~ek=our2Xp?^}SV&*5HKl>$*}zS)F%H
zc2<TI5lP;I<X%gZ5`Yy$P=r;fw9A3mG6Rke4=+g~hI0W_LFJ>^Otbwi*GYnP{l%H?
z{;#nAKR};^fA|la^1sEkH1lA!R!{E0|APfOp1qdGqkTE5)wTYxX0oQx_s04b4VHsJ
zpi<pwAtel-kTA^yo0x`z;yGHjcC5yAXF$iF8kRHK>J1|i1)&_vRKw;!P)**!FFTvg
zoSFkl)-+;sM-Tz3jvhXEFq1gNc<j1AwRHn7a7v8^52$!MkiPBu5kOvSHPeh+s{8si
z6*;-UQh#1Q*IE_1uu7Rl0(c48Zl<O8iEjs1>{O$Z?Wk!?ycy6^O;n%$+@zG06cQiH
z<7~9II-(%-hY$q6`|~gBqt0uNEhBYxPpfwajft3Ey^mxuOmufg0Y(I$3R7;_<VT`X
zViNK8-8*^-G?PS~dj9?DY?QeZBPAte7C8O}Gj=S2iHV7;@Mbs&PBu*>_B}Sw(MP*K
z`3Zn&ABp((?HLKTgH&Kol8`^{jF~P-Rk`j@cpmoVxuw7w0@HjBtD~FTpF1%*IoYZd
zj3jICz3RfkLSQl&mT_}1b;aRg76Hf*qZNnEk)tS3T4jiQ00s`h`&TfvtPKPrT24-t
zd-v`cb*K8GgW3D{7vfa+iZhWtbHJgouDi;`hTSjVfW@v9Zw<EtU06xGftwJN;YeR1
zR*}xm&P>hP8koyekk(yusxKY3^1gH36IQ^?$m&3W4w%H^uY`Ad71v2P9kvKgxQ~_#
z3EQN8H=&=l$L<aqDtZ<CK0VqSLnl<C>_O_5h!rskj+W3bSdx{I2@VWQv>$U-6!5$*
z@8VJg5MRUunYdEQP3o`L5u1Za=8``U{{8#&2M-@E9?Zq{!>xD`yeEWkF7@sbwn(+<
z+GzD)rG4IPdrX!_wbKH?XsPGVp9_*&`DQks7)PdtGHT}D6Y#(%F4SqdzA{iC00P5^
zSIx<;YO(oPO)K}o!NH$VpJ@prtaJk;uTo)?0j@C>o0#p#9|O(E&(9wyOQaKod*UXQ
zELdW&lc{L_qkdIKXmJ6kAyuPZVFnm~U^2MLC0HyWvDT}@<!D^-X>(&PltDf7&3VjF
zs!cbu+17{$&B(|#09Fz(sokAPLbSo9FJHY%JltE;T=JjJSv=t~AFaY|nEwEv0R6|r
zVER{)sUA3F8i~WzzhLFD+RRJM%*=#Dr6kfE{mCk8@}mb@#9^a>usTe&)GQ}0&P_$N
z#H8HdjZ!46Qt{{~(t&7n0Q5yt$Pc*_&@zLjS!QOrfk8p6T6HMlE-<@dhmA?AxoA$?
z%w@YuxYMLU{f_3T4Jtl9%`vy59DCs_VJ#^~Yqhnl93Iz6GLe%dO-`5nd=R~_7W$~C
zt4qxUf=HtdNy%G+(~G@Xj7OIJ!HWa6XaaN}q?XCZ)<$XpN)I6=rm`j`*%S*vFAh79
zQ`yB`h~p~7Mj^>tlna&Xdva1z@oP0l670hk?-cSha=?p(4xBdA=7IA#JUv>g?sq*q
z*?TNrIM4wcB5Q>SZEn2vkh`F2z?0<G7NObbF0<6JKHlG$1ryRrO-s|XiM@jNFg&pJ
z$B&oDR%-;auAg<~$Sg!pF=4%-5>wIP*yw0AI#}RbA~9+4@pa3?<%Rac2?CxBC;Vr0
zl5xDQV#2dA_M^&+;pZ{0JFMv+?XD11!~x}VoO9Uj6k>Eff@cN73Rj@AjRkMv*u=B>
zgFMBLdeD4kg|KwUw6USVu_>g)AF*E<oT+vNZ>h^$vgm-;UUbhJzYM-zDzkLdNfbaX
z5sQIT!4hD)%!92twC=9aU{MA@_~6J$mBl+Sh+A_VVH9XTrJL(a8>2}y`GHX5vyI!x
zz@|j79SlyG<7R0h`3r9sunFCES<<MsRFLqN3oHW9iCk@y9?3L5gGeK<sNtxaQQGkR
zdpbN~V6y2@Nk|zFC1sPd+-f+JP7g4R$6KwjJe93jA@UJ_mi0a^8oNuS8s}9jfP0fs
zPzc4v#oZn>&NCdy*S-pf$4|M`EasJtjwn4pJi(&++RlV0nF4?(Sl(59{QL+%Bme^I
zCqCCtNU5ns+e0s!WN3sj%obsr`fCXLx_(@ilkdSD^I((lssWppd-?JsyLbJ>FQ7ZL
z3&&15z*A>N(+J=pPOEy($^ol7uAR*Rt(vUnHXAOpn)~GID{o+6P_9Fh`Pd)r<<}oz
zC}6_xVoe9cq3m#HDb4NVaKXOzC`PPX4k*^B(-M94(Q0L2GPsNR@1O91?{1sfYdg+b
zl}}rhGaT@?(8}diEM{7jf&{zKNJTM(Y_t!NX{rLyF-oq&_yJZ`{Y2UH25Umat~t1&
z@}WzZn4*EhX7-<k80h>|0Y9yBduSu+U>$=149cfy6902|e|#lFl=dwzU*BG^&z-Q^
zW5KS2#N{M^UK+OCie~2eM>MQ@qg9S!@~eFc3aevZ!c&!X_H-N2X{e{B0%#XNqx<ds
zmC=?lmo4#vp;xb7RfW|aF49J%I{xG<C@9b!odbJrZbPOw9E$AMG=ybftVM?vib~8Z
zEb`Ito8*bbQ)OjktICm02kucU;KSv~u390BgS8<yNJv(yT?)p6C{|!)sQeLPJ^d5+
z<wUf#6ZXDzFVNYk7V5QCg%Q%L$m#AKz@&#dgX$qyRhW053cc0lw6LzOu5jQ@gia|Q
zMp}|yrAP|6hY6c%0Xbim&VqZUocdAeV~Dzw)m0V~->jNe0T&MkRja2P<Y>cFq8iX?
zJ9iPsgPE}Euu6D@<&yDnE#EsFGJ)G*E@&_%Zxsh_t8aCrQnWfJKVK^7q4?(MUWa6f
z?F47B;8K3*>^d@0?=<DhP7_h&B;S9U<FmeQu`}N(5V#FFn|V(%H!9ocrm0Q=93Ht<
zHNsf;6>wAXgy>p+I;2uDP*1N^JMDBW<<;gwE@NT7GYPoBR&I__Wa^hMUw|8<4ai`3
z`S<9`3!hukuW|tt_BwMgFi3!zD~O1AZ6aG+)%`%?(7;t#SU3_G)&qhp^$MHLL_zN=
zA@_<AGuOQl`^D}we{jzagu&m?StjPu=9YbPKBzfb?S}Xi92K?XYLX@q(-qBO-IJ?c
z*}`e4tUOq1&Ip-%&wA<APi-`rV1>F>(1AbNf{r!cK3uD=rSxw~LLkf={EFvSsdln0
z-+C@6=*TDuJSPn(VQd{draMj&IwktRWVA{dotYp2JO_CRDgextE^pfOWS>8O@3gy|
z0Ws6L)DdW$O&e=}aCnxW_f02oJ=JF?(1MtyUTKHU?`hwHgc1e@1}rBzNYLzTcS|Bm
z0O4k4qqrU6-NILGAU;rm#66b~bg}KMl#Yo6Lijn|@JE6N`X($K{O|DCE(W917eqvh
zt|ov<z(4}Q&#(ew^jfZY1s`W%fo9OTL0+gzcfG;QORniOCp*flLiqe8tj`n#`1l6P
z0|ldFK_h00;OmiMy*8qThK3Aap@9@pInzU3`T37*7kh-!?omWgQ1Is?tTkQ-NbzLt
zSi*pyYTVH<Pj0YVk}n5RVIhu3FhOa(S&D%4zlf}wZ%+SoI$9|O;twE1L_$JgYqm|a
z`UKKk+c!N^u3W#~2YIgT7R}QyUCNA#i1cljg1Bq)JEu8x^^b~PHF)8+=%}Rw22v@v
zO4<9d{_*2SnI?6I_YMJxuH$x<Sq~q<J&e~wVikTH#Vs$Jz#5bV(-|zYOoEhV3b5@Z
z;9>yBjCwOeIPF)nA^T<g<MTZr#9syH?mV%GM-xGK?31myGo%M#R1tFK23~Z*i#AsZ
zqFyhMw4JpvO_()$dxrW)+d$*8G;7@=S&c(^Pqt;zNjdnpUI+@?<L_`X;W>59$Z&BJ
zfk7H=Og5m8gRCZ+K>{K*{{Hbs<P*z82>V6gB3nl76*8JnR0#k$19V43U=Uw0>$ms7
zx6rX3Jh3565D~MU4@~{udk1h<Dec`^ii|5mrIOuglKqfE0*^IHqs)nh^69i#h4cAP
zt94f~0T(tJR7S$<qSO}2TKMD!jWGawqp_N53PS*!6<~_=66m7FK53`{{Cet3ONY?)
zyuX51bp}Y<eCZqf5yk6+o?n!_+0J6`;gaTACYjqlKcp9cq_BwVLS5qCJ^mO@TN%J<
z6o&ij6JB0kU!I^*$9R;|yFcG#;Nnv4;NJ3yb)1c`YWlrm*A~u{<ROEOJhnJOXbM+{
z%f!TlCQy+Go2ij4I>lS3Z;8f_sLbJf?Rqp#cRgH4L$g8joC30-e)#YOr2Klp$FRQs
z@efRwG6?}34~-9ok*U4+;NzeF!u<bF60u&$pNB<%^S`Bl=wttj6z~6Y%_RK{vpdn#
h6;5$M_&L8^QVgyjFP8FfGWw*#f)dYjpJ~1QzW@cz7P|ld

delta 108535
zcmZ5oRaBK-v_(-d2$gP;l<p2`0qO4U?&ebg2?+t|?k?#@y1P51q`Tp6(Er{s?vp$)
zIA`y@Vy-#oGO7vgdn;T~F9&pri7S81q;(Y?{n2Ij89}nw>o>`#8tI;&l*n$jbsoI%
zeUtL(v4Hd|>@RQ7{28-FNlU*=@x4g)_Bxt#r9ZuX*{B%yo<6(fTD7~Z#r1^yaxS{^
zYdpgm$~U&E9s7XZ?R`A@Fa;!Wnu}ongHtZHu`RA}qN6Ei=KKp^Sz*UeC@frht2)e-
z-vueYBg@Zn&fMj8e8!n3O#p7*@b~qsGtcPJrRq}xua~z(2pZrYzNh<%YqQ1pbgx*R
z@-&2z%q9+TNS>X?!<^Y|X)330tn`&XU0ib*Fur<SfOCV_P$oNwx(chcIDnB!!bMwV
zlUzg_9N;CDLbFRsO8EH`G;e6N*e{83aU%Vr>ERHv#ZZ#5DP{h)=|RK|0S}E2KANQd
z{1_^k{}&<Ck97*3#daoN#wNbRm9oC(lYE7Z*YOqg#YEYnsfN=fwhO#VX@*##1D?1W
zZAB^Z&Xn}QPZ?UqQA=V6ZC+~W$deQ%#*FBStp*Bkj1iKd1ysJL(A~+L)h`3}cO&pC
zOHW^L)^PcW%AfG(?R0#sZta!Dre&m0R<HOaFQq@pR>hnqedyrzC{cTNNLxVd3|T5`
z{D5GJs}{yJ7;&#K|An3RxmUtPJcH^yF&DR2zOCxglKe>4&XnUvgc-rkAz9>ettr|9
z4Nkw1D%^aZ-ehH{aE+Fho`ReC=E>)2bgSzYEbiEAj91B#&RgfFxee=BX@eL8RsBXM
zmX{$(u$qGgXEiP>RevsF&#N%)b{Md9!<PD0?w*G!^D0rz8$1tzSC;OS9i~87MEr*!
zR>{3zmR*VOrCC8V*gbX*a9S!;DaLI)GyI0-{0iM66_Ux_IA~FXcQI!<L10HS-FcYB
zq=IGd@kM$!u{m`X!_ND$fX3AeWCpRX+!rSHk(wz@U)<38-Mn(qC6@+@kW=S*B)dNs
zhEL3Rr-XNLC2I73Qmja3D%ZL`aK0)#zceYRUJbmSpEhEAqIP9VXKX#*jhh&bBeHOE
z-Z;NYK>*!71>XHr+`j~QxWD1t)x*FxnA_t*WYp4%tzZ64(LAfK9Sj&w`gS-u&+~IA
zN^!85EFDij6<dVWH|Ne%5Hma6iks?5%)e`pZ8&}k!-$2j8d0_^l`o#lz1o!RIZ^3+
zIB;|wrBV~|uAsi(>Pc-iLRs4<e6%_~NYAMCG}ZlkWhIODZ!eaPei6f4MZ!*zU>aL9
z4F~ee29Kz?Bc~_aS2%$&6}e5KcQ>4jWW4wMJ{2A?9&U8mGDId*P-^W)@;{quKoL}Q
z>W`}3k_*an@wesRZtR%&F)4P_*u&SrB|$O0W=oPp_B@cCOJX>oKY~Ib9g-QooYbT(
zwj&{=ist8l3R1|sshQpdQSs0S<qr;7c9E0{t*-bxT8#NH6V9a;6@AjIch2<B$*Uu&
zy&^ELhJQ<ZK|js2{^3ge_OgvgSWT<lCtl<!LQ=LPdO~b*DHb^hasmNo5l_XtDB<&+
z#=|;%C~E)(KWD9Sz*2?gEqmqX@jFEQE?mE-e8+JPX~cao!J11a{jj$V2IdWnhybtL
zc<`-PVBL$u29Ed58RjN=C$}^-G(HQq?b6vyrPznb8f>K`5v6!^aG#EOdUAUXj%=*j
zC4cMb50sAEDzP&SQV4hRT59QRc+G$AC2A8?u<<U771C|&X^q{0rYtEj{j|5viWuM;
zu<r^9%Buc6uA<U+vES)jX9aORW6t>^fmM9d1xWilku?KYSmO)0cSlKKwCRYskN20i
z{YiBF!=P|@s2#8?_9ZS%#H>94Y1KS15(8cj#QpohF+}&T5;{Y<FN*ESrX3!PYR`{_
zlldb;kp|aAMV48P32pI-3%{p~TZHD)lHMS}!ZwTM{dBu>uXwO9xh+y>i-8d{l+;Y9
z#$6Nu>gL`+6gqisE!>x4KTHNRi-p48`4Wrs%e&kkkZnL$XQ+S+_m2yBB!vRgLuu~x
zd&3c9zr{3bG$w{p{qD=3zWG;7A%dH9<J>;C+5-&qhlmS~Rt#&fh~F`=c@=Dk&=ee4
zM@<E4cKFxn_N4!?m`D$=pxMqr_%PHO*W^Fi2S%Q_2A{OdAFWLZLbX#zSpf%K;?gJ<
zYTSWi6Av$+;FAC79?j*6M50`7TNw_QoAne-S730OLAijBZz*=ie!4f?2OTHLrV+J)
zpmabiU^JTg8dN7Vpu3P-X3+iWE?#~}VAbB$6!6pgbeYBBY##3T3s@;BsrL>JXsfGe
zjz~1Q`Jqdx4`@rWH5b(~m@TY^6MbVNP4j-m#zzZjSTHN}^-5=xmv_fy2V%o?J{t^|
zo`UV37>T<fOw73LfHEIQ*|8^OJ)pb`PPlK3PJe3{8DZcm@A;Y95$a`^k-&UsvaB6y
zUpkFo_SI!om@G<2Tfto?T5yr#PQFjY5w>y9hpfdA@*3<3GQRD%o+A{{MZVSF%$w8B
zKH>b5>pC1mUDESfr}=z)+VR=Z<+suw0?%JOhorLK>n63%Wpjfbuvok{-##!=%6IsC
zWu?gDoKpg4LV)jx>iP2nDU`>7pV3}Ws7~48wdm6HG-oH=xb}0en{BDXbbJiTvMS(U
zo56CCXN|1?*qaZcG%WhHaw(75W2`#Io%WLxhSztu=rqt-XncRb9A&2uFJELJLi%mL
zCCqM&UyXZ8ghYh;s+bDN@G(Zh1|Mv)*gEOQp=<WK=8Qq%oHV50f;I3v51X;AKyne0
zwn!;`4<2iISoWFj@ws!0z>HRROzX#W-Cjw}z5&;Q#K>M9K1&MbLX8hc`KG_u)>4>h
z`Al_0w0XJT`4!_#9W%%Zx&GAp14-HZq$l5(z{kmFKYH+n{Y=aI;J1Ee=t~!@a8Ys4
zL&%fque0RHv6xNyJshz7D*EKe#^Vi#ij=3j$i(mSu&wj?JuvV!v0G{+sgL7`o{6LE
z+2-8)b4z~g{#AEl-A^>1S20S`rsc6k{A|~XXE$q+JCkA1ZD(FTR6bNHzZOH?5Ea#d
z-4itu5>ToqN6r*j^D#|&jnv9y3OPT7jqEua<}-oE+be%U!Uf>)wKn?`ZBAP7xRu>g
zdUBl)4{V~Y?Hny0p);Mv#U5uE_Zw!9+q`UBMNE>B{vIo`b)4!Ir`8>a%wxNHW|4ra
zk|>xfDOJdOnBxXt2KDjlYU+)KviMcm`1Yh(^_U$`e>*j3bU{&nAaRVPV`QY(DIX!3
zsX<uf@+)|=eOzm8P@@pX`<xBrI_@V#E+0h$=lcX<5>oiv739K@;h6V3RhK(@J5zdn
zs{+pVp65TQ^9#3SkTXGmF$qND2zDsN<w~^1_gR2Iaglkg>vZZ}@AXL0Y(+<3Ogu<;
zDwSI=jaYYkDipNg)bSFY<yEe<tg07DGTLT~?^F9EX`L=So=8mlg+x<nkM}cd)8X4y
zU+!+^3~x9s?=#Ol<@QYWL%r5K_aqob8mjCT2%y5*^-!O0><Dn7!cSQn&c)=qeamdY
ztp^#T?Ok}4mwBMt=`%1gW{7l4tUo_wyKGmhvqAZ7@Ve5R8%Fn*C6~T?SghgIZSz2f
z*Fy7rNP?AYAr9|l!&?ese{cuO3tJkUsDBd?{$F#H^He2>JhR!+gK^=x#Wol2<=54+
z>&M`iHFe=G^JgZUC!}GC4Tk101YJyY?0=||Sc>m_)Q~_6N|?pu-ehGyR+(jK>vbg~
zTz#aav%>eW1p%!e$8kQ2W=(>Ke(QM~T2MgLa<FS`$my)-=v3=B?UJCynLMQPDmD;C
z#XYC;inTYt3X^24w<k<2&aP$ZwFDPCf^P{ah+i|<Qh<)G7Cxec2Qi;e2gL0?ss6%=
z=yCb<v9t1ue{hZU7Y5a}Bn8u`vjbsyx%9g(BXAt{r11U+CZ+4vqm=L(!ueF`%=T`)
zESW)v7FWhC)`XVIr_47>99Xg(<B*`59a+ZM@%rt70cei;L1k7(0=J6gDKgkr2LlRk
z`8~yOv5}oUAwnphKV4?{cBNN|J^Xrr4TSr3hFMF`SbWRbhw{Ahw4P#nUOgqN!Owb)
zgYuHL+v~yl{uDjkZH_B?e3F)r<dS5<!b4QN!y{(T4Cm9UBw1cJOO%{`Blw-x9{G-%
z+z9`&9_MSpo8ZiOr7u32j;M;{pwd2-c2E0t2ZN!f(yaiGc1|usIu`RViODW*qw47?
z!>Nh7T1HI;2;CpVs8wZi$@4F7xL~G&&x?#=y=-iZCj1QQ68NQ2>SIpjvX>)zp(ldU
zMnlW*aJXtBUGWHQIoJbKpEmAN@K8Eld71$%2*bF`@K|ur+U}SE=ga{9ois0TkRl3h
z1@!P%Yx>K{%--c=*;tnmORl({BRvQ6vubuRn((m!-{?e(4DDg&7PaG*;8RuGR!6q#
zFSW1R4T*(qmLlJ6o$Jt6DR%}WaC#;O*Q@l(zqspEa#U=3e@l~p{6xd6PiWVfvyQ4?
zhXDQ?$~t^hfT#Q`JC$RDWeEybg0#;CzN!-wQeh&CkzzSU10bN_Y53A@UDZO3+TC(7
z7yZh=jhm5w*BQ&f?T0M3VQ4gp#QNxaEJpKG#{xJ(w;rkj^@H}NdIEJ)H*rtI<-Kd|
zx7vo~FC7eCFk8&AIy<ztOrffKP-FTf3usJ<Nx0xh+4-vPu8oJr;z^!WLECYxMSj+j
z{RN9P<T~b?Ey5JR2iw9l3pJODL50qrKOqcTo724%6P5%~q?g$mis9m4`I1Kp<>N;x
zCx^Bntm~>*cjYIG{2zwS(pkGcg_UYJ-g~~1J@4{JF*t7_?zSGwTyA^vCl1+^<Z0ic
z)dhe&zGbFjGAR(X^~x9g-rAb0WtWNL!{2KwI-p*IQdPGmojo#giYc_0ygO0i%+}v<
z9XMBohvGPAYmx8cDe5hD=n-&z#vXYJwT`Azhy+!>w`uW)TC(I!;$n_kT=(GrT%xo_
ztFY@@tnq%>!vwQWana9Qzv*{v6D<Mb=vPfe`-Mf;(5UdEi*16fo>iS?&`Z~u&Yrb#
zvLTCM`(S0s-M5!N|DbD%e{1^;U;d!Z#y};XbDnxF%!dGk!hoMgXttYJFk=N&<Z<1g
z4E)Fv^U^?|99uwm#;t06s5I;0Ko>%Q+IRfsgKyR2Pi5Wn{e6uCD7yp<ca!l56h)CH
z@HiwMD{!jusjR6Y4wgex$CW06TKIDcbQ=2C%^v~+vzQR(IAUBJwNFCL0jjk+UJ`k|
zOxW4?v3Y9j+1f<{quCHvxQatM@?DvjCHaf&e96<uo?~uwVnk1V7L`v{`)a@pla@&$
zFJ$Z@CP@)9KQL0=RJAh$$?B--6`P9eONTQDhZqlXVj6eeec6r=G}V%+G70hFwk7n1
z{`HqzL#DeW49raJnX<Pr)cp5XBlO`fs!w^IEk_REj)<6uz#yuEk(g)PQA+b@p2=D6
zs^e;Z=Z8CZa{7CHNqQ@$dHXgZOv4RNa4-bJ;p-XnlIN#P$^hzbZJq5efu>@hf%2v@
zQCs7=*Jt2kV>Wrq!ufqeB;;bk<Pb)W+hocRkNg~^Iu*g8X<t-Y4oAMh)a83C-1k$)
zjveH|nu-(*ILi9-7e&R6zf%2>sHRVv-kT~dPTkcU0<VAQE+-BU4l-d&AN=Z}0P#8n
zmo<?fR-{j<rk<iXA9*YSRT_U$5&0w*D!mn~a+MTUu_t$^66p?V+`d#mk;*3P>0dnf
zRajMO_)PhoG;szK0Qx1HdMd$A=&}^(FrgOxvi+&Xm}0W5&Q>q56(|>0M;l+aXN_V_
z+2c1M%c2bwMc~e?{+@byc3FQ6rIWFT5>hA0<Q6+R(8zw2a3J9rr_yhqq7f1Cr?L+4
z%YF<i#M4}vY0y1ad3S`DsQ-8nU-GexO@6gHr%+YH3Wcm+*bWX9Fy=Lg1$je9^iEqS
z%mr89eZSSb>*;@2u2P;CN=OrmhyNux<QZN9cWXh-h~<|MKg?*T^A*t?yQgUWV&k~(
z2g%B3<fuV`HA_=AFIOG#vY5W=r3<`rye13bb$!NtL<I&hfr`%g;i6`zw|%>hwARlQ
zcU2WtL?0&hdTj0R>PJY3gPdnokPxS1EPhMMmC&MU_>`x@Q&!mj29s%DoTRS$WnhMb
zTdl>c^BXW~@7Y4rgB#DKa*uN<xLbm^N|nGE(Kym`P`_NLeqcBlPz=Vk8LraavR0TL
zvsr0V_zuw^qSYLl!N}8+>k68Q3VUMNc}?+158b02$-CR(`!6KBjxo;2usxzA2X4|!
zF=!F;iv_D}(^@r%k4AQ*qLhcqRePD8wsD|MZQWfy8$QNgKJ}?9jpIjMwie^UFr`M*
zB_nkC94GDbj@YE*bwWB5n9mP>W{sQB;^Hf-iI%uM0gy0m{+xjKGGeJl=QYdqcvx7-
zr}RKrfZFCtGR~q@YcJjKby6*KwwDCk_Y;@KOg{KslA@#iwprObjk{jvx;a~6cnk4)
zMjvmWF+Q2H;quAR(%<xzY`=*+oNqtY%{87nRIWRSn^!)<I>tC!cW=MAv{)XFUheGg
zZ>{Sk11(N&iMwHr-yk#|iysggkMJ)d1T6zsvRNW%5B$cDbENNXWT`<S=*Qg<?v<r;
zj9r#AzChX`5=^Sqd}PzU`5ZcAD_?z#qE4Wj;FkN5;$7}k2}8ugmzCS@LrVpb9Nbq|
z5*8PQ{9tY~pNo-cuj6ziHYxi$VcJgC(VeoL<qRKFzL`Lg9N4?PakRb$A5{&TU;KD+
z%@JnrXj5!QSr)`IegmEVL{@BN1#a&i+fAi264kL4XW-{oR2P!^EfB{In9x^_2IH@a
zoJ=rx>RmW*mfT<@EEB{HWSpv&`6aI<z)(KHQNv($REOP?tk_?ycd`ryiXrQ(%p=IT
zn+wUBq1Yy4v&C}#bjSc6UR<f{H^LRtt@IAo-%;+rq;E}AMAtE1ZbY45iJPx0GLoF6
zA3%jxZQXibw5YLp`DC&*<BwaGndxtiK{$bRQ(=+uUG?dNV7el+HFNXD_le|r(g>P{
zj`^D0AEg%*HPhp$wQ&7O0Ai;ztKIhYx;}N+P=X^zB`=E^Q-3mNdv&q0YkO$&rcc~;
z@YMU*D6prSF*sco8PDaK1~8-Nro)bL5Z8@i^iqxrLt+6u)*&a`#{Sl&3DKl6@ve_*
zXcUDH>%npu7=JXP&Y4u2hbk5aHlCmB>=4=5{xsrn;Rdi?d3C%ab+-A)?Sd30qR-Vo
zp!7HWIN5!<^ZzdwuqHT04OdBH2TTH4>x|LrDM-V@@ErT2N+(RAt@A5_0$f#Z5V<80
zSD%ahl$qq+ApCqz|MW7hY`R}&nggHr4moxwE5}e*ZJloww^T-Auj1fqRUFr%H#BE#
z5pSEJ?<n}*fw3&x?QUv2ShKZ0Ln-sQWbC-0_>tMG>8L#dc+?y4r-f@Cg63LE1<pO&
z*aP-M6);z|kQtgNCia>9b`_@P;7@g$&?eZ)8(!g*FDJ9MZh$pKzftSa6ni4`J+ikY
zk|%|~fAKV6|N4-nsX((aS=w+4K2~-I3){}g3lPC?8K|PEOu64V*4`ES7AXF|mlyd~
z&UL=v)2uYjX4|}r&Rjnmh1^gp(BeIt29eGX(NK7UoA64Q>sZ7mDblxik>ZK8G*^k5
zcxCsRbli0pUv(B9A@KYp7#py31RPWm?qWVAw%SP4_)0=HEzUOduRBBoNw|L#eda(b
zTA4*V%a8XjpbhSG7drffRDT+;yE}#oQ;N%m6t0r07@zOc&#SA|l8kFg*3nM^S@$f?
zHiRZdiXC#x(w0~cn<hNjZqre1Kwv3+VzYc;TxTQ444Y4|OFb(d%o`jMjnW6M<EN`N
z{ud;7VKC9!FueYTm|0^TS28_kco?@DH+6<auGO`*i+R{kvCGk3pHK0?h{c6BT5;z_
zby|8iRVJIQ$Q;#QB2y6&REMxAS$GJG$$T`LW-LMX+W&lO@`wx6LgKT53gRFpx-zY!
z-<V)P5z1Goy1b2!QT}PBOWQ!{8oO=UK(y0ly<~wyV_zHXjnH3v&gSmJ<2*m(Ts0YY
zIv+gNfx`me{J4?X;-_zlJa4JDR)>@|RUEOT%zCm31uu`7w~nIb4wGaCp`v9c@b1r8
zf8zF|;U4wN`vnAuJ1;lSuaCGAHFz-ifII->aplOqc@2WW5C#5lk#X$hYTpO3Rhxyu
z6vLU?#5tyz$?E0x$-oIy;VLwg_rnVD$9=?iS+K83_zw`780c?y99mK>hh)w#j)dTX
zDl5!q#)@eSbt+A=0`Bd*!>rFh!SqW&pqNz8T*~B*eoVM{7F@@g*ud=Gz8KGgjE*Y=
z3Ytmyb6r{68SDz3feG~WvD;T8e}20AQIgPmUDe{K#hijcQ!W`Rp40WEwKmHu0$yTf
zr)GZYWM<lydfwKUpAzCOFfU)dmP_Ug7f+WNJWPh#H$cTf@5do&bmRcPPedvJh99qd
zPYQsGhq6yU%wR`)e41?#^j7E1q-ZI&W~ozDnqT}q=?PBz45{irKw$Iz--L)zT<oa5
z(A-bAx433f{X<mS$f(D+T~k+p823?R=u6m7*=1thw$Vy855NC-slElNcxX@q+!r-j
zohl=52FEb$+9Qi?1YcUZ!<f`UQo$XHOtH&UeYh2B*|2kuK~l{APH#{5@sr7(PevQ3
zk+f>-SonpP!RF8xWaK}4GL_=l_xy$E97_eK-Qy}4BZO~-kH#=C0L}pb(Ji6rYG^I{
z+s=%keyxx<bm|n?#W%R|LEMj&gAjp&gA@!GQX5sIMfZ8^De#X1w2w+517Q7^<7r_W
zTYYTl53ibW_|mVXX5+C1*KGriXJHsv(IZ;vt&{91`RhF&p61`NVrJXw-_&JunUtp4
zL*g*_c{R|Ue9gG%4|8(G^I19^5p+y|4E5`Um{0UI5#nOQT0fiT?s1fQCzS}_So<b`
zvbbG?lzuwk2lOSH-g}~HMESp2C$)+H%bCX_3U&Pc<M-)fS}QZ~x-WzeY==GsBQ4H!
z*2QsDMb0aFblhsFehlVCW!1|I1sYDv)R<pO(eQ6QA}uqELdOEABZ&(JF!?G=#|GT$
z`y55l>)n?CM|s^Gs#S9ZJ9Zw9Fj*kf(bAO=om@yC*fI=YMQ=uakoK)TLWk*fuMU6r
zd#^xK>2UNTAVOgVVw7|?x4eN!wN=VGv}}1t*Wt|UBhTSp95hYO1LDc{`p?~{;FZ<-
zzcV=8<1-7xP7cZMVey~A@TG#}>T{7VtYu{4?if%&n|K#zQ>Vlg4{sMrCWCFaWnd&F
zE+5%HM(W9d2i*>EwndJc&kJy`U+w;~+z2P7+hX+7Hng%cm-_QSC*CEvX}8g!07s+1
z(UXOT#q;HD8I0XlIuv%stCPf(<_?`E<II)>VYiPD)b5J8toT*^5mTA5=j5!jwllZt
zFZT9+oTqZ|1T>MypM-1ZO4oT%+ev1nb8l}p#Vrh8S3In+_81pdW3~T|Dd}Kv(Y|Rk
zZtknW^p_4&D03WUx&ww!lOw-nA{868Vc7g+FyeAUOkLAVvmO_@HG=L%Cy==5w-l<F
zM^hH?;`x&&o3A~TnQOk1(^#A#7beT(Ru#DtOd1!j+|dR%*cq_2E=xx;4H@&g3r<@&
zvp|dVTwj278FkSf=d&4FTBlp5+lv3Q8NSTRXtIP2^Jquuy@rBK)v~?!`J*fDRI>53
zjZ@r)i*FlSYS1mJkU=@;QJcm^UFowmgSy+nna8yBCFvrkxBGinVopfWA`S*v_;qyt
zK=bHQPrXYE0rUR`J$53X*GSG`32SRVxFZ2-24DCy)r-m$j%t0abBE_g76UZSqXtja
zoaCuXTg#Eq2dN1JEDX`G^8x9!Jeg|yP$o-1$!ZtDA7@5gP&LBYB}|QaS`@h-E({#B
zcYc7U80TDEE5F0VMhJn<I$d9ETDjLbRqfK&_=CgZv6_GF;t487X`W~a{(@#ViM*Z+
z6e+8^AVpjCFAAUL8G>1cj=oHLB5%2}9K&k<-3S&HZX0zPpm=1JzV|n6=t3VP#VK6_
z?<-4L+`p#%=Q~K8_bw)H#jAPbcM7{zYs!<>7<gd!h9ennthMCDT~>Q(j3)A^tjB}*
z*D|j=B?7A~L9F68XlgDt?FdFisqCF(s?rW%%`s%0e)ofSJ7WXyGEr5;@hmyBOK!d_
zmB(fwwm6CGBjgUDQXg}5qbNq!vR-lk8vPY&X(U%75(&%l4Z=v%7qy@QkFVWu4Q)+b
zLDAn>W^GNsgG%qAqEad@yt@W9cU37@nV7L@VFB9s4HH`sko~9?^Zwr}y(Ogb<6&|4
z`5c0|-Tr61(Z-VD^*9==uU{qfYu~*=(Z>6Nj-FY+51qqkYc!z}yM6Mi0DVnt@@MiJ
zns2c;n-laEqqA*`aWsxgV*{J@hX55OK&F^=J`zL}Y@xXL_KRmR=pO>BWX$qXD2P_O
zv~RA=V1yNUVEu#E3_a9y5wmk8G7KVm)T^;i+wy%8xf?UsdMhi}7;GT=*ZzUM{k;np
z`%Grh%dKU)xkcr+tF!4fRyO8N{cLWxYGQmnR5P_cAf^@$(6eO*wYS~^q&-oljj0@M
zywZQ4U(zo+2~6)F(Y_^Gx8@yXN2bWyvnVC@jrjK5-EP32FaKWTS+`Ft>`_Elg;V6d
z`V}cJHmB7Ojp5(HQV@l0WBE*A?ldYD0&{1`OP4=8%!pLe$E%~KinA5gXSbL}2+_=z
z2niavXWKPZYWwPa0L&|~GPZS9z57ZsrkB|sg5$AcaX2)YPe)T%g0mZRo%4GR9|e&b
zk5bJ(4)t5+oig~yEUGVie!gj}j>@3?SUML%@YhH!l3jo9ui#U$LGP=$TF}xrxLtpK
zjb#3~5$L-B#F)q;oie9B+t(CJn#Zd?#?N%5zB{z<6yJZ%sU^Ln9%jUpDUHurO%+}s
zee9)OTLpbr4x><8#$8x_<>3y*-Hdp>P!pK8_a5M#-5@$<rcBxXVhA4Rn$o<QTkcP?
z$!UsAHpa~vqYFZGW8bmnqN5f=g@qas*G=0j9T>b6wl4=v_mGdU_j$m`C|cDH7Y`VM
z!uksTSCX}TefDB(_IhxB+L<Eo9M`893}REqxPbT8AXAjR0^n=p$=6E8UDf0}hz{n$
z2(e5gF~H#5c@XTs{FisWEtMZ&C~>hB$$~+sk|^}05edpDeb?NKi3Cl^$$G4!cJG9R
zsO4)|+)Hr)9Lf<3SXpc49SdiF{wytD+ecsBKYMZbK`e-rnLS|S_ahwjKY*SO^2-)X
z?~z4kGFocx-Ys|!y1g#~={s)!rlv1YO`){&yez<tdKsC(sqr*$tdg%xdFJN)gQ#zg
z6yE>4XaNfc5~-1-e6suyXX(B;N&wf$;|j?^KH$X7T3b039#m19<It8B5zx>eAtNK#
zDLlDF&mJ+sQRB3-+Ru_oh$DBhT+`E+V~zZqqR>Av)R(<!s=#P|c54J9FD_P*Owqvv
zUWH%xV-F=oYN;qXCRWc_BF{?KIdWPiaS8=xOHWgb3qlNo-c@($W==bSTFThZqEf;9
zvG4I%vWsfU6}A%-Y>NMV$p3xDXeuVmKFhVW#hCCeXEvQ;2h=T*&WoI=QXzBP^_T|;
zSbpn!o2>=u=|nGaJ>q4B9kfggI)57{-if{@Nutt|awbgFJL10Z2qavKBxoo4JgmF(
zOO^?b8b?!&^}|W;nw@m&KZGVNY`0DqfWXs33WyIU9!A(Bv%aqhb4ou7*Y>(73!BC$
z-Jx&!Jdl418Ai<1KvFa7pE4Y;s(3h)KmtH-wP^5&8Z|xvWWeF9FpZ7Xr3v@hBJGnW
z;UzSMS!}p2@Pzb?mFw4BXH?grWY3c`EgBkJM+#0Wvg-O&@+;Sl&6+9UOI1qC%o=Q@
ztAq~C#3SE$y-b%Gw|j73DN<~lT<oF7{A6@cl%F;fi522FPiCu4l{^X;6)gOPOkTxz
z?Y_T5RGz0CkV009bwm<O37>rVdxg=62N5XD#rgACEV~4y&>278%>w;wqS|Kwd??Vb
zoQ&;|?QA)oU;pPJc>I{j{YA<bAU!{7Y;E8s<{?!t#L+ga_>N-@;RTa$J2CEO_Z-|^
z9f3)}f4T59T&345lKas@OFvC$Jeq3#Bmbs-rv9+lXn{oSu<ZFPK~y<q4~{XamAxHJ
zsceLG4t+x0cE~?D+dg}F)C4x>#O>?iZd^-GIrlzYh0|>~@2FC7)^miI@{IPRc0j|d
zwb0OfhEje|a8nRJ2c7GI)=)S8X!W-6{TDeR{x`a?r`aYB9$_3|M0dX0y0#%x&K}hF
zAKrMrr^*o?`vAMh(P6DKJ5kO$T|X=`tW{F;BV5v^mp-_o@}A-`Y7KZNdyXexN$wC5
z1L;j(4!t0g^F<)>1bd|o7>BPQC@~1cv+T_6t@ApE8U1!OQvTAt(}W-=_`j=o87|sN
zVXHValZg$W-9&v7EDJ32jBS}Nm_bh{F<Vw(yJ<^`3{FCMihkPXwdIJN0u@0<>U%KY
z=@~bV-Kv(&H!CuZi`WFKxXGZ}EOo%PM;%Y%11=7-G46kT;zkz*xbyToctvnydPf3;
zHuc|80vWZ3b$N65Hj?<aK6%gX%E1U7sH#1e4*2tKXPiEc*G<X0_$yoZyNiQ6+FP7c
zx;h&lG#s#ui^qK{i^8(HFL|Y{OxIht`fnOr>qNxoo78XWr(uj34!G)&0!<-0Evz+1
znzsbZaX(s=^0wp1O9Z5Q0l?APmn1!350nP2OLx<i8^D)Ov8Pg$I9!0;Bmz8u*MO{7
z{3M^smORWwu$w>n+zncOvNXc{=li;{68di(!__m=Ybtd%U1zX+%|jC_2-Jwa-HXbW
z;s|EM0Cy8Kh`&vo*VIU@H#~wGe22UMIUtu>A8Zvk;=vL!fVOMJpvA>8?%wGY$&=GU
zE-0Qgp2U_|h@dzAwQ#c5g&-vI<fXH*q@OOAs+}GfkC%tDAtuFl_ho?Vf6D-%=<jS<
zvDBWv@HGH+)d1wdYa)Y_VE{!;uvIP2Ce!q)Itr&|yW_toOgM#Y?5m?9FU#Po0{3YP
zzCeY=j~jcBeJwghdhMgMjQ}S%EZ)z;RZbT#MXW>g^fR>0H@(E%C8r$mRdN;5{t98<
zQ>P*t_q42BwF_Qv*4kiRMmJgHy6+|ohN^-aj}~%wia90}imXc!CpJQj1xp2{l%?5Z
zOYz9J_g|mz-`6KFA*R5Fg8`EjMH~eWDFTeF?Khita(}*$XZAX&d_Z_;`8A1PSkX2y
zQ6pDb5EB}mz&DbgOGMh1#6;WD_AC4Xp>6=f$h4aAE_$9yOZ?~fDwFZdfbt)F_*SQ|
zsAg;}nEM$wj;jkN>?L;J7}K|Xj!N&AO5fL;n?M)(%6UNFMq+?kP@J=2=Z$l2OLcjp
zBr-5bxNhftTWMipA$YmNf5&oTxt0|sli!n%|7Uy}5d{V=VcjHR;y<r};V5c4ypHV>
z$)B&~JRX7p{k-84X?$a%NasPU0`w8IRTyB0)7lt8;YX&h8CnMYTzq}h)=Pt{y<C&W
z5YRh-8~NdJEblpBx^_nhHWkFfHBQXn`LwCimDvEdgRl!9nKet0?eD@eoO4sEy00G>
zT9xN@bX0Co`~Q|+m;XFj(1@r>S<F}QX)7+WPf6Ee8;my$bq4}!GvZ(hgAe?;XEMB;
zpnGUja&9-4`smC;IF+f*wd11-eAL(M(S~weDHI?sysUJ8ketbKupgTm4OD9o*AW^_
z3h6NO+ioX)%fAf_@7#Lk!ZeMexlpm^CV0sz4;_THa+*d*=(KRA_H{>%nByYQ@%h8=
zW5@TW)PGHa?}q_6FmL+O!Hs^JfI7LO8?Ob)gNA+>pg%<JvMJolz>E*-r;F{5^nTAU
z*kRNZIoYDWPj71L^{JQj$lU!A2<%e_-Qy*bH7h<g<4WN`D>B^g5$@>iOWV$ZoZ%C<
zh3g_b{h1sel(wNc9lJ;Spg&q%8cp2$F6KsSjxni>IuEP-H;{4v`3rnia-|Q;0W07G
zxITByKu{1swXJS)3%WWZ0{U{4vJ%6_sNgA1gJHK#(2lZCbRagrA>Oa4*4&SWpfpE3
zkgZ6SF+KAbgu1<a9)(x6<@^YX?@Yne1^j;cGT}F41|8}wxzj)f`V##C(;92zxkyzO
z>YoS0#-mdCcNNBKAo5>z7RW!D8{z@14+(I6ZsXe#5il;)(}8Jq>)eeNyh`#Q#@>FH
zZq{fzbky6~r8Vog`(R+L6(LHAEizXKid?-mVlm<Netdp)(aW*opHB`nC}KM5ho{Vh
z#!TzIHmx~KDi^ht4+t4l5zRDv^Rdx$vqb-9lJx&(Qp{B2Cz>(!M&M<N7FjgAEn=vr
zAbE}J_EKH@W*!#R`*+8;xJi1Sz!FYBMD9;OTX&w{O7%!ZJB8wwI}+^Eh+2n|u32%g
z9Y=HfB49P!N2bEVe9B*AW2Ptub)~G^Z7E$6lZVG3E~AATuk18n#!fFY*rEBS6zR?m
zk!Q*LeVr%&q<6j%mx!?ww+5~s#3lG^XP*Os!XJ+9(U3A2qh+$~FJvm9Ad2Fw{H}sf
z&Ixk5*xMOWpI{rSvAltqH-}-|J`b2<GJzL28aaPPeRKQ{#1`g_Q?I$X)pjV)JCR~m
zrVXG`+twXAIk<*?3bcH+P=l)<fjG*g!236NV+axl?g|Y-3oLKG0)zXai^qiZw;k9v
zU2T1zf;T=#XV`bLZ)EtFkdS8()%5AA=PuGgYqhPu$<9Ny8kA3>>1|cYiO)yZr*s;(
z-EL63WThQf8)Sp)L2D?owG<Q8fqz!K7%u(SftaQD^UGg<C8yLD&L*Av;zbQ4Are4L
zN(w6+`So<hu!0e=D1fEY(b)(PGF3~S0&?1e5B|Y9QVIT0HN$PZ3whI>=c9-Lu#d|N
zOVW#Of2)Mee^r8W0CYwe(z<2Tctk;IDP!`RL8bh`hl1kEF=@l750eV`Pl@i2Ec%a_
zkn0Gf$UNU4W~uaG`*by<h9c-An$DL$rnRXQoE%)+wB9*P!70h>fd?5%UI{jaY=i11
zEaQe^68Hp6hn#in1I4EwYB`PoqLwql6O^uW`Go|uhL<Q#Khn;~4i0U=klI#fEyjL?
zA>o79Ov}tvxW_fhbneU+$6X#PD_oB3aQ(3J8J!V4d{LR>oy*Z~iz_t3+;z0tT@9i1
ztCG@WNHmsb3I!w}|M^r37#?B3EE6uDpaBm0phGnKOmvCkR8MzTW>HO^VC&+s1l~+j
zD}#?nK;Rn+1_V<MQneXQ<9S`OJA$jJt01Er<?hO|=>KJe24Lf&uFWKM2rVCW@f5uC
z=K}=I{-E|%soj$Xk4r?!x|4VO`0iyh&>@+BQwR^VJZ`!A5*=4%%f-zO1K3El)-DRI
zENA>XWzh%gFQw{sHo)wWcVqM9*IE#2O+?(w-d+krw%Sc}Fz#U7Zq>l!u{;(h&in2*
z@Qxi9&LPiW+i^Dp16_K{MNN+E*IIDNc16GONG-0U#Py>@iEGoOp&@4{7u$X4h+v8T
z%IXLJj)bUOp9kPGbAE~ri+Pf8!vJM0;B=wls4pxrmX~kh9T<kJtqRT@kiy^5Ge$9q
z87@JWpK5Pui0Qr#<gT$=YazE~3}b*MC?pwkCnh1t{k8j){k9~!yB#tXXx^MM?uOQ1
z!bdbe**emX(Fo!}15lo>+$sq$9i^3<0%{Yj;mzHh!2gc)Epu_VCjPWxPIlG#0Cb=W
z&$YeTYYHhlExBcCWNV8lY+#x|5fnDCc_htyS?S@%RACYK+tauigUgsbOha91b+i_g
zi0g`e<?9uAlz7A$aV=j$yP_wCgtSt;Hr#5WKFyF{8g8Ir>ryu8ZH9n8q^l?$++-Vj
z%%+U#c779fhC2a&hx>pl{oz^$7hSOp?>ovjgd1qmyslLaN5Xtx`uFC<U@>lI)bLz)
z>QqW%Ds&59sd2jdVv|C{HNSG#pI2cP-B{GejHH;;0BYN}zBUm^-5)8Br<>kx9T`Bt
zAY8i1ffe|<JT$kqqC|m0v<|Iw6GhdqAmt4y5(73%%!x<B#awfd-BAm$&;eiX-SUBh
z3o^@<hc+-b;T&9twAZ_0=DMJ2_Qsn32uz{4XpP9uHugTeN6T_vqS7cUH_a`${wV{+
zy25iVn^bMV_?p8j>*kvqF;GF$i4=<W)|MGO0s<ZF_j%{5$d#29X(_4FhNO^?S6Je$
z^{Z5PW>r;H?+0CZuZDps!^yG>21W1}&>~#8>#)Q!Nz#L8fG=W6Gtp)>xGIT^@pPlD
zod->mK$Wtf=o3-~0^xuqQU>$!`f(2yU|K1--b+8_D)H0JO(dp)IKn*$Z`vtrA%S$6
zS;qqV?s-FSswDYl6eNs4Pye{KsP}t$k1r}PgUt%1%7Y2Wru*wvt=gQcF${330^iuD
zx&9lQ{?)9M^`~s-*zJ%<$d{RqUxOg;wYiHNEDCCMQ5i~_ATt3E4{lxVr)^Ei(p9!<
zJoR4*dALP}6`*+K%0B|pbFT8nt(n6URTl4hFY5)ypJ_&N)EB*pq}nZi1KR5O{7!q#
zGMC3q&U<5oxkFXSBKx*rm{LqS^h5&7R653YOp-2cPlXk^1bgH)dvNodEV^+h<MJrv
zR2_mCom%IAmXy#Am<%c$((no#$E=-iaTOZs>xYDgKZZVedkf}Yz*iX!zP-A>o-8$%
zy~Dp*Z_9K$?<*?a)5%H9+S{QwD1ynnm)7tJ)(Z)}>M((0QDe<(1Gg*rStD3q`tjG+
z^h<U)#$(bMo*hklmVHA9*2S5cJJ;qQXskj#`|TrL-mue(C8pQcu|6Jf6(~%OflNgc
z(2K7ev*7m@jGu_9@>0A2SaL1bS#n?VD*oJGGSh}_kF5N7nsy2I_OifDK6`|5hdyMn
zL_$l?pc@$UCK|TeO>$1=^33Dc8O-Pb07CR&NocCwvT4n1d$fEyO#uTW{VOcXorjzU
zM;w5ced)G`4UD)dcaF%Ge%TLDFjBI3B=6|l43+qZw^Vt<Ap_|1^|H)E4kNX$%#IHz
zaj^CWuw9KuHmvEFWR-7+#!*(gKU?2dTaLdIf-imS>Ck(-YQC+R>7mVNCL{Q+dUoKF
z_`uFp{pb@=g_CJ%3P*}7_^6BBx6v_Tlm>4{&7)MpudPVMpv;UTSLYWdff0!f3F4wM
zX4!XOSGm~9WDwtg`70Wlegd2<LhU#_P9SSNMEKc5f8Hg9x)KkiF`>5`lr62s+-cf!
zD>B<Wpn<z*nt>=T`SWc9kXdrt>H6sIo8&NnlAF97eQLuH)`b#@g4Y2w+>pdD#P=QH
zCGrZ8uCvl4`$5-CI1J8>_Csw$=E!#vnG}td+j4ixd|CeREDn=|^qK9A%Jaj64O)v=
z2vfmSwpmk+-|-}`efgUu1mcladM*g%m9p+9&*OiSXBQC-O)Lev<ii2`C6-flCdJo<
z9)((Kb7+E@x}7!1qPP`{KIdfY5H5G~+^?a*l9@wR#y=lR=4igh#&?48M+~=OH#l6-
z=Zk&u=biGE1UL#<r-Jo`?6Ub@D(DK<4D^==0}dl@J!YGc9CPW;nw&M_BtV46HEqb-
zI_TQAS&;67!)cu9gEd8`0UK9_(^n1vhoK4%hGX3I6X03dx~k-W09>p*hX*#)66a$+
zU~R}bM-&}1ZzSYoGp7TNSYX(sZ1-@9==r?_w>bOCb^B=z+R4JfXMJD<qrW^fpJ1z&
z*KJL&@%a7v)n|f<5mQ015fsUxk;-)E3W3?Xc7dz@($qzLZ-+?{>iC*^--*J?=Fzs3
zH)l#Abi}~w^>Fa{%Yes1q=MoAOG0V=5iAb?;714SUAgn!F5{bAdL(zd3}-ii;EP%N
zX-4yuJMs@UaIWT@avuHk#J98t=n(KcR#pz<Jmau{@Q;4qz=aV#Jy^QN;Ig)lcfdRz
zl)rjbKks~Weo9{Wo(pIClmjLodeYxm(oJ-?u~WnlyRS|c@(ICmX_X5_Y%K#sD)4Fu
z1N#xML;x3rw~G1u8Xumd<AF(>GY3FmG28Y*`0*}*d4sOix?nD6tzj~lyvnR{(IW;Q
z)|M@T1U#|20P<4vL>{$BrBH7ZFs(j^L5MBaUR@Gr9i_O{fLi2<jmF=>sBkgD%3~r)
zU9oTkp8&GsgWQ0IMSQi8Cv`7HiNHMzi$(C4k^0dYZ;<60D-CdIzw1--vIn475^FTz
zIQ^)dG2@)#3izCS<AWYOS*g7?7zB1}1ZJy6xq=ODD-*)Pf34WIH;TT{Seb}foR>q%
zhk@o^^8;4>Nl^VM0j2$~)?pvua<M+d^H)i|P38mp8%1^|UHRX|fEP?>O{ws8E@dKz
zs##?Pp5ItYs0WStTlO;;u)Oxtbu`knT><qQ9W7%z02ZZ~=X0~+t2^dGcgtI!NpEtP
z5HvnfyW?lbU(T6n)cIadE55Q_p{0Wq5z(>NfZ-#RR6G+Xy;;X*DGK4(!ZgV;+1a}c
zd3??@(*Z=}Ml>uzYX36p&!fLnV9H1Oq0k3K1>-q<&hX-Sv=z3bpj7~PYKD7L<-$zO
z4iAGp?%xZmC0mFJXfxknM2fKm@ql1~!HuNP&eCD!Z?RCApp<3HagFg-L7#_${S_%@
zf6CrcosjR-X)}#2$fI~yW=VlReN#N^1*kN@-wR6!J)~1a3=7BM)lhie79jxykfth;
zejU~D8D|pfvb{Pj3*+wJ+25khx`fqg)+Nt+--IHk44MQM*;}PYOOC6Q>73pUqoRc3
z#eM-psSvd4fZAoh#w6+|ninv1-*HU-$#Bvn=jVbxDVT~yRS!YE7X}@Ur7(_aY@zo9
z1^Yx!ig+ByA+VLAS^m~bo8x}p{qr?9;%uMv5U2fMyrJXTxk5|)I2%{`YJ$3l1|mS_
zN~}L><$a2)+@q0>s>fXm%B6Cd7@@-rQQ&`KQWuS0NkiQ>R*SyX?2zcov-&5oyuX)N
z1aKq5L%Px#>RZ~qLP8<}yu)#W3J{DY%hB61R#&>xN3mqSCyGf3wPGt)MGLZV#+;$l
z7+5IVO(<Is_vvcO*S^!Uu)mZlEBzAcscTC>hWq-EKoYK8#fk;`yt<lnGafoW%_bU?
zNezq+JDl^klG1iWU*ieb4jxWs(`Ai30q*Ukfh=1;xYb$bc&a?pok?Saz(Fi5#q+yk
z0Q>3V$1m}0{C<0308au`xL|LtG`tetG4azV+zt2(NSF;!|2k5nvH$wAvOEjm;&o;Z
zP9e~)`L%gHiZ@mG38zjm+Lpn7Po4E-y@92LF~5TX`@0Zyuu<}Tu586HGbF_61&nLv
zMQZ^^wG+&D9DC_qww(__wU4Bhei=F-9!q79l4(ySNtKFwKE(O>2|@a0sJx8~1r_{H
zAK}gGx(gP{8l!~ivT^&Mz6#CMjt0%tD`<<6dgnYKx1Q&FIF`Cj96%0KaP>ik!I@II
z-xAbCN8-n`Rukey)0+lcJv)jh^hkuY43;w-3pjS!+EvapIzEUQbjL^nARr7(wW2u7
ziyOUJBEuV8dVKj)o{Hv4znTVP0?x=l_XDc-D2L*~_Gw(W*tU7o{jmH4^0}umeA$V3
zXd+P)CG`7OS70!rrxpe+XvdCKbo5O4B@EtnzC!n{JtOsiW41k5KojEJXAjGy=1Z+Q
zZU0hm-Cn#sPCUb25g6e)H&D|Ce6P}Wk1=9@)VdTnZ|(qF2p~w2yqzaoPgtpPdwxF!
zI|{nHTS1!*C-aYKA>eAxwRil|hxfUAL(A|fzM3b6pihG`&=vDXwSZSGX%H(c97}xW
zShq!k!~X#6TKN00VDTcUvPzfg^TzhU9_JkM4(IS|j+h{TrPgGM9#T+}0Q(UVhKF=-
zs9hW>B{dj39zb0U3&9WqbyX$>EO_=FssUHUHKZ~1=)Fk~>fMH2DqP-!QlLSCYxaV(
zNOhi%JnBRCu$Z28EAzWu<WXYZeMdhW%A5YsQq0I$x|F`|{X3On4|4CKO?0J;xv)Eq
z#M1p|u|od3`AiA%=h&cE9u{y3vm~4Gq?}gZuiSOMm-e6kUHDrLium(F2$A4yh7J%5
zMjR%1;;x*YQ;kIv6HQk*UywZb7lIy}UD`ieB!~oTy+Y~Nw)EA+y#qM2%VS%e0vr<V
z79<}bb>K-zS+6W|eFd{!jquJoAf)&mt_$KC=yX+Xd#0eT)Xf~isGS`m!&D<VJqpkH
zEjv<;;FOm2tA-qhuLL4s!Uro!b~vZcsCzF_nVN2hfudb9y=Fan3a4a+dNIjwM=h@6
zaM7Q)z*JG1Y#;eKj_Zb07;^ImRuVG*Affogq}lbypt!h6Db55U1A~QU<!1ZDzP2G~
zeh&l-`NcPx7hX9O-1>iWjTk6hD<Qs_nUOU4NDDS4v`ioiZ^_ID@>1y@<=53-@`(+~
zM?O9{S~l)hWGgp)nb-KPr?`0XWlUCQEL3%hLm$9``~b((FCXN;r+h(>E`Fe~Tc(AM
znwR*LOzS}_pS+aIE-{hU>!)6!UUoyS8SW5{?^~j1#-~MdYx~xrlZNs3ebLVtOi%I<
zUM#UXlE+0t?T!n*lbg?s=l0u(RfASN*{$?i@q#SRq86x;l5?8imyafziQ*=F>uQic
z<6@8UMclt4OTvW#DNxh+eE%@SesrU|)|L?5`Nc*1L)-G}aA`k-K3X;-J2qfx+Xgm#
ztoDtCTIvyc+@aS7B~?A{O3IT%mHMPGCsfX8kgkY!`XOhTJ>rJ-tE>@|jJk@n9&2C-
zMg}6>@5J4(ScmeVOA!v}`}($~uDNNy?=zZ6ww;|<$kxYupF7k##)d1$!{D9m@wt01
z)GI1qXs!JbK7Dq$aXPbWl)tULNet7*bu5v7SR1a+Hdg`{ku+qqaensJg{La5XNmd~
zq-dqMbG?hGGJ8l*>^-T*S&qj6XZSV1eP1YDJ!5s-WvE(~$7Z<C#%|jud~Qx}q+~a<
zGpd(Fi2Fh`nqkDU)Lm7PtX18Vsdf-U<2iLr<%@x-XWW(#nrl0I`?T6dw9_&*YOS9|
z5~eA>xSD1rVpwD)Wrn0TwgbQF+@JrGMn9rZem)1)U*BvF|MR<@(-w2B<(zYznJO&d
zeyt}GvqsiP(xini!P=Wwd<oL@_)SNGxL7yt?MZDfF!vAXHcy_<5=Oxjqpe{;N0NwF
z%&1aySBsGYO5y5yIruZ7$;%m1GEu#&lg(s366H=jZ0`&wE=Hez6oR;rxDL6gE#|<M
z*3a_R*=~GV<e3$M>fqb9rRdn>X~fKa*@iEo-YTheJ726T7#bQ$KpS-ByvT3Wj8@EM
zvcq8dX5h#L7iH~m+%h+qj>b?ngG}T8xhxR(Rqp)i)=eViT%Q^JPjG8Q@l!(Yo6+F$
zF8QbojH|W+<O~|Fp{Rpq3W9;*rus4!RL*IbCr+W^2|aPABx4$(;mI=A{n*O`>!n^)
zmo&wTxs-Dw-~lc#+eJ7WVdfQhy@IyozxmyJt!9*D?*2;y0t;(F@AE5m&1msS*P(tT
z4mpd(HynyQz}XE8!_yz@F)KJ(P*i5WkY70FDcb5?1nKA9mg2Kz)2+0zVEOx9nLbf9
zKx#LmA4;=F$8*+J@OBn=vPLs!pBGoqK{1`%{a|x+a==0x>xZD(8G{*5)*+FyyYLJi
zQ}L!}KsLnH*N@$+E|+JPq=&|K%tHC03R1Fj13E1x0;T$yMtXVOXuir7-9_lzFEv4=
zmEFzZb5Q;8M8EaExrB@xypKbEwl_RQp@)^2hMOZD*indo(T-fcgbLq!zG4xU_^LmH
z#>dKKoDU~{*7^EyRB3u*HYuRH;_4EKjIzT!d2XOC^jm25Cv@!hT&8bN%omg<%2hgx
zf4Zr{(b=^<>M=l8T?_o(@yC(&&$q5wlG7n)W;clWlP#P_+ppp1L0i()L6J=MJFGp!
zoz{4_Avd$s<d9E5!4-Y6EwnJ1Bpa@C#?%n&BJL7~g5NyIuiA|}T`db~Tpu|CTfjR*
zhY}kX1_$UCV_*0-jf@4gWYVjiH8|kxLr}r>_wYg%0=H4S9s>q*_T#nG-4m4`VM3|U
zN?HF+(l{4VXrgn7J^rcw@*1q)WKGmn0P9B~E+KxpDq};xxqSr4V<Ir$R#-;nd+VN$
z=2BHfJw?gs4yqzrswu+}LaBm0MOPUnJ#KzD?(~hCfQa?Vw8D)4M-swV4^6;llh~cr
z$onTNm_EZzMrma!ILjaz&W$K=4Ezm+OG52!W1n?54Sp__3E?PTBSuntCDZggD2Yy~
z6$6#8qdqY~yVW%C18Y4Vh<>-DYmM6LE(>XELd+v#v-4$+f$T+sZi6FD{>-oYUFFz{
zWAUaV7xe`JS30j%Oz*P5pmj6wF0w#{TmNQ57b}jkl3}wyp>jeQ(*bpj$9~7D!`;oe
z;jG`1kNZ<Kx5a-iuw+Re43>ePtiN%TPKzcH{bweSjlRa1SxL@;lg@s+bM>-wkb||<
zmT9{hk<!sAQ=>VmVfEo~;+DAR(&N6Z^Un;f9+fzvfs<y{CbbHNt%NilJ>Nw=>}R{C
zBHry@3m8Eayp+)_@m|Q-BCN&5jD|O(46I4oFF@#%Y`el2(3W@V5t*H8W{s*~*<q+>
z*WPrt{eqnb*0FDA@wz_cDy6%s>OFq^dWLB-h1N7UoL_Qq+8d6w?3jaaxJtcKny%P~
zE6QaHwcW~z;_w?>!a=!wS$tvsR-pkw?ec6)8JeWM3b~~_*dH${ajX`>)Z7$kk<dS}
zAVs$tmgkZ}VmZ6w^f}FIHX2ABR3KD9Frlj?)DXX#IM^4JunZ&u0o9}^4p(3sXqLaR
z-y&5liRIfjc43fyw^IL9+VsCLrjBEK{=Tklbn~5Zp@=1nj&6`L<6JYxr$R_Ek&xui
z%X#SJH63a5KqQJbbF5Uh94M+YwO1HUx3U3lTchM_=`z{t?z_!)o3}8pF2`(**08AK
z{D=-^UYnFXQAEQ366=@*4q3W0f>bv3(>siQ=yhkWL_1dzmx9Cp&~%k?Rdr7nL_nlL
zIt@Ttx=RG4rMtVk<48-3beBkXcY}0ycju++z6YQG`++a~Zk@gN%&b|nX7<zptVXGw
z&H6iUD)#MJ61mmp^8`rAwt-A`bq{SFK#N4h{)?DQ_}d0MTa<SAY5vL&=&-uXL;Xjv
z+SLu(SbYqZthY68$?@gsmBm-nYawv-N!B9&2**IcL{xvc@)RBSf_X+cUw|CfZ;D>I
z5rMk9zA20m44oct2r+f;I#>1Ht8&DwKJmWaoTaZ2(Ww^qJut^}ais5RyN9QMHRIf8
zOaHf48X^G!0ft6KONXolMR|X>jdQ>4>|VN~AaoAN2rquw1_obF9)L8e(e+!L+E&`g
zpscKO51wgwf_N-B3&&YbS7q9W;?PnoM*4F3-lLT=92?K-@<k>#CsGX;(ys@@Mk5Ri
z)MIdQd@c%D8!%-qmxa!NwsZW5mbv`{{aHeON!6w+%%P_F@gfD4c82yN<y3^q1g{qn
z^JaTI{`sn8!jLiOKFlwtw1)TB*;->jS<jl@T$az4y^8VCpAS9j>vE44*7@&<28IhA
z;AFXF8<K~kj~RHB?ND{HfhOa;v)T1H19m^5(q2>6<wM0U^aa4~V{6~VH3#DB*RMa)
z74cXlhlGZf_RGGqyN`Ndg98c|SP2J=UrmWXy$$TC8s)p>jFbD^>Z22-{uG8Zox9dm
zYPm%#FGRw5iDPz)z|ZnDD9<w1&=H2?R`$8NZCg%3t(b{WbIy+ads~$Yn^2eu3ZZLh
zbA}(o)utRwSa$*Hza#C-W8Ld&e>nzXOZRlRj`!>ApQ_ahxuJE9=O_LWIR)pCOVKe?
zhSoujqyaM1-~l$;Vi_0VQaPpJ^71Z0aUmvl8RxvoYM1aQH|2vn1EER&*w>2Y)x5aY
z4mizF==J@Nh6@{Yt$vYLay-A=G(_0h**Uqnm%&XQEC3KF9TpkDRRwq6w0eG=<9>Zu
zQC?pDuCGQWw=qYc3v)_EQxoO85-PNPGX71u!y_Z{;uDXYm@}yMf&Iv5{`ew!w53<~
zZjmSLd*Bd##NPIrq_Zduyjt*VBfKJ4?E{LD@}J~fn{YB0t<5b3vhNRJb>_Vd_Q$+U
zaCH1GfN^xNaN~5D&A{sj()rKZIll)gh1p#1USiE3k-DScOG2Fa3R88YAw{`#O-k=~
z7!<f%v|BC@9yW$`KYf#z_l=EJN@d)1oo;-nC9kf#Yafrn*S@qqWUHttsM9IwGks^S
zg`dkVisA7o>ES{XW|EMt<g(dxG8d;f(5WZa2+$4RR#sJ1<qn<s#oK*UTvi4XF6vhj
zagm2xot<?Jg@=cK^X3hVM04AG!|{XF=+Y|su0IAxnqnnrG(U(8cka^F9=4!gWtCdE
zi>G&OPhWkuFcK3|#wus({SnU9Qy{{!$Kv`TvgL+JrUa)}%XG2=dq<Y+?b)U=p%YNl
z58P6STD>6u?bb`vTNhhbH`57vB}t@S>c27_!*jX)<yZ{qjiU5)JT7vWjjA_i$SDF|
zFTddiR5+b=_fH3`H%b2fkYcb@QD3CN9?zb7vfUTo+Qpb>77Hmf3s-l1LC$3EP`wU@
z9dW)!bP(Bk!Vyk=F<-Bso%wNU={nmA?E6$)VkmsO^URkVwiM{^L67_8%`b8rAPS$R
ztee{&lSB3B$-K7{JqLZHOhSESw0gnBd0V%%`1_y6#N6l2^U#tX>c*^uH^&qCMOt?_
zFtEizqg5w{o=I0C+4qf9?lS#m!$I^LT|KgMS3B=y{7+V^N0hPg#|sv3QUP~VgN^>%
zl(|Q~@CLpDPnD`)AWkv4ddlcJU}+j(SP4<2!6ETL2kY!yU0^($zE-w+Ten!{@H=JM
z$1j>ldcm>O_MsuiLG`9DG(k#~*Oz-#zm4@l1FgY4v)1?U!}`0Y)tfuDFQl_w1K8Y5
zRy`w*yIj8|xTdSV2hyKj0ak(%nlA~?#B7mGHF7w2(4KtzzI`^Wy&v7*zdBsfkj}<U
zwY7s#9(<wEq<aZ06u{3rA`KdIUlIEjlng#GC(ki?pz;`(%6VHKy}X|C`(|P!S4S%!
zqUVA@C%L}aW4n%)u6L2r_p%DTp3ChY=P-#y{cl)C^<PW&N<eqp!BX`WbsFvXBJl#b
zWvH|j<7c#s!90q0s<BtP(MVRXCTElHUi95*Ut|y%0aUm7=bcG@=;&f&4IjQ^uxGnm
zP;L<<s&rvW#ETru6QN3~+E?E@vQhFA6X?C$f^gPYQ-*u+`-Eq!Y`sVr@-KFUj~;7V
zEgyj{U00zH0m`034{9Ts)4RLu-e}V9(hb%^%foX)f4tKE*hTgRt3I-e&3OL}VhmqB
z;vzDeIxFelQbh*$M3^k4qeE7EbF?C!;FA*^jnY;KaMwW`QRI#I9n3Z~opk&m9fSkb
zlmuZ25+Rah)s-S5OV(pB7<Z-|HQqI^aeLY|r1KnselCO?HdsX(U#h#A#Dj{*0k*>j
zbJSVWRRPq@-22POp#?(GyEa{$o4Xgc77Dj}V<T>sd$f}l%C*Mt|1GDY7|AXEv(T%K
z&q@ozN2KiS7oJTetRwKv*zP*L=umSPav`5|<(AZxg2L}<_7cHMJaL1KkRDq*0jcd6
zI3ayc#r8*@%eYId0_dxR;XPk3;$G3HqP&G?Giv)mMNK_ue6T2&fMS%sbHG(^g@)G=
zx3<1LR2K9nmgnl?42~}0k{2xBz$#3S{k~Sf8Lm;bC^iM4h>l*aq?FN7YEX~w7nA-H
zTdEp&k*-+P>caYy#_wjDRPwEcDBtq{xV&G5l;6oPkg{4&<-}DqRXTZBkwq*slyP7y
zQz}^-_z^4%pmHJ1FV5hUQhHZM8&ae$J1bHj&zrASZjTm*tse`_<*juGep)8;l0$0c
zd&KFkcQSGE_(z%V|9bH}w?j_yHL|@L?>!F89e*V;m{Eq-f~7P^(rb!kAlw~;$RSL+
zHD17>MB!u&h2Z*KH+*7I@?2;Y1d)DTc0Avz2)fCDLIEG<u}`Aze^tZTyU?AjAdn-p
zQS&=xt~+o9Q9i~ce?5HAIHEYY?@32nFnj83MQxzG(uKyPF7ZSlIdIH0V<KVX%xt<;
zo{Rkm;%L?mt-ZBTF<+PmD$au<KP85NqZ2?(IJAaSs?Sb9yC|cR&>6>djR{;Zf)jDL
z)b_rREdX%LEs$NyVgf!BW<pwePi$lXs_*sdM}OJcE);A(AeQ!cqqsY07VnrNMYsR$
zLHAbhzLEXpKdaGF{yGQGj)S8Dn`QR;s7Jl;zusiypWxyE&#S{;tS%^HOo+-8o~^y1
z9lNYuJT~TbhzCOlwMH<G&R;x>l)3e95hl$f=LxPRb25Sc+W9;3qw#n)WfI>Rdis)x
z|47!(8^N!acLYvWs?;%}7LG{6LmR4(>hrevYsB(!0Y{R9{eb0#Nc~J0Te38m3HRu*
zcwm949Z*zc>AL5qV>(?UR)5dw?aGrIaopWq3om`otB}Jhe3I-FTg;NI0;o_KpOt8z
z6TnynwfGEA2WKDf>koy=fP@8Zbf4{ztFSn1u8Y?|0fz<#Sv>;C6%bx~Vts;K2Fb3h
z9-a_;TM>TRho*ftF@P87>Wtaw@38oDpy-YagqKmmr-k{STJj2e;&VXViVm$xNIbB=
z6eGXi_)Jq}=S!-jN+%7^OY%w`=e}jZzoo3M$KjvYq>kV>;(w>>%0>fT5c0&R$Q9_J
z>(NAX`C)62VK{|T^;l&Nw6u}dOFq=(DuceJy_7}vpy-9|Roqboy!36=g+@Rw#wB>r
z-@uSU6jt|Sw<BADKNEUNRgkiA)V4H`9@(-m@Q^rCaa7uU2;G85W4K4xZeOy!c=b8E
zGLxJ(7-R!J_DjQ)XXrU-dzvt{56!FK1JLPM7}mxEE=6-``0Ub=`Q`9cYi!PXBX+&U
z=})$sL$o1jJUBvfdZbs@mw*}11-&nKgx*fa+3#Lb;|VBg&-^lK>CfZY`Wt=b{l1YO
z@3#i6!|vX}ybwOgy?>7kCeJ^XZ&B?`A`O@mzs<yoG+DENSbFQssI+WfVC9!PGpsfj
z(lEE_G0d@Y28WP6Q_5c@bru-<AZO##Ms&TSZ{6K<=I#06nEm1R`+!nQ`YFz3QD!0S
zofbjLg2OM$$HE%H4A5%de$jdd|J?MKg}W;m!sX=1$c{$p%DSsqtol@ZLIEyuM1WOc
znM%Gyj{e^|yge>kV&LI|+Jl4fJq7=-u$#YyJ-6`enfFWC1b3^7#3`DEu`5R%J;OeD
z%W7Re?HJE{Wz(^MntR4y*|pKcV9Q3Vx-^Kk_Bamh?odc0=R5R|&N>%hUzzgNsdFJz
zHN;127(38eE=bd#TB|`)f5{$r2<3Ei=Mo6#l5KvTWzKS8Ag6FgJd3g@nHM!V5){0T
zdLDjIx@w)%>QcHAaZ&dQjQtr(K`k5?oW~!QrQuxnBLNO@1^^7^fHSrWCjz8pu&_s!
zXBK0;5l1T>|GQ0LMsl<%*YJq|ygH75gCgTMDwWyT_JGIS|MI#Y3%>t@A3tN$!sZM+
z#%S?l3*0NQQkw{2f2JAV&YY(%H6@&ONc9LnF{zPuz|N5Oj8t$`c<8(b@atPdip4;R
zwFzo+AoJxT1wN{~{zw`*1rsia<&U*nULw@CwdbGfpvuZN{}8nZZ!QF??H$fr9#rf4
z=n3VqhT#AKcA-6@rHl=HLPug@aXt3AzOgsPRA4YIbGmBpuH_;uzGN#;X(Eau#S>bb
z+mK%^HlhhnnttCf?*2t3qyqVCEAaEPwK=XjEeYr)TwmU2j7<&IA_l(W3sAdVwJN4d
zQYP`17zj?;9DI%NAJ0X&zfH>t9*T8bm7nQw--rJk2Gr6S85MSY9305Mr~yeM_owzr
zNUf(2N9L%h)62lBENwvpm@0L8Wo6~q2m|)y=Zs^1nM6_PQM-N}=-<fy)OsEnkfiuC
z9`*b1)m@|wt%eCNeM-IeR)Re<z3QUOBtDS|&%-=x-JF?f(QiSs6CbZB^%Tu+nP9nJ
zjC}uw*4cSHJ5$h?B5N)-!G0VR&v!2BAkNt-3~h9&HdxxXube<ysf<82dDr(<@=Njb
z>h_+bEd;IV0N}>Z;OQltlNGmn+lyvJvHzEpA*U;wk-BFpn^fp+vFVDzC+oBp++O*7
z#kxAT=-4#hK10FclDmOY+DD+2w}s|C(=qFs$8^*iGGS+98s{sSI*;3l3Sph3n!P|#
z12^pwQhvc9-j-5g1wLrtR0PF-+%I4K;bVTG^)cklKY+0wbAs)KeNSVsz*{eMy`6ns
zJ30*$4@GmFZ7Huy>;ZZG+4`Bmm48s<N3!%kcJ>Rst^HZs^m6dwrowJ0WK>N<`|nNd
z#q6i=Jnp#GAd@FB;^D`a9IwxwSi&P(Sedml+KbP!S3Lg0&}eJNxZHkB>gmpky*uF6
z2#BzlIU*(XzJeoUH#3pbO#hC-%t#T1>GCo%15K$Pb#*+TpAQ$z&!WJ&Y7>hyTtqm|
zk3x`@p1GmWdor&8p=`?R%7&Woj#hX!^P%OTfv=H#mUhQRAO7YHFcH@7t2e7=4gYP8
z=^W+)uv#4Pdayd3dh;e}l29#WSMuN60yf{y5OHb331?trne`Rc<dIwzz1h9;C;A;0
z=tm3pnVz|UxBD3XgRFu%%Gh9Hc4evF-<=JNWo#tSbTcJ~b%{K{3Yyjmu_dn48%z|y
zd$K^(!Vvg4%`=k`tDIlTT6{`;Ta4e~@ab^Gu(RK0w)~0ePfM-8Omh&RVlI&T?)npC
z3)J!d&X`u0Ydx}kP)F+;%Z4RPMTMUe|D84`Y}rNg?y7`J|HQpV#yxoqdqpsZp_R%M
z{KhXLhwrXVL(Fkp4LVh5&fFI@)>J|Bg&6a8*R<lot!kDc`5Ya0^5yZdM`(j-&|h(W
z>4&k_uOZ7l*Xhy*5a0ya?Q%lk6zLT;$5jR*B?QvM0Wx8jUUCfF$pu6HQefr5ckrd9
zCCMqWrDhI(mNi%9<44+Dh76EQv1A=xAo-lyK0wGylf&sjhX`3l$(jzZGdJLb%Z<$0
z5cZ<6)WZUu86uvI&l16>EbS>QRyYr47}Lo{r78XAau@)%ll;eAn~R?ap;wwr$iIVn
zre|Dbf6J#`;bAsjiQLrYmIeLt*#%Nj0Np_ig+p%TLn8s$Sl=WNb`HC6kkad;{nP3@
z1s?SsUn05=Vf#esu)*rzT>ANu_SC|}%N04$vk^KIiUrA$>76J;lkA>>&PIalKVCIR
zrKP77vO;;guBPp3f>~f6qQHj`wiaKyvbkBqRfT_p1x6j*bU#|K$Pmm^K}e+{JMXx!
zb3FTN{qu5?CkMEBx;ux$x#j8DlWRT}cn=n+^axV#-}C>1Yi;@Jd4Z;#HX3~0IwuMe
zTt+=Ew)UM1j6Y)hAt|#r1Xe-KfYROuEN8_wiC`ioy|tHE+==m}MQFQaVCHha!?cLY
zQQH&gsT4+$RL8}H$aFJ=We#cY0=klCzfnuLtVVsxAnh-tt*aH#J>5}3&W(PJuzXY1
z1bO=N>@3%>6o)8GpYv3?IB-zkXq^%sGj;@a{gt1DO?o`3<m@^01#o&=8N2eNT<did
zDPcFGjZDU-5F#=|d#t@y6woJhDrzuVkSGA*iXdH-U-Jfe!?U>qbbZ#{J>FyJ?&IdK
z16-HVQPX0x<rDb8rUrBm6`-c{F$!lT9`I^<Bpe62zJQ?bQ0w81G30VVDyI9Jd|R3b
zHY`5h8ETQ$VLvfo!Jd4(zeL4}Q;&Gv_ARglQ`9fQpoIST$xlO&2%)7a5v;C=d0SJm
zfHFoZmg2>fG_MT}n23`59~ZNy5+vw?gUj6@2t*4P9dNMm*4R4|S;F!-Wd&rQa&9hX
z=C62HR`0r(Xe}pCGsp(b(a_Yd7#0P)Dwej6J4>D5l7I`F!oe45WE!_098f;p^TK%L
zE(J+wHH)DC%kTGJ|IhG2UD?^7#fq+#fmqjxjlKKMi_qGkksl_IcwrEL8LU3ajulx&
zst;Bj_w`8`Jc*6a=*)TZy`1M%Yq68%aAkTG-S_v<g;ZEaUUB<d<=k_HM_w~Qb^*|{
z8EaZW#_^VZc&!3ez5w)4)oFRWpYEjBM?_+Ip4v^8I|n#^%$~*)-#EQCjvmW;1U>HT
z(<T1+#yb!+Bf^~k1&e2hE)VS`sI9su`y)$?4*{v6M#vrcQ|b+2_rEgG&~2%?B1Cka
zY8TJIL7zER@v!)pB9>`<VMSfo0PYbYIln!s2QD8{TV3u-_izi=jLIaJXtU2$>BILv
zQ{Nrm3>ymLyjn+p4kADG?z?O6w~X|Si))~TcTabdPXBHjL6R7Nm)yi(EjIni@MrJ4
zGb<UsxmyJ2H^Fju%Ks3`o5?&jh2?vsYW2>v?&#UdUGWfmg@I+D&2qbAZL7$w^?f2z
zFPL95P_{g5g-g~&sGfC-y}{k=z+l{6mA{37YtE=dj+b9oYkfFi4k7^`*+ho8YpRtE
z%m>u=mg?ZEa*jqhn3AO)9xjQtBAo2co>>+H5K{NiSI*sZlUGD%4=hn$9%?_`mGL|m
zXwraB&}5+cMRCbtcg`Rsw!n}Z>i3e39lZn1KK>jUoF%Hk-udBJD}cVF%=ED*{@|H^
z7E*p;VeClhzT+wN=YX*QK7(;^?KDJX?qSMSst2p#im3Z<uXeF!MAmSyYEAOs4QtNK
z@oEU3RHLW{75;H}OPRlx1C+)kFEpJedq+yt0x0$)TVH#e_^GaWg<u?0gEB%Egk3*r
zp`DINCZKnQ$n|d(YhUdBo7gy6!QgeW0TL03m#v7L%@%KW(4Qo9@1t&Nn-CF40!@n#
z!SW{M<sv_WKRp#>T@L0B)gvWWJ5j@0=)~EOnP9HvO6*(V>RV_5ViE=nV#VweFry#H
z&$-ngT=Q7}>brC2GEHrOkXzEFN1DNbiJfV$`ME=7nG0PbISKxFh*~JxY5Sz!hc8z8
zQ_1uoF!uk>E^K#Z7809ED^hP}s2TQw-4my8Yys93b3PfQ@gt%oZ&}_O$OPJ*J&1$U
z6vn8zZEQqu|8M{<+#k%Rd!j=?v(-O{as<?Gsuvp{hnW5mSc;DpbFOqbQy|vh2t^<f
zQvmBKhnk*D1ecpRy;1npQIMcGxs-G?kb>!H`HKy`&63~>X`ry*(%$OvH|NzMyYOS9
zCf0EH?laMwH68QliE=oZAO_qrw(_wAEou>zF=D&d2~nw>Rx8gp7y|oij%gL-3;Db-
zNy|$SnCZMev8hf|kpJ2V{Qg%zfhp!*?mpa-qk}pl9H@9o4=#&z9cno)m7@~Y8a+6W
zjn^Es+2b8IXmx)HpnlZR;PKJm;^U>o(nnvOP6>@5=VOC)VQeitX5W?MtN`AF77h}0
z1?SdP%7F7tE$QH#mw>$?RJ37*uVL}5<&p@p_sjO_?ov|u;S|$O?1}vNFWC&q!SU7z
zXrY-St4OA-IIWh*3&n%pDCPp-0IN5C++pt?t~p*UJgU^gYr|A44<I7Dl;2(lJe+n9
zZ2n2z+MP$|YPjk=h`7G8XoJ{51w);i3sTba0G`*(_TrrGu)IYXi3U?}Tn8D|*dl9l
zT|vDVv*mX6on9L7XM^)DvyHZJh~Bed-X7eRvaQFeV4%53<@Vjau8L1=d_JRf^x{6x
zZ7_90AOZ%v(Bi|q_DMRfOWxDy<0YB}zrn)+8(Yx90wPF1`e%q*9slh(#z86v;ummP
zo$|EFaYhGW<wL|n1RSJ&IQZy&*0WwT`8H;iR}@c|G!Z(l9k!5(hw)gAxmpnM90M0P
zfngp|U;z%x&6Xwp*GAvKP}kfPHscC{zcbO73HBvAnnUd<{KCVPO1&K#l3twQWofWC
zmXhg;)wpsltJeUEiM=PF|85PqpRTmtBW6ew5C@A)VEYTAm<nO_;oy%mMoj$i2KD;j
z28zeeGzmlfkm?4YCW%|nm7QQmfrjq%#VqHX-AvW}?%5>y@NvX?hmWbf0yFh6ZO8Hw
zQmr5(;iXxCO(D73w}0bEYuWw_BpeBT(X`FSZa=lJd0i_N@`31z356J<sl;ZFfaLMz
zceH%<gO7?0vPVvu!BKA1`SJ^DY766FfnokSPn{<KXtYn6;-`##t+C^7NF}C_knWf2
zSJ|B$=-dpBPr6pLfv(3z0n}pkYw8XoGflKi^oB+@!j(RmtkD4z&rsGb=s!yA?fx4J
z4Ua8RAS(Z^{dG9Z?5s8q(M`WMSZbeI4sZEbI{2vij%$r>C#$3Ht8B*r)lX4+?zR0j
zg*{F-cd2eSu;U+$-!#p|V;NNRToFYVhTv!sFi`=MkCic&i37syWkF|s1UEA-(;`+*
z_xjb%Bl$gC1n$NGz7C7iTlE=_P23$E`#s<@7CeC)xRh&4ZJ-deUqJRB;5HBW7dBqW
z<<_~~(Qsge7Ld{9@9rne%*o2^&zYZq#rt~ZYlXO~->Tb90#OQLSKl->U56f>(T1x|
z_<!W;+#~RKP=)Gin+ZonyW?KBUE%JW+XSwAzOIb6v%mb3UF~VSJ?5t|3)GqOuX)TU
zBmkFmQXH~6AhF%QTF!_uTILq(fIX$l?qJ+&7eTimJy?RhiyT_ou+Y~VPWd=LdK>jr
z2kN8WjE^0hXZeb=U3=1ps|DeQTz;Wj>JF|B`hhMGI%n9j@0}j!!`hn5M?$*qS1o+0
z-trt3KP$%l@A-gSR*MOeScw2GI^;pc=28<M%!nQRomDB_uZ0%J&W_GhAhiQRfv9ME
zLaLJx5AQgNwOMYMq#*Sg+gWHcmk-H-cZX$*=7;l|Kp2g(8NmqHVVrM2Q-?q_P~ifx
z?~!s3MNCW3)?3`R@~E)e^2CZL%@yiB$Mc|-UGE!80Hj4j@jgpRMi>lxU<dP-iQfju
zpbn9=YaxoV{Eqo)#}I`|+p4#5g1js6fJB;p#l@~-Y!Rv+PY~a;ui{O9HM+$=Hf?UY
zmRI$eDdBBSLW%<c@tqlwUh8272>*AXD{`jA1B|C02=|~cNl-|Ubd#VhGsK2*uq~4g
zX5YmUP^`wU-n^*EF7mQK;u0t5XVi2YZ9_|`>SyZ3`Q^wP_j{xKP^-=dwlw0;(BqP|
z_4q$$ll}Fdv%zo*i7uoXn!h3kz0Be8_cUuI<SKp~6l1V2O>>)W5*!1E0WgGsjN3TK
zXDZDP<sb)jdkDb&zn0)D4*+0{xL=x$)>+jXpax?7;hT-PagdW(yR(+X-+~izR>7-~
zHvZA?r-Sv0w10ia&Bk{7Mo9bK)2aS!;kphlT)FqQEXNW55|ce5Zm8p-Hd<o`W;Fl$
z^mT6<e@!~NHlYky47^e8yn)$0eqEDPh-kd0Crr4#f;CrV4$DM^2AtjDS=e5mm&!+U
z*yEvnsdPNo46%RRwt`D}Bdns5^Y>nMkD`np6;km9jbT~h^QK#m_apHYsnH0zirFM<
zOChS5@}W<wyk6K=<!q@vIrg~3sl^fuqU?YN|4a?vN|C`pZedKoCU;{8r;h2i>m_6g
zC57?3QSck<<}NbmJ&@M5QNGCHRYZ4837*RzM4$g<`2H~D9d2twGMS3WeMdjs?eRSE
z5J^%>>2Z}m2&03vvxG>c2~n(@<t0@^=KWlV!otoo`CoU#S~zJfMG1!pBqIp+nz-9{
zZ7taMGByDS!8;7!E76wCQRqoZ1-=!_A19B!snPb81xD;lX9$;X%Xm8Z-HQ%MIE8oX
z(unWuXu-Tp^jeGtkluJd^Ao=s3CzzA#kMdrX?a9@0u*s&5{2ZPVNuROcp5X;IJ1$Z
z5CXvQtu0NY2zdL?<T`DYuW08(_*86+@^i6ooW?CBUWcr;xPPL?U^RL}vQ)hKR$z;;
z*&zw={|kAzyHMe+*xa%=Id}H3Cyy$;(to1IymR{=q_p*mj=jBu6Q)6cFVB8qI;$+b
zwsI>vR^si0z}wM<$2&MBL1CU7omYz+@F%I;qmdfr&Sh&QX4gx9W1jN|yC9kkoVsTQ
zp=)R&&qdPLqw2f=y96iY+oaZ<5b4unGvF`N=&=u}D+j$tV4hzt3tw72UUa5EePs|x
z4ELQsCJYfjSP_EKqK&E9$k8#r0g{%>38fFDjTu#XH%%gQ-Ah2@0Nk?83v(RsAZ$(k
z`C%V)?e6@lh5;>Q`I0`jW~A~|_l=r3)m*(TFU+G3L4c^rl_pH|Sl9T@HVydhdoNvy
z&Wg}R=w_Cm+E8XSHgzsT#c^snZESW~N_kh?c%MV;`mwP0snoaUFFP;ul}!@HM#OqI
zQfan*ejbE)KkBV%lcUwXFczy}J7gFrJN#Mmtx#(6Xccf<xoQ)$8GTj+T;pwd4w<Su
z+g-2BbGGNwx5;!KyIFu#0-$<C`L9dyze4IOIlO$u#p}5*nh$)svItGy5_{4bQIq-!
z&M&g;&#Z9%W^nW<)rm*uB8=79SlG)O+&GAuFf%G^{$zTyQwdo>N?BwkBBJu>8T7!d
zt~pfe3B-cCZ&}KCI9<Uwt<AqteacT|;VpPm0CnHUwGWza1bEKdRAjTKO=qts4BvSJ
z*wOjx4)M17?=1xOWk5!Ai4yJi-1kxYUa^u{VRJ{{af|D|yV!%+BBges@OCw^%>u9A
zDIuj@Jst9T0yb3d*>OIn>U|d6Vy{u|TnKNM$?mHEseXJ4y04yAD`C!5bcG0Ed%OA#
z*c|c)QP#t+Kra|iqM+Ar^7cg~{pc%|b(xcN^}S53=H3P6ys*+!7lvUUQID8Y4fd^l
zK1W}0{@?TcN-c!jz?1HLKa1E=xi<D*Vj%#)ix?Et%sj_A>3zQ@#Sfd=9;R{nPCfY?
z@5hUrpxs9usVs454G0G}B3%~BL5Tgk7YL;F;(%s;;z<PA?MhGqma+uJj#w@U(_zJ#
zxl7wRHt2;u`pDTR8ecov6(8n*TReeFQ{K$o0sCyu?1K=TtUKTw(u7PbhN1*kHX>`)
z9h{-EF>j1@*oF<6L1C94y~D8qsA;1f9{CN)e%!lGzo(TlS*az8C5qnHV5bExR#K`Z
z1QaS0y;iIW9<H)HjUfUha=KHvw2rmDh@MPD(!{D&^^2LHl+0j9KvuW6$F>9}?T7zu
zIfh0auNag%2j5%TA4zQW8}6~tGWfGmJzS7Ys%I)DcVB`mC+uaBo@;Xf{N5hg5Q#WX
z^j9hP5tZi%{A0hMnzTt2$c{Q%sD<pL16ZV_GV=1{E3_i<8-s}~)v>>x*V~$@<dV-#
z|F+Qb&XfS8=?$LVl6wC@w6Vy!F=Q5253b;RlTkiCe%l9ok;+UQ9=pZpP;^-IT%rSu
zMSn}U0L*3w{xO~%gN@x_KM(R}N>5E3+yVNfKXYy05Sqa|SVWLNr!HcYEaGn}RevV>
z9!q%V(ZPsBtl1R1nnpiBviw_zC{6W=#$rYNvu47$yHN4i`YwTQaWo!902j!O1+{!a
zgb}j=M~tp~RYyD0)T05_3iXZ*YKi%?O@t!#uxVxF5iAQ4`pik+EDVo@q{Tx*sHnKG
zlej-V>J<X!Kk?{7sN5w#_$?Ew9HVIPs34MRv&NEC&XhqLrSapa+Rs1OPWa?jp9=}P
z2g|R*gUqm%sIy6%MRvJefoKF9o%;ahRu7D<QH`5Y-nU^bMTRkrcV?5hPT^bsrXii#
zRLigOdEK#k-Cew;rp{;o*LkpxrhlY6@bem=8r?dB^@r9(PiVZ4Ee*|m;ESFq3?`Gl
ztIm0x)dr;&N@fBy)gji{h~w6ZE@;pO@P+T{@!it07OTdYWG*$$JXMJ1BMu{M4#-0d
zOsM|Qa?Cd>QrF6pq^!li{FS{<M|cdnPBFkcF+q!pv*r$lX(2Ky)iVU-u3_@30YXA&
z3(*PK%qo((r7VIG7`*qx)EosKylCpC`+{r1n#>HpKUP+YgxBGdW-prRoFQUt*-r=w
z^Vkm?{ftggTZ2euPuM@F0XN@t(4ZB5;A4pM6O__<x+Yq9$r7$i1&%WDouQl+YkOC(
z>UV9S+ncZ7zScw$&}bI~&moXyzP`Tx3JD38oQ^JRx_Aq&n@1@e&$Y6$5(yQRLO`JP
z!Ua5_xnB2Isu}~Y`<~HH9${qQCI0EYDMAxBp>R~%Vinz_IU8EqaBNbs2BrocqC2k_
zBVSqzo|hdGL2&#hT=3cnyYBXS^Gc;~g!i*~u{AHJ)^eKSXohJk08gUWke}N0NPAOQ
zgT^8D$bGUS-Sh1T9votdN(pSq$w@hjc`9(FM@_|RUl{81=fE0{P<?*<g5BgK8@Gy}
zwR%lASq7(j<ETl$?VS`sgq#gIBO6ilm_HcIxHSs*cN9bI6&M6snhV{I7mEZ7@9uBG
z@iJNSGHJYh=UW2+*l~;98%dU1R)&Fxhv)0(2R?R=$Mvz}-gv;><w17Ad#@#n=s8Q#
z(is)rZ;MyvO>I0T!zttI)}4W<bCo9Om-|ZPF0ub&tdpsgV;w@w%5@E8UNs?cbF7-D
zx)zQOd!m#Cy1W)@6<L|QenLTmX|zN@c(hh^E1C@*P~T*lT!OI7sm0G2sPE<=AB{p0
zKVpK)e=JezL1E^op7OX@nBI%3UuU19Jmu+G2m{XhwwoGG20cjY*7i2D`-SOrsSYsY
z)z{befr_fqWQ5dsBn`D~L;yNiXNSXBL)&$f)q~vB`gjGBlUoDvl6L>5MYyn66IP3c
z1JpAx?h{k+wNiYCBmO@uzta17!g21k^%8P7e1y)KJ&~qoUJ%w)wAQbL&>Zok$ia!s
z4fJmADh(;WsQ$pg77AJeS1a_*Trk0#tv9QydtnyesoplnCXRv1%T`i~Tac*Wl^KlK
zFzm^d`xzA!9!)R9Z_T<p9@FzUuR3`G+=h&o@3Gm2GhEj>pXMdc<o)4t<#yP!{AGVk
zTyW-`Kl&;_lo-w?b-f0h{3Jp_LE*SR8LW^ivvjDqpVHPQ7&_gERohC)c$&uV)o|fl
zs9bE${of2|`re+t2;V_o5IwWY{45Si$676Y^XcJU#O*CH9Qz;Y?xnGoPrx{0(UdJO
z9*oIKvc{+LSM`e$YhSD{Ul0u1GNV^3`)OpTTw)grxtYybLhY+7>Q-8frE=3HDX#=s
zL9V=~XD)(Nc11Gx%q6qBrYPu4fFR`J5UPjmv0a+WS$oR;e%OA!R5))%`}1ys_Jz0i
z$1749oF4ikE01@0q<NwTp@2SiN}rgCMR=vn)SbqPIj91tD}%<^AHT3FB;7rSOtE}X
zj|9W6ljq~*Ois5}s!VS$88G+*^Zn`NrAXnFK3Gru_>r7~A|P8V@*E5!`UVCL(UrLC
zJq2ekN?Th6n3(#r&<Ml-_>MiBE=Fx@Mu~K=IuXJG19iZBZ8i`)SITZ)*Ne>E_}NRt
zc2uwic6hsUS!y;4O`wYw3N!|v0~u4N3>=rPFFp&G+0|r!WESH>tL?{Qhc()^PxHw*
z9xAUW<+16~M^7`JbAZHTHKigPgWYXfU=DG}CgJlW#edlM+f)T1*@S`0{OBzRjY_I=
zf+#>W$j#9nP#PohSCMA-2yJPJybCSG>8FkV&F9(lU^56shxhVuA8HT*-?iF{&wc~p
zOBs?1jy1}==q5rXu*8Ko6#zi_CixY?XqKpl){Ig3iFs#cdQJP4q^%7bziZutr<JEl
zGGV{etGCBe>0VcAdq<r?=sGNk-@wbnR#(6K`N6ff1MQ10D_$JjB5$o52u=H)s^zjp
zW0!}AinoiNM`VL{)(NKL1yVn9&g8aNVN1weFrceWqG_9Eyg~OWihY;T8ZN)<!S)ij
zM$Y&ljq$rlWc!&WD>o_$>7j!{q*$@s#MRTWE^DVOULkH*n0yn$eE7bcL!Z(_J)hhI
z=bAzeP{;EZ8MoHY#4p1yv_dyATbcUe>hk+uxURQ_>9*i{xoQi^rB64q`PK)c)1y4}
zn00GOG1XZ@ZJU=~i?KHm>P7&G*6J-nSsnf2k+`=zLnzskYNDPxT5)(poG=e9i!C6X
zH{bUuX-#bFuEs0FV@iL~*SCSSy?cZtVpGr%keTxB!t2<U1k3)Tj=H8plwco>3x;32
zprs`p3<<Ym-ZV$|`*-ikDk?hq`$PO+Zmsly*2iQu&+6;zTUJ|}D;dw=dAk*#V<D-Q
zm67qCLNe}?{pMFsPtUUI>KwC)KT>JDiNZl>YQ&(6{b@81$7H?8`TW+Jnx2`NctFSx
z;cQApCwBWm*3Ql@r`PdfcT9aqp>(Fwe2S(goJb6Gv*qAx=jIZfkLH=o#y?CKsVU?`
zD9x3@FhwPsrs?hhzJ!hB->k9qxDCuhKVU)kQz95;BH`PhpWCW^6HA}iPxV;vSCQQM
zKT99=UVPR-G?u?|-PSGwNGxO6vG18_$#m^_Q~G*yMZA($7BaW7i+l&`SmRnxt1Esi
zZIUO<W2De<&iK@viaYFX$;jAKN~R}<%IQ31sR!!wJu|yaRhBuU0)<>J6-QsUsUbp<
z0;%Ab_QUBsjx8#%#s`*uUqggrYcOel#m<e#R+FQxzs_HI>n{p4pQKD%5T?epOfZZ*
z-QCa`Dpf?<l_T+o!A2(@M<eJRSn5E;U%%}l`CR<eXlZ1n)tq*1TuIQZ9jRT~Ysg%#
z+x{Yn&7udid>L@ZK)Eb2r|ow<9ZyUh9UYkzPC30n0q6s^tgI}l_l=%%iKe){J+p|2
z$QLm&7Mo>WZop~3t4y!+BVCg=(Vy00?PfxBbac^3Qcwr^BmsUP=x{$!P?UQ=x~Bn8
zZs3H9QtFH+NP;=b$nW1D_Gd~*9<NtMz}NQQLUyRBsozpj1-%!GC<A|0ASzK$n-3f<
zuSu5kq^l1HIr;Y$07~**Jeo2pE>3x(@i}Dr19e-RQw~bjiwE!qk1+(zaiayg62hj%
z7h1)GpEauchNpZoSxDuK*S+!K5b5a7lYCaN^u%8%J5QseUvkn^=TK!kTRp6DfSkpe
zHyOTMaDcCq`HjsdRy*>wT{jRARW7atc`r(g2XmSp=e@D2#wBfios)k2$>wz>WRBMX
zaI5c3-LJPl;X$V+qigb*KYv$Bpz4XKOsThF*;|HArl9_HQAi#2=nF)MguEJGn+HB#
zf^Ol{2#k%0uoV9#k3TqH;rqNacwdmU12{>o#%ke385n~>P-Xt%c(L9gk=0B-@p@--
zbF@T@=hv@a{lt(ZmlY<*-EW}cy?+Uq0e5eArj(2_davPnB#|X(tusgkKewPDId{(T
z4*-1vgZhkwcb2hYeOq!&jN&$s%I&1z6Gi~`3COj2d$ru{(}Gw=Twh<mej)a{3-ftZ
z>y&c3KVVfqed4sxlh4=!>yK453F9!6G*@)0vUb|Hm4enW+t<z+D*$jD^zMBV0jdLI
zjKBg@$v0|mFQTAn_1?eGWPcTdR-!l?H>^l)auA=ps#CKR(GEzUGP1+y?xYAXz0ra4
z?jlRhK3;NfF@4JM0lx-F!hco54VK`e>}gsSGp)p&b9Xk|B?MMfWZQjJbEs_8hH=^~
zsMc~rK^>v1?iEWN06$d>e`@_IHe`*3EXO+t<40dmU$3rL!FbKO`(RrMN=~eVD?4dB
zKkw<3A!|djzyfIJ)=noa_ttwyKYwCqyKUigbaoahVLa|;M~<4Zdfp$^f=i>vocNJ}
zfs&4{s0!6zyL6|#m$dQt3JD2`_pA#Wrmn883>>(Eg^dkJnjqrg;@aPw>fK!(A-A=7
z-K*AE&av4IrwfcYjPR-4LA<V)#0(8TfCBvlDM}OsxV^nS%cX|ca)VywJ5GytBb%Fs
zSy@>yrc(ttdU|@B+uQogi4q2yb+$faz(dq@v4#q_m>f&NL&*D2pUOI@S7AW&<Apvz
zaC_O32LRWxf0MK9LPNlyY|IjNDu;Vi@EXwA-4i{Fa2!77&KzVYf@=_9du1ZvGTl`3
z>hYwnj2&m5i<Tu~$w@+puKFdKNqey7C%@=OqU;@Bzh}`_s&_uN1><!Z(Ox#4zY5OV
ztI@dtUpO_*Su*LbCvh_<T(wjo9fzJqOw|B+W5H3cDC+fXn~PnFDp8=z#5wY~S8hZb
zTbp0C4Kz$o4Xa4T=ZZn%M8}0>cj%`)ik=z>jbOm!JS_2^PFs3k6!yt@bWWYcdrY#(
z%L55aQaep{Vp_(uG&3`s&wZpzgl<>$0m=_}aQ#{4*Iwr@rz3ExmO$&*i^>!Rf|!l^
z-?ZGUBVV}lv6u`SZVx8H+}zyQUN5;gJpavTxvU)=SlBK8`0=CA{Z3s?ZLG!11C5Md
zDhQ2))BWNj5xXThKY!Zy?}gCfRDLgRn39r`6do5ey!b*qwd(Tnx2&u&L&+R0rlaKS
z){B2NY?`|hm<&n?uls3BW0rxZ!^)8aZl`^vYYSGw#l^+lg<2ak%o=9%ci{by0XLv~
zGn*}2V`F*&2Zw}$vbh#SR^k|uWjyH`itvmNUfESAr1r)VEom0w<uxvBx#C>=H%6(W
z?R7yk_%k*-y~fNOB^m?a%TMLP#ycJ5c!@{4mzRRQ86!96P*0}vfpvxl2w}Yv<F{Jc
znO#@-`}}Q#%%BRd!H9#Cot62B{dR|B6AbP@RO+z^SH^7*syWR)R3@?AlQP?DyIUP?
zGrgx^b?kU|{TK(-<O&P8`uBQ!1R!W7fBqwMi?LA%`Lu!J=|<`;8@Q<t_933FcHj}*
zUslR5;(Y!=cy+r2q&t8|9|5JV;Zj;&9xD`&(fHx!?B@1%tWbp!^eHr9*RuwZvGuyi
zGPxTIHm&rJZW}$}Dodp^2Wm~OOfcXPKVuV_0IzbMYP#J{h5_S>juIU`1B3D2cn(aJ
z)j}M#!XL#17o$nVG+0<zQ`Y{M0e*<+)20zw+lp=oCqNpXhdDF)e^5(#x?3_sS;?zs
z8wd61DMF=})0LZLw-~YhJ{GagmOQCy0%`M)QTCX*?WuixsWjY!I!7MMqcbR3rrnuT
z20QU<ze|eD<Xv31PieXijNm%%OYq@>Eq%<Jk4PZLOuTcQ{(+t{xM|XoRau?5S@YCp
z2Ea(i^>Km_E*U<vRXPt?xAFFzpmkIs_q{RP1Y<yum=`0VqeVoe4VwTlxZiRmqod#0
zEoTL_wMm1}$+kx`g%9VeL+LeYVsJmW9L<Yli-xP#T9aHaI>;uLQ%WU#TWcwKg^tc>
z<p3igcPo)TZ<6K~h^OtkR?t$dVh!lr1~!bD(lwuq=Sa>j(3Zl2PDv6FaN1W?WLrFG
z`Ey=XQ?m_1{>^qoGQ-6>JI}klT;7M%&gc9T^nZ&>OBSp)a%@Trr_P77<&=<rQMHaQ
zap;ZVwoi7=d9w?2NDWa?NYwfgQa}N+SZ|xvvBnSPuU<OEX3l;GTFQRkb!)-IpO?xB
z{TY7&n`?e$tOPr(9>RbFQi0zr=%D&()1qS8{KaOIVt1QLam_*4#epuFWqBBt;%>s0
zCMrYd)}SFMhJbjw+2i_;%gOS?jsUdXk{z7C-3j87;U+f-3_i2rD>E~*x(0115zFPl
z45JGf7%hMP1Teru#_4f&sC514@c6M=+eYqwoK`~ONxO=Lf+FX_dE$>WeS}_qQa`pX
z`1H>QO0Ckk-1SVqqoYIO#KFbIWPdVWDuol%#^z+7ilN|3!xQq`eY)truTqNm`N2DV
zeGUG8*|emW`x9VuhNCjuDW&EjGbrdg+mZQ+K?F!EyT6`hcLwW*V|hj<hli=w&hARj
zR}SrMAcDg!e%Ix};JMg5_<{3kGStGo-EBtWZMVCG#(!=69q4Ntn<KGVL3qh=@?G>j
zBV2zp4iKDeaTV>&4-Ym2Fg%gIbg>Y%z243!`4UeX`<h6M=QRX5CLp1}!3`4AU2d5|
z1L%BXP9FS_9!AB&5DPX)6_Hrl7VL!RBnN^KvLI@5Na1Iq*gvHtD3ru6zqaitEGM(}
ziPd-`Wl=0*kkg%DQsinY!oUbl<?wv8P;@^!4_}oV^4cOOtEkOa{Fm$3ErHciLN@a*
z5RRNdAO?3Ad-&m^aSU4h3bFv7(l&csS92OL>{uKDln5zJRJvqtpEqTJ%j^k#yr$Ev
z`+0NY3Pw29R&Tx)uWOQtTv-$FBfv9=4jozc5)-lWMkR(aG?rbUWqtpV@$OyLED=o)
zuJq&==`hNqf?#S<FN6IN#pr|teb%0YS^bl-4_CMzcsQ_fZ$SUcqp>0&D)-*W1#@!8
z0c{86aFE5t*jcaMV~3v?v?0o+(r;{(&=KqTDzhBfELnlas-7Cmmuu~@X}{W%R$5nA
zBc?XD3>Po!a;oW%8U=dF@c7#XYTf9+d|adjZ_2gQ<c?lS#*2z`ye#S6dhOu5aH+qW
zkpD8^s_q(=aP<%<*DBJuz%hwD`drxTLFsXg5WdT1i{h7-XsUS5&BPx#1rgZzyBz)o
zXLI?oF1#5#)q}77`8$eRBHQoXR>9Gv$q<epJj-0p>*K^k^pTaoMNOG(G+30|EBK<+
zqXO0<VK|czhBt!pOsr+Bye&8H1e1Oj6})DF%t8kQQh)_V|A(r^bYu=%>&6JOZ##GE
zapI*^=DlWU(3VvH=<*z;m|6#`L!pDUB~y+9s@FMgTL%4xoRcn;5OW&Cg7*5?3o*IB
zEH-T+J7*%R24B*rS#jPR*q-89!M+s2c)9k6$)G1jhOP!glJ?k5XJ=<OcXx3-5ui6;
z?+%4=0i<cRkG&84HbK)JZ}xEU({FB`XXECc-*o9urWuC*8rWhrpZp;jPROL&2Gat4
z@-QAw>5HLOz-KW=S!#4CJzadw*13ta%-C|1Rdv`{(5Ei^Z-2694(94MiQ<MHF{%q*
zCMKrb!SdOizwQ~I#!&)=+Tu|kM2?JC;sB$EMk{!~H1AN8l^Y4gcl_VcPP9i-MzkAZ
zH*NIcU~5(_VY}78(g=Al2Pp*A*zb=jgSFMc1E|LLaIM>D#K}Joi&EYx?12XN?BW7t
zN`iUFlPbRQXo)m+KV{7iag+y>dvprD^*h8`!8w`+Q}dBBxkh`3*K=i?M2_6_AA!M)
z;_R5O6$*kS2<xw@fBZGR54=5mvS8}edH6}y+oA#|<^_m{O6#vp=*Y!*8OS4fz*pJ9
zpi7qh6c28NxS4LeuCIdft~e)sDHRvqJ-3HS^j;ldeLW9&$(aH#d_Y9Hxkd-`(14(9
zcBxQ!U3poFllNNkcmJ9okE!|lYE0ncXvgI{Rw{?!dGC<Cquf_+S;%mI`20G?f4nq7
z?U(E`KGJ~v5MX%h49Ko`4Bw{~dh3URlQQ!D(VlssRdXmxgol6|B0^YWCBHY@lA0`u
zwESLewACG_&85yRoX~~(o{;T$JY~WH{p*1f%R4R)1A&&*Zp2hy@>jmL(V`AO;rFbv
z(_{1l8}$Z9Y7oC}@sna=$~7!DI34hiLe9@=w6t41>)V4BWJffj6Q4k%=Y<TiP!?~p
zDW$|Rv9N3$)vg%#$56vq&X)NEV^dB4ZSg-34*Nm=%E?zIcdOUTVD|o+>@SXl`V|lq
z=arjL)!eAp*zh4ePpdNc#slC67*hLULrq+-7s6H53ApsaznMH+q@=NQYzvZ0fJt-C
zmtDZ@qKpa-m(td71ObadjM?g{hFc{n=@)))Z|UV*>bJ@gU&1c~KjHn1KSN+B$H&!q
zZ==GkBSSdoU^BdWCKp{$MW!T<sUt(spYNy9rjFLxtoI{`)ov2l0xWc;v68CA>F#fZ
zOEkkWW{PYF-xw9$W7O4M)6weOBBd%Lrp1l3?eEh`hh6EMiIjbWeQZrUOev3u$+JL7
zy_FBxK)gcU8ht1(drLqH`S~kqDmK$sYwz#S7VDneRnbV!g?o*6dB}TYm#@8_@<_uj
zUdA)4Gvzs!v7yFkZU7haaKC#rHxj`+N&3fX?wF=CV)9<b@k8Nm&A7%9aYs1(BV>$f
zo9Z-QA!GgJ`_;EDw{kjr!SoV+PjT&LND`iO@&CmH_W}0$x7w=lFTIxfTdrcoX?z|N
z6Evs?q&x0iY=S$+iHFiqT!)btQBApik4f6S7`m^{Hw278Cn4nO>iKlLlQWQ+nYmSR
zeNftZezjl|KX<R3J$AYt${6Rpk}I3eLt0x^Wq5PCR-|4XQtz<C4AMN{DN#OBjg_DP
zDHwyx{V8Gr?-U*H+W@<b9wm>I^8}jVXe!xX$;pzcs;c!!y9?2KAWe<XONU<~^TE|d
z`6LNQo*6R76v8lH{-vfc2cDE!PB%<`JvavbEhrPa%KUf*l!w)vCyb<$^!@s-jIlIP
z=O&}d)Yp?N#3kyW+iH>ylHn&e|I5m4zj2hOk=FIs904fqji1HD5=bT2%klYB#$(HA
zbM^JCG}E_#{lXG--!MW^5@KqVPLfgo{LT@u^>um7*|cyYRP9g%b~pW>k|pULyqVp?
z^-a%ADQYeEE{oD1h)mE)VHs&}eiOJVf%J|ttU*&|Bk}`dk{^$9c4C@PD|$Nnq2P&u
z8j@KNF^7i4v9pU$pSP}>knBXD@~NJ}Y#PFf*7o9`Y|i`1?B3)p6%CJ?^X^g&WCq~T
zgH%~FNKSjJM1zv=(=|=Zt|`Y=l!S&b@ts#=%wn~|*YZHc(c{hzDG3-CyTHh7abAzu
zsZwZ19=y#|qCX(y&M=^pL_y&~X6MJ@l~Xq`F-ypAg^(;iI#l1$<ASHZv*vG;drtMq
zY>^t~<Qsq`)75cj=$e%jJMFGJj024B?y;7jODt3@FsgE^POQz@q75#*5YP#TOj2%T
zgd}~a)0WDPfW1-7<W>Vu3dXSsNDAuCmgza}j=}-}kdl(}jVAW5Av2i2-QM1Q5fM2%
z>m`HC=m4?S9ImIUuY0Y)#26WmVG~TjJ9~N%ij^h7Y#J<2@KVX7zyRdDyu@VbjiF*K
zPj?o|#p>{a*T+k42gSAc9M;6)glsgAP@%%9U-9w1-Jy7M)fRZK1AN7!kLeZ@z+A6T
zr4&II%vT=nPg*@s+YxW>?+vW2BeYAY7#SnMQph>@f5=9pfWCpkWe24~S0Z>4N7DGf
z5|0sxCmauFF#)jX0}4|?nhVTaAm;;&D|Xre9;MSJhJX9}X|$}zzcdNq_y>(8zF&9_
z8FWR&yBuyN0ewpgbM<V0pfO9c4L7#7JA^f*(CWbVIOggy`jpAb8ZVg>V=nUcj|OU9
z-_#MjT!_YU3MbJi&Yu%3)H}c{{MwlCaY+<})ZGC1Ndfh3_tJt0KuhD#rkKOmDv`6c
z36%gTtP1iCW!d`X@e!d$=ZueT>FVyfR@Wi7LiG_Td8n9|pG<aujLbM!>BoAHMzO`;
z=x@J!RdPgQ#>=1WD+GxbqvPuMHM?BnF~*bL2+#MFq`hC#Ci%9u--m}adv;0qiNlZa
z1q~o`47M8+w^*XJOlxuxcR%So<-3QmSx$I_{*@l~O)HUq(UJxJ1cWI+4$m<QCr~E!
zT-4brrHvlDe#8X5+^I#=WI<-ZMcYH6?FVEk8xHQ~)2{)AE^avjXlV?#<k+~BUl&xS
ze56z{A{RdkrC$e1wlcy$Llao%i(NqOk5uAkaLdg`Gho3Esf5JDx6yr{SXerjm(#9}
z7FMDby+GH5ghJPvGu0K0EtjttM$Cn!p`iip8h8{|S9K%E1kk=V8{^Tbe%rp{=jR6z
zYuN?3@lcQxfT5$OkLokb+2G>hdgcxO$fQc>>#u8VfEn%?7`fsA?^eq>W-z%gFDv`V
z$B&4J*zkWeT?J5;UDpKx=}@GkrKA+;MnFN3lJ4&AhDTaax?2P!FWnu|-Q6hN-T%S&
z`)6Q=87_0V=Q-!>z1LoAZRgz{@W<No3MQr8@Bl(>`629Q>)^mBa332RTW&rc3;<3z
zT@w=(GkvWbA~70Q0|SE^*TWT3ovOhR0#Z`1$yKhVpBWj`O#tpJ+rF*Z(ZcaZPSegd
z7!KQq8!|f?oSL*yhwU=&&Y^ffZai>|f2A`rt8WB9ip*E72ZfkQj<FO7mggJ4)(XH7
zG<`H-FH<ll{x@bU5w-OB$ZHxq_kM4B_*?6!QVmwc%nV!%SXk|c_60R4wk}aQOYi)g
zIMZg~u;>o)(`n2>CA-`20h@{R$*sJqTd$p+j<z*wV_(pp-^)v4++KbJA<PLEnauwV
z7BaNA+3H>@N{ef}ym`S?^LR^gz{v;SSlgFoQcgATcx(<erc!2##{kBDIEB72RFG(e
zU$7?MUZmO4;->MkuP=Y$jALo5gizmE{KKyFg0<$Y#n>XLd#!wYLTCD7rAoEAUu+2B
z#dO(^(Khin#SEX!gXAJKAxuxZC(3ksdIsGJO;Rtb+|Ox0dRU|;UO)e<j9+}i>q*L3
zA)Wh0>^1CK0%K#jWYgflbWiJ$^C5pZz)9qFVmSWWpKqH2GLIvG)x=#_r=*aAi=}zC
z?xv|?P~e<|&v~aA1f3;S_yzB4tJv2%+6FEa#Kb05*7NavF3_uqi3z;I^}W4;T!m}~
zEm`nHD@;)^Fqo{0<YG(R6BO4fCbijUNx;n~I=qgBOW=iOvXA3A8^U6W;nF0-ZYdCd
zp+L-CYJxC-9l~xG-P#}*sq%X|B&bCUOO}Jh5=Y#@wN&t`eUiJxo;<AnIMd9p;1_Q1
zx#ZVMy;IPCY8I-Y5?D@C94O#>m*x^=py%LXWZk!iLN75r@I+myME$slmq+w7v)QMw
znmk~(6l@1O5*+i7=&~Nl4YgNIVjeAL?2#{4KjGmU=xKSMI_ln6^7DM`2&8XHSF^{g
z%xclarcYRGmY38vKef%m!7^;=Fzg!D-8LRPm@blsgmPByM$`_?<cU#Svc|S2akOKH
z7D>ndK6}AQ@@w@NPbFib+yL5%#TRpvfN5a7nIJwhw)LDQYo*0_NBbH|Ru<)>it3&^
zxqDH5yt=%vxcS`4_7QfLV+gx@?%@8N)CC(6TE*0)%ir9ZqUDK9^QNXPB-i6Pas^c$
zZ%#KpXmt4lC%>}n*6of3vlm4sa%6B8>NqE@s5~lK7EhQLPv8%qT}(cLfB9+}n2xay
zidgVWzk`;N-Eu1uz}dU{``cSv9|;Kwf$lvKi!K}jGLWy89Em(hkm^XXX_Qb^B?Ozq
z0Hi3HzkWS^`m}XaR;a)Cb<_70t#*jH<gNK~cye;zKnkB)i9QlCHW~G+S6Er{Uz3uE
zg@uLv0|JIK#IZ$0M0!ZqbF))dhQ-N^`V%=G(p>;#MaIMgL`C6JQc|AmFQ%o71}WzY
z_~oud_x`Cr>S27SU@hmWBVrg;N<mY=-rnAHEawR+pR?ibFHgW9=#)d+u=h?sk!{|-
zg*#p!>|9)IsuTs~Eg4^>4KCbWiWpTCk+*^mls)`AZKk=z3kpWu_hN8f<jF-g7gxz$
z8JAnOF1~pFhEI0V-IvRFGGyANO$(F!w+#F)$?#tP(#YM1#PbFZk*twQw2mF|$0yEy
zIiC8?sopQ2*7Jsh_T@vlBPHkTD|0ofZ!M6y6Z?nT>o#}`SMbq%Zbj^l^9RpftKin{
zlib)VT}Ae0Vsjs{S6NTTb(?aOvo+9SwhS~JLCTV8=z6aLrPrJlsu?Y1DBagL)uj*2
z$DhC2<)b70c+OC1N&hYAsru9>SI#&Kla&MORh8wf7E#{ig}x|gSErD6PcC{)!(||b
zwJ1W$T5LPJcJvt5GF7d37q+zf9Fc1!3zNLq;_qW{L_T{@JE=Nrxy#JZXQEl}!q8Sa
zv>v$`Zn113DV7D5flGm6uG7-ea$_hR;UQ0H^+%;8<)fzd=W+PUVlqJD*I8U#`~%d&
zVDT4oRCjiu9YQ{5CWFpU*byg)9UG@g+oy(>$>65{PWZ<62^9uo-&6D0PpAt3^;kT8
zhlYhEmK|ly#qaPJ!`^LroP{+3!@|NsQ&ThVBp1{?*uQ!|(ZL^}Y?zI4pCY)=6YfV3
zzd@~_;$hWRN3mnE{&6LllTA(a3dt!#vKFZpKN?yih=>WNRWhn2=K4eoBh|He+#PXf
zMC&T;yk~u4F5MWeJI8(g71RW|tuu=q2WZ<#BAJLbHznWX0=mQI6WMjwY;5cWUWDer
z=xRJEBInkaKr5`o&f)iPFVcN+EaDk9a5LD7sg}qZL_%P!u`#zsmn_p5gsDe*F{eb{
zK>8@9Q8VI%RKLjv{zj4z>6rNvG*cB*W)L(;x90Y~v*MGOYe8S5HTjgR;vZ5+C9`17
zzBwG(d}n26zF};uT4{~N&w0yJ+c=We1%-mG!lZV%37|*4P2Fz*DGcG`qWfV(TrcvR
zxSsr>(g0z<#mMnn-~+Ibg}Sf=E$JErf_XcXKtYc9IV8$%`~&_(=*wu%tX0(WwE8g;
zvzC5DR5ca!5A>8m%$+&EzrB(2=C2C7T=dE^{1_7;Bew9=iE84ey!8C|hs3vrH$MLP
zsD%+e>jP|41cSOdt=ar~o)~ZM`nxZ_do3h%H%})=X0E8;{>XH+7v`c8f~eu!3L~Tf
zU5J}T{ENl%g_VA`_ROHERV2Zy^wmW0X?0@%*uDs6Y<|dC9{&3#sfC}*TjTk^$mBs(
zkMC0~gSJmc;%x70FBMtZjhGr5=D$!nx@XX4$Bj21MiYm1Ym5IP`s@Aa>bf&?kY|+k
zHV;M{+8s}6-581?@EXeMGzixHPQIc&vVGFkXMT5Sc*61Y;hjKDl+uk`53o|M({Nu3
z{$>&_{v~}xqy$ve86s`7T!tR842IwH8M&w#xO+5wRm8hUBOkw$Gif;)^!0eeE+|Io
zyjv5Z`!em8n0b6CACo2(2+J2Z_472vUkC=zuQz$E0Kh_7of&z~7cMn+GLoG)*nBR?
z0GXorh&dppc=#)qVQ$c`V|>5EBbFVPe|^MSaNYcTZn}2ytp1oo|FJ;|8SD7V-GV%&
z295Zb!ON!5CuQ3h^q(eQcJ;MMSqH+B5VR>n%)r}y7POLknSaJ#;lTB0d!vIRPvrIL
zOgMg0*Co_sU2a;PDE3JP9-lK7z*-$u4sh^ZGebFE3K_U1NSigu*V}{yE-93!`Y(`(
zYP$OIMmWrO#ozAZBnmwBovW9T8m9!O9(*bKxUSQph`^4pEqgcHW<Lba1biEuZ+M^2
zaM&sXcas>D$iTi#mQTJ$#hly|(6|?f9j%H1<<eAhp1svmb9*XLH0rPurnPfz#s<{?
zRb<6+cklkvnnWZezds2b-H9&=Ek;E!yOtW1WZaIfROY^}{ut$&75!F#%KD8^3i{hy
zd_f~Y&87fsumVetSSPIdrNs(7czf%@3a^`q6B01IG%43|T3WuWU6^goaqJBBy&EQi
zehG(TKiS5XI4oy-5kZuI%|LoOtGHY9L&<{sVp|#1V!nAlRIk!8Y+1zfhoc+>{F1kc
zlJB<(u`rW2PtS1F1t-!+46~xDh$Z<LC4U~Jf3VObUmzZ|3B$+4B+q!fmStGBy3LV1
z?|7n|RVrd0k<K6&fuHn2-SUMv!TwJ@XqNWG3;GRPj2r(!uY9~e2-&%|IFG&Odis~E
zFhVSY?pmF9dUAxPY7+$<egFGKT*Nb6Na4TJu!sDjg%PZ{`J@$&h4;Phl;}6J;dLuR
zRyMzmuL8v%o)=wHC<M~YiF(hg>;4MQ;7g|Im$NZhO3T?b1A~^v^G1}Qk4N8Bg*LhL
zol_#7k_u?0S1}^a`W<yRl)OVWvDfo>9^k=tu^#mYj4#nY_`c2v`)f6mj|cG9Akxi)
z@7a2tQ5il|(_PHZ*!SFB_OFw!c^b~(EI+dGx8x%`&vCmf$E4LE`y{kt#EG4LXiCQ5
z8`aD4aiPr0&s<BF600+TTBV2$IXHI5gPGF`KhF~#By*_}p7EVdZ08(ZK0{+YQWem{
zI_xyaT4=bP&sx_*=cE6GDwh)w#ZycZlbn(801l{xzGKs!Ve4Y<fnZ7WYZpOCcHYNt
zrlKfAtiJr#=S~X?`9_-B(ij->3E`=Xw*I(<!^;nYF@m~;JMdaia&xEpRZ6C~yEo~8
z?`ijM@}{o+wrsY8G7~e7a)+4nsJO`fo<pGs*E5EfeY$c2o99}ecVQOMMG=L=i@Am=
zQU?9o4rce3udnL49<kN<3{pMZ9XX<)zV*{CKgA`Jqm?}fB@=iaBBlTXkA?^f3+ti1
zS8DW;*Ky-BlwmMm?<IqSIBNcM4F0~k^0ye-`q6UMh0}Ab-8b^D?mmkAHrAYL*cdXf
z#e7FUhc2nX2-pyPxk~q~rP~L=@Jg8p=Bjbu6GUV-x`J`7)GGq<Uf=&7H+VDD{1g-~
zcbrCu#*cSNvx@XEm&yo#R;KB<S5JReKfpH*K4L@@h0?HAyH-**ROnhW_LfZF25gL2
z$Q5Lrc9E&7#%UqVWtw~>5zy)qOrWGOogo~eU0THWLjObiGlAePS*kLfz=C@+ht16k
zf%U6u9~it%2tLhcu`rY3*Z31k7R#R$K$FjGgev}<#|vEpGz>NhGg2KyjZwK7^%3(A
z*=Q<Cc+j)5wP_oPkeXKw-U}9ww2hJxEO;E9VjE@FTL}nwD%^$gTA!v~>}8=by3K}R
z#XYTuG@~LTV;@r!L$YuUsm>W5^HGhJN)e8G3?Eak@uf@DYc2HZ!yHqs=|mGrtmOm<
zm$0@YhX=w;R{7B3@}k$62j~rl+1N7cP!nfc>mcdojPHvym1rA#FT6<m1B>)|YF!EP
zMDW7A4huGh8UI~vazP3FEp_&PB*)AvhFAG7*P?DNR>OW~7g6vpfhRiHc!x(Gg1tU8
zGHH8=+Fl|&trToT-y>l>Sf35=>OM!O`VwZ&ptwx7aoAyyJnM*fT1%Z9Tp6V>&ZP`(
zFGUf)92q&AwoR^Q+}o_b?H!l5uPWi_Ln%;iw^bm%$fc1AUF%tX>Fn7ME4YvO+Qo@r
z(Y>5#Ap{gLttJG^y??9n{^WrMzvYLHR&|MURD@nFY3srtlbMtS9v*_AfNS4SKW)s_
zd>y#I3Wt(*^Daf=^E|rG?d-}%LoBr8?-);pp2iuKCzo&T@!V2cls4+k$#tCht08lg
zH2&x*XfRzT9FEzNa;R{Nn7}-1GD1dn(hi~6z8q$IDwO>yI<gM7vpo7ak48syN6-1A
z>o&>yJ+Ig;1vziiRMp&g!|cVAgAW#9=lK$q90-qYN&l#(M>;bC6_9La2^6Mi(3YqW
zt<c<aSeCW6SEhaNUFA<Fb({q1bztm46Co#5OnCk)6|c%-Xj|?4tOzWQAC>3Wv%Y-W
zkQu(#0*ZLo)KQDG(H~jZR^n$l1C(h0f)UDjc=vwQ)<_n*R;^>Y)oi89%^Vna4A>x}
zwN4gH%=-m9D1cv04N)UH9Baf#TTB$Rban>i<h)b}dWj-JItqrLrh489jJ3O2UUDXv
z_Xg#cYgtQT-1lt=>rFV-o)b$H2^*Mm%B6Qqk2|a#7IjEg(Lml$oU>J|+W0X&aWEs?
z#;}hlJ&%UaRAfVH28L<svVtpQ6P1cc{cl68EZ0cX%0+7%pr%J&3#4KVaIe2~1%&;~
zn*EWlzaHXG#p2q59wXz4&t5D*DeD-K7f&bNI{q_nm+lwbuXiGifika8T7;N#%x4AO
zrfRUNCCY87I&TIcDjm5n^wnd$9jroDxk_(sSOF=1<9-NS2-ZSa;Oex}EQ=P60Kv+)
zgY6k#89Zt&cxVOd{gawMh}DeCZNJ9z3!KWrwZ+khJfIte#IfQ>2tvdm8QJJJNvNZ*
zC1DB-8ORefEwU;s3SDkToriyvTNW*@p}QoI_2#NqG5{>i$<~*E_@|P^^wnjLhL=Y#
zPQx)q$@PFwc9vp|M$FotG$zwBG?Hmu*Qh0HC!@y!`9AwE+1jItqH_hk>wZuN+BwCz
z!#RTndso*7U`NX1c0>(`hM9oBxUpgCezF1oIU|D#&`*A5X8MDwEsi#5+?+$e<Lq4)
zYfE3ZFU6P{Ri~Haey7flmvjlB;2ns-v-uRQ;JhNG6yLuoop!KJm%c_VKKjc6g0!z~
z1zGK`HxMX`A&sj>p+{rx^evau?>A!)OLLqKWEtv7;KAf_i;>Sygqi|tl-kGh@9%En
zqMs3M($i$;^@^fW@egmG-0-azp{a0G$sdiqE?S+L5|u1(Ac5CV$eQZ4uRswD@hQ@e
zFgrcD-!OZ0RbI$XoA6y+i0FOU0i-P7G>mGJH39GThdR1!{9s!t-j!@bsob-ksv(>I
z)1^1rTL9yIPreTvvtHH3uYXEOGsW?Tdf*jabI|a?t<lgHs@qRCtxnA9Dq$614w32z
zfWK#^nbiAzy(6H#KtItkL%mu38h;7&-sl>knvur*YwsCqX8wIGUV1&K;$phF**)i*
zBFUYXM?k(h-|XQ@k&W*DGWF>SDFOU7>syTujuC|6%Y!|ncn`dq*6);T+CTI(?zcKs
z5kRTKU27OOM(&m@dm35Cg?R{P!?+Trv(_^lZ?D4GAoe<L_%5B{(e95h;`W*Vl3Rl~
zc%T12puSQzD_GE|IxIGnl+XU=!Va*2emC9UebdxFBbH2&5qEWcASO_9bI0S-$^4Le
zDVIN9ZaEzZ_5d6>$DQGnqEbY^S^)07J5#ZGf46^6C|G|pYrkEKNoUUITPs>Lsi3F`
z@B<{YfOwZ;&dpQzJ8hnCG(rs@f!R`*VKvvDe;MS$5?g5v1WOSqjy};F4I7Tn*X)t}
z(o$Nx4%76FjeqxTapfrHjgYNKW|ht@YGEfTU>#fV`Lnj~gz5s2HnTej?z!4W+($g1
z^_QEepBuLxV&01%R%x(glJRLqdG42}!cSwdy01A-=Bo1(#onnDyXRk5U&5FYJPm&-
z!b9plk@YV)8>&9!xgr)viduLa6i4fKDYgaFkMwA0aCwj=30BVFRk)|5Hk5IeKzoZD
z?CP79dlY;JC6Np<4jectDC@_<MdIn~*_|3;qn81K*j{1%efjiEgohkRL|my3oc5->
zaHrAotgjT1){mFe-@qB$v$@NXigmte0P~l(g5O(m|BjW7Q%Lq$6^*3j9Er=Q84F#$
z3Ym;hM`k5TmvzcsM3z<JJ|y~po9{Ji_LQ4m*DXg+LNYvZAVYR}Y$#;X#Z!CJp_oD-
z?lkJJ!{rjvCUPa(rjw#9y;8Pu&XQ*82_-TRfOmx3=rre7r1`)vNMwxvVECkGWCaEE
zHTwzH9kOiG-9Pr4*NuE)XC>o9zG})<OK>aXepMJ$R%CKu5SA-Xv*(M`Rx~{R!gUja
ztiQqSaxe@d>>OS}QPW?W?v~id%1Wzu7^MQ1lY<jb!}Avn!F?C~JS6N^r{LhQwN(TS
zJ*$p%XJ<!_VY@?%3QouC6i+;i%m@N|r-C$?0)^xE+H^sM#pJ_Ag!2Z3YBXGGV`JmK
z6uvkBW50OuLKT2hfLqi$H<t{+3>-2ttY)BJ*_QHE!SPQ75S;5tF4x9Um+i!>n@BL|
zEy8`1A~z0ep(GX7fD2}||NUCB|7o$#TG0^o*JQZ~OVR#~Z^k!g$lmx?m~};dsVAih
zg)#c?MF8^fpO`x(nI-Df2H)D8cU$6SruGDxh*pdqk}_-Ij(g@=*D5n=hie6P)X87|
zG}kk03gUeIt2?2!UBh|yV|-|!9OA4Wkz6b7>N7lXJ$@sSy3u!613BnzO9T(BwI<B*
zO_bSH(1F_yfCl7Hox8ri0-Q;jyA<yYa}cyR06GxqENrq{emCR*0R)x9!G(%JNqN!i
z*OaA}6w{e9JmhtoKRdsu)-?M*e?xL9Fwlc#N-XNuHnP4or;|rg348ZZ>3}pDLzcuh
zI{vhRmhn}{&Tpw)Oi?6D&0V4j+ZQGGsp|(ev(@(0&*z~g7agjwzYFDL3n83TwMtjZ
zy$#BV$$N4kWUEG-8;C+3Ph2QU`+`=j_p&o0CUes9Twh8whQHNQ7Mzg5J2%4rU-S@&
zYFY)6-fYwU?c29m+t=5Nk^@Xywb4LM06zpsr6Xzki%pCEPtbhjCP9XG(3-jL=Di|S
z2>lidPjn5K+i|qqO|gZoORs+B?E2%hP41j-tg6bm$?FXVAP9<MNd_X7$Zpx*_EyQS
zRszV+5jpalpR#jpim`U}BH|voX|#ML$&E8s;zB)L&()#*@rr@+a8Bt+!CsCaPIuKj
z>{}7?fZ$$7Clc^NU<Jhm(ZS)%$idDiC@NXP&KX`gv=aP6?sfBZdZHAJmsju}g)|NF
zzZoMLuoe`#iZNt=Z%)#nb-4ZPLz}a-Ip1&d6360SAjLaz_h2zAkuapmd{Fm{Ogyb~
zmtjXa$}QJFZ?c!V<UK0&uNJY7DrJy9ctjDkyeNQjzb72Mh9^;!QBG2BjtKz(srr(u
zIo+r1&~7|@WRYM&ER|0My_lWx{Cp6$=oD)tsBOfWsT%bW^rN1lYX3t<{CT_9h$2D6
zoqJx7R#msCz3g#a32e2qdI1NW_Nz3G|Mj=NK8(d*loy<sxw~@yyv+d183_3<>m9q0
zXKmpt<ipj$Vt7R905Q@C?j}Ab7p3W2GiFDze&@F&#?e@W90IFD*iA#v&V09}mCvH0
ztEYXNx5X<;+Xj<Bm`h}5anD(qBOY|qz%4JWdDE?DXDv*_xFBt25@3_vgO0WpIpSbq
z;Ok(L|29V3S0=wM-^AWd8}p{H7Y`SP`Y%Bodk^Ze#3X2Oi!VAU>6@Qudzm-EvpIJ*
z=(@71tg4Dj+7`U7RdjIXQtVVMbmpR^H0w=)*3NhGk2vP4!eV4`XDL){f+4=&H31kg
z<-t9iW570-#Ga#!ZJd-e=!V-EYN?rb?k!t19$~x1Iw_;`;x!F%iWsBP0jJs9)mJ{M
zs*@=TK~PnhUE?m8b8_h8Gm1V|_Wxu8vMyzVO<k?O{;pXumD9BKVgoBKDGw+Miinos
z5u#0zc7tkzI|m$K-!9dxr{>$<0*>us$*%PK_f^JHHVgGNUo##uD>7vO8;popp5`O%
z+vhHI#20pi(V%cpzga=6YnpA&f)FC)7w1^Hx$LuNFT_#3>#0gvBLnXLxE?z>c6SBK
z+```EjitiN;l<h-E@Oo<%?ax%dnaszf*%n$=!J2SwWSBIE-;32K{}cYh(bxhDuo|J
zo|v9+yT#mIcjl&lEsT)Cjzf>=?%#6UM*nZ~1SJ4m4X{`lr4JgRlaY@Pq(XdV_2?RF
zDJcwtj_>rhcR-QCtknoAu1$MY)%e`De;-92T>B+<B^gVejfydUO_#n;5;1(~=^U7U
za{fyxDP9t1k0wRA^K3JpyVNxM1i{3?tQGdFaw+HYBW6(iP)!?ZOB#V9kh{?myUnCY
zee<_R2rtAgYF+r&gNazu@dvwYYilo3xV*?MK4ys>tlszB{T)fZc8aLBBxig##cwhZ
z(yZ>8HP$Z0mKy%@&QhMTJ&JWsa--&Zu@QGyuC|qY_`7*O?nXy!k}iKi>p*PBw_7FK
z<P+ffB6dN2XG|*!QKPfF%MngF@UfEHkFi@WK7-JSsje^omlAQb0wNb6<c<{S2munj
zuR4tO<;yRi{lR8F_VPKW<rgr}gObt*h<JeP#CK#D5brku2W;~ky1O*lJbFloNPj&v
z=E5Vp`vHA|FZ9&{65~_7<S;!1*at*`a@X>d_yVaM2XOEiJNs!t9|OvTwp494_O+cO
zZN788TWg)m*A&9-ED{x5t;tfbJ%p+3UEohhCI(c!u<u0$&7?pg&`g?kf>Xz^)aeek
zwHs7ZZs0_QNg|}AEWJ-FrNi-<Lpb6nW_YrjN6p#L6klMmdE#DYU7uz-$Ae=*r-6oT
z((*^d)KcqKIJms{6-4wIA7Lh@Cb%Ms*;`BF!=Mi&uP1joSti0&_(=}0Fydj)$45HP
z>Qs=5RCHhStItxRVAF$I!HAV;d$P!6=!Y=i-6A~bo{9~-Bkij;Ct$=$oK~%UJIbfx
z+Y>Og3UfR-Pyli`#rsmgWt04!bVhLsmV}99Ou+q6gc$bY5#?Qr9Z9)91AbV>9UD<C
zpQtX!wPb3qUqAgr2so0wPehf+`Cj~I6i;4%uoq3x;xQ&eZaqe2(7r~qOI-J2!a%T!
z_eucAzP_1`_tBf3{?6*+UFxf>$*dk-oBS;#HwxUl{@)Y*KF)(D1-9em-@X=j9TU*k
zT+53>uBGsbCpE#aN2xX_RwT<&glG3RTp(2V^ABgaznOpw$l{sT_KJq375VSKYOC_m
z{S<;`Dsa%N)b(jpUzcw#$t||dJdBGxE)^La_eTnqt76&f!vRO`)R2rgMoQU<yBgg8
zLS&Ee(yb$28K4l&R$5suHl_>>4cSjZ=%}d6Y0#bC{z8MQ+We)g@M5DE5vL^~=+@Oy
z9sa{VI!c`!C_ImFN343uJKt(@xGQQigjWu4fxb_%MwB$F2&tXfy_*cvB0APz7J(l}
zpvuEHW@*Qs;fZ__*H>sWsZw2=1R>Pc&yk~59O$uA$0-VQebKzwdJ2aC0j^z*(q&Z0
zrjt8oV;P*19xDGTy%M@gg{k0gGC|8n#;kX#mC^NMJ9(X|en98#1Hh$s>uo9}KSq?S
zL<K;L`X(i~qKq``W-$t{aDk%w$!^*%lH2nO7PJv1CF~<t&c6EZeNr76Y2QVxt&eJP
z{@p={E$_^k92layp^f=EE}%cA<Kl_~#Xn#?bb@T5+IIO7U@Ro_It6>*UQy<E6!bp?
z4PCTd@$=07FV^8%fn0-av5ID_=|`rg2QCctPGt6D<zBy#(`CPr9g2`<OqMxx@S3@a
zP~~!2ImLG`-QP5;4|K;|cLmGl+n%P{2*{KzP=*8i8+V|!DM9{eDZe>sfuz67tG6Yw
zskq3vL)lIN*AKhst4Y!4CQSZqt%g;2kjIPi4?WAD=CW}3v%QF5kB-y^dE<Ft(`_b8
z2S?9vC{l|ejlOPeMSLdd5IE?KcYPHj>prnQzs$ZjM^cC{F}Y!@+&-D}4jIz)JI@e?
zWeh@fZHL>+XP(dh1w|B*d0Sb*;Ht8+3_w3-x0u+`-T~O8Fb)G<)ZqqQMu~%q;eLTM
z%Q0(#jIXUq?$+SAh@@|3{*vRX9VtJHcDR#d?s)4}<T2|TU%9}w91Cjk4!<G}N0&F2
zILHo!J3iLaPA@pgXm?J9{}eJrCB3nPCb%m-D@78ycSZ^6pUjTFH@{xer&Q|6j&1%*
zvKf0n_7vm=C)cjj58w_v;8C7^L?3ilYOp_VN<H(Jrcb^-X1NvgM5}qftCBHO!*|p=
zbN<lT{`y2k*Avmv-uKnOk_nYGoZ1f`e|>uTIzCcK$Eg}VmX&o>kZmYtLAUYrYy7v{
z#Flrqs3sGn{$W4U6JU-Z>QfSvA7DYzPjk&(On41*!jpQ5PA(lDXnz^&|7{^*7VX`p
zIy|sC{^~E-qMn>}7+g4G8LoflpiZu*KFwniZ>zlXmvwu(&*^>6A}O3x`>l>DGk}Gl
zEh?R>O2;`RUdx05k~;8>eoMl#OO$wK*$$W@kx!j4_0fB@nk2$vPZHUq-Ql%k0@Gi~
zKXG9}EBG@sQ8E+ssYe?gMk+_zz0xymdQ&?Q60FGQBHr-|3R>+wE4Dc^4#qNWX(?cx
zj7jP><2rj%8%qAUBFd`bjH*iR;3Cezbb`-cYq5|{fLFpC!u}NcSrGQn%Fw*sQn3*F
z-#*yI`IqV~r%x8cgbY;bMeWBpK#zd((v!c2UOuwDF%|crl{Lc{@7`bkwjA&2Rx!BN
zMns(Gux^UPCUj}&ZsnhXZm#>nbr90>Bn%1cV}J#*=MGN-hXLWjaD(l?pG`g(Kp?WT
zwohjC=z$bAUCDx?3;vl{<C*$YO9T_(+Ds-OzW&fXiDlqEYW&V+1xssbCAG|uE)%fw
zgR&YpfSUVZXWx@C=ci0H_(_R1Q^Cw8^6@~nMQc1sH1dizGA;GnoQO{fXoXqbnBtT9
zDu{B8lK%LYN%_jHzJoyd2A&|CAuE5P$Lg@=)}fmG3u(eEMdh-fT#B)uKoVfqGCf!A
zV>m*%wcQT5!^Oq9Gw-Ze*@#kMo`;ryEavWPSf&r>IK{lxLk|tc;T#yz6VJ@wk7~or
z9feqHh!)cCT!XGAH@yG5hW$OVJD%19&;w-8#cs!t@q>5}D1Y3&-m~$3c%@yhszCXW
zp^BvZq*soLiw&v$5CHmypg`#AlhUCqpjVD?e-p8@O~mb1vobM;`<`?|*qo&yb*28r
z(WrfRLW17<p4=bAH9wj3dOIy8-c{tfH)2+2`AVwsi%-6Nvi_bfmq?+;^bgkt9AW4~
zNWDSW*RScI7zRSq@!tmz8^Csvp=V(IqO{@NKDYGBG@}c@@vh-FC@j_=WA}lLD7GuE
zq}nGm8fypQKU&$|UWM7JzHBoU<;m^A<~8kf(sQpAT*mW<b|lXv+n#G*`=p=oG7N>3
zC@|{0&BLQ6<vK*LqMHT)R3?G)n-mHmApF}O&v<xGjXQ+nM?UmX-#^bdA^9&?M`=Z9
zUE7~>0Q8N~UzkWpRkV{2;4M45;@n>b@JaZ~Pk04D?b^{eokZUa=&~D@Sf|XwW{(rh
z(N@vFs6t=9oCcZB(S(Z&ks8UB+7DtwrT;@xAmj=AgUUx5JZt6v837H>WYv-|z}6nI
z7{TFf80z#Fw@`2A39!7hM0!Gu{qa^8zMiRN{E_2ZbyOm|Zj$7*zU2Tu({m~8RyrY{
z)}!V8u@kJ*^|7s6TYlnNUSh$aP0gV#i(5`_aak41d6KTMejiH??&o&WVMYi`qztr>
z#Q*S5GA2-u45SLwvDeI2+ZkYyaDNUGAEz6A*q#2!U>CFpb~oAS%}TIDwZOT!_91w}
zvb3|g{#Y!%#b;{h_=6D9M}EZ9W*nCNDacr9HbSN|vSbn0<-qsED+d9?299}mKlz6w
z%uk`Fc!hVy!423caDZb18BZ11(efE9bbtA1eA!#pRUe6VhDJCH=U(Ictv!X2Hl}$(
z(Q!MHr__`VHT%^eprK9G@{GwWRb)zpfvLo-C)ccC`i8E>l14+5Je(=McG;~iBNbcJ
zFcZH&F5_9NQ0_r_wjP33S5R*NH-}?ee^H~E-~T?dzq&DGyalgPsHs+4Z`}VpV(#HK
zB{K(L7p8(WwaYfWp3Fzr{5R}(vH$`x)YRd+5I}tFAqTnzM+6Wx4MV!*wfg1EpuD|`
z+2egKpH1Lj)t1t&1?RwKx{tD0`B@@Fm~yo>7T40<Cc+GB<{jFIgWA!Y%fB816NZl-
z<{sc*J{nVLCA7I{z(bshB(w`KdVTY`whOW9Y*E}&a|aM@zMbf`L|O!5+S3wX8R_jW
zzXSeszc7d3NYnO1q^ut^;!=t#-NKi<$GhNte)akkAz|}c6aC>7$+;N){Zpd5?Hma4
zri9z!nzHCWlkTkl5GJvIFx*O$Tu-Pm3xR&cuvF?&$k(s*TGrF0S-C+34^RBMXL+&8
zPH#J8ZLffWzC?p407KgfF{43_AYYZonse)CFs=q&>@KByrx{CJi%LIbU)IrW*XPE)
z0|_>kC_D3h@1K>a1wDG>{0*s2HusX&^1mFJ$hq>>Ir8$6JaJKkovC#`EbW^?z2<@w
zyuUh=!X%;wV6uVpo8z<GS%`vBwDr3jW*<^a9hO(Ng=e$de#oJZ>i@`mj$63tqqDl~
zNwy~~fJ!t`(XsOHVWwNtGkiMiChs-ncV~Ltf8*nHjcxmC4{ei&M^i@9_`c@Yv&55T
zm#+SOlZz1bioK>)5EIH{HaEz3vbQ(iQSDV+lDq3eT&=cwX8flN1hHJ^w$A>%MIpy%
z0!LfwYxl>XU(N&xq=fG36h*}9SRwj2kI%|#i057@VMDk3VOv%jqJaB|<ML8+i<Ey_
zmN!VrNbm8z1sc}hWHp!YwXTaq^m44yb6folm)8<8c{<v@P7?Ca&oWkmqvihN@C@3M
zTozyQebv<x<#8&7pylZvf}h8FBW@pxL}-d6k=b6i)lxH&77z)pEA_3>LfijxV-eEm
z@K~Ais_cXANTX+QIfSHR$)~ayD>MEC-Dx+^6L32nJa;|_qHjx4A$vF{^S|$$q$_px
zK-#Y%e+%%c76>$@qNAf{u->yob}#-B1ii(qUN<+lUu4=3Cx4HaE_OeLR&je>xuR$O
zJ`%p@;&Z}xzb5Mj4zn4EM5`8ANf?#AH1KXO5b4?<xtfu_t1BN@sUG>SRRq5FL9Dur
zVw@ijKh>#)z`-^?u9!1qI>Q#<q7(AIkj1Y?Ww{@6u%#wuzwbWT^63pY*rjpZ<~Phk
zAshMXO)j96txohZ`Y-=!@3w0Idtm${*E<ZR#>crv<x4;{Gr{h<jA1F=-Z7pUalH6@
zy0dvQ@LyOxkBKg@td&jf%boh@cvLW$Kp=Lw{dwW!+`>P<0?(M0XyPxL@0@1*9A)$u
z0!aK&P1S^5op}&CCT+Som9-*o704Ji1z3LtX`t>S+BCym00#U4bU{R2P|aZwbM}qZ
z{ks&eMF1^iq}4#|B;&l)7sn%5<}VKPqOuQla-cxqB{Tn)c7x>PY(kR6*A2k_thB_2
z_nIXnD~kZ6f)Cto5?d9s-O~q<&sA+peg5w~o^Q%9VNf7c2{@@Agv)SWPZ%5{f0GEQ
z#UdB;snEQlXdWg(<FIWY9LY3S1r%9#84{^F6Zf%PQ0;KOa^}D8G<5bZ3k6h7vGY!j
zi`7KMKU%W%6?ENl5~$n6Uo8RbhH+!bes&t-O7<_R*)9&5R`5fY-RsnuS!|iP<mkw!
z@722J|KQv6MDh7Q-P!9Isz`EkIMY<WV9&IW`{wbif`V{x4X=J-VF9oo;XlXecqf#X
zvn%Y2JgmnSzjR6B?=qF<X%bJ&oP`E04{<oqieWllKB+b|OHawLyRIqmk~AErb;_ql
zrPCKKic=RSgrZE!BSnh^XfE3`#dNsTJkN29!aM3q52*qNJR@iYd_S4?Q?O!V8Mkg~
zeo*)%)D;A(ljDxU&2r-LG8tMI8~K<0o7Z<J+Ouc$q@9AkfCxkVtmV*>I~wT?lIo1=
z@X=xD!`g`5o&EZ-_keS>iW(lCrH};Izx^P<<qeTfWsd-U`_;wOf!96@*a$X#so!LD
zV#ri{<+$SOqiN{rzk;gl_;Nk%`0jdtv{+BnxGx?kX1D>7qKn}&I~`PLzpy0!ot7Hl
z0%F$EHd0MZ^EpWVIU7w~Ul2#Txf%IsbjNOXBMn^MPgZM>rl2Ev8xj*QsxV}0D5Y-9
za%<J##`T;sVeb16VN#UmWAHFZQEhWhU`+eciW$^A@61e;W+|1Ez=!i%AeI3%;ZNZS
zvfk>P>M{5lF33A`B;jAR>mIKBgQ4LqN6)Ju_Le>Du#;xVPq)#N=#;`tyu{rly>^z|
zw&Ceob@9A?l4cd^z(=Lc7eYdR5YYqu!viS<;65Fpq#prWws-yv5N&}ET@X+<het>I
zr=^jDl6>`aQm1qJaw2G<Gn6y~SifZL67dTNykV<5y)(PLI_1CG$jDmU2A!^}?bm+?
zQhHnbkcZ7HfYKr&PTTz};KweMz}|HJRL*~}&H_zuNDpxTHJiLiRa8^}oaA=g&t*2A
z7bCoWqNjU=SQHjENP5c$da!Q5{wTk`LEBU0#Y4gc2N+7UKBer!KPBju$a22*Ca6jl
zAqONZ?-EI#Q?=)fZha^2{E*dY7U@72^2uEQP;b$ZLbVG4$UWiw@)u>4-l~33$VU!3
z2>)OBSoU#X;tljKcjWcp_#ukWrk>gTxr)#K=p?X7t<2_LK=#1ALLDfMxowx4Nri+w
z&X`OFQ;i;6I1CNvA_|b`<2c=p41qM#t#fb?I7a|CluIC-0xy4newe*>&gN2-^MPN}
z!kYzYFe3)k_<#_c%GUucR9X=OKqLdaY-hfXz+#(kU!8qTl~y|OcY#{jo~jV=Hb6qh
ze{fX!C7U)dmaDMOYa@bLZu+p{5aB^iml=S8OO4N#^~O?*w8wKwiW0d33ys}HQ0}Ln
zpW1-}iRUv++`+c{pLzp$+_9K>_tf@fnOtr_NLtZ4Y5_U@CJ!@O=h{{kST?~IyyfWC
zzVVUXXI=qH0buB*Xu+>R;9*ADhKMU@C}}WNJwq%M`|p<Zl{2y8qm+KeKeRZRV@GB-
zhJ#*H?^y3)c73+{U_k?{xn59F0lCJ8BQJP<U%!7h9Lbb0S|Ad#m|y}6C84WJ0b1zL
z&zu|-&})^E!2-KZG)w*oKE9Z@w;&woi-87e9IYo%kq>X5X6XP{HxubKAmbYV1&^z%
zt1a*?0Jd|_LjnOccXfNa_i(k7h{rxtBD5X&3IWD-WODL%b=WHbU_`^IYhch;x8L|t
zO6ohk>?J}QxW(@K`#_cip|rMj%?v+&Kn8KqD0C1i#csMAC>owPSZNmRc=1e!$0bt1
zCI>!A?i7CXRr)ITgV}+YliJVk5h(wKPctPZUL3;}!}M7d0ekRn-(cs?8bR&Q=zLBH
zP>1Xtc+zFhf6V5GDd&!ijPHxp{X%zB3eLEJtJ$3+dNT|Fb?9HBw%(qd`DmzhygkRx
zlDW^!)JEJ=nbYn4`zf{gI%j2Y)DU!mmyApKYaq2^w^>NGX}tVfXmxaaoWO1O;?=8H
zKz2uxKf6*<RkfZKr!A5{&gy;R8q2KF0m7E%)kXgV+Mf#54-yi3cQ=<=B_O=V2u|cF
zF#_cb6k1ouN6ckS&B=)m)La>%Mk~L+y4>~w^$jo`NCv=e(B6C<unMYYt@4@zG^ANz
zW21mg|4_f3lP6$+()X(>go0%#ti#pT@o75~wFC~(j*SXOsy|YgOpyC}5e+rV=Z%EC
zj#TG#bPAM_Su((^wC@@PKB3P>hnXPI(|67UnO5em55NH!F+VEF^<7CEe_e1ByAnp{
z+xUyoZ`smK5?mgx`bfNV2yogrR#K8@J)@V)bg}>q0Ijo-e9bQ;UjX!$DYTFZ@yEIK
zs=jsyjXm8$zMRX;`Co&M>_dXv2=euSa;#X`A4?-tweW}}4_VxLh-*D7_{*0c8dcUx
zGsfU90^4=LZ9RoAvaZTw<trVbkT*YAgmqHXg`S<c8r;v>sqR~$Am$DX4AdWX5PJfT
zk{cjAdH7k3^yB-PDjV!(u=4MGt1y1P_j@=hW8vZHDOmwolWz@l!!DcTDHTE>Rk<38
zp1mUMM^FMrWp<l?87K)t5c;qK!xCGmtcS+LB=0!O0=)MQ%-Ak?2p{cIRKj(LH@7f>
zf`_+htf)0yF@T1Nghy0zkF(gnD6!3jS=ZEyet5R|>5XNk|GuG+>e%zE>Q6v=vogGh
zQ@5`fu-e;aSpIxYVvG5vlcQ~bxcP<I?8r9Q{|_6Gg1<O<*Al*qiGkC-AjzP5YH^lR
z2PiDsL9-bZmv*SiMh8R-Q4x_xyjqtIhI0kcz=VhaA0OYydtRziOiWCfG4;zA6u7JN
zeVsjCASHji1hiUW<h}ibfH)%sO!<Hr7_-1*Vq#)A;3Saegn^A7vGh=ELH^H2atz>7
zDMKi05dUhy?%-6uEuFi<xl2D+-!lijql@xW<P>f5ZXliYB!knAQ6y%s3?%ALk7Hbu
zy!A*D7<q0%H+xhyygkkR$0d?#HOmkbET&o#P6cH((HmYBPk2LOD{SsW#5C;#k^nb_
z9Pnf>=<q@fmH55YIJ2SVLrXv~G-dngaa#3I`GBuIMnN)pzA^_hBC(LDY<W8lIs!V4
zSr7FFbouv|nwliQY0ns3*&r4C4(^uNvovXG=?06WHN#D>3YlzCm~z~ulf$EmSvvXr
z;O`ilnzgmDLc;s~5+fs{qalK^VM2HDpxtx;<Y!NxKu>_i#AmqkpqRz&TVUxkJiLh1
zRBc{S9nhcJZ?_IWOSX^KnWY>fazA;d_DnCm{4He@-D^}(rQ*8_B)*7YC*2P;ELLcK
z`FiuZN0Z(k`)9i9m46*~RdMV1e4?xU>D2p;B}81${7UR__kEeOD0RwrYby(5Wv%0{
zErr?xAS=4`)Th5VqC#`?9(%k^42~TCdUJeeFUDM7*Et~#U>}SdLv{<g0l&=>AY?^>
z=_dtfZL(aW0S*V*{gvAq{)$W8zktx{S16k_;aTrWpJecwEmxG-uX7roAWS_wLsp}$
zlO-@ao9@9~W})5<XhiJ}mV~`;PgNSBo1>2p`IMAYAE_G$9UUDU(15lBM>47Bz5Q#u
z{qJnlO3TXNZS896&h4B+X=rH!fv23Fnc0JRC$N}7mywa_`eP4RkmeQ7(a{k!sss?t
z3w%|8e;g?2|MQvqrPme+vYX}gb(c<ep_>B)U|?qiTtiIR6YU`}+`F`fYpk%mJ~HmR
zl!Ryr8)`~K6rZ7|DCv@)!k<5hdGT^P7>(hD>N%ZnVQuZ!E5vuoh@T25f`QuO%gKum
z`|u+#Dg93tvTyIIC8#5L*(h-a*4>V5*^U<+22+S4`x4Gmf#K$F*!o=TDz!Mx=QRAC
z@nR<U1zzMh^!x=#!o;6L;y|A|^O>l)5$A5Gn?>Fn4HwTh1EIZ7p@;U{GXyTZQ#niI
zFyRdS`#6o4J%vBxEm=zp76yCH9DI6SSUGW&AYDv5c<&RvA}#PnnhU>nw?F?#mEH_j
zl%`<JwytzsB@&4BH!Vo5MA!d%A?fYgB4?+;e<L_W=IkMNXH(ga<TiKJ_Ay?lUya*Y
z5~VzC-<ol-RNkjdNEHo$!F9kDKq`ed28d)H+~$A-)_%jec7Pq|{2}XtoKJZ$-ND2N
zGaRN-rOaat!atnjuKFo2Sp`pA3ZKiP17MlN3i^lj4GkqFCBj+q3v(c8Qg!DIQEzzW
z3VjO>CIp%-u-&OxSxaY1fP-CMsz4G*4;_H8_v(W<uro8o$%c4-Sn^Lly_=LsICZB$
zPvwPY8*I(XFXlo3?pLvj7mG;owitpnIZX0rUxYlTP!r8+JlR7F7-JxN@3Sf@LdVC1
zdl}`TEaGg;7$-leo4&-S&@qxmIznfxW_Bu+U8wFgNiHPBc>D!@PX(3EE)+BF%dZLX
z-k^rS9xGF~W^_hB+Ut169glR)wj5{ZO81;h=rOTEv5VbO1kGz4&Pu)emjwR|Hn~F8
zvsPfJ5b{zxb6wZ#hT!^Ff_%=w84~pV6=wZija~&QKi@ybB0+h{rd7mY2~DxU0Vfto
zpdd=oudQQ*M%VxRQF6auYVqqe0LMYc1#))g1SCUbT_72&5VSU0myh*>S^e2J5xiLP
zQ}1jd9slU#0+woLx(rz^wOp&tIbB)BJExQY{YDM;@IP5Q(Pn&|5{QsQ_fOrEY?>Xc
zYtD?7)bPugmeg5*Tz<w(=Q^zDin1DRMIWX)2GXg&Y|6cU8K>30$E&niqVE#(O5Hg-
z+ejjAkG!l5zcY86Y1!iR6b|pVvR!f$8tHt{I%ku3+#9b?aX-p+;~}yIZe*MH1u0$6
z|DQ|H;)@QR6LCx+9~pVyvBn~u>03JVr~fA0;6PPBQAHoynPlV0Wc_+$cWWGcQn5c-
z`)2P&k>#Rah*+ENdg^;v=^D27asB<>HIq(bvcJFoQT|+r7!@!x0p^lKpC43rQ=Rkk
zDWGQrKa};6;KAn5+S>Z^7C5aX3wZo<46<2lM7s_Q2!KE6=UR-eQfi0=6-}5RboCOh
z!Toe)tvd>M;z-fre){z3WN)7MKl<P9j8eUQVd44+FvEg-NWVdj*-K#XQC8Lj_>_U_
zD0?Dob+>CvdNY)Sr*)>n0)At$NeJW;?_t0?!IVLZet(9b88AxS{=}#hq&zr!+qK~t
zuE|I0WIl7Q$Rf9gX7cqs*$CH>BWi4U2}V9f3)YcU4z&s=5N#jE$--tC7(0wRvX$Kn
zzF$P6t@hz7w%NVPQY19tbkR7dq@$z975&<%EMPx>(?H1M@#4ImC$(B2f;#OZ1s4r*
zjC#PgD&l?-jp14w7EX0G6PEv`C~BmN!HoYT6%C&UKv40kzx<+2I{gGxfkOkM89iRX
zn2-OG^7k~I?{p$Ed2hO1A-dGtN$5~9Fa&cgxMuWR`}z7Zz>*rSw(^0T@(9>c9RO1h
zm705y^)Pte17bSvlP4`8+3p43-^4(dPZ?ariAO{=A7hy)!Tk*qdGL_pw)F%3{fi;s
z^7(BrOQ*^EK|cwu6qT7hz2(*bsU%KVk82fQ{@~IrQ04blt@kIX{OOx#9T<QL$AFc$
znMF1gu$^N6a&s~w+dt1p&uy-Vvz1q_>ne!ztI8)^U8MT_w0+(}e=X~eq8Ae%2e&EH
z$J!I_%gq?_Y>JFThrHvG=%3b*x0p310%NI*Gli=w_TxtWVC1`N?(?R5LYF4?NN2y=
z5C7REi!80~4eVjBk7Q_RU};JJ+$QKvsMcfK)*o=-cun3g{K44{wKGy7;k1*ImszcY
z&*RL%#G-eEsOVtq;NWrZZP}op0k|uHQB?$}jM$%J?(OYi5V870U?rXs5q*ssJ&j@p
zW4#^L2WWVC6E{b*M{68R!0SME4;)ebh}ev6LCk!N`CL&Cm>%gqG>Q?h`9B8*AwTwm
zPmu}gHe?{r7&o7;vLOeq2*$vmft!aXecZg}daq8c(US*gqI2YC+0He=n}FoS0IV@M
zNIxLu{KNC}^{;fPOW+7RSi2~DW)hQwWoNc1#E`^m`koh@yUWQ5{R&T1zx+-KjM`Sj
zx@!`J`$D5KQ=8Bj@D_Ze&G=ZJsWjDA2!Z=hj`PR@QUjJC2Buo1j13ywD{t+r(@Wnb
z!I0ACW63%fvW`}#Kx9zBLr=+yZH8iAbdC7<J+I%P82UtSs$$1vHNEPHJ3_Es?arCn
z$7r_k@ud2yc6G84Nk&=|=1vkd{Lb9+1&<0FeGhZ<a2wJ2U+DwOO-xK%vwt&tf5QqB
z;^_E<;`?vLEfUI(#2drDuUvQKFmx`?=F{DK9ifV-Bh6J)GlHcddZYgiO1YZW&IiT!
zyLaz!fbq>X*jk$k-Zb7yptgKS^uXd*gjx*LOifL(08=(#AfUc^JBgqA{=M?+1MPa(
z63^L%g+`fcd;$Uu9%8UOKk>hdRrX&eWH34_#HT>FxA*Fyes#F><TXjcuLa>aG*cce
z?d%3iW^_cWB8ow{Im|D|wCE9=E6o1H#ny11Qt-VIr7x_^tLN6JTYWa}N&@#Z_(y(H
zsg|X`lLRo0dG$=QDXJ`FMD!xv7-#8VT`-qsiqE3NVI(olMnHnwJ5@;VK4fY>Vgs${
ze)hKzdw8+k?FR3e%oyk2!Nh^a$bki#B<`xqmu|U}gNbL5>85n{wfnHi2ury&K7Y}_
z7*<6IOCe?J3^VBQIxoFzhUX;Xe%BZCE|QJS&d&DtU~kl2dJ5rq6?QrdftXi|l1cyc
zQ4uI3B_t%^2r&YWh-#H8%wGZ~-Xa!?>6EM_IHq&nE7<J^^P}YSxAwwnaIi8)*ID{S
z&}4_YNzlrUF1>~yV=>QnEH=4D{GIpB3&}63NH?mYctsi<=-`I4qwgJwTnZdE>%6jD
zs<!5Rdc{|x?Z5~lY{N%5Q`w?fF34w_&5blkOS-gk`{>AJy4oa$(XYCOq|ix|WAL%w
zKhDx6NkG?-*Rx=lTSr%TwB(8vO^8Y|#L@uu=gB8)biKA=`qS6Z)`NLwzOjrYy&Hp1
zxJiSP&>Is)P;Gzo2=Z)9_~ek5>885POWJA3UgELr?tO2F@Hj;{A&YqD=ZUWK#M@A^
zm7{L{w4T)W#M~#l&VdbgEPB{y`9Wp)4~uF#H<yy)>UcmwM#`V_w?$;`XMy(8?kJS~
zfP%qC$D(_Cdyj`UId|AJc~3yAkqHUiQL=(xyA4}`^D$stsofGj#BymcWyN^4sj=OV
zojzdrD|7Vwfx8*`W?lK<G##y97xI)kx6HoRi%@cL$@u-k7S4h+{XwA&Yg03E3<xpR
zoToq3vXx`VO;owgm}cO%qaZygh2t6aXN0;63^|JB<WH41!<Aji#kaRU47XTbydYDi
z^+l~7#U-;UEH}8=e)q={!BQ%49~UG+Ul`$tjpTyTCts%y$(0smmfdvUN8r0CnK8df
z=fSZ4`OK}z=n-K2a>aFXr?zs1y<=h3f10wlt(3pI@y!Wb+a~X<1R(9&jnt;2WTgpU
zkbt60Eq|iHux;!uEjfl4*uW_>!wHt|!P1bS`09nl12=CG0A?B?t3tOYqer_S9BXh8
zM0JD1i|F1!=GXGMMK?h33wZtWfN$3sQy0o+cIWUqs$*(qG;(T9sv&4s?yYp`EUg4!
z0L|<Bo4^Z*s$`rjV^-9K;a1%xo3TuJThhk&Jpp{R#SDs~Qn}j!x|zh#XPIy?FoAmF
zmK*LPE`^5p`nAXM@JQVV<#RsbD?w5$+9-Px4UYN+`~`Dl7$g1OBW^QYO|}cBXEVI?
z?ZZ(#uH^^6MsF(^gD0vYZ(cN})xV_&4ncl2>9$4C;Ay*-+`yaPNXE!*(Qp{<R6<Qh
zNCeo5^P{lj?%!8jvB0O^SpA}!P0IY!=})Kr@`JDmmV5;6+2Z?r<{ho0r1$Ip=GTB7
z<_UHbsen7XI&4ICKY&1+f{#ywM~;U1`9rXHFkw76IM6*<YEkYt0ub6eF)@&qLof3?
z@IY2#!L}9RDRfDdC2^ng>p)ZYvS7q~I%g^1ZM6>lkvEr_AYq6SR+Uq+$;S7R9<KfO
zq9S^`QS9*KTA$0@*fFgoCQDqD`8+qi+fA8T>8U-sN&R%a3U?FxD8ER5f2Fw@Qf#sn
z2};+B2~bdKPTF0vE}9|D_)6M`3BB5H+g_v@`gvC|aWZ=?6#mo;skXYcPV!BK7z5Bn
z*Aspj)-@bQPh&uzS6cm@D8zkvOx}aBjD6?HWIi^IY3zc?$1(f2kaqdva%T76z>kMn
zfT|76+iPpTgYg*QvIIor<Zu=}k1=^u3O)M(^6Cok9L91(K&XG80udPH%K$4+kYSs2
zg;QdX@{XC?fze5GSDg1QG_|cyQS5z9wx;g2koafO)cNG)<@)}9Ny{T8<IVd&#@;%t
zs_u&#MFj%{R3s!-T3S*-P*4=4QyP`-ZrCUw4JsfYUDDlM0@5YjAl)Sm-#q&MzI&hh
z*Y$Z+Ji<A9uRYgXGsYYPw~Cx3X;;DJUVdelNm2gZW)9s@$Oo8K!bKitHN8!>8`brX
z;*_DkUPF@Dp3C_ww>5JDT~qTTZfBxwhQ0Zj_Sk6#3o>+8eV~-gdM3?Y+E6ubRrb~W
z@`Nca*Mo#&GFrh7f8`Fvuf)MH3-af|^KtHx{_Jif%gIGc8V_U>hxZ-&3yk&}d2O8!
z3bs2|o{D`gVFr^s^^>3}etweFp77W6!1h^=uf+mZ_Mh7WSU|#wqCm#Stb}l6b`}W|
zWIlIo`9h6qY^6VHj}@c<ZXFt_n9O#LTd<GB$j+DSbY@$=hL&ij`UGIo6roQI;Z_)t
zlAj$l>^fKy{a%{xpbK{(_@nz|nlsB?y!P;7^3>4*8&LS`x#3DVbCZ?m#uSM$Xg`w@
zK{js_s|NH^+HH#tg>Nt1*VpMq#I)2Oiu2-3s@HY)#CQ=%Ql}y*hO>m)^A?8VuE|8a
z6E%vD)SET2_K)Hi7^jEQCp-M!JiiiVqBLUdZ+6*Kmsx@}4?|y`^vL##h|AnsT&IbE
zn4~Pt*mOMOj9%chK(|_3vTzSo^ea}DkAJyQZzS~OfiKa4du-pFMjB9^g+rh_V<2PK
zZ6|`TUU+4|M%AU91`sK8kcL(64XDnofo6!%RkvS@3?aV>8JI4Wmqk#drtsMh$wnzZ
zJMz@7yy%ian%<jwhu4a<qoVlD%Bq4kxrArpaI3Kvu4?k2(MKHLfVtU<0E*f;C6U>w
zdHIw=UP1FK6Ub9nV@ErTEw6W4u1j+#%v`NXfg1E2VTM*pSEtXN!)Z*?HE-kJwOs(6
zA;*-IUe!UhX@Okkh(#0HQmfXg`R2TkmS?+v@LG3R3%9Eqie`*()z#>Eze$`-6RGU2
zn~#4=HK7d&Fvki^l&RA)JD7jiGaA@e_zr95-8GbFMZi@CI5KQrEY$D$0LRQ$BcbM}
z@bI5dKW8Pn4qTDFFepq)YB=najpOU<8~)gd3o^`PSR38OmttZwKR;83PUt<TqLPJa
zCp9%=+Zb#Kv+hCuMJ|~?Q8077`*YTkk6okhRlh#RTOsE6rvCR4hSS4|{;ciib%w7n
z5RbCDK#tB6+i(FXufft+qxa$q$#1lmO$OeqT0;-RZ3Y09dy9U3XAgOfq{OxI3if;Z
z`#;!gudr?`*aZLSu(uv=Q{%tU0Y6~*W<wc7C0OSP;`0jXvO-x*)-k`&Zmx4|416DG
z7UhvqVJuE!#St^Qu19i9O8r?%pF){EqUO@}s)L66(JSS<or%7lorW2aB%~cQ)4wsk
z?O!inU;iOT7#fgGAYiMmK1~9#K*9;L8JbQF?XF{$+CAHfitiG>Cz`jVk2HidumV&h
zHqQ4PU6Mz+FS*xApH(Ye?cwP+dSBuXq(oUNMF|QSilGx5zkkb5)W;_!{YZGnglJTT
zCnVfI+X_O$$!4MDcCw;*o^C5HBXhH?tW1|8F+TLR>K`6%5y$b4y;rX}GL1%Cy_={{
z-{$8}t2dLUhSCRoc4bSJ6jU0;SrZzKS4U$d_9wD-G%mv5$!qFb(pA4?Fu*B&*rY9_
z^!h`VJDHkXczT<K<(F5-S5%LN(vT3MCM*o?6z!SCX=7!6b{!2-C9cc^m7kKWrDtrM
z5g|hPiB0(tU1$QWju>V1%G(u{7na0++j$%W!)>EouDT$MZ{qlDRyu6|+|$$MH3+kI
zFInT$6o6gDl7Lu`k7P=L7~34&84B|YwY%wg-RbLh!1e;Zv`8Nz%~_nf0biTwf(#_D
z1`roDVPZ6WOO+quRVfZPI3&5mAYe;GV#?){M&8*h-&k);6Wo&C%%-yW4G)|X*e?e&
z_w)CI#MZoIz}Y1L4ATTGLj*)bKcP>?P}QaVaI1wj7Pz{+5)%`jHygYxE+)|M9q|rl
zjUQa<(dS6*NA~j9<4T_s`G4F_@M#ipSFypWCKso<6{vYvcl6o}Zu6(eD9`LTq7<5J
z`{M2R4mZg{ieCH|T^}RZ9N0IF(+UOd_G7YV_l6RYUl1rdFEe~-arlxH_Lvp?1?vR_
z`R3cU4Gi63aTlWXxj8aBtL)6sl_+p^C_Cz#JCl$f(gwdu9V-%hIoi)T?F3nL8hmpw
zlr;CnG_g7-Otm#dm5vxkRn4oDgR1<b=9+KrwZTSWCCbe!da5H6L}-!Gw_-D=PW<~k
z0oFa0(~!duE2sAcchR?rb4@;rj)@8Kfcbfc$B@4KbW*=ZTqYtnb-G@$@w2m&=KlTq
z86-v@7ESycM_TW3QtjwL2%)_R$B;g82XgJr<={?vjj5uZp)q+Rq+etv&CfVuk1`BL
zM5(etQKH9FyO)IuuVD7}sP6f*4Ikk$t5$PkGRDXr2L?m#T&x--|EA;|ohNXN>V&)D
zJwhv__?4_EsKi1M@UAVBLX?MKe<9!jQY#|Sb4b<OMT^?+qxW4CHv0Cv7h2v74NdY3
z=9~(tJRqU9nqrpPxztD~RE#qj$dud1HD=V^(1rB^R}ZsuK{#7O*`re}Imc7?Ue*H_
zCXTfsA^Odjb{T@jf=wRRI;J^0hfCcgyF}M78=IIU*E&V^k?hX7*w4X=t=J&?9K^;Z
zzl*xV`*+W3;^N{eYmLBMs!lt0?1}>d0(Jul6FwVp^k+i#^MX9@PVwrO-pWdTP{_cp
z^`dZnts|uc@kT$XdR}CJ8H*C;uZ|+~Se9K&ukR||sd2SvG7h<XWaIPl5N$sJHw_Hq
zq^4&Y*wJU${gp;W#;b@)4=n+Vb1dr58_)k#dr99kP=`}9^|e2UGh*m`PqDcD=w@&a
z`UM=;dL{bZ!x4rqnQCC_smr$ZE%}}|+YswBVf(>LAi~!#2m|Bk--RbRAo!EiAji}Z
z>r>7o;lrKkhhLDv7b+q#`aVADbrr7d_lDEzd5qTHy3fYO2-&IuDy<S8!Cn^nH89&+
z7O6VLd4aE$yh3{C?cXgncAuj`gu@zBXzU@e1o*N#X_t4$-_{9*yj45ul1GPj*D;e{
z@=&9Ra48h$%q6`yHEkW7Jh^v2@VDGG=fh3L7>%9Fof9QWliN5A<(5Zb?wdON&(H^5
z2y(9JXd+N{2_p77hajl9OZ5_$p?x*}_5HGHij0B)*@KlOoY~Viwv0-W8dED<H~EW>
zY)HpQs<S@&<?yXtvAbtDdhK;2Ixc61)2HiXnPt8sBEmzZd|9#DWH#rcXvww5<M>3&
z*edM418q4^d_L#CI97Z#%WH3MS1gj5VE*FUb}3TCHTxiLeCbA{R9_&|=0<P%`tjc?
z*_hB7{;`8_inQJa`s4m?dhb#BM1%T=$eA-r(P3$@-V^JBxuxZU8VTJE%zMP{e#c!~
z5WmuSHk@0#vAGTs+MTx!lIz-eT+jgaq&&;zgi+JF?^n)tcpwC_EIamPu)*Clxrjt{
zK7Xgz;B7}M&&Q#=Aevlil=HESEHt8mlc~604T}O@*bMg}OG^H-r&MS}4fp#BP13a6
zKWeHo+j+9Fs;*7=eSAx~=H)T%iq}qVMn3Qj*7(6=Ye8>0-+h}r?n?tSZx-~W&JFez
z9<@&lzpwZnC_Pg{c*)_YoQyW_>X{vK!L6>3;+akW#^~QX3*53$BK@XfO6H*dZq!X7
z4jLNw&zk3_Rk8<8dHB{@4UB3e_?5kvKSaHRQp0S<p8?evZb~4oqNJpu_iW8!Zt(pv
zg8xUHj{R{MwHqcX<m0kc(7gVxd+Bg5dLz_K!{AaKbCFc9Bz6wQQ@2Y9;TcbJM1Yr3
zjoIUJ=fc9P>D+Z7!|YT)KsSkY?)#yy3@s^Es@%YLfb#6|1qdZ`8OY|Fo{BOFZVpA`
z6uiAabcgL4IOl$xn%TTM^t}!AYo41UWg{$#cGYOQQJ(ydt7vhOJJjzLjY|DXB`s7-
znNS+DJw(yN)*i^O@Mp>TJnbp#7gqZ>)fmuhkrS-%qs9bgP!+LQD4-0vzY(^zIil>{
zk>lvK=p1e#5>b}^;X>EA)aK*yjK@taoXDrpPu~lhh_srF6|P`In&1{qk;SChof1!k
zEn>@}mgCHx>XXkp>|c<(cwc)teW(R}v`?#5lqV#)T@o=}dnvYy-)ZvLw!tzdFgOeg
zIbOv`R%rPuZ@<n#+MZs%7^`;OY9=J2v{B?^)t?(=4HZxm3Ki)H$Bq5&@Y*Pu_D?~|
zRG4=1YmJDDml!472&IHA+3NQHX#Mohn~zQXFXojow0>;wd1J=bVt%mqPBiXef0dVU
zgcc!C-U@XtiDySuKH14O$^1H*rPbY*xM?~bQp8@<WY0nJJw}$`36_;Rjjh;wmMo&c
zg(uGU(1=yB6X*Fwe|pDoHVGCidFQ{8%o%w&K6dw~&+d@6Yy#ev!G@VDBbK@V@hbo_
zSzCBUJZ-J<W3KNK&1JsBTk*t6Z!=fHH(1FJnUdzx4kJL0?Q3_r%sEIV{H9lPUM%ut
z&etynO!ZB~rVLirr4tIyZGV2PV{3`W_qRM#r-*%rx^>1;#zE`i6>eJc@I7?c1N}!l
zDoQWNkN@Dip6>6Hs+L_LP;+CM37swM{@{nLy)^ACK-zX8`eu=c#Ce#A<zNA25+~yR
zZ-Hib3GnS<C|e<eQRRD|0^3Tn*f(yoPT)6Q-h;x6g>Zy;)TEOLJ+F;j38Nx~6i;;9
zVU?a~#YL7U3!keD1bO_6a$F>Ki}8q<@#~Z8SDFlrP4oph{880p+rSaxSz>Gv=JlY9
zFR(onktNrFBl{=D0t@>wh(Y(ojl;>VlL%ZozsB+D_+7S%nD2!~2S#m|i^S;B1!vCn
zNIu7`YbD-a-~UMx=FC)jP|m(}tn{ejkm&fU@l2}^RxbVAkp)a_1+2Z8OsUVavq;^s
zFkUq&J3nn+6sQ)g5Y&(wi+OpyqA!O;Zy_mp_H=yle9Q%k5?m`$$p3bI>~o9>KrIbA
zStvrca?f|im#&lH0|Aa$2F5FNI*jzyAJ-YQZH@A+1|FK$-A0@F^Xyi|%I%12259!h
zo8Q-Jr7s@aX3Djn6p~)&eT>da|2T-LCw~=);jY?KhSrFvJUVkoOkol?!GP+U8o3(%
z2gc}LjRYTr%%ha9KsIMC?gE7qCf-gD{#UIPExU%!c##MAVCy|>l_U;n!#^l|sM*=D
zg?-2gHSJDDO5zNf3A8+`&pLQZ-kzj%$pF{c$>-OW*}AeJ!9kFMuXjXtC#V8cz%uWz
zG$WK{9k;WeLILfm=Wue>+jA8j_7?{J^_hs_QP+gX2*{yCRtSM&(B)b{H)it81*H`1
z#Uy^X>>8N<n6^mF;814N^LML;-S8o|())d42+ay9h!xLNAP%eR95^Z)d1MgUIQZXU
z)?X*6S`n9!C{iScD0R8EZd`QuXI{O+?PNAm{wSs(8$b3t5>zqOFMA_k{_a-a)8zW`
zF1(s4v=d&8UuoxQ?k|R|l1ljVdWizDPe@N4Bo*z)L;uX#SceTp|1(#i*Yyb(b0Q-n
zi?-nfnJa%JRSHc5Z&Tw7#Xl|kJ+4NGiDtgPpD$}$67%K5+doK=XG5DEK$+y0i1f_3
z9VLi<kV#!{8i=1ql&d5)vWJF76Qt-aM$0UP8D~fSezdJW@oczMI+YPqH=F-)($}fq
za~}c*26$Wk=8Rw;Cav{N0dQ}`9%9d7;hva%5LW)$I1+Y^F6)}<i!hw$J6oRyhaVwK
z4hCgnT|+NUrW>Ekt=vSe+TLSfid@?n*xdTWUFu9#j5L)R27DWE;S>6NAy7$!BO4zt
zS8B;+^!t03Sr#PV<jtF!d3R+kR?dUN|0;pc-rdE}fSBvK=h&#+@axa1w=Z;9+`rN}
zH{DZDQu7HdW6m~%r|Rr(8;x^Z`$$zAUiI=tijTE;Oj;f}+k4Nj!YM9ldelY7{Qj=J
z)scu;Ju28T57=_wSc^AyMjB5R7N5|lJ@}DQB`9&xrQ`Zrva|U2rJOq+ePyj-kNjy4
z?%i~ontq%a{=Ww?7Nh-{6LmR76r>+DU%tu~X=)guBI++Ry?}|mxrKqY?z;7!cQ4H3
zIx~gO+`^LPTaJjS!zZORA7>1#m@1KBoB`@LYSuoSeKoZmlUV1>?-ka2m(KWx3iK(L
zQi8RtFg=ccF}3hNW~@f9tql`-pZ+G665PKVA-UFgdrs(J=ZR!o)~sG*M(~u-1-RJ4
ziVCZ$-}HY?@!1`h&w-^?dfwv>zPo-czf<P47W9K&$=O|WEKWTA#5&f7sC>R=;rHbZ
z9EcWvvS|v}k>ib26ArPP=9*;8kt|&G$lEE|d+%L%yO*jp_QDn=eo80k?VY=T&Z-;z
z1x^4bq`G~Ucw_mBx~c6I>tedbGl?u2s|`NnMY2r8IdlEz)e%}{eK+KV@Y|bfmp}Hp
znum>Ss0Nk$FAG2EvNf+p&~zK=a*{_zf-k(9=@}YkJ<$x}u&-13m)vQHxPJ!2-L84h
ze)y)fnN&H?58|xP#F0OtFb_=(qMEAnOBC470sZF(QJEfvA0?Hq7I50X2l$850hF&~
zoH`m*tR|iCLfR%KeLm`3<0KqozFjMLfBFdlIaAf0;~s=v?7}ve<v~R>q*`9kAFnIU
zZaR3%%?&Y^<*wJ`ejM!8zW1Ah6ql=Ro;gSwq{U}UKl1JHB9v^{BCQ<ddbeAdV*2fE
zs#v1#ShHM+G7gNRgtD8b<sL=*25j!>*mDz$)2t)sY~MVK%kLdxqH<R6QK>hI<dXth
z_edrY4M|rOoU!Q(eWSGt>QgD?n<h%(c-Lo7YCFF6)sSj!d6)Q1xF*EQnaKk@Hm{&y
z7Pu!%Y;d@R+)MP$t*xPo5sH{Sj2KOl2InRP4TlCrqvnwa&3MhPXf@w!9}9H1k)0Gy
z#eEbfi)wS9WY@Ce4Uc=if0z8|K$&zb8X-BlZCbjWu=<0bHFK<7Hgw&IBQjBNHGU!S
zQU=+?O5W3<N5civ9XdqS{$AKEtyaGpBpd=)U&SV}m*w&$I&;}-O$Pd(F5X#<CToA0
zf5zij&T23xt|+PgtbS{U^^j_JU;c=w#O^Y#s?KH0w``>C9-4CZ646z&$P(uTM4ORK
zQ*0N^zqx$E=LfCyQyE@*!L|a|R36ncn8ZcPH9lqLnz~_Sb@D$j!8@pLP4~=0=u9oB
zq0<=R$<MQ4Ahf<R%-X{J5GuSv9UHbC(<1tt7jwWfwt4z5`j~mm8%I`rOU-JY9z~za
zxP@aXId-^Owd-hngBTed-57i_KPNBy4lmyG68fY5N@0bON5bq(Eep)S(c$ro0*V2-
zSB|F%Z`Br;3Er$&;c&K6K1}z3NbNLb{Kfu@brF}~&TQL1wzFGveUEajNorUoUiM^L
z4~1&X9RxO%vn_`9*kgQSd-u<#y;WH6NMyP?_{ff}&0ju846opPTeh?8uI;<f-E>X2
zIp20($^P=u3}Nqf6BUyOKF0yyF`&g>8iCwc?gtu^6Z?Z;gQO;!TSa?3kctOKet9mc
zaTCrXC{HAh*Ghl5Y+c_eI@-O&6&|wqOee9xPv^b-6I#NTjQ2Td@|rHPxWRJpZy)-h
zKmvxNhy^s}K1F`<%+5uJlL6|O2HdGwPfvw&ZE{N(Ye=5F(-VfEVhQ4iJU%JBsvW93
zI#^0PMizT}cjLnH&7TKzxnf6W|2>uTf8XaOzi+#5puZl+_(@I9VqeAm+Y_;CM=Z04
zduwyzDvx%)8wA^Cj8wZg38$Ek!ZR{M0Qe%Dau?DLUpr~`O80LT3n5M<yrGe@8irr<
zr*-rka(jv9*o<0Agsq(UI7&_qxUV_sU4aKqAC~|3H1huV?qQzQnTpPFnc|eZ*Nka|
z*NK)B@wMhIyn>b6CL1_KhQ9qk9NpZ|a0#>rL_{ugafODh7MqT@R_#w9Z64R~4gjj1
zU2(p<)W@h&=q(Y+WdWkjK%m%AYB7sy7HsWiI?IXy4A;lckCKx!G&7SCg>47$-^-UT
zr$FjJ2f8XyUqC_Elr&u$0rXDK%={hRXbdC-Q9XZPn9?BR=I4(h<Fhqi>ZLw8$&TH;
zdFM_Gk`n9MI^Pl5)_k2pAPOZ7*4@2A;S}8R1K#~LD69dVszHmjyTZ;a#`VYm5a;eH
z>iZcK9X`sD%_|~C|IFNiGPESQTd8kn99?@UM7i9wSYdyODZqlSxcpzdFr#b9`7p-+
z6@BZs8{=c`SP6=C2fC~bUw<OlUS|652U}WvK2yHP2QUWoie8+boidPdL#hj9HPG+K
zHm_*Wo(Wabcmv=)eA6I*f2xNMznp{07a-Qd$R9hZ5rST8V&Bqhz!`vESB80PomQ<D
zD=R1l{aGJZovsGJoyGDmdQ@@U(ecOz<9GVeBov1=9G4Jk0^seR{sRILC`CrF^k*=P
zT3uZ|2V*=Tz;e_H=oyC90*4)4F0%>oAo9mUpaBsW7&u6l^I*Oyb7{-)D!1M{LGY^y
zF{KBk3?wa$_!}=7WR3`=_%Qa#-MoTog<piMElAdFGknsJwTQh<{js1V0(tV+6<<Z}
zWuh!rCo#uM=e=W@yU3niCCMT%nq1NXghc2>Y;Z7)J&>4NSj6jf`6_iWz0P%W*qOgc
zO4<(6cTeBExhInNd3DuT;CPi?OiT;~WdnG%0yzTagX*)WRXO4(_IS5wzXe2w@o>H;
zfMajL4@a=;L5I9fAK{Rdm;V9(I;c567L>Qq*SiVmnGYCPxdjF90N<#sr}vJ77^SQV
zs1W!z4Q?x2cx!+n2ZMd(Bj>L0j2x?CW}AcXpzI0ap$@-4zc7q;=X|TUXU~ZqVF~h*
zGsQE5^k@JFT6$glP)<qDL+4s9%{4Lq*1LoV$tJVO^e2^zJ%_D>z6`7*xHJduA{fp2
zGDg1+JkdFZ{ug%LWw<9XtZz~$)QZ;0HDx6A^euCF{R~Mtmq?B4X}4ifrdrt-mYS1W
zv?;y|`5+gIrs2HRh>6h+#WD@oL!HXXN*?C}R+yQY+b=6G-`?$$ddEs+{|6b?JA`s%
zcI2t5F7)3}O2UKqM!00W=(V6v#ZdSbB{2-OH{ii48CVF9j_xqZsk&SXI}8@*)uBsq
zaWRU%<@{{F2LF$!q~!cgmmo`5vcrzx=(r6q_<`c{?Cfj<K(l-T0;uWeQk(cQZx*CC
zXWaXF9im9UpW#CpQN<eTgE*P-E6*jpDX(77!KTBy8R^&k!iW1KYizDs&*--$1+i57
z?!J&}i&^U~?>~EZ7j=+Bkg|ub{k2ovrl4M}sh8;RjzX7h7A_DG0bUZ{4femf-8ryz
zq1m5oH8YG=ziY<v7omhUQ>ye>)qp#$z|~p=GR$=TGf-AkH^|G+MAln^X^kuh`bW!c
z7|<w1{NX#ZIzdzx4S?v*o<hcVK*<-h12ND*%hJB#-!m@I+yh1WI_>oDNPGkk)n}5+
z!IhONZiou|2$EjfVWBIq9p!6?G^tDZBDJ9K#OOu%EwseQdcGcW8iere&V|LD-`ymU
zmH)X4UX{UKQ9=jDHut#)F7jk*W+%@AX=9zgfZ4Ob=WHi1J^{@>hZVP)P8z1jH+)v}
z;%A3lh$8WzzI?VKFr$lyHk$&(D7hr)4*N$?RoDhEm}ff9i*jJ=rL#DoBD_+WI}E%s
z7akx{suHc3ZSa_KEZc!V<^Zr{kDybD?Wf#7+kc~-c@r>u3L|vGvU@TdsCH&{|C^UY
zk}=G`^?nPbGE`1pST)-6^2vGZOms0P+Ow~W$El*n_Q}GL)a!rwf(#RPO)p0E3|72X
z#o>G3fKO`v(ZK{xdRaDKGTZ~n5Wbm(h35B{uE2Bz5O7{`qi{XA;09<bwGt_CL~*w*
z+Azx$(oAo%^6^!OneFfH{)Wn5<)0BFM%7|};5@n%*Kg>slpJOcmfN<IDQal=;E5;3
zFT*b}F*r|xX(ZA~<%(-nRa8d)yQ8G%P;u}cm8|!{60;cWLkK2}l`3XT?>KQ*_T%sy
zw?H#$Yd87m6$lEWEKG*s{i&E!!+KXCV|R~n_Z@>i79!%QZKqJNMmVl6yasZaIPs5+
zMqSdw?Mk?~I&aFLa|Z&n>8It^Y0LX8{Y503`}=yZS#E>M?dlL40khgmn7aw*G$H^M
zgIVO{aHoh@kcEY1s!PbsBm`fj%{M=^rL{Hv#<|FkO63uDHw=skU?u^fh0HT*xTIet
zlY;bX@w>0Fb{wtCM(Y#BV!uxk4rVt8|FfTjv%*CJkAe~g(ysU6&O0B%-6%r$9`R1}
zbvU;4_Z)1^UH~Omv?+kAf^Q5(D%Q(0dhev5z{F>29$5k~m<DrJpvd|F&f9n33E^Cw
zaN4L+S)0s?+%C;luP7d9UL8{|Glbn@rc-P-iSuMITb+baSNLJYb73v5IL^Nsn5I|N
z->_NKy!eoK`J+Q51>G3Q*B^Kl#~erZ$+%QhSE&CbNFGQkUXe(LdRO4oAcEjxvDA-g
z<)+PEZ(B1k=MAIDioE4)*5%kzz;$wRvN%!2_acxK1z`liYhpJg4Tt33EjS%6X~FWM
z+Qfwhee)v8t)O~+!8<T8yZ<0wIFwTyHD?EK=B1&$cLB%O$@z(-J6Z1EpDF)?Gdw(u
zpgQi-1ZEilN8ZDq{xgo&P&`7`kj3~6)$#MGim%bc^=dtYOE39ddJGT$%fQ?FOq`K#
zmrLXcp2r~M{x+`{maLhb?aAdItUIijrE+4cK4dWJW3^jXg9m=~M49xI!OqSOaF`fR
z!kE=xmswIndvR``5W&Z{__L}E|IrA@h~=m@5YFiWXuasmmrrt(2=ds81Aa6&vpa16
zHnhrM97N8KC((9Shl68Xk9<CSxcFoV*BQc@3D8K!WFmz{vfu4_abdAPgAnFEen6+{
zVT8A@ZvybD!dwj`Fb43bvbIx$!m(RcQ90*5pdP2AO)I2TVrO6smnNp<rf=JR{Wl#D
zy6UHIY=Vk*odxC}Z*!*zeaAV=3CRB=qTo+66n@c<`TfMtPj}uKjI$gxQLB6AD<{Uf
z{JDdMM<ZTaR~O&9l~aB6AT2Eor3-};OjQURg;sX7m5;CQOaS*p>h>9QdmbQqgi7<V
z=O?rPCPrzX`IGS#_p<4D&OZGApjDf^JUl}78MCwCIZ10@hUCI~R7~_~UjDR}r{?=)
zHs2jT#vs)GfPI@AuF5V&lN~M^JRmVGO~CJZoj_AocX1gGTs#m^CiyZwEVtQ%Auocr
zj{st~S_;Rg`=pJBk57rtRWh2lXk-gU1VLf#78zLw?DD0tGI$J^L8%J3XY2g$u=Ch&
z-gUo>oo>a9x?>az6%1{mU%s5sqi+s^sBn&=s0V?(7R0#>$)o^)tKD)7B$L}=D92!;
z@)69;0a$nYmz<OZ0!kSxDm3~<DkzW;<d!JzC77vz7uG>z8{~32L0ZS$);7D(q5nnu
zN6QA?&1LKIZG*A?wDmrekE)l20#d8U_s(stWVWbpr2_`s$+(+$pxQxA`v}kF%{3U}
z;V>8vDYo;P!CzFhy7BMie3s0wdbs%;Ngo_8nj*w;sUT}u&d5DVw@fL(zJ2G(x!&>R
z^X?}AqPxLVeuY!IDfAd*#a1e%;9&9b_NLQ}%i@AlD7{J(hTY!&@aEl5U{X!Wd;orA
z0fyc(ZTO2+Iu?5p?>u%a?6+j4)s+ZR5>73-m;;d*I2K0bd=D54lwU+nW%^v<nLTsa
zy1psHmC&K4yssi7c0ET?pomx_M7?$+_N=u|XSAD8K|^ceBsP>|W8`C9j{m*5!H&rl
z(Z5dGJMwdt-=zi(#Csz6qviZy!;e+|=S_{;+p<=<R<=JH2cz9I)|ObQ{A&fsp6Jva
zkqU9-I?`{>&(A;9Mt<JFA?JThF%6ibs|AO}`^`mZS02ek0N||p&pbxC%5{swQf{*Z
zwKD~GKs~;GWO5UP0ktwdo@sRVVSh7xn(@&Z5_`|bcc@7HW+Q95;uk!gdYtb}U)G$0
zawpiMw`!=79Im4jsn~{3kt({lx9aEqNw#m|G_W1p{E&*rj=L!T1+8%MW$0*X0sa+w
zqG%ACQ=>c=$kS&=2ck}*&tlIPQw3W8Y%|O^A97uA>;7D2bW$j&yFmX{fcFLcI*#GR
zNsl@blY3)V%1ecu^cGhx1gq2&s%trtnCDG*=dYcx>f94??>XT|-7D(SI1CjLmd*8v
zsz|}JgG+8ONSy}(ElM+4qaDM3G2RPKin}1jlcm@O!5}zLN+3YZx8;SO{$X<@fFC%}
zf7A8Vdv|4MrM;j5E`Yjm9|2Cr`cDLQp|KiA*WEkO<SuH7+~CN|aJ*84`O&eVIXb@V
zD~x^&Kg?LKDuGJ?{0sa!Mh?(96n*C*YU-Kcw(a?YyLa7!1dhA`n*qns&z6?F|2`JV
ztYgQWlcD`q=XjTBmp6|WUURkkV&M57>?aOY$QSUW{zdIUMH(Gim;MyLii^TxA6E`u
zh%({X+*8bsDwRGa{oewNaUt-*;{|$BhqjPPCcn#8f0<<*S`MFDeYDK*58^JL-se&U
z75h_ZEw>20P+RQh{e?nw9JTrA_G_fKiA#@@KcnMr$gHuw-G6oq;km5BQ#;$S5~Iz|
z#NU6*ZH*!``}pkNdPBYTaRy7i?cw&^wuzqJ!nR32!g+d4f=2FeUqeG<t^`!+yDA|Y
z9*{&?cIvNj76G@696plCL!4Reus0^R#+I(V$wrE0UNXI(eYEMstGyyKs$aTs`r;ai
zlU~4<)LEtG@Ii&3e!E&IH=MY%=F>q|zo~R)Mt(I^^8DY6pw1w)kB*9>d+?wfS+@n@
zCT8_=^d~R~=diUuQonI_dYtvo<g!d>kMYc3@Y%K>N3cf*)*flkt>nl*_^`277Z>(o
zcDOaAsT4KbqMlMKf>Fgp*J>!=-G%2WdEa)>p!~6O*kK+#+<&IV3vlhr7#J{Jhx5e3
zf=wlhzzqoKwvmfwFuv&Uz65lEoHdI^mdQ|41>VFC7+gB5ALxH4>*5vc+*YDw`5YW_
z%`GyP*7Y8ffs-8_fYFV}3RoF}5~40%wMK7FKkgY6bDFc$ZE6(5?E2rGni|2Uf@2MX
znFq8_F)@KPHG*iM358p8vW<mSVY~Jg0fizoQgBkvlNwcy%6w#CjtK)%RpX1i>yqVx
z6~QwzZ_!rQ);fGc!xYQSBfYZbd_n;&^I!OKLntKwC}bGYp)fNF0q0U@Cm<u78<qPm
z{${+~#(1RA2MB0GC(j_=QboOYYACb=WD~N@w}PywNCPO@5dr+*9{HU>A^Eya(XtPo
zLhf^QnAelpIvBZ9bgkxoV7#@9rP=!Ze7nzzsq3G+gUu&Vx6E#TytkrMeG23lY#;=R
z=Ircj55xNoFb_*lM@LsvCI0f|Rb0W7+@ijlLf#J|{AoIk`P%`<X!+*+k^hmeCy$<)
z(g-$!lh~CMZ;!!c635t>)!I0UE&t#p<7GUpd150pH3i4filc$*Ff)*cxl%pcG%WGr
zrBAix_^E`u>Mct%Tt`VFkLW}P&v&RzjC#)4UCS66c6R$>gvyuf7ozR-hx4gfs&;RI
zGUZ%})He<v7^RQvc|DW})lx(HL$!;Dk)HGO-U$1-%X%dG3zs!)rHwt3?r*=f<SBp4
z-Ryc2O^%S%c1O(S=9SdX{h29_%PWY7uIsJ9#;4)}g;oy*hw2#7E;<%v^l!dn*w3YY
zi~oEWqZPZOu)FoCL-fMpw9o}ny%|wnysopHV)HH%gz|TF%>2TCK-C)wNF6{`u@v?h
z<kRH&_kPI%62~Vb<TjKNHuv^8D{A6z2!&Gu%t5i{<PM-wlNbNI##)@e%!Nd}jmUBj
zpbuB1v0NM_qlzj2c_0<r+|ri(<`!W|6pnC1$i<Bjs_==X%DW3#Q`$ISum<D<<s<R`
z`uUH&YDUxA$#IbVj}{$q$qky9FA`#EHR5mEec$P8iv>B0hwkTQh551nt?O!ZTnaHS
z#y!(JdnizsO4fG+xw0|ks}{4=q<-)EcVe}=?@Z;*Z|=`4X!368Vj(+TZzeWvIg}*T
zwB~+!@)xh-bJ${U&&rLxEf1ooosi;|Ddogp-$6VBJ8iJEIO=Bnv_9~~-U$?wb6GiA
zqd3dFZ8vT_omJdDT(tfEL0l0fz^lQ5AG58V?kYjydhuA^e%$ik5Lp8GiE~msduoNu
zI0^41i;8@}cbe6la>1B94gA)vT8B+-qwq!ZN2@_{<2e0=O?^%n4RRQ+8$-0NTFz|u
z@D#h`mQPsW2e`vYeo2!&xjo#4mRQ!;sHWiB0V8?f!W8~c$T~tIy|xL?X83q1XDU@j
zBS!(>?mRt(59Io64S_h|$6TUC(SM_r5K_1oIKJN(JU_&Rc}a5A%;dl8*lQ+?gf#DC
z1^=5uO=R4*SgC<;txf}lrD+j+nGs9%Jh5-~^97H<^-FiCH|q_&+p`=}?o+n=c=jtn
zb64vmr7wN$-1s5U@WbO5bCF@H!fwK$?-=?g?!;iD1^Z;(%&65)AN_w^zJEU!s=LGf
za2cEQbvZW21N(fPHQn{I&=NBKMZ`<SY^uA2FEb`lKM9ft=v35`uU@&sYyarhM9}--
zM_LE3?r~<JD{6W+2h|O5s5y^?Z*~Z7VrEun$&5{~CpA$$V7hzc6cM%nt%hEh%y?S{
zSQiJj|2f_<bMEJOqS2C9+_JJVC0RUDDYjm`rEg`HCXh4w&MAm=KNi>=$4@9GM<L7e
z?}n!02V`O6p6pwwSQ|C`aWMKh8H!WUTle&mfNksbt?SXUQ%v~$l~<<C`2GEYoC3BI
zOQNN@*0<O;8nC1VPKf5QW1I1yG!Bh0-ODHdGwM1lD4{3B&CXunTmh3l*x^1m64)v%
zPb%|!s*w4&%0=Hs+-tr<arCzC!t9hr^YelqYp;?MR)p<tT57ULo)m^b>%d3wmfxA|
zo5LGH1QUg1L}Fn*E}JX2rB}L@BtxpJE6S9=zQeUU;6Lj+XZgS0-w`kcAP&lQnk)qs
zQM$=MUtgiD>`E+;9(^h(BSUnvjf_HdQCAc~8kyv=5JsJbL5QOT{+K6wkA)c1Z~F2U
zSQVx~NVglsdR3eT(+aa2|8F<8X`oWGt&D!SwO!o|OLJ1d7HRqq5qZ_(Rc>l62`Q=C
zghu>B5W!28s|47k@)ro=<XhqxPB&YcFteIX<niBOzF|f}F#$k9Iu^x<`HrrKfr)_8
z<c{poc|%;EDgkQr7Pb9<U|Y#C<`2XRUvWJ@wS$>zVqW}v&nV|s)g6}{9Yyyp86|Rj
z^b7I~YruL#-6>!i$Ne<W(^J{O<S&p4ex9j0;Z7xxoYrJN7uMIO6G^=3hEi030m*Sh
zDOq-UanU<0j1VT>z9zvZa&8-J<QPYFdLFxU;D)4l`FZj<^3498GmQT2k^4!C50QO$
zYZ~;mb~psPmq^hn-A}SYBL2B4A@3ty3%`#wz4{z;7!%#y(y!{_8u|;D2>qJ|E;qav
z{|PL+a})yo`gWIzB;YanB4Q@fbDKi1MJwLNKQrY)=j&x+*j#_9G=shVHo5(OM*NFA
zqeQ>ddg+#rj}H|MjS62OxaE#c(N{lR9>buu;J3q_w4tkQcxZ5HGc)MYdgzScUTc8q
zG}{zm<7$VelD}(vjK@x^^v81&ZSej%EqLN)jNgzt)jJRX8F|5r8KuXKpPyw5dQW>j
zTU@@ZzTEw`4*g_<Kj~|?sTLiR)5|V%i=AAn-rn=FV-)J{z>LxPSDI<TuL5gUS{%0u
zO=6#>fn|jJ-(%~!D=tFWlY(omy&-aIHnBxaN@QixKP*V_p89cL)}(oqI#Vg8B;p-e
zi|uel>GkWEDk{Xh(Kv1ZZ=3XPz`XYiWBp9yI<j(epr-8lgwIFr(FdLr--yI>4{jMt
zGz^ah+}w~zJ=gY3O(aJsUbc72{aEwyKw{jHqD#1=-|WQcKb9oiO%8`<`Bk#vinCyt
zGJ%Ji5ECIzb2e1$JLflXmrbA?>1B&lLOup$kxk=DPRG+u*7i3XU)H1duz6keT=)+3
zLDCg67JsGTLCGH_zQJWO(hY@g%vmV-p0h!%5vXK*G|Z-#Ct*F^AzbX+h?s7DYmgJ|
zF2(Xxl2g#>zST6v=UlF9VCGFMt~i7g`y^qxH;R%5QXMuMT95-9OgHv2*1g-tq8N#3
z+hn^muVb(GNj4l_gdX|3->bAwsH>9y#_K<_vQk-NAF(&F>CFpMHNBV}Kf-YgjBMCc
zvSXXCvf|&If3xm7f91oG0L}vvz)*WHP+(54ZcUQ#+7x(tdA-xqYwpibB5BMc*>*O4
zzW30*BsJ9=0d9gkU;-j;Cb)f@)#;Bh8j64k+8oQ2)b~8-xDP_v?2ESF%<@gfpD#T)
z`Ep%d{oW7;ze)TMddqWjIud3Aw5iefN4QMid|zN4K2sX2gan0uYdj+pc}JfzKqfuH
z?0gf~&i;wprP84%#pKZat`pZGSrl+Z+ppx*P(j%LT8p=;hRhoQGXeHbL+oDWroFz^
zcm%T){Hx_Y*EKGTUt2xA@t8cSaXMD_R>cO%MASp?L4kva_0x5exYdr7zNOZjD6{L$
zrayPe@cY~N%3v%jjL_~14(U&*Yj;37jEy*8Fo2BT-u&<nv~!Yc89Ykw)i+IuMxG(j
zL#JDgl3C+YE$<n5KG}@(<UG+llBo9T?s1VZT3$=IaxW7gv4K4l{Lf3y^){*-A{meO
zgJKQ7SxdzgW?t0UP`LfR_14(c8I97-jH@%V&s=WP;zV8Tl6oGMDvob)A^)lFuv5-O
zChM+330EXUF_)=OLd-F2e}#wtdW(xDGDvIijY}`bIa~Ez;X$INwN}66(*D8Q=7@H3
zUtfbnjUa6KXj^O+=Xf+UG&jFTq&O}otu7-4CIN<=)z3~vPVR5GS+VqPUBguBI<A)Q
z_}Qs)c>1#Hc+A9cAwwa3G`A5Wo!C64?b3wJ%D~b>Eb^7A@y8E+;NjWc-260bj4-oJ
znLwW1Gxyw&U<ttL);6MO(g|+4TIxHk(dFBosy5~w%+W(G&(fNMb+(G>=n_5>+nRhQ
z4yYejr;)z*xu?T=pIeG<R{k5#NtG~tvm`ykBjM_MXPD_qaslN<pQJv;R<xnR#jELd
z^lBD5PjDQ1a!J?REa)Oxk4e)(CU3Ti2ljoRocxtyCYT;Z+nMu2Ox+CXuKk9lb`uno
zmWP&w(32okcZCS%_q4@2J(TbLn9YXs8~f+bY8@i7VhqQsLU|{9UOj%Hk@3n4TiWa1
z_%%Bjri1>g(kn)ZHpA{p=^Jv~>7$n}{aSAA^r(P~NmV4p@v4qI_<&F?^^OXDv3v5m
zta;JCfdTWnQG!IqdGCA?mb&`PBOPt+pN2&bSXnotoa8%V1vM2j6!s^PbLGL<3_X*!
zuS_~cDxqJ#B!CjHA@Cy@j+N$RDrNxrSWd#MF;}F((IRD$K_3uGxq=jz-vn~*5*_oR
z?|)c=x!=DLePE@^SIGh;X~Vp&SCPl<4zn{e9FWmG7ZGXg>XL$f-2im@WR_1EU*S#w
z$A#hVpPvnT5<DT{`2|F&*c1Y4K==N@*4WfE8ydMOUgVB1p|Tzv8ygD*ST>U``vyR6
zk=*&_HnV1(lWzS4bbJ!pzRPl%Z0(+;ATWfDt`v5uwszZU$6EVTfSTe_n*R@uL-a%X
zRLRt3JKryj&22V#k4Sl|Pjdq|_%nYUXgfG6-<D1JPHb&>zvkQf?8qyyqG;0eIBVkx
z?oM0`Ng8~d`iP5X(<ZCxul3~~l|OT*D%;1mi`IHa8x_8G!d*c=`U?t)g>R8YOt8lg
z`>`t&&T(U7YETqg&0Oi~FW_Qfk%AKOw}(2uA8{#M2TUM^`jWJ}YYkns=X`eSjO8Yf
zVXBmvMhl#uSYZ>fK7sG1cIB-pEW=ueeq{9>;u=McF!@dTGn8dR#kCvCac%tHsfO2V
z1Bf}rpkpbxa)g{7?;An&@*9su0?<)qL!*vx{Lc?OsoQ*K-SNUl(3#(bktBUcYaRa#
z>%EHVx~fLLNzP*)x0oNMba*j5DvFd)=~Aut7aVH+7LP|GXL^~&cZM5H_jmJCTjoq|
z?=0G2nkm=R?`Ws!E^l-#7&f%eiSfx)o@a~G^luCyufrbOCx6>F(afY5L@i^S;p>tU
zZc$n7l&6wab{Uh?juGPI&u7(NLNzEAm}=sI4=IFb^ZWPj%h32&c=E&zq~6;epKVSz
za9S@)2axe8f{KrT5!*b}#8ZF(B^_@68>|PwNtwMG;^7Eb0rMP$mWI;wu?=#5y1xPp
zDFfkb@8mNSnZP9Vp}$+c5aYB9ZL=l#FGX{6^N41r<Fyh_VD(bYQk3|{Aqo#r9zJ_}
zd)s=d_JTwVf7$C7??nQL4S?6C{A9aLAyEW7qC+=1IoSaE*yXc<d>oFu`q<?B2bqfb
zA~V}=+&nz5AtXcK2`RbN<CiBF8RiZfKM$j%q*PbOLAzuzM5e-P5%|@a|HU8Y3hk{=
z<(FRtxFkRE3yGm6Yn7cSzr7Y-p}{wwtrRxuWKfG$^3#}pvTeBYkc8uSm?bpy%e@Gy
zqTMi+%P%<@WdqfDV<@v;^-K>{eLLDj=d4UByIgY~cVdKrD0D{uHDCFmElMY9QBF!e
zKF+`rMB;GJTfVpDG=EOJWqEGace^#oX@7nD&z}KkvTy*`-r-^NxS~mC)T1v+&E!t2
zl}s|PP_h-Ml^9A{MeSjiBWq<wf-JQ%*{4s>D>tfE^V<rY4=s!IWws<1Oe@y7B%*lo
zki|&=e3k=A!X7y1WuUC+gYJh-2j^gjK_(ERCg9fqoq{5uF#JplUNahw6sExet-4wX
z>}ZukHt*lRCk9(Y#l^6V7C~6NA21`WPW$QTVf#~S5KHyU<rk{)ST&jfw-{<+-E$Y_
zjOO3P>i2jR=<J6e)=G~J9QBrp2zp(xlkVN@?LO-D>%h!LlHzYKrWH*x^>W+QD9wmX
z_MSXml9F(-7kXOi^qgsFC9v1aB<b8Q#5A39Tv^KTq1aJ&@UEp*+WR1!%#k!vcTJ`G
zn9^3qt0<dB!w>mF#3G8nTy<<V@&1iofJ3~!qvP`D7v!Ijq+7>*)jOShUy_{G%Tk19
zp=bO32SGX(A;B&Qxq$P5^0a9@xV6~O{m?-Gvnb8%f$4=%tqm8Xw7=_!)yzE{=NweC
zQY>BZhu)mngF`lM%&;q!;E<WioplrEd!4siXrG6fRqZLF+}z#y!07KLNM^91lPE<I
z{Z+)?rGuKnc<+<x)i9dk&~FDz-6P{iTRx(eLu3oldMi0b92;48zJz)0PU`l`$zWK0
z{Xvzc>dg_K^{e8$Y!K~|*Bc53hDTL1cDGSCI%QeyigBFsTQ`OGv7)JB9g%=5p{vx_
z`i?%#hA|-v3yYE5H}1rohDpG$D6hZzT>^>xYJUp%M=ctr>FSWSpaD!R_}}Edz&4Vb
z*#lYkFG=m~?GJ<SNk|w(%Yo#rO}!b~@w>l+rP5uF|7`E?hxHqV_8T&(&{qLW!7@l$
zp_G4$j6?rZWmOfmDCLVld8p;oJ)>qM(}a0r8Qh>xpKk1}jSj-$$?7)lhj1C~4Pqss
zNiD4@%2H0|M-$CcR}IVhZVq(Ta2by8`*k_kI|}*g2A$>KEp@3*`^8>pykqb1vWx38
zGnP#6H2UeEZ95%4T-42*KmB!`Whj<<?gw*9Q%hHy%u$;A_V%Ap%_qah_2pu4TH@cl
z84ojv^a5wG_&)%Y6H0|V2Fee7@=?R0ul6o)-zFp>7YDz1cq%yS6b`Fx&t!d3#EX@B
zQ{u+XEzRHuOQZ}Z-~jBB3;;LKQ&J}CQ<l_sOm>c;cj|**J~}*Hf*!&rU~;EU6Ah*T
zj+mUzkweo)UwmS=w?CAb?zkRIds&*bUambLwe>n9=0*ypC=!WZ-JEkKySf_^!*H-N
zH?vis)5hph(9}4K$e4c3(%RcdPClP{@L9CCPa;Xnueeh=@trEp*9~XXO+QIC=u0l4
zTKf52Gn#neptG^HKquYAIu;Y-Q|H47amVGf6e>}GTzsN>L>icYUxlw+<~Tn@O@G}s
zIQZ3&GXl!tg>R~E82(DVm*%{;I#TosSOz|ql$5xhY`th~Zf4--R<f;7$MKrYJgxI@
zox6}D-z}V&yH^iE5jgzrf{PPCkhn4nbJR!sda*9wp?a+(LXVbS+K}gjU$c|?t6geY
zrU}33d|*s-`A)jOoWTeickvHl%L3z(gIUD)Lw{+>V8#%CweHfN@-Im?f>BOItf)Ji
zsy7&C6FGYy&s#nxMcr(YvB{&lP{krzUbuf(DOAzalSzEi_1HwT7YfWa(AnQx9hT!X
z8n8Zeh9XBO6_aLl<*#>O+6&tb3<>bCza)XqN?H2?VBB7Q*#$1!m5(H*h&0Tl{$5&A
zzHLvKt;LlucFmsJ0H8{s4g6`na;H-^=nlVqE}*zFu8`Zu<+q|qH?AES3D<d67hx*4
zD9TT&V9YWN$<K5gpHF00uBxRO<=+0Syyf87|4BS$0SKZNSWx#8x9y;kNo?KKXUVgw
zve)p=a;WXx;#inO7g<^SDtMcqN19D(Lgt#nSBYoOJb3NaGf@E@c+4Z<{-G9?5;)XO
zZXW~^CedCnuhiY$8K=(G(Gn>jpv?*?Yg_SU$o@oi3e-a=XkkPreWetEU-FMph`rPY
zZ!d_15^Rde!npe$P(RVp(!My}-x$o-X)CnfdW)<L<q1bbM4(P+Vs2Bl2x_#{nLsl^
zHS_qY;EolkSj71_-m0swv&G>9lWbB>FP?0O#e<HrJk(vGF<W&oQOc)0Wp6Zz%;D^d
zeK431l|&g?s4ZJ}ztM>TQ6k828ucb9HX!0Qje@?`-KALbzJBPkrU2rkRkwhi2B8sw
zvn9A;`qQ|nlCg0nba-XxcNCwQo9BWn!WWgbwl0JeSGiDB)z!5cT7{_qFx<-|z=0|3
zQ2EVp)1lm_V4h#B0bcj`c(tIB3v|S3zvhKT_ZunzHlZJ=^sv~-&@PDzD(BuzGz=MA
zt!h7M{}QCUv_((d6^xi49etDRq!ca6?0qkuOr7*IXWnyC0UvdD(otb0IWbO1aZL5X
zEtBZ|>U>$L)X``>oxN)hPN}?vue{YW82%mK+95hP72mboE8xze){|U1|ILU>6z~CQ
zR7=?v>tBL`q>?-KVdYbSP*SDFwF(FzJIk{GO+@@q+Z$;LqF})d5>sbTq3IGluf!qa
zO#;VZ06lDKa(1AZPH01sLIH2<S4QOofKd|g*+R>NSispnx!>tt>28K4tQZrRbb(&i
zK*DoBYoLWy!oiIreqgx0Zd`H6i>gAHCg{sC|Lu-<cBaIlvemPzY!^_sryw-`a~^Tr
zJ7<y($aNekS?22?B1;aasm>0W<?G^Q=H`>QGSKmBWA;dY`urmuH?2DN`1yx*=ifHm
z3a(rTa$Zp1_JZ6YNp=D2>eX&I6gW)BcsR`_)N~7!VF&h;m{rC@I8Ar|zEH1VWK>zO
zmzNg`CYYsOp}0Tclmli!TYwB_0;kjB`zz$#&}o&2mQcd11WhoCK3O81^(lDUukd5b
z16disa%KqNN=Immj@g^3HBd9ya~luEshU+Opu%xmRmd{-fQgil$?LKk`{5vFuMgRa
zBy`oM$L_vjZNZkS+c|h@%lu!GJfeKNWa#3xygJ-+t&kIZBGh6`5d^4O_m-LsE@$Az
z^v+DyD69st(EVv9)ZV%lBBfQjo!=J{>L`Tsq7E~T7yFyRVycC2U|DBJXsih|jCBNT
zA!AIQD25pAVZij<TrzEpy^s|D*0a1(Om1Z(5Xp*I5ST<!P6Q{0G-@oA3wok<c}p-t
z8@taVy1l#mi!z8`)=*kQg-m367jpv4BMr_|>W>w&PbueLF_%39#+Tp%xeKWumkUi^
zo4;{s$--}4-95A!;be4(sb3jP;pOfUjkM-@_0-1bC-;xZ|F&r8eV)Q%a;G^X7gn1B
z(@MKSVlETf;Kxd7w6^pNd!@(+p@v_H=S+?>0ULWgyIBg%V_9Oh8xvWcScI*HMFd8Z
zZs~Pt_?61a%ETP{(mgdUE-rhEi9x8S9&(j1*wk@_g&!omP0Qv8fD~2i><s-r+F4Jf
z*%hYfgA5Lqr^L7Jh4#LkuAC>}kT#o0aq;2iUL8?W6f_$ipeFP{3h~z(`Es;kbX`VX
z73!lNh1-Lyr?Ug)RtF@g-<CY2t*zq0uoK(ovGSQh-%wFe=>y##F%)*;itYj)P)ASC
zs3+lh`RSigYEf9U&|za+t7JMKIR<;EEM@uoUl+yTTW@S^T*@k$=mpSoM&7&TWyr=L
znnlv6T67CSjSL>S`3XlCe@-7{6XOk6xZV48^rO6iME-~60qGrAGBN%i#(VN%qs(F3
zq`hJItGF>y3=)39Ee%PFd`?Qn*Z*F(YPdMf;St8I^{=(2B`eSvRFIbsVI=SqrwN8+
z*}7jr32>~U;J=C>U}v)n^O5DjM>ZjZFIe0r)O3!Iiz_&mxr~k;)sF>d@SRTaD{;IJ
zBYtfxkzg@P@!VZ&W1KvsHR+ASaB(@}cXe+c=tZRM&TBG$<Na2;K^4X=-f8tRXYK35
z?f$L#nDk+uO6P@3s0)*R-k+_m3M9wByl+zeDm0LDu(wZzTnU!%6LNMii`xM-=lvB5
z;c^!?AUXGCwgaO7AMYTq^JC6YuNa6}LX(%nF6U6lY0eI_T<v8aP43WH4R*RKchsdw
z&!)6?Me}>{pr&PKgUg2#jsy2AI@_$G8*9O?kM4yg2RW+5?AGHL$fX;Wq^7{!s9Nz6
z>Y~tCy=~V<mXO3CfnErlLt<ld0BUUrr66KfEjRH(y3F1vO#2|xN3zz#gM%eM-T{L4
z_p}VFgtK!cd*6lBCPx8Flc%HZA!<u$YrS5_$#{>bDWbkGNC;2`4BaQ|lJR5UT^>=|
zcpxV=mP($*x!K`_r4iP&5f(Gbye7`_Xt5*jRZZh71lRP-TRHg*Q4iBoZ4X8Pjm~l9
zzn6zbXpz3;-LNHf-!hwO(wM8ww1kfx-_5Y|_nTe?mhtS7junFm)xOZ3m1|3bc(Qnc
zX`gd$uhJ*az4FRWT3h<kaIc_W?(qxtlo(u1x!eaO{xD$ved?%aw@)sX!C_p~K;2h<
z^pzokf7kOhRb<s&wTTdw8{1-rE+&Q9E_=)lhsW{*!vC&B+SEGqvUIX_Ij0EIW$WI_
z;Im%?nuBx>UGxT0P0Gy&{nzE&&Z2*v#%J7?r!5rP;aKc_*04~WK1A_VzabN!w|Gt~
z55Iwcf`xNwQY8$#Le^%KJAKpTBI?}{v{Xj)O0;y+$s5R?%d*Ixi)F4P8BhFg%9N0Z
zaGH&~f*8*|(iQu7=T!1ki3=L#dhGUOcGX=%2x2p5D+1~E?7uKhC}Iqz#q(xS&MoO!
z)II+DzG$hW%vknNIND0ks5t)`8L9exmTGC-Zqi0U6wSX=TQ)#sI=zRk66sCByo&Wz
z$vbD;a)%0k7gx_q`tMy!_m?qFold#qs#%g=_!~pP%OMg=uMGYkUhofvZ?M7fZrxH-
z_?tk(D{)a^GC<Gxe_y#PS{`3e!GPxPPJsRQUD43Aq~Uh|{X%_-{J*~b|NmvoIx_@6
z4{*F5-s%gD`NZfoe(zl6>2}6Vx?r!C@z8%)c<KHWh8x!1H&RgplFCWZh>84i-j4#)
zogDns?^UB2Z6;w13IB(-w~neR>e@#!5CsH90cn+zmTpjyPU%LvyAQDu36T<zZfOo7
zai~Lg_n{jO-QC;`zTf-Z|9)fq#=Q=39F)D++H<Wn=kq*st+}Ov^my7^brJ<d<a0d+
zN@ROfSiYc6a_&>s=q_F7mzo4?<i4)BngXT@=ie0blgku0vKW*$%>OtZ#Y`e!Z0^+l
zxogZz_u20K50v8E5Q$lAqQX-xQA$8|ZJ9^@MqNG|se`sGfgvkCOM<waH&#RUrjN%T
zBZ7X~$j<+J{`&W022W9FKgy^_Wlo)Z{8%w5Ab0-fyMuY+!h8V*s#-cxV+IqiShVlE
zH5>2RniCJh1gV-`-#hPuBJ>7)PKn*F1D%bf+C};3ucj7~G})MBF}v)NOzhE+QSiAb
zdfYErJLa=5_5Rb{=ezg40!c!q0jOnNVWgRQ#!8%6Qf6_oMfhQItGovpiE5<~9M3AZ
zl;__Cdv_OH@vrrHV>lz|R%<lCqE&-eN0j1p0vAob|4i<+?&pP%qWO-ASH&&LisrHD
z#Yo=ihv`ola)DTF+e$`yaU1CWjEz7C=Tt@i-WK5hJ33M0Z&4f~AX+sdy|)46-8{v*
z$zALJ9u6UiD!_{=UOaBm+1>5nOn?hc<&ABeTugs-j-t-KuIQ#-KV^JY__VeIH<RE_
z9uTc$H<ja!@qDHi|K&ENxCD1p`$-<r7zwHlKzb3VF@NxZ)+KoE%*K}Pvhr7aJU|!G
zIiN#==p%?v*{?zoFJ769+?cDBOHACkx`{1B<%pLCg~v!O`B)?p+0n4v^8qGOukjH)
zW6{~$>vY?n=r$&@8>8}-E~`QwL7#c;<4gB(rRx<TnUS?oR4B&GX|ndcgoxM{uort{
zX}_ARRJ2jAcfYSwgXaXMDS;I9`IU#nSTmBEn7t?0H{B=*qmmATTpr+YcV8d%9BMB2
z{TcNAqa!5~lWcewm;m~$;~Xt#$h9P=Y!pSW4CU)_m(=XGsN_H*@v2SB)oTnm;=A}o
z#1hp}@ntIKdbI!Kn@Ew7)fzLEf>Nb;cyx4hJl!{5<jM*@KYn)-kW{RmXI1cc`*(CQ
zwMK~ajDSGiTur;7p)qd$0@9eRxR?cLTGYN^YU@ms{dvXhCOI}%trRS<F<U2bg|&u6
zQetECINV1N%+90CePSo_$JZQ-qT;>~-M3sr)fe&Q<$iXzy0Qx%yWFeyz27ACawhki
z1E1(br1svo5F{YRJB0h{c`oSAD+@l{{i<9(xR$&E9V?358hi^w72SUoV&ey~pR(kj
zAMN~eZb}%p(?@khL;s~_xnC)j+Bx#`8JeoQ#Q8?LyZ6=b*7bD()4zrli9mldr)LDs
zvI0x0@^zCA`aBq{_9;%Te*1;6Tm@i>W!uenL*mXUm^%Ku`6Bu4dg@P+RM0wU<wteP
zt1BsgUCM7tcjfb2({$mUX*VNHTfudjwS0-d7Ly$8uG96|eA8MeD2lhinCj;5>mI_R
z*y3d>F$@`rL~rzKC{oF$pC-PPSxG&f9yw;XS7bIfd6I*O1QMq;_)zBl952*92FEo6
zG<=tCxn#JzhsU($(<ug4I!U}hVLaIxZ54CrDX@j#A;}+5<O2RcN=MRwngV3<Wh$uB
zg<Conm-1AO*(S~4IaVLowP4mMd$K?Hmo`E{ss9C9KP4O#^>b#2ICD?3zw!KqWQACA
zX9K0DFS$3JP<>FLPF?3_Q~4QuH4NK~cl--1rU{x^^EGHYusLPL(2;FM)Vrrud|^@i
z=<X8=$|GUWZ~pu9fAR6o|NUQFciZ6q7ncG5+j->L_I+@soZO9s7tYesl2|whdZsj=
zOwkAm)~!~~neP2Y=sDU8+hkRPi`>p`X>o%*LTa1wHvau;M%Vsl1(f&n_y7Cx-T&bu
z@c#ew8~^|Q!oT?Qf8B<xGd{|Ri>;0iTr!)5*zO%2tE4g$lWSZLXcQ}hE*K3Hr)>BV
z=p|VD{(%I!mLhNb|JVuhrx$*?mfAh?`#|T=G-`#son0|_tSB4IRNlVD1=KC&<>Y=%
zlvzUWJ$RH27KhtreEy$HH8Uq3R)d@CZ-w(ft>j<7QoxOMCLjk~pMXKZ;z$f7<C}iy
z|8M|GJuFwc-W){oHEzyF;CVo>#5Ot>)-A6EQ*<K*T2NbHvDNfw9dQn3I~>|boGaj(
z|7piY9k$j$^z2y*crsE1lsuHc%n|5d4OwErhswAW-oQGSzbHU4@-|FZ72y5nx&#%=
z`!fe?{i)w6I_7QOF);M>ApY}TUI&bEUZYm!_*%#N`*!-2xBq-Q3Uh>rOthUdFwrCw
zu2`tee|CDBaG)CBlX-Y}h+}QfMJ#+H$jZh+=uf=OnA-Vw^5zUIAEY@&Z0vqUvT(7a
zr*a7CJDB?e9^ajv2jg22ox9fGNs|2vlPK{ycNZ-$ihvUw#LU^u4R5at>lehnH>>#r
zIex8|_X>NAfFUqzKyj-U^X=@>Dd+WrrC(ip^;uY0e#gYDhM!M0(6M$+Qz~*-a>UzN
zZWr8F@@1fIo2E?p*}iaulZ9MC($Im?%rFJLy#tJjPqIIqa^Dm0?CDVglalbR50srW
zZ5KC#ctMqt;EQ(Lp5H@g!gSoemQNnF5?O=i5CQQ~E$2k_Qj_`Z@c)>po-+3cT^meT
z*)F!yaBwJtD$FQYa9S?&0NelsisM=G)}g_{Ak~a&(X&;`!B0R-rsE5owu<g(5~6a(
z*F$kVJ5#Up$cnNU|5Fi>^*b&y<L}&8VPPNh4=XZe#Lu_4ap=0YZC)OkW(TbRLh43Q
z%Z0NkWBhK8d^F;qqBWoT#%we@?A&=f`xyY>_dPZqlNH40ld5LX-25q^XUekYUeV4?
z0s2nQ&_=a7)?Ig$9ZE*5+X$JQtaHfwecYlzzVIn2DJjbyFQ#0+(xlhuyv-`eeuqHm
zZy08-!Vhk@H1zzqF;<iY2!Q2e<=cWoLZU)LL%|2sRNF_51Peb2f{x@hiumV`id$QE
zs0SqD-Qg{IVOisw+jhBm7V*+I7f8+;e#a^td3g!G<(xh$sCnQr<Z1?vfD2t<Xr><g
z!v^k`6J|y2n}1fC1|qT!Cw}i7A*WnBd+5&6z8GNx*wce_BKq64L;Y`D9mDObq|u1%
zwWq)r`~)UpBqbTCF(`AYJaq2t>WgNe7S*27W_ga(94;;{352La4&07}nZP)el$2EL
zqC8nX`pZC|@8YJrTguSLNa8k<L5FvKDMMirsJ$qfh3eN2X)&GgnY8o3UDqaHdF$$!
zdw|TC<5KpcP*){hEV3|h`JMLjD%#xCIG=}f{bmCJ-Sqi_q=m^&TZnZMUdI;JzKxlC
zDyxgnL)*LR>87=GM~abECy6JKVTt)DP$kZ?pSL&2C42QcBE8}elR?1YbKT}l)(#&1
zU=U>&o9>d$H18a`b2Qf7RKrY}Kn1Fe{ovry=wa~m=1#GaOMQ#URZ>mW!rIF3gs@cY
zup_nt*`SHx+%KCIMDd&hbj;Ii9&J|qHD7aTPwkVANqIA8wE^LokwE*F$>wo;HoL}~
z2#Vm<!H;QhVY=b0Q`%%+^>R5go%k6YTxhvU0pk0PM>aXN?0TK^lvAqYqkmrbkwShH
zjE5Mj26SrbU}bOcT|R}NE)iyfZ_^l%ZtxVcrbKCMVDdQ{>-=b$NCMLt!!<^#D#MU9
zzv2Q9h5^l?j4xZA#|9Gtj~81EwkK4B<4yyryN#qt9AGnRLnX~N<#H*A+-Qh)E;lE(
zq8IqFF6mEBV3a0M|MMTrf36hoRsMcTtJomm*6%Xww?5mxF9X?+5a+SmPq#81m$E)t
zqp>HrP#t}`J1Q$wpiohlTpL^Wp{`I(qa#w8A(opPGFO9b;LLScM|WZJbts!9d6Kud
zwzkU3ODdf7Y*n8!Geb!`_0IJ5ySdVVeqfbzz^o(O8)2nUW-$Ps!Ia=X233h$!JJ9a
z)a#AqIPhVXNi&^LK!8v%#ksOPJZEvI5c^!KKm$0J?vuTiP!V>nG;y-Jnq{h^-B$cl
zU_GH)UT*be`=BtyEIV~eeIrm><EXckfY9xxCQ|Os<Dz6^<@TV@u!0+EKqw4EpU<|O
z+S;a6*It>YQmpM2kl?B|kXAj27tYaAmA0G#KExkyIUBkvO^hB~hf`dYw^8DfE}m^`
z8HOKjUHTRu#Qg(C5??HpcSoQqvY;9avV#w>y~$km@breBctsb?i1d#=@g@qF;yr3G
z^{q5Qcy?9p##i_1GE#smJ2SPxOap4<+m?erS6JZK2&3Ju;Mktd3X$%dS^gKuH2HFo
zAjs$~;w7GNJk0*&QH|i@YR@TR%z42mKH)*pi<_b33J)cYq!1SM;pJDJ*Dv2VjQQtJ
zZ$#=bQex6e9gxTsx~aa=wIZ`*(nS~f@F9JIh5#VlK-_LXe}!6{Cn{$3Q1$Mb6bMM5
z<$?;MqbX>}Is4X%SS%{TDFm&Vvf5bg>+CePF@kqtv9k=P0LaL)+WvuO#EW}OUV84;
zdFXp5qeX|9hDdJw{cTAw(}&3U(sbKFN7>mvj}97{P*~k+5r!j#9&8Mp|8o2w!+iIp
zugub*t8a8cNHl?PrOyM6LY#SPq9P?Xx6ngT|LnrWd~Kd9OL=dZmpBf7^Q)VbbbTZ@
z1B32AtnvxiIL_1RO?mObo4{$!dFGHT<JBxJH>6=vHn1X7K$|adP^{{p<WO3Cuzz^8
zyoF0@yc{yk14Kq<Bw0?rYo49;#l3qFuXP_U{0{+EYq)8TbYX{WwGY7dzWTPh=yuF@
zc6JRljcTQ(3=#QhVJ6iY{-9V3#)UevutKL+vL-AbrfU-7gC*rNOM?@4mi%z^3AFy_
z8D;2<`<#IIAb;Iu&U8QrlII2}ia*?n+5&QMU#1&d*Hp4d<Q4O91Po5A94dV>in2Oe
z<1}1BMxRY-!nIC%Vxd<ZG*wmRrRR?8ssWMzES?6^6}jAark$Ly?*%IIP9coRELs@Y
zqyjtKu1;`m_5E_v<T_P6*%G7b(`vIvsM=m;(5SsofxP1{!LUoQgo0DgOdupPBl!em
zsbJ}b^S-ze<GQ%vfc|LNo0a>D=y#706y$|p&s}phcSY!V1(WFR98ppL2J;24G7bGh
zY5S}MH!7AAh^5D?gh;4m2Q=tLY3tiWS#=;ErzcD86p-7<m92ao9Y9S>gKacUpL%$?
z#QJcSlCG4FIcKcs0*FvJZ2ie@b~HY!?y&nqoik{7M_)(5<bxLx?LXWovT%#e6)lsD
zN(6|nfuB|bQfbs4skbpf7Jf|I-j(b3@gw?fG3^^x1-A(KpPy#I@ibXzpEny%sf~v^
z7w-Og#^7R{r+v9k<Z*ex1TjYP)RfpIuAg2&wNrpPsqlqCRtGM2>2QPN-TuGk1%n#c
zGg&BPWG8rcxos58Z{U20of<C>edJgtrPG^On5=S{pAlbfqVN7`V2l0!0|&=Zlj)hP
za&T~7M|3Q;nnAIwp-R@+aGfUW3YiGc3!#!5?TmahELz5^<fx}T>mvwQv=ib=1zQd{
zfE-P!GTd4EJt0UL43|NB&5!CpZ@=B)T2E+bBzoEvw|;#^JwL};PZBavjnO87l;3{a
z+1WeNGk<Y{3q4Ow-QO#&g7e+o=uT;)ZQh(u1)ts5-=_lpvT3hab@;E7rE~N@kCJ;2
zY^wr=7{Kx$*%^oGTx(~*;LaPWdf}T+@VB3Y5N~pe3TjHBjHctCNb|fIsn}T+$a>G!
zSfd$-1{6SrGLc-iMD3er?N6%_De^~Jg)>`|mGExbH3NxUPc4}Gfj;Ld=-|SK25|{r
zrTD?jp+K}uRjxU{@96~`E_K^YuNNRS1yaHpW{ai!q~h`-C!JryEr)8a&h58&YS<mC
zCPdjs#xDGF!7Ht@i`R>g$hFiH!9iyXJer%sN1ndm<KG7UaiDN`vrZ+~GhhAuvg6D8
zP-;zlZoYc>;Ly?X$<X0mU)?)u&hoLXt}+WmIl%BMxxAdTZtoGl?NBgS1@&%{-SNrD
zW)P*pc0BvLbY#OQVA;=eCp9=#Ks;gg>=sX?tr4i(t^TN;EmI&m-X22q)n$MAZrvY`
zJw+#5GG60cAB>gsW0aCy<kkcfIY64&^w^((Kje+L+bhtZDa|ce8AS30_G)|&uLIYc
zj}MYOKHgN?0>5S;y8>V{+C(SYLn+LK>vHbi)%^VIrCM}alj%k32u!8F<<U4L#W2eG
z-$pw#S!<KdYcLu?lk)|`^F8EEu3AmiHY3@sLFHsX!9z}*H2ySzsyB&TCopSf!>w}L
zY*Kz1fPH;@aG=yS0b^sSN;BFV%YSZ<zy4iK5Li>y`%~+2oD*obSfw8jV0Kl1Lo;cl
zP}2X*!^u%lOP=~NPaljoN`mh{Fh5;4b0{x>xFOYRYV7i75}75%*VfJt_rk1lXKM7E
zcfRaUCo73Km06kKWt^P1WcSvozxRQ-LB+-X-u@H5o8}*|S+kjIq4n#r7wYE?TreP8
zr%xJjr})t}Hcg|Vd~iG*w+2iWHo0@>7t@5Dr_#YNj*H>kXAth75SXd%G}@C3#z_#n
ztoD#)>Qs0`AH$~$3YtW3dwl|NEL|#y|LF_AS+<6)M#fm3Tg@r4#7mtogjU8~UL@mZ
zFXWO7aI1Z9tIiDCwBYvuOz}N{CPYi)G+_SfWp=WkW}z<0BrOfU)^}})Fbx3#ezxA-
zWTm^4rV^#Z5ktwIj3piEIh>jqPP*(}?p6`MzV=k_sh_z())CfqL*HId(-sZpu4EKu
zt%AlSe1vLto?`xT6REUpj|Lv7Ek8f}Ei|ZqbD`*ZvkpvZOb4GqeugCQ<-}`n40SP2
zD~`ZsjUvQ%PT-mk2u00K2DoZiEk!a$p4|4sh5=l+Hbo~!-SX<nQQPLT>ip}`by=nt
zf7K!MRcRKkt`Lt->_|CXgaOF0uDi)7#TU<8-wn6Hr}X&-a}s%c{b}f^ML`cbzm5aV
zdjL?&tRBqXD5gX!?VJSC*jH$@Td=GxUe_41#~;n-B?m&t?gl4U;I`d2AS(kcFBYCu
zTPo?>Mu32&<r$S2F=(VtRr9Gts$^VML4u1dIba29LZfV^imA+eQkj3+JSG+>3+M|a
z;$arP`&&-@{k6enr>BZ0LFeCqdsQ`#_J-pYaV2K3z<<$P5%0FGR1*KtTNlDJQjl@}
z?H(>E^;nlA^U($^CAkpPdDBkUS7~%<c10o_1br1#baH(T!fB88_u1&S6&HlTV&8r2
z8x**FPQ9<j*d)Ra+u~p{qjy-L;{BZ-iABt`8av4d<V6kX<AS?PfQ1SW5Pj6Gn=eDI
zs%hc~1`11ie@EJIl`?I|&`Ch|(UYBf1)PNYDi`rpc8qVw;yue<KO=HbYb4Fjqw0Xo
zI>aHvX|~6w{gwf2cCeq}iY49*k9rkP$F>9rR}NK0Y*^TpW|EA~)aYe`6;3M>{{U!r
z8J1UmS#S)LIi9_aj-S&tS#xP?XJ<bjKUInsd5U;fT_(9AE+#YZ8x;e#(f*t=BLk|P
zum00&M(VqpQ*@U9-nG?j60Za39om78uG%z;pQ-1Y$rpx*JS9yp3nq**&jX39xV&9i
ze)D8#ma)lIz(bP0_FvQUauq;w@`Nvfz9>=1&Pk>r=H4|jw;}sd77gCP%`FqEbUMav
z(;#6+{`mM{s38awL?m64Lz?rp=w#({flJ(X&J%I(9@AGFEPo*|HjJW`cn;>0@Pc;x
zNisxR{LN3xKJxZEFA4Xi?8a={o&Tmqg(oZ~eW66m2piiGu)xVEOm>y;Sx4mW^5J)a
zm1-Qp1I~!r3jLPtzJ1HdBg9nQk}1bVZ>`5+M>hDP!el`%s19_Fe4AH>5>Lz6nTySF
zl2JNnI-9*s;(FB${x^<rBsG8AbW|d0=*?NLg1B+8+B70xO<@v?bTC9AK;|eMhFmeX
zwkxK{alHZ{1+;|{P%>pnw8cdjEAyAF#K$m<<Z}Ej`#X}D`{2HIeLYrd+eX>w>>x-&
zDaGfnJ=b5T={&vvU};TPUK8RMMJkYMiH2=D+$?)I5GrWixtfJLU9DER%pDn*YejyO
z=Fx@@X*gG-U!VM2d&_L=yK+y1q*_QrN7sRhMRmA&xw_ykP_@%}eQq%PIC|Dtr4l5y
zM_ZS5KfR|xALA&VXybcvLBq==ix#eMG}_Q~3&W!ASJvR*l-QjvcMuO;ULh`*PV>8k
zlXan_3VO<AI3*>HE1QC)^Pjyba;&`SAfh8G-#-Ac27_Ux4h-Dxit!7w>%c%I#?ipU
z)AQqVIs`r2XMlu{0>m!U>ae^iw6vcUFqNR?{6K`Y8}{m&BwW!Y!BBMJw^f%?Ps)d^
z>l9rI51z8oVg`+)$(aW-sMU4b<4oe~lSSX55_euhu{dxskxA}p{xnQuwC24C2xiRi
zl@$uLJY&z<EnZ_s<N#wuMQw{m3*mL_H}!9hAPWQ@M~5#6afZ^Q4mP>mRki#><&9)Q
z8ZX_CX1AkPx62riN{wbCX{KwCgB|vB_eEuo>x<-xYM?HvMo@-SDbG?UdOp;&DXAtr
z{S8ChdJAcEwwj4p(c(S{R?kGGC=<h(#znZou>NpF++p&UDj*A(m2cy4b1{(%?K)g5
ztlL;am1H;W^Q2r^%9i|WSwlD^UGsKhV^-&le>%dp?Wl19R~>Si4Uvv5m{f<!RWYc1
zuhOB?VPe9cxyaoJ@$@+CM_YvLuQKQJL8PKbg0uU%b_zoxDYS64_I{0-SZxky6}lfJ
z3aHhok;5Ib0Xe3b=1mI^m@Ax5ib`vCYfvH^=hnH%5#%bb-%pXkqD~mD$JbH*B=M7U
zwO^!byGrefNI2@%wuF4v?)xjdgAX+bxWC=OzQc*eVyWFl$U$#JQeu9Wh(Kz00<j~M
zDAn;~rT~q%l!kdK5?whfWM@{RZi-Frqqeqsjeay><$5KZwGD+nbwxT_j+Lk=m#FDz
z4%L~!EX2F1BCN~?6ruhznct+-mTb5V>)psq4z3@pBrL9J__YsNjQqCbjZBD03)A8;
z86XiFhWps4giD2kw6vhI{OO1!Ka-M?RT7o;1g$t^K13rC-t=f;Z9aAK;xL?oUjQg7
zJ^=Q+_*+Ba$G4!MIHreCaN+qVo~kd*>Qznyf7ONy#7|qJwrXmHmx90Cq!QGL9d;tG
znBrzy?rcN?2IWVKHe3!nygO^3<%|8w%ro|`+`82LX8ux%hYC9nPkKN)qoT7kbleow
zxGAk4DzBd7x4r|x>{gSRX>ZR1K&fjqON#y2QdOtKx=I_b#NSi1?2};ODO_qrsmxHL
zY<&fZ3UhayMwuQtpW4DhKht11-}ZyDrP!u}2T33T*TdYQp_mv7sy4sCpG6$jQpzt5
z>4%o>sl-7EXd@#jx)YRv+xukm-xB@KX<gR?naTdi-pKijCxpir?f`OP%TU0%r|3o1
z&1hwKI<+<FTflm!leR<xvUr_}et3kqfa)J1mMxJyCu!^(N61Xe+ch4&U6p3{FQux8
z4!WqX*P3vdKn^SUq5iKPF2aR4Dn3bFyS-bE(h(xJ0uEL8b;j@_$Btl1hDKTE!hGtE
zbw%zJ*%?rqE>oesJuV%7*e?2_Q_T?hYQJ;CP7IQ*ukq9jPpQDZhbp4B+Q|#VfRRa&
z8O~XKv`1sHlF?w8Z7@>NcXMQNPV93E96>%^rvW2g@(pwO)Cy5-IU?70aVC7cNoc5S
zXAm~zIWkasXrH7HszW_dTk9v&QijXRYx~E5<<WXSoKdbtNi>4z`MJ}f!v6c%!~Pv$
z&9{Z_cVx74ul=B9CB$OpEq{qkEPaxZG_c2w$axO=Mthfrs<;mZpZ70}Z@?Lr)0UA@
zCc%Xtkv_^;w|S^)4^CMLzU^_40GNpPrv?+Ep($mLi1NTxeCm>AC@7<IC_e*+JYm1{
zB4~2<j}M0rq8;=5U%i>p7LEsMxE<AV+N3WHLnzjeF_K-Sb|<sfSL{_|4q(JOvkr|c
zNBn8=56MPk>1+rdM4Oz2aUqxROD%20eoD%Y;>VZews8&}&RHKsMkE*B8%Si|=^34N
zLz`M%FBmvqi^RvBTTHWsjsQ75#<Q!_N<|Z{wBl!3b)yF)SeD{k8AU+hmq6LNLFx@o
zsK{5Z>sJ?XkW$mFWw+aTkMfR*L+t@31uL~E28YG(I>euiIF5o{E(UX-Tm_dHz0fvu
zaFv>kR(vVaSGJgD6bCU@AbQ{8mg&eZJZW~%tw$d3H>j_6FH3L%(k?YwkMqh~!tx0H
z;2?jOs~l!lV2sbGV5eRM%D{%#D(y)zQJF}MvYgDq&;pZ}YV)pt$jzXpCV$IF_|iSv
zLZG&vU3}YE{v=synD310nDob}lp3|&#pHk@B1+tcTd|d{!n%bNXAsx5$W&+3;W?ry
zCA&ylkTwJP=cf<&dDk#{uH7<{bgDJ`n6z)v^=oS;O5>Gw&sH+0TP22bn=H5e@B0(n
zyUiro@4*@W!HHDh52%8J-cZ@(Xge0do}D$exTjucY+ZDlWuaqWRk&6T<nialKR+~+
zlU*FO>d*AJNF;1iqXXZ!Jpy`lZgv^KQcwq$OOR_x{o?54n50FS&uScyJ9C0F7V4-h
z%X8GwLA9T*n>y<ib8Jq#4WL$sEAmpP(h0cBZk%i)HfE&};nMMuaj~iCo5bMxQdc#4
zGkNMy<;U9@Sylw{sfa;@QVGP7-MkwE>{ua)lnW|1MmB8Rt8qYPj7OXYwC_$3_E$z&
zZH{Z@6DeO}|I2fofYJ^o#r80bxcoE<kkv*@yr`-tUo;t#l#~UQ^59Frq`_Dbmk@L&
zLv%(ZB)#cy44p2E2vCQ47#De+fUHAuh2~Ys>FBE7V3GK5Q2$AQcfC&!k^VCb{&*`L
zz$Nc=07U>NiuWEq2nFB4RxH(GWc>57l-z?p`I7(sJqwB9(-a6T+23?pdBo)*nCNmE
zm|z|R)QT|1WZ6hjxBi{n>J@Y{OzgtNwjx*>-zwHgxlo+e%r3lphPg&vR3^CqaGjZh
z@IS7^4t6_TUWkKd-snMSJCMPkv3}kKFb>prRduYZm07p^==Yx93eDuMBP_ECYGm$9
zqW(-JZfi<V%u6hjzrc_{-CXBRr)~_|5ZmY~{i)kAO_ve4fBD_Hs%O`uqoaDI2Z#MI
zm1x)MuY!V<nVD-pFv#GN>Ksf?8ZKg>P$y;<*R>pw{pHeJF^&(eWi55~Q2`%d8o1VP
zi;Rz^_NTy(5Wf}>s47*$;CH@YSRa@;>nR0U?Z6?WB)ju8goNjLVfvv1>cvIQN7a~u
z)!{EfdXRTjuB=%rYLy=T+R6Lhl`SR<O7o<~x3FR<Tp^HL-K_}_$_U-m9F(yDWLi?n
zmb0m#?=;umgpmA{m6H(h13q0|1*)@6y&PJ+DRGV-_@brtpzVhBg)@HmGtmbR#lWgs
zmfvVIJk}YG<kPKwSA7U9Z<(I{E+W=n0pBoKx_bYF=~o9u>_byGPxB+mZ2>#O%s~E?
zZK0Ipo$pDRcdtkH2LJAVyr9$MZG-*03+aA*0>QJM`eZ7Mnxf_ltbH|-$dDJ-ggmlc
zBe-;2dM8l8=)S+SD78KW?e)}vROss&G_NhOb>d1$VICGXxv{-@;w7UV&MTkuQj(Pw
zT~e|#Ch4OM`Fb5`4k2}axo3Q1Vy#`IsK8mmQr@Npz#1-I#LAwQQti!Bnb5!bD(U8$
zdO2UUr?9ZP{ZyMDSl=vniyvvu%&=#B6M)a-+`nVsb+~|2u)Os4UR8NP<vlH4GA09g
znznFxxm*NfbYAJ;X0$+`|Ao!0#Y7_-ow5Q|^z3<jM%Z&BzefhqQD0QmJFhXK$AEd9
zdo$V=YVWbF@mC)8G4QYla{?EMTgIQPu6o3%npzqYs*CO30(y%-egrRW1jiH1Q2t;_
zc8yaeeOl;|@H?{Q^o$IL^o5%_dx_%PhViMOK?zi0J8n(|`P0KfNB0s_)&o)xpYlv;
zt6i=e5fz*lu8X6y<IeL7`E{|cOt!yRsS)T7F#8C%?VeRV4cZt_WHJ&wO3pi=e|$S2
z;McD8Yn*DpyT4i9-Y&Lcm%LY}?pvy2Co9P$b4H|~*0d&b-o(S)lux}4r=PDMhc3yv
z;|vL02jRaJJwPO6G4~g_#>Eyff9F9hW?dqlv}%11gLy*9(~9jM_p~>5jpvL!je?+-
z>Yh|$<b_mcI(wH(rG6cPnV@LLzOU^)-6^xs-!AL|<Z=*PvGIF)D<4`<FZvs`Oa7=n
zrNDbxOh*+$#_01RUZ?(dsK9*2p8gr6*CNpaV7{OkA?0cOcKUfRP|~L8{pd!*h=>+i
z{m9_`*eEwcbKQ?SPWOnPhJLk{;v>JRHg6tos~mRF+ebIqKn6sYY*XY(Wau5DVT?O#
zM|5M@0glV#AzQE37k$%WX;36!aess8%QQomS0tZ*ZfzM_&f&JWoZ}a9dRlN&eXLc{
zPM+*8^HH3pho0VrIr~A{mB%g&R_PeJu#YiDK+<-C-#BK!eV>|9wD>%q^VM{Xc}*L)
z3r$cF8GZVnVjP7B50bWqlb$eKtu8G%(3vJFC<CpvV-G!_)O3TLoYANae~_BGd2>jf
zel!1cv@D6INngiZTHKX8X+*iYzw$xaW%XWhQLSU>ANzxGPu3_hM!%w?VS*4p;oXx9
zYu0RqHZ*>ZkIx?d`%3+G4ZpdM=i~Oeup;K0b#>3#2IpfZ$NBxuSa;3}vKqcAD@wrm
zUS4o;%-CS4mO^Wh<>0GLDXCN4NvtHTeUC(TNHQ5Kp_eeV&^wiPM0}#`yE6krakZ@>
zl}fC~R>}NV6gUAsf{kevvGKEWwkVzz&txFtPYRrz985EK39dGJr>n2K<Q2`>`<%@B
z5XMG(BIUiMc@o>+M!2O}Fx;Ds;cvj-EK)<y9jg>SYfkIMBc15=&DrshN7O{-v@azD
z_#9;l4>}m=R=P^?QDQh3!_*g($avAc&Zv9kazyi0Bh`D{+SpE3eT&jIg2>K$`HmCo
zeiTWy7C-;a#cYqxdS883V*7qO1r86KYwyb}sb>7aD{}XmY{hr41cB$&X@AxM_Cldl
z-Jsa=U*6&J+FOGG*~UiuX3;XIN|eqY?L*xy@?2@&{pynKTk5g;_%b?@Rd-%=Q^2`a
zdoiKT_Iucl#|L3U%I`Jt_LjZ+?h}R74F$c<c|XcH*^yyo`CJ_OEOa7ITZZes(sZ96
z#$PDcw4QNJcvAUvka>Orhjr%|cucjhMo|6g<qu~iJAAg2_Ss&)QMl7vvtxQ~OmvF6
zbe}IfOmikK&EFGh|LaXt(8A_$eW?B@Lu1dzRi>oG4YNb3(K2Q0yC!mborFC2xVg>I
zWtNTd#(-ZsPTL~{g4qhMAMifHiBPFOxWM9cVJ)OwVJzb0Ae9BEsgsDb+@4NyB=xLK
zrjnpX{#=`MQlKxi=;xx2kXLn)KR~~cWfCl8J+{;2W4b=YFq<3l<X2<KW?J_1<C%Qb
zVNgB9AA`{MySF>NY`Mv0uWTGIjj{6Po+oyw!8C&CNZ{M+=T-l+mjJ(ixGb^AgXs*3
z8$(F2w@?8v!&*b2xlrQR$R;9xSvC`5uTa^<MFV+smYm<zepn*}aFe5C;wKnwpb@`^
zN_`diu))J=XwSxK*yQttZ=jZe78a*#-Iw;aRM9xTxtl)I<B&ii^5>5K;R@JK=KQqf
z{D5yk!X8xzle-_Y5oSj2=akH*raY5iD8hs+0$RUmaYIWjTcQHeY&kfG#zpQ+nGbn4
zh01`n?|uz^QE_z}w@b9V4z{AmdcSc{C<jH*lX3fCLmZn_+lcO=fKK0?JB2e4?dW5w
zA`7xfR=l@&WL)Hbmf%O1iXRUPU$ccJzF8S_yS`srHj+okpZD>|*25xi{JkdyARpl4
z+NBbvqIn_m6<ke7>uNt}B8hog0kd)6N7~<xg-c2gbbK`DruDPlbFdjqS3AFFW*Hne
zx@bHbQ&oKinr+unLyU<rXY_0{IN10yGx41wZRYVt)2daKP7iQ6#3Yxa*o|q@L#M@q
zFb^_|P(M5H>?_FrHa_bPW#s?_7_O8*UFK+^sk)v?)U>|)-4$DGi0=<ITZkhd`l6|<
zZGM?78uCWv-7hhUOFJ5G6)aNdsDw7S@)}>fad$r_mJfKCA<=>lfTdS!ORK$QZ?v)N
z2cLf@RFn{+3frCj<#}jhEmEo3tDwU)Kp<%CyIkpBpx|bN_Gs9R*h>xAQa=Ef+smk^
z{TVVk$JSo8eSS(L@*$k(;Nb7QS8OT6pG1vCm4gV;>%)6AHl);VCaY^KxLXX7)dTsg
zLz5#@D00y^j_dWC^(c0MKUM$Q_3+1~{&%I=F#R(>_UJWgz3~-9*D{aSHdlJoh_Tyj
zH10AS2dJH`(lWMSb>nq%=OV+`&`{Q@l_<vrN*0aiB^Zz-!}a6C4u01)vDhKW4bA;O
z1ZO=0Q}C`Gdpo+DV}9+_HcU199GDU~XkSwcYGP)U%BQjm)TaD0J|;G}Y$bWwfjmSv
z>6&n!8Rix-alll;!6|yrvQ%=-zoF@o+d~BAe-}+Hg`GaWfnpI45~e9b&rE`sNnyq1
z!m0A$1b9PbJuYmZ95TXTmk}<QR`C3=EqOB}K=kdvR93<ty{PROjSYA0^Lb+wQu9$l
zUWFH)l`lW`DQ%45$9v^QxjDS0@x3<Z++T}5P<=%(Bk{*mKC&LM4AY60&ynk&9r`Yw
z53JWFJG1_lP~-cHjWQ{doO)T-Gj4^S7}uKYAF5|<*4Zd9O~i*)f58u*v#@GyU8JrI
zkI7~iNtNF_3DZZRCmqh}MPy9`?H?G$gyN%TgyTcEn}7yng-coTlhjyW)SctiKPqH_
z=ZE#J%`r-7SkIHg2AWx4{wG9z1fFf<7InnEX8Y#TEyXw;)Y{e>_vKmua&jPHBdYva
zQ@5*Q$UP;kmF`1IGSeGD_og7(x+MdHml5TgjWff)uamZ2RMqs}MTW~Ob_b-t#K8ze
zF_~X(op7E_PfND0Sy@_HmS~)aJcJ+}3n(dBPx@_Vc%6P;_wa*pyS20Z2hcsp!kA(F
zEG`VyOY!YIkS4mnb<j65ezPF0ff7T|Pw1vF#PbKn8r9(f&e|CPyYN}3E!~m1Z|r&n
zRFt;-+`h*$SP5?KJWQLs-8=K(HXVYlkocm~((()qRu#?eo_7lG#NNGIu=#`|tRyb0
z9~<<7HA^m4anS7uQkKBe13+K2Azo!kF=nNtKaBm{NPPv_cRKlMWHsA$R*KnbH5C$3
zNpKYpU$ip(_V5W!OAE2A)SC>k5;J)IcYbIaX4B|M9DCU4yRa_t_V(tX@tjswUZ{k8
zH8FAasN;jj4o>NHS!kj-;in#b0!gFt2hNjgte_W6+U$P><x%z+1zM;2e~o|JI$a%B
z1M!AX{Bn<ZbDn_B;|vL5XHJzG&f^C?D<;+!d~Q&V#0#b|J9&kYv)GV(j#0hY!TABq
zq2=aH;RmS}f*w)Pt)tq0pa0kMx!f(iZ{)gh!2y03kTuwSn7X7JZ?$D!B{9CN^2d82
z;Cq#T#c2{^u6=~@CMI_qxWc&!2wGarInA3L4-b&BVNI^??XNM9#IDCwV_sBl;*XH=
zHpEK;)ResbHAQ>u3z!BhRb^%!x#i1d(qd-zd<!fnV0<UyE1V3z-DE*d9q<Fn&EzE%
z-MIZ~#Xt3dOQA8oU*TluQ9`xG%`}L^fyFhfd-SB~YHPXjP+K-uWOpR=_wQ(ZD?K~W
zOXv9Vd;QZj1Wr^`W3e$d?#nMZbV8Tg0_T0w(@pblh&I=|LA1OlHmO|BfkU)8osiF|
zcj~e=dwDz|>D&xeq-ELPf6x9tMaMq$k!DVu$SK5@;!Qv>wTzkcE-BX2!k^|-06*dU
zMvG`xuidKrs~wqS^~KYww6=JIge20QzicdTv;ID-{PZm>=}8|DE~M)$Fndn8w$cMy
zoEZJ-iznLugY|rXAHYo=1a)3OtaC6=)6&#<GTW2l0|qY<h)l&gIty4EFyi8>)8NjA
z4kramc(~8;NF>dyu+V{wpGHnjrAPbCJq@*A4stm%tgSN-x1>F7pKOGXE{aH?SNaOG
zIorL>&JZMrKP0#O(4*+(jU%l!8+9BEcM>QY5u?QEMB=K4QX89cgj8LTR;sH$J08YF
z8PTfjovC=ZaDc(g$<o}{>KE<(V6Uk1fE2voa?7GaLWYu^8Xo}4jhu<Gg9CDyLmuA3
z&;W+3b{aZzKGyis1so7M7T_&E<QiMl)794RWyn3R2n&~ARQ+UxXI4Knk^Au|36?`j
z&0lWhO%%6%RzOOf5hYzMZ_k#&LU=*{N+RQQO!Y-lY|Bx)!Gx<Ddu|=zy(HDcZaMJz
zM9iX;3Zy~6KPuGs2B+G){~>xV2U!URvz3ERnUcrqgh@V!Q&be@>R9T4eRg25FSor&
zut|9B61p_Mj`?D%7H{llZ&=ZDSLDQtYDB=&#P+~22Se;crYDht(rntgfOWKj>SM3q
zfMfYb$C;U!yLq!eHR9KeQ?s*Mi`IQ$0!K+ZfU<`G%#v8eBS|@!xNEBY*@K+Y5o35o
z9wth>hycd!%_t)?>xN+1Tn`gXL-U7`RRiG2kuH11@8^G5-0R1U->Gq`8-;}PmF?MS
z0zjZiVQi=%C-0pW%q~z+fca$IC(v?HIcu5@ONq|1M6%0_xa%Da5{{9oDbe>9fR?@#
z%6X}RVGl0R1Jw(TSJSWs_L9#-{t>}FHOUoVS#{h=8UNkU0e$ecARY)wgR$rH@wwPu
zny9<7<!O2aq(0x5KpoF8WDk#|5pE$4lBE!+`;}MB0y2r+zr~>c;q@$Y)+Vcz7uR-0
zEml{4#;dAvZ(`@mN8bdYJs?1{tAji({UOMWG#R?r=C>t`@r~2Nc`>XN?@Tz_tzDwd
zBiY6ERB773Bx+@*uLsv_v$!=hLXABH7Nyjj;##XvY!A+vw(%*c9ozjbI=O5`n7||l
z9v+2}Q5fRk%5Nog^beBGCy!vz?7St_Q^QaCdSboZpWcSeEd#3`s09VjOR?+6`u`5t
z!|{pp^MA-Nq^fEM_=a8E=Xoxd@=yM+shh+pTzvh#CGw4?2lU;mkFCL=GD^sqqK_uv
zZb7A`QC<~V0*#kN_{q)3s+C_m@o-0GIg?Rv)TM~Hn~31S^%q>v>c2d4as6vIEVn5w
zEeGTvG<!QmXv5~_mt1C-b;#^|y;tfk!r0NU6>=0uxIF0>Sn9SXg1^IXXZbk&*RC!`
zpgL&gbfE6>q_4YiR5{|EU{Xw@cf^|n8JT>T7OyDU%nrhr`MOrjt4l3z^n$@`JiBKS
zdF(b!=oM3aA1?`0x9iA+R*5!$OOw9a6=1I*x92HqYyRM;nX{xW!`SNQS1~b;_^~rB
z;05y!c@{Q)sxs9C<;dSyRk2?bP~qiG$9eJuRd`87k<AQYeq=vc9cA3x*5bPxquR(?
z@@io#=Y;FI(xOWxmpT9T*21UH7R1j9)TAudvQD^K_9gxtpo2*Aw8D&R!)hGW7r2#!
zqPE60HYAKB6%>p@U$trsWq!3b%sIK-MzvPK|D!PUqrLl&zgtNa(eR)1Sm<^2{|Feb
z!795T5hAS!s&>zGKYWt)wh-mnoJrqf9c%Tm@vnv!Qck7>sRXf`K_QDaCS3kPO|Djl
zVt9hQ8ih}rwtc^mJP4BjC~omof#QC5l29Jpj)eWIu0gvE855*}6O?F(!Abk+;~TWa
z7u>PGI~yuK^8uHMhUAYx&?c^IN>FouLeDVsMxrCr>ALmlqnbTNN@QHKN=gZNlSlP#
zNlC+C`;YE$c}At_v~KTDMxMu%vsN>md)Bpi_TQn;wY<-CCjqRo*&tn?Le9P&-tLKP
za}>7nVuRnw*jovkMV#yzDSy17<CS0+SL~EI6v{V^{yN($2a~oQ6U9xGYiXD>mAq%G
zU`64zYTZBcnz?jfPH+xQ`jjfKt8dY5qr+6iVc!4>O)S3Nr{15inknt5xFgF$5|+yU
zc3Hd<{JHe!E%0;6&d>SDU{%~fK*M*J?f9K1UF~mEQ=ZB4t9)9HRIzzWgZgB5n|uh?
z>cY+YB0i}%Y1JjOJYUB(0G>0T`NE6ZA8UY+73_qDOrp_b@R$gtOSwugD?OLk#$J{v
zrXh$oEb0ilyalF@X8zg$n)mH8aIi(&APAos$7A9WKmdHzZMb}pAqyBC2g)mXt)lR}
zj400rnhmg8sEb^N@UF+;$?QATUp8F%9t(Zvv7gg>2Z~J`v<>jB%w>DFy!C^(|D6zg
zm!u(^F}umQ(Mb(nmDXXuBc;94_;s*WrGfk-@yY$3l-j&1_eijc-tr0wydo)~8wkP1
z=X+TwHL0P-QejUGN<3n4Lvs+ZHGiNrv>NTLp2&z8;mNUf!dQUFknRxkcWlSFB#l+>
zD35>3xjLWPQOHIVl&r-5of86yWTIBWsojSi^%r82k7&-w`F>){B+NGk$Gh~oB*fEs
zwn>RvE{syk&b4jvj@ukjsTA7-0qxzb;1J1P92-c)<F!+;54n?_Y;?@TxXkd`c|WiX
zI8=p&Z04%$ugD~2I}$sip+O#%&MPkAw&u;Y&ihb+mYdl^1V?+{ht-Z;%ba|DTVm(3
z(f$Oq-hgorgfE}%$>hX#m=sl7?00^eac*}~yKl*USlLHYj%x1PZ-Gc!T#n^s)lqpo
z-H~74=v84+OnutfN9qVe9Sc5P0<~8Jgzr;Tu=APqtDKRM$)38>OG-0q&x72Q7E0cL
z5}YdObKkIU_ZA@zqlB!b_S90rK^1WkUB~gvK}RQ+yIdeW@9Wb<`Kr#Z9ClM>j?LtL
z^TG@kN6lM`{GOx*1FT$5>}*8k0|V7u7U=1ibacC`eJ{Zp;e71wCf`?Fn50$_5#`Yg
zD<`{R%T8NCX|czVn^hDqU+$|6J9Y;J6Cm<8_+v8uUxbg%0`B(F<dd+WVNiVCkMukX
zs<Oky(;ylcg)=*5x;%({qgYEa2U{CLno{R4fgL#67bxPyb_AwzIEMKlf})6-*{E;W
zmVR|wdWA}rVasxy&6W-0!a?-u=N06lg?>{+Q)yv(^(?q+Gp2M^9~^a1X0ksW5a&@r
zBkVBXo6O<QGLmWO8}S3(5~Oy&s+XCMm}Cuqj5I@C`m+%SI7c4GvPZ62YK+AF{SBZr
zdDL*Ro6|PASO4f@N)giz&|y`1^l+rMW6D3Y=l&c>972pTiQNtE?;JlKQ7d7qA4npJ
zXuBl|Kar7|9lfcQB`=1~5lMVg#WtikUxk}-L#T$e?b12$hgi>!-CyC%om3W%5T0xw
z=BUQR*-`RFGYK)Wxg<>O+iCN<0K{JSZ6P_1Fla>ZQ>I`dNniO6P$EjRm@}5iP(E1r
zmiAkj)&+E%Q^cFFR`Cq^nHZKQUX}hbCi&5t(bj$bI7zX-u>nuW1ehg_Hyb2Vt52_F
zLEr_d9KXS7)*m;OKhoXkt>cevaMu0?RN}5rrb9tF$9k7|$9GvK1n4GZ4$^a_WZDw^
zBvglsON)Cm=^O(JI4aa3z^a7Yf0y=S7zlUgzo}K2F?`v3a(Lii+_IV)Loj0fu@aOL
zd8tBMK#4r^0vi;gr}&dhuM}w`#ZY1uJI`&&*S6-B3!QN0@~u90Z%q)e%c-jWOqFtC
z%2KyZ17?zCGd?vu)?xA_5_q-<>FojqIMBXd&GouL5?ec7Dr{U0uX&JCSw2Y}IuL6B
zPmBzE4g?d)$SEZ%P4Xy5MX7Y&lYR?1X#VmLK2q45t@sv{psl*s#<19b=_m$$3o##P
zoF<{T#j-<XTg&ngqng%x6;TmgOFg_{bU@_muxG=Kb;qva8Ko!3BKPUJo;R+yLyt(Q
z>;8D>XgJO0KIgB^I#z-qX5u(v@wTlt1EuF8ABcV#mM=NA0tIVMtmLn4J*w4O39dT>
zraX(YfCJ%n2o?n{9<%L%j9nLOv<YDZ3USrWn`}69{`g5vA4FaK!uRcxPVJa!0a#qJ
z-?6d82Hp{=F=R5zoim*bv_5t1rC%|+Yt7`0<bkn(ggv!r+0>8uL~`mc2-fF?ayc%x
zH(zc&OjRYCeU^z1D${?*i7(-^w~EVa9M+Ok(+iny%%`f%c8~pzvLjr1IVCYXUk#FZ
zokSJpIlo=hhU`PF0l5c-577p*T(836<r-sIyp|-JG!b2AUy*T{$zLCA)wu+0h<B(G
z5EWT8e@%btel<JU8rSlG9jb|+ryiIa<@43nl^Y=8Mz0^7|NghoFrsuKhVa!bx&qfa
z<J6C35j%<eBKz7iQZHx7!%^N!H#ig8VJmo(#zFbonZH2gNX57g)5m;eeTs-M9qo@d
zFrqW|L%V#U%5O%I8_8crS8E^-a#|=~iGeQ)F*EYGR?4?*)Ys$oK%z&rihx|kXh)#k
z94$0L=S%7VUvGN)fbr2I8H$YlJla@+!Vu%$iKzqzj1^sT-G;%xVj!WHk<#vY`p_C*
zN3PvRnin|T)8iUhHnPw<z2_8FrQ!!B!+LLa8H`$66QrfHMstr6#B#5a?!W9rR(mW_
zgl>)bgW$7wHLRPEPtG`=tZeHw181l$gL~}res$wK4%t%cu6Kgg;Uy(!rft2C%Ioln
z^YX`vzmi5}X3}$$buKZ$#pvk=$Cr()1kyMx^a0n;FOQX@EDFs2cL*=-LFT^u;Wilu
zqET*=$x%1h^0dg=MgMq90Ns-S%?i)70OPUumBKAkQZfyoFk7j0H0dHv752U`I&&NO
z(*rHe4gxs?nJ`NA%i-ssE>&|LGm(i3I#U`NCgah{19kn{*<uo3d2s-c#je3wL;;1-
z?#Cv+>iNq}Fu*hz6f-Dj=rD+0=xc9+N(Pq*3&GCO<41EViEy*vNk?V-f(x=(F$`V}
z+TRk|0ye+vXGeDja!No9Oj-2zM|$Ji-ptGc;dWH-=$n=ZIF=km^rDez!x=Rj5j64V
z&s=uyTD_Z(C^H)afNy?;h`yyovut#u>D7h==n@mIY@dLxK0;gbld1?_SL0iPN>940
zQL7~pE7DtS{ic~51ZhiRs$5TEt5gWf9h_GVodZpg^S!C36g-!;er-;h`c1I-&xR|%
zzQ}h*ZfbI9g)RqZU6__TuwUrC*mX8<&9TNG8i9Hel>*!os`ef&@|qTw0X<A#cXi?k
z&QI_>F>0hhx?tKu33kN(HDg$%f5_1KXe2m-KR>^nuMy4MYr^PT+tdMHaOA$@iL!jM
ztz5`$sxa+@to;r-85rg{J-1n(EZnno)KzgHz9*@2uvGBCqL#D=k)X*UAra8os|&h>
z?x|WJ8Z?~^+nZQW|K~oK<~)V3)qRZrZ#dok`a^oe2kj5fJ&qiut-|C{;rmh+$i9|C
zDJQ>#G)ZWbG#lLS!gdmCxc-^7tc9b+JLPCy0qqz9!tk(YQBXwSfKb)NH@3tNph#IY
z^M^}NNSUqmwPp3~9fW8FjTGAgLsj~5gnt#(Hv~g6JupDUZ@xd{eSTZC#SD7An2vQK
zyYmLi#4Nthnl26tOBmhDR|HdM?2tVpwhUZc7d4Fm1S0lFmhs=+CAC`2$`4Z7_s9}?
z+C;NTqjAF|D7vvQ=O!DjP*yh=n!!GoB?&b7X!Dyl@9;p^jOFQoDS7~WtNc<LJr6-+
zT)X+D15H9QqiG@b1~^&JYN}YsQ`lJoe8AOpw*Bc9{8R5<9dnvPQeB+}>!(_sMo_Lg
zGiwqV(W+^vf=&QmnK_x~^8*%LEC(Ovb(dPdR3QS{(!z~qhos<TGnpH<4Lx(LegJHg
z_1M;SK9o}@LuIlIa6I;4rHou1X1h<kWol_Xn&LWLkeK!WTq>Z@DrTY1BLK-mufv~k
z;aIs0o!vk3>!Ba7js&&fJ>y~a{32x|qwYkOu;{kXmZ{HmWOo0ft3LBVlLV<M+Jf1q
zx8UL-NhBdc$by2NG!P&z&CF1vRgo%g<7jUVXnKR(PlOtuT81j)ncWs)lQ$1%%ORmu
zhWr=CXnk`%DGawskt0O@I;n70x*2~`LZpXi!ptJuP-!2-;I#zAMT&m+rZ1=#%p#g>
zn8<=FCVFvUZF9%DZEsz=yb4|^uQjaaSy)g#3#Aj(e!Ljn|5!foXBZtj)b449d;-wZ
z<C5n=G_yV6J4tupWN(kugHVC_Q-s)VR2LFV%P;C0252a4={keKgGk`VSxr?DFICbD
zahi0zXgT`mMGGsx<ou_g=(n~tfSERF_ShM`2OL9JQ!;nf&jv4(s`_S?O`A~A&jX*)
z4&)9`2x1NJ@${?>#qC8!A2|ct&HX98!RKL(A83(T(b{uRW1|CbZeDsEzUmks2Qz;9
zCa;?&x)Zi7)%QAm_8d%=ZV0Oov-rUt%O4J=wE8+0)#V3;BIZz~#e;=I`<7thFK!b(
zXpFw$F|oo{U!G$rD}!8^HQk=t@1Y|OCBc73=o;J@s{&N)(9+U%&@r|)brv@`YV9oR
z6{z9HBt?(DR2b5LSP6K|AgQcY)E#9wBNRwgy*U}lAE4md+Y>tb*SCDK7eJ+?R1b|-
zLIgT;SdXFi!D9&szZx2RPSw2dQ&B@jRXQ-e<hNtyL@TaHicutQAW9C)aU>?#m6v<X
zgo>!+^8t*A#d0OQXBo)_Z$e@wE7jEB^S<{Q|90<)unpXYwO0A`f><p!mbp^an(Ab2
zI>qU!p^;wqMg;Zl`O*{VL7k%$-!k)2xpQs!MN4aAg6Y)-OFl4*#*h0y%H+d&(&r#>
zt7T<-|A8t}zCdo65X>%8(DM3&`rf4lU}>Uj9T*lBJk}X0fg%+YbQl*XpYgWjb@I}o
z6tTts)7o1{RrP)CqaY^A2SsTRm2Lq+X+;4K-5@C-Eg{`oK`<!kF6lfXC5<8=-5`x1
zU59SC^XT{eyzjVw-22|&xC|IFj+}k=UTdy7pE>8V)?;a7XOp?s?BNlmQnjv^tgyfv
zG_0B!>XVSO@V=!P)zT)OtTJhC@}{_qJmf4xw`DewUZ0(r|9ZBCfkul%xa#OY0~`q*
zv)AvK4}}fYjjW@x9JFm(I-xW>%FNDiiUFa__-Ib0JQdSX`YySPnkzt2SUprCEdL8l
zuQ@NjkNC6S)B45+UODl3x``g4*(MJLidn;2P^>HuZgtTu=9!bvo}lAB#~=J#BJ%3g
zxb>BF#0lDBq=H&Xg?3C4>SNh&VwBa|FCGxu(cO$bTz@Y(^tdxOEVe(L9`W0^Z%krg
zj#5xKf4}f-Z<k6|-*bMKn3^Ce!;^YxC`FG8Foa6Y6L&^jsB&BX-uX3(d;p!0AL}iQ
zvj(&4zhGy+_@ss)3x|lCXV0P0%JFqvo?`yDx~^BjIu)2rO!Dddz5X0gH&#|UB9tmK
z?ld)e<F3lq-p?nR=E7F`h`AIHaW0(kDOG#8K%4UAbeWP1Yp~m0@Lb;v3}WZ5n>F7*
z?d;NbZ)P=rkSU+J;T@Wyu`z8X=OmD?pX|poplT#5<xmz=RPVdT3=S`jHdfi+yAubE
zo`QpsJqr4JURotjWp=Du$$2aLO_r?sX2}zgS>f`8(8kZ0$Bq(6<CRD~n?^1gN(+fS
z7aqY+)W?o|rN{XlP9OdaLO1TD#Yo?R+V)&mI)V9wg4it_G}S$c4Vs98Y1d28O8yxW
zbs1{$_c;j#?(rKuyFpw0<#qim$FI$0bZ*|zZBryS-6N>Pa)EaR7!Y&ohb=NkbIQ<t
z2~bi*Q0EvW&$_z6`~Aa%Z#LfI*jNv$>0Ed>m@DX7se7b0ASrw~va^=;MF!aGE$+f#
zRNCocOb$==$Il^<mu75T?ZDLU2WZR6d@Kv67aonp7BSeDo}-$ZXCmV(F1Q{WLE(p(
z-)G38cdYH}(0k~ST}2I9%=Ql)XIpe5g7@9ZUfvg_!GZD&pDmT_!O%id;6i^Is^MLG
z3nNDjj^V?HfbA`)XvPZ#i6%4*n)CWv_JK81nvgM%XI`z#8UA>#znplFm{|L9OQ##v
z?ulD(ishojR!=X~&dAoc-7dKt2=`M7$mpXk0J<ihn12BMPaxgAINm2ffD7ewVklqm
zY(uTr+#&wxENnTw&oHGRMg+OVt+Ke2WkH!n(f;)#pWDz%z!Vgum%gE_5*~ry82D;l
zu5!G-vU@<GJA%B1EDnkZZ?1;i=DwX|=ht9K#Pn}gaDJbN6{3-WOi=#wd!An7-)`+~
zZn}Kq{*2&|jem=b^b+h&i@x}ISdA`1jpZjzWUYzGodWq)f!eI!DkY<D#cu`Q{%m0r
zZ<vqXjo<z+GJ&xDYm@4|!1v8k0%iBs)OK#1b#pN=-aVnggG~Fe6vcQvq2X{F+4ZB%
ziyUKLnoJ54d&|UZ#n9KmZnb4Pek-{$#3R~qo8=tHqer-%LxucG?DWi5k<p1A-+gH8
zW|~!H2rfAnYF#cIDsJfLp9%Z|#ISz7iYC&Vq`Qkh6vgs&jfUTBnDtbawI9E+U->@t
zP*Q`$G5VG6b=P=#xhLeYXh}TG!uK^z!vsp<kRwu#g7tOcG0=#8_n=PwV$Q2H{a^$~
zW`0=GqV;N>FJ_WZ)DAkf7W`#J-FEfUtG{C0dJ7`@I=9u-?hOxyB*Zq3tLJYxAM79P
z+LVrRtvw41GcYrjU?p50KxBff;{5#$%rxk9V&+S~s++&yf?_ilFSo(d#{GeF)dPXA
zeRzPLfQ>ZJVs&ym&C@)7E_N7E>&tnC@n`cSsA_F)QI9%U<ZjokYNxT&>(lK27?bC7
z#$)K54Xo&KG0s9ZwvfyvDv0#Xc@`$u&3cFka|G4#7)eM97}hDI+jX!Xf0KD$Xsnw3
z46$v2V)VaD7<2y}apgBWu4Hxj%D8y`YmtuuDDW{fEFp2+5j70~Nfoh+)J-~_&y#rb
zeK<2rWT&U9WBNWLuO)+()fzUBJzhB3bn=e{@|d_Bw$!VzTyyWQA!Rg?)6cD0h4<k^
zKN6GQG0vqF^;(A9-ST-Wv~Ph65YT>{ro=qu>UeG<4|3OjL*iefO7Gu4ao;Aw@k@-)
z<0(OM#oD7>F7n$=v&3u1y>)gy3l13_JaR_%s1s^xW$UeM5(fvDp`Q4ubexfwQGJ$7
zPk7@FkW<+28r%!d9cLL6XY5d7L#H~}5gv|c7dQB?^b<=`T7>w$2hf<QJN5jFtM`W4
zd=we^dVf)qrlJ3(OJu!ev``mw3QYKj{Bk-E<TPv+NHj4#o0!LIHI~wYyNl_4g3nT9
zy`L@}E%x=GQ4SAt7^PU;Ak1%nzJ!D%H=ta?v@eK`fOqvLc{4FSm=}@VRmieT&6B&h
zuxzHvNdqi@Xemn)62tw#-yEh=rKbf1*MU?K7sE>oXo`<Y^sBNSf+){Qjg^HmoWjfm
zUxk#V6Wo_#p~;$BX%a<~&dV9Kh{|EwC}2f(4c7&<JQ=L?*gTD6EtBx`4kO2ZFK^e&
z*y0QjR8k-I<?KaBNKm=h6M0wKqnXaMS<IuImcP@`zQKt(-SK)%*r5dVqxIR$tRC=}
zo**LT=h3H~a@DK4Z;mBUxjwrEZ+2fF5O9na&fxJJTtq*8Zu$c{`!m#nqe-U?P)797
z`u~{vIpn?`&DqnAo#ZF`Xz4PWDF&UGG8>!yM|l^rDxQ#&7g&v)PcOAZnagniX;H?6
zsI{vhM)KVDxrty)x5peTR<-0`A-Ug0csIF$f}?8P)<=HO<JUopOOkN@mx6mp`_Bq-
z<7v#uSRGEKphrx*?0Q7?J~uFQgt^w-UH9F+8<;ac@R)7L{g$F{i;tUI(sR2Iz@H*9
zG4oVjN|B4FbQLHZeQs1hl(BOoe8J90YGG$bBFvZ4uR(dyZl=xoH2hB(J5uq3joGr3
z|A@pjx>#|*=F{iNJTqmA-fdlG4px(IUW8B#i5{<@gc`OEnxYeMJU_o3{z?Q%uBfBe
zYUxTRx#$-=$++jaSt{9NGgI~L%!V&+=eQB)1D&xV-QNiM1uWq#+Dp&p^%`eY|834+
zUZKZ!PTqDS>-dg|eRMyM$@ruC3EXu&p$3Fy8k$eU8?^u(N3Xzl^0Te(g_R$@{p=KX
z<e;2h6F_#lHn!#{Q0$4Dtm?(9SNYlz!gr6pP1;35S?J1dfyo?tC_Z62WTC<My}LE6
zPR}EqM}$S>Ra(W40o2u?Bxjy0)-TQyjM+dsUlrNePuTYo^Cb%9!Gv3IP7UOelhdbd
zKB-FZn~!z~Og<YCM>st{dJcE@=?rag1WzVkwK}0O0r_jCJmbT3>cc0hhGiVkqaoj&
zWsqWFogd9-Ulrk2H97R1Vg$Yx6dg(8Fa2t(V_XP#4`|~|IvFC+Zfk;=@<CMrxi6tx
z-OkrYioAm=_Q4bpS(UzEnbEYs6fSf|(yd49#?`H!-UzqdT<ln6>jTDG7ci6kpu*sa
z9;}_tT@q+i%<6Nu-527dzW4r>OoK1sSbg<%QKAvT@#kY=N@{$-5p}L`8%<NUnl;dG
z_q~wvL5hw|vnrx;atH~#hmgF8Zaq9Dl$WQ$Ys}>9pglWG-t;W@I?p9oqTO~I_lRB6
z9&G$em%02poerA4cwE-gEG)$M`!40+#wF1r(DNgPVu@9HD$#Yy99{-~!Pw$AK~GK0
zzX=ML&+)c@A*tS6x0!Oa#UJi?J?Bq-o$5Owr_dV}7DWGv1|7AhUQ5i6oNLV%)#-^U
z8sCB0Wxwk|0M&nunCKTX74;g?(eqS>Yw7Gz0>j2HKOYzG4#*oD7?1AcB#0ikRkdqm
zsWg|-L{7{v$(X;m;Gu?|EvtG1b%v}0u`;o&XHGIgRqE(jU}QkGUFfbc-Ku&xYi7$T
zg0A@%wKDpFHC(7{5r?+la$3W{HsExwRNjo3=Gh>t;kwMNVW!N?q{14SbJe>lu+gSC
z2HkHhcFJ9}NR+Rk!QJVmzZ`PdRGxDekaoZ6)kwYC3Q75{`JiGbj;8_pMc%<w8ZfS?
z<#?8zkh}@wHnPSA$p97V@yF$EM3qg%3AM;FJo4EVOP;Fj?t5NA?w$N+B2(f^)pa5X
zz_uYC8o75|hrUC%$FzVhzq9`$zYFh*A~@%gST1|LvOplo-sa((S=|{+cM@EtoEWu+
z+8G>+1GcmNq(-;5s=Ycrussl%c(R_T3(`)jdAuurTa*Z{MrRJCnndxyVqQQ%0oR6`
z6W+vz1LbB^*1y5?#vKBnNMY_j7j`)N(&W_eeE4@<SuZ2=MUq1b*Jx`#OQU*9bl=-S
za@&83HQP4%*bDj#=0iA$CZcO`Aglj$wE2W#vs;JApgr?S(Z-tb$f5Z6X;RGQiZ|H`
zA}r&^cn&)6SqdyDHa2K-jpgyhA06>;_GM`FD&nbTDHZViW;?6I43)*4fjyi{>sEo0
z#l#S1=!}e}erdQ77_##4SmkRp_Xdd?E@ttt1-&(ZqT+S6h}|JgL&!yq&B!Bp3YoKd
z=}@u({<~{wv(dcHPrYY~UB1Hpv*Vq!B)PgR(8>CTkxt*jK+NKtAeF3aQx4NCG~CEl
zQA$w6Uka|<`{gscuyk-(yqG#W-PKb?E?;6TO+$D$Vc0DSssTbJhBOV_<je6?Hhgze
zWRh<rC%v7^RxO9utepsFjuLd<HnBbhTOV$J{YgP`@AOMu%VDCb1;puWt4C}sb5^C>
z=CxI7VR0=<w-I*s;Kw(LGhaiv%gk;z`XVUs>7+2o?AMCg>Y}mn^VV{gRZO3+to!=x
zIF=Hr$00spWa$$Wz)K?Gn;IEL6df*WmUil8JNx02{RQcwowfX(dpkpGZ(FW~g>knw
zUewdFs5)>-HxyLUMXJAT_uQ=HH&P0i@<~g>5?%YhN1^hLpd1eqT<M6p_eQLaiJ22I
zH7C3HZ<cw&4oRla<t9~hFyT6!HJ{J%uu>T6*4!BuF5>0;ee)1Z11K9MdybzyQ>mJC
zD>9Rv@vYf)tuyu<D)9T>+Db7^aLI^!$OUm*+bZnntk5|MDIN$7X<}xBRy)sgS^q^V
zg^L$r?DpB=nzvG-`fJeb&alLNT~>)gBO2o9$L+Qk976E@9<#u<e|bhqN?;AOsduzD
zB;lef$5-;qLR~yZOm%mU86lEO>MC+&tPcHd>+TJ!rXE^bWaa#oYh=i8DTk<xRxG&9
zn^@Ex&tJWFIkGmpMSYMowsx<`UvQ}CQ=w5<;@u9Xm6x$;J~P(aB~W;!<K{^43L(mE
zb3lJ|EH0c{ps6WX-XzEF=>VD!bU)v=HkT38hCjyESp|DC`0g=#y3tt6MSq`qcy_I4
zJN~&awb(Olp*6^P4O$UsAr8~N1!Q0X>-+J7qTjw}`IgI;*~ku-cnTB^ez#=-Tc3{-
zIx`ys)(Tk)+_eK057Nk2jOjwV#(UzI?;2v!hsSWKMQ`Pqf2$r3X#P|I@Wb)bq<p-i
zvvkoC<L$!|h0UDkM{ap-VR1{@_i`ix3=zZU{y?M4;^)uoTX1D;&`lJ21s&ofZ4Sm;
zd<i-4-bKYmU7lD*UA$W=u$VK(POlYuZL`E`WN&0l(%j0bYIwK2+!nJTkMkHR=C?Qk
zb1U5sM@Q!;7IjQq-tB+B%~e)>ORzLE>QfY%K~MJ!C;KGtpu72w1UZA3K18JO%Of<g
zCwk;84?+)VCYMn_`g478VV<5IdMK0T`o2|WZL+0wrU@0_zui*G;S<XVDiAK`DUoyn
zR*>R_caItvlT=vD7N=2yG}cvDC!Op*7wCV@C1;D`^#%aiZEMTS!d^ev9$~z_EpRv4
zty9T1JJI?`ebtL$aM{vOTuzf5A$Qs1Rm&XTry9It6*qkFxUDIRR_TA>lz=xodi&pM
z)Xd62G@_8<@f}s`+vy6X<E^<xl^<E!KXn;KQ;>zI>%J1p`u^#a-CZv3kJ33L_GW#b
zWXD$2G!;EWWM3@b8A#oy;0?<tyqeH17my&Sv&6<ucKPQk+*b%SMPEmrg?dXBkYLw7
zeg2+uJvKm9@8t7)Up>Jr6)8b$v}@mhC7ag8i_h8FR`%}6rZFd4_hcVvYbx5=RV=)a
ziJi1l5e`g9r2X%?H)R8GwIT?9k&fsBK$*-3s}z;Bd78N-#=Pp9r@s6gJoQ0M+p8k&
z_#MI#t|@D5#tc<L0=X4F`p9hC)AO2q3+9>rAw;r?ssnT5?R)*jtY-FleI4=2!YvQa
zZ%r9!9l2P(pRL#o6e$h})g9S2N!~Bq)!U=$lgLuKVmnRY)}A;kVapP|w+AiXNkF8D
zj<}ariC)JWjfy7xH)<zfBtig8=vx_xy?8?Yi~I>>2n(a~kUHZe$PKB7l&oW(n*88z
zPfNeLw2-@VFS46QF;k@oV>3_@ls+*#;|skAZF2#ZVhS@_^Gx5=RJ_j5xUnurp~dSM
zoQdb=0`d;$J1Y{Qitr<sDz~+@h=K6TQE~3Rd%U!)i6-Tddc`af7Jel@GN>>wp#SsD
z6Hz%>5(Ex>svFe3-;Ta|-W6hkaK9xP#Dh#z7<6|RH5^u(&`6v;3*{s)r*!$RO%jK<
z|D}I)hQ*}MrQEoq;E?6jHjuKl@{r(pwz;TXCr%z0ZTaas%NP6E39~=)zXmL16fEVj
z%|qD)f$51ubtTAT1?6^gL*MQx>>%3V%Zs(~B}CtR5dQ&{w)8$QLnaxXL#J!)N4S*l
z8K;K(3>M$W@#`v1<9=E;z7)<9K6doH>WVjX=AI33Xf7I+uG=&Y*GFSo**<QpZ(W}S
zUKMaX%zl9n*5au;*vo*LuVMGMpy)MZF!Fn9TGwb>IOo~T84T^!cX7g@9~#pO9G7g2
z&lWY@KQ(SWD)+`Dp{6)G;a^11dFM})b<FJ<1&*lhCB~d08MMfDBZCbGFEuS}G9u%9
zWFZq&lDFpIFl}$k@@6u0lY%iNHB3*8_V%3*TzQBdq>55O1mvfY+HavZGq;wedU|HF
zRI;4ypPrn^gmkxehjfirgh1}y5qWEK;UUnyo969spy~LDIfD^r<)42pEs;HjJg22D
zWpg1#EbDh659!OH)!;|B^ws)i=juXDwqn^nWo9q8o=g-fXsgMc42&BbL|+E>@1U>)
zmE`sK@6r!mZQ3B#*z*0W_CO$rHNv)mFZ->A&<K<Tda(DH>l-NAi_9ti>kf7Y{^Xf+
z!`0NVxh_majoWSKLD+(C8VirKl#A!MJ`QXTqDawPn56je%zgJe8xFvOw@YUkLV;Ky
zzbbOyEX~VAG9b73+V`W)pA)ZkD-hl(X9IjuolDj>2i|M%4v8pUmA$*LDwF6+XuONj
z+}xFcB}|hbB<*ah<~`gY=;=#Y=5l&}|Ms2C4dJL{%vxv=F6MZxN4ulR!sY<lbfJKs
zF&qzo_OE3z!g{sgm#JkNVAG#A%f{Y~dyCBuKp}3Z?+3PYkeST^f^Y+5yQ;^*z{U`<
zVwE2k!|#Kea9v>dUTkcE{r4Ny*P-9*6EaVOv~T_XeQ?L{znsi#<SsUWi?X~ZTKl8+
zC6J+82kqYCr+tP!Oz)-1DqaXSiPt;d6IkEtk)QKV9LXTP>b64%z0m^7(g8QaG^j!j
zy8UB&1K|i(yS~B}wPjj0G-}~{_ADDR#S0}0=PAquHA&q(r{Sr0%tGasoUjg8^_*Aw
z8w^i~&(zaHBG7ydzrlJ0^~l(<YAEX!rx*vOtxWpy%kKC1RqFB=%g+>JX4{xUnQ}ky
z7*9FoPyg=EiH8O9_1)zNU-8MOM)ic8$MPmfWm*=qw4NI|fNY)$U~dTIkJ*hl^7xwU
zPys>n)6!NXwAHelh7s?)jE+y!y=IUq{k*47=a{}Fh+M$9&;C={pLP|XQ{~|lddKnO
ze?z1zScr5R9R73NdxQWYA;Y<}Z+CT8smCHfJ*)4~WBTl<I&v%vjn)X`bnV@FtyHDa
zpaSq|L1pOfVpVC5bdzq6K7oF{s#j8t*}n?2<}Ybf1lN4bpE9s$iSVI$YH5|fQ^r?p
zm8Vs(SsaNQRdsN~Zky_stmI_50*6Ug(`AX=aj6I5XTY4+6WO>hUz>Sx?CtF6i@!v9
zTHM7f?<q7RAdmIsGVIIhWv|$s*d>>_^lW!{(3Y0phoDhz1Z{RQi#210%>^tE(WC$Q
zM5OeU9}b5&_Uw$2FDRWbgRr$DD(2IRW4A3$OVMmC((0(AYex8ILhr8rhK%<IJ1d%3
zBqP^<fRNf`A!1a8DV)vIk+06ndsCnhVJX%`+F%PS-lt~@dH{7OM;g7xMlmG?GZSeF
zfs&8-wyyn)XpZgZ%pLH?15SdJbNWu7JGU_tYt3!c9Cl8ax{7&c)wNUk%k)Sw<t5h*
zYAIG0oVlS^pDUhYyTe0I4Gm<BQRMo-!+aY^&w$3kp;0GoKf<u{k!;DABN3`UT(*Z-
zpSaDP$MG!T^WWY{_70_$n@Rs2scu7iHwfySr9^EH*bfKqoOl2UDRwtEwgGrXGAFxs
z>+k)zxNbjv>tuJ<*>wUMg~>8f=86{@LzRABy($T8W8bsf)yOYINZ5VOJ7M1=GA2}d
zL(=#cba#vSgVy>6{n!wJ<SfIbXI2fSqBPK~ggq{D(P9r^tf3TeFXV&2SlYL?ZZCo<
z$q-$oX5ed(_2pgEBK5>1$uH_jXDO|Hec$I^ZH0#eG<E;Aon&bq(zv!ZrmS2#j+zf1
z1sABKWhBznrluCY&FrmSomlQmD)?J>V^1#nc<LhUL&W#MxvQgx8n`jN=8u_KW=svZ
z-3%s~vvR+~&LY?i)(j2n*yHasw)$OMr-e_r!Bxh&Jg49-{gjgkTQ7IVtLv~7szg1{
zWYp}>-Y_Y8V$~Cb`}UpMADwf!9gm$qT$7?-e~&NICKEh&bm8(b0URQ!sBU@Q+}L>g
zr3peSWC};WlTlC@Xh;v<`!)W6?^$a5R9aeESVY9Br)FmF3JTVaqSUI~kN0Nd1f2rm
zz-i&Rtn*$rL2pl6<Gnk*S_OULjgeAO%s{XqlYMS);cZ@{<5o#$he_wEMuBS-4F@gi
z&}j(JG!%-Da|VhM1QAT~_-nR|Pj5N<5z7f+Vp`ERtwNV8dI(K;(YUR)v2R+W1M}+>
za5NH*T`vGXc?#Rb2Vk#%{ElZSFfgb1w-+SKoKEvU{NnLoOI4~^R_dUWv?7Y`_wy%u
z7Dpz@X4>WCZ3nDs1t;IqGz}rzM+>?Tar#pSWEFTem4+=*iD({>vhE->kMb2v1GxC^
z{9e21?Tj?oYcDTtQp_f<^wjk8Sl6=vwW2|A)O%|Mf~^0HC|h0I*H^p6tTE3$cf1~B
z@JoS}mK^i<m$8$x;FC}P(q}j(q!YRZ%}kBs1`9TiBjRHb_pqA4>$JRrZzh2l&&CeK
zj9-IYwDZGAwN6SS#ZRsez(#(Dk~SRQ^ZTBj5gHNi9Uo2%bw#qD1$fj)?!k|b$xmzw
zfp)SnDF69s|E0j!Z|shT2$byX6q&;B6qeG~w)iDKh^H(X+8bwL;3a-3$8mD9;0fir
zSO))FaO)5uAV$&VeL2nZ5D%E&^Yi`wu*U?sU72?Yw#-UUo)c`(F!~S?d!X>Va>u&A
zX*_6@hVvHt-ALhC4jUeNpQ1P+HW^u);?flP3Y_v%v(Y&k16YrHd())5lhGOXoSDNq
zW4Y*H7s%v>jqS&jL{4u3hwmQ~wv12mvPC8TFc5;ED!HN-L1$#5toY<zL9)IwA|wA_
z(fya`A9DTsS1Qf^5AdS6OlyT{S_b;UPXfq&D0q$1jL66-MK^OC9JqM-3?4;3m9XzZ
zR7Y!`s8y#|zDbEt<pZ81lvi|CtEt*#thbnnDtF$%R!aGQC~Z3$3-0h@3poC(gH!$4
zZEwuB>b)aW70#O?CsapzZadV}s_~y*9F~PXQ@Zf__2V04r02!UZ;0Qn{h0Z<Q1Q`a
zp8<6p2Opgl&4Kc#qWBBnF6O-ba>YpXebNo4?hCTlPCd{R6TA3&Q^t7Z;GA(p)Tz`_
znVAEA#@S0eH3uy`ck7<bG{uisX#_ubCE(7W+{nXyjrxk>{NA7QXKS?-F_T`?Go+@o
zukZwHxpmI?{3Fi$JhQ*;V_&AHM@|avEb+e9O$UaWq;9)-W}9oVS1}~wL3^z%dx*Vb
z;E??(Y3!LSm`B$t@1bd&j&tvd90XO%{`@%NHEF#-P^S>Ym&f~}bF19&J`yByio`*y
zN5UlN@<^;_w$cU?6%ryBA3?`^0Uggcj+ZTMpHdjKCp@3g%gIpwZR0nwfWxj#gqXLN
z3g_)l*^}5qJ6@`Pjz`W{s=6AdTe+-=ZymH4?wgOglBY`#@CJ3A*ju4R9hOsLwMu*G
zzdq2cu_$GzK8`3KIe1w@zg#H5r;sh`xI%~8-ygsWk1;`pyQZXCn$-YuS9uDnIVT>+
zXV08;HG97aSJ%iqju%vW!1Cwmtf;>FDbZ7-`X(3?6}y8~#NYJy@Km{(J{nfT$FO_8
z&TqL>%&_-1L8nUBf{?nuFZ6oXNR4o|zWP=X6Td?Fc4U2~xJp<_YPz}X`jxFfE2U(Y
zv9U^3B;-z|6FS@fob&7Ic!A6ZHlHGnTov9i%4az^YxvB#xNY#Ex-nzwsz!f~HpNK6
zZ4k*B8JW%XAnI@%qd(UPhirG2DF%`)-`T2%Join`iRJBkKWyD)_o6CnmZ@`@Q@*cZ
z^c@Ku$R3jUCVcTorJE04RbxeWx^&s)BC;YMk(0PgOs8wP6DBD;@O-c9v8U($;#;k0
zZrLM1NwzUF;vH*=!Usj`6Z*w(-*;zvzUce#=Zvn(i)T2coG0R)iW&LE?-qTLNgT6J
z5nJBgurhq?nrm`bb{Q##bH_%v4_lhsTXK#+)rx<aiV<CC?$0ZtUyfSQDw#w+HfQO{
zo0$G4)oTgcoYZl$_2PBTdt*hB{WD#*=&<&taG7}kNBZQb>*kkZlcv&;)PZBCkB9Di
zH~t*5qC#?p^F0cV?OMz6?1bswe~NKf=koU(I{s=E&r4TQrX-ipE2zhon;v$*e)s3|
z{iSF|;=WJGB76LbcRflJ7!h0Bj*8KByf?kGDe?|NjFMO3#9H{tKihq(SouRu4awn=
zna|lX^Lesb4g(*JntyOIIZ-NiAC9^?v{iP5t4?$(8%mfxMqlT>`bW-h;U2#3HD#6N
z<Gg|J>{aBL)(PqOuxT)&%1tMGVw<c@yt~Vx5FaD)+M+(^L}3w`GxaDuNg8V1Rp>bW
zBqtac&*VsF5SXIi7qV^f(?rh<lcC2Vd{-W!5YSi^vPty!JFGjWFi=!g6dh5s{`n*A
zt<6=v?aXC=LSCa$hGmqf>hqCkuTvfgQK`5aA_G6s5iG;MIt!;7<^~H4*`)rQ>E}S0
z81PyvxhIRB@-EUeVB9n3XvrcZzlgcLWlE8@#hfJt1FBwPrRu}{s@Z~gRK;T4Q%4@b
z1RJv*Ybp;|_xEna691mfk@oTq*V4$G8suN0r!u%jyS4s$jiSjl`Q<Qi0ei9MQ0)Hj
z$&s|3dT#HyI(h(p)P*Z_g1^3MX0w@db__fqWVGZ&EYsCjqmvfVOU7na*oBs9!Q}kQ
zabk`u&BQp{Ewq7Hz;#C(&mhLkLn@n38S^nBn}#a4^u!*0z1ro=j7O=IU%$1&<gaJT
zqVYHDZuf%2+A<rLdx&009-hpPs7Xxh3x1`U@j@f&c<aYAVeUyakI%N@B-TGuQ^RP7
zoLG>De02(;Z*A%w=8|Bo_F~4WelIGoKa0oS%gg5E2_LP6rqGa7>4%G|uK9DFmr+VX
zt0MUqd->W@C>FE0QZ{>|6j^Mp&B+*-EO}Pr|AQYwXmriy0Y^}o3a39!1}SV5Gm8Q1
zS5K<Z5#dww!X91yazWzj#k&~`*0<{6{O`Xi2QBm{QV)aT?p?By@fmwHSB@lxGiP?S
zsBtl-zWPjbw`&*vnAZd^3=awMm_kTcT(s(ZnbPO>_O37bby6MEP3@py{;Qlov&b{W
zQ^Z+8Qa(OpbU<G57eQZ+B_%;sb_Pf~&`~lj|2;kX{uu?#w`GJdfRdH&_WcWHjbT$v
zDj)RfX|9V>Nd?n6m><8%em|>-A1^0+V4HlrhN_fHXtLK^H7U&MoK`Y-EiYM;?cPL_
zWGRE_!Zfnd|B@q~UFH&Y+;DK@AHf4wpwan@{i=y*-4xE9C9twQ);sdH#t~{Xa*B1X
zTfK|xC8|i==WXeZ*50|ubo2UtOK!T+eg?JDV#~C#E=6-PGNz1{wBN04?BmGeLZ`!A
z_5;s8hak~3la((z@}pcaDK39cbGG?<@8H7wGAGOpRQHAjAJh1gf-SqmlNVL^IvlcJ
z^pghSxrABo$`1(bzlQU`(pzXL%<Fc$_D$!u5DJQ6?O(56antP$JulDo%w%x1Li0{Z
zEa-WkWBJ)3-_a4DA!*Y3d7tC&gVlN~U_M4cT$mrq*C{%&_r0!BqBJ9Jbpor<v|Ya9
z8VS)yNy#xac5;itch;3@)HN&Wt8)yR4vm%zWaNq~&JVliTC^<ss7qPcYW1wykzFx6
zb$4|h{l5<~vBu=A`0Yg2QzM}6+h$k0j_X`(xa}8h(xTDIM5o0BbY#ozOKmPoX#9&8
zvSw0?Ti)jlZw?ZQvIrC<h*X(<0OxygtmI+Jc@Lqbv#4&`leo0jk&%V>f6mUuc`5MP
zi<JIO!1;m54$+jioeFIuNQv?b8jtMobA*vDcUI9EJodBSjLO~CN?Gj7?ks;jaCo?5
z#T!$`=e&~Q5|SFZubYOct@Rr_4&rLN^VyOWj#2G&yGW|;@b}<toZbg<&j`{MhFH`V
zGpe$JIpq|;fttk4<BRpZgnhh`I@!{T1EDoRK}v9*Gek?i4CHz^9dEXt9*xL8DZ!GX
zPyM`wZu7fP&4Ue2a}rJY(&Be;EGjbJrMgzJJy%HM*_%RZmG|xS-^s}IgV9QF0g2MZ
zG0%sI`TGyhNXdPz&~3B+?RfeIamh!f-;x$LF>Q6K=u?vU{b$B4hdEFby_4L7mr}lM
z*uV8xKngouhd$p$F?Kv-5h~JE75{sl{(b64U$q{{Whnpirfz|Z*jON*RA{^8i%2<p
zeo#nIVrRp5VJT^O_3FChOpHx2IAek1U)<z1;W6!gQKcnJny%mM%T&5H&-JuluXlHO
z<ec`IEM&;{ZtpLiXaW1pOp4RzRLw=Ssd~1Mai7Q9>C7$@Q{<|RcREC_3$3{js5TlM
z@3?%mj~2RK8Mim=av@GGlI3_-yZ6ite+PjO=lu|wo_fRNK)d+aqmwexlc9dnFWUJA
z9a=aG*YU^jvFbAt625(0EUluVCm&1Au4&5O)g3JUzI6r|nOB-@!}rd(3r!W#Wj%O_
zAaL;1B_TmcP1kbGyOv7xdkS2E0_$HSrl-{>_J|xpd*)%E2p#H-5a$HdVZuu~VgY>B
zOtBopB#LRh?*}JwadGA{wrlwK_`l0*Lo@I&GS@1JuU)&AdX|7fgF5A#HZHE3WTvo;
z^4{JaoTeAr)|ert{K(e_6#~s=7V25#QTrhO$6L34cXVV)5DQSPtT%L6MYc7#FjYNK
zY+5xb)rvc<W?QDc(&z>HFz!c(Y;vbr%RB{dwTzGJWq%Km!MMYfyR9Yb)?5j&pRu(S
z97Jf1Ui@6UQF+qQ(xR+8Hn?N^;K75>10M%IfV_<eu{-)=Ef0n=(+fn95O)t@ryYLg
z{E^mi60_US`Sw(7Y7UcM7^1q{rx;nI#rOBVqsV=eqoi-K#+@UcDk5jm%&0RaTS7S1
zcePyVrfOT_H(p58HEej5vhF+Qv^6q#i&Ip_=##B9;&oj7Kv$Q!>vqBUY$Du^=+4At
z8k$7SK2n<re1)C08mi1w2wu5;gg%+(XX~1{aE)Q6n_HQ%HXMu=)vzo4e;WEL78{fX
z-q!NZ2U$-B_nl`~@H;jZ_D`d_-&U)utKRv$B6{_&Ro*9du^OQcNzFhsa(Z)@{Kh;L
zyFES1oDySR-|ZE?JEqXL_V<q!&M9J$Q>v*>2Q3b>wHEew1SLy3H0%`_wTGSQWT!~k
zL+Pp<PEJ^UjMtPK{0$mO|0IJjU+3nzRdvv-2>akDqu%yzpggSCwvn`4dw;e+(SDin
zVi=&J9z%)68IdXrroU(V;|S?2^XJhjoICw<d4HdV)AW!Pz1UK`i1eLIb=<)aa#3{g
zsi}+KRaANWX+2(!y7@<M=>g$<U7}}K!lv?Pa$ntp@pt<hvUR!D|Iu8-+`v9+4k=mY
zUWZT?|9F0<o!!hr?9sWcwqtLjz?$-hZnE*tk&r3scr*X?t_q|Im{WbV#$*(t7>a-e
z(u`>>q{QskKEeL<#7$~pkP!CtG#~kt>{@J@lB~X^GVn8k`%>M)kH5o(xep)6i~nhu
z_lt=W!OGz10@HuokT;WaRai$YWnQ3y#TRXk>5PmPU37x`;Za^7-bMbx&tZv?j41p}
zJv|XmPZCD2aQu$AV7fCtzM65T5j9(Ws%AKhnvNV7cm2YJf_qnTRiYBnhpjSa?mlOV
z4X`72<7v!NwiSpENp-z)g(TxuAHG~dn6<(O_fvRzfv0dVN|oGyC0KB92*tntmm=Nu
z_y4Mm30*t$Uoxi7O#f=NVHB=a{?*?C1+HW!=f6XQQ~OX6^LMUsUcUJ_@qd2hKV$R%
zdST@M;GX|?SN`8!81UDhufL=5|9Iu~Vu6<#`)iFf;Ry*=4wqDLd+vnC#<C&^0)k~4
z)aOW8n3xpw^^@tCZ``=gM~UEV7=uFsjTi*+3$2ynh?I75aS5T*)6*MVcJ#aYRSLq(
z9<sNGf9X<~KABqME1&iC_20H-bSUsv*m<^vodmP9jCy)wx^6sg>frGCo2cTIs_FK4
z!Ms46q?k3z%v0aKeJipU;8-3iZdqW;yp?XYE<ph9l3P|e#Juk!w?@ibL9Ly)6OU|<
z&Lo-dR~pY+i?fd>dGX#)@#3XUc3edGV)OciuY&@*vBPPcz8|;!5?<ldwg~Tx*lB*T
zNv1Q~UK(5gov!+KI2x8M5$xr&32t<#x;KOa%JRZSFK*!!e+8ZUu#k{jG&D4pKR<n4
zxQAo@>PeyTJm`Ie(g~j&ZeYcp;qA^>|9SWy8~YzqgW&ux4fQ`$bE)T!Y|cZR`<5T#
z;&PQU<WM1?od){VH~g;Re$sSwEb%<rQb#G(`4aSj+%&wHNkDY<s;Yv*4M+~EJ=}7p
za5B$;n0-2EFG+=bOhHnJNl1+0ImQ4Q{WS#~p=;9rai{;AB@l-0!^Np($x%o5gYLZI
zN))qlI#$S!H|+d<OS`3?T<lLiKZ@nE>;~!Y-b_VWtPbCRu3O$MTH3t8b4b$JdsP1f
zka10VHH^7$v{48;KIdgTrJxm+o&}}O`C;ptz64>nu=x1AA3q*DEDx!ApCwQNb??3_
zxXu?~G3&P6k>!tX-w4E=@Q~^0olQjY`R!(|RfzEOKUrH_dmj`u;C1HW@5L5r=32_Q
z)h)4LwLDE#J0hH7pSoY2hDeZbdTMH_!2ly8BP+Z`&vP$rrD{i)fa1=NkFRkmFRV)8
zIAdkG<nNByc=vJP^xoQ-?HCv-vzuOA%y~oOsiCgxl6LCUDHRQkOi<A?1vzvz$<qoC
z4-XGFL5Muu<p7@h&Wp)M5JW-ttoJE5j@th3Q2DU1uvC~fM0oA{RrBGJJP=EBm}w<v
z*DQJrx9xgzwBt}utey!PcxsJj6dDW8a^j#(3fEN5Pt?f8-AgXB1LcD;j~yKlS7*L`
zTftZVM@NS|7Z=xlPvY_St_mDFRE}C+2%QdcwDoy-x!)I8!7zgD$t_fjkjrNQ$0a6F
zQBk;~ePnCVqBV-kd>}ub&a5Y0R;}Um1idLQSVB7&a_o*W8R-7~mJ(;0OIG<J3EKxc
z5rWgsS770XhSH+U;aRf`T{q`$0@=PfyFeSKk`(J{Ypr1--V?LZp|JB@Z-&$B`0m}i
z=B6e^XXh^l1_rr-cuBrwzR6aj<tQX%dv#oIA?o$((<r5b?PUu#Qk>O#lZ&{{RUrE~
zzlg3;18qoHLP=Q7@)(PJMff8lGGB<{y-EK5mjG)$4xzYfl`HJFZF+pTH&pFW;pOe!
zm!-@|K+c`on&N=t9LZU8s1#zqm{+<UgNxz5${?d4Cgug!CM$dd2eiEhF+spqwzjsL
zTLXF)aDDT^&uMO1XK^y+e4Cgjah<bhJa-?UtF|?xOP%IAA9RQuur1Ex>i({Mh37h#
z{N~z~E6Iu}5@th1W@&*Hdo!`To`<_J1m}{}nwpxrG8I$kP$0)$u+W!d0t=0I88i%I
zgxyN2PmT`^kb!gO&Mj^a!^X1*-oK}&3>5`nDbhOQI_H6P2xDsT7W(sI;rU_xhJ}Wv
zz?#5*&i;Hx;^zJ$deIs=i0xCv?fr0g=#ElybuCYqjZObm_hzwV*h=~F<F|l0w~ltk
zE%rC(H9tLlCE&1-3Rb8QNA^;+AFaoULKeEyE>3eyq=kfrZoz0rd=U{8)Iy=KGm6z=
zV}E?Mb{J3tMpJyzIG*+1Jq>tK862SlSg6Hlxr3Cn^wcBimyt{Y%?;DGS&~85vC5pI
zrDsl`o-1F>R|c0PvO9s-mLROPTDcLZs)(z>CL*HSJxIHc>zrtWv_zP}y}kGM?`1_-
zek5Vn{4p?~&d<-US!^DDyw`!4!)L^B8T!HyJay2<aiC&+g7;GCk)&kB#R*V`)QUTM
zU-kk4frH+40ejOd!~LCAPLzN%CW-1F$$;Od6e?hGGx74OI#$GAWx~~E;^dU?$&inv
zlZ_WpwHzu62K|e~OLYo1HlM62*Ca+A26UQRT6&fS3)$r^)>6WQ-zR@B_c072-U}ie
z-szDbMZV`hKlO5b{rYvKMV#>&-h@sByA}hAg8SL|MT=9qa}!^^LsDT9ScHTsd$~OJ
zH-qiBD{12P&);`QnK<*3duhBzFJ$*I@r3x{(NSL!Vy2Ahp6|(!rVb4aRemmnvpQ)n
zP;A<D3)KzQ#V&_ei!Yuy!F4I6l}o&|Ogo>V>2z%cAe8&e-9X29F(rXg0^iCehnK`a
zoR{%LW#VuQ?2I18u2m9-=FH&1schR`9##eCguN#gmS%`gYzp?Cfgp0reR87`F>a$k
zDapvll!@QQQ8|NksLZsq4?)DP{}tXjkQG_27J)Ha?9b~-JUKK$brY*=Co{;#Xb@k0
zd1ilmIp@QN%Xv=<CG2L~v5KFmg_4W8+GR>SChcLV28>VfCWJ*!4($M>U!i4_J5|ea
z<;oTFvC1;K`GUslJ=Gou$ZvZA`gjk~wdK-r^Zwi?A<p?Y3OtuW>!0E{C&NrxjC^@9
z9SEK&K-c-#84)i6=VYvxUH{!wA7OZJfZA!~D<xm;MWbiWOhG?2Bz1S!`OMj~nsMh}
zs>&sb<ej5&S0E=RHwv5|E=4oQC$wIv#L1=xCxs4=h)9Phg_un>1Ps+6HGn;Ozq~Lk
zwjS3#I@lSQj?`=S<Er(Wo|!RUo2=hIK3H|w#pwI^_+(iO<WulHkNN`{gm6eS045l>
z$7P9l?9X*c(i98)gX0I<f_b`Cnyl*iDoRQr;90JblV|-yEiea5Rvr9uZs7S^X)vz;
z;Ln$D3<&;9H~4=6#Qy(6r7Uxu$;d;94)8rxVECr*Mdet3p4O;nP*~VC01EKM>xD%H
zVerU0uD^UpuaS~w_!3aEK)?Yh-G!cvpzE%4H|rZ47s2egWqTGfdD|fzQ~kqF6^P#v
z46J1idPdn86`y$_Jp+S}ZsT+e3<*w$Wuk=E?n3bNwSvf=CN+FcuUvF=GzFKz>j(}V
z&B>CYqC3z5HEWCEb)AiOqC;)WwDrt&rC10U$1A8@7L1qk3+?3)_p$6U!rml37O}r1
zf*c&H^SeA&HT*X-d1;Fs;+I5Y$ccwj+2ZVMhPs}IDquJDPzKrGi;Juf#}M$T0ENsO
z1E~Q>1w$ZbllybV{Meg6PmTeZ+Ao}pbaa^y-~3Z=b@$T4n>TObU$~&?N5XF3BYU^c
z{t@;TUfye>z`@y#|MpFcTF6-;q&H8ilxW(@2v*tLRFG&td+Er>%uFUQ&1~@C<>2{?
z!KQ6XN0D-eoP*a|%*@Ooso>5mz_81gKDS(>qRIu>vjD3?&Czk&rSKUTcvA>HHhSlm
zFR+L|dc*{ikf&MP{rbYyBJ)0$Ag!I^>>X<<srL!)>!@P$KB-r)UM=)yh3dM_)5n8x
zdaVPbQq55_`~4&M{oA(+;KzCG=MV*$y@~<K>Xq_EP>kZW{rz~y5uAAVsFiw!<MPNr
zvBdx=-pwsGF(wLG`!<hZC4Z7irp|-CD)_f;kJp^+9~`I;R>F96{|Kf9V4%FT|NZ+T
zdIp9CtmG}IdS0WULD$yS5)CK1ePfSoNsHz&Sx9hOVdgWg^F;`TKIwZWc<!6!=~R?x
zJaEqU@;b$9+L_cl45CaqRc<@xSOngc@}_!khSJB^SM4{)KxXzxO}ZI&T~Y`++Z1pK
zIxckrrrlg`;j{o#0XStD?E;a{p@qHF6#l9{Uk?^335Rwj#6DBV6>RTO!8Z#V8ymux
z&m?^%vV}gfMRlosBR(wu^9~lJV@^|HQ_1%QgU=U6$_nYQwkTjfr;s)q2cg^d61mbx
z_rofmi=^hYFLCMLSB|^wj5<k&Rm=+5&G?tw&wuRg?ZuyF|Eynkc586}-A~h`)pxK{
zM<~js=%2vK$S4WJ&#Kr=f&j1ru+FNul3)Q&7Q)C|@I=jRN{ZoCgjmDMNW$iz^g@tG
zth9hv?*((phkxTr@3YCmY%|D2-9VXlCcP3l#5|<tw|xXcOpt1om6b_7eynhEd_<4J
zI`6~1jh1-3<}l{<wY9F?Cx!O=n`UK@sw1<r*dbe?rw8>Cuz`H;Su(hX<Bi0VIn1v*
zD5W{j+>D8i&LW`Xn_I|1N7$FV)vk8m6CYoVQKCNB$;wDjGBwSdoSGUv0F`4?7<w$8
zq60}?nQ?%pS;J_}Vky82q4wK!9<#@Buu3<3<VCZbR>w4}-OCVgIO+xm(on-pH(LPo
z?x?o|QoJp9+6?rR(`B+s4qy;)oG)DktGh-hdiaSs>$^I@e}1cxdmEq?%dM=W)GpYu
zyEY|@GS<+DasiX)Zsd=bvzw@Wl}=X(E}q^>{_$g`2t!!nY3<4i3QrxEes&43bS6I}
z^~^w!$P9I~ZaN*7Ps8ER3PEpV+ypQn@xF*jd)$kTwFz%R-OcU30%X~-B03^MZ9xc@
zheJ>WtYK+t<)ZQCVNSV>9)CYSyC7CenpH?dJr449Dw->;`(DssZ(jZ@1Rb0t73=D~
zbRr_6J2n%w1_K%C>E`Wm{1k|wV{QzeW%idZ_vq>A!v)?zQWD2+M~6aJ@0&+2j!BtB
z`Z~`<o6yDU+%2iRnubLvMhS;`b8owDjRS&>-Uw?nE}Z{*jwXzbU8lTA%uc0rVW<X%
zu8TfWTWP^EI5-#pK_s0S%nZF%M|-;?O}L=zrl|vj%T^&C#H5{#=b@46j0pBt6aFND
zVe{`<^E*MSIt=C=JjOiaF?}#x#8gyE*9%>@%<CH(A~e0y)6>fb+^SZq53t^<w)Xx`
zWKB&Cpc<pnVfZE{C7m7H{{HT&(G1}85f?ks!Ouq0(NlpmqWL@Hh#pwI^2Q9(ModOF
z*KSZc>;UrXj|D1pEg<_Z8}GsID@G8#g%jOK4xN$llbt{+{>f?~m^$0pcL4#A&?=M}
zvSf4rU95D~YFNKitOtk}3rYZ3ETEC29GM<${KX=Da7AvQ#_N4<5ljXD;zc`aCo7j3
zn|lao#Bl7oMW|<jpG=#!vM3zvh1-O=FdMJ-NDFj<s5c^6q<bEU4;7=HU#Bd@9G1sB
zJ3DPl3C^GYKnJNOQE9vuRpa;XVS+*WWai`5RiI?AvGV18W~&EwOoc8;Vxy2BG@lI&
z4YyV*Ru}3?^}_5+m=0heAbtTk`RlE>{iOo-Fm;5S*&}&*xg#}3T(X4?)se&-*iuw6
z0XD133|pIOtl&J^oeHYA^7Hq%*E<GKViYrT2Eoq9o-iS`fQl6Z_*Zb6QXw&St#v0$
zJ{CLRojSD%<}n9wvbnW2tY;n$-nKQ-`1JJanPt2jL}lPO&;RY)^E=ZUM{#s!v+ePH
zAz1A$3<<V8mZSt@H5zGWYx{nbgqnJ}`ViuP%HYl++(k<jXunVc3bx22;eRptnwy7*
z9tAkCFWzY^D=Lc2=k@C!#)ZU!>xBXl|71$Re6psXLK#+Cz~8VsEx{Z$3lAd0W$*hb
zuu?t>RDYi`QrES~qYMy7z179@=Q{y7(xFysPO8<=x;n7yVWu^D2%>yeDby{@C%?a2
zK~9dJY0c&<YZd)tVfGJ=NbZ0hQWb*AQHObIJ`a())M^wGFpte(&%NnLrAF|2qi(BE
zMRHiV@G}sh7F2V%2_FAXZwR1HXNn|Y)%K8uW|eEPlCpAmWaO-_R^J_2P&7A({1^hx
zCqR4iV^+JEU-Z!U3s)b312-n`n03=Z$YT+O1FEZf=6eCsLxg95G-J!_eV<&sz<j8@
z$O{_W1$%)ltos9>1OE8~9YR2MJL&toY6@QS-_`B_cFeaH`mkPiyfl<i9*c<KQ`^Mj
zi*ng4hf5$TM~+%??tC{UR@FHu<<|AwGuoNP2DPzDmzBiOmg#Ol4d7w!v+Gq?j<|fQ
zs~e5Xkc-dW+}vEafbC5>-^La|6}~kDP;UL84Prt#IOJqw?kGNen&h^3Dz+`Xvs1xg
zbxg;Pl+)4L4r1CO1XWHhbr~6dY#O&yGw!;`4o@b5i;K&B=^>OIAYIef*IyW`s<@6E
zzJLgnI^quaO6kFaS5`Ggj$j2}X6K-S#%DdohQdaln&T}r(eTi+<4$l2BO;%mqOBbV
z!uKOP?6+=-U1gM0ff^!Y7%IV|>N4;v)jy4A!t1fQW5p$rarNMZpEcnQZM*}SF!3R&
zfEEJCc^Z!F;L3^+=!y;M@u4Jw%|k=zV)K5tbWcn?b%3Njq{x0=$>+@*=pXKgIIlM`
z=4^Swcpx|lFR?&RWe6Kr$$5;!Tn@1IJKYk=%Ep!}RXv`H3f=72vJQ7~t8(26`|u&j
zxGg5!^#E#Jzc~JpzIyp`5=_i|5OQdkR;Zi%k$cx&Z#H}3Nb8BOJ{GajM>O8UA}Q?0
z?*#u!PEJk>6x=z#c1!%AcHUkMk$nnxQS1F57%cMS$&;Zny9_?dp;QQ6W`Mh~$q7UX
zT2MKK>^Kv`7`C8zmhgHmw3y<Y+HvvDGO@GEva02F)|?z!1B_BT*jdH8Vz8-HNd9uH
z$Ez0x(1`+0E7~jCzWQ?=$Z;XoOTyF6A%2|Io!H7;nVm_%bzx1Ydj>(gP@mZb_&5aM
zbfzN_4S#3~?i%otN=Je)ulufzY}D-s@F<{DKQ6v1n4>a%oQDq<p;>O9rIxFa>@@D7
z4mYF#Hh}#g5R*6Eoa=%qdXadUlaH^Y*9EFZ$sPyWW(X8BG)taG1K87|m;NKG>ENSJ
zLd^$DQo+x1@N>NfQ7J%nRsn&b$w!&FfRS@xyhBpO{Yd5kn-|@~@?a)hG?X|MF5}?f
zx}zRFQ2B^OaJ>1SJ}FE!205Knm~<rc`(R%aBE$L^oa~YH>6R^+vWGRu;kqB>XU}5N
zAy%lVML=;d4H7l%<A$W_*ZDFF35`sChxfcR`J1Dul;!5KVT*RIlkXUtsb{CAQedqQ
zC}3GqNh`mwxEM#{7B{yNfKml;7}ypCEIYQq34n3;!zl&qGa)`?a#rt(L3-GW093Gt
zx}Ug&M7DgQ2uxWROf4JK%yUw(Z@k`iTehc2;BQwJ9(1zc(H~_+>^1-Iy$pEm|8pzD
m@PAZ&|BMax8vn;v-a6?eWq(@gFW-p03`$(?QNGwyum1}zUQOfx

diff --git a/dev/articles/lazy_tensor.html b/dev/articles/lazy_tensor.html
index 64487e81..037223b4 100644
--- a/dev/articles/lazy_tensor.html
+++ b/dev/articles/lazy_tensor.html
@@ -387,7 +387,7 @@ <h2 id="digging-into-internals">Digging Into Internals<a class="anchor" aria-lab
 <span><span class="co">#&gt; &lt;DataDescriptor: 1 ops&gt;</span></span>
 <span><span class="co">#&gt; * dataset_shapes: [x: (NA,1)]</span></span>
 <span><span class="co">#&gt; * input_map: (x) -&gt; Graph</span></span>
-<span><span class="co">#&gt; * pointer: nop.6b05fd.x.output</span></span>
+<span><span class="co">#&gt; * pointer: nop.181509.x.output</span></span>
 <span><span class="co">#&gt; * shape: [(NA,1)]</span></span></code></pre></div>
 <p>The printed output of the data descriptor informs us about:</p>
 <ul>
@@ -434,7 +434,7 @@ <h2 id="digging-into-internals">Digging Into Internals<a class="anchor" aria-lab
 <span><span class="co">#&gt; &lt;DataDescriptor: 1 ops&gt;</span></span>
 <span><span class="co">#&gt; * dataset_shapes: [x: (NA,1)]</span></span>
 <span><span class="co">#&gt; * input_map: (x) -&gt; Graph</span></span>
-<span><span class="co">#&gt; * pointer: nop.6b05fd.x.output</span></span>
+<span><span class="co">#&gt; * pointer: nop.181509.x.output</span></span>
 <span><span class="co">#&gt; * shape: [(NA,1)]</span></span></code></pre></div>
 <p>What happens during <code>materialize(lt[1])</code> is the
 following:</p>
@@ -484,8 +484,8 @@ <h2 id="digging-into-internals">Digging Into Internals<a class="anchor" aria-lab
 <span><span class="co">#&gt; Graph with 2 PipeOps:</span></span>
 <span><span class="co">#&gt;            ID         State sccssors    prdcssors</span></span>
 <span><span class="co">#&gt;        &lt;char&gt;        &lt;char&gt;   &lt;char&gt;       &lt;char&gt;</span></span>
-<span><span class="co">#&gt;  nop.6b05fd.x        &lt;list&gt;   poly.x             </span></span>
-<span><span class="co">#&gt;        poly.x &lt;&lt;UNTRAINED&gt;&gt;          nop.6b05fd.x</span></span></code></pre></div>
+<span><span class="co">#&gt;  nop.181509.x        &lt;list&gt;   poly.x             </span></span>
+<span><span class="co">#&gt;        poly.x &lt;&lt;UNTRAINED&gt;&gt;          nop.181509.x</span></span></code></pre></div>
 <p>We see that the <code>$graph</code> has a new pipeop with id
 <code>"poly.x"</code> and the output <code>pointer</code> points to
 <code>poly.x</code>. Also we see that the shape of the tensor is now
diff --git a/dev/pkgdown.yml b/dev/pkgdown.yml
index e3bde776..510fc9da 100644
--- a/dev/pkgdown.yml
+++ b/dev/pkgdown.yml
@@ -7,7 +7,7 @@ articles:
   articles/internals_pipeop_torch: internals_pipeop_torch.html
   articles/lazy_tensor: lazy_tensor.html
   articles/pipeop_torch: pipeop_torch.html
-last_built: 2024-11-26T10:52Z
+last_built: 2024-12-02T20:30Z
 urls:
   reference: https://mlr3torch.mlr-org.com/reference
   article: https://mlr3torch.mlr-org.com/articles
diff --git a/dev/reference/DataDescriptor.html b/dev/reference/DataDescriptor.html
index 0b6c3171..483f81a1 100644
--- a/dev/reference/DataDescriptor.html
+++ b/dev/reference/DataDescriptor.html
@@ -263,14 +263,14 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> &lt;DataDescriptor: 1 ops&gt;</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * dataset_shapes: [x: (NA,3,3)]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * input_map: (x) -&gt; Graph</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.05a0f2.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.ae2791.x.output</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * shape: [(NA,3,3)]</span>
 <span class="r-in"><span><span class="co"># is the same as using the converter:</span></span></span>
 <span class="r-in"><span><span class="fu"><a href="as_data_descriptor.html">as_data_descriptor</a></span><span class="op">(</span><span class="va">ds</span>, <span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span>x <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="cn">NA</span>, <span class="fl">3</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> &lt;DataDescriptor: 1 ops&gt;</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * dataset_shapes: [x: (NA,3,3)]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * input_map: (x) -&gt; Graph</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.05a0f2.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.ae2791.x.output</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * shape: [(NA,3,3)]</span>
 </code></pre></div>
     </div>
diff --git a/dev/reference/TorchCallback.html b/dev/reference/TorchCallback.html
index 7ecb8ae7..9cf7ae29 100644
--- a/dev/reference/TorchCallback.html
+++ b/dev/reference/TorchCallback.html
@@ -210,7 +210,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-in"><span><span class="va">torch_callback</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> &lt;TorchCallback:checkpoint&gt; Checkpoint</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * Generator: CallbackSetCheckpoint</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> * Parameters: path=/tmp/Rtmpc8mJ4Q/file1a057ccb48bf, freq=1</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * Parameters: path=/tmp/RtmpzHy548/file2541149855cf, freq=1</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * Packages: mlr3torch,torch</span>
 <span class="r-in"><span><span class="va">torch_callback</span><span class="op">$</span><span class="va">label</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [1] "Checkpoint"</span>
diff --git a/dev/reference/as_data_descriptor.html b/dev/reference/as_data_descriptor.html
index 050293c7..a55806e1 100644
--- a/dev/reference/as_data_descriptor.html
+++ b/dev/reference/as_data_descriptor.html
@@ -93,7 +93,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> &lt;DataDescriptor: 1 ops&gt;</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * dataset_shapes: [x: (NA,4)]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * input_map: (x) -&gt; Graph</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.fa4271.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.6ce271.x.output</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * shape: [(NA,4)]</span>
 <span class="r-in"><span></span></span>
 <span class="r-in"><span><span class="co"># if the dataset has a .getbatch method, the shapes are inferred</span></span></span>
@@ -106,7 +106,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> &lt;DataDescriptor: 1 ops&gt;</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * dataset_shapes: [x: (NA,4)]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * input_map: (x) -&gt; Graph</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.ef35d5.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.9ba523.x.output</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> * shape: [(NA,4)]</span>
 </code></pre></div>
     </div>
diff --git a/dev/reference/materialize.html b/dev/reference/materialize.html
index 13d9bd5e..c43207d1 100644
--- a/dev/reference/materialize.html
+++ b/dev/reference/materialize.html
@@ -132,86 +132,86 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
     <div class="sourceCode"><pre class="sourceCode r"><code><span class="r-in"><span><span class="va">lt1</span> <span class="op">=</span> <span class="fu"><a href="as_lazy_tensor.html">as_lazy_tensor</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/pkg/torch/man/torch_randn.html" class="external-link">torch_randn</a></span><span class="op">(</span><span class="fl">10</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span></span></span>
 <span class="r-in"><span><span class="fu">materialize</span><span class="op">(</span><span class="va">lt1</span>, rbind <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7209 -0.7791 -0.1200</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8322 -0.3782 -1.4658</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.9278  0.1341 -2.9706</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4715  0.3795  2.1664</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3711 -1.0344 -0.5415</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6425 -0.0933  1.2480</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0744  0.1560 -0.1901</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.9826 -1.4531 -0.8456</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5194  0.4390 -0.9260</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2598  0.4431 -0.5103</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1876  0.0088  1.4593</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1544 -1.7849 -0.9684</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4133  0.4283  1.5972</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1213  1.9914  0.4018</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4719 -0.4199 -0.5979</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1147  1.8712  0.3224</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2409  1.8200 -0.1913</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0128  1.6925 -0.1613</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0527 -0.0583  1.5428</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3041  0.7279  0.2507</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{10,3} ]</span>
 <span class="r-in"><span><span class="fu">materialize</span><span class="op">(</span><span class="va">lt1</span>, rbind <span class="op">=</span> <span class="cn">FALSE</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[1]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7209</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7791</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1200</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1876</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0088</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4593</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[2]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8322</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3782</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4658</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1544</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.7849</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9684</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[3]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.9278</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1341</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -2.9706</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4133</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4283</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.5972</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[4]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4715</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3795</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.1664</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1213</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.9914</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4018</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[5]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3711</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0344</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5415</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4719</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4199</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5979</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[6]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6425</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0933</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.2480</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1147</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8712</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3224</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[7]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0744</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1560</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1901</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2409</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8200</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1913</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[8]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.9826</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4531</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8456</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0128</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6925</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1613</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[9]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5194</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4390</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9260</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0527</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0583</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.5428</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [[10]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2598</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4431</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5103</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3041</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7279</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2507</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-in"><span><span class="va">lt2</span> <span class="op">=</span> <span class="fu"><a href="as_lazy_tensor.html">as_lazy_tensor</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/pkg/torch/man/torch_randn.html" class="external-link">torch_randn</a></span><span class="op">(</span><span class="fl">10</span>, <span class="fl">4</span><span class="op">)</span><span class="op">)</span></span></span>
@@ -219,184 +219,184 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-in"><span><span class="fu">materialize</span><span class="op">(</span><span class="va">d</span>, rbind <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7209 -0.7791 -0.1200</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8322 -0.3782 -1.4658</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.9278  0.1341 -2.9706</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4715  0.3795  2.1664</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3711 -1.0344 -0.5415</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6425 -0.0933  1.2480</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0744  0.1560 -0.1901</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.9826 -1.4531 -0.8456</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5194  0.4390 -0.9260</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2598  0.4431 -0.5103</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1876  0.0088  1.4593</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1544 -1.7849 -0.9684</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4133  0.4283  1.5972</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1213  1.9914  0.4018</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4719 -0.4199 -0.5979</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1147  1.8712  0.3224</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2409  1.8200 -0.1913</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0128  1.6925 -0.1613</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0527 -0.0583  1.5428</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3041  0.7279  0.2507</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{10,3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9499  1.2267  0.3966 -0.1104</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.7213  0.0694  0.6292  0.4429</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1034 -0.2247 -1.0382 -0.8711</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3446 -0.7110 -0.2305  1.1702</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4673 -1.1430  0.0230  0.2525</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2598  0.9152 -1.4227 -0.0557</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -2.2192  2.6817  1.2324  0.2709</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2812 -0.9407  0.8546  1.1632</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8758 -1.2456 -1.4916  0.2601</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5359 -0.7665  0.9438  0.1353</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5711 -0.6314 -0.7923  0.0621</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9379  0.8469 -0.1051  0.7225</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2402 -0.9588 -0.8237 -0.3911</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2728 -0.8515 -1.2257  0.1618</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5972  1.4824  0.8495  1.8945</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9398 -0.5522 -1.4726  0.3737</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6116 -0.5517  0.6216 -2.0019</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4594 -1.2072  0.9961 -0.5702</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0091  0.3094 -0.0514 -1.8272</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9225  0.5034  0.1760 -1.4413</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{10,4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-in"><span><span class="fu">materialize</span><span class="op">(</span><span class="va">d</span>, rbind <span class="op">=</span> <span class="cn">FALSE</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[1]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7209</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7791</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1200</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1876</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0088</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4593</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[2]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8322</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3782</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4658</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1544</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.7849</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9684</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[3]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.9278</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1341</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -2.9706</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4133</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4283</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.5972</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[4]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4715</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3795</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.1664</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1213</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.9914</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4018</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[5]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3711</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0344</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5415</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4719</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4199</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5979</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[6]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6425</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0933</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.2480</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1147</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8712</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3224</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[7]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0744</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1560</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1901</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2409</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8200</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1913</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[8]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.9826</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4531</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8456</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0128</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6925</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1613</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[9]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5194</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4390</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9260</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0527</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0583</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.5428</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt1[[10]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2598</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4431</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5103</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3041</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7279</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2507</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[1]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9499</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.2267</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3966</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1104</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5711</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6314</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7923</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0621</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[2]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.7213</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0694</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6292</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4429</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9379</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8469</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1051</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7225</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[3]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1034</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2247</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0382</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8711</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2402</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9588</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8237</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3911</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[4]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3446</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7110</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2305</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1702</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2728</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8515</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2257</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1618</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[5]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4673</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1430</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0230</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2525</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5972</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4824</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8495</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8945</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[6]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2598</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9152</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4227</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0557</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9398</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5522</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4726</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3737</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[7]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -2.2192</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.6817</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.2324</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2709</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6116</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5517</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6216</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.0019</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[8]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2812</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9407</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8546</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1632</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4594</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2072</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9961</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5702</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[9]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8758</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2456</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4916</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2601</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0091</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3094</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0514</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.8272</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $lt2[[10]]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5359</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7665</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9438</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1353</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9225</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5034</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1760</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4413</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
diff --git a/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html b/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html
index 3959006d..4059f1e7 100644
--- a/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html
+++ b/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html
@@ -274,35 +274,35 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-in"><span><span class="va">x_batch2</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> (1,1,.,.) = </span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.5945 -1.0047  0.0898  0.4695 -0.2213  0.0279</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.0332 -1.4109  0.8082 -0.1438  0.1753 -1.1941</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3561 -0.5219 -0.0985 -0.4688 -0.4243 -0.1469</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.4475 -0.5437 -0.2712  0.1507  0.7601 -2.2758</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.0112 -0.6460 -0.5519  0.0411  1.0907  0.7192</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3734 -1.7688  0.9091 -0.0721  0.1462 -0.0802</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.3536  0.5422 -0.5122 -0.4266 -0.0916 -0.4592</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.8402  1.3915 -0.7842  0.3108  0.4223 -0.6371</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.7064 -1.1143 -1.0290 -1.5323 -0.1813  0.2049</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0286 -0.4911 -0.4830 -0.1494 -0.5065 -0.0469</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.6232 -0.3729  0.1369  1.1254 -0.5872  0.6150</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0530 -0.7461  1.2229  0.3075  0.2254 -0.2231</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> (2,1,.,.) = </span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.7234  0.1754 -1.0752 -1.2271 -0.0561  0.6646</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.6509  0.3914 -0.5936  0.3779 -0.6972  0.0340</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.1612  1.0706  0.4121 -0.3568 -0.2576  0.6846</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.6529  0.0693 -0.8919  1.0970 -0.4039  0.6887</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -1.3010 -0.1821  0.6225  1.1529 -0.2668 -0.5625</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -1.1438  0.1229 -0.4223  0.5343 -0.4232  0.5311</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.5584 -0.6006  0.0852  0.1745 -0.4200 -0.5432</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.3563  0.6179 -0.4114  0.0964 -0.3026 -0.0497</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.2654  0.0973 -0.1497 -0.0353  1.1315  0.6398</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.6037 -0.6516 -0.8290 -1.4187 -0.7332 -1.3848</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.3522  0.2617 -0.8548  0.1738  0.1908  0.1583</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1979 -1.1566 -0.4095  0.3285  1.1347 -0.8909</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> (1,2,.,.) = </span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.1896 -0.4246 -0.9544  0.0690 -1.6374  0.4171</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.0969 -0.5784 -0.3845  0.3485  0.8230  1.1430</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0990  1.2661  0.0993 -0.4821 -1.1926  0.2399</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.9176  0.1813 -0.0787  0.4138  1.5121  0.4998</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.0572 -0.2697 -0.0084 -0.7172  0.1109  0.5671</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0192 -0.3271 -0.6918  0.7929  0.9149  0.5626</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.8666  0.1124 -0.3811 -0.3970 -0.4885 -0.9129</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -1.1868 -1.8638  0.2943  0.8698  0.6637 -0.0425</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3130  0.8464 -0.3372  0.3382  0.7500  0.0065</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1226 -1.1815  0.6687  0.8796 -0.1691  0.4671</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0456  1.0095  0.9920 -1.1715 -1.3439  0.0340</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.5361  0.2356 -1.7181  0.9764 -1.5450 -0.0521</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> (2,2,.,.) = </span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.7294 -0.2803 -0.5699  0.5159 -0.0475  0.2107</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3567  0.9719 -0.7477 -0.4505  0.5632  0.7329</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.1405 -0.0291 -0.3008 -0.4238  0.2812  0.9680</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.1327  0.6031  0.2286 -0.0610 -0.5833 -0.8078</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.0608  0.2079  1.1355  0.2255 -0.2593  0.5523</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   1.1174  0.1390  0.5125 -0.4994 -0.1428 -0.2354</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1027  0.6353  0.7274 -0.5460 -1.0755 -0.5056</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.4653 -0.2932  0.7616  0.9773 -0.1486 -0.0574</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3783 -0.3556  1.6875  1.2363  0.1012 -0.2989</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.8542  0.3894  0.9231  0.8197 -0.3083  0.0473</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> ... [the output was truncated (use n=-1 to disable)]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{2,3,6,6} ]</span>
 </code></pre></div>
diff --git a/dev/reference/mlr_pipeops_torch_model_classif.html b/dev/reference/mlr_pipeops_torch_model_classif.html
index f190ca42..1989ad21 100644
--- a/dev/reference/mlr_pipeops_torch_model_classif.html
+++ b/dev/reference/mlr_pipeops_torch_model_classif.html
@@ -274,17 +274,17 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $optimizer$state$`1`$exp_avg</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0260  0.0059 -0.2193 -0.1682</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5636  0.1927  0.8308  0.4352</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5376 -0.1987 -0.6115 -0.2670</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1067 -0.0173 -0.3737 -0.2571</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2698 -0.0843 -0.3516 -0.1529</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3765  0.1016  0.7253  0.4100</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3,4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $optimizer$state$`1`$exp_avg_sq</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> 0.01 *</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1396  0.0058  1.4116  0.6514</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   2.8716  0.3722  4.9687  1.2888</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   3.0630  0.4077  4.3327  0.8889</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1868  0.0053  2.1873  1.0252</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   1.4240  0.1388  2.7922  0.6212</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   1.5315  0.1404  4.2638  1.3239</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3,4} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
@@ -294,17 +294,17 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $optimizer$state$`2`$exp_avg</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0467</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1375</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0908</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0748</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0581</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1329</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $optimizer$state$`2`$exp_avg_sq</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> 0.001 *</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5587</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3208</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.0020</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8718</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7947</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3835</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{3} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
@@ -336,7 +336,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> Empty data.table (0 rows and 3 cols): stage,class,msg</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $train_time</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> [1] 0.067</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> [1] 0.063</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $param_vals</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $param_vals$epochs</span>
@@ -383,7 +383,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> NULL</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $mlr3_version</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> [1] ‘0.22.0’</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> [1] ‘0.22.1’</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $internal_tuned_values</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> named list()</span>
diff --git a/dev/reference/mlr_pipeops_torch_model_regr.html b/dev/reference/mlr_pipeops_torch_model_regr.html
index 09216734..dc8d0966 100644
--- a/dev/reference/mlr_pipeops_torch_model_regr.html
+++ b/dev/reference/mlr_pipeops_torch_model_regr.html
@@ -274,16 +274,14 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $optimizer$state$`1`$exp_avg</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 1 to 8  -1.5397    5.2623    7.8226  593.2802   -3.0858   -2.8836  328.6158  -12.0903</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> </span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 9 to 10  -2.3402    3.6805</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -6.4192 -38.0977 -76.4209 -2787.1460 -45.3608 -48.1255 -1960.0070 -215.3138 -4.9757 -38.1129</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{1,10} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $optimizer$state$`1`$exp_avg_sq</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 1 to 6 1.5250e-01  1.7056e+00  3.5828e+00  1.9482e+04  7.4272e-01  6.6760e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 1 to 6 2.5316e+00  8.1738e+01  3.2276e+02  4.2917e+05  1.1383e+02  1.2935e+02</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 7 to 10 6.1135e+03  9.6809e+00  3.1214e-01  7.4889e-01</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 7 to 10 2.1708e+05  2.5621e+03  1.4196e+00  8.0366e+01</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{1,10} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
@@ -293,13 +291,12 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $optimizer$state$`2`$exp_avg</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3394</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -12.3812</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{1} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $optimizer$state$`2`$exp_avg_sq</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> 0.01 *</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1906</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  8.4702</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{1} ]</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
@@ -336,7 +333,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> Empty data.table (0 rows and 3 cols): stage,class,msg</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $train_time</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> [1] 0.042</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> [1] 0.041</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $param_vals</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $param_vals$epochs</span>
@@ -383,7 +380,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-out co"><span class="r-pr">#&gt;</span> NULL</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $mlr3_version</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> [1] ‘0.22.0’</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> [1] ‘0.22.1’</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> </span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> $internal_tuned_values</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> named list()</span>
diff --git a/dev/reference/nn_graph.html b/dev/reference/nn_graph.html
index 14c515a8..f5586236 100644
--- a/dev/reference/nn_graph.html
+++ b/dev/reference/nn_graph.html
@@ -119,22 +119,22 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-in"><span></span></span>
 <span class="r-in"><span><span class="fu">network</span><span class="op">(</span>module_1.input <span class="op">=</span> <span class="va">x</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> torch_tensor</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0771</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5067</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5069</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2827</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2935</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2439</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1360</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1873</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1472</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2764</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2794</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0572</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1514</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1721</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5291</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3046</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2580</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1551</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2086</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1070</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0621</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3388</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3877</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0435</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2469</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1769</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3112</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1107</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1484</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0044</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3122</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3985</span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> [ CPUFloatType{16,1} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span>
 </code></pre></div>
     </div>
diff --git a/dev/search.json b/dev/search.json
index b26272c6..af7e0aae 100644
--- a/dev/search.json
+++ b/dev/search.json
@@ -1 +1 @@
-[{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":null,"dir":"","previous_headings":"","what":"GNU Lesser General Public License","title":"GNU Lesser General Public License","text":"Version 3, 29 June 2007Copyright © 2007 Free Software Foundation, Inc. <http://fsf.org/> Everyone permitted copy distribute verbatim copies license document, changing allowed. version GNU Lesser General Public License incorporates terms conditions version 3 GNU General Public License, supplemented additional permissions listed .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_0-additional-definitions","dir":"","previous_headings":"","what":"0. Additional Definitions","title":"GNU Lesser General Public License","text":"used herein, “License” refers version 3 GNU Lesser General Public License, “GNU GPL” refers version 3 GNU General Public License. “Library” refers covered work governed License, Application Combined Work defined . “Application” work makes use interface provided Library, otherwise based Library. Defining subclass class defined Library deemed mode using interface provided Library. “Combined Work” work produced combining linking Application Library. particular version Library Combined Work made also called “Linked Version”. “Minimal Corresponding Source” Combined Work means Corresponding Source Combined Work, excluding source code portions Combined Work , considered isolation, based Application, Linked Version. “Corresponding Application Code” Combined Work means object code /source code Application, including data utility programs needed reproducing Combined Work Application, excluding System Libraries Combined Work.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_1-exception-to-section-3-of-the-gnu-gpl","dir":"","previous_headings":"","what":"1. Exception to Section 3 of the GNU GPL","title":"GNU Lesser General Public License","text":"may convey covered work sections 3 4 License without bound section 3 GNU GPL.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_2-conveying-modified-versions","dir":"","previous_headings":"","what":"2. Conveying Modified Versions","title":"GNU Lesser General Public License","text":"modify copy Library, , modifications, facility refers function data supplied Application uses facility (argument passed facility invoked), may convey copy modified version: ) License, provided make good faith effort ensure , event Application supply function data, facility still operates, performs whatever part purpose remains meaningful, b) GNU GPL, none additional permissions License applicable copy.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_3-object-code-incorporating-material-from-library-header-files","dir":"","previous_headings":"","what":"3. Object Code Incorporating Material from Library Header Files","title":"GNU Lesser General Public License","text":"object code form Application may incorporate material header file part Library. may convey object code terms choice, provided , incorporated material limited numerical parameters, data structure layouts accessors, small macros, inline functions templates (ten fewer lines length), following: ) Give prominent notice copy object code Library used Library use covered License. b) Accompany object code copy GNU GPL license document.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_4-combined-works","dir":"","previous_headings":"","what":"4. Combined Works","title":"GNU Lesser General Public License","text":"may convey Combined Work terms choice , taken together, effectively restrict modification portions Library contained Combined Work reverse engineering debugging modifications, also following: ) Give prominent notice copy Combined Work Library used Library use covered License. b) Accompany Combined Work copy GNU GPL license document. c) Combined Work displays copyright notices execution, include copyright notice Library among notices, well reference directing user copies GNU GPL license document. d) one following: 0) Convey Minimal Corresponding Source terms License, Corresponding Application Code form suitable , terms permit, user recombine relink Application modified version Linked Version produce modified Combined Work, manner specified section 6 GNU GPL conveying Corresponding Source. 1) Use suitable shared library mechanism linking Library. suitable mechanism one () uses run time copy Library already present user’s computer system, (b) operate properly modified version Library interface-compatible Linked Version. e) Provide Installation Information, otherwise required provide information section 6 GNU GPL, extent information necessary install execute modified version Combined Work produced recombining relinking Application modified version Linked Version. (use option 4d0, Installation Information must accompany Minimal Corresponding Source Corresponding Application Code. use option 4d1, must provide Installation Information manner specified section 6 GNU GPL conveying Corresponding Source.)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_5-combined-libraries","dir":"","previous_headings":"","what":"5. Combined Libraries","title":"GNU Lesser General Public License","text":"may place library facilities work based Library side side single library together library facilities Applications covered License, convey combined library terms choice, following: ) Accompany combined library copy work based Library, uncombined library facilities, conveyed terms License. b) Give prominent notice combined library part work based Library, explaining find accompanying uncombined form work.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_6-revised-versions-of-the-gnu-lesser-general-public-license","dir":"","previous_headings":"","what":"6. Revised Versions of the GNU Lesser General Public License","title":"GNU Lesser General Public License","text":"Free Software Foundation may publish revised /new versions GNU Lesser General Public License time time. new versions similar spirit present version, may differ detail address new problems concerns. version given distinguishing version number. Library received specifies certain numbered version GNU Lesser General Public License “later version” applies , option following terms conditions either published version later version published Free Software Foundation. Library received specify version number GNU Lesser General Public License, may choose version GNU Lesser General Public License ever published Free Software Foundation. Library received specifies proxy can decide whether future versions GNU Lesser General Public License shall apply, proxy’s public statement acceptance version permanent authorization choose version Library.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/articles/callbacks.html","id":"building-blocks","dir":"Articles","previous_headings":"","what":"Building Blocks","title":"Custom Callbacks","text":"heart callback mechanism three R6 classes: CallbackSet contains methods executed different stages training loop TorchCallback wraps CallbackSet annotates meta information, including ParamSet ContextTorch defines information training process CallbackSet access using predefined callbacks, one usually interacts TorchCallback class, constructed using t_clbk(<id>): wrapped CallbackSet accessible via field $generator, can create new instance calling $new() Within training loop torch model, different stages exist callback can execute code. , list stages available, also described documentation CallbackSet. history callback, see runs code beginning, validation starts, end epoch.","code":"tc_hist = t_clbk(\"history\") tc_hist ## <TorchCallback:history> History ## * Generator: CallbackSetHistory ## * Parameters: list() ## * Packages: mlr3torch,torch cbs_hist = tc_hist$generator$new() mlr_reflections$torch$callback_stages ##  [1] \"on_begin\"             \"on_epoch_begin\"       \"on_batch_begin\"       ##  [4] \"on_after_backward\"    \"on_batch_end\"         \"on_before_valid\"      ##  [7] \"on_batch_valid_begin\" \"on_batch_valid_end\"   \"on_valid_end\"         ## [10] \"on_epoch_end\"         \"on_end\"               \"on_exit\" cbs_hist ## <CallbackSetHistory> ## * Stages: on_begin, on_before_valid, on_epoch_end"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/callbacks.html","id":"writing-a-custom-logger","dir":"Articles","previous_headings":"","what":"Writing a Custom Logger","title":"Custom Callbacks","text":"order define custom callback, can use helper torch_callback() helper function. example, create custom logger keeps track exponential moving average train loss prints end every epoch. callback takes one argument alpha smoothing parameter store moving average self$moving_loss. value alpha can later configured Learner. , implement main logic callback using two stages: on_batch_end(), called optimizer updates network parameters using mini-batch. , access last loss via ContextTorch’s (accessible via self$ctx) $last_loss field update value self$moving_loss. on_before_valid(), run validation loop. point simply print exponential moving average training loss. Finally, order make final value moving average accessible Learner training finishes, implement state_dict() method return value load_state_dict(state_dict) method, takes previously retrieved state dict sets callback. Note access final $moving_loss via learner, otherwise implement methods. created TorchCallback object associated CallbackSet R6-class us:","code":"custom_logger = torch_callback(\"custom_logger\",   initialize = function(alpha = 0.1) {     self$alpha = alpha     self$moving_loss = NULL   },   on_batch_end = function() {     if (is.null(self$moving_training_loss)) {       self$moving_loss = self$ctx$last_loss     } else {       self$moving_loss = self$alpha * self$last_loss + (1 - self$alpha) * self$moving_loss     }   },   on_before_valid = function() {     cat(sprintf(\"Epoch %s: %.2f\\n\", self$ctx$epoch, self$moving_loss))   },   state_dict = function() {     self$moving_loss   },   load_state_dict = function(state_dict) {     self$moving_loss = state_dict   } ) custom_logger ## <TorchCallback:custom_logger> Custom_logger ## * Generator: CallbackSetCustom_logger ## * Parameters: list() ## * Packages: mlr3torch,torch custom_logger$generator ## <CallbackSetCustom_logger> object generator ##   Inherits from: <inherit> ##   Public: ##     initialize: function (alpha = 0.1)  ##     state_dict: function ()  ##     load_state_dict: function (state_dict)  ##     on_before_valid: function ()  ##     on_batch_end: function ()  ##   Parent env: <environment: 0x5603df3b07c8> ##   Locked objects: FALSE ##   Locked class: FALSE ##   Portable: TRUE"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/callbacks.html","id":"using-the-custom-logger","dir":"Articles","previous_headings":"","what":"Using the Custom Logger","title":"Custom Callbacks","text":"order showcase effects, create train simple multi layer perceptron pass callback. set smoothing parameter alpha 0.05. information returnede state_dict() now accessible via Learner’s $model-slot:","code":"task = tsk(\"iris\") mlp = lrn(\"classif.mlp\",   callbacks = custom_logger, cb.custom_logger.alpha = 0.05,   epochs = 5, batch_size = 64, neurons = 20)  mlp$train(task) ## Epoch 1: 1.22 ## Epoch 2: 1.27 ## Epoch 3: 1.48 ## Epoch 4: 1.07 ## Epoch 5: 1.09 mlp$model$callbacks$custom_logger ## [1] 1.090038"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"quickstart","dir":"Articles","previous_headings":"","what":"Quickstart","title":"Get Started","text":"vignette show get started mlr3torch training simple neural network tabular regression problem. assume familiar mlr3 framework, see e.g. mlr3 book. first example, train simple multi-layer perceptron (MLP) well-known “mtcars” task, goal predict miles per galleon (‘mpg’) cars. architecture comes predfined learner mlr3torch, can also easily create new network architectures, see Neural Networks Graphs vignette detailed introduoduion. first set seed reproducibility, load library construct task. Learners mlr3torch work similary mlr3 learners. , construct simple multi layer perceptron regression. usual calling lrn() configuring parameters: use two hidden layers 50 neurons, training, set batch size 32, number training epochs 30 device \"cpu\". complete description available parameters see ?mlr3torch::LearnerTorchMLP. can use learner training prediction just like regression learner. , split observations training test set, train learner training set create predictions test set. Finally, compute mean squared error predictions.","code":"set.seed(314) library(mlr3torch) task = tsk(\"mtcars\") task$head() #>      mpg    am  carb   cyl  disp  drat  gear    hp  qsec    vs    wt #>    <num> <num> <num> <num> <num> <num> <num> <num> <num> <num> <num> #> 1:  21.0     1     4     6   160  3.90     4   110 16.46     0 2.620 #> 2:  21.0     1     4     6   160  3.90     4   110 17.02     0 2.875 #> 3:  22.8     1     1     4   108  3.85     4    93 18.61     1 2.320 #> 4:  21.4     0     1     6   258  3.08     3   110 19.44     1 3.215 #> 5:  18.7     0     2     8   360  3.15     3   175 17.02     0 3.440 #> 6:  18.1     0     1     6   225  2.76     3   105 20.22     1 3.460 mlp = lrn(\"regr.mlp\",   # architecture parameters   neurons = c(50, 50),   # training arguments   batch_size = 32, epochs = 30, device = \"cpu\" ) mlp #> <LearnerTorchMLP[regr]:regr.mlp>: My Little Powny #> * Model: - #> * Parameters: epochs=30, device=cpu, num_threads=1, #>   num_interop_threads=1, seed=random, eval_freq=1, #>   measures_train=<list>, measures_valid=<list>, patience=0, #>   min_delta=0, batch_size=32, neurons=50,50, p=0.5, #>   activation=<nn_relu>, activation_args=<list> #> * Validate: NULL #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response] #> * Feature Types: integer, numeric, lazy_tensor #> * Properties: internal_tuning, marshal, validation #> * Optimizer: adam #> * Loss: mse #> * Callbacks: - # Split the obersevations into training and test set splits = partition(task) # Train the learner on the train set mlp$train(task, row_ids = splits$train) # Predict the test set prediction = mlp$predict(task, row_ids = splits$test) # Compute the mse prediction$score(msr(\"regr.mse\")) #> regr.mse  #>  283.838"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"configuring-a-learner","dir":"Articles","previous_headings":"","what":"Configuring a Learner","title":"Get Started","text":"Although torch learners quite like mlr3 learners, differences. One LearnerTorch classes construction arguments, .e. torch learners modular learners. learners free implement construction arguments, common torch learners, namely loss, optimizer callbacks. object can parameters included LearnerTorch’s parameter set. previous example, specify explicitly used default values, Adam optimizer, MSE loss callbacks. now show configure three aspects learner mlr3torch::TorchOptimizer, mlr3torch::TorchLoss, mlr3torch::TorchCallback classes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"loss","dir":"Articles","previous_headings":"Configuring a Learner","what":"Loss","title":"Get Started","text":"loss function, also known objective function cost function, measures discrepancy predicted output true output. quantifies well model performing training. R package torch, underpins mlr3torch framework, already provides number predefined loss functions Mean Squared Error (nn_mse_loss), Mean Absolute Error (nn_l1_loss), cross entropy loss (nn_cross_entropy_loss). mlr3torch, represent loss functions using mlr3torch::TorchLoss class. provides thin wrapper around torch loss functions annotates meta information, importantly paradox::ParamSet allows configure loss function. object can constructed using t_loss(<key>). , construct L1 loss function, also known Mean Absolute Error (MAE). printed output informs us wrapped loss function (nn_l1_loss), configured parameters, packages depends task types can used. ParamSet contains one parameter, namely reduction, specifies loss reduced batch. wrapped loss module generator accessible slot $generator. can pass TorchLoss argument loss initialization learner. parameters loss added learner’s ParamSet, prefixed \"loss.\". predefined loss functions stored mlr3torch_losses dictionary, can retrieved using t_loss(<key>).","code":"l1 = t_loss(\"l1\") l1 #> <TorchLoss:l1> Absolute Error #> * Generator: nn_l1_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr # the paradox::ParamSet of the loss l1$param_set #> <ParamSet(1)> #>           id    class lower upper nlevels default  value #>       <char>   <char> <num> <num>   <num>  <list> <list> #> 1: reduction ParamFct    NA    NA       2    mean [NULL] l1$generator #> <nn_l1_loss> object generator #>   Inherits from: <inherit> #>   Public: #>     .classes: nn_l1_loss nn_loss nn_module #>     initialize: function (reduction = \"mean\")  #>     forward: function (input, target)  #>     clone: function (deep = FALSE, ..., replace_values = TRUE)  #>   Private: #>     .__clone_r6__: function (deep = FALSE)  #>   Parent env: <environment: 0x55a43f05f830> #>   Locked objects: FALSE #>   Locked class: FALSE #>   Portable: TRUE mlp_l1 = lrn(\"regr.mlp\", loss = l1) mlp_l1$param_set$values$loss.reduction #> NULL mlr3torch_losses #> <DictionaryMlr3torchLosses> with 3 stored values #> Keys: cross_entropy, l1, mse"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"optimizer","dir":"Articles","previous_headings":"Configuring a Learner","what":"Optimizer","title":"Get Started","text":"optimizer determines model’s weights updated based calculated loss. adjusts parameters model minimize loss function, optimizing model’s performance. Optimizers work analogous loss functions, .e. mlr3torch provides thin wrapper – TorchOptimizer class – around optimizers Adam (optim_adam) SGD (optim_sgd). TorchLoss objects can constructed using t_opt(<key>). optimizers, associated ParamSet interesting see : wrapped torch optimizer can accessed slot generator. Parameters TorchOptimizer (also TorchLoss TorchCallback) can set usual mlr3 way, .e. either construction, afterwards using $set_values() method parameter set. see optimizer’s parameters added learner’s ParamSet (prefixed \"opt.\") values set values specified. exposing optimizer’s parameters, can conveniently tuned using mlr3tuning. available optimizers stored mlr3torch_optimizers dictionary.","code":"sgd = t_opt(\"sgd\") sgd #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_sgd #> * Parameters: list() #> * Packages: torch,mlr3torch  sgd$param_set #> <ParamSet(5)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1:           lr ParamDbl     0   Inf     Inf <NoDefault[0]> [NULL] #> 2:     momentum ParamDbl     0     1     Inf              0 [NULL] #> 3:    dampening ParamDbl     0     1     Inf              0 [NULL] #> 4: weight_decay ParamDbl     0     1     Inf              0 [NULL] #> 5:     nesterov ParamLgl    NA    NA       2          FALSE [NULL] sgd$param_set$set_values(   lr = 0.5, # increase learning rate   nesterov = FALSE # no nesterov momentum ) mlp_sgd = lrn(\"regr.mlp\", optimizer = sgd) as.data.table(mlp_sgd$param_set)[   startsWith(id, \"opt.\")][[1L]] #> [1] \"opt.lr\"           \"opt.momentum\"     \"opt.dampening\"    \"opt.weight_decay\" #> [5] \"opt.nesterov\" mlp_sgd$param_set$values[c(\"opt.lr\", \"opt.nesterov\")] #> $opt.lr #> [1] 0.5 #>  #> $opt.nesterov #> [1] FALSE mlr3torch_optimizers #> <DictionaryMlr3torchOptimizers> with 7 stored values #> Keys: adadelta, adagrad, adam, asgd, rmsprop, rprop, sgd"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"callbacks","dir":"Articles","previous_headings":"Configuring a Learner","what":"Callbacks","title":"Get Started","text":"third important configuration option callbacks allow customize training process. allows saving model checkpoints, logging metrics, implementing custom functionality specific training scenarios. tutorial implement custom callback, see Custom Callbacks vignette. , show use predefined callbacks. , retrieve \"history\" callback using t_clbk(), parameters merely saves training validation history learner can accessed afterwards. wanted learn callback , can access help page wrapped object using $help() method. Note also possible loss optimizer. predefined callbacks stored mlr3torch_callbacks dictionary.","code":"history = t_clbk(\"history\") history #> <TorchCallback:history> History #> * Generator: CallbackSetHistory #> * Parameters: list() #> * Packages: mlr3torch,torch history$help() mlr3torch_callbacks #> <DictionaryMlr3torchCallbacks> with 4 stored values #> Keys: checkpoint, history, progress, tb"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"putting-it-together","dir":"Articles","previous_headings":"Configuring a Learner","what":"Putting it Together","title":"Get Started","text":"now define customized MLP learner using loss, optimizer callback just covered. keep track performance, use 30% training data validation evaluate using MAE Measure. Note mearures_valid measures_train parameters LearnerTorch take common mlr3::Measures, whereas loss function must TorchLoss. now train learner “mtcars” task use train-test split . make predictions unseen test data compare scores. directly optimized L1 (aka MAE) loss tweaked learning rate, configured mlp_custom learner lower MAE default mlp learner. configured learner use history callback, can find validation history $model slot: plot shows epochs 6 30.  important information stored Learner’s model $network, underlying nn_module. full description model, see ?LearnerTorch.","code":"mlp_custom = lrn(\"regr.mlp\",   # construction arguments   optimizer = sgd, loss = l1, callbacks = history,   # scores to keep track of   measures_valid = msr(\"regr.mae\"),   # other parameters are left as-is:   # architecture   neurons = c(50, 50),   # training arguments   batch_size = 32, epochs = 30, device = \"cpu\",   # validation proportion   validate = 0.3 )  mlp_custom #> <LearnerTorchMLP[regr]:regr.mlp>: My Little Powny #> * Model: - #> * Parameters: epochs=30, device=cpu, num_threads=1, #>   num_interop_threads=1, seed=random, eval_freq=1, #>   measures_train=<list>, measures_valid=<MeasureRegrSimple>, #>   patience=0, min_delta=0, batch_size=32, neurons=50,50, p=0.5, #>   activation=<nn_relu>, activation_args=<list>, opt.lr=0.5, #>   opt.nesterov=FALSE #> * Validate: 0.3 #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response] #> * Feature Types: integer, numeric, lazy_tensor #> * Properties: internal_tuning, marshal, validation #> * Optimizer: sgd #> * Loss: l1 #> * Callbacks: history mlp_custom$train(task, row_ids = splits$train) prediction_custom = mlp_custom$predict(task, row_ids = splits$test) prediction_custom$score(msr(\"regr.mae\")) #> regr.mae  #> 7.122375 prediction$score(msr(\"regr.mae\")) #> regr.mae  #> 15.27983 head(mlp_custom$model$callbacks$history$valid) #>    epoch     regr.mae #>    <num>        <num> #> 1:     1 1.777395e+04 #> 2:     2 3.955504e+08 #> 3:     3 1.143863e+04 #> 4:     4 1.792927e+01 #> 5:     5 1.759594e+01 #> 6:     6 1.726260e+01"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"a-torch-primer","dir":"Articles","previous_headings":"","what":"A torch Primer","title":"Internals","text":"start sampling input tensor: 2 batches 3 features: nn_module constructed nn_module_generator. nn_linear one simpler generators: Applying module gives 2-batch 4 units: neural network one (4-unit) hidden layer two outputs needs following ingredients can pipe tensor layers follows. now continue showing neural network can represented mlr3torch.","code":"input = torch_randn(2, 3) input #> torch_tensor #> -0.6687 -2.4004 -1.4578 #>  0.9376 -0.9994 -0.3993 #> [ CPUFloatType{2,3} ] module_1 = nn_linear(in_features = 3, out_features = 4, bias = TRUE) output = module_1(input) output #> torch_tensor #>  0.0980  0.3119 -1.4607 -0.4350 #>  0.2679  0.8104  0.1816  0.0485 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] activation = nn_sigmoid() module_2 = nn_linear(4, 3, bias = TRUE) softmax = nn_softmax(2) output = module_1(input) output = activation(output) output = module_2(output) output = softmax(output) output #> torch_tensor #>  0.2918  0.1502  0.5579 #>  0.2890  0.1217  0.5893 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"neural-networks-as-graphs","dir":"Articles","previous_headings":"","what":"Neural Networks as Graphs","title":"Internals","text":"mlr3torch, nn_modules wrapped PipeOpModule. advantage network structure can represented mlr3pipelines::Graph object made explicit (can plotted, can extended manipulated), compared e.g. writing function pipes input series modules. PipeOpModule can used wrap module directly, usually constructed PipeOpTorch (see later). typically single input single output, although multiple inputs possible (module called multiple arguments), multiple outputs possible module-function returns list. input output channels must explicitly declared construction. now continue recreate network using PipeOpModules. can wrap linear module_1 layers like : Note po(\"module_1\") equivalent po(\"module\", id = \"module_1\"). mechanism convenient avoid ID clashes graphs contain PipeOp multiple times. can use generated PipeOp familiar way: Note use $train(), since torch modules anything maps state (filled empty list). single hidden layer neural network can constructed Graph, can training . can now use graph’s $train() method pipe tensor whole Graph. object allows easily perform forward pass, inherit nn_module, useful various reasons. Instead class inherits nn_module Graph (work R6, since multiple inheritance available), class inherits nn_module contains Graph member slot composition. class nn_graph. constructed Graph, well information shape(s) torch_tensor(s) expects inputs. Shape info communicated integer-valued numeric vector; dimensions arbitrary, e.g. batch-size, given NA. network expects input shape c(NA, 3), since first layer created nn_linear(in_features = 3, ...). Graph multiple outputs, also possible select subset outputs use, change output order, giving output_map argument. module gives us convenience torch nn_module objects, e.g.: can used transform tensors just torch::nn_module:","code":"library(mlr3torch) po_module_1 = po(\"module_1\", module = module_1) output = po_module_1$train(list(input))[[1]] output #> torch_tensor #>  0.0980  0.3119 -1.4607 -0.4350 #>  0.2679  0.8104  0.1816  0.0485 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] po_activation = po(\"module\", id = \"activation\", activation) po_module_2 = po(\"module_2\", module = module_2) po_softmax = po(\"module\", id = \"softmax\", module = softmax)  module_graph = po_module_1 %>>%   po_activation %>>%   po_module_2 %>>%   po_softmax  module_graph$plot(html = TRUE) output = module_graph$train(input)[[1]] output #> torch_tensor #>  0.2918  0.1502  0.5579 #>  0.2890  0.1217  0.5893 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ] # the name of the single input is: module_graph$input #>              name        train predict    op.id channel.name #>            <char>       <char>  <char>   <char>       <char> #> 1: module_1.input torch_tensor    NULL module_1        input  graph_module = nn_graph(   module_graph,   shapes_in = list(module_1.input = c(NA, 3)) ) graph_module$children #> $module_list #> An `nn_module` containing 31 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • 0: <nn_linear> #16 parameters #> • 1: <nn_sigmoid> #0 parameters #> • 2: <nn_linear> #15 parameters #> • 3: <nn_softmax> #0 parameters graph_module(input) #> torch_tensor #>  0.2918  0.1502  0.5579 #>  0.2890  0.1217  0.5893 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"modeldescriptor","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"ModelDescriptor","title":"Internals","text":"PipeOpModule represents nn_module fixed specific tensor shape hyperparameters. constructing neural network using operators, one take care output shape operations match input shapes following operations. complete Graph matching PipeOpModules can constructed using operators mostly inherit PipeOpTorch, making use ModelDescriptor class. ModelDescriptor class contains Graph (mostly) PipeOpModules information. PipeOpTorch transforms ModelDescriptor adds PipeOpModules Graph. ModelDescriptors always build Graph specific Task. easiest way initialize proper ModelDescriptor use appropriate PipeOpTorchIngress given datatype. use PipeOpTorchIngressNumeric, used numeric data. ModelDescriptor S3 object contains Graph, information generate data ($ingress $task), tags build model unrelated architecture ($optimizer, $loss $callbacks) well information necessary extend graph along given output ($pointer $pointer_shape). $pointer identifies output $graph PipeOpTorch extend. Piping ModelDescriptor PipeOpTorchLinear, example, adds PipeOpModule wrapping torch::nn_linear. $pointer now updated identify output PipeOpModule, $pointer_shape shows shape changed 4 units (3 input ). model_descriptor_to_module() function converts nn_graph, functional torch::nn_module.","code":"task = tsk(\"iris\")$select(colnames(iris)[1:3])  po_torch_in = po(\"torch_ingress_num\") md = po_torch_in$train(list(task))[[1]]  md #> <ModelDescriptor: 1 ops> #> * Ingress:  torch_ingress_num.input: [(NA,3)] #> * Task:  iris [classif] #> * Callbacks:  N/A #> * Optimizer:  N/A #> * Loss:  N/A #> * pointer:  torch_ingress_num.output [(NA,3)] unclass(md) #> $graph #> Graph with 1 PipeOps: #>                 ID         State sccssors prdcssors #>             <char>        <char>   <char>    <char> #>  torch_ingress_num <<UNTRAINED>>                    #>  #> $ingress #> $ingress$torch_ingress_num.input #> Ingress: Task[Petal.Length,Sepal.Length,Sepal.Width] --> Tensor(NA, 3) #>  #>  #> $task #> <TaskClassif:iris> (150 x 4): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (3): #>   - dbl (3): Petal.Length, Sepal.Length, Sepal.Width #>  #> $optimizer #> NULL #>  #> $loss #> NULL #>  #> $callbacks #> named list() #>  #> $pointer #> [1] \"torch_ingress_num\" \"output\"            #>  #> $pointer_shape #> [1] NA  3 po_torch_linear = po(\"nn_linear\", out_features = 4) md = po_torch_linear$train(list(md))[[1]]  md$graph #> Graph with 2 PipeOps: #>                 ID         State  sccssors         prdcssors #>             <char>        <char>    <char>            <char> #>  torch_ingress_num <<UNTRAINED>> nn_linear                   #>          nn_linear <<UNTRAINED>>           torch_ingress_num md$pointer #> [1] \"nn_linear\" \"output\" md$pointer_shape #> [1] NA  4 small_module = model_descriptor_to_module(md, list(md$pointer))  small_module(input) #> torch_tensor #>  1.2548 -1.3432 -1.5458  0.7297 #>  0.4231 -0.7686 -0.7230  0.6464 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"using-modeldescriptor-to-get-data","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"Using ModelDescriptor to get Data","title":"Internals","text":"ModelDescriptor represent Graph nn_module created, also way Task processed get input batches. torch::dataset can created calling task_dataset(); task feature_ingress_tokens arguments can retrieved ModelDescriptor. target_batchgetter needs created extra (necessary), since depends ultimate machine learning model, looked far. Use $.getbatch() method get batch can given nn_module. Note $x $y slot, latter used, account possible target batches. $x slot also list, since able handle NNs multiple inputs (see ).","code":"td = task_dataset(   task = md$task,   feature_ingress_tokens = md$ingress,   device = \"cpu\" )  td #> <dataset> #>   Public: #>     .getbatch: function (index)  #>     .getitem: function (index)  #>     .length: function ()  #>     all_features: Petal.Length Sepal.Length Sepal.Width Species #>     cache_lazy_tensors: FALSE #>     clone: function (deep = FALSE)  #>     device: cpu #>     feature_ingress_tokens: list #>     initialize: function (task, feature_ingress_tokens, target_batchgetter = NULL,  #>     load_state_dict: function (x, ..., .refer_to_state_dict = FALSE)  #>     state_dict: function ()  #>     target_batchgetter: NULL #>     task: TaskClassif, TaskSupervised, Task, R6 batch = td$.getbatch(1:3) batch #> $x #> $x$torch_ingress_num.input #> torch_tensor #>  1.4000  5.1000  3.5000 #>  1.4000  4.9000  3.0000 #>  1.3000  4.7000  3.2000 #> [ CPUFloatType{3,3} ] #>  #>  #> $.index #> torch_tensor #>  1 #>  2 #>  3 #> [ CPULongType{3} ]  small_module(batch$x[[1]]) #> torch_tensor #> -0.9548  2.4764  1.7547 -0.4855 #> -1.1006  2.1895  1.5184 -0.5936 #> -0.8588  2.2540  1.5650 -0.4339 #> [ CPUFloatType{3,4} ][ grad_fn = <AddmmBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"building-sequential-nns","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"Building sequential NNs","title":"Internals","text":"sequential NN can easily implemented follows: Note second nn_linear need informed output dimension first nn_linear, since ModelDescriptor passed along Graph edges knows info ($pointer_shape slot).","code":"graph_generator = po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 4, id = \"linear1\") %>>%     po(\"nn_sigmoid\") %>>%   po(\"nn_linear\", out_features = 3, id = \"linear2\") %>>%   po(\"nn_softmax\", dim = 2) md_sequential = graph_generator$train(task)[[1]]  graph_module = model_descriptor_to_module(md_sequential, list(md_sequential$pointer))  graph_module(input) #> torch_tensor #>  0.2554  0.3238  0.4209 #>  0.2483  0.3749  0.3768 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"building-more-interesting-nns","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"Building more interesting NNs","title":"Internals","text":"One main features mlr3pipelines ability easily represent computational Graphs. ModelDescriptor / PipeOpTorch setup built make full use functionality. possible multiple inputs NN using multiple PipeOpTorchIngress inputs, possible parallel alternative path branching, possible multiple outputs. Consider following (bit nonsensical) network operates differently \"Petal\" \"Sepal\" features tsk(\"iris\") manually split task , shown wholly integrated mlr3pipelines pipeline can automatically. can use create neural network iris tasks created . set $keep_results debug flag can inspection happening: make multiple observations : can observe ModelDescriptor grows passed along edges graph_iris. Note $graph slot ModelDescriptor often updated -reference, time inspect intermediate results, may contain complete graph. However, see $ingress, $pointer $pointer_shape ModelDescriptors take sepal.-path differ ones take petal.-path: po(\"nn_merge_cat\") unites two ModelDescriptors contains common ingress. pointer_shape now reflects output “cat”-operation: 2nd dimension added : Multiple ModelDescriptors created, since graph_iris multiple outpus. makes possible create neural network multiple outputs. need unite outputs graph_iris using model_descriptor_union() can pass model_descriptor_to_module(). need collect output_pointers separately. parameter list_output must set TRUE since module multiple outputs. PipeOpBranch disappears resulting Graph PipeOpModule iris_module. PipeOpTorchs graph_iris add anything ModelDescriptors. branch interpeted graph_iris runs, nn_relu path followed. iris_module therefore contains Graph “relu” activation: ModelDescriptor’s $task slot contains Task features used create input data NN inputs. can given task_dataset(), along $ingress, create torch dataset creates batches. , output graph_iris can used: resulting module multiple inputs multiple outputs. call first two rows iris, set debug $keep_results flag can inspect happening nn_module’s $graph: first linear layer takes “Sepal” input (\"linear1\") creates 2x4 tensor (batch size 2, 4 units), \"linear3\" layer 2x5 output: observe po(\"nn_merge_cat\") concatenates , expected:","code":"iris_petal = tsk(\"iris\")$select(c(\"Petal.Length\", \"Petal.Width\")) iris_sepal = tsk(\"iris\")$select(c(\"Sepal.Length\", \"Sepal.Width\")) graph_sepal = po(\"torch_ingress_num\", id = \"sepal.in\") %>>%   po(\"nn_linear\", out_features = 4, id = \"linear1\")  graph_petal = po(\"torch_ingress_num\", id = \"petal.in\") %>>%   po(\"nn_linear\", out_features = 3, id = \"linear2\") %>>%   po(\"nn_tanh\") %>>%   po(\"nn_linear\", out_features = 5, id = \"linear3\")  graph_common = ppl(\"branch\", graphs = list(     sigmoid = po(\"nn_sigmoid\"),     relu = po(\"nn_relu\")   )) %>>%   gunion(list(     po(\"nn_linear\", out_features = 1, id = \"lin_out\"),     po(\"nn_linear\", out_features = 3, id = \"cat_out\") %>>%       po(\"nn_softmax\", dim = 2)   ))   graph_iris = gunion(list(graph_sepal, graph_petal)) %>>%   po(\"nn_merge_cat\") %>>%   graph_common  graph_iris$plot(html = TRUE) graph_iris$param_set$values$branch.selection = \"relu\"  graph_iris$keep_results = TRUE  iris_mds = graph_iris$train(   input = list(sepal.in.input = iris_sepal, petal.in.input = iris_petal),   single_input = FALSE )  iris_mds #> $lin_out.output #> <ModelDescriptor: 11 ops> #> * Ingress:  sepal.in.input: [(NA,2)], petal.in.input: [(NA,2)] #> * Task:  iris [classif] #> * Callbacks:  N/A #> * Optimizer:  N/A #> * Loss:  N/A #> * pointer:  lin_out.output [(NA,1)] #>  #> $nn_softmax.output #> <ModelDescriptor: 11 ops> #> * Ingress:  sepal.in.input: [(NA,2)], petal.in.input: [(NA,2)] #> * Task:  iris [classif] #> * Callbacks:  N/A #> * Optimizer:  N/A #> * Loss:  N/A #> * pointer:  nn_softmax.output [(NA,3)] # sepal.in path graph_iris$pipeops$linear1$.result[[1]]$ingress #> $sepal.in.input #> Ingress: Task[Sepal.Length,Sepal.Width] --> Tensor(NA, 2) graph_iris$pipeops$linear1$.result[[1]]$pointer #> [1] \"linear1\" \"output\" graph_iris$pipeops$linear1$.result[[1]]$pointer_shape #> [1] NA  4  # petal.in path graph_iris$pipeops$linear3$.result[[1]]$ingress #> $petal.in.input #> Ingress: Task[Petal.Length,Petal.Width] --> Tensor(NA, 2) graph_iris$pipeops$linear3$.result[[1]]$pointer #> [1] \"linear3\" \"output\" graph_iris$pipeops$linear3$.result[[1]]$pointer_shape #> [1] NA  5 graph_iris$pipeops$nn_merge_cat$.result[[1]]$ingress #> $sepal.in.input #> Ingress: Task[Sepal.Length,Sepal.Width] --> Tensor(NA, 2) #>  #> $petal.in.input #> Ingress: Task[Petal.Length,Petal.Width] --> Tensor(NA, 2) graph_iris$pipeops$nn_merge_cat$.result[[1]]$pointer_shape #> [1] NA  9 iris_mds_union = model_descriptor_union(iris_mds[[1]], iris_mds[[2]]) output_pointers = list(iris_mds[[1]]$pointer, iris_mds[[2]]$pointer) output_pointers #> [[1]] #> [1] \"lin_out\" \"output\"  #>  #> [[2]] #> [1] \"nn_softmax\" \"output\" iris_module = model_descriptor_to_module(iris_mds_union, output_pointers, list_output = TRUE) iris_module$graph$plot(html = TRUE) iris_mds_union$task  # contains all features #> <TaskClassif:iris> (150 x 5): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (4): #>   - dbl (4): Petal.Length, Petal.Width, Sepal.Length, Sepal.Width  iris_td = task_dataset(   task = iris_mds_union$task,   feature_ingress_tokens = iris_mds_union$ingress,   device = \"cpu\" )  batch = iris_td$.getbatch(1:2) batch #> $x #> $x$sepal.in.input #> torch_tensor #>  5.1000  3.5000 #>  4.9000  3.0000 #> [ CPUFloatType{2,2} ] #>  #> $x$petal.in.input #> torch_tensor #>  1.4000  0.2000 #>  1.4000  0.2000 #> [ CPUFloatType{2,2} ] #>  #>  #> $.index #> torch_tensor #>  1 #>  2 #> [ CPULongType{2} ] iris_module$graph$keep_results = TRUE  iris_module(   sepal.in.input = batch$x$sepal.in.input,   petal.in.input = batch$x$petal.in.input ) #> $lin_out.output #> torch_tensor #>  0.5751 #>  0.4885 #> [ CPUFloatType{2,1} ][ grad_fn = <AddmmBackward0> ] #>  #> $nn_softmax.output #> torch_tensor #>  0.1594  0.4310  0.4096 #>  0.1743  0.4296  0.3961 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ] iris_module$graph$pipeops$linear1$.result #> $output #> torch_tensor #>  2.6239  0.8791 -1.1404  0.1583 #>  2.4387  1.0496 -1.3065  0.2425 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$linear3$.result #> $output #> torch_tensor #> -0.0769  0.4376  0.2229 -0.0915 -0.3279 #> -0.0769  0.4376  0.2229 -0.0915 -0.3279 #> [ CPUFloatType{2,5} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$nn_merge_cat$.result #> $output #> torch_tensor #>  2.6239  0.8791 -1.1404  0.1583 -0.0769  0.4376  0.2229 -0.0915 -0.3279 #>  2.4387  1.0496 -1.3065  0.2425 -0.0769  0.4376  0.2229 -0.0915 -0.3279 #> [ CPUFloatType{2,9} ][ grad_fn = <CatBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"building-torch-learners","dir":"Articles","previous_headings":"","what":"Building Torch Learners","title":"Internals","text":"now seen NN Graphs PipeOpModule created turned nn_modules. Using PipeOpTorch even creates ModelDescriptor objects contain additional info batch tensors extracted Tasks. complete Learner, still necessary define loss-function used optimization, optimizer, optionally callbacks. already covered class representations – TorchLoss, TorchOptimizer, TorchCallbacks, Get Started vignette. use adam optimizer, cross-entropy loss function, history callback.","code":"adam = t_opt(\"adam\", lr = 0.02) adam #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: lr=0.02 #> * Packages: torch,mlr3torch  xe = t_loss(\"cross_entropy\") xe #> <TorchLoss:cross_entropy> Cross Entropy #> * Generator: nn_cross_entropy_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: classif  history = t_clbk(\"history\") history #> <TorchCallback:history> History #> * Generator: CallbackSetHistory #> * Parameters: list() #> * Packages: mlr3torch,torch"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"learnertorchmodel","dir":"Articles","previous_headings":"Building Torch Learners","what":"LearnerTorchModel","title":"Internals","text":"LearnerTorchModel represents supervised model (regression classification) using torch NNs. needs nn_module, well list TorchIngressToken define batches created Task. TorchIngressToken hard-code column-names Task used data-input, Learner created like therefore works specific Task created. (Generally full mlr3pipelines-UI used problem, see .) following uses sequential NN : training model, set hyperparameters. following calls $predict_newdata function plot response surface along Sepal.Width = mean(Sepal.Width) plane, along ground-truth values:","code":"lr_sequential = lrn(\"classif.torch_model\",   task_type = \"classif\",   network = model_descriptor_to_module(md_sequential, list(md_sequential$pointer)),   ingress_tokens = md_sequential$ingress,   optimizer = adam,   callbacks = history,   loss = xe )  lr_sequential #> <LearnerTorchModel[classif]:classif.model>: Torch Model #> * Model: - #> * Parameters: device=auto, num_threads=1, num_interop_threads=1, #>   seed=random, eval_freq=1, measures_train=<list>, #>   measures_valid=<list>, patience=0, min_delta=0, opt.lr=0.02 #> * Validate: NULL #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response], prob #> * Feature Types: logical, integer, numeric, character, factor, ordered, #>   POSIXct, lazy_tensor #> * Properties: featureless, hotstart_backward, hotstart_forward, #>   importance, internal_tuning, marshal, missings, multiclass, #>   oob_error, selected_features, twoclass, validation, weights #> * Optimizer: adam #> * Loss: cross_entropy #> * Callbacks: history lr_sequential$param_set$set_values(   batch_size = 50,   epochs = 100,   measures_train = msrs(c(\"classif.logloss\", \"classif.ce\")) ) # This is required to evaluate the logloss during training lr_sequential$predict_type = \"prob\"  lr_sequential$train(md_sequential$task) library(data.table) library(ggplot2)  newdata = cbind(data.table(Sepal.Width = mean(iris$Sepal.Width)), CJ(   Sepal.Length = seq(min(iris$Sepal.Length), max(iris$Sepal.Length), length.out = 30),   Petal.Length = seq(min(iris$Petal.Length), max(iris$Petal.Length), length.out = 30) ))  predictions = lr_sequential$predict_newdata(newdata)  plot_predictions = function(predictions) {   ggplot(cbind(newdata, Species = predictions$response),       aes(x = Sepal.Length, y = Petal.Length, fill = Species)) +     geom_tile(alpha = .3) +     geom_tile(alpha = .3) +     geom_point(data = iris,       aes(x = Sepal.Length, y = Petal.Length, fill = Species),       color = \"black\", pch = 21, size = 3) +     theme_bw() } plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"torch-learner-pipelines","dir":"Articles","previous_headings":"","what":"Torch Learner Pipelines","title":"Internals","text":"model shown constructed using ModelDescriptor generated Graph PipeOpTorch operators. ModelDescriptor furthermore contains Task pertains. makes possible use create NN model gets trained right away, using PipeOpTorchModelClassif. missing prerequisite now add desired TorchOptimizer TorchLoss information ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"adding-optimizer-loss-and-callback-meta-info-to-modeldescriptor","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"Adding Optimizer, Loss and Callback Meta-Info to ModelDescriptor","title":"Internals","text":"Remember ModelDescriptor $optimizer, $loss $callbacks slots necessary build complete Learner NN. can set corresponding PipeOpTorch operators. po(\"torch_optimizer\") used set $optimizer slot ModelDescriptor; takes desired TorchOptimizer object construction exports ParamSet. works analogously loss-function. also callbacks:","code":"po_adam = po(\"torch_optimizer\", optimizer = adam)  # hyperparameters are made available and can be changed: po_adam$param_set$values #> $lr #> [1] 0.02  md_sequential = po_adam$train(list(md_sequential))[[1]]  md_sequential$optimizer #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: lr=0.02 #> * Packages: torch,mlr3torch po_xe = po(\"torch_loss\", loss = xe) md_sequential = po_xe$train(list(md_sequential))[[1]]  md_sequential$loss #> <TorchLoss:cross_entropy> Cross Entropy #> * Generator: nn_cross_entropy_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: classif po_history = po(\"torch_callbacks\", callbacks = t_clbk(\"history\")) md_sequential = po_history$train(list(md_sequential))[[1]]  md_sequential$callbacks #> $history #> <TorchCallback:history> History #> * Generator: CallbackSetHistory #> * Parameters: list() #> * Packages: mlr3torch,torch"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"combined-instantiation-and-training-of-learnertorchmodel","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"Combined Instantiation and Training of LearnerTorchModel","title":"Internals","text":"ModelDescriptor can now given po(\"torch_model_classif\"). po(\"torch_model_classif\") behaves similarly PipeOpLearner: returns NULL training, prediction $predict(). po(\"torch_model_classif\") behaves similarly PipeOpLearner: returns NULL training, prediction $predict().","code":"po_model = po(\"torch_model_classif\", batch_size = 50, epochs = 50)  po_model$train(list(md_sequential)) #> $output #> NULL newtask = TaskClassif$new(\"newdata\", cbind(newdata, Species = factor(NA, levels = levels(iris$Species))), target = \"Species\")  predictions = po_model$predict(list(newtask))[[1]]  plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"the-whole-pipeline","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"The whole Pipeline","title":"Internals","text":"Remember md_sequential created using Graph initial Task piped . combine Graph PipeOpTorchModelClassif, get Graph behaves like Graph ends PipeOpLearner, can therefore wrapped GraphLearner. following uses one hidden layer : Compare resulting Graph Graph trained model: Predictions, (can use predict_newdata ):","code":"graph_sequential_full = po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 4, id = \"linear1\") %>>%     po(\"nn_sigmoid\") %>>%     po(\"nn_linear\", out_features = 3, id = \"linear2\") %>>%     po(\"nn_softmax\", dim = 2, id = \"softmax\") %>>%     po(\"nn_linear\", out_features = 3, id = \"linear3\") %>>%     po(\"nn_softmax\", dim = 2, id = \"softmax2\") %>>%     po(\"torch_optimizer\", optimizer = adam) %>>%     po(\"torch_loss\", loss = xe) %>>%     po(\"torch_callbacks\", callbacks = history) %>>%     po(\"torch_model_classif\", batch_size = 50, epochs = 100) lr_sequential_full = as_learner(graph_sequential_full)  lr_sequential_full$train(task) graph_sequential_full$plot(html = TRUE) model = lr_sequential_full$graph_model$state$torch_model_classif$model model$network$graph$plot(html = TRUE) predictions = lr_sequential_full$predict_newdata(newdata)  plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"mixed-pipelines","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"Mixed Pipelines","title":"Internals","text":"just limited PipeOpTorch kinds Graphs, also limited single PipeOpTorchIngress. following pipeline, example, removes Petal.Length columns Task fits model:  using Petal.Length Sepal.Length separately first?  examples hopefully demonstrated possibilities come representation neural network layers PipeOps. Even though vignette quite technical, hope given -depth understanding underlying mechanisms.","code":"gr = po(\"select\", selector = selector_name(\"Petal.Length\")) %>>%   po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 5, id = \"linear1\") %>>%   po(\"nn_relu\") %>>%   po(\"nn_linear\", out_features = 3, id = \"linear2\") %>>%   po(\"nn_softmax\", dim = 2) %>>%   po(\"torch_optimizer\", optimizer = adam) %>>%   po(\"torch_loss\", loss = xe) %>>%   po(\"torch_model_classif\", batch_size = 50, epochs = 50)  gr$plot(html = TRUE) lr = as_learner(gr) lr$train(task) predictions = lr$predict_newdata(newdata)  plot_predictions(predictions) gr = gunion(list(       po(\"select\", selector = selector_name(\"Petal.Length\"), id = \"sel1\") %>>%         po(\"torch_ingress_num\", id = \"ingress.petal\") %>>%         po(\"nn_linear\", out_features = 3, id = \"linear1\"),       po(\"select\", selector = selector_name(\"Sepal.Length\"), id = \"sel2\") %>>%         po(\"torch_ingress_num\", id = \"ingress.sepal\") %>>%         po(\"nn_linear\", out_features = 3, id = \"linear2\")     )) %>>%     po(\"nn_merge_cat\")  %>>%     po(\"nn_relu\", id = \"act1\") %>>%     po(\"nn_linear\", out_features = 3, id = \"linear3\") %>>%     po(\"nn_softmax\", dim = 2, id = \"act3\") %>>%     po(\"torch_optimizer\", optimizer = adam, lr = 0.1) %>>%     po(\"torch_loss\", loss = xe) %>>%     po(\"torch_model_classif\", batch_size = 50, epochs = 50)  gr$plot(html = TRUE) lr = as_learner(gr) lr$train(task) predictions = lr$predict_newdata(newdata)  plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/lazy_tensor.html","id":"creating-a-lazy-tensor","dir":"Articles","previous_headings":"","what":"Creating a Lazy Tensor","title":"Lazy Tensor","text":"Every lazy_tensor built top torch::dataset, assume familiar . information create torch::datasets, recommend reading torch package documentation. additional restriction impose dataset must .getitem .getbatch method returns list named tensors. example, create lazy_tensor length 1000, whose elements drawn uniform distribution \\([0, 1]\\). data stored -memory example, necessary $.getitem() method can e.g. load images disk. order create lazy_tensor mydata, annotate returned shapes dataset passing named list dataset_shapes. first dimension must NA, batch dimension. can also set shape NULL indicate unknown, .e. varies elements. Note case, implemented .getbatch method, even omitted specifying dataset_shapes auto-inferred. can convert vector torch_tensor just like : added preprocessing, calling $.getbatch() method mydata selecting element x. continue creating example task lt, relationship x y variable polynomial. Note target variable, classification regression, lazy_tensor, must factor numeric respectively. , plot data:  next section, create custom PipeOp fit polynomial regression model.","code":"mydata = dataset(   initialize = function() {     self$x = runif(1000, -1, 1)   },   .getbatch = function(i) list(x = torch_tensor(self$x[i])$unsqueeze(2)),   .length = function() 1000 )() lt = as_lazy_tensor(mydata, dataset_shapes = list(x = c(NA, 1))) lt[1:5] #> <ltnsr[5]> #> [1] <tnsr[1]> <tnsr[1]> <tnsr[1]> <tnsr[1]> <tnsr[1]> materialize(lt[1:5], rbind = TRUE) #> torch_tensor #> -0.9852 #> -0.0672 #> -0.0044 #> -0.4205 #>  0.4658 #> [ CPUFloatType{5,1} ] torch_equal(   materialize(lt[1], rbind = TRUE),   mydata$.getbatch(1)$x ) #> [1] TRUE library(data.table) x = mydata$x y = 0.2 + 0.1 * x - 0.1 * x^2 - 0.3 * x^3 + 0.5 * x^4 + 0.5 * x^7 + 0.6 * x^11 + rnorm(length(mydata)) * 0.1 dt = data.table(y = y, x = lt) task_poly = as_task_regr(dt, target = \"y\", id = \"poly\") task_poly #> <TaskRegr:poly> (1000 x 2) #> * Target: y #> * Properties: - #> * Features (1): #>   - lt (1): x library(ggplot2) ggplot(data = data.frame(x = x, y = y)) +   geom_point(aes(x = x, y = y), alpha = 0.5)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/lazy_tensor.html","id":"custom-preprocessing","dir":"Articles","previous_headings":"","what":"Custom Preprocessing","title":"Lazy Tensor","text":"order create custom preprocessing operator lazy tensor, create new PipeOp class. make convenient possible, mlr3torch offters pipeop_preproc_torch() function recommend using purpose. important arguments : id - Used default identifier PipeOp fn - preprocessing function. default, first argument assumed torch_tensor remaining arguments part PipeOp’s parameter set. shapes_out - function returns shapes output tensors given input shapes. can also set NULL unknown shape \"infer\" auto-inference, see ?pipeop_preproc_torch information. , create PipeOp, transforms vector x matrix \\((x^{d_1} ..., x^{d_n})\\), \\(d\\) degrees parameter PipeOp. can now create new instance PipeOp calling $new(), set parameter degrees degrees used simulating data . , set parameter stages, always available, \"\", means preprocessing applied training prediction. data augmentation can set \"train\". create polynomial regression learner, combine polynomial preprocessor lrn(\"regr.mlp\") hidden layer (.e. linear model) train learner task. , visualize predictions see model captured non-linear relationship reasonably:  next section, briefly cover implementation details lazy_tensor, necessary work data-type, feel free skip part.","code":"PipeOpPreprocTorchPoly = pipeop_preproc_torch(\"poly\",   fn = function(x, degrees) {     torch_cat(lapply(degrees, function(d) torch_pow(x, d)), dim = 2L)   },   shapes_out = \"infer\" ) po_poly = PipeOpPreprocTorchPoly$new()  po_poly$param_set$set_values(   degrees = c(0, 1, 2, 3, 4, 7, 11),   stages = \"both\" ) lrn_poly = as_learner(   po_poly %>>% lrn(\"regr.mlp\", batch_size = 256, epochs = 100,   neurons = integer(0)) )  lrn_poly$train(task_poly) pred = lrn_poly$predict(task_poly) dt = melt(data.table(   truth = pred$truth,   response = pred$response,   x = x),   id.vars = \"x\", measure.vars = c(\"truth\", \"response\") ) dt$variable = factor(dt$variable, levels = c(\"truth\", \"response\"))  ggplot(data = dt) +   geom_point(aes(x = x, y = value, color = variable))"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/lazy_tensor.html","id":"digging-into-internals","dir":"Articles","previous_headings":"","what":"Digging Into Internals","title":"Lazy Tensor","text":"Internally, lazy_tensor vector uses DataDescriptor class represent (possibly) preprocessed data. similar ModelDescriptor class used build neural nerworks using PipeOpTorch objects. DataDescriptor stores torch::dataset, mlr3pipelines::Graph metadata. Per default, preprocessing graph contains single PipOpNop nothing. printed output data descriptor informs us : number PipeOps contained preprocessing graph output shapes dataset input map, .e. data passed preprocessing graph, important multiple inputs pointer, points specific channel output PipeOp. output channel tensor represented DataDescriptor. Note id input po(\"nop\") randomly generated, needed prevent id clashes one input preprocessing graph. shape, shape tensor position pointer lazy tensor can constructed integer vector DataDescriptor. integer vector specifies element DataDescriptor lazy_tensor contains. , first two elements lazy_tensor vector represent element DataDescriptor, third element represents different element. Note indices refer DataDescriptor. Internally, lazy tensor represented list lists, element containing id DataDescriptor Currently, can single DataDescriptor lazy_tensor vector. happens materialize(lt[1]) following: Preprocessing lazy_tensor vector adds new PipeOps preprocessing graph updates metainformation like pointer output shape. show , create simple example task, using lt vector feature. see $graph new pipeop id \"poly.x\" output pointer points poly.x. Also see shape tensor now c(NA, 7) c(NA, 1) , can verify calling materialize() :","code":"desc = DataDescriptor$new(   dataset = mydata,   dataset_shapes = list(x = c(NA, 1)) ) desc #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,1)] #> * input_map: (x) -> Graph #> * pointer: nop.6b05fd.x.output #> * shape: [(NA,1)] lt = lazy_tensor(desc, ids = c(1, 1, 2)) materialize(lt, rbind = TRUE) #> torch_tensor #> -0.9852 #> -0.9852 #> -0.0672 #> [ CPUFloatType{3,1} ] unclass(lt[[1]]) #> [[1]] #> [1] 1 #>  #> [[2]] #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,1)] #> * input_map: (x) -> Graph #> * pointer: nop.6b05fd.x.output #> * shape: [(NA,1)] # get index and data descriptor desc = lt[[1]][[2]] id = lt[[1]][[1]]  # retrieve the batch <id> from the datast dataset_output = desc$dataset$.getbatch(id)  # batch is reorganized according to the input map graph_input = dataset_output[desc$input_map] names(graph_input) = names(desc$graph$input$name)  # the reorganized batch is fed into the preprocessing graph graph_output = desc$graph$train(graph_input, single_input = FALSE)  # the output pointed to by the pointer is returned tensor = graph_output[[paste0(desc$pointer, collapse = \".\")]] tensor #> torch_tensor #> -0.9852 #> [ CPUFloatType{1,1} ] taskin = as_task_regr(data.table(x = lt, y = 1:3), target = \"y\")  taskout = po_poly$train(list(taskin))[[1L]]  lt_out = taskout$data(cols = \"x\")$x  descout = lt_out[[1]][[2]]  descout #> <DataDescriptor: 2 ops> #> * dataset_shapes: [x: (NA,1)] #> * input_map: (x) -> Graph #> * pointer: poly.x.output #> * shape: [(NA,7)]  descout$graph #> Graph with 2 PipeOps: #>            ID         State sccssors    prdcssors #>        <char>        <char>   <char>       <char> #>  nop.6b05fd.x        <list>   poly.x              #>        poly.x <<UNTRAINED>>          nop.6b05fd.x materialize(lt_out[1:2], rbind = TRUE) #> torch_tensor #>  1.0000 -0.9852  0.9706 -0.9563  0.9421 -0.9009 -0.8487 #>  1.0000 -0.9852  0.9706 -0.9563  0.9421 -0.9009 -0.8487 #> [ CPUFloatType{2,7} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/authors.html","id":null,"dir":"","previous_headings":"","what":"Authors","title":"Authors and Citation","text":"Sebastian Fischer. Maintainer, author. Bernd Bischl. Contributor. Lukas Burk. Contributor. Martin Binder. Author. Florian Pfisterer. Contributor. Carson Zhang. Contributor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/authors.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Authors and Citation","text":"Fischer S, Binder M (2024). mlr3torch: Deep Learning 'mlr3'. R package version 0.1.2-9000, https://github.com/mlr-org/mlr3torch/, https://mlr3torch.mlr-org.com/.","code":"@Manual{,   title = {mlr3torch: Deep Learning with 'mlr3'},   author = {Sebastian Fischer and Martin Binder},   year = {2024},   note = {R package version 0.1.2-9000, https://github.com/mlr-org/mlr3torch/},   url = {https://mlr3torch.mlr-org.com/}, }"},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"mlr3torch-","dir":"","previous_headings":"","what":"Deep Learning with mlr3","title":"Deep Learning with mlr3","text":"Package website: release | dev Deep Learning torch mlr3.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"installation","dir":"","previous_headings":"","what":"Installation","title":"Deep Learning with mlr3","text":"Afterwards, also need run command : information installing torch can found .","code":"# Install from CRAN install.packages(\"mlr3torch\") # Install the development version from GitHub: pak::pak(\"mlr-org/mlr3torch\") torch::install_torch()"},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"what-is-mlr3torch","dir":"","previous_headings":"","what":"What is mlr3torch?","title":"Deep Learning with mlr3","text":"mlr3torch deep learning framework mlr3 ecosystem built top torch. allows easily build, train evaluate deep learning models lines codes, without needing worry low-level details. --shelf learners readily available, custom architectures can defined connecting PipeOpTorch operators mlr3pipelines::Graph. Using predefined learners simple multi layer perceptron (MLP) works just like mlr3 Learner. , train learner sonar example task: Next, construct architecture using PipeOpTorch objects. first pipeop – PipeOpTorchIngress – defines entrypoint network. subsequent pipeops define neural network layers. turn learner, configure loss, optimizer, callbacks well training arguments. work generic tensors, lazy_tensor type can used. wraps torch::dataset, allows preprocess data (lazily) using PipeOp objects. , flatten MNIST task, can train multi-layer perceptron . Note transform data -memory, applied data actually loaded. actually access tensors, can call materialize(). show slice resulting tensor readability: , define complex architecture one single input lazy_tensor. , first define single residual block: Next, create neural network takes input lazy_tensor (po(\"torch_ingress_ltnsr\")). first applies linear layer repeats layer using special PipeOpTorchBlock, followed network’s head. , configure loss, optimizer training parameters. Note po(\"nn_linear_0\") equivalent po(\"nn_linear\", id = \"nn_linear_0\") need avoid ID clashes linear layer po(\"nn_block\"). Next, prepend preprocessing step flattens images can directly apply learner unflattened MNIST task. order keep track performance training, use 20% data evaluate using classification accuracy. left train learner:","code":"library(mlr3torch) learner_mlp = lrn(\"classif.mlp\",   # defining network parameters   activation     = nn_relu,   neurons        = c(20, 20),   # training parameters   batch_size     = 16,   epochs         = 50,   device         = \"cpu\",   # Proportion of data to use for validation   validate = 0.3,   # Defining the optimizer, loss, and callbacks   optimizer      = t_opt(\"adam\", lr = 0.1),   loss           = t_loss(\"cross_entropy\"),   callbacks      = t_clbk(\"history\"), # this saves the history in the learner   # Measures to track   measures_valid = msrs(c(\"classif.logloss\", \"classif.ce\")),   measures_train = msrs(c(\"classif.acc\")),   # predict type (required by logloss)   predict_type = \"prob\" ) learner_mlp$train(tsk(\"sonar\")) architecture = po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 20) %>>%   po(\"nn_relu\") %>>%   po(\"nn_head\") graph_mlp = architecture %>>%   po(\"torch_loss\", loss = t_loss(\"cross_entropy\")) %>>%   po(\"torch_optimizer\", optimizer = t_opt(\"adam\", lr = 0.1)) %>>%   po(\"torch_callbacks\", callbacks = t_clbk(\"history\")) %>>%   po(\"torch_model_classif\",     batch_size = 16, epochs = 50, device = \"cpu\")  graph_lrn = as_learner(graph_mlp) # load the predefined mnist task mnist = tsk(\"mnist\") mnist$head(3L) #>     label           image #>    <fctr>   <lazy_tensor> #> 1:      5 <tnsr[1x28x28]> #> 2:      0 <tnsr[1x28x28]> #> 3:      4 <tnsr[1x28x28]>  # Flatten the images flattener = po(\"trafo_reshape\", shape = c(-1, 28 * 28)) mnist_flat = flattener$train(list(mnist))[[1L]]  mnist_flat$head(3L) #>     label         image #>    <fctr> <lazy_tensor> #> 1:      5   <tnsr[784]> #> 2:      0   <tnsr[784]> #> 3:      4   <tnsr[784]> materialize(   mnist_flat$data(1:2, cols = \"image\")[[1L]],   rbind = TRUE )[1:2, 1:4] #> torch_tensor #>  0  0  0  0 #>  0  0  0  0 #> [ CPUFloatType{2,4} ] layer = list(   po(\"nop\"),   po(\"nn_linear\", out_features = 50L) %>>%     po(\"nn_dropout\") %>>% po(\"nn_relu\") ) %>>% po(\"nn_merge_sum\") deep_network = po(\"torch_ingress_ltnsr\") %>>%   po(\"nn_linear_0\", out_features = 50L) %>>%   po(\"nn_block\", layer, n_blocks = 5L) %>>%   po(\"nn_head\") %>>%   po(\"torch_loss\", loss = t_loss(\"cross_entropy\")) %>>%   po(\"torch_optimizer\", optimizer = t_opt(\"adam\")) %>>%   po(\"torch_model_classif\",     epochs = 100L, batch_size = 32   ) deep_learner = as_learner(   flattener %>>% deep_network ) deep_learner$id = \"deep_network\" set_validate(deep_learner, 0.2) deep_learner$param_set$set_values(   torch_model_classif.measures_valid = msr(\"classif.acc\") ) deep_learner$train(mnist)"},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"feature-overview","dir":"","previous_headings":"","what":"Feature Overview","title":"Deep Learning with mlr3","text":"--shelf architectures readily available mlr3::Learners. Currently, supervised regression classification supported. Custom learners can defined using Graph language mlr3pipelines. package supports tabular data, well generic tensors via lazy_tensor type. Multi-modal data can handled conveniently, lazy_tensor objects can stored alongside tabular data. possible customize training process via (predefined custom) callbacks. package fully integrated mlr3 ecosystem. Neural network architectures, well hyperparameters can easily tuned via mlr3tuning friends.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"documentation","dir":"","previous_headings":"","what":"Documentation","title":"Deep Learning with mlr3","text":"Start reading one vignettes package website!","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"contributing","dir":"","previous_headings":"","what":"Contributing:","title":"Deep Learning with mlr3","text":"run tests one needs set environment variable TEST_TORCH = 1, e.g. adding .Renviron.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"acknowledgements","dir":"","previous_headings":"","what":"Acknowledgements","title":"Deep Learning with mlr3","text":"Without great R package torch none possible. names callback stages taken luz, another high-level deep learning framework R torch. Building neural networks using PipeOpTorch operators inspired keras.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"bugs-questions-feedback","dir":"","previous_headings":"","what":"Bugs, Questions, Feedback","title":"Deep Learning with mlr3","text":"mlr3torch free open source software project encourages participation feedback. issues, questions, suggestions feedback, please hesitate open “issue” GitHub page! case problems / bugs, often helpful provide “minimum working example” showcases behaviour (don’t worry bug obvious). Please understand resources project limited: response may sometimes delayed days, feature suggestions may rejected deemed tangential vision behind project.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":null,"dir":"Reference","previous_headings":"","what":"Data Descriptor — DataDescriptor","title":"Data Descriptor — DataDescriptor","text":"data descriptor rather internal data structure used lazy_tensor data type. essence annotated torch::dataset preprocessing graph (consisting mosty PipeOpModule operators). additional meta data (e.g. pointer, shapes) allows preprocess lazy_tensors mlr3pipelines::Graph just like (non-lazy) data types. preprocessing applied materialize() called lazy_tensor. create data descriptor, can also use as_data_descriptor() function.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Data Descriptor — DataDescriptor","text":"natural define S3 class, opted R6 class avoid usual trouble serializing S3 objects. row contained DataDescriptor S3 class, copy object serializing.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Data Descriptor — DataDescriptor","text":"dataset (torch::dataset) dataset. graph (Graph) preprocessing graph. dataset_shapes (named list() (integer() NULL)) shapes output. input_map (character()) input map dataset preprocessing graph. pointer (character(2)) output pointer. pointer_shape (integer() | NULL) shape output indicated pointer. dataset_hash (character(1)) Hash wrapped dataset. hash (character(1)) Hash data descriptor. graph_input (character()) input channels preprocessing graph (cached save time). pointer_shape_predict (integer() NULL) Internal use .","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Data Descriptor — DataDescriptor","text":"DataDescriptor$new() DataDescriptor$print() DataDescriptor$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Data Descriptor — DataDescriptor","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Data Descriptor — DataDescriptor","text":"","code":"DataDescriptor$new(   dataset,   dataset_shapes = NULL,   graph = NULL,   input_map = NULL,   pointer = NULL,   pointer_shape = NULL,   pointer_shape_predict = NULL,   clone_graph = TRUE )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Data Descriptor — DataDescriptor","text":"dataset (torch::dataset) torch dataset. return named list() torch_tensor objects. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. graph (Graph) preprocessing graph. left NULL, preprocessing applied data input_map, pointer, pointer_shape, pointer_shape_predict inferred case dataset returns one element. input_map (character()) Character vector must length input graph. Specifies data dataset fed preprocessing graph. pointer (character(2) | NULL) Points output channel within graph: Element 1 PipeOp's id element 2 PipeOp's output channel. pointer_shape (integer() | NULL) Shape output indicated pointer. pointer_shape_predict (integer() NULL) Internal use . Used Graph anticipate possible mismatches train predict shapes. clone_graph (logical(1)) Whether clone preprocessing graph.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Data Descriptor — DataDescriptor","text":"Prints object","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Data Descriptor — DataDescriptor","text":"","code":"DataDescriptor$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Data Descriptor — DataDescriptor","text":"... () Unused","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Data Descriptor — DataDescriptor","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Data Descriptor — DataDescriptor","text":"","code":"DataDescriptor$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Data Descriptor — DataDescriptor","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Data Descriptor — DataDescriptor","text":"","code":"# Create a dataset ds = dataset(   initialize = function() self$x = torch_randn(10, 3, 3),   .getitem = function(i) list(x = self$x[i, ]),   .length = function() nrow(self$x) )() dd = DataDescriptor$new(ds, list(x = c(NA, 3, 3))) dd #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,3,3)] #> * input_map: (x) -> Graph #> * pointer: nop.05a0f2.x.output #> * shape: [(NA,3,3)] # is the same as using the converter: as_data_descriptor(ds, list(x = c(NA, 3, 3))) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,3,3)] #> * input_map: (x) -> Graph #> * pointer: nop.05a0f2.x.output #> * shape: [(NA,3,3)]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/ModelDescriptor.html","id":null,"dir":"Reference","previous_headings":"","what":"Represent a Model with Meta-Info — ModelDescriptor","title":"Represent a Model with Meta-Info — ModelDescriptor","text":"Represents model; possibly complete model, possibly one process built . model takes input tensors shapes shapes_in pipes graph. Input shapes get mapped input channels graph. Output shapes named output channels graph; also possible represent -ops tensors, case names input output identical. ModelDescriptor objects typically represent partial models built , case pointer slot indicates specific point graph produces tensor shape pointer_shape, graph extended. allowed graph structure modified -reference different parts code. However, modifications may never add edges elements Graph destination. particular, element graph$input may removed reference, e.g. adding edge Graph input channel PipeOp previously without parent destination. cases better create specific ModelDescriptor training Graph consisting (mostly) operators PipeOpTorchIngress, PipeOpTorch, PipeOpTorchLoss, PipeOpTorchOptimizer, PipeOpTorchCallbacks. ModelDescriptor can converted nn_graph via model_descriptor_to_module.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/ModelDescriptor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Represent a Model with Meta-Info — ModelDescriptor","text":"","code":"ModelDescriptor(   graph,   ingress,   task,   optimizer = NULL,   loss = NULL,   callbacks = NULL,   pointer = NULL,   pointer_shape = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/ModelDescriptor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Represent a Model with Meta-Info — ModelDescriptor","text":"graph (Graph)Graph PipeOpModule PipeOpNOP operators. ingress (uniquely named list TorchIngressToken) List inputs go graph. Names must subset graph$input$name. task (Task) (Training)-Task model built. May necessary aspects loss use etc. optimizer (TorchOptimizer | NULL) Additional info: optimizer use. loss (TorchLoss | NULL) Additional info: loss use. callbacks (list CallbackSet NULL) Additional info: callbacks use. pointer (character(2) | NULL) Indicating element model . Points output channel within graph: Element 1 PipeOp's id element 2 PipeOp's output channel. pointer_shape (integer | NULL) Shape output indicated pointer.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/ModelDescriptor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Represent a Model with Meta-Info — ModelDescriptor","text":"(ModelDescriptor)","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentCenterCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentCenterCrop — mlr_pipeops_augment_center_crop","title":"PipeOpPreprocTorchAugmentCenterCrop — mlr_pipeops_augment_center_crop","text":"Calls torchvision::transform_center_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentCenterCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentCenterCrop — mlr_pipeops_augment_center_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentColorJitter.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentColorJitter — mlr_pipeops_augment_color_jitter","title":"PipeOpPreprocTorchAugmentColorJitter — mlr_pipeops_augment_color_jitter","text":"Calls torchvision::transform_color_jitter, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentColorJitter.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentColorJitter — mlr_pipeops_augment_color_jitter","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentCrop — mlr_pipeops_augment_crop","title":"PipeOpPreprocTorchAugmentCrop — mlr_pipeops_augment_crop","text":"Calls torchvision::transform_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentCrop — mlr_pipeops_augment_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentHflip.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentHflip — mlr_pipeops_augment_hflip","title":"PipeOpPreprocTorchAugmentHflip — mlr_pipeops_augment_hflip","text":"Calls torchvision::transform_hflip, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentHflip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentHflip — mlr_pipeops_augment_hflip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomAffine.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomAffine — mlr_pipeops_augment_random_affine","title":"PipeOpPreprocTorchAugmentRandomAffine — mlr_pipeops_augment_random_affine","text":"Calls torchvision::transform_random_affine, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomAffine.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomAffine — mlr_pipeops_augment_random_affine","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomChoice.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomChoice — mlr_pipeops_augment_random_choice","title":"PipeOpPreprocTorchAugmentRandomChoice — mlr_pipeops_augment_random_choice","text":"Calls torchvision::transform_random_choice, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomChoice.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomChoice — mlr_pipeops_augment_random_choice","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomCrop — mlr_pipeops_augment_random_crop","title":"PipeOpPreprocTorchAugmentRandomCrop — mlr_pipeops_augment_random_crop","text":"Calls torchvision::transform_random_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomCrop — mlr_pipeops_augment_random_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomHorizontalFlip.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomHorizontalFlip — mlr_pipeops_augment_random_horizontal_flip","title":"PipeOpPreprocTorchAugmentRandomHorizontalFlip — mlr_pipeops_augment_random_horizontal_flip","text":"Calls torchvision::transform_random_horizontal_flip, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomHorizontalFlip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomHorizontalFlip — mlr_pipeops_augment_random_horizontal_flip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomOrder.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomOrder — mlr_pipeops_augment_random_order","title":"PipeOpPreprocTorchAugmentRandomOrder — mlr_pipeops_augment_random_order","text":"Calls torchvision::transform_random_order, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomOrder.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomOrder — mlr_pipeops_augment_random_order","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomResizedCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomResizedCrop — mlr_pipeops_augment_random_resized_crop","title":"PipeOpPreprocTorchAugmentRandomResizedCrop — mlr_pipeops_augment_random_resized_crop","text":"Calls torchvision::transform_random_resized_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomResizedCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomResizedCrop — mlr_pipeops_augment_random_resized_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomVerticalFlip.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomVerticalFlip — mlr_pipeops_augment_random_vertical_flip","title":"PipeOpPreprocTorchAugmentRandomVerticalFlip — mlr_pipeops_augment_random_vertical_flip","text":"Calls torchvision::transform_random_vertical_flip, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomVerticalFlip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomVerticalFlip — mlr_pipeops_augment_random_vertical_flip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentResizedCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentResizedCrop — mlr_pipeops_augment_resized_crop","title":"PipeOpPreprocTorchAugmentResizedCrop — mlr_pipeops_augment_resized_crop","text":"Calls torchvision::transform_resized_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentResizedCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentResizedCrop — mlr_pipeops_augment_resized_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRotate.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRotate — mlr_pipeops_augment_rotate","title":"PipeOpPreprocTorchAugmentRotate — mlr_pipeops_augment_rotate","text":"Calls torchvision::transform_rotate, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRotate.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRotate — mlr_pipeops_augment_rotate","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentVflip.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentVflip — mlr_pipeops_augment_vflip","title":"PipeOpPreprocTorchAugmentVflip — mlr_pipeops_augment_vflip","text":"Calls torchvision::transform_vflip, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentVflip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentVflip — mlr_pipeops_augment_vflip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustBrightness.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoAdjustBrightness — mlr_pipeops_trafo_adjust_brightness","title":"PipeOpPreprocTorchTrafoAdjustBrightness — mlr_pipeops_trafo_adjust_brightness","text":"Calls torchvision::transform_adjust_brightness, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustBrightness.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoAdjustBrightness — mlr_pipeops_trafo_adjust_brightness","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustGamma.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoAdjustGamma — mlr_pipeops_trafo_adjust_gamma","title":"PipeOpPreprocTorchTrafoAdjustGamma — mlr_pipeops_trafo_adjust_gamma","text":"Calls torchvision::transform_adjust_gamma, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustGamma.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoAdjustGamma — mlr_pipeops_trafo_adjust_gamma","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustHue.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoAdjustHue — mlr_pipeops_trafo_adjust_hue","title":"PipeOpPreprocTorchTrafoAdjustHue — mlr_pipeops_trafo_adjust_hue","text":"Calls torchvision::transform_adjust_hue, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustHue.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoAdjustHue — mlr_pipeops_trafo_adjust_hue","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustSaturation.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoAdjustSaturation — mlr_pipeops_trafo_adjust_saturation","title":"PipeOpPreprocTorchTrafoAdjustSaturation — mlr_pipeops_trafo_adjust_saturation","text":"Calls torchvision::transform_adjust_saturation, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustSaturation.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoAdjustSaturation — mlr_pipeops_trafo_adjust_saturation","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoGrayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoGrayscale — mlr_pipeops_trafo_grayscale","title":"PipeOpPreprocTorchTrafoGrayscale — mlr_pipeops_trafo_grayscale","text":"Calls torchvision::transform_grayscale, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoGrayscale.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoGrayscale — mlr_pipeops_trafo_grayscale","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoNop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoNop — mlr_pipeops_trafo_nop","title":"PipeOpPreprocTorchTrafoNop — mlr_pipeops_trafo_nop","text":"nothing.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoNop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoNop — mlr_pipeops_trafo_nop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoNormalize.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoNormalize — mlr_pipeops_trafo_normalize","title":"PipeOpPreprocTorchTrafoNormalize — mlr_pipeops_trafo_normalize","text":"Calls torchvision::transform_normalize, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoNormalize.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoNormalize — mlr_pipeops_trafo_normalize","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoPad.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoPad — mlr_pipeops_trafo_pad","title":"PipeOpPreprocTorchTrafoPad — mlr_pipeops_trafo_pad","text":"Calls torchvision::transform_pad, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoPad.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoPad — mlr_pipeops_trafo_pad","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoReshape.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoReshape — mlr_pipeops_trafo_reshape","title":"PipeOpPreprocTorchTrafoReshape — mlr_pipeops_trafo_reshape","text":"Reshapes tensor according parameter shape, calling torch_reshape(). preprocessing function applied batch-wise.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoReshape.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoReshape — mlr_pipeops_trafo_reshape","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoReshape.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PipeOpPreprocTorchTrafoReshape — mlr_pipeops_trafo_reshape","text":"shape :: integer() desired output shape. first dimension batch dimension usually -1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoResize.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoResize — mlr_pipeops_trafo_resize","title":"PipeOpPreprocTorchTrafoResize — mlr_pipeops_trafo_resize","text":"Calls torchvision::transform_resize, see information parameters. preprocessing applied whole batch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoResize.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoResize — mlr_pipeops_trafo_resize","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoRgbToGrayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoRgbToGrayscale — mlr_pipeops_trafo_rgb_to_grayscale","title":"PipeOpPreprocTorchTrafoRgbToGrayscale — mlr_pipeops_trafo_rgb_to_grayscale","text":"Calls torchvision::transform_rgb_to_grayscale, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoRgbToGrayscale.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoRgbToGrayscale — mlr_pipeops_trafo_rgb_to_grayscale","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Callback — TorchCallback","title":"Torch Callback — TorchCallback","text":"wraps CallbackSet annotates metadata, importantly ParamSet. callback created given parameter values calling $generate() method. class usually used configure callback torch learner, e.g. constructing learner ModelDescriptor. list available callbacks, see mlr3torch_callbacks. conveniently retrieve TorchCallback, use t_clbk().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Callback — TorchCallback","text":"Defined constructor argument param_set. parameter set provided construction, parameter set constructed creating parameter argument wrapped loss function, parametes type ParamUty.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Torch Callback — TorchCallback","text":"mlr3torch::TorchDescriptor -> TorchCallback","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Callback — TorchCallback","text":"mlr3torch::TorchDescriptor$generate() mlr3torch::TorchDescriptor$help() mlr3torch::TorchDescriptor$print()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Callback — TorchCallback","text":"TorchCallback$new() TorchCallback$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Callback — TorchCallback","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Callback — TorchCallback","text":"","code":"TorchCallback$new(   callback_generator,   param_set = NULL,   id = NULL,   label = NULL,   packages = NULL,   man = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Callback — TorchCallback","text":"callback_generator (R6ClassGenerator) class generator callback wrapped. param_set (ParamSet NULL) parameter set. NULL (default) inferred callback_generator. id (character(1)) id new object. label (character(1)) Label new instance. packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Callback — TorchCallback","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Callback — TorchCallback","text":"","code":"TorchCallback$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Callback — TorchCallback","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Callback — TorchCallback","text":"","code":"# Create a new torch callback from an existing callback set torch_callback = TorchCallback$new(CallbackSetCheckpoint) # The parameters are inferred torch_callback$param_set #> <ParamSet(3)> #>           id    class lower upper nlevels        default  value #>       <char>   <char> <num> <num>   <num>         <list> <list> #> 1:      path ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:      freq ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 3: freq_type ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]  # Retrieve a torch callback from the dictionary torch_callback = t_clbk(\"checkpoint\",   path = tempfile(), freq = 1 ) torch_callback #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: path=/tmp/Rtmpc8mJ4Q/file1a057ccb48bf, freq=1 #> * Packages: mlr3torch,torch torch_callback$label #> [1] \"Checkpoint\" torch_callback$id #> [1] \"checkpoint\"  # open the help page of the wrapped callback set # torch_callback$help()  # Create the callback set callback = torch_callback$generate() callback #> <CallbackSetCheckpoint> #> * Stages: on_batch_end, on_epoch_end, on_exit # is the same as CallbackSetCheckpoint$new(   path = tempfile(), freq = 1 ) #> <CallbackSetCheckpoint> #> * Stages: on_batch_end, on_epoch_end, on_exit  # Use in a learner learner = lrn(\"regr.mlp\", callbacks = t_clbk(\"checkpoint\")) # the parameters of the callback are added to the learner's parameter set learner$param_set #> <ParamSetCollection(37)> #>                          id    class lower upper nlevels        default #>                      <char>   <char> <num> <num>   <num>         <list> #>  1:                  epochs ParamInt 0e+00   Inf     Inf <NoDefault[0]> #>  2:                  device ParamFct    NA    NA      12 <NoDefault[0]> #>  3:             num_threads ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  4:     num_interop_threads ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  5:                    seed ParamInt  -Inf   Inf     Inf <NoDefault[0]> #>  6:               eval_freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  7:          measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  8:          measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:                patience ParamInt 0e+00   Inf     Inf <NoDefault[0]> #> 10:               min_delta ParamDbl 0e+00   Inf     Inf <NoDefault[0]> #> 11:              batch_size ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 12:                 shuffle ParamLgl    NA    NA       2          FALSE #> 13:                 sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 14:           batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:             num_workers ParamInt 0e+00   Inf     Inf              0 #> 16:              collate_fn ParamUty    NA    NA     Inf         [NULL] #> 17:              pin_memory ParamLgl    NA    NA       2          FALSE #> 18:               drop_last ParamLgl    NA    NA       2          FALSE #> 19:                 timeout ParamDbl  -Inf   Inf     Inf             -1 #> 20:          worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 21:          worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:         worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:                 neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:                       p ParamDbl 0e+00 1e+00     Inf <NoDefault[0]> #> 25:                n_layers ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 26:              activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:         activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:                   shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:                  opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 30:               opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 31:                 opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 32:        opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 33:             opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 34:          loss.reduction ParamFct    NA    NA       2           mean #> 35:      cb.checkpoint.path ParamUty    NA    NA     Inf <NoDefault[0]> #> 36:      cb.checkpoint.freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 37: cb.checkpoint.freq_type ParamFct    NA    NA       2          epoch #>                          id    class lower upper nlevels        default #>            value #>           <list> #>  1:       [NULL] #>  2:         auto #>  3:            1 #>  4:            1 #>  5:       random #>  6:            1 #>  7:    <list[0]> #>  8:    <list[0]> #>  9:            0 #> 10:            0 #> 11:       [NULL] #> 12:       [NULL] #> 13:       [NULL] #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:              #> 24:          0.5 #> 25:       [NULL] #> 26: <nn_relu[1]> #> 27:    <list[0]> #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [NULL] #> 37:       [NULL] #>            value"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Descriptors — TorchDescriptor","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Abstract Base Class TorchLoss, TorchOptimizer, TorchCallback inherit. class wraps generator (R6Class Generator torch version generator) annotates metadata ParamSet, label, ID, packages, manual page. parameters construction arguments wrapped generator parameter $values passed generator calling public method $generate().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Defined constructor argument param_set. parameters tagged \"train\", done automatically initialize.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"label (character(1)) Label object. Can used tables, plot text output instead ID. param_set (ParamSet) Set hyperparameters. packages (character(1)) Set required packages. packages loaded, attached. id (character(1)) Identifier object. Used tables, plot text output. generator wrapped generator described. man (character(1)) String format [pkg]::[topic] pointing manual page object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"phash (character(1)) Hash (unique identifier) partial object, excluding components varied systematically (e.g. parameter values).","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"TorchDescriptor$new() TorchDescriptor$print() TorchDescriptor$generate() TorchDescriptor$help() TorchDescriptor$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$new(   generator,   id = NULL,   param_set = NULL,   packages = NULL,   label = NULL,   man = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"generator wrapped generator described. id (character(1)) id new object. param_set (ParamSet) parameter set. packages (character()) R packages object depends . label (character(1)) Label new instance. man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Prints object","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"... ","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-generate-","dir":"Reference","previous_headings":"","what":"Method generate()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Calls generator given parameter values.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$generate()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-help-","dir":"Reference","previous_headings":"","what":"Method help()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Displays help file wrapped object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$help()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Ingress Token — TorchIngressToken","title":"Torch Ingress Token — TorchIngressToken","text":"function creates S3 class class \"TorchIngressToken\", internal data structure. contains (meta-)information batch generated Task fed entry point neural network. stored ingress field ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Ingress Token — TorchIngressToken","text":"","code":"TorchIngressToken(features, batchgetter, shape)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Ingress Token — TorchIngressToken","text":"features (character) Features batchgetter operate. batchgetter (function) Function two arguments: data device. function given output Task$data(rows = batch_indices, cols = features) produce tensor shape shape_out. shape (integer) Shape batchgetter produce. Batch-dimension included NA.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Torch Ingress Token — TorchIngressToken","text":"TorchIngressToken object.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Ingress Token — TorchIngressToken","text":"","code":"# Define a task for which we want to define an ingress token task = tsk(\"iris\")  # We create an ingress token for two feature Sepal.Length and Petal.Length: # We have to specify the features, the batchgetter and the shape features = c(\"Sepal.Length\", \"Petal.Length\") # As a batchgetter we use batchgetter_num  batch_dt = task$data(rows = 1:10, cols =features) batch_dt #>     Sepal.Length Petal.Length #>            <num>        <num> #>  1:          5.1          1.4 #>  2:          4.9          1.4 #>  3:          4.7          1.3 #>  4:          4.6          1.5 #>  5:          5.0          1.4 #>  6:          5.4          1.7 #>  7:          4.6          1.4 #>  8:          5.0          1.5 #>  9:          4.4          1.4 #> 10:          4.9          1.5 batch_tensor = batchgetter_num(batch_dt, \"cpu\") batch_tensor #> torch_tensor #>  5.1000  1.4000 #>  4.9000  1.4000 #>  4.7000  1.3000 #>  4.6000  1.5000 #>  5.0000  1.4000 #>  5.4000  1.7000 #>  4.6000  1.4000 #>  5.0000  1.5000 #>  4.4000  1.4000 #>  4.9000  1.5000 #> [ CPUFloatType{10,2} ]  # The shape is unknown in the first dimension (batch dimension)  ingress_token = TorchIngressToken(   features = features,   batchgetter = batchgetter_num,   shape = c(NA, 2) ) ingress_token #> Ingress: Task[Sepal.Length,Petal.Length] --> Tensor(NA, 2)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Loss — TorchLoss","title":"Torch Loss — TorchLoss","text":"wraps torch::nn_loss annotates metadata, importantly ParamSet. loss function created given parameter values calling $generate() method. class usually used configure loss function torch learner, e.g. construcing learner ModelDescriptor. list available losses, see mlr3torch_losses. Items dictionary can retrieved using t_loss().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Loss — TorchLoss","text":"Defined constructor argument param_set. parameter set provided construction, parameter set constructed creating parameter argument wrapped loss function, parametes type ParamUty.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Torch Loss — TorchLoss","text":"mlr3torch::TorchDescriptor -> TorchLoss","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Torch Loss — TorchLoss","text":"task_types (character()) task types loss supports.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Loss — TorchLoss","text":"mlr3torch::TorchDescriptor$generate() mlr3torch::TorchDescriptor$help()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Loss — TorchLoss","text":"TorchLoss$new() TorchLoss$print() TorchLoss$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Loss — TorchLoss","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Loss — TorchLoss","text":"","code":"TorchLoss$new(   torch_loss,   task_types = NULL,   param_set = NULL,   id = NULL,   label = NULL,   packages = NULL,   man = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Loss — TorchLoss","text":"torch_loss (nn_loss) loss module. task_types (character()) task types supported loss. param_set (ParamSet NULL) parameter set. NULL (default) inferred torch_loss. id (character(1)) id new object. label (character(1)) Label new instance. packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Torch Loss — TorchLoss","text":"Prints object","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Loss — TorchLoss","text":"","code":"TorchLoss$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Loss — TorchLoss","text":"... ","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Loss — TorchLoss","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Loss — TorchLoss","text":"","code":"TorchLoss$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Loss — TorchLoss","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Loss — TorchLoss","text":"","code":"# Create a new torch loss torch_loss = TorchLoss$new(torch_loss = nn_mse_loss, task_types = \"regr\") torch_loss #> <TorchLoss:nn_mse_loss> nn_mse_loss #> * Generator: nn_mse_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr # the parameters are inferred torch_loss$param_set #> <ParamSet(1)> #>           id    class lower upper nlevels        default  value #>       <char>   <char> <num> <num>   <num>         <list> <list> #> 1: reduction ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]  # Retrieve a loss from the dictionary: torch_loss = t_loss(\"mse\", reduction = \"mean\") # is the same as torch_loss #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: reduction=mean #> * Packages: torch,mlr3torch #> * Task Types: regr torch_loss$param_set #> <ParamSet(1)> #>           id    class lower upper nlevels default  value #>       <char>   <char> <num> <num>   <num>  <list> <list> #> 1: reduction ParamFct    NA    NA       2    mean   mean torch_loss$label #> [1] \"Mean Squared Error\" torch_loss$task_types #> [1] \"regr\" torch_loss$id #> [1] \"mse\"  # Create the loss function loss_fn = torch_loss$generate() loss_fn #> An `nn_module` containing 0 parameters. # Is the same as nn_mse_loss(reduction = \"mean\") #> An `nn_module` containing 0 parameters.  # open the help page of the wrapped loss function # torch_loss$help()  # Use in a learner learner = lrn(\"regr.mlp\", loss = t_loss(\"mse\")) # The parameters of the loss are added to the learner's parameter set learner$param_set #> <ParamSetCollection(34)> #>                      id    class lower upper nlevels        default #>                  <char>   <char> <num> <num>   <num>         <list> #>  1:              epochs ParamInt 0e+00   Inf     Inf <NoDefault[0]> #>  2:              device ParamFct    NA    NA      12 <NoDefault[0]> #>  3:         num_threads ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  4: num_interop_threads ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  5:                seed ParamInt  -Inf   Inf     Inf <NoDefault[0]> #>  6:           eval_freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  7:      measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  8:      measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:            patience ParamInt 0e+00   Inf     Inf <NoDefault[0]> #> 10:           min_delta ParamDbl 0e+00   Inf     Inf <NoDefault[0]> #> 11:          batch_size ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 12:             shuffle ParamLgl    NA    NA       2          FALSE #> 13:             sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 14:       batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:         num_workers ParamInt 0e+00   Inf     Inf              0 #> 16:          collate_fn ParamUty    NA    NA     Inf         [NULL] #> 17:          pin_memory ParamLgl    NA    NA       2          FALSE #> 18:           drop_last ParamLgl    NA    NA       2          FALSE #> 19:             timeout ParamDbl  -Inf   Inf     Inf             -1 #> 20:      worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 21:      worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:     worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:             neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:                   p ParamDbl 0e+00 1e+00     Inf <NoDefault[0]> #> 25:            n_layers ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 26:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:              opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 30:           opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 31:             opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 32:    opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 33:         opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 34:      loss.reduction ParamFct    NA    NA       2           mean #>                      id    class lower upper nlevels        default #>            value #>           <list> #>  1:       [NULL] #>  2:         auto #>  3:            1 #>  4:            1 #>  5:       random #>  6:            1 #>  7:    <list[0]> #>  8:    <list[0]> #>  9:            0 #> 10:            0 #> 11:       [NULL] #> 12:       [NULL] #> 13:       [NULL] #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:              #> 24:          0.5 #> 25:       [NULL] #> 26: <nn_relu[1]> #> 27:    <list[0]> #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #>            value"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Optimizer — TorchOptimizer","title":"Torch Optimizer — TorchOptimizer","text":"wraps torch::torch_optimizer_generatora annotates metadata, importantly ParamSet. optimizer created given parameter values calling $generate() method. class usually used configure optimizer torch learner, e.g. construcing learner ModelDescriptor. list available optimizers, see mlr3torch_optimizers. Items dictionary can retrieved using t_opt().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Optimizer — TorchOptimizer","text":"Defined constructor argument param_set. parameter set provided construction, parameter set constructed creating parameter argument wrapped loss function, parametes type ParamUty.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Torch Optimizer — TorchOptimizer","text":"mlr3torch::TorchDescriptor -> TorchOptimizer","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Optimizer — TorchOptimizer","text":"mlr3torch::TorchDescriptor$help() mlr3torch::TorchDescriptor$print()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Optimizer — TorchOptimizer","text":"TorchOptimizer$new() TorchOptimizer$generate() TorchOptimizer$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Optimizer — TorchOptimizer","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Optimizer — TorchOptimizer","text":"","code":"TorchOptimizer$new(   torch_optimizer,   param_set = NULL,   id = NULL,   label = NULL,   packages = NULL,   man = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Optimizer — TorchOptimizer","text":"torch_optimizer (torch_optimizer_generator) torch optimizer. param_set (ParamSet NULL) parameter set. NULL (default) inferred torch_optimizer. id (character(1)) id new object. label (character(1)) Label new instance. packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"method-generate-","dir":"Reference","previous_headings":"","what":"Method generate()","title":"Torch Optimizer — TorchOptimizer","text":"Instantiates optimizer.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Optimizer — TorchOptimizer","text":"","code":"TorchOptimizer$generate(params)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Optimizer — TorchOptimizer","text":"params (named list() torch_tensors) parameters network.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Torch Optimizer — TorchOptimizer","text":"torch_optimizer","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Optimizer — TorchOptimizer","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Optimizer — TorchOptimizer","text":"","code":"TorchOptimizer$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Optimizer — TorchOptimizer","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Optimizer — TorchOptimizer","text":"","code":"# Create a new torch loss torch_opt = TorchOptimizer$new(optim_adam, label = \"adam\") torch_opt #> <TorchOptimizer:optim_adam> adam #> * Generator: optim_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # If the param set is not specified, parameters are inferred but are of class ParamUty torch_opt$param_set #> <ParamSet(5)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1:           lr ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:        betas ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 3:          eps ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 4: weight_decay ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 5:      amsgrad ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]  # open the help page of the wrapped optimizer # torch_opt$help()  # Retrieve an optimizer from the dictionary torch_opt = t_opt(\"sgd\", lr = 0.1) torch_opt #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_sgd #> * Parameters: lr=0.1 #> * Packages: torch,mlr3torch torch_opt$param_set #> <ParamSet(5)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1:           lr ParamDbl     0   Inf     Inf <NoDefault[0]>    0.1 #> 2:     momentum ParamDbl     0     1     Inf              0 [NULL] #> 3:    dampening ParamDbl     0     1     Inf              0 [NULL] #> 4: weight_decay ParamDbl     0     1     Inf              0 [NULL] #> 5:     nesterov ParamLgl    NA    NA       2          FALSE [NULL] torch_opt$label #> [1] \"Stochastic Gradient Descent\" torch_opt$id #> [1] \"sgd\"  # Create the optimizer for a network net = nn_linear(10, 1) opt = torch_opt$generate(net$parameters)  # is the same as optim_sgd(net$parameters, lr = 0.1) #> <optim_sgd> #>   Inherits from: <torch_optimizer> #>   Public: #>     add_param_group: function (param_group)  #>     clone: function (deep = FALSE)  #>     defaults: list #>     initialize: function (params, lr = optim_required(), momentum = 0, dampening = 0,  #>     load_state_dict: function (state_dict, ..., .refer_to_state_dict = FALSE)  #>     param_groups: list #>     state: State, R6 #>     state_dict: function ()  #>     step: function (closure = NULL)  #>     zero_grad: function ()  #>   Private: #>     step_helper: function (closure, loop_fun)   # Use in a learner learner = lrn(\"regr.mlp\", optimizer = t_opt(\"sgd\")) # The parameters of the optimizer are added to the learner's parameter set learner$param_set #> <ParamSetCollection(34)> #>                      id    class lower upper nlevels        default #>                  <char>   <char> <num> <num>   <num>         <list> #>  1:              epochs ParamInt     0   Inf     Inf <NoDefault[0]> #>  2:              device ParamFct    NA    NA      12 <NoDefault[0]> #>  3:         num_threads ParamInt     1   Inf     Inf <NoDefault[0]> #>  4: num_interop_threads ParamInt     1   Inf     Inf <NoDefault[0]> #>  5:                seed ParamInt  -Inf   Inf     Inf <NoDefault[0]> #>  6:           eval_freq ParamInt     1   Inf     Inf <NoDefault[0]> #>  7:      measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  8:      measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:            patience ParamInt     0   Inf     Inf <NoDefault[0]> #> 10:           min_delta ParamDbl     0   Inf     Inf <NoDefault[0]> #> 11:          batch_size ParamInt     1   Inf     Inf <NoDefault[0]> #> 12:             shuffle ParamLgl    NA    NA       2          FALSE #> 13:             sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 14:       batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:         num_workers ParamInt     0   Inf     Inf              0 #> 16:          collate_fn ParamUty    NA    NA     Inf         [NULL] #> 17:          pin_memory ParamLgl    NA    NA       2          FALSE #> 18:           drop_last ParamLgl    NA    NA       2          FALSE #> 19:             timeout ParamDbl  -Inf   Inf     Inf             -1 #> 20:      worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 21:      worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:     worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:             neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:                   p ParamDbl     0     1     Inf <NoDefault[0]> #> 25:            n_layers ParamInt     1   Inf     Inf <NoDefault[0]> #> 26:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:              opt.lr ParamDbl     0   Inf     Inf <NoDefault[0]> #> 30:        opt.momentum ParamDbl     0     1     Inf              0 #> 31:       opt.dampening ParamDbl     0     1     Inf              0 #> 32:    opt.weight_decay ParamDbl     0     1     Inf              0 #> 33:        opt.nesterov ParamLgl    NA    NA       2          FALSE #> 34:      loss.reduction ParamFct    NA    NA       2           mean #>                      id    class lower upper nlevels        default #>            value #>           <list> #>  1:       [NULL] #>  2:         auto #>  3:            1 #>  4:            1 #>  5:       random #>  6:            1 #>  7:    <list[0]> #>  8:    <list[0]> #>  9:            0 #> 10:            0 #> 11:       [NULL] #> 12:       [NULL] #> 13:       [NULL] #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:              #> 24:          0.5 #> 25:       [NULL] #> 26: <nn_relu[1]> #> 27:    <list[0]> #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #>            value"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_data_descriptor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to Data Descriptor — as_data_descriptor","title":"Convert to Data Descriptor — as_data_descriptor","text":"Converts input DataDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_data_descriptor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to Data Descriptor — as_data_descriptor","text":"","code":"as_data_descriptor(x, dataset_shapes, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_data_descriptor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to Data Descriptor — as_data_descriptor","text":"x () Object convert. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. ... () arguments passed DataDescriptor constructor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_data_descriptor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert to Data Descriptor — as_data_descriptor","text":"","code":"ds = dataset(\"example\",   initialize = function() self$iris = iris[, -5],   .getitem = function(i) list(x = torch_tensor(as.numeric(self$iris[i, ]))),   .length = function() nrow(self$iris) )() as_data_descriptor(ds, list(x = c(NA, 4L))) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,4)] #> * input_map: (x) -> Graph #> * pointer: nop.fa4271.x.output #> * shape: [(NA,4)]  # if the dataset has a .getbatch method, the shapes are inferred ds2 = dataset(\"example\",   initialize = function() self$iris = iris[, -5],   .getbatch = function(i) list(x = torch_tensor(as.matrix(self$iris[i, ]))),   .length = function() nrow(self$iris) )() as_data_descriptor(ds2) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,4)] #> * input_map: (x) -> Graph #> * pointer: nop.ef35d5.x.output #> * shape: [(NA,4)]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to Lazy Tensor — as_lazy_tensor","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"Convert object lazy_tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"as_lazy_tensor(x, ...)  # S3 method for class 'dataset' as_lazy_tensor(x, dataset_shapes = NULL, ids = NULL, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"x () Object convert lazy_tensor ... () Additional arguments passed method. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. ids (integer()) ids include lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"iris_ds = dataset(\"iris\",   initialize = function() {     self$iris = iris[, -5]   },   .getbatch = function(i) {     list(x = torch_tensor(as.matrix(self$iris[i, ])))   },   .length = function() nrow(self$iris) )() # no need to specify the dataset shapes as they can be inferred from the .getbatch method # only first 5 observations as_lazy_tensor(iris_ds, ids = 1:5) #> <ltnsr[5]> #> [1] <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> # all observations head(as_lazy_tensor(iris_ds)) #> <ltnsr[6]> #> [1] <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]>  iris_ds2 = dataset(\"iris\",   initialize = function() self$iris = iris[, -5],   .getitem = function(i) list(x = torch_tensor(as.numeric(self$iris[i, ]))),   .length = function() nrow(self$iris) )() # if .getitem is implemented we cannot infer the shapes as they might vary, # so we have to annotate them explicitly as_lazy_tensor(iris_ds2, dataset_shapes = list(x = c(NA, 4L)))[1:5] #> <ltnsr[5]> #> [1] <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]>  # Convert a matrix lt = as_lazy_tensor(matrix(rnorm(100), nrow = 20)) materialize(lt[1:5], rbind = TRUE) #> torch_tensor #> -1.4000  0.4682  0.0700  1.0743  1.9243 #>  0.2553  0.3630 -0.6391 -0.6651  1.2984 #> -2.4373 -1.3045 -0.0500  1.1140  0.7488 #> -0.0056  0.7378 -0.2515 -0.2459  0.5562 #>  0.6216  1.8885  0.4448 -1.1776 -0.5483 #> [ CPUFloatType{5,5} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callback.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to a TorchCallback — as_torch_callback","title":"Convert to a TorchCallback — as_torch_callback","text":"Converts object TorchCallback.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callback.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to a TorchCallback — as_torch_callback","text":"","code":"as_torch_callback(x, clone = FALSE, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to a TorchCallback — as_torch_callback","text":"x () Object converted. clone (logical(1)) Whether make deep clone. ... () Additional arguments","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callback.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert to a TorchCallback — as_torch_callback","text":"TorchCallback.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callbacks.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to a list of Torch Callbacks — as_torch_callbacks","title":"Convert to a list of Torch Callbacks — as_torch_callbacks","text":"Converts object list TorchCallback.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callbacks.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to a list of Torch Callbacks — as_torch_callbacks","text":"","code":"as_torch_callbacks(x, clone, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callbacks.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to a list of Torch Callbacks — as_torch_callbacks","text":"x () Object convert. clone (logical(1)) Whether create deep clone. ... () Additional arguments.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callbacks.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert to a list of Torch Callbacks — as_torch_callbacks","text":"list() TorchCallbacks","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to TorchLoss — as_torch_loss","title":"Convert to TorchLoss — as_torch_loss","text":"Converts object TorchLoss.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to TorchLoss — as_torch_loss","text":"","code":"as_torch_loss(x, clone = FALSE, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to TorchLoss — as_torch_loss","text":"x () Object convert TorchLoss. clone (logical(1)) Whether make deep clone. ... () Additional arguments. Currently used pass additional constructor arguments TorchLoss objects type nn_loss.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert to TorchLoss — as_torch_loss","text":"TorchLoss.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_optimizer.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to TorchOptimizer — as_torch_optimizer","title":"Convert to TorchOptimizer — as_torch_optimizer","text":"Converts object TorchOptimizer.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_optimizer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to TorchOptimizer — as_torch_optimizer","text":"","code":"as_torch_optimizer(x, clone = FALSE, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_optimizer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to TorchOptimizer — as_torch_optimizer","text":"x () Object convert TorchOptimizer. clone (logical(1)) Whether make deep clone. Default FALSE. ... () Additional arguments. Currently used pass additional constructor arguments TorchOptimizer objects type torch_optimizer_generator.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_optimizer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert to TorchOptimizer — as_torch_optimizer","text":"TorchOptimizer","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/assert_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Assert Lazy Tensor — assert_lazy_tensor","title":"Assert Lazy Tensor — assert_lazy_tensor","text":"Asserts whether something lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/assert_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assert Lazy Tensor — assert_lazy_tensor","text":"","code":"assert_lazy_tensor(x)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/assert_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Assert Lazy Tensor — assert_lazy_tensor","text":"x () Object check.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/auto_device.html","id":null,"dir":"Reference","previous_headings":"","what":"Auto Device — auto_device","title":"Auto Device — auto_device","text":"First tries cuda, cpu.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/auto_device.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Auto Device — auto_device","text":"","code":"auto_device(device = NULL)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/auto_device.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Auto Device — auto_device","text":"device (character(1)) device. NULL, returned .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_categ.html","id":null,"dir":"Reference","previous_headings":"","what":"Batchgetter for Categorical data — batchgetter_categ","title":"Batchgetter for Categorical data — batchgetter_categ","text":"Converts data frame categorical data long tensor converting data integers. input checks performed.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_categ.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Batchgetter for Categorical data — batchgetter_categ","text":"","code":"batchgetter_categ(data, device, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_categ.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Batchgetter for Categorical data — batchgetter_categ","text":"data (data.table)data.table converted tensor. device (character(1)) device. ... () Unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_num.html","id":null,"dir":"Reference","previous_headings":"","what":"Batchgetter for Numeric Data — batchgetter_num","title":"Batchgetter for Numeric Data — batchgetter_num","text":"Converts data frame numeric data float tensor calling .matrix(). input checks performed","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_num.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Batchgetter for Numeric Data — batchgetter_num","text":"","code":"batchgetter_num(data, device, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_num.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Batchgetter for Numeric Data — batchgetter_num","text":"data (data.table())data.table converted tensor. device (character(1)) device tensor created. ... () Unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/callback_set.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Set of Callbacks for Torch — callback_set","title":"Create a Set of Callbacks for Torch — callback_set","text":"Creates R6ClassGenerator inheriting CallbackSet. Additionally performs checks stages accidentally misspelled. create TorchCallback use torch_callback(). order resulting class cloneable, private method $deep_clone() must provided.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/callback_set.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Set of Callbacks for Torch — callback_set","text":"","code":"callback_set(   classname,   on_begin = NULL,   on_end = NULL,   on_exit = NULL,   on_epoch_begin = NULL,   on_before_valid = NULL,   on_epoch_end = NULL,   on_batch_begin = NULL,   on_batch_end = NULL,   on_after_backward = NULL,   on_batch_valid_begin = NULL,   on_batch_valid_end = NULL,   on_valid_end = NULL,   state_dict = NULL,   load_state_dict = NULL,   initialize = NULL,   public = NULL,   private = NULL,   active = NULL,   parent_env = parent.frame(),   inherit = CallbackSet,   lock_objects = FALSE )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/callback_set.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Set of Callbacks for Torch — callback_set","text":"classname (character(1)) class name. on_begin, on_end, on_epoch_begin, on_before_valid, on_epoch_end, on_batch_begin, on_batch_end, on_after_backward, on_batch_valid_begin, on_batch_valid_end, on_valid_end, on_exit (function) Function execute given stage, see section Stages. state_dict (function()) function retrieves state dict callback. available learner training. load_state_dict (function(state_dict)) Function loads callback state. initialize (function()) initialization method callback. public, private, active (list()) Additional public, private, active fields add callback. parent_env (environment()) parent environment R6Class. inherit (R6ClassGenerator) class inherit. class must either CallbackSet (default) inherit . lock_objects (logical(1)) Whether lock objects resulting R6Class. FALSE (default), values can freely assigned self without declaring class definition.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/callback_set.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Set of Callbacks for Torch — callback_set","text":"CallbackSet","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/equals-.lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compare lazy tensors — ==.lazy_tensor","title":"Compare lazy tensors — ==.lazy_tensor","text":"Compares lazy tensors using indices data descriptor's hash. means two lazy_tensors: equal: mateterialize tensors. unequal: might materialize tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/equals-.lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compare lazy tensors — ==.lazy_tensor","text":"","code":"# S3 method for class 'lazy_tensor' x == y"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/equals-.lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compare lazy tensors — ==.lazy_tensor","text":"x, y (lazy_tensor) Values compare.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/is_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Check for lazy tensor — is_lazy_tensor","title":"Check for lazy tensor — is_lazy_tensor","text":"Checks whether object lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/is_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check for lazy tensor — is_lazy_tensor","text":"","code":"is_lazy_tensor(x)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/is_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Check for lazy tensor — is_lazy_tensor","text":"x () Object check.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a lazy tensor — lazy_tensor","title":"Create a lazy tensor — lazy_tensor","text":"Create lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a lazy tensor — lazy_tensor","text":"","code":"lazy_tensor(data_descriptor = NULL, ids = NULL)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a lazy tensor — lazy_tensor","text":"data_descriptor (DataDescriptor NULL) data descriptor NULL lazy tensor length 0. ids (integer()) elements data_descriptor included lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/lazy_tensor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a lazy tensor — lazy_tensor","text":"","code":"ds = dataset(\"example\",   initialize = function() self$iris = iris[, -5],   .getitem = function(i) list(x = torch_tensor(as.numeric(self$iris[i, ]))),   .length = function() nrow(self$iris) )() dd = as_data_descriptor(ds, list(x = c(NA, 4L))) lt = as_lazy_tensor(dd)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":null,"dir":"Reference","previous_headings":"","what":"Materialize Lazy Tensor Columns — materialize","title":"Materialize Lazy Tensor Columns — materialize","text":"materialize lazy_tensor() data.frame() / list() containing – among things – lazy_tensor() columns. .e. data described underlying DataDescriptors loaded indices lazy_tensor(), preprocessed put unto specified device. elements lazy tensor must shape, list tensors returned default. elements shape, tensors can also rbinded single tensor (parameter rbind).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Materialize Lazy Tensor Columns — materialize","text":"","code":"materialize(x, device = \"cpu\", rbind = FALSE, ...)  # S3 method for class 'list' materialize(x, device = \"cpu\", rbind = FALSE, cache = \"auto\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Materialize Lazy Tensor Columns — materialize","text":"x () object materialize. Either lazy_tensor list() / data.frame() containing lazy_tensor columns. device (character(1)) torch device. rbind (logical(1)) Whether rbind lazy tensor columns (TRUE) return list tensors (FALSE). second case, batch dimension. ... () Additional arguments. cache (character(1) environment() NULL) Optional cache (intermediate) materialization results. Per default, caching enabled dataset data descriptor (different output pointer) used one lazy tensor column.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Materialize Lazy Tensor Columns — materialize","text":"(list() lazy_tensors lazy_tensor)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Materialize Lazy Tensor Columns — materialize","text":"Materializing lazy tensor consists : Loading data internal dataset DataDescriptor. Processing batches preprocessing Graphs. Returning result PipeOp pointed DataDescriptor (pointer). multiple lazy_tensor columns can benefit caching : ) Output(s) dataset might input multiple graphs. b) Different lazy tensors might outputs graph. reason possible provide cache environment. hash key ) hash indices dataset. hash key b) hash indices, dataset preprocessing graph.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Materialize Lazy Tensor Columns — materialize","text":"","code":"lt1 = as_lazy_tensor(torch_randn(10, 3)) materialize(lt1, rbind = TRUE) #> torch_tensor #>  0.7209 -0.7791 -0.1200 #> -0.8322 -0.3782 -1.4658 #>  2.9278  0.1341 -2.9706 #>  0.4715  0.3795  2.1664 #>  0.3711 -1.0344 -0.5415 #>  1.6425 -0.0933  1.2480 #>  0.0744  0.1560 -0.1901 #>  1.9826 -1.4531 -0.8456 #>  0.5194  0.4390 -0.9260 #>  0.2598  0.4431 -0.5103 #> [ CPUFloatType{10,3} ] materialize(lt1, rbind = FALSE) #> [[1]] #> torch_tensor #>  0.7209 #> -0.7791 #> -0.1200 #> [ CPUFloatType{3} ] #>  #> [[2]] #> torch_tensor #> -0.8322 #> -0.3782 #> -1.4658 #> [ CPUFloatType{3} ] #>  #> [[3]] #> torch_tensor #>  2.9278 #>  0.1341 #> -2.9706 #> [ CPUFloatType{3} ] #>  #> [[4]] #> torch_tensor #>  0.4715 #>  0.3795 #>  2.1664 #> [ CPUFloatType{3} ] #>  #> [[5]] #> torch_tensor #>  0.3711 #> -1.0344 #> -0.5415 #> [ CPUFloatType{3} ] #>  #> [[6]] #> torch_tensor #>  1.6425 #> -0.0933 #>  1.2480 #> [ CPUFloatType{3} ] #>  #> [[7]] #> torch_tensor #>  0.0744 #>  0.1560 #> -0.1901 #> [ CPUFloatType{3} ] #>  #> [[8]] #> torch_tensor #>  1.9826 #> -1.4531 #> -0.8456 #> [ CPUFloatType{3} ] #>  #> [[9]] #> torch_tensor #>  0.5194 #>  0.4390 #> -0.9260 #> [ CPUFloatType{3} ] #>  #> [[10]] #> torch_tensor #>  0.2598 #>  0.4431 #> -0.5103 #> [ CPUFloatType{3} ] #>  lt2 = as_lazy_tensor(torch_randn(10, 4)) d = data.table::data.table(lt1 = lt1, lt2 = lt2) materialize(d, rbind = TRUE) #> $lt1 #> torch_tensor #>  0.7209 -0.7791 -0.1200 #> -0.8322 -0.3782 -1.4658 #>  2.9278  0.1341 -2.9706 #>  0.4715  0.3795  2.1664 #>  0.3711 -1.0344 -0.5415 #>  1.6425 -0.0933  1.2480 #>  0.0744  0.1560 -0.1901 #>  1.9826 -1.4531 -0.8456 #>  0.5194  0.4390 -0.9260 #>  0.2598  0.4431 -0.5103 #> [ CPUFloatType{10,3} ] #>  #> $lt2 #> torch_tensor #> -0.9499  1.2267  0.3966 -0.1104 #> -1.7213  0.0694  0.6292  0.4429 #>  0.1034 -0.2247 -1.0382 -0.8711 #>  1.3446 -0.7110 -0.2305  1.1702 #> -0.4673 -1.1430  0.0230  0.2525 #> -0.2598  0.9152 -1.4227 -0.0557 #> -2.2192  2.6817  1.2324  0.2709 #> -1.2812 -0.9407  0.8546  1.1632 #>  0.8758 -1.2456 -1.4916  0.2601 #> -0.5359 -0.7665  0.9438  0.1353 #> [ CPUFloatType{10,4} ] #>  materialize(d, rbind = FALSE) #> $lt1 #> $lt1[[1]] #> torch_tensor #>  0.7209 #> -0.7791 #> -0.1200 #> [ CPUFloatType{3} ] #>  #> $lt1[[2]] #> torch_tensor #> -0.8322 #> -0.3782 #> -1.4658 #> [ CPUFloatType{3} ] #>  #> $lt1[[3]] #> torch_tensor #>  2.9278 #>  0.1341 #> -2.9706 #> [ CPUFloatType{3} ] #>  #> $lt1[[4]] #> torch_tensor #>  0.4715 #>  0.3795 #>  2.1664 #> [ CPUFloatType{3} ] #>  #> $lt1[[5]] #> torch_tensor #>  0.3711 #> -1.0344 #> -0.5415 #> [ CPUFloatType{3} ] #>  #> $lt1[[6]] #> torch_tensor #>  1.6425 #> -0.0933 #>  1.2480 #> [ CPUFloatType{3} ] #>  #> $lt1[[7]] #> torch_tensor #>  0.0744 #>  0.1560 #> -0.1901 #> [ CPUFloatType{3} ] #>  #> $lt1[[8]] #> torch_tensor #>  1.9826 #> -1.4531 #> -0.8456 #> [ CPUFloatType{3} ] #>  #> $lt1[[9]] #> torch_tensor #>  0.5194 #>  0.4390 #> -0.9260 #> [ CPUFloatType{3} ] #>  #> $lt1[[10]] #> torch_tensor #>  0.2598 #>  0.4431 #> -0.5103 #> [ CPUFloatType{3} ] #>  #>  #> $lt2 #> $lt2[[1]] #> torch_tensor #> -0.9499 #>  1.2267 #>  0.3966 #> -0.1104 #> [ CPUFloatType{4} ] #>  #> $lt2[[2]] #> torch_tensor #> -1.7213 #>  0.0694 #>  0.6292 #>  0.4429 #> [ CPUFloatType{4} ] #>  #> $lt2[[3]] #> torch_tensor #>  0.1034 #> -0.2247 #> -1.0382 #> -0.8711 #> [ CPUFloatType{4} ] #>  #> $lt2[[4]] #> torch_tensor #>  1.3446 #> -0.7110 #> -0.2305 #>  1.1702 #> [ CPUFloatType{4} ] #>  #> $lt2[[5]] #> torch_tensor #> -0.4673 #> -1.1430 #>  0.0230 #>  0.2525 #> [ CPUFloatType{4} ] #>  #> $lt2[[6]] #> torch_tensor #> -0.2598 #>  0.9152 #> -1.4227 #> -0.0557 #> [ CPUFloatType{4} ] #>  #> $lt2[[7]] #> torch_tensor #> -2.2192 #>  2.6817 #>  1.2324 #>  0.2709 #> [ CPUFloatType{4} ] #>  #> $lt2[[8]] #> torch_tensor #> -1.2812 #> -0.9407 #>  0.8546 #>  1.1632 #> [ CPUFloatType{4} ] #>  #> $lt2[[9]] #> torch_tensor #>  0.8758 #> -1.2456 #> -1.4916 #>  0.2601 #> [ CPUFloatType{4} ] #>  #> $lt2[[10]] #> torch_tensor #> -0.5359 #> -0.7665 #>  0.9438 #>  0.1353 #> [ CPUFloatType{4} ] #>  #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":null,"dir":"Reference","previous_headings":"","what":"Materialize a Lazy Tensor — materialize_internal","title":"Materialize a Lazy Tensor — materialize_internal","text":"Convert lazy_tensor torch_tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Materialize a Lazy Tensor — materialize_internal","text":"","code":"materialize_internal(x, device = \"cpu\", cache = NULL, rbind)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Materialize a Lazy Tensor — materialize_internal","text":"x (lazy_tensor()) lazy tensor materialize. device (character(1L)) device put materialized tensor (running preprocessing graph). cache (NULL environment()) Whether cache (intermediate) results materialization. can make data loading faster multiple lazy_tensors reference dataset graph. rbind (logical(1)) Whtether rbind resulting tensors (TRUE) return list tensors (FALSE).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Materialize a Lazy Tensor — materialize_internal","text":"lazy_tensor()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Materialize a Lazy Tensor — materialize_internal","text":"Materializing lazy tensor consists : Loading data internal dataset DataDescriptor. Processing batches preprocessing Graphs. Returning result PipeOp pointed DataDescriptor (pointer). materializing multiple lazy_tensor columns, caching can useful : ) Output(s) dataset might input multiple graphs. (task_dataset shoudl rarely case try merge ). b) Different lazy tensors might outputs graph. reason possible provide cache environment. hash key ) hash indices dataset. hash key b) hash indices dataset preprocessing graph.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch-package.html","id":null,"dir":"Reference","previous_headings":"","what":"mlr3torch: Deep Learning with 'mlr3' — mlr3torch-package","title":"mlr3torch: Deep Learning with 'mlr3' — mlr3torch-package","text":"Deep Learning library extends mlr3 framework building upon 'torch' package. allows conveniently build, train, evaluate deep learning models without worry low level details. Custom architectures can created using graph language defined 'mlr3pipelines'.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch-package.html","id":"options","dir":"Reference","previous_headings":"","what":"Options","title":"mlr3torch: Deep Learning with 'mlr3' — mlr3torch-package","text":"mlr3torch.cache: Whether cache downloaded data (TRUE) (FALSE, default). can also set specific folder file system used cache directory.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch-package.html","id":"author","dir":"Reference","previous_headings":"","what":"Author","title":"mlr3torch: Deep Learning with 'mlr3' — mlr3torch-package","text":"Maintainer: Sebastian Fischer sebf.fischer@gmail.com (ORCID) Authors: Martin Binder mlr.developer@mb706.com contributors: Bernd Bischl bernd_bischl@gmx.net (ORCID) [contributor] Lukas Burk github@quantenbrot.de (ORCID) [contributor] Florian Pfisterer pfistererf@googlemail.com (ORCID) [contributor] Carson Zhang carsonzhang4@gmail.com [contributor]","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_callbacks.html","id":null,"dir":"Reference","previous_headings":"","what":"Dictionary of Torch Callbacks — mlr3torch_callbacks","title":"Dictionary of Torch Callbacks — mlr3torch_callbacks","text":"mlr3misc::Dictionary torch callbacks. Use t_clbk() conveniently retrieve callbacks. Can converted data.table using .data.table.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_callbacks.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dictionary of Torch Callbacks — mlr3torch_callbacks","text":"","code":"mlr3torch_callbacks"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_callbacks.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Dictionary of Torch Callbacks — mlr3torch_callbacks","text":"object class DictionaryMlr3torchCallbacks (inherits Dictionary, R6) length 13.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_callbacks.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Dictionary of Torch Callbacks — mlr3torch_callbacks","text":"","code":"mlr3torch_callbacks$get(\"checkpoint\") #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: list() #> * Packages: mlr3torch,torch # is the same as t_clbk(\"checkpoint\") #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: list() #> * Packages: mlr3torch,torch # convert to a data.table as.data.table(mlr3torch_callbacks) #> Key: <key> #>           key       label                 packages #>        <char>      <char>                   <char> #> 1: checkpoint  Checkpoint          mlr3torch,torch #> 2:    history     History          mlr3torch,torch #> 3:   progress    Progress progress,mlr3torch,torch #> 4:         tb TensorBoard tfevents,mlr3torch,torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss Functions — mlr3torch_losses","title":"Loss Functions — mlr3torch_losses","text":"Dictionary torch loss descriptors. See t_loss() conveniently retrieving loss function. Can converted data.table using .data.table.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss Functions — mlr3torch_losses","text":"","code":"mlr3torch_losses"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Loss Functions — mlr3torch_losses","text":"object class DictionaryMlr3torchLosses (inherits Dictionary, R6) length 13.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":"available-loss-functions","dir":"Reference","previous_headings":"","what":"Available Loss Functions","title":"Loss Functions — mlr3torch_losses","text":"cross_entropy, l1, mse","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loss Functions — mlr3torch_losses","text":"","code":"mlr3torch_losses$get(\"mse\") #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr # is equivalent to t_loss(\"mse\") #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr # convert to a data.table as.data.table(mlr3torch_losses) #> Key: <key> #>              key              label task_types        packages #>           <char>             <char>     <list>          <char> #> 1: cross_entropy      Cross Entropy    classif torch,mlr3torch #> 2:            l1     Absolute Error       regr torch,mlr3torch #> 3:           mse Mean Squared Error       regr torch,mlr3torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizers — mlr3torch_optimizers","title":"Optimizers — mlr3torch_optimizers","text":"Dictionary torch optimizers. Use t_opt conveniently retrieving optimizers. Can converted data.table using .data.table.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizers — mlr3torch_optimizers","text":"","code":"mlr3torch_optimizers"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Optimizers — mlr3torch_optimizers","text":"object class DictionaryMlr3torchOptimizers (inherits Dictionary, R6) length 13.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":"available-optimizers","dir":"Reference","previous_headings":"","what":"Available Optimizers","title":"Optimizers — mlr3torch_optimizers","text":"adadelta, adagrad, adam, asgd, rmsprop, rprop, sgd","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Optimizers — mlr3torch_optimizers","text":"","code":"mlr3torch_optimizers$get(\"adam\") #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # is equivalent to t_opt(\"adam\") #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # convert to a data.table as.data.table(mlr3torch_optimizers) #> Key: <key> #>         key                         label        packages #>      <char>                        <char>          <char> #> 1: adadelta Adaptive Learning Rate Method torch,mlr3torch #> 2:  adagrad   Adaptive Gradient algorithm torch,mlr3torch #> 3:     adam    Adaptive Moment Estimation torch,mlr3torch #> 4:     asgd  SGD with Adaptive Batch Size torch,mlr3torch #> 5:  rmsprop  Root Mean Square Propagation torch,mlr3torch #> 6:    rprop     Resilient Backpropagation torch,mlr3torch #> 7:      sgd   Stochastic Gradient Descent torch,mlr3torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":null,"dir":"Reference","previous_headings":"","what":"Lazy Data Backend — mlr_backends_lazy","title":"Lazy Data Backend — mlr_backends_lazy","text":"lazy data backend wraps constructor lazily creates another backend, e.g. downloading (caching) data internet. backend used, metadata backend known advance accessible downloading actual data. backend first constructed, verified provided metadata correct, otherwise informative error message thrown. construction lazily constructed backend, calls like $data(), $missings(), $distinct(), $hash() redirected . Information available backend constructed : nrow - number rows (set length rownames). ncol - number columns (provided via id column col_info). colnames - column names. rownames - row names. col_info - column information, can obtained via mlr3::col_info(). Beware accessing backend's hash also contructs backend. Note cases data contains lazy_tensor columns, necessary naming class nothing lazy_tensor data type. Important constructor generates factor() variables important ordering levels data corresponds ordering levels col_info argument.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Lazy Data Backend — mlr_backends_lazy","text":"mlr3::DataBackend -> DataBackendLazy","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Lazy Data Backend — mlr_backends_lazy","text":"backend (DataBackend) wrapped backend lazily constructed first accessed. nrow (integer(1)) Number rows (observations). ncol (integer(1)) Number columns (variables), including primary key column. rownames (integer()) Returns vector distinct row identifiers, .e. contents primary key column. colnames (character()) Returns vector column names, including primary key column. is_constructed (logical(1)) Whether backend already constructed.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Lazy Data Backend — mlr_backends_lazy","text":"mlr3::DataBackend$format()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Lazy Data Backend — mlr_backends_lazy","text":"DataBackendLazy$new() DataBackendLazy$data() DataBackendLazy$head() DataBackendLazy$distinct() DataBackendLazy$missings() DataBackendLazy$print()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$new(constructor, rownames, col_info, primary_key)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"constructor (function) function argument backend (lazy backend), whose return value must actual backend. function called first time field $backend accessed. rownames (integer()) row names. Must permutation rownames lazily constructed backend. col_info (data.table::data.table()) data.table columns id, type levels containing column id, type levels. Note levels must provided correct order. primary_key (character(1)) Name primary key column.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-data-","dir":"Reference","previous_headings":"","what":"Method data()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Returns slice data specified format. rows must addressed vector primary key values, columns must referred via column names. Queries rows matching row id queries columns matching column name silently ignored. Rows guaranteed returned order rows, columns may returned arbitrary order. Duplicated row ids result duplicated rows, duplicated column names lead exception. Accessing data triggers construction backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$data(rows, cols)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"rows (integer()) Row indices. cols (character()) Column names.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-head-","dir":"Reference","previous_headings":"","what":"Method head()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Retrieve first n rows. triggers construction backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$head(n = 6L)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"n (integer(1)) Number rows.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Lazy Data Backend — mlr_backends_lazy","text":"data.table::data.table() first n rows.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-distinct-","dir":"Reference","previous_headings":"","what":"Method distinct()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Returns named list vectors distinct values column specified. na_rm TRUE, missing values removed returned vectors distinct values. Non-existing rows columns silently ignored. triggers construction backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$distinct(rows, cols, na_rm = TRUE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments-3","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"rows (integer()) Row indices. cols (character()) Column names. na_rm (logical(1)) Whether remove NAs .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"returns-1","dir":"Reference","previous_headings":"","what":"Returns","title":"Lazy Data Backend — mlr_backends_lazy","text":"Named list() distinct values.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-missings-","dir":"Reference","previous_headings":"","what":"Method missings()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Returns number missing values per column specified slice data. Non-existing rows columns silently ignored. triggers construction backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$missings(rows, cols)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments-4","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"rows (integer()) Row indices. cols (character()) Column names.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"returns-2","dir":"Reference","previous_headings":"","what":"Returns","title":"Lazy Data Backend — mlr_backends_lazy","text":"Total missing values per column (named numeric()).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Printer.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$print()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"# We first define a backend constructor constructor = function(backend) {   cat(\"Data is constructed!\\n\")   DataBackendDataTable$new(     data.table(x = rnorm(10), y = rnorm(10), row_id = 1:10),     primary_key = \"row_id\"   ) }  # to wrap this backend constructor in a lazy backend, we need to provide the correct metadata for it column_info = data.table(   id = c(\"x\", \"y\", \"row_id\"),   type = c(\"numeric\", \"numeric\", \"integer\"),   levels = list(NULL, NULL, NULL) ) backend_lazy = DataBackendLazy$new(   constructor = constructor,   rownames = 1:10,   col_info = column_info,   primary_key = \"row_id\" )  # Note that the constructor is not called for the calls below # as they can be read from the metadata backend_lazy$nrow #> [1] 10 backend_lazy$rownames #>  [1]  1  2  3  4  5  6  7  8  9 10 backend_lazy$ncol #> [1] 3 backend_lazy$colnames #> [1] \"x\"      \"y\"      \"row_id\" col_info(backend_lazy) #>        id    type levels #>    <char>  <char> <list> #> 1:      x numeric [NULL] #> 2:      y numeric [NULL] #> 3: row_id integer [NULL]  # Only now the backend is constructed backend_lazy$data(1, \"x\") #> Data is constructed! #>             x #>         <num> #> 1: -0.3872136 # Is the same as: backend_lazy$backend$data(1, \"x\") #>             x #>         <num> #> 1: -0.3872136"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":null,"dir":"Reference","previous_headings":"","what":"Checkpoint Callback — mlr_callback_set.checkpoint","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saves optimizer network states training. final network optimizer always stored.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saving learner callback trained model impossible, model slot set last callback step executed.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"mlr3torch::CallbackSet -> CallbackSetCheckpoint","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"CallbackSetCheckpoint$new() CallbackSetCheckpoint$on_epoch_end() CallbackSetCheckpoint$on_batch_end() CallbackSetCheckpoint$on_exit() CallbackSetCheckpoint$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$new(path, freq, freq_type = \"epoch\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"path (character(1)) path folder models saved. freq (integer(1)) frequency often model saved. Frequency either per step epoch, can configured freq_type parameter. freq_type (character(1)) Can either \"epoch\" (default) \"step\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-on-epoch-end-","dir":"Reference","previous_headings":"","what":"Method on_epoch_end()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saves network optimizer state dict. nothing freq_type freq met.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$on_epoch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-on-batch-end-","dir":"Reference","previous_headings":"","what":"Method on_batch_end()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saves selected objects defined save. nothing freq_type freq met.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$on_batch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-on-exit-","dir":"Reference","previous_headings":"","what":"Method on_exit()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saves learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$on_exit()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":null,"dir":"Reference","previous_headings":"","what":"History Callback — mlr_callback_set.history","title":"History Callback — mlr_callback_set.history","text":"Saves training validation history training. history saved data.table $train $valid slots. first column always epoch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"History Callback — mlr_callback_set.history","text":"mlr3torch::CallbackSet -> CallbackSetHistory","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"History Callback — mlr_callback_set.history","text":"mlr3torch::CallbackSet$print()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"History Callback — mlr_callback_set.history","text":"CallbackSetHistory$on_begin() CallbackSetHistory$state_dict() CallbackSetHistory$load_state_dict() CallbackSetHistory$on_before_valid() CallbackSetHistory$on_epoch_end() CallbackSetHistory$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-on-begin-","dir":"Reference","previous_headings":"","what":"Method on_begin()","title":"History Callback — mlr_callback_set.history","text":"Initializes lists train validation metrics stored.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$on_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-state-dict-","dir":"Reference","previous_headings":"","what":"Method state_dict()","title":"History Callback — mlr_callback_set.history","text":"Converts lists data.tables.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$state_dict()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-load-state-dict-","dir":"Reference","previous_headings":"","what":"Method load_state_dict()","title":"History Callback — mlr_callback_set.history","text":"Sets field $train $valid contained state dict.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$load_state_dict(state_dict)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"History Callback — mlr_callback_set.history","text":"state_dict (callback_state_history) state dict retrieved via $state_dict().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-on-before-valid-","dir":"Reference","previous_headings":"","what":"Method on_before_valid()","title":"History Callback — mlr_callback_set.history","text":"Add latest training scores history.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$on_before_valid()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-on-epoch-end-","dir":"Reference","previous_headings":"","what":"Method on_epoch_end()","title":"History Callback — mlr_callback_set.history","text":"Add latest validation scores history.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$on_epoch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"History Callback — mlr_callback_set.history","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"History Callback — mlr_callback_set.history","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Callbacks — mlr_callback_set","title":"Base Class for Callbacks — mlr_callback_set","text":"Base class callbacks inherit (see section Inheriting). callback set collection functions executed different stages training loop. can used gain control training process neural network without write everything scratch. used torch learner, CallbackSet wrapped TorchCallback. latters parameter set represents arguments CallbackSet's $initialize() method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Callbacks — mlr_callback_set","text":"available stage (see section Stages) public method $on_<stage>() can defined. evaluation context (ContextTorch) can accessed via self$ctx, contains current state training loop. context assigned beginning training loop removed afterwards. Different stages callback can communicate assigning values $self. State: able store information $model slot LearnerTorch, callbacks support state API. can overload $state_dict() public method define stored learner$model$callbacks$<id> training finishes. also requires implement $load_state_dict(state_dict) method defines load previously saved callback state different callback. Note $state_dict() include parameter values used initialize callback. creating custom callbacks, function torch_callback() recommended, creates CallbackSet wraps TorchCallback. create CallbackSet convenience function callback_set() can used. functions perform checks stages accidentally misspelled.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"stages","dir":"Reference","previous_headings":"","what":"Stages","title":"Base Class for Callbacks — mlr_callback_set","text":"begin :: Run training loop begins. epoch_begin :: Run beginning epoch. batch_begin :: Run forward call. after_backward :: Run backward call. batch_end :: Run optimizer step. batch_valid_begin :: Run forward call validation loop. batch_valid_end :: Run forward call validation loop. valid_end :: Run end validation. epoch_end :: Run end epoch. end :: Run last epoch. exit :: Run last, using .exit().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"terminate-training","dir":"Reference","previous_headings":"","what":"Terminate Training","title":"Base Class for Callbacks — mlr_callback_set","text":"training stopped, possible set field $terminate ContextTorch. end every epoch field checked TRUE, training stops. can example used implement custom early stopping.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Base Class for Callbacks — mlr_callback_set","text":"ctx (ContextTorch NULL) evaluation context callback. field always NULL except $train() call torch learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Callbacks — mlr_callback_set","text":"stages (character()) active stages callback set.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Callbacks — mlr_callback_set","text":"CallbackSet$print() CallbackSet$state_dict() CallbackSet$load_state_dict() CallbackSet$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Callbacks — mlr_callback_set","text":"Prints object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Callbacks — mlr_callback_set","text":"","code":"CallbackSet$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Callbacks — mlr_callback_set","text":"... () Currently unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"method-state-dict-","dir":"Reference","previous_headings":"","what":"Method state_dict()","title":"Base Class for Callbacks — mlr_callback_set","text":"Returns information kept LearnerTorch's state training. information loadable callback using $load_state_dict() able continue training. returns NULL default.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Callbacks — mlr_callback_set","text":"","code":"CallbackSet$state_dict()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"method-load-state-dict-","dir":"Reference","previous_headings":"","what":"Method load_state_dict()","title":"Base Class for Callbacks — mlr_callback_set","text":"Loads state dict callback continue training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Callbacks — mlr_callback_set","text":"","code":"CallbackSet$load_state_dict(state_dict)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Callbacks — mlr_callback_set","text":"state_dict () state dict retrieved via $state_dict().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Callbacks — mlr_callback_set","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Callbacks — mlr_callback_set","text":"","code":"CallbackSet$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Callbacks — mlr_callback_set","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":null,"dir":"Reference","previous_headings":"","what":"Progress Callback — mlr_callback_set.progress","title":"Progress Callback — mlr_callback_set.progress","text":"Prints progress bar metrics training validation.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Progress Callback — mlr_callback_set.progress","text":"mlr3torch::CallbackSet -> CallbackSetProgress","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Progress Callback — mlr_callback_set.progress","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Progress Callback — mlr_callback_set.progress","text":"CallbackSetProgress$on_epoch_begin() CallbackSetProgress$on_batch_end() CallbackSetProgress$on_before_valid() CallbackSetProgress$on_batch_valid_end() CallbackSetProgress$on_epoch_end() CallbackSetProgress$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-epoch-begin-","dir":"Reference","previous_headings":"","what":"Method on_epoch_begin()","title":"Progress Callback — mlr_callback_set.progress","text":"Initializes progress bar training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_epoch_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-batch-end-","dir":"Reference","previous_headings":"","what":"Method on_batch_end()","title":"Progress Callback — mlr_callback_set.progress","text":"Increments training progress bar.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_batch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-before-valid-","dir":"Reference","previous_headings":"","what":"Method on_before_valid()","title":"Progress Callback — mlr_callback_set.progress","text":"Creates progress bar validation.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_before_valid()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-batch-valid-end-","dir":"Reference","previous_headings":"","what":"Method on_batch_valid_end()","title":"Progress Callback — mlr_callback_set.progress","text":"Increments validation progress bar.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_batch_valid_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-epoch-end-","dir":"Reference","previous_headings":"","what":"Method on_epoch_end()","title":"Progress Callback — mlr_callback_set.progress","text":"Prints summary training validation process.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_epoch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Progress Callback — mlr_callback_set.progress","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Progress Callback — mlr_callback_set.progress","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":null,"dir":"Reference","previous_headings":"","what":"TensorBoard Logging Callback — mlr_callback_set.tb","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"Logs training loss, training measures, validation measures events. view , use TensorBoard tensorflow::tensorboard() (requires tensorflow) CLI.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"Logs events every epoch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"mlr3torch::CallbackSet -> CallbackSetTB","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"CallbackSetTB$new() CallbackSetTB$on_epoch_end() CallbackSetTB$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"","code":"CallbackSetTB$new(path, log_train_loss)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"path (character(1)) path folder events logged. Point TensorBoard folder view . log_train_loss (logical(1)) Whether log training loss.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"method-on-epoch-end-","dir":"Reference","previous_headings":"","what":"Method on_epoch_end()","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"Logs training loss, training measures, validation measures TensorBoard events.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"","code":"CallbackSetTB$on_epoch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"","code":"CallbackSetTB$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Context for Torch Learner — mlr_context_torch","title":"Context for Torch Learner — mlr_context_torch","text":"Context training torch learner. - mostly read-- information callbacks access argument ctx. information callbacks, see CallbackSet.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Context for Torch Learner — mlr_context_torch","text":"learner (Learner) torch learner. task_train (Task) training task. task_valid (Task NULL) validation task. loader_train (torch::dataloader) data loader training. loader_valid (torch::dataloader) data loader validation. measures_train (list() Measures) Measures used training. measures_valid (list() Measures) Measures used validation. network (torch::nn_module) torch network. optimizer (torch::optimizer) optimizer. loss_fn (torch::nn_module) loss function. total_epochs (integer(1)) total number epochs learner trained . last_scores_train (named list() NULL) scores last training batch. Names ids training measures. LearnerTorch sets eval_freq different 1, NULL epochs evaluate model. last_scores_valid (list()) scores last validation batch. Names ids validation measures. LearnerTorch sets eval_freq different 1, NULL epochs evaluate model. epoch (integer(1)) current epoch. step (integer(1)) current iteration. prediction_encoder (function()) learner's prediction encoder. batch (named list() torch_tensors) current batch. terminate (logical(1)) field set TRUE end epoch, training stops.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Context for Torch Learner — mlr_context_torch","text":"ContextTorch$new() ContextTorch$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Context for Torch Learner — mlr_context_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Context for Torch Learner — mlr_context_torch","text":"","code":"ContextTorch$new(   learner,   task_train,   task_valid = NULL,   loader_train,   loader_valid = NULL,   measures_train = NULL,   measures_valid = NULL,   network,   optimizer,   loss_fn,   total_epochs,   prediction_encoder,   eval_freq = 1L )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Context for Torch Learner — mlr_context_torch","text":"learner (Learner) torch learner. task_train (Task) training task. task_valid (Task NULL) validation task. loader_train (torch::dataloader) data loader training. loader_valid (torch::dataloader NULL) data loader validation. measures_train (list() Measures NULL) Measures used training. Default NULL. measures_valid (list() Measures NULL) Measures used validation. network (torch::nn_module) torch network. optimizer (torch::optimizer) optimizer. loss_fn (torch::nn_module) loss function. total_epochs (integer(1)) total number epochs learner trained . prediction_encoder (function()) learner's prediction encoder. eval_freq (integer(1)) evaluation frequency.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Context for Torch Learner — mlr_context_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Context for Torch Learner — mlr_context_torch","text":"","code":"ContextTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Context for Torch Learner — mlr_context_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":null,"dir":"Reference","previous_headings":"","what":"My Little Pony — mlr_learners.mlp","title":"My Little Pony — mlr_learners.mlp","text":"Fully connected feed forward network dropout activation function. features can either single lazy_tensor one numeric columns ().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"My Little Pony — mlr_learners.mlp","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.mlp\", ...) lrn(\"regr.mlp\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"My Little Pony — mlr_learners.mlp","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “integer”, “numeric”, “lazy_tensor” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"My Little Pony — mlr_learners.mlp","text":"Parameters LearnerTorch, well : activation :: [nn_module] activation function. initialized nn_relu. activation_args :: named list() named list initialization arguments activation function. intialized empty list. neurons :: integer() number neurons per hidden layer. default hidden layer. Setting c(10, 20) first hidden layer 10 neurons second 20. n_layers :: integer() number layers. parameter must set neurons length 1. p :: numeric(1) dropout probability. initialized 0.5. shape :: integer() NULL input shape length 2, e.g. c(NA, 5). needs present lazy tensor input unknown shape (NULL). Otherwise input shape inferred number numeric features.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"My Little Pony — mlr_learners.mlp","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchMLP","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"My Little Pony — mlr_learners.mlp","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"My Little Pony — mlr_learners.mlp","text":"LearnerTorchMLP$new() LearnerTorchMLP$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"My Little Pony — mlr_learners.mlp","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"My Little Pony — mlr_learners.mlp","text":"","code":"LearnerTorchMLP$new(   task_type,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"My Little Pony — mlr_learners.mlp","text":"task_type (character(1)) task type, either \"classif\" \"regr\". optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"My Little Pony — mlr_learners.mlp","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"My Little Pony — mlr_learners.mlp","text":"","code":"LearnerTorchMLP$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"My Little Pony — mlr_learners.mlp","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"My Little Pony — mlr_learners.mlp","text":"","code":"# Define the Learner and set parameter values learner = lrn(\"classif.mlp\") learner$param_set$set_values(   epochs = 1, batch_size = 16, device = \"cpu\",   neurons = 10 )  # Define a Task task = tsk(\"iris\")  # Create train and test set ids = partition(task)  # Train the learner on the training ids learner$train(task, row_ids = ids$train)  # Make predictions for the test rows predictions = learner$predict(task, row_ids = ids$test)  # Score the predictions predictions$score() #> classif.ce  #>       0.68"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":null,"dir":"Reference","previous_headings":"","what":"Tabular ResNet — mlr_learners.tab_resnet","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Tabular resnet.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.tab_resnet\", ...) lrn(\"regr.tab_resnet\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “integer”, “numeric” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Parameters LearnerTorch, well : n_blocks :: integer(1) number blocks. d_block :: integer(1) input output dimension block. d_hidden :: integer(1) latent dimension block. d_hidden_multiplier :: numeric(1) Alternative way specify latent dimension d_block * d_hidden_multiplier. dropout1 :: numeric(1) First dropout ratio. dropout2 :: numeric(1) Second dropout ratio.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Gorishniy Y, Rubachev , Khrulkov V, Babenko (2021). “Revisiting Deep Learning  Tabular Data.” arXiv, 2106.11959.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchTabResNet","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"LearnerTorchTabResNet$new() LearnerTorchTabResNet$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"LearnerTorchTabResNet$new(   task_type,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"task_type (character(1)) task type, either \"classif\" \"regr\". optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"LearnerTorchTabResNet$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"# Define the Learner and set parameter values learner = lrn(\"classif.tab_resnet\") learner$param_set$set_values(   epochs = 1, batch_size = 16, device = \"cpu\",   n_blocks = 2, d_block = 10, d_hidden = 20, dropout1 = 0.3, dropout2 = 0.3 )  # Define a Task task = tsk(\"iris\")  # Create train and test set ids = partition(task)  # Train the learner on the training ids learner$train(task, row_ids = ids$train)  # Make predictions for the test rows predictions = learner$predict(task, row_ids = ids$test)  # Score the predictions predictions$score() #> classif.ce  #>        0.9"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":null,"dir":"Reference","previous_headings":"","what":"Featureless Torch Learner — mlr_learners.torch_featureless","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"Featureless torch learner. Output constant weight learned training. classification, (asymptoptically) result majority class prediction using standard cross-entropy loss. regression, result median L1 loss mean L2 loss.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.torch_featureless\", ...) lrn(\"regr.torch_featureless\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “logical”, “integer”, “numeric”, “character”, “factor”, “ordered”, “POSIXct”, “lazy_tensor” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"LearnerTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchFeatureless","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"LearnerTorchFeatureless$new() LearnerTorchFeatureless$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"","code":"LearnerTorchFeatureless$new(   task_type,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"task_type (character(1)) task type, either \"classif\" \"regr\". optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"","code":"LearnerTorchFeatureless$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"","code":"# Define the Learner and set parameter values learner = lrn(\"classif.torch_featureless\") learner$param_set$set_values(   epochs = 1, batch_size = 16, device = \"cpu\"    )  # Define a Task task = tsk(\"iris\")  # Create train and test set ids = partition(task)  # Train the learner on the training ids learner$train(task, row_ids = ids$train)  # Make predictions for the test rows predictions = learner$predict(task, row_ids = ids$test)  # Score the predictions predictions$score() #> classif.ce  #>       0.64"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":null,"dir":"Reference","previous_headings":"","what":"AlexNet Image Classifier — mlr_learners.torchvision","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Classic image classification networks torchvision.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Parameters LearnerTorchImage pretrained :: logical(1) Whether use pretrained model. final linear layer replaced new nn_linear number classes inferred Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Supported task types: \"classif\" Predict Types: \"response\" \"prob\" Feature Types: \"lazy_tensor\" Required packages: \"mlr3torch\", \"torch\", \"torchvision\"","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> mlr3torch::LearnerTorchImage -> LearnerTorchVision","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"LearnerTorchVision$new() LearnerTorchVision$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"","code":"LearnerTorchVision$new(   name,   module_generator,   label,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"name (character(1)) name network. module_generator (function(pretrained, num_classes)) Function generates network. label (character(1)) label network. #' @references Krizhevsky, Alex, Sutskever, Ilya, Hinton, E. G (2017). “Imagenet classification deep convolutional neural networks.” Communications ACM, 60(6), 84–90. Sandler, Mark, Howard, Andrew, Zhu, Menglong, Zhmoginov, Andrey, Chen, Liang-Chieh (2018). “Mobilenetv2: Inverted residuals linear bottlenecks.” Proceedings IEEE conference computer vision pattern recognition, 4510–4520. , Kaiming, Zhang, Xiangyu, Ren, Shaoqing, Sun, Jian (2016). “Deep residual learning image recognition.” Proceedings IEEE conference computer vision pattern recognition, 770–778. Simonyan, Karen, Zisserman, Andrew (2014). “deep convolutional networks large-scale image recognition.” arXiv preprint arXiv:1409.1556. optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"","code":"LearnerTorchVision$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Learners — mlr_learners_torch","title":"Base Class for Torch Learners — mlr_learners_torch","text":"base class provides basic functionality training prediction neural network. torch learners inherit class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"validation","dir":"Reference","previous_headings":"","what":"Validation","title":"Base Class for Torch Learners — mlr_learners_torch","text":"specify validation data, can set $validate field Learner, can set : NULL: validation ratio: proportion 1 - ratio task used training ratio used validation. \"test\" means \"test\" task resampling used possible calling $train() manually. \"predefined\": use predefined $internal_valid_task mlr3::Task. validation data can also used early stopping, see description Learner's parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"saving-a-learner","dir":"Reference","previous_headings":"","what":"Saving a Learner","title":"Base Class for Torch Learners — mlr_learners_torch","text":"order save LearnerTorch later usage, necessary call $marshal() method Learner writing disk, object otherwise saved correctly. loading marshaled LearnerTorch R , need call $unmarshal() transform useable state.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"early-stopping-and-internal-tuning","dir":"Reference","previous_headings":"","what":"Early Stopping and Internal Tuning","title":"Base Class for Torch Learners — mlr_learners_torch","text":"order prevent overfitting, LearnerTorch class allows use early stopping via patience min_delta parameters, see Learner's parameters. tuning LearnerTorch also possible combine explicit tuning via mlr3tuning LearnerTorch's internal tuning epochs via early stopping. , just need include epochs = to_tune(upper = <upper>, internal = TRUE) search space, <upper> maximally allowed number epochs, configure early stopping.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"model","dir":"Reference","previous_headings":"","what":"Model","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Model list class \"learner_torch_model\" following elements: network :: trained network. optimizer :: $state_dict() optimizer used train network. loss_fn :: $state_dict() loss used train network. callbacks :: callbacks used train network. seed :: seed / used training prediction. epochs :: many epochs model trained (early stopping). task_col_info :: data.table() containing information train-task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Learners — mlr_learners_torch","text":"General: parameters optimizer, loss callbacks, prefixed \"opt.\", \"loss.\" \"cb.<callback id>.\" respectively, well : epochs :: integer(1) number epochs. device :: character(1) device. One \"auto\", \"cpu\", \"cuda\" values defined mlr_reflections$torch$devices. value initialized \"auto\", select \"cuda\" possible, try \"mps\" otherwise fall back \"cpu\". num_threads :: integer(1) number threads intraop pararallelization (device \"cpu\"). value initialized 1. num_interop_threads :: integer(1) number threads intraop interop pararallelization (device \"cpu\"). value initialized 1. Note can set session changing value within R session raise warning. seed :: integer(1) \"random\" NULL torch seed used training prediction. value initialized \"random\", means random seed sampled beginning training phase. seed (either set randomly sampled) available via $model$seed training used prediction. Note setting seed training phase mean default (.e. seed \"random\"), clones learner use different seed. set NULL, seeding done. Evaluation: measures_train :: Measure list() Measures. Measures evaluated training. measures_valid :: Measure list() Measures. Measures evaluated validation. eval_freq :: integer(1) often train / validation predictions evaluated using measures_train / measures_valid. initialized 1. Note final model always evaluated. Early Stopping: patience :: integer(1) activates early stopping using validation scores. performance model improve patience evaluation steps, training ended. Note final model stored learner, best model. initialized 0, means early stopping. first entry measures_valid used metric. also requires specify $validate field Learner, well measures_valid. min_delta :: double(1) minimum improvement threshold (>) early stopping. initialized 0. Dataloader: batch_size :: integer(1) batch size (required). shuffle :: logical(1) Whether shuffle instances dataset. Default FALSE. impact validation. sampler :: torch::sampler Object defines dataloader draw samples. batch_sampler :: torch::sampler Object defines dataloader draws batches. num_workers :: integer(1) number workers data loading (batches loaded parallel). default 0, means data loaded main process. collate_fn :: function merge list samples form batch. pin_memory :: logical(1) Whether dataloader copies tensors CUDA pinned memory returning . drop_last :: logical(1) Whether drop last training batch epoch training. Default FALSE. timeout :: numeric(1) timeout value collecting batch workers. Negative values mean timeout default -1. worker_init_fn :: function(id) function receives worker id ([1, num_workers]) exectued seeding worker data loading. worker_globals :: list() | character() loading data parallel, allows export globals workers. character vector, objects global environment names copied workers. worker_packages :: character() packages load workers. Also see torch::dataloder information.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Torch Learners — mlr_learners_torch","text":"seperate classes classification regression inherit . Instead, task_type must specified  construction argument. Currently, classification regression supported. inheriting class, one overload two private methods: .network(task, param_vals) (Task, list()) -> nn_module Construct torch::nn_module object given task parameter values, .e. neural network trained learner. classification, output network expected scores application final softmax layer. .dataset(task, param_vals) (Task, list()) -> torch::dataset Create dataset task. Must respect parameter value device. Moreover, one needs pay attention respect row ids provided task. also possible overwrite private .dataloader() method instead .dataset() method. Per default, dataloader constructed using output .dataset() method. However, respect dataloader parameters ParamSet. .dataloader(task, param_vals) (Task, list()) -> torch::dataloader Create dataloader task. Needs respect least batch_size shuffle (otherwise predictions can permuted). change predict types, private .encode_prediction() method can overwritten: .encode_prediction(predict_tensor, task, param_vals) (torch_tensor, Task, list()) -> list() Take raw predictions self$network (predict_tensor) encode format can converted valid mlr3 predictions using mlr3::as_prediction_data(). method must take self$predict_type account. possible add parameters specifying param_set construction argument, currently possible remove existing parameters, .e. listed section Parameters. None parameters provided param_set can id starts \"loss.\", \"opt.\", \"cb.\"`, preserved dynamically constructed parameters optimizer, loss function, callbacks. perform additional input checks task, private .verify_train_task(task, param_vals) .verify_predict_task(task, param_vals) can overwritten. learners construction arguments change hash learner, required implement private $.additional_phash_input().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Base Class for Torch Learners — mlr_learners_torch","text":"mlr3::Learner -> LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Torch Learners — mlr_learners_torch","text":"validate construct internal validation data. parameter can either NULL, ratio $(0, 1)$, \"test\", \"predefined\". loss (TorchLoss) torch loss. optimizer (TorchOptimizer) torch optimizer. callbacks (list() TorchCallbacks) List torch callbacks. ids set names. internal_valid_scores Retrieves internal validation scores named list(). Specify $validate field measures_valid parameter configure . Returns NULL learner trained yet. internal_tuned_values early stopping activate, returns named list early-stopped epochs, otherwise empty list returned. Returns NULL learner trained yet. marshaled (logical(1)) Whether learner marshaled. network (nn_module()) Shortcut learner$model$network. param_set (ParamSet) parameter set hash (character(1)) Hash (unique identifier) object. phash (character(1)) Hash (unique identifier) partial object, excluding components varied systematically tuning (parameter values).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Base Class for Torch Learners — mlr_learners_torch","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Torch Learners — mlr_learners_torch","text":"LearnerTorch$new() LearnerTorch$format() LearnerTorch$print() LearnerTorch$marshal() LearnerTorch$unmarshal() LearnerTorch$dataset() LearnerTorch$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$new(   id,   task_type,   param_set,   properties,   man,   label,   feature_types,   optimizer = NULL,   loss = NULL,   packages = character(),   predict_types = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"id (character(1)) id new object. task_type (character(1)) task type. param_set (ParamSet alist()) Either parameter set, alist() containing different values self, e.g. alist(private$.param_set1, private$.param_set2), ParamSet collection created. properties (character()) properties object. See mlr_reflections$learner_properties available values. man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). label (character(1)) Label new instance. feature_types (character()) feature types. See mlr_reflections$task_feature_types available values, Additionally, \"lazy_tensor\" supported. optimizer (NULL TorchOptimizer) optimizer use training. Defaults adam. loss (NULL TorchLoss) loss use training. Defaults MSE regression cross entropy classification. packages (character()) R packages object depends . predict_types (character()) predict types. See mlr_reflections$learner_predict_types available values. regression, default \"response\". classification, defaults \"response\" \"prob\". deviate defaults, necessary overwrite private $.encode_prediction() method, see section Inheriting. callbacks (list() TorchCallbacks) callbacks use training. Defaults empty list(), .e. callbacks.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-format-","dir":"Reference","previous_headings":"","what":"Method format()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Helper print outputs.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$format(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... (ignored).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Prints object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Currently unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-marshal-","dir":"Reference","previous_headings":"","what":"Method marshal()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Marshal learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$marshal(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-3","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Additional parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"self","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-unmarshal-","dir":"Reference","previous_headings":"","what":"Method unmarshal()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Unmarshal learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$unmarshal(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-4","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Additional parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns-1","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"self","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-dataset-","dir":"Reference","previous_headings":"","what":"Method dataset()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Create dataset task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$dataset(task)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-5","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"task Task task","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns-2","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"dataset","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-6","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-6","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":null,"dir":"Reference","previous_headings":"","what":"Image Learner — mlr_learners_torch_image","title":"Image Learner — mlr_learners_torch_image","text":"Base Class Image Learners. features assumed single lazy_tensor column RGB format.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Image Learner — mlr_learners_torch_image","text":"Parameters include inherited LearnerTorch param_set construction argument.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Image Learner — mlr_learners_torch_image","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchImage","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Image Learner — mlr_learners_torch_image","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Image Learner — mlr_learners_torch_image","text":"LearnerTorchImage$new() LearnerTorchImage$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Image Learner — mlr_learners_torch_image","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Image Learner — mlr_learners_torch_image","text":"","code":"LearnerTorchImage$new(   id,   task_type,   param_set = ps(),   label,   optimizer = NULL,   loss = NULL,   callbacks = list(),   packages = \"torchvision\",   man,   properties = NULL,   predict_types = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Image Learner — mlr_learners_torch_image","text":"id (character(1)) id new object. task_type (character(1)) task type. param_set (ParamSet) parameter set. label (character(1)) Label new instance. optimizer (TorchOptimizer) torch optimizer. loss (TorchLoss) loss use training. callbacks (list() TorchCallbacks) callbacks used training. Must unique ids. executed order provided packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). properties (character()) properties object. See mlr_reflections$learner_properties available values. predict_types (character()) predict types. See mlr_reflections$learner_predict_types available values.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Image Learner — mlr_learners_torch_image","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Image Learner — mlr_learners_torch_image","text":"","code":"LearnerTorchImage$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Image Learner — mlr_learners_torch_image","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Learner Torch Model — mlr_learners_torch_model","title":"Learner Torch Model — mlr_learners_torch_model","text":"Create torch learner instantiated nn_module(). classification, output network must scores (softmax).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Learner Torch Model — mlr_learners_torch_model","text":"See LearnerTorch","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Learner Torch Model — mlr_learners_torch_model","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchModel","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Learner Torch Model — mlr_learners_torch_model","text":"network_stored (nn_module NULL) network trained. calling $train(), NULL. ingress_tokens (named list() TorchIngressToken NULL) ingress tokens. Must non-NULL calling $train().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Learner Torch Model — mlr_learners_torch_model","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Learner Torch Model — mlr_learners_torch_model","text":"LearnerTorchModel$new() LearnerTorchModel$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Learner Torch Model — mlr_learners_torch_model","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"LearnerTorchModel$new(   network = NULL,   ingress_tokens = NULL,   task_type,   properties = NULL,   optimizer = NULL,   loss = NULL,   callbacks = list(),   packages = character(0),   feature_types = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learner Torch Model — mlr_learners_torch_model","text":"network (nn_module) instantiated nn_module. cloned construction. classification, outputs must scores (softmax). ingress_tokens (list TorchIngressToken()) list ingress tokens defines dataloader defined. task_type (character(1)) task type. properties (NULL character()) properties learner. Defaults available properties given task type. optimizer (TorchOptimizer) torch optimizer. loss (TorchLoss) loss use training. callbacks (list() TorchCallbacks) callbacks used training. Must unique ids. executed order provided packages (character()) R packages object depends . feature_types (NULL character()) feature types. Defaults available feature types.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Learner Torch Model — mlr_learners_torch_model","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"LearnerTorchModel$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learner Torch Model — mlr_learners_torch_model","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"# We show the learner using a classification task  # The iris task has 4 features and 3 classes network = nn_linear(4, 3) task = tsk(\"iris\")  # This defines the dataloader. # It loads all 4 features, which are also numeric. # The shape is (NA, 4) because the batch dimension is generally NA ingress_tokens = list(   input = TorchIngressToken(task$feature_names, batchgetter_num, c(NA, 4)) )  # Creating the learner and setting required parameters learner = lrn(\"classif.torch_model\",   network = network,   ingress_tokens = ingress_tokens,   batch_size = 16,   epochs = 1,   device = \"cpu\" )  # A simple train-predict ids = partition(task) learner$train(task, ids$train) learner$predict(task, ids$test) #> <PredictionClassif> for 50 observations: #>  row_ids     truth response #>        3    setosa   setosa #>        9    setosa   setosa #>       10    setosa   setosa #>      ---       ---      --- #>      147 virginica   setosa #>      149 virginica   setosa #>      150 virginica   setosa"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":null,"dir":"Reference","previous_headings":"","what":"Class for Torch Module Wrappers — mlr_pipeops_module","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"PipeOpModule wraps nn_module function called train phase mlr3pipelines::PipeOp. , allows assemble PipeOpModules computational mlr3pipelines::Graph represents either neural network preprocessing graph lazy_tensor. cases easier create network creating graph generates graph. cases easier create network creating structurally related graph consisting nodes class PipeOpTorchIngress PipeOpTorch. graph generate graph consisting PipeOpModules part ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"number names input output channels can set construction. input output \"torch_tensor\" training, NULL prediction prediction phase currently serves meaningful purpose.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"training, wrapped nn_module / function called provided inputs order channels defined. Arguments matched name.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"mlr3pipelines::PipeOp -> PipeOpModule","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"module (nn_module) torch module called training phase.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"PipeOpModule$new() PipeOpModule$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"","code":"PipeOpModule$new(   id = \"module\",   module = nn_identity(),   inname = \"input\",   outname = \"output\",   param_vals = list(),   packages = character(0) )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"id (character(1)) id new object. module (nn_module function()) torch module function wrapped. inname (character()) names input channels. outname (character()) names output channels. parameter length 1, parameter module must return tensor. Otherwise must return list() tensors corresponding length. param_vals (named list()) Parameter values set construction. packages (character()) R packages object depends .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"","code":"PipeOpModule$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"","code":"## creating an PipeOpModule manually  # one input and output channel po_module = po(\"module\",   id = \"linear\",   module = torch::nn_linear(10, 20),   inname = \"input\",   outname = \"output\" ) x = torch::torch_randn(16, 10) # This calls the forward function of the wrapped module. y = po_module$train(list(input = x)) str(y) #> List of 1 #>  $ output:Float [1:16, 1:20]  # multiple input and output channels nn_custom = torch::nn_module(\"nn_custom\",   initialize = function(in_features, out_features) {     self$lin1 = torch::nn_linear(in_features, out_features)     self$lin2 = torch::nn_linear(in_features, out_features)   },   forward = function(x, z) {     list(out1 = self$lin1(x), out2 = torch::nnf_relu(self$lin2(z)))   } )  module = nn_custom(3, 2) po_module = po(\"module\",   id = \"custom\",   module = module,   inname = c(\"x\", \"z\"),   outname = c(\"out1\", \"out2\") ) x = torch::torch_randn(1, 3) z = torch::torch_randn(1, 3) out = po_module$train(list(x = x, z = z)) str(out) #> List of 2 #>  $ out1:Float [1:1, 1:2] #>  $ out2:Float [1:1, 1:2]  # How such a PipeOpModule is usually generated graph = po(\"torch_ingress_num\") %>>% po(\"nn_linear\", out_features = 10L) result = graph$train(tsk(\"iris\")) # The PipeOpTorchLinear generates a PipeOpModule and adds it to a new (module) graph result[[1]]$graph #> Graph with 2 PipeOps: #>                 ID         State  sccssors         prdcssors #>             <char>        <char>    <char>            <char> #>  torch_ingress_num <<UNTRAINED>> nn_linear                   #>          nn_linear <<UNTRAINED>>           torch_ingress_num linear_module = result[[1L]]$graph$pipeops$nn_linear linear_module #> PipeOp: <nn_linear> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [torch_tensor,NULL] #> Output channels <name [train type, predict type]>: #>   output [torch_tensor,NULL] formalArgs(linear_module$module) #> [1] \"input\" linear_module$input$name #> [1] \"input\"  # Constructing a PipeOpModule using a simple function po_add1 = po(\"module\",   id = \"add_one\",   module = function(x) x + 1 ) input = list(torch_tensor(1)) po_add1$train(input)$output #> torch_tensor #>  2 #> [ CPUFloatType{1} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Applies 1D adaptive average pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"output_size :: integer(1) target output size. single number.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Calls nn_adaptive_avg_pool1d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"PipeOpTorchAdaptiveAvgPool1D$new() PipeOpTorchAdaptiveAvgPool1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"","code":"PipeOpTorchAdaptiveAvgPool1D$new(   id = \"nn_adaptive_avg_pool1d\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"","code":"PipeOpTorchAdaptiveAvgPool1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_adaptive_avg_pool1d\") pipeop #> PipeOp: <nn_adaptive_avg_pool1d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: output_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Applies 2D adaptive average pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"output_size :: integer() target output size. Can single number vector.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Calls nn_adaptive_avg_pool2d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"PipeOpTorchAdaptiveAvgPool2D$new() PipeOpTorchAdaptiveAvgPool2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"","code":"PipeOpTorchAdaptiveAvgPool2D$new(   id = \"nn_adaptive_avg_pool2d\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"","code":"PipeOpTorchAdaptiveAvgPool2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_adaptive_avg_pool2d\") pipeop #> PipeOp: <nn_adaptive_avg_pool2d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: output_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Applies 3D adaptive average pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"output_size :: integer() target output size. Can single number vector.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Calls nn_adaptive_avg_pool3d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"PipeOpTorchAdaptiveAvgPool3D$new() PipeOpTorchAdaptiveAvgPool3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"","code":"PipeOpTorchAdaptiveAvgPool3D$new(   id = \"nn_adaptive_avg_pool3d\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"","code":"PipeOpTorchAdaptiveAvgPool3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_adaptive_avg_pool3d\") pipeop #> PipeOp: <nn_adaptive_avg_pool3d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: output_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Applies 1D average pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Calls nn_avg_pool1d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"PipeOpTorchAvgPool1D$new() PipeOpTorchAvgPool1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"","code":"PipeOpTorchAvgPool1D$new(id = \"nn_avg_pool1d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"","code":"PipeOpTorchAvgPool1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_avg_pool1d\") pipeop #> PipeOp: <nn_avg_pool1d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(5)> #>                   id    class lower upper nlevels        default  value #>               <char>   <char> <num> <num>   <num>         <list> <list> #> 1:       kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:            stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 3:           padding ParamUty    NA    NA     Inf              0 [NULL] #> 4:         ceil_mode ParamLgl    NA    NA       2          FALSE [NULL] #> 5: count_include_pad ParamLgl    NA    NA       2           TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"Applies 2D average-pooling operation \\(kH * kW\\) regions step size \\(sH * sW\\) steps. number output features equal number input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"Calls nn_avg_pool2d() training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"PipeOpTorchAvgPool2D$new() PipeOpTorchAvgPool2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"","code":"PipeOpTorchAvgPool2D$new(id = \"nn_avg_pool2d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"","code":"PipeOpTorchAvgPool2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_avg_pool2d\") pipeop #> PipeOp: <nn_avg_pool2d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(6)> #>                   id    class lower upper nlevels        default  value #>               <char>   <char> <num> <num>   <num>         <list> <list> #> 1:       kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:            stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 3:           padding ParamUty    NA    NA     Inf              0 [NULL] #> 4:         ceil_mode ParamLgl    NA    NA       2          FALSE [NULL] #> 5: count_include_pad ParamLgl    NA    NA       2           TRUE [NULL] #> 6:  divisor_override ParamDbl     0   Inf     Inf         [NULL] [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"Applies 3D average-pooling operation \\(kT * kH * kW\\) regions step size \\(sT * sH * sW\\) steps. number output features equal \\(\\lfloor \\frac{ \\mbox{input planes} }{sT} \\rfloor\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"Calls nn_avg_pool3d() training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"PipeOpTorchAvgPool3D$new() PipeOpTorchAvgPool3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"","code":"PipeOpTorchAvgPool3D$new(id = \"nn_avg_pool3d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"","code":"PipeOpTorchAvgPool3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_avg_pool3d\") pipeop #> PipeOp: <nn_avg_pool3d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(6)> #>                   id    class lower upper nlevels        default  value #>               <char>   <char> <num> <num>   <num>         <list> <list> #> 1:       kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:            stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 3:           padding ParamUty    NA    NA     Inf              0 [NULL] #> 4:         ceil_mode ParamLgl    NA    NA       2          FALSE [NULL] #> 5: count_include_pad ParamLgl    NA    NA       2           TRUE [NULL] #> 6:  divisor_override ParamDbl     0   Inf     Inf         [NULL] [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"Applies Batch Normalization channel across batch data.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"Calls torch::nn_batch_norm1d(). parameter num_features inferred second dimension input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"PipeOpTorchBatchNorm1D$new() PipeOpTorchBatchNorm1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"","code":"PipeOpTorchBatchNorm1D$new(id = \"nn_batch_norm1d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"","code":"PipeOpTorchBatchNorm1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_batch_norm1d\") pipeop #> PipeOp: <nn_batch_norm1d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(4)> #>                     id    class lower upper nlevels default  value #>                 <char>   <char> <num> <num>   <num>  <list> <list> #> 1:                 eps ParamDbl     0   Inf     Inf   1e-05 [NULL] #> 2:            momentum ParamDbl     0   Inf     Inf     0.1 [NULL] #> 3:              affine ParamLgl    NA    NA       2    TRUE [NULL] #> 4: track_running_stats ParamLgl    NA    NA       2    TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"Applies Batch Normalization channel across batch data.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"Calls torch::nn_batch_norm2d(). parameter num_features inferred second dimension input shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"PipeOpTorchBatchNorm2D$new() PipeOpTorchBatchNorm2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"","code":"PipeOpTorchBatchNorm2D$new(id = \"nn_batch_norm2d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"","code":"PipeOpTorchBatchNorm2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_batch_norm2d\") pipeop #> PipeOp: <nn_batch_norm2d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(4)> #>                     id    class lower upper nlevels default  value #>                 <char>   <char> <num> <num>   <num>  <list> <list> #> 1:                 eps ParamDbl     0   Inf     Inf   1e-05 [NULL] #> 2:            momentum ParamDbl     0   Inf     Inf     0.1 [NULL] #> 3:              affine ParamLgl    NA    NA       2    TRUE [NULL] #> 4: track_running_stats ParamLgl    NA    NA       2    TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"Applies Batch Normalization channel across batch data.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"Calls torch::nn_batch_norm3d(). parameter num_features inferred second dimension input shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"PipeOpTorchBatchNorm3D$new() PipeOpTorchBatchNorm3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"","code":"PipeOpTorchBatchNorm3D$new(id = \"nn_batch_norm3d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"","code":"PipeOpTorchBatchNorm3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_batch_norm3d\") pipeop #> PipeOp: <nn_batch_norm3d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(4)> #>                     id    class lower upper nlevels default  value #>                 <char>   <char> <num> <num>   <num>  <list> <list> #> 1:                 eps ParamDbl     0   Inf     Inf   1e-05 [NULL] #> 2:            momentum ParamDbl     0   Inf     Inf     0.1 [NULL] #> 3:              affine ParamLgl    NA    NA       2    TRUE [NULL] #> 4: track_running_stats ParamLgl    NA    NA       2    TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":null,"dir":"Reference","previous_headings":"","what":"Block Repetition — mlr_pipeops_nn_block","title":"Block Repetition — mlr_pipeops_nn_block","text":"Repeat block n_blocks times.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Block Repetition — mlr_pipeops_nn_block","text":"parameters available block , well n_blocks :: integer(1) often repeat block.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Block Repetition — mlr_pipeops_nn_block","text":"PipeOp sets input output channels block (Graph) received construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Block Repetition — mlr_pipeops_nn_block","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Block Repetition — mlr_pipeops_nn_block","text":"Part documentation copied adapted documentation torch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Block Repetition — mlr_pipeops_nn_block","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchBlock","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Block Repetition — mlr_pipeops_nn_block","text":"block (Graph) neural network segment repeated PipeOp.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Block Repetition — mlr_pipeops_nn_block","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Block Repetition — mlr_pipeops_nn_block","text":"PipeOpTorchBlock$new() PipeOpTorchBlock$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Block Repetition — mlr_pipeops_nn_block","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Block Repetition — mlr_pipeops_nn_block","text":"","code":"PipeOpTorchBlock$new(block, id = \"nn_block\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Block Repetition — mlr_pipeops_nn_block","text":"block (Graph) graph consisting primarily PipeOpTorch objects repeated. id (character(1)) id new object. param_vals (named list()) Parameter values set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Block Repetition — mlr_pipeops_nn_block","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Block Repetition — mlr_pipeops_nn_block","text":"","code":"PipeOpTorchBlock$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Block Repetition — mlr_pipeops_nn_block","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Block Repetition — mlr_pipeops_nn_block","text":"","code":"block = po(\"nn_linear\") %>>% po(\"nn_relu\") po_block = po(\"nn_block\", block, nn_linear.out_features = 10L, n_blocks = 3) network = po(\"torch_ingress_num\") %>>% po_block %>>% po(\"nn_head\") %>>% po(\"torch_loss\", t_loss(\"cross_entropy\")) %>>% po(\"torch_optimizer\", t_opt(\"adam\")) %>>% po(\"torch_model_classif\",   batch_size = 50,   epochs = 3)  task = tsk(\"iris\") network$train(task) #> $torch_model_classif.output #> NULL #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"CELU Activation Function — mlr_pipeops_nn_celu","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Applies element-wise, \\(CELU(x) = max(0,x) + min(0, \\alpha * (exp(x \\alpha) - 1))\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"alpha :: numeric(1) alpha value ELU formulation. Default: 1.0 inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Calls torch::nn_celu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchCELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"PipeOpTorchCELU$new() PipeOpTorchCELU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"","code":"PipeOpTorchCELU$new(id = \"nn_celu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"","code":"PipeOpTorchCELU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_celu\") pipeop #> PipeOp: <nn_celu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1:   alpha ParamDbl  -Inf   Inf     Inf       1 [NULL] #> 2: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Convolution — mlr_pipeops_nn_conv1d","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Applies 1D convolution input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Calls torch::nn_conv1d() trained. paramter in_channels inferred second dimension input tensor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"PipeOpTorchConv1D$new() PipeOpTorchConv1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"","code":"PipeOpTorchConv1D$new(id = \"nn_conv1d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"","code":"PipeOpTorchConv1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv1d\", kernel_size = 10, out_channels = 1) pipeop #> PipeOp: <nn_conv1d> (not trained) #> values: <out_channels=1, kernel_size=10> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(8)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1: out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      1 #> 2:  kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>     10 #> 3:       stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:      padding ParamUty    NA    NA     Inf              0 [NULL] #> 5:     dilation ParamUty    NA    NA     Inf              1 [NULL] #> 6:       groups ParamInt     1   Inf     Inf              1 [NULL] #> 7:         bias ParamLgl    NA    NA       2           TRUE [NULL] #> 8: padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Convolution — mlr_pipeops_nn_conv2d","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"Applies 2D convolution input image composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"Calls torch::nn_conv2d() trained. paramter in_channels inferred second dimension input tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"PipeOpTorchConv2D$new() PipeOpTorchConv2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"","code":"PipeOpTorchConv2D$new(id = \"nn_conv2d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"","code":"PipeOpTorchConv2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv2d\", kernel_size = 10, out_channels = 1) pipeop #> PipeOp: <nn_conv2d> (not trained) #> values: <out_channels=1, kernel_size=10> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(8)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1: out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      1 #> 2:  kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>     10 #> 3:       stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:      padding ParamUty    NA    NA     Inf              0 [NULL] #> 5:     dilation ParamUty    NA    NA     Inf              1 [NULL] #> 6:       groups ParamInt     1   Inf     Inf              1 [NULL] #> 7:         bias ParamLgl    NA    NA       2           TRUE [NULL] #> 8: padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Convolution — mlr_pipeops_nn_conv3d","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"Applies 3D convolution input image composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"Calls torch::nn_conv3d() trained. paramter in_channels inferred second dimension input tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"PipeOpTorchConv3D$new() PipeOpTorchConv3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"","code":"PipeOpTorchConv3D$new(id = \"nn_conv3d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"","code":"PipeOpTorchConv3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv3d\", kernel_size = 10, out_channels = 1) pipeop #> PipeOp: <nn_conv3d> (not trained) #> values: <out_channels=1, kernel_size=10> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(8)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1: out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      1 #> 2:  kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>     10 #> 3:       stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:      padding ParamUty    NA    NA     Inf              0 [NULL] #> 5:     dilation ParamUty    NA    NA     Inf              1 [NULL] #> 6:       groups ParamInt     1   Inf     Inf              1 [NULL] #> 7:         bias ParamLgl    NA    NA       2           TRUE [NULL] #> 8: padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Transpose 1D Convolution Transpose 1D Convolution","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Calls nn_conv_transpose1d. parameter in_channels inferred second dimension input tensor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"PipeOpTorchConvTranspose1D$new() PipeOpTorchConvTranspose1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"","code":"PipeOpTorchConvTranspose1D$new(id = \"nn_conv_transpose1d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"","code":"PipeOpTorchConvTranspose1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv_transpose1d\", kernel_size = 3, out_channels = 2) pipeop #> PipeOp: <nn_conv_transpose1d> (not trained) #> values: <out_channels=2, kernel_size=3> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(9)> #>                id    class lower upper nlevels        default  value #>            <char>   <char> <num> <num>   <num>         <list> <list> #> 1:   out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      2 #> 2:    kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>      3 #> 3:         stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:        padding ParamUty    NA    NA     Inf              0 [NULL] #> 5: output_padding ParamUty    NA    NA     Inf              0 [NULL] #> 6:       dilation ParamUty    NA    NA     Inf              1 [NULL] #> 7:         groups ParamInt     1   Inf     Inf              1 [NULL] #> 8:           bias ParamLgl    NA    NA       2           TRUE [NULL] #> 9:   padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"Applies 2D transposed convolution operator input image composed several input planes, sometimes also called \"deconvolution\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"Calls nn_conv_transpose2d. parameter in_channels inferred second dimension input tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"PipeOpTorchConvTranspose2D$new() PipeOpTorchConvTranspose2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"","code":"PipeOpTorchConvTranspose2D$new(id = \"nn_conv_transpose2d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"","code":"PipeOpTorchConvTranspose2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv_transpose2d\", kernel_size = 3, out_channels = 2) pipeop #> PipeOp: <nn_conv_transpose2d> (not trained) #> values: <out_channels=2, kernel_size=3> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(9)> #>                id    class lower upper nlevels        default  value #>            <char>   <char> <num> <num>   <num>         <list> <list> #> 1:   out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      2 #> 2:    kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>      3 #> 3:         stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:        padding ParamUty    NA    NA     Inf              0 [NULL] #> 5: output_padding ParamUty    NA    NA     Inf              0 [NULL] #> 6:       dilation ParamUty    NA    NA     Inf              1 [NULL] #> 7:         groups ParamInt     1   Inf     Inf              1 [NULL] #> 8:           bias ParamLgl    NA    NA       2           TRUE [NULL] #> 9:   padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"Applies 3D transposed convolution operator input image composed several input planes, sometimes also called \"deconvolution\"","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"Calls nn_conv_transpose3d. parameter in_channels inferred second dimension input tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"PipeOpTorchConvTranspose3D$new() PipeOpTorchConvTranspose3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"","code":"PipeOpTorchConvTranspose3D$new(id = \"nn_conv_transpose3d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"","code":"PipeOpTorchConvTranspose3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv_transpose3d\", kernel_size = 3, out_channels = 2) pipeop #> PipeOp: <nn_conv_transpose3d> (not trained) #> values: <out_channels=2, kernel_size=3> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(9)> #>                id    class lower upper nlevels        default  value #>            <char>   <char> <num> <num>   <num>         <list> <list> #> 1:   out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      2 #> 2:    kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>      3 #> 3:         stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:        padding ParamUty    NA    NA     Inf              0 [NULL] #> 5: output_padding ParamUty    NA    NA     Inf              0 [NULL] #> 6:       dilation ParamUty    NA    NA     Inf              1 [NULL] #> 7:         groups ParamInt     1   Inf     Inf              1 [NULL] #> 8:           bias ParamLgl    NA    NA       2           TRUE [NULL] #> 9:   padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Dropout — mlr_pipeops_nn_dropout","title":"Dropout — mlr_pipeops_nn_dropout","text":"training, randomly zeroes elements input tensor probability p using samples Bernoulli distribution.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Dropout — mlr_pipeops_nn_dropout","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Dropout — mlr_pipeops_nn_dropout","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Dropout — mlr_pipeops_nn_dropout","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Dropout — mlr_pipeops_nn_dropout","text":"p :: numeric(1) Probability element zeroed. Default: 0.5 inplace inplace :: logical(1) set TRUE, operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Dropout — mlr_pipeops_nn_dropout","text":"Calls torch::nn_dropout() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Dropout — mlr_pipeops_nn_dropout","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchDropout","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Dropout — mlr_pipeops_nn_dropout","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Dropout — mlr_pipeops_nn_dropout","text":"PipeOpTorchDropout$new() PipeOpTorchDropout$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Dropout — mlr_pipeops_nn_dropout","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dropout — mlr_pipeops_nn_dropout","text":"","code":"PipeOpTorchDropout$new(id = \"nn_dropout\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dropout — mlr_pipeops_nn_dropout","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Dropout — mlr_pipeops_nn_dropout","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Dropout — mlr_pipeops_nn_dropout","text":"","code":"PipeOpTorchDropout$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dropout — mlr_pipeops_nn_dropout","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Dropout — mlr_pipeops_nn_dropout","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_dropout\") pipeop #> PipeOp: <nn_dropout> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1:       p ParamDbl     0     1     Inf     0.5 [NULL] #> 2: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"ELU Activation Function — mlr_pipeops_nn_elu","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Applies element-wise, $$ELU(x) = max(0,x) + min(0, \\alpha * (exp(x) - 1))$$.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"alpha :: numeric(1) alpha value ELU formulation. Default: 1.0 inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Calls torch::nn_elu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"PipeOpTorchELU$new() PipeOpTorchELU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"","code":"PipeOpTorchELU$new(id = \"nn_elu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"","code":"PipeOpTorchELU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_elu\") pipeop #> PipeOp: <nn_elu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1:   alpha ParamDbl  -Inf   Inf     Inf       1 [NULL] #> 2: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":null,"dir":"Reference","previous_headings":"","what":"Flattens a Tensor — mlr_pipeops_nn_flatten","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"use nn_sequential.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"start_dim :: integer(1) wich dimension start flattening. Default 2. end_dim :: integer(1) wich dimension stop flattening. Default -1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"Calls torch::nn_flatten() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchFlatten","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"PipeOpTorchFlatten$new() PipeOpTorchFlatten$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"","code":"PipeOpTorchFlatten$new(id = \"nn_flatten\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"","code":"PipeOpTorchFlatten$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_flatten\") pipeop #> PipeOp: <nn_flatten> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>           id    class lower upper nlevels default  value #>       <char>   <char> <int> <num>   <num>  <list> <list> #> 1: start_dim ParamInt     1   Inf     Inf       2 [NULL] #> 2:   end_dim ParamInt     1   Inf     Inf      -1 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"GELU Activation Function — mlr_pipeops_nn_gelu","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Gelu","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"approximate :: character(1) Whether use approximation algorithm. Default \"none\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Calls torch::nn_gelu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchGELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"PipeOpTorchGELU$new() PipeOpTorchGELU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"","code":"PipeOpTorchGELU$new(id = \"nn_gelu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"","code":"PipeOpTorchGELU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_gelu\") pipeop #> PipeOp: <nn_gelu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>             id    class lower upper nlevels default  value #>         <char>   <char> <num> <num>   <num>  <list> <list> #> 1: approximate ParamFct    NA    NA       2    none [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"GLU Activation Function — mlr_pipeops_nn_glu","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"gated linear unit. Computes:","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"dim :: integer(1) Dimension split input. Default: -1","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"Calls torch::nn_glu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchGLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"PipeOpTorchGLU$new() PipeOpTorchGLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"","code":"PipeOpTorchGLU$new(id = \"nn_glu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"","code":"PipeOpTorchGLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_glu\") pipeop #> PipeOp: <nn_glu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <int> <num>   <num>  <list> <list> #> 1:    dim ParamInt     1   Inf     Inf      -1 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Applies hard shrinkage function element-wise","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"lambd :: numeric(1) lambda value Hardshrink formulation formulation. Default 0.5.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Calls torch::nn_hardshrink() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"PipeOpTorchHardShrink$new() PipeOpTorchHardShrink$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"","code":"PipeOpTorchHardShrink$new(id = \"nn_hardshrink\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"","code":"PipeOpTorchHardShrink$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_hardshrink\") pipeop #> PipeOp: <nn_hardshrink> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <num> <num>   <num>  <list> <list> #> 1:  lambd ParamDbl  -Inf   Inf     Inf     0.5 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Applies element-wise function \\(\\mbox{Hardsigmoid}(x) = \\frac{ReLU6(x + 3)}{6}\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Calls torch::nn_hardsigmoid() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"PipeOpTorchHardSigmoid$new() PipeOpTorchHardSigmoid$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"","code":"PipeOpTorchHardSigmoid$new(id = \"nn_hardsigmoid\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"","code":"PipeOpTorchHardSigmoid$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_hardsigmoid\") pipeop #> PipeOp: <nn_hardsigmoid> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Applies HardTanh function element-wise.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"min_val :: numeric(1) Minimum value linear region range. Default: -1. max_val :: numeric(1) Maximum value linear region range. Default: 1. inplace :: logical(1) Can optionally operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Calls torch::nn_hardtanh() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardTanh","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"PipeOpTorchHardTanh$new() PipeOpTorchHardTanh$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"","code":"PipeOpTorchHardTanh$new(id = \"nn_hardtanh\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"","code":"PipeOpTorchHardTanh$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_hardtanh\") pipeop #> PipeOp: <nn_hardtanh> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(3)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1: min_val ParamDbl  -Inf   Inf     Inf      -1 [NULL] #> 2: max_val ParamDbl  -Inf   Inf     Inf       1 [NULL] #> 3: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":null,"dir":"Reference","previous_headings":"","what":"Output Head — mlr_pipeops_nn_head","title":"Output Head — mlr_pipeops_nn_head","text":"Output head classification regresssion. NOTE method $shapes_out() access task, returns c(NA, NA). PipeOp trained however, model descriptor correct output shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Output Head — mlr_pipeops_nn_head","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Output Head — mlr_pipeops_nn_head","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Output Head — mlr_pipeops_nn_head","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Output Head — mlr_pipeops_nn_head","text":"bias :: logical(1) Whether use bias. Default TRUE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Output Head — mlr_pipeops_nn_head","text":"Calls torch::nn_linear() input output features inferred input shape / task.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Output Head — mlr_pipeops_nn_head","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHead","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Output Head — mlr_pipeops_nn_head","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Output Head — mlr_pipeops_nn_head","text":"PipeOpTorchHead$new() PipeOpTorchHead$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Output Head — mlr_pipeops_nn_head","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Output Head — mlr_pipeops_nn_head","text":"","code":"PipeOpTorchHead$new(id = \"nn_head\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Output Head — mlr_pipeops_nn_head","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Output Head — mlr_pipeops_nn_head","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Output Head — mlr_pipeops_nn_head","text":"","code":"PipeOpTorchHead$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Output Head — mlr_pipeops_nn_head","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Output Head — mlr_pipeops_nn_head","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_head\") pipeop #> PipeOp: <nn_head> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <num> <num>   <num>  <list> <list> #> 1:   bias ParamLgl    NA    NA       2    TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer Normalization — mlr_pipeops_nn_layer_norm","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"Applies Layer Normalization last certain number dimensions.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"dims :: integer(1) number dimensions normalized (starting last dimension). elementwise_affine :: logical(1) Whether learn affine-linear parameters initialized 1 weights 0 biases. default TRUE. eps :: numeric(1) value added denominator numerical stability.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"Calls torch::nn_layer_norm() trained. parameter normalized_shape inferred dimensions last dims dimensions input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLayerNorm","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"PipeOpTorchLayerNorm$new() PipeOpTorchLayerNorm$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"","code":"PipeOpTorchLayerNorm$new(id = \"nn_layer_norm\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"","code":"PipeOpTorchLayerNorm$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_layer_norm\", dims = 1) pipeop #> PipeOp: <nn_layer_norm> (not trained) #> values: <dims=1> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(3)> #>                    id    class lower upper nlevels        default  value #>                <char>   <char> <num> <num>   <num>         <list> <list> #> 1:               dims ParamInt     1   Inf     Inf <NoDefault[0]>      1 #> 2: elementwise_affine ParamLgl    NA    NA       2           TRUE [NULL] #> 3:                eps ParamDbl     0   Inf     Inf          1e-05 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Applies element-wise, \\(LeakyReLU(x) = max(0, x) + negative_slope * min(0, x)\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"negative_slope :: numeric(1) Controls angle negative slope. Default: 1e-2. inplace :: logical(1) Can optionally operation -place. Default: ‘FALSE’.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Calls torch::nn_hardswish() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLeakyReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"PipeOpTorchLeakyReLU$new() PipeOpTorchLeakyReLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"","code":"PipeOpTorchLeakyReLU$new(id = \"nn_leaky_relu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"","code":"PipeOpTorchLeakyReLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_leaky_relu\") pipeop #> PipeOp: <nn_leaky_relu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>                id    class lower upper nlevels default  value #>            <char>   <char> <num> <num>   <num>  <list> <list> #> 1: negative_slope ParamDbl  -Inf   Inf     Inf    0.01 [NULL] #> 2:        inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":null,"dir":"Reference","previous_headings":"","what":"Linear Layer — mlr_pipeops_nn_linear","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Applies linear transformation incoming data: \\(y = xA^T + b\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Linear Layer — mlr_pipeops_nn_linear","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Linear Layer — mlr_pipeops_nn_linear","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Linear Layer — mlr_pipeops_nn_linear","text":"out_features :: integer(1) output features linear layer. bias :: logical(1) Whether use bias. Default TRUE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Calls torch::nn_linear() trained parameter in_features inferred second last dimension input tensor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Linear Layer — mlr_pipeops_nn_linear","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLinear","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Linear Layer — mlr_pipeops_nn_linear","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Linear Layer — mlr_pipeops_nn_linear","text":"PipeOpTorchLinear$new() PipeOpTorchLinear$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Linear Layer — mlr_pipeops_nn_linear","text":"","code":"PipeOpTorchLinear$new(id = \"nn_linear\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Linear Layer — mlr_pipeops_nn_linear","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Linear Layer — mlr_pipeops_nn_linear","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Linear Layer — mlr_pipeops_nn_linear","text":"","code":"PipeOpTorchLinear$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Linear Layer — mlr_pipeops_nn_linear","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Linear Layer — mlr_pipeops_nn_linear","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_linear\", out_features = 10) pipeop #> PipeOp: <nn_linear> (not trained) #> values: <out_features=10> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1: out_features ParamInt     1   Inf     Inf <NoDefault[0]>     10 #> 2:         bias ParamLgl    NA    NA       2           TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Applies element-wise \\(LogSigmoid(x_i) = log(\\frac{1}{1 + exp(-x_i)})\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Calls torch::nn_log_sigmoid() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLogSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"PipeOpTorchLogSigmoid$new() PipeOpTorchLogSigmoid$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"","code":"PipeOpTorchLogSigmoid$new(id = \"nn_log_sigmoid\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"","code":"PipeOpTorchLogSigmoid$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_log_sigmoid\") pipeop #> PipeOp: <nn_log_sigmoid> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Applies 1D max pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Calls torch::nn_max_pool1d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"PipeOpTorchMaxPool1D$new() PipeOpTorchMaxPool1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"","code":"PipeOpTorchMaxPool1D$new(   id = \"nn_max_pool1d\",   return_indices = FALSE,   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"id (character(1)) Identifier resulting  object. return_indices (logical(1)) Whether return indices. TRUE, two output channels \"output\" \"indices\". param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"","code":"PipeOpTorchMaxPool1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_max_pool1d\") pipeop #> PipeOp: <nn_max_pool1d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(5)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:     padding ParamUty    NA    NA     Inf              0 [NULL] #> 3:      stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 4:    dilation ParamInt  -Inf   Inf     Inf              1 [NULL] #> 5:   ceil_mode ParamLgl    NA    NA       2          FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"Applies 2D max pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"Calls torch::nn_max_pool2d() training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"PipeOpTorchMaxPool2D$new() PipeOpTorchMaxPool2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"","code":"PipeOpTorchMaxPool2D$new(   id = \"nn_max_pool2d\",   return_indices = FALSE,   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"id (character(1)) Identifier resulting  object. return_indices (logical(1)) Whether return indices. TRUE, two output channels \"output\" \"indices\". param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"","code":"PipeOpTorchMaxPool2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_max_pool2d\") pipeop #> PipeOp: <nn_max_pool2d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(5)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:     padding ParamUty    NA    NA     Inf              0 [NULL] #> 3:      stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 4:    dilation ParamInt  -Inf   Inf     Inf              1 [NULL] #> 5:   ceil_mode ParamLgl    NA    NA       2          FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"Applies 3D max pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"Calls torch::nn_max_pool3d() training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"PipeOpTorchMaxPool3D$new() PipeOpTorchMaxPool3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"","code":"PipeOpTorchMaxPool3D$new(   id = \"nn_max_pool3d\",   return_indices = FALSE,   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"id (character(1)) Identifier resulting  object. return_indices (logical(1)) Whether return indices. TRUE, two output channels \"output\" \"indices\". param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"","code":"PipeOpTorchMaxPool3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_max_pool3d\") pipeop #> PipeOp: <nn_max_pool3d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(5)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:     padding ParamUty    NA    NA     Inf              0 [NULL] #> 3:      stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 4:    dilation ParamInt  -Inf   Inf     Inf              1 [NULL] #> 5:   ceil_mode ParamLgl    NA    NA       2          FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":null,"dir":"Reference","previous_headings":"","what":"Merge Operation — mlr_pipeops_nn_merge","title":"Merge Operation — mlr_pipeops_nn_merge","text":"Base class merge operations addition (PipeOpTorchMergeSum), multiplication (PipeOpTorchMergeProd concatenation (PipeOpTorchMergeCat).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Merge Operation — mlr_pipeops_nn_merge","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Merge Operation — mlr_pipeops_nn_merge","text":"PipeOpTorchMerges either vararg input channel constructor argument innum set, input channels \"input1\", ..., \"input<innum>\". one output channel \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Merge Operation — mlr_pipeops_nn_merge","text":"See respective child class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Merge Operation — mlr_pipeops_nn_merge","text":"Per default, private$.shapes_out() method outputs broadcasted tensors. two things aware: NAs assumed batch (almost always batch size first dimension). Tensors expected number dimensions, .e. missing dimensions filled 1s. reason first dimension batch dimension. private method can overwritten PipeOpTorchs inheriting class.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge Operation — mlr_pipeops_nn_merge","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchMerge","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge Operation — mlr_pipeops_nn_merge","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Merge Operation — mlr_pipeops_nn_merge","text":"PipeOpTorchMerge$new() PipeOpTorchMerge$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Merge Operation — mlr_pipeops_nn_merge","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge Operation — mlr_pipeops_nn_merge","text":"","code":"PipeOpTorchMerge$new(   id,   module_generator,   param_set = ps(),   innum = 0,   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge Operation — mlr_pipeops_nn_merge","text":"id (character(1)) Identifier resulting  object. module_generator (nn_module_generator) torch module generator. param_set (ParamSet) parameter set. innum (integer(1)) number inputs. Default 0 means one vararg input channel. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Merge Operation — mlr_pipeops_nn_merge","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge Operation — mlr_pipeops_nn_merge","text":"","code":"PipeOpTorchMerge$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge Operation — mlr_pipeops_nn_merge","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":null,"dir":"Reference","previous_headings":"","what":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Concatenates multiple tensors given dimension. broadcasting rules applied , must reshape tensors shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch. PipeOpTorchMerges either vararg input channel constructor argument innum set, input channels \"input1\", ..., \"input<innum>\". one output channel \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"dim :: integer(1) dimension along concatenate tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Calls nn_merge_cat() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeCat","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"PipeOpTorchMergeCat$new() PipeOpTorchMergeCat$speak() PipeOpTorchMergeCat$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"","code":"PipeOpTorchMergeCat$new(id = \"nn_merge_cat\", innum = 0, param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"id (character(1)) Identifier resulting  object. innum (integer(1)) number inputs. Default 0 means one vararg input channel. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"method-speak-","dir":"Reference","previous_headings":"","what":"Method speak()","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"cat say?","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"","code":"PipeOpTorchMergeCat$speak()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"","code":"PipeOpTorchMergeCat$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_merge_cat\") pipeop #> PipeOp: <nn_merge_cat> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   ... [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <num> <num>   <num>  <list> <list> #> 1:    dim ParamInt  -Inf   Inf     Inf      -1 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":null,"dir":"Reference","previous_headings":"","what":"Merge by Product — mlr_pipeops_nn_merge_prod","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Calculates product input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch. PipeOpTorchMerges either vararg input channel constructor argument innum set, input channels \"input1\", ..., \"input<innum>\". one output channel \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Calls nn_merge_prod() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeProd","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"PipeOpTorchMergeProd$new() PipeOpTorchMergeProd$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"","code":"PipeOpTorchMergeProd$new(id = \"nn_merge_prod\", innum = 0, param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"id (character(1)) Identifier resulting  object. innum (integer(1)) number inputs. Default 0 means one vararg input channel. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"","code":"PipeOpTorchMergeProd$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_merge_prod\") pipeop #> PipeOp: <nn_merge_prod> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   ... [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Merge by Summation — mlr_pipeops_nn_merge_sum","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Calculates sum input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch. PipeOpTorchMerges either vararg input channel constructor argument innum set, input channels \"input1\", ..., \"input<innum>\". one output channel \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Calls nn_merge_sum() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeSum","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"PipeOpTorchMergeSum$new() PipeOpTorchMergeSum$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"","code":"PipeOpTorchMergeSum$new(id = \"nn_merge_sum\", innum = 0, param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"id (character(1)) Identifier resulting  object. innum (integer(1)) number inputs. Default 0 means one vararg input channel. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"","code":"PipeOpTorchMergeSum$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_merge_sum\") pipeop #> PipeOp: <nn_merge_sum> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   ... [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":null,"dir":"Reference","previous_headings":"","what":"PReLU Activation Function — mlr_pipeops_nn_prelu","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Applies element-wise function \\(PReLU(x) = max(0,x) + weight * min(0,x)\\) weight learnable parameter.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"num_parameters :: integer(1): Number learn. Although takes int input, two values legitimate: 1, number channels input. Default: 1. init :: numeric(1) T initial value . Default: 0.25.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Calls torch::nn_prelu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchPReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"PipeOpTorchPReLU$new() PipeOpTorchPReLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"","code":"PipeOpTorchPReLU$new(id = \"nn_prelu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"","code":"PipeOpTorchPReLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_prelu\") pipeop #> PipeOp: <nn_prelu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>                id    class lower upper nlevels default  value #>            <char>   <char> <num> <num>   <num>  <list> <list> #> 1: num_parameters ParamInt     1   Inf     Inf       1 [NULL] #> 2:           init ParamDbl  -Inf   Inf     Inf    0.25 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"ReLU Activation Function — mlr_pipeops_nn_relu","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Applies rectified linear unit function element-wise.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Calls torch::nn_relu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"PipeOpTorchReLU$new() PipeOpTorchReLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"","code":"PipeOpTorchReLU$new(id = \"nn_relu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"","code":"PipeOpTorchReLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_relu\") pipeop #> PipeOp: <nn_relu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Applies element-wise function \\(ReLU6(x) = min(max(0,x), 6)\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Calls torch::nn_relu6() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReLU6","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"PipeOpTorchReLU6$new() PipeOpTorchReLU6$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"","code":"PipeOpTorchReLU6$new(id = \"nn_relu6\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"","code":"PipeOpTorchReLU6$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_relu6\") pipeop #> PipeOp: <nn_relu6> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Reshape a Tensor — mlr_pipeops_nn_reshape","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Reshape tensor given shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"shape :: integer(1) desired output shape. Unknown dimension (one ) can either specified -1 NA.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Calls nn_reshape() trained. internally calls torch::torch_reshape() given shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReshape","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"PipeOpTorchReshape$new() PipeOpTorchReshape$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"","code":"PipeOpTorchReshape$new(id = \"nn_reshape\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"","code":"PipeOpTorchReshape$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_reshape\") pipeop #> PipeOp: <nn_reshape> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels        default  value #>    <char>   <char> <num> <num>   <num>         <list> <list> #> 1:  shape ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":null,"dir":"Reference","previous_headings":"","what":"RReLU Activation Function — mlr_pipeops_nn_rrelu","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Randomized leaky ReLU.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"lower:: numeric(1) Lower bound uniform distribution. Default: 1/8. upper:: numeric(1) Upper bound uniform distribution. Default: 1/3. inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Calls torch::nn_rrelu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchRReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"PipeOpTorchRReLU$new() PipeOpTorchRReLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"","code":"PipeOpTorchRReLU$new(id = \"nn_rrelu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"","code":"PipeOpTorchRReLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_rrelu\") pipeop #> PipeOp: <nn_rrelu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(3)> #>         id    class lower upper nlevels   default  value #>     <char>   <char> <num> <num>   <num>    <list> <list> #> 1:   lower ParamDbl  -Inf   Inf     Inf     0.125 [NULL] #> 2:   upper ParamDbl  -Inf   Inf     Inf 0.3333333 [NULL] #> 3: inplace ParamLgl    NA    NA       2     FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"SELU Activation Function — mlr_pipeops_nn_selu","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Applies element-wise, $$SELU(x) = scale * (max(0,x) + min(0, \\alpha * (exp(x) - 1)))$$, \\(\\alpha=1.6732632423543772848170429916717\\) \\(scale=1.0507009873554804934193349852946\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Calls torch::nn_selu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"PipeOpTorchSELU$new() PipeOpTorchSELU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"","code":"PipeOpTorchSELU$new(id = \"nn_selu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"","code":"PipeOpTorchSELU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_selu\") pipeop #> PipeOp: <nn_selu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Applies element-wise \\(Sigmoid(x_i) = \\frac{1}{1 + exp(-x_i)}\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Calls torch::nn_sigmoid() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"PipeOpTorchSigmoid$new() PipeOpTorchSigmoid$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"","code":"PipeOpTorchSigmoid$new(id = \"nn_sigmoid\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"","code":"PipeOpTorchSigmoid$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_sigmoid\") pipeop #> PipeOp: <nn_sigmoid> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax — mlr_pipeops_nn_softmax","title":"Softmax — mlr_pipeops_nn_softmax","text":"Applies softmax function.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Softmax — mlr_pipeops_nn_softmax","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Softmax — mlr_pipeops_nn_softmax","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Softmax — mlr_pipeops_nn_softmax","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Softmax — mlr_pipeops_nn_softmax","text":"dim :: integer(1) dimension along Softmax computed (every slice along dim sum 1).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Softmax — mlr_pipeops_nn_softmax","text":"Calls torch::nn_softmax() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Softmax — mlr_pipeops_nn_softmax","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftmax","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Softmax — mlr_pipeops_nn_softmax","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Softmax — mlr_pipeops_nn_softmax","text":"PipeOpTorchSoftmax$new() PipeOpTorchSoftmax$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Softmax — mlr_pipeops_nn_softmax","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax — mlr_pipeops_nn_softmax","text":"","code":"PipeOpTorchSoftmax$new(id = \"nn_softmax\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax — mlr_pipeops_nn_softmax","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Softmax — mlr_pipeops_nn_softmax","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax — mlr_pipeops_nn_softmax","text":"","code":"PipeOpTorchSoftmax$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax — mlr_pipeops_nn_softmax","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax — mlr_pipeops_nn_softmax","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_softmax\") pipeop #> PipeOp: <nn_softmax> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels        default  value #>    <char>   <char> <int> <num>   <num>         <list> <list> #> 1:    dim ParamInt     1   Inf     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Applies element-wise, function \\(Softplus(x) = 1/\\beta * log(1 + exp(\\beta * x))\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"beta :: numeric(1) beta value Softplus formulation. Default: 1 threshold :: numeric(1) Values revert linear function. Default: 20","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Calls torch::nn_softplus() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftPlus","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"PipeOpTorchSoftPlus$new() PipeOpTorchSoftPlus$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"","code":"PipeOpTorchSoftPlus$new(id = \"nn_softplus\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"","code":"PipeOpTorchSoftPlus$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_softplus\") pipeop #> PipeOp: <nn_softplus> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>           id    class lower upper nlevels default  value #>       <char>   <char> <num> <num>   <num>  <list> <list> #> 1:      beta ParamDbl  -Inf   Inf     Inf       1 [NULL] #> 2: threshold ParamDbl  -Inf   Inf     Inf      20 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Applies soft shrinkage function elementwise","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"lamd :: numeric(1) lambda (must less zero) value Softshrink formulation. Default: 0.5","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Calls torch::nn_softshrink() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"PipeOpTorchSoftShrink$new() PipeOpTorchSoftShrink$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"","code":"PipeOpTorchSoftShrink$new(id = \"nn_softshrink\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"","code":"PipeOpTorchSoftShrink$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_softshrink\") pipeop #> PipeOp: <nn_softshrink> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <num> <num>   <num>  <list> <list> #> 1:  lambd ParamDbl  -Inf     1     Inf     0.5 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"SoftSign Activation Function — mlr_pipeops_nn_softsign","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Applies element-wise, function \\(SoftSign(x) = x/(1 + |x|\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Calls torch::nn_softsign() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftSign","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"PipeOpTorchSoftSign$new() PipeOpTorchSoftSign$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"","code":"PipeOpTorchSoftSign$new(id = \"nn_softsign\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"","code":"PipeOpTorchSoftSign$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_softsign\") pipeop #> PipeOp: <nn_softsign> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Squeezes tensor calling torch::torch_squeeze() given dimension dim.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"dim :: integer(1) dimension squeeze. NULL, dimensions size 1 squeezed. Negative values interpreted downwards last dimension.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Calls nn_squeeze() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSqueeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"PipeOpTorchSqueeze$new() PipeOpTorchSqueeze$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"","code":"PipeOpTorchSqueeze$new(id = \"nn_squeeze\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"","code":"PipeOpTorchSqueeze$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_squeeze\") pipeop #> PipeOp: <nn_squeeze> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels        default  value #>    <char>   <char> <num> <num>   <num>         <list> <list> #> 1:    dim ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Tanh Activation Function — mlr_pipeops_nn_tanh","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Applies element-wise function:","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Calls torch::nn_tanh() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchTanh","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"PipeOpTorchTanh$new() PipeOpTorchTanh$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"","code":"PipeOpTorchTanh$new(id = \"nn_tanh\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"","code":"PipeOpTorchTanh$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_tanh\") pipeop #> PipeOp: <nn_tanh> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Applies element-wise, \\(Tanhshrink(x) = x - Tanh(x)\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Calls torch::nn_tanhshrink() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchTanhShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"PipeOpTorchTanhShrink$new() PipeOpTorchTanhShrink$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"","code":"PipeOpTorchTanhShrink$new(id = \"nn_tanhshrink\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"","code":"PipeOpTorchTanhShrink$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_tanhshrink\") pipeop #> PipeOp: <nn_tanhshrink> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Treshold Activation Function — mlr_pipeops_nn_threshold","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Thresholds element input Tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"threshold :: numeric(1) value threshold . value :: numeric(1) value replace . inplace :: logical(1) Can optionally operation -place. Default: ‘FALSE’.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Calls torch::nn_threshold() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchThreshold","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"PipeOpTorchThreshold$new() PipeOpTorchThreshold$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"","code":"PipeOpTorchThreshold$new(id = \"nn_threshold\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"","code":"PipeOpTorchThreshold$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_threshold\", threshold = 1, value = 2) pipeop #> PipeOp: <nn_threshold> (not trained) #> values: <threshold=1, value=2> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(3)> #>           id    class lower upper nlevels        default  value #>       <char>   <char> <num> <num>   <num>         <list> <list> #> 1: threshold ParamDbl  -Inf   Inf     Inf <NoDefault[0]>      1 #> 2:     value ParamDbl  -Inf   Inf     Inf <NoDefault[0]>      2 #> 3:   inplace ParamLgl    NA    NA       2          FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Unqueeze Tensor Unqueeze Tensor","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"dim :: integer(1) dimension unsqueeze. Negative values interpreted downwards last dimension.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Calls nn_unsqueeze() trained. internally calls torch::torch_unsqueeze().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchUnsqueeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"PipeOpTorchUnsqueeze$new() PipeOpTorchUnsqueeze$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"","code":"PipeOpTorchUnsqueeze$new(id = \"nn_unsqueeze\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"","code":"PipeOpTorchUnsqueeze$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_unsqueeze\") pipeop #> PipeOp: <nn_unsqueeze> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels        default  value #>    <char>   <char> <num> <num>   <num>         <list> <list> #> 1:    dim ParamInt  -Inf   Inf     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"PipeOp can used preprocess (one ) lazy_tensor columns contained mlr3::Task. preprocessing function specified construction argument fn additional arguments function can defined PipeOp's parameter set. preprocessing done per column, .e. number lazy tensor output columns equal number lazy tensor input columns. create custom preprocessing PipeOps can use pipeop_preproc_torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"addition specifying construction arguments, can overwrite private .shapes_out() method. overwrite , output shapes assumed unknown (NULL). .shapes_out(shapes_in, param_vals, task) (list(), list(), TaskorNULL) -> list()\\cr private method calculates output shapes lazy tensor columns created applying preprocessing function provided parameter values (param_vals). taskis rarely needed, checked notNULL`. private method responsibility calculate output shapes one input column, .e. input shapes_in can assumed exactly one shape vector must calculate output shapes return list() length 1. can also assumed shape NULL (.e. unknown). Also, first dimension can NA, .e. unknown (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"See PipeOpTaskPreproc.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"addition state elements PipeOpTaskPreprocSimple, state also contains $param_vals set training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"addition parameters inherited PipeOpTaskPreproc well specified construction argument param_set following parameters: stages :: character(1) stages apply preprocessing. Can one \"train\", \"predict\" \"\". initial value parameter set \"train\" PipeOp's id starts \"augment_\" \"\" otherwise. Note preprocessing applied $predict() uses parameters set $train() set performing prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"$train() / $predict(), PipeOpModule one input one output channel created. pipeop applies function fn input tensor additionally passing parameter values (minus stages affect_columns) fn. preprocessing graph lazy tensor columns shallowly cloned PipeOpModule added. done avoid modifying user input means identical PipeOpModules can part different preprocessing graphs. possible, created PipeOpModule stateless. later point graph, preprocessing graphs merged possible avoid unnecessary computation. best illustrated example: One lazy tensor column's preprocessing graph -> B. , two branches created B -> C B -> D, creating two preprocessing graphs -> B -> C -> B -> D. loading data, want run preprocessing , .e. want run -> B part twice. reason, task_dataset() try merge graphs cache results graphs. However, graphs using dataset can currently merged. Also, shapes created $train() $predict() might differ. avoid creation graphs predict shapes incompatible train shapes, hypothetical predict shapes already calculated $train() (parameters set train also used predict) PipeOpTorchModel check train predict shapes compatibility starting training. Otherwise, mechanism similar ModelDescriptor construct.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpTaskPreproc -> PipeOpTaskPreprocTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"fn preprocessing function. rowwise Whether preprocessing applied rowwise.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"PipeOpTaskPreprocTorch$new() PipeOpTaskPreprocTorch$shapes_out() PipeOpTaskPreprocTorch$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"","code":"PipeOpTaskPreprocTorch$new(   fn,   id = \"preproc_torch\",   param_vals = list(),   param_set = ps(),   packages = character(0),   rowwise = FALSE,   stages_init = NULL,   tags = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"fn (function character(2)) preprocessing function. Must modify input -place. character(2), first element namespace second element name. preprocessing function applied tensor, tensor passed position first argument. param_set inferred (left NULL) assumed first argument torch_tensor. id (character(1)) id new object. param_vals (named list()) Parameter values set construction. param_set (ParamSet) case function fn takes additional parameter besides torch_tensor can specfied parameters. None parameters can \"predict\" tag. tags include \"train\". packages (character()) packages preprocessing function depends . rowwise (logical(1)) Whether preprocessing function applied rowwise (concatenated row) directly whole tensor. first case batch dimension. stages_init (character(1)) Initial value stages parameter. tags (character()) Tags pipeop.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"method-shapes-out-","dir":"Reference","previous_headings":"","what":"Method shapes_out()","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"Calculates output shapes result applying preprocessing one lazy tensor columns provided shape. Names ignored order matters. uses parameter values currently set.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"","code":"PipeOpTaskPreprocTorch$shapes_out(shapes_in, stage = NULL, task = NULL)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"shapes_in (list() (integer() NULL)) input input shapes lazy tensors. NULL indicates shape unknown. First dimension must NA (NULL). stage (character(1)) stage: either \"train\" \"predict\". task (Task NULL) task, rarely needed.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"list() (integer() NULL)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"","code":"PipeOpTaskPreprocTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"","code":"# Creating a simple task d = data.table(   x1 = as_lazy_tensor(rnorm(10)),   x2 = as_lazy_tensor(rnorm(10)),   x3 = as_lazy_tensor(as.double(1:10)),   y = rnorm(10) )  taskin = as_task_regr(d, target = \"y\")  # Creating a simple preprocessing pipeop po_simple = po(\"preproc_torch\",   # get rid of environment baggage   fn = mlr3misc::crate(function(x, a) x + a),   param_set = paradox::ps(a = paradox::p_int(tags = c(\"train\", \"required\"))) )  po_simple$param_set$set_values(   a = 100,   affect_columns = selector_name(c(\"x1\", \"x2\")),   stages = \"both\" # use during train and predict )  taskout_train = po_simple$train(list(taskin))[[1L]] materialize(taskout_train$data(cols = c(\"x1\", \"x2\")), rbind = TRUE) #> $x1 #> torch_tensor #>  100.6242 #>   99.3046 #>   99.2164 #>   99.0469 #>  101.7928 #>  100.3490 #>  100.2591 #>   99.1940 #>  100.1057 #>   99.6664 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  101.6418 #>   99.3561 #>  100.5870 #>   99.8496 #>   98.2892 #>  101.4310 #>   97.3548 #>   98.9675 #>   99.2925 #>   99.2994 #> [ CPUFloatType{10,1} ] #>   taskout_predict_noaug = po_simple$predict(list(taskin))[[1L]] materialize(taskout_predict_noaug$data(cols = c(\"x1\", \"x2\")), rbind = TRUE) #> $x1 #> torch_tensor #>  100.6242 #>   99.3046 #>   99.2164 #>   99.0469 #>  101.7928 #>  100.3490 #>  100.2591 #>   99.1940 #>  100.1057 #>   99.6664 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  101.6418 #>   99.3561 #>  100.5870 #>   99.8496 #>   98.2892 #>  101.4310 #>   97.3548 #>   98.9675 #>   99.2925 #>   99.2994 #> [ CPUFloatType{10,1} ] #>   po_simple$param_set$set_values(   stages = \"train\" )  # transformation is not applied taskout_predict_aug = po_simple$predict(list(taskin))[[1L]] materialize(taskout_predict_aug$data(cols = c(\"x1\", \"x2\")), rbind = TRUE) #> $x1 #> torch_tensor #>  100.6242 #>   99.3046 #>   99.2164 #>   99.0469 #>  101.7928 #>  100.3490 #>  100.2591 #>   99.1940 #>  100.1057 #>   99.6664 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  101.6418 #>   99.3561 #>  100.5870 #>   99.8496 #>   98.2892 #>  101.4310 #>   97.3548 #>   98.9675 #>   99.2925 #>   99.2994 #> [ CPUFloatType{10,1} ] #>   # Creating a more complex preprocessing PipeOp PipeOpPreprocTorchPoly = R6::R6Class(\"PipeOpPreprocTorchPoly\",  inherit = PipeOpTaskPreprocTorch,  public = list(    initialize = function(id = \"preproc_poly\", param_vals = list()) {      param_set = paradox::ps(        n_degree = paradox::p_int(lower = 1L, tags = c(\"train\", \"required\"))      )      param_set$set_values(        n_degree = 1L      )      fn = mlr3misc::crate(function(x, n_degree) {        torch::torch_cat(          lapply(seq_len(n_degree), function(d) torch::torch_pow(x, d)),          dim = 2L        )      })       super$initialize(        fn = fn,        id = id,        packages = character(0),        param_vals = param_vals,        param_set = param_set,        stages_init = \"both\"      )    }  ),  private = list(    .shapes_out = function(shapes_in, param_vals, task) {      # shapes_in is a list of length 1 containing the shapes      checkmate::assert_true(length(shapes_in[[1L]]) == 2L)      if (shapes_in[[1L]][2L] != 1L) {        stop(\"Input shape must be (NA, 1)\")      }      list(c(NA, param_vals$n_degree))    }  ) )  po_poly = PipeOpPreprocTorchPoly$new(   param_vals = list(n_degree = 3L, affect_columns = selector_name(\"x3\")) )  po_poly$shapes_out(list(c(NA, 1L)), stage = \"train\") #> [[1]] #> [1] NA  3 #>   taskout = po_poly$train(list(taskin))[[1L]] materialize(taskout$data(cols = \"x3\"), rbind = TRUE) #> $x3 #> torch_tensor #>     1     1     1 #>     2     4     8 #>     3     9    27 #>     4    16    64 #>     5    25   125 #>     6    36   216 #>     7    49   343 #>     8    64   512 #>     9    81   729 #>    10   100  1000 #> [ CPUFloatType{10,3} ] #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"PipeOpTorch base class PipeOps represent neural network layers Graph. training, generates PipeOpModule wraps nn_module attaches architecture, also represented Graph consisting mostly PipeOpModules PipeOpNOPs. former Graph operates ModelDescriptors, latter operates tensors. relationship PipeOpTorch PipeOpModule similar relationshop nn_module_generator (like nn_linear) nn_module (like output nn_linear(...)). crucial difference PipeOpTorch infers auxiliary parameters (like in_features nn_linear) automatically intermediate tensor shapes communicated ModelDescriptor. prediction, PipeOpTorch takes Task channel outputs new Task resulting feature union channel. one input output channel, task simply piped .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"inheriting class, one overload either private$.shapes_out() private$.shape_dependent_params() methods, overload private$.make_module(). .make_module(shapes_in, param_vals, task) (list(), list()) -> nn_module private method called generated nn_module passed argument module PipeOpModule. must overwritten, module_generator provided. left , calls provided module_generator arguments obtained private method .shape_dependent_params(). .shapes_out(shapes_in, param_vals, task) (list(), list(), Task NULL) -> named list() private method gets list numeric vectors (shapes_in), parameter values (param_vals), well (optional) Task. shapes_in can assumed order input names PipeOp. output shapes must order output names PipeOp. case output shapes depends task (case PipeOpTorchHead), function return valid output shapes (possibly containing NAs) task argument provided . .shape_dependent_params(shapes_in, param_vals, task) (list(), list()) -> named list() private method inputs .shapes_out. .make_module() overwritten, constructs arguments passed module_generator. Usually means must infer auxiliary parameters can inferred input shapes add user-supplied parameter values (param_vals).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"training, inputs outputs class ModelDescriptor. prediction, input output channels class Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"ParamSet specified child class inheriting PipeOpTorch. Usually parameters arguments wrapped nn_module minus auxiliary parameter can automatically inferred shapes input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"training, PipeOpTorch creates PipeOpModule given parameter specification input shapes incoming ModelDescriptors using private method .make_module(). input shapes provided slot pointer_shape incoming ModelDescriptors. channel names PipeOpModule identical channel names generating PipeOpTorch. model descriptor union incoming ModelDescriptors created. Note modifies graph first ModelDescriptor place efficiency. PipeOpModule added graph slot union edges connect sending PipeOpModules input channel PipeOpModule addeded graph. possible every incoming ModelDescriptor contains information id channel name sending PipeOp slot pointer. new graph model_descriptor_union represents current state neural network architecture. structurally similar subgraph consists pipeops class PipeOpTorch PipeOpTorchIngress ancestors PipeOpTorch. output, shallow copy ModelDescriptor created pointer pointer_shape updated accordingly. shallow copy means ModelDescriptors point Graph allows graph modified -reference different parts code.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"mlr3pipelines::PipeOp -> PipeOpTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"module_generator (nn_module_generator NULL) module generator wrapped PipeOpTorch. NULL, private method private$.make_module(shapes_in, param_vals) must overwritte, see section 'Inheriting'. change construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"PipeOpTorch$new() PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"","code":"PipeOpTorch$new(   id,   module_generator,   param_set = ps(),   param_vals = list(),   inname = \"input\",   outname = \"output\",   packages = \"torch\",   tags = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"id (character(1)) Identifier resulting  object. module_generator (nn_module_generator) torch module generator. param_set (ParamSet) parameter set. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction. inname (character()) names PipeOp's input channels. input channels generated PipeOpModule. Unless wrapped module_generator's forward method (present) argument ..., inname must identical argument names order avoid ambiguity. forward method argument ..., order input channels determines tensors passed wrapped nn_module. left NULL (default), argument module_generator must given argument names modue_generator's forward function set inname. outname (character())  names output channels channels. ouput channels generated PipeOpModule therefore also names list returned $train(). case one output channel, nn_module constructed PipeOp training must return named list(), names list names output channels. default \"output\". packages (character()) R packages object depends . tags (character()) tags PipeOp. tags \"torch\" always added.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"method-shapes-out-","dir":"Reference","previous_headings":"","what":"Method shapes_out()","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"Calculates output shapes given input shapes, parameters task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"","code":"PipeOpTorch$shapes_out(shapes_in, task = NULL)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"shapes_in (list() integer()) input input shapes, must order input channel names PipeOp. task (Task NULL) task, rarely used (default NULL). exception PipeOpTorchHead.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"named list() containing output shapes. names names output channels PipeOp.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"","code":"## Creating a neural network # In torch  task = tsk(\"iris\")  network_generator = torch::nn_module(   initialize = function(task, d_hidden) {     d_in = length(task$feature_names)     self$linear = torch::nn_linear(d_in, d_hidden)     self$output = if (task$task_type == \"regr\") {       torch::nn_linear(d_hidden, 1)     } else if (task$task_type == \"classif\") {       torch::nn_linear(d_hidden, length(task$class_names))     }   },   forward = function(x) {     x = self$linear(x)     x = torch::nnf_relu(x)     self$output(x)   } )  network = network_generator(task, d_hidden = 50) x = torch::torch_tensor(as.matrix(task$data(1, task$feature_names))) y = torch::with_no_grad(network(x))   # In mlr3torch network_generator = po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 50) %>>%   po(\"nn_head\") md = network_generator$train(task)[[1L]] network = model_descriptor_to_module(md) y = torch::with_no_grad(network(torch_ingress_num.input = x))    ## Implementing a custom PipeOpTorch  # defining a custom module nn_custom = nn_module(\"nn_custom\",   initialize = function(d_in1, d_in2, d_out1, d_out2, bias = TRUE) {     self$linear1 = nn_linear(d_in1, d_out1, bias)     self$linear2 = nn_linear(d_in2, d_out2, bias)   },   forward = function(input1, input2) {     output1 = self$linear1(input1)     output2 = self$linear1(input2)      list(output1 = output1, output2 = output2)   } )  # wrapping the module into a custom PipeOpTorch  library(paradox)  PipeOpTorchCustom = R6::R6Class(\"PipeOpTorchCustom\",   inherit = PipeOpTorch,   public = list(     initialize = function(id = \"nn_custom\", param_vals = list()) {       param_set = ps(         d_out1 = p_int(lower = 1, tags = c(\"required\", \"train\")),         d_out2 = p_int(lower = 1, tags = c(\"required\", \"train\")),         bias = p_lgl(default = TRUE, tags = \"train\")       )       super$initialize(         id = id,         param_vals = param_vals,         param_set = param_set,         inname = c(\"input1\", \"input2\"),         outname = c(\"output1\", \"output2\"),         module_generator = nn_custom       )     }   ),   private = list(     .shape_dependent_params = function(shapes_in, param_vals, task) {       c(param_vals,         list(d_in1 = tail(shapes_in[[\"input1\"]], 1)), d_in2 = tail(shapes_in[[\"input2\"]], 1)       )     },     .shapes_out = function(shapes_in, param_vals, task) {       list(         input1 = c(head(shapes_in[[\"input1\"]], -1), param_vals$d_out1),         input2 = c(head(shapes_in[[\"input2\"]], -1), param_vals$d_out2)       )     }   ) )  ## Training  # generate input task = tsk(\"iris\") task1 = task$clone()$select(paste0(\"Sepal.\", c(\"Length\", \"Width\"))) task2 = task$clone()$select(paste0(\"Petal.\", c(\"Length\", \"Width\"))) graph = gunion(list(po(\"torch_ingress_num_1\"), po(\"torch_ingress_num_2\"))) mds_in = graph$train(list(task1, task2), single_input = FALSE)  mds_in[[1L]][c(\"graph\", \"task\", \"ingress\", \"pointer\", \"pointer_shape\")] #> $graph #> Graph with 1 PipeOps: #>                   ID         State sccssors prdcssors #>               <char>        <char>   <char>    <char> #>  torch_ingress_num_1 <<UNTRAINED>>                    #>  #> $task #> <TaskClassif:iris> (150 x 3): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (2): #>   - dbl (2): Sepal.Length, Sepal.Width #>  #> $ingress #> $ingress$torch_ingress_num_1.input #> Ingress: Task[Sepal.Length,Sepal.Width] --> Tensor(NA, 2) #>  #>  #> $pointer #> [1] \"torch_ingress_num_1\" \"output\"              #>  #> $pointer_shape #> [1] NA  2 #>  mds_in[[2L]][c(\"graph\", \"task\", \"ingress\", \"pointer\", \"pointer_shape\")] #> $graph #> Graph with 1 PipeOps: #>                   ID         State sccssors prdcssors #>               <char>        <char>   <char>    <char> #>  torch_ingress_num_2 <<UNTRAINED>>                    #>  #> $task #> <TaskClassif:iris> (150 x 3): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (2): #>   - dbl (2): Petal.Length, Petal.Width #>  #> $ingress #> $ingress$torch_ingress_num_2.input #> Ingress: Task[Petal.Length,Petal.Width] --> Tensor(NA, 2) #>  #>  #> $pointer #> [1] \"torch_ingress_num_2\" \"output\"              #>  #> $pointer_shape #> [1] NA  2 #>   # creating the PipeOpTorch and training it po_torch = PipeOpTorchCustom$new() po_torch$param_set$values = list(d_out1 = 10, d_out2 = 20) train_input = list(input1 = mds_in[[1L]], input2 = mds_in[[2L]]) mds_out = do.call(po_torch$train, args = list(input = train_input)) po_torch$state #> $output1 #> [1] NA 10 #>  #> $output2 #> [1] NA 20 #>   # the new model descriptors  # the resulting graphs are identical identical(mds_out[[1L]]$graph, mds_out[[2L]]$graph) #> [1] TRUE # not that as a side-effect, also one of the input graphs is modified in-place for efficiency mds_in[[1L]]$graph$edges #>                 src_id src_channel    dst_id dst_channel #>                 <char>      <char>    <char>      <char> #> 1: torch_ingress_num_1      output nn_custom      input1 #> 2: torch_ingress_num_2      output nn_custom      input2  # The new task has both Sepal and Petal features identical(mds_out[[1L]]$task, mds_out[[2L]]$task) #> [1] TRUE mds_out[[2L]]$task #> <TaskClassif:iris> (150 x 5): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (4): #>   - dbl (4): Petal.Length, Petal.Width, Sepal.Length, Sepal.Width  # The new ingress slot contains all ingressors identical(mds_out[[1L]]$ingress, mds_out[[2L]]$ingress) #> [1] TRUE mds_out[[1L]]$ingress #> $torch_ingress_num_1.input #> Ingress: Task[Sepal.Length,Sepal.Width] --> Tensor(NA, 2) #>  #> $torch_ingress_num_2.input #> Ingress: Task[Petal.Length,Petal.Width] --> Tensor(NA, 2) #>   # The pointer and pointer_shape slots are different mds_out[[1L]]$pointer #> [1] \"nn_custom\" \"output1\"   mds_out[[2L]]$pointer #> [1] \"nn_custom\" \"output2\"    mds_out[[1L]]$pointer_shape #> [1] NA 10 mds_out[[2L]]$pointer_shape #> [1] NA 20  ## Prediction predict_input = list(input1 = task1, input2 = task2) tasks_out = do.call(po_torch$predict, args = list(input = predict_input)) identical(tasks_out[[1L]], tasks_out[[2L]]) #> [1] TRUE"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback Configuration — mlr_pipeops_torch_callbacks","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"Configures callbacks deep learning model.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"one input channel \"input\" one output channel \"output\". training, channels class ModelDescriptor. prediction, channels class Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"parameters defined dynamically callbacks, id respective callbacks respective set id.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"training callbacks cloned added ModelDescriptor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"mlr3pipelines::PipeOp -> PipeOpTorchCallbacks","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"PipeOpTorchCallbacks$new() PipeOpTorchCallbacks$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"","code":"PipeOpTorchCallbacks$new(   callbacks = list(),   id = \"torch_callbacks\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"callbacks (list TorchCallbacks)  callbacks (something convertible via as_torch_callbacks()). Must unique ids. callbacks cloned construction. id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"","code":"PipeOpTorchCallbacks$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"","code":"po_cb = po(\"torch_callbacks\", \"checkpoint\") po_cb$param_set #> <ParamSetCollection(3)> #>                      id    class lower upper nlevels        default  value #>                  <char>   <char> <num> <num>   <num>         <list> <list> #> 1:      checkpoint.path ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:      checkpoint.freq ParamInt     1   Inf     Inf <NoDefault[0]> [NULL] #> 3: checkpoint.freq_type ParamFct    NA    NA       2          epoch [NULL] mdin = po(\"torch_ingress_num\")$train(list(tsk(\"iris\"))) mdin[[1L]]$callbacks #> named list() mdout = po_cb$train(mdin)[[1L]] mdout$callbacks #> $checkpoint #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: list() #> * Packages: mlr3torch,torch #>  # Can be called again po_cb1 = po(\"torch_callbacks\", t_clbk(\"progress\")) mdout1 = po_cb1$train(list(mdout))[[1L]] mdout1$callbacks #> $progress #> <TorchCallback:progress> Progress #> * Generator: CallbackSetProgress #> * Parameters: list() #> * Packages: progress,mlr3torch,torch #>  #> $checkpoint #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: list() #> * Packages: mlr3torch,torch #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":null,"dir":"Reference","previous_headings":"","what":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"Use entry-point mlr3torch-networks. Unless advanced user, need use directly PipeOpTorchIngressNumeric, PipeOpTorchIngressCategorical PipeOpTorchIngressLazyTensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"state set input shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"Defined construction argument param_set.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"Creates object class TorchIngressToken given task. purpuse store information construct torch dataloader task entry point network.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"mlr3pipelines::PipeOp -> PipeOpTorchIngress","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"feature_types (character(1)) features types can consumed PipeOpTorchIngress.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"PipeOpTorchIngress$new() PipeOpTorchIngress$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"","code":"PipeOpTorchIngress$new(   id,   param_set = ps(),   param_vals = list(),   packages = character(0),   feature_types )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"id (character(1)) Identifier resulting  object. param_set (ParamSet) parameter set. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction. packages (character()) R packages object depends . feature_types (character()) feature types. See mlr_reflections$task_feature_types available values, Additionally, \"lazy_tensor\" supported.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"","code":"PipeOpTorchIngress$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"Ingress PipeOp represents categorical (factor(), ordered() logical()) entry point torch network.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"select :: logical(1) Whether PipeOp selected supported feature types. Otherwise err receiving tasks unsupported feature types.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"Uses batchgetter_categ().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"state set input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorchIngress -> PipeOpTorchIngressCategorical","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"PipeOpTorchIngressCategorical$new() PipeOpTorchIngressCategorical$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"","code":"PipeOpTorchIngressCategorical$new(   id = \"torch_ingress_categ\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"","code":"PipeOpTorchIngressCategorical$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"","code":"graph = po(\"select\", selector = selector_type(\"factor\")) %>>%   po(\"torch_ingress_categ\") task = tsk(\"german_credit\") # The output is a model descriptor md = graph$train(task)[[1L]] ingress = md$ingress[[1L]] ingress$batchgetter(task$data(1, ingress$features), \"cpu\") #> torch_tensor #>  5  5  1  2  3  1  3  1  3  1  4  5  1  2 #> [ CPULongType{1,14} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":null,"dir":"Reference","previous_headings":"","what":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"Ingress single lazy_tensor column.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"shape :: integer() shape tensor, first dimension (batch) must NA. specified, lazy tensor input column needs known shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"returned batchgetter materializes lazy tensor column tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"state set input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorchIngress -> PipeOpTorchIngressLazyTensor","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"PipeOpTorchIngressLazyTensor$new() PipeOpTorchIngressLazyTensor$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"","code":"PipeOpTorchIngressLazyTensor$new(   id = \"torch_ingress_ltnsr\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"","code":"PipeOpTorchIngressLazyTensor$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"","code":"po_ingress = po(\"torch_ingress_ltnsr\") task = tsk(\"lazy_iris\")  md = po_ingress$train(list(task))[[1L]]  ingress = md$ingress x_batch = ingress[[1L]]$batchgetter(data = task$data(1, \"x\"), device = \"cpu\", cache = NULL) x_batch #> torch_tensor #>  5.1000  3.5000  1.4000  0.2000 #> [ CPUFloatType{1,4} ]  # Now we try a lazy tensor with unknown shape, i.e. the shapes between the rows can differ  ds = dataset(   initialize = function() self$x = list(torch_randn(3, 10, 10), torch_randn(3, 8, 8)),   .getitem = function(i) list(x = self$x[[i]]),   .length = function() 2)()  task_unknown = as_task_regr(data.table(   x = as_lazy_tensor(ds, dataset_shapes = list(x = NULL)),   y = rnorm(2) ), target = \"y\", id = \"example2\")  # this task (as it is) can NOT be processed by PipeOpTorchIngressLazyTensor # It therefore needs to be preprocessed po_resize = po(\"trafo_resize\", size = c(6, 6)) task_unknown_resize = po_resize$train(list(task_unknown))[[1L]]  # printing the transformed column still shows unknown shapes, # because the preprocessing pipeop cannot infer them, # however we know that the shape is now (3, 10, 10) for all rows task_unknown_resize$data(1:2, \"x\") #>                x #>    <lazy_tensor> #> 1:      <tnsr[]> #> 2:      <tnsr[]> po_ingress$param_set$set_values(shape = c(NA, 3, 6, 6))  md2 = po_ingress$train(list(task_unknown_resize))[[1L]]  ingress2 = md2$ingress x_batch2 = ingress2[[1L]]$batchgetter(   data = task_unknown_resize$data(1:2, \"x\"),   device = \"cpu\",   cache = NULL )  x_batch2 #> torch_tensor #> (1,1,.,.) =  #>   0.5945 -1.0047  0.0898  0.4695 -0.2213  0.0279 #>   1.0332 -1.4109  0.8082 -0.1438  0.1753 -1.1941 #>   0.3561 -0.5219 -0.0985 -0.4688 -0.4243 -0.1469 #>  -0.4475 -0.5437 -0.2712  0.1507  0.7601 -2.2758 #>  -0.0112 -0.6460 -0.5519  0.0411  1.0907  0.7192 #>   0.3734 -1.7688  0.9091 -0.0721  0.1462 -0.0802 #>  #> (2,1,.,.) =  #>  -0.7234  0.1754 -1.0752 -1.2271 -0.0561  0.6646 #>  -0.6509  0.3914 -0.5936  0.3779 -0.6972  0.0340 #>   0.1612  1.0706  0.4121 -0.3568 -0.2576  0.6846 #>  -0.6529  0.0693 -0.8919  1.0970 -0.4039  0.6887 #>  -1.3010 -0.1821  0.6225  1.1529 -0.2668 -0.5625 #>  -1.1438  0.1229 -0.4223  0.5343 -0.4232  0.5311 #>  #> (1,2,.,.) =  #>   1.1896 -0.4246 -0.9544  0.0690 -1.6374  0.4171 #>   1.0969 -0.5784 -0.3845  0.3485  0.8230  1.1430 #>   0.0990  1.2661  0.0993 -0.4821 -1.1926  0.2399 #>  -0.9176  0.1813 -0.0787  0.4138  1.5121  0.4998 #>   1.0572 -0.2697 -0.0084 -0.7172  0.1109  0.5671 #>   0.0192 -0.3271 -0.6918  0.7929  0.9149  0.5626 #>  #> (2,2,.,.) =  #>  -0.7294 -0.2803 -0.5699  0.5159 -0.0475  0.2107 #>   0.3567  0.9719 -0.7477 -0.4505  0.5632  0.7329 #>   0.1405 -0.0291 -0.3008 -0.4238  0.2812  0.9680 #>   1.1327  0.6031  0.2286 -0.0610 -0.5833 -0.8078 #>  -0.0608  0.2079  1.1355  0.2255 -0.2593  0.5523 #> ... [the output was truncated (use n=-1 to disable)] #> [ CPUFloatType{2,3,6,6} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"Ingress PipeOp represents numeric (integer() numeric()) entry point torch network.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"Uses batchgetter_num().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"state set input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorchIngress -> PipeOpTorchIngressNumeric","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"PipeOpTorchIngressNumeric$new() PipeOpTorchIngressNumeric$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"","code":"PipeOpTorchIngressNumeric$new(id = \"torch_ingress_num\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"","code":"PipeOpTorchIngressNumeric$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"","code":"graph = po(\"select\", selector = selector_type(c(\"numeric\", \"integer\"))) %>>%   po(\"torch_ingress_num\") task = tsk(\"german_credit\") # The output is a model descriptor md = graph$train(task)[[1L]] ingress = md$ingress[[1L]] ingress$batchgetter(task$data(1:5, ingress$features), \"cpu\") #> torch_tensor #>    67  1169     6 #>    22  5951    48 #>    49  2096    12 #>    45  7882    42 #>    53  4870    24 #> [ CPUFloatType{5,3} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss Configuration — mlr_pipeops_torch_loss","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"Configures loss deep learning model.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"parameters defined dynamically loss set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"training loss cloned added ModelDescriptor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"mlr3pipelines::PipeOp -> PipeOpTorchLoss","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"PipeOpTorchLoss$new() PipeOpTorchLoss$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"","code":"PipeOpTorchLoss$new(loss, id = \"torch_loss\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"loss (TorchLoss character(1) nn_loss) loss (something convertible via as_torch_loss()). id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"","code":"PipeOpTorchLoss$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"","code":"po_loss = po(\"torch_loss\", loss = t_loss(\"cross_entropy\")) po_loss$param_set #> <ParamSet(3)> #>              id    class lower upper nlevels default  value #>          <char>   <char> <num> <num>   <num>  <list> <list> #> 1:       weight ParamUty    NA    NA     Inf  [NULL] [NULL] #> 2: ignore_index ParamInt  -Inf   Inf     Inf    -100 [NULL] #> 3:    reduction ParamFct    NA    NA       2    mean [NULL] mdin = po(\"torch_ingress_num\")$train(list(tsk(\"iris\"))) mdin[[1L]]$loss #> NULL mdout = po_loss$train(mdin)[[1L]] mdout$loss #> <TorchLoss:cross_entropy> Cross Entropy #> * Generator: nn_cross_entropy_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: classif"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOp Torch Model — mlr_pipeops_torch_model","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"Builds Torch Learner ModelDescriptor trains given parameter specification. task type must specified construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"one input channel \"input\" takes ModelDescriptor traing Task specified task_type prediction. output NULL training Prediction given task_type prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"trained LearnerTorchModel.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"General: parameters optimizer, loss callbacks, prefixed \"opt.\", \"loss.\" \"cb.<callback id>.\" respectively, well : epochs :: integer(1) number epochs. device :: character(1) device. One \"auto\", \"cpu\", \"cuda\" values defined mlr_reflections$torch$devices. value initialized \"auto\", select \"cuda\" possible, try \"mps\" otherwise fall back \"cpu\". num_threads :: integer(1) number threads intraop pararallelization (device \"cpu\"). value initialized 1. num_interop_threads :: integer(1) number threads intraop interop pararallelization (device \"cpu\"). value initialized 1. Note can set session changing value within R session raise warning. seed :: integer(1) \"random\" NULL torch seed used training prediction. value initialized \"random\", means random seed sampled beginning training phase. seed (either set randomly sampled) available via $model$seed training used prediction. Note setting seed training phase mean default (.e. seed \"random\"), clones learner use different seed. set NULL, seeding done. Evaluation: measures_train :: Measure list() Measures. Measures evaluated training. measures_valid :: Measure list() Measures. Measures evaluated validation. eval_freq :: integer(1) often train / validation predictions evaluated using measures_train / measures_valid. initialized 1. Note final model always evaluated. Early Stopping: patience :: integer(1) activates early stopping using validation scores. performance model improve patience evaluation steps, training ended. Note final model stored learner, best model. initialized 0, means early stopping. first entry measures_valid used metric. also requires specify $validate field Learner, well measures_valid. min_delta :: double(1) minimum improvement threshold (>) early stopping. initialized 0. Dataloader: batch_size :: integer(1) batch size (required). shuffle :: logical(1) Whether shuffle instances dataset. Default FALSE. impact validation. sampler :: torch::sampler Object defines dataloader draw samples. batch_sampler :: torch::sampler Object defines dataloader draws batches. num_workers :: integer(1) number workers data loading (batches loaded parallel). default 0, means data loaded main process. collate_fn :: function merge list samples form batch. pin_memory :: logical(1) Whether dataloader copies tensors CUDA pinned memory returning . drop_last :: logical(1) Whether drop last training batch epoch training. Default FALSE. timeout :: numeric(1) timeout value collecting batch workers. Negative values mean timeout default -1. worker_init_fn :: function(id) function receives worker id ([1, num_workers]) exectued seeding worker data loading. worker_globals :: list() | character() loading data parallel, allows export globals workers. character vector, objects global environment names copied workers. worker_packages :: character() packages load workers. Also see torch::dataloder information.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"LearnerTorchModel created calling model_descriptor_to_learner() provided ModelDescriptor received input channel. parameters set according parameters specified PipeOpTorchModel '$train() method called [Task][mlr3::Task] stored [ModelDescriptor`].","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpLearner -> PipeOpTorchModel","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"PipeOpTorchModel$new() PipeOpTorchModel$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"","code":"PipeOpTorchModel$new(task_type, id = \"torch_model\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"task_type (character(1)) task type model. id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"","code":"PipeOpTorchModel$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"Builds torch classifier trains .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"See LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"one input channel \"input\" takes ModelDescriptor traing Task specified task_type prediction. output NULL training Prediction given task_type prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"trained LearnerTorchModel.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"LearnerTorchModel created calling model_descriptor_to_learner() provided ModelDescriptor received input channel. parameters set according parameters specified PipeOpTorchModel '$train() method called [Task][mlr3::Task] stored [ModelDescriptor`].","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpLearner -> mlr3torch::PipeOpTorchModel -> PipeOpTorchModelClassif","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"PipeOpTorchModelClassif$new() PipeOpTorchModelClassif$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"","code":"PipeOpTorchModelClassif$new(id = \"torch_model_classif\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"","code":"PipeOpTorchModelClassif$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"","code":"# simple logistic regression  # configure the model descriptor md = as_graph(po(\"torch_ingress_num\") %>>%   po(\"nn_head\") %>>%   po(\"torch_loss\", \"cross_entropy\") %>>%   po(\"torch_optimizer\", \"adam\"))$train(tsk(\"iris\"))[[1L]]  print(md) #> <ModelDescriptor: 2 ops> #> * Ingress:  torch_ingress_num.input: [(NA,4)] #> * Task:  iris [classif] #> * Callbacks:  N/A #> * Optimizer:  Adaptive Moment Estimation #> * Loss:  Cross Entropy #> * pointer:  nn_head.output [(NA,3)]  # build the learner from the model descriptor and train it po_model = po(\"torch_model_classif\", batch_size = 50, epochs = 1) po_model$train(list(md)) #> $output #> NULL #>  po_model$state #> $model #> $network #> An `nn_module` containing 15 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • module_list: <nn_module_list> #15 parameters #>  #> $internal_valid_scores #> NULL #>  #> $loss_fn #> list() #>  #> $optimizer #> $optimizer$param_groups #> $optimizer$param_groups[[1]] #> $optimizer$param_groups[[1]]$params #> [1] 1 2 #>  #> $optimizer$param_groups[[1]]$lr #> [1] 0.001 #>  #> $optimizer$param_groups[[1]]$betas #> [1] 0.900 0.999 #>  #> $optimizer$param_groups[[1]]$eps #> [1] 1e-08 #>  #> $optimizer$param_groups[[1]]$weight_decay #> [1] 0 #>  #> $optimizer$param_groups[[1]]$amsgrad #> [1] FALSE #>  #>  #>  #> $optimizer$state #> $optimizer$state$`1` #> $optimizer$state$`1`$step #> [1] 3 #>  #> $optimizer$state$`1`$exp_avg #> torch_tensor #> -0.0260  0.0059 -0.2193 -0.1682 #>  0.5636  0.1927  0.8308  0.4352 #> -0.5376 -0.1987 -0.6115 -0.2670 #> [ CPUFloatType{3,4} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> 0.01 * #>  0.1396  0.0058  1.4116  0.6514 #>   2.8716  0.3722  4.9687  1.2888 #>   3.0630  0.4077  4.3327  0.8889 #> [ CPUFloatType{3,4} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$step #> [1] 3 #>  #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -0.0467 #>  0.1375 #> -0.0908 #> [ CPUFloatType{3} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #> 0.001 * #>  0.5587 #>  1.3208 #>  1.0020 #> [ CPUFloatType{3} ] #>  #>  #>  #>  #> $epochs #> [1] 1 #>  #> $callbacks #> named list() #>  #> $seed #> [1] 1285070263 #>  #> $task_col_info #> Key: <id> #>              id    type                      levels #>          <char>  <char>                      <list> #> 1: Petal.Length numeric                      [NULL] #> 2:  Petal.Width numeric                      [NULL] #> 3: Sepal.Length numeric                      [NULL] #> 4:  Sepal.Width numeric                      [NULL] #> 5:      Species  factor setosa,versicolor,virginica #>  #> attr(,\"class\") #> [1] \"learner_torch_model\" \"list\"                #>  #> $log #> Empty data.table (0 rows and 3 cols): stage,class,msg #>  #> $train_time #> [1] 0.067 #>  #> $param_vals #> $param_vals$epochs #> [1] 1 #>  #> $param_vals$device #> [1] \"auto\" #>  #> $param_vals$num_threads #> [1] 1 #>  #> $param_vals$num_interop_threads #> [1] 1 #>  #> $param_vals$seed #> [1] \"random\" #>  #> $param_vals$eval_freq #> [1] 1 #>  #> $param_vals$measures_train #> list() #>  #> $param_vals$measures_valid #> list() #>  #> $param_vals$patience #> [1] 0 #>  #> $param_vals$min_delta #> [1] 0 #>  #> $param_vals$batch_size #> [1] 50 #>  #>  #> $task_hash #> [1] \"60543108f51f96bb\" #>  #> $feature_names #> [1] \"Petal.Length\" \"Petal.Width\"  \"Sepal.Length\" \"Sepal.Width\"  #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.22.0’ #>  #> $internal_tuned_values #> named list() #>  #> $data_prototype #> Empty data.table (0 rows and 5 cols): Species,Petal.Length,Petal.Width,Sepal.Length,Sepal.Width #>  #> $task_prototype #> Empty data.table (0 rows and 5 cols): Species,Petal.Length,Petal.Width,Sepal.Length,Sepal.Width #>  #> $train_task #> <TaskClassif:iris> (150 x 5): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (4): #>   - dbl (4): Petal.Length, Petal.Width, Sepal.Length, Sepal.Width #>  #> attr(,\"class\") #> [1] \"learner_state\" \"list\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Regression Model — mlr_pipeops_torch_model_regr","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"Builds torch regression model trains .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"See LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"one input channel \"input\" takes ModelDescriptor traing Task specified task_type prediction. output NULL training Prediction given task_type prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"trained LearnerTorchModel.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"LearnerTorchModel created calling model_descriptor_to_learner() provided ModelDescriptor received input channel. parameters set according parameters specified PipeOpTorchModel '$train() method called [Task][mlr3::Task] stored [ModelDescriptor`].","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpLearner -> mlr3torch::PipeOpTorchModel -> PipeOpTorchModelRegr","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"PipeOpTorchModelRegr$new() PipeOpTorchModelRegr$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"PipeOpTorchModelRegr$new(id = \"torch_model_regr\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"PipeOpTorchModelRegr$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"# simple linear regression  # build the model descriptor md = as_graph(po(\"torch_ingress_num\") %>>%   po(\"nn_head\") %>>%   po(\"torch_loss\", \"mse\") %>>%   po(\"torch_optimizer\", \"adam\"))$train(tsk(\"mtcars\"))[[1L]]  print(md) #> <ModelDescriptor: 2 ops> #> * Ingress:  torch_ingress_num.input: [(NA,10)] #> * Task:  mtcars [regr] #> * Callbacks:  N/A #> * Optimizer:  Adaptive Moment Estimation #> * Loss:  Mean Squared Error #> * pointer:  nn_head.output [(NA,1)]  # build the learner from the model descriptor and train it po_model = po(\"torch_model_regr\", batch_size = 20, epochs = 1) po_model$train(list(md)) #> $output #> NULL #>  po_model$state #> $model #> $network #> An `nn_module` containing 11 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • module_list: <nn_module_list> #11 parameters #>  #> $internal_valid_scores #> NULL #>  #> $loss_fn #> list() #>  #> $optimizer #> $optimizer$param_groups #> $optimizer$param_groups[[1]] #> $optimizer$param_groups[[1]]$params #> [1] 1 2 #>  #> $optimizer$param_groups[[1]]$lr #> [1] 0.001 #>  #> $optimizer$param_groups[[1]]$betas #> [1] 0.900 0.999 #>  #> $optimizer$param_groups[[1]]$eps #> [1] 1e-08 #>  #> $optimizer$param_groups[[1]]$weight_decay #> [1] 0 #>  #> $optimizer$param_groups[[1]]$amsgrad #> [1] FALSE #>  #>  #>  #> $optimizer$state #> $optimizer$state$`1` #> $optimizer$state$`1`$step #> [1] 2 #>  #> $optimizer$state$`1`$exp_avg #> torch_tensor #> Columns 1 to 8  -1.5397    5.2623    7.8226  593.2802   -3.0858   -2.8836  328.6158  -12.0903 #>  #> Columns 9 to 10  -2.3402    3.6805 #> [ CPUFloatType{1,10} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> Columns 1 to 6 1.5250e-01  1.7056e+00  3.5828e+00  1.9482e+04  7.4272e-01  6.6760e-01 #>  #> Columns 7 to 10 6.1135e+03  9.6809e+00  3.1214e-01  7.4889e-01 #> [ CPUFloatType{1,10} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$step #> [1] 2 #>  #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -0.3394 #> [ CPUFloatType{1} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #> 0.01 * #>  1.1906 #> [ CPUFloatType{1} ] #>  #>  #>  #>  #> $epochs #> [1] 1 #>  #> $callbacks #> named list() #>  #> $seed #> [1] 1984753776 #>  #> $task_col_info #>         id    type levels #>     <char>  <char> <list> #>  1:     am numeric [NULL] #>  2:   carb numeric [NULL] #>  3:    cyl numeric [NULL] #>  4:   disp numeric [NULL] #>  5:   drat numeric [NULL] #>  6:   gear numeric [NULL] #>  7:     hp numeric [NULL] #>  8:   qsec numeric [NULL] #>  9:     vs numeric [NULL] #> 10:     wt numeric [NULL] #> 11:    mpg numeric [NULL] #>  #> attr(,\"class\") #> [1] \"learner_torch_model\" \"list\"                #>  #> $log #> Empty data.table (0 rows and 3 cols): stage,class,msg #>  #> $train_time #> [1] 0.042 #>  #> $param_vals #> $param_vals$epochs #> [1] 1 #>  #> $param_vals$device #> [1] \"auto\" #>  #> $param_vals$num_threads #> [1] 1 #>  #> $param_vals$num_interop_threads #> [1] 1 #>  #> $param_vals$seed #> [1] \"random\" #>  #> $param_vals$eval_freq #> [1] 1 #>  #> $param_vals$measures_train #> list() #>  #> $param_vals$measures_valid #> list() #>  #> $param_vals$patience #> [1] 0 #>  #> $param_vals$min_delta #> [1] 0 #>  #> $param_vals$batch_size #> [1] 20 #>  #>  #> $task_hash #> [1] \"907e8c646e24bf41\" #>  #> $feature_names #>  [1] \"am\"   \"carb\" \"cyl\"  \"disp\" \"drat\" \"gear\" \"hp\"   \"qsec\" \"vs\"   \"wt\"   #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.22.0’ #>  #> $internal_tuned_values #> named list() #>  #> $data_prototype #> Empty data.table (0 rows and 11 cols): mpg,am,carb,cyl,disp,drat... #>  #> $task_prototype #> Empty data.table (0 rows and 11 cols): mpg,am,carb,cyl,disp,drat... #>  #> $train_task #> <TaskRegr:mtcars> (32 x 11): Motor Trends #> * Target: mpg #> * Properties: - #> * Features (10): #>   - dbl (10): am, carb, cyl, disp, drat, gear, hp, qsec, vs, wt #>  #> attr(,\"class\") #> [1] \"learner_state\" \"list\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer Configuration — mlr_pipeops_torch_optimizer","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"Configures optimizer deep learning model.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"one input channel \"input\" one output channel \"output\". training, channels class ModelDescriptor. prediction, channels class Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"parameters defined dynamically optimizer set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"training, optimizer cloned added ModelDescriptor. Note parameter set stored TorchOptimizer reference-identical parameter set pipeop .","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"mlr3pipelines::PipeOp -> PipeOpTorchOptimizer","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"PipeOpTorchOptimizer$new() PipeOpTorchOptimizer$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"","code":"PipeOpTorchOptimizer$new(   optimizer = t_opt(\"adam\"),   id = \"torch_optimizer\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"optimizer (TorchOptimizer character(1) torch_optimizer_generator) optimizer (something convertible via as_torch_optimizer()). id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"","code":"PipeOpTorchOptimizer$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"","code":"po_opt = po(\"torch_optimizer\", \"sgd\", lr = 0.01) po_opt$param_set #> <ParamSet(5)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1:           lr ParamDbl     0   Inf     Inf <NoDefault[0]>   0.01 #> 2:     momentum ParamDbl     0     1     Inf              0 [NULL] #> 3:    dampening ParamDbl     0     1     Inf              0 [NULL] #> 4: weight_decay ParamDbl     0     1     Inf              0 [NULL] #> 5:     nesterov ParamLgl    NA    NA       2          FALSE [NULL] mdin = po(\"torch_ingress_num\")$train(list(tsk(\"iris\"))) mdin[[1L]]$optimizer #> NULL mdout = po_opt$train(mdin) mdout[[1L]]$optimizer #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_sgd #> * Parameters: lr=0.01 #> * Packages: torch,mlr3torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":null,"dir":"Reference","previous_headings":"","what":"Iris Classification Task — mlr_tasks_lazy_iris","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"classification task popular datasets::iris data set. Just like iris task, features represented one lazy tensor column.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"R6::R6Class inheriting mlr3::TaskClassif.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"source","dir":"Reference","previous_headings":"","what":"Source","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"https://en.wikipedia.org/wiki/Iris_flower_data_set","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"","code":"tsk(\"lazy_iris\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"Task type: “classif” Properties: “multiclass” Missings: Target: “Species” Features: “x” Data Dimension: 150x3","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"Anderson E (1936). “Species Problem Iris.” Annals Missouri Botanical Garden, 23(3), 457. doi:10.2307/2394164 .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"","code":"task = tsk(\"lazy_iris\") task #> <TaskClassif:lazy_iris> (150 x 2): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (1): #>   - lt (1): x df = task$data() materialize(df$x[1:6], rbind = TRUE) #> torch_tensor #>  5.1000  3.5000  1.4000  0.2000 #>  4.9000  3.0000  1.4000  0.2000 #>  4.7000  3.2000  1.3000  0.2000 #>  4.6000  3.1000  1.5000  0.2000 #>  5.0000  3.6000  1.4000  0.2000 #>  5.4000  3.9000  1.7000  0.4000 #> [ CPUFloatType{6,4} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"MNIST Image classification — mlr_tasks_mnist","title":"MNIST Image classification — mlr_tasks_mnist","text":"Classic MNIST image classification. underlying DataBackend contains columns \"label\", \"image\", \"row_id\", \"split\", last column indicates whether row belongs train test set. first 60000 rows belong training set, last 10000 rows test set.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"source","dir":"Reference","previous_headings":"","what":"Source","title":"MNIST Image classification — mlr_tasks_mnist","text":"https://torchvision.mlverse.org/reference/mnist_dataset.html","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"MNIST Image classification — mlr_tasks_mnist","text":"","code":"tsk(\"mnist\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"download","dir":"Reference","previous_headings":"","what":"Download","title":"MNIST Image classification — mlr_tasks_mnist","text":"task's backend DataBackendLazy download data requested. meta-data already available . can cache datasets setting mlr3torch.cache option TRUE specific path used cache directory.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"MNIST Image classification — mlr_tasks_mnist","text":"Task type: “classif” Properties: “multiclass” Missings: Target: “label” Features: “image” Data Dimension: 70000x4","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"MNIST Image classification — mlr_tasks_mnist","text":"Lecun, Y., Bottou, L., Bengio, Y., Haffner, P. (1998). “Gradient-based learning applied document recognition.” Proceedings IEEE, 86(11), 2278-2324. doi:10.1109/5.726791 .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"MNIST Image classification — mlr_tasks_mnist","text":"","code":"task = tsk(\"mnist\") task #> <TaskClassif:mnist> (70000 x 2): MNIST Digit Classification #> * Target: label #> * Properties: multiclass #> * Features (1): #>   - lt (1): image"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":null,"dir":"Reference","previous_headings":"","what":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"Subset famous ImageNet dataset. data obtained torchvision::tiny_imagenet_dataset(). underlying DataBackend contains columns \"class\", \"image\", \"..row_id\", \"split\", last column indicates whether row belongs train, validation test set defined provided torchvision. labels test rows, default, observations inactive, means task uses 110000 120000 observations defined underlying data backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"","code":"tsk(\"tiny_imagenet\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"download","dir":"Reference","previous_headings":"","what":"Download","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"task's backend DataBackendLazy download data requested. meta-data already available . can cache datasets setting mlr3torch.cache option TRUE specific path used cache directory.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"Task type: “classif” Properties: “multiclass” Missings: Target: “class” Features: “image” Data Dimension: 120000x4","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"Deng, Jia, Dong, Wei, Socher, Richard, Li, Li-Jia, Li, Kai, Fei-Fei, Li (2009). “Imagenet: large-scale hierarchical image database.” 2009 IEEE conference computer vision pattern recognition, 248–255. IEEE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"","code":"task = tsk(\"tiny_imagenet\") task #> <TaskClassif:tiny_imagenet> (110000 x 2): ImageNet Subset #> * Target: class #> * Properties: multiclass #> * Features (1): #>   - lt (1): image"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_learner.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","title":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","text":"First nn_graph created using model_descriptor_to_module learner created module remaining information model descriptor, must include optimizer loss function optionally callbacks.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_learner.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","text":"","code":"model_descriptor_to_learner(model_descriptor)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_learner.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","text":"model_descriptor (ModelDescriptor) model descriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_learner.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","text":"Learner","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_module.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","title":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","text":"Creates nn_graph ModelDescriptor. Mostly internal use, since ModelDescriptor circumstances harder use just creating nn_graph directly.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_module.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","text":"","code":"model_descriptor_to_module(   model_descriptor,   output_pointers = NULL,   list_output = FALSE )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_module.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","text":"model_descriptor (ModelDescriptor) Model Descriptor. pointer ignored, instead output_pointer values used. $graph member modified -reference. output_pointers (list character) Collection pointers indicate part model_descriptor$graph used output. Entries format ModelDescriptor$pointer. list_output (logical(1)) Whether output list tensors. FALSE, length(output_pointers) must 1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_module.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","text":"nn_graph","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":null,"dir":"Reference","previous_headings":"","what":"Union of ModelDescriptors — model_descriptor_union","title":"Union of ModelDescriptors — model_descriptor_union","text":"mostly internal function used PipeOpTorchs multiple input channels. creates union multiple ModelDescriptors: graphs combinded (identical begin ). first entry's graph modified reference. PipeOps ID must identical. new input edges may added PipeOps. Drops pointer / pointer_shape entries. new task feature union two incoming tasks. optimizer loss ModelDescriptors must identical. Ingress tokens callbacks merged, objects \"id\" must identical.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Union of ModelDescriptors — model_descriptor_union","text":"","code":"model_descriptor_union(md1, md2)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Union of ModelDescriptors — model_descriptor_union","text":"md1 (ModelDescriptor) first ModelDescriptor. md2 (ModelDescriptor) second ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Union of ModelDescriptors — model_descriptor_union","text":"ModelDescriptor","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Union of ModelDescriptors — model_descriptor_union","text":"requirement new input edgedes may added PipeOps  theoretically necessary, since assume ModelDescriptor built beginning end (.e. PipeOps never get new ancestors) can make assumption simplify things. Otherwise need treat \"...\"-inputs special.)","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Neural Network Layer — nn","title":"Create a Neural Network Layer — nn","text":"Retrieve neural network layer mlr_pipeops dictionary.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Neural Network Layer — nn","text":"","code":"nn(.key, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Neural Network Layer — nn","text":".key (character(1)) ... () Additional parameters, constructor arguments fields.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Neural Network Layer — nn","text":"","code":"po1 = po(\"nn_linear\", id = \"linear\") # is the same as: po2 = nn(\"linear\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":null,"dir":"Reference","previous_headings":"","what":"Graph Network — nn_graph","title":"Graph Network — nn_graph","text":"Represents neural network using Graph usually costains mostly PipeOpModules.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Graph Network — nn_graph","text":"","code":"nn_graph(graph, shapes_in, output_map = graph$output$name, list_output = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Graph Network — nn_graph","text":"graph (Graph) Graph wrap. cloned. shapes_in (named integer) Shape info tensors go graph. Names must graph$input$name, possibly different order. output_map (character) graph's outputs use. Must subset graph$output$name. list_output (logical(1)) Whether output list tensors. FALSE (default), length(output_map) must 1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Graph Network — nn_graph","text":"nn_graph","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Graph Network — nn_graph","text":"","code":"graph = mlr3pipelines::Graph$new() graph$add_pipeop(po(\"module_1\", module = nn_linear(10, 20)), clone = FALSE) graph$add_pipeop(po(\"module_2\", module = nn_relu()), clone = FALSE) graph$add_pipeop(po(\"module_3\", module = nn_linear(20, 1)), clone = FALSE) graph$add_edge(\"module_1\", \"module_2\") graph$add_edge(\"module_2\", \"module_3\")  network = nn_graph(graph, shapes_in = list(module_1.input = c(NA, 10)))  x = torch_randn(16, 10)  network(module_1.input = x) #> torch_tensor #>  0.0771 #>  0.5067 #>  0.5069 #>  0.2827 #>  0.2935 #>  0.2439 #> -0.1360 #>  0.1873 #>  0.1472 #>  0.2764 #>  0.2794 #>  0.0572 #>  0.1514 #>  0.1721 #>  0.5291 #>  0.3046 #> [ CPUFloatType{16,1} ][ grad_fn = <AddmmBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_cat.html","id":null,"dir":"Reference","previous_headings":"","what":"Concatenates multiple tensors — nn_merge_cat","title":"Concatenates multiple tensors — nn_merge_cat","text":"Concatenates multiple tensors given dimension. broadcasting rules applied , must reshape tensors shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_cat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Concatenates multiple tensors — nn_merge_cat","text":"","code":"nn_merge_cat(dim = -1)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_cat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Concatenates multiple tensors — nn_merge_cat","text":"dim (integer(1)) dimension concatenation.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_prod.html","id":null,"dir":"Reference","previous_headings":"","what":"Product of multiple tensors — nn_merge_prod","title":"Product of multiple tensors — nn_merge_prod","text":"Calculates product input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_prod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Product of multiple tensors — nn_merge_prod","text":"","code":"nn_merge_prod()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Sum of multiple tensors — nn_merge_sum","title":"Sum of multiple tensors — nn_merge_sum","text":"Calculates sum input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sum of multiple tensors — nn_merge_sum","text":"","code":"nn_merge_sum()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Reshape — nn_reshape","title":"Reshape — nn_reshape","text":"Reshape tensor given shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reshape — nn_reshape","text":"","code":"nn_reshape(shape)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reshape — nn_reshape","text":"shape (integer()) desired output shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_squeeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Squeeze — nn_squeeze","title":"Squeeze — nn_squeeze","text":"Squeezes tensor calling torch::torch_squeeze() given dimension dim.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_squeeze.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squeeze — nn_squeeze","text":"","code":"nn_squeeze(dim)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_squeeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squeeze — nn_squeeze","text":"dim (integer()) dimension squeeze.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_unsqueeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Unsqueeze — nn_unsqueeze","title":"Unsqueeze — nn_unsqueeze","text":"Unsqueezes tensor calling torch::torch_unsqueeze() given dimension dim.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_unsqueeze.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unsqueeze — nn_unsqueeze","text":"","code":"nn_unsqueeze(dim)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_unsqueeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unsqueeze — nn_unsqueeze","text":"dim (integer(1)) dimension unsqueeze.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"Function create objects class PipeOpTaskPreprocTorch convenient way. Start reading documentation PipeOpTaskPreprocTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"","code":"pipeop_preproc_torch(   id,   fn,   shapes_out = NULL,   param_set = NULL,   packages = character(0),   rowwise = FALSE,   parent_env = parent.frame(),   stages_init = NULL,   tags = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"id (character(1)) id new object. fn (function) preprocessing function. shapes_out (function NULL \"infer\") private .shapes_out(shapes_in, param_vals, task) method PipeOpTaskPreprocTorch (see section Inheriting). Special values NULL infer: NULL, output shapes unknown. \"infer\", output shape function inferred calculates output shapes follows: input shape (NA, ...) meta-tensor shape (1, ...) created preprocessing function applied. Afterwards batch dimension (1) replaced NA shape returned. first dimension NA, output shape applying preprocessing function returned. Method \"infer\" correct cases, might fail edge cases. param_set (ParamSet NULL) parameter set. left NULL (default) parameter set inferred following way: parameters first ... fn set untyped parameters tags 'train' default value tagged 'required' well. Default values annotated. packages (character()) R packages object depends . rowwise (logical(1)) Whether preprocessing applied row-wise. parent_env (environment) parent environment R6 class. stages_init (character(1)) Initial value stages parameter. NULL (default), set \"\" case id starts \"trafo\" \"train\" starts \"augment\". Otherwise must specified. tags (character()) Tags pipeop","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"R6Class instance inheriting PipeOpTaskPreprocTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"","code":"PipeOpPreprocExample = pipeop_preproc_torch(\"preproc_example\", function(x, a) x + a) po_example = PipeOpPreprocExample$new() po_example$param_set #> <ParamSet(3)> #>                id    class lower upper nlevels        default  value #>            <char>   <char> <num> <num>   <num>         <list> <list> #> 1:              a ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:         stages ParamFct    NA    NA       3 <NoDefault[0]> [NULL] #> 3: affect_columns ParamUty    NA    NA     Inf  <Selector[1]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/replace_head.html","id":null,"dir":"Reference","previous_headings":"","what":"Replace the head of a network Replaces the head of the network with a linear layer with d_out classes. — replace_head","title":"Replace the head of a network Replaces the head of the network with a linear layer with d_out classes. — replace_head","text":"Replace head network Replaces head network linear layer d_out classes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/replace_head.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Replace the head of a network Replaces the head of the network with a linear layer with d_out classes. — replace_head","text":"","code":"replace_head(network, d_out)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/replace_head.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Replace the head of a network Replaces the head of the network with a linear layer with d_out classes. — replace_head","text":"network (torch::nn_module) network d_out (integer(1)) number output classes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":null,"dir":"Reference","previous_headings":"","what":"Sugar Function for Torch Callback — t_clbk","title":"Sugar Function for Torch Callback — t_clbk","text":"Retrieves one TorchCallbacks mlr3torch_callbacks. Works like mlr3::lrn() mlr3::lrns().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sugar Function for Torch Callback — t_clbk","text":"","code":"t_clbk(.key, ...)  t_clbks(.keys)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sugar Function for Torch Callback — t_clbk","text":".key (character(1)) key torch callback. ... () See description dictionary_sugar_get(). .keys (character()) keys callbacks.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sugar Function for Torch Callback — t_clbk","text":"TorchCallback list() TorchCallbacks","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sugar Function for Torch Callback — t_clbk","text":"","code":"t_clbk(\"progress\") #> <TorchCallback:progress> Progress #> * Generator: CallbackSetProgress #> * Parameters: list() #> * Packages: progress,mlr3torch,torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss Function Quick Access — t_loss","title":"Loss Function Quick Access — t_loss","text":"Retrieve one TorchLosses mlr3torch_losses. Works like mlr3::lrn() mlr3::lrns().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss Function Quick Access — t_loss","text":"","code":"t_loss(.key, ...)  t_losses(.keys, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss Function Quick Access — t_loss","text":".key (character(1)) Key object retrieve. ... () See description dictionary_sugar_get. .keys (character()) keys losses.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loss Function Quick Access — t_loss","text":"TorchLoss","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loss Function Quick Access — t_loss","text":"","code":"t_loss(\"mse\", reduction = \"mean\") #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: reduction=mean #> * Packages: torch,mlr3torch #> * Task Types: regr # get the dictionary t_loss() #> <DictionaryMlr3torchLosses> with 3 stored values #> Keys: cross_entropy, l1, mse t_losses(c(\"mse\", \"l1\")) #> $mse #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr #>  #> $l1 #> <TorchLoss:l1> Absolute Error #> * Generator: nn_l1_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr #>  # get the dictionary t_losses() #> <DictionaryMlr3torchLosses> with 3 stored values #> Keys: cross_entropy, l1, mse"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizers Quick Access — t_opt","title":"Optimizers Quick Access — t_opt","text":"Retrieves one TorchOptimizers mlr3torch_optimizers. Works like mlr3::lrn() mlr3::lrns().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizers Quick Access — t_opt","text":"","code":"t_opt(.key, ...)  t_opts(.keys, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizers Quick Access — t_opt","text":".key (character(1)) Key object retrieve. ... () See description dictionary_sugar_get. .keys (character()) keys optimizers.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizers Quick Access — t_opt","text":"TorchOptimizer","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Optimizers Quick Access — t_opt","text":"","code":"t_opt(\"adam\", lr = 0.1) #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: lr=0.1 #> * Packages: torch,mlr3torch # get the dictionary t_opt() #> <DictionaryMlr3torchOptimizers> with 7 stored values #> Keys: adadelta, adagrad, adam, asgd, rmsprop, rprop, sgd t_opts(c(\"adam\", \"sgd\")) #> $adam #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: list() #> * Packages: torch,mlr3torch #>  #> $sgd #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_sgd #> * Parameters: list() #> * Packages: torch,mlr3torch #>  # get the dictionary t_opts() #> <DictionaryMlr3torchOptimizers> with 7 stored values #> Keys: adadelta, adagrad, adam, asgd, rmsprop, rprop, sgd"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Dataset from a Task — task_dataset","title":"Create a Dataset from a Task — task_dataset","text":"Creates torch dataset mlr3 Task. resulting dataset's $.get_batch() method returns list elements x, y index: x list tensors, whose content defined parameter feature_ingress_tokens. y target variable content defined parameter target_batchgetter. .index index batch task's data. data returned device specified parameter device.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Dataset from a Task — task_dataset","text":"","code":"task_dataset(task, feature_ingress_tokens, target_batchgetter = NULL, device)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Dataset from a Task — task_dataset","text":"task (Task) task build dataset. feature_ingress_tokens (named list() TorchIngressToken) ingress token defines one item $x value batch corresponding names. target_batchgetter (function(data, device)) function taking arguments data, data.table containing target variable, device. must return target torch tensor selected device. device (character()) device, e.g. \"cuda\" \"cpu\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Dataset from a Task — task_dataset","text":"torch::dataset","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Dataset from a Task — task_dataset","text":"","code":"task = tsk(\"iris\") sepal_ingress = TorchIngressToken(   features = c(\"Sepal.Length\", \"Sepal.Width\"),   batchgetter = batchgetter_num,   shape = c(NA, 2) ) petal_ingress = TorchIngressToken(   features = c(\"Petal.Length\", \"Petal.Width\"),   batchgetter = batchgetter_num,   shape = c(NA, 2) ) ingress_tokens = list(sepal = sepal_ingress, petal = petal_ingress)  target_batchgetter = function(data, device) {   torch_tensor(data = data[[1L]], dtype = torch_float32(), device)$unsqueeze(2) } dataset = task_dataset(task, ingress_tokens, target_batchgetter, \"cpu\") batch = dataset$.getbatch(1:10) batch #> $x #> $x$sepal #> torch_tensor #>  5.1000  3.5000 #>  4.9000  3.0000 #>  4.7000  3.2000 #>  4.6000  3.1000 #>  5.0000  3.6000 #>  5.4000  3.9000 #>  4.6000  3.4000 #>  5.0000  3.4000 #>  4.4000  2.9000 #>  4.9000  3.1000 #> [ CPUFloatType{10,2} ] #>  #> $x$petal #> torch_tensor #>  1.4000  0.2000 #>  1.4000  0.2000 #>  1.3000  0.2000 #>  1.5000  0.2000 #>  1.4000  0.2000 #>  1.7000  0.4000 #>  1.4000  0.3000 #>  1.5000  0.2000 #>  1.4000  0.2000 #>  1.5000  0.1000 #> [ CPUFloatType{10,2} ] #>  #>  #> $.index #> torch_tensor #>   1 #>   2 #>   3 #>   4 #>   5 #>   6 #>   7 #>   8 #>   9 #>  10 #> [ CPULongType{10} ] #>  #> $y #> torch_tensor #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #> [ CPUFloatType{10,1} ] #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Callback Desctiptor — torch_callback","title":"Create a Callback Desctiptor — torch_callback","text":"Convenience function create custom TorchCallback. arguments available callback_set() also available . information correctly implement new callback, see CallbackSet.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Callback Desctiptor — torch_callback","text":"","code":"torch_callback(   id,   classname = paste0(\"CallbackSet\", capitalize(id)),   param_set = NULL,   packages = NULL,   label = capitalize(id),   man = NULL,   on_begin = NULL,   on_end = NULL,   on_exit = NULL,   on_epoch_begin = NULL,   on_before_valid = NULL,   on_epoch_end = NULL,   on_batch_begin = NULL,   on_batch_end = NULL,   on_after_backward = NULL,   on_batch_valid_begin = NULL,   on_batch_valid_end = NULL,   on_valid_end = NULL,   state_dict = NULL,   load_state_dict = NULL,   initialize = NULL,   public = NULL,   private = NULL,   active = NULL,   parent_env = parent.frame(),   inherit = CallbackSet,   lock_objects = FALSE )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Callback Desctiptor — torch_callback","text":"id (character(1))` id torch callback. classname (character(1)) class name. param_set (ParamSet) parameter set, present inferred $initialize() method. packages (character())packages callback depends . Default isNULL`. label (character(1)) label torch callback. Defaults capitalized id. man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). default NULL. on_begin, on_end, on_epoch_begin, on_before_valid, on_epoch_end, on_batch_begin, on_batch_end, on_after_backward, on_batch_valid_begin, on_batch_valid_end, on_valid_end, on_exit (function) Function execute given stage, see section Stages. state_dict (function()) function retrieves state dict callback. available learner training. load_state_dict (function(state_dict)) Function loads callback state. initialize (function()) initialization method callback. public, private, active (list()) Additional public, private, active fields add callback. parent_env (environment()) parent environment R6Class. inherit (R6ClassGenerator) class inherit. class must either CallbackSet (default) inherit . lock_objects (logical(1)) Whether lock objects resulting R6Class. FALSE (default), values can freely assigned self without declaring class definition.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Callback Desctiptor — torch_callback","text":"TorchCallback","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Create a Callback Desctiptor — torch_callback","text":"first creates R6 class inheriting CallbackSet (using callback_set()) wraps generator TorchCallback can passed torch learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"stages","dir":"Reference","previous_headings":"","what":"Stages","title":"Create a Callback Desctiptor — torch_callback","text":"begin :: Run training loop begins. epoch_begin :: Run beginning epoch. batch_begin :: Run forward call. after_backward :: Run backward call. batch_end :: Run optimizer step. batch_valid_begin :: Run forward call validation loop. batch_valid_end :: Run forward call validation loop. valid_end :: Run end validation. epoch_end :: Run end epoch. end :: Run last epoch. exit :: Run last, using .exit().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Callback Desctiptor — torch_callback","text":"","code":"custom_tcb = torch_callback(\"custom\",   initialize = function(name) {     self$name = name   },   on_begin = function() {     cat(\"Hello\", self$name, \", we will train for \", self$ctx$total_epochs, \"epochs.\\n\")   },   on_end = function() {     cat(\"Training is done.\")   } )  learner = lrn(\"classif.torch_featureless\",   batch_size = 16,   epochs = 1,   callbacks = custom_tcb,   cb.custom.name = \"Marie\",   device = \"cpu\" ) task = tsk(\"iris\") learner$train(task) #> Hello Marie , we will train for  1 epochs. #> Training is done."},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"mlr3torch-012","dir":"Changelog","previous_headings":"","what":"mlr3torch 0.1.2","title":"mlr3torch 0.1.2","text":"CRAN release: 2024-10-15 Don’t use deprecated data_formats anymore Added CallbackSetTB, allows logging can viewed TensorBoard.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"mlr3torch-011","dir":"Changelog","previous_headings":"","what":"mlr3torch 0.1.1","title":"mlr3torch 0.1.1","text":"CRAN release: 2024-09-12 fix(preprocessing): regarding construction PipeOps po(\"trafo_resize\") failed cases. fix(ci): tests run CI fix(learner): LearnerTabResnet now works correctly Fix tests run CI feat: added nn() helper function simplify creation neural network layers","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"mlr3torch-010","dir":"Changelog","previous_headings":"","what":"mlr3torch 0.1.0","title":"mlr3torch 0.1.0","text":"CRAN release: 2024-07-08 Initial CRAN release","code":""}]
+[{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":null,"dir":"","previous_headings":"","what":"GNU Lesser General Public License","title":"GNU Lesser General Public License","text":"Version 3, 29 June 2007Copyright © 2007 Free Software Foundation, Inc. <http://fsf.org/> Everyone permitted copy distribute verbatim copies license document, changing allowed. version GNU Lesser General Public License incorporates terms conditions version 3 GNU General Public License, supplemented additional permissions listed .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_0-additional-definitions","dir":"","previous_headings":"","what":"0. Additional Definitions","title":"GNU Lesser General Public License","text":"used herein, “License” refers version 3 GNU Lesser General Public License, “GNU GPL” refers version 3 GNU General Public License. “Library” refers covered work governed License, Application Combined Work defined . “Application” work makes use interface provided Library, otherwise based Library. Defining subclass class defined Library deemed mode using interface provided Library. “Combined Work” work produced combining linking Application Library. particular version Library Combined Work made also called “Linked Version”. “Minimal Corresponding Source” Combined Work means Corresponding Source Combined Work, excluding source code portions Combined Work , considered isolation, based Application, Linked Version. “Corresponding Application Code” Combined Work means object code /source code Application, including data utility programs needed reproducing Combined Work Application, excluding System Libraries Combined Work.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_1-exception-to-section-3-of-the-gnu-gpl","dir":"","previous_headings":"","what":"1. Exception to Section 3 of the GNU GPL","title":"GNU Lesser General Public License","text":"may convey covered work sections 3 4 License without bound section 3 GNU GPL.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_2-conveying-modified-versions","dir":"","previous_headings":"","what":"2. Conveying Modified Versions","title":"GNU Lesser General Public License","text":"modify copy Library, , modifications, facility refers function data supplied Application uses facility (argument passed facility invoked), may convey copy modified version: ) License, provided make good faith effort ensure , event Application supply function data, facility still operates, performs whatever part purpose remains meaningful, b) GNU GPL, none additional permissions License applicable copy.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_3-object-code-incorporating-material-from-library-header-files","dir":"","previous_headings":"","what":"3. Object Code Incorporating Material from Library Header Files","title":"GNU Lesser General Public License","text":"object code form Application may incorporate material header file part Library. may convey object code terms choice, provided , incorporated material limited numerical parameters, data structure layouts accessors, small macros, inline functions templates (ten fewer lines length), following: ) Give prominent notice copy object code Library used Library use covered License. b) Accompany object code copy GNU GPL license document.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_4-combined-works","dir":"","previous_headings":"","what":"4. Combined Works","title":"GNU Lesser General Public License","text":"may convey Combined Work terms choice , taken together, effectively restrict modification portions Library contained Combined Work reverse engineering debugging modifications, also following: ) Give prominent notice copy Combined Work Library used Library use covered License. b) Accompany Combined Work copy GNU GPL license document. c) Combined Work displays copyright notices execution, include copyright notice Library among notices, well reference directing user copies GNU GPL license document. d) one following: 0) Convey Minimal Corresponding Source terms License, Corresponding Application Code form suitable , terms permit, user recombine relink Application modified version Linked Version produce modified Combined Work, manner specified section 6 GNU GPL conveying Corresponding Source. 1) Use suitable shared library mechanism linking Library. suitable mechanism one () uses run time copy Library already present user’s computer system, (b) operate properly modified version Library interface-compatible Linked Version. e) Provide Installation Information, otherwise required provide information section 6 GNU GPL, extent information necessary install execute modified version Combined Work produced recombining relinking Application modified version Linked Version. (use option 4d0, Installation Information must accompany Minimal Corresponding Source Corresponding Application Code. use option 4d1, must provide Installation Information manner specified section 6 GNU GPL conveying Corresponding Source.)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_5-combined-libraries","dir":"","previous_headings":"","what":"5. Combined Libraries","title":"GNU Lesser General Public License","text":"may place library facilities work based Library side side single library together library facilities Applications covered License, convey combined library terms choice, following: ) Accompany combined library copy work based Library, uncombined library facilities, conveyed terms License. b) Give prominent notice combined library part work based Library, explaining find accompanying uncombined form work.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/LICENSE.html","id":"id_6-revised-versions-of-the-gnu-lesser-general-public-license","dir":"","previous_headings":"","what":"6. Revised Versions of the GNU Lesser General Public License","title":"GNU Lesser General Public License","text":"Free Software Foundation may publish revised /new versions GNU Lesser General Public License time time. new versions similar spirit present version, may differ detail address new problems concerns. version given distinguishing version number. Library received specifies certain numbered version GNU Lesser General Public License “later version” applies , option following terms conditions either published version later version published Free Software Foundation. Library received specify version number GNU Lesser General Public License, may choose version GNU Lesser General Public License ever published Free Software Foundation. Library received specifies proxy can decide whether future versions GNU Lesser General Public License shall apply, proxy’s public statement acceptance version permanent authorization choose version Library.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/articles/callbacks.html","id":"building-blocks","dir":"Articles","previous_headings":"","what":"Building Blocks","title":"Custom Callbacks","text":"heart callback mechanism three R6 classes: CallbackSet contains methods executed different stages training loop TorchCallback wraps CallbackSet annotates meta information, including ParamSet ContextTorch defines information training process CallbackSet access using predefined callbacks, one usually interacts TorchCallback class, constructed using t_clbk(<id>): wrapped CallbackSet accessible via field $generator, can create new instance calling $new() Within training loop torch model, different stages exist callback can execute code. , list stages available, also described documentation CallbackSet. history callback, see runs code beginning, validation starts, end epoch.","code":"tc_hist = t_clbk(\"history\") tc_hist ## <TorchCallback:history> History ## * Generator: CallbackSetHistory ## * Parameters: list() ## * Packages: mlr3torch,torch cbs_hist = tc_hist$generator$new() mlr_reflections$torch$callback_stages ##  [1] \"on_begin\"             \"on_epoch_begin\"       \"on_batch_begin\"       ##  [4] \"on_after_backward\"    \"on_batch_end\"         \"on_before_valid\"      ##  [7] \"on_batch_valid_begin\" \"on_batch_valid_end\"   \"on_valid_end\"         ## [10] \"on_epoch_end\"         \"on_end\"               \"on_exit\" cbs_hist ## <CallbackSetHistory> ## * Stages: on_begin, on_before_valid, on_epoch_end"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/callbacks.html","id":"writing-a-custom-logger","dir":"Articles","previous_headings":"","what":"Writing a Custom Logger","title":"Custom Callbacks","text":"order define custom callback, can use helper torch_callback() helper function. example, create custom logger keeps track exponential moving average train loss prints end every epoch. callback takes one argument alpha smoothing parameter store moving average self$moving_loss. value alpha can later configured Learner. , implement main logic callback using two stages: on_batch_end(), called optimizer updates network parameters using mini-batch. , access last loss via ContextTorch’s (accessible via self$ctx) $last_loss field update value self$moving_loss. on_before_valid(), run validation loop. point simply print exponential moving average training loss. Finally, order make final value moving average accessible Learner training finishes, implement state_dict() method return value load_state_dict(state_dict) method, takes previously retrieved state dict sets callback. Note access final $moving_loss via learner, otherwise implement methods. created TorchCallback object associated CallbackSet R6-class us:","code":"custom_logger = torch_callback(\"custom_logger\",   initialize = function(alpha = 0.1) {     self$alpha = alpha     self$moving_loss = NULL   },   on_batch_end = function() {     if (is.null(self$moving_training_loss)) {       self$moving_loss = self$ctx$last_loss     } else {       self$moving_loss = self$alpha * self$last_loss + (1 - self$alpha) * self$moving_loss     }   },   on_before_valid = function() {     cat(sprintf(\"Epoch %s: %.2f\\n\", self$ctx$epoch, self$moving_loss))   },   state_dict = function() {     self$moving_loss   },   load_state_dict = function(state_dict) {     self$moving_loss = state_dict   } ) custom_logger ## <TorchCallback:custom_logger> Custom_logger ## * Generator: CallbackSetCustom_logger ## * Parameters: list() ## * Packages: mlr3torch,torch custom_logger$generator ## <CallbackSetCustom_logger> object generator ##   Inherits from: <inherit> ##   Public: ##     initialize: function (alpha = 0.1)  ##     state_dict: function ()  ##     load_state_dict: function (state_dict)  ##     on_before_valid: function ()  ##     on_batch_end: function ()  ##   Parent env: <environment: 0x55fef5065170> ##   Locked objects: FALSE ##   Locked class: FALSE ##   Portable: TRUE"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/callbacks.html","id":"using-the-custom-logger","dir":"Articles","previous_headings":"","what":"Using the Custom Logger","title":"Custom Callbacks","text":"order showcase effects, create train simple multi layer perceptron pass callback. set smoothing parameter alpha 0.05. information returnede state_dict() now accessible via Learner’s $model-slot:","code":"task = tsk(\"iris\") mlp = lrn(\"classif.mlp\",   callbacks = custom_logger, cb.custom_logger.alpha = 0.05,   epochs = 5, batch_size = 64, neurons = 20)  mlp$train(task) ## Epoch 1: 1.22 ## Epoch 2: 1.27 ## Epoch 3: 1.48 ## Epoch 4: 1.07 ## Epoch 5: 1.09 mlp$model$callbacks$custom_logger ## [1] 1.090038"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"quickstart","dir":"Articles","previous_headings":"","what":"Quickstart","title":"Get Started","text":"vignette show get started mlr3torch training simple neural network tabular regression problem. assume familiar mlr3 framework, see e.g. mlr3 book. first example, train simple multi-layer perceptron (MLP) well-known “mtcars” task, goal predict miles per galleon (‘mpg’) cars. architecture comes predfined learner mlr3torch, can also easily create new network architectures, see Neural Networks Graphs vignette detailed introduoduion. first set seed reproducibility, load library construct task. Learners mlr3torch work similary mlr3 learners. , construct simple multi layer perceptron regression. usual calling lrn() configuring parameters: use two hidden layers 50 neurons, training, set batch size 32, number training epochs 30 device \"cpu\". complete description available parameters see ?mlr3torch::LearnerTorchMLP. can use learner training prediction just like regression learner. , split observations training test set, train learner training set create predictions test set. Finally, compute mean squared error predictions.","code":"set.seed(314) library(mlr3torch) task = tsk(\"mtcars\") task$head() #>      mpg    am  carb   cyl  disp  drat  gear    hp  qsec    vs    wt #>    <num> <num> <num> <num> <num> <num> <num> <num> <num> <num> <num> #> 1:  21.0     1     4     6   160  3.90     4   110 16.46     0 2.620 #> 2:  21.0     1     4     6   160  3.90     4   110 17.02     0 2.875 #> 3:  22.8     1     1     4   108  3.85     4    93 18.61     1 2.320 #> 4:  21.4     0     1     6   258  3.08     3   110 19.44     1 3.215 #> 5:  18.7     0     2     8   360  3.15     3   175 17.02     0 3.440 #> 6:  18.1     0     1     6   225  2.76     3   105 20.22     1 3.460 mlp = lrn(\"regr.mlp\",   # architecture parameters   neurons = c(50, 50),   # training arguments   batch_size = 32, epochs = 30, device = \"cpu\" ) mlp #> <LearnerTorchMLP[regr]:regr.mlp>: My Little Powny #> * Model: - #> * Parameters: epochs=30, device=cpu, num_threads=1, #>   num_interop_threads=1, seed=random, eval_freq=1, #>   measures_train=<list>, measures_valid=<list>, patience=0, #>   min_delta=0, batch_size=32, neurons=50,50, p=0.5, #>   activation=<nn_relu>, activation_args=<list> #> * Validate: NULL #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response] #> * Feature Types: integer, numeric, lazy_tensor #> * Properties: internal_tuning, marshal, validation #> * Optimizer: adam #> * Loss: mse #> * Callbacks: - # Split the obersevations into training and test set splits = partition(task) # Train the learner on the train set mlp$train(task, row_ids = splits$train) # Predict the test set prediction = mlp$predict(task, row_ids = splits$test) # Compute the mse prediction$score(msr(\"regr.mse\")) #> regr.mse  #>  283.838"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"configuring-a-learner","dir":"Articles","previous_headings":"","what":"Configuring a Learner","title":"Get Started","text":"Although torch learners quite like mlr3 learners, differences. One LearnerTorch classes construction arguments, .e. torch learners modular learners. learners free implement construction arguments, common torch learners, namely loss, optimizer callbacks. object can parameters included LearnerTorch’s parameter set. previous example, specify explicitly used default values, Adam optimizer, MSE loss callbacks. now show configure three aspects learner mlr3torch::TorchOptimizer, mlr3torch::TorchLoss, mlr3torch::TorchCallback classes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"loss","dir":"Articles","previous_headings":"Configuring a Learner","what":"Loss","title":"Get Started","text":"loss function, also known objective function cost function, measures discrepancy predicted output true output. quantifies well model performing training. R package torch, underpins mlr3torch framework, already provides number predefined loss functions Mean Squared Error (nn_mse_loss), Mean Absolute Error (nn_l1_loss), cross entropy loss (nn_cross_entropy_loss). mlr3torch, represent loss functions using mlr3torch::TorchLoss class. provides thin wrapper around torch loss functions annotates meta information, importantly paradox::ParamSet allows configure loss function. object can constructed using t_loss(<key>). , construct L1 loss function, also known Mean Absolute Error (MAE). printed output informs us wrapped loss function (nn_l1_loss), configured parameters, packages depends task types can used. ParamSet contains one parameter, namely reduction, specifies loss reduced batch. wrapped loss module generator accessible slot $generator. can pass TorchLoss argument loss initialization learner. parameters loss added learner’s ParamSet, prefixed \"loss.\". predefined loss functions stored mlr3torch_losses dictionary, can retrieved using t_loss(<key>).","code":"l1 = t_loss(\"l1\") l1 #> <TorchLoss:l1> Absolute Error #> * Generator: nn_l1_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr # the paradox::ParamSet of the loss l1$param_set #> <ParamSet(1)> #>           id    class lower upper nlevels default  value #>       <char>   <char> <num> <num>   <num>  <list> <list> #> 1: reduction ParamFct    NA    NA       2    mean [NULL] l1$generator #> <nn_l1_loss> object generator #>   Inherits from: <inherit> #>   Public: #>     .classes: nn_l1_loss nn_loss nn_module #>     initialize: function (reduction = \"mean\")  #>     forward: function (input, target)  #>     clone: function (deep = FALSE, ..., replace_values = TRUE)  #>   Private: #>     .__clone_r6__: function (deep = FALSE)  #>   Parent env: <environment: 0x55ffc865fba8> #>   Locked objects: FALSE #>   Locked class: FALSE #>   Portable: TRUE mlp_l1 = lrn(\"regr.mlp\", loss = l1) mlp_l1$param_set$values$loss.reduction #> NULL mlr3torch_losses #> <DictionaryMlr3torchLosses> with 3 stored values #> Keys: cross_entropy, l1, mse"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"optimizer","dir":"Articles","previous_headings":"Configuring a Learner","what":"Optimizer","title":"Get Started","text":"optimizer determines model’s weights updated based calculated loss. adjusts parameters model minimize loss function, optimizing model’s performance. Optimizers work analogous loss functions, .e. mlr3torch provides thin wrapper – TorchOptimizer class – around optimizers Adam (optim_adam) SGD (optim_sgd). TorchLoss objects can constructed using t_opt(<key>). optimizers, associated ParamSet interesting see : wrapped torch optimizer can accessed slot generator. Parameters TorchOptimizer (also TorchLoss TorchCallback) can set usual mlr3 way, .e. either construction, afterwards using $set_values() method parameter set. see optimizer’s parameters added learner’s ParamSet (prefixed \"opt.\") values set values specified. exposing optimizer’s parameters, can conveniently tuned using mlr3tuning. available optimizers stored mlr3torch_optimizers dictionary.","code":"sgd = t_opt(\"sgd\") sgd #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_sgd #> * Parameters: list() #> * Packages: torch,mlr3torch  sgd$param_set #> <ParamSet(5)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1:           lr ParamDbl     0   Inf     Inf <NoDefault[0]> [NULL] #> 2:     momentum ParamDbl     0     1     Inf              0 [NULL] #> 3:    dampening ParamDbl     0     1     Inf              0 [NULL] #> 4: weight_decay ParamDbl     0     1     Inf              0 [NULL] #> 5:     nesterov ParamLgl    NA    NA       2          FALSE [NULL] sgd$param_set$set_values(   lr = 0.5, # increase learning rate   nesterov = FALSE # no nesterov momentum ) mlp_sgd = lrn(\"regr.mlp\", optimizer = sgd) as.data.table(mlp_sgd$param_set)[   startsWith(id, \"opt.\")][[1L]] #> [1] \"opt.lr\"           \"opt.momentum\"     \"opt.dampening\"    \"opt.weight_decay\" #> [5] \"opt.nesterov\" mlp_sgd$param_set$values[c(\"opt.lr\", \"opt.nesterov\")] #> $opt.lr #> [1] 0.5 #>  #> $opt.nesterov #> [1] FALSE mlr3torch_optimizers #> <DictionaryMlr3torchOptimizers> with 7 stored values #> Keys: adadelta, adagrad, adam, asgd, rmsprop, rprop, sgd"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"callbacks","dir":"Articles","previous_headings":"Configuring a Learner","what":"Callbacks","title":"Get Started","text":"third important configuration option callbacks allow customize training process. allows saving model checkpoints, logging metrics, implementing custom functionality specific training scenarios. tutorial implement custom callback, see Custom Callbacks vignette. , show use predefined callbacks. , retrieve \"history\" callback using t_clbk(), parameters merely saves training validation history learner can accessed afterwards. wanted learn callback , can access help page wrapped object using $help() method. Note also possible loss optimizer. predefined callbacks stored mlr3torch_callbacks dictionary.","code":"history = t_clbk(\"history\") history #> <TorchCallback:history> History #> * Generator: CallbackSetHistory #> * Parameters: list() #> * Packages: mlr3torch,torch history$help() mlr3torch_callbacks #> <DictionaryMlr3torchCallbacks> with 4 stored values #> Keys: checkpoint, history, progress, tb"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/get_started.html","id":"putting-it-together","dir":"Articles","previous_headings":"Configuring a Learner","what":"Putting it Together","title":"Get Started","text":"now define customized MLP learner using loss, optimizer callback just covered. keep track performance, use 30% training data validation evaluate using MAE Measure. Note mearures_valid measures_train parameters LearnerTorch take common mlr3::Measures, whereas loss function must TorchLoss. now train learner “mtcars” task use train-test split . make predictions unseen test data compare scores. directly optimized L1 (aka MAE) loss tweaked learning rate, configured mlp_custom learner lower MAE default mlp learner. configured learner use history callback, can find validation history $model slot: plot shows epochs 6 30.  important information stored Learner’s model $network, underlying nn_module. full description model, see ?LearnerTorch.","code":"mlp_custom = lrn(\"regr.mlp\",   # construction arguments   optimizer = sgd, loss = l1, callbacks = history,   # scores to keep track of   measures_valid = msr(\"regr.mae\"),   # other parameters are left as-is:   # architecture   neurons = c(50, 50),   # training arguments   batch_size = 32, epochs = 30, device = \"cpu\",   # validation proportion   validate = 0.3 )  mlp_custom #> <LearnerTorchMLP[regr]:regr.mlp>: My Little Powny #> * Model: - #> * Parameters: epochs=30, device=cpu, num_threads=1, #>   num_interop_threads=1, seed=random, eval_freq=1, #>   measures_train=<list>, measures_valid=<MeasureRegrSimple>, #>   patience=0, min_delta=0, batch_size=32, neurons=50,50, p=0.5, #>   activation=<nn_relu>, activation_args=<list>, opt.lr=0.5, #>   opt.nesterov=FALSE #> * Validate: 0.3 #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response] #> * Feature Types: integer, numeric, lazy_tensor #> * Properties: internal_tuning, marshal, validation #> * Optimizer: sgd #> * Loss: l1 #> * Callbacks: history mlp_custom$train(task, row_ids = splits$train) prediction_custom = mlp_custom$predict(task, row_ids = splits$test) prediction_custom$score(msr(\"regr.mae\")) #> regr.mae  #> 7.122375 prediction$score(msr(\"regr.mae\")) #> regr.mae  #> 15.27983 head(mlp_custom$model$callbacks$history$valid) #>    epoch     regr.mae #>    <num>        <num> #> 1:     1 1.777395e+04 #> 2:     2 3.955504e+08 #> 3:     3 1.143863e+04 #> 4:     4 1.792927e+01 #> 5:     5 1.759594e+01 #> 6:     6 1.726260e+01"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"a-torch-primer","dir":"Articles","previous_headings":"","what":"A torch Primer","title":"Internals","text":"start sampling input tensor: 2 batches 3 features: nn_module constructed nn_module_generator. nn_linear one simpler generators: Applying module gives 2-batch 4 units: neural network one (4-unit) hidden layer two outputs needs following ingredients can pipe tensor layers follows. now continue showing neural network can represented mlr3torch.","code":"input = torch_randn(2, 3) input #> torch_tensor #>  0.4430 -1.7493  0.3801 #>  2.5443  0.4085 -0.4808 #> [ CPUFloatType{2,3} ] module_1 = nn_linear(in_features = 3, out_features = 4, bias = TRUE) output = module_1(input) output #> torch_tensor #>  0.6725  0.3022 -0.3800  0.3413 #> -0.8161  0.3485  0.3559 -1.3795 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] activation = nn_sigmoid() module_2 = nn_linear(4, 3, bias = TRUE) softmax = nn_softmax(2) output = module_1(input) output = activation(output) output = module_2(output) output = softmax(output) output #> torch_tensor #>  0.2155  0.3400  0.4445 #>  0.1626  0.3668  0.4706 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"neural-networks-as-graphs","dir":"Articles","previous_headings":"","what":"Neural Networks as Graphs","title":"Internals","text":"mlr3torch, nn_modules wrapped PipeOpModule. advantage network structure can represented mlr3pipelines::Graph object made explicit (can plotted, can extended manipulated), compared e.g. writing function pipes input series modules. PipeOpModule can used wrap module directly, usually constructed PipeOpTorch (see later). typically single input single output, although multiple inputs possible (module called multiple arguments), multiple outputs possible module-function returns list. input output channels must explicitly declared construction. now continue recreate network using PipeOpModules. can wrap linear module_1 layers like : Note po(\"module_1\") equivalent po(\"module\", id = \"module_1\"). mechanism convenient avoid ID clashes graphs contain PipeOp multiple times. can use generated PipeOp familiar way: Note use $train(), since torch modules anything maps state (filled empty list). single hidden layer neural network can constructed Graph, can training . can now use graph’s $train() method pipe tensor whole Graph. object allows easily perform forward pass, inherit nn_module, useful various reasons. Instead class inherits nn_module Graph (work R6, since multiple inheritance available), class inherits nn_module contains Graph member slot composition. class nn_graph. constructed Graph, well information shape(s) torch_tensor(s) expects inputs. Shape info communicated integer-valued numeric vector; dimensions arbitrary, e.g. batch-size, given NA. network expects input shape c(NA, 3), since first layer created nn_linear(in_features = 3, ...). Graph multiple outputs, also possible select subset outputs use, change output order, giving output_map argument. module gives us convenience torch nn_module objects, e.g.: can used transform tensors just torch::nn_module:","code":"library(mlr3torch) po_module_1 = po(\"module_1\", module = module_1) output = po_module_1$train(list(input))[[1]] output #> torch_tensor #>  0.6725  0.3022 -0.3800  0.3413 #> -0.8161  0.3485  0.3559 -1.3795 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] po_activation = po(\"module\", id = \"activation\", activation) po_module_2 = po(\"module_2\", module = module_2) po_softmax = po(\"module\", id = \"softmax\", module = softmax)  module_graph = po_module_1 %>>%   po_activation %>>%   po_module_2 %>>%   po_softmax  module_graph$plot(html = TRUE) output = module_graph$train(input)[[1]] output #> torch_tensor #>  0.2155  0.3400  0.4445 #>  0.1626  0.3668  0.4706 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ] # the name of the single input is: module_graph$input #>              name        train predict    op.id channel.name #>            <char>       <char>  <char>   <char>       <char> #> 1: module_1.input torch_tensor    NULL module_1        input  graph_module = nn_graph(   module_graph,   shapes_in = list(module_1.input = c(NA, 3)) ) graph_module$children #> $module_list #> An `nn_module` containing 31 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • 0: <nn_linear> #16 parameters #> • 1: <nn_sigmoid> #0 parameters #> • 2: <nn_linear> #15 parameters #> • 3: <nn_softmax> #0 parameters graph_module(input) #> torch_tensor #>  0.2155  0.3400  0.4445 #>  0.1626  0.3668  0.4706 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"modeldescriptor","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"ModelDescriptor","title":"Internals","text":"PipeOpModule represents nn_module fixed specific tensor shape hyperparameters. constructing neural network using operators, one take care output shape operations match input shapes following operations. complete Graph matching PipeOpModules can constructed using operators mostly inherit PipeOpTorch, making use ModelDescriptor class. ModelDescriptor class contains Graph (mostly) PipeOpModules information. PipeOpTorch transforms ModelDescriptor adds PipeOpModules Graph. ModelDescriptors always build Graph specific Task. easiest way initialize proper ModelDescriptor use appropriate PipeOpTorchIngress given datatype. use PipeOpTorchIngressNumeric, used numeric data. ModelDescriptor S3 object contains Graph, information generate data ($ingress $task), tags build model unrelated architecture ($optimizer, $loss $callbacks) well information necessary extend graph along given output ($pointer $pointer_shape). $pointer identifies output $graph PipeOpTorch extend. Piping ModelDescriptor PipeOpTorchLinear, example, adds PipeOpModule wrapping torch::nn_linear. $pointer now updated identify output PipeOpModule, $pointer_shape shows shape changed 4 units (3 input ). model_descriptor_to_module() function converts nn_graph, functional torch::nn_module.","code":"task = tsk(\"iris\")$select(colnames(iris)[1:3])  po_torch_in = po(\"torch_ingress_num\") md = po_torch_in$train(list(task))[[1]]  md #> <ModelDescriptor: 1 ops> #> * Ingress:  torch_ingress_num.input: [(NA,3)] #> * Task:  iris [classif] #> * Callbacks:  N/A #> * Optimizer:  N/A #> * Loss:  N/A #> * pointer:  torch_ingress_num.output [(NA,3)] unclass(md) #> $graph #> Graph with 1 PipeOps: #>                 ID         State sccssors prdcssors #>             <char>        <char>   <char>    <char> #>  torch_ingress_num <<UNTRAINED>>                    #>  #> $ingress #> $ingress$torch_ingress_num.input #> Ingress: Task[Petal.Length,Sepal.Length,Sepal.Width] --> Tensor(NA, 3) #>  #>  #> $task #> <TaskClassif:iris> (150 x 4): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (3): #>   - dbl (3): Petal.Length, Sepal.Length, Sepal.Width #>  #> $optimizer #> NULL #>  #> $loss #> NULL #>  #> $callbacks #> named list() #>  #> $pointer #> [1] \"torch_ingress_num\" \"output\"            #>  #> $pointer_shape #> [1] NA  3 po_torch_linear = po(\"nn_linear\", out_features = 4) md = po_torch_linear$train(list(md))[[1]]  md$graph #> Graph with 2 PipeOps: #>                 ID         State  sccssors         prdcssors #>             <char>        <char>    <char>            <char> #>  torch_ingress_num <<UNTRAINED>> nn_linear                   #>          nn_linear <<UNTRAINED>>           torch_ingress_num md$pointer #> [1] \"nn_linear\" \"output\" md$pointer_shape #> [1] NA  4 small_module = model_descriptor_to_module(md, list(md$pointer))  small_module(input) #> torch_tensor #> -0.1274  0.1809 -1.4397 -0.5290 #>  0.6121 -0.7253 -0.5453 -0.2486 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"using-modeldescriptor-to-get-data","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"Using ModelDescriptor to get Data","title":"Internals","text":"ModelDescriptor represent Graph nn_module created, also way Task processed get input batches. torch::dataset can created calling task_dataset(); task feature_ingress_tokens arguments can retrieved ModelDescriptor. target_batchgetter needs created extra (necessary), since depends ultimate machine learning model, looked far. Use $.getbatch() method get batch can given nn_module. Note $x $y slot, latter used, account possible target batches. $x slot also list, since able handle NNs multiple inputs (see ).","code":"td = task_dataset(   task = md$task,   feature_ingress_tokens = md$ingress,   device = \"cpu\" )  td #> <dataset> #>   Public: #>     .getbatch: function (index)  #>     .getitem: function (index)  #>     .length: function ()  #>     all_features: Petal.Length Sepal.Length Sepal.Width Species #>     cache_lazy_tensors: FALSE #>     clone: function (deep = FALSE)  #>     device: cpu #>     feature_ingress_tokens: list #>     initialize: function (task, feature_ingress_tokens, target_batchgetter = NULL,  #>     load_state_dict: function (x, ..., .refer_to_state_dict = FALSE)  #>     state_dict: function ()  #>     target_batchgetter: NULL #>     task: TaskClassif, TaskSupervised, Task, R6 batch = td$.getbatch(1:3) batch #> $x #> $x$torch_ingress_num.input #> torch_tensor #>  1.4000  5.1000  3.5000 #>  1.4000  4.9000  3.0000 #>  1.3000  4.7000  3.2000 #> [ CPUFloatType{3,3} ] #>  #>  #> $.index #> torch_tensor #>  1 #>  2 #>  3 #> [ CPULongType{3} ]  small_module(batch$x[[1]]) #> torch_tensor #>  0.3122 -4.8047  0.3374  4.3038 #>  0.4304 -4.4378  0.4278  3.9666 #>  0.3240 -4.4393  0.2842  3.9762 #> [ CPUFloatType{3,4} ][ grad_fn = <AddmmBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"building-sequential-nns","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"Building sequential NNs","title":"Internals","text":"sequential NN can easily implemented follows: Note second nn_linear need informed output dimension first nn_linear, since ModelDescriptor passed along Graph edges knows info ($pointer_shape slot).","code":"graph_generator = po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 4, id = \"linear1\") %>>%     po(\"nn_sigmoid\") %>>%   po(\"nn_linear\", out_features = 3, id = \"linear2\") %>>%   po(\"nn_softmax\", dim = 2) md_sequential = graph_generator$train(task)[[1]]  graph_module = model_descriptor_to_module(md_sequential, list(md_sequential$pointer))  graph_module(input) #> torch_tensor #>  0.2329  0.2873  0.4798 #>  0.1835  0.2948  0.5217 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"building-more-interesting-nns","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"Building more interesting NNs","title":"Internals","text":"One main features mlr3pipelines ability easily represent computational Graphs. ModelDescriptor / PipeOpTorch setup built make full use functionality. possible multiple inputs NN using multiple PipeOpTorchIngress inputs, possible parallel alternative path branching, possible multiple outputs. Consider following (bit nonsensical) network operates differently \"Petal\" \"Sepal\" features tsk(\"iris\") manually split task , shown wholly integrated mlr3pipelines pipeline can automatically. can use create neural network iris tasks created . set $keep_results debug flag can inspection happening: make multiple observations : can observe ModelDescriptor grows passed along edges graph_iris. Note $graph slot ModelDescriptor often updated -reference, time inspect intermediate results, may contain complete graph. However, see $ingress, $pointer $pointer_shape ModelDescriptors take sepal.-path differ ones take petal.-path: po(\"nn_merge_cat\") unites two ModelDescriptors contains common ingress. pointer_shape now reflects output “cat”-operation: 2nd dimension added : Multiple ModelDescriptors created, since graph_iris multiple outpus. makes possible create neural network multiple outputs. need unite outputs graph_iris using model_descriptor_union() can pass model_descriptor_to_module(). need collect output_pointers separately. parameter list_output must set TRUE since module multiple outputs. PipeOpBranch disappears resulting Graph PipeOpModule iris_module. PipeOpTorchs graph_iris add anything ModelDescriptors. branch interpeted graph_iris runs, nn_relu path followed. iris_module therefore contains Graph “relu” activation: ModelDescriptor’s $task slot contains Task features used create input data NN inputs. can given task_dataset(), along $ingress, create torch dataset creates batches. , output graph_iris can used: resulting module multiple inputs multiple outputs. call first two rows iris, set debug $keep_results flag can inspect happening nn_module’s $graph: first linear layer takes “Sepal” input (\"linear1\") creates 2x4 tensor (batch size 2, 4 units), \"linear3\" layer 2x5 output: observe po(\"nn_merge_cat\") concatenates , expected:","code":"iris_petal = tsk(\"iris\")$select(c(\"Petal.Length\", \"Petal.Width\")) iris_sepal = tsk(\"iris\")$select(c(\"Sepal.Length\", \"Sepal.Width\")) graph_sepal = po(\"torch_ingress_num\", id = \"sepal.in\") %>>%   po(\"nn_linear\", out_features = 4, id = \"linear1\")  graph_petal = po(\"torch_ingress_num\", id = \"petal.in\") %>>%   po(\"nn_linear\", out_features = 3, id = \"linear2\") %>>%   po(\"nn_tanh\") %>>%   po(\"nn_linear\", out_features = 5, id = \"linear3\")  graph_common = ppl(\"branch\", graphs = list(     sigmoid = po(\"nn_sigmoid\"),     relu = po(\"nn_relu\")   )) %>>%   gunion(list(     po(\"nn_linear\", out_features = 1, id = \"lin_out\"),     po(\"nn_linear\", out_features = 3, id = \"cat_out\") %>>%       po(\"nn_softmax\", dim = 2)   ))   graph_iris = gunion(list(graph_sepal, graph_petal)) %>>%   po(\"nn_merge_cat\") %>>%   graph_common  graph_iris$plot(html = TRUE) graph_iris$param_set$values$branch.selection = \"relu\"  graph_iris$keep_results = TRUE  iris_mds = graph_iris$train(   input = list(sepal.in.input = iris_sepal, petal.in.input = iris_petal),   single_input = FALSE )  iris_mds #> $lin_out.output #> <ModelDescriptor: 11 ops> #> * Ingress:  sepal.in.input: [(NA,2)], petal.in.input: [(NA,2)] #> * Task:  iris [classif] #> * Callbacks:  N/A #> * Optimizer:  N/A #> * Loss:  N/A #> * pointer:  lin_out.output [(NA,1)] #>  #> $nn_softmax.output #> <ModelDescriptor: 11 ops> #> * Ingress:  sepal.in.input: [(NA,2)], petal.in.input: [(NA,2)] #> * Task:  iris [classif] #> * Callbacks:  N/A #> * Optimizer:  N/A #> * Loss:  N/A #> * pointer:  nn_softmax.output [(NA,3)] # sepal.in path graph_iris$pipeops$linear1$.result[[1]]$ingress #> $sepal.in.input #> Ingress: Task[Sepal.Length,Sepal.Width] --> Tensor(NA, 2) graph_iris$pipeops$linear1$.result[[1]]$pointer #> [1] \"linear1\" \"output\" graph_iris$pipeops$linear1$.result[[1]]$pointer_shape #> [1] NA  4  # petal.in path graph_iris$pipeops$linear3$.result[[1]]$ingress #> $petal.in.input #> Ingress: Task[Petal.Length,Petal.Width] --> Tensor(NA, 2) graph_iris$pipeops$linear3$.result[[1]]$pointer #> [1] \"linear3\" \"output\" graph_iris$pipeops$linear3$.result[[1]]$pointer_shape #> [1] NA  5 graph_iris$pipeops$nn_merge_cat$.result[[1]]$ingress #> $sepal.in.input #> Ingress: Task[Sepal.Length,Sepal.Width] --> Tensor(NA, 2) #>  #> $petal.in.input #> Ingress: Task[Petal.Length,Petal.Width] --> Tensor(NA, 2) graph_iris$pipeops$nn_merge_cat$.result[[1]]$pointer_shape #> [1] NA  9 iris_mds_union = model_descriptor_union(iris_mds[[1]], iris_mds[[2]]) output_pointers = list(iris_mds[[1]]$pointer, iris_mds[[2]]$pointer) output_pointers #> [[1]] #> [1] \"lin_out\" \"output\"  #>  #> [[2]] #> [1] \"nn_softmax\" \"output\" iris_module = model_descriptor_to_module(iris_mds_union, output_pointers, list_output = TRUE) iris_module$graph$plot(html = TRUE) iris_mds_union$task  # contains all features #> <TaskClassif:iris> (150 x 5): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (4): #>   - dbl (4): Petal.Length, Petal.Width, Sepal.Length, Sepal.Width  iris_td = task_dataset(   task = iris_mds_union$task,   feature_ingress_tokens = iris_mds_union$ingress,   device = \"cpu\" )  batch = iris_td$.getbatch(1:2) batch #> $x #> $x$sepal.in.input #> torch_tensor #>  5.1000  3.5000 #>  4.9000  3.0000 #> [ CPUFloatType{2,2} ] #>  #> $x$petal.in.input #> torch_tensor #>  1.4000  0.2000 #>  1.4000  0.2000 #> [ CPUFloatType{2,2} ] #>  #>  #> $.index #> torch_tensor #>  1 #>  2 #> [ CPULongType{2} ] iris_module$graph$keep_results = TRUE  iris_module(   sepal.in.input = batch$x$sepal.in.input,   petal.in.input = batch$x$petal.in.input ) #> $lin_out.output #> torch_tensor #>  0.6367 #>  0.6419 #> [ CPUFloatType{2,1} ][ grad_fn = <AddmmBackward0> ] #>  #> $nn_softmax.output #> torch_tensor #>  0.2731  0.3325  0.3944 #>  0.2820  0.3441  0.3739 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ] iris_module$graph$pipeops$linear1$.result #> $output #> torch_tensor #> -3.2734  1.7562  1.3484  2.9473 #> -2.9881  1.8728  1.3176  2.7498 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$linear3$.result #> $output #> torch_tensor #>  0.1362  0.5117 -0.3741  0.8576  0.3130 #>  0.1362  0.5117 -0.3741  0.8576  0.3130 #> [ CPUFloatType{2,5} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$nn_merge_cat$.result #> $output #> torch_tensor #> -3.2734  1.7562  1.3484  2.9473  0.1362  0.5117 -0.3741  0.8576  0.3130 #> -2.9881  1.8728  1.3176  2.7498  0.1362  0.5117 -0.3741  0.8576  0.3130 #> [ CPUFloatType{2,9} ][ grad_fn = <CatBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"building-torch-learners","dir":"Articles","previous_headings":"","what":"Building Torch Learners","title":"Internals","text":"now seen NN Graphs PipeOpModule created turned nn_modules. Using PipeOpTorch even creates ModelDescriptor objects contain additional info batch tensors extracted Tasks. complete Learner, still necessary define loss-function used optimization, optimizer, optionally callbacks. already covered class representations – TorchLoss, TorchOptimizer, TorchCallbacks, Get Started vignette. use adam optimizer, cross-entropy loss function, history callback.","code":"adam = t_opt(\"adam\", lr = 0.02) adam #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: lr=0.02 #> * Packages: torch,mlr3torch  xe = t_loss(\"cross_entropy\") xe #> <TorchLoss:cross_entropy> Cross Entropy #> * Generator: nn_cross_entropy_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: classif  history = t_clbk(\"history\") history #> <TorchCallback:history> History #> * Generator: CallbackSetHistory #> * Parameters: list() #> * Packages: mlr3torch,torch"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"learnertorchmodel","dir":"Articles","previous_headings":"Building Torch Learners","what":"LearnerTorchModel","title":"Internals","text":"LearnerTorchModel represents supervised model (regression classification) using torch NNs. needs nn_module, well list TorchIngressToken define batches created Task. TorchIngressToken hard-code column-names Task used data-input, Learner created like therefore works specific Task created. (Generally full mlr3pipelines-UI used problem, see .) following uses sequential NN : training model, set hyperparameters. following calls $predict_newdata function plot response surface along Sepal.Width = mean(Sepal.Width) plane, along ground-truth values:","code":"lr_sequential = lrn(\"classif.torch_model\",   task_type = \"classif\",   network = model_descriptor_to_module(md_sequential, list(md_sequential$pointer)),   ingress_tokens = md_sequential$ingress,   optimizer = adam,   callbacks = history,   loss = xe )  lr_sequential #> <LearnerTorchModel[classif]:classif.model>: Torch Model #> * Model: - #> * Parameters: device=auto, num_threads=1, num_interop_threads=1, #>   seed=random, eval_freq=1, measures_train=<list>, #>   measures_valid=<list>, patience=0, min_delta=0, opt.lr=0.02 #> * Validate: NULL #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response], prob #> * Feature Types: logical, integer, numeric, character, factor, ordered, #>   POSIXct, lazy_tensor #> * Properties: featureless, hotstart_backward, hotstart_forward, #>   importance, internal_tuning, marshal, missings, multiclass, #>   oob_error, selected_features, twoclass, validation, weights #> * Optimizer: adam #> * Loss: cross_entropy #> * Callbacks: history lr_sequential$param_set$set_values(   batch_size = 50,   epochs = 100,   measures_train = msrs(c(\"classif.logloss\", \"classif.ce\")) ) # This is required to evaluate the logloss during training lr_sequential$predict_type = \"prob\"  lr_sequential$train(md_sequential$task) library(data.table) library(ggplot2)  newdata = cbind(data.table(Sepal.Width = mean(iris$Sepal.Width)), CJ(   Sepal.Length = seq(min(iris$Sepal.Length), max(iris$Sepal.Length), length.out = 30),   Petal.Length = seq(min(iris$Petal.Length), max(iris$Petal.Length), length.out = 30) ))  predictions = lr_sequential$predict_newdata(newdata)  plot_predictions = function(predictions) {   ggplot(cbind(newdata, Species = predictions$response),       aes(x = Sepal.Length, y = Petal.Length, fill = Species)) +     geom_tile(alpha = .3) +     geom_tile(alpha = .3) +     geom_point(data = iris,       aes(x = Sepal.Length, y = Petal.Length, fill = Species),       color = \"black\", pch = 21, size = 3) +     theme_bw() } plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"torch-learner-pipelines","dir":"Articles","previous_headings":"","what":"Torch Learner Pipelines","title":"Internals","text":"model shown constructed using ModelDescriptor generated Graph PipeOpTorch operators. ModelDescriptor furthermore contains Task pertains. makes possible use create NN model gets trained right away, using PipeOpTorchModelClassif. missing prerequisite now add desired TorchOptimizer TorchLoss information ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"adding-optimizer-loss-and-callback-meta-info-to-modeldescriptor","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"Adding Optimizer, Loss and Callback Meta-Info to ModelDescriptor","title":"Internals","text":"Remember ModelDescriptor $optimizer, $loss $callbacks slots necessary build complete Learner NN. can set corresponding PipeOpTorch operators. po(\"torch_optimizer\") used set $optimizer slot ModelDescriptor; takes desired TorchOptimizer object construction exports ParamSet. works analogously loss-function. also callbacks:","code":"po_adam = po(\"torch_optimizer\", optimizer = adam)  # hyperparameters are made available and can be changed: po_adam$param_set$values #> $lr #> [1] 0.02  md_sequential = po_adam$train(list(md_sequential))[[1]]  md_sequential$optimizer #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: lr=0.02 #> * Packages: torch,mlr3torch po_xe = po(\"torch_loss\", loss = xe) md_sequential = po_xe$train(list(md_sequential))[[1]]  md_sequential$loss #> <TorchLoss:cross_entropy> Cross Entropy #> * Generator: nn_cross_entropy_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: classif po_history = po(\"torch_callbacks\", callbacks = t_clbk(\"history\")) md_sequential = po_history$train(list(md_sequential))[[1]]  md_sequential$callbacks #> $history #> <TorchCallback:history> History #> * Generator: CallbackSetHistory #> * Parameters: list() #> * Packages: mlr3torch,torch"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"combined-instantiation-and-training-of-learnertorchmodel","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"Combined Instantiation and Training of LearnerTorchModel","title":"Internals","text":"ModelDescriptor can now given po(\"torch_model_classif\"). po(\"torch_model_classif\") behaves similarly PipeOpLearner: returns NULL training, prediction $predict(). po(\"torch_model_classif\") behaves similarly PipeOpLearner: returns NULL training, prediction $predict().","code":"po_model = po(\"torch_model_classif\", batch_size = 50, epochs = 50)  po_model$train(list(md_sequential)) #> $output #> NULL newtask = TaskClassif$new(\"newdata\", cbind(newdata, Species = factor(NA, levels = levels(iris$Species))), target = \"Species\")  predictions = po_model$predict(list(newtask))[[1]]  plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"the-whole-pipeline","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"The whole Pipeline","title":"Internals","text":"Remember md_sequential created using Graph initial Task piped . combine Graph PipeOpTorchModelClassif, get Graph behaves like Graph ends PipeOpLearner, can therefore wrapped GraphLearner. following uses one hidden layer : Compare resulting Graph Graph trained model: Predictions, (can use predict_newdata ):","code":"graph_sequential_full = po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 4, id = \"linear1\") %>>%     po(\"nn_sigmoid\") %>>%     po(\"nn_linear\", out_features = 3, id = \"linear2\") %>>%     po(\"nn_softmax\", dim = 2, id = \"softmax\") %>>%     po(\"nn_linear\", out_features = 3, id = \"linear3\") %>>%     po(\"nn_softmax\", dim = 2, id = \"softmax2\") %>>%     po(\"torch_optimizer\", optimizer = adam) %>>%     po(\"torch_loss\", loss = xe) %>>%     po(\"torch_callbacks\", callbacks = history) %>>%     po(\"torch_model_classif\", batch_size = 50, epochs = 100) lr_sequential_full = as_learner(graph_sequential_full)  lr_sequential_full$train(task) graph_sequential_full$plot(html = TRUE) model = lr_sequential_full$graph_model$state$torch_model_classif$model model$network$graph$plot(html = TRUE) predictions = lr_sequential_full$predict_newdata(newdata)  plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"mixed-pipelines","dir":"Articles","previous_headings":"Torch Learner Pipelines","what":"Mixed Pipelines","title":"Internals","text":"just limited PipeOpTorch kinds Graphs, also limited single PipeOpTorchIngress. following pipeline, example, removes Petal.Length columns Task fits model:  using Petal.Length Sepal.Length separately first?  examples hopefully demonstrated possibilities come representation neural network layers PipeOps. Even though vignette quite technical, hope given -depth understanding underlying mechanisms.","code":"gr = po(\"select\", selector = selector_name(\"Petal.Length\")) %>>%   po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 5, id = \"linear1\") %>>%   po(\"nn_relu\") %>>%   po(\"nn_linear\", out_features = 3, id = \"linear2\") %>>%   po(\"nn_softmax\", dim = 2) %>>%   po(\"torch_optimizer\", optimizer = adam) %>>%   po(\"torch_loss\", loss = xe) %>>%   po(\"torch_model_classif\", batch_size = 50, epochs = 50)  gr$plot(html = TRUE) lr = as_learner(gr) lr$train(task) predictions = lr$predict_newdata(newdata)  plot_predictions(predictions) gr = gunion(list(       po(\"select\", selector = selector_name(\"Petal.Length\"), id = \"sel1\") %>>%         po(\"torch_ingress_num\", id = \"ingress.petal\") %>>%         po(\"nn_linear\", out_features = 3, id = \"linear1\"),       po(\"select\", selector = selector_name(\"Sepal.Length\"), id = \"sel2\") %>>%         po(\"torch_ingress_num\", id = \"ingress.sepal\") %>>%         po(\"nn_linear\", out_features = 3, id = \"linear2\")     )) %>>%     po(\"nn_merge_cat\")  %>>%     po(\"nn_relu\", id = \"act1\") %>>%     po(\"nn_linear\", out_features = 3, id = \"linear3\") %>>%     po(\"nn_softmax\", dim = 2, id = \"act3\") %>>%     po(\"torch_optimizer\", optimizer = adam, lr = 0.1) %>>%     po(\"torch_loss\", loss = xe) %>>%     po(\"torch_model_classif\", batch_size = 50, epochs = 50)  gr$plot(html = TRUE) lr = as_learner(gr) lr$train(task) predictions = lr$predict_newdata(newdata)  plot_predictions(predictions)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/lazy_tensor.html","id":"creating-a-lazy-tensor","dir":"Articles","previous_headings":"","what":"Creating a Lazy Tensor","title":"Lazy Tensor","text":"Every lazy_tensor built top torch::dataset, assume familiar . information create torch::datasets, recommend reading torch package documentation. additional restriction impose dataset must .getitem .getbatch method returns list named tensors. example, create lazy_tensor length 1000, whose elements drawn uniform distribution \\([0, 1]\\). data stored -memory example, necessary $.getitem() method can e.g. load images disk. order create lazy_tensor mydata, annotate returned shapes dataset passing named list dataset_shapes. first dimension must NA, batch dimension. can also set shape NULL indicate unknown, .e. varies elements. Note case, implemented .getbatch method, even omitted specifying dataset_shapes auto-inferred. can convert vector torch_tensor just like : added preprocessing, calling $.getbatch() method mydata selecting element x. continue creating example task lt, relationship x y variable polynomial. Note target variable, classification regression, lazy_tensor, must factor numeric respectively. , plot data:  next section, create custom PipeOp fit polynomial regression model.","code":"mydata = dataset(   initialize = function() {     self$x = runif(1000, -1, 1)   },   .getbatch = function(i) list(x = torch_tensor(self$x[i])$unsqueeze(2)),   .length = function() 1000 )() lt = as_lazy_tensor(mydata, dataset_shapes = list(x = c(NA, 1))) lt[1:5] #> <ltnsr[5]> #> [1] <tnsr[1]> <tnsr[1]> <tnsr[1]> <tnsr[1]> <tnsr[1]> materialize(lt[1:5], rbind = TRUE) #> torch_tensor #> -0.9852 #> -0.0672 #> -0.0044 #> -0.4205 #>  0.4658 #> [ CPUFloatType{5,1} ] torch_equal(   materialize(lt[1], rbind = TRUE),   mydata$.getbatch(1)$x ) #> [1] TRUE library(data.table) x = mydata$x y = 0.2 + 0.1 * x - 0.1 * x^2 - 0.3 * x^3 + 0.5 * x^4 + 0.5 * x^7 + 0.6 * x^11 + rnorm(length(mydata)) * 0.1 dt = data.table(y = y, x = lt) task_poly = as_task_regr(dt, target = \"y\", id = \"poly\") task_poly #> <TaskRegr:poly> (1000 x 2) #> * Target: y #> * Properties: - #> * Features (1): #>   - lt (1): x library(ggplot2) ggplot(data = data.frame(x = x, y = y)) +   geom_point(aes(x = x, y = y), alpha = 0.5)"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/lazy_tensor.html","id":"custom-preprocessing","dir":"Articles","previous_headings":"","what":"Custom Preprocessing","title":"Lazy Tensor","text":"order create custom preprocessing operator lazy tensor, create new PipeOp class. make convenient possible, mlr3torch offters pipeop_preproc_torch() function recommend using purpose. important arguments : id - Used default identifier PipeOp fn - preprocessing function. default, first argument assumed torch_tensor remaining arguments part PipeOp’s parameter set. shapes_out - function returns shapes output tensors given input shapes. can also set NULL unknown shape \"infer\" auto-inference, see ?pipeop_preproc_torch information. , create PipeOp, transforms vector x matrix \\((x^{d_1} ..., x^{d_n})\\), \\(d\\) degrees parameter PipeOp. can now create new instance PipeOp calling $new(), set parameter degrees degrees used simulating data . , set parameter stages, always available, \"\", means preprocessing applied training prediction. data augmentation can set \"train\". create polynomial regression learner, combine polynomial preprocessor lrn(\"regr.mlp\") hidden layer (.e. linear model) train learner task. , visualize predictions see model captured non-linear relationship reasonably:  next section, briefly cover implementation details lazy_tensor, necessary work data-type, feel free skip part.","code":"PipeOpPreprocTorchPoly = pipeop_preproc_torch(\"poly\",   fn = function(x, degrees) {     torch_cat(lapply(degrees, function(d) torch_pow(x, d)), dim = 2L)   },   shapes_out = \"infer\" ) po_poly = PipeOpPreprocTorchPoly$new()  po_poly$param_set$set_values(   degrees = c(0, 1, 2, 3, 4, 7, 11),   stages = \"both\" ) lrn_poly = as_learner(   po_poly %>>% lrn(\"regr.mlp\", batch_size = 256, epochs = 100,   neurons = integer(0)) )  lrn_poly$train(task_poly) pred = lrn_poly$predict(task_poly) dt = melt(data.table(   truth = pred$truth,   response = pred$response,   x = x),   id.vars = \"x\", measure.vars = c(\"truth\", \"response\") ) dt$variable = factor(dt$variable, levels = c(\"truth\", \"response\"))  ggplot(data = dt) +   geom_point(aes(x = x, y = value, color = variable))"},{"path":"https://mlr3torch.mlr-org.com/dev/articles/lazy_tensor.html","id":"digging-into-internals","dir":"Articles","previous_headings":"","what":"Digging Into Internals","title":"Lazy Tensor","text":"Internally, lazy_tensor vector uses DataDescriptor class represent (possibly) preprocessed data. similar ModelDescriptor class used build neural nerworks using PipeOpTorch objects. DataDescriptor stores torch::dataset, mlr3pipelines::Graph metadata. Per default, preprocessing graph contains single PipOpNop nothing. printed output data descriptor informs us : number PipeOps contained preprocessing graph output shapes dataset input map, .e. data passed preprocessing graph, important multiple inputs pointer, points specific channel output PipeOp. output channel tensor represented DataDescriptor. Note id input po(\"nop\") randomly generated, needed prevent id clashes one input preprocessing graph. shape, shape tensor position pointer lazy tensor can constructed integer vector DataDescriptor. integer vector specifies element DataDescriptor lazy_tensor contains. , first two elements lazy_tensor vector represent element DataDescriptor, third element represents different element. Note indices refer DataDescriptor. Internally, lazy tensor represented list lists, element containing id DataDescriptor Currently, can single DataDescriptor lazy_tensor vector. happens materialize(lt[1]) following: Preprocessing lazy_tensor vector adds new PipeOps preprocessing graph updates metainformation like pointer output shape. show , create simple example task, using lt vector feature. see $graph new pipeop id \"poly.x\" output pointer points poly.x. Also see shape tensor now c(NA, 7) c(NA, 1) , can verify calling materialize() :","code":"desc = DataDescriptor$new(   dataset = mydata,   dataset_shapes = list(x = c(NA, 1)) ) desc #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,1)] #> * input_map: (x) -> Graph #> * pointer: nop.181509.x.output #> * shape: [(NA,1)] lt = lazy_tensor(desc, ids = c(1, 1, 2)) materialize(lt, rbind = TRUE) #> torch_tensor #> -0.9852 #> -0.9852 #> -0.0672 #> [ CPUFloatType{3,1} ] unclass(lt[[1]]) #> [[1]] #> [1] 1 #>  #> [[2]] #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,1)] #> * input_map: (x) -> Graph #> * pointer: nop.181509.x.output #> * shape: [(NA,1)] # get index and data descriptor desc = lt[[1]][[2]] id = lt[[1]][[1]]  # retrieve the batch <id> from the datast dataset_output = desc$dataset$.getbatch(id)  # batch is reorganized according to the input map graph_input = dataset_output[desc$input_map] names(graph_input) = names(desc$graph$input$name)  # the reorganized batch is fed into the preprocessing graph graph_output = desc$graph$train(graph_input, single_input = FALSE)  # the output pointed to by the pointer is returned tensor = graph_output[[paste0(desc$pointer, collapse = \".\")]] tensor #> torch_tensor #> -0.9852 #> [ CPUFloatType{1,1} ] taskin = as_task_regr(data.table(x = lt, y = 1:3), target = \"y\")  taskout = po_poly$train(list(taskin))[[1L]]  lt_out = taskout$data(cols = \"x\")$x  descout = lt_out[[1]][[2]]  descout #> <DataDescriptor: 2 ops> #> * dataset_shapes: [x: (NA,1)] #> * input_map: (x) -> Graph #> * pointer: poly.x.output #> * shape: [(NA,7)]  descout$graph #> Graph with 2 PipeOps: #>            ID         State sccssors    prdcssors #>        <char>        <char>   <char>       <char> #>  nop.181509.x        <list>   poly.x              #>        poly.x <<UNTRAINED>>          nop.181509.x materialize(lt_out[1:2], rbind = TRUE) #> torch_tensor #>  1.0000 -0.9852  0.9706 -0.9563  0.9421 -0.9009 -0.8487 #>  1.0000 -0.9852  0.9706 -0.9563  0.9421 -0.9009 -0.8487 #> [ CPUFloatType{2,7} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/authors.html","id":null,"dir":"","previous_headings":"","what":"Authors","title":"Authors and Citation","text":"Sebastian Fischer. Maintainer, author. Bernd Bischl. Contributor. Lukas Burk. Contributor. Martin Binder. Author. Florian Pfisterer. Contributor. Carson Zhang. Contributor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/authors.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Authors and Citation","text":"Fischer S, Binder M (2024). mlr3torch: Deep Learning 'mlr3'. R package version 0.1.2-9000, https://github.com/mlr-org/mlr3torch/, https://mlr3torch.mlr-org.com/.","code":"@Manual{,   title = {mlr3torch: Deep Learning with 'mlr3'},   author = {Sebastian Fischer and Martin Binder},   year = {2024},   note = {R package version 0.1.2-9000, https://github.com/mlr-org/mlr3torch/},   url = {https://mlr3torch.mlr-org.com/}, }"},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"mlr3torch-","dir":"","previous_headings":"","what":"Deep Learning with mlr3","title":"Deep Learning with mlr3","text":"Package website: release | dev Deep Learning torch mlr3.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"installation","dir":"","previous_headings":"","what":"Installation","title":"Deep Learning with mlr3","text":"Afterwards, also need run command : information installing torch can found .","code":"# Install from CRAN install.packages(\"mlr3torch\") # Install the development version from GitHub: pak::pak(\"mlr-org/mlr3torch\") torch::install_torch()"},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"what-is-mlr3torch","dir":"","previous_headings":"","what":"What is mlr3torch?","title":"Deep Learning with mlr3","text":"mlr3torch deep learning framework mlr3 ecosystem built top torch. allows easily build, train evaluate deep learning models lines codes, without needing worry low-level details. --shelf learners readily available, custom architectures can defined connecting PipeOpTorch operators mlr3pipelines::Graph. Using predefined learners simple multi layer perceptron (MLP) works just like mlr3 Learner. , train learner sonar example task: Next, construct architecture using PipeOpTorch objects. first pipeop – PipeOpTorchIngress – defines entrypoint network. subsequent pipeops define neural network layers. turn learner, configure loss, optimizer, callbacks well training arguments. work generic tensors, lazy_tensor type can used. wraps torch::dataset, allows preprocess data (lazily) using PipeOp objects. , flatten MNIST task, can train multi-layer perceptron . Note transform data -memory, applied data actually loaded. actually access tensors, can call materialize(). show slice resulting tensor readability: , define complex architecture one single input lazy_tensor. , first define single residual block: Next, create neural network takes input lazy_tensor (po(\"torch_ingress_ltnsr\")). first applies linear layer repeats layer using special PipeOpTorchBlock, followed network’s head. , configure loss, optimizer training parameters. Note po(\"nn_linear_0\") equivalent po(\"nn_linear\", id = \"nn_linear_0\") need avoid ID clashes linear layer po(\"nn_block\"). Next, prepend preprocessing step flattens images can directly apply learner unflattened MNIST task. order keep track performance training, use 20% data evaluate using classification accuracy. left train learner:","code":"library(mlr3torch) learner_mlp = lrn(\"classif.mlp\",   # defining network parameters   activation     = nn_relu,   neurons        = c(20, 20),   # training parameters   batch_size     = 16,   epochs         = 50,   device         = \"cpu\",   # Proportion of data to use for validation   validate = 0.3,   # Defining the optimizer, loss, and callbacks   optimizer      = t_opt(\"adam\", lr = 0.1),   loss           = t_loss(\"cross_entropy\"),   callbacks      = t_clbk(\"history\"), # this saves the history in the learner   # Measures to track   measures_valid = msrs(c(\"classif.logloss\", \"classif.ce\")),   measures_train = msrs(c(\"classif.acc\")),   # predict type (required by logloss)   predict_type = \"prob\" ) learner_mlp$train(tsk(\"sonar\")) architecture = po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 20) %>>%   po(\"nn_relu\") %>>%   po(\"nn_head\") graph_mlp = architecture %>>%   po(\"torch_loss\", loss = t_loss(\"cross_entropy\")) %>>%   po(\"torch_optimizer\", optimizer = t_opt(\"adam\", lr = 0.1)) %>>%   po(\"torch_callbacks\", callbacks = t_clbk(\"history\")) %>>%   po(\"torch_model_classif\",     batch_size = 16, epochs = 50, device = \"cpu\")  graph_lrn = as_learner(graph_mlp) # load the predefined mnist task mnist = tsk(\"mnist\") mnist$head(3L) #>     label           image #>    <fctr>   <lazy_tensor> #> 1:      5 <tnsr[1x28x28]> #> 2:      0 <tnsr[1x28x28]> #> 3:      4 <tnsr[1x28x28]>  # Flatten the images flattener = po(\"trafo_reshape\", shape = c(-1, 28 * 28)) mnist_flat = flattener$train(list(mnist))[[1L]]  mnist_flat$head(3L) #>     label         image #>    <fctr> <lazy_tensor> #> 1:      5   <tnsr[784]> #> 2:      0   <tnsr[784]> #> 3:      4   <tnsr[784]> materialize(   mnist_flat$data(1:2, cols = \"image\")[[1L]],   rbind = TRUE )[1:2, 1:4] #> torch_tensor #>  0  0  0  0 #>  0  0  0  0 #> [ CPUFloatType{2,4} ] layer = list(   po(\"nop\"),   po(\"nn_linear\", out_features = 50L) %>>%     po(\"nn_dropout\") %>>% po(\"nn_relu\") ) %>>% po(\"nn_merge_sum\") deep_network = po(\"torch_ingress_ltnsr\") %>>%   po(\"nn_linear_0\", out_features = 50L) %>>%   po(\"nn_block\", layer, n_blocks = 5L) %>>%   po(\"nn_head\") %>>%   po(\"torch_loss\", loss = t_loss(\"cross_entropy\")) %>>%   po(\"torch_optimizer\", optimizer = t_opt(\"adam\")) %>>%   po(\"torch_model_classif\",     epochs = 100L, batch_size = 32   ) deep_learner = as_learner(   flattener %>>% deep_network ) deep_learner$id = \"deep_network\" set_validate(deep_learner, 0.2) deep_learner$param_set$set_values(   torch_model_classif.measures_valid = msr(\"classif.acc\") ) deep_learner$train(mnist)"},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"feature-overview","dir":"","previous_headings":"","what":"Feature Overview","title":"Deep Learning with mlr3","text":"--shelf architectures readily available mlr3::Learners. Currently, supervised regression classification supported. Custom learners can defined using Graph language mlr3pipelines. package supports tabular data, well generic tensors via lazy_tensor type. Multi-modal data can handled conveniently, lazy_tensor objects can stored alongside tabular data. possible customize training process via (predefined custom) callbacks. package fully integrated mlr3 ecosystem. Neural network architectures, well hyperparameters can easily tuned via mlr3tuning friends.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"documentation","dir":"","previous_headings":"","what":"Documentation","title":"Deep Learning with mlr3","text":"Start reading one vignettes package website!","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"contributing","dir":"","previous_headings":"","what":"Contributing:","title":"Deep Learning with mlr3","text":"run tests one needs set environment variable TEST_TORCH = 1, e.g. adding .Renviron.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"acknowledgements","dir":"","previous_headings":"","what":"Acknowledgements","title":"Deep Learning with mlr3","text":"Without great R package torch none possible. names callback stages taken luz, another high-level deep learning framework R torch. Building neural networks using PipeOpTorch operators inspired keras.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/index.html","id":"bugs-questions-feedback","dir":"","previous_headings":"","what":"Bugs, Questions, Feedback","title":"Deep Learning with mlr3","text":"mlr3torch free open source software project encourages participation feedback. issues, questions, suggestions feedback, please hesitate open “issue” GitHub page! case problems / bugs, often helpful provide “minimum working example” showcases behaviour (don’t worry bug obvious). Please understand resources project limited: response may sometimes delayed days, feature suggestions may rejected deemed tangential vision behind project.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":null,"dir":"Reference","previous_headings":"","what":"Data Descriptor — DataDescriptor","title":"Data Descriptor — DataDescriptor","text":"data descriptor rather internal data structure used lazy_tensor data type. essence annotated torch::dataset preprocessing graph (consisting mosty PipeOpModule operators). additional meta data (e.g. pointer, shapes) allows preprocess lazy_tensors mlr3pipelines::Graph just like (non-lazy) data types. preprocessing applied materialize() called lazy_tensor. create data descriptor, can also use as_data_descriptor() function.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Data Descriptor — DataDescriptor","text":"natural define S3 class, opted R6 class avoid usual trouble serializing S3 objects. row contained DataDescriptor S3 class, copy object serializing.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Data Descriptor — DataDescriptor","text":"dataset (torch::dataset) dataset. graph (Graph) preprocessing graph. dataset_shapes (named list() (integer() NULL)) shapes output. input_map (character()) input map dataset preprocessing graph. pointer (character(2)) output pointer. pointer_shape (integer() | NULL) shape output indicated pointer. dataset_hash (character(1)) Hash wrapped dataset. hash (character(1)) Hash data descriptor. graph_input (character()) input channels preprocessing graph (cached save time). pointer_shape_predict (integer() NULL) Internal use .","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Data Descriptor — DataDescriptor","text":"DataDescriptor$new() DataDescriptor$print() DataDescriptor$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Data Descriptor — DataDescriptor","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Data Descriptor — DataDescriptor","text":"","code":"DataDescriptor$new(   dataset,   dataset_shapes = NULL,   graph = NULL,   input_map = NULL,   pointer = NULL,   pointer_shape = NULL,   pointer_shape_predict = NULL,   clone_graph = TRUE )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Data Descriptor — DataDescriptor","text":"dataset (torch::dataset) torch dataset. return named list() torch_tensor objects. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. graph (Graph) preprocessing graph. left NULL, preprocessing applied data input_map, pointer, pointer_shape, pointer_shape_predict inferred case dataset returns one element. input_map (character()) Character vector must length input graph. Specifies data dataset fed preprocessing graph. pointer (character(2) | NULL) Points output channel within graph: Element 1 PipeOp's id element 2 PipeOp's output channel. pointer_shape (integer() | NULL) Shape output indicated pointer. pointer_shape_predict (integer() NULL) Internal use . Used Graph anticipate possible mismatches train predict shapes. clone_graph (logical(1)) Whether clone preprocessing graph.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Data Descriptor — DataDescriptor","text":"Prints object","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Data Descriptor — DataDescriptor","text":"","code":"DataDescriptor$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Data Descriptor — DataDescriptor","text":"... () Unused","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Data Descriptor — DataDescriptor","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Data Descriptor — DataDescriptor","text":"","code":"DataDescriptor$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Data Descriptor — DataDescriptor","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/DataDescriptor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Data Descriptor — DataDescriptor","text":"","code":"# Create a dataset ds = dataset(   initialize = function() self$x = torch_randn(10, 3, 3),   .getitem = function(i) list(x = self$x[i, ]),   .length = function() nrow(self$x) )() dd = DataDescriptor$new(ds, list(x = c(NA, 3, 3))) dd #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,3,3)] #> * input_map: (x) -> Graph #> * pointer: nop.ae2791.x.output #> * shape: [(NA,3,3)] # is the same as using the converter: as_data_descriptor(ds, list(x = c(NA, 3, 3))) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,3,3)] #> * input_map: (x) -> Graph #> * pointer: nop.ae2791.x.output #> * shape: [(NA,3,3)]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/ModelDescriptor.html","id":null,"dir":"Reference","previous_headings":"","what":"Represent a Model with Meta-Info — ModelDescriptor","title":"Represent a Model with Meta-Info — ModelDescriptor","text":"Represents model; possibly complete model, possibly one process built . model takes input tensors shapes shapes_in pipes graph. Input shapes get mapped input channels graph. Output shapes named output channels graph; also possible represent -ops tensors, case names input output identical. ModelDescriptor objects typically represent partial models built , case pointer slot indicates specific point graph produces tensor shape pointer_shape, graph extended. allowed graph structure modified -reference different parts code. However, modifications may never add edges elements Graph destination. particular, element graph$input may removed reference, e.g. adding edge Graph input channel PipeOp previously without parent destination. cases better create specific ModelDescriptor training Graph consisting (mostly) operators PipeOpTorchIngress, PipeOpTorch, PipeOpTorchLoss, PipeOpTorchOptimizer, PipeOpTorchCallbacks. ModelDescriptor can converted nn_graph via model_descriptor_to_module.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/ModelDescriptor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Represent a Model with Meta-Info — ModelDescriptor","text":"","code":"ModelDescriptor(   graph,   ingress,   task,   optimizer = NULL,   loss = NULL,   callbacks = NULL,   pointer = NULL,   pointer_shape = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/ModelDescriptor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Represent a Model with Meta-Info — ModelDescriptor","text":"graph (Graph)Graph PipeOpModule PipeOpNOP operators. ingress (uniquely named list TorchIngressToken) List inputs go graph. Names must subset graph$input$name. task (Task) (Training)-Task model built. May necessary aspects loss use etc. optimizer (TorchOptimizer | NULL) Additional info: optimizer use. loss (TorchLoss | NULL) Additional info: loss use. callbacks (list CallbackSet NULL) Additional info: callbacks use. pointer (character(2) | NULL) Indicating element model . Points output channel within graph: Element 1 PipeOp's id element 2 PipeOp's output channel. pointer_shape (integer | NULL) Shape output indicated pointer.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/ModelDescriptor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Represent a Model with Meta-Info — ModelDescriptor","text":"(ModelDescriptor)","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentCenterCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentCenterCrop — mlr_pipeops_augment_center_crop","title":"PipeOpPreprocTorchAugmentCenterCrop — mlr_pipeops_augment_center_crop","text":"Calls torchvision::transform_center_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentCenterCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentCenterCrop — mlr_pipeops_augment_center_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentColorJitter.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentColorJitter — mlr_pipeops_augment_color_jitter","title":"PipeOpPreprocTorchAugmentColorJitter — mlr_pipeops_augment_color_jitter","text":"Calls torchvision::transform_color_jitter, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentColorJitter.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentColorJitter — mlr_pipeops_augment_color_jitter","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentCrop — mlr_pipeops_augment_crop","title":"PipeOpPreprocTorchAugmentCrop — mlr_pipeops_augment_crop","text":"Calls torchvision::transform_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentCrop — mlr_pipeops_augment_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentHflip.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentHflip — mlr_pipeops_augment_hflip","title":"PipeOpPreprocTorchAugmentHflip — mlr_pipeops_augment_hflip","text":"Calls torchvision::transform_hflip, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentHflip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentHflip — mlr_pipeops_augment_hflip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomAffine.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomAffine — mlr_pipeops_augment_random_affine","title":"PipeOpPreprocTorchAugmentRandomAffine — mlr_pipeops_augment_random_affine","text":"Calls torchvision::transform_random_affine, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomAffine.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomAffine — mlr_pipeops_augment_random_affine","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomChoice.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomChoice — mlr_pipeops_augment_random_choice","title":"PipeOpPreprocTorchAugmentRandomChoice — mlr_pipeops_augment_random_choice","text":"Calls torchvision::transform_random_choice, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomChoice.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomChoice — mlr_pipeops_augment_random_choice","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomCrop — mlr_pipeops_augment_random_crop","title":"PipeOpPreprocTorchAugmentRandomCrop — mlr_pipeops_augment_random_crop","text":"Calls torchvision::transform_random_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomCrop — mlr_pipeops_augment_random_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomHorizontalFlip.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomHorizontalFlip — mlr_pipeops_augment_random_horizontal_flip","title":"PipeOpPreprocTorchAugmentRandomHorizontalFlip — mlr_pipeops_augment_random_horizontal_flip","text":"Calls torchvision::transform_random_horizontal_flip, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomHorizontalFlip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomHorizontalFlip — mlr_pipeops_augment_random_horizontal_flip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomOrder.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomOrder — mlr_pipeops_augment_random_order","title":"PipeOpPreprocTorchAugmentRandomOrder — mlr_pipeops_augment_random_order","text":"Calls torchvision::transform_random_order, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomOrder.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomOrder — mlr_pipeops_augment_random_order","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomResizedCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomResizedCrop — mlr_pipeops_augment_random_resized_crop","title":"PipeOpPreprocTorchAugmentRandomResizedCrop — mlr_pipeops_augment_random_resized_crop","text":"Calls torchvision::transform_random_resized_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomResizedCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomResizedCrop — mlr_pipeops_augment_random_resized_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomVerticalFlip.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRandomVerticalFlip — mlr_pipeops_augment_random_vertical_flip","title":"PipeOpPreprocTorchAugmentRandomVerticalFlip — mlr_pipeops_augment_random_vertical_flip","text":"Calls torchvision::transform_random_vertical_flip, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRandomVerticalFlip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRandomVerticalFlip — mlr_pipeops_augment_random_vertical_flip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentResizedCrop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentResizedCrop — mlr_pipeops_augment_resized_crop","title":"PipeOpPreprocTorchAugmentResizedCrop — mlr_pipeops_augment_resized_crop","text":"Calls torchvision::transform_resized_crop, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentResizedCrop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentResizedCrop — mlr_pipeops_augment_resized_crop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRotate.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentRotate — mlr_pipeops_augment_rotate","title":"PipeOpPreprocTorchAugmentRotate — mlr_pipeops_augment_rotate","text":"Calls torchvision::transform_rotate, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentRotate.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentRotate — mlr_pipeops_augment_rotate","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentVflip.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchAugmentVflip — mlr_pipeops_augment_vflip","title":"PipeOpPreprocTorchAugmentVflip — mlr_pipeops_augment_vflip","text":"Calls torchvision::transform_vflip, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchAugmentVflip.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchAugmentVflip — mlr_pipeops_augment_vflip","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustBrightness.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoAdjustBrightness — mlr_pipeops_trafo_adjust_brightness","title":"PipeOpPreprocTorchTrafoAdjustBrightness — mlr_pipeops_trafo_adjust_brightness","text":"Calls torchvision::transform_adjust_brightness, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustBrightness.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoAdjustBrightness — mlr_pipeops_trafo_adjust_brightness","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustGamma.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoAdjustGamma — mlr_pipeops_trafo_adjust_gamma","title":"PipeOpPreprocTorchTrafoAdjustGamma — mlr_pipeops_trafo_adjust_gamma","text":"Calls torchvision::transform_adjust_gamma, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustGamma.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoAdjustGamma — mlr_pipeops_trafo_adjust_gamma","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustHue.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoAdjustHue — mlr_pipeops_trafo_adjust_hue","title":"PipeOpPreprocTorchTrafoAdjustHue — mlr_pipeops_trafo_adjust_hue","text":"Calls torchvision::transform_adjust_hue, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustHue.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoAdjustHue — mlr_pipeops_trafo_adjust_hue","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustSaturation.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoAdjustSaturation — mlr_pipeops_trafo_adjust_saturation","title":"PipeOpPreprocTorchTrafoAdjustSaturation — mlr_pipeops_trafo_adjust_saturation","text":"Calls torchvision::transform_adjust_saturation, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoAdjustSaturation.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoAdjustSaturation — mlr_pipeops_trafo_adjust_saturation","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoGrayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoGrayscale — mlr_pipeops_trafo_grayscale","title":"PipeOpPreprocTorchTrafoGrayscale — mlr_pipeops_trafo_grayscale","text":"Calls torchvision::transform_grayscale, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoGrayscale.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoGrayscale — mlr_pipeops_trafo_grayscale","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoNop.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoNop — mlr_pipeops_trafo_nop","title":"PipeOpPreprocTorchTrafoNop — mlr_pipeops_trafo_nop","text":"nothing.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoNop.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoNop — mlr_pipeops_trafo_nop","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoNormalize.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoNormalize — mlr_pipeops_trafo_normalize","title":"PipeOpPreprocTorchTrafoNormalize — mlr_pipeops_trafo_normalize","text":"Calls torchvision::transform_normalize, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoNormalize.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoNormalize — mlr_pipeops_trafo_normalize","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoPad.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoPad — mlr_pipeops_trafo_pad","title":"PipeOpPreprocTorchTrafoPad — mlr_pipeops_trafo_pad","text":"Calls torchvision::transform_pad, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoPad.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoPad — mlr_pipeops_trafo_pad","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoReshape.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoReshape — mlr_pipeops_trafo_reshape","title":"PipeOpPreprocTorchTrafoReshape — mlr_pipeops_trafo_reshape","text":"Reshapes tensor according parameter shape, calling torch_reshape(). preprocessing function applied batch-wise.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoReshape.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoReshape — mlr_pipeops_trafo_reshape","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoReshape.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PipeOpPreprocTorchTrafoReshape — mlr_pipeops_trafo_reshape","text":"shape :: integer() desired output shape. first dimension batch dimension usually -1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoResize.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoResize — mlr_pipeops_trafo_resize","title":"PipeOpPreprocTorchTrafoResize — mlr_pipeops_trafo_resize","text":"Calls torchvision::transform_resize, see information parameters. preprocessing applied whole batch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoResize.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoResize — mlr_pipeops_trafo_resize","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoRgbToGrayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOpPreprocTorchTrafoRgbToGrayscale — mlr_pipeops_trafo_rgb_to_grayscale","title":"PipeOpPreprocTorchTrafoRgbToGrayscale — mlr_pipeops_trafo_rgb_to_grayscale","text":"Calls torchvision::transform_rgb_to_grayscale, see information parameters. preprocessing applied row wise (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/PipeOpPreprocTorchTrafoRgbToGrayscale.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"PipeOpPreprocTorchTrafoRgbToGrayscale — mlr_pipeops_trafo_rgb_to_grayscale","text":"R6Class inheriting PipeOpTaskPreprocTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Callback — TorchCallback","title":"Torch Callback — TorchCallback","text":"wraps CallbackSet annotates metadata, importantly ParamSet. callback created given parameter values calling $generate() method. class usually used configure callback torch learner, e.g. constructing learner ModelDescriptor. list available callbacks, see mlr3torch_callbacks. conveniently retrieve TorchCallback, use t_clbk().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Callback — TorchCallback","text":"Defined constructor argument param_set. parameter set provided construction, parameter set constructed creating parameter argument wrapped loss function, parametes type ParamUty.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Torch Callback — TorchCallback","text":"mlr3torch::TorchDescriptor -> TorchCallback","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Callback — TorchCallback","text":"mlr3torch::TorchDescriptor$generate() mlr3torch::TorchDescriptor$help() mlr3torch::TorchDescriptor$print()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Callback — TorchCallback","text":"TorchCallback$new() TorchCallback$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Callback — TorchCallback","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Callback — TorchCallback","text":"","code":"TorchCallback$new(   callback_generator,   param_set = NULL,   id = NULL,   label = NULL,   packages = NULL,   man = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Callback — TorchCallback","text":"callback_generator (R6ClassGenerator) class generator callback wrapped. param_set (ParamSet NULL) parameter set. NULL (default) inferred callback_generator. id (character(1)) id new object. label (character(1)) Label new instance. packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Callback — TorchCallback","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Callback — TorchCallback","text":"","code":"TorchCallback$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Callback — TorchCallback","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchCallback.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Callback — TorchCallback","text":"","code":"# Create a new torch callback from an existing callback set torch_callback = TorchCallback$new(CallbackSetCheckpoint) # The parameters are inferred torch_callback$param_set #> <ParamSet(3)> #>           id    class lower upper nlevels        default  value #>       <char>   <char> <num> <num>   <num>         <list> <list> #> 1:      path ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:      freq ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 3: freq_type ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]  # Retrieve a torch callback from the dictionary torch_callback = t_clbk(\"checkpoint\",   path = tempfile(), freq = 1 ) torch_callback #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: path=/tmp/RtmpzHy548/file2541149855cf, freq=1 #> * Packages: mlr3torch,torch torch_callback$label #> [1] \"Checkpoint\" torch_callback$id #> [1] \"checkpoint\"  # open the help page of the wrapped callback set # torch_callback$help()  # Create the callback set callback = torch_callback$generate() callback #> <CallbackSetCheckpoint> #> * Stages: on_batch_end, on_epoch_end, on_exit # is the same as CallbackSetCheckpoint$new(   path = tempfile(), freq = 1 ) #> <CallbackSetCheckpoint> #> * Stages: on_batch_end, on_epoch_end, on_exit  # Use in a learner learner = lrn(\"regr.mlp\", callbacks = t_clbk(\"checkpoint\")) # the parameters of the callback are added to the learner's parameter set learner$param_set #> <ParamSetCollection(37)> #>                          id    class lower upper nlevels        default #>                      <char>   <char> <num> <num>   <num>         <list> #>  1:                  epochs ParamInt 0e+00   Inf     Inf <NoDefault[0]> #>  2:                  device ParamFct    NA    NA      12 <NoDefault[0]> #>  3:             num_threads ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  4:     num_interop_threads ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  5:                    seed ParamInt  -Inf   Inf     Inf <NoDefault[0]> #>  6:               eval_freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  7:          measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  8:          measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:                patience ParamInt 0e+00   Inf     Inf <NoDefault[0]> #> 10:               min_delta ParamDbl 0e+00   Inf     Inf <NoDefault[0]> #> 11:              batch_size ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 12:                 shuffle ParamLgl    NA    NA       2          FALSE #> 13:                 sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 14:           batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:             num_workers ParamInt 0e+00   Inf     Inf              0 #> 16:              collate_fn ParamUty    NA    NA     Inf         [NULL] #> 17:              pin_memory ParamLgl    NA    NA       2          FALSE #> 18:               drop_last ParamLgl    NA    NA       2          FALSE #> 19:                 timeout ParamDbl  -Inf   Inf     Inf             -1 #> 20:          worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 21:          worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:         worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:                 neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:                       p ParamDbl 0e+00 1e+00     Inf <NoDefault[0]> #> 25:                n_layers ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 26:              activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:         activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:                   shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:                  opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 30:               opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 31:                 opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 32:        opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 33:             opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 34:          loss.reduction ParamFct    NA    NA       2           mean #> 35:      cb.checkpoint.path ParamUty    NA    NA     Inf <NoDefault[0]> #> 36:      cb.checkpoint.freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 37: cb.checkpoint.freq_type ParamFct    NA    NA       2          epoch #>                          id    class lower upper nlevels        default #>            value #>           <list> #>  1:       [NULL] #>  2:         auto #>  3:            1 #>  4:            1 #>  5:       random #>  6:            1 #>  7:    <list[0]> #>  8:    <list[0]> #>  9:            0 #> 10:            0 #> 11:       [NULL] #> 12:       [NULL] #> 13:       [NULL] #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:              #> 24:          0.5 #> 25:       [NULL] #> 26: <nn_relu[1]> #> 27:    <list[0]> #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [NULL] #> 37:       [NULL] #>            value"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Descriptors — TorchDescriptor","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Abstract Base Class TorchLoss, TorchOptimizer, TorchCallback inherit. class wraps generator (R6Class Generator torch version generator) annotates metadata ParamSet, label, ID, packages, manual page. parameters construction arguments wrapped generator parameter $values passed generator calling public method $generate().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Defined constructor argument param_set. parameters tagged \"train\", done automatically initialize.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"label (character(1)) Label object. Can used tables, plot text output instead ID. param_set (ParamSet) Set hyperparameters. packages (character(1)) Set required packages. packages loaded, attached. id (character(1)) Identifier object. Used tables, plot text output. generator wrapped generator described. man (character(1)) String format [pkg]::[topic] pointing manual page object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"phash (character(1)) Hash (unique identifier) partial object, excluding components varied systematically (e.g. parameter values).","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"TorchDescriptor$new() TorchDescriptor$print() TorchDescriptor$generate() TorchDescriptor$help() TorchDescriptor$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$new(   generator,   id = NULL,   param_set = NULL,   packages = NULL,   label = NULL,   man = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"generator wrapped generator described. id (character(1)) id new object. param_set (ParamSet) parameter set. packages (character()) R packages object depends . label (character(1)) Label new instance. man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Prints object","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"... ","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-generate-","dir":"Reference","previous_headings":"","what":"Method generate()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Calls generator given parameter values.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$generate()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-help-","dir":"Reference","previous_headings":"","what":"Method help()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"Displays help file wrapped object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$help()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"","code":"TorchDescriptor$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchDescriptor.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Descriptors — TorchDescriptor","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Ingress Token — TorchIngressToken","title":"Torch Ingress Token — TorchIngressToken","text":"function creates S3 class class \"TorchIngressToken\", internal data structure. contains (meta-)information batch generated Task fed entry point neural network. stored ingress field ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Ingress Token — TorchIngressToken","text":"","code":"TorchIngressToken(features, batchgetter, shape)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Ingress Token — TorchIngressToken","text":"features (character) Features batchgetter operate. batchgetter (function) Function two arguments: data device. function given output Task$data(rows = batch_indices, cols = features) produce tensor shape shape_out. shape (integer) Shape batchgetter produce. Batch-dimension included NA.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Torch Ingress Token — TorchIngressToken","text":"TorchIngressToken object.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchIngressToken.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Ingress Token — TorchIngressToken","text":"","code":"# Define a task for which we want to define an ingress token task = tsk(\"iris\")  # We create an ingress token for two feature Sepal.Length and Petal.Length: # We have to specify the features, the batchgetter and the shape features = c(\"Sepal.Length\", \"Petal.Length\") # As a batchgetter we use batchgetter_num  batch_dt = task$data(rows = 1:10, cols =features) batch_dt #>     Sepal.Length Petal.Length #>            <num>        <num> #>  1:          5.1          1.4 #>  2:          4.9          1.4 #>  3:          4.7          1.3 #>  4:          4.6          1.5 #>  5:          5.0          1.4 #>  6:          5.4          1.7 #>  7:          4.6          1.4 #>  8:          5.0          1.5 #>  9:          4.4          1.4 #> 10:          4.9          1.5 batch_tensor = batchgetter_num(batch_dt, \"cpu\") batch_tensor #> torch_tensor #>  5.1000  1.4000 #>  4.9000  1.4000 #>  4.7000  1.3000 #>  4.6000  1.5000 #>  5.0000  1.4000 #>  5.4000  1.7000 #>  4.6000  1.4000 #>  5.0000  1.5000 #>  4.4000  1.4000 #>  4.9000  1.5000 #> [ CPUFloatType{10,2} ]  # The shape is unknown in the first dimension (batch dimension)  ingress_token = TorchIngressToken(   features = features,   batchgetter = batchgetter_num,   shape = c(NA, 2) ) ingress_token #> Ingress: Task[Sepal.Length,Petal.Length] --> Tensor(NA, 2)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Loss — TorchLoss","title":"Torch Loss — TorchLoss","text":"wraps torch::nn_loss annotates metadata, importantly ParamSet. loss function created given parameter values calling $generate() method. class usually used configure loss function torch learner, e.g. construcing learner ModelDescriptor. list available losses, see mlr3torch_losses. Items dictionary can retrieved using t_loss().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Loss — TorchLoss","text":"Defined constructor argument param_set. parameter set provided construction, parameter set constructed creating parameter argument wrapped loss function, parametes type ParamUty.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Torch Loss — TorchLoss","text":"mlr3torch::TorchDescriptor -> TorchLoss","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Torch Loss — TorchLoss","text":"task_types (character()) task types loss supports.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Loss — TorchLoss","text":"mlr3torch::TorchDescriptor$generate() mlr3torch::TorchDescriptor$help()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Loss — TorchLoss","text":"TorchLoss$new() TorchLoss$print() TorchLoss$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Loss — TorchLoss","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Loss — TorchLoss","text":"","code":"TorchLoss$new(   torch_loss,   task_types = NULL,   param_set = NULL,   id = NULL,   label = NULL,   packages = NULL,   man = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Loss — TorchLoss","text":"torch_loss (nn_loss) loss module. task_types (character()) task types supported loss. param_set (ParamSet NULL) parameter set. NULL (default) inferred torch_loss. id (character(1)) id new object. label (character(1)) Label new instance. packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Torch Loss — TorchLoss","text":"Prints object","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Loss — TorchLoss","text":"","code":"TorchLoss$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Loss — TorchLoss","text":"... ","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Loss — TorchLoss","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Loss — TorchLoss","text":"","code":"TorchLoss$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Loss — TorchLoss","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchLoss.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Loss — TorchLoss","text":"","code":"# Create a new torch loss torch_loss = TorchLoss$new(torch_loss = nn_mse_loss, task_types = \"regr\") torch_loss #> <TorchLoss:nn_mse_loss> nn_mse_loss #> * Generator: nn_mse_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr # the parameters are inferred torch_loss$param_set #> <ParamSet(1)> #>           id    class lower upper nlevels        default  value #>       <char>   <char> <num> <num>   <num>         <list> <list> #> 1: reduction ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]  # Retrieve a loss from the dictionary: torch_loss = t_loss(\"mse\", reduction = \"mean\") # is the same as torch_loss #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: reduction=mean #> * Packages: torch,mlr3torch #> * Task Types: regr torch_loss$param_set #> <ParamSet(1)> #>           id    class lower upper nlevels default  value #>       <char>   <char> <num> <num>   <num>  <list> <list> #> 1: reduction ParamFct    NA    NA       2    mean   mean torch_loss$label #> [1] \"Mean Squared Error\" torch_loss$task_types #> [1] \"regr\" torch_loss$id #> [1] \"mse\"  # Create the loss function loss_fn = torch_loss$generate() loss_fn #> An `nn_module` containing 0 parameters. # Is the same as nn_mse_loss(reduction = \"mean\") #> An `nn_module` containing 0 parameters.  # open the help page of the wrapped loss function # torch_loss$help()  # Use in a learner learner = lrn(\"regr.mlp\", loss = t_loss(\"mse\")) # The parameters of the loss are added to the learner's parameter set learner$param_set #> <ParamSetCollection(34)> #>                      id    class lower upper nlevels        default #>                  <char>   <char> <num> <num>   <num>         <list> #>  1:              epochs ParamInt 0e+00   Inf     Inf <NoDefault[0]> #>  2:              device ParamFct    NA    NA      12 <NoDefault[0]> #>  3:         num_threads ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  4: num_interop_threads ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  5:                seed ParamInt  -Inf   Inf     Inf <NoDefault[0]> #>  6:           eval_freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #>  7:      measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  8:      measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:            patience ParamInt 0e+00   Inf     Inf <NoDefault[0]> #> 10:           min_delta ParamDbl 0e+00   Inf     Inf <NoDefault[0]> #> 11:          batch_size ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 12:             shuffle ParamLgl    NA    NA       2          FALSE #> 13:             sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 14:       batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:         num_workers ParamInt 0e+00   Inf     Inf              0 #> 16:          collate_fn ParamUty    NA    NA     Inf         [NULL] #> 17:          pin_memory ParamLgl    NA    NA       2          FALSE #> 18:           drop_last ParamLgl    NA    NA       2          FALSE #> 19:             timeout ParamDbl  -Inf   Inf     Inf             -1 #> 20:      worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 21:      worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:     worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:             neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:                   p ParamDbl 0e+00 1e+00     Inf <NoDefault[0]> #> 25:            n_layers ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 26:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:              opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 30:           opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 31:             opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 32:    opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 33:         opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 34:      loss.reduction ParamFct    NA    NA       2           mean #>                      id    class lower upper nlevels        default #>            value #>           <list> #>  1:       [NULL] #>  2:         auto #>  3:            1 #>  4:            1 #>  5:       random #>  6:            1 #>  7:    <list[0]> #>  8:    <list[0]> #>  9:            0 #> 10:            0 #> 11:       [NULL] #> 12:       [NULL] #> 13:       [NULL] #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:              #> 24:          0.5 #> 25:       [NULL] #> 26: <nn_relu[1]> #> 27:    <list[0]> #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #>            value"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Optimizer — TorchOptimizer","title":"Torch Optimizer — TorchOptimizer","text":"wraps torch::torch_optimizer_generatora annotates metadata, importantly ParamSet. optimizer created given parameter values calling $generate() method. class usually used configure optimizer torch learner, e.g. construcing learner ModelDescriptor. list available optimizers, see mlr3torch_optimizers. Items dictionary can retrieved using t_opt().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Optimizer — TorchOptimizer","text":"Defined constructor argument param_set. parameter set provided construction, parameter set constructed creating parameter argument wrapped loss function, parametes type ParamUty.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Torch Optimizer — TorchOptimizer","text":"mlr3torch::TorchDescriptor -> TorchOptimizer","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Optimizer — TorchOptimizer","text":"mlr3torch::TorchDescriptor$help() mlr3torch::TorchDescriptor$print()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Optimizer — TorchOptimizer","text":"TorchOptimizer$new() TorchOptimizer$generate() TorchOptimizer$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Optimizer — TorchOptimizer","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Optimizer — TorchOptimizer","text":"","code":"TorchOptimizer$new(   torch_optimizer,   param_set = NULL,   id = NULL,   label = NULL,   packages = NULL,   man = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Optimizer — TorchOptimizer","text":"torch_optimizer (torch_optimizer_generator) torch optimizer. param_set (ParamSet NULL) parameter set. NULL (default) inferred torch_optimizer. id (character(1)) id new object. label (character(1)) Label new instance. packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"method-generate-","dir":"Reference","previous_headings":"","what":"Method generate()","title":"Torch Optimizer — TorchOptimizer","text":"Instantiates optimizer.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Optimizer — TorchOptimizer","text":"","code":"TorchOptimizer$generate(params)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Optimizer — TorchOptimizer","text":"params (named list() torch_tensors) parameters network.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Torch Optimizer — TorchOptimizer","text":"torch_optimizer","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Optimizer — TorchOptimizer","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Optimizer — TorchOptimizer","text":"","code":"TorchOptimizer$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Optimizer — TorchOptimizer","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/TorchOptimizer.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Optimizer — TorchOptimizer","text":"","code":"# Create a new torch loss torch_opt = TorchOptimizer$new(optim_adam, label = \"adam\") torch_opt #> <TorchOptimizer:optim_adam> adam #> * Generator: optim_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # If the param set is not specified, parameters are inferred but are of class ParamUty torch_opt$param_set #> <ParamSet(5)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1:           lr ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:        betas ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 3:          eps ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 4: weight_decay ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 5:      amsgrad ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]  # open the help page of the wrapped optimizer # torch_opt$help()  # Retrieve an optimizer from the dictionary torch_opt = t_opt(\"sgd\", lr = 0.1) torch_opt #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_sgd #> * Parameters: lr=0.1 #> * Packages: torch,mlr3torch torch_opt$param_set #> <ParamSet(5)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1:           lr ParamDbl     0   Inf     Inf <NoDefault[0]>    0.1 #> 2:     momentum ParamDbl     0     1     Inf              0 [NULL] #> 3:    dampening ParamDbl     0     1     Inf              0 [NULL] #> 4: weight_decay ParamDbl     0     1     Inf              0 [NULL] #> 5:     nesterov ParamLgl    NA    NA       2          FALSE [NULL] torch_opt$label #> [1] \"Stochastic Gradient Descent\" torch_opt$id #> [1] \"sgd\"  # Create the optimizer for a network net = nn_linear(10, 1) opt = torch_opt$generate(net$parameters)  # is the same as optim_sgd(net$parameters, lr = 0.1) #> <optim_sgd> #>   Inherits from: <torch_optimizer> #>   Public: #>     add_param_group: function (param_group)  #>     clone: function (deep = FALSE)  #>     defaults: list #>     initialize: function (params, lr = optim_required(), momentum = 0, dampening = 0,  #>     load_state_dict: function (state_dict, ..., .refer_to_state_dict = FALSE)  #>     param_groups: list #>     state: State, R6 #>     state_dict: function ()  #>     step: function (closure = NULL)  #>     zero_grad: function ()  #>   Private: #>     step_helper: function (closure, loop_fun)   # Use in a learner learner = lrn(\"regr.mlp\", optimizer = t_opt(\"sgd\")) # The parameters of the optimizer are added to the learner's parameter set learner$param_set #> <ParamSetCollection(34)> #>                      id    class lower upper nlevels        default #>                  <char>   <char> <num> <num>   <num>         <list> #>  1:              epochs ParamInt     0   Inf     Inf <NoDefault[0]> #>  2:              device ParamFct    NA    NA      12 <NoDefault[0]> #>  3:         num_threads ParamInt     1   Inf     Inf <NoDefault[0]> #>  4: num_interop_threads ParamInt     1   Inf     Inf <NoDefault[0]> #>  5:                seed ParamInt  -Inf   Inf     Inf <NoDefault[0]> #>  6:           eval_freq ParamInt     1   Inf     Inf <NoDefault[0]> #>  7:      measures_train ParamUty    NA    NA     Inf <NoDefault[0]> #>  8:      measures_valid ParamUty    NA    NA     Inf <NoDefault[0]> #>  9:            patience ParamInt     0   Inf     Inf <NoDefault[0]> #> 10:           min_delta ParamDbl     0   Inf     Inf <NoDefault[0]> #> 11:          batch_size ParamInt     1   Inf     Inf <NoDefault[0]> #> 12:             shuffle ParamLgl    NA    NA       2          FALSE #> 13:             sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 14:       batch_sampler ParamUty    NA    NA     Inf <NoDefault[0]> #> 15:         num_workers ParamInt     0   Inf     Inf              0 #> 16:          collate_fn ParamUty    NA    NA     Inf         [NULL] #> 17:          pin_memory ParamLgl    NA    NA       2          FALSE #> 18:           drop_last ParamLgl    NA    NA       2          FALSE #> 19:             timeout ParamDbl  -Inf   Inf     Inf             -1 #> 20:      worker_init_fn ParamUty    NA    NA     Inf <NoDefault[0]> #> 21:      worker_globals ParamUty    NA    NA     Inf <NoDefault[0]> #> 22:     worker_packages ParamUty    NA    NA     Inf <NoDefault[0]> #> 23:             neurons ParamUty    NA    NA     Inf <NoDefault[0]> #> 24:                   p ParamDbl     0     1     Inf <NoDefault[0]> #> 25:            n_layers ParamInt     1   Inf     Inf <NoDefault[0]> #> 26:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 29:              opt.lr ParamDbl     0   Inf     Inf <NoDefault[0]> #> 30:        opt.momentum ParamDbl     0     1     Inf              0 #> 31:       opt.dampening ParamDbl     0     1     Inf              0 #> 32:    opt.weight_decay ParamDbl     0     1     Inf              0 #> 33:        opt.nesterov ParamLgl    NA    NA       2          FALSE #> 34:      loss.reduction ParamFct    NA    NA       2           mean #>                      id    class lower upper nlevels        default #>            value #>           <list> #>  1:       [NULL] #>  2:         auto #>  3:            1 #>  4:            1 #>  5:       random #>  6:            1 #>  7:    <list[0]> #>  8:    <list[0]> #>  9:            0 #> 10:            0 #> 11:       [NULL] #> 12:       [NULL] #> 13:       [NULL] #> 14:       [NULL] #> 15:       [NULL] #> 16:       [NULL] #> 17:       [NULL] #> 18:       [NULL] #> 19:       [NULL] #> 20:       [NULL] #> 21:       [NULL] #> 22:       [NULL] #> 23:              #> 24:          0.5 #> 25:       [NULL] #> 26: <nn_relu[1]> #> 27:    <list[0]> #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #>            value"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_data_descriptor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to Data Descriptor — as_data_descriptor","title":"Convert to Data Descriptor — as_data_descriptor","text":"Converts input DataDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_data_descriptor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to Data Descriptor — as_data_descriptor","text":"","code":"as_data_descriptor(x, dataset_shapes, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_data_descriptor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to Data Descriptor — as_data_descriptor","text":"x () Object convert. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. ... () arguments passed DataDescriptor constructor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_data_descriptor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert to Data Descriptor — as_data_descriptor","text":"","code":"ds = dataset(\"example\",   initialize = function() self$iris = iris[, -5],   .getitem = function(i) list(x = torch_tensor(as.numeric(self$iris[i, ]))),   .length = function() nrow(self$iris) )() as_data_descriptor(ds, list(x = c(NA, 4L))) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,4)] #> * input_map: (x) -> Graph #> * pointer: nop.6ce271.x.output #> * shape: [(NA,4)]  # if the dataset has a .getbatch method, the shapes are inferred ds2 = dataset(\"example\",   initialize = function() self$iris = iris[, -5],   .getbatch = function(i) list(x = torch_tensor(as.matrix(self$iris[i, ]))),   .length = function() nrow(self$iris) )() as_data_descriptor(ds2) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,4)] #> * input_map: (x) -> Graph #> * pointer: nop.9ba523.x.output #> * shape: [(NA,4)]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to Lazy Tensor — as_lazy_tensor","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"Convert object lazy_tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"as_lazy_tensor(x, ...)  # S3 method for class 'dataset' as_lazy_tensor(x, dataset_shapes = NULL, ids = NULL, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"x () Object convert lazy_tensor ... () Additional arguments passed method. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. ids (integer()) ids include lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"iris_ds = dataset(\"iris\",   initialize = function() {     self$iris = iris[, -5]   },   .getbatch = function(i) {     list(x = torch_tensor(as.matrix(self$iris[i, ])))   },   .length = function() nrow(self$iris) )() # no need to specify the dataset shapes as they can be inferred from the .getbatch method # only first 5 observations as_lazy_tensor(iris_ds, ids = 1:5) #> <ltnsr[5]> #> [1] <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> # all observations head(as_lazy_tensor(iris_ds)) #> <ltnsr[6]> #> [1] <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]>  iris_ds2 = dataset(\"iris\",   initialize = function() self$iris = iris[, -5],   .getitem = function(i) list(x = torch_tensor(as.numeric(self$iris[i, ]))),   .length = function() nrow(self$iris) )() # if .getitem is implemented we cannot infer the shapes as they might vary, # so we have to annotate them explicitly as_lazy_tensor(iris_ds2, dataset_shapes = list(x = c(NA, 4L)))[1:5] #> <ltnsr[5]> #> [1] <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]> <tnsr[4]>  # Convert a matrix lt = as_lazy_tensor(matrix(rnorm(100), nrow = 20)) materialize(lt[1:5], rbind = TRUE) #> torch_tensor #> -1.4000  0.4682  0.0700  1.0743  1.9243 #>  0.2553  0.3630 -0.6391 -0.6651  1.2984 #> -2.4373 -1.3045 -0.0500  1.1140  0.7488 #> -0.0056  0.7378 -0.2515 -0.2459  0.5562 #>  0.6216  1.8885  0.4448 -1.1776 -0.5483 #> [ CPUFloatType{5,5} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callback.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to a TorchCallback — as_torch_callback","title":"Convert to a TorchCallback — as_torch_callback","text":"Converts object TorchCallback.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callback.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to a TorchCallback — as_torch_callback","text":"","code":"as_torch_callback(x, clone = FALSE, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to a TorchCallback — as_torch_callback","text":"x () Object converted. clone (logical(1)) Whether make deep clone. ... () Additional arguments","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callback.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert to a TorchCallback — as_torch_callback","text":"TorchCallback.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callbacks.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to a list of Torch Callbacks — as_torch_callbacks","title":"Convert to a list of Torch Callbacks — as_torch_callbacks","text":"Converts object list TorchCallback.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callbacks.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to a list of Torch Callbacks — as_torch_callbacks","text":"","code":"as_torch_callbacks(x, clone, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callbacks.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to a list of Torch Callbacks — as_torch_callbacks","text":"x () Object convert. clone (logical(1)) Whether create deep clone. ... () Additional arguments.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_callbacks.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert to a list of Torch Callbacks — as_torch_callbacks","text":"list() TorchCallbacks","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to TorchLoss — as_torch_loss","title":"Convert to TorchLoss — as_torch_loss","text":"Converts object TorchLoss.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to TorchLoss — as_torch_loss","text":"","code":"as_torch_loss(x, clone = FALSE, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to TorchLoss — as_torch_loss","text":"x () Object convert TorchLoss. clone (logical(1)) Whether make deep clone. ... () Additional arguments. Currently used pass additional constructor arguments TorchLoss objects type nn_loss.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert to TorchLoss — as_torch_loss","text":"TorchLoss.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_optimizer.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to TorchOptimizer — as_torch_optimizer","title":"Convert to TorchOptimizer — as_torch_optimizer","text":"Converts object TorchOptimizer.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_optimizer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to TorchOptimizer — as_torch_optimizer","text":"","code":"as_torch_optimizer(x, clone = FALSE, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_optimizer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to TorchOptimizer — as_torch_optimizer","text":"x () Object convert TorchOptimizer. clone (logical(1)) Whether make deep clone. Default FALSE. ... () Additional arguments. Currently used pass additional constructor arguments TorchOptimizer objects type torch_optimizer_generator.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_torch_optimizer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert to TorchOptimizer — as_torch_optimizer","text":"TorchOptimizer","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/assert_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Assert Lazy Tensor — assert_lazy_tensor","title":"Assert Lazy Tensor — assert_lazy_tensor","text":"Asserts whether something lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/assert_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assert Lazy Tensor — assert_lazy_tensor","text":"","code":"assert_lazy_tensor(x)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/assert_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Assert Lazy Tensor — assert_lazy_tensor","text":"x () Object check.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/auto_device.html","id":null,"dir":"Reference","previous_headings":"","what":"Auto Device — auto_device","title":"Auto Device — auto_device","text":"First tries cuda, cpu.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/auto_device.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Auto Device — auto_device","text":"","code":"auto_device(device = NULL)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/auto_device.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Auto Device — auto_device","text":"device (character(1)) device. NULL, returned .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_categ.html","id":null,"dir":"Reference","previous_headings":"","what":"Batchgetter for Categorical data — batchgetter_categ","title":"Batchgetter for Categorical data — batchgetter_categ","text":"Converts data frame categorical data long tensor converting data integers. input checks performed.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_categ.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Batchgetter for Categorical data — batchgetter_categ","text":"","code":"batchgetter_categ(data, device, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_categ.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Batchgetter for Categorical data — batchgetter_categ","text":"data (data.table)data.table converted tensor. device (character(1)) device. ... () Unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_num.html","id":null,"dir":"Reference","previous_headings":"","what":"Batchgetter for Numeric Data — batchgetter_num","title":"Batchgetter for Numeric Data — batchgetter_num","text":"Converts data frame numeric data float tensor calling .matrix(). input checks performed","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_num.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Batchgetter for Numeric Data — batchgetter_num","text":"","code":"batchgetter_num(data, device, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/batchgetter_num.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Batchgetter for Numeric Data — batchgetter_num","text":"data (data.table())data.table converted tensor. device (character(1)) device tensor created. ... () Unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/callback_set.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Set of Callbacks for Torch — callback_set","title":"Create a Set of Callbacks for Torch — callback_set","text":"Creates R6ClassGenerator inheriting CallbackSet. Additionally performs checks stages accidentally misspelled. create TorchCallback use torch_callback(). order resulting class cloneable, private method $deep_clone() must provided.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/callback_set.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Set of Callbacks for Torch — callback_set","text":"","code":"callback_set(   classname,   on_begin = NULL,   on_end = NULL,   on_exit = NULL,   on_epoch_begin = NULL,   on_before_valid = NULL,   on_epoch_end = NULL,   on_batch_begin = NULL,   on_batch_end = NULL,   on_after_backward = NULL,   on_batch_valid_begin = NULL,   on_batch_valid_end = NULL,   on_valid_end = NULL,   state_dict = NULL,   load_state_dict = NULL,   initialize = NULL,   public = NULL,   private = NULL,   active = NULL,   parent_env = parent.frame(),   inherit = CallbackSet,   lock_objects = FALSE )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/callback_set.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Set of Callbacks for Torch — callback_set","text":"classname (character(1)) class name. on_begin, on_end, on_epoch_begin, on_before_valid, on_epoch_end, on_batch_begin, on_batch_end, on_after_backward, on_batch_valid_begin, on_batch_valid_end, on_valid_end, on_exit (function) Function execute given stage, see section Stages. state_dict (function()) function retrieves state dict callback. available learner training. load_state_dict (function(state_dict)) Function loads callback state. initialize (function()) initialization method callback. public, private, active (list()) Additional public, private, active fields add callback. parent_env (environment()) parent environment R6Class. inherit (R6ClassGenerator) class inherit. class must either CallbackSet (default) inherit . lock_objects (logical(1)) Whether lock objects resulting R6Class. FALSE (default), values can freely assigned self without declaring class definition.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/callback_set.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Set of Callbacks for Torch — callback_set","text":"CallbackSet","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/equals-.lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compare lazy tensors — ==.lazy_tensor","title":"Compare lazy tensors — ==.lazy_tensor","text":"Compares lazy tensors using indices data descriptor's hash. means two lazy_tensors: equal: mateterialize tensors. unequal: might materialize tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/equals-.lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compare lazy tensors — ==.lazy_tensor","text":"","code":"# S3 method for class 'lazy_tensor' x == y"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/equals-.lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compare lazy tensors — ==.lazy_tensor","text":"x, y (lazy_tensor) Values compare.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/is_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Check for lazy tensor — is_lazy_tensor","title":"Check for lazy tensor — is_lazy_tensor","text":"Checks whether object lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/is_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check for lazy tensor — is_lazy_tensor","text":"","code":"is_lazy_tensor(x)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/is_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Check for lazy tensor — is_lazy_tensor","text":"x () Object check.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a lazy tensor — lazy_tensor","title":"Create a lazy tensor — lazy_tensor","text":"Create lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a lazy tensor — lazy_tensor","text":"","code":"lazy_tensor(data_descriptor = NULL, ids = NULL)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a lazy tensor — lazy_tensor","text":"data_descriptor (DataDescriptor NULL) data descriptor NULL lazy tensor length 0. ids (integer()) elements data_descriptor included lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/lazy_tensor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a lazy tensor — lazy_tensor","text":"","code":"ds = dataset(\"example\",   initialize = function() self$iris = iris[, -5],   .getitem = function(i) list(x = torch_tensor(as.numeric(self$iris[i, ]))),   .length = function() nrow(self$iris) )() dd = as_data_descriptor(ds, list(x = c(NA, 4L))) lt = as_lazy_tensor(dd)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":null,"dir":"Reference","previous_headings":"","what":"Materialize Lazy Tensor Columns — materialize","title":"Materialize Lazy Tensor Columns — materialize","text":"materialize lazy_tensor() data.frame() / list() containing – among things – lazy_tensor() columns. .e. data described underlying DataDescriptors loaded indices lazy_tensor(), preprocessed put unto specified device. elements lazy tensor must shape, list tensors returned default. elements shape, tensors can also rbinded single tensor (parameter rbind).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Materialize Lazy Tensor Columns — materialize","text":"","code":"materialize(x, device = \"cpu\", rbind = FALSE, ...)  # S3 method for class 'list' materialize(x, device = \"cpu\", rbind = FALSE, cache = \"auto\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Materialize Lazy Tensor Columns — materialize","text":"x () object materialize. Either lazy_tensor list() / data.frame() containing lazy_tensor columns. device (character(1)) torch device. rbind (logical(1)) Whether rbind lazy tensor columns (TRUE) return list tensors (FALSE). second case, batch dimension. ... () Additional arguments. cache (character(1) environment() NULL) Optional cache (intermediate) materialization results. Per default, caching enabled dataset data descriptor (different output pointer) used one lazy tensor column.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Materialize Lazy Tensor Columns — materialize","text":"(list() lazy_tensors lazy_tensor)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Materialize Lazy Tensor Columns — materialize","text":"Materializing lazy tensor consists : Loading data internal dataset DataDescriptor. Processing batches preprocessing Graphs. Returning result PipeOp pointed DataDescriptor (pointer). multiple lazy_tensor columns can benefit caching : ) Output(s) dataset might input multiple graphs. b) Different lazy tensors might outputs graph. reason possible provide cache environment. hash key ) hash indices dataset. hash key b) hash indices, dataset preprocessing graph.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Materialize Lazy Tensor Columns — materialize","text":"","code":"lt1 = as_lazy_tensor(torch_randn(10, 3)) materialize(lt1, rbind = TRUE) #> torch_tensor #>  0.1876  0.0088  1.4593 #> -0.1544 -1.7849 -0.9684 #> -0.4133  0.4283  1.5972 #> -1.1213  1.9914  0.4018 #>  1.4719 -0.4199 -0.5979 #>  1.1147  1.8712  0.3224 #> -0.2409  1.8200 -0.1913 #> -1.0128  1.6925 -0.1613 #>  0.0527 -0.0583  1.5428 #>  0.3041  0.7279  0.2507 #> [ CPUFloatType{10,3} ] materialize(lt1, rbind = FALSE) #> [[1]] #> torch_tensor #>  0.1876 #>  0.0088 #>  1.4593 #> [ CPUFloatType{3} ] #>  #> [[2]] #> torch_tensor #> -0.1544 #> -1.7849 #> -0.9684 #> [ CPUFloatType{3} ] #>  #> [[3]] #> torch_tensor #> -0.4133 #>  0.4283 #>  1.5972 #> [ CPUFloatType{3} ] #>  #> [[4]] #> torch_tensor #> -1.1213 #>  1.9914 #>  0.4018 #> [ CPUFloatType{3} ] #>  #> [[5]] #> torch_tensor #>  1.4719 #> -0.4199 #> -0.5979 #> [ CPUFloatType{3} ] #>  #> [[6]] #> torch_tensor #>  1.1147 #>  1.8712 #>  0.3224 #> [ CPUFloatType{3} ] #>  #> [[7]] #> torch_tensor #> -0.2409 #>  1.8200 #> -0.1913 #> [ CPUFloatType{3} ] #>  #> [[8]] #> torch_tensor #> -1.0128 #>  1.6925 #> -0.1613 #> [ CPUFloatType{3} ] #>  #> [[9]] #> torch_tensor #>  0.0527 #> -0.0583 #>  1.5428 #> [ CPUFloatType{3} ] #>  #> [[10]] #> torch_tensor #>  0.3041 #>  0.7279 #>  0.2507 #> [ CPUFloatType{3} ] #>  lt2 = as_lazy_tensor(torch_randn(10, 4)) d = data.table::data.table(lt1 = lt1, lt2 = lt2) materialize(d, rbind = TRUE) #> $lt1 #> torch_tensor #>  0.1876  0.0088  1.4593 #> -0.1544 -1.7849 -0.9684 #> -0.4133  0.4283  1.5972 #> -1.1213  1.9914  0.4018 #>  1.4719 -0.4199 -0.5979 #>  1.1147  1.8712  0.3224 #> -0.2409  1.8200 -0.1913 #> -1.0128  1.6925 -0.1613 #>  0.0527 -0.0583  1.5428 #>  0.3041  0.7279  0.2507 #> [ CPUFloatType{10,3} ] #>  #> $lt2 #> torch_tensor #>  0.5711 -0.6314 -0.7923  0.0621 #>  0.9379  0.8469 -0.1051  0.7225 #> -0.2402 -0.9588 -0.8237 -0.3911 #> -0.2728 -0.8515 -1.2257  0.1618 #> -0.5972  1.4824  0.8495  1.8945 #> -0.9398 -0.5522 -1.4726  0.3737 #> -0.6116 -0.5517  0.6216 -2.0019 #> -0.4594 -1.2072  0.9961 -0.5702 #> -0.0091  0.3094 -0.0514 -1.8272 #> -0.9225  0.5034  0.1760 -1.4413 #> [ CPUFloatType{10,4} ] #>  materialize(d, rbind = FALSE) #> $lt1 #> $lt1[[1]] #> torch_tensor #>  0.1876 #>  0.0088 #>  1.4593 #> [ CPUFloatType{3} ] #>  #> $lt1[[2]] #> torch_tensor #> -0.1544 #> -1.7849 #> -0.9684 #> [ CPUFloatType{3} ] #>  #> $lt1[[3]] #> torch_tensor #> -0.4133 #>  0.4283 #>  1.5972 #> [ CPUFloatType{3} ] #>  #> $lt1[[4]] #> torch_tensor #> -1.1213 #>  1.9914 #>  0.4018 #> [ CPUFloatType{3} ] #>  #> $lt1[[5]] #> torch_tensor #>  1.4719 #> -0.4199 #> -0.5979 #> [ CPUFloatType{3} ] #>  #> $lt1[[6]] #> torch_tensor #>  1.1147 #>  1.8712 #>  0.3224 #> [ CPUFloatType{3} ] #>  #> $lt1[[7]] #> torch_tensor #> -0.2409 #>  1.8200 #> -0.1913 #> [ CPUFloatType{3} ] #>  #> $lt1[[8]] #> torch_tensor #> -1.0128 #>  1.6925 #> -0.1613 #> [ CPUFloatType{3} ] #>  #> $lt1[[9]] #> torch_tensor #>  0.0527 #> -0.0583 #>  1.5428 #> [ CPUFloatType{3} ] #>  #> $lt1[[10]] #> torch_tensor #>  0.3041 #>  0.7279 #>  0.2507 #> [ CPUFloatType{3} ] #>  #>  #> $lt2 #> $lt2[[1]] #> torch_tensor #>  0.5711 #> -0.6314 #> -0.7923 #>  0.0621 #> [ CPUFloatType{4} ] #>  #> $lt2[[2]] #> torch_tensor #>  0.9379 #>  0.8469 #> -0.1051 #>  0.7225 #> [ CPUFloatType{4} ] #>  #> $lt2[[3]] #> torch_tensor #> -0.2402 #> -0.9588 #> -0.8237 #> -0.3911 #> [ CPUFloatType{4} ] #>  #> $lt2[[4]] #> torch_tensor #> -0.2728 #> -0.8515 #> -1.2257 #>  0.1618 #> [ CPUFloatType{4} ] #>  #> $lt2[[5]] #> torch_tensor #> -0.5972 #>  1.4824 #>  0.8495 #>  1.8945 #> [ CPUFloatType{4} ] #>  #> $lt2[[6]] #> torch_tensor #> -0.9398 #> -0.5522 #> -1.4726 #>  0.3737 #> [ CPUFloatType{4} ] #>  #> $lt2[[7]] #> torch_tensor #> -0.6116 #> -0.5517 #>  0.6216 #> -2.0019 #> [ CPUFloatType{4} ] #>  #> $lt2[[8]] #> torch_tensor #> -0.4594 #> -1.2072 #>  0.9961 #> -0.5702 #> [ CPUFloatType{4} ] #>  #> $lt2[[9]] #> torch_tensor #> -0.0091 #>  0.3094 #> -0.0514 #> -1.8272 #> [ CPUFloatType{4} ] #>  #> $lt2[[10]] #> torch_tensor #> -0.9225 #>  0.5034 #>  0.1760 #> -1.4413 #> [ CPUFloatType{4} ] #>  #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":null,"dir":"Reference","previous_headings":"","what":"Materialize a Lazy Tensor — materialize_internal","title":"Materialize a Lazy Tensor — materialize_internal","text":"Convert lazy_tensor torch_tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Materialize a Lazy Tensor — materialize_internal","text":"","code":"materialize_internal(x, device = \"cpu\", cache = NULL, rbind)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Materialize a Lazy Tensor — materialize_internal","text":"x (lazy_tensor()) lazy tensor materialize. device (character(1L)) device put materialized tensor (running preprocessing graph). cache (NULL environment()) Whether cache (intermediate) results materialization. can make data loading faster multiple lazy_tensors reference dataset graph. rbind (logical(1)) Whtether rbind resulting tensors (TRUE) return list tensors (FALSE).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Materialize a Lazy Tensor — materialize_internal","text":"lazy_tensor()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/materialize_internal.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Materialize a Lazy Tensor — materialize_internal","text":"Materializing lazy tensor consists : Loading data internal dataset DataDescriptor. Processing batches preprocessing Graphs. Returning result PipeOp pointed DataDescriptor (pointer). materializing multiple lazy_tensor columns, caching can useful : ) Output(s) dataset might input multiple graphs. (task_dataset shoudl rarely case try merge ). b) Different lazy tensors might outputs graph. reason possible provide cache environment. hash key ) hash indices dataset. hash key b) hash indices dataset preprocessing graph.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch-package.html","id":null,"dir":"Reference","previous_headings":"","what":"mlr3torch: Deep Learning with 'mlr3' — mlr3torch-package","title":"mlr3torch: Deep Learning with 'mlr3' — mlr3torch-package","text":"Deep Learning library extends mlr3 framework building upon 'torch' package. allows conveniently build, train, evaluate deep learning models without worry low level details. Custom architectures can created using graph language defined 'mlr3pipelines'.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch-package.html","id":"options","dir":"Reference","previous_headings":"","what":"Options","title":"mlr3torch: Deep Learning with 'mlr3' — mlr3torch-package","text":"mlr3torch.cache: Whether cache downloaded data (TRUE) (FALSE, default). can also set specific folder file system used cache directory.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch-package.html","id":"author","dir":"Reference","previous_headings":"","what":"Author","title":"mlr3torch: Deep Learning with 'mlr3' — mlr3torch-package","text":"Maintainer: Sebastian Fischer sebf.fischer@gmail.com (ORCID) Authors: Martin Binder mlr.developer@mb706.com contributors: Bernd Bischl bernd_bischl@gmx.net (ORCID) [contributor] Lukas Burk github@quantenbrot.de (ORCID) [contributor] Florian Pfisterer pfistererf@googlemail.com (ORCID) [contributor] Carson Zhang carsonzhang4@gmail.com [contributor]","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_callbacks.html","id":null,"dir":"Reference","previous_headings":"","what":"Dictionary of Torch Callbacks — mlr3torch_callbacks","title":"Dictionary of Torch Callbacks — mlr3torch_callbacks","text":"mlr3misc::Dictionary torch callbacks. Use t_clbk() conveniently retrieve callbacks. Can converted data.table using .data.table.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_callbacks.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dictionary of Torch Callbacks — mlr3torch_callbacks","text":"","code":"mlr3torch_callbacks"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_callbacks.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Dictionary of Torch Callbacks — mlr3torch_callbacks","text":"object class DictionaryMlr3torchCallbacks (inherits Dictionary, R6) length 13.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_callbacks.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Dictionary of Torch Callbacks — mlr3torch_callbacks","text":"","code":"mlr3torch_callbacks$get(\"checkpoint\") #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: list() #> * Packages: mlr3torch,torch # is the same as t_clbk(\"checkpoint\") #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: list() #> * Packages: mlr3torch,torch # convert to a data.table as.data.table(mlr3torch_callbacks) #> Key: <key> #>           key       label                 packages #>        <char>      <char>                   <char> #> 1: checkpoint  Checkpoint          mlr3torch,torch #> 2:    history     History          mlr3torch,torch #> 3:   progress    Progress progress,mlr3torch,torch #> 4:         tb TensorBoard tfevents,mlr3torch,torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss Functions — mlr3torch_losses","title":"Loss Functions — mlr3torch_losses","text":"Dictionary torch loss descriptors. See t_loss() conveniently retrieving loss function. Can converted data.table using .data.table.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss Functions — mlr3torch_losses","text":"","code":"mlr3torch_losses"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Loss Functions — mlr3torch_losses","text":"object class DictionaryMlr3torchLosses (inherits Dictionary, R6) length 13.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":"available-loss-functions","dir":"Reference","previous_headings":"","what":"Available Loss Functions","title":"Loss Functions — mlr3torch_losses","text":"cross_entropy, l1, mse","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_losses.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loss Functions — mlr3torch_losses","text":"","code":"mlr3torch_losses$get(\"mse\") #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr # is equivalent to t_loss(\"mse\") #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr # convert to a data.table as.data.table(mlr3torch_losses) #> Key: <key> #>              key              label task_types        packages #>           <char>             <char>     <list>          <char> #> 1: cross_entropy      Cross Entropy    classif torch,mlr3torch #> 2:            l1     Absolute Error       regr torch,mlr3torch #> 3:           mse Mean Squared Error       regr torch,mlr3torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizers — mlr3torch_optimizers","title":"Optimizers — mlr3torch_optimizers","text":"Dictionary torch optimizers. Use t_opt conveniently retrieving optimizers. Can converted data.table using .data.table.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizers — mlr3torch_optimizers","text":"","code":"mlr3torch_optimizers"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Optimizers — mlr3torch_optimizers","text":"object class DictionaryMlr3torchOptimizers (inherits Dictionary, R6) length 13.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":"available-optimizers","dir":"Reference","previous_headings":"","what":"Available Optimizers","title":"Optimizers — mlr3torch_optimizers","text":"adadelta, adagrad, adam, asgd, rmsprop, rprop, sgd","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr3torch_optimizers.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Optimizers — mlr3torch_optimizers","text":"","code":"mlr3torch_optimizers$get(\"adam\") #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # is equivalent to t_opt(\"adam\") #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: list() #> * Packages: torch,mlr3torch # convert to a data.table as.data.table(mlr3torch_optimizers) #> Key: <key> #>         key                         label        packages #>      <char>                        <char>          <char> #> 1: adadelta Adaptive Learning Rate Method torch,mlr3torch #> 2:  adagrad   Adaptive Gradient algorithm torch,mlr3torch #> 3:     adam    Adaptive Moment Estimation torch,mlr3torch #> 4:     asgd  SGD with Adaptive Batch Size torch,mlr3torch #> 5:  rmsprop  Root Mean Square Propagation torch,mlr3torch #> 6:    rprop     Resilient Backpropagation torch,mlr3torch #> 7:      sgd   Stochastic Gradient Descent torch,mlr3torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":null,"dir":"Reference","previous_headings":"","what":"Lazy Data Backend — mlr_backends_lazy","title":"Lazy Data Backend — mlr_backends_lazy","text":"lazy data backend wraps constructor lazily creates another backend, e.g. downloading (caching) data internet. backend used, metadata backend known advance accessible downloading actual data. backend first constructed, verified provided metadata correct, otherwise informative error message thrown. construction lazily constructed backend, calls like $data(), $missings(), $distinct(), $hash() redirected . Information available backend constructed : nrow - number rows (set length rownames). ncol - number columns (provided via id column col_info). colnames - column names. rownames - row names. col_info - column information, can obtained via mlr3::col_info(). Beware accessing backend's hash also contructs backend. Note cases data contains lazy_tensor columns, necessary naming class nothing lazy_tensor data type. Important constructor generates factor() variables important ordering levels data corresponds ordering levels col_info argument.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Lazy Data Backend — mlr_backends_lazy","text":"mlr3::DataBackend -> DataBackendLazy","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Lazy Data Backend — mlr_backends_lazy","text":"backend (DataBackend) wrapped backend lazily constructed first accessed. nrow (integer(1)) Number rows (observations). ncol (integer(1)) Number columns (variables), including primary key column. rownames (integer()) Returns vector distinct row identifiers, .e. contents primary key column. colnames (character()) Returns vector column names, including primary key column. is_constructed (logical(1)) Whether backend already constructed.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Lazy Data Backend — mlr_backends_lazy","text":"mlr3::DataBackend$format()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Lazy Data Backend — mlr_backends_lazy","text":"DataBackendLazy$new() DataBackendLazy$data() DataBackendLazy$head() DataBackendLazy$distinct() DataBackendLazy$missings() DataBackendLazy$print()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$new(constructor, rownames, col_info, primary_key)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"constructor (function) function argument backend (lazy backend), whose return value must actual backend. function called first time field $backend accessed. rownames (integer()) row names. Must permutation rownames lazily constructed backend. col_info (data.table::data.table()) data.table columns id, type levels containing column id, type levels. Note levels must provided correct order. primary_key (character(1)) Name primary key column.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-data-","dir":"Reference","previous_headings":"","what":"Method data()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Returns slice data specified format. rows must addressed vector primary key values, columns must referred via column names. Queries rows matching row id queries columns matching column name silently ignored. Rows guaranteed returned order rows, columns may returned arbitrary order. Duplicated row ids result duplicated rows, duplicated column names lead exception. Accessing data triggers construction backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$data(rows, cols)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"rows (integer()) Row indices. cols (character()) Column names.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-head-","dir":"Reference","previous_headings":"","what":"Method head()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Retrieve first n rows. triggers construction backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$head(n = 6L)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"n (integer(1)) Number rows.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Lazy Data Backend — mlr_backends_lazy","text":"data.table::data.table() first n rows.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-distinct-","dir":"Reference","previous_headings":"","what":"Method distinct()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Returns named list vectors distinct values column specified. na_rm TRUE, missing values removed returned vectors distinct values. Non-existing rows columns silently ignored. triggers construction backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$distinct(rows, cols, na_rm = TRUE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments-3","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"rows (integer()) Row indices. cols (character()) Column names. na_rm (logical(1)) Whether remove NAs .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"returns-1","dir":"Reference","previous_headings":"","what":"Returns","title":"Lazy Data Backend — mlr_backends_lazy","text":"Named list() distinct values.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-missings-","dir":"Reference","previous_headings":"","what":"Method missings()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Returns number missing values per column specified slice data. Non-existing rows columns silently ignored. triggers construction backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$missings(rows, cols)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"arguments-4","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lazy Data Backend — mlr_backends_lazy","text":"rows (integer()) Row indices. cols (character()) Column names.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"returns-2","dir":"Reference","previous_headings":"","what":"Returns","title":"Lazy Data Backend — mlr_backends_lazy","text":"Total missing values per column (named numeric()).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Lazy Data Backend — mlr_backends_lazy","text":"Printer.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"DataBackendLazy$print()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_backends_lazy.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lazy Data Backend — mlr_backends_lazy","text":"","code":"# We first define a backend constructor constructor = function(backend) {   cat(\"Data is constructed!\\n\")   DataBackendDataTable$new(     data.table(x = rnorm(10), y = rnorm(10), row_id = 1:10),     primary_key = \"row_id\"   ) }  # to wrap this backend constructor in a lazy backend, we need to provide the correct metadata for it column_info = data.table(   id = c(\"x\", \"y\", \"row_id\"),   type = c(\"numeric\", \"numeric\", \"integer\"),   levels = list(NULL, NULL, NULL) ) backend_lazy = DataBackendLazy$new(   constructor = constructor,   rownames = 1:10,   col_info = column_info,   primary_key = \"row_id\" )  # Note that the constructor is not called for the calls below # as they can be read from the metadata backend_lazy$nrow #> [1] 10 backend_lazy$rownames #>  [1]  1  2  3  4  5  6  7  8  9 10 backend_lazy$ncol #> [1] 3 backend_lazy$colnames #> [1] \"x\"      \"y\"      \"row_id\" col_info(backend_lazy) #>        id    type levels #>    <char>  <char> <list> #> 1:      x numeric [NULL] #> 2:      y numeric [NULL] #> 3: row_id integer [NULL]  # Only now the backend is constructed backend_lazy$data(1, \"x\") #> Data is constructed! #>             x #>         <num> #> 1: -0.3872136 # Is the same as: backend_lazy$backend$data(1, \"x\") #>             x #>         <num> #> 1: -0.3872136"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":null,"dir":"Reference","previous_headings":"","what":"Checkpoint Callback — mlr_callback_set.checkpoint","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saves optimizer network states training. final network optimizer always stored.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saving learner callback trained model impossible, model slot set last callback step executed.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"mlr3torch::CallbackSet -> CallbackSetCheckpoint","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"CallbackSetCheckpoint$new() CallbackSetCheckpoint$on_epoch_end() CallbackSetCheckpoint$on_batch_end() CallbackSetCheckpoint$on_exit() CallbackSetCheckpoint$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$new(path, freq, freq_type = \"epoch\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"path (character(1)) path folder models saved. freq (integer(1)) frequency often model saved. Frequency either per step epoch, can configured freq_type parameter. freq_type (character(1)) Can either \"epoch\" (default) \"step\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-on-epoch-end-","dir":"Reference","previous_headings":"","what":"Method on_epoch_end()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saves network optimizer state dict. nothing freq_type freq met.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$on_epoch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-on-batch-end-","dir":"Reference","previous_headings":"","what":"Method on_batch_end()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saves selected objects defined save. nothing freq_type freq met.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$on_batch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-on-exit-","dir":"Reference","previous_headings":"","what":"Method on_exit()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"Saves learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$on_exit()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"","code":"CallbackSetCheckpoint$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.checkpoint.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Checkpoint Callback — mlr_callback_set.checkpoint","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":null,"dir":"Reference","previous_headings":"","what":"History Callback — mlr_callback_set.history","title":"History Callback — mlr_callback_set.history","text":"Saves training validation history training. history saved data.table $train $valid slots. first column always epoch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"History Callback — mlr_callback_set.history","text":"mlr3torch::CallbackSet -> CallbackSetHistory","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"History Callback — mlr_callback_set.history","text":"mlr3torch::CallbackSet$print()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"History Callback — mlr_callback_set.history","text":"CallbackSetHistory$on_begin() CallbackSetHistory$state_dict() CallbackSetHistory$load_state_dict() CallbackSetHistory$on_before_valid() CallbackSetHistory$on_epoch_end() CallbackSetHistory$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-on-begin-","dir":"Reference","previous_headings":"","what":"Method on_begin()","title":"History Callback — mlr_callback_set.history","text":"Initializes lists train validation metrics stored.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$on_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-state-dict-","dir":"Reference","previous_headings":"","what":"Method state_dict()","title":"History Callback — mlr_callback_set.history","text":"Converts lists data.tables.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$state_dict()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-load-state-dict-","dir":"Reference","previous_headings":"","what":"Method load_state_dict()","title":"History Callback — mlr_callback_set.history","text":"Sets field $train $valid contained state dict.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$load_state_dict(state_dict)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"History Callback — mlr_callback_set.history","text":"state_dict (callback_state_history) state dict retrieved via $state_dict().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-on-before-valid-","dir":"Reference","previous_headings":"","what":"Method on_before_valid()","title":"History Callback — mlr_callback_set.history","text":"Add latest training scores history.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$on_before_valid()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-on-epoch-end-","dir":"Reference","previous_headings":"","what":"Method on_epoch_end()","title":"History Callback — mlr_callback_set.history","text":"Add latest validation scores history.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$on_epoch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"History Callback — mlr_callback_set.history","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"History Callback — mlr_callback_set.history","text":"","code":"CallbackSetHistory$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.history.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"History Callback — mlr_callback_set.history","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Callbacks — mlr_callback_set","title":"Base Class for Callbacks — mlr_callback_set","text":"Base class callbacks inherit (see section Inheriting). callback set collection functions executed different stages training loop. can used gain control training process neural network without write everything scratch. used torch learner, CallbackSet wrapped TorchCallback. latters parameter set represents arguments CallbackSet's $initialize() method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Callbacks — mlr_callback_set","text":"available stage (see section Stages) public method $on_<stage>() can defined. evaluation context (ContextTorch) can accessed via self$ctx, contains current state training loop. context assigned beginning training loop removed afterwards. Different stages callback can communicate assigning values $self. State: able store information $model slot LearnerTorch, callbacks support state API. can overload $state_dict() public method define stored learner$model$callbacks$<id> training finishes. also requires implement $load_state_dict(state_dict) method defines load previously saved callback state different callback. Note $state_dict() include parameter values used initialize callback. creating custom callbacks, function torch_callback() recommended, creates CallbackSet wraps TorchCallback. create CallbackSet convenience function callback_set() can used. functions perform checks stages accidentally misspelled.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"stages","dir":"Reference","previous_headings":"","what":"Stages","title":"Base Class for Callbacks — mlr_callback_set","text":"begin :: Run training loop begins. epoch_begin :: Run beginning epoch. batch_begin :: Run forward call. after_backward :: Run backward call. batch_end :: Run optimizer step. batch_valid_begin :: Run forward call validation loop. batch_valid_end :: Run forward call validation loop. valid_end :: Run end validation. epoch_end :: Run end epoch. end :: Run last epoch. exit :: Run last, using .exit().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"terminate-training","dir":"Reference","previous_headings":"","what":"Terminate Training","title":"Base Class for Callbacks — mlr_callback_set","text":"training stopped, possible set field $terminate ContextTorch. end every epoch field checked TRUE, training stops. can example used implement custom early stopping.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Base Class for Callbacks — mlr_callback_set","text":"ctx (ContextTorch NULL) evaluation context callback. field always NULL except $train() call torch learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Callbacks — mlr_callback_set","text":"stages (character()) active stages callback set.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Callbacks — mlr_callback_set","text":"CallbackSet$print() CallbackSet$state_dict() CallbackSet$load_state_dict() CallbackSet$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Callbacks — mlr_callback_set","text":"Prints object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Callbacks — mlr_callback_set","text":"","code":"CallbackSet$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Callbacks — mlr_callback_set","text":"... () Currently unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"method-state-dict-","dir":"Reference","previous_headings":"","what":"Method state_dict()","title":"Base Class for Callbacks — mlr_callback_set","text":"Returns information kept LearnerTorch's state training. information loadable callback using $load_state_dict() able continue training. returns NULL default.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Callbacks — mlr_callback_set","text":"","code":"CallbackSet$state_dict()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"method-load-state-dict-","dir":"Reference","previous_headings":"","what":"Method load_state_dict()","title":"Base Class for Callbacks — mlr_callback_set","text":"Loads state dict callback continue training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Callbacks — mlr_callback_set","text":"","code":"CallbackSet$load_state_dict(state_dict)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Callbacks — mlr_callback_set","text":"state_dict () state dict retrieved via $state_dict().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Callbacks — mlr_callback_set","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Callbacks — mlr_callback_set","text":"","code":"CallbackSet$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Callbacks — mlr_callback_set","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":null,"dir":"Reference","previous_headings":"","what":"Progress Callback — mlr_callback_set.progress","title":"Progress Callback — mlr_callback_set.progress","text":"Prints progress bar metrics training validation.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Progress Callback — mlr_callback_set.progress","text":"mlr3torch::CallbackSet -> CallbackSetProgress","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Progress Callback — mlr_callback_set.progress","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Progress Callback — mlr_callback_set.progress","text":"CallbackSetProgress$on_epoch_begin() CallbackSetProgress$on_batch_end() CallbackSetProgress$on_before_valid() CallbackSetProgress$on_batch_valid_end() CallbackSetProgress$on_epoch_end() CallbackSetProgress$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-epoch-begin-","dir":"Reference","previous_headings":"","what":"Method on_epoch_begin()","title":"Progress Callback — mlr_callback_set.progress","text":"Initializes progress bar training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_epoch_begin()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-batch-end-","dir":"Reference","previous_headings":"","what":"Method on_batch_end()","title":"Progress Callback — mlr_callback_set.progress","text":"Increments training progress bar.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_batch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-before-valid-","dir":"Reference","previous_headings":"","what":"Method on_before_valid()","title":"Progress Callback — mlr_callback_set.progress","text":"Creates progress bar validation.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_before_valid()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-batch-valid-end-","dir":"Reference","previous_headings":"","what":"Method on_batch_valid_end()","title":"Progress Callback — mlr_callback_set.progress","text":"Increments validation progress bar.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_batch_valid_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-on-epoch-end-","dir":"Reference","previous_headings":"","what":"Method on_epoch_end()","title":"Progress Callback — mlr_callback_set.progress","text":"Prints summary training validation process.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$on_epoch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Progress Callback — mlr_callback_set.progress","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"Progress Callback — mlr_callback_set.progress","text":"","code":"CallbackSetProgress$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.progress.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Progress Callback — mlr_callback_set.progress","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":null,"dir":"Reference","previous_headings":"","what":"TensorBoard Logging Callback — mlr_callback_set.tb","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"Logs training loss, training measures, validation measures events. view , use TensorBoard tensorflow::tensorboard() (requires tensorflow) CLI.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"Logs events every epoch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"mlr3torch::CallbackSet -> CallbackSetTB","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"mlr3torch::CallbackSet$load_state_dict() mlr3torch::CallbackSet$print() mlr3torch::CallbackSet$state_dict()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"CallbackSetTB$new() CallbackSetTB$on_epoch_end() CallbackSetTB$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"","code":"CallbackSetTB$new(path, log_train_loss)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"path (character(1)) path folder events logged. Point TensorBoard folder view . log_train_loss (logical(1)) Whether log training loss.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"method-on-epoch-end-","dir":"Reference","previous_headings":"","what":"Method on_epoch_end()","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"Logs training loss, training measures, validation measures TensorBoard events.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"","code":"CallbackSetTB$on_epoch_end()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"","code":"CallbackSetTB$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_callback_set.tb.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"TensorBoard Logging Callback — mlr_callback_set.tb","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Context for Torch Learner — mlr_context_torch","title":"Context for Torch Learner — mlr_context_torch","text":"Context training torch learner. - mostly read-- information callbacks access argument ctx. information callbacks, see CallbackSet.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Context for Torch Learner — mlr_context_torch","text":"learner (Learner) torch learner. task_train (Task) training task. task_valid (Task NULL) validation task. loader_train (torch::dataloader) data loader training. loader_valid (torch::dataloader) data loader validation. measures_train (list() Measures) Measures used training. measures_valid (list() Measures) Measures used validation. network (torch::nn_module) torch network. optimizer (torch::optimizer) optimizer. loss_fn (torch::nn_module) loss function. total_epochs (integer(1)) total number epochs learner trained . last_scores_train (named list() NULL) scores last training batch. Names ids training measures. LearnerTorch sets eval_freq different 1, NULL epochs evaluate model. last_scores_valid (list()) scores last validation batch. Names ids validation measures. LearnerTorch sets eval_freq different 1, NULL epochs evaluate model. epoch (integer(1)) current epoch. step (integer(1)) current iteration. prediction_encoder (function()) learner's prediction encoder. batch (named list() torch_tensors) current batch. terminate (logical(1)) field set TRUE end epoch, training stops.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Context for Torch Learner — mlr_context_torch","text":"ContextTorch$new() ContextTorch$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Context for Torch Learner — mlr_context_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Context for Torch Learner — mlr_context_torch","text":"","code":"ContextTorch$new(   learner,   task_train,   task_valid = NULL,   loader_train,   loader_valid = NULL,   measures_train = NULL,   measures_valid = NULL,   network,   optimizer,   loss_fn,   total_epochs,   prediction_encoder,   eval_freq = 1L )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Context for Torch Learner — mlr_context_torch","text":"learner (Learner) torch learner. task_train (Task) training task. task_valid (Task NULL) validation task. loader_train (torch::dataloader) data loader training. loader_valid (torch::dataloader NULL) data loader validation. measures_train (list() Measures NULL) Measures used training. Default NULL. measures_valid (list() Measures NULL) Measures used validation. network (torch::nn_module) torch network. optimizer (torch::optimizer) optimizer. loss_fn (torch::nn_module) loss function. total_epochs (integer(1)) total number epochs learner trained . prediction_encoder (function()) learner's prediction encoder. eval_freq (integer(1)) evaluation frequency.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Context for Torch Learner — mlr_context_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Context for Torch Learner — mlr_context_torch","text":"","code":"ContextTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_context_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Context for Torch Learner — mlr_context_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":null,"dir":"Reference","previous_headings":"","what":"My Little Pony — mlr_learners.mlp","title":"My Little Pony — mlr_learners.mlp","text":"Fully connected feed forward network dropout activation function. features can either single lazy_tensor one numeric columns ().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"My Little Pony — mlr_learners.mlp","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.mlp\", ...) lrn(\"regr.mlp\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"My Little Pony — mlr_learners.mlp","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “integer”, “numeric”, “lazy_tensor” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"My Little Pony — mlr_learners.mlp","text":"Parameters LearnerTorch, well : activation :: [nn_module] activation function. initialized nn_relu. activation_args :: named list() named list initialization arguments activation function. intialized empty list. neurons :: integer() number neurons per hidden layer. default hidden layer. Setting c(10, 20) first hidden layer 10 neurons second 20. n_layers :: integer() number layers. parameter must set neurons length 1. p :: numeric(1) dropout probability. initialized 0.5. shape :: integer() NULL input shape length 2, e.g. c(NA, 5). needs present lazy tensor input unknown shape (NULL). Otherwise input shape inferred number numeric features.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"My Little Pony — mlr_learners.mlp","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchMLP","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"My Little Pony — mlr_learners.mlp","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"My Little Pony — mlr_learners.mlp","text":"LearnerTorchMLP$new() LearnerTorchMLP$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"My Little Pony — mlr_learners.mlp","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"My Little Pony — mlr_learners.mlp","text":"","code":"LearnerTorchMLP$new(   task_type,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"My Little Pony — mlr_learners.mlp","text":"task_type (character(1)) task type, either \"classif\" \"regr\". optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"My Little Pony — mlr_learners.mlp","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"My Little Pony — mlr_learners.mlp","text":"","code":"LearnerTorchMLP$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"My Little Pony — mlr_learners.mlp","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.mlp.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"My Little Pony — mlr_learners.mlp","text":"","code":"# Define the Learner and set parameter values learner = lrn(\"classif.mlp\") learner$param_set$set_values(   epochs = 1, batch_size = 16, device = \"cpu\",   neurons = 10 )  # Define a Task task = tsk(\"iris\")  # Create train and test set ids = partition(task)  # Train the learner on the training ids learner$train(task, row_ids = ids$train)  # Make predictions for the test rows predictions = learner$predict(task, row_ids = ids$test)  # Score the predictions predictions$score() #> classif.ce  #>       0.68"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":null,"dir":"Reference","previous_headings":"","what":"Tabular ResNet — mlr_learners.tab_resnet","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Tabular resnet.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.tab_resnet\", ...) lrn(\"regr.tab_resnet\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “integer”, “numeric” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Parameters LearnerTorch, well : n_blocks :: integer(1) number blocks. d_block :: integer(1) input output dimension block. d_hidden :: integer(1) latent dimension block. d_hidden_multiplier :: numeric(1) Alternative way specify latent dimension d_block * d_hidden_multiplier. dropout1 :: numeric(1) First dropout ratio. dropout2 :: numeric(1) Second dropout ratio.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Gorishniy Y, Rubachev , Khrulkov V, Babenko (2021). “Revisiting Deep Learning  Tabular Data.” arXiv, 2106.11959.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchTabResNet","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"LearnerTorchTabResNet$new() LearnerTorchTabResNet$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"LearnerTorchTabResNet$new(   task_type,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"task_type (character(1)) task type, either \"classif\" \"regr\". optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"LearnerTorchTabResNet$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.tab_resnet.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tabular ResNet — mlr_learners.tab_resnet","text":"","code":"# Define the Learner and set parameter values learner = lrn(\"classif.tab_resnet\") learner$param_set$set_values(   epochs = 1, batch_size = 16, device = \"cpu\",   n_blocks = 2, d_block = 10, d_hidden = 20, dropout1 = 0.3, dropout2 = 0.3 )  # Define a Task task = tsk(\"iris\")  # Create train and test set ids = partition(task)  # Train the learner on the training ids learner$train(task, row_ids = ids$train)  # Make predictions for the test rows predictions = learner$predict(task, row_ids = ids$test)  # Score the predictions predictions$score() #> classif.ce  #>        0.9"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":null,"dir":"Reference","previous_headings":"","what":"Featureless Torch Learner — mlr_learners.torch_featureless","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"Featureless torch learner. Output constant weight learned training. classification, (asymptoptically) result majority class prediction using standard cross-entropy loss. regression, result median L1 loss mean L2 loss.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"dictionary","dir":"Reference","previous_headings":"","what":"Dictionary","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"Learner can instantiated using sugar function lrn():","code":"lrn(\"classif.torch_featureless\", ...) lrn(\"regr.torch_featureless\", ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"Supported task types: 'classif', 'regr' Predict Types: classif: 'response', 'prob' regr: 'response' Feature Types: “logical”, “integer”, “numeric”, “character”, “factor”, “ordered”, “POSIXct”, “lazy_tensor” Required Packages: mlr3, mlr3torch, torch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"LearnerTorch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchFeatureless","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"LearnerTorchFeatureless$new() LearnerTorchFeatureless$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"","code":"LearnerTorchFeatureless$new(   task_type,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"task_type (character(1)) task type, either \"classif\" \"regr\". optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"","code":"LearnerTorchFeatureless$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torch_featureless.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Featureless Torch Learner — mlr_learners.torch_featureless","text":"","code":"# Define the Learner and set parameter values learner = lrn(\"classif.torch_featureless\") learner$param_set$set_values(   epochs = 1, batch_size = 16, device = \"cpu\"    )  # Define a Task task = tsk(\"iris\")  # Create train and test set ids = partition(task)  # Train the learner on the training ids learner$train(task, row_ids = ids$train)  # Make predictions for the test rows predictions = learner$predict(task, row_ids = ids$test)  # Score the predictions predictions$score() #> classif.ce  #>       0.64"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":null,"dir":"Reference","previous_headings":"","what":"AlexNet Image Classifier — mlr_learners.torchvision","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Classic image classification networks torchvision.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Parameters LearnerTorchImage pretrained :: logical(1) Whether use pretrained model. final linear layer replaced new nn_linear number classes inferred Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Supported task types: \"classif\" Predict Types: \"response\" \"prob\" Feature Types: \"lazy_tensor\" Required packages: \"mlr3torch\", \"torch\", \"torchvision\"","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> mlr3torch::LearnerTorchImage -> LearnerTorchVision","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"LearnerTorchVision$new() LearnerTorchVision$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"","code":"LearnerTorchVision$new(   name,   module_generator,   label,   optimizer = NULL,   loss = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"name (character(1)) name network. module_generator (function(pretrained, num_classes)) Function generates network. label (character(1)) label network. #' @references Krizhevsky, Alex, Sutskever, Ilya, Hinton, E. G (2017). “Imagenet classification deep convolutional neural networks.” Communications ACM, 60(6), 84–90. Sandler, Mark, Howard, Andrew, Zhu, Menglong, Zhmoginov, Andrey, Chen, Liang-Chieh (2018). “Mobilenetv2: Inverted residuals linear bottlenecks.” Proceedings IEEE conference computer vision pattern recognition, 4510–4520. , Kaiming, Zhang, Xiangyu, Ren, Shaoqing, Sun, Jian (2016). “Deep residual learning image recognition.” Proceedings IEEE conference computer vision pattern recognition, 770–778. Simonyan, Karen, Zisserman, Andrew (2014). “deep convolutional networks large-scale image recognition.” arXiv preprint arXiv:1409.1556. optimizer (TorchOptimizer) optimizer use training. Per default, adam used. loss (TorchLoss) loss used train network. Per default, mse used regression cross_entropy classification. callbacks (list() TorchCallbacks) callbacks. Must unique ids.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"","code":"LearnerTorchVision$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners.torchvision.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"AlexNet Image Classifier — mlr_learners.torchvision","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Learners — mlr_learners_torch","title":"Base Class for Torch Learners — mlr_learners_torch","text":"base class provides basic functionality training prediction neural network. torch learners inherit class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"validation","dir":"Reference","previous_headings":"","what":"Validation","title":"Base Class for Torch Learners — mlr_learners_torch","text":"specify validation data, can set $validate field Learner, can set : NULL: validation ratio: proportion 1 - ratio task used training ratio used validation. \"test\" means \"test\" task resampling used possible calling $train() manually. \"predefined\": use predefined $internal_valid_task mlr3::Task. validation data can also used early stopping, see description Learner's parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"saving-a-learner","dir":"Reference","previous_headings":"","what":"Saving a Learner","title":"Base Class for Torch Learners — mlr_learners_torch","text":"order save LearnerTorch later usage, necessary call $marshal() method Learner writing disk, object otherwise saved correctly. loading marshaled LearnerTorch R , need call $unmarshal() transform useable state.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"early-stopping-and-internal-tuning","dir":"Reference","previous_headings":"","what":"Early Stopping and Internal Tuning","title":"Base Class for Torch Learners — mlr_learners_torch","text":"order prevent overfitting, LearnerTorch class allows use early stopping via patience min_delta parameters, see Learner's parameters. tuning LearnerTorch also possible combine explicit tuning via mlr3tuning LearnerTorch's internal tuning epochs via early stopping. , just need include epochs = to_tune(upper = <upper>, internal = TRUE) search space, <upper> maximally allowed number epochs, configure early stopping.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"model","dir":"Reference","previous_headings":"","what":"Model","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Model list class \"learner_torch_model\" following elements: network :: trained network. optimizer :: $state_dict() optimizer used train network. loss_fn :: $state_dict() loss used train network. callbacks :: callbacks used train network. seed :: seed / used training prediction. epochs :: many epochs model trained (early stopping). task_col_info :: data.table() containing information train-task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Learners — mlr_learners_torch","text":"General: parameters optimizer, loss callbacks, prefixed \"opt.\", \"loss.\" \"cb.<callback id>.\" respectively, well : epochs :: integer(1) number epochs. device :: character(1) device. One \"auto\", \"cpu\", \"cuda\" values defined mlr_reflections$torch$devices. value initialized \"auto\", select \"cuda\" possible, try \"mps\" otherwise fall back \"cpu\". num_threads :: integer(1) number threads intraop pararallelization (device \"cpu\"). value initialized 1. num_interop_threads :: integer(1) number threads intraop interop pararallelization (device \"cpu\"). value initialized 1. Note can set session changing value within R session raise warning. seed :: integer(1) \"random\" NULL torch seed used training prediction. value initialized \"random\", means random seed sampled beginning training phase. seed (either set randomly sampled) available via $model$seed training used prediction. Note setting seed training phase mean default (.e. seed \"random\"), clones learner use different seed. set NULL, seeding done. Evaluation: measures_train :: Measure list() Measures. Measures evaluated training. measures_valid :: Measure list() Measures. Measures evaluated validation. eval_freq :: integer(1) often train / validation predictions evaluated using measures_train / measures_valid. initialized 1. Note final model always evaluated. Early Stopping: patience :: integer(1) activates early stopping using validation scores. performance model improve patience evaluation steps, training ended. Note final model stored learner, best model. initialized 0, means early stopping. first entry measures_valid used metric. also requires specify $validate field Learner, well measures_valid. min_delta :: double(1) minimum improvement threshold (>) early stopping. initialized 0. Dataloader: batch_size :: integer(1) batch size (required). shuffle :: logical(1) Whether shuffle instances dataset. Default FALSE. impact validation. sampler :: torch::sampler Object defines dataloader draw samples. batch_sampler :: torch::sampler Object defines dataloader draws batches. num_workers :: integer(1) number workers data loading (batches loaded parallel). default 0, means data loaded main process. collate_fn :: function merge list samples form batch. pin_memory :: logical(1) Whether dataloader copies tensors CUDA pinned memory returning . drop_last :: logical(1) Whether drop last training batch epoch training. Default FALSE. timeout :: numeric(1) timeout value collecting batch workers. Negative values mean timeout default -1. worker_init_fn :: function(id) function receives worker id ([1, num_workers]) exectued seeding worker data loading. worker_globals :: list() | character() loading data parallel, allows export globals workers. character vector, objects global environment names copied workers. worker_packages :: character() packages load workers. Also see torch::dataloder information.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Torch Learners — mlr_learners_torch","text":"seperate classes classification regression inherit . Instead, task_type must specified  construction argument. Currently, classification regression supported. inheriting class, one overload two private methods: .network(task, param_vals) (Task, list()) -> nn_module Construct torch::nn_module object given task parameter values, .e. neural network trained learner. classification, output network expected scores application final softmax layer. .dataset(task, param_vals) (Task, list()) -> torch::dataset Create dataset task. Must respect parameter value device. Moreover, one needs pay attention respect row ids provided task. also possible overwrite private .dataloader() method instead .dataset() method. Per default, dataloader constructed using output .dataset() method. However, respect dataloader parameters ParamSet. .dataloader(task, param_vals) (Task, list()) -> torch::dataloader Create dataloader task. Needs respect least batch_size shuffle (otherwise predictions can permuted). change predict types, private .encode_prediction() method can overwritten: .encode_prediction(predict_tensor, task, param_vals) (torch_tensor, Task, list()) -> list() Take raw predictions self$network (predict_tensor) encode format can converted valid mlr3 predictions using mlr3::as_prediction_data(). method must take self$predict_type account. possible add parameters specifying param_set construction argument, currently possible remove existing parameters, .e. listed section Parameters. None parameters provided param_set can id starts \"loss.\", \"opt.\", \"cb.\"`, preserved dynamically constructed parameters optimizer, loss function, callbacks. perform additional input checks task, private .verify_train_task(task, param_vals) .verify_predict_task(task, param_vals) can overwritten. learners construction arguments change hash learner, required implement private $.additional_phash_input().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Base Class for Torch Learners — mlr_learners_torch","text":"mlr3::Learner -> LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Torch Learners — mlr_learners_torch","text":"validate construct internal validation data. parameter can either NULL, ratio $(0, 1)$, \"test\", \"predefined\". loss (TorchLoss) torch loss. optimizer (TorchOptimizer) torch optimizer. callbacks (list() TorchCallbacks) List torch callbacks. ids set names. internal_valid_scores Retrieves internal validation scores named list(). Specify $validate field measures_valid parameter configure . Returns NULL learner trained yet. internal_tuned_values early stopping activate, returns named list early-stopped epochs, otherwise empty list returned. Returns NULL learner trained yet. marshaled (logical(1)) Whether learner marshaled. network (nn_module()) Shortcut learner$model$network. param_set (ParamSet) parameter set hash (character(1)) Hash (unique identifier) object. phash (character(1)) Hash (unique identifier) partial object, excluding components varied systematically tuning (parameter values).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Base Class for Torch Learners — mlr_learners_torch","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Torch Learners — mlr_learners_torch","text":"LearnerTorch$new() LearnerTorch$format() LearnerTorch$print() LearnerTorch$marshal() LearnerTorch$unmarshal() LearnerTorch$dataset() LearnerTorch$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$new(   id,   task_type,   param_set,   properties,   man,   label,   feature_types,   optimizer = NULL,   loss = NULL,   packages = character(),   predict_types = NULL,   callbacks = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"id (character(1)) id new object. task_type (character(1)) task type. param_set (ParamSet alist()) Either parameter set, alist() containing different values self, e.g. alist(private$.param_set1, private$.param_set2), ParamSet collection created. properties (character()) properties object. See mlr_reflections$learner_properties available values. man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). label (character(1)) Label new instance. feature_types (character()) feature types. See mlr_reflections$task_feature_types available values, Additionally, \"lazy_tensor\" supported. optimizer (NULL TorchOptimizer) optimizer use training. Defaults adam. loss (NULL TorchLoss) loss use training. Defaults MSE regression cross entropy classification. packages (character()) R packages object depends . predict_types (character()) predict types. See mlr_reflections$learner_predict_types available values. regression, default \"response\". classification, defaults \"response\" \"prob\". deviate defaults, necessary overwrite private $.encode_prediction() method, see section Inheriting. callbacks (list() TorchCallbacks) callbacks use training. Defaults empty list(), .e. callbacks.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-format-","dir":"Reference","previous_headings":"","what":"Method format()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Helper print outputs.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$format(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... (ignored).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-print-","dir":"Reference","previous_headings":"","what":"Method print()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Prints object.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$print(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Currently unused.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-marshal-","dir":"Reference","previous_headings":"","what":"Method marshal()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Marshal learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-3","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$marshal(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-3","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Additional parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"self","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-unmarshal-","dir":"Reference","previous_headings":"","what":"Method unmarshal()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Unmarshal learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-4","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$unmarshal(...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-4","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"... () Additional parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns-1","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"self","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-dataset-","dir":"Reference","previous_headings":"","what":"Method dataset()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"Create dataset task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-5","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$dataset(task)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-5","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"task Task task","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"returns-2","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Learners — mlr_learners_torch","text":"dataset","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Torch Learners — mlr_learners_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"usage-6","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Learners — mlr_learners_torch","text":"","code":"LearnerTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch.html","id":"arguments-6","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Learners — mlr_learners_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":null,"dir":"Reference","previous_headings":"","what":"Image Learner — mlr_learners_torch_image","title":"Image Learner — mlr_learners_torch_image","text":"Base Class Image Learners. features assumed single lazy_tensor column RGB format.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Image Learner — mlr_learners_torch_image","text":"Parameters include inherited LearnerTorch param_set construction argument.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Image Learner — mlr_learners_torch_image","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchImage","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Image Learner — mlr_learners_torch_image","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Image Learner — mlr_learners_torch_image","text":"LearnerTorchImage$new() LearnerTorchImage$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Image Learner — mlr_learners_torch_image","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Image Learner — mlr_learners_torch_image","text":"","code":"LearnerTorchImage$new(   id,   task_type,   param_set = ps(),   label,   optimizer = NULL,   loss = NULL,   callbacks = list(),   packages = \"torchvision\",   man,   properties = NULL,   predict_types = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Image Learner — mlr_learners_torch_image","text":"id (character(1)) id new object. task_type (character(1)) task type. param_set (ParamSet) parameter set. label (character(1)) Label new instance. optimizer (TorchOptimizer) torch optimizer. loss (TorchLoss) loss use training. callbacks (list() TorchCallbacks) callbacks used training. Must unique ids. executed order provided packages (character()) R packages object depends . man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). properties (character()) properties object. See mlr_reflections$learner_properties available values. predict_types (character()) predict types. See mlr_reflections$learner_predict_types available values.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Image Learner — mlr_learners_torch_image","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Image Learner — mlr_learners_torch_image","text":"","code":"LearnerTorchImage$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_image.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Image Learner — mlr_learners_torch_image","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Learner Torch Model — mlr_learners_torch_model","title":"Learner Torch Model — mlr_learners_torch_model","text":"Create torch learner instantiated nn_module(). classification, output network must scores (softmax).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Learner Torch Model — mlr_learners_torch_model","text":"See LearnerTorch","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Learner Torch Model — mlr_learners_torch_model","text":"mlr3::Learner -> mlr3torch::LearnerTorch -> LearnerTorchModel","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Learner Torch Model — mlr_learners_torch_model","text":"network_stored (nn_module NULL) network trained. calling $train(), NULL. ingress_tokens (named list() TorchIngressToken NULL) ingress tokens. Must non-NULL calling $train().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Learner Torch Model — mlr_learners_torch_model","text":"mlr3::Learner$base_learner() mlr3::Learner$encapsulate() mlr3::Learner$help() mlr3::Learner$predict() mlr3::Learner$predict_newdata() mlr3::Learner$reset() mlr3::Learner$train() mlr3torch::LearnerTorch$dataset() mlr3torch::LearnerTorch$format() mlr3torch::LearnerTorch$marshal() mlr3torch::LearnerTorch$print() mlr3torch::LearnerTorch$unmarshal()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Learner Torch Model — mlr_learners_torch_model","text":"LearnerTorchModel$new() LearnerTorchModel$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Learner Torch Model — mlr_learners_torch_model","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"LearnerTorchModel$new(   network = NULL,   ingress_tokens = NULL,   task_type,   properties = NULL,   optimizer = NULL,   loss = NULL,   callbacks = list(),   packages = character(0),   feature_types = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learner Torch Model — mlr_learners_torch_model","text":"network (nn_module) instantiated nn_module. cloned construction. classification, outputs must scores (softmax). ingress_tokens (list TorchIngressToken()) list ingress tokens defines dataloader defined. task_type (character(1)) task type. properties (NULL character()) properties learner. Defaults available properties given task type. optimizer (TorchOptimizer) torch optimizer. loss (TorchLoss) loss use training. callbacks (list() TorchCallbacks) callbacks used training. Must unique ids. executed order provided packages (character()) R packages object depends . feature_types (NULL character()) feature types. Defaults available feature types.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Learner Torch Model — mlr_learners_torch_model","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"LearnerTorchModel$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learner Torch Model — mlr_learners_torch_model","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_learners_torch_model.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Learner Torch Model — mlr_learners_torch_model","text":"","code":"# We show the learner using a classification task  # The iris task has 4 features and 3 classes network = nn_linear(4, 3) task = tsk(\"iris\")  # This defines the dataloader. # It loads all 4 features, which are also numeric. # The shape is (NA, 4) because the batch dimension is generally NA ingress_tokens = list(   input = TorchIngressToken(task$feature_names, batchgetter_num, c(NA, 4)) )  # Creating the learner and setting required parameters learner = lrn(\"classif.torch_model\",   network = network,   ingress_tokens = ingress_tokens,   batch_size = 16,   epochs = 1,   device = \"cpu\" )  # A simple train-predict ids = partition(task) learner$train(task, ids$train) learner$predict(task, ids$test) #> <PredictionClassif> for 50 observations: #>  row_ids     truth response #>        3    setosa   setosa #>        9    setosa   setosa #>       10    setosa   setosa #>      ---       ---      --- #>      147 virginica   setosa #>      149 virginica   setosa #>      150 virginica   setosa"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":null,"dir":"Reference","previous_headings":"","what":"Class for Torch Module Wrappers — mlr_pipeops_module","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"PipeOpModule wraps nn_module function called train phase mlr3pipelines::PipeOp. , allows assemble PipeOpModules computational mlr3pipelines::Graph represents either neural network preprocessing graph lazy_tensor. cases easier create network creating graph generates graph. cases easier create network creating structurally related graph consisting nodes class PipeOpTorchIngress PipeOpTorch. graph generate graph consisting PipeOpModules part ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"number names input output channels can set construction. input output \"torch_tensor\" training, NULL prediction prediction phase currently serves meaningful purpose.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"training, wrapped nn_module / function called provided inputs order channels defined. Arguments matched name.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"mlr3pipelines::PipeOp -> PipeOpModule","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"module (nn_module) torch module called training phase.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"PipeOpModule$new() PipeOpModule$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"","code":"PipeOpModule$new(   id = \"module\",   module = nn_identity(),   inname = \"input\",   outname = \"output\",   param_vals = list(),   packages = character(0) )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"id (character(1)) id new object. module (nn_module function()) torch module function wrapped. inname (character()) names input channels. outname (character()) names output channels. parameter length 1, parameter module must return tensor. Otherwise must return list() tensors corresponding length. param_vals (named list()) Parameter values set construction. packages (character()) R packages object depends .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"","code":"PipeOpModule$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_module.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Class for Torch Module Wrappers — mlr_pipeops_module","text":"","code":"## creating an PipeOpModule manually  # one input and output channel po_module = po(\"module\",   id = \"linear\",   module = torch::nn_linear(10, 20),   inname = \"input\",   outname = \"output\" ) x = torch::torch_randn(16, 10) # This calls the forward function of the wrapped module. y = po_module$train(list(input = x)) str(y) #> List of 1 #>  $ output:Float [1:16, 1:20]  # multiple input and output channels nn_custom = torch::nn_module(\"nn_custom\",   initialize = function(in_features, out_features) {     self$lin1 = torch::nn_linear(in_features, out_features)     self$lin2 = torch::nn_linear(in_features, out_features)   },   forward = function(x, z) {     list(out1 = self$lin1(x), out2 = torch::nnf_relu(self$lin2(z)))   } )  module = nn_custom(3, 2) po_module = po(\"module\",   id = \"custom\",   module = module,   inname = c(\"x\", \"z\"),   outname = c(\"out1\", \"out2\") ) x = torch::torch_randn(1, 3) z = torch::torch_randn(1, 3) out = po_module$train(list(x = x, z = z)) str(out) #> List of 2 #>  $ out1:Float [1:1, 1:2] #>  $ out2:Float [1:1, 1:2]  # How such a PipeOpModule is usually generated graph = po(\"torch_ingress_num\") %>>% po(\"nn_linear\", out_features = 10L) result = graph$train(tsk(\"iris\")) # The PipeOpTorchLinear generates a PipeOpModule and adds it to a new (module) graph result[[1]]$graph #> Graph with 2 PipeOps: #>                 ID         State  sccssors         prdcssors #>             <char>        <char>    <char>            <char> #>  torch_ingress_num <<UNTRAINED>> nn_linear                   #>          nn_linear <<UNTRAINED>>           torch_ingress_num linear_module = result[[1L]]$graph$pipeops$nn_linear linear_module #> PipeOp: <nn_linear> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [torch_tensor,NULL] #> Output channels <name [train type, predict type]>: #>   output [torch_tensor,NULL] formalArgs(linear_module$module) #> [1] \"input\" linear_module$input$name #> [1] \"input\"  # Constructing a PipeOpModule using a simple function po_add1 = po(\"module\",   id = \"add_one\",   module = function(x) x + 1 ) input = list(torch_tensor(1)) po_add1$train(input)$output #> torch_tensor #>  2 #> [ CPUFloatType{1} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Applies 1D adaptive average pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"output_size :: integer(1) target output size. single number.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Calls nn_adaptive_avg_pool1d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"PipeOpTorchAdaptiveAvgPool1D$new() PipeOpTorchAdaptiveAvgPool1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"","code":"PipeOpTorchAdaptiveAvgPool1D$new(   id = \"nn_adaptive_avg_pool1d\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"","code":"PipeOpTorchAdaptiveAvgPool1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_adaptive_avg_pool1d\") pipeop #> PipeOp: <nn_adaptive_avg_pool1d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: output_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Applies 2D adaptive average pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"output_size :: integer() target output size. Can single number vector.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Calls nn_adaptive_avg_pool2d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"PipeOpTorchAdaptiveAvgPool2D$new() PipeOpTorchAdaptiveAvgPool2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"","code":"PipeOpTorchAdaptiveAvgPool2D$new(   id = \"nn_adaptive_avg_pool2d\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"","code":"PipeOpTorchAdaptiveAvgPool2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_adaptive_avg_pool2d\") pipeop #> PipeOp: <nn_adaptive_avg_pool2d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: output_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Applies 3D adaptive average pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"output_size :: integer() target output size. Can single number vector.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Calls nn_adaptive_avg_pool3d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"PipeOpTorchAdaptiveAvgPool3D$new() PipeOpTorchAdaptiveAvgPool3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"","code":"PipeOpTorchAdaptiveAvgPool3D$new(   id = \"nn_adaptive_avg_pool3d\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"","code":"PipeOpTorchAdaptiveAvgPool3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_adaptive_avg_pool3d\") pipeop #> PipeOp: <nn_adaptive_avg_pool3d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: output_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Applies 1D average pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Calls nn_avg_pool1d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"PipeOpTorchAvgPool1D$new() PipeOpTorchAvgPool1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"","code":"PipeOpTorchAvgPool1D$new(id = \"nn_avg_pool1d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"","code":"PipeOpTorchAvgPool1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Average Pooling — mlr_pipeops_nn_avg_pool1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_avg_pool1d\") pipeop #> PipeOp: <nn_avg_pool1d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(5)> #>                   id    class lower upper nlevels        default  value #>               <char>   <char> <num> <num>   <num>         <list> <list> #> 1:       kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:            stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 3:           padding ParamUty    NA    NA     Inf              0 [NULL] #> 4:         ceil_mode ParamLgl    NA    NA       2          FALSE [NULL] #> 5: count_include_pad ParamLgl    NA    NA       2           TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"Applies 2D average-pooling operation \\(kH * kW\\) regions step size \\(sH * sW\\) steps. number output features equal number input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"Calls nn_avg_pool2d() training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"PipeOpTorchAvgPool2D$new() PipeOpTorchAvgPool2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"","code":"PipeOpTorchAvgPool2D$new(id = \"nn_avg_pool2d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"","code":"PipeOpTorchAvgPool2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Average Pooling — mlr_pipeops_nn_avg_pool2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_avg_pool2d\") pipeop #> PipeOp: <nn_avg_pool2d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(6)> #>                   id    class lower upper nlevels        default  value #>               <char>   <char> <num> <num>   <num>         <list> <list> #> 1:       kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:            stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 3:           padding ParamUty    NA    NA     Inf              0 [NULL] #> 4:         ceil_mode ParamLgl    NA    NA       2          FALSE [NULL] #> 5: count_include_pad ParamLgl    NA    NA       2           TRUE [NULL] #> 6:  divisor_override ParamDbl     0   Inf     Inf         [NULL] [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"Applies 3D average-pooling operation \\(kT * kH * kW\\) regions step size \\(sT * sH * sW\\) steps. number output features equal \\(\\lfloor \\frac{ \\mbox{input planes} }{sT} \\rfloor\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"Calls nn_avg_pool3d() training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"kernel_size :: (integer()) size window. Can single number vector. stride :: integer() stride window. Can single number vector. Default: kernel_size. padding :: integer() Implicit zero paddings sides input. Can single number vector. Default: 0. ceil_mode :: integer() TRUE, use ceil instead floor compute output shape. Default: FALSE. count_include_pad :: logical(1) TRUE, include zero-padding averaging calculation. Default: TRUE. divisor_override :: logical(1) specified, used divisor, otherwise size pooling region used. Default: NULL. available dimension greater 1.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAvgPool -> PipeOpTorchAvgPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"PipeOpTorchAvgPool3D$new() PipeOpTorchAvgPool3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"","code":"PipeOpTorchAvgPool3D$new(id = \"nn_avg_pool3d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"","code":"PipeOpTorchAvgPool3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_avg_pool3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Average Pooling — mlr_pipeops_nn_avg_pool3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_avg_pool3d\") pipeop #> PipeOp: <nn_avg_pool3d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(6)> #>                   id    class lower upper nlevels        default  value #>               <char>   <char> <num> <num>   <num>         <list> <list> #> 1:       kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:            stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 3:           padding ParamUty    NA    NA     Inf              0 [NULL] #> 4:         ceil_mode ParamLgl    NA    NA       2          FALSE [NULL] #> 5: count_include_pad ParamLgl    NA    NA       2           TRUE [NULL] #> 6:  divisor_override ParamDbl     0   Inf     Inf         [NULL] [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"Applies Batch Normalization channel across batch data.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"Calls torch::nn_batch_norm1d(). parameter num_features inferred second dimension input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"PipeOpTorchBatchNorm1D$new() PipeOpTorchBatchNorm1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"","code":"PipeOpTorchBatchNorm1D$new(id = \"nn_batch_norm1d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"","code":"PipeOpTorchBatchNorm1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Batch Normalization — mlr_pipeops_nn_batch_norm1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_batch_norm1d\") pipeop #> PipeOp: <nn_batch_norm1d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(4)> #>                     id    class lower upper nlevels default  value #>                 <char>   <char> <num> <num>   <num>  <list> <list> #> 1:                 eps ParamDbl     0   Inf     Inf   1e-05 [NULL] #> 2:            momentum ParamDbl     0   Inf     Inf     0.1 [NULL] #> 3:              affine ParamLgl    NA    NA       2    TRUE [NULL] #> 4: track_running_stats ParamLgl    NA    NA       2    TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"Applies Batch Normalization channel across batch data.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"Calls torch::nn_batch_norm2d(). parameter num_features inferred second dimension input shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"PipeOpTorchBatchNorm2D$new() PipeOpTorchBatchNorm2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"","code":"PipeOpTorchBatchNorm2D$new(id = \"nn_batch_norm2d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"","code":"PipeOpTorchBatchNorm2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Batch Normalization — mlr_pipeops_nn_batch_norm2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_batch_norm2d\") pipeop #> PipeOp: <nn_batch_norm2d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(4)> #>                     id    class lower upper nlevels default  value #>                 <char>   <char> <num> <num>   <num>  <list> <list> #> 1:                 eps ParamDbl     0   Inf     Inf   1e-05 [NULL] #> 2:            momentum ParamDbl     0   Inf     Inf     0.1 [NULL] #> 3:              affine ParamLgl    NA    NA       2    TRUE [NULL] #> 4: track_running_stats ParamLgl    NA    NA       2    TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"Applies Batch Normalization channel across batch data.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"Calls torch::nn_batch_norm3d(). parameter num_features inferred second dimension input shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"eps :: numeric(1) value added denominator numerical stability. Default: 1e-5. momentum :: numeric(1) value used running_mean running_var computation. Can set NULL cumulative moving average (.e. simple average). Default: 0.1 affine :: logical(1) boolean value set TRUE, module learnable affine parameters. Default: TRUE track_running_stats :: logical(1) boolean value set TRUE, module tracks running mean variance, set FALSE, module track statistics always uses batch statistics training eval modes. Default: TRUE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchBatchNorm -> PipeOpTorchBatchNorm3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"PipeOpTorchBatchNorm3D$new() PipeOpTorchBatchNorm3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"","code":"PipeOpTorchBatchNorm3D$new(id = \"nn_batch_norm3d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"","code":"PipeOpTorchBatchNorm3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_batch_norm3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Batch Normalization — mlr_pipeops_nn_batch_norm3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_batch_norm3d\") pipeop #> PipeOp: <nn_batch_norm3d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(4)> #>                     id    class lower upper nlevels default  value #>                 <char>   <char> <num> <num>   <num>  <list> <list> #> 1:                 eps ParamDbl     0   Inf     Inf   1e-05 [NULL] #> 2:            momentum ParamDbl     0   Inf     Inf     0.1 [NULL] #> 3:              affine ParamLgl    NA    NA       2    TRUE [NULL] #> 4: track_running_stats ParamLgl    NA    NA       2    TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":null,"dir":"Reference","previous_headings":"","what":"Block Repetition — mlr_pipeops_nn_block","title":"Block Repetition — mlr_pipeops_nn_block","text":"Repeat block n_blocks times.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Block Repetition — mlr_pipeops_nn_block","text":"parameters available block , well n_blocks :: integer(1) often repeat block.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Block Repetition — mlr_pipeops_nn_block","text":"PipeOp sets input output channels block (Graph) received construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Block Repetition — mlr_pipeops_nn_block","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Block Repetition — mlr_pipeops_nn_block","text":"Part documentation copied adapted documentation torch.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Block Repetition — mlr_pipeops_nn_block","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchBlock","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Block Repetition — mlr_pipeops_nn_block","text":"block (Graph) neural network segment repeated PipeOp.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Block Repetition — mlr_pipeops_nn_block","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Block Repetition — mlr_pipeops_nn_block","text":"PipeOpTorchBlock$new() PipeOpTorchBlock$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Block Repetition — mlr_pipeops_nn_block","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Block Repetition — mlr_pipeops_nn_block","text":"","code":"PipeOpTorchBlock$new(block, id = \"nn_block\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Block Repetition — mlr_pipeops_nn_block","text":"block (Graph) graph consisting primarily PipeOpTorch objects repeated. id (character(1)) id new object. param_vals (named list()) Parameter values set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Block Repetition — mlr_pipeops_nn_block","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Block Repetition — mlr_pipeops_nn_block","text":"","code":"PipeOpTorchBlock$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Block Repetition — mlr_pipeops_nn_block","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_block.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Block Repetition — mlr_pipeops_nn_block","text":"","code":"block = po(\"nn_linear\") %>>% po(\"nn_relu\") po_block = po(\"nn_block\", block, nn_linear.out_features = 10L, n_blocks = 3) network = po(\"torch_ingress_num\") %>>% po_block %>>% po(\"nn_head\") %>>% po(\"torch_loss\", t_loss(\"cross_entropy\")) %>>% po(\"torch_optimizer\", t_opt(\"adam\")) %>>% po(\"torch_model_classif\",   batch_size = 50,   epochs = 3)  task = tsk(\"iris\") network$train(task) #> $torch_model_classif.output #> NULL #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"CELU Activation Function — mlr_pipeops_nn_celu","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Applies element-wise, \\(CELU(x) = max(0,x) + min(0, \\alpha * (exp(x \\alpha) - 1))\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"alpha :: numeric(1) alpha value ELU formulation. Default: 1.0 inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Calls torch::nn_celu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchCELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"PipeOpTorchCELU$new() PipeOpTorchCELU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"","code":"PipeOpTorchCELU$new(id = \"nn_celu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"","code":"PipeOpTorchCELU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_celu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"CELU Activation Function — mlr_pipeops_nn_celu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_celu\") pipeop #> PipeOp: <nn_celu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1:   alpha ParamDbl  -Inf   Inf     Inf       1 [NULL] #> 2: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Convolution — mlr_pipeops_nn_conv1d","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Applies 1D convolution input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Calls torch::nn_conv1d() trained. paramter in_channels inferred second dimension input tensor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"PipeOpTorchConv1D$new() PipeOpTorchConv1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"","code":"PipeOpTorchConv1D$new(id = \"nn_conv1d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"","code":"PipeOpTorchConv1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Convolution — mlr_pipeops_nn_conv1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv1d\", kernel_size = 10, out_channels = 1) pipeop #> PipeOp: <nn_conv1d> (not trained) #> values: <out_channels=1, kernel_size=10> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(8)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1: out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      1 #> 2:  kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>     10 #> 3:       stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:      padding ParamUty    NA    NA     Inf              0 [NULL] #> 5:     dilation ParamUty    NA    NA     Inf              1 [NULL] #> 6:       groups ParamInt     1   Inf     Inf              1 [NULL] #> 7:         bias ParamLgl    NA    NA       2           TRUE [NULL] #> 8: padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Convolution — mlr_pipeops_nn_conv2d","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"Applies 2D convolution input image composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"Calls torch::nn_conv2d() trained. paramter in_channels inferred second dimension input tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"PipeOpTorchConv2D$new() PipeOpTorchConv2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"","code":"PipeOpTorchConv2D$new(id = \"nn_conv2d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"","code":"PipeOpTorchConv2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Convolution — mlr_pipeops_nn_conv2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv2d\", kernel_size = 10, out_channels = 1) pipeop #> PipeOp: <nn_conv2d> (not trained) #> values: <out_channels=1, kernel_size=10> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(8)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1: out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      1 #> 2:  kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>     10 #> 3:       stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:      padding ParamUty    NA    NA     Inf              0 [NULL] #> 5:     dilation ParamUty    NA    NA     Inf              1 [NULL] #> 6:       groups ParamInt     1   Inf     Inf              1 [NULL] #> 7:         bias ParamLgl    NA    NA       2           TRUE [NULL] #> 8: padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Convolution — mlr_pipeops_nn_conv3d","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"Applies 3D convolution input image composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"Calls torch::nn_conv3d() trained. paramter in_channels inferred second dimension input tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"out_channels :: integer(1) Number channels produced convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. default 1. padding :: integer() ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘TRUE’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConv -> PipeOpTorchConv3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"PipeOpTorchConv3D$new() PipeOpTorchConv3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"","code":"PipeOpTorchConv3D$new(id = \"nn_conv3d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"","code":"PipeOpTorchConv3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Convolution — mlr_pipeops_nn_conv3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv3d\", kernel_size = 10, out_channels = 1) pipeop #> PipeOp: <nn_conv3d> (not trained) #> values: <out_channels=1, kernel_size=10> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(8)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1: out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      1 #> 2:  kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>     10 #> 3:       stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:      padding ParamUty    NA    NA     Inf              0 [NULL] #> 5:     dilation ParamUty    NA    NA     Inf              1 [NULL] #> 6:       groups ParamInt     1   Inf     Inf              1 [NULL] #> 7:         bias ParamLgl    NA    NA       2           TRUE [NULL] #> 8: padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Transpose 1D Convolution Transpose 1D Convolution","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Calls nn_conv_transpose1d. parameter in_channels inferred second dimension input tensor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"PipeOpTorchConvTranspose1D$new() PipeOpTorchConvTranspose1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"","code":"PipeOpTorchConvTranspose1D$new(id = \"nn_conv_transpose1d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"","code":"PipeOpTorchConvTranspose1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Transpose 1D Convolution — mlr_pipeops_nn_conv_transpose1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv_transpose1d\", kernel_size = 3, out_channels = 2) pipeop #> PipeOp: <nn_conv_transpose1d> (not trained) #> values: <out_channels=2, kernel_size=3> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(9)> #>                id    class lower upper nlevels        default  value #>            <char>   <char> <num> <num>   <num>         <list> <list> #> 1:   out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      2 #> 2:    kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>      3 #> 3:         stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:        padding ParamUty    NA    NA     Inf              0 [NULL] #> 5: output_padding ParamUty    NA    NA     Inf              0 [NULL] #> 6:       dilation ParamUty    NA    NA     Inf              1 [NULL] #> 7:         groups ParamInt     1   Inf     Inf              1 [NULL] #> 8:           bias ParamLgl    NA    NA       2           TRUE [NULL] #> 9:   padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"Applies 2D transposed convolution operator input image composed several input planes, sometimes also called \"deconvolution\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"Calls nn_conv_transpose2d. parameter in_channels inferred second dimension input tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"PipeOpTorchConvTranspose2D$new() PipeOpTorchConvTranspose2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"","code":"PipeOpTorchConvTranspose2D$new(id = \"nn_conv_transpose2d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"","code":"PipeOpTorchConvTranspose2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Transpose 2D Convolution — mlr_pipeops_nn_conv_transpose2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv_transpose2d\", kernel_size = 3, out_channels = 2) pipeop #> PipeOp: <nn_conv_transpose2d> (not trained) #> values: <out_channels=2, kernel_size=3> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(9)> #>                id    class lower upper nlevels        default  value #>            <char>   <char> <num> <num>   <num>         <list> <list> #> 1:   out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      2 #> 2:    kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>      3 #> 3:         stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:        padding ParamUty    NA    NA     Inf              0 [NULL] #> 5: output_padding ParamUty    NA    NA     Inf              0 [NULL] #> 6:       dilation ParamUty    NA    NA     Inf              1 [NULL] #> 7:         groups ParamInt     1   Inf     Inf              1 [NULL] #> 8:           bias ParamLgl    NA    NA       2           TRUE [NULL] #> 9:   padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"Applies 3D transposed convolution operator input image composed several input planes, sometimes also called \"deconvolution\"","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"Calls nn_conv_transpose3d. parameter in_channels inferred second dimension input tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"out_channels :: integer(1) Number output channels produce convolution. kernel_size :: integer() Size convolving kernel. stride :: integer() Stride convolution. Default: 1. padding ::  integer()` ‘dilation * (kernel_size - 1) - padding’ zero-padding added sides input. Default: 0. output_padding ::integer() Additional size added one side output shape. Default: 0. groups :: integer() Number blocked connections input channels output channels. Default: 1 bias :: logical(1) ‘True’, adds learnable bias output. Default: ‘TRUE’. dilation :: integer() Spacing kernel elements. Default: 1. padding_mode :: character(1) padding mode. One \"zeros\", \"reflect\", \"replicate\", \"circular\". Default \"zeros\".","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchConvTranspose -> PipeOpTorchConvTranspose3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"PipeOpTorchConvTranspose3D$new() PipeOpTorchConvTranspose3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"","code":"PipeOpTorchConvTranspose3D$new(id = \"nn_conv_transpose3d\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"","code":"PipeOpTorchConvTranspose3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_conv_transpose3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Transpose 3D Convolution — mlr_pipeops_nn_conv_transpose3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_conv_transpose3d\", kernel_size = 3, out_channels = 2) pipeop #> PipeOp: <nn_conv_transpose3d> (not trained) #> values: <out_channels=2, kernel_size=3> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(9)> #>                id    class lower upper nlevels        default  value #>            <char>   <char> <num> <num>   <num>         <list> <list> #> 1:   out_channels ParamInt     1   Inf     Inf <NoDefault[0]>      2 #> 2:    kernel_size ParamUty    NA    NA     Inf <NoDefault[0]>      3 #> 3:         stride ParamUty    NA    NA     Inf              1 [NULL] #> 4:        padding ParamUty    NA    NA     Inf              0 [NULL] #> 5: output_padding ParamUty    NA    NA     Inf              0 [NULL] #> 6:       dilation ParamUty    NA    NA     Inf              1 [NULL] #> 7:         groups ParamInt     1   Inf     Inf              1 [NULL] #> 8:           bias ParamLgl    NA    NA       2           TRUE [NULL] #> 9:   padding_mode ParamFct    NA    NA       4          zeros [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Dropout — mlr_pipeops_nn_dropout","title":"Dropout — mlr_pipeops_nn_dropout","text":"training, randomly zeroes elements input tensor probability p using samples Bernoulli distribution.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Dropout — mlr_pipeops_nn_dropout","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Dropout — mlr_pipeops_nn_dropout","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Dropout — mlr_pipeops_nn_dropout","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Dropout — mlr_pipeops_nn_dropout","text":"p :: numeric(1) Probability element zeroed. Default: 0.5 inplace inplace :: logical(1) set TRUE, operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Dropout — mlr_pipeops_nn_dropout","text":"Calls torch::nn_dropout() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Dropout — mlr_pipeops_nn_dropout","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchDropout","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Dropout — mlr_pipeops_nn_dropout","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Dropout — mlr_pipeops_nn_dropout","text":"PipeOpTorchDropout$new() PipeOpTorchDropout$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Dropout — mlr_pipeops_nn_dropout","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dropout — mlr_pipeops_nn_dropout","text":"","code":"PipeOpTorchDropout$new(id = \"nn_dropout\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dropout — mlr_pipeops_nn_dropout","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Dropout — mlr_pipeops_nn_dropout","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Dropout — mlr_pipeops_nn_dropout","text":"","code":"PipeOpTorchDropout$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dropout — mlr_pipeops_nn_dropout","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_dropout.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Dropout — mlr_pipeops_nn_dropout","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_dropout\") pipeop #> PipeOp: <nn_dropout> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1:       p ParamDbl     0     1     Inf     0.5 [NULL] #> 2: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"ELU Activation Function — mlr_pipeops_nn_elu","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Applies element-wise, $$ELU(x) = max(0,x) + min(0, \\alpha * (exp(x) - 1))$$.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"alpha :: numeric(1) alpha value ELU formulation. Default: 1.0 inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Calls torch::nn_elu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"PipeOpTorchELU$new() PipeOpTorchELU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"","code":"PipeOpTorchELU$new(id = \"nn_elu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"","code":"PipeOpTorchELU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_elu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"ELU Activation Function — mlr_pipeops_nn_elu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_elu\") pipeop #> PipeOp: <nn_elu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1:   alpha ParamDbl  -Inf   Inf     Inf       1 [NULL] #> 2: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":null,"dir":"Reference","previous_headings":"","what":"Flattens a Tensor — mlr_pipeops_nn_flatten","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"use nn_sequential.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"start_dim :: integer(1) wich dimension start flattening. Default 2. end_dim :: integer(1) wich dimension stop flattening. Default -1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"Calls torch::nn_flatten() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchFlatten","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"PipeOpTorchFlatten$new() PipeOpTorchFlatten$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"","code":"PipeOpTorchFlatten$new(id = \"nn_flatten\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"","code":"PipeOpTorchFlatten$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_flatten.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Flattens a Tensor — mlr_pipeops_nn_flatten","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_flatten\") pipeop #> PipeOp: <nn_flatten> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>           id    class lower upper nlevels default  value #>       <char>   <char> <int> <num>   <num>  <list> <list> #> 1: start_dim ParamInt     1   Inf     Inf       2 [NULL] #> 2:   end_dim ParamInt     1   Inf     Inf      -1 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"GELU Activation Function — mlr_pipeops_nn_gelu","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Gelu","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"approximate :: character(1) Whether use approximation algorithm. Default \"none\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Calls torch::nn_gelu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchGELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"PipeOpTorchGELU$new() PipeOpTorchGELU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"","code":"PipeOpTorchGELU$new(id = \"nn_gelu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"","code":"PipeOpTorchGELU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_gelu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"GELU Activation Function — mlr_pipeops_nn_gelu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_gelu\") pipeop #> PipeOp: <nn_gelu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>             id    class lower upper nlevels default  value #>         <char>   <char> <num> <num>   <num>  <list> <list> #> 1: approximate ParamFct    NA    NA       2    none [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"GLU Activation Function — mlr_pipeops_nn_glu","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"gated linear unit. Computes:","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"dim :: integer(1) Dimension split input. Default: -1","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"Calls torch::nn_glu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchGLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"PipeOpTorchGLU$new() PipeOpTorchGLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"","code":"PipeOpTorchGLU$new(id = \"nn_glu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"","code":"PipeOpTorchGLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_glu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"GLU Activation Function — mlr_pipeops_nn_glu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_glu\") pipeop #> PipeOp: <nn_glu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <int> <num>   <num>  <list> <list> #> 1:    dim ParamInt     1   Inf     Inf      -1 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Applies hard shrinkage function element-wise","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"lambd :: numeric(1) lambda value Hardshrink formulation formulation. Default 0.5.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Calls torch::nn_hardshrink() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"PipeOpTorchHardShrink$new() PipeOpTorchHardShrink$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"","code":"PipeOpTorchHardShrink$new(id = \"nn_hardshrink\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"","code":"PipeOpTorchHardShrink$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardshrink.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Shrink Activation Function — mlr_pipeops_nn_hardshrink","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_hardshrink\") pipeop #> PipeOp: <nn_hardshrink> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <num> <num>   <num>  <list> <list> #> 1:  lambd ParamDbl  -Inf   Inf     Inf     0.5 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Applies element-wise function \\(\\mbox{Hardsigmoid}(x) = \\frac{ReLU6(x + 3)}{6}\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Calls torch::nn_hardsigmoid() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"PipeOpTorchHardSigmoid$new() PipeOpTorchHardSigmoid$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"","code":"PipeOpTorchHardSigmoid$new(id = \"nn_hardsigmoid\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"","code":"PipeOpTorchHardSigmoid$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardsigmoid.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Sigmoid Activation Function — mlr_pipeops_nn_hardsigmoid","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_hardsigmoid\") pipeop #> PipeOp: <nn_hardsigmoid> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Applies HardTanh function element-wise.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"min_val :: numeric(1) Minimum value linear region range. Default: -1. max_val :: numeric(1) Maximum value linear region range. Default: 1. inplace :: logical(1) Can optionally operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Calls torch::nn_hardtanh() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHardTanh","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"PipeOpTorchHardTanh$new() PipeOpTorchHardTanh$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"","code":"PipeOpTorchHardTanh$new(id = \"nn_hardtanh\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"","code":"PipeOpTorchHardTanh$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_hardtanh.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Tanh Activation Function — mlr_pipeops_nn_hardtanh","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_hardtanh\") pipeop #> PipeOp: <nn_hardtanh> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(3)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1: min_val ParamDbl  -Inf   Inf     Inf      -1 [NULL] #> 2: max_val ParamDbl  -Inf   Inf     Inf       1 [NULL] #> 3: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":null,"dir":"Reference","previous_headings":"","what":"Output Head — mlr_pipeops_nn_head","title":"Output Head — mlr_pipeops_nn_head","text":"Output head classification regresssion. NOTE method $shapes_out() access task, returns c(NA, NA). PipeOp trained however, model descriptor correct output shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Output Head — mlr_pipeops_nn_head","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Output Head — mlr_pipeops_nn_head","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Output Head — mlr_pipeops_nn_head","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Output Head — mlr_pipeops_nn_head","text":"bias :: logical(1) Whether use bias. Default TRUE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Output Head — mlr_pipeops_nn_head","text":"Calls torch::nn_linear() input output features inferred input shape / task.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Output Head — mlr_pipeops_nn_head","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchHead","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Output Head — mlr_pipeops_nn_head","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Output Head — mlr_pipeops_nn_head","text":"PipeOpTorchHead$new() PipeOpTorchHead$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Output Head — mlr_pipeops_nn_head","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Output Head — mlr_pipeops_nn_head","text":"","code":"PipeOpTorchHead$new(id = \"nn_head\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Output Head — mlr_pipeops_nn_head","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Output Head — mlr_pipeops_nn_head","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Output Head — mlr_pipeops_nn_head","text":"","code":"PipeOpTorchHead$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Output Head — mlr_pipeops_nn_head","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_head.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Output Head — mlr_pipeops_nn_head","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_head\") pipeop #> PipeOp: <nn_head> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <num> <num>   <num>  <list> <list> #> 1:   bias ParamLgl    NA    NA       2    TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer Normalization — mlr_pipeops_nn_layer_norm","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"Applies Layer Normalization last certain number dimensions.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"dims :: integer(1) number dimensions normalized (starting last dimension). elementwise_affine :: logical(1) Whether learn affine-linear parameters initialized 1 weights 0 biases. default TRUE. eps :: numeric(1) value added denominator numerical stability.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"Calls torch::nn_layer_norm() trained. parameter normalized_shape inferred dimensions last dims dimensions input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLayerNorm","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"PipeOpTorchLayerNorm$new() PipeOpTorchLayerNorm$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"","code":"PipeOpTorchLayerNorm$new(id = \"nn_layer_norm\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"","code":"PipeOpTorchLayerNorm$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_layer_norm.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer Normalization — mlr_pipeops_nn_layer_norm","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_layer_norm\", dims = 1) pipeop #> PipeOp: <nn_layer_norm> (not trained) #> values: <dims=1> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(3)> #>                    id    class lower upper nlevels        default  value #>                <char>   <char> <num> <num>   <num>         <list> <list> #> 1:               dims ParamInt     1   Inf     Inf <NoDefault[0]>      1 #> 2: elementwise_affine ParamLgl    NA    NA       2           TRUE [NULL] #> 3:                eps ParamDbl     0   Inf     Inf          1e-05 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Applies element-wise, \\(LeakyReLU(x) = max(0, x) + negative_slope * min(0, x)\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"negative_slope :: numeric(1) Controls angle negative slope. Default: 1e-2. inplace :: logical(1) Can optionally operation -place. Default: ‘FALSE’.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Calls torch::nn_hardswish() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLeakyReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"PipeOpTorchLeakyReLU$new() PipeOpTorchLeakyReLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"","code":"PipeOpTorchLeakyReLU$new(id = \"nn_leaky_relu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"","code":"PipeOpTorchLeakyReLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_leaky_relu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky ReLU Activation Function — mlr_pipeops_nn_leaky_relu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_leaky_relu\") pipeop #> PipeOp: <nn_leaky_relu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>                id    class lower upper nlevels default  value #>            <char>   <char> <num> <num>   <num>  <list> <list> #> 1: negative_slope ParamDbl  -Inf   Inf     Inf    0.01 [NULL] #> 2:        inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":null,"dir":"Reference","previous_headings":"","what":"Linear Layer — mlr_pipeops_nn_linear","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Applies linear transformation incoming data: \\(y = xA^T + b\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Linear Layer — mlr_pipeops_nn_linear","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Linear Layer — mlr_pipeops_nn_linear","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Linear Layer — mlr_pipeops_nn_linear","text":"out_features :: integer(1) output features linear layer. bias :: logical(1) Whether use bias. Default TRUE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Calls torch::nn_linear() trained parameter in_features inferred second last dimension input tensor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Linear Layer — mlr_pipeops_nn_linear","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLinear","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Linear Layer — mlr_pipeops_nn_linear","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Linear Layer — mlr_pipeops_nn_linear","text":"PipeOpTorchLinear$new() PipeOpTorchLinear$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Linear Layer — mlr_pipeops_nn_linear","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Linear Layer — mlr_pipeops_nn_linear","text":"","code":"PipeOpTorchLinear$new(id = \"nn_linear\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Linear Layer — mlr_pipeops_nn_linear","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Linear Layer — mlr_pipeops_nn_linear","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Linear Layer — mlr_pipeops_nn_linear","text":"","code":"PipeOpTorchLinear$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Linear Layer — mlr_pipeops_nn_linear","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_linear.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Linear Layer — mlr_pipeops_nn_linear","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_linear\", out_features = 10) pipeop #> PipeOp: <nn_linear> (not trained) #> values: <out_features=10> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1: out_features ParamInt     1   Inf     Inf <NoDefault[0]>     10 #> 2:         bias ParamLgl    NA    NA       2           TRUE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Applies element-wise \\(LogSigmoid(x_i) = log(\\frac{1}{1 + exp(-x_i)})\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Calls torch::nn_log_sigmoid() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchLogSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"PipeOpTorchLogSigmoid$new() PipeOpTorchLogSigmoid$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"","code":"PipeOpTorchLogSigmoid$new(id = \"nn_log_sigmoid\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"","code":"PipeOpTorchLogSigmoid$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_log_sigmoid.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Log Sigmoid Activation Function — mlr_pipeops_nn_log_sigmoid","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_log_sigmoid\") pipeop #> PipeOp: <nn_log_sigmoid> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Applies 1D max pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Calls torch::nn_max_pool1d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"PipeOpTorchMaxPool1D$new() PipeOpTorchMaxPool1D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"","code":"PipeOpTorchMaxPool1D$new(   id = \"nn_max_pool1d\",   return_indices = FALSE,   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"id (character(1)) Identifier resulting  object. return_indices (logical(1)) Whether return indices. TRUE, two output channels \"output\" \"indices\". param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"","code":"PipeOpTorchMaxPool1D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool1d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"1D Max Pooling — mlr_pipeops_nn_max_pool1d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_max_pool1d\") pipeop #> PipeOp: <nn_max_pool1d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(5)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:     padding ParamUty    NA    NA     Inf              0 [NULL] #> 3:      stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 4:    dilation ParamInt  -Inf   Inf     Inf              1 [NULL] #> 5:   ceil_mode ParamLgl    NA    NA       2          FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"Applies 2D max pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"Calls torch::nn_max_pool2d() training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool2D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"PipeOpTorchMaxPool2D$new() PipeOpTorchMaxPool2D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"","code":"PipeOpTorchMaxPool2D$new(   id = \"nn_max_pool2d\",   return_indices = FALSE,   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"id (character(1)) Identifier resulting  object. return_indices (logical(1)) Whether return indices. TRUE, two output channels \"output\" \"indices\". param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"","code":"PipeOpTorchMaxPool2D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool2d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"2D Max Pooling — mlr_pipeops_nn_max_pool2d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_max_pool2d\") pipeop #> PipeOp: <nn_max_pool2d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(5)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:     padding ParamUty    NA    NA     Inf              0 [NULL] #> 3:      stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 4:    dilation ParamInt  -Inf   Inf     Inf              1 [NULL] #> 5:   ceil_mode ParamLgl    NA    NA       2          FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"Applies 3D max pooling input signal composed several input planes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"Calls torch::nn_max_pool3d() training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"return_indices FALSE construction, one input channel 'input' one output channel 'output'. return_indices TRUE, two output channels 'output' 'indices'. explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"kernel_size :: integer() size window. Can single number vector. stride :: (integer(1)) stride window. Can single number vector. Default: kernel_size padding :: integer() Implicit zero paddings sides input. Can single number tuple (padW,). Default: 0 dilation :: integer() Controls spacing kernel points; also known à trous algorithm. Default: 1 ceil_mode :: logical(1) True, use ceil instead floor compute output shape. Default: FALSE","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMaxPool -> PipeOpTorchMaxPool3D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"PipeOpTorchMaxPool3D$new() PipeOpTorchMaxPool3D$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"","code":"PipeOpTorchMaxPool3D$new(   id = \"nn_max_pool3d\",   return_indices = FALSE,   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"id (character(1)) Identifier resulting  object. return_indices (logical(1)) Whether return indices. TRUE, two output channels \"output\" \"indices\". param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"","code":"PipeOpTorchMaxPool3D$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_max_pool3d.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"3D Max Pooling — mlr_pipeops_nn_max_pool3d","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_max_pool3d\") pipeop #> PipeOp: <nn_max_pool3d> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(5)> #>             id    class lower upper nlevels        default  value #>         <char>   <char> <num> <num>   <num>         <list> <list> #> 1: kernel_size ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:     padding ParamUty    NA    NA     Inf              0 [NULL] #> 3:      stride ParamUty    NA    NA     Inf         [NULL] [NULL] #> 4:    dilation ParamInt  -Inf   Inf     Inf              1 [NULL] #> 5:   ceil_mode ParamLgl    NA    NA       2          FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":null,"dir":"Reference","previous_headings":"","what":"Merge Operation — mlr_pipeops_nn_merge","title":"Merge Operation — mlr_pipeops_nn_merge","text":"Base class merge operations addition (PipeOpTorchMergeSum), multiplication (PipeOpTorchMergeProd concatenation (PipeOpTorchMergeCat).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Merge Operation — mlr_pipeops_nn_merge","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Merge Operation — mlr_pipeops_nn_merge","text":"PipeOpTorchMerges either vararg input channel constructor argument innum set, input channels \"input1\", ..., \"input<innum>\". one output channel \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Merge Operation — mlr_pipeops_nn_merge","text":"See respective child class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Merge Operation — mlr_pipeops_nn_merge","text":"Per default, private$.shapes_out() method outputs broadcasted tensors. two things aware: NAs assumed batch (almost always batch size first dimension). Tensors expected number dimensions, .e. missing dimensions filled 1s. reason first dimension batch dimension. private method can overwritten PipeOpTorchs inheriting class.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge Operation — mlr_pipeops_nn_merge","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchMerge","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge Operation — mlr_pipeops_nn_merge","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Merge Operation — mlr_pipeops_nn_merge","text":"PipeOpTorchMerge$new() PipeOpTorchMerge$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Merge Operation — mlr_pipeops_nn_merge","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge Operation — mlr_pipeops_nn_merge","text":"","code":"PipeOpTorchMerge$new(   id,   module_generator,   param_set = ps(),   innum = 0,   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge Operation — mlr_pipeops_nn_merge","text":"id (character(1)) Identifier resulting  object. module_generator (nn_module_generator) torch module generator. param_set (ParamSet) parameter set. innum (integer(1)) number inputs. Default 0 means one vararg input channel. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Merge Operation — mlr_pipeops_nn_merge","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge Operation — mlr_pipeops_nn_merge","text":"","code":"PipeOpTorchMerge$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge Operation — mlr_pipeops_nn_merge","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":null,"dir":"Reference","previous_headings":"","what":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Concatenates multiple tensors given dimension. broadcasting rules applied , must reshape tensors shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch. PipeOpTorchMerges either vararg input channel constructor argument innum set, input channels \"input1\", ..., \"input<innum>\". one output channel \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"dim :: integer(1) dimension along concatenate tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Calls nn_merge_cat() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeCat","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"PipeOpTorchMergeCat$new() PipeOpTorchMergeCat$speak() PipeOpTorchMergeCat$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"","code":"PipeOpTorchMergeCat$new(id = \"nn_merge_cat\", innum = 0, param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"id (character(1)) Identifier resulting  object. innum (integer(1)) number inputs. Default 0 means one vararg input channel. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"method-speak-","dir":"Reference","previous_headings":"","what":"Method speak()","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"cat say?","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"","code":"PipeOpTorchMergeCat$speak()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"","code":"PipeOpTorchMergeCat$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_cat.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Merge by Concatenation — mlr_pipeops_nn_merge_cat","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_merge_cat\") pipeop #> PipeOp: <nn_merge_cat> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   ... [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <num> <num>   <num>  <list> <list> #> 1:    dim ParamInt  -Inf   Inf     Inf      -1 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":null,"dir":"Reference","previous_headings":"","what":"Merge by Product — mlr_pipeops_nn_merge_prod","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Calculates product input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch. PipeOpTorchMerges either vararg input channel constructor argument innum set, input channels \"input1\", ..., \"input<innum>\". one output channel \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Calls nn_merge_prod() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeProd","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"PipeOpTorchMergeProd$new() PipeOpTorchMergeProd$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"","code":"PipeOpTorchMergeProd$new(id = \"nn_merge_prod\", innum = 0, param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"id (character(1)) Identifier resulting  object. innum (integer(1)) number inputs. Default 0 means one vararg input channel. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"","code":"PipeOpTorchMergeProd$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_prod.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Merge by Product — mlr_pipeops_nn_merge_prod","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_merge_prod\") pipeop #> PipeOp: <nn_merge_prod> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   ... [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Merge by Summation — mlr_pipeops_nn_merge_sum","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Calculates sum input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch. PipeOpTorchMerges either vararg input channel constructor argument innum set, input channels \"input1\", ..., \"input<innum>\". one output channel \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Calls nn_merge_sum() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchMerge -> PipeOpTorchMergeSum","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"PipeOpTorchMergeSum$new() PipeOpTorchMergeSum$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"","code":"PipeOpTorchMergeSum$new(id = \"nn_merge_sum\", innum = 0, param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"id (character(1)) Identifier resulting  object. innum (integer(1)) number inputs. Default 0 means one vararg input channel. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"","code":"PipeOpTorchMergeSum$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_merge_sum.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Merge by Summation — mlr_pipeops_nn_merge_sum","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_merge_sum\") pipeop #> PipeOp: <nn_merge_sum> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   ... [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":null,"dir":"Reference","previous_headings":"","what":"PReLU Activation Function — mlr_pipeops_nn_prelu","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Applies element-wise function \\(PReLU(x) = max(0,x) + weight * min(0,x)\\) weight learnable parameter.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"num_parameters :: integer(1): Number learn. Although takes int input, two values legitimate: 1, number channels input. Default: 1. init :: numeric(1) T initial value . Default: 0.25.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Calls torch::nn_prelu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchPReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"PipeOpTorchPReLU$new() PipeOpTorchPReLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"","code":"PipeOpTorchPReLU$new(id = \"nn_prelu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"","code":"PipeOpTorchPReLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_prelu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"PReLU Activation Function — mlr_pipeops_nn_prelu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_prelu\") pipeop #> PipeOp: <nn_prelu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>                id    class lower upper nlevels default  value #>            <char>   <char> <num> <num>   <num>  <list> <list> #> 1: num_parameters ParamInt     1   Inf     Inf       1 [NULL] #> 2:           init ParamDbl  -Inf   Inf     Inf    0.25 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"ReLU Activation Function — mlr_pipeops_nn_relu","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Applies rectified linear unit function element-wise.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Calls torch::nn_relu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"PipeOpTorchReLU$new() PipeOpTorchReLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"","code":"PipeOpTorchReLU$new(id = \"nn_relu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"","code":"PipeOpTorchReLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"ReLU Activation Function — mlr_pipeops_nn_relu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_relu\") pipeop #> PipeOp: <nn_relu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Applies element-wise function \\(ReLU6(x) = min(max(0,x), 6)\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Calls torch::nn_relu6() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReLU6","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"PipeOpTorchReLU6$new() PipeOpTorchReLU6$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"","code":"PipeOpTorchReLU6$new(id = \"nn_relu6\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"","code":"PipeOpTorchReLU6$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_relu6.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"ReLU6 Activation Function — mlr_pipeops_nn_relu6","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_relu6\") pipeop #> PipeOp: <nn_relu6> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Reshape a Tensor — mlr_pipeops_nn_reshape","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Reshape tensor given shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"shape :: integer(1) desired output shape. Unknown dimension (one ) can either specified -1 NA.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Calls nn_reshape() trained. internally calls torch::torch_reshape() given shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchReshape","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"PipeOpTorchReshape$new() PipeOpTorchReshape$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"","code":"PipeOpTorchReshape$new(id = \"nn_reshape\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"","code":"PipeOpTorchReshape$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_reshape.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reshape a Tensor — mlr_pipeops_nn_reshape","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_reshape\") pipeop #> PipeOp: <nn_reshape> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels        default  value #>    <char>   <char> <num> <num>   <num>         <list> <list> #> 1:  shape ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":null,"dir":"Reference","previous_headings":"","what":"RReLU Activation Function — mlr_pipeops_nn_rrelu","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Randomized leaky ReLU.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"lower:: numeric(1) Lower bound uniform distribution. Default: 1/8. upper:: numeric(1) Upper bound uniform distribution. Default: 1/3. inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Calls torch::nn_rrelu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchRReLU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"PipeOpTorchRReLU$new() PipeOpTorchRReLU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"","code":"PipeOpTorchRReLU$new(id = \"nn_rrelu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"","code":"PipeOpTorchRReLU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_rrelu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"RReLU Activation Function — mlr_pipeops_nn_rrelu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_rrelu\") pipeop #> PipeOp: <nn_rrelu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(3)> #>         id    class lower upper nlevels   default  value #>     <char>   <char> <num> <num>   <num>    <list> <list> #> 1:   lower ParamDbl  -Inf   Inf     Inf     0.125 [NULL] #> 2:   upper ParamDbl  -Inf   Inf     Inf 0.3333333 [NULL] #> 3: inplace ParamLgl    NA    NA       2     FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"SELU Activation Function — mlr_pipeops_nn_selu","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Applies element-wise, $$SELU(x) = scale * (max(0,x) + min(0, \\alpha * (exp(x) - 1)))$$, \\(\\alpha=1.6732632423543772848170429916717\\) \\(scale=1.0507009873554804934193349852946\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"inplace :: logical(1) Whether operation -place. Default: FALSE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Calls torch::nn_selu() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSELU","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"PipeOpTorchSELU$new() PipeOpTorchSELU$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"","code":"PipeOpTorchSELU$new(id = \"nn_selu\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"","code":"PipeOpTorchSELU$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_selu.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SELU Activation Function — mlr_pipeops_nn_selu","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_selu\") pipeop #> PipeOp: <nn_selu> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>         id    class lower upper nlevels default  value #>     <char>   <char> <num> <num>   <num>  <list> <list> #> 1: inplace ParamLgl    NA    NA       2   FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Applies element-wise \\(Sigmoid(x_i) = \\frac{1}{1 + exp(-x_i)}\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Calls torch::nn_sigmoid() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSigmoid","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"PipeOpTorchSigmoid$new() PipeOpTorchSigmoid$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"","code":"PipeOpTorchSigmoid$new(id = \"nn_sigmoid\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"","code":"PipeOpTorchSigmoid$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_sigmoid.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid Activation Function — mlr_pipeops_nn_sigmoid","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_sigmoid\") pipeop #> PipeOp: <nn_sigmoid> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax — mlr_pipeops_nn_softmax","title":"Softmax — mlr_pipeops_nn_softmax","text":"Applies softmax function.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Softmax — mlr_pipeops_nn_softmax","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Softmax — mlr_pipeops_nn_softmax","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Softmax — mlr_pipeops_nn_softmax","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Softmax — mlr_pipeops_nn_softmax","text":"dim :: integer(1) dimension along Softmax computed (every slice along dim sum 1).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Softmax — mlr_pipeops_nn_softmax","text":"Calls torch::nn_softmax() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Softmax — mlr_pipeops_nn_softmax","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftmax","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Softmax — mlr_pipeops_nn_softmax","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Softmax — mlr_pipeops_nn_softmax","text":"PipeOpTorchSoftmax$new() PipeOpTorchSoftmax$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Softmax — mlr_pipeops_nn_softmax","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax — mlr_pipeops_nn_softmax","text":"","code":"PipeOpTorchSoftmax$new(id = \"nn_softmax\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax — mlr_pipeops_nn_softmax","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Softmax — mlr_pipeops_nn_softmax","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax — mlr_pipeops_nn_softmax","text":"","code":"PipeOpTorchSoftmax$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax — mlr_pipeops_nn_softmax","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softmax.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax — mlr_pipeops_nn_softmax","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_softmax\") pipeop #> PipeOp: <nn_softmax> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels        default  value #>    <char>   <char> <int> <num>   <num>         <list> <list> #> 1:    dim ParamInt     1   Inf     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Applies element-wise, function \\(Softplus(x) = 1/\\beta * log(1 + exp(\\beta * x))\\).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"beta :: numeric(1) beta value Softplus formulation. Default: 1 threshold :: numeric(1) Values revert linear function. Default: 20","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Calls torch::nn_softplus() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftPlus","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"PipeOpTorchSoftPlus$new() PipeOpTorchSoftPlus$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"","code":"PipeOpTorchSoftPlus$new(id = \"nn_softplus\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"","code":"PipeOpTorchSoftPlus$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softplus.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SoftPlus Activation Function — mlr_pipeops_nn_softplus","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_softplus\") pipeop #> PipeOp: <nn_softplus> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(2)> #>           id    class lower upper nlevels default  value #>       <char>   <char> <num> <num>   <num>  <list> <list> #> 1:      beta ParamDbl  -Inf   Inf     Inf       1 [NULL] #> 2: threshold ParamDbl  -Inf   Inf     Inf      20 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Applies soft shrinkage function elementwise","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"lamd :: numeric(1) lambda (must less zero) value Softshrink formulation. Default: 0.5","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Calls torch::nn_softshrink() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"PipeOpTorchSoftShrink$new() PipeOpTorchSoftShrink$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"","code":"PipeOpTorchSoftShrink$new(id = \"nn_softshrink\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"","code":"PipeOpTorchSoftShrink$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softshrink.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Soft Shrink Activation Function — mlr_pipeops_nn_softshrink","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_softshrink\") pipeop #> PipeOp: <nn_softshrink> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels default  value #>    <char>   <char> <num> <num>   <num>  <list> <list> #> 1:  lambd ParamDbl  -Inf     1     Inf     0.5 [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"SoftSign Activation Function — mlr_pipeops_nn_softsign","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Applies element-wise, function \\(SoftSign(x) = x/(1 + |x|\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Calls torch::nn_softsign() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSoftSign","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"PipeOpTorchSoftSign$new() PipeOpTorchSoftSign$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"","code":"PipeOpTorchSoftSign$new(id = \"nn_softsign\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"","code":"PipeOpTorchSoftSign$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_softsign.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SoftSign Activation Function — mlr_pipeops_nn_softsign","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_softsign\") pipeop #> PipeOp: <nn_softsign> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Squeezes tensor calling torch::torch_squeeze() given dimension dim.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"dim :: integer(1) dimension squeeze. NULL, dimensions size 1 squeezed. Negative values interpreted downwards last dimension.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Calls nn_squeeze() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchSqueeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"PipeOpTorchSqueeze$new() PipeOpTorchSqueeze$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"","code":"PipeOpTorchSqueeze$new(id = \"nn_squeeze\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"","code":"PipeOpTorchSqueeze$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_squeeze.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Squeeze a Tensor — mlr_pipeops_nn_squeeze","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_squeeze\") pipeop #> PipeOp: <nn_squeeze> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels        default  value #>    <char>   <char> <num> <num>   <num>         <list> <list> #> 1:    dim ParamUty    NA    NA     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Tanh Activation Function — mlr_pipeops_nn_tanh","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Applies element-wise function:","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Calls torch::nn_tanh() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchTanh","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"PipeOpTorchTanh$new() PipeOpTorchTanh$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"","code":"PipeOpTorchTanh$new(id = \"nn_tanh\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"","code":"PipeOpTorchTanh$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanh.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tanh Activation Function — mlr_pipeops_nn_tanh","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_tanh\") pipeop #> PipeOp: <nn_tanh> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Applies element-wise, \\(Tanhshrink(x) = x - Tanh(x)\\)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"parameters.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Calls torch::nn_tanhshrink() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchTanhShrink","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"PipeOpTorchTanhShrink$new() PipeOpTorchTanhShrink$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"","code":"PipeOpTorchTanhShrink$new(id = \"nn_tanhshrink\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"","code":"PipeOpTorchTanhShrink$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_tanhshrink.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tanh Shrink Activation Function — mlr_pipeops_nn_tanhshrink","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_tanhshrink\") pipeop #> PipeOp: <nn_tanhshrink> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(0)> #> Empty."},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Treshold Activation Function — mlr_pipeops_nn_threshold","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Thresholds element input Tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"threshold :: numeric(1) value threshold . value :: numeric(1) value replace . inplace :: logical(1) Can optionally operation -place. Default: ‘FALSE’.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Calls torch::nn_threshold() trained.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchThreshold","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"PipeOpTorchThreshold$new() PipeOpTorchThreshold$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"","code":"PipeOpTorchThreshold$new(id = \"nn_threshold\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"","code":"PipeOpTorchThreshold$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_threshold.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Treshold Activation Function — mlr_pipeops_nn_threshold","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_threshold\", threshold = 1, value = 2) pipeop #> PipeOp: <nn_threshold> (not trained) #> values: <threshold=1, value=2> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(3)> #>           id    class lower upper nlevels        default  value #>       <char>   <char> <num> <num>   <num>         <list> <list> #> 1: threshold ParamDbl  -Inf   Inf     Inf <NoDefault[0]>      1 #> 2:     value ParamDbl  -Inf   Inf     Inf <NoDefault[0]>      2 #> 3:   inplace ParamLgl    NA    NA       2          FALSE [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Unqueeze Tensor Unqueeze Tensor","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"state value calculated public method $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"dim :: integer(1) dimension unsqueeze. Negative values interpreted downwards last dimension.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Calls nn_unsqueeze() trained. internally calls torch::torch_unsqueeze().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> PipeOpTorchUnsqueeze","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train() mlr3torch::PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"PipeOpTorchUnsqueeze$new() PipeOpTorchUnsqueeze$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"","code":"PipeOpTorchUnsqueeze$new(id = \"nn_unsqueeze\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"","code":"PipeOpTorchUnsqueeze$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_unsqueeze.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unqueeze a Tensor — mlr_pipeops_nn_unsqueeze","text":"","code":"# Construct the PipeOp pipeop = po(\"nn_unsqueeze\") pipeop #> PipeOp: <nn_unsqueeze> (not trained) #> values: <list()> #> Input channels <name [train type, predict type]>: #>   input [ModelDescriptor,Task] #> Output channels <name [train type, predict type]>: #>   output [ModelDescriptor,Task] # The available parameters pipeop$param_set #> <ParamSet(1)> #>        id    class lower upper nlevels        default  value #>    <char>   <char> <num> <num>   <num>         <list> <list> #> 1:    dim ParamInt  -Inf   Inf     Inf <NoDefault[0]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"PipeOp can used preprocess (one ) lazy_tensor columns contained mlr3::Task. preprocessing function specified construction argument fn additional arguments function can defined PipeOp's parameter set. preprocessing done per column, .e. number lazy tensor output columns equal number lazy tensor input columns. create custom preprocessing PipeOps can use pipeop_preproc_torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"addition specifying construction arguments, can overwrite private .shapes_out() method. overwrite , output shapes assumed unknown (NULL). .shapes_out(shapes_in, param_vals, task) (list(), list(), TaskorNULL) -> list()\\cr private method calculates output shapes lazy tensor columns created applying preprocessing function provided parameter values (param_vals). taskis rarely needed, checked notNULL`. private method responsibility calculate output shapes one input column, .e. input shapes_in can assumed exactly one shape vector must calculate output shapes return list() length 1. can also assumed shape NULL (.e. unknown). Also, first dimension can NA, .e. unknown (batch dimension).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"See PipeOpTaskPreproc.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"addition state elements PipeOpTaskPreprocSimple, state also contains $param_vals set training.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"addition parameters inherited PipeOpTaskPreproc well specified construction argument param_set following parameters: stages :: character(1) stages apply preprocessing. Can one \"train\", \"predict\" \"\". initial value parameter set \"train\" PipeOp's id starts \"augment_\" \"\" otherwise. Note preprocessing applied $predict() uses parameters set $train() set performing prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"$train() / $predict(), PipeOpModule one input one output channel created. pipeop applies function fn input tensor additionally passing parameter values (minus stages affect_columns) fn. preprocessing graph lazy tensor columns shallowly cloned PipeOpModule added. done avoid modifying user input means identical PipeOpModules can part different preprocessing graphs. possible, created PipeOpModule stateless. later point graph, preprocessing graphs merged possible avoid unnecessary computation. best illustrated example: One lazy tensor column's preprocessing graph -> B. , two branches created B -> C B -> D, creating two preprocessing graphs -> B -> C -> B -> D. loading data, want run preprocessing , .e. want run -> B part twice. reason, task_dataset() try merge graphs cache results graphs. However, graphs using dataset can currently merged. Also, shapes created $train() $predict() might differ. avoid creation graphs predict shapes incompatible train shapes, hypothetical predict shapes already calculated $train() (parameters set train also used predict) PipeOpTorchModel check train predict shapes compatibility starting training. Otherwise, mechanism similar ModelDescriptor construct.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpTaskPreproc -> PipeOpTaskPreprocTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"fn preprocessing function. rowwise Whether preprocessing applied rowwise.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"PipeOpTaskPreprocTorch$new() PipeOpTaskPreprocTorch$shapes_out() PipeOpTaskPreprocTorch$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"","code":"PipeOpTaskPreprocTorch$new(   fn,   id = \"preproc_torch\",   param_vals = list(),   param_set = ps(),   packages = character(0),   rowwise = FALSE,   stages_init = NULL,   tags = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"fn (function character(2)) preprocessing function. Must modify input -place. character(2), first element namespace second element name. preprocessing function applied tensor, tensor passed position first argument. param_set inferred (left NULL) assumed first argument torch_tensor. id (character(1)) id new object. param_vals (named list()) Parameter values set construction. param_set (ParamSet) case function fn takes additional parameter besides torch_tensor can specfied parameters. None parameters can \"predict\" tag. tags include \"train\". packages (character()) packages preprocessing function depends . rowwise (logical(1)) Whether preprocessing function applied rowwise (concatenated row) directly whole tensor. first case batch dimension. stages_init (character(1)) Initial value stages parameter. tags (character()) Tags pipeop.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"method-shapes-out-","dir":"Reference","previous_headings":"","what":"Method shapes_out()","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"Calculates output shapes result applying preprocessing one lazy tensor columns provided shape. Names ignored order matters. uses parameter values currently set.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"","code":"PipeOpTaskPreprocTorch$shapes_out(shapes_in, stage = NULL, task = NULL)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"shapes_in (list() (integer() NULL)) input input shapes lazy tensors. NULL indicates shape unknown. First dimension must NA (NULL). stage (character(1)) stage: either \"train\" \"predict\". task (Task NULL) task, rarely needed.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"list() (integer() NULL)","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"usage-2","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"","code":"PipeOpTaskPreprocTorch$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"arguments-2","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_preproc_torch.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Base Class for Lazy Tensor Preprocessing — mlr_pipeops_preproc_torch","text":"","code":"# Creating a simple task d = data.table(   x1 = as_lazy_tensor(rnorm(10)),   x2 = as_lazy_tensor(rnorm(10)),   x3 = as_lazy_tensor(as.double(1:10)),   y = rnorm(10) )  taskin = as_task_regr(d, target = \"y\")  # Creating a simple preprocessing pipeop po_simple = po(\"preproc_torch\",   # get rid of environment baggage   fn = mlr3misc::crate(function(x, a) x + a),   param_set = paradox::ps(a = paradox::p_int(tags = c(\"train\", \"required\"))) )  po_simple$param_set$set_values(   a = 100,   affect_columns = selector_name(c(\"x1\", \"x2\")),   stages = \"both\" # use during train and predict )  taskout_train = po_simple$train(list(taskin))[[1L]] materialize(taskout_train$data(cols = c(\"x1\", \"x2\")), rbind = TRUE) #> $x1 #> torch_tensor #>  100.6242 #>   99.3046 #>   99.2164 #>   99.0469 #>  101.7928 #>  100.3490 #>  100.2591 #>   99.1940 #>  100.1057 #>   99.6664 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  101.6418 #>   99.3561 #>  100.5870 #>   99.8496 #>   98.2892 #>  101.4310 #>   97.3548 #>   98.9675 #>   99.2925 #>   99.2994 #> [ CPUFloatType{10,1} ] #>   taskout_predict_noaug = po_simple$predict(list(taskin))[[1L]] materialize(taskout_predict_noaug$data(cols = c(\"x1\", \"x2\")), rbind = TRUE) #> $x1 #> torch_tensor #>  100.6242 #>   99.3046 #>   99.2164 #>   99.0469 #>  101.7928 #>  100.3490 #>  100.2591 #>   99.1940 #>  100.1057 #>   99.6664 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  101.6418 #>   99.3561 #>  100.5870 #>   99.8496 #>   98.2892 #>  101.4310 #>   97.3548 #>   98.9675 #>   99.2925 #>   99.2994 #> [ CPUFloatType{10,1} ] #>   po_simple$param_set$set_values(   stages = \"train\" )  # transformation is not applied taskout_predict_aug = po_simple$predict(list(taskin))[[1L]] materialize(taskout_predict_aug$data(cols = c(\"x1\", \"x2\")), rbind = TRUE) #> $x1 #> torch_tensor #>  100.6242 #>   99.3046 #>   99.2164 #>   99.0469 #>  101.7928 #>  100.3490 #>  100.2591 #>   99.1940 #>  100.1057 #>   99.6664 #> [ CPUFloatType{10,1} ] #>  #> $x2 #> torch_tensor #>  101.6418 #>   99.3561 #>  100.5870 #>   99.8496 #>   98.2892 #>  101.4310 #>   97.3548 #>   98.9675 #>   99.2925 #>   99.2994 #> [ CPUFloatType{10,1} ] #>   # Creating a more complex preprocessing PipeOp PipeOpPreprocTorchPoly = R6::R6Class(\"PipeOpPreprocTorchPoly\",  inherit = PipeOpTaskPreprocTorch,  public = list(    initialize = function(id = \"preproc_poly\", param_vals = list()) {      param_set = paradox::ps(        n_degree = paradox::p_int(lower = 1L, tags = c(\"train\", \"required\"))      )      param_set$set_values(        n_degree = 1L      )      fn = mlr3misc::crate(function(x, n_degree) {        torch::torch_cat(          lapply(seq_len(n_degree), function(d) torch::torch_pow(x, d)),          dim = 2L        )      })       super$initialize(        fn = fn,        id = id,        packages = character(0),        param_vals = param_vals,        param_set = param_set,        stages_init = \"both\"      )    }  ),  private = list(    .shapes_out = function(shapes_in, param_vals, task) {      # shapes_in is a list of length 1 containing the shapes      checkmate::assert_true(length(shapes_in[[1L]]) == 2L)      if (shapes_in[[1L]][2L] != 1L) {        stop(\"Input shape must be (NA, 1)\")      }      list(c(NA, param_vals$n_degree))    }  ) )  po_poly = PipeOpPreprocTorchPoly$new(   param_vals = list(n_degree = 3L, affect_columns = selector_name(\"x3\")) )  po_poly$shapes_out(list(c(NA, 1L)), stage = \"train\") #> [[1]] #> [1] NA  3 #>   taskout = po_poly$train(list(taskin))[[1L]] materialize(taskout$data(cols = \"x3\"), rbind = TRUE) #> $x3 #> torch_tensor #>     1     1     1 #>     2     4     8 #>     3     9    27 #>     4    16    64 #>     5    25   125 #>     6    36   216 #>     7    49   343 #>     8    64   512 #>     9    81   729 #>    10   100  1000 #> [ CPUFloatType{10,3} ] #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"PipeOpTorch base class PipeOps represent neural network layers Graph. training, generates PipeOpModule wraps nn_module attaches architecture, also represented Graph consisting mostly PipeOpModules PipeOpNOPs. former Graph operates ModelDescriptors, latter operates tensors. relationship PipeOpTorch PipeOpModule similar relationshop nn_module_generator (like nn_linear) nn_module (like output nn_linear(...)). crucial difference PipeOpTorch infers auxiliary parameters (like in_features nn_linear) automatically intermediate tensor shapes communicated ModelDescriptor. prediction, PipeOpTorch takes Task channel outputs new Task resulting feature union channel. one input output channel, task simply piped .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"inheriting","dir":"Reference","previous_headings":"","what":"Inheriting","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"inheriting class, one overload either private$.shapes_out() private$.shape_dependent_params() methods, overload private$.make_module(). .make_module(shapes_in, param_vals, task) (list(), list()) -> nn_module private method called generated nn_module passed argument module PipeOpModule. must overwritten, module_generator provided. left , calls provided module_generator arguments obtained private method .shape_dependent_params(). .shapes_out(shapes_in, param_vals, task) (list(), list(), Task NULL) -> named list() private method gets list numeric vectors (shapes_in), parameter values (param_vals), well (optional) Task. shapes_in can assumed order input names PipeOp. output shapes must order output names PipeOp. case output shapes depends task (case PipeOpTorchHead), function return valid output shapes (possibly containing NAs) task argument provided . .shape_dependent_params(shapes_in, param_vals, task) (list(), list()) -> named list() private method inputs .shapes_out. .make_module() overwritten, constructs arguments passed module_generator. Usually means must infer auxiliary parameters can inferred input shapes add user-supplied parameter values (param_vals).","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"training, inputs outputs class ModelDescriptor. prediction, input output channels class Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"ParamSet specified child class inheriting PipeOpTorch. Usually parameters arguments wrapped nn_module minus auxiliary parameter can automatically inferred shapes input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"training, PipeOpTorch creates PipeOpModule given parameter specification input shapes incoming ModelDescriptors using private method .make_module(). input shapes provided slot pointer_shape incoming ModelDescriptors. channel names PipeOpModule identical channel names generating PipeOpTorch. model descriptor union incoming ModelDescriptors created. Note modifies graph first ModelDescriptor place efficiency. PipeOpModule added graph slot union edges connect sending PipeOpModules input channel PipeOpModule addeded graph. possible every incoming ModelDescriptor contains information id channel name sending PipeOp slot pointer. new graph model_descriptor_union represents current state neural network architecture. structurally similar subgraph consists pipeops class PipeOpTorch PipeOpTorchIngress ancestors PipeOpTorch. output, shallow copy ModelDescriptor created pointer pointer_shape updated accordingly. shallow copy means ModelDescriptors point Graph allows graph modified -reference different parts code.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"mlr3pipelines::PipeOp -> PipeOpTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"public-fields","dir":"Reference","previous_headings":"","what":"Public fields","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"module_generator (nn_module_generator NULL) module generator wrapped PipeOpTorch. NULL, private method private$.make_module(shapes_in, param_vals) must overwritte, see section 'Inheriting'. change construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"PipeOpTorch$new() PipeOpTorch$shapes_out()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"","code":"PipeOpTorch$new(   id,   module_generator,   param_set = ps(),   param_vals = list(),   inname = \"input\",   outname = \"output\",   packages = \"torch\",   tags = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"id (character(1)) Identifier resulting  object. module_generator (nn_module_generator) torch module generator. param_set (ParamSet) parameter set. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction. inname (character()) names PipeOp's input channels. input channels generated PipeOpModule. Unless wrapped module_generator's forward method (present) argument ..., inname must identical argument names order avoid ambiguity. forward method argument ..., order input channels determines tensors passed wrapped nn_module. left NULL (default), argument module_generator must given argument names modue_generator's forward function set inname. outname (character())  names output channels channels. ouput channels generated PipeOpModule therefore also names list returned $train(). case one output channel, nn_module constructed PipeOp training must return named list(), names list names output channels. default \"output\". packages (character()) R packages object depends . tags (character()) tags PipeOp. tags \"torch\" always added.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"method-shapes-out-","dir":"Reference","previous_headings":"","what":"Method shapes_out()","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"Calculates output shapes given input shapes, parameters task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"","code":"PipeOpTorch$shapes_out(shapes_in, task = NULL)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"shapes_in (list() integer()) input input shapes, must order input channel names PipeOp. task (Task NULL) task, rarely used (default NULL). exception PipeOpTorchHead.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"named list() containing output shapes. names names output channels PipeOp.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Base Class for Torch Module Constructor Wrappers — mlr_pipeops_torch","text":"","code":"## Creating a neural network # In torch  task = tsk(\"iris\")  network_generator = torch::nn_module(   initialize = function(task, d_hidden) {     d_in = length(task$feature_names)     self$linear = torch::nn_linear(d_in, d_hidden)     self$output = if (task$task_type == \"regr\") {       torch::nn_linear(d_hidden, 1)     } else if (task$task_type == \"classif\") {       torch::nn_linear(d_hidden, length(task$class_names))     }   },   forward = function(x) {     x = self$linear(x)     x = torch::nnf_relu(x)     self$output(x)   } )  network = network_generator(task, d_hidden = 50) x = torch::torch_tensor(as.matrix(task$data(1, task$feature_names))) y = torch::with_no_grad(network(x))   # In mlr3torch network_generator = po(\"torch_ingress_num\") %>>%   po(\"nn_linear\", out_features = 50) %>>%   po(\"nn_head\") md = network_generator$train(task)[[1L]] network = model_descriptor_to_module(md) y = torch::with_no_grad(network(torch_ingress_num.input = x))    ## Implementing a custom PipeOpTorch  # defining a custom module nn_custom = nn_module(\"nn_custom\",   initialize = function(d_in1, d_in2, d_out1, d_out2, bias = TRUE) {     self$linear1 = nn_linear(d_in1, d_out1, bias)     self$linear2 = nn_linear(d_in2, d_out2, bias)   },   forward = function(input1, input2) {     output1 = self$linear1(input1)     output2 = self$linear1(input2)      list(output1 = output1, output2 = output2)   } )  # wrapping the module into a custom PipeOpTorch  library(paradox)  PipeOpTorchCustom = R6::R6Class(\"PipeOpTorchCustom\",   inherit = PipeOpTorch,   public = list(     initialize = function(id = \"nn_custom\", param_vals = list()) {       param_set = ps(         d_out1 = p_int(lower = 1, tags = c(\"required\", \"train\")),         d_out2 = p_int(lower = 1, tags = c(\"required\", \"train\")),         bias = p_lgl(default = TRUE, tags = \"train\")       )       super$initialize(         id = id,         param_vals = param_vals,         param_set = param_set,         inname = c(\"input1\", \"input2\"),         outname = c(\"output1\", \"output2\"),         module_generator = nn_custom       )     }   ),   private = list(     .shape_dependent_params = function(shapes_in, param_vals, task) {       c(param_vals,         list(d_in1 = tail(shapes_in[[\"input1\"]], 1)), d_in2 = tail(shapes_in[[\"input2\"]], 1)       )     },     .shapes_out = function(shapes_in, param_vals, task) {       list(         input1 = c(head(shapes_in[[\"input1\"]], -1), param_vals$d_out1),         input2 = c(head(shapes_in[[\"input2\"]], -1), param_vals$d_out2)       )     }   ) )  ## Training  # generate input task = tsk(\"iris\") task1 = task$clone()$select(paste0(\"Sepal.\", c(\"Length\", \"Width\"))) task2 = task$clone()$select(paste0(\"Petal.\", c(\"Length\", \"Width\"))) graph = gunion(list(po(\"torch_ingress_num_1\"), po(\"torch_ingress_num_2\"))) mds_in = graph$train(list(task1, task2), single_input = FALSE)  mds_in[[1L]][c(\"graph\", \"task\", \"ingress\", \"pointer\", \"pointer_shape\")] #> $graph #> Graph with 1 PipeOps: #>                   ID         State sccssors prdcssors #>               <char>        <char>   <char>    <char> #>  torch_ingress_num_1 <<UNTRAINED>>                    #>  #> $task #> <TaskClassif:iris> (150 x 3): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (2): #>   - dbl (2): Sepal.Length, Sepal.Width #>  #> $ingress #> $ingress$torch_ingress_num_1.input #> Ingress: Task[Sepal.Length,Sepal.Width] --> Tensor(NA, 2) #>  #>  #> $pointer #> [1] \"torch_ingress_num_1\" \"output\"              #>  #> $pointer_shape #> [1] NA  2 #>  mds_in[[2L]][c(\"graph\", \"task\", \"ingress\", \"pointer\", \"pointer_shape\")] #> $graph #> Graph with 1 PipeOps: #>                   ID         State sccssors prdcssors #>               <char>        <char>   <char>    <char> #>  torch_ingress_num_2 <<UNTRAINED>>                    #>  #> $task #> <TaskClassif:iris> (150 x 3): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (2): #>   - dbl (2): Petal.Length, Petal.Width #>  #> $ingress #> $ingress$torch_ingress_num_2.input #> Ingress: Task[Petal.Length,Petal.Width] --> Tensor(NA, 2) #>  #>  #> $pointer #> [1] \"torch_ingress_num_2\" \"output\"              #>  #> $pointer_shape #> [1] NA  2 #>   # creating the PipeOpTorch and training it po_torch = PipeOpTorchCustom$new() po_torch$param_set$values = list(d_out1 = 10, d_out2 = 20) train_input = list(input1 = mds_in[[1L]], input2 = mds_in[[2L]]) mds_out = do.call(po_torch$train, args = list(input = train_input)) po_torch$state #> $output1 #> [1] NA 10 #>  #> $output2 #> [1] NA 20 #>   # the new model descriptors  # the resulting graphs are identical identical(mds_out[[1L]]$graph, mds_out[[2L]]$graph) #> [1] TRUE # not that as a side-effect, also one of the input graphs is modified in-place for efficiency mds_in[[1L]]$graph$edges #>                 src_id src_channel    dst_id dst_channel #>                 <char>      <char>    <char>      <char> #> 1: torch_ingress_num_1      output nn_custom      input1 #> 2: torch_ingress_num_2      output nn_custom      input2  # The new task has both Sepal and Petal features identical(mds_out[[1L]]$task, mds_out[[2L]]$task) #> [1] TRUE mds_out[[2L]]$task #> <TaskClassif:iris> (150 x 5): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (4): #>   - dbl (4): Petal.Length, Petal.Width, Sepal.Length, Sepal.Width  # The new ingress slot contains all ingressors identical(mds_out[[1L]]$ingress, mds_out[[2L]]$ingress) #> [1] TRUE mds_out[[1L]]$ingress #> $torch_ingress_num_1.input #> Ingress: Task[Sepal.Length,Sepal.Width] --> Tensor(NA, 2) #>  #> $torch_ingress_num_2.input #> Ingress: Task[Petal.Length,Petal.Width] --> Tensor(NA, 2) #>   # The pointer and pointer_shape slots are different mds_out[[1L]]$pointer #> [1] \"nn_custom\" \"output1\"   mds_out[[2L]]$pointer #> [1] \"nn_custom\" \"output2\"    mds_out[[1L]]$pointer_shape #> [1] NA 10 mds_out[[2L]]$pointer_shape #> [1] NA 20  ## Prediction predict_input = list(input1 = task1, input2 = task2) tasks_out = do.call(po_torch$predict, args = list(input = predict_input)) identical(tasks_out[[1L]], tasks_out[[2L]]) #> [1] TRUE"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback Configuration — mlr_pipeops_torch_callbacks","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"Configures callbacks deep learning model.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"one input channel \"input\" one output channel \"output\". training, channels class ModelDescriptor. prediction, channels class Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"parameters defined dynamically callbacks, id respective callbacks respective set id.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"training callbacks cloned added ModelDescriptor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"mlr3pipelines::PipeOp -> PipeOpTorchCallbacks","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"PipeOpTorchCallbacks$new() PipeOpTorchCallbacks$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"","code":"PipeOpTorchCallbacks$new(   callbacks = list(),   id = \"torch_callbacks\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"callbacks (list TorchCallbacks)  callbacks (something convertible via as_torch_callbacks()). Must unique ids. callbacks cloned construction. id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"","code":"PipeOpTorchCallbacks$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_callbacks.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback Configuration — mlr_pipeops_torch_callbacks","text":"","code":"po_cb = po(\"torch_callbacks\", \"checkpoint\") po_cb$param_set #> <ParamSetCollection(3)> #>                      id    class lower upper nlevels        default  value #>                  <char>   <char> <num> <num>   <num>         <list> <list> #> 1:      checkpoint.path ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:      checkpoint.freq ParamInt     1   Inf     Inf <NoDefault[0]> [NULL] #> 3: checkpoint.freq_type ParamFct    NA    NA       2          epoch [NULL] mdin = po(\"torch_ingress_num\")$train(list(tsk(\"iris\"))) mdin[[1L]]$callbacks #> named list() mdout = po_cb$train(mdin)[[1L]] mdout$callbacks #> $checkpoint #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: list() #> * Packages: mlr3torch,torch #>  # Can be called again po_cb1 = po(\"torch_callbacks\", t_clbk(\"progress\")) mdout1 = po_cb1$train(list(mdout))[[1L]] mdout1$callbacks #> $progress #> <TorchCallback:progress> Progress #> * Generator: CallbackSetProgress #> * Parameters: list() #> * Packages: progress,mlr3torch,torch #>  #> $checkpoint #> <TorchCallback:checkpoint> Checkpoint #> * Generator: CallbackSetCheckpoint #> * Parameters: list() #> * Packages: mlr3torch,torch #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":null,"dir":"Reference","previous_headings":"","what":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"Use entry-point mlr3torch-networks. Unless advanced user, need use directly PipeOpTorchIngressNumeric, PipeOpTorchIngressCategorical PipeOpTorchIngressLazyTensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"state set input shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"Defined construction argument param_set.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"Creates object class TorchIngressToken given task. purpuse store information construct torch dataloader task entry point network.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"mlr3pipelines::PipeOp -> PipeOpTorchIngress","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"active-bindings","dir":"Reference","previous_headings":"","what":"Active bindings","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"feature_types (character(1)) features types can consumed PipeOpTorchIngress.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"PipeOpTorchIngress$new() PipeOpTorchIngress$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"","code":"PipeOpTorchIngress$new(   id,   param_set = ps(),   param_vals = list(),   packages = character(0),   feature_types )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"id (character(1)) Identifier resulting  object. param_set (ParamSet) parameter set. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction. packages (character()) R packages object depends . feature_types (character()) feature types. See mlr_reflections$task_feature_types available values, Additionally, \"lazy_tensor\" supported.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"","code":"PipeOpTorchIngress$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Entrypoint to Torch Network — mlr_pipeops_torch_ingress","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"Ingress PipeOp represents categorical (factor(), ordered() logical()) entry point torch network.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"select :: logical(1) Whether PipeOp selected supported feature types. Otherwise err receiving tasks unsupported feature types.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"Uses batchgetter_categ().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"state set input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorchIngress -> PipeOpTorchIngressCategorical","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"PipeOpTorchIngressCategorical$new() PipeOpTorchIngressCategorical$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"","code":"PipeOpTorchIngressCategorical$new(   id = \"torch_ingress_categ\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"","code":"PipeOpTorchIngressCategorical$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_categ.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Entry Point for Categorical Features — mlr_pipeops_torch_ingress_categ","text":"","code":"graph = po(\"select\", selector = selector_type(\"factor\")) %>>%   po(\"torch_ingress_categ\") task = tsk(\"german_credit\") # The output is a model descriptor md = graph$train(task)[[1L]] ingress = md$ingress[[1L]] ingress$batchgetter(task$data(1, ingress$features), \"cpu\") #> torch_tensor #>  5  5  1  2  3  1  3  1  3  1  4  5  1  2 #> [ CPULongType{1,14} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":null,"dir":"Reference","previous_headings":"","what":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"Ingress single lazy_tensor column.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"shape :: integer() shape tensor, first dimension (batch) must NA. specified, lazy tensor input column needs known shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"returned batchgetter materializes lazy tensor column tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"state set input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorchIngress -> PipeOpTorchIngressLazyTensor","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"PipeOpTorchIngressLazyTensor$new() PipeOpTorchIngressLazyTensor$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"","code":"PipeOpTorchIngressLazyTensor$new(   id = \"torch_ingress_ltnsr\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"","code":"PipeOpTorchIngressLazyTensor$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Ingress for Lazy Tensor — mlr_pipeops_torch_ingress_ltnsr","text":"","code":"po_ingress = po(\"torch_ingress_ltnsr\") task = tsk(\"lazy_iris\")  md = po_ingress$train(list(task))[[1L]]  ingress = md$ingress x_batch = ingress[[1L]]$batchgetter(data = task$data(1, \"x\"), device = \"cpu\", cache = NULL) x_batch #> torch_tensor #>  5.1000  3.5000  1.4000  0.2000 #> [ CPUFloatType{1,4} ]  # Now we try a lazy tensor with unknown shape, i.e. the shapes between the rows can differ  ds = dataset(   initialize = function() self$x = list(torch_randn(3, 10, 10), torch_randn(3, 8, 8)),   .getitem = function(i) list(x = self$x[[i]]),   .length = function() 2)()  task_unknown = as_task_regr(data.table(   x = as_lazy_tensor(ds, dataset_shapes = list(x = NULL)),   y = rnorm(2) ), target = \"y\", id = \"example2\")  # this task (as it is) can NOT be processed by PipeOpTorchIngressLazyTensor # It therefore needs to be preprocessed po_resize = po(\"trafo_resize\", size = c(6, 6)) task_unknown_resize = po_resize$train(list(task_unknown))[[1L]]  # printing the transformed column still shows unknown shapes, # because the preprocessing pipeop cannot infer them, # however we know that the shape is now (3, 10, 10) for all rows task_unknown_resize$data(1:2, \"x\") #>                x #>    <lazy_tensor> #> 1:      <tnsr[]> #> 2:      <tnsr[]> po_ingress$param_set$set_values(shape = c(NA, 3, 6, 6))  md2 = po_ingress$train(list(task_unknown_resize))[[1L]]  ingress2 = md2$ingress x_batch2 = ingress2[[1L]]$batchgetter(   data = task_unknown_resize$data(1:2, \"x\"),   device = \"cpu\",   cache = NULL )  x_batch2 #> torch_tensor #> (1,1,.,.) =  #>  -0.3536  0.5422 -0.5122 -0.4266 -0.0916 -0.4592 #>   0.8402  1.3915 -0.7842  0.3108  0.4223 -0.6371 #>   0.7064 -1.1143 -1.0290 -1.5323 -0.1813  0.2049 #>   0.0286 -0.4911 -0.4830 -0.1494 -0.5065 -0.0469 #>  -0.6232 -0.3729  0.1369  1.1254 -0.5872  0.6150 #>   0.0530 -0.7461  1.2229  0.3075  0.2254 -0.2231 #>  #> (2,1,.,.) =  #>  -0.5584 -0.6006  0.0852  0.1745 -0.4200 -0.5432 #>  -0.3563  0.6179 -0.4114  0.0964 -0.3026 -0.0497 #>  -0.2654  0.0973 -0.1497 -0.0353  1.1315  0.6398 #>  -0.6037 -0.6516 -0.8290 -1.4187 -0.7332 -1.3848 #>  -0.3522  0.2617 -0.8548  0.1738  0.1908  0.1583 #>  -0.1979 -1.1566 -0.4095  0.3285  1.1347 -0.8909 #>  #> (1,2,.,.) =  #>  -0.8666  0.1124 -0.3811 -0.3970 -0.4885 -0.9129 #>  -1.1868 -1.8638  0.2943  0.8698  0.6637 -0.0425 #>   0.3130  0.8464 -0.3372  0.3382  0.7500  0.0065 #>  -0.1226 -1.1815  0.6687  0.8796 -0.1691  0.4671 #>   0.0456  1.0095  0.9920 -1.1715 -1.3439  0.0340 #>   0.5361  0.2356 -1.7181  0.9764 -1.5450 -0.0521 #>  #> (2,2,.,.) =  #>   1.1174  0.1390  0.5125 -0.4994 -0.1428 -0.2354 #>  -0.1027  0.6353  0.7274 -0.5460 -1.0755 -0.5056 #>   0.4653 -0.2932  0.7616  0.9773 -0.1486 -0.0574 #>   0.3783 -0.3556  1.6875  1.2363  0.1012 -0.2989 #>   0.8542  0.3894  0.9231  0.8197 -0.3083  0.0473 #> ... [the output was truncated (use n=-1 to disable)] #> [ CPUFloatType{2,3,6,6} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"Ingress PipeOp represents numeric (integer() numeric()) entry point torch network.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"Uses batchgetter_num().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"state set input shape.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorchIngress -> PipeOpTorchIngressNumeric","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"PipeOpTorchIngressNumeric$new() PipeOpTorchIngressNumeric$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"","code":"PipeOpTorchIngressNumeric$new(id = \"torch_ingress_num\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"","code":"PipeOpTorchIngressNumeric$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_ingress_num.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Entry Point for Numeric Features — mlr_pipeops_torch_ingress_num","text":"","code":"graph = po(\"select\", selector = selector_type(c(\"numeric\", \"integer\"))) %>>%   po(\"torch_ingress_num\") task = tsk(\"german_credit\") # The output is a model descriptor md = graph$train(task)[[1L]] ingress = md$ingress[[1L]] ingress$batchgetter(task$data(1:5, ingress$features), \"cpu\") #> torch_tensor #>    67  1169     6 #>    22  5951    48 #>    49  2096    12 #>    45  7882    42 #>    53  4870    24 #> [ CPUFloatType{5,3} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss Configuration — mlr_pipeops_torch_loss","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"Configures loss deep learning model.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"One input channel called \"input\" one output channel called \"output\". explanation see PipeOpTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"parameters defined dynamically loss set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"training loss cloned added ModelDescriptor.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"mlr3pipelines::PipeOp -> PipeOpTorchLoss","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"PipeOpTorchLoss$new() PipeOpTorchLoss$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"","code":"PipeOpTorchLoss$new(loss, id = \"torch_loss\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"loss (TorchLoss character(1) nn_loss) loss (something convertible via as_torch_loss()). id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"","code":"PipeOpTorchLoss$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_loss.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loss Configuration — mlr_pipeops_torch_loss","text":"","code":"po_loss = po(\"torch_loss\", loss = t_loss(\"cross_entropy\")) po_loss$param_set #> <ParamSet(3)> #>              id    class lower upper nlevels default  value #>          <char>   <char> <num> <num>   <num>  <list> <list> #> 1:       weight ParamUty    NA    NA     Inf  [NULL] [NULL] #> 2: ignore_index ParamInt  -Inf   Inf     Inf    -100 [NULL] #> 3:    reduction ParamFct    NA    NA       2    mean [NULL] mdin = po(\"torch_ingress_num\")$train(list(tsk(\"iris\"))) mdin[[1L]]$loss #> NULL mdout = po_loss$train(mdin)[[1L]] mdout$loss #> <TorchLoss:cross_entropy> Cross Entropy #> * Generator: nn_cross_entropy_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: classif"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOp Torch Model — mlr_pipeops_torch_model","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"Builds Torch Learner ModelDescriptor trains given parameter specification. task type must specified construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"one input channel \"input\" takes ModelDescriptor traing Task specified task_type prediction. output NULL training Prediction given task_type prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"trained LearnerTorchModel.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"General: parameters optimizer, loss callbacks, prefixed \"opt.\", \"loss.\" \"cb.<callback id>.\" respectively, well : epochs :: integer(1) number epochs. device :: character(1) device. One \"auto\", \"cpu\", \"cuda\" values defined mlr_reflections$torch$devices. value initialized \"auto\", select \"cuda\" possible, try \"mps\" otherwise fall back \"cpu\". num_threads :: integer(1) number threads intraop pararallelization (device \"cpu\"). value initialized 1. num_interop_threads :: integer(1) number threads intraop interop pararallelization (device \"cpu\"). value initialized 1. Note can set session changing value within R session raise warning. seed :: integer(1) \"random\" NULL torch seed used training prediction. value initialized \"random\", means random seed sampled beginning training phase. seed (either set randomly sampled) available via $model$seed training used prediction. Note setting seed training phase mean default (.e. seed \"random\"), clones learner use different seed. set NULL, seeding done. Evaluation: measures_train :: Measure list() Measures. Measures evaluated training. measures_valid :: Measure list() Measures. Measures evaluated validation. eval_freq :: integer(1) often train / validation predictions evaluated using measures_train / measures_valid. initialized 1. Note final model always evaluated. Early Stopping: patience :: integer(1) activates early stopping using validation scores. performance model improve patience evaluation steps, training ended. Note final model stored learner, best model. initialized 0, means early stopping. first entry measures_valid used metric. also requires specify $validate field Learner, well measures_valid. min_delta :: double(1) minimum improvement threshold (>) early stopping. initialized 0. Dataloader: batch_size :: integer(1) batch size (required). shuffle :: logical(1) Whether shuffle instances dataset. Default FALSE. impact validation. sampler :: torch::sampler Object defines dataloader draw samples. batch_sampler :: torch::sampler Object defines dataloader draws batches. num_workers :: integer(1) number workers data loading (batches loaded parallel). default 0, means data loaded main process. collate_fn :: function merge list samples form batch. pin_memory :: logical(1) Whether dataloader copies tensors CUDA pinned memory returning . drop_last :: logical(1) Whether drop last training batch epoch training. Default FALSE. timeout :: numeric(1) timeout value collecting batch workers. Negative values mean timeout default -1. worker_init_fn :: function(id) function receives worker id ([1, num_workers]) exectued seeding worker data loading. worker_globals :: list() | character() loading data parallel, allows export globals workers. character vector, objects global environment names copied workers. worker_packages :: character() packages load workers. Also see torch::dataloder information.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"LearnerTorchModel created calling model_descriptor_to_learner() provided ModelDescriptor received input channel. parameters set according parameters specified PipeOpTorchModel '$train() method called [Task][mlr3::Task] stored [ModelDescriptor`].","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpLearner -> PipeOpTorchModel","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"PipeOpTorchModel$new() PipeOpTorchModel$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"","code":"PipeOpTorchModel$new(task_type, id = \"torch_model\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"task_type (character(1)) task type model. id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"","code":"PipeOpTorchModel$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"PipeOp Torch Model — mlr_pipeops_torch_model","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":null,"dir":"Reference","previous_headings":"","what":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"Builds torch classifier trains .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"See LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"one input channel \"input\" takes ModelDescriptor traing Task specified task_type prediction. output NULL training Prediction given task_type prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"trained LearnerTorchModel.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"LearnerTorchModel created calling model_descriptor_to_learner() provided ModelDescriptor received input channel. parameters set according parameters specified PipeOpTorchModel '$train() method called [Task][mlr3::Task] stored [ModelDescriptor`].","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpLearner -> mlr3torch::PipeOpTorchModel -> PipeOpTorchModelClassif","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"PipeOpTorchModelClassif$new() PipeOpTorchModelClassif$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"","code":"PipeOpTorchModelClassif$new(id = \"torch_model_classif\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"","code":"PipeOpTorchModelClassif$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_classif.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"PipeOp Torch Classifier — mlr_pipeops_torch_model_classif","text":"","code":"# simple logistic regression  # configure the model descriptor md = as_graph(po(\"torch_ingress_num\") %>>%   po(\"nn_head\") %>>%   po(\"torch_loss\", \"cross_entropy\") %>>%   po(\"torch_optimizer\", \"adam\"))$train(tsk(\"iris\"))[[1L]]  print(md) #> <ModelDescriptor: 2 ops> #> * Ingress:  torch_ingress_num.input: [(NA,4)] #> * Task:  iris [classif] #> * Callbacks:  N/A #> * Optimizer:  Adaptive Moment Estimation #> * Loss:  Cross Entropy #> * pointer:  nn_head.output [(NA,3)]  # build the learner from the model descriptor and train it po_model = po(\"torch_model_classif\", batch_size = 50, epochs = 1) po_model$train(list(md)) #> $output #> NULL #>  po_model$state #> $model #> $network #> An `nn_module` containing 15 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • module_list: <nn_module_list> #15 parameters #>  #> $internal_valid_scores #> NULL #>  #> $loss_fn #> list() #>  #> $optimizer #> $optimizer$param_groups #> $optimizer$param_groups[[1]] #> $optimizer$param_groups[[1]]$params #> [1] 1 2 #>  #> $optimizer$param_groups[[1]]$lr #> [1] 0.001 #>  #> $optimizer$param_groups[[1]]$betas #> [1] 0.900 0.999 #>  #> $optimizer$param_groups[[1]]$eps #> [1] 1e-08 #>  #> $optimizer$param_groups[[1]]$weight_decay #> [1] 0 #>  #> $optimizer$param_groups[[1]]$amsgrad #> [1] FALSE #>  #>  #>  #> $optimizer$state #> $optimizer$state$`1` #> $optimizer$state$`1`$step #> [1] 3 #>  #> $optimizer$state$`1`$exp_avg #> torch_tensor #> -0.1067 -0.0173 -0.3737 -0.2571 #> -0.2698 -0.0843 -0.3516 -0.1529 #>  0.3765  0.1016  0.7253  0.4100 #> [ CPUFloatType{3,4} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> 0.01 * #>  0.1868  0.0053  2.1873  1.0252 #>   1.4240  0.1388  2.7922  0.6212 #>   1.5315  0.1404  4.2638  1.3239 #> [ CPUFloatType{3,4} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$step #> [1] 3 #>  #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -0.0748 #> -0.0581 #>  0.1329 #> [ CPUFloatType{3} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #> 0.001 * #>  0.8718 #>  0.7947 #>  1.3835 #> [ CPUFloatType{3} ] #>  #>  #>  #>  #> $epochs #> [1] 1 #>  #> $callbacks #> named list() #>  #> $seed #> [1] 1285070263 #>  #> $task_col_info #> Key: <id> #>              id    type                      levels #>          <char>  <char>                      <list> #> 1: Petal.Length numeric                      [NULL] #> 2:  Petal.Width numeric                      [NULL] #> 3: Sepal.Length numeric                      [NULL] #> 4:  Sepal.Width numeric                      [NULL] #> 5:      Species  factor setosa,versicolor,virginica #>  #> attr(,\"class\") #> [1] \"learner_torch_model\" \"list\"                #>  #> $log #> Empty data.table (0 rows and 3 cols): stage,class,msg #>  #> $train_time #> [1] 0.063 #>  #> $param_vals #> $param_vals$epochs #> [1] 1 #>  #> $param_vals$device #> [1] \"auto\" #>  #> $param_vals$num_threads #> [1] 1 #>  #> $param_vals$num_interop_threads #> [1] 1 #>  #> $param_vals$seed #> [1] \"random\" #>  #> $param_vals$eval_freq #> [1] 1 #>  #> $param_vals$measures_train #> list() #>  #> $param_vals$measures_valid #> list() #>  #> $param_vals$patience #> [1] 0 #>  #> $param_vals$min_delta #> [1] 0 #>  #> $param_vals$batch_size #> [1] 50 #>  #>  #> $task_hash #> [1] \"60543108f51f96bb\" #>  #> $feature_names #> [1] \"Petal.Length\" \"Petal.Width\"  \"Sepal.Length\" \"Sepal.Width\"  #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.22.1’ #>  #> $internal_tuned_values #> named list() #>  #> $data_prototype #> Empty data.table (0 rows and 5 cols): Species,Petal.Length,Petal.Width,Sepal.Length,Sepal.Width #>  #> $task_prototype #> Empty data.table (0 rows and 5 cols): Species,Petal.Length,Petal.Width,Sepal.Length,Sepal.Width #>  #> $train_task #> <TaskClassif:iris> (150 x 5): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (4): #>   - dbl (4): Petal.Length, Petal.Width, Sepal.Length, Sepal.Width #>  #> attr(,\"class\") #> [1] \"learner_state\" \"list\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch Regression Model — mlr_pipeops_torch_model_regr","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"Builds torch regression model trains .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"See LearnerTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"one input channel \"input\" takes ModelDescriptor traing Task specified task_type prediction. output NULL training Prediction given task_type prediction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"trained LearnerTorchModel.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"LearnerTorchModel created calling model_descriptor_to_learner() provided ModelDescriptor received input channel. parameters set according parameters specified PipeOpTorchModel '$train() method called [Task][mlr3::Task] stored [ModelDescriptor`].","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"mlr3pipelines::PipeOp -> mlr3pipelines::PipeOpLearner -> mlr3torch::PipeOpTorchModel -> PipeOpTorchModelRegr","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"PipeOpTorchModelRegr$new() PipeOpTorchModelRegr$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"PipeOpTorchModelRegr$new(id = \"torch_model_regr\", param_vals = list())"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"PipeOpTorchModelRegr$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_model_regr.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Torch Regression Model — mlr_pipeops_torch_model_regr","text":"","code":"# simple linear regression  # build the model descriptor md = as_graph(po(\"torch_ingress_num\") %>>%   po(\"nn_head\") %>>%   po(\"torch_loss\", \"mse\") %>>%   po(\"torch_optimizer\", \"adam\"))$train(tsk(\"mtcars\"))[[1L]]  print(md) #> <ModelDescriptor: 2 ops> #> * Ingress:  torch_ingress_num.input: [(NA,10)] #> * Task:  mtcars [regr] #> * Callbacks:  N/A #> * Optimizer:  Adaptive Moment Estimation #> * Loss:  Mean Squared Error #> * pointer:  nn_head.output [(NA,1)]  # build the learner from the model descriptor and train it po_model = po(\"torch_model_regr\", batch_size = 20, epochs = 1) po_model$train(list(md)) #> $output #> NULL #>  po_model$state #> $model #> $network #> An `nn_module` containing 11 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • module_list: <nn_module_list> #11 parameters #>  #> $internal_valid_scores #> NULL #>  #> $loss_fn #> list() #>  #> $optimizer #> $optimizer$param_groups #> $optimizer$param_groups[[1]] #> $optimizer$param_groups[[1]]$params #> [1] 1 2 #>  #> $optimizer$param_groups[[1]]$lr #> [1] 0.001 #>  #> $optimizer$param_groups[[1]]$betas #> [1] 0.900 0.999 #>  #> $optimizer$param_groups[[1]]$eps #> [1] 1e-08 #>  #> $optimizer$param_groups[[1]]$weight_decay #> [1] 0 #>  #> $optimizer$param_groups[[1]]$amsgrad #> [1] FALSE #>  #>  #>  #> $optimizer$state #> $optimizer$state$`1` #> $optimizer$state$`1`$step #> [1] 2 #>  #> $optimizer$state$`1`$exp_avg #> torch_tensor #> -6.4192 -38.0977 -76.4209 -2787.1460 -45.3608 -48.1255 -1960.0070 -215.3138 -4.9757 -38.1129 #> [ CPUFloatType{1,10} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> Columns 1 to 6 2.5316e+00  8.1738e+01  3.2276e+02  4.2917e+05  1.1383e+02  1.2935e+02 #>  #> Columns 7 to 10 2.1708e+05  2.5621e+03  1.4196e+00  8.0366e+01 #> [ CPUFloatType{1,10} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$step #> [1] 2 #>  #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -12.3812 #> [ CPUFloatType{1} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #>  8.4702 #> [ CPUFloatType{1} ] #>  #>  #>  #>  #> $epochs #> [1] 1 #>  #> $callbacks #> named list() #>  #> $seed #> [1] 1984753776 #>  #> $task_col_info #>         id    type levels #>     <char>  <char> <list> #>  1:     am numeric [NULL] #>  2:   carb numeric [NULL] #>  3:    cyl numeric [NULL] #>  4:   disp numeric [NULL] #>  5:   drat numeric [NULL] #>  6:   gear numeric [NULL] #>  7:     hp numeric [NULL] #>  8:   qsec numeric [NULL] #>  9:     vs numeric [NULL] #> 10:     wt numeric [NULL] #> 11:    mpg numeric [NULL] #>  #> attr(,\"class\") #> [1] \"learner_torch_model\" \"list\"                #>  #> $log #> Empty data.table (0 rows and 3 cols): stage,class,msg #>  #> $train_time #> [1] 0.041 #>  #> $param_vals #> $param_vals$epochs #> [1] 1 #>  #> $param_vals$device #> [1] \"auto\" #>  #> $param_vals$num_threads #> [1] 1 #>  #> $param_vals$num_interop_threads #> [1] 1 #>  #> $param_vals$seed #> [1] \"random\" #>  #> $param_vals$eval_freq #> [1] 1 #>  #> $param_vals$measures_train #> list() #>  #> $param_vals$measures_valid #> list() #>  #> $param_vals$patience #> [1] 0 #>  #> $param_vals$min_delta #> [1] 0 #>  #> $param_vals$batch_size #> [1] 20 #>  #>  #> $task_hash #> [1] \"907e8c646e24bf41\" #>  #> $feature_names #>  [1] \"am\"   \"carb\" \"cyl\"  \"disp\" \"drat\" \"gear\" \"hp\"   \"qsec\" \"vs\"   \"wt\"   #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.22.1’ #>  #> $internal_tuned_values #> named list() #>  #> $data_prototype #> Empty data.table (0 rows and 11 cols): mpg,am,carb,cyl,disp,drat... #>  #> $task_prototype #> Empty data.table (0 rows and 11 cols): mpg,am,carb,cyl,disp,drat... #>  #> $train_task #> <TaskRegr:mtcars> (32 x 11): Motor Trends #> * Target: mpg #> * Properties: - #> * Features (10): #>   - dbl (10): am, carb, cyl, disp, drat, gear, hp, qsec, vs, wt #>  #> attr(,\"class\") #> [1] \"learner_state\" \"list\""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer Configuration — mlr_pipeops_torch_optimizer","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"Configures optimizer deep learning model.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"input-and-output-channels","dir":"Reference","previous_headings":"","what":"Input and Output Channels","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"one input channel \"input\" one output channel \"output\". training, channels class ModelDescriptor. prediction, channels class Task.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"state","dir":"Reference","previous_headings":"","what":"State","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"state value calculated public method shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"parameters defined dynamically optimizer set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"training, optimizer cloned added ModelDescriptor. Note parameter set stored TorchOptimizer reference-identical parameter set pipeop .","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"super-class","dir":"Reference","previous_headings":"","what":"Super class","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"mlr3pipelines::PipeOp -> PipeOpTorchOptimizer","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"mlr3pipelines::PipeOp$help() mlr3pipelines::PipeOp$predict() mlr3pipelines::PipeOp$print() mlr3pipelines::PipeOp$train()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"public-methods","dir":"Reference","previous_headings":"","what":"Public methods","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"PipeOpTorchOptimizer$new() PipeOpTorchOptimizer$clone()","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"method-new-","dir":"Reference","previous_headings":"","what":"Method new()","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"Creates new instance R6 class.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"","code":"PipeOpTorchOptimizer$new(   optimizer = t_opt(\"adam\"),   id = \"torch_optimizer\",   param_vals = list() )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"optimizer (TorchOptimizer character(1) torch_optimizer_generator) optimizer (something convertible via as_torch_optimizer()). id (character(1)) Identifier resulting  object. param_vals (list()) List hyperparameter settings, overwriting hyperparameter settings otherwise set construction.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"method-clone-","dir":"Reference","previous_headings":"","what":"Method clone()","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"objects class cloneable method.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"usage-1","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"","code":"PipeOpTorchOptimizer$clone(deep = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"arguments-1","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"deep Whether make deep clone.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_torch_optimizer.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Optimizer Configuration — mlr_pipeops_torch_optimizer","text":"","code":"po_opt = po(\"torch_optimizer\", \"sgd\", lr = 0.01) po_opt$param_set #> <ParamSet(5)> #>              id    class lower upper nlevels        default  value #>          <char>   <char> <num> <num>   <num>         <list> <list> #> 1:           lr ParamDbl     0   Inf     Inf <NoDefault[0]>   0.01 #> 2:     momentum ParamDbl     0     1     Inf              0 [NULL] #> 3:    dampening ParamDbl     0     1     Inf              0 [NULL] #> 4: weight_decay ParamDbl     0     1     Inf              0 [NULL] #> 5:     nesterov ParamLgl    NA    NA       2          FALSE [NULL] mdin = po(\"torch_ingress_num\")$train(list(tsk(\"iris\"))) mdin[[1L]]$optimizer #> NULL mdout = po_opt$train(mdin) mdout[[1L]]$optimizer #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_sgd #> * Parameters: lr=0.01 #> * Packages: torch,mlr3torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":null,"dir":"Reference","previous_headings":"","what":"Iris Classification Task — mlr_tasks_lazy_iris","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"classification task popular datasets::iris data set. Just like iris task, features represented one lazy tensor column.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"R6::R6Class inheriting mlr3::TaskClassif.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"source","dir":"Reference","previous_headings":"","what":"Source","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"https://en.wikipedia.org/wiki/Iris_flower_data_set","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"","code":"tsk(\"lazy_iris\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"Task type: “classif” Properties: “multiclass” Missings: Target: “Species” Features: “x” Data Dimension: 150x3","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"Anderson E (1936). “Species Problem Iris.” Annals Missouri Botanical Garden, 23(3), 457. doi:10.2307/2394164 .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_lazy_iris.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Iris Classification Task — mlr_tasks_lazy_iris","text":"","code":"task = tsk(\"lazy_iris\") task #> <TaskClassif:lazy_iris> (150 x 2): Iris Flowers #> * Target: Species #> * Properties: multiclass #> * Features (1): #>   - lt (1): x df = task$data() materialize(df$x[1:6], rbind = TRUE) #> torch_tensor #>  5.1000  3.5000  1.4000  0.2000 #>  4.9000  3.0000  1.4000  0.2000 #>  4.7000  3.2000  1.3000  0.2000 #>  4.6000  3.1000  1.5000  0.2000 #>  5.0000  3.6000  1.4000  0.2000 #>  5.4000  3.9000  1.7000  0.4000 #> [ CPUFloatType{6,4} ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"MNIST Image classification — mlr_tasks_mnist","title":"MNIST Image classification — mlr_tasks_mnist","text":"Classic MNIST image classification. underlying DataBackend contains columns \"label\", \"image\", \"row_id\", \"split\", last column indicates whether row belongs train test set. first 60000 rows belong training set, last 10000 rows test set.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"source","dir":"Reference","previous_headings":"","what":"Source","title":"MNIST Image classification — mlr_tasks_mnist","text":"https://torchvision.mlverse.org/reference/mnist_dataset.html","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"MNIST Image classification — mlr_tasks_mnist","text":"","code":"tsk(\"mnist\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"download","dir":"Reference","previous_headings":"","what":"Download","title":"MNIST Image classification — mlr_tasks_mnist","text":"task's backend DataBackendLazy download data requested. meta-data already available . can cache datasets setting mlr3torch.cache option TRUE specific path used cache directory.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"MNIST Image classification — mlr_tasks_mnist","text":"Task type: “classif” Properties: “multiclass” Missings: Target: “label” Features: “image” Data Dimension: 70000x4","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"MNIST Image classification — mlr_tasks_mnist","text":"Lecun, Y., Bottou, L., Bengio, Y., Haffner, P. (1998). “Gradient-based learning applied document recognition.” Proceedings IEEE, 86(11), 2278-2324. doi:10.1109/5.726791 .","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_mnist.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"MNIST Image classification — mlr_tasks_mnist","text":"","code":"task = tsk(\"mnist\") task #> <TaskClassif:mnist> (70000 x 2): MNIST Digit Classification #> * Target: label #> * Properties: multiclass #> * Features (1): #>   - lt (1): image"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":null,"dir":"Reference","previous_headings":"","what":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"Subset famous ImageNet dataset. data obtained torchvision::tiny_imagenet_dataset(). underlying DataBackend contains columns \"class\", \"image\", \"..row_id\", \"split\", last column indicates whether row belongs train, validation test set defined provided torchvision. labels test rows, default, observations inactive, means task uses 110000 120000 observations defined underlying data backend.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"construction","dir":"Reference","previous_headings":"","what":"Construction","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"","code":"tsk(\"tiny_imagenet\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"download","dir":"Reference","previous_headings":"","what":"Download","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"task's backend DataBackendLazy download data requested. meta-data already available . can cache datasets setting mlr3torch.cache option TRUE specific path used cache directory.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"Task type: “classif” Properties: “multiclass” Missings: Target: “class” Features: “image” Data Dimension: 120000x4","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"Deng, Jia, Dong, Wei, Socher, Richard, Li, Li-Jia, Li, Kai, Fei-Fei, Li (2009). “Imagenet: large-scale hierarchical image database.” 2009 IEEE conference computer vision pattern recognition, 248–255. IEEE.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_tasks_tiny_imagenet.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tiny ImageNet Classification Task — mlr_tasks_tiny_imagenet","text":"","code":"task = tsk(\"tiny_imagenet\") task #> <TaskClassif:tiny_imagenet> (110000 x 2): ImageNet Subset #> * Target: class #> * Properties: multiclass #> * Features (1): #>   - lt (1): image"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_learner.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","title":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","text":"First nn_graph created using model_descriptor_to_module learner created module remaining information model descriptor, must include optimizer loss function optionally callbacks.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_learner.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","text":"","code":"model_descriptor_to_learner(model_descriptor)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_learner.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","text":"model_descriptor (ModelDescriptor) model descriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_learner.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Torch Learner from a ModelDescriptor — model_descriptor_to_learner","text":"Learner","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_module.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","title":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","text":"Creates nn_graph ModelDescriptor. Mostly internal use, since ModelDescriptor circumstances harder use just creating nn_graph directly.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_module.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","text":"","code":"model_descriptor_to_module(   model_descriptor,   output_pointers = NULL,   list_output = FALSE )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_module.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","text":"model_descriptor (ModelDescriptor) Model Descriptor. pointer ignored, instead output_pointer values used. $graph member modified -reference. output_pointers (list character) Collection pointers indicate part model_descriptor$graph used output. Entries format ModelDescriptor$pointer. list_output (logical(1)) Whether output list tensors. FALSE, length(output_pointers) must 1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_to_module.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a nn_graph from ModelDescriptor — model_descriptor_to_module","text":"nn_graph","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":null,"dir":"Reference","previous_headings":"","what":"Union of ModelDescriptors — model_descriptor_union","title":"Union of ModelDescriptors — model_descriptor_union","text":"mostly internal function used PipeOpTorchs multiple input channels. creates union multiple ModelDescriptors: graphs combinded (identical begin ). first entry's graph modified reference. PipeOps ID must identical. new input edges may added PipeOps. Drops pointer / pointer_shape entries. new task feature union two incoming tasks. optimizer loss ModelDescriptors must identical. Ingress tokens callbacks merged, objects \"id\" must identical.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Union of ModelDescriptors — model_descriptor_union","text":"","code":"model_descriptor_union(md1, md2)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Union of ModelDescriptors — model_descriptor_union","text":"md1 (ModelDescriptor) first ModelDescriptor. md2 (ModelDescriptor) second ModelDescriptor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Union of ModelDescriptors — model_descriptor_union","text":"ModelDescriptor","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/model_descriptor_union.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Union of ModelDescriptors — model_descriptor_union","text":"requirement new input edgedes may added PipeOps  theoretically necessary, since assume ModelDescriptor built beginning end (.e. PipeOps never get new ancestors) can make assumption simplify things. Otherwise need treat \"...\"-inputs special.)","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Neural Network Layer — nn","title":"Create a Neural Network Layer — nn","text":"Retrieve neural network layer mlr_pipeops dictionary.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Neural Network Layer — nn","text":"","code":"nn(.key, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Neural Network Layer — nn","text":".key (character(1)) ... () Additional parameters, constructor arguments fields.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Neural Network Layer — nn","text":"","code":"po1 = po(\"nn_linear\", id = \"linear\") # is the same as: po2 = nn(\"linear\")"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":null,"dir":"Reference","previous_headings":"","what":"Graph Network — nn_graph","title":"Graph Network — nn_graph","text":"Represents neural network using Graph usually costains mostly PipeOpModules.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Graph Network — nn_graph","text":"","code":"nn_graph(graph, shapes_in, output_map = graph$output$name, list_output = FALSE)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Graph Network — nn_graph","text":"graph (Graph) Graph wrap. cloned. shapes_in (named integer) Shape info tensors go graph. Names must graph$input$name, possibly different order. output_map (character) graph's outputs use. Must subset graph$output$name. list_output (logical(1)) Whether output list tensors. FALSE (default), length(output_map) must 1.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Graph Network — nn_graph","text":"nn_graph","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_graph.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Graph Network — nn_graph","text":"","code":"graph = mlr3pipelines::Graph$new() graph$add_pipeop(po(\"module_1\", module = nn_linear(10, 20)), clone = FALSE) graph$add_pipeop(po(\"module_2\", module = nn_relu()), clone = FALSE) graph$add_pipeop(po(\"module_3\", module = nn_linear(20, 1)), clone = FALSE) graph$add_edge(\"module_1\", \"module_2\") graph$add_edge(\"module_2\", \"module_3\")  network = nn_graph(graph, shapes_in = list(module_1.input = c(NA, 10)))  x = torch_randn(16, 10)  network(module_1.input = x) #> torch_tensor #> -0.2580 #> -0.1551 #> -0.2086 #> -0.1070 #>  0.0621 #> -0.3388 #> -0.3877 #>  0.0435 #> -0.2469 #> -0.1769 #> -0.3112 #> -0.1107 #> -0.1484 #> -0.0044 #> -0.3122 #> -0.3985 #> [ CPUFloatType{16,1} ][ grad_fn = <AddmmBackward0> ]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_cat.html","id":null,"dir":"Reference","previous_headings":"","what":"Concatenates multiple tensors — nn_merge_cat","title":"Concatenates multiple tensors — nn_merge_cat","text":"Concatenates multiple tensors given dimension. broadcasting rules applied , must reshape tensors shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_cat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Concatenates multiple tensors — nn_merge_cat","text":"","code":"nn_merge_cat(dim = -1)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_cat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Concatenates multiple tensors — nn_merge_cat","text":"dim (integer(1)) dimension concatenation.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_prod.html","id":null,"dir":"Reference","previous_headings":"","what":"Product of multiple tensors — nn_merge_prod","title":"Product of multiple tensors — nn_merge_prod","text":"Calculates product input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_prod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Product of multiple tensors — nn_merge_prod","text":"","code":"nn_merge_prod()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Sum of multiple tensors — nn_merge_sum","title":"Sum of multiple tensors — nn_merge_sum","text":"Calculates sum input tensors.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_merge_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sum of multiple tensors — nn_merge_sum","text":"","code":"nn_merge_sum()"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Reshape — nn_reshape","title":"Reshape — nn_reshape","text":"Reshape tensor given shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reshape — nn_reshape","text":"","code":"nn_reshape(shape)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reshape — nn_reshape","text":"shape (integer()) desired output shape.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_squeeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Squeeze — nn_squeeze","title":"Squeeze — nn_squeeze","text":"Squeezes tensor calling torch::torch_squeeze() given dimension dim.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_squeeze.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squeeze — nn_squeeze","text":"","code":"nn_squeeze(dim)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_squeeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squeeze — nn_squeeze","text":"dim (integer()) dimension squeeze.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_unsqueeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Unsqueeze — nn_unsqueeze","title":"Unsqueeze — nn_unsqueeze","text":"Unsqueezes tensor calling torch::torch_unsqueeze() given dimension dim.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_unsqueeze.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unsqueeze — nn_unsqueeze","text":"","code":"nn_unsqueeze(dim)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/nn_unsqueeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unsqueeze — nn_unsqueeze","text":"dim (integer(1)) dimension unsqueeze.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":null,"dir":"Reference","previous_headings":"","what":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"Function create objects class PipeOpTaskPreprocTorch convenient way. Start reading documentation PipeOpTaskPreprocTorch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"","code":"pipeop_preproc_torch(   id,   fn,   shapes_out = NULL,   param_set = NULL,   packages = character(0),   rowwise = FALSE,   parent_env = parent.frame(),   stages_init = NULL,   tags = NULL )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"id (character(1)) id new object. fn (function) preprocessing function. shapes_out (function NULL \"infer\") private .shapes_out(shapes_in, param_vals, task) method PipeOpTaskPreprocTorch (see section Inheriting). Special values NULL infer: NULL, output shapes unknown. \"infer\", output shape function inferred calculates output shapes follows: input shape (NA, ...) meta-tensor shape (1, ...) created preprocessing function applied. Afterwards batch dimension (1) replaced NA shape returned. first dimension NA, output shape applying preprocessing function returned. Method \"infer\" correct cases, might fail edge cases. param_set (ParamSet NULL) parameter set. left NULL (default) parameter set inferred following way: parameters first ... fn set untyped parameters tags 'train' default value tagged 'required' well. Default values annotated. packages (character()) R packages object depends . rowwise (logical(1)) Whether preprocessing applied row-wise. parent_env (environment) parent environment R6 class. stages_init (character(1)) Initial value stages parameter. NULL (default), set \"\" case id starts \"trafo\" \"train\" starts \"augment\". Otherwise must specified. tags (character()) Tags pipeop","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"R6Class instance inheriting PipeOpTaskPreprocTorch","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/pipeop_preproc_torch.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create Torch Preprocessing PipeOps — pipeop_preproc_torch","text":"","code":"PipeOpPreprocExample = pipeop_preproc_torch(\"preproc_example\", function(x, a) x + a) po_example = PipeOpPreprocExample$new() po_example$param_set #> <ParamSet(3)> #>                id    class lower upper nlevels        default  value #>            <char>   <char> <num> <num>   <num>         <list> <list> #> 1:              a ParamUty    NA    NA     Inf <NoDefault[0]> [NULL] #> 2:         stages ParamFct    NA    NA       3 <NoDefault[0]> [NULL] #> 3: affect_columns ParamUty    NA    NA     Inf  <Selector[1]> [NULL]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/replace_head.html","id":null,"dir":"Reference","previous_headings":"","what":"Replace the head of a network Replaces the head of the network with a linear layer with d_out classes. — replace_head","title":"Replace the head of a network Replaces the head of the network with a linear layer with d_out classes. — replace_head","text":"Replace head network Replaces head network linear layer d_out classes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/replace_head.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Replace the head of a network Replaces the head of the network with a linear layer with d_out classes. — replace_head","text":"","code":"replace_head(network, d_out)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/replace_head.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Replace the head of a network Replaces the head of the network with a linear layer with d_out classes. — replace_head","text":"network (torch::nn_module) network d_out (integer(1)) number output classes.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":null,"dir":"Reference","previous_headings":"","what":"Sugar Function for Torch Callback — t_clbk","title":"Sugar Function for Torch Callback — t_clbk","text":"Retrieves one TorchCallbacks mlr3torch_callbacks. Works like mlr3::lrn() mlr3::lrns().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sugar Function for Torch Callback — t_clbk","text":"","code":"t_clbk(.key, ...)  t_clbks(.keys)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sugar Function for Torch Callback — t_clbk","text":".key (character(1)) key torch callback. ... () See description dictionary_sugar_get(). .keys (character()) keys callbacks.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sugar Function for Torch Callback — t_clbk","text":"TorchCallback list() TorchCallbacks","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_clbk.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sugar Function for Torch Callback — t_clbk","text":"","code":"t_clbk(\"progress\") #> <TorchCallback:progress> Progress #> * Generator: CallbackSetProgress #> * Parameters: list() #> * Packages: progress,mlr3torch,torch"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss Function Quick Access — t_loss","title":"Loss Function Quick Access — t_loss","text":"Retrieve one TorchLosses mlr3torch_losses. Works like mlr3::lrn() mlr3::lrns().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss Function Quick Access — t_loss","text":"","code":"t_loss(.key, ...)  t_losses(.keys, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss Function Quick Access — t_loss","text":".key (character(1)) Key object retrieve. ... () See description dictionary_sugar_get. .keys (character()) keys losses.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loss Function Quick Access — t_loss","text":"TorchLoss","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_loss.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loss Function Quick Access — t_loss","text":"","code":"t_loss(\"mse\", reduction = \"mean\") #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: reduction=mean #> * Packages: torch,mlr3torch #> * Task Types: regr # get the dictionary t_loss() #> <DictionaryMlr3torchLosses> with 3 stored values #> Keys: cross_entropy, l1, mse t_losses(c(\"mse\", \"l1\")) #> $mse #> <TorchLoss:mse> Mean Squared Error #> * Generator: nn_mse_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr #>  #> $l1 #> <TorchLoss:l1> Absolute Error #> * Generator: nn_l1_loss #> * Parameters: list() #> * Packages: torch,mlr3torch #> * Task Types: regr #>  # get the dictionary t_losses() #> <DictionaryMlr3torchLosses> with 3 stored values #> Keys: cross_entropy, l1, mse"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizers Quick Access — t_opt","title":"Optimizers Quick Access — t_opt","text":"Retrieves one TorchOptimizers mlr3torch_optimizers. Works like mlr3::lrn() mlr3::lrns().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizers Quick Access — t_opt","text":"","code":"t_opt(.key, ...)  t_opts(.keys, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizers Quick Access — t_opt","text":".key (character(1)) Key object retrieve. ... () See description dictionary_sugar_get. .keys (character()) keys optimizers.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizers Quick Access — t_opt","text":"TorchOptimizer","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/t_opt.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Optimizers Quick Access — t_opt","text":"","code":"t_opt(\"adam\", lr = 0.1) #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: lr=0.1 #> * Packages: torch,mlr3torch # get the dictionary t_opt() #> <DictionaryMlr3torchOptimizers> with 7 stored values #> Keys: adadelta, adagrad, adam, asgd, rmsprop, rprop, sgd t_opts(c(\"adam\", \"sgd\")) #> $adam #> <TorchOptimizer:adam> Adaptive Moment Estimation #> * Generator: optim_adam #> * Parameters: list() #> * Packages: torch,mlr3torch #>  #> $sgd #> <TorchOptimizer:sgd> Stochastic Gradient Descent #> * Generator: optim_sgd #> * Parameters: list() #> * Packages: torch,mlr3torch #>  # get the dictionary t_opts() #> <DictionaryMlr3torchOptimizers> with 7 stored values #> Keys: adadelta, adagrad, adam, asgd, rmsprop, rprop, sgd"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Dataset from a Task — task_dataset","title":"Create a Dataset from a Task — task_dataset","text":"Creates torch dataset mlr3 Task. resulting dataset's $.get_batch() method returns list elements x, y index: x list tensors, whose content defined parameter feature_ingress_tokens. y target variable content defined parameter target_batchgetter. .index index batch task's data. data returned device specified parameter device.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Dataset from a Task — task_dataset","text":"","code":"task_dataset(task, feature_ingress_tokens, target_batchgetter = NULL, device)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Dataset from a Task — task_dataset","text":"task (Task) task build dataset. feature_ingress_tokens (named list() TorchIngressToken) ingress token defines one item $x value batch corresponding names. target_batchgetter (function(data, device)) function taking arguments data, data.table containing target variable, device. must return target torch tensor selected device. device (character()) device, e.g. \"cuda\" \"cpu\".","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Dataset from a Task — task_dataset","text":"torch::dataset","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/task_dataset.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Dataset from a Task — task_dataset","text":"","code":"task = tsk(\"iris\") sepal_ingress = TorchIngressToken(   features = c(\"Sepal.Length\", \"Sepal.Width\"),   batchgetter = batchgetter_num,   shape = c(NA, 2) ) petal_ingress = TorchIngressToken(   features = c(\"Petal.Length\", \"Petal.Width\"),   batchgetter = batchgetter_num,   shape = c(NA, 2) ) ingress_tokens = list(sepal = sepal_ingress, petal = petal_ingress)  target_batchgetter = function(data, device) {   torch_tensor(data = data[[1L]], dtype = torch_float32(), device)$unsqueeze(2) } dataset = task_dataset(task, ingress_tokens, target_batchgetter, \"cpu\") batch = dataset$.getbatch(1:10) batch #> $x #> $x$sepal #> torch_tensor #>  5.1000  3.5000 #>  4.9000  3.0000 #>  4.7000  3.2000 #>  4.6000  3.1000 #>  5.0000  3.6000 #>  5.4000  3.9000 #>  4.6000  3.4000 #>  5.0000  3.4000 #>  4.4000  2.9000 #>  4.9000  3.1000 #> [ CPUFloatType{10,2} ] #>  #> $x$petal #> torch_tensor #>  1.4000  0.2000 #>  1.4000  0.2000 #>  1.3000  0.2000 #>  1.5000  0.2000 #>  1.4000  0.2000 #>  1.7000  0.4000 #>  1.4000  0.3000 #>  1.5000  0.2000 #>  1.4000  0.2000 #>  1.5000  0.1000 #> [ CPUFloatType{10,2} ] #>  #>  #> $.index #> torch_tensor #>   1 #>   2 #>   3 #>   4 #>   5 #>   6 #>   7 #>   8 #>   9 #>  10 #> [ CPULongType{10} ] #>  #> $y #> torch_tensor #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #>  1 #> [ CPUFloatType{10,1} ] #>"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Callback Desctiptor — torch_callback","title":"Create a Callback Desctiptor — torch_callback","text":"Convenience function create custom TorchCallback. arguments available callback_set() also available . information correctly implement new callback, see CallbackSet.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Callback Desctiptor — torch_callback","text":"","code":"torch_callback(   id,   classname = paste0(\"CallbackSet\", capitalize(id)),   param_set = NULL,   packages = NULL,   label = capitalize(id),   man = NULL,   on_begin = NULL,   on_end = NULL,   on_exit = NULL,   on_epoch_begin = NULL,   on_before_valid = NULL,   on_epoch_end = NULL,   on_batch_begin = NULL,   on_batch_end = NULL,   on_after_backward = NULL,   on_batch_valid_begin = NULL,   on_batch_valid_end = NULL,   on_valid_end = NULL,   state_dict = NULL,   load_state_dict = NULL,   initialize = NULL,   public = NULL,   private = NULL,   active = NULL,   parent_env = parent.frame(),   inherit = CallbackSet,   lock_objects = FALSE )"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Callback Desctiptor — torch_callback","text":"id (character(1))` id torch callback. classname (character(1)) class name. param_set (ParamSet) parameter set, present inferred $initialize() method. packages (character())packages callback depends . Default isNULL`. label (character(1)) label torch callback. Defaults capitalized id. man (character(1)) String format [pkg]::[topic] pointing manual page object. referenced help package can opened via method $help(). default NULL. on_begin, on_end, on_epoch_begin, on_before_valid, on_epoch_end, on_batch_begin, on_batch_end, on_after_backward, on_batch_valid_begin, on_batch_valid_end, on_valid_end, on_exit (function) Function execute given stage, see section Stages. state_dict (function()) function retrieves state dict callback. available learner training. load_state_dict (function(state_dict)) Function loads callback state. initialize (function()) initialization method callback. public, private, active (list()) Additional public, private, active fields add callback. parent_env (environment()) parent environment R6Class. inherit (R6ClassGenerator) class inherit. class must either CallbackSet (default) inherit . lock_objects (logical(1)) Whether lock objects resulting R6Class. FALSE (default), values can freely assigned self without declaring class definition.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Callback Desctiptor — torch_callback","text":"TorchCallback","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"Create a Callback Desctiptor — torch_callback","text":"first creates R6 class inheriting CallbackSet (using callback_set()) wraps generator TorchCallback can passed torch learner.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"stages","dir":"Reference","previous_headings":"","what":"Stages","title":"Create a Callback Desctiptor — torch_callback","text":"begin :: Run training loop begins. epoch_begin :: Run beginning epoch. batch_begin :: Run forward call. after_backward :: Run backward call. batch_end :: Run optimizer step. batch_valid_begin :: Run forward call validation loop. batch_valid_end :: Run forward call validation loop. valid_end :: Run end validation. epoch_end :: Run end epoch. end :: Run last epoch. exit :: Run last, using .exit().","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/torch_callback.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Callback Desctiptor — torch_callback","text":"","code":"custom_tcb = torch_callback(\"custom\",   initialize = function(name) {     self$name = name   },   on_begin = function() {     cat(\"Hello\", self$name, \", we will train for \", self$ctx$total_epochs, \"epochs.\\n\")   },   on_end = function() {     cat(\"Training is done.\")   } )  learner = lrn(\"classif.torch_featureless\",   batch_size = 16,   epochs = 1,   callbacks = custom_tcb,   cb.custom.name = \"Marie\",   device = \"cpu\" ) task = tsk(\"iris\") learner$train(task) #> Hello Marie , we will train for  1 epochs. #> Training is done."},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"mlr3torch-012","dir":"Changelog","previous_headings":"","what":"mlr3torch 0.1.2","title":"mlr3torch 0.1.2","text":"CRAN release: 2024-10-15 Don’t use deprecated data_formats anymore Added CallbackSetTB, allows logging can viewed TensorBoard.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"mlr3torch-011","dir":"Changelog","previous_headings":"","what":"mlr3torch 0.1.1","title":"mlr3torch 0.1.1","text":"CRAN release: 2024-09-12 fix(preprocessing): regarding construction PipeOps po(\"trafo_resize\") failed cases. fix(ci): tests run CI fix(learner): LearnerTabResnet now works correctly Fix tests run CI feat: added nn() helper function simplify creation neural network layers","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/news/index.html","id":"mlr3torch-010","dir":"Changelog","previous_headings":"","what":"mlr3torch 0.1.0","title":"mlr3torch 0.1.0","text":"CRAN release: 2024-07-08 Initial CRAN release","code":""}]