From 376e5c1a488dd6616369608d74d43ba33613233c Mon Sep 17 00:00:00 2001
From: sebffischer <sebf.fischer@gmail.com>
Date: Fri, 8 Nov 2024 09:52:12 +0000
Subject: [PATCH] =?UTF-8?q?Deploying=20to=20gh-pages=20from=20@=20mlr-org/?=
 =?UTF-8?q?mlr3torch@bd80db5b5249361a874c0d61d0945d3616f77211=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      |  59 ++--
 .../figure-html/unnamed-chunk-37-1.png        | Bin 123484 -> 123689 bytes
 .../figure-html/unnamed-chunk-42-1.png        | Bin 123441 -> 123566 bytes
 .../figure-html/unnamed-chunk-46-1.png        | Bin 123298 -> 123400 bytes
 .../figure-html/unnamed-chunk-48-1.png        | Bin 122702 -> 124119 bytes
 .../figure-html/unnamed-chunk-50-1.png        | Bin 124094 -> 123416 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                | 262 +++++++++---------
 dev/reference/mlr_learners.torchvision.html   |   6 +-
 dev/reference/mlr_learners_torch_image.html   |   2 +-
 dev/reference/mlr_learners_torch_model.html   |  16 +-
 .../mlr_pipeops_nn_adaptive_avg_pool1d.html   |   2 +-
 .../mlr_pipeops_torch_ingress_ltnsr.html      |  46 +--
 .../mlr_pipeops_torch_model_classif.html      |  27 +-
 .../mlr_pipeops_torch_model_regr.html         |  14 +-
 dev/reference/nn_graph.html                   |  32 +--
 dev/search.json                               |   2 +-
 23 files changed, 248 insertions(+), 244 deletions(-)

diff --git a/dev/articles/callbacks.html b/dev/articles/callbacks.html
index c55d112d..17260a19 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: 0x55f6ed50fb38&gt;</span></span>
+<span><span class="co">##   Parent env: &lt;environment: 0x55dfb6df1b30&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 deb9c3c5..678472df 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: 0x556c58224810&gt;</span></span>
+<span><span class="co">#&gt;   Parent env: &lt;environment: 0x5584a92d3ea8&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 b7599b72..0f82ec20 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;  0.9197  0.6295 -0.9055</span></span>
-<span><span class="co">#&gt; -2.5884  0.7595  1.2294</span></span>
+<span><span class="co">#&gt; -1.3766 -0.5136  0.3212</span></span>
+<span><span class="co">#&gt; -0.1381  0.5962  0.2744</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;  0.6356 -0.0022  0.3491 -1.0918</span></span>
-<span><span class="co">#&gt; -1.5718  1.8213 -1.6298  0.2341</span></span>
+<span><span class="co">#&gt;  0.2026 -0.6605  0.1249  0.6521</span></span>
+<span><span class="co">#&gt;  0.4501 -0.2476  0.1827  0.3494</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;  0.2569  0.4082  0.3350</span></span>
-<span><span class="co">#&gt;  0.3488  0.3890  0.2623</span></span>
+<span><span class="co">#&gt;  0.3464  0.1966  0.4570</span></span>
+<span><span class="co">#&gt;  0.3344  0.1942  0.4714</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">$</span><span class="fu">train</span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="va">input</span><span class="op">)</span><span class="op">)</span><span class="op">[[</span><span class="fl">1</span><span class="op">]</span><span class="op">]</span></span>
 <span><span class="va">output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.6356 -0.0022  0.3491 -1.0918</span></span>
-<span><span class="co">#&gt; -1.5718  1.8213 -1.6298  0.2341</span></span>
+<span><span class="co">#&gt;  0.2026 -0.6605  0.1249  0.6521</span></span>
+<span><span class="co">#&gt;  0.4501 -0.2476  0.1827  0.3494</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,4} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span></code></pre></div>
 <p>Note we only use the <code>$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;  0.2569  0.4082  0.3350</span></span>
-<span><span class="co">#&gt;  0.3488  0.3890  0.2623</span></span>
+<span><span class="co">#&gt;  0.3464  0.1966  0.4570</span></span>
+<span><span class="co">#&gt;  0.3344  0.1942  0.4714</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;  0.2569  0.4082  0.3350</span></span>
-<span><span class="co">#&gt;  0.3488  0.3890  0.2623</span></span>
+<span><span class="co">#&gt;  0.3464  0.1966  0.4570</span></span>
+<span><span class="co">#&gt;  0.3344  0.1942  0.4714</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;  0.2935  0.0055  0.1483  0.6004</span></span>
-<span><span class="co">#&gt;  0.3216  2.2073 -0.5215 -0.5771</span></span>
+<span><span class="co">#&gt;  0.2228 -0.2205 -0.0451  0.1615</span></span>
+<span><span class="co">#&gt;  0.2096  0.2013 -0.4630 -0.3194</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;  1.7664  4.3770 -3.4375 -0.1343</span></span>
-<span><span class="co">#&gt;  1.7621  4.0245 -3.1681 -0.1304</span></span>
-<span><span class="co">#&gt;  1.6213  4.0500 -3.1483 -0.0796</span></span>
+<span><span class="co">#&gt;  2.5036  3.6613 -0.2827 -0.3227</span></span>
+<span><span class="co">#&gt;  2.2548  3.3180 -0.4614 -0.4464</span></span>
+<span><span class="co">#&gt;  2.2830  3.3477 -0.3081 -0.3362</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;  0.2308  0.4097  0.3595</span></span>
-<span><span class="co">#&gt;  0.1968  0.3787  0.4245</span></span>
+<span><span class="co">#&gt;  0.2415  0.2577  0.5008</span></span>
+<span><span class="co">#&gt;  0.2506  0.2512  0.4982</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">
@@ -673,15 +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; 0.01 *</span></span>
-<span><span class="co">#&gt; -8.6465</span></span>
-<span><span class="co">#&gt;  -9.1034</span></span>
+<span><span class="co">#&gt; -0.3686</span></span>
+<span><span class="co">#&gt; -0.3300</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; $nn_softmax.output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  0.3961  0.4251  0.1788</span></span>
-<span><span class="co">#&gt;  0.3951  0.4239  0.1810</span></span>
+<span><span class="co">#&gt;  0.2793  0.2349  0.4858</span></span>
+<span><span class="co">#&gt;  0.2875  0.2328  0.4798</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,3} ][ grad_fn = &lt;SoftmaxBackward0&gt; ]</span></span></code></pre></div>
 <p>The first linear layer that takes “Sepal” input
 (<code>"linear1"</code>) creates a 2x4 tensor (batch size 2, 4 units),
@@ -690,14 +689,14 @@ <h3 id="building-more-interesting-nns">Building more interesting NNs<a class="an
 <code class="sourceCode R"><span><span class="va">iris_module</span><span class="op">$</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; $output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  2.1381 -1.4562 -2.3068 -3.0385</span></span>
-<span><span class="co">#&gt;  2.0602 -1.6112 -2.1434 -2.9874</span></span>
+<span><span class="co">#&gt;  0.7996 -2.4717 -1.8110 -2.8502</span></span>
+<span><span class="co">#&gt;  0.5982 -2.1502 -1.6179 -2.5826</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,4} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span>
 <span><span class="va">iris_module</span><span class="op">$</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; $output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt; -0.4947 -0.6350  0.1133 -0.4768  0.0452</span></span>
-<span><span class="co">#&gt; -0.4947 -0.6350  0.1133 -0.4768  0.0452</span></span>
+<span><span class="co">#&gt;  0.2892  0.8388  0.0890 -0.1588  0.4280</span></span>
+<span><span class="co">#&gt;  0.2892  0.8388  0.0890 -0.1588  0.4280</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,5} ][ grad_fn = &lt;AddmmBackward0&gt; ]</span></span></code></pre></div>
 <p>We observe that the <code>po("nn_merge_cat")</code> concatenates
 these, as expected:</p>
@@ -705,8 +704,8 @@ <h3 id="building-more-interesting-nns">Building more interesting NNs<a class="an
 <code class="sourceCode R"><span><span class="va">iris_module</span><span class="op">$</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; $output</span></span>
 <span><span class="co">#&gt; torch_tensor</span></span>
-<span><span class="co">#&gt;  2.1381 -1.4562 -2.3068 -3.0385 -0.4947 -0.6350  0.1133 -0.4768  0.0452</span></span>
-<span><span class="co">#&gt;  2.0602 -1.6112 -2.1434 -2.9874 -0.4947 -0.6350  0.1133 -0.4768  0.0452</span></span>
+<span><span class="co">#&gt;  0.7996 -2.4717 -1.8110 -2.8502  0.2892  0.8388  0.0890 -0.1588  0.4280</span></span>
+<span><span class="co">#&gt;  0.5982 -2.1502 -1.6179 -2.5826  0.2892  0.8388  0.0890 -0.1588  0.4280</span></span>
 <span><span class="co">#&gt; [ CPUFloatType{2,9} ][ grad_fn = &lt;CatBackward0&gt; ]</span></span></code></pre></div>
 </li>
 </ol>
diff --git a/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-37-1.png b/dev/articles/internals_pipeop_torch_files/figure-html/unnamed-chunk-37-1.png
index dc7f345e4249e7a176e18bacf922f3f4d67ab8b5..cf3a7b7d7d34786972d2758687f6afefc88b2798 100644
GIT binary patch
delta 109200
zcmZ5|bx=^v_cw}yf`p{Bh=d>@-5@R9At@js-3?czTR=csy1To(ySux)-bH=BznS+h
z7{~G6yL<MWPp#9<@ENV}`8^!a84iKM3Dc+R$VhR=y{E(p?r1L)&NNcozA97P?r1+i
z^nRK6^|64=bG+{_v3wabM9GVPNb?~kc(@<WI?<oqpw=sezNgP<zESNeX?8m0zM73J
z_l#v&!|-FP*tPNR**PGj4^>2wpuG(8Jv`%L8{Os#dwo3Vz?^gGEhlUj0)>XjY*&UF
z@jIf#wx|1;&6+sgjZHhyCJVsN=`Y{RIPi=dU#UJJc1OK?jidp7@I5(*Uz;f)qI=Hr
zgr_c)Y$kq?L+bn@7UtY~Tk~Ve#!7GLlZ7?g{x8>Px%jt)btQ5Gn5(d#7W!YLk#W5%
zu}sK+7v%3Qok+V!PEPXeD>SEXzR)N2;__7HXXC>`bko7#UyR=7>=+$}-xBlCdJ$ns
z>CTN}QuuxsG7_&*bStnn{60GVJ*JoyjZf-19$~vDCgOO>f{}*(6`tcW$Kq76JX=Bu
z=XYhrq`Q+chlR557)Q)VZMAr5WFk%znHW<e%eL#Nz-J7T4bEfoJ%R4O%t8GsKzA>k
zsJ!?Lk+X`+M^xdIKYO=byt1W74(}Z!eS&&fgMzf~1X~4jvdodK`=fZRy+JJjwR3do
z^sz(YNv>)bry%70yc|Spj|=y>%UA~0IZ`fe_Z%zL#YKhT^xa82ainR%jzKx}i%*m9
zay2-8f-4Aeym}HoLV0WN-qBNWGvC5}o5D7~X~yM_zIpLHA;Mw%;w-Cf9XEO4MSn$~
z!Kv9*@NZbn0lo7o$CZlBE7*$)9P3>MT%FLxK9&3JVM@D>SMvbdL+qZeGii+@5E>r)
zA&^ySzrn0C{=Ez<s0QoD4*vFwB`O63_2>G2xDL;;ZId9`tc}BFrDu*NET>59SVp@K
z)0tFoZCt<0?8P@F&AhPocr2iCjfl=5=E;3&XcM8C*!bNUtIydz3tMWjKOa45jz_BN
zTVB}sv`1oCCs({i&sU|g1g6qYH-`?_B^OtQxs|H{H*-@4jBsk#R&-x1#<~dN!|+As
zPcQ1{_Na)V`>i1SvlV}Kf-=lkfA;!eKr77MNgg_8aoKi*Z(}6SDw?ex!)fm>C+9^@
zCRH&$E|Zzv+1CQo(AuV~IVw_S+dB!P&ysU*>g4K9p1?5TBCUp(EJ^1`WO1)HX1I-)
zI~?^N-~3Xk3V)MZ+h-0}U5QlE`jrT)h7bB|&~ldK@}s<*MQhoerM*u?|4xadLnMgS
z%2>me@~Y1DSIn_J9QQSTKvY>)qv-t#C!l=UNBDCT9xxtlbXqY)Bv8Hmv=_nubg~XZ
zP|3dUSM|1hV78;L6$f{H`}ofZvD^A?zB(>Rs;M<AvfmU40qk6oLvelKREjB(?9kPO
z=DPxGG7{=Yehw%%k+O@1=}jOt53Nwnz@S+tS+UUSim#pNs24NIY*K#ySIt_7G~dkZ
z8nWtZVw1{euShTHr+C&sTua<twZ0Zs`_$$YEAj;CcZMByTy#M(E+r^(VgUycH>LY3
z;q#s%#65Z_rw0WguvXaOszP&SJqmL~?V>)HPCpWV;5!7@6CjSWX3@z!?5Tl)c?lyT
zz$-r%bmtyWgLqWO@xCe5#4!8xj+U0zYu>6&CWEOM?<hfot(Yvln2-+s>j_VHR`=nt
zrFomw@@L)t;xQ{_cBTO;;VxdYPud&qbKiPiw+bp+dXz*9>C|_(L~lToW^ZwPw6@Rl
z8Q|;i?i&fls_q<tlJXC+<rJ>-+?eiBha8cB3cjgal!M)fs(xJDv3Y{~Pf297Y>)a=
z@2Gcuzv=phK;v*#J7ibtjh`QnT6+MFRa5_P6xcmb_wNTrkzG6`wFk2h3v4K+Y#)qh
z&5edp_##6QdN=v`X6bfut+Da*%adQ4g=Uj~zeItBZ4%8cbiQ^edoVw-BT{4a;ziWp
z?<Nv8?tFjHH~0I0VN+&R!+$UE!C}BMozLr;D>A*fx<C8<a&_413}x_PzA^rfq%mN+
z-<sGjZ`i#!Xf{e8iHdrwe*f}kFaI@Di1;?;OIELQ_2CQbhsg7G<_v4F$Uk1-@hV!r
z#!|Fp9WfH9+T~xT+n3p7F_alv!Lpi#h+t?mZYY1Y^$){21)Vm}9j{FaLe-PU>HdeE
z5;7R3YTN;%;}5Uk2q=Gcjb!mepuAmbT^S0KpK%jRQDm^6#<+YI+g#v+_hf&j7drVZ
zm;AaF6s0Y4E~CMuXJCy`zs`J8iC)+9`+E5)iCcZ&Q^0TUQ)Cy0GI+RSFJYynrQh4y
zVy&)X*`d&8<%BFIJ$P4?p}C-z%4}*r6z?4!VU*nv9UCd6Valx7+ar@fS=tqo5r7BN
z@pK?eW)ik*d^qNoBtG@79ZGv3ZO#5W{lVM&;)M5OwEx?~C`bZUc`we@jxnz~4Fu*o
z66CBgdsAoyGp?^HLgg@mTXXLx(Y&KHcfy~eILtQY_K>vzLSKVDMJKZI>psQ+Q^c?K
zcFyG83>@e8ET^F;nxbwr?WT*JDZ8h~SANAm1rQJsAnA<vI=@?HGq}MFSSVPVYwI5`
z<~v$mS;=?3;FQE47vMXlMnE`}#&{g?4GWP<b<&!!S%<c}DI@OIsgHZzcv~H&T|6+|
zJePxQ8rM;QHKJC$CkIq%Xe8RlVjknixHWdWZKtOUX!omV-`|n<<<AFZezgC9ikONF
z>9#IQ8sC0-KISGF93JAWVk9KX#~1+{bhyQ0Wv?59t=Z$0ISPew(vtfHX%KZBHQ||q
z!}XfHSz7r=&}j3+l5cd6FC3Z$rayH>wTQ3l^hjy;_B-XqNAzg(nNcz4X?!@&F<M?*
zOJsh>XQVBn#moK1rvQKQgh5WosqoV#ByCwpPkA6ogrCEH{NN?~`6rLVW!<z8R7cz}
zQ3)_ZC==*!(&Z^}nT_~eZE=0ddgUp`V)X~}KTdU0Nc=g6tsMw|fPuH6^<q6)Z45``
zbj;iC9nSsDJIWImPo0f*AJH6MrC(B(&5uoEXF6Bhx>)mFnDhhhI<otq(!pYdwJ4gp
zU%%S%x_=D^`xk$fr(_DK5>J*{BR4mkM9&Fkqd;K8c`ERDXJs=uOaPwf(^g-+<!Lh^
z_eYn+?ktC+L(5+`)^?_ku$j(cqEAx4^yz1eS)#VCBL9|^`4KI$eUjuJqt+FG&SSND
zZW>3R5-*q~C7s86l<EAe1nTA2(fm9T!s1h5>D`@d-fes`<#(o6?}(xPK=K4v+rU7r
zLm^x$O@pMuu_0)yZS2$9fJPp^#|0ZW>jYnsxxAG09R3^-CLzVYQ$aoy9iI81Lv^XW
zrz5e)yTbosA0Y>>#wW~*LEaDv#xMYjBgnRxfGggD$ZH+~%|#YX$NtQt*8P!^@rt&t
zm_(q?WD@u1WKx}-$q+Dxlg5g8mR7lv(<=~v%W4@fyie+tdS`#>3MV<`6C6pcHP*+l
zLq}v?d9}BdIkaKFbih3Qgxf8_2lGbr!i{(cWw5+Uz@Hj_&sBY{zTMxEnkaE?C<}+{
z?iI5s_h(2idH>SAw8RzDT9<*5F;%2XavkA_?W#?!#u8&$53SsU8%F1jC5ygmNUZMp
zT~mL%`+U<}aGbeZ9zO3?-76|mU$6#C^V=F7n12uAvwy=;-c6M_;@oogGmIn89iEAB
z55LYQ>%JHKaz>8aCBDogb0oAZ(LvDc3qeOiZJVE}WM&e(;u?}zfpIf9+*_>7Cn_^6
ztvyZ@B&&};X|M2!HzQ&7;oHsqqFs}GO}~xMiWTVpYbnSnI{0kHZDg{=Ppc?!VLBV-
zqJj;SQ9<{af>QNuL*8#$i=A;p)AJkI+E0QD?Ll`WRHSGOR#af3tA!0K6GF_VH2yLB
zaFyRVkzKEzJa+iF;u}<D@tr|+?YE-Quk%A;1^JZwDI;(a3YYii7bfMKmgB^*Dw4S*
znY6Yp!gSdI+h!-mZPvKv@+Zu<${e_I9Al86nl(l0=*jxsp&l5Hx`8F;1_F0VrHQh5
z=7;@?ulU`>2=LGy+#nJthd)Jj=x(J)nLX^LpAD4zO{#Ho_h@YM`G?Z%i{$PC8{W^#
z<^$h8*AFO2S?#O`>H1Q2ceUEBeCCrfdnEr`HZ&|)#Um_y=3IX+r9z4Ytx2-z%#V0E
zxh>)i4W$9mRV}_}?#rOGIpyzOX?B=Ol%Uf-k#R|GxQD^e6PXr(M>{5#AZ^n*nD_)o
z=MnXk#G#~kolgdhxkz0f#AsCIvM6(|Zn<D4gD&z7qTMYm4aR-+YU221FlwXD<TIAS
zdmuPL8H2&4H~3st5l)08Rvhg9s!!_osCeGmUAq|rEC|E6$M9Hiz{2{30sma@*&7*N
z(m*8)f->mg-KVMVr_=jakL98rht0TRx{p8Gnw(d%i_wOS_IpRho2F_FF*mE7tOT8@
zTD91*RerBVYtttcwp@&OvwflcuHs{de;lV<LQt(rj{@TTq>`s*`~0^x`NoddJ^zYz
zlQv_g>NyDT-(dRDqg=v|4H-!s8!U@Zm@=ewA>gS_N<xi;E=G=P7YTrXqMJUd^SY|3
z8jXwDLKgP5Z!0$=|DFS`t@BSgJpGVJ49WG8_qdEENp`vLg3jI4x$1{)jkUz;<jxXq
z$V>Y-TCcS9OHpn05SdM9SsiTKnkO;UU1@N95(G3R#Uvf^rLDcy_twTjq6wwWE1;bi
z)_fleslMEWDoSmWt!81Wpu-(u+WD$0rNBIgZ(osyEKKN7#e}6m6@5-?33=h@TaL;j
zjUg^hZExE;h<j7<{J#C9Q~txy>Dp^IXRy+BC;JF18FP-0l!CJ7WA18kP2_hb3h^l>
zrOtX6%r61t@h&kElTCzZELOe~^;B0~FFB5%94)V{XoG$YT2;-OOvdo=8II6?!rpk1
z16yC+O~7mgA%@+om1&Nbo2ZA_k*ojBIeWwz)H0GpB@$Tv-m=*PYEG9ojEOpKcG`co
zd4<suslu*fy2krq9|v%sg8afPpQ$&k<IVnK*w>Bu2YLAx(1`G(%N^qF?p5t2FiY2&
z&Y!k&vZ0INdEutZ{~0fv|DbE4Z%f-WU(SH`Mt?b<L$-P~%m;s@JpaODEUPVCn9*En
z%9t+D27acCxoaT39i2yd%B}k4NO{K9mM)kWv-jlX2k(l<UrV~?`g-g8G4_ZV?wRp0
z^ouH4;Bjy)Zop*y6FDO#d|cbc_G?WfwJ-!KY+8D>rVsuB=}bto98r#TTBji<0M(kG
zEQ+8lk#zKmH%*Q{UAv59G#<naQ?X4!zi$&WWPg#JH)S#f0?tOqi^z#@q6!IWo_fqM
z$!TN?LSG!kq^P3i`iCnUD|V+LIc>Gi1x6wVGGWX?!CwYAalUlizij&l+DfS;*|^v+
zt0MY5-`cC~L8HAQ24<$VG`YJd8vZ{^BlO`fs!v47kf#K2M?_3SU;tCmK+G-XII(FY
z+weSV)o!(~<HJ2XIa^*|l-Z7I+PMo4)o>;h90-14>v;~L@)XJ@_G2z<Y43awFcJd;
zlsAcq#)8nLHkAkur_n<W-p3D_go_D>Ll`?|iz!ti;@dCP$#4!$oBZNZc*+f?PVYP6
z-a=Vxc5n{XRHR|Rr>r9&<`>vCB>A9FPn|ZtH&R}hyze>0?*GtTW(*J<WJ4D}_|!uF
z5;cmBYa)THC|{F|+(a|Qc}xRT>KmwEd&TD|zY?r)l9Eudp|q_K=?bjhxl+WC&Y<Y-
zTR3dUt0>lg`tgknX(|%{`bArxRf6oX<*2Y>LQMPQ`jWojh{>@!m?PpTzMWqksYh!|
zAHkipA!<aI!|KluCzxJco<u#rsy%^HDA+?tXnxCP71-I*%85(bl5vbt>$XkOiir49
zTlo8AJcbqGX(~_C>zXaU{{&Rbe>#Xa<!I^_zgmrbh^l^>VtNCftt}OdNfmN#_TVwS
z{Wb=3?)49Ezp6LgeQ!!tO0z>qXhR5zz9$4fC5+>4$*meT`yT9r6A5)Ze?80YCYrNQ
zKc@3Rs{AP>W?(?o;-n?&sx4tUljr9Yf#-HN6v4bsPq~k&0TB}@>zEtLZ?b>YyZ7kR
z`nl4cs*;N6!}uQ8?Ok5oa7hVp=2;abCFmFnUXgRfHLL1B;VE;I6ZXBuVLFf?tEof{
zNVRpYHl1;J38?nI6*M)lfgqiAl10Vc9JF1m3}{6CaQ9*DQl9#u{(yf0plj1y#XTi0
zFx|#8GDgqZMFU^2X3`EsoE6=a(T<ndkjl+zN<_Ho9A`>BIhZpaH58PHs5gkq*+2sn
zO}=CaJ^jznBp3yx!uVW^oA)YWP`jS7xD{)qHyFza3IlveM_W&1%9LqpzQiiq3lL4f
z5*`s<28P3G4pH@0s1rTtPc-S1Z7bd#CxYfoV${xXeYEr`+Pe}pvKp+NKU=)%ImB81
z{My`g&Fj&Qs_&k!$n>$Ml%&wZaAst{pzQmi7IjPyc8^@V11fz={IChyAYauiV$$7s
zHAwkKS6I`W)oIY#`BFrm_Y)+&(KOZm7Y1D?_?h3sa!9xduB>_?qOqWy)Gtyuxaezk
zST#aia9oaa1-G?R=hmzCheuWxW(>Z_OITs$eR*d;B0s0KqL#aGH+a?R4yfulYzC0}
zJ1^SQ7-HPl&v{-biSAztpnM}l4c&J<%fY3m<L_#w4=Nj-19Kc62r9MzBA=ri679@9
z-S$s`d1UGoUtGSm5{Lp6x9Hm~>5LKa=(5yd3)(()AC(;9(@8pn!$JcNLqfOIoys*)
z105?TcCJe{*wo$kM#jA77X&bzH<uw!qd!AqJKD_rUomfz?V1QPI0y1TrnCHFT=-Ht
zp9K7F<!}xHc|Z-Zxmc6)yp2=wtjZ${s$(VkIuxxR8i}E?dk!`W-PM&GC^66QUH0HC
z)<>+~M&7VqTvS-&-@UofoVD!ZpBB%K*<e4wyR0e3f#pY6q;`P4Yl41O3Ej`;C&d5S
zqaoeKh==kd#A*&^`kEc>+ex|7XFxu~*F1Vir-Zz)u+g*McA`qHI#GE-B&?Hn3Rp>@
z*~Sz4P?Ao3Jfg+6g_WQO?~Jdrl&9Fj*hd>XcTa5|?=>(OFld`*AdNv*M}&Ow*rVS>
zJw$U-Z6dy%;n%rdTuoQwdRq-?9J^?7sgNAe_vk<y!|q-hcZgucZYPfk^n>m`rj}=p
zcBfIfwQ@>HI&xms;)>X+bX+Zjzh3Pm4!k~p4bBa=o@>Bp{;>(GbtPZzFAg6-cqRkm
zq_TL;!Y`N(4bI+i#os>tnsOBPiiEnNo9PcQ_I~x>goZnnJc+#MAS+fOk%Y{mJoY*}
z09c!0AfrCTb>fD+joK8ozAV*elR}@2wG|~W=jH$?(Eef~Whnk`3fubS^>wckbFPiY
zh1%>LW);>684o#Z-sp^j-X-=~Ow-*y(d!mrt)*3oD>|ZxDj}A=gyQuxN?*w~AeD*|
zM7+Rm=snR)7{DFcs?d~y@Lq%)5n5frS;rnz!|>2Xd&suJxhgoBwkj1HkNG9<zb}RQ
zgVTSzoZ)MNqvFAXeKZ#G4P#lC;b0|bz2EAbOUL={Iv{@w+uUF{lDTR1XH_d40bzp3
zSz5cem|j<uR3`!L{k|V)lK;2!gNY_m+~*7LMs0hiq5Uh505g`=W?I_k7FsMDPgy#;
zM2W#dB&4xeo_jqtj-2|wKJQk}V%kE^6WiQzGJDmg$imOBQ+u-nN3$!@F6pN`H@L8^
zHtyC1A0CWv5eHP4LRm0+*hJ<79~(QAJ$5xpOQh;fvWg#Y-gnz`SsSk^S^pGUO0t$9
zYjg}{7M<*~89ei6E!IH-M4B{^t@=vo@W}GZtUiJsv#Ddx(C-;#<E^rV)sgo=S;<if
z?9}_sv^<kiwL!N|M<3dFVbt;7mUfpP;G_=#1epqH4=h--1X|fnT^Yv>#mM2!@jVug
zyy4QhJv#*t2OSe;M~If<sB;D~;@E^-0NaHx&EXr^IPUy&Kn^rXy|SPF!LdoHIqXL*
zlUCW9bn#r#wdCjhWuPyXHW!ekV%=ljVD#B^1uDBc9ytcFaFb`N5xQnYE;|v;Gpv%+
z0ZS=?EJgVXkt{rEd;8>j(Hwwo9o=3H;6^U3GRW`Y22kUhA8bOx%^ovq>|0`6lCZu3
zY&OXT@w1+>ghNh*fb-kpOUc^5d>~x9G?D0bTR|BdT=2%s`5Z6SF=%<rY!%gN<sCSx
z5$v@lWewtvD>}H$CRTIP0KupsbE+KsN1qK<2{EYJ0aEbXX@_F$`$L1t%%a(fKW*UF
z=nnyb02S%UHYRYm&C%sNJm!dDFLblIz53vdL=G@H6&`$y8s1>;73!o7-uH|O1&IhJ
zO~aC?4mqBM0_?S;vD7`*>LdNd#HRL+UV$9Ijtv8ymK2}iUOk+-<oT|mUP>F>BH;NW
zKEe&aRj|vH@=2lkD6eR?$zHuqZ-z6ohuTSp;XKD#(v{^#Gel5eUh#dW{Fzn=@RTF>
zG3(4_(9&XL)1z}!@>j3Dm)#AXJqyUUF;9uKQ(`|M^l@4mM9IG@OV6iTxAtIV-c3D2
z*y4!cbV3DRyFeWAmtj4TLH_UH69cmO%=w({o_tC~$i8`o6ae8IoE!P4C%etdqG`jt
zE3wTmr$IMTf(b!O^i%KZ>$+zqU<u_jeRn?TIR-fx!UlW`AB~0kIleq9Y^l9cBV-q+
zez-ICM^)te8SoEb-I8Dk5_2#F3@5XScT*P>+P+t49VupNc&}>R|51y#<7zeBEB7Tf
zOb(=V5LS;13oCEPRN3g;C-+P9lGA1&_A1+dEPbCaPv#>9J3+>qgU|ZOTDoiQ01Z)c
zojt<3A7u7F{zIS0au^=(>$uK$Z^Ibc>_Wps%+I2}js6JVaU3d{dvS>jq4AT1;$|1K
zs(K}2#^O9p!0fdl_sXd;DljsHO5luC$wSa}z&LNL@0Y)GKr`!Gl6H=(T<<-~?jCDz
zt=-Q0W@S~#ns0*X>Q#zrecH~0JWy0@0a{YRF(Lc&_`sW#iF+#R%`Wy&tF(~^;@TW-
zOk)Z|Elm0_ZgxyG&`mzLO5$!D5XW#-L|BoZ9bR>!T+A6kej9Y@554>q=*h=Cm8aMN
zD2@Uh1Q=&0y_h(~@L%<~)y1l3!8~w<XGV8%;5ktZ>WAXQ%GM5O9O5cla}u!0^un7p
zctl-I;Z#VhZqVK!1V`LH{9PecM(=??dxB_G8zmy^i~*y9E@m8C(~qnABNj{8@frzv
z_uR*g19EMw;@7-)uQr(say4D<R~EJ8*S~Gt{0%d=^VVVQsbwKUzNM+qmew9%c2Ay3
z<m?3kTMZr58Sho?vf@0EGxL>2t~$be$Cvoux%1yXhViB-4i$_&IW<*pC`sa3<T|IW
zIehIIOGdZ-EHD>Il<R%A+@8!i0W{tn*3@-ky1%-nhq_pY^);~O*bI&5!?R+~;yv!4
zNlv2{OXs*zXD%XnoODVqB7RkYUTG}gU)W_U1!jbT*zBjZ%RHEU0~Sv}vlqvcISfzc
zpqh>z@kPaX=H9(aZ^SfY`XxY??(7M6Dv<`V{=fk5$;?yJN-}p;aQ=ypEf0=~v3g$c
zOWMNac=c%=U;8ocvTqResKb;9`H3CX9z(iRZW>bo2^maNuIi1nU3z2ED=kRSA`>WE
zY+Ksb9Mq-L2fm3-P?I%`kW`Gu=Gd{*GyI^~B)o++J5&Pa_jxL`+bSO<Rqz3F^Q@6`
zsj=kr7Lss={R#oQqrFR%#;!W|UHsQMeT<bT#+%wDC5SFIcB`(jy*u1%c3iA3f*EK_
zXFO3=FhY8zj2pK+gmKVQz#+kREhVuCJ$K$!t}yz8@6vnzJN915WG(kHAjR9azqj1;
zzOXp*7PKmfY?k15=a&U(8EAKeGsQzd@)u*0cY<~UROmkqGc4<oS<OcmV63kFq`Ey`
zooH5L6W|Kc16`v+w(Sd2K;(I+FbuxREh{ihR-(wSr)_(8E}!Hi&`F7Zl+GQ8kJa}J
zJa%5UD3sG-J)cPeT3#1nHpO|KQ>=SG3=na}_K4AU*{;{)PPUb=NQF`%=4a71_+wbG
zyBr=WNhwX^9@dV>RtSiPA6Z?VZDjc9=E>YwWMcS#>KDQN!TBI1kqjn$=Z%en(HPdT
z^Qu0K>!oEY4{~;<NMH5EW2ER3(%nnC+v{SUi43~408pz-XIoTi<26sv<q&6Ytr6i=
znw(mX3F)EWZzw#%U}HG2ZOMs`<s&z%Y>$PJof)t&JTmt38dGZZBNsPAu=OD?ResvU
z@27=pM-9t*({<le02~<@=`+8WA5&JIptCwibuvW?iH~TK67wtFSiiY8Q)XkKQVO7k
z7f-T?c$!SsR`6w{ZN>NA4|D(DE<rNFKxsW#OWgy0QPPd_4X*O8iO}MtHViPhS!?&s
zM02Higj!R(FSOwUY9`U`mRO_bP8OZts4@c(WHAI0sIHb)PO-)utw;j>f5wh5Otj_5
zj5_J8-{&)_aN$xC%{`*cpi}Y<P_Xs@+u0o-k2Jr&ssDc9;JP~xG8!Sa8R;1A4yh2c
zTbeD$9B6<aXH3JvK64h56yhr=06wo8Ppv1lRdu<oF(!aa)m7HmlDSBGY+ifuMk3x7
z9?@sgpr+JrUnx+s`>VTuOL$tv{Fa1@A?KRy&!jaZ{!b1IWVJG8@pyc}xWV}V$Xzu@
z_9u`_Bmj3YoY(L*yGLOco7!9RonAcWg@zq?+Y^_tX)Ha9pbEz;OIoLUVFO0Jvwu<m
z<9%kPH9@y(w)*lm^NJbyjt4e}cxW}JVJHi>&V?m{s<x|Qu?ip2s}<}(OMi3J_AzTn
zQ{Rk<2749LYKmR>(oBJ7*z}cKV?zt%=8+Xxq<tWN-wuJ;#ddO~I$l}U(p}1JSi(DC
zgkg)I&O5NV&#Vr2&^aA8&Q>|YQg@e-17m;UB)EqwVY0a}tH(%uYxMMZiofi3>A08l
z{-b47{~Jz$<<_IvUmbHi<tc)U{Jw!{@V)n3QRrh>D{y8(SwU_4ZD|EMGqMkz%Yrco
zKgin8&(O7|u3B%hg?~$mPmvh&66KQL(U7OaY$A#m$|FV}uehbUMmAm@4e(2Dn=9uK
z${Fp?bRlv?a{cZ~a_iOlUiHn{`ELfglKGJ{MY=_wv55-PH)(;}FX^@!eYDc<2zB&C
zs%n!Q47w3l3?W}l_AS53den$j7b^Lr=rdHl>oLoXW39g`dZOfyf2uP<@C$4}DRh<)
zR_K7>(cBsNGx-BJr$*{WgJ1~J_Fjf#&)#~vkDG^}CwBC8wTw1k!t0-1?Utx2hL4_d
z-ZpX)LRy4hIu4qJyn>@*z9xye;K$5BbaW-@KOHs`;Ut{xFJ>^oxPMx(00@MPva<OZ
z6Xl*0CBTjMZW=x=;0#LX%zjfmL?TaRFgr>9zlCn^>3!sjTa1~)L{Hlp8R`Mxn#YmC
z<1kSVw;~SY-dzRmh?D@>R4is-u4!p&CysT_-@SsuR07=m;E`zvF*`%$9`EhMWcwqm
z{V~hZk^bwX!pc*aJ5|xKxNc!qM+hsOk^#I%2L3*gy1G9VH-MBrae0)c#s+;qE*7x%
z@c8vnSuT&)%F`c;v{gT0br%~}!3r-0`VZ)dlrdCWw#axKYyUk1`oD*iJ2Jw)yiMil
z?YUf!C1+YiaG1zoG+30|Qq0t-m~R81;W>e8&3H-i&qs+(c~+b4$l^V3BRO}_fuqic
zB>#poS~RD%rRHgHLPIk7baP|vHvHO+>UB>j4EUFGvJd|r#-IO<c5h@Z&w>TW>pn0-
zu;-=sHLho(e+bX-@C08Yaz$Z0D#fheWKBv2<1xP+9<KEZ;hJ!_cV$QKbLd_J(`9^1
z_$ku(dzERM+N*UzWK+!^cpm)L`rnqPE0|1a{6Sg*o^aSTb?|j5nrMQ=$HO~kc0rRT
zw^*bcMkR@<V&!?pO>dtSs}w^e#&wpU!QT;VM$x>#Z}oig$WnR8<xIz~h>tNAux&TO
zNN732#GJx^%Vw-I+4GSqrPtBU#x{pD`N?2-D7wuq2YV1($5+R|?X+pl{U#|rQ3Y<5
z`zZ?4-bAF9dKF65-%boS`Cpa*SJfN{&a?rh-*bQePBnvg<)z~jz!II0r7f=|Z-tR;
zZ)1g!^ezoF!v~RZhkcCuBOe6>|91=$WV0go2fbio&MyQDhqSvQK(?iJ)xnY>be+MN
z{sgZi{X1osSzBca{_G7RWyu4VG~?u@b16E;5*nLo6_fWMck%8zHuTrTC^`2k)*e(x
zpN>m~rI5foG<y8v_yya>LCpCIsq|0F7X1`~1u(Q=#2qh^RLX9=?_b2d_P?6+0_mG5
zLN`JZ2Da}tn#iP7OqJpm?CK8y+q7i(1bd%VVLJ(>%q=!Pv@AT|@Y{f~j?MQ6YZT5l
z3S{aBd8Tdk{qBfR8nWeTbG(8=YC4%zUBM$&cqL#~eZSiqlCk#C^}V{|L2>btNL4>e
zP*BVGvlJ))TZ%;4413*O(WonH5`gM}GBB@K@9C6lD;`*0A4lp|?)(Wy4{Nkr;}Bl4
zG;l=OW~26IT@ZNL7GGwAfPF)2S8Z|wjTRF!r%{Z-uLNP2_yO&$i2Di_q<v_1J~qbD
z!iA#QWDEmjMbZpW-=|Du&)HN;w_1MM)kcg|{TZZk%RiXztHzuCzM(qI?Jh;5#INn=
zYO)^RB*ok;adA#^`Ns-7iN7W2=#nG>%pSy!&}<xdC=g7MYJNdC6GR|;ld+*%-Zf{3
z2V){sn|799K@BIqbiFJ1y4f={<b_{JtekE*MP{VFAM)GVjn_mqbiQC{8N2ks1K?tQ
zMCj|0MDF?59zk{V&1X>k;8%n2{w`SRpmAyrguXHtIR0#2y4rtC%<Wl#_%CKNBp1k2
za&b+!=*d&=rVDkV2-eZT)^7z;J=*Q%Zw!zEvML4#Nx?gW1^X6wkVwYOob>8^`s@k1
zv?*M+{#qq%<o=;>m~u+`(-U0AwWE@6+mriD(Mq(h4+jZf(eAvl$Odrio*6r{pX9Al
z{+0c_{_6k~VWNUiG>x0VkS=PlmJ+LrF*X1lA0&L6hP%-?$XTj6j#C`Ab`2Oy_H-6T
zj$C2JE6M~@M&&6Yz<PogSQ`x(tm~yEeCsJh<{!rx`#eaBgkqlpTZ-t_8=8>pZpLhr
zSo4fMk*i&PqY+>0Kf9DEeZ}4J$0hR?Rq9K^flfH1MyzpG&<HFDV05~C_u!4SKNeH-
zTWTA7Sour}f@}ofC*VzTHe~-cu>VASft=5}OEB+UB^^DiR@Raqd@?kZ*;N*l@qSt-
zX)I6i<&6(3Yt_Q`=nFij2!t^OIr<?v?@F7n<>nW~fN-;x%hn}e<HABF5)ps~fKuud
z`N!T86%G4~nP9#YjDc=Voom39?^YxnsRN=L6P0>&uJC0)USP{;yVSt(H2e_O0B%xt
zBa<RiZ0-*dC|7rmi0Uc3)1%njeWEdmo*aWY7m6gZS`nZhGX3>&-6?4Q8Pwh?88S!E
zsll`2WC8V8By`WLstQ^3?|5arS?kchf>CaFx4=#S41`ZZn?YcrKCZ9jg?}3xrEQ=s
z2!uf}%AeLa2KHyIjh}9<0fmPE$RdMzwT7sUJDi%`u>#{IM=&p<r<VH5sZteb022Qo
z^@Mb{)gth^BpkTDDPupLR<lA|s^a2Zb&DVWw%-4HH<290q5Lm+x$uusi=j8Mtx&2`
z9IzAa2^jDYG)x?)K|sxvIlkBNC*X(%#jW-zks^K%hTfH2j@dr*<+EIV{IHDn<?*o)
zx2qK+y5_AnDC@zj)=Z4~s@T3+hnVSYabR{qi973$%N<W>ZOzqPCE3TMvKOXLAKz2?
zXr+e^b{jMRBN(OMPY3|2u>LnAya(?kt#o;{**bsVxUB@zw)ecAKf{4}MsRn?)kf*t
zsB{!ZR&U1<9vJ5z<K&5~gN<bdvW4>F<rn>SoqNZzh0kI6<y<cyM`ow8Yw2Cwticko
z+qZESPbwxNiwT&57K(1t$1J4_R%yP1FjHPF?a}`Cmul6Q=!BL&D5*#!99fG_FaH*D
zCdt1rm5dRS;Vc#q1o|;o3-+>tfnOTVrl|C*FPJF)H7Vs221Fy^fUL(Yof&`Z9pKt}
zS7ZU{tap<rSJ{Vsi37cv{?n!{$Y{*bI!5mQ_)R~i90rK@M@F5u_X4-`XF3z1?#E%9
zwHAN`-X|t$daf+DK%UK&JqI`Q53fMI4EZ~#+|x&9RWHezPkWQwyz+0{0$;nsO|y`3
zP=|BHqmUgRlQH7La%(s|gGwrogKF76;$&a#a=~0ymp?$KTC*HkAFL+Y5$Z|He;Kb?
z=a8mWprE@}nF7q1`B}Z!AWZTxSGhz;EFPQX(`UV|Z#!8PHcQ%=3UB@#Budf$9V9ut
z7yB)hh~pe1<P2|Oc8F|S?OFs9-B5rYdgt`}KJEtvGpMfSTZEGX-T_iebbHY*!)C*a
zaaQ;fMS(w)m2r!SymXdw9Y-cC=eK}?ZVAE*+9dqPoiZOWVgm_`>I6V|L2EjZK4<!d
z1&HX*p1piB7`EA}DvuTZ%HEwX|H+x&Dt&{KaeHN6g`ZEC@Okos@m%XHp{6}0Miq4N
z9Q6T%0FXj_MN73}>yCgTMKr5kiOA9Bc-BKSUxoBFl`DRrN+=-VX#hJ<LrWbuDX6U`
zqP&V6Bw#y|lfJ<M>EZ%Q(_A~kK2s^Q`~zwL=aaez6Dr#%4gMA{!!r8cxS;*Z^bx~U
zTD3kD4$}e1rQZH+_BJp;e-J`$Khs~{M~3O?6Lha;o%e5eYnAPC#e3Ah8bv&m&KPE8
z*^*1ZRabph9@|&P=RmpHdu_Kl5K-U(YrAVABH~X>%v?<`dT#@}qlSt9ivxMX`^M^U
zf#*SIeIR<Zy!?57eQ-bSYCAj?2q@5Sva<NwSn%;w^3jnw9QXDqW~ZEt-D;hjl;E$4
z%243a0^t!~jgvA)>Ti>v`LBV&N0r2)&nMeeCk`c0AW%xQ@Ze@Ksr*j=q818_37BD<
z^|(J3A<*#`wRj^`SgbJexL(4I(Zc_TkM>Jk2l+R70F8|aw&;$7m0oZ;IC`wTY42Ow
z_NjZqS{Mtu+Xvqu;JSa>n)gjpf@eP^^s)1*1wlrHo+JtcCED`^=p5zy!=MlhB0*OT
zWWY~>R4?L1*tF5F^8(N)%-Xa{GLYjQYO@81*|srP^CcTGT)?*nrKh=!CL@thV2$u#
zp-G-fSy0xuQB*`K4thBqX={&k%6-C4(ZEObqYWj6okIWuFu;6#gQ+RvycP#?0+caX
zl<w+v^&M`i<@T%Zd<vd_)=kwiS#$S`Cf{*-mIAabAPS6FNB-Fh1);wuqF)RMaYgYv
zBF4)S?=RYz1Z60ow{Kr^v-`(~_s)-*%`Ie%aqF%p<%K&D3d*n^u9;n5_TVoj^82X=
z1tBkWUZ8KQ-C^zqsM0xXzM8Zn${gcT2ee~Ev-josbb<C5U|tmtD{sB*ZoUJivlQ<V
z(Xf>IJHs6_1chBNZK*zRB>QHU3vKC@uX>Ol0_pj+MJaT1$_A3fOC1^U^FtXBlDrB#
zij)~m9Jhe8HxY<Jo%EtxoxFnq&s0$S;`pV%8UE`pDz5{=>+o5JLo0?4PdGe$B1w^Z
zG3Ry@=mkCDiu5+bug!>loZW$Lf$DgIPmXNdIBty#hG56K2*W#Ox-4Lc-Ot<1@aq79
zKl%5S)xD|!k0BzVfeq;lCdOU*V9*M+0hDjWLIxs7FqU1cH)2(}r9E4uNZGm80V7Do
z9^H;Pb#h9938V?_<JVR}kA0E-4X1Tq^-4JD81DEcu7gzMOOQSq3(5mbIn?=`$%`a%
z34IX2?W&GalEEmwJ7vqNi-Eo+4t=Y7`pm)|d34_R5gg@*UN5l?bJpPEq=Rf_&M+|B
zfO5#S8qVy<)mHhEoXU-M4nnJgjeY=dpD0Me%}*DTlPS_rcs&06dgc|S0qt1e?FW!I
znRWUCvlJiQrPEiVWTbdxyUq07FAc?FY1zz?#U4|gm<4upu16^pQYqf?0J1hPv_oUD
z{CrYeRh8XiXZ=B7%GO0ViLPJKGVIJ2xY`+G-)&>Hx4IQDmY22kmV}08`{d@FUV!=b
z;zeX|Xz+W!^*4=@W{aM4wAq^6LHxBmhvRm%ItqD<Sq?=3g)-KHb>s9*OgZM<e{h(u
z^c&yAY3LVJ4pLO?w%hHhW;HG62TMkk_%$Y6>O+{rguaPCVq*gm5{Tj84mpDE3ahI*
z5fFS&u^H*<`I0im$EsW0s99LBDI96Np-S^jm^}%cuZa}Eh7jsYQ_|TOP98o0F-5iO
zExF!nI&2J&RqmPd^R`8*&weeKSo;mdS|E7S01c-S*2I0hw?|UoDG(C$nJtV@ih$QT
zXx4lO*hNsfeCAbe3CK2suR<v{{I&RipMcjTsC+9jf1iE2vlTdPh)RT4Ql=BYW$ALF
zMVhyqm+9WY>j$y+-3fdJ3<6?uXz9F<R`y49NB=?I{<Ny5f=)|{n27I6OK<?x-q)w!
z4d0;5%1CJ!2=bNY%WdReCNmijm0~G|obo7h&aH7lA|u6NE@GX1Vau`&8*906Wd;k3
z)pvJOH}-bbZGD=U6gdq}u3Jw_d(WSLWhPl&WLP#~^dw7J6$wIFSkFxJuGsXeaKr2~
ziFES_wec|Q>IEd~Topu^AvhK4NyA%G?{tkL&iI<OlkDubg@uJM`T6<j3h~6))GF?E
zbv)6`CYks6S0UN<ZRGvoX@(`{;BY(B#C1=kU^L(fY+yj1E&};X;Ho;F<6qWQV}_~l
z?_VrkR~wJItf>@PlYMh_%C|kxZjSlt#A5dVm}|vEZQg9R&@Yt=^G{cXwmEgsQUBh2
zc;aeq0UVZ7gL&Hh-{b80^o7bmmlt43XQsV3w|;dDng<*5H1LEirLK5F)l1M^SRQ3q
zRI54}z4Pc(|8C=n5-psovKwWIk=f{y9CQ~?2_XQ08unV{enqI~o)7=}bnWW&_-mVH
zjOG#;Y>;rAD?HF-zoWP|<rJxQVP{jsoI4-@pQLz$K7BH}D#fP~we!@@RS3*^yH26W
zz$m*ufXO`gNeCTvXzy<_2su-3X<kM(I<vi#ydO|D{|JtZ-jN2bx1^GHtdpuER}owQ
z?}SoXc^QsKns5Y6zHSdW#_CEK#D1#SPFk9G^N=5%>awm9NiM69X!rBR@K0A-Gv1tm
z!=tgH`y`IdJ=mqflA^wB>Td0COzb}Tl=%(VOi9n%&dgr-=WVM>4L=4E@cXD5Xj{+h
zZScSk5WBQEO5OVz08(j+E_AWo$3H2vi?<Wa)F0ogaZYu(5U;78Y~*4{BP{T7UmjWw
zK6p}LS=wp%7KKVxSR=9u?Q0W{_)8E^qi@-;ete!2+gYNMm0xO88J*Pn1_A2T6w}h+
zCTOKj<h?4;MOMIjF}RdLWC8M#=|>4MAE9If#J_s8#$V<s5V^dfYqhc{0#hk@y;_^K
z3W+X`>|t(CPbIEfzjd(V%UEK*en%vj)}%kqq0H_G29cLg?3dB2$CgD5fc{uyES=sI
z6s}9>(=hkXlujRg(uL{6B;YJLt*TfFg1+W}`2qv>C@9k8#<H!7anDb`tmAVkX_3xO
zM3}lpg*q~4=fzjpUsDdBb`}MXS8gEV=(#`u#1bhg(R9E-c{0r%A4_Bh;37UtycSrt
z`vkeEq$1NCZ$~APy*qgtb?Qr6KiE0{m2s)QkVv)bm(?8Fc?4QxjfCKC#4{@~RXwN{
zi9@)|FuUA(EJ*Fut9Os09diuW_D?L^jEY;WW6zObYBl%XD@iL)kRsgQzv3HyGGYEF
z?5A6JI+e&}9UFbxm>`2)V+o#L4PV!iuo7eiL=w`#j;@}2QUNEV2jt(6dyX|;-OGpH
zbn)lvg29W?98ev#otjcyMzpD&#Cd7h?l0YkGw*dCAU|HNIlc!{Kl*wZJjka;ojMEc
zanABXv#XN@BlvS?96{W78Y`ITUOHE<@V`gqaqMj)n}Zdko`Z`=wNzwi`3n-$)n3{g
z(-{2dMj){B;TZ?Ne7^?B-AHF48Jz9I7Zb8a!_EbgY<KJRUhE~w<pnc2O4$KkPM}}b
z`ZoHG5hO8}!LwMIHohh4+ubAHm9-Ek?uF=Z3RiNg+JoO9=i04S{WQ&SWzMfW^U)6b
z<M&x!dcP@`9{Aqrtql+YvBL*pvMWb!%%DHT0VpDWNEyjLY)4-)^vuSLA(PKx(*cDy
zg&EAsq~<MflOQA_IzYI}i|*a<-To^EJeF<F$>&72p6RMe8EXU3Hqn?>A$_^??hEBF
z4O#fA4A;@ViJ0}O+E0$Bn(7D6xZP7|%Iz;v+0}_bR7l3?m`A?7vjPXCgeovCeVWTZ
zY~yb!Xosy>>H8s_jgkyRDG%<YNPv|7OY^IT+Ig3jJL8$!ba0-j_qYP2zN@MRr)zpX
zH1-;k?Kf+dK&cIg{V97*#fEdyJz3ZV&=sY2E<V|6=YsaqG+vqZaq`B!$y*X9F`Q|+
zv`z}r65vjO0>yps`~F5iYp~eAlNI~#O3hCZAr*R<(GwN$mp}-9)pi$*iXRw69yv`s
zN`0+@UJpJksT4TC7;l!mqT7a^9Smr-;8yz_1Ow@t@g1$<prU|VqLB6V&g52Q`MUJ)
z^$W$HAfuSvqCv_{mfeb5?VE^-jp>`<Gs06EZT@8zC1Ulaz2n2n)lRL#5E9@eHjEEG
zMf4J4W!d;>oE5^+_c|A=7Q36;B$CtSIlAVBw-NTu(>fdhOFl-Goaay}xOQ{2p^~Mz
zeG657=Nr_K1FW>EeL9z~Vz*l8jf_v2%pck!5b)1|>wL{qw=T`Ngutu><KMmg0Zh2n
zup;mYI9~+eRTYYTtn>-L9uF@D#P|<S!5e&3Y-~%IViS!@qp3XF-sPP0Ek9yUOB<dp
z0_g%`9*7nm(JQ2E^Br>;?a?=lGsnIWEW;8{;Cg>q9V$r`ds7rsoX8bGb##X~(GiLH
z&t1X$>#n3X71GOAShq%;|9VG17y|OWO#>wl?lYg`!_{gP#VY&=hgKaO>Gn&MEJYMv
z!szkZW8Kt#$Q)fc%?@v~eOWxZp&V19%y30%OLJqIM=AW;+{+vFiio>p{nPlvcPfWY
z9Ks?;!-95kGqGz|cW-<HoAVi73v0aOV1f&qg!z-#wV^qvuja@r^04XQ;s8dd3jY!%
z1ik_Uo=SehH1g8~$Dq1>8{OZ-tj<L;s;oxCQN_v$$pg`X=GN>+=x-Xu5cV&pvgrr+
zp|VwPxxu)mc-u~Et8Qc<a7EZ;f9>H=&O$Raw%tQ*(%vgdPI~z2L<M=Us`PrBksi(L
z!KXd>x^Wv}Y!sM~B$`dGHiC}e2pCb$^X}dRLdp_c(zkdRzNKHJzq&omCr02a6qPhg
z@!&n(H-KvDl<qurv3snlPAhu{E>4XwF3ya8#>cj{fD0lcXYe4)B<5dadyu$NKX$Nv
zfMIAeP$cG4d4xMMT_l$w&su4Zieu~1z5Jxxc>1{4W{qo28F3B~WN?f>9z6v>;De#8
znvV`AXAKV17xo%2@pJ{J{Y)fAWmc8Y=R|_0kRBQkGaT1MYj5I;EHi2(?Ek>i!@stZ
zj28%m7cLHEY)@3<!;r|~aZVz&R%(d(i<`7}h`(lAd>D*<*SuG+CrLQn=uNJne>f~?
zz<746++T9XT6$CIuDVt_VA+r7IQGKAg2gu%Mkh2xsgKEQhJc?09SEce?;zT$al$m+
zut<Df?eQ<$u`?DoW4Bc>fH~Ti>`L}ekeIU(Y(HCYGcr$E-)8}qVtEO!e}xMyql(Ub
z%B#xNf80n|#x=-O{a{~+WL*o9+_&_t!H~mmjS_AH6`OTG5RY7JqyfP~(30xyT3&<J
z&70C#V8Id-b=urr2@!(KC5U-@BTR~k9_xbiWt`YmZStI0G|UHZ>kuR~4UfMveFgU8
zbFj?&^(KsaS0S?wjeRCn(Rs`VJlF7IYg~eQD)i~{^CnNJseKhwu_*+pMx|0vZEqfm
z%{hetjbu4Tig`2>vnIf)^tJgAU1Q%)ebLnCnj?33Q>ZHX535h4`M2W2pY>B1?na0o
z#dCN9VFBEmN%|6`FOByC3s3b|JU_V3`((P{t4SRVG3PQ`5%-kuXn*<eHM+(v_%l~Y
z5aww1I$!8d7ppLm#~=XNh=;$`%ZNz>>FAkL`1=;_2VDe;t3;MpD!8Zd6%m45AHF|(
z5cFF%%-L6^>_HF)lf9A$zX5ads?OR#VBivWLyALPqR@C9FtaUfuQ+orZ($H~`FNbX
zdsDz&p}R=TYQW8kcj<o=wcRbR2T45#ArirXf%GHvV!Rp{&|6S8eZhSi5UoJ9yLrLE
zz3rc*&+>cSCjASzme&aGA8J}y^r3-qT-qETeI*j`lx({j3tXYw-(dWWkU88}su_{L
zdBoy(yZ-}LJQoEyWqR&{@iwUD3OOiq5|{NlI((Mu%j+NgPU{4Zc#|nV$TWDSkHh4h
zC2C_Mj!Hek$p5OHKuJ+^hK9!X^bwy3ANvN^P%RI#!1EAs+(*iL(9t<-$97RK`&zeq
zB$Lyq9xwiyq{q?WIdV0f6Blt0$87oMt^sFh_`vM(rUN7*^yw91d4*qEF@#RP+|lt0
z?Cf6U>m;eP%ibg3#a6kw?MHv6vWs4*53L?&cSbu`IlL$Rc7S=*sY7b_t1uybexTC6
zU|&L|;qLJshswTH>`62lGe=aQP*JJkmTf<(uO;o%e)}fdIxelWmzay@VVFu!v0cPZ
zHq-7<Gho7bUmtN<4)rQJ@_boCL94kj$~M`-p<7s?Z=2!Lew9nejT(tK;fQ<Y7($%4
z-IVsRZVSuq!M57XI=2@0P23ehP6?NB&7=DAwAoziIH;_Us?F)N9!(yXnu|zNV2?cF
z=f~;C`v|+8HFq<-Sxy?M$?u~5+D3KO=DU9s(9Wh8cMVsoOR3mx3Y$)T49vaxJvbHZ
z?|6E{h*LDt+%B9J!Nz|hG*xEfAv;~7*t8X}kTPHBR|S(3JAZ`rt~+{q`<b33fso^7
z3togd8z1CYQeS7nPmdmGrG11QHalxTp*I(z|6u)O`_u4b<<R#(0!|`(a{Yb8YZ{I@
zH>$58Hp}Apncn<x0M;3*%^`x;e6hqh+f}~HPSEy5aq;wlS<jR|IQLx4F|l)8-#etW
zEdmdO@0RP$<Q@PM-}PyL!h~SzWdk)1yD5Y~>M>y)$K~I+{n_s<G}O{FM>C-Q^YBqW
z!`{N;s`{!jm67Un&GiF@26e*3t8=rraTOoycFZq1PpH!6)0rl>XkOTH93dLZvM*J&
zCC#Z^hdBW1=Ot5QU^`~@6-Qr5YMVgakWQ|P>tGODd20N5!6kL??p?>?PB$6Jl`9k_
zKA2(hh|#!VzJPXEJ3PFr_k-@KYS0@Vag=XySf@E7ne!8H@9IOdk3X+ga}=D0WIqe@
zb<!a65=tXAo`8IiG<6SB$&%t`=6ED0D`3w{5nfpv=c$PQ!PASaArJ2_ty%a{QFgD5
zU340E&W5_SP9|6)!cSCO4lEsj%f#mnv;JE&owgo1Ik0A~B_Q}o<4&z}lXlZC67Ytl
zc0gKuS(9Oc-icK&G?*->YQG(^h_c(L(XYs0=)$xRElK%ELpUU<ETZn)gt+m!=xF!0
z0Iw~Y3%t~)_VPo0&p6*{e(_=AZS|WCW-ZsqBHJ#se#a+0S2bTd3_fi@jdoNeJ2|E!
zi$Mbs7|*9|IKevS9-qdxx(520S%oa>uofFhP=8i-IB`=wT;%YR$&lYxV~xwl9HPTm
zi>Q)t72=Gp#c$kDN3^iET`TER(lODobEBu3JKUQC@;SDrBmbEwjcDpa4P%&-Y@hM6
zZ#r62hNNwshb9Eg{gDv*yJ!jHSe6iythv^7T1va<X|c75{Rx;}f^!^rDSe^7Sbrh%
z*v<PxZE^ME-)bUGm3V>;*(}5Rrtc=H;JYw(ZMw;{AtCGhOc=$4+Rk`O^2T9hXig?$
zBspfHrxT}V{N_fdndl}oaaz7x5B+kHlYWtet@0EWQmh?y-&bIG7u{35F&vLlH)SLv
zi5n2E#B=4QamM+<byaZGWjZ1t9_6*(p7tTb;yav^7f3D3sj;MvZ7kOpyey@?1R^55
zSfH+b)@HAByJxGE)IR#>_VkYps3_!6jOn#MD^|2MzOs5w+7f1_YC%OG>2G}L@&Kp?
z{R?}L{2Nb|98zm{S&X^z#po#UJeA5>rjU*J+rWSv*tSuZW?v;di_a%hYL%#-weR4&
z^?VO$g3Yx-={kC1<}troICs2$7`?x6=5%mMRVCOG@VP%Ce<f(|eVV7&+vouZDHgYt
zYRXjYqg}4d)>S3_p`VCBj*URJs~=__6WJk@7n$59_?xz~#h6*<XLM(aj?qMs!7D0F
zBR<Px+v@1cqw_)~^UO8JA{5^H7Mmy}OBSd7Uzngnc<f$`q2BG6)IHTPkMPPQ6i9hh
ze#HM;cyGSHccgfWJ%4&3))7hPHVU!ZzL4dO4>oJj8hr2e#eyXTlU~sVlzE^Miy2gX
zPRCDqF=7J_dusI+dcq7`!ROg$Xp+XS7n)lKV=k2X=D&JU+b??r8O-UJdlX2Jk!V%j
z@#Tzy{}W(cbmc^f8JY0B^u>F%<~@9l;*ms`xlL_J{8t<#VU7V#3-izAS6FXz;FF3h
zme&)gF)&xhB5O;)Yb_N)qzl*QBUIZ{mEXNb%<h}R{A62MZ>Q%%&~;F7P@QjV&5%Ur
zxeEC!TgoCkj*73*xtre{%Ny3#*sApjgxm-NN^uJerTrQR4h$FD!h@779gSWuWa**S
z8p4^!8htjB2qO0!bQyeaT{_A_rTYimsgOLrzp%ZCWw+1`$-rzf<!VTEI7&FeqT*R~
zCms;aK5(wN#HW+-I=?N7@MgpDtWADWG~q6q0={>Cb)~KTDqee8Cr_S@78uxjPDH7y
zcktfhWpVr*W9AFuuOlNuq_ddH9nVpxt6{-MCfTI9rjLYE$NUTq|6pp%zWZ&bpfF1V
zWgILGI)r_|Q9#$W>-}&qtS!cliO32!he@MBUF84q^pycsbl=wqf=Vjgh)S1qgVNod
zN_TgSw9?Yu-QC^Y-Q69RhIjD!|K1Px+r4vV&YZpXT5HE)EBm{7gYAV|llSM+#7Fv3
z=)rN?W_?7^`o}$G2;qExR;1IpZH1M6iGPhYKVJh%x^2pbEr@foDs5tjf<<M%!PuM<
zOz;s!P(LvZdx*<Z+mv*%pOfDW10-QbgdJlfPg3|d_j^e|N{qD2S&+?%ZB1z<FpQ_j
zXuO5Sp%|H!Q}cgowM<D#P5p(6>cAjJR$ow%g_YBnL%3u{_baxA1MWI0!?P}Fs#b|`
z30(t9UhSi%GZ#8W)XG8`ZEtPlZ?%k$?zRdwYcox^C@14TQ5+V}enn1?1;JY(d3hJJ
z^%uZ5HAh-k2W!zkgr$dd3E|jpel-<nS6!4hxE;$yY*_y3k+9Dq>4EvNBWJrq%YFuZ
zwa=R8&=|kHw-82*(%c#LMjHeVx-1E@1trHeN?0YgS&@PLNGJoCNYpV645q9k^q5$o
z``WhR7;`EPT1j|L<IdwtcFA^*b7cXhH@Lv#1~Y$f4`vO11rDMMw@}8JLCz-=S;3QM
z$?TRE_Lo5qKYuP6FDPhexG8#o<3mExoeD&<<kH4eSom{k@2>G8IOn21g)l5g3=bp?
z>M>|4ZbtQVBpvl9RL*<g?}`VI(^9rBk5#pMl<AKdRmMto4=@@QJXuD}8IcD&k^x*#
zRcc%MVSOr(DA$9e{KkX%$($MArN4V1LqcdVNBEjmi`^uj`>mCGqAp6N)TVBGmZAVM
z1xm%JloHkeA=_L1<}r<@Z{ixos>b&9crL;DHX*?zaANQQW!B~VS;bAJmm?KwGeijK
zkUX$gx?ro!DxqNa@wsJ1crpu^0QTi*`F_t6!!A>nmzA}Rje*ZEtFOfTHEzS<Crbha
zHMKy&RCMzkmSc8!c=)(<wLQlY%_DJ0Zu9!O4iwm7iIUx%`8(+VSX+gx{NX51XiP->
zwkLalHR}W6hMU+~J_BRBg|!4Ligr1p97V5K=Q2%df8CZLc?H7~m%AzOP<!x+F69nt
z)38&Kb2`d)XN};;Ssi171pa)LpS!w})#WHA109KGGC_XZ1X<Xq#r3Fet4e&$f$GtN
zNxUJq{#p?(yJK8`-Nkv4_70IFoVRCqIBH>G;l3MjPsQ?93v6E36iRe<=V`v9Zpq&&
z#jZyiNg*`ZgLbwmdaqf)du%gj2Rd~+*ShFLFazBG%!Q0RG%%)2HSOu`<Eo9(n-mt7
zzP_+VOt^OTnZi}M1Q<=vbmZizsi{yh`>@l~_B)o|$8f+mGN0PB%=nk;%f*kJGtYzT
zbkD*r`R7x{RYLyW!td3nIN{5~e+RP*nGMn_(cJy!@g_Un4FrfbHr1>be1&K@t)DJG
zph{b0%U<{7Cu-|Vl0xytI~Ljmss9Od<>qRHsg1Ns^g8@<nCPC5PqlVlXFAzlZi|Ic
z7U)aV(t0}~45cqiu8yT$Xr@U%+#n#<&OS|(Fi@{=YYf7aWVc)K2d{V~_LlhRTtPA@
z)C?A7s*q;#7+|vEzSP#Ln>AHdGLw|H!6x|ah@bys0U^0M3O#JrNKF0aASA=g4WQWD
zk&o0iQdl*Hb-<NSbr~Nai}2uydt9+hpZg^PH;nBa5xOuW`EM3eI;M!{%eb9feW47`
z6P~30`7G^M??Kcfc`wzSW30_o-{P9SlbN{fHVZIFnLR|B1GMV`H^^F#X3U<%Mg74m
zc{5*GiW5iTmR2Tm7c%-C1*!VXSgZ|BP{bN}I;ff+i$D_`E4%(WO_Rhwm`r5kZaIbN
zH6Y0;+`mV_##T30DR~XwQ=-n{@0-&2yHBt4b1n>6Qq|G2m0E!5Le(qviFO?&q$EXf
z6Y!L5<Vr)EFnE&iPv*~$)RjaqnuWGpfy?X^4Eq4QuK8817+`z6N*9Yr;^WYOQdDd)
zvfMPqWHR;+i`BARDzm+d@3f{!EfwtgG|Sc<A<o_JKcuUOBdC}u)rH`0?MzEqX>-4|
zSHkpd_>_h1VAxvzMalw^?D!N%Xe3VJ8*m?%5K;4zQW2o}`wv~o1`Zl9BJj9`AW^ln
z&9^1-M0m8-++JZp+nifSB(#-`6`oWasxtjW@K^q-D@Vk0d-i_W18HDUc%c~P48e@A
zJx;SzusGXep&!dg7eY&Vu}+U$wJQ*zr1`RLCa1eQDlKk=AhHqINS6KqZ0su5K^ZXq
zq{`N1z{Tc*+_u~&NB`^yD2d}eV4uV#Sn3{06D<*)q3K;<zaOm}nP56TWBX9Bvc5Zy
za<iKCp^@3}a}md~VjaGzr&yNGuD86(O<y3s8iWBR+u~bRzQ&ofc5KJrDNC5Ic8ie(
z$AMN_D!O&kU%bb@iCVv_1{|t$a~EiEq4<c7x~E5!+S=KiD&9p&vG+i=kp2=AJQaB6
zu+^*v>im+li(cU&J>(Ey+O<5DIjeVw5Mdu8X={H3%qe-nX3tQXB&j9?dsCM3f2(Of
z`8RY8O!WR<n3!f>srb3YI?Lw5a9IVTkLmc_*FWsbBTbiOoC4(Z4fL&g#|Zrx1sSGR
z#u&qzZz$C4<C5-XX14-mPtc7Bgw7fBm#|i%;*M+J^p}>@Y((s{FE6Yqi4#5|?eM~l
zo2+JNjc=AgW3dqjGFoZVlg#7s3$`LzxM+;wM?XjWrAL1I+_Xs%#?c9oQj9Xvp~t{*
z0HA~=$IIja_@?5|AEeWjOhtm`B{t(hHc&LDr`O~4=EYa`b$NFe)lbSrJ5~>eLORA4
zU?B$W{h9sdW?ws8bBKGsCEVAqPj!|!gcoK`EQC;YX{sZ+P2;;n$u2&U;W0PPzVE)z
zprc%?$!SSvt2r*T_}7Xkt10&~`i<BGoQ4C?1{zAgNj+g404Zg)lR7x`X2)84eJS@1
zh)QNJU2hq8z)r8>Fl#zFjj+D@P?x_BPpN|Uq8wGhS1F6WX_>#QrJE!JN0%{ooEn_$
z_H&qw7Na8>8)i>$zkx_uvGMO5IKYFb>^TN&)X4q?!yFza_2XD<-;ZSfWWwDAn6_=*
z;;#x6LL~3GG}TU@&7ahx=Buj<AJ+IKNI!!h9fXC}PF+-B;DoQ4z}ha=A9e?c1Oj-Y
zQc4_SJ&=`OK_^@s>ZYJM1$B!dtaP@Iy!%n?#A$={bpsv0udSer$a@)E`Jhg%UGHo8
z+2cwi(GwaXF(lfukZ(^ASR4vI>_mjPH<T&a{A*yT#rcnGPlihtV1`A+33W(q<j&M8
zXVoa1+Sw7_!$J+TB>b7vqjY@8kmpG|2@@vomC`x#Uz%&#WCsaKF((Tdv)>XgSDT;o
zy?kOy+baFYMJpO3Wzxn`Yxi{WOtT9A4-qPmF;?tfZ3Bsm40!Ic$IQAz256!=)n_xp
z-0#%s38<*>v|OIC;;W8ik9k6fmWT5)T&>T{zfbUtC{T=y-W%xX9}I3!txWw<Z7hg!
zgZs9SR+3R+zrI_KqRLo-h80@WrN^MbxGW>rx~I8Pz8{R)DO}vvN9EJao$`9n?Xhiw
zmF=F>UmHk(3%YB`8J$yULpl4pRa<@<zsC`^%=fua)RXKpF&Ey*d!p_kfRM^H!L(=j
zH5fK~Ys28p6H|pTFvaCxNx$(@>S{h=Nu#4KKEwZ4S78CS_iwVwWccHP(yP(NX9UHN
zHNsiK6CrL<_KHA8GUVucMTcOafeT9y=?;MONQJ=??!|&OtJ!N|E8Dtq*ry>D=v$hf
z7}u~>|Hwkka@L>jMoRDUKK!BFapJVPKHup$fiOI07|uavbRf^?&~IWM-;}Icu(`ZW
z8=I2}ixSwMb4TCGhJHU<%1+nZ6uWA0L>1Hm9Ed^1FJ0kNR%LsQ%@~@W2wBy<ds+X4
zQcY8Aw)1Ta!I7kZ)ZDKqt~Ga+1b3w033^Ix*PO<6Q7ZPH;dISU{FFr;4>jEAajrnL
z^4G?IRmlAVlxN!Y#N%IB>?=WmlLx=riXQqniQI$!Pg*r`q>l^dH|2YUcOx~K$r8Wz
z$l6rj3=y<0!C#V!(E6d(iJ8IysavDNh#ot{g#`JvZ0a;1DB-KK)^6&*Mr%$_X122z
zY@zbDb(I>=1g0?G{?}n8{(l=3+Fg`BC*-)RyA383E-4>lpyjFwq}yF(V5=a6QZY{f
z5$F07_nVN7&=em1lWqOXPT}wmM9x!eEv=+>Ev`)bM2P`_cZ6xxd4uNcEc<EuLLfH7
zvEb9qo_xE*vmnSn!!w=<Gvo@4my5|B{kWx4+BRg>RSLO1oH(Ig6}?fW>iMN~ZDTWF
z7UFB(kO#X=W7;f#ZtPB%t5AExq`MpBt?^b}=yDh~Q{dzykIj(c!fqkKTO;0ZOqlF7
zjK!kM>*Fh6;V&Up+n-Z{%Kg!6Kf<1*&IV&B>Blp;o@xA#I&&85$UQ%nW1Oe3P2b@0
zxzwZ|puwD)W7yfE()8Ru=6oygl8M=e-QR?7V5lv9jzs@AJ<VGEKb=Arp?o}oK0$)b
z1vW!Y(NZZrLisb)_yqN{T60AL7Mj-0@3RpA5CM_Wc=$%|KwoQL7tDx@r_(K*B2~_U
zYDadOguCTCA@Ahb3NjJVAVF<8pN1GZ&o&w>tNaJ$jtP0X^f5KvlYQ-?#l%KBUQ7#{
zK)*<zi`JD;Ntrt#)?GF~ID#G-xmZaD8z$d5(OhN@vm|$`GxHE3Q+G*fvJ)YY`2+7A
zA+_jErg35!*;l`dD+$M<(hs4>=`zC-;(e#Z{4;783EOy#?CdX&Jkvy^e=6}fsZ#}J
zEBdF5+k7HpukiBxClJ>7)!)LuR3FrKVQlU|rB+b5uF;6A&Qa3CU)WGK6^Uh~Cj(!v
z9%!^zCCQo82tbsTi#3W3_P;6xMnx%RZNv^4fVryDYR5N<iY>~;wdYQhgC!5TX`~pN
z`Y@>#O-~yO{nI$o2!_Sl?AeJD2$(f!%-!8{$So{LAPdB#qL7(){h?ijNv2#TuM@c`
zj`D9L+9Kiqwkimgpx|xg@(BvQ3h&+SuODk3snpaj9s+jl*M%GHaUu-ksInkycD_Re
zS@Tgna_97LnGi%?PNTPsUmYYC{^As8Ae*wj$`zIop_;$O4ob)==_&);l4vcK^|Eq)
zFyP1(6g-_Dh&``Rg_iV-1aV5<B-?-K>IqAV162gHW38vk!_k;*^LmiGCC3!_qsTCZ
zo&Z!m3L`B)t91X=XJw|wRy0V5IbA~6<*CLZlHxZdo>{C*i=-&c)^Vud4@}PyMydI-
z%WFS&mPcYc95Y{B|6BZZgNXl-8$zN6S80{wC(c`RP~ozhj>ckmqTyyiTn$O~O+Y8=
z7`f0g9NEr_mCQZ}8QU-1Wxb>ax+PpHDxp+WZ|hVGH7-8@k1Pr6?GFq+RR<0zD7^PP
zNC}k{<i*n8-h#41?{7{oVHtnQEjX(^zAZuogNnFbn1mbSW7EW7_=cy{Tg$b>?XJgA
zDZ|m+zTLuN^Ndr=@4;#nju#A|6H`xg#d&A5XA=ZM3MIPeT*h;{tx~{tIK_u<$+xW2
zYKfk*D!>3{VZw3$?INDSgQu@pMmKY7M9{24tpYS*REnhbe&zY11Hxs#3*tX<fI|TO
zWjLY$4LFbk1jqi|3DGuZyV^{<Y}sPgpuB7S0f2fxZ!RO5?KCi3Y4U2t?UEXlZ1Af_
zhhWq=y>ggi2Z1mp#5Hzoh+#R1&<<6e(d1V&`Er&!sH#%r|GM7nZvEBM&rF?GlLa;B
z2Q}o*v5Z%^Nl37BqM6Ba{AQh$pCH9#IwvmOQZ9jGXxhSAv@K*tMm66B1P?z!P6JFZ
zoOC;sDbHLM)z?(qML%2^@N`wTV8X0wl)UpiI_}`k1-My4v~~k+dPR98AI4fl0+ze4
zQzUgSP7px-lK1r8TMAZ4>KoFFi$+!vwz3cIva}ot`9nSuoIse@;A=B8f>Q)DjV$bR
z3-*dfa;hBKRJM}0W=x|C1bj(t1Gr3-S!}M<Z!VS*)2Dr{O?25{41=;Z_}wYZexi;y
zV7$Sy1U#2;zNTwUp)$d^GC&zkHYiT%yy$#<ll1=sX(!)NDhwy&K0UjIrb04g@~Bog
z84OlJH?=sVFC4hfF5eaUQXf>4*cz5TG{lMhG-~7xI)Alrkl5p?6=wd61dHO1I}Sw)
zw{4<iP<ORAhi3jIfe?4KB8!dWf(tm{3j50v>6|D%$6F(D24`%<O8;3tL-JQ{?k~Am
z`E<Lks^#SRA>SrIo*OP`Ev95;6^zrB*f*%4;)j{yO*O9D@rSb<mrPlD-fmy}cs6{U
zZ#mJ8HRsdykDWvQoU#<O$u&<?tOrvhkbr&u%o7gN(k$Y3iL8}8`sccN4iJ*rVA1b(
zw6f_~p!*Ga&&1gq9CQCB)xC{ZY~|bL-{L;dShx#EpBP*f1h$#^r7ic{P7697wLTG^
zO1v+WXuq5goNjt#oO`!-`Mi>Avr&V)tV(f8E9vuH>xZUq*K~XNM-G&0xaFJ5#*3%y
zmy+yth{6zw(F<0QjYbBz{YPsRvB3NbS{>w3%rYg0P<;0M8T$0JB5LJHsRqoch*{5Q
zArP_12MEqk9c!%>`u2T4b<rKok9!`o#pkZgYJcqP#^0!zA?!37op4@!9=q=&nV~t>
zN*hy~Xk0-+bk-+=6hE11-+!iRrhC3~x@XRO{{08sb8`(h6u{v2K}{TaAPA(n14*I3
zx~w07Ns=h`X)~6KDNboCN*k18&oo4L?jSIhXs9nfzNz`eg6L+<ICSh1@~|jjAcZF+
zWuR8a&V7GZ1BI`7O}tmO7R5pME4C@H&dH1bl+t&WZ>q-^0_#a2tHw`>XBp?m>rwBW
zn%k}d8<b~(+_mq#u~Yc?m4}+Op1<`Q2A2G_58#;gR+Qpkxp=G3MYpkkK%AlOq#977
z>Rkc}{}r`QV*KCc{3DjN9;XcQ)b5bqD=VvDtjDxWNck|bP3JEYna~X?rRk~*gjKP|
zOA!%q-47@!ip}MZYHus8%iT7TElzdb1Edju=PY?ai8XgH4fJR1CsLWUUsI|(mNT7O
zuqYqmZ!n6VigusdaX}IP;o%CeGbIp2-rggPfiafv%hssaPY=)sNiziR$FO$F9uMfj
zm^xZ$jsmy*tWdXsa(v^hiRaO_JeAU*eS^S>HGt_OUr=R-b4~J{fN)jL1|JA)ez2_D
zP`BR|U!aBgx*s^p<qD!sGU|DhZ*H5o5Na1xkm0m=C%Kapm*m<8xxf_UJcF4VuN|*c
zOA>?J8ap3WTyM3mJ4b++j;3L84YcsC=`O<Q%+`^2Vgs<^o7k&EMk2J?d%ql*h_FqZ
zL;i&f`^0}x<+mDuKAFobzx?>FO06TgD`K`{S2Vyzeqg!PY`cAJtI)aRbKJWg<Eb-R
z(&mS)VDZ`zm9q|ibj;26_jJ3fa<@=0O{o<KF>>pwEe{6_{sfC>Zw#gIJEvNh&HbBY
z_m)6Fopm(I`ZYlUTsv@plD*B@>ePsm?#XU@A5F#FO&9R7JR)OYiS+7F>*+3)6HF6k
zf<WOpp$fXF_^>N$kQ9w~$Qk2T@y5>21C2hOEJ}<evcaDD;V5%}rnuDT@h;}z6k;m!
zT?yLMz=70d$4w&pfG!_4m2PnDG(dUwVaiIP8?_#S*VWsjm8TJsKKw_eI$`jJDQo6<
zH2_PZ0T5Cr!#)mfF7?s0eWrBG4NT$8*pU!2E%GK>Uu&8OQ`3OJ8AMhiju82y)lHge
zrF<ab{dx9={og9qK>r`rD5rHl*(hSii@R!h!`5U<yo2&2_T&EVO-<vQHzR&VMTg{r
zQlwXlJYZsv>|$LO>JG^@F(-S3rY03|<b25ofySAucp3&w!AQRS1*2;AY)TKJW1A)7
zg(r>I-x_{eMlZCdat+;akCa_ifTMDl!Eimx8Lb(HjA-0dDEP(+*#C-}&Qb>Qtk;!<
zeBTn%Km^>?KOB|5$izS^go@W3u(^8#Rr&U42r#7S>C1YE`%6#OQR@UeeIfh~f@0=c
zu^wIh9X>EO{`UFN-l15#oj<}5?A|F-VssQ_VoW|B==<F8;xR&s5(K?NGV*3>u^ecd
zV}o4I!D76i`QBUp0rFvKaO^U%uJ-1iXE)g_e8DBuI-3$3T|VaI5|q2Pc=V7W_SRM$
zU0Kt94<0xNlfjG@;2~xuIgbCh7;!e{9rt2t1l5fimANLvl@*7yYS@tiA(Z5aMbB^P
z<ktS-`hDmBtpEFe)^G7kiQq&Ba!wu%a7J8FHIo{geA0o?)*sKp1url>)u9+I*#D?Q
za9k?ZNpFvQDk(~8ElMOz#&}ou#?o|HP#6?IegO?JD~gVImn8UeEgy=a2%c!(C_1x(
zV(EiA%JyM15qmlCdFg~u^Vn(SF8J0?lnr>iBcbxR39J1KLg&rUgAn`cbHl|6%L&8N
z*G%yN+SR^CN(Q`eRYZzKI04@Ygg!joANU+S6)Zy2{ZcE~dh~6WeYofIxLMZ+YY1%Y
zfTK6~n)hw?J+Al6tG6{Cse%n!io?>A1h6=DU+|lt2$e<$%l17J?MLVTq@S9tcj&%7
zoUL74YcBF~GB{dzahfo^W<o981l2j(j)kr;mUmJF-!EL}2L}HoP!^;AFLVD9gsBps
z3fs+d#j@r)^8wUt2P*{vfIxVwvs{1eA7~tx_<S-MERK3VuLFE$2h0)TITYx^;ElGS
zD7wCMrPO7&CL~9HvzC1YW{1X9=9Em8wU8RD99BZVbSG6CHC6=^T=Q}y5pQU1&ZZR!
z5H1d8Z91LXFj_lYbq>_L0yBD`D9RcL-=O@ZoAUnJY_?9U$J)07a6~IKOA-y@4xTfs
z-aVTIItA`RV?}P7uL?o+G;l{cJS71fjM3IM76nSwj5-hEul*{El`PUNL|ek(4Jyvv
zijGr?vXS2>OLpHz!Xq?7u}C>AG9U4{a54X`XvnCw`WU|$DG_iHQ0nN|PBR)P<gT{1
zKq-Q$@X{vKX%0yKgOHs_+pC*mC9Z^-Gu$Z4hYI_dhnMJELiPqHT5H*rHN9few6?ra
z$MUWO$~t;lV)>H}O4HSqC-kup?)+(~wUuIg!>UlxtoG?T!nwXNaC*p(omI25>@>|2
z;>rW=sfCtW{Y~wDS&Q1~#S7*WF$o0KxrI@)ALK)T-Wt;j+*(Zg?>+tI=$fC_s*}~r
z_bBE248>ACGYr>UPQ-8RQRSFH*H2~38!aNmHc*#q#V{Umdj2j`Hc$#o^(k;j1ry3!
z6H-ROo8mQyS(-BKii(GPrsBo{C%7c!(1GN`@pAC8zB`vSEGhQ8mKriH6;g;6EgRt6
zpiuAYIecsrdW!4TJbgu2*X8W1Q6DDoU5F4|N+;ISuI{??5ww4RJO0zZ%q8C>g++36
ze{g5|{u^=1dX8;-NJNB7^?XBLlU?!PV<ZU;MKt?~<mTjGZgeaxE!&nPZ^okrl=VF@
zQV%cprN=7I%WP&QcX5|C0bO+u32AI8VEIR)2KSf|O-h{e<tipvW+HvFe1*QP&5bZI
zG!YZR?GMg$L;!O+T3d?(%&|j>bJmHU2OQ8U4qo52uNNFUoLgszwy@ZFWb#^k*TVF*
zyR2`_{UJBRtu3xQP3)(`{XyK(TIYy(y<<d&C9b4?B}f0`(221$;O~F;<4C3{n(m)R
zk0SZs0}jpG31`XuL>_=<R6IFQ{o%7wH_Oh3hCRL?-$0PG^v-BD@y1qmGc&QfuVFno
zMPHPy8u{0DS~Lwl1hAC&(NmxDHGvN6Ph$~rE9N0{nC>rISadzZ3e%2~!==VV`iSLC
zPcj1qY@o(@A=ylWJ;i6^z<9Lmp=-H6!pVlwAYyFYcNCNbVlR#bKB1V&d8Enn+X`FF
zc#a`bkmx0f>vi;7&feTZ-=P3rJvr_nLs4_{>C^h%S!FJFD4m4$i)8a^NHq9SO5|{i
zndlL=wBG!@=ARE^2<+T)Q68T;;fw5VVLe8E07Qox1mwT_y}4l66wZ!(ao8eZzdeQo
z=VN9>CoVfgR$0;#lTq;B`UWE7&U!{Dv2WM%g@^Qn+i^uVA+Ua3jT)=0Yx!T^#Sxw2
z-w5t>DQlanOb~d1G{|cSXLO>-#B(~wt4+e6(0d@!cfM}q?Mo)vX5rBsD3SeNF@Xk0
z=wrHAL!I5)d$6!-UA2S<3AcvEv9L|1z4Fb%2Y^<LLSyCVU54Ao-s63~>OMqS=TmX$
zZhqAVt`qxVeBR!IloDE2k9)%F^s#+5qUuQG44%&Jv2xYeX9t@Dd?-*y)>@mL=z&~3
ziFfo3i@*|@ViN&SOPXy;wpXpx+-4#j59*&1lxRL~cM@ye>mY$fqu4L2a@_7|6fKo3
zE-UrwW}*Exu)Cj+?)*&A^m;nyr&e@tTbH`%Z?2Kql9Z0+j6<AF%Ha5J$Evm70Pa&|
z%Ahhg%zH&-JwCN}9z}SotH#bFJM+~$3HD~Uc2VVF8is%}_71`jeu+nOyypdMDfQp>
zqL>@7t0cmSE85i{D5`7eU*!w`e+jMk_Iy4U?c)pPq&1)S`!r8A@$Iadt5sC~B62?@
zLQ!H2C7atR9s>EiV@OLQN{RD4p`=YI?Oi#lSq{1kIBYiyOi&y=6A_<jSE4yR)*cx}
zxd6vkiT0-kx<-3WkDew`HY<bKrB|qKE+PC4)2EO(OE{1<UHy|*9KN2b&CyVG_QrF?
zdS08y4O(dsoq}1sHQd!dkGWi*?DTs;VYa`Wf{~)1_Sbi<Q;jeN*J~FIJFzp%RI*0f
z_=}#<)A`C`F37?q$BONhQnU+)r#9ITke!)|eO&M+0{XkW(Ad}#JmI-~@z_Gn%aWEz
z16`lRwP+nlEo;c$pw^ktZLgrFrMz)cQKaqDL8A_WNxh%&rmQiifxma=ROeC6LvMdZ
zgKk<X;gx}9s`P&LZROTFJvK&uti=H3yW@C=m0*c~joyyA3|gN5nwNn^u3(p%FSWu5
zBII7GhEXNA>iM&;PZIs`VU7rQJsl^?4S#buaZGH)xzgi^%`&)R3p&S8uY}GGwyk8p
zNU&PLk8cT<Di+t&c{K&s%Miry{6^FxH5VUi8n{-QNK=)p^}4jA%mc_Bok|S-e;lm8
z=;1ad_gn}lLVu&|fbb-Q9+*+!j-08!d<Qx<h%}mqRhi;=`2<EDHyXSQ85Y5y335@A
z2)q4Xd&_5yJbd4+AQCumLl&Sr5}kV+diB#DKSlUsxQ*k-vxU8PPL?F1p#z*sQI2}9
zl;$xSF*NI7XCw@6HmN%?P$%o&N#4isB0LF+^5BfzIeGxH!!v`~`dE(7MZsup^U#M(
zhsib3j0>St7U};y;{P_@U$HLt6v@u66&Q$Kt7%T;awUc+jh`{uM@p;}B~L^bcMac!
zODB`}!S~gqVE%#E(1G+qrwbN`-<pTBC8-re`(QnQ;b5botm#0qz8N`v>B$08KML@S
z^a@A_2;EAVXYa@mOS>zX0YDhu)bJU_?TX1H*w39;l63u=^=lj@kK;QsmU2kPaU{CI
zg!%fnc%-*9j&7R)+?hgT<V_AKh1zvmeQFqu%h-wb0>$&f{dqTHX4KRekP8O}9eLtO
z9Mabc7P_X)jZ@f%Ok#-$d0WUtr$E2a!vc%X#O>EDwl7#QoP;v)?Ipq;A$Eywi`ch-
zv}e9WCDSZ}DJ*HmJJM<Y>1Y}o70OY}Z06TLE(*6|gQLiu{+oI3`JeVHOEy=>@zrCD
ziOb9VBNrYacY3AfjpJG`VkZWr_L&D4>hLDod)&cVSi1Bo3c!;bka79F{M_{={?bz1
zVxIlUDTlxni&xOatajY-f(X7Mg1t_kg-Xl3bMBFMZTG1EMo=Cs6i_|pad=oE!oCvH
zchJU~=}(65dgNI(7W8mIHgqJQb2O6KA?5nIg98>F;on|pJ~>q|8(u#Pi2j5qZb*>A
z^o?=OGVJiMc<_V)>kkqfyz?m@&vh;Bc3>JxwvIlpS-*nsc}TV3l94%K?s_S^-;%E1
zkmY#;7gou<jL(tEsTHeYtkY94VSpgNYMS2R>EgA68R8E7>K~C<B7`#Ee%s3wIj|^C
zAlB+Nvi=(3=FpAcM5$wqm69usc(By!;W3)b2(I5ufm#@ld_21Nx03Nsr^2b*f2~@@
zY6oqmOt^E+?wr8WE!o=I+7}X%$0j*D_TfZ!d3B1H*_`W`8)_>NWYI0{*(fm|%FI}|
zLAvJebXu(xAWh1c8ATq*{|c9iCMgB~+X_X;n2PEXLr?#3{Fp6i1=wqQ7=M8^#<_#m
zWXl7vge0@ozDfTtZ(}h(H*Ow}gs4B7fTC9Uak39-tv~p?mLgT<E0w&o7P5!%R?ywQ
zJDw|FT$8-3TKltAQVJO&>GY*2!kWaT2}vfN<#3jji{(GgvfV|ng|fPI*gAd=&IEHT
z7}?=qxq|pkcvM`#P=RbD-s(Gqb+!uNh#cHmRYKc@laZOZPB!6;X?fn`MAi9l&TZ*V
z=`{ijyTR65!=+WW_N!}x@^kdANcSxZK0lpNsimP=37piZh+b#Lki%+y%TgLM&~t=q
z<iN$tG1*k<NUy<{wLm;oFi24Lh2L5f+(K<42?x>djEcuU!+utxf?b{fa-=76q=~@V
z8k5hBFNy2pkJHH2h|ZSKnuJ8(ZY(+3o0#Qp>z(>n7@eEqZ^kmU)#pQxd?P~Dqaa{r
zW|zhK<B^d4&S?qwMUwEqzdIEDl2im63L{Zq5KKGjJ-(?%E@yLa(5bh{XcCXIp_*#L
z@euvUPC1;u2D(ls;BqJC&z!xNZUM^sibw3ZzMNhu9$1hmrTwfiU?O-gw(xg}CjJc3
zihe!OCZoP*Bh@z>89lj|WP>k&L0x_Fci<X*e=r0CfRd$UC19%jyrTif4mQg0v@mhd
zI7&8zhM-(+ozii#H8F>8Mx6Dx`^VHxUuT|an!hEOKpS}vtn6<oBhU(IJbla|O)hEe
zI<3G!d`WZVlhV?LgoV9_YHx2hT2kYYX_TLG@?_=YOc))NIls6N2WFbxUFT~oL#wI^
z!~UI~lYeHU68fqn$e@)fgC2daPjY~-NNHnrPa|)x9;3|OI<;rwdVFKgbqp&e?61>1
z-Lh9?M1V4WVgZZ#l+jnOU^QcL83L_-)4cV@PXjfwd2jco!o##^+RU=aYr-BbH3|vL
zN*!$l6n#}Izyv{munY6){O|&<)Mj`>JPqtJZe)T>9f7c<W`-^7X12ci!>fRd8U*)j
z#GH}HHS8<6bGZt797#4Oj*N<`d=IHAa@MW$4Pc=merI?HzB0ch<pl!gA8qzUP-{0Q
zN+z&aKipWuBOq8?TNAUf#e$Dr?|32=L#vsq-NKX4F+S785;tUU>Ehl6(`8=LjnHVj
zJxD=w%wjeh#_bA`Dbd1kRsJ_eSHpi=$lvtat-GbEms&yFn0j1S&XnSrenem9CGc@-
z>f8M*DiXiE{6LmGg;};a#ruQP?(hL{h@i=8@Wj@I4tX>y8FQbE&8^$?>PGQpy>hrv
zPS@N>)1C9;M?b0_W5uDK8u-yXdUWNZM(_`B)GaJ%GlWA89`CMZ7Gn|<C2VXMTp@=)
zA%}BN4-Xe3L3qk%It_e_v?h(M!2N<{OVqMfvs)dy<BKM1Z75$LYdRTggOtjuscAJZ
z+3_UtyVq|LwU_l*1hQLUT1Mb`O49bp8ID-VPpbQb`=Z?O<_;WYnaCzLi4d(RGZGnb
z8~F-0&>Al8rT>Ct3&f4c$-}j9PN?JRaX9As%8v)K;o1U&!yBE4Y=D%38!jx4Gx*%=
zn;Syjw35b)CBcG^$tC0j30g2SyEdVwZ$j&xgH7VSa~NZWNhx-Q({L?g;Dcq1I_XcR
zDJ&3(1vC}p<oI%=lGyC`Cx@i6hTkC~RvI-?S>G3&L7(hR<cM3=m`vqocl?V~i#BG7
zhvdD3-nm>x8?S*mmm7*#Phzh+TC^jv+UbAj#>*`d93gymZ&gF;v3+F)X{v+$92z!I
zgf6wIL>JbENOmD(%3KTUqR3l@8yOQl1ff%=3cP<b?Y6tC)6+o_u+_V@FiSkzJ-ckO
z!v(O^Ob1-BpCi&lMxKaR$_$^tQ`rzo7}u%Eo;ZB&ibw#TkUhbZ9e<sZHHDvx<IB}#
zHfx?HmCfWVZE<d3_+q<PDr&hZ>SpZpN4S@!*r6S*-uI;C)#m2UyaFeJDfG-}1Da}!
zQ+sx94`)3=*!0la+S-+BG@z90<pnJ#C&%V;N&XT>SXfwC*-eQp3UTR9@9yu>&@!m^
zQ^$+|N^F{c2>Tvke?Bs|>SyUuxLmF~m0H}RNrs<mV`>G}WDV>aBt5)UGoHgCAA-G8
z{>Jm?nt1Z(H)=>HHu9llADLd`8FB`EvW?uZDscOWCSyRmgY$NO8K3ID{z8ZvRK26Q
zS*0oZ7mQbrT}J0%AI#gdGSGKNHan6f0ssg7(%XzJfg<kCSm%+b>dR6A2OYQ;?#HO|
zw{c>yO-pGFcs{e970Bno@3r6FGt!$k3(Srzu-C19&Dh})|7K)WSB7V2956f$|F}lo
z%lu0hLmP53;-iB`45Kz_RB}*X2>Lp<%L$fQk9WsK-EUB0Qc8kPOyv3F-Vt{u?Ew`d
zJXB5YH>Yc+q-gmHl=%4gR#sLYL3_VHZa5LjA@7t_hLNl*W-*z}9bc%cn@XH|Q3Bpt
z<LzFwf6XG=xnqiT;G+mw-g*@{--dt`xt<nUCFgTdBMa|d;?vsB(jT;P0egB#&z2l0
zl#|xs?QXGumdk81zq~>aWN@<w=uCOV$_C7QYmntOp{+CFs38j=B6G_}cYNdAw(f9h
z_TJ7)7L(d`lG1m%nec1Ihe1oEA$$=o%z>JW<r=5&MTwb+%>-A?a*O7BZxPfxOoLNH
z8&Fqd3l2*%5V%&LFl&4pxcj6>^v^nos2C*%(yGSBY>W)`;f_n*DRSZn0gMm18Y|T(
zb|q>->M}Ahq_sD79e?qE59X<b-SEx2y#t9uJThX@+-u+W>&QA^v3~~#!>024BTwCB
zI|&H9R#H+@QC0mLmin;Zv_DCaEtw!BBl8{tf!OVh`_i>Ml-rq8h@+#Um+JMQ3WebR
z2@OT5uC8V>pO5T|AhX>a1&9MT&@nN~&E^=ogYksH-%d|Y-`w3X7>#~xzS}PtPkg4J
zr7gaU4I$)KD{$Vwa#1Q$>#DWhFq$nT2EWS2$vIK1$w?|1-x<$rN@>3=Fq97(J?O;6
zEiGqD)BeBFgLw4R5)YBIVmx}a4C+SvZLQMRi>Va=j>iZNhKpEQH8<}uR?Fz1RidA8
zWHGSHJ9ykcN84ZvhC0Nb@%E-NSNg)0bJERAF$^cB`y8uT39~E#|Ft}xmptqa$0#O`
zF>=#lpXjjO9-}S&{(!c^fK1^|jp24!Omq?YoguUTFL<?AYzM2nZKrx4g_{2AkGM-J
z@wwCi36~xox#KK4g5?bA_2rnI-!+J@IDJNQuahQRBr_()9p|^^w24xD3&kMxv@C6J
z=zlOwxQ~A{OFA;;5e4fF_;aU#n4-d>f|-kDGnR~*gRz6FSP4>0YL#yl217TXyZ;j$
zoRymT3+4^->3SgD`oY1#A2c$`pxd1kV0{nR>+3+_9qf;y5|xt5&dvS3yKA&PlpqXx
zQ*0J9Y+v6Q(~irUsRHGVwY9e0v5ecR8SR$~W@h?iW`3D2(H^-N;SFtRd1i*41PJBL
zx8ous1;D^l_xyDK@?Cb|iHn4UWOH-VYP0tPcvr2ZKRdA5tbX|JNKA&{(gCM%uV0_u
zpR~YyH6H&mK0a>Zg7puc)!^s26!!<aHvK3}VaS+4Rl|@lep?jPTo9#dRKW5=GVHcL
z_e(6I=&b(o71UxpRsTdS3X0_MYZZjTS=E<%lf}U&&qA_N#ET~|pDD{^N~MsVk^D%J
zzu<F^tZ6$$b3TB4UKHT?0G0$rr58=NZz!i-P7rKM{&e6;;8oK)wgPWp@gBn8{|4bC
zvAVI>6b@6{CrWgdZGryKyC)J8)iF$opYx?RPlU!#S-LF?b%14%u{B}reT4KQGO=9}
zTdD&Z23P=@s8*D1zEBNyI9K6cTU(3!htu&`1KhC++mGx3JA0++Oh{&CCPR<g^^zkZ
z0Y{{lmzT|TmBA41-PM7ApI-g(BZw@1f-VZ9J1>8K+n6O`Z>5E|udlBvJS7)Lk@l37
zge2-#8+>K4?6#z;s;ZME+9@T?cL5P((&4eOD#s36+}1Bni`YWZnEsDI5ijDfS}eMe
zv-LGRPz0aN1XUHljW8{rsQg47hL?5QM6i3Ea-Y`w)>5Ej5n*e@A&5#LKzpS-4dgoK
z&!lTpvnPIA>dBoMSihH1-z6*cBOdgZ%XI`ILCv!7NF#R5@S@zcqdEXhQqG7H#y`Nn
zNg;*vDhSE7|CyN>0J1X>)9$f5myVBIM*;GsAC>m{)T3=-PxnP{_C9f_QxV2KB3gX6
zH+|KJ$CssZBOEH%!vD5&vtv`Ylt<B6-J09rhqB+e7zr0%=no>o8xscHrJ)pau(YzK
z-f`JE=h8t$<Shv5$+);0)nLa^cpu&e3w6}2c!(=0VUkAt0So6rOe?nVZTHHB5O6&#
zJFrLz3u{G_=6dxjFc2XSljajWeVLtu)z6>A{wPG*mMhIrhYK~K`UBCLXgd$L7hKni
zwpqL@jw|j0LPByXDqw+g`;V_-$;rdhMd~=54rW=Q6{P19Qc|Rpl<>bmfAs+qQwYRl
zq!D>c-M$D|H)lOW#!L0~S+?lRKt_KF8#D|IRU$O9jDkWZXib*Z8WnXxSd#ylK`t1R
zf3S4CCDI=&&Ofd3O}0Qk5!pE`M(aOp=SaPm_qKxA8V(f%jy@+Nn#!joRWNMKYUZGK
z%89R9VJ7y%GP2#)(z^^Zf{l6gb9*ryWe}T_y1zXAd`^lVbWoB2O>(xdF5ky7pGynp
zp)gz5bn?MrWlVX5@`8B}vq`HI6gRA%KUF`SSZP{5E`gOP8ZcZc_tPWT(IjpA=td{$
zmrwRHp38_7UGJk|$u6L*icb1s>qge6GA+<oE#Z=im^ZMRI_GPm--9I8%|U7pZ>b9d
zcb<x9b`v)eg%$!B^4ZX*Qs+CYkV*eIQu)4s1cp~k91c5j`D}@`K)U8{B{LUe!bE4_
z=hf=QM#qxl%FxS4({ahSq@<+F`tzqx#ru1v`uc+RkRx`to38{McB`PLGMg>+yt#3L
zhAw`pap(@hfvT;qw>e&9*U%`|(wpk;_5%SuizTo;nED+&px64dW%DhmZg)_f$>QbS
zgt#RDnKP6~;<(DcXU0Xcu&}s1TBxaftB%9`cZYX@@8RkGgvo67G#=a2)AKbrAnCLN
zMVUibmeHh9(EG*Gd}`G|fZHEzQ9Ok(xpZXH7~tBzBtqkFw|+}=GI6%nyvxLqrP<*H
zsN8%N+WT2gP8hPIb8;-u5+i(m=;wH#V2$<U9K4v1mAepSYS-xXsBlp>%0I_r&~tTt
zVQDUWwOTjC_zt9SMPe|nj6XaI$F(*hD*AOy$S$ldZGNtr>b|7M9PL084~~jXJK=5D
zZIR6RK9e10EW^`zv(Gii1*4tpzXC)M*I}<rf9I+_73}tSEPNpW>3;4{w>$VeZn3Xm
ziY?qDA|pH3y8>}|J-9qPJTCXA0%K!i;WxM;v%31Uv)L1SH;@HOabOt?%8J(qS1y=z
zT8OPZM1Y&**k#J33A1J6864Eu*qAbgcwk_lYP}r=6zC20`y(kZwlx79<|xB&-=J`V
z8NRQrws|whXXLUgCbu;hjwHtq{yS!Dy*kSN3^i9#F{R8|kiVg2V~QT+6NoMs-;a}@
zps<hyA{;n87c0$})R2(1Xm6^Fe}M%Q_Sq}Pih~AQiSs6@T347>dAqn#Pdv5p2$r{V
zUQ(KQJznzM%6^TkfG4OjZ#+CjWOjY|8xtSgEf$y&gYa~OyFT%$)IuvGf%+p;>{+1p
zr)N3j9f-hR2jV!dATAK!pXQzjdRq)xLyETjVZayOkAaKZ?}vofJ20?uy4Hcu0iLqg
z;7|rSjd8EdV$BAl$=nYI(?uJ}uE!LN(?0LeC>Lii*4BWvEOX><6qMR~^7n&`Bd6^M
zbQ)7Nkp3g86&Y6C;H+!Sdb@Y7u3OCCbaZs~PCy?YkneZDG^3CVf}e+*jZm(eRqr4?
z)(~ml$4Gx7k62pG`uC`)UJVVd!P_+X+5ND~Aq1S$dW8RZ6Ect%402gbDm7mpx3C{T
zkpsQGMIKsJ0;WZuGnK{pTr?MwQ;QkU@kX(t6<TT$R^^_UDSH*vKCUVZEf-R@s)~<g
zK}!gsnFMcX4^GVQ8cpUf2j7XLH1?NYTcuB((xPKx=&(>&{ysX)zikT!%m2l!t*lZV
zPnIu_mlB>Ic6h-;NAO8ra7i4iB_R~R?}tcCLxZF|a{%aOdpzE49rG6{p@H4<5)z2u
z9%$Nc28xNYW)5gfv|R*6`&I%SvNJOymo4Ti^SbD%73H0wU!K0&e1XAqni|}C5-KVx
zd+x`_$Dr{>fk0>d=t%RiQ!q+o1|Swl(r3W=^9TMj^BEVrYK)bYT^Ep{<qo!I_zdU6
zE&Bwj`^}k=J}tgO)ye1+lLo75kR9oGajOo4z%IxksS+`8+77$>0~C(&A9O5?)%D|G
zK!))FkEvix`UL3?bvThtM9l`@nH`6XqhaG{;n(Um!Kl%u{x|3HbvkRiA&>EqoK!l>
zwyqWLRg0mw$IJMWQNXSZm}T2A68`DoS^i$$p4KwCoGm(7ehWd&AR-P_Z1lRb!c&q#
zJc=ZBrEz9D@jj_YkvY8H8J^)ngv-TqagtgB?lXi?%841$K~riI=w)0~WmH&+eG!C#
zQrx=*J7|_z0tAbdXf}V~%C9Tof>DC7L{T2U6cW_nMTr4|8A8FaoQ|}hX%EKI!(@#0
zj%SHG@8{+8FLsC5&Va9}c<O9~IoHGmLOZFEtjuIO9dy$2oD9Y>R3aj21qJkjnUbA`
zlV;F&;(%B1Ci1xP6ee`ue4cAK$t52y@!VL^$BJJ5j{g3Gf`9<JD?dq@U~qiym{EFO
ze5?-Os=Gg;5D^=z!_*x+t8+5;1%l~{h4D%jJ#v{-eT7#@_OrdC?BtFu@($wRAfuy!
z!_PX`9Uju95*cRF=Miul7-vbRgPXsyyZXcHwY8|^@YeX1metjesm(3D#jDz^DjM)m
zydS-RM5|j1mplUGi&WrLIx`K9h@}LqztMo*RdLT2ysg*5<&V?dyw|=C4cD)5R}afH
z3)L?%3`39j3YuI=!AuU)?wBRhV@lEwg>w#i9>1vz-i_YnV04Vl<*VA@CNu(9uC6dF
z-qxbx(6)Q$-CeuGgbP-GEVCT8>*KgMl#vy`MGYx$hgXPq5W&KvqkN`9es046T6Ex>
z9Q|Yk6Ib)ir*H9@h57J|7qia;yon2TJ`a@*DG01o7M!66>IZl0(V`_}COyW;&&^3b
z5oKA5ku|pFyWa=WG>sU`IpMRMw{-jU*e2~sFH9))^4sd*7b0`QO*gGBcFqJ>bp=wU
znJ~}~tWL4aU%meR{`Fe+ScY(n6m>NKJU`i>HJ+WFfut*%Gejz91N8M!j*^ty#~ugX
zn@z6Q<5i~0-a1W9^UNF^^P7(S36#Un;R9PtMx(#rkZ?z9EVY}TpIi-w6Z;}5<gvM3
zm_WZ*bh-%7+_Cv?nXdUJz4EXjzfX<--}?ax%3SRxUi8r8`>OocadALm{i`MGbNKG@
zX$-}mEOVlNzX%=~tVHWSG?>Fa`nL?(T6&OBejwuWcPud+Hl$k-J7+ETgi^Pv{-s0n
zB`TFWt|O0Mo!#Du3Ow9pGx(XNwY&JC59T-rtj+UCWQo_kXA$KsRs5EVFsbMzha5*)
z8zyaXh@mu?(5+qIq4NL;w~SRE{4zNb&Y38;gWQ{`oFj20VWJz1&dZ8@nWOlq3}fw!
zE}&Q8-r?r(Tos!}=iWP^yUeeYlt1Y&qh^MR;13AOLV`s}4lzdiUvW{l@sW}|4ez7{
zQ*DkvUAQg$MCB$$N}r%|rzBXYk^k6-g$m*_#9~vYF#igu(*posC2LB$Xn6$ey-KB|
z{!8f(bCg;ZJSmj=4Php5G;xr`<ennEI_1-5p44R$gc@G_Gio9;IUPwtmF39ir`k&t
ztdQ+-=KPG;zK9G8Ry?7X1AS)CHq649M|!7&F}W-CvRS(K4rY`jPN90=Q}z|1R@S<i
zO=j~BO(#p`<)wfU?vF(?UB5REgCB1k(24Bey*|Eem&6?^f(~eJZ{RI#G2R=`^6~YJ
zx(DgLz@E)OG|h?A;J`qEo%8Ma795!rH@Bsv!5v5<PY)|b;w!^v6s(oH2M2>d-^-xi
zj}8Uup{-!JGttxLyA==;c<-b3J!j&4qjK!<@g8LpoPRNN-^hQVB`-6-yi}^G87@6p
zJSXtLox~e6yxHHy*m!>=L|qh=SSpmuJWX>KKvIz{w&K&!Zo4e^>9U2%Wj5j!0vcPn
zA<Vm&&v7iS>O}=2lA=Ky8#Auqk?*NU(z~7bW5E-4oM2*95Ccp8)=(wUks)o~))<^_
z_KGUb1EKkuzv*%iw<p_Ze$e=*Er0Fhk7wI&IcF{<eK$1!CHnD!SeaJn>sdusFGdw(
zneqhfG50{fGQgKJ<(ZRZ7tFyaj+XJL?Dm0L?Ax;?DW_rKXOewfZ^LGdwHwbSes(5+
z%1p%X6`>qMc|C(dDHTCoU82xZfAUpbq}IqZ9RTJ>*O2$H?`AwdB(KT|QNcwaS7*0{
zn<i?x{+f(6<+{ILP+^}};_b$bM2%`4$MX2dOCx;Oy6$^7x}0|yjTM9N^E-5%FoP;r
zk+1gn|7r=&@!%wgBu`V3F+x@<9m&a}Pkik~M+b@yO||1PQD;;B)J2sqerW_|fNgT>
zp|QihUpq~$OTVJ|ahuS|Mk^f)#FQ{WKj9<X1F8gC_4v5BYR4`uPj^&kWKz@>*F{1Z
zYgsW`l#k6|h<4<%FgETRP2*RsGQqIj8P@Oo{Q*>evgXSuqs06WaoYE%3fMg!*jgUW
z5p;V(iXd!T^oogO(#c;~Sz`cj(4`I9Xbss!L_`Eqi0W63B9YVO!ewSwfay68{q?Kp
z!TA2=mgm;)zQEHnJgYN`=g_jc@#t;$<+Sey#D}rWzkX+M#xa{SMYL!rer^i%iFfS(
z>9Kb{#~WJeG3_Q3-}Zc~tg)6I)}~uLy8I!#C)|nJ4<*OWV6Z<=MFkI-$*IfCd>yvz
zgu|~WGiVu+9T6S&_9$^x6QH<NOLmF=Iz`06miD}jiqMiGq2-Mo&n??W1J08?Abakf
z^A>I$fafuW*21WSE>&zbvfSN=eh%CQgOAa^{^)(jnjgU~3^f&TaFnj0c;8Yn?bkUJ
zYN}j=VoO*e7QZ9o(<MNE*qkGvLYknyNztRK$`C>Op`K}Dyk+NSmp%5L;&HwHIE-xx
z-=7c`UBZ$|=FYU08u30&PG59W5!fRh>cPy{R%C`ZZ1J5H%BqjubU_-YECvDa$pzOk
z-(6@2w3A6l9ElvgYuvJG2PQ^wdCZ{`<VJ_;+PfXGbavK!Ed7CVvaiOo#84+P0IF1H
z`<<Rsetgul%a37<(cQhT)nsA|<qLEw94ZsXbymoO3(&mUzM;6vEp!+0Uuv}^ydke{
z6w)|U--COFQ158IO1C$R7}PV60RTu$Oq8LF>eXYobm6|ey%i7?JOg>}1*lg?S+H8I
zw!-(AkEO!`2w1dwt-oH)&d#PznfwU~+VsWJTJ8LU2Ib@Lzp-KKc-9O4Vklp+0BU!>
zipsL(kp<jCS>L_kvYvU3s?^TU&ug8|J~cQT@ukUt5>k>c9!UKSulA>)UT`7s{1}76
za{~q5&*0DyD2;(<zgR;KRXv^`9X%e-gv|i6e63l%05)4&+r8x`&a|{Ni3FBl$kD>a
z`ns0{X$Gisfr^-x^Hvn+aDmmy(zPpnMR~bc3?4J8r3~zvtlf660}J&n90JuF7wKiy
zpElPH4i-pzjQXtAv^@BWb$&+&K+&3rZ+vt_Y+v>@rl$lc33tBBeR&Zyj|Fra1OR9M
z>-H#{ny`$T%-(fTj?<Eg)%=w<)vyMy^pp0~<8xSB$_eckI+BL=sn0C<+lMK_ty9p=
z?+V%KTpRqRPwhX>Qh`I*u;?3x70es_d)NQ2q)r&6c}Lg{Q;8ifA6v8T@xT4CayX>d
znWfNpH%Tma;N3Oz<7DZvYar{*6A)HiR<I!wGjpd@>vHb<+QF>}Zg6Nd@K-=YLyawq
zKV5=X39~Y?FxWr}<P+*5EnF^tvWcijZ77KRRYDHl!jt`~r1qs6f$2)Ju&9UZ?LoiZ
zr3-*<W6S=4OIKx*OUeE9nG(ORy>B5HnT|(mZ9hrt$lE&y0l(MB_uV)IP_+MuhS~YK
zMrQP)DQ)~j90VudoqiK=yMSOKdlZBop!e{Pio!53FaY<`;2?N*SxpYeA059qYCEmL
zLc<JL8Z-ZFe+@PweHIF8ou0;wlknq9dr|N=dO~63O6Fh@FeB@lK;aqOYSfP)ac?2}
z_z_B6TwISqbq{col9H-4Ux?0=&jGbErx`X72)o@M)nu*EsaFTJwtl0grWOkwX>0Qw
zPU5@(f2@2x8AmPk3v`5|Ex`Yy=_-J#>e@Au(kjv|(h`Dn2`JLtjdXX%rV)|u1_9|%
zx=TR1yQRCk?!xcCGdQCXhqG6`?^6q!o0}6KY|hD{WKgeI`V&b3Gc`S}l%$Gl7AS|j
zFg2yhsNLYwHMv)Q`tj4J#k0%)QZSp&4;G=WUAG4YcU*XAc{z7-OZ9Yx2gU*KjP=}+
zK*ii$f`_9`TMeV-tLrEYEo1Xwl<dvkl^#XX3pTWTRLg>C2^DfbOah<E6fF9SNLR{A
zBj2LHQwgyY?Q$LQ_sz3;yK!jPrU&qTJKq`Qe>zX3ua*2zG6e}hh+E0dw}k?NidvMC
zF;wAGSN+`3sk+u#737^d?~;)6ThfLlen897-f&(sqm*a!tSDLbC`fWYln?*%40VJf
zUB{nt^1W7fE#pvIpG4Uz0g8luS|U%=$OJMt_hP9?+=mI16wCX4V>JsdV|zI{tx*FK
zUB_pe5H<O5g!}fN<~Mf#4jgY5@tn)7rw5-Rx`r&>0}3b17redF`H;BzJYK>TzQ}QU
zirqclf{Ux?_yUJLj`yVZ(yq|i`d3BudSY#<zLnvV-qQBTpYpRYUrkO&>e7FElYrAM
z^pnED*Xrm0POYJ_d<D>6Y6t8{CNTSecBN^qYj1a#)Ald?!ou1^@gu%5{Zk)FS{jl@
z>_TC-{re$~Nk?1Xxss637%;q#4H4+Cu}e1g_Xohtpwp7c!xyPEd5VVi#<Em4?w4Dl
z;wnJ5IjD%h#`k`B0|N`!6UF%8na3;|s|lKOGsc)TSHz_P;bUQ<<sQy#8`IV(8>RAl
zIy9(N1Y>ZF)#BMFdzU)KYp7$A^QRqYc=JhxsbA3#tlkUp;3{&**47q*S_<Cfbj5)p
zo)3f0L7(;PoeizPBcv6Z?tHGU_*4C)nVVDCo555rSd)uDVYzr0GupG9Ljtv(uTWd-
zPA98D*PTEt@+c1_<YyUUvm1ArjCoobrG9(N;?SE~#Mu$ujG!2B|C=!_B~ztajr>!j
zU%L7%tEOd+<;I)4lX--Sw$N75<-2@z|9biDGxMiR5}PYwX7(u02{A_R1HJ;1c;=^o
zlLRV7WA+3?Al?ePJ1b|C1*YE04S96DfHX)1itIM9aD$%DTpb-c7dV8@aQOtu*NG4`
zj%_$YRJVy}0qVOu5cf&n+@!6!V&$TaBW#1~eK>6q5l3=E>XyxGLd3C1H-Q^<3DT!q
zKG~m5g(^e4BNTmwOb5J~K4FcO7VB*Y{X2P>!TFNH;~WM~2HVXcaz@RX_nMkSwY9YX
z3s~CN_+(?l?0IvZqtoog$*rtS^D4UAWv96MDJrV|0@UF82o#QuDGCV*;qiNLmNp!t
zI5}Zg|Ax&E5C+6K!0HV;zG4&cx<u5}aK?yJ0xS?70pStSt+twfii}L_oMN2j8c3by
z;>%_}0oV;h3=&>Z05yY+#oyn52&hXCF-bIw&(VH-RbY%i*)A+S2Ixn*?TUzUi6+#Y
z#^?U0yE_w{Xcg9rX?pE}!{9rH-Ol(6u=!lzU||ykApZON`wO0zg3tDUf_~=7(xTo=
zmc(Ta>`Q07kGJ{gM6(yUqKEt$pD$9Q-7>12W9hWfvG)-}N}rdXoO;R}cJ9`_y|Jn(
zT=tuy%?PEKX~fr@Ar%|dc9zLggl3dm6Yl<acg>N}r6;83k_N+<N@r1=m=!qe&WCb3
zGvW-5JP9+hcQ|6?4lk|=5Yj!6ox9!sb=N!od!GlXSu?h2m1PSqjJLm7<J^!%9c^DU
zT+FY`zVtRdc4R{+Aw&<WcK+o)O0vS21=s5)Dwj$#^!W4C=&&4fY>ldz4oDcef0u?g
zxwn$l)}{Km(B^DOkl4ohZ#SUTv7Bkdw~Ki;#1kh)sW`IMShcq9&)-{U(mi}SbUO1T
zJUkA>oMN+wGoB!R-@MEHrk;_Hdhew?LiM>af^ECB4==ytQ%Bf>>GuEX6J0=oKXNTj
zdoypnZe60oh+n*DgHpKc6IhMITJ^zPc^s?j(=C0)H%?AY-|$%j-|BZNx)PF-lAc}{
zoVl^EOM_DbV0@irdV$zR$ZWlN@Q?C>1GiTQNJv?uV%3CcT=s~y4ZCJ|;1H^;WMgDx
zRJv9KJx|c+T3$4a3%mn|8EAO<tiCz-Y$o{MP;fyk+UiejnXNMF#PhOVQ!pYpMvB)V
zT=Ua(GrgiGmn!IS;IR76uFl=EfEGou(kpqGv^zJyIiBP}-dbZz@{CpvSw5h$IdmC&
zf+Ii_dErtYHxnhAw4SY?>&xQ8U$+lLavCyrSMOifcB{Qyd--`~{mUb~sn?IZi}f|c
zdYm4zjTTQwMO3z192>)lF{)0}xDY$9bPzudA^#(7MymJinLB5)gq?v$4ze~(c+2%S
z^>>l}3h}`+PEj>}xPPPuz5~N0q4TGdhIx~*VlE#~^%cd%{-rXDc_KXKJ6G=JnV>&V
z7*_qk8?~dYmj>&4Te=~QL0ITK?35WQ^bJA~0YL-l<CzaIiDg5FkibW1AqRP3X-QW<
z5Ep*=c|>d)ymx4{?YJn;L>NsBd(H|nW?I7-k!fo`JhGY!swgeF07F;a?~u2WUcA-e
zSBsuG1`@FW(jp5l94RJlD}G&^l!=8jzV-1hL@tT+*%)A*!X4DrY0Kp`^gw%m5A}Cn
ze)C*N>~5J(iOT+ab1f*gwd!jn;w1_js6~;8cA)H)gRB>->{sw1Ahr_wml+3KNrF9`
zGk5ro#D*=liZ?3gmmgbZD6!#VJK#yB__E2KV!lnxH+q=Z$zfV<bv((}{o1SQ=3f6?
zc=J?5N}8$AXN8u7B)Tz7lC=-=`+j;6E|zZpL+V;dLy9Q+t?--&F}|C&czD;#Ry={<
zjoFi1k%{BCC*7OouODB@mu$|NO6CwyO_O87D87f!z}oZ<dH&LXO4robB`2qVhA_0?
znR9q2)i>#y_}w|H3S3d9w_arq<DqZq57gWg#JrwIKBSNtZJ0(4bU_kIUM%=kCFf)4
z$C+X-sIWH`jP_CGzoz0yK)&hn;#wycE%AxK?i?{qU!UI_Hcmphj^rN&HUgN0D$Rrk
zY9o?|DH8QpDfP`80~2>u8>ywwS@CvGBl1N|jg^YG=i7=r1|A&0C1@Fo2G@xQ8j%>;
zv4XjAh+--leTV~p3RwgQsa0;|r`FTzeBvmP!*;`8#5DaGobs?E5)xQw3MQ?to_?^f
zmazv9;<KT0Z1fp+cJAjI5ox`(&0)5-llbyXRPqs#A#<AKT3K;k@Vq!e(Otq3$&+96
zxgTd0XkzJ}9n8TC)|B4DY1m&MvPXS)TK;SAY&~2c-cSZ%!8HHK{DC4>)3uFwC5oDB
zml<Eb3`dD0Y=&Hd&L^v?oc3X4=alG1ANR&<Mp921#OUqj2t<PQv`S_Cym93nk5LhY
zg@=H;z-;444|h&6_5lC>$ZKs{O4+V28htVwu(H*OGKrXzmg~aeV{DwMaD3kq-s~z$
z2i&+$jUhvaB1Tlgq$*Fj%xsunae1k8JD<(6y_Ita%*S07{(UVh&ujTj)Ien2PyExo
zvXVQa$v6SVeO)ZvNmT|NO`I1E98uq}+3Bk0AlZf+;>H7aAix#fB#AzpJlUg@Qs#E8
z&#Wl_{+f>TYenbTa^unMxry@P;aju!+T!dyP1^0ywA3eSB|7?S9Q*=UZ1CW-JE-iy
zX1`r+^`Z14X#Uz$4L#{I+r#}WE<iIsX*p2nd94tbzH|FkniK8GkZPYCwzRq@a{0jb
zQP=v2L+QVjEJ*gE0nAes%eqtJ{sWz7?j~>h-K0NIv9g+^Ou>a@AJv_67`n9atxiHY
z_(mvPb?s^+XlG7{_rWu1i0@ZO80fSrUpId_i}s>0n#89Cx_E37`v6^c*##HIxXmR=
zMhlrN8*9V|k6$U^{t~p5BcYt}hEVs4r9<DnG793fvv<2LcJDw1Q5)bJn0m193E2ze
zPLg<Ang&w4oVA1zZZI|++g~s<D?ka$BN-T5@zHw5+-U5FjM)ycF}yL|CMOmNkM&q5
zpL*HaD0UeST-pU9YP(nNf*Y|Cf4UGb2NiY=lHa=&jn!9&*PrrQ7*SKwJwd6oVL__O
zj<SCjeHIO^^hBCmOS|LLe!MDmWDOvCC?gxN=0|jA$^MBRsfxC7^3(Pw)Xq9JSq8?V
zty`kV%@YG*t9Aql4UR7(8tYP3qB;6hbg;Btt&BhMas~IwbJOA4^PGqQ(Yff@>sbDm
z8rMGci$QIK0Q}QZin}J%vPQ?yW3~~tYo6GmXwxEBeMZal9l^niKs_psL)U-1N2O??
za`8||l~<c^4}fRope&KhVUvQ&P)Vxx*48Hi0s=`K*3zJzws9y*tsX@7)m0DmomnQG
zN}Bq!qRHRzvKb?*fKQ~S0}}f(^(~|x7r?GXJ~RRDf~MvlLCwcixvit>M6Au0@%5bo
z`cBkag+Iu|a<<>43Mb0`#-`V}y+PboYeNccS^V*>r7xox$~tAsCh@=#9y?bk!sAJf
zL|y2Vl=J$iw|U4M_GTg2%kdTOGDaAp;upq7#13`!u+6{#$7!XO&^nH;;n9HIXnc9!
zAF)p?Vek9B6JWRDU6x1JhfN(v_V%dNPS%-5>`QV9rw9%wOypoqPF)ahKI>^!;*hX)
z4EM&0%=^~XK-948KU$u_MNJ%iEMr5@DW+sVbj-6+QWAVCk#-X|masFjv(YHBq}0h(
z7AI_MytU)N=wTbFP<W9=_hvc&K)a+Ys(LTahPL&li;b4n)b#zYqYL<<9aGkVF1{^u
z4Q#QEl3%M=*26CgbLw$AuS7StoLI8EmjT|_v|q-E5AEH$us4;%WQdC`{7n(U;o}jm
zOU}q=XqbJ<+jFV)xzM|Wab$AX;*u^dE5tdIPVwzh$i|0o?f-T*^*%6$vGgk6l?y{j
zHnwup*8%VPlFt)inVcFFx%;7|mnNE=L2*rCAu=>!XFVnIZ-9-5#eE$eN%wL}rc72H
z8duC~;~U!Re*!26__Mm#`CqI26vjD}cb1~@Xhue^Yi&{*==Zl8K*1+(SDne$_q0g8
z!$zO*GM`H_Y_(^Z1|SA;X$R;G&W^Z?+!X{f0icIzGsa!+{ZU=8R`403vJmZTQx}^&
zy6e%BvMR~w$`&k2<|J(ji~a0S>;^3=mf*Hu*uwbBcoj^%ZrKlL;V-s5)s}D);+6~#
zkVzI*%2TT3L$CTl9^H+1d)Lc`Yr^I)Ln67`yfMTi>TOi@%tYlZ6t!|>M>IG0WUA>A
zEUmdtSHKnTp2s;qG<`oFRZ4!&T+h|u8Fm5@XIVvH66<U$?Qbs>-#&F|fGFi=Wm><f
zUS`;yQOVB5z}~Qc4k-67p{wwxUO#OQt?pXQw7lH~7n93G;jw;HZ??aur^KZWTC#xZ
zH+TKPud{5AZSrMd%$()MQ&Kj>6FSYTsxW5yr&K8$r%88X=KI1s&*uy5zx=OG;4?VB
zDdu+E{s5?4lleO6j98$$tSkfY#1$6PFC=<-=y`;Mg^`hwTjy?-4M%~f?eOS`oQ|$%
zj2y*Rw)dU&s9&y*3#OwjiAEDT_1^I^on?ITi*^ClI=KYL)RM+eLs<`-N$84swoZ;j
zuE&LMM=ZNR-#)*kG|MalNBTk7fW*<y<7ql?P20GFehFmn&~26fc`ur3+#E>{(kCS3
z0@Oi-)j{cd&~?gEx1heVTh21sV&5#7a-~>)3ijI1IauikXXX-r6jHP4*1IEZD)Y>4
zf@!^JBO^*P+e7${Fi{I;oG>Fd;w9s#Vz}e9uE_;sZ&OiNnc#*z|F(z=DAuUVCA*C5
zH0osxL#;Q%HoWy_gg*+iH-tQ%d!;51XZmK-zn{ybc~zYP<4ox>RpS<0X@vs@m|XuL
z91@QvP|jF1&aydm@PSzMu0Q5GiW{D`)V;pkLIFpK%mZi19Nne6<-;Hpot#3c)T>Mh
zU@9l+LwU~%GX@O-9oIr@Yy9udY2;3eT!&Ca44rdf)LdhJ8s)7_(wOCB^$zU-ZIMXC
z|HiA);28d3f4+XY)}FegL>{w6I+go5KE9Z(?OOn0uL7YVkT1R<CYA!-AdHQJ!zuv6
zV`5@#?r*O~R?C3$%l%?MPmwBtOARm|mcOTDTz9*1X?w)h4@&Fo3_76MmuoSpoC?@R
z_HLupw^w2`T;0PJ+fzZ40x98=bNtt`Pj^R?v=1=hDz(LJyqx6f3>)5<5#L6mlzwQ*
zCz>M|*Pj{$n23_?*><zQ110505f}09{Ox1eHfEVqZoY?stHL&={^A&bq7c@wpGLf}
z1m8HJNPGTElfhR;A6hC@HBjgC<R_4RKNuI=^*Z|KH#LRniv^)=4(|J+F>bQb6o0!a
z0<T6CtO4P=uj?{Z8$W4fl$z3mA^?g<vEEEuIm?-VP%0vd>?Fk%TrPodsQ1nT!mn@>
zO4}CQ`006IITUNcQ@{goKR5&e+HRhcU+<=M)V?~Fv5Rr<Lm>z*m8UED2Riu`2ir~Q
z>|&@)F7eXhstjK2L!}toudsulMeXH+i2*0Hhcl^NzLae327L-$9YmAf8WA@#ZqkKD
z82i#!DhY2*Izz^+_jK;?%-;ja`KS2ZsB4PwvwUs4iywj{OO_vRk3t@il6|wB$<0`n
z83DzQ8N{cMLl$RsL>9EmLMHpaHbgsOOKMI}y>NU47#zY#1*C&8h%YY|p~&7|FX8;r
zt*4lnfc4;mM?$jy`y2Ts3W8Z>5|B({5fh`g)H&|py)%iRZGV@bU78C#1h<P1ccj8W
zagc_UW%cs!yReM&p`Ejr*w}h{dOkwr>kx56IsnL_(v*&<^o5-4nr8+qbA(PU(>!<*
z8XV#%<$t8>-ix1VJ8+}h)%yAjO6mQsk$KC?`q}7=5QNa0V%p}&tL+~!ww_9J^)>Gp
zzpba30uZg~F}XM5#EAd<fpvKyx_s5`HP@_}U(;BYn-|kGF;&>rTSGB<iNi*)kXb5$
zay5HVI<@GH=QgJ}y#y6$BCV3Qa>nb@FH#2#QFJ6vidKh4NO<4m;IWL`ke5~*PfYZu
zJsIs%tpBcDI$G{LL3KWON0=%?KR-cifW)sXpJP{o_!ni_gMagkCT88i9i$aDO`Sv@
z{7iIZ3qWXZ`a!nRX+I!>!9bi52FZ(e8Wt>hkZ95)wn$oa>MH~dd7bKCyxso#(Sq}L
zoJ_nziu;OilnDDsd{*sP*z$GgWTZM03x4LHL(T=#pd#lHz9?3qC*WB*X?0!m^u)v?
zq6V_0mlYR7Ih;JSx9rQv_~Or^*X^&C5Vnx2P_`TuKhY}X8s#l%rk#=_L7;l=%&^w^
zU6_;Z^JGCRYdnKaPS4OXps&S`zy63B&^(BZbGlK_Z7rpJNLNf)3y{EC@tSO(o@BlJ
zIjfle!ddXk^5mx3^9B#Z@E;2e#~YslyKW+A=@^Adqwhn1RaG?&4nMunkdq@vvA6HT
zMm(LtCBY-TX;xfb2?z*)UgC&5^8nB|Q*k1!UIYecp6P+qxo4)*{J@#Umu&1=v~l@F
ziB|8?O4pOiwRjym$%ur+#J4~}3JVMS<;xdc08V`);(7rHjP<@mVHX!pnBI7%QZ{5z
zBAA6qQmq6VFJzCd2&>f}N?=&%eB93Q%hu+0>2j+&m#)}hhaz%28$M|r%*aJjCTOqh
zw>JDdK@DS^=}tUD>i#0(L+`xssz^3z1()PQuF8!fmru0IIm-6;7x|<)ncZ3uFDe*z
zX}KbWM`{aW_jOE-J4VYpqikNqf%e1(pbklXYR<(QGedqjemo=RwTI&Lk-e{2qlfp3
zpZN2RN8Z^#K<jN2(JxcG16XUFi;P#yWOUOf9K0$qog9Dv{sA+<@0j3lsqv_JxNE)@
z)-y^QL%I@pDQ^Erl69AM>YD1*WAU`Y@eI|tqhbs!_JD5YPi?{zT-1HlE>Lu2X}5wO
zRyrn~jZ#7q99OV3%Y~oD2~YcW$6QKTlVCbH5t1hiM476YVYOR^|NH=y!HZS4YohP!
zpo1n65~)U0t_sxRK*#*k_IUT=cU<!&1bm&nBMZo0-zfIq8zf?O&CkiWQ<Tbi(jUr3
zd64^ncDE>%Zol`<|2lPdDyXX-kf9@M-rnBglNT4wu_=!VIEqhU@EL%`$bkeFo1tc8
zikI-2RiKey>JVEzY4cCi3=Sn@J3^_Q-kO@>Syt~#4t!D^(Q7^_uq)JGR;E1Ss;Tpc
zK&;0qvXs1zPLB%-h<!23jWLF5CcF3A#A+neAR-=_YW1Yb7mG`tslUML#r7a;^S4i0
zib8*D3s-YjGZ-0LJI6_IAsx&V&$-ZWRG=EnSu`FTzF3KTogPk2%@zH37q*DV()QFY
zQ`-B2f-O<qu>xAQooPPgyvpZFnfLHNrSCTRI>zZ%|96r+;@Xm6{@__*$4F}uNi^cH
z{@?c_pQHbXi)N8gV1laJ$uT7lcU8f(V;@>eICfk$M79lBr7~24l9mwOQ?)mnKfBlR
zqv@uP(*9)11$tb75!jrR27zXFWu<^Tr<tiCl-IcgN7qz&{~%qXdFM^9M?;T+=ufKn
z2yn<hcAJk+yfUJj^);|M)v~!@TRAs3wq2ZCT3T3$$1BLoOTI`3zzj}eB7Re-@S!D<
zU9<3WvnS97;yxz_Gyj#ETVY)>@;&?lUVB7$P+xJ(WreAQ2U;w1bg~>%%@Y7Oj$Jm9
zuvhJ4RCw~IM7_v&E^gEG32;3tEHFn;M5`7#voTj3+|2xHZ_T@5DK*ry-uXJHqZT2y
z$b};(6}yR;j)^8uN_;Ri6nP-E7*@#*nY(zuBn|5BaAiBJXH=1_p!zpwWC;^_Tg|P!
z3J+v~WEcYjd+=II)@G;wc5ilJU}0h5T`Mwpqv5>f8-(M=TxKgsy>XpoPCr?_-U#OD
z0V9>>jB{Au<vSD>JGmKky@TZse&<p)){3R-d<m9T{2Nl2+`nw|7jF0(l|QxR_i92s
zp1}<<77hf(imkJGB<MT{phI$ZBMnpa!1{kN!NNf2Rh9jQ*>5C7H!_XT96!fItiQ3!
zGe*gW1Y9ADj{kf1k=<Agd~T<@kKF4crVf+?O>Ji!)(ak8RUFvZ^GR<TO-FxFg~<bq
zz(tYEwO$PM=Qvb~)6kpaOQ>H*AP`437V-NF{qDULTs+1HaZmf+rMZ?=#+5X(%kJ|(
zJ)$`G@hkn4!;}>C+-KFDdGhXv@0C>BA_I6hf7u`fs9hPB02=YxSSiE&0+F{cc2}bc
zk@bC3jp9=;cI;BkNdH$uo3_UwdnH`K6HALCm4p%ZRWy`5zCD5a-#Z9riKcex3M1A?
zTJpSFiw6>Iw*DkGdi(i=1U!*&60Pey^U_A;I*+xU*paoeGG>J!S}<h1-+7bV^jAKA
zYQ6Hvr|TbX!7pW1d=?v?T2HYbTFCx3mTXRfyEtrbRK?a{Sb7U-V12}2Bh<20?g@f=
zjW&e#=~s6{Oi!mXj*Z7ON#KB2@ksCmL4~>RoTksYDtpZ4d($QVfOu509RIrhZT%H>
zSA>JJFPzGa>_K4Mn4<;1@cDC#M^|<mM&$e2yPAX>+8fOa4p{PwFFu5Va8t&>4i&iD
zdPG}~EfUTMWWISc(oOF%{Yil$*F;Q?UFhYPKFPp)Q5=)g4Cb;8Cj<V~y!^$gw>az~
z=O@_z0cd9aU}Wd(opXVV(_;D;LZ!MC@aX{E;OEspk-%#}nkXqEBm@hTLHez}2syI>
zK|v=|`r)x&4`2$!F=7(D`!lBZs4EJmNu2v;BvY}x`B$s#)$`74Ja%a4>xUuXa_X5c
z-97&5p8!f=w`=i8WL)Ql9R>358&Y41xC?(^>}%{fW3|2I-egVF;YI+R=31DwUU4=+
zEJ2sxEfB5u#mCb#UJEHht+^_;=qc_+isAT<{ZPTXp<!h0V_J9EpVWM0I)GdpE)oAc
zdK1!J7bhJaT)6_>8MKG_sPat9yKtg1;tC1xY-7r22RL@ayxW=m(Wg|DYH_qPC+UU(
zD(<`9<}~a&+fd&Eo6yH6(;8DTdtbLmu9)1gfpx(Hz-q%|_)<wKec*pNJy=X9=5c~~
z^gYGBQIp?i2SQ%G(H~L?iSq^bCeyzt0S*J3;Rx+a{J>Cn2LHrc0pmm>Udarbpb`-#
zN8YvTbwpcx4rpjo)Ei09(z}Po@nH+6RgN`DETmh((FRWAG06yDU4Mr3l86`X+RBG&
zSa~X9YldZf9*byA^^*4k4$9!VdZrRrT#1|G=x{x2^rg0{=*@WTcsjqv;sV@NAIVD&
z2up(lX@+;Uu}x`|k?PlZu}=2te(%}xmYeMG23-}3OD2DXoxjZIsfUwA&YK#nW5wL<
z4pq39aqAAQd&pH@O)r}2(H0SXKeEwfoV75ikbjaY_p$^8j@o{~{DPaj8_63<Z}zYy
zx&Ef#Jfq7lAnmclpV#wYF!kSt?F-4W!yV)nCIAL~w#rf;NcjB&1LNwofraDY_5>3&
zPaQqm0}zVa*>*<+xiB7!!5`vaoySU@Ij%SvZaF$V%1Yy&6Y+1saMrQBPerKm;EF=f
zT#|+OGqYfcn?d`}SK2nrL0eB^sR4t`>__W0)EW7$#=yz5zNu||{ubI9-iqay%?rTH
zBCpNf78F@+Q;CP!<k?f}EAC0RkOT$pQ7S?nio}wdOul+e!N(Kw^Fuuf=IEn~ik#IP
zBNe_*f-$!b9K3#><lLl9UlC$U%m}xA)+Uz^6oJmu<@O=1#OBPc72V`U6Z6k#OORYd
zNr78LSGnH0p@9)g8gyy!+Ri?DGT^_0;VG58exgF6WT$>OUneN}y8HVn>FB-!_VTCo
zBL5??DJ_iwkc01FDgG&VYhM2N6cdDI<>pP<ick&mmy1uzD3y{V^LW+yK+A2N9z?`e
zoW3WrX){croo9IJ<}{#6%@A%)cJ`wfgBAZa9U@ij<*KzKRXZ}pE?KzSqc%NOefk=w
z{j&Vf`bUian{aic!WYY_sIo;lH~v2Grxt3<{EQYnVt=RUm#pCiT~!WTU;sNshWW0;
z?7V+Q^XmG-drqizC|#$0p<RsIIQmtN6FPRcZ0l)A8S|5Xk(A7Go%SZHwdk}n4`&*5
z9*8}l@Z$&@cZ$*x8{Kd#>EzO#iR{ki)xPX)T?}7(;D573mS-+H{4ZeeO-8bjrIOy&
z;?`q5BKtkv2E0ezG=QC$<}t_WYd|ZECYw_#xVEp1h$7pf<4i^68C_C;3o}oRmfc&^
zbiv^zNF!`F#s;Xg)=7f8I<WtI@#8`}eP6xH{hSaJ>s;i)P9?UxACJZSjRwC4xfId4
z>9?3Kd;6~{J#BtQ(5)#*EPV)b=!MLX0gvGZ@%m0s)F}}RY_Z+s1Z*@EXxHB8wDajH
zpcp^!N*-XQ3tHJe6w2+*5xd&%xj*vS;CN$ZgyrTGq&7Y*kEHT`0}Z_w`WX{5pNc`r
z4c-Rc`H9Om0!QA6SKmv!z=vz9WC6{#L`Nh5!I*re*UB<-8ex;)A<y|xFf*Tsfls%(
zZg&kFcCCKgW1Fu3uy?^{-<%&7=l{t#JwdJg{Cph-P$$3aI|*J$Ye%^$#btS!LxBDY
zb*682$Afan6z4nUpN{~lcDZ{fd6ubOy03&4o61E{oa`J3PDAUfbvHQ7#e(Y)e<G}v
zEaqw0yGasAroxnpx69(@f8oJHn6NY1Q{8K)F)a0t!^jIWuw65G=ap*y>0<#n1NGyY
z+m8(0r#ReMKO&Vc=_)7g^z^S);Gb}&cAa(1K;2d9H<21NF2}vY?_A&bYb};g^Kpxr
zVakPhr!j23wS89N<yctx{}kc<v&X&$emPR5xl%x1x#4jzKRcz8UwQD>(#l75G=*US
zbjxIpB7s|~;e7@63$vdUA;BX;j?@^prHB){l+<_fFE-DV+!k(v5SAx%@S$Mi(r1k)
z@A8KyflF4fBtm~@-P>4HILW`drDHOyN6&yVJ5$M=?9Y>zB2DK>)2)$jMwlokli-;|
zyC<EpNVCSvN`4d9wRG?qcT0=rlCfe+xSIo%!er_v?fra`3eQh>q{MIDT^b|VpHNFu
zn<T-*&DN59Ex?Cc8vx700U<T0Uk$(Jh0ndV`oZ)5KL{eT6R_})Gau%Ufy=wI&aC{g
zmGbNlNzCH;MMjtkf;HjCqxLxn#`Y_SQDXXerMLi#Zf*NvhE>_!XW&wG*ijx(YJ1b%
zYS3jri+5$p)fuh?*H`?;WG)K<kR*A-cYjumT!OAwh_`j1teZ#44EWejC6G8Yt5gRm
z0qp?J;Bq~FEhZ+md4}D;NmlD}^jRw4KLX+F4^=5`b#{Qv0M-zot*=;FJp;;qsD9k;
z`0%ULcxiR*tA@VJ%gPL-{MJ)95IpxsrdQC&C@{5NI2J1clPBP~(@|;b8|iK*-bJ|%
zRNU5f8>afrP9H5`1it(8e-A(~@1xl&u{*GU0Gk{IEp0v(;DW(=M^ct^;_r&zqjc*s
z1;j_~%3E`4joW%QY6F{31QnRfW&#CmpYIc^zJ<6C$R{{Dmr4=hy2FxSW=;mRlO`Bg
zo1HpJ7tMVFV*m)Vnw7V>IuIrpX)=coc+v~u`+I9n`yAI}RPmzGsrceKev)hx(p7)@
z4;D|_OTzprA~ZkW>e7zj<!p%lZ>f>I0R3HBbhI!y%xP$9*lI;gOlW|H@GrZ{zge0y
zxy5yR!+zesASvec^HkW_KF=Ikob`{Q0-nx$K<<4gf&@N1Diw~lME#;--MucOK~F4_
zJ1j0x7Ux%uJ=H_50yp`U_RpWS-(Nj}Bsk)+w_5H<qvBN`r5m6%=ryMQGUcMWjfA>n
zh+Kk`0XHK4=ZnPp(yajo)0|+kas6}tE83TF;o3Probv>7l@EeeriZ4d4}GLJe3hX^
z>lMiP)S+Xd-{*FU=#R2r*7M0zRd$$#%+FCD{u_zq<<9{B1kP27=z4TQ_SF_2I2RX1
zPZspYPrFRzg(zAi+!EqerNXn~pE?FTAz3nKIOA@l)qw+gqAfN?(k}-08H4yN?nLM_
zZ22=`4%~6yuF`;XAp1a=Sl5IPGO>a-;FZjXQ+fihCx_E|^>5+}Yn`rHgpESkrr#!u
zi)ga+>Lh*yoo<;*OPSkHJGSPVea*M|@Kry)YsB|Oe&iA3bPP6L59^xfU%rR;N`JSF
z^9_`=UOH3_ili3k9)%YwC@@lO=2fH?k0)-De~``R4(&^7>_B${_8}<1qek<Kf+q(M
zwIs2N{%`u#4E9O#Edf)sMEqot8bwq+C|=trq(CvC*Au#GMNHXeieb&+ANU})?;G=N
zXAQAAsO3H8S34Bd`;qYM50r|gI`ldZYU2BJ{Y!Tu%#E}a4mGT`sJd2&mc+}mM$Gx-
zGzOYecD_;v5CcoUV6rF>>q?|ONQ>`R+4~2OGiNW%cWZgqiR0~|H&=VZXuO~(DckKJ
zmBme5nz%>$5*k7bIKU|WH>!61Wr72%ZJBStmh*Iba-iY~LB<_wIC*q(YtM&2vPuw4
zG0D0XLdIxsqoC0>|Kz_$XG}ZM)%DF(U20~!G(>H%>#~hbD7G@pG;BG-<56LQOwH7}
zaPSPF)da{4I)DF;VFW$g+*tdg(xDA*?eJi>rtD0_&RIDkW~%s`5Mn8tx14rfh*A*|
z522%6M{qG#@A)%8j3@Li75;d(E`Q}^@Se%}G8;Leo3qkL00>f)*UHS&v<K^go1>uQ
z7+T`qE~x_L&}|S<4lfiz3@j3t?lZrV!Ykvp&L?0@M{3Wy%Xk{y2y<gvj7R*dfuEuP
z+mcb!gU^JI7Y(J7nE_F};O%k}$t4@fz{c{zRCEM+s`(q~e`9GD|NRAa`t8qefmxl0
z$9<%#)ekTme@i0LqTAQ94bIaco<ZDRl@X`g*;B6z;a7}Z^9K<Kt-9cyySci?GWM?7
zDPS@^)@GRB0JZ}b@|#M-%Nnt>o1m~$U#Bnwox!Eezo~1bxuYQbm~G_P9Gz^a#O=m(
zk4C9^d%5)CPOj0?TomZ`{$hL-YxJYO5!)y$c+<B8g|@b)u<l?$QAGGU_C6Crz`})g
zzi*f?f7UBe#waH4pu8@`-{44MH|EceC2|<OXU;oO;O4RYj924Q$bnyy8*@u3Z&hha
zcim7m-ap5WO?aarFmZ0x+7sdNjpqft7#Z2(wna&Qqb6b-{8dg+O6@@rDhY*xZ@WvQ
zx%7y-y1p%ymqB6gY|4zgx)mxQeb|s{mBX|fav+|h2C7$4F|pdk%%>tOSM_tb>0WgX
zTue-Z|33{@ejhiE_?tD(7bt_8hh<5LlsVe=aI(0XO6n#&OX32E6MPUShPvj<bkmOk
zivy5}%I{w<S{fKe+CmoOdxN*_n1hUSyr5zvuldhye@M_m>ycOw3*Y|1He$Y@YjDH}
zGw@p%NYv)DseRVe2oEY*No-K&4BccIn<>%5flAWY%N4FPBJi043{#3v$0S+hHyk1a
z^{FI8rZk`8{6|dh^SuV=4c_I<zukHn+)EZKHHjR}h3y`Uk><#C@>kSr54yqgyd_vN
zvkZ7tY}NXbao|!_qf-9OiS)pkcpj?Ui@NC1#RT2iH)AfR4V%2n8{fVm4%CIEN;lZ6
zoYY?FWom}k1-?!+e3sv1^^Y&)k0f*&HagM<LvcOqt_IrkkJR>G1fFZG`<ok2ki_7m
z*_3MM?CkaRXMZ^TYYq`<R0Vl!7o&wwi-KqLI;G+D+05RO#pQc|Ixi@FzDW64I6=n2
zaW$K!&fp@8l;XKlcIDTREus~Knbo4s==p5P1S5c;MjOpko~<aSk?pb6?Ju7<Q?k3D
zBE;oMyU}OiX;$k3Qq8QJ6@pt^se<E@teBZ-gS>42v(OJv6*Nh9?a7XEQzyb34t{;G
ztLvInsHfep+JhN{KQ<K}aOa*`w7PGHTAR|nCxq<eY<M`+-@xY1*zd7D%9S_G_f`Gb
zH{)XzZEI_5VD^D&p)YL^QddvPcP4*aIF)IjV0dg7EHIi76%Q6liYZ`@x|rgwY}_<|
zRX{_X!!>tvQ~Z>_%%bbYWBjpxW=u*HQ+(@YRqrOBN{hPvET2`H{H6&hSVdG=d;an{
zu-1IYvTCKNFnX4Vgv7e+z?hw%N4P3{fehzrKj|>@ZC9H<x}JkJX8lozUTS?4bFNR)
z)jf;;;_H$NJ5i5s0_+>hB9nUq_s3@gQFysaymo$EGt19^x!g!&eAFh(4kK*7*ikbD
z5_~>S&&&It%<t40tOwFUq;+CRRD5N55*lJCX=#Ij>+~4JXq*5|xk9s-r%_*mAi&=-
zFfjhmUFBw4&DZ_F5PO^w`?U=f7N)K3BQ>uny@TWlSSjoKg4jCDOi3M~n9jZ25=xkY
z?<`jC%|XZV)<niy6yZo#sr6$P+kf?MT`<W2Xp(Wi_jG#^qgY&6+l(E|6B<i~pZ3ll
zOpP6zg^Sp!tvTG9IAYQcFfz$l({C#*D(DyhP&rl0MTXVr^zXezW{q!Mo?@E%&XbZH
z1jc^9rY*%4_EmL&*rW90SEg*1$ozPmSsW8$N6adV)XD>aXx`>@Q%+o5{P5%?3v`>0
za!8Qw5?8-hYrg>$lNcIR7UZ<F$Urgj=<y-6Zty8TDaLFJC*~0W4XXGotBaf4XCtSb
zyHOBh;rch2TD*C3=j;}!ru$NP;{A{?zmJ=lE<#OwASH$c{If}4LqF_ev2wZbAjw3L
zT07ueYrXEcOG-+B6Xaw<!+QApM^wLot*n++WNlK?be?Nx5UzXDNocL;>fU(i|KV}!
zKFtfUl7uxh_5BbnU%)Gi8_)VPMjI1_wSAT~X8+SyrQC6Rj<87a@X^;B(gazjZoJUn
zRU8zeMoj>7H!fx-1WE>i!<C(JJVjOZ-(n6L__>MqxRq`yg*%<0nMYE=QB|QDUIsR+
zm0LgX$vjrfAWec37+Q!01l-Tx0BX<>IAIM84CW$>5NPAsU5^cbH`29ha4=n9@akma
z3WTtLFNZ1CsAlxsh0f<vl?y)>jxW4jkfMwIs@>@J6Z{CC&V?#VLV$>5IqppC&e!9b
z@9-R`zgty(47vg1v#%<^&z}YcsZ6&O_J>UR0NAq+xUGfID@-1LZTKeuKhF+8!+MRw
zn)B8|i@3*QN|GF@5(DL(Gr#P2KR>lY1tO1^=vV_#`@?s=0c_4V^n!=icIDX|uE0uK
z**0qK_?tA`aD!t<mG9md2jd~hkm;@k_*ZP5R`fufQ^1y19MV`3q34TC$^u?Nb1#%s
z;@>L^=2cduMo^*bZycD9afWl`g$HPAb1yVjc-}Z+5OKcO)g_CJjLcJ{bKoZK=;(kz
zga(W0NH)vCw{$jh99H9j*UZcjAYVvMQPE#Imx7u)$n|7BQ$8Q4wJ8JyNWIxDXEG*t
zB(R!-gM&BEu189NXFN9S?s}*6apLgu@^XR*2(wB85cvV5xkqPbYnMkWAQ!9cb=PC`
z&3{2RBE`z_B<(=STL%67xeVeYD5<DAV{&gZ^^Q>%9*oxiEg3n3JrlpmXodh}gox57
z8Wi`d+7Sz1YT1hv&YMy!{z&vkPMELIP&!RgwibM#8H%e#CyZ(peA3RB2@#t{RsP|H
zu_#PH8{auidAPx=Z=QXx>@m%E0lc2!zzV#1XaxYYU^+-iz3zJ6WYU4WRNOX(l&m_H
zhk~l^5!?VDW{sCIiZDsFb$D!H#Q|JrdwXB{d#FnDp7IpBPY!HP{nUp6Yw05F96>xM
z3M|g!evZYtN+~Zdek>}<mX3}9;DFJ0)Mx;4Obxfky}&L7GM2i28TPuY#j1#tHczW(
zLJc^mw6wH+=>pB~*qNA0N4nH2Oj-f*Wb5o~ywVx^yR-A}!GZPIPhg~2E=ixBojqBN
zmN_02czE>YfId-Wy(*j=;O4Uxruy6D{~lZ`MfCDwG1v1Q%vPT_n`=f>oMx%-$Z?iP
zCmQ;c2%Jz>`>ut{nBcb6#FgrxqmS~vfdd0Wt;illt;!g#1YhW5K657_*q642Ib7BR
zSMUMVr(up>KFU`$0aN~Ao<UIo6C)(8l+0iT&P3rzoEH*HqBua)PNdZkZJ!*@;&5*M
zC<9oY0zg2SZDlUtma_|6#=jF#@3OzWiclAtCrj7B9{*Pf1KF-HEvwA6?&5UM$mY+K
z;y2`|EOzz>roI>J4T{g7|MM}5&+aZX5~fOb@f<YRe*$NM4By#H@Mu=RLujCA)ksX`
zafUqv;SF<<bB*28(}^659?=I)H(|{W_sQVzqj~!IqF^aMnfhfIB6=7n_}?1!CwRe9
zlLb|jof4R!?Y@0b0k&-*ouT6pn?4J@CyQ=9cff;`K}TTcebss>r|&X{PB1Ily+~CB
zZnptOJ^>`*4C3E7Ej3R$69W1qMY`hMRq&z;%KQT;j6}QV-Qo_F6fR5f(Msqk-FDP`
z(?eqpQ+JU<)P-a?fXQ5ycbgEv?1hkxM`GeP95^1)#(;+!qbXN7)z_99Zp8mhK>wh)
z@>B+x#E%^!;g0rbxHs;z>*j|`tqIA=tUyBl2+9J1lBlTY`&@b`>|1i|FNIv`j<&Y%
zeWl6W6Mz2vG3&s3{`|@Nbbikst&N{uByf+`J`vL1emo$akpxjYP+4Uq10NzGAtB87
z@81g?(ZW$8mmYZse}a(=Pna}vbn;sGb*<ui7?qHv3l|uGJ!$TnNiSGkR-7ZHYMXTr
zHDh>?z-Y%R627Yl#FxqpqF)uib5G`7cwtUGdt5!dGtIf?4Dh}jgZ&~h6E!i%qVn39
zP0!*d+@W!m)_0JQh^AdYvV&VChd2*C=nBIOlz6?=*mGNsEQt8Dz-_eVXVs6`1Nb1J
z;U|&jDYrMp7YU8dmA7T1#--MnbysghmH%h~3@49_2pDHj3i^X1FaA7TN=mBHJbBe%
z%d=8C7sOh`Upd;tR?bq(7k>SQwxwBD7bgG-9`uWijEs(k^2db>+$R|ICmrooFTQyG
zTu@NZXSnMykJ&XO@D~t_M5d)_a|`Q$UF)F33L#S{mnfT`#qIQ^+DpC6iudHr)J(|W
zSHyPXOL`Uij`$$ZpiH4flxgdxN0Zhc^LwW1wSPTlbs50C30&>Yraqc%z^`}~R)K3U
z-43MBqt(ejto~h?DsLNSvJmJ9u%ylmd+~!UGAzFU(d~WG*QoK}+mj<Z5ypo4t_dmS
zu&UD=gO}#ieSTX%5B2n)7Jp6-(Ar|YK>?^}+U+&MD$ZZ$`hU=ETYHIdVVp;U3vIGK
z6KlQ}y<gWfxDZ-BH_O<(y^A>z%FU$?!DTX9XmAA%QQN~ML9e?rm8RQciN|V6QnHWO
zm5rL38s;(e4dDNh1ik&Md;K4*)qa(iKe4f`t-r8!3<JA&AfOiYfDQS0t#@lI9}9SI
z{;b&<u;7D}02LKZqnZ!atjJfTbP{Bu5rBNHvZ|`bu#w9v=xTXm!@0{%;Pwy>%w5Cn
zv3!&J^u`;E@Pa-P&imBF7%}VD<nT{@hF*y0$-m4NfzP37tNZRtyelB*P+Yh#^xoj(
z^H!A4g7AXxb8rRSFe5~e#Y8C^rEl(Rqu;-FG-`X#nZD88-rwgw$meV%#A3YQuHD#_
zWCHS7eur+HLY)*KzFEi{hiJU)-SY)rblk56i(<z9O0<WXFSBFce==m>o2V@;n5Sf7
zOKum~Qw%#2+Nr~J?w!tG>X^uo`?=8@`f@{2L6SXjiu(fB@McTbFtX}8X$}^ah^u_(
z^~S|`1FY3lp7z}RtqILLkX#)*-FL-yTWlkSbTZJ=2XQjD<y}xcAFJ28v;TL!;WLgD
zZQ_~Hk-_-pD8^qrZ>Nj!M52Q+&*_R0_la!^*+vUFij^fmoeLh<8c4bUM+d22Ym56q
zql&3!I0+w&Du{m=x8IyoQ-d=Agx7f3^e;nHnE5)T=CDs%1BPIQz<69P2Z4eTQAY>T
z+1=fh&5r?Aa7L|Prw)MqqCLdH#bvcwdfyrTm@lFjs*j3_nh1&mFm``|#S%R-3?wY+
zVNrX{7h3R+yttnfOFXON09n&0tkSJtL&|nu5d8@Z-8yV8XWnA=aFbTc-o1Z+S%&p)
z)NWapcUnvdvh&&w_rr5Q#9M}PD}<Rpu`-|?XI5k+Oj&jn5JlN-!{Q&|lK)0<b5C+6
z;=z`bH9{t3f;`Nab!TWgKS&Gem->)KE&H^KKXLXX*zFDEQ#&-LYrulTBn~_?<*ehf
zefVpP+&ozY#r#Vg{bK&9SpH}|8-hKY>)d-8qxDgnGz5cmLspz%8M0Jot<%vl&ozkz
z4zMRZl9Tl4r2ItQloD~zNJ(`{6VlRp7Egh{O=&^E;7`QYsr>wWmHG7O==KKJ6Bs-$
zJA>44jq%!;4P9%UvztWVb<P+#vKB#?UrOb)eFe;G@G;bufK|&IS0XLA?SPFhJo`8E
z1`OUkuVxiDg?xs36}FsjTvZZA<W0?OXg+QA-bjCA;c;~0Es_X$lge^GI&f%%y2jzC
zv-Yd7X0NM|S9$LI*T-I~v#K{+zwY#_$Co+JQYk%hytFfXYs|t<oV7a7*S63Ph8;Bp
z#&T#F@aAeyz5icJu%7BODbt=j%>3|Xx6VIR9BOC4L|;ynI>2Hni$*~HcfZXS!r|;p
zzuHO#yslB!ga#+Lj_bM$b1hOREWT>@w{(`?{isZ)vf43Hsp|st;2FoDAp#<WI+XTB
zvLgNc;YpxdTNqH=DVvhY$jE@&8~OcYiI!TE2bZg>EA|C0F0MUDs(j2V0WAhZzeh+y
zg3$ubX@z0sGw~-+p70*Fz`;Dm6#+XTxFQ!^C;H*jr^k?JP_crNpu$|`s;BlkP2ymo
z@ks)|C-<i?qI}hF46Llt_^d|EKz+h35XM1K$dSY2a!6LFQi`Yx{OuMq72t}gP?4;B
zyPO7k9T>0*RzQNnFUU#!aXi>mTxi2bd_U)u6*AmJjLM;7CAVI}fNLqRTqD#9PN&a)
z1>s2d^r@CMdz#HFtG#8XCvWW0*Af>kd^cc0UP-j!p`K%FbW(w&^T|%(7jw+^&KSic
z#Zb_2IhB1EAy)7r{>SaD562^RviV9iS8hW^67=d9s2$Sg&^nqChyntErtPD7<8Po9
zrE9*MSLUAwnRjnuJ>J^U>@_qN%;)~r)!lCrzIERqk!)$(NYAu3=`xQ$#r~d^<$8BE
zS=cD>`gL|s-84W$B;DM&Nj(mqfC@Cv=^>PmL;3nXGl=9B3riFT%!>zqa6?7>)QgSA
z3zV=~?I2tb*<@ky1`xeD=948V9caLn2_$Ve=@Jk3clgY@Fg#AXdB5h{snb2$DojTU
zLY>RYuhAABqeVZv@6Xka56m%aY`eMU>;p%(Di1MuGi_8TEWP|}7e(Kf^5g_@opTV{
zSzDPK(RpWHyz(ez+}D0iAwW@mt<RO$DhBDRjTm;({wccV42U=5%ylD`pxwsN>?t(V
zH4H0}_7B_a&cN-oxqwLjmRyjoo^N~_wXy;C6a1T~Pj7korjM41+^Wob??Y%EZa(8j
zPg?3q?cy`8<7RLWbkowV*_d&&KFrAEc1XAyZ;%k}?x`L3NbqvMYBQ>3Qm^;|Rhmz-
zfELl$&kvwWpk-!p*pvl=0U;2q#K@>3Qze_P*q6u}1SBY+&6D03NCjV3Q(HS+ZY%>*
z39vy@K^KTkqu;{?@nerIGl;D{Iz9$@gc+k-%!!iN`XCIr-ub`?h*r?R#sX$P_zw_<
z$OGgWE1SbmR;-!(dRhp$I{ak4-{ZKO8f3j>Di?wX<<;dN(sj^d@E-lfQufFuBrP0V
zVz4ej6US~&DM-rSb8?DT5S1x)>x_7uKE<or4Ig$V^t01zpVDDBwMv=t<e~0AIWvcr
zHrKN)UXI=UDl?1i7^>q@ajl-{4~}YhW$4514x#ebGw_6EX^eN-+As*@n&EIuFa>0P
z{xjhdP2fR9Hl;>uqGJEX@@mC_cqD$a#)ZzOC>!lG$8GM857}fAHlAu<nJ=u{<P{%d
z;#`Q@%MI+myK7&KPpDj~PZ0<;326S!?JV>4mHsMBqqo*;W=2d8!3SC2ulVWeW{#Yp
z^ZgEp)HR_@S#XlEn+~){Af;BV)T>IxY_^I4bgG~~op4QW^1M;j4Avp8RPP5*<T2wh
zbtTE35!gf@W5$3(T-o&@Ho#aeaU8V)j8Wl~lY>LiWD`g}RUQxlX@GffS&{wr(DD!_
z4^0nk&GmhC-16#{Q1}gGZG33&EX<&i3sZAlmo<|2mJ-)*7~TvO#Zl}dObtiz^$5$w
z5AD_0I`Ur&iwhVBCiXZxDoK#Me}CRw&$Nig)00B6=<zNOP&rsN%RPP8REiJJrL{oa
z`0|qq0|uk|$%!Rd@TY;p$qndPH3f>+@HmtV?}SX{;Q`gnyVILNJSWCd8?Oz2nI9db
zEZBUxJuNhra<4qi8O>WbO=Te}4ni0#d<-=FEhW0M)bcHO#no;swP?OR{>=RXD>+8e
zOc}>YdI0{?g*8Khy7P2LB7UN^N?PS<h-zcY&eoQgxHt+7qF?6cKr6>$0&xim`YJzP
ziyN_E5Dk|s({>N3Ho5Oo<mfQcpk_EaA?YBk-JHsE#!1LIJ%JciyY4BIuQqT8ha22Q
zLv7j4oho)o+OF>2+HjK!&G*UJS+m0wUsseK?<;$9+bqjAuna`O9xS|p@x5`%JKkSv
zBA#lzbe@VwaK379vPdArjPJoQF!>RHQ6i2S^x03IIjjfr3*9QCB^J4Aj~29fjVm4s
zk69sQUA`T&ihZ=RBf(jF?``LoMd0Qpr0^*RLBxpm!&;=n?i;bLf#b;yLOj-u=Se};
zC6l%w6OV^qtv5Zrq&<U`rrbezK8B>Pgy7G(u*VZM<B{oFUE-VO`L@GJs%Ks@4~U%R
zkDKlnaI^R5cmPr$+PV+9pE>E9-{?Kt2n%4c{W2S;dv99%?AbHfl}mc#+w%kKI}Sz=
z?aF)nhXTa=YHDc#8EO7_IQTJtI~@*#h>M^d4M~w`dA!s)A+^=r0|o>%s+`NKaRl)J
z+@o=USG7BLK~I~j*B3R2U)6z|4`ON8ab2=bAaXiPFu&gIj1_FJpdAbhAnfQ`(tR^G
z(@qW)g1b$_pu^q^3RL0B9%tAauH#W?U_xpV^<L}Ng|#{hL`tIH!@K8~t8IbVQCPk)
zxMairi`puUKc+V|9zs1OE-g>hOL9dDimBL%7tu-W5me%l7vfY)1|%N#J`$fsF4ylN
zQISv5nP9Rxa)&j203aFDE4R_eUmq`EV>gNlnmW5X4ux4dpzLy^&dKbL)NgOck4CBl
z|J~5~R4#ZNSB;RvO0(k}01@WI57Wl!N)|FHoWljR;2FK5pa6j{cxZ@di1ep|ptc>R
zNtwm!zvk%uTJL<*?mHk(?^6I>cbF*-4HEM2!BX6T>j7@BORYsY@l;mc)-5_0j+8on
z`oUse*>p;!YJ=fD`y=#&l2me#cUhc2+Ae%P<)OaaT{v9X>*sb34o!s|+6<nBMBi=8
zvKIB9>$1>x{Zwf;w_NUb8mjyHm(e=}v6I8^qgF<7h&&NrT!;#)4)?qgRQ=sVn4W=V
z{ibV!zGI$*q^t&YZp+!x7B|Ko&E8Kf^U&ZKVq>I^_O-9Qd*Q*Ba;w4Ty&Z|zzgDaa
z9&`9VD)PLB>x|T0ezNuL!jw4eUYSwU<2>jM*Zy}vkrR*>;y!8`LzD|oN8LfDDToc0
z!%+Mf5(or+tj_z8CQL&^1CUjHQL=xuxX8)Lfx^7Az!uyEFgNwhjbc&YJ`!AhE~((c
zUGrJlG;~+-(1Lj`=cQ|NG7ILnPP3p{9xj&MSoc`#g%F?L-|MJNVfndHbB|34W}>Vv
z7SvY`^-8~WNt<IzU@Z?yI0|62keZeqzY4RGMsxeQ9spr+qwV~7lTrY?V1iVAGq#5(
zMWicj@{x3*rKA<-thDlnz`{dhMZ)XNZHxoDjxqIoZ`&t_n#>}8k=TP?Ki)W~_+=H|
z|D9rjXj%OQYHn4>3X+lfwx42{u@9<={f%UGoR#@+w%96^x%o^2Pz(eVK9QZF0ND(O
z{TvWT%h{sq1PD(uNOAZfFA)enow*S5Yy7o<KstaP--7cP6q81kHDD98@*upgC(@Fm
z^{;8cI@^6HflV_*C{CsQKdQbmEUT^S8U&;S3`!)#pi>$_6a@qnq`Q>vZrBK7AfeLT
z(%s!s(%m4^64D{?&0Ej&Uf1`-a}2oIYwZ<tjydKStSowufA;VYZr=OKUPD8fy{qKG
zo0-&O5n#K!rOf>&?rl-ojcX5LmCZN@%PhFWdt$PQ-_8Um8IDpQrfLStR^EB+_1H!b
zn@w_U+6ZBov5lTpZ1JUXc<OMKw-oK$ReTA)IBiMf$dtrKlX$bqU#&|yj5_-J=XfVZ
zv8j=Xp`2bM_5P`@^))>9g@TivRilwrX?ye2EwY8^5z&xaRvyI;a>r!R&&&h=s6~qX
z75s;TgZtrYGDLTGxsN1=?B2b5&XANblQ^FoZ1zAGq2}*Xl<4aN_i`n>2vF(-3{mLp
zaK7W&ZTQY4_|fNzywoOY4$^POM>BM#q|pe^%ae!auH-#W*u33Dt3w3h(f$gMixF$}
z3ASOzFe>8ZLVhHy=8TQf0~z>jyr`8{+=z(pzx;(AlvFUnAZ$w`a&c{djon+BEtz;Y
zr7`68mey6m?!d`|c16*P_L-cXV87lv@=MMy@fyZj$biy?66-XY$|QNhyu*QNa^+&H
z71Gq_t<8qBdW-#-!5UuTEoW9`F9%ho2v>LCKK}VtzkyY|<Cf<l`5bmvNz4q*<Bk>N
z$<E7v>w}hpDvP~Pc>_eMj6lJxdVkyur30iCc%9FZ60qg34r|9b5455In?`Z*rrmeC
z{2Z^qqQHFU`s`{d)?KMiuMO%&?9~Na?fnpal3*8Jl6c*7(|4KNrf<8*Wp^wsM!C(f
z)ZX>ay$HN_W3*FyWzSU-!M^*=@Anivq5hWiE1PZW3>Ed*&04IPPiUUV0rpkG=Emy6
z&ZDVcA$giL^2Tffi=`a5&UjwaS(s`QliakNut_C%%_{5H7}31qnNJrQ(Io-Z^7~yu
zuNZ?rM6gPPF6pTbdAwU|p1D$4so>7=?hQsk@#?@_CT(AI5Yvq-qzI)%vNvt*Ru|Sw
zUYbOcR?@}Rt1u?GoVY0nlxMMwL8D7LoNWP=m~RRQM1tn~rHdFKM}lfrq)!vvkIQ-o
zdfI?mi9!F{@>o%t8Yjwh!g)(K^dPc>?Ze>O9(2Qjut?6z-iXA^V~zeBAr~Z$>`b{S
z`jYcZ$dJCa>v_wm@JCMC@P~Zja!APO`E*zEs#wFS6xJKJ`r%75XEt87L1KfI(i0$d
zDa;WvfKbvI?#QuGzE8f%MHbxm#;54^gV?C6wr)m^8>_xfpJ=R)&NGjV`)jWgv3R_;
zb9=>q+DhBKAt)xZeb?Wm#hCKfmhpmY^n~aNKazyqRaKU;V{<hqjN~ILqrXo-Mbl8=
z&-ta^75gDJ>XBZcS^?ItO!|H&(=f59a&-g!o$XU0$FRj3(cH@WM~4Ci$<e++T6|82
z7rz9?a;=rnvM8LniRO)TlDy=rVjfRv>q+t|Jh-~8Zaiq}cAht~mNi*RO@8UE-$9M&
zd*oZjBmC+Gu$?d-$|T8AFUj2YBqIy=BPC2tO(mr=n``t%Rk@&M6h$=;!qnA!!x}t+
zV2MBn#rR*ZM%7I+igFb4EBe25<nh#O93S6hF<<K_y_aOuE)<?=emGo&ZK~AqQa?_*
zYV|xMs+ejxw6v-6f-W~X=ko5T88fn^5br!>GRzubIOQE`^^DG5EOqq0Ks||X#xW&5
zUr9ESpBfY^^+8Wa<Ee}~X}5N6SPPeevi#RVy_OMCnU2&f8iim<#yj`TSfY<E&8B6m
zfa8Hh=*WAHpj<^yiT9m9jpWeCcQ0NVDqgt~@YrMeoQ!1XmEM%u!?XdVsPDIsH^%O{
zlCR5n1!KFJT0$VrE;8sOg<8hjckg_;^ik_jith0kP1$e+k2-1ct*-*K+@^yR%*5no
z$?|%e=3_6vl2~2JJ{;96o-)6xfDsdv5#_u7j4oIy=vDD74?QaUr+s6rnfviV9?!D9
zAPO)5{EtODpA{6sG87Pq+~eEze+<Zqv_Sv6(b5ek((IZV-|!^f3kyN7aj&_?Qls+K
zshewremCAU7^gEt8UNx=SK^ypYtX;-yG`vSAI;slhb%4`W6OBb)+Rp!otQ<&Fa3IQ
zk~n3(U)ts0zV_NlI&}|yMLEJIG^2S7Z#(PcJd5rc8R55UMMnq@{69E6ETA@>m`sPr
z-TM^x`sfG8rpC1?w*z(SgB^L^do1q1o=IH_6%VM&qh|`R@EjJxa1SA1t|f&WD9fF-
z{MhwVa7f0knq%>-dndIUUtMO<ZHlly<7QnBT_dY^9ZA&eKg82U)|e`}VJ*VC2LBOm
zQxl(&h@pMt?7ZWHSzutGG51K%T0?iY%=8*wUmH)<g-uA+C4g-VIS}c=6MQ_pm!R0K
zG(F;sS6$j&zn{yrHExnWz#T<{t9LQDTTyPdWMEWF1qm4v+x~559&>bU>7Yfiwn_W>
z?=i)DJzuX@?eRw&oj!|iZz|twzcPAXjW>Bmd=RmGp>p=eUYN2+2`^yrtb+qx!VcY7
zq}(iadRB*@xaV7>$Y*E%v+V$;dYR|HBZ=?zq+BjKVu-$te4Tma)O+xGZd5>9<XjK?
zO^G6J;6kIt^dhztgEuAB8I{ItLdld_FEiY%>?(LY#1G@N)!uY#8szyJ>g8m*vvRMG
ziZUWw)SXJC%SBuK=!vYU>G~M1KGZ%9UNto}#LiZ>u0FL%M0tN$@M^T*i6=Kq%V^^V
zs99{D10czaR}F=JenD*!#Asq{OmfqimZCyjp-T|Tj}Jq+tUB2lyLx(zevKCVMLUP~
zF-b+=-{?xDam+tkH;{T=SwSj(^w<iKHO|Bez{4ZB97}ZnUZ2HNqa|;LW^;j#W?$L)
zE6cCecJne8r_^6983#S6AeUKsEWu^OC9&{G9d}=b5c5rGwo7(<b6g%@fz6^Um?z@c
zdVfBXO6-}Da|y)TwPvCzHIgr6^vR()6aBuJJbAF9IQfT#0Y_b~>XDXOYgs-Qa*f);
z<m}mC*LA7t2CbY3t<wx{({aMp#=g)hJq^0a1uN|Yb1KU34b$zPl|9BG5ty3IePSh;
z6jN1!rF|{G<WIrcOYKXj3*-HbH-?8#HPUW26D0Q$J9l2c#KTvbx9$mA_<g@L3L|G`
zP9HH|eQ+H!Ihtug#kTLE6VbvNLi76ab=lS5PikX0f?Jhq0<$)|hRUj&wk%@~9cbJm
z(LN{c#kZJ=j*;O|^%V`zA#8sFSnX6^$5O<ZVBP+cyV1I8nnrx)>eAT!FoHy)zo5un
znDdK-O6}X=wDk$o@*}r?&!;L7Hu(-QTI&=YV^^g3D3QrcnBdSLS$o0RNE-Dyc{;Ik
zowT=JZ30iSaSaQuRKOdN-n{YaBZ;?N^BHZi#5A}<T60RStDvv9y(zyC<!Js$cVu{P
zMhA6I{<yfUJh@aVlQ=z%>oPdZ_&!=O*`)EF)wB?h7!(#p0JpjTwi41U7Wdp#aQet&
zChVdhq?Kc&JWIpy$4AO5%TJun0y(m7U)rSwjDJCaY|<W|cj`3$UGJz_NaqytG!yZ%
zx;FQOngaX2c<ud+Z;1sHfs8F$5<K5m(!WpUSQO+esvq7YK6d`>KpTsSiogG<A)Gv$
zG)}#Fwufc#&BkN-`Ih#j5ONa{ZpxlD-F@estzmoe>E|nnC&;5}`RcqNt#DljJP5Te
zMet^h>jCgJ$?K4E^}I&4ZDeGVACj=TBD719o<vM$ETeS)ovFkWQ<tVQLk^-o$8m*I
z?U(<4fO}0;(eNJm(fZ5`vu_=lccO-{;oNw}Y%1?3#b*$UQ>qnw+_+|GR<57N<*1Gg
znV%lA@bTkZ7Jh<!!yChro}WDEGxQDpPhK>P5>dBwCm!phUW|>KnLg4!T1@4=ysCU~
zhiY)k6i3L-y9kM0{CSM=CSX<R;PghD?_&e87d5!rA_P{YKDiquo2t5A=go~DgwWu`
z(7hLx`%!eE4Uzrv<HOER^BzU&gC*}!#H{Exa?OzWG(<u|rXf|bD+s`7!nyD9^dBZ{
zE@%rASe5wD5th&_J)^BAAvkh9enu@GJ<+&otNt%SdLR8HIkJKI&1Jsm;5wQ+!n|sX
z6ody^2$}?s&=j5*70tQd3SD%|go&b+rH5&X{NU+5bZucQ8?2#CAg$%aC>rKwBD%8!
zPFGwdf+&J)%!hRJ=?xijI4?$VY79wY$d+8tosh_4CfzL{Np5HhlpSW8xn!Bd@@&6@
z5z^5@`=MX@-mKTResYe+=X{s3Tz{ti>}nbxRjpygNOMbT0K(#`Y@$@iNfq}1{iHA}
zzAIWiiP&-Gqwm$*O*%C}fmf!?gpok0WW&jKL488iCH>b<GUm6;GUZwFq!Q7??ljub
z-z<64d*AKeZm3(Ah#YE@<Ip}ywpaK*3%)krmVa7dhXkBlSe<!Xs+oSn-uCvv6H-7%
z`l~f=WGy(|CjZk&?ykXSu;-u0N!{@}qcs$(w^2AGGL-ISGG^dUQPC@C6o?zaD?<^{
z(YA4;fw(`X359wxdE>TI!Z(|Jzn3fazx?vsjK<iL1z(GwSi;ja@ZKvY@%P)C-gaBk
zc-fM^-Il%6B-Ot{fBbBDIU0{4V9+vCs|`pmWykA-mjZhej|2(xYB*e34@-!C7AWNv
z{ix_Ex8{(EjGb&o{|o8lBPva)@&-0F%kCVngWush^|(juKW<ucr_8D1BJw=BT6cQl
zM#w&cyVKK1hTCDIZrspCSh#2X6^(GoyZgO*vSIyBa>aVa=aCkFqkr?$yw{R+VH1#W
z^H8x)`$|*8Q&mZTR@I6OpZidQp+zo4=3G@WOk!!&abYdU>KXz4Hh<CD&*v($Sj$3{
zMC<R~6de!ceHe2mZaTWfuVX*NA5cTJ{?6A9K<LH>H4QUOSPT`yeY4rG*&K}ZUIf=_
zUskD))S|agpB3I&YVSPqDpZ|h)s~`}XT2^ZJD*$;>Li~PBD!_@I(*{&?l$IqlHZ89
zR@Kk7`Y!PQXrHP3o<Z<#43M%ffpNYw1G&^oOOl5*rz;1>n2@ILJqV0V7IvyRzbQ;d
zsN33V2E_zlJ&8r7c}mS<`|2!^MxyowFr%nj@?Rp=cc`W~8w={2K4eCKB53B^sfz#_
zpYdEuv3mZKPx1*JlC}E2-z}dJ=Bj@~dXTGLhp}?a79ujrFE64^v$~^z<AzhgVP?Ze
z;%paVslByA;#$Lu)y^ojv@ol~n6z)_#IVmSTM#K|GyrHPNb|*yRwJ!A-1rBYo@Gb&
zAhUo;eb48lpBMgyvO|eP>m^C@eVj@6scu&W$HzTxY~U~a!UZ)Fo1KF@NP-7G9O#L=
z7NswWx^p59#|A=3iG!^<0{^1Pq_e-)-vXLkygYp|phnaD?!U7yJ+j&ib>Y2`lTphX
ztt0FHE=QWoQ1$k#i0?CRnQtFtX~3LA$$@Y)%e2g|nAR+%_kQb<3s_1~W@Hsvt^A~U
zw8oY;d(4DvX*Udw`0O_|GHc^I?mP02YKW*{ld;lb2&xI^{taF)|5()RtLom}wKrRU
zC6+`TBcv4)pvkN$CP&)T-Z`|b%Euqv_hkEnT}sTeoA(8-DKZIRRO8|h+)7Y*QF9zG
z=n^U-;o5nP`Qx1V{L=EafZCjc%5eC|n1Z&GDDv-}>EV9FtEQ%=9`1k@`b3z&(zwr+
zUm->wQ|#ix7bm+|R?K>?-d3rxDQ78s>Cv4baYB{}DNk3+&{vus!pp)Y9A!d6m~Ls~
zMthZq<$r^f0&;utRGq?#+~B3QESGBr^X8um86qmS_daR+eGIzcsP49Q2cmQ2?@`Lf
zh<}3*5UYILP(PTz*VC9MMXVo7?&H5uzc%ULy%{t&pO{x*F>SidwH*>yaDd%t*U&L?
z`ug&!xTiRw#$xUd_;<TGUCkVuAqt+H6Nfi1F(@-CjWy}%bj$irN&^De=kuHc{ut=h
zEqaXUa*X|P|Mwy<BT1Z}F605$mC(5HE7knZr)*;ME*ow#nWg!m#&ECvn^_Bv$@~S5
zG}81Q3E#(Sx^C32;?cZqXkyh(soUPO#*ZMosPm>NXZmNV!A=*39RS7;bACVG{FPzu
z9<P69SImdsO*S|3;1gZLnc>BPz<Ra@XYQT1*IYWn1^np_Ig#sLy;p>DB4Yo&YSGCU
z{m(oQF+D1wxhnHeRhmuwZv84T`9QHJy5FU(-)d-<$i{~2Ubw|2HcJ2b@D-hx+>x`9
zy5(76KTPactNRFtQ*<@0<CF&K>Y@Ax7~0TA|4OdLXoaqDB)z=uV~Qt-L+tJD!Yo2W
z^$AO@UPwo2&u6#-W{aFq5Nn5-kv4tr@J`W>P)6(0Pw=$GRW)`#KN<fW{Q@2Zfrnaf
zJvzpGx62^b|M_F~*Lw{GS#Q^p<kMbjp-zp@xHdjtu2(}r$VWay$=~@TK7E~A>Wa0K
zFk3V`XCv}*&izV*$NgRt#m%?UyMp=%!%)%o`GeMUzWU%tCBSRQ@alU$_d86=pN#ZX
zuTtInDD~vUZl|*kQps=6V|{!C9ApdI(9Q#2|9m-4uqt{&tF=~Zg|RC5mHko8ctfOU
z#PzJ|uvOEgyQ(o>-x>n1SWQ{&c$fT30;~Nm-o_5Im`8jJx8{Rs?6%&m#~=`n?Uw69
z=Dx7|w&~CBvmkfwWhLd>LxHgm*Gpy8aEivN>$67Im_5K*HN9$aF6a8*^qqym1E-g|
zCNI~q%8GoqUK0Q6`j(Z-e&D(LM&7IX1LdLR4FSnWD6BJ%4v&<hEpI#;_O%BW_M&fz
zF!sy8$1!_v%;}_shEY@^>T${C7o7JoDcf5B8vNGnv0KUb79gW<$e)yZoqA5B)%I7r
z$%d*GK3u7-Mce%5)!wYNDe|<4Ehks9_V4ETv8WId+fzn_v>lg)%YFKcU*_Df%e&>G
za#%s{K0#pkuJI6UD}Sju&rOYg+u~&@Wr@RTfE0K)LjIzOR6GCv+C8V9@AC1Yk&%(T
zpjOKPU2w&7Jg|FQB=(d=ew7er$NeTbPW`Z@K`}XpT_a5U5MRc$t-oQ#{cz{%uQtqU
z#miS3J!nLHgVVBdWapx?7&>Gu6Tf%luq%z#zO&`uotJSH=1BiG2gs<u&&S(q)2e=R
zotUcOnL-l6h;YbKTfr{@!AR`;Y)aq15pIlpb`x>EAfZCf^~IdOd)o4u+5;a5ia$C3
zyd>m#>Eh~l&+h+`37nZd{3ois!)xwg2vlJ18#Gw0SZ5Hs<zT#86cTNHzft-C>+hZ1
z_mpIbbbIKHntqJrYvt}O`5VSd@-CmEsSX7YQ57fW1OfNxrz^5#M|176<whH?AgDBm
z{(0kitc*&=qEpeLHS1_~J;Zs9wtBbS=4KKFqL;GoxDH%7<dr+h^oFpZ?12M4YINVu
z7EgFSK@<*4%6JZwmaSy-MXY0P?=$Ju6?{|;kCZ0s(V_4uE)M?7otXF!6Q0P0+KL2s
zr3MJs&<Di(=FcYNbkd5lv^<2JjmPZw*~Y%d<Bup*3tuZWQodh1zqN76(}d+g1CXOg
z|GxYrLa013>!^ZpSCw16#(KS^!j6m#g)<AzczJy`zjDtZ-li^cp;28sI>k5tLUdgb
z@Th%(J1>gEY}`-hh2X}^+=kle@^cCPI`+4~Q<vKLXtTFhUll5{e$jzeUzZ-P`BIq?
zLx{yf1G+84u;kHWCvR3n?2i6a$FD{^ql&USU+VenD)@Q1OXBG$jWjd|0wxRh!WQm+
z8f1%aF5xsyxAztP$wd$Fp*p1wNN8<FZ??Gq=zHlaO@v9Oyiw_6RUO%*nowWnsAAlF
zhsu8=`}|S8F20<KKA|GG@S&|m%M>$#<pq<}<+^9HWo6-<21ovDU!RW$m_i}DRA%n0
z-p&OpclYWX1-O{STliA0*V1+Y(s17*w!?u@L|9{N%O+n)Rn6Zo!r>>pLU?Ws&p&I(
zo8+u(ddaTsspfxwYNO*by3#I<9CMK=Jk3?PNmeHvH1qbGzuT_AETc&SFRARPm(TI=
z@V2(MMYD!%SNiY%`jyExW-(s;0qCK@lEHMpTgY@Y9nQi2L^WR*9XO!(f#Q$?Qc(Uh
zB5_b-xM;PRNuO%}YxFa?)Ab_*1MS`2Kt>VB)+h^XZhqo<vd#yDdhZ5Zl(LWe-jE8)
zd+^JbtKf9hMCCNu0E%NPqXqLyDC4`)N(bhpo_PA3v#Yo?P`Yu40uM-UB)<Iq{>p7<
z;M_0wFRZps>)$J|oDw8u(`*BV04Mn+z!BfIQoH%v%zr}?l}FB3CG1`fFZ`|?HrIy0
z9zxVqzf$633fJ0bFZ0-Y(j!f^#}8@@I4!>^|MTV))s%&M$$G^-s*vu96VZ-$x)Mjh
z_3!?-O{ZR)sn0st*yur3AzQV;8&zZi`U_KQS2Q0HP-y3aM6_2NgWH>>*=w37E9{t&
zDQXfDF(5mfAN3%kC1zw`on6iJRXY#-cnZK27LY*Yl?Q;xz<kWszy38|lGX3g8bq@M
ztnTk|!qH#|kcHPav{%PN5g-JL0VpDK|2m3<4^k8?s`*|aA;f=9PnkEESy}Jiy5$X;
zJ4)k5K%zt`lz{X$QV6{CnAq4nL2ZZc0csQT2KXaJuS8IB*6Z>k7p5B-cSASZ7?h4l
zW$zO@ZTRA1&(->L><_f+sa@3%Q=X=O{$Xy_Y?0>E-_;|Ell6YWMs#d>$v`b(Ixay@
zE6|BR_;hyw4aKZbnvRc-?gWW2YdgDSmAnp>p00RABp&)Ht^#>O2RRj!le+>OrkEH?
zQdadRCMJyF*$QQ-y775gFr`4by8%`5<zGq?a5-XtTvKQ^!Ul4IsAgTji^&7T{XTv~
z{o(uoIe(W)z0Lwf&`1yvhohcFt?&(evcLFk_{!$yW)3jy$}&W{W#FS>R{Dxf6rlCj
z2Z*Y9BX;1`3XBv1Xj1^;GbGk7U;O(jHM_<qMhO_FB@=qAHx#QP2$vYR92~?|qj0k1
z69&^S*%o2U6p|9Iq+V7?a6kQo`3gf<{-&{_H9vNV`16OVpSrpw{)L<9aJhDfo8gnn
zBNwepFc1m6;GJHQP|bl7&f|m4AmuCr@D_3h4VLn9#RuU)kpbqMU5^rz*W%LB4p`)&
zj38fQU|ZYTW|sF%O~t}uz62aZ1zF)1KrNkye$CiInPvf!u<A_JD&*V~4Hmy@28mB{
z@@-J%(1-2j@SqiDLmTnfQ$hn0jRa7{m^X_4c}x5`s5{RomNhkHibu{|#Mru`bu67m
zfbR|s3|{f^G(hU2V`B^a^y!oO&i`fx74yK%06<>6OnpI3a5RQ+K(;Z+06#B)$jett
zS4?(chPkX2%j-&N+x=5w;B06Lv$~ZakJsc5I=%jjMu$}*lz6#@n(*wdcfKUn%=9cZ
zdz@cX^a@z1Xei4`l__X3F-N0zT~3rLvzn?T0lPNnwvx{%cT08KsKE9D1$srdKS1xP
z>Ue43ay|(ZwxKx|(^8WGIy4lqC8Jy;7`7q?I36@h`HtF7Bh%ADC%=l&vUTdtqFS;y
zeivdD>FfbHyIDUo5lM`M-j-}?1$b-a^6`vKPx&w7?{qU2IZjkH-mqxM^OIdtdT3_k
zd=KXvudmZfDC{?rpaCeWH^S@kUpA~%ykQW`w!IgI-KlV#E?A*+@_{l)Xd4G*6a?!H
zArVva0(W4)VRR(^x%z`9lvU@SHJ$U!d%1n-nmNwfKif@%rl8Y`$%+OFaVXL$=$t~q
ziAM+wC?{<@S;2mD*iEkpUO2j<VA*&GS(t=Bx!ih0;^TMQKw+{4zB+J6D_c8Bzq4&Y
zSwUrtVN9MtjvIA><|2c7#{6v4mc*JJK5QFXp4iseO*C|8fm?dy)vkClKj&809!<QC
zz41Wr-*}z`2wmtPJp_n3E^ZN<iBr@R7g#owS|!7hxzvJ=zn+4ZC2-m9f~6sJMxmiR
zX3$XXF<JZ2OphLN%hH9mOpu55!s6`gi-q-RnEx1Iw=fkQkI+=<@Io*TKm&RfG!_W4
zS8s`m#g$0P%3`M%d#PL8<KjXG^$W?8R~=-|oo#L9hMZY$)#i^}w;*pSU%5jqoJH(Z
zvpyH#=)k-YW~bN`y*~Tl)z7r0Pwqd2x&+Hi^EDwE$FZ?<+j-^%mn2$A*)B0JDKc4i
z8;C3TZiONHri+3h7=pJgF8X>Cu+^z#C_amS5KE@x{+WbXnZsdK@fn?hnA_V-!&+At
zM*L1zo!S`a6Vb5$wF+;b5Oi+f&}xRBo3R=<J`|-J<mgZ*G%(lH%Ph6lINjady*hoT
zD;d~}!^jM>zQo3AF`gkM?_Fw&jnMF2x5OnRuF=p~?XjZ{-^9}+6E&h#Y>kMK<jw(V
z>KDh{rxp2@!g2rY6oNFdgUgy8Q*%$*nn@$bnlb;RWcT#_4%`@)&W)p!iNSfOuYU&(
z8qt106?@Wg!a}K=f4wgm+uG$y>D!bcQ(O>ngk436fXgW~Wxe7cciyZySS>~lHag}j
zp|L@brpV4{A+vq^tVZcS0<%Mr+B1!NZWpbrt@C+yP&q{|u9h4wfGMwwF1|g9qG4zx
zBc;DwHjjDyk_wgJ-u;(0y-6o45%02)4ft7xAVl#oJ_3olDm~Z3H15qlL8zar^WEL;
z)+YF3q=kuz*_$jugz}F7N+eteJZS7SyYiv)U_(q;|K+Z<^G+8Z5Z${<q(V__;mg;q
zDWv^wddp}Qk*!|h2YaKh;8m;o!S#FkI0-R2`uabW)~Rf5ZBh061eR%KGJ-1uVc+9W
zkfrG4dq8_yULcw8EUEX-eKw^CO>vwgbh2+BH49`d{8u;QkUjW-VN5YDuSV=LeVt?Y
zRczWdm#w+$FEYyKyO4z8(NcY0k7Ii@c-1I^<^@f*z_VCj^g|mLW&+STSzwI-y~OVX
zj9&#DR*B*7^$#|t_PQ5N4`z&DTJIoqbV(!3A4jhTh+n;W)%nlwR<rLpnVRRPn+Qg4
zmI}M)*^v<*K0bP?$QHaN7jVdCF0sbJ!I^!HO^t@~6{WoIcqzFLESPTz3PwHfI0iZg
zs3InS2cbhpGVROpFT1Fa#zvw2dAmGsl$u_{n?iCI#<8&f%G`*k>XTPit#4%0ejk+|
zFpPd8R1-X`<Lwx~F~k|A>p{W7Pvm<0u}($!@af|`t^Of~cpkt;ZlOUR-kacHg_3z4
z&m#k9-#D7}A&TJ86D^tNwVYJl)cpULvEyScN=GAgXJVj|1viSo!L*xI=&KI=)&)}J
z|EIqO5dEXSI&>!{z8v>M;jHlNz%P;Gjmh3(&Sw#qd{nSd@ZTmN-2t!mpEQQODGmyw
z@dQAx@LdYMUMZM-uMIOLw7yc&d6nPYo;0+tnab)}%FOL1izq}uMP)>)r3WaB!~EK*
z@(8C<HzlyAQAB3wV-ip9n_XNK2YFLCvs6aQtZ1M!GuMd?I(1%xz1Ooip-6CM0BS&7
zTpY@(97R(nA@PO!M(ljXkSX}DeusNGLdfMF+%gi$eZ7SS($pTiawy9qrB2Ko<xYq|
z)hcZ7HoYTiZE3;EL6ke8-K86P()e9>8KBv7PcZN8dlTjn^NV>hpbPa)D>&)__{ieW
zvMl_V&o$RyVIfi`ZA$&%R^zB}j28bB&^%j(qF%DlA3xR}1rI+radRVsDykXLt$&l0
z{sro>)M~#eCYH`SfDWRo>T*R{TXDm_#OEc(GCNXch4bs`_pOnQWU-*jHw-jO;ysT+
z9}i^-1^r8YSmUd=czH{|os_7vf@ov56Z`nKSV!gnY*T1b=@e{L$Qtsex*vwxcdXbD
zM(;b1iTaQ@Kc~B<QD$lI`$r4J^||t!LF4<^U=S33PM_inue@69s8sP&etB4oY-bH@
zf$;a)pI?0YtW*BA<*OHV-k!6hxs>r<axtNNRFikEEeqcXqtOhaoQfeo8~50ayyqWW
zM|IP2!iFtY`Yve6V5=BS{t?Dqor)uT+=RO^fv!lJ-9!5(famFJK6dsXxC#rw#{4Y~
zn52tFbk%v>+<0Eo7SQcIh+N%r;%4t{x%!^HzR2a;!gijzCX`*bP4F4o+OXVsE^WQm
zrtVe5SLe}RowcX_RsH10oOA8blDX@1if7GrbffOCvH#tSmWuG1%Umwx9&{rXS2$=b
zpC}`(#`hv&H>NXez*Qd%izPu>=fNQKIkiCw0z_ssAp~M})Y+w!rL6Xsyq$C})Q|1e
zG1Elnm+wJ47Pru}#1As|2F{>0Lv6d>Jhl9OIj;LP!n`tH91-Azt!pL5lRY!R)v@<o
zv8e9~Je4j3Y$qQ04P8B`um!bpo+KOYUDyO0<ioYRI<)1PItZeE>A_t}(xQoY0bx3*
zt?lWL`T0K65*E$!Pw<8J48N0`rQJr&dFo8^72~PcxghS*aIJGR#Gv7;Jpi!F8lAT;
z0kIJ}GsK{g1R@}9+}Cu3Lz@Z84IDa$<y_n&SF6^^Zx{{N9fYAC0zT_~v=7jOrNdY<
zM&r1QWKtV9X0Q+}ER?mmoTlb8XsH3qqG1_Sj9l(F+!T3#<|wVubptRnV`0y8%I?>Z
zHoD%QA()jYqaelRiBvr~mY2f!(|dW-AK$C8bRW<9&4Hj-7L{Aw^ppO+B8MKi|0baM
zc@xk|VS6i^zw`A~5?Ji;9qdraU}j@mM6&W~3mpYfvlJ>Au-fI7w&^X3@tr@l*o+}J
zkwo+}dKg_}R$5|P{d!Ljo&A#@iXiA4@dFxBZzC?XsrLDuXp_0VFf{WcU4zqu|I$O8
zzGXF!&n4iurf6POS*dKEmAfnpOM$8O;P>y0^3!k~&$)`FxFYC%0Q~t$C}5hFz-3T7
z?{*@ut3b<Am{qpIajKgXr>!9l{Mi^E`6*e#B291bgy^#8W=_dBk^N3;gZcQ^37mFH
z(>OY+VQph|GQrVbeBBh-cYgdg=r_h8+1jXFG_kz=FmnQ&l=wj@3k@`PNm^!@@#|m>
z6gUc^@aHE1m9mr%ds){}9aJ)Bo|lKeGU*^`@>g7qWG~inkb(zH$v2+s&!I*^=*YDE
z>XDpsviI3El%Q-TWP3dtE&h))1p^1#^9^vXHZ-I|1NJxY$<Zq9SLEepYv8|PuglWY
z(_iY8jIUCm7T0_ABLtIgVfsiGES?|#DxKH#7nHS>B1NM{pT>Msk^*o85K6R{BE-gy
zh?hu3^SSs{W`b&E#+mE5*kiMjwpjkIZzw{4g%xg9=Xp9><;0E_85s!?*#+xXDyJ=2
z@fY$T#fAF{vmTBCrv3G)s<M&+6mf0!`_KsDV>xe?I&r80O~B~&;L@J0dUV|YzszM!
zBknI)0$=y{hdjBSB^ya~PycDlKAiAOr4TJm`>n&J{{?c6E%3$T0!+$pg2wF^=ii;<
zAg`_}s&0RE=}hqXk3{FftNp;PE!kvv!b7PU!mP@0Pn9~m_XR03@cpHoE1{%B&sB(<
z+FQHjK1CY2$o%i8=He^IH%!za=YH#Xwa0#7T8sgPUag=u&;XXx(jcPUvc3R03S7P3
zs6o9qjr6CJ$rY&wYNU;0kBR?rhcCa1j?*{+ZrWyu-XSJmiCowZ@rA`TbS9EJ;a<2*
z9CiUbXlzV1>r<y|)174?23|8`zBssvE%{Nha`o(MA}elH5q%9E!MAWTZgdL7Wr<3?
z)VzJCM?@AnG+WjczCP0ft#@N?(tQSE(f5K<!m(|uDz&#3^u29;vZ#u8&}gt3I7Ov6
zj|wU%kM*$MA;-aS=j3hq*XIlDiW?X9e|${<#;}XC*;S&}$L8q8cAO%kwdQEvu4QmD
zG<7-^a@u0G2>5mobQSk@NFpw=ORICByg+~LsX%0XTpq7&>)QEEb4B)@9n=MbdRKZ}
z%IsUBd#89xOe8@Mjv3$)!D2n2zh54Zkgg3Y=!sZ-KkXh8%FZIHqL@iMMQx5gC0xH@
zrNClmpZ|bhU|WA5sR{)S9F)Vj$`8xO=xl77RybHR8MlyJnb#{8-|=m~UJhLN(cc^k
zh`1T0>A#}C-E5Ub$FDgY9rBq=>Vl^hH?2!Adp=Fl)u0op%YPp-Kt{feyb3-2w`lVV
zoJ6Th9PmUZIVe)<HEw{a=)hOoilWOu#-)WQd;E+Zv?OrV`*h94cV*StPd8DYUBPMc
zK<3uTA#gP~8kBXmUXeDYG~b>{D{t71;Qyo*e=5`W)t-ETh)QdXtlrq$cc~t;>;dP6
z(65WbM;Z_Qh%s)sjy6SGBo?_1Vf@Wob679!_#V6(+Z(gucd6{%?0bK_@V8@x_6f$}
zLP8E#dL5^*dF<<QHPm?qX>~NhH|^{Lzmgvvz-l{YVr4KqDlYu75&UTDApt;?NBs3M
zFK}$;8WWv)k^I62JCXpO<mhWY+T_XKBRuFY_DikS7U8Ofph?*N#<i3RLM~<7I8W|h
z`*r+_&w7|4Z)DZ43K{9VX6S&4%@chK<gpVEE0_K~69CYQNcID)b&>NUBEeWfFTgs=
z8un4oalH1C0CK&0i)jwt1kd~TGv{LWITmnE!psv7QW~6jt@=4W_=%@3onlA|xxRgs
zb>?j(TYqEyW_B1aO^f*9@+I?*`o69YAL5^96`U8f>*~*E$A?&H8xuT|84ApicT;g@
z!Z~vad;S?=jsO#m{>(@3qbH+{R*&4a=@b+-{o9Q(>&M2%zK!8%Ftac<PES+2-LGhr
zhV^Kn>Twb3Ynnm+t{86N<`)C-fHhneS%oaEp&elA4&E3&UsujrVxV9Q^{bjjK5e-j
zH{GSGzKpfm+E}2;5y2GH#o!xCHka6x&xpY%f)LSp?A&_Lz*ENKH{<YScT_(Ws~L@E
zXS=$INL|ooPM1=$wI`Acs+fOQSu>B5mV2}61GkZBZRIpFQEvY?J~qTl#7rMEcm|?q
zmI*};***aFO}^hzbFWiw@WZ$1z^e0;%dC}UdYzL>#O6J3)+1+A&_%Ef-X7kKm7hge
zuL&m6U&)iL@9HTUKIBBPXq^rt59Y8KXgn_+`1V|TKW+a8>Sl%)|Kri(*swwgDLOP%
zsQ_RMjT?ry-WY%X5*Q=L>pG~-k|n1gfiq~k;!YtZfW_}JY}dzPvm>9e>^NfqRfw*>
zc&95hD>%5=^S$xxndCo)kjC&*pUPebbDax<SVIE#{e6WaPEJIfw!c<z<LNceTu=u)
zYUnmoMd|f7$jIKcD;{mj`!u{LprWD*P}g!G!%6OY3Qc;z>mw<HUen1h_K<I@yO@l1
zysh(DX!t`(n-Ni6Ov$+A=W1u|m@<;mup7lq{*2Dg7Qc@KfjE2q?YHS_olX}MfsXH5
zI`}#>o`m7^P#*gm=9JT_D$WuR5abx=y$B_w5*%cnhPFKxNr-AP4SGX84)u+EhvywL
zJ2-tg!litncp^j1M(WEMvKmT?A69ya@#KhOc{i-(|Hjd!N%k?@VY&CU$oZe(Pspds
z8fh?tDViRgY+5}mIWD?dAuZNBMO(JcwAsx=Js7)mZ6)~}%y}07d!RX0Zl43_8&)YX
zUhYrD7f-(8l`IjI1g=MuDyi~wzkd0HH!Qe1hNZxFlx&;Pq?^aI`BeSsA_~cSXD7z*
zRsln{jQ#m0|C3G?^05C^@Z1x&-2+~c-Vd5+?vF3zghc)?z<X<G^YLxnfH5}-)-;<|
zrRr5}9NO%!&(B|Q?>mz38$NoCu9<nZm7#o}?RDRiiAP#5!03lk8;9A1FCAZQZNF*w
zStnk#23M%_<4Sc#uKxybnV@Txm7+UdL>BBQ5$N9-CIkk*p`CT#Ro+Z8Vv1RMu5*Nm
zv$60AT4g|8m){t!UL<f>Z1!z^sUk0{=xTpTuCQjn?s-U~ez0jgg^7ZJP#?`r{?Twt
z`R>h|_2aa@mm8TZ5le14vje?D&g+#f4@X>!RXNbl?_;c_7T*~34HHf^DlhlsN6L+_
zZ|yRm$mxexl<tEWo<)O24jlKXy(Z#m0=`YJBGmuZpWZdP<j)b;+yu(agxhgB^^LwJ
zIXJeovVY3A#8{q2FmpFZHL>WM9cH7deVCb;0H?-?mBD@pQd)T}dcK#FuEO(*SG$Oq
z&8XsSHrL;<2o<%mxI~AjI#cr+%&x9Av>nj#xEG2JD)b<Pw;&Vx;UwNrC({Hau+G|j
zssX9lF5{=nwDwPAEB!YHw%Y&o!}QfZqq~M!Xg3sM<psxFR#=E2U?$Tu+I-k7ENeS+
zJ8ynt!7ZSOxKS{BIw>)xUH>K2VC+{1aiKH{x4vXFFJfT~J?*JB=O;7g5uC3sLe%px
zugrSS3aZ0m0G2GDH{%!-6syQK87<M4y*+nJ+MaD;P>n@Z39peo2#AhfDv?%wFDP#|
zIJq-7k7<wR9qhuf$8Y`&J2_*O3i^zq@djk-BSoR4{%SY--J`oTsefNsb3jgcy4M(z
z=Wn#{fAUo3eM8wE>n?>z^c8k1qe-{4ijx40$+PC5H@j{)ijLVQbGPFv*!>kG&WasY
z4L`nrUk?t(fXeauy_(o|w|qiy&!?=sJYuftT@fI$>aC=tr1#8*S<p6W4zyh%S-Z@L
zOZ&u^mOY>;EtF<$=7{jL2WSk6%g7>OqUEdw(OtTytZz>RtaC4;ofCFdST!hCtGCft
z{y0wEvtFFr)HZ}BCH`8&yne^IkC^2x_}($~U4*u~5D@GeFit!S`hZa}-}RwC>R&6@
z<d)w@Jm>ijkD6mAgUo`JccY6x)%VV+JlEgCU4685UgR`0_dN8Jrd0$niAo73Ovo0!
zZqUvmUEs~~h?2#Hn2*gJ^N|NJ$(~#czx6F45wh~og@>A6EMwC9(>L&@xJ<sx(4>->
zzb=)Oly#}g?d*JP+YO#@QD3{#oNCif15BNovebD_CD%2#BLQZ&q}W>|GQEKW1ZA=d
zziX1)YCrPhC}XDyx)O@C->(4ZjpR)gAWTEceaS;m7O2q76+Ll;esEnk2n8{#(b(P(
zW8}T?>Fz4oI*lCPn!h&|PA}aVjK8i@q0Q4mRH4p#qQfR)7wNH#zC|IOpmaJ@tzvP}
zVc7e?vwwPFB1I>NY`jN*UjaU0Xd$sSbF6T~mz@|54Q&g#S2mGzpi{1KpD#~w=r}yH
zbt|6uvg?MJ;A_31p_L6;rMq8q!aTS{tsXcOO&^+9m=0OkzRtWTw6y2t1j_u!(B5)a
z5>mloFh2dcv_xMtrm0^)kuI0)y$%4|8fdiaI7*_uNa;<t);ejls638*G+t&~6}Zly
zKA7BIh+wJ`D<>DOIvS^R>}8%*K1)!1InLKlr6ZK-A8aySWuGM<7^A3c`_}uGC^4y)
z*q$J-oUUqqUMD6zeYN32yokx~UyOKx*~!#RMlpPo!de3xO`(zQs%wac5>4${LR>sY
z+9l?_TBwRCr&o9!Is3P*b@l<GtD1ibB49{+&*r28^llNGTTPbtO)Q|*yF})kWzrxY
z#;3={V#PJnI^XWfuf6{LrYiliH}kv9F^6@BBM%itcgK7G;|*k;wA~o|J*{2&bG7tM
z6+Bq_i%H9}+PE$MJ1d16(+Urb6+g7TUpw$-5^FI%^%E)98xpb#cb2(+{W?;po|CDV
z(JzoE#{<ZjT8{FRU5~KQq-Sni3DkH)hfI~Er2L?ulm=TX3nWJ1>yM%8iML-gvF-{<
z(_poW9k|72!SZp1_M*_>;AFr<RF$)og6U=1xX;EGA9zdyI%WnvNPZta$N|<|>A0Rb
z19jo-pdx-m*qP5_Jmt!5@pPy)<6XL>2x9X*=A#@rI(4<ZwS?~;L#xs-psRV}0Thh<
z{HmOsoRt7^lz<vtQMVqkwoZeaYvVLXh<RHKsIiUz*gSUvvoYLhHf*og0#p@(ha;AE
zX@nB}o4K*xVlihMrkIwQ+>}n}aswY<!O8{1E70<O!r*-Sc9|NnT#?^u@dqXAP^sx?
zY?wy07j*lZ<LMsv`510hrmhyRDr9p0`A%9=Z5Tr8;TlAGb{wi+ytcZ))WZf(2BPLe
z5E>Li<-h?N`P)vXA-58sBqX?g@2CGn&rlIBAD;>gNXKI@6gu!6zj?@{DIi}$;^K0_
z!q)^86tE6*v}@F%8aE7iXc9=Lwa-}-AFcc;>WUTF1cZqP6f?XlQmSQtgZ7#ZI6M2n
zq$vI3yxYEDy8@-5;c}b#?s5-6JEI_fq<a>T0D(cKdWk94f$;HivhP01n*}oG6qst0
z?-87?+wBPb$Cw_RYU(Y(G0Su(44Ku?f`fzS7fjK}<@(PsSt&&nG(2@~yNNp{o}5~c
zwrG!O#?n-44@D2N?Fo%#M5-URv*g6qs+zRo9n6Hbr;OX|p;l*>&4IFS)HVic5<A`v
zeMCOrX(?XW>~kv{&pBz+-)1J?6m{Kdz~(_L#<>G%L~7f-_;`4fAS5b+^6wCI3||cr
zfmJ|2Eu;!{UAYp|p>9=w6kKiw^JLne@&VKXazXv8bKTJl?zGi`w4iC%wH$eoT&NpL
z5D9I0fx|^krcxFYJ-q}ZjD4|A(-~=Tu2QIfYHR{qw`qV9a&3_LZ|$ZftFWS#XKFo2
zC@3-~Cp7_5PyzJ?mO;4$cnMKLE`_<(+fquonrencY9=NoU?~}AS+l1cK*1x0aq$v=
za%lEvXW>}_BBBw<=<hC^lH5dmH^$U?M%VYf*P&o!6^>E=aFVU;EQUh?e<q$WeBiFy
zpA^D5`BUU(#uqWEp3r?GMkfcq((Yqv4WrXk$1;1lJR{35WAcUJMHY6$rr(0&nV9cB
zz9(r)af^o_lu09nGTfe9LOnJK!_3iS8)IJSp#K}i*1iW`28A*&5%u52Ybrss@#T)v
z2a~lveso$TVtD~^N&5Y^xZt~4ti3lDNTh*Yf`IE1w?PN-^B8`)h6bZjrDy@WuXk2J
z;X=?#dN2uS506nZvCDXZ+QU#`DpX~G$N}o8XPz@|P|-F6NEqJ7E?I2V(PA`@UgH~Y
z1RE1`Iq?2QFBsM5=GC34!#zC=;-VVa1<*@}r~D^ZwLlbF!<fo@ULpnQU=dEiW%w*{
zp4#Q-XVEgHOhrl5R)TNL?2f&^)<h&S2pR>_$6Fm(c?YxW@cRZw8+Epr+8=xm9!$fD
z$$_Jit1v)w{BUO`mNvXDWg<B>!y+|2>Z#Mn=ZnLiE{=r7S%f&9^j6B>MRJZ^YirBr
zb8e{|$944wA;`#2>~BlHRVqrM4}yZHE!a-)iyx~zUnw<SDY)1_Vv(|RL)zg&iPI6U
zhtRCS-CshQ%4w*{MSCH|_@VV~hGLD=Ib(wS;PqG94A4eMA!wg|hh!{%KY(xg)14`o
zrPx#o1SV0e^hGQmd)~|z_^Ef}IPLIQP9FMK=cU9UuB#chtTrdg^6r>`&_WBfS$9nJ
z#~CfR^%A2uRdLI`i9RVwJr#1Wyr?*IA2#f-1Vr^L`Vc)}7^Mbdwz5ui-OEMap7kxx
zC`%8S_q%oNT$sLT*LPpzG{ti$6_+(-RCisc<nfHials_pfy>g^?v7NKr$v<hPVm{p
zQ;IR<rd4-R#2p9!Pv6QuRVvG^_TBNg|I@?ZZd&-+>kFuXR<kkiZyqxS{!$ETwbfq;
zxI*qem)HIU`Ca^ngKq|Ig%ZbgbufySEhs2RrKp0pDK=dTg<r}3P!(;L;a7ljt;%;X
zN&>9Lvc3QQJ&vTjp`l?xt9p@Pz?~K3?F~x?8>(nNt6{K+_zJREjCbxlf!jklt3)?-
z7hIvGg$!Nh_PolF^D{ggoKJTG$T>v;+;h1`Qef}i{^c%~u+Xhrw+3t6oKU+A&GGw8
z0!e&NsOV|x2>a(%ERxN7hUi^Wg$!}8($fN&tU5L8?x}%m^4zZm(_9+XkpY5Kna?fJ
zt&8-+*O|4b6H_p7j~9;eJ~~mN20!baE-$=!%);1O>c$<&#V4^sf6z{PIc2u@$ARx&
z7_fOV;Rsf*b*~7dId{YQ4P>UflCna-pJcR_s^^GJ$3;XitZdeoFv!WJ>Rh+rC984H
zN*`)o3Z7w|-NjEQX+;2d%Y*9w3P>Feg34Ox=g)H1v}J!j*2Q{W;ZOeUN`$fw_kf#q
zbtEtT%a<?2H*RD>aVE}Yp$h5`%h6V~eE?SKSWf_&qw(WoAv(u{vd$clZ!$e6txbEW
z6nEXJ(tolGlQ0v^q+`t+rW>3G7G5L~{^o5X(*D^X9644AXS;twlzS`kjxaSdfp*q&
zE|t}7H`fetyW_XXFI^m+2z0_<Z1Bf+dw;Ay*7##kLVPOC$wpWqllDtM0PhlBh@__T
zH8ij76ql;pDBj;BRGQ~SQI1r~t2Tj_8VQ<iU)7x|a~083(x8bEC>t5)f0jg|crBvd
z5>hdhO(0tNNoCy3DzY?H3`Y%?wzgNuXhc%^EmYaxpybM#)PyI>tVSR%CAoY#!K(I1
z{P*wQe(&FZ8Z!@T*Z<Nq$8UCMfeHec7xsGFtZJ;m!4U3@Lw_$g8_xW#50Zl($iscu
znM91XF!`FN%HDl~HbXAoUW`*^@JwO;GLoT?a!=-~$yzp#*v5~xLbIs@>22lt>8o0m
zF5ch!SxfFnX4MHFnChWcL%r_+OKy=Ulh|tSx0z6PXS<BnG42x!$$kT^6Su75tDRPQ
zfL!HCe<?K|OM>cJI#jST<x~Aemdt>pc*g{eN~0+w$VRl|%eaC6c3biof%x^K$M-)+
zJ9j0m>P|~x38cZT)iE<ucXO*-r@>s`;OdrCKS;$D1MO!D3jgp4puYrudf|}UOcyd?
zJQOH94JvWDF_qPk-11^7YcYV&$>X!rrdp*~#Ta46<8tD~hIeF@ieb;v7H#6|me^54
zN%(c7k-ds6A@jCXTIy%(k8P`@&TLLj&5V{|Y{W{*xq><#X4e(fgm_E`U%;JP0zKkz
z_C7|H6Ys1*Sj`-K1eJhWlCE$mhsJ3iMsa~G&oGpO)#&WP2bEN`?n)j*?ok3xurH~C
z&m&k!r9U6k!5uNp*!`}9Wi~Vg#;}G!BlZg!nLwD7WU-H=CcttVuXM-)b_fp?;)n8d
z>R6=FUUXAU+1|8D4t>|PCQF6bcb?hx4D>g-G_uM+`_z9;KE&o;S5X0KSZLaI>(85%
ziezq6O$$jdrv<X~CB>xptd(rdSBv>M32%R8QJyKiLM33Ea>oRE87UAE&+3%Arsq~4
z0|QYJfI->BP=%e*`gDzkWtj$BxU|`Db{~m~rX1Z)n7FfZxwi8hmZUrgkip~adVeey
zh?QBND3fKcJ^XQ<l~T}t`6lCLOs-ayGJNY7`|Y2S!bd-Yec6nSo@j42Xb@)&WddI=
z6M|Q=mx5>)zjz|IHbwn2f9cgl>lP6kD<A(3Nm>E}05=l$-JuNA(0xNDR$8$JCAL3G
zO6aKZ`Ho%i-W{Ca+C7X7Hy!v@78>gMjLItdrkU>O7?EvNiBp%8I=4+erN9aW$DyRw
zutmKu^D#?q&Az&1KM?L1Br8w%CrHT2RSkDMfZ1t^OjVYg9j~$T828TlpHT2wW<xc^
zbh6xLdxv?TB&lV4cR<z*l&g*p=Yx1a6!;d3jPbO=5kdz+r;Ti1d@?3Bw%p!YP-Y#;
z*Gt`;(4=<W%d%bWrR{<c70`rUSlSM=_2^cFL<tv|jJG#TsxZp#glo{%C7lQne@EuG
z%_$_m40tR?poZsVo2dNkRm~w{rzF6nUKNhym?TpePxY`fFD#dG?c3?6nQX4E96Olv
zb)C#S6tL~z%X5qCvv{ntFSlTXhl?w4yBc1Cab!2hM`c3E-VEL`jHckbVG-!SCc=dz
z1~d+(016i}{qDN9NOlAYo-1%j^uwmJAJKAR6ckj4AkSfSun%@u7I;zrjh|1VpV6Jd
zR;7UD0t%sYz=Z<6FKHkX48x9fI6pl&+3ItCr=q-tdiOpT3HpKIg*T-FD%u{valz_L
zM4z>%Rcf*~Uwg#Z5I1s`P~2~uN*OMSn7)m7pq7*um04M83)c<5<0~FJpI48X3$*z3
znUsKUo)uE@HC*!eo~QC=wTL+CS6ICiUbdoE#h6KllrUY{eJiXNU4jAJhjRs{Ge@p8
zDxP0~4ngWR1372hUM&y=iOnuWcYT$h@lb^NiE-k|HV8uim*{1W2U6?ew!2JwphoHz
zyY~AhMAtcxn>tUQ-f5qi(E$#cu-OaGQbGU5d6H_##FRBPH9ybghK#&F>X&98A`<yC
zyP}B3ln}+bx@hr7$==NO<$VYDt}us>arEc=b@IZ|OK-FUKc(iYhmsIq`7x%78kg6-
zE7WSwiE6egrX!K+M|_h!QVustL)sT!tQM@ph%)or7_|C!=!B0u_*|`4zuJ;*gwca+
zS}#Zh>X)eFmsmYKn(f1z&dz4p+nlMJoq@s?Dw!Bck-TOxQ>z7QQAJ&STI{`L<qBP_
zkc&JmE#@t>v*B_#vr~aNBNIn&?)<m*VWjv=Jj_VmYu9<p9KY%+%Xi-jFJiaV3JS2N
zuv)NBkGbq{uMSHkiTzDtLd+ai`35YSwEN)K=ZSS?Wy2_Q^hfui+1m^<F5RZ3eFO@+
z1E5@{lqN%UbaKLc|Gvh~QqKr$+2k!r2ygvi1qLyT$qA`9+<*1$l7YVd(~hbi({o7E
zyS;I<yp`Oljefx2>|x_T_WR<Ce%XX{aWQ3mOQPXNa<NvPPx}TWZ;J1h@+zrEX2Wa_
zl}oJMa^loD9okrAPSq^iHDNsQp}(KP_|9OPE92BcOPcQ@YOd?);o9}~X1=mP+5kx+
zaJdy>b?O(W>fehbCMQovrfb}o85t!<H0=U_hDoav{-J1#VK(-Ghik(=<cY+jj)o!~
zy}i9$u6oa+dAovF&=jB01-lA)KAzq`tm)tQ{)Gx@WUGl98A+SCeX3s?cZ%)sr_PO;
zgN2{{DVs?|%vzZCIIRIciPI~3r3{@K{no$p`9cO563^F$k(>e0+#Jf+dkh;f!*-!#
zb*8Qk#ToC45eOctT8NN9-&`Ha!Y_sQuln-=oVDzu+9jr;^aJSW39H3s>1;%MpKUe!
zuW6(ZwEO~{e1+p4I>x?4Wehfzj}OA*><;SlbqSo89z_0m)~`{O8XqXxvXUGZ#w=W<
zq#mDXiabvmTFbccQH2IITWA;ifZzig-)+3sqbf!wO(upv5u}f0YFzied?58*h=Rvy
z@WI9h(Jybm$yURn_8xYtC|GcA_o%$^vIlzCw#TVc_)|V>q&OWsZMF5A7HFY;KFLln
z!$8#~7o7FOe(GZxSr)ye4|7z;W|q?eQo|N7aa*`VY&=ZczBhXQi>CnFRGGGVa8hOg
z5BIXqTp{MTkX!!WheOl;AvbWP>}2!FPff{Mh=4j5w)L6NNMhQ1Ql@*Rl1u}Vc}_|)
zI*yJac6T@#c~vvX?)aWao5+|TVa0c~Y^w}PkFy4vMzRJlMiz5ASigpy-ba|K9hFWw
zwy)jIdqA5b^Eq=YMOri!=cB&Uf^X-V!j7o6WLK_P-kNr`7PL<l3S#|xlDpz6CTI&%
z(T`$>QdDlRo*p_`9J{1`af%CZG>j>b1q8)(yeI&awsZ2MKBq>d$yBKlMDHaHyH~w1
ztJ6rjedB;cozon*m=P&8;?jEX(-d%~yul#czn`9veR=BwgA*WbgC0K#R?kM!Vkh06
zwPVXNadAFJ*K5Gbx{{GeU>0WFvN0CHdS;5a@s-roME3{MuPyKUNC`*Xk@ixW<6+J+
zEFwReV9|Dd^6#Myn#|BQZwCp<_nIBVQxZxP><6bHC_!tDyXYK(!IGhCN-7JkL!IuK
z02Yqj@(x*R3k?C^a^^RG6CtndZuoxaikJ^Q#c}?78$%rB++7<&`tS46UgWwId-KuK
zHfT8f_cv%#ZOLAIEG*yu?{B`Ocv^U9q3co>|3Bj1I;zUAYa7Kte~L*esdOVPt)g^y
zBORNNZf*=fTDrR%BsbkiZi!8I!=}6W7WzEz^PX><@BDY(&w-9T@ZRfQG2@!ooNF!P
z>t`Yx<!=x`yIFUc&>YuBC)r&!-`%KgF3=DzAnyRuWyWz<NqQ^?08(1b@}FLyE#lDq
z`{fNu|J9HG$H$dM3^cqSz(~Cl^9`h=1nf|@^P<QHPwy^-{;r+LpT0GCfbY)S9I1io
zFQKhWZ8p7C79xq)R&8!bQc~MV10vh2vRY(@PF@i0HPwlRCiK1m1I>V%flfT=a$5Lw
z|7g@2KIz2Sc3Vugyk-N-o@hKaP1^lJGx7Yi+M?!XXdB;ag>X90(}F+aMEm#*@W?Vk
zheQGe{8w&Tv4@+X@6OW+U}Ai^Vo-1dtAm7<$exs#Hc5XNI%K+dNEEZvWMit9S#Ej%
z4;6>z)%aluE$Km;zWz!~W<n2!gSuh}Wm&`>Bt7A#m|WC{HD%4y(lT48Z-<0jIaAUX
z%IzT`Os=zk2P*>xoAhAfs`s@y&nFpEJjMVMT2Gw$TUc)mSK)1t+%PS7)Teh9fliv6
z=jD(7d<X5riy3q@CiN;OzYGQqa>ZgX=~l7Jj5!jskp09{xs5kH+7(&?Oxmi3c7GRU
zp6vR00v<c(skV(ir&<T`&LyhJ&|&B(AbX%j<@ujhvy4Gc8vK4*ug8=*OKORX*NUy)
zh1Q<Yd{8U39l)hXOLHuI^XEyKWS|8ec=_YC>Z)9aH5j?7e2zZsqNW6lk>mM`>g7)P
z{C}S-4x1}!{y|<nS<u}rp&cn}`pCO^m?Dh8Vo|fDGa3~(4$2Z4OLzA*Y`HYi$3fht
z!8puy$T;>{uYHfh^tPjx3}hEezWVXqJ2<yA9>#Ds#)Q=-lSRP*qwR`y)xr~mx1Z`I
z0^Pg$WvJM1Z6ciBRlK*0%eTop7HvBlTdn9y<MwK<1TM-NEGoKIZ6oYNTO+`>kk;7G
z_X9mD!mMp<!sFvNP%L5$oRQk^uYfRiF477Jd}G25oX`NFC3)rml-<uuz_P22m#5`8
z51FEWa#m7;^GV~~>gC&oPguYm`QOM{FxE58Sf0muOjvMfGIEI*zea1DeT_J~8n~rU
zJ;@lFeSZED90h~9eG`Z9h9wO;IywytOaHJznA%6(ABE!7oD<e`bgbgIxr&#Dbrg)H
zNeyqn%NsU}=!S!#Bg!7bvQ9IONJsxMWXdjwl=Ey5wU521T)Zi=6X{f?0xJ4NLd-Z~
zgXlgpae+{H`AdDhs*)eF9j7tUI$mnj#u_-&hWk@~8cO!|Z&jKQ#`=Z^9lFK8<`6DE
zzTK_~M^~VV2|aQ5;k|n_%JqU{W*jFceaTo^zm6Z>MT3}3HF&r!^=x7mR!KTIGsQz<
z7XD=mw;uzwnfcJCn3H8^c5Br*u#~TM{}xe4aCv&JJ(AW`wy>b0^N&YG!+CH67?IJL
z1ioDkegT!!xBLPxM5{`Sf~=AC7=J`1jOn^DgCk9wGTodHqI99a9frGlphHyY0#ejJ
z<+6z3$@NDVfAt;b&UJIqJ-U5HK?1Z;i7I}*a!kC;`PWGDPB!~U04pf@Pro;P`P!8y
z^{4fZuGboVr@(khf`p&tox6po<$6flW{A`X!u^e}U`&xRey!#vVdT^^>!5<_yk<{Y
zyg%Q276#Q9Vt82cR#$udXQ7m087x~VW+)sqkeckIWOiLq!(nW6Z)05EnPNp8X)CHv
zNl|U|S1Eu4A;RY6WIQB8iO+&8R<O6m=l+;xkC#Aw-aN0$F^?;kJQcB&h|ajW*!SyK
zCKsm*!PvkvE?Fz!=R4SE_^QAFkxtwV|Et7gq%-IqEV$3{bC{1N()%1HGCWb7rU!xX
z+}68k8Wt6%3V{l_rw4=ib};m@ygEjlzeBj5dW{zxejk*}0tOV#=Up}L|M`!s7nQ&8
z|9nD2dxi}j`upd9`w0EN|N4LIx6uDb-~J~$@vrg!Z?p10@#eqpfc~$&`R|MWj{pB&
zC!qhMWd6^r!JqN}-y1$Q^nX&!|6hh52CUt7#=?R=xG_oUI@uQiF{pz>1=GM4>}AG%
z??9{}2Wm3hG)e#UdPhmmdyew*@^+vA7Q2cM#G0VQ0SR`SB9jI0Ugw|;OGfYF{Iy?b
z9uA689>e-9H2nNpASbH`qCIS4Vn#q%jBeP2q>I?~7CK)Xx_z)sVGSgp5A5J*|I)vi
z;`;+%Ioa400D=7tP+3WUf$cKm=+b8aQ1rSz1n-_dxd}Kn&6+!&zu@1fmseTm=F}dz
zUA~OGW5XWoOYLb#{;TI6%ZxD|!;989X5(fYBufA4BHEO!j`MZ`0+9O@qM$_gKJ?%L
z?{b}ewdU&T>W8Ui`)Z6!$l2J8%>wSG`3^p7eR09QSfep9k?D3aId`_8tGw-u*qceQ
zKUJRISUUMs+y!ZVf=}#L8%K!FxH@Wjkoz2A;!K@PT~$lFu*y#Dei?w>l3~`wzEL=7
zF%0rY6ryB8{;#(#LFsxQ$U4|Vy>8&V(6FeemGD#7#)#-e6lN!p_avkD<gj8=P0{kg
zan<cGEY%vVIcy_~f}hqlZkl&GKgX?`X-9fke@n#rK@Px?ck?WYMaIioZ2R)c45Q7J
z0$lY3dwK&$!&BR3#u~ogNmodH9cXJqwJ~fabEuGi%~1m#Ewq1M3zR_sTHh4x1N03F
z8U`V#dfq%J1Rew_hLWjzaLY$@kD)3(u4TA$<H-)ZiM3LLG5Y)*!1<SVexPhkM<dMD
zDy>o<CBDuc%%cZTKlg3G+8VMvU~M>S6NM?>vuKaEZ$gd#YRl{2RRey@f@+2lOXp=u
z5X^qFzb}m(uED4n<S*0Zd@xvI9!L7_T@5c8?=1I$#y0Ct=ci)2<L~JmEqk-1%MlCq
z@MFIFM(4c^dpBw#D__TLUDOY((jUh+0uYrud<TeI^fA$$tlzX|4Ji9+aLA*j3ApcU
z4WzTKw{5QDl~z{1b^`~r8WB~vok54!iGLN9C{FRZq_`V5>KKzJK6T~ib<=|ppTzdm
zF=kG5mX9CQC%X5cLNjS!Ql$TK6LM#hC5mAi>T;M&F34BcoVm|u<S-JB`gL|(hSUun
z6-RgrQc10)7uNM<mKb&gtM3b^cDuV*Ws4gH)Xj4UdN1IJ_GWVsX+-Tob4#||7zr4c
zmxihgN+im+wL9&&4G<l{3=Nc6vb`S|=y<A=G7Zq`m9_U-N`5YEZ87;lo^CD=MVA2d
zMRa#77#e1f(@;}qX<R%qDO2lGoO#f2M!O)LbpFW!FxsZW1@RzF1dgAb`Mh!Wu}Q-U
zUfDsz+h}sTfb9IKhiecm2dPGQ9xhQ=cfYo95XyU_x>Og%p?T0UP?bBZ=nNgpj6SI8
zpCqVLFI91<-6?1JS%L$y)?VWHSuhE6WC1p9Xj=2?bG4$YB=@mmb!|j^p$P__$@zDk
z>+jbw68(|2XKy?)9Hk#S%+Jm316tR`9EbQ5^C@hZ5z;(3mKW>&B@-EC9I!E@+LNoq
zSUUYMK=0t)a$roAoi}mfStjKLzj}2bVILDyeG(d~6+UO)<R6XCe0fDK+B{x1lohyv
z^IXL-9FgA>LtR#_Y00FWIvknb?`Bz>`;|&+rM_ayHCg-*Cxv!LJ~^VSYSN+>WCPTS
zK7NKki9)(jqFcl#yzPeKH2ni;^8<yEzOiHV(CJB=`v<S;lXD5m$-&a#hzYAuH6g{m
zrUL?Sq{9}T>$e?U78nwrk9YRgXt$x{CKNEWGL-%66Q?#st|G#&N`lpLJy!;Ngf7D=
z5!#M7=>B#=NC6lvusYy8Qqw*t#S>HuNLFGvt`tA!A8N#+vXvhYa;{aUiuN0AuTnDp
z?*7cNHplxm%8$$D0M)5zpkZ<n5*?cAGVLtPSfPGZT@`!EVJ5S#SKT%^WHfvfBHI7^
zXi8aEO>GqCFA#iC*i>r#+IHm!E@QX)M!D6z&_lx4+j(bofNRQuG^-CiX&=5y3vfAH
zi*?#l<a#AD?`+4gldf;-^=&1dSlRUV<NB3VpTum3Gkd(AaP_OB0R*=zi>w>|&@e+!
zOxW=;41W+g6VbzFazR<kY!K5nVWa$9;V+*;`{jPgmm^z{&ydR8%e&I_C&9Zbw$>%E
zBpO1z`Xo~<9a>KwuaLb}=AV=D(B;tdVv}!*zEI=L^^hba?MlnaB)i>_yY8z*NL?JO
zG^19@A7N@xRfxi=0oFT)S<kDm3ys!aJEAa*{Gl<S`WmxD$SG$a@^qex&F18OnW@aQ
zI5J%OERbY4qxNj@6^jN#juOqm>%QcdYceHO5hM_Qj$O4q9b@^i`Jc)Kw1(EycLg)(
zGtWV&yW#zahUL|pgXU&ld{lo7l0uZ{5%Hmr$>LPtS8sCB4Vc1_W8S$<#l_&C3<4V3
z6(A~-ex#M!E^^1`5ECx!s34^Rtk`He=8OGjROEh=JFQNji-!b~Q#$kh^<hv}6o|^P
zG4)GCLs$sm#6jUY4k966ZkZ7W8Qgw*<7Byac1Pv;qiS**@>jp-|0bH=wElU`n^?uC
zFJE$pBqZl)Al`v%j<ZA3V#4WMQcJB1hqBw@JY22w#fR8&)gnRdf88Gr)(s{PQajT=
z4f{-aj6;%r<2Y#AVJn>6&SG=|Lopo6#~)M{`RH-U`)MRqu|}#wz2tB`augv=#ALG8
zdjC&w^X9`lFh&tKa6L)LJ_O?kbM{YNeSWHQ#yCokdtf%+rk)-S1JOwBc!jdPMFIbG
zQTpJ##B`M)Rmso6><=~{zI}5ne)ne`WrCk<LpQ1WB{jKor4$q|%3TT#H%66nm=jbh
z6;$DH=b2L&j{Vvf9lWpYmiW}7&ITR#<9~J2f{jo1)d^=d(d4TgZ?ASu&HuEPr2BFP
zT_z1JSFv!)rGk95H_?4Kuw?<0=&GI_b+FjRkb{q#L3MXh1(za1-KzMZZM))R_zx&L
zJ=XpB!zjF0mjv|*ZhO?NIu;t20^=@K4&tyfP@(zL_ABn2T4+Xg^WpJqDG&*T!SYIH
zVTS2c-K}Ll%HhR+8;r=cU@i>(9ZuL-iFB@*N;-y&`3cKr<K4lWw{8sljw;G-teW2a
z{Ro$CpKf|hfdk>53Ak`EJ8T6SXuq<#LHn266cTE{n2FP>-y(j9vWdqm`H6JdeT=`~
z%%o;J^TS%Rfb~$kx1&!vww%^tPTARfl-Bdk9n(`A1a;jB1%J*&#g|(*_xHLGxq_i>
zYEBDLG0?u}W%VbYaw<xkF$ZaDUoF+2a*?k4@TD!DEJHiJeN<E=g#8?U?Q7@XxHb6r
ztO^ZnF}8j*&%ygbeIu5wKP!l875Iex=0nZ<C8~;?^b*M&xtJ`hWpp&@FZMSEQ{a^T
z&oVOXOwTHvr*nv(Cc`cL`$`hslaHWtQA>LDe9nk`rJf&8J}BQ07VV$ie?TN68A={~
zan80dHs!og2UX+XKU?|lfirc)c$I_v)p6IsY980ARfpUKTqkviNSeB`-a1`Ib0E3a
zVs%K7Z>$t#;dgb)7KJ^KIm6x)sX7-0!|nCw8PCNdVFOcc+;Nu<U0y<ErmIz5+$#w%
z=WHw6(#N^~nrJ~F!y{n2*q}Uw9sbZVHz0NZ0Y=SECXXa-Qlz2&E^|bvo076WH^X!N
zH*EgXiimEC24jWkdm-B2h6i>|e@)Rf4I1VPR0K&zs=!CuG@Sk{ut!&R(iG{c`PbZM
zp@BzcZhoqsy!Lk&rM7UTjge$BA5m+#M$S{{dG^r5HPm$<`!ggL+G$tb$+5B?Y&r;-
zo+kR)t}KWe7%u;sHd9KUc~NFLDAIg(Bs;9c;HGH^w<V=e1SDWG7i`FMdS_ltN?ch#
zT?U!2B*`~kCcE`$1N$lF9E#=7(h>tg(SxPjao&(`XXib`_|LN6ZYYPXzAY$M302l1
zj|6fW0Cy<@*YC`KrR^edu_)Mt&iLH;9O~jS8D}Dy+Gk~i?8^=5HLvG4+a8&7x=`$m
z%(tT>QB<!ZPtQalEi+awNFceuX4#d?4+Na6z{u0Se(@$~;<t3rCF&02`X4U!&WCF%
zw`Ic~m0nnn+)IQY3Q<!zF1ZReSY82hnPR3~mPE*-bR*;0%de+~dmQ<bW$MGc7kR|Q
zzomg*N1pc>PV=LYj5(}VG7?=DNu{NUg*XrS^*xl-Mb-(~DL#HA=hd9d*O7hdhR|{=
z*2#TDPM$R!idYfg+#iM=O>WGuxxpq2Cf5W`KWqpjBtd8p^w7e|r@$}2P(0^7EpJ0c
zOP-mz#Ov(mJod1)hz;k}TuXMlMpCz{a|lUY6;asrXB=TqK;RY@>`m#KbdS}L_Salt
zMU{5;xH?<-2?@iqlrLYgmGg6iN`nSU^Lf}@hj?rECLeQ$L}f805<1ChRc-=dikUpK
zG=Ro8OtGIQ2NkR)i^G%z^b&tHpaD>o(`#pHqR{@x*=>}zw$3bkD()~oq<`RGmW<3*
zq1G~gv{+t~#6|A;=qLc$pQqOPRN1fkzPx04@uv9c?Dek_YUku0CbCPH9IY##3kIzr
z9NKb3H;hRXA79HY^+N$|z<}ZO2}3|*zyNZrKPD|rpE7}|5KaDcWJ@V}W!f*}XHhQ;
zJY@_ol&p1Ozvf9-8TSdUvG!_R$6$QJH>e=nv*IdOrt(bsrIMf4%7m=;(FQ|FN1AAT
zPk)OxRR0DilB7K}l>Otrc;~O)q4~(N0nUp88tlzW`N<-&Q=N|_4}Bn5a@xKvKwaqR
z`uZA4zJH{3>J1PQEqfr}LC~>JH=z5DQ27zEioR}UX74LAra#5sytp`+i`0f~73&$O
zn4IiQENicVDP8kY>>s=wrkyaPJ1oA1-@aA`Hk(B3|Gwf^*c3QDRI*JKxDvZ79rrsx
zl3?|umq>8}Dw{S86}3F7l&f89H*QaEcGl}GWy#07o~0RZMke4aEhW#SOjK2?sufu^
zWvdtXB-KACA_X_*^$|VDE^RmUJ2(;2pya|*`bsN=t=z@v!<{=}j{eE!YB`bz)CP+S
z#)IvlHJZBcvZE1Zw_Vg$j(*A7264O}(dH6sEF|GvZ=|f+V_17O8H8RN^k@8wBtW<v
zL8^CX;RJ(1G)Q}>U~ryy;-!>QiRv_i8mX`aU|#8l-L;D4M(wL(6tnYSdZ=*I2Lhql
zGd0qxMZaSg!(rDCW_G2!sIx0CHZ;fu$X3f`S(pr|sM=Q>GHTB*{&nY&fu{!^T8%O4
z^Vczl^UAp#Ma2!V<aA=H2}Akiv9hp`;A9;23YX%hUD#Y<3ArdheM~#|2M0723j@tA
z4T<=p9l>!&$8L+JqE5@bSZ)LyKgci2t;5^d|BQt30dSn#DYe3T&P(O~IcqyAC$%m|
zggoGb(2r@$$`8;zXIwmr5nWj`$4XE*ZcoBL7tr5Xd%hf>xm+;A<*+}ucJbhPF|K_h
znwBjod!d6#cv8qw&_yQgTPKn-X_30ur^y*0sU4CN_OrENbr&%iDCo^c-1U!CS_W;y
zblyk5)>Nxq#JH&Q_B9Oh(D~!#rfhh*RAKhediR-e)j&3WaImRXVZoYW;S$mbGemwL
zIpH+szMT}(-;bkNmNUj_k#-|~ZgI<~rk(&4-bkLimAh*e%iYD4qW)#7Xl5(P$D35I
zHkhlPQVfDua?y^_L((mQn1tl|K#CP(Auqt+=73Ul1BGzXaV~PB<%NLf_;M-u{t)eg
zLQ!m7=scL2<$^#wB{_n(j-QGS4fLsbhO^NiQ0LVlU^$D|Ve=<7M0@k5R>KUKsEjMP
z%9iWV*9gl=%8GGK<3w@YHRE4pwd}yH{ydGm9KS44wb|cKL)g+Yi+UObgYeVW&pZi{
z=qM480t<Iez%V!pPF=fz2BEUE?cjGoMmtLkr)EPWyPUCh>PHr*GqZOMI1&Y3=W4QV
zk!`9=Q{ql6Sy!_=?b%DJlt|G9VPEte0hMG+v%{Di-NjUu@?02~qKs!Y{GgmGhGRZe
zp$)#{JSaZ2X=|!%{gXV~YQxwvT)6-ln3Dre9fPUsf{k)dOrd1~OQwkBKNXJ(w{O>Y
zeBR)3*Wj!_k5Mw4q!i5|#>}($>lV&0B)H=ek_8?U_Aj}e9c+Jlygzj$u$C(ul<PJx
zL=m&BZ<@m$cpPL;(t$!JFe}b-nH=*C_(wUbCujb7w3^L+r4k{3j^emO)@be3(4_Lw
z;N8cMkDyb%!-(oBM|JJqlhhX@=Fb&BL=NQ@ti5dP1RF-C78EXuw>$t617KwI)OnM)
z$7PRQb1ZF#tKA%kOOBVP$!ap0Sa+wAu6?;h>`v?uvG@;i<o$=T!%<^7@-*)$qitU?
zXkvg)gOFr@CeO*)5~=1kx@^ks>;IeN>=2@BFjZ_9kBQjmpr*QG2i}r9yb`JDa&T<0
zA;#C6^3^-s!XZmcg(B!ONr!xLPR*pRP4d)WSS#q}RHDOXFu7e2Y$Lz_T0_tb*5@3T
zF%;h9mT&z&K=@y$V6OvKab>LnVms6?gg$NQR!nO9!Fp4AOo>B%74PXce#qrq_Tjqg
zhvKDC=NS*~2QDhnTwK2QAN51?Oh_Vzm`<e@(;-2VRcE_ZF}I%<q5LNY4(1~z!hTc4
zboJIl<;jyp-lg`-jH3Mwln)QqaUyeM_1`C7#DP5^<K^-Q292D_4w|aD+4lV%h<GTT
zuR77rtRyk<RND&9e}1CO2k{?CUt{48E-q8$)EKSx30X~DQ&gh%%ZfA{@1V74bn*S=
z&w^{aNQFsPQLOZFqBRl-{>6-giT`vLHAFmVd&M(fy8inq1CQ<dNMw#|j@6V_u~5ih
zwcDzoZYq4%*RgiHb@+6q@F9L*fnB!q4yB<Hs^p>if)Pu$bbA-nnIv{ScW+dwwX-oi
zbh&<fdgqqtrge{ormDiuhN0?0=pYYIUcZt~)I3LcHIv|Mt*k)#S8!}Dj}}3$fqHp@
zL&E;LnI`|7tijSkmR>%-n%gi#RiWJ;X|pDcrQKR(g66?M{V8p?;ec=cxZOZkq4T~<
zt@#WF6Ktbb^CuM36Ju+R_=5cTfjL&<R5{d3jrrK{z<lHlh?$S-4gxvWM_qcki0e&~
z0Q7MybBYbIv1Uurg!R<hB~{@*-4p`>e>!_EY^dkfKG?7mw)zX%au7={ikRU3RdUzt
z#U<dK($(=q`^ju@fk{!FW3HR769=#$vej&;auyP-RsYB}lxN|8N8@QBh~%72c6>2O
zwj%Zi&s)VhI{Gw~k7|sa@vifHZ$1FO{LCM?B~8nkbUo1y{E1!60Mw}BknX-0yQp)P
zTb$O`km%ur#zW^-MT13bx0gk(ZVAux2fHQ8d~<T5IS&<)Ne#Nlky;D$o?~Sa)oyp_
z)V&Y$qEg?aIFJtJ$6?VQ<hN&fvVN*n1s{cvGb~^ftL@v>Ppv5m2!RQSCw}hKd!d(-
zf{jBA%wGQNh8pU?BHq_-hbuHqmS`LWmx{b5g6UtYM79jlC<WlOetX)CjrlAIm|^2z
zg*FSH;J$9_on-B{2n6tEL6gvHnU^<i7$_P7mXz-+0RAGDg(|5C4l~BS;;sC1@x+4Z
zyc``J=gmX)u8cQ-BqbVP=?6<kin;92q%*t!gW+mrI~wZ`EAR)fak1gHp#ebRu(`KM
z8%<g^C)A$W%u}kKpYKc7Pf=X{Is!|DI*}rx5EYyz&~Bpbg4!DJv?A2|FRT~_>t(-U
z62Um4_gC#|v7bLn+Rvk}<vOS?+2pxEv;xXyOIKO1swJ|rmx`x+INUBJ95`ob_gKZb
z`2hL`Q)EqXVaNJ%CXISIs)|NnO-AjME655yd8#{)dvNSr#X<zH`l*Of`}+70L&56g
zsd|Z59U(g;=S~^Ad3ErZujHc-pqCL&5>=DQ-iC-_tUcV0Im@9zqeQu$&S1PO-Rr@7
zaDvq49afz-L`d3lLFIa0K4xz#`D!{HlK}gW1blLn^vGp5RMd&#o*Jj}#deV#lDkxr
z4*Ac$q2ja!-Lh}g;-ECb4gk9wmzU0`yN}W#d@|AorB6GB%2d)^D0eLP7#dOMx;=wd
zHTaqI)j7VV*_<wcaW6^_`(v4L<K@??g=WM-A{R#4@r=&m>%B3KN(v$=>EX14Sld=S
zdE03wo3MG?9Obj}n2WWpIyW?aTS@Q3(^*d=5N2i-cPZzd2y3tOEW%1R^h4q@>xQ9t
ziwX|PxWTA&IwWX%zfqe&0WO#2POujOvdP`;LD!>3labTVfB5Q0cN}4lX0;SRy9Jgi
z^@H<l`GnaV$kfuM6)Rv9l^IaObft(8Np*5j5YEW0Io&+(!5>LWZp&B8b*MXgJI#Ng
zpWVJY`0KGiF!@2AvaoF9TPU4)c}W}})#2n978k*oIY6@~@b0jsl#OAkQdRNGO*_Tf
zC^^JHoe=xAqyfiQcJ@maStwgOsP|kp6St=nwI>Q<8T9;#Dow1Ryn%1*DDp-0jp)FM
zFGb2Bz)se_I<^PcB(YG8k+Wd+w^r9QBrNUDE6N(dI52>L3!zrg{R}vwv2j1jTiXX}
zWM1ejXE4Se0A;|z!da|N`2*w90r-0MtNhm8aPx{>BhaJ#DL&0e?!IfUdN~Hqm1#g(
z!^0w?v!CRCFwb?tJ+;lo%D2iLy5{qU1Lm_0)OT;97@7Ns1LPTAi$p5;3@%~$s8t!j
zk|CTRf}{1&fyMstGFXRegx?g7GH7@#I4ZBK+FsN4Mj^SP-#^3P!rOjr9*5P%LMqto
z=|i!eZW%fIKu`!^hU=#b8~UA8pvktI@i|?MK~t99w;+8>NFcQyWELF!aA8Ah^|mv|
zRumcObhNx0%*fnq<#CRxWcivHB*O(nLi%Tg>x8dII2%euAzJoMke8Sa@&_Th=O&M>
zX+q4)m%Dqf(!gaFG#3+Y3SDZ}&R*nO@zp(4m}RbA<?n861-HB}><#5KoGt8YC1))E
zoWXY%R@n6_t#ahfHR{5<J3Wh%`|}=yZ;Mh5=nK}>h_Su$*AkgflTwA4fKc7}Q{PU@
z3ki3v#CCZ&W2DioIOjOxxXYW%`14zZ!FWp@Bm-ZWH5en8f!5hL%`e~z7IaYzqdQCB
z^mv})|HaybO4~OLRshYIL9db(-KAu_S9if35k{Wdnt~tvv?|X;{s4Y1cTiiCns?cH
zrj++I+QAMw6EsOE^y(oYgSuIleSQFrrt|?Qd<jz8SwdhvmHUa44+4q2Yw?9SY$`=e
zG6J-@U3?B0AJo)Tt?iZFFt$t0_RL3$jEvPJJj7(WBmrJkq4WE1psdF<W_T%E7QQ{a
zVrH$$Si6g?I0cu)vT*0*!e)*k!@13Om=Gzsu?O0{&CN~poUp^}ib3(<D%mkP+TU{w
zo10R<|G9_n;1KU$Lqa)=&Js-?Dnq+mp;W)HNv3RAHDa^LPavH@%E2CR-}5H>i*S*m
zSM2F9R74G@+kL)qHdogCT!)kq*EKfpf^R>5B)qkdeaeBm#Aau$E7Sf+Vtq0+3AYC#
z;Fa1Ah;{wxervg4I;p7ib~w1W^wq0Z4PB$rKD90^mTjfnP~fFF<3wf0(0~z_EBm)^
z{t~->e#K=)mFg)va$5<vk2zch2e5Cyee(AQ|IIl}<q7c@7I++j(@b|KVk0ap@>UEu
z<kc`kfm*R8BONO?qBpT!dS-qNYRZ)6Yiya~gKboJdh<{U3B`$vrAs8Yw7j|EuA!R4
zy0MZ)0VfL=D`sO&O-lPlXR>;mJ?Rp)(`j`x25Bx|X)|p3>SBONq_ex~jFT)Hn@65?
zx2uyTyz8Ei?~*4;1T(;!?oJ03Z3pXGu=+{03Q|h;0IVIKnvv{UcdA=Z8V-&P8PDB&
zzw;Fj9<Gd?I)YLwuIMO#x%{E(M};w0_4Y(-Q66XZD0_a5vIsf&`Q|izfd4C2wimf+
zVP8TpA8FQgCU+_C93ASOeQ|QEem(4BAxo28bAYm32e(%rSqiZ{XT)7vh=_Pfj3w?U
zK|gIa+E&#2+4W$ZwRArVf;Q$`x+U6ruD%zoL{;ZTr(@phT`~+)pr#J%f``U+8$lKe
z<N4!JM<cdV3EbIJ^XCMb(jqfG+V1l+i*ycGhPp{`VXpM_Z$cli@Ns(Z2sl~Gg*ofu
zF7~0Hy(KUH`SRzdss6Lb_WgHc;vKy@O-(^?TNH<v91X(W80uPxaJRLxX)UtG$~w2^
z5}w06=A-|TYULwwv0$U+|EzV`CZX|7`;v)aTib*6XAKE6e(N<@Jz+|x`GNl3os{H!
zSWHHl_?#+htJ5)VcuXZ6a~V_jrsO=o<a4H?Ioh39s?FC=*M-9C^WLN@&FfpViO1i3
zwy`W|<I>}+%MShcq1`aGq7tF?h>Vgcv4aHa?3!=e?ui@aKD)F&9LN)Jj%&8M2Uclb
zSEpc@D?g|P-6KC8orU43bExTymiYzOQ{3LqE3bu)ot6Ev{rqT5-Y+i=v~_`AO^T|7
zi?wDw#<SiIp^U~`S?jhT8e?f`#}H<fM+-A$PhO2SgX(cwrInj~z6Ycp4A_k9&Yt<N
zU@9y-5!JYQt|q#_b+@`FNl))cQ^c2g?rLlZvpFJeU0N)mqKzP7^!|A@XCS5R_i<y~
zk|jqyk~_#|!IWB$Uqi&{1A4}tJG{fiMDyy${e7*g7X7@{rD0#0cg_i|13({Z>aE$m
z1>e2iheq2)u9^u7HeS0V0QS~v?k5Xxe!gR|yo=DO^Sptc+Y#}H#f4&&)C2LtZ$IbP
zous~bmCSy1cs5epoSIPg@Ads6BYiaq-N0xP{GqPU;sG`X_UDh^hF5(uKa{8=8H`LI
zwr~ioMjL7qk}#g_n(ELS<+5V4r=+BZCgHKm5$7_nfGIqopqn52qCa+GcRC|vAJ#x~
zaT$fosar$-Zoiiiyq@n#XOD_RHWwC<E`%QGk<B7>jBUu?z+MJBs#IvSB#6^pD^M(d
z2_CpnxwB3`wLa#98{VP$JMYV3Y55#PG>&z;uh+eW#2MV4ERV0*>9UT$;2DR;X7u$W
z;qN+xjhNQ#__@*zT<vCA(Dn=^3Yj-OnaOTML8rVp&V#;+*~l75&L&8UkTqxM_~lmb
z{7gFCvo8I-XZ$|~SgBZ>>yeWSj2;rC$)SE&BuA^gJzPPoHPLuD_tEE<nC2w(_2ZZ9
z*vP~B_HaYAHJV`YJeViz5?T;y;^}nf0;BO}a8Lj{)7?9tHlm|zknBHF`dQkT&ZB@t
zBTeGOD=b;(E*{eqW_|qOAaBv@6o)uHeDMo;Qis!bn2{l2IS>-cN?B~W@I>O_gV#^-
zrKaJQ!zOW1s^;?7=>5J-?8xvPt)my<__;>$&o>;EDIMP01f8#&Ii^D5!lXusx%39K
zU0=WFCyh6yzIh6Cu44RETkx;*lY3fkjmb^RVxkkFp7;f9uVY7R$3`lqq<<F^j5*fi
z<(rywwzP+`mQY@6SdO6Zu@izne&-g?omWYMXnpXLnorv*wVZou);qL!b?3Hit>;c;
z*@uhFp^ecnP}-j&lO0uVL-JwQ*Ezt3k~!Mbh^=s&$z>5ezIa%7-f^`(QTA5i!A4`q
zXSY|JY=fh1B9!x!7H{*2_^>^;DmL_wPB6H~%`{&>zPGKea+|<%qF@j1kx{<kuX*<u
z9bNn2ukVp<fd$O=W(Jyz-@DDGc<rmS=FwcO4<S+`%+O%bbY3p~AH5?VU{^Da#?OH6
z5JvMSJip90oeb;g>kjLg7v9yc=Exc}^{QhH0#-NP9Fv75`gCWRT9MZM*+dXnj)u~R
zm<q}}EVu`oZH3KuQj~I|cSe=pXJ)LK$x}wMZucC?<S1OP)CQK&teHhdNx<!>u_}nG
zojCTpXYk8qQp@dL-*E2nbAm}JF~cD|?Np^)RW99EB!}z1#etkR(YhZzJkEiba38M+
zyvx?#6;2Ee3r{Zz_%9XQJ|e}e`SM%j@V53sLwXlmNtt0!R`Nkk5K2$4XUU4}hZuRc
zI^1g_C5H26s0ewU>z<v)2DGk1;uOkt79>-U=SES8f6Q{svK7zkSDgHkV7b=MCpfs5
zqZ+rjy{+jBm^Ux;2`M#MtN7mKHxh*N<(qGl<(ShgErslm7p9kbiEch(>AD#Zoz})g
zv6P-d^hkfuV`tm954$f9dE$$$vP~SY=A^gyy4PUCC^n(b8odh}kPDZRD%NoK$oG`~
zpZN1vuiY`>?>+-nU(TJaabD1mSK7@PG{sGLv=nNC7c52&=DTj92;O!a^3$4syy%|C
zV)Nah+%PW*(%G{{x7E$5Z@}wk)~lwlx=+#e*qpew+jl?OF5G*KRx(yTSJq2JLrpV#
zF|Ca9k1tuc>kIgYf2{46GLtWuvxKuf!L@OD_*}O=D8}svV@fY8oQzt<e8svPOJGzP
zW!ZH%k{GX0BMGB7E*G$Ie||iu@@5nKH|V%MtMgY=Nb;|+JP>DxEn_MUa|*GHk{VIS
zavZ1D*7G90$lejkL;LkhV}8c+@;tcx*8(V)qv%ZPff&Og(O`pT7@L>bNSk+Nub!@P
z3&~ck<VTx&wPk%nU5V|ZX<ZB5ru0Eu?)>ww>B>#i*)qlfL0wEUqZh)@DhxP1t-1Iv
z#)~c|D{KRwV&e219N(t*e{fQx9i*n5Xu4-OEaj-C9vrUS5YsGN)|Q1d<=4uD3o}3)
z=CWA|-J?*L==c$4aCClIfJXL0(689rmzQ_XQs#e%upQA<7kytmdo=huF*5S<cEh~6
zx{i@)6GFLpwb+pCuHKw1@>7r$73F~C1uj1ibPt29xmt8knhvXStmd-Inil*1sgv`o
zVuOvNy|=*W``z8Y^Kluz7dBbKHC_v)$jP0|kK9|AlvEw5R<CWpahxP<6xY);V!ifY
z7z?wsjaq@0gHYgJ!`$aun5J7!kBGLV*QsN&46Rf4F8?{6hF^f*@X5y*6_stRqh9BO
z;w%RTGYNlHn`3IL)ztvr|Gq#^WApC4Bm0i{HM15^?P`|ye03R0br2}LWA{<<*seU0
ztKFO1wB{zjlVpyo72U_{biME2vzVM|f!?hyuLgRDQ-A?<cz%eJlAugZdg8HUVUY)I
zF>oJq<nG0}PHOmQ*T*Cj);0FDkS`JQwVqKS`U|t~_L$`tv#^k+riJC2tbR)~x(T52
z@pRTx*P|vIsO+%b5oM2Ncv{gkHCP&9yy0)alCU)r=n$AVJ6wAf&2IdrEQ-L#XF^OY
zBR=#NUSd?6-`$`cbrtgJt(I%iVOTd`9c#hCvtbp(xj(mu!G)VBq65xxR-v5V@9qer
z3*T3(_5s&O?2LUYMz#T9+QGyqL>atoGEH1FzyN)Zu7321hCQBH-F{ZzLa07tvr=94
z#oKq*vQINHLqr(|^TsUZKO)BZ1WgQKJyqXhvAY&qGL2?d*A>k3_`)e@wC>&ijjFnF
z`!!ELV&Loij?a1d)e9!$?33!f38z6*cIh(907Zt}I+(77@vn7{a#z>Gn>@V<nRWvi
zNZlcbnBP|}Jta0acZ%Tb^OBA23(CM-;jz6JcxD2r*`w!XcdyjMjRsqWTB4}N$_+cx
zpV-G&H+0RLG76IZ3tN79aqjm0cHs-QFZ0PU0X|;rd;KtU({S8Fd&hAV8wQTR?sISl
zUcOyhHP7o$=P1M|mTNFj@K2PaAq<?*mVzH@X58OM_`$qS<1OYvvJ*W$9jO-I#8KN_
zHcN)9fQ3A>M=dRieyl$_R{DRU!=@5)*<WV&WN*XJ9#}4EZaGcD-m_Xxak+bKkF4$8
zeC^D>UuPBZXcckB$IWUs|H#5%dFo>8Er8*O(j%jtH$7s?;c5Y?(KO42(@B+3-UAz~
z(6hPv>2fR&i%rQQR_4aumT&T`R3#fq*aBMj$X;_P!Y-y?C>teg{Ek>PSDTs?F#{mL
zej;!w^nl>_;2?6#4;*lxw!EZk`ti+l2Ju0q9P**%QtQs~`|>#cRQ5&}dh$G$0-Ymc
zM^+|&ewWXw%7RXoV_~N4oC&W%PIZ2N+xSa*@Y7Lr^j|pl{r4ptFGC3*)?S?q4gaHU
z8KH51)>tG9p!a`6f^)(1vnwRsKhWnEiJnd*u6_B#HQ}3V6(uh@m=YJd^VG0vlCnE^
za@6sJ>`?EdbY3RQGN&!*r^@6Q>lz6YFiB4kt`o2o(uSVx>smvJS_?QhB|9ym7F1NB
z@-iawR8%+BkC{_!p7c8pN}&H@8X7z>*O@MYE$hJo6iAaLv<eO!<uVNCvs|`<b3M_M
zr4eW!=MU5gzXwn;7NXgZ3A&n_IdV~>k;I;DeO4fCMD#{<r(w(ewAr?SSx@i$SXnWh
z7xB2q;2QnADRLd^muq7ADZ)T21cC!ooxu99d>~pvE=|<@3>xp~KnAuN>8IQQka$gT
zOEosfs^2<?sxo;eVJpWf1UAw_N1-lHo0^RHlC0nh@SD!_=m{~2l>7?iucv2JBAvfA
z29^SBgIa`O|DL>mt_VpR>HZ0LQE$D+48xj-?i?AYk9rpb44LX0nTTFJ+92Td9ve}K
zVl`UW7FrcA_&vB+#$hp17Ai`vLO>9#p263s9xhU_N-SwI`GqDQRnDx4<jYyx(*nzf
zHWFpbWbo0_R188?<yV$vJLBHR$Rz>=F(~=1ledb?@&j#FipwU*hT`sk2h2ajV6ITP
zpP2ZKB9oaCv@2)>fSN0;ds8)Yx)KlcA4(l6YgU|dTVzrHm!|PLgWjL#1NUJ;PH9cb
z7MIpd4`Okv(&DyuWO(CeBCBHak+o9q)@18+b0txeGQHJS3Rj5hTdqm8eni+u>hL*J
zo!%nDDc-39cA>|4j!+Nv(EhBx-rhD<!Ceq7_1ZL^JI#@cMn_|ZN6Z?VONu{!tcB0%
z8$o<Phxbv}_*>1px7Y3J8(8Z+1;OI?F1i;brPw@JSiE`k*EYIv<6>C2Kp{k3DKO6q
zr{&R*lE<M-m^S0GvBpi#5*tw_#<sx*G-V8%s_N?gj#g0rv2fW$pm>y(gK%Ll9?pyC
z9b}PmQ#_c>TZA9k*`UA3EuEd$R|wS0H^5;wPS^tG;HP-kCl1a5m$;gr*4VU)0f_Pd
zPxO1wqxCtOxC|n~(0(^f)@p5Z($T(tZXfVt#W;k8C0K;Kl8gIy-*DvA#yK+wo7wgR
z@f~3pT?!=UOB15}Q77$+FgJnA98vz?;2b$-P1E!jv3$dMpU`W9gcB>&)E_mS%tyk<
zAIg}ocIjD?8!x@+B?H9H;mV$o#?$DK=p`jny_&0tGbM1|j*Hv(PF<XeoEM$|4<?h1
z)adGdM;kmdyKo%YQ_r|?ADszuy&^*e5~jY(&a*$)KW29yt$utLMpx(N{%GLzAX~m3
zOdxhx*prB0QC4-ylyil5EEv7FZ5V0N+w9GWh>nwtjD*hE1#fC}4c1Iq$~Fs`Pca@%
zr7EF;D7a)>;?=n22iasf6Z2(VonF{mqNssnB7l(>#3;a8yy9SMfsnlgYOaO$m)C*z
zw!c^X9I+DIMDy<S8o?2KZ1iQg$ae%2d0|b>+gz>Ogz*e^UK@T!#%~|mL615<T1wZp
z-zy|+?`l;=Sy|%YHR<WzlSb&j!_cm62h;i_JT~Y(X=A(MnabnNPNtc?hp{TfS8oF7
zq8TJ5qxoE^Ev2D&|Jv~MsXdoxHJX~=#Y8E1$wD_bB~2}5zUCMetnK|bCv)L*yKwtg
z`tlO<$-FcdffW%NB<$kFxt9Ttf2h>frXL=L8IsVLkfjpn`OS)++<#d(;Ny->$-x+?
z*Le2(C0h3Trn;5aVCaZVfzCEmh0HBnAus5`-cCxj3zi0bcQ9Re%QD_4a>mcBJ~V_r
zk_~(9(<;oaPZEYAfa>%Q)IU@*JS_|}GmEsCHk^swlGVr%bwz*8!_wW3q0K`8SfO9k
zjI*YTe4W3TJ}xb5?@X2x#ho7Js&Q80PjP)u=W!@ONwGD9BGaZuy}Y2BdvE0F#37;6
z)89iK#4hskyIo!I;P7V9#{vUGW`i^SJ5HPiAc9!0%FlKiPY*w2!4Ukyed8K)Xzy>e
z{-o2XahRjOAAX-ePY*cV6mcv(FbWz)>hA!Relokc94hB{cKqD^VQm|TK16Bh(_|Qp
zRF>c5=W-<yvi(@Dzxe!nACdr&WqZ(uLf=A_D-$a;7;Q^3F@wDguzyjdY=^?F%=wa&
zo#MPA0ZTDxyzKf#Ls2i^#*erk67pEKvr#uz=UT@lzx}KhB*5zW4-X>VXgK~E?*3-R
zk(Z~p?mU=M$Jg9L*_8WJ7<kbKt-}j=kss+pnL%9c7E?8EXmSHGRtiZJ5t%dAAoZ)T
zSU$Stt#Y&(%bw7l`ii^=%e(Mf(VmpVSIiooEn|LRK{^4Z|1#uF)6>&rJe21~tr_V9
z(B?R~q2cV7F+%ymrB52>*3aR}Q}h^CFgf#8e)-tsNXvqPqWCNo9r9Do*vO}UKA25*
z7u1ezK)!Zh(LcTjXQ>dU(n3wh!r})js_t0R@kBaE$LVO>5=w7!IaaJUgw@qvt;uKg
zO<_bntDhkYqs58Why(FpAwHIf_#N64KKhWy(uvEv8$6B($1gm&?#&Xy=5w5s0n6>0
zE+ek<kA_RE|3Um}YHb4C!Tv{U)Y30<<bW%bxWCZ|=&LDi_n7A5d$!so!Fip^yUS8B
zNp^$|ckenMS1g&e*a=yVbrc`J2iSB9H5wjy>sD#q(4|Gi3sRO6r?X>|z&470mq|+A
ze71YzF`K~>l&L;n`8rfbU7xx}8wj2bJ8o9d*ExD#%ZiOn9*o-W6>J*~H%}Fz4B;^j
zJLa3(yrK;Jy)WVa;Ni3BlIo$TzR9ZXwV#QqcvfE>s_)KfMIYw{m8R$_c?s+4e{2}p
zoGfj7usZluo{Y>wf4K;VwcdFQpoPNKdhnK+ZoMjdl=}-Btxm*)RJ_#km$DDdkBfxy
z@(HGDF$X_~rI6w#)HQ%0^f{j;^irU9msI*Of!r1s|ATp#!<l#-x)TQaFK|unS>p^h
zifYE46ZG@vP8h6udP4RayH5;h{^JcF9E@*jJ;9gNF`?00W+G)~^#izqi(Sw9VTb0!
zTiIcmV->UKc{HW=Me`~m&#ulcg$GR!v|poQ7@S@2B#DNML_yC=piULs{k0q$tUTPq
za<lkXzMi$!$#+t$_db;u&t@l&Da&0VbqD&RD_5|<s<WP}Tdf!Sa4X-%@6f+y_>7FC
z8i$9x_9u9qYfim-nRy9`RtbJTN>`I~6v72E@=pT0r-`}RU{p~UT62t(mHiMlh@YOu
zMoK5w+?<9e3-%vpIfcT5h!fEpz!DSoNN};Z@ow2Y?pG}tobF4Q3~k$%s(9hMfTMxv
z%*Zaml0O@Y<DIDcM2vYIHA%G$YgngID)+M-Z^}_)&Q!n)k<X5%N@-N0VyyY-ybHfD
z4ZFm}{bbP+u)gmR+2&thRhAnX@Q^8pNDwa90b$$?O3qsbS4b%B_0M7f-r3Q8@E?E>
z%i2WK<)-3zpL)QBrD6Gz8nIi9p1|n)MEl$$;o}lFS`W^BS+n2F@*rJq$REGy=gW&S
zN{3vw;-Z;Iom(n^IAQVmcI9?O)-Kq*#@FthoE=J5;{7;XM60pOJ5YYw@`I;X>(wKO
zIQ3onT-m+(OM43H|H>eIi}_>g=`~6cepNc2I#-pL4nFCAfGSU#7#Q;fGcsdl(Y7Wi
zJc8xA^Ab^NK^rDU?jb>IFfj6TfA>?aJkPyt`{6fB_N(9Usi|oiWhh;u<nv2{NxC!N
zC4U4c+?|L6ZJxWg3(*UCO@<2T!F0RJMqbxVR%sg5RM*~0f>%^i1`I{(S4V~iKZOE`
zggzR?D~_uvMMNn6x*<vGJpL{CHhd_*H%oz3=WSs@_v&w4_8-bB-#!Pz`X(lc$T1a6
z+2tKJ=S|gw;@fUiM#nd&nY;X;3x5&(WMran$w@T5bnj5;CjDz7sC(=?fgLebVND#(
z_|atkQ@#T4L&Y*h!1enqZCVie`z#}ysbmIaN{-)(mBnxd#Ty>Q178{EbXA<qJul#B
zM!OmKh#VK2MQ5DYy0dh+8EH5lFkS=g5@LQ_^+V_-rSb||sXmu>FM9^G6Rc9yM2=H@
zd-PBS`t;*XKS(ZpuA{Iny;MVMx%~0Kd2~?UU`OyqMXJ-$@pA<`i55BqS}ktd>x4k5
z>00AN$Cz^Cf5=Hl8o*IrOn|-U$II@r3f@u*c;>G|w5C7R)zkka{p;C{?aNN2os>SW
zgtXfQgXs)13C)=^U)}mh3&X`BHQ*~Bly1#MQlK?_jFp!us9a!8$9#`Yk4)(S$+yyk
z!Yh|?l@dn}!*J9Wc9bmg#}j&5x_QJr6`Bga*^5JVvk)aL`CGS?f(XBSR_I(Dmz|Ij
z0~}*-g#g;N#ZeMBmBZG7;*}PNLgi)TnmA)Ckk2^JVkrQ4_Sa|G?d)8waL0h^ziH%J
zk#YXcv%VnPrK}yYpiIwtb}JjZOsyx6sji?{QpNVca)EYx|HwZ8K&-n7J6`;v+3oDI
zOK3k?P;f4s>|Slrk5Ql;SVr!4tH#EbxXtpKG0>|J<2ObzRDbWyulqbxJGYDRtlo|j
zgp<u<6aGXP0xi9AN1Za+3Y}1T`m~OrSvR1boC`)<R$F4WXx;#3j;9^5eZ2V<FrG{P
z_>FmQZ`qcsR@rc7bHV-1r~CmkYTIVI2Pjeh{R2uGIzNNzqrvKn=9!P?t%vA)MQrQ_
zAX29^pB8`jHo2F>!Fu-@+=+^=nXI_tr@p#@%YR5SC{d8y(NlO1%<2z1Y#_)5y}9fp
zZ2bstDMl%cAnBRi{P6(Z=F<)u`Z*~%7QoDQZ=_7Oi38<!;$ydCQYPY^h#DB^qC#<^
zZCii5yDlNQzosXKS}HB7NUzNEuf=0B2{-|}Be2CNE_nI&#r$tw6Qf>`IG`o~RR_t#
zBwU^E_(VsKL_OP+%S5YfV-D$Xtao{WCa^K8t~W7-5o5#0=617Z!;D6m)(If$o9^oE
z?M6(W0N(!g<sBx*efW=%w#dfFduk_C(i?md$r^#N>FALOgJya594tTE7LEZQjs}`E
zvX>zAEGvH6Rp8aP7KyBJU%>W;bmx6Kd&u&$qJaBT`X=;Bci_=!{&dquoW?cLzS^<M
zR`zoNaMC-^tCaC8J~=E;9Qm27&i3|Iz8||P4{UcJ(ckOgd2eVCs-@9%kVjZP<vZ6=
z>up@^$^fXj9UEBF=oaokTdoB`Dc6{a4Gd|mVP@{r(Pi{Ias1g&9f+siOM{G@FLW?P
z7r30)gs}}1shP$nU8W%Le@hy?eYu}a@3!L<f02tcRU&?UyWw0*6U9blo-<a!@A!U;
zA7A~<kU#EXlfYF&bA$lL?`pS<x9sTExWxx6G9K^!IZV=$pQDcR9IOsRFJ>z2;=W6#
zlo`qQLRSp#7N?fKA<F-FQ+(Y}_SxVpF%L;I)jia5Suh%w+Nl)<!&kmWyw!=HAIN$N
zF5M^K1njRy#<A3i2z-%-SwVP)US5Fv2vJt>4@l_PYWuV(g^d=9|0yU9^B$J^@lFTa
z@4VE7DYevWqI4%M<Iehm(rHkt$#k6J>@b%zJEVZx0b~%`tCZ-9-k9pL=ZnJ+1Xr2o
z23UbOK6^fK#qJCmq*z>ZR8{Y5MMiHM85V7Kqu1T9ld!emCrPaR)oxT4i36BvKr=5_
zzjeE@vM|>YUi!d3jt6m^z?ajyRhzVGq$@5kLC;XRJE%v`jnxfUH2P}%wx79UD5$wg
zRc%`b!PvcR=_x^~W431q!sW2qM!b{hm0)hfG$Q|AqX4~5M>&yetrG^wY#7cz+z=>=
zijCSn{8n7UZ^f%*IL7Il>ysN#WCi6dEv9!JUo_}QQIcD+BriF)foSRcatO9>Hjpt?
zw?&@$Bs|2oK9w(jePfNi5M0`6XPe|%%K(_x@1l~;!a<6&j|E|f42{zrp1RSR05CP%
z!V8j@-)pkjhdR6c;h|L9ot4Z{u`5S{f&U3rjj!FM$i#Q)1`*2CuO)h4Pgj?ga#opM
z7Bgq0S9Ddd-zzAvcqqOd_3YQm%3Moyi~bkODShLQnQc&%R&BNdZRdiunWVpoh2?ls
zU6s;QYSsdFetRfO$)+C&vngW^xje_TMIZYG-nhJq?u!IEWKOtoPZF`9W=0r)?z+XJ
zRko@Gj#$p|RqNgw^?dAXKil4wd`^fdC8qP#I^9sQ0Pr&<@aI7lupXXC9ve%8=>m%a
z4biR{Xus^2>AfkCG>95x0~CT)ZXVydg-IjG;IM$g@WT)7R@}_dVJW)43)X;@-YzTC
zab&f0c4BK>Nx*N?oYXkEw5I07LY2Ra^US_S!pjjJ*&eV2GUJ4GnD2mLQAI}|QCICB
z><fMy&&@(!B>&Yt?bQ!4)x%vs*Nb&CcWA6O@*$N}NLcs>S(E2pcUbazx>tv)C<+5f
zgqI`i3X|0j-1>eSJz_&l=C5m5P*Pr+W7Z_mWEZdaL|-rdT}I8NZR8U`+yD#N(%5;Q
z3D4+`!`iQzhQJi^um4`5CV@CW;6d<R-}}~nzS~7V^BgpX@4};CN)UR+;u1&O;C@6z
z=-ng-xQ`25DPbT!Z9zC<CO}g9ZbR!toY&;bPsP9Eq}{EGhvw764|L;uP!&ykZw<Zf
zs62lz`(n6mR?*PiebE))JxHUw=d*+b1-X*=b{p75LL_{rhVWTK*JnGX$H~v{V$y}q
z9El;is+pFwEFy4$%0Nc;9#DLnRel^yP3nSQSjo^(Vr|k-wAKX)ZVr!*ezw<^wo}7w
z;+ndzeBhS6F@}t+8{RC`)zj%6ba#syD@3Fw`2KsCU$%b$B6vHJmF?9V%5b?1nZalB
z<b^L-=wg9&b2T?Nwy^XIjtCg(3=>0>@b}-3@soqnNy!9)(n=o?nwfvH`IL3o>82T_
zP`CTPYI_T)D7&{^d=vpC<Q1d^loXH_kXAtHE@_eO92y=BP?V6N1f)9$q#Hy!2I)@e
z?uP#!-{0^3zW+Jvob!F_tTPK-v*ej)_Os)@ulu_1y)VN8<yGELFOB$c|AgFCkOPwL
zZhw}nK4)eAp|p}5;mO!%%R*$R;cWTwdx!Pwmrd|Ujj7=P10mknlOdOamVkiBz3h|9
z++6PT3<5be8B1s5`4^O#HBR@`=2ubWfyuvoxbRhYwZb8+h=OBQ013Y1;>e)7o4bDA
z&=S(E6+L5&M7=Gr*18`ZuJbZTGDT)ccaXE^%24Q{?%h8y^+s~C<Xf%dP8kM5y>%Oz
z_Kpl`1IKb|D_2W$4r;-conv}{GPbPgo)xRwS*?*gTAjpy*OB(h+xHqwVO2#+EnTx1
zy66%B>6P|mRxM_B)@m<##z-{NFz<CR)BfNcCr^Q{h(15I7i^AYssAsHrGb`WJ4)4S
zKg~_HJ2!pSZ`Ud3>-m33#h|NlS&T7~xVdesnR&UMpZ7(YJ~I7A#8GXJp`pf-vAJdT
zL4|)dGq9r|+MUqH&ko&BdHq{<VG@$ue5bduels}4GR-TE8Dn;9L1W`<VzpUnMGF~I
zfzv&j=vCyU(AZ1k`{7|<3_r@|+x1!<=j>U&wnw_r5cC&);T*ar+EMp8Iwx}Grq!=+
zAadk3|6NFtScI(+Z>~3WakQ-I%Fg<Oc7%vn&c@)<r}a<bXRRBgB92E#p^1xX(dlRZ
zMD=fbzvgfR-hE&H^k^tw1w(3{2<CpUEfdSG-*~9|>vv=%!0O^(a(OUnge|sye++vQ
z1Z)F|Va=z0xB9|5r=}zcuGquL@sfi6ifFsZj6GZ==fXS+%~!}i$J~g&M?SOg418Sk
zMZh<_`W99(=u*Cu$sF+C*@6eU8Z-wEHr^u0pBJj1KixvLi8DqLj!cvf9fNf{rCo&3
zR1)osjw^GZdk~hP8MW_VFP@mO35a(F5O^jYt<!D$YkfUA88#ucK&qVG2Q*$sq^u&7
z6*68|>jMi?h6CDIxPIN+2L`-+Z-e9G4eOkDkl=Ugr5EH(Ec~G(HQqt^_&ctS1+LzD
zp?1=K;kVk|v^drz(lw`j5;6{um!7qNfVa+#`W+eVEWoS3<v4S-rE<gvcC9KIf0mUo
z-4@#7IgUTlcH1YzY;G=bEX=IW#MY0km;p5b>AK}H<yQE<p<>+#TOsA#!O}y@kskyQ
zModu6&83Nd7ZarJ1U?+J!dY;-blWDQ@-;*eL`wvAceb{+#2QfBf#9kMy!%??T~~ik
zw!wbQeWlkCM5gLl>i0e?{cLOSQ|zWJlS%r=AlzS;%bOvl4*WG)U3fBJ>5CscQ|ugl
zQvr43mEN9N&AkEWO^<@!m&sz<qBQZw<PmpKF=U{R*_KfcV|wA9D;{Cs`#bt?t*D$Y
ze$5M?Z5%~tsrUYB4lJ+6<ZkOE;%U0NA_L8R8?JV8srpc8OLA|-+|UV}U?!S)5z#}8
znD=sReabj5tAnN0&ka0Z+A#b?M3~S|GeIHDidn`H83fe`Gd?<hzc+0E%IZ~vu0BQ#
zNEX+X@f|g%gS5!XB2^t;T^A@-`^$SCOGqqMX)QwmekfL<CN)GO((|d;8gF~Icg_2|
zoZP4@D~{hVV^BZE-g(xYdsW_%gTn<N?fX<7Z?#_MCH^CVVAw0C#3>!)gnNQtqoXbX
z);F;;gO++#QN+>p{+636A1E{>#&xx4ZOr9#6xk^Yrso1(;b%Lu&j6Jy&B5N-*A)~+
zJ%+qm-qaT$y3@YfdsC4Am;E~=AXI>#G_o@J>318BBsedmBEUCB*C*(zzXwSXKqb6R
zOC;bs%FEHhFj|Q@Cv*x1ZR0tLCf2Kf96wJl)n%0CJtx-1#q})TTLMH0fTYIHEpCKq
zkAjrPj%T!@y#3^Lg4{DL45qtxzq8LV%CQ@mdRW|OY_}BtRe*o8;8dg%^Us)XO9v9y
z{Gj8iGbgC|7B4^%_te6Zk&m%uJw`Y{to3$gT-=%us0*@5;w|&wfqJ@Qe=zwx&NYPy
zKGTC5Da(t~na!Cb=bFhSj_seIq)YkXagCu;PuI6DvU;;42DfV_gttT{;tCxn99$JB
zf=i3I0C8yExC$3Mi<V6f+W3X<S00-9O#6ezdjeVZN3VDMb77Bs?aF#a`kVw(m)L%v
zA5gluIMSZSkOPAZ<?A?+K0c5zyp1<=T3k%Fuq0ulBPcO_;uo=gFWwkn1fI4u)w1c3
zMH_@~t~+Ub<<qd5w_uCuHr#oTf+xqC^qy@V;8y3(Iv;oj7oW7)vZrSly*6nZS7Wja
z=Hljj^(yl=h~lP19HUOBAAIcTUTS4ElJ?s&a4%@7@CbzzFlJJVN!vJv>sI2}({V6q
zz6oPD;HF9qksS?uX&ljcStz>8_#jdYF)&;czQnBdho9+Ekl$ZTE%KLq&f_^H4j_d)
zGpAo~CXI0blzWD#PGX)UoLO&Ek~nh)Y|{iA8@dQo3EH#GXZ2W8{03{#ZB-(Qi3CLN
zCM4OD&^)e>kqbIDR!4gmR4LIW#y{sE;0Gl1PETMztVY06eEwU-Kf`-}`58K5{4vt|
z<uDSts7pcacwQ8Ux8GNra`!PZz5Z7^U41ASo&}*WaALiqjG}b!Vhp|gb@I&?+8S#p
z29gyIWU;-xoLo+C0LmOX-bV$m*E*<}pMT;vXa+a!4{Yxe@+dI4^>gkM4myN&e2#go
z+k2U&*TDRovi=d8Vs){b`w?bIRKl1>ZDMNjv!$pB&1W!=3ZRP?<7GIdD6Od}dkOqj
zgpR`1U#f@lycMIh-<50Q1wjcqfz0DND#vmE8YpXl7A^ErdPcCHo0#pLzcGf@@Z09}
zDC|bnAL0PpbMed3+jUd5wxyjTn#XaR8YB-C;=e8b)^ffTEI{{Re({wONP~gU>S?m-
zpa+`%+`|rW;t|5Oc6KHa5lwA*Nr~&c*!>mM#hL9M)|%Xn9$5S4i(lDt+7|cZrS{I@
z0Mve9byf&+ZCt*MQV|^spg=AXd3uzihQADoVZOx(kh{s@^2G~x69~Dk1<?#-V<_?s
zfg)PGgPqr?-&{|=4%Xoeg9@qI@K(;YYGz_tVs_WE{oI8;mVe^9n*`wie1=NOXbYUN
zTM*iS(UHvctYa6;SHLxZ)(5D0w>LrIlkpN#M+#tIbTx)-?&1MOD<B^YKM>rQ$v1=~
zlJhDva{L2t=`PlTl#bC(b8d~x@)IYj9SsnGlW(_b7G#*dvNY=`LW&KO(~y&*c73o5
zdAZE`M`i~#AFR=WiVMp&*qET81N*Mvc+a<f^7^?0t!-PV?>&f70gVD;wjiBWWehoX
z|Lv|iI+*tpXV2gTyquu5(Rl55*eU9{<mGUFNPXyKL3dci4XpE`NH9M<OKqFH`ZJe<
zrojxiDy4{#{E@=bSOZ^oiANy0JWlp?>tiTK1*GW-x%c(}zUv=d9GF9La&yio+X$!f
zZP7xRCGzINPm}d}N~q389WMB=RGFFTPEv+csW^hEnn7W*%%g{C!&TiNb(aH2k8tJ^
z`_Aqx5m}pw&E>Yp`r0;Kqv7nh?YK-=*XxKxf-RTa+Wn4Nkck-;_N7-1d=E~&4%b!R
z3ngS~LWYG!hh~-%TfZP^ER0(7=Aes#f=W|A%-LB!0~X($u3*?Y;XgwmfG*wX7xN7y
z{KcOcD!IKlJy{-X#TKc4W&FAelmh;E{ko)u-qf_{a>Nl3FYe`D=Cv*4<2B?WA`=yn
z5nTm)je^D>FfqUHH!p?BGUVM1PP(vPLzDW~Jr0KViEweWA$P&AhxPe%k41>&czAlC
zdNNiXo_OO&Q&m)e0gKbea4h0P9-V-|@wfzOHn_J$$+r>|DVwf$?szMqrzh<$alh9F
zV}h7NP+Q)*6>CIiq%817zpOVHK}7}@xW!(o)b=ye%C}?Zhb_55n?uBabXM~s-~kyC
z5F=|qat)xqj6E}BZ=~lUXrd(ieAh|v<izQ?6XkM<RLwo{?AN4kAdomUGPdbDik4V8
zKBKXzag3w&cwWj^<Z*&dfl)z*D8RgcDyMCiGd@sjUTiZ=9Y60ODCD@@sRXi|zumoQ
zZKIIrS*xQpx)>3)vp48)9*s0ziia%lOu6}>V^Q1MTB~XsFj0IH&nV_SU6@4c{p@o2
zcUQy0CpvoaT|+^2kLy%m#`tP5pb&y$inu7JpbrM%aDKldt#H=6k4CGgK*j+So~0ea
z4+r56t|p)o4-|E3*KIuwb~LmbyWa3NLBEJQr_#V)!6qm;W;&M@W(e6#<h9CU{zJ$m
zJAPCcdN1|5b#xCVOxdf6q`z%J_()BMcPaXdwRbffJ9o;vR;LS8VS|_@qtxzAX^29!
zOBf7=b&Rgt)L&xftjpX){N?=Oc4mGkKRPP{szN|sx^cUKYfM03#>z;sQdLI$*@N=$
z&Rb<G(C?+zUr4d0lqiZU*~m|T=>gHP$9H|o+8*pzNp5POwDe}+Cq^zFT!PkEYck!K
z7)SF#rws310Z>C^Zt6}CNX4DP?`bjlyqYgwm_$T4vC(~=!%;vCfSOMlL*F;r@301}
z4aP~l8%QbnieGJbhBPxdJrcd6OGKn2Fft2SSjpOXkb}m^AmtGqX{u(SWG#K9iOA9|
zIez=Fs5m)-BSzuD0{mK$3HQqeE>P#T2Y69X_*>%a)G#^`1PaZM&&5+%v{ooOO+Z1_
zM^N#|D9E2koP1|pn5IjdYJmvmS!O9Q5Iemrc0{#FpSe91SPY7r9{~!(u^cLqq<s_d
z<2CSh$vKarXE&D{l2KyGmDRfO$T;*MFV%Jq$CdA{&`=(qSUnLl#Bc2!!=a(H2E~0c
ztn3eHc|Id>Fk$hrXGQ0w(Uwc|`b&<~Woom-Z5-@IfE@$1X9t^0qql9feKrKEdocHg
zeg_VQkVZr}2tUTtB{GH>%XR_9!kizwcN(U45C0Y*P1o(!nqK~EFU5^#$M$z_8O(-B
z=XeKuC%J9F-QeWw8|@qqhM#CPvG!>o6$uUxZAPAm#vYo}(yrZ|u7$I5^If$Ay1U5c
z$N&%%Ig$LF5$zUh8_oPE5jkWTTyAJeR-U#x=?h4-L%>KE)IP(fx9C|J{$l+t_@jwc
zy{O=Px+-(rQmvGW6^nD!d4oS?x%8ul^hHVAb)xY$HeV+)Ne#vLDc^b4?D`_hGZp9j
zi&EWpK9~YW9yalVwnwC}JuW?)*=MYLAzI6@MZ2@bY3y4;jT?|k0;w{I{ubvad<ne6
zb`G2nV}e}v%2BDMop~pyEHqdipyrpu?Vet@wX`yok+ivY(%k$qZq4BNhLw41@HAu!
zDi^zSOQUoHWh70^h{9rg_+NmEilw3VZkY<%ZmQ=~qEoTLVN6&ACZh!+>rY7K!95&L
zFx`y1f`ZBWMZUCR?HlWbZc48ze?CRJ96@`l{e@fT!qbtsSf|-Vw%<6<ORKujv7>l$
z8pN~Z8ENMJB~1Q>n39|8WZ*0SAN!WWWzY6V{(}WTSHEpXt7QSV%yO?w{=k#>)L@D)
zR>QZldCc0D(cE04MVr^+9!S#zj!mY$=S9BeumIQLA7TZv6gG7dnB{Vukd|(k_;~}e
zSBkB!9Mmy9c-F2Ro0e1#Z|Yf^5><6_TAj#<$z_$QJHG38ypQ+8*tor`y9qTJf7o4P
z&BmdkMZ(DFre(5pm*frwm4dt#1xO2(9PP9W7WneUT~jUQU0N~TyJD4Kus~eBs};#s
z&X#6nVa$gr$kZ?-^w_!z`Sa^mqHc6MM=L8!%PzdIfOBPd@wd2a1<0n;a*MwCSm7Qa
z%V1k0Y`Hd{?<Ri!Op%Z8(O2Ggn8jvb%Ymeo>urAqh9~2reLtrRmKa?^b(~xW_wuh+
z^?T{7Yi;w|6xkR=_>p5c2i1vmaRI^3_B70+qotD2KV@w<e>CkJT}0PyBQ;brMr}Is
z?hq<yPzEvI9pmR=WoAVN@z`v?%)kox9Tl0X%)*toG@%d_v^?=$AYOOKKD-<|GD0g}
z$yiUQn>V}a#lgbL`uogcwRcM@ltOa2r`RQ#(9<)l1Lvc$L5D4nvIJD)PF9ndF{CkE
zUs7{9tjH4CH#DBBcmSsa`%13s;(q=hs%UK4wKQI`D;gc4H9b*ab+5nW_?|{^>e7<*
zKdbXE3R!@L>%T?DkORC3pv*v$8@|De9h)X@@hU6ZHNXp91J!LUld4ye(X#oBscAQ`
zS#v#^+$h84Oa<OPhS(_KgIb=BMH#3L>wWtr@1rqd{t9HyQcrfs!h`o!jz#&WHGD05
z`c8(4V=XeQR>#H42ioUI?+~jL_3WE<zTp?q?2bar{y{aukK`2`oeMMFbfEzF?Vmm0
z4{4PCnWXPeiO^Prlb<U&t0bauU;g?xs)J3YCy)R!Oo?fORYup-y^!gcmQ<||6l4#1
zeap2Z2bI6EUKMMTQpO<Ik)RE8;9|fBrIRJsc-7y9_rokyLc41&+jTtMVGa)%f-}d@
z)+Uvb^%Nz|_yCKj6aL}trH;C99eYL^S83^G?qso?97we^jSjuA9{qAf!&Ey~bXo;{
zNjbS~wRve#Pf1M-&{YOe=GS^ps$bdJI>x@^&|h>I7>hyodaC=u#s)fHvOE8NAPf0m
zwSASpR<m=|8_RT6JdcvG$}~X*abf+?*V*`YGYL<8G`gK7qm!@>`jrngqg-qU5nyzB
zn5>mD2A-SAFoRkbu!AiL(p}RGe$rHr1hsQu8^U_<r&RUi-s;3!700$n*#Io~kEe{*
zKgn^L1hKD!Qk2KUERj<e>lIYN0MScE_voydmhU^XjyJ~?2FZi`%|7*#nJc=~!(wwW
zAW@g7_Ut5+<q@0d+z+L%J16Mf2z}k%ksN17b0hn=B)bx1pzV1zZ80-4o(o;0`)_Dm
z*HbQvIVL9MmHwGw7U+USVZtq?M4X0DtsiZVm<Mx1CAfC<<Yk%w;llLc_7FvXtpaOo
zWt9_1iGf6|XD~V!g?cbcM$rZs+E4?Z#~H2I1UFvI#=IKvJlq+rlmmo~Fkcd=K7eE_
z%`Mxr?DX5%*%mT$XFU&+m9vlcC!WL1@F{w~_7lNC5ek;e&HI&wA0v&zJP%X~)OSx#
zI-hnQNz;aL=I%Q0|BAsqRUmn(KI*y%ZEHVdVN0E+v@c%s=A7aiT94z7nh#a@z0d%w
zs^&-6Nx;md7)!QIPqyDoxF0&2LC58^JYtD&?CU!lJGH0}{-e0#+!}3GrGd32f$TXx
zv9f%v-MT3Yzfg@NZ)O17S;$}<LFE1__;5#j6K+vYClVNn=83GPNO(I=<R<`lG613E
zOeZz)5@!y%f1GsfxRQ4kRL^x5ydSK-01<@m?OUxAKOwZ=@VLoyex|ekQ=Xpi77d8-
zU%oJ_9Y+bl6UGH9Z#>>Jl75x>fm32OJd+w5T~1UH8Lzu_=p33rTa@(0C>~>FWjdp!
zLaQ~j`PW!%Fuz^#$zr@`=-$$Su1z(u|L1%*j!;Vdt-~u5u!)DMUPs=-1#fafPe90-
z`$7Z~aVc};<2MHV;mwo8PmIIRFM9G;S4IZ<74#1(yqrV_0w%8P=jinSs-d~euT(lB
zI|NJR*jUWEqV(yBNoHr4+elHu**v~rN2dJp;5(2TZLt~ESPG5=ETq|%*pJzf8#<m9
z#jl2p;Z{e5<P>CuyWPVOpeSa#jWe|=9aDR3l^Ey&4#xm|g|GY0Ni}8Yh1PFUVz3bg
zRF#2E4g|C0+Hk8(EHo!pPKu<xCfm_@U?9S_OfoVNTF$7HmZaCS%vKHR9n>;iSxRFu
zGv8@bQ@nW>Ni1i0n%Ltmr=}KFe~AvP&5x*zl+<nFm!Jx;s8AbXpFmLAOo@DD_gZzS
z><UY$C*J&Q)ER7<C^9|2xh}uvdPz9ruejk+>nw#ME9<_{YDm2)`5zrb_ost|<Ksul
z9H_aoBj9)V$&On!f$bS215pGzu$$DhzYjJm%JHbNbDZ7IRzEU+6efKP`D|LdSdPAB
zwIp}ZEe=!Qw@d9SvN{%*rMMWjpGCFlR*Z~&_nH0h63to_1m^?x^$~PI3^Usj`tj{T
zN@g&ug8fUUwhQCe**MoPh6ZR$b;Try=Uw4R1q10R8D^H0t<!FAH|F=;&%4FgL0Kk`
z>6wA3Fr1nQgWT5@Dg#9eW_#`Z&|t}xGkO|>8mRjq_Ly4Z+XGR`Z$V#J6{sH>VFL$o
zo-+UwKf--=A(&rAGg``H*y9@u@UP7H;r38b|L=&XL9#@BoeoMVC3N{97<Ljd{y0#b
z=NUgQ776OGnl!Ahz2q(=09)1$HjP2yiu*Y966{&LnfZf`!t2vNk>`~FVQE#uR?Agu
zHXv8uzI9*AFaIj+v74F@G8OURP;`}f0=~0oi&b~P4l)Wb7t6UH4u<yUHnxRaT#Ts|
z5xx6(u2Ri^#m5o&h34XtM*DsUem)sqkiOuQaJGpJnA6*S$b5YChRQAI_s8%`W)IVa
ziJo>-Lzjt3N#|G2Vr4y7Fn;W_7ouw|j-YwyB{7jgE02dqwmungnLzFSQv3D8aD?j1
zuI!&Chf=NRYUAWLK{N*EG$=gE-K@;ZXt2{Yd1YU_LR-hdxo6(nXBOwPqV+17P)r<G
z$M6R`JN74WacVir&|is$4S;tJX=MV2&!M>F)n3J&G~#GZ-!(f>Zq^mzPb0e8e>$w1
zqwZZ{#b#(Tf#;X&FOYiSy*%*~)7+T{7^aWitTV6Vg_(#Dpv3pMqN>pL68qT8YtTBc
zJ5zCQe^~AzGe_%>wl7Aa2Kxstt=)asODPbzpPz(%`FRN~KLK|DZrNxd4hbmJ&=#?s
z`pVuci|dB&{=ioDh>1q7>W&O{ZyAN1Z_RL-5zurfA!|mmoNa7m(-FHJS7PL36t7Gy
z_CK1_3(Ml96MA?EWSC`0?_b`yd4-j3YMRB$e}k8Y$LnYD9qLj<pDG?cK386{&A08X
z`;@}s(9#Y9K`1LL3o9%s`8hIjjQpw@?+xV1^lIG0z$w|HtJydIr1k7Y0&i?JsW3+G
zox91M>>bP871UhS{u;{-ExH)QaVl;Vg#^6cZS3IHtAbazkSQ@SO+hv+#aQ^q<M!fF
z`pRw5l>B@Bo3(D?zX=dB>GCV%aR55bK=Z4+q^RvJqrqf4;k|CcMHh=sw|A%4{|bxg
zKJQ-|B_-B1ZpQ@B*<fYjO|&59D<czW5Wj|RMW-U9+otNWTTW5p^A(;<p9nPbXL}cN
zd0Rlfm)osR5^NO3Q8T_k*rV$pYc7>EX$0TuCHGPo+gP*_y~EyZ46>{k8--SQd&DNe
zChONzIlO4{&8-@9Wl{+M1cm5=hVI5U6$iN#w!Q@K0EKo$c*`L9)8B8VuCT&ph;Php
z$)D3mAM0{*alZxwSbQF=HR`Z_14?)zBdf{xJUx9xTW4;5^@nrni{4EAajZ~YupJXa
z@Wx;9u^o+aS|4Q~IXYfeN+DL(SU~}<BndCDl}Up3#QeOed#VG$VJ6o6R-5MuHAzc)
zT)9j;Is*#u^=*iQBjAQ>&af5OKD8Wi7baMBewBN68FFYtC5Grir~(h+2ntw1(zOc@
z-=T;z^Mfvtr~Bi^?F8=oE!^b!RO=NHv5HoTJ=3qD`C6zr#J6t!%;ZkCfWUgadyG(f
zdzu-j7q@zwTT4|te%hZ8&&9g>3y_^CF!S1?QKH?T^zZ#dAS1I{@twpE;Y3v4SIFA4
z+o3rfS5Gc0!%Yw}@9xsRxzGOBCzoec*Y2u>nPG&{;)p@31qFrH+T#UPRj;AD1O7C5
zn2r!#$ExZLf6V8C*CZoI(irGVo=T%k7B4Rwx(L<$O5)mU3{l}rm0co7T}x9jcs$eJ
z1TxdH^nYn2lRlZY@zFT<8aP0_&{)k>DU@)WH4^lmdyn%?5fkf%x|A^bz=>^{yY4EP
z$Ky5>{L5CslrHe#X|2wYmy=@NS@JO*o!wBl2xnnW=urMgy552ZA_W6qs3U1Kv=XVr
zd|W+4Uzi4cx4gUFdwhF?y#o27@wQUpU3Y6K8<VunI$ww2X{E1C$#JKLzZ87V++WxZ
zrT-$;6iV=Tp*mVGMas%JlH6qe-9?y3&GFYSjm<v?4|)@08t4_)ycOSg;_y^T(JTup
zY|9>LlP<L+$$WuN@sbMv33A&IN7Wn$CwF2vM%+A5^qQf{`8QNf)O2dG&p<lnBlKAp
z4U{S~%^p7G^HQ~kEA{*EK9U7uu!-BAEWoqOHnm-%V7S5>%k$vo%K>`@%6O^=75ww+
zk*8b?42gnlGFF5d{G`(uR<e4j#VF`Rggo<cl80XUSv!Z*%65#6KQ=aN*KAgSJG%R3
z!flr9uNy{&MW*W}oaADi&#GK_O?r}N=nj`!6kM(JJ*x+YR`Zuzrpt-HE%jYuSMWW+
zhu0ptjYZX<R!5%36m0fsBlZ_i*f|5Jveeedg!oiy+wtRb@4Rgh3?VHPwDqe4obhF-
zp7usxP(y+R{iWNsK{L-|5TR+7%S1}7C(^NVKbrPkrV}Uz3rgRf%Z2p9<1&WQ?)^&C
zLJp5fe1H!|eBeD&3bl|-G$3KX?iiZ=zGy{GnagHrsT1xb=`$azo|DMWZy!k?ygptn
zIDgF`8CnXlA|p&O6{CR$Hp4Rq7+F$OFKQ}1;^-|+rbP$I{~jwsI(L<h>~$M_Xq7`&
z6p=xxd+|bdvrd9Yu#8n>B9cDNXgSABp>!@pcVAxIenAKq-iq5_;y7%Fi@;dm!IKD`
z=F&mzZ$p}_^=<H9K6wfPDbQtwuF9-YmD}={)sUM59&ey)*G}iNbQky(E{vMXonjWB
zAMGGT&PsM?j7o!o2RASXt?o0RJLQ-Jr{>{;n|*D*2y=E-@`CB<?+i*}yYj>%GKs#_
z+A8a@hFtX-m(?<JLuEx@-|i2X>^~}=X{~ke<(D#CeBIP7n@jnmCYIPUu#BpYU|vDG
zO}1aR-iEXZrXK6<D8yNuypiyHJxHN1ALzFBYyLEm!B7ria5CMFv%U0L65Rr24X~8N
z`x0tp^$zmcZ`~U5l0b$h8tDz}5V&lm811k{c<!c6oq8MLr{7>{MVa+yDv2e7x$Sd4
zZ#TKWA|Bt?Ewts>NpMfKo<Qu^&V-&q-K-!4{;ATD#vM>-eDg+UJ38|&^}RfPKHEY4
z&r!_9NrgX*<z@EVHU^VC5}B)evzc0_3XE#Kh7@0+Ta>l*a69GwT*`W)r(rhMn<bjc
zGwXn~ddB^*2bL|QxC;e8huz-94m=i=Wi_~r8~fS3l6!$Y=(>1zx>vqdaxoG}NC(|<
zj(@HbQDaCadQl$WyX()ryL&J`D{(NiwHfXh1|7pz<6NWp`gfxZl=WOnE>HTYOXE)J
z4D`xnytox3E!;8+l@m7<hj{ly`;Q?GjgbCqF(tBGu}~Uo6{kOKgpTd#g+J-BvyE2B
zT=K=Wl-TR8_>M_F&ob|SnXd#XsDU1>`mV5MC4Wm@f2|@z=WH?8bG4R$*Lf#8w#>IU
zBY!$?1_tR95~kFo&{Cn=59DWig&ALP$j4^=>LSH)*Sj+<yOvpBAa*e=_Smgi_QLeQ
z6aDKyy7B3j<HOwE?0JF81hH==)g>_ob>@hI`tQ~0m1r}PRF@1j;D%yIg)-Bz%8A52
zE*Myzbk^^$i5bewu;jrs7*N3%7vEpB-x`$Q4;`qv%Iixxpe#9a_;i@q6JbPVHIVe?
z6mEf>_3>OSd!Kuk?5|hXF(@hjT}uErN$Rn@cPAMV7S<pLt<s=D>ZLbR@-Yh>d_0Wr
zi0I}VpcjN>Z&3+Zft5=!T3&vBA<HrJ=iu}#3Y?m4X(10)go~y7%gD1&xCD{(=~v*O
z;HUciv}HvJ_q?jnzc}&!g5i;LUVioBZ{ICQap6?lkLIseM=5+uT4GfEVj0`f&hdq=
zmtBmK=NbBmpDLs_;1WcSv&oRw7V2o9<3ZybK2<prq{?w`-(AlSfw}WGao%0M`CahO
z;l;;zpcx;7KSj`|o0w`<A_vxseIADd<vAHnt;^dLkQ9t3JzwKrZGDFg-ydKey}-Cj
zNt`{kF=pis9%v-J*yhpL-v=)H$tHo0o_=hXr!4ga(HF`*terOF*C8eX)rmx~$gEDx
z@edNCa_?5^#a=dxUu?`v#tVlpREjj1sfCT_mjvP2j>hGYUm2wJu<0(RUsdW3yKKBR
z`g0^14CYP;T3L8@wpO({xj(wfd-@tB8f-b8p?Z;KVrqRbtNuK|Kl^B5BS*k=G8$NN
zE|e1bEIt$Is(f3h#?E?J$7lXE@rpy+f-5Tnt8V8JP2aZ-F{{qHMEIn!(&~z%+j6_3
zF<JNDN>+ZjEvrlqXQz@Sz#)4Dqe<7b78X%Hht&rx;CRfHfJV{s*PgBhbyy<172pWV
zjLuZLQlmLCpC{*@_@Va64?jo*&_i(w(Y!Z@f3s7SKxRKIE{vC7`2*eh(aOF3wysgl
z#OvsV1LV(dVj4tn7Rs6HkLTxt88@&Ml^@{|)_S#<DRMgVA^y`%{6V@@``oY~_~`H9
zsoulHqg;DEqI?RsIKgH~t&yoOj~93Oba*#4pWnb-4aQ8g-u!emp!i1!74(`2+Mq&L
z8KP+|ckY}zv?_`m56TKmZsH2sO@&swprSiX$pS>(_d~(=<j<zTJ3*iAuar+SvkS{7
zPCS;Jbv?tR881>P5v6c^N53*#kJ8jMFi>c|@u%@mMa18braul2=_5sn<}YtMbdQxu
z+@kSKiNq#|7K%Mu>W`^%p5=xnYmcpoD<@+w8!oDnrH*vE^+<*HW`$+?_AbsIcFa~R
z&m--o<Wqsn!%mlcZLxzR{BIRc*i(_8wAK&R%7=Dz#4g~$musJ8$BMQ1>3s_p@qh3x
zf0u(>zT?C^Jz-`OlCo9}v2Y$!kx@;>!r#B7gL-E+wO)R!J4{c2o@AGbU8b1<pOkOw
zoql#AfwGeY=TRp5o`J&#;nuv2|9TuB$v316RqF`3lFjbn{q`LDgVjvj5?+r)GpIgQ
zy}DxALV?05p1R5-?FS2&Q>o}Ns%#6K@~30K7-ytjl^DYH%Zh|LgUu@(zAGP<@}nt-
z_0Pd$K>ud-0ZvWiZ>P%N;An*q70ZlHc%!Yo+;ZkP;VJXN@h#hlTB#Lsaux?cVoKt4
zV2jl<r}Y@;qf;{0XixK5KV>FxPTFbkT{P|&r+(!K4ard@kx;G9lBN<3PnVDi%`uX8
zh&*j-=9;{U|M=537C|bSsNy5Px4zd%X*jG5`AjTAhNSZ4?R$-_G4*+JKrcq)L~JQV
zloRD)I|ILlFp0HGz&-eOCY*SM@;P7AscqPBjdq{Bu6(6Dlb=>oBWw!DPO8+t>$o`V
zr(a)yj@;?6OSj{A^e8VxIx4pVYG~zAl0#5nVQLCUrOn&WfQPZ_FOi8nlUR7N2DR@|
zXlN-14$BRILzF)gY<;08-mUU;-oCm)pWUXNMoX+sg~6LOLo0f<w759*3h`O%<;#ah
znCuC+8-!Fq17>Dk0oB^xRkFy%+c@6^#V3toSsp%Ae)`nU#@9D;m6nlK91#%~ZMg7+
z$q99MhmtD)DRrif#%~wTPyE+ht05b{q2ztj=I58N>d86!Ld5vSp)l$B`T1{}zg{`g
zzHGnf{Fv9!@H87ojl1!T`0icl`k;XM3kn<@9H7O*Bqk<mFl-*O=Y5Gnp|pH2M^R3V
zjg8-jlaIatO(EaN^E})Jp=~i`hEbrpcjmRV;H*-y%L&D#H5QKk)@ondh2mO*fviMG
zV%}?#l~y7qu;gM53Ds7Fv+hNAEAI$X@ionyo>in6e&QD<@SYGbG=FSU!+u-^&Kn;)
zx8eSp)|g2iFzX?@T?Lcj4(foS{M@#MP)k?N*dLBhP6o3*uVHBUorX?3B~j@0mwv`;
z>P7gRE#&-|cp+mc6&V!|rLvBfXi8dG{Qo9;7FM+6WA!UVDsV+=sOx%;ix>w|2|*vc
z3TeRqRwV}O@5LL^EJl<(8Iz3X>UgPaXHVtwJSoH0Fsav3wv2k(X5~)6YRKOTntq-i
zuX89XoVr%%x!tBtOBJZBf;Xmo)^UJr=fhkS^h$u3M9alh?o2~N?5Dq_L;O77gO3E8
z2|Lt#mRjeg?HjM|Jxdu%&(IjDYugx_F278&7Bfg%7tc(f5PHFXw5BU(`R5!w*v!Ui
z&-ak0B7d2buXWaoWCn87=5P!Z3^#;|4$e$J9?o*zStS~l1V_xUI6##y4|(Jax5Uzp
zpkp+oX*VPDT!5I?j`)KMPw)OQWk3Y+;r8{xM_seM=b4GU*oMen9-6K}tuumEfTucb
zA&u7Cv3w`&2NvTQBx-*f?<7&+@CIOUMiz53&#HW$zKyaUMeF%3XTJO30(>Qi(6Ss&
z(mbJZ>s$aIvUR;#$pehwCVzjSJ1_}P@o~P7^Ok54;BO2ztv>ww4_O}`Zl6A^Z{*jZ
zO6~!Y0xWH9Uql|s`@VUj+4{y`Z*-KU4HZp~$!l@H-VgSfs8RO0c@K~iX=-9qyv|w~
zuSjfSCW{D4x_{yyp!4mfCJPJnQ2Ci=lsGM&X>pNe5r=bvAVYK~gAC5+kg_NEd7AFK
zY3J|WEi5^_q{4*Fq`I5`Ehz)U?mu1U_`69kcV?WLf4!Ya2OahId%x%uaA5b}um8=(
z|I+{ZFJArMy7hnQV)Xa__gnp6y7hnQ&d~p=_jbz#{mM@_tG6P;!i?Z}>KOfxBcr1;
zfhhuN)!J0pBrJ~}sjH}jp|zi+`6wZbJyRe&ibw+&d=77r1&8H=>k#x11Of3i^h^Lp
z)PoD0p4{WcYg&C$sO;-o2hI_P=$W598JT-rsAFJ|aQ*uA-&?3wYeS$(wcsMiklo``
zHc@2!Vek%Yfb$YN=l#czH#a6K)r8zOO9XLp0xe^`mEVDz$oEI22_6LC4L$T_-yWM8
zm+#-d>xkDhLm)OIf<eENaE$@{s<sbf#&v5hB@h6iv0}mNeiSBOn~~5{*vx!_$09kK
zUWGE^v3buu`07CWoFpu-%(Rna(itU!NDHHrT5V@aEO~Oruj<v$k9TAh6e1H76ZvdL
zUcv9d;0^1e#Rok?ThfRGpF?eYa_*nt<KDNLl>Z(jSc9AjP{;lI#qs>j-^!l=UERer
zfc;B<Zd>*4Kf=cU`D)Xz8J0R!*I}z~BO>m`3&Ld)13;u36B}Fi+fBbgCnu-<J(QRn
zf|ibMu*6dBGtspyT-U9s{2xEW$peyaQwhGL!hTBSus*^?CE}@$xJyc!33?D*w$1SG
z?}I?|xc+5P0VexDFQty@RT%75<Y2DWTIsMsZejOlj=C&9+pz&vS{S?==(v@bv=iEn
zmB_S)P^qQ92p%Z6H`MpqDE$-|SR_d8r!(Us$ZIu_g=^)t*UfKz4ju3Ce!KZp-zYW%
z+$0TIf2rejxUK?JpI^UzJ&>=<5BgKGsHjL&>c$KmSlrv%8I~OjPVwdVuxaDi4AHZ7
z9-r+-Yz8|e85xz=uTuz#h)Q2IUOze8)5O`vT=J?{&V#QF@qWAURDae*@L)MLltn<G
z^5+IKGqbFW%r8>N&^y^{t5!=Q0I%rb1DN5DOhT9_kUih*$LF7T1;>NI+s;q+c-=Pj
zfU^Bs@(1d^NG*fxq@?@A#KbFYVRSkr7GZ;Tt_>jk{9qSn(^u?nSD+6bd+eHk5h&2F
z9(v6U(>y(%4~_ixEe~J?k;~(Gv?kj5#RYt^i<}M8DUl8g4D>vjbb~h&Yh^jkw`jIv
zDx<i7Dt02Mi&wolQ;$|+uXJ9LMNIjg*)q`6x6L}VapGa#xRK?tw|Kf?4|Oi<!w_je
zRbB=JmzSPk!C5!KvUq{UUl_eYyr2e{IjR7V;58`&NyNuMO}^BA+5k~v+0V!d8DHPN
zqs|LvXAc*2=`CZv)MxS&-|y4cjBHF?5^2I|tn1z%K^O<Jru3yYqu;yuM7$0o=uOSd
zhjrLtb)V5zxPe2ye*owz(#s4Rc;z=+!Uo{gZd;mmb_FRZDFZbg&bjljEQSE1EU%+Y
zMB2sKQS>MykOM>vBujjNbUHB`+;%3#VWRe<-%U?<J2@f|h?D?)${e4wLyE^1Uw|6p
z{U*-55bPupfb?<-3zZQA6^_Omqs3BtOQ_+aX$bk{%U|%#3GJo;d{(!T+py`~W>Q0G
zgk9}{<`fT1)T}Q}1`zChYsmV7%mh!_ot>TCy*)M1yt?yUNY3Z#;>c<+C(=9z3zj7l
z&`!7m^ByRl_RZBY@JOGypb~N&NW3~XU+l;5ZhM1)y+8AgfJIPH2V7B|p_FJ~Rvl+w
zr%lA9m^bMFTaB2Q(3|u<+d!m+ik?av85xl<T~{7gj2BE7a9&&nS28<#NlSm+teC%z
zp4h&=kS|F(#*7?UBW{B^Sl4TA^=I)_Zd(PQ13EbI78Vz+fF5S7s85ye@ho=j>1yt2
zf8wPHTpd=LmX!1qAj`1imHtfY-)&)48)f5nDJXK@zP;`PmKIu@jO5YFq@H-8T<hTi
z5+<xykq2wTa)>I|4OvIhZ56NrPEVGgM7ha?@bFZi$qB|Z_t`rv5_*go>pv^4dCx4K
zm7ZP_0}D?%S2N!Qbd>dE6<k(H$vl~%LF#VkcgO0<)zRY2&AKx?R2^6p`fii(4sd8Y
zSW%hi2hJ~_`p&`v`^><|C@UzGLgTlkU1k$w*7Y^{>sPYc{XRJ>X!LtJSPPNh04y+n
zSGJ(5YM_mo8N*|qW7HJzXDJi`8R3Wr?Cd>iu&Gk(X!QJw+TM0R9d6Gyb=vSa%siFp
zYkY?d9ELRmsb3WXFRzBBrRDeWjYmm0j+4&My}Q41Ngy8Ejy*259oM{e?HV_p#DfP9
zs&-l^;lq%=8?%_0q4Q!_(Gxb9;ZtCqJ5*{us=jy;&@2PK$hrOnIXStuSsa*SDA%30
zHZbqb@1~tV_%8t8*jJ$en@ZRnOh{mTpFe*dIhtWu?~CXuIBV>af>}`j1OKq~^5Tpk
z5UGAml%A7QSs{)ul0F^`XBS^(WhKKH2DHrv9OGvaHUvFBwJ>mmusr2+eCj674qLSv
z)Q0P@EF-Kg7_7302O=W+l>BxP^x!Fmz;#w?JKnL)nBNR_YduTpvqA6e9+?YZGz@!#
zabjU%K?|u8Q&L(37e(^WAd!xNA$>yZCQ`c?h!fO!?BU=x!j!2VLcpYg!-isFdCWs@
z-+Ph-W)Erucy66?yLjKLvuQa*LB$*nuVvrQoUgD)z4Z{}9}s|+>Wmd~E7;xLrH`k|
zFLqjx1^*+5kbTsE)!^oRIBJeaQ%QZH{q2UIMoMbxV9Ik`nZDFeQS_1qBIq)LXC8B4
zH4XvC2{G*GTuX8#V+?PEZ9mIYN*sEBi-^Uhzsy!w<Y>a#8<TSTnG}#-9LQFalW%n>
z0QPa`4gU5s&}0fR-x_*mE-n=y`i^$kPN(~pBHIlZ9bH|I$`!6IFL>dr>MyphXE_3A
zcbvDs?}ApH1zk7*h6FfpWJ@q*E+p~cHpLT%B=-7qr>UT8qK`N@ls=Pk4DnBUr#5{c
zV5OxM2UEWP;~lzzJZ+xb`D=v@Km2idt%n|g#iSG}y!&y`5%#ic#l`Aibx^0$DIF0W
zj-Je5e#6&CREB}%Dd<XDxFk&U?SDe7i~mLh5bS?~1b>ji{|XWOkNDxgb+Ml;Fs5JD
zU8})Qnf<gK{9Ynd4qwwm&4<OtT7iiIC`nCabkm5(DxEVEyfy`ZyLrgXt$O=|1aQKu
z!^lKx@9k$`Ij_z%zT;NtU(G7*d8_~gX@D}k>`_=`<N~v-VhndWMZZV=WVM@DLHFil
zHER5FqP2-MF%;Z?Tj5d@9g@M-G&Q4uFeuo6FBcs(7M+k#fF2+)JMC)^qUbHBYCI@-
z%;arb^GHA8{$!8toTG=!IX0HL19lVk)DjI}o-+IU`>*~={ouMWW(xlkDm}$S<9|QZ
zDW4;KB}e}I_n&gqpMM9ktJ#bSv2D}Sh8dPDV4JK|DD~js#yC(MriB0|)&BNJr`($W
z4ak%`tdX^gql&V!ssO>rt@GnaBqt~1;Nxp_MzO*B75E2TB|#HlelJb%VX$*q0D*5)
z3qO=uovQW9p0m^Vy|}mvrg|S+`9#Xd$V-8Hx6xJyc&U1TZ`5atbj$4y9#urf#<Iwa
z0c!?jD8}b>vRBIki?lkJ8$IQ{r?Bw*_cwkyGqa4SN7^abYMD7oiK3rELZX^xmDh&y
z!CIsTV#Xb7i$3|AYjLr$IZO%e;$Ui{|Gv}{e{jBsTH9agCyi)*tN<>QO_d7Oba5#G
zzBFmJ=>sbl*Ds#wZTPT0Q}txQFDO;YD@4e1KMPEWv2y#Z%6za!6TF+nfDrOe4<7Xc
z^rY~s3b_M9tfU)p8}-l?J|13PnJ+1uPTH?jDOP^|@}C>P-4B6*Mx$VhkwPO`dHIN@
z*@y7+N?g0jNjtP{0Bcz(kdKnjrWbg&{Q)&4%Z;&82AucM5;~3m#<X*y+-=KXW3pNc
z*kTn_g2=(3mLapSa1D2+UT)6bX>3Mz^Jh})<=$k5F`MZ+{qb`9VO)-yr0zKWTp;$m
zun6LiUUa8}tD-A>fVinng+nrMb~&Zi!yPLw;3p?5D@(F^?ol%c87A!90>^vGA42*X
zD0I}mjTMBR_Y5O8eDIaN=ZF#@J4_|$lGWkU6T>5gMiv_0=cWF*q^q&k1<V?4XQgNW
zpze62XWsKoWw<~;GAc?lyQUcg1xeR1um(=TY6g9LrlG3~RsUbkXg}WIcP;(X_htZ<
zfH9FJc0-kdm4^>e^2q~pOKg^%my>`NIhkSmdkUjt36v{vyZa-?p_X9oEwSB?Li@IG
zMg;^tpq!Mq>aXfT?g`KW%?2Pv0*EjM1_sNShBt^bfOsu0&rgUCPv$;zbf8d5#pb9X
z023Bmt2LCB(Ou@15#b;hU$s^MRd;L{gCRQ|uD?u28$1pU6p#30R#rLmSZQeKGpyJ=
za5~vtAoZClO=m3F3Re=%2HqE#0M7zs|H?p_?f6|98V(sj*Y&TLdx=*O9_Kk4xg|EE
z+%o2*VE>OJ7~cT`vGb%lpEE0PA7Bmjv*$S~e-p-JWYa*<`7h`3P_fI0{g36&YmVnW
z^SYviru>FE!Kid>_ev1NAY?CowzR|&G0Jm_oo`F<@bVtiT*!ZW@+-pRg{0)yOvQMX
z^7zgLZfpW-8(Hpg;EJsvRaMo%dti!i6eogUs8}siF~Rl2ZK|s6@Jo#Zuu|g?T>y+#
zcJV>jx9@FFpuqYx%svB<B`7l#o7#OQB@`LBW?eqsy}!TDeI@|bN9N?k(X^pi+}xVx
zFSui+o|(A#+f04!VRx|P5-vbQ`sirFY7V_xVD^nS(S{@g7=L{k85s&;cZb07{=UBb
zMr<03^Wz<>v65^;Lc)k|-}KBN8pTXC8t?5l^eYR|hU{)rx-07ge<_Jwl%y5!wP=<o
zj{!`ju{meoO!wlm4_FfmhiZ4?x-&o{pCU?JR{CTQRQjz}PwPO}RsBZDbxe>F5v77B
zltX}#(X@g<BxSC?`IXBUc+gCzy{=f^+zC7ljYwy7OKWaHe{5a>Ye9uLm}6+8teLNq
z9&PBW4JLyYfU23@z$_8d&={KCc?|T;d8XHVjwUk!8UmILrS$4m(sc|B)ADif6Zz%K
zjKJZ~K%#}kSWcrK9?qoWAXu&32%!-xaGUng0>b0U0O<U7cp|I~tPSwKOe^b<cRsM<
z#8a8X;2|!nvaYPbVm_F-LQ)Ah{VW#oJwL`d3k(g_a&U0)w`Iv?Swc74oe0VWYsIgM
z11PqGctQ&}`-4`oOBG<=1q8G~5Lq?lu|x%wh?6DY=f{rEB_-WMgv#CjrWH7jMOj(0
z3_MrTiK(rj;?aQ|4OXC7ZarP+V_N31S-B)kr^E5Se<fnlu72jQG&!Td-re2ZzZDA;
zQwE{Ou9fYxL=qCmZ*O2E24Wh3AI3m9;=jR?`tQa8%mMTJY(^F^H$Q)HIBvg&gf4Q^
ziv7PUiJqiMR>!g=kO!!w^MUny`5Yp4TVr8G+Iz_N=ars)Pca0KmsF69rmk{K=6(hy
zY=B_2;Ab7c6R1M~O6d9@a~v~?iq=h>3A-#S`^NxVcDOmoDq{sS#<P8zo!3A!SAd3k
zKZsQ#A|ugkTKV`!d*Q3bx42bUn%UWFy2L=bIALQ?L=jP947G+(w4)8+F%FtVOOpNc
z>67p{8k-(AH#d8fiJ{}wY7~fOvVL5E1$AUYiYnDv2cz84(Se}vju+B$oAStv<+m^B
z?NuJNDI=s4vwpqmx2mo-T+V`4dml{zUFC0kv@xzV``UhqCBd2ueBDz}7Xn;(BqV%w
zzL&6JH(3=qLVEWu8aB!x(C$0mptee9(|IUQ8@NfWm^08WbX1M1hiXTLD)$CRP)#~f
zV2-W<H<ACkFIEHnec}=lgSB36GJPLte6tdLPqVvR(K1$3UiSVz@zt<#5Sr(3(1Wj6
z4*dgj()XeDQ8i@E?k7#8L4HZe<AtRquSHE>1vPDL?XeQe@POM?Ioic$9wC$hPA1uP
zVA3$4T*1oEkdH|L_$D_gslmCrbD<9y*HREp19-pj7AH4e$Ze>V23jY-w(3;5-$Wpm
z<+e2iLQn*~zH>KkI>{*@K*G$=fVUig!0H0L+LAzdIFpjq;rgi6c-h8tHXY3bfNI+9
z$~*1KwJHj=O$P@K@;{LpxHlG0g6;qo9JTiZxK?x&{0^I7Xm_ELb)AWYrK4B`#5txW
zCIj-Xz+*Oi{zE;b+e}ob02^Egtj<813=w+i?ql-BDa_77vuFfG!eOv3un5S$Az*|I
za*&JNacf`>?VX<m^G_*{m)U~E1AP;x30VM$)rjxhNv0O@TrO&*?g!KuOBcUs$6aC)
zlGPBQO=Sy<tjtgl?^^CHb|dJ;#KbP<X|5i^^Cs<%g6@fU?52Sx+`V^iU7&m!WULZ|
zA#D&h#d4d<nz9?-0&krJ<4M&Y`GSH+;pUmWi@UjjYq}bMwdiYjgTWvJOa+CIYu=J0
zI)IPkvn_eui~|sAmN_Z`K_8dDKi-F?0s!ttn?_Y|yLL`I6tDYSAd}U;Fz64MJvzXH
zsgD|mie!KoGpZW8LXQIIKJZj3Y)9xvsRe=I!X^POu9~VUB9_-$bMX@J8r^D79?2;y
ze*soCBNtg$)uGLy>Bx2LEw+@i37QySWYLabNY7|}qIGt-Q3}2t)wT^@+~e^=;x#XK
zc!6-~AQ5`zZ@>N*0#<T$Ed%<GBgtUaSb`t}4Y?1urjb16J#BX7xu%^_s{`5L0eIvo
zfG0sm!%G``0DP!;dRA%Xy-bFHx7De3%}0L(kgkAKlLmlA86Ee$YJdR{lm4R)q+38l
z?RBs<Z3yl<u#^ZD7`KLG%SE#Ta2*6p9W?~HG}5c4X|@-H-MoMS>i#C623)@~z_p%(
zqK(gCu<OJ!lF!r+R@1>?TLCTwhk#%hICJ*XG3|6gq&@s54o+WRx?Cw3i`t9Bah;;q
z{y4a}gCJB4rV`RLZV3tlWU2YHn&abR?z0s@7Xav;i6$)o-!lePyV;k@$D=LK0BL}>
zunnN^lWqgY@!9-NIU0|ExPMMED+hRmZ19Und;XbZrG?TSkg?z+Gd^Gc2t<_@WMxSM
z4rc$nS_^3~<d42x4H??C!PKi4I-}R@p|d#O(-kHF#5x`;0OYa0ZiB|gR1lzd03xNA
zL3<~4^@!#`^k=tz-UG%EmgPD+I_i18(|Q01Jgt!)fTVcLy0X&M!QY-<G@-pb>?JKD
z;|pLIwLrL?AA(LkxU-=h04!+?pGH&<9d4q#J(5++3LphRfH2U+Celz0O%qWFLM~ar
zf#!#x^BSP`;V}rQ{sM2&FTc;NcR`5p=k@<AMWFJ3k$pIY(x`a_qg|wE&~21?t|(a`
I{_5@j2VVBJ3IG5A

delta 108947
zcmZ6zbyQW|_dSgAK@<d3x(!NFN*a`u?vRk~2I)8`($XT`QqtYs-Q5k+TsnXIqR;mo
z<NZsA-pko%@3q&OYtFgIYD1oBK`!X!f*kOPM~_*xuA-wWs-He1``>-CifSC6`)OZC
zyigR9u*8gCV^h)5$qAsOeXTBKZ{+{_Kh#Ek{+ZeE$~_;1+j7bKor1}i<x1n!GsUNq
zr+wd8&HCcxY1c-k9NsV$H@hsb#PGdM+hD_9g-%ywicb~72DWe~N)Az_NT-FQl!T$e
ztswg2O-3W4=Jr~L{LO1k+tb>RXT(<?p&0h3*9yaXmM^|N;lB!JarAn^|L{c%l;BW3
z*CIGWcGW~rVCVE{oqG}2{@%iHp0s{MdU*Bp%WsMEqL+-Q70xS$S7Q?VYEc*-m2Lt=
zgm$ehL(xr5+)U5-{pxHy;=cdy{#%JvXh~C(al=!W$_wci<xEO~Au4Uow#nhon-_cx
zKKd}qLTJkdOXz*7rC|ITVu6*}x4KXc4r6PwZwVA{xkHY-*9%;(Pirn3F+T6y>Pbd^
z+L<9d*D}x+m^s#6(NELTN|_ehZu+X8Lcb?4-{9LeS9F^PUPCtagQxOC_=`bT>vn&5
z{N2=kT06}<IDU4=iTJ(kqcR&2{W(@GV_a9uNMfrG@pn>Jk?rR!X-|sWM~`^B>x4Z0
zD$a0|WHKgoOWa-2?RU)=HPvv9P+^aFYQGe#@xA2X{gzuCv$(A|lDRu#s)sfs)Hx)N
zd2V*7m8Ze!l^jh+_n`lm^=afF#LPm+`}T(73)h;}@`_1a^a0k>WM#)c5z@cwHwb<W
zZVV=`KSVdX^-A&#Fy>f!6@RD>(-3yIctGqN)X-6JR*VA<MX>kvEu}L2bsoMTVWi_a
z$&f8RArn!5el~!yDxE|Xr+SG#^_AP37iOP3fv(jfE)-bf8{i_7HToHHHDJ7+pYznH
zwQr57MlcEFMdRdd-k0uMo^vLfYP$cD;X1G|3&}IVHM^>tL00=cFD!$nu~a;xAb&pL
zX`*`4$NF+{vt9l{u~iF(oRmGa=|MIo!`Y$9tJ?E*lf0_6K<8aV76KmnYu3b)3c}T(
zez71jtM-hB+c{;h+HX)M&#y+ByOrO-6sWNo>Q0FX0!ePR=3Y=6X-;}#dlgR`JhiuH
zwwJQt=3Tu0O;bWdz+&vS=TX?V3^iq+#*)9azm=RH`2Go#O{Tn627!eD%>bu#SteJK
zrha8<Q$VHK88-0)F-oQK$LqYj{wEi8rNpJe!TPW{GNzIx=yH%(B5sy|(QLbVW&M|}
z*1;Q_q2+F(*vR;<FMFhvlspOU!Sxd0;Ij+l;5)p*Pf!D2_!EAFKsU~V!xC~MRMd%b
zt0(V~agvQt_XxwmHwPp7ZQa)m@d%>d=r+EU=(U?(E-7;3vZeJ;8A!SwcC>fNQZJ;A
zh#_Hsu7e7_F!7B$iRO;guoXYrXNg*yotyK5gQ%sn;GZAJQ9qChbQ!YQXvf0A*}udf
zRg#S;9iL$1&CqYZ`6Z~S-EO$!TfchS5VhhPuzgj7_$c|FqqD=xYF5DS<hvFBi2QeV
zzf~+3hUPO_xAkXw$eU6WABW?pLm--z#71r(yhiB^jeI!6;S)GaXtUgs-!Fu`hk)=L
zK~(U)g32@pY*QLJYj3is&E)p>@X+QVFLXabRbVSLaqEY`{KQur0mJ(R7L(guw9=G5
z_BUcuN)HldHQSmBL!~%kjv-uKRFNxm3kpX1qA;4ASr8BDz4H<obEzHBhsV0>>%{Cw
z!+1>$+bM4tF6S!1eSOKW*Xq#vzp_4y_a;4xs&>xbgK>6+=NmE~?7f3;ep;x&*j><k
zJA*EQUA>$W=+5*TxzuZ|!j`i&@4f3Ez(T~N0Vg?#vvN;Jjq8YrCE>ubx3#U#Y#K?y
z=UPoUiApt&@U#t@$*b#gBCvP#20MaD=oIkvEBDdSW9q`L!-LaVbPthrL%Y6YLoNQ=
zGHI*Tg=ZjbcIv*;%*{C@K@o0Cmx!-UsAs$FnMzB1;_Qr=oWX{r!CZ_O;SbWA3>Ja}
zU5?X|7Y6=C&vfQ!Obo3CeC?+%&_2QA=i!T+)?&yBxlY|WF5n*W0ar;h>OEx9?}5*0
zIG)*xs}XInB-urN`Mh&VW=F`9yJ-_pxlm7vV3<FL-KDFg6BzG8xMX>5dmYC2iLP2n
z*l9JL3WY5{?)`iRQMI!<&5!QZYz$>|`o*;f<_Lw^W}!4CX00PZXLN%8gX^tF`?HOK
zNyEP&vbH2Dl(G~05tx~1^{&9?)Ly4*@v1L;d193REmSN?E^FU-9y#&E4S|Y^>Nyt|
zDl;?8#V2x9N%UY4Axd>T*NPs~RJmeRx+Rx}b!?%xPvR8=8=-gxMWzDww0>X96t>68
zw~a2Uon(DI9pOZ6h5^F~R0Jwjh)!k%0iMHJnuY$mSU*CAVFj+P^cVha`Yb7XYXj6%
zDC<KJDGE2&Grh0<Kj^|!x1%EuZKBLwO(Nl7`+A?<NedJAKKc_<+t{8XEKo%J>u%<q
z5aKU+zXZe@N=AM(cvQG}b-pa}zl&>0Rk4>|NJyCqNrTXEdtV`aymRnMmNIW+L&q<E
z*vjf^fyJho6v>Pr|B*Qg%ApJvQeXn_+RwKo+XIOipW*{`iDyp}+$KIwb0CI96Ui>B
zJJ?o?o&L6>*NBe#m}03VXdtFHcvFCFbTn5>8q3xH0e<XD+LX5vcbQJMw@Kh@6tnRY
zdCC&V5S9<Cc=|b}<cHwMMLHR|pe^}69(464oK3`~(D4LTdXv54#PQMfpzqBme7Rr&
zE{L4tg?H0e*Q3B**)4q8kVFxQ7Rj=lg#L*HrK>8d-+i{%WqO>tE-FP%qv|P^nv{He
zxTOg%F4}Oga8|;@@VXag>lv-TA=^@*y`-dpQlIS|Ee_V0OCzLsVklBcf3l~3lx!%I
zHNs*tR`-m0I8ZDt#eMnbieW~m<E5FG&8!>6F=6@d??ZPHY!a8VfDKQFnZlG97CMVL
zk5<<7v<<d`loQJ{x0wHjij1LPf;QFU8up6b=5j_Uk4UwtkvN;(Wr-mjD#q?#gNRM4
zu5@%>S~A(Waxo2s$$|%Ovj;sOPG}q~CcN-ANTOQNBYT2f5^O4>{DjywB)8tJw=&%2
z++?(qDSxfEyIZ>rRigr5ePqVgWCpR{f<ZRpBlNvz`+Zu<bGGma!uE$8rQWfW(DJq)
zPYp{ZJxnS7msBl<;yIp4e7kTh)YgTx_tf_IO|<A@)YL{9?q05k9o5u*qOFnkhK9_r
zY}&+=Ss&}26jC;+oVKyrKq(FsnC(NJ+It>lv)&2@PVkq1!YZt^j{?8b3gO|%YzL#4
zshz!@DXRk?zh7NZ_g>cfq{-`oeSnwOvfW%`)+Kpi2>&a{EuxU<_r+9Ym#5Fv3nKy|
z=vX;|I;Af{LVtdb5NGy=Y!Y@EuD!EQ8&cNjJ;c01D|z76+2b|x2AdPEclz6JvlR_1
z8Lk5+q&4~3O4+Nk8EQ7nXS5I6ejYj55+Y_0rD(TX<3)4LTvpX3B)_(c;Sny!RbpHk
z7FVgh2#F4D;(~P5MlTcxB6=Gde1&_bQt5iuXUaZuu2OaMl3875AbfYWaMu5s5k4o7
z*xWD>weX5BG{H(dg9uc+@Q7zZ(-r~+4mQ~<D0dgp#QR^R`Q5k^(oGx3de4I-|4KA_
zv3PI;e~zuKp~c^#KmXTiL2zex?^=L^Pj~_pqe8s;v-1*@GbA$H%Tol`TY~S4UHlrx
zck5%4mVxMdcxs%OnAYLv!@_)T3>I^&1-&&|J7&HlGhbVqa%_KbDNyBbVRXE(jw3iF
z<74y5TAq$LuOq?Ub3H-MR8Z%u^N)3}k7Q7hWxwdQ)k_o9F)<LWZ`aHVBj|3P>WkKf
zbmxWCLfVr_6O*1ZWUC)ovq{_7Z%K@ppe@gxrTG8Gx(jaip?^{SAiFMBuEfS+qMgO)
z_VXm7BDE^ob=1X9TL>O*W1FUlh_1<aRcniokB*L_53i+}@ccwdPvP1L+6HV*Nx_6i
zig->;&&VoESNEl*&Cy<ICZu3;QB#PI+FnG;nPF)pyyr*ASZe)>kvEJkyPn#1JM}pe
zZN<hrriF<v+zR*YNTycoHT~sko%_j#Z|_sfWOa6Z$ZZnqw&)Qj9c;;9_R}_;IGue5
zhcH;U*d>YAsK}{w9m`%0xC^_B8m$q3gOt?xU$hpQLXlF|yDgeglcUXsA?TWR%2`IY
zE9D#bc*QQ)&xXRf$3}jOnL;;s9&caqF-%b;$8B$wX8_<Ti{;f?oXB7|6DF-GD2qj4
z=SG4YOuw4YKG5^CEOq^Ufp+WqY7i8atHw#_LBT_rLZup%>m2II_nnNKI<w`#KV{W^
z;cZzzWZW!SKhId0Coqd6&><mhaC)(qU3?_BNr=a^v(-CtY9`vls#)eX^fkLq92D~i
zc3v8eu<&|my*&rSj$Sc?mTy^)oey(O@>Jm$F~y|wh3VauRNP+q5{SUE!+ic_n(0Sd
zOPALJ;x?O|4I6mlvzOejsN=PYiEj@DelOAcPDf-sVI9}!6oU~Zw!cVP>6CU<>y@u@
zL_=^np%A~GHNPt=Wf>=IAjh6Oem`q{Qf^%xBj7A*TD^WdxKhK7Gw?dW*<56~+Kfos
zu>6U=J?dEA9J}-9#)&Q@b9Z`tzhuErb0SjCL^7X!HG-ALLSjjz&(4YCxGVgiL57~p
z#cFCD{pAi^nxKPa-2A2LbquK-_Xv+;c`M^743I*P0_i%W0>~MUSgc>TCsE<WM!|+g
zgjg?ZU%`{Gtl~ceE!#?eMv9PR5%2=LOCCVsJYtPLr`v1Rsm!Tr{ns)jiMzASi=VYE
zlu2nK4Soq~to;$<g9uz4uG)_a2yr|5yIY$Y5ydmwIQvuOva+_AqZXR^3LRi_pUxMM
z<866pck9;G)gIrAKv}1##Xnv}%Og_hDb=O<5LEHOrll2X$y6`_ZC<lWgTkK+tk!51
zc2EeUKJ0%4oeyJqNv_}Y>$Z-G9}a6*eG811Mb@jh|GU_M$hWuGbe3~=;x>5~@P&D~
z&`?b;2{IM#4)c%b-!3i1zqNfvICrqUsZQJLP=&+gUI-rHCw%=)+O-oqgdFu4CHBaO
zV%*oVk&kL|RB{>#pSk7@trgb5wsl?#P{(a)hdi+*M4(x}<>BH%#H>M<g7?L8x_M#G
zupvQv6BZL50f`)S(9w_x6ho|5iqy36wH+TGzhZowGRES#cX+hh@f~Tpv$JbqDoviO
zO(ZNn+B+ifm6RP%=6og+b?LSkB{&FQk=XyzUq}j--Y-gBBNJ3=MK2rtC{@Iegct?<
zv`t!`TqAcx%xskPDi>)#>(EvxkGxexaf;iULI|e!i(Hv~y!rX7+6`VbaVPtqQi0ty
z)E9Hb;`z1xeH|^20sJzldJL~Kz6Ae3q%ZX^xvYYMqE~d}PleP}lIhTFh62!)i*pw9
zW9H-eC|@NvRqoD(dltJCnw~hgGf>u^ar>u`Gj7u!yl@Bk5K5-&XrAc8H;pu4XWjJ^
zp3iJ?`g|VaZl?DTSDN%CJ?HnM5j`{FIP2fTAwP*PwGFi8cU?yto4U|5e~nYUx*_H0
z78^9Fbe~MzdVi0sRV0~?wqdFd|5TWl#dMUI$ITOEoB|%CFaC?sRU(Z);W1buVn^6H
zTP))chqO%#5(1I>A4NI+Dr@rrEOhf199R6U&t7-cx|odAxD|E>#*`xN&XlKkhzV)C
zqbsAk^{>{Vc8K{=Gv<P#MiNgkptB(orNW6A^4Tzfb2mYsR&sG9nJ!0@*GFyqzBo~;
zlBv<G{_}o`IDu<bd#l(~akmBNG?zwbFpLPYej1?~7$2?{tz%QfJXvc0lxv)pjQn+3
z3u7C_+7a==%hbfg7cR?x6KBYO2dvU96>PmT^x7hES~dBB<wRvJLdhSNq$085P7hhk
z$%-Z0m-IWyZ}>7MOeTxOf@%bvT^~O1ZmV!P!>Zm@FMsOYmaa!365@CX(NdRsd(H=w
z@hswGDzLL`e-kghiJGiR9m5L&^>rptSMb5ktF!&!8TJF3c*-k@>oQ&6iUL%R_qub=
zS{lZ$AGtCwXSpCpd}Mc)D{p;5OU%Rgri=`<tu51rkNE1theZNz2zhsT+Ui+x1JmtY
zYb*wnNvR|J_hDQECa5w{wp}(2Z%goYiAs&CeQV3e(wF6Y%@f195m~Y}p&71{-ZW!P
z3G*2l)1wa1xv#}A3@W1+EcfqC)Y~D2^L`X+pL=EZTN-{{`Tub$DP4A;W!Z{!6C3y<
z#InPM@Ae92v;HBVR(N9*4KG%8;J0n>TSs%^KU(JN5|9l)OQR2z0%!!gE?f6gTm%;L
z(cBS6x~b#Fc?YhHu{ueeiH|6UMVZ5{2s;T;nqtMS)ExMa)F+NP-e2M_R=aYSUyX-`
zhkAIG@&wuanJ-TVdD-<yOj;+^LVI;vsG3o8dCFDG?~)Q1XT|2Yqd#u?gmK#a*4uYL
zD;=tZ0JL|nOb}2kD6wwH&>-!*+*@xuPy`*iXa+Z~J$9|I#wFi*L*Y%`^ttmC-XY8u
zz>YXj;UX&f^G_R$^)lL?R4mJX#klRLITbm$sz${->EYrWrGf|jfhMI$*Wbt~tU`+!
za%*5-`PGhD;4W6VT<UdBq<KvZ`Im)+_Sb2qNBmISNy`dAcsO(O$k>{#i<s118yCD<
zid=Ryw~JlY@?2Eu{Zdtd?c-&v75zbzGqVfnIk)+U0bv9C66jFvLTbtnaASasyT9MW
zg}SRIFJ(`mS;rMOk*QWuO3)Y*eN*9#L49n+9hmS!@=`))!{<hY-5Ii3o1p*Mo2cMj
zolhcNfASi^@NNHd=GQed2cVgK0@eVCmm@;!+XvPgr7+s41D5mi$`V2ttJ~a!&Ah<O
zce?lumR)4$`Z|0bL|r9%N8H<2(_I>=@GJi<<Uv0CPr*{W+c9<8oSn)))jv>O)^K)S
zt33cc0ldgTQQ1D$r~3vS!&SLIbIsx}zo#cF6CM`HPu3ezMRz{c4+_LMSne3gK)~Z(
zqsuWP7AH2fFm_q5`g-b-RYJs>J>tDpU+IkTYdzcaeHwq7(r}7y{bi75`4Ogt25L)Z
zF<0#;4WGK?kI`#|R%R`_dsl&>(ZrDDg`X|u4Yj<wxT5Lyu;0xq%@^RxTnG4s#w#7V
zSCIQvI!C3}a|ei}EsYw_r&`Kfyd_dKe=*-nAZ;OqlPl~h2D{mdIA^u7yk?*MbcXFA
zRGdf|{zVoWvJbM0J{hGrI!5CtHs-9Op$G{lahQ)Pnb_LCK4Huw)${{_I*e=9+%ZuS
zL+xI`phVQ%)D<)mrDfk^EPm#tacZt^iVCZ=(PFG#KYq?fiI|V0;b`6Ea2nry+wnGR
zDYd}<IK8)np<5$0`}BFti94!Lw*K`YM_Q2h_W&&jz)z#2z>S+j(#+$>U97>;8a$kV
zw@*Fhe@ckF)<vxUUCV99*sJP5`I<X~II;8<wb89xJN#-K<ey*FCXpgJO@4P-MxC-J
zw3)}7ufHMNZWG!j{G7D95WSjL_-UHbXk^uiX*g`M=px&u9Tk4>ug3qY(=jZN`DkGO
zPvXX|F+Tvjiw``#<F2mlUefQr013|N<Lh6opD2%dgo9`ayMKO^^h4+Urq;aXdc5_T
zNMpAiGP7fT^Jb~toXy(aRM1|P_7WFOdweEOoulI0iZ3c?1VbVSBwagR-*@QK?iP_e
zP-A{%u1x}{(1l|CI*d5o`7vk}p@iTqVT5X`d_0|Zk6@}{$6|c;>yCaJ>z+^^C?qjb
zJUjuDvDALlIKeqrL|eX`ygU!d2G)fiZ=>**#|@EJwZtdyI`Jj*>$rN{Bi;V1g#VG=
z_yKNh#efWlSjqwRzx6VxX)Rikz4pLOp`itzkNL0?gM_{PCYJ`mhzHV<idNpeQOjOT
z1e^&+g&VoQQnoz$UCq9w=FpS812zOsZ}%D;O#6y0&NH@WkC;IQBm+5~Jc#uzqahs8
z5?)&5s#ad!Mn|a@=;^+xXE^sDHpO`co(>w6ADnX|GcmaHmp?xVB)Dnq*36gz;P-BR
z;IVQd^qzCvNY5g;SlJS6`r;vnrfSKanpw|gs?g=pnC0z{K3I~RZ2c@=SQFwS_A12S
zbaT6YMZd>^5{{=ZKI3`aJMJMWTg120`PaxnE+>Z8Rjj7meDNxxt>G_nKu)L(GFuF(
ztrJL^&sgnmNNgYLs$-xu6gvTk_z3X1@NoFA5X5|(V&Ct&6n6w17>C<u%Ju8h@F#>)
z!m3O^$4=O*Mlt}sRkF+d1e3WthW_-+hzTnug<Gq9zFXIPtou?&-n?%B`YTE<%u>+$
z`r%uk#STLw*`Ie||1+_htP{k+#OHVagx)9VoI1#yBNG@zqTyEoRrvUZBt0XnS$q)*
z=n*aDd>zS;?ZO)Od>=mgEtAldE|;}&GBNylmpMk<F!;$rwWn{$^DxYmLFP9tZF632
zbMZ!C4=m8c`5JcTQ=0NpMsbz~3X1z^a`tSFHpIZ&e)CcP`7f_={P&{XfFT{;JBZvo
zh0eNJHN?{1ym_OuF>J|P5{CX&T-5hoLrYg>Q|0$ub%YOMzV!BVd<>DG?||3UEb21P
z=uWq_EHoD9ixQ*x@)nA6`Y=qIJ=Utr?XbO?-Melk#q1w`Iv`D^LYa&<{zcKMZD#N#
zErRDaQ(k)RBj*bt<alhHA5e{@g_9Ag&AkCOs=8+?sTn?Exx#8H2Tmw3@x*o6&!@aJ
z%dFHj{iX^y{z~E|V%(GX#wt^sI%YaSV(mO%{_<%Q|Jq&0Z?FCNljc9Uo?bOhV!C_~
z7p-uRl%Dr6$aNXWI{n~`=7fVNuuPCs!cKUZf*p_lC>a~-vt#oKj2AC7H7R{SLnso4
zsY}(-&EMZ3c<S_QuSZwYH9=%>cKu^?erT!2q%z|Anx3`QrPvj^<%^A(L5L{DP}^KO
zmfmv%C{nyT7axl6WSU^we;q4#3ZTx;oOT826OWPJrHyHAj^MqrMLx70`Vfq<c$MV_
z#qpH}qtWoLSC+32mqJMB_}O|u;3_P#CS%-w41fb(^yFq74U-WAoW756tSywLgfr6@
zUHAT!;r+hkj@;{6);Q!2CK$i?1i;D@mu0;3O8gDNn9ZLbWPNL{*bpRLIU`@MbnMQ5
ztaT(1J&;A%AXL*@Uhx9echVTL{?$IeGSH{VQ36x0nLb$>#c)pod{fmZyhV5@8}Khx
zE<MEpv4=#;D<Eb6004bs#xnu10RJz3qqU5nHtOV)zB0Pyw?MX7j>VU@y|zK<m%2f*
zL&E6!{g}Em#(vhP$a2djkMt60@7~_l4@&TIp!c`7g0htV+|$`{dECCYV30@h@wcrf
z8y}nJlUr7VlC3<K@H6d{49j<5gW1a*=-8xl(na&f?Yb~+dn185Q{m#cydbw%r;Cb;
zs#3Y2VV|q^*u!Nj6m-##e1tV*vTskvV`qBEz^c}pb{Q%hFP_LJ?lk;h7$2&Fy^A*i
zWC;YQz+98P^uJ9wKV@Vek-^+Z=!%}<fKG>mu&cA5cYNDzY9(sY*v9hAFI=~|8L`{)
zQ!Oo>))PLUZ-n8qDmYmSBshT--rLG!+bI=5=TMVkbfC$H<neSl{l^9Cb>4NG*4Bd9
znOD3ZGU-UxSU~l9SU8Tz_>HvoI;5hUL4=XIxIBmS+-jZZuqdpXIG?VEQmc6+MSVGg
z+v@7~UV4G)tj=RebI0vjw<C8qOrWn^ffZ>PnFyrP_aB^PL>`c^2X6PU)Hyn^H?9Xq
zRtw+d?<t9Y40UkG$S?LfXNbki3+Xe#B~PwM&+D->YEd9jLU>n=){no_O%z?wT%nt1
zd7F^-<i2OVK!-wSdi2n5-uO3N#o4qbPI7w!<O4@_KweV+!|sOos;%}M3Bf0lS5jC<
zrE(2mcTZS#{X3vq8dCJVR^D#zOEu=I;yGcN%wi<E*c2`CI@iZg!nzu-6W}*O0Gw<}
zdqd$VMC@+1$7!GU(Kp$T)iWcpAq~*>QpsZ*!}cTUQ+(%iUP$LX@l0VyBP~VacOX@^
z`~O0+NW0bWR|dsUW;bm0rm7}FT)wLIj1ZxoKJFqIG23J)V9!nDx-HkY+B&YjZdbtG
zqz6$KEaU9@fCBflr+K0!4g;#2`<hVK=%(xZm`DT=Z4{JrK^nHzAy1sEB#t%sPEe5{
zJS6|7U(!lT>f@;r*1=)p23Oxq-5657o)n%=-&--@uK1#=64G6$$Y8TNa2lmH#bkby
zerbUgmKfQbAK+iMxpB~ysKUuguN>HWA(gJ(Uigcmz0i0ya@8L&(eU`|{V!Mgc6<<D
z%9rFzHZ&58m91hem{CP<2K|iAOA8yFTW_&E88&ZWik>SAO0?RkC!n~-@*OVmzgBw-
z5VmzUM5kuJT0hPnS80WtcU()t9t_Z6#A)P!?f`pH*sIBNArUx&9f`_rZKIyz*U2MU
z%(sLtY|KUjl8l`6r%p@%gjD^}(=lnx5puCIorv35@HC5=j1qULe*yaPiHWP3#};ua
zy+5Nr0iN<1jT$Au9)u#17Xay<=ChKu%(RWVJh6+xpn^iFvCTz#1^OGJX15|kBBGUU
zi0;ZlX$Vo#U3-1t%kJWe0nQd;`SS(`sBQ5u$75&D%%tl$&Z59wjI%CRyTU?;MVNkF
zj`}{mOF8{=Px@xuBsU`AI1aqs9AqP-A@~0_Fnn5OZ8TPeZ0S~X_$9FURvM)H&SK@|
z#nGR9P)w`he46^PhCdMWY2&#;qxq7HCun6Kb{Z2Ym51hZ-JAwFmTzg0pWYM{IM^HR
zc45*c^7+9v3$1G`DN9FE!g8gksS%;vF&gK<CaF|9DplgSB3)=MuJpbd@;LI^DPIjX
zLjIiPAN5o}JK`DDpyw?R+81}1%^AazzDjroePM$7UoCh^p9n)rl7fA~?phuyAq9<e
zQZZB)A*+<e$ilQ(iqFTmve~(<UhRu8zeDDQ9EL^WX!#ugiJj`elzHDsqsh%eISj>W
z(g?BEkPOOUw*?iD@V?JC_RiEf`5m3bp0;I<62z2mik{x<A8^^QH99(?StTL02gE?l
z@qrkS)6uKP=$DQL<H@|^UodX>+_ptQssmNy4@n5mO)6uSBKyx%#+Vx_Me|WcH~#R1
zfGn)eYb*FSi#U?~Lk~+&L-BAQd3XslqpS`xv$ReO`Th9y0R{EYR0B3tT8Eo2c^j(C
zclPDmgcSfPoq(omn?rJ;EJxlgRa3%Ivgf}3W0v4<_tv=cphXz+35|*K@U+#~j-Pq+
zmNEZ_6N%6<!TElhNVjZ>>NA~YQUBnSf`WxPQ-!%xg-;f5qc>kiO|!jAwNyx+>e;(;
z1qzfRRq$QgX8s??BqC^GPa!jsluP9%P@EFS@i8juqY}HOqD+>;1Xrc@-ip$y?MIbq
zT^$Qa=+h^Xwzf9T+{ZTza#^DjT-CaJ+n<uTITZKj`Z~KZt?2)z(s%!*(uXwDDQ*tM
z^120|P2`f*L2}z5#M3r2k;P9_tm+uhD!VyhRc44I<Dq+MuKgl?O=%=+{M#4Zxg<ME
zV)Vslr9UgZ&`e6bw@0E-*1M|0q2{TvtfqgFg7|-tqCP{2_yp=E&1mDX>kB8aji}jH
z_``abWk;i!|LI6~G3P+gXP(NQj*k4x#y(*jZxtq{JSv66jeomvohqWDKfsOI@Sa4^
z=*{M%@;Bu)myJ%hIM5ar0{w16gkr+oeg(>ne}?Y;#cMVtJDhkP1H**xWIy*|IveQ9
z)>#Npr{#4OIS*b+K*`L5i)?S+6t)##b|1x5z~v158|&xS6nkxk`;TdfxI4NemNOVW
zEqktW$;yN%52po)?FXNh=V5sb3_HxtYnYB-s@;(;%Ku?CK}W~Z)>5br_fc}jqxuT9
z3?iq=EcLl+ga|RA(|fu!mHe}Ar-P0_DGW}bV%fVTk?uiD?*5iGV#-WV?YvWvHGN-s
z?@Ni(l`YBbHi|}qMi2lr$_y6=7rw$qs3;~EID$yWL@Xp8;t(Y1E>$JP(bNa1x~2QO
zfv#4TEAr8+0RBIL_($-JHDr8P{__s}RT{WoDwwe9YTgoggxA*!+pC(LKchmSRI2B<
zFnGp;ES+TGnXV|Kf4&1Q>nj&`k+iYY@s>b1&G3LiDJ@7j`*?r*<fQ;Fw^x>=#*(fD
ztTXmHyHH0K`;SvulId+mhiLHn%3gk?O9bSsB!=Y<)jVoAW2?)wn-0F)qz$qE!Yu@y
zM`WP%Rvxf#1(u7?i{w(fu`)(+7Rn8tmERxJU0-P{4zK&YB-TyAKpCJ}AAA2@Mxl7-
zxPd~nZf;!d#^o6L(c|xx0QeY6-zX~T`<om#RihSw%$h#GcV-#&6-uF_m7lr2b=qKc
zyq{bB82qR^O#M6fA9YCb0+4TPhPg?8(uU8P`-*g>)GUy4>i10A;r<PcfkYbQNG%hK
zRzNrdN-DGMxA-a01V9V~+lJ4v?A`$C$ox-PQtLvrP$e%7RFkOzxbx_n-nZW+=_+a*
zpnER$gzv;gR^{4h{C|RkvwfCBEV{UEJXXbp$#?ZpX%EgvErq5i4<p@ejc=E}kzUg_
z=_#yvJ>N?2bdTufT(r;p67O^@Sa)mgfi5cYVbL&P@1$B<H3T&{5d6P_YHi?JjAmFa
z%<6`EdUk1noCGAj_wDKG!*x)2Wq~0HklKHdwdv?@sYH|eo2(6%(7_;{R@^;wN@^oM
zHNKNZP=wWMk+pXhzI-V}nlvti`?<EaafWpA?*U@~3>dnCK^hp_C%#*w0{wp%`Rn7y
zfq)KU<v|WyFDD;{o3XrRHGi^?HBBYR_DdTsKpTGGi;A{sY2v<Ulr*ofL3cpJQ?(H%
z0xTgQzGqNW+Ve_L@h{tOBUI8dAKCq3U7Tk9vi<3nCsw(a{P6G-u%6Ia=ZFm3e}`b=
zu2ZL3Yt(;F(C5Dslm#aUxbR_6xL!8r*{q#e06-BIgXQCwg#?ua<vEt`2!-Ql3>cZr
zw>O#MULh3NnR<!kX35a2oxX)HU?(;Wir3^;2QP!q=#C<gm>?muz(^m@At9$u_A%O6
zWs)7BBhR-Q4@Q4uZ{8&o#pDyWFx<PpM&t##=>Dl-%iCn?BIjIsfFO*6J&!Rlkz%Cg
zy#lpNpsYVD4G@+l-7Blzx0udO7k>)>TISP#iQ-9VSXwC$!a2t#c3(td_yZ5WXs1eD
zY@rtvR}TthlYga=>Fllehnot3xS~;4mh6;}e(~po6*O`K<ID|Hb6!=)wF4WQ<DZTH
zX1ch4XzPVZ8gSu22q08GEvg$^3uS-ViJ$2v0v&(f*eDFxTfAB-e4lM?0iEXo@4*A$
zp25y5{B2z7MRaY7c3GbLd}BtoWU2{LXJah>B^mktG3hA@z-1#c5q}?4^Zz~Qf#g)u
z;4mfmA0~W1?WoD0wH%ii{`@Uo8VJx4$;FJPk7+5K{g=$HBjaPhMF4h~LaU>ehKaDa
zBS&!Qo;dGtCy)0PUk6)8voBLlo#N%Gxs5(Qx<IOnWlbp9!bO04C+oB+C1~RZan?c;
zo`}u#ViH}upo!dm5;*;$>tucrj2HU@YAuE*{%@I|{%>!gFxA}jIzw@b{~<m3to3xK
z-8EG{!B^bnLqPSHFUgF+sWazb2Fk9cG!%v~PYkHKeb8&9yHV=uoagW0VR@}|1pIF|
zN?;>8(Di)oVO#A1QffkCo_wI<>jnqpdDk<R%E{=7aZBm!r?yrHf8@6ps;=TbTg0jN
zo)g>d&%z+&^1KI!gDJD1FptWt{@pp+5mpV4yOlmk11}8R9ThfOkAPg(O{^IbE85_l
zh47dF*5=ku?ccOC$pBapM4&o)_i~C)%i0F8IowmA&e3q&V6{f9rLYj}Z}>J}pV(MO
z*WIO|t5^l3pY!2_=WJZzO3jAMke)ZGUCc4V*bLGD8&iAsI3YJL<wL`KUO=0y6?b}e
z_WeF1>G->>KmPy_|A;DDOi5FQhGmJD)3zq}_wAAirgjrl7{^6-@h?FnZIBMn%GYKc
zuIA{(IY8f<K9=`5#vY5tFrTScKEIa=aNl3Vau?T~ET2Fm&lJ`OvgH8z4NDJt$;TII
zFWr*?oES-RJa762M!at<_Zllj87$^&TKw}Vl?6)g*24c^8GTYZ)gExuG}Q+H;1he-
zv7m1Q9>{Ha8LdU2Ir)|_`As;rJ(4CURVngjkd2bYvy{*6@w@)<F<rAwg*<}zSm-GI
zcwhQ8ZO3{rDKD+EnIc?L^0R(gQA#8Xr;KToatAZkO}7|XN*5P`6amuA`e!nDf=&L<
zZY7(Yb$oPYu_Lriv=tVO)fd#ywNUL!G}!1ffU|aMnUoSj<m6YZW>92Zg*@1ZJjnJ&
z?*!vAYdUGuZ5?8-a-e%HS_rS|KHl-4e;0qOcX(51vpBMt+am0}@guc}N|g?1T!LwG
z!}dotOt~}HY_FKPTpvh^$YR;&X(%8m-YFKoUC(`T;1!+USG6Ay)QJw?inQ8><Jc=g
zW8ERE;XMqNi#A}YO#0h7)a3l&dP+UR1G%0V4O*BH|KWZEdF<IEa97u6+6EFJ>Jof*
z^oAn`k3Dp6ZVg*CLZF>Xi^<Uj1y*TqjtH)Elf);IYb{^??`-#7be@o1qT;~cC^6`u
zWPvZ^;em`y_NYX3b&x)n<#5^_UJso)Fa(5<K;b_!pC1;<y<uTV|Hu?e;(7vmQK#lo
zC(*0nu@#93tV}@J{FqFj5pz---e1q}MEz&?%3*6T0@!KFEOI}Da~tD9KK{7^#OXa6
z4twLT{R&rB^cEJ>Z|g3%f2J1{c#Ql~`A!!%utBJp`Bt8g4FQnw)ITy(cKe|61pha1
zyT&$c=c7RYc9q!^|7HaB|1tswD~*~+gY3GaP_~dPmO_j_#G+(+^d%+TT5`w)^_~nL
z|0x+9HUYaG#ozN+=;+NhJX_szsI(^cxSCPN5&|*=f9{ypvW(Ar>6%X8NAlF%((m0#
z)}?MF8wQELed3aqbHy?Vd_d@e%Uv@yl-K?SFi+qc3PvPCvK97&7(kPGh{7!lK6UQI
zp)$$2I97mfUX7k?KHi8I9i@1d0FkhQ|B2(M^BER#cw)c&vhxumXm<(<Z5sizJ9UgF
zT&y%}I?J2qeDMIZ<YCbSo+)&KieFA;lx)Zl-k#6{3@nvn7LU*&mUSk!RDG&vzOKHg
zuJ{6|2U5gSO=-m<AK=DDW;C!HaT+{E!*);LB-PANS$g+B+KB1jmPvu_jXIEH(|Kii
zuZu4aP>|Z%_YMZNud0lwp{uhO{q@(Za6j@4jLN@hu>NKxnAG4IF0tyiv+HL(&?@Kw
z38Vtp;J=B&jE40?q)tB{H;F3I#fj~oFSq8AQwrK{{ox4Oz9B|6b!n?PdWtA4EeOd!
zq9G%-_gCs}jL%kq4YvdF6RVm87h7!K3)ActrUpF6y#N_DWw<30WR`s1C7Dy_w2eA@
z!bijnzy;;n)Ki2Rg17lOL!agHLopB-m4g>}7Tn@tE9(X!PwWYF<x#iesq1D1^!*>B
z862k=pUrbT-@6qT{r=);%1Md@n)w#msplQ+68v;ILr2d{74e@{hwhxv4<1ony+94M
z&EO778BLKhHJY#{I-dI8l=q42d(Y16j_#LqyAPh?I@lg}$T5Xm6r0`XUAZ~zcb>X*
zb4!D2GY_3G)w$dt_@jR}0bi9$-~lP&y=)0U!N6LVNJt4OSzoTb=8ff!0?&VQlUu3(
zsW+Pnh3n?(ry^=bic#r;!DB??q+D!v{I3ZpVX#FTP9jmV9To2Gb?KL&Fv-`c`$T({
zO7$PehJ;5nHc2P60{#&JV7$SkW@dIL`{s!u{Imj(@4u!6k#?3Z3N;@B=+>Tps^8OH
z?D%ID3O4OEnFFk*&1R3bMZ$&r`h)xwx)r5{D&X{!8&;ZSe9K4)ING9N9mcW&bJTzW
z=2&p~2SKk5X!8j5wB4XRwf+?81i0%|-c*^}rG{T)uTh2FF$r)#Qo{G)O%Ye|ohAkM
zb`sBZeYY&@<ZVis>HA1uzXu(gdD0#o9S`OU@x4kp3|vTyp??pqWSSh7kkkGf>&O(U
z64UXZ(|IV7QMYa6GY99ib9y>WCJ&a-3ob4~L`1}+qm_B?o9@HILtv1d(i|8VcoLYd
zqY1;oD4&^8Pui$`ezUo>>FLT{$l#x{Fi^le`$@&iOE*=n9d~Wr3fkPu=cqQvp+W)d
zGHjKhRk;HvfQEJfRLn<5(*b}Bsj9Ca0KOY>3>hr`ev75E-;o{NQfa3H>yEZ#44DeQ
zk^4&RHz{*~-Rfxc%^>ju4v!07Yc?kBU9^h$K$lmfYkBN9i8DoDtRga4bFVw1K^J(w
zQrpGr=gT>MQ$ecyQrDe+^Pf_IX%biqR#s1Om33Rw#6|sAOSRED`uh!63>yJaK;;lb
zNU3Cd6lr*1Hut5HM*8Pwt#ZUqh$mXZ6nF5~_!i(p-71$+QEes<W#xyK1<sDz5gp5`
z`};LnIcqN2aQb2p-gPGz@8h4{BS2i1YtLx+u`*kQU@hvfR$$0G&qge;%al(hu(7cj
zFeycdii%R8ip;zwnONxk`Ct!fGz-3Db1W_@>geja-!L;%qvYlt$7V@#b8}-hRgwh<
zy9%LGg?e|j1H)^3`!TOmEuVjiOn$G?0|_3-d0?azp7+HOFc?}7%*`Aj_{tEsP)-gF
zld_Nh5gL97HXAsJ1PmX{)B}!MP8NL4yYHwSjIMZcDHXcRddB*=`;=j^Is~%3nMd_z
zp93FT&_)S}5wNvw%G1<3*j_(pxQRVK%%`*lM&7q}YD`B0x5#;}ihqr*FNURcj7*w$
zkHy&wt|Lu;pWNh_5p8lnj`soSkABVK-Bhz})0)xtzva{QSOo6tsSMV+86>!X2EPGd
zo@0i$ywtVPzo~PcB8-HTzdPvKRMPfTc)-{_3dRIR3cwh|kjdHh<y(h)Jv)T^6V;bS
z12@KiIEv<4wMIFF{m#<4E9$@hnY({XZjWCkL!&r4TVwva40*z-$|s|*E$Dm(Z#t0>
z%syUYdpV`CsaW`^8??vly?Imit6y!Jv5@w1Q3@w0;mBPV6_{G<-S|9z{F;z!Ut2XO
zcD?zPqA6keP_nn18-s%2tb&mp&t|DX`2KZQ-Em9mg0KvN3mXG4aVFGRnzwh%)b$jO
zL3wk426-AG*4wq09ul1(=N0i&<7mee0WfRY^GnGq1{4vm>49VoG+9bMN@9<Y?DgL{
zZHJ&mUg-TugTP*7MyqYe7WrB#o36pKmkh46M+7d{60u6kV*Tue0<Nyqr-BkhYTu@I
z5oM1tCvCy%wKVH_fN^}HT>m3Bq(pru&(!|yF?^XiySvx^q{$<ECOB<IRPK#v0h1;Z
zRi*rcCLi|)!Hec<p1k%~3g}<JD74zAI^{k=;~c1H&t5SwEV4Vkl55|YFH6~wDF_-H
zXicAiX=NcZ;1Y9}u4Q~&4TgMk&o^}uRMBp+<YTK=gB&b|P@&*H%nk{0gY_7%2#rsb
z8jrZ!ygWPsF{=wKEJqQDX167*Y}YsMWJ}1~KuLjHPVC(#y?TCGsk=P@ZoSDG6Z%Sy
zkE5>0<XbbRj}qd!6TjBb!sUkL`Z_@an*-1NUH@A5AG5aAHeiaOht@3WV8`B!SUtON
zX$UFd3W_zwy)MWbjx)HPd+ksIZiv*fJAxLl6#&z?vc2Ekbz<dzO9VH23r6Rh7aCz`
zQ1<Umux(_@qeqAZo_d^e)EpW2C0u2<FgjIE!eJcfxB->Bp$_-Sz$JW4<hO_0E9d0y
z`Zs5F$3nNT9|3)^fms~)%3kQW`=+IRrfsfM@>Cte4``$JuTON03<3<xaqaq+dG-;(
ziMIM(6FMG71k>cdBPN=!zvr=rn7%Y<%`KwLD?KsUdqz(x(r|HVbxiWjE1dE{n-cKa
zgyPx%%uI3ve15Xh*5JiKzY(L+17wIu5_~@HbYA*$x7p+<VV3{`Z8%`CKOSzPs&iF+
z-D+1}oXDg>C((e{$=F13XB{fp$kQ+aTWpVu1iI+A5N42FS3Rz^B1c{Q#(L?wuDoo-
z0YWQOY-i~yKI{%^+$*Ymyg73hMT7tL?k@vH6<*ydQ~tA==0`@*={w*Ewkcu~t8s@w
zt^m7MQSMH}gE))9z*K7AI#$bipl#?^<rc`f8+dEY1G){G^<d95UdISsQ*TIKw&cE}
z_+vOZ&bBnjr++yV-E36q1x!Btxr5g-FMyE~_<L-)R+xa#5S#bx_sd<2!o@KiJTM6W
zSG_>z@Mwn7XTr1&s1Hz)%J+mXkI=Y+QUR5-m&@W{%JIFRIQ9|uxkIk;{{7F}$Lzh+
zM_RDoPDB%hVgLLA6YS;Qy>JGD=Ru%At}bTj?>N{r0ITVF{R=3oz$o<{?EU`|uGId2
zjJJMR=+DO<it%gkK^8N==v+c*^WX(=4{Gb)_Ky}@4{9T*R4bT~AZ&5=iazENcm1qV
z_`p)fmCRoW^|;4BrUY9yKl3RulyM@Wq5hdl=E#+%t4hiiut!fxF_m=DJ_*VC2B5Sy
zOdOI<OR}?V1_J5~K1E=A{P1JvQ}|q<-lA`%8)6)R9Zi{H=d9r~NEvk=)d1(ML=;P~
zwWWLGvG~omd6btNF9_V+1kAQIjJAiflyWDoZvibY>L1WDGKDR`JdLvYyEi}d4)D|e
zU*f2@{Zk{06ib8PonH0}6t}HEl{#p1-%C<FbE&6VD!DM;Ih*Jdn&M|Upbr#fehU_G
zu=&9P^`~y^7FS=o(k$(6#U;qu_S1v;$EzBmo*nVI-_0F8c|FL^vwJ+sYCw`8xcr<_
zD4h{QpVqTW(xN;H(ciIT+~5DMfxD@QNv_g$Eb=Vs4Vb6^*4x&`G?HFbyY`Nj9Q8`z
zOrq-OC#OtZk_3`u3K5`h{WT!wVk*vU+VW@XjODa*SWLw=wwbD`?>id{E1&#8_d5^C
zKq7RrV)mvmNV*vho9E&Xi-DU7CTHF1ZWB|C;D=<hB2*-=E0c8$u1sUyd!J_*J=K?-
zv1+3;#lviG_`&Q}*Su}1SIVdj50%G$NH-Z}OHzjAem!;#{jn)jRr4%5M(v*B?fO^v
zezr7!xQm&VR~w+TjnQ9!dskmeo`98asr4Z?&(rB*bh3h#c6xlfmo}%fPmGfM@EKqo
z#w*LNwpo~-St4q!tCUaJAsad%gr(ANa<^-M;kJ8X+-E&wb?DTk1mv`YSpH?kG9Cht
z3V2cYi^L!)lPrwuorO-$TeWM#pBgOv@l&LB=u!@!Qaf&Ih0iCA)u%4e#$6*rBM3Bf
z;s8YGy{PpR8DE#p4oJ`t8<`!?TuX=-OKWPJjPEGO2Dn<wWaqbSUiBV_kLuYcphFsT
z|H%g>A6G7XNsMTDcd>W>Kz@VK$}D%vQ#myQ3J$DA?R+E)>YTQbNvdt$O6Jh5`dwr>
z;{DZU9lx!tuNGZxiEhu<U?B)#n$D<UH2jugek_^O*(pKBvGg)D@}>!(Zg{^nU^u7U
z^vlVJ<@8o%AgkY~?6%NjdcAx&=eC-29M9Sq!@msSmG1Dfd^Y1jBt!u~WHRify<w=H
zAj!~=h$eckGl9PyH)C}^ep8JgOkv<yy1_qCnQBE&Y=*W_yk%9ovCn3E>0R-b)C2g-
zST2P1(u6S`me~u(gOkmCy=M>hEf6<M5Ge!7XpJV56QxRd&0@-Tyi{{vEG=08#Te-J
zEe_|7U`*kCT2r#NArZt=t!<)+Y_81<2t9cAB)>puSFUeby++-<zMwz=W^K9h_rnA{
zT4+gN?hy=ukaJscNjlf&*(U76ruSB}C9i_bbO*QBJ0-*)M4;5rB}F?8b{*YMGitrq
z#A>oo*Bfl_FS70BpSv~D%~c(o{ynTqG*K4@Y_JC#-0vYU#{32R7)*GfbHw^fOgReb
z8K%KD)es-?LPW1@T~yH$0}aP<fvpIOP%SD^U_Y4@J}|ih(LU#1VH&QV^soCOn|0n^
zcYr^Ad-fBaBry(}>)yB9jAv4*((gtizXP+x;zYVedi-E;hP@C%1ZHMwSbO>p$c4ls
z0CGcm7g;BrBp2c8uex3*jm7Gy>@8sQ_SYI9;$!Sn>zgV?_fmu>>h;J)Y%JQ%^3Lf!
zImVO9i3Qh(GCyA4_Mm)XMm1Iayxm@j30ZGPu0J}^^h_}_9$3Uz4UY1xbn8U6Yn|nW
z^L;I|?0tQJR$3Uur*bAXQH8+c;I0Vz9Ssk|@<_3(N~aKDNo?w~Xl&_u27Kq)Nn*2R
zx4QRe>fQ1;aDykCzg1~zc3o?H@YFAdv+yGa=Fz2R4BRng*>yU5skZZ!>;-P*kv9p_
z%|W`dn*~kmz@ZWvd8%W){BBt3wtddchwY;~8qux6=e=6vWJrIHNk8@BOSK=#9;nvc
z`|E8yjpx68@24as6(Q;G@ra4zd-3C&Z_c@pNzU<+WWFqs<V<QfSC5QLu1$o+8U8OU
zoNImeBl5>P?n)i{Ll}QYc$)MXz1bf<vX1kg@l?~I5|~HH_?VbMLn~)Yn`1l)5eTG{
zHce)5JcTab?$08Jzaa?3-fbQl@}|h;L@clzEdCLJ6Ak7+a*I2B1KxoF%AIQ=>>$GH
z7PU+T>X)|{9yKM!278n{SMO8Gq2P!>;%3Skg~f_$AL*9XS>%_n=OQBEpQPWMmAaj=
z)NjSKY9OfL@>NMkHEZ<e@|)@-s}0}7yIiR@ScjaLt6gghF=+YI28UpUNJs8EvR>fh
z{R><br>n#zb?f9}0X7wmW;|slQjbZP&z;jWl5uuihSFggD{1ko8&ko(b8|B3y7T${
zS}X*3=N;MDeKPfPC*~S?SPr=(`v#bfj1@~u^zR5o&Wgw%AHQUv49CboCx3OkbSz%+
z4BFYN@S(`94vVmg!;GzGa=B!3dvuAzgW8bkXPGjnI%4mC2Lai>VZnb7fUjCZTZAop
zg)J*ON#o<Fs=*bKl%RwckE6;SsZo-a<zIX8#Qs#^yosNTmCa_=Yw<3&y=lffN~2iE
zgN_)`uNbjq7|mt^jA&`-nOGlMw1z;1@>FzP+AubO*IlA-o%kwW8%|v8SUuL~Byqk#
zjMJ#4CNeeF_l(Wp(;}<b68KDd>)FZ7Mp!~9oJa{FN1&7nyYwP0Tj;xgk?b~fJ#}0W
z39o0$mowdPyp@QVKPS>^>1<7Py_ZY4bAVgl)HU_fcXZFn{x+Y&(vamx;odl8_<^%D
z&_kFGpJmO9vuEftM|74etS_~T^b^{b1d5Z~(ccS`-UIWV+u!vJYWWIx!&p3l8^kZN
zPS{jibYdvd4^7kfu*{o@xEgZ)USf;<&|)z3<`u=LcKY4EjYt07w^(&+*?aZ<vIg=e
z2_<+mE-)qUa=U}nc1V{GvB2@<$dtazJ~}Xv#}cY0QJ|*(MyG$h?hqPFAa8sH4kQ$U
zI(}MuYqR_qERj?uao0A8aFUGT4lmQ#rQl-pXC?jolEc|$l@Kqm#XDcC*C+mUk7&wX
z(xb;DYUe!z@}Vxiew^Rp1Ndesda|s>EQIb)Jx*`cV9=r(hKlv(E)C3Giq$Ij<v*vj
z{%@W!VIgn1XRZzw#bAA6{$9@Yv5$RjmgnQ%@#huT;qz*~Vre?-H3#45AZ$0$_i58u
zdTqs<Ta9sZtQg3YY;$KetV&~L<FA*y=q@*3`J=$Jd~rXs*~&6uzo~rTsP?qb60yCh
zY*aL84aIoS&k#!QZ#7~musit>l;a-Bq+y*ly^@w_*o=G4;h@Dxzsp;5GAUMqvsS)v
zR}(Yye>K2IX=hfN>~mjmy^4OY<Fh<iU|XtlR$4*0mtG584vQPfxe-C;!<D4@ZgnUT
zs*+vNvpG?7akG3pVpoIpRDI&$iAgk_l+3(gsgNn3K`Wu;BTWwC)WHwaeb;AtH=n|7
zhKx#TZdwzMs@C756ww#$4%eAo*&r~w;L3&I`y@+<MAv--89;FF>`OiGZrj+Nu$*dC
zZ-F(#;mkCQX1ms9a<z413f&1aSO3`G#vEB%#MHREtR6>3RmXkWKQ^>crY&937L2@p
zgV0^_RT)Q2=#rZwQqI~|`9=(!S#8?mUpXDa6ji1tY|UmL|E!;jdX+s-e$9_pm7@NJ
z;B}ug<zkEQ$@K+Gx2B~;)4}l>(%aSW#FM2}W$OU{Ezx`${q96qBR?8O^xM>1Y8(^1
ze6wn6q3cRvH)P3$)?oMcp`jRX5elvXAqv5p>lgUePdEHwZwSz^*3)bbCR9&oqdNTA
zV#0`O;~Uz{!IsR47oXgTp%%nN9Wmf2ti!oV<fE_gW{#8pAj@Vw+&z5KImpFcVrv(X
z-R-c;`h&xi{>`=D+%K|E3q1*m9L4MF;>ja;f!;K2rgXGUj)xhK-@>xDenLubC7f=R
z9T-M@|1%EoSEK0-Zf}ysO?+tb3tZ)ZB_=f93sgPaM`hj^(3OcOMDu29vcC;f%}vv8
zmo->Dwj4XzJuBOmR&SvWBAzlCs<X6bq(l!5kM?&6|1?pJtjHAv#lBZj;j%g!!0``2
z`6D?$)mQ6%*qUPH4xOK{f?O%p^2fnf<r>}U_}X0CxVOMZQ+q`>U)OZq(UYle%Fj2x
zYkRh@Z$Z`bQ!$VFVmFLvt~iLP7H}82JA)9*mlwI`tJBIHCAMgbG#Qw6Hk?++tVKnI
zx)T-Sc#Mmy&bGmKg31WR>pNEjTP3Pt(EWwkU&?{dk|Wy=+B>rz#_+ak7o`2`KejD`
z#7n>dLbj%dS2d*bT^NxXq&Zd3><m)GU6-)X;5wz98t@A<YWIjm%H1)Z6k785*yk-a
zMSmz73YW=dt*E$sGL!Nu-UTg9a=jY-14V1Egg6-kC1QC4y2{j84swn|_*{!n&daxJ
z6XuZYLal4R)98M}uu<;*<2ui8n1rH1!7nmJr>8#TE1j_z^>i~wK85K05vP@Hjs}#5
z<d*>L+}7Gl$ngJYy2`LBzOM@?ppr_rh|(q9C4zK!gLH>93@Oq`cc*Z<ba!`3cXxNc
zgTMd#fe$>-z4P2TbI#stt-bc1W)VL}l5cT*?KB5x?pc}~L<Kx1TJ4$Tp$Li7a+om^
z>UM;SkFBgS#j!70%v$$!iZ<JJzM)#^be^=&^?s8`r{u>m2G|)IUI=ZEJZN)W5&z68
zzYCTAXZ+0^`sQ7o7!(xLGc?pRYO1JVTY3>Hov}E&<FI{kOOW!j#yjLG@pb<A&d3h+
z;TLKvkGP|N)zv;X?n%UioxprnY}Glpzm#F2<<NHwbfZOUm)3t1*_#(scj9Nc!Y%8N
z)1~5qjB(<C*j<?M)OxBwTbSwE>j@5qIf*TLR?f%V8e=nRI!<C3jjdU3r`|xEW2=#h
zt`oDQ0BbS?{_79EB<^psMt_Z(2t%66`qexdHg~9SVnAosi20;cLS<4UZg9EVAk@Y3
z%xWZB4?|mT+E0a=P+@LrdkKoSyLN$7SvXyHYHk6r_-N;46&3As8Gf|#`?L5&fc)fc
zpI%ZDk(qfrUR98ty&iP$hh9xgb*vl(R)exWTmF>wIA3BXuP*Or7>SuNveULQM(|w;
zVz;p4pI@YJ7AxGl6Hrl?6tg9DM5R}{42Jv?$?7zrtY@*Wx_x59>p$N(u?o1;DOA!%
z!0Q9d9#>LF(;33m7VK4Wr{q*4f>V6Ud&?uNqx*Hno89MGZEx<)D`PJno9ge6{KRwq
zT%KLHj+)Rn4lpn)^NnD}nJKFy1-k;F4t4K>kj_e-Yy_18Y9H$(RF;MV^K1X49#LcE
z|J`+cy7I|S0PpYV%0t-WwmySz<AYy5fjqDbAvZTSCnO}aa=@Q3qC!H9W5y4~#KOv$
zFza9151vynH8s7Vn8U`#X1@sF$lMec7l$b++Gu9{k<+=a`9y$N4gIj&V|8lJ44Weg
z(a}}lI>!}GbL++!$@lO5thL|)a-<waZm_%0%WO+IqF|uU;o=x_u~r<;<ADz&3q;_l
ztf|Hgr_`%>^=^Tn(Z*y7|I$)zX!MIlkGM=73%8<gfXw@hg~JTdr_Z2Gzr5>af(^~u
z@b@2ePA9d#+Ou;@*`p({*+|CJlqxnh_M$0bZN+2f$meflBBF$XzO(C8v$>*juJO@k
z&i+zWB*qGNPp>3=G+cExz(`aFEJ;AU(caECe1RNR*VNQB#49Wf1?GDw$12A>Hs$A|
zhl==q<=v5^+Z9nddIIg)RJ61&qg}mYls2vgFTUGixC>DD<>M6?JqMVn-l2<Nwd)I-
z@7dHkDaz)wB~EBU=B5&Qj_*o8cmE3I1ZN&FZ!?>~g-bqSipKq!qh|rECKAr_!XupR
zm_zIN)er99WwF7ck7X)U8q9;^aT;u|-n{2s;qd1ao94Pv<W_FWv7f5RG2No4*2g*H
z^Y8}03)7b=%Qn=cI?JIDnI<jX#yLs%5cWf&Im<Gx-eolhdV*x8d~k23u@;*G`z>Tv
zGmE2cl}CQ~uzhmz^%CIKRrx!yX8n})0}o=?jS9bP7^?o8{xpV*-jxM_`6<}+V^@l@
zXS$qsz^7?5%KpjU=mb(i7A%E0fY=oG(He;A{y25N|EJ*+{D%$xmX<N`p7E#k&3;;f
zLmx3>9K7Z8l__sBp?%y5Oii*&JkyilEIqNWdZ_fm09Lr2AA%=d{;Viw@!Zn>GE<|E
zYV7zVyk-?nxrD_UHEOavu$UI(?C~?9@nP^=j{IoBmFy&4!dEb<nc7=lVEgn-aEtcs
z5KVK}jTKp5QFlW%n1Wr>q`Gt|+j;ZJ=<a0Nn<rXmuw7{e#6Hlk%7j7F#MZr~8dx%z
z$N4lRf$mh_!WaWfB$L1I7?yr!cYKY&171#fIZg!GA1+&i(Am$Cz<W~2NN_T@i;jD3
zfA{NUpkj^c%SW~eqzWpc5D~5mLd`rnv;jR9{2PPA`uopYS{sO#gnEALeq2<+ZCbJN
zaP`c3rtj{mi<j4bYc<^Nh)1R7{RiqXsDOV7=|_(V3zJiqy@l$@3*}3z^MrGPCc49>
zO;u4Sc?Hoq!r&cWe=0mit&KD1IK{<Exen^e##DZrM@78>42ZnNILd=-dL~#*j6J>C
zjE;oL=!vsz`-fO8JpyLtmWmKUUBfQ=jh>-Q;<-DaIik!N9B$|{fc%mGqzW3OKII*m
z!n4snrE)T-eSwsw&t0pr9I);r!n=T^BJsi{h@6+0O#gzLQ0wk9Ni#>XlaXcik38<d
z{!&zsj{W4}@CO%sb7U-ToBs!xEnPbKwdCkUm1*GXNDcD-eod@q1$zq8KQokU*R$(t
z6E-Yh4m|8rsGv1dcmd*HkA;B`$(s{m#hbH_K;aChF^a1BG7y@cH&+=x*#Ac4w&gv;
z^&=doxq{2y47ZCzZ=o|gbVhRVL*a)Igq*Lw5Qt@{{(t#h@!=X*8x#_1zFKOXVGt<a
zdsX@(ep%#@)w8`?qP0xfIs5zW`F`%_CHXvUIb5|&0sl2td^!y6c-WT1oAAgad;{gM
zgp=R<I_I=7Xfmm4M{dEB<jv_u=Om9WbG~v~JREu=-v$kzRVL4~1taN6*g9V5Z+{u5
zCUU#IhTR-;C$1Ecc%s@fnZ4_!Hr6PtjCH6vB0I!oC-BV(Nr3VQh?x1PaZbi~wx<a|
zCL?ewfrK8`qohj6sBBkQEt-6M%CD?)ZYgyqNEWTNwQsj{*Tgv5rBQ3p$1}D`f)(HK
zh{C*sU);wQyDIEri96?2S^M{np-c)_!Pp~(KC6e#YK3}sWVlD+#Wm6VskgC_4Y%cW
zjnB|sSW3+MIkV@Jef4*tz28)Vx#ep_Mg+ojqpHCYnFHP(t9GaxI)QzNR3UX5=Tz+#
zBSLtJq?jJ|p3(USkZ4C;Mmd-f?D5xQc$ktJq`Ak%W;99-ohYC5{2ex(&L*NVb}U`#
zhsE5=Q)&D?+Ee<2dHQJ%$)p0ifyW-N$}ubknf-JeXPlB|eB9+c4y5J%MTAKgfXknT
z<)`jsr-9k3;-c#t4CJ3oiqx!c3zt@oPp261`~J*Jt3-n>i)6p}Z`r{z+x$LqX81uf
zD&CT3k>$R#;ig?htn%>}T3TrsgNeD{c>z0l%irizrS5UBC#SiFComB_@Cf*CzudGx
z`H*%Dorfw>HvX!wjf{%#Z6aZx1jNGrJBmL4@kFjN(+y3#@-dUYF0_kMt_h8e>IQ;w
zpLJkMEFJ@JQGSYd^D}LLT_v9JZOI=eJ;zcWf>X2JGan`z*ee)JKb);uTsM$>f=@uk
zf8asm;ak`zw1%H#b8I-7z=!*y9t8sW80)I#^;qi>tSv3pNfUr20$f#Kcw21xRHd|{
zViRNsSuXX&F^6l{D+kG3bvVXLu`GLM4wa|KL?@)NY|<9myv^DJt6GiQNHDV~E*MAz
zqG82%eyk*po2#XjxX8DKU3(HpoQ->;%1slI*FCHpu0Kbzw=leZAMtzcav1}6CS!k)
z7(DI~V<*{;0ec=v&SKgIqZ*N%>U`wM>=umQ=&8P-fM&486P}ZDF=1fsmH$KUyY@H#
zs<4sviF+98<pP0`Khx|0MqOs4`mo%y2EHm1LHs<OO_*q=+42(%-EE`n0j`xd(XrWl
zR5WuvhyAK<j*j7!Nd?c-r3{PNZ0gtYJbSx2T4}&4$=_T+BiTb;*KkYR){{%^b$5?+
z2wclhYl3L5yGXNMxC(c-hBqicqzx?fC*uv59US@ZKOMlM(m#--wl3J%pa~i1Nr5T0
z!1zioIqr>sKo+*7rxR)GAh#kO@HB}9!_s`lsc0sJl?5H_8;V)B!r-XgAOQKRUP})F
zpn}QkMD4Sh>AtT?Fk;m+_won+;E>B&?=Rkg+$RFRm=u~+(uL9=mz}b?;ByWGZ-#0d
z>jgTlE6D;$szgF@{B{aIy3-{Gf-me3UbT)r<#F)j7!53St5}CpT&{n1zWB(YCV=E-
zQ~ojqHd10)m~T4<Em9q2r!D=8xUCOpN(?Ta@Kzuoi$)}C7(AkIPpHP4U-7Y0A>X=8
z&Ngk{r9=O(TPW?@f9Ll_-Pk*6qJ`kfM(#rt?kUPxAjwUPvpm>9Z|={&(jWN(TpmXz
zUc``%VIBTw(39OE8i_~Z(#g@>lIzgyhWUj!IjAv9BJ{K#<hQfH8EB(^{s!=P%c)oY
zM9%5T-8=deB3Cd0f~~0fpL`-|f@^X~sztn>&fGAtmIxD+-CmPdc6R;jgh+FaUyvF!
zCI?hbP3=K11jikVanHH0<|_^H^P$>o;iLC_c7s@Rs;V;>BYGd7Q!#Um#?9~oegy>?
zy$^`QOcsFBqaVCsb7HblRqmx6J}G-R<t`K9$vkBQ?M+2AUDMN?V=F`6$OKpXr-vMN
zzD$tmn}<AsLLb?!MEKo(xO%4AlBei@NB5PjL<HP(+q)cM8FS=kd}I!1us}Oup`4jz
zYyiVFCi$e^6jew8hWgg)Eq>BZD}<)Kw(G!_(dzboSb)>})FwgJz8&Ssd^Ji1epTTF
zZPEQ14A4xSNki8)I3Cxs96QaO_%v}G@;BI|LK!40Q<UtYP>HKH6sysQNE_not;Lr)
z)(-8`S*FvbEMTr!|0t>u9p$HP;2j!;Hq=BnVge}++Ss0m52<;zIz$`?XHBF~2(X^l
zI`McM63KX4VF%jg^@_{1wT-<|aniXbU=uahdoA$tqOhfqVNxzy3r1VHFVmEDC9fj6
zM{4Mge)@A<?-&$(^WVv%r0s!kX{%fHua-14&FY}+8)kLWh!OV(aqi+{F2gZ7CO!ln
zq{=6#CU9QLg`Hz#WBVL{NY-y~>|fFTynxaI=#r#glKBe1xr9_%z{rmXgOp^xv;@W+
z=XehLhttm+MhPV~Dfz6IVe~Af6B{CVdSzomIgN;p6Xv#d@EV^)Ouf(l&1>p_{Qr-#
zalC*_r)1CRDh;xWNretsX|sm?DDUs@eeXWZMV}l9d}OA3`^>Hb7+UGL#va39r>D98
z_IZj+1aUZ;(8!5iMx?$AFRplr<7Dk4q&}X}{a%yTT`DX`z5u-R2W~QWU%_y7t%o?B
zxVtCqrsjd)xK4@3ynEk*!bvG;ur#<u=XoAXCrTLxXC1#}edcAraUw}~JUr^}UO!mM
z?<xg^ALP|l+h!IhzVWzvH~h6gc71`GD(%dAL6I$EZlSd=6=t~h9qi?8i_a-^(Vr1V
z)~KGnx%-zhRz;%zx8Vp=7D`L)KT_Oa)<wG07D+OEzgH2lz-_VQMPjGo``KB(^nG~q
z6PP~C!$#Fr?WQ6`@pyL*PT{g^0dT^GTB$Aq@N|Om=_@=|nEBEHaTo91+t9=+cB*0-
zfmdofV&A)nT6yygD(zl;Yt~(C1a;;>o{~VFW;fq*IlU=jmMx-WP2Sq0-zs>{Pt2;S
z+6y8p%|7_mdf@5*-a*X9dG+Sq#|QZka>!A>8m*i)X7*N@{iKc`!^~GTpx`kPoREe!
zi0F`h^mnC(wdnQ*QjAEbgco8$X+ebco5RiFTm~m~7pr;%S7rX6|DXi#2l(efikYbS
zf6avI7ur*$Bx;<`t??c4wICcKUqCerfJ2|P)V0iY9-pqTqvs!Rj%8{@C9D`3B%0;$
z3c$h*iBN`HsolOZ21@lh(e_oJXcU%ogtr#k2WE#0dUyw%O+U^gP~}&~vY4<VH!Hue
zv9bQu)s_h6T>0UBQa1fhZ{4OaN}YHO$Q#?Ag>re7MGTgTIf2X%j&I$ZT|(b9l&;W3
zJM$hw?23vj;%7+sdaIgo;Fi}HqFXM>Y>;+AU`SE7s%ZyA$zTBI){&XjCm=YPZZIQ;
zdOT9_!IH*w5ns13)22RPzouZzq#0^&t8-;09UmgeW~6~NCvdxmOH!oynXfZs!QU5D
zU{BS{fMAYt?&vE^aa;ixDo!;-+}~Nl9tEb(<XPMyuN{N!bJ;X%jd;IvtHD-(4P7U4
z<f~$BA3#J`h8!l;t-4tIzxhJ0BL6$iJG?f8Zfta<r%<$4yEgpPl{bxK2A2;)+P5y?
zWP!i{8C|09YRz=FR$Tp?KtATruDkXJK3)6JGmaJvyYrIm51oaPH`kpjJQuST2qEr$
z*$;&CwTTG~OEO?+%NjD+aektht_9d?%ru|>W4X1@bh>X@!GfOMB|Cj<a*oAW)bjch
zxvCicn00e-SItptiA{&t@T;=1zISW6EM(edHLR#>w9~gL<xYvKsv`SWby{r_sCAe8
zIL!dX>({&2EfYwOmEN%rdB5>UdcBt@Fi0AG<H#L0eMzd`{>`@ZKRF-*v`$WuK73l?
zp<weAtk1Ac^5B6Hooqm&WX|jyJnAW1DDw0m#SLZbkZ>8|($M5BN#o`rdPx(>c<=iA
z#}Pd_1*xwmbY92wezLue5`OVhAma!|(Vx<3byYUut+mM!q%`H~rKN)B^IF3?Ve?uY
z+_yrC*#^U{ET#)FMSj4VaPdRW{=pfu{%eg(CYZHUR^Tpa;j*%svQvwj&V5Xi3R|nN
z#w{TVJN2~>dy<#(Ac#S1&zEDVuP{Kc%&_Pj!PFQ=Z}1epC_`IkWP8OFIvIrR)Y!5-
zb*Z<tIt*T^<UX$t8Xl)Zc8ikstC%Ca5h<<1q`@&MFlbu>K$<Dn{Q#3i-V_*9Y3Xke
zDnho@bj6YP;c6Khw=aUd-DKMIe4=v!uthQH_rhTxzaQEH`}$YgO~3RGiWjov886+x
z^~&1gBPC;-FIt2BHB%oE+&!5wYT|MTAsJlM{>xAb^1Z2ZooFW<sDH~D3RG*paX=jf
z?Y4CQTNet}{~Az)X&1D1`0i&}K)m7DN4v~fx=E2x7+^j0V#I0omEvw9o$G%XMCX4P
zgbZxN$<uj)r7_p|pxoHGlVUHmSoNMcbx;Q_mxCb`_KH{R%#nAOar~BoohFnsji=Dv
z@GcbHWY6Tbi&sm}4jthO-crLIW!)IC`X&HBR0J&T{Ep0on?><R0@qG<W2%F}{={0U
z8D}!RvcmA>${JKa|BRiW@jdW?c^WLTEVmpffb2kWHoW*_+QCvdIS42=4_q0Sn<!|A
z0GsZ#{n=vzNBYP4hqk3@uB58>$)H@lHw^`MCCpB*E$#K~hIn`N;WQBEsCB21OJf99
zz<Q;5^u|14R>`JHj9W2aLOgbPEe-~1nlcxu#&^^nc878MkdHOd&shzAy3!yz7rxF{
zDgvHBR%1s^#HV|UYmt1QEA@s}T2`Gt{(}mdwdfbS*Mqm!d7WOe<M#i}Gu-;0M3Rx-
zWf_0^2+ap3@GohTJ5u3gWRv66!R8z?98?B_LBy0EaSrJ{X(y(W94J($?V{r}jXxuw
z5e>mjNYwc31@f^U{ZMvZN4e+B6&5OY&6><mGRLoAfB<9!{rd|w2o+fi5!H<N$h41-
zgW}y>T=i$SIEAJ5Hk%;zn7*D6u#3#1Ss=gI2ecm_Yq(EDpW5d5<5i64Ma|_-R^>%}
zy>*NmQCo=Cj1AVl8>Pv|1RY{*>Kg8~fl*Xht-|wc?}zgZYbGPNH^@6kUAG94f0{G=
zflsID+W=|X^!))NvSH4=?bJhv`#I~G9p;5}$JoO6&OE9{M?|GY-C3ns&NfB{hp#!n
z{_WwBcmUU3?Ps$|W)leQ@|N@UGD~)>VuSbF9J=<i?iFF;MLK-O`|)n6$M>TD$cE8-
z-hahVuF1-ySBnXizd>f*vZHmz0USHQaffs?Pw-B=u{<kpa?(r%TbH(^7I<nn;czxn
zBi+uY6Gzl^+Y3P;(FE(I0Q_HS=bq+UZ}OKd-WF9)jS-}1(zTlrmH-`0`ol;yjisJ!
z!lo1NaOnjzXN3VPIHhNrQ&BSl71?xf03Pw$WLNfGCR^z#B=%(mo}|DWDTsmF$0q$(
z8(5eaMBWiJMXqljz*Ha~41zNL%C_dsKu>&>o+c0Q?kYX(Wd$*y2w5ww_>pV)Wa<ZX
zjRNuS$p|xvGKVrm`E23TB$t08E=ts~+-W6>P=h_{XNt(mLN-p-P!n#S!@6<DZ`{h?
zX92PD@)5AqSiaqon<q~mN)W=Y5ASOAp{Ke+D5+5JoF`(2K{#8$6eMd-?a7DWXw^p6
z+&DlYGyK6?1t8z=_aQDGnSpnQL|4(PadD(TX?q;0NxY1ls*SIW!wpQ$l^%Hh)t+`q
z|0|zN34$B*bg+yv_TidzFAk6yScLD`*RmNbMM<hKxVI-6KMaG7p`ZUz7!>$M;kLQT
z_T`kgtF!W_4%CJs4jdvu5i>ZeR8N(Abtnz6(eoH;LT7nm%~~>BKdL^vx@3fIeD%=w
z`+;D`^D1Aw`;N_|Z7aC#=I#or_IO6oL39&ULk4Ku7lD+}Ak8(7&zC%XRPmvV&Z3&>
z?sI<o18ZUi?%xy7HkC%P>BD9kx?{NfN_D=(>fph`&66_~hTn0Hv03K)n_%f4$_Xs?
zL&3&|GiJKEt3QsOnU9nM17s`P@HW+co2V0L>)&B538_$rN%?qwwn$4+6CKWR{mPqa
z>R)sr;5a@ra+qV*b?aj7u46sV1mROfO`h+-{IgZFgHG&U9`E?Ey1(1m5!Yli5Owsf
z$%#nrZGDfpM0R!!y0Me>JsglUR=TANg0)6Er24o|9ol9ZBLhuF-iiYDZr`X#sU8QE
z;6V@DOYtD26Xd2Bp*d2ajvF)iG~f~m>Cc-+b!n*0*)5SlnVk2(gT*)T_13{`itebi
zIRB_%#MCB~*9UA7Mwayniqnm>pOt5tMYHH9xWf9A8xlc7h_L=XU>YKlC3a=dyn?aA
z?*3urZnE5}mzF-2UrbP&4eV@jL&U+a-fcTRVR(l-mMc;lc>8Hhh#EJv@2oNxuw*R{
zI3{Eo&QnxdhA(lf*cdV;+df$r7&;7FZSyucl`o_0W#}3jd^;r|gMVcjYQ6VfP4?d)
zrfQb|Z6zR~F$<~j&?Q3qipM!ip&-tVNj@5pHKEZhp>S*w#P_vr5Ah3mm-zJEU^#%v
zoLhp1;JLyOKs;fdcdQc^Uf$K6Ew8t>E+uWp@)f>zdX_t(!E~GG3f;AvmvyeW#uP-m
z)M%P!M>mN8XKo<vL6Irn`{>Owo98MjsTH6g|BtZ<H|iT2<nd6l{pC@W!-JV*QyE5R
zY9IkC?4eFsG(%7u{@icU9R0QVHcZTs$LFuJL46gh^u-h9DTfz)1Sc8z$5cXQ=3;k#
zWPj@Zo$@<GJ*#t~9fw`!Y0lNi)gx;vCRT?G;{U0bxT(hVI>959)`H>U6|@uBGWR(8
ztr4jseCz-54f9#4V?}3S2E(NECcn(Hs7`7Y{2#lOgM#0qMtc`*PE1>H-4{PKEn~Gi
zgMUQM@Sy2`SMW_;p0=DXzNdi9jgpG0Zubhbt2k`)MdWEe!Ga_xN?+UR%kLytY_JAN
zzICc|8wnR=7isHo0tjTWdKhhv`f5zq*rBetO(`haY|u)opy`0Rb{nYKqL1W%t+dCO
zFlNDjy;L7I2t0o4XhMm`)VdM$s#k$YaYZxiiSHemt%yfWjl;&0{L{E)?lRyitdmKm
z-%BrD#4GH6trmi?hShv!iuEnnjcep2gHLs4s%Rtl`St+6+W%7GFV`HGxkFG+4$+&@
z&A#>EMe1L6CI;`~<xfp5h@FicS-Kq3>hRa!Np|?6&_8aT${$yzxnR`&602;cpBh=c
ziS|4@O&SzGo|gH*5R{Ay6cZ0&$;U^v7o#+nJDo9!$hamU5~pb!&SM|Amf3a^z6*oq
z>E!`^IBtcKtG$D`D2C(=ezCEH-9=Xv?_6}FPG2CgR@lI>gHTjRoGM}K-Au_UB3p5y
z1Sm<JO;<3KHt}d^{H<LNHom!R?B;meq;TUMgRjm-36VV)0gvv3CeL2%fPQ<~(GEv3
z`-6592}EfF-ZNa61Z0&|A$<YvywfUYz=8i;j%XACU+K_7oUE|qMaSq$YlaW(RN&Rh
z;aMu+MIm+rwqHLIg@4*{rghu0On)-ZGI7(XCP=FIw}EN?fB&v-uL^>o>sQb;d_8kU
zp7{kFL<YX`cNi~%5(iSHMA}M6TJY+6)zLQ7W}(>u`9mA&*bqqv&|VJ<M(sj}B5j?%
z3dA`KfL@$RbLjx~>P{fyAjaRq_ig6sO7+V=RKs%T%cNvZ&xomJO)b-VRNlj{S(VG~
z)9<(zpI%R*rcQ9>Mrjzsj?}99*$4i=m*A>|x;vGxDlJ@_P(<2@`dYR18fN<tjxCV}
z(P&Ti3w{^C|853kh@fBKq)Wxz=Ik7*9M<O0jb^<S6(=Gak%aQd#@I@7Xd@TY@DwZe
zW^bI4R%(5@4f0Vq)p+1)SZ_=F1lcY|L@J7V&hGmS@ijM=ND^Q0!yCl1_YllxHk8sU
z#>Opst+}&yt}6E?d9`oF!AKz&{+p;Od^({mH$s4F;<wRBhe!7fig)4NcEjhN%W9II
z7q%GbzX+A7{mUj%<+0{~`5@3$A7>!kH%wXu2dOZ!z_;^ty8P(<OqWW>V>JSXR;j8*
zZ^1#i>Fk|YvPWW4C4sNta4UZ<?8weeDxvLT>DbBH;ax<BJx)t6+@cj6gs_-b76Qn7
zUg<GUC!(gMt&J43Hr&}d{q#IF>EyQY87bxEJnvUVg*vAkqUqe#&|r|%{_Mz?s(I<G
z>jdz<#m4IHX^7}+pdyh^R(SP-ccu2d;pJf^hX!S5Gc57(qUjI`e;dL0;qoZIHsp#|
zg<(8yhl-A>vuhGTsWLBCN2eeUBOc(tFc3$Q(Dh@NeDhKYzP4Vi?Tcq0tLqiZ>)%^9
z4+YjP&Z81BL+b~Z9WWnM4m+<!Kg@5S9>#8dQq%Wxv4wE}%+g)vCNoMp=Vnf4-sGS^
zHluGJI39RYr~jsFcn5do+;*4ud>cjkkXrlMUwa8<|F*3HRs-B~S3TfYV~c-f_0=Xk
zHbNGed6-(~j9n=Zc99#U)^(PD<OS3SYZ$%q$|kdW{7Qc$u52SI1FSG4HIZcf_5~!#
zHU_MClCIx!{V~pIul8(;>@dKIW<%DgDgtA(j#FJbqLd(Hwdb@kc@|S{=U96fwX16;
zb-dnl{qN-rIdRAbZeZsl_f{xk`cmOlR%yO7y;sISOXOCQLQwtHO-r;07LL=?;0&Bm
zl^0uBc5+=$ofA${OctY-T0!ckVfQ>tCz4a&JYQ=1!^fohPwta&iL+7YE~!$?u}XYA
z(*dXkZZrpP76aL=TUmYI@^K^Cetr4FRJ`0XX4t1$J5r{r{1Iqeiaxv0D-Q~;f0~$l
z%V+(%3g2;9yFcNedzVe_nXJD0Uwpj%mZRT+gJFB=6u_Hs+(nhcZmA?Mp*#EfS6_Se
zkM}Q@2sJjyc~c3$MKxLWNQJII(dr|9l<xY3G-Z3G@Jm0}9KaN9aK3LWXHJu9Si~?<
ze$<TufJQr1SYYo@_{P#KcPp8A#_bV10CC7qNqT|ieg`(U{`6?(`_l2Jni2O8pA)Es
zlwC$0Mq_V`Sgyh(UZFu8+}BTevqdN<8y(Y3bhV-1A#WBf;H0sIi|0lA3Y?kQuyfyN
zG^7knE57>S^BQ*3(jw=lQ=%6{B+#C;a%=^7&>`>#>P2koU~&lq)+7I1{e}jxXX%L|
zgz*)J+&5DQD)AeMSSQpHeo25OXY*pqrL$9|A3?}vh0{G7FvRU2pmhE>3sceee|eH&
zrf!<qvz2Au*W*18A~9!Xh1Rw8O3$kfEGnH7Puh1wnCR&E1Jkupqv|S60stl@ukg{G
z**mGbN-~B6LZE3X73Ak1$=Eb^>Tejna`q<ApQ3;^uYs9|=GEnmwrh9=ThnE^9X`kV
zrL^dszUUbbEv*4wte`vhmi9=W5Y3@OF72I>bb2Wl-N|v)X;s^G=+({H0bBCI6Ah>e
ztAsG)o_`cQ((NyL3wqiBm{0I(4ldawP7g;-)Z;T(Ya$!1*$BCVZyp#ZAY;RSO};PP
zOKn;+P#d#6?%~5LTbBy4OE|aS7$kVVKw?wjQZzSgjprQ3m@asoOyk#0rYB9?flW0$
z3P@0SFJb7lf!QTgBLZg)N3wlrutV+O2Q4d?cX_>xS2Y&$S+O<{k!m&4=3T)Y*r9mh
z{)}Ax@(Z~3S~ZlFyDksrZuX$fuh7sasHk?*VlruLHwO~x#KoRT7&~4Z0x?h1J}a9?
zS@f@*`p=JqX#Cy2aWV4lOt_>FMK!`VBdTUQr^3LkG~UmyvGX|2*ed0a#0%uRP3!Md
zMa_YWoB`V@fPIPMamykddZ<6E|BM)Dsij-Wz44MoXnIH&LlR1vM=$2V-BxQ9BJj$=
zzjbM`eod%Z6SJHOp(a{9FRpt20h3LQziJqBE_^`9)k^k~KvV{F?Dk6V2NYA{z<}*v
zhn|KXAwL}om|{ZCdb86cl#QsX@&)CJn-K>>*MNblERfdcvoc+!SAG228I_a_Ke!+B
zSTeynf`Q$E(fBFS+d35D@Dl<F=3T><H0)+KpXLWJZVkmsXkTm6wX}=hynmmiNOcTH
zT^UqGRzLo9A*aec`Mz5nyK&>!X?k-2(_90WA|>e92;Zjr&2l)NJ9+xg?S2jAqB;gr
z`8*Q&JZ>YRqZM&ho@tgXL1|gF-Z3J^_N^+=_B7qQ1Nq%I)v559(7QMDH$2y=GOIma
zo8z`d*&GxFgRC$RM;;#!2_HQu@~VaHKA-Y*x|nVl5H@O=*cofghVF-1C%um{*!pNh
ziN?jDLQ`!=l(Jm?GBRsb68~_f0|*zzZWbOW#vs)2@QOd^?h)hrDt$@?15Snrtz*qo
zdy)k&VKo+F>=pcW3k2r8#BOvf<$2q}#0fa^(KEA|%9_aGa>Ne#w=aH)*QyJUY1j=L
zMkXaIZJbFMWc(up=dwDePgTC-8f0R8cu-n*=Rk=UXo>SPu_L3rC%~%{h??h*W&MsE
zG&{4kn>U)s{2HAC&+4-4>gwvN*RNsU)6#~pS+2k6oKu3}INRIVy+T3BNl#C|Hzwk8
zNdFRms;ag?Nb~VI3S>UoTCYnzO^J*?RmyD#|5(p4%u+5ay|Q64ud-4C^_iWMzm4b4
zyV?$)a=w%2Dkx_H-J{*`Zh5<d%5>v#-j(X660#VnedYy`_xj`2ps#XgA}B%|(qfX$
z0sZ9@!UJZpQu`kQN}+1yE(J4vFOJlmEw>EeDNcJN27kj@)AdO-{te5jDjuyBIvul<
ze1z$2?N8uS_&I9a$JpoKda7X07>!qQWz*KyRkq1Fk<kD|OI%f}pbV!Jelk&d;?V&E
z{@>}|C<?VI3%n#Y%hl_XHPDQ;vcE5BW5bw{k>PN(m{P1&Kb9`|Lh&LS>ZZknAG+;O
z_XjaZLmONj<H1BF?Rm?;`ET7W_gm-c6B`Tvv4~<pb4zatfejC9b6@qs)>esx&b-R}
z&3<ygS8zwrOb+P)kv0bR=Pvy$NTv=<G}fT)aWUHJ6@!`BiyT^947cxG?B$crbK-`K
z=@y+frWPdek?5p7tyqWGxJGsOr2Fv~m#ABWLYuN(w?Kl|!4I}$<FcAz?oC0Pcb$5B
zdp}Z8lp7BdH{EO!z+f|I4@`q&2CHZ~AVB6MQUh>*B6_oC*dHrq`_Ft>&0dLLDH2L5
zgT~b!17A%`v_I_gi|`wrKO8TIx>q=sgPrnY`NTalpTZ8Sg{hN4GnUe#6Gu6AP+449
zZ)^Ea$jjYt+GcFe=fU%d%I+OA)ev3BV*Bej7&IYQ8s;oIOLI?NaipNY85bZqo~XMw
zNG6MsnZ#U_fY3wJbh;8x>d?$wzIm&^uiNNv8*AO)wb{i+PwlK(qoWn!+nJ9W^D|S+
z2l8@wm@zBp&S#j_{^*%})O2(?_FF^ATt9Y2^5l<RA|NDIEXLGaOt^Yqo$riqglkk=
zY0dpR;-pqwlVMeuMjKxg0ltbKeuPB#i7-H(*JCq7U+^++$!965!?08T-+a0`Ki6Df
zIBpiFeSW+3<S4(PI5ormX6+YslV6Fi`GVIxRh6z!reE)FsxYSW;gQ*ZYfY1>6fu*t
z%z}MVy6SnSbA>l!Kw?aJL!P8`ioh<vq-kbj-cK%H!Dggb41fd>dKmg3Bzl#3b@rvi
z-l>Sm%EfvP!{t~2MO`NvcUck^n@ySHMj6_CG1x`c5rA|jL$yUel6&h-&#S{Llzjod
zuj+3H*qyJVJzoh<w;Pw?H@rSxxxT&4oG|+g8VRwnBO@c*MzchpGf7ERRa2`PlC1~0
zCR0mGY4Ny*fE;3|h(f2yGt&8`p)*1>_)DDtVQwr&=zx2)Z(qfnb$;!&mS(y)%Cz}=
zQ8Z&DU2E+99<Q{un5X7(#T)(+SJf#3S0305QdRC5dBTK3!hpgXe<^{4+GHq2z8ezl
z!R|0Y>9lX2+U>kW4yARWcYR|<F!S}CSBePLNE#L3#jA#S)c5@?#84Q*+p#?}98-PH
z_*Gh;9N@Wtl)Xw2hi_a+ucz^w_KAYWj23i%yFc7k!kGf%UC!DS#;hi{Sg(;~aRc`K
zS!5&v35S4kaMz+vxX(26BZ(vXom;CFHWu%dlRH&yj6mmf-MHp8e9?%(#Y>u>sfuO7
z0EB+cjpOLJ%RPe2)~+H&t_Y6yo0~JRww3`8Jef#H%_7ws+x7_|g<_fo;OSu_{Or_>
zqwvc=9Ta<H%RfV$f9S%L@xV_WvvHgVyCMJwpStr42~EzcEo%totu~Uki)9R6^A^&|
zjDZagTB!feFtp3sjP9(-d{ZYKCxDW<4T@8U)A38b0bzD&DeE$_FvH$0^V+Ztv|~7>
zQ_xzV4I>DO8XLAUAwuN05B5;U3EGp{M&?sH3`*~vpG3(S@?(;sd&94QY)mH&&XxwY
zV-bVxwcHM;5$e0Wb>v}oKmUVeE3`LEr>Y!>3Wh&TKUGX(SKyC*P4K>R1B5Q5MH1e6
z$`xDJBi5-fbV*U>n=2$g%!4K#;Z0?RZ|-j`q`UhT6OCR7nX?OmRh-c63s-#u1K929
zQiISYD&n_qy>oItfzHp5A1R;MQn`KvnU3ea#l-YWNg>jzciaH?9lef5AEv**KdpMj
zuis>%t7~h%p`j>)iL7f0fN|=|NX8fPZYpYO{hhI#5F+l_^z`>dMMcLK7lz9%-tKn?
z)f=2o8k(A#H=l`kT`Q!|{_Z${?VORN2Hj)%a%A8~ZS3qaJ?<_g6IlXT%_rl3ED9g|
zxqv%j9_CSabZ-BNQc3h|N<&}GT_fK21?qJfabv1ph9k7lkS+uAodYU>`E?EF`IVQu
zVMZ=$mRkk{k(tVk$Adxna#B1g&_it#Q56@>_|38Dj2HHmW6{IQLQ)#~P(ucHD6RgL
zB)X*K%P^(nJpqF2BKUJVGIq}3PcndLLh<R>%JWZLkIt^xH>RB88GsSetNctXuPFF?
z2HezF*^(+<V_QHI1vhEdn3O9&zpyDmDUVtLon>*YExL-ZZx|`dp~qc?_PDCa4)?*>
zK2&B{s4BrDL<0HP78y6hA&mu{fTDg@Mjn*WFw8Fa4QXf!6E7OrpA!<70~Zm3uXnBe
z)>wm!ebfD^1o+56>+E`q_Y28X9y}c#9qD8aIl}><`SBJB4Gpa;l%P8Zi{_iObZ$*e
z{Pwnq(QvBN)=&x_mjmV3uTy62=iw8DDr<*_L-}%9*AG`qfuW%}6&2Ch*&ipNI>Tqf
z{Nc^bPfGBA0TKnR&4lP^VP6E4nxm$>=g)Eik32a!IqRF7cI!QFH-}RC!AFkAs4Eb&
z1xWr4RcAOxc=@u#Y=RnwTq;pmU0vPs=ott<Kd|kCOrDx4jIL`cW>={UrPgSX5s{sG
zvmqgkAH9>EsiY))MFSlV90;iKs_U$9z{j@Cm$a4;T(+~%ie?HjA`7$Eqa>JH*sUIX
zjPoJ>K6HPN?HE}qDOR>!KulB)@}B_D8S$||4wIF1Ov_4g;^sFwX@{~5L3VGZc?{HY
z+KC#qYOWRnqS3Gyxh-9PPragTctsOg;#g+5-5pL%_tA<B65x>A*%B6^4hK#1IhP-2
z8<=XJ8%0+T03YD|=TcyS9GtLF%~7c?RquEJGhJ?sW@~F(`R=M795+3d4Txbq^0BkW
zfmlBZgrF<&H)mVS-+R#CAR*B#OeM0K*X$M7GvXpo-9HqmmE*EmsK(g-Da>^NYj(F6
zd!pd5kd$=_Gto+<1~D=4Qd70|#(ek3U&_kMO-IsUmpt$2C}dJUe*RoKc0ld6`n;$@
z^F?F&?*i2U+~f*IDq!S%_0UiqOVad}8J62-ao)$~Q?>U~62WfVc2iG>zN<{mH26*R
zvNsmpH6L=WFANCPX^=!7mp+2DU#r@kCd&C7ZdBt#YrA}co+Z5mtEAPdmLWADgQ51)
zWdtjx?C$PIA1`B0#tv9QO~QRek1Y&ABHK`O&-pSJx-UI!3?NWFxa)Ak+Z)8BHehk8
zyn?3r=`iT)8T7tb;t4gEeNcxvs*^3iWm|iT?||$b59hpzTO^}2tE?IPb~>h@LjLt|
zEAMmwVv`3#XlH1eB?HKe>aP~vp!c@lzP*Ar-BPcttgur5@;vROFk!*>1d$H7HlJ~k
zqobqIv9XnbMFgR|@ssVWaBcfLKfjm6z`ZR@a!QI?i7sqlVBki%Z*Wi$hx;`n7+>IZ
zmzq3iwd&6QaBK~yZPjk2(mb90dQ0{zBqTdO|CMj6mo}7wjLa9JQO#sDfLk73&UPv(
zElo~M{l;q&qJe>pEdmaSMexl4XIvoRTyMnbn2ZCtp);I8vvLP2gx=oXO46{1l=SqQ
zH(P1!oW-N6t-<8~5_6K`-&saRe7&0RG82^_*K-0gD&mgcZ5*dw8#?Oojy&{wbKKOH
zZeGA|D}QRARWv!u9^h~+mwlsOkYmF^AieJW76ev!tb3;J8i)(q=lv<FZUtBK-$)Ee
zfri~zrM~n-n8ZDsZI32GuGMvBJ+-`Ym>Zas!3FR$%GUgoX1FUa&cGgLHe{$#3{Li)
zG*Os@7q|J6T8O9VyCP|C)FJUB9d@J<Jkp$(ZDlLp`C#Z#cyJ#U%AMIDB=FY+t!-UF
zCrH}g54eMWsn!uRapawc9`3@0?nk3p0H^1Dh&XrU?`vPmVk;4x%&e($Dvmln=U1_x
z$arnhR0`L-h0u{5hk^LdzIe<=Fw1Le?F&vzgU^2{PK$w1D74p>?8ArR?HzMdQ?US4
zqUXm&#O=7gJ(^`PUE+Ox?F<K3d|&N&d%gn$wiwtP&T}TGKsCQl%+4kPBO|hG){6sR
zT%4(~C9`gRV9OE>Z_=MHgf^)^UI#NY>F)k5P`TI68eQJq6$dv!e{jiWekjk=0=E9d
zA0-yEzuSLs!AA$;O(v^(j&gg<%B)EZmVYPh7ft6rPchH-rhK`;f)+z(VI9T%#a}ZU
z<J6_ow<%p0ok9jcMumAi0cc2|<dwwfXg}fEpC_!{%lXYyzw+R!H17}SkWGP3JI5P*
zBB(w;YXgoe<y`lE&86pK;lcy2v(;IEU)W)wa*lUn0waa8DfDDU&G2(&)D4F-7xQxH
zC{N?%!dx9_pXieBj|eB;1f|HF6{5lS5>WSFCkHOq#{jas2eI-euu|S+>Emp(_`S>h
zsR+{v1k|w}Hyd$;eD3j&4^RsaFgSIz`6J=+d-8aCdY<o21cL*LsMmOJx3jS5r;YUV
zTEqDtRUfxP3D~ty+)|{w!4Jb;pO4ExsS_WQ)FGRY>P5KsTwJ$7`le!HuO(t>Vbei$
zAPFQe<@7ZS%4xWODIk565(5*Jm`Ak`JQ4~Acnu(lDj4MX$Dpj#W^l#@A%7l22K(`6
z9x<S)Ag$D2o<p;{Dnzqc5nwUR!|0Ld7tET_J=zJ;nfrNNX{u^{d!XHFqBv9#iqK$Q
z<t3psm4Cn0%BtzyKa4>8*JuGL3|OQ`e$<7#k|gQBd7uyr7W=@@iWr)MjVNP$KgZsd
zXQ=zoAi}OHV8XbD)AUWym453|__rfzLfe3QQHq7F^|DyI6`jo~WF_DQn6jv`@bUY8
zq7Z;%qqaaeNyr7h_O9Ns6l7MXKI<TDG?}mYyf+D1OLaReV4U=OjqwMVpE3j^)P^Oh
z02Q^)p~4-wWM^$i44R|Wm#1s}OR@|!G&E0IuC82;j(z4{9UY(%K_-O@ONs8P9tz=s
z6cs(k5$EGpL_BUs+BD#ibFC{h%Y3TX&(Cjpb5ps{C3v^M(Ysn9Pr5Q8;92_Zfb^Px
z9x4*Tip{IV2Y`VB2?nftnWd@_jjF4mj}>B(HWpjS$m;A!oNot-n`_U*+Z5m9V;ND=
zq1zPcSu0~LXXkJ<qJwf^W!E{wrYVHZWIhii^zK_4kNr$7_n51yN>ETRBYre{M^zQ;
z{jvyn6x8+)4t5sm*&E%iHi4%*643M*ACLPCO8^jOptLNv`ie<Nw0CaIF2#d+OEBjS
zoOlS{0Xhao`%Jkp^m?IrDtv6mfxMETv8dua?JKu}ynG)CyZPk6I&SVC2oG#m7?A{+
zU*kFL$&qmAKhn|`-8)%XSxKgF#ROtdtZr;5oppSz{5A!U3O5=uM!&_t5;0qKu`370
z!H%ugg?iXf4acswGVDda;H00UCLbT?U%%7O>K*#l+Bb{MF2fHNVSOMeYoXFmb?`N)
z93J;D38sP3m^iEZoBn{TJ!lq%PRJZx)X9TBkZ)k)-%V`-2S@G7#$3cQHqofbx}kt!
z+1fZ~oe@9Ymcj^D-0J?3v`zKqOTft*){MDMvp`Xmds<#mHlMabF{(caN*iW)VB#s*
zYr+`m&XihIqX@$HlCtWq(K-0$s?d^>(jq~5c4<9jzCclih`7X1L7kr|^<<#!-Ufxm
zT8<GCgB_X^TXtXe2T^5X49;j$da6leW2!hEdEUGR>dqVMIBA3ho^m1Jm&WspZf)1P
z+^-g#u5WHWd)zwo_4TQieSa;(ck|tUF8<w>`XldnCHZ3`f|BL%qW^6FS=r`b(izA#
zKU**IfM`B2C`kX|_F^E7KW%3`j~Tpz=i`krvFG)Q`T~?u3esHeww2=4XPpTW)#jn9
zw?gxEU>l_x+Y{L(V5KdtE#0NmTwPU}XAjNbIlguP7qb}_FK91^kw1VR$D<W&9FB;^
zbelM0RY#h2xCJf`xv7K7-IlMNkZ2oy72gVe74BI4`l<9Q2CE!KR$#7~e$W9<R6}pl
zqCpU}3l}4l{1RVE-@4RE?k)OP!JWqPi>q!*v+P59pIdieXeHS@#btA7fS%UrXg-gj
zy?gL|g~t&$N%~u9l9|>-9saHJ-$dIE)x+P9Ru=r4e7ird_x6&1f<|G0sp^Rg^*~!j
zOzzP#e2$A_9CDNu;(W<p<90*PnC2FyJ)Fu3*xIxzi=SDiCEP6E(^1l5vg{BJlt9xu
zyE>ah+PMJNdRe>TSMuVmbA8v>?<8PMBg+!0hv;>W!U21ucG5S@Rsg*wwC3Qmt*q}}
zum-8Gcf2J9=XJkwWLS5cirnG=q=3Rw^fG%bo{v%qdm@}FYGr#v=yJoMXVOaorK&)Z
zb*ks&b{IpVWyy~Zf@elKv4;C`7VeeL`&Vs0*1CXDnlx35qs7J;a%T(-jMuMT<vS2I
zY7c`j6xJ$9aiD#z3$eG}=_pqPsx8v#YF}w?VNo=56wR)Z{FEz8X)>1mS}Ku64?MHa
zL~S6MGwOK<fQiucVpnOS7HjwC3`Rqm`>B}AjFVJUWtRontG32&UB(j66BXKoJ5&Jw
z2Od)`9CVqF<L;-P;nvSOj=1?~6M22dGUM@gk27T<q%sXd3Q$ja>bFONh)%_ZL!!e=
z-%i=f{dKu*O8w-CvZFJ&Q@aCSByrm6GN-tqZNxm$w@fyj_P*Fv(DP(lYGw~3b7T9)
zU!ox7rgALrEe<<l#-Z}f_5M@rpj$Y=uW}zQcJwHjr;Rpc#_!q~wq)EqSl6kOF|{W^
z6@V%&|Djj=^^N*B%3Mn7ILO{+`PuN-5cfkOVuB~^vRTEPMPH^>1=Z`0t>|^_B*iwt
z*9U~H<8NpU<yVg^VC($!>Gyg2r?m3MoOT)Oni)%GOJCFlM7YwwM%`?&VOaA8c;a6O
z=!Qq&Lp=32wAp|P<R7*PikyV{XoHqHY;a$>YSeu$o%ud%$e@>Byj%%)G#+u2$*f=^
zZB+|+GiImuaLe;9T>Qyq<vXcXmI$?M-+KuZW2Byj_`b<uWxMUE1LU+D!ccCOLdShC
zE4rMQM;U~;s`YP{H8kr*r0P}NfzU)p>)zgpyX*_s#E9g^sx&XFRbOdA^p}D)(uIw3
z0&<$}cL5hu507_8!otFex6X=+iZ^z1RhHH+15;Dg?k*r{7om{$aJP~)x&<-W$$FGb
zVrAr%rmeCleH^o4A7-vh8fr`3kG*&v7rT^wiTvcAU$1;d-fv%UH38NDVg1uBiog8V
zJ#Xugx+e!k$7*(qsc5Ws<pykWfJYxm^YHrEjY7lCg=<lfRC0-k3FGAbEoZRtU_?`(
zgKyq_w(xedDM|?q5;nQ5$rM*4y~x?(J{?N&2Ua>x?<<`{QE~Z(=wJ9ph;NJ2^OsRH
z?Wkz$gj&NS`I5Mz^K|ZjNYO4y$f)OE!%yyP9vl>x)^G&yA?Ln&sbd{-Ps;iJN>nB)
zc_r%sPBjjEK~y9UVL~&E7;Zlf?hnUlffHc+hwE`^WU(9?j;VeHAVKS`6UBHRHF(J|
z6@hK<zrvgbwY!giJ>X;nn2RL*19)@3f44L&&=Vzpg9Jb5s=@>8*>n1O52h6@kF;5i
zeMUBYndQ+^>7$Hn#%nR;)v3Z!>q@d?tKxhbH&yCbR_GD(gOZLsa4GF3tSy9R=S8qF
z{ui7rP<z&-{RuC&PSR&I8V_~89=G*F*;C)RrbDdQnjTvyA&0*OpY=gulstl4d1;BV
zQlcN)?gV~L3_LKarFLAw;EX$J_su+F^`$zhiOOIf7#w_Tp02Og!ikugbpPK>&>uHf
zZLPE8%KvbVp^z(CP;vzUU%ma01~ut4dsg#U85(x>{3cT?)rSX_Si|$}(ewS8^2yRa
ziqs+@xUGFLG@P#Il>Cp^WEkX<`4=29pH+b5?r;+DP3>vFe*F^17-6RvS5Qz04O3&(
z_WgFO4PC4(so{Srz<dddJ)GEj*zDOpITP0-wr6tzy$>(-?+crJv6L<r@YW*ff>AD@
znt89$R`Sl`D}?`qXN9BLT@50aHSsO&8vFs6L2n}Ug;Ka5x=AO8649tR?{#D+u&e-a
zJf39)t>oo#=g9Dmlc96s5FNf35(MM7EO9Y0uw(YEn7f(so~gG~cw)UHGkA;0k&)1t
zlPVuP!I25Pb`>1e>jRcIi5;W68D>^9Ot`At2gcGz{4bj(j|GvAA)wro@~-PfkU{0z
znid@OF1(l0>E)ElcqnkB&nbon0yahs@xx0ye^YN=rLU~^2IkIroEC##*7SBp?RuK`
z1fzGAeT5K@*_Y_k<ybh6oJjq}ZMaL~Q;-PQcsIPha|gmF-AV{4>V27#_~9%~j%s_M
z$(0^e+`E<qt>p(*@Fmqo8#Me(My-22F(_+Sx^hN*5@~6lJ~@#%1WdtY*8nu*e-ZdB
zU<t-XhHE=+*>Uubb|eL+`Y0GpXIXKZF9kmm1qW0X9GlS5Ku}teCDK_(ehBd^I$~Ap
zl-N~doGJX?PpW?HS%ZAU2KCU{RHCw}j3(FeIe`lPC<8aq`}sjGr^5k|ByZ2xG7Aa{
za&U8lQyU|CzvCh!fEq_25E!@uDnwhu{7<w?R@);PN?~)kQUuuanmz4k{2>qs9al;i
zC`YBbY~qCxaU%$P|Nimuu^K1b{iGA8NVOEsOS{qa<aQ6rtk?E3k;PR0@yS;@FDpIW
z>uhVd+@KpZTKtnj-lBTV$xdEcncaps47qf25C7ABJMef$;^PDOl$4aT=rYR6%C>-+
z3#2?c*I=kCt%SJfMRBeU3=ikYr14c)FQtN_LO8uv9moQMB+0>v1tHVZs@l#g{u)Dt
zHd_;Sjq#kEoE!aviGl3-J~sSZKTK%$umS&^O0pNZmQbczH)gl1T;${?3~m+R(psl;
zvXXzoHd_eL>X#TYYS(D?ktF}bYvy`amhx`V8deWAyHitdZmV#YZA5+gzV{8&5QRkb
z@8+q7-Sxx451K0nM(=(37d5{Ls$0bribA-1cKA`flRmj)$^WpsGLo`f4o0_;wR0Xt
zU3E<UGvjy_@ztP72s$CCV!V^A5;Na8<9j$3fZmb{%n<a7e=Bsoz$43^L<}x&Ih?x-
zR_TMeC$}_M>IqC`K>1^L@z-9=;!Yn+wPz)2`A<|myM3wGu)2^JN*roe$yIl=Ik5rJ
z-0>uIJ#SNhgq;djM+y81U+`NCVR0Q?m7Dp(%i1FQreb=xpDrmevEN)%$(c>aDj?`S
z8gfJfzU67$T4^37eG~P~9O$AXD${W&NKx4b=4Lu<wBLmF3bL0p$a0qA{)HqoZjjoK
z=gH)OXvEpg&A`$!RFhtmJc=}~-`2(^9V9u`ADz;`ts*2uZEecxa{3*7wIaypyM~5P
z6C{7i7l6bd1pFUP9y$Y$gsoW92sDF*0vvkUr^h+#=1&|PxFFs7&ZKtZJTx@qu=_U<
zR2(9blilYyK#K4Fa51&9#GqE@7ZHI@OGnoqK0G@5o0!-2`2T3S3ZSaGc8zoiD$*s=
zjg&M9(%oIs-5nc2K&87u1f)|s1*99KyE~=xF8uyGgEP*d4r{M?-=|g|bIYIeKS_*o
zenws4ZvcEZH$UGu)DjnmZ8KB(h;PFHPC-3I19vA#5@i7Z5`!6nP~YUh&r_(QvvcA6
z@~jlhrt^b&uf9Fc2PQ9E_~iw&;oYmVo3}7qJm*M61RSN~=W6E%?;Fcr`kDM)$SSHl
z5s(&cui0%lGw)+(X+z_gs7NSLLVU15<U|c;TogIaIju*jz+q+1ws#!YqkVE``tz$c
zt4dA?8~TN?@^{mVB0)M`*<5I%;_s}>=vzDZabJHLW^9=#s)(ukaRPK6pZ)5VkTf7r
zXk4?luZ|qlwQ(|O(lv}k!b2}g(m4y3AsO(&yY65ceM_1V+@UU{JW$6p+S-p+c8MxQ
zk4BQgn?EywGTWlVD9d+VL|U0=l7e6rsB&BwSq?2wef^1MS>r=o>kUza*QO~ddBxDm
zT0vomo?rrj&jD%g1fBYs?fj=!Qho$axJE)G`sf^2WYwO@HtJ176I38&;JaVVmMU~B
z{pnb?d|bzDWN3pTUD}3wtTY>2VE%Npp7i&}CnWdIMCpuo1q))Jo_Tg4!e-iUu+ZqC
zTEYgVM)TeItQ16AQ&Y1EsEo@i)%%)rut}JL+6L@Wgv9i}0QjU)?`)AaKW=N8%I|e<
zd@xrxH8rJDmmVC9L@k$^=gOaHoA~ED83jdlW0SBb-HwS;T!OyubgVrnpU^texg~`6
zP;_qFP`0>w^WN~mHjLKpE4+{|0#Y-VQv#Nmenkk?O<40CO70{nyfy;gfz8}2Ih><J
z+#pePu4->g%;w`nkAR{aBWXA3FV)ZWmdw{O=0*gL+<0;)h`ueRke63BHL<?6HH(eS
z-rIt=vD7HOveVUlH5DG3Xv{<$UU;H*?$wh%O_&o%?QN;2duM0lV`zm)JMtwjM7q2(
zztI!oF*ZOVNue~FKg}3nYKGNjHOR}2?Z2pwj!xi#)1eFU;Vu)vMy;Hx<7II?mv#?S
zo7d-|ej{QDx11U6c}LJbiHhj?Z6Q(YPkn&1$B!M$KlQF_HEE-5c>4J+A#XZwS`Qe)
znWja5XSa60jEfjEb$@Hjich$3k3E0;{*qo7suxR3YgMMyB`<!r^S9wG;k`uQ2Vx98
zCO&}+lev`fy$kg$1Htrr8=gMu!5;=*R7Aaj!Gw}c8F1Fwn4i#vr1IG_XRcKwXhrfX
zmX?P)pYznlIBW|4d)%yma!TcP3L+N|w_p18WH60aTv!<P>gq~EOX~$5o`i-5UI2vn
z^uy6oD;Kw_KF#Zx9{2sKNl=MAPLd++n()(8=eG8CkCiXfV57$^t0}Fg#F1bmQc_l)
zZua5ri)S=?xWA*Ipa3`C+Pk??v$My3{*3H-aUkpN?w+&CcYnFccRDOmq*e9RWp|<+
zdhi8!y}hd|6(3*fM>r%_Z-BpPRhhA<r=kY>0}?VSH^VpiY<Kdf_m!jT{){L<i$Qah
zDV@L+O2F|Ie8<;`!n3@0r%O*UFih`nuiz07#@%SZA!{gpZVmG)Z&$;Zb=8gzd50(d
zH`Xp!#U^IdQj>JYAvgJ%(EbXQY9j&tC^3#Mne9!Fr~Fapel6z>WSgCG956+nnT1t9
zuc<RdDBE{mBcG?Tpw=3D*X@TiNA@GKcAxq%1in<Skov~4-g(bQhRdxaet6<J#^&Pz
z<0+F@A1K`IcV3I+S$_62?)M=#oS<HO^GnukxNzA0Fzs^_)*19evBmuMvWvX?jJT6c
z;ZKDQSarI%+lYkF*ynlQ3O%ewp0Ic{uR)2f(ZV1`ZMv0wf3@?M8H1a>`S7<h2x@hn
zQ)MOYl}CoVX7rL=&&bav(wy?!tSFSbEpFVIuNwtkYESMwJa0^L)4)pCzaVw(*XhdV
zxM5+aLzOO(8j|$NQE4r1P<%I6>4l6ri1#jQN$cDm2kZ&Zr-npOWTf%B{QxB^U=VwX
zjDmxM*~|usfY!?B6b}%a=tNwRVDOV)hBP(tuXaV4-CiDZEOag|k_Kd&Hl@pijrI0^
zRm>85V;SoZs3|<pO55snveq-3E@5bBxOFzVACxPXR^S}{8XQ4@_be$X;kZ_L|Ll|3
zSVtnX3geD=@y3Fvbg6l=lh@EuM)*!5OfD9ujaZhjTN)uu#h5ZXlU7l>MrbmiqV0ua
zG;4N&4CWY>WmYly-r#Evp7J*M(&@qZD!t<T&1f0IO{OTE7z$Mlq-g?uLm%DR*d<v$
zM-oext8y<kJF$mb0`;sHpQpW3LvFtJ$PZ*f6s36DZq%vXx+q`GWxIc!*CPIDeDOBt
zw$8wq^1_92Ux>2hc&>oDPMwvOo*_`dE8pi6Z5#2Dvq1jHmvJihCww9VZ0D1-6lt{*
z{qd_5dxGM=mKZq*lvAEu%X^!0KNpHo@TZ<^b{>cYn`|C*w`T~*M&Kg7xVunXxk#m=
zhTg<{--9h8+P6P){$WuQ4}W17Z0D|=bwU1!P=Kl3AY(gnWQQ8~ov)P;2gkJc2)u^m
z$f^wPmf|cX=o#%q$?2{3=372cj!dU!tW6;!+cib8FnMXZmK5v$CU81pbK}iJQ)#76
ze#P+gZ|(0T9a@Qs$G!&~_{Wn}k@gT<I|u8@+|?(9B`h4FwNr7~h$l3S0{9vEL%en9
zIt(<@w8fK{QoYn5#P3+-1<L_JkMU&{8iPEtm5bIN6U!o-qNwqDeEV^oWcOjCl~<lJ
zMfI>ckJ}IUxF6VKBw*5Swmu+3cvl<T)6a2J)B4H>#jSK#9>Gbc^JI1wlEo9YW3h1b
z)FdX7&pA%qcQH;kJiu9Rf%>mg+3ZG3Cw5j6W$;(7ce|(lD*5tlAa|aU9Z&=;_{r#9
z3{l6FMm>%avcJM%fBMqd$qDO4Bp9G};74BeeknRFsWDBcZ;{W`BU;GR+5WQ*KMU)C
z!t26nAAz+`hLl0s2eRqk@ru_|SJMzaUF`Zi>l+>W^NUhkhD;C;pz3AbSAMn@>HdX*
zE$68Ayy0Y3agyhDR)?2*-RM4~DVV%t8X+?42`*NFJKG=CgH63v9QS>G=^*@cC;Z6_
zr!%jEJw98{1%yPFfbs>HX#$n=iK>DH#zES(y$tcsvkqS?6CwJSCb+NEyUU&2e!p#`
zWqdvRY_618N_8MT_nc(I8z>0GIJQi5NLg26xNkPAaGoZx(e%)_jZUD7GvmV6BVd8A
zmgV7i%l%wCHfbnHFv8xqi*|H~S~)f@qD`1qH>>(1yx=pr5TtfR4ffCZJpVnIPXx~n
z=iuMvslS8Bh`BuM(P-cB+Ml&F)tJjtX3Sy;8T6fRpvu;^Hwxe71Y%tcFScKPjFu*A
zLa+IePE%YU{~G@kG5oeVfd64u!a;6142dXPFzPCw8&ts08ulq>n{`CMf#>XAAzu~e
zLdR>RbIYGC**Mz!EefS3F{2;pS796!i0M^zLJcu)pr-Z*G*ijoxV|p!y_>TlK)4}`
z>0_aIDUyyzBr6e~5j&{6I>^}nGR>h;<nD`>m(bVTER-+(Tfz#v$Cm0pH~r<ZOdjM@
zrr+9OfmU@pnDy5PY}$m-!Z^c8iK)Z+5IL7;!?0T~i<aPA-HdS#sJH37ta?A7X1+Jm
zB83*LhU|va!oo<iFkWV_SL}zVrY_iDise)&Sw>|tDaAa`0GAiD2IKC3a|kv)g}nZR
zEhbcbV;nyk!h{?Fw8>66`C?<TBudTd)rAxl*e1AtE<YReJ$Qe=nDq($8KjKxZz&ug
zzHX&Q7H-;^azZZjz4~+s7h*nchsp{i&E+|)T-?!xO<H~CZpS+K_A9+Ok{){tb3=md
z&n)2;wWSpz%)m``zUw&H{wCz^l4^-_v)_x8zIFfgw2er}Wxr12N3il$8Arg%yyop)
ztI`XM-z=wnW(CC?HVZXRk#Ma8ZIX_%BOJP7asx%`Xc*$y_Z_?rZ+36E0k#5J_pk`G
zTESH7^14NhXEb<y)S>OFk|bwk{?T?WgJ-*@E0JVL>p>GUa@>Rb7g19hycNT-;64L0
z9KL;|_|7}IremH0UxC2yMX+je$s#Kw?)h$qLYUhgQ@wgpDzGnifs4jkiu%x=(<=9_
zKS!h--SbfLWqe%A#JUv1yXR15>byI=tBIr$A$i@)L2wy`WrVzjpM%k#kJ+oZ8pkI@
zUs?a=VB!UN3cHyd)kOJH<430JBRLt^(0vgf`rFqnkAFL&T#3I>+<8MQ(Pwh?`ytBv
znw4gqB|GZ$hH2`clY{P$)Z|vFvtzhf*-l)Yvu6SG-2y?Nj^Wy0iD&T8yJrNEpw=*a
zW^?z+D;9vYpo0-D9G|cx1gBM27CGp9DsAVJLGiS@_J_qT0}H;jF*(qciY6QGctpI;
z9XLK4!i&%FEH)p=Xf4TszsmW5zY^p7%LDkhG!B~}x7FWYA%h-O_mR(Zqt;K0Bt=V9
zhJBsKzi_6B^;O_P_g(Ac4ofx&pmBLujSWu~EGIWQiEMXC6P{R1t-)!?X@t`luA}K8
zYIsBanyb>BHz=;>C=!VaUy<zk2AgJl#MRl3!G^8`)^;M7Lr$P%jbteL1^a%eIo6Rp
z&Ts4ynkn3e@C*6En1xcowgPo1Hr?rNu66KpjaUoS#|b}3UIQkclf!CRQ9adr-h6^L
z^l!s=10RA=me&?(8n7vB3e%b+rW%uU>Z-VL6_@#?k`WDbO}-13h|SokaD)h__EF7p
z8|~Da+~tnUJ1eXtcN<r9^d7unVX-e7TXdtMqt(PH2q8(>fREuuHw#hC)f`cd8>}H2
z_XIZiV)a&#;rABjbXzKsWFlJ$a=f)CR!SoS^2T?BXfT$sYm5W*8<W&6u?JM<7D#rS
za6_oR2~)J035BOFT(|-en@8V?^Fqe?ZhMr!i<ka82!!EG&*2YG*84`x*{Og}$|&L(
zz$+o?&HnZgF+Ik7XT6(^yl<l6T3SL2A+OX%h`vOrNJ3JNu3>}!$uBG{fnQ1{y6Tyz
zZbFCZq#OMpg?l1fLbXyi&8oRv43qTtyD#Aw@}ety>Mrt|#Tmu1(G37==!;BNMSoip
z_)ryOF`~nR*Iv!6Hg;GX6<2aVF#Oq-+{@8=+xBvc&wJuTtV&0AZy?AeUt=b3i^IYm
z$|<B}h&@f?3FT1UXb0U}s4d9j>K#|Q<b2}SZ4tql2tSqfjI|6-&bP2fw)d0#>T)}d
z61JtR&4Awkez!4e5Om``kD!GXH5P>RFoche7X2-b_|w#+_v(kVQ=U=6tv80xCgfaI
zOYXZu1C<<W!r;8|bbhW`)x!1LJ}u3FmcdbUgnE`c>vTu2J722%sui=E(sUQ)l9l9s
ztkD&V##SPLUlX)H!zQ-@Ra0dLv1{1n-YeF$bvLoheB0W0Z)lB`+M3I;Z{LQ8J!pOs
zcgiT3<#_CoPhw){_V?fqIgvsLbh6~+Sm<9;Wvs6iG`ZKD=eBPxRB^ut_2=V(w2Z-e
zbEh`n;Fjxkap1f)yd0mQt)(?uZKFL?Z4(kVUK~ffKU+i1$47(~LKr7os8-zesNPJ(
za|)(G7KR(EH_|8Z+F)eQL9o8&i4F}xvZ~be)gpqR>6De!LeOX?H;Z(R+k6vF`)RlK
zn>-^T!B+bIl5elSXc`Ja(7Kdoc#pe;y(7O?xvW}}PU05V9^xj2Z3UWkU0&p1tJx_!
zuG3x4?elxOQ!gj@Fn=F1I?{!(INN}dFZ+`(!ha1jXB4}x{2XpgLxw6FR66)T)syhY
zTF^n1g`WVm%Gsxy^xgAJj@bi}lhYN8(N9X^j(vg@c~LE+hR}v6%(V=@)_yRm+gEWk
z((4uu)JjixUYzia3F~&5s#6^a2C_17`g*k}X9NkZXCkLRjFvL}-I6CuDkkQYJc(&d
z_U`X7EtJ~n(C^I|zy@$6&*|-{XPKr%3<q%UdpDFirB<Uw`JzhHd%4-0c71S0ON>`=
zNFFQ{{bb{VCh^k5x;7#b7Pn@#KThrEeqMA!^e)QoolmB4EB#yT{~#4pzklh;nB&n=
zg{<^nVOJXiimmxNCu)F)cMZtyJeg|pa{Hc@Ma#&z4g}*wg1)IMD<6R3?ve9`k&}}n
z;<ERvs9+sT=hsdrnlNVvdSs-k8FV&kG{}GWTUh!L87AKe1Eoj)?U@HN2BTM6bmLbn
zvz-+kIonG=K>>4lJxhlD`loB_ZX4uSH4zI~!icWl9%i$Ij3Qs8lBOlt;+K9F!-R6%
z(G=}pV_ujpgkCS3FccM*=-q7w3+7~Zo0HOVt7@my;jIU}vs=pGhScf`0;|4uL6A2B
zsA-hG?+4OMr80HsMfFi084`JuM97#KEH%i3V$A034cIqMTk*8sS4Ulk$?-5&VbJ7J
zWTqB=%wTF?`x%-4<BVL_-QYTF{BV<GZELl6>#9_0pYwe4SBRpLzpl3A%AWA^Ml7^N
zJ@~YbN8$}5Ms44C37t83!uCZjAY3HjC^e~*!A2U{bM+`C{vx@k@Eh;v%{Fizr?a>P
z`wjJ*5Mg$p*=;X1Wl@}vHr^bv?3d(#7VBb1RYF{^^vi8kBEGa7y+*9>2J$#tbhToL
zud1Id@F6)dXDRgJpRfO9KX|5cgSRUrsrNh>_b~FBT>hw3*YTp>Fn!5P7FrqktHVR2
zzHPZaxfO^yBlze2XZ};CT=tg4_LhB?*9!lt7x-nlNc{deL#^DgksN8<a1ws0$r4?l
z(t;Fd=zK-;CwISuh8lyHJ_QCk>`lD^pff>O(4*#ykPz(`Q1JobULtopdvdG6H*mh*
zE<W6ml8dYPYTH60H3u3#;=}d;4+qh%WtElh^I!hOn4k`5{iWY$R(=WDDXD3pv*MG;
zHmt1R2F0cEnsjil{YWKebbagDgcB&-Go1T_&2uHVd%bg@$hXbG_o_lGORO*|+K?`;
zV02NpE$w*?+t;UdcXYA>@Wa2wlAm>V2un%i#7Z^sm8wG8)t-6e0fGb}y(K%G#51f<
z^<7g##Y#WM5r^(>qKx{I@_+4}`DzhtDuw+`m4=?i5Mb3(*by)jsOGyc>1s;o(*#`V
z3VI$^n+LzzGr~(<M2p|q8C(=S-AlW{^(>aFFXsJ2a$B4uweHPG@v?ndm<3OvfMM(I
zn%vo|deH}3G-A7PBtR4Md8as0wV7HBJHsPP9v2(hr`A3VTdRcXr$e^+5ubdE_28pt
zXuYN4E=JjQdrs_z{!L_dQt|1!F<F&St>?Z=_DFz?*iDz$n<c>rb?#$)39Le&c^5h^
zdP57Z)#%t)lxbb*zv{|)ORaBeXEq;s5Ubq5%@ye0X%urWc62u{N~%C^K&d?mc3TAF
z*Yr@Dp}t^<-K-EfTw|PCuBh&h&Y4bvq$ngG&%1XMayR(b4ecseD>I3Ri`2so*7}Ie
zH@|H#H}wWqZWh?GrRETpJN2``TDqm>)4o1#BRwP_J>>t9T?rqccm1}vv-3xK*7F^0
zNlEC96dD>stk~%c?g1bv`PJ7~J32aE;!3;m0?0T^i6gW@3`XGL)(L}Dpnqq)aF_Sx
zyJtI}r+;b|s+Vp6g&WK*VCht;B-R|c0e}Gz?8Ll0ns`R-&cruHfZ8w^OyzzrCI(yP
zygiB-vxE(wc>oSg63<)JdO@z}aWqKKXmJ8NTYpOaI`FQh$ZT^-W^+Nr!}(Sl<coxU
zWoW%kzCc-x+t0ASoZL^<YRD-1#9r&X%h)Crhww2f(;0Fw2*cv)$z{%@rH#2(BuJS<
zpnW&n`LV*tBRjQDXa8RHe1ebD3Rbdf<*`A*y%5cs&v2Ucw3X$zU~42ygVMhiRllfv
z%~}__P|7cJnFa3ey<i@+5s>T2_vVU&`aG*`F+QR12?5aH!=i+YHR{sJ;^%y%mW?H5
z?dhA<#g}Hd5<Za+qWKlC%FO6VuDtowFrM<3V{ul}58PB5KHLs~e&%!IUrqbwt#}s(
z8x{6lQZewuB=_H2E}djT8u_TrKYt#75s7^9^9kn~g>hs`Dr1#lo_}!)oDq&3)qTcK
zqFWEN1dNXF51Z#!Yty3<)#P5IwkkuGg5Gswy)k2HRwG7@uXR(GZ{S~?i>hDRx1HTN
zzd0u0s792>2AB-ZAnotYx}E0!e}D#ZTbkLqTVu<^z1Nkay**@KwTpx>FEMxvSf@~Y
zHbg%;AX$3w28s;RP=#lof(CoWW6W}W``BX49BBo<o~}uQYtTMyE`@573nnAhh-8{;
zJd~liw2z93Mtlj_dgJhHI$^Y@K32+5Odx)OZjXR_g?J`=LUQ;v)~e@~KATMUrrFc3
zlWe_Yj`?h#XHdig7mBPTevcWGa$<b&k4kjyw@5m0d=AGKEc{HrdM3jF0^jlNqASvq
zGYZ7BNjZf8@pF66WSm&5@5Q*WW!1=f64Lfbf56Khn+~3<WUS0N-x)`i);7fqCZ7?w
zS*tCQO?-{Zdg_ylk#!iGY5Y4kGA8z&qS_NWUiv*kZfKQwN8hI7pk+PjugjVh0!qhP
z?wU%<5`qvFd7e7WyCz{;dYULbaeR+0cCEcRVh79<pn>W*t=*BhIz$u18cIu{)aaqV
zyK`D~OH&Pum`p^co%q8Fj9%~k)&HH=nNL`~f*$wgw}_>!`v-5;FBck%L1yihqn1{(
zCiDRde)-^aOdpCliJZ{iqiYONa~Oc-ERFs?T{F8X-f=a(#B#SmH=-h~C{tTrEU(*s
zqasogxk@|%e^^H4(p>5lmKIE&mL8s6B(tMiSgG*ALh+_x2T_Eoy(#JK!9b%dNdf!A
zjMKs(da-qhtfNM<zRJMY8s%witl_F_z&{3M-_A`ZFV9WWDw@S-PHx-nCh2G?AhqnC
zbiB9vZ^}H40(}_>0Kdjq-`rf?(l{$x6cv5D(7)KYKLC=$`>|3%?Qg&M)1gkyCqDO$
zPnnJdI-$DPSl5eBJZiD3zXJKqy5;)6(a@8EmJB0(kbQd;Ho|nFYNyp2^19=;>zs@#
zw%T7f#CP_LmeX15?Lh-j!1u|WY}5tQ1@?Us{~X!&Z69|w?Sz!Uv=kIlJUl<0r)uGx
z*m9rM^(<1XI<3}T`%a?#d$a8CP34b(nmr+oULCf~cAh8Q_j;Rda<SH7wWp1R`2V#8
zjI(;`N{&XI&?(rflDVZGt6<;ty-g;@-;>Q+N1V7{OTxBIuQmJ?#KA>Y<$qILbf$0_
z&jyP9q_HqP<-9+aiPyg>e5HA5CpW1tjw@5N#Xg6ZCAN<b<KloHjAOeB<P<MAU#iuH
zw<$?(Ey<nwo*5EZaUG<sINx1#1b#rSEcRS4pooL~Cj5Tdb+kaKgeltu+fVJw@sflW
zzG>_K4}Kg1m0lW)p|FSl1vu_=f!ndta=ZX2r&d<*-rDM1T`)JDDqWrANGJ3r01<LO
z@sGzxk#S<Z9b8<<FKULCQ@~D5Zq_)cBqBq0P9o>dUON3-#owNK+r_imx85HTX|YZ4
zUHU*KTRT37ks7Oz9Z;BKPm{2Rwp|1nJrPgX`>?i`P>|+cD!GY6!w|a_z9yVRq9P`z
z;a-kB+IwU<j;CF-L+@THMp?F!!wsIBzYgLu#_)n<a*-@qfHg-dt&kf<g0j`Qlj*7<
z;+-2i5R0)hHS;|CLm!(3?tk_7nI0aRzx;SE$?qEQS_^DnV+Um2Ypm$OjzU&2IZ^yI
zS!iVPBCB!(<&_-e8UJ;e{tDD!T=vIuG2sx=4aR=R0%^Ulh{&S^)u=Nx6Q~KnAp?wO
zk19em90r<2CRWyHBog;;w>Lojk>%|%Rc5$>p$yh1R4d=}(ou{5PgHw|>#qvq#3j2q
zp}y!=m|~J08Y+4$0wMqtoS^*tvgjnXlfkoqiRh`4x<<H(q<F=0!KyA#c{ElQ;eLIw
zU5m(G;Os3;i2Vp5Hf`hGMxm02?(H_L(ZLiWx)qp;a<%NT)9a-N$Q1|L4j!dK9ax|F
zS5TlpEsdy9TTxWP1-<u!aWdC(7I=r$DXwU(ozvkf0o-6dB+vZujeJ2|B#H2C{98D7
z$%j>Iaf1l8QKzLtXFR2=v!c-|tCv~<Ruqp@W+Qv{#LOpH!Ro6o=C~Vjn@{f^ie8dT
z=S-CqKC3Y@;1=Nwj#oMu^crhNZ9LESpdKI>j_k8jfhuig<FV=0U;tZ;<P_|&$O{B4
z%_8P3=y){c46RBgUz7~b6F$8{eyR|G62Jw}hGapd%mlPge=17Xl3=G2gL1;aHKrIv
zDX$P(;Zp>3s3_09i<h%-mULy;O%ma}XHlE$B$m%*;+rl7({<0N)G*?xNkNd4SE7@4
zxY6ZZfeQYfnQp_Y3zK%iY0&SE9`^{SmLxj}B#oEe(NMZ&*r7*wh^S^1PSGwWbq)`r
z2z)j=#Dtbt4&Q-O+X~l+SV2p3_-{_|)a|k3<;H7oV%YeqJU>ntLA@Zpr{Ed<&I**z
zDLnVxf4=Qe#)87b^U;j)26|yP^eL;3>#F|wy~bSz?|<cX<_EC_d36rJ#aCNTw}Cp^
z0toMOb_0Wh{{ji=Pp=ExqGG$d+;7mseJI`=|L5dlbJ3UBkFB&akRZR{-dEhNUcXp<
zm5IYo$NW-qb_<=4%jkDl&ky*_kF{kN3|;XDh~j9M?6ud>+JQ^FV<HJMF-UxqFzK4K
zD%6?h?@l>jM*sAI7<4W9CLEZobff|;Fu%TCr2lTKKk4ZN8Kt|68h-V)=;^0X1%Xyt
z?4$=UhwN~5p)CtHY5u2k$z2-3JfR0+09^Ww)3$m1ZDP_|1g`9q?lvwwEQBBPe$hy^
z<bZhMlN=;}76{AOhy|CllZ@oSj%%v-o+{yZ7|BRM<1=RKf7mHO0Z(6QFR0qwm!3Sf
z9=^T;Z{EBC%Bo<%Nk6(7fZ3PJVFeznMIp5i&6LOQY9)GXuNoBj$;9DG(*k9;M~#K!
zM(~$Po!##k)-1M;jO}Iw$vl5UXir01e*w9gL|;s@M3aS;2vfiV8ayFFSHd9~^_#5T
z@tekbWJVI?;5DU0XM+2l_~^*Eku5_})<y>mDVNhEzSt#So5H!SI>s(Z0Us29>t))O
z8DdN!ZE?JfFV!mT>6(Ov`VSZi8g%$q?b5)OAiZgAS^1|VlTOp}@{|u^kP*(WFaUd;
z{ZWDM=ucm}@N7T>+1I_V7ejI?XcjZ^yxd1^Bt|3Pzgr-ZG#L82Ird)@{)`}q_e8M~
zP;!h~mES}|aNIt)J)S@c{7Ip~0(gL)z-lnF@k_!5a)I~tlB$qxywVE_<)_qt4ZRR`
zXc4{*t!193LR@*itGJ&~vpaRjIC!4JT7OF@bO_+m`lTR$DqZrT>i*U9QuKUh)g<A;
z&ze|Mrz5NuD`l9g+3&4M)${kkUadl#p@#_SDee2QzDW8tba#lISnQ&NGc~60TnbJ8
zYNFBcVln;b<5P>&`_)%)3m!y)CU@#tGhSTBdpk86kbnbck79zshnLPSK?h9F%;{xe
zG_wP<m1t=j*syv}r@XjWcY-C|gE{g?eAlE9<Zn;u2p2W`0lB$6uVHRTt(&EGKFTqQ
z_o`T?U}ioYMbi>UY=XRR*$vz4C$>HQH&!z04~Mpo^CB}H?Cr1hd=mX^$*dn7u_4JW
zW{D;``jJ|ZXHjFM?WMX)#Gy*hEDvr6X)MUv#xISx&h`^@C~dp)>987KQ9eUzx}bP0
ztZ&o{B}3|9k45837K_}huH}g3vJ4h2+lOQGB&SH>I8nEjRCe2GHotwU4$s3}x8JsB
z_{dKnCRzZV%W6V%>j-7T6q^@0M0w+60LcaJ(lNXQTDDjGR2XfY(<N#*iW+ntYd;+-
z-Cv~=>z7ipaf(`EDt-ZTZJXBwB%AA1%|!eIJpB~}|Eu(B4>=B={k}A@izILg1RQ|`
ze~N@}a;&j~`C6!Yp5VYu&`4e&^XF<|VYqk)0ZpyE{ofxSCZOMlMc5!*{YgNk*wD?l
z&kC?!)vvEkDL>R?Q1GAA1txjW#G-<_^BBcWDx{mU7SVlzD>OA5-I;*#6OaGs?@2v+
zFRoaTjle&kyF#G7n(>%}t0Wu``2@H5c;M7YCINoSVBH`fAeT+Hp^75UZC8uh)s8IY
z`K^Jr)~OBD9v6fA1yEri8{?(vMpl{1hr@%<b`3&W?}Jw7rslnoX#T52{%}Wu?hQOa
zm?ZR}cwhJGSWoZxI7uwAe3hy_LptQMJQg;6E&Fm7n^#HfK^vB|0E}j^vEo!rkzs1j
zm4y+oft8tk;4tM)PwAxT?`HEU;xRC^k%ay>+n6uZ$fg~L|A<_j6hH=&KbLw^@O!|=
z@?ic?_)@uCBTzkC&(%f&?mH$r+P_s{JQO1u=<6m`B_3DY@Hca5j+e%3-)R+%HaGmL
zVu)GmuIN8i<yZnDF2;)k%}oF8T*}QiYZ!isFjE!<Ed^miF%s4~A}`D?h=BUlp^Mab
z)o#H=iojIWo1LHf#iOGu@ftFYHt0FM=sh3)doS<Ig?846y>xVlfQf{Pk}`)W2p7D^
zPYr64wq&48qH>*;1Zv0n^*4lOy#AgwRNn4ylbumGEd+N~l<s})P&VDtesFD`>!nne
zV89FF$vwEG*IGqGP21(_+(2sMz$+K!#<3gU1)3j==hH#wS>*QT4Uipk%|-1*E@zcY
z@?O7&CBLy%nM1XRp&~Sdc_S7+Dr}jw9grhw6q`Z*zqeT{o_I&>1(X~6GgVo6Tb*kJ
z0J_Zu?!gsK^~WJ?HW?=5y6nR_?_ZG6G|Y!iMR3uMDq!P?DmY$J<2<iAhf%b`9DbB3
z_;7XR8LyWd-WwdwwG3OijAB~9Trzv=irPKd7N_*J;)J`}8k1W6F!%6}A%Ivj@FOM~
zrT%`M27A=^3DVf>R@h4|APL<17|L4lLTOr1FfIVC2GCdw@8!H+=8WH|=WkC6K%cSb
zFWN?Qww6EK3Q_-kXO?#{dzt1v2eJsVSp+a0IhKsuIQXN{{PexvNPhos#o61v@$z>&
z-<w|Vng1c1Qen3!z{hVTkQDUz={YNQO3Mk^n<P+G1n3|J@GTk2BbP;ars=8Y1^RFw
z-C#2vgqYr!Mc}{jdM(QX)#OcwpL*g%Jj8#x+sr<k<!zI&GvBx0g;youPwfsiQ6EX8
zH2|uQI>(dO<fgJLvu{$Q)r9G!^%F(UbDZrJ+Gy;yb`eYBYf2K85L3oRN+lnS{*pWn
zAG<c?+6L##ePK(f2RwrxIn}+MpiP2+v?#l?dl|TIqQ>**0Z?(c3{#e<F;`doeo)KU
zG79M`B9A&UZis{JUbxq{1do81k1GL8G*7Cx7k+&S%*^_FRhDrQ2dH4`UvLC0{Houc
zjBW^1VaeraTzHv$PhW?o#9GqAd;0=IgrByD{=>G@Vhm1<N|p~JbcIA)R~X$sJFqkc
z1T@DJYfO+|AQ2CD>twNkrLsaVSIssa;llMeACe7~Xwnsf?=>N9-Rv|9PEb4Y;cQF;
zj2<1{CQ?h~M+4r+u4rxga5y$`dE&m|GiWvd*Onl_pyAz7`%Z4X{UZ^8MNx?7rW>CS
zWcIOPzEbWs^k6m4w-WRlg!#(^H+*;X<P{qPWmx_D^9V}z{9$pd7kaxf1u4DTC>q7X
zGOubkxA@PVF6h=Ku5PaG^1gq^k^~x2U|(ME10Gk-3!udS&5k;7qu>nMG{?`SVbf8%
z7G}k^O3NA)F()f=^ilVVDJ!{r*qw^j{cX#Og2Q^*Ncp~2t%VeSco9y({D~ukPp!63
zY5Mm#nMiG(4Nrq@o%%$%BFD(1MRcG*ah_qh@ELLy^_k8KsHrCVz{tkHv*o;ueS2<3
zWhzOt0Vr>l$SvAto0Wq(K6B!hJV|$Ue+~Q9<w%bxcKkyd2i%eJ>r}>HS#_Ap<eq;t
zQy<^&D|QA-6H9?h8x+~D@J;9{X|de&P+*Wk=QSI1L<nvE9=}t(d^}U`SoS3T(FX5(
zU75VYNQ72%(AHjGU&k}kzeIJyWClM6*)41(RT!*enkp0b7Sa8i;PB(XLAf=60Umwd
zk1Nfc{CO6lI(i~4`|5YI>&^BOg=!1I($K@bT7%85#M~eaAgU#xrhr@osm-z<4C_aW
z8%yS`;9!>wlPhWvB76lrh(XZbMwR^QX#|a!*OWYkP9@Luc+orvW&}yzO_A&F@J%Fp
z+IJ#*&%g^=lgaR-C2{1fsqPwXh-$?61I&edv+<LT;dxE-e}A`-B!1FqGHV(p4s&Lu
z%)r!ee31|2jI2(@FTPxv(38#=m9;AVSMmMZWG<ev{Y#YkM%E-z?Lzk<AIzPFCSIU~
z%FfAo6r6qOzh}BA?;@esXQH8D`QJ2{<f?YA0b)Fui)2+0$A<1cLb}F!{!n?<nQigl
zczL0l7t;If%Tx4I7N#k2{#FrDMYVSUdC<!ITf3ObqoRyIj^C?^7?&jT^>>>yIFsH~
zG>H5aFhsxe=0nXP+RzJ{9Q>83Gki}!W(C89(>{s2boMi{n95p19kKDoJ;2iBn~LVq
zWkl%`;lj(6uS-eFWm~A(>cjq75=bd*OVi3_+{6Ic5*vb^;zH#5g#SilPIVM=blm3f
zAfR4h9v<%sQ636k&V7f;!cvA}%Q?|)!fMSB`>DQxwHPFFq!YG_Kc64oapwCw{`U5A
zbMedv*)3G9P&8}gj~kvpexgrwRf3-9mmz^0biaX=w`?DNBRd`br$j_8pOuqB!%P{D
zSePVeBc|a_7hfsAM5Wf@g{nMTDvu(?_vv))Ovw^njD(cf8}~5z-_)=Pm||g@U`ION
z`v0-2DGWJ=J97*6NcghoLXajF($=<T=S5IiW~l3cV*b1Hf#Eg(1|4W4I8nC)V5re(
zCfdf8x+4n=i;1ka*gR;6Gxp}cMK?@N*U$X@V>kAw@CEu6tUEhMeW`|CDb9^N?oT>}
zG$UnEDP=`~XCUikRymh@(h=W_e|11f=J>YzUO<FoZ@00=Um+Zf?TCWphCW7Y)E_Rk
z@`=HO?do=GRntxu>x+pbnZ*YRfBX%2eySf`UtN_02nYt~<Plg|S^JyN9yc5Fqsklt
za8|?Oq9H>p-#sr((^6r;^$*Q$!E9JcIf3DFKlV3g&MihbQvMW|00RP2z&lG~L#gx;
zEg+1K#g30A@fx>2o|pA*PdQn+cEcC=%bk3NA(9^>;{0T9C@AJCGP9HZPliBD@X1cN
z{5IDQ`2d*%J&=++{IIXs*tjlI9|}D)6LgcH`v_&#*}p<s`C3*{^^eBdsj&#Xzv76R
zd#trS-E@C1p5=b#6MiIiaLZb&U)3|8a)x%aIg9ifOoT=rFD6eBSx+4l;dX;_rTEy`
zmmqxTcNZzt55(lMz=(Tw(#H&#q2D7T&owz7*xfxml6WZ|hoE*9mDMo^L7pBOi;jg!
zr?cp<oI@p@h_}r4VOL73CS|6F`9q04ckYV>uz6lNW~>M#+th|royMcLK8+2LGGx@w
z5@y1|&SJR}N|EW(|ASNcQx&GQC*3|*Z2;w6OO4p*;lOiHekvtKg~%1R54Jz|auYZH
zK3VeboOW~e<R`s{sAS7-<^!6_)g}bi&#c{PTGGQex;x+}_GgEVNahEKO{+kZ2w<E)
zwE81Fy1@aB2@3}Y!aHC<HJA&W&2hF~8~yk|EZGif0SQS-G8UFd(2yVX(w8o+IIfBu
z0{QJQsM$`=(FFtqy4T+>2o~!$#0Qgj=G}S)K&`)kX}M>v&S|;@5(HikV6iy5{X4`4
zw6WUN9P1kF1W$pl_t9#<H&+(}NRO3`4KZK>DQ!9@0OI;LJE;qpkJ7h`i<TTn9Na-z
z+IS)rR`d3=yu{-Z^z!?$HbbGHW-7s+pWO>5Ye9)6jgS5<mT)mW(_2uf2=Yu@sC1A~
z@l7@5b{oZ8pIjBlEmK)p+YO2-)%FSgK*EY&U2;BEKjHY{-M@qKfLr{-szh&hc1r2n
zw5BktC@Q!g&bC8XyZ0Z))0^D*S_{Xv9~Hy^u&BAc1w?Nu2$*ATxs3)nHQLctuhvJw
zFm?7PvDsZ+0c$hF?RCxo{E;X(=lfXzU0jctPi+_jmwY<cYj-xt6|0T4KoHR9xb#hF
zS6dqb$uZCed<0SfBO|9>-H=YAzCZcONg(xxahIBks=}ZZ7KcIO7tj&+0y4pt*G`zY
z()?d7{{%EKIe`pp6gWR~-hhw+dT)&q6+bZuQ*=g~lxD2`p6F5^wviAmW~dD^G6Y0Y
zG};bC>FAXG)<Y^zw789J>y&~d%<M+pN`reF2}j@<K|1{=9MFor+h<6i%qbXmcP+sf
z6^_TJ^6rlBEOT*U@nQHkV(^)*GmbLJH175aev+$vQ`v9gQAMxZ-9`bfGY!?)k&=;-
zy%!e;L2)CyXR-0|FNlc;z#m25$mTmyr1@wm1$GMzA}%428{pZL-q^?o+{eLM#Z<3f
z`?s|T!|=H7eQq_Nk-wy(rZ!^1>uo{AB+(H#2cC?6V7Wg!JzWF#asW_PYW>SXdG*Q<
zI6zoEPtAW4Nee+jFrcQR=HS2vF<8Fnz?|PcF%kcE(L1KLmdh#q#4Bi;-u)ph>0hsL
z6BQQ2$T{fxk{bJ6CeEJBWsH}cC;7%irDDg8V|C6eetE6ZteDs--56N<=u8RqWzHgL
z1aV5fWjGN6$(SNhTo?oOcH=z2!0+jcPhAc0IKF`B59%P``~*$=jNnL0Iy1jYmXvhO
zFRT6JL2Ph;YcJf|4=1y>_y3@cSIEKv35oF-LiBIdV>-)!R(>lF;6LWBdl?&RVCS<r
zqe;x|2=50%e-=iYg}_yf6(-Wn*E@x?)gbxD7$}l!>+2m)w-k%)zD~Bbegy0_Ng5>A
z_4^B!CctT#8;^vqRwuRTRVF@=F*1e$2XI2y#_q26y#@%v;{wJ#5)zUk{g!lKU-9?%
zm*I&AF?oyUY)D8*Bpz$e@9%HyfakZP(c>>2*tg8i;qAJ@i3j+z|7|ut#;lJ$ndSbI
zsRX#Y`7lE9tOTmxu>P+hE%jO{fH<dWdBL0;f2@g9$w|?8U1hzs46B>gx`?><QsG<R
zrw_D0+PMMVYu)$hLR(@MuJvw4jWE7GFSbf~6RY7a_B254e0U5&YYJpG$8No<L&sSE
zHk;P5EIP~m^O%L)tmXYG#yi#Uee9hGK<fci{+04m+xRLj!?(~E`cJ2<K<8<vqJs9X
zC_ui*-h<r^@WHjBkLhwOul6VE?j35$LRr}Tt0XEaIs#(!2yA{jZOWZ`^4m2YcQS%A
z;V}k7CY37+fZMC_@OCcMc8eIlRwD#`r5!JLaB*?<e9lxKqxbHwcbOmm9}}KKm&W4)
zYXI_JWM8*xKN<P{Z>j7^TprnuwjHVqSs6!M367CzizvJ|&ynoaD@A{jgkEjhmOTbE
zOljo(m;r7n?EweFb81re)nx%7Z`cs+k_=sVkE|J)O04ys$k-8oKAlv-L;~OdIN|q7
zQU{*et(%O?DQYZp(o(0M+V8#<#~!6YILIy4fxZWTMlCbnQwQRG<@HNI#&U`@NHu|A
z*bj3aaF?nx)wev46)*U>V8Hd_g{hF1k|o)_#w)>as~(sHQ5WTaY97s&JUv{@P%Tu8
zijDoGr+-C+KS^(5)}IIh!b}47Cu537Mn<Z>Qa|S506o20{U97B#0T>W1bN8_#P7b5
zOW_RTw3#t5GgJ3Pudc3!0cjV1cJK~xAEl$<SG#0OCK!0Pr3hQ%_c!R=JYh=s#ry>Z
zXvZ2y$3bG*@!>Og&wxMCh7#nLTzKAv`JCpKaQfAsnr$PQ&kk*l4YWZ3CEa4Dfx+FW
z{-ee|y6&eilxf2;Ue1K0^!*}-YO9Jzt>v6t6D42OcJ!{}N#0)Z@}V<Vtn%3<_H-hm
zxRb+)L(Y7jUe%)BzdiL`ZQ-St)PDb<v2hZ++()f}^+xa1{k;S8GbRb_augJl$22oW
z?P?ZK3K9UeXaVBn=fA7AQBuyi824-<0b35VJb8xl$2MJ}n5v>8+i814?FznWl3KE_
zzn}d|V*z^b(W<iSnAC@4Ra9Pr=|@G{ezy`%;VvyDB_SoXveFTv)98LIxA#w(w3E^5
zSN)z*wyS{&8%ICtx^3cj&Ifm1M5!G4TIs(b0NHLzil_uGRw@L)mGd54I%6^@A-dF1
z$;BVY*67p)YtG;|tBHce=-f*YrY677jw#6FEz4>e(gK@YWv^;++8*9*tH!ps4*_P^
ziDh#JaT3eT3&+4z@n(0#z?^AbrT;iU<fk$%u1YE8VaGq#)2b;bzK`r~PiCfZw-s=Z
zl9J*uYP|<rIq**z0RCed9tr91v{}aT{=Sd4-oW%Uz8_#g=2#WAMD-JxCrh^0JWtL&
zzt_C3tE+>ltzTHUspaf?fs5Pab=VXP40a&ugp!}X$@@D8$Y0X33F;|4G<1nydQ{*6
z+YLzfv9`8;j9vLJqGzFE!EMo}3!IHW9)YM;Tn$B!{Ht&euzsFVSd^>(>O)m=q;0*i
zMW={UX^s-Zv{GKK`WdN*$e3d}lg2iwRxkNZi-}t8o_ycPQJFSVhfoS%#sU=;9|@Y4
zve2XKF%I|$zD%+&1~yxA#?q>h)IGnpmN~gExG?=m&G=IIsqfu$nu%8JHxEM#KEdLU
zswpPX!=OxW_S@)Zu%9A_Um#~nDMk`Jk9$qC`xTWF{ze{Nx4ftx?MW22C_Me=*RPRb
z>0p)MxA}z#C_hC?d8VD;tOiL?xR|yHab;}wv=8)q4e`5~O0b$Oc<GsNI=y=e(d&%1
zHjc5dJqfj+<XuJCnXbO)M*Wv7MJNYFobQ`&Z}w&2GF8xXcNF1?#eBIqrzuA4JZ%_H
z-X2d;E2IS69oXufwuZ?zhB6{r4VFOcQa*bL$kT)YCaJ+4Zs(osq9+v~_Rq4*db$F^
zk4d*czblJrv%+Op4d!tUi-i+`oF-;hSJyn1-jozzXwvlL3DW|R(GZx(TZ=7%Kovzv
zMfIbz9L(|lWR4KcpQa8VkugZ(@y405FROz4iku`iC=a`N{hsqNa1q2Gx2X-+MW;Ma
zfX>Vg|EPtH<^Gwsx0+`5-bW$#7IOA6$9sd<+_(kzYqx8qNGx0EEdBBp!3N=`T6jw$
znuaLH+7W8p6lc(y2P+n{a30#}Df=OpD_yOA5~cUv$*OXIukyhM<J0%-=L8z*u$UxD
zsI`QnhISZxgFzXiOT}F7m(Q#vx<&@RAmIo~ui?$!D>D6;^h#VM)v>$!fbz?!#U@K3
zx6mzAG&tOWg(Iyyaw2qVY<7lL%lMby6F=6mGbxc^2m`^OCz_IiifV%XM{e%PTOVMO
zj=R7lV9$H)s--ofda=2&!3ttlV1h77^)^@4T$mD^fD!Hbt{*G__07w-S%|bp;e>C2
zIocJbH>TQVmh;a}0jA4U2P~wz3ld&F6Q=c;zcv&P9fi#d&K*5G9~)#T&l#OqGAJ2M
zxPXeeOk8*N-fLtaPV<=DcBL&v;q=<=EcBYFqs)e@dK7)KSr3Vy-?7mgUBBtFD{gn#
z;g!CW@@W)8Kp7|6JbW#7wy_;{IPBp}`p1NZ5pN?cXi~0qx18kxeXsFTnp8U@Cgw_#
z?BR=<mMjF6|INCQ%T~C$DkH8gR-pecvn8jzuD`IFk93busNVY}mt}ZACZDOMd4g2x
zvEXx|3wBU3;G64EJpi#mAa5|;eFX(3=*ciX+y9FR(D9WlNaf|_O$Sp^gg|KFV*mlL
z`C(rq#Kzj6tVx2`dq62jw@-+VkMc`h-zTu-tG~OD5DcaObomMfuxO82Vc3jVkCM&t
zLR`>>01?$pb+Pxlj?Ykz*-$#nV~|jLFs3>~XM9{-2Po^4_`O`yNV`&^##2U#fxHux
z;4s6b`hws9`=nJYmNlYMr?e2q><56dlTm4AO<u~!K*MDE`dE2Q!!Jrx<MB}Uc3u$V
zFkoN}^LwrR*t?rf`FhWDS_F;S#I$H~QJ~5;s|YV8Emm>8oQX0=oxrSmSS#F#scDq!
zr^ie{DLDv0#_K(G*@@bHzSR!Msam5>t`}~<BVFl|ZclZ<or1Q_B<vyPqPqfyQx*fM
ztqlL=Xt@(qC1bPcWe)vhuHd|f*hQLkJIvaV#wKP9XMCidB5oY;cel9hb;2`v-{X4*
zyVSZ$>XUJD#!u>b70=v=h=>#xr8M21^cfo)A1!ztIy~H6s5d--;54=En_yDk$Lz=j
z-y569f1r1>tV#yCk?Q^3x{dB{j*pKYSxew0(8p#29!MY&fwa7t063ll@*}Jlo11!F
zF-d(+pFgHca4rlC?y$K7tN!1w`=F*(<8=ar|745Wj)3oNZPnOwkY#24ou%+r*=dLl
zHJl)>B5xY*<B<%D%+rruTEel1;MO+!4=;^3R3Qc=4lWA@>H1r)KuuNB+%2W2n<|G!
zw4(EE($G~EdQtz#_?I%ndHka`iG!=d_PM(Qk-7U@<Qa@Th7R4wjS~4|MlSX2Vur{?
zwat{<n>eNnMx(atu%a-*fewG+Q!(<o91|j`K7);no2^=RTD$$Juz)6h=^stMjoQCk
z$KQMx4biBvB?hS%uYkoITxp-1OC5lOY1$JFQGy}|^vP$&XmL^<z^(c@B*X|L&H?W^
z3i!H|6hhGJnaxyM?9DeMJQhte%)r79Ld6`0_%D{$*8^J(gmT9WS?{V?E7b~308#UJ
z>}84sgYa&Dkj`;FYiBlHVM4_1n5|?pH0Pw4uVTH>m=qy!{rTv~{*NWZ0S-pKlWuY%
ze$QCy4DZpe5zebdXZVc?Df9tK9W64d1Akn38(NgQyYoCdBOmk8i`R4;-`^gJ_i*_<
zSE2I`NEtO1f<Jh2W(6&6ZDd=&oVcqfvyAH+uIE+vsF@h}GJ5;L<RhmSnm_}iV0f1H
zc$b5HlegSBgLAwhA}%ELN*qbf^2ou@z%1vRE)ba{kml}2Hmh`X#7G+JcfZ&F(KWvq
z<5$gSZ?R%KzM2?H#ri_2yCG3l()anGUWq%Woxhv3epzeIM=&2i_+!1Pz=E}p9nN31
zV#D*VhgY0b@peWZ3W*_*etml}B93f_?Ytjd1jgnuDESQ9G4sj+YxWs2v8=PRbJ_)v
zY`b;DU5+ajjz5}ifvt0#yM%l5gPB>^T2D+8$gqjK(ARXwiPLQQ1>)(ZzuJh4cX{pu
zy}sqTHJB=&c~}Uh`7XpUm=@(6Y?sF`TFTL><#Frz?Rnuc8jvjOco2S#PZEPGK528{
zL#ZGIZH2rB1#~~Ju>xjBC%YX{+g{3?+L?l#73=wMWu;vQ7zw72OeDRo`OhbQwEd9b
zQg%L*D{+;=T)s(u7dfVXV;sA@RnXBYYUSJfl;Y~h;{Ntj{Y7=xH~QvlyO>{}4PJMh
z=kz6NE6yefTJc;6<M84>wB?5((RSB*PE@Kx()v6L<1q_Z!QaoymOTDlKv0WfK}?h=
zV9OA5P#@#sbzWvTB(TDq9T}e%mJ-AEL}1}GWUISdpy`V_vGuNVVA<z!w8r%G^z0u$
zU_rDlo#~d^WI-l?J3jEzbXRA@g2z&OMX?{UnbIYwaJ}wd0i#9jcI2YaJVX<a1$h9K
z)=8KD*`9Vq66AYMG|&(<fv0X{C*D_*nL3M$l#BYRf8dtlVhv`vK)r$P(006#e-o36
zZk2)d=%5GjRju0%-66)oe7bH3^s#B$zD96w;A7nv`+V<rL8e;fhXs}PG;e4SRujE1
z8R0lhe(Bz`&>~}OzDtT)y06~0fz;B#{Pgv-0XJ*xsoUDCh`7m{@8nhf>ZS`w^kyRB
zZx75eyZk#A7Ml8(W`JCKa+ixL>&rfA@oGfGq|q>btY*ETtrVuI`qQ;-vhv8iG5kG^
zvUDGPJHxHQ3oH-2MMN&Ef29_<!+1Op3F+xt>pM6rQik78dv_2P7MdWhpPpxyLKbz;
zo;_1oy`(q0J>R#wd&@LEJ<SK2%0w9|9bH|`dguJ{h{yAz)7jw4m9z1fgDOjsHdM|$
zlfm))A|4hoM(Or;g)e9G>(vVB2-v^c--qX4U8vpUl}LSbQ!ThcSeSzS+|$*Rk2}Cq
zU2x4M(okX4>Zy@{2xX-tPyWd4hG{YOlM-!cjZ($1aFOP%JQ$4K8webVHV{Qk^wVxy
zZrzaL4Qi2BVPgLk#D2+%;42mI`*oQzVil?5*$PT^v=>>3NyL3b|L+hft1rg`h6-3Y
z-CmSf_OD5sd4_V@x4qAV*vY?Bj~<t#yYPgA<&8PS{MvPF89EWYfQ{QMDroMcr7YtM
zjMit<;JuR`HDYEtjVm9me)qVj!JX1y#&I<X`8}OG&H?yiMf}i1ouOhKLHKsCpbm@?
z@LmAy2uSGB5$R9gfvX-zXj12i`+_F>LK-eE+x_LF=sy-*cUY<ncM%GN`z_ytZN&F_
zQk{YrlE^#8(>-iZ-JY!HmrknE()Yw-;6nRlqa2h~@MAzk$X#0a#_dz3(j`;Qeb*C>
zX!JTBp80vK()eaoj1jJ>k<DWomZaF#slw%P&LdB4b6rsf++U(_Mk%e|k<C}XzZ8$y
zSRQXn2zQ$-rkNS;MU}oh|8gbeP?Lhh5bno&2bC}#_&bI#+p^H3Qy~}*6J8zX3~@E&
zUZ>k)%TP@;Rjbe*aE`SFrx|<**}*Ng`Vok$-QIt`WZ8AV>}#~SqR+^q{%?fH-h*u$
z_7J2q4_n6hdIA=L0d!2g9}CP!%B&`tK<F}>UqS*dNM)|DWa4$*dqqVR1hTjxEsm1M
z8ZkxDZf%xYKq2`w!WvZ`Nxh%0I4TV_{b^7`-0Q=)SeqY5H1_SisKtY12(!HZ$`MiE
zPV!4)9kt?v%hkuq;+^rGtisxZWe7vkCP?C`+DQ|3e2M(@#|+0siRM%Bjy?>aQCE_V
zifVGHBJ`W1)!h8q$CD*8kkf!fNf*Ff^EhXvRX_9>@4`=O(T1mq7u3GyS1CQof?KLH
z2t_2P&$xT9xmAKFuJ+I~Li(U$irjKBX{1=lM8Z-C$CWqDXNKJqEI(-@_}@tUmTlcZ
z&;zhbJhMIq&Hi(sOuYIpLl)ZzK&c&%x#0Pf?Z70r+XT^Q%c7#9Ccd>Jc|4H2Bgxc4
z=pPqa&2bAo$Z33+hsLKlbbm;<l&>~ClT`6tiBs%BVfFvG`s%Q(((Y?QK|w&J6bXx#
zP*PG+X#q*;P(m7{>mXqup>!kNozji8ba!`mefyc2_xH#5UgI_MGQ;znbD#U{d+)v0
zUTgE~&x)m^zt0l0m-8PJJDVB>rhnnmnF_IrtQae~n4vEudEq?5{;t05s<`&ynw{?F
z%-CQBy*--4ZN73_vLKa)ik;|cww!r4WQ__JQN)=6Bw3ZuNHB!)>4(X78?EH~Haprl
zZ0(_a#=5t+=*iEulVy9IGfl>D&Xh`f2(UO~7i9&?1TtN0GT~j?*tke>+A&O#oz3*)
zWEQ{B3Ha9DB}kW_@{et#$8|R>x=joq%EiU)3F+z4`0;IvCSsUDhx%v}sUu*l)P&p&
zde4Q1_&f9Mb>rjFw9$4#VILmCn-|D?T1VB=7|)b$+AMR2msryHYIW>J(EtzMa^K0)
z*`f^H-#YKNp)NOVG~A{cRj|WcdtQmx>`lGM;k~nQSCH~ZkAbqid?irvaFf#8VdF8B
zQ;8z$gd*B6jy1pkQV&Lyv5b1OrnL{}G4MNlM-RH*3!PYJJa&0wgWlA?uILpzgRv6W
zqkvIpo=jCb{RBVrw66S?Vu9@vN#g6K61_?7`R+@EDz3r}mDxyBN_HtlFPv@7678Q=
z%L8AXwmohm2se6K%;MWC*F4_<fdsWf*m)E+->>DtO}c3#b)UCjZ*1=kXevF`1I4no
zX#R-)%$4F}D-%$n{WD(SXb-YKjKi-`zlwI-f|QQ&k%9`A)gG&$41x4!UNU=`M+V!6
z!%HVhyz+@}*xnz%4Y$3j-IEdYo~q5&{dbG87K>xzzzQ`*AQ@w$EyY00Cl;UD=DI|U
z57wm%B#6@s%)`00xrb|hU+m{XcFKn6HSHY||Ee+17MaJLTMNDxGSna<zN@|RIPsQ4
zdU3bvfZ7|E98RB-c99b6?x^{u+`#6Sqrneac32aZE9<PMH?BmD(LW89=sveo*)v4?
zehjvY1p3h7`o_PhxvP1&%x4in&{OaNcjw(zxFZCXgv<wvaTBA!pvw(t%-hh5fvHuH
z?<ED<@xzmo9_S)RMlKIJW>La`QYeQJiF&D3EDS2@HHf}`JqN<c3gh|=RS#ve2jn0L
zvWYHA;#xGKvI;0UjRckQ-9IyyY-}1`5`6~jIzr5!Oucnnm@baS8{(H=)hvr6mtQ(m
zZ-hl2S_@WC`K4KhaJoM*7!&lCtD?Mbx^tsrL;4phBSAoyhkO150g8y54vw_78|$BJ
ze^6<ScdpQ^g>OxElX92Q`ob^Fwsi!|CaKkiYPvPn;QJ|_dyJ9@f6*Q@80Sdn0Cu9R
z#-pMHR(x@zYf8MbzEQ7|dlbqLJGB>#fBsMlGKnbP?MM&z>{w1q$0l#1K3cpTesHB?
zc|Gu*evl6{iGYpz=oDj#`EJH3bMdFvD%xF#3blJS`DJmhy-qcWWX?5&G;n=Dx&C6$
z!8vGT0MorjW~)-^N*6bW!KZ^K&?SdP-YWpo(#j_q8WQ6?Unnbwy=p>KO3d$3Q&X28
zPJh>H;<jEUf}XOl&pTLZ5E14&xmBd7XgD-8dTYsgV1PwiTUDtj=!TC({}8)wU2!zO
zOu8)PgB4?k3YV;++udQDFL*Gu*l#^v+#NJxTzKab)6-shsV74sXWOUQBdg?^BQ17G
z9IHai%xog`7>agt<q-PE-uhmDUUxPQwf5!N>9th;NHl)CRrn*qVQIOk_q{Ama|c<g
zo&e?{{R;7H^1`1D%RAjc%^O#?NRO)M(0~d1*>HIGYh$dy#>MyVE4exA9KU*&7+&k$
z5Mmb?NT^o56D6e|q|=frmpquKiwlKaO<i5>?I);hM;#D*iukD-wUIITy}8j7d7X^I
z(1#euyYNkFj%|_g;~p<AEJdeP)?ri5Z>i`VP4exvHZ%dvX-$t+HpBn1aDsQ4dEnKj
z0W!cS!E1mSdM6|-gCsB#0l9j5dFT5lTU$+3Gn139`T6;;nsTWf9dTJ+nY@$?{kyUY
zkWlY3f~9OLGkW)$@QYVnFW)6U&n!jPB%mQ|{?d89>gbBt$xO3!>vJV}F9|~1a@C_7
z-Bl0Puxx9FMmX7Mva!dRg8H%FztwJ32)<dz_$PZ^5DY{*s6-!{?r5X_CdiNNY;l{p
zk`kf=;c(RhlP@p(%MR8HC)XDejo+&r4u%CHw-|iLY&E6j$wp2!cdl)|;mkNxF+VuS
z*e;x0UU|Om`6#h7FaGj^>iJ;Vk%BXMSB|&v11xuRUPoSel0CZ{Ts##TN?nQ1&slJy
zq`tu5UwqLf1D=I?`r6q=5a9qh1&9x4#SFMdMMael5Xb#AV!u3YR(bS!bW{zJCZ=`d
z8wn9nJwQO|<<dCiKGLw5QtnYnU8f<}^;nlAi7#vL`ixoci72Zlkn^^hu~jXMj+eC9
z4yMyT4G``RW1mI@8|Re;{~Fa^H;rP44c4k>HU4!Nf3b~<%FOkhy?AWbsOD{&Nl{9)
zH7uK~p3c4<Y{Dax`Gbd4;-@RZ#E4dM(ySGOa<bgF5ho8gwf-(_@h-x-vPzyE6ElE5
z#T4WdZ*Lw^eDulR>z>jV<O>P6fyn{cF3x7nP-{?JkLQnUC6ZV+r6Z$QWxI3+P2AVK
z1u{70B0GVXk00?#J$oy0q;87o;7$qc;fjh1B0Don2lvW2f^2wc9V+k<D6%V}%1Oxb
zhsQ5SiHQ}6>?p{|&8^B^&O3_o^8?1WgxD>8lEST_`{~M{Kdw=kPyphBP!^(Z!o-%+
z2;p0>-(bfMG1IS$Xbr@${>U^sN5Ib?n@YmYK=<UCF}iW5Q1N4T3fz9e;<+m`BmzDn
zzx*<Ls*=KOKlnt&BmBpKH@OwiDB2?1$pZ+M5(jyoUccU-H=j#GSXMP!4=djEp8fd`
z!fOXgIL_z3Eg$7mz64y(Vfx<vI_&iP+^^n5pw^j6y0N{}&FvR)A;sUYJtsWgyvf|)
z`OcSPk*W1}28X<c`eMZ6liFRc_+|`WR?VL@MRqXe7AK!E?;}#a1<QR|E%!{YZD?$I
zwC6_KnKr{+Cs+h+Zv1y(mOMB3j9+YxBQ(lN_8LN&d+dDUe&HC(X(7Bqf^(;Jq&4XN
zRKnd`UGhVf{@<}m9q2SKu6QiI?DnPJIXYvrciJ;5IUZMYZXF?5tKEv(uU!zkPN8_A
zTuu|{jC_K|ixI_nQ*F%96I+hK)TuB<*0zmhmkg+Jr?S7$dgFR1UsdCF_{`GuYG*J_
z$zr>}Srv~9ksiidmRy~GkV8YfZlri|H^eIE_`V@8-p_Ytg)aVky=zowK^Plcq*`BD
z4BF{bSjy0Jh3`|X%B^{?z&8#ORrpU&umv?CnT5BN&`mnsJcPn+44!K(^zTh*qK^7$
ztZm`i<RYoq@o^mIwto7r5%TdLYEhlo{w#IXxd9O>=S!~Nze_ZjJaL35e=Bm#*$Wa_
zDah55`MvQ=N=r^`&wU|3-95M`$n=FcfOOOfmvd|<=1bS0QDEy>h-M1nnjL^l*$I$g
zLutid*%%8wX@&-M5stQ{Uu~nACb^mODt0W_9PEEO3%Cb$y2Q1eR^X4Gk6uZ<dp>8c
z+i2mu_<B={Jdo@I7x&H16|H@{oj?5rRKu^9Vox;UWXrStqC&J@GeBOo8)2zbSVyiN
zRbsMjPJPb1CG#tL!a`Pj5<&FQZn5ua$oId^`NtR<RXVYUW0<{WJ_!FA%_j7<vHm^l
z_D)rm0uIty=*GMk>S>hh79w=F*K@Wm`DwYOzWba#e_js{>wSkf*%T;3A4#XGYIk}(
z>G}BKNs3mlqo4MI%vl2~R#bG%#IfSheB$*hAJZNE@eS>$iQIsHzq3FuWjv^P;C$!r
zk^9^$nTo3#yb4`}MT(wSb`MB=7chTV^X!$<s}Niq*ztwi*A+bq4Y;CfeOmp_k)%x$
zP3z_EOSSNl0q)b_FIYE*vmZPptz%4Hu6#STbU)PQCeH3|IIG8xk=m7_n19p1=_GSG
zlvQQiW1g*6bOh9!1LLM-s*bVE<?DJf(FV__q(e>fl*=hLmYXxB@szyEV`ys_Y4646
z98*l=Xb)f2olnC#7Ify)wzW#%x+S7?Atq&+)J+y&bvs`nihWa_5v|B2CrRfymFi<M
zZ{rJtsO|P@nwHx4JyNx9Kv_%Iqa`$M_7D@%*7SYCi~KHRTe2YKE?oM{(ON;nEMR*2
zt}9<W-GAcA<kCzISMqTjqC(#$C9gLfOm_RY#o6lXvdW5uJQ+v)1QI#_Gczc9P@2L$
zraMr_Ucu{H&coZ$y!amyJ;sKjYI+(&k+^?T-mk)`U3*}({WYTDTTH9I)i|ll1!Z1{
z=Wpb>2-mi+{>LducWkfbs<?2uePZ#hFSeHJm%O++Q==B@B&m$X)QuU=O?SDnO_n<n
zTdV=iV3or=@{De(&as#5qMiji5bvWypGJQlrjR;J(LCU1zFcMVX}&2Bp6FE|h70N2
zhEk@TmQ6l74aG6)`4w=d;NxY;in;A`kxTn?LZ<dx#l@u1w{`!$DaA*N2t4}|TSzLL
zjrtV6%3}^ifrG0H1P|%*10uMeq53md_6<wqlOAP}s-n@DZ<|j=hxz?AHGd!F7jJDl
zADgW?y?`LCvBbPu=eG|Zi9MCVWhxnZ?n@~}S{uvv_x2m3-d#tROU3*gYAOiZKP3V9
z4ePx5bO7qFMJ;i<Jn9_zE*S~-JM$UpWkt_XotJ^Y=K7^5+Q2wDF6maCz5UNKN0H<9
z(@BsxTak5rhtWCZot6*?#%@z;k0Ti?4I>-JeQjRc|GiK}Ff*Bv$&u9sLoVxY&HJCd
zY!6u<NGm3NNMz}-fLx6rv6H^An9g>3ru=A~geB2$D_Am7y>3F4GetrC-r=X}s^PD@
zo0oWrzajN%WwR^Q?Gpi4Co0>TeZCA99+EPXKgFy{(B$vv4EXcdd1>b@k*}%Lo`7#u
zJil$#$@Q_uy>};HfNpxjUE8r=VTN3-z^bH-@hrL$H=CqVFA53x_pONA%1^20H5)I-
zV(9>srbOk#^Q)lZ_P18-L!s8UyNS<&A}pJ6!p1_$)e60%1h2Z*Q#^iX%qwhTNrN`d
z<n)2<uG^#l#WHa|D;t5k6hRHMc;`Ebg7Q;xb@fZ-!3-J?G9xcayGAH!irvJXo2Q{M
z2R`$a`oPkK)T7sq1uCa1S$4a|JZPZzT=Z(qv9&DGTaaN4^1^?kzI44W2wYwL@gM$(
z^8Xeq=@(H&i@v7v7b=(>A<YZDD3!<S0m-0EcW>z#hCOLMP?73~M({f4UmJ9BT=J<<
zGp0s0kib?otlS9I>5#E7wa*@`BJF5<LmAJFB!527=lTf&;_7dv$nI`gG)S4+z2vs+
zN>w(VGURQ2l%IK;svzF#SIYd*_Q*O!)y|pNE*q7bmEHSEoP8M*G4j>Sm~{8=ipoy^
zh0uF#T3uxw-rm=Y=3wi@QAYBq2m0_as=azcS=Zd!xBZrjJD}@j%T|6uG$MMRmKs-{
zfe*9%>J{Akzuw7I9LMn3tGoE_2Gc51)(tgwc4;oW_Z(2S*VNFLwjV<kZK&I!e#Efp
zU}yK1kpO<MH@O)1YGq1N%<wVMoj{54cI%`oXr0ff7<FxD=ouMw4@j$Wk)aKE6&0Q~
z_6@j4&tO}&K5);mO+%a&v%mZC+W77-21q>9mp`VCSmCkvEf42<!`VT^_}wim{Tab+
z-IBu+?c~B@Da(Ju$rX!$yc~;oxR3~$)GNQ4$aiT5L*6N8s1LbzhfnHRnQD?#;kyK7
z?)1z22+Q8}cD{Q(8dtHB@`?v57tWK9sFNQ4z0VHzO~lYA9h~sSk}<*U(7f-}49{cq
zvcU1;dz_WtRnr{X$)0E72lK8q?{{`kfDhdJ*T!c!<p4vt<5cB;$;;(?MUN3aBT5;`
zP@V3{f<E<Jv2{(@!f8w+*cXuwD|C$Da_YK0PVhxUrP>^sTlq=C%0e#i-9xpaS-Nv7
z9UECN*Qq;KnfWp{q&{@G?{?JL6>ZF|<gHouEd0_*#O-<>j0bL82J?7LJ3%ZTZ@u1B
z$3cV1D%Hol|2>D2r_J|`P?uxKMd$Cj8LZ|l=EjlAn)l%hgXqp=9UB@P*!+9TEF-rf
z1%Fig$q6<YJgH%;F}G!E=$TguMY>4(lXTR~ef#7D>mL050==-l_3o{8)xlgXOywpy
z?RPvckNQ}eG58<xVOP5@h`6;Bb^e58?jrvn$L2Vs#YkU6<Zu5@{<K1>dD2(t_YaSc
zr#|0?9w=Pk^b_ZAzV%VGqJs|J8TlVzz-Vf6^{c9!w3|ag;@G4p9}7}^pKuzBnhg%b
zX{<!q^jrJ$f^SAX1m2&_a!}*cH1xQruludz=2)COqt1VaxohP|tYl}}#R*aTI9h<4
zA<Mc(5{59lC#?#@V~w}o&y~*Z8yfnA^G+PoLS1uiOM*Y!&)=+rdCH3~D{jDh*UiSA
zXTW&^tJnU$8C-*B5AuqYO~an4ug_N^hx@9JJ4M~NUG@Cse|qLuy6kS_3GcIvv|M0G
zE|iH;S!x||cB~7$5*ETC^^KnDZG3Thyl(5nI!UZpaY+BPIjl%Z`AYs!d3E~03ZpZ;
zDJf-eLjGQf+G^3=Fl(sko*l+D)%TigxHTByt|Xz^=(LF>$hww)TfNX@{7sto>r1k1
zPG)u&`rkhH)t%eqr=}u4FWeoAX`$Ek+{^=sl2tKWG_?1<i8D1cW6n)5!P$O`;~;k>
z=~h~I%I%Y>hq}RaS;lrtmm(Rn?2~05IN4K<`(g%sE1Piss&^wZORVzWr!p@WDOWw<
zSKj%V3ki7>8irkLSJQj_uz>CKFGb}#Mt&VY!ghlnPm}o{k)P98sQM1_><Lo?Z3?;w
z%o^o(atr26gXOQ>^YhtwqOztP1DZ!^JYf%h^eQiA*;m1aqZ{L_`~BfliuFzR=;-Ue
zF4J<nUzwCgKef18r2pxviWTl$B|%JWb59@ApKRW)K%`4sAfq?eACl%Y;r<)$8=Ie!
z9J-v*1bFo|4L|qTf^`o0uJ$L>cfPSDn0spZ2=SGgt=pF%&d+r}vC?Y8mY!Z*elP5g
z*EBTZ^NlenUuPS`pCINlt*?-vS)!i>uhM3olHT3qz}j)+!>5?cLE72YTC<TNOV^RU
zzW(I9|AaKMRR8k(l_dyCSh8Gf5L-DBhKV`@fn`3XpvdAXJ^|;K>;U?c^ZTu|&h-@^
zsc28LD2v1@e|9Zd1sS?<8Q~AOwc|ouxc7)upd9;rEh68(gF%Oj;&XODP&Na%eTn&s
zoNf0`>?ea&vr0|Op#`=-qN1Nj5UK0SvL{1VE9L)3Ccy9eu!7TGkPyWq4N#o!Rxa7g
z@sSYM&AXhPN1r@Qnx@isT9k;@oJJS9=9p2i-tvl-O<{UAHUUdSOEh)F(Yq-u@gm)t
zq71_(+grmysCFOip!e;bf5TVrr)C44KHTp?)22!UcJT-k_XorAW-&}Z?B`!`+r+in
zcUd;^_i68z&lQTPTw%;P;ynSvl&BFF&|utX?xBxfY!Lwh`gP1I^rPd<AL-86j9-_O
zGzp+P(PQ)MQtTP;^kiJ2btz{6Wk{T{m1X{;A2^GPSoGAOywR&Nd3n=pA`)A`*|$Kh
zx!zPj66vnA+wS=Yo$jgSl~2++f~Y#MK$Cl4$~Sn<7SDiubrDhu$RZwSuUJeMQO<SG
zc`~XMmYg`==#^70i4zlb&FuLHHN5Vl*7n)?p|K!-dsN_Me<Q|qjSpIcqAV{i>~9KI
z{q}}!p~+<e!YROG&b7wj!iyjU#~FmsB+3oiI1H7s?VETejhO-K1=b7$iEiHHx^+!8
z+gfvv8Fc!8A}8~Q#b%~QTiJxZSyCPoeJ_}G|Lp)3y?c^tJx->3vZOdE>jz$j2bZgZ
zC1LZ?xQsj=9;MXoBQlbG`}QkfA1ax_z&z5{SF&z41<is{`-LcPgi0_Pu#(lZ6F(}!
z4!6s=C_AkZyRG{`N=C8w^rP*u%r??!6H&D5`udkJ#qg)OSsPS!P*m92+4m@?BP={J
zj~Xqlxt1W5D*+0rheSe1NT_7Smp7DE8-s$!s)WH52z;RW!VCPA2X`lzuTsGbx6?M#
zObe5hxjJn*+I4o|QTXj^$#E`Mm?Y}cp#_@<(9Jb0%nFCSX=r5WLw`D@Z?EdOD>BsD
zzC4QqmJr48F-nbN@~r$Ey+LzVB)P47FDJWF+RWZNBDe290AL{C`(rURo{+4Jt6hC{
z<#!E<y^XNg2)`;<As68!cIs+sw;t<UVpPgP>7?v}X>^MPAOSDH7_pvETE27)`e?vm
zd2G8rDqh@FUyrioLfI$StdEKvtdC`TfdCe~hmTZ?yy@kV<V!oXSOEw|(YknT*E6gp
zfZ)k(vw91dvows1D`^;()@Nou!Vkdnfe(Q((rl}fFgE?S;J64{5qE)#jUX3ZCtyPL
z6?~<DksY*;vgW(PWtsq8XXB9~JbnFzbYCTkkMH{T?S^Xih6bhKurs)8>Jr@|tny)w
z_epDDWrLh=O_by0Ii~ppI&6-2Ijb1&gA~ST)xS+G8_iSw`1ftRf~1QMx~L5;&gd%t
zijMyE=8SlQ%;9(`8T}9}A=AnQSy=o)kb4GZ4*V#CvZEse5doCU5^ykpm=XXgZ_thd
zoDJn{3zHGY2OHARgDh-{d7_I0Fv}HE7L9u7dp`-g1wN`(XZv?xK~Nq9gtSsJii%(~
zh9o2;c)<VLL@o}Hl;y6D`#e11K)r^ZWwk%G#Gm%|1_O3kMMYs4d<El!D#Q6zP^BL%
zwPt)B!Kn+24_>c8ME*iRMFY}LXD0_c)tCgmPqB^aGuF5AE{9nHh%=pZ^Y7y;K`Ne0
zGFeaLl=M9kn{#-GTfAHDa`7UYOlDI*-e6gHwhV0JvNE7GV!H`n{$dF=(j^b<p;f1U
zv(4*20wPn53yQSv>7ps|ZrUc$QB$c~EODsci3c3Xpio6I<82KP%6-2}%zab_stX#>
zQ`8bM44ywn2Vyw}On$6RR8i5=CVQQLH-i1?Rugz|^6m8^a!xz#TuSZd6~-XL?sR^%
zphRqRur`8S4G08dp^mwgRRTa$FfRdWU7-;Xt&G{_7}bEvEWyY~gY*9We!+*H&ll$h
zRYCh;AvXi2WSM^w?RTn%$E?>!!(c&X78aU7flX9Y^e#JlK|f!{jr`K4R+9Hv+3@B>
zAYS<)j7+7Ug^KVw^{aacH`R}7jox8uX<y1%%{q0Q+h0UujU%#8j`nyR_9Ho2*RRNc
zf7#_WiU<SwRqe4?UaNkR!&`0UclO@35`CRECRc#vh_cvq)pR+UOFIGb7+y74nW!!8
zPaqc*@m+u5f*FWyv#mBN4lr_1vmJ(h7xNKjipm+~qk;!OM6>4p9uA`|S6!#-z9=!Q
z_rHgC5zYG!SC;I({U>H-X3@_3%4nR%gXZlziU)Q>ZbR9}u!FRo^!=^+D+y+TihlI)
z)YMW3>hJvnE9nO+m#wL33*XsHpYE|I{yPl(-<p)9O-1s^KgbskAv6a;p4nj<Xy~DC
zs>G|m!m0b!r}1sABXKGP+5;oOpEF$|RJve(0N_|z93Acte}CC7p2>}TF_zApC;?}l
zi{pW+G;bqz1|`O5Zi~3#cf~*kZ5a{qqilypygj^6%4If8;1J1OJZgyxjH5k`!Bv>b
zKnp@TGC4zVTfw9`-KGtGVh?n(r=Zpk4A8L|U0gj<y9W;*q#Fksr+%g*!-XYRG8E*#
zNovN0!!q~6gSUKyl6+<J%eSfr<y83DksdQgCC3hblc1`ThQ2K7hTc5zbKSmN>F*U=
zmVER{+~<d)m-}}T5v5FP|8(#QMR<UCQ60_iC@L-8)Y(bk3NQk`D_p&Lp|qhP!!IHW
z7E;S}xt{x4`1tq=<Lhf{?GUT%=7N;K8G;^S(xqqVecLdwgfeZEpH+^4s?Sv%H>D3E
z9j+_b8O!s|Z)2E+g@p|~(zEspVG<?Kfa~Ilnth-jIGt6D@V6PL7Q?aSYiL)&#AhD}
zwXUphOuTZkB3^JhT)(zNb_{-wHO@dq+_VnZDCQOi6c6~QtTFL>`dmI(tKRqPuUR)v
ztO);&KkwEd`p1<ur)KVeA;=35m+k5AEGW+#8tG{7=J#m^oicrGN-Qk8+P{Hf&1$py
z_PZ~l9#viaEclUfEpXIO@Dpm-3Iu9UV>d9KL`X_10uk7V|FfST!KPAdF*7r%3Ja-j
zx_>}`9Q}FWSNn<reKcH$&Bn$?|Iko_Jr=mRxY)BT3wwgZFDjt}@vS|Sl=8>yXSq41
z{L%jf6TIYC12SsPV^ijoNIhvdY5gUq?2OK?Dc`JqiL7X_qQ-t^WyOvL&c(Apa38f9
zF;VE?UhB`ivpI0z7*8y{@8<2>&9JeaS}k6O^8@VUHamDXdOuubsRuP#gRi)?fL^jx
zi$Xxi$$=t_s4<~|`dx$GW0U4pzgzC^?wS8MVOqRi^pZftP9I=Ja<KLeoIZO~nZy0f
zW38zxWHF~_Sd~TWpA7_tktrO8rjCZ(f)^*Mhl`Keu40xsbmWZsUv1j?&|ONeZM$#q
zgKy<N3>KM=m0Ez)1Xv=XCjZdwmJ&nK$OWwyWzbM73{L1$qN0Ba#e>i>vCyuQk}4%{
zJ$Uh%&L}iPqar+jR=OK_dnyOF*$uA$iqy2S+ErMiu>ANjEToeWofv^ek9({nfjZyv
z1@=S`dGGZl-^!OF&_qydMIC>rtrDFzbKJ$Ak(fufUEZUW(ZnO&v@><q^GnvEY31Ts
zX0A!IF;EN9Aeepu0^pUybnatYAa234CUgZ-X1-1X&i2j@<2nmH{S+)OY7YR@u#Yzc
zIP*vdEacC~NN)`=J=bR^6NBl4Z?w@VkBieR&km|C@Sgcm@*hrpp`v48zz96AhBRkp
zt#~4<xAq<^3^tY1bys-HP~%5%V$uYB1HT&>7xI)!e!#%s1+3+MdcV(KKS*>x6a|aY
z*VHT=szU~Po|g$*-hp82TS^J!zk9slQ$cZn@Oes$RK1vLy!NNR=E9TC$V|S;zzuIH
zXOeu;=UO!9=YHq9EcNcb5<VlVuDDld-c*8=aP}BBRsDa&Ro?aD372$R=G@%e^8I0>
zzHurtGL^N}>9o@o&5Qf=^xvcf&jKI6Ym|SA%v(+!<~4jLC8VP2ggXI0s~v@zy1F{$
z-<G;j9}QAMi|(VUXrWR07BBH2JX*$6oJW3dRNm^?Kl*ncc{D)#noax2D!jlH=frXn
zI1-zYPhKeL%VUMP?#coK8ydL;9lxH0qnNzMd_28LN1F4?%POgdPwlsnS+b^dc?OuX
zd;zHdI2|RzAAm4NM@JW7I2-cWcc2*vRuwtp|70UzIPw@QvE1C<K?&N<!QqMBrWR~M
zAxX(=%*@QzzR8N|%0T-#IoQyz^|;DzHcsv%`~qGT`_<g0!D6!s1tvlmnEP+^@)`vX
zLIi{w_zA2R%J^gkeBNN3tbe7P?0!<FTK?^Fs6=(hY+oB5v!hOtoWSpXWqCg{^ZDM|
zUkjmG-y43tgz4l49Un0171mVy{5>1{wF*~a``QfvM+Ych{uiyOHxym?O1Bk?+AqEV
zi>@YT`F=)e%IK0If=!L}_oA=z9_#(k*z9qU@20VwEhUC<raD&~K)l9-*&<@xg1%Qy
zPEI}tMujRUzWjzpK?o^*x>T|0*dfH|6>{lJMK;4;I$<9&-9qxVZ^|QZ#^5PVa)e2u
z^cO*uD-^OlrRB3XBQ9_kFJEloZ_9B`x2Y)~AmX{=*J2c1tjRR|RqLr@PTR&4Bf3K5
z)wRYCV~E)|Zr1(^5hm>2Io{Yq@n&EX@BM-40Mw;se|y2bwc|2fOgYm%#12NT=H}Lo
z(8c`~y=DL+oRi~g&vk2#$%O{r!a&5%&dzN<J66baO}=iTR&wZc<nr`*FY>*7&{&l5
zG9t2rq5jCTUWhM(>kacoTEmh1w~Wv{*V;C|9Hke6REtySuM2-Eya>hz5eQ!ifLUkL
zk&dLM!%r{IR$L-jHvcWQ>wQ{)3|K*PEq1`8lnH~T-sjKq#;w;hgm2!wA#NC3A1lj$
z-+Z%oxe9*|OdTIvFC+56=0-!M>@b6?x3}EaFNGS-NRdpuuk+qa*Kz3b&WuykwP*ET
zH40sZsjBvFNl->+64_#Ne#*W-RmgdL4m)ZrHi~g~%k3M}RDK65$EDHL!D1`iiPy(h
z;%f|CqER=CS~xO}bW%ImfgLayeDxWyl!huCxn__izihgBj8O<hFf1An8u}F&yi>KF
zsJaYzY%07I@Y~Z2&XnJUR@SdoCDn~b9%CP}-Qm=ooLv&yP$AL&BU%0jSckR0*s2rm
zhLYgCCQqAqp6Pb4a#NeuAdA{%<&8^(^2NC7|E|LnUy)^#SvfC3#}UdsBnYPrAqA`)
zLK$ij_2<u@vvYIG|9!0rf&RMFN=(c0?~89=9_$^SeWWARqjA0Z`Qqd``IzBb^4|A4
zV@49trv7B{iIoR!d3czy`o{PNf-IZWBTGV3-^f%)-~an!GWg>4!Q6NqUOv9^md8rW
z2{!gfP^rKF&C0`RT%+U%uuk%+(2!d?*d3F3@47@AWP?Y|Q6NOeS>65KZjOlPPRZHP
z4e)@}n0b8Drv68PAf*a-O<ns+%zHK_zW$r8OdFYL7w7-`LNfkzmOL9U6!J1rQX23w
z=_5Y7`kF?r;Q-Qh;^GSsnB_Y_MnCpTiJ_i&_gd$eds!7N2$f*XC@pBeW*A+@`?-!3
zzWGDvUT424LNdDyEyR+QG3;;TAB{4%{LV{MhYt1c8nmeeakY>JQ-7TAl$%Y|n;GwI
zE$9Bfi=c+b4wWb!J^k<HeOFgBIFYcQ!FGJLb<k3~fxwo`jPivI@+KaRxf@mzsxC1n
z4`^MUqOEiXW>?9jUMb7x#nimzY0FrpG867ibF+v7lb3im?-yPAo9CKlhmV7xd$Z(l
zVKU!z^S>|cT6|kkeTR%pTv3q(Eio|>3K_Kf_wO^{l!w3!6Wt~WoYAZ4imDawJP_13
zGsa9Nd4~ypgCq_0HtV!K-ImQq-@LL(bGMw@#MG}k@tb4RJCX$=8GofqH64>`&Vt80
zACgZy%B_mLYV%RM8y?-)h;U#`J0I_A!$E@fURU?Vty@>BDQrh;;Iz3^4Ts?^uvtgL
zCZNn1EQb2wpEnKhNMVCEPpNlB5c4~;#DW4MG!YSzd>!KV!7QWmx@Pt|s7}d012(8H
znadE+wE<UN;t`1O*@Kj9dXf$M31R_nArFT`5Eq8>GfK{&r%-Ie12;yi`zkS^Nn>&~
zCRtP<RMbWmizYsGpTaZcv-@|;%eW0#t%y7E?Ur}p%Eldk=4yYkL7zver7NFEZ%#G(
zKL|<a>xx^SK1%zuSm)$BOQlx**BfXkN9lKs*HF_SUS3`drd5w%?&D;mYGQZ^;%lL#
zFd^`+7d8oapXQ$s2|LdTMW$!-nqa@`cAUQM<Q6eZCt(q9xwD5B#DW7J$(KTZH<F=|
z2958ppIdLtsroC`M`v?>+z!D^Xi_g(K2OqSx>}Xb%il;)AA-(-PKG$FFsT$7H}7-`
zN`n_u!y42{?_r-Cu#<olA#nA@d4t-*KWP4NxG(m*rVoGE!PO5lknCQehua|^F!(!S
zH!gmi2p*$YeXHb|=D#aS=%FQMKsgjX`{LVm%_}DAZ3IPJ*si~#O;*@X|9VlJl2rKn
zkKUg{anI)xkn)b5U1-bq5{e~6i|POjW6@-EaTi*b^9!{91Usl9^)fUjJ(o|2!K9SZ
zarH6yQ;LG!yRNP-knbYKCS<PMYr9c)bR}KmEl{{5=}Y;0F-jd=$0m~A&M#}d;7KB;
zUp+at1wxzdC~fED^o(9aeFI+d*$>ggYNU*YIO%gJ>8#Za^7gvKZ@ef5J9qEDh`#Ox
zmuk>Z-J43*qT2;l$0N@sC2u^cdN$Sn$i+_kMrv{g+v0^V{mH+8llD6^PXWCZB99Vn
zcF|A8oG{mu^m@)x|6YAzh!fq<piU?6=lZT+!s`Shl#r$JFZ`zuu75s8Y9lPvxGg^u
z2MA8bCbGm@<TU$9TevTAW(o2N9(hi`zHJ^NXl0(Mi57pABpS?oj%Qir!^2e<%<L~j
z5=&OJ39^enNv$go$fIXQ0l}KJo^;$te}PtT24Z?lQ`~Y03WI{}jyKtaS=90CT-YYx
z8f#ilckfmIxYN#SM6WHh28KP!tf(MxWrKl?q9i@bx)hEZwh`UBMspMTyazCRVV)rG
zs6qiRI^BT$AG&g_sBYt`s>Kftm4m3))o~BGBfm(S%axwRBR+M6MHHwjaL*UEuTWGP
zJu9P({+ESJbcP8Z)Xy)1ddE>0#i@_%(HmSGbs~S4tkBBAZV<1?@48U$KunBD!rcd?
zraPbCpzn<gy}>gpbpQ5csO&v1e-lA{zJ0)FF?n`|F{(xB?Vmm5>Swfkj^ug;o|`Je
zPJy?k+<S892}QVQ8=apdHlBCVxc{8cbIx|!5i+`BnPy`gH2k|(?l}xZ8hdFV%^qf&
zl^;|&%3VsgJF(nro_$D_4K2<uuB#cZA^Gra#baFSS4II5tv;<hNoFLSsZ|3tb*Q&6
zT4Yp|Nu&?zHcV|4V`_jAJ@6oYyc*iu&*R-hRQpZl*C&p6sw>cM?P2XZpOy7}B=QLq
zZy4JUcsX0)JmrtC@%5AK%oJikc}p^=NiVo2?fd5Dy}`h55i!^^RSq{|Cc{U_sMq<P
z3p;srW5mVy+1rm#VKYT>BX>XHHg$xPzW66nLKekgtIqzjoX>G!TthLLREXU>0rP6$
zED8PhRF0l-szrQ3U=HU-;Iq9S4vnx0j!&UN)d!~s+A3Cf+O5cFACYORn~Cyd0=gi!
z?&m}5DA`FSQqBzCH`!x;5i$CiTQi(gZ#dUC**5BMWdu%dE#t<}+$dXO{dA*759@!&
zgas2|N)b^}>banIcIS@XeGh{+W^#c{W2^Lvw~FyezSmRcx_?`FeZ;lc&RB?3o}LHs
zg}KU+D@av3t%%E@*lJSULkiJnNLGo?S0!A{;cH)i?WBxfvbvV`aS>OfMRP)(|50mh
z8Ld<J?!X1<|Mt%S|96~nm?96A3}Rt{fo9?nC{t4U6l_1SlKH^+9n37LTa18SERz_d
zVrfA!(ecKIs<B8?%mX#LdxHE*I7oTiIC&(+(qosq5ECy2e48%6Z`FW8b7OP4HU=zR
z9I~E#92lq_D2}Pq-1=|Z3XNU|Z27aNr$T{Jia6+!tPq>E1kwj5K-$swWU)oiDKB4W
z#e!F1E>tK~A`!UEmuU$|zt_SrqWJgA5~`mAFD_UYcY_;l2>n;IS+uK>e$OZb6I~EU
zaEF;})nr)01Ym@*sCP1?+iR%@Hkt3_Z(7>j%c$?`#N4I%lb^%Nm~HakN|a2!Wp_US
zl>wzf1JG8fHekmvpZ#-#o$Phc2}DiUtT0(v)85|h&ko%OMVB><AfK5d^WrMx=o8Ma
zdC@I>79{{|G6N#7lWZH^5fygAA@6jr9G}PLiN<y{Gx!zs<Ucy;m9D>!z;Di$%qJ<X
z`p%6T+G2K4%Y^@^)HYaML+Qr}VV20BP03{zX=22G<HWo#lgd<1u5pF&17Mq>dnc0z
zk6LlctMuu&u&j$@^-%2Ue^!+&yib)qlGoN3G=gMhWiu#LAga$##!<sj014bi`{|Lq
zp@(iWn$>N`2;zLg+9zvRBGapLI3xLYWB-=l71^HDVvhbKP_RByQ~K{1Eqt;8D@prg
zruegGu7xk)<X1p(!*u%B8UUoW+H&pKGeB~+FwSksp0uE#*Wj^4{gM(F?h*q?8`V#@
z_bm_2Fr5D8y5#qMv|{DG;W`pS^{H`u8KM05qO}QW;(wORICAHF#H1Iqn0=oZYi-le
z#PVa4`D}66_Ey$9iP~k_dh5yXa(6~%W*Rm&!YC>1XV5OnZqNpP)y>Xl2-W9`bF>w`
zN1nygTEce{|Ek_`hd}TCIt<!$tfx_1crh8*_j=D_FuzVXcmt(mGh1#`SamZm+qr31
z){dW~TjJ`S{;01vD#!$s|69q>=J(T`?TpGrlN_@C(`n@S7Jc_uS1G=h+TJ|iKPFq(
zs2t#yxR+743??)P8z3oazx`o2dU|zL66#iXoltJA-6a=i8;E3#Kopoh#}BBM92Y}i
zj$0{l*>$!I&3ik{m^>Nq@rP1fXLP8O-D9l&gz4F-+7eIcnfV3#wVNbyzP)4|2bC9>
z&?(#}UkXSHU{&e5cFFJII}m0_xYIXHS>1TfKl}vI>3{Icf4uvjZ7|)aD695o^~1X`
z{xbo%^i30IT>i!my}{OzTW1~g>tVSuBQdAGe+3BKTYUFg1t-H!YM<I45~wDd7Qr|&
z2)zR?r4!Fx#MHSU+<TV|r2&D06=0Yr#oGAIn;ZFh?HF2Zp^vSg5pYlTrqEX$#9f2T
zN19QPvwRQln2Nbp`orG_irUfY5lZpbKNK-eET@cbpaE9;al)<HuJDJnRl}ad%2eBf
ziSoA6fnamvHOj=bT75C|@1&QWeu|)&U7nJ>vD&U!^vQJQ_+!(`Oah1dKSy23R1gvA
zlbL4phmnI{i~<yQZ&&@O&WTk*UccYGx-4w&ce3cRA-R9S8s(v``S@M;GY$I<k<z`2
z9^s(~X`tb&J)OAaYtp*&`v%7!-rpb}t^kGOwOiE@W|hsZ=$H`hlzb@I4P+YjK{F_^
zn$e?#=6=(txM-EH@r5}>Q~zp`u{8c_^KZjQKro$SO5DG`ah5S6TSelq?`1TOH-CDw
z!7n^uaW=SLBON(Lh%1<^$sv{ahOj&PL9rm08w#6u{+B5?eKAzFu}Raq?=zzxPBTVG
zcUiTC&M}z!?K7E6*m*B4zT;{>Hz(*-vY4*^wt22L`7(@Q8smL43_>f@S^bUTY~Ju2
z#wmI>l8_*M<l>9e)&1b>bk-E>DR}C*P&%b2l`Kb9;pAsyqMT}I&7yFiE!fJnIQre{
zROo)+Q8bG@N0LvlGnd*1pX2xObIae%sbh;63V0<QYv>zb{;>z>hM0e;)l%HkkMn<8
zTP5`Je-FAAKX=FS3|Z)m$%cuR+A(VMXSr_og{;R~i&v241+)J1AYLoug?mxrqCXo3
z;(m}R$d@@|Dx?xFEj{`4<lBdVXSOb9B~PT1K1_6*i1RzhD%0QICcu>!l~?Fn8WeT8
zhaSyLh=zt{9ub6`Wi;f|GTto@E+vSt7&kw#^2sJoY72ZYyg72`csBFPZg<Jkxwok&
z?NPkNvh8cl$jQ5xrkcK-E~kgJt<YJ|DlhNgf2F0RH8ltD)$cNLG;WssK2__Ph?I5{
z^*(z$7VBH1a+z55gzIs#S5j^!BvlMA<(*{r2T70r@;%s-4@e)mzK(&01}#mVsBrFe
zdokHBs)B)LGA~n`zOJ0)xjoP+)NY6wE-{QQli*86pc}88sdyP1JLaQb8fwuD3zy+p
z@x%7r%P;}pHA68S+<Xhcniua5rzAAZg8B^mpz)w=7ul}Y1M}=DdhOx#ckezk87b;>
zM871Z^go`!Mn!UDB)Rkckba&@CdzsK=T;}@!O{-pQ_Ghqr2t7uZwv$jqc>BP>kN8J
z`V}$MOEqQoJ9>Cj0%|^F9R4veG0Xj#>Cp2v1o-2XFPvX}Fdn}L0A30_Qk*MSWZ*N_
ze=ksGYF3?Re7xm3qO=(uox=s#&}QD!qlh9q#|t?j9i6zcR&*_`{ZT#&l`;|fFH&?5
z=wHKQ!%4l3WG(jZz2Dlgn0P6be6xsBzINZ8$C~fexB!V~Uq+fr<~w2;jzQ-gc)Vz6
zEe)T>A_|ZJVb52EGMT{@t?ZvehT=#DeE-I=toPMUR4VKgjg6|!4PCx;DMdO)Ae_fK
z4LXW(px8#_I0`jgHVhTuQSu#nuXOh1bF#2V|3-fQK86W&4rtP_ndXnI`6qy*a-qqH
zA{b|}l$$^oNu|P}5M^((w>l&TIBJ6n&xP?v(l%a`VRrba5_4wy3%OI2)G##9vp`lZ
z&`wg!lAGLN_NNFtB=>At6J@s9fEXl0(4fC}?>UGKu#~SsUCUy5`Uw-kW``c=>rmF`
z@xngcsJXvYC*jF`S6k>Rd4@PUiIY~(&d!$nCPNbrbh`TnGSA+A!TvwNKh<Y#%&5NW
zu$Uz=6R2nZY3$OCrah5Q3U(2YSV>}19L|k9v(<e<&y$tKTy|l|QJ$y!GwBqyh^}7R
zd()cKVvjRVe(#PWOr;8DVUil2GwPxQ{JTGrzy}z>aFRQl(LC1TP=|j9C9PhNrjYeR
zp#*qj?4od?val|lZ@ujSLN$Q&%dj`)Cuqhcfl)xql9geD7xD5~SurSMj6IpiMu~tB
z2ps+bg_Tq=rl6;#6#*cjU)>fC`xs|L5%uq2{!xN&T-B^15lV5?8<5{AAQvOxOhiPK
z0BUMbuS<iLA4~U}U*J6hJK3MP>eNypRrZx$zVe3;AHuoJqD{-zIDKwC76Lp~Fg_^b
zX9@o#9v+^8kdRx#X%z-h#$SHbbgYPWMZ2doG>v0(Xsp+h>r$&4k@ntNKfM-hTTv30
zROb}=0Zp7D(IxxtPmks@rTU7LjS5ZrRSGK#(;Wl^W%MMLG&+hpm6LCGI>l!6UTc1s
zYVoU=TQp7QrzN?pkZ%Ex0`|aae*$(>+I}Kl-$lg<@mr<@r*%Hmy8mjsiN6;7P{HTJ
zOgDGsCk5iS)WXfQ$bL(30Cl=w3~vkLPtYwpIzBc89W*jQmvRh@4&!vrkMojZ-GI(I
zjR&VRER9!`HTqF$&dTkJ_>y-4&i+34ZGrL76HkxRoyvo}R^}dnqxw?b3~@!3%>}En
zn~s*$A}7bk%P91T3OSAaMo4Q2;92Q<>54eJqM;#TIOR0~SX&BUC!WR1k4XBmH8tT0
zGgywt`w6JbmQGG;?Jf0)05c#071pH%j$VQ|9rM~)`t{Lq_hL}R@eGmZww&4}u7VVI
zof+HpIrsKz`2J>7_(jdNr{&aRH>h8Kcu+AcTACcp{_8V4or**3^YgtIh7vX7DjbMK
zWo6~gn2yqMbbEK_`CnMO`7dQk3IUZ~0RjdzXSBklxk*-zfx9pHKj;`OC4QD=5?oWQ
zRFBz1UVkg|4IFrZT3a-__}+vWfBB^GvWxjU<Q9#gq+Z3v#h<U;zAx6W58Bpi#nfW9
zW7bMxjZ{x9-{&&PFSNIAaPe6!B7fOx$1otW@;%sK%!CbDG`MoIyYxT|>@bwRHHEX$
zB0FU(S!L^>OhgR(Fhu_JohgwifgLohH0_Rdh1>Oa;9Vxw0P(blF?D=2_fQVQyrR=e
zm9+H-&V!ExC-=BWk(@q><va8~M$8RZi`8jS$5UIKvsT=lyq)uirlR3~gd8;$%Tg6y
z0`1CNr!sj!g0FV>;Jj>c8c)!Vj16ogAtP`%Z{iK=MqNjauxB@7{3D!O=5j6bx0^O`
zOEw+zh>W8Zug{ZkTh4xk{397eKIu3)izMF-WNTIpu5g)-(utJ<u+-_W-lwFgp7fMD
z6%LIC4Js0Zghs5Tt?kj)TciGrpAW~(>+lk@-a#7b0N7bC3|3yo!fG)rguh6#0>{Dc
zF|{d!Zj^F!e`P=#o}7OIFaQ`iI5?0vonT)UXhr9I5qbMIGz%myZ9_1t(QG;<QiwI%
zgWSp!i^D~&PbF-0Jp6Wr-=%)eF1q-wk&aZUns?Oo5Xy19t0z6GS^Kz`*fB*N(bOd2
zK%-dtNU|B5aBoF|P{M{1b@!8QDP&<)f)dNQ6wE&FI$PfyH=k{9?I}HzX+Ehfl>j|c
zg?M2+Ce2E<Ccle(H#^XK^G^V&$<nrWs+s*d+0x?T{=3qXDv@8m=4E?xMts*q$`wWp
zr+4TVNz9M245?8hqEx_Y92TN%3$*L7bl<-xGM!b-IT~4tsJ`tp;<$-*>Cz&cBuWsU
zgCZj{03`IG<UjU~+CUvNkU&U*I?S6XYZU5}%B72`<<z<1E59<J*?Q6Rz&epE(TU>9
zesHu;&}WnGmIHl5L?_(^XLpBb|8Fdnl#*%i>t#}UYwi<*ES<22{KSlSD#`6^N`|$4
z#i`b3d1rk@jWRh~w1N?+o#2(NJYpKob8byBy!oBX_KwQD#myoXnP?`tOr8S<IkY8)
zfaU}cOp*2D-w#E_^?NTq|5>sk$y#J;ZRzW2VwQTI6M^{4w$0BVjAD2Cabv}U82rVR
zI-wR}x11G)P_$gOQ7Px-R9;(K`<^pG_eoe_M1*7i<_O(rrJGXAM(mrv4pT6hN5*dO
z_cpqjlhf(#sdZTHG)j<a5urz8C`pVcHtZR9mUTx<mAk6cd#+mdEcIi_O*z@U8Gq^q
zO8+n<Af9F6G^~p3pg$$Aea0-<wBn`zZZk_&kjXXF6L%fq@L7qgo;0Tz-~A%%YfQhx
z9id-EE4KVMGp0<OvCxeq$*f{8_~8dA>UbP>Q+R;Rm8Oth>Rbsb3=;k%EE-Pz@ANvO
z*Y@pO;~;tRPl$0jPOV>PixJfH-yL^EHd28el%|*&4n7^Ym>X`yZNiGo=-Ch9Y=4W5
zPyQDWVwe7RK%2@2*_D;blXb@ICus8Kf-I`h$W)8rQTqGV<rYK#gs@gcBFCN4!j+(?
zcLBp+#P2aX*~J<WH$QAwtCW>mIC4CPq!3b=nBs;s7h8Vn<eu2M3BgB5a{Mk;v0#qr
zXMD3`^&1?9>8*XmK`(-+<zOQhe&`W~I5Pk^!(Icarl>y%OMSBp%LYX`{{%QUpDiuD
zI*0@kgdfN@W?P1p6=anzGl+E)2%i1|Rj1&^NVgL>AiOHhKo8AA+JaKv6z=Smh{y`$
zDvDB4zWeLQSUMP!d;sZOFmjD}!p6n{c}*YuGM@yROikuU1?`291Zl*;h4cBb+g1ms
zG%o4F)k4w7=OFIay%i`w(D=`7m9$%9eo58K#Nx@*^;MLzmt|;(i=2INr(iSx^p%&b
z;IZ!`#m*u_N8k@qi^2KPKbZTD`i#)D6j7Cz4?Ko|8_}KwX_^7x4VFAnkUy#9>-j(!
zlV{!ueq~^gqTzC+D<%tIFFWLxwXcU)ZOZmHCbE?@?B~2m@KJ#|@FElh-=y%ryzsXI
zMJ@iFHnvxs1#I_=4<?y>`wSJp?LYy3ANMf@8m5@2R7aIeLxT$!g09!xnGvslWYCc7
zW}j9Xn=QqWUce%ERZJ|6(q<}R9TWA`ydcO7<Zrb9m9te(YtNsM>rZ{RY<I$~d?wW4
z;@^|%Y?Nfwb@;rcT}<iXd1p!;@fnL+m%hcA^?;baz|kzBST|rN_XGv2EF(5hOAkZV
z1vYCckP0Fj-Xmn3#wozJFo42%dz)dPFpheA^ke4|%&i}<=6AAdRlDl=0pe;tH3tcT
z^-}jMTvBFvH8r)LaQIR{cHTN({QJK7iqahPpTY%^O2|56fuOenWTka-%4gK_gF!WY
zHrP^P_{rsx4OJP$oB_5aoIVemqcx3?SBEuNs9QX2qJE6G*K*OW5cHL9O<CZLoJeoa
zv8INcVeLMfZ)>|$Pi0@v?#Ln|`v!s;7!i<lXTw>en83dMMNkYqE9<8BrX@&I%<*W-
z$ov3Hq`0nlk#z04&rc%S4aduWYanOVu&Jat410g<cGBoa12UNiSdu&fOVJJZZ6@>&
z!rJvfL|Nis%r-_bODz@3Nm5F)Ct<t5+A~}o%rzV?NG&ffw`!&iF#z0XI7izZ<u^ny
ze&@iv6=xh??j<0p^#i)|eN?g7R2|Y!ThG>5^hqm@G!LU0h(d8pRcFtY6X(R9;`e=J
zaEUMavx%IlKMx9(+kcAlaJYjURYju7X!n}*s>D|*C$dZ+vbZiM!bUE38mK=cRa!@v
zMHpCkUn?j3HZQelqRZziV&XZxFBrD|Jzf_S^8vb$)B->{{jFbN0{jwcvbv@w5nz>-
zP$H*W_lt>*^H9(JW~XzdeVOfsOMeAo(qWC@0O3v;5?sLz3;)@*wM+=qksNef$CSLx
zEP@gIjzu7q<%MUPcq-fm`t7MMr&}UQ$rv(M3yuEXy{J_oJsCQ8b{YA+*U8pNKQRAP
zI{;sgp60%gg1h?a0xs&7=tyIy`t~+kHJ8rB?pW82J&-SR!3`t-&0~!if60@grhCXI
z6y<QOcP1N2CP+5UGM_Sil?MW{AQCUFm|iG)f+c^AIdihTrA4;N<qYRyZDS)F<grDe
zNbE+P7C_VjY*Nf03!gj%l6wYL*1v^b=<*B$w&n8KuTM3?qv-Hb?%9WRAhH&Eb6heT
zeX)nqxK<y!B#}=pF5;R|{#4z;g`^v+$UMFA3F?Z`QhpfrnWT+;smy{u;B}Ci0dk{<
z$F0VdcSmGz3qwuW=zTjLs?b!$OqK2ZeMK9aA`pMfh9W}2pB)$!u$u^!yrxSE+-%{$
zZ29x&M_|>=H~{gxuuGQL%QpH!iN*>!+|=&Nj;RXo(8~`p{P5KE`-_O<+~vO|KMMrY
zSreEFt&<ai=+lUA#Ew(&5;=bL1C!0&QpJ{!;+lcEf_CCr>_;i(k~I8-WlBSh|6I78
zNAr7s<QeOf>0@&IFA$%I;Wd{g`TmL~K+MI(rF1^rC=e0{qwESrh|@?;a%;VE3DG7<
z%VutFdt4<Xx@kF|<pCUB{oUG7C}0_T&{I4lX#&PeHr7vD-yKLeK9S(6z8bMB?4lW?
zgf(`Ls&(k(P=LeSwbl(|$9=Z@bF;nFQeTv;aP)dbQZm^qk)U;_f3A!n1zstmv*SGo
zIHAh}*(~rs1XV@Bk9sz$F0z0(nCDQ8@sWN~y~4qIY7QPPiK8S0O&peNmBN;wB{X^2
zsBn(H@3!MK`o8K($;IM#BnD}1HIsG+_2}u9Z1)zkRI?WyI_m1&-PkSad$ym1CD<N_
z3Q`4#cZ={dNjtbi6ifxG^k&9)5sWJ39iyQkaAmAtVSZo{YMfs9!uQ1^*3<P@U&jcj
zgX&?zEA7h*vek}@e`9M~&(F^rC}(D79<XGee(<FdTyL>LW5``Ftx$g9aNy6zq2yqh
zzFVWQ^j2yMJ9%DwhTspoaX{dYc!3b>za>*IBz7?u(l{-SPrr$PoC&qh5F>w?0x}$}
z;q*6R=7$SE^H|k$m4LLy!nnN|2^PBQfq_Eaf8HLnNg8Gj3JabKv6mgh8KLcZPs#C>
zk}zbN?ZYIG%@>j0aw!x3y(PQEGfT<ixlp1zcl)+Ngenr*%dGB742Ls6jmmrb$YH-j
zb#FBjb4O_|_%Fp+w#?Wc(XqXKgm<|3I1~y;C>J=c`qw`4Z~sZfY{nf#wJlCISor<?
zVDklzKr2=LKTi;i?^9Mc8bjboc7@8>G)qsJT(VKOo9A%VN=X6ZjQ6|ywL*P@=TG=K
zHii4VQWTTyrz1GJf;$EKldI0NSoGVw!lTN(GH&f3>?w2}_B(7R-9dUQ-_$*sp#A6L
zUAKj*aevjOlHJT0*q}%lSpO}-i%->m?lB`S`C7qF=0{w*f<nOX<3yEQM@~gEGrYAi
zi5HFs-9H1whRe-jd>=;ti1G7_OBbPIE{b-_R2<m2{?F$=dupLteFAiKlKtVcV&H?_
zq@WesDBtld6Gt$#%;Ckri;wGNWk?%cNK>X(Q*g%&3W=bluR^$aJ^o2AT)jF>&^I9M
z-Uyz#Vu$|kTKsz7Q5i_`??0o-#QrXD<9htKOWpdvpP-31C%SSyd|3Z~Kk-kfG~rW6
z*CH=4_}_oy9-=LePJ&pbVzYR!c=B`bsS9~d1<EOhDd!#kKjOYRs>-ih7gSV4@TUk!
zt8`0uSadg=4G2hgH!lVvE!`l}vFYv*kj_nacX!?ee&;*)kNeL#;~V2LfUyVc{jQqp
zna_OYisL_HriZrNf8Q$Po&Oo{U!QvD<MID|Nm7MkiOL-(O_7>P`+`OdFp(go2AszP
zv`itb{V9KIef#A!ng_;*cHrA<6Ck8tO1mxf4)hzpROu0hLN3<EL7m-Dy4{89Bc=N{
zrTKjaWwpt{l*37`YGn?ZNDCJ%dG>Psc0yjd?lL9htqj+Ebh{iQJv}cKwC!JrV_!^s
zk@<vZzWmosc^=yTY~K>%9)ZrwoUe{3%qB~<9_)vkXVM&`Q+ld*@x5YKiFS~Oq8(L3
z@Xk}hc<qgw+nF-05Qg1Pv2MzL_(f_E<tI_H{>SA}1Fb)g9MtbhV8g^Xhv*45#pS=h
z2Vt#SuQ&+QTmmVVqYbXIbX#fgYAkO)`+@pS6^bkti6i7!Ldr*8j2R}(9{xsr`x81;
zxUcgzLtTkjoG)=qxEi7`c&&c#<})5Tpu!$83=;)P3V4jcnXnWCT?z|@Rv8X4=j@!P
z8)+wNoXZglbnOV~zh$_BGN$#D(j?n!EkzXFw<h{_cY-%Nw`$K0Lo>{?Z-NLEMULxt
z39E5)bi%4B1(T}_j7RRRK4bJ$FS6ZJrhrmYD{_Tyo+q1eT|OaVs!&9YyX4#-nax=B
zJDR*0J*;tMU>7kqp-g)7zI1ro_K3D<aHqb4zJ6+2J2h+WI9w(|R+}rV);o-!Nk-bM
zu@PApLUah#TsqlT66rY#5b@dbUA@878p80|4<aWPmsDh&D#VMO+k!AR9oL!61Nu35
zY=?>4aJs#G)#|%Ffmj)(=ySQ6ZYw=?CoBQ`2f5co4xFO}F<{-bcX_oxbStKlvT{Om
zk|InMT;~<Tc{_{<)dGKEVS$F1cdTD8G>U?$yPral%iim){VmP&PV>?%`r2T<r&Wxq
z6Wh&KTR3$qYHDo@F3_N^<GNhkjBQ)}9%?Y)pj|QcjwFIq&(lJVp5T2N->bS3<QFO)
zF=dIO?IOLkVb6=oB-{FUJ3Oeiwzi?MkzYeNYUJYl|C&57#4bk+zXlfHpKuceP4mkb
zHqtDmMHW6%01oq7Tr1y@d}178@`zRBLif=+gccDJLMFBg1SCs}ZF110WBB$VN4s&p
zbJ<Hr`S=M&b?Jcfb62%`m!S5xF2;uv$N2so+4MlJQmb-u(TIxb-n$fJWi8xRr~ym&
zaaJ8f!71KOPQ8iP=^e*d_fR;EpAm;a+7i+8tE8=^$r8Z?IDY6i<G9TPECZSrmQ%{g
zR5InKc_A_yq<Q!$U{Q3eA4da`89kJ-k+XaARw0VFbr>k-9O)uQ=NMYVv}ez?h(pOI
zkV!6{t5aw<Z=uD#rHNu$cn#0YNHRbi`np@h1?xVC^d(R|B_T={u~=TZymd1(g;HK$
zKW$Mrf4Ln>Bk7(hvRD@|ve#*q(F`W<U!Guzk|ek~PI`cTpSqOi?_Az_d_9+!2eLB+
zoNUXgVuL3q8T=<1;_~7yz7D{R#s!S#vAAuBoL#o7SCNSqp0s~a>RD)M8U~6-+iF)<
zj-vmz<pj^F6h*gq=wkd2nAiN3N+%U`^)rOS=z_*b$tQH~M{RB(CEre@`7+Vg$n4*e
zU)-QbMW$42iOjL0g-m?r)x~*MpGvYQoxBgx(vmK6qS|G-T-`$XAW!y@;)9#c_zPWX
zz-<UmxReS{I#uGD4<{pe9affGL#(;tGUX`X3~PhEI0y*V+TF6cBLBCrdQey$+AQc8
z;H|IAOq00z2TItTnC!o=P&{6PzPKsnKk*X!k3Q{x{;+Ej^q*+Q|GebC1w;SI&ir3g
z;=29+*%AGpRRa1C+S325<i7=j_W%EE{x<ZVn@azi=7S*Mf8Bg1mkJaIaWS#p(}T5Q
za89k+M1@joYwH7abUCmm4OwU)!)`u4+)wdeEgVyf_BaA}_f+h4@%q9OA|unm!QGHC
zh}ez6;oic2&t4Rctbcm&FPN#Trj@EQVTpQ5N}3CH&SYp-J3qk042y_}Fy)8=ac2rp
zIZQyh>lCC<?Z(VXohsnhGYWiJs->*d6N2;=kV9cWo>B>;Rr)KT&d<#y0Ud`QoEjdS
z3~nuGH!p=;%3yy#hGMCag6`Oz25u+`CI>-pQ4!lCl^Y2P{6DAC@Rp`mqnzVNOH0d_
zuz!8R!$mQgq-v=U9IG!T55%<d^YbpJY`{yLot^Vbx#PMTt`bNH2>lhLZ|Dipm8+B!
zoYB0izvf(e$IUvQ&b1m^a<Y`;y~x_K-E}P+#;IcDB9eB}U~4D06sTep3jXclExzTt
zbrqLZ9hGkARKHS;njXvO<1wUPXax5A-sAuhj*clL!k58*8?Gg0e0+RxI-s?LTl{oL
z)WXUW+E{puR5^!?$|s+O@Q}HcC$g_U?l^nbj_mD*hD9u`KrFJT^B)adD()ygKacew
zUsznE<>Vad*ZWDd5ITEtHG=EsBrYwD<?<Lc44$IpGHuDV4+Kv-Zf-Sj??^Q`ly80|
z_+Q;o$3T1VKofM`_0WKgRzBbqWS|78{MrNCod+uH)(R(U^g9WFIsn*%Lzr{D(@r_m
zz{*906;)OBao6_6-5&;}CL@Czuri~NfLN{!1vNFF`<P1A1VO7Ksl}^<-}i03^%gC8
zFwWRQTlQWsuu9kr=JZEvREGVCrT1z~{ImV4I9)-6fAlvE{i~i%)&`t_+5`+uN=<x5
zdr#7QwkSsd_<U?|EY`hRkr}@!d`AHf2(T=#z3Qvdu?^2!*2vDq2M43z<a7$`+D4Qx
zSPG_NoOIgtJp<qxX;ENHLm@H(YO{Dn;Ql8!AgCG;mx4V|U|W=esOWcIiJG75H8t^!
zi4B%*Kr^}KNKP@G@r-E=Z6zc7*D;5BNGdfmd-1St{JCw_D5r7y5VTpNFaawTu^$5x
z#&Z<0cdr$z>ToK${&KgLTm-Ifhq<!=k${zauKQVBO(eT|xl<Fw*0#Qo@=2r8a6Yb-
zkcT{1()!Z8zrHyZkMsY{;9OnZKE7-#PuJ5kz&kA3UpQLz)znuGud0Ts^>%WKEOKA8
zy2c+}ft`pTBNa-1a(1SYKJ@~jY~uMK7PU+(<s4n{pG`*9yqlE{-yf4XGT(;h&Gl#>
zke>LI5c83TB%3q$R-|g0HGT7>9X9UG@1RFWapjOYtPlE_joBglyH_^jRaTO;-_Gk!
zhiz`Ya_iNZxHxoagZ^|W|3GFEMh_^IMjF1qcdwOvMb!BlE)iF9g2QI-@A5r%yP22%
z!~%*AM^$b8DSs#A?Q~nzt(6&bI6OAqWs|!U(Qunl*u-IDP)UYAE7b3yoxgYAGF?Tw
zS~2bWfXYPeY|YI+Ifasq0R@|@zORD%CP%$Vrw7Fm3noxCmr9j?46{OG%Vh$&k?NQ-
zL2rIs!-Y@IZaKn;EUuz2^(Z{ISG`?%<<dd0ClQ-!e`?IUyocg%MZRPxw!^MX#A%6`
z0dx<z&u0LGsY<8}qikB$+UK<T*hij)g)Q5R?!sDON~dw+mL_G$T@Le~6b@L5zTY!p
zP?>fJ$@z3rL*Oe-Nqqz?Nu=40p0rRc<WwW8G<O+pTBiPd*cj;>$7{GK<@WSr@oGwL
z4Kn!zgXu%l&9tlg4scQ(3=X_Z7HEb;VCxmS11cW>&W-_(?d;d<vPyH7>21)@RJEQw
zO((qv&;3(};>K@~<E~zw84J-NR9S4SZKo=?+GlK6Ar-AefrGWM<8d&ar@N=|EMf$^
zGa};R5_Gk1ZjjkyMkyK>6Kp9TZ16nEngFNcCj$XJ_Fxz&YO-LP6<#J=-4Bk9iL|Z7
zP8A$MLI0Hz#DIG1z{7_FXA?|Aor0H{OU+P*$!Kx4p}qM9G}T?Z{HK-tAV*83l*3GM
zbuSm8qW%TOhWv^db8tAHL14Evu^>xv)7JEdw4Dzo`$3#PvT;~=>A>s`C!$7JT*gv~
zITyLber_LrTPpL#+4!-X65-Ki|2KY@2{ld6L*%+3h1k`Gtx;5qBV$dqWN=Lss|=Rr
zjjjWs)Tt>(<5YpIB+<;e-_qJqX#KS!azn`vES{+^GD`-Zvil*w%~7%0oIEJbRvD5)
zD#5{N5JQV@XG(&6ik(?X40~L1uRj@UvC6jxk{e!Z^85HQIK})8sV#TXL+Kn*%bI6w
zjaz0K9~|u|P4xT|(OglapG<DFlt{J-Q6n10oR;&*Gc^1~BU)%3*X_^wDgTaaE#QXK
zFS!{sKoI-#_PGX0*6EZkIXP7&U-)Ap3F#nmPJVmT;N&DRF~-1u1wm9G2zq~fb}Smq
zV!a0^p=%7M6ML|IVC6(lwtd9lxNfaMLPs`4DF*VMHcs@Q7{GsIxv$G<=x)i(s}mxY
z>XucGH+iuYAF2?X9IcdDmQ?g$E4X%RDLzy&qGeS5c-XU&@2~We9t#Aar&%_8`UN8!
zFw5y55_9MI1_|UD62AK8>6x=8YPmVLE?f_W>d6=pv&Fn|(Zxp|9h#b}!^!!Cba!qV
z35B7fGI{76lxKe-U%47IrQdsozJh;h)nb04O268hT3jif)#FN!tI?kmSFpFck8DQh
zkr<4Y#PLLhrF3=1Tm!v7&R?af?W+f#Q2t%}jbE_G$o;c;x9DOAD5}}feMVkyVPU0y
zEmz^s6yZ0Ghw%hlLc#l|!8klYP?XZ@p3I1;^G>c_aI%4UO)sMOubsDfUOx>;Y1LL!
z{K3hU?w4<QL*&KnE<r7OIAcmOVWQvT%TH2wsq-6rd=Y89xfSx>QECLobGdC@g%QW&
zaId#V9kD6ZG#M2oZoGCD#3}({f4NW+*)*D^GN;C*S|fCJplod)z<6cvv$!GXIBBy>
zgjHQ;nL!d)D1>WKm|C+PoB8dImWMH2IP)X^^zegK<DEpZiq%RZp4f2rNgL^p`YWe2
zf<}uu)jkNdw_1D$!)WPGZ@TswKImB+=e0{0_96!+3O&8GXfeUnMN$a)>x);|l4YzF
zTwEFosF|pevYM+!#<VNb{OOkI*L2m~>5O(vx9$dpgJjCBPCoP}Fka}@H9lM!I^AlZ
z7!3-A&kuuUIlHxrv!CqKbxuK|QYJ0SrE<BneAIt#hyBGyvYkU)R1{hci$XQhhZljP
z?8ZMICrI^x^TgI#h%0esb(J=L`~#t#sD?#GWlXeYcqUpbEq1b*AIq}m;B3sS^3$=<
zOWI74q~O&Y!$>do3nw*8U=~d3i0DtBd>&whiuDEUKP2W0qZDItaWdMr8(Qu>+mic{
zkYM2ZclLb~0+Q-G^V;192fnqES{>0z3E}ieEkaJCi=E#MV_A3Zt&Bp48@6jW69d(D
zbWxI&kCfG|j$sIAlkIAP(IaQK!?msGeA{x4t+8CPm8&9CL(5hQ5Z_mCR+pp^2)}pA
z`^z0aumL4QoW5>I<yw;MnXLZkz!e%9ovDByl}Un|`2m97OxS$){@z3PWpe=hC`y~>
z1W;16&RAwb!j5I<uG#U7kN%F?Q!r*M%q~tepzT50+VPA{mcaUvIhqswdN=Rz5cdGG
zBqIu-S;dw1YfCBZ#l}n0#8TqnjWe=9Vkpj3QK{xB8cOjeZ1q*@c^})>F>tRR>MN?n
zUD<98Y3t|;Djmt5Mu9*EOQ`V_I#4bRXW-M&_q2U^5bCa0^TuW{yZ>^oB`i2RXM6fN
zC|QGlv}BNdV~iF|wSK)D3x<Q9MOiT2No8<8E!=^X*r^ANajcJI2`VE7V?t8X^{Oif
z?7E*H7><qJR#N6Gsp8|`p7e_5a;RB{ET&bSQ*o-`kBoq>je_*)HBy(2k^3jufy+M1
z4E%~_O~JA8Ih-{y*@5n-YTV`J@6EM}j0$T`Y;tb7x*aTIx*Z(*bLQofUU$bc&|fH_
z*q}31sN8cSYEG4d2hzpM%*{q+Cv<)((Q%1Vih1?37H!wPa4XfT9Nd^pGD&cG9k{=h
z#kpCjKB*5C8f<yo+Yq#eM_{?Uqmq{q3w9px?#1-iLFNc21Jn%*ipi>`N!xFkbUSsP
z;~<ulRO~j0AJXOA?~gf}ngeEP0x*A-a~Fr0E1p^lDD%d>@2YjyGR2UVo~NhOG<51Q
zPh^anyZECE?7TDq=aEv&6BVr5*JA%(@w*>dxS1^C-q%`lYvY{2U>QE^g2m2YrHW<_
zZ!~B4i5A1Nm$%MwRH~!DmHPG%nDMfk4)RqCUWkA(@8j&|bCNTMJFwA+No}147MO9c
z-+$DfHe^4P!cNxBMLFKB(p&#}RetS>|MIY=1%rA@X*i;fS0Rm#n_nY=k7pcui*y+T
zsfmFNZ-daFI7AKR>dM5XhT<M5w!zH}hY&aO`cIzH5yr#WPSMpz9<!wGaHv`k=Km{+
zPm%w65Ps5E@oAu-C<-TBo>n$@MrP2uyGKJ~{saTR#Dc&&U`jDgj?CQm4&%w`KZT^C
zYh^851s57{;I9Gmhl)v0zk_Nwi693Wc+KgbY+kDWME;7Xv>o^@F!aZi#i(JcknOTm
zkWsb?r)4MfDPWwLECq!Jie4qBIa|f=8z!JE;wRap_x1;;%NE6@WmIbMz``!!X0*&_
z*Oi;bx>2dY$gX6_2t<s0>`Sg0N`L<j`6Pi5G!8Q2+z(DDE!8~H|8+>rs~ixf%*W(@
zfB@4o!(KEmb-z+NQ9yt~pcHCUt#*2Ez)it_UMmxNlv`6BHt`2gF~A6I73S_bXGn;I
z*>04A=etZWsdQH`k2)QHpN!`#hOgatx-vXqsbwP|ZA8aFCvj9{nkn6zvSkE~viyoB
zF~_4)*NoRZJ3O;a&(pbH!fG-4C^)%S#a^x~!6KvOSzm=GOhlNPj!J|~AZB{=30<~Z
zOTfzB)TL}*pwgDzmi(+`sRt-qgukNIm}BkBiXhIRz|G<U#T%7g7_2Q1S!4<&=SL=*
zMYT93yTxYf7p5}7pkL!vd=6nMi5CSrDc9zMfW04T8l>rerb($VGbi9w|HdI)B~ePN
z=pB%<kBUuG!Qko<knwPd0=kw?A0x#Yk!xw=dEM1FIV=?OqK)Yn8-ng|D`XZ84_A7}
z+}G3NbcuZm@y<UzK4!e*s@hHLm!)dK$&&|b;gB=h>Pk+gf`q`NJ~4NcZEum$vF?Jg
z*hyd<-RVFsmDs7<0&RrJU(|ValS(*yvg@)~@;!ZVR}~Fh9Ze!(3D^zwGUs^JQ+{7x
z><{_W9Bhi9y)FSnb|OYNn)Q2K$d;UK?g&e(1NWOgAp`ywg_FmMw$=eMQo8Cp3cM)I
zygVT-mXR#yI!#Q~|0o)q8J^!%xAK6bGR`Zo3=pnZE<5p>Nqsa`dPaYR?XtM!dra4D
ze)jEJ;hh#}QIIdvU&=WTOdZ?Ed`yy~Z)cn({n4(pD<m>F(_*QsKj5ubXl!^+x-oJf
zNBj1{77>#L)7-@b^vAr^<eaqdV9Y0sC(`lFrE-=|lJRQL<poG>Zpi!N;69@k6K^fS
zadCO${B4j#Lh+68HzMh@m$z>Xy!>UHzd<BlV6)ru@BzINrZCR2Ltad9$Q<aHKn(`Y
z>|~D5Jsv0<-G5~5<2AsJCYC@@b5Bk3?wib7`a?k9pf5IHt^DDT&NZSiOqFOzrxpHT
zmeMi4cBqEn=g+Ffp~cyC0@DV&W*|TLBTn3V;BRR_nz6?D6dizAa>pK=+9lv`4vJqS
z0JmJwx6~9FQFs{&Jk3~c)f-a5-5(u`d-n`_k-LJkH5(u0XY8<mL6)rdKxOP}#k+<(
zq;3{~wSQ4(AJSlm{<G^KT66Jn7hhM|K0*<Z$#C|_aIBR)!nqYJKH88L3nFLt3$Rvk
z!hzYoW4|`Fv5??_K2O=oX(yL_zNPDN`N{Hp8ttL!)y`bSLNRBt&X~QJ{KpR-Ip4jN
z5!26|T%v5r`QJjsd{8!c_;QdH=~^SHdtWJ8ynN+1OmX(^R)w?yt7F|!8@5_JQ;()!
zbS5Hn9Q2B>1(b=q{|rQbor77Czs|B>SL{E#OmfNDObjH&UpzcpXLP6q$LB`v8EEH?
zz5JEQh``SP;g;Ugf{|7m%|-du&)|ZBgQn~e;0xmytM^hNMzewK*g1|Y1$E?`H%0r~
za)7AO(eSF5b{}i31yG0wGon9G1fenVn(<H)i%SeOeDNxtIY#GaGl>^~7fVm=*T66v
z8uynjZ|@evkojVByKb*&zsFSC_pT#N_SZJuoHO5!u>=+__dHG+vsPj!A5IZ8eAbF%
z#ZYY659wc2xFkrRZZTCJiU;TTnJj=T_x0Dp`mEy;hT<F0#({(GKt&znOxQa`WLAV$
z>SnP?Yx_r9>vhl#&e}8bog)rH8l(_ROW>gFuI-BAFDD=gq#wQ|mD86F0KV&B>R7_g
zV6S<XMxT9crpUHB-!?@+t5z-H>d*_~*aXYqw*13Y&`id9?3Wv1r6#51MnXxa&WZgg
z#{ZBg&vRAwXRM4ltTcH_6jM<s#gRC@N#H;HzPW0(pw`0S;*_D4)_%QgaV4W1y5_;+
zt*vrn1jqyNh=5fBYy<>(#s(`BiS_JZgfHiQ@j@Pq(Nv^}k%7KIGF?I}CuhfHJCiE#
zQoGtWkE#TP+p$QI1&t!enP37vf~9I46LPVkB03jQQ#MXppP_awnSmA)PGiS0Wxds=
zn9w)`6Q_AtL~AlwqZ0b|Mi}(imCcvOL!q#Kwe$6~YCHIef`Xk^mOjk3MQT-!Ds|XW
zB^Y4&@>0y{a;G^I7a9|h#ydrX&{HpsbBNzxH`U-cx6)gjPoD#aHk@v%+d*0SP2Vcz
zFINYc`72hI2+yd0w}{Y{Z1mI(O<JzhRE&8<X!TcURO+$Eu+LPGcxLJO>T*Qc>L#7S
zn3r)6GPGm>7#>xh)qt*c=;FQ(ekSz`YbFXbXt=o{g2|Gw#VHP<2_JL^3Lw7pMF&OO
zNkG-Z<zHBA_SoDT^~L)3N+;Gwi8=nUU}BnZey<|dmt=!W&D*WH%IrRzuU>dW@}tRg
zFjD*NxcUdQ!6JLVQwJXvfL59sRlf$7cq#aKp?rCuF&rWJ8KWnguB}TJA7q;D&e`%n
zc1L4dt3WEbTWt6Xw67KRc`}bu>=B?k>MFM33`{bdwWr?-tbL~LXA5mDiL5H>r^194
z`#WMz)1zy(yp5|0?2Girht+WlBivPs-U%|O!G7(vjs22}Iq07~ht*Woq?sEzZ`xy^
z-9q3>t-JU;5Bdl7uP82wB=9ZmFhh?Y1Bns*VIZ#ru|J3tCsA|~E7tctU}Ilr2@9-;
z1X)1YkVqIsW`4mlPS<o0&lVr5{<?d`AySSd;@z#`v>&;7p%+fR;vI9Bx@V}gsxAA6
zX05}I&leX5lRoUH)^p+Vac47a#iu%rxJnr4y}l0P$^vLN4y_Fh6=M4E3d)8dKyE$V
z-7#IDEYl>P?oGFzVN7@hV@Uv399LpKtGmFO)wQ6Z6~oa!yO6SVs;RIV=&A){m1nZs
zq`p>YhUJyj1nyRyN5=+}d^gE)$6O6tg%`FaGdUk9mL=r&6&CDwg{~Oz+mCAb+RcBw
z)K^KWXpX4pGJtqW`WK3Sg+>P@c0&mu7WVHLGDcSQT3ic^t?CZxL-Pr?TnrbnOI)$i
zZq-Iv6*94@9pii&lLh6P__i%*nPs!zLEt&%=O_=PEt2C68UET2-UYk6T~5*!3@VSh
z<8#xN7=oL|vT2z3vZ9q5m}GA(0L9@Np!K0~KP_@GCi_*(L1pwMo#6e|g@_COK}jH<
z>jb(i?xCMBpu^Y|CVGCbAbFDQB#`$X_T<lpJ}Lf6Ez=ULa2>R`R#zX4>ut2N2}kD2
z%q(hINJhI3&NW}~DFpA!M_^IF%FisWE(tVDXLZ($y3~{AO^OZcKfN{HZB=nLg}UA_
z2-v>kK0GhMlok`|>u1+))VDizF4(-X3$R|dKQJ8<E}U}gEOe;*l4CX{Em(8@`SbZX
zoAZe=``Nxpb)l<&NrGhf#nzS`fGH_&Dt6-)arV=B3AZZyq)ZTD9?9iqDio%Y)ZE0R
z9(_6?>~%swabt6sa7KeMKovFa6D)!KW#VN36bz*rtjxwMWf)cmX{<YfJJl(~u0^p$
zC)esuf}Zr6=4@EnQQB4dR)R}vhi>_5D`2$b2k8nDk5Mo6NS4AWF%LSb-iwP#L`;}R
zoM^ppk%vX&T8#e6ymYU~l$MNIG6Lhi>g8qJet8?Xw%2wh(7SNJo>s@JGVCGt*B8Py
z*<oY$hU3u(1f}}b$>%@$SvhlGt%s3Aq@>k+%UJw<wYh2HIj^p}qFvBgtJ{_1HETFy
zzYPZNS~;4p9!J#E*|nt5bs}!*v&ODEa9-$X^_QpEk7xe1C`M1X4&&LM!!KT;W`!9d
zagi(#Vv=8D2qp&$v4jHwm+{Fq4-r<1o<41lv0RMk(3^yu;$Lj~NFKA{!l4u^c(JL8
z`-cg3%C;1%%zMXKAjUG|4oXf@FU3*YDKpw~A0fulAaV2VFc{g-Nv;_5>FY{<3RY5|
z{lajh$Z%Fo4|m<p8QdI23{gt=HDD6j(L$_mTSvG-#(y_o^{IN{H}g%PA!8BArXu^o
zl)j`(u({3e!75Ek0aeac%Nkmp-kAepsLDITbe)%3Sq3qK6^S-g(SwX<q%W^=4JBUW
zx`9C<V6An;9kdEQ<@#qLA;I97y5lV;W3Zp2z3E56hxqw@n<F^}3z`*tD0uv6___@X
z>ZN$-j)9aE<#zGs%)ajajFeXS2W$tP2K*Y|UNkigD!@(ot67#oD(MD0sHx~3ZZSUh
z>USE9FQw2v<u$d-T-SX{{VcKUe6W3f;m`>VG@de!x~ZdnPp{gZ^5)|ullgB>51MVY
zA9^*;oa(Cdm5=X0)8Rt#jW1fzDrCYsJx>X4v@b>?<!x{U3_xW{Ce?JXyxtJl3r~0~
zeOA*if*EbFL^I!6%Rb{2J}9TW73@Ygc)xpU;?gpDElV4tnflpYhf~AHU|VvAw^(WY
zDN0pndtcGQ&EK!>zxO@3+$SwIA0p8kF$LJ_N!kus!0BK|8vP27_A5$$&w1XxL%pIH
zja9yrUs+_;__Fl<H&3x(tT=I-F+TRprxyfqQt^$6L$w#0A-MDdlS$}po$VnC!NC6t
z2@2X7_K9{+^2o+TSJXsZ2eoW^Gg{Hb&Il|b-kkY;ThF_9)0N+S%`qlUL)GBP&}h8N
z*bNn$u5nZ!ZXCf1^s@R2SQad~^llackAn4@LJ`uZTN|aJ$7jnr1+HrxQfDUl`nhUZ
zw_C8o0AS|T)ivN^a@wyXE9W2ZY22u`b!b(DciUc<bf$9FBxF0JUH<HdB@#QydX;s0
zMl-b8Gx=3_G{#bLsxv0Hg8k*eDa8MU?WaWAUclHZ9`PFGtZ*Wp)2sCfJI9crvAP|L
z;EKwga30uCA>Yc%Gmi_(I(Sqds*PR;8iy<4tMQ3MLr+{pWo2s6QD9M(P=Aq>uPJZB
z+NfZs+kF1|Y?ssK+wwQw-k&b*fMJoN_J~PXi_nJY80iaZ8`0qPv)(s<2VD%2Db3wE
z9kfs=>sQHPnVaht?P{wz;3SKB%B#ToppA|?q#fNqU}j{?{bPL_`}uC)p5GQSsmr~x
z)JVz{#v_6q?3iP7?*i+tC==t1u{7h8pDt_BggwQ3J8vk%tm$dcQeX%NVomldXp?7<
z74_&Km7%KC8dJ~u1W~I-0O4!v8w%MQ;d`c}tk+t4U*Iu5p5AB*e~+<kZ)Ltc{!5|X
z@nUFd*qw5IaLVr^3!J{Am5z?;B}c&bVXm4}7uLqgJ3CF|R7MNE+oLUFYHEI3$EY*R
z`K8LN5)@BjJ3FZnin8PpHmnq$GMDh~p%{+B=^*?nW;5$AEhn56!BFHMliKs<8Vo#(
zi_YIdHvZP_ZTEcxUeNP=>mIcD2=Cr>6<G4V`WR>M`ky;D-(R8ILDM*2DZ`^f+J85Z
zT3?=GK1`8cv^Q-DRlc;fvuVmR#!mnIfcxGY_6a{<@Y@|R$+KmJcweH<h`spVLM@9%
z`pwM`SD*gr`e9>Q3E?*m44xta{Mx!+u#sUf8Kl8DRabF(zD+DK=27lOPoht<cz%`|
z`i1If)Nsr*S36PQe^N^?lA=8K(Y$#t?l%71)>{mhP9GiiAF8r{aw55EYX6Zobw^bY
zIFH0@4qV7`vavn9IEkfFby&WCZx&mOiLt(^0iLT1!?PnsQHLZjKEiUs5T0n&V`!Fa
ze4neCcl`GEavm#?F3I<Mx6=JYL{MpVRfOb$Q$u^V81$!{q*$2#AfwWFN_5i$@dETy
zQK*r|4J-x2*8I^4jPL)!$CHVnTHHBu*?LuAVp+XZyw&htA8b!Tqex*Z%$q*>iRE{A
zR6jJhILPkP&IElhuqB$(xUbt?juBf%)EdpHscU<bb3n2*o${)=rZm#0O<WT5p<||k
z3uV!MQiyLqBH2k=4gX9B`<Q#51=P{Wpotdi>$ifdkViq6UYzhT=hcUS?w2?e3k!}g
z^&R*Umf!APb9?hCD6gy2;H*=%ruJF$pn?Gx0dBn(v{lpkQ<|7`e;@7N1AO;%>k>ot
zpUqEIvpRQc4k?#%r=AZZJ=<;ccQ&To_jfHT!^Y9_I5>24Bn0iH@LNj6bn{r5XU@!c
z9gSBU>;{+J>^lrmCI#`|l)o97=Fa@&5D-7O&u-s=h|YI7+iz_N5+y$;A!+IyKfenF
z-rm0OwCs}*w=6D9lAxPtUDmSSTtp^BB+6$wfmY{q#dVrY@bnkZmSaRK^~4V<^Yv%k
z)|TOGXqcQ3!kjXlY|g5s?;oixL|{sD3@TEGx>~NrYPN)^D|g`;FrnmqSoDL<#S`;B
z@bIMSlFbl7sD{Bhd0W_qmdo>#)*wI9)4^MZakby`C7X92f8k;_dBA!bHYL7!>ks)8
zo=UDA?I{v?%n>!2YjwIJ^|RV-H)-MbNE1!3H;vB<KQMz)tr}0kma{R~v5t^FR-$lt
zZR781n(g$0`bWD=*X)7$>!%T6tlB$X>jDlnnsc$WR$qd)szl!FQGBV6u`z=tlU)09
zsp@mO9J0TcaI)(Ok0}$ya7TqDebX`iP}iGx|FjcI(wsXdG!w*bn13XYnp_oI>T(Br
zXI?V?(SzTW*WF?=4=z~C(%^l}Zotuf4=q?rz9HQK&x(z2_O-OuzozY-*F9fEw;#O0
z26uONobF4QrzW!uw1hyn`>Ce&%#;i&69wX9WBJjjY*>{y1^;yOI;{DT4$%DgM0^{}
zg6R=AFT$#*p+1|4VM4;dv=D5Wz?|xDW`x~~`x`x2p5jgWW|cJIDjbHfA!0Q20vS%8
zS1ou6NPHJ=#`q=Z7@vJxtTvvWcmU7l?b|pr0A!P^BpuPrjS$MQ{T_#XEO%C6XKEA1
z50qL@k%G%re`it*L*uco@tTP9-3OBXeva<rkucpBC8Sb7N83zsBNYJ~R-H&%mlX>Q
zT9jg&?%R87er`}FDG@q`PgXsML;tQNXc?&U_`*v0iuI>mTucFXbX0*=12u`s6fUrV
zJWf1WRS1?@D5ay1+b?kbG)270;qB=jsA%M%+}_}Cq2@EUioLb%*Xc-OefM27W+Nk>
zvC#H~ht3;jp|m)xqa`}+zk9i2VpdGaP*aPI>SshKwY5L~SxL{TDtGnl;<ys#3|5G|
z7>NK6Q1&BgaH+1)&Ytr}L1BG+yXIL>48@mI=%*}g)NVZ$xq-o&DcSwc&;Ne&!QXvW
z`6v1sBiH#MR4sKA<v^0bpQmL49D}Arf=@EuHGIa&YNg9@BYWljuEA$WvmV(MU?(>e
zS-SKbr>#O|H}g~XTgL6fe1&I(1O@K(>XFazS4<1tcHWrM(|_-1(>^|BNXCm8fYuxj
zo10nP`=y`K;wYRQXxnm5<Sq*Cwl1YB((0oXEv7KWt+ERjiy_XPN)1#7>SA+id+`Yq
z!rkd!T>j3von7IYqN)3|Z9#IRz_(F}-Kx{ipPG>_m;Ex3tmOyYQ&I9K?ln*ACGG_x
zO(=535&Apc{|NB7VP(AgET1yFR||0r5@^9R>!~0!ln*1{Ut`jyya*9Jwigw2F5^ss
zGcI}OhOhc*<;mM<2~1@0+uXXn%8ymIZjW#(m!NteA>p)iq{O4U#BdQ1yleegBksg#
z808^>tZ`yeAsL+rCfbLUMT8wC;>v6iUh{djz*=){Flxoj#2h8Poi&c<6&3W#WBk`k
zTm;M|P@nqW-9L`;vYzg~)chAa=+?GTt)?pBnslsje{mE>;aD+cZSvQx67wPav5v_A
zVsKk|(_lC<BDc?ML2JDX-&bD3=^nc!2YXy1WVTp6XYh)dZ8orU>Up{+6lcT2%1k*w
zmdnh^JmU(x8>ey()gR$C(jXKK{R5e(KW@f<@w;-l_uiVj`qpT{CSu4g;4uMlW7yYs
z^nMRdilA;Dl?3BGSZNnP_8<+}KdZOBikg9O3Ba?;g%Rs!a_Nd4V>aJg)$Jbx1S$51
zXUqqto7}m$yjQ6^8zjm$M3a^tsaJV4$diCqbA|39G6oelcMW_^V0zzUbKrc~MENsw
zfgny)Jn^cc_GG@p2^4EVAt#t4K5e|$#?%fFaCvK`K3UmLtGg0kJcS;A_@0MZTDPqG
z*rI8vue8)$V53MrKryR?O-zz|)8OKHOAgMttvRal;K+1#;Y9hJp&(rpdi;U3E@av$
za7>VXWFX#yl;EYToJOdAq0F%)0X;Fw6V2ji`Iidm*3rr*e{bfjM?^0L$dbGb{;v_D
z1A*jg!&~T4QSb_U|3H8Cq;t1@aCPW(K(=2QSn8GLBqQRl;vpoc^`_=3Ix4}nzV#0^
zrNxM6bq37^|HMvS!EaewiWDe45?hRh92Xb0r=io|&Z1UYL~zQvyS2MB?5kbY@<Cq~
zTtZy`GI6fTX8q|!jLWo%iGNto<ERupxOn$-z*tbgRAfyO+3hTpL5^GnOR1wo;`9)>
zDD?CEDrOCDb*xZundGw$)^ga;KAir<u9HJaVa3nwc_@t)dp2bo4~;oQPBkPATd{@{
zbLtuySxA)X=afap$fe5ukd^&C^Of8uGdRAP1)oY^D)CK5@=-LVDYsp81XVy5vvd?(
z`n4E3i}%;SvxE80&H9efnut9ke7_49Ug<^seT&V2$}qGi<jGI*Z1|WxLtl1gZj@jh
zX5+Iy1b{LqteNzV5u$oQhAm@|MxeWqO)IveWBF5a3G;c+NYG_G36YrSY$^Ylqq0Qj
z6d2^6%2`fTZwa$kRSz@>0gNx+DVb|jFc<-R!z`LJ_i^E3-3tJV6BF=*$>B$A-My2;
zj)%%-JV#o23(2Yad*cD9cHv#={s?d8E0dvbDNAoh+)p8TCcoV3hu;6fz5AcHxBo%c
z1OqG5z($CTd34Zcf@WMCerWGFrfLIxN%rSkL<odkb0sgA<$Sx$2)=ay*ywG|!s1h#
zJCXf(sJA<!oXl&l_QjW117s(<y4uq7y9pzf>vwXQ&ip}K4!~~cJL`|u<<U%vtf|Oc
zi2X%I=g#+b6u<2Q^X)2g&c2}-&cImmmpO}7C|DW~Ds^MGys)O@;x(StxYG^U?3!OV
zfY`u~iuX?5#~=DBS>lN1-i}aKl^l6_&46<6R^pv0^U1qH4oCXKQIQ`EbQ*Tl6oy^-
zqe&URVFw4N7GGNAWwF}umgoXUU&`?WB3z${MR4jI8%}7QpSEE*{08S(v+VA^Vt@5h
z%O>z{Y(}*3v9lHV3-4N)*n#M2ej)-*oS`xkwuT|!M>25>Jt?my;{}SFm(~JFEP}0^
zo!J`u{(O&Zcr*V1P1u=|?@A2IbHWK)Gf7V1l<4-dre>2Qcl<ZiyL<s%UR<r>d_8+k
zBT;BwoDyf2bQ*-hux|Rjo#{CRU1irDCPU-hfX&IU-igX!b;Wn8w)7e0+|;48(IX?h
zgYDRy_s)bnvqnQ}s-yZ!IZv7>vIU*}2x(~MadCUcTfKjVU;3>-@hd`QS3rFr&SvMP
zKFYZ{eliRr3lpV|P%*rK__!qstp+xR{vaDZ=TYh3%DZf{J8Y)-Cqe6bdC&`Y4vwmS
zh`tAQ#tjM422EQSNGMy=W^k7uSF%^m?e#=P63mHucr#o?4)+Dd7O*t17K+g^z6Qa8
zv9`3w7k)<EMfSz<UzvfRM(hF-e>&b^0_0>*pDg4+&LuAmN?Jm1`zNNr>wyAq72|+Z
zMPc17w6_^gNeNBBv&!N_C3!7!g%6Hoe#B;Qr!hn?6L|X%{*8mc?@R6*FD>*Mxu`ox
z+3dgBUXuApgAjpi?&=Fh1)EcyvF!5mUl~lMKbGo?fG1`wnSf)JY^Xv+0egO#3wQZk
z3h@NN$JnDsQz;b21}t<lS4|WrmNu*(V4&tP_iDD)O%8;j#viC#nJBn1J;gyDXY*(y
zWU7}cl}KgG2wRj$)0D_sSU6dJ3(+DDCK(y&Zr6ce_}2MVxkN|DsSLT}o`m=-<{#`l
zBU)~g4_3<)fBpAFoA$nwyA8<#J_wf*pEXpy)2cS|G&FSDP}e^&i|*}bo*A97GNjRw
z&s(JU*m_?9L4sar=<bH6EMtm6kP^%%QPC&Mle^Ka+NsEm&~FpA6Rq}ZrLnbnaZ(2U
z2t-R}X>X7BmYIVK3jiT#=^c$$CL!SybZ0~G5hyd({qDK%l=68U0?%!23;m^o)lBsy
zK}%Fw4d0%U(ot2dZ>pbIaE!M*5S~E}ZZ%P3p!-gojX>ELO8@z$EAjYU&cb|<oc!mE
zPpCP#i~E)iDTThQ@~v^6oSC_Db_t~>K&N1cOVNw{mbuxe*6y~owD>^CUTv<WHJhLX
z)=sO_Gd@K#!aWa#hRvCpryg;gEZ|TpX2Q122r)38uonI?{;b<Z%YgPXrPAfgsFI?x
zHT^kDA~Q3&f&PmiQA-=!pCJjPZlm`;m>Lc*mrq~>3@X1%3zwzO{oNpX#jQMLGrDQ^
zUjfyfnHlmYRSj=}T^3$0GYi0u@P>keP5cRa@^3?=d_2B62=$4U5oBkY@!Rz6j%;;^
zH!ij<r1H^|7%f>c0X8z<3Fe(D$7$-@ReY-K#?*(Yl<5P;j0SpO5^Y;rS6S@kJ}Q4x
zUpRGHH~PgZLy^|R)t!EprFMPP|4T8BmLPxN%Ej4WmexHJI>4CFyE__&ltZWm<0Bh>
zg+MQ2rKJ(Y8oVO+CO43t%$GHy9_?z;LuU-eBk7%E3x_5H1Rjx9w{R04YPTO4(kjs+
zHh>1<#9c<V$5YbsTj(3$7%*wZ=5OE1Mn{m}w<ZKU?7Ra02^u+88E?$uK&&mR*vtv2
z@bV@<K}Sb>Kd++5W{fmBa2u-%cj;<rf;_h)Rl!tyHoK8=#PwKd&as@!<YIiR$rI=s
zFIft}#Fp8z@>vLbw}kUNNOao@DKI!ufbrmDKNpjJs<3o0tRp-%wF95ok26)OezHw|
z>tq6BGKcY9TI%DVsP~VK^zAs)-yiU+TvYv=-Q9mA2owyGGQTR<RLJ$1IqhXG%0JVF
z-S@RDV8bIH^h_mz9+QY4weA-%XKBdg^|sN;d0#Y~k6k1evvJcCL_|U2<m&<6A~}jI
z4&5&XwaNns@?NogEDOZrACH;Q(GCDMoA~p`mmTj|N%4gb@R#v|=hwmIPEiSYR*|13
zs#!iuKa}!Offs`2?tam9i3zrrnxz9?cSb<-cx2>*(Z_fCVl)w%Idb~1Ula;le74}Q
zSU4Wi8UW`oe`l<Btu?zhX3_krVJODnh*)1^fE^`TU5%x!+sj`1a5yGfURJIixem8)
z`EA2Z7n7lG&pKSmoh9-}_Y6~NV=M<z9HI7%U>x5AFb?R%RLH`;>RH#Ah)kkvMjkVV
zepxP_Mojs%j<$5vo7b+dU)-C0!W|OZMkpe^Ycj0I5kUIhTJABRs6cr#(&#)Ho_@`D
zap3S(?_-dx?Dw>Ix)*zGbX-B9pYJ%jG}|JWHM_cgr=}HB@x-r~t*L0vnj5n%-I?A=
zd|TA~5OWji2v<lOu_r5$L!aDGntO<W)pvRlLiVj-V;{d{%1=)&Vbj?{wM>u^wOroY
zD@nK7_{|oNOTb8io7ihH8o-M{T)sRuPRuERAygQAZm7fJH5;N@V1!Y-HPJm<s+o1b
zMe;TAbsV=PbEp}oE@9a8guTKZ5oVUN<_(NP*WvM3uc^i$V(iPp**?~1udXmbjvC#v
zXvr6~dL_45z(~D@WFP`C{N2r+{_`STyrGbgk)8TqUbA6FE_4!i9X5Xh`OK))l>r^i
z^O<N$$JS^RSijQ;qdEZ5KfJqj4v|DPLX6$5%X>6I%loQWE8)!Te(|W?As;p_<mMrt
z@~hAnTkGMnhRH3XmFE<-ULre_-mLI4E9|YPtGUItsaO4SF~`VO>|^aSX%5Zm*06Q>
zOO`^b{8vjMHT+S1sc0->oorJg7tRHol<`A>vC~5Em;B1PRF$h?=O-al#>6cX^`GKV
z&A4!!JedhgthipF>qhyNy4cNmLO+ENxy7|e{>VSsM3Y?BpMhStuU3gFLnqZxQer|L
z(rUQ)K}OsjE5Q>!-I93QnBQI_JhoCaMcmWBoO`f1^nE1%$LGGvxpTgSc*guHb|ZL%
zGOEJ3!r$&nE+gNXbVP1yIIXlKs1Z7al{(*<(+oQ{{k}i@METu2?T;Vo${>=7;^v1N
z<LC-xWI2oT1sD1cbuZiq`2xfdi-oLsD3Qv6O7j588f~U7zQYiUJ6L8;xdti-!pZ7q
z<3R(nI=Bg3?V0T=*IJBV^)ZpmL(<V{?nw-6XTXRT=#Clqf=@Jf=L}QbC@LAj_?+Y0
z5wq&*3frwYKhmeU{~7w1KOD!v9NX0VCt6lc4Sx=|Mn+zl16%#n%5^am7F<biSDq%&
z!LU7QM8Rn8L(5E$W&FwEy#eC`w^+N4NywJ+r&#cCB)GzDl?(S^Gt~w=JN&T3G@2{A
z$*m&k?ytRFtCIYg9qmyS`AG|_6?{qOE3#|-;5N+W7zz^OQMpIZ`jHQr!Jnot)zm%s
z4d9Y<&3<MVGanK03f?eD3QX#qK1ZBJ?ENVIuoW})!RWkD6ySHc*j;EFU8?|9`zfmF
zt;MQx_s(EV>-fFEuDjEgZ=L{Yk|>t>6A+8mvegIxz6G@EIu*LSYT6ag`MA_GKH_b3
zVGPNTn5$Kxpy#p7J`lk+X5yBQ@m+~>wGnuTzCCsxVdC%Na;iu#DODaE*N*oBkAP3w
z+=)Avh!3l7Pq$^DVieih{0*<!KjlrvJdlXwIJTn{GLU@GG<8bYgLo4DC!F_O5LWUT
zp+l*Xw?14ZV#C%bh8AM>62BVgYwPPY+=Bu;2k~9D<*B=&BeHMuGytU*fSDChI%Kf}
zwKQVpwi>hbeGuXp)LeWb)1JYMpR4W}Zw=xm`zA-9JMVcRVz^xT>rKjRlpFAT8IDe~
z$6h&iJpavdWO=WlKC&rm=rK-$kf4l%asY`>l=cN~s0&OzUw0Q$1sb_(@zrzQw_Ou+
zQG+rm^4@oTh^3f5;|!i8F+VV-LY-X3{e7MjCwCi2^v~Kp9$9A|aI8*vD$+g$F~jW8
z%x&hJrK<A{(_v<(hb5HF@BrVXH7Rx5=E^Z%lxUMmx7}xPVkKj0`1m_ZWixk_kZQ#U
zT|FbG6rvvTU!0H((fd4;i&Z~yiYLH!nyUgfWN)9mZ+f=Og~u>H@%kko4XkEk=74hz
znh<%IaF<I?R$-URqd!4^&2Vm#5TPjUVqrmE_Y3<)ws#e5%W9X7HnKwR2tm+0=VGj-
zq`u*A6k;^9*<3SxN0Z0$>3art->p=|ET@fabo+6?CTOfF%7G**mGyE$Bz0zUSx8>{
zF*fHGc`4k5A8t0NRb9{9B29)`18o{Tv<)?&reRvjXA1u*pLJqo2&ER8%WYYumYTJe
zzCBF@y}$08OOy3usB)va$a0X3&B6L(yZ^IE9_uGc{Mb7W!hDvFNEc*XpCDpfUO_&G
zO52~GL;v#14UJ;tj-K~kGPC-UmTvv-=<x_6^ZK9~$P-m+Q&=>8_TmE7(V!{QnTlmI
z1zc1_z`iYS`P?mfh^gGA63nvx!TNZX^4KF~gxM)NN6)2i1mNy|Vfl~SMIH)|)(u@%
zn`Nu9$QHKQ^{(2b8;jk?jfmJ=3Q9tO=m`bxpTRi9vit#+%o4_I5~5vb@HaFkPa7cM
z-CEm1am*62tPZIxJ<}1gs4~l@p=Vgg@{cj`kJ&@Y4;Zrxv)}{jxUY4@S^HHp$BvA%
zC)@ghY?YvT$dtUe?9T0sr^T?&_3oPNVkt*S`m|hlec$kXpc%wY+;du(wt~QT*^;zG
zg|eMykNojYej2Sr(T_J^)vXp2Q>eFd%anm$6&JU4xV98KM8)QI{a6#;Qx@%hEcdc!
z0mbF`>UgzYaVeW&iC~KDO+jCx@|g`3`d5mbZ&L*u*QTO)T60^FaJMG$9s)T%iqo0A
z@ZO_=kwqZYi?(;_!ZEZU{kuhj_*tF08)JY!#LTaox{v1<)^BW~0k&8|5A#ULpaF3s
zi&Hz}6AWM5lJcmS?sWe#)yI!LtHUT_vPLofOR~IwMFd1MzJDMRZXbB!2hEnl15Jce
zsobT2t-aN(_?wq_Ia7wS_#Beva?G3|3LwD!;I483acBmMk<%}yWVn=DPkl;_jn};9
z$0$s9?0$C*yvr3xEirm6;i4|y);?mlSv45BRbncVs0gBix4FN$JaQ5207es)vraxt
zB*1i^u&{`=wKX*Vj$Gosx5^$6B-zZ%{~7hzXekFaXCH?oTD!!@Hqthb@RV=$y%+H0
zcnpFG@Y#j;w`KqdC`eKD_sD>)z@u`ZZ684=*CAA&dFyd{2Cl0}ecfUOIk{rW7L`rG
z`*pnh6?t}jhS~o9$2;;|2uQ7o?p|P_AmgXyD%U<=iTmw;bevSl$v90@fQ(ii6w@M0
zD$^y%@numt9O%<jFBFc%=fXEPV=wK|Aq3wz_tS)TUQ>g+%K>htRrVq2>qIQ#;@5zi
zy4K5AVj{Q^j?mUJy>)5c$#u~AWYigH@r6T#&?F}pToi?1$Ui}Qcxl7>_G+{tH!%Xe
z^Uq5}*;h~ztltzYS3Op+ivk}YwSbgM6{{}6l;BZO*Co=Jo$he+-Usb1bLb#Cgac`&
zvca3~Kq`g!lKUvF=HvmX3ZDJ@mdAOfn1Ne7b&G4g@Xl6>q~uUgwQESo-f%%bH)8;)
zPoue!Sk@_b|Js;K!Xos4o5G`S!yly_pJ!r!2!SClPn|T#>wYU=!X>*u8vNm|C^8;-
zS7=g3TCyR82|e67{H@&bO~(#@an3U;@ddt<(>z|pha`jDN*Z_-K4#s#Fq{m`m*ntE
zSKCt}!HwZ2Lk2=lgkWk0DsM{h<(%PY=?A^vI6$lEKRxN|R7ZWbvW6&JG}jP5JJX@&
zmwY2kY3+H7><e$LFw>-8YSFJ-OeS;{6+G7&6x;-e@7k?>2N{k=g{9-|(8Ge4^4?#$
zKcTv^mXV?WpIa6g-#XZD5ZRMu;D4-Pz<wcVybu_Zf`&&5eLa%-mdCM6AgK9ZsYPOR
zt+7H4wyl|))A7*IU-G3km|Zp1|1{_Pu;F9X5AB*eaxfA8;~=3{;Y6opejan6Vv%>5
z2_2HEc1TG&>dta$MzNvNv2GAHA)A4ge=MlH({LoXdp3ESNJ4T^RcGVC&kYVgzBq_j
zL6(@sxlKumQSh+)qlmb`#>UtUHA_vFmW9Af$swnF5)&Q}NRrjC<;fA>$$izx=~`^X
zD#2;70L0+)$97UC*<n%mdufvMO*@b-GiUFX7tw#ms`XFMHD^ARMgRC7Bq=On&?0=$
zKhdewucafeASq#iXKwfeJ~2)ZG0n{Uqqnum8>D-_!YiBees)PQSKn30Py`8*;pk`m
z4U;8uZgyD}!o4&ZFhLoRQzw(&vWhkuUBWejDERp(agVQDgtef-X5k{cA}V?^-*&4V
z+ruLu>#RzZR#cYDYYgaU=jK$6Wzh-3@6Rpu+?V(D38G`qvc^o22Tt1&s1B3sy`E!q
zXZAKW3p&n}RE#1>(e3c|h2N<ubJ}{|>Pj2h4iXsd!~)+vt1I$(DdU0En~WPFOCJGZ
z{%;8gOfpxfdS#>cVbYlKeFyN6F$`_EMCQ~tOWah-s<S_An$HBkx#2NdPwP;l<41kO
z!_~3S6SEN!x#s*mKfC--gBpOlXu*r0YcIDsHfmwK<Z<=`80ypB#?1CZ+e}~jhywu<
z*NzoMnGoj0>XvWwq|MLrt(>KKD+ZdCM6qq?b82$)*;PyvT?l2~(K#9x`~d~<fgmSA
zl%2h8ZKW@|B>fP5m0X<V-{I?dBM@#Zr+_9AtOED*3e)?tRx#Px1TUn0#m|&WlbTw(
zKkkZB-7YoWRlFK$PMOA+cgF%Qd>4graglBF!6#qOMaKu?;zjP7%1TVo;7FA=9p9F|
zzM%)uwLPsj2sQO^o+@^x?F{{C8YlWQ>#QU1f5*aAICaqKV;#opPgVaYUhV!9ZS_2Z
z7Tvwl@BExs$9zfOs6Mt1W%1740TZI3%<|MFj&<$=7FJ#Bgh|QG{>s-I1Z^F^BUPwm
zra>4Jom{)Rg<-aoleo_hIUM==dvf=m9_hiZY92Y${u@IY_+J9i#Ld9)gNGz-3Oq5L
zJylgxOwk_8sEOiuWVEMY5|@mDFh7hg4COkv#g3OE{$BsJK<i2r*YN1{;<W5s*xBi;
z4CjT~H;Dr#B&Asblo<RfNSuhUknkAaHyOW?U05bI5Gy&RWb&9IMNQ^X-?Ig!rQdns
zd$d^<IsW_(#%z0t4OM?k%&bL)GdC5j1#v0Ta9D8PVso<;hxB_SS86l+mIe0PBwihY
zB<CDTe`*aYL6D8He&GA(PkUp>?>j2I>ecaXA(<|(hsx<f+r|D*YhN8zRoA_H5Jg3l
z*FXdW3rPW`8<CRkE)nVOW@CYZf*eY^yQLchLApDo8>E{%kMH-rzwh2ZZv4h&=<p2L
zXUE!e&H2n{K5I={8vgs8<V%^xM02O@k?!OC|KuMR-gKxy_#ANn_({m6E|+IR6|6j#
z<`pKd)Mj&tZ$o}B^k~DIYv_4LMqqef(nF|>`0^{tJ&uj}%R~QiN_?tZ`k?g&GeXKy
z58wx2TGe(a44d*(b&9RkMktT!>T6UvmzrPMLDea@CO=8@OU*PmW=%lOJpb7;9YR_*
zc-6X4xC9qp<h-!`%J+a;uWR#j>e(mDnHgW6lPe~k4Rk1P8<**A&^0hh-L|<TFE2(*
z^PGyRyntBl4$+T?KZBaoKaP}@3=Lkrd3{*0|G7NQ{p#lQdV`bIu+NiWFC+HStXwsW
zD-YIyUVJSf{F47FO+l$_I`S*sFFCZLV$5ws9EcWGw~4>X+9R`N-#jSRyHszxR$ne#
zYf{m$s^k{&d*4ATeu=~B5x(Cm5KUO^pL1}X)zxjK;<JEKMD2LnY~T3y{*`P0(6jH7
z6l_Ej?v9-28OrCsd|T}K*@;B#2E&tM>p~5`dsXFU)S<;X)C`H{XZX)kihRO*1<oO7
zdgNh5<^Fp+MG~A9zux?{&O*tGdLHB$M>ERJKlP6LcOWceX=oNErhh(Vz=yBrGBQr9
z)ahv*{W4JFVfES;@#{|?Upz<w-h|kZVl_}n;~AuHjH!V!H{I>(l@*jWW;Y1(-A(DE
zE%w5Z^pB#<Mc)#VU6>A79g^Qi(ut}L9+lg$@wZq0%#f}kPnL1{JXFS^qAnY&V0~ZI
zWo6`GNZU@z`Me<m;a~G}*wwF#>x6ydjkMKCdt^sTnH$+Vo%ZWy2!WXVP~>CtxxHa`
z+4n4a@K<Y-?Qgsa#@m^PMh6~Y+5I37DlITagxrPB5q2?dk)1y|C58oSrLxD&1@k@3
zp5^;j1jbyw3cRk!tH?c5KktI$=-dGnP^c_LRJp%BN2@$_I)DAqXgNWFnx;V&Ki%<@
zFOBShU*5@OCOxO0gM4gZ?~t>y>@Q&rcPuDdZ|zvd7kQDur@C3f^_BOJ&mii}o5-Ob
zkfsSx3x<qU7zfAJyt7#H%s&N+v;5jG&t<Pa{j|^-^tnNL-9&{vI)WENFrcQ?_coqc
zij*?j@Y@f2*L8`VJDj?<xAlo)r&04fdr(AEAxSP&pM~u&XW6s>CHZ)F$s#o+BO$=&
zx`i$N%bXBOeb3WV<Q7%0@iV!uy&czE!{+ttKPjVae%DL#Ms60j5umR-FI(B?{##72
zUHz0|3gS#i7t_M@WJ15wKF@tzkG`l@e`9C?Q`U0iN<n|}0ZKs_<;K$Jo3Eo;EbfJ@
z4t0ryAvwQ>Bmr1fM~nDgv)-l|=2dwfKHF-ygYYHTQ1+2>e&5Jx#@pKHSaP&(YJz$Z
zu?Lj+`iVEU=KV3UFvXDUb+w|q5_rUwYATl>UGh2mLTsE@O_z*>_1#?6=|>1`8qJM~
zK~c2y$fbUbc;A8y&%&Z_-v`@*w??>6O06SL3$n-k@EwLpuS*MyQU~kn=WlelpEMv_
z^Yu^|4Sp}wl^E(y;+lAqFprCZ4A>YP`wYZw!e!;T3cEb--aF6&VPq|3B9Ecanp0v1
zyHZ?e|2Y}b(F_uR_aModTkc&t;yu;N)T2uqR8#3lc0<90-e=ijY_7fkv8k<QXyAi#
zslC2%c(9|&XbCZUZqA1Sf23c<dG%WJ_xrHx?HsZ)EGKi1kR<3okYu0=CaC~Ar>=Q0
zRKV&Z@QgaTFI(EEb@y#XE&xs#^B+0aila8AwrB0UqrkhEu1a_4;&_8YPfucPoY3L?
zd1$rEZWDa!mitcZ{Tu1S-O1?;>QCFd6%R@--3>sV&o7L9+MW{0fjk`my(&<LK1!Lr
zgH;MNdZ2zom0L}dH#ukAFIN|neInkbb>>B*yp3&ru;Nc5pqvb{G!ns}u1dE}AM=iy
zkUQ|p^KHG4d79M{*=mT~!SIPYvt<dh(~GwOB1qp45F509;3?mDkU8mgI<4wGg9PQ9
zufC2BPw)HMIa)ets4HnL`(#|}La#~0XJrdZUOLk24+O@(vZ|VV@_&u>meWBnQE9WU
zQ=a(<mtR5)i%$bj<c#a)vV1{3o#QJ`)SU=UasKx^lH<12@>Gty14gJQQl_-N;s&oP
zHGSXTfI#Fz=7E1g?9JO;AdRf)jz>gZn!kifyCfYmr{|MPHv^P4S~mG#zvKCR0V9Kg
zv6Wm1erh1%PR!~dpO-voZd6l0Md>)1p1q{e$c1zXqO|y|-#}gf-iy+>%@oKI1sK?l
zktQ@orNrXETER0?91+wSE+UOuADaFsX*2Oph<_7v9>LHv7jy*#I29_wER0^&`HN6I
z`6<5ZE!R;Vd}7=$b;N6ZAx3mvVt!QKApBj6gudGj6okVF;yr-@XmIYOC7oPpj3v6B
z%gi`A#n+#=w9FCtyII@qBAio-JTYftU;*dhU*a^hq2+pn4}*fv9vnQroAb79PZ?3I
ze@}w2Ep)ZKf7?xIEctx2wdv=F?J42M6=veFUL$m;Gnq5sg)eI#H)9STim7nrUP($Z
zmdIH|P;(0G{`tdeN>xv^4VL}d`Fzf*xZZyU+$pE-+Rf0Q+HMzo?Dd^3BTlwP^_>kL
z@IyOwb(@U~o0o}`p?;On%H>ZR1pG)Gl83B3WS;#@{5t$C54F$_DB%;n7gJobQZQzb
zUgN1Nh|tJrZFM^B3@xZ1Qu&(qVipg6BVV|DKQAGlzhTQ{UawC`v$m$m8|oiMGQ%#3
z#TgM7eRyLe=nWvfbA1xyP+|Pk!Pb5ntiD)XE$7$3#meDQ0fVE7Hd3o^xk!>+>PwJd
za3z2GQIoN%SPZP;au{Q9%~)%yLYif6ULr!;p8gM=c6UVlPm-TFL$J)7E}C=9YyCKr
zWUusStB}qKWUxKe>dngYpEQfD>3-<Vl;0TfPt8<kHjIEQ#hvuY1&@B@`11R2J7H-z
zDJYIA)=9~cYZt}aG$;3uMZPS3$PR`Kyqb^Iikj!TIFtMpWIwfSFJ<p+@XDAEs0K*t
zKd0p4suDsrC@pZltF7~Ox2*Dj>qdUt-b{A+2kkulSAl_D2CX>sNd|o!p2RB7yn&Ps
zcg{{!s4^MT8OWHqW|Os$F<wYiYu%IN+KW%Cx7I|U?#Q_ACmE=q)K#XbJ{I*v(clOy
zTqeC2^najX{J!KR-9x)i9IpNJ)Q+SiEWk(b{=M`V4<KnH&3qa5&pg1!-f4LF*m=qx
zq_cMQ*f$zyB*sg<s<K9^Avc&;j_y@Wo7%UE5A8`N6n!tWrB&wNtU74t8xI`X-WKAd
z#cp{6JM`1%wTjPfJnrPhcPhuq-zQ|4Li)O{zgsGvJ<|5$d&KAK!uNunJ0h)|8{Q_T
zvXcNf7&kAi=NG=*{u+C$&nn8ofIJ{qzvb}=9o_Tx)+vSQAe>hDf3$ypg`q+GhhO(b
zF+CMDJaY?eRLGr*1IHpnS=m(LQ%NXV%+WNX?<?dkeWbi^B2zC~Y|qE_vx8UdYOejf
z)JSd0cRh~D-qG?epb(hlQs2$!_}lK8k(xZnmZQqv4|50J&`0`R$FHISjE4q|Z>UAc
z3qAcLn@<nJ3o+7>O!dE@9DkXY>^e5lo!#yX=MyO5aU!~ja@2F~D&EO?OsPK}ZmH(I
zPTnP>py5ic$DILP5L7<5wDMkqBttM$X>;qVf4RD1Rt$Yr#lbb*QU`%7g^j%|>`y9H
zLAoCsS+3>K6og$eBI5pLh9A#r8Xpt)rtQT2_TJI;$m8@L<I0hoix<~NJ5B~}k!%70
z;RIb2vXng*qoSo6F9|5c;tr2c@_4`1L#_~zzi@ZP(j^iQ92L7IxBcdhj2t-K7XdMy
zHE1yko0akaCuXLb+B7%q@5`OVYVONf(Z};-BqAAi@v<M&vWsLTzjm>z9SQx-;{r;a
z=evqaP74!26;JmT3+v&Y1F+HI*6WObiM;{V4yo9rs2DaHN3FRB(B617ntu*EHBkC+
zJ8so(hsDm;{*}^GlMQh(l|mPbOe+A6BpyN_Ce%}KsTdoA$H~iaU_<e$SIPQ?;{M}a
zpLPEgo^cVs4z#|guOtIf+2mIYuRW4iFPZ@F<yv7xORMdIBnr2H$PluYq#~n_1-_CY
zw*Dv^Y;5+>!0$WV2B96|4x%5N;)Off8p5qEjZ@NBdBBvavs}{|;^LS9mBo8G(U&?`
zF~XRFh2^#OSO&hDIA7uvu^{irdj?Owh&g4rsaN=(fAPh^raU9_!e`%m)U+h{4MQfk
z)I&q9Y>*koB=;?9b?HbQA;BB8_v6+<nAP7$O#9#gqj+<1M5v@_B88~2@vAy~2JYtN
z8f+5g>{g$#nY=32Z-%X$S18Fb)hqz=YFq=q42{LKggD?{ex`__n-b(^KizKIZRgQ6
z7@y15q~u}ni&|}UEO%2sa4i14HC7w(T<jOpNqYD=xY2NJ+9R)o2;K|Ss0EcXyI4TF
z1+|j3o#kf`W&yzr+Ecvy!P!KK+A<yvp6{)C^5P_XL_&0CB|h14Zk~FGb7;!k=%e77
zRH26&ZB4?8iuARkov)9h-32c>6gfw;LD8v?!O;g+OG21BnVb{9dgMxdh5<351z45)
zn3239A@FC0sTeH*%GUf>`?nD7m<m6KiiVPYUlEO@n1nzbC=Cy(EseI1_^znf?c)wh
zd#xV?in8!B03!_`4N4CZ#w*#o?3foE^tOTH2C=ZH)4qUwt;#=cV9(+zhTM6c<^0+8
z3pAy1utSI-d@{s_aA-EucJ*dU;s4$-9bxAU+o60Azeq41FKNQacG9R!!nQf(t4Xe}
z-$!Nff-0c1ZP+I}DolLMep|dxOInN~Ip+h}YHJS5ldLj+I&wqbh5nA_2=7lXA*bh@
zf*DRiSS~4{M!Xs9-Y}TnFG}<1ernFVFK?tNV)_+yL)yG7=ilO#Z@8C16q_Biig#Kq
zyEACFbK6PiM?|Gb0GqhH{DHk;0|P^aM_%yfnS2geyOp%;M4)gElck~Dc_y7fTMszW
zxo*Z)!|v~G!LsgS+`}c<=D|0^ss@L<0FL}D#;_hato;?<L%^@4SDiK$b{ilnIo6l^
z1JEB^n(i^u&MAFTxL-4#Nn{}zJv5Q9x8hdLt9R{`gz*CuC0;Ep{B?h7@*@LPIf1Ue
zY0Cf^KL!{~x$HcZ^Hb~D?arL*O|}}pL>fs763Ch5f~B7X-o$pH?b=oRiO(8Q+G{vL
z`o-bEsv*)2P}_zwL--bAHl8KBdGnu~^{m?m47o$CKm!%REBZ;!6~t~_79_1SRpr-K
z3RvZ>zok!R5VQDx&V2$e9DpUb)}_c>E_NGdwIaD+TbD$97;wy@pa4|&%%g~sZe(OJ
zBip$s&{R^cW4P3+!N%oIYDvkq`P9;4Q%yTB4s!S2gMbgk?66?*w}V6S+iZu(nTLzb
z-|MFPyq8DfUVeZ1@`I#!1DNs^-m&5#mJEHSr)s{l>~5wfh+fKB$qnQ~5oHh!!NQM+
z2_b<mj9+|6e^0cN#KJa;2A&H_2G28xE@bl7@65Likd4ZwDhNI#zeW`8A(xr}3oBM!
zk@9V+596&+MHI?2_-42xvEl6_3Yknw{tW{z^{~O=rIB~5{NjzNl~vo?6_>(Z6%~CZ
z;4spAH9ga>66?Oi@y%$2y=d1zEI?i);Xhys3I1Bu)_Wwg%Nnyyvg8-Tw6PZwu^pA6
zszy#iiV-^T4U8<0L0E*Gc`x@-2Yb%^<txDuCA#^pG@bDB(aij~1IMt*5u-NIZ4P59
zX#+~||HM3gr30#JfF7~2&u}rPMXS$c1)CU91x*6iD8j=^k`CDq!DT)NsKr18*iOqH
zRw=xBb3jdHWYa24Onz)UNwL2qKW=u`zIW;?>kk_R9^WXW-T%MtnmHJP^ll`Ov)4Yp
z`-GP3iE+5BHK9bObSW>4mJUl}*%a$JEq%?X+^JxQSSCeY{*Cjr={S@Q;b?J1jyJY+
zBz7zc!lECO+S6>dEaqI7MuM12vlkEBQln<>y*seV&2-lapFE+jh?h2-N+4uA(MoYg
zZl2P%%VOm0_ms18=Z>79@P!VI_Isa74P4Qd5n=T@J^z4e-|^IcNi;hw|6ssb3JO@-
za5c(bC?+p{?D$<m0@lPrg4;tQT6}za&xk1vV@_Ds;Ou7uXzsn3&?S6_ibF$OcXM+<
z!)=L4fPeg1;}B2d=VF{FI+C2wem+(BCaNtdX;aV(RdGm-ZiZP|)6^nivLVx<-JCYJ
z1gGpCny(t-NzV8LC5{L)v;5%8`=?fn#;$!b2J$T9{B$FCmK|!}#sM8>xy61-IrAX}
zouo;Nh06{`Zcr}!z|rL3Muce{{)4~6Bm)0PrW$$k;y%Mx455L{TWyWEr#dNZClBXx
z2QgEe_~VhA$=QoezRScTDelD0V@+uDOm%-*xu(z?D~*8%KWXTu1m$qyMcZ0*mtHj3
zTPMEOa}ky#WX(G^-N3Cl93&tG4cjXfO9Q-D-zXwtAD4^&g@WiWlPb3`+K%**JOMEm
zaBgu0TK2T~4&|B=pwPO{$)LSv1sXSNz#=!z%|Y+jT^5a`#eHL>pgZ{x8tu*x?i8f9
z`>szduC#xBJ{_U@j+^tdovs%XZ43*}wLSACd(IZOoj!f%|HK0r@9$_(&y0flp_GzQ
zs21Y3>?b(^;|kmhH>>G~zOQ6v>KNy+Fs*Rq2jv2b6tD$+UcxU7@9Kgxci!#{BqL?K
z;PHb}KxS#Mfik&WK2`4b&+3!YQ1Zmc#E0=^4iDkwz(T0`UN0!iPj%0szVbM}fGt2#
zBc!))4Ltb*Vn3yKA`87DP=rfH4Cqk>+7CVoeqT<Tn|~W%upRzr-Cc25^|Zz%7E+m$
z?8=%QrhB0N&2w2~_vL2%NciT=!mlwrK6R2ak))xZ`_y}YV`aA*4^<aQC4Q{8V~##R
z+4If|{8!ZL2TAd-*Ym@bq*!BGfb<y+cKHJrJrik|#1Szk`*9t}lDUZE)fOj-f4*{E
zA23Bljk$bS#-RuxpdpZ6<$Y?D_$FESb4J09y5p_+iP!KprW{}5@@e5&SRR7P>*zM|
z>ynQbA!vZ8WN+R16v+e9)YWq1(XUX|%jQ)Rdqn2bvsdSrYN2Sb`@-|Ph#wzrby<V7
z4G%(@U9&QGhf@mhGYG0WX=BVXtr#S4dBqp705p9uRS<reY_6TQe-Jn!LUnbhNF>#)
zoHjz`*$)_voUbZCfmj>&-2irO-W19j_1~_(-7|i1hlW5b3E3h7t<qZer+@gglTlEt
zI=(2RbH~TzE%o&XlSTVIn!qOK0&!Y;o#pXiQF@yC+D`xWC|3!CqiNoXuLEV2AKwds
zrrf21O(_jltK`ky1!~i1=2u(XpkGh=XobGvi)zdYl%Lfe{a2!2NKW@U7x93)@WJp7
z%)|ZFl4J`r?j3De8I3n;nMo3{FOJ5!KK^XUQs!b=N1T41tK@aKcu_`8Hg<Q>h`%W2
z<_RX~z9+D*KdOiX!Naz3vRt~V?frLpI{xe`s=tpn+vh(OF|+6m(lXrlfBv=X@Ge*U
z)FMv<=(`+^+wTGHhW!7k&Gi81qGPrJ=L0--smS{7A@3^6Pjzki?hOoBQ7a>_l}l~o
za`KTtLS%1dOiJP3*7A$*8Agau-#z%T)ratIfEY%(#E+OxY8X3b{)YSl61$kgw5kiO
zKfiai1YY8Ef2)vo$Qlew+Q9vn78*DH*pevC)@t`mXOme_VR40(AXrN$v+XHM#REg%
z^nSajgoiJ=Wk_bKFDX58L1u(dv&+0c%Xz$>cl#I92R1vMzPj*$)ie>3m!c%;ed*v^
zjuyRYsM3-Ww3>!Ql^x0ctSi-w_Yk51Rm&HP!J6d`M|c!CjG#a%@sIdkuS>bwpMPq@
z>uAaftQkg!$ZNv3v6*;bn$sOs8XVAET;;I75W1=IHEezNJL0fdFyyXKL=Ve6FM_|K
zuo!EEjQ5MWl9q`ptTW6e<814pD@&E3QQhdd?RoC}g59T61t4H{m9ns2EW}iT+>o`{
zPpG2omzi#Bjd<*BzYnI1)c+(QM|0167I7*`PVXcurC+>qWp=w?+W=Y>;X89W%dd`a
zBQeRA%_Symx3&({1C<m|7>4=5QTgwpf*Ppex1brnGcpcdB^3t{j^fRtk2pLbF@3Zc
zZA+K_WwVEEJa@I@D(cA8?AFSf08sc5Es~84*%QL%WU$`F;$hfZwY@x}UuDG{18l!#
z**?Dwc@}!oX)sMdsQdCo^<QhVc<abkb#2o%fzZI7xNE+Yi8+)MzU3RZc$Z(f?v5~V
zbGXjGjXH1(2@4S4B+9b_A<9s>$o|H%&=_BAs_rtZKsDsM0~2t<)Epz0e8ztdIB0hC
zmru_UzqfY1`8*|p){<*_C&_iIH}eu`HFjh0m$b#fVU9JixE8uUKxH8}XFvVkjTL7i
z#H`g#%^;_*`r?**5lX(opkkj|{671B?b^gjZ%i&L8DGtHZTdtgU|)J5;CblS+1`zj
zJ@yzsr>2<&;iJpTN$S*$x{<nf1lbiW@3eTrPpa=&>vTRj%X4`Db(?R5lPxhzQBHdF
zcx}9+{<zNUr(5;X2m^0_f{m%MSR&dx?Uaw`7V(Q$hFcy6<lN736R$Z5olDYBdcNg;
z<{k3(+lF<IBFUSu&`_R<!3+3mMyf=fo}Rb*<XENl9KKv8O5NP->XH)^7stxU&#xO8
z*mIXi*Wx`~W`Z7a4bY>NY)j!Rf3b+*o1)i(Z+6G1LYkcp!jZN8&7V(ha4IdTehvF(
z*@veU#jdrMd*!|7?^W+!C^W?P-IAuH9CLa_4AK?XdA57a)Q=`Nw@LPm8nV<yH_93l
z<dSxlYU2i(ep}g<v@^M~XN(|UD67<(KuWczZ^?(cpJ%P&IEhylKV?i;@<P|N#YdBK
z!*%W{tZhn_L1>J2Ru~noUF1+_?NK{bH6#u(7v5ZL>IK$4weJ?_J7=Y%>TGmHbd58#
zH|QR6dExgTH>Mq*_ouy~ru*sg<^Az5{N-FWoLTnv&Yw4>cN?QgAf;fr<l-=uxmx4A
z8=$Brytx)KH9PV2s5D5DwaSGb{L3(uaAhRn0|KAc5X{QVH|vkDpPs))hD#wVoZ+}a
z5DXLtLdx3|E#l`N7Aar_!ua6ez2egKo_YT*$Qn9zFE8KvS?&9}E+H{@?MUclceiMB
zTTNS&_pFo@-txd}U-!!A<fosI;z4!QjlqZpRc03ok?*w}$O;am091iufUd716Ei;`
zC|y=tIkLDB5Qh7dGUODo<E!qBlDsYqD*t*|!~>j#zmmqA&{!-WaJFnsgr)uTZ;dJ8
z)TvO!e|}2}ITwhD+^AT`4%wEb*BVy+#wh97LU`urEn(Gpegrqj2Q3SV;-$+_rA0=g
z^LJj3++N;a_%XNVx_TYCU4BDszV#M5kD27R6FWyIEa2gf<a{aH?woh3)Y~rmM}aV9
zsPA1C^iXtgP(eZIB}ndXbYK5>%Cskq5iG>Myc@)5(}B{zLTuUIuwiF66Z#}xCO#=6
zAMb(gl$U!!7JX8^@Td7}Eg$F)veORPnw$CqcP)F6Z%vr>JE~ZjOQRw2p$nP3LT`3Y
zU@h|0ZykZveRdF+e_9)S3nb2nTEYI~%u2_OZ@9QvEA6rZ9HF6T1r<TS9IIEZd|s9Q
zCxTRo!HadKP5*M8=#Dh(2w*Mr5UeAA%Munr9Pmp^{<Zi@G3X0T#uv8IY~-2K45=IL
zcLX%*+CAj${z-mbM%d!8wj{ga(APIjYJmM@M3gc0?6YU&H>_UEBwUg^i~Y#ko0&xD
ztgNG3ucGomL55xCrJeW~`tls_4<xcmi5GCM2^n9Iym5wK?b)+S@3tiLR}atY2Y)@2
zFe)*7$U!;xiK*%^+`znEXSNZE#PkpGpp2XN;dETQX;ATd5ZN;OeskvLjZ+zmTb|AZ
zCh3g=S^cDMah$iV=A;aVv2wb+n;6@Y4+?7t476R$Cr=f1T;bj)lhF)g;XI5O4{#U5
zR%1?I=XQE8MamnH*b#HL_Yz)4nj>a=kclAExjSbpf}%oG)uUWwu}_eVua#<5{GJP4
zc693VNU42Kbg4sUtC{)|dg89xK!JLIHX=T{>h^RPy`xG>Le+bx%-)~e?Re?6SZm(O
zyj7x}Ub;b9p;i?cZFze&z2fF+=OXOxs;rXoT9Zqep9XW$b@SUbK5`F9MSD9aqh-7N
z6<981CL0{=bfxT=AvY$}k1lA27#W*wT;(a#e+#v^jPzvgxP8P$;X&4-nXGk;Eh&RZ
ztdgwk5IQWapi!i!NHqYrb#rGn6P@7icBr7oUHxEdtxNRso3{m%{cpv%oy5#)YiXF<
zhr^3R(ZV|wX*5%qZACnkCogCPTGG&p*G8X|p?wn;>-!L-Be=|>|DDu|cwia&6%(H3
zXnCsAsJl;;&q6%6$ZK=Nsg!j^D*rBr3~TM<t-6)?hoXM2MZ!+++kB#QH8-vvAI=HB
z_v-7e5b#p46O8HG+}-Tf78#`pF<N9ucmLAO%;4;Mx;&WNdmcYsQ6aD%??b9*AZAb|
zG^reAzhfh7<>@?))TOb#dIwucBm~Aq(GDa6UAh%6h6mN@L2H)tg>%|U_F7uoUgzk)
zllC9UDi@l?(-}3`j(aN-DoPzYa|*;)?WrXt5U!O5T+gU6;j!(u6k;=r3p#&gp|QB2
z^sBeSQs>k`wBEQFTI+DUabt7FM-X!dGO9PqsSpaCcC8=v$1&n>-fq=&EKTd)A}lz5
z%TsEsVk@2)!JX<*siGlC{o4C??%>uSlQd15j)2>SO`Q3GX|B}tbd*CGTqkv0Bzn<N
z%9mo?G7|o3UQRe?oO~BS=nf9MaWCU`ssC2D>qiqb{bZnWAd&C-KI&rE`ldZdaiT*W
znW{HbiPmDd{%5nyMRr`R8}bTGCz<U*X=}*I0|%u2%ht;lMMm=0BkNoZg1rNJ*j`2W
zok?%{sm!o`>sIT@TVu8*0iMHJ=M{GQxB>qU9KO%!M2E7kM!O}(M!Tm%`SsXnPo`qC
z+KrKZ1OLub&F+(~S+)J9X5*VzV-e)^o5+fRTZ+-rQ~~9&LavnoF7B4oOEeDMp2k#_
zT9&DoK)ckfWP=93JgPst7WVMCvRyNOFO?i@<{gsVnv_dAr05t`2)oksGK-PO$pz&Y
zWM`QUC*jJU3F=Q$mCtlJwBtFh5eX~UGLan5QSD|kRr(U0O@Hx<lDST-FJdrP9AasB
zl_)iYQnJUe@6x%EtPkdymzl<T#!uF?!>Wu6PAc`^V*V};u6W~JQP6)a(2!>{g0LGQ
zMm81B?c|z>9=^B@W*gV1ngTVKim*lesCk&=dBT~qIy%edU>DR9)`SFOrgIF&sRuPR
z&WF;wi@(A}jB6aM)V?7)@z?MvO->i^h&*G3x0lkj8?F$$D>97Uv?(|`KFaXL>*4%d
z`+X~0Bx%uKwrDY+*GjGQiIym}NdB6Fr>SQw#noTl3q7JRRWo5S{_IblZv4AvQ4r={
zdj4!g<k^?LrP2CKfLzq5>adhvW;p+oaYl0e=T7-4q&}ruHijv5@Y~@&4Lj9y-ay&v
zb0y_J5}kD>&sIFPdLz-TV6Rnq=6GczH}~3~jinXXaJ7m#H0ec4=dY&+)i>lEvnJ<t
z*<i6yQ&o>IUhoiM!V|dn8cV4GW8|o8ez=X4bx(A{w7;3}vwv#|S~A4igV1*Nh;RL6
z5+!GPX?qGv5ubjG9`xn#t<9RZoX(4Z%a!ilxE3kAE5k;6iWMw5CWM3RM*nL6M&t2@
z&>S%wQGqLoJ?BcgEfV|Yh~{K#@0W}>Y(U=3r=ZB~CpT|vTp!}I?I-=xZbKsJ(3vk>
z@;!ZsTz7YYo&{Rpy+BYCQ7h=7J+&#qv0F0MdHLs#QI7ZF0SDdO>@@vW4_sT-+)50h
z>+KEIW1Y}UdHJ)J-o$<B$-E&&*^_|<+Rpk(^_9$;>4oM2_ZAB6J4-X&NVd&xCVIYp
zZC$D%*4EP1bnM0K(9k~gZB1EocTe7V)^~Wjhg6p%4D%?M{yT)9@}gtDx_DjjC+DfM
zkKc$SQ0Lp-zr&W&G1HYmg5xqu)SYL&Y1<kvZP?pqR>~b0%PMs{-Op1y`&TC&0og4o
z0W?g6=0z`K>waG~E}i@n7TcUR$46U7N9UYv`);S-`Lkmb3dMcl9D_Fzv^LCI8IwGe
zHkf}+fTZhbP4AqZ==sea=ZovyyXS1XW!8(%Weq>7S0DPl<i2gzCurp`lGNsZb;g&C
z{BuQl=HzOzq7K%Hvq>jmSk+=#*u}qFHKP1PYJ9GpSS;2W-@cjTXTRDdFrE=yJ5w67
za`UKMeb==@&Qr&5Oczm9dH$S*GE-B@Y%~dJci)q3eYUjF9qvih15^R*Z|vRC@WA{F
zInPErM^j5{X4cogUrUMrc<g80A1vMWkC<T7zTRasE`rU@WHq;S%TSTc6Y#_i^jfiH
zA@1qVLd2QK%C*_HmSy8ZP2SG&s@)ZJPCZq`#m6h3UpXC!D-kZTh@T=Naz-qEccuCM
z*_#dJ!TBK|BUWK_SkR>Q6C2CH4<oUGJY}{z_t=};0(Px>A&XrH+tX`P@jV%)nUcJ5
zt*IgpML#e0_j~!B2)%XD(6KrQ3$dO|=sQ0034@qqwh~9n-LCCyr@WlSbFl1nXL~+H
zJ^k^$zaM82tSoQ=`6f=^c~CUok!-~`zukV*$fS$6b+9t8u2}!-)iCC1P9E3I=e_CG
z{^@o#4#S1wXwUpo3D(+_b?(Uf4|=%sqY9TLyBe87aq%;i6|er;y$dCCUUpsSOZP%x
z?XMh8;rF3ZbLadj=#rw%eHN0ZrzrEOt^B&Wx*@hv(QqRgD-32?ZAh_1y+vZu=Ka}B
zI`819$C~tt%*QqPc_L4=$*yXpKJrZ1`E}WDPa^3GDNh<s4S!G=K4F9HvBSM<Ax2_k
zd#=L0Pg?WZ&r7>%3iMiqp1Ns1)}&}Z&tPz3uXP6Y0P{ZOAhdh!)MkR6ed!B%x9x9@
zlqrq5{Uc-*S*t4CK1UytKCpDtqtxq7(Hm<Un;?IG9lKAso9)mW!i@NRU-xHadj=~v
zXw0f8_)SCa1dBJyN)6p}x4No}gBT|&@&O%>rdc>^>XvKDEfG>Dt@w=%+I8oTrHu==
z8^1`=31_qjEw7wJdXN)v;wq**OV^Ddp3$zGJB^5lK#V1<f8*lfPM1~(BxA2j+$blw
z0j+~@E)%Pe#(&qq!cr1R;gOKv-`|HWF9EF$$<O2kF<X*#*=*8~&Iz^+a>U)gKi$5T
zDont6dv&9}-8`hV-ifM0Rkm@>ph!L9tdez!#_HS~KR-X$<D*AXXK72^x$ZZQk87vZ
zp(WN`5jb;~g-Gph_sf`TY;XJf;or$#`c}Q^deYw9EUz^-*lrCE{;fZ5_ltd*CNIHJ
z#-(bebq1MpgQl$e2hOXjHgXyzp;dHUF9u9EQYYN^1*YoJ36TxbN{l{Rn;!!h@dW7T
zKW12xDSYbIPce+V60VbTTA}gFV&-x?HUr|&)=zYqBpcR@&M#%M-U-LMX|!(asZ$({
zkLf>z7e5*OaDvFSZFB#0<gI|KtE;i1cJ*u8r=tz0GfxR_^<<{w)pmSyJDka=8xp>l
zDiRbHlu0{WC<{}Zfa32K4^K4(Yc46g7XMD1g3L*YwKnxNa5*SDMk%wAKSv|N+f}7$
ze|OA8#axM>IG6U*=K3nk{^MHCM14K&tsQan&Qf#5VGntE>}ttx&;OXMZwOQFyK%uj
z>F2fM;|DoP?K;suoWF*W9vFoczj{x;ka>8j%ztvyOYo<;Gzw)0{!iw;KCfeybCE^z
zyL{bJV?-;XFF7HYs}g)IU(rc|37=|0#u=|DGBzc)-6eGaqra!r#Da0V5$S8ryVd6X
z{ilxn*R`EAwX=p}Uf)5IY7oO%*_C)Z#XJgc+>1(;-KhdZp$GP-_uTfn8HOMoV6E*i
zIptDCDvjRt;{Sa<6xlnt`Ip;g5bT2WP)e4yVMS;2F;};|)0oOZB&X%KvcqD^Lcp3_
zyS<so<A?w97-w*`Vu%Ou1g^vLKEZyOZ{5sFH<lw?b(=dOK3S@9yj}V;39CZ%&I<EU
z8tgALs^+PqLQ3@am&rcB1%<Kf%RF~hQ)4S!)JP%k`4fNa{yLW(vO(k;-+R^ga!7OH
z(U$GtQg47?RaLLmi+cu2!5f<!<f2&HOU(`K?URBza=${E&<QaeGCl#-V)R}D`$R-&
z@&|XH^4`05lk!av?&fAtbZF{?=ViS-tE`t6v#@bCY(R@vV7UK%k~v4MrZHgSGA>&B
zMNF?OTwlxEA?fcsL`}p}{ky?zwSeqDjp4lFV7KnSfAEEB{uOIbWZwTYh2_3+?$2J_
zaEv1ruGYVfU6;90{;!Y!n-gQ=FXrd}|C#?j@Bit<aQ*-1d;Y&W^Z#<<N~QDb3VbJf
z@m<~BnZB=SUW6rqr=3C8))=5+XZlq2vvcRp(L8z-NPh0@*`NAT7B9vg@$qS+r`Ogj
ztgfjv&RA13Fi4}PmzHSDM$N76tAKN*fdioFCnH|%B72UEmR6JoeF!-<71dA7dp0-E
zf_`YzU^ONdvzF}Bmh9xIeQt##B{s>uS5XQ~C+E^Ua&nNzD?<h1(C7LaCvMsYqcB&Q
zw{Q}P?!W}KZBM*DDrfrjp@~t6Zrc!cPn~tIJL^$cXvcPsNym0n>fX|uR{eClcCXpi
zaCK6wr!0&uamsg5>(=c&yAd}Tv>u6uaXDmhI;@&n-A5@6f7A0>Etw+|?Sd9b+dcBq
zsu35Snjw#0)?G+UNcaiO;z`)d^Et10sJ&>5;Ba26*d-Tj=Q}O;#3OnM-yeHdEAy{`
z|L+@TEwg0z;$J8k5^_q)C1{0*PC%fQ-JcN?L}dg=MrLXw$BO8l9324`v!R|zL6q^^
zx;jA!Nk1Rs*)0!rF0!MHU+cow|MQ*k=mh=NFzQE-THvtqi~lhNg8%8U{C~D;gi1Rd
z8|9(cm!TTZHs%;V@}xUSQhvD5%!KNy$I$FRuC7|4Y3StC6kPoYi*e6$x5J&@f_|;^
zpFf|cPhq8CL969>;m@TpxoqREdG&sGv|B%-dTu5e(eUzWjk_Oa?ko?+FE5)42@9vW
z?5+|szWNXs7nk9Cp;}yK{mi-=l4R7hU>oIthg#DOOX58}vVkU6*#q0qNYMN9XGK%f
z3}{eJuB|;LhBY?fx<=@oZ`7R>7Jy2jgAVUxF3V3IK72?}#tZG};2Q@U%?x(%;*5WO
zxp(v7*$K%#g)<;Tb{Kan{pCrL;fob>Q?=MK7JAh)XlrZp9Ie-uA{$M#1b_V%^ZolX
zQVwfD@!&_focBG5C+FspvnvnMT88~dxsoug0%81@`Z93>uyVR6qg~??63Ah`EI`ve
znZuoxnWd$4CWAJ^M)X6t>t5)VolTon{OmIcTOlXC2d6KKBbFmi_~(#ks8dXb_10*Z
zSX--QYcv!Hh47ylBmn*CjXUGspKK-bS#8myMCG8j8TkUV+Jmipi&*hIi1SPb8AZFm
zXnIR*3;-)MKE~_K*4Cb?|19ryg_K+rI`}HppPhJU$O2~CmTURznW#a3*YpPx)XWn}
zd7qnfX!6R-XC1G0GY4Th{u=kDDW>W=BioA)VaZ@qTCN7FoDgF~)fN)nJ!y*Ru<f~)
zOXd$FRFD0~j~|<0Q*}qjVzqoj9lP5omb%Ay+w+|Q=zQe#M6Q9H93nZV{j!Q~3CfY7
zQvE{pQl?7oGqB|XbU;9F3_rHrd<TEOhRtU<h^qE+hVS_p9KBbsQet9|r|6zyE8UgB
zJdsO;^gTN(!vwc)XV7bw8V=;>=#gb&SC_f&&xD{8)GaE4cJVwyLYa-w*REaLnhm#D
zTu6>qvbD_zR1z2v5T9LjqBdM)p$xBSRGbC%)Q(X?>MDM<`0;uzzVk{!e^Ysde6p1D
z&Y*#+H2(Vi^H*+n2;d_;2_HUO_)Nqk%%tB!WjE5wHlFeEI%S_JJE{Qs9{SdQB6=Wt
zus7wWn61U*a~&(iC|N493!K>Y@$NWb8817#O3}D`890gi;6{DAu+}MV*IcSL!5GKu
z^JFr9WIR3a{E*y7MBM(-&|m^5Jb?SgG*Mi#Kn{`cURkW3FGokNqH@m8rTQ&jWBnh9
zEEW!%V_FIV=EPjbWVR&~M&7^Fb|DUWVX8*l@=#<o9k?sH*q2E%LqCxSqrAP8T@_r+
z%f+QGDvEiba>obTbkA|BhyP47p(XuDTDVti96T8qAG_1zT@yG!OdJuP5)}6cq2g?+
zpRrC=&ekeDTrC?df*3Ggx|AhvZ(qo}H-X(6$)${G8m}merSgcCRjF(6!2y<IGy$!W
zA>5kxr%ye!+2wKI${PFol{q*#ynTGS`A+wxGqtO<^7LEf-r`)&@Vyk{J4J4SlJN$&
zRdQ)!w9G+00tb02dGYdPJ8d#vSI1OcoGZ7fL^&PTV{Ts&`8<6_rYs>Lfr^Di(Y`G5
zIu(`{6+|GIW`X3A(Y%@vHv*!gqw#8GEG=`)%YO@x+VyKTH8*#|%b}CHR80a8{*=gD
zDlRZ8%G~N2ie+XbTNXwwr{_scP0ebv5alcE2~9*PL*<?*F{|mVC6hB+^Ak01(Fsti
zq2cB(?|A|v6kxkkekbDK!c#kdrXGyT<5k+|Ju!j_p`+vD-h3l@(XNH=WO34ffB<<@
zZq(YOEoZ^2#%rQoa3VS>>}p)0@rfsR@hwk;ib^!Hh#SvVmjSSOO5N4f1&)kDR1SgA
z_ecsmhJqk7Syf@;Xe^8c8V+Qt7X@cCBr~DPTVX~hs;H!3?umw`3H<*f8C!SX>FH_4
z)4S#ND{7*0FtwZL-eAShVBWMC5fKq4>Rq|b9s@?TyY1xUq*mo#adI@rCyvf4p2TG_
z>Z(ioih9&h1)UMzA@2-fAj9~ON<ae-BYD&ewikQDcw7p}3AnIGP;0;U=c2{Y!k>xt
zU<j2ye)u5M983osEYHqYxHHP+Zt8{QxsMiA$4WM{E&g>T=RCr6zCOaiuD%q>>!x;{
zLW0Uc0v>Ve`z3~T0@LbC7$*y}kgo7{ZOLl!)D?s`2j1(02Le+Kej(&P>g#D9Kkk5l
zmSKIcWyG^L@wVybPbwRU)01NsyF~@xiE}b501_QDp*t(axWZY@x=xM{p)D-XStRKY
z8Hw2~K^}W#J@yGp>k%89vUmu6@9sE~UF(HU`{<DbSY1E#uk0yzvVriGQ;kgi@L_S;
zG2LG3m-gd5Na?jgskHpEG=%o0Y8B1}Xfzt*S|HfjE%u1^L&wvhd?Q^uQ2oF85BO>F
z-^Z%|!k_<lVD*0jelq;v_R+A<`>1)pFDX}s@6>jo`RL+_8_XFMQVgS!autJsV>Dp=
zyPzp$%-65CSWE|GAl`U;d&^b1A8&OCAo$cUpU>ly1sa7ZiDX(qIR0h%^OvQ9qi;)#
zgab8)UQ-~A#N7V1>(@sQDW1SSOjGU0ZjMu%SK$I<js1)8PmDb@4ed#hjn8-8-?&zk
zn3zaS2co0?tX{6Q@&}@7IWJz<`;w}6KUsPg_hw+hRxKjIo|?wghaNan1V!}`Gm?Qa
zGBPnyJTAt&tD{OHf0CrYw|{=I4h??Gnl<_O_$m$-(}_8(6T+B{?nC?zO-coyO3LG`
zz-7POfqp1{<FCJ7SQ%pmX`T1alfMH*Q|rcY?;cdt>}c%EbZ%jq6N6t>D=<zBH!n?1
zdhmDu<vP(m2z+ZdZr)t{;{nKcoCB#GXdh5girhhta?ML-ZVCwrQ9phx1Cym2JaYEl
z-kx>QXsKQL;X8F|YHD->7?#OmPipD%x8@rV!Nw|FmR5?!mF(?zobq11dSy8HO)reo
zE-kBckw!>RkP1#^Hk99)fQSa%AS9f#TB=mqK9PiG%Ftjo6`E>d4$v&M?G$IINa#+M
z)|MgbW+oscOorCFwD<2n19R<P|5ZysLb9Y~u?L<y0IUEo;pi}CJb))4<+Qcb?vE8X
z?@$>R4S;2%oQUQS!|{6z`96Rw>**8LVTj#Jfd94iHtAKr3C4@~DM?95<t|;nemw}9
zGw+VMNY{E^mInxS1{+)46j{Ak1khZ+Ju1@$BrOA90o@n_`f!GDRsd1y<+)!aK%_vl
zdn;!#T;`Cy1D#!<!*d!0l>>;@H;9SV{Jk{d)2vU!l2Q}JLyYG;Vt*j@%LBOrm>1a0
zP?Z7LFC74;-fUxgdp}*#NE{q?v&{YX<PWuS$J|o8r6d4h&^8YYG$kh|XP`Llkr3RI
z2EB6bQk#QZY15ycSJl@o&jOI4s@Km;1uH}+6qxiccEkz<McF{JY4G(bA2)vQLV}RN
z{}#asqS|m@1B^%~yBf9B63PTIHC#?!o;I-PC%6=r7u6nW2)Tvg+Bh^g%@KU!oCQXA
z`PSnYL2C6nX35XDp2P{3xWEo(MaF<wApX#@i3=(2Ji(Y@)E_4ixPG+RUdB*lKC0%&
zcT!wvI%qw1CYKqk1`HWNlc(~(#W?_sVmJNsD#z_BB^vaYCd^>8*ypbVl4HCaAfrbP
zZ2){5Axi;2+X*pe#=0ahIyxIhtSltZlKnQ=Kv`3SF5JUGJ8}|_ad*(i+3lU7zJ3y`
z`AAB7cBNX^;C<_vrXa%OViPgD<?-YwXE7u;HWnbO-I_M^Pc|Daq~_#QL)!w@hsMng
ziRO2}<RHpF=X60=+91+^YB1oXr^rG(Q=>#~sXv<`_xNFm#{F*Q?ag++vg|_$7kXj(
zxzTX5io?2?LE!eh2(>P6uRHDofeMq95X2%L>%7_T6uA=$Lx9Xh8cf!GOd@B3M1~t&
zeCSk!R5FL-`U`pVV4>OYwIWJ#@*B;$;MwKoAW?eE<*>R;<^ZO13F)Y6hN*E!9__T*
z$`-;oTm2e)^>DD#&E@^-9K6U?4l>vE8lSOB&YkR0bGGiTF0sftm?NuJ5D$=4UGE)r
zr$gk&f^FE7Cr?(b5?7|6mvf25c!7y>s9z>08K$B$LCyZh2`vIHRCxC64Gun;#_~Z=
zj{=OlG=wZ=$d(AMUylc$D>nykA&ku;ihQA0S)0ygRDEH&>#uh1%})t=OZCAHartGd
zrRgmJRRp9@)TYBNDyeOvHU<z=R!o#bC(Ja%c96b{N}7ib5MoNF(c)qibc~F0ZU<Zb
z<nUrL>sAH_2Up{*8+=LAK~WMCY5=K~o_<P36E~)*S#<z}f0PHKrqvGuwcMBQ8JU<c
z{z6&XO%ZHC88DJd3t)>Iq@<e07sJBB7PZ#4D-Tx@dH4;??@_*!lar0zT}<PUQFCPp
ziBB%W*D0j}LruQFyExaX`3l0mo-ddCQSlgZ>IPQ-3|bv4ym%4A<pZ9F)M4b_!m@KF
z*vHIRGC`Wj=}BeJX$5$NlC{%a-`gCMmE7(}d)9LweSMX{)0gPdq%rzoygaf_4$K66
zD9{GKe6YKg4dHw#8#zvg!jqgFddziGvI1NYIw4lTYjN-Q6z-)<HWm)%PJdGh^n1L`
zv<eIrTV2*g1%sx>qotUToo_K-Dc4ZG{{7PG12u+swZFcMLS^AMOKj<mg`uHg$uQ2@
zv(M2HQU)J~kv2MC^?XCwRc=q66fiOvd-;+nnz+C3Zk-&!JPJi0rgO{N*SD{1tumcZ
z+f}w}FmAb9nvae(tF(!!_Aw31Y%M-td5U#W;Wz8k;2+Gb#AOV$vl!CKm;}Q|$xy>k
z+G?NtZn)z!dXtD~q4Ee~NHcTmr4s~3gnpA3pOjO<oC3(V3YT3mG^E_BlZ9~wcGzdm
zY}K5<GY^)H>B|k2e)EjAfl9t%=eZT^bcOzt%ukTL?9OZ+M?pK`w#Z&I#%R}X{)v8a
zFjS)<HrCeOqmbP*?aoN^?k6;@Zch2}(eFa9L;GR|!c3$<ZMyE`^*>#|F=?s8m{T&O
z8nIu#{55D=_)^4Osi6x?F%<?*j~RfYn^Vo3b2Gl9-~dV}UAUAx)oS5_C~pAAC{)l!
zmpDTSgTvZb#ef+W78Ws!aUl5uJZV1|UfMLT0fY_+lOfLPIy$iw(oxGNYvV|?>C%Xu
zu}_~4UF)w}uWd++lo^e-)3T1$iSFIIx7Zo)MZOU4$J@nFbtJdQ4kP5UR#7xk(#C@|
z3{D}9o*XVO<Mqe(uQ_ioCPLCGY0{S=Yiyjhv9U2?k`BQJkPIgE0;hpNwk~CmVFiC(
z3R2dx(nfqge(VgSmJfu8_XjTnRhcXut>o%jo+J$_;~&N$^qO0Zl+ne0C_B!`Ki&^~
z&O=Pb3q=BjNkQ$;#4Og@m78r`1<>3cTzBc6VIT}x4BBo(+*ys2Qd0^BsN8H`1b;$Y
zUgmlKc{q5UZ->@Ji1}##i<&dLYe29WwMX#~b35iJWof9ca{v1EYZc?)l%1WOx6fjx
zIQK0~eomzWKF4|5U*rb(9SX@XKnfiduCGDnZ)lj}yt9l6t))dJkjB|T;vDidUC8I=
z!Ngnqk<;Tob&FK+akv*Qe9^fDb%DytN(78V9<I9$Dee*wV>7eijC$S>$88T+ifGBn
z>v%-Kt_qws^+nN8!#LT_MuG<V%!|tWZu2;8B)ok263Qjh?xt6vTE*@34bH{_h|znw
z@+EwPW-9b4kKE@X9Y*xl_uah}8cs4ZOtMOZ|3o%mwG2m)Vyo5h066ndp;@|Ox)L=7
zg>Z%QHky3U^x=3g&iV7{kS+Z2j<np|8c>(mtpS9jn(6n)(9ixc^e!^w$O-^^ddn~*
zsZyB8q@$ywXpP7xNpu2lpg6k%NsT<XZ*b3Aiz^o8zdg&?A4xua>QRW84JYxP{PqiK
zbI$c{n%F#!NKf0f(1UzkuQ@nKg=|U#kW|OUO!KISr<bQ^Fz4#XgF2rZE7-~1sw}TD
z9t?lh#13PFhp+VXFmuq`&#(VrdkNk>{R(7@25k}k;84?*GJAlxxlF_;DByKP5fb7y
zt~r28NMLYq5<K_9_~}VeU#5EU@!>9}Is`@~2cEtck~9DtPvNc6InaXwW{&Sv6%Ogo
zP|e%=L8v_s1l{6LK|0{;bU<Yy@SL<Zht<_j%|=TVF@y>nmv1#COmOpN8q7-!dIpn8
z=;)AH87^9j+l@pN0goxY_wYc`X;*5%U3Oud9NbW9Ow3cvg^I1Fn1R{Z-`gve{y+%P
zW04!)PgkYeA+v5h9){)tS}zlw$OL2HQTqL{&y{fal=1^$&Lr>0VVKjs;Cy_p$9}V4
z8_R@=AdU`xq#8$JQQiq@9^ikEIh~If>+UHA4o)(Hz4Le+oM0h69{k@#&7X>Z%nVG4
zROD#t&cRHGHY8*Zx|tJqujQjqSkFZT`4#lSlD>WW1~?kH8BQa>r09gJ4}__>xko1J
zFt=s!@-n6>piq*L@rdVl-F4^5kCaWMb5m3Cp$L;fpA1GN6|?&50_Gr<`w(;ikIKL-
z%k;$*&K(~?+6m&0AQE2JVobk#hRVHX>B`xCS(;_Gvn@mzYE!$)otR1gljjvu8U#Vo
z<1ybOYke%)ozHm|K4N|*CUXTbMvM9RFUlW5{Wns-Kj--WQvkX1PibZQ^!q%=GplPI
RqnOi+3P}lOJ$v!ye*mCsktzTH

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 eb93e89f44680e9c4039ec9365ec2cf69761463a..e80dce3e3bd3d9595321833d4bae36e62ff354ff 100644
GIT binary patch
delta 109044
zcmYIQWmr|;*F+Iiq?B$E2|+*_qyz+{1tg>q>F(x;bc={|OP6$qbYHr=`_din2L1it
zPx2^u&pvyvSu?X{Q_yrjulfG(Wp3y<$4jj{kKbY!B+}4+N_p@p32&|cA^r0oy@G-d
zAFkCLf8IVPpS>Qs_l_#O95Sx>=>VJ7pW((D`E9)%`-CZdxYoU(f|pN7zu#)-VE#IT
z=DLcyHM%Cf-&4J+J}o@K;%rpfs)$8G&ECrN4O}}6pLZ>{E@t|g&3^h(Cexo|h`<rD
zpO#^}yuuZHK;FbJ=Ara}+!IsqV&&9kQ`BS;?HR-7FU#Zf!yYV4ZEA$v*GKs|{RdY1
zs-C|4+^ekGkI6;d8qB*t*a#ybD08wlWLjcK=Bm-qc2Ki01xC`{Z|jzPuGNUjuzeb<
z4JQ_QLHnFym+CO4xTSx*mdbpM3_{tn{#a^3dtSyMDvQmkqBE(2{--~aNk!+8)xd%n
zDz}u{+ou~5F#VA;xWjH7e2r-}G@7Ai+>zbPt0I+tcO*D8A<jQ3(~Yhm0A=5|wi+Wh
ziAl=!^R>EcP0vUA{Iw_ejn-`HUoizI!&x1?FmDzKX^glnjF#yVL>#+9K}u!RvY8!j
zzC$D*+sn!wop`-PsERk(f1zZ(<fa}$W;GcO?wu{#RON|n3vfn2D#P_KJnnVb<mIlf
zD`n`#!NIh1Lewv-6%`SrQC%smZ>}nrohpHk5|#(_>;9tL?JJ&BtFb|$u;y05NDvy(
zfaz#`R<tr{K)YK8TRCfohFrMLWuG<X?W{f``vCQ^y-Q(Av`%0m9z9|axL=pr?R$LC
zVLfC|faDluv(1966ZRdZa<fBh=@$uVx9joD&>6QOc)Nfb_s*~S%}x@$GvSRaXUP$o
zjn`?Q%f+Ee0ZF~*ia(KKD4u<aT=v@DNg1k>E633x_fLau=xzp^3Zb1_Y#&){YtfXl
zjj-2S^|MD<EE3+ldn&dOnpw#WcX4}MBC{i;=Xw*dbok2O1i0r%j8A&ahj;KMd~E(Q
zQkuk8sx{T`cu{hiZj@KK9OS%C!b*t9u=Vks)5W1xtA6_1$K{tovk;u_=4cI)_Rr#g
zeelG*cBF;E{q<+g2!mP>u5HaRNs99~|M)kI@vl5_vtv2xA>}z}nov+O-~zj_4M%LL
zl)8Ehcgz1dK6=TK8_MOHZjz|65sF~IO4vnj+?u2kqv+PrJmB-&@>qN405MXf>gS6*
zjqJy$4yHs#L3cX&oWnDI`@==gpIYu3bY=LN%YB1l{<xcqhQ@@xUmm@PWkq*6#;hQC
zi0ZQHNXTxi<Q0x`^TcHTJ@8$F@hmzROzmr4#y8g)REpL5PuS{xygGZ<{hXD1hCT3;
z2NXK;dW>gSERQaaICD)jyQ6%QX;?Q8@THu7n#13Pi%Y>*+!~wuqq+{T(cu>*kbl<8
zs^Ue5(u+#W5)p~I7csqrQ4w_06^3Nvfi4Nj$%1a@S5vfEg-6jR9j(P{%)Wz>bSBps
z6c0Hr;Y49i979RS?5UVuexc%1cW`T&H_G^M^JbSQ(KK$J`~kTc5zmouTG!P>91Ejv
z)>60L=<M3gHtC?NnWw_!lr_+--7W&cQv|WM0w30&xN|QTSt?TGC7d7;IpgtN>^?<{
zF3v0%zplHuQ2Po?c#}HZYqr_ywA34|-0!qVj4(_eqZGBYw$vM}+1B0jODun-z{V&?
zDHzvdx~ZxEVVy^en14vtB8o9bP|KI7M;rqsGn~v}4P(2O`skOO?!J#8?iU@fQ)e|<
zZBp5Y<N2!jhowEnQYt17OsmkBYs~GE*jV%aaNm~gR2#YR?XeNO;P+^$eRD~bd+OLn
z<f+ElzB|N-pr!TMJ1LV5?A|KMzWaPG$r)-)*Ww7S*2eqNt-D!`SGjcWu}K)D&b75?
z-oSZ+Ui0^4vJF7K0hMCf+I+~!-r>r+rzhuTi1pbMVNNTrUa_XS`S?^=Sv*V++0xn>
z#iPrvzW=qrkAMX?XC=Rzzrb$ozGvL@o6$A2=8>)~fz)RlERV5~5LhILNZppzMfk2>
z8<nI{N#NZ;4)42vxatcT_7}1w<Z<>ifAKgW8Hc*OzIZ%u7XBXh#LFw3MA{6^aKx2{
z7nx76#=G0Na`Fko3H&|)6-}=_{OEBl_*lGW%#~WzVu*bVX}GLL#S8z2AUgeCLRBrz
zIw3Es3a8_hjz5%><yL9mvz&it3>tX{CI}{IhjG<Hymx9RDoAaj2Y)GJ3;*f)6MiN&
z-U}6symLZz`8909^$eUB->NX={F}G!iJAI7^F#6aV1i=Y8RC0o<#*E3?90oz5SfOe
zDkXIjf|j0qenW+%>Ev7*e<(-(pt1)CHC^0AHE~XCOp=^}Nd%WXkv)Rk6CK)=;`=`f
zGPy2Ad4;ZJVF-z<Hg7Bum2W-*1J@ARxvj7VvLb}t9(%hLCv->QeZI9}Lg}ltu@aj8
za;L}E)af)-LF(Z9=l>igx(ME4M}~FYeI$#$dMbu{#81hOa6b}cDyvN3L6>zCn3ZS_
z_(~yEZ0dSZlr6$Ltr+mftH*e2nPS*DxXE(vEpGy9<asNPh&odkU7#;X1%_VB?J$@w
zK(JVza=)fXRczel6dDvlM0#As(&9B9Us26ipRkxKXBgkA$)-9&o48+&MIJZTB$0y>
zXGE=JcWy5sVOTh!LN&h$5@C(~cI(j*%afaz=+N7qZ+HvRSs&jTVD6`*>$G6Sjb3!>
zt6{vBax!{i1mzO6`!T6K<Y|G=z~+W_r(L+=Ybw_Ydl!=?e2zNUjDZ=jB$B#s-|;16
zweP54JsL!A!Ny(0B69Sg&&Fde2w|&pVG)8?xT!M-_|@c&RvX=6D$A37|IKH-{z0xE
zrLY&$FRErshmm|qf`#Ub%tlbAIZfQI7R8&r_rFBWLP9tlVF_U(pGFBj(6R-o+GKfl
z9Hhu)8uS*UA#trepL(sDq-Gy4=8<Mfm%@4jDZYf%bblLZR$seE%!%GM*5Z<$j@gQz
z{q;!3h|#^D!5SHZgn$tIdaJQGT#>8BRqW|8p=5c~M5^an5l%(VJz{f8*(5Th)v_22
zeNIbHl*=?6)O~`+%^xfu7H<&LB_Ml48(tfbDTU-cr8;@43SEA+ras;~%?i`i=#EK9
z=-kw=w5BL8uUMHZ=8$sY^HB{Dz>nenA}*}0fonMu^4Q4;jYImXV+^ax(M*|_Pu2EG
z*;#ki9@F4!kJg%|4q@}hL{^bGix0XWlf#LCW4-#<*y`k<_+GKH<}K28rpxEcs@1Hv
z9%5w|?3?5(1t))1YmhcaNECuS4*wW`Qy_1HLUF2Z3uS0=@sE_t)se%VRgoc!6@v~x
zdeTy2*>5W|fAQQqZ*u!St&9bi>hOKONKbOGb^Z?OZ`!$lG3gJEYK1a1)-N|7b@Yyo
zc|iWRo#~%djGlpw3Qfyn0+q>0r1zPxN-Yw#S6T^}FA)&ax0>3ze(r4GMIIT@*;Jmx
zH?juS9QF&?Cm-+yB>CZ7YF>sV4Pf+TbUqKHC%(R_taT41Fr!Bv*pE^OJl%Q1zMb3F
zA-NRfXKH*&yV{{=U?8~w;l=?DW3dgZa9X12>U?G~;e!z9Y^13BGd;qMBB7zT&2yH5
z|9hx|d_J+j{aQ+HwrF%-(b?He*>;@U${j>C{9D2wJXDh+PQhtFIB=W87!Qv3X{le0
zXYx)PK5%iOG!?0lannFgsOE|eEdI4lQzkj>mdtWm0gZ7w+3GjtV2zfSRGSA@!Cr~<
zJgjf(@EUC+dKt@gRm(Kf75Dao!-E#w;K1005YZUf<0+4!@#dJhqTq$eVhkk5imu4X
zg05o)rRv>2c%q!;)(NfI$)#egmhjvvOeQpr&?~9}2a(LFR@~?%254&{QZ88P+I&84
z4v`HP(56*$>kMc6Sg!T^Y&J=!r7+qoxf^UsKKn<7lL>RqoSb$$rexs*zXx=5F+OA(
zaPPSi+Qhz*b9qh#<&gws#%<J6@=>A68*`l16%yq7=|<_$6BG7d59MYIBFDb22DD?g
zDmW;p@=ssgaI#$H+I5HC%eW4=)mUh_E!It*{gd8?fSe$<sv@MZJX-mbmZ$vW4fE`1
zhkm);<_dE&&eF;(XpId$Cigxh%7w*6DSCzfobp_mPWhN^_S`E$duBwaz55-FyqZm~
z+Y!V<4Auv$u6Q8Bm_VC%O9+zB{&k>afAtz0e{1R^Ei=so#YX7hB&$kXI9;^1p}yBs
zeIw#qyj*FTqJql#z&S!O*}3H|`@^M0K>Vbfc|($EZqTVJSaol%h{j)VfmCK;b*N@?
zA3tiMPy6C70(;I$`Q$D}p{>Nqo!%T6HnwnUfdb5G=e*#T$o=rw8nQN%C>LR-`xDlU
za1PKqGDe)Bdmjm2Y^q$V+x~P@QW2uib1R5KmKw)&9*W@SH<0p$%K;Xf8Vr;6n_o1S
zA3Z<B+$)sF-)};_p{Hz0^fw$wwq)@e+buNF+U?bgAVeQ0*v*&{Y_|-KhkM6STOF=i
zNs~>8)rHC^eDYet!f^`+<V*5poWY>(cWByC1wBe&XO67ga%tQ7k!H=R5|7;uYR!sa
z@9tx1^H?!Ot9#l?h?z6Lf~G4-u^$$cmD`PxJvv9Wi8~=V<PL1LAA8;aGb6v)#5I|8
zwb)w^_A8;Q{K^AZj&Egk$Yh{H;t6KeG-D1qBZa%~{sfn!1=d1BlzOARLf81PA0(ql
zyx=3F(PV!Nk5#^Wc4s4fVcg{@z)%CJm+fL96fZVecaz)_X(A(4b|wRx+0LvfZr7~v
z>kr-S7tIVOWC(x;KgQ{h&p&*ZN2=cAczww>o-tmf`{I3d!ggHa+yUS`2*xA0zeF^t
z7L&=OTq`dYoK6lOVJ|`}+H0#`;nwmsE3%n`gT`#JvyJ?P_WN=g4lB+z*1R;^)hm;#
zAqe3p5qg*S$7{BE#3w(qTNS?n=J}&7;My1b6eruZAM6{!ps@n6zR_A>MI3d~AN}@s
z5)BO<c~oZU_jIsWrX#P38mX&0TEY`BSIZ)Hiee-Jiwf+MMxbx41m}yAFHMP!66J39
zFpK&xb@iqFiRt{PL?j?pC<SGyz-|}$=?jKr=8k^V&0J#y<tp}k3kG96bGZ~oEIYQ}
zZ#_6zPt;Pob!^Yannk=6IY&I6G9BR}3^QCi+aY_WouJ0~!Fr>9HG6*}BzY^Xf@*|F
zA(^Xg2^sW0q}YepEEH&S)T-MhQK(-q6b`dGzzyaHyv71bLu2CFha}S9iTggIrXw*{
zrH<Wj>`SyX{$Ml~E`5U&wlV)?W!AT3n$LQ_v_9+Ir&aK>B8VPC_!jz8jUaW6o}$LV
zO-OzNWu?+uMr>D>GLMq_t<pG;NLus$P5=9F^RO%_CwxKMdz*kh{d80~n1L~*9Px4M
z$Xlz}u4@kH-h7nc;Di6LmONbCE473da2lG*WE4}wZ!xki<<Sag(9lKCMMab!PLJ(M
zJh3{Dgqn5v<%|`Jcuxw=>+cP)@0jbNc1Em~4K9*7>c1^b-687%K7-<x)SeXB$iYIg
zV1>gO<}TCR=U=H>E303NR4IV8^sPuWt=Lz%FJbJQ_+2BV%M)6BNhb`>sV<a<g!C;I
zGbk?u696j@QCYI^ga*C=8l76qEs6a0mrTA5w+8NmiWN^LlhSWX<4pVbA8n5iI4g(q
ze?c7w3)8>3h{pxRo4ouf<$lz`Gc7x%DH%A-`W~0N&iLP@#SuoTP~$ovi8y9>#?4ka
zSd1*9=V0slb(mtNaj*mF%i_d-xg+}v9Yd6KbW`oiz0=8RLgjV*5*4AUZ6QaO4j00~
zLXU}9H6n!TDs$Ht{=wkf0%E)w$J>D&2E~Lm_ml*T6)Y3Ug)NAa=zKEz^Mwh<e6==K
z$dPy3=#l<w1pXy!uWxj)<;LVN3Y$YC%oHinbc8%yAzH&bFs7jehYZWnkOMp|rDEq#
z1@S>j<|l$otjvD~vK9O`7!C<_K7SHX6n`#O){`t}%y}}U%>2Y0`YP_Flag0nith>P
z#wdt<EVHjklV9XRFW*;;>-s~iy523n1cP4fyP<e?cVC$_7$4`-@v?<7>hqr+3+Voo
zmI5GpPf=Rx6>GsW4hr{Qs``GMbshtv0oMe)YxGogr3gW3b}rRsQ;tt*!~FpO^{t`E
zW*=tLkq?H%9&kd{69@H!U3;|)`RYe2MFb1){6j9zd)Y^p5nmM@zR6{J%1!A%nqAwZ
z##MjBU#oXLj3areg`<8JG?dLpp4G^ey>wh1uM-*Mt{5=<MpX*uAvmblAGrD4|FQ#Z
zAY~fPbjhCBXfbu)mlQHKk=-U<vs8M>U2_MR4C9bp_84-0TnZ8U5yP+NEtoDRqAI8^
z>iIk?_O5r|?3ePwU#Ho&UIAUUFgQLUhQcqQ?Z$3K@6)?G=VfpB6ZVvYCG-(DKb67t
zCzlAB<+SV9RxDTr#g_H$xqo<##eHj<of!|v8P2GN8K!MwcaXN}eSTI#RzPk7<7$2~
z0fO(Zm}@>E!0pOmEKSTrk|L#36;-j_gaHgy<FECn;!PN!gL%a@=gWo&giPA!)WO+n
zBe~qae&yjA4@_%3fr(G8jOedQTJ?W=IC(|>#|N4%Dr?<6hzY<yJRx!#qPYR-T}^7y
zj@uXVp}+|(X27^@VNvfn=5IGAsA9oZ7>QW(@|l00Z-Gp$dRq7og{=yLfIAN+kM6AP
zj&zcc0ppG1awJ(UJBr=>P`$N=*&fNL#Jk+M^+gvfZEMvLHM(>oo-Qe<du#2WO$jx!
z5;3;`=`o%i!cG@a!1hOZGils}TCV8w)fk6jl15*ytMYTys*O1ZLPUya=s2OH&HfUo
z_J$}OZs5cj5p35w@f`L0AjFRG#N;gcTDAHBmD}VMH_Gk&%A%`=s7YH1R%CFbgN?n|
zXW4*v0O@w}Vpy^iSfy?Jg0CI#>7k-ZDOapYU4&eG+$Q{MQT>OJ%{PMFaN!N;>kOM3
z@2k@x6M|KL6wZP(;ZH*73u{Y_zO4tU^r~Z(hh#Do@T0|*Uj@+ayF$5Voy0Mi>#d=w
zA$MQ7G#D#Sa823fc!f`GZJ_W8>GkR@PFAt2U3X9E)3|D=J)*^-Er}jhN1nF6s7xiZ
z*f)CGBjwb0+-wWPMJH=Ws=N+z-7a3s^3erUvL@OSH!7c2o;6&opoARc>OL8G?+D`d
zk3Hx=ZlsqoY10nAx_Az12T$M4XLx6O!eZ4%*oD`g*|B1*m}H-v{(g<Z@<?0Jl+c08
zG9Cgh%sIO2;0N=_?Z%V){|6p~qufI}%PTRxMnRlKhPbtK@lQD@uHqr4HOrCF^C7J~
z5>3C<AjBfvOTPlYk8BDqP(lu;H^6wqf<Gn+Gr%h0hL4pG1FpVQU$%Lc!+}&0<xsZW
zQ#4RlyFz>=di9)(oRG+#Y&_($c#^I195uzUZsfhq>tV(OXv9)%H*{0&n4U0~sV4rK
zq{LmjVwtHdUuL8o;z!uo`2LYb*JM_yWJTXaBEYWZUGRg?=b|Kj=;IJ<SJ=x@ZL;WF
z9WL6hSNuL=`7VM!@f6zx3sox^4f&c)N!t43s0s4hgd6q_ku6gRRCK&v^x6E3_SHAo
z*#yP?xAS74mQM}F9_MVuR?Hm}5zQDjwa;i!L3{IN00zG&#ECmivH^m-*jDIg5;YMU
zIV<GAYSg|`7u(Q&IDOu#$RQ-8<94!-%s?6hc#+Qa{NqivUJtta5rB%gPj<a69QgF^
z0k80E@TrMoK9`$uS~6WnI&e~5io1?$tv4rBT*_r9WcS}^vN;WnIW-ZpB8w1D7eLVs
zUoYfZiN`Lvkxm_1H&~EZi0o^YGiS_*OAp6yP7X(2Y#*JdE=w=r-t_#`0saBa$M(k;
zB?eEc@Q%n&c0z#4ea^^|d16sXM9o>Yg~$#wY*`C3Pq_<BH@^az8tWOZBXjjTZ>pH~
zJFBgZXU2oC5W>Z~|CrLr+PtoTtZqB|g;M)Jb89F=th2|ZW|?Q>C)74A#sApl<UZ&G
z@7I;F<R7*0FnP#g&EzQ1*<bU^vq=0jL0vARg#UK!=4hJX6u1b&vL`Xu({3y>cNLy_
zEHlkv4>T9v1x6wpb8F4es~Th0QVU=vkpU;z`ZSg?mm3YDVOl=un|F00vT^zVcYF_o
zzvbQb$L{xtr-GIC(a%=0-ioxORiEyX*{t*W&(tLdnY2CrBosVp>oQ>`6nU|p&T7!V
zTwB31AYcB(A?JjEV<X)@NUxL>$U2JcaV9_zdVlHa*I3{51;LAv|5Qly$FbLY4HFWO
z!%37qm#fYFmCNHWd^`a?Clp0?bj}L)MD^n&V8dg}TW3cz9M=vSvB^kY#)@j5eXC=?
zYzvL6dqRdK4Uo+!90iah9>S8(z2$go!`Iq_>xHg<FWG5-e7#Upi#`<6(fW`?7+39t
z>1fFjnk10Q?m!{Pn8N<5sn)GIx)72Maz?-;ApDTT6E2l1*Jr(>W3#~@FQFfTIa?Li
z*83DVSilNw82%HU#;3kE=!{JowC(m~s!rI)CvgC0ZZsoinl1+pc)!48iOyS3ji3K0
z57&Py^&TCqdbN|ZLU7W%&!#S(3M>Y_MV)+>!tdx`C8VH0r24_}RN$oLsNZ=f9r(gB
z_bGn}&?j9cyzet<OFg32c$YKLSdmp!><Ql+g{hLg5DPyhoN}~#QKo5s^G+Ygvj4NE
zPeR^fd&TD1^nYJqVk&Fu7{8-HpfoUi57Bg2KwsD^6&|BXTG_yZQ0z^qbk`Ua^?l~h
zD&$YgCnNi3Y9%vZds1JwT15ERA*e&JZ|$9g)Kh8}5;|^bNlE(mYZcVLZwuPvE#Y6+
z!!|OR=HA&>^NnudEr>Z7Y^uvAFhDcx%i~Z#4@!mmQUmIf`A1x$(3VD{FDeROaSqI&
zM#rmumZvEi|1PzCM)~3=(BCGYTO1}(uO0EAQ{FrT53Iu_$Yis#mXMQ?BgSJwao*nH
zk-$pu)dD831kwF*domF)#xfWxsAF}$=`FJQ{~>I!{{yu7bR_w5mtj1Fa1{?AYJd&$
zqD_jLg8avR`jO?;ARkbuk?9(m??MD{I6von=c<6%>wyaUwZpQgNA=0XEZ;saBe4d7
z)pV$2a6?Mfwi!OF_bHF~X3B5|vpzlI;Y)?dkCb!lv@~^nFW+dTY3qDz{5enanK1eK
zo+bjT@ldWLxv1S`%euM#_gGA{FT+hwRQ1h^TW}YDwT6XScnER3%>lD~XRS8^Z|0P=
z3_9yR$d<V^9B}_xuWw1P!O;_j^MT_{rPw27$a|r_UP9`ZlzBL#18lV>b#9T~EAw)k
z6Kc&Kw?N|vL<g(!l?*xuEk#|hr8b*v-(6KBthYvCC#a69`8;BC`tMNq15%7Ht4N;?
zXC50|pqo)Gw#n5HRU<h<QJ!b{cM^3NFJqc8o*?FuZ&hI|j82zKau!7ExoQs6$!scU
z_Algv*6acP2j^gVTH4P`$sA)9+o}jW6G5qt*O{E`+rcWAb#>E=x|G&*7DRY<t_@%c
z6S@0YKMkt<X*cD)u8?wbdfu8UICH3D-MVH@6*kMS61pLL39!ms(Jt*Sp9q6N+dcm1
zDg>h-+z36_wd-|7<eFL{A$jA{%Ua)=pIv!L9TJ4QCmVphGi+<Tk#<t6SilkXxjtJ%
z2#jLA%+nlzGg;iV3*LUKHYy;wBT!>hQ*L~xVr};ZwU49^4?1?A67sG|A>6g^qUPsL
zo!qU2lJDMS@1p5pb3Aix%|CN`%{}585E@1AF*7Mz7r`+%S$ur+L?U{ntlL-E7_JTk
z2}vyN?cyt89HLk*R}K}%*f-0`=>AWR`~0G8-9MdJctK11YoiHnz@g{n*P>(y#+uqE
z+^tNk1)n?DVY}SLsf1ZI{2|^DJ^rh+k9&o3oF|u@{wP+RIFR1H$qg4;9{b^r!3%n~
zijWRf&Dlc)p9nq?JvoPpWkB%b_N({=j19UL<d-Apjs#YQswsHy^NUEUy!3H%xu-5G
zFy?`SbX{Wp4IlRnux{6QZRUA>$R?_SNsMd%v<-UC6-m#IyVSd;L+_1?ZSc#!NQbxx
z1B3ZdS_mZtedZ?HU-oYG4_PvVK;ifX7C$GnhbkvEBg*VdrD6II(rkpe-o-+FR`kK9
zs^&nVq3V1-nLZ5fvRy2o#yl`D1EFq4b7_7X_KK-wHjc4WE0L{QUlN`ah7YiCF|D=8
zS^%<t(kLB~S!>+twmf^h@J2TB?2owGx?}i@JO)R@qYzMPRKN&!t|JuGSiU660Ca>{
z_3)(<x~mTum$2ZORjv#_MLMmxC@gn~37%%Wa1Xa0uo@cuRL@Fi8}&{_oCnMVRq~W?
zkkWTt@;~r;tLMb*%zo+X4uukIwXTiS?-{V50L|kYRQ<>Al8ClK@DiUOhIwU!gmuqM
z2J}@EH(1Og`?12LSNt(CSH@$^WlJ+bja)OEO*QVvuE0C_P`z{X#0ewjt(`bFH{V)=
z-w@+>Tq3zoImvX-*cTr-*AmwQFYIgbf>+ecP?peoR#;{fr8OP|Z%L7r`1BQ{&`?^Q
zVe9hyWLMJ@okx4j6>gu{RF@sBw@$<XWpXs%WdI6x(i(k+*Z5svjTbOV!I63ah0@}X
zw77b^Cq1khW+dmPtGVF1=?UaU<XoSnZeb;u)7gasi~(4;c^+d)P}l4`9J_k0CSJzA
z-lMMBhW5ix*FSB{cC>c7d<3NZ$V|N@N~(yn$dSEtJC~_!|7{8zP~Fi@zXu@mh2CU!
zbB%MYC08P!>jd}VZ-)Dd3rkb({BE%lfYA%OM0}d-eL%NDg&(aP7WP<1yy>(qCZcCM
z$l-J<^~7dViSqa=(ZXW)^Pdi^mztd601P2=txFqea|a^TkkZJpVWC`h8n>VZnH0L0
z8YI(coliC5s_t@&9-l8C^H<{_fZ6Z5dP-&x5vgel=6ap?o4Oj%tu^3kw~$ie^^a_>
z`*yCl+_5m`Di(B9i@|uCI*e&ET*lmy#d^bKSkFa8c-VtTjPbB$tDvAo-#6bW?p~AX
zm>5Z?$1mV>E_03BkhqKp4Q&gj$}!Cs7#<M54h0VHO)oH;Dt{}|5LDNiF+`|R3Kz;J
z4;e6Dls7;v4z~)ycnv6;hBHJ-K?Zrm8RJ=>t)c_Z<r}u!GNjY>(s0Bc)uXjmIWO&G
z$+|WJ4b{?cxC|1!;p&uw5LyI`IP@=fSikqr<{5kp+#&?!*X&vENf9YnBy(cLq=J((
zekTsIF-{RV?Dyx8-DD!f0t22>GI@W!;4vXlv$QxWhig>n>KdBDbQpUU3h-7!<G;UA
zljJmf;m27y*!X7*5lcQmJSv?bYo;6r%C9aX7EY&NV)(k~Y2fdVa_tt>aUS0R>m;Y6
zz<K234a^~&tnT?En5fk!8?U`vwmOLWA1+YrSrvYnM>Z8U4C)H7Ue?nwGE<myQ4mO#
zo15Kq_b3t#hv^zFe981eJ3T#(jEr0@VLqvQm#VlhJMZTapzpN@Eh_$T=H}ooYN+^o
zZV~^t_F=o`^X=td&7YPga|A1ZzV0I=)cK_Mkswu}GUK;a>D3t$IiB?Q(0I?hj0Y@_
zj3?JQbXGl;3`P_}gv#IM>S`MXsN|AU)0Z|>6q>L6F0jtCVE79IZ~lV;&umCv9^r}E
z8JHx5Ogr{}5^95qrMN99mYAg)wF&drMsY_)`fgC>rKVFo!Lp1S5VhAb(rd>W;L`^K
zVdF$`*|85sY(SCtp**4j!4}*jU4BmZ%oDTtyY)jwV!Gkfw*)Qazvdunv&s~E*va>N
z^8nce8HT_c0~KlOv*WI8a3wQ$X8#@vn9<wF*G!~A@lInOgufyMX_Hr=26LI7gBk>U
z<G~|k8K(<dw~IwebW_b{(avBkt@m8#d;ZY_Nv#&i4POIhW`w4P8U2zm?aq$Qt6FPj
zW>#{3&K#-)%cA>B->DxQ{7|QI`=)9AD$^ceLbsI@-q=}%xQ$kQlecMVR(DzguVG<t
zS~%katrJ?5ay!mCeB+p^NcT_>Tli+S<=`MdGnJq$etEP|kU#Ty!YLOFti-K0O{o9N
z5}f<+EhmNK^<K6w83fTks90_7)FA{_owKHji8H7_`Wb=Y@(>BB0wQo)dca@i1q?w4
z5b4ohQ2hV-+CL^uk_4QOlf71i*o#9-tNw+K2oP^6I78Yi{JzF=@o=5ZL<33T=S*-(
z09WBT#|I<CGwHE+tq(P|42)byc8yBzfB|8ruf2@6o@$8asH)4(dZJN|@&~UoDU|^o
zs5b>TF&2Ae5dyLuI0<otK{N3DyRS~eWsIhdC~jZQ+4uh*q`v}&?}X~~Bj?AvBUhHd
z&EiPr<~19ylia9|?k|VHAK6(Pgk_ia7q-K?mm~7i5qoXtpRAXWDJ?4lry9Aslmt--
ziy@w`X)Me+D@#&VD&7&Yeq4&4H@S0;c8xY5Ya^`de}2u6;p)RLP_^BdRsPo!U#IwI
ziNAiED!02ni350M(661M8($#i&IY_<oQZPpnKK^RJed7QepF_0?n_V}d#$K5WPjX9
z%_ys^{6(V3jdkl#uwub<qa^KchpV}Qj~1%VP=~-kJMwTTAu{r#l_UJwksGh);A*Em
zNj}YM!OM}~mnp%|MPt*2t287XkEuV_+>9K2Ukjsuyd`lNI)Q;wQOAvLwbv&|!JB76
zjr_a~75p<GNF+4sSlZguA9@jM)4!1S*wK=@PYL}?^6Yed%$VktUx4UBbgft2={aPE
znAn|he8|U|clyOpZS<1B&+;*tSZkZEe2Aj-H(DO5MtlTuHB*}h10KMyI%Do-ynhE6
zSd9&RkA$!!z<GEF4(-Vp{jXMaq{QotbjRH|GBA%H+9A2cst*kR0g3<!c1gyRMe{XM
z?E}lh6FFh2ArF#~ATc&vs4q{~XR^7#F7|F2Qd^q>v4L-KQQie!6w(?~hdm&&<y{4P
zZ-$(svodZTdD1~0uAG|j(Cjc~cPRReUG8;ocQ!7jts^6du>5>Iob4z|Rd@Cx_w`}^
zGVR~#zt4Z2i6&fJIEKN)U_cjBc8)xj3#ttOlTN@Hb&T}<;=;K3NY46mZ0%X2$Bzqk
z){*!THhIQ@TTJ_-v!{rc198}Lqh+eDhmrIpSWI9PPh4n-6a~!=enby36&IiXW{ypc
zX`Dsoitzb%qtTGR%`M(#y+7w`e{}n$_{#DaHU}(?O<km->8-~7dT6w8{`^jiGOl^&
zx9`|Wb85+UefS2uYrS|0=&r_<*tb7LNd7NPBun>kj1%EKm^RF*Ayup)L`JLr;8v8y
z{zQ?7g-0R+0uAUr+niM#4K35D4Yor+asodjo2ye&zQE0OCuc{`8#JB`YI&nNkb3I|
z_*#8OhW@5aGQ!t64AI^=O_rdnh=(w0pl!}>eq4@hDt}O#r{O!3ME|^5^rwm@Xtyi}
zb4st+nX)Z{6r|GW=2zp{Hm|LbfIv(hz`0zg%&fox3P$JPSDC@$RIPW`Ej$1JIFu|M
zo5?U~ko!+?$Ujqg>6k%EN6VZ1G$=RjdN1hY$7Fc}1#YJ<?73&aK;S{x+_10j&wS`j
zr3hdGbaL<C*((pLoZ?OEpynTA{QW>qU9M*%YmYGO98R`-YE~h90eMXHpvq51OJDP7
z5(1lE&6~&0u4)={o{bPXp~ba<N~~`xzMzZ0JsHs3!+-GNnGp_54v@E+b9#>;a36QT
zupz=}$*;ON=HO%lpD=26Dy@~bb$MlJr#VOT!Om2$h;bm8Xt9#LLX*#iOj{Fae!W$H
zZgK8%W1<a|wY*vjm%%?icv)kVBMy>3r+ZVaZJq56Ndna4!BHXE_SaADkKimqdKQ<9
zMm^>UA@J>)3W6tX&H?tIb;}Bc|2;5;??-Iz+`^#XWH(r_yLfXrqQtwPTD9$yhU%9<
zJ0WO|dMa&mfm2W$cZ>=?slevo)8J$gP!)-vt?}NZHnh~>+$8njKZ5Y_qxUB4i^j@y
zKUNs$h>tLwLq!(Ljf+d4`|J*QM7oX|bDI;NbgLr(1AE$4b1vyrpxoh@`q=?9nG}(<
zv|aP@o12)bC#HhvzZRZ6Y&4of4`Qi8N<Uy%Ajua-3eZ+v_Kqmz_I+^mY1hk~sEcv^
zW8ymdPPyPM^?;}#?nXV*=JDn#$^Aidekk~kd|8T|@yS||lJ5gFTI(Q%7aY_9FulN(
z#HS&_7>vKV+bR;}rZ{>)^g<vX-je!J6Oh0rf5tQgp_0EZM)SXTV9EG-z_FGNToPnK
zC-~RjeNwJ>)d2_v{9R12D}pIZvjaG-vf1#fJ`N7<uoC`S4CqZl#IVk%CAB<4_2|Qt
zsa1>@v@d6m;WP9~CJJK}c1*yzbtonf?i4T&r&G-B>t9Jd#E%-YvN)#&kMTiR;Z~^m
z`v30`4GkhAyVM*#19G=bJXq_zHEvg(&kwk%z-GVVcrW|Tud->3-g=a?Gaz!-!vI1H
ziafq_15MNo#udD+E@5Ovd^i!(S|F=bvu?t)g9e7AT(%B$ma<7sOBpZyrCWp9W<X$P
z?bTHgbME9NhmN(g3R;X;GR>xPRXNJPdJ7l$*C8-V;(>+DZ;GMQx7o^RX+gnp;kM#E
z12-t90_n`;K0*lOGo;+?*0&zu#B1a_0GYE%JhdIeUvE!@0Qz>eeGs7#)=fTD0&*4j
z0=E<4QM6ViQmfjO0S^kiJ3tO4;XE2s@z1puChnb>iAJRg*@h_e-tJ7YUK{moX6eE1
zoA=p6lA6DRb-B&@T<h<>%-QV4S^ges0eOH!vR5R}#LS!|d_L`RFhE|b;S4fe>R*7d
z@*WJgIkx1J2-RaS%Hv>lf{wK7Js0-teGvs-h<2o=Vp9H;OkgN2{@2q8msI?lh4H=b
z4@O!8csb|iuuL(_#bv$u)&AM|^UY}~Sy^(!cFRcR*C~C3N*U}Biz)`fcdTT(AL+^4
zT~K<wKN!0`eR1*fg&z}Vxq~8S!i}2q%uxI9jA5`Eb5qxg!5xItg=DcmMEgUmipIrn
zEC<S|)@+0kGB$d7|AK4(KVJ;TieeF!AKIlF$-nphTZB*YDhT!g`C>vcg@3dIWd}Nb
za`IYv@v?1Yvc2vBXJs5fI5BBok=#D5jamN)`}?ntNWElyYBPqSQMy4aPKohs4qBA?
zdN(9A7E?;VjbYfX2RQWtKBN?@y*ivNY0EgZ>6K|u`6F0b?STZc17m{cluWom?srO$
z@>mVWqJnEg`;txbRmwuL-JxO3M|9kU{ht8$$)MdK?U44nTR;4#q_ATkpd^k3VJs6z
zQ_N*09?%rp*lb!)&APTsA=Pr5qX%M^p`Z0rbxc=01FG=0oKaKc;eE==V?#Dke-T3D
zFGBQ6SP8NgC2WYP4NAQ^ZDkXdrJ<#LPEHl8i3^)OCWEGz)<){IMk(^cok@jdIV)E%
z+^*fwcbT*)^QfCVv8?8v-#9{|DL;9b4q1Tvm^Q-e1rn!O=6aaG{_!&g`!tEl)f%c^
zrwwQXh~((BoxCuh)%IHSknUUKEZ4y?c5_Mib<*}^WFe*TL9NACQ^-ZQQE$t^YJH`6
z9U81drmf+YVI*US<B}B2APc@B99YhW4-iO$R1_5roic$+Lnt5@-67a3opwaoV-`c@
z5AutmMDM|w*-l~x%D%{HSMtJ~?A!5mDpm6nyF(so46}w&8^RRon&!xMmYZ(rF8-H?
z{JaQydxM`#%kZioqyaKPsBaBSL|E6E{m!;pwKeM~T5!>^m-WL)p}4kyss~(!KXMjT
zU#fDVK{Hz}J5$rlu&{)PMtd^en||RwkC*~n7EkN2>6>b%c@5@tU}^*dvAJ^F1#E!X
zJ(2AL0Dr$S=PxT@du>53&EUm>1n8{xT_KVr4>nGS$lV=94rSH~nON-ErSjcQJZZ{$
zd+q^#b)SmVAPran0|+cG#$d!Z0ribA>FVA0dw+qGW6jSm{E@gUc~pCuFZy_Ei4H#&
zq07%>O;WG}D|@T^@R8`DY7*z5#r~+qNzOx`X*3OCrKavL^EK}D-Z)l0Mvl5al^KLx
zQ6MH(7OLfrI{>9$6%3j`3fN(Z>BCtb(0WhDFnTH1uCUH%Fk)ERqb8<p#;3==&AN7Z
z_t3>#2ZUxoxDF&Oo_ohpU35`_#92#a>OSY@IZMdZ35udwVf5`q3flm2S?r~&#wHT~
z<lX`u(P_7HQ!2^G_XA6-rW;Ywc~IifjcFfCjEaWF^y=G;<3+^M%8I<Qa&=u|Xy{`?
zspi^cdQuZdN5?n2&H@($%>(xxp$-HP;4P5laN;XMkw|Z`C(-=ytuwW*c6KwtT0qud
zIEyIZO3lj4#IHPAkT04&7T6?_<K=xDV?*Vv8}`Qt!sMHqRKfX9!56U*>sgSCztYN8
zpm=H}T+KMbtsX4nO}6)o_*Y$z2akQL;*lQZHRck7aL1=V{tMUb^h79o`wrJjcR*;>
zyTaAgiT`+lHzvhDnc`rOEcWD(5@o*-pM6dt9y*^e$~oBvu?h#n4?-v?DEQ~HuW-@P
zI^!13{Dz1xPsJuxIH1`vw4y{*cuiKZ{*}sw5?SKMnTWO5yH!@LcsR=iOEt@%pL;Sp
z9;&jgj^}_Y8Fi0TWeROuU<kzoToH;wGXgpWRwKIv50i*hPgx6;T$${D&!5s;|F$U1
zdymFd!9`O!Unde39j!<c;jpu`_*o}$D4DHy*ycqtO836gc04pvR8-X7(}P^!(2%U?
z>V2_U1lu`2)*ncdVpI3WO!glPro07rho_5+pZETBdVuW$rNZ*L<&QJF{T*;)MI|Is
z=vt{1u=Jbi){N)^?=c$>_Pp1e5rhr`W3epY%M2iz3dAx2S1Xx39Ui)wW^<nHu{r4(
z2{bM@ImtF$FKCWGXLIv?5d<&I9@5i4sIzl7U*YYBcI}rt>mobC>vcwljT^zJ9$vMl
zkSaF-LY^XBul;~k4UE6J`qG5#%yH+UQrl$M4)sl<bG7v$_R`7^IzD=K{5pOOq(u0J
zj!|{Rqu%|cIs9k>ESFVhv$zQr=q4t)m@nV<?m?<-2yRJQ1(7RL;1>MKUpzGpFt%b6
zUU)npWWWhRb+aZz<K`976C~WwxC&0b%k}DS0_vZ*k=5JRAoZ12@{AKzb>gie3=$Yq
zO26SQ>7uxIe(N5tfajU9-OTOnG|}8=1C06rG9`l26m|J0+X2PC)EN7!lCm@c*9(g#
zwjU`tb>spbVaW*w3z2|z05i-Z6M5PkC0Hbu^LOZ*x<YeaSv)li6fO92D!wY`tJxA$
z@W&6X;g%_%T6uPmvHW;?Xj8~6m;JlHFrdN~=L?weY@XXacl_9{>n{=fU@>3VgQl$;
zqUpi1$?gEjP!{e-BCco85k!AVJ-^so{-3fhS)gMoD*L>+Ma(8GJGwtd=)d=9_0<*2
zt*(Dx;!e&Re?{Pf9Oqk$*)nD;n*ICzt!7&PcLU>xn423qd4%+xUfcfTV|6wqLSSNk
zTIs_TPS@#k#)9Cl%`OWuz$jP+4Z@QU&9{Or^!{d%a;|%)A&sl5#8|yiehKB#@L}f!
zpxKnsAM+>^)bBUnm&rY~G&@}&(*oUv>`Q41e^XbfM#Tc)kj>c>Sf5kVf9rH}gsW>*
zsDo%ZxMC1F;rMB5UaBh`(d7I$31DHOvNowLApfWP1YFIra~qEXY3NB+G|^N!>x=X$
z?8V7TJp7@|(_VDcSg!bn3lnrV{@{f=9#KW{c8{HH-zvKw>W`c(mS&$w=5#q6q$?HJ
z<dHZ_YVtcpwbj2jnN;cjG)uG;JYo>E_Ik}3!KH4(J@0o$p~T;5$>YMgj2m9l@fqsI
zXm(zG&9UYs)X3np1PB)xTE#R<kS%Ps^QH#310NE!gPX+Hv9mDT$<@1O5qKOoXMj4L
z{-rJB^uqZt-gRe6o)b)nk^AHn7ifwwkQ=<Ok8x?u`xK<A9~?{sgjSH;l1b8pB)A30
z2TO0x)Kf54YdRu9>vJbpXb8o`%!v66VO-!64+wL%*q=VXJ6v~TvcHUr(K7fXWhQbn
zMgzS~6vU5!rL;wQ<~!V|XT9UI>zObmx#~y9x;0ZN0$&TxgQ2ywG=`cMDOw(`qJUI~
zm;A!bFuZke97kP%lt4`Tt>&}@xZr}&GkxfA0#&^j9i%(&*v!H$-e5(bzYlw((1M74
z-3{b^UibYcB*h`B>G`sC5?iBAe>v^5zh<H@bHxDap#e*!pyQ;Acd{JOx0tw(>xXpi
zt!=Q*Kfr*&$6tpWLV?_|i{E<pfM4PuHygy{E?W;LHzo68`Oy_sb{JZRA=K-%WCx_3
zpUQVHkC;Kc2SYpa5uhT0ppL=v2MI4Ra0)?^Jyjmd%u9pn1s@8xXqk9Uq06>O!Capc
zwhETSSxq(27asS_5@aS7G{K5LK5MD({UM@G)u%Uj4&}2R8?g9;#Ak)&fn2WP&Rtfb
zL0njm6zA0d>0J(pgc?jo0KE|pv^JR+8PUJrXJmu-A$t5!w~nZl0o+8$95Hh6+PlFK
zn_fOM;+mI#y@<hwe;M)HD3G8;(ApX<6m&$D$^mZaQ!wrYUbYFzRfQ8f9~ChI(TIOH
zqSnIjcVL-4Z}N;V*rB3)o3v{4ex|w{5>tJA7(mqbu*^7rV_v7>%lwL}EC`s1g;a?P
znm7cpIp7GB^SwqD!peqBIOc%k&d_}p8WZ&n#2^agpS8oJn@9T}0(9i$*#7t_`#U%6
zIenJN`LQB3799(6HE-~hIq%y+ubAg*A>=INXhBJLaD#`-%S@HC6*sS-$_D5e^Sm5W
zk!5S!WS(HfcC|=mpIEqT^i6TR4f;0_?)a6M=pZAzfWoeX5ZJW|pn6HKt>OvVbl1Ur
zi?VPGr~UGsws&-6ZS9g(Q(ua5JVS)atbee}sN1ZJO*z)$SgzRiyvwc7SNfJ90r)yD
zdaC)5wuzbMG9sA&@Px$<_A6zKAkzjB9+w7Lm``M+NxUeUJTOVLx~}^c_FZA_-*m<O
z*2D6B{gRs{s8ion%omNCg%=61cVhL8fn-z0K&kG|aEWZTU%#J*Om_Km%<mQYTLV*9
zVLrM-cUN2AF`P5ROQAjOTgA|p&`|#)35?)Q4=54!9qktvl2M~SjrXFXJkB&F+{{;K
zOyet8dy1;|U9S0s`$DnZe*VY^Sp3xtj=7f6mcLMAPbz7o{fcer@4O2AbT(EIV*|BJ
zI2YE&-ldBIwQxEn=wY75vY)AbrSusAF3?hAN|7_QWn5V;N1%$E|2=<jq^%|AX5?_=
ze;fUqIoS3dCJyANpWShHa|SA$#!}tb;O#kSqvK|?qXR4L#K>khy=hx}&r`PbC*WGe
zn!2ih^y?_GGKAmuSxWPjDFM_7FZv=5;unH{WOO2=lDg=v%zBIGiadnf@tQX!#GZZs
zEGNXfce6iC#ZdbjFVFl@-4Ika$JM+)nOBo%ZBu(mD0~TLqiqvg>7Qn-Sg*QqnB^97
z)T<4F5U+OP2~ttN*py3wSBtP6Z+@V~31*0Y>*ez;{E1&wP}MXzJu5krCg7Tycv#HO
z?p<By2a^tgIlltsS~=NmQznW$b_s?m6u16lv-fJv8A#~~vO0c*+VP{-u3hBK>Q-{p
zu88WsJ-#`4Q-3qq$>=ZIU^=IG@?%wwFAzl1*d0oVzLuyy3Y0W$>lS|oJZIU6qgi_k
zmYd-1;)3~3(;%$ZVEeKl(fE4T<lOafS&w;2b0WuK9@AdG8MtB7LY0~yZ#-Pd4cyDO
z>eCXkZjiJeuXHCZ9VIAE7T}=Ip5<TK-vLKy*K8^-(OPs)9C?(WL3HHluaoRUTy$_g
zx$aq?pMj~==k#Vw=#tZNBsj^$*IoVLj-Jq}_9N<mvX=$}j)@v2d<H6Zt2Yy%y?KV4
z)WkiqCi~ORRxD3|BL%o~Z1h1>*EkvktfDBfxwrSXDA<-!T9xYRmo8^|;Y(5oAS(Vo
zUwfaBR|_psDnbCfXw1juK{uHQZkeW2Qv}xazT`sjY;5v##MAw_@WnJ;4{;>EpS=yq
z4!TZBa@C(MZD@DNWldbq^`$Qh6MP~hGZpBZhz?BzX44kl#?H8di^m5&5Z2@JEx5}q
zwA=nyVb7?!Juozh#CJ{?0IBH()>!m`11?DI`4xb4@jbG$_<EB4v7_^6&jcA+Hac?+
z(1PI6&^#Sk!ParUFB+F1(2?k9RX?gkg!?QlkM$8ikg9l)2@F4NR=K`nopV?$_1YH`
z^MI=x5Q5s|J{Q8&eV#yZg|in%Q^-7Ole6jsceJ^_DSnC5=YAc0O$?&{{zTf>;ErN*
z#nyiLdKK!jrkC3hW|%0rbbQ4^=K$)$U`qGtiuC5(C`<-xy$)Hp??$K<6S0s1Y)oxw
zp5YX$_a*?z=APgXZg6V^jmC|{NTuu!gi&NYRi-vxwR$~T1>ecY%5`t7GhD0rPWo;Y
z(<yZNo@|y^@>i03vUlK@c}No_pX#1{Qu>XEa<`w`lkkb)!+UpseUo;A6~jXiX#~Hw
zKX*sia3v*^VRnWWNk_dZ(u0i`ub!{sByh03XycTYJ76}6+cF%kvPFDr7{@zfq6J;5
z285G&J2oyp-N~CjLC0x`X#YhfwcJ^dZ#^@@vK#WC#;8Csy?ZB*_=5gB#l|#c&WdZs
zV6;j|Xqe5H!ITZl0~_*Rs{6>kayPnqe%+X7rt0?Y-%uA77Fl;i!a_l)gSU;|phRCW
z;j7$^=iwsyq7z9)1YG+3W^TEV-&jqsjAp{uF_SziGIsW(VuKNqFH`({?gfQ8!<M8o
zJPEOj_+ZFbq<LP@VYfi${18kKHnT=vbnq!`Tz*a%6TKLQ{VuWp+_eZPjcpGzh0Jd+
z-7gS+pO`LO*NVS8X?CH4QTxdXul^73m+7!Nt)0Ln15*-hu<2CNt|id0r3Q?F7@k?f
zpWJV-cD^Hc#($&}-mZ?2@LRR~^18I|suYs{s8~PNXl!X=rQ770p07BnU^I})_;~8R
z{0~=uq3oF~TPRExwoR6%!ydEBV$1p3c>>wD@F+MpK=FH3sgXkg5)YQ5trgzW1|&X{
zM=35ZIvM98T32KvL+zo%yLloGb*1<0Ohh!+9|xOVWDH*hd861}pWH*Zj4wwlDLs1l
z>6THqynAvJJ!?YKx-4WIEptWcnbt;)DVd%pOlIJlLyf7qb6<(SH18_D4J{zs^t$6~
z5OBGyz9)}ebt6WihU2zIS{dLtLvctM_^#J11}0h3%t$^^iOynE)Q6uqzb&U}I#8Hj
zPRnR7i?Kt@$Y4dT0~)Q}jb$CB?4?HzXZP;vH8@R+!o?C)d8}E=7YvXkFOfdb4iawR
z;00ENB6${9b1qcDJTz4=!X3j^ghr}nk`5CIz7Qzcn-j-P5x$b^<lXBM9UbSwESkrW
z-a7p=@1_TZg--EHQeU3nR>zzVxbDp&2liW0atUATCYXnl#iRMu;tuByWzG7fy&8;}
z-xO5jtNT6oH3u!+e?lYHS0tU%sQo$=yQXVCk18cTZiY|tXsza?3?p^l(+v&(mSZ;!
zQ9iu63qJo^Nkg%LQIn?V#ktJsL)G+gdg8@di1bZ@PaWcHI+Ar>In4yp(UElZMZ3!T
z%y)q*4dci6<s-x2Z=Hw;Ar(Nif6^RNm|_Bp16SJ-WoqPg)WQ<ne9(E#DEaAw!}IMa
zDBO$>NxJmDCqw#6L{l&5-|5eigb^?moJA$I>QIRJPK2a5KdqsHWS4TLvhE27CU(!J
zL$3(XUS6%2s2tWz)2kjZoOM_=m_@x+hMgNv)t!4{Snit(If>srx6d*8^1zspGlxXx
z+Wv9I2a{`~C$e?C0ZGZf#`!2KUYMLX+@rR0@v4cz!s1JC@2NdnA|Y((L4xEy%Ar%Z
zCfU>DQOe(u;qam=jUBiA(Zt(I^t@$!F9dp^yGRtq6eDG#VS#sFsoGywQ^PbN*;-%u
z_|JQ3_~k=tbEOv7*CavGv_m((>U<0Nzqg53qC<JgMR3oa(8GRUTQTTy@3LZa@um!T
zTAw5Hyi(J*$K*W6S5}~Qqp0oo&SQr+FOJCN&~uZkX&x1gdDbRJPX17a&~x1aW;6}N
z4hp#<yY*h%kTp!qY!-qG&mx`sWETQ;6WgiJ;V+G+N!rdZd_sR4Kc=i`j|_AQ^hWV7
zlzCjYWvy5<6wsY8Ev#$EmjZ{p2^b-H-L0ud3@=4Sv=;*mOo?A|f4uJ1ZXzybM>qLc
z@~y*vJR-N8G)hv0{pIx&2>eEZmA>l|Bh(TVi4`VsIhRZGgb{rQTdU5*D*U?HhbF|E
zoYi^^b-XjN9&{Q@YM-#r3I$)qykO)LUGF>GywWDT`KLh&<6V<+Paiz~0OxPVC4Bb~
z0-X?<Bq#*Zz0Y69^g#r1gj6qFF)j*gNK>+7d<)7fMnvU%$Jc$KkGm*!HN9U)`DdO|
zb0NDrKGrMp+e?1g!#%)*@L!mKR6}a9LG<*}>*wbt^X4*zrx=Wm=n+)gBWIYmQHJ$!
z7n_@#fNCp8=1O|}-iOt&Gl37Z4NjAyA{XpO=9&WOs@@6{)y{RYE(Oe?Tp@?lEcBsa
zp);1u&|cNPP$t;_h~klT6NlL~djh7pOcgI(9b;J6h+Cp{{@g{m7J8=WNSKq&Wpm`I
zX!mTQo~h$(W`)g~wP=t|jHF1i-Ibq8ZmQvkWr_a=d^5Is$DI4eV&oF>blq68t0w$O
z@OJuBjsPn9v$;haqeD*j$M%Uu*2}$DQa>V~a4mhfG9)l8(D{cXa<4VO#0ky;fRjqE
znXW0v0jd1rYIuBo^>WAbAj*!rRQhOyX1<ehi$Ak#dTSe*^!3L5^@TfR@l@Iq+LFjR
zDVeGhN5N(%#i&{XRdv>%tY7wZE@Rf1pj(`>^^Z;TuY`sk#53pg4T{{3(K2MdL$|+*
zBhT888skpyFX1U8adyKovM;|p8*xjH_t55xaMD5H!pg{agc=HH4#JlryQu*4G%7*H
zNI>wO;tH=Kb?R9!2Y+>GUYDRDv`wyetnz;}U1d~M-`56FkXDdRr9tWLP>?R^?(WVj
z(yeqi0@B@G(%s!XbTjZ?{Qch#uvm*>4Rg*udq1(yzTk?BulK2*4d-SnTi9Nokrhd+
zG+YyW`F!1;KNUxjVP82xP5ZZ<B?p-uKyxaDUC8+YL>a_I$JpNd{(7Qax8r%>I$~B;
zbK|!p3<jBD4*8q;ED<k16$V~muezORGefEzJ5Mi^V2=pJpJ`%ldNeZ2^mPbsn9Hq1
z#Q}~TV*^|2fgXv21~mPm_uM_kr9Rj)-@35mf5wG+*lr|bqO1}@M*K!he!B4j{u=~n
zQB@pi9fQ2IAp>@2_U&Y|fhVv>W?VUBXb_nhZq0wi2c8U3n=Ups_E&m3P~2Ij?pD84
zjg~n!c5sP{5@10TP!})=BqQd}9E=?lmWa?=d*VO>&(8KOc-Z0VcgE!`9A~+vKAq|x
zZ<mI}rz#h2pjl3*;RvYd09MI&+>vkj0^W{g@6UC|#=5sv>{uJCGs8qcV4Wfc)*k8y
z`oGrGLlPVMhvlLI(X20BAk5<!W*CbV!y8k9_KmG2nGQb@mWc7BYQ^9U*b^C(g>sd(
z6dtpivQD+me}riobruru!+OxNe0NZ&EW36pxp%!BtS;c-`@JX;3DDFk&;4F59b3_6
zVeL6!D50;!LviO5%|2BR{D3bfef%W-yS0T&PF-C>O>Ki_;hVg?ow%oP5J4olYjKW3
z!Kf*IugYXmrn2Pi7BW4#UCf-Zy|%S6W%OPYdFdiLM=i3=IQ|{~s7jooymm--rAPRs
zeqq*_DOg8#Oor8;0GsGJ(>l)OA=@X5>iz-J=yC`NBvCpb)Ii+>5(dwgOX1JS-KfTk
ze5*cY8tb>>zfls3yrWZCZN_@76+7~!8ozSPPd!%|`@KciWYYQ0xM@jCCdKeh5vzLm
z6p&YfZIJ8JKaO;$Nhnu9KtJ7@KGowb^SM*8KrmN%pxM?JDBOcS39N{*#sL$8%QLU-
zSbn}NjXgX(v~{-jWWoJtG{K4OV*PH=n8nD*$h3Ms)aI&tsGy?aY@KR>I~CIK_(p6}
z>kByii2d>bK~YC-_^Mi<(&oT^9SxFGCzoiHH#PU>&3F{O7eu%1DEhVebk8~yU(*B4
z-=Qp40aUR_fV9wnu20F>$ldjR=H8A&l;67@CT5dz#;pi>XLx|)@4ABtd*W=X0YxJR
zv?mjq{Ff&_V$M>uczX2+M|elOcFbh=<-Fs*#?H-tvJ0lsTCqEqi(7yF`6;?J>vd}J
zQ0ze!ei%sZV`pdQGt4*fv^fA-BeKFQh_4Zy0jNKDsJK~|U`3C8UQD8nD2;q?ro#)u
zkMB{d&gfT}x8JX-NT%Rgo!?(JtW=y;gjTu|EO5P5(NR;Q!4HQ|twFmi`I4<NasM<v
zt{@>P2}_})Yq8+^$iC~yEq;b%=NoMq2TB=sv6edbn4$cS=v-<hu;5|fbo0-uw##4@
zNCm)5Lry_xJl9c=QfI)7&JU%hhp@$KnJ$;`n8*;4AI*v;U3>rCzZqX=hk3L7J9)}L
zQ+AOzd3B4RtA3I!20LeCb>HMtwP0sN>%1l|wXrUN$nz4Sn8IcEMh@E_yvf|#vm73>
z?qK*k&3GtER;_YVk^7=9w)YXr1e;l01Vog!u-PmSeXptucfuFb(f&MLu*At(zfLmQ
z#4n7wb1TNG#GZw``Y3pSKsze(L&0UXdnDWYM<EqeEwMf{Ikb-}y5K=L`-KgG?Wk`5
zOtb@~Zwd?E;jJ@VaiFvDY|3dk1oyup<xBoNHX{h$m=*tK8>&{M{eTD~1iU3PWwYh>
zE{4w<gDPP*GoHUR$%?aegx){b>g+LaF@3T8{Ax>`A0SZC^2cii&iZqWq-%G%+Y-}H
zi9?h!kUJSXo3>r`>sI1|Z?xdOx$dOybMc^uh@gdHjn^gA_pFIjHEWyeLWO9}7M88X
zr=BhD$7WKLWZIcpW;}o+yU$U8PQ|E(nGkm$6JTq2#M1>?YHOu(F}WfT7^^;hU8rAV
zBQ!L>ey|cVxQT18h_?QAa_o<t%<(;5bY|3Wb5GV3Tf|%=N67b}-<<CCa`siPhvS4Y
znw{XRI1!9lh7g=hN23s`_1P<}mL+P2GzuOF*IpPT3!11H$=3nM%yhaY^dH4{X75a+
z&{GT-ANF24A-1yH{p&wreIx$QQAho2$$q}Ih*xwEb+An3#bewud05bNdryk9g6&wh
z@k$-t^T%KR$SBuUj%_KtHjkC#UXBd4*!x}A0mc`?>QV0Y2z}*ruOX`UFReOCzHFxQ
zNuAd$naw={ap1yfFELreTg20^o+Q<Sbn&(GFn={cn<eFbtqBtzs-s&5M<8!YVa=^i
zdhO{Bk3h!s?k<}d|EGKDdVS&L_Nho^IO4o7>4Cu@mI+jD?M@yE+kQMY9=nL^VT!Ju
zAmdZLpU1mmxaANrDHh57v!o($Y2;6+c}ffOi5RdD4LeXleY-6!nX5czaHh)+l%cdp
z1jK7hw^-X|!g`Cpie*+dD8Jl@wR|$9m^E9~8OkW#iH#hZ!G@gq@_4RQewI)+Iqmc%
zIU8L!F&m%1vuFROCWME-(|Z<k@JF{}S1KUaBjpZP#+%3<g;J2m)No${IAR^CLRcVD
zQ8QscV<fCSqqL}^Sv1?|Owj>$_I60*oiW8EZ4U@huO2`uhRzEemahIPC1^?)5aAZq
zU-5z+aOa>DncBIPh_FhO)t|&5Y}VYpwY{8SXS+hN(1zkwiw=w?TA0oO=HZpaQ|bf_
zi?@`ui>)w~eO=>J0@>7+cS=pAKpg$XUDkBIc<92{!~PlbsToV=ilYG!njzWF_NCyq
z2qobA`=XhR_IA$uhV~NP8@^`eR8(VE=*xHbyNp-Zp6c+dQa`I5^#|Tl)LWGLpwA1o
z(nd7`CV2sQ$?Z9U7hACpcQ;hu-m#@k9sTpb_a*+($j@*zbf5IcfF=`ITuv(*kV|4;
zD-}#$B-}8y4Bh{Xu}wf7C`@0PZGJKpmGVpq8RwlIJQBM~^uj)RW^2<e9Ir);XfX3y
zBo>CDpCV30Pgjes&E~VnK(Jp1>C}tCNiVFAo*P_abC^3nSIn=*ha_^KYUG)>ys%-f
zd>qnY!V7mKc>>;h7h7-;R7`#EHDg<~>qnB9y*mw2zqF#};nxZ6GZoHNzy#SM+QSv0
zaddf)HweDjhe#y*GX=oG3o+ZVw(Q#%N<lZ<*_E|M;L$O)tU7VSc!)vTPP&&j{La<8
zru&}%(+k|k@h<qtDGW;AUzawIe;!Us8DG0pHB}<z0@!v8=t$%CbI{8>_+a^e8iqPp
z0>P|3CdmHrKB}Y1GRZ0ZYcw|0d`YKxY|NlT4CW(r8+DLUvHFN4<(3#uj8_oM9fd29
zp)?W`xu|5%?iznJ#KQ;3+v#QjzmO<R2<{X4H7|4CG}H3y!$Tsswd`0hqfwE0(+y|?
zmYwC&=ip^1t?6k?_Fvwf$iVQC<fP3=R`rw}$Pr#c)V-ufx|&aOKBspKwSFpwm#U;4
z9yVBM>sgNVm2_bbQET#Ok;gD35^D(>UiKHUST+X9CzppS8)fub#Z^GRg3RGfk`Y5S
z1%)#Os1tza8&`LGBp#^iA6wyl0dTW!L7t?)n3%w=<hBix^><kOrZaRO2rFa8Bo6i8
zebLkjO}Ocfp5>g^TE4;$(!l&;t&i<!>%*`7G4A&ZVszNzkfjw%K^<lJW2S$-SenQR
z|7Lxel6us91e;rb<`oflf>0NXswg*m1gc0#_LuuKrJ5<zD&E^OpqABQE|>&oaBVj3
zmT-ZYBr*f7?zXypdd&B1-C;AAm{Q;q8jP3;2Qj$RBBBENPGKA`WYYzgAB}Nryf|Y-
z2AnH*cJgLxH+63OE<5Jv-h~fGj6XP5$Q=C!c_z%UEzr|+&**(5Y)<$zdxs07j-KQr
zQCyID89T@A{$?7mVD4of+UVL=ZIEhmt}LF|h${G4+%`s*5J9FxS+(F=1F!|B<iJ_p
ztPss|yRTv{6CiK^Qlfky4K++#43^hH6a2cvvj=U9$GzO@vUy55#4ONd_EFy^X7!5`
z;XXTwGEuuKE$!ouLmDB5Qz5Fddm&3OnoOGgA_AKx7Z3REOy*(R3mg}E8{a88KRlcV
zl}nrJ$lm;*tg#K{xg)AiiQp9-ct8E=&A$uPV+mh>JU3RD!rb*EMUc$<r$*ywOE=D#
z6}1H|^zqjFcFJq~kv@3=cCJ4k+m=qsGjRS_rNssYv!Z^VbZ~Qymf%-O_Z9AlU)RGO
zeceE_%;^&KAz=#c`EcijtM2Q@R_&3#;=sQ_D-xm)I#%DzvD&qgC-&0#KZikrB@EPr
znaxU>zshwHIfXFgLyrg-SR+|HB6KER9=GTIlHh$?>nq`(tYaZMJSPU`#PwQ0-1vcA
zL0?6CT+HO)vs`?mDUTyNzg{!_<1$MfCKu4hL=NE*Hyw#nx#YB_Y$^VcaMvqIs9hhQ
zK*1x<|MR<>%QA(PW$NMU<CQcGV!ROGe$)cflx_Hs3r!u%(YbIzaA`?Tj2@&ylK&Rw
z4adv>l}u6rD!9r%ze-tLTBRjpiiR6j1X3!VsdfGmVty%abNc8s&*VX1a>CGr9IOGM
zp~*PQ(!wXF=*5as9L%jX?TfeV*$F73%(6LA$w}em9p5D}IXsgQ^pDj;8`5{Mj5jmK
z=GS+LRr6*wJ!3uA)Z<P6%J#uOHgE_P5G;X_ljKYwYorGI3hy)qD<^w;x6I|TbEp*+
zzxc?=p#FxPii#_sC>kh;TD8)p4NnSUs(r(>joptd(I-BbLrq6g9FAt9qtevzFuIGs
zvj8VjpH}gJaZ;^>EnOH2le_s@PgENf^<PFWK@UI2N_obasdS%cUJ9D*DD0G59>y~D
zhag)lsere7$r-~aec~8SDBqbE!>7l=)&;o3W=xrtkIey--lUa@A?wVRM9qRmN7wyW
zIFP)M^$PBAMHQT@(#qB|sFlHSNUcB13<IH}nYA7CCDlj$oqa*Uz~3|6D4|Mmv|D_(
zRV7`06ofHMw30dh-4(D#iTX#Z@xLbrGQb%5Cpr-q7ReRd?r-mD?39<de0jz7S*+y!
zsUh&Oe$EWkDS4Zqh!zvEGPNk3%kA~rR|eEG4#p!)QP~*OfB$yNW%kbXVy$2cv69RZ
zo209Ie$mYB7@P2qa^Omkf}b=$(DByFe>&JyyGKW}A2y6CgTAg8SBu$vr+;7=DT(PK
z>`_CW@4#t}JRl9j>DKt-(gD!nq`W3A<pv@t@Z&*(By+?oF`xN4q~eSTPpbbs--vV#
zZhm7GVtA^0RT7auNjvCv*MJaL8n(8oCIB$}L$oDq|BJM??jd9kF&rf3i0o2oQ4<Td
z4;$&A@CswSuU5B5gZNOj`sD)$eUBxxR(ZIxN$2V>bSCfqfjW;{wQ3;!tV7{LtLi|@
zm*lZ{CB8<`mFI!yV~L}*`tb>WhO8@a{qTi-H2ON${M4UWm|+Uq>mMFvkw&1-DD$PK
ze9Lz$he)PLKiZ1KH#-lxId56C_8vZ~yx?{#)25JgeVUMzYVUh)rV8Fl{kg$Q=LWaq
zV*=^Edf}Q8oi7*bM7RJQs){8z71fll2pP^KyuH$Fh(x=9)qK&nsZYB9hePbDQMctC
zvP=w2|M`CkzbNTH0Wb>|yoJhr-4G6{&0HS4vS75^OmP}V|JqCtfZ1KNaP<msU79UB
zl;lspe<|uqoKihHfczn)eF0m__;1PgU11cDGu7*RV8>G99<EaR{ucx)He)UT3Z;op
z={Pi8DnaTNBDTF_FRKUwDpeF&xp^dz$NRYP#1#gaF=|NO>)pS-6JQJRmJJikEor$B
zc71Xt7fU7IYqN2ZIHCOVR1(IK;T3L$isi1{e8U$QcLd*7EiNg{9p24Sg$x4R-2r5p
zK~5n6L=xhWD;;+nIGB~LR2u>wSv_4U;2KtJWea@eYaj*{Ic}Z<V_{>^X)x2KWk;*Q
zrnvpQ;bMRA-aGx00;_vVH_zuWZFCOH(^4o~A*wPEnd!Kn!dm)W?~E9W<?zdfC60_9
z49Yd=LZ!_Z<!mCdL6Ry?242wMs|>fNBp_YvA6lJ(`(G^OJe}MJFP0_DDwSf+Z&`oa
zpPz}4nuYRx*%==Ft*Fxk6~-s+?>Xv~-6^|B=7{MU%@M3yKtC+CB@Vn5^nvx}mApMe
zS#~!@i4>^)uIIvMJ?4F7rPI#y<ia89n`XR3HGT=HwY#YEbhfU9VD16zk6!DwUc){s
z=_U)kbblp>jR$$qm6B7;J;XR_ysLf~Xlze7&$58>O$(uri~?&Pg39XZnpCBBaiNt3
z^~^ccm*l5C>9h6FAbaz3wB%Z|WmV^86%lvxGjx7DW5qjNxY-}(a9iy)IkCd}wsqrN
zf>!s+Ju2NkBKtr9$k@n_xA5YHK~QNy5>#z<R=&guA860zQ9FJ3z?iI;Q+k%-G@-R@
z8-yc;?7uR>1htK0D}DSdWR52C=-^VlQA@HUB7Q<I<GaSMYtgN5A%FWuC(J}q=V5-y
z6Ht84k9G!sHZZE5S2<r|riILOtcA+0KrBcLYFvDe1MHB$t-g4Ise1~ZziwT&(t<We
zEjmqhEELBYj>Seb;?$Yzxl=dIzNj*9N#OAnxo4lMO9wf-=SfF1E~>k`)B_2W<o-0I
zZwZ-IU1J`_JLK)iQO_9PPmuwaN`v-&^XoO@P3b&4FA(CDTa0}h+@_09%n!+|hym%^
zi_v2s`0egt;jMea++fH~3)z<7i?sKtc(?lRq#MIqdF$^vh=qu7uLeay*dpYnhuA6w
z^&4NzGB?1#-1dbbR6l3u5+AGPCjRcH70T)1ifUdQFlNK2dhnQKCqD1;L&>2<Wm$_C
z2Thi}+1~#81rOUI0nBj~EY+vR7Im?~sMIL{LPnZJ^C{5rn=vf#>)SmkGAiRS3Jjb{
zWU<|LGq`&1yDo+QM;O3MzEU2K4wgrRL28c}R5gH^-AdHrJ9cm0A7@#X&tm8_(LJkf
zPPnX|Hai%pe#~@N{<gbvp<lZ1+}kEH6Qo49oA2J-^`FruCk+{Iw_Z_W0Y(&#gqO@#
z@B&I3GhGa}=$0ct^OQkxMEBt7k(B3s@%+_uTKd!$nzOqycRwICND0EnF_m|fmzAfR
z<h^ee*;42+@0AuFZ+d(Q)vD3Ot_7{V_GJkdA3n`Tf~xdRS0v(*JNhfLddq~ys`;x-
zJ`IfNyicNDTm(De9|VA-{4e*gUxT&8H#>S^a8E9%EmICZLg?^GvX|RJOj<h)S=+<p
za8*O`_K#B5a<Ma%%de`_X@=?ZYi`T9Fw0yV{9+SN6#ppLw`2IjyK)gT)0(nXVOj+^
z)$oe`Q%*%?{<nNcR>bX~q%K%I{ifa&=@&9db`y0Wu#aZU5{wX6adv49qTNidS|Q%M
z<H%C&w(u^Sm1b!*v$tU<J|))GB)SRc7q%Ok#z<g~s<Kg~*TUoK=9pI<g6OE-reyVU
zxR{(6cZ4TmW#8a5`cr;Qv2Wu)Cnz-(4g|OcP_a&$zd@xnveN}x54XxTpbyi*H=FDM
z@FzxysyO+70{?FNIv&<`Am`O=<8*~gfy9^NJ{_<HemHZh(Lzdn<@!<)l$aHh();^4
z+~2f9k)?jqG5mRz!9cg!W;0TL8a<W}7F*R)jO<0++U(2}B1&c4Cv&Dy5v$(lqAD+0
zpN$-L^+iXyFGBXIsHQ;UJJrUvte6E*=zFZNdWrc{>GDTVf_y(SuiMKA?6DkaMuTOq
zu61<#|0*rQ|2Ty%A40@d?CpWuj=;68ey>W(+%@uhEy#zmB`Z!h4iSqDK{X+GwLF(-
z-0GO7%0L)g-hI8_{P#5KT13jBt!SX>0~(~nleBdVqjx@(2<^*&XU$`M5Qv4r8v3aa
z`^Bi)<iJ!MW4_EmyR5w{s0G=C<TjPP<G~4}UAU`4S^bqeay~FA`!#f-ts;IN;~mqF
z3^X#ka|{1)U!JWmbDL@56g2L5c&8e$ulJUJc<?6%_%<q5m*)_PZ!Q)%HVHRkOg^F&
zM)dZyhu4UhU^W8}U(iZt2Quai3ayKj&%5qs)L8elK-e=|ZAu$B(<Cxsy<s|@MS^Di
z2)bS&4pHKv8Y78N$$aWQ*w_$abQDlxLsagj#*d#}aiJlIJ~;7g=73Q)ws+mq4j4-O
zuC91_&XnVUO#E>BbmKJd8RERz0pZ=<fL&T@Us0wx0HX2&F}wSJSoVg(nmSof^)y|-
z25a1OFU|A9kG&TE2de9k=3NN7nsV(vLhn$gAFuEb8?mFxEY`3V*saEXiL)$cZG7sK
zTk@+M#8}S)$EXIp#yQFV9bh8eLDW7IH*w0eX_&jdhiTi_3GYRLa9tz@Zj^50n)0zI
z@cdI3QOZ1&FVCfCZ2PKYs^;2qe;`Ic+i)dI`tZRMwGZKm^maAY<f(zx&ZKQ3I|z57
zaNcj`6;&-EJZaUyl2cM9?);^7R($K<zE6RLvL`Rhi<1z5NA$5!m$K6%ig!lo>!bU%
zoK<(lbTux>XZnMSfryal#r>Pl8}dMs#^?Ra&nFc=)5Fu~n}2*?<Md}Ee+`x)&Cv)o
zPrk23H=5s9nY*0gzVfbp3!NsMO>*e2Nt5=%<Fr7rvOf60;_|P?h(!I<4D4d-+-}Zx
zTLp|eCRKu$!6{`4M4Lu=@}Tygtbz$DapRz%D4gt4M}ygjlS7%x0BDT<xtsgLrT|-<
zN?Q5jA|^VX!quTdv4EPPo9YmomHo<TK329e?RkN?f)lZ;XXX}0J@#$gc^Umc-et#q
zhU>pqAA=LOzPc8sk%0rkjJK0qi&<TeS6sn(Fs>X2dQ*0qtNRNW1Rg3|u7k#*QTnO9
zm$~P4EuatV2L)Da))x0rK-wX1HaT0*{bn|khL40e#HNsY@a$c<jEsJqePorrLYjlq
zSU+MG@?XAaHG>~GZNuF@LY2MaCi}KBM`2`kHoRXF-M+nx-`L&zAr^!_gizJn>XrX7
zpAw9sAcCa-l;6U^zV1}-+9gTQ)NP*5WR-ZuwF5-wI3+fF_yj5}e-NO`?uSk`$Q)s5
z;$~CRwJJlaoVk<GgEDTv2XXhp$VF+nit{Qf7L>VI2@8qxg3xQ0TO$ny{a_)YtmSCG
z2C!SA=E_@isGC_iufFL>z^<^X%DyZcu8Q-Yc#dYf=>In9JwBF(hnP|J;<xK|pg#;|
zc20+%;HJYH1&hLI8NAt7aB8-pYohej6k|CaNiLtF^xfh+73^DSyED7uOR%{kx5l)Y
zQ1J^>{<qyj3mv$5Lq>(4aguh<upqfIAyLsxsvDTuAb6X-Z2O3hl~q2rdbkk~TB=c8
z`fGUY0V*FkGSj0?R-089<`BFG#B5_5?5uycQ3MV(3rm{+4RjqFCI2_vG)T0&b3teB
zD_!Zm?%q~>gquI~opI`)bL)`dRQR^JnW1(5R|Q-}p5E_i#_~qvRkW|_QY-nu5MWua
z7N6x*i{o^*&N^&mJb0J(Uh8ATRyz0zW3UWYch~&X3(M<P6192(&XWUY87l$dFQ>UP
zBXhms*uLDOP#+=JKpk(sE+t)d2m6J$X!LX99TC~P1GMyRp#1L}=gTv?;q%_r?@pC9
zrxF}r&)@u*R*KVW`56VvT_5{+b$1{7tXEkw>zvi;G`^W%oc1rdMs1reu%^)Ym(a~6
z{}n|)Q{bDr%Wk+tZU<p3x^YqIfttA=*%#1%z`$63tO!ii;zKl_4v3ksduN>1)?K@u
zl&&Ps*98Cl4Kfaz#^bRK=R+^?;P$!r2OW*q6$oOe@9<Y0X|`S&fk+^-u3G*Auz%;3
zw<5n$_g*U*6FT6m5LuLKR%@qBB0FCmo**q63hZaRc$c$PadzvU5)r28ovA27U~)3)
zGlr5F`!02Or;TADNV_Rj=K4Z!7I}kVm=hb7fq^R-Ys6b@{1LOC!E$PJT8+kEpHam4
zm4BYY+*2oZmiIpr#<=+e5w+0p+Iw}(C_uQFN*zuYpby8Xc6-%<a;PGa!dmm{M#d80
z>e`P7av1qkZvB(dlDNNzDsydbGJ|pV%bPq{OQ+ai*I{GV)}+*epz0ZoS=0VUtUqV3
zuS28{BD{h^=U{mpGD`!U;EIP{_$HQkK?Jj#?W6wmerL)_iP)HplihQtvh4Z)UIL~y
zvO;GEdOLwRldHA;V?Zc+T4<_<O$q<T^TGo3m1%+xG@93}1ztPVvHr4R7SUvSNC`0m
z_%--7Wy}xpC_VYJ=TAEks>F(?Y$?r|w2r+AcuWS2^&Th6ht6S5xpEA`twxA>he{&>
zVR4zaH*+01q`OCGS5FDI$>2XZmjFnU-OUi?5J?zz;K<SfxdfkmB^C)r*7wpCv{jLO
zt{A!ASlg*YH39L8h!&Mh4B`eWSwsDN*K;VK!2X9&BUgSxUk(B75gQROZg-_KF*zRc
zJ(S4h4R$#ih-QuXJ2F$Y2Q}&!w&z>a%Mz$^Zn?NMhaZmVh+^$$Pv3}q@|2waMQ)wU
zB82S%UTgv0cPahX(USEx$PT)%9yprsn6MmN*X!KI(KkAEd1k05ZGU;9NSpty$~(&~
zc8B9e9^Fx3&erqbh5A)`NV3j>`caP|bJ_9C9Lz-;cT>+@?ut}GA-Ri|(}_K@>5k`C
zfO+iMEJ3mEIhX4bD;0Zh#rm8L-sP2uQiRPhu>Una7&`X^HAZ`Q<ZUL&hkC_gG)7g`
zp@Lv*;$Jx&k{(>o6MGVK6QV)6_QA{SKU|`-o1p1y)muJU!89ZkU)$*ghfV0DTV1fJ
zl$xkfDcl<M^U_Mv=~`kYV7y48g(%7r)G-Lwi<YL5g5E?c?;|4m!_Bd|$=Q`VV2N8k
z;H>w^h6?ZDqw?v1T_s_)&=AI5fWOJvi$HEtR%7mB`B>t^gU;IM%6UJkk{}DhjV$Nl
zHe6|{?$dR0xl@I~=wZ1Ei{UfaUCjC8`ilNvu03KsUWq8<w|ZYqxXhL7IkcjHMR{|K
zNX<G5YKb>BNJSL|(Y175!0)AqKDj=yvx=I5-{jVN(D(|<oje%Lb87ZhaL3N8(5mGY
zY6NN=-})InXKVI1Ap)ppKV~$3sCT!oINN!ONL3Hr^D({f;GlZexi>U$R9zWE9xe=?
z8>C=F(wYgvOZMwwD6@mRS@@Nr8eQ(%2aT1eT%Lxd5ss^TUIRGHk?BT5g#uHwqIz7Q
zS>8P%0KV+{Ra@2QpXf$l#|ZAY+9ysXrzcIXb1`>{1N}UEpzBeh{)fuCMeS-*AIfp+
zk4<J_kE)#5SCu`b0bvF-MbP7s{#*BKrK!FpV1M<+aqTMvR34n7?ak><RDyp^X)phG
ziVc7q1*R)sF+1doAAdDn*Xk6KYbe#Q81JdOu+~=i{ni*{ipnL^3p&l17=LsWy;N=<
z^KGhRqcQ|Dx#g4lvv%D%-2LkmIW6CriCHAqXGy12(fg*=_H~uzSXC%@eTT0^*4+pW
zkY7`foyT(RdB$lT9`M|8XM8J}AjOsrG=e>xYEWkz+5x29%Uwow^dn8H``MKbeIc?+
z=12V4nvLpx>R2EF)^zHkUy$BDjZt^RjtjXnqXQK-Q1zfx(c#}@2AdpnUvZko*kM%u
zvEsj|w*&hb;i)8ax8MHF&#&RjAotS=3$)^@yK%FfdHOxgiTr3jcQI^}5C%}Xvf_UD
zLoaXl<^2_7BOVOFidx8=YN%iBbGGxt`u{@M2by8x$R-<D+X!Q?PF96-4?6@f#f(Yo
zrEWv6JIie={4$31q#n-9s+^}7-n^1P9pe#55*~O@1onvzkvtuBm3&d`6Aw||w|<lc
zDZ;bnU%Qj+U=EC+gT@+*+1kfN-{!)l-kK+<kG#z{<Lk?+itWQx?^P=iahdmd&V&Uu
zW~PyGi!1;>w}%E=iF>$!cEsvr8cF^~y1JayK0{J6Z-V6QNF!)qCy33ndwoT4A4+*F
zP)x384r0S;+T(_hpj0Zw3L}%2#@8GK<|{hiMl2r}05+s@>vCc6fcBD9j&&{j6BJtg
zsh(n53#E$nzm<vaAS*xlf<YcjGcwjXQ1o4AT=z_9Ha*fLpCKB1xMyslL+uFlBOQON
zB2T0+rUvd(SA)=<yD))Ch3xNNa-HW&&xTf334RfL>6oeE!E|wvdUeR|uIc#1#FIKr
z13ncXn`W8+^WTY0wL`z!2Ki0-+7U(9$GPC!#&<XLuRN_{H>3Md=3@KT<-YfbN=Ehu
zFxzgMva-ju4wl=MJ)3TakH&mF+fU*TR}4WbkFT7EmpR=6GREfQlhS46uL?rF)X@(H
zPvVcOO@;Ce4zgmy*Z8L?=3y)+8M_DoZy6&nzx!QUsaTzZxZ>pR_>L;{ERtM|CKqQC
z|I_gD+HB*EoSO32vHFSK0VG)4E2o|5r^PRgzxd2iD0ZaacVtYcS+Eu+-_vkR60}Hd
zZ^nrvgi%i7Cwj<%gZmVy2mksi%Sd>sTA*J0_|t%V&ge!GvoKZfU=`8@MFvdDL0A6D
zvl6y*+)*_NjhOqSMv4FR--2%#P2VMovpTh!@$Mct4>kAS81RLzILgi8?iUZT*9IVM
zEp}tQZKa9ZhWoJpiT@wf2h4xIJPyXB!e!Kj+1uMgW%@S+*7^h3wif2<#SG$ViHp(&
zu5vEr+XmNd9Tqn)bpU?U-d3zZYBw-6Jy7XhzCh{1-4Qnx0vYnrsd3O({Agk@BBveW
zRUQt?*y<?zg1wFI8zgVdPiJ45arKMha1`n{<`>dEq)3QrR(u)qQh#}}D*HsK*DDXw
zO39At&Uc1{d88YhFrYGI+^jxfZfe54JhhO8Hv(f~AVWt0IPnoveX6q0Iy8M-ADq~4
zX(i7vEJ{U9pk@9Dlci3~ZMrSEwx@~2a%S%7ZRV&BKSV4Es-yO+J60*Pgpxhw?~K+u
z&%v`}#*`12+I!v2v)F2V%J#jIX-D91LF5jdE&JWK|9pxAaw;lJVq!GGZ{Oa%eS7A!
zKJ;f96KHz%>Xjp8F)=DCDzx=rJQrE3!LfZ>S^4X~4n3!)BsM(|0pZ>)a-wXxhg6))
z7fyi+{<L|Gwe;aq%I^8i(S_TC^92@o0U@<hRj;jkgosMzz_<_}RkTcz&Iob8L(J`y
zlDww18%b=)iTBa#;~BI;<NnZo>OPu|={^8W4Bh0L@?g@Z{(cqqq{&VT9vux%_0RIk
zKIE<k^77NS;5-PYaG7|87%kmX1nS`yYm@}8Bv=+PgGmDFVcIrB*0s%}km2Ru5GJ;a
ztiki^S{tH~)Id~IC4lc4>rT#<8}zC+IMNWXTdw~7MHUkqySBY8s;Y{c%H#6e1vp;5
zIGl@KUDf>-o>Sr>!kj#`dwG{c5JI5cOy#;Wk`6G{FBGY<d>|t;8wW?4Dm`b#H!2y`
z6KllZyp|U3GN#RKuat~W&7725ESVEy@N*j6sReYi!}Kbg$0%cwK;?oPAX-*+^B!Z@
z;%l<w2DDMI&bD*-@n}cVN8Ev?V|wjt_3_=RI(XM<zKFsu+ci$BeOI6mCKX@A-)RRN
zqSvJp5)y*TVNGZ;Q__4`9u|he%*-4V7G}8I<OcI_F(E6JqI$01D7;9$&)0r`P}-`!
zsNLXLK>i;yc2F#ox0p-_LCN6IO0pf8=uG1cqWNAVLCy3%6b+nznwgw@mhQV<dQ~q`
ztVhsmabKJt>J;d3@SeuwF(JikUsWQ*=bh2kzq4=!#`33l&@oAA(y~K8qikwrW|{BF
zgv?}D<fQl=iG7`Q_5)d)ghOb#LwPUARDSbmjOp7Vgzod{*p^$#@M&hBvz4Sa;qS2h
zv5{FWN#Oxv!EbroO<J@p!ijHpejZTOgA^=W-pGhzsnKQU=5%9*J4Td>oV+XRvZO1#
zbQ!aWN+zYQZmGrV{*GAtc{A1<KmsRdrp?raC4zuMJ=%$0aAHD<`*@qxgn|ZlP#L=k
z;RNai&@HEIb6&%!JXs_&o*@b7NUWL4FTn%M32UlMAaRU*liq-!92ruH3^qU@8QHZ0
zg9*c2sY2t+u4wd2!KsCCpstn*iH6Utt}Y5VUBT<WJiN7r`=wKqo0Wfv^hw6<Vg#D6
zbNg#}<9_U^?-$sKG|36+p05!LT}|VTX~}17@56Ass;RrvP+7Ti-ynjBt*34D1NF$|
zwcZ)$BPz$XG<@JAu$)fkOwzTaAp4Ho%U;)7Jz;pht*rv;#n_c|Qf6i}iHV7X+>Sv?
zh01y+CNAOG*z5a$u3y^d>XITN_Gh8thyM<k{uiPgEu$LhZ)>@cYEl$2s!O}*Ds_$~
zZIeL_Gf>yVb{w=SWMbP&c!AkCu<?NU3CPVo12UkyX)_nDKez)8K!?%IGSi-rxLVfo
zMgQQIK3rz=g6_tw>v9mq5l?#*9&$`hlP|or)eF8w(oteUHi;jPhKQfH<y*+&!yV!=
zYTX4ka0csC$2&1V>G!z>rvHN>E!M@g-!kzV2uuk#HIuLe-eDimS#?S%FJQGb+W;4K
ze}M?Tw-S%@L|XCa+(L%HPxLSFwl!x)PiegQjBoj?teU4AtOvm~-1bOP(>Cp3>p{ft
z%K7AJ#gs3?_XQL?IP%c>Z&uf@eBco1lcn+%3T*A|JNo*9%gPvgB8WVyMc%C^6$WN$
z#y8wU8X2r{uz)E$<-ljC177QBQ_PNWPqC=fyOO7%;r@$^?ffGziMCHfK!XAi$25=p
zE)6-bl$)BxdnBd<GjQ$eTm003l54cO-@Vz>`FpM=HjGJYftAy!`l906a9{~h1lp&@
z_Wp9Vq(8JoEfPWCLb=;9Ge@VS)yf9d3d!9O+}l%t?UOGrtDHX^*<$FdK!0~5n`8an
zIlEl4q3v689zIce*)&QB?@GJDV-X8=a`2@}?#PO;E#v@OBtJGhX&D8Wv4I-ReE4DJ
zU>lTTsv%mBwe!0F^2k-z=Afqz&Hu_W7*u)*Kl;rIAHa*#_=fM4IaOp&l?`7KWPo(x
zGYS9?CoV1y<9IY*Ewl12GB~)srw5*ZfIvYp79fZu5g0z4t3*Ua?H(KZmB#0BcG^p#
zqp8u4k(-<QaI+D)=rp5sb9<|2V<TIrT%5q~#eIK$T<e-rQNbMdQLUpZ81wmWr>C(2
z00=-Q8Kz42v}f9Gm6VZbag_>0BQhTaI4|A#2I4*$tak;oT1=y8XlQ6^fpe}4Fcp7G
z>23ALR&(2Ac~5!#aJjqK!>OGq22W4tzt%iW7tou3eTWNBS*Z}Wnn?<>qPiKzbGA{m
zWph-K=1f(!{DOfSZ&QMHseb+m>)z=;85)*yp3*D0j}ku76gSnVzlV;rfCB+Rx=jj`
z%8P=Lm@kLF13QY0R~0(z(aHm_=LH31!NRM`d!!h4w~sT${B8I4S;DPEj`{q=V?~-D
zpa2H)jS4O=_78a*!@jVQzp_IGHIh$MVzfqN-UD*Rk~7oseLpotZD(wgu3OzCN}ntE
zFu%b4!II)nodND1ED(Rw!Acps%i={e7aBkjK-rfAvHGsGDz?~a$$U9^$oRsY&%K#A
zy00r<$6FjkkhdqD=;9I*Syfdr%^ufhVEW%UJ{}p%7PBC9nEIpS>fzz=1ie1LKW>fw
z^-IXjjqCLE^rLFocgRvBi}@sVe}BohfIS8}<)XHMflxbOqvv#hq4j4>%x5s=f}p8L
zt1%ATrl_{IwgHTMK;U;k7z-a?Dp(*PP3JP+w_7;FDO@7pYyJKG33*+AGPFKK$!Ci)
zS<aT>k0hGnUZ$<0A|dJDo^8QMB{KVioy>R3|GFf)oO*ME8!PM@(WPF?8DFRi7yd{b
zA_asC8r75iYw)8$zl3bXRWLF5UirmgHDjLzytlM3iQ+THz2HI5R?u)~#-G0906;qc
z1#;obRIN&_niued><f}ZV(w0`H&SK66J*5V0!b)VDX}@l(;XWsnUfPlTas!*hy*+I
z#WnRHhO-7P`L}%ipVG>B!MDH$QbAL*+`&nAQUjfag53JUk(l%IZM&aJup=aa4zY1g
zAKBXmPJhY?gL`Q0=4jdpPYZ4lXSKR?Zv+u6ScS>Cxazfgpx{0vG8;+6M&e*&j~6IX
zfe63Q@_dM6h0%WVE7+P`xHQurONYnn$_VzRheVE~3#5a4HUn@*RPW>G%~=}%{rTR|
z(2(`3XO_2W>Z-J?Z0T7mxM$}*1U5g@(`9XK8Ph$_5WzPk`}C=J-g}DA?wK{bFHxC8
zH0vtngDFu@z|eOCO~zs*WA<0oA%7WW6y!CbAAc6+f^m)qlA+AS*9_gAdBIa7&QkP>
zQ93$Q+0|hF0}0{k#D~>C-5RM?BEcX)e`P~p6yRhvm%q068^$ud6-lH!cEq=(C@M9*
zSRUe&DE&zm^0<`!ezk-QwAmAT75`YxpO!3Nq<RM`j9k}moESwZMAy>hNv_=*oOG%=
zUa>E-*x(aqhFick6$<C-Uo0XZLh#WLDHx%QK+%vwzp`d9-t4c!7=>W^HoCz`(6xyW
zbvZIwKnBfENb3{yxclhj#5T?AN-Zlp`yl3x*ZooLm?^6#^m1l%XUBj!F+V?_l$Lg1
zAMbUu_J{~NpJVr9K|ulU*<Ui4y88NZ+cn|w@p1ZLG<rI^jW!=R6hh87-$8g$Y1juu
z4fJq1Y@@Ik57ynB&z2hmhll4D6`@4;`L*kkQc(pCC3CRob@-{1n0R$1f(J@dlMv=;
zp_Yo4HY6+SgB0aNHxmXX=I!JCad1#jzEc>ud(aRva`K!WnpK#stu5Z$&1mdM3wD*$
zEb)J6g&PYj%7ezZ8+jk+jluyzkKF^I?ixawBhEjA^p-E(Wr-p{xtlIsC&=3kzG(P|
zOi6#3foy8jn}I)maJRNquLoDxciQW2Cr)fg+>1L~cX#*Ct9Kx0ir5{MqIF~!w%Lw0
z{J|dUh{r3C3d%gP4Qe`Sd2B^wr!s2+XLH;GgcWiJN2hl!#eZbKqNf0C@Ykda%B^RZ
z0Tg>>A21v0XsQ7|rwk5IPyM{rS*{NQatpktJ<OXbOG3mO3n3GmfgJylvb$K4V6_TN
ze$L{_aL69o5GWs8hM@E#;CP&Yt#xwPN(A@IWV75ea?T>)r~V=uLD)B#$a1@z9XWb%
zIhe{5U+=K<Vs&Q+=q;{Y?tlKu5b^~F-_*J;f@69m&F|)CXGe1-anJ}k1!qdMxEy!i
zfVq#1g2F!|L-MIOmBj=N1_FVk@O$BN)Jo#0RaaJKJM0WwF4p~8>kO>)hPrEe-x~Eq
zk^th3Ch1;R>sJUQo^3^HmGo|!*<-6~YwchhFi`3N7TNsk4<2ekTLbYE;I9XIULzvD
zCnxs<^JnWcQ^=kCF$Mm=gVvP`w!O%+yVuWHWh@i#^cL1rE>2Bae|AY*yr5iykOawO
zEB3Q_Ccd<xqGprA`Q7&G10kOl?!$4TtNxNjIBh*L&?-voM36t@Z{e5|0G&M5-p#%M
zWM`eXpbv_@jrWWO8w8HTD>%gxaE<nBq2#$uJsx>6-IEbm8;2V`WjH-7R+ohwPT&a-
z50Cq0!`~*-$}c&gp)Qmrbo*u_n&qTN{N@Jb<H+|gqz5wL`s>2H#FIJ9s6{N{Aor`a
zKoa}aHrN2P=Bf*6cX#ek>`(gRIpQyYmoG<4bp$XlFw6;$GXshL*tfH%%Jk0Y0A8m%
zmj^STfuLP#GM4oPrU^t$@f6ZeGR4a0<B|-pZ%Dwkt*FY%`q3SV1NS^z%6K?M&0w)w
zUdImp)vL@gYRq?6M++*Y;367aqAdUY`}Z#FIT+ro2gCwbg!xL-BeP9FMq;PV1{*U-
z22g-oZELS9VvOsaXs-{kS%)uJ)lp^mtzSBtN(g?Q5LL9f#xSn5mpNh^&9O$Vn{*7P
zf8;NUK(=``>no#G=+T}V%tNM~K4|4YX%DupRbY<O4ATl4WOswEN}J5ki3PzpXXzLl
z_9R!BcxYPYrsjL_Z~%X01~XV|ad}h_ntphh&KIJVJUTtt3W7rjs<~YtgAGmx)b#Y*
zV6Ys54<8}oaV|5P_#KMa2mUnL<mND4q}FLwyQI9O8ciizVyA=iSxijlqXUeD!mV)X
zJowVnNkv&Lh1#o*D}S4Og7BWuNRW1-iHU8o3975lcdq~=No>eUd^3%urJPn)R?m5)
z{(9LXk<;$OaqAQR^$H-s=W#{O%S$LLJK`R$q1wCu(x%pWX`clRTuIQ|iw0?dHWR0X
zg^tH{x~3Bh=x0swmB54T%Ws!UiPoU>46>uDc(f>>d32Dr_{z5M5kGdDhT@HFfq|7W
z)++EP_d%0}G8?hF=Zs<7?8uRo<c-idtO8YO4Jh}U3N%m|Z5bs<_Y>$N1af=Zg3tcT
zQcg)p*<EQBfZkv4=sexw7*{6x`Mu<SI1$1As0z~-^p@=J@Nmx(w0|W55e==xnZ&$0
z2~6K$h;cgYPf|oey`a?1+BWjH<H)phz>|6v3k%Enx-$qMKN5E;Tw+nRfw*yRww&y^
z=^PD=(+^-~UpGigOB+mL3l$^rdZ(ka(tto;^g-##|4n#YpCS2s`jDb3mUr9foqcme
zrIhWg;Dwr0X}%^zoEcP*HudjigA;whr_2Tyr^2Q2!?N1Uz#)jmTM(rwNdTe9oOXr`
zU=!rGnD{6h$CssLl&*odQwQdkqRl>&v!9VB9H<8v+LxCQYXun|76PYf-#fDO%X0N*
z#(MnaiVkjR!T^8K;a*Ky<U$}fv%XmHZI+8iPFdm4$F6Vj0~$piBaC(w?){?X3@JBM
zk=**8+amQ23<E(d@`9Tc5HYb)49qA=9dFVfM+2Wf7aQ59+ch@uDjgfv)o;>Tp0T(X
zJ3JiQpuBv$hmEWCdt&NAZ0T6Hg^K>mQf^4$?Dz2ORNi<HVJ%LAmmtAvwKE|rKtW3S
z9jt1)toUk8u1tz~GQ|~<6dT%oD<%!o-77O0e~#xG<p4YF(FVZDq6K=T(Q>y>r=Xw^
z6dGy(G67Ezw}9C?@X6%3B|D3=5g`JI*HU#ykQe+c_x7#aAu5)~WPWfGt~9F>CJ2o~
z3rY(pnV7m&48m+IgPTR8XEhGBzP<b`FYkVA357C+;1@K8X7HywaGL0vq}8)jA`dL_
zT;HzxqN`Q`rf*01u2aOX>Ar>cVKz<V$uf9hc-`&gvDJi=zD6&-h<Q2vaQ6zqrHGP&
zY9p=pQ?ZrVj1^<2&J7I1%69deOxxCX+dE~a*J_8$O$k*wh1b0FDhZg2HB_BFGGx@?
zeI4{s7?a2&{iXF%P7g~Q^`mB%IUe_hH)o$sxWf+of$`}R;t0G`ng^C^S$*e}Eb`VH
z@Lh<C%$7OAtV{U=eDmsx|HR1jw$uZ5u4~RA4#%m9w<kZq0VG?j<|J*SHaczdb0rd;
z+^n|0cZFfeu|(h`J2x7YUw(~_QBu+~CcXzAlO2*Hex}EHjBM1>S~p7J%I(~($BCCx
zx%Zj@Xip8Pe$jThN--|>R)<1`Z^27Z4G2iNZd(TZhMbcwl;`HOhJV`Ykrra|qAWIT
z4|Zlk+IoZdi_K`Mr`J{)=3l)2LHfGmlR<C9r&LwTr$^`(rE95HV?b%CP5lOV4~|~D
z>9y@hjy?oZEAVx>(S>2D!HI=>nx|09Kwn=9@Xl*0SbyNC*ObW?4L6<4e+vhXvI-VI
z6EL+&rE<rD>e+a)MlI4vu^fw3+A4HCltE&kr7)?8&5%Io4Kcrab6U7$tKM7_x_qI;
z&&0k0l2JISQr(YMTJl)N(W&pgC|p=yS*i61EGRSX?$x?pg_r&u=tf0EPws4-VYvn>
z<E^P~lV=6Lf8x{-EXR(L7)={oo6Iau;jW6kS_(Y51F-cCR30~Bt7!x{qDk^1w`N<U
z$I&gZ-fX|PjLnD<`He)s8BG-5zD)A8Iy<g~lEf5e{Pz9Se57`v!A#L5;lm31g*p+n
z;b@P%1o<2dsS4bw|2wg5d-ExQ&m{uD*V<s;U1sk#vK1GWPxQfGWPva8S3}|yhghTY
zu)H7Jg()c>hj%i=88*is&pj9HnE9x{=>`K(I--SI2hu(qj8$7~ltLTVpMjO&lguzw
zbEjmBmq(^Ylrv~Q{-O<pJIW{spA^jyMCJ$?O&O$czo*HnP5*EaJdBd?h=L>ukYS{-
zQ8Q2^SKAEj%-ije&=U^jT6~kG;hSOkX3HOWG0<oEWX~pseQY>egkv`QTRhj`-r1a<
z!Zp%R=xJXGW@T%e&3v)`&|<RA!9k{!kF<DZDroa?2zuj;LE?z?FE;Rpj$VXxba)69
zwQhkGg^HKAX|b1#j7&0K1E@B4nlaGU*3KOmh-cv7wA<fv^I)DJ@xJjfNptJTiEa8F
zQps#=Y}^FBS^(n#>pLm|JEPfn4*p!c9RIJkZ(nB0jP6xEaC&|8c)CNa7Zx9QJRYq$
z{uWbnt&IIg@+^3nE%0xv@k#G-Wu5&g%CUFjTS+=Hs-n01VNRi~`oP|y**M4Bl9Qs^
znAcE_aN;Y&wyc*ChjW4*ThzZt<^nXl((eguDHJd^Zf~mj$tk+TgY)f{-#}iRF)z1K
zV0T6xyn{%XZ256rc<+dFw~X#AU-oCvC|B-R*LNN9AWYliuAz$X<pgIxe$s)nh#D=C
z*1vWk_!`_YT&BjL{~Mq-SQ2aNZZ=W7?FYupoL0?fw6f`k$Gc3QoQq#$Q)N6+MW?dN
zEb+Q9C~<o0wlbo9nnfar(kly$2`ef=!8()>v!>4zzpgOxqR*YFybM($RI3u<4iLK3
zjiIWG84D50^({6lyWq~7AqLG~?vaN(Y~_fWsmOBp+|N`e_rSRTm>I~tbWc4+c2~bS
z4}=c*mofS*k|ln`gQEJHgi*Mf_9FG#EHi{*4qSIMkndVEo$dX9Gr@uKRduP90^;FF
z*P?`)HO6wQhUB}rZBuX&9kf@i1qa@TYlmQ(E|)q{&>=sELiWb_hya5Y8Zt7SOS6vm
ztv^t^;@vp7T#FwbfTQDCesRz51@UIEHZ$u3>$qyQIoA31kilfW{IftbUrrk<@hs!)
zPZe@`KX3{>J+N%_MwVW2ZZRpPP|2o|b8y6;oSfKWj=E4zL`O%XMyN4rrhYwH{<F-+
zQJOsU;RoqtQ*LEm`E}_^ncdk3{dIUi)3g7{z_O$&*XYT`bkdlU#~7XRY*)rumSiA?
zL1pUeaaNLZzn+j)pj=)GbmR4b2?I<n-a|Ujjb)95cVX|37Qsjs(PdEX$ytrOp)Awc
zF<-|&L_tCE{i0_8_s<XgM-Df5GsNK0WyIyn_^8q?GhM`qhs<nDPlw<mosu44xic6$
zQptWGPJ86{HI8JT76aW)4xX#hC{mj;tU+^TGV%jeKz?ha!GW=CD|$Nnp&*H_8lqVd
zp@5hK#M!}z##>iSNOmGX`9x1~HVuA7YkTo`Hs}3h_I<MNz)}}W7fx%*r@D&~H&r`0
zQIFV0>%@z;rb4-_?4!En*qQ)<^Wwmnrrsj&r#@<NhDx=6VhjxA5QU`or|T7Mq0Up{
z81l*=7R09?7n*UctQr!KD`#I0uiUzUiCKJpEBIuE(V^y!9v2+Foi#t3-ZLsjvqdJD
zV@3c&rmOwV&^0S5cG_JRf;GCk$JruV|4I2HSlON|Um|SV_@tL~h)Dy5EAej97>m@u
z<pW_n|H!3&d@d*G6(DkF@w~B}ueFH;<!2cg8SudYhRk?iwtsl=1X=hksQ%r8u_Reb
zz-;){e<Q%pugrAp13Wr$Sa>+9dU0HSK6M)K=q8iG2`l*d^GkmIYwXBp!VVgsNZ+lO
z8cwG*Y?!<s09SlA^KW-gzRHCZ3=EN>I1CjQGxYC}UV(VJ#Ucfd#kkD<(i{dHS=$DN
zTPH~3zJveI_V)JfrTf|08K=k9$JWO?I)TR<GIn+)2218IDfRXB7rSG=pyc~oHeCi(
z8Ex(Cgears6%>S}q)<S;>P%}#sqV?#OEjRU2q+rgqLl6;_$K7~E|sey6;-mey`$g9
zThUoLVX-+nRm0#;b}76#6dl2(MDPAi$DC6k^Y<3Ham8M9_c*x$30v{q(LfE-O&#9L
zg=h?~R3e??f;quLy#t)WuZ;=ROQQVEX5D>DJpPSs_tJv!KvQE-Q_LZfO5~z#LM1>5
z3#cI9P?oK4LQ?THI%j-zOILTJw7MYN;N?c*50avQuS|Aq)c!F?0cX8O-m>+0;G4D<
zuH?QT*#g%~v(DUU1j&-lWsm#NcuDjJN-G1EkzXaHX!*96o5+ZVH}0qj$?Z6>SX3^@
zo0e!g><Bs*C4>>iPa}W|Uk`o__z^Y%N)sV;6&hF_dHZV_-~D!imfdRZ=@z#1sE4ex
znb^MF`^T@F+x9nGNt1sv|6H{FD6}O<rn2DxGi_tX6i@58&c2ryDJC|>*9Da+A1PH=
z)O!CP{Kw&pdpTiXApMZTA>wn7DNxJ<K?DartLgXHSS+xb{551A92_jGtp)P58l7Dz
zT++c<c>y<IW5#ASP7VTD5R7zyx^bdZfN&-$CuahfXW$e{=lqdz;_90tBhf)#Knn%Q
zQY#fHDU76~q#?7~o+}upRTeW5Anp2<kl;GP>Fw?92}(t|D-0SnVQp=KA3uJSh#YBe
z_o3HnI01b!<?Bfx@uSRlOez^h(^1mJ#YKsc*4$i*_wU~ma@oIx;rTzBt^z2luIr+J
zN-NSSA|Tye64E8oozfs74VOlgPH7O3c<Am9k?t<(?ympf`~5RG<1i@Cz2}_0*Is+=
zb$IKfUTzEbG10UT;^KCJI(Pr7);_ioFcfYiO{dr|U$))dUhgMp?=Kkd`@G`2IpHDx
z4!J+|3N`d>ZPXbtx3V!?7rW@`d8Eb+@lbTEo~vVSHS!BuihleVuymSk#qI`e(ij~G
zy)2HkH&_*5Dass|2_Fmhqz)RfIFm-&a7v7xi%fq_oTR7bG5Vh7tp6@!cC4HRt*&X|
zmB(=?bV~I#ugI?|tDD@>vrKMl?)ocN-60dusV*8ZB-d@N{i&^MhI1O+R_Dz9ixEGY
z=*qIPy`E(w(DY2Y$hS7qUONr)SC7-%`~s2&AG?h~R>4chel1#AzB3<AHrS^`MA{mq
zL)tjo&v!V;tjBp(9%$9Bbg`3j5sGe6RfMYd1gB1cB!yU?)@(PR8CtChp&l{Rn;rGR
z&7}-AwV1aRA~_O;Pa3PA!J>+bOIC7IsZ@A`er7eW%lEmV+0&w^dBQ`Dp;6PnlteS)
z#&6%?oG>|es_`Ybsn!GV4_>IxFWJP-o6BQbEfYXixC35$RlwTBqzm-;56&3wZFcPk
z_yHrz3%WckTq<55mn8xoXJS%&x3hDq!a{YeFEOy9;*0-wO<kS+)@ZKO#7pyl5A+0&
z;o+rcXrr_8bs$P?m)(kqU?lP9eoqy8Yuehol3J&Du@)-H(Foe+ipbG4(Uzuj(TJwX
zysdA=;<==jJl}TH{@v+p(<2YeCcxwPn!(MdPI?=;oupP1HownjHvI~=*?!cZ_|wYS
z<1^EN{N7vn3*H#kZS<k_xfg1avzqaQb-apFij`md!f7BCi#ttmlV`;g^gI!?;)b;K
zF}M1qHvON>mdpn7R8{Y1@E9jk#1u!_mHa!@)Z%pci6@)T<ojbw^tbK@=^D^yhV{M%
zM)%t2QByEhxK=AZds?bz$g*Nq{H4sKJxMi}RLK_F%(pO8Bz)$SHM%|VTqkyLfp}~U
z>oq6QueM_dTQy_6+^}gk7DvoYvOp7+xu5Xt(`WZ7l2$5oQA$QADO^7}yxYjq6=I%7
zoLDSU+9arOdW@0f5W?n~JFtHzb3u-eTroM}w3h2qv^1V+-qf^(=zKhD^HH5!AKis_
zWJ!VpwUFx6^Q<CP;Q27TWXG_dIbjr8E`tAXQj0DWN@1$U+t%;@Z@u-7j*q#W_K3im
zE)S-QXjOmqzPNC<v$roPE^Zze2m@V+C#ZPMfQ_+r$gfurrHndUOSHixA?XL3WU5#n
z9<aWC0Raz9k>jHtMne--TWPv*4KAR3D6?OeQ^{9rY4Lp;6c#3)tMnJB02CCU=Sxe#
z_<BBhaB$GB%VjG!+j*y?ORek^fVq~7i!HuJ#>Pu4E25^RZvo$7GFhxYW<H>o$F*w5
zT~vRB3~E;*LQZQV(5M29vOwUY4$_$@$T@i2c5lExz%5X<_>jEOLiqf7vkxK)AUKP5
z6%d4~2SRse5WJu+X6Uz3z=G1^f-UNVoYmrFqoAnI`6KMA{iIR&3SXGgp?S_TPBpva
z$>eMc{uaWz?X{h{7li>B2H9Sfp{D1bKAS;Ueb@86>568Wh=yn6@UAsN4gw{-@`9L1
zI>}<<Yhy~AlWusnds=j9G4MovYaqKg;Op9=_&0m%S6_~qK-Y1wtdwp+-I&G|mQDCj
zzP??}Gh-IvfG<DSSJEeIRPP3)Htoq~JBov5t%E-M_ffa+@M*)4bP6=uvOLx^4L4Of
zV(E<!6@*cEkq6f7Ze{r93vOw&_?=C~XA@mYZ+phtNx#LqZud7RuCYgU_nDq5ak|#C
zBM`>@xqngr5%OIWt1n-IckG*Pj{ov>kg@JjkF~^rB(J|xOF@IS&}_B+tA=c}Y@K)P
zZe0H$EPQWgX8?dFBaI$zlyXU?!zCgjBJsRV41oS+cSc1)LAg5Lmj`ul$hVIE{!rWH
zPG;-bY8sp6<z-(T-3lG~&Q3D`Z-Ep9pL@6%YNbp6gdelW?{Rg)FP{lEm;=wIV;ieU
z|Est-Z8Y+z?&8YI&W?^Osnl=Jd=5U>5_w?4g?W*Pg?toDTkkuR!zYeX;El{Eltf8K
z??hA5MLLZ(YR^t%P?W5CH{la`^STssI5SViX_?YIBL;q_?-lo!Mm&;ERB~NngGLl1
z^ahW!AO1B3k5FFmSrv1A>s71*b&K2X<ole~l`d;p0;Vm<#*UmNWut(q#{Jc|*%i8P
zcE}~R_75D<{Xs0k8Ae-2)mBWkWYz#8JZ+7QnKg=3*~~zJMx+OSa^#J{v(l((aXj)-
zlK!<wqAvx*2#j9hdLo^4o{4@R<AUZP$H)s`+~2bOKe>vY5ju${)z6s@H&A%r&Mb}&
z_AM4$&Jp49Epyk_52peU{IY7FPJLpS`gd>Ja0`wb1{Hymx!k*jac6%H4rjrKd=F}j
zww?_fw0p|99_g3$EiEmbsEp5A$zRd(Bf$|;E$2nh<r9B+YbDk7l7!DCmm3=PNkqad
zEyM{|6!rv{#0B9R{<v}8BRXEVmg<hBK}im-r`Y<Ge`#jCzpfvJ<7EFBN%f@`Is8Df
zNikzeOQWcaN9zTi*NtY>A^vqeYm|2xH-%(J#g&lc=8BKCn5V?!0B!Pk)ctYKEF}F9
z72iBj|CkEwQgcrD!XNhJPPLE>-n%#?Y*nv^z4h?vo8s%=w~fJ6W%)A!4rY$e{d7OR
zoyEA+@?|?$?^}<Wt6SX*k&P2ESvf|S_cApnNmN6q3~R6N&y({o$yEDavRBHQvP4nL
zBD1c<_^vuqVO>})c!KS>nG>54@nbi~-5VBf9)1xhg?OT<oT2TBBAYETL?P#|+_lcl
z9TLbaFN@1r>xUJUqeH}$zt)L=d?`jXEAME{IfIfJhbO33yS4Ts=A5n}>rK<woF@6_
zb;oRy>qU~vo-Fv4g=eD}N9pRK&%pXcb9bw&TS|wCEAJlc+a4!Hhh8!|?xBNZEfh<T
z6q55-_M>ujDijz3Tj)EdGoy6w&zf`XAKR-b`_i=RvTo6i+R4Ljl=op;T^Z(D#YCHW
z43UsKPT(v4EczH89~eK2s>%7ZUYKx%?zNwnK*5K%urt02d(PGEjWzZ>VaDa@aGbc`
zmx4Bxrqu~zALJh4aKr*StF!7O+#?T-ksR%h881r^H=&xO>tj4zr)08IgV%`oHNX3?
zhuO_n#orYvB%GqZx7MNhwe=i9=;15xUA_HFgmjh5r<F@5OIB~*@3(fW&(Hr<j&atf
zg7fTalfh2-oH~Ya6oDhRDy-|&_Ed^4U=&k}&@HVvpDze30QcDp!`ZD9tH1hX6=`wY
zo#+}RWD=2>^u9NAWamds=s^&?*|AKR6zz6&r3%+|wRDtoR`ff5a%*P6WR!QeI0D83
zT21~K)I$?X4(KOr2BpOxxv}@w^*=H;6UN7*dbpFV<g~PWUAZvZoMpFM=oKBp|0NWT
z5wVRSd05W=Is!iigNFE2Q+2l{P1%C$qDKY27h3SiOH;c_NBzts4jYQHC-+Sv#80|E
z#zRL-+Ppkd&;WNen3WE9&!A-`V^a)WDrB-3V(j;=SBZZQ2nhQC4+i|E4o?aCt$AGQ
z6xGQpek)+-NPL%mL(dSAA?-L+Ec^1RQSj(&zsGzK3wF7lEy!K}nRn}FBGPf<gha+r
z6a=Eg&h}=~i~|Bd^{?;tkse<MHRb+y7%(13EMkK<Ua`ybmA%PRXf~aqeB+pdB|ctg
zxt&;zE@Z+goVxAh9yus+nLhp=BJPj;lvBleAU}G>2A6P#n(HtI0^qp(9cf8jzeG?d
zN#S4*w$HAR%W;j`PhZxWV*U`QwE_R;w%hJ6#(P$ywc{9-V~$<?)+PTLeQ_SXWwXh{
z71Kot4m%XYkh<BK>$`DJKyks%45P{)Y<UxN4rq(6uJ7sV{bFj%`oKTTNY&QwfcI%p
zTVj)j+U0H3&{_QFktgc1vTgyzt0UOOyznW<Pf*%Cv*$gDl?7745EBUmqv~IsKIF=E
zvVoj7T_#)Qq?8_GnJ*ckTS`)_n^JCP--OYM8{jeB;;-PWHjVdT+1<4Hay4Dz?R1BC
za)=r*Ck=FUcAyXF>9%#|JVK&kECI|DR_3}CDHH0USHIk%A^ok~Fgbq{x(O)A3yGV4
z7m<HrIw&(i!XMB5a}9M%cYj*-yN?9r{@Y$X1^&%*ZLi&3i|C?=!Xe#UqhuMwzU^bv
zb4&TFdQME%8t(z}hr^H~qETfXqEmkQl0=bG{vZ^e^+|}>#{w%*qNAgu14+(Msj>7!
z@t$EgSMMSB9AU&9bCl%1nWOzZ@%quy!r?a8uW7?fts5xN?{{BywcAkRbvVIS#cEQp
z^Zih_+jEugn@b4~zCBXTj5k+{3yv3-o6Pz4(n_Nu0GshX4%TIm?oHu7h;U=|=UFf6
zG`Ey)8)36aiKvZW$G1a!yoLk%UaVDBtN?!E{s;=*p&~iIuGy9$FXW+xB9W3<m5f?L
zeQihV&ruKw{ey$?wNXHHnwizRL@$WtRi&5fBb|$K4+o882WZ=Ny#BE4Bx!s3N#&Dz
zc8w%B0)7zp)znlA{KaK$PwcWJ&-EDgg8V|LnP$B*Pb*$eCIk$RM~0^p(`qRzEA$SV
zG*Pgh1)N;D6d^gs*rMmJ4tA(t@g$?@YORcJLAN=`($0VD;%M?$tT-u!r4H^KwWXvq
z2*NFI*dLv^PKR!~=w3~$ydSQC(Ru7?vUQkm7WyLz47=1up_<uM1D?-|H(y$;Zjk^e
z=btHLy|J;CBm15jPdz6)IEyoH^_ax?hM`3H9VhPQwkd}@cGnO^odR|q-c3i@Pw}Ex
z&_gZ|CjNAUun*)RKwK*pcDxcLf3zGX_${T;=O8?)+lcS_O4Xf>%txfDDd4+2UMSk1
zLPa|14{|JSYl3LaBg=v<!e+`J(~-}G&N-jhTaUaL-HyzpnGTgxUQjFWl{ZM2kWMrq
z|9n(foqc@9s4Cq(*E!C_T5OKpCTj&PHqq-#VX2~vmo0BsoVeJC*1+m6#xqX87AC1R
z;tJF~)8%N=rdP$jX4MyA!5NpTl_gT!Gq<xx1U$I$+ziU%HW#(nMBcgVsu;67DaO0@
zXsRul$2*?Sb{3w9?^l3{rnX?Ni}H9Se+}iC5j_h6miS=(g(f9?ZxxaDXnm;t)*k(~
z6~R;=#oJMNp<02qA=JUOHki~aCd0}fcF$}T5j~4m5-1O!7we8z5y(xyRAB!QA5VSV
zH%<%|>{QRKcg5fBtAL${$hy4f`)e%mo$W{Dou20Lz7+IjS5?Edgpbi5oBtr$XRhp(
z=^O;Bu6DhNlSH}>h&@PEgMi-|!uhWh>?#XTCZC@%D%w>wkh_m!^@1B7r@32qUj1-L
z0mb-8S{8<t*q5SyvIj~Xe4R&h?^kWj*04Uvh-DKwoNi}ZTYXdtG=}n&Y3b<q3wNHO
zppdh%Vc8$g#K>BV=eKlq1p?*&$H2E>y+=x|<Wt=5_(!1*7t2cypYonSSh=>f6dg3M
zBd0%(R(neLen803f<qy_b85_f<*=wzvWf!oeBqd_YSqSv=KdW1gA2&@WCnkVkyYhG
z8d`?3>axC7$R{Wl5&PZtSy`?Ssn?6tHZ(o<m?xHNfMfjH>mO#2Ig<u6Snu;AXEItt
ziIH>1VS|c^)8!pv^L}KDwv7GE+oh_6`z0pS7!b;M(jv&0V?M(#nxe_`K0#qi&2cjj
zLHS6PrneULU07A1(p7rv?ce#{sK)&egb?)kFz@9l<ryYz3Vz%|?4^k;L}FTr+Q$I=
zQ=0c6eO7G^l!d9PZ+{|d2*qMKySoTv%hgAG;t6p)5tqlkpnCa9D_SF4Nem-j+UH)U
zMTTs>o0KYA6A^0Kz29WCiz>@BpUQjVv_CqJq+|J2rNDT5oz~`jmtLf|pY?&~Pl;7r
zwFZMjRcyt6_YzUHyH}3<9o+t7&lLYd57nulc{SJI#>wln*FL7a^=Puj-sI=cpU^{h
zf2BJriO(gnwN==_z+hv(ffp_`H1t7c8ySh&pDL)EK%@d21Dqm!%pU|+Dij}u>=1%`
z>IM?X`)88T-Bli;xJv!|4{}pMtjhWv*_<A&q^W)pQA8r!a<?wKZHT|ycQY{joA0b1
z$TWrREQECv+$_z`o}c3_3sCNX-J7I8Ed`SBVs|G5Ybw~PMR&&|sfvn5(b>^Sc?=j7
z!6%|cBgX@?g0xy=vr^q31)D-G3;_r^vwLGH8)LxOGxqv|4X5#&$gi=H`;2U)jgh>g
zA4V)B*G6o8uZ1%`GhPlz6~G_;k$h_cTZ)4*ZSrNpOfQ|Ey2^U|Jc4fd{D}Cp@hTS4
za_2f*=$t7OrHU>7NG561cwrbWUiYMpy}S~o0tXUYRahum$IQQRBL%%3l#rZPkIU-5
z`=gleTOm1Yt4r4miy@VFzqiZsP(8nOiYh;-ciPPK8~bYl55JwxplEwBv5Q@3zH8Yv
z(p2x=5EwQ3-O2sU4%7o@g+C#I(Z14FhpRWQyn@Uc(C_XYV08l{HAH*+hk9{iSX&WV
zITRR^-kWC`g|9EDcD#A58rm6Kr?JWIS`&PRq5Q)1smJWwe)^RSn>y5IO8;ANQX|)w
zPCv%YD;zcky#aqX0wgMTm+Pb}N2Lo2Kwka&_3Ozp$OHq_C`nwlkNt(cH41(i_r$ih
zx5E`%&+!BRed-@3v|kh3W;2!a_vg@k*Vh-T3$dCLR$L#zls7RT!+``0kctX#+v~ww
zD4IB)_%lTZZpezp!u62}a9a-La(ilMwk;;9k258SD-ksHT`pUQL6vD`DmOUcfcn!X
zF{n4+GfFCpW|yo`iKgdv&e%Mg2xm6#me16{N&FwlydnY@lAnd-oz$1*r=(ai+H?G=
z!uWgFG|NF5^II!g>ydQo53FUoZ$^38^Q{fl1m(leuG?*+RbpkhS)}ve3f@WTyhxna
zq*La5tk#_#!|$vn#cY(fB&=h$T2PP(@q-={%<+AJPPWPZj_gkAIWA9#N^Jq}Z*=$k
zOsQ2Zw07LqOZUO;4{!T6@5YxMT+;{07YtdC&5=>WJ!OSD`__{}h%~t0@n=N`_b9bo
z@>T1gdd1HM>x=}9=vx+P1lJlw5Tci?j;vy>&)<Y+28Eognle-vR@)wb$sG3vR&J9Y
z=Qhq;K4QzcKN)T}UvNx9l|xn|`J@(!nVEU~<0BWl!}2L{s_GRHoy36g?FyonfV0q*
zcY&}g@z_VHv0Qu_t~e@>-;cml3iHVf?t%HLiJq6!h$(xHl#-gdCK>BN+JH4*?-ljE
zZ=*HS)Yoc`?B=k;a^jJwJ5qiRXrOJ1ljWl1nk^U8Nk3<4V#EAJy~C5RwfXv+F7be_
z>u_os2H~QpyO7e#%EDDsziAaErLV2CME3rV8qWNmc6Qw{s$4kh>y!EW56_`HIy%0M
z`2sGS_9|+;`j#OcR_eGrJ&peUxgUYSk>&Vwg~dbT3=Xt1?ER52f$#+a6s7pPcW@CA
z5&3E*K0p#`Xl>m;TVoH@gm8L5Mf-)ah4z_nUj%>wXQPA;d;C5X-y2t&&y+KDcqF#6
ze4PqREcQjO_e7ixX5B{53kk8+eKy&5U-WMCHZ5-aO-@pW!h<ics_o9{TamN4Jzoep
zcxFK{FS&wDiqaTppopsv1PQZ<)Ix*?^4&p1k^R~Q4);>Lv?)paDSj_)iGDQs_dNYM
zNYq<v!avbT0<fWIoe4QjfhfcGUKz1-&ZiII;cVdgJGdB6(SZ`Te`UTRB?QI~D$v2z
zn7hfbmoF0^gMOId(UJ}K&9>v&{SBvoyo(~~3Iw*)A+$a1)73-}nB_4iw}XJODKSDq
zMx<37Qw9_h_RR1MII|U@{uPmw9=@ALf8cVHC9t*lG0Bx%s+?H(Y@aC_j87!wU$B+F
zWX<A01$QV^n4J8S1dt-7?S7^+hMalhuI8M;h3)*;*w?~xlBtSe4Bfq|;T4B#^AkKq
z>8Df|5Oc}}{0CzJBFI9JIJjO6t)N*%^2i?nx$LOKF|V4j>|e6?`FtOLL%a_ZSvs{u
z&z~!7UftbQMX@&B-|>Pqev$@y7+Hb~O;8gOITk`XxJu`6(Z74)_$E_bE9m<h_Cw^F
zsm;l0-X)EJ#Lh<r5k1@Ef;+-}#l<SakZ~(qE#M&5Vw6}4-$bRxe)o^ToN+~GduFb*
z`^MCIxX&m&4vBK*xYFk(w<2R-o;Bv$z(+xOQEesRzcsLx?3FYIhL+AT5?o!3B#+sU
zFqF?HcE*b)11yWnLvK=>#N_NKJEjoTxPevPs&E6`zPLF{#n#4wyA2QDLgLLRMAUCB
zAbb^L#P;5ts6qR18&9IGQr4We-Mqx1_!po6Pu$&}nb}GjQDlAyj|dzd$E<UeV|#X#
zTW)Z^U?+3Qb5!bgA#4{_N&`#}Uqvk~&_BDk7K&cMmMqFBCn`5btF#7wlAO)y-erfj
zW8uRKxbtEuys9Y0Y_#X+1G@{*msn!~9b=YEwJ2%8LIKwDU#;)Q^Lqt95-8thN$QcS
z>K4?O+}0~OI5@I;sHv$bgJc0&|Gl>tb>Ua~1qWKeu7V#=b4>(qe@L*rXY72QwS|wM
zH)j`<(Gi(FErozpA_1ugMeoWdwrAF&fNx}PED8)MsDs~4`JM)4if(Dcm{j%jb#C*v
zctj}MKqj6|U}JL4`8%69;G+4mytIb9Ti@1Nh=O)r+RVh?rt1$1@>b-qy@{cZy$S4H
zjE;{StS;Nc&Q=HQrmzS5WdYJZggE*h6lMvD3wylLiHYBRMLNnnaj|ASS)uhxr_!n_
zPT9}k$66usf$7Unr|O~87cHe(%=y|o!KBj9&DB6@8h4gVeyl+NNRhWz1;yu|e|`x3
zRLyCoP(T9aA5)Gg-HE5FmG*B_bYSF5qsE1)jiNeI93Xh-Em{qGbtc1?6c_R~VC)+$
zyLZ*tv;T#M=MOLF9B!%YMLLwCwJMW!4oO-efx@W&Urf-k&-35Aul>D(G!LddN*_9}
z9F~*Kpv$7Vbd|_y(`YfH`R$wazz;BaX8%&08USIwJU|&{`d4qMqRcNVw)8P;hdexh
zgux27BUI%r{q$#MV~z1=>lL{Pq_7t^&QFbu-=N(09gc5rL`GmILsv6~O1b>-KrOTM
z`zJ9FPx!Xx->^(XXuf?$Fd4dMgW8$?ZF!zCw8~Rgiv~e3@tbO)gzyW~6E2sS+v~2}
z^q|5BIgB`zi0-~Ehi#Pq9zZ-ENXbCPXOul?JQ<cQHya7Sq1TA6v9`3Nd(cVU-i@j1
z(rY)`tx7?gcFxY#xoud0U?TZI($`R8TavNp-l!C_X1W+Wflu?bzpH=l#rZGg#2-@6
z_o$Lpy3X)nT&2$0C-5frW~~LkDi?FS9@Edh4IrN~(vdP==0)Ldw8S7^tJm0UcMF;3
z6tOOR=SD{;<&ehqxwW+i5fGiXIA|rd1@%7XVnz@jOea|yEZmRoSED4YL|NU_D=fd2
zT59;pyGpss_Fk=XkQz5bVngPls%?K^kM8DtxbCh=xWbV<W>HBuyhaTt1nZ6CF*x^#
zzV^xl+)3b=o}@7quRAIUtmK$Y7E<GIh@Bl{|Hq)2`+@hu?|PD^+vI6ERf-6P7$60O
z=Z2lZ`T*UsSx!C&22H}kk00i7mRkL@X5Nd64#HH6hWGD5yMK5!T_iApM+Us(r56-Y
zi9>%&Dv7j9QB>ZvI+M0N(TI5FYI>-q$*dl)ux>g=pc0xYGTUgcct2`8Ke*7a^JYQK
z1{5JKsx`KhDY^lko%T)g9UV{=xHVE!jn)k3h45(X%L8#1RL<g}{Zc}>a=gjHNbIG*
zDP{StRty?3Pk6Xnz6%)3zdEaax?-WH{CZkh5_F}I)F(zZFCFKc1*fV%z8aUS#3JeU
z>G2<&tiAHZGSk;)funy?PH&HGL8W<9TOOVQ4`sKe9Tl1QTkB+`(P%;c=oBW3-AX|V
zUO)4wbC>)|5wH`$+5Kb0GG3%R0CeeKEZ`}ebR2{6=5W@k^$2Jza#&6K#ZT>Sg@J;v
z#C(hfZo1seaEl2sU5;zDI1T&{dufWW!5ntDw=sX}D7~>r7_>Vv7Du;kbM7Sqtoxs<
z^)*JWAL-d6L!hMQX&P?&<b#DkZh7h8A3b?o>NAia<SRwP=)1dgjdL((eB|`2>~GBd
zQEP%zNjan7CEned9)l%s(A_zqqH%E*u<qguN4hsM*j{@#W8$h*Z9d&0I^K9$7Cq~c
zOU5?^M{}jsq4H>e({DWVK$7knx<WA_brMe9&}zVgEWBEWlM0IjI>!f8Tb<&TX5JO&
zh#{7fTwKzC4a8hbzt_jtqg`>G7$!a7`yOd<`~_d$RGhO3579v9P*2PtmBsz+e*y2A
zG3$Y6?KgD|jZ&)_j!K&aIW;x4nr(m3j{~B!GN*l={QSMU>@=H&MpROPBp}mYh`E6N
zZB4_eTJGn}t4N!bt6_#@jRgEOjMx%0rl%(<95-i1AAJ<WwWgsIAb|4m<?el!!&l?>
z-5+1()VDI!sGX{8d*&3oaBpqFtDo^aP?+8w55Aj18H0rdx3LzX*=ut^&szD;$j~)8
zC1jsYR9WrufV0$Ct1^88Mnxb7l;{;7-1O^SPRHDF%rIXF)c9Sjleh*)OE4~l&AiHV
z640c+lj@kB3Z1_{H}Z#Zf*ew3;wGH(L?Af8L3zxVF9ofPft)q!aFQw`A~CU^LG-`+
zgDHZbCkgB!*VC<z!{x5-zCH>bo&=BUvl5#H0S2w=cXD#**>X_wPjDE2=P_ISLmEGX
zL4{??UrK9~yVQ4FOre}0lf$RMe-ndjlN?CIZk4hrH#zcCyk4Pe<8&vfMouGTQesP1
zllT+Mg`gl15!kH?lon<wSG$H)kz}hEDM-)-1eq+Q%1|SzlT@IqP_**mrm0DXNs2rJ
zRbs1$)+Ffvw3n$iFFZ*(Eu24uE4TAXCo%mtwv*UUb9A~755ycV1RK7Aa?PQiQe1ME
zfF?5C2On4e!()IYL5%z6ar=jgfBSgv8R!I|tA5V}cPqgDS}6VWT_|k3W=<P_=y_j!
zp?4EN`xlxK(ygl*NjqkyriQ>!bppEu<S6J!^}wgZM0GpwL@`g+@RmBAQU3UXuR<l{
z2U9v-Hz%VI8dm<ygX{RK<<tPqRfa%#0tw!n(^KfJh~3E-(0cPOVt24-wtW879`CQW
z^%UA`P7=zSQ=!#Dny5r(_cP~8_e$XeuALDz>K8L@@6E3l4ak)LWXCoK5pBjo_oJ8q
z8Jt`@V?EG2Y)KEo3zPv@<p#U+g_JWtS!$p<W4aY^N3MCltC}%g!+Qj-rBVZ}?Vt!%
zJ>fSF4nF<;izei<aO!E^eg@Rkb-ct<4&W)6j<SOQ-cZWA{XxCLciTf~dH0#xWSrP9
z%pfhkz#&9qQgT9~AP|bi%r$p1;W5eyPwc@zxwJ>X*j39||L<i9G3o3!)jgaLKaJO{
zQ7`T~_1t+8OYPSPFy(Y2U6o@q^a-KYzXXDuESj$IvR*08b<9N!3s-+%>K@dbVXPC{
zDP+Z_bQY3&g$~4~vTlCG$B7l0qPl_AK6Vo}pv}im1z}|9c&TRT=X_UcCk=jL!Ch{f
zf2JQ@s``%hsgihUwH@1&CWbE=ugCi5B5ZpWmM%*9cM{9ufBLF_W`gy8SzWW64&9cG
zLxL}MEG-!KT){ymxwoV1bM-F;r!uqajpQ!bpBIDK*!&A5-(b48=DXtmosJyh*PyQP
zc?Izu5mX+U%zkr{2yK)DpZ5W?T7Qc+Dk)kQy?sRrx(L5?SQ(#N`TOzE?Hc?nSC|We
zzb}YjSL>E+ibinth)FrWTN1kXVkPu<iWc;yWzBohjz;XmX;&s|xfk)Le*iIX^@Xbv
zcW}NR8^EIo++MA}Cv;GFxgy_RWvxIwQHzMwm(!k>j2dA(VmIk1TC?`CJAMwxEXT#=
zRh>W{*qNNLywG4=@v_b-qWjs#+_z8sBGsN(ETE?m?#LW*bD{;FdS{mC06bjTSA`^k
z+W0*<#Q%^80TpKb0~&+_(0FK;Fl<dRY#<WTto3bHX#&3tVmc1An!PTC(!bV;_9m)R
zVd1TqvYL;->=>naRDAaV>R|_K`$YBaf>G&mjbZRjIni=g-lH`Y9JXD2;gJ+F_F7WV
zh68H9`nDSjhyU-t-hMO$?Rk|RfZ!(UU05G<nbg!s4UIqXiQnKd>g4B!KcInhxgjiA
zw=~ovrk&Y3K1M<-u4(k?SJCrvESN;Y4~er7)TDYXTZ^rrpXevA*njBA+v>cyG~@ma
zR9p5=Id+ea^s^P_>kDa;4}Ir|GB~csXy|(PzjeIj>{56u8%~GwAL8)i$rPI}>FVqp
zZg6t~78B+M&JQofSr6NI-QIml#{hCnqCDIIw>XtKmUc9`Nec&l0XP>f<1B-QZf$?F
zr@DHb&L8ly)8)1$RB29cJp|payd}?mb<Qrf|M37qIS~N*>qzVmPt!o#Q$M5C_Q=!0
zhOx@NX)2qeMl@lkSAmAuD6hn^6!a3^1~BSe(T9cYf)0JNBn+cd$c3Q)wxhGej~ik)
z5Rrk@{a4!J*$RQazP_Kof6$B;SV=$eSe@+kgQA&9P7_v&%Z_1;<i7Mqino2QKFTgt
zJbpbv`r7W6jKB$fr9@twFgkzsxQ+CGOv;sa1yOxLvkH0*LA8#IkQf7%%Fgo1wC_n0
zUo_xYnW3ej1QpOR&=k-VTIp<46jXy_l>S@+M;xvqEauMg6!(q2^AP&9z)12}YWcKc
z_Y)fShPE5-3Rgj*fdGd701O8n-0~}prjUi95p2+qJzBPVrHK|QrZr+X1?L<XB#HUI
zof!BXU_bk{UTWDr4(r(;@K4bfmp+_!Lu?NVwWOcHpje{+=f1jl_sx%P(9%FXgVVLh
zo!F$mB;X21Y|>pn@_vhhdYiOlPdS+5QI@wWk!`qHy&oh9xEO)nM`CPVv1d;53R&YV
zxmrN!x?96n3DqQQXEU;b031bJmiv&XZ&m?BmPB+z47P;h*KX<fyN3-$N$FY8)PCs(
zbafO^BsiPTivrDoIn{y2aHbm*3oHJWNBn_-NHP0C!F(z;9NYJtDbM1Vi1=<(oELwB
zu-%sSm7J$XioF?D+E#_Z*;`suT__j&A4xAv&&qo0MXOp!r>#@=DMVuL;rn514RQkk
zxtvjTr_oGM6-E6R9>G0r4T>U5VG-dC0kQ}mpdIxtzY9iUCM#rl6u9qFQP;rOe9se`
zgVzTCEbXl@c-7#S+xyvQHm`K1h1=@)AaOKQO|m_;dX2bQ1G+@K)HO5nFL3-@51yP;
zIz8H=)z9a=A$a_#@{O2|{uGK}8Uss9!7avfLR?-5C~aZ|=+s>2iiP@DTGQ!v_>(~4
zGhhCQAg%4VA-ua!&-k@bPM)Bb3dV8K<H5@;&sVG-sTTjj%T{4;Js*zsT>n>jljl61
zuhGOpe{D|UZVFry3uu%_N8v7;^NI>FC4iMLDJjY7q1UWrJ3V-aM+ff!7sJ6|*#O1Q
zCKJD*aHP!$Z-IsA1><fRVAFOD2x%+X8f>?h^0)7Bp{=o8<Z=7eDK(Pf4^9vZm_7N#
zB3Cdae=;7W%$=W*JyEu|tNUB_Pju+&@+W1O*p~_~o5y6oLxBYKhTTKGuS=uL0kSV<
z+nl{JnoQPd2>B?r$ZFO)lM)p;9t3gVGBlEuyYLVpS1|G8NaZmvC6HXK5+n_@HRMvI
zx7CfRB|UuLSEKarDQE;Pt?)y-f|s+@1&gikAN6<WlJQdo_DFW`2s))O#ujDHwO4JN
z?*)qLgcj|FBev)P81L%h62s85YWER?@d0OG1h?k9pY=`ARTy7XtDi#GfuUzUXQ4(5
zzk%{Ld9^5eq~vy+efS6#@@lZq_1bLtt#Z@W(!l+VN`sY!#Lk`s=u}HL_-0>7t{3Og
z^)3p^0{R$;`yRwKYmW<j!TRBy%}Fa}ZN<guX-7tH<oEi!J0J6{<RH1xBrSIFjymTn
zkGV~M<`Z!zqa=kWQv*+El#cQon6_E+rCWzv5?<KrcF0&CRW<(i+(JlZn+|SnG^upj
zVzxown1%Y(@7rG$%DYG0P!~mTn?3I83i(@hAjf7`Y|_>J(ZSgRQ@T?1Amy5D6tp-1
zJ1rq8nF~Z`<ncG!dr|TN>a|I6aVA?!4{>418?rUH)G-sT@U&Kj&E9h2AAH03(G8>t
z<1<meZhM9n&eLe#dW($@F3c6{r5?aJQ6S>GeEIq9y!nQHxD{nyu_tKDjw#S2)3e~I
z=afO~ZxR4A5Lj1Ww!g7IG40dTyP$^}`L5~4>+pZBdEVH70AQ>Pc8dt_@P0=2mJ%&I
z1XO_8o}R=-gPqSc;RaH@DJIA&f^motH6|0DFeZ5y<W5#kvzoUga$VzHO#chkOWu)@
zF|t*XAGxrdf<jYHxsO{-J@E&>-+6|7uBdF4p=_Jh5(-a5`8XZ;;_|Pl0Ls3f7;N>@
zrg3>)@V;<Yq}qm!0L{FjwzBQHOf$3+mEUl+1#`b#jf1nCtBnC=Nun$6e`n<Dl&yva
zo~5XuD<kc*itG&|^a;8481AL%_==}l{#~;$RbcIC&r1EGVTGqFc9$&8#-h!TLlJN{
zKnEb%;K<8ED4?Lr`2&in<~JG2WB=VJdsmq)he9&_!eqhK@=gya7aHPRY%?v6l_o{t
zN)SnZVvQw|poC^&W+Mwgy6LsdIZw=NLq}&vs0ozV+HW&%yHho{;aP{?q#c)6zfGNW
z_6DDZTHk{;Vt0Z5t9NQhOJn1U|0Dv>N$>~?VyOf#ll;mZ0ys_$hte9~($>_?3Nx5U
zT{HR2Y2#NqvO+ff@!}|{0S-v!wy8gH5SHWXeI{mR5`p{??f^(Aqhez}ZD~Io9O2|_
zYLzJGGi0tr-=arjbx!8%&zdWiaeLgk_yyE8Ot(#EJ*}-%F_kMM!7YxhL0p5|S>rD{
zSMUo~o?FUGdNSuSbIRZ>__ouQGZ$IjeAO@0$&@^Y^sqEozy9&{jRy#SSm{B&)RjqA
zUt;mZO8h)0s(VfoOUfQdSRyAXJ){C{8*Gmpef?f@fs1KNUsb|M6Vhr3%zHn)o1FUw
z>d3l(>-J2&)PpMUC&P1%fe_uNXSr-ZIxo;D;J!dh6r$2YXup;|qMFqe^J)|g=v$2+
zyfgrR>ke8&_4jwzfCUMUXEnim@xt5J7rs>L&MZf_sWDds^8tllGceL?Y8(`^F-RgU
zrbl<mbTb_5%a&+Ni}PpMUgz2%dFAB<&&=;<)Q={1sW1Fq_MTa1U{Acdz!RK@m!RY%
zH&t-+c~T(`8xD_4kOlih3euF`nI#4yt1-u~V{0qBVyWk1gX?okSVn7r3=UDr>KLDU
z#LM~A=;uV?Lm1?<E9W6jy`z*Zz82ZvQ=`sHJ#r;u``TfW!R5%Eg7y?01QC;uf$G!%
z0D|&Lt@<;Wco<1mW+sB5py1W%c2~9SvMFHiYGzV+okDBuq4iSZ9vpzo4vvovFAkRK
z@2<AFco*&luL0=+e8FnU`F*pqsXl$GS`aMIZA$4Pamzn(11<|du-{$pa?Is?@Cr<O
zpcC^PfT8dQ7m&J2_BFLpb9T409sRxp_MQ2LM4*se+}e@?V<0w-my|$V|0H=t+#6&i
z0Rc$B>2&|02YWi9TLk<t50Qc%($jMV+_@x#Kf?$?lvzqHfZ{eW0R2|1PXSdXZ|B0W
zU~({4Ya;R(l5ENTWHzzQQ|I<>dMseM1W0a;ne-n4G65mqi8&a?lGGF)VuoK}f5{Iu
z%;Dk9l`*d~G*pWSO89we(#{b==+2<8@%Eg*1vX}mgn_XRq{U9RH$cmq#P1dh_CaJs
zgso~RPNvp$H1)?H>=xtNf`ZT;A20@6|IXOB2!0%}i@#1vc<r5lM^k}{+k|r!xcleo
zZ1(|w?-97oR9aciRE7Y;sQ7^aDcEnc7z#Ix=W}5xGZ|>KfNp`Q*MmU?4ujV3V0_je
z;0O8FWpli7(Ef#guYyG1@^F2hc7tDR0R=FTc#K5qm9%jo@3p<BQVyS)Q0@pZNIRM{
zj}KDuez+lq8vAbP2F=oK%BBdL=@cA*$;XM%u9D-#FZn4KFTbbUft#rAntz?+*gG!k
z10iW1#GX>{z2V_EAd#|UF<WZwKG3)Sz<W+vWpX$rGReli#VA(;A$z+p10(Ui=X9e7
z?^AT5S4>RdV8#cGYVah7{{3qJdg(v-T<qeVItNnt6V=N;y}-l!4G6=xG&I85BN0(i
zZ9t+8ZZPl*42()m#drta3%I;A3qo3GZx=mr_1K&M?^3$U6NvFA%S<pxNJxkT-0><j
zLc}tbPBup_uCDBEE^Ng@2#g~Q4^K~xf#Mpdx-(Q@(qZ~-0cewxlLvcs`<R4;7Jok8
z4KleP?l-X>{|g6iOC8Hwrn_TX{T~0G@G5u#*)dAkv}+Mg(7Nf$K^!YtYZBWPU!pPV
zwj%+nh-KO{8-UimnC6wuudl#w7Fm(Q2WJWUfjYK*l61d~SJyQ2ZoHz@Z7)@$fwZP3
zsW`iC<aDenPswa^%c)Fz#J`@V<AbNytnFU#j}0KadH`xU$Awv$j|W@FI&y3+>3hvg
zZN#-yI9%SppHzqDz?B<7?hda94y98CqXJC~o6UR@I45f<RzM9O&t`_nz`#Hx=t%_Q
z{aaC0wVoBHBMcj3(QieFr3WGc#wM++i-K|F)e1F`==EEC5VJ~L0OS$C!NhC@ypyey
zlgV;tNP9=e(*C~9Xo*Y$+h$7Y)XdB=pmdMVp#^9JOy;Z&r0T9!hCL`!!NOQP2vPw$
zXdwp;odyb^Uu^NF+t~gz$z`6*MuxLUI$zqW{7~-nUWpVl!V{=TDc4SX%Zx?SP6Cs4
z7izu*L`Q7Tb>DV4u%WDJ`x9>-*s_|ry-JgZNwq0rW0iRxJvMMmiSQh3MTcs6L|8RH
zvjo1Te-qmrL09g^%2jJb^}u?A=>@aN&ZgFq?WdP|%jWW+w=ALR4SUZnH%V-8OjhA{
zB_vQqoA>|qjpC=2la|J^n5-c~ZcTMz9HW)B`3Z#m0+(Fi-M0xkuQ8QVK>q=DtJA#8
zdLfv&N^$;+1U*cakwQ1Nsy<aLKpTRvn46taOA83bvNBei`okv+9_Pf+?I|w!d%%Y-
z#R$;PeN+rWX2c@Buj$Z3NPfot#zBqBZ0USEjk&n%CN(!aw-ZQbX5ScBO+59Mc``Gi
z|5qJU1k;aj5Zc6L)2j&CDk37A`Yh8^dEFs!WeIY?JKnFfUPuw(ILF5a%Q2q9nBFMs
zBYhX^i;k>Aqg#Tl9cZKAspv|ba1<j55>hYmc3|LC`3P78F>i@cFDDSds4iWAT8T`A
zN2<Xes<WPSfmfrm(sU*MxBo!=i!Dp0hl_vTAr|(<0u}fAdS=Z6%rg8)gc4b8|B60@
zE08TDBqZKt(?X%anlYs#zhv85TYvTzX?BkTdmZzRm-zTd?|`hnM>|`!NCyr`CR#TP
zYp|C=Tm+UWE#u>rCwP{YmT+G~Lkq{5`DjFaR3CP~xkLu~Bj*5)fECOiP{kFz7QK&d
z(W628qhdQ%`y)rFEjvVT<3v#!Ow2AG!yV*hPN$Wy_gbR!Ym=#NEzIOQ(dH*x_AH1M
znJka$N`}@_QrMe36@p3S87B8-iSENf2F=RzJ&sLS{zlQxW**btDom@lEaZNE1N(c}
zQ_77PY63j*g|{Q-@|qg{trn<#4iCep<CCR(n9P$PzL;+??i#MC>G_E*?Qj|cFeU(o
z4pG4=cu0o;4HTE(Tbi41@z1=WlHP}X;UNal5xkX`Idj|*8p=jS66?cb6*ow(HnNei
zDJj+`CGIzeUFD`jJ~mCavA=&yAR{9O3j~P!cWbgmXXWI4JvliU#=j!qw00MCq*a0z
zquYF@(bM|}GqQjGe1?*Rh<xecA_-Dp&fxYCR&M5|Gv^!M<i}t*XF+vMiVO(9I)<rc
zJ8Jv^<vrdW1?yrEcBA^Sp;_VKHR|T{Ep-%`j0N{W&{pWFe@zX#1YjV6sUhA$Mwz`G
zepo3wUv6bTERNN6sDZ!Iy2K7DH(sPNRO2-h=>6)8ktnQ&Jwvx6709j16qR^*F)=`1
z^}K23zavKC=%G>dcbTH8jC+qIYNm?AImzm9LSftC#LVyluk+yK#Hc%xq8Auj0ht>*
z3EyupYS{`*%N~q>fU8miZPD7w3PN+9G7L|^y;{<+CZ@h22HK#Pc68)`V_-Nx6DaI2
zY47Z0@wjrJqo-GIbms=zy3X!yq#CVI8A_1?6+?Y}z3oczcOl9LTc@tTpg;t4ug1s6
zA39e6-rU^W>}l)=l6vZaZj4MK2O5yX_M8680gFNNo(a)EwiOmIqrv>+yhot8@+S<U
z`+i8~PbO#N^T&SSgP70Kr9Olcyoh=IcKaJL2g3H9=%BKy>YTsV?{BZZVT*!U_eU{0
z2%G%x2q^r*rCgKGu9qnzeh9FVJ?~$4Ir_|cJa0dcj33z>f1ZLHOVOHWvZSbg%DWj{
zOo^}a)o0XTBeVC{oPdpNMcv9TBioa}xj)Q0kh$jhUu%3q?3)aFQ>C421J?IbV^~qL
zl$8V5l2e`>3hJ2!jg%FS1>O30x%)cc*^9ggUF8|~+-<S2@D?k|{#dTn6Wpm(=lNc)
z>!}doEu}b7VGh=L+}dOPX9Jn6L(Drk-IXZFpsR0V5UI`_O!;zuUsVGwnDBEQ`v?T%
zEKrelcIE;mA*&lh8NewH&bOtxZog4eUn`7+A5N{wvti7BWB9!~yb<UNj)hKDS`os{
z)haiIifi{9gVY7*e$etXn1r9e#>NH&0EOIWPEHUZDoz{{%AZgl;9z4je_nhCteE7L
zl$1gYQIL^;@VjvU^s@#aCWICd3T(`V-pbw9>rOe>v)>ezv21^VaQJMAKBagotLhCX
zb44R2jx0j!`hP!MT=<kqOo`kb2$`6suMS5zmoli{c3;LXpc@Y2>-mKZqqh*zmoxqT
z4s_=RPDwd-5=ZgUnO(ZxdQ=;`*7LTON(X5y(icWk%O8CojqK?=Kr2RUl>}-L2pFWw
z$kl`&m)_y*^!TNIT-?QT26qw3bj){n2gdTrUsQMAW@tUQ7YS<#n#X+Oq`D*a!sXb1
z({w^}baipEAsPB68<S)-_QUO!2Jidts03`l<*ru80t68_I_=u%i;D}_4hJXIyI%u6
zJ+PxYGLeAjE;&g^GS>kD2jF;p;Hloq?}#q~*2?ROvAzXXWBVra(ETk>R}(h7gtt?x
z9}<AlgbFF;;86TjQT50v#YMSMFTe%MZ0!jA)vM<|NTK-Y>`67v4Ai!+j7w?yl21wL
z&Q}>mPvzSG)fYiF+KN6Tb@J}gpl-^&ris(;*yU4)R!jam#xiI))@2(@#_f^TWuSNE
z?$a$<oF2nrw=3HxH6as@#F=n5NKcg9D#WFg>-~Vxbp%mt#M*q0i`Qv5Xwa(odHmKZ
z>&3sKyv_Z8Ex|!^KhB$f7iN-Ki@gj)UpwnhPm+-~8NQOm+S~?y1co<=I`fr;giuHT
z;(&m!hH*D({hG^u!?_wb_CY~GtEx4GA!5KY$Qg{}!QpXOs`-1&&dtTMTj0QfYV863
zt*ouZ0vB(kb^y}>GRJW{-zzA#0b3bP7rOo8qB&-9d3kvV+!Emj>zr)BAiDREX}b~2
z3u0o~Q~{5{O$SlX-G&-HI%wCo+!%nD;uAgKF-gShnD5S92~2i@omF=dxBb=)*ZjBq
z`RKiar8Z!Wh~xDL{>cLX03{A^Tm2JrROs<&XR^c>M7pVIY9z2jCbV0|w4{!{N4FMM
zo`}e~vCEh~TE2ZngA2w|#Da!RCfaZ`xZkoG^4}fhLSJ}|J0^Sm%4x#Bu#;7^a+M)Y
z*d=|>#qNyno5QK!?Uw9sR{#??VOpFJqgY=`yS9zYD^DjZYLl1UvVCG%DocNN#l9DF
zl<(sDgpBv%_x?Nee}83vfT>@0!cN@oNb|q*cNOoe9AExUFF!xYaZEMv^>%q0dayrt
z&GZQRuInD!NS0ivJ3Ik0_rEgi>~3*!k!JhT!+;|gm3yd)`7RcH$M%|Bj@OVE1g|zS
zM8(98`g!*cz@M}sQw7)<uul8sF18m*A^kNZ#2CPY2g3-(-oZfy&>=TItU1|VNPT#5
zpxU9C;iuE84hLnEGT}a?VseUKuj_vJ{_+fzlX=NCnw?XZ12MJo$vnop4i6SrViBny
zu5xSKuo1;WkXyB4a8L&jTTfm&|6~_+R%O`Jc}dR5pM7OGmD?g@s6L$4)jg1N#f7M9
z!IkYwDowq$rrTFwtoQz`AKfo_qdP55%zgnPU7bHCS?b3XAA&qTt5Wa%yEylm`|RA*
zX^WO?rA=20ZMVa<J9iX^rGfSicj3xv1}8x`1H+Q36hYY?m&7_nkF(pkz>cMrI?u9k
z3f_w$(;5hv7UhEqSu%<els;d@GFs;5l0gAf>aeK@ilfhVd3;mgHf3g~JuRRze(*{K
zPTN8dFr)kXR~Xbx@yv!#KwYJGu-JlqhJu8ozdzqF>$KqE?Ch+^ga@|k51C&_%%Ii=
zJM}@C1B^IAqoQ7dZ}9HdM}e_t`<shH<O+WKf%jR8C118xKt(A5##YTM2!Q)4iQ5h{
zI07xe9LlOgymqtX3U$78?N1t{bl8)V7e8%ncx(a|TX_Hv^TAXyE{Rq(XwK88++Ouv
zRx58Y_wEz_r7hK_v!`ba=zqjK=NM67DrTq!FvA!p3vD8xy;Z_7Evr~8n;Bl7tW)wJ
z;$1Jb&-;{KR*qS%g`m#!S;04s7=ecdao0RFi{35C`rVZQv1<RDaL9$5*dvHURoCfe
zx~erd9;O^-$dUjim|m|-QNFK1<Z|U&o-f5uDQHN4!GQKJr&Ui?92^;SU6;i-xzgLK
z+%-FW7;?wO*519Sb36ar6ecKUn&`D1XM|m@K~F9ony)H2V^INY0RR>QmB4nr{6eEg
zPf>eOs7Ar3Ug!risJegk+xBte0^8ePa>;*2b3elB@26YKa(-_i&ZFQmMBB{F%uHA;
zfGU^P1UR4xgX?L2kGlxk8g(!rmmSkC+!w8KRr3xlr}N_9w0bD?j8Vh&&ll#fnafDN
z*JJaZ_7R?26^7GFRP`{6x3I@3jVh$+7~@dJHQ9EAl74E0!M`(j;b8x<@_R@z1Ab*@
z+ojJRUhbK<{C=dSj~<~zdr>Mr4P`jIR0Wrp9O3PT>le#q3twlj{R@ka7AWuZz5T-W
z<*RI~GAfy&afh>2)HAxve*(%l6_@dHaSI!sg^|R@sKAUbWs5G@4*{YY(QJ3;2e&@o
zM`*7m@$~w<%=TW$fRQR9GCt_LFS(~q3hJ}%-VIL=Bn!gH$xUSa7cgAsxw$z>Ik_jz
zdQ8y+g9U}kBEUHK((&E9Wlu%mTm5%D>N8WUP3n!pW*o|&$EMm^Ei-bL0tE<iD+)hi
z2KK*meEz(ta(8!$Qoci?WvCTWfL2iV)E+w2Lrv=QeU5t^HJgg~E>VCuuf9$;HB^fS
zvnaSMv$;<2rB`z!F|3}q*hc(};Sw9&sI{T-^DC6Kj<&I3M0^d?^o+9M*7V<LSiu=~
z`t}5tlD=;4`ciO2J>Hn)-tdIkRq9{9*`r)uo@OP@J4@8G=%De>rPnkJKPJyW_-O~#
zxc}vglTTU?8{lAe-CHh4wb3kE1{WAO(rXRuliD<=vA!*H5YdVz>Fp=@JtX2T5WhY=
z*<cF)v}vmIV1ne$&D|Z^=l?9XWfSYlUSeS2_qK^VnGjI>nixHXQV#?l#()O%8n$a0
z7%1vIoan#UU0ArX(PCj`b-6hNBjE8eT)ezxmQyS+QecnR-nV)1P22Kjf)jj*EGC}Z
zn9|=<x|+y%GE#@fJ<?dZbSphg@9p;8!Y~C2*A_I1$!lp5=f}F&aGf)bL5`VNBPrjQ
z{k64>y5GO|o>0#NIoQftCo$kL{7u3W8f@Z#09EeHJ}ctES{_A*SA>v&yjP5*-ec=?
zzd%izUA#M<58MlIgdZ2{-du1a{zYjP!ymO(TkUvBAPqKFtl*vC@N&`H0t(7XKeE2)
z*ZUPR&6`>mSD30{Sx2WL5y9V*WY*nWCpNjj`4()vpdQi;jbW%=FXA&%K4q7q@xhkF
zx{d9ujn|DJ@Q7b30>*NcX#Y}{@K@yg#^2Kb682O9e?yJ!3weRdOm{bcdsO;5K-$K~
z57H^>(qqzNQcK793GxEvQfs3rN29vNY{LL9&X*4wNJ?XjDj7bWDF?QA!6zL?Y=_5!
zO_WQYwL-c!?wWU{n#Q&9|KyKd4JYaiA;Rtb;J{R7hpb3$gf(CC6|-hywVg|z3Zgd;
zi~b%TTAiPFE<Y=@``Xzb;3+TW)tyPyrGi#e;mbKAGDGvN4o;2$Z{e{3ny;grl<kQq
zI!QtOrn}ADx1VSQU@mt*MI#N%$*sC2lBs$o+s+m@+8*_b_Y@fu1)T5vZ+S+j-~V>(
z+=14F1uD5aHI}li^2ZU_dK{&x++?baLJw+z-mPc-96Q%`tZuP4pN4AwOHgD9B;Xu&
z0ZVul*vW_+V640Q`&H1D-hU4ONrA;?o)3<Xj}MQ<=&Q2*1K084!v}!Nr}Jzd>;Y$k
zUC9^tPt%7pAqRVBj$(Hevk*%q=)nEeEDXz~DVdKnxYImvhG&x1vXg5y(aG>tN%5X!
zf)~a>TRdMvVEM0^>KO|n0n1Kj%ks)%x)+{J8t2P!H?hdDBE9{;&1ukR{b%7W{kdrW
zkIK!2yGwFKwZs|m#Jy<L`)%6`?1Mk=D#lM{+!mq;?D1=BTY06JD^zIopnTH$=U<2P
zjK+>r-=fToEnXxDavhG4{y|+jzjvoIAL~Q=<OJ?b)m$&6Je;hl-K7HY_aTB&s5LZi
zudK9x!={DH;ui+?$N=@R^CTC#_s%!Cx$M=f)07*jl^DElz%_<R>gbRFgl)1hn2t)w
z(QE$ih1_u5I#EEit(eb%&-?b&y*47>$dhRvMMZ`GBkZk%vijcWQ4Byt!62kn6r{UL
zK|oqsT1C3MIVe&hpmaBaBAwD9pmcYqba&n7)z9}gckbLf_YcMqUJmE%v-gwhS*!KS
zFA5V<a?UcQN(!RyJIZc?-zqxH0`hk^atseR@L^zy{KQc`tKs>&J;NOilVF8A?$jz#
zuS_!Z!qh6U0Cl#WAd;qHb$*g=xu5gn<?->fPH5f1C5B1P{o#%|bsS>H%u9nAzT?gx
zPY_!+eGUyZb<<l-?xQJqb~+Y1CBo-LYkO~N+a8rhw~TjuOVG+>>d!m%ObE5m##Z@I
zvD?)yv8ZFpn#Zvvu18SxcR6kdccP(ddjJu4Sc&#CIk|<rCi2YZldC|}N#V95f4DCd
zPX7v05tX)AXD5e)!^2y^OG?8Q@E(DKq|oGkq6N8^T^|FHv2Mz{-&q&9k504IeZ(vF
z?z&M>;5a!uQW8TpCZL+M5#9fUYEtv9f~DKeqm`DQY&AYjRdInD*d4rS_`Laqy5q>j
zfGn}a&&R!3gONuI_YL*BrCw@Xl@`R9&~9w+ipM07edzdjfp~`S&AgTM7$TT<hqty`
zmPdIy$E)0}6F=P2wr_pDAvgI_OEHZZFI6EL7h42XWbajWS{(NRvXrVN`{xE+WcR3f
zZkumEC)i|bx<X<2b#A)$PH+cRG~r)@S;!}6U$}iPVwDrv#iJDaI9W8HJriS5_0_r+
z><gp|IA$aL7Q2hXq&$fAdw@c00E<X?i9c#7(YjxwxUs(%Ui-7A@1Cf=Hu*}*O8+L&
zg5t*2oh<td$rSy;*3H5df}i0-nZNjhF~sx3hHA)*C36Ncc^`3x2dxXfy6dS!?&%q$
zSlPW~Jz?gEdAzyINB4{y4|`jRX~7{+zRG$>GU<m>Ct1={<YfCvG<sjfBihHKfgzGC
zJN9yECD}jjyTY2WXaw0DhnBYm>C%MwMVd8*ExdZs=jOBgbAY_8Ou_a+%zKf%!o{Av
z?oWTRn!*CEU<ani*WR!IYYe~HRiQ|?t&2)W;Gv>FV$f}XgmT{pm~YxVTu2BB4gCor
zBg-A+8dTQ1%j`@@NtXs)vdKYd1eKiwM`VgYN4$Q+Gg;YrNPLvX4QMMLDC7(%YHFI=
z#|Ow*bPfpqA&}pTB@PhV`88ur$EMvQ)n~~5TA1;z`G;h8y3@n)wyY)5+9gT!_vN*F
zwzB8;zEyH!y=7A6Nv{gY?$ixU1l}sxMYtny8a^j{_WW&F*WN{*)82+vRe+tezgNdr
zgUP(UU{`>{;OICcr8!7Z55HmQ-a!Q6)9Y3SVO;}H4HBJqYgdwbe90lEP+FLIK+eQK
zprLRFGmXQ$;JO4gQ)18JH!c3`q`5S`keJ{!`Wp-KZNdh2ePh=H--I|$^E0?0=&EIW
zi?ir@DKTC_T3o6J#Ec20cdJ`So(!#HRcbS-`i?DS8-?6ICqxgeiug1j9HKIEj8tmt
z%<ToBBBAPad|`H(>*|vXx&{W~F`Q<GP&f_S=i}iCE3te;LzDOC1#p^ufw~`ZGDD%?
zgd!|EKN_euCRJ2YdTMTtWDhkNnm%j@7igFHtnHwCdW#@ndAOK0+30uar({#dfdWip
zWl7yPn$NNZI{NJE%eu{9sG@zupu~tH*Dz~v{k~~aCR*}X-4<J^iX{q;&&%bPm;-oU
zmTY2UL%Yvv!v*bWA9`=PUKsis1+laSL5Fly1Y78Jx+;SFkQ^0vL|TR_?bAv|p6NWB
zM^L`@6LaJ66Gn@tWw#E=?wvh=$M$$)QS)3ThP;ctci|Ly^-6%CzY@omSIg@+y|C|>
zg)%%xla!vt?+Vg9@Mx&ADHPvIf2eVbhyjOk$Q#zsGb@t-wUjiLX5Hx9!YN`|dBSh=
zIBD^V1H&2g)seHtnh!x_jg2ufuZ=ynS(FQMLCh<E$?79P#r!{0e?z9_y8L7j2GCNw
ztXIOYlt%-0F**hYE3gUEuCql+qYAErLqk(gcI>hu4#~>$ib&?XM<xBo&KcF|xAXkY
z_sIMfMoLnJA*g{Ss*JtQ!3^<(INEQ_8Ht5M1m(CSx<`#)qv+zzTmgL(g(Ffv^=F8i
zjmqM3d{$ULW~$n&MrH3?^^`T~YeK3rRE%VS)cVUTsDHRPu-#r^$&Q;lq95IYFk;om
zYoL68ZFh2MCdo8a<6tl<^bTzRv8siGQgq{C_59jGn(2$pqRFLY?KzX$A!H<ArG-c`
zLC*!dy<{iPVe{(|ck(Nu0I_QrmpF?#Gw7E#@0;5@S7uj8d7kLt@y6j4i3Nz6O(_#k
zd&5pVH=)jkUQSBaD0dDf<S$M+bWd*&yt!I2;eO08G&BTxy87sC2A~7}9YU?@lz`_L
z6TfvaCjf&)CghOmUh7MwJE|GGf|EFgWGN`W>>01q+qMGt$?3O|9~=h1(eNe+Hy+}>
zE=hQ^yh5?aW^|`Sj_7Tcryd)xh1e>l-Q0c<SKV*Yv`)jdoNKo&`7u3kJj?qR7naf}
zH%;SrFZqejoL9)B4`$HtMo2gQc=NiL+`wcuXvbxX@kmgRl;oIX0a+;0F*MPed5SYS
zW#pdtUier4)W}zS`{72H>&^#GQ-7wUYk#=@>Z9}TjLO_8>bzel6PDvOAE!=(K7O00
z`Jx*Ec@tA=5p3DZ&Yt9+wH{enS$K{XR?ePP6}UODV$PK~0b-t60Tl$%`NQMqgm`$N
zgpR8#E4d}rR%h+SNM7FO@l9cNE156Rb~}IMJ!lQa!?hHmBO;n2S#;5$?1B4j6^kAu
zA7Uu!?DICW<|N;ir4g)cdX2AjYGM=P<K4N-i)*i!I1<_j;kNIPeJqxLLPOg`W?oNq
zwz|EEe?TR}QR;o}7|giVSfI(r{@O6eJ2*wyT{fRbb^92p8m^mH6MD%SeYfU<*b%Ll
zppMHNPxAo8T!T?hHp6X;k&WA6&bMot6s<P%67L#{!ourany6X3=9pY`$R=O%g>DnW
zq3E^!ZhCiy5!a)SGnxB!f!9h01}Yw_@p8uoWO)jh9oa%D^te{bZ`@j%$H4}>g6gHq
zkS?LVHgfGRjd;v^{exePvQK<t#6oB`!ddPZo15y4oPuU=0Z*<iU3FD~FwHGlED!m%
zuvHTtqr~!<tz9m}_apV(oBKNa-2T)96@dl&S$!oHGaKfyM~7&xpV2<w+f8UN5gsDE
zLey2zKUK|i=4H0s?-4*8Z*qy!m3Cj(u2}vmf>e&adMDHxIJ4!q?aFa4m|CUU+tX#H
zq|b8GF1`L!_EQT19Fvf#w*<Sm*2<VB*@?wnHw4e@m#Nj{1kO00(`>lA5P3U!_J2&?
z-D-LM`oYPRdtZAg4pEmU!{^LLqw(4N=4NfBR}ED$e$MrPPdq6v&N@H)n{se{Mj8;e
z{z01Fk&#+CcKfT-1buI!b5R`(yXAi6!y;(QqXF4@vqi{IIoP_%x~hVcz43j?sSP2r
zSKoE!rq^^1hwE2lnLGVQL+MpywQO4$25l!0a#7t(c}52vs~=|<swCbk{xDoIy>r<9
zlh~(mjXMd?VYQFBB!tzbq`09NQ8%7Cz42$}$$T2Siom%a>-Ii|<pOp}=(_{eh)|-Y
z-6XEsa+c?&sr8+fqZS@Rpz}S_ZRN<7=CgLZDY;oCxTon=^m~wJnC<1TKp#$lgZ*US
zSIV%1^+w_ND)%M{+IQ!1m@j|r9LuZe)^h%f2A+m=f`#CqS>~V=Kn&jBkn#Kday9}%
z((%r;4KIx?v0x8~p)+AV$-V8@)NW<k7mv}hE%qj*NLLAD#GfC|J$&(5siZC*OYh4<
z{?q5Rk^2cXjoC;3udeY5R9BqFZMC&LxiPJ<|6sLu!<hLpx8+$-0i@m;YsZ|4QsR>S
ziO#nx+>0xyO|!PHZy-5_++o;{8-L7_d@&T*nz(IcGKIy*_R^d`w0HATkQNa$9gc_m
z=MXF7dG(o%5pp8o4UhIuFRtXoXC1$`Te$kNk?2=Xk3lyx1Oizj?tkB_u&+iPgZw1u
z9(?(;lj9@dFQf3myWVYyqDK3ANAZQ11L8uuWexh9_EqUzNQn@e0?n5DXSAI7x1}db
z4!*yAh|<HNGHLG|tW>@q?MrFY1$8w{HNQnklaFupTUG2jYuL=t1~;dMx6s{){ULQ3
zsM~Gk@{FOu4r$}Q<j?r{OLlaebS?wgmsk%^wdR%C;=t5K&H0t^lmUkQXl|z;8zpiH
zs_J2;vYrNVNbTG2Oa5mx4b0TqQGObxZw@kXJ86<#vYJ2-{Nw4Qgt~W$vh!;pn_KR(
z@*n1uX_z5Db)3RW#k+0oAm`nF$KcHGf5~Ag)qtlebOZ3y5t_KJLl2!(mtWynR&U!G
zXJeaZ;e0{V`ljf_9K%zGIKW|K{(IaZKGH}z#JroDkEUKq!d{?w)pfDKk{<Kc)eD`2
z#&-l3tc1n)zHyu_>fq9m%<p_-=o!2IF7B7F)IAO=tOT61`KJ8NG)k-4G(8bj)kvc6
zC*^-Go`haoR!H>6wzb5-)kx!ZO1pxiFJf=BS(1Y(jr)#rXv3V|z@0}reX591EC&;T
ztMLDrw^OOyz>2_WqA3W88G0h;U}%n8Y+S0IOrK2{he>5=X)|4=7S#31&@nD3#m;W_
zvxW9NY}lr*u<}>6P6zx?cx34qN(Uc@lef>+cae4yf#TgYM=LxBel`Vlw!dAQEB9#g
zGHQ;S>2hA^b+N`}N2>{`bm!|KQvF#P1Bp$L1}!(gnI4Vrk?i{FQ?8eH<Xrbv6MAlL
zE%=TX%04cM_}rWN?zT$@KCH<y?*=1*3UPW~QBr-b<=XLj${>%iNq`CaogcYO#H1WI
zRJPNOi<XuzSRKZW*p$g*S1h;-of-%x(i1qu_;fcpYLs<nPvtO>^n(8KjgwF331&~7
ze|QFbIG9Mu*4eU~oL+8EUDuv!{gSexwM7>Zl;ESINEftiMeVAI90}ZfTj?)S8KkRg
z(NeiCARL5sX>hm8dD=nz)nJ&sQEXgDxa_{YK~2H_WLNXZcN<he7a}QkHv)W=py9-1
zgBtul41WP%K2{aEoJAU{f*pS4`BkMM3DvAmLv6p?)mjeF714Yr$V5lmL*7t*ximS{
z9AocUcV%VZt2WDaj}7ymog05rcle8+F6S>~Sl`@mrbPk@txmSZ<TN6OF}_@`8RGw2
z_f4yO%EouaPF_DvyXDCx!96^xxfDA3+Wy+LvN(|}C0brY)HYr5sl!%xz63X^xV{3f
z_bma&s<i_<L(y&7teUHCry90K_3!24QFLeT>#b%rf2Tg8rhpWDRD&}`dH*hZv2|Y9
zzZ^JOe=GY5?0;dep|r}^=jz;S;zoe>B|hpH7Y6h2@K50L5B~lafOVFOPWEtS(X?_Q
z14;axkV4!ck)@#S-fMU7NB)48{aJwWWTkgO2??3ES41^|si|3RL6Lj5y(}~`IR2T?
z_spK6>%2YP-QUANJ<iwJ($YbJ)B3w5zNpbPOk9H&-LY9|vD*^qjGtV2-+oWM_WY~m
z{`OC%2WZvp7j~CGlvsX`bdf>sJ3Hd7ODQCd#E0N(JEXZek&VsS{a4|*5uTe-vpheq
zpT!Svuh+9Ogdi)1QuMdibHnG1PFw?r7tuyfE5C=1wWX&fM!M;k&v$OE&Zf;N@iY#N
z1dQLtv>DMA$+gRHQ^^U5pyS*t{&U;^v4PUj!<c0Zr$BynPDRdkLR!TCT@;-huF*rV
zOD|K3ie#<-?UpN6{`t99De!D!NeVY@#UK5Vv1D$^f8oM*A&sLp>ks7Hg2YF}eDgUE
znFLf9iH<)BRouvM6_~}n$2gm2P+&Gr3_s_EBi%==nkC;;tV8{XwI=VIf1};&i+i+{
z7M{e1tY=#T$S(?lJrluC=opGA2|@qfqcA(V0?K;FDE%3$3=P1Ski6MR+GI0nN)A=F
zP3-Hr>gHVjip#jUrqjpG-^cj)r&_W%o;>33zX}1APEgUi!=w!-?$I>~b7KM>A^wOS
z65Kvwq1m<g=0=BzHY9M*psqpKmG_{BzH<O+z1u%ULEx)_Eyo>}=^9mU_-;J}4I-AF
z4@3VY&p)#y2-I$)ZU;=jfmiW7Fob=V@$PC{CQR%^qlm+tY$sgHR7HlqOm<IkyGNg)
zi-lmtW50JXp@}c<ts_~5A=E}uLN&Hio6(cXjdkJ|tBi5$l>uY^oWWKIEfHs$@O!z_
z;MSh{X@9h9&Q|Oh#QQu7M!nZu&jkD1P9&XHTi34*JG-k?@DjBXPErs&JLW5Ae-{7m
zMTK?|F8WT`M3<6LL&hlV?u=s}xL3;VK%J=S!h7v{Q{{L3s2f}~QMGgs!_tJQH~5yE
z1rCd);9c17pUFQ&UQ?b>6EG6o8Z4oVTVEWr^c(lM_Rv}7e8S+)E=J_7=Pp(yAFtl&
zBaJuC`srwdg`u_UpYfH$c+LHI9U2@56X@KrTKf;MuWzj1c}AOjn#jz1ZBcz6q{|1o
z+_3)ik+<X5e19=lq@-G&Qx|-h-JbK~!GK7BcJ5z!*&8xQMAysN&MmSNi%#jPgsq!C
zsw=Y)W~DYfv^xCh1hjU|7VB;uKAzfoyF!ALbS#y5pmSv!pgq~Qyo+CjIv83l8^xpB
zYftC$_e*qNs=Z!CFWE-l2zoS_;03LZMz@1P1&<0Z(j$A!p$N9bh;oJ8teiae&ar@2
zDI?ND#N8o4<kGSIgXtp7Cl>FKcx83tKgLdh(_2%Air@)R6+X6gKawwZzIA61op<5s
zh1ZWuJ07xY(tmeKe@*FXhc)hZA;iyR!sR`|E3<5=s=u#gx?eFetcfegt){UgJR~9h
zQi)T|&fSCWbSHzis}?bAgkM%rlnz-hEg+}#<c9<-BY|A@OEM^gEHH>@{;-}4Z=Z5D
z+#$FQL_~jV%eK~KsBB{JuHO5ZMFS0oLF3gIeA`kxEw`p3luBDLoy$EF2uOQBef2DG
z+BjAo47)&O*hd+ODd>!W>9Ei!=(U_(@rdth7%N>#l~UR<`S1B(`{j$f8;Qg|$iDj<
zW_#H^Q%7AQZ>C!>MLuki?%FEGQyzCO{IO+RDN`3D;mg&bhIM3z7OMV(AxluTimb4v
zqDyDIzd6X`xODm2hSJd;2b8<l!?0FhA^r7Ur+d;Et5}^+tivr<_U0JLG2Y)){^@OH
zVB(6vZKpcbWG%FMAK{sZR4O(+VdR|D#nsZ9nhM{Lx&Kf|^HoDAPL-O(|G)%ZqeANv
z!t0BD+`qY4pu#KMwyx5KTAqtJU@%&|0#|^PA>1^sJo*nq(`3xkxy-9L=CY$_tz$dz
zaKs}T>jRG=yzZV31`gcIe0^0ScEfxk*bFoVOrMcBTV*xZ?-*12`%O&j9$7ZX1M>x%
zgQMUrWprvLLJaA0A9CC=zaxw|t<{y)#*~)y0%TFeU8mx6avV`ETgvceoVSC)MvLyo
zRd(IR!hZ(L4$`?7r`@;GN=L@7pY@TGvU9A@CcdS<boUpQep8<Pe1s#$(KO-k(a{5h
z;XOY45#KR%$t6rM?!C`_r~kyA|6mh!x4z@nAN?_3|B`7-evos;z=Df^<e=OA`JLKJ
ze7Xy%W-GQNGRD%u`CSbS`HyK63$P#He<p}gGI>i_Q=%(HezbzF&`x!$a`z4vrY>Lw
zgjHX>1D~e-wyKa~G;H!h5~QIT;mCwtVqQVg#O<3qk<1>YWa%*T@GesiTb0|yYIpZ;
zVoAQGXu;j;Mp+0fQ9*>2JPGctl4ZO?+vfz&va?EEvTEDz7F|9i#5&pPx2DR8b*e9v
zC|5HZDbC7MEY)Loe#<2&Se2&;j)ba#uM~Z}mZ8+Gc`P8V`8o{~gVYeRkzc_7!sz2x
zq}}XD-oq*~4R!YulJ}Jbmy-5(f~AK3?VA~vs6iGaEC1A##PXZCRY~h5_gscJsNndH
zIR)E0c&l+VM1vCG%+%CdmCQMirx+Ym(#jvn*TaSeIv3Ep&w^Azj_o<{(7z}&yopi_
zP|eo7_vn!?0ub5lQJapbMYY}3P}w-%n-mms>gvR5YHBEzIv9ia6dCy!$@b}<2=kq5
zDE)Sn<Rui1pKEGL1OCim%#JXGN+uZqRAJL%PoL(x9ovKEj&Fj_92Nz1U|hF;Q=1MK
zesygHMGZ#)31DGoo(#rFQ=ynOrjZ8!A<*Q(@O&%icC-P^f2{T^FM5}bbWROrTg&*h
z2Q*R$9ie^WyZp6+!K=zUex2&BRC;T-T|riroQ2Uxjl90U>Y8BOgX!e4G%F*&eaGq=
zYpb|xj1RDrf@$Aj%RglVj&ZTEcQ}mkKxFGR;0mtYVNPtef{utdlHP-oC{r#Sd&9a3
z^B!y#<3P9~E-HExAHM-P+osrC84>|>0$oGz@-C<eg)`~21LB=_)PVU8EC(nlegc*j
zrPB5PIMhIO>eJ`X@?SkPwaztCx_$vj{udBm(~i2Ioj3sr$IoO@f2b)p4h)olC~ojA
z1jWJy#9Ur(0DS)mM0$UQ3aC*WZCzb?-J9QZ&0B}ba~dsmIrnaZVI}tU4=T_9tf09W
zxN|GbVzV&Ba^Kx?gPGs*#gO~n1N!P~L=W-zC6r%tyTlw~rv5vPb+ps=aSEJAPA>>D
zO85+wkjC;rl$InK03==@bEiB!Jb?1NKuAdFy!*{@cNA_nlFNb|#uT8gNP$N{@S2?%
z`fT$6oq#!t_aH4C91)?qx|t5#^twR;jbU`E%)|ka(o`sE6rgh&0{A-oGDs6f!vlU4
zqZAbtMg8aexcXc~Imn&>xSmijJ3c<Xa7IK+63|~_DWBn^C|C%Ojec5G0o0Izr-TfJ
z1c&}{QqLlYCIT=CrC|V*7C%LB1(X&@zuzn<YszH#6-o&wvjh(!%Zy%HY4L)vs4`wk
zda3suzn{NW!uT=h;b4UAQ+Uhww`!G&rZ<Z_a~B0WEq8Q)dP*Y`JCfMdF8(h;x&S$D
zvzP`pfdXhQU(j*68f@K#PoOhU!~%rcK`NOj%$4hA10gq$GJga}176yw!)8)?<S{=t
zw+XW!Q*(0-0C%ypw0u<E(bpFSBC@vtNyyg0Uk|+e8b*5-QSjrf20|Sr+J16=#~U7?
zr=W@$HoOG&;B?uzNInPixDFORz8F;<H+&B`V4Z;4*g7^f1I4dwGEf0U=vG34gX=)f
zQ6*lgB^)U(Umh-^^T8pj18PP1>@7k<arpCDPA@3+?SfQwKM_^Ho-wBhByTTVBEJ^J
z3nOJca`pRZ*f|0Fo(->Zzol;>>qR5Fr!~F`vvKYi#i=5J@%@#<DR!C1dXN9bS67zl
ze$mH8E$Msx_?X7Bm7^jwQgjcd4Vdkgl+fHk&_r^-BZ>{v1BB@c>p&RPqREh*6&O{n
zHjJ5!0e%ndyGBX&@J#h?zlJgaUCYrzJhxRcj)8mu3}d1(TO+f-;FHlXd4}(Sb(>1*
z(a-SBot;@|(O|v=*!}>+-zQ-{Rz{9?jHJI19}z~=u(S6rv7l^%$!-+?N2p&jpI5M&
zlz6gLbm6#uJFwM8FSL74bAkB_PX*bfbensO6dqMUxaC#9|LP_~BHUFLE?f4iM}wSR
z6<JJKnsh#;w&p)Om}!mS3W5V1N~m`MCW4wZFi-3_jfSp`K*wyW134>bE)`3$E}CBK
zNzW9pFSlR)R^bN7H<U^kypwRKnQejrvmP-kfl&jUb+5KSQl5*?r$&txRbG&@ynb4w
z3zT}zAmLI2Lv}eDtxIsSk(c&wxj>JqS?qVXv2Su799gTZK#_W@E!%tN#*FWO)VYw=
zR^cF#2gCK4NgGQ<++qIU#a1!wjSk-bXZbj+x9PMd>5^_Zm&H(I3|^sW800a^bLdy6
z{^9%wl%zoh1SQ_s+pFa2>WX#%sv#GkwkwwA790|STuAfPQ2CW5E9370^>SsWa+r2P
zP4@spkls+f9$0QHG?=AA*)b`mN~L%MGmw%+qoq$&i2;lt8ldJ?RwMJ#$3gG+zel>u
zfD#NHwZ0XUp!%UazIa3IxD4N+WiWH9Z4a7>+UlH;7jkMQ|J`4RurDvxiq0^Jz0McC
zw}Edw+n+r@zhU^kWnGyTAJp(LUxKEy2P;wjcL5P)HXQ-}P_H?aY)!x4Z@z6VPd6-0
zRQL9p*Gx<#q)A06ym;{o2;|gY<_qL0-vB-u1w#i~{JzDLii!#kA?!*!c@;wjx@MJ!
zbaa`*C(Adt5iz3<Gy*)wl#~>ZMJPFnfEUo!>3&{X=+jM^)w<j@XO4|a=N9d*qgg^O
z_wPvfdiQ*DVFrW~#K*6<+pu3>bbsCt(W0uu@<Ua*+f>qMcUf6)c>ssL<y<rVTIoz6
ztwHP_&}Sal+${D^{X1x70BRhhpbUecV-o<p&2fLsVYv|5Uu&Kzg9!^>iUKLZxo^Mo
z&nlJw!yDU~6h;R8ULm;t_U+sLR#cm}l+o6!;paIydLZ$gH8c0bD;6(;ACG1${_(dR
zG&IYK_|N<&Yxfd!Kp`kt=j@b>z5>lgq~=RJ8+}Py{a4NCBE(Gg*H|U-gN`y7fCDZ!
z)YMVr;NT0CAR-z_$j;F=;Zg{adcg8-p`<48@N6p;^g0utpaD%}uhB{Ft7c*<V0WDY
zkh?8BQ@@+{HKL1;76^%m)V@}V{D<rfZ5UZfzW17w%BbruK&8U14-cFm%lb2~f}9H-
zWb37STdMAe@&?Ol<*NTz!@RO>o51Wbe=jRc^z%Z(C0&7jC*Ck@ir7nGAIis1NQ`A>
z!dsh%*-=3Xh~ivC2*l5u0dNR|R~Nu;s=^+TFgG_pJViMl9vg4?exAenxbwo+xXXH;
z>!CG>7I!U&i6SfDps=v8_;-61Ip$TAv(L}=t1tZN%On&6NhNf2bdPR|gXsqUitTnj
ze06SMmj3@k)K<A2^UihMU;$-Pw!e`}jJ&M1kz6)`$_wTsb}@77@VUIdQQi9+1YLv3
z@Wy0Q&w=Sm`K*Pcr!?7#&Zqo6F}sgcgOOML8Q+cn>Li*m91)s5dZ-nUua=VFenf)?
z@!Zea+N9yl|0h5-a0F<{DsBA!?>JUt147lgg!2<>cu3d1A%l-3T-nVmdR{ZE0ExKA
ziExI4Ob1Xp4M@&_jzo(K!GN;JPg8nUR3-H%Lt1L$iy_P0x}2KsI}L}T;Z6S)(_XY@
zc~OU`w5F<!%)`_UxxiI_CR#RIJ2AaaSkD-|uhVR_N*rwP)28^9k4HYPZt~$WKGTbh
zi=)F60#~O#;}|B{fFgPHL;PL==-76#8y&h-{{GkyOucFRNJPqQ=>bY51AGy3LZ_kN
zY}r5-2~=x9^}4jYoXz&n^KeEjNhKvbkUX*rOjpTN%YTCfo)D$!TWC6TD?m&H<dU7H
z{itC;^JQ8%*8dl76!`}nwSs$Fw$r|Tvmj5igs1`qqak*XF+O5Po%r0)kciCu&tFl2
z`n7cZ?A>9R#Y?WyQa&@kBz`~DdGRi!grselGkq-X6*NKn$)Em<C*&V?a0>EQEpga0
z9z=OgnAYbW0mLhLbkX#ZC`RY`P*trE{U>B9N8oWLd2K+bRiFF8l#lazPqr2x;51tZ
z4Kw}0&l0Z>7#Dpc<;fqfG;ATD$&`hmgaZ(vTq2drQUTTOJFr^1MWn5MY9lj1<nX=5
zgf(ig%8d^Nc!9Y<d}pM$I00qUp7URF4Gp-%rrKYrQh2LAGe9oRn@|rS+&HGLwtP9H
zvLmWY7|C^N@Q|S#on>||@A6L7ri;M88juxbzJt}v0~6d%gDqy+b8Fv0GVm?;yebMw
zS7OPmHtm3#1d|u%IyHb4pR45x`rHWk{^V)}D3M(w;|+yShW)7{1Zj1qRb(etW@}Gm
zZK+c6Jaa0Y-Xhs}fGJ#Kh@pu<`lex^K}A;MDHL%AzrFGx+_CnYSMU6GPXyDz=GDIP
zt)Ry3-HC-xlj)cD&kkRvFl~S_ffw|%t-0t2rW@7ul#bLghKp1`+kP#r-;trZee>qc
zxrQ+mlOJG^dzXZ8cTe{x*kHT{+Tv(TRR{M?`jPLbA>DpiR1BE$EbYFjt&X0{q2Bh4
zYsi(ygBgNF?!y%2GY_OF;#2S(Fu6Q)>Zf+?Z%kd%(n|7V+u8Q`oi(+c%`9+5)0Df^
zlw?9q4X|A7LUR9ANpHwR0|XKxh3O!nGqgnH{A^{UBnZ@-!U=6IkULIZxQ|d$qU3lo
z`|=$(U-Q~6{e&5mtT@z2h`dBK^?g7o)>(YDyqkG;;q2FY1|t;Awj^Bi4uB9hX8aRg
z!_bE$TI&^Wefz%kM3q`*i-Qd8!roiuGnZow`l|QCQOg5s<g4Kc_W=xU3==T+Z0=RJ
zrh|U@@&<@vBMJb+fSCbE0{&Zsr_1xd#w^aY)CBg0@+>RJtC2wc`tRJ#wZ9OdkCjr_
z{XL_lzawr$4-Z9|YbhF!o@i;f91hi&lhqD+pX!)RZEbGbb;$oa0$@Sv0YeN@^AQ$4
z-rg!a^TS@f%_?8%XlbWma<kB$7l9jm8ioSsUg#LdM#qSB5;Gcjt|UVDUicEk@Twic
zNpJadNJW47#&eKUnmgcO`MgdFyTm`7e<o4fS$8|@!r<Q7($PB8e<n5gcfrwplL$sF
z+6ky6F;1vx^(1t@VGTpp=jN2%kHMP!#Ad2^xf+UE722^wq|UAMsH|Xf6~y@5u==U$
zSL1K29T|{#72e_wzf^7TxR|^sz}scqC33^6WPtemr`&O9K=rX_W2>?gWyLF}`wQpX
zQou2q-6qiLkHSWy9z0nh-}&vp@85oadzUOZK0LJAoEAlcLAYlB>Hz}x{nZhJf6#tn
zI|Q7Z$H;Jj+qhip*)H>UIr*wH$+Y`Wti9%LbPeDb#sMa*(Y_;2w@j5fo8BdV7ehZ4
zm`Z;$Z6BX}?CB6mA@$!%KMm2{y=Vb2Kq4X{G(1AWls@ANz`#z&w4q(;TvSn`0c3Pl
zIqGJ+fj3c0o>g%%hRwVd4}s2u-<TET&V(>(ak-|g<92mte_=Tr%Fx*bszrv0gLcVf
z-gmmYr+=1HJ-7I_m%M1NcKq8{kX;z5Fm3ED4+Mc1xdfT-TTmxNLzM)8;+G8#64^J?
zKmE`^O)k}S0&s~Md}OHgX^*WO`@?x=Ulwq&Ixb?1^}<60H4kJ#!E2l0>M~ktH8-O8
zMfw$>uuhJaa{$9xcCrWgRQK>A+<IIn9$pBbXR(pkhnvs%qp*x3KYkf@z)L2!mu<$U
z$0Mf4ReAX+z(_xjhask=spB@3Yb2-{KP7xNagzN%_kBerp?0JiIp11El{4g#6tN4F
zGc)xh78SKHPayK-$&<rXJz)_MCm8Pfdmxos-mqIa&)f?<E%n(xZhzgYO8=9`rM%ny
z8MhlXZarPD6Qv%FxeDXBd7{D-1(rxOwt*LyCO>qKjQ5qSRl53HUmZN(RcSwQj_i$W
z30t!u?~d*)cj*RWwnExQhmii4oV+)5g%vF;FIWBXi@0XQqFM&!$RO?f-O<5R?Y1@h
z8<jf;G}e)OfsGJ{IT$s*8L417>PKAMBadCUHu61RQ$8U4Oc+!D8NZzSsD6FtWaP1T
z8);BLB|c*bVaeryxa-hF4|ZfHf4!Yl8FE;Q^X9@01ZL*qyeYJP#IylypqcTxxnYtQ
zCr<M+zc}s!_A|k_$6Je+_uPI&m201qKxv9qNa%kyE}(G<T|1*nCrL65{B8V8xW&{j
zUqDU!T~1EU(CKxr8+D{z8bJj|X>wio;g3X1#MPfTaOEDh;I){N!Gu{MtDoltVoD7t
zG%D9EQdw=rfRNJCMO#vWY&^#Ss`{O~+wWrOJZ|%FbJ(QmIgSwfL|+qgKiqeKnzj$C
z3wIq!V~IyU4nlM#mdOJ@;`hYo+8`!3$*m!&Ylqb=&Hu%d?D-|035F|p?@Nu?pS>;N
z1X1?+wmEBqdp~z09NFU$<qo%xPYgW7)$1lSRFK7r7>1KJ3U8Ns4Hiv7_xLbH!mVHV
z5nY2E%|UWjA&V@p?z#vbhSf0`Sq3rLT4(E^rC<^0pxqul+qt3?9j&ooaXd+#R=TW@
zZBgvWn5~Sfbis^ls>q(Xm1NvsuY<aJa&qk{*!QIfXfdHqFmin|y*?oB$!Wk^*C3Iv
zMpz%An#*!liQ`NF7wU`HmI5494vh7wK&y8RpeQ~jkJhwXT5#sQALrhV2>e;Fd)jwB
zKGTqRMT_*#PPd_6<H`ujCw0yJIGLJsEJ#&yYUXziMrw$CnpXK6;^o{ghzD#rXJwZj
zs{#ZCv8T>eEO@vj&Pe8$;}ZD4V?rDUe&hVG13@JyPVhzY?>>YCC!RtgH#&u-d$Lqg
zP)R(#hPG#nLq8^_^4hun0ZnvlJnr=erqVj4>K8i_-ersqJJ7P$#d5!JLw6y>q;h4=
zYCP+p^8DOZfIP2q`pR$_;a1`+mL!Xvib7+K8oc%6s`Qtm)&FDEy8!Hr>fhW?8Lv4@
z(|D5dt_fJ^=ht+cMru#pp-}_S*s&#^4^*%PVY2BS^aw1KF_}bbUGgmNV_rjOQolsh
zEYWTJksk~W^baDrU44eB{g$JI;{$Y#3I`vgp(8)hn6|enS}@i5EEg?FBjmK1nY{n!
zJ(t(K4ex*LSwzb%3hP_+x^bOy%92!c2)+5WF+<@zW?2hnRi*S!xbC$W7FmMSTXxBL
zMN?Jhf|8kQPfTM<yiZV-@=K)ble+1xH>$rWUad$I7Tb9+y^CqJX_dp_-r7IYj5KNq
z&bv8YX!fC~ube8loLoX$FGSzj`oAj7z3g-`rVUUI4Yw$P8M2#fH-*{tB7))+Ih}j8
z=`-bRTs8)*cVE+b4IOfq4%l?jTP-Vm{j*m?_I|u4W=3f3o^uP!Eu5yL>K88rf2M?S
z!8}u!owUvgFJ>>#&fH8IlC!>M9zHkNLnbJ7uzmg)3XF1*B=Mr?<E5qC(#T>DlsubD
zAW@dj%v`(pik(<ON{W(|HUFd-${8BL5VI)JadAiuZ4?#oM+OOKGT!In(70NWJibD@
zy2xS0TuFt45#m=Dd(qYl)oq0__w@m5bK_!Gn}Lx!WS}HyEMfCM2P!;@M^C!}n!2GS
zsu_|{30%gv09a*sDtOpF)<2R$uK9Vw%T^@H4Uxi;QfYOyZCIFG)c$pMBzGkJ(WNxJ
zL(k5kM<3pR@^z?dT8)j~@c(YNGWdA{Igf+!zD8NmD_u`{Tq3l~;^h^#Yml->0aMeu
zQ*#@_FZ>FENdCxtf@y0N?UDaopeN4|V;#icX^RZOt=1c|U$i_quwo~c2|NC#L`zCW
zwz;)c*U-=q%ntrdMUd<yEI{#xWlqlF+pQhTq#NS62EZqn6%u=cV9RJ-4)E@=&2P`*
z=P`N1E_pcQRwH4or~)tQUb{N!&AT>KVYsrQhJ=7XUjjns+~~x!dxoZF%qt!5T`zGf
zpH3-e*=wPv_yH>6`B#PtQtc~rMnaE1W?aX7<>IY}KE6sMS#eJ*D>UkVd)Qllm*6_k
z9|9MmUux<D6CY?m77XziVO6FR!X!0<$%Y8xlR!X&Hr6=ts$u1QuC!0m6hLO{v#<HK
za54SbB;^hcD6b|oJ9fx7fA@b3#79v!$IR)41%uU3!dG75J5(`$$k@eXc1^fEHkxQ6
zpiUp@aUSPccxxE<YKooQA`zoGDF6vD_+4-TRCn==1e3x(dxCx!p(ktgLv@nHlk$p=
z5-|T=pZ#h`*R$5Ir=;TrpHtu8$!AsQ<mH{2n~^gtxnR34bkRT+JNw=3X$(uP@mhQW
zg5dc0K-u(5iO_u+n8^kA_!jP$a@^H-*Mi=&M0Ya4UNWhK6XQ00)fJ6yt37+&+49>Y
zf@=4czAhjWm^UHL-z-m=OfZ<zS*@)v-Xl}=pK3FF#z6Jvsa&1duF=7pe+4(gmsBBs
zLdDsy50eKZBz*x@|D`YCwXTWy&r?Om$;|<)-HobtwX3hlR4lrlH@;y-5hU|EtbI&L
zp#*wGD(Jc-CMI5;fkMMMi_N?=%-Q;BBNkIj6L81=YqudcT)f%*+9)U9OO8=k_PUTu
zvh@^&d@d&u>~Si3cLG9+{pE<<8z3K%@#EYxePfvvE#2nTlEh#vhL~a>-yfM=LG&+i
z?`+U7?-+Mv3<bTFbe8%XUHHam$iEV_(f?@B-irB5(zmvJ)oh1Y$meFx3i4*9+~CK0
zfXL|K&i>_n7d#~xIo`b2A#kHPO1B!Cv-<wNzBQmH35_FmKV7c*!_kk#Z!j$6<!{#e
z;2P|&jTHqlJbLsJ+9haEVhDt!i}R@@JdpXX7Trzbnfk_(YSHr<d>E=K`yuR$?Kmu)
zho$gg@+GI{BQ)sOAyVWo*zto@w*0b7t&GM<omNIvUz{H)$Yzh^6lxvtUohYHK*O}m
zduu8HUj9;4GiRr6-QzW9n|Fo!|F$?+WpbVCxTEt8t90Dy(eKw!sTdoD-Z-ZDUDLiG
zaAjo|O<+BtA&J7^R>{7}cqD72>cqJp`RTfO>PlOB&mz*C6Kip`$-H&@2NB0Fz8_GF
z4kdKJyiWX6!=kbQIIy9`5!Q%9#0n!>V2@Dv3{AqR8FOmG#|HLph4CRK`||lh*n0-b
z#0&U;q|VdV?wd*r^+scq|9NsHGtVwpxAj=A)2$B8-u-}$a8JCR>1$ZyS9arf7$-t5
z`()$Do1t^JlaMbGWrUe}VU?<HlYf0mn$-7od3y=@kU0L~gEO5hDfi7aaZ$^pd{STR
zPuK6KtW>Jf;FTFfZ=T4j*A{nj(ik-#+mVuc&z7}s$FR%9$aT1Oa4zlsLF+Rgn5-kb
ziiU<(8r8GdE6y~LDKOHXA(kR#+wd!4j?i^Y6yZHf`n+VdzVMBFOd_+2EX^UlQpN(=
zELmbVw0ipuLvvxz!vW%P=-(wgkbo*pFSLnfXJ#l<x?Y&pjKKJFT?+vH`+;C@aV;1v
z@kK31z3pm?!_v0}i@sWpa$DE^n`sYIf6W>w7%Fja{OOkNBNlTmlTOr3){R+Q%R%Cw
z&wsTgJ23H42#Qm%m;7?E&J$0)q>R|!?B03jW4Q<{)ZMp5>j{}OoDg08n2|##*3M&5
z>hdMrO`MkQp>~YBi%3GnuCI1<Vd6zPyJ{#3s(k$+;Dl6s9*IbiiDFIL*aB7uI*J4v
zxtn-+N{&vz=CH<yiPqj?xTl<gm}m7a25K*{Eb(gbR#XFsL#))F_Fk@fuDT9kW(#?#
zq?`CeM)v-`YGx+y&9#eYXo-?mym3eR-P^&bQWodW-D^2o_{CalBJdRo{5Z4D6k`S$
ztsVRpLZhN+3WmOa`t-?WO}^-5?Zv1z{W0ju8fYQe@kP)C2^}(3F*ZiN0b@wUXG-yJ
zFvG+e({=<6TkN0$^{gWp;P3k)Ljr*RZ&1p128saCQ^*83dy6Tj3DnL+8<`kA+Zz4{
z7BE_4TL=gW8px1O!6PHff{xkl+L#6?b-4KL7qAOB!fYYxPka^iC)k3g0iAL7*FUeA
zA<&`yg(6Oh7DB4XT{oTp)t5Asnm+^v_E+wYX9kdPg>(CqSRP*;XS=-zm_b5q&d^*V
z|D1K9hYzTz%F0R6v|Hm1dOq(Shg#33vW9feQ4nWz4&VGXke4TYMzAk4YfJm;_|?M|
z{_lmiX|{?Y@wCg!mK;ozYnJY<#b6@vdeNTW!&x0m_H^N-t1##5vp__CG;Xj&_pU{=
zsGl9WsUMB77p;ztv?+(vGc!NGgmJACwp}5J80-RH+29PYMYzq)%^AU%*T2Q(jFCY-
z5T{wR*vC6o{bu&%z(@J{b5eU8pUFadoL320V_C?qdZ2rQd^tNkrek8t8I)UyV0Wmk
z0Dg{s82rIr2;LHEsD|Q9B{0Jgv*}BJZ3=zhQ=bGxaMDof=UzU7%-00Uj?wH!!NAv`
z1TALZ_H>-~yWa%q>@#SK_rXLphxsVCz;-KJL;J$@QSuFPZp*}t{04<$G%7Bxfwn$0
zWJBz?&#fZcX-%_pO4<I_*!v~N>e9y6P9(dov(BQ$-&NO~th@7f1`M_fU+rB@qhK0F
zR53i48rZw_a5dBB-9`wFnsHH!m_gc~b)X!KH)i!oYJF_UuJj3pr`(hyVK-$Ee=?N8
zjUXX!z^DLXgQK|!9XwLf?@(oAMS_H28(b&jRf4XYVwQSQyZT2I9fdJ0lyO(Gmwrcl
z7W8&o7Hy4pm-@=4yh(O|ZY=!R9krL8|0MaaL*KqHLtX@Ge$gzt-Y|biE^xe3fcg=%
zkX4p8$DNlMWFncL=ohGkP|NiLjRgE{`k1yaP#!dFMnY*sd9MPI&2wn_48V<L8~1(t
zZOBe5AfN?W_uXU;qe(2fjfTJuU{DmiOJ$Dh2=q*$pFZ8CInO~ur?ZtZ?t59Jj?E-4
zE?$n>j6WAGjfYZys2_D{@~q(eo~5+I3L!>+c8`jto{FqF+)V46eXB+e*>->C_G)GC
zIdZ7+hCOU`&UE43_2})^g=X2d&}+f!7!nMgX~tAE1NR=zN-itdahvCS5n`9W&S;cf
zyDpHJNSuUkvv2{m>~rCkZ*m@db+)CBXxzm9mYrzaB`Nqrr+?Hmis5YSQI1lX-7*!B
z^t45?h(K*a{^jJ5S>Jqn9G!&oPM3HGGIFbkTUFD_>x*=h!mQ%n=H_NJkIffpst`cN
zq;+;8<NMn7Lwf3Yuf5a@ADw@(DH*emp^}MKI<SC?j=AlR9f4%xP8-*RVZ#Ufvb|vw
zDGo3z##0|AK+HE2y69ddQPNZ#X2Yl#OlLg~pVpo&`}FC?^*hXut51==80Wo}uP}4`
z6eS$8JeG+!I_`-|HM}CEdaS2f@vFi5+c@x`RlKj)A6O9=>Yv7)|20Lom|fQ0D%+jO
zkm|B5C6b()w3|v8J*RVSd5Pwys!6rl_nbBNlEJX5hleIx6wzm$KTj<8n~A$4Sh_LI
zQqTRv&A%~>si`_6`FT4s;)`kX3T_D5flE3SuDgt%%QW56J=Ef&g(@0%V@c4eu8$$x
z*k7IcBMCo1ae>_Z;FSbzvCt}%Nf;!`7UEr+InDr0;cu2yW+?AoJ#~{(EEa!_00<us
z)tq$HP#dyATj*sCnnX7QD3SvG@6ZopCbV2uix2~^^6Wp|Y5u9x6gC!9;Ot<hc|~-X
z_G$)0?Y4AdE30IDY0VnV(a&i{-*9AoIaQrtv7Ynrf`v!kfoGdwcg-^+1&yx~i8xtf
zSrqY+^y&3Z<4(nA`C{7>u@BW;Q_nd=#JJiM3b-*OwCdCR%+kjD!ooj&U-w2*S3*A`
zTNCoA2lW_1zdXO55F8OQzxBdt=ur0nQ~N+b?JB1n-4z?`pG{3BP}Dab%uNDwqmZ2k
z19$2xyP!-H{y+<wbr0iR`jl0}Ok^d#0ZB~WfRB%lRB(QNzNCPPre^Ge#YPefdK)qr
zQtJUf{S`#S=^j0L1_x5V0Qq5=wi8Ik*UYc4hkAPG@B#Z?$b$f44oMJ|KfH}!VCxD1
z2rmKty88OQ5(^!Nh)XrtE7}!mnbn$)(2pf8ih_wTLf3vq6D$UVD}*N4pZ$<uuF|q@
zuV=fZ$YIdBa9F!)Ba=pcGcBfRno^+WX&JrdZ=c({%Tg1eYy5xhe+|0404suo_s*1S
z5;vK*C^vrf%s)VX#9X@L_G&B*W;bCxvl}wFtlQLQYESr~k}uH>uoSJeyf%HCyaDQm
z4<(MMEg59nU<VSJu@h7~W|sE-Tu|&}m~6e?D>(z)feK*ocY(GAHK*9z+DbqAE9>QU
z2^+%ocAL|Dy_V$8os9LWV_7gtO4io-u*%7RkV+`p9s_yO-PMwbrIa8++6u6JRdm~5
zOwlS5dZv04W(*e2XY#}pl!$aNg<aDOVxCc!82*0GW6YJ2*?OTNDDO>OM{`TDr04xK
zDXFtsWqQxl^yp_IT!zb|9Azt2XCI6akqJw9``^~Q+~gN31nBrQs6r&DbV{l##q<=a
z&_hZ+hHclf)D*P`KFd^R=&j$JBd30~aBJwh+EV9CnKk;SY}Py1hy|Pr&$biJ4Waez
z49%oDTV8C{op6J>r6^s(^t*#H5;XeTn>~^}-n+sXDp}E~`F#Z@1B?M=d@2Zl1djGb
zEc$>9(aXoD#gM)ECG#Wjhd(z)G)mpjb3!LXQWAITAnxD*J;8#D<K|;!arkTQ5D+d6
z<miZoUQ7(;BP~&-wkHpZMwc9u`jNah5sfJ!a6i<ER*P@G0pI1BcCj)E@*=95cz>^K
zE{3}1(U`V)aN%0HO%4Hqf!h6=+*!KLw2vMLi<g`vg&vsAksB<Z>_3o**dw3K?r#2`
z=(Pyp_(KWc;4FvD`h{G;IRo@X%oB1T<qJ;5r&9@aYz3Z8U>S3=T^W=Q-htf_b`%Rp
z8>wJu{0dwr<$lg#-1{jf$KutbR_?o@80BX*bQ1MG2<`x&b8Hvl9Ks!2Cu=>6fUFb+
zU%^+#@Q^egNFdo_+R6G{Em>S~)c=dZV(QCli}CXT`yuPCRK`~d8Q7V0&dMRz(ig1a
zs~4D1kCHfj7~M*2+o5R9OLlZzE<R@)dOEY`!`a%^#ZaB8uXJg*Vkz)yM!qfd2p~Q{
zsLA*ppO-tV_b((Mj1tZOnxlJD1*Ln5V8MgQOL}kc#%%O~xJA9#RL!eZOGQ=NDeb``
zaJ_ZHDVDO6t#xCK0<x|%*x9A&+_>LzJye4M2bc~re3Iajhe4YSpljI=rNRv%eQ9oP
z?uM8N{B_C)`U{UAM&u%=mS<jjjGQ?_Ztit^A`9{9asJ!yqf-B%={@*DyRB4aqWjm)
zjm{5ih<&p^)&DS2{MbZ4kg+E@c9t@x(AN6we#5JvYRZtI0<))pBVdpq7ILqkl8t?|
zm|)p6w$z`M3H(RDjEla(Ts0^&C&bF8lj8VHPfzy(cf(aAs)WC;u5Je)cv8Wyb#+nn
zxpgwbEB586xEP*x?mGbv5C(k?J&!aW%zc96BN7aF0~%CJCCXX2Ek>9s@d+YnPmJrm
zkgZ{}kfAXJTd^XPV#+po6EcbNTU@j=PmMRXAH-Kxm604R^)gFg=7+{deydp`bXQgW
zYkwZg6I7!K9FLO<*mXP{<~<kCOD9tu;G_P%LZQATA)Ai#(=BIC(VLcuOT^PhFh%P4
zT9u_}#N@C^PS^uCcw@n-<!fb9XJ=>loD%@zN&~E#;px$?@o4F$S_y~6xTc8rwXXMA
zw`XS*lT1XV=6)>1_rR`WciMV`T!++hP!}L{+KxXoO3asr3eus){Vg&w@_rK#7s*qK
zg;rbG=uP(LGy}$^2m;c!C^m6a$_3T8F&=&o+Kk-a<(aEEvB9LWM6Ms!x1Vzd)fX8(
zkg#7E(uI5l4UN`|x2tWp`CKEzm}A*^Vq(IuKeBQ7pv6Q|W)N}ew{qfH<jplNb|;w-
z_dnx$>hy!CY_V#=ZlONMDPyYuun=+h+JYw=_4vFxmX^82W+Pv~G-d)P7AtXpF#uk*
zfp*Lc-i^HcA&D%3sr9^sprGIp9Kj;{Rdqm;N&`YFtjkadmc;J-bl(ITuRw;VGPQ}}
z`2<1E4%1<`WI$@s($pjlv&~(dNg_o?JyZfPgi2n)wKf%wUqQlY+6j*M8*so8{K$el
z24%CbBSbFMM7=Y9zjP2tsgTT3efq>HZ6sCG=bf=n7<bSE8kUk18@y+QlELpx;%9Pd
zP-}q}WZ0XW(uh>5=u`-)OSvP*>Q+MhIi(cw5x<3=p56PLKbK0!w^|z8euDqFXql_0
zv#F7;cwv`3`@|H0J@mVIK#WAcS&V9J8ej8oNT!Z=6qe)PeENChDyLc|Xc~Q85S*w$
zMRc(b>GyaCPH8&7DpErpI#<yo5&SQBnY!h<6}+@Tz?d_M#6jK#0LTy59nwz3S|c>A
z{oGGB#Z}WUK1c5}AGv#8cQma*7dpJSZBk<L{DdEUAV<!|EIveK(Xv>0SxND+<EQ^l
zDDBf1PZ<P`wj!L7BmFUM(~rdLX4IaPaz;%KG<*zCl5TSJRZcKCPw!gnPPJx*F^*<Q
z+pmu!S&$I^HsvHVn(%0@O~bDMFc>kr;q$5LOBXNh{GtrXfS&`2Yy|2Z?d+V$_?rKi
zBZ%X*lYH{z@_n?kt_llGhL|0Hb|HD<TMxrJUr$xc&v8As7*BW0<Kp_$tvUZ#NNpf(
z)~9f8wSi9viPv`<mMSzvy$KJ=EcX<ozs%<6GV05kH?Fx|GNOU$vDggyP<XQmq?g9P
zf<p|wLGUz8IyyRYv$KYq(+y>(dm|4eAe;^9GTgCDdlV=|=eW=smD1WOEpf=V*3i_n
z4DoPp?yGus$iBz~oRvVQI_!a3RzE8P1E!@_#u%r%oi4EyW&yScGR!*ii7HD%-Z0MC
z*j?1;H7-lETi;=E%flGOql(zmTE(qYqD|I=J9sVz>stYTZQ;fh&)P(fnj3XvuenKa
zDz}>H;H2k4$bJpO^W^l@ZnsaBj)zBWK+`6G97ssyBK&TZ;%&6Y3zTncfyu`wcD5I4
zhFWxi^Q%@QnxiFb5lu-H!!z4)sqv)cHtw_I{AT+ziOd}_cbx>mAl%R{%`My|^K*xy
z6Ol{Xp0oMhzd6%)mdt5-vLj<$GMrO><;VTa03SbuQ^jC5S}G6pUrG??%+1WCKssdv
zLERA`AUJ@6O=Vr>H6wZj1VeNcqk#TFUIzh3DMLjC997jXm!6CEt&b{Il2<-ywtFUL
z5y2UeFSquVeo;sxEtS8B!Zm7m-10XM?mq7I-pPH6Pqj^VE6e4M4oj7y1(y~%7Z1t0
zzp(ci%d4IyRQ|Bt{kvL|4`KVH;|iTT2RuaZnE3#N5-2=ICZHl%JCWpE1sj@40qfak
zQ*$039<PJ)UeIrC{!W<x>32I@8Z+_g^5%x!p&72ic52s18t$G{la0R+H-}ub^)uoh
zqLROq{poaV$LJiwzEPXctYvXb2E01Vk-EJn)nOiCua8HuGtZl<Blir~G<aG;IYlDW
z{M0cO>JkHM^r$_EhIqu3_S*hjJfXhHqZ$kf33Ha*no(CD;~Q+S{8=KjP|cdR<2Y}N
zYCfGS-D$CcIKP+q@zw3H?$z!jQzIAeS=(kPBLy{Os|NXGx9NmwNUmboDZi*BKC;>|
zkW_O}P>-PP%YT;jF3mFJjoqwIOHB8ch_*ylj(?wB&gera<P#z)aQW}8py`RJ8li<r
zHfgwg6Fm<+g1l+?sD(Ifh3DbwZ~o;-D3}-z<)oAeI2H+;h=-cQAH~HO{;W{N+aFUf
zJ8(}6G~;>krXlzDL^9Q>`f~o1>Q~ug@3|_ue;3F1@*sSicKZf-YE_Db(gZc;a_`$2
zN#KEzRFhcT!oIF#sPx{LJxl8ge)&U|Xjgm}Rf>l3KJi6rh0xq(-|kz%dRuIC-|ad2
z%*hyZ{C{8Hzw-CGiLXl;Cfxuqj_tB~rlGB^{WUNo6^N#rv>smAu*dNH#_>_9$on(C
zpt5BfvsxW+sTN2yuT!++B>nq!B;Q7(<`|%<<#XH@wg25J+M~GOi0;3x3%)_i5uvg{
z1*N7;od16OnCJ)IhPSu(|M|}2VD{XQbUL&OOa|irK2sxgbK}1&ctqTvME~#GOmhmj
zsJ~L$fqk6JuXV8<snsuOiH@N=M?;9-Y}oQ_B(B7g@{r+3C;;DvPbTbOLOzxM*DHzE
zVE=bf?GNM?XqNYQ;+mzKpNqZz5Q~!zPKh>G-S?W{j`hEnt0xX0Cwx2|@0b-={DSX%
zG^0>g!eER5QKEb*<5Hr!Fv6&+6k~|1{BKz+ZeCzQQ@RU*)|sPO8E32Zz4ZDhM6a+~
zsV5jYOjRbr>@IX~WX(#*lg`Qs=;P3}h(KO^TKMocnnu(1^dLl|eYVu4cw~QYId0HS
z>FQRPc{aoUMcrFRRrz)8!XSbQA}ZZVN_Urvf^>H&2*{?pxd~Be=`QIOaMPh6-QC@t
zn~t;4=l7iV8{<6VyyuHQzCB<J2IJoAUTdzo=Dg-L=UNMq_w;qlmHpeKAFC0ET?*!x
zhC$Fg4{5J?a<DR@Ss{sz<j;tlFu65QJkh_#Cl#0+*Cms0mtlowI()S#(BYhnOqUPm
zWK}$?@v+mW2*aXX-*~q}ckq7Pm*rpI#RapNjU*y+oFTg{^_Bm##Db<|YR1XWSDp$k
z#g&8ARsF%#rbi3OlCs84_!vlk@K+5U7;VMaTt)5p(S&ABm4-{1-=gWW%QiHLYZI!W
zgAp$kfgh%Qndo`ge>*4wUSd^RPjej#%@tOi);e~y&r?pR$%XycADEd^C;anNHtenw
z^3NbMKIidLGbxWZ6Xh<ach-r32<!lzQxeucrS|0S`?D4QJo&hcJdBB!1v20&#b7B$
zlSph>wBxmZ-zZ!@bmvctxluN54g_KE;hK?=kxOgr?My5zJp<IrM6yek`iA9WOLpH=
zhnhO)tZ;P8#>|Z`tA8lY%Pr}B*|)7tc@scIhfse!Q%W{#wG3T4Z?agR_Mp%~*IeJe
zC1VNB(3g8BV*uKcJ^TLUkirsg&*<&Q9j!NEjTUNT@e8UZl`GEI(6Xu55G-=aqG&Zt
zvdgG65mAMjvi@|Ii5Adl2?-=}(Fzva#PwgU;v3&ZH!3)L+-)o+ai}JtKR<ZNbAGtt
zJ%(v{Ay<Ud-POMAI_%@Lsqhn8vuafh7P*Lcoz$h8y3i<O#_gf^Q{<%1H$irc^uZXz
ziScbgh8@A3>z@1a1*&=XKeZoxeqKE#`x8Ui`gY9{&clZ(U%uSEdGnP7auRcDzg2tA
zX1-Z$r_m|VX;ejw#zB6`(o($yM?4dNi-u&BH4Z@+7+I4K(P=a=S`MM(71yJe9bH|(
zKQrQCsojz*nb`8}K}{<v>ZGyt1^YgC^gy?`xT|%BC$&${&x^f}DWVg@l?bAC-2?JB
z%>cO->*ma&5J}i`a3uoDyLXl9KoL*my5pOGCJJ&u#okgOoS8$K6&oUcy{tDH#awwN
zzHd%=YQP<AXZ_@l{vpY$0Bhk%?EC5yLy3wz-BY#kNQEn};3Y``4xGohshV9IRYlML
zsDOONL-1t0eMPa>l`l(31oV{!YbAv1ggxgmV&|{HiNv_kW{oT@SNeQeE)yC5!-KsN
ztnZ5vhbt1NpHXu++%G5fp8RPon;p8*yskhaJxmfP*)34G9~@SG@w-2{hFyZFUX7iA
zbGGYU$?a0>zqaNX;f-1F1crb8Y-Fcx#^7?R_B3@%!nBPxeOV&>!5afdiOL5EV%zXI
zgbC!8``7GEw~q{<B2`q>erIVPpvEa#)-VKCIE)c5AW#fSGevQzP&m*tPoHS1%~^Ez
z>u(5GZ0WtYhN<s|@EJyOXG)kV0Ze`Cf-&xF5wbG~vFI1Z-+8DV79Y<*cU|cj5ItqV
z;>LUu=Vr(qE2RDCW?%_HHwLITGEx7B?Ru7rYcu2J&Wc`SG5>4LM8!p3C_tmpVl`{U
zqFx1pH#7>$)BTaguGsAwB>7(y-acI0b?=B=j}tYkOag|trqc|vWZUNK=yNqcj`P5>
zg-Q8aXByg{|M`+UH~5m3m%22K&KRATnuGtI-QU(y{NILi{doilX;%?E^!GEH0VMwS
z&l?{B{jZ$Fe>Lr3Ab&qKI6(G)KTdT2c2)nm<^Nv8H~RN~*7B)K6_Qi)ANTb#V=h8z
zOk<2)i*cWeNJ!^j4pJVL|34kx|7I)@(*KKt`9B`Z{~zW5zs42+8!`V!Dc`LBze@F=
z5X=AFmj8PV|FwU*{{J7V`G3?5=>O>g`j0LD_Zt2?ek3H8|03oab6KGzDd`U`jLZhy
zg~|ERmZYq#>W2^6<7MV~AYqsXDCs2TKc)`6e=@WE_Dx=`>U_I8Ksz}mCL1K7Vc;V1
zb&yv8W%?jjzUqbk&otNOO8+!vB^U&RE6C77S|G!!s;(XbwrBxdal@Ddk&Ebp%g2X7
zf-Os6yP|Pm=^s<lkt>ZgJv;kW=<;yjrftBY(4_S}^ZE4z5D|cvT?5<aqN#2&O%Byl
zjYI!@Z3Dc~6&OvsuT*@AvKd)ft`+~lWGiUuHTEt)NpM34UO<yi<X=@p@^n>>C97E}
zHE01q803GJc6Oj{1K|GI+1dSTyVmURuC6<KM<;lf4(;wmernw78Hs3FnqL-N=-9!T
zp~bIm7;#OygAeDZKa-{H`5d70Ys0&7E~4jksr0Rt#*JH45;f7;``yP^Gmvw0ykC11
zeu4(EXYSg!yH4OG;K~FR2_it(I)~_0+GHewoy4|e;jytfpnj=*aOGZ9zL{LK{Mw>M
zL3^o76}mDWKn8Nro{|*tBT@-Q<zo@%rOs9Nm2Xw@@g*@MN8S~oeIfynReMLrfgNG?
z^GZXfYrTGNHaT!LwIRepk>a1XZ;Ayt$r~PvnwEA0lpd>r+@qk*<9~^x;&r6%uMInX
zE{_W*Q0kQfc)BWZIJ}g-E>H=6kN}<EvbMGc@CI<P+w=(UgE?DYV{2X#_LGy=*|nE7
zo?R9rxmyBoyAHgjWWkIdn`T4zF_adv-yn9NL4}}i6hb=sm3#@^4Z@hXdv0%f+K0Un
z#5m=~MUTV6K|G2>_^8kUlQ>r|i@^5|YNfMSMT7NO<*dPM<ea|R$s;Krua(Bv-E?hh
zR=ch^-DinjD=1DZ*4Df;=TP;O?umd;--mqVB1}NavF{4pm(Nfc^Bh^TTL&53p?$Cq
zW++1nKZs1|nC%(wKg9%9gDfA|*x1g%wnj-39&T=Me)SbPCZ@e{8-B!8&tJ&P7nvBz
z6y8Rqm)1A1q^!xOns;@4P~>8ICQY50iRaq6DwmZl6|n4LK038Nm3oO<*lRy5m}lmE
zN?y+$FU<~=)#`AO4yxtf(sCEJtlKwhf;ih(7t*v`5QC%DJBRCEYv|jArZRnk_*g6)
zs%v*EXE#hE#@-)Uz22pYpJFz(C<HZ8gkNl!VPkxJzCuiwC$(qaq9YTT08<p}mbW=!
z`i4CT9Nnm@8U<2eR1(O@$Tz64bWA(vmNC+{8Z~qa_ajsdHq0o1WH%SsQls06gulse
zqwLdC_l@*(Nb8dvM>`>%sS2;SkDuv&oAvMIn-$3Mrjmn$%+^T-?3V_5Ow_L#5Phwm
z4fp0Fx3_mM_i>4ij$GS2+Z4W!<r8ahsqU=0<l;XDd7qO&X(m3qS0X#pS(bRr!}%E<
zT-JkkF`*48BTMRUc%hrpj#^{<sTxS$1?15u%h0hZ2nuFO)zpL$`HJiuLkv}g&Btpw
z++4Z1eA~^-GKV%jg>jVz<w=?q(n{AcMYS!#Ya=7NQOT$C+I!q4-P%EMC)ZRGlmd>u
zF@?qcyDU}hE5XA|5yK<7$5+7vT6x3kke8!KZzy17W&|v%d#O?WXgrHjMiiVsQMMLr
ze$y;@?+0l6B79vZFDrbjUwex<m@a#`8Sjl??^hn2EiHRBeX2%vyN@>2UlOZ2?H({b
zlwX(;U$&ZxB}{*DccJP1^f<2&_nLu5pqLD~UC{LQK-#EP`SUMh`S`Bdxlkn!mOPME
zbX`K`T0pe!Pyau5(vv5b9oh0lMK##K12>;g1-rtjCsDNz7hA&Z7Z)XN-Zi~GAlj^W
zR9uKnl1~0vyQnqgxjX!As)tr-nKwg$c8QIx{dM(D$DG|dq=xA8%nXupw*a$oHh*@-
z{>J5`F(e`)N!J?k{c9zQ)C@H_!NyEwsp#b*2XKNy0NBr{IA2Weagqx1DUPX<C^h|2
zmv;`gzBT18?-bp;bd7dqh#i0N!sXh(!+s;Ae!Y@TZ*_gN8^5c&;QIU&zbRR=A5(^S
z?P!>(-R+7ii1@>Y>fyL5lf3C9HmQYeVpU`n@M_?HJabdp^Jj&^2i770lHRKi*jt!?
zuHk;ULLt$=prenY*iBZ2633-l9XAAbzD715FP}+VQi17pUn_+j&TL0VxCX9N*@wR9
zAHl$D`vvX^x%oVM5vlPD*eM;5Eod1qL{e!kf{H?^FfFFjCk+UwFu`@h@v;enlTbY<
z5?Bzk$SvIGcbflLM#h_+`RE-?E6t^{C5M<Kof%GvaCGzL-k!>ep~<j3_|rHNGntAF
zDhtvGZ-VR=zl-vtD-ReB-g>`!^k8rV)t2z%o#1}@+9~m1$uRvi3LUb%+2RR%7wJFW
z^@J#AQRQFRxO_5msHQ-HP_)VNhS%7DrPCqnASRKM4W;4{a_|aEH;|BE4*46v<kA4V
z{ra{#V_15pV0?#xq2>XVgqH_ibjv;(vIlC2P7?UvvIe!pi>s^u>vvAHI?=Jv`-=mm
z4JWiQoR2foCBk58dCNEb<Imbzse>Yvqg%%HPDWo>2>z8^b{s%XIC<g|+$G~%nl`In
zq!uoUcRZ<bjH|gVvUB%MXDULyFD`#<TGdG9Ic>>iEpLfXrG<o4mBH4U|9bi7DldOx
zq^P4XS1X4dUe9w!y9Ry?l)kJmEn=hzKeJz7E$pY~3TR8uGcbT64NBKKx+Xe=x;IIy
z?db)ygvQGhoaDa}GyElvRPG_Oc<N8}=Ye!<e(KCIX>ZNtoer+%1F37pvaZa}h4reb
zSv>cv2?Xss9!m6YFsN<3S9OT(kt@|i%$KNyr5jt+wiUqtuBzM{60rE9)2P<^x;8A`
z)#C~htqxagx4QzpT-#upP@&{h?wNG4eI220(!^<!UtP4B_dDNgYJZ`qy?6|)prpoB
z*Ol&ps|aPC0cuq>yTF{El7HJSznRx>@hm&33O*L{HnN4x$=pf-56Faf__#ldvM^st
zg|Mhogcm84>N?2|iHb5mhevjNgv7}d&zpmO%)hBAyl^NB&NNOqP6*c+tMXWui>o%S
zj~fi`Uy-|9xutwl#t3315CMDIx@>rzbIyO-h=@o`%gegX)`BinI<UwiyXJT|Dk3Je
zWhSab*_1NZx>QVYH+PE0ScHo#%r%{}ZF<lT4+dC@YSWpS;!L}W7hurAn%db;f8=;%
zg!ba-C0&k-`w|iS>`;nH*y&XP@vDt;>mqsz-cgjuzoyZ0eq-$xlCnjeSeh`w9^`M{
zCv89@HxyERwBDlQiVI`Qd+{OLo5_gU$k4>n2KH`sBF&swhKTDW)5gew6(R|jkgU+x
zGGqWF8OjLpIMaYEi!`&!)e4lhQl1Q9-V1x7=q8r=AS**EEKpdO$YX2B<-p4ojA3VN
z{;!zTo&Yh^J05MNy_jMiSUJF3&5DFf#+I4JohXWTk6=%KRb@)$TkH&Iso8Bl@%c6T
z3&wyYQmSEzSuaP4rd!lje^tl_T^1DFc|b3`kPJm(c7c?eeY8WL`^cf%L6yCbI@2FX
z@7@H~l1#lla+DSovNyjDpGEz0awI!C6$^G*eQ)1>xSk|x*A?etxH@(7+uqeNP@GQ1
zM<V<B^{dty{UbZ)KMnl}Y@XqRB|_2I2*$s?rdSi)^j^_N6E3LpH~){6eG4qQ*;`ip
ziJJAsMIKihmIYI_S}kE=H}D7dQR;&r;x%xB_G(xZ(_$n2Dn)-u$>FbK5tE~mC4QIn
zh!Pu{B~wm%D%R)4U2!IhDa3=-?gDbr8gz-$G6O77(Xe*~=8bt%?B~FF^f(_44$IM7
z39hP=GKK)6a<2MgnmjQ;A5y#vMoVPm<dVLrOgQ3Hnh(@no*8WM*Yde%j`LWgTOU7A
zsqZEgk~ds3fR7!{x{|NQ`vw?_)OzW8P1Xn+E-!xQ*308uDN@N*OZz13y3dAK>zx)I
zQe|!<By?GrAUs>mN{)<=R$Dw;(kffu5P|UNfrLY5c;)3L?I?-QZzlN<VeN@NZ^w^V
z=b2f{V9AmQ(^L-V-71LB1At@*>t580{yu;wgzR3;SWM(#x*o5^t?2d|_}p0!I{(}@
z^5JlkbT`GPMQ$}LG-T~PERHHiukpjw=r=!Lj&TVIQo!};TXGTu<mXt>YF?Ls9WR^0
zWwNvGb*1&!X}4KjEVUgC;^Xasl;Wb*UcXnTr)RJa)v8s65|r><vQ$=)fUpwVAJlsh
z%bU|#u@jDKLnbpO%VA4ra~apU8dLm9%;Yx{FaL+mc|n$VSWMFun}-x$G`;0oxec38
zY=+=(vq3jvG`S#7dTy~&ht1(u=YYejW07@r#jrXLqr=yg)>A;4;C0v65eD#FjaL=I
zwae%_iURxj1{4>U$}CjsLK`D}F@-vZ!?|!5(;>}Tjty68<M*rKE#l7EE<OGewc@qO
zZep=~d_<%9#nZLlvS8PFsG=W}lL4ADd0X1-l6p=}1o2shF7Jq>TZ9Ow8t#Os1-RG?
zDGS|nKSj_@PzMQQ=8~_@s&57WySmbJYmp%uFH{M&I+jM0;DLY0U2>etuln^OuFffE
z`ov<{m!TreikD=aeV%Pey<lCnyew+kUI{!t<K0P9J6W1nM`#5PMDxkK5|&f)IA6F0
zsc1VVypiaeEiR(xpcR+X8P}M3P?Pzl;@lGRrRd?pS7J`V%z`iucFxtN=*YNOXD|9c
z9DfDMAIiLKkW&cJX3S1>KTdcT2G7%|9bDnQhc&QCTohhkRHS=)Moh}<QFDu-z)knU
zCN7woI_JfQFGdflRNvuA<@AtFLN$KVn**Y<5+PSR*ke~i<>k4==S!TVe9nqI<z|`D
zE)FUtBOb(<Xtj<_OY1YwG^<0JH?{~%>3_)H2+7m+hB%F{71(giFgD(pJ}MZh)g5z<
zD=X3v4QXy}g_W+sO=eV{`$mR0&h@&MLSLzZb|#bnU@#<Gm<!$-gxE?lOKfeqjP{is
zqt#77XSI_rk3PR-G^%xc-J|%q8Wl)Oczs06C{2Of?ON&Gj={})pD*3yQ!ExiOy<6)
zDABJJ?ZAA@8)EMlI`(1kxI>M;$H-RAIdYsOZ{%4EOl4zz$j&_c(-Ki*%iCvelg5l1
z>)reG@SW%)C}L`^_i}3@JFXt8Y|h4ioIjz(qgFPQTL~38*!un%qd)NGNESa9b{~-n
zTEAq2!?JbXSXH*%D^(vX)g{d^GJ+w3bu99~@5k*=2uj?5zDQ@(_1Ho)bv-=2w0Svd
z9ZqwiEA>)Y#Vdmqe`FbAsW1<{(&42z&^o|>_-Hp%+2{4~BB^o<Hf-gb?>$|69zt6o
z7{&8!m1nZ-zED5__6PPH3H2CIJag1UZYypA0@8!-p_am3zGy3JgvTG9uoS!LryV;}
zy`u@jv*bABvV0x+lpHQ6UCDcvnrUPt8I~Je+VnFPoveCFeBlOfB|y?r;mJh`UNqrS
z3pj!nS*ySPFwMG+H<YneGBzC@SV*+jX=@JE-zX>sGvA?b1?DdeJ3m%Z<T9V5*y6W&
z)wV{MtCjXiKb~h`mTYQqW5FHb0VIChev8{>nM=?E+9YMqv3weP(~aU^K7(cNeAt6<
z$#AG7B-@JdTp`z&VwuC>io};jnIy7V6euV|g+>Vd3BL5j#pMt}66H*y8kOF)w>;E5
zBTL1g<7XD?5?h!Jw@~a^Ih6Rgt1sMwmdLVZtud<KB-!3fmpm?D=ykh6!SYcFuFLf~
zIpR1t5fm4^4mQEn(Q7QK)x7yNn%RdYAT~?LD=GfWH<*?>Z<R5uuSJ|qhG?TBD_Oa;
z%D!Ew$G2oh-Mi4^sD2zD%5S~YC^oOIi#g763lf_eD-RL&*nC77w1U;WKGb^s<vdtP
z!gqu(K~`p_Oe%by8`f1kFLE|rXk?GM)2n*?!<2ep?W@~L&SGhE{`$IdQK9)AERz-}
zhbr2oO>@g(OGYA;imOO%L<eAd1Z=qX7dGt=4myb~ox5<1q;mWY+du4X?yXz8KLc*`
zu)OvtQ*~#QPQw2_x^glIGszz_>XG}*bFjlVcNNS9oZf`UoT-yv?S^LtIZGCI#YdVX
zDVdn8^RJ7HJ2KYeo29g+*I`2IITB7W0I(#sb}DK&+$#u{>mApr9y+_a9;j7A477v<
z*`2#PM8C7F&nm*d&>+|>%hS*SIA8g_e2R@3H|ZORf#f?_40HnG$vyE!AhhlO<~Th_
z%@mhxJIRenoGtbt`-<O1T4Ib(vc0V@(bi++(}G`EI{&X1>!dnz2uRfX{nr=b{Tx3I
zL8tkx6+=3;(y>M?dPjJ)GmIY1ogK_RJ&o;JAI~*x5+;cJT~O-+yQ<>7tUC5gs`_DD
z*`nb=rAs(0sDg!Qy@o|Joh~Qs$~04=Qok)UtN8Q0Bk3|My^f7^QsochmEx^Kvg77O
z|6!|es_N>rSW+62Q2}ukVtpxk%^wZ1WWb8U$IU$VQCIye8jplGT~|)9fF;T0rKpPo
z%aEYJ19H{qM<H|!JZfAhD2Fo^(W~2KOontuTpN9d_iOh?h}_Gp^Mud0u*|3H^*UyV
zS&}7(%Tf=%h{NCY=t}f|#gjq2JeC?{q^N1NcF0ni*_njW@T0FX-05=GuSUb;E|Od9
z<m&6FgUt{3`ck%|*TVHY29z+OO}3LRCEL1GR<bo+GK|#_{)Jf(<eLzt+CsZ}O3iRH
z;fOeHrE)kd^}cBmntio2GF|}jj{A~*3$FX&>^b(2^;Nb4Fb&bee)J=}I|aevR63+u
zS6@e=iOJF0c-^aYV)q)4QZf5&`9j?q2|m6v{R4+ng8Nme8yUFr)HmxYg+jS-5cPQu
ziL77_GD}VR??L6Mnbe@D8T-;Ru0UpPkQxpS<w6{Ep&T0RY`2$=*&W9(z+vaX2i@VJ
z;b|RWLaF(a+V%z58TO49BPgWfIK<nMr%;{}?cqj2vL#W9e8B5G$IX7GqT0EBcl+|E
zEcKa~u1D==oUnEHCAwM%Udsp8)l57YxXY?JQSXmpU<<dQV|VjuXg0z2Ci*!g3W_T;
z&r-Cp<;u0a!&GPR#RVUGTQclLKQ)|^Yp3j&OL0p^dwZ!#JMgIKD?1fKZ~<uZwVS(}
zooljgSb7USqe5_(?lPOlC|vE}m^`fQ!{EnItm^WglKJ*NA27G4W$qEiy4W|ZvnTIF
z9B$6zN_KS(ea}17;j9g~0d6Vu0G!tZf~{8H%V&mem=g05N$fL3R3%RBJIpY$prAf2
zcb6NdIIK1!?_jY<>;>Ug%mF>@XjeRs<KjGCM$UaStCvf3&Yhjb$20XLyiZ(&Q7<Q1
z^-3l8fPQ?RNeeOHf1z1dQCDwa>b~uOfeLViJI5{zVTr?(HIPFV5+b~^*T-d=YowkH
z1SICLol<^w_-ts|7EQvSs(HPMEV_Noj-HcMzTn$NbZ#@0tKXmf2z~!};B3K<=xXBD
z5+6y1up<bKs&=BC!_sLj$-X@isrRQZRdl(mOGUqTx7Rgq;jn4h+>rmck!u5SnrWW<
zh$t6-rz%Y!{ZGiMd>5Fgj*DW4L_EvMbigK@&9>_{72Xw!ev~-AF;0Vd!{Epy9N7A`
zfVqR7ca)h$wzY@@*i+R>ep;9AkgmJlrxJF!`x(1K-LQ#Zf8W$b(uFm%0FBF3JJU7@
z;DP$WdH@0=LK6!djXd<`i%X*m<S%kAx(b^S>#EEA$!kP;9UCJmFr!T2b9dd?o?{~q
zqf)#eNJ)&Gylb>hYNYE#LVSRu+c_9&TpLu=D%<e4G-6=r$?TXYA7T*Qsd|p9MkWIj
zU)x|y12$px*UZ)6fs`u!gpt4sAFwt;{7`X{?hkDmy)>h`-F%o@K*e+1KLO&TmnZ%%
zd-(DNg@jA-c?NZvBrm>Cx*ALU=ApVRsLKuz+K%nXTrQ_t?G+1Zc-aA|Y&4QiFwj4f
zZwXhD9D4kXf>chn55PJAy4Nc^sv=AStwf~emh~*9lNIKz!#$F~#*v(rj10}nbIa>1
zA{{gRIL=}3^5(W4v;Htq@xk?Q7DO_t6dVN9He09W#0g=*BXHUesM=q{K9QZ+d)aI`
zAVRj%Pc3Vf#{_1f7bA9lLm@2nDQ_Vt=-M@L-0duK`cyzxK#sb@<IHtx+YTl*PZxA<
z8l>sA?<=fZuQBw}8*yHtHdHHPiZ%0R(9boB+lXdg1Ii6<)hLD<D^pcxl4cV!P@?&f
zUu;+zxD_zb&$o!U7;QS+3qb82xam$|+ZniSBB=jc4o&`43fL2?Fe7v<UD;wYv??bY
zPK7X4*8@hjU8vi<AZ{b}bw^}a($nfG=7su~N9l)lH&NOU%7wjQ<El*$HnED73}wTQ
zt=Xqr0#PYq3ZY8?+aNwnXu8%X<i0S_8;&Eb=A^SoqiHgah#>!AyflX1s4Ic~UHuT+
z@2x%@o%Kqhuk5b?f%DbXs=E+}bMEp5T(XBrBXuPz|7+h{`RwG^bbOUI345OtZ}{nr
zV@wqg`6e;^@hb8bw`)A{$DB{T*yO7sObK)fhpLU&Rq6}pAIpBYOsai-OBZY$JIrSU
zl3B-D(_9_M^U#q|uYM7Kv;{eukf;@mCy}ck`xkEX^TKbT4CcCgEK8K`*8>5l#$JWt
z^$Splp`%+A97t{EqNZ89J~FhOj$=Mref$l>i8@EsDMr;}w{Uo8EN`SP7ErI;3``74
zZG4qrV?araG)z%IvixW}qp$ov-3V+g_a;C)^VmGNiL>kU>Y$eg!v!OA*tCERO^ti^
z&ObOm<0_GVy_sxku<FI>^}|1`!>-^B>iC_^Pc4y)O|rPJj%q+6CCO|v(Km1~_y9dD
zB;cvxFTpm#mF+amK1FtR>dKws2?urw$Z|%5YE~#JVsY4czlHIPt19L<V3jmj1UJ^$
zYZRr!n<0Obe)XdJwG#swH%<nL{%%L2Cr0a>UDW778Rz}1M%Ao#5iu|CNB$j_?)QZ)
zl3P@`KR3dWWoN);bXPCIeQL=?p~TB{1OvWoAceCLqGPgR)8eCr4mhV^47=pqI1B{W
zABtC-7i7ihdx2H%QZG@qRVx@S^Z9HC3G!(eg-a+rnydO@C^cm@Mona;p%&{dbzDGz
z8(Lp|*=IMR)`%CVz`8tk@Ui0wov@J7gRFrJl^WE8k9Q)?UvGeDV-=jyIrvHFJi;6f
zj4Pf8W4+;}OV#!ImCJ6Gsicsu`rDMS{(k%4K_OGrO!4@YOzeW%VRlqZQ|1TID;0(_
zE_%PNx!#R|x`9J#DfqRz_r4i7A2o2lz%Gwu6lkuFO;{=%4;NZy;+DIH7%}T)I&Tbb
z^zLCxcDKr_%clTlFo!{)`NNx@ZGZFW&LwA6^O@9<O!utFi82Nu0jUR{mCz*W>gr(f
zc~PGZHw&+)owai6LzrnGOK2u_tV4D6y3{ecPeDfqrRACf+YvwPp*YIzH&%|0=3ev}
z?(0kl#dM2^s^<aaO;b}oVsCmYsyE&*Ub5a}`PxPmd)vg-!xSO4Eo@_$86dQ>Z9A*@
z%r900CnDml={xJZSB;CpwRaB>+<OObZohc=_lM}kg{k_(Z7*;SX+qDHZqIy>*0X|a
zB+fim(oX3d<=ih)Jx`d0eCmTqnp%c@<&xYIB-~ww#w?|<9Hj#Tc@q=FDqs5fd0p8%
zh0stT_ex0Fh|@>vn;40`GoiyD;J9n?>T-lqW#RVOu%%MPfLh^xD{XRkLZthKr5HP%
zL#7c<gql1}G&T__)Oav+{ksM??-UV<PdJk&o_OhCIS@Dc!!DSHYVRe;a3tRUs;k{F
zl%P%Z{CJp)C&hGS!B?7oNnH_L60W(!Ujw_k9=FdozO@}MqjmDO2?lN)(Sk}j*&8~0
zJvaN=(WSs)&vTE@S+psBP6RMvEYEGtPtI1p&dS_dwyZCOE~Q%w>IvK4SE$E?tEtMq
zG9!6ZZ0%NS&CytWdvA$DZlFDF4jvL-QbHhdE!tq9Z&_dzB}v5_AFrU9lk$dn#Yhr8
z`3fFv$!%^loffh?{G$z%Y-MIuQlG-|0-UeStX+fP!A9v4YG-1gWaR(!eU(61uiVR8
zsQRF3GwUiGvQI+&XHu>(UwNxNj8linGGpU={65{_VU3V<&NL9n{IQ>v7tioXCH^_C
zaw{J}ax3{<D2czE{&-9B-pj=gb#VurX#CuA930$6@>$OvJ|!4TbP;5h4O`m3S9W-i
zN~qz)Cooj?OR>n4dsEoaYbbQ&P<A(p4#%tt!o&Q@fZaSocs&(m-g#+HNDn8XwdK&=
zBxxmp$TdezRdMpXetW-kX!^a^=xe2afDyCbM_a?`!Kl|0TTr^gVkFWuE-KFZF!_y(
z^*v_1WtYT(@$3zxQBBP!12#owOpL2j%${j;k2Zt+jMI}HgJRBjqvK?yb&<_$1@~-*
zp~R0WIm&@oZ=arbn}#{RDC`!TC?*wkud(9f0AN=YJ1Vp(YNPMMjn2EZ!f9NP>vIr!
zO-pOxFGSKv9R|MhVt5bTGrEjeUsM(}`?o*pvFggW%eNJ@J&*;`7EZ2S61aBj%>G__
zqgszZORIJ)IOe^>;>&jA<EyN!st<en6Dmy{=Sa;UPt;DZpzgA5Zy&{v|H+aX%@-Zl
z?b@Mc^Mo_vGTGzfp-9c#_^^_((`G;s9UV){z_>p?zvJ78WFB4aVemdX3*iaN=KaPd
z*0Lo2#=h@&uRN~J(d^exQnFa?dO!L5!`TZR>;3e$!dk?|=hzJ@2Q)Am^sA6rJ4q;f
z2kkg5`twA=BC{vF_z||&S+Qw5Z<!2{Pom#F5wnS3rBZc8rD8h=24X1RSKr@VLO1kT
z$wr$NaM2tX8D%SnHBWDqW+lI(2;CKT!r`$!AQpV^6AQmVf`#taP<JqMfgNocG*uZt
z0%`b57A%j+ZdIOjGET1!L$OgRZ92XA-bX*m=AlIXhev~HojX&QgS%VH<~3Iviiwv3
z6XUb*4P<;@wtdd(Ub7djG+c6XKDvc;U?iw$I(c9qj?8-fn~)xnjhm;K=e!fMOjgV4
zV3)WjIuqMccYx0p)uzs^|B%*5_MO=C4XrjEq}W$4jqz!P*C%{TZ6Pv)k45;A=%@=T
zEc5S`m1~#npK%7V9pEX-E4Jt9@IOLyO6zX*->c7EWd5)n8WP0Cf{JQ$x)xhA#`I;r
z#D<3P;>*-Z_lRl20gnaiel1pZcK-YCj<$te&Vh302&&Pc(_q}bqs&af{xaMaBWa_i
z?9><{0xM2NZx7p<+!2WPT5O`{BkO6S+%r0sl_|1cuLOmB@O^IW6-sV<FO!{4jP1IN
z$iC`IY6^JawHO`2x_*gDf*IoUbpZe0*7Wen;~MYYSA#Gv4v};Jc)FbXh`q(1hxeQ`
zw1&nkvO{QTjb(ndO$OqgW9@fzPw_9@u}6RZw&l70K04&&F*Ao|xeQ;Nq&{kHURYcF
zZJTCV^lqZLl+Rt@hiIxIoBM=?|579-(aL*8bPPF|?%eg-(u#evZg9X#6FKw((`~$n
zRtgQ#G}$X}OiURY=vZ}!RE-cIqbEobuU;sZx^%K4)lJj{ks5ceo0Xc(2~z#snG4OY
zc1fFFxdp|ImvNNsP}kSoQlg~PCE>{&9<^4nkEATMaa!2*LkmLL-OI5}kd}Nw6S@Dp
z_ZPnbe!*2Z<-55XA?}9W>%99xXC>1t*eKI1hSi^Ty!>^BD#zxQGfq_0x><NkY}1br
zVbL7Lz00pxH~q}CAFSbG`&yhW=jcH!d>5_>bp^`RQ_(QV7{}XGCcLn$2~f<K4=pPg
zWw+++UN+*Nc;jheu4#4BC!tP2Xv4owZ^@!nSlU(J7qp0D<bH(<juTs={Ji_3$bw+Q
zi4y52F#0Ta4vqFwgr7x7X1blwwT+GOPyBlc-o?TXoyGk~Q^K)G;@YG4a|9ZWQ%U>G
zl`UKdRExMd<Q07=tK}*EzQ5_rU1^HY?-xNMR`5~-^FglVO};n2CEJlTu-SmG7&We&
zIL)6v<;NdCpb-<}pD@$oB`3zfpeVH=sI$OWYobQ*C)XD51<w4y_LXcJ^l&Gbk~z*h
z1t}*~`bhs+<O1gmsVtWu1;|{H0h-<!!{uT<kGHgMZqCY>k^CY8-#}tu(aRYj@U`Jn
zD%+g5QjyW-Frf7BXg3YIK4@iIL`Qw>_fe1e^=|tw1B7N9Sa}*w6dyz6l11r5Q<QT~
z=65jR&CyY6yK6HW$m;z1@;;I#tyGWei2Fmxw!5B1cQ~(aQ1dUWKn1;m(LMICp3R@0
zE+Htac{-+Gx%oBpD^8yqWOR@6N!4Yg*Cf1EGxi|kM)E<6&|F>JZ1q<VMOH6yHcgb+
zfa1%enX)SK6L_lZja(+;h>_O!>t#t}NC}=>C_KVYEvyvh_@B6A8}*O)_vgmzkxugU
zSn{DDds<oe+IXk0#+GKdo?^qXR{2Jtxshc{FxJX~zG+8phuT`%yrsir;Om@>8XqFw
z__*%gdY*${dmJ$YM0+dMmX8J->bv=3V+pYLr+8xGW2}Uke7ZZX7e5EgMX1vMXl}vB
z+3!s937o0$p{Ps%jz{CXJD$JgZcy-?<V%PGPwbQ1aJ_M2o*djcP0LuYprDoc+4~_c
zQB4D|pvepRCeu`p%AP6H0|ofK$W;BMv%POztWbI1c#zBc>Lh%MZSM_L+<AnOa%caP
zt7&47l*q;Xp9Wi+2F9=kmS~FAN*L+6KDPXjO^7U2MlZA_CM;x2hC$h=B{3+M3q7tF
zuQ}RKq{l@+b9G_OQJFeEu=u1%7jWkY`R&9SN5o7S-^A88rGb-!U4=~<8I92_SDzMS
zr+Q=jgs!eprwb~z$Jp>@(7h{RoBN(kK4*iKXUg#k9@F;^_Fkqc@9Ph_pRiPr;P)JH
zJmoC?K08IGbGUAxkW4ljLr}W@6M725(>SM3|M;l84Yd^98rUCczJ85u%ZTqg8CB<K
zpAp~j(7m_a>i>lSRP)DS*$q0O0z^r@_}MaMZ!J0@Fi?L#N@p}au775L6pKOXeuko=
zzdv7S#7q5mkC2!qC)Lll<LwR)3XT&lCbcEhtHX4QYP#X+@0ZCS{%+S6*}nS9J9TBl
z>V$-$8k)@+hxcD{lpuYTq`I&?d{gIsHO7CgS7U;K#DSt|U<uO{|KZTgPI#lV6XLS_
z6NN&ZM1tSNI)~D!3Td-?r`x=2L*!O`-9MiS3Hp=r7>rqSL`5By*_kf;2UHRoqNCnJ
zeN7_!=36%Xnh7HmZmqwJLmwgOQl#4xrIh7tKVLz#5Ye^-;PaF4Hz#24()2c2XXlnH
zUvkv1(Vz)ycAhlI)^W0c@#~YP85J=Lest+LCaUJ;_TP;WG4D>sHA`vsiKuP(09ED0
zFEB)4YOEsRAMJyJsIoxv>JC0UKAe5?@^SsnSj&MX><Y3Fa7Tk{c7>CUzg20;-5HM|
zCSn)(PD-B0-vRK(hk+XBpqQaM4^owLR=1aPkDEPGPD!?cf{ZPPGHtI>wPay1Zjk}D
zl<s~)crh2-@}7gax{YweR#wk`|H*jA*Z*Squ>LGPD=O(ywmvI8mRi*mRZ|nl@^g6*
zSM<Vs8~wANedC(OHPt>S;&o?%iBDI+3TL3DOdECDG772)_NWu&u;C5CBK(+2)&^{y
zEk*QY!<hlClkM70`ummdJlu6+30~4-BdSaqh+*x5dlKfs;@W6$KNdOqtk_qF)=k&A
zFH$|S=2}6uT2&k+UASPiL)w85n+?16T9rxRjvMvW0EY`sJq23K$)KYG;{skKq0%NY
zY~Q9+Mtb|pwZ#E9QdKoiWB2dd0@&EWJqZoV02(^6C47zGEI>d$I*B)EnapHjeOOH2
zr#8M6A$Fg<=rVpr*L5uz9DV&-OqVPZ4eii$yU&wBW#FzJOE5%5RlN6GqTt?iW9poW
zbx^f9rCaUE>Y&C09rT5%H6>FLJ;y_*@W6qAG{Pt`u|*s_3bkjf{Ty?rayHL4yH~XI
zC@*U?fW$>p-%<Qx_($BkSbICyYDG^b??`CGMdmSWnGp|#-2U*+SQH>rI*l=V8mAXZ
z;f{94P+{QqbmZhWiiU3n5?x&{QkDBUyMxPglPdT|d0@IjnTK0l5p}g~EX?nw`V1!D
zNIWh7{$*KK_BQ;`_<i)-bj<8nbb0x2IU5uGb=D)@EyF}PuU=NxyxiDZWbA5ID4(HJ
z!@y`ribMWc#V|5P9R1wQ<LR>4r<1YKT}7)W!rYMKG)m4Z@A^b0jkcWh<;D4(;mzQM
zR5muQ=1?R6L)nTm$oUKnKN=f_t7>~o7BmkvBp}^!)n@E8?W}yQLe8V82qTed?R<&p
zq^ZS~JzdH@w2;0sENvw^AT$I194zrug9T?Qiy0)Niss;u6Pq@0lvX4~NDEjNi6M^V
zpsxuzm6=R&t7Lvb!~X_USA2gWq317#mJNn!nuE4G%5t%Ehzo|AyZghCNd-7v9@rNb
zvl@42Ka%N>)=te?F$LQ$O=-j5eCOt-DAIQIF6R%`o)C%~&!kqG7_>CVU3XQUTG_He
z$3Xth)~DP4b9yiWISHX<EoQ28$%unH#H*V4eK>3SU@EsKt1M_Cw`U+X)F76aQ2$Bf
z@R=EYbR_cIfmUN=5NlE!8C}vot~sTqjv;6zO57}}vp_{e7y2O+K-AU%Rxak7p-W0y
z{U~m38H#-RBGIbYd~*GVYf~yL!`#6{BglY|I>XKJX@u08#{0-9{b!N$uy@^7dX{9S
zGay1ytuc!FaAs9fTzQN=;v<=%Ke6@@<Zl8O^BIBjxBgL<8Tt7ROE9?%L&&E{3|2PI
zm%2Gms(<8*m9S?(*)cwQe!o}n!&Fp1fw+N8HR|x#uBGRwdxKJqzKx;ixoQWx{vA_>
z=jTH7*<Wp+35S~UIll)y0qm`3o-d!!q|Vyre4w(8@m$bfdBu&Hs!AD~5N4(3o?8v?
zPo7s+-ukLDd}7U>*IYbY;!nWY<0$FyfX{FV1jK8QkgSI@Dtct@u<=B5gZJ-F#*(<%
zgc3<9<PU4x(<u(Pk&z@A_0k`t@3+NjyrFtLNkY21xFVi>giJt8#O-2n?41V3?aQIY
z>0l^k?T3PgdJ@Wvwf@Ce+<xK2>vy|oS6)76C||%OFl6Ma3!c8HxqxQ3`SWu~SGws2
zq<U=aV1JM^{*YdfV`F$mmNskuaNFniGXCHftgcplM`=M8a^7|%7A)0ZBLCW)d-U+Q
z4vn!l)Q<%dNOmSJHbe{McP|`8N|tBe(MPruwS>IZULMh@w*Fo=AHTB9S_{p%AU58b
zzXl}7T&%{)IPCFK3$~KF75m45wj6jFWV*8Y^Sri-wRr~ro)w?eflK{sCStgA^`u9L
zCd$KbwVw^)Pt7XHWHKZ(;E4ggd_`!E?`2z#fF6Bih}7<AYZ~o+s3uTNv{yTE=lfCb
zBMNrnYkya#p-g+yOPIyv3e5PkGR4!>lrI<l{$d3#1s)dhBm14F6%`N3u%(=gAUx3;
zAp{xiHy8;EzHvcIt&aAWFF}+PlmUKBmxp9o>beqky2y|HZj-BR8B8%%r+mGkeoBN~
zO?BGLOknMsj~bm!$8k0b7!gS`wX<96FYRDNk76|rtt?d?eKeItY=@HZ{~;@G&&`2V
z(0~#h8r>5SB0Q(OhzlXzBH1FG+mOK!AkGgAYGY=i<fIHw?6W3$fVpDzfww1!tZg1t
zDZt;2*0O##c%ta{5rC)GNXeuz+b{NNg=uJvf%gfw7hYo>8v!~7=&Z!EXNzy$C}_hK
z`3R!Ftc28;n*Dw&Sisaq&y=S2Tr0oxP>*hGuzRNBR|?N`29&NHeu8f2@V+7h-@}1h
zod}=_#rBL`*9Z;y*3O=~#fdI0zN@ZidkZVBZ&Iu`Vf9fwTi`BY`PBWh2C+S+=dx;4
z;Qm2sVz>~97;%)9t+)iDK^L{V+l|(vrtTkEbm#Upj(UhHotK%KZW-czLUTlW+tQ^j
zOx_!x%nW@x*zQvTP|Peah;V!3^@tHO7=}`&{6jtYR<DXz-Belo!=d39cn-0c3*w>L
z%CADTyLXviXsX!?vpGHNaX%1xN2o!CgEODABH3EAj}GwS6Xa?4_f^9LEXKzA0|R5&
zq&<ga2nH5w9<DgkTf2+BPc;odCyY9B2F$9VM-$n%caCzcKcU_+5xP43l4>U4{%MQH
zK%+OSdYTw-J4EVvq4NCPs(KMP0=qlOV0cAOL0-|nSh^Xw!ZU8imahI=z4=rkaWS2#
zCscTr;|=7l$EC}OJTRe<JMTWo8$Zy#Z#dC4Dn{*8J~ikT{7J=K_un@A<7a55F_aYh
zs&u}egMjszDXgUAp4E?)4&>UkBDk=LRvqDBv`Tc*{?*>VMU4UeNO<_z-kkU6`A8F2
z1}QXf^%mfBOv=qL@1gn<yvgX?8jn@0w(HlVMRZxb(C%qlNKB(0BgO8)7C4@w7hr|!
zC7-*a!^iE91hMhlz9?pVD}@5E^&YJ1nBFpJdqNF;6yN*h$Dv(ejlDw{Vr$#96SI~r
zb8)tK@1nw;mbh?wQPP3=QS2C)p#zhdBx#d@Te0azaXSkhLSw|(J@MkQR!^8iL2YcP
z4np#3vJD#^ys$m{yvLI}!M~;<1~I{#@bx2*Ty<^Lo}zrp@W}WKv_>ZVUiE<`Dc>(3
zzO#IdkgQ0WcUludVQud@ZNJk}3w-VoW!@d=MTRF;GY9nOq$_4LG*6OjXtq;y60v_p
z%^Tid8pD9==2dM51ko~5Vo-{r&>XI(av4b<7uM_#u;lyIv<nD%TNg>XalZn5TL<Mg
zR(966+9BIO4IH5Dk2#@_)NF0WJUBiTU)5ofiD#%DoW^1xN$0iD@9Q~k&H3&?FAG!;
z-hVE0>n+V2#@x|;u>A&R&4wYDV7Roz5ExtjHCfNYE9(=M`7X(h@4Z@=eNf_eK}Oa#
zS(s?eIZzj-6aCTq<&}R`aBMMKT9tk6ozj0tchJLa-%sX_EO%u92@`02fFBwW+4Vch
zgCv<PBWkhmlb3QkfVnnC!^-2lqt?8f3D@q&lrZ?2aCenM09icphcGH`n&!<|{$SMF
z>b1$+c^S1#@if6k?~=E(y;f`763oq<Y+{ZgBaXGSUC_eu+yfOe^u?Z=J`mxTSXqBg
z3B?8)3&2+fJ9AFP#=nL}%k`+~j_`f@oJM(hYa&fK&#$#JvorwvX=x^Vt$BwgcGTmy
z6-D@2MlyymA^6s(s5os5d#20no<GNVV$XMrHIP7*(9qWz;S(Ya!lqyRrcr3T<D~}s
z>L4PJF0Kh<7~|t~bkNtV9}oYmh1V?#-OB$`(a`sCTq<~QS>J_bCJ9UQ)`jr#kmka4
zqA)Iol2YOOm3_ljU$l^LE!}R@J0U2$^P_C=LI?Yj;<*fKgB*E!XOEe85ttX9IfF1=
zL92*<NmO7p{o;LUhk0|9+liHEkXes=?-d}zHukqp<pGvr<zSm@q%~R#6mo*-BJk_4
z?}_F_07_#YsF`MX^-!FK`tHWA7b7{}?bufh*LCNifV{y>TBz-vYzrNu{G`BGw3u-z
z{$aKJ+sll3==^~aU~ATY?_ufOL;ts}Pjvv;R*)4#t$><d*vlhnJD)%ZhokX~>Dlp*
zGX0jtV&MLZZJ_HHfO@WO;V4L!%y@k7Nw}njhy_)x(B%4%?5&z1rZ1H`0%>S~kY|f%
z+g?)MFdsZW$WeYqFt^cU*?31^yT=)@ZQl*{3Cv^yD?!A{19mY4I+i;kaUMIIFS7B6
zo0XTcv(Fb<H{Q$5Ern^dwD<2z@})?;vP|Oje>-vp4G~64QUSs_h>vt46>UGjpRL!L
zNwcEb=M_M%M&o5u+LAr_BfLl>Ha4*}&a^fekEtGeYo3e<v*bh^O;5k^M5TSHPU;Mw
zURYF9X0`9w*5dlsay(d<TQ;sWKr_%V735xguf}HfnZG6sZ$>6Z^`LnUd$WQtpge$M
z2jrw>j4#eZWE-dYywV#eXe_2BEGmy&D#iiT<^yV=#BN>eHrNuZKI@TJ)a}OBV-^Fe
zbD;N1+GOh6^n$eIMA*)!!yB&bcl3<B>eC{}30IS=PH~g%#nvcdwi??-s$3VEd0cv2
zmTi%CU|=+Pj^A$MY~{`VlR$tTe*7H%W)~gnmA(<4i200PY$`H?*n?l+$?VxJSBy-4
zVTqoh>`w1TB4poBO38U3Fk!Qiv(L{si}Y-c9Wc)AvI=~J3~l(x^R)*KvCUF0@mZ{e
z=m;4MD5HW10)KcFUAs=GmpY{wMQ(Vjm?{i~J!E8NSt+6v2LzMb!y`Ck7N$z9go1K|
zuE1LNCGMMiBb&qJ(9JW0vMU=h>q6^jdA#H~?KGjTq^{XMLd-ov#B;tuojn8#e9yk$
z!LVBIKw1Aux6q!Fr#Ii`l4G5yR?YGWyy7-W4i1Qs^Ly%Z;WAFrj9Eq6zxqe-=Dcdy
zoVw$@HEX@kKyR1S0>wsfp5;&f^h?KyQ=uMM8QSaAh>tHdoD&3T2z4U7HW0r1wdGZM
z&2G(OpvQhUxg(-`zh!(fumD5kXSd>UyF6H8r{r}_oue_OIctZS@FZ*r_>7Y0%gTiE
zkc7oz(sKYlI|+rtG(e@V*+F9&Y5LMKTcIxt$1Tl`z#(Ya@5r2RX+_4yfm>v2bZqwM
z*xLt=tA*Fu*gqISy&@W0V~FqP`<j1Zva`#0@RK=X$d`uc_OLFMt$(KQ;Ya8J2-C^F
zfXv(LLV4CKgL9Xn;hKqO5IKJ}uQm14P>c~v&b|A@H09Yb@ojW~g=H<z^B`tvqk6my
zA^HbvwTpM2E6K{SfVh{~GuoS9-x&5yja+HsM+&f})Jsd7{EoV&oyS^bY4Y@IJ8P+G
z`XoO`hT{m43*_^L+$xS(3fKU1L1k$8&OpC2AQ1&Y)bfUAp_c3y(nSqO+FDn4!@BLS
zDX~r78B6&pwsqVIetdFHB+PFgH2}5cP~qW)Dy#n<8?>vtTlsu<zuw7a1lfPlW|EPK
z_xhue0bEHz0Z!ARwkdM=r>juacl-Vi`9VP^Yl^%DkWL%0h@~RTH>SQnK<aKstG<Rh
z73GL!#`Sp6WNViJAktToojwAPMsg2>_Nmgb<U-W;cEXhfTY=ytA5Jp#>hsLV?r5;P
zS%pJ*_9h9tq-14SVn^mipJ1`6$W~OLk)GMDD~m2(#FqZJ%aElHw>YtBVUfCjKmk+n
zDeC<zQx4JV=?2Tx=zCqQ6a^I@K}R0^re;6xwbs0>3Ih`u`i83JxRukxX^t=PGH(bL
zwKIkF^J02xor~5ycLFM$bd=u_j?=uOM=tAr&Lv#Pyy+iI`F0JKG->bIHIh~t7cM0x
zL8marcViM_CysAK2lAW)fb2J}J4b_fE_ps+{~J+}-7jv7+kLJ~`;!Kqdx<!xdgJL=
za8VhP)Ykpg*3M+8#LsBNO2lHqi-=%;%da39!UUC3oh=v~qN42k`>ia@zr=bL9hEAa
zIEXykf-1xT-j-mxJUuz&^O!CU=+or3&~%hmLUE8Zy=>|GVfhQtdn8pjnTV6HE-hq&
zwT*((WeXj#^K<=YLeekAXl#6MQThnfi?K`xX8!ngi^Y_os)`?b@9;iel3+*lQ~WPb
z8^dgx(#pwd#8C+ac-fd2vYS8!IZ~odLIvpb@PKhlZ=P|`iH<SXr8I0QBt8QbKiMtf
zHLa*RwsTN2v+uzG^%mmY6eH6~ikh0y?H^(^p{!ta&FI`UP;2G8;xUToJ~(zX6>B|~
zy5Xku%tu?R@_5u8RnD;(muYCA8ZK*+BAGbP$k?NxdDsHXSo|WH!Fla-@`=F?axgWG
zEo1SKOS3RD^TpnB-XS<Tt-H=LGyQR5oS0{tP{Gy~5ogf^+}7@9r2tmdPe!UmJVa&g
z_$6`Aa>DhVR^9aXSnBxl8{+pWuqu;lxV<y5Qi*1N|D-r3e)U)s3jSB3@2>am+nZSR
zxwg!k-*0Hr(n<}|&ei~Zx5mPjLzkIRj={ruuh7CGNNN|veyfY~zuVGrUKZ*YsLQWp
z#_0OwDATtu*#IwS^3HN1yRJ%QDs44Fxu7Myv}EVYfNO$L+)?H0aXksF5I@wb>)5_%
zkTNtU-Gc6Qw9eaT3npG(Zu9EauBkfmsO1J|cJ2&C9=qBvwP3uvcue#Hrod*CVWd)+
zs_b7{Nf-jRY@86PvT`+?v9KuAmhM^ny_8LKZyF?HKn`9opV2xmS7_71<QMSJZ1rk9
z4lTcJDd>C$=O6no*#HK5w-o;FV-ZiSP~1B^NDFfgGy=$tKvN0C#KZ1}!H>t}iOo_A
zE#G$<UW4F)`YC1jX2VMmVVZ2Ow$64Z5)chmDu^foynk`nkKu;sG`msF`VcZoK7n|U
z-n)<aUOhf`pjL)KH<mStxnNySwEXP+TX7G`hotHaK4)zyNtX|gYf|WPidXlb*cxt=
zTOh47`!hrSF-K`D6Me(4c61O6f3n;D{WV4Q8<o#LGt?lG06gWf&TB(z=cXnnbq(%{
z!i_2IPOiLL0|)zY`Y;!%o~}E+)sw%swDY>xjBMIbq;{JP-y_ZHTgWbKKdeyy(4Cc%
z{%T{dB-f?5dl^$}&Kct25Jkxk415-0I<rfkmy!Y#4yqStUS58VH-a4wqjFjRwZXxZ
zD7xtAiy0|r(Vu$*Q6y9(b|(xMS7sjB3h+o_7~Mf*6<|%#U8qp);u0!<mrTjP!9jDp
z)&*)fhzAQ(JEa4i-S=<nCJUxYoaJ5B^*U@3NzRx$){m;ka6ko+UM6DaJSB_bA0G0_
zBqcy@la8Oz&|1?K+j0Fh4L~mOmEp=)z~&_vU@~o*v-Oue(yCOv=jj(3s%xu|tNb2s
zF+3s3&!+&>;=uzh-O3`Y@LaVk1{}%SW$Ws7+mc>e6&-60K9kBykYdx`+WgMVBijG-
zRh0~+GN0$d9HgLYFLU3tvalrge0<b2ON?V?ynAT0_d84&F5-w#DI7d}mqSfwZece0
zenD9|$N;8&aLUcbSw)%2mb(J$|I&|p?4wygllA>eV2xm>v^2=S4Q-TKDk2zHs~kW+
z%;YsSNVvSO{U5cxbyQVr7cRcVIwGN>h=8C1(jd}}AhGENNfqgo?u96zQqlra(y<8v
z=}-iur9oPxyBqG@p7Wji{c*?djyrz$GH?vWaIdx3dSlLK&iTCaQDdGwJSpy8;g|X+
zMTCImB$c6d_BoJe2pNBBEMw+KP`}=+y`>0seU~REPlBka!}Rr@OOjoPiLmYdS@cXP
zVVOjg<}48|?kST#FTNN^2r<(tZ0v*>EY>)`30+^;ZhWRpUxoAYev*?yJP98O`{aIa
zw}V~&`CHX6xk?`LmVyf<x37ta8r`i~Mh4n~-W75-t2y)V-TP#IJ)S}+TtpahM#k2L
zJ`e9?GSDPePC6X>W(iU9%gA6FnPuhZF&Vd#5?A}oq&%Kojk=}5U`=DYG7W}HQ`UlL
zgi7wlYZPl9Y2)DtCs{pbJ<poj<Zfw?`_@=z1L^7LfN$P+RLvZc)$2486xiL|kg9{y
zC;~rUDC^vIOyJi~J|k(<LL4wR8pxaM+rJ+$g4q|C+*qlU?6+IjB&KL;pXK?resP&+
z+Q6WVPT1P${O7u{_UXQ{tv!M(|2m;pX{vVU6i8$X{w#WcPcAKWW<26_qsejI{m(kV
zx1nsTO+@6a8+T)N>H%_|Rx$wZSqgDf{_Zf%hw7t<Vl}dJOToQGE1jR^su~1=@1Prx
zL}yZZG()LHiGg8m{P%YeRzkSG;QsyJn%@ldj@Pbh39`jHNCx+3kInCAod0!&)R|^E
zTwO=>p%K^p%<x2FPlpB#RdsA!TIBibMzv;5NcI~vN6DuYB;Ac2vKZ+;am?sPgNL8&
zxF2-Xv_j(NPP(F7;Li$PO>LzlRogpgx256zpZX4RuIEiyuV0#-IbHjr^qWLbvZ=oI
z%^szZa*igRPM5uJQ`a9`n7l6Ww+oz{baRx%A(xRYD|UTE$xyuCx542d($>EKVa|wF
z$(@>y-WN@8AW%R(L4sY2EJrdtW3(W7CN}CBeA~LFF-FXfo{2sQ;Pq(%!P%d!XMtkV
zu3z&7cu%XyB7LOA=vysJZsAyB<T1(jXE4a3nd&OkQ5!WLesjBxTGcg|B%p2}#MQUC
zZIfK$M+SdER&#a4=N=BCU*~@itjLs?qKu^evJt%^aqgVGRcEOj&cB%Q%me+WXTci<
z%ymZKH!7@ZYf>k~3n^$y5Thowch{9|B+EJM9ykuY{njO2vN#?5ALhW>gUKtqivf4H
z=XB5WM0{xeOjZC{TM0+}T)Wm{zo6At?lC2$M`+sXw4?JB-yX9f3axl2CkBbC*%#30
z$%>6978N(@KsxnHJsJ0&#da8<hLrql?`c@L{&Ud~ilv14hw1fC1;3_4X}NJpW~EZs
zFWukRh%Q|<XxAxwoB86Dwd2L7>J@zu&JSOX0v6c5IrrY3-#ExkUg__>*%k*jLY6Vy
z{evg(+T%3cW|6;IzOu0YXc5cXXrgB{*w`{b>BaE?+O<oT{B-h_MqRJ^E7^$mJvNhF
zvnQPS986cd=sgZh0$rXCT+%z7FotxS4Kr<&x@&Ifp`eakFkjy%(x0QvxW@e2U3MG0
z2h9!?*7`M+(2tG(ZhSzfGE-Zt$yCzOe(+ppU|5I}6N2xbBEx>fwJXicVg$+NGzkg3
zMa`4_+MS;v`8HTG0=OC|sBTyVKbF`#cx*S6<;r;}KBPV`De95S9<Cjt+W;ieJ%i<J
ztrjiS!s7fyg##39=nd^z!$=?TD-m%o!PDnCEVWWuYW;fRFk?3U?dzSshk6E0Wim|o
zh(9$oO<dzpNot}f3m3zG0ZF<7Zw=>LkV<+JRmkr5>k0scuMOYc$Eh@|_m^I_ce1OZ
zD!UkiU(`^%I<=XT={~GcsI?uuN@v^L{C>2Cb(w&GuW{)4G*d8A_Fme@yJLUn*Y8&P
zfUO<ubx<aqof{2sm_#J=J=RG8&p(1<EmZzH2WwOr40Sq|QI!tAsheGM?o?_^E3x@^
z=;1tGD=c(qh*bN2gWdYJajs5km_N~Fg^tHykU2vu5;H0Hl@)Jk#7EbgAe7j2;`a6d
z6sP9NC$o*?CGJLinv%dv58cwPubBP)2pn3G)$)skSR}jeedkE|FB1bfTO~F?jlJd{
z@HMP~4{(!+N^CTS>yddt$cs}i`J5g?<lWw>!f{CO0GMbH1xn($9~>4IH?+Cg^Pwo?
z`D7mWY|D+=)2q-qpA49y;wjm)`bug22)^`v^)6QG{|mDpl9DuilV7Y%^b4%6r&doP
z$U@|pIrw~myA-OS(+X^%SAtt~5E+hz44;*oixe@I1PX<hN@3bBfK$Gcw9fMj*7G_|
z{7zjDQKJI3-Vm{X^`bS~W*VI*R)vB6G29&L<aUN=tCh*st@JrBV{A!qJZe0!i70mZ
z+kfqDLTqy8QiCtzYoFoDIo9?4HC3?u(|1wJ*HRN*n2)w{<8}>xUlIu$kN#1S-p+L)
zI=c1nPzJ~iEt2>X5#vEySsHxbhQh>xWxv6-3ddTFxrH<K4XkcQ*8=eIp^X2v(jc8I
zT!bkz=NZ)b!&TL<%r4Y<)e`Vc>Yq5!U$+{DWq+MO9!bdHyzRivNJZ5bNcLs(Gc&ra
z$q*=tbF$vNKQ%XIYmZR^pDYQR#Ov{SsCFT`hge^NAtZkuQ^se2xURWS*M23s4ZOy#
z8*U}vbI9X|+(^8f(iOVP(4$=%`Vv6Lp}iZgppanwgXiou+q;&dZ*($K%EE}^j?7t;
z*o={$zlO;d(ew;P)7w>as#&hV$v{Z6TDhG&_4}!%Rr|+AqaIRogD35kH-ZzDv!g#5
zUXP|p#pg>O6J;&rnkOYtSFhx>lz#bhl(F7$N>+3AHY7tf+?*yMg}tD5x;(j%iq=^E
z%Se~^R-^ptQQu7<ayc!(4h!UrkiJ&uUWp9w8Zr?WIplqH;9xiG4=;d%fqbg6kAEEV
z(ZX&e?tgJCsIjFJB`v)#%Y`v#hA64dcbC7&Yf*q45YP%m@GA=~%Gw9^82c1;)Y}=q
z+>HKNF8;Csn@umd{1WB7Q8K1PXP%d7_T=f;hP{>BAyS&|%U4*DWnpg}N%vw%KAk%*
zdm}R!tr-(-yPMr%-ym5}n{U6P2O$Nd<Q@j9eJ_WqmnQ|^JI5e`4?*Q$r^C`6V8WLS
z4`Vo~l2gKkvit4W1Pj~QdTK*Ak_4&_ZlNe!-BbE(&-FI!%zw$b{ZcWj7syR7g|p3=
z`lp~qbNXZ$d5AWosTwMt8j(z!4;ZsKJY!Z&3Q5P0VYe$?b{|%S=0TLbsbBFTo3{)c
z$tPUg@84Vc2d8&`v9duH8_~f@=C+_M$<Z+u5P{0bCdLhHE<Rq0a**edOI#`+u+w6I
zh71XYhbx?9r5_@6XLtayDaa?xq@I>!dxta)tbVB&=%1AIKA4DY)0S#(HUdVdwNqZj
zX7du1tOxHsxe9Tqcy>SR(1YMBjdxhWqV>Jfi?3xj;ML~`Ex7Xof3DZScf!i_fvXio
z`kg!ullXt9&HG0<l%<_-O=EP#Q$-&LD>#7_pY3gglHXHEmb5`Hi?0R-#SPTP#)aFf
zmltK_oYN(*S2f(|{~D!7VWp!W|KuS0Qd~jZ%a^+oJ~Qv6X_zkhr_`;OKO%C85Orzm
z-wea`*900DK~3Im$o|W)v$wmKkU#_wXTUUCbQV%z;J$1M?pd$yLEN~Tl3>)kVT|PU
z-(ZZ8=G~b@t_Z}%p_y5dI_9?oN_Or+o`F-HPJ=FDG3ZVD&iodZe*t{CnTCMhM_IOK
z9rlj4G)TIu$ZS(XX`OC9{#K3wgp(WMrIy*NBh@nAY3W8%>`8X^+BN&_oypqk3G<8k
zU^jdB<)pk;{I(k#2Hvl*yy~0_7JlTkbq;_wLTU+k(s^f=p6{q2l5wyM%XP^lumq4h
zp1#EI5HvHl`0bcgsA1s8JE{xlBHi-0^)oD{x`wgZ>g~Xk`b)Ds#**9MUO09h-HJB0
zvlc$V+TI0!7_c(<66y}+c{+`Z=YyXM(&jY?GN{lEII{aLGxiq%Bi5*)0XaMx9on=?
z_c&d7AeXC{|E!xB7LU+EzkhlEV1G~NU<mN;npo+d%q>?HH+U3yQlT!oI@C31Y`az!
z7qD>|d9GNY+^el#Qm|k9FZ_EO_)rwDG3o87RoTb<RiQkI=zLkr5pR7l){V9_Opfzp
zF)16rSbGwpo*g}!M$^Z~rQ0QrBS1t_3t|1|1fiB!dgU^rtv5Km9fd{%VryQP9^%B0
zdR(fj@-toejjg<2lcs^ZAQZVOdeaJPg#D`buln<uB{HZnKtjD<gkcI2i(7n@UJpjw
z);@i{170R2$3lP{c%l}DMh$J(WTN-ij4cpB2j8j7O4zJoUGo)vZF7i4zcVpU+8S}u
zL_iM$C=Ll%ZweHiBeeKY$;IlN&~Y+a>)mV4`{CWUiC5GXa<V2%kNuJ}>0Pc(!Lv{$
zGoTs5q@x9Gd4wxhb>ch{Y8>M>D+OH3xhI>pE$1dD8tmK%vDtt|$Hh9$yU9pZ*lFNw
zhSOh#8s=r)V(EU(yOT@HUn<86FS5Cug~k4CDEbodJY53nN4BN*M~$O)uQF>pcA5yM
z;P_uv$>}BwCczb{G*>R95J%YXGLosOmEk!_p5I2c*_ek%<~>GQ0|Fm$uiUKung2%N
z_%IgBe{4zpt1<_(T2RpN75V;dGgSFb;U{{GD(Xj>kFbos$D3`T(JhJ@J${Pk(cYoT
zzI)CS=H?hD;SPI?#^x4{0Z+bH-08Ae&nxqb!l5uS5JSSASzH6@c$naR0*V<$6B0<8
z63XEl{96o}s1b)4^ZcE%dCv6*qYnKey@sz#Xu+2u)Ar1HQW`aUg5QP;pHJ?Z{7eW4
z$kK2R=E5`$uU^-;9%xnIcMN$S{V2KeMUt!pzcp*M=k^WPl5G7?+pfy@d40s8)b5LU
z^Lzx~?DhjY`5GU+Y1L;}yZ3i`Z?Hy458hkRecILazVNqOyIz&N@{`MGA91J6xn~A-
zKoQ!TM+OI>Uune7P!*@;{f=O=wu^AI0$llbMuzE4-Pp=UF)>{wV*H*SQ#2onEtt~V
z-ekFHc(5K>SPlq#;6S9M>9EHDR7IYI$mf!=#Eze#7Cyb#e$%F~=*MB(Xe1<FA2!sh
zC@Omwmo2OH<T^U)98Ge6juWbbIHoK<9KlvULmu&#JWaWB0rITQ#``_zXzQ(4Cfn%4
zoZY|Q^88sHhDjHD9pFz}ow%wV03}2Nw=-)13}AnsMhluLjVY7}<p~8I=mE?0byX(4
zah|SZu3}F{i~I3DKEpNXbGxqG860&ZP%qebD=$1UV#1;i>jTHYn37w_zgdr2u|HXW
zBxRt@+1z9AUFe-1av3sF-<ygs$t#YAUPsrAdo!LxogO%IcR6X#7i;ORd>dE@qg{mp
zad^wr!*30CoNSlvh=_111x+WW)r~!RD^j1ChkB3p1@b2h9Z0pcVVubHH8co3k1wDg
zWB%}4JrWWT6<kxnf788cfr(W~pZM<G67^>@#Mg+3iNq|l_?CkR>{r(sdUKy}#-8f3
z<(Qc>*f?PnqcI6R0pdocYNiw`%k8R^Gb*RWbeLQ7xiMvPk!RZMqBJ!hn)OPUY_WZo
zbZXe14Fa8VhT9?Ech7`N)4(YT)0s((ljxz9Ji^7r^q%wO*+Qdt8alFC@+{)1sXRl2
zU9F>fGk5I?wWQ^@HZUjix<9BZsw{K9qXDR~)->R5X(v#{HU!$aQD*Y;Iz0Sun6kEd
zETg~M&%oSo!DZNY=1tl>e#2GVrP%0*NDbBc8(gkT@T|~(Uy&|?gv$n51}Ze<rM+1&
zq@SCHMx+^mOhum?Hyxv@pX|GC@A$QQFtc+b!#^(d>}eeSd$;n%lK0CTmdwhWfbj}!
zCFdT4!h-n)ef7>&bJ_@L)0ri;y#4Y(L-QvuFh2Eyl<AoqQyZ3A4hb@hW>!`jf<dM#
z?m=0Nof~%4*r`YU_SAti0IocY4UK#s-npHea+Pyd_vQ6cHUHkPIn&ZI&`(?&)%8}!
zrld!>GgUdeiHW)XOLNwjrjfMY%4$KkbXz2J@gtO!Kh71-lO7H6j+Hyo1TJUqm)T_*
zyZn{psbi5XoF}*D9yQvT{`q;AKiLVBDnfk1n2`Jvp;KcJ&DAmJ*xP`@7Y(~GC6s&t
zk<N9qcunYf=NIN2G!p%5Mv^A}fC*qW(y=g72~hb%fKYV+k@M-=ZuEh4mpTQ-A|7vJ
zS&u!WN5F(ByWYi(cuP#N?7Gu=^>BFjofP#489Au%Q&~R5?|m0;k3zJ#9<%2mpWsT=
z2RXT!OoCIcJCH^oijH_05gpz4J1MayQY|6RE^|)SJo&_I=fkJ_A2-&Ye=dev!0(gB
zri|w~N$qK7)i1S5g$GuA&jXjg@Obw5be@iwtX53l`>~zHy6?4KyvY))?VI#GgRdzi
z{`HeF%H7CSkE^fFWwEH`@3|I@dStwiQ(H17AWh^ndKmS%lv!TCcY}P8vsCGbBeE)j
z2Aeklq{&YIGmL&KL`|y!%68b}LZfL8MD42+oYWaaknYj#TmygNx>Lx=M^UESlt11M
z2@*LTu_w+^w@{x)0+N(dyDAbe+)tr0hXgGC6gRJ0RHs;WuHYexj~2Jrs~(*MV$~FI
zJ%EP1>SUhgBz&SOZfGd0wadmXbP)Q!h@-rsY`xxVl-%>#osdG~E9(OseL*+dTfO>7
zq{a_0LvyRG8zo%b|1+A7@x|6yzqtj=o*r-Xp8yiyTyNv?z?pU=HO=7J<65FRUz5PO
z&d;3Hxw#+Y<);+gL-i(sWw$P4a>n=EeaO&eYf#F~72JM_(saImra0ci-F06eg5g>)
z!Qx2*TKVTAt^4I}8W4N$?kT_0^8iAT5U&S;7Ze@2rZ&&ij(mCX;^t{=nB(FcuIHgj
z7vb(WjiY!Vh=fW}w`T2=dpNJ>c!r%NRx6I%qjV<*-r(HHzP5!hkvC9%E7F4|zbA(|
z)6>0!g8-1ezx`t{1o&ZaM&Oc0f=d7t0)Yzr_=oli<m~Iud;Cvxa6<Yg&h!{6p@Nfi
zq0Q1pD?+c8NAgt-jnMQREBts!EgS|iP?gcuffNs6!uzwefDb}OYM*@qX2IOZc{+^h
z>S9Jnain)?0tK1CC8WfNTtRjs-G2h#3b`i~qAyBykDF~GZcb(83m&DOKSjcuhj%9G
z!i+C;w3^u*xHNXyP;m!~=u`ThbErPgepSeGoeecA>=^@v0DLRq9puOJP+WhKQJU94
z@;483GyKX!j=^BTDOv7_#VQY?Th<PQ;nIPhz>Q|Zw>nWsQ_pt!)l&be{#(~bTZ9UG
z`n}q2F=H}gJNC#Rl5hQ|q28-_|7fzKEdGOgf)Uey{LAXfU2ObTXgP77<OT^apW}Jm
z2$Yf#)xK<9t>+gExUFc*GDrv|C54dv3vp{<2b)iyuEaxal+AH++toTpHA(fpTgL;a
zEp|26#hoprz-Uyswf(5|nzF9+g$#u?xzLyS@7Y^cZUSsRk|j&94>yi3c8vUVUaa|;
zSbC;dOD<aU%Wfb)Fj{go_&Ha_zn`t<bwVy&P}EXN+L<?fP?AV|T*+&A#pjxBgT-)r
z@rPm8VwKEZCZ5Y*yYi*Tb8>bk>QCkZN;o(T=ge8DLZqCHb{;rNH1@g$4Oum;F2;*N
zfEHPGd>_hcTUH0w`CFSeK7Zf~giwi$HsD)(k>?ESsrZ%Q7yae-9|1Z=A%|O|h`j25
z$eI(vqF9X7&^`F3?c&UBV1hu1^8m5>rDNubS%aXhCT}?Lpt%MLtzWxZ!!HVZ_^W0e
zutiEyAF}PJK`Xze`c2vLJgvt4Y;MIlO-^?%&bU+N-0K=W4muW=gP&GU=d`tL3H7J-
zVH_AEtGDl?3Y{pCXYKJ~DRWXf*PEx9_-=fCy`@c`kueVI={St$oRDVOIWH(rr4V25
zwq80{El4R1QfZXe;SkpQa>MJC<iaoCyud^FwXLYr^Ct%09B?k3e1KJ4|4L5~c!((-
zG5CkJ^f+ZflwOcexNkK{AjA*nI8=zZ+L1iH>?KJDWZ~=e!i0~qNsrg3;~?Ta11IH*
z8)m{I*7lPoWk){6;--eyceER6+Pu@7*;PnY&)$Rcs;Ef3dH<?R)0YRDKPt#Zx%$W}
zuQso}{+<^9@!w(<67&#bHD0lH@*LtfXh1ybz99#KmFtP=$=p-Nx`eK{B_Aprk(Upw
ztHn&k&!Ti=%kF*cAVajtEMx<knCPjPsKRb`7JU1C4K+SF|FsM_4h<iV(fyxrc{@B0
z1kDL=o~rHW^X)tqvbVAQ)-)V`xQ=V}d%A%C?y>qu-V7h%*y)<!*q0te2@c&n&h+7P
zl!(0EbGEzZ&f%?DwWAaWp-{*cj-b)Z3=_`ExNc8>eP_(8{UOq@!+IE;?nmhHL#1X#
z+0NGV_&XC&0ZO@M*sA&5Ld0n~90KB6+ensX#GJUX2U;NECPhd03{nA-S3;FCR7)Re
z<v%%;>XMPh(lvl@_GIBJA;qTYkvY8@gOFxHOh>;ZsunouZ1Cua>9jYLG-)CtAa(zV
zmFtDUo2HB&Gj|=`0f7?_oXRRvj=n#6w``hlb=AjDo_)8WWZicupENRkp@mN`UQaG3
zYos7605^bRL~!&=BO2<}-3hy5EO2r+aYp|xm`e6wx#VeL$E^)GXL&N@^d2ZwYa1=B
z9CPHcNQ`o3%E$Q=g>$~=NN$h{j^DUI7%H0iP*3SaW;FekD&bRKYaTpu;hoRIf6&(2
zO0jD3>O7_>2Tx2)49}dNjP{WKyKg{`OJz|}cvqJ)iiw$-jf<<AvO6oPL-^zah!7W;
zcBh>ey!PPI^5u|m%UdDiRoqY~jz8Ewf6Ql#=Ys@AxDuzOwI=NSl_yW0x1;GNTIY=H
zF+in3$vhxEj0iYNf#5}iHe=sST_s+tFUJQWW<Q2{Zv?7bDhFFnzU184EiE6OxgkWZ
zz9{^?Tc7obe52!^B0k{yR^jKqBc0ESBX#oGxTdFAZn_lM-Gbv&;DI;}tmqk#H?KZ0
zFPLLam^cDuxRYIwdWLQNQdRQ+@b-4J3cq{`lG!->Sob1B<LV^MPo7R4zx;j;7sCeU
z=mdnGU0UgQACw&zr#E@n?XlnQI*rk+2Wo|Lx?BF}5jfe-LqEe(UY<)ej^bf$D|(kM
z=uKc*X{tnj|8X#O3xQLZw**?hoxFJE=lv$+a?6-=U}|2?*D=4~_5MzS6r&kPr}@UE
z(f~0-fZvrGJem2X=4t+O_koPeW@@^~kAw_t+IM}j%P;)&5%P+s`5hCJ6Iib_2~Sd5
zZL3GR3mu?|TVuY2tlG5eS3EGYVu2lPwC#At*@cZG!wz|=&VNEypOS|2qiO0ye}vGc
z$hiBJ5<GhWxCn4kDc7PN8<}_k^ILEvrV~z-_?ASRg7dK^laCwsitxH@?I_;{?i{!A
z?Hv>xOo{`!=+n&bS*x^^cadR!KA9JCw;eZ4uZ!$(VaSfe0%BhZ3fCOs=nHIIg)_}p
zT-gy{IP)RlP9ZN3XRQns(}SQt+UR19+SW|Q9|pp64)*$y=FpTy3kC*8HZBf}w{sW%
zliWI_{X?#2dnX;uW`bfrhhI8luvcEbUc;8Gn4FeVghypC>E~IT%aYb05ionD^)<_W
zUe*D3OLKqtj!n;}<_6$NB9LBL7>P=bp2G-<d+!`W;dZj=iP-WUPh{Fj|EPzbq}mih
zelm={pLs@9l=@P@#aDN9-&uKyD}@*(zrKYljLc^-E^76a_nkdRPGmWJeC;y##~kU)
zSI>y^C0*$9E_(Hf_^-L>*5NVA_m7_9zH?-c-dS*27l;BqzrA7pFyzT}Ec?uv`Dr~t
z2fMuZWu;h7(@mdV`ILkR)%BImRA|)I6H<7CFC;#9_VsJroj+ZlN8jSGvS`WcRXdUk
zWF)6|h3k(if>NH@Ahj0$d|xpism+6=nrkWT(^DFyd4+pHcRX-LKd$m7DACtQ&g^HP
zFKS}DTX@lato27O?wdBsSMBW~o4Q|(rhZ>PQnV)VEr({87fxIMh&wuP+5Ahh$10HQ
z1KI6T)+KwNeKs1Jcp)mZ*;PIE>tm<0r0rkYb?jA{ruTB+EgwGmP0Zu)+agW$rMD{W
zz5Gvq8jyNve9|qwP^qRR;cMk_OtW>=GP&=%#CIc2rk@)L>^!lD@Wf#`QCo|v-|))=
ztrGK+yOnF7`9^jUKi^$0;p0;LDD1d+3%x(x|MJ?r@j?NSbX%)39NDV!6jFCi@Vv<|
zvtv@9MAghpr&trT#CZD}wke1+x5Vnlhi@cRIA$}ABb+(JJChtXE%K~LN6US1-2H=o
zI7FwZ)80gS?Xvq!+a{>WjeaG^Y?L%|6d61Dhn}V}cM}hY5f?YLlYR>yQ_Gu`n3|u7
zlhdX~*ST(Gv^sl9&UPoo&7VDQK9}cLG}B{Rj7P|*A-rPolL1ZImu+m=@Tbj9-nYN>
zuG;ajtBA$#Ra^;9n@yWWeoVV9{}^z@_+N9PL7I$6%G1yW%r%qhf|UJYWkK#EYt7K9
zrK-{Ux5yk`-H$M*fOph_`$r5@#5}BJ!JKHxt(7>ps}x$2JNTrUgO^{Bc8ql>AU8c@
zb<#3xm`;@ByDNVS9Tz&)sFU-3G*VI`9pui%RsJ*MPD!=?Pbu7Y1efHb`vTi?e;6a&
zV=0OiN8XJjp%g2l;wt;P$pU_2B{i#Wp)*gm5R=V?<;v~g@^ZtAmhytR2T?3FDZhI1
zjJr60wOBuscAZ<U(0q)qw=fawNB8+T6*IrHrUFI>BSXJt#p@_~Qn<?x?S)VHA9z-~
zhx}2gVt}RlR%)O^VT}>#f!Rj(ZGLR6vjxcT9VfBe<mCCB8LAN1RiV$4@iAOvZ+F0A
z|8Ow<s#;C8&`6miH&QJ^TWH7PJpAdUU+=+Why5LXZ6d2mJ@{8m+J(cZ@~)<4`sC<_
z&$+X6OM%+<J6}_!J6c#bNL0EiM-J?lhFQa-SAKLSOPbjPTlEr={OP4=#*14?_c5RD
zpoBsrJNLv#a``y^_FQUd{C#S<1G5)8Q@i^tiV>ueugv$%PZV2bO+4Jj-)YI6NOc^Q
zst*MxFx|hxHl&p3^X5Fq=0+apY++t_WLUN^{*4qT%1ZqPn&XMn8J1f5qYqp3Jj6RM
zU*6wgb?bHAmyD@&Ox{M);#j(tEj+8M2f{Lr?P`9!&x@jKnfPI6Ah}S+mdq^ur-NTH
z;aK1+?HApd^<-Sl{YKpxT$tt&g_hNO4(8;k4o&pZFrXSG7D{jEU%WNHJfdkn>LS+R
zzFUySx-X&9XW13*5PS9aOm>U+%Lo<4M50H7%k<<=?@(<u?;_7hzqluP4XG8h6<f_~
z?+=|E$=F5a{TS20bi+TDn-#htUtTnScba{u{`r8A7Q<@&v48ny9$D`4Q&TJKM$53I
zTm8B;UN9-$@?EiC0>6dRb;te4AjU-3_AQsvxH!HVH;?AZ;hd15ZOgs3BhOUgT<ttT
z$x!=o*>b!(MDb<0LvMv88UEa$TuGxY6(TI0O?8@6zApdX(LyKM)qdA<-@aa&S*u3r
zO_OBtiG8_+Z<562iMQtc@?vg|{rP%=zZX>83;whg6{%Mo_t9?7R0&OK5nE6#g;une
z(8}{G)wvgYIOLK^=jS@9wl-%&nw@?xSV%-JdR8O2(29t#o^Jf3&JDXVsMbifD!={@
znLwR2h>WqDBO)R;T?DCcw`e`{r?<FTAP(#B2bL&m-Sw|#G;3@YHCuBzHu@qCxVWn&
zd^qEscy)?WkG*=?u)|Ypj(sFmXAiWpvV@V(fpoh-zhvG17UOUfEq`nrf;ns6Z$43u
z|I=mmT~M45b46al$FB?7k_+mjU4<3<4vV+Q{2l)^R6-?^V@H}0pA6fy_(Eo5#1ft2
zJN>-`3SkQQNNNX@eZS_nGfKG7P4dX>#qpYHrF>kQ&DgSWVP@@=vb0tV*NRL}QQq||
zvBEUhEo52gME)i-B8!K8N)$@r;fw25j{5q#W2uGJn#EfBOCxzomRgb72`K@x7{1UP
z=s}**7=dBrHEfHD3`tUARp^A)a)wgnRcbj_)+3@-ACpyu?<vYh$)ImUYdMbR8(EHJ
zwjCh7ThI^Dc%@QIVT|`}=RcB4r6xI^-glPCI1t6q&9C!Y&32L2$HUh>wbF5-WG1c2
z#~m2cjG=h`voXF0zQw`tp$CSNTbh!v?19+RSO$T6ioVqPv2q$c_`R=!EQ?(i|2i`N
zfjWDzH|N7^w>7XCt)B3wffEWP@Yh+rtDSiU=XLwb!U#T~`(Mgve6RdAd2-aE*LLiJ
zM12(%-5hv5lud0Q$2HMbQhd~Zdj8so6yr*z?h+MMvxT)3se_>mrgDSJ=j`sn(w@Fq
z`}#^Y?cY+6C;<w%{RR?7p#%%oqtYOX;@HuLoTXp<Cb!d!)yP5H{TseXgTW_iPj25<
z`%q-`sM;`F=wyvn+^?_Gn43@)$r$nMZ2W9~DM`V1ln(8>nZO<q5~iZY_V<)4`YVo$
zvjtLp-RiyA!ElE|mHqRw;&5YvDfqk8py{J6+45zreAOMMefhI2QE5V(URH}gQ?=Op
z&vYOH8ljuE3Gujuc$L2=Pe{5SQfZa^c(Z9=Rr6TMu<?&~LWv;E-|m>xDa{ZUXmM{b
z2RfU6y0X389k`~?kY<&UTlLFUSFa@K(8E1EG&;2xEm=%I^`1KO!ASZ3Y(b5QNK`hH
zy#7(BLVQ<DTulbr<mSjI!{_0n!&GhK`jwyA9mpX%7{3N{h3N0eNaO;<p{)CS&`=B7
z8SGn31-l}X8+$)POZC}HmgxFe`0j{Kp4ZEhw7QYTH$_sWMdn`P5K9e>ml>b22iWd(
zv)mVz6J0RbsD2p77n`*DgD>LRKY}{Nm(GsbLdRBH9&+_YI_~bd%Gu8)jaz+7&#NNr
zf2D9O4rcchaX_QOx448F3oWr3N0hnm*0hz7ZkG#3eOG-{6yCE$HOC$+&X*~*YDm1&
zp!2_->EP^flQ%Z0t}!T6&qmQ_zb=(c_ayH1H^Z3F^Fj$qF0vcL8()a@`xRYurV;(?
zvQbSNUs|GTFf!xz<*LWRGatF`b+}oMJDduA7Q+&MvG~IRPQnQcin&%*?Ud~`mk}IA
z<UnzpIURKPKw?MxX7cdi#yZR1`*TSS_T9a4{v+`T3Hmoem^4C8ycALrLa2K$uIAYU
z?Z24Wmy(lFX=w;#7^ZA)RFaAB&nT_5t<?MI2xH*x>_<(X(0}%W-wCMLoDIUcmQg!T
zk|F96wYn$F&tHKXrc-`+=w8NRyJnA|;@{v{^MYV=CJ2aQhfYJptM{@QLAR=&Ui{(V
z;j@i3dX4KLPo9LP6nz-RZRCyi$PLbT{ny?Gk6Jd4U!88{TH%mkVF5uZHzU*f!Qc`x
z?TZK<mZ*cW@ySUlePLmLqA$|9V-n9mhfn6(`_K#VtgNgG`<5K!EB)`Hq7YhreagMv
z)wOc2i<PxAC&R-!Q`7MWOxLeorFO%2B{VG=no?=+C`SpLJ~<r`)@Brzmp7v)MC-%I
zy-|WD#fy~_nyO?$Oa>G~izdHi;u?iG#tYEY-Meb{=vmoDqWcFfF`o_l0lb$BD^s|h
zjH9aptF|*9ah>^V<>O1ZHKT~xMnTi#&~tf4B)$iGwIDOpbg%j35l;M<8SGq9lK+~V
zR7svci>GtSOxUyKh4haa6-n6A^#_FCAH66GyrLXLPs1()&NnIl-8*U7KOhI?_473j
ztKtQgi`nTyOGDTXBrFc-E_>=}N8Lh`j?Tn1|D(F;g=hu9C(-egZt2u>(UkpkH0%qZ
z!TRr=Mf8`3@+2m<99HDG)g1-?Hqt$!SDjx-cV#a{l$+NU^F>BrhNhvd>e+rSa=G48
z9~m07weV^Ubb!mtdjIzo8RB%SI1wg4#RxOp7SyGs1aspHCyTjwQ(L#eU>uhrQ^vYG
zbMA)I{F|+_O693ed1li8Xp|~w*hM;=F?)fk&>cIR6c2RU!uKdPire>CFFSX5TcS?8
zG0WJVO5EoRW5-<2#3r7H(|Eqn{@(Q^IezBYC$n28af!lX0zLC8e?N)W`6SSP{KWU-
zMa<5kO3cxSt)Q@55K=8@BQ-|BqiS|w{<F9@{DM=qd3I#)`+sDVicg)@OT-<(6PrAz
zlO5tQS!A)H&a7FYS3}O1W00=U0>~GCp<g4o55tjS?Boz+$x?SOC+HuSMhGq~L-dqm
z#W{zStOp|%S~vf=ZxjlQg&yk^`}o~#Dx`;mB}&G{)bvw-CKV-uWKY(LwztYoO-)OO
z;>`C&1Scfazj>VMU1$tSAxTMp7B_+4dRI3MT0+FjT~w4w{tV-_-&6FgRAOJY7Fvu9
zFi)+gc|5e_(^W!MY43df#gVUdoT;*oj~lA+II&j=h0=TLl=k<KpvIFBn4W*e$L}Mw
z;{N-G0#Ncle>&6Q{wqV6^nd@u|7e+w|3A3(-!JgLxij|r|6jhMe)ehStOU<}G-$1<
zH;T~QlT+d4<z3&|i3$!5?tT6kNdilF?dnyjP#hebXN7Cr_pBqMqG-|Zi4vX*Ilpbo
z!a=hVy}q>t5~?LUmvf+fd_+`KdTp)94c{jY<eGHc+#0ZgLa$!E`s_KWCNnLqq^GCn
z`qtEBF}EELIryVc)kz`k<$;2_`M*$EBw)qFG&SElF8xf^t#EvV+&mlcjYCc-;{n_x
z%~V=B&{gc5`BkCo&X%iIX{P1=Jibp>{U}%CPlY;nczty4KxNJ@c()m4Kkx9nD;J}A
z^OVT7`>kI=m3Gw@GH#W`Oq#dQF`VX+puc4O{0>TesM4iyxoQVGNrSfe=4Q8o_J_Yj
z<nGzeDuQfTOC!=KxzL|i!bjjm`gpoE+V$jUhX&1Bx$1q6fZPPGAAR+N?w^(TznRJZ
zkBPi5+s%O?IECVTpsXA!pCH7HHtWlY_V)H>l1I+enmu~-Xn(6;p9Z}JTNxzUndQ$U
z-W@134qo5bFeLE>{XBCAaujFn+GL}{;s6_3+1|eB>({Ta^J-%i|5dFdtiOu?fAFNR
zFi>7<>Ig<Fm3z#+u$3!}RzbZbk~y3HvRBOX?%JeQrE>vR%x|(WIQzlNb0Fvxm0PjQ
zCoUnOb(2tp6BHXrh1_?=i>sz0wK<Gh-!s5dr@`_f<F@L-%J5aW?JzJh%7lc3%nz3p
z!f<9}WYA?(;jCV{>U_$Xxtjwz`?3xo=e%~<<W%(C*L;jR;&p#-gI*no(K8AO>21t(
zMCGwsr=qNFl6EfoXcd_#)uT~8R|)*Yf`yMuuMiRW566dxr@$2+ARk}8QhmT-GcMwP
z1!wiHJHx5*t>u~%&k^UzK=nqH;>*@pZUwvP){wlsJPMcJmv!M1+`s-(L>@eN5kw)F
z1?q!ZH!q8Du5D~|_v<6N^T+Vs=3if3xQ{0m$LEmM@C=o+IhUTmz{aKoQn{|{O|-q`
z4(5CN`}2!~g<7RHN#H2AzMQ9=&C1SIuJgsm1YJywG$7DJ>2(-Qre3N=;sjiju@699
zwKMAW*56;{@#92@|K24eBs_ikbTeJ}=uKw@iW<$RksD4uVpE3?dH{XCrxY!TS@oFE
zF&w6$u%uaqg@wWUmOCt%ZO(R@b*BYVo5G4wZ#XkflVJwa|G{{)(j~OLEw$-28Ojhw
z7CVa6)%a(eFGC;~2P}J7dXL^%!X56Sn_F9voX<L_%IjxA;M;5LeOl&J3+!d=Jo|T}
zGUm1y2a^{UEqwg^81LR~nCC%dy?H`{RZ*m#t9@~r<}+GK>K!#`jeY)UdumnBz2uTt
zzjnEOX7%1Q84bv00HVS^Vs6#87FLkV_lYKU<wz?iXn1;7Q)fP@^ZkibIOR>Z#iqkO
z=y`ZFL2&WM_?MS5H!-JcE8JmLLsN#VDgt-#yrOOhdsbe#a%Hnm+j=hAI9|!ZA_w#z
z!{g#G*429wL!T{GHm2J~R5(#0YWHO$U!NaO^*mgYtUBIbNL{w>&y9<BTjrhaK3)0!
z%gggyt968ockYN%Ng_c#J((&p7EUe9B__kA&wI%TL;^$MrXf_4(xF0ozlnM?RT_Ui
zL}d-lb*FQh_c0y8wc=KU1O$CIoraTMoIc;&*x2~d9#wgP3f46yG4Yn*?)XdWAi<WQ
zt-FVPLeQ%;oA1fo2d&Ai1~TheYvdrdSLu{$SS3;^fL7Mhib<1?%YyNk+a9vkkdk`6
zzP`R$M<{Iav&b|?7^&v7pG}mH<9qah#7nIYqfLv3n-SIP*2=&r&%ttKc60;9KG+dJ
z8OVgG)q02LRRVCM*8Ey~tV1<apSG5UG}>ah`%e=JBo>?ZGjMW7_(Y;ul@OlhhAG=W
z>Wnk-Zo>!Q1?-Qu`Wa#8VrATn+G0{+Gxuj@pqPj$DOu-+UwfiPjpw>jW4J71?od%}
z&1dIjJt{N;2chQZ_*p$yClfi|NYJWq#H1-C<hpLnKe~Bcgt_sKIZDxccGkq{l=IH=
z2yyx;MO{%bv5_G}%omToK(5B~h#Kuja5J=SA|p}<WfuAS_p{WLl<!t0g<e<2eQZJP
z#PTNxx+|Psr6)^`ac^(0=^ozM^XC`RVy(JBlo12Cxh;mX<yR4^5+D<bPJH5yG`Kus
zTKJ;&6H>#Y{Vf>_B69M@Q5Kzd_+CzrV3|^zu8t#LTJtVACyDtHphHQyt#VdZSFa3h
zE{|w;l{pA&?*ApCmJFyQmB(h>i+U}Bq1J!*;2`_?WJ72QyK#H?z+DI@q}))W^&p)A
z;-#S}r%s>K{BX|e8g9)O0}_c$OWd6@om{h~Gqn{;<7d6Dq@<;pfpG{;QO9UK`fyIf
z0b%%{P?+zv;g+qA38m(=9@VL?W->-O$HIolOiUEPDxIpThW9>4X(rlu1c9+rKDkxt
zx@Cq|mX(!__IZrTz{A?p)Xhz3uFWZPo{$0?xEyT5fM9eh)UlI6gVxY>hyZcWXKJ!2
z;@R%>U{O)gBDe?W+#!+x2SVEe`yLfM^#Ln>8!b-!`1H77_3p2W=&sd?I%avCT87J)
zFPr@=%A}rKYozQ4EmttzG;s$4YcTSr;59>?F(^erys)OnDCZ;?7n8+-0>jUyy;%`@
z?rY-J-uTW*$&$flT`6esGy|ToIH%F3^IWwTA>=bhTHmvUrtI!IpFMX@>%(7OYV99*
zll>@MrRL`5Qa{6%3kEL_Ib@Vt5J1Abh?tlxSiaaX<bx@Jb<vyuL=0<OV1m0NH1wKc
z*=qf#e*ZS<{3yOZ;YVrz`}<`dAD>L9#F6sZXMh=f>L7q3*35tW0^T-h6sdAsC}^Z8
z<Rd`&^O6%_w2Gw})pHb|KK=Nqu<+WA8<_x9=6uPl6hYiHC+M$Qng4dU<p2MXDe`{~
zdKKmU2q_~+tUyD&e|u>N6EwN?*<$dQr$A^}7$yjt*HFEP#mtGa6^=_?{;bs*rKLQ*
zdHVV<Uc7kw_O0^DXw@J42QJo+0J#}z8lx_WaZEF9p0N2oVG~{v91wu#q-6W++nb2@
z?Hfyjg^eDT;0Wx#Y|jW&jPAOFV1bv@dNCh-QAYIr`}aBAcWsy-eM%Xq?&(oYg$2Gf
z-|M<F;zWanLGR5_O3vap!YFFqa^#C6rp<hWd-1c+5EiJs#qKNw^|q;_L!&=ex7cmR
zDmxkbAE;)L7oZ@1bpzaKsec8)N!O*~eh!O)d!TPCRb5?uUruEZT7)_-zW;1JmJ6<%
z;nuC%Up?4M$33tyg4a`QwSC7?RMh)tvAM>xnkOW<7cMA)VFhTI7HeHyR!;RN?D&R&
zVM2{N;o{=j&;GcLR*quPeS6J`@eV(~c7pr*8>^}<b#ZZV`g@@LpjF(3wVtpeBV86F
znYD*6S^<QV33Xf<#h}CP&b?*KRkX3mFLzqWbsBY%1XOpuH8egxx8DD{IV}D8!NT-A
zckWyzA;}<S)-pNVUDE=gUZk<@Kx?p{o0s>Ig5Xv-fSe?N)(*2j9<bINeuB+l2Wgh>
zp5_Qf&Zpo05`XwG12l_C?>~DDaARZxUmrQz0-Qj~_Vm?3DF}WxS(RkV#d0kVf#aBL
z*Vx_YQBlynCnny3O{ZndkzkdI-sTS;AMEHO$cglV69QHXdo`?f|KV!m4hooW$(5Ru
z>T`q?Ie%|UV*ca9vn7zyz-}0~m8ue{SOBW#u;E;%qm7HeMcg_2bz^E~CL8=c&GJx*
zbpkBQbZ`n-x3Ui&?HVbmR=i5CdW4IITzAh$aeoezABlG?v6F`^4V5gwzU@qti|ql=
zrM&TbYHALAd@p?L?Qr$uFF4Ga`62+px?t*l+wj1zs-mJ|itdw0Q3x4WqO!kRUe^lG
z{Wyc2erFsB?DXHU;EILiL5J3=axEsH6w;XxG6qiup8q<d(DA7E*^50j$jMWm$}t39
zXa{Dz!)bf*k;8lsnx3A1s#)6v-fkY=B#^R6d$4~eIar86okUO{Jly{Fl6DTA_ID0B
z{F{75n$LIKwLe>Q0%nPbjny`aT+f>2#Xd;9H#UN$oA$Y`XJgF`W?eyG?G=~xsQVY3
zr)b!ySXpGWa+NExwb?02IXByx43Uy;nND+4lM<}j{jCLBG}eg%HqtOk0Bjx!@tRTl
zhfFZ9-=}QpA%_&pWofHGKVidT51TFOXCpkrR>H}lq+}4elDvG7+ba_Z3E#dv{XFyB
zGKsD-(chm8TUhogvt`tD#?li!wUUyO-UcJVPWt+ww1CkrVF1nIQ2WE(=N{M~h@!@J
zN}5@>T(zrAe7_y}0I|r{Mu#wnZ0l)#wutV_)zx0yZDFa9O*1tEDd6QH$K_%AZVvxL
z*?57n=639QTP3**s$&<ha9c&WAOYaN9XN*RMoPruE-kHoeggPWBs#P;KmXpw=3IB`
zXFI;QX|GeKWzj|ui1;rlNJ-I!8i0^E&30Hqf;B9c(UE0vN!O9CnieQx5vqi`0Bj2%
zn1)whJ{bILsa>`_18{s<h{AO?Nk*ueAvbTtid#iRMP_kkpuk|+kx*b&<c%e6p7x`D
zaM>}>pCiwg9cgqYArvj!S@~I<-D8*i_Kzf%z*&HED3Ld#t_#dG)YK-mUZ=#Z6wv5S
zxIb3+nA2)l69lZKWo4NP2w?yR3yoz$_1u<<m1n9V+U4-q77IJ1`M9ku9$icw@z@zj
zWjR*od0O<NWXx0BOxUns&;lOOj~YRPmNLBtr5YwHcjb3en8To9R75H3{rk#-uAA?u
zEl0|W2aa=cn07LTm69bgJr1^0r=!^nBr<O>WZn6D$zS<n>{YQoK9WT@3#>Orr%XAo
zYC8txcs~O6bBa~VR;TCU`uy#V2`IF?RrepP)aZZt@&y+U53xT1MAI@hrtesx3%~G6
z%LWSZ0*}mc3k(|d?l~m^YK;21{Ot|V{`gBm4v3opiK&7sS!RNdgP}`LvD+w~o|%bq
zNl?$oS~xiPc)S7|smbG{#%Xg_X=nRAn?ds;Ei5jGn+!kG=e_-a^-<$!cMR0_5R_T<
zyLXqi=DH<4_BT8APxdvU(=xw(dswyAr~P*{{_-4t*Bc_*tHi|9Ld%E0qU$E!)lHPn
z9x!Dmy6#OyMY-g(8DX)11c=BV9qdF6OxIu6w+}X{B7sQJvEu0CkcMAYc{MC|LkhA>
zmu{gUMO?ofr_U4g<_$ekNLhEU2Ud>dKz>rI<CId!-?NVe^Qc-Hc093yf&y2t5u05y
ztY|(Bx&xTkw?)}mzoMUzTr=B%&8w^Yyd>?@`hH{sN1Is%E8{*k%VtLUXI4ZT0?B4Z
zk0Ee;yXYSfU~hc_(ZhYYjgQDv0|S|=jylATFe<uf%GPwO+9Natj3Dul22Y8Vh=`Za
z;i}k80cXV<H=u`h9GHe>Jc#9LGRz-Gz}s8{NH#HA{bQ0E&F{FF5lRl<=~a@5&7d))
zBqr8Y#l^+xKGTOxf!jg1f-t7&;1Gn#n}=&W4jgwJCv>uX<l_Z00|F2{bQd@o`+i*~
z`nz`}Zr{HBY0yGk!c)Dz=Tt;yR<5-;hgr|}DnX~^;rm0&V>LCc^`q??JpK2u-}mv4
zuf=tNSr{3L<9A~D#Cq$NNFBTqwa4n0^IFAbYN^9B0XNu-+;*0)>_D2y1S0BX<ixW8
zvoVv&{K2mnp$2o+NoHNC^d&_t8~c5)UcJhYiDC&&fgoSA=5Sr+%@X*Y<%*T+`v-i;
zsm~CXYv@VA*C!G~gahtut`8H7m7&LCH?ZE#ZEX%67_*o(fF?!HCnfMnIi%Q)T5Tw~
zEe7(RdQ|PtXD=@SB9i;SqvW=YEGl8)vLom~VSy{e(lPf9o5ioG^Fzp`Abiv#h|>eQ
z?=uOAdN74x_`7$v(DP$8NP$l^DTm4UX!SyFMUh!A10c@QjwOgBV|eY-;JqEt{*e)B
z4hwzYoZ;7y#X?<A&B2557r46T)=uDO02?_7%d;R+GB7s1VD*YD20z_DxIjqgWUUZ=
zUmSuLHEe+%wfY$FJxYX)jLYH&_@OL#bSw<Q?!dW=;O8KW>IIC$tX1^gqg8NcIGHi8
z@{^`M2%_)9S{pg>JXk8}g_!B=nKR$-;ob1P4N~t|L~P!l>-y^rLTe9+g!>!V(SWF`
zd@_jt>Xj?MmXpEK0EP&oZfb0lJlXHBd3)2mWN3s+QX!5n6&?W!CS@JW5JD5(l~sfs
z%3b*nn(-|f58+R8f=2Oc{$n0Hx=w7nqQX8|zxLI8iu?7lJ&*R$q2Nb2T{lg3CVx*(
zCPy+O+Gcy}Q`mG!iPZ=-noE1FBo!=^83gRqrcI$VxP*jJPe_){R{_YV&HWL_jrYR3
z){K1GcxTb>bOlz26J4i~RKu1?dM2hE<z;ICuCDOTG9&uw@;AYBX8<^(Sw=S6xJP{I
zLeHEzW3teP*`K65(VWPMLbCEn+>;VE)V8-UbanTNoY#Im?Kb$32OtWo;_mj$*~G-;
z%~Ct%ZM}?DQl7Z><_p;Dy7N73{tV?6L(9&t0vqrL?DJyBC9S2O#WLp!$bW!0Ff1AB
z&3VLLY(1uT`aD5T;)9o}d3seBw+CwwAf2aSUweNCPzVN+3bT++fsCsLm^!R2`eO|b
zw$&PgZniA8+`~KC>p;{KggiJr_U%1jv+;QzmIAy1Xe}`}>b9Z==tx>n2rw!Zid4El
zpM!mSjDC$COw#;lRfR7;SsLUH896vq4|Y~+ew!OdG*?zf+=jg>gG4fQu1~dW!EH?8
ziJAe6W4{PyD3Fv>T1ZG}`Rp3jR4t|hmit+Q)L;`8sWMS!V6P(XJ76_56}3r>Ch{-9
zXpmZf8D1724kG9M2(iAuh4E+~3Uw+5D``NK^bre`A$uVcO3b2@5o=Z6^y}dtE%%ux
zf2u*M72e)BfUet*R+dL9w6A0D{rBQWDb4}jPD2hSDfg$M^dX}H=|e-<;q2_ifKnhJ
zl!qj46g40+)rpA-Vp{B_0z`HI1yPAzhDJv7O|%I;VG2TzjQA-jDPe9gG0IqkQcQ?o
z?L6x7oA2MHEXS&K;p;#1VO6uo3;tkbFKO;)-!Fl0c%bjOm#UtlnFL5-4j{nWfB;np
zq&RJVOJNPn@$s>`B-ZW=#5{>fGj;g83<m6nRD*viUGOf7*p2+>r&bKI0e}92{6C>M
tHrn~~BXAX)i2L*Nzx?$7{SOl#3Df+f+%O1zC4l`}w7C4kT+zqg{{!&i0*3$q

delta 108233
zcmY&=cRZHg|33*uNXVu#BZTagoxS(U$lfE{RmpZ!WbYNS_a0ey_TGDwy*IzZyU+LW
z`2EEny6@{c=XGAM=Q`!Lq87EF7OwC>eGNLV-Xmg*;1(s*(!EE*ZV_S{$iUHjN1Ni)
zCnV&y!xwYm^<XXM;kPbI%9bmw&axKwWB!Yom`dLS*3}<@Tt(+j(hQ_~#JmYBNSTqB
z;-Y64iSaoni5*Y8j!1|jp{{m{$@b26I;ZK@Ix;>#>Gw-tId06|ZbqlPn9^H)p&7%a
zx@j)~m6Hpt2hYDgzM;g>1wVuy>?N&^)??Wzj0W&lB>3ti4e-b<Rc#@Zn$A`fOAdB!
zSz*8EwN7iUMAqg%L_*A=PNQeM-`J6j$cRy+Z@nmHXNRaEG2fy)A+D?XL~!G$o!6+z
zBiD-ul6KfJ5*%5RSyUtCA;Ydx(2czMH~7F=_;}5kIIgH*8e41H*|S_7qExoB7fckN
z(|5yL#8FC);Z>XJ7Z^;nR}9lAyo%Eqv(7UZTJO!Q<eG*laF^2cE%Rq3Hy2Vp3>dcs
zEi5Gs1;!yY@IF}*>A&f3U+4Mt0rv@EbE1q?RplzZM>TJNq{_8$t~JytTh$H=w_s*x
zPLqieRFVI1R#iEg<`vHV)hE%ow<$#=VHTOz<qh$>IKP#JwQ0=jGN;e^mEU&5H#qCg
zbXahj$9QW9t?Q{^e|zf0)tYsFFtGB)U}Zj9?(o5hR_E@sUwcJ#;(|$m6Et9mQ-D+J
z#5fvKuoM&%`UKA_Ks*V8Zm#ilU~h;hxQ~fi>lp;Fiy`bh-?|*NMEYx~mtDQ7h6M`I
zu@PP);2pkZ@#Ebr(cQg|2*cp7K?EZ9jko4QTtZkLT6^~cn*xRl(&*{lpDq>*`)?t%
z@o5eG3F2<RiO`hEgyS>Ml_hYFSn2($LQfts&7R$}6RU@yr#tLv{doA;p7U~DuW||Z
z|Kyz1T(aQd<FP>J&DvGA_cGl`368G0VvnG_t`)`~xyswuhiK<@)yR3-covae^Sv;E
zsnuUOv(YzoFfqdxADx5tca^I1r}0CkUn;y1F0bmJfJYbUe*S-jm8qERveI49XBzxK
z8#>@-e(v5uo=Y2lZOD$D$+58Zw^dlKR>jFj_3ah4)iw9~+|HJry{!Yu7}zv47^^<u
zKEVCAlmzpWSlaJ_YGaaLW%_3{@dB<0JaPgkveSN5dG1vUe@h9(8q_M1M+k}z2i5#B
zIXna=Py^!BgvEj(@1|PmV+JBYU1NhPLkQh|aQviOc<BT6z)i|()@-^Ap6cBSd%p9&
ztsqc4$#y|Kyd0j))eA3<J@~$U^F&v0Y;}gc{3ElvyPwHfvn4Do-5pakhxID=aB{J<
z*pthGH!xMcq1o`w-)l=B0~I<pzN9NWRtvRa7DcYgt0=l(#g_w0PsR>AJ)j{X22X0M
zw-yE(DK4XJb*5~We1Snhr$>{{zp^@FEiOCB`@udB7%&wbH6K&XJg7C1HXm{=_Z<43
zzWa>m%xR&@41Gs9<E}br&34i5p4hIlv7hO+*$P(HC!Cz0g=bfh+IxDEvYK79-nc$P
zYv}iF;yTX>F)BY|LOR^52nf#*Bt!(2H%&H4rcdWCu7)gzeodI)+@PWDKj3qJ6rm{`
z*H6A4A#7nR(>Y&)pkCrA`ynzj3Te`dnu#Vzj4!cyA|pkYD;C2AiPZxC-mwiO*YoNy
z`rL6))<1h^->vc+{*erMKPgViwR^?+qSq?3^4Up&U90Gs2&6X!yD6p_<ISX6w~I2H
zc73+vRfm*UWX3l0`S#dui(qQt+`io7g&V@x=dOR*&D-VBbyxSR7D;r@coD>?pAy7P
zhDF8r=Vu|%j!&eFehZNHa3VXFU7lo|^BSvhB@9e$&a)%WCD~X{$P|Rbxr)ru_WC(!
z>F65&{HcVV<m5)by<tE&b#ztSArWg^8L~Qg9d*pKvh>~R5y_R%hlrE6uh~l8Pt=xO
zR_&pxGPx#!v;RP#Dt{?9gCa|L=>yg0RD}2XR`6!IhUQ(ko^alXvdQ}GBUhsTZ$I{!
zV6o0i;UC1g$V-zxArW)omnJQp(%LJgK}E%Ojjd_?R@In4wQj_-lp4&OqO>}y8tV`$
z&yJaCLyM|NLuRZbYO((eqkldv|Eo&4tP-6{tKHy;1`c3|(<I@K&glxK+b2HHP>N8V
z_7=@hWi{QTbi;`G(_d%taDTR=>|1y`6+DF&4O4b|$Z9C?n9p}G?}mcSxDy+?t(U2o
zH!eQ6tNoSMTQVb6)UUpDtq=cGYU&Gn2ZzGa(({eo;#fK==BEhgHQXqxEENY|i9d@>
zS1Ma!%uH=QLfdyJva|Xk`C*LFVdY~YLOOM;s+bwpD<uc1^AMNVdWuApxBcs%!cYot
zzk}@6z#QJ)j;6-HiG@&?7aiKwHtEx_J$d|O`w8Qv^a2&nP8@hLX`jP%ZukB7QL=~7
z-M`3k%J;c%=P`@YgS<t-yiGkGW2QdI9qDn^(D|(T(4D}!Kx|mYxQeP(eE-6E=LyL!
z4WU^(E<Rq0;^($^L1|K?U<7oh^)`eSWmgFv;NmV78AI(PAv?2g6*px91JUGhkzwDT
z(i6LDI*~M{F?IcJ6~3u6A$R;3VvS&wDMlE-@e#AvtsZ4{AgU?@L(@>HcraD!*pDr_
z-~GaC(}s`Yb}{`?6E0&_?<$r*=n=a9_R@1{_f#Ze==z7nbDH*dQet^!&!}ETO8pi;
z-I|0nYFS32L>&@;?def5{pp-v92MWaEUYFOJ7&$6umcaD%2py|WhAMfj|hhGc&R=A
zqWk#%!RX*&;zm>}E5YobFN}T#+x>vf$LA~yCfygSp`@++PSc(j)%mh1U4g$Vf%j`|
ze6D2eAgcJm$2tUm48HgBiC?B1!<yt$6rpUhv0Py)<)7pclA?8Od9+>K&(2oNyzQsG
zsE>#$@`DHNrp>Q}-2XPMG88OV<7i3JGXW-4DO4Mei+ag~r_F^Qj|DSElIa;K+5Yl!
z^jfYAd~V&Gt^P6y|H!-;)cwYMrl`7;iHh-@dyu-EhMod1LQ=+C7AI>K+OyLpauT;N
zco--IkOG(mmG&-#MJX5Kmz}HeT=j9isw7M&<}+Wu@EY?^M{J}}b@Pq3@tF->Y4FBk
zG!6|sRHh5uF9~S)__Yuxr_$>ZUzU8by+ru$?t0*&%aD~~L}*GIA?J!{B$G>cWhZL8
zY(PtsZbO}m&HX&Ve!O{TK>^Fqa8NS8*ns<x;CI9UqBc+Y6V#n@w`E$!E{e+~ZiERZ
za@CmShqrsHOsw%AoUltGnQi#oJ3}PprHh&8+iQF%9%aTqCu6BioTg}jAa7C1=~9t$
zl!ZlbAwj^elSs+H72>S7CZnFEPmfQp<z{|QiOft;Rg4ZQ-|i+jyVh{yy+BX&@|KFN
zL7>C?sVY0PbIR%cb35(k$5yoZ7>?Z*i4vLBXIbYRPaP+2QXAfpz>SnlX7-Yl9eU`s
z%U?ty$JkHV5eIsDqZ}+njcMa=s|D)5D3EJ6_LLxxYnavY*E_-Vkafm`VKQLDk*XJj
zdM=r@3P>Q)v+?^jX9%B|Z{ieUme(_ScW?A94aC+`cG{(@Q`ADfc?SKF2ujg8(dVpk
zQ?d8CJGsNcM74Lnoc!)C4D#T+c!=P}f5Vd}-XmyeWIsw^=%7_CP!_LBF_ZR8AS!g0
zK*HE<|BALkd&Vu50@;;|^TP4NSg_%k>l>)SBsg>Sv2QP0Q&$_TGHatNiT^l}S+6(o
ztz*jv>AZMCe)`7+subH*s5yCYOu{B4tz~vhXa@(;>_whcJ2~YpSIx6;GZy~#<7s1^
zmKppGY?@EH7MIn0^E8tE5L&CE%Px8W#u@?hHHM=oGRf(-eju7X9d45C6=x<#uMm_s
z@_K$bG$%{)?(AO3{ojiP7W|CNog&>MJL2J0y?z@JIMO_urh8#WNh4Bpof+F7JyCCr
z{jg|0deW`$WEp1Ko!d!jv8<d_UtlSg!M@P)yIaK1Z^q72`_KvVrius@{A?75!It8+
z@M<Y_<Y7!H8M-@D7=)8HUb<gm4<AA4$4;R_4MP-<6zCqN5Qw19uY87o^Fi*hmpNbZ
zy@=TO+(WwD(#tl!3q~1D$LCxNxbF2t(zhUH!9eJ^_+%qxEfCG+9ER+2ajIf>2M8XY
zg#K&nzs4faX*eA$@gqZ=%$P+&3#xOJ9bJ~Bf}!B<v7(R6gxSj$<|-M(0=1}bC;Q1G
zEW_mmWskV~D&Y>aflaAga?^HG3s{MQ3b;fc-O)^v^6?VkEd!rF8m=N1T91q4g+Sl(
zG+0nSQP)wY&~k_5y7*qrhEOw5r+4iIrY;)ql3T`1Ms6TB7+9|gCPX&>>5={bpRn0B
zS%;j{+1T~ZSBbMnv+(THF3%PLRyo%DlAo2_wa1m+Pjch*q^5yL#7mYgEit<7o49UW
zuL}CzS!CcclDTcOsez=k6=a`&dS*^JX!-ImS2$&Q!S}SeE-7e2Ei*wqs~Gs;;gH(-
z<OVBu<?5LBnSGgYjm3%;Tl>9J33;vexTEWFk;}7uoO@HZnmu}rEt{Vt`%#i|UhPZv
zWLCsX6&Y@T9Y>WL^J99rn<Lo`hSMah6<WXlLe}QR!Xfo0@0IP|dSFZ*Pj!<=aaeOe
zeWhP**Zt(B$2Ez)1uq0xzyB%G;;M&`43iQyx}8;;MrM3Mqe$m#LiHM<Rmg@!>+GJO
zMdrdxslZ4NKPaR2ZhsoCOUj4(h{?3kF<N_@r9Q*w4ivY2SV83<ol$7*LEc76mhIF>
zS<W7`=TyVI`7_>FH$0IwQD^p0zBp-nt9kxVe_0(%>%=&{ReNdw41wsv=^TFKZ?Q`I
z>JNNk9F=&q&8-(oQ_RjDG@wmEZ7?}$qvE|eu5ep*lb`bHbuA}Bp~JJ5Ie)zpZlOI<
z;ZH9f`JKoqaVYEH#$O?OsPfHb$ZWFNkeR|06YH(O3~vtol+1Q%xwekB6mIKGTqRa?
z-J}b8VR&8B`7${UNnxD-Ah8sz#KAEQo(w0vT5`<YivI<rDBc~sUo`*G1An?m0%x2u
zv7j;~B=7yP>a%;_NXRt^Kddusuc#ncPR9|&caC&Nl|~1B)qBV|JcohKr(||8Hs$uw
zu3sNz)$K85>-7D`k*ysQ6C=xz*}3&oj6_+tsSg7NLVsjJMXij!RORu2%6@NfR)CT(
z=>x&ym^9k4DdFP7(I`7suuE=chu=^4?Tahz=kK=%^e-8kytml-VnIs&m(4OjlA&hc
zb`Yd~3`Z%r1v4Ueinu<xUCPq0NZw|!&_)&ckRrsazl>0bonc{>T;X7^VB;L!a+j3#
zXLMqqV*WX>!pzJR0|wymoHo{%+^#U_;!Ra^Nr|16_O?nt#8=Z5d$B3q&0yA_V&fEb
zyD|?H82-2CKFV}>fHfPXQ>Cp1VoVG?V=7s)Ws5Vv{&lu>c2FGlsK*I#|LIyO9f#_p
z*?ktN-ZWb7Fbdglf~Rj~H6K-{-s`fQE_AT*wOG_OE&chSC#`oBI$TU^=7jbU9u{F6
z{mfjL+w{#5N78iq8=6t(>2olR1-|etRh^hdz|9{G-9TCRaL{LMjE66%m{D}>umLb~
z;jk1e8R2_9Kt!w^ce^BCzx`X92nV|YaW~0(W<^nj2d4SXpmH1#IBQ1p(mM_pH5y)$
zjER^pL0rvGUUul63>CWiDMk^qTVPF9+lV4Y1THl3GLoM92G!ty3xBF)tVw}`?!he}
z*g3yrV7-X)wj<A#x0X1f<c$vwDti%-2c{~YXs-~I0jiSk?dmb0{dc&~ds2I90}-07
zsK*r*2c${aoGm0@cR8@Vv)lOMuaBkH5t|wGa_>V3GiycaY?cYZ#gOzk@g3=FdixKg
zW*OvhV=VzLKgC-zo{VDC64Q-Py3Lz)tAQjlhpRViQlv6&rdPf$7&0H<mxR^7H8vn<
zqc(QAD5Z6p&XX0E?t1k{7WM*QDrIwwiv~g&;{go$Y+LI-!V_bpLPuy<ERUI*hQar%
zg}6WFxbSUH`nYLQ&AW$mGy$XdJM|I4H`x6s(dk1v)n03t_zibf*F*-klO|?!TLyNU
zF*_ixVIBu>zME9o?c25sP|(!2x2UbA^bZAxjxz8Zi)SceyMCnS*1<L}^Kzo{MgA6t
z9cq(f9j&E4>u@<71bz14wtCe5Pyr&HSeYp$e|oH&#U{3tE{5aWnlvVLF?N2TwmDrJ
z+`dzgtd!L6Hca#N)P~dWz|bkGkv1`J`w0N(A?9r~@^`0LGZ%0e^_oK@p<{-0vxb+Z
zGI@0I6cpEJ^E9LS#`-cFjKh<2PPWI5J^iqubaf#tFG#NBx?nwGbu%Os(R8Bvf)rkA
zIU5BCS-kocQ9E|#OS(V5fgm&?z<Yz)y>ABCmYwy5_}5Jx|C>;pD(p(%+FVM|M{M`9
zoJV7yEL!hxAJqAWR(%FoVGHgVVsh~N!*-Ik_K)6WVFBBf(38`}Q6_&htPgFgaeV}4
z^oJ11uP?AVR)&d_Wy?9;d{Q03wC!^?>K-d5S6DTM%Tl5t(pAOo*`Xco)0M?xbIPqY
zFU=MH1~2c-@HjH@-4Uz@e5a&I^=!VGGwO~0aJ|tSvAdPH;pS4jBLS&_!`ZVx$h^sD
zm{`9jmnqSc6wos+t##nbJGKUU?L&F?dFi?4y|P*l-tI=8{3o6KahL?JMiimh{ez2s
zhjV**yi-1htwoPdQb7<CfGqvSAEeGY53S7C-}#Bxr_U}~fra+xt-$z{si$W<8oO^_
za<&Bf-U6R$%sHa<?M7=EwE07d1s^1c=_}na@@>HJFO#B>B^~b>fX>1@?M_bMhD=@v
zjoC0(fRA%+Jx<%5L=q=M(sJ;x#HZ^7NxYU{p&Rpy>Fe`@MU@3OX8Pzc2P?4}x7=wY
zKa*9<pRwR&^qFV7edKug3b!*^=%0A9P#=BzmlteM<z~ZI2cFf4&6ir30DOkzd3*+F
zeD4_AGaNHK&x_2n)Hx5y_-vzv2K1O3Th8%ph2;>LN9c2XVd97c+}kRaq}tn@BOPt?
zBTmQ-f`g`P9>=7_`8O-;9MQT=xhN56K;-I_z1rdS>JB@pvZEpx72{jGY9azyYdID^
zk?|Z!?*>OnO8k0im(L_*i%v~j#bPRz6+#jdkkd7O;f@b`#*wypW!w40`Rth=<n|;&
zvK4)>LD0fu)r!-x+Mqs`@B3vfiDu~erxm1`S0@%r(<{%b1+G{T&{ugyRmIG!EO752
zG1b3zZAZ>oJ*0nbLfmIA!&#<}a^0EujEQwVpP6t^ZTxEaZ9bV!NLd1e!}|2{b9PLt
z=daTle%+7vEwsc37KO&SD@ebumd}>i<$-D$s2ANlEv4~s)i^ER>|j-DsNNv??05YP
zG+`imO@v&hH?g^$W2BfRsVO*ek6d$}ge->Pq>cfAN|g7KFYyAS?7N-MN|9r()M~&i
z;%?^xAuT9(wX)Tp5JRd)99g;uN(qpcTE1@9ZJYuXwPz=`Hwpb*N%>pn*C>IX0Q@_{
z_GXWGut$3ql^ZDXe0zS`!ki}gq?_~Qz9)R|_!*I;!l`@5c^hQ^hoB_}N2bqg*Pnx(
zt0LQ_-<Y*-+|$#0@%K`eCEQy6Tez6mrAiYdm$%wCuxA-N{bQ^T$%W0d$3!5hvGr>E
z0zPBY*7nom1nx9dOB5D|-8<vcJjP!@43rLPw!b8<8WKzC@GydpU*@bF#t*HP?y1U%
zscFB3>!zLBX5a9wj+Im`y-ygSSV>m2IQ&(zT&Q6hEX`SIMJ4Kk{2_#^^ne2mecuck
zuz2$HW1?HPQbjkx=0@p4Tg3Z(J?;#(kPhQ?wv}lgo^|#)g<_=_XAXkU$DSQg(4Zw?
z+;U#3_U-eIqoH=jofnqz!(YQUW`&TYW6zM%ZcdKDmjm4i!Ah)~;iO|4vf$hIRHkFQ
zsCXsbzkb+pH`M{#_sa!P0s`ub_^mjUqhg|Qu!y=jvx@|<7{%e7^M?>N)(V3moI$`{
zU_@p1CpWAWg~og5@?JWZ>{@D`GEGN@l-dFx;d@V2Kk^=|L>-bHFY!6xFRj`$+LCMH
zNbzbt>aP(~qNkbixOnX3MT6gaI#?{+plB#d#yFl?q3b7pln9|l1iEd(zds?A?wij$
zePfjg<x3hFnD-EqU&;C<QUHlUqeX{bWNKO1nPj%Q^BHyh{=*}pvTBoU0H+y|$zQs*
z$(*n0r1{k0_weiZH@JPxc`gX}zQT_7-Ep{K`!`7|do4E;T(9$5c`pxn;(2RfC7<0T
z0-;b!pau~0(120h|8wL0gd<^aRfddHl)IdHSu71T)RxBN)tE$FzuI@K_s$>ojERG|
zOwuR2ib@wFmi2uXy5suAYjdZwdXcC2q+<GlWB{-xGqW&fiQy+%1MOZyS^ZrM#>}6>
zpe$IhU9N??(ny9r2SRF-(4!8eab-FPZyJ?fEUMnH%J%%R;I&o&8v1j#mrp4SaW)gC
z%r-}3F!{vihUa=v$F8)bTRwnB5Krvb@9uMC#mGN>`?lbd{|6Ni{>gpHI6UGU=K_Z$
zSLAN_)l%8GppG9wn{2*n1A?|@5_SmDlhuUron0fOpMF5;V~V4o5ZgQ6wKmP5M|2dD
zc6IHr$jKx#4-A0KVu_nNe<zb)JGypUB2{-(eMf!#_rL##v09+7sk8v1eoGAd6qCWm
zUM`!}Q59RxO<*R4$xY8l=RP|X*oU`_l+4Ch_e9;$=&NL&n40A&Z>x`UVkUdFXHo>T
zM}#-K%tnVLnn}UO@O>)xktK=tYt{;(jMc=(yA8y8^6#~V3<6UhlN1u)BsAEakIAwO
zgk4vIGxAd-F41wDHFl;sv1P0!mx!LBdnVgR)I?RW@^_W>YN4|ulaox|2td(7=Jq6_
z$Xo=3aT@}1`+2VMlD$e0@w8WB#2L{zS0uz{3&a#~N@(WhsD*k+uHnz!Mn1}xZgGE2
z6sGP=_0%&vzLmFAN5cLgOa?r{r%hM~)OLHN0tj?vD1gyKYjP<?M+pKDW+JyazZt-n
zt96Wh<8H-2jY44iN1X~_Qop=9bbu`WbV@c24M3ny`=8e!nogxU#EcSGtVuUs!)|RW
zz*3KrRu}SIFFO)5%lC3ugJOKx_C`bGqmz6_S`-^Z0)-A+1rSXKQl+YdFt<)4B7%N~
zhY=us8tEc(SJNx~sk&rL`7OpJ0jT6Fjpa#u7Nc?aGuKZ;t7Q&z<nDMY3Lp{S`~w!j
zfV<QXfxN>jt84v~Js)Pqbw~;<j*z0NEj>fWRv$3=h5pPGR1sdH>UDVp=iqz0*msy@
zVer1rVaXaYpEx1tN$eJ|Kl&qgis2?zzzg2)T?i+v#C@8F?Ab`l>N_CGFYTvLxpBCl
zACx>m*G7JOL3Yj&Ku#$JX&O#^ByIXWom1syNhRafkgk?kNq=obsogvY$1*vob|%%*
z)^6EHwkh=Net}Rs8T`|1ww;q>Gf?yDM9*fwM5&Un*2jT|f90|rN!q=bLP<bFCXUhj
z=+_+%Nd5V%DlY#{J9|*4&FS<_E)z!0#1RzPY|Q-E8~x21#Jfk{t%CpvpH}$Hq2~uy
zB;Cn?qACV0FRjLcAQM%k<qQn>OEv)ASDkw(I4J6yfS^EOyQPdECKK%jFIacp5Wqw=
zkprdz1-oN|Zdu|ib7*bweSqC?#iO|0mR%mh;&)LHeow9_ehb%}chRwX4(!N?rC83R
z#6>j^h%I8v;;FuYes3ekao=hNMkV|9S?CJmrvM*5jGF6EKmJO8z#_pJ+ub9kc*y5o
z+QfF}u?!jg@7Rk1E9Z1GuLu_(zu)zeqm+=XLU}Rg6<Ieb^Wjz+L7Ya3*|N!fk}*f}
zUv(S>=iO@@`-EWK#&5O=0@0xZI$X$nQF?)qyrsE@8FY;8kj2urpVBVNh6tF9uEDuW
z{%GAe0K%I%djyY+{eaVS=gDt+_hg#6S7UhuPaSe%dfYsyUTE4!rAPamqet_S*WV4M
zHR!N=T7xN2?2a+1>In;|A_dA>bDdd(yKKscsTqec87bLW*g~ohv?4<)W9(*s_8)r5
zn*Sf!3<3w|dCYqIo6U1&p9V&ojhP?Yy_feId-dcl;Gb27*JC&uQgs(cD7ASy@_R4B
z;>etP3BckaHq<qpB_tUbTenUo*)+Yg@EDCw^rq<p=GvdmckeH1^C^9jUa5Jig<iH?
z+rXhusjttjmMe$%UaDzZ?JsogfTJcQ{fF!$#N~gP>&_t`6tvG(yg&%ZAPPM(DzKAm
zNAm<h$2@h_Z)YdvOS&Ni#~$xj5nz;W*4`u@4&7&eb(1bsbl1Gt{^x9%2jm9hezZCz
z13Ypfq`%<v^04{+ZJ7%Q8{a9x(0vpro9E2TY$hXWlcsZ%kzY69W2gB{tJ{SigaPK}
zxTjWP_VJ*B$giY+Cd7nq*1ZSlCN5x@9g?to22N+W=Tv|pz1}mvZTsw^!2IaSPQS4u
zQ2`)?q2QAC%H8}FBUN-3h|yAKb)T25C3-`{?kGTP%(l`5Z3EF?l8&wXwLp*sT(#IN
zl1VL;5kB%e!Wn<Nv9)eSXLWbBEHJ$?&gC2fFrnl?{A|U3>#j`DMT-o#4D5?p=@Nzi
zYbRmCBe#OVbMt^fac4zf0yKv#GM9Yp=Lxmhi3)f)-Weba^(~U=#Wd)F;dW-4EzpS%
z)&Ab@mf`~VG?TrjWu}BG5_871lzJ(>^&76&et{K9Y}tBy&JI(9$>Qz}>i`k!haO1}
zkIB7@laeX`={v8b3hm85!7M0(yDi(hrJu3wZ|ZeRsr91N9`-!BoH)r2bBE+Pt6W)a
z&X~R81!8VJL(?r~*?f0&_kt1iF=$jUkDbmH_Chw5o=$^w26&yiF?EknO?ur|uy`v8
z5oY-d0J^A3K-J9?!+u&BIQMDVah*r?AG%=k`6qCi2Lzd1BJP^<k0tYhZ3QCy1FkJS
z!VfA$I7$w>@n?L^WN}U9!|1>c6OuG~{qp;72*%7-kq!x;Fdvd1y#g%1b}_}=8Eh-T
z@ffb#BeLFOwx9=N4d3HNnLr!ADquD*^IfrW?>FpQJS4mpcq29;Sp`Q;h^fLWsIQ2h
z4{hi<9!BSKbs_$*`yQPTpj%8FlC=USRypgX*IuxIIfda=>>+!Wm|Vl}ugLie3)#X0
z(4DO!lK1SN{H6JV>p2mE#VQKQGBXhJyV@d3f0zaap;>4j43s1?G4-5=fGHB0e)>)^
z+~6e;@@CpeY19hbx68D2fmFkL+(bg=u8&EW9nb<7`<58{wM2_FUb_xV^j|+*&C{we
zw08Gv1X9-=DO<(E!U%*RJG*|p0tL}oCFHXzBa@%{%gW@ik)h;?T<a^3bjntOAL4h#
z_49wK4#W4PIE)`}OffRpie9zHM0r<bz<!_o`XZUp)IN+39)&KcbE4;=13eEK8ro}C
z)<KI8yU=&ZytejkVj2z`@BJl}-|jq2G@WyI_z-%`_W6HX_)yC)%+uLnaIw(N%0NCJ
zv^LNPqczmk#LOZXxp{wdH+b|cJHXq!-T|N=>=VI2<`nnZWva7uJs195JGHI2xLiDM
z<Xaq>iEVzo7S|?;HbxKCorH$-kB+8B+MZZHdb<F7t4(|Y5-o01&<8*`vnLBhVOrU4
zXC4T^cz9>IRHirz;Eg}5wynK+@^wYwO2Eq`X4^ybVfl<KOS&>5jsjmG?uW%VBGJS5
z$>UFK0ytz5=C-G^iaYMi>MYg&4F{?TC*qEY4D!cb_M~1&nV%E8$9I)Br1tMTB|^((
zsHtKMI9;@XUx3kjD=eptBfKK5Rj>$dmk4@(J%QciJrefa7Pcec^&Lq|ZFhb^n1E2F
zr@l7zymW*8$J%aR*vUz9#i3Lx?boE*qw8(el9|~LTz3M^vySJZoTwCePpvQ}3$!0R
zZ-M-(tI_W?y2laC@CwqzBM(RdO+aGlj1<^iAkMI`jhv-S=k|z5uO092PL4B8vDvZ*
zu4u~HCk|H^T>_mXb)o5UL%ZBbvI}m8y+dN@Nk9ID6fCCxqjvKC*m1lt;s0d`H0{7*
zGQrG&EuU3wBS?@1e0c=z(*p;z1D2x>Y&&C|Y=zZb>oMrwjxPgU3Gba~VoveDQy>I9
z?CXqB1kUTupe5rLwO5~0Z7r7%mP|7HEY@&_vS&-te&tqL<>1?!t`=ksi5(`ECLS;H
z!^ftMLx4;}LF5Ga>hHuiHq~PS?cVh!0`a8CKmiY=2P7&GLqS~N51^v=M6=}~51@>K
znRTzRB}yRQZZh<?+J)An{*l<ScYsf>-7&zlv4f+>_5+5Uc$tHE*oEdS9A8Fm!z$t7
z0yztkote*={E`}C(cZ4^1!Yful&&(gn)X;f`KuZn+<mLBW`^$k`o+@fxqnEEUg&JU
z>eg6rKR+%IN~LF1o4UDiUE@1^IN?aD*cx#MFQx?kS3#mc^)2Yq6$^i=13+E(NYz#)
zcIk0`4yjUFPl@|)1ja4B+q1xmfB6|!ci@?O-cx^ZDhk|eJBtCoHon9H+gi_r$WjvS
z9LiBYyik0$poBDH*7{Q<R&)!-ZAiCqy8z|;Q-{5fMLbb)^*Oq7ud&1*e@fn140=oK
zTt3TjzD))w<otPVaW_%Rf&;R-9O@@KE6DkJQ%D=^cm2dU58&=D6|N6}Z5pk?qx3CK
z5K#Nn9vhmK0e*R0-i<SsgCGIh19SE1irrgn522K3FLB1Njctnq0@zW2>Lq_z>5>Z&
z3g}8i?2^vjSivuo<Zc1i?bKR1(VPRyWTKaFq5bjiv98#uU>D$BKe2Pe0a6OW8qdN@
zU6%`<(l+X057fJ7jQGFp`&JY{sI-!s4yy06*aZbR6kHXH;Vf{?NiOD<Y{Exm5YNNB
zqIJ*w!=A7OiuQI%_)rHSep0^M>TUNOw0R9SxzO`yG#ENtl+0CBMg{Qw)L!K^&xluZ
z8{UC`eP!HQP)#Sv#$YKQj^5eVr=;gIt8{l+ifyd_@g_yOh(Of@ls+)qQYjc6{;pvz
zq~?bM%2wsW24ONLo|bLMgh%T1o0`wSo9I{XA=vJk!zA;VX`{TIwXE;w&95BsKnZ%h
z>cKKy1Z;M6&E#=H?{={n<GJ5S3ccYPCmfronNx-PmFxHoW#QdLPadnJtaygu|F;@F
z09caEz;bGW(E?!AD4(*+<}VQAG@>j!zr{i%HnxlkC8oip;3;5wH8aCHfp?Uo>|ewG
zn)TgsGaiAqtMd(~n!YXy!b+qnUWf{=Pvswr7Wmij?~%dbIW+l+7=;dwtd;qH!xCb|
zy1MDpso^PXENwmQ3EW2>{RCz?HE-^84+X)0b^Vx4aaxQ45&GlbU~Zcq{55R2Q8s^O
zIYc5b)aan&4B%xEdm-hRqa%VtPH%EALSQF(TjZK^>oEH{=NasfcRt{9AV|nRB_Hi(
zuri%=dH|Gf38dNbVdadw4wv=6FqA)0wpVyLctK!;QPOkWlk(f=Fu4&_@COpxCw{2;
z#(8I+o74h|@ASn>0%1g~PlnSU)ob1~jdl6fc8LK01RKbo^J+cC%N0y9Zw}nK4@t81
zJWm8qhaEiF++Sd;^&-AEU-wt??unkPOrd9d#`-cI;788ORXSk&29kNh4;F%VUb7aQ
z2AbThfgL3AFIqPA4gxBN8%)!Vfzlu3%GO_Rqu1?==!{8-HL-!+-nAOu$~Q#ccR2HC
zqqu)&BMRb?HoK0o$8le(x*N(3<#83$dT%p@^O++D?Qv<*yZi89S94!oHr0%0H+dd=
zJ%&rG#*4}YNO|Jjt<x1Y+1?wV0atzWpfaiGX@yp8ps7ldt{Qabxuw4<G7<mO1$i}R
z#JkcFnrcsv5%?t8n&BBF5mco<f0}w}`DH`~>HVL|s2y%q1>C3IuCd^o)Ojvm8yuTW
zz+%*@LA+2=e|7xJO!!E@`+ythe`iN=F`MJ1fI6vW<lrO42wds!JCvC5*<&_C#wc_B
zmT$8G;~E)4o1I}g=$m>B^r|O>AHRbXB^3vJm+#x%c@i1^CwhF%D*DS|CKGh**@uT;
z=-mIno^_#<GzaLUyK-`2PjSA0Ziu6Ib(p{CY@>P0Y>NsHjQ$)p12ZwwBC*;E-e<F~
z3d|S*k8_sqkpA=K{35#LW#fbOQf46X4_?ool@jq2EB|tDi%}nA1)<4Gw)O7r4Se<E
zvmY1od3V*-7xfRch`^3#nsmmvs>5Ok+D5P@!*R(Rd!s>zy5%psO2i=Kcxp7RF!YxP
zI_&OjH#j=kLG0UTEsGwnRLv5rpoTn6S@)3uz&qQTbf&}AQYkOceP=I(3(#(#M`Ijb
zpo&H5p_?58_0!|qJPiP&$_f#8d($NUe<vw+;*WoUY7)FZ19SZ4cgw3RifaMcXY7Xv
zfEG5O(xGMd<8#)TJ>4)leN0?Uo>%nHp?+E_;ap~i8xGA+=SWBeWLi%aadNEXswH%v
zH-rF-lm^q9nx9*`fFwe+e|@a{uLTe=Z@rKFh?*N?;5!68XO+kG0Rvc7@W01NW%=KP
zNNp=AcUtnxn|~F3MgTu90He+ZWHT^eBaa-koZo2T*bf-c{*kD`OKF<SgphX_r#^d^
z!1)eP&ukEr6;4bJ^A!v&#jZj^9p-y`Z{ur)@Lx6--k@;6gTgcPoh`NF))L^Eo#FTM
zpde6uw5*O<+r?;_ccWVw>+CD_9StIT0M1tS!APOwq)hcOv0nrXsXCRH1*ahVDUOh!
zED<%e@f!<hG8_xyY*fqXjm)V+w>>g=Ocy}}a+r^P^#isHta*$M!>q}J*?e(5kjmk7
zDWe8Sp7hf{9kh4Z10iLWe~F;yWVDz)WYP&0D+L0XRPgYqD4!Cox|;10cbe8#(e%s~
zL*uPLF$;5fD`2r9q*JyL{D8|>tU?Vsq0e7!)`8J3W3Kp0)_MkI^}oF1{|&K)&&C^b
z_bS&&3)UO~PLwn}6*nvsL;vvxmJ=?k_UKdf)9LWLqNI)cuPF7uZA8iK4X)bQZYe(Y
zIXj~b#Eg*i(aifjGw%HytDz~2<|h%a90-(|gxo+9K=%Er$0a3c@W>p7;!fiX*j<19
z>VwOPBeA>@f;XWcYsf|Zb}mKW!9#u_@K-*_u4CMl3&Y8TTh$n8Iu$9=v06kPa%D0=
ze9a8mURs*Z6Vc|8Ggl-bqM+C$oi-b)rPy*0N!HjPW3wh#qsU`cWydtigoC(>0^u%1
z)*%5yX`E@VEAAK!%JZg2;4<?qpse?l1Ku*q>Mt~4IXEmdQZvT$i`aYM-VQtN%DI(w
z;Yn-ZwLS+*@L9t;Kkp`1#J~P<2zX6%mq`r{xhInRwVjLtE(p}V3x>_D`V5|K)i}HS
zE$h9_IpZaYE8s$yJLd_r^$DUvmM+;F0Y%u_{Lb+}5t<)P=8}`X{sCs{nv+~qFR4lE
z=pYEgmJXEx<I-B2zZqa31FO}RN|V_SePa@)boxF^-m>7VG*}_$)I*_5U$YcPaSWOs
z@;L`HG(Zv{%j{ymZvujq6-0D(tTyBVxAWToQbz!?`B4v!^YUW^AZe?qnz;cwf{b7n
zg*DHy#!z>FyJGSk*X{{GYw_9`v)d^wqJDnG_VbSSY)HSI++<L}Z^sk-#HMk`1MSxR
z%?cO<dDSdm$-A!kEAkR1a(giFo~||cxGIqg^{>lkvjL^#Gw6hK>QDtAsB2ujrEh>_
zWF7ESdYC>Of61Z24WWI$)^fdxn_f=L1Q{+Heh*9_{324Jim|P|K?LR+$k~W1ID%Lt
zHcoLAkS=Tw&L8r}v;0uN$z<6Zc3sx(&Y?xqf^Nd_^yGy5*Bd(@q3{UvXxXwMQsAXO
zZEBU~{2|@;9TK@!1;8PezB0my9Qo+R2cLKbLd6BMP}rv{V#M7;Vj<897ALZJjM=Se
zTSbjBrYNb6uz-9dqP}J}<DN~vm>35_opox4st4se%<<@%nD$pKJ3#Bi(_!ZLB6*Bb
zPNdjE7)MTD(t`GKTjgt!&&#5y<L|%+k16wlwA4l7Cjd+N<c6ea+cUo^nXd>v=aDUT
zJ+}|sM=LBmNLT1x&0gfdwJ;$MmVO1qninrGDX}xzZTaTZMqb4<3*IET99~2f?Y;J3
zv&olW49s7`*IHRw`FD<Af0b0gU7l%fu?kG>tM_!Daob>MgAuDuFRxi0WdDWolUG0^
zotn}PhQXLw!e(7BqwsKX<&~9-8<GP89umRy^$wn5m0w<ZBrnxHyI#L+zpCZPCJ#)V
zn@UEXPSX8cCe_Rn!o=9M*p!sya2`6IO#8;H-$%}iu1=_2S=$bqT$WJMw$Zw4i+63W
zOA-hWRSut0AA7kWiL3ES2}0J~p1wDsYQ{HTB{Qm5wR56sdt<2_-sbp9%)D+P9C8k*
ziNrX_|J|)!OI-ebra0}Uoq-srZ>~ETUfO8<Uos4J-1sjxbmG#0EW6pVLvpb!3i6nQ
zEZcaA?iT%?xu-n?iVLoGBA#gI=!94Lw5qa4^zn-mMG+*|>k_=0(8ywp6KgWbPTj9k
zi*56JYTIcNJ5g)A*KufYeEj8-_358R)V|Di%W*>+Q|Y&)&mk_f#C1Xra0vdmr{1$-
zhL*qqiwV#t(YVG10kR^;;`u6>l0y3;wk+Q%UbEf6hUJx&_vyo_Yy+y!FH_OF_gs(j
zM;+|#|FpLwHbNa8zr;L!FAoRh9zJ?x)a;9K3xHDthk~i@+OW&Br4{47YbhkKN#MY+
zm`+^Rusb8Z(NxmJj1vV;_^r|ej=0{_-pW;NdVE<L&i5SU*7c8)@ElJ7sF4%}g!wkZ
z=s?VIY@_o#gzdzd425J&d0>idxY0uoB3QA%{ku@)SC*LFdH$7dAs`IA&~VCgKrWEf
zl`%&#Hfq+o>wIsgYF{NoaWHxCzHc{Y2vDbaMi`A}Ba^n3%A8Grw^EQI2jaI&jVFt=
zjDPavJ>0Zu5gQurz1<lVQ*n2|Pri^6W-fnCb=X(|k}OrXImtae>hCcSfta^-x4E<f
z$BGB1*6bNF$SLkfYyfQ5qZ3zy0CtP;w2kbqvrBMpi{nV`Ax5i@{rmXaerx`Z-=t2h
z)DDp+<P)G4j2yZcQxT=@7s&3Jvwgt~eTDFTu%nljk+yPiEKVPPy!tOakO|i={yy!p
z>VPW4K?ngludRJpCf3O`+eOJhbscgCcI~d0!UUvk5M=T?_^?)pTnd;u<7_Qw-M%1o
zG#rUo*#fXUO&s;!d|3qC@c=)Rx7C3U?k1-hDw$z?rl!>1Bc0PE*~xUgf`<U`dyz@M
zEfCy0@S>g&b|s;s$VW$MgwZS+M+~cDLr*bPgHM+v5HJ}?<yzKZ%YlRIfe+N#VsgZz
z4ChU|ceS{FC8QQ^c&p&z;|^GFW)4~L{G3VwQeHLR5e^8FJ$7J>AGE+!(@(iWWheiD
zp#mc^a+P+y?Y&PuIQ4a_zp9E)k6n?$647##qR)^*a65}AuTQCiP`|+%4?xe#E@YYK
zq-IkG$N(9S*V<o`!ndvrHQH0&Cu}C?c>1jVZhMeO&*fr1d{RPBc7vdGbmfjvN_K|$
z^w^1xWM?Fu1Eh?f@lZrq<<#eI?$d25?qY+KJbd6cu@;A(0n?UYC#qE?*)DNlDn08@
zAU4nCkr!Tt=R-!BA`9aNv~&@%X#$(+uTOd!8Y`1xF|EaK)(t`UM*pV5kn{YY#3(b8
z|G`fAX^!7G^PLCZ*46e!9CAUn7>zB0AC@)a2cTg#_oQ(F{HQzc0vq`O-W$2`CeCUt
zFp$1R<d*pOn?ENGxpVNeRXx>;{qil+$RN()opd#TPhhjj6`Ykz%(CRODl5@nF15d0
zRu~u=e6~r7I?5jfqzC96T}7T7uJ#?5EGG+m)h1T$4ZMHzPk0Y)KP6e)9~nFYN{^lW
z(SX1x(hGWu2p|;S-oMJaV0ByFBfex;3(cP!n{ndZzjTjKOC}51)<J!HAP5cauD|)-
z+?fQToMM+y{M)_>4j^oODf>^+?v#GUweSX{(rM}b0HHdyTbO*n{*UxA#1RS7j1PG}
z2DEu<sc!I1kDQHv9&_a?`Ot=zPomp4oJMbJHiiL$vcT(2qwk_&<%*9Es*wn-dGsmA
zhLr5u_{m47js}(*=$FqX=fk%Lp9ke54iH))hh6I+_rYtB#F$@c&l%cMW-U*=19z{#
z{s-<97%SpZ06QT10dKZ>O;?8P%BAr^w*lAVD_FDIN#3vDi?03fHO5>Q=B9mH{0Cd6
zYQOk7FGAYAo$U()<7w~HkY~{>yLWBqBvlb?0wD^}uRzG>>T%@6TvN99nqw<48U*wD
zZ{(i?Uj^{s1>-SM5GJv+J<e*OqhC3GO03Hup3GMAvsa4PA7ff>cciLCZ*TQ)|8j-_
z;Cg0#X7}>MYEMQb9YAc`pR?}!h9A>VGz!z(aKya5?J<8)W-nZt6@cOOVtYRYQe2ED
zcI`WhK`bR0e;D!8*mHd6{CiNAvN^cIWiC&4`xy_%n~$3DGwzc2{+fdSNprMOAb)^h
z$rnrf13uP;B8AEX?&7_Z2>v~@o|I#PV?cpW=)hq~hQJfLLMa|k&CH&v?9IW&OeC7=
zT^&g!qbWSZU?xH{1O2Kim-5Fzl}s#1+2oB4$A)}bD*eqZcIg6zgrqq^!0u+ux+zbB
z91PCqO!NW3*!d#?OMOOzp=5!d8~=B4qR>4%PtsZ-{BUi5Yt~&LW?a2*Cy)8i^Wb@E
z@AR`h%~u)$>!!i5Li_U85~nMoGJ}(wK|H*0!X#gx1dwlt#6(%Hz9PzKr35nfT?-lh
zm!P4IYHAAO6P3ZR%twe;=lli;Calv$lFg-rMUu$Sm)*{~7wKD!ZuR%@2(KJO6P<_T
ze%8yh7KsxeDo}Iffnj&y1o~jTiJJO1ae&QLWk1*q>zl8uZzI)#$DJxIR?8Z_57EJh
zdPgvJEvJ*KrCwIl29qS)-!y!Si?y<4ygRe8Z|~Mygds9KqNjE6nzBvwTg3N3K&E=O
zoy<n6QGc#P;&o&U6DcW2%jrhW!zwQ~!$`p8%pcXpW;*s`8eUXAh98uwUmBS?Y{-`d
zrxtrJ9z3VS2}4hLi+ZdH?Xq}NM<!p7hRtemb;N>D!quj=!0e5jg5&MBv7Pf)t=j>C
z7aSvyJTi<yU77Yu$g|AZ88?H;^b^IY{&nx^DmM_&I|)x!^5-I#T^Keke!|*k%BEg5
zwlH3eSe_U%tfLzvohPN*uJ+f$$VC7Ddoc)@?~V*hlmPe00h9!YxRlN)QCwg7r2MUq
zqKnaS!saez!DH{hA`w*oNxTHnZB#WnPEd7U27yH9?~m&3%@vHd=3U5jtCb9F=P%yI
zpeAu~x+W{;D7>&2OfiZ>LFW8+-@@?OtIW-Gc1?o02ca&Pb>|!}DvT1f%0@pzJ^2+t
zVQ$b%Ynpn*xOx2r-1=;n2xVaH0V!wWD?NDTq*B%}RWBpM%CUvAXRmCA;{4`{47eLn
ztds`O=@8H8)=50f&+|isn%Vw<1821({_l;@Zwc`lD+H@;WR9-J&%7wON3!@_hQ)|-
z<E`2ygP#lVAmgB4CE5JVJAvY1lh7hQ%WdUlUHxbN4!!U}KD#_UGx}(weeV<Sj%Tb;
z^f&q+;W+T@7l+@bpALO=?r+Zi2|$V?xIW$BWj<4S6I%nlTAOlyb!*&rmuQ)?&tcSA
z1Xma0h^l?f))>!OjV;{_1Zr$QzO$0{H!btf!Tk=9i-o{F1`x`vys@~^K>!!VJhVPi
zEp_Qv{&<MlwmYv<sl7k!MxTk|!cP7bKurL<fC4X3`SHhTj=20tYg34bOB%mpiQgiX
z*3;6zV|oRe<ZrqsZ<#9yJfVW1p@Hzp;?1JSJ`0XuMpKm_KDcZIZyG@YMm0!%gDwj_
zXE=T$uEyWnTBkP%?0<xGxXw@6tDnPejYMyAWRd(sDy5BfpT-_~N$l`~kS#)AvQ|k4
zU7n)$3~?8^4HFF$AF49lZi%n4Di!@;(2p>@4VuAsl_cFO6<!N1z3#3qd+9`FsTG1J
z#`kSFC6~snz(qqL@KL|Dt~+SHK8Ap{PW+!cl|==L>}+oM_t!+e!0ZSd5cEt@s4$Ty
zCX6D|(vB8i`)u=YhCPa*f#Dg;<_p&PF%Snh3^{{0Tt3(Gx(EZE>D?pr#%!-lw9zbD
z?w*}X+)wN}C{bK(Ekp4ir~T|Kcerm8)!>@AstehtgBpIMWQzvKPa(U5Uo=mu!4HV$
zofOY`@dwcDcCCdkijw9>(&Z#5Hub<YMXnMjOts6XrYZV9IJUt7xIzS!JLH0lGExSf
zqt&h=9g?&4mreK7z~zs>&K~jXX)6ZH$_-tWKnH!O$pHi5=J8_%@p#W_1gZjIr;ULf
z{qh2h7N2Q!$aLsxoCUJprs8A#;XfGny69Zsc$<g!m$}8<u8k(R|Ni;D5XR$Qht&Rx
zEkX*vF;mBOuJYZAPPSEwI0})_?hUvzJ^zOA`RylM@Y`U$0Hz3bSXcVnPe=Trq2XQp
z7r72A($N|v##ZhVm}_JA6xe(6ZAOY$j*vUB((3fxz~!k?JC8l8-4ApzeoANd3-nXh
zM~3%Bd~{0HHwMk5nJ_QQ6NGbSscUDqvWAFqR{zM2i=WT4ak^DMo~Y457kPQBHQDWz
zCGP0l-3;jOqQ)Wbs`5^=o7{T@Grn{vawOo+>7k4eGZ9}^kb7<2qdRrUQw*(|^SgdP
zF@{R2dOXL*#s=%9$PqqXpFVDjmCRY>t=4Be%f3*^9_HQK4|NKO^SE&U#@s?=gOJ(w
z!<>_FcXO<OIR565neW7zH#I$Wqcbk#+q&0})zDhqCa~(0isW`$8W#Q0shtr?rjQ6<
zU!jY2>3$g*!De4VeC@OnqzZTC;&HE~`3T+kOQwD3<2xX*zX^2$f9}y#Fn@kCDLuMz
zEhF=RITI_t>34fXkrG**y8dXkQmCNF5uqS>hy}G;>!V}pw8R7fUjK1RgSjr|wM{Fn
zXObJF8l*Xt&Q`;_mhLVrwcAUKZ-1cI&Jwm=C~7G8FkyL!NT|-YqF0h(d-LE7L3Ip1
z>wcWI+ekST97Mmy&lZId9(k>y3~6rQwA_uuyzi8^ms9=e8)IAb3Qal{<LjCUuTF`A
z-2JOhMP8o=lyO#`mzcGlSnAaz&+IXIRAYi$0A7->wCbyH!c*E72nH9TN-jp>h<G^L
zE>B=G(-J?w?J@?(ofEJ{v_iQTro0iY8c%mJDK9vQ`RlR66?(Tybbog@SM(IImmj;F
z-$yiVx|r_yWR|3*T}3$Hc~98n`W_vgIqdigdYexBJz)w9`g+uvqq0`L7`1G+!|G)Q
zZp<W%XgjBqw*K)vWQww>IytXTc?n1tQT2@hIs)po3FAgat5{fJu8t7J%+|d%R!<Rh
zF|(@uOU3Kc-ii(^Inl;v=ocgAD!cN;1j{lS)(<-qV@B--HH9-4s!cLe`^^G0PNDP`
z#&Q=W@Vw<VJI_y+tnXF(V)K2lMMt^2ziwwI@#Xd|LuHqWO5{N&z9if7H$nmZ4WsEM
zLCo#Xk7`$W5WLPI#~x(-4?LYSGXbjty2x~R?ab&JN81l1?Zf^i%$jCVZ}+Nc-1Oh?
zqNez+a8-Hmy@Dv%;ylf!KX0cD_XB0&X2gsUmGv(OsU+sd*sR8}IXPV*E8)jx$2Qr#
z29`t9sgGe#ju+O*L~hpghBV5wqUN%S9EtWGP^W!H)~mA|hRUOAS`TK$^pr$T*#;l%
zIQtDxp2Y0)TO%g?k(M&p8Tr8QIORyFsb+Ug>Oe5Vpi^zo|7`2!qv7QvYbD)pZ4+2<
zd&>>Au^KMs<%99H%{-f(z9;3L=QM1~DzAI+QPitwW6)emF;2(r4L0<{oJp;|6XfAN
z#%rro%QR$<8-sS&yGhi~Ux<0C)J^_aNiaBaPvfk*poy>5i8*ZdHgz1xmc5nF)G#IP
z?$$?DMdx+WcJ#hThr?Vd)(5k=?(ygZ=`B5G84UBEtx1_FQt%YpKdh0G&mJ}+pKaT5
z;4a>ptdR)PlT|UpG%k7Q=|qTZX4g{Sk}!KiEU^hKa7A|;S053!6VGlI(o)03I$3wF
ze$Q=Kr`7Jhp`Rsdi#-t!FE!*}M$V27b_@13mvoT(R(a+S*)WrYku)VOBZ>{mI&8cb
zQZm_@F6m(u8?~A$F|{Ij#iMc4XSGRE%8qHN%^w<TzA>q)9U@IlVG-pKAelk!)+7as
z6$bYopLnK);p!bsu67gZhHJB48dF2(kJ&Lll4ze%H}DOHAW7*_GI61Jes~z&tsJrZ
z-OB}o2a0aB=`E$W>Z9YGJhkinl4*-|TtDQ8@JYeu{Y~mtvQqO_q{?BEc&4veNBM=<
z_&{(!o6Jy)Mdg_{p{bSArr!CzhN}ubI|!<m^!n>{sFD%gP5n#Kfz#6#d2)3cz5|T2
z`N98%31F1;b~yE<Un))J2P;pj^!^IpQ6f{-{r<As?i({%Bbx2?Xvf)okwzi=!HzqR
zMnK$n&4-tK`Gh(<yZpd`Yb`cfnG*QA8Eh0{a}%;S*}%gZK~ysC2=%z+_4x9Td&Yql
zS48)USj(Nv{R}nrB6J=uxB3md>$>0QP+OOP8wzU0R#!r3twir<jyIeY?M9TwlSx^_
zniGeyMk?Bn$?B5m+f{}1ZM~-7b;Pg0eG<m?UEM%*`&|nG0-AzCv0fZpCnRnjlZBJ~
z%wNSMJnrxn^x}um3voxoHuB!$>x5a(V(!Q*DZTeM1Nff~3T*Q7LJ&{1;f)%jh8spu
zzzcr8j+sTtBWrSw-99$2)X!wQ*Aiju<Sj<8%VX=y5^UCBaz@WKX0H<ri5L8Jc>Ts&
z9X=h9QzX<eAnlOCKM|&B$=Dd(Vfq!yhqwR9%K}T32gFN;&~PlX@p2uX>;9)7>E2Ia
zC;J#HB^fjll4G3}JA4uY-w6)cRKWIt9D;cn3RcovvDr7_aXVhs{dU>PbiUdU>XrCR
z$8|!^Sj`8*cK9N&ibe89=C+sRTrzbJ33(H<_$lOb7^L3fEg8#e#)q1i%2|@cjv$Q`
zTs>BVPMDrkt^Yrot}-mDu4{uxNUBJeh_rx!Gzg+}cZYO$*CCV+C8S%ryE}&NX6VkL
zJHCU@`~Bb-Tr<qsd#!cX-m|RqdzsID@S6QNVcf*>_(hBVS%-O#xWv0c&zen?5;d0}
z*EV~QiXN%H<?V#~_b_g8TO)EMN*qqthGo)ysm@=F4*0VOB22LoB!{l$5r1hp>e-pu
z-7zYiGq}^ES3{pczp9J;Wbt<X2B+{jRqA5&Q+P3-K;PjvaiAP6kFlV=w;wU$CD}WF
zCrUd5XJTkFqhsyA;lm*Ox-m>qO-)Qr4m8&?NAvUd868oI!nI2vx*s)nKXy_YJZMm+
zX_ttUlr&)F)t)3bk>=;`!VYE#->e&^O>LLjo2jpMEiwqiGjVj}uVI?a#~AZ#E5PD5
zxg+E}MX-U%n#0}I1VTRDS$hs13LH3*%T_ALhRr5)!yrowcUn?m!-%%`D{{>ykFy#<
z@72$5?5@q5vWAz_`rE|&UtsPcBe#pM;-shsjAZkawkjWU^Q}B1Ty2$o>^cevZs6VN
za9Xc-xMps-IXs(<Hs`t%n^}vqV4>!F#N+!Bqf3A`Yh^b?M^n>?75~gFRCp~5Q>Ze{
zfAKL<67aN7%gFHU>zf*{&a1D#NxA~tAgS09d!%@lV5TXFj+{AqmHHf0sv~<)reb2R
zYh6OTWVU@|qaB~-!ceoA`OcmI*w!7D5W|NKF)l)iLWDEBELc+Vk}U69q>M7totg_)
zfSYNvnAk5=`s4gi)<pCwd-e+`0frl0pM;HrQ$;Q6wfv|_%#WqnHLKu?EuDfYQ?_ZX
z<YA#I!iv!!>RHOT<mO!y39MUVCdEJ+<<NTJCv~aJM`^7O-Gm5iIFmDVoN4kknuP#=
zLD79u;GZ0MAMv6;<}0YvBe#!0bA$;4@R;q>Sn_Js?Cpz1;qr=#j@i0%rH`t8*^VC$
zp>$wtrlY4HHLI$!zUZ3F%g)Zeh_XzUc2)xdDH4k7>j~i!jUN2+O7isj^HXOlOkcb}
zu9j2IP@FY&WPk22JZ8U)f!^dr*k;M2#+wtM0@sDerTE5hdHWS~d!`R9I$shCfO#A7
zA5Qf^_@8^9_)bs_J6=FtuaPu74nZ90Uc;!rb+ZJ{l@##KngJtLu>PzcWixJdAgyu=
z+Y|b-QETQ_i(Y;08iRI+s4@0i*e}P-ig`2=(JoEd@G&Ov%}dLwzD)M9Hd?`_^o1qu
zX3P937+86=thH5Frzk0m{Q{uZ=d3s4r*(f#JIsU_{k+CK5bMC-DQoYl`O5i)<<z~h
zveK0Dadox*JI3a|KGaWy7~GcFH(DXeCFz%sI5;?pii&XJ&8<__=XaJPI}VZSIIF1C
zvmc(?%e&d~UFuEXitNN*8Ud^Yn-kQp=4%2%a#XZp$YNsLJ*%$P0Lz*G4)lC`zP6SF
z>Xfb(`#Fi*$vc!{&xB+7P#6;XyA;<uP`ZhT53<oYIF7}%q<b=kB`22@<W~S>Zi(~T
zAI>>e??Wl~m-<w!`P=jKk>L#GXxwYyEcO?28T%=C?%&hndA0ZbOdSCc3-l3&^#Xlw
zWcw;XqhZJsu16G5o&M_zj-U0UnIOL?dib`31d30#!UBp!8^RXM&{q>zLBZ$0{+^LI
z9bcNHEGP6HOC1-wTEbuFl*j0pC@xzfU=m6Sxt+)u{s!nmw3dpq)^bgRhvT8U;a{F`
zLk5@VKNM#2T{zRndeJip2kkS`TTZhNB>sGXK)`pv!~r<}#?<?{J1_1hvGyq=QS9)V
z&Aw5Jt2J%<E3Licmf_{&!s~mqaO>MYPWLeRW_mNqirLO%k|o+xjAY&Oi&zTq<_;M7
znp&MMqu6{4#qLcv!*i7c_sD(yL|eP#w68Vr7~wkLFLjzuX{e%Jm6U2bke)yPqSc*L
zF`@wo*;3`*J}!^s*96Wt{Q+cP9oQug-%7sd6XEFFqgl-pEP1mV<Q(R_J%A%^TOeW^
z?&agxXPJ1p0K8Mz+RI&Rl7B8-)i&Dy^Q0rorvA0>J`u<09Ph_OQ;tH7kZRu2;MsxA
zh(Bf5nTCDCa*Q8Gu`;w`gjR<7iv2ASOFu>d6y|zj3(iSm_GIsCjXSe~{@p^B@$-LA
z?Tb-GJkwIr;O-JXWU4nRoz;sfZTu(RGrMC;nHr)83CMS>GS5ZN_psYAE>{1Op-cB;
z3=55|&-FU239TcdV-ePrY*w`73q>*`R0nY~D?5Ibd(eQM;%U_;Px~3z%_JbSWSr=J
zFUufrYqOD9N#?-hH1v5&N|zn`bY3AY6v=Io%iARawoB3s+Zw#Pp2Esy<GztelvrOT
zz?1CZsl6HhzKu{x=7I%cdx5*)Y1f0^7trH{waU6|N5zwi+8U+;?ay6{KOc=I)(YhE
z3U7CFn#+HHkm`Bb{k<hqFR%tsD&#4hC4ctCfHC-&$gf<B?%g8A;xETK$kN(Z;U<38
z?EVuG>ywQ1Fo}!gs(m%%VYy-i)VAS$mdN%H!sLgZe|BbaL2U^Svo`*BNx<lwC^qiX
z)<d*Zy<jK+9*KF3t-J4?os)YPS}JU9^tTbvwm+W1_lcvr+ylKt%-iO>#3iF9@ee(T
za2B_lRv`8-3@&@i5U$1l`FU${M6;t3!|S!lsfcYczr4H;Mjg~Ox-1i(DNO4<-F5FF
zA3DL+_qv>ToXLhH>urSaXHJ<AoX?2A3mC8{ol`HWO5c(yu^l8y)nkuWJv%avHkxh;
zzU&&aAA37N2(S^n*bCISC|8<KkP>DGT#0am2smqbUQIL|VtxF!cHU(Ee5l!~H1nm`
zaHW;LAAvR(5+zc-R3lPqPwSr-W8x6@Zu(v=cZ|mA7f92;q4zvIY`k;0*@)>S`76`N
zpR?;L&kL=Xo=EzrV&uQDwYsRG5bY_y8>;vB3S3}ipTz388ni!}75F=A76jxYG0UCA
z?kB@M{UvqxCq(j!X$&?J{d)PTAQ|@tqzgUy&i?wjdAhp;css;=ovZjIZ!%Nra9YCK
zK)Bzxdnuxz^k@1=*oYNbrMb5_mF!E=q|Kg4MX6){7X6VCtGV_qPSmJqYA<YUX|&#~
z2f+MFmew(Xocn>%csxL&Xs$ZT;@$&?D)U8*_~1hpQz@(_hdD76!z!ob!}oh=(BG~{
z5WHt+lqQYe%4q)xEcwjsu0b{E`7$gF_p^*o>R21}$SImocSw#?)>r0MpqrPYs!PA>
zN3X8Ipm_@&{~bN?J7R_<)t_S_wJpe91OVdxotsohC(S<@gZ6)AM{bS>h>3k?w7iw<
zD!JD!s8KO@I7>c(hYM+T#i{63gk3d|x(9sMhyBgqjKZ--*PV{2pk)$mTAqnadSC77
z%v7*MBIf7{&mrj*rvXdWcMMf|2I5eKW*7X_9)a@L9A#It(!^?yuWo?+vD5B^K`NFA
ztMy?H<X2$Tf&2h^*M=_WHn}v;>gxQ?w2ayYZCf%Ry@C+O_^eZ<tju|%HN&zfBiMPG
zL)^FUw$^xbw&7*V3C5qMxqxwG6<rpBAtLMCyycmx616)`<FVC0E34RFZogb27(IAA
z1cVUZPg0a>5LsO>NdgW|>_=LJyiTHOE`ts}lGa(T{up-L+rHEYb^gJx(BgtmB+_sS
z(SG)VUY|yE$j6HDW9h4qtUPaI!7l}RTO8)2D#{%0q5r^Qtnb->8?2`|EChc0)5&k)
z;T3{8>?J3^p3luOYXO%d&E_rK5VKF}q+SmibHW$P5I|<j#$Zh|+#)GN#D)V@hQ9A@
ze^F8+{kelo+b0UwIyxKn^r(MhA6hzq4-t>w`A`kyh1yrYuex|<&rd1#)vlO4+HcqP
zar{Ar37oB5)K2nB2?$SEm!sJe>p@G9DCw~$;U3O`_%l(A`grR$T7~1;+r~$S@WRJH
z03-|BT<2p-8pc$A<rU<@GDT1pAbc{auwIL6Gs0`yTh-eZqEg4RR%$&sOe=VhgZ81x
z_kM21$ZSUS9@$RBx7mutKUaEhb)5v&-_d39Q8&X<7hLZ-gjO8U;y&(~+lBw_P!MTV
zLQYO<9Y;E&>dE)-$Pzb4O6#v}h8_W92j?o;ItC)cyb*ER#-98u6?YaZgwAIgXq<KT
zD8DI)Y6Q}r*xonucmGIB2I97PJ{MsN0uMD)<J_6)nC2I`=_pSRdSrd}sl;VO**wM2
z<Qp14VVR8xvPDBk{ZE6OF1cU0&O^1}KtR&u39_x>3y%tMk6LYzpWy^%?AA0cM=pnc
zgLQxb;pmErsSPUZZ0XLOpc8bf%7hdl1SO1M#dMedFzB-~j9ktoN_eAhGrh-E{y-sp
z!0jf6-H429sj?bWPpU`<t{;ol5NV`7Ze;C-L^(D`6mHFalBGkn^z+Nc_@T^{R_aFJ
z0LJ&#?N@4j>N+3f1+a(bDi^V4wmMH97Z314+6)_JTWpB(ZX-X!KJXfc7DbV~`dV){
z*p!K~P3Qz#YmhuGkuL?lCN<eo6A+SW++U4*x&H(*6kDNR&z~+O8GQx@Fg_+qa^C*P
zlxHiV8|}&=fxE8T!L?r7UFkRzSu~jIu>O_o6$vd=>;uG6ZER25=b7H5dY4jlIZ&R$
zh})%L3jw{l?{8JP5qIkF3ot;=y)w^;;;uq}<CP*8sT3$|9+K%U8>MG|tW^9{Vj&})
zh5_$tZ|-@PwL{`!l<BxT3n;C<F{YGPOA1oY>{y&6tMQgLsjS`_+*q9tzni#UCzgqQ
zZjGj33QT-kaKVt{(Eh4qi+J5T8@>4^y=ru2e=k_2mY+~-&%<3ecT7P|5wp3vFT+&o
zRc>VxO)s5=?!P_JxsmSJ?E93{V6?~CpoH7r*G%Az4QpZ%F;okhoX=OJyYp>7Ml1L5
z;YgP?XS@1_v`V;%=&hX7C2XvJ|ILVX(h!K5;db-MFI%ElNoq>T@25-rB%_eL&@%MU
zK3x;mM8AluT%5|69qe1a`9fN?b7<11)yk655g)f8Pj=3Id8QVZddo(3GnY1&^xutG
zwJiT@q3~9u&2Wl;3^-hi`(9^l6(r_ZM{U(Zm2hKzS6;tV6OmC!(E6dC33Sgwa2|`l
ze?#AeT7-Nd$UjS(bNn$8_dd7+^`qjR?S*YujpxS$f5)y_NN`DDfg&h8%cjIe^cn~h
zS>5%QRNMo}yhV@8Zyy!e#WXPGenozD^?Fe)eF3f+_7qd(@;1r@ZgnD@&509Y!q&}a
zcge#xZK)KAN=A?#CCxYiabP}G+&=gtfn?t)4NG8Px|XetbGZwnyNZaS8P2Al=Qskr
z{HD#3gVoJ{2fB;A|I=e}msdA`Hf5r9sfmoL#HLqN@%}!Rxu#`?Q*p$x`XY`=#IGvY
z9*jRgb&3cvaZs;c;`Jtv2rZBjP%p^;>wBP6s5g9vu?^7g6(5G+Lw~NwNOzVf1PYzH
z{kW0+qapca7rwFGjg5}oyU}1n-3IHp%mUOHvFEX8t`=XjU{dMW^lW0ajgJ3>XD!LR
zJm?si`E++OQuFoW3;N67spZ3CBYUAtumR#Pe8r1<x8e#ZD(!#j`h&DearDDheD%T4
zMj~=)FaxY{tj=byV3jBmvK_W9zWKW@ppD_emXYvKsK4QT()jOAhuYWwzo>Lo3EL0T
zWu6Dmq#=)D;9e(Ui<mWd@Ayat_wABwJc(SJT@P633&!&TWf|WJ+#7d+XjYeb-(K{(
z$(I_wwkqVH7&g$B)ev`e=&8(<5)D){1tC`OXYrOiubT4zb%>3!&q8Qi!fxW=r1-`-
z(%yrI_7*Rya^oeqxXkOnkI}tMKF~Sd7va70cOYlm<fOgzGMD$|puBK4Y);oAA~w&R
zs<@kl1?e<m!>P203c5>nvKCT<n0C8BoTissSp6>5fXWKQ0%$WWx<pa@fO~FRrwVL7
zpzPR>UzEQQ$y14zyP;c2a5H7@tI&6DcFbyA%Ij}?&K?T9E&LCZlZHyVIX>hmFAq>Z
zSEtF*_RjiI#yP+FRm$R@X|6}WXN=lpO*W^C78OXYHOmU{S<rhE-4o)a5zJ`A2{e#n
z@JBoW4RS{0|6wtp=NGecDfm2}ybaZ14cWXR7)gBPsk`{8^k^RBZR7n{jkkSMmTR{X
zGdI1d;(8|PFTtD0F7P-bGCk8~;P_Lrjl^(Uh7$TQm=?Ya=Y9P!C$oCPi97W0CaO(z
z{<Ff4V?Pn6Tar%leD$!Rn~Ya%=U{3D?9zDVZbcqd!5*kPE^645l6GuP=Scjl+V?xm
zeuwu)Y(J^cTI5&iYl&T-EKV2B@oE6#1|iHp^8lvZ{~k!Rw@s#gR#>DioG#j8y23+D
z7UBEyMV+_-aE&hP9Q1G;H-}!YMO`_e6x)08s31dqd1=M_*Nze{w?99)7DON~x0d6<
zqmV(f<be5*sV6k^!PDq-+~CI{71yEqogp&LF*Ku}ZLs(_FV4X}JRpOUFZK6hUDo1t
zgY<U6dlx)|R(!Hj%T7IWW}#oI{>DT)cMY1JfWlMc)7H>z^>yVFAjy>gX`Ja&6&dKT
z1Zo^7OwPw{<h-odeuS%CBj2lOfp#&?0(tM9Nve*7MJUxA($^2wiq%~{qNiA5sk6~*
z-vue~gnz0{P3m;;0`4RvktjTK9wNvuA6?qOza%VTNEYU8pUH6s<^8{%*E_*(mn)zc
z7a@i)suEMK{{RXW3Vyct$?4g?2d9S!x<_C2M`x!`&mAh_vt$5~x#su0Uo!!MOAB;M
zH<{1;b+;TK!Nk8`=LX#qDJ3zkmVvvtVBaw}eZt2%Ex67S55K!;)q-&3x26yeX)?_q
z^@8v1d(LW2O(t6M7sUG4Niguttc*x&hb3bYl&@bJqXw~EBe_+qKfFY_0$8bbyL^Q|
zDV#SjT;RhbUx-!S2}hRv4A7my#wIFk{C8BQR`dVyFL+`{ZP+9HpJ&4hDPz!4Z?ts%
zLb&&tSVGB^G1EFh404E|@o{V_CnGHyhw1LGV%G!5vwF3%+?lsnE}zgVMmx!5JOk3}
zQ-tCNjQ;SkVtE>}EkO8Xa$ZGzx?(%7fcn+5ocuO40W67U&RY0EWi!|Lt`uV3$uJTE
z8Y#2){rYiI!XX!#C5-yx#!<a83o@c5C~nduny{<PJxm*uYf|C`zuJr6{I)h|r6&Al
zRe!t`_}_yY@{RGE(u+Z%^Kh&^Jh7|RArPEBd0avS0St3k0_sqf>fr`5@AQapNOztq
zE&#!f^#?`!9fhUuED5?!_~;cGf=wZ;ywju;bjnS4tcG=6=a^(*U0+y*mwq7f5?4`Y
zRL$sw8>SXiW+M?kPaEGh>DT|)CQUZp|I6r9rj<I$@m`!@<xUb-52v<WxLztKB`91X
z8L{@>;+BDy(VJ|t40G5D^b!VOF=@9Jbyg3n$SY_g>U*ZeCZ1!5UidUyMH}Fjf}GnC
zlc)X=Zs!<u7=klyYqts)!$fSH>m%GTD|!dAFCTNA3;i0fK~*;_IRrsu3g}G0>XuE4
zQH8OCK=^2D4?#3UC)_=dqsHs*Euq!|N%a5~E777BdMn#F$jc{zzRy{7I0)H*Pmv$b
zv4j(QTxIk9o(O1XWM|MM8<t#9+V<ncN#bcMq6z4zwKp%N7%f1sKN6U4yJ*QEeYR<G
zWDZs~er&gVlf6OL<YaB~jcMbM@najFr(u|Ul@3;ofXov+D=NZlmhe*v|02DB_D)aO
zwEyUcoKrJcCkJ?G<#DSW!r$}isbXu1MX{cg`HY|EMC`xuHj7t^9-)0@MLM=-!%ROe
z8spO#r&rb&aErnaK+keL2`Xkc@D5c5g78uAj&HzY7rU~ast^|-U?Gy~WST`Z!-0~5
z0#-J1$KktGT-aZj76lXM`u;NmzHWXUYc#(b2V&s(BeF`95ZE&F#pz1zZFB7lK^NhY
z(am4S$V*bVT^*NyIv(`wo;wo-Bm&45#Lzst!K`HmV%oBhUUBGi0h81V=*{s5P&3~R
zqg<0^aiwNCM%@##nI<B1<YmXSdeWN8BvLk(#}?25J-r3LNi$OQQ`FO(CMh$?*F3=H
z@cMFJjmfzCYp(<3wPjrI-rSrI{4t?F*plWNe(j!k$mVUkZ=rya+E^~W(BA9ITg;8_
zGb&#2|Nmh)ORQO)IBK7rHZ^hYfqpQa(%|*L@0H9JJ642PTJaLeO@Pc9rBrFI{>|=+
zYJB<I`B4@)Zq0c=qNO@KBINmQ2rckW89@D?Rf<x7K{_N7@>=!myrA&rfHl%-fWj2g
zh!$=+-pn&@)FORpE}U79S}?$ZnMHH2ld0L7nL?U^5oGgRN2C0;JnbN3`|KrDTsL#E
zHK-_HbdK}ogrQ`nb7qz8Tbf-Y>HWKusQK614xS}$5cle1L(umD%8eh~sy&+{ClG*N
zZp-bHEsS1==$;O|a@e=iiAUUfSKWSaKh7Jo#BhSYH(8N8^;Qrsxs*OUXG)vOuiM*t
zeA%E&zqhC=`ky5)4N6c<oQn)IAw)}!0lyXx2uV^+_~=l@#Uf;on25Hwp;tkgjxRR^
zl`NP|%fx<M09HD%Bk$tx1M0``c$(D*4`Ea3OVL<~Crq4E6=6&rq%FUtr_ZDNo^E0m
z)cpNlfor}8!7W(4Sn({ULVC`H@4r7PquR6P7U#DP_D6s;%MF82Lbbt4#Zd3E@GeTN
zFAVcsrGQ-&(ZRuY-|Qr+Vp!tnj}aay(X;{7jnsrhw^RpxIhOoS59n^)BBXn7-xB|$
zJ-R21xEG{5PehP19#Jtyad~+qRO1&O$cSDNlYEzZdUb;hsB6u^eC`@&|0Kt2B6pPy
zmjR=YNWIZ0q*r{VFR80Z^<bAoy*%+kAi)5`Zs0ift*=U7;>$cb82Ld5$c*l;Q*Y5s
ziy~8Ra817<NqSoy=c_P_LFEF!uStym(D@B%h`1xj^2K|Y^xse5jx47m=F>VtIZt00
zDSqyN5rsJUPdX8b`bWNOe{<E<vR>8C<9jSv#&Sn1=Y44TRyP{GyHLR!eSy1CN4W)~
zmbp%z^fY5y-Y3oixP+7~bE?KitxVn=_zSafAzkfpCVEKH;an7ZJuxUP<)1|r8!wKI
zzG$J4O=Owg;>CuYunto|6JghG7j>qzL?S%g;o;etACjJqJp~p-|HlSL^JivKcOujb
z&29{0gy9^*w=SW)4P;*(U89^cOXS@=<~V2A%x;19N=+85EZIK_q2_7SK5~{eC$?2?
zq+0f*3XTN1ilCtmgSTYCryZh6`%r^lrR-M`K@aZGlk+R(&RmWp*HH11855fe+m5d3
z!Ci>xmd)mkN%xkyIqbKFma>?(9Tg6uA?U&^&r`wiOxL?x|2tt=`RPA1JwquEkg_mR
zWb-VSG{iI}t2sTiyHL7Nh_uh>ZJ-_GaWxt`y;O)@J@Wu3Xhh`<_I82UP{Y#DYzl}j
z32sk5StNasAm+KHy=Cg<@&SFt5dCZFuZ~<`Z{&32OrJ9Quy{BTPO*RMh(=vaoKe3I
zg$&O^!Y(=*Mz@AmTtuFK;s)49pBu44ILKC8+b!Q)@`W~+-y}6?-SGtykA^inhj&s8
zKr(2}Q#IllZatbN&?nt4m}dG`aNNa0_<Uv!K5uzFGNwD!7gIM`kp&kijH@?4ss4hh
z60iT!K@%lGH(?s3AIw6!*~gxI^r1GnZ#0bEJtz={8|D`WU%-GsIJ+om*+@@xyd{5k
zuPMs=WpOw`22D<j-j<e?WwM5kiNo)MJS@p)NQv2G^}M5S6meAS!0G>5x$EQl$}zgz
zkw4NF#=&Ixj8640FB2RPOPQS=8xX4n%o<eYV$liM3d4ip5lHXBi#{d?v+Z^C6z-1+
z<PH8~Q9}E-h5&0BTJ^}<luzgbY<(_RfAj<d#cDSb&VTp$vhUiJF_M#GlxdJy4OA5D
z%DtQTE#7Z#4W2c7(oj6Ya4W0NYqXCrFClFULBT&*Xjr`!D3-M2e44~VqNb|hmde78
z3tb*W#97dyb!vCg!VqdrV)Uhp-k-{%#s|pH98a1F1AKfzRvhOom0zBA#%lh5B9CAt
z9VY3}kF(IXAjJVG#23=AL=z=U{DDd$LM3Ayv-PTlUeF8f>6Y2%+3*uSpki`ms>``p
zCS3N>fh@a063oy&KCc+sMX^M8o>wha<39`i|15z`2>%bRq{JTfc=B{~FKGZ$SrZrr
zV<P7}d;@|AE!f?wuMHVfW}G+`-r6;dKFJk^N^y@41|6Dk??jxeZT5E%sISj`BA(nb
zKqC$p9Rb?*K(^H=idA6G{Fv}%zeJ-qP{yba4~xEt=v#>RVsei!GJksh-)Fx2-`!qn
zt`)y`ePnDwQr^(hj%!P~28g1~4=ZLl-RdT|YQ!_M?~kmAuPk});t`^V)RLPTUfP*Q
zgJ1hFRYPq`JI}Avd*h5pkJfwc7?1})GI|Z&Dj?ugf=6S`h5mw&JzoC_{YtCJVI2J7
z@Lw+o5q#Q(wm}EZ_q*}|ZyD3-)vtHUi;5KSwNVj_!sT8!E&+c9nWj&peqzv$bja+a
z$usMAJ33<Sv08FAv}i#_rBYvWfqe0Sq-ZXz*Fbl;L}sEJD*HHQo!tl$2Q9}V5BKd=
zj};%Y7`vqc65X$)jqZ`4qNsW^Q4<QR9dd8gI6MB`njqi5_d>Y!0{r2f)gLb@8LS8h
zUYOg07;79jZ`vi%_?6Tlb9o<;zlw^#hEUM~c1twdaoXON(Z!=N=5oxe<I{UGyUP-p
zs+*L8*oq|mojaB_3NyO=?-^tRD2P%d(|I`<P|{ppVJ1=va3NcGkC>ewu3&ETuD474
z3AJ53A0HFoDH>b*`<e$Qsjld7yy3Az%dF;c-vFoMR8uXd4Us61D$mi68;=BVMyB8&
zu;Hh#Tl*n1xA#aE$@}0G=nNq!e<H5M@{Wzq{v!(xa~L~<Ai(~mjcA=L+q`L=XNKMA
zfG&=HR_pi{Osnh?@2(&8J;gUzCQ2VNkHPGWS~FQiJYUP_FBiv_SCv@r-Nt}4w*YGw
zXJ!KAs2FxI!BWgqcPrD;v>YQUTfliVj~QOP01oIjtgx=d!61E&md)*K83tZ_A-T<0
z)gEjZ-oIrQe1%n<8N7Q<^xF1iqSp_l9*$J!>!TWgt!%#<?k~}3vm!DTCtaz9!E&lY
zs`|QZeyKFQ3-buyl~osd7l4-2avx{zFMqLjY;#m^H#2Sm0&d6ccVA{{wV?$~G1MBd
zbnsBWt%@vd{_RkjW_G97!yVb3AM2$&G+jWlD}DH&z=7vd!^&wgbFqUOZp}SY?(^>g
zi?{#OTER+TIvnh!2CSO5OrQer$d`YXZ7C|v#%qiJTKvT}VIL4@oWow*+ZwNL!v7%{
zbPxWt%*D2kV^AA_5ig5tH*OHp|40s@h~oFmUF_;XXG3!GYEr*md3^V_r(9^<?A+#O
zL-e=v)ZqtNn5>A5vDN(*`~)EwP;&WoFOkw=V0Hz~BGc{&=0K8FP?LPT6&|-$bZA;z
z^O~2I)~g>g2QD98?zvH#D}iZ6o5@!-F4VIk<ApP-s&tz<Qcr47CB0-xi)W9k#pIEo
zyUhsA>HDn-9*>FfD3skr-0+wV{eZUx^^3>ekuSa2U$Jkkt-{&{9}{PFaqhaxABvVf
z5=hzTpnRKL$XN9iJXi^;hAVV8e`S=3^TPURP6PtN|B1Hb>XgXN)Y*gX&q{27KnDK?
zOd3nPhj>mFguIwOl}1>Q-`d7=kpR~7ev#3+$rgb(G<@HzZ*-A<Q#2@mUN67)Zd>>@
zYp+a@(v8)Fl%Xx|6BPm@T5FxfWHuL$#&eAjTdd<^XR^e{OJw0DxxB-@>7JAd4&d6a
z24mxzGHJ#wcQO6DCq<p(;RlC{YGTy~7J!$m3ydp*56)wVGpbiWdDUoYIJpN5q$y}=
zi!L<?v#jcHu;C94l}yuc82wOH#<`IxR?kY!n;jUko>FgV%1xcUtAWr|+J@k`&9xT6
z$AwWj4Qr^tWWi_?tqyPDhXH4x{MlicACvssEP2=P2Q4SIyGZzPtk%VMyd56#?64Qa
zd<=zV*5A2;c0jQb|IXsOug)43!TO8jq+Y(1FY*Y(D*qm1wpkfzqj2VLsYLcNk83+U
zef{2nfjjf8Z8u_muf{CF*I?;(8g}z_Cc5MC4bE}kmIEv714~(LfV1hcdBOWsNu!=3
zIoyC}pV-xJ7Wsp}M@R|i>tFX16CppjYrm403cv3``Ng0oh(Q|7twXa!)Wfo$xOwP;
zS!f6O<Qk+Pqobj&59}ehKe)@emFt+hb8Lgy;fUBm&-_!zK_Lub#jk#Xhu7{K$D!8Q
zC;F7f-xK&*=f3JdX4fpwGj}{PGdc;HL7t%I^5<~II?l!S!-Eb4(=~0fPssvj7KuE?
z{fQrcp)}T4EHP_DYs>^vxoq<8BbqGOW1t@X15;WYnCT{5LpU81)g*(&ZDmhseBWC>
z>?AoQH9i&TpE(~?M-O57E=T3E-8!#){q}8(u-MTvx+);piUUh`f;Fnt-c&Gj@Pq9=
z{6;s8b=F)Xvm(tFV>&oLZNZ^CX>t7T62CizkoPVTBs|?C>0iMoY!S5cPjbR3b1XOB
zryt*O9;N~H&-OA5+rH_pipTu*^C0y1cHf-cqzOVpT0}B0GUW@lfh7|+D2umei^a+e
zaKkALNzTasRO1vHlFD-s;GB6oh=3A^TH$iI8tSnjEK?)grCbn8f0rb!ESN9U;W>wF
zH!br>WxDqrlW{?6EZgsmF-aEZ9u1sXk$?3)&)9dxPYc$hx^i8ttvmPBF+@^)q|^)!
zHW?>2w1~a4;#JiqxH<Y!(L{Vi4{#aU04cZ0o<MK!Gx2g%A`dT92X%jb=HH7oSo%xD
ziCop>;e`~YeeD6pwE-^UUM?IN)L892@6t?<tGvf=hdq9c6A!B_JM0G`Aw4y{u&9It
z@o(Rr!!a{6e-r9O4pu2Jk~zY~!^0*ak#co)J%G7g?28Sj@EKCtJ#E`h2`YdFzc2`a
z<YZ*pWwUZV)u2?Vheg#CC{5O5lsVh$Eh9{`9{KEzs!YrN`W;X<=N;0*BV8O;pv*=X
zs+*%c_xjV%h!gev`og{wOGUbONY3PXHh3Qs^XQ8H^0vIupynawOOj*DFsuHQBU-mI
zjb3s9rUz_jv$(n<mo0^}xQu~0+fUQ_4&}aMP`wa-0ow~l{cU}0{F|BQp>7C<TwQUY
zRA3Z#baws|UpB5{g6HPVh2?79X(@csZpl^%*ubcG?AB32a2fOt=BkLf9RmCM#Gz0q
z3KkYMH+KR&0@6r{4!@wFplnrS^C3g81$*Z79uX-`vCCk}ax#yz6<}#sV>w@KvD8e$
z-y(cb{8XqvYT=5KF)x>IZWq~+pD0I(d{@pS+lgUR@87exlWbpa;fV|}74-2e<?}$v
z!sKLnC5+7&*K(Y#gD8h}t4o%_S~?*reRJU2qI<9S!nP(-IH!wkq+-Wb^Syrs#B{uK
zPXlp#0zaR~dGU$=-(^)+MvFyJm6=WEpC0!2^}VB~4>_2xIUN-tgFhOV;b(R<>j*hO
zwaqfQcn&LRc~EzEAI%Vogg$*KQRk#ok#LcmJ{_C)q62tROctvF^*g}4J|R#Cuj#OC
zjn=df{Yd-?2xF3U!<gsPp0gI3ydYsT@&CIYuWxq&nCj8{;L0Z`be5glg4WMdSP@iW
zK)a@kZzeI2T->lS?CUsLM8XMOWHs?S(^GQPal1SknvS6^Qak`_?rcxs_SM3N-AfPU
zgPA)cm(ZcJTBj4U<On%?`x7Th*tcxE!=!{Gr@S$g7~!O^OsVVDFTtVA#>RHGGaisB
z0z_!<%Bm-ijL3#eHM~}72KOr|k5l1=({!;`qPyF_=Gq?)N^3Yj4S&fZEoi2xV)6g&
zNS)Qa+6p-l=R_&&Rf=qgdbVr?lqz+k%fo3RFj}_v(636;|Ln+;O6<|~r!dT|BIFhH
zOdwTdrR!%f0%Hy_UH1=i%*g;t2oSvxM66ng0n&0%P2=;-#t<<#SHoIsVO89jJzr3-
z2FikW$NJDhIonqFFEhMWM8Fa>>giasIkPjhS)5y}4v<)xqN;4pWp6vS0|!?v(x>8<
z^Nntv_SAJPkr;bCv!$z`Azqy(iofT9V99kEr+O+ULML||8t*UWp@7R;*z@Pl(<aTB
z69>U8+ppwgDGv|c{#Yi+(9q#)5r)pL$^93LX=!gGBcp{#J`qYGWd6HYq1fa5w8Pu)
zLg*F)3a_h}-qTH%IGNQBA>Y)04dmlA1+59{zV9OB7DEBBvFOK^mw9(&+uZ3(lzb_m
zCAY6hk{e7f7v9kU+;mc<DKf?qpb(9`Ac9Pt3V$#hW(MsGMX^2}lVx@O?LDL~7t6VF
zug@U;-`&ZTG|*HpH{ZQh-ZwnSsW8`YQ_bMa^wuG}QdRpQndM8K^05`+9Wm&(ZwQMP
z{LuOtH?{E!gt*~_i;%xEr0+Pn8yD;5&-T2B7Lo3o-_fuE-u4ta7cjaXX4tPXxjD>B
z_NEF!4zhp$ox-BeZ!htvts{QR{B~s_ftCd<Ozx)3Mfa+jOYpK9ZRYf+JO6rLP*YR4
zcXY&a+OBqC$^}Xk6C~Ir@BPuR%4^b6pX<!W{I`ebTSr8?>F+sJHrlZRV*YO1FvU$H
zkeX%!@3HPJW8Q22N{-xi=YH0_o2_`4lJ6aW*`6-{i5sKIipWhYLZnD;{5~6A_frQ}
z4iQHX0i`R(S*@Y&DdO}u>Kf{!`!na5^+SBJxFrNqR6lL2K3~Hjb;?nm0kq9ZKc<Yd
zOl$&Vg{zxC(;u{D3*J|Fmsq<>i5)uG`j=!Z11$%-xj8H?70R<Z)*LRi7y~fOX*Y!w
zoxVGsMP)k4y>w8UG1Sz_dUhFm)}4%=Nzkl9x^!5BX#ypVO)vHOU5n_}?(;g3-W2-Y
zG^Ugh3+%bvq2X4Ga6ezJ(&>G~7ZnqOPg5z;mb%HldbqzWD=Ta3?)E=(*jVb$6p1(i
zI<aM5;^O*eW>VbWoUe6Y@fT`o^t}dym3LQb5m1lQ!K>R_eQRr(08A2>qc-H(2IrEl
zQ$=Ova+?*wZ0TfiRaN}z>S}Nce>57v|NL;ZHW&BE$i`N65x2UwrtwL-Nqcy;D<q!V
zkq&ILT~kxIDk>^&XWMO|1gxsn7V&`6mei*bu<AcwE&El~fdA+Mbo`%4qv_Pxe*a>Z
z6qBlA?9cL{?v>2*S`OK3LIdOy=O1#5ve_RdChSjiD~r-1?UoDoKm?#LC%ONj(#!YS
z)TWX6afsZZzFFlIc?Jc%t^0%9I;-KU2IP+A<Lc$H);i3p0B7ztT^TTcjjHx>4?Fbc
z25*u@aqrr3O0&-{B<9<rM)KQQA7!{=?kg2eC|R(awZY2%Fpc<diE|V=R)oTw@ZFy*
zMG1|)enruu_kUHMANXtX6-|HQX6Q)aRJcFCctKNM9${PcLR;opR6IOuw=QgV=I#qB
zd8d+K(F47uq4|7e%B4{SK+vQe0j9YPSdKShb;io{MHYcaPkDLy+}vE<%fkiUo9#^S
z;o4jstsre_HhXX_E-pfCAtZMXw=e@kLpdcSAw9i!sa`j<o;PRT@2-v&m9=v*#&!NM
zY6Zu};zdzOt;XtjkyBFsOh_PPVP#dTGQ+{*zx(**%a<)m*+(xxl7Yb3$cT!ZoE+*f
zE_HX(&y>XN7!3ZgxVSi{%`%*h*EQuDB*OS?dlVyub(AH6uZ@V9cp#Ee0xphOAE~ac
zZm{X!dIk1}hom;xt+|oKz0OrUQx`1oh0F;?k2y~!{H#`w1Wl4P?Iwl1PEWoUP8xCR
zYzQqztr0uG4Cf5XN|C#49s?f%gE^QHlUI^|!ExOYru^Yep+d*w&#xdgJkFW9JpJPq
z00U_}cuX!msv@#-Qr+c_F50Vp%`*?ot}P^n?2;U=ZVjC?Ef#x}g0#Iqs?kXI!Ssp`
z8|>}fEnYQaf~$06j+$iLVLOOilScz`Jq2YJKnDu~Zz_t|#*ee$<%C<|DvLBwvG}P_
zFzEnWB=H8PFo>z<KAd`;^imJRF=r0lN{EYZk7gpr$HzN{UvDN_<8xRMgXMC5Xg*tR
z7>dvQEUMR`?RIM@x&OxtY{I0RvIN4MSqr1}p5NelRf&3JxPIbpON@_K+XBEP16C$g
zbh2`D1AqN`309AOI5C%?u&{8`{N!clf0m5q29@EjQZuqzv+u&W8@&q^m2@tJ7`aLo
zk9IhdhUY3injeyjVO@d1!gD2I7sgkg#fn?zn5*Q)ygZYGr<5Te8}yO6Z4JS-_GDR&
z64N9@USt>lSyg2*ab!XS2UHJeg{R-NZ++2G>!==9rkifnxBMV;_^{xazFbW4-4Gq*
zqq`(?_nYvb)Q6On3L6($>3@?&ZLKA(+{!Al954<ZyQ@9V1kXrR6(b%oqQ(TjuwE1b
z%|cPS0A#H8eBvz2uAgx}MTp|`F*AB#P1l{TA&+h|dR8@oiav0t?Y5IOH~;F@E8g=7
zh0gA7c3SeAZX)ZnQIo5iy%JhFx^@E=IV~+INl8gdA{4NlEI2QDwjGR*kJo{-0N%#l
zUVpLC)xf}DLK#^F>`T^5pZzh2kXnb5dDLocR$Nb?_Va9o@$>_)?drGH)z#d>!cGI0
z?;r7dhKG#-@WvI%c$Ti+HH&c0it=(%QBk-w^XcNw&Q4i%_26EE{A3C)u6S6>V@hr=
zO@aQK!OYPTPQkBVi9Wz1=jP|Ht^|-(yJ%O`h2u#*LC2S|c|UNNb=f<@9{Mu0!d@9*
z`SIp1CJlpjOw4Y(vTo}Db1jlv`F_0XcC0~LFfp*blCMbYz8n2RRT^WT-nS!57YW2w
zRD;Pv0@I}uBwj@w^h0MU4YN9GC?(AAh|^1ma2i5H7h^=hLnm7pTy=ateXDkc%vB!)
zx(^`I7})nKiHevdfo(Nk?_cbEk5~$HFU14ndM`TKB~=qJ_!goUTT-}363H6~^@5j0
z2>`f5@<qvqb#jk)8Y0gg<VR-bSu7W$3SCx2E|{Q5uw=y>;Y<5~>hL~f_AI0gzYwjT
zzmza&Lh_|NwzMshiA);6&)oiQtvhV33_`l*hxJ*YWcm~=ixcqY*)j|^Ts5uT|MYXT
z((wWXrPO8d{`N9;e0&<5LMeRiz|(rUo$m^Ppr=5kgo=S7Y<b!Ku_&3<7!&UB@Gyzj
z^)*M0J)T-sMFj<j=pFZ_n886aRjAf+I?O-3-WxSca5<qMGr8ymYr1r~JBAF`>12*B
zudK9{e(x}gDxES3=X_km)$5I<thAiJXygLd?ga+MMmW!5DYx(_z?;w{<oBek{oqs?
z`}4Ig?BP%<(Zm}Pe@%G!!kO8TU_KyVx<weq<DWjh9qojnlif-rCXO}kY6?o}A~JZw
zceh5Q>i2d9;)m@^WE2r#Ebeh%-yzrM{?oBKey&Ao<#qe*b<SZ8y1_-gN>B<fN|87M
z59)Hro2;g89%>-EYa%>)Wq+-E`%&6xDj2V1_bJ;z#IUB_0BUHS!R*rx-hzH7<lT^*
z<;k_96<XCuM72rqfUxbnN(9SBMZo(y`X4U`_wO&4jF*}{LkS(N_7VpMBsYgshbv9T
zQNT-uj+WcrQ&O^VE`gP!wdHHcog5C5i292yHuEVU$-YIq#Q5Rv8tw^?Zc$1-N*5`g
zLhXm)f5pYQNS9aE*C!yN;lhnHxjU1(@0TI1RSVm95fBrH3TM8d*`KY*Q3H!2XB4M#
zzQ)>mh!f;9ph~GKlyfRjWaiPesb)#?kWdn1d*Wm`a9U$~pimRp$nzVFYDRD3W{#Qx
zpx^V9g?*MeSDE0|u=O!9#@?(1z9pMsd2B>q@%HwJim?&L1dF>$Y1$Vg==wVGLyq>@
zc4pJBKy{tApH^5OR4^c6X{46NFyc9GvP_8&oZLyKr>TE=OMXPnWA{i6Wn`qGgEOE}
z(8FZB1-lqt>dr<040JYSWM--XO>S(4hKBp|H4#njm*tLo%2OpeI*Gra^^Wu)qtxH%
zi_vsmaC97E1DiF{rK7Tig@t7zf^1};@$JM=0$Yd(sq1=DbCt1m>b-ycs7YZH6B84b
zh{y(;0NBP4U}V6Zt>PdQ)?Bw*Syi=h3v1b`p10E9oycpvJ#2v50uYk3z3GU+Konj+
zzC<>wQO+>wql?3Z2ClV#o3fTAeo*nvAD@b;o5hC5^?Zc1jI_!f+bYn}4-ah93dHyD
za`&C;leGlX0pn-ujNTge5pqK5W6$7YsV<C^;EUDsPv}4qFqiEQz${~)lesWLGlvoL
z*>%e9%r1TJ2jHcstfa?IYaK{JoN(0^2EJ(lCl?p9LC;Hnbb|JlhjS+I+>e-;SmWcv
zt;fwy4qPY+&mZu*Fz5o)U0D3DU%%Wg)0GQ?0t4UD(gtBuODEYcElo#^?VaORG4T~-
zZ<Yx0R8&-$TRU%$w5@V|nAEg=CYj6~1AfBODGh680C@1w)6+A9ot~YoA4ABtnr{y~
zrl)Imd_shO7e5LHD)$?(L`J{-D&0*h9(-bB=LG3z`r(v(n^s}!&#|(4@|``(p4abA
zVK#azCat2TZB``@rP8sUw*^T<OTBleHY$4KO=d|UctQlR#D>4I<37=`$}MwIz)h2d
zGX7c-z>U#|9sU4KU3jgILU9zWu;8(iK%MFk@D)L~&j#(}T_YgUj_mNP3TI0?t9@wJ
zr|$&SkUTq^x6!a~Rn>`ciaHvMbf}4p(MS%a|0jBA-a%jNPvIsvNn!ieJJ6x__XNsh
zf676LY75MaMf-e!aEwxACHJwY2saD}!_|*3TWQ1Ks_>3a-up?GIR`0*qG(G&RpQ}6
zYt@wz1Xs2B;6cZ^Dsu>Rs)wMi?pyoKfi{y*<+^RpjE0ibw%Gp+xRV1V4wk_!C0{0?
zlMN#twkeY!lx)7g&}hEdr5_s`1LvCVWWExJ=Q#%ASp&H2H;=$>V7mn*JG1Y3{WJt&
z-EBJ>(}eanrj|Rq5KD?5wRRhZIc3|4(=%mO<XK(UTMO1(%imJl*6Pj|xt>?;xV(7x
z22aA#l#-`KO#-?~@+K!0U;vr+oaVR>JS*ho6vxTqYy&Kpruf2VZOPCl%lQ}GyWBt&
znhXD6KC_8$QlV7LO4xwH2Uq=FrUgX>54P?r;mn{WVg3CkXSr7|g%Ed|D$Xw<G-iJp
z8c=rclRCH%+(B4XzIQXgI$RX+YL{t679DxZl4@lHv6ApWtAxje3;RY@g*yzUFMGsI
zPafuJ|3Kw*OWTJk0dreuUkv_p-?4MtM=|3b1j+9+t=dnn2m(6Ls^t=Z%YlJ_Eo&hw
zsvI$hfIo1_t1Mm3wmY>r9u$ct^h+K!1-FW9sp_>r!r#K|;0m;v$A12#d7B^Yp?3}1
zE%CPVmaAz<DtI-5+$7BNGWJ(BmhV$SnT9hOlO+_K=R$X-Z_AoKD#a-+<w*KB1z^vr
zjFW31O0%ivumF=u6t{^$i}JH(BxqWntY9AQEu0*#z{}@fURt?b%o_Jo<IcALKo<NW
zh)Pzrwp2{{)r<aw;4wXevgO4k&KD8&x?k+4i$suC5$+w#7J~#&7?597UAyC`)077B
zzscC&7k)?>s{v$Qx|;*>jptJuytn%mAfC_c)29a}Ffx%{TCP?@PAXvuNap4h;cro}
zNRy=@H(g8R28YPG(S|*e>n_hP+DCbwqY>dnch#+De~Wru5z>?xTWY<>;?H|cCDZbN
zV^`<$cI69p#rYee#oOdaC=udAqT1Iq49@tkc}VhoBPj!GW6h_UbIvKP&6fB#8x>W3
z7L9m-_(`zpV=7|j0WZaUTlbrGlD4n#s9Xm0D}v;8$7=i6J51pWi@m=HhR5EqD3%;-
ze^PfE((G@0v=qu2_02<{;}Z3Guc6SR@>P&IOYS;Jy(}WP3VDaw6lH-5-EDi%%BVA^
z^r*MQO3U6Udt{0f%FwDwh4?UpmnKHDIreZ0cuc#}HRWLD>`5RL9L$%Pzs$pXy<&qI
zdX#(r7HbP`d5Ixccu=MYI{at&Px=SOD3$;}i}boGMd372glP%c)B$Rvd0M^d<ahgJ
z#ppNNw2}<yC^l2+zLlEFEnE@ikgP|W6<w<2Okva=^<nw;>t`%XIpqiaF~rxN$P`A+
zfWVs5U_SW95)!nn?(dP2c4Of^0*JP9O{as0<;IH{xSDRfCyE>3+h&R8kN4hLIiUK!
zMQN8ZtE%^?tGzl*Jm<K--b(%U?VI9_8>k;$+a<DF9=Z(9&o_9u?u=zsVJ5k`E+h<`
zLCAphyqpvkGpdrZZZ#5YZ0u~w_;wAzx(Tth-f=IE*VQiBLLxT_oJ77OZ%wKQy>7PE
zE-Y>k>hgT25SFqXA8TZ1pvh{vboB3ywk$H(2bW9ujqNs82^O!d_;CLqNvl$P!bje;
zG=gvxWa+%qS~ex4eR*STMzTNT`!lH1f=8&k%b**KMlQ3++efinug$`;mgN8n#L!D0
z6yskiM~IAdZ*VE)8d;+@u#Lwr4<I{t9xV;?N&;H>dR+zfflKa~h3g;tko_AZutn}Q
z(g~Hb($tdMi(%8haicl|<C%2z)W6IJw07|st38|mS%qdyXgkD?_K{S=w>9RJNAB<H
zJ!ki?j0VH>FZE*XkugxVM(zNHQbrvAdWkksPea=8##DB!T!U`PynFUs$>9t-h08td
z2V<YnO`q#Gw$}N|ql+_G^m=!RC24zlpVhukxjB&fKw@!5rcWXsBDOxf<@NU?g-reT
zV@>_-byeHmkp8aHx4-WwW<L<#GJG}WmFPJdQo2~O6X2x&bLhiHJF-*-tlH}&N3+bN
zhQ()PjJ@>`h?t5~k4s-nMjUqW(_j$C%yS&$^!bA@r#!Cu)DuWazrmCf{y$8R3Ou-8
z+qt;7WF2_jgXdTz<K@2}XjG{WqA@kfKDf7dIeX#Nl$PpF<jNJQmigA%Z?S;jx!xI8
ze5s0qibx?64j6z42lS|tK%W2e1vb_8f{bA(+fWp>Ok`@R47jbb;J4@fcJlJ_sz@tj
z-7(&^y0f}Z4-|ek3PI_1?(_2U#7a-XB%$1-jH4f5#!*W<H=0)|;H)%7n+jk2CaoS&
zVZHH)pGa$A!gpkuP)XE#q4qV1yWUTRKVtbfRhu2ib;^^^`c@Qqct!UHb^RoObwp`t
z=)v(D>LH!vGC#bMacz8GS_GPdoj1X;0f~hE?2ktluWN^aB0}M0BSSe_37rH=L}woE
zb{=^7qbimZ+>d;4+&t|yu`?GN;`>k%({oX_2#XyD3m2F(ZS9`ri_}Uu7@au$Ok8`#
z=aCM8;DPDSHW04AW(M)bN-10jUxe7bW>x+Cz4{IGU<G~rt}P=tAv=s=#c#8e0in5N
zpbkUC_QdkrFo(Z#eXvR{*~5Z{t?81hsuM}Ko620ZAh+(J63zISm&cNs#8=`Q%fvU(
zDz=zzLL(JdeS2(LIOyby&*7BU`%22q7ijtbUgNCpvj?I0>kpo`f55lgL$$sVTPv1S
zpueg_v=IHuvxU01q!w7zK;e~4pW6D@bW6aa9YK`$6_&Ze3Ns&Tkz;ETecIHht6+<`
zM4{@8#^~{mDCWRCI={|Oti(N?SXU)fO{Ofd903pWU&$X(!0vHb5l+IxVz(~J?X(Yw
z3JAc3hlkU=K{a3@0GUC1yC8_iu0Zkc8U(BT(?skRsxkNe{{BybpFbu^a8(r>VVM8+
zwn>WnA(wO#C%k}x!TaXsX6K^Y^J$$5y)IO^Cw(u`y15q=K%S-#>d&*smtn$RzkGqm
zW70MNXLwvf0*mAB1kX)80BT6|D#g!rS385AK=|bnkm~IXH!v^&9_D`Y=8et8zM9wV
z0f*Pk4s~IXdCH7IsO{#~)_8#m6Day#Z)ZkqPvl__4Gr}eu=Mry1%`!fWW{RNuhpVn
zg5T8L^D8P!p&42jf8|Suf|c2ioWb=8K->ItfgV?#*|&UnU;$_)vK$p@&2ZknZn}O>
z9SbX1`;F?zz{Hh^0efIX+~v2>K83_uO0KNozSI*@mwY|nqOMefFxl~kN%@y8vzlAz
zUFN0|va>bb3ObApf#B|ZNDy_*uE2t{>hPoSBlk1L%_B?M$+m@gwkqEezt*AX%n+wV
zI`ZR}YM(AND$RhfH;O=s+S4DLJH}WE!nqq;zo#cJ+T9!%5zoTHrEE)SFc()5N4_I3
z(x30~IV?fXFL&Mj`K=mgLLHD?4r$HzYpAhj@nh2PO~p~cgp@wU0%o7PxvN(emU?9d
zeY6?n;hOWokLv%)clv(Bw6f64p49LsWlT*>@Al&&V6tvxcXghin4g~Q(0W&3^B!4*
zOof|&sitpJONW$(&(!8E5iXVZ-t>fzg{lokT^;THB8lOI7y;s!m(SD5x!ulateehp
z`P|N0$H%cHBqTtc=2x#lTwI*4v9SoK8Xuw@e!hg-KSxy39y2>x@1=<m?+zvK3k$<w
ziAPre6cj*rj??Ro^X+7UR;linpL@4AH{f1tKUo5<R*;_|Ae5Ds3Q$GGTyAY|i-NBg
z_t}%Xx3|~%d<_&AKq*M&;T9AlQBYCaq0mN;$NhtYov9+Cmivpu=71rSwdbHO@cgSk
zdfVh=)!mhbs_IKZ!oJ@WB5<WuRYmNj0OQ2}(e#yJQFZUvbcYB?he4-=fRunV(mAwr
zcSs#lMClYM5s>cg6r{VmyG#1rJiq_THGCOhnAzv-JJ!9{+SmaI4adjFIt}i};c2sG
z+uy>%8h6i=gh3^>x7hp;K7D?la>0t$7NC$lUV3M>t#muE>eE@I<kUD(UE)Hsl9dIw
zcfZZ1N+XjIp;Y7M`cOmZl87PU@y-+VPW??>>rH=)%s|MxXLnPy9#6oS*hUtCeDMq$
z^3O#Jb#N5J=R9Y}Ks?GEnb^I=bB(NiE*b86A5FTu6Ib7R{P&;e4Zd=H>Ik=+l&bWk
zdTKUFMa#xVM&6%klx<f~kj;tZh?o-29Rn4X<!U>zefGXfKI`2iGIV&FlgBq9PWt^5
z`ZGF1OT$ilCptClw}y|oIixG~((jo^yjDr9ED6-?NCIl;-@Ntjn$LGz#*G|0_P<8=
zB*|IiJ@Aokp)~5y53nu`JyfYpkCnUNqd*s2<}OEu<ONdkJ<nBX^s)7I0^=CIYil2M
zu*Kx1P!G2ae5KoH`kI)<$J~gc-z96JlJ{#u+D148kpe5_nE8OkJ5NvYzf2}Nnyp|4
z0qjn#DpBAQhw$x@SXS}u%#7no2fCLR<3sgDqFJip&BZ?N+qV-Q1#l|A<qSbZ<8q-|
zpT4@c(8%w!Iq;_^Z>Lrs419TAc2q0(1M3WjPPYKIy}nM!sI0^ilh<&)lm8h@@=4&0
z;>dPyri6ZbHyu{<MTXTlpp%=CRzoD6l`6Ym=6${{J*Nm0kXhEFXQE%;yKXmN38oXU
znhXgb5jgO(ReDMnf0%VR8$sq4>G8JIIfB5PY#aWf@|Mmml;}2QAdx$G>vDj;ukdi7
zFKm{O2AN@&|D8E@Irl?hIX==%BwkEkT!w?1P|rQ}G{0v?#Zjh4e|i3cO)|t|-Ahe#
zP1>`f+*imrn2JL>72ajmt$8`|pvWfgUr55<r;D-S36pA-4Z|rSX!j%6Ov6C7wAR@t
z=Dr~m@IUl<f8@$H$&tzN#1<oW!p_EByX75EEZOK6nA>;xLA8`H*kt%!sVt0mmnF_N
zH|$ubgznp_tSkO&3wDlA>vvFK_?EvMtx>yYb4DVB{U%FxGbiI^4O6C5=7hA1&!t(r
zey(Cc>p*+R<5jgP|KaV!jd8Y|D7^6Ts$-g9?{DDCTK0-4)tl)-nY!nhh{`fq=;IUF
zqB11EF*rZ_Trsx>@Q)p6KBk>E@82aQk>H{^>`lGKp?KR7gh?a?Nl8%)3sX5cIo;mf
zp`)W44*e9tq7)!T4c@^Q2N}t(;TQ#A!+l9S5div&N==mq11ANct5<A|XHToD)Dk)n
zas$OIEgAV;_Y434PcGmd85oE}fPq|8RD^_zijRjU0tgT2n3(w$X>_UX{s6BDRLfT{
zwVi*9Ln$By$;;z_^cNTu^xA&uokS>cy2d9Iqc9cLgi|mIIR%|&iO-!!Je&RiKrFg@
zdo#eC-g##%WH>{T6a0aaM<pfq7yk@DKfk^-VUW3euPMKR3_I1|)r$BwLX0;5P7E17
zrAr!qfx_&)@1+o7^JTIsVr9&?*khP>R8a{;A5ue+y(`#4R(;yJS0BY%dDt;QFvI_I
zI_6K--EhSobQvhKZbFDP9gelNit#$AW1sg<5y`#MvQU*w>+{(bWW$#vM&`-;qjy*w
zoE4nD^Atk_vCZN9(qJ}$DeFgfvre4P8pk$@q~G<#%Nq`o^p|x*o;&u|^LY=0GFO$*
zlVzRykU870&VFGQ9S6L+#qfGzGW}VNo0;C*O6}1sgbo_-n6$qRZp1j#!lzZ;tG&S`
zSs<x?re}TQNP`~yF>|KQsYum5+}II+MqoHV&km=3ruO;)ZDD{j6P>NdGcAR4(!sPJ
z-dVP&J$iS9+ro;`#NNu*<O$C{>HEcP=%ocgjv2NYNG2jR5n9^XqX4%6b;;jM>6p+N
zH*m25*kL%3%vX8PJ~x-_hlt7nmOY^Jqm@<fzvERBVIQxs$>Z~L@$7u9LOQW73q!+J
z(Akm**13jgiOzG-w|HKkZP&AAg851XY}=uo(`-QXt#p(eZ98z2F*AcUN#_aikd|9x
zJ?=^24*VLicDi1cRFCPOaR+q08jCn_A~Lnpl%&Kk&!lF0j;$7=?15BA8A})+8T0+Q
ziAfy^@3J<T=TwZi<yj>LVuJ*~E01O6Z#^llg0!3WFu7l`h7<<6H(0w#ZJBU~l}1GA
zC^1y^Z%&Q-9u}57R1sc7+VYN9@3^jV`C``yjrMnZ1FAXU@e0vNiNnkF`#Bv{dWwT!
zrjCnaBPay4=l6*}l47N`3>lDSW!7I};hCbEol1u_soVGxwFu_CUiC>6UCHh3uojeV
zYsFm4-cW=o`d20)h6@#O?N0yr>9=ohy$9PhoUWPSbnfJI_WVQ*4L^|6`5bCF7MUo=
z>iPyXV!IxMbM22rx7^zt<5<CV_aC<|c;K=I$a38#>9Dfl9(@iBBpzu5&-wXdS**}o
zNw$|o^%at{t7)o4rkaHsqM!e~ovS5%PQCjP$HIibbd7|-{dM5*<>rOZLt|lu4nz6S
zaEs(G+t5mOXm#10%A{lQ`@6P3Xq!3lT?0%VxW!L|Uo3j`9leGN;-OpRJgei3@{1uc
zVaPi7$jJ0^6wyGmi+;C15GD6nLHpsvw*;0Uq4m3Cq~~ek!XMxKjj@g}TYS`En~&op
zOaFWG{zrBaY{T~FJ8DL53}|hLa;62-%`R`ZC2JsPzJVarfwY&(j^`(4;lR)CZ4;Lg
z;kfWF^TU^`&Q8tr54W#(xXtE%rY-DzOvBn<*|^E6>dNwhLSGZN*ViOC!zegn>Y7ur
zge3p>zeF23J$xtI+uNsfG5n9<${kAS!d`tX!!Io<^!}Yy_KK*ojM%JLLV`TSo^84g
zTA*i()G~E9OY9-@R@W~-`(4)d^w$HQ$XxhnW^yyF$yK-URL#^f-k$8Yzl#1SRr&VO
z<zJ^_4<nQhmlxMNUo#w@>RZq6F$}Q!k`wZhME*%iFL30?B_(<wTh!tRweL5XYuMPi
zoSLqgcb&f?VGk%(N1bLQsEYS6IefNB06E?t8UgS{43xIw)5zcdekgIM?BXwAV0j^i
zzBl<yMZCXAMVrT{DZ(~6gLI6e2$$rzI~nj;Lq6w|fLqqk*2WJYT0Xi6;J4#xDSnWV
ze_`qA+nihMs;N1}+9rx&=;@KPHDyf6cyo_*fu140$i8NZ{V@;camPPFg3?hscz8sf
z?8Hf0>9g(&E6DW8Cwm_jvA|TiiCm@bD)T%(e@b6IL2Gh2kY5*rs%P9{buhL~YPejw
z#(rH>N6Y80PEqks-z<g^Y!WPZw8~%8aR_q<-;R-gtWGn%;j#Mio$I@+!g=1Ca(P3e
zkeK2mt`H3v2^Yr_;H6x>gVrajC7JX`K1MG&Q=YvsC@WF<q-MJ*GD<&$niztmEBbmZ
zw`IY1>7T{+6!*bYn^Z<V#h`)yuL;WN&d~+mcc#wI=_}nU3a2Z7D&_M1+kFOFuh6m_
z)$0tb5+1RPi6}S0Zw<RzV8w_pQsvK5Ig%&D_|?e@oX;G@4XvcLAi3wU=2-ddo~zfc
zcH=b{-eZ0MIs(Q0*K)G(g&x&6aWcmA-(rnB=wk5F`ap=cm@6W=TO{+|JypitQ6{|c
zxlaNRW;$an9(%Jx_?RU{t#slVt%;*;J07{WPRQ+`5fU~U&qDYkYy=)I5=idIA~_oU
zp21}gn;$wLD%kscAC<^%Im>Z{TffDu`xjzid;-gqCx;6d_$que?R+Cx=T)H4vD8MP
z*3}+oU}84hAA7*7J`E+<+FD2O)}ijeA>b+c!@-i=9c)8FaM-<)rVt-(Pyg9)Q9#4y
zjos-&I=Nm6c9mJ1ieBYxngBO>x45mX%&4u~WK%O#n=+b@+cFV4gX-?mazWO0@ca{1
z;`bj*bD4OF2`Yw?w4picDpH9u4DR$X!}qypt2oLEk8l2J)TD$7GxZq$bjJJlf(MnF
zaGI!+Iy@6y|MJYSd#>x@JD%;g)+S8IjPPH%?KyDYWXHI1B~^~zcg4LdeFW;(HsYEm
zKR|NpNbixHd{$pirO=$~l67a<H$Aa3({2#GwVafskP`S5LRdVIPS2@d*UH{UQTNur
zIPNSlQ>?BDXR~ESwTPnmuU&>I`@iyRbc*4u&RGVjUX!bihiI=0#_vB`A|*yPb$|93
zI;rNw4-Efyc9b_Q*{Q5`cCX_BT~t^|Lp(GY^>|&lr~x}hoUZ2@9hV;9v0Y(g!LY&!
zD~d#b7dHyt-``i*&i?>i`^wS~J5?G!Qfp0v*c~uU4_qrZtM&S)aLXStS|3sI?M5ne
z=1WTbOF!?Ku~zyZPZ2{S>8&rU02alK*9lIriW2`(=nrt??>*>LDlR`iWI)R0dsz>T
z=bGg1u$G&C>Xtfxzr=T_sM}_lbY9OS`>Ir!lh~6=F&8>IDo#S%j_6+UmYJj0#dRNj
zA>$o;A$B(Sihg{^R89d@uLB6S6ZzaYdIM{zebES9Hdacs2KvuB2#4v1$QCd9&-EXK
z#!vetwa%~EMz|*j39T)U$Dysu_lmnm7{~2%d8RZ1J6{8;G5b9}+dH@s6et8lRieT4
znfzEt=&hO2C>Cmae)s%RekOcJvF$LKbvany-%{^yWIz*aD?c*0?e%ZvMYj5Zq?S&n
z(PFcv39VhQLzJ@bWyC86dJCm*2CH@<KSzgsbPhe0C#z_c2Xsx2ITxXHRyAn_7P+&y
z@X>u^shD<b3#5QKgAa^9t|m1fE)aZ4kw3|ve!aTOwad&#ht0I`_qR0q4?ZPR*;;14
zBkvC?Ug4`Vr|bDhBi8H=8`~m$3~D0gEuoZ4x`3_27}vApLkVVpkNAbssKygYXAY7m
zJHgJC?B>;>wBi+Lk1RaE?2p!0LWXWxlgkD=cm_q47|D~M&-t6XVtm!hXnEWW%Fk@f
zx_%TFjI_}?DE>;HAC2O@R@EK-^jP67*#e}9Hpi5ty}xj|xzG^C^#}so?T(*$x~XRM
zr`(YayQ;d$GtVQY=?1cCFZ9IchulpA(9?+tS&$YRDym4F>x*4b2`)G7@%r~xk2FKJ
zfA_azBY}?;9JS+&5jHKHD;pBFb#TC-r6WG4p@e|fjA)N&VXh5H2@PVRT;~@my@Gnb
zo#&1-Xe*Z~V%5}mu<3V^HvJl`6H<pS3&U;kQ4F^YR7_5rf3wB<*QdK4Ss3*D7aO{=
z*K;>~g2GQBs`i&t39M{q#TB&*+7QA`)pG!;|D&Z`uq#?b`bsupU<VGRcXcLR)QBM5
zm6z7T>!llr_pev*Y*$$rKa2r*{vMfj%fp}2<2&&h?#sM7odEf+9}LYht@GhHM}|)9
z`K^3AOK|1>h9{pu=N(~Yr9Tt4ojb@1Vj}by6fYx()2er7Dw%ZaU9-Tq+wDB4HQ(6G
z4E_E<#>{-sxA({=W@g6XEZ6o+1l_~(?k>PNLQ8|`(L@nfC;1{ly^+qyVsPpfWvnBl
zlW~4By7F%`DV1PTht6F}vhB=WN6q}ZH5uR0md0vUSF~_5T0MtV)ki;4XdxnK3qio&
zJ0$tdubvHXU-Q+uC&ZRsNMqjhh)m6EcEJXs7OQ0=E;Z5@`9ZO~g{^dTWLR)U?EGP`
z2G1Vpv-hc2C4L$*?!mh$Bi$_K4U1v>%1qPQ5GMcXU+Ib;HaBnZp5W<}pHor~GN;mh
zpm$-ezNHljvz7@?On~2lxJ@)Gc?b^(K}CNCxpC!G9_cZY4;D^;O_bS&)u{VJ)#}bg
zFSaluK|tT?r)&8DPsKIKRz5MV<3W|!Gj%$uKa%-bjrew&m%Jg_Q2$L|4Sk$^Za9u+
z48lXBe@Jd}7xY8Z;@NFu6-YOrB$b*j<3>MND9U)*Fhm?C1GUEwCM@R7YWq!5m_0)L
z4a}K+^FNT>>`p3tA7W@-X0PU9GN)h9i8KGpU;g%NB!VsVR+Fr*dsBYjzNsRXfVtRv
zNl6tLtfxDMj_=dSt3QAK03OjuxtRhSIXOA6^VSnE;ZGKZngn<YlykSn^ND1#@J!{O
zA*GA#LgyRoGogDD(t3<ZO*^eoCU(lv8V*ZxTFJfJ*l9<Tbjxd5oll?O^$oG5;zdLV
zQG*gJ<H0jSBr66XIZZc~lBD>H4CX^-yR4EuBkrVWYUuryS|4-%fZqJ&>7uftacBZT
z@<>8Gf`n)K=*|?+<{@L!4=qObm4r!qiyc;EXh!d=Kg;&W6q92ml2BMc@k*41GuG{q
zG{VjJ&$BOboiEcuv&4;aRPCHcv~r!_iO|5W>k!_05Ynav%h424?Y~c}rp))@$EEgU
zGwbYwbQ!B1*irSuqY?IYWDv{}zAgo5{#)iScMcSs2m)JiepW;yqrFDY+4~bUL2ec`
zblBHBVhj;7jJNF&STwUeTb}Ps>+*5WovF+Be5e<JxesJ}DT_A#({5jq02medy^VI<
zhr(p*PbZU;rA}_KmdEm6{2R4?<}Q(M`%a0vZzA+Hew`Z&TwWAVSapTJjwr<eWuk-g
zv3#!336lPue?1Sj8!DCTW99^q2o(2*Nm!CiP4PaR*55ikjVbHu5BZ&u84C*I<IeP_
zp2OrwiDHto;v#+PveR+A+O<7b%LzasG1lULm|5!iG#nh!^Yw1{PoK5{`T;vPw|JKF
zaGi@4kl?^;mYZU?2jlW=uv>k|9vyD-_Pjq$dtkRzU5{9Q<Gd?Bc6b3POb-?>!p}(r
zhexkG?IYay-t(s+;9Mqqb!#OB42!VylxhAN*KqHL#wehHFceF;8JXnhS5YHq&TT2T
z=2+J-m8NGW^tBHSzlB~+W{i=6__rcG<K5oVgRmN1C2RLfH;a?!zO?Eif$!u^mESRD
zsphSJdO2G*{&JrLY#NB*O}yGgKC8;{hnw!43P#Q8BG+>Pk(mRdE+tJE`~F`!2E+;s
z6J(mlS7KPjYf;9r*=v~>B^9f}U?qq(TbQYtdF|Zk-pFa`0;BLWwJ~^@`(Qp!x?Or;
zpnX&s%Mo`v9;t}6i7ArZu4*VT$U~LYdwNHCo%9L}p*7jBA*-)twmMD|j^;+VVFb$T
ztlH?CnAEuhaQL<8-oGO<N6Wo;9LChX!Q8&|r?F;nJ#>DEHk}xW2uM^7U%n2zjFQ3R
zAX!#8GibR{P~|%&k;2dS1`}1@A8jK?Gsp}g7;hu^gN%$c^k(lrRZbcxnF?8D4{V;a
zVvbN*%Vk}H`Cm|M%8#r-C*+Y2I4#3nam$N>5pHA;Fq|WK%P>-t#HX@Mvc{Ej+vU5s
z<*qXiKbx84xD9m3MwhaYW+Ha`TX54FcGu^2wQk3p>S9CcsS7SPw=B^c+)*y$g&NK#
zgDa^H5r1b^Af>2Z+r`GksR>3%<k->B{Q3SP(a@Ap1_~1VD}>{o-ZS~_+cyT*_*7Sb
z3yY_l6Dx$ld7Y1IkPCXmj^}GkUMeDq<h+!KFfG(;`~mo|h)2C_&8j5SKxt!mFjtF&
zf&w_xf67dHiTPcbO-)T7Nfv+qO2Gjh?u-FKgFp4y^Tbdb1<9veB^0PWSHY)A;W0kj
z)Z0)3&cbjX;3l*PQEb0)cp~3anXII14y<}GmMbar?awYLlC?Y|3j(=n95-LK0Z{Gs
zWx)3JNed~eee|H7h?+9(ct(MZJy9l{O%8<#-H8;RTjk6?lQqjHuL-253LQ=c3#P18
zUnIQQH;Y_ZKg@$NH3+;9^`sCTyu^@DDAf*&X0McWFReZVQ(Ioyk(O#;ZE<*D=1(|)
z3$`v!-dU4{gnWQc_9A<Ie&hzWKiLg*6}+;Ckc)HD{GGAy=y2hyWHV_*kZwE9ULe-4
z+^JqVd2(|QXkQ}TUSqO8n;-6+lMyMue-7Y=NLBevlp@NIxUwbww-ZA4Xv&bm#NgND
zKrF|Y@+=N4wG!7-qrJB7jzj*z<p-Pfo+!<Vn*$+9fEt@~=JDG|x)on+^fszfCLiQT
zh8=$e$a&gRDc&c#%f0A4)uY`lPGeI~=Lj?Pq%wK4pD=ah?Q&rLukX1`9d16p2b1#k
zUVzn&SK*@7oY%~y0s2w*RH+gC7SNMjEnXfj+I8Sz_zw*Mgk<#Lj&AGnz<lIm3~<NE
ze3ktK)6;5Cfc)5eU3)*Hy)B>>e5~q_fGu#<rZq;o>!M{0C4@HOF|0|aS4rmHA5xq9
z^n@8ix{W6`ajzJm)173(esJs32>8<ZBnIYXhORkjFvvl@VrsC@mER}^d=4b)EtZ$|
z+m;4XO^3RSqy?d{hB0dyr*9?BC9SQqi4uN7xZ}o_*IQ>g`fmj%6WY!M25bdHH{w*s
zXc)=8bdC<m->+Wu?|6RkC@rb}bvd>S+Rf0=e*|EjN#Dcx&d|uu=B1=SfOmfX)Lpdz
zXz3eRsG;=lx3S5Vhz)RmjgijKXn#twgJ3gqL;n&tOeQQWU$6*1b;|QCiBQp7<@-;|
zMoXnmuTe2$WP~AG?+Go{e;#4ok)1za5<ZsUXh;9+F}dDJ;2s?;G~TFh{vLOOJU<`q
z@4~dT6WFq|vPev<K}gp(rsbH^zFt}w$%dMOUvR<6f0(nk{OEpEeyij{_VO0rJtJa9
zPRQ6ax24+Ksl7`P8$_ej=O2!+ER;8AJ%*-grbm|(rFwcYXDWZ2{{Eh{ou)Q#0K)W}
z(~io->{i{ljm4S`n^47`jm}rWSy$rYTn$NW6Hl85#_0&SUafZx%%F7*bP|}Dl0dMq
ziz2eLEM%7>tpqk&82<n2_b;G+=jN<ebE5~~wC7Ep*&Qom-MPvAg^QgT_)2a?2e%cj
zAX7HkwAm-dcK+VDYjD68cLSS9j*++)vd)ZW_?{IBfysJCNomKxd=za8Y-yrTrI@sD
zf~WY?%ch8el=GXW>``+_TEHbSb2Ncos#iRJLW-oJU%+C98_g;Iv&zkp>x6tY1wRp@
zE%BDG4&k#4bwZ(W<y4C#s}85NS3<l0z11hAFF?cSd3n&^6UPLvS#3)(C~ztlx6flc
z=iXJv6Bj%8<8`A+Z@dI$7D%T@C5~qfanDxH2cI?cX8pqGhEQbxKbe5cyN-AvfqE=z
zEr)6E+J_<m=e?|p6Vnoa(+=>dp05qdbN{FqS>CwWiuNN(snhlz%Mn2Q$;`S6I1=Pq
zH7P5}Z5XqE`#+#l&iSPl<Nj&-e7C@a!z0YbN*K`WG9H#6t%+bs5xzhO-SfVAGr~8r
zv*h?1;;5AlVD*0fU$Q@92hf>YTOj&6`~s1_SCs4i%y6DsKE`b4pFhWY<%97L_dQI-
zv26Oaj#*xOFJ81wwG)6vpi~C&p~ST+Ys)J=Hf}?)m!!Bn_KC`u!X!l7eycaH<EZCk
zVPWw$oa**#mlPvZ$$gpV)jwu^0~iP_pd}9m@@|?N<+r<wdM(E^^(5Rkah7xoRwJdz
zj+`o7U5@E;m|tNN;>|ip7AEc$mZ6oh=Ohi28BaBjd&t40xAi@Wq3e?a>JJmN2_jWb
z2MB-i#rGz1qw-ZB(+Aw!VcAHy6wm+LdG{{DApD`0498eZcyh+g`CVQq*^}YAHYg#<
z6<gHwg_zqb)Umhhg(m&`TcUqehHJ!}`U<fNHktlc1&yIUUgYzgKfdsYQw^Uufj`*N
zk^wl!wF^Wqi@BQD9EqVF9q;3rwc7lVF+nm54b4<Y6IyS605kdZ-p3;7f*nk31A}Ab
zY(?`WXPlge<u&86@F6d&$>B2NArbK0x9*-m6v+(ba<4Y)mCMwao$T;SKR;VS)T`8U
zzZ|dco^iWoPyf06GZQ%7)aJe6&Z~ex6A6IAkOeP<dzql9%7WecwYPAEC~i5wpX<zd
z#{=j@Btun<NYcHvdIi0M(j{1wyFf~!{6!Ejn|?vZ5VHWOkak{n|JL;{ajZ!HXOE-9
z`uVr|wVU!UKki&`#Ohd68?Acyn}2RrG`fhO%8u+AG>ZdB<PvG}%g08Wowk3XdDj!o
zBj(aGvniNbAQNodxNhSD-TL_FVxLRAZ(QM?vFY^z2@;cTHrOTBQzh{DG1}8qHZv85
zf4&kza1~aQ15@o@TPbB}`io-VHFJ0Xb!=s<=Pn0)CQA7ynP$4(PRJ#%vB8H0BXLch
z00JaL>wD)!f~nK0z(JU=8FymBbJrs=rhoSwLHrzA7bnmspK*T8DK$#3P<JDGqN>#Q
zmzypOoKs!O%_5Y4KYLG4f&6lxtEct`>aY3Ux5MKT<zLVC=5v=|dfpA=o8Lz|M&`S4
zX&t4YXX3r8!@JPbxa-tJm1F7Ky6<^uedWdH1(7siA~*?h{B&ty(|E2Rp5MCfV707f
zF`8;8z^IFgiJ#txShK5@;4Vo^KKQ3~HmIa)_hITWY=H-1{`#r2S~*|_LAw)$*0WV%
zM@L68vq3>Y$AiKTC1!(^%q)ku3L+2pcO-m{uK-V|;pNePx8mnAhe|pC3yc``Qgw;=
zQ*tx15lZix19xVjlL=!L|Dks}Gf*nN?4@rbNgrB)P>p9yEqcQ`?mop|87Y%ZH>-T>
z1JWhw$pjS9QWT*tLC{#)4_ppLs-`-mmY)%+ZGB}Y-JJka2fNX?tOhIfB)Tlf>IWB<
zDJm5gB0foy&~=Re+1wA%RaBCV;)6lK7nZ={6_H%fu)a0ecA1I|kBedPemHfH>lYOJ
zF|@qBp!oAM?ni1Bty2-o!Jx#1sKmeSYcZvN`$JQSUZD9ykLy7!ogyx8!a)Kk-pgR9
zw6?hk=yB}}jh;a5#zw@dL(Rq(p^`2C+Mor_*rRXeKN3ynkf9FSO!xr~AEW!KfVF%A
zEq-381ZvXAa{p;$_OE#rL3BuY><bYoaY{JFVnI|q9e0%F=uZ24l!s$fCO*J4q1l8<
z#6kU@jaT$6*MTCA403NVZ_2W$jZIWxno(GG`o7B8jkLHgOZ1aKQ)jp<4eC8VI4k!V
z$>uwFP_DPnD<$Mj@mm?4?;pYgUPgwG4>G@-<o5RuVeviOsx-3boUCna@8t+hLx0?1
zxKfTQcuX9oD1qV5)1r8(%`JQR@y#4a4ikFS>RI}8WCZ-|pIfoJ5cdC&`E<h?y6P1h
z8ykz^U&t)5iW;J%$K)8e(`_tp(#>l<Pd>$Ih1l1C#iT+#BHnZD^g6+xI#Gr<1(%-#
zG!VXeF~P&$D?hI(nyFQ(xoJbjwyc)SkOh;$Oudl=%SjSM3Z-3K2>*B{QYNWqjW#ry
z+pj;3C59H~hD#o!<H9k%M~F8`Ht}TJ#QueX$C_{>7N`Ocg($`9Ukm0_3~a02WCdk6
zTBX+B*eQ6le&zzKvG4Fy0)BH1_vK19^W^S6=k?5Kyh$LPvA@Q6%P*=ycT^ou>&vVx
z_A!?rc0LZ>1ZEA*?~&Y{0w>5HW9Iy{za3L=&t2a5lziv3!f6XG9cq(!D_Llr<&-%h
z0!xgotbd7L7W=n<4eRQbMD(*;(dg=b+X>u3Mfc6~kL8c3Ah^N4xO>z4==x6`_cK0K
z(2KTIQ~%{q4+lnC<Y0h$X&>va=tSfuE{%ydPzm9FUeJNMT~=IEb2(ZtmEjzQBo&4m
zH0b%8bU%QZOb3zLOq15&#98WFe+j>ZaU3zVU$2%8ur#~VQvw$1l?Da~W~MR(qTAy#
zN<ow)CdsX`#R=8Fe$$n(XL`677E9je7r;kY6_?a)A>70^Pm9ymGb~ZO5AyV<vtBa`
zXqYZEWwDTk^Nlzou2M*gkIj~#2`}qCrH9RbEDaD2sa~`^Zhe6>(rQeD+p=ouFzFYz
zn<UDE`Y(k)V3bX9J9Ys4Zlh7me5{S^z1H(|_}-fD%q&~{4ol74r3kWoi3!T@vh#;s
zsGi`3LYL~V(i%ngz;vriFPohc!mZ6EC)4n3X!`*Oz655%m!!Ba>0Tp-$zZo$7+5ZO
z7RRxWcRxQ>whTZ&@+a`x|7T4FcAZFDI_6`7nd^}IT83FtQr-rNtlzq@r$gnH3x4zb
zpqRa|5&O1~{wk$OZf=0nSV!(RJ0}&`B~?Wi?$zhn;>N_`VY*?}P{98pqG$>}yx!xD
z4;R^OZ?_%bDVy7Arj^KL7%Q4?hOHu!BN>!<HR2zWXh_kWDWp0ntMn|v&{nrRWCql-
z4Wf5P80VbUH$22PknL^;`^&+jrX63vfdP|>kp?&O6kADJo>n>mxU?hRxATBu{rFGq
z56858A&rdT63<K9H@0(s0el6779!dCmJ+Dlf4&jzT1q|s-q6n+Sv3OVwO@=%A6GvH
zX9Zw$;IEpY<;%up>dVnydOQnXV6Ay`*;WBRMqS9A5<cSp6od05GV!~?uBrYi+*-hZ
zs>5mnW94Tz*Nlkk0u3umTh}g4EXNF7l^{IG|Ihyxz<<bhtefYeriS;?BLav&b)DHB
z-{xgu^qgoo+qBr;#VELPj$KwDA0wSvf0+8h%C${kKI9tvC+qff_$ljwkgUVNeC3<M
zlzp^4GKmM=)CN2M#_iHczEUqX2mZlGH^JTT$>h5^3p^5;f8Wu_y(t&EpkAAK+p*ja
zVBhoSvt&IhQ`6*|q2|eN-fgdnYOIKafzhaiGSzyQ6u35z56UMJ51pwA36ir_HVlk)
z+_e(!?!3#(%PX%nz|WV{vqpD#Z)epn1{S2W40F#U={S!{;DXSN^XZ$^-;d3Ry)P7e
zKUyRraI!rGLuPV=9{v&CY=V+#LJBI04v3(c`ukiTTRIDr?oJ=4_2i3%qz-f20TWsH
z<w2oC+y-I!Rk_LgLSg3d0OX@8I3lMZ1QHWtnoLJ~AI4xJ+w40P!I0z&;@{73^?(0X
zaEAi}iQv`Rq5j+pij)D=u%ZR3RktI@*3R+NwtoVD)%>6U_{}p-Ryd*MkJ0b{7kHOj
zKc2xI0%cWiKt1r2S#R+}tf`;$;`lEvyAJ^nt{~|`nAKVkotA@m4Ml9ebx<bH#X}{X
z6lbj^O<G6jN&oF~8mHsM`ybBMO6AxVmCTe6i0G(nQq94$?eq%fHg20pYK$z&J;5fE
z&~P%-BU42Bt#@KnIW4hGS{Mz}^SCUw{uehgZ>JL5%oVow5e7Eg-pDZouxy)8W;%iz
z`c-OO2Vj&Lu3G8~BNo$vfekfhRQhZZHX9d863l1(h9BGaS(J-5Zq-{3d=i!y?YyPd
zxVJGl3LfJ{8JI!#wZvVWMK#x+#BNaJ|J{CygL0ws%O~r-y$KqHFdF}Q@O+!t2SB+y
zC?sp0m%#WwO1fZ;@!^4R4w->?a*@ynGr;fwK#QekL1|Loqzi_t;+wC_3Fg$kw|Mp&
zNXVKSL1y+uEqzFDd0X5H4J=jI;~Wf6(cd)O)H}G&NPXov)w=Dwgp}>8GZ2}v0-{pg
zP~G4K2#yMHZ=YfKy?G_6YuJ4%{EP8*TgKfb9vBIVN;=2cK4X^8)1@;zSysdWL<uC^
z1M@lf3DNA$YwDP!mJ{Q@$~%jG5~!y|2t((8;Hf!%Qmp>FsH8=E{s_;kr#o{L%$~e4
ze43ikHF7&mw8aOV+X(5@e2*bo-Szz9Z@>=9{MD6}o&@b`TbsRuM~>xkljFbYLY*nc
zmo@xMnizZPU8*R>Tz(D<IWlp~Y3yWb(&K)G_<+$>f5-K`dnrF2FnaD>jo(2$|1!Zm
zpSRLd__enDgmLvw>VH#_=(K2-{oXDX%lNRJk2ugGibx3y&^%0^4M*2eSs{*|hNX{W
zI$5b0YW((L%8qL_8wsoi0tve(CajNp_tF@Nz-lRjmwwc!jho_Vmz!7L7cdXoBwX{$
z9sUCVhDu^%8LrkMXt*eNQi;s&!PyU}yb1O}eGLMv*_mYzbDx?t>ymL|7K~f?sh}q=
zeCbWW%xIOSXK=;U)oxd1I;=ddmydUa{UZGUgE2^A%37Zzw72wyoeOxJovA^@lO)>D
zX-HA;tt!@#?gfiMt$kNDIv~>HGv2!Ra2f0am|_5YeMSB?$`V!KG$GvZLt;|P12?KC
z=Rg6U6rk$Vk(nbJZyx;p0v+nda;P;oAXzDAzB6-SC1@kt1l9CEe7rz`ncRUcn2L9H
zL<iFt?ng|+olk!YL}-hZF*(`Ve-JS56%uoA63VSfJnZ~e8jB*vevgulV6Z#xUZIJN
zXCNV(i>{2oer_>>Ng(eTx7Ur&C**^bChnXb(nv9@kos%XDn<*cDExXjhx@ZgxBjE8
ztvn0{^Tritj~s4ZfAmsV??pyN?mW<Y415P<Kr|3Vt4g*6EAN|!V&K_s{6j(mnr&;j
zEoPd@@0@sgesP$8L`Iyq<eBQ4YyBXiq%IQhcSVrhuDs%?AwtzrRR6;CC|0%jyrnoP
zjJNDY>{RbLgmq&n;HZNC-IH*E7gf?x<gQq%7^g-DYf;yqXHcvvI&wLM<Y*ZmJ1^ER
zw=I;;=M(l}$?qo`NP{VU;?j|NBIw~Suz@CGbFd}exj=W1nDYTU*~{u`;lFUb+!P%j
zBOB2@%1a1^4CqlNfAV2&#@T6q_dr{tsrB(ij3H!VDpQirIr{c9{&}0G(6=prSjKX#
zUGru_^F!{TZjITPB{ySSUQ_UMV!vTJEzk8hjemfc2R~J2f(hCZt_#}7w6NUtr2ej=
zf_==_<7^@RiK%3Fw?I+Kmty<2O3r*-8O(^9?SyR)eolObCyn$WMqsl{HQVfWS^xY6
z6(IaqqV2o7=6Orr=^8Gmt+=I$XDGkq;k=$qMPV^^H%q=HeH<-jcIgD;er%Cpokz20
z5BSKf!xn_aKsEsguaBkfeQqi?Bj|Q2%UKj2UWZXE7A;Rv)eqfAAwjRCwH5%j{wbO{
zi=d<l?=Q|4Uf<GSr{ShyiwpGXG0C=hixmzZ{i92Z`5x7Mv~kZ&K0+nVt&ot^yDqm&
zEdM15^QF4oTG+Ic#ro`b+t$K^;T+^vdx^oN^W$T~`8pT4hO-e_Ab}_L#DCm4Lrb}`
zY9Lu7$eS){eA}C!b+LzAnmPVP*p!?Y$;eXS#%|5eGokRUXgU+Jb)h+aFR<mTyp1Nd
z5*qJ_R+>>z5_>kaD?zq+*8zw?Y{d_ec%1q_wFaNaannB0g+nc0b`~<GKSsOAnp`t%
zXmUc?@4B#zY@JU6t;QK-$5m{57c0MdvjX^vAvE>^L^$j3bnu={#8>!K;6JX->jzIU
zB)9)!+jFSQd=>{sT%Woiim4g<K*PrV()I4v>ndFV$ujKoq}iZ&e!g@VlZZKSXlhS%
zbhId7fd9Ka9sC5;-sS-JSoAu@j*5z!nw@=_zaflGNlDpB84Z?zdd>-%q;|x=KFCl+
z2F@*L%nW85#Fj6sHu8p)0c$oq7#Xf78iUdkY7WL2vD?cI_Par>4%hWyrTj;_u;?vC
z%kvxzZvBIu-)E=-QkyKGmSB)!f~H4t3N~^1<a}dmRi@cZa`-SMiok?O`FqqiiQLxC
zeoAc(Pvc(`MUU@5z?loil6L?6L&=TNVyfnk9jX%#A?ELkgAz`AjtILr<i_u-i=a+Q
z(@JotUA}tZ$5wyj$8keh83*9PlCT@VfjL7qfbQaoJP+phfOfBaqdy5$h^>o@X@J3g
zGGkS|yym?7`?A!iYip*`>eD9_AU>9pv-O5x^{y-t0w+>ehd|QN{BWNHq(6>FOCsPu
zrMz`n`9q{{-0=m^t-uQU+XN1Fc0F9^2SD3Yr4`X+v3}Q|KS5x2vEaUn4aD+RE^<^T
z0MWg^zFtsj&Hz?blX{+oox8go6i*+w9{f+n+Sll@L3jusFc?nXslmECl;2W(|1s7i
zAa>=NS!&4*YrlT2c;!R7J|*oPn-&qAJU2&xYDYxgpu6J<(RV(NG2FNQR#~2kkNIEz
zIBMa0Mke`Df>yUdaC=Lu{;36rflQZus!nx?n&;)>DO|Bc#`701{1H)c=8iAR%?2HT
zMx^O#gRHEq>;P@TpFkE!;79?_vtjjcb1-f5_UhE@_zw{|ue~n-6yus*J@@1@hJnAt
z!PX%yl%*^UP>yLJ{-*G@o3D#2)UF;W(&fL{n+}K@Pf=i*Fx3Y(H$4Sf<p<`_{tWjs
zX3Z7=u|0a90RDROZVkGN;;;X5JQ85eB|Iw&eWKYHe_i5GT6|;z2vj*0ikVHPK^!cg
z5qqF*2`**761df8?yE)Uaum5LhThqRpV17e8vdR<awcV&4DTvmo#ie_KI6npyZu4K
zcKi3U8?*`d=%CWs9#0D^PpODB2}GwV%Mt?77VA*<zc?#mUQ(4BNsB>AzGS8-?488b
z33!@^hk7$rHe{$c<ka-^Uw~jIR)z*hM@L4Kf$$v|@>m8bbPhs5!k)-!_LP92BV8<*
zo`FFme>^%Np{wE3sUl+>(0?gvYNEXZ^VZ>Q=XLnSwKZesx4Kt@ORWLDH$GrK1{RJ7
z*ffv}c@f9S1i=b7fUVcRt1E}we-1$9VHE3qe0pkhb+VrM`}Z&9+@8zFXr-yGeLBE@
z_=JeatPc{qZ(;X9Ic$D;`WU&6*vlMCI(vpcuXMbiR}Bb*e^}FtZ2D5mWutmrMPA4t
zH5fBf=@yAlZ>4M?&Ou<os1xjxYhZ@~006+XdSx|?3^z5#PMimtfaY!D`X+tJfy%E|
z2WhZcs@lZlBRbEKyEsY9Sc~lyC_`s7$e$Dn%Q%5|oGs-v_{IOftC-$eA0LnNj)}-D
z;JWlRPey49%Q0UqL2Rsn)u+QAEed{DEMKsj7L=Oar7I4VnaBc_^=$nF12eNRP`B{%
z^E-jh&TL0koRUJp*7WZM5KDo<beVDY(?_S1lS#XH^_+qZQ5cN$baRjmqDp!!daTmZ
z(}4<)oQH=8_zx%mGYoimc$EEK;BRqwodjMfLg*SWQBQG!&c*;pCJ2jeN&r?~?R|66
zd7%7Xc_7R|0xzhIUwQF_{IB1AQRS{=Rvc{y0>TTJ##E-oB=tanAd+W!%cyA2NTd3T
zIP0wqq!fx)XcEJE4M1dpA1&l0oq%+0DixioI&uOP;ifXzRTI#yCgf8T_QE+ws9@8g
zZgl2`h~cJ+zx>uXc%uVM6&@?GrMb?GM}#SC7Fl(=TC#*~g4LBEh0Rxvrsn&uLyIlG
z2HLiXJJQWUPtLQ)WDR;ascGbqFmF0p%mC=dv8=+E{BO@-b2j`!(y6kzxcD&GjHHF-
zfa6gCW3dBD9txn261<o)l1t`|YQDeT4TrOH0rU^}7JLN(2TA)&T3`XD2i>_eJ=}W&
z)=tN6akK4ElWLY&Fb<scLW35PI}^1G*JGj;r9dYe7Ki=gL?5dxc>Un^7BbF9`+`=)
zB{jWT`J6hYwMy>0vpU8>Rb}wib3h&1+QsNE%mOZYWW-A-X;r>yV$^@1t$Z{1NI>2v
zV)`PZ_Zwm!d30G#Xpm78fN)cv9jVFwYa$FiibQBkeYRxBg6x%j)qZI6GmYubG%_>k
zCCjnrpxShE-+PIzk^yC#Wk@*M(rpgTCt-p`8251pfMNi#w|fG`cQ0Bw;HGXmxO~2m
z<&^tl2O7}Cedn@x3hYzNfRg_eE$y#ybs@>le!1!G(QG4ITfODMegi9K=L)^mkgs16
z43=8_;;O)paN&R}1G~GvZMefsg+<`!&wv`C(u6xPF#*@n0d;Iu;^JcBEj<#~0^g2g
zs!;10et2GqwV^xS9FR|QG|LU=p+KQz!3ktxUbEovD!Mph;t@0<T%<$GGVC9)Rvwo0
zYRom+b5HFvn}E2C59_;I&IQmek8(z_bnBc}>(K8Xdzy5d3_~|E_!XAe`HTO83IfL%
zHNX%Xj-U#PKT>!LT=wf8zFNx#tk&|n^NqSX<K9gY{qI&eU}`#0<|}t1(Ly;_`Kjv2
z%|<0jmgM=P5mQ;hT2vGP7|$312s@OVFB(ik5+>6=C@3^rrHrX=dseFF11^Z#KPM-I
z%2_(){E!0XZOytGRUwhXe(4e5g@wJ45CI9_e^)Pr9Ro82JVHWoC8em=o1>B%I{1a?
z=%-Kd@oT-VI1<?On~0JgCryFZ*`w*KN;0FzMY2JEq?L82e$*&q9sYTj>l|Fq)m+oF
zQPkg%85ADQ-i^EFrqHDL9F&~@NMD%HJ@V4Zu7Yf`Yf8>vaVfV*vb3m2SSqwbbB398
zCP4r}IUZq5g1wz${}<yha{L1im+r)}DJ3P5{RV^95uftzns04GqF9RN@eToeQTDqT
z8I#U8on+74W`5XX$EnLUyV``kW^M#FpQYx*FCN{yTy_XRcBExzj|83$PZbmf>sKs+
z)&&kofmgyqMi{SaJ=CD)%$TIs*(AS(uWK6{2upijZu>^w7c{J_;UMIoVPaxV6zR%(
zcz9gBw|bj`lQn`Cqaoa~!`zZTx~&Pk#9CWhHMO*cYaEO~&2oHs*_<&SA$<QAWPwDW
z(*r8?=!M4%xqOdbzRv|td{{Ds$G~T3xiQ3g6|dS9EdyZ!8Ee>Sq#g=Omi3G-=cIa_
zj0lVsz$P(OuWNG&6eognv3_$Q*M_z?+hlMRM;nyMcNHb>X0A*7jJ&eBCOrh6dw}1t
zu``@!)?&k5BPh7&MevpT$Tl>#Uq|zgTaY#3tvyLoC@Krqy)}6g#gAQKiT?%q<)%@g
z5(qhLUWd<FFepH}yK+Y?fAn}BFL$hERj`i4$=w4?NT`{aLpAh11wG>bRujyifJ*be
zn5NhV>Qnfs8i$PGuA#b0F4Vsl`?E|o<xZROa5ELTUX&3(_IiP02>u`NX(M@KO<eRa
zxZoWusb-1`oay9xGI^q(AOuDZJ%zB+Wt1XG@#22b?|#L2gS319Zm2Rr7wt;{#xG3~
z0z8o?iWbU9+k)>%X#yi<JW?)hR%xTY3vo(QCl8*kEt_sMZl>~czDndg6fCahdGqE?
zj%rU*5>V7?dc5t21rWG@`}&ob%lru_CFvL$^EyjGL-~IOAojol;xi;A39OIIpVC1K
z+13Ix&+aGH63^?{Fe&*_%H`YhdP{CF$ou^JEsmO<*pJb?y!5m(_uwJ-W~Pb}$76c|
zt%SV$wc9hhVD>F6_7O>oV1saV81bICv=3hCY=j2itE)5dkySRa68tj<o<rU@j5Yd+
zG+w{<j&oaEEZjI~9U%@<ZR`Rmf7A$i<hiii5@E~>)3!l|BU3B^FQhlWKOU~MAj+oF
z+%EjS%g}lB*AnHTtd#j<h5vxw;*{hTSv)$<-<d5bZ_mpYcl7fj^k8vth5`0ydF0D6
zdT#krA{5R5hz2{lx+a-^*Vmu$c<bru#a-f3ysZ#$1)BuQx&ihyS|tYXa2XO|d8ZEf
zZxhPMt!?ssgijYd56AiL!*WD08t`f541go4?@8wVR4HO*-HupXqvlevC9p6<b7<rw
z`2bXt$}s9PSB>j=p9yT*YPA_s@&WbhlO~dkKKT?fTry>9mF&=mf3)tIVJAs-X^zus
zbK&@Eln|4x<n}L<YIj(?^Y(q|?5=(C^}y367=)`RqQl*FRbNti=k77e?@TsH<$^AQ
z`5D=4IgaxP#^vRX<Eq~(k+r-L9DLFj?}kn}Htvt~pEYR5x%o=T^SKf3SbU3ZIaqVZ
z3-9@xO50H3!84os65y+Lw4JS(&wLdyDR2li-LL|)QM%W!OK%z7GKQr<OjQs#MuGz)
z<KmKt`g(p4LQsAJII2LWHW~;mPj0{+#p`{|1)etr$~wXeY+PK_C6IE@1Ke=`KbJs4
zLtJETa4-Uu%Ik@*koUDca3(y7M;{&@KHHxqN|BcYF2xl5>J5%iQPbu~wgMHLY2Wu&
zAgb#nGV1;N*8uo?^`-Dv4@9hn{oVPn{iqfM4GugpuRXPPm9<Erdd9GMo({v|4p9r(
zcp%i&RjNE%_mb+i#_7nMe2Z->syY*s_vDr{M_#?7W+al|5S0iSF`I0EHp5HLX(V`m
zk?VqL@@ERKFR+3!!iDB}WgBlf@}y%jX2L!^JP|aQ*s)(MPwk^Yz+m>Y`$NA(miqF_
zzC?bBYsDdzf7nDHN#CPq*prpIY^8jys~y=h=t95Xn_%zJq5A1nj-m6}S&nPk#{j>p
z8{E^0qQ>vOe+Grcnj=}m63)~>{rw358F`Juyu39%b<^M2nxsHI2wGZ7pnr!dZ|s09
zwzlo^(WntLw~xkLkKg88qyP=EnkZmMi>e3pB*^2_)pk^%qRuDa<l>42#AGcmRzwu6
z$JD<!S6klWii=6aY7DaecxU~vmN&{FU{xqJ2}1X`mSzX>4zBlBSqH%C0me@OKFTlY
zHM3YB%L6^}RwXDy`5Cuo3sow5Ac@G4ruLVkBAvNLMhQI+`SY$am4JZGvP{{Cm^7tx
z2aKXId--#rc;&JF=4kESmals$6lf&V-&)y1mk%C<`g9M4I&20bey>K_D*xwL9Qfp4
zO?~<4-KVpiDL=%7nC<&3M_QYyj=HIGoq=9SJFaM59jGlTmdo_vlW&o1{P9Ln<J9y_
zne&D-B^k-EaV-RFs0#F(Q-Kr{kC^z2L`LiUe2QWF=Thfw<zK&k**oS5Op9hIzqVTt
z09FQ%of^-)Gd^h;3K0<zu$ub=;@@;LBn(91xh48iQfR<GXZfK1u(Q1#F`(5~KD*r*
z&>~ID3)OppL7gTr2m?{11sKhOCv2+cI{`+`=enmEG?QAp33{3-;1mW{`4txz0r1Xl
zZWac8H(FN&)n+$cTE)uC`CCnI6gw>^6UDrLCaU9dHr1t9OhFYDd1hFTj^<%Bn<s=H
zAz}HM>E55p%M?@5I05bU=TN@>4WcIzs>#ddgM=Wpe6G(E<E1_Im8>`fo6i_AE2d#b
z(>B`<d%Af-;VDkjd}`%GS}=4SMt^1s-kjE690ZCue^>6Vk$6G*Now!X@ZN0vzYVeB
zP46S}i}`&hW|zj=#`C&>i+koDnkXl}m|t@Uz^U88ZXRK<YGDW|0s7BRTvX=}_UTjE
zT_Yc}W@CIU6SP#HW8U=bYmw_Mg}2A)q9oPx3f*_BG+M}DX>yb@cE$%UEGgbPQv*)!
z-@kv~9*!34H=i&a=Vz$rlK@_%QCE1OlaABCAK?9$tCIa_0l|3KTzACGSmxn(3HtQ)
z*^7&d^O@=zsAj5X`f-1`7;u%j+$)glmBzH~-S%8IH<5#L3Ot1D$;49Y$t-K%BDYM>
z62?h{G326!(kf~z=G3dA`C9env9UPWY>T*#5N6fP6z!~7&xfmk<4obCnEelIdt6KL
zCTQB@oq07BKhc?yWd$F37gDz@7kiJ$l(<cN+{}cyg%?qx-OrdIp9?Yv{){0)_50Zl
zVdluAW5g|=LzWzyzBKL>Z-#(*S)cX3-|cHzYN_1&!xKqGZWrl`Yk21K%<g*s1!LQH
zFn)g155$A9{D%g2P2SdXZoX5p(wWAlD=SN1(h}$p+v_-fL&9w{sBc{AL^V5hkoDZ$
zAh+n47<hJKtaVQ3e5-LqU<lhp@nonXEfy?F&G9SpFYB{i!iv}H{$+48ABvW2vc1yO
z5xSh{`+7Db#L0L|oS7bOZ~0)CXhpA!Iww9?3HsOCQ=<&5iG2BXC4JC~@KbkDUTtB%
z$j*$FGF+omOL=KN0sWmp^MKj{#Z)mT<Qad9al^UHsy~&6e7d?Uv+a*nSTQuMWwESi
ztccO6R9nJ_U6t9I(!@>!7ryZ>B&tQ^!{}_m+ds8Ix3@jY@-_lFa|Xqq^$vzc^ep_M
zln|6^b>8}f%XSSOZ<-|$a~>+ef~$)s9pohN`H6})BO-VVhDc(y>YlU8;F_9YLu*^l
zh0^-UE%sm)eu#_KMjgGE_&$TaD7;qSrV#zI(LX;Of*Opo+?p<CNyXgZ(zAoS;B$9f
zYT8fw&Z2e+_++jA%Qw5lAvU<<VFg~Tk3PEJ<!F;rQnY~U@OT7xm6dB}^FQ)$n(nb_
zGo%?p<t;Oq9WO595l~{3Z=u_h7k`K_u1=_;=FSh=0~fVV4+{tI1Y;2ncZ^Rl8-_nU
ziTh?HiwkqH(KvUII5U|1g<T?q+*!$0{6WkQ-sb~i+*kSbwtVuDgAI9eT?&SoH0(6G
zmY!Ga@YU{Z0flj{SWsF{vYl@j!;$XO+hl}eWP=Q>9f5|E74#2)jRv9gQr%H)A1B03
zJjK^;A9^Meb!kbvq{`~$9mOQ6a&_$|B_S#QdpE^<$C(`LZ>qG1*ew=GS)%z<jnwkI
zYf?JG)~~O=`p6p17Zgi{HFbPR=#-sBd0ac-e2A>BZ)`JH<FGyp@_M1mEwGhSZW1(X
zL(1Ry$dULZH1s`GJZZ#E%uZY|gsKGopxoZaj*Cp6iREBnAM=~7G7MdHoJ~E$|33A|
zfjIQ!k1^NrX>&8}vZ78{*OtQ~i%j!G3Q14F`1MGV!SE+I+mtsLYTP!Ix1zI81#*}&
z3ED5E&xLVXM&AFOa2VC~J~H3cU4Igmf_+C70T(097hVXF>K(upy98dS5xBu8bFnuO
zdXb_~B*(679g_>pxR0I})e(x5!9bb0D0rN%Jh#7%BHOlvlh*S`S8?F=81hD7ODr6g
z2OuasB%j{o9d4@+7=OHD>Wy8oWn%Gi#+S2Ix%=f(SaDQ*!CueFJ6m$u+rOuGk|e&#
z_}>yif#yC1dP@N8VTKisC>&0krP1*7Yx5}*vXBC3upPuIctDvl0K8%KeiKj`z84pN
z^d-&j1fKj3-kdvOUY8xHGBNkBS(dk3PPDSP%amHn{p=Hz*X~?T5ZFocDQtHQ29FXi
z85bMy5krDp6h@G|>~Z8bhdDA?<`1F9tj^_jv+ynnE6m=f%Mna2V7csaORhOCSaGoK
zHTp5>oGBM{wRqK#8)ooqPxEG9!=_prB2R=G?&lF$cJ!@2G&kM|=Dtq&nZ(pkNT6bs
zh0uA`>f}d^Euk)W6H+#$iydBg6|*XQm)rNm=difzm-_s?fjm)(O1aB|<x?mkwQ$zw
zzY0h63z`2oW`n%EJm?z{6lY!nSA|o+D$E$&?&$(k>R_|fPrpTFKNB*kk9jIp{a9OH
zF9ptoz=O9FSiTlM(*S1_oZF2;8o!vB{S4^b%Tn(a)ml{ce>T3PLH=X?^|rKg+0f{E
zcr*REjQ+S;M$gUM1U;mUTP)dL{QtOm%djfDE?Rht0!oNViiC<vNr-eR2uce`BOyvi
zcL_Hkpdg@hcS=il2}nzebazNM-`x7V=R4Q=fdb0rUTd#4bIdU^KyBLmthx6slb55N
zLF2+uP;mXmaDFSEx(HKd`;zoVzo`BCeTE_*ClY?Q3Zw&bScv2-k*I0i2$QQ~(1lsq
z>SyhZm|8K>?SZ73w5SJ}L#+Vk=02P@O-VnJXf7u{%(e9<z%zUdQRD7ArIQskxJ%>X
zXX41g+^lxW4|QrdbEl2>2-Rtj7kYAkd9E+sx`6+D0+UR@(b{3zb9>tw2tdE!V2**-
z#0Cg9gdjB#ngNl7PCo_Iie0m_nIN8i&(`}4EH|CacC%2h-_v2G(ZB>wPKNLi5?7k>
z^(-5Euy4KV-B&-Sc`GOB3>YH6oH227-%`HD=p>=T9P|#T9h2fe|IU79_<Cv4W|o3l
zLXzXd-KyrFzupE@mB~_g`(+?Z?n?KnE)1m9&VaPw-4%+}GPB7S?%1~34_qGWQPB>p
zD*7Z49xn$g+&xt*ZNPr1bLpwliH;^7<L?^dsGd|^VMg<ZR)+54CXbS)kXo)9-9F{f
z%S&kIc1ORygR1qm%N*I_`RpPM?kggj5l^;Wq5U%)kx0r+mTyXPA8BK?f#kXpD8$h$
zlZyV=ffyUuK2n^I4zmScMnOVRfSkE#I;|FM+M=ngFFaCfzYDQ?Na#6%k#>T<VM=~u
z`p)~er!{ai5x468%Zi45cyS!D1EU_l*z}1byn83ii`j1cYEW9-aTLXFY}))Ia1Za*
zkL`T!Ql5Us?5MTs>+9M=F3wJvm#}qp9xu)wGjp~o`)ks(hvAfS1b;xrWqso#NkuU~
zz|p4n8k@@49^g=ZPmm>fOK<ssX@sTsfyMndY*!0b!;wvXyxj72^O*hLB3#Nom-U0p
zSwn4|`IO18Rqcd-yQ))X!OlRogs-J_M{ik9&DK!Y<^F%hwV+!YQ_~g7ka@;KY=URU
zR?xL7wO+i2<#A3+OA9LUz8)f!PF?*Xa&lLJ+adu#soP8aZCPYMY6Jo%>8(dnDPhQy
zr5#5fU*C?-f{aD{!{6AWH(O&BlDa=oIcS)^Zpw7bAomlK`f>MXhg_doiqLpPg<=(*
z%F7H_bFO`X7=;?j<nigNMQc4eDrBqJp5}8UG(ZZ`)aYR`qd%j~7MW*N*qmu3>26l>
z@n?nO^!17J;d@)<qG?L}r7M`7>%NmnyO}pn$r1aIxH9V;NkV=>U3Lu1;}Wl*Z(hqn
zPyu=JwSW*3Uncrb9zEm@Lte>*^UxsZXQmqbx!LorMY=_v<il3FUP3|n(j=C5fjGa5
z|3H-o<7kZaROPqs(fa0`3dVQ%Vg7Mp<FnH7D|_!XbHf<Fn_4RuiGMls#gIV8C4U$t
zsAjI706TH6<&l~;D_QIK9?q@EG%Ti#(ENgeB$(3#=x$EyQA~7nt2%dZaIlJX7NDjS
zUcC5T1GKdTy3ssVUV~alRnr3~DnoTE()m-WlDzTI^a#bG&ENowPIqCBDI-l}*rzS}
zRxdjNr+B3lcCn^|KF1S8W0ke?2BnwX`C0?kS3ei?`<2?8$%WF?Z3vPT_a4pO2)YFK
zr}(n1xuVWj!R8im`c-P<Mcp*{S7cqppJO6FCV3aU`O_ZkKrb?sH)V8-??7U<x$m_+
zPK!Tzo1WnLeY$1hnN+=&_04<9j9<pDPO)qvawxKGEWG~0#j716i1XMPDanAY&Zt;?
z?;?Yg-c@0CAwT(vY(JT_0*ME+kiwu*yN11~lXH5Lc-P_Vbp*CNKWxtC(DMszq6;vT
zx#&k;)P#fC^gI2<FJ86caKnQe=NScU$)Zd-KW3ckY<|&EZ+kDO1+oCOjYz_Ork<T;
z@QK?Xxiv3BxekQ^=s}RM{30PE`_Y%7kXhLUA_YLvtE;c?os^U$+H5FMTuh?r^ZQdc
zYr^1Sw*h;4KmND1_);0NPhOjeADhHH)NClLsif{*3y7t$U4C%wa#LVbpl41zSsF=>
zeeu>UgXhY;_{;uD`n~Ppl_d7XifYuK^*paDUzb4e0#I?~XBr$4rS^2O;n`2i;IEOs
z``ey(xg9Ddi5Da{;&ZGis5IBdD*WpDn?;76m2Bicq<`>&oFiv!EuCF@3+Wi`l0<KQ
zUH@QJ>!5o<$&H~fwqU`wU6N?CMcwr&g&m704Q%9oWO!J>>HJ)^Z(3AiecSP+Lm1Wp
zQbT9WO9=yLe!uH8*}rpS7w?@Z%W4#qlsru;v(+SkLffXM@&~<~AtHNdP^f<^DERp4
zQ(-Bs8qn!MMxC=MNWZ};G>fTu1D`WeZ%a!!ye|1+2&ugZJLg~@^&6Wn)>HgZ`e(JC
zCl83%^Ar`VnLR&Q(SJX6anH7LNnS^iCr8-c+$p{s+ObecWkWIICnPv_k=8O;IH|0L
zO{zAZ0>z8JsFW6$NlMUM;IK<@hw9_mvKX;;QtE;++jae*k3){1vcCxRoc$SI+*`P4
zypkIeH|UaJ65=<_W1g9I;C2n?mb$lD80Rm;s5Op-$$8Ht?}wQBTep;R?+feaAU%V^
znLlk6t4YANB@x*N%+uj+tJ+i4d@kOssJyg*?}&$W9&|tCC#veeK0(+4=XYU;7s{i>
zoSSL}oo}7aw4OhIt~mj`?G0n6zYq9^o0`NNn{m6_IilVUpb6N%Mmwn6iW_QbQbAsb
zIROObxy7`3pb21hIT!iW(!>?Yh`BVNR@zH?SSljPWbop-w(A1_uLw6&1@a0Klbd?O
z5ANM}A?HGJQ)O!8iq^)cZa@!0T-5v?L3ub&6b}Qbyl=iuUIR<ANr$fVy<L|g&yE75
z_BH!s*Ylq?6>*%acU!H3bG>^V@@PD!m==#uPi4_o(wSPe5ov68caJAVZ#@3H^A&6p
z`NeZxPevnizTfES@EPil1F6i030D8?vGc<ZCz!RiFy1UqJm)OFyj!?DoY8uV(%YKG
zu}yAvq=RWA(sjIzs)F@Df2l#sP9w%)gCnfL6)Ejj@~Q5*mG05+HB#XZ%N?gF>IXZk
z#1dc35ed{M(+p#+4)OB>Bf+`IzQ}8SdfICTQ>R1x)pNwrRc2Y$voTMt=V!W(CVek_
zz{@zhuSwWi{piOJ&7#X^x_?-VqiV1~;z5d!Y*C?%I<vq}et&uOGU2g~3<a3o>C8)C
z_=^wUP>e)XXM6eP^2K=B(l{<ZP>I6ZEBcCjMi_ZjtgaXNd2ReDQ4wl!TsjDNEaA2{
zEXg`@qfn11J50;{8qFlZZmw;U&5s|k#ve6AC)XESrw7J9WxCqxJ<<?lswk|hQ9n9d
zXF&%E7L*PGYXzw?31ee;E?F50SGdykdf&eR#6y;PNe*xv;gVuZ%*?!SiE1As7G}pr
z=J#SJOYCZtalFqK@urU%NbSx#2dU<)-5zVsq`uC#TuQuQT9;aJw#7=W?q^wc<RiI5
zupxarx?NOPzahPZc7U7R_Q)co)F{`>;bm|{MG;eRzZ&kk=OZlwHkQ=<r6Jkih?+sy
z;n*A5MBi)5)7!ao2&yiPeRO+86XWGMjkw^qQ<+hY#dgw^GYVLb`zrhF?c^WfWJx!5
zcE`H)u4UPVR~4Kuk)cfaj?j{9Ztw8P$X`8Pr$CA5k9f3_;}-vwtApUkQ%_+cTU<5w
z`XUd_djlCMESj3iof!FjryNS_+~)flH;APO+_h=JM9waJ9mNlwFZv;7)Ql|rXpu<R
zZ1c^#onh)JK-enP_2y%|d;YqBm*2}B(greXM;C`JzS*e7$5*E0_wKCrZOF4;;5J|D
zdKmrNK5y-aGv3!*(l_3bu+lQMlGZqT6Mw^zJ-Qw9=Ctu9he9FM&cKrqHHUhCyl_&2
zyEezQy82k{{^zx=XI@9El@4$t$8+9Z*ZA`o?~v3>>MvERj1FOhJCJW~Z^y7l$E%Ah
z7ssJPKS-jz%teUoJ+CD4@a`<%g5Wwhaq?o^A$Fc}L+>i;*FkqzxM`kh<;(Sz1}0fQ
zd7nzHJGGK&x0q`VDPL{s%+bG@`-+1l;7od}WH0LvS@YnrxDv=>k|VmGuNNp%vTwZL
zLlq-=#qS=+rM_Y(8!9ue+U)pD(pvxep=?9q*5`~&Jz7FzNpJe*$w#4Vd}UQMkn(UB
z%sxw+)g+{4AwB)GSRj!Z2>pv~@n3O@&fP*OWOGf74=sHCKG`n{mOt4}-{EHPi?mo5
zl3Cu^m5^OYld0{Zmq79>88ER_A}@BuybpdQytdGo7qJ(KB>r=gAM`SYb|#1+GiuDx
zlF+vvn^!)g`u0XxBR&(ChxTEa^)W?u1-2?v&2rsm(x!>g>bpkz&YpkFIi&;=?z@t#
zVKJ=)V`w=(py+UWFN|U@CU0tY4*n@uW#oq}4T`jS@l+vi+qTU;v}xI<<e&M8*QW*5
z9OU1({$ozzU^w)%gIEuf@r;?5K<I8mIZ<f~CNGbNceZ?VrN#y9Z?`H6_lLXA>s_du
zG`6P=?X96mZ2RyVH_5LqEl4OJ*TGY<%elT$#G-X&hQ#(~8>QNBk8JXsU^?Gr(RsH^
z6eH24PP;VL2sw%7{4^9U92TM$k2eSGC{)XE1^Kb#Xz7{_lQ2nd{OD;IddD4jo|opW
zU!+@Cz}!HZn7GmZ3JiBii0!rB{SnQ)ua?@Y_;9&dLaoHK18Annx5<PrcBf0->s%8|
zA&PM<S0~Agqm6E#U0GA$(r_c4NbPoBWM_{?iddXQ+Xr;X0FdcfFEPzt%xO(d)j4|R
zOvd-rB-_Fdi)MaL%p(b_{%y5&jnWDKP3E|Q3Rw<Ax4jkWw9{7X;KnJ_-RFm7W|GRk
zIgYE|c&Mcixl?bPW%@lcv&)madiM0i^j$4%=IS-6!!=PEGv4~xyET@Jzj*dJiZ;2a
z5xe?z%139-xAW{Q()TQkS4_&5N4*z#bZ+tqsmUJa(BbhtWbiAn>gk!yp~8E2C<8@_
zC>xpoj>lYABE~v^gvcwoL7OA-+f?OkpDmwwKaZZbb=S^GxH0f8`Vg~K2oCWWT|b*7
zLQUC;UM-ov5b<#lWncUU?O1>X#V$j+{dk)Kk5}LVFTFqjQ+M$S&sU8L#!|DnUu^x6
z(FH?Xh2gqj6n7AIn9ohgw8Ah09qa_=)~vov9)sHE*!tvm)}D8N@&4lY&C@55k|_Cv
z0FL`j7N)58A{MXjrRdUiPfsoTYo&1(uJmAK{T!_6mtGCuXz2}}A^n1w)A>iYMG(!-
zOfM|DJ1vtyRQkS2iMbfvWC4TTCpE?adA&_KYek{*$~dn69G#zBxT46}(2Zw5vlA|i
zl1o@vJU~G;SFe#C;g8AQqabGgWA$R1Cqq%9EuftFj`D7KTW3u%PNlNNu5D89#>VEL
zyI(fGZSg=?f2TN-dA}MhA&vfJdU*>k_0`lOi#FkFIfBb;GOc3%#5sv{ev=cE#tvD_
z(JcnI7iSy-W$%6t2|mX_yyg<S?|ETjkRx96M3$1jHI}Ao_G3tsx$)HeLax=f`sJKL
zmfWO@%C-M~LH#v2uE6+`72*x0MaMk5o<zjPDw(eA3J!kj-Ss(MWt3}0O<2gciql{s
zq5>!tQD~g<IT-RN`lFw{>-(*d;~&qBDc{ERFVD>S_*F3-e!u<mdWcKQT)9{*O~3$C
zu%DG@YohjCYp;mfB^gcTUj`=GLDBz~Cp;2=-*|)$-G51c0UL+oNJ#Mm`F4LGR_c+V
zv^h<=`Ax3?ud4Q4OrsXbE=9*C!3lycOa?82eUT5i4vRzJ=+Xa<zSBibibj2f>*JU{
zc;<ymi54;8t^yM4!}0g3Kh+#_Y^HghMDFRk)%DHH)BLye4mY&%p5i5Td}{VaS2R%m
z0()~l1$PkQ{q%;}vGDn<(BYk_nZuL&QTwrMMte%K``>O<QsV>^y2(|y4$_!tKc;&}
zb-BboFe<Wfkf3wAugRmRiJ&Yg?AxEwyIXNd&JpGbSK?$&3omYSJp5p@?~OS%&`KyX
z$DlRG8aZ)v2PG`F`|XwOze9D=LN`!!0DKB>KVoRW4bf^DMF`3$NJpye5R;C*_USyT
zbhWHvxnHZ82hsxnIK6jn-LKvCF(z}=eA2RuHr_r}3HxNxjXBW1Pr-$-99Li}RY(q2
znCcBC`n`^p@{-QDGO18=XKOoe9|PqPM}7Wls<S_23?8vnQ8oqD@|t|O<?h9?am2ZZ
zjF%NAJZ;aP;wbVvvgjj!@mA}Cul4tJLD7+2!q?HVw}b?+`JOsysFf4`toR!=Xb$*D
zeOL}|M|+0lV2SKkGdIqjm?X*hxH>Ia486<0dP;~(iUT}(5zr+tEPayu*z1}==Gpo~
zkUEvRlwMV;E{6HZ^pb3$;wz*w?}A6pp=e(pIfq^4fh;`a@j5?6YmQT(Czd@UR-n%U
zH}@wQk&1rjajZTQDIp9;yY~f{tb9riO#S^ijCps?K{=I13yE*R9r(Mui3e=tpC~k{
zW6tnn8}}3hr+)fSMvpsr6_(y%rW9(9kbb(qI=X(Ky&;)(I;fAlzMINr(<&UE-GW=`
zSXEsg1~;SJ2m&I$@rZPd-BUfh`;u8z6Sg_1Q)0}cZ&1PQsG{c2Ai8OG6T4<qW;X?I
zVKs-yc-bhLdfS|4>lL|o&A8(9&svw5kM;*HKW|Kq>j#IDm#f7lmA<pzy+hGaS0TdK
z)QPM_(tCYne~UCB?4l8~d9;M7aqhUD9P2NP-{&&d4?oamvly%(=w`@znKT2o6_#eo
z@2TqWZtKU9@jG3fgGC<m0_^sS?p@_|0aX)GYph%snvUF}m;m3<JSAs#rC?P>A#E%R
zfQ>*;-8|bCX0|{<E)rB?U~X;wB&|0DF=LAuPJWKS>q<@T;ngY`HUC!CO4JIySd(18
zD)<wW_X;QCy`(6kTRenvYCG{*A4kR3jqG(RaF_CN?a;HAZzirJbGBxU9Lk5RIdMld
zjISr|d3YF6j6s3(7tc_^Wd9qo>SwO_^sQEXZPE_@E0S?Z3T1g)5(maBwWlvj_dICx
z?GRg5niemxz4bF81u8r?S3f@OswZsLiBIckE)9BU|BK<_=O7;)<lb>0ma;@%|L+HS
zV@cK@7rOh>ZvNes$c)5jjOt3{@}hoFLJSSWnEo=v*k!+50|czCJh9aqxIP9Sn@12U
zkXUvdH(#nMu<N}u-o9X6j>A!NgTotX`Z<HC3CBJZm$LuHcxUzyuNg<Tx_>>_$5v%x
zbd)fg!MCvW^9p-0*I2GwyRRWM5+C2dz6dCK?<H$%FC{hkPK`U@=f3T$;McETchxdL
zsn4H08}Sd?*b$x0!~rXa7i~)`KwXNz!!d`M^zDDp0p5z=L&J!lcUYJ9cu)3mzCDd7
zADi}5v7+gUPYZRQ7j^&Ou<JGz{xhC7P|Q*}slS#|I^2M-xqk6-WTc(t&)Vs8S7t6)
zO{wrLOY)z0{)1~W%q@<qgDHL5Sq&3aJ&}(u-pYy>{Y<@w@cY39NGze4tv@BRR4AZp
zVXPk-#_vLGtF0+0#GuIVi^njEcl2Hmc~G+P$jmF%o5Vl+F#~;``AUB+s>PicF5@70
zw@U#afsbR3!SB!KLsch%@qQh=NbVyo-ph+(8@1pXR+!pyuNtfvFPy1d5TM1oV|pb1
zs+f9SVyK?DQ%%EJ;Nn8keU0whKXNmA(W1WZkt-b<eNE?qnDbILWiV!-5#ySs$urz%
zOpnh*F={IFqgxpDUp_SH$x((eIHM&ZQ!;7I3BP~<vn3*MB&vTdn|j)rGeSx8sY05i
zbgN{WJZI&~eF7!&Bz`^016xx+v-yEg>N9ek^_52pF;%njoPSMRx3$#0ViuK5V?=25
za`cki2$NMDwXr1ap9R%dcG2+-COS&e`S%VE4nT4OBx}noXK^?=6~k7Fjfd(f?MCXp
ze<VBra6aQGvN<D8!ld>cbjDiYl3l{Vc?Ad%e&AEKcT^IfPmv5`21rg&XsDQy63NzF
zhc3wQ0iqEmJ8D?17oOuGSmyx|KLuaD26Q$NC|CnPd9`G`6KH`KdQv1)N`Vv_t%0dg
zYJN_rrj<|-8vpw<L9)LJ4HXW;(dc;i;u?w4K|WW66b${lL1hpSD^yUn0L<(xl(F=o
z0hZpsQ*+oI9d6@LNv<A53a(K2L(_|~P+>$%q|1)*bBcoAV)>BLcv=a2li#nUzlK!!
zgHg+D`h^^gXNm*w7bf*IuN2|jD|8>Scw)l+D;-l4u&81IRDa?01^NU4QqQs+fmjga
z4EX<iCnPq7=aGsNj5*kvS1uJ`VDJH1K{P8BdI%K235xWonz>KYkY2Qme5!Pm(xf0@
z66&<IffpM%&nxPk0n9l8`wZ}yuuq>pNnp+_9gagWI2NQp1dlfwa7yC%?EFQ2Nzf=j
zXd(kQGE+@%a`Jl+b}2R+*VKtaYt$R|-a{iHfKd|cUY2#`QuvL6;^s_xjVQkpupb`~
ztf{H#9OGmn<w}~WCyVO*s~1M{3D$%z>`}@W#|fTmV|gql-smhJLe@E^cY2Qgug!U%
z=TjO(8{&6L6^gkueVM#p>koN1$XDan)_dY+RtnM5dgI^VoS2#tULPaQjXm0EAczn`
z1VMWA7>Id@NJw6>lects&cTl$De%OPJ+Qung(<K8Nt2Fhq6y%yhm-Vx+z;Bg&|ay7
zSqL=v9DW)t@CazlPhvDs=S6>Uc2sjFq8MOH0^pf<`;*?khKA7C5Gc~Yk3*sO3QYF{
zpr4^5ItC^MK(`J(WwA0^0l*}0&D?q5k4Avd2|X`CWk2D@JFo2*-Y=jj{Wz5Zvij15
zeH;8oDM~uLf!XbS{nvF=Z)b^nT}xv7(TG(ksm`4o@<^(BvuNttryY-vr}1)iUyQ+t
zyCj^GwiWz$fG+-!kW%LSt)BOp&9wl3|F^>DVBeg)Qn|q<G8p;*h<*qoqPz(+8WS@w
z{%8OmoS@dKOF%gR6$!w6P<C+zVOMMds_%7m3CDsUr;Fy7gmWz`utPNNa%7S+RPzSi
zeM4h?GE|%<YR{o5Ff<Z~J%CE|Wx2Hu;KywQn^#th0g!$dwYxqpTVguex?WIJghoss
z@70`TBYCnC5;%Z#RCQp^RzK@s%YjZYaI0UTrlv*%LNzof0Qp%or}N*==xu=hd}#u=
zBX@7=Q`Xl@+D7E<+?eBG%$1yD1SH;dUo~56l+6OR&eKSIv1>8}lnamJ@7=Wz1KDd~
z=l&~|WaZv0u-5w;MLfqAiZuPGkv}mW^wQSVZGhtQxqaw3GL8z_t!bdBT=pk%N<fZo
zoy<=REr6C#>Qv;kR)*~g4DpIi$6RCfNv*HmsGf}g=y7oJ{@da=Z6qi$KXS!{QuQF1
z@4#Tyseg~-`HL+!HWn>~g)$x~&F_Atv}ezOS)jF(CKFSSy#AViBt*dYOE#4P*yB>Q
z@s<X0ZZcflPWl2{KAwh5%Gxv^*?GlBPEW(Hp>);jqaK(ukWOu$Ti9FJXp+|an<ZV8
z`#4Z!-6I-1wy9kvL9K{A*cA8U4``OKf$H5{XN<aLFy=D}2?jIDjlH4dHxG^VJ0crv
z&Q8`DkZE$i)_xy&l&`zL08x{^xoa`q^Z-N-VqFdm2eQ?hrvrpKJGjO_)zlzWq?~;3
z_E{f!T4hSB>Nr3XuHPSxIR(*DK^T{MTln|%)~#Tef!Pl63uNlQBB~4ae(G(Ex99ZC
zqVQ>m#ziYnJ3bF*%2U=qMc#WN$Ye^dIUA|I85b{JdB!9m$-k4^_oE}%+Aa6J@WQOA
zOZ+REWFp=9<i;^oqnMec1hDBa`H>HLp7II`AgFO}0t95R(VAl+yW%o3Si!eVRd1E>
zaxh`&>FX=cuEOp|gQ*JBbo>a{4{efAfcFRf5q_3RWJA~?wn$u3@_c%ptC|TL8ym_1
zB=3uKTS@FCWo6C&E$OiuJNmtuyhFv8Mr0>oGTHBM!?+fjuiRinS=SI+S!f(6qiIn3
z<S!5hute9m;|sY~#-!RXC_AxYzQmuq1yHB)%ohv*K0H?o15l_pY_C<FyuRuo)sFUb
zxNX@QW!HjQUQ7>q;`Y}d85yIlfbW8Jrht$EOdsZfBo40~(1Q*Y81{c9*6)hV?mYki
z_Bn*uaXB&Tr<zi5wuF+BlIOnP)h#u{E))roJ#<b!1_YEGLO^}7`{(I?X`y$Ih)Mf$
z_pB){0fTeosftDkl`Qg~tMNi#r_Qk)b9U@kTOGKsFkGGw%r96v5iH+TmIY8M3ZPck
zTwxH&>R|JYY1PgEgMKvN6A1T>t`>VG_rJYFK%fs=lN3Cb?*UAB#iOp4Z@p3>?Zj#O
zIn`j1(cr*AIM-;h28ZfEO$h^p66fQg9VtyxfY3*bCf;IY{Tb<iKFNN<$|_RF)N&K)
zZdb?VuzfbX?p}h}<YV<SZ@+;7M{&6P=y+^NG+WMUSs`hEHhW!r(jlA_gL5ZZr;LeG
zX09jESrW=<qc<G}Jmcc11)O$gfN~9JIyXW8_#7)RuP`(RfOxViga$C!hU8u)BI;Ag
zV6)%QMgfR#iQQ%2{0)8(pdZZ-w++j|nlB52vBC`A0;;2|zOh?;d{Jnb($*;xo9-(Q
zq5=yI;E60m%B)u?Aj`70%RHlgr(B>QQp$LEn^K&%OPxvR-$TE5aFil@*7mu$HSq^-
zr+YrtGJi}v;<?x5F44G}dvd-g)lgFJjR`sUn`ZE1ZZ$qh$o$$`C%ozB7C>nRu-13D
zxJr&s6BBRC#0hr5_d#t~w#v$3Z`4i|IDkKX{J5~wdfQek=?f@R#M)1NB)7_B7%Ulc
z#B~LZnJu`<^3R{6=uX^NG)ilS&1zmIcp@IKNd$uXH&Op_(txRxx<~l`;Ad^K6sYj8
z7SKlscmT@ceJR!Kov_`Rw{JW#2_<xA?&+&iN&d4#I#;2zf(~}OnyN%jQ`vv?)tzC|
z8=1XmHgE^f^#aC%A6}FL%PRJJcFmrCx+o4b#4-+p6$F!kLG>%;*_1C!U}$KQ(|T1i
z^l7wKOFN>t2KmBZs7w=(tZ&x|E?>^EwY5+@Jy}PBK7X#&k8^IS+8a}!4gXee5ph>C
z^vaDJZPm_4uwm*ce_+^YR={`p>#bLu#1VCn#!Y$2%RkRD=2n?h);9b!b$lCX)BVpY
z+@Bk04H9S6>xKiw{TY*h+O3f^_<A`n_}dG~Bn@6f_Dg8sOh`8xuN!}sL=(w=#s+#&
z#6EEV+0VA9lt_664iEI4S#_K6Ov@MVL63@=nHhTS=)t#d07(F$Q~7lb!INzT@U>kK
zVP3>T@ScE(FucnkBCJwwO)n=W2db$3Jh5uodY~kXRv7{GX*ckU!m9yw_bgic7>F7|
z)b1G+si6@<OZ}PGAH2E%uo?J<X~ilhT0;Z|HO*~pa|Rtap0ua-7(BGMfnhrKwaC70
zL-YmGp{He0enX?6?||g>kjXG<EEh%fVDA%=5;T4Z3jA1|cY)T7ORdmM+<Ulw17|y1
z-~;;BLBKB&yZX-Fy1#8@e?$NBy8`@%rMA4oZ<c8V_(azqohct-o#j0et9(ao;)um1
z*Q9tf=}o;<B@4a*uwM7I;<K3nN^DfAmD$hswlCdrFOdmRPtM{+Us8Z3r(5xZq8#vy
z0faNF-;$M9&m=%u!5|MKx8?N|;Fa}ZLm<LV5MB@Xy^C%nll$nq27J4!JJ(vS?5p(3
zzGy3>cO!8H<8*(9beCn=r#gKJV=MJ6E7@l6rk4!$(?@YeHG@I_mNIx7P_JT@$UACh
zT7vmL7VM9sF7Ax;3O^*<>N<z6l4{vu99k!C<(Xp~*<giWE?Zsf{!X4;8x<AR25;4`
z8Uzt}PYTb_J~!M#Js3(g^F&`*;=%s!DAEPX2jAzX$I3k|50sez^r$~uC_0li@0v}H
z<@nf)%OC?Yo1D$s{2r<8%UgetXZ%`cR?Jnjmv2pt?{OVq5d7U}VU1i3b;qA}J?4ip
zUx#mN3u^>qo%(mbW`k@*PIF`xsD;5hh-RNIKOaZV1Rm<OKT7IGZ9qy?0Kh~5wd<~0
zqr+>T7Yy}H+X}^DT5Lj{Tpv;I{npf%(@ye-W%IV{$|Nl8kY~92ycC*?sH=id?&swZ
zK?ndVB)E?C(7m?N>wH`BpOauuz;bPE5&e6xi``kORkal`e{y=fC5ytsBdZ1562Wcp
z%HaX&1c>U=048^T-yS?HXosf4Q+~Ssu<7?A`uK!wBlj_~Gii=PagGX>s3*Flej$Ng
z9u<TqZ&lyl5;$cO5<H{FiaD2hp|)MyKu7uH@>^oOJ+A;$_W`w6CDi&`D~Eey{}qOz
zhL`8)W(Ok|v%`?rWj_CPQUQ8yg8LKaeSCb-G}^SZ&Hw#cgHejI=?;V;bN-Z;#J_)f
zuj`9(I3@NyvYi3M<5xvr9(Q`n_e14Sxj?N8(724I?kg$^ixa5qy&i4yYp<V#nUYU=
zkK#Y)E<C~tfW6ToBjp2I0|SE)-5kwl*~TH|V3n(nHzkeI&?V0o8Gx81+uwc(tz^}-
z5L<jvjzVFmEPi`pzlcXfUb=eF{r=O6;`9S#%2xD?IV%hc)HY-<$Gv7`;vXP!HCfNj
zIXnGt!*4|j0Pe~5&s5{4iN5~)rb)lz$tCG~3I}^IVKQ5Cw%ses54vXqGUBku={+W0
zqke$BzA=7GGCfGDW5PBvb!jpODUy4}=$6E@@f7RzvfQvi>DDOrrJB7^L~WY-%rN)c
zA6evQco~D)d#?MkGb3^OKg2t~7o~VRSp4<3@CUo363YfJ504yZe=gbz2?;T4l%gJi
z0+z#{y`}nfI0M-+{&2PAlBfu=u9;gsb%{G8Omh*qu&fZATceP1u`G#HM3c-{pD#{r
zI`Y%KDcfk&4dQL2D4s4;t_~%;PajcANxNuqmOa<}zqh!1MWw3t1_gzbm>3oc$hN`2
zq3Y}FTX8Fub2R8#Hb_WFLf12u)T?ySPbtTkTN^rZG)V-&AoOK!lOu&yT&V5nBCPpY
zY+cHP2WSu*f7S4ir*bmqF7UeR?wl+rE++?V@@=&qFyM6kZ@*0qV5yL;#6bdo?(>%~
zZV-#1O2%FIgwBp1qR{$JpdpBYN#xRZ<1lif@^7f2GrDSYLGYBMJfOxi%;w>_y9mrV
zhni9I+vmMIf(aj{@mRJuI50vu4N##s?con*ON4-VV?YMxqLz}A<HP$FB%{x^@IHD^
zw@5u}c+K8lophx?&q_W=JK?9cLmq~Rq0~d)4#Oc0UR!@if5mxSVvj#*v|0Ucw=AUd
z4)M_Fz+f}vU^eU%LM1?sf;39gc}H;!(ls0$oK`ayLQtgx&uWIY4~c5`l0GWjAh6tS
z9*rJ$!b+-sb*pn|IK*G1VUf#$ThomxicZp^%JPrx`4F5ypSla-WliL$u#mCcoNAl(
zS@mFr+UndA%fmsD<R*=;)(FHl@5j1*#|XW(=M6NGNtJ&=!L28~1&(fmqoY-=q64V0
zcU#f`PHWrh5|VXsaXDB86w6!#fzXqr{Jf_L(D9LLjd(K9c^o*SaH$o)lk;ace&tI1
zuZu0BGmUfR_Kh<B#~avHrw=|%7BuPVnt8UZHsU|l)fa)TDqvH*f4y5+c<qvRqKZN;
zY%i*$O+Hmxdu27l0PJ1CH`EM)%xsgltGl5L1Kq`W$KY0_WjeR3eEJxQQypEh|9jw|
zd`CCZuj|U`@)`Ck`OP3&%LdA%kkL;pFB{n+BDdbNKHVPB&2pJkfZmrx_i6&?&jx(s
zQ_e;<b?h4o53;4EY7wrQAOPPG3OZoXvV^6IYiQW<5)dfkR6?Uc?6bq;PLkR`6wn0k
z;jMJ`6OsPgucw5Xc1}B$|AyC}?D*`4aV(5^9PP%tn<e@i)3|Yg@8)icoS9{P`>!W1
zz0xt!9LbA$E9VA28nUI9o%%+yY(({9_x`nj??}N?M8Jauyp88Z`})gEEHQQkp2&W~
zMKgOEFUjUkfHnwIsc};Jy$m|uCmM{1CB(`<3eut<3B<BH6egWN#AJ=s{^P(}$0Y0j
zBqgKF=Cl|{jC?9gi#OD;OJRo5jRBCgl_Tf8oEj?fS59Y0DMn(T?D*uw@}H|f63-o5
zr)NA&uX!L4#e3c&>D%FLBecv~ZuKDpe7yBo15>%97^wBV)i9-Tl>Xd?Hh$x1ro&KO
z<&}%JQ&W@n#nmO{bymb39#y{QJ+#P@+QXe6V#)h1R~KB`XFumZH$nU>F!qe>#q<W)
zj-itlmWgyZ&D?%syR)1UKBRpS!}e|QlhYrK4S#I90ABvrmOTlA!?-7Oc0jORxpf&n
zWd1$J&is57IDap70i7(Z6We0thNzfYO@PV_Llwm={Elcr@3PzD_kOF`$`TVpL03Z5
zk7Pu}!f#+_vH-@b_PuOs|F5&5*R0<hItg02SPH!nl=s_arHr75{KO~#_TE-@Yk5&~
z@;@)aemxUi2AooR-FM?xi8Z6aiw%R}@{W-cJ~O`3W`d-#S~oM75=Z~U2>FasXbE4e
z(MM@?*Wt=-n1XD=xrT9^jK#YE(}wBA;^gV6>89_wXP(O19VUK2D|4HPQ``_a_uyw+
zCBNC8Tk&NLexl|S!uby0e|rE|zxySj^c@>sN>p+2H{Pl%i8?@HltfqYNL4C3WZ?Kw
z$x&@d*oQI8;K_&UA($CYVhoH;N{?y?AWgh7JePY5d5X321BH#h|BL?N_uPeG46}^S
zU#dRX&$?*58!_^uF4zf2zFAZRI8bZ4Zca8|G~2)L26F>{2yodSe*PDMK?SsL26GIi
z%{NzJ?v7pO9*kS8^mG@rigo?=eZ%qAkl<7es-G#WeAL)`j?9sFO*I&ZN7pMte)1ie
zuQZQ7G2@Y_PvG7r#cA`lFZ|oiF85pY)*5u>^{$4|@a>(^hK_F(PGt@Hmyi<#4FyJ@
z0E>=-^x7Mwq~42W&WCtdz#U?3&HmP=|K&|6>=_ywJ`ocO9p{8Oc*&lt!XMtwrs~n(
zsJ0Eo-pcou6osn0q?>YIR7{cp#B2$B-*0KN$%QMp=0M2ILdsq|ccRuQ-Jztn=BSSB
zAMWNKF~<IHwaf0Jm}8$mf8N~1Q3P!PtEwp|kah(AwZm917rkXUl?((-lfIHET(rt~
z0Z`+~iMiKuWFj9m4fC}P+dlW4QDeo7edGS0n<DSv$p4`RBDV*S*j>DQ*>vdzfHQ*&
z>=sj&uixb$K(c&$$e0C5G%vHf58*y4bsM|DLqeiATg3GD9=yLYNa?G$K_%c=DDUf)
zm&XdLM>YFXiX?TvD9e>dsqgN-3M?-g8XEV-MK0cYJKNi$HN+i(Yj+SyO(#n*k881Q
z-qh08&)afI=T@BmX>#Mex)utu2}76oARd%G_7S2^|L}ju4ZmZG1RqK3HUQtPqoWgw
z4xKYEkza9d8o+9(miocvs?bj@*#CB(m^E^Nl^ZYQcYWKWXI`McN5hBT<|vo0ysX+I
zC$DQ}23(A9{@1ThOi-;7Z9t1VN=V$Mr*HY4;CZFq*j7eOczp#Q6ARg|Ymbv{6jO=4
z_-OA>iP{6cTlaenRk;1BI`bZ$pW5uewY-mgS$}$MUG(VO%dDvQ|GiOIy4`qVU!laL
z%9ngaSf$aV%kSRX6BoR3+Q)WW3rYbbxc4Ry{ae}Z;pq&9B^|wdjl5%mLfvG?TXM~E
zUC)~8uwO27Ng@K0*jCf$(6OF@arWuF+rQVBu7Ayb&B~J6+t%z3{VlJoPeg_Oo%RR&
zguk6{3SfQuKGg4?Iq|*}i@lYv6f}~1dV2xzj4Cu6uYPbDGFal$Bc}(OMc685OKkQJ
zu{2lXi5!h|S?i{7^X^hY8BMI}Vg6(gabm0^a$>~_f&P^mMx0Y;Ea+KL6&J0CtKsE}
zioAVri^l0HXSrDe?Z4jUlk*QFq>jsMh}%f3gyn^$LUcK^Pl;}*^W9#HsCN#icF**-
zqtT~-Z=+HrDsnxj<vY}{fC-2Zru+*{hC}1yZ@Rm?GkiK4KRa!Pf_FHF5s4#G<(Y}#
z-g?3WWse`Ov`-m)Q>v`71j|r79LFmkTRjocy5Fy%!xx^nZwN-?28+(ErDssr={k}H
zNCiq}w!V)LhQB_gezp4<rR|I(#dk>`_u|6nmu@Snb0OQI|4M~}C5iP=;-u&rQQi9z
ze8t1=C~8r_g_%Zc1gR^v>oniX#4cNHS--S)#+nYv<J9@sGM3=pr)LIBeH|oMuet%{
z3D7Q>k~EvPZk17S{J<N7KptK82copMl<Lo)*8;j5Rk}Ckv<UJHBFT5&Uzvqm(p#C2
znzQ<V>5!VGp^N_KD6z0cZ7+$B^k;d1MoqpuhpLW2)UC!7Gwch$$XYiRNiNU*pc$|J
z0a=FSt5Ly})%y8M&_KSx?dpsFdU;ag+)CGrVmI)?7Dq;e$s8Vk&r0bUG4b5iQjo_-
zJ{#0L$?AfT#e>sZGz`+OH*gljwbzdpY`TmupK|+_3TSOD1>dUN9+TasFn>z;#!N6q
z22pmIeP_jOs}S3@lm2+DMH<l4m8E=8=t2`NVG&S$>Ndz?w}r`~*<ol+gkj`fNtAm(
z?u7gJe8sn|q@8G$im*Uu)jXEW9&TtcV)0w>>F6FM4Y2<bB@kBe%l4qu9e4T!(f+q-
zk2szh7IGKyY<8IxIYRb4#u5RpNJL3zm0U>fp9#on2TXUjdJuXg>$_YJS0duN%lr7n
zo9nXgmdKp?kem}3IvOpg+}jiiKiUyI9=0(}&hpHQhJhfc9=uxpU4+7;;Qq{{UL3EK
z2?%gS5_<^5RjR9icso@#&U7^xe3}Ha;Myn0<;SMT8w}FR$oJ3JqMpfEk`$|{j~rJT
z45rbbBU)bRTkY2XNI}4KB((X^lILvx`ZpH+=iwm%08BQ`R=)yMiI+`fq+?S5-B%?m
z?ZK{<vC3<-UB!4#dM~pfM6tPkh=-{vPI2~zGfqsfBg(g!T|4zJ@dB6X?mD$lZ=v|!
zD<rgPkKuVReY|_rtu@Tc#|-X=bPP*L+!pca7rSId%@e*+e83IHzn!#FWtWiRbk`(s
zoH~Wr7}q2yaBTL3_cDQq<@n%^cw}nF5Mx~1774S4O2*vo+{uI60W43WfL5t)7OhGx
zC^#Q`8^*gQHba@z`Cnc}MTZSdWr~feBN@GvRsB|dHs#j#hz(g!j5%qD@bi!>Mb*q=
zqg165VFA^0FW1*|m6UC6k6@xuk5Z;A%eHrB=I@G!jh(t3i+l}p{IuZ4!dU1?XrgwK
z#+mzaX7O4VJq?WriLWS>*5*f&mVkrfbK|RSXDC4q?94Bz4g*h->U|REsx2VE-9*k~
znM}^{_Kt63A|y`*6WkS}IZ$nU1f=M4&!2yWns(MD!h1jT3JhC=>A$Jmk9?pjmhi_n
z>ZsizfRZm&E=eo{hAkm;k&hE}-iDGTv6GNwkB1$kQSh?CYJ~;~ko0rB?Ry5J(;%W1
zg+mz?6qE!7+J!m<oB9P*LqOX-`EX}>8BMYP%q_k#4Rrj~?tAU;tDuX687AFm{CjHo
zNcIvm5@MOLUEO#rEp1+(2;QX@PocBf{~uW3x!9{njCyw>B}RE0KF&}r*M2L)G7X<$
z7ZnRW0>+)!-`-`r_@pZ;?(lNTJ@0?s)o~t$pin3}Qtw^U?mB@eV!5=KrJ|YyDc$|>
zlfJdc2WqE7&T5Ct)<vr$%?*kqEOd0wV4(gd6m^R%XC-!^s`q0W9$8>^W(If^CiPz;
zIk81$Wo4}`E!{wR0z5q*3QVoIDFTyWGR$UeM3v8O?S82V%;*CGG6r2^M0VE4wRm`V
zL|u4mO3Sw9V;<`Jdl67;vW5lk(Vs$_dKp?>gD|x&!Jm4ECq@QMd=v>H=Bk71<2724
zl??zX6fr4j8tgRD$nJMSSt8z3sHI18nnc1*r$V1-5rnhpEd^;d>}#R@=jPs~WVDDu
z7S%B~`4&Zn$n_-HGsUpW4kG~}0yGsrSF4qTILfm890e?^-WF%6&pCWkc(^l|wpFCl
ziIWmTH-aeN3tAH6k}tu|k>6??CVkrqJP#;TI@v(>zG&c!n(3P*V-@|Y<{nb_;@tl*
zCxgWD6X>-MKqb`&&t5@6VF!{*#yfYOfLKln6@mcs967YK;9W<rA-ag{FVjY63}1hu
zC~il@!>3SO4r`n3fz7sQz~@>3ZXTwslX93y+5Ra)V7T%t|Ep4Vq%mg{H2*T7@%OY#
z@F<JQap9Jyw|AI*B^1^r329}5;rs{FE3gA;yXX7&bX;7@z=Za*F9T7)wIMQqdL=Y{
z655z6q!hot_%dXy1PDVwEF|--m5S&#ed>l=()ISP(H$xRd(d)){idZ~&qW<DTKgLH
zYnKe;G6PZGc<(HV^-5T8%A25h!$t3{8EfvlLev?OrG0eIadv`wtWrc&TVIAHQf>_z
zH6QORy>l|#0mT`Th!crD%i^s7{K%tr{Z@&V&r*SESdvU4$qX)m9rTSu$u{)76cx%3
zsUnlH!q^8VKoqi6P32(30ln>yA{r@&7{$M;)dZ<e?_(<~nui=pNc>`{V3H&;a;0r$
z`~ti)ZR6uwz<}2Wty5wmYg_dqhpjpBKq}|$?k%MF2~#VvvPJ?-;FBB%pFmBT9f)lJ
z)B!y#MI79td1o`|2N%9-#3?e6+m)V27q6vb`O7B8L5H&ZXsu!g`g+Si_%Yx^%xJw@
zG#F}-1u3{YaZRRjuF}!&Dz53))|NtVsx*h`NU89=0F+Zab8zwT`(4kS-(E&?YrZPy
zY98)S2XMF?@Ar&V;nYnF*jL!Ds+pJFVIu9+_%mGPWG5STzV>Ex+2sz0Z5k#QBqzE$
zaP2Ei$I_ju-@@f3vFvvw$kmh`6I91`2)I5AI~VOPHy&=T+4*WLsz=wP9_L&qkZ#4b
z%gwsye3hwcT<oysut&NmnQ!?SBFh<Cy+&8nr}33E{yehY7hIQ1@6wgO>39*ny{I?A
zf%fu&t$jc>cr20`$1&$QwY;3^P3@u*=(8ewv^#7j8Msz9AIs(^1RP|y)zg{Pf(Om)
z;ZW;)RvbifnYk#9DXYPeH2yM3K*stgun_u!*8+B$sU4!mUplQKMezUD0L0>(oJehz
z1a}mC|Ac#+KIdEJ&ZpGQFU$`H)XjFRc~E$-PD>+x?+){3cqtynOKaBNFkn8^_6=aq
zOJy(1?z{8Mdthx~Pgd$n+am8ezdE9sglX0pnkCqSM2RQenD0;}3QEqo+LK%?ANN(l
zPqL?^LFk2&CE_@=PKRD-?KFnR=B{i@i;Rs^R&-Y70}OZO<JkKLQE9W@Wd{HaNd>e<
zI`C?|?MW7oXfg#FkuQU0K&KXHJ@h7w*lA5s`K4ki&)p2*&l1(7zIyeRcx!w6{XZ)C
z`T2p#H_R&Y`)puHO&ZkNlYf_(<$_=dxFA*JeyDp$4@3!cJK@Bh&F9Ae=FtMjHIV&n
zA00TT8R-|_d;WwlDwAw)eE9I;=WLkP+M*?jtahz2s_4+yQ~pElwi088j&_SomTAbC
z0lnyf(R|#EqP3{I^Zfh*f}&Q<??OpO1P5O%U040)0Ne3UI@$E91TRIdmgZskYh?Z$
zdiBg7zH)5s>WlN_w`E&UlyRoS@zP}?@$SMBW26HLYQ@GG0L(dok{qwYW*qT%DCK{?
zBXF|a`-W=xRX&k}^Q%{h$q2?^;)`6(-H>iQt8U*CA2sQBf(N4*og<I3@1U_2MW&-_
zP|5IzVbEm2>bU!jY;qw8pA{i@(jR+*F$m9{k5)^_xy|EYMzk+koY0S2=rk*4T`5N+
z!>sy1jM`;ir_KhA>B!)Eb6)!$bDAXL9~ZP2&`d5am-3GC;!a4c$!))p#Jhy^Uq|C}
zzFi;@XbSu&L{HnsjQ^GXrP?t;&h3ah0{u_O%O9)HXUmlw9Wo+c@tqsaq3BpxR(v;Y
zFEzHdc7`r*Gf%m>-92t#XUJ~PLfe9tuEA-uT)8j7CUtk1<2hrF40ihlZC>|wy&ckj
z64Cr|xa>*Uu9VG!!^;SaXLb=%_xz{?)XomN5O$|sV*$V+S(a&FGCH<>v)pOV_YI_8
zxkpQyf3R<O*AQuNy!9eOUQ~W@X{g93tjV8}5kU}V7#UsyeADgQw_kHcy<v+84hm8*
z-Zvql(yC*e-TC>$tO|1Uc&OdR5^n%q=}DIPUXZMyT#d^~YZ4KPcx3Wd6>r&Adf)wL
zOLFzZMP~#2DofWhJ}DUfejrr*L&Wmi%J2P?*2?};q-2nOnBPrrarbv%vdws`QxPlr
zJWSLZQjSX2C4Av6pS8P7J*+fmQOq2EQ@p&e>!xgo(%S<jsd2|;Ch;EN8?ph0$DNfS
z1t8qn*?$M^zmPlGD*1`ObYb9o;*UdHA`G+zCQ~EF`QPX7YgVcRvIDmQJ_Yx4Ft9;L
zZm!wmiI^Jqp&zYxCqj}Z8@!jneU}3zrmuh=5Lf-Ku*+Qo&elza^N%n~mb>sBNSSj(
zOC*Z3;4iMRdTG(I%Cp?n){f3E<?khROtW$pUW&5C#5UWU#@3*h;<`D%tlu5)6dTw!
z+c!Qg$PiYrO<f{XsFUMwdLWJTCy}QQD%u5$gxn=0CQgxw;avo0_q)oeMDS?kp?D8$
zt19!>&f2JY;5LA47zL^}V$q6=y`XN_`ev=Ce6yADt~m6!s?d0{M61p9ptv}#FC{gw
zf>`=a-&Q7A70h}p0eUaEL$CMr<fsV1Y)q=(p1`etj_B#70wx65w#Y;DFQFJpOIN+O
zm<#O+98MOuri-b0_xM*wqG?BDzfz}f?-RX=masCr|94r<8cuD``o9iJH5Z>Ga1@Az
zy_J*c`68{xd)MsCR9^G=m$=3XlP6CDsF=jbg<LA{Nk`sUj5F`;1pq`6)Zc%YmVkrO
zM8FosX)uRH%cMcqvp?x+fMrC<eV}vO0ra6{Ab<_n^*f{8rGHag_P$KTpAO`2vsomU
z0XX#~crV=aSKi~!UXxJRwG=e*!A~z+G3&>X7r?S{82o&MPtIO?rS}}`=k52Eeq@S`
z9WGpt_2_G}wu9(Jy9klP!>?4xLZ(8;ZOR`n)ZY2Hf7RT7src6(d8(eA-&uKI=C~9q
zS!h#t;Y~{H^Fcn^wrf|Ej>Vb7E`N7Pn^o{;Xb#P(6N&3iamLmvP4u+@?<)}qOyRoX
zJumG>Q(RqLEu+?9-(Ns>R_Jo#2p)JpYZ#nT4yYj+K!0TO4?;g!lz4Y@#Hwo%9W}DO
z@j>^$XCA<OK1=BccwPu{C80m4Xl-qspi3(qnS7O|b*!Z9HfYz*0yIvo(mo%sOPT0M
z6$l<;DX1RZeerW<bkr7`SzIPeN>FsqfiApxag06ij`#A)gTJdI7G6E<*l9gePiH{-
zmkq$&8|{WS{oda)lK436FkrzmWXPLm>hD51uIzit^@OL`denM+t9)y6u=2&hkh_OR
z7Y++p!%52+7KkGOsL~5MJ^hsqRuLU~MJB^afHN)UH2xKzo?h7@yT!$`bq_i$&YLZ?
zK!ltPV7OEe5B*`xnW<Ujh?E0#tO(k%{lHi>VIS0z3@MO8&^>8+$X;frr;~ta%m}h?
zaCbpe?=~K@jNRzk_<=J9&Ts)7b3eqXxT6E<ax`d@h1r&HZbGWAzux*)^Hut>*um5I
ztRBm|Q3`cM^}aDr1BuKSk0=hNRs)kiFU{{WqAd-As%0cET1fw+mEAwT?l%yqRU^N$
z*TuMdG(Ssv<*Vz@$kq-S1*N4+Wf~5Qu5bH|QWFCFl%;Z)t27RV^g1FHr*uJ*NM*&O
z!fu_pn@lNF5zQ}whLs4H%$j{u%GIp8*5zF7kXt&1Q|6<(Vv}<elnYRZNZ100fS@}F
zun?dPbR*VmW4{iOY3{_!JZ(3iB=g0yOQ{rV57V&x2s@2!mxl5DQvciGo9p$5iRz5O
zzZSQQij1Cb^B)Z6%G#L4UZ}9<meeWAq+`APt12ITK~b}}`M-HlIUnFwap@MAQgDjl
zM-~UQL?rbQRs@HTr{(=vTIya$NCf^Iavf?94B(KiDcylp3?z#BzRp*{DKdN1I5B`T
zmX&oM?os;0I4Ctb&U|^=2WbIR8YI#0k(%1ry(c9jmcHl<IH8Sk6?JCd;(*N*bzD$*
zQ<MDKUE<zXla+Jx9M8?hN}RuNaV_*{gx*Ak2Jg;$7QR_+thkAo#p*c^OBNcSpV$>A
zTQg~(-0J%>`U8AcYpYaHztD5NEzF_NR@SNMX%4`^;8XIxfLjrio0|(5(jxmmuSQ*t
z;5~&@tHqepsZ5@zF`Kv&xGogM#JrL#DfiI96W91V1wb|TmRd|(Z!gGknEZYb__}0R
z5-BRAsBKJ88b16=II6-Mu6U&<^taawL+NkzZ<Xb?7<q{tK0u|Qv*hKy+4AJB)?d{8
zJWnGXr0!7%y;@sQD1Lx`Zj`1DTLcjqkwmHbg|yxbuu-CZH@HIM;^L(41z})dSOYaP
zBR#$NpoWznME!DRVIN<W5^r=yJWN9-HVX}ViAZd1g<8R-7C7p(#|jL`ETUeB-3!XG
zUuvps3%dO5h)vSY#r>rtO>VC2cr$jZG5smF&E8ps+Bu!a4v#Gj8)h`ZwwFxl8xj_x
zoW8zIicI07Mn9J;&Q~z(zkwK2%%36y^3m4j=BJQhb;m#UKo<%O$B;^gtg-Cc5ghx$
zY;|SxGT?9!r2@Sz1#<^@lh>rM;{(4AmSs2!IeNUdl8}WF^{_nIH8-h6A@!8+6h+h{
z(k5o5Gtu7I?=cNGoCto*=mw6B(w<lgGl)~~6lyFUQ1+z0=r@pidJ<Pv_Il^9`O&@t
zFWn1dgz+yc5ZjhXbGlo%9*;Zk#T@Rfvk)@jFtV3mi~*pJNaHdd-u+Fo%zg&yxJpBr
zSpLOMYZOC%EP+<o;O@=4PsyuZIwCdJ^-9}y^gKO%B0WEt<fEmTay|{%<c=sN#+Ldy
z6%M#^dCgWaOOLP!RJ{Bhp=tfo$V#FnI+*I$oyOF3H6(~`^sxKc0jA1mMFD!tP$;S?
zrQWMy%W+R;G{mcij6-`Dt@RmDA<-?($(dS}FF7IyyWP3;TXwfDEzT3l2i?rji6>fP
zOyNluQSI5aZkY{)^Aa2IGB=jsWmE}6$@|~<OJ6t2`?nhwWJ4;MAFGAk?$49R7EZn>
z)Uwu&;D&Lf%z0J@s%eoXEo+uZ4otZIFZV%7_~rDV7=ryvcIACfDR%~g$BTC6IV8rc
zBz^~33_N(ODANB;>!}U0@z%57=YB8$4lh^thl5<Y-Qkqq8>14G<Cb|peROJ2nUUK@
z*`^J-@~gY}ZSUg#EwdH(9E&RsLd$wv1zqF{6XH=xyBz&|^NT{OlvEI+^0_|owqmwn
zwtV(4CjX%><p*{<itR6*Lu}bpEh_x!IGDv<PDwlmzb2|>FSiXY&AT~W`1``o1^G!f
z&~h1=RkmkLw6tkYn#Fqn=Qy1ah$4$=NoK$|`YO>7IV-66Y{y&`p}`1=3Sy+sMY#Fi
zW1wHa)u&VU+&W;}=D$o|wEufxP^kFVK3WpE{~jEb{&PgF?glmWdIs-*-$A9!DAnE|
zA({I>-x2Srl^M%Hv0heV|L^~}KEAr~_ZPsg`2RmeTpyS9dyt~E-~9`~i32<Ig6^nh
zE3qhJm|1dDV0K|@32CVw-`C8!I2OEY{=>9@{wC|Adl8jTM1;E6N8-x=USC_~!vFd7
z`TzCl_5`LYZp*B^7)(-62hCcR6KOC>JysNcvzZV^^SAzmq7%T-M3thZ_9%LPFZvd3
zb@@pM#M%#-)b$JZ7s@)>RIjp$7G<ob;QZYU*Q>CQM?By3UoY<s>KXih#JzP?m0kBP
z3>J?fq9~xGO1HEiWzpTak?!1pu!$Q{M5Mbxl#Weqx}>`|-QC@J7W)34bG~n!f6sUs
z;Ml?)D`s5tnscpH7mIJpqvfPBfE$>hvCB*|6z29+Y1N3i?<A_N|77H`7W;w2%`aHa
zIe$zn#)rqCZL-qx{_(xsbu=r`9P*cZZjK$CDP5bBu6I)#YeCdUT1=Z9J&-WCG?a3^
zY&|3>W8`J)E6?kHI$HN3gcE}ntEiAss`{|9!n)WsjU)TBomqD6G?LQuSnRr)c!GUk
z*?Zb$>xwEOuGdd)rxR9L|GD)8?bWOAxgj>Nsn&c#c)cWM)gf^ms-v2BKhzL;rTwWH
z@&Pm^owEs|%qqu(f2aGdk2@g--<tIr%;MBOTM_-7fArlBXV{s3NYW)o@XD>Fnheqr
zl(9l7`u3lf_y+vBK8LP)QL7|V8n!v!n|~sDhz3Ey*&H{n@~uBt(G!Ca-}*7O29d0*
z6;ZJyRaTe|W33R;y+2t7uQH%caY73Hxl%U${?;{jtOg6gq_MTw3O_WXWkr27rTv|V
zw&mU+oKrsKKb96FO|x}tSVuH?+)@QV>f$!qg4kFRs|Z=0lF(xBFp9xv5g|C)%!XjY
zm?>K%#bFi{p}3Q$-~<u=JR4_8OHm3AwdKpxtTuAur;Xm-yY-PzfRKCf2{9=zsr3A}
z#kvN@%f&a+(*Bb4?;(3dUL-q=_=pEn%$Ap=RR=yM-r*2`cxhBcFtqN#L*l=~oL+AW
zhj$^yugM<Y-wKzrv_t||S*ax)HNUXqvsBF!caJ9yf6?nj1=O}|-JISlNN%o{s!hm9
z<~R!>ZsszvXM(1{ZQ*SeYpw20SjVvdq4SYHrQ?p6B;3MvlNMr*AdyZbF;FGJm~H0i
zI=y7Su2??K9FjhB_N;r_8Vp$&38vMPo40P!v$6G#8ic9|>Ha8|NPD4aOUJ-2;YQ%&
z(%~YZ*}ewxILt51Rf-j;X0gR|B$HA6R7O=`r}EuzrXIm=y!#Jr_2zI5z|q8`05HZ-
zwwR(2rFf1Zb<?zt%tt5D2HORWrOzD|kaf?>J5t;+AmuICTz-B5b}gQp|8sT*2~a$<
zvpZk;&cILrv=XGzJ!Fn=VPUDHO+L$3&UA5smbz~%8oGF0cgIq)P-Ne%E#hby7Yw$N
z=prt#k>bYsYcay`F<CGiWsN)jL7OYdd@m3y?`;$4wuvcZ*GiNN%IK-^U%d0z`i0%O
zx>D7PxO`FTXhw7x>v+a`LiKaG7=9W=_(!19RDo5ispB0Vk(fqhLZW8D$qCY-Zthg6
zNrDd_-6u;z9`5hDeEM_kkKU(KH!*P8HC#BY&s=v{PelP3z@CwC+i87gat3SB?{(+)
zY9Xp^Wcqu#;~xcEvA4`v)vl(o3fM0`T<|QqeJ~><+7d)f9@w}MY%aY|3=fIkw}3un
zfP4l7ZWNwhUB5O(`X2-MRf5^Tz_3LZ)jDW?;J~x>+}U5Nl&3V961d@l@-DA?+@`-X
z{p?OlxWhOEVd{}j?eDgo=0tODMk;`fw=xR3)be@om#ol*G?%}KogE?$N}as^v`#2l
zqy`&feyLTuXM|_{VI@Bt9KQ(lv50HOF=k6M+iAL^f9i1W^}&lLGl(8c!_pr8x$W@-
zFruLa&4Ks*%B><BCtOp*Ek}OZmme-2r7PH$X@1YZOF!@<%CPdfcPCf9LW};-T`wMA
znN*=HH@C8Ug+k=oqxE(9zpbG9O>Oq)69&cy*&ERR)n@-6=KKHgM=<`N75}|PT+RR0
z*MF-K(0}dCe_#Bc{{OeSj{*JH-u(B)fBXO6YXtP)6n-H0fBOGHwrYa8CIqa{MA^&=
zgEeW=Al0uTFCPr*X#0ySG_G=lAgR9R;QdEe06obXMm;4ZC4~mbFs-E%P|3}tSt1Me
zomPxs*5Uv=j{+iuMe%P20|O&o$>bUgFdHBflnIUjHwEk40Ko;w4T|z{2np8xn1a=~
z>Xir-IJ-t;+@^_ig88qjz`z(&HbHEC{qkiAEC5jk&sIzp#|O^_3@?DF_L~!0V1ZTz
zr3qm009J(|3I5Z*F%^?*6T5&>`b^m}Z%fC-)X|3i&!5~+nWNnYm+c5Ni`27WS<qj%
zg@M8Mt9NgTKLPBqDjIx6(E;)bPG(j-8WtZ9o3mh!$eg+~`r0Tw8{9TZrTAESQBYff
zEW#&d)t9AO!X-g-?G9|yiA=M<^K7MDlaHL);W!6dI#oh|u!HS(EN%{VAmujfXYZ@Y
zQ)uvs6*av&%-YZbgFe*<dOn331r-o(pcnmVa*O<K^Q~bBKoxa4fr93*;LHzx`2=oN
z_JRtnH=ZGS#SBdw>-4g6Ll#_C5$<=M`wU9NE;M&2DO)0o+k|d@XzM(1xSb|+1&u%s
zSPlRt@N1s@u$gnn2D`DM!LT5Y@PTZL(g_MnI#mJiR8Scp`|Zy62Q$GU-3jKu7!r)e
zPx+iF`%=xThF$TJfQTE1F3+iWSr{25ulAumI2amI1v3H2d8dqi>`t#P{P>1tyVki`
zijZP#R(DOWx0ywo{k=>Yc8t@Z@&);d{Ct{zzf3crRDaV)N`0eL;Zc3_^h;2mELoRV
zVLRh_p4(^L!!;#M`C!0>nhFc2Y1MsQ8KhHx@cTDe{DTeacsJ&wNJoQ`A$*tmUj3P4
z>maz8El=DTnP-8U(bO@<goGwZ=i$ZVwDqI2&JSEM4E<)G0|Ejh&+c7UG!5gr>J0)E
zAY75Z_RGER0n^I7jjaxW6Kbx4HQgpZZ*MsvA$K71d}=#4=eG_I9d6_Q2Mly|$y6B@
z6t|`4=C3=PZ>DPR4fv{kml31eiR?*`REi+z*dNI)N2k6$#nwzbUHgOST+c>k7HJB+
z;&DYDN-v|ACzlG2u|q>N_C$WQ3kZ}Ye`N8iJe7vtdd-D-uiM4k<65XFxkh1p{6ED6
z0LOYPdgSD&5eWK3FnDm!^OEF`n&+Qr7(8z+O*1esBHSz7TRdFxff*`?ySTc}FKU=f
z!rlfS^KD+Tv$KaqMbU}fxqCNDw)(9)WB48Xm)Ex{#E&1+3XW;z>w8}nRj}$cZb7=Y
zN^}N;ei88<;IF@Qc}Fegx-y7H`w^#!y(@Xc<i;BmmAr35vtNg-RIk)aMY^2u;H*N)
z0wl+|S+`)z8@DH!v=eo#WMqXMirz31j1Ohxo%H8qmL)!86XOk!Po(nlUrHT6-P^Bi
zh_19NrS>CA)H%=Db6EbPjx`t%y;hC0VF@uQY4&2gbEjVBmb*{S3BR-byh0Z}55?Qc
z)xn;ls>=Qp^{_Wh<3knRunn|Qfy3H>0(OMS$R>8-;gQYx&Pdx%v1Tm81a((js$9rL
z0;P%m;yt21PT?R-PnF}bo-$8)mSVHo$vlJ^TAhDu<CHfz@W<tZ&8R2D{Nztd9!n;x
z53GRF7H|}pF<3%mI$2c-bDX|et5t(T6CBLB)Swa?8<ZFn_iAY|28-ay#=<UZl$}JI
zp}Nx{lc`E&#uX^@nz+oJqVOdQIR?j|56uPWbH;G$b#o%ht-7Aqjk_cg_`uej(ig`0
zgr0$qC%R>Zj1UZg$?!KO6IKL9>&*IUwTkIzQ8q|LNt9rhEVl%tEQ!AHurwK#wAx>%
zgWo<?9lUpY0QW6YrDN);>v7YY^D;vV9u@UG{SouB%kjfGyVAYt%o9_gzD6p6Vwv-@
zle3cxHZjD}XWX#JaAh~oznH!*lY@!KAJqS6Bunfnms<?yzXxX!&>ak2mEM4=mrB`W
z%&&z8mzb^v#(5bxZW!xc3M5a6rzl(JPL$8Eji1!9AS_oF%Xm|E-xt7FFWpYmnxv^#
zzNYxufgPHX1uml-aX1RK%%nyOm8`mAvuzG>7`*8JMDRievL?z~@xU~<KtNqjCmW7z
zY<?70oxp=Am$#XF?l9@$j$?KZWoqqbj8t5@bFGhVw(QUsvJjlniJe#lNVM`b!pF;-
zDVH+^30hjr{+fMS;bgVn8mZEx9dWM)M-4E-8d;#6AW9s|iKKSDNJ<>jiFYy5)W#E&
zndmAT3S1=>8&W<eNL~Y%M6s{g8kMb#YM<@DeOoy~opb*%@1@09xlE6g_{73r3PzKU
ze*O*&W5cFQbHsix?fLaHG%a6yb{=!Ol0tCdX)~)#sPgo>fURQiqlZd;>v=om<yxst
zf4$2QHvr?Fl*!2qgVVAdKZJ^#GHejOB}tUUO%R*ML$Q|Nxx^-B(O^41)}Hmh<h9WJ
zyBh2tfkK%eFb)b?Jy3meKq4j|SdCuK^L^Sac$wE`<DBJ7Qcg+t@@dWg_a2PQ_rPZI
z#>XdSX&%0raBWEsPEuFOaQs<#C+A@Kw3LOXVLk7Fkk28=GT`Q4{6kLz8u)%?xX5qq
zo>w>t@65o<>djYA#a7TrKdg+5qY-Y)q$v!D>6J@+wo=8t2#MW8^5Een5iJeW-UrV=
z3iN~R+0;$hB+PN<Fv|1lp$-Ew3TFPdf@>oTnF$Fe_A5*2z4RR5Bhuen&cPB{HB?dA
zKHEMF<!Pkm=l2&_5Wl~H<(&KG&$DEsKmbkK%T(!{4%;_9v6#f{n1pgA_4umySPbSD
zRuWfpl!lgZ>X>iIjI+H9-aBbUi#NB^?;oc-l*2TO#Dn+Rv$>rLx&A(^^y*>3)eS$n
zH>y>F?Yh8;Tpx99(>pJ<*Qi|~ltFohr{{8%h81%F9z)g=wV(ZASOkLryF~HXmQOyM
zluCfRX}Q00cfky@{f=D5`^CU39HE!|=h-z&AeMc!y3mF&?f*d>8fwvJ&}U@Q^Sf@5
zpJs5m-$&xIcNSs{O2|<w&~}tos(ZsCgu-vD7a@~BZS*UgX=c@ejyp(5r6dG5d2UCk
zmAMTBL>C)XNB4j4HIg}Byry{P&ue^r-3e=WD)(z#iEQfZ#oJp4(ClvA*@9<3=e+{x
z2bcL)OY~!;`ieJX_xhI+aJ0H7P}feMu1x0`Wmv{oxyRZ0{;nBIx`fYyI{fU*h|5m$
zMY&zF0~Iws#OnaM1=B7S&K5X+<AX~kw|Z-Tf3-vf{|VlVXX{_WE$9eQj)8$6lQW!k
zC~}sdtIp9EN!Pare8P#&Ak0&Z$@D%!%Kgmu+fh*?F`@GJPY%pSg+oOAWZMw-m70|w
z(Z1ihYm1wfJ-e#<y#z0kidC}vMl%P}fhmDjzT9e~TvkPVy_cFI77RQik=r^NvHQsx
zAMgIu_{`r1JiD60pjRxLnXkvomkW-erKT{*WO#vT#F4FP>VQ7+Sy_sNujS^bh>G3X
z(~y?A22+N6kk5N7oN5`0TQER|ou!mQ=h@j<_f}L-1eu7V(a7vo=JlVa5OO(Z!G6Ut
zPyNqGf8YSIs5oQuq&9_RiS*Uay;Oy5Iy&)=W<|em)}FU+;jP5KJzqiZw|19b{TQ^h
z^laC^<B4N_|6Z8B#&p}^PgxX3%?}_Mq2548I4PbhvOljKO@8c;?vlYV?cZ|v>vDeJ
z0dfO!$F7;TH93#3JXy}EMmZc>lFCR1_bn?b>L>AXSSDRHQbN5b5I6G&wpd4d|LNp;
z8c$pJ>QcHwYkN`4pt-WK><`^&&c5c?+-WkW4in4;Z^T8#&|Of@0C^s<-PW8KRcbN=
z8^=5FX4&daQhWP|&hP)+J8+2l(_x4k46P84MxH@Td_0>Hz-+X3qbaVZ`==xJ2zVnS
z%4$q>c7rs-E$$T}vGsdXd`u=5jmwW5Lgw(<y<2X_<~%&D&7A{d)ygIt+L?;Q>YT18
zb9yVIiIou<*{Z?4efy2v?EBLJf-T`83QGs`8bvFaf`s&7fCDC9%F2vEb`->?SCjmQ
zlNRE!0Vr*M)oQ<&u#ZZNDdXMB!9WK()lCbHjT0o++p~9<iQn$sr1@w%j`QJi^dhWV
z?=#kl->DZEiGvN&t)HK1q*sE20_dJmDFyfE=(vs#{q)xTX_{du`-9Eawos5qsd`iJ
z2MhOR_w<qDKC}Qioi8@NzLK{qLA|p1pBt-wzNML+-(1a1TvuN%qxS31FiomG&%V;T
z@FRuja_0m?i^*h_R47wTB7F5U;v&<~-qg9=<!V5M=>Cw?PIfyY(R?4{ubsQ>Z_Gz5
zBa4^}E>y6;uDt&A9VsXEOZO+IbCacCY&9gh>3aCtpN4uR(@0o5lz0$KzvAlYrOdEH
zdGveXr4Tb@Hx<e>-d<AQy?W*A*EE&N5u!en&xATk*(zH-^P?e@nzU!<dUiz1qO_@%
zSALk*71BjZ@k--*4Cg?(%AR~Sg!=Gp?@%S=n>CL4fGW#g$vcWgmR^r@um1BM8mMrs
za@$m{*gqlz>#F^so&m+!81Ai0A=++zbEUhAR8NI`HcQ0bj-u!vZN#@wE~gMkq0W#o
zN;H(J>;x=V+i{gy-&DB|!U-*5z7+Fvt?||V-rjOcaQTHsZi5WV;YvmhmL^j{OOi=-
z$6&uUR4s(BLP8S~!s)n{30+B#LWIKr_w^jQt2;jD%AUWNyM+^AnF;!ZgV;7m*4Nkk
z@%@h>h!+A38d-Q?a7L=Yg`#1I7+Bko?SRe<KN_QcFdYq_wZF5qk7m$z&AtvwF7=Y}
zyL_}I2Z4Ta{)3IqdOb)#X3Df(`Z<QXC&tyem1?=?5^~KNgZ9_6*BaeewM6jn?BH#N
zh;$W|FuV2QVg24O49p6b!uzPxC04-KIg}3k36QoJuh{e4k48{xVHV@~XcY?4&*Uo8
zcYw}Z?uj%WhQs$q_N>|xYrnTF+nLtJZc>TOvTP<dV_eDyY+E-utf&^0<(qT)MX_~2
zeN@V-SkkfGsnr)q+06G3kxNQ5*4zo;-CpM>&AMEx=bLVfORjgXCMWIaiaPxa%0&zr
zVf12%mEKMRvc){+p8Uv;6HBnE``iK|qvFo`BHAp}qBzqYO<z_dR>$Tyt;dGXx4|C4
zxUcVoKk~6nrFGhF<>J5sOfK|8Qm-Eer)NDohM~=Ju%cSUx#W)%7IFgq90Pa6=IGnK
zc)s){BD5kmM^gLj7me%{rQ#xaHnu`U6Em)|s*Myl>XdRB;6CxnYx`g;p)8uroGT~@
z<FFud(=L#fpU_+M$CM(L2zJvb7+8<K>Inwntt+0m`x8`B5nsBVp#pc#I-96Gm8mb~
z4}UOsNNlO`J-*@n?VVkPkS*&{P3HAi4r33Kl^l<|UVS7l_gK^%Ye1j=v7=Yc5rdER
z8E4K%6Ld}QVZeED#wRCoRDev?%ag<ueSOj3jPSgMnkazC#zg|Io==&S(mFYq^;A%>
z7;0>R#-w{d1`K@Nts5M@iLn(M($eyMypm?~*YZ8DpU)}KDLU6Vv()N-?bg$z8lTj@
zAF-eN7Tnvbl&a3zz+sj0o4>qmu*nhP1{_56$?3K8GKAl~CK}pqUAYim)ZWn4Ytzx(
z*XfPG?1ocP8IAf>@B?VHDpSwEOaZ->j*#H|?9q=zpr4QOP#TTyG~T_<tb`+s@4`DI
zRWv+C#&DUJx%erpU-&-RqomJIp0sw_H%|^~F4*jl7jOB@@%+)-FicFA57x{P67cZs
zZs*DKAJre%Lfhu)?Ibf#_BqZ*GC9GI!8X%r@%iHJWjDM}V9>js5<eb#&vtr5TPCLG
zy1f@-s+{T&wOA^9KpN18?z&Jp^(SSqgOtrjkzI?IblL@KA7K+J<opNU+81`QjO;V#
zUF%(&((64|L3+f{d^EWwj~KP!*hB7W1p=b4AB{7eM^M3&%YiP7=}85RS!<>$*%=mC
zxF${M?}`yEBc%7%r6B2WO3oaOCWUz=WBW{%zTY7<D0Wu4_3?jDy9vdFZSbosBL@BT
zYp|M{D>)p6IV)s>u2G`z-jkm^u~%Z@cvmGrO)AmVonI${TAzBx&vi`Ux^*59EwDfb
zW=A>0Njf<tV>Rv4cyhY}mirno;uWv8IKBT-`G)=(0zHZP?)NGk-856^FldRYy1ca8
zR65`5(WBsZOW-#c8hyBjCWi=almi(>14Eo)XdRdjhuc5pQG&?$vg)(UUAe&VAg~fq
zp9v?0h(cqxPMc8<^B<-<Iwz}3HdKZN46Ls-K6ai<5lh!g1e<!@$OI8t1|bn+h$L@t
zI1)<s3NY<!xxvo{1qrB@zt$eN4!g4ZU7x+?tn$*#GQ;OjKPcWWKG;aUxvIP0HgWy0
z)*BrM>4_IFrkSdyX+YgCf~$JgQ7&!D=lO^38kn11&Ib~~%mEuy>AhFOhcUl4<G6^e
z_{Qwx`oMlyfRc;&_YtL>3>B7DD20KjKO~?<acO9%Bm|+etbH<D^K9R3#>Dla{~=Lt
zrik~#u+33!nXjW{6jw|J*3Fgbah9HO3yoIwdeeblCd-J~9L^KFS!LIg)y<I#s5)Hc
zF-yXofuh9SZz3q2E|^$vAc55Wn*fP^Mz){bHXwMITp#;I7;%y8+CsMiZ4ZS+?<6$Y
z$yB%bkdCg+{(8+{wmh2vp|1k`iTHEzqTzm?jIXd|v=ZO~(ACv%n8h*!8kUc#7*K0l
za{KWD62&OfL_25yTlSkNLYe5#hgCCWv)`)_Tb17G_IoADr2I$;Gm$Wp*xQ+URAMSk
zq5PCw(ZtSngKgSw*-*fM)j(P&X=}d^H<+jds|jZE5HW7M%0iNiKfzn5*Y~h84BDW&
z?b}3KwGx2z={YAIF+SGf&eX0udH=OMxIfR({@iXCqjEUa%hncde;6~1GClR*?F@;J
z%eMR#FPqYlf<*l)v>aMT{CFGlyiSYG%6_HZ&2_I&>N@nu5M{dWuT?gW;vOr@iRYS|
zA1nl;rq*sKRbHh1+jRJ6q!jgd|Ic(~+)AnI(OltRDZ=??Rcgm6ZY<nC)$+-6pqG8c
z>x%guDRObIjJHYO_NRWcRtc62z6wout%$P-onaa6v6@ojg-XGDXQahtAH8tu2b)BS
z57(I>F~nH_6)-8G2;a?*>u3NA=8U+^W}3bDH9p)ND0}USv(ckvv-|{M->@Q|uo=9!
zHiIvoH_Wk?;YKL6se2_|X?KBJ`V5d8(ZNAEZ%}9!z)dE))dKmUKe)5RdVjI@ATq6F
zfLF9%&9tIzQ{EH)aBS)5mL<>5js{y>adwnDxo6zG6`I!AM%<4ci?Iw$7#5ji(M~5<
zxlsDK+d8_#q_IEyy+U1}x~kgJ%yq*7``ClY?F2D}UYSg*0`3qYAaD}fG?)X^Lf6&J
z*_LwT!?7yS#YWy-3G8B5JP^x4g(5JHI<H@wDpkY0O|uxO1BenqNDE~PKFFD_zDOiV
zEvR2a^0>V@Jm=F6-ka-q`)F-b8)l#^nls6*nE2^abS|$l5D(=HxS(nQEWu(V1?V_p
z8~Lg=Rp@FIgh~ONzq@dVS}k;(rRMazJ9w&cz-;?|snmC=A2!Hkl16d+b~MLfs&2q2
z;55Q5hUo>#$st$vtrdA_?lc&%4)R^d;3O}>NX1tAwx+-c?GUhvsmFpdO>I^_&WtN+
zZ`l4E|H)L&UKV2|`)avvKwN>jr+-}Q=$fhK=0-AdHwwvGmIWgPfN803av|cxg<`i@
zwb1x12H1Gxg#)cJ%Y^{m=y11AI55~%WBFPtE)c&+6i;bcma9f3+a3~I(Wap%NjbEN
zh+SQPI)v<OUxvS0KRMqP-sL1+#t$e{)vQ!PriiP67Urmo%6JF+Kaknu_PJ6)iD1~8
zjY(cMUS|X4GP<oSj?64xlUXIGiGO#wup;~h=}4NKqiU3?Yl)uo`<qzM@;iB3Z}m5!
zselFnflF9*PEP#lK(eIXyCV&3u1kboW4mrZ%vm3wSg%ASGgD9NICOeAxTO1VJSp$O
z(Jc0i$xYpM>6^OindilBlYI6|TS+bv*yyc@B)#S3=pvI$gbR-X?%9r)`8}_bCc9n_
zZrJWspkP*3zvLC64atfMxi?-`N?QmV%Ws~MX2zuk&%Hda+^{Y!q`C-gJVyyL30!EB
zXHRB<DD%W(EOVtN1r@RR2af(I9MZ>m5hxiWh0SfQ2(a!Tw}s+@jk>b8VpB94pZKGE
z%ptjKxiAhkX@TvkgIMo+wnnHE8KhnCwa_@j+l+gu654d@F7D!2SjZdWJeifD5<4dv
zP}u)b!xjM7$Xq}e_Sk~(t!>)6`79@ppOs09p~zeTU4x-Doa+{~=<S`6m=ZKS;;<D6
zT(-imv}gmewr_|ci|k5m$-W<XEM^`4TG1rsav^%$;;q;t>~LG`TA0|KLiPQf?s5Q3
zK21RMrL`p>cD}gS;^${Kh+eoriFYxnB`$_%1rF=WM^-Df40G{x9o2OLHfDMM^lOIO
z{>lUc(Uru@8s4aI|Fp*E@(u(mC!AT@Dd=*e6|Ee9-C1c<5Yc*GgMn(m!gk4<8fX3K
zDb9&9n+SUjoAp~3BgK>1He)DLXBhNhyxfhnxH0{~qcmkD3KY{bImlvC&bVPV{2XiE
zn%7=!e1wg#U|CCSAE_cz{xAfkvpM0z3~IVTid_BpqFlY&C<k_m_zL*ftp4`9R0gB^
zY*G6A05_`3m7!#;{HhE@qm(67lsi`X2R`a03<7d_e|!f&y4texUIHJKUSDXZ1a{mw
zkLK*JRuCB<DkxXyradKpe$_*a36QdX+?=P7N-2hV>oHEo0Qt^(i+NhULLlPRfB)XG
zVNYsLprJb-b5(59utj{O?BaJA#lzb_Jc#qWHZkk%O%_LE)n2Rp%SX&Eq%r!uwqSy}
zF7Rgzhcfst!eD6}nqUT12w*IwRX1*E3Q0#~4#gK!>F|;kl_?&u?#u_sD!Q%e?Q9?M
zZM53+juTJqo!iY$vsfcpFOG?vKE$7~jz6wjwkx|F3bSLD;VYpET(xU;`m~{708gwE
zzIw7dwikYd5YLrej*<vBoyb66$%`~4Rm;p`Y8ur1kJd<jvgMM4_bt0sdLOHnD*Rv1
zcO8lafhd>`ow4pT@j1u;6GcfX(H7V@Z)=8cb;0-Gcd|y$2^d7gW;r?VRssgY>sKo*
z$lI*ca*!L<Wz}RA8_{?HgaEMC_j8bLPw9OWYg_QN%zOft<?seaFXTD>%y@J`giiwj
znryYN8tqFZfx2{z1^#nBcutQEu$n><MW%^HR@LwM)yu1+l-D-+qfvHI3ppA&Z)v3I
z>G-v#f4|O9sVqaTJ<fhsiJ(n!VlQ4)L%DjvVlJ|kEk}`sS(1R3iK09Pb?&eIhya%H
zx^P~!pt=w!i#)4z^;%%mNW2N<t=cc?SAcNl(E}3A?w`{sF#i5TF+Hn@KarmG_06xJ
zYy&NdSN*JvR>p@X8Tt%ELW}Pdn0|1#C6Qhac$Ii=L$lZMvjHTv6mlkCDZWMQzuYY8
zSRDugjANLHKf|lRXV7H+;k1^^RxV|_-h*j>kd=CK?Rr>rRJy$lMDu7?_;T=l>f&-Y
zAn|Im$BMLyt1nXg>uDZo8nK?zvOSfh8|DH7d?o0@lCfErZU@N%2=k;{`%4ZJHswX8
z>8?$)g-265{JrU}!yCRyC3|-*1jA^-YAYseE1|(6x$9A`E`@Vjay*(yw!>-r<%A~Y
zvM-opUw9^0J)y2D#qe>MwN~?`(I&DkrefljKNQkG?pdU0+;i=w`q;P@_lN4+ddT*s
zz*5#CR!^MCujS>0+1)?wKMr#2Tm*#^8UrgE<hNUjJb$}E5n!OQ8M`PTEFAfSC{RgW
z{I%HaQx;9#OHYX-XXiYF3(X3x*j4`Gv{1!~i*wwf1n73vzLhZB6J~;ix$y7@WDg}A
zB^f6z$J%o{Ke?Q*u@~;8LF=>i?UIchr;)i!ks6s%$#}Or7YsucX=y{-N~2=i7onWl
za;cc8(<8kKc6<5Crel_};J_M<>hs@K*o0Q+G2WjRh~`7X@rdxHj1@Vwk)t(3WaCGs
z1APJdagfpL`#1i4f4RICG=={~I6OG6?#s;t>x&bdyUFtL>)qikOy^583*U9-sb!Dn
zZSOTwZ5gTF@S4GkI$0*Hc*OeM%24q0C;yJVc3;0IL#*y`zZZub*rVyP_cTbKM1@I-
zBB)`V(wvsL{e9J(<y~4I?CUuzA66vaxfz3e4`FVN9D$<}E^`ft;cgmV#j_h*#(X1Q
z-Fz|9<+X4Sjo)DU^jWpLR!<&d>3v~Im_Uf&v9C&_-}LBkf!)D5IxAJpdFdal89Y%2
z`o@M@%^W>-Vkc4z4GC<0b!1q!=grEI6F3KBY-Nv)@76w&K!I>kzV};aN2*^04Xm>)
z1j?lk`z}Y$^kLFTijAck9ks|?Xz(q7(BCixgBET2rjLEtmjAmpv**JH#3HMyMg`ly
za3A+jrFHI(e4KtB;CeuH+(5;Vmpe6IbKtwPn;4bdk!|)UCZek|P1UX9err6O@TCop
z2YR!jA<a^2g|G9sM56DpG3Gozek9T3ElsG&!Pjb9LWD3?Lu_}gSN@=(e%GwopffpX
z`}a;5$v?MsE*BG?x_;r-97U|$m2uf;YA`TxRM*&+Pd)J6+iz@ZJQ?8p-rAyf+VMDC
z^RB^aIy(st&SC=Yq``-#d$@=bkEj3q{eHvGUj|`f*YXEyDAv0dPt2d~6W=I>%B-CR
z?z%nhkBBF;EDkQm7-$rhG@folw&)FW*>E^eQ!@@1eO@HER5kOPAQTC>_F=l^L+hhf
zM3)?ni}hY%N)GEUE3|LrC}9_$=ey_AUAdx`3v*~^5_c-XT2Th(Ms%ltCzCzRFKDzh
zh8xoQE}p$m>c3ucvL-pPGU81Dg|%pPXL%eH7R`A2uith(v02z(xnp|dw#zkfW|7``
z5ocjxou8MYnK(DAu(-z?AKS}TCE$$dUfM^V$-2?NN`V>qapo>QqW`;6I@-lr67B~h
zV$T>jdbZhNFnR3rxJ{ptSRn8rJ2m@wH9C8a*<JE^Vu;VfguUfX)UeD8s3M9G|IW?X
z1*SQDef_uvds@oS9ynp3wr1^W+$+4}wY#v*u?Lf_bIiubpa4J4oEx{SkE5PcjB^J5
z>}6|YIr=ib*g0qtyUT6)V&-!IZBE*Uz>f;Z4mXdk(}SnY=wpBUo`aMWzJU~~rW6_7
zhURBr+O5HvI#uwA^uCvu(EHJ@8|=%oU9`*$v*#PHzX-C&MO?qWg_PXzSRgtc9^NvH
z@PF4C-O~06a}J-7eW&i21P2=m>?8cEwcy{qr+5_}-J;+ydiJWLuQ4rjcel6aY6`a<
z)ilr}mfr$sXpE$<lnVb8cW1S8oOc$>;qC_^FN~k?T_+o8avEW^fI1qZI%uq)UY@Jp
z{bh>O867F0a~&yupD532F>3jJ4u79$QD2u0TiO&heR+%~v~Ps@`_0ZLxV^`5nJC8c
z!8-Uo)-7v9OcCXKSavfHu_ue+HdNt9YSg)6USRTDld^_f=RsCE;#X(Wwdj$ezT!=)
zoGcOr3JMrR_NuILC@;)DjDoGeX?4pR;dgy&IUOD=DJe`Hwlf#_npckydErBmsB<-N
zTY1}(hgM3fDds>y%^wQ}(~P2CCo8k>jh8+O4P7!12@GoKm#Gp;m)|b`v8M4qwFI+M
zZddJ0LG@Q~vIgeme&lu|=eiH}(cFW&s;PN7m5gH8N@G7+RR+i>&*X724YTPy0ydU+
zRMb5^I|hEV?XCLwi7oFA{U);pvY-R%@dGD@yB*`0qN3+1Ztco-As??gQef1ycf04D
z9_segOwfjkQ$uGwVrpu<qyJGrpoX}Fe@tvX`w@wAm-dgSY)&Xs7$;lBRg;|9jLVn!
zlj3uojUA;q0(8ZL$al|mjy0^NOpTgGn$6lL;DNRV>kLll8kOIKj+aApzBo8)+=u*<
zzP@%v!&M*1xCjW|=^L_}j}!EM|GsL+=d3sLxY&f|PS|%kS7-K&%ng-Y^2;*@MS6Vp
zGmUNaMSpvuU#;~}xSPw+FKp8f10N?>ef=b)%ERtMh(SzcSe1M#>zz4UYOCRf>DK5H
z!$#ppVd3b<mxDoi&G>gH@orbc?ku7{H?0%~t0XHIOuc)N5HH}R`1~~(d$b_2sV@s1
z!{RP9a6Q-Vs0I<%8z@G3_M(Jkd^|>$I%0RP*vG!HODhp7Wl7i^@>j#WuqVU%U^uE=
zF;Pl!lZwh^Yf(HduYmAU)JE;eYh>pcTcO~seOgra@8KYP3NMUY$K3(-kbz51FeB1(
zCQq#F*Xs+p(A*gP^T3Wo;ZMm(>pa_4L6(ZJKl)cwHg5hI{+;cSYRsdbRx(+B_7IIU
zYPq-GGs;3ENJX<fK%;oo*Fx8-$v<FmS{EE#-=?x$Hbd?(=KDH5x#Ba?%b4KCK!|I%
zXPZ6pG12zm>E`{Rv77;(sHn#e&+0x##za~RaK=fuUFx<pybz~<q&_>lgYD?h^}5kw
z>~*51w9F+}mg}ZPoZgpr@8MJenAi}-+p49(UjyjWkqR{GVf=uy{ybssO3kxuRv=t5
z1I9=>31^K*^!NpbB)YxfT40LJEGSgyj~RA|K0y}^8%r-K@jC2VnHhDLH!3+h7sIHk
zF}B`~Di1|PZ>=U<BeK++sv;c~IW>Ph^lsVHC=crDifyWG-crWfFKE4igt)BEibC&r
zdbq`tNt&yj?pMSlDz^T_ti&90IJs+3Y|-gZ4#3I<ugfE61qFh=4dqQ641u+tCq{5~
zE;7NEurD{VS%Hi}{v_8w0J%pY1<-~`IoxkuD`aWIDm0x)$!<pI%?`OyM^Y0IR7_SD
zz+4&E#z~1E>M4`He5VN^5ENaq|A8PkUYDrUQ=55o|1-nRR$|9~2WW-x$??*7m)UDl
z)dy~#Of)jPLH<NJ7?@QZ|G;)g`vfneLmmW4j^ty5)}1(&&FUC<Cn}2D>O7#q3lXEG
zz44<ohS+Qpin&+SvBF2A&8;#sh)h~Ev$ADk_2%*N--UkXDeRf8yk^%h%kuSYrlp&g
zSLc1U9EBlDg$H}g>9MKpf*iOYv>3TobN5ntC?<LUFFP(zvT>JmbbBKx<gCgJmmLcy
zjR4Nc8ni0bQV6%58oNx{?yagjdqB|QAH%v<^5iYHvql-0O+&+&ue_(n(b*e7j?1e?
z?Qw*+yo6Zw*yhElDwT7L+S2<f1a!<%HsbTPJ3c-^Vw5fe#aNpT*t(4em9a8XM(vI3
zFQa05a?)RIC=-YV25KtSZ-3Db#^%f{8htrb0IY8GuzCUY3{itE&-U^YN8VRT#AZeY
zV&7FhU)$rg!ockG-`m}r5_NeFtqnK%sGD9`zJgqD0W+wuuA!&Yma!QfNle7X@PjEK
zmbYy{!N_>L&;%8U#=4iNn7*{Ika5`P_QS~^IEuVt+rCnf-{B$A(H&h5TRA8@VJUMd
z-(&630Y(Ax+JinLCj-q61JQpq?~9k0=+>P#6ff~Ih9iwt*;y);e<+*iSb0Ikew;u)
zn%ldYdPdZZD#|}!7pXe-02orW!ij-~;`y-Sx*@6rOrN7e{fi53$ERNrl#n@NySM5A
z!)id>)Os=$7#r3x5m{ntbBpbXmU__#^ecEFH6@CN%<YN$bVtQ1f6l0%@j^$rh%_FV
zhARxii`jF6wKMqH`SA}C$mnbKXzTvkn8xLJhbZ-I)WHQKQJUIvhJ2<C2m5?`SmvXM
zFda)>Tah!|m}1QCj>_8r6AeX0TDvsrb7=?6r26RGph-2&z9gPoX)hoh*-?!NFXqA{
zo*NI%=KgtXw=Q^g(a7+z9_(x13x~6EzW8CK5P*nHU-`<tJ0t1>Jtvll?T?<~C&5#G
zGMH<|0r1Gx=qE}fRtQ>%qq+u3`6p;+_qL2Zl7k)$-@FNIukW6u<3$X~oyy&dfx&;Y
ztq2lur)5RczySXb*gKOrJG@-M|8n}2RiVHZ<r!!3`6LLVW8M5)M;i?fd*;9m7Gu*L
z|CIqxNN&6=P(fAHSn)0$&oFH$ee}?9Ph;CdZmiQ=9ht3j0+o?tsf14N8?s&@e2H6H
z<{mxj?Qij}A#n3uBKcBC#;Jh0<HKq1sxid9Hg-IuP7x+b8xmo5&PS(EVAH=o*o?0A
z=uS<&o(sbFgT}^(5i0JbLC>nVxGL}-xd(K_L4!i{K~qLX5{B}3=!SWYN;ykscLSG}
zOlL*iZ!@_>b=O7!R9wrJFUmwOoUA-+s^ee%nIA+;=!LP0`~c7*w17`n^E(`XiP4!8
z1zbwAN%M*_)^BIW!3kHCgSp-+;zcQn!f!WkzD<7w>`5&fTWNgcI}NLxTl<Grflz27
zPihD8GLg66z`xRZ!T-)f<v!7CxA_IV0qRDajkcnETrJ{kbbV)U@CE4zx8prjUfv6%
zgXU?!-UzUF%c7aZST39<L^PL+>v^%Hvqv#~kY*oebZ^;B1@%D-y^IVqMfBo&KK31q
zT-F{~Yt2M|=taD}hK(7dV9)Z1K>0L_UME|oa<N>AR7Q%pNr^O7>3S0zH(R|BefIVQ
zGd<JInv_q6*Um0WC7PREDDf(s8X6)WHxtFK<?vb^+mFf0{cA!Uw?N|^)6&qUNNQ?Z
zfHafZ8G$*mDa>nZLWlkSL~4^~F|<<X+>mISY@rN^nV`CQeI$;i-~1EN%htk3_gj{}
z0bLub(^28{Q_IYELw!41E^E22rEy_GMiUoAOBSi0k3l{-n2nu|o~f>9Bs!dotb)(-
z5flH43Uf_;_diaa9$CS?rY)^=wa-Dcct1i#$k2oisEwnlT5ne$srKlrt^mz+%Hk{%
zH3)~{8);TH#`+*_wPt5x$ql5zVlYh8H~kYPl3>B^<HzI>A9necI8W{jq?}Shv6*g;
z8dF@lLF`G!N}XDp>*ngh9U*0tx~7&ik|y;wdfo1^$riO?{fO{sbIX*&mre%rw2B!-
zn-;{_*m!JgE#_bJS{dmue|VNVSB`|KC_>hZYi!SYdMFL_p9hIr!|i_j%3-M-J=QZf
z9-1m0$MzpkewX}6fikD2R`ilbc^p2nVevl^^Ns0gFcY<J0RS!7SYj2>I>jBt3o-Rw
zolN8>7i9g+oUK05G>q;{wSAkG>)Wnz>A*|pY2-joNzGEN?gGN%@?W!%<9?0YiL+C#
z?cDt}DJj`>v}R}f8=w-?bB>b>+Jv^A9)!@5HzyXZLZYJb>6z-0f!ipG7|V>I%Gr&p
z?_@t+q59r{G628#!-K@wHKcx?Ilf`rmAM7yrAuyfuplKp$~7_)KIMFSO|!GVV!}$U
zPS|3Cxmf#`GRC{bg#u_w|AhupL@rU@#9~obr*ri+Y2<h!DfKP3IdQ7NeE174j(T(_
zQr#uT;UbIQ?#q_%*XkKE<x&$ZtjQJP!_txAXJMk>x>zW4Dk@%Q!ZJ5TQ#i-q0?f=`
z-Zz03wS2IWscgEPL(<gNptia=0e;_9j~5@`@`a!Z5del2dP>ht?)}aqAoCz^BRovI
zhVU}WXgI!-6RBUs!D%qCl?XTuwgcsGMAs&4%fbE`JyqD2uVh&{cY5V(oQ#dqjK%l=
zTl9~-Z{%#%Brh(o9*1<j#J9PL0f{&RLz_|>hojuss2>)J7aac^gy#3WzSz8^^j#Ez
z{66%+3-(4{tic|4Kf_3WR|^mo=z7-%PkkA~pxku=;jedhw||y9Kh$Jw4`K+bUnI1C
z!Tc%2n~ua+s%AG;P?gK&e*8eVet2kaUp0=h;Ym)Yx%r~i*oP^NO$E&qF_)V!U$M0}
zVQar4X7mFHgLzthp0l-?<pVepX!jgi$R@emSXJjPA)4U%Mm6e?dv3^47vr=$Q&pGc
zS8+>7fk8s}<fMC`g$y8^owl~p=&(A_L}nvHR)aRnTTa{t=(I%h>cn){(d4iLHte?^
zyqLMOhIh3OHXFJK>l`na=m2`;YHYk~w(vfJnz=+ZQXxb3euQ?-r=c62lu+_kCx9_z
zh$x&{;b28o(L~KoKs~9ezy)}WpLRiq6p71GpwCn3i{i~=`L@!O)D)N#ka_gLV-_PL
z-QM1zhxko}T=onsHr$-i@M~NBfxp%DzTU9c1#rf^{6XK38=D8MHeYYuG(A6yntpqQ
zDo3pvUsQkOuH<GMBZr8AC8)3K$&=<x{20gI)Xq}<th(feUX(BJ+)I7^q}&qNLIuHR
z#&7Td<D^z-f(>nqcSRpgWNg1A`||8l+)HcLFBaT-M4?mT4hp*@IN2_+f9?1eugSU~
z(uZVfc_zzy3A^{aqtuRGjqKWA!uI;q6AE!Aan_(Z0P&Bvj!fK!^L$pt#MX`+FWzpq
z7FLE)ede{-pY-%8)GGRVbaw4-lKg_*ay=jgOtV8y90rAuV3+xco{%+v|JjzfQ_C>Y
z$ll_^bAc~J>%7kK+q;6o>GC}-4c;Ig%wYXg==dX8u<rTsQvDO=u@djYy-OQ-19{|f
z8YFeLkhFbP=t@J3oR||=zBm0u4NTFlQv#W!<v2-H5+ZuDuZn+&n6znBMB4f(*Qbrl
zfgq)0@rziohlT#L8&g8KmjkiiOKO+)M_-a8d;{3@l)rpO{Db?z;K<}E!Mk^cT$?03
zzY|N|A#&6LQp2cqH2OG$Tl6_>a%p%Npct0VEL*#Q7o23}%s!o3NpF&C8^=giiCUkx
zlkmA8XAbP_4`Xx2XO67-cvEojKcskjz434j$!Ta;%v^D>o0H{T(&{|^**ZtkWk5t-
zj<v0hf|c#X2lGQqAO?0T-_nB~s1(45+}NX8-8xqI{Q>{e#!iS4sqUpTSG_(0)kz3m
z_^M|wq#sxurHK$`6HinU^?wTe`J5p3Wm{{T%47bInAH_0crv)e@|!#p*t5<DQ{x70
z#}tenJX*Zd#wo5;<;>g1H_u3&NZIxDgdNsxaSiGI4Y|j#_eVEi_le|mOz89$nJ8J=
zeE_WB5pcD|Y0<*Do(_ekjucO~x6>6m<jz`)K4v-FdT5F|5$G&4#7>C3XB<q)Mcud-
z#TI^CS++8EBh#~zDTdlt_w}Y_+NZQJzE}%$N7J-}u+T#db9LT_gl^yElXc4=hI@hn
za*HcMfDc*ZEt>83r8^oK35<^s2^gH{=lSaOi^BE#NTOoq?JlVKCt#y~v!}kumhvHb
z7>@S=fMe;7*jia~7Tw()x=@v=(fY6pOinAd!|jLcm{(K$D~b+_oVe?B{IlHv<Nn7&
zqoyR*iczL-&z)|n<ma2Ltkw5danj%xGe-K{xQd#7ZF5GDq_;E_-mMZO|9<00uXB~N
zu~N|D4nax^q}kFsy*h=xPezlP<;#7K3~Oqh^_Q9E`ij0@9fMwzEK+d?m?GNM<M#A^
zQ<#8tXz-nzT>(U&Vm?<6yi8YY)d-=v_F}R7cJ_X-+3xxAeSoi6*;%J&X^d8U`5o!L
zL_lXx-lm5t{rY+(W$Gd2O3|vSTcf#{jO473quxQa*n(&ELEX8Cu5Fm{%3O2S^jo@M
zm95qbivB|l;Fh*)tLhZf6hUM(_EyE?&A+xjEehaW+;$HoGjCs?)ST2;48l0S+jA|k
zTlurIhgePS0{#Y+wgrMnVq;r!HRVnEyKBu!@&f^+uUT-5c~K!4n|3(*(olqS+9!`-
zyt@YZ&OB&d;^S#P`bT%Ju=%0R^%uz5inBFy@HCDh>PAgKd6i3P*fo}qzQTw$D|hWR
zjuR^yReSPe`*29-hVp<KD%-%oM9q_=oAM{O%ww@Yv+)&QaVorp!<E*xdlqunojl(^
zUV2H`F+TqK`4wwrKWYgvCmZyekhq%-^GJ?VJS>&Jjd%M~$GK^IgrfK>D=W&HvxmR4
z{7cjqZFXAQqVf$6vx$0U5ysj|8uM}KA;udUjUc**y|RAv9*5IsGet2IaA>yyUGaOg
zAqtl)A%p#*t|@h9V@XK&^|Oauo0O@N&itAd16q~8v6`eQFvFpWJ-%DKt?VGm*`6Ep
z0r=yLFNr)vgEMFMW#35Ql-F-pxeP_V<Nc4J;ccs92e4@FS(|8Hp8R=Xj!jY??+kx$
zM^EvFw$lJxTIL{SDgH=ywNrY>{~tuU9t3!~uF;RWhQ>ADAADxbdK=e1TB@d=s}ETV
zVhE^K<#&4fIxmB5*52Y&5IhpvOtF{Pn6Ly|R%nx(s>r<fE6j`t^4KrDoS{O}Q0o|&
z*XK`v4jkCshpdRr2ynWyY~!j;nKFp*Jl-1^oVm`?Vn~1*kE*Xyy{5XRA8J~I-o6MF
zFSVxlPAt~K+1PopA%ZcnN~X9t2W_lZZA@Z=r`!=2KXRyBAeuFFEq{zs_Y0B~hN<zK
z4_M9>2exrq&fSVuLPGChzXn%uk{;pfJq`P*oEddI&&^qM+sA);FN{Hzh!|j@NYv?5
zD-u3wm{V!z?8}f?@l7=G%b-ZfkXHSZG4%}|GdR2mC??2ENW7(ZTmad;>M2EY->=Zp
zbi<tGUY`Q-q-sn|p8ao;A+;6Q6Z$Z5p6WSgkXqr12^Q{BTIC5+-Q~PV8`$P_*en)5
zl!qIp-mxY}5y@lh-pM6VwEmQnCL*sRhIPC;tW(t@9vs>iR5d|qV6eUGAGdD_vAk3T
zwayosj8al+E&^Y~APt8gS=qFLj*eJ9VauV$oh{L|Y{p^DVc5T#c650EHf;LS(P^{X
z+?|^RuL?7FO*_i{_I9s$bd{NhCD!M{6kWM)ws(9=qC&5bJG^YEcsp3~xsIib)OZ{I
z={=5xy}g~~5eUPPD)+iW3OMbOHQ64nJ~S#1X(~4&;ftuovl4?>cannv4TYd;!~{eJ
z;OnUX1r{)Pk<Cdh!2cQKz|19mok|tlVY820^j*67|8duL*vMeD98~Mb+knikf%ukA
zh29AO>T4IObu_7+wg7K`@pyxWeFqF)Bj8Q{QNQ4pS?3i>)bx)@zPXstZw_TS(6i+?
z&K-R}$HF}N;cQeSP2!2ldmyNN=NH<F!D)8N=AfzZVOY$SZv#+>5!>%W{Rv2Il?M^?
zlwL9IXCVPS{jI@(psqJsf|-ofj>q*0GX6~VdIbQB$z>A}n4|!lK}$T43Paf3R>`*=
zvg7|w^Km_Oy%)U(`FJ;VW#3M%(tbyrQ)bb<V64}%k1t`9hoYipNq7xe88!=Q9P!Vw
z0ZuB303S4@3E;K`5?<-HN#*<_mw)iQzu=O^v<64S4-QwWXB!?*oMW`FqEXWD-js~>
zR#&Y3jbC{vtA)Ds&FY3Gw5OI_-VuvdheG+i++)Ne6{r9%%EJ7NjUmQf#p;FLDeP+K
zdX6a7ow?@QaVMu2Q!Z~={12`0PwnStTxt)=zGd7Pst@O^c7qY$evz0=PT0pb^uz_^
zc{DhB*H$)DEndl9|6IwRht5k>wfK7CsAFg>im+q`EW*%I)S2a7LMvg+7wMRTqVz$~
zi#$(qKpZ!*ymDX_Z)=oV^*1rCk+w{}d>MVvp$5Fy+lV(Pw<^b7Wej1C-2pXtG1srN
zXdPO9Nf*Q|ThD70pI=oAvrPE@OsXJZF=Hqz;eB|~ZbuNy*EaZ@5G48*LlhQ|=0Xqh
zpUZoH0W`CVU>XZXoMvoEWccraJ|pPyWPu3kNXv-xT+)2m3rGcF0wUIv=Cj&E8@~h=
z&+T2R^?jXjMy8`QBf^x^;y2VmFN&Pl-AeAaBB$ULFAzzcvxjiAH(LQdYJjksG#_d%
z%j8!5sC0_UxIj%^!LGw-B}ozx7Iok^V>)AMBr>MW;AGI!byj+a(S#TL5(<}v^Xril
z5?;8R+9Qfg6wD^MLlnM=i#1I4#48^->_-o-kMQz#*G_~p+`e6(lrsIRDPzky5oM&%
zdAY|wX*twB)1h(21_RsFM`UJT*cjXL(9wSEal6#s=`)dkwb(KDiWYc^Y^3=>8PfB6
zIPdkj9UP8^_96m@9{2T*dJ;$9P*ExVO64edSKrRgW(MSYeh(s9IdT>BBy%P}+Q7h;
z2Y5k_;$jO~Z7qfzIYry=16?6(1smV`oMWk@4|qIBw8e0J{4p;s<9i~24q1?_-<Il{
zvo$A)J9F7Qsxa-#1{;wm;-ex53$V!#)@o%4-$2Kll(C_;&*#KDGBg2018C)V<H6{U
z<sTOgTC23iHA}%fdA9|j@hZUkS<_7bhcKh8<~%gs>oMttqYvvsULo*zx@>gs8y>aH
zzl@2QHZ!Z+GMw1kG4=q(V_(Y4N_Xl!Wsy_^7H?l-D?Ai%aiP!gCGG&ZaUN?{=*!XG
z-TXUua*Okm(tlB0VSdpMskhEWb)JUp>@u)ycN7#H88k6p@F0AXvKYUMx4qy9)Uhy>
zE*@ZVvN(ImTb$*LGoG>O^K}CLqjVDl1%O%JP88enNJ(m}z3iCs^PU}czPN_P>=_ss
zBnFZa_mq_#OJA~=D2Uv`a(Dz$J^43WxtIr2RI=I1K;O(nUt8CjKVh%6xuA<w!;cJ)
z8GWqKr=c!Q<~}h`ZMmrKB*`5TOs?v|&BafOK?sKvIe6AIyn*u_H#DdVvSeowuA0ND
z@#8<Tvvbt6Hh;TTdVrI|w!6!)*tcair)zmn-<JPbp1KD&bXAG{l5Hxi1_O3G%h~Cl
zj+cR|Q8%i!5U>eS)?#=;;Rj`R!GddNs~rAj^O}9OD~BX%QBkbvA6cmPH<WT5_6-bm
z+hal+cR?8A*Q<>_^elp_XrIKFpEfks#-^X2yg+FuVDaPsZ$v=$0JxJXb`G$MQkXoH
zBHRBM@BmMN1?vB)>tbPHY-!~a6z(_F8Y+$fl3RPxKJv-WDFg$O3h$DbTUf)b(+*l)
zwIUU(cC?YQE1UtNN~9s4@tG@NYD7C)UMDI7E%QQ^J_$xp_gc_D)K-F<e9Lcz>#Ei{
zg(<-XzsA4fj6NQ)%WC}kHEcH>I<CshWJ^xLm0%Jxf$Pl*6Q@?#k|@ls92R>g_K>pT
z$Z@KK+zLgOkc0f{@J-L8#RXkMU60JTY*<&7*nH)W)iM0$`6L1YiXh|QePvGH-OEMk
z>dmMBcK5Y0El4_Tyh?xP;S<M1r#Aekh2eiS^%ihZZQuLwAYy=kO1GCzk<I}H1wrX<
z=?3ZMASfsZ$WYQqHzEw3qI7pRjC6Osd%VBz{l9+XUOvv8bN1P>*0Y|q_I~V!CA%MU
zEM{H?2|*82p6hF!QrXg5sCUTOPWykTt1(oJBmCUU2}_lMuUX^`S(PV>c8g1gl>qP)
z+}UC0!x$;Mx;h)UTj#k0@}pnLMhlaR5(3~bhv<FFk9o#ci&_=9xiFAIIk9cEbG?~T
zs9yB5x&Pe*J>mVX@BiF9;m=orj?~p((jz*Qp&K0{`l#aKyv35zvZ1r854kK1%gZ3(
zpVG$74L?XTT9REZe7-fUySZ^EG|V!?I)epsetTJMlSkaGTgoAnOI~+bQ0(!r!KE7@
z7@soI{nKziT6;Qmct@>u@Yjs~w%Ar?^FM|K!f!u<M7aiA(}STjq4XLT)1x*+z8PaT
zC||gzvR0(H|8PS$=EIBcJghNewHG2{`z0mlG*an}i3M|Wr}f#xQ@n@&EtK~ysy;Fq
zkEU)<>r_bLQU)u)=8%}J7S$$XfXa&0P!B8obQk0d@9pfR)<x1u-Qrdb4%TwsdMH18
z-7?XE8*tQWG2v!ESS`Pp3^n)$^l;X9L@9`&6EMs~L}hM{CVsKE!S$MCWRC0?+{IT1
zY#<}@yE?sfeaK$TX9%-o_fV9;O_qa|QIrlRPF41e)2#MlCmXA~+qk1P{FL3GaJFsu
z{quNjt%2scyFT&NgHt^*zq;6cr>hbgNeoO|NM^t2rPfJpPE$e!h0GePLJ>K2u6BBx
z#iC4g7syzgJ3EIMb5Cs~FkAb}igWSS+KQ4WOcO!82|_|Hz{pMhu9`qCdOAzqpnUZb
z?7q>WVMU&eT=Et|hokGDJ_zhf1%zB8PVTB_IOa7JT3ul|oa#%iqbkc9Fnj6OwC&p%
zFB!zb{5T5=4S-+5;eQ5xVKF)J^cXZWO%dhqcr92G+s_1R7ZxFB0o701W&%3-d9Jz&
zbL>8d$DB``IK=huFz|EI07l)@qq4uV85jI!w6}LUFTsof70fSRN;4xg6u&7u2#P(R
z!^9Nkvmax8_aBh(PvtB@yCW&cZTn~7BGU<+*S}XGg$WQ0eh%FUyRwS6g#{d*xlRWE
z0`ca#@#5*iXU3yAz1xD(Q9sKmxL(pflybhC=hB4?3ftLm6Iaw%dL~W6G_y+F_J~uV
z>)!P>mtx*9NX&qG3B~g1&nvsV4GRmQ3Eo`mEJ<wit;ZW-?_Rfnh!`Dz_4WUjDc;{R
zcc+IHiYCwwD<c+M$viw>Ywoa^7wP^UYV@mk@B60kf5ikfx427f*?^_m`uK{C)$9_P
z;7#QHt|6R~;bcv%+($HgL@r@%1A3bsh9*o+zr~WxEq@!`1(fs;fVg1A3Z&V5E2m0y
zW6=CVHi*+;zOd-sSJEK}`JG&ooE)PO<Xn&~qC(D3I5|&0fTXQ%XV16HWVXlGcLsIS
z@%Rb?%PW5T9Pf!KOVqupa!k5L=1+#>yG>G2NdtHpV`f%%(C2$4?Y7;~{_$l@pj2Pg
zCtnKhOhTd(5n((~a&gN!$IEBVYxAWAX{7iSlsJ_}>upou(cpET?$Y~vm91JG`JGq2
z=%A;d*877{1IOaV@cM-i%B9ehBX3Mwv#%yeJV^2<8k2TP(0%OcXqg!MC{XRZP%^@?
z^S#=M@2$kMf2?VXN<d8sc>Mju#Fpy`<Ub9%kNH^%{#(M-%zwwf1v=h%7hB2`WJGXB
zuJ>2VudhjwtvknK^8lOF(N*EZ<CeAX+XD#&m6Nh?Gvr2!VFc($?m`2aDJn|mu!_Yl
zt*hfhS`e~>EP7-lsPbtk9;a&8gT+0Uz&~azpb$vhjFd9AwdqQe)A`!q7}|4{smx;R
z<%~I(M8~8wVh9n<Pa|+bG@Xwg!a|Z0GI`3&bbJ0(;Rre{vr_3L-hs@7alG8}W2Ug=
zF8efYU{>yJyp}pWy`K=%)8{5m&mba=BApdIzsH>$#UYR;6rC~4I`e_4SDC&zRZRuo
zds-LS@_B6nOPazOfgJ*pEFbk3Z7Mwj_H!-2l9KK93_&v5#-H!qBOS}dWJSgwhb5<7
zY1{#oIj5^eqz~uLNmBB;?6wvMVJ9a;T6>*H`?IW9xe`r5t@4;V3x6W`THt&pPs)jj
z0aYyZaygv-knebmtb-6cnTGF%69GZpnT5?iO8uXcYy7}4>*L*fbg<eQw>7&K2I`q)
zD{cykKzp5iSBddLSZ=Ry_$kGk24{UW)@5ge2$*0yM-c?i%}fWgb44Ob&GzI+8?D;)
z{Ib+zYG>ge+sw@eI-5f~<FsnM4sOEi<>LDUZ$JKzcDv*GH)aIcgAAYL{L>gIDM}A-
zTFg4<VrHyv>s5Q#)wUG6TbS`Tq23dHxO4fZ-nKU_u^=gtuO;UAq!`O3N2P9{X(pM{
z)8+=a*!$>M*fj~B&Eq?sP;6*p(e~+=$qKcIUhz#MhZk6p<r`_0Y5sm@Vh8vZ#XMKt
zvoVhIEpots%+HnG)h1AOE9K@knjV{fZ9V88wLzKXd-kD+{%d+DUpPMW=XRs<<k1&Z
zP1SHZ{|4IPU7a@5jE8sgXX4U%j!>}%&u_2`KIj7MX;O(>NL$y+X<ZP8J2au1l(59%
zbEyXEV7$^6Pc9BlJwqc69-A+$B(+7$0)nK!y?Eszd8ssBYj;yJNQ9Z$%!Wv0=IslD
zVMZNOFKD~sjjXJ&L*7Q7p(XspD_u!fq~N1=5RWVADi<4{f%RogS3Lj5^tXr#(`f!?
zowBNS%gHH?3TrG+v4#d<E%N30Qbhzs935LsrZ_k>p^cm3cGKT@d=<0t1jjW&zKCn6
zH3Y1uxi{+Pth&-XDz0+W&5r;{#H?01Ha<`sgHpQ>{kI~w7ywS8{)qVD=$bN2KXygQ
z^-=xMThCWJfw*=k7j89IXWHq~(xX!n2nyfyan>56)VtUD!Wz=y4fr&78YdgB-$`gW
z$*JKG5b*b$!#+L#ndTJh?s5HKNq8Zjo|uq(u2@c2wzF}dE2??c`x}oK_v7Tu8tB-_
z+NkTSq&dG(V&nyD#7^fIj4~JJIp-xJ{uI$dFyIu$3zHdlSAR1c=+`BOSdt;bzTHA+
zyyK=pz;4lE%ooFx!(C+VW`;-oQMu4;DvRnuP0P57|MucT(w$MpU^do2!{xZ!;`}en
zFMy{}%pY!}e@L8sZ)Fo4e;EaNhqiZS&raa#cK1A^jrbCmqTp$l+{52W;u9M24g+`|
zGNhr1P^PKyA_n%dW0vi7v$%%cn@7=(2a}xE9K*YNV@j_#rXzp(OL!KJH!gR9dK)EB
zi0FDIT1+Vz44h10_)Kd(p7_Hl{8`%>2_rM&^b2qI3Z2bN4Lgu<gL(pY=^|w5*>&q;
zrqbe}C_!T1xX-oTqJ@Q;e4cj#k%_2DvUZaKYB(WSp#zSNLPQRk`uqQ{rs;hPItPi#
z<+jWUtl*hvHP1k=bg`J&#yA$Jv9V*!zE4b$&d12Y7J>{4e`9X-(`#B}ov&wXrTfq7
ze9Iw}VN#xQcZ5uYo%ss`Cp=u0)lkdg=H|6#XZ*)}!nXW|tQ1YfHh+SqvTI{pz<+4Q
z$2qJ7z0iz{TctGa&*A`S=9x9}k0BpC>cU%pJ#8`f;mc2MI5Ds5(FEjhRSNOq>c(qX
zC9XP+ViZ_9pSn7V$|b6Fb4}cwDM8T-3YrFHIDc4F-18=ldJJ1mH;sLL%O5*EkZBV|
zuEUx~ZI?AAl#w?JK;A&YN;uqpvbpVWvW=57J2HjycKw;I`88NI_9ONZH!cJ-Xh5Fp
z;_W|CVW2V!gy8^zs9F<%V%&3Sh5p@MB;dNgrDWe$sIZrm@emTXpUriGL>0`Lj-W#p
z5Drj&Qpo5YP-7Q%lUZk-1cj!)b%bHp1H(2oWTD^C11XkTSfOE?jSJ>>ij%&%Zuk}e
z%6+!W!ulfapV!B$hKrPdaV}5jch;#20~`Vu5C2LE{!FHZ&L+j&v^~h>{!S<db7$|x
zZ1PZjPkl9B4;0q5WB)?mf$Npqb|KeR{vid%*V$4wbt%5l^}<h~C0zTSE?9G&E))k}
zbN^S#>;H%sucSSG=cx_c<(8j0=EPd`FAO<<Q`!8L3u4E(ZGKzR-#YfHA>(A;c%Zna
zY=q<cNXaVV@x!@s7$ncgSbBM&)p9(f_t5C54Vo`I+%?;%{qSpLdB1BbfOHG~T7(H~
zSMLH;8YrMz@^sxQ<WIV|5IpEcX$~o4s^uz|@SL#UQf5&ow@x43h21-_3ywt+0XIW$
zW;Q>s>4a;{#=~P*qzN!}!y8bK%#I&!;^9tDtscW$ifjSF=x2oTV{;QF-ET68c!->A
zcaV*f+o{jbv)g+A0{?U`1-HwfE~a_+6J?mfuaRHx)$v>f<Kq!J`Yj+v7+{Oc0KQg>
zF0f=N6J@=5et}%iSYAQnYz3sXq&>bs@~qcafSN>0jA>f9$u<)y-|`EjtA2YV*6<~#
zL#-P88z`w-Xr@m3)YEOQ1zFsj?73O`RQmL$U$X7QedHScS+7SdsG_Zf4hU%|;V&qp
zy?}L%c=;&v`&Z!Za#%Y}D@BCAUxz4IZf&Wf_RY*}x~i+Tce?}SypKl3FFaWi4LCn6
zn%aH<<-#C)oIw<OKO*jh%!eC(^4*u3XJ>3>`@2Q^g2xlPJDp1qN$e9~rOnDmJdgKn
zg38gDYco5*!)x!CRSHQ|)LVMqdpZQhjOBZPWML_c-*0JJq{P0PSy{00E!U9!KZh%-
z{g>WBuo}?O{L%B8*3-M?<+TTVnf&NZu=8;KAmHOD_`|}pF55r03IBGppAP<v3+-=0
zf!!Ci#w2&i$c8&8NIOi*4v!WlQOTJalhq8?#Y~J5dWtXIjv2?|Nk?Z=Z$0E~J@5ke
zeDRzcX+V>-tDGB!>om(G%ojZ#0xbcY2<$vUQ_)>IMV_Dk_10jBeR9HwXK;OwhH2o$
zb{EE!Y7Z8BH-x>yrhd}O&m!9spv+u!rSWpbsD6`HXN%)RVot5A+UUUONZ{pQi_;_i
zWVP7hADoQI`?;%bWtMj4oi7dS^*4s7L0-7KVSI7`P_G}y1VMpI3IAwa@B}C~&xfvh
z$_LHA))?`P5=4G%>tdwkIamLzO4|g8)@`{2<?W1<O`lr02kkY5DD)+zV%ubW^$o4j
z;>S+TN+M%0oL{%ziHfDtfB;KcIY@ixc(4#PoNYLZLzPHXBEYx2fn0AR3bt~I>osV2
z_CZ4C+HB^Kwfn6A$oBarZ`m}EHGiNGSRE{~4Orp$Z-DNcp%GENN)GV8tdw1^4?X$L
zTxa;9cs1Xu-xJohjOONQEn0jQloZCspp;Igy+^80eN>3+2=FSkLYAU{{hH@;(3N}H
zIQ8o$WUmlkQ#Gt@aA?=A6`zq>;nEz2IucWMbX=Rtipyua-EeWw4-i{Fjg8y8x|@5a
zla9J;tvNWAG>91)E7VO;gv1ZXspMW;kwaTycxA`CEyG2=d<huJrF^Zc#{1W7k_;Az
zYcOSU6?4JH3dX`pQWZ=|U=+R!6xOfC-0b#=RaAT_yDVkl!j<JE(BixuD4R*kE%xqX
zrF)btgKe#-<@#cwo5Y13m<L@cLP@7Q*^xS`&#heXJ}@u{PL2-%Y5>LPjIRwjx(;6!
zUa#r((bv@37PQH8Fu(<hC$JB7dL!hTTdL~9qc%3QX|ev}Q|4#f-LD5ou1aM)coXE}
z_$1PDtpd0{NSvFQF_<n#I|SlVcMdgv0i_CrvJ8QTi8Y@h9wh1TXFGMC+3vYkAolVW
zEk8SD3o)soa_U8sprjpVSF0USD`RuZXV-p5071^C`r0MckQXI=Qw<J^W}tQ%vng1+
zh>aP&o|4H_p78V<&aJnT9M{%-V{Pcr+iJ-k>H&dshERyTwS-BvP~!>GUZr6{PciBL
z7wz^b(l)V8xv86(fxs11TPX(2@VUR!L27b9U3tZ2H{X_&Vm(6Ukh;CPB+vu?>@(sa
zWkJDO(v<D%$I;7o2Fh%P)C^XC|Be@9>+Oe<_6wafrsnGNwnVapV}kClwfrt$^RWGV
zUb7S?h765V9|8pNnwGa}a(X<NZ2rZ%#=WQeQ!i!zq2t8IQBj$q+*q9+Nij^qEGdNw
zE?Wuy>)dZwf2~ND+-qi54hD93B+Oj8?)<L2JXYhN%(?6UsPx*nC)QdO#p|z|bV!X-
z*1iR5e^;otJdgWHy{D(XyrfTjB_o-sPG|cRcX)1z7p6UA%u@`Nf(v1Zax#mnM^~Q5
zfJGpPhzkWL^o=Y*-%=ZhqkBieEwjo;Af%V8C5uv(hv4Ld1qa36^<sv`c|m<omPLI)
ze2w_ncHr$7pxEl_kHbo#uB5Y7-pE(+qmAaaM!#{op#7P$cERzFG*bUTT5^>|LLb8S
zo_VZlAlVT$<CF7FE_&1a-bOK>nLFxHC5OBLq_$59w(~r@<c$I)A86jBuJS!k`$Qy*
zON5qk2Z%FlIN1M;RIAbT?9B5Cey0P@M^DKiij~mwxRi|*uS&VSFI658B%y+mhMK6S
zFhDkBNn{Sxq=4#A00ck(8PKw>+lZmwa<36203}+gcF8k~Af%rMwI1EmQ_zl+b()*_
zwfcw5%-6a*EQ1Xv@qdO?tE;N^)UNd@eQj<$C^fr(@k@0ijI0GzAd|kFYb~6#yxs_@
z;2->d;WRnAycYl0?Ql6PbOsR=Bdra$UVU{3`-lup-ZVx^-nI^1oyj8V*3}bAiuwt{
znYtDk&d2kk(3iLOa7|usjsna`K1dL7`^I-r5oQHH2O*6It12rsP2YBejPit*Hg}#5
z$&XjBPM4J{^?jICS;a(r@ciNeZh_q1&Bp1bcNkPRUV<ug56ujHpD4CyZ-<o|iu-(s
zm-Jl)-eF>59_kJ4H*nY{d)#kse*h80S=}if*!lu3+8TFnu^0>OojljSdoLB#SpjC{
z&RbU#2^@nMJesy)^OQHBpaO8$WG6jd8#QltpV9ocuzo^;==Xk};%d8le1r}HwQBIw
zR7W}0;*@lMhMePpDEX@r>98Wn=u^f;P?K<D=VC_EWSA$eKrz?d>=;89VkHz6aC{z`
z`DiPf=@xDya1pHph|^Oe&ALgN&#&f2zq|#j!QFl$%IisU&?sQ^pVCI`?6#M_H#Hfl
zxtv-wuZ)O%^^}Ju@TP0^Z5~1~pS8O%F;U-m00$54#8G_?Ap1x*BJ?Hs%GT<$m+gfn
zWnrO~+u<#WqUnNTK?pLa>RDM=svMix>!|EMw7a&?QqT9;TYvFhs<p4XJ6eT@ad37}
z(wqV$M$p|+f$6F8O_6(Zkh3Z|$~ndCKLoHd%ZQExN^P5@Ny$jR3owR_d*LY+sgC68
zSt)TD^V>1ye!qh+EnV-uB9pWm>3M7-QA|xw2oZ9dEglKm7pNAga$EO4!~4b9H_#01
zHBGTfjaPSHqu;W9-8&Eo*;9!%t$oaQvGWrBC$6<(6RgAi#aF$v6Yd*I5eBNDgeOb&
z+ncGDr1$0~!?{W1C#)woW|L~(-(+Ut9Q8UZOVxSfBtDZY8sYCKlr8`=7_Rb36S+f>
z)k?}8GfRt`AT7e~af+3+8<UmA?z4yvtb4b_+In)vgnJz(yDgF@o$dhK{X@Xu_tdY9
z9@}f2@?^2*7?teZvT~!vlLupQBbz4-jUh=B`%<VdP!GAYrqD1w`RTu!Xi-|UZF?h{
zlr&cR<}p~td=E6pPM}dnuttVSi<R?FQ60F|g%?3z4qmf&zpmQZKYp26ZEyT9*w22|
zNv!+2Gj%f0rJ%tRGdfJI*<5AdPlGERB(2~2dnASfN9^asl<qd*nt&BCS*0kx@x=Q0
znc23`^SrQMF7I@EenvLkGiZpu*mImo;6IIE=i2G-;Dcb!E=~-YIUnn-1$IFBOa~#~
zb>;n}gx)BQnv*{#WFQ%LGCuA|#}bFzDnC_hDefrd>&OU-pa7oro3Z9c@9|g?_Jvi)
zBsGtPikvKTd9c5h8x!0P5H2jI8AtLMbbH2wT?aA=ZhJ@b1WQq13r~ukc1J=`7!Rl)
z?0}+WiQxhv-=4kmq*`o<28CGZMEMw+qTlhsuC>E0W8dN;_eo8s9pyW%&2;2m!T&A2
zaLP~GdcAIoG!9nCc%Yp(19b5i6_PJfv(p8Zuk`Reu`3d5Tm#<OH%Nn63#J!e`DAL8
zLR^OeB=OssA2aa61E`d^o<pb=%(=z2pH&RA#>Ps=8P=9niKzPP2*B&Y=0GBkS)67<
zL}<apP8Jkn2sk>9Q&G9G^;3>bh0E=W{pxP2j|i`JpsXeQS*FlHUy*Z%02A|;mq2Di
z_H7VvkalEb5E46ya>c>>OO|4yjV&H6w#Mqu#TiOc9}I2R04@lM7vm=YSQ5Qs&%Q_B
zWd5!kl`&w#rB2Vl^20KBa*`#&Uf9PO0|i$RC-k6{(k_PBhE%}6Y~f*x{X`RB=Etk&
z(r^1{UTeul2ZyT{tbkw*_7?fXyetJjDWP)C#!d=wR3#o*pnwjgDe;gMNlv3d!ryy-
zAF?m@ghG+?CwG&t%(y_>1#B_{m7*c_$y=3L+NtW2U>l1x0b-1RN*de&WRuyUbr(>4
z*MD;R>I+V2Fyqc|jk|AMm%0d4a_c|nO<#}^x8t1S@mQZI`g$B@N+lT^tFT5#X<S{m
zH}=8kB-3w;(#=0pW{z&_?BwF(d69NZ@CdpfiHeE-Ry1{gsH<xOqgTqkPgeF77ccLx
z?}0xYenJ5@J9m;&VxyuIm6dP!Wo9z8w6q)>YrE_LAxOuqDIp-Hc#xV+d*>&d_F?TI
zpdyZ?;&glVXF2D$=~@;4Xy%$5T6NTth3}rdckm2H@2<p;EnWayPx)Hhg@tyTphcB9
zA&W-IhPzs>&;BteoNNSz0MoLFSvkgN?o508mEOd#@cd4n<oRa9-&X;m`<q5)ZWf)#
z0cSTrRMlNeU_M4gq+{HU1z@xO=F~f|bNG#+$;&EbnMj(LG{nocnIH8nXT1`OU^`JT
zzD5ChCpKTpn2r{=`M`6qQxw8hF}_6DgOYLYA4m4;TjC#ezx9!W>hHGeTlh=OtAlTm
zrGwUqP)+Z(elJLkd>zea+jhYL{=EyK9RFA=Fd+cb!CHfN;^Ob-GCQV+wGu>tNJIY;
zehd6^bKa2g3ZGt-5acp7uSO`?efIgTguJ5~!l~&P*b<<$0p$&Lj$H?r`6?y4Kj@&L
zmAvmCKj2M2%FCg5u>Q)(jEIj8O^POxOG+f|`v#}-PWUI{LgeagX~mDFPAvs4XvI0~
z{RfAu30x2WfL&L-B^AxHogtt&b`3dk+Q2N#faMqV`1*RS9*}hjE#dT~8Buj0mgwza
zKtegBJd&88w0U)Y<UKXD+>_eEd!U9?%KElQPj*&heB{2OP3^J4NPlnabjo9TKBgcz
z>hW71p}3(9U0`V5A}i9}bf9Xa-S8ZZ5dfB_%fqoJYvb8Mp!Dxc0w}Au;$PT*>zEc`
z&vVOl!|w&vdiRxv8eAHLQWQhs>2VOtmU;}mnW>V#(z}vxhQ_ZI{#UUuAMv-i{`XT>
z)3e5_*ED0-$>GCY!_g#C;m<?k2N24iKZnws4OdrxC8yx3b-Ldyvfhxl)xgf9pfb~}
z)X7Q#2v4(<iaaL+uvN6;jPUNvc{7L(vjg$!QHz@K@!UnyZh-&#<7+*<(p+}7@T{|o
z6ILcJeYk95CJU<N;!aO1MFe>(#s7zn-%CdWu;HSv{g{dNaTu^2OTpM6Eow18AC`_h
z3`mZDG}MvD{g0aBWBa!TSbM=DXN2#iUkt+y#+Z8}vciKbF<<?6D~JwE4J#LJS_tui
zqT=n`0)y<b+m#Y$#LGVeKd^f5(R3VOTwV>cz4>s>FCg8G-IO3yg}Ww=`+Ru?qwm+%
z{FJxC{VgIl>25eg#}o?;20+zWiXilUA00;qJZLB%{J&>OBg&5M&nrc%XtsfU{e8#X
zZeTYa=_l>q_h}T&Agq&pBe0d+em<nMimXZFy14EfFFfz{n7D6>40rBL2_t$!hj)q@
zILbFM*@J_5!7;T;C#QMV0d5@DdM++azAQO7oCUNCQc=U1cd#G6u(~5hc;lJ2gai|P
zTvU74>E*jPhAfyRQ{<^{5!)zVk0>h6OTNW$r6&%xF_0lJ1F4UmQxA<q)v34Ml09sF
zoWvkwa9!P?yl(?{d`&hkwANo9+FRU;J<F5oG<*S-ft``dM8sB4W98Mp{5>4jn?iag
zO_e2RkZ)-f73C<XC)d2bS|tYO`=MuzbxI;GF=`vH9qFmrTdZG<yp6va-yUu`V?aH+
zlkcS!Mac~51$8Fmo4)?cpPg+jw`O=~IZ-LGUt4M!LT~vzWk!t*e+T<g*iAK{Pm$yL
z&<~-IsGQLD5xaZ8C#w*V@xZ63sKXt7%E(-#pg##iVn@%&ck}DKsOHt`++mus(_|6e
zdgGdh%SH6#_asMot<$32rt(M@JlT4+QbqLRvO=L61jkFfsO~5C;4ZDqG6Tl5h3fj>
z*3(d$37(qQXF^&L^M}r}Uz*gY9^@R}z`SvQil}I}jX#@>7H>0+CxN#%xNQ&TVFnFx
zGfJhOx2j3XqFb}WL0HQ?-{dwsxHZH(J+|lQ<GL@0T`}2P6Z+5UC;e$Y+eW?)H({6a
z5U7}$iL>~u`3A{+kM(s<{zTcx%IEZI=(Hn*HQN`Sl&fUm%b_wbyHy`TP|#!!_uhZ<
zo8Phk>?L1Yea~0@HQ22)!uSk&)euczcO@Q4IM++^3EhI!AI$DbuLri^!3rnOVgm#J
zH2*W&+56<N)p=00TYujev8;%L#I|o0loPLfgJ{vbbafw<tp2(=@azC(GJSDMOXaqd
z>{bJwlegWPvii08n~Z>dG59gSw0t+gLjOb&=lZPRr=-Mz>jLp~oqHAcK2FDUOH|pL
zQEEMrYaCM&!Nabxe)5SL^X{AI^Tqvy!~98p6Gs1lR(+z^ZYN*2P0#YsUqT#PR%8e>
zZ{*7@Y;PZu9C^7kTI3$GAzyAV>&$Ex1DSLhk$i2KR89le-+1{vs8lgT<;d~`b>07^
zw|<~~O7zll?0(((+$HzpJ$#y{dV^a(%(Qpc*sREjGNH>KI-6#`8M*5Zk>^XfMvFa!
zgnvSo2yqjyKeSw$$2lg-!y$3sf@B(jk+daruPGkyOuuv|tk%`8=2E)iPiQH#*O~X2
zcrKM=9!XlL@Ja)8Xmy_4I_vAZs`~3$a%D$J0tXvGG}bGKVm{kTMme1ZnSU^@s(E=;
zt3G@Xj#=DcpYIbD)-HaXn5tAjg6*zLHYvNF-H_REF)Pkzzs!0Gxg2_4f}i2iNg^up
zf`S{_VatM|kf5cT7Ehh)_VektYHqKb(9&N|^(32c!Y`%6)=a=|99}!sZ|QEmeBV8I
zK)Xg6vGs?P4<#iJds4Or(<?srU2H!rbHBXBKqEU`KN!b-Pcn9>xVWBb!g_2N?&3AY
zhkftPcVTYDb%;0kW1l!pLRzcTOqPNMgGGDb^UMsYkr|b^A7p{#(l1zxjHPiKs$-=N
zRym*s1k?K8Q|VOkusB5r6cWGC>k;c04><AR;z!b(UTm#bKeZjC&5B4S6VdMdOjUTp
z{9TX9avSwrao?Bwr%F;7Bul|7v8$PP7Y_&C?{dqjKtdj;*mz}~y)Ckee@9@N*?oTA
zv4&waO`$-YT)g>cGO<Xu7i%cvOX|~nI+#9KsUNUzYf=4GL-W1`nqfpsA&USC4{UG^
z?W1uaay}c))vDTYvLMNj5!=3nLyE)wcbLMCYf&3l7{Q-xaypV#2>w?~FavE)Sd9(W
z@ut2q9~ZRzKiLmfKYtQx$YSWFVbe3&4nbBh>;AzYf;=zYK9NHVRsE*PV@=jzajxRM
z%ztrqp1mvVz}V^J<JX-fQK7TSq+qq5OxS(WYvf`#E>V2d;Klphngoa7U-`C13Z<Q&
z&0E{W>osnJB>AI2|90AoFUuzZyE`&Ix^OsXVCcI({IEVvk0`s~Qr^teva55j%&}9Q
z3}I(akv}<!R8Ps`vLdBqkDV+WDzGUWb5nFGzb@$Zf`$(j#L`h)wUWJ4LV@WXe@_U5
zy=@XazZ+a--}XjkAjF!Vc~-6D(8=J!p$>c6POjGB3)Gv$1s4%e7!|vUQnwq=V@o5q
zYJSgk=01i?PCO|Q(bU9Ee>!PXUF%!P=(3v1w<x`uB6*92_t8_%o`m97^Pcjvf6(Iw
z1k^(mchbIM=n1b^+vy><owKNQ>t$D7*j*PJxLR*1h!8y>&T5sO*5@FzE|sybFWhX{
zQb_^zFVSyNn`?wrYFqA-17F!DnCE*Y3Bh4p6(qj79&8Qyao_*F{f;Nvq+Vx<=1Dzx
zaOU1B55MSU6`J0{nXtR+t#RHSWiufSToKzffARi<n$y69>2gfr*+me|FmG+LSdGW`
zpPNTGGv$j6N@pe~p@@vuD=r<as;Xg#2+RfaAD25=oU#@ViE@kez2%kW=i`wsK?`D6
zuk-OJS@$hw3>3`Qie7ZO^djul{<vD7N#g%$p`Nzpb+X<0*}Bgl!+Bhz4LXaY?vf@!
zU@*tE(1o2xWt&r5d3g8V`zzjdK=4K7>8xuOrd-8HUE1a$T(|o}p<RPKkIT1uUZ<wj
z1-@{svr`thh_tmbzO~BP1DOGa!}ay_)@zM`Y6DA$v-nuoi}xc}7a{Q<Mh7=-joCzd
zua7pePLgU4Pw~aqb|Wg(4F5gB9yas>bl2c=tbU%$c5QC&L*G7%6waZTsJi$%Tj@^@
za_MQj4m*dQ2T;s|JdJmB8teYs<#v%ktiYqnCQxbKLs&{WSIZr4I}@2kVE%)UOusmf
z_Qu~gNyST>D|{5&$&fv0>GBzyN3RHn+RIDVV48YVBeylo+IY{-G*L9yJ6nY-BEkxo
zF_*LEWo}<KK0OjHdt0TlgZHbh`Dv^8CW-Li(_kp4P1cl)9vNbjBWeD_d8h3G-yPiC
zA@54Ttn&t(h)Ow?y`LzayfDjTA2IjEb@IHh$&2lE|5bhuu@k}liv5R`x^`+8y;V6l
zT~7KG?s|Uoaajo{2`ZErI5O4=x<(dqzO9-5tzQ8ff)myfMI|KUcXhqaH&)er<9qeO
z-U*U%hfh1JRXiLeAtY3NLEoviLNjL^m(3u9>wkkSDcx{qT!>pfSv7$%lnqY9+M*?$
zJepH`fnBSgW(STO^L7uS&-EL{N(_LQ9?5*@ks!+JTO-YpIwjdX1xhfg#)uFNMa4(Q
zSjnPPer}-SQonq;52Udh*GP$0@UVXfLlT>Y@g2Rr8bJER#@Cl9o>rJv0^!UpP(0ue
zCeqXMfPg6TsjmWGUf=3LlmLcnjSYW~<Z5TI>=mqLdV&5sVsi6Hl;q;#Vu|Md8wXm=
z_RGl!T0kjSnIXo+|G2ZaH`Z5#JnY!I4gA*7kaBTx;;t8ql>!b1(brGj+v|aRV{q=?
zji@883o4u3+{>Rc*MrTxgWj>WnU%KB&U(3}ge-J4U)V<6YBq~Bz&h@vA?HKTZwk7m
zA+$15!c)E&l3Mv%I)B(tHPYpme6j94b}!&^dl4IAvv;RHsdTFEu^4mL+i`heV2?Z<
zH~IcDK!N98^S<|-3kJf3l;|2_$Jm~8OYXf!QBy9UP#XDXH}G^Z*2JzjZ|5~O7U_XS
z9kVsfq`@%4FDomacCbOIH=Ko}{rNNR$^Y4%HCk-_RBT=H{->A1E~=icVPfMY`9Jq-
z%f)06ygz37)QE8ZU4bs7Se%M4^hu#?`qgoo5mvYeDy;%CNc^E-usNRJ5`pWoEhnJv
zBKq&Qhn^AEJ|$WmoX4K#`=d?O?(nDlLkQ>Edus}ObBmt~%ZQ8$niHB2H*cwHy#Bu{
z{y|?6zuSiiWGf00++k~LRlC};G5T94z%N{n?L`)yJAeCulj!NXK{`E9p@5qS_Q#^H
z^{dE-XS>{7^LtdIX8=@=<lGN&eUPqox4ieNZe4z>k}3}w9+snEV`Nu71>)ysW~-B6
zUqQb@gT-;X3UJf<*C{fdODps0;{WP>z}z2zJ_#Pi!5i^MC67$mYpi{!+CS*7#{`h-
z9e1n2U`arhT(u5-q8!R?iu$+tIXKMj@2uYN+>;}9E|fYB9}Mo+L0Vkix^$dN4NDZx
zZ1jWiS&b>yO(KWC=X>Sq*szfW8oHSLd$IdJ;DO{=<D2MH?g9_i3|;;e4}_t+QLMwh
zPflGrhi}TRVoxDCs!kt%iginy>Nc^XsgXIGIHv-FP_?1{;>Hbj`I~G+$wG88*mv<O
z1n(9sxe7<8039K2fBz*CVlyzmbDxMCoN8%d+m0#nQ=nLnOnX$a!GWvvGKXnt?DClT
zM2Uj`Tmz_qbz6rD3pSVTZjK#^f$=I5_a<><2d38l{m4Vh_AW5{!^Lp%wWan*_133E
zPvlgps;YosQ52Ak9(H*1&ENk95izkW!r$NDq;yx{i2=})=0qzRmHlm4O>}H5lg!@n
zaqP5Rh2Vn%&<Jp4@!Z@Sbhl7yGIMe&*j4Bt2p>K)av;~_5D?J7z`)o$MlB*aorIwz
zN8u!d>bjWD)>ujN+8tOP+4T*zr_Z0icio!GO%id_rovMD#VaS4^9<Z1%TiXk&{Oi3
zH6fhD-FC+|%hFSpviZqLGroH-Os3ZSsY9)~pX!B6lUeC-HA=G7th*o#mP$@$*3Why
zc2Y^A@BS==m;*uI9mgY`B<%7-@WFFfzU@q{*71y2F}>xlj}JUo;><zkr}C4Ec~=e$
z3@o8g*+Al5lj`<UjwfD6Hg!NH9+5#};H?yJ8^2ue0nFg1A#>ywCLicf9Lp%f_$*78
zJFpyFWfc{dwILQ6<<1xmBJ`Wn!qhw4+xvmi7psib*!OSoyq4&HL_TwL1XcgoV0jBW
z|9Rs=2}3sar%SsmwLOqg9?X&>dhoyye69Yy|NoTOY6)NQ!XgVWS)izfVGLL!YraPD
zNP>O6*4EctmFzsZ1b)7#J5Oz$fVen^UbX6{PoIQ64{{z?Ej>J39h4r`b4voE?uB8u
zQ*Q|-$%j%3?fJLNICgT73)t!MG2Kwq`I?pc>(_Ii^UaEhI<E>xM@K>+31v53ZMieo
z6#VH^i7+jU=2;U?qd63p#S+`;GQ$P)4!4XyTL|Igb=#UU8qLniVign9-`(BSu6+CL
z;?jc+>%@P*&VYHK+GX8J{zhu~Tj3<ng-=XT2!!&++0c+`W>(f(esTSP&+&A2Q^0MV
zsQZ4L^D1{?DQ?%7$LlQ=_GQAi{IpG){ITK%?2|7pFAx8uhp$0-`9;c#ieIpCa9mDy
z3_0{`$JRc=tX={Ka<!jw9p!NjI+D+FfSxSyTajMP$Qy2$`q{ytLNo%A2AmK4a7e{!
zv?vMO&Sj;G3|y`WntV1FfbYB<t&yjcE~VazrPx|>ixZXtPTNzyGuIGCBVOdOzaoQ}
z@!E#}hJOEM5fs!6G-Q7C=xKL?z-Y?V0aMrv)*~6&moLB2Prq%({V};Y5{RK_9P?O@
zRVJR>G+5YWEyK?b23mA>xNf~MR)UtdqG#da>hJgp^S)UK0>nlkVs_0(s~g(!ln$7w
zqFI(!^FKczepS4K(HO;NQQ`qVYuXu;8MX?ImuAzmh=|m*IAZMR5u&9#8@S|D*fffw
z==I#DU5K+W6ek&&n82a6%-r0{poJB}cz)zYq=D6L+mB@?D(nq%je+PKy$qGGb54ma
z%$sDE<L2Cc45vX6n5Npfxw%$-sqet|imi+5)WMg4k5g~OR1Dze;ZajmgdWLQ4dq0*
zZqF2A;SgI)mfK?E;;Nx~lK7ne^l^ueV9e>%pRKXU4A@pow|L!ZY<-<5s3apJ<GEf~
zaj;R^%gWEM4pec$S^Z~c>%~ahNq(f&a7U>sCQOy7d5Lfr<~@WjevvwA0M&I(inwm%
zpi-dZ!vh%259rz#ez5~J2L}fa{OL~zN+$=(t<zu<$3HQY=nlkG%on7@3#Eo@BM#b`
zQgZa`Jc+@~9+XcK&J=Q5`E$PA!0GB^VUe>{xrm2LEvgAxL7a&>_X;p4e!bXt@Ae1a
zi{+q25}yU!0udu#kR(_|@nSyJzGtgUb-OJTXPpLDMnL`;#z!wMZh&y!TauAW5!0iS
z`rNl!F<bQg`;%o9Y8^ah^fMVx+1q&um#rx+y&6}fAQvpZz@R}NF@&_%=+JAFMo)e4
zh8zw)Kq63CSve{xsSr#JoiF$NioeZa&6Ng^bPH>MwgO4|V)a7p)pxh4`@lMiiim)k
zKE=Rjl$@hNCpXo5li@DC@MCv79Bk}?!%>4`Ao}ccG9MJrV;(l+bF7!2cH_IIw|8wu
zsq_Kn4!kp#6WuiQ($+BWQ3INiQ4I|ZU~)Zt`0yYcx{zESDIDyI;~7qUhMB*3BWF>t
zog5aGW;vLZe8YXTNDn=?AQd1U36BlJrBGk{ih+6SB?6C9@KN@l7F`u==&_}xrL2Mi
z@Cm4!vVQm-ot=Z=dGs>Cq^W?JLUjNBqvj`s<m7n?cGWU&vpyAjdwaa*J(MAXH+HbV
zTviNBi)2m`^%z*7hgo%08Km2ceK*WCmXMT$$A-wa-D<30hYad<5xmZS6h}sHdiTA<
zyiG477tgB%PEX7=K0DfETix&w7d(96w<gtu12Y|~uuq}ijb?2Ox_RqXo=hYoB4aRH
zAy&8r=$=vFH&#^YK1Q1W1g}NkqmV(^cZq%h_j}FEPOtlI?J8=dxG{EY(aj3}{P~{k
zRHYJPs2sA@0bk6Kd9iLuOeEM);dy8!qnxi^xU}87dJkq$3MMr0o9IXVHJf-a9Kxg&
zIzB#Rm-roUY_}9*sMcfOG2{&_2OHekTN$Wz1NqGK%kyJ4aq(FvAAn0o3U#ujsXd+}
z{+c9*#TND7H{ZPTy*y(9>W`51NI^fi1%UG4UHgT>qX!DKO6haQ?vBAu!Z_<sBoMrg
zOD{trVc%s{Rim`aEdG%}+ubs^ZGE8RA!9DC;0?@tJlw0CUY`v%bg$WjixejI=U{aZ
z$W3}z8Bdhk=Ev%Leky*0DfVY)z8P3B*^*__9v4)S+ul0}cjsALTpVy5iS#VM;5Rij
zfo1k48g%RWw;*cpK4x*oU@m8~sARpp$?0b$jrJO*CMFis)h-j24(Ugmli9VkqI_0E
zFM)CC^TWV&{!OH@V|H)H5Fik3AdXmnedPr{dimhNgSC;ugg|_nBD3yzEh>CJEMg8x
z?F_9_vt0Yxde4j9*1|B%MkAnogZ74bYWcvo&;~Vs`}4=bWn)Z9VCrpvc7>g-%X65R
z|NqaET1P5jIzfuhZsSu&P1}LVIttoU7`C`yZZp1c>XVU~sYQiW^~iaEHm|9=B5qql
zuQ=;6Dk}x~Y{nG98-t-$nyvRa+-b&V1ATBCkMfIUyt;IuLzZO+P~lv4R(B68C}3A$
z5%SnuVttLe1lq3fb4DSd)7r(}OP`a4{Re2|Ti-sr2L6}o+R@Q5&=x_5v>E^LumbcI
zGaa2IdYo&v>)B-D#i9C^&txitT<h=C{Ku9ZwGuO2Sy*jC9-}!G=1=H1i&9inOr?f<
zT7XrmF8!DLhzj`o!A{d{8uyNA;C||kCalT1jNXHxV~4;W;K)H=(Mg9{oF8p!*SePg
zsK+k<H(IC<`uEzE1bffNObiUU#;sw$?F8;g8X6h`$EyP1jmv(wz^L<c&;pnr+DyPs
zl79Y_pb~uh6fsoib^LJI&J>tqA0VY#eK?{+L_m-Zupuz_>iD8Nen3|h>YR23w@D_j
z%goG_LAY*CWP%y<XRnhJZTCD6*D~E^Jfww$gizgxxVYTRSFu(dQ7qMGYxzV(M49M&
z*-dK`Fv+V*NcfKyo@)=E3vh7ADTrT|fgvvhpQDX0XT!x0`4lmqXCAKy(O;AbCC|*s
zVID3vFo2o^aaDk6G_RXqFK*Cxt<g)D3f;4D0}yO{(-mnyJL9AW`UQR(?3)Z6z`j||
zhB5kOtwh3rFc0kan9cR&x&50rZ=$<v+;@#a$@ygQg3`wh+uPe^<>g0!7byNF8H9#J
zV7@Sviyyz&0Z<lT(}B9<86BH0C=i#T%&_?u{CuNSyT;YF#D7N@3kwVQ-e<WLo8zU@
zUPl}Ki9$|8rREAgps%s<@r&Ce!68&tklf%9g5edfN!u4A<8u4iqRq*Q^=J+K{w%qK
zApm)}_W^ueAJMTN2Hy(SYq032<<KsB2^`Vz1k!rhSs46P(}XlNZFH4lXhEZZr5kfW
ziLmII(x7M594_#PFc_h6>G)9QD-xN3cs^@q00U)vg@CEXr4lZCetO~s5cNRgJIq&v
z9t{9SZ$~q`^`Zfl%jSgI>gp;{hmHy8;I-KXpzwgG$UEMartwitBZbE{&nsSEUA~Hf
ze!O@+I0VegtEd{(#1GSuKs>6v+S*#jQdQOeYX=9jI|l6OU0~AEP>#UP9V|rxa&e^_
z#c}^tWK2xfCjur}1UauoYPi_h%OHH3n8DR3;1t0oUx06T-!#Hx(Acx=O_RSPDpA-a
z56}d>0L&+e`Doi!KsrD>ccRQP(p=^)I3^#gw3;p<bQ_@S^H~7Ozc@NM!KFW%fg?4*
zi&U?~Thk|T&}tNDI`oIvWU01Z`GA;#kea#(s4w?kxd<msRJ-H@tzHHh>2T^{;M+9T
zj_%&Qt6lG1l~KyHDsp_R3YKV!@0kq1E&ahn5L+35*AFKRJXbRT5iEXvQHrWgfv&W&
zEQ!zKgk7wVaCF|F(fTX+Z>IY}zqFhTgRha1QP=KbYXS{)Q4$vu<2vKP`1<P`S#CO7
z+AfuB`Dme@Wk*mHh0DzlXFE~G;@}0qMF}O4ljd|u1`%B;_)f9|uzR&T(Cy0_QLs??
z8R9CY9AD)cA8pTWSEblamaF~o?MW0`j9Ucey!y{(zVY?MrYjE}UHXtm7^Tnx*>NK#
zew1*QMMHFj0TBNnBOyVNX-tUCR5@j%jzH`db2^r{$OT#roAIYZk6`YM=Q;-i1A`@)
z+rNtx6ck3Ms+=G`yD33{)1NyyILOM#u(;rXQK)i4$s+W<Hj&(;b<sVKad)SjdS!%b
z>}))kWNXg1YC1(vb^0Hqzo<W5N#I7Ygm;?_O5(%mWwgs}v_7ol*bO70Z26=-V64lA
zCPBb6T;pahGXVVQtEuG3$VDMmx2~^TKwerU&I1B2Ma?X*Bd_c4vaY?0#R-h*AXrDa
z%bl@oY;3Q<=y`4;M|o|=o_q)l^lT~6uhT9mDanjU1rt2^P=CAO+82DlDkwO%Pe4kV
z)7bbFSPt;__wU~x<#68%_qeK-a2jZ4>IJ|3jP6P$FrupfWA=ebKb!`=z`?qG+v~^J
zrt7o^Dhbh3G^!1Z_?D0Mx9X$q?N^tx>l4U))D^T1zMkB<*b>U-mJCFnKfHe*C6blK
zx=)JkVfzIT$b-bq`G*f5at#`M*G7v|t}amGoxF0H85w^HD`p1JxL1ap+vH;yl}Ogd
zk9T@qQ8DK7*^r8S_P+miX|2TJkIu(zZaN9wSpaFWFfhFEI^N=N=>na&B$Re20pi7R
zKps;Y5ad~32L=Vj*4TpZ3oYy&UAZ~6Pzge#<sRd|foy6hg+Po*5$*pP_wxbppL1y-
z?g0%4X3y(va`uXG)kf)P+#E_FqR`M=jx!{UCg1OurqNTa(gBn-)M*rN7CIeEw2!7+
zk+|ql$%(p4!07(?`eZTuw_QkB*pcql)%i9M%fPz+)zT8<kpd<(yFes}x{~2u^fJKM
zF+1q=PgFm9_UsIJXe0<6OifMEa<`kVUfk@ul{@}N?;thsI_gpYA))q8eH9R8V05h7
z(Vp`J8wkz96j?x*CU>H3x-E#V4|T6V5ShOrCVsjU=?Kt=PYuekUpxN=IywZdUS<Gb
zGG`EH=YswKE;6#)jLmh)+=ZP=G5YU=zyQKyQN1Y!TVr6=#dtwKDx8Q4`Ow00se=DL
z5bdo{jum#(!PCTqgc%WZ(y|D#lR3;n{km$|EvJzJEnvSjaw`D7v>M2eNd|Vby}U;+
zyWE0mWPlO?Ci~i5DTYY(;<br}=m*5aSs<8sg#hx}#+SgYW`;S{OjmE$pLZm!PZ4~F
zsNkMuYZB;QMsxpQ?te4yP4)y*<VXN%GqbbVI5~Su%kwPb0XR`MGD-`)N0tVVmj~GX
zYV`b5mA=oRTJJJ|L~_ZZTEOW668Bhl+p)^|?7)(Nk<kj6Rua_$K<8)yo0p5aqZr$n
ziMD>=<<*sy5%iGdT;n?i`2&}BhFJH<vo520XbF0>(GB1H^@#utPXQ41Qexs@dsXAQ
zX;Lw1S~1xPFVT9T299L;tfBAm+o{TWv<QukP{&KIHAt^xJ6v+!mu%_<%x=kTZ7+M5
zLEy3q_^n|A^jg_5`ith?gE_iZ(n<}f=;|f{C{zHTLhl|He}dfn{QO#fdQc3z*24Ho
zve#BWk%q34QPE0w!g@2l_`${Tq<uqSx$Pt?GxN(p!?q{)PS4NTXlW(jo(FxA3Szm<
zLHMx=_Ipe1;0t|JZd2btK-fzO2&VzynCWFexOs*Wzt%F+shQpqa)b`RTLy=OsJwjn
z5moZy#S2R?q^#`hL!iNIKhA>z){h!otxf%LGVL~d7}bP2XHlj)ulP|^cchI=`s67B
z?M_OJTRnXhz@4oIvqII2baTOY9Bz!OAWTZ5y>-YzpamAaEFy+OCku24S>XKmZGe!R
zkz#mg=CzSRapwliPu2^7G(M-5E^hmV^`e?Gqu<yso<CP@zeJC$eZ!?|T*YX<h60G;
zt-ut_t!MzoH+kVZFIZk^pLccADy}ZP{+pFGp!e9RC~u}A+7z(XkM*>TJU%-CAf*pr
z-aaK6H8r*VTBTF?eNN>veBH;8#_-?7c<om-p9Oom83|%wAZ^dg#ifJ}Lwk}$z_nsY
zLQYZun}JYO6<Bs&_nkMmWZcieI16qkV&2A6m68fXzc+NbZU9hJAv(AKc397`onfu-
zTL3x^xw<^he--<9xd*h0+GFX#;rgfok9iL?6l>s{3uNf?z=z~KW}mm}j`fE@9MQEI
z5D<Vg?Ih;V{QeYtaR>l4RZUGapMa(c;{_ekKm?3*13EY}ZtA%mmpViMIG5Rsa~JAX
zDR0*uC!-q#oGto6z^Dv=<uV3<3?=lBzKxAd?*0K7A7EjjVGwX3y-yth*2|WS=>ano
zwuarzE^GsqWGeAwtw$+}-Yy7_3e5*{$iMFZYd;qJFv*~yx?S}uV+aA-TFnG8pw)mB
z7lF~?GyHiIoSC==bf2DpfAb?rM?=-8(K;nYAFi&B8`i-qM>NlaScO{-QW+l~2b7g!
zYZw(0;82+X;Ae=)$SfU<z@?2MXa)=h%hPprC1H(6z+f0uGS8o>mM43zv7kA$<*@zz
zeRObd#Y}{L{4t32!2j7F0B!-`(R>j9{u_tX5=(<g221S#eClfo0lUB5u9mOPYY8|b
zVW*YPXg3W=DKv2d1kjb<yh$BaLcdtC+1D&fZ7Q^ZLjQTYtC@xo*^Q|VC|611vuoP-
WTz2kxvYY77WM0TWFOqoU|NjBcN-xI%

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 eaee34f8a40cb7d7b3a8bd23753d28c45e397eae..5ed37234d0179ff282378152d303b17bdd315e47 100644
GIT binary patch
delta 75313
zcmX`TbyQVt*FB8Vt#pGT-Q6IdAl=>F-5_ilk&^BbkS^(#?(RBtcQ<^S`+0xg7!3Kt
z!#V8hS~1sLbMm_r{zn&l1$7g^`soV>xhq1bLaJt7ku&vARjV2ftqt))6^<g6qUyXU
zk{YvV$cHbO7+$6)=}Tu18}AhH-GWv<+H`NWHX%FwJ6<CVYL0C?JS#g|k%t;Gjv|*z
z<*7+2N^KJGo+wH3+xzP{Vx_81+o!6vr768zVZ91M4#jKaB=L{tvqwe%Zwnh1=44D%
zoit_Ve9c=0w{l@a67G~;lGPliP1ReTfwqVjSsZtb-PZJa?LO7@!<;)W;dd~??{r<C
z?ym+j)eq+?jlNR4pDZ;5)YZ8~KNTp`mg%-2olfN%GBQTJJoA0OWgbrDnx&bc5^oaw
z@&)fv)<d-WPESlsOu0xI2$hVd9WPX(+8)guZT7gpV>3ko-#5qOV!N9}E87{5zn#yh
zC;ZW1C56(;MML~uU}BVJsp-K0Pl+ynbn_CCu_O7?ljhp02E3D%nIl`QL?jZ`{(vxe
z6vN}*R(t2q_V+%5QsKIui0QnyxM6=#zo%`M8^Zd`tePXuOa1@`ggpd<e#Lqr@-!r7
zB`n__9+*Ul_r|mA6^1@?H;|e|Stn*xa0WwB*>gd&k$)Qs8r<vBZR%d2h@EeErV5t1
zkNCH(Q?n2$nP)uRfoQG410OGm*yCuf&qNVp9sWV(4HsKKf{f`)f*w!36YR-%{%De?
z-nMfWt`i7?S;s$MJ8H$Lwnu&iZE2`onUdnUV?usuXtSfx8^!!&1YXo^?5VR{^03KP
zUdvWY*H=kmY+2`qOpZd}usmIC_yuxL*fvZo>uTD_?pPm;s$3H@fM86$=7Zz0qr~HF
zj+~EflPxp*{QZ1dM0)u08-K#@z`2K^)KkH=58`=Lhca0}D|vm2UnL5L;|KbPhq|(g
z@6YD0C3k-BCsh@4bgM8P$}s0tB~LSC{UcK-j*(S5c#1IR5}+3oB#u({Sh5#8kERf<
zi@iww$q@WcV-%=K(6jo6hUot;7pWARkN^G}T^<q~yv;OQtXBTdDkCG~ZqRC`WSqxg
ze{C(g?XeRb5ISv#C&hFARGz&~AF9o0w%HdQj?V_?%b-=uY}oZfaXvwwJ^KB7m~3hJ
z!noHP32}Q0jjqJ6;q&==bZx|3<<qF|r7O#;_N?)3w=CIOiz7-)jfvIvSapcuGOa?4
znvL^?B`hAfM^dQTA-EMg_scFoGi6WxCShgkGMfQdYI#4h8T2r)^kXUso-3L4(uKxj
z(-Q89?lUqKGwmjNJ^XydI*G+g*TqqR?f2VmYsjI1WX{p%w(;3f(s5wg9CU)%sVuDf
zpiBz)e%l0g1q)L5A8OSX{%9>VQG6a7Y_T5CaP9AY=|Zhxd*mx$@|1l;8k{3_o#H6P
z9?JOujniHoWpPZ627UUva$V!D>gn$5m3kTU+LAPAt>4+jsd3>`sv?Y!m{FCXG^mmV
z=|!r=8rM(XYRfa2jzn?QcDLTh{*mJvQE7*8t9wL&H{)=yK+&pO0Z-WUm&2f^v@h)Y
zg@<^|dQL7f4m7O@c}Rzp?owZ%iCd_UcVZ|Ipvh)EVOPDc{b)=ZvDiEekqI+!!m4Ic
zyKP{SP(~w?>Kf`mdJ!wfUC8x`jZ8|>_~k|YM(`2_msk>wW=b*b^Tv#wRa+}~{bbF#
zf8;VV+KAr0lQ(Y4lS*WD-0TyvTA{^45`4bhc+AuTT2@b-ot&H?9y`P{<$9y*-C@Aq
zbTRI5O8j@0ySAaBq22i^Dnd?cGH&i9J)gTE;Q%;`WjO^D?47%d9SIqkysu$6Un#RP
zrT>VF3wie@$=?rj$)}6wxXDQC4l~93ll>cjHR5SLkSihL-{U{#Muj}rj%_mfh9+~o
z#>=ck4QDbf6cibh#1~=Zok}LXSYV6|UY)r?&yPT?BkZ8>SVhrsb_jt|HK?2xH)9Ew
zMq*|LA7?f=_?o{Sv;Q($CvW_x$rn{qg2@oK=$Iy9v{NZ!^CN9{N8H@Sn_ksmL>2ke
z0d+xuz4yA5&VkReCM9SRPO@J|4*9mQHg1fLz5#gxVciFNHAHjOvM02S3#d5s8ph~n
zO5J^!O=c-dja#xgh7p>LerBL;R~r}cw2M50HNqFaE-@HhFArN5o7NPbvRWef?WBL?
zeH5S!9cCvQ`BPRP#PC4AY;Jo|bvQPriL>X_#^rXX=@fkOLTyPXrZVX1(gqTep~G^|
z-^%8K@5L>oWWx;>PPuY70ox|WiV9A4#*Yoit44@Q#qDO00{QdFxHsCo4|0`@1(Ox1
zJ}#*pNA$N+AxxB(`r{1p1s+v51s|nK)E9YAd%QA5Gmv~*8A|g@BNXg~QWsD)E!cTA
z{_7CN!otEs$!sAx;!zmb*k4^;U7y{AYBXguH3`f{GqxhQAo{@GMBd$j%fU?PoDcZV
zU<!v-Utb?oZc0reqt0)Kjh-|<H#A>Pn}yY_tIO5slhy}DpaE<S;dm_mKYk!MIy#<b
zs?o-OqM~xTKGZNBPU&81aLU{_+a5~38w`L)-wr?*Ts_|!yz^iMKiJZewl|7&Cnrj9
zAWB6k_y-#Yz!9%nsySX^AnJNJC;a{U_vf?G{{9c_>~Xa=q$>)_b+$UChFB;Y4OH&2
zF(k8f#fo_YPT^nP1@3IE?>)VidnIH(-f%(RShaCA%}isZk&s9uG{d9K?oepL`Ip&=
z7&VN6>iS*2NA@eDU;X-u*u!>uwVc5;(^-)Wb>+dpd2f_nf^C+jI;Y(yH@iTIx_h}Y
z+K7_rFUV9$^03AgtEq23b>U=U@mE_>h`W}*HB(Abg68Kax!5;CT$zT_!gMXuI;{`v
zqBEYzFU>PHHG<zuO0~M(QD|~J&0|lV4PN=<_~L3@IsAZ{Nggk#?tqRaq@&mT&2m-F
zf><0EK==I4o;|nCMb7js^9?4?&?z*4YQ@8s@~vOrPKpth)OhmTjw7lvjv?)y1+nrb
zE1oU<IzDD1Wp^DZ(<?`jCYkM7fEBq={ON}@<wlOe1q&xjP@Y?^B5H~vbQ>giQlj4T
zHjaD0)vK*?A|0+5)8J0$D5e?wK2IDjGrS83IqDWl4}1j62ZBj|GMm{e007}a{`b3c
zb#-;J#KGu(<Gw9NYa1Ix6cpo~k@UL*6g*~M0uD>8`JEr{$@_+f^VG_9)wQ(9WBZvo
zIseF}^TNQxLo1ffn+zskH?}4w;tnP<jlOt1Ki=Km-5J>0MoLo-o{YA)`~Ca(8R&^5
zevOli{odKxy3ZmmSNqdb1q#qgN|TRIRaI3RHM6=KH!HyG^%y?Atgq9~$atO<?m#?U
zBJemoJY1+%$Cxjd3HPt3=TBExqdf1W<f(3xYXsQtaNPNMr1ZXg=~5~e^8E7Ui#-UG
z-c}tvxy?0^AutKmNOqG7NnQbTgg1BiTC+u(rFOPTkvN$pZ}g91`ZtYnb<8BFA7-A<
zIXoKX$0mshEH$KN)!|->-zM}NE;a`i@To~rePdW^@<?v0dt#)>>Q$lfoU&CQ;Q8$4
z=2*q1QRgffJ$&xyr2cgAWiO<h)=Ee?t_eRpj#5PZZj!su^@MKyv`h#H4Qqu)qnxOm
z<W!rPK;&)6QZ2so>3HboEZx-#WBT*>;feW)wxpRVVx3+#7-L%}B>sk(gU#%5sgxo?
zR=q*Lg#Bzz96KwehSQ5Y;x<DaJKXrOW}QzlcD-7w4J$BRNaS>!SkWUkT5usVCz5e&
zAbBd6yY~6+vM-rRrEDDVZQlUi#1l|{?YM_m<Oy8QrPc3InR^%1kjJsnwxYTY6ROI2
z9iG{Q&4w(c?nK(!L>Vtb6Dh(_^F)a;bvoa*a$2^p_sIJn{y-w&bETI`pr@%F1|uU$
z2=Iw6U!0q>KVIUYRi^f2okSOU{MGjIOb8DTpKb(;$K&`30O1HtP1C|vM3qLBDZOqp
zo|M#};bKF5{W}Z{MMJ)9=iT_#QiU8im10$fQ+WZF*vibzmDei;V`;oZgRRyM;8!%?
z9m~ECBuD5#re0rBS_(&Kuk;s!psq|8qT{agD3{De<la)tk}epoA_SRHun7cpHa)}8
z8M&7?1GedNsh`oglxZm*o@t+>8qEw&w)JwqXvw)%OjaH~b+RJEyFCi4nrddt_(vU)
z3EoG>VU-uJnGe5M9wPY6fbd0ZuA#H>>g=uN%VKB0&dSZ$cjqhM>I?=Gf0kTCk{*2`
z*G8^qXn`zZQ^x}xPCHO$;^_J=hdbV~KI20KaCt=d)@!4}&2b|&^QFDc^URI=%Sw}M
zUw@0D&TN@&<55&TkF)a!KDKnxCe{6!kT*w=3VF@lBScf)=e>otw|ypu^S4f+xXhYv
zd%ZhK&;RD;V0>DjN%i6^sl>PV1;vp#zL@?yam#aZ=6+)~izKb*8TeC>g9E8hak^3Z
z0oZyMxi!$~=f3MDOfc-lAnJ+kTipUbuh_C2#rW@UQ&LhAi{rgzireGoI-?J-Re~w$
z^3oYh@@{`v<XsRvfC<|Pfsiz4M>aI@a9GW3KHaRCZ4JbKqM%@L+LjIv|M}7+OBs9C
zjjIP<X~$}NfP<S`Z(rXJUtg&0At1T)>EVWhmSgOX4B_vbocS6nT<?eLJjbm8JbZiu
z3yV;&ex&ocZ8x2?Zh*b)`-9E6n|R9&#Ou?iPp1zjZId-tYH8`|EUc^@^Yh8IHjAuo
z$J(}#=Fs?f+l!H1%>3x+_nAKTG>C|ZB|1$Bbr4TvF#0wfc?1OoFZZW`uu`o$#}kOa
zcW-Z(pK@b*$w|h;qoYxJz*9U}WrR-TVT9d#NS4ub8yrZDS(nRsRBV?*_&#WHdl?fB
z^IftkyLYfTKX;MDNy4%_CWtlf7s>FbJ4CUC)Sa#2R-xi%p`8yBS3AtKU=kW3REs-s
z`yuCFWAkRm2-lKBFC+i~bbDz$e<K!9wGJS&$Rc_2-)f>wzSTV+4D@5|&$lNVgZ<^-
z;!`j7VzWdliMywo+1=x~;_#-PtrR*rzB;~e%15DS>eznsyPl9s5?)bBm;|zymaz_r
zy3i#OiwzV$LvAN`LLDKzLN6gt#my~8@ONcS5OqS8s>SLVMgc4retFVF_5pGmYWWhC
z!&)xKVQO0o^qE4MMaxaQ376EHMEzT0)kn{WEpRg3BjaC-D`;oU`<8@8&vrTxAFGUb
z_NQ+QuDtgIER5(QR#gQ_gtfT}h@XN-vlZic7mnKgL#F>qN=gurkUmmV|1jtb#;>*m
z004(Uv+FnTHO%?Ab80Fsl9+Fz(n!3;u(Gl;8+duXJXz7(aA|ICX3q!{Tj2(4y~oY6
zo5zjLGMp)5o>V!^Nf0t_h9~(nIJ*K^C*5JVVa=8wWBV(rtGoL95eW$iy*u8)tiif&
z40u2-J!<YvM>Xjkgggc9633o?UkZ13KYl#AJOKKZQM8BM?WtnY4yA{NGDjIilJN_>
zmfAkZq--m8eR4=CGJ_AntaqTu9fK)vk6iB5Eaf6++?^doUhkO>n?N`iP;HE=@afW%
zJPi&ZVqj@UmLJcjjP+R>o~~d^yEkLQl|7;Qa80sTC^5&A*xTKLah2+tE3ug2?yfS)
z9S-!Kz!Hl)3MnoV3~es`aBqFd@8PePP!*${`b*>4@mc0>u_I5}Ru_>5eCJiAQW7Pc
zCz-Ci*w=T?HbZ+y>J>F~iFKGJV1IG>A<Kd)rTQo|HUBtA<uH8-k8_U;a?F*m0obiR
zhAx8-B`m2(q*^<|-Q=!FRlpFT<&qC=c_bE#GK2^4x^(Tn6@K?K5sMVoH$EU=_bEwp
z!N}S)ekJpCT_W5|eW)$1@Hw3)RYV$1;!MGlLiaM-c=yTg<H|~?1A1<B5Sr)A5v|&z
zt1jn_ZJ-qL2pqR~RHTl-Yj2t4@aLi_l-yCX;N-C({Y>%(6e5|vpOW5XxRN9%Q5m=r
zWG_j;C(E<vgnr}OX^WW5cdFD_o7>xl1_lPF{liOis@c81XP0=)kgO-9*GBW&zXKA@
z>%AXpvzxGB)+kIXnEEVhF5l^@-<`mUru9(S+RZi~$0J>O0D^;0EOR@3%*)Nx0}Eaw
zWReD%m;38$JKttE&TonEKPJ6gYV;JU0JO%AEq?~b-1+X)$p7P4VAk%~-2G>lmzT$M
zko-?Es@xm{inEPBLAsazq>Vgnno{KS4R>{y7QM;A%%FhZ3275rce*^G7p&9IB<LIZ
zw$-&l$nE7$I9r}Z#q)u{nDvlY=d6_Evwm0^yTu|VvefVZk^m%AyR;J+=rjTpr=Gca
zn_^S2;3Yg(3wo&`+(%K1VMCowjn(hWd~cR=jCL-+N2PDXp77ZTh%FaQ{YnbSt4UYQ
zaZ`%j;AyoZHjA>|nVFVX#4uuE<?Mm<M_ASnj#qf75!4+=h88x6J>g$FDc7#7fFRe&
z`h|CdbVcT}<sdho($K&npbF4}-DkHJ&6!y))MyKWgwAbt;YA+lNP*yCsr8wOMxqjr
zFm-+RP_~9Q0@3=6QkbF1z}nC^`sjc6(ZQxbNvVMevo@{LAO_8twje$g57?g<l{*Qc
zGm=74I|O@V9}Wn)eJC8$Z3_K~<o*k_lE+>7ymM;kch|OM{7*pV05DuGSCWv51RniM
zS+Xd>oB{DJSiI=yaKg8i{bG0LjJ}1EAJCUtJCw_7q;%0y!fRjEcRz~ILJ9f=0b<{a
z;q4DBskOiP&3H=V;KY2qQtJYvQBY<7W8$E~@=#zLf5jv@kB=~~D+jbD9yRastt$HJ
z>Ik+XKk5<Sazf2d?5J`(Gk(~T)i=0Rj=gd=OS%#rpG+x01AQnhPZQVmb(xZJpA#G{
zjq<d<dZH)GUPO|{C6zc|oVHIaL=Mqe{K}guN3+BC2B^9qX5Ief+>K^e8x0KMqzT|q
zwR#+3e+-gNJB_eY1+(FA$lGT3+;nVZaR_}<M}XStmDtpD8iBi0qAeP%F0cj+1*8(q
zR}rY8lY$Id7^RITo~=0Sf1w$usC0khjp$E?m19uF2&K)}@U})X&q-)D^5w(TY;&0K
zI!1=TOV1G)2!62=6DU`W3#9Po*ScU-nr!ddKJxA$PbjWu$FoV_1RVKQRQg@c0VD{M
zk6A&Og%S?*PsePg-gnm<pHx{$TH5aTT5p59B^3Vc8B0qKpQ2l;na9FK)mAad-W(P9
zwULrE;pn)W?7)8H;_PWLRIyk=-*t#wcgow$mKN(J5k<ktWF+ACTp^YA)n0y)^S<P<
z%xd4EreH~s2w1GR-w8D~$rlCZpd=k=SGe|jsJSJmTJ}H*l@DVXx4V2pnNuNOC*fie
zwH|jZ$c-@fx%Lo{u!$tcya5tpQFUI|NJiyC`~oi_I9G~Gm>f@dd*d70soI`X16XxM
zab8y|XnifRk|Oj>kAQqwkaNT5%R+py)+?KK#V|Z!^U-(Wd=^aC{>teLI0SicAZO$Z
zqpkufS*^q@H+IYq#Tk#h4^DBC0?$G?AmbXyn(Om)UodSsWyz6|aX<Wq9Kj@_UHtTV
zZxTL8q)7<IgcTp0v2XA16`Ndb{Qdna>Sm3mLcwYkPFv$&pq=w(wj-dy!4kVQ;=z1b
zd}umn2^Ea%Fe)&afOwI(jw*c%6R86(L1U1u1Y7Ful+=SP#)ZBUyI2pUv%8mkr3|)5
z!w}whP{oCKR&zt#T!HnKvnU%6%z{)@#+HHRCy;J5*zV)HyU%Y?RBS{z@<K7}T_9Dk
zafeFR`KKH5{&(@QaBU2yfMEO5MB^AGs_sE}m=}fN=!!pZK_kB+`*1`g<J*VQTr`z!
z+OoY+{^GxXwZ}C?aua!(;nhN<esG#UPU<F{{YP^6-^bDMs&!3Bs_!hX)A5|_909a|
z3$?gC74mhDfq{us@rZYrO-g)nPj0(Sz0;H4-UVALk~@x~Px}M{NWPKaDjM{tJcIn{
zyvAO5S}X8pF1k5=<ssbYVWuPVCO3?En>SqU_KC4z+S$`=j;+wDIP@TK^z3St6Qus6
z+!#=<V<IPGjlacqx;O;MGFI>qbeFrSGv^6Gl{Q_1)DPHir5~&FoVkQutnk!emyjBA
z{w?68@JU3m^)nw3@Xa!i@zDijV*=P5o?uo1;QbCT2T-}smTHOX>SjQmE_cTRNP$O%
zsY~p@#58HjoD;LEB=&k#DmB#Zemba!a}IW0Av%f_Q7H)g>XYz_xwBNOylcxtzcJd0
zhRpFcb%7Cb1TniFj~wjE7m<V*v=h|s>CXJQ?ZKU>gqSA5gb8y|LR23L)f)dy%T8d|
z{&m>zj?x<kHk&`DCYGzmyV{`|pZw`&KBE1urPFw&snm;MAjQ6&pxS(K*7Y$JKQ-yR
zeQfXhOB?!X)4{0{gZ<2JpY12>BmmR5HPc(m6hnVhB-T~H;i*40>4D2&n<GgRbyG7f
z=gFah3oTi?rw_kaib<Z~5}z~n7l7#{&(MU-QJNiW>XP%OPgxRVD^ftJ+`wo@+4>^$
z56Z8YCKnrnqD<3%H9$*UoiI4W!Fq-F`_tRq88_+Z;irA}4+SP%;E?sUs=WQBNHSDT
z#wN;-9C41XVZ3G*fkBmC0Oa(Ch^A#EkFT!+(fOqyROHEV73Jn=dgAB6MY!Bqk#^IZ
zBPvMRdAuA@P`Cvg<|#~35~SsP_EMKe77KLHUp!GR8}~&MfU~NL{eO+e=+w?$CY3YK
zV)75W`B>J!i$8z<Ockp|d0y=s*y`FFi+}zM_5A#tl*CUmU+u6V213eu$E}~)qF6B5
z;f2c;90G9RR7?P0gn-lHJGYHEv4gyxFMqvUrC4K+C!0l?BE(n`O;TgS;Ps?+8JAgw
zo@j}tPHXSjVIP^^c1)gv9X->4EzHJe=f3Sw^Cgzu@F>+Xg~E-r(z`w~o@H_N`QMXP
z_e;h;*LSq|qM4vE<s_^(djyazM>1VW?zei<8XD_e^25MFn_Ozy(+3t@v5?Lq`r%s4
z<TpkgA9UJNTf`;mI@=8YFbY!3P`hyQynOT%bM5YECWQLVY|Lv*(wc{tx7|!TYx=L?
zB=0XA=0`GajVkO7r@WhvboxiamXTC%G2F$+iFttAL7%)nslBUcA)<uA3~5CfWHzl5
zWuqxM=zD-)h*I`6dWeyf9Ye*xa)j>?IVyONd>VtID-rcZ7EYJM)rJ0^-Qa)fg2ajq
zTE;ASEB=E1uKoV2E@Rs$;aR<{-G>p9d9l7wfmH*>h^6UI4G-TuoNgOw`-~qO?D@fL
z)P$jVMQ;9N{8DgorzHIDzo6mKRuAcA&L^rZlz@)T&i;{+?Z>-|G$9~kXh{0w$B*Xo
ze<`4@udmbiJz~<*h&nnt&SN`0JUpr`r++SKLl33VBZwr=SdA$!wBEhfpcy|*nNFLu
z3W<$Zyv2Q98Zu_IehIc&f%%0&Ec`@C+<lb86isR4#j3w~(|10l=<l?kvX^kyv013^
zMh|?)K>9$Vd|KFF-=M|JR!8qj2)1QLim#p?-}-568j{<awiJq$t#7?OD`kcmO7Mj2
z2*DLz?w3WX><TW`pDoOFRr;2nh8lyV5uw|(Kj*00#pA_*l-y*c5awKSv9g`>T$|#a
z@9%wls-5<UMdptv>rNKqg<Rf3nJVk9x`5c$QcKcpCu8AKPw1yxc4f&5x}V?DbVhwc
z-!s3!QBPMNFJ#s|M0wQ;CuU9Zb?!p!=#8m!Kb<raP8b70{i^MMs4d#cP322vjl_o4
z!=e?=a0O==p2Rfq<k!}612xZU{JFII`v>$7ihhfCK+o~U8FhRw#6l0j(3htsk^rDg
zKJ0u${gT^RKUi_g+0#Rb@?r`*{iHW3*rWI>Bc*TTO31H}2+qSIamjm=fvbRQdJ~lo
zsP8kq+u4m384xo69gNzn8yQ}0y!SilKaB1#PuV}!t#du=r&yS)kRt}Nf|K<QMv)`}
zKcf}a0Qn+i+J$P1&TP?;0;PNa4Wt%3=jM`{oB1tg%VfTMfq_cpvfD1LlYDl3pa+{S
z0jG6eqMlDb-%0aDbLaScg3~CU|6kEk)!6v>pu$3$o*uYz&5!27rK(j{%WqYm*(1S;
z?pMG4nMJr!5O0!JA~_XX=w`uS*B2tOJVkwDW9hn9qg@VnTE_+wwqigiE;Th(PjqQ$
zsHE?uDbS4kb{EYj<(`Emlw|@lp1ysQfJi1cP+7TFdGKP*d~jA5NNmSU02{WJbsNYK
zURrEON75p5vX7N!P<>nUejrNN=6tLaQYCn!84gZpII>Ub1ZN2P1~6CF?ZetXKd_)P
z=5Iwc*MP)NRXH#Vcz!9>%riWipt08E#LW9B*I3$=fa<RCjXYin+nx=xV?wyJr)w%r
zH=uQGSTINLF6<MyDrP;-F$MRRs$%k^XxhWM#@dFAfqCsM)=-;bVwZ%1A&e*JK%@fh
z$yixDh-BDUClqt3iqs|0y)X-jI-f!^bNI1Ylf!45fB`@}XWPh7^D!tYu1o#C&Q0`4
zAI0jevy`HLobxfJe`-Ec*mI-ix)(DKOYq&x!EiSlASI_xD3wXB#%AGc2`<P|DRnlX
zv<1H7v0TX&O<R;0`NGm*AvroZRaR7t-Ca1k&t`&X4t^mqK*2No9`YVq2$Bm1GT!Z$
zf4*)c;B6l!*W@Wh?vQwQ;1g+U*8VTdUG2&Yvj1**!@J+u%q&My0(YODA*`){y7}?%
z-@nW(Ebmt+bRCy}mi>VKY-&oGoSY0bH#etm`lpl3Uy;>nhJGZCr}K|YYK9xNKUo|`
zzzZB0Z<y`v?NE>BgG}9&-#`{9440_~q<hAIG5MJ?9jOyY3kwU20k}A=wwQ$T1?jA~
zWSDX?ggkM|*zvoTQHzCRVorvagpG$;Os^-H%2_!qb2FQR&+>z|&C3A)F*dDGzIo8Z
zOx8RhJ;X7B*b3NOjTCaXr^B!{!7(>7w3KbxPTzyrmGmH{6&sj=`S@#JVLxRF7N8N<
znhhRuqS0>qUe<M_rMbvrIs;dx%|Eg&Wa_Xs<#R5}?>$&0P9db(cJ=`TO3hu-eL>Ov
zTg@+fSg+LI9}_m7t5MSlM#_KPQ*V#JVbfxwmTLJSqPxFa^x&vDe2Grk;I7rYw)8$j
zX3#7q1&eB0_!%8j$!VX%b`%#-N-(m}sig|CnJ+{~)%6Nspfhta9<^LAg9I1j!!9Ig
zX-;NM6M_#(6kjwoUa$g;632+O!}IKBaKq5B9N0))4j+I6Qo;Z;RwraUz8v>>qocpY
z`pJp$C4(>Ue0OrEuc=7rsL@-xsSYl|{cKAES-|VFCc?z;9(Mx*9^hXT7b#U!*n%v%
zz^KAkWP7#^(@j{#*p12>x0xKa|4roC?4TT>$#S~r(L_k74P#NQ?2((N%5s|4^X3@!
zv%bFm^F?nYF`IrMc-=17vbNvZEd`;)hBjtu<PAxeK63DoSZTqh>CmlisI_%zg#;xA
zfGjuI{_}f%fwM2m4Fdsu(Vr;37}**G1vq(tG*^}j2D{*PS@o)gFXA1o`o9LO!h}D;
zkNcb^kIgp5W+wJe=R7g%q`zX*LtAp_wpxnt4}S+)dDsO@t9RrZ;X1H=47}ZK%FUxR
zG};N2x|4KQD}0cY`ST~Dj5Iv!0G|BY&#3iXNX=3kF(4u9N?Aq%0k`=w!=A&LZa*|D
zYhAK##_(UeyqjGxrqc-=IEAZg&adyT`<3f`4kdSe8O+KgmvgF5mJm5?!N6=L(?Kbj
z&4s>7aVBFE*jn46{@L8DO_aWTc3CLQQ}^?}S4HwXN(WXTD!qy(i4gnlWpu;ISa<h;
zAoc^$O^_}sawrhJiVY%R7-8xFl#CBN&bwn9BOLW#xBB0=%9UA}n8ddcVPYEnngdDE
z%Dzp<=HmMeo#_{kkdJY1j<JhZbt`{Do{Vz^((DHopxWM&US|((if-emPL=50hl-Gs
z8-Zu~ZHoX^%3F(x5Z=s<x9x_9hez>9(+Ds!l9m~qF_*XLGiIf|pUV9$C)4ua`N+fi
znTC*XSx7u##&ybePV)sXBm2?F%*5Bo%;t;hcV5Z7^+M##<URC3N4(RQ#>ct;n~+(%
zR}pnFF=HRMrlyn+4-bFZWJaIea<)^=ddHO7oLyo4_c|q4w5{BrIQ}O-Kq}$B!pV8O
zjY*i=geJ5yO{gJc*K1ZnB2Mg-x4~1fNtllIc515VkF~}vQK7QrM-e%U_nt9bxA4?h
z9^Bjw3gdM+F3^jHwicy;b>6@0exsd|URY?3UuhvVrJ~txHJee@GMr0115rA!t6A`X
zl~6KPZI&B3z;!#6{veRR@cnizq`P(nBtiXLNc#so`HesbJB><|2X@obBFC%KXq++=
ztQC38**rn*L9c_AM6$RTt2Hx?nq5Gfza{hc_KZupdQb*5<=Af}y9mnCp%)^ZQTkQ`
ziK(cEv0Q%W&#tDti?c&!4=JSIq#~KBBSH)DhZovLy}*4wkk*vw!$Od>7~4BCTXfsg
z-H-~4#KHLW@x`dZDOa&GLVD9NhgIiZ<iFp3O~v$d-75eNbwH1V+nx;M>Au!Rgoo$q
zwfWo$q2sfD0`6u?weAKPG^?4-!$H*7Q9Rg`(#EdHOs~@xZG+fgPeI)4xCH|KMf*2M
z{(jW)0FqBCUmA-ybc4MmsU^J8?&jN0Z+%A<yba_Q{(rMS3&dd`tNx#IfKTnt!BMCO
z)m&2K@ClU2V2>%LG#^^CZWcJBpNqkDZGs6->Loslk*bHv*EvSwQ}XQb(V`G>)ef0d
zCGbo|-%tfl6H${p+N@+X^JozN+?4~kw!2fgxs?=d=^@Sy+K`h&%a0E^fM-8a-W_xa
zSdeh3m80_eecIJ3)-m~kB5MD++`zyh$4lscQ;0|*%Pj)rb&;Jd3V(e#FRH6cQe!#&
zTP{=Z@_0G*CpI0(OxJteFoJtY)A_$9cLy~VlfmG!V*91pX0i5hs}6v84vrbTvYX}9
z4NE^}7g`2KT#)sTpz(4yc3W%EKckk{c&SONQpWN8Y}Pgn_l*68>6Lq52<l|SahVx&
zQAFF2-lZUg*<s$DSoB=l^i@-YQ0rA%Mn9!$X@UThjtb2>&}MwBs@c}BEFq?USQjYo
z`m+;xvcAZ&tgyKae8z;oohv>??1DH4IK!3zzJ}gdpjzwGb-Rop9;;3r9%>1wfwDeK
zm*B%j=VEOaZwmeRrcbeB+Z-yT)+>LEvi5r9mdjJe6a6qUh>?P74Ebu8sD|l|2&AbH
z{6Em#9DRp;FdyE%8X4t-S;ICtwaW#eN89s#@#nv;TLVCRwcSd_{&exi&W?DK>k-l?
zw^R<Rp6jE<XrVLM2Hh6V61}#JtAp9eEMaKTkk?V7xdU=6gr?*qMZ<!f@y8qf<xhJg
z-Vj6SYkOUT#)1)e0(mZ)${mY0LenuZUP0hYR%nELVI@Ln2C8wASbkfDVp_J`>|yP#
z_Y?sSiVaXU#BzZ-q1@a}lV|DK-JC5><l#IU5Kw0VP1i9yjU;z0XlR21P1{j7-M)|%
z(KfpJuh!&i9S|%%Oux&c@2{M({y_MvcniUt<8;vGTR*tFUZJ~2^9M0oG)}@w&@RzY
zfWSNS_Wt^Wn?EwMZx+g#z{dc;_%ZVU*3G`9r8IE$RSCTKynXxjr}fUs$>{NN6A!50
z!@<FU)wNn_6%g>c&IP3#uo!KxKbXY?*3f&O>r&^%^SE#Ig<m5E4)7;<B{`m2lNcz~
zF$dDXT~npoq+tb{U$T#UYton9Sm`arc~DLRww(4%<X+sxW`8-<M&K5}Hl{T&40J{8
z0tZwVVqdvd%^%4g<2mZnwAjd$H{^V5(1NGG#E`9cNz>XGq)Xm9;!ByLQm`ERZL=i&
z1}Ac*JX11odML3<q4Jpim-0|ShocqZ2nx2`x;e|Wb3n-QgffA}e1_DuV@YCc&pOoP
zC(XDs=^!LU!2=e=SN<iCwQ!0xAB+kxF))aTq(QZaTAB7*Uo<%pk7FPxB8ZKN5&8BF
zftc5M%yP!`wTdza%ATO)ENel3ph0g%2oR{8Fc$f$P-LJ+kAfZ&$Ip9BB=S%PkG-t>
zE(5wobNDALx_CT#F-)21f(uz<dt$4;MOSNE0*3bq;8{8Bg_<F$AI7yN%!%@V8%ZfC
z(f1pi4WFQI_gRhU_P|LYLK9o>=fNh3-;vRS2UN%E{9PZQLFf%}B<G?EyOL@f_jyv(
zH?-`T;{nz-*J+5W6xI|Qg4aK7f<tx1V_Eg{;V^eo!gv1rs3fW7f3@~}QoQ2%F+YXG
zc>vnm3lixczWmrvA)|S^8`ld*S+Ai1mVRmcW;i<oYisKQ)l!1#A{8&3RLs}Y!ss_<
zg^SV(H)=R9(PD+&UiL**80*m>xZ1I%r4U|sS`(oq`VrtY?;AI@&!XRPvHV7ux~sGp
zw*mkCTS&s#l0=>Cp>$R%VbSyrPcv{J2YK2&_?SAdCH9@i3K~>3jvS+g>09Q;{jk~k
z6z^@%8ZS1`yA491y@Z!JqD`?L6ru-mVowxC7EmY$B;wqAS_(Sev_Wosk64HxRX?M|
zUVK9f6GBfESJlh-SDqYFl5-jz7#`_S(1Lo<#-`qXFZ~CBVa2lN(vln?0m4Iwu+g@b
z@P*vy;;q=H%=tjhQ#z?FpqmZ;Z1j4IndWVghSW~Mp9ts3^Qm{S<Pi&CDj_!Dr6#d%
z0#X*!4R(7Z2>*kL!giyd+UZ^o&m<a*u@*T>C9VnvI-kd%blFt7n(<!?j$a7CxjRTN
zopO{}Qg}@JbQ~a_5J9F|j?Ipm(euu%60gHeU4^CbyX{v)wir7^eH|yYqH)(W#iEa(
zQ(S-FR3s}GDdu-EJN{)u6!3zgkI3?CnQ4k1ox$6mkM`ybt+)s|v%73O_QwAO-p|nO
z6}1=V^AIQoigx|@M)ni-!pvM(oNP<*`L*xa+yb8!P(zHV#0JV))BoYvo$qv<Err$7
z$A?ghfJDMs`>#E4@jGkD_ZkmiGrjnL(fs<IuDtQZYUuU|udYmY^Dz2osopUoP~BqN
zoeCEn+=h%*XyAfW?t7D<K-MF0`CyS99N|N(@)rh~{CK%JT(fj$*ITbmLVTaYauk?8
ztx6O48?iXc7>z|5QCW$IHgc?7iv6V%N9XKx;xV!eq!8sdw?r74)~P}#zsL}Br+2la
zCJsvb<K}K!CF*v1v^-D+X7K!L*l7CE00e+4lM?Gg2gkEqS$VCuDXtT|l@FhirNzKW
ziQO6!)F+h>ZqWJMYy+dy-bGY~0{D^O=M^sof@)rei%X7GlK<QE<;i?<f5mEVXztc@
zrb$nVo}o53zhR0cCkwo2HNXlh^&^f?_~(TB5clT_Iv$(Fq{l}F<(VQHC2PxqVm<H?
za#yZNC0VDQ`Nze5GEKsh8z@jy-LefX81ov58<r5-8fDIk7lCT!I`aSWjf|f!FJV=M
zHlB#r`3EtdtNrt%$M4_2<%`2o$Auk1L^gpW@d|<l;+E3r!A8>8Khgym$0p>XGIs17
z1yWMe`_l;xy2O@y@|wA@y{!x_Ep1Pa_~}4e6XXHFANWOxzqWc<=J591#O!>xdAsQH
z<j?>hrKB7^`50nBb2u9I2*CwE;5ElP;w1gKSXpBd7_>kIp9;Y;^6R}L(=kx^wWQ)W
z-H+jv1;OLG5BETl;bi#6i=>1sxy<usUH`I8-kHqEgG;`zmX41~<%>Lx69p<<aN#L0
z9iN(Q^_sd`ho!i^ocSCS>X4ha*HnOrV1~wtwIg9rF2<jakSPShvple4S^Q<w*(xGq
zLD%v!3zv~b#Y{s(I*e*HbVR!TKROSA2@#~tl@t{V^x6dR@uR3_5fBi_$jCk~ZPS7W
z+WdEl7F$bg*yyQTsej;Ge1<do8#gEBSW!a~3pWgIP?HVsR+eQ8$n(KP>6Iw`bp&;m
z^VQaHl6fMY9TNz&>b^Dab7@y~IYA=kaQPA$zLp^eoQj&~z2e^7%v4M$#opSC;1WTe
zb~c*)_=nr)v6Me+qutnv+XLO!ph#D)H$!xn&6@I$Db6Uz@lJcWt<vk(0fws6Dzeed
z`2%*`{*Dh}8vyP>h|-oN8gXFzJ@m`P3xlirp+>zk|91~C?PDNX*QvKR@Nt~&?G4%J
zjoR4WHr*ZmFCCG|srX{=V(;zuGl$6%g{g3N_u(K3ilvj{6WI3rh^osQpj0vZc+$7r
z)b5Y!XQ_zV9L2&U)<YMOQKDZN2o(Wt1EPS<qCckcRw3CZ>=}sonH>SRC7@iNplMC3
zY1e<SrUoB~+1nlW_pvUlmXUiMf3+)I{!2J*#DHH*B{i3LmC-P|c7wbB*+uR8>f!VE
zw}31`^9+8ncI5rHk04E3;_|3uXNCgh$IwW?%zM<o>G(F)%>-0Bfz)U7Nw$_akT|&}
zMroVT3KkkJ>tFvXH2$Y)_|0Q0jV=f0iCnfz)G<x}RUMg9#k0=bR#2%sje;{Bye7DM
zk-5qEt-t<yYt;?d-4^q-#87J6?E<CG1U6PPNyp$TrO_~p{tfNRkCy2>;{N-B=6B6p
zM;U(p#t5GZ-P(9<3W!3j0{47rc%7Rcj&e6`e+kgvgk)@G${xS1PZe%MI`z^yt=iVS
z)%fEB2|X+c&tGiqIwaCDe-abgGi)F5YIJdA!Iq^2yRJml^$8)nppNsMK8$0-`w}lG
z#D)d8Eags&!rWdh@3t?L4=3<)|7BOAhM^|G;PvrNdLh7eM#>(3escZCvMtJjXP=Ie
zL9hR=cs9fBhkq6q7~tgO<QIunvt>5B$FJpRE$FYyG{5X~m4xImV*v$)tCS&qz=!y(
zD5{2ejUl4th_u5YL$&!P3l=jpJ6KHop&~SfG)YJ#=qh--q72;-4vGdTDSi(^fvBVz
zvE>I4)j)SN7-yLhA?Csv#KV!5fC@@ND~xBxCeUlcJ%BMR5ZSJkn{(efKW!OyZ~kMq
z=vahwn#P5);*#i2o39wX0|A`lr%iK~wG;w%Pxy|B0#HE6X@^E#!~Nf^ORe?S(eZA3
zbo%ykZ}gvsP{rMx!*h#A*liq^?U%2B)8PuUX^DpTYQ>r}9#`}jAy=kbpMMq`_~-%x
z0vCzkc!#pw^4foznL%1c<K;lPkiIj{a$y$z=m=5xx=XOWd!$APXI?r_vGOs5WddE4
zRv4A=sCM$7F2Pe#FxbZ}>TGEc1semT%wD1`;iInOBeVZVZ=JfRRIsf9AWc^P{=K57
zUxKhKV@7yNs~ADy{gKNY5d{UWlp_x@i0<O}7u;UraEQGUJxfydg}$SySiuT^;SEO$
za7Pt8-!9UJpw1Bm#g_^g?K}oWD-;ef(ZA-W2)+{+W5zal@9FTswDHd7M82xavu6|X
zO|ZB0{ciTBNIYr#7eH6qYCKTVwf50lFi<QrLcjYNdX(?dtO#5yC8eZbU}9c0i@v`!
zK83N1A`v)GoSc~W|BvI^=J{)ie5okUu-Nf)BFgO)@+RxECJSNS$MP}LCNBC<d~C$D
z8Y1Gj%l_VrsM<r{B!1Fi=D2gcut;g%7WgYWifFk4l&5~m<yYlGt!(<nOu=x&kkzXG
zK^9I<zwWxVFksD@mHIh$PX(Hb)!^6#j|48Xp?ZHQF{<}b{f{JmQc?<=4}`O5#ug?T
zpLz9TNj-ce59mpd?|^XM?tVOs0CijO5<-VPEh1ZHr1k%Sh*526Mb=^QaLUaE7Xq((
z!09XpLVq-s5ZM}UUt9f|G1kDsP`cDnHXIF!>~aGFo(@cqV!toX%EjQw@~9VX27Zu;
zu&(af3QHhvSUYC3XfR5@9;C^wJiF<&c?IJ3?1^gS29<eE<=j+50a_DEm<`|0+_PoS
zZ1or!oL$Fla}Q#rHx)k;%(XZ~+XCKr-y{v`-@gL1rU+t}lLBuZ<^-qXTShDo-YHw}
z_8RbU_oUD^R2n$7B|<{gvcxOm`qGII^@2rDE|{sYz^${1O)r(+^Cya68Kgp|$FzB#
zM&_f$0OvgpBJv=j=Vtg8(cCY`w{)j2Fepth0s(n}F4AZqSEfHQP^AbMs{gMfWQc%e
zf@`vY5O-8d#e4n&qIJtVhusue82vsc;O*Q*4swqpj~>1DP$kfrq0r^8j|XnER14(&
z7_@Sksx!ZzBRV)Z1VaK}4>``Gfky+J72w1%PiG7+k7M5+y+J_d(cB%pu@(P@jCypn
zIMR7#C#F4C4v-4yw7L`3Xfnh{j+wnH40=CR`GJl+9>in#^sTrO(+d$vb-aO~E$IEJ
zrkOMr<+}_xDhv?fcWtd}3HYK1OB<cd5ITQ1GfyUHc+jK#vyFmxCxFbQlWkT09_;D=
zmgx~>gJUJ$CHuwa2{7<pneH>eCtq&Zf^;bVNd+wMiZ|TFR1byz^`8#jHYIG_aOw6+
zhfQRcdwH?6*&t&{Qe-k|hf9`9M9>LS6Qbz;pr@t%2`acj&&fz2B4&Yl<;N<sQIKxu
zoSUDIzs+lZRZe|)o&G!6Kv5Y5Lw)i=VlIq58#Ecqv2Zri&X%#%qCLl?9pCXblrdmt
zvO61<5kqp6-<0tH4I7zmcLnwI;PnO+AF`i}g>4giMkl6<8a_gU0;kIAzulGE9`vPQ
zq|=Hx4W74Sz$X8FM3R)HA`z%fqsD3gXZVfP!$@t$ZNLAnRV_j>&u~Ap_|0v$jud@(
zjhu(~re2s3W02Ezk4!C*kT7sFn?k@lgh3N8$!WXP_t#{Q+i4pGBqu>N#{@{~+~1tY
zf!Hrh8`Pq|n1K6%!}+R^f&yw#Ho6CTW-`66@gpOnPQv-TuIa%e7-kw>U0d_hes)LM
z1*N24!R6KkpjteAxYQ5}N)?^fyAb0fDIy3s3|Ig8IfGjV;9ukYp%h3Pg`g1|ZuCSD
zf&0;io0E;hL+Mmbo7?p;rUI?HGTGis75!6~^29_nzkA~O`T6hH^JWC)<v42JUUM$-
zv#jf%YJFSZRFk1xW0m6^!545p=2xkGbB=Pi7;UIh7zbMWs5(jbJlX~I96(G2G7&vB
z%{LclKsT8&s9uR=oa5P7_tN}dF7(X*_;%dWJ=s;oFTCYK7flH*1_yikoS<XP?|7I@
z<sL6d0Wy56w`cx&^Nc!;U7+VENx+K}l-k0^)R1jWh((bMgPuYpeo`D9F?Ds&S}+8@
zcmrJ1lfBeXIk`#ZVdN=lWVMMqIsr}=AiV)TJnh&`tM|iPogFDE0sHFdTBo`uPP3Il
zqBlP1I6%gsryhGAN#~>2stpJ21+M~Q&_M7o<$x)f{UzYP!lr+6c^WkU13l#<GsG~r
z1K{DOe$GH*#e&uAQ)mZQt9z8vKYiaDD5g(*ty={-!y=?wxtM35kBx(4mX<$WvVRM?
z_f?_a`tTPX6OPkte<xj?`^pII<UaOt>PzW;m6KyU4h9PIn&|JTkf$8M|C%d!_7km6
zJ8#@`{`%nlT~sOhQ5db@1B^~QKRxjHJaThbPLc8Rr!6gME7M|w_ZAou!e%*DAaW9}
zSE2vrVtW|K5sM&r-E)_e(1S0bp`oGI_!|hmUbEU<*wd323YC!aOR4G()pt^UcVErY
z5uo|3hw#oDotQUHGz4vJd%LGlDZg)UP&8ZKeyszxxupdW4ebx;%z&H6wX?GW=~!`R
zXLfLTe)mveGKd4}aq+lpiK3&UgZVd^Py>Ld)h9bs-Pe29{mSVoJ2%G{>&o?J0$abd
zw|r@Eg4B>Cz6iUa6x7id>WF&LR5tM-%?SC1e-zuW4Ta+~a$kXNte1=}3Ce$yLL~~4
zM!nbNMpvVlp{;H$W1N896RHuW(n6cS50IWVXP@}%f77W|!<cV@B|iS<{C3U*!1g)(
zAJz^`>`YHmhqC1510}mPg^1WTHuVW{(weF`oH#ha2y*1F{FnaOdN=F<Zc=ixLW`#h
zXg0c-;@#WZ16@gPsi>&@0|F}DKk8~}X~Cfp{ZK44K3#1u(QQeQ6TJKMawVRJ{gbxx
z9VTXHdE0XeX=;AH*|QY@TA(Pnx$$suacgat2WqTl3$z>l>OQc79?1NRs+yXuVXhV9
zo8x76J4hOZR03Axaq-A1A|x1<kf2fOzv8(DZY~JekGk5^=~mq_s+sb&#64&&MCi$~
zEb>KCr%)|>v%&xQ_BM_yQEU&=A0x@xyp0e+{BIDWB!Uc(0gbA8Nh6=y*B4LS_@xUe
zOVQaauT(<~Oicc;yZ|jcPR6O1;Mi@xjWAu(-VXCO?vxB(mMWV&tq+7<j?O0pJtGil
zmzUX}M~kyeP+sC{F`b4V1aZ@3YPj{~en^MR5rLGz>MFhE<Le9w!nh%f$z0@zNSuY`
z2vs~l9NFocf8f-VS<0$6JClmJIrVS~$E0~W=*{`9Sa`c%Tn<vQnU~j#c5Lpa`XEE6
zO4hCw14=lcP9a_=*x!x?*LQdOb~1qxutJWO=``IwUW{HXB(B52!M&|88zqTeWeB7m
z|1U-Ja#a$2COV3|vc@WV9!J!_4dno!x$$d)x%*A51X*nRvpb$E2&`x70`w04<ra;V
z5rNLyn6~MX>;PoxQaGVmKLSpB257jDj?hoWjy%ivJ{s2F=9BlN@#aJ%j}y2;G^W=c
zQb4P&<YfXEydE9R8s7r$^J&QJU~~#rZ<lZ>B2Q)h+_Q-z55vkXvL(l(2Ow9s0e2cw
zZF?ZT^Lm@$m%9bef;#lCr1(-dy{LQpzctfMOg3PxCdS0X1g~A};reL4-htfh1oAac
z5xZd|C;G2ZZ)Eq>RP|VIj<J=yd%fyeU~ur;u}Y)fkBikdi#V_LrM(uWC_^FvPj)ci
zx)&CB0a7p(q3QuplBD`@TJ#Q4_w}=>DQXC-e_NK4{TT71sH-_yb;Wt~f%gRgKi2%v
zaAzXHRf?Y^L%cEQ0-JpLR~0x8Cz!z8#9QqetX$}0mxE(xV0EIsNeglgq?AD4@Q-&i
z$xT@}Q&n_6e36ujzumr3#Qo$qjq(b4s3ZTWllKq;=owhpAU7GUIV+w$;hKu0z))?h
z>xQ~6!&&~%TnZAAB4>`V1L04EiCJnw77YzcvJ&0x3-`&95Lk;ZhoEO{Z!#YpJd_+T
zL6u9?S=ydK&hk*{;ET98be3WvSexiB&&=P}P*EvPei89^rcX}~d-gdskEy=Bbq88*
zOa|g;rb;wo8=Q792nZy_#DXE<fiSy=26q^nn7k(?%@h*NjEWkqnk#vYB7QaS>C)1q
zEqEae<N8J@EdqI&QjT<(&$w^qC$Nvht2Yf()h-JqZtuw7Z9;W6b1cGrt8o*5kj#50
zqEsl~U|KmWNBS71^kNOC#+C$V%PdgIxHtvWYyais*WYu?Wbs&QeajRw_jG>?&$r8c
zCBPY>{+6vAlux~r7x{P>4XhiiF>H!C@{a8Fw{59fl57Cf@sgk+5{j9VrDy6(ujBg;
zDdBA8U!Zm@C=Zr3Ldduy%yJ^)w^Wm{iCK;h_uGEf!M}a39mOBm=#T+0sHDR)`)94z
z8ulKWq2EcKdOmqdYgGOup0#DBEodEiI&AU?;gA10OSlUO5#;CRSFbdL1<h7#piL!>
z+rb}fzUN1EE397E)*s2qkvFG_h=`!n+k9Sk2JGhaQ89H<7%f2(vI(vE+&lt059+Dj
zq2b}i=bQaWsi{?f=`cYGr_IlR0BG<J1Avw6{QR7m0~-j#&6^TCEt<9jQPH<j%^J{N
zk!#rf6GTPp2M1bN%b=s{ANU6d@Im7jQqb=okk`yv>wYDsZ1q}Tj<@wGLBpE<1a^I@
z%B7?S?JXSvILOTr6)-Ma)>^IwepHkA(K5QC1puJAA8#$k!|~?55q|jP8*=<bHE%gj
z%QbS-X-ke`pXbcWPG&oc7*3n3a;Rg|*L^2Qv4+`bUPJ^}6FpOc!|x-u1-p|c3qbd2
zyZw>y^oT>gopADg6<b~90q_5sG_I^@o29ZdO>h5$LR$UJ=X^nBOAaSFsXREqlby*2
z@?*jjv?sQ;34znhhfkjZVUcmNlxdSWZ8ZH{-@JJP1zHF@S6QaYbO?Px8(UK!=gZ>-
zIuwWt>HbDgJ~-}<eSm6xc?QXd1F;6jEzk`qMj6|>^of#kzlo2jMp5F_)XZ%1_g7@k
z>q809U0G?v#>SS+VTDs=3JkAz1R>|WkNhvRam8;+5^w~I5QL^;#9A_BgQJ)_<_1A)
zdnA>JJ?mjq0I97==DXXw1a=9H@|P;Nl{J}oJ*eI`iAH#xG2ub#xRx8mID<C|_&>8c
z<b;J`;As)4xrMV)6kwwB{HbaE$+A9@i4H*G6gU+%8*c{gpBw<GY*#7^=%YJ+Ia1r$
zn+z}8l$h<!mYu5*xl|Ds_>l|*iu<1Uw^3Dg4@|q&^JDob6KL&N)5KY+xcGZCM9*eQ
zexi&aq#O5JZ0|Nvj+SNkA|sIC9-u1@W;T=p%Zx~~fj$KGS^{-%=wK2Rr25pHwa*_9
zrF>UxJK)z&4;6qpqKKAaaycfC_yjrin<&Q*f$u~R6nsGP@C3~Z&Uf~CK|*sxWizkx
zc2^?u%h$=jw#4@?$HseD!xN=iT%VKoJo8sq^`TTt)Z0OTErmU~taFzN`rK;%J$PO2
zMiV}452xN8%SlQaLYh6GE|0X%HI8A^2TO^pZIFG0*Z{MUG>qQz@PXWae8zq6Tu&gV
z)I!yu@$9!M?)L&MIh=Z(Qd|*)|BtAz45+H>x&;+b8l^!HmF{j(P>>L$rMtVEEsdlo
z-ICHE-Q5k+-3^CsxC?#W@BZ)y3Uc<^bIlxMOtL+D5pT~l*E@NY7p1mJm$MA!@UKj0
z%+W$$Q@cA0K2|cw>aTtrZz1HRNN}gm>Qvx&C_XDzwX^kMV;_NJ8<s!4f5gVPnQ()#
zKdH7bD|$L(O>A2rwZ8+w({No&H+5VTu07;AIE*qb#h)Os=MYEnbB@;PR(3ig?rlJI
zT`HC_-hi25XiV}XrXa<|g{Oc+BJ-)lL))HAyo%#k)RUxZ;j|!=Q#$m>k7v}-zCN9u
zFk<V4F3A7n99nj;xnJ6WDRmo|quLhQaN?t?0j?~7=E1|DL1B#5$&T6CL;_X=xIti}
z9+#Vxva+%^`vbkbD}Z^|8^i29I5^06Rc<!<^owP~4M#=(R^s*bH9n6W72NK0#d_=4
zXZWay!6L6fwUY)yp;4I%Krp<&xpXc}pxvzk174CZPo7cS6LDwSCYF|#V%UtImKgNW
z94vo^2$XBZ1{YiWm4U~CQ?GH*2W)hJ`eP&f=t;Xed+`W386?nl1*}@|#P6yk#DmFR
z@xQ0u;aPF(G%oZb8`b_s>XXr!jUCXVc)8;;A%rHXZ;;mk36vFN<>3?A*DXk|*jf${
z82(OKvU!A6`#&r}s4dw5*N-Y_T8Q0sDYY=IufKV$zkWh}=DKc!)0V3)3w?2EkI5pb
zPD<SFDNPyjxQT_lrA9nYB9ZR0))K10UpuuP82L4N&1TP%sOj#Fw$6Akh+(QcRzZ2f
z&yPGyDN$&+=lRO5(v?(1Yp7Cb<4>i`1-ZLFNCuce%oEzjKt`s4iY!jYR4uVUR+_+V
zi_hm|_4DfPWL(p^9s(sEsHheJX?_X(5lR~Z6k%n+i{A^L0g34Sc;a+_UOEVu{;<t#
zIQ<DA)G}YGRoiU}jTfi`I;E25E(tE(=X)=W5~4Jz?1s1x`1rOJ;3cVU^v<fPsj1mi
zNJEJ{`kQw5cP{rs;D9pDSlt(|G!cZpTq=5ejVC@&>=%9G26I(zm-e1BKV3f3#Eo>N
z<7MYE>6PGF=MsBzp;W415!S`pU#m-<%uGqy9$~`iiYH2VW-L)LeswfKIdHgfe0Q)=
z>G5ZOPAIVE-fA%)SUNg7xbwu9pKM!0Oe}$lJ&Me3Be)z14e#_!2zk6c)AP(i=YqA{
z=e*g}e}Bht*l+0{i^6%<aU_(iq1fXNY5=pD%5Y#b&+zdBMAAUVAu);H^|wwNGJpzO
zIrIxo3T4PLgMbLY#j#A<4?spMoHn$uwIx1XZXU~J{SpbCIK}mB&-v68nyF;|`ST~=
z>4dJ6lT*t0_K-;#en|TaGdemtNaB7w?5cpO!rH;19q`tXNZq~xKC<=o`2k-UzuW-G
z|BB70Ilw8u`9bmVQ|{>b)z#6A?SM5X^EKQK-lGT%<Ak*t!sbj0P{ncXa23eprp@<0
z8?#Yl+F2gnAG!n-RCmGXe>d5BLTmUogCa%IeQIVrI!AzA9H*nJ?c?MJ+6#Sc-|nxA
zJF-^us)l%+!)f0m5T?xf&UZN0xvHw}{D0U+CRqE@+2Q^`!;1a(3qh3KY4UB#57|M6
z)#5$GYC8IG!-8;3tRMzeQ5;go>#gy|DifR_k`5!Lr?!iTP*?a9z0CCGl|?VtHwt{F
ziVtbk2&wKrnss6B-fCAUH*0K#1AmFfWh`lP|AFvCl~9M^O#Sm@zB}TNCgU^>uTd5j
z7T(O)yGx7?oJxU^_zJ|&Z{NN>-5&vk=?v`wN}6IW0h1QoLqt?6Yc;w4w)S>EX5Drs
zgWi}2`wdVDjPCHaljj-8W~!!7PunzYoU=eJjAHl-d<g6TEv6!qMP4)InwUQ;UP^}c
z_D8AX&(fT?+AxCZcn}&}dhp6ryukzmKRAUW?Gs)&{Fi9yH=d5J&PbTg>_1W$j5)9@
zbH!F2AoHTbjkEq#kkaPpfg?)QGNn_K8ff?t1ceL>pLcD2>Nz+SV4mdSQED3M8a;OF
z={(Vmo~SvNX4pX%jt{@He0I0ySxWrJS5bP#f2#dhc|;7$dPn&X$*_W$jGcme{)U7-
zcj!3$)Q%7<*gKv;Q0d-31c-Mw^@oZHBQjrx`%hLvg7IghD{if578I`+wl;i{qbb#)
zc+L!HU5c&nnbEGDFH=w(O6}g>-n;kj;ejpM8gAg88gTLQ@;cTi(r<gnzVyLNJ$%Aj
z9kf}<pL*PbpbuYappX)mc<3(4k??_noH~@fC$q%)#4Xd~mKROcj)A2hB>#DciUZkB
zCj(bJa!4PQ-k$WM6y<3Pgt^&LbIyJ!>9In4tlYwEGTfUZhgp=`8s(b3JXNd*NKY4J
zNV&c}wq+D@>-Rzh5&vh<a5I@)A$D+!>_Q<v;QfT1H*Tm{(GmPcW7}7QS?7_{l6%i*
z<}Dnr33sENniy@P*Pn``1Xr3pmLOYd)qg3aX0Sg@f1G^72T#8pvBf3g9j*?ca2yUq
z#}AsGbkohY7l>mR;Gw<VUeF<Y60I{)XY~xJVOk0{?QMMF6wShA`u9)+`)1o>fAUwe
zmUhgZRAgaB=^Z?<%#up{Tw(D%rBn@ebVO@&oyh7@V^@4l6z${lC<(fC2GNOufgw+=
zYIv#R-ncC<ZFiO2*q0&lV8k?8{O20N$}2k!^a#fy8Rr)#4e6R6*LK6tSl1HCk9Rmz
zJI|ljpvfcGnx;uI!jdF>i?C~JA3iquQ{G6HSdF3Aceq(}K!JM9p8rv}n&dX%ZHDa7
zt)~Y9y7AQ7+XVGlG2U;FdJ3APIU66(tUnQc7~z0=m-s6o)4z`sTJ$G^O@FM6LFedB
zL7x@(_^PBV5g4F5dX$Xi$}DS4&rApBBCilw>&WsTOKxEr`YMRqS0U{`$f)<|BdM`w
z>hi|#(rXqA=Jj%t9%Z`$&5Kx^KjY^j6nXV;*OHmdVvf;<)?YqO;`=bgaqoM913oM|
ze`yK|_}<yMa&t(vc)8(nD|cp>OxW);8(!Mb2Y|HUNkp&H2-C3i6bJx6K;!en|8^}`
z+9t3j?4q&RK%Cao>$1&6esJ<xy&1Z@I};%)U4p=O){a~1Lwx?LhA%Z)7(0P-2KiA@
z{f|)!;g~jjsed6?5oSuYjFs-qAXykRg>i~Q#9wg6$sn<wo4Vw7&J+tzVYA&G0$4*u
zqmg{5ZmwU&c2mvCPbv5|bvQ#b)ExAncwv6S3@(?K1e$j|cDpAL##Jhx9aJa;gX0RZ
z9j7aszOTWG9B-ExsMaT1aSS;8hRO!FQLlMC2$W9o=0@nFhdSG-X(N{`?2*n>34(vg
zoTu84K8F@Uc$m`2DkqBew;Lu(R-{Y%75<%*%)#f{MqaH(xchp_>10$%Ok7+RUA{4X
zcyRF3T7M$S7k_^wewX7SBZeBseF{3dV9@hZWh(&J-ZlPhmhR<t7|qAatE=W7&#Dsf
zAiXp%rH6|$8J$;)wC+s!sJ`fqR(5NXkabCfWoDxPHUYWx@MAX*g8uZU6Dc_Sq_i&O
zu##m1ZNq&y+QJq=gflFLKgmNy98W2uY50zYOJDjsI2ASgw#q?{Q6VKcrkP~UnQGc;
zuX1<VX5anNp82iWfcwh!GVGnd=$jxKSh%bteH1eO22vq{&LGj}qb()94D6dDeyHAp
zk<w!}w-+<X_Ntglk(M)N^6}0ev-rBG?GA=WGr@m0^DaZ{;6Y(Y2`d0HB1W7(1_DEF
zSGyHPWm7Zv;j{Lzo7P!{1@zjg{{;QUA!lcg1X@8uC5F-@E*oS)Z8+e@?W+pxqy1(X
ze$h#wL8pJ`hy#-~|FsOgGZ3Mf_qbIIZy|Ma!goh%c+jQd#7<W+U>^S64mQVq{wj`!
zWHg3veOxE#C;K+rV&qx93~~!;qLRVHR*<s3qM?uFtuJ@<r|i+-yK~Xn+;Ik*XwCZD
zchtdx%FlT!@gN3An?H?LXpJImp6Wy&f*X{pZ+hPJUJwyq>4(Wee-dRI^n7o5`FGG}
zbH9@$3vCQH8;);*OLenZ*v)B)ME!nl_}6;gF(U`zB<^7lOOVQc_lU~-0pptnj~g3R
z#@6=sLlE~@sbLPA!@i95=P%E$K#qn<Ol&qp3<?nc{Cs-Q=#@1fd;-psyWIR-WzwnM
zpoyLU3e-*W!eWP?sJ0L*ZV|1$puX{|f0E4;<AP3+=*fTZ&02K(#Vadnb?Qn1SL{>r
z1*KO3n?a7UeH8{@hpQQ#m<ple=N#n@;$br~lv=jOyYB}rzV`W{acZ|Rp9?%dqBNxF
zQvsyRx}f0X1pJ<UK}G7eL9$!&Zfa5p$!Q`U$_r*I&)2k2K}F#yvZ)!))cs&@#cFNr
zwaES^Z3&F^bq8@~(6@G?UcDA2e}jafk0mH~eBcL%R6v4CxwT_IqVwNr&=3bP{&wj3
zC6l%R+Clea!o2&Lx~DGa#CGN>mjJ4dU9L3V`aP(R%j%2cygO%wI&eqOV|c6GC2E!t
zxyUH^{8>qiKmE`7%*3QWP%N2n-Ii_P@yO?o!DcEa*^K7CS_--b#fvgDg}>G>ro$l0
z7&ooO{f%0+OiFnEQdsjAKbj^fLp;|s+3@kNkr@3F&E?~vStiDH8xJL-Ra)gN>^R%f
ztv^9nmTuy2w9<5RN#{xkptL5qGX({g?YNw1PP57`U?NU#=XxW{F+E1F-L9<l6;Fjn
zCBv3=G3ax?IlumD>fbL{qub&p+;7fXV0H|W3qqXZUx@R!66fW*=NviOE_DCh7+)C!
z1dV&Wq#~PDae(7#4<X_V2ncvnIKI5REDTaoYdbpaeW*$n==SM>PWC;yt=t?%!@QPD
zMn;BMz^$_JDwsR^-Ig5>V7*B8k4)L|u+GIOeM0nxPn1%+ej}qjJ1H@$#k6u;9~4M#
zH_g~BLR;a5hd@h>GdPTP8>+uw&VGESIh0QbH)X#onk(f#QE`KZzIPeI+Hpt5d`ltX
zjf(_{;$0K5r+kdApBdk#zBYLTA9H*s+gIJ^1TAtjqwj0&pK5Z|-!bY~-UbPfKT}n8
z%^5>YSLo2N+u{!J>*I|+=UU7<r6pR!ZhvbWvZ&M2HsU{MV7Sgek~I>&f7g`FES5ea
zDUTa)#8g6RWy;C*z4hc-^b;ezi2Xex%0eoLD8JWjM@7NY5~K)`Hc)Nr?)Z~LDN3aP
zg$LCyagFMaEPYbjzjn81F?HJh^culAy1naf$k;1Ci*%{en45kq(-dj!d4>yeq<i|<
z?o=5{?fm2<cZIL8gG0G;2N<JZ0fNx|(1fCc3k>P@>&`U7by-+h#l*yzFYuwUsKi~p
zUF<y`$LXM=q5|IO?tnn@!+Df$kuurSH}oH*f;%C`sI8M+_lgeV`CV(cW0&VH<{S>_
z&Ucy|XD65`9!Cs+EDMfnI0-}a^fR^h7ZS3V6yla!Mdv?p^t$iJi=*DG+AZm%{aVsc
z=pygAwq*$M3E<Li@fz?=|8)*suWPf>`07!HII||h*G8To%cef-xjA@RVO+Axp32h6
zu&5XMJ@gq)yhfHV<31%#X`XED-9RT*TXXg5o7_>{DyAaLr8y-Agt;Y9cL!n1^(XQ7
z0QmHMXU@XHp#K{dXs3{ETK2ift7@I9OQF^guu`)V60ybohPV~8A$cyY>KL3Bt<Kn3
zy{3+Sqtdx}wB>wGo+!R#A>49S!Vp~k(6@}1!+%o;AI!j-)JD@MD)~{M=_?1MnM&}?
zP=G!dXc8dI8L|;x0D@I<Nl8oB=3F9zsj2B&#xI<JA)|L$BR_|XKIQi4&Azn`8{9sr
zN;U%mQ8PcHp?Sg(k!I=L<%NE<1n%<q^O(mhBO^f6q`lC9U&YZm1|8!xE3k&Z{ms6$
z2fTe`2tG=$N3fjE`RFgXr8}h%kynOccP*ZYKyt$k<A51U&(L6DZ&mn21<eK$$=Iz^
zCE?3eP~y<(bK40PM+LQlp@QZg6MaQf=LU|dD`#qm2-l54{=k|M#;2B36^@OSb*x;@
zxm-rEUhH_naL$!#Mu3WV?847c&he{v>)_=c!uy+0V|$*?su^VuJYU7!_xXw-XCkT%
zl$7hey5prRn4F%TR~SG7LNv~WwC@m<w2^DU=$Idvn~#rAjCw@2LN4wrkzguVeW2Vz
z8RUTI7{-fI%fW%OrX&>8XvDUL>%>mjr+y-;ET+cg9$>i^Lb=n~U+XxLBUTgD8DhE!
znr_c%1)!jgB#%VTfsf5iLz6sRDl*eoq}m`6wPI1GW{bQ0uKg&K^5Nj#ZHM?V;@!^I
zPxY~!Nw$=ybM-O^6Tif2;y$#F+7i!{L7q6t4Q*fUKBn9i0YK_3e~|B(@r3v!S2*ff
zyv&#SD0ndylTpjBbvL_5wEHi=`X}Q_kuaodihE-x{(hX8ZIgsD8vlm@#^**f`U$y~
z@_Qu0WiMZ-=5m~@{O<>e8vYuE9b;pP$hDDgWEym<c`uc__frSA?2OFJiXzXR;y^0a
z2y@5B$1@KK6BC~U&hKit;@6qnA4)!$<VxBao`5xq{Ar<gS`{H45PA?%@!=Y?l08(6
zq_sf%Yy7@0EHnshOBsLr<sskF!`<1Yad<YG592LW@)J0sV^R1X8XO_sHPG@d_jdN0
zQqR<8bS`2akB;H~htm*LP!T~Pl4h!yQ4fZcCwRGL8QgH2jo1#SF^<j+=A1I<-_VP0
zY^Pygy6r?J%fzAgr#JU=v6QzT-bBlBS8j7s^l3P{uH{9R)$L>H>pBwGei}KB{WYxv
zZ!~@)gQ9Oh7%B3x&Z3XIo&(#VwJA*?6G8k>^Wtl>=eyan#cw@@w;U~@rmFmo2wTpw
zM$<>Pp?zP+XG-z^Hc8GNf3(sx&3e(wpVo}+WI2RaVq<K5hC<d_XE3rgpv-f{p7DXJ
z8u$1}p?rTO&vVS8dtB}o(<YpjGK_Wef}vN#8Z#`%V@zyLoX25%pg_>$%j)=xbn8vS
z145Zo1fDZu`5hQYlOP`jlzDg}_S&b9-D{7|bnS~}kfGff?Et~v+Qy5bhqX7&H@zy>
za7vtovJ}S78gsX$`<M7g4(4A@WMJt}eAk`m^Mc;#>Pm!lAlIv8@|8Jn!DlpHxj*8Q
zM<bS`QYj4lV)l^U-=V#|ej(N^<f>HLw3CqKHdKM6f`>v4p%kJ;XGmY1{)Ko;zZKyD
z-N()V*G^ZwC>35`*{c^D2;c~4vXMu8^g;X9Y@+CAV<BkH!!7LWB)G=+&fy5jKcd9A
z>HA$(NbB1YM%4VK;ks@1?y%TDHe>;waVhOp46}BMa*)q#P`i|JU`T$VP1V&_z%bJJ
z9#4XJ)7Ti4G-Sz0W3#acaZtCZyQ1}4r1#FyTNLNRA;pZ~^SD^CE<G)w;CFU?%S20b
zxP+7<+r)WC!x}0fZ0KXW_)S0M=HU&jK%+*|hA!kwl)VxDjQHkH-t|=3TE-CmgCYTB
zUP{;MQ1|pf?Nf%(8NtlLr_hx12m00hnGnhI9Tqr<pMo#6`zcEKePZ~x)<Wj=u6R-5
z()VE?b~?0r)-~BD66ci&%MMFXNkvO0vEpw$=Q`KTy;$zJ{63K^g+K<!>E_1>Lqw>x
zM_oPQi~A=$gWdTnywkt5+|&a)Utg~X=s-^1c$ddCHG2C()ts8kv9To!+fby}>cUnV
zG_b<{K=55-_jZbx!jCyEt?qImf5UCNT7l*FtMLc+DaEE;&r0Q5R$e%rU{hjPQ0mU=
zAqq~Etz1QiGoD$<;^Gd&%k?*dO6`5*lB48EOUL@6JTe@0H2M&y!>cV`jd;?KlKwlS
zv}EW9dWXjPd?IuA@~ukBp{2D$+s`<A2<sJWHr|qnsS5rzo`4#3_ld;RHCzzmxnGyz
zV%_DZ(Sq}3HF5v5{<k@y)(@{(ryD5MXbv`?pbT%G03zn%#tId`lkfe>wW_Bz_b%WF
z1$3`P&$@ECpTx<sp`J3h!@088su6t&tG|N3QD+e@twr5Ch5O-;su)<^Q`iKpuv(Gj
z!=5*fXe}A*ujL@mxvv=`&W&eu>gHXW5-Fux(Bw0dF#T-vN<-AhL2JCCRa#}((52_8
zOWK?-RIK0BP*9FBTO#=Egz&T!VPB7_`Sdfs`vY_envx&eT-IBMIAi{M^iIh(YrA{>
zmZbImg;_|~I1d?<y&it`9ayV_Vi8IhnoE!HOHg|WPlSpG5a2>o`f|D~hknWkXHcF>
zP%c))clWM&rEhDu*~=IDHucdV+Ser$Frsqay!-p$KA1S)C6%zjz4;TTen&+leHmHr
zze69Q{xh2vD@%vli6J30giwX9!L><r04+@>1Wu$k)_?EA(_rpwI_S&T@W)tX<3~Cr
z3!hokdf*zz+Wjo&U4BrKRyItWH(Zy9PBLXE?b7GZzUYiS!h89+#wIe;v+QG-er8?P
z047dyO7BD>cKUS<4EdT6fyhFLRmo`K0ZF;IH4Do#bc(mKQU?`3kR*Shh<?|aa`u?4
z^M_aGQQdXh`@>emakUN&EN6!qJ|%QeiA_`+MW3qnF65ucArnwlR@+yj!Eb6A_0_iw
zT`rb-F5yP(h-X;t>(*<E(x5ZB4>e!+6z#;>BoLx6m^}Q-QyGCM`s1zXo3tUORY=(R
zYO~4NfDh&0t$Gxryx_i_UqScVC{|ImGmjMI$>YabEq-YFkP9lA50$h?T6n%_D9!sU
zj=eR6?Ug54vgg~KL*hyc2u{_aoFtonJn_~%+S`H3OMM)?66;||NlDpH&8$#5XIo+h
zbo93N(an<`E{6vRY=gCr=urcQyB;#6AB>0V0`W=<ew!T!ffPF2@h0I--CBOdO=Guf
zL4M>vG#QxJQ0d(+$IIQBCXqHx|F8(nL-Ga9(V#?IgWnd?R3(r`k7vpHLfD4e@^vjA
z0m(I{B|2e1e2Hy#*s_*khtjf!_Meuz3!_mCGj2Pd9d+xA_9`wO7Fbiir1Ki}*7-%8
zoz&lGA`}ofkoRC8fU>%4xoCYbUpA~9+kGDy`3d=@4VyF(ml4CCbMmx5h@Fkk;o?3R
ziRgke<973z#~@x3m&b;)&$$|^5O%GuFq6rN;e7VAYiu(mv}E{bf|%XbLxlMvC}eeW
zt1n@B+@ab%EsH)l8!gSWX;MC*)8}3OutsXmb2e)QgN(T@nTK<yt+OA<_-RonU9W6-
z{g)shTMw?AujiRP8H+$towi{m?(>g|4jdi|bw(ukAA+jw>4wJ}KHj)oUpCdG>*-+<
zI>bS`I$|HhGU958JnIGQOlBVQ5>-uz|JSx*a~^-;gL#&69wD#4+~CV9x6&FQ!JuRR
zOZ;`^M^mrqQVLpHlux)btnY;af5$-h3{lqv88M_is0Z&8f2d@XUQ=xf*>JC8Cx5q5
z!g1PPS>F_E=zngTi~>@?iODD}nNnFBd%JWc(Dx4mCwYIozleKX#tELuIUBkHH_q;?
z&_$g}uDU=In1HV^JJTCxAI6FwLX0jrFoc44!O24yDgORVnMr8@-n(tc<!#0mw53d*
zMKAoj|GL#iM?PQ2AQy>M`PeynBznAOw0iyZf4TP=PYH6Zg)A!gow-)_cC9UKzCOR$
zItOW^YvZ?Uc;MLUa+RrST%}D!D2~Q%yQ3S-mkfl2Jaw=6_z8|CUe)p}FegY1dXG=f
zq9O6BvJ@J6jio@x5*(xX$>CuYo6K}C2020JZWUEuhDk1C>jedor6VWtF1_F@#s3xg
z&sd2#k;#0D5Wyo@O^>CpHaERH^75rY=$nf?etIxVDLtfh{$Wn@;5p~BXOT(8(pN2}
zp55akw}TUrojy*P9wlCWuSBp29jYfPISA4CRiH92Nq%D1v>RWn$00~Xm-^RD=Bl(r
zuFHKWb}z7wMRJqF{vI(V2=ub|-re=j4-7j64#oLTfnPGYl28rho*W-#DAc_OT6HM=
zd~+s65_cZTVtVEweO`^=HuWee7v9}HqHx`$v6iAJk9=lbOizt(c0R6mpvD1@2!aM5
zo>3DCzKgkKol1sNVl^WC9FGWpOMzG%?0~i<*HS~6ATCocsm;_J#tG+?y6ukJEA_v_
zpj6iT2qwcaVGdhwa#N~_>#VTGMBhU5!w8*&EqKxJPS6Bv#ulS0mXBZJTGgcO5CIxY
z`~}~Z+%Y;lI#Zc|S%6Rl1fv=7>&=L&;M^tRpt6f5!w_ifn7{jQp8a^@DO#QQqz;(^
zQc%GI*kXyWjPiwHx=gOK<pgUjlVQ@~Ozb!|vG|4Z!4y?xf>aISiF2iU&2KUJzn?FF
zxLT4P?<`@hIrK-XdX<=p8H<C|Y2{_LT&}u1;j!6QFzzwUqK15w^$-g&g=W&g#9g!Y
z`Y8toVa8*8mB6005>(iMb7kDyiYR*g;a$()62}Y?8KW$A?Phj7{J{kZZq{eGID-Rt
zBBQTWb5~;tN;=1I>H`3gp~Xw7Jw+yWv4qx&D|<!EU}Lexsxn+`9(~ujRM-FezN(+!
z$xi=GUtDvB9Ta1&;mi>EXV2-g4H~bi>RQ9~_feFwzfLQ4iDd(gSf$K+p!(xd$B?F#
z(}`5U4zEP7FVVY%%HBzO(F@Svbel70F`fS`zI5{0G>gusD^Fu#bkokc9$cCTHGdH_
z<>M*ZZvA~ypO)_JHoO^ztQI2eBEfJT505+#!|#ynm%Mkl<}<xz8%9R!1=wa1rYWSQ
z4$C%r*b<y~GAn}?^r}pT#~%g>77|OuCD6>71=CQBVO!K(QIg9l?8@B6*XnYpqyMJB
zWx&#I`<A6!MuSrwyIC4J@0vvywb*{2pC}orHfUVDru&!UB?h<3*Sqe?aWphQfu$3A
z8w>Eywwi}=VG^q61vtGXsNUC&eNl1^yM{9JXHVumW~Q>0sOnr!n)~}hfuNvxqOfz3
zOAsPJJeI1M@12RHWq(BW#Z0wvVW^~(Iz0`JIiijQP2>Hc!D*tY8t>$XTsdTKQ4Hl4
zlbt4TkVH0S(wPON-H8fm3dSG-Uz-Y`^VQuBBkazb8J1n$aee(Yqv5VArh|mq=Dg>I
zq^)fzJX+xD!S11H(Pri$tI2leb?^@G`&n{YN^W~*JA3&mMVE<U^42yiWL>}I`uaj<
zllzyxVh9@C#g;H_J+eEIzr^RXg2zzRM?4In<N5~I(DBBNBlPst?Ft$LiECj{ne^Mg
zRxAH0wrkUk--)6)IyI&IS#TlBN#YT@O&PqUtC0+XA7Z29c-JlW_t7*w*!fI2oAW!B
zugYGExqWg+xp~|XNV$0TxnRtG7t&sqvX=Rr@-02l>7Ms~0j(wi%*tN-P|n-AYCdgZ
zQocUgQzBdj=+8NznQ#exvpwL&&a6j~cU)-A=ury-1Qo}!m;QoyGe0;^ykcs1h;F%@
z9@pO}UC_K?cjV0EJHlr4*UG^DADG~D*E>W+*CkZHb717hZi#^QZ6Q%i@ulHY$M&ZF
zN}&#njp111SGS_xU%uG%%Km#&(0&UqD})$j0&DVT7wRq-KKIU#gc9N_qipfC^rGY?
zAG0TU3Y@cND+`FBXu~T)hMH+2XvTdYxZw`|GC&)v7xWz&FDNFn6ReR+S`ks>%ibh7
zcXIhu=wg5AU?#fnU~X-1SL#r$_{`zGa%r%ELBE<J{6}>#EX^v^pf7Jk8Kf)XfQYi{
z2D*BbK9g35K~lcfuNXpu^XT08_>4Zi`cnUVYADy7DB(GAZ%-ld#Pk%4$KZ1C*0ev!
z3v`Y=y%WMJ-`WYiWRMCycXb{rbCx67{u3_3Zu*&5{cqkS)8uO|(CMq5^`cQm`9)Rb
z(q3Y<lJAC@T~Fb`)m1M3!N_Z<QerB7Hn-ez-eMP<1`WJGMvYpdU2HJu6f#yTp$IV@
z(;NQwc1&uu0GlAW1V1r4@uJB_=UqExgh_>gXPGhX=jrR^d2*1oFrPOSpn4SLmdSyZ
z7x?{=kwcxK&fAf@(Wc%dPn4SZH<6}wt2}m|t~(~B&w2OJ$LYunL&L)`7_1Qt_W>Bu
z47qB`2H;Qlu3;^xL>%`>^VOQWKiP70YL(vvZ_m9uJ3nlJWeF*7y`5Ggm0-umbW8@I
zuV3kO@Dj)e0le2e3J)wzXazu<3!sO+Ih2OvQkMkegkS*EVY|=>4_dhZ(9clkdS-AA
zTVykV{!nSo0F{FSfS$EI*-`+wn!{NDn{QXrbZ-M)W1EiS%@Kb*O}Fs1z9Yd?kf#Ec
zL9x-$OCe$5q3sj9or&=9SE4Ld3vd2V<k9Pj_kQQoSnb<F&;SRJ$~QSVMMG5m(xslr
z^IblR#CTshOrVaNnMODFzr=VCq6pU>4`?9F&wDa7)mQq$1P3)6os|KzeDhNku2W)S
z;^ufEDS(#TZ)PY^M@iyiww&WDRk@#LblcVp8R-EmL~kN5=fExy1M%_oJvitf1#}RS
zsdDpp#!RU!;oQ;RK&S_R_2btC?q`aH>z+s?S_4V^&U?Jhhp!<@=_I|eJmuE5wpH2}
zcw={CT1-q9GhARTQS$mJ;Qp6bEehc>sC|%^#|FyXRcHDDNtdg{2)0Iq8zzHgHTR%)
z@XLAtIl!8x?UFpRoo?^ow_kUeS@rz@zVI?%kjKJcweA&%>ZU5+>1)1PQbbZeN(|a*
z#$+D_EC@3?(J|SG+)tA%+2_AQ>Fxj~8xhh^vW8%?k4`wc>$JMD8m)<_)#QbMFqfQ=
z&<ofuR`&K8>NT|}O?}CN$pE@K3?$A@_huzTL=X`Hf+hVmE~AE|g98T!9}F%#JKL88
z0`Mokvl-iWA|g6W_)LT$Ko{+39MXh}!Eq6CTJ(aBDp0~}*JAn>3!LCcmh({`r%YsO
z;(IWVJz5{=0L~wWR`FXzgin*!3trxM7|e>|V_byg8u*WOCjm^VzP^5Ex&k9#F-R(F
z8T=hUVBbR%lK`UY>AKd}7+-#ku@`coAO^|P<ZmPVRLo`~hCa#~Y?+`H9rEM_d6{{$
zbdDnCtd?2@bos&9R!sq83g)eSn<CRR^xBM#P-QgNwCw2L?Gkze0<L~qC;<xf!^(*-
zVf>q%vV#VC+S^8!9ya8Ez_JD~H$C6XO8#8$#@*lVJfIKQEpWA`Q>M<yDd0eAZ!fo<
zoSh+S@Ck|cKig)2eN@53+)wK?1|%5kTETTlwmjXZ2Ow6xi^J86z%nEZl1EyAfa9#f
zJTW~z@=NBG!9bGwlC30w_5m;}-2GSyNQ<w`8Je0(fM9vP@<aM?$M$ILV-$mWT)af%
za}>@Bqam+**2L~^-{r}6mDS<{2uw2D*0ldS9M=2rAy1<)n=@D?ru$*Axg0;niqNC>
z=L~lz1}?ilyLlwEKSQ#Cgc$L1--oq`9rwHa%y{V&5(WlEe#(EfCQZ$;PeN?A1#V1F
z%(GYY^{~kT{ic7I7dTi4DYQ}p0egqzyV@ynwzz|Ulsz*(s$>Eev$XY@RG8o`7aULq
zqB=g`Z@BaDZ0lCNv6W8XlJP2`ngU!7_>Jib%cLa>834pbZ2U$he2A@Md2oEZGODDB
zpau9OtGFda9X*7S9%IP@?u2Q@0k^&sCd27p+jK$Nj_K%fa{zyBOM|+mu4u5!Vda?(
zoix^bkzQAL^PCc29R=|mpfas^5E#<oVeQWzc5^!ad!}~+=n6KM>6u%-7*VOmGWL10
zhaLq{hXCo%3RL&O_LoT22!u)?vT3pX<*uKij=jZUI9;svsGk?1ogF_S7Y7L9sBaW+
zkLD^Z>;mY){k)(Sh_RVXKt-Owf88^!u5WH6>WU`8S8#_3s@4`bQCG*CavL!K*>woa
z$OgDen;_+hWqmXXJSWw1Gi9wbJ=@6lL!W`KuN=`$<oVP$8xqs{+DHQ-lk?Hl(`)S|
zU5sQ7&o09SdF1*el4Ij&DEz^H&vRjO)T4~Itbz!Y+H>-Mih{Q=&|cs9QWxZ{|DDzs
zP;wU6*E_&G|7&Py`}8!?7a$0W_2ug`%C%`q8XDwJft3}=Rld|FSWLXUH4iM$zzYD_
zI>{uy*hpF>z#+rE2MyfwYyi{5#UiQg>({UFoh4sXtO58D?w3lL$xmGidKfMdkt$@t
zjN)3G_tpibD7&w5SmJ=}hz7cS!>pj&*CCQt?7$-wU}Lqg!<|vNZHywf9i3O{8dOI(
zmyf}i^nymf^VK@zVCL=YQCI%IGAU9<D_Uenr3mYkoN9Me(*E_6Qp7C>Q&h#uYYl*j
z({6q==d_Fl46r4JSr<4VsA`u0IqqU>@h2VK*@(0tGvH{IgsS5Mo&I}BE!Y&r4Tff5
z#D7mq0uWIqAg>2dbpYBWRonqPJbYk*=OVGB`13?jru+qXOy{L&(cYR2{)aI`-e+JX
zkXo(N(2^#I$M2M{^iHI^+wV+b#?pFbas3ciLdbnEq5Du>6i1lvZI!Kv5K7*1!8YQB
z@0lEV!cBN$VP3T&Zru5e3=WJ=;H=hq2>`;q>hAUmCBb?M?2TABeQBoO1!6Y{lq?3l
z_+VJj4TMBNTm<B~{C8Is7WsSy+vA0d2h2cJ2ytd_UiBLQtUesfQJ~FjE2a2h2!jw2
z5mi_)fLWxfrRy6UjxVMy{PJ8d$wb;xK#Nja9uW5L>g;YGy4|bd6aAaG1$2w#%h=vY
zxL@3~kZLvI#LVKGiF3EuxY8wv|1^27Nhml|AbCjlL{L!hUgi>@6uF^WzInGJG&q1`
zWwdg$vSNaxRm@wMb)3(zfM%;#BWY%<?H>Z3Sna!IQh=9$0~fsmtmH8Q3vr6CfFAL~
zPyLKw?S-bRKZq*;PJl!eR1x`3x%KZ8+~4c7RPXiXu3?j(WvZKZ3pGCy2~`xyUeU(4
ztPiL%y!~hMK4K~un}L?gaowq%M+}}nvA6jW#XTWuMo;ng`;N&%`NbApX3aiN;Chqz
z<L`Uy6+vltt9yI>U@aURSSg1fcys&=@YAZ!3=Iryz`FndzPCn3M%lKp0`5Gyiup^x
zrEb`DzjOfc4-n>=0nhEFEiKhJpbB7qaoC+Y=wWR5mXw4}b>ms6QP(X>>K>8h3wb9e
zlLD8FkOCx-0G;81chi?4qZME|Ahp5DSCjzX3IrxRz8+Eq#f-iF+v}C58MkEbEVS@M
zcm^`TR@}J?vA=KTv5Wv)IS1;k4V?s%)GKod)YnFfebwDlM83@U+O}+Twa7dV`6&h7
zCNhNdFm4LX#IaX9K?gc;fYjrcmZtl2dERB4UPnhKhSd-q(4QFWW&nj#VM`6979A1P
za>#XpR8+qPC)*I5*ApEp9d-#Nq8V1dQYKMZ_2xGp(BE@1fc5RnHzeQ3A)@(ov1j0-
z$t@9g(`1%)qTZ)uk>VyInZFl}Jfj;E@Tmz1)}W53f0I?An7;gK$`&Q*RA53}F^>mG
zeBUcY5>%^zko@?aVN+&2Y*_64nuW$Yd``}P@Zv%%cMS>(3X(`#S_d~dkhl;?O8^e$
z;^LxY43j5t$zb4mpQ8YoF%VqBfzurXNvG`kZ>IyDd!{R3Sc1h)?;wDS-P@A_`Is`G
zm4F<Mi<dW=D+#Q6?D|fAuY_`svc<&wB9itK{GYY3W)#3sWzy^c$qN!lL6vJ9=z*{%
z_@U*6g+{RRs}t-W<O0M;0Jh8EDf}H^Oux$Et&}^uCbi5UE|V;B|Dx6o9USF2o|dj5
z^?ZdEm-jSsW`|i=CtX(XU8Q`nu<%4(6mjcGNZ5X1HMZ<d%Lq2kU(ft&f&EJ7Gj%<>
zj;omDzT-sqsCa%e_2Dy7_ROZv!&sNWS&R9FdA@H870Kx$@h@og)ryj#+p0fT_o~nU
zkvfAgCBEdwOur6%^L-f2b{(2ifztDs%%@7VL5@Kln)QE3)uPYnnXP{lNfufdm{8x|
zURg%bDK~?}@9_AR%Hkdtm74eGXn;D&FTV2kUSdBrv)6TQw2_mL2@RRxZSVC-LwIbY
z`a;ZsCVhr(nx7lr5_dI+rHM!Azt=4<bXpMlsNn{B{alo<b#orW+M`2ksc_~sI|TtR
z9;KT5O!jC!Jeq6y>moVfPqD=U{+No26bN$>5!ir4(&hOOsrzLuaQWK5Gs{dDHOpwQ
z)^!#QaL*^=aj95P7&2w1$Nl}-9|S(HKRYHMxwGOKdh%5(*RMchupiCsdE;9&YZb2e
z2Q|ykq~jXhETyE*j&9g}p95rhBR1H7KV$0m+@<Ze9g7%4)=pix@XwYs^bSaBK?MTJ
zl~r6v0Hy8$0(YNsElIaPTB8SChmp-(Q792-D4@6%Tg>uEWZh$USANQ7<ByfZANJ)u
z9(m~Q`1}jNZvLcM8@UBlWGZy!(<N?Q;i_<6PA-{^R9E}+Jr0;3B=T#$`}le60mYy9
zt0pa8Xha0(w`_JUb(_xj?(Y8M{TigxSeOM{g8+^jAqCt-koty!9W$%Aw(v=AhhjoP
zLic4EfN|5Qlm>u^KGXm~|9zsPwUiEPTWdeTv_<HR&gR3Nk6-ecofNHCPdbA3M(x(7
z8u(5QBujao2Nxhb^G?Z8U*!Fz0Q<xqtjXpLhr~h3)!9l;zApj7*S~|ZDR%O7YzxGx
zng*p$zud%4AqepY7wgGCVq?X6qUgn>q{{7h!1ulr0GT44lQSy*TlO11V#c;@Xkh5I
zJS)Q&EZlk)+=`cqLF|Mfwa1fq=>~d@m=M70k(=+X8e)tPrrH!Qme_A|Tb-(P+5US$
z0N7y+Fp4$c1ZH+YDS*wj*6;hrB40Kd6cFQ@kojL=f4~E@%~LqJWM^buKN5;WnJIQX
zLgaZeVGPhAS}B3>S&wV)8h-K(oz%Y!D_(_R-E!j~vJ)>(E?ph3$<4wqqp-s}t#9MB
zq2m61Snx{;l-<J1pt~DR$LY2<a>uLwTOW}$<}+7D#=@Ct=;lO8!*!QkBI~|wjlEHK
z$1i-6qfz^NLBekS7G*naK|tC_dDVf)H^z)B3KhC*75<K^N`!{eQP&I8Xk#3P&5kuh
zh8SMI)m_S|t;=c3*}it);TqS~eez3Xw3O?=4?#xrItcH+KJ5bAD;;Y<5CN2Dgcm@v
zq~hGwaM2L}(xasA12)xgr@!k6=uOPs7?3yY%snKsq|v+sHVK-!H*V+94Aa(y5+g@J
zlabz0leboyS13s)<u94Hc+%(Wfr98HYrX<?l41Y%s++;9<^<b}Dl56*KrL@&Wd)Z4
zppI1tbLZ#hGY>2QiEgjnq^yniucvD>!NhC~lcRVR`K@v728r8c@g-JqG2A2hToKSP
z<^wo-NPuLIJd^+|8jv+1e8l4^w|KK55tVWn=S;Tb=WP9-g$c=X;WEM;_GV(i(yWJ=
zczFJRs|E-B`C7%i$3rbZ=>Y8t)Gx0@Ls#~u>)l=d*?B}vp+C|q7o>;X{t@?b>J)8(
zQXk|w_{{pG=_p0y-MlV}eV~6PEF@2bo6Ir|7DmqOf@r*IA$ZXGtm^)21284K21PyC
z%OrSq3bJz*p6B3H!~X<JN~u?MpXzoz;dl1~y)^v2&zQ0S!!{1`1zKdnmL5fR<bJt7
z(!^>+$zR_mu(y16F~W=MU!^431ph0QKtTc_1Pv_g4rfU0odS#f2;h`(u+vk6tv}gr
z$3vp&Sy}6I9uk0D0hlpG86#Li0YU`9A|7*9@s{P+ZK2DQgIkBVOK$vpPNi~4?fi4e
zi16@=-Lf;eVCEr{Est-h=ZVE+9l{RxGIH*8bpLZqDc9F*W4GIFm-|U%5v>w*G&|4o
zi^02ku4AHuq-|A^`G4@vR{oA91Pw>Wvj=#}56>68F@7nRP=Fft6+J!u*(ug55hm!?
zjg64i6&soC-IK&xH{ZpQb(@`L@>N7UqF?)n8V=v|zx<L+?WZi#J0hjK(?wwvIAkJw
zCAy$xXgJ&2#@{kG7KIL>5f$zCDsv?xo+N(6aCj4Ief8!jlbo=A=S(oR^4RpLoLnSP
z4(}Ut(N*W9Jnx=O;u@=PT|5RB%iY3W*Uk4}7C(A?{XaAL90K?QpPU+mBE{QK7;78B
zHokz+PzqjNB2rS)A6G=cc~W~!*CZPn_<T!>&A^_$g_Itu)OhwmnV%_)L5}2?{8G76
z8|<nbINC;%J8)pU5K>LG*&eU@A-kYXitubJ4tGVCC*ec;bW0+72q7-Fyd@KC%IiJd
zR3(8t&4vORqJn?io$;>LBhY8xdaUEndeBQrmQZOIYus|@FRtG&ulvt`(=B_a;N>NH
zL6~R3EYvVu(R{__?QPQ#VH7f&DtE*F!<ABL026{Wz&TN*z$DgZ%@(Pl;HJ@q8!Pk#
z{*Gl*Nkh}lj)`d%twlPS#gMJ(_;1M`6xyb!00k7HCXUyr<?~*t@$oYx452iJYWv%o
z`8@f+8jEr;jWtrtSn#At34K<pjzI$D_?cWf8{D3=k`yomlO8*5{xABWUb{?85kzR*
zag6o9kypgk)Z4w^a0e_?GLffD@(kGGuy|MJ!dHwe+P2xDGA$#lwpm{1NdUseT<M(~
zF^oApsnAc#{e?f1kaz)`)b0+R#Aqht$z9<k@@yLi)Ao~xfPY?Wnf%iK&Kicv3SARK
zrinP$<0{VF6guqT8*&KgXBiy+ww!Ril>GlbU)vtI0uOK{L5&jvREi=u1%OXhNiI+p
zm3MnEv;XAbwRlBI9ie~`+^I?Z=R2b4G>fQ@-&30jZaP~~w<n)3l`M`jJ(i&M$dY{6
z3su9JzRF(OX|2Y{#GSlT2CsAM`zjp%{I7>)#3;{IQxF;OHFJDHqi<=rkeM(fV_NfH
zu1)4M3Sti!fI7a)m+Dw=uz(9ePAXLX9b0p!(1%WPCY(ERwCWdvJ0K%3NfQ}Zr~yT;
zd%4oQpBrz{x^xEcvdcScS=|yy+x#`3D|UJN()wzbCTr%%ix(-Kx@P<~5hF^4)buZ^
zU1Fn*EG>U+elVIaM)ytTr1$&JV+Gh4Mv~OtWsI7@7eyvT2?ek`b-6_Snw{OPV>0?2
z0EwrG>cg~d=G=RUxrK1kpOWhs>b896)JT(C7LmL#Kw#ztBXnYj%cOByDwd3-PyxA?
zJ88p()K)YswL03c_yC%&(VX1aHJ|dwOL05LQbUW<&~2G>FE#x%-}>f%VW|7ZC`v7$
z=~Ybw>Qjjh1sJrO&BvB>ix4>;J-C?iZg`0I>g88yv9{d6+>5TS;(!`*;Fhd`>H&kV
zzcHmHf`rbBDR<anl~`>$9$j7>e(*m9?<jHX#7msm^={I(Tf~&}=)2iAx<p~|elPdY
z(*I7Miqinci4pXr3#F4a3I}Ed9QTN~gcPqi@KIY^TfxS~@2;v=Rz=*KAnW+~a(O8d
zExGwBCMu?a|DE1b98yLE4H*-LP;@BReFv*u;Pf%b>wNym510IgIXHIox0gn*c2+v+
zuUW9OCeP1x-^v^uthN2u`Fkv5(MtmZMFdwxozPpbpW$<k6QF&vQ{D%L2*oTp4l}wz
zml=^NM1d<sNixZF*=(4decQ-rEe7&$d!CFouPS<B8xO(XWrs7+e_2I<Ynys;rwRx*
zAZ>ykBhy^0{sLxAD3C<bEu!Wy^ibj@rQd#jjL_#;fS}n3{$^&NeL7}4Es-?+->a}L
zWHLqE2bv3qWkyDGwcG$E{RI@<X+uWHGfEk<iz_QWV5MXMn1VbN3Rq%8$;&rO0QIk^
zyCg<LjyElVQ<Xwfhj5_Q(%xXH1HF`H9b(q^pArXv&afrV0{78m^!F!&|9k6klk|kQ
z5@|y#xK3bOg@u$8Xkx@9<*RBN>I$FUKKd}?ZQDdDfMcq}haY@wK962j72SfOP3F#9
zx}KF-v$)T7gj@@Cz{8B74s_gZ$^TyJJLM9mAU{B{uvu<LvzV#W-W&uOK8up5C~)&Q
z!zZwkuXlp_1X`o^a7p{KJzkSfw39k`99w(xk#_>UQH(PYxRX#tzyDfY{unk!Papq}
zB@Ei&EV^T;b6#L`sxnVHrqJhB*E?6gQC^3iHwO@6sf{<Tf_Qxi*7?rUR=tsXm<L-Q
zgy4-iwzRd;8Ik1=yNnK6{DA^<9vSYFmHyM57X;@9+xfI(Mt8gJ`R_6+YO~=#7A#9{
zapkaGR^j?;&1r3A)6_GaOp#wbTrEMWmJ?%*lYr^yY2%C(9UlJH%#2PYANLE7{T39_
zq3w(LGHgmfz)iq`jnVxSAqS(GSt>sZ+a)NCvpF>vG+8yJG3woRSAR<@2FSW>4%J_K
zz%W(OpL;s?H@luRwUnPfPb*oNwZ%+qd#1j>9J@E$Euj1_%yh>9&joPsc`97KMB|`%
zEz#fAF*DoUI5{^;a`Qk|fcSDEZKH%f`?aaE&FHWZf@QR5V8YVY)%6vu#FhngTQKm6
zH=Hp{c5hD?`*F`)=1Gjrg;2=eF^dSy#DDch(v}=;S}B2zDtusAx=Sj;c6x<6Cc3z|
z)m%tgD`C%gVW%|X4$3OUj~+qbS})ik2XjAx+P*Y>hN1Jkg3M~4=t?(b*zH-4MF6Ce
zFnL|ZKR6pd1M900<!4CDro%sGDsjOAzUM}qM!4>CQ(cY5hR$YNWF&3}qFy|LC^SSI
z`L_v5NoLKlx_6&>?SBAQgaQEZpjk%STAc4JaF)sv+Y}FK5zu-9TqOnvGoUQQYS{m1
z;VVR$lmc1=`r+R}r4}c!qztOqqi!$8u50qA9yL$$bKdAG!AgDUUd{CSc#8q~_AhJp
z50sDgc>f$6mD!vMVfO<R%5z%-7MD4<`uTcc<C>x!LOa+Ol~4f6662{%Ku3BSVXh@V
zFVIVLlv--=&tA@IdxP0B>nB1`pe&3f*)onqfBNPIYo=?%m)=VG9FNLfzn6y%EDSRK
zsKmUamlJg~JWS9u9$D3<)wkrySB!UEb$icM&OWfNAy2Wt;)hD`U3F%6ET0JU%mqL%
z4Uh#^i2fzBK*;rjr_y5?2EQ|x01fkLCihVcw~8Cf;Zw@y+g7(_Xba)|XzYhHQ=Xxt
z8C%ANF^x|o*W`4G&XUvp>%~y?80ddS$$U$aXqh0ApJ_PFac}!mV}*vklK;?mMRaLH
zop_|73(cF}^0_0PNrl&b-ljw5c*L`ARZuIZeEz7PyI^g1LPfI9SjE*r;cQXDZ`@`#
z5XDIZ+@KgRDttItW~v0_hXOC}c4i%&eyrYh97xA;%sw-MN73I`h<YLX=@I4QR4N8c
z43hVs!k$qwSiL|;Pe?{l`z48E4gdS>d)*mT2K{rptdECRX{BD~Kh0%_vuZWR?F)`3
z_y*jeADPq_$Nguguy2!Q3@i)gpt0C=@SR-Z_zxr}T~+-Wj!L$VMmvtp#D{{9PpzlI
zd_1&86#IvjO8p$}d&S&Li+q(|)_UnRJ&(H6ReNw%&ig;NkyDx;tmX3%;uWU<kym(v
zE}Ize_{3~PT-(!gkKq{{9Gq3`Ne?vEjljA>s;F8g(qCs(=~|FE9x5+l6I_0K8YSSo
zE>j2bIBRn&Zn*#&dyc$q&CxA>c#kVgDwZW<!3lH}4q=rrY@on_&;c4Yj`rMeRx)!M
z8e5|rDj4%NX&P1Q4E(lN=7dx0Jk9Fz$ctoTgn!a%lIoW^LY{NHlLjpB>1unj83Z`E
zHyKM-T*tdBc;sS<h|ufgMwo0Emix9Tgp6#NlD~cXryMOvU4025BP**%r<jV0YQxJ_
zo?fl;6K>FZ5R`cLxAC@=r@_K1Wjfr9^z;qDV<P0XNe9C9ltFE&f_K*?3w*hwIbi~q
zAAmI4w}ys>sS3+NFvZdYs$Z8oB?EfJpMeU)vCuLe=ULTJdu!``cQ-H)3j7Yf*M;R5
z%gz6R33jppVW4i}O8GlUEw97QAmCK}0lmbF%gY?ldLd-i@d37?vM~czW6^&udV|4#
z{0-2IUW1nX`}glR!FY8L^uq%aTd36Pz0PKQ>Og&p(9vc-wd5?-&;Z8VO3@D+8}q8F
zR7z|sAb?$@5g>mL-dfIcjgN`(?2!#&VLfZsT*Om03n}H#dGdZ_N%wT<$m|AK*;|ZB
zV*Mmzh(`ji7w(flD9J>@tE&oXw%?H^6?>|g%zV8XCicrl`R6(@Q8yeP>8H=^tRvfw
z&d!uTVaEnW2S7<ie744s@vs&{;QS4c*@FlL+|Hdf8$Jz80nxM7l@%k<GzX8^>Rd5t
z#vvxFq7rw1kYEgmJyQ+_Oy)C{GWT7*%j3<ogoK2bf{B_lGZ#mI^rG70kHHut(S3z|
z2P~V-VdrxQ34e8~uK4#^i}B?!!2&S8Njf_L)vYlY7sl~9|JKpf#euM}WUQutNDC-$
zT!U)Y1f=(IJoY(&td+H%<XX^q5!|>6zYm(AD4}(AbrSaGfK?Qf3$OJKNh;}2{7C(H
z0#0s3RIKub18Gon!Tz8I<E(qM%6OsDn)COOVRr2?V)tULpE^&a@TzC^ShD3tC|oW>
zr|l%NXoGHqap2$}IJm|-j)H52@7SZ|&`-2c@ifKwdX`u|<o22f68B>nKoq;Le`#12
z_-)sMJuX)95D*b_z*A(9K&=5-ge-ai_a7E8%Z}!j1z?R2r%DXN;@FH+Crb=-t(V(v
zxQJCMED|qIcXNRTN5q8(5R{VyE%~DMV1hK0Q7DI+msb@C^bLaIrVylF8u$MuBqUy6
zxs~WRHw3gj^ou_qPZ}m?gDyNWDyqDNMFBdgKpwSRCiWK9ClDY%g4rHGE0pEPXTL0&
z%mXwHdX3u8fZ<mGbO7_=A0j4<i;sO!n@e4svFxnsa<k4S%u_UNvX+;R$nAmPwO%;*
z3l~aXN%i^PsA|%m%NbWq%64FG_QwswsElDavg1tpvAdVnt0rZxdJ!avT*_l8Zf+S|
zwn~LEEo3chr!|{_-0mtSgfGy3k=^8$+i9l%sfi=j)=;UljYbT17OqQj)+iPeTeT{L
zzc<_dKz@%m-{skJ66Kc9A3^d8Zz8G4$rB>jYhYe!hxdc)i}Q0Ma601B(u{Uj05bbp
zER6qj;=O7rZU$&F#j)rI<|q|<XjBqCdzJ>qG~GQt5&$6~0~Ag4IO10~H?wmKs>NxQ
z3h1s9zuS$gJKozHG6qI;m)`fVu&}+FeE~EQzR%yq**(O?A@QcbdC)K21Y&9mpFVw)
zr1$|A75*vG76x=dsnMLodFTR+7zUp^&io7rc<k)r(yhlrO-1$L=JI3&uq-LWF2S5T
zp*neQ<?u@dkQPDcAUCeE+6$Zq!i<&G#Bhk=KjL$1=G-I-J=i~TNc-Darn8kK9*aEN
z^qd-QB!q^)sSNjo9M<1{cuw6ZcgMM}bmDP}kbzq~T*0;97ncUFJNHS}B+gEqeKXf+
zvjum!AuWB{RQ&9nxb$dry*X{&)aRVt+zi!<O)$|J?@tr$*?)U~%y1E6;yMYhBeT!;
z1TP+e!fh>M2^@_miSxVffjiPy;26Ze?=^sp-#{S?jFMgo##TF7Ycwo^&dUEI?ybY3
zYTx})3=9x`4M31kDJ7*FL<J<Jn?Y%YZWv;r2q=iiP|_eh!_Ym1h=6o6(%sF_c^-Jb
zd;hNUJJ<PlZ>||G9M`P%#NGGjx$h?e0%e^mU;)H9)>trrD69YE?k?fpAdA4|cS4Ho
z-i^+>@{c%$0m{~&jcM-#y>nzF#b>1^Hho-w`T%V)#|LL`2T^G~J5P|SsW@AQjV-1`
z%bm>#m5-%Vw8ejj-Hux&6Rn_)Z#Ecq1)Hb5>!5<3u+!Mkl!>bVyHLlMw3hAV3{~UJ
zoo_HWy{NuH;FI+DFET1~N){?|S!U|-uLUsP95v}ZW-2Y-(&v=LJ~u~8FVoy4G_|EA
z80~Jfmq3kIX&w_{6K`!J2ebl8rJ-#`D&U<-644DdfAijj8jL{zCqusgs_e1v3@+u?
zRLeWcssNS`IzpJ6XVO2Vqf7a11o9IyNT(s|(uj($pfb^7Rz`vfKq-;_^UuSe`p4h}
zB)?2IXc;0WYAOkqpcJgqrXY5|3n-3oBofK5d|R$j6QiTvSx3jiqlSZdV9+fW-V)Sc
zhLXUgTZQuP6S{@lQY*X}u+N5d5$G(nI%)A-7&>H5zf8Kj>Gvy9fr(2^^nx(+C(gty
z10tg|-Hs%w=oilO#@SK|2L??!m(F%v?~Xbehon!78(iuYC>gd&ix+!rODn#A6?ULd
z3aXi+fR-``Co}_t4m}7r0IrEarUMc?w2E#l(PEYK7Ukqr?jCz@sr;+1rJ8HdiXE|`
zp}kYyv_@vPdIip~0Tu}bR2c%C;4E@p(g2#8+-f><%y4H|qflyU(+4*;2G;YmiXJ2>
zWk3<pPS7>beYD*bq$az8y6&)SS_d*N3ZcqJ&|u~&z=LQeN76%$Q`@POX`a;AHp3G-
z4wq&u$2`$Rt=WceyryE;!b2nCOAks_H)_fPjgedOm!AHX>xp9RN%vwny>|U<&Pkrd
z=3}ZzC(k7g#o)VGvn&S;1$<zFvec@`nk#y1tLKdz5+u$TWPltdWI?nPIBXB+2O(*f
z8JU>KJ*;J*!8{m|I!Y~6tpWkyLugPL<xgXywD{4wFTe_fhlSB$N7w`E>_<=DO3iHg
zXf`h;hmIj0J$fVoN|gz6oP|z<P7cRT$&Plxxj58ll}D8HQv&8P;N4jqCfBLZL0Oj9
zHG5A~rwFoizfXp#ZcU-oB@>XL@fZ&Blju^2($&s*V0MFh&U%Y+A<V#=W$6Cdya=K!
zX-4!fNW2ruKfcA#xXg+d5GJeSkXh_4tU=4N6d#eLgMN%(=`BF9AtCq8R8aaX9i;v$
zJgb52AzJ|oo`IUJcKekkT_x7KGTm>)kJTnEK?d!y4C3^7*BP9}K!F?y;R6UG1LeNu
zD{#2(#$W*;9zuw%%+>O_1j7zcGsfp8l|T|Gx|3JgOrTVITxn`OT3g(=snHbKSL`lq
z+^$=g^Cy=4xL8tqKqJfO)4gTw9e0FUsC;*2@zTr_i4KFai9FxOZxv*o9(Vki$g^D8
z4QYDEaoo5PVv~TrI|OM$Zoxu}-p3;|N+wOfAbSbOvj+sZLCq>LuZ?<gWNJnN0;M;*
z)RhuIzfgBeTunzuCwb@=jY#H9OH@?LaGLXoGfoESy*F_cd=>o)H$BD#@*t0GAwZ^j
zs&Nk@bbqqQfSyp@AiRxPUbX^FY+i%xe<uie7klj5Kqkh<=Ad&F&?qqXNP?sfL!RSr
zM<pzoVsVL8-f|~y^=&=PP3|u$Q=f>}Gcq!>3)YwCpG}K^-LWSr6s3&~HFRO`J%|Ry
z?6kteoIKHyUXi7*?dR{VHMLBR^8MJirfdkI&6NoQnnHNksTe>QfyePi1Fa71FkCI<
zdv(T6z_@e>G^S#722;?RWefWD?Vl<D)Iz4-q!v0G9ARt`Fvh3_9bN%E%Yi!^m5|Hw
zlc#H(BbDxj%Pv^@1;EP)uxmut=4P)HS&di(flUmY`%J}np(sNkf-84o!^2&b(&kOq
z7-nrFJ==t|TcuePXF62-peA?vtQbk5xa7aTUrBz19i$(gP0r_6rPtBc(V^pZ>#-)h
z@xyL}`)s(-7`Har*!U!egMgD;4;AJ+*DJvR3(9BHg3cADKhsz$?w{!d`UPOkh0>3B
ztW|ApZ?l!|SDzeMg4s!d%v97OE*BcL2AZY+=le^%_LHw5cen5*ovu6X6Mp1QTl30Z
zpz0Q-fZf~Bxf`@%mE&a~vA8%|g#|>NK0iPIVoouJN9OGGZYhF;Fv?p~F-{;E6nZdU
z9WK`?F#i0NT6ng3;)2=>LVb!jYuE8_#;$+r)V5l5B)HwNjy<oDR_?_b9}TR|r&yzZ
zyH5+sJso?jem$hOws;KQKYuW|*%B-JdBkzlT~IqqsM5Z~`vDY8c50jCP|)Lam`Cp1
zL1$Z$wTUOZd^wIBCbs#9Zbr!9?b{@f8_L?4YCN;}xMJJj0|jXSa*00EI3V?jJQ<yC
zjTaFS*#bZ+IB;hug`6@$bijl!7E%W-%SAzYL#M{O3bfuOYUhuZKfrg0^ICM%^V*K;
z(a_LTOa`*>fZ}+dYn}R;v+-ap*s{!+a19_)oPl<Q1Mq;K4TxGLR0FO4VWyM7E{9V6
zSJ6u6yQiZuxRa!86~{CyEc_K&m9p-3UZuVSX`=p);AXvU2kl*8{{m0qU9V-g##GHf
z5{I~(jvRC-73=M$eD;P_OKg&$a)z_jOcDec0_bcBmD&b_d8Riu7CVp0_%*$&fdLk|
zH!K-MM(P1FUx%_h{-vd*X9-}@{pm12o0c(S48Wc+Bpe#zcYwF{x5}E`3EBj8eic>I
z<;&+GcjWZHW=ZPbq~UpwWr3<AD}eJ1{Q8xhYdB()lGDI_CgK1jz~bYHo-_0&E2mWY
z*2FEFdF&kQ1V4ZoCvdm7g}sgVYQoGcqNnZ~w>(FDHl-)zkKWxzEEnf))oxi=SkIgH
zfm79t9$(wz2GOnD`9wE~HF2Z9Thq|rwXLpJYEt%L%xiFJKLgY-1pene_?S;%Z5C-I
zC|$5PRs+>Shx_{rUDCAbz!fNRU4tnAG6Wz>WU7Skl{4-mx^~Sj$B-bRnJqrIH+IaM
zgY8q)mHuxc^5mdxDjrXnIX0X%|AyGfw(rG->T21c?{Vbak;UX|AE27Kr{?D@N09jq
zj<d}1kowXmO4ralVfTcaPJL9&n=6*1sYJ~m)M3p99uyp)u~7Qh5)o0+h=_=^Guj75
zmXWD|?wYLlu;nWE>cOyq9jcB%RHRl??*cr{S@B@Y`Oahja*aD<7=IGHMvpe8UxkhB
zDrDzwj>~&5#B2SSx{39gg`72>!osTn9uR=Gol}%dm|U4jgT_2l!k4~PKQV&Vr>QI&
zWZb{YBg)p=Odgz@9phw-VKUw)5+AXWc|$9R<87RJjkg!5cn$hwHQ{iY{KIy>zzTAj
z3g%zTZ=`W_jeg><dSML|awq<p0V&3##>erl3^#7v5XmKP0HsReqHjKDFb2xI-}fc{
zD-(6-P1A52pUdbi4VXb9E?}6&Hh1!HPmJ3<>BXf>n;)jT{hP{fE(Ck2AaeXiI;9oh
zl98*3Ga4XZ&IBgA^+16nNX)7MwwsxlNM=??Wdft?0GJ|xy+k05HI285gq08YZCC|>
z6mbll1I;&+$KAl??ebTXkc)#q%3<YeHi8A#T%uUHL5*Kx<pJEOr_0~&I66UVFiKB_
z5aI+Hviq5~(&8+)!enCXln(a8)sSL~S>@e9*ffk;FAAm~49(Xf`_BD#9|=%uM3&YI
zR4ZE@fmmYO(m{=^wrFl?5S~bud`CE^_R|BPx<43!UaOBgHg<L*1EfcJKzD<)@!auv
zw;qp3U2Ahloqqjh4Jjr<AK%P=Fda^1(X1RTWfNss*tdpKj1hsq`s_WShhx`C6yN?R
zA}W&lIUse;RmP(`@Ft>gr90QYyQV}o7S!JX!S>_5#g@(84g1yhi|6ZB_DNMin(AxK
z-B4me7Slp)7wkxM7J-@r4P%%X<%Vr>wyt!Cr;ub;O&ro&kb!qSYVknkQGyCeK`7?w
zifkCwO|bb3`-fMS_G^7NHD)3^tEzI8s}}p5PqAwY>kX>i4fme@(uI>f@k2l|8XmB+
z%58br9~KLxkkx}RqY&oI@8YKC7Ck}PCH=GuYEMopHW1ru4iSv4F)0gjQ44=g1T3Hf
z_qmS&{^rl4&|9rZL+ZknI?=NW-9MsTyifAQj<;RC1fgWUl_3N3`aDp+_`t~mov~J?
zKpo4&j`#X^o=-R~LzM)~aj04eN@1%{qn_j5E28!<qz$LQ>BAp+KcDxZG}PM`Uq!&{
zZR}YS5-7WNO@rq>3Jn#R;EfY?fuG-e%PlR1OMYajknq~qVr?ILbiRH*4wD5wyJyI>
z`QpPjAj-lZ^acqWOi|5<%Rk3~i%BUhg|as`F7Hq`E0q0?x<dGbpGi{%c53Hx4aCa8
z4{3bM#ZeHV`N=SMd;Q<R5<G9ZR})Tl{@*}kCkukc;$mW>>dyc9gg~|_%ZFb;;O~Eb
zf=tyND#u>@DPZ135Y?h`fi`MP<pS~l`xg;3v#@%RVB7(O=gubm_c#5(ii>*go%Ho@
z{{TK;WhXhm|NVb1a1iKS6b_;3J3T%^&*s2VrZ%q6jgZ}@M8yg__rByGNh%?EP?q)L
z{PMT2E>{vLeO2M+-kv`7LUEjEJ;c_9?$$pscseY+CQyq~4G7!Xw}s~M(AY^QwWe#&
zZ)sZPh@|Jcr2jtehOZAfL32}RnZUN!g8GVp=PRQ1@YT$x+c}JW?(vYJ?AM~bqOFzd
zZCVw2OS`K*!#J37&tUPes=)d4pWWcozU1TZ4VjShAQ2Me7`CRSSFolV(Jr(rW09$a
zbz}O{I0j#9`Yl48O<E)CM{+_D;hqb`=L|~V|KcYx!OVE_Ek<}p{*=n$KNfnbPs=XT
zeqZO7%jJngX!LFDT^<o#r54;A68d-f@4W}()w9}e*N-W-@>A=3EIRgllPBWHFf8@7
zHw?ZHZ;Hemn5-N)pwB|H1O)eK!Frbc-HV0_@i$k4TbHVIb%0x!S9yY6Ymb}x)qH-{
zbCZ`~B<$hQ)_q68v70XjErEHN)c#K8dz8l7-!#xd2uFJD{7Pohi!cJaWVWlfn7M~y
zO7mmtKSybWrMf&S5T??>T86YNp37>22ssKd;TtjqI+*4^f9zZ-OrY7&u&LEM%r^oy
z`SJ^9W}DxYV#_JZ!bWy>cg1n0VO2Le0*o>yH*doC1Pqj%ZDqPQ_S|OHxjmW&I&AD3
zjUe~J#KNxR5mU{tZen`vn-ItJ^z@%pl`T!_G{;rO%%cP!`T=XcnLI(8v`DYe-P5^h
zGA)e8fK|yRd)Du>1{4F86q(c4+*+RS^J{8rYp=4Wv{pJej_AMeGb$W1p&PH)=8IJ@
z*=N#5oGg8)C$v54$RSW{X;r%^UT9Ni6ygcOS1t*+zJ<r?+K(319q)-=5TvS%D!VbQ
zH+_6segV|bj%ZS|Kl#VJaNwHTd)co(rC-1)pL=9vv#VUqt<db!()skMnp^*b*Bgdr
z3@ooYj7#-Xp&cAIM^oK(S-GMn_LW#3i>!?o5o_S-br@ud=^jrvb}TbQ#mBSBZ0_v1
zI5OBV#kOgz>#c9`v`m`VJ4)JR=*24ZFI$1Aao&eruBhk;^?xGQg()ebqd|#pWE7Ka
zE(<qaauhl>Gd8~G*~||T?=OxYbPo;b($F}~UuNd^{&nG}A0;lZRrqf-v*4lhX1oP5
zm2uzbUvY4B>5PDsjQddVo}}t~v1}^}EQimT3e!X_Q0f?zPCsS}wfQd*F`N?-U#e?v
zrEl)nD6en1UuBT1Ya{3v+iMrk4L|}?;&XDWv8jY9-=63`FkYBbO{0CgV6o6~*^iKL
z@$=9Dw)*+`?UXG3S-<EL<nH0V&|;R*I#gua{mw$dvxx&1@uX+(5?8c*%;xl6qMcoi
zQlWDp?y~_z*mJ>oph(5s4Z}sKkfWjxgT)^*Xa&_gWPqe&JhdnbHNRS6N2$f+`CS*?
zvYP)Df>+K(Q&K_2T7{)L9jm-=V-+TKf1~@y<t%Hv(Nh$F0Xq#cTMN=zzm6<e*~D-3
z_hy`=fQ*B^?Z){~fBtjL?LrcjKG1?5>E0+;v)1lY?-O;4C1jd&lV3Z2yTsfa8XH9b
zl^S@B#QtU4zu)?F_N_QS!YlCMykttpF;SfoXDr|9Q0`eIe$_5hD&kpjj;1EA$b!zm
zHV7;CpV6MJa@42NsHhxEa?H}8Bo+(KCxRa(nY<>-^S=Pw-vjIX@6Uh!JYWC+@5g7?
z{EvJJ{oe!nAA9n@4?dg!|HwrDf947OU;FVtJpHo-#Q)~>IsWDU|Ghy#|GU#aTmS#*
zEeDGoo`EH{OPiiHwuXRv0-O)f+W;FL9-azn=@(lKr~p?$xhGv=ME~)>!y_n@PrL5c
z5hs`eT>RY5vl52I;1DWEb-o4}_TKrf6c9@oQTo^XBp@i$09Pw8hJkA2Kfw7dl=txg
z(e>-00RaKo8fZSNzeyleVF?oT2PcQnn)B|6D;BHwp9k6Lq%G7wxPLzrATBh(z11R^
z)j?Xtge!V_Wd#G08uRt!2I2H=-#K6M#kRHe{l_@<!80rT^xy=c@=$s~s+(!)=@sSw
zCG_k@P42D_uIzj;vbDA4*Z9vj2y&Dy5bH3g6$H*1WI#hC5G|@2F9(4!3kwU$(OEkI
zqLb6%&cEhV4rSPnM0GZ&haSyS4(^U|_!ZYriD51JwM(KMzAc00V_pHib9FQ9r_QxD
zu79Qqv>D$AO*mgb(vQliA;>*W*97L;UP-nPw_`}2&p2sct_05j$fPKG&<AW0`Y{L4
z;<9f<w|faYWZt~dG&KAnx9E6v?Tfr?l+whs7N$AB0!yTJ59o5&C?BH7VK})sla@Ah
zQ=1%Y%X7kS@ewntca+|*dt5|<a4|@VQn^OGcL>)H+&!&lyYCnUVvakHFEW?Q*6jSW
zDYXQLMP+4WyV**Ie*nmM85W}duZ90CTlE1qVP_6}Kahhx-@CKe>nj}L)vIWL5vd=o
z1~J0uHdEkTm7{v)+I>4n6B=zwTc#%-g<**~eSpDX>9W#QurfpnpaW)H+@v&jejM<k
z&~LCMD9w}#tQ;#hwobV-BUvQjRu)CH&K}Ox8XAP2gX~<v&<cu%45i_`gmT52`(3$Q
z<|ZXj=e<)t{%$g2!a;NWwp`ffxjK{fM@{ca@oP`wc5SaD+#0X?Ju_jrc52P1TU$fX
zYU=2Rum;DuzkP=42qJXSNZhDz27qqTN#NXUCP1hm4i+-xjy{l4Q)56&sMNpq!cOzh
zLjwZ?Fhgv?v|RJ-)W4?axh)nI6(!PU>Kdi`&O{q?FxFAC7{9<xnut`H7JLH<NdiZ%
zKlbICY6l;N!``pGa8@fKHf(ppGAJsll^r1#2l7Mw>0TKevm8o_HeIyATpn)bz`;Dz
z=l(aFSi+VIt%d%6$3vT@g#$zh#0d$e+Lf#}IoDRhEeo4s;@NCzZKFRquv@-eRQdg1
z(12)dZ9-8tuRYgo6An6V)r5qE&CTp0+uTP@p1`uDpr8Oo^*8$c{eAUJ=%o&8BoV2=
zCA%uw{RdCPhxIF*vd&WJT`AId!z%*s);8`^KAUxhR>WNN7-d9L;0yi-&b<$LUP+K~
z#5e!xHORL=uwT1?8LE=Ti5`CN-zir-rQZKkNmAK8LOPqRm6g0!ox7FIN~ptHfzkHb
z{(?%=k$BG~6DT8x>rQFh##j3uxOClV(<yloCKcS-DTXV@I-S$nTY?DoaZ&!^;ICi%
zi6M^4@d9<XpEL8h$5)+3xI0)tlN!+<k9kjGR^iGR*<`$p<?!=d=e6u;?CwIeWDD<h
z2%(VOW~#Rro_Al1y@S*?Y_7d0Q|6ROhH{`fSju%Z>DkyXwCAeL=qt0F;>F|Q*sb;$
zk^|hwTh1bnBtU*1m&Q(UK9KwJX*y8<A^?T}=h$MQI}JWkQ4I6mH~FRi^Lhpn*>@U&
zi-=3e30)bQj}o?F;~H+8d&gy=)FY+iJod1dC>Ld-Te>uH>r{9Bc0)_k5lQUU-gL-}
zHJ>Ho^uhKj??~VcS3IFMD<VyosSTTwc#(~)>Y$HD!JuIIj>e7iA$)-{yDwMnPppDd
zqrC+ta={eb{&wVc-Kw-u`qcI7l)Xivqvh$btqjXXUXpKpys6xRP*?Ysmi&UYh7L$(
z^epPnQaM+dJb{0ke5wxxAMj=yOj>$K^>s&;ZiUKp)6pJl0#NLX=9Ba1J*$-M`7r^q
z`Z^`7w#7p`E6@<H-S|(#j^p{15H*l0lT)XB`+DdWb-!y+RAyjqF5}LE_9Dg@e&zDK
zFHCc(ofhgNR;}(Op&m5g6D~bjn=`oQD5UK5pucB)RzcS5ICy)>{55(urouU3G@es#
z%AbTSfce}Ge{P~=p|PIOGGX_}w2<_$re$Ml^0@`%g`YE&>Ap*yLaNq4Q*lh$r-@!i
z4Z6(dmC%SxH*Yc*<&qsel|f`JMVQb``LR=*Ah+6n)feDn{xy38mBCHHsS7v=o~^Q9
z%H;9Z$HI_2N%p>6A@7jAHTIXRgXKCwLa4XGTa;W~)1LZsP!B*i^*&>L#fDaAgYW{O
z?=LKB6;|{&{D!x@_YJgB+T^=+zF6a%%1<@dFWP+_ndlt7%XDvcDFyMURx#Mwq-_zI
z>eXnZr>x^wy0&*+MkRo<FP?_wG=Ip$MgIJ|KG5TFLZ#Orb?#uzr@p)Eh{*m_X!M(r
zK~OaEcATtiBt5yP%Ly_S3j6$7G34L5Ff#!s+HcR@Lk`*P=m73F5<@2g0%I-+R=Rb3
z6SV9QN5dl8$Sgy&zU_=)J%5gpAn<OU{KgIa==33omzUEt&2fCRRy)m&OpQ@TmHj9p
zS_>@P#lLk3>@lKdc3b`xM{^#aCQwrZA(Fse2xCI5ZOz9fiWtta2USkAkiAvv#N=w~
z8^t*WZ!T1e>3Fj?WT{v31-!)HqK+7-dkHW?&hPxrAD+bZ2uw27Thkdye>|3MZ9+rG
z$EGj7D*DitgJatU*RP(YZVmuj72tfg$;|8`+8Wy|(!M-$2uc~Nj_8yqnLmeoh*>V5
z-z?(#1rDFbB#T}8#dYX~UEcuS)6P(?;El<GzUJ1pzPi%k)bYXx{`xl)D0q~RJT84g
z493Qh2M635Xqa}j6mn}gz7$);_wTBd11IqbGNk@TDUIiIq|<2!WNu*Skd<>fmX0|v
zJ!DrWzS-~SI;u<qF=%LYkt{A%FM*0iQt4NY_s|6n)(q8rjW)PT_}L{Thd`U<zY<l0
zQc*@}=lA|O5B#%xOJC7g3$@c~+2-c{z8+Kf1~`>b#fY=Pv9xF`efE=^nX%~4jxtc{
z8&8r`vaaXa=nCiq6o?U_Nm;6n#oW!Z0GB_aQ-w-fk7V#Wv?{(ayj%6$r{$L;{xwkO
z=*9`{B-V~Bze{aZX@8E^u6TD4Qh{$YD4zU#eP(cFE<P$Q9%CWw!h&e#sj4#dz3An|
zdl*69mnO;0uf7U95feL@Y18b>modVKyUVhP9qVwg>84eEmho#B4UhYgkM2hgWEeme
zO`%h87?W&Gu*XY)0<+1aI-lkJ&%i_^P|A0YX*;m{UDKB=Ll=Bvh4LQucn<bA>8)+h
z;^nI>okKtNzTBXGE&KGE7q-|cxhX{|(_%%(CTFw?72pr=!cJ{%k{c}5h#zOW;;eeH
zmV7AF=RS{pFB&87g+=BVnW;bwGN!j~k<=_)sXETRdT)MdX>{^mxBb!C7KVD{R7o9U
z_0oVI?R|sRu-q?M8N5SoVCMmXbt8ZCeXkhws8$V@gf<yF7!{KSfE{=Bzt!_e*S5>7
z<|L;ajRTQVx<HvjtjWt1mYmMEcp>=TV6R>QuI}J-+Y#8`DQ84%2-fSr;;U9J*hy6t
zCq~hi0PRM!iKpwW^2uDcs}h?QJA0eVP0t$}{#bSifPeIB!VOKX{u}7<kvL<;yM+M(
zVD2b{c_vnllL|0G-{(&t(sjP7Np~KO6Z**goUl7ncU$<yVgLZJO;luLY_cguOKE){
z?9I(Mi)ThC#QyPB@AQjn`qLTU=-s_M!{Ri#Wnp4$($Zt4@vSCNhzFh;dVGj7Thdr%
zFMJ{US}?_+Kd0xHLaZ}AyJAz81rafk+8HF%W=w<M$to8}2iK@`$b#KP;EIM9Mhv+w
z4s@hL=50A6el|mh^=~JC^OSAhBxdz*FIfy<dfAKh(&bmv-q3P%-0nQ2T^N#-GZx0a
zH>h5Yqe?R~u~=Q?jlm*<;p*u0jzokW2uG}>!&xFzi+HNGM{XPgJVHc#nCkov>5%Ju
ztT=hoSflitVHb2l<uM=M&UrM@7l=p*1gn7pZYUjF>dQ@+LB^<E2bv_fj~f*Uze?Cl
zzwk<@B9dA(XU%1`YxdGr=P6>2rHu=vi<qkc%{^8wYt(irbQzOQGk+Q*zBx!Cd$Bg*
zHkRGbso`px&X9N-*eaz>ievlvyz0>s-NwnOzPWPp?Zlyjr4>kYe0FgJvIH`JQGC{!
zIyfjQBqPJW2%aAMpjWjJ^^Ry`L&MNCvZOJwf7t6PuOkjt#ku0qrdRPoyz&4^DG8fc
z-8?7Nv{z^G+UfizTFcF0_f7m}??=nr{wlWE=A#g@8xtqqfnPJzm8pO5TJDMk$JJ=j
zw(cu4-Mzcr=J>_{ihO{JmzP(=Up>XJqSb?cz1Us|u5tnzW-k6NP&mM*T7(@E1+NY(
zuVoJ;x%OLidKAt+dCI1cD8kvS(1*pPdmZ-Sv*yARBGPmp^x;0jM%3}NZ1;b8dY)Nx
zrF^K|C@d^16i;qho1C(^`8Q1Gq}Gi$i*zA<5sV(tzsbY|2*D?D9c<yoc->b;)Hk;K
zkXT#XH;pzTKkOWrOhuK<OHA4BP01@)%ny(zh^52msGM=4Uw~w%J~c*|6r<#f71*%5
zJ1-_y#R;`4w{&*f4@$;JxMJmF4h}sP_E$pW)a>hTN5y5I!O0LBkqHaEu(1(@QL-Jh
zn+OESLgRZcUXhA4WOsKjFuKB#3ErQXubi<=*%`H#X}xHs5XCpmo!!xUi3CmxEB=k+
zyGGKrJY*A@n~?!Koz<llb4F3IVBBE)@EE{QC`G$(raLt<k48rss$HH>T|pei*f_jK
zEJKM;c4@slSg@Q13OyUp$IByKMWL{jQEdkOLu<b1E$>`(Uj{6EC;n3xmfJMKGX9)%
zae!6{$YY;bo+lNaFU>CMptnR7bOSVjUj1PLJW+rK<p>~RVZ@H)?3$fEl%a&IUH$eC
z!VzDj2TN3j7AKlp+xzD`(km){6D_T0sU94air<@tth)}YDxw62+sweT$3{g>IAgab
z;u47vwC6m+EOr)He-u!cydW1Tg9Hm#-8v;Q%Be$EsO_kct)gg?gKkE5dTW=$v^CH%
znwTH4l)Y>{_h}eaYp2AF7PvGcHIv)4^JV$4`nO7Kl9{WkO(;_|l#;IS+h)t7)ggR;
zj-&m#I``Q;%R8_a9sWHtl2n~u@%9lB0pJf<!9~~-QCez-<;ZFkdv@Mr%;u1%jLMk*
zwrthKZ9QBchRycPSyl?hXH9vcRimsjYBcx7n`B82>1|L`yAX4}f>Bq_cCvHgZ)HL`
z<`-H}RKt61^d*#G7*Er3ZS#<%lvT20@&~9SmL>7}6V=Utn+<VJk`|<}6MXwXW6roB
zxrh6<_c=|RsYU{FS%EMdwg_dgaM*aFs$NqWG6>qh(;O?Qb^~CUddpwKalQSJk?Omg
ziXkSKubZ|1Y3yNzI5;vInxJ9D9$)gaLEWf>dPWEd?BdHnC}glyd@W^xFBVAkEXffe
zl&<KnH2iO-4D-Ps^@AQqWk{8td)k;7As(C53nLYOdg|OD{(=Fkn9atE$0asZU+bCv
zbQmi8{)s_&K8HY(GDFP>II^#i$ii&NQJUDuoEpV?7CupZkTB>wKJH_#gomzBLCC@a
ze8Y_!Pt-}INbUIkDzvw|AhM4db4rLcR31ots;T*!n!&<?W2AlK7~Kel{(?Ph?p#Hy
zUAbK8QwxF}ZLn&UGm&EHx)xeD@x)27FmsU7yNBfLUEmgZ+<4NmPrUuBm&2KvQ^wCM
zw3UVi7`X-pqKDy>k`PRJ78GFgWioQE!*eXWZEbq7Yc|uaf8cPExXBai+)<31pK7Fy
z81UQ}&(<ie3@KX^i`QD`1@6Xz@Gg8aolOTSm=CEg1>Vqvwa(&+M%%PAAN`o4&_X3A
z3zh|?(brBVU3}>DL$%5n!xwv#o=FY~vI<V>?xZ@HbvPt+hde{{^Jn<VtIYV*$|}7C
z$Fq%5q5#_eiVIL>u2uUb{VSt%GANqyT+1KuEElCD5mEUwH*?d7sR;i^SyXJ_$ZM+(
zoR~PEa5f*#RKv%KIZDM#A4^|-GrRp^dv~+D*cqNBVgBi26<eTOM`!w90Zh>6KL4u2
z-XnmmSbD$Yu{6~vU2!CaZSHg(!|@ZJvR-r1_0#N&;6<YmJ~agnj?M!6Urt0E<bfe}
zV0d!T3u~TFR<m1@W7WO)VX7okSGw<mZ}Guz83t3+r<^QuyRFCiv2W_-kU7_GD0O7&
zps4RX43ig1h)FkTR5Z8L`#=*9^nmr>Y{>QimMirx`2b7=q;MC9vn}?;?vwtG7p<_Y
zZBRJa{`pR{C(A3kQQKZb%A%i2o#dv&;w?%k2L&ff%-`ylH8cXtYCjKKPF^BYEmMZd
zP&{YL_goMwGP-{;q7%%~sU^thEVZgJp}yVFr<tW0Ke+iUnkXLoluwa^1JcF#1%#j)
z!#Wm0I;HpLs=gdg$Es0dul?_-=0EKkWi^SgMTfLi%K^1f$!t^6(3td6ja=NEuFE_g
zq)tm-1{6&mZ)|41wSF)d;wBF|oPDDwvFcXhaII1S%8WSP%u#^Hti!viz^eErV@BrE
z)SHFEQ<0oQ4+GuhzoxgKTia`h2(kSce$hhn^4S&f)ahlb;@sQGIIOEiNT$xgoa>)Q
z7W(qpFue-dmXLU*5DRW+%|b@m!r4qH3OzlyvhvHrz<y+8iBfc{_7z6ZUd2)|PA!84
z9&Qo0i?c@3k8~z4Kg#bK?k^649Uka<9pU(b;bZCYRc?0+3$~z{5`C~7y`_#B4w-k(
zNsoqEWAS3rE5Og0V=cL7rz38aSR|9~jwX*frvR6pA<KH(?$}dWylSKn?EpvvfQE}r
zajQY?VUZDOXerUGFUwZW)j$0CZc=AM#TLrOWh^ro#rVJ24Dafy#1B<f#WuSz5(J(t
zV-*ddNjz5+tVf~eK%lGIF6!s$lsqvFlPTr5$aH?Ws#4>>4FGyu;~#GUK-C0s`$lWh
zd%%~x)^jurW@eUZhkQB?yU({-9}h>+ho!PuYT%BrSLP#Z;=@D3&`)(;Un60ahl+En
zdGFm2ZszN}qVzIvSoKiZdb(MSd+*jRKx_g|YB1A6#;=~giwQN}wk^G~c#dNaRFK0x
zV8KNkAOLJ5DqBs~rNkdLc37~We;&7pxNw2v?X#N!FJF$uC#lLbQFf!E8FyBn%3Vj)
z#n6lPKhjy#>rv<*)*IPx&}|0sFuFWXF9WI+ZVC!GIA(%hozH-KS2LhC1|gwhp52}5
zO)ii508;c;52}x$BB%0<vc)m2=$RJXgZZ3Ua+}g`96u-1Y;CI~ycQ*!ChRhY?!GZt
zKLS>{U%+J$QZP3@<U`EG*f0!ww$ocU&D}hUxd7Qn2<l8~tF7GW)6$lw;RWC)v4bVG
zO8?UmB|#UD&$;>#v>EPy%w>Sw>ZQI!IzFWeGtqc$6hNZ<EirM_pj5`dW=I=lDXw!G
zRsFX@74Z0WbCqk<oA+mg;^QFkllXos?E}8E>^l5D*^gVfOUl!Mox~JfImI?`CC$dW
zcX%vTJL(fk0T_-A9FK$1&>XDxKaf2RT|3(<^oS57hospzS9ywNrM+a}p?Xv1mSm~g
zsnoWdGiJcJ8ypr={+=5G%HwR0whGoK2z%_y*2lFEEyMLf*wb?G<?Yl?&MU8hYF@&R
z%|#s`=<xYkMQ%q!XVJ!noT>`<hBrc?F)p>1%Qpk6N2}$ks^c+V;`H=KD$-!I4r>f3
zsBF?)snEZg;<!jeo>6pbC=p@5z-%%*3tNxjvHVJ-oA;`N9rlRhO`&P^Q6*gCI`$Rm
zxnZIQ5C_k`H#tI~^i|!(s`;2Z?7!(3g3-A`!u=wGR9_!=c&y^H+}W-V7AaMI8|Ugf
zcs<mQx;>r?$b|uZT{%IKfq~`G4D`iQM0ER?I&}+=RAz{4?W}>VLZb4rdl;0!B!ejS
z$hd<UKF)x=L)iI&n{GKFJT+~!(G3EOeam6FTZw)xMVoG43as6&A5b9;ySqVq^_$Me
z2>+Aqr57ecR*OB!@nn_T=^kr756-+rZ_gDI!h-|&(O@uqV*!V-U4a-^1mo$41`?fP
z>ylSCGI(va5xo1;a%W;f{W?+a+SI5Lj;w9O^WTtAT0Nts&wFEIGs~`7HljSYg-HMH
zW)a06=67&iQN38hsWaGPF-o(vQxNU4jHD;A8BJ})@43bB=E=NN7|zo^Q`BQM>w3|R
z4*flep6rm!qJC+~uj6A77}0N|Tq>_;36G0mjBR7Yi@(_U-K&Nfu2xk)_H@=C$>i$j
zNtXIZ>F;Mg7n&mE#-*_hSL48IN%dk(9I5x<sDlu9g(cUuzG($r9`_zH8FvW}DMcHJ
zzbrD|qoWAxKUE0NpO{tV@?T}3S|sLp?J8om5Xw4p;$T`8Djcl(PYat;-_qJLn3=SA
zx6`^@*`CkIuk(MxEHdmRdA{k*U%;k+6K(v;(gIgst57!4sq)C7DW)q1x&&GgS9&sD
zJhUR)_D9L5UTy^?47yHaQfGaBz-PHs`%*D@Dw`L(d2xwGITT>!C)P4d)$58N?nlaK
z*A@1nN&CZOx`Kg~ii&<xF7m<!eVJ|WU9ZnggP!9*O~}_}RbH8Y((n?H0MAL{uZhnA
zvDXE8sS0F4!Fh6MChZEZ<vchDwzZY?-b9B@Uz<0u@`YnVWoFF|ZD$-Q3okTfAVfx<
z@w=DDFW}=76yoDSktG@e`Fdr%kjd%G^C;BMV-VcA^cfr|LOOwXYmd=Ph&>l)R;M*n
zd%G(4!{;=L{^9$0`ZWEr5V*UXqFmH;MWc<`iy!@~gQ`!)MWesZce`&kupWt4#J%Fr
z%+=-)Ut8t6@f6^*B*8xcdT?%4<xF-E%zT_t{K;VK3JC<Rxa;0YFMNl7X+le9aB@v@
zrj&;9R#eASwjOq*{6hQa8z!2Qi{zA3YipszvrV);pE&{O8XuxBGv`q6t9!BB1-l$I
zFQ(2Hwdip^hcYLvTH^2&z-KY}PSLX%?TmQouZwCbNLCmOb(;*Os1S=<Q`ba|smGga
zK&%5HhQWE2BP(xGg=#4Ipf5+99f?e8j53^A3jn;Mh?#6lWa3G3V$qocLH}E`N11Jh
zx0rk25x~^k)A9zOr>$`kudU8D{z$jVe;|k<L8i~1T+Bt2&suN3WHC7znL1^rqR;mx
znAxvJy1uD+h}UIsH>TPt`eytWbw$V~hYATK58>VcO0|#?;l2LQ*?RTJR3~<41=niQ
za=vQ_W<?nb3!>}mnk(ye`ZzwFOITO+fQI0>1Tzr5KsXNW>J&Rh0koHKsj%@mpjYVj
zZOXq@XjsKu&Ye>#eMPW8n1L2ecx`%(6AwG=>B0gL$+aiuU3QSiZq6Bk1)Oa?+ZKWM
zu654^{}kwK_wRD^yz9ROY@o>SQldgsVy=Ug3w5Aq+ROZmt&vHFvy4`Q*t1t)C{^=i
z!IqNEjfP#QDW-@B%dil$fx~rnK?iwcZ^<XS#=5$f@4dZ+!a|dS^D}yDKcD0Vb+q@S
zdpGg<=|W)OS3<mW8>j9PRhx2pJZZ(fd8$3oT%@y&JGn&3X<q<4n6&gd+ItP`q&~^U
zmEr8Hn+ad~K=*gtgaXX=66pTExcWkuxybzeMYh$|oCkShm%p>%IgM?;Sj)ewE1g(f
zm1HS09~t2du7px&2xHQ%b&QAh^>8N^t8^hI5V>o+W1`W<=gPlXRo+)GrIM05vI$?k
z-icA)CYc{RaC;OcE12;8iwdS6dowrYRNR$%@w?BV3tmG^uiRgy{Ag#4HTcUjZk~te
z)bJ0;J2&+zTi*voZS3wCE`4x!!wTR$Y?Pm7Rc;ThEekyGJhYYKqF^VRosNvWO><Yy
zO&($$vl{%B*Z$s<_mDyd(Oa`)Z)nOmH4DAca%x@aYgPGt+02n!migYj=9ytx#o5A|
ze*DTTTdT(r=V*dZ*67sH1AT?p<820z*=fC8>8Yt<Qn{31TG~e`$-9)KZRAV*TqV-{
z)p6TfP(gd)!)qVTU%zVJzg{PLK_}M)wZHrIE2P|RiqT)?X`yhDiG!SYMOnF_R{3=`
z@A9~p_ScVYlhsa?y-N4q6y%WlGPxw!@ixf_&%<qGbj9VyTg`R%lCeP06ZKd_TFTVq
zE;8YBRH16{1J>;=+qMKex%ORq{k6-sSubj5my`)Ag}n7J8)LR3?ZXC9DJp1tE{K0B
zP4U8)9ei?mm8oPt<?BBjZ(W^O<v9FrhedTvhltP2*c<JcwzI~2L4r#b{nTR3wPlyP
z0|V{zpW&9i_e+j~mMz^Eqa!U%EZ<4wmYh%#6Pp=0QPFu;q3e6=_rrO9IN~j)X27Uc
z>9b|i40{8J2Ox>CR$?dT*@2tZ5XIz_UrO{-0@>o2&Mo<p(NB|eEfcSPH*_5=2wEMi
zdr3K6N8u}X3@U#cC3;JE+h12;Wp#U+Cz^8PX>Mxict}}4J5$KQ97)UWU$A>5sR;Z0
zwr$9k3#A(+6yiAV@We<=@3p%mVfvLTB18GqQ@VTIUG)nt-6Ghc@MF%6Bj}EUAO5LD
zWsM=})t4^>)(0A2pVmFdu@?4<WU4;$pDf@K!C>6-RXwB*5K|^bUWb&DAksFaaYK&7
zuV7_M`>UTne)sCyvpl1K08uZAe=h&~N=`Vzo?ayTc=)h1^V0>}>N>v%{MQ|vCz`V<
zJjF9PL`8>fm|QPp2M0snJAXxugf$j;O`0#u%WGA5F?F!-C-+lUaeSd6{P{z*HvDP$
zc0vlJ+v#5l1p_0gu^bAI+zwDq?enW&wC_TPo14(P0t8^cLN)ZhpyR3Ibr%(-PAe>L
z*&Y?#h|p;(q#m-j-jN?&9{fxe(WKv*{b8r5WOD!O%5|>;``KM+kqEwL)8aC^=N#RB
z5^rs7SCF5smpq-)Gq)|25ZBFBE#!LJ$GwLZ&AIBwNq6Bpw8=_}=nYiQ>Z^qG9D`w0
z+^GJ&M{Dgyr+#wxTZ~z0#lD4cGjfmA^sP*@`^Y~|4hy<VySvc7E%@eHWi&Y{5#iJ<
z$E2FEar~^a0DU;rg(pKA=>4*c7rrCtaV6Jz^^>iM133@wPh&$&)TK*ho2@rS7b#yW
zTTo1qd$$6LUgT-_i{+=-I0*>XRD1RZ`xG}PQfubr@z3<4Coa(3LsgZ=vIjZU!<6qZ
z{f|7!4FNJr#j$c*oJAazLT%<7z15X*FDhaKw)^l`Vstws_x^&LCZnr_WAM?km0OCO
z^kcKO>9n5&n}I<T`?VLp`uT~sE)f6Aw%|A7pX)FQb9#7j3`HW;hMB(Szvt#)H|SQG
zje%S9|85LZ<z|{PwrOa-6tHK4qlw?RBz)yr(o2G_^8L!Z)7Mq}cpZlmu0zAgbti|(
za%El!muI~)5H7Yd^g}7@dx_UA$<07Sup={Pv|k2K?)vFgGqI3S?(c?;?vR?KD`HDS
zz6_TR-`bdG76ymPH(*CTXmKN+5RBOlw6_<|Ih71l+R@w@-_5($Z-CXFGrFK)J+vUa
z6RN54H7oQsM=-U%(emHsisswV1C{R6##Id$puOoWcSQz9x-eOHhd&SF^Mm@2nz}ll
zoh{tfU$^wbQk1hQpES2B-&fkEu$93i6*R>`y9kl2tW$~1pFhTnjfp90S$ps7#KG7u
z_2|GyWA6Xa>cu{&A$jl#l<_<;Gq@cegDO;r)@%QHzjzXy?t4fe^GZUEf9z$J%7zvc
z8)+5)j0niscx(Efz5Q*1n|rTVzuXYK>mKYHncF!qJ2r2s!o^*)M_H`X7I@otKhI3_
zi*w~|g14sTKQ)g%^4;Fr-jihr6F>Npn#~$-{T~LD_~)Z^(7B8qAoWb}B4_Vz?T_h8
z@$W~A<%9(DK4=Vn&K&y8wjP2z1i?qI$v+Own{30`7Z)1oPdm>pi@qu^*omlwkGG5m
zRk|MWH2C{xC+@N{N=XR~Eo%wTJ|HH3%vDHH;>}_qS??^A;KgbGWvG(aPqse1x{9Jt
zeHXnS0_`w>%2O047`I}_MxkJ36GJ(1vH31S-e*fgYfcRfY3U=1pCIw7!-wb`JxtGi
zLI=UH($(45CJK7oMs$I)E$}a=_<8$jBEmbLKN_$H;1`ppoOPPeDrjvtuJ3Q!<D{8G
zlfc?<t_AZpM1@D-2gV){Xz_WztpY`h>F-q24o9u7Dm)2maX*M<5Sp69^bSXLI9oMl
z6PyZ#TzXb$%>r8*hC+eT96dFPX#XoH=bcoR{pF#inK#WKtuG>6XMTM7Pjl!?0<8iQ
zU5v~U6bp7FDYNJcKi0)iCwB1x!Q!T8<4!BlI#Wj8H$rs(Xa)t*e{;&N@A>lYn~xX}
zc#f7b^*2=9%@}mDnEcgRg89S3J{w>y?9>49<(C@GN#pJFi%qs8ES>cC`GsxklX_J~
zB}n4rWQ_id>F%xn)|S5!yoawUy>7LtxfItb|K<A0iEgWceA%}c#SBuYU`ypb11S+^
z!+xKzb=2dNeXNVhlc=7a8j__^FS)t<kCeOgg5Ui7_2*m1;dZ9;b{3zl%gV=$Vp8pW
zh%j5sP0R9&4rX?$IL5NKdz1`pVq8@~G^orC*hFXOPlW8rmrBKPTBv|Y=;r5Fj`0P=
zL0Tnd2Ell6%%g)qq^pF~C7toeK`@R6jVlyqq){iqfdn1baliFsVIg_mGp@?`SI0$M
z@6l~bHS8%bl9LiMuHMz{_ZD3<c>+ZgxIMQeC`lBQ8=#IY9C^yj5Fs8yRX5rj?pKBF
z^Puv%g>Y&Z82+Nbq-f9kNW${Vi?qqLl*S8MB6+SD^qXReg{3wnS`NSh9O|%u(+@29
z_Q24Qze`L!x-_btD@L}x`Zf47C;Ao*BfDr-sM(YMm>?e>4s#fW@j?>jmp3RV^r}hy
z09()<dv2}H^(EcRzERWMuWGiZBps~jLojA~nd3NeA<?To>(gsS>yhB9tf$8!BO7@I
zzm=F7%JV*jeQu(^ckg27a9gnRT9xG0F;nM+u-sUdjU|mPd>?;V!uQ|9bxdxjdlm=0
z+ey(FUJWTJ<M40?K{hD<+m<{*>YeKyA(XYQ93r}wr$PL-gm3*JmSqJ6%g_DoT8vZW
zUsv4b=0WMes;Fq+D{4XIk>CFWry36m3fi4OJ#O}=x4mx#8P8$vyNm`V`&F^&$CgbR
z+DK)+knr?=Bt3ceX|_Cqvc)(qjx*XJ4GFX4%iG?ffW<y585Drh90d34>jA$25$x+)
zLC9&;D8`<e&TZOrGp@C59@k6l%0f*4`ZYWA<f|84O@1-zL)KKw+Cb{ISz)vv*vA`|
z$L($15Q&dR1s9_W=<8V|E+FFheF)5?58M%UJYu0^yQ_1AY`jouqF@jpv#cFVKFXxe
zRiTxHB$^cJj0|~`Xf(7?SY$I;q|+MX8=#C*b8>X-^Dq+c6p;p+;;&};!Z1|ZuNo86
zRw%rO(-BHxEMaUo%C)wP)VubsG1{?GBxjb3{=YfzSISFzH^SuGR3v>?@;uqmQ7}*@
zw<xc)Zm1y!kcaW?+?2UBhI@3R_I(Oy0MVW^j%OVx2v)rvYE|Kl*3kpo*Gj|r<&=cx
zQ!-5xkadoaBTs+%PcgY$4y@6enG|v`=IF$|mzRo)WYap{(ZgJqu~z(OaN(^R_Zp&O
zNs%tb9pB!i`ms?Xk^JXGOR{Tlq0?(<bge7;-BPJNo$&4sw9AETa!aog4}z}3H7ZcO
zaH=+WL62?9dJ~iHvEPSDPmd#42%H>kKg55&?!w3z_nkAEjsM5qhRXfgrYxPM(<|jd
z+}M}8K+|vTNE!N8#uluY@+4*-#*V?D{Tw&Y9t>f_4s5<H>qX@dZuOCMQy5k78{?iP
zRh_$)(l357Pf@xA`}5-`p{brZ0Z+m3-E@Vlo&5tNCQ6ey8XAQ>9<ImOB!co6??V%=
zcR<LxWY@8=1Ime|HYD*Qs+-qIffeu{Tkj+E2|BGo5Kdfp-P)V2XZ_A>D9M}txufs{
zL;R*pdn<~!mN+J+GNr=#omwgT^gXgWl%y|Z75RwL0vTOgprVb<-)*(k5_l9>`5O+#
z-33XQGUrkV^_-~#mmr}ZkF&$$JU?1Gp{tKD(rs!!M}`XeV`Mud!meT^{jVABG*8dy
zh`(v>wW4yK<8>&`d$_f*K#*O7@b${y_&r1V*2ef#ZZHYQRl;#2$6elUvh;fc1IXy`
z7qTuly=@d$4<HeGp^8C)J8xeQ8!#5FGXS$V?+dF+P@r;FUzyJLskOHuL4NMVg8aG5
zQ>G3M#T}E}T*{Xc`ZcyuyV(kT0v3XOcZWOOc3t}fns}`yBaoP<5nY`gwMxwOz<|s9
zB5y;_*wsITmN-7XOK)P{Zh@DokTOi+{jb%7K0-G`$4Ng*eKRQE|6csr-uJKov-}b#
zWyzagSd<C{?2sh9w6<bxjplZ6OW9)3G+*4;dBDFs+(dr|n9qOMdJ@Jlvh+Bo;PfJ!
z+VFnoq3@_gO7(5`jO7m(39RIcqPX(-dDCjlMl^;8)kcW@{QAVuvRX^3Yx^+uM%5mu
z!Ct-5qVH#ZYw!)uZ?cw^r9rZC7xFB|s|Ze^)jM2Vr+2T9R!$nq-!LXo+|i6LJrc6X
z{68{O#Pe}2i-<}uoELZJt{|{9RZPi%I1w1snwnZ{9{>5Z>@S~{^WgEMQh7HYI<;uf
zY@wdP6XN{XVi9c|8U7Z?|4I0}_B_*3k**wGsoiY1wkQ2t6UIi4r%X9lK^fH@?%E^g
zWdWJ!=z9^76Xw<ma<^{5DER{;X|`F&-x`{Ktb8G%A!AW&Y~t|lcEIGFPc1AD3ybav
z-@Pl6&AKqR@HKa3WKKFq4XLZ~HQS%`_sxC{pB>9^1NJ#Py^FX)M@WW)y`lFffhMtI
z1s2=gR8NsLz545cIxZ8{A<UbP%Z_aC<x=!g-=4^xsor&VBz%Z2nwT<H`)Y(SA>}k*
z1<il03glYs`1pG~;wlRq?lG$e@0JAn5niW=Hs<@qX%ZPu(D`em3bxYB!Yn}J;`eRd
z9LdEM`kKB>P^hyD2Stsp9C7mTSgjAQUX`|F`Joc{6sh_4UFs<ndR5SRx8%QE<`3)U
zcdo-GLV*FQ4J`|t@%qwd8QzBNwa$c)gzpkj8f>)V%kVkX@<(b~QnM>H@iMX#(V_vL
znp$$nXpsp`3G)IlLANg_Fq&*i!OI@1ch%s$np{vJ;0ZSYL9x7joS$CuxZfINypl|(
z_9UWDOq|Bs&%kx9wstVkHZiuY$lGhNH`lkAl#qepqls~Mt47!p@z8IrA5?4OWMs4Q
z^>wAMip>v1cX_vcBhcjH&E3`Q%~OsoZHdQalAbvLtRkeX{STDYGSzg3%C)}Zb(F5o
zLQduZpWojoS83W_rusrc0xVE8*%&%reCZ}FgGkCmZdri3jSjB1^q|2j$HfJ3{3M>Z
ziG{<q6NN(^mROMN#)SkCzw6=zBjWlyq`GGRi=}bH!M^{Xyflb-f;0bPh8b6~U+(DC
zy(X4V&27?;L&v93UfCWdGvw?0V-Fy$YFC3`mS{Ewg=k*SC$?|mzf?t}jpDJimHPUh
zWMvpc=)!;%3Aa`9MS0{by5Ih9`<lk<jP#ur%H8Ep_EZ(6^$35CbtO!+Ra)4X`9|b(
zYBQvLOf|~`t~uSKcU0G}Pn9dvX&o)IR`K1y(q3`82JG)VCL!o1nEG;6fHt&!1%76s
zy%U2ux!QgR@)HEAdhh7s&3XnqW=_AC`Pbb}_JmskZvE;RM7APT5C^5T&GM%a$l#S<
zZLv2z@uGn(JfCxc+V$Ueo3J$VO$)tEQO}^4Qz=COtirt`mg=E`;Jahq>PuHN>x8f7
zI~^*z3<pA{Zrrx8ztBhLP}~J`xJlD%1KL5M0V7+uthc)$uE62%UFyb&R@QW@k-CR#
z>De^A;ne)R#kSa@ZkXx2*GfzTz~-0;I;q-QP`STGEgbwJ?N`I!@9)|w=evIsk`p`1
zk`hgB5)0gq{W*|&Z&VBsIQAkp&wltdl8E?ecnby8FQ~KxZ)dtnGF@&9J>(G$19KMH
z*1I09tI^){+nD~RM~FfX<<txa4(L#Nq)dMAPq&$2SWT01R3u2C<kGTi{g$L?1!`cg
z@EcKvkB=w|ws$5y$0Uk#dtMvrh%;%#+$Y;wP2ctRyJ|QU$`auwAk?45xV7py7rkGP
zn}*1v-m->q0?*tlreey};)1ybl2-2Zq@CVlV290wcwN;xT#e<WX{>ujpLh2vhhN^Y
zhP>~c${|~egp}lnj-da;pYt9(1xCLh4J#d?<GE@DLsPPouRET7HRQ)d{Y=(%rI!x#
z4w`wz1nm72T~}wNRPG3i>#ResE4i#~O`4JvsRy8qJJwOT3}Q6|m$Jgk^@TQc=DzAI
zrB?UxetR%bUg@!X&*R6Gn*;7-$mDAJ*2;0xtz;=|hyL564<SmxW^%BfM4wc_<sY(v
z1C=3puf;^%Bf%UdM313ioqb_bA^6=IP74w7)xY%(mZfDe5ii}O$RH35^z6^L8m7xH
z&6t%idzfl_Ye%wiqrWSR>Wf0s;o`QjdNoqqwfiz=s!Uj*v#y3WgD!^)QRnjYH<KZ2
z*aoa4r;_S^(suEVE0*wfXms0<^cx2O_RvvF{&9;mnZNx`!Nyz^3RKH7?ai}6<3UF}
zsZelZPRk@dDpM&d58~u+tne?m#(nkH+uB8EPv=#`;5iiEq0!_ug^u5UI(;@ov6jIU
zBN*b`lPzLPx||K0-G{?0cN4cmHPl)I*I)lXWW5DcRb3Y~d=(Kv5S0c2r5ovz7Li6e
zFCyJ>={kslA|>4*NT+mzv~+h%cX$2!KF{~Q@Ar-0p<^)Q#yMy2wbop7&AnD2CMMxj
zUHoXwi$wkp=7ue0ThG9@HV)J)aSH(q-}&Dy30TIcKHhm@&FR5kY~d8OT-O{=5}~aD
ziA_J}Lg=33^YpB*zO;Qs8=b0e`n3P_71jc6iQlk$@M-f|lAw#pIR-%2@2#WddE?AL
zuO%4|uRD;$?C<XJm{8rC=`Wi9SIQ@TU0y@*`HhN(R_FM6z}xQb4**uk%O>lH+p9N=
zFf3=ESn&!VF!D*ac%MFPJDj`y3Tfqgd6yJSq|#=f#8LT!OvODj&`7kffS;$)T;HZ^
z1^uVScnyj9a|Rx46B3dn;v7%v<xlwWQN1k^$sqfU*_r!B8KN*;#ozyyW^p2Tn^({P
zqrRLwZFOZ}afHVqcwu4f?VPt0-H3S;NAcRy-`9KH^s*L1scoOr?-7o6Ljgk^nJ<Ui
z1IBATFqZMoxb4y*k)HT&&!ZMu7+B)JrpBEq?QMsEE=uH*xhCD7;4{PAb$FQi7T!3L
zXBo#Yz*c6{)6ty&aP`F<H9x&&4UZmbNppL8=1;pUpDFHmC0}X<+3m{3`}k(ltC!fH
zF`Rv<Yww_ETs#)1g}InP-D||^HT)Cfdk5oe5sZ<5Bzn;4CWK%pi;I3l_A^ba5%KfQ
z{L1@6E_sH?gOw$JSht{ul11Ipq=z$T`Wq<QEh~Gr))C8ZfUWyzA)VZ-Ns+<mIO_%2
zW(}iD){!*$_*eXvFvd5SZ)igIu1<yRIXUxY5w+xT{Cw;dXNu6bCA?aXdAj3kH4v9x
zHQYYHW56qvBRu^R`Qtxirs!jHny0l*;w84n{1l6S*EiRDJKu@F<j9-yu~)EQbdo3j
zR8mEh`{>yR^}X}US3}j=iVUGJ_Us5O1Ihlb=6v~Ws^3$(D6v!NIOacoq}>yb-SUzz
z>wV+(G8qH%^#LK{%xWAKxXH^ZEI<N2EhyigP5r~-86f!+_jgVnd9O_F#Bgevgl)@%
zyzz?5iST2jqnee4rY0COgvRN=#*Fjk>5^6j|4$|V&!~0P?3v=)!y<6~C-?b#B@oWe
z-`^@`kI2r{f9WzKawJBZcXN6aUcMz(pyKmd40?}fh+CzO9PFJULH{*x_|k%+V8k_{
z_P`!^d_>Kgqe`+#^$|5!_YJDJnwlbEnk02>GVyofJ%8QlN6cqm*@q=oeAsicTe)t}
z+z7m#-_NcT;Qy|n_~$co%^&(9IJSEH@tF0&%|6Fl?{rq7FCT*sFhP>@>U`t_z#qa`
z(0colTF>UI-coMnGcQ}c2nIX_uVd!)<pchTy5yBUmxK)YV9Pq9o*GGolpF&Jc#6{g
z;75aJ!7^#8%GfLNqxBf@|E`k@r~kdeJw+-*!-Jt<-54U`U-m8UF*g-y59O_+@R>-b
zvAs{#^KuXri(YOUn>O!I3hzUx{WWGYVKvTLz{)^kOCQ;FZaXsBmi%6j6tvAJ@t4vd
zA}R7Xd!BaeaXkkT$W|22TpfZu_?H~N&+p%95<#Sb_kS1bqZAeDhnL315EK3~UwVw0
zro=KE_}5s;KBx#$e|^W>xdv^7ugX2M@Zz)yd<fv2>L)7G+bT7oIVRD3H3O<}z){LK
zIL4oGElCHFRDCp#+)NG~TQ4)Fq?c$Ve4Cy%Vs7lU?-*p&lxo0>IA(}ydjqhL@it$}
z<`?<z>hOX?-d>u_LHC|FKzliq#vUtF80FqOIh_CuQ{O_rvAF+1o8_dagY_}iy<OaB
zz29d&0_Px|ghp44Ew#^HI7L;f`$?xOu`pdf0;QGS^t1uQ@qKxkQ4hI07)$oEd~V-p
zopvPC)i_Ls9~eI({!!91xGf;n(_v(JXe@hblg^<QPdinnpH|2trKx#Uicqhcn!^TY
z+fw%>wv4heGE*NModGP#h9!cLHYAj!qJ-krQ+J3_rd*|*;J>KqhVQkHx(nS6zZ+(i
zSYyJY?TuQkLpb1Nfa_`*IlIR)TRHKEBQH6jiNczi+Q9t%?`f6`t6Hk?w=rYN%BS|r
zH=ly|1nWGCcWhwrM_b;B+5T`(S+ki0{F8D(dly|^q2{v*Th4qG%T7zq`~rvk?Yxc4
zN5LKV18(A{BV-ojuh93LjRS`S#{j;8`(un3RLpjrvG2~W>eg5UPI<~B^3cfQL4K><
z87YlK)ZTMr2m9*?fPr76{P^DDBL*JczR{k9kJ?qMat$_M;sCB_^rht@$xQ^)UD^2R
z4}QOZ&F;>|TH#272K<&S3enA`$D0grVFw?SqW#y)D<HPXn!3?+iBl35>60LdG-G)d
zagSG3drJnKWyMRTriA-M)#}6*w$7{FrvYZ`3%#I3HhEhg1s2I<@mTw{u~9f&r)ebb
z8TZ`Rza2AZrZtvWAgJymESlDL@rT-Stw~DxCN&U{!7{s!rImk2mr1{#HsQgkn%BcB
zvA1qC*%GbbLhN30f2{hV(0?$d%2f5(=QYK`b$dqfQR}M1NugS<X3;JTcIfVyKa^0K
zZC+!3`=ajB)N(ui-1efzX3x%1-&uv}rfYO>c;}O4)st2bng;`9zIJySgF|86<3%Ch
z$wBr^LjI#ka(tK96=zlJ4WjY?15~zO<wie$NoBahU1Zc#m^}98uB_>LdwMCR6KZlM
zq6<HWv*zN`?oxjE_j?-!OGisOwfGbvI57LLq<9&?B5=FWG>|l>YrliZn*Zhslctij
zTxh2?={-E`oe3N)wi~)wjDv`Og=?=12WNO9kkM;zSusDEb)p!{`$<o&<LGpL4_$*X
zp>)XO<fM-}DZl^B?-UNmRHV|;v$l6leq$-<{VdFp=$qNQuev7z3Vr8iN3KFb`E_|;
zRVeh;yoKkmHXe%5640`w8~xhu)@(c7ddKW_$|om`qkXXeHbr_SL}oIO{eJX~&T3|t
z{*{BKyQc<^%tN8A3E%ab{O?j!fmXADyR3f20%x!KLddcY&#tHA9$z&-jk26*8zbQ<
zx|x(4)f>7&38?Z%w^1PM+Qex$Fv{@!Ss7P3Iq~Fll>rXFz=5^|{_KCI3Z@;_=^Jlv
z^%b6V3rez@9G*PQhn8?tm6Z0KOEV+ubn8|eGs}Y?0jg}-T<k7ew%OTQ5yO?2DJal9
z6!vQ8A*uLf*u@Rl;9lUH%G$JIW7(V!?!k$8?jvxnz5e`Pxe*KCG|0YL?4uKQ7&_QG
zSt<(8OzUu6Q&Vxbl^!|mXV5f`jO6R1li@%pM5MY2ypb`od&irYSX;w{L%BSFzKYce
z22rqQ+lpjZeXP?0H!>mh_%N+IujcO&S*Cx6r+mtZ`e1Xaf%k2L$&rzj<!cg2cd;Rk
zs;4B9w8C>WLWPqKx$*IgsuIjAr&~>Ef;D@Kt%_dg=(Op-9#konFpL3$2C&Ms&}7+4
z;!a{!NwJMOk2+W<+XHFCvU@hMI`XpS#AIqwm8zp;P8Fr|75XT#b0v%bf%C@3$V$Is
zld$LOT`@kJoCIY(p1N0Y?;Rwu77DBj0J{;f>F+r8xVLHeGd5XrXV>DJlH0%3kKh<D
zgakk3`P|*gXEFo5{Q=}BdJ2taiwY|N2Y+>R5a1!`jxW4B-80u793BxV0cw7J_u2D$
zy4OdmsSAV2C0C;y3sba@JpsY=i1TzNXiKMnmPUvj9o`=?tZ~#pR4Zin>XJogX@2=J
zU~W3-nOYh<F<1#q?mKV%0bEQf)u;a$SCeo8s7_$(xIz-BRT2^|8k(au1)Fs`-CVhM
zhwKp~2Ig)uT^%U>HPbDLI*cLfYDx#<GJ9<qnJDugtmT^v9#?9Z2GcfFYHY5iM;?Dj
z?AfenI)oeh-Pa%@diQ*+enHjP%WK(#*egh<yXPYy!Kl>0+tW=<NhR-tbm4c`M@Lrd
z6e6^z5M%hfvvglq4dV(jMuS?4iJCW**k+fXV4)>{okYpV!9jDp)0#F58b}q?gVKkq
z!Jx}*3}^kji`<L)ey5!cnK{Jik8zC{jxs&+jFZj7UzrsD<M?TjNxH4y{rjLNO(*U*
zVgrEpkj_y{C9p6bt7jXsVjC=eq8+P*?(H8M3P%9)*>Gt!G$QKbr+h5yM~_C}RfP}3
zJJoNx@TBWj>@sr<iu)0&x?pLt`S|D}q;<BpesS}NEA_pumNWXBYtDl8PR&&w(}K6K
zxN~!|ACdWnV=gXuWW4{Ey%5)&v!iOp@R5EtHNBO!<+SNyggO*#V6Jn0!_CH7&6x2b
zXT^Nbdk_!XM=PJE=EwJ=X&nG>ZOckVHYa$pYY7|@YwlJh3^#qCdMZ~uo8xB(PROsH
zl2ZNmb{}9-1E7jR8~N6@s08Ty;wfBPx81TP(v50e<{wBdhW}w?Y{(6|*CSn(ov-~X
z%b5qlWC-`!KSzaeW=oRFfqFTz>T$Vd_eiH&)=qO`yU$RGUhsPAWxYCf(lho49E;T>
zLym}FXwM7mg|OZ^%38hJEPsI~;_>Ew!_IVb?>u6pQx&AxFH>bql8CrJgVKrPM&E2F
zsMafKN{hITZC)L#I=*s>iqaCUc*&KX2Gu5<UHMGqzBmqrs|4y~*Czg{iw0%<lmWIN
zhagU)=T~*LIi4%41ru}yY;FFz&)VD5FLG~#xkm~A7*Z_}q$=f-loHo(xWtW5ZO*h`
zJ5W&tj-X+etX}z=Q}g<0ahf<(6m%2u<+ZtQ0m73g+WC~LIyABC;MW5uiPop_P&71V
zBE&r!cyw&6CZY{$;MTXg$-B@0!_VFmHI*HSevQ$D(qHFjG5mYs>>dm*2vDg^bjk4>
zPFJJl@wgaJdhdp@VowV2eK34ZnVuEXy2`qEG*g(DJN<G^BrwDClskWua|dW3)~}iv
zWzWwz%9DgXDZLYQr-dZ%ySRe)2S$dzXm2ju>kMh=Kj_fR6Lw0Kzg8QO6eJ|k>*KJi
z7*72F^0)6&>G!x#^SSD4Q+<%N$dT}pbNgnR0&1$k;m$C!9l5ok;lvfEaw=ov{jLJR
zd-Z!5974BoTX(m)%y*aHTS)-2spIu_8DeuoWSMV$xv&n>5%Dg>hM^hY5c7f{DCejs
z>9jpfw+LeYllbV6qLdf2EWIN_<YPZcp3b|79c+(uU4U$&VmH(5I=2DN8u@l@GKboA
zd09tRZp{7J*{#KJom}qy<Dqd7R*0^8FEPDVJx9re#6<1b0K8t`T^Eg}Vd0I9w_9kF
zTglgfg641(N<dT@Hh0XwzF)rAEJCwr>p=nJFk)2+tEc2C&E%VB3m@o;T+h$OHHNyH
zgWzP=;3pg)&p9O!M?dpqW+mm74z2aUi2O#m?@vfX*q`m%%#DuwUrc$=(;p9{Z*0}>
z1l&oy9X2UXJe*eUrmV-p0xh7~!01o>{Z;-Civ`rEqEzU3ACvU0ns62MZ3+EZPs1&r
z3lyaY1>0FM-=ILiBVuCCO99AfK+JNf(SH7^_7xk))or^-wi-@CLgy)g*n}5jT)XgP
zm<*52{&`-eL&5iFb0q@z*@>bD*((#GL?sDm>w3G$7eZmB2bSMdWFV<;0!B;AZ2Akd
z33YB*$|xv6?ePaxi1*6Msu)@FvtqEuY>;RYAmYGzi?Q*W0S_%6tDKTs+P@>9cg??I
zg2qJt;+?&BQD5Nt-qBy+?cuswolzt}@5HU2FdDP^CTRY5Ygb;ieI!rK{SO}$Z%3NU
z-)oxQgN0v0N2A1tOZ|<trAH<pb3h0Re2H=^sJs%CMFjB=|013Qwre7wJ!o%#ez}2l
zGge?NEHrJQ;P4Fa_!@Ds1B2q1=j1bWjF}V^n)t+`t27r@dyz^iEr4y74GJXePuV&C
zxLffvMPK%2jl(|R+g?%8O5KiV;L%~@O0ep&IrN-KFkDZ(;1^(DR&J^j5M$PR?O>*N
zR9V&sAgJ1Px<OK>72g?|u!EknGm=;sP{MZy{oA!Fem8IIjQmDN_00>a><v4NMt;4m
zJ@Ku@nBSFs)p(Zd|I)&&77X}83gKJ{8oamnS+E?S8IkbbNsg=;(lhun)YEfZu*71@
z-A_DKOJrg!zYs1IH&P6JmIwpS5zyYY&bIx0vp`^SHm4qttw?uPtEV8}6VpxDVC+oo
ze8=!9y5MyDFRy7^Ct2d4vLZIXx*0X*l<{9iOQ#2J4`6v$gd~Y+eo_5=D$T<5VZW*i
z#^mJ`)jK|D&j<ZnWnS7Ev2bv(qqvG)XWE#_*8=jNV{yGlnDZN3v4IDNFSwZ#Q+7Q>
zq@99$+3J(7F#)rX&-0TnF>Q`vRwA?a9aZ5EkPCav=#;euZcL&|l30Xc1f1P`_zGXN
ztaS>1m~9qANOW*AETv_DWXs00Q2@K*9FmxEqy7LwqE?UxKAo>2NXloko?RS*ovj`5
z9dwo1Tx7@q-5gqbA8fB_EtIQGkjwC8IX-?~K&|}S<q)c4uH7@uPF##660W-l9Gtof
zQ%fK;2u!N@KAlPd$G`-oxc1ncxeer_m9DtQD{yewf0K~Jhw3cz5NU@)a~|58ko0U7
z4lQgC@QolW4`pTxgq9c|IF0=_@6h$mP>ZS^6a4yTf@6Jop$)LmvwTiZy?JQmOcPN4
zmonZH{EE!_2}y_t>kmdeWGE?vg@i&s!mHIId{)Yic&Z(iLY{oOBj;n{QAox=)na1!
zaxtJ#F?S@7kS_*O|HMC>4POTt#3NzuJK)Vi*A{{`eTid44Oc<xpfa-rqzs3_$=T|6
ziz-unq_mxtb>4}JaFAc1m5veJwiR|&x*Xys##h>!PTtE$&vw@&A}aFQvdt(n69q*^
zW#Mxjlc(JSZE@{rM=0|}+wM{fOgZgO8Ucbn=UT#m0A({`(>FqNIV8jS{0@E+=8COO
zYVWwagrV`_bUh`k78rZ4+(A^`2?mhC)D74H-z8Sp*1;OhVb{g7Buy=sn6S2_jzACM
zqx`bC@?HVd!8n|@#8Z?;T%Un>j<|dTj2g`Xfxs7Jt83E$)$A2H2&s%)%CG65gIKc1
z*sk&IeF6V(_pA8tc6RvY$cYy{1G533mFXh8V&`%h^^SK{Gl~026G-{E4A^&~@GW`0
z!pO_Fhlzx<l+h9_^Rg9l2saj63pxyJU*>w^6#zkF4Q+`2cV}gW2nL-k%XRgxLx%sM
zl+$n|_>_>+<a395&7N99frPJf-RfC{^W`4Gw}#2M_I3{N_zhMIbz}h0D!lLbUOIQ>
zn7AF(SLXb%@6G4Wc}b?^Rb}M(`lH9MG=#jS(BD7rs&$yBvnSnGefkiCXm8b+CDZ7=
znQ>=mq3{?HIT1l$vEbz|E;i%l@r4n!7n>B3Z_ix+z`*$zep{LY25tGo0kGi5>U&m$
zpO5}Nbt|0son9KK)vkyl!}58rJ4qbo!Mj*Q+(o29G*%)2*y`8WY8=*w@;7;TMLO|7
zriyun;Aq9eJ2Id0J0*)x<?vec4+o?T@C^a-sE4DW;f~VvZA;5THZyMuj8>nAzWvkP
zM=Qt!DVW6L1T-}L`Rxr-|G>%0Pe5~+xDWHS=Nl^DUf#{+0aaPPlOBSys0|mhEjn<@
zN9u#<i@H(kC23=|{Kc$cb0I;*$7pOa5I7AMzwGMkeIZ9D`)pHTNzKL)@Xkx9b}0X3
zkkQ$`=%vZxStuZl@bfh#t>5(sOFzSI5QE;pg+Mhsk4*4hvsUk6VxS@WXdv0Qnsih%
zzN&=aO07cI|MA;m2FMa*N=$`07tk?!3eFI%ul`ggCqLTq0mOAhnDT=F{{do<y0lhP
zN)~Z&EzImB9uu1YhUn<XcJYl8yFl;#%|KFo+&_Z^$>I!=zg}@dJ|_)@239Rx7)v91
zLkdiEh6^3<kwgX%!!q@B^Il$m$m{NQ2rhCtR0CCJ8z-!m)~4Foc7qSk7l-QORQr-|
z>qJfRR2Y`8LS*(FoM-_0nzi8G1EvG~iMn=*PJs&0Ed>M7%O22VfE7t15%j@k=)xOb
zIpu*HK>!{Cg;Ua`=l2C5&Up1ATkEb`x*CFwE3v&|*J#Oadnr14sG~3Xlb%;b2?We$
zjD(x=ju+t|Hi>~jeFLF%yYP)6;4?sZF{jMGr)#y&Aanh!oxi|%xxa7dQ|44fN?L6~
zveVWeUs7fX4qE-yJp*y^;5OPp;|X*8WT+v}G15xjjHp59Q9Cq248+WBivKL+5Ks-2
z5BUMgC8Vm5tw7Ql6h7C=%>{uE>kZnPHXqAUNqK*nZ(5zN)dys<F+?PXr<c3^x%nqp
zdt{t|JPYg527Bkc)<(fx&qPlb%|z1ptb(GjGRDaLWQ6c^*-s9}v<6M!G5(C(JRWTU
zRlD=L>brZb7eQ{_D*p|hkBPsG8_K!Rl_v^4{JrH}?nZ4V68=7RFo&<fPiJAOXG-II
zWpm^Mpk%-dQI34ahk1^c)oB6uVo2xvvRe$D68PHBlB#9I00vJe7n%)BZ1@UF4M`J-
z2hE<6rmN7BcdjqrvNSaa<kNq8>t-^-lgYm{++YT1ng3u=WdNUxESz*WSsn3{d4HUh
z+b=Gspkjz?>1KNM%%VWw(#W_iKe8+4pkvh%XqNF7Lh~Mkv@~~y0a6d-qM`4UIwJ#_
zKm&CtC1pxVFOPFnKwc^qx>{-O-O;<M3x!ZEM(wDCLIH9FK)%1T@sXsmvVB)qWoV*|
zn^BkwO(eTT%gd<xGo5@-y-Kf7rPDfYl{^dCPF8E%3vGzYU|V*!SU|e9YtZ~m3`YTs
zKm-GB%PPTB)hg*RAl*u|7WL@=$$_kXV+XyzoOqAFL6;2cFOefjRlAkHt%7!K8Touq
zgA-S-4KmFZO+#wZsh^^!;o>8xIUcGSRbe_yqCN7sLGPKF+3=X{4h+&1-qo$2t&CN;
zy^2SQBD<Dtf%X=_sy0wOF8%uy$TI1&6=D6^Rpec8_gE*PSx2%}FM%gszE^*r`dzpT
zbdPH7dKo*LcI7atPuM{;F%>ilc_*%-Qs|Q6%}5XVfl;`w9l5}T<ot~b-xa@gb&|yr
zxBX9E1tEo{*b++17E35~7KJsrT_AdTVO2H762n6a+#KD0RhTrl?GKa+<2}*LJlo0F
z4lUYe1%$V6veOP|y7E)Y6-7KYuzVKoaR_Z$x|QW?7Xua3(+N~xWa&HK-)<Yr{JAJX
zw!v3t!QnNP^5yXW5o@IM1n;h3s<Luy$r4(xZjC5#gZ`+#&d2M92F;)|$l1k&CgJd6
z<f%ZNpyN~4K+@I#;j0!<=j~u-`r9oSU#l4xH<;<E?d5rFlu%~Ll-bkKX`v1|GJ9HC
z@xHWrg=Vm*>A6e{NYbkBabD{2AdLb^;fk;Pwp~eS|5eXy6o4Z@m>;UA6IfQYqdE-a
z2s)RGJV3hGRU2th85;BYwJ|r4OjZ~GIUf)gxj0WoxY4%mZY}0_;GECRf4nL+AeidC
zhx3F)Qpc-tn-K8qZIA=@(*B)m!m*zMVf@3XUf->S3x%*Q;Xr!&6{rkX3+6-$6jgG&
zJV3g{fGd5$b6B8m=I2h-VR&@tU;>KUEVH!|$xc=zoHqYUS^K{r7brn2T`XH^;k|Lr
zSnh!HLHR)5K^OBV+=7EoAeX&2M|5$Vki-D8_#i9o#BE+y-Rob(Lld+bB(+E%-i03G
zzo;d4cZG)L%-wx(x+j11BIi`PcaRFTE}z8dmGenxSh(^kcaM%P-%)EOtR>UrF&T-B
zu{D|3&Q7V#imrvLe|7OowyHV`fy`fD``1iAC=J;<JN&2%;VE!B&Yi&g9+%H2Pab~@
z=@BuAhF_6yZA<w|OK81gGiz-HMHC<k@~bmDmaCvL-KqKPBc{Z_BbFP$HdDiv;m)JN
zfET)t{<@|0rwRi&S|O!PZc~Tz_td_)NWZarohgPd5dcu75(FI+ed9!L6wS#AKTnGL
zcvwm9j?K0Dt`E(`wv~Z|lQ})mZa^byS#w^&mc-n>iJobOCSy<O|1X@<{_~R6b3rdR
zO}1nrLX(mzE{B^?nXc5vSWQU{PGmAZUj%gG9n6f9DJmZ{`l{jyi3X~S5d0N+b2K)m
z4D^NkNo~B&aR?>H;Ek@K33LUM%=>*mxt*f$kFZS@ZoN;iHlWp$s@#Q1BI$I%!To*r
zrKc^Y7cz6b?n_i*NYAfOPu$q#j>z{#1-SdvGofuT=Cs0W*^!05u)a!ngf}B-|Cwsg
z&sD_#j{-6=G6tdJMP21Rt~K3H1eLl0BTac1(lRgt0c{wNw*f{$cpzOV7wVv+Z*1($
zneCWqTT;>telFMbM)x=pu9LT3a`5bWif>NHxfN?CccUJWW8nHA#ahQQH)*V`vO#8N
zLwSnVOaiNcASbEywiBU(lP)c>H?p}^XZtZ;ir#Q(Szq7$Tl^=%x~U6}Jj8RSsIps^
z|1opJ$;dtG3jy9*0vs1gH+ee0mE|l9<c%;FeltC($-(ANx{a%R$zgB4=a$Sa>d1GA
zq$mwcH|y>S5q~O-Cjn>}99#kTCk-nrg*YMMzDR(Z<;t<b=wLI7H-g^m)y%Bkt{29#
zJ@AM{>>NXJuW*mXQV-7glf&AEM=uJU`dFB$86Z$r*Cxp~62E~Gj?9xs#;S08oD~tL
zUoL0%et{3?ekRa(rE1|XFE>JbcsY9w=!KUiOSZF#+(fh_`^{hfL#T=wm?xx$_$Vr<
zrel;(e;usB=K57td>#1lx~vUz_t`z?mi1ioB+BZ0Tv^F}#TCVApp|zTL9{Kzt?AC5
z&cZGu%@-RJ2o`!qtHQj6NT~Wgn$v5V$YKG8U?pF%T`woe))(pwM8w1BD6x4@Oz1E-
zpFQ|}qCEhL)^4XIK>P_(x+l8)7WvULgUzHi1SU=F?EalSK$NyQ!Kc&HnNk%`0(=rs
zHz()(Dvj`eD6D_63XUc~anhjgcKy~|0A+~W^ADq#KFKki)ojf%3v(DK=vTk2(6&t)
zY~48o|La{+?EXo&YH!C?U|zMjZ;Yj~;hxXF`mcIO6TbncjQAXWjmGR)X$^FbL~=;c
zgG%kNMFN1-fP8Q77ik;z@;uu3$tK9=@*MV10Tuc=S`yONtMdcMx`7|O;gFIRkdb-g
zxGCmjMlY=QXe<{S=+QGxdQn~?`n}r{l3+lEf;>>t;VY5%AYeE*1jwZN`r!A>>)(JK
z5gXYzAeq^_&ZGJMozWE0ZKHb%(Cd5I)b<3IG`dV_l$dmCAhSbovSw{(_MV<LL~=KQ
zG!~peGHOP%w>rTAg+K{oFCmmUujp$6R>3Xt{{o@2F>v%{A!Ti-fca(?jtkD-SoFP!
z*tke%;P@uvqJR4Y1}Ege_({C+m-vsN$I+53ENRL)CeMu~94HF<w)oyk`TCeh24@xF
zwtPBj`Elp0ifaMX-`RzAF>;+Rkq6zLJ`D!euqz*LPi}m+rUeN7w?LN7?-=|mu-mO7
z+4(OM55gYZbfx+GZ~0Swzb0A*!8^avXnSob1uAB@SILBsA+ebsZE&bTF31iP3<7?Z
zacvgP7=nV}G1+QiF3-3<B?f`EHk^#7dUpG^0|07B!K*B$e~qxZs><H2&G5qAZOwAg
z(z>m90p;9S#0G?%49^2pp0F^o{y^0K?17sZWRcdiob~E4B?`((YYUsnu6}1^Cx4j?
zGlOw=1}PODKG2=~A?SOR&EciAn<Tq7@(FlxP=S888WaiYrpDWXEsf*%wA`S|lJ_f%
zj^@q{M4)sIj-aryznt%(qy^{p@_VfmsHhIlPCS7fdb#XZO_79Lt1l7Y{Zz}PSEVDY
zFe6wX(sVS>ykC!5|EFS=mGX{sXV)=Yka-P%d^!ZcnzSD}y=!s;l}U-pnNGk!P-#B_
zmgwWBNIV%mH@Hb{dNw`O1&2_<QL-(;FXMQ4+N*rl4R}2z?%0%}Z#8b(7=ajjxN5th
zv7EzX0*p5YnUW>3@LhqSYSM;#E-A|H@ZD+T0TTH6{@A6!yZ=0&&$kbd^)F%t6WVg7
zgfFrTMj*kte^V6;92Qqw?y&hyy<50-?hOr<AcP~sq0Dk~M5>g%Go?WhiX^|*28&Sv
zUAR{6CP-2+^E%yNWUnfM^P+I7MK6@c7914WY{D+$<{7)xAkiLEW7a&mV1#u~BEIsu
zV~6KZ+UY)zE?HA~+00`}*-)F{oGEvs5Jh-qJE8@vuC5AGQh7b+NOHa-_Q2>{9Haw%
z?d>t?9G&rS`O8FepB?buRk0D>ZHu(z!=<+fAvb%PMfa2ELJi(P&vgK><`TabtU4*N
zd0bB5{Avr8Nmg#0^Y)rQznpxcj6<*2El$Z7a3oIqAmLq#h#DbG1`Ha>7loHbn`B_w
z@r4BfQba$xkuC~Hf^vIvqy+@iVE=`El96$|V*pf7z(m2J#Rs%al*I1e{);$o`JR?l
zC+)V~wr7u+`}p{{fca%HuUrqBlY(a>(42&~=%<}-F68dlAG1&|FT;>$Y;5&xt}Ly7
zv9Zzotc>VC!gPAm`ZzHj+dCxp;cB$kmmrCm@PhPZ`y}Xk2MrwR#P9~yAmowC-KDSL
zRrG+?pYHB3QgE2wevNpK5a#Z?h4G+Wz7g7JvjdkmBy{crAcia;)3#<JZj#7A?EXL(
z9;T$x?cZW}@}?cBHct5%NDlW%GV#N59ao0tm5LFj7vr+iOw5%+tLXJY&i)d#T#kWI
zVs{VW@}@Ws4?#jt57^dU`&j2iEMI|IjbfHcrKT3pN<J9;CPUcZN=sSeWIWY`O@5+(
zZ9%WffII-qI%;$8_efL1fR~Vi^<N(#LjoK`%L?8vAk<kzmjZ_4p`$y_ev5KdSGU16
z+J-~xda+3z5fN>mZ}{gWIqL=FhEzR%)WQP(^qI`t4>k_Zz$buhkF-U=m`W~^pHZ@K
zGQ|BRfWP=X5n7@kn|^`s<aj8lLDjb_+?2N<etsTFK;jr2B$utC1V6u}MG;2{#0(-k
z2>kp-Dn*5Ym9>4i8`pnU!F$g}QA^FQ+Fo3dgQMijE0P6$MtX?3wZH#yGjV4d`t76Q
zMoI@SxRR2XloUEFDk_SQghcqO*+>V0o=4;P;VL+@ci5ct?fHXK|F?UifJ2%kH?rsJ
z^Q?$y2qu$EF-T7W8nZ)_S8WzJZR>)gAaxPaD2jW?V?9c^xW3eK8KPjc^5ag(6^$Xt
zG(4)jC_$|i@)VHrVvSYl1n1T?Un2O03RtjR{6hR9(H`x0^T@;ur7-_f#0Sv+P$9`k
zLB%qNn!qwJ$Tb~-2IL{$zM(Hb@+w!a=O<0^!hO?^cvJXc(BeTposdRV90~#J7@+}B
zcTP*;xY~8o$p#(E)PwF0hx^qH8jkai?R~`{i$G9gxg~k^9Xmszd8Ajq(C%hk=+Xk^
zP_4k&tFWMUiuvuX7k~$5)ed~19@&zwJz)ye$-n_uB3U%m|MSBQ4~WXoAB3>)L(dmN
z-+wS{`NpGXGC@*!a2jfWRM3M%?La~Q!NklTm_T1Ju_!WWcr^-nKe~gt_E^NhoZ*sH
zhk}MVXgJSuD=*J<W2wda(f0cM6sG@D@;)|?JEKf}%rWjmWBFB64MKu~IVx&eG+!kt
z9Crawq9c4do(<1UybVfKR)Vdq9o!b~TnbE<xL&;lmp(y&(++CY0Clo)aQ<!uEl{oH
z))oftH4AilL*UG8snf4TF>`)UPz`BmbH6nvz4n+m?4VFW*pIv4P*jA4OZD`&HVxlX
zcNcK0IL_4GRKvP?(bm(Ll&t3(%4#CqlhXC#^Yk|wdgYQQ$zQ^8NIl}&ei@)KZD`n>
z%;jwGZOm_t|L_7QZ0stEsn{3;4!&Ukr8$X#8*nhtzfR^>>YsBAK)02Fz14U0@p|`C
z10*-DUPo_-v*?-WTmOuoh3e(!24ojK0pe-4+LD{xjhi0!L(XNvXJvWCcAOkWeo!v|
z>9VP@Ta7~Kyyr_$8Fn#Wo4*ta24QB=&A~(`WT!&iN=Q1Fm8`t`>aNHoLPbv=GR~mF
z#|?{34ypMh{^=6uKSisvLO`{p^h$RvgFGiREb)`hyN{qO%KU2DuC&~1Tp8!HmZmCA
z8b;?b0OulbJN0CrJ-v61eFWT{h_v;^cay)B{)1c=lHXFf_-ZA7`)atKk^ijLAy(;p
z<~A$(EtP}W1mt7lhi!D>?RC3`_p7NPKK`e%?|gp3^!?+X_fzc;VWuB$O|9!`ADjl%
z{9H+Dhz*y~Q*f@kq*tbsVxnv6?8cN)9=f&It^fHm2ZuXY*f~s=;p2BgbP1j3xa&T+
zjj-HQik>%5L~-GIjw5(Z%oe(vKG$?v4_Wgl{=SJ&rT2k+UWXEk;rNPtkNVz^E=6O+
zj+hgES<-_0eA}3Pu8)t;B;kB!?iGUL5r@Lgm9=W?KYdV=4FyggpXK0aYE~~-5scij
zhOFn7Z5qZ{>yD*X-DJw3U1w{X%uHuQvOK~Xs3sNQ+CQ!sKAuG@r8Wu~Df*~%T_j^7
zH8|D)`acc7(tbp)`o<~Pt<-AqaI>iC6mw07>y|u!8d7xi^(F34SH?^GrzPxM+ml*4
z)y@<LEUyEi^k>rQLwjy+843zPrej*7H?1N4j`104nPh_{xua%1A3s%_4^qZmR!DMf
zb;9wf(WJ)~23O39aDK|IuWKc`z4eUaF6e({1KCV7NfFOxI;*y!)QUCql8!v*RO#}X
zDO_5l9id@Q6+5i0vkRayb4g~*WhX<n$T*`4o|izUEh*tTT5n)nR;;$RJVvvszH|!O
z94~lX#VEzy+^^abXWF`5u~t<%+ljzmJBiLTZ|*gG7=~iRfhn#qEu<87^v5|jLVc|o
zn&%r$BO7dh<kbva1ZM<X6jdefnb-aHhFv?h<lDz*+!X&stCBtRSv~735Y(!={h`&?
zVybdmY5w+~Zl?9VJ_OXS_Ihh3BjYkiwBR`kHr4MIoG$q)*gzku+?tF>mD1<(nbz(%
zM$1dfs*uL~-D$fXaz}~y(a8wA*QXG(ww>rQ=(@_zbr4W!B<;3eAi&p1>@h0NgQaR2
z-lkQ*>zyR4k^eB;g@;@y=T`F|pK87AHZcnZ)xZ6xWk{q%W59<nr)@QlChTfn&e!Ca
z_3~bRcA`_l<8=XkE#-=n`-^U?0lp&<PqPSx<K>bC57Ryna}U!B({5(;LHHXnl&1+(
z%yZ5np28O@lBAF1Y8UvyziQ&Q4~OMneEg`jcbbW8rC;%sXR|<}D<W1;afdr2?Oecl
zGtM=cvF3MnW?SIk_gO&;nwQ808n(l&rOU{QxYpifR1w~rmfX2Xda>SG$!PR8?q*r6
zigE8)_Xj_h|F_|9>bFn|w_N;2eBI@UQ`;A}lP_hHPD>g^lut-|#h9fha`lrv(;Jh$
zvT>?hFNXRX-Bv$d8tTlG3wc*(%ux4~UPVTaa56~W-WWws_M9mykJ6Zr`G>c*C9=@<
zuXfN^yUnf=Uydj(^Cr4EO%n<VI0Z^{fTp9cA3F;*p?h==Ri%)jc<PlEcaZg5dDCI3
zf1qL?4VzYUR6&|<4RVcCW1eQ!j>^TG_DmISu44-g2Da*1Y*YuuRa^fQyL;tmXPRaI
z@S!7T`+{auimO8@Jn1EzQj-;TukuxS<fxpxGIZGhneBXwytvG9ME=V$sbJ*i)X+qg
zC+)Fwxv_-?cfyR_j+4-<Z@%qBtswhLdRP1T5RhFyA|LS2V>$*61;r$HleRZjdP-TC
zbxG)5-pPz$V~tEf)#})DuXGOUW%^ScHzPe>s=6x}b-u*ZRcWZda561SQ9l{BvQwvj
z`TY0BACdeaItsB{H^X8wBD-86&DbBUn>dkSPH!Q1dE-uFLi1;M*_FYhNLvNQSaEeN
zos{>_q%K81+}gdj!d^OI4pJ_4(tRVGuVlj__G=0^@0Q60#Jq7c@1_&vGLFDVK;=)2
zq;&Ug5EB<$k|aV+i0+EhOC@Lv8G5+b-oYscc0VI;KZ`I4I3EjcQpEA+yziH;Z73@3
zp-_>Q*%juL4WgPskU%k9%;#=?FwBLk`6Vv?GegIJ8ouojfy?;nZ%HR5-P2icq+>Q=
zvSJFxl&7yA7-y0`V%DjF2aH<v_g$WD?sbG0U!u{`6sNdVSmc7IFzC&aXtbE73%(Wn
z$%0|PU9M8zQTn2y(bF$NU!V1g`Q9^=ec%5dF@eg}`hN3AK$x%MLs>a4@bv!W!BSR&
z4f5oc&9il;zT-RG@fb7K>B+n)ay81F*SQiK)5X5<^W+*w$1xUsntytP!n}D>!41yL
z>5Vz=X7s7&k4D#=m^xf?<HgZ(F>2>N%E;EQbKCHQG;0{-H7^tYbPldXXPCFHFtbz5
zjB=}0lv(5p$r%j4<keuQ(9YJJ(5gT6HnzFcj6b=`*(8Ic`=;$`lBx|jB@h4Y*F9fr
zcY=e9*qs**OIuYhRIktdu(>L9yU34MJrif!ERrl>;O1nVRN#7qo<CkI5iP+re6j!Y
z7#ir`<l-#%rTFy8=<PwpgoD|&t=!h2Xl{GZaKn9U6;l!Kq+|hwklq-hWAT5Vpf^Fv
zyaXFcwZmu`m?plPEi3rTGQIoZ^vztu)*z_#&`~hSh6LB<Yq5_S<IK6pOnJ@Z3dxZ$
zE(A{v)AQ#{@B5_<M3=p5BCTMcpo0``n9sxz>K^JG`pA6&gUuh)xx8z-mGjaFqM$e$
zI3aagi;uLgo>?uMsZDm?NGgo(dWqTNx>fk?!Nc5@w6d;mJ=#YT*uptxE`pwJL(hGu
zwpZ({+5=LL_J5I3NrhY?qCo%5Tt=FIuYTh(20Wslia|+F=r%C532Np-bUC$p9%ZZT
z2y3hi)HWn_4ktd0Uf*e*%jw@9Ek)ZM!xF3>GhLaSuL@a_3WIMrr-O&Z>T)T}ZIcnY
zt+i=toH46iS(&r=@Adz9^q0zlPPJmRNQ-A%F37{~8*t8}So?ZeuGhA@h;{1fG%rEa
zi*tHG0m)ot2;0oyLSJf7;&^kjS@JgZ;|yyBuV;^pfRm1e*NsF^)8=}4c@bXHDyv`(
zy;nPn=Sg#jn?p;!YRuz*@5k{8Dr=|1dMkm&_LR-q&}Z_a3zLVXd`2P0Rx|JqIzw4>
za)b_vWc99<8>;hvMyls)n6;^;K54-j;`LQrlAz#$cjowTg?OjTHL$rQ{Kj%~zW)2k
zS)b@!82>EWcdR>Tk^XFwjoz$npT!?juV_71-54;K{9Y?VL@Q4>x2_^<m^P_9D4Td=
zn;keOEjNbRt*wpAET~^SlVf^uO=J@Mp`MAMSV<F8YMX*&V}6~9l-EN8OH7=BiyUGJ
zT)){Bqp}`~sNFjc%rS^FNy45ztM<!T<i5SEeqSw@ge}y2wOxCUC0ZNr7QwP{_b_kl
zKAhmWRL1&Bm128U-~RdS)*NKD`Ec6WxoC_T=7^1IgSx!3!oa~Xv_Xi6ACn9<PU+yn
z6?otS`7Ot&sHmz>I_x9Q(aua)=piYkPd|Q3ubL8i7zYR9&<<xvElT%&`DQm7y2{1C
z+HkipIxAj>g@Anag%rcQ;~kTQIrO(mDJ6=&45j?{bKc#cdfYhkJ{yTyYQn?C{o|QL
ztAIA22p$|A71e7n1^yTR8{eu=K6$~4?lBn$hlhBNA3u*z*7MEy8vOM;RpKA0C)Xjt
zAeaLI#!u6KpHmKEaNrkQmh+xri4QtdyeLi$69!{*c+vBgoN5I7P=$rTJo|eZ2@lll
zZ=GJNfyv7%{?xJa*Wi=W=YstfuZh7=%e%X@!oouEF{8%Sv#aFj=*-$$j^t%AF)_nJ
ziQO05PBm8yXPN}w&hvUdAPmq#_~7((c4jrT^57+;o!0nkVoAO6`ryl(AS$V;5VAr;
z{Pk6Xb|XT+IK5}O3$L;Wqk9ww_({2&-``$VHqCXCB%ulI<P~xdbHfQdbO*+?t-QF2
zm<H#EFq`*Yk-O-6QxUys6_)ueaF)1ucsqgi<?Q@?B&XgRMcunP1iJ0|$Y2ZnCOf9M
z%*j?{?o$gA>U457H>cnN(;=-V`uhLRPFF-_2{!%VmB`NktcToYPIJK(t{Y7yMLe*Z
zlk@HWcwYWMAA&53zMvZMu%jbvi-@E|MB@}IKohLayYCkN_*JjAk>v3I113TL@r~yg
z@E_lf)@@^Vp1A})TIF@v^pPNNXez_{!p`Y6z%DhYeR^9cb{FOUaUnl2<e)vTWAiaJ
zDI!US<ph^}IT==GkCdr=$ofWV%&S;r0##m6(74okXDZWg@LwPOduk+Ui2tsm$oH9o
z!rOhHwuS?!x?ZP59!2Oz7Ui~IhjOxYqcbB24+lTTBU%QHz2s<n7Ug0uQ?GQhMQQ?5
zyzqMW_4Rp%zMAkq&<_T4d4jz75im{S3Utp+uu)&D;zJzw%wAs@@7}Pt>x~qB%8;*>
z5u7Z~ZjAUpX8n-P<)is~4$L5Om!LP$amdhtoOY@0jN5hbt@oqkSJ?pja+!}S)=ia-
z^fS6=Yg;lB`RO|v{`+p17vKwtQ75*Er*psDKi$xG#Qpad!=kvgLY+4AL28m&T+`MS
z2MP(_>)?XtsQ0x<(eUk!Oc;e(6~b@<IJvremr4E}BVLjavlNC5I`+T2n?Eu+mR^e8
z$7hKE-`I{qBm1SW-nphGhSj(tH_0l)iBWGj?XrCpmDF)+Dw8})F5WX?mfP=-&r#8|
zNl{_*KV3{P!tdOXiT?fP&RqEOJ50zQ9zx4Djxd|!#Sq~QFmpz&T?Fr?oN`rF6$U0|
zL~?TSh~4`XAD_E;`1sOcfL1ds+2?tw7nPjMDg|Cp`mfPNMn}_vAt<NOGr$a=<Sp4(
zfp->yIc3O)@=Sw{l~uvILOTri>C+E(#2U<?m~iLLorBZW#WGd|Ka^<CpBSdHDWq*Z
zS3k4y0G2~^dskKD_3O`0TjMEUyqE?ls%i_ToKTi1_>c@UX{CHO@%t9If&?yBdsZ1{
zZqm=1U!FB%yY#`NYE4A!YE8UVt{fYUOGc_!#Y>EP@`7PMh>45`7$3vVDp!--<sTBT
zfZ=3Gg4&7vj>WuBUc++1NHVF@X+#K>+1cKL+eUUKc=kWI3Ars}1_uY1R#!8RH^;O|
z(L|VET8uK)ozA)g_>c_2Ki$Uy?}zC%>C;80{JA4y7NH-PVdnS%mW8FHtn7HOBqgQP
z6~&B)ylo1YT31KMKw-lTgOvIBuar1WGvt4kiQ3y2V_;wy68iBwti5-91}pm=N-61Z
zwx=(p1g5*;J$a%JURM9v=bv8aU>nuq%5cJ>pLE=zhawtA?!bOCfQdkZBsR70c4Fw2
zGqU93xw#@9h*%+({ybsUEPn0p@6T_)^p&N0jpBHHNOBDBo}c>jXMQl+kNdod@<D`b
ze{0)jJq~+_1)MCoXzwa&$7E!;w7mAb+^(1`vxv60x5ovuxU6TY&H8_ce)IP);-`R-
zi#9&|Z34w+&_{PU_Oih|Ld@bWt`7USOnMVOjAdqIFbE03_xJa;e!c&Ab?wTCdgg<u
z)1w=#cHA_Vzx$)ymOs&LtEfF9EKF(ZVt7~ujDp+fVr>`(k2V)fOV^Be?9IBMj1Eh7
zp7T6i_am|X#ed&h%kbvvJWjy5;OhGNc&{0I6Z*zjps1+$9Rnm6YyB_4pu0aCUt#8N
zMMS*oFE(=7ZG^#n)5s^#HNPAN!{mu&@Bfxg_BilYQc?mtF_kvSMQnSrEWDwiK|Wb1
zAB?A`2Uoq@a9hvPaAnVBKKy5+61KrRI*&SkXvq!7xN|jDWYEs(yrUyEShHK7TM1S2
zf}vypXrbP_m*sl6axs*A(=t#4qmn{LM-NNeENp;-so0Y*VdmXo)G}03;j{{BnevIa
zAi6cbQ~~qP!62zO+fx<MRMa#yT0}VBAwNdI0`qoVogY#sYRZIDTEimRIpvr&eu;#g
zaX+3HmmP{|UieN0rvDWWvk5}(ClOReCMIJVEU@|j<OZ4N$uKi-{e}nHj2G68{^f+3
z*MRwfpvAJpn>RV&3f0>0D2ih?E1vGmP)oU=?l3#bDk+6gWs84##1-|>d-rdRFK9yr
z##|j8kLfXTajAmftay2{?byRLmIYqKj@i870y9@MHqHQh7NU{@)A0tuR6<VZ#=~l+
zS~VplWn;Rk#1E55X=Q*P=Ft_c=S447eLSk|+6Oy}LC2w?qo<d$sXxz1^13i<ZENf7
z>r((7s`a4jvU2O09JPEc*C=zC_l99i0tK0O$i>1J@}o(M!bn&5_F&TEt&+atB(iz&
zlkKU@FJx}e+jsBY5p$VGWXQ#<ZB)kF)N6u~eT5UYFmpUGi{ahRH&S6jH)jSTx$1p?
zU&G8NJHlzX+)pe$Z?8{V3$U@UMz$;F3c!Br=<DlCb%1%GOYYs>(|qybh1>i`wBb5;
zXD-uz5kLJu3$}D@Y{*%QLkhyzX=8ejE_69Hmj?x7{O|+bK_*N-fj7l$C<Ey|s<+Fh
zz}(=Bt1A!W_rwW#R^u~i{OC;-$h8<Nz^A?cKGJb>R4T06X<NGL;aL{g{nr<ph13Oo
z9UbqarN0%Il&FHO@i6Sp9v5yc%tLLUTGQkq7MAMm&Gkr*D&q`vd#zRDl+Vu24)G2=
zou~X3Tx?e}voTSUxl?~>-Df2#Dq6i$y@{MN2!?};l$i953hdSzsH(v5142Tg9ag@2
z!Rk#Wij9z$20X!WCu3ojQCAeweK#*Cii!E$ocBTIz%t_44E%^WP2W4}z#N~^f?3&0
zoZvi|GHiFcY6!WvAZZtu%GHDZ6mc0+j`zP|QiC_wm&*?P7E}b@lhn-2fW%q{GXoPu
z(Dy`LU0spiM`gA#oNYc;o`3i5UEuP>#Kj#KgLpOyYIc<1@Fe5DBr7{DSiLvc%bki2
zPt$Q|<uwxSiGYKD77<t^+)R$Lrlwh|_5<h>u(K#5s0$3+0DG<JBF-CovH@zK%9aYJ
z`4$)mUN=|oNu9{Ii<T*LS8;|8%<9Di+dY!1UJ3Q=`|sZ(TxNr>!yLf1D7EuG3;wWp
z=L|o(D;>B1sj&Nxo|)Pa!_??5?uY0T8U@o=_iQ_jy`_QG2Z76y216)Qz@@wmn=FT{
zEWiu7vP}hhQtz&wg3U`UWj>M<d3DW~cxT=MxuKhy#E@Ub#>NIr@BPQrLqu9eM$|dl
z0@#x>VR1Y*aO)?SFx8Jf1l|gr-QB}m6D4Ifb3)6jtL_DTU0tAYuf_4I@U3Q;!})S_
zuH9mrTI;>{AHk?jdL|}WC|A93bh-)=3C^&scMtG0suv7{7C_Y5<LKkCv*14ATb(e@
z4=oP@!ePIpRa7FhN=;u#A^mQtNgr924vUnDkb(zr^Ko1jQB-h7*f3_ZAsZn~Xl*!K
znV82~r^;Zm+$xs<?((tlJ&Mp8Sm(G`uJ%RCguPDfq|SREA*?;o*w|PQIP$2FuCuYR
z5uf(%`$*+X`SnILaus7^<5+`><IPN9cZhqe-sY0Xs9|JiXbLvh<S+eCCp$B_wY9H+
zt9?t#@6hjk2gPB+D0<%*b59@O53O34qWaq#chXlbicg+A*%-}__roH|Fz$)dB*ped
z#b;sH$%^GL$+4NMcRL!8%nwHS{Xsrc85s>bBN-#Z!YF_fSX(o7+!}ueY_?9T!rCf&
zbnuQZB#QRv+4mW1fL6xJt)}vV7Z15@X1C5g)6&y5Ns;5-vz`FcgZ<DEaN6pB$XffO
zvXYm}a$FJIAP9i(z}`CU7sp24^5yAnv{{^*WWHU7S?9U~_^aj;oR`O>K+k73n8qNx
zdL0%XE_g}9$9JBv*mvW3_V@4!a-2_n$J_?^o*zz~ot=YUpHP-MtPeh|0FQ=_l2RPG
ztiWU`rNI73dJ{sXJH)p6In`%;*<Ldt&554DGJpk1aMSrCU48_JfIr9t-Aq?|6V)XD
z$*v<l{qu{xo6BAI&KYlSSRIIl#6ljGU{RSM@CU?g@MxO9U|UR;TY+5>3UD2hJQny&
zSpAoO-`hrF<HABqO`Q!!a<*FYV2bPO>w`F}4B(98kQ~p5!)xTLy?K#StIra_Hs79W
z&;w)X+)tk_TO08{d*pew-(I~OrHzY&^AkC+)@n-abPj5Gii6Vy%{U#d^d#X}r>Cb&
zg*oj^XM+v0IVi)b3as1hcq6@H&P$t*k8ia*EH*ZqS|-Z$V5y_}ppy{~4=)}0S=KXJ
zI5e_IS2dP@sWo!R1AGl~SSVQGeDHh6ooP+hx}z|8fbZb0vO_B1g>QnNWoKp4jTGwX
zi9=r=sd{+Sn$#_97SzZ))#y@61|L~E1IRV8?F7tk7NHAbp5Xa6uOyFy4uo;mmI3l+
z%>{zNNE=w|xaIBjrOo^I?<0F^z<k}HXM9SJzWf|N?&#={mX{xO->u7hA?`=_uR&0k
z3txWL2Dp{Oq<2{GbXwW6`vN-NEc3c{6TChe(kd~QEArX11o^f(fLGZS+p~4rHO{-E
zz!8j;m?(H2_V8d}V;BArFL<X~1<4H!|7{K6a@m{zU;uV*2AKG|z8tML+7?7I0>Uu+
zA&5qsQQFo6;8g*-4Ymtn%sQ2dx;0KCz{Zyx_<?g))zE-6EOp*RQGkoGuBaJzT;-?N
zHTq|e(L5F}a`kZyN+yQWWeB7OH%4+D=YM~c?&AYt6N`vF?e+PYJAlu_BkFp(sedl(
z06wdDUFBXrl&OFWZ_)}1g^IHjn`4FQAY55oht{grLBdyi%}-m`CcwC$i_=}b)Ch?A
zLtF+hWwq34!Ksy=crL>~kC&I1BjeLk87@KZQy9z!;7b@x+U4eaRWMq+%sjoF9MXyB
zw&FMWs1a9M$|)D~QZ#J%hbW5Y*}@lqo4+T>XM%Lf9)XAi^O5hM+<zN3TA-T?%E8eB
zs}c0dNC>Z4V!RF<n|;j;1W?Bqcw~Yy(f|R^IF9KW^|FC$xCWkfpz#wL@TZ%6Vw{YO
zGT=tWRMmB<*JYnnd7K|S^qS^M=Ret1RuD|z+p}$LZ+AZ_99kcaV%EkbAz_q4;_P}b
zd|9bN;qBXg0Hdf9`K0D18Ztrt1QuIiV<<Buid7FP0BAI7{N`@}YZSoE=c~zrW7hyZ
zv&)r?_kpX;*=9pN1bFj3VC1dM*us|54Es1ZI0nHaJiS+7255I_X(^(Y!|TdLIgAJ-
z2uZJ8i@WVuqPw2)+Ww&LumniOe=0mI?A!8qxs`VIAucX%46ucRKmjmD^jsKPn_FI=
zsi~mvs6Ahammw_y5vT9k63kT<J4IX^ne{ybz+oQMMvwGz%(@~TjI2%XMF8vQtGwK8
zXc%UbxXiD;{%c>=4cua1<>1MFFsZ<+irw*|pWW?hq#;Njw1@xfpA|e_iML?@kPijz
z5tEarn_KltckE9Hz~~%gqz(vB26K+H5nvU1VmTzkjy7@&qpEiD9gNMsnk2ct8H{!O
z|7tq-cqsEdj(1+0YMtGBb&>6~^g3-VMrIs|a;qGdGd9;qBkZ9=5qdERgC5n9Qyqk6
zt$RXZ+{$Hka@#hSWoX(#guxOMx!)_y`TXqb%wIER9?$dqp5OERyg#4knfaa<>f3g^
z1W6o5n)>g=e5Enls@s46n01JcPKt8a^hu^zob#Q!ws|10XjR{%d6!GrOnputX&&5H
zjSig^Xp=p-h<3{&TS1n2ETQk_mctZ#q($qn{TUBsQbq?Xfd?B-BoXM>I9G9s#bR0D
z=<$kc@-zP)Mbp-9KkNzcx)U9(QNGdA+Ip#Ai33-nrmbxcJxI?0O~`H9)a=I>`tjB2
z^p~X**uCbKBYwR<u}A1nfrVO)N`@SJVl0S$?be{2RF59*N~rhazmL(Mym!U3`ZWr0
zz*L)jM>$1(KPGeSEOIeil7&WXRho5HhA$J$)ipHG4jjnt{hNYAQX`oi!8<7)Hb0`9
z0TNjKxNy4m+Uk5~?V7_QFWSL_ri8K4FxQYN9(aB;aQ>aGsfkH&g?^<0TQ7}*-<F1F
z3@hX(ItMTXJ>5P!<=!0@YHDiTC^>oOU4ws@6KyY)zJQ?l@iUf|lWkl+=^7YR_4?kt
zapM4!p2(~7y`b3Gnxe}HQii!g=TKi*+FpMJ+y8IedZ|=w(fk6CKPCz4@3_mTQ5`*H
zyEvMgLG9k_=rWw-_7(Rpa{zo-8^O7AECD5SPxg$KBr2?D_2pm6;fsrp@1qHMC|$NZ
z6kG3iHzsD$c@3iGkvtM2!+`t{k^v!C)lKM`c)i{r91c*`|J-lg&%nSR2X1aF!^HOR
z@MuCjJ8?Bs>(JSffvUyp4pjan*%>nck#tsY%{sm;{L`ZM`W3Hbd3jf~wNi_Wj7%4j
zn(hlTz2Lm|Ho-c7-01k)@=(C#{tDj|5U3N+6tn&jq@F1Tf(1&`f{4L_L?*O)s$YG3
zA@Gor4R?J-Lk-b}iPF{-$H^{3IJVHI6u(rV@CdnN^Wf+oD<o<q4dDAp?nrj;+LgWi
z8?%4J_ffL&LZ|Y4i%Cq^q;3KGLLMI>jQ3#O?Q|5^riz!YxfK_cUW>m5A0bxY985S<
zJl0*y-46GG;rK4z=;OO@R%eBboEUbc<QY;6VcPrgoh3@tv4HF`W0JN7ZXu(rnt4H#
z&Y39hEc5Q5<>cn7GZ>=MihQqRz=<7Nop0KvTmaDGRO~eh3%)II4H;Gt4ddLmOx9*X
z^iWNgoTmZ=z-cuYS#MCa)S&>f&LU=8y2S=$opJ3W6KrsXxXn=tr#apKVewu1vtQXC
zUg}S^=R5H3qv(eCG{)}hEMS$kL)`wX=fz@ye^n|a^h$~A<n2~k{<Fo)9#t!|!gL|g
zG~DT9P^-f4DZ~pe$dpq}ScqWG#LSFCw@%Y@P_asVoc=TRGD0zn2=_ssu9B0n2DAU6
z1_uY*fAT*M^B(#GqvVYb6xK0T*|02l#$BXCB#mA<{&WZX$mKrgePPqrrM3N>_Zoy?
zFs~57!V&iOe@~F~x2z~-YQ-P$`nWeN;af|)lP9nH)vjK}1PU1eMkna_^{O(FHn-KT
zEfnwQ>7KYMip!#~myXWPsmPiQVl`+q3)mccH@8gmdbf|0IG)1OR4Zu+w|=T>E++R2
zqfYC%v94M!G2s_~j`&l2Hsl1+LBm`q{_|<bVwrore0_a=9u8ytft13xJ>pldYzX$o
zgDdUe{c*)pWE4Exq{3V9UvDSbM=aI``)-7OMSBi=-Vgbd7SxwFEKF42zyG1d<YjR-
zlY{%`nyac^t2wCAcF55zs#l|R`=)nA<$NwLFNWsL<6BKn#4Lur?Vhlrj8cb%wZcO$
zG|T?g6u(1wKD7|1l@t1zOMgYT_i4MvwD5li$2|-m2oO;D>({^s>Q@T4;_!H9jF0>H
z_Sd)|sUpsnq<)wyWAmeJaF5<^p62KUj%CB9drc2K2PHi}CanU{a$&^KGyYw_Y>ofA
zfR%+sXquWfiY`~;Km-rqk@6_n+8Fdfz}|B!^SJDTXsok1!?Wc_Dw9#OY-}F$wQVBZ
zMVm8!QNk7(<~tU<CiQIIy!jH?H2FAHgAZmR_T!b5*i>^C#M8gOIO6up#>l9sOQLi`
z#j2?SJMJewZKIqVX-e+&#67Yve95Ko>=R?K99dDrfA?gekGl}r0^7K%tE<aSItr7%
zIR1vpB3S_(csC*j?x8dynC%L;SXa6r_$M?fyxP7MF?+Wv8$3p^;gq_CTh5kR0U4~~
z%fOqG;Gt;PEG$#C4M`x1NCPPgE6X$KvdIo!VxSxTwt-<r^f!XW5Bm_XI<!yricz})
zKlIl&ZI5K?5kdy;y?YN(!Mq_I@!4)KZmDT$i3jVjWJHkrm#Bgp>;*V1hBI|LIfJ)V
znw6wLsL{4u{3HtA!>s{Z$Y;*(*VA)?hdBMvoLz#~^W;bCaUK-JknIcBSXY4)+`TwI
zbbPuVkYR^O)o#a%uK+Ym&CPjmZzRGYH3mw`%OVT9wrX0omFZzL)*l1EJTdt!f%+_I
a_kyg&j=DJGc>E1{gK27IQe^zk8~+27*)^~L

delta 75259
zcmXt91ymJn*F_pBr5ge1?rx<)x;q7=J4Zl3y1S)Bx~03jySuyVAAG-`#agU;?=W}f
zne&{p&))m|?1cN(0ar%V091tie9!x%lgeU>M&+~C=P`@o8dJ+hs_H*t)lw>Ox!22u
z7ANn1At0dfie3*74K=k0qL9Sdsllh;960I89eWKsH1DK(kl%QuUwAcd?JiG!r0nFo
zst+pQMoxCl>}d@$DPgcSn6zLwi}+^r@aImu&Jpqoy%tYkU866-c?l56Qz|t`OpNzz
zG-1IVF4+odb|6G^XdC-<13{cdR@*+J@ND&1TbcRG|9HVI^ZOxRDy`+S?mKAVcRvzz
zUY?&WD6x3GA6=smmKSPmHja*Tj0I`pf-PrC-te&(B!&!Zw7i59aEpXs(K`lT^^KLR
zt*ybfC@ss|Jtss(y^jNERDLPuN;&RN2Eia=F#*q>-Qk4N;Cao*etj<QQKce;{c<{-
z%FPPQ*5>tT>P)|fUfar6Eu5>dXDCdNAytd=ld=BMU_(pLxNm5E6iUjPw7JXltCVOR
zTnd7PmUxDfh|HCp%C6nqgwZ`wEp<Mx(T$IQ3hik%thiF|FyOaKaL$nP&eMXpE&uxc
ztvjc{gZtdZn4O7?y9KGB+}=oCHg67PQCQmw!s1ZtAUqnTEqU7C15S+w$F~jBFP*1-
zE$E}a3RM>r3eMH5zmjUH4S&w<hce=3t4ctV<mit4^W(P+i%2wg*%c?sk*08I1m&1f
z>kqZ&<L$IwCJF$>;_Q~+%n}Ltuyu$)#zb+Qkqy~InF9OEiJ88KivSKs%rUXIis8x?
z<N_L|-FrDoVHcMrKfbifvKD7N_7{(ZlScs?=((i5@!qv$4*4M|8nVbJxSNXJrpW%J
z4lQm6S6Zujl+~D0ECY)?@%3^H^3Elr=gu1hL;(tZ0~-JmnV@Y7Dbyq~^ZO{<@ls>|
zIgHV4-8P(<oZjc5`yuJW25TWL>NXDw1Rjg{zl_QVJHNp<P^|SRsx3=;epsW{Dsg5O
zAA7<3et>V67Ax@N`%cQrTG{Jds9OBXWFY=?GZ7kEbc=S2_Y-8Po15DfW2r`^*iL6b
ze!kiBIVvFd*a}DNsXrRk7!X|f(Y`KKAzO^u`A{9A#(Js0H;PQ!L`!7slh~(sKUjow
z`!{;1pSUX?7(+vceovgy?pG0qBwbCwEI2rf*WHhhia3uLf-%0>=p1x*QDb18wey2r
zB&qyvWUfe?OYS}^jP;V1kzC|Ox*(tfq3jXG3<a>+k~79p!a8DfAX79t5odOq#aBG<
zokD_*4UCu5SMDXYy{R<l(iwja+F2M|x;+iyZk#f!5?fH%G(J0;J%-L$M2r_3E#zxF
zqL3xH(%X}o#R;=lVk-pFJBKVuLX(@{k{R)fF*LU_q|Oe7D%fQ8EBWTvIfm;x#8QYo
zlmZJHr#pnoJ!l$rwgj~$I)<B<Q(f09IkKoVC8>y7Mwx{vv7wWyzL3vBJ+qy$0pyc%
zDuq*Y58fEOs*I7>Ss}ES*U*#``S`TTWhZZn8VA~r2L8G|vFI@_hJ@+NB{OZ9c9J*W
zHIhl5jg7;E_+Vz1i>{F~lCm1i5w4Fc8UW*~(?}VQyv01&f=N;_94BQq6KJyp$VUU1
z-pA5tX;3}r%#~&BrDyp59Ok1{WO=4oR4A0Kfx4@|MEgi`HG|Q=j^7g=&0zma2$P|>
ztmg?&PfsD=ZVpOYSS%)gi&(DE#7PM}-)=l+thi>co;W%<IJjT!cimqdXjB*j2u&V0
zd2Sccj1(^kNlCc;9@oYl0m#rWFlPgdEljSb`o#~_!CC8#wrh|D1qD^#<|=xjFJt;H
z6BF}KaL}dnl5qZ9e7!6Q>MOj<ReT4n4lbj(^NW9+A9h+#M&=st3RdzW3=?x@o?K7q
zE)Qj6K`jKt2MEbe!pe7$*F@C-Jhw-rztTPSZ-bAGf5VS8$jM!XLiH8dZ~sWG_31=L
z8(XCrND_WE9(h`LKZrA>f>*3AW^9tjZ1f46x(<IAqGi&5df1=RMY!Gz(8f$u$E+?0
z-9qfP^?a#Es|6F42Afx8?BZ3$_l$iBgr@fBh7VyMj`fJVe<PQerapQAPy@{~bWx8L
zL3J*EN!bmOeM?w!rw%%ME5Fb@G3jz{w$+K5;_Vzy@TTeR>SwK#3kfBjD2_-d;HH-+
z1SHtYHnBE;?@l`4)SzqyTQz%4c~aL|8X>Jre;;W?{BjJL*q;l18eK2nk&JI3Jb3qq
z(oJfJa-GVmFQ{CG_$)6LShP|@!J6`+`25l34UJZN)&6M`C(Ze-`MwZC#zX$B)uh!V
z?27x)sw&pEv9*vOJatd>Sc~O-*7{yF+UXjU#=GU6$U@%?8_5H6OCz?$fjqemRlG1c
ztk-zJ*3j4g;IK2iHCt}<NnAW8EiLV3=}WO{OniKIox_gn#SkZOcQvDTF>^3otX2Lr
zU8okpZaEzp74=4f-I7tSStf=;mJ4|1gRq*f$};TAhl}vMS#r4>px3O@-|UV05f~T)
z{=9hV>Js(!^`-K<gwIu)vDz#*=HDl=TdF>vN0Nx%uKJ<=wcnERPL~0{IXSsM2%YkF
zH#_3oXeyRu9gvZg%~dH>WwV(4na1maMogRmJUZCh!y_PMSuOOqU`!RNeJR9&>2N(p
zSkB4mxjOozT6|AH29_bWI>&7YFx_`ewuhq_p{y&H@6r~3si0O%3O7PEI8u&^f#bB-
zA1Ea+{TOSGYG{2iQ57Lf;n%0a-r_YkkCUewSYNXU)MK@02zqPO`&byvJt`~)A;no=
zWhlzxX%q&<%Q2PH^y;s1LP`Fl6`*!?y^#U~6lOS&sg}akTz#jyg4|OL{hRD;we{Kt
z8d_;&Uq(%QbdoGVj*)fAT88*he_O)1u`a%MMt76e`!0?XLBkT|d;KmQu5(APGtvT|
zmK-O5K!<PdY^>8TMf|gNtN;{+ec~RXH{P_pGh0-2+j6`%3QuR=)`|s@5w#ZEkrCSZ
zu?Vdh&ITEIF8f#)rofF@Zokp-kDFe)9HJ{SmZZ=KF~-)}P7LoR`Z$ZxSiIXKtW?SN
zsY?5x?Bj+TV#+0RThjOOD`O0O9^l`l?pa_m^iB45LVpe&SU-H_<5T%P;>(SC#(#ZA
z<m2NjD=QOXNFa;piyYXRou5a0_pY@>r{R1Cm&^787$|fVBT;1HzvQWQml`;yr>A?3
zSwh3Y3bbk&goTB_D_c6xRhamf=#?9Oj%U<Mf4T65K|H;>>RMbRNRo+P$?)@odIP+9
z6N=9n{%~_5I292Q@x|xI{{B>9fkq`oUS8hagPECGk7a37H!<*-b(F=bo(i!ul94Z+
zzz|ET#_)2Rl9HlSs>=`J6R7WDVX{F%PFaq-ajQ+<PdqavI<%**_VLZ9r>FV4z$H8?
z>dx)i28`*<!%3No@H<^^i@0Dx0EpdJ<Jfy(ttpwR(p*{0BcK=5fNGu3RqE|ys2`4&
zzb<~-vb4vX8=WA;GuM!wQHOmgd>h|=xY8KAyD%d~DM7y&R^Ps|@DhnOx>1P5Z&6DX
zMwr>a8+<4;{lrt<Yg!jfp!&imdB0XjGymn?mNjg4PS88mXExnGOAo<Z6~F@{QRgBo
z9Qxj@c?zw^5&}<MrfT7x_xeK@NAaps3S-_g^%K(uUQr`$*gBn@7|~$M8`={?3)A5r
zZVOU$X=V0nT;}WL5&fc0mR8qdcxU_tcsPBt73;i;F+^W$y{X}wY-2NxcyrGwvS}=V
zaGrW0S~KvYvZK;`@nOCKKx!xeDd`mP_aqZH<n&5HKM4k<3Ic0sTfo;r`r@S1r)xAy
ziK>cnv-U=-L=E-!WRKG&r-E-%cCi)(N1X3714byG!d=)~)@R<oW*oKAzjhny^^WG{
zR$}1<ec%HR_FU@j4!GDEQEK)ai%IgnT?>4Cx>@1k;@TykZT$@>WUE)2$`&k5mDr>X
zrEr?{MU$&*YGN%6cXj!%uCAgjAmoacLfZ@_vlplqYx)H42PK7id3j~HKaF+F%p`0N
zrBoXAY($BGMQehR0DP`OYcJ^(G<-Nc=iaWI?rQl0+MdbsS>Bw{>)9X}EA%15NRJ!5
z#0WopMl1?l1Nc#wc*;3_h^&7ryB)45J`y{Pv^Nj6NCefSAM58E%{>H~V4)m6#uN=!
ziKIXiu1I<gT2WK$e(4xXewpfg?^#@jjw`i$M^=Az2C4b-(9x%}ax>}acm-TtL89Yw
zw);ubp^jHsed!)v{t&jQ<Ax2Z>nl5cbbXi2bunp628iPAEimF-^S3Vdh1qyN!)4%g
zt*pIsxPb-Xu%wetL;YTjpzikQnE&CRBg%<xffg}rm01U=c~%5=;@)lk!T0kI3FR&u
zBjmcRUzc(PGrl~b8yTVB9&vbjO8!YBNV~Wtpq_dc^#)js8Xk*+PSqJ;-I&Q%{;e&Y
z9Z9272=rX)p=_uRL!oqwwuFp-T|U4a_khSbS{jMpKk{}2b3|KbXQ%?J>cp=NxOzfF
zVX*UXZOh5eceuYa$70ZyI9YBg&}$L6zP|3=vL758Do`oJI%#<kpj9jRNKGB2QE6&m
zto*p*1m-5j=3C#@Rs8@YT;GNUUNJGT+lL2$)#X_Gc&Q;@p}>d3YW^KMdI$JwZ>E%3
z<&Sb(Z!f&Gv^0%sk;Kk$+TC%(375w;-HNx?<&wfl(+v><0>a~253%FXJSLbZLPJAs
zot^Wn7HS4lI3s@lrYb6OOmPv*;0g;3y}g|HA}JxEe>7j+_;kGhM&E~ofdcJ17661l
zrVBMzi%lNb#KgpD>*$wi6cid4S65rLE8f{saWo^m7{c$;%f)N`)Xc}RlVolz<y4X{
zx13K^hY6SDLQtBJG1E_xbhzpg8u`fCoa4nB_y5RptJ@`1iq~AMVLKpUXQG@B5LVeu
zH=*Mj!Iz1<ee?;*zrp6oiWaUR1EA*P{R;O|`Rmho!ZZ(-XAnhlm8GhqOeAWb5BmEs
z_7~dbjGc?MdU>I`u@@T6Q;1yMX7ulTnH7gN^{gdP$#B(ig_BtXqo`u~T%U$e_gKH>
z#lq7GKYL{5##p1*@sD@oI(OS&_d3^mGfCZeU*-&|*sg8rGR2D`B1_j`0Nq1K7ITsA
zcW>Umq_@>tPZ*}u?k`3JtL&F~w&4^wpRA{hXGc1GOu!$jd7!)`;zQ;Y$1ZOFqgeU=
z;+bbeFNETJWO(bh+KS)QP0meDSUH<7@5veby1UQOuAm`9+@I+w>lYX>YSv&(O-w$6
zc*MuY2Z>l<G1Kee?ylSwI5Wgz)Gcx#2GhOc;cQemA@4qBNLUy$6fBZ{>-RT!rAlD=
z2;2EJ#wiB^*JZC8on<ezS|1IOy>Z$$kF)iz{ACHJgXtmSE|0zOT+-5Hun5u7)7#nG
zw|94oWiWx^>hoE*3F3Q-4Q<oC-Hp{#S&M#?5n6i0@6*7YVPHhAB^+!lr$QfU$$I;S
zI|m={7D;21qNEO+B+SRz5=be6CX~H+xuS(Or&dMK8)12!Q7x;*{PiYFGq68xR!fh=
zAJC}QN0fPYevvv2fFY)3ZbOvMvZtqb`5S1NWybw9V#<)tEhMt2Gp-a<=5XEG)r5we
z?381^knXBl1eDm}M64k4gl~H11h6Oe1t9^=FZtce5u~(am=+5d{05nMkKfkpX%y7_
zq-9Pal}mIp3PN5e9?z-eUMiC!E8&778l+9#-#QPc|8-O8ZqcAjt~m-yNjT0{IZRu`
z5x8?%-V+Q*2W(c``}jMTLq3x4y-}?h=4x<N;4Gt;1vH)VHPgCzY{XwF1#q$4_O7h*
z%6zB5?Rkf=T<_Gi3n~0AI01tg`FosSg6>nI=Dg82)3}w4({)MD)3QW>Tj`0{fx~~6
z08Fq~QjNQ0eWMRhpPO6Ii5C;Zuy}SgY74c`7To3*!;oUIQju-Li3l5@kHU4PG~Xx}
zV*)}eBS@@SNo@k+_PAWhk(g=s+KjP$%f$SQ(L&G6l9TIC^f7&>#S3hLBt^<J3RD>C
z>gqV2_N+H=RrZ)|A^BnDn7+2WVi&Oe9T-_9$Fp9lH`!%?5q1s2Jw&X~r04&Et@`ZI
z_AdV{Xko$WSR@zHYdK@Y9zeS0W|83p7=tUW(^-Ei&QdnjyWVn-U;X715D$Rg7W9g@
zvxs|}q+Pnl!nPbPup{xm1$stGUv+YE;dD&!{m|G6uJNiSJZ_K+Q!RDAgJ5uxacr=R
z_TnF0CZ077dt2|@2QmTim*I6xn1@LrQ!yIKm*v(vvV}UkE66?s<_|G|!O3p$7i0iZ
z(>WK62z7M}Rn^_IXR0cc7i!c9Nri`v-9tHxaVUv?7;%2FGl9|PzfHY-O_mafs7R;j
z9*&2p=_ln5p66aW#XY~Z;_%QM%ByN7EH14r?D;6KOE(tF2^Sl&M5GJ}#Jam@dFvFK
z<c#pDcco&cmzbH^XMO{M9p2&n*AEY17v6(fw9y{K-?`W%40116@f=(YarxHS9BkH0
zX`0BI`)NOA<wCb^o2#J-7y63vU!tQi0?W&(!-tQROBa58kGFZHZ$f2}m@0Jsw^kJW
zogXd1Wvc%(`K`BW;vuN{9#sw#e|K@@i5BjSRH}h4K-F+QqGL-_HDmqj7v!;-s5VM^
zoF_ATB~^KyX0GAQeTMqY0^4}>262nNCDL!npAFLBP8uef>kjtcaCI3q=DEhvT<_(P
zrqx;;P>Z5C3m$r-#&6BslK580WTjn@Oo08d&W#?LAy3GXp8^q4F6y4vke4x~U*&)#
zVL;(CU?L*LFuNB;&_YGnp!rOXjFD#TX>sfDb>=N~y4ZES3r0N3TTB5bb`M4>2h498
zXM|eMkKsjwY<rTp`ewdTNj-kMT6{m7yUv~t)b~_<1Vqh7!AeO^LA%69Yh-REzCFd(
z+dol?VuR}iP=eG?S9>W}jmy<W9epTqd~Yw%y9%zn3uy<~M3Ar_aGI_DK5@F&$?xqm
z>^NbaZNB%wm`fUuK^;*jQItG<qB1-Y2b)m~ZGPH`lZ!bvtBDR<UYgh7_x%u?Jl=95
z{SG<{CoI|J&FXyt=I<$*w1+zH^)Cf2SUN~(38#m5?B)?s<p`-#(Wbv=rhMSO1Eweb
zgQ3c%VSC*+4lN(($%!<%JCO8d-$UTzOGOaAVOql0<cUhY{I1vX5IU#Ve8Cl@5-NUT
z^Bm6DY+$DtYfFKPuEhEXxqvyJvzC2y2xLAuPqwb0M(7Nb)Zbh{M&aXc=w>BXoxojj
ziJ!ftcB6_JEANr_>&g!ah*-V}1ki-C9$pF}1jLcqZwYcgb~C$4;DdlUsOZpfcNfyZ
zqP;{&{RA}yzkOPasSXc%7M<831t#X9!u9@xJ!V%#ItL3!mheP)rgGaF^wQ0aH^+}W
zqn%w5FuT;>80o8t%CXc(2#VhniC9@85yuDSI^`PGP7~L_wf=zxJ8f4$;~kb~s2x<w
z`AGPdrWGa0F%dd!wf+V%t2|XD$nFf5s+<y7S;K><ANlZS6Lw!}sr&gt>PX+f#Kw!!
zYsRD!#bf*WU9(uMSaoA7cZGn-SAH7D6WV5_*-L?@`ax8lBF4bXEHEMh)oFi{go){$
zhX=n{8ocjm`)^I&NPvbN-=3fDXYWMesrk6tfq-Et;f7qVT&i{>yl0aZ^(SgR#CX-3
zD(?$Xly6VSsw;k9Xf0MG4H>?LxAo5cKsiBTs3~AkYa5Kx0g*li-(YU17-nHUesUj4
ztFHUy<@de!soafm8rjS12N~Crl6$08C(MaymcXa#H33KGH-HDB*w2!B#5uRBy)%UE
z_-T-4y45t(UZj2RvPPjT;REYDZ?mHg`N>$m+(O4<$|mdSp7Hl`hIdEmSyRr<BK}sZ
zW}sl$b0d@#yNV?YhFf?#rW&fvqu#;b-A04H7Tx7!7D3w)-RunnBCVFD)3MsO32^sN
zpqh6a;lD^_02v4>ZH1BNAc=RHoPGH>@hxlV*-gz#->{U~WGH)A_~y~$vKhfj3)XWz
zt<DeT83WIDVi%!ekTqW$$tfwOu-4`tnGv1O?Cpy`*2Ga%;F0?hO(Kgj(}l@X*Sd`p
z+p=W3Hu(mk>Ug}R^C%~uf_ZOZCbt2>LcQbm&#G!KKnOG4J=!kmo+=wKH!}xr!Z{6G
zq?F5)-($4@wv~JUE?--N^%9T0gTvKvzIX)D<LySoue{O^jhn_SxK9EpLz^U4T-6ph
za+pu$H@^$c?lq{H33TrTAWl%9#j>@|{VNWZzU0{&pIBg|WM`Gov)PS$ZfRqBqQB8_
za|B#%a7xnSLadrWxJ6y3HK&MY^)<V#w9`WMREZD=y?ZNDwCMJX7B)0l&)6a2?bO$s
z>krpfnvfHvZMEz<TBYnxeD7<z^{vf6gUmMO+#pC&q6Q4DKD>BAtSo)cu<)zIx?(*s
z>F}k3xc9oYN|OA}k0`rv)p<o}V|enp4JbNbD(2~}<`Rb^YHfxC!IRjKwvM}xMb^ow
z21u@kSj-XrHbG`*FT_U{GEKkaRxH?K*5~7Jmh0kp0ZA(Gl1H%e?kwUxuPXgNJL!L-
z>bWWH=$Xl|e(J`WTdcKKg48?d28nQkVW(R?cIvZ&0>W}zwl^M>ob}g41zkXJE+s#E
zl*T{i&0F`>N3HL{_+t#(HRj;j0latm9W`d*Bz29ppJuV?^NYK5-;$7rIYxC5><(2c
z5!Q4-R2ISFGn|SAM~-L&$^22yKOo?HI*9X#uU%<6^mV!j7a5;}?zFL>fcolSCg1z%
zJ~0KGU~cW?BqAe&m;#`P5a+O3{2LXFNo#+#uOb>q1uufDl&@gvj+827?TqKKQ@tj<
zX&yE$9TuiPk~-5TQ!fsJes;XRkFi#A4C$WCCUuSjkD->f5*-WM6S1c)w(!y&`2m7n
zfg?0EcVLiLcszX>|5<JG$1;A=tMht|N45J<{Q>I68tcX55AgD{PWg{2s)(Y2Jy(vU
zYr;UAEA<+@C$p1D5V|P4&5Ho29(?}<6N=!crIBn!Nea~+AY30^dmkEa9SC&EXxU37
z%kOL3KYxY{I}G=`i-PcEa)h-`7s$oQ-KJ43=x7~Vkv2i52oRO8Fs1FL&ADkmqI<+^
zY18uR!B}`a2kHPWH(Rp5iETL+^I=8wW^Y%Np^R`XfixOf3-!<}^n~?w1%z<&_9Dy3
zY~ecu_A+if@A@FHhUg71pDv243s#&(Bz^2$iW3{Kh+gzk%!KmPeq&PNE4Urusa;$9
zZYSKD@^(}Alr2H*`1OF!4ua+qSQcD>K3T}!A{xLFe)}JK8;Nd1Zwy2sTwiVT<KmT$
zX=JmU4lOHVs;I221IcG>2JH{5tTCsjwz|NR6U5=s5x3Jm1~)hNkAQ$}%0RF;FrWH^
z`PTo9Dad_h$sF*z;#sZS|165daUQX5OR6U8m7;z`YytJD5Z-2k2hQi5&L0z*=76}}
z!>;rz7)K(w;;+!Ij$KXlmaR8wWU8Z?MTg7yA!Pe@i+0BFv9UNYNeinX47p__GMcoB
zKX)^>W^r&;2X&GL6aP6;;hT~JA}z(>O1K%h9`)F@B>Z7Wj3MJ-F<m0oo61tM0Y70q
zh8(-XT4^LnikP##zZnQ$$5j4s0z8j;1&%@rSD`TuYcpC*EEwyURfm&--WekdcD#()
z_Q#k>mmSsxi`{)6Z`nRcmC=Ssq%QS-ms~6RfObAvb<FeK1vV6@5l+YiGos^(APP&Y
zVzi5Q4%?ibL4`^gwdVYZCGIOe8)Ka@1Uqs62&WcrGI}Js7h6RU`a3|qEv01B*rd!#
zWsGd?%{Rh+W;-9{;J^+9CNAi33L#yR9SP(g`5m{kAk*F5O%Zx)3d8=WJ0Z}m7@3~j
zTXH4nQ$PT#Hpx);(7RwQuluJD-+fIYYUq4qjM6_A82u!R@0v}3Yu-2hnfbL^g78Ep
zx*(+Cu+*=ykq?g(u$GaRe<z#Fwgz(4IJlV4aJf>szhqN52n0M6>zxklpC8>oPO)!j
zC|53x2PB-K%G@qYm%SS{8EZ#;HH-5=xytiyE8%i~>cA~9yRvvE!v)$TP_vkVmKIS-
zNhvf8S-1MrD0rmlO}&57qd6b<T+)Y_HN`rF@T~(wYd|MFb?{?!cIi8OZ|FFwpr|P3
zrTehln@IECE%(8q=SY#P-ck~4Q&ZK=7oGOX_m1Bv1!W%+V>{LNf~4YCEs?`16h&xw
zwEk?}*;Z{?+B6PC21%D!RnXV0Xn*GLTHuPw?)ODTb}HOK7<XRbj>xS;28e#fh<(Q*
zW99e%JOK#Ai}69K;|P0c*|H`WN(}nbn%|Mx9HlMZ9v%o*5h)a^7N|cQD09%Kd`@Rk
zt1oV-K~}Dc6r;(fA)!GRnGi1S?wm|r_G?)yZZ^`r3;8(aQo{V4y&bXQpeiQ6f}%Z?
zW32rPQe<M)ngha~jO;!lQ99v9oUiA1hzrPPdjlyKxVYxj<LD~IIe|?<%HeCv9U#_x
zn<UeXza+SoJ%-4{O#j?24mJ=_^9@<fOabiA_2ksTQam@v-9rO`U0Nk^w?r~>p(brF
z_qKb7E92<X-QzpGQ!;pI$(i4QiJTCE?}cV2VsZ~5`%WplIQqg;xoG|(KK}mx!x;%=
zK*Nh++h+4<GePLI(d5mr`Ym&?TvGt+?tuT7CIXiiiVWwgc(HYci-{+KrG8)GgwvAk
zp2w-S`ksgAzTsi9+-QbL4NWXSLBaLT#KgqF(2%rQx^c4}Q|CS0_rXCK7FO0bW##2<
zgEE1_(7DFfhjXtCHyB_1C-NY6BL`q#kgLomD13c=A#S%4^nzt0K^`lfPQ$mfl#zsl
zq)4YBUcl>K-_Wpqy(`3gA`cnDa=M6`6+u_{om7MjMfQb9#zc`0BM1FUJUZMoy358#
zJ5D*QuVz*ULaPT$u4b*!U|-@UH}*ES*x-UwqvDE(cjmb_D-E!JF@;_n1n5RzAK*Pb
z&qUJ;R~S0ijL1k-OdQEn_9S~9qM~n2I&#vE(qFqR&!C0csv4uiLuBnHN8ce}czb;h
zYBSyQKHh9Zs`(Wk^0eHz1(8J|lEn<Ty-XB<h$lmKz|`(OV<<;Q>brCS%={d0{VNq&
ze?uG?6Ouo@;D%cVB$W1_KQVvPyW&rYA0cJm^HGMo6O@q|Dp<+k*OSo+&vlwh$GE<d
z^D@7kPx+?99hmK*cCk87|2JMCGpsI`9a*cX59Fql3M_QO?w@WR4a_k-GyRT(L?h8E
zm*YD>Ev*V)(ib3!G9MD^+_{f=Go;V=)U<4$o!oV75_;SN1a4{<**dim<~P@wXpGH#
zmLxkUQiwj9lQ@hrMsT;x`*sx#zmM3aF9=l1AUIinkupRl6GRj@RT8Bk{8u<~+g1^l
zK?!M#@!|eHezsXEuBR^vd_9)`g%|=767n3>cOL(~A;W;qDynp`-vDD8hs$1ovo!FX
z%L_nJ(08W3z_WU&ncm-wEY;|w?#i90r8!}=3jK-zE{`l1TL^^}MUb|!RJpi3<a13$
zi(@j+%3X3`7@=f()s`WL^Cha;*f+IGanZMJD0yl9@YRN6Fo2BILmBX%&wi^f>0{Gt
z^Ecx1*imPMYso0PGMzEINJz_eb`K)}*b!2$YQ;BFvUz!7CB&gQ2XJJP!4ccD&DDzy
zn8I?-6ea1)lgM9Xi|^^vU4Cko*EnU4lqJ0ud6!@9Kb!C+r4$Lk27#;Ib$#PNZ~f=e
z3(M|K?*PFs7O+>=GW5mH3Gul~anAIfKrznC=X)gh9FtPmIyOgG+RskCqY4PW{FRb?
zE~hQsDR<z2amXfpd2#a$A4K%hF=EM4H?o)b*+7fd_}@yKM7FVh3`8aNUaW(y{msYS
ze$(evTvT+PU}R*3jF$2No)Ifa8&m@X8w>Bi{q=*TwMj%}kyd%CAtSUZ5M53Ke^qU@
zL%=b~vv;vQSQA-FPyzUAY%CL(9QIqWQlY@+z+_|fP;Tc&fRa-DkB;THZ=nU|b9l&q
zkdtJ@{QIka%h$y(d+|RL;AX~#g&k**Nn7&nvsn}aNF(yk<&`Y0<&_+yyaH*Bd|OJ`
z0<06n@mox*FHU#mi2u9(U=U(ZP*B_rfKvAC+#KeD06?~JX65^(<l&>%!ukRAKP-}W
zNYF2dNL_dD5>Iz!u(SIM(DRk=S47+RYiq&C>|zrcO%=B4WPkMwlv-m;wWSM_D@@iL
zaum1T6WIr!Mk;;mUJ)-C>}u-^dn-^~fIW)kAZ__%L<>{%g^1UTp8@B$&!%>f72(^a
zLUI7ev8q@bR`5Rj?+Qhd3{RnLKs9!*%m4~PtJZp}167c&=4m*M7qRF;LUnorcfxY=
zx4|kE$f^ENivu&O`I@`f88W+_=zhmkht^9obXKf3DsRu9u9uDRBq$8_+7b`{KbHht
zQSoW_U)jt}o_BV|OwW*|e|awkQfvXZ5EZNwbLFOQc&8tQH0?dJtia9xb@|zA`N8<@
z)*|gbcD7F)zuTc_QLv46Dz8SS@`hsPiy$noI1T$JQ_=+dxV>@g|1Q3i#SeQ0pU>6y
z`8-qD4;~qrzQXF$C+Me#n~@i{-O;SO0ZyCcK|Slaid<nM3=j+G>MR3OPdQP(S$p?k
z-=yVRryL0e+~mup2s>b?;9;W^@;7hX$Z!@Izk`H%HduS)7?^&s$P~aDd1cm8-V)31
z%>TccK>fYf1Kz7$=+<Wl3|Te9mFg$G1PVsp1SD>nCBM+Y@3RCn4;-z_W7VD8V2))e
z3S<O5BA)<F3&`8Kc5zI75tT$`<Ny`lc)3kH+B>rz`+#PpG$<`#g1=c&zC{wwe64dL
zro((|?f!X)2lL*(snuCA?TA$}la8;vG^)%|qAxdK#Zp~;es?+Izb-`H0!hHP7R*GB
zyQ4m!&JdlFA`J=<dkeLUUiX(-BK~iCjRBV9#kvkq^3<sLrt$3ls<VRv#+@KYAMgWr
zN3|;)j)~yvw$n;6aIRN7pHx^biKu%J!ya(e+Z=hxcUT@RW5&cGn1sJ_0An`qJ?%@C
z1)YT4K<=d5Kh|JZj|l86`shiY`8#I4!xEb>nTRm|V9vFGnfncWz5O|`DmF_5#n11E
z=vW;qWBuIiLC(Ps7;>(JAB8JMLCF%XZUp;(`EVVNpF?Z#<Q036t-<H@&h94rR(}_1
z$&jtg-N&lD+=b%Ek?)hKh{+K}F!TpnRE{2q(ITBvk(Aka-i?`C<URf}N|gH8R3XO6
zY=_C_O^1)Ky>&orEdb11pa@%v**JPHlfai!`2l2cea4{4Bv+`yd8<FJ%4Q`UkKJPR
z;6S>;`RMIOmlRH$VNfH86g-2e1ADn5y_R&at(zz_5ETui22a}9zKqL6q~st?&5RkJ
z=LPqY*EXyUDw%uBK<XM41~mT@#|<KN`HsmO!KwIY4}XB%g}A^H5z#^z-wY(MlejA^
zgU~ISZuXS-*1P|BDAqw(rAhoS3gY5wm?)nqGfY_yM>09S)m5~2k%zghcVFvU5-$of
z(1*FLfmO$^1bZQZ#pa^j>ygNUBp7};x!8UsPhy0(^x{H=K6=)|QVgX{XS(Gs?=FcR
z!>T|B6wtZ(mH4pD6<(@)sw0I%2hO6U`F}mEOl*H78Q9lcpRA1dK*62f+%Q(8fjz6o
z_2C+5%mR6E4A)|>hwID3xtMf*4=jntf7z><Aa!Z)X0CSJYCwtXgH#A!94AYT)A{7(
zi(z=8zATjkL;CVav#fOMFtq8NFCfkwG)kG&pId<(c39^^o7#dUOI71`*U!H9yrveJ
z1Q*j@E&U-Nr~MEg!t;EJbZ{_)sOt}iNx5<b=aO(wee&5f)hF&&AHNwWszeT`^LXF{
zp&na;c7^>>BY2Pz5_07kY%Xz&RzohTgKO9*bE4`A)bmvPb^#J!V|Poy$Ol=3m=cjj
zk&6$0(FWchS?W^1^t{l;ZT?3gzw6IY97ydqW|^<Ck_I~kFmY=)xyNOTN7!HPDwJr~
ziAYJkZEI@-x0rCDM#`(w)b@_&3Qw+~7EV}^!SnrFkBywZAHmW20ZVpFq)3YPd$SHy
zM=QA6jvm6UD3gE%J1v#~fr>yQ+TUdJ86Ry}z5DOE?;mO(r3RPqLC*%iG2R>c72k(B
zrbj)lcq(#|kmy3+tDCh#+Q>BV6OxIqi+^qvwewM|w%yg=DVO_tZ&fc4f4_G=t-%jY
zrJH`a54%1{fCM?x?x7{$n2-d=L9sx|Exh2E%<meYkOvT&KXIqnIn&>{BH?gHxWBPl
zQ;l~Fm!rf4!tIq$XLJ&AQ0i!4Lleb8*HQI?@P9qjS`v#Z@^bs^?5uo~yVD=F(q}pj
z%GWJL9^Ge*gxdn$V^371#e=ym-X8Gv>jquey#q&F^oO&zAioAj+`9BlNXp_0hwJR_
z82J7HV|}|?`jbDLUoCl+?RYUdFZ%9wfW;L*sE~)a2;SBA=GoW12fO{&lMAsjvkt>V
zED~}<EpAB@;#8U3sB=4RTY=B@TY?O5by3WI-A@vMM50niJF7K<IsV?d$q|cBr8oQr
z;&kHICF|MDg8GJkOx7DPxZ*EX8w5v~082;ClE5@S5|Yu*H_6<-5ty?nLIG&U#?q2=
zqGyrW?OCb3=-XD=QbS^RRy`o&8HDIuVLYw>JXcOY5eS+d_6VjbqJ|Yp@C+St6dK?*
zNQ8aLgQ!JcZfJ{nZfKXFNdB>SA9jzDKDxd}M;2Xl-&W%dsI0<NnRG<}dUouaIm(3<
z@L*S)4H{L?iWe?%#i-$v5qWBqG{5fUc(7cK-4~wN9nXdvqBnjZe<yfn@g6y)rLRAu
z+ktf*O?!l@q6w&!yTla;ky1H`#}NZT8wg^3bKkR7&A4-(&*6!B@m7p{eDA&n?;W->
z+QbNDPq)(sQU#!H4~%@f_2Mz7L|-*l%(4oTmD!v0z&Sxm{SG2r)df(ZF}mFpg#UVn
z1@a}WOQ*DoZa+!ctgkqTk<xq;T#$=v$o{)OK4Q|gSHiOK+~T&WdvGPf;bNW~6sUza
zrw~7OO0xFw*14`Lb!3PQHM|nJ3aS!R6UUzdBmqQrtILDmpy>j~j&uL5MYU%3jh9~@
zVZq8@{+ae+yW?GFYn>JuMNcHK`3IArQnB-gFiq#K%3s?qj$O+kpOOl{&pUU0LCX))
z<gld(YEV$?mjOZNre(r1nQQg~<&RYU-aGV$peAfdP{u0yc4%jR_RCsd>oLV?oTvN&
z_y|Hfuu=Lt`3v!V!WW0*T@-8afauhBe&s>9|CFVf*zZ_6b{FT&t3>}}igIQDdz5de
z@78pNP){tpAvQN9(8ZFH_+PZ@V1$)GWo_IyJII@`FGs-s*d#hFF5ExY6u~G-TdsQE
z2ek10Ub!ZgVwrsA8w13>GfYC0>d3=19kUcd7;@_g>lWcb!`*4|Op$W!|F14@Zv5)d
zz`(}oTF2pHT?`^N!#hI4FioAsP?TuUYaOR5@d|<D^Qi1AB!VL-D0$I6#MQnXrbhN5
z3%(SP^2RJgWp#dDY47St0Wl#HJ3HsCb2?_VrYrzTx1cq&H!RfF)_I2O&0Phno1-o!
z%@eQ3Y+z8|MMJNF=G%u0(w3Xo;C+o>rrhY&u%ACyLEj!F4^KT?{Xgp`wyLHS<O2u&
z+TdETHYy^`X>wQ-D{_j{FO`JnN%L8$(j#NDyo-TD18uJIl*!H8eBD|H$6UKYz#565
zY1LoSb*bJv=_w__{&*IJ=+g@L|7Kkvw4RFt6^8Dq%5#U{(>n*c9TNJ^O^KMaJ*h0;
zPVH|;Y0MnV>@v1IRAqwq!9&!u`Tv!!4~-u|kyKVvGE3MG*4UUYUAtptg%|(;yOcuk
zh|_+>LgjB->={=rvH%$^nTq@P81&*$n+!@&UCxr+>Szfj!F;tZXd%F1F$j|D>t~m4
zHl3(Q+T5bfrKAhXY=lwA7yg5GY#I}dGteyCVr%=ra*veKr{yJvZ)<~0%g-LPvg6$r
z#Sic|<AbeGs;n>!U+>y-vF&kjVn_HC;x1SMKzg=)a&R4}>we<9own>=S?~_bG<F?Q
zxCv*-6G^|XTj@So*F}IcbBQhy7qw@YJy7_wrp?moutNFYgAit7^P_adrq_(kSS$ve
zB$RTgtwTdYqbdK4P}r|o0FmNm>rya9n88nuVRQ<)kv>;5KO%>A?Gsv2{Xi}&9L_{7
zK%zG^Cu8+8!#*D_m20?v$_W}`<=xv?P~6>w2p4?OmAC$U;KH0<7QM5<8^8`Fc|137
zPHk$_fA~gqW9`oh+F*a3XhUfkxz0do^7>5xVsT<St<7BL96e<YL#QBE5RGtAyS94x
z><J0T;Wba=f*soZS5lCqEpmEPvH{fSAl}e7;4$$W^=;ZiYC4*%Z0qTNA!s})&=OZ0
zCt*S7WF&5Ew8Y_%_}^{24kPLA+AP&O-3{P#T4VI-fbXxx<*63j8rTa`*Bw~Eh8(p1
z?3ce2e6{b|9SyV*dyfu<T~2<?^BN0Ed7r>GKYdufl)G7yL_nbm_-n9dHtV!)GJFpr
zfIKwNrN`014K+#S`5X0lLV#mX4qw9FkWML}CMLSU=eleW)!m&hZ4t(n)9-m}Ey8`(
za_FkeLFS<m9h7ap)OrQYao3OD4?T}}`Np|89c(Jnq)N5-b%ksnt#E$ztlvFvli0K;
zQ+FXo!<cu8%n!%~6eS~SAH0HOu2)dg|3xojWoH#z{1LBEsd?ooJ|97Z%Y%IJOHZS~
zJuL;JUZ20^Ooq!(-%L*M;ggY)o+lhk78vawzwXX8{|c)qPr0W`NvUIoB6@OXX+t`{
zthmfl?BGK>P+<i_!nC~baG~l4XbtiSLNcBC9uLTo+la$fx%k-p@zVx_5J*59GXd2^
z)923R&YtC{YZD8nA+c<pl=j7B`GtVrKSSu!_=_(6=#7?%`7pu8auy9LFukNjyJ(gm
zkwMmV!R&UBMPa+W($H!C1_vq0jfou6zADjH7WvT~Uu5DpXes2dL8hzb`j<|qw15hG
z49|-h3z{3k*33J>vb$S^=dgN7PpP+2W6T6EoUuuk)9;x)p$WUckbY#@cMGJ0=V5jA
zof#OQ1?GBtdq<aEd+T;YGIs8Ca68BJv&a;ecs-2@=)U{<W$`KHQ1ihkYDk#|a{$1h
zePXIYo2(kM=0!fr9V@j^XR#~;oDbroL4RVM=P_}C^C=GT9*#x7GK!uW6fgR#?XLH9
zbsVt@$=?3V@G_KrTw=<`rgi%(^_WI9?cV8Us+`U7=7;>M!`b{vqp8X7?;Q$MdEd29
z7&uaOS~5keW%=Gkfr16Cmd7jlic%;TWTbnomm;_hJHmYQVqsfAN?c*zu)=7qGjdJi
zgnvVaAcfZ;6Ph>*$yXtpyxMP4<h)wX&F^G3M#L)CTibRB?^&mcG+05`a!^>9*K!uT
zJBzDtEvO3K4Vaso2OR&K%+bYp;CvyK{XLzPYD(tvA`R=Q`F@k7rYKP~ouziv)PsYL
zl$Vu|Mngm#8(8x7>Os~Xl#q0nrw}`{es-v-I#qBh*!)p*D6<MxpnDK6*utuB%oq?b
zw6a-AJ;2Pt;X_`#8Un1zuu#3k?5RLSB{7}AhQj-Qf&mfF#zwhde1Re?Mc%edVoREZ
z6Q1ZNSfoa1+G7#z;fXNTbNv;l+g-H7kdEQ8xIV!{psmR>&W!l)ztA7i{-(%6FBVR6
zdyky#TJNxzE$}#BKu_pJv-Zv0N1LK!yfblrD-{QQ_e*b&SK@L7b$=rt?C?{%p7W90
za>MB$Rh*Q)1L8|KNuxSg00?;;zdoNgdp#g6HNBp{k!hG<nTFnRDx0V&0p@lt8Wa$K
zIBCkaR`y?e?9H<(L!8NUGK@*VEzp`C7UFc!Aq0{x@TlC>Q}b5qzdgi{27Jm>FDRX}
z2dA=}4-iXKcDx7G-G6{)HIO+-bi{OrmJC{?et&9$b6E%BCkozbQPI)u=0HYHXPG34
zQ+Ng|XkN>EvH!`?B(9$m+o9)@1$-%s?M<`AY(@!OOUN>dj4cZv$^yBvK^-?aDWreA
z(doF-_SV=&3htuYyOD({&)J9Ng<Gy!%!crSsP{&YPjmCxh~4(wdUTrYXMkviLZ`1j
z4mdRRYI!c!%4Mv|@H|IACE{bcr1)3A2bDIB?fB{731nNf0>^-cFK^4>KG5B&XEClc
zVKUIORm|A%qRvFTZTYPy=w5HX2MuB}8p5oR4~mFnzb!>o#b|0v^f=T#8Kc36%L+I#
zk>!`Rzf=z=QptXxLzbgx?^1TnYy|~K5F!poTSYk@vKPsOEj}@#D>=Z1-=}|0DH*T6
zkpLYe9cgs%fK-5z<hv-JxLkgE-YbK7#=ykK4I<DbDtJ=%FGaYDsS<GY7L?K6*2Ql`
zaIUn<gp6m|xT+ZoFT_(_oZotmp($I)T4*DP%;}$s7>|vOK|@0`8FeF3$fZj7zr{>s
zHA4Yejke+8l1@hC<KyE16v<a0lU|o|`_;S+7}9R}Nd7$q7XoY>vG47z<!Y*s)mWcu
zHmp&EoK*w7pLVJD#=Pqf>ICIiXYVFYXIfgYBektRff}s&SQaN}uSVNA0f9hZZXA91
z7GGHvgSWD@#^%0866kB605<tRvvVW{ahPI3Yug+lSk5=?nNffI3rCA{Wf~ywsT%J)
z@qAObfB7|aX~d(&#m*Mf@Mp2|t4_)Kmg&J!gLo=N#)vR{&WdkyEW_#i(sFVr0s;cB
zoX#64Sov(RFwz;&&<toax-xs*9Fz0#;DJU`_v^zxvkb3*?2QfH4UoMP2gT^~0ajOc
zchUqwBB*+k0W2Pu<J1?R63q3WSDj9y;ztbSm*|89@r~{<T(+<8mKt0b%Sr?uK-Oqr
zts`*muN~Q|I=bBC!Ntovktd7ae7lYa5=Z$H%wCu{d~I6ghFy?Xyr5iqcUU>XrmII^
z68W0z7=1t@+PrNMf^#5IvX7=KTi`{sSk<Ro`{o?!ZUKlgR4F*WJP`j9!0&wS>(#st
z(toyBryv{g;1+23jwJ~R+%ty+PDT9muV-tQ-q3c1b|LDVm>lpELS2%|>HDtW^W50t
z#EBn41Z*QLgR-7x&9}ZFbQtMMoo3IZfdocYvta@>G_<!LErt7Y!*E#x?k;w)T?N?K
zVn=~ddDG!k@My}fvt<U)mS47y&(F_$!SXV?LiX0*_xfnT;r4V5((7cUB?Dv<Uzr|o
zp3yhgq^$wVX)a+bTfEYCtsU|RB+P~?Oa|0zzTtvPZMgW??h7Gj7KRkt>s9+gT+5-p
z{FY%7I-2?PxlQ1uqjwhCUL`V%4!9Kwe=A4P_C<k-n?Ni;%1H`)<%dIGrab1ijE_%I
zW)>DDDy}ptL~0B~W?Nr2?v4kDWC$jG%`;pK$EqS;N8!em(5v@HhWk`BC74=&pav-t
zUq{@(vYX(quGGVb9>nSp&Am>qwORpW9%!n3;+KX)Az1AULZ2+shyg%_8{;huII?+o
z_^rLY9f#lDG0UZ!QTuNsxJXK%H6D~bKdY)@fk(u~V#G7(1dC*4f^&VK1siJgw*uLu
zg_C#t&GXa4@2UOAo)nwrq?Ma8gAOjwTPAC3>r?@+<i$npSSfOBM%~VZg_O+95AHX|
zR*_z3*Vm>Xhyyx1J0rm{iWXFmZ0_ps4?f%IL4W_=pjY&Hoaq*R`NeemAGA?1Hrn|C
zwiM<rXdiSCrrjU=n@FEXwlxaeqZXusBjV@TD-zHeQnEpgu$Rh^?RXlej5XkfKP#k=
ztxr=OFVl@hxmeNev6bi5+m+bB`PS+JbonrI0L-~hyncJ!w!g&4=m$Im8&wXZlEv&W
z2tPzb=ybJ5r2Q*QzNhwo3Gv~C=*=jb?#nsYvwVw@XJVoy2m0H)+uPr*nQt!EIY5AW
zC`afHl9iPeoO^2n-^tLV<mXQVM^x4~H$Q>o5!RPZOH0cEkemhs%&0e#q(~QNO8z2n
z_wnUQJQp*Vru-c~{(w!(Q#^4>LW0_}<<<Ta85I>gHa2#R)j~H&G!`i42J1es*l+gc
zr&m^2Zxz=r8;+#&SA8Ax&6Q5Tsy{9q2E{0`Fg)CP>DLHwT>~98cpL|vSvs_<pU|op
z^R>k5Xv{_ET3a;oMN%dafp0x);74B}T@$6V9c|1c$?<g^eo&yukD(QXfz+5nlA?@R
zjD8>eEiX}?YQi)eUafmWXITZyg!*ZV0B9T6asL&M%;xfAe|l9PQp3$w5N`Nm7GHF4
zNwuK;w-vl@u)9ERnF&5x=$Rn~d(vppjSsqShVwKD$|yhOh-w2Uph42k7K{D)U$%dz
zYZ_!e?(ZdzcyugA8=CFU{ZUxNVxLv^R3JHl=Htg7;Jk%p)o_+*pnQR{=hc)NOmxE+
z_nkBs92R4EZ~&Pxa+SWX%@6L4E9lP$=~`ktN5{>J5y72@lNPg^<0a4MyB+r<X**Ds
zgP3h_DU|?Du|#6OzV4MDisOHz0ds>X9r=BABX2c=kK5y!<~^j+Y>f|n#*uxiLVQRq
zbmT7hvu!p`2KMY<)lmLuNcOGlPmi|2<f#N^9Eix`|4n}*Dza*f2C|6OPt9~3IpEX;
z1&ReVxeNcGfiv=LGnS)E0$LN%_Noy%xo6L9E0E!RD#i`g2qtq|OW!ZY^ff?|_V9tt
zHB>d;bA9209n`IoX6;q>#eZ@5p|Og<s_2cxVE=SOskj%{-2DDh0!O{eNqA5YLPsEK
zKFG&@1KmWtF2|XQ1;N@oV-j<A8{GyL7Mi;o3Ki^hbTWmmA|fIXyP(}6%4)s_utIDF
zH~Q6bf;e+Hm>I#zOam*cv=&rVRaFR(=~OB7a#i^WUHba<vC%xpS_&Wm)4AC?t~Kvs
zON;pY7g;D{jWy)k_V%7AI4&%l8!yG;Xy=dHysFav)vrIEDId1VIY7C<+h!x)#>nzS
zdy~eZI+-}nQ$pFbEOt-_P{-e2z-%EDN2ea+twM9_*WM02>1xti>na(4>L@L%s`2Ny
zv!y7weMYQg#qy1|vn&17VC`3>r}z5IzyQxuc8uu{eR}vHT<SfJ@zy#wXI<awNk;s{
zmyMKzh9($<e-*%sH&_oPK{RlFFw))C#a}HaBZCA1rnE(1aZ!Xgf<oTIo|Iko<41{}
z@wvL)6SxQp$&iSM84tZYxiqO_&FVx>8zNARH#RojZ9**#Hj7>z>hE6%>*^;lF*ID<
zJ35Y6@*jRA4K#jGQYs~fyI3NDBU4lUQ=LOHiLRIxxwMO*38jmp6qFFF-t9tkG_~#S
z^<lpSOh^-=OPzAZ5Vi^@LBCd#s1_IK*tcc(qn>IL3$ggJw3wL>r-5t!c)w)U#Zg`E
zVA%UZFK2`OfG73Km?!T`XUDG;kiEI{+_wiSC!3zy|JH@1X)IF!g17V_!DXJ<DMdRj
zm=lkK>65zV#pR`OS1_iriOG8cf<fTd=MZovZfn7&rK_W3qo6SvoQi`um?>psQqVMr
z(5kJ%cRZ(Zd!QQAO58Nw^9YWan5d4Zo=f=o6FyHS$+xbKTdU?92?GNI@Qb}#Gfc$b
zdtafRAxKw&W6H0QIA3jPWN!Yt_NQlOZ9P5tLJh>;w?1G{GFvZl#R5_$@0(tQ^!AV)
zkUCndcgB5XVi?F>DfRq$H9nWd7W&dNgPm6AmJsruYNbjD1y-Gu$Rhvq@-=TFZd-4^
zqKF25Hw%smZm990`W63Pfqz6I_}xYKcfdGmM^mokd=+~p-`^Cwy}>skj0Ds*d!S3E
z=fSVVXOJ(^7Er+(O4S`)c^gy*8k(A$!ewAYDW64(Vd<+Q6#loAYi#Gmaxe+@srA@H
zeGNBG`=Yt3tS6z$7MbE9>pNjoR@>X#w|95Uj=S={etsH%ZArj6cbN=<CLvysbAoU_
zUNk?&*`F+cYFz<dbc|Dg=R3IX;P^B4oLCIt#SK<U2=D`buo`A-*V*IX<CDcm>RF?q
zqhBqhC1@6WbgitcG@q*=2WO){vawao)+8k*4W{#_fU~X+-~h@v{J+_W^bjKtHEBd*
zmG8tB(%HHrA3J&y^cOK_agZs|NT_g8_`VbW%Cup6^i)dDDXOLP)Whc+tiuEBqyIi|
zA}$T@0UTL_DndQ>irmmTu8E?wlwgkM0zB2SWI@k^J;=?1;sS)nQjHOvp9}vRuKV3v
zjO>}7m|M^N44jn2_W6fVN}T2<l!lGfm%}9&PuZK7*0Y%=P7hQOXlBx%^VwnwtX6kP
z?DOOCG*Z>`idBwTDutTG01Z9;RF;%^L8(f%p^Y#a>(5UChfXsm5$GgC%{SfT%OgEx
zsFXLJ!w-!oKT{rtcNzuZ&<$SbZ&_glkOHy8+wxMl>i#lcU4%LAu`gB;_`ef@C+UO|
zFIGHT9^HX(^*Ue}ubNlvQMW(D5YM(5&vB-AueCDT!x$PbP@)t{0`}Z%!GSNRGQ+O7
z4BCI?|0EI-^`BgVG7rr0$#|}`^<iRSqUS?QbhLVz{u>B>_e-iXH^h3zT8FY~`rC`a
zMz;$of)@j23YcZE>{ObK5DB@BJ0H#aYwDa9ShuLJ%2n&all*WU#YKh<KYGX*hs??i
z96z~U=w@oHAD><YzDAEH+;rHQIPa$-mNCOxo5!8$asI$Aq&}IniXA8k4|vpBtqWgl
zeqv_4h#uztKd!zqpsJ?r8bmr(Iz^?s8x)b0?(Qz>W`mSUsdS@AcXvy7hjdD(bbNF4
ze!d^C{}hC?XU|-*uC><ObuY*w?hDtQ7o9}A#vabkdcH@CxY0iJxXV|L+|)evB24jY
z@}t%+tT51c_b4kh;JnRXnQA)8kL>K!d&8wk&n?x=7F{0&tWqG3XtX@^y54o8RD$>&
zJ$Hm5WJ{^6e(YPA#H?f179lik!O7_T2XCC-$#<D+pg(~_v(gF=^a_2Az4HdFI7CD}
zU<?TV8%VcPje%G<p352^4)iHzplbQ?<Hy?mP=Ej0?o4@qB3J0=&z}XmxGhH?`hmj_
zZwrs;ps%kFgm@32m%t~2*W=j4t}hUjqFU8<;cEF*yu9%(EicRWWxyBV;&~mObwyGP
z>4q^03#ax((SUM3`LRd@_VXu2g^aVHywYg)6@mlZz5x)@<xE&{63hBCtk2&*n67oA
zc|!fR#911)DTLyp31!4RKPgSSbYDKtZX`xh_lJj?a1tM6D?b(1-YVa*M4upyNhnRd
zzsz?bGwqoEZH;*UM)>?J*?}6Jc3*JwB#a{=gh{1>=5B+lVn&l{`-}F-RN9pAMe3s`
z%}W<e_YQP7535(~fP(g@U`ghsf()_9ES|u%S3OQJY+nY}Ko;XxYVt{<zor&UUjKYk
zcMCPd9(X!a4fpBc)_MPrOi(^-Tz2ZYK_S>H;dN&G5I>SaA}RgqoO!E0Fer!((D+5q
zksm(9)!^~E-m3;Ukm*8$2L>@`&hTor-F)KaHG!VHKTHVkbLE^RpB5pW@v2a*9O#nf
zh-!g!nd;fI_Xx;XFr}<F8$`ls(UQnz-QUClr~w(TL$G!i8tYqJvGsla;2C#CFmW!l
z53V3CfQo<JZYWKpc{riIy0)ff%`Q#YXLfW8wvE)jM1L-)FOYzbBeYcM{cgsGo`lCh
zBW@&m*Z19V$TS;CBAhOFLqLAC@P=WQzPf9J$b6yJX2}d6m=Hkj*D6d$hH`^;6?^=C
zRl3OWX#sziiq_3rP;mNoM(4_hHRbj4BhFhpJcNJ;TO@=pWfrAXoE=}t=C2wVJ-)AZ
zPdJbZbmZbg^wQRjZMx(ia`m0Fs$~HUEkOT|K;c*Jv?=%O`Ev&sm&`Yz_=5L0Ct$Hq
z1O)26yK*5F`i;Sqz8oL}u5+dX)BO#w&w2Uy6szrY>zublffJ3>^`8ZNTwF0pNfcOT
z2g>caIv4aty~q)Fb5jdsWMoiJ!FFIM0gD+>qyGNm33l-SaIfeGh5oR`F@#08=cAab
zVzbzkLM0PhVlnazwDF;U-v{e|RtI<%U(nJ@qBXoJarZxWbuv}RyQt1ZOPIAYcA^QI
zp4O1behxdT(mMUB5jXW;tmB6KrDav`UnYu72B5{{#n4AYh1QpSQ9JHyLyLLRK9jlB
zR6WSN|3iX?4!vR~am0siQ?R!7J|xOX?iGLG=b%BO@?}rz7gnu`m<z6u!k84DMclE@
zIfAV#H9@=FyOU15ikSp=)dJCQe$1o8Jd=?VLQJG0J^t6=1B@y?b~wrLlaEPMDblJR
zUQ7d@424o`InUjz)M+7`QKLSUFCSXn9A@>!!ens1Q7+Uv?vY^`O%J|rlqrM5!ooWJ
zJ(8hvYjdEMEk|ZHlnM=i2qEMBHi_HTvG^xT7+}<bPWp83SM^T^*;*EEZf;;H63k}Y
ziCc5O-&Wq;2lgW9=r`?t{UU&R*5}L0${3Tw!G};33}|i1dvj1`y7yQq3SBQ{Vq!jN
zDO=*CKPU22BR^uxSUtN{CDLmZYQnH|?hFMtAq2PsX+&tq`OjIdw2ku*C7r@$lTHtI
zr&T;xa=0Q>(Xr$<n^<^mvGyw7n&{N~rvxOKA%`vWO9$?o@zXY>Df91~9l3-S*&q8_
zi9iL(eIq_eJ-w2B)y;ANYR+3i$wiI98KrDK+)lgU3<cV3Nj?ITUkNEv?4k9%N=u3x
zjK);2Cj?M+8w&&O-)MZvH$B7VaR?`V?)|A5i_~Du|FxJ-<g<-QCLOYr!#`)=UuQ`C
z9;?C(x0<MMtrLy!SV7do#>W1h5ra+V*(mH5Y-dOj3`zy3>f77G>$u0@WMZcc8`mdX
zddg?&Gt;7{b2$fm;)_G?Fl+BIQTn=~c{2YN;=V2WY^i@E+t_Iv8fI;54V&|$2-T4c
zW{Q3$3hGDr2*+tNlG`#LS@HX4Z4_lMuM=7WvK{<zj_KwDHy)dr1Q!Og_EzmfC}zpv
zS${JeDjzz#ii4jgvow0h+eRf;ELs1;(h??oPqp_Wckcv=$}l;tK)NcR0#mX#91>8O
z+m6*bk{4ISp<nNrG<w*D4T>z~v(u-Kz9DNiJ2J0|Xh|8gPiFE*ju;sHsUFiTzcF^6
zwNc28o|52_csKI042${4BcG!O%|1K<viSt;yWvB>N3Au8=B!}}?!)mwdq)DLE<eYQ
zzS}?b**LK#M-S}5X49y)BTKK)IRr9!CdZq34&Psu*5+HzH7+~&ERCm+#WNm(rdzE@
z$MVPtmcChU?e}=RId<^EkTR4A$<$;u*LItk5KqN3FKTb-x75FSY_6Y$dAQN(dSSD$
z`|ilyMn5>1wHQNhFwDhCd|xfE`e3`7_MK$X;H^fv4fPEk!YzH$WEQ0^gq>szVnfCD
zbbSkDBqu~_u8uv5McwN5rGy_Xti@Q5^+(=l#B%eKn6c9mIwaqS_aeqiqWsge0pxos
z$1)AQT3~m1pUy_<)H|zLGO<<AJLjh3dK{QN9Q@IOtzPSJBX|(8+igwZEq}daF6~SX
zPPtPy#>P_7AHLB2OFC#s^Oa9{2K#ax?dTmrUmOJ{&G8MLp0{yoh3(%sj3c^cD4x;u
z9)J%lZeG`CdmRz?8%g$#)yUZ7HX}M=;?F_RBc46>ZE9y^hiMpD5rl)<ciV>|-X~IW
z4%O;Z8siF<R0;~s8sTWOo~={i=@L|7NC=^3P5m>g_xO-UM3Ij;?X8wu7el$4jEs5U
zn>ctvRwMd6(W2m7)ncN7Sz%N*%|mOIts#Vx72+3`miIr)VZo<`vdZ=B)aTBYl{n^P
zI&V3J|4{GgIDL8z`ci>EysClJ^PJTJTnTw<<(Xo9GHDH5CTpTCx_-h%Wx9QCw~);6
zCAzem4oFvF-xxvtf)hpjI%CXI^KjuLS-7lx`h+lC9Djf?7uPMHw<QkK(7m#apfG<0
z&VKT}=AV|1U`!F8tp0TTzXw<JL%l{r3qT+ANl8iJ-1jP{B&DU5<WyQhMn8YX*%(U0
z_RG$u0qLaFf*FvfC;{#WDBT*IMPPqI$VsoVjW<Z{vpKg8l!^2T69n=rB)RF`3#o6?
zl-6!7=AsBOf<;3Fc==D-F8F$RIG!zOiT6ZtKG4^BU!FT&@GH8g@lz_(4fx{pj6h4T
z^`U*MPayiA=A@D{Bspd~!Smj*XCcB@?i8j-xoGkDD3M-qyQFx#26hT|6bt-Nq@b1B
zy_Sp?_W|ZJmgL}0J62t~EPNTn)03bnYop<vcn*)JTe(ti$jh8o(Om7ca{F;o-8M^U
z)s)>XDukzVFOT}@0@G&o%U1uU0uXZ$LkZl0(+?0u=)+|SiN1UZCvtO$pjF82_7<d2
zfG+oqeO->z_oe&u-R2D>B`a&%`W_K@D8zsYMUxS>D9DRti1zfM=7_xQIZ9`cI(+fC
zpO&17*^lFP0(%0!o1^UDQR87=$;meBZ?!&lq0v7A&f5)6Wak7K*i9EGI`CiTPg%9p
zoHwhZwU0fICaStJQcIX1TE5O_J$Oc6nTSep|4BZE_zn5-HFPW=-FDGJf0!e2biQsE
zL41+UJe|W^<Av}Ml`T`C2}4W|!}L0>FXsFHc*|G3tAn#>Z0Dkv$Aub`h4IS2S;2A2
z$Cp)Y@}PN1cN{BDW=^;lIbvbb(I!>)zU1EldKTA#-T(*qmxH#AI0OV;p)5`D*)6|{
zbOoA#{;6;7AdXpA8emgQiS59zq@=Ra6j&RL=F5SDME;A1QJq^aPw$;f@XB<Kfx8P_
zRiU=l?%}e%kN5$l{i{5ztf1Dl@)90+_I)cd6Cg8YK%4?y(Vg~%8oI&8S6!6?oklTB
zp-CSc`?vRxNVEkqCHA|cO13sq-VED?_60+jY^*oDH_l|pH0IPpnlfWm^0-7Id~6ZA
zN$PYR!>ztly>t{J$nL!%1!0_3#rkGi80fR7J8_^SAk;yelhsKy(BK-*xwx(_9^R=y
z#G0u#W;)_2Zp}#Tede3xK8l>jkK^gg6Zc#$9{=^<o&!a3KU&J40OsEWz!Qkwr2DgI
z*XhnwBp|i08vR5GO}z#s0;!<;X8_7?RE5G85e{rkY7QY)OQPZa`O}w7LM>ylrxhke
z`lu|9Ia-Cw9Y;<cUT4JgmlN+Zqz(;vPx&wcB54ul^^grur7pJkx8I`d^L&~SAx*a$
zxXP5yZ;u)}+B)`m$Yl#*t8*B%zF8-|#cmJn`nG83en#u<f%?7NVwO6dPlS}-*|4&n
zY^Gj{E?9UY^K_C+wFDau+0C9+C>Rlh2&}z3<%F2|!tY!D^fRxGc|d){r!1+z@w)oJ
zx`b$RBqu%8@#dTEq|feur{f&^Np@cJ+e)hmjao;lMIY##Oi@u0xZ{yE>d9wH$K>cW
z@<&8a?Fy-*fiB<sYAc)5b~YNH2~E$offo$7NckLpwD?rBd>r3$1`tj{a(3mGvza$5
zOEv})0)NWjtR)IVU42|$tHEcP^uFBBd(8|avh(PAxCK8{UOsWVpo#GLH8s3>`CXd}
zu3~pM3CglfHMx&i<g2)rA56W7`x!_WhQx#w!F(}MEOX*2<j#0O<eiTQ7k_%M^k)|H
z(qP0@G;-Hu5&RDa0|o~+^^t57BWr8<plvS8S^juHAXO+<{Pj4E=H?XMv^}|vm$r~i
z$EN1p2J7tZt|&nygEi=nKI3@lYCatt>-TMe2IiJ69zV*)R%({I_O5>}icLo%JK(*W
zSl5azE$A^+cW&1#-`&=@vcLFf&0NvXJJy}h#1!uqx}P7qzdh3$-uZJpBmTj$=q-9f
z^~}r+kYiy2^j&eu*W7$_4j&8rlUa&2KuJu@%#K0cN+ewcz<k5~-SwM@p8oz95EWI&
znv;gMYgJUGa?Qb<y?t9--M$^e^WHcT^&tilJee7U%BOs6b~=p;dFRDyDOY28$`IVC
zCz+=$w=0&%y*~H(EzZ*tMS%w-bhrEA;?Fe`ET;u{H!XEl5>H;tg(-3O`|PSnW87vs
zudIFt#j}B);FAt<eN}G83#||JkcZHsG3Z!p@!SxSIX4Epkq#9n86-c1mUUZ;#fy`B
zIq-N0p?n^Cj2#diefl_ZpqsX%t?t}UEuYEJ(17x1z(FdmduYal6_e6iKDT{R0oPrF
z-RaVr`Xs*B;;>ZwV`|eIPFXu$#7i!0E}nYW!e-c__ij%kQ)zkWw~-oz5kP6a@d=Sa
zAY;Ung1dL+ZmqN5jTvED>rjwAK;_?w{j%Cbd#LqZE#DM$N5C{EsnI|i7$OmIn7#GC
z1{Hulbj#)H>dMO}R8wwXYdg?@TS>}B9RAyxpOYBVib0o}aUUC0sOK!dKs}GW_8?q5
z<ICaLHD84_0%RU}KuLMVpRi?F<8`^D-H0_XE7Xw@dEPqE9wwZQ&AXUa856F7-e2=c
z-RMFsUv=eP9W3&u+)ZDrE(x!j5esVRRL@|4OMCre)tj9SVF_kCy^T$CLp{8FS`r5{
zVOMdgV6DQ|d{NVJtDOqWJcZoEGQ-xX_ii2Zh)wu{5Q7T!StnjMyI&*G0T*9nqdDc~
zBhx+u*7T_zis^F{Lq7f#=5t5$OHEv@LVN&lia)nK>@8mrzF!n_7+6@iYw{EVG}vx)
zUCQ2CN9Cp4t?V?A{f&Uo%j~{w@8DnvQ1;O0)P2z!CI^@99Nhe;I<w|(BvGCahIh_A
zD2Bu{YKwj87(SkuW1ZZ-M-3#Ok8ukrOswb@`~Iw*@AyDm&HISEG1Vu)mqo6#b!3LW
zM@upEOh^+{k;&3(XyI8+riA`adf+N-7W2g@4u?`Omah??pa|3+42ILUh|A8#h=DM<
zI=qiraIs>0B`oqq_%)4zwc9--zE-}TGGw}H!d}6X0Y{bqFQs*ZkYr$Q-*~;bs*R~0
ziS|P$ECiRVCh>^;^NCiDZnLI;aZsYaEGTVzOicZSIfj(;?VixlYx>J&zb>9Kwzg(K
zTnL$)zhzuER|GHL#>Qr?T#RpXP33gR6a2CBpBfy*^76X&`l9x+n}Rxs_P!Sy5Eur2
zZ*XYI9$xhZz@ASU_!hulfC)0%9x5ZKCK8#f;|>S1joE&?C``bEE<%l5VFsR)UUKTS
z2K{aXjkOFTtn`2Fb+?w{v6K6I$A0hhAh!-v+t~a-(p3d1StjU;x{3)!);~!ToXdGS
zZ~Qh3&g0D?^b6nChhY#B!i9inq_p(6;;)Uzq?ubwHuSrnXpFT<MSO(9U)SH89M8G9
znGI@bhIV<<lX$Hdin)xLf47T7NH-0+FIO<jQq1Q9Yac}h`C<#KF5lB9;L|FiOMESH
zbIvjhl|#g`oLRo-t+8%&T$E)xB=(0K$?MPlmk)Xqp)4<K>+GjL5i;g*-R&9dRE;(N
zXrL%A*xan6H>F(HDZL=w!)|0?_~qs_`I5tJeTPUS-p9W$TaNWI)3smM0c8lSX;NxI
zTFY&_eSbity-YHmOyx?mK|I>nAD}6?0ypRDX(d)8^#&Qi4qVGpX*WLlx;D{m`ZCyr
zyU1y&5Dn<H4#F&klJSroT+W$Izs7FvtxDu$HNtA3g2-ThYTVQ(6(##f#3b5vr}P(A
z30`zHi?o?7*JCcf)gF<)rPV5~!<s?j$M!BACYrLX$1><3h?2adk@J|dN_dAiyzkBn
zSiddQ4~35{wjv2&6|=iv19YctovmO+0U?EzxT394s@f%%Flv<e0kJ&p1(@sU!!Pab
zCVM6IFW`wrBWGjXx4WGzDxif>Oa=HhU53`*tM-r!7h;n&O4WFzXg;}4sFa)zu-F~9
zFD<Q9W_G(@_uN9bamSR)Hn6rH`t62Kz<1i+OcJwF)QGe(scyE9U!9|_iX2Y>JuDRZ
zSnK<RFAABS13EwmknlmF$Mc3%6C8Ai`JeZrgfJo$@}G5PONX=flyW{GsmgN6Zj#2H
z_aKN~@v+>M_4_lz%UjF(?q`K)2p?bl(W%S<{gJuYODj5P<2(EEi3}VD84r5Q>By6R
zy~3J1tuWc#KG$<0^FW;Sk5Fs6`PQz3^zy+U;^FbeHRNA82BNko@2GFaeO+xd^vpaK
zB6mrX-S+|xViM~6irZKesTg*N6wN$~8j$GmR?Tld=sps1beXk@9VE~8=J<yHur~IY
z6xUl+I7mW{Y_@g5HebfqhKS#d{b87RriQ>D9Xt7Hqs*=4TUOqb2QQ)Vn3C%(3n9;5
zBYKr0tojwheZ+2NL%sS{!$}V~z3U|}!G=$EN5f8s)||H24|V$d=r*OBH0C=1-w9vI
z<<#>GGqDPldB)+tPAH`9wc-*U=(}l6ei0YxVW@rzp7)WaXwkzgo`hSP%A_!471Uv}
znMD?9A8TJ!{RB1iJu~dp<V{n%fM2#`-3nuOQbi2V_l>O5O=UhNy|>YM^>!SXO^QoA
z(PHGaT8bftyJ@(%okR2~cvC!e^vjJS+MZus0iQa_86RIaSbO$8HanrA1ZkKsYRKc5
znV!nz8pk_aWY;@~AxD-^^>NtB1UH!#^fSch#i_voAA=x*TjMf=K}1t(yI6li`~D2s
z&AX0NiNeXl$^JA{n`y1BkGH*a7aZbru>>!4u^Eb5cfR37y{WfD@VNUkleMS_Lm6Os
zR%PGw5RLL@`!kbY_Pw1i7e~)OH!^kOkvmY`opL>sqxb=gjPUyoULDRcaO7<3M+Q+~
zN<f^bP-n6d{!8l>EWF)l(9+?P$_IC2z=6o=;d3QTzQdg}eQ{k?tjPU#OVvx8!wqbV
zZExN$1ahks!^5W<X8dl?0<|4@(o0F+Vkz9|IuqS}2*zX^9>&|Q^UE$&N^baq8uIli
zUG~MV&L`B9!5LDl&)<LEv$mM8O`F&sn!iDSh}q3MC^zJYe%3zm0_%mhn;wL$QiZI|
z^}>W9bSEhD{Ue>Z2Ey|@b}g4(-IWvGdQaBXBnaqiNVV@T@bUHDA3TdWQE`G(Cy^X)
z?cn5%#n3{q3dal(hkp7kA4ppx{kK5EW_a$4zSp*)j9OU}`0P7fqQFSC7xW(_y{Dw2
zC=M({;o=b#(|NcliC*yp1>pqAa?3yc3$b<m3DQ@UUt7$S66pJ$J`QoUf9XbrP-L=3
zVs32Rl|goWxb3yLytJldqtmF@mD_&r7llMzn-FJ0BF~%F`4K_8P-{uZ!C)uF`=uu1
zzM_0)<WU4~rfxf=&7GqvO$l9SHVfE_X?gmEe_Fk+o~vJzY5CO^8QDDFeg{n+cYWdu
zOcRDnvfm)tR4qkIxwj|uNlW_>B}1o`u~t&)WY9mJ>3qQRX=2g$XQ|Kq>uLevMP3OS
zbuHv`<Keyo#h&6=sdK6S`kuQ4)g|2$gI5Uz=3q><7WXCgtvIB`uHvMk#T4JZ_{eYi
zw<NmBnSrCD>*}6-4ADr}x&3l?b5isq@RDlQs<2p$5-+TarZ+S4#^AT*9(PvrBu~n?
zx;-N8Sg84KGybm5DI<<4EEhiGc3FpBXDCpta8&R6dos><s%DuB1BwX3XFtR`qGA#v
z8Uvo3qsK_UNg9N1Eyrp_Ku!McObn%T<cmdl`C;CYOn8g<IEIbUm!>=TS4{5acS!7`
z%DisgETG_*lC2E_X0f2x`k|Lm)obMU@8qK>KC~Qp3KDjD^x8wy)#e}$<+Y!ZQoSMp
zhlb-nT-*{X_T>PL2o4xEQg=u--jdcvr)|<^LOItz)aaN*r0YKG4EjjR%GDI3HMpG=
zEcJmJu*Zi9)K&)O=9lmDGotd#E)xHPq~co&<WovZOJjRMPbo%MQ@6=D9b@C4PwQw@
z<DtSTGKnx~!TRoQ*N{1w?D<)zD&YU((W_^=HkyqWO*N?8${z?y@Si;C`)CO5ZFUcQ
zEv`>)YN(Bj`tQf1M8&HDk2MybMLw{5H=LXuJ7xx=Dt?P~4>QCyByB74!{PbdGDXh^
z&~OOF04S()v6<-^n{ZtVTPmKnht*O(Y}Q6fni3kmwZ9%vn1V2CosiB&d;3+^<(e<g
zW`47~t2nF7f$p^?gxjx@%l~1ZbG{E*z)Pa<1<n6p6gtgMnkbayF9Ay-t1#e|dEo1h
z8;Iz&h%d*uu{Wt!zDc;9NH6k(U?}wXiBWWF>Cjc{4p&UE`R&v080{k-jmqD;ooDE4
za~~T!Iwve_EN(2+yT{v)){MFcdQk#t3m+wFUb5do;Ah8#`=D+h^2eiiJ|7?-ol5?1
z6*Ru$x(DH`!2i)Wf7RQ`qA2}n7?<WN*jX_*R*Lm{9HHxH>C-38X&whL%)JHkXL=<K
zY~GV_dId;A^mO_xY5|CtOZC);@dT))if_p#J~nEr6Oc6PR{QsM4o%5I5m<BY8Hk_@
zWX>O-Al)Yc8saGuT8iHsU3gQzdndFR1nOtNS^RSGLoEL<bAAYkVl1NuJ1j?y%u_L6
zkApc?0BtAZoBaAHV)t-+TpXOc*?36iHj6fSvOsCv&>{pMpoqO{)poB+f902l#ma=k
zEtaA09E5lvBL7AodO+b(F$}$irJ}?GGsjzJLYrFDR^dz;c2(riVz|RyhB1K5W(q8{
zjay*`Ncgq@a8zAS0G%8@M5)PcQoKFWd7IPX%}JMMwCwl)lKZf!Rc*;R#->~9vpYu&
zONX9mzL{068HD>-|M9)qhmm<@<R9Hjgmlt2f^fc*D&|f=VTq@BptQ<FAyoR*w^5F9
z=<radC*81bumW01D3duA{xL9UbT_^U&=uC%=8KhmTl#c2Q$kHLU03{*MIZbW?)u?n
z;&99iY8KJL$(GHyWl5!;sQ%%0iDjG^DbOic&U{nviDY`6;wda2U`DoN#NYo(K~ILN
zk_J9j+z2u@AL^FU*J?Ci`n&OZSppVXmQ@8V8^%n0$!@76d}lh=Z#O>(QFQ!?isP+b
zC%jL(WYBj`Y@6k2raIC}u2jZvGf`tX+;*-L)Ca2|&0Jz=t>>{xvR-zM-EO#Exq@yN
zH@ZBa!24Me3_cQgtt*xMyg%6Jop6=%6upFGurm#`wQjw?B?z!hv6Q*Yk3I|G24e+)
za3S#{G<eBb=6++^;Nwq2DozPr?NP~}@Z&KnsI&j{&8jlET0}Ph%%Y_4H&D%&q{T;&
zO^KEO$R|V^Cf+HhEB#WT&nxO53g`aY<VDu;YG_*NR)Fe{4yyKyRMt{%?_Ut1uRXj7
zcM#tz))s;}TfW;(c>MDVze6&bf)vl%$E2iTDJf(;c607IPBjDt3*TFNdQc11lCB#^
zEUcnLK`G1Yvc-krg-nC*d+}_Bp}2LCXBm~kVjaE-YbhsKTZ=ai?Vh@2-<?&ov2!n;
zbJI;{zOm4UV86ByTEjIBQH7Mv)?;JoH>WeCY%Y!I<grM#*AB@8Ize;ZwjvMgyYNv-
zjH5Smg6fY+XEp^T%#{q>ME7lWK_Xyvn6yexo_{M@wp-(&W-?jU*q=BZd8mACxF+>`
zDf5)AXy-v{RWHU`4kNES`dvg&pv^8RDbf9i^sMu~{a!r?)}km7Wp8f(nsY4nRdh}0
zh(bFm^?2~h&4vIS+m<eJ;`-Lld55c9Wl()9OGfke`<pn+VP)#)8UP9yP(Ja}fUSuh
zDECkdo4V?-zl*1c7u-0aA-ffZP!!YYBffJuS=m}X7zS9ajm_S^L)IjL)9o{LmQ;VM
z2%III8~7kp$l0GBO9xBjm5uWYJO*_ZNB20rcF=0_F%ZaYSadHQwyUn^!|m${W010i
zFWqH~SnLl~TY;vh|H6!h1iF0BwFrsM7P(L3>Hof%Ou1$=+3mW}^!Ud6x9m{#CBhV+
zY${a&kiS#NIYMh`d_Eo-*I}8+xNETdP`U3h<k(=P)>i#$q<InDL8SC5yVGhSPYPDG
zl=$Uw?Qj4%S*OuGIG}#=iKjCxl9~L-URu^opr6<t{_Z26Sn2f`q<iaZYc6%>WHoN$
z#7n{)KM5U5`lx(@*P%>@55{xGK3MA8tT_IZ!rIpwN-*}S!6SL2X2wsBN58tM55?i{
zd5u;KJmKX{RP=6=u-V()I=K-87AYMgjY+&W2jX^O;-BMj0y3f)^n`r{M+=^Vb|eeH
zn3Hnqf9r~9r*BDE4!-A>IommmpLKK`&h4(AlDPhlOOV3QGGOD}_w4dIKovt@h-g2G
z+Y&pEOu$t`ASfhw0zz*9&Q_oW0~6FDm3%z0O08F?6x<!NxpQ%L9qXrh$5k!6Kyp#b
zCX)kRFL;{no0CNvu<zUpt_{d{gr{e7|9*@QAu%J}gNQpZd+OGH<)}Mh7^gvG4mtw9
zUq0w&=bT(LUZ)5#F=5Kf%Yz|_q@-j^XV%i!dB8vX<-VPpO~~&|y9m5>n{Kn^)6Brt
zFE$Ge2|(VZQ15DsghAK>xC)2I$Mmdbw^x5)Zd@Suzo4dO2j0!xc5~g8)>9W?&JX-|
z(LEbefiZlE_oX8ixu7H{IKbC7KGLKb0CqQU%!NcH-o0Z53;=BPBAwQ$DIRF?Zg^Os
z>2ial%4X)Pz(x;EnhAgpIjy4aOL>UHVGOp8P=Y6qAHUZt&IUfe4L8Sq0IbHYtUHwn
zPk3DZB66zH?9YQ@aFiTaW~SG0U0vRsdh<N^sAU>kRByZN<LP9y*7<M1E2_~EJ}mjT
zXijA^&|<?*3#Sqd2~zjh+(eyszevT04NbOykv;JDp#)4<EHZvvUOO-vv8dDnUgim#
zt{EfUfv8X7MIhhQU$0-kM#aQjUyT%jaSj0g_nH3N6uA%p(M{y3hLI(?`8)y)1@M^x
zhhrD~(4i*fY&nPRaXNtDz4rF@ZhGKocx#V|1R<<;MVc?S1$o#tJ8wM(*Yp^H)d!a~
zC<ChToDYCcA0i_I0>z+U7jQ)gV$!Y}2G;yAf>TtKa?=T3=XiXFklv)i)o2x1f+M(-
zJdHOqA9{N0pEGN3X$zj=3cA1{k@@>eMDMA~x908Oq@|q4CrM)rp<Uh+gOmfo5@(Iu
z;K86}$K>Q#!^=X!ml<_W4#+(L2mpVvonMSLmzJE|-x*E<<8PSuHK4aL{UfJ8!NvUv
z*quNG)?&c=IXe31S1OtA!NDx>#6PmK@ICJUu5`Q-%Hp`%i3z}QfH?x9SGmiM+S&d*
zOv(>fM3e%M07*dWOEWWS@V8~arWRHG<cbd_nr2{0fI|z=G@yO``jwRs!^GVD2|4-T
zT2Iu;Kr%Q!jDTYM!*RWLcd?nGP%Tt0dlk6*!x;5|g?j?vt5{xSnV*ciM`cI}tqg%U
zE8mpreVtUSxH^rcdf)06!mH^8xQu+Zg%pr#XJ2+CYu1*|;G|}G-#U_AU9~BQ4CgFq
zJO*jKa+|As_V&@g^5W%6C9i2lY=X6B&Gbtuuh9&pVKWoM9SeJ3N2+l!PXZ@PqtTPA
zFP5o&hVQ`V5?sLo=YIQV>IC;Cup;wV4d*#LJKF$&BriYz#y+riSY3zW^u>#(PI?)N
z569$1){}0%v;o5lW(Lw^v<GO`a2gHn4ykGwplly6)=vYscEnx?Q&Ul~nSL6TI|NpF
zoPfpYmR4BE1ZYXiQS!n)PHzFX7S`$6F$3$7f%&N#2V*P>;V@u83TUO9H<u?spo5zf
zLS=-1=B^0^^uzMUkpe)*U*2}^Eydzjb+!HEIJAx~bMj+hw0mf+MSH){I6p*yLHlOm
zqq(6r>5Cl{t66da9u9NVf1MR&|5slIbF;j9#0~b1pOP}P-c{C`z;(;8Kb{Q^L=wKm
zJ3glkxWhJdmbKd3H0Z9?fR)v9^oL-_JXHVZV3Fc4?BsceEqsyQVzj<@S4$(G%GlaS
z3rrt=!GUR%MCyv2VwL?ORttFI6I>(vljGx+$|*er10aMsXEZa6jN^*G$rEwfzz{cn
zbc-%Kp8g6Uv<r6Z_a}F1dH9@cuo-K9^v@RVJQ&;dlZS^=)Zr_W4Jt&S(?Sf2;PwPu
z-ZZ3_V4^qwYwa(EF%ajHw`ToRN_HZFg7M8(#`bTuW8b3K<53wFHgHK6IK@%g(TE>D
zed@I(J6xRkQEfggmnxXxw&+vcQiBxWSh<qGVU8x`mcs9X*3hWm><gpOsO$U00D$fA
zxR2!o+<>yOxv36q2Mt7!3!#19Y{J`gHM|7zkJ0%p*a9#S4=>Oe2!(1+uL1){0{0|N
zOC^I06Q`IrBO0(ntDPUw=x|aX31*=dYe*bB{ywO_Y`G@bX<UAE=F84#sS7cdll!@4
z^04m)!9nqb+hQ_e(=(3rppY&9G7Oxkpp<e5;fxt8R;mbhd#X4RP@-Vc6u|MnxVV5j
z1l#uk>@|6Jfp~-<{pj44vYPb`(*rYe^~04BJv}|0I_Hmc%2_4gb{7!6ANT?4GKglI
z;XDi71LeFRN5FaDD=^4;dkezh(W}rp_wOMttgNi;YxcL1vsiy7Z-CW5<osazMBAau
zq8*Q~Q3mSkGb?u%9PyMjHE&LWL2}BcwD-~#U?x4qc5aW}pqoAQB_$?2BkDQ*eq`TK
zpo_`0xdA8b6J_agt!mmd_T5ncKz2-aPR{0jU>^^8G%5EAH^Z#{YWHrNWg3#PEH=x_
zkV@xLs8`0(SU&cOfuuWt62}AFgad9Wz%=d*Yqj9qs4}Ou*vQY$9%Z)!TuKZ=)|LV%
zg3h6#NIV9$5|f_CFg+I-r2{UX>12T>AgqM|TS`#3sP6JLG&cGR1TB26(Y0vQjb@TG
zGBQfNIDfWrJDLjL|GSk0AHOpQoq+L<3cQi%O0GUgM2$PO*i0*u8C#q1`>EyCtt;a{
z#3pb;#TwsF<_!;TD#jzrEJU^k&2&3X<4~I?IvtP`68Zs#A;P<}T}`{Wnpc1qf}ZR!
z+u-4}H2BQMw5PSzpISDN)vTWonB{b~1)-1Z`~inCL#&I(I!mw!#xFnOBjL6|f#$Xr
zwLgP+3O8I~shr<YK^<j&2p($es|Re?oQs)>69X6oo(9vo*upb?=~D^k*9^*}p26Oh
z!KS4SjK57wAdabgq2kzG4EfFN$k5@0X*|h8?#yJv+UHnu*$Us(Oy8$kX;OQ{27(O+
z?)caQv{f-Nu|uGmLx5Nf+X5eSxg9K!`GM6vT<|(j1WrF*G4$Z%Ja{T7*dVrA@3Mo`
z0=yt$mp{O0FlX`1x^-~9w@E@?u1#lHQc_X?xrS4~w`cfR_Q6t?dI%QLD9ObDg=xgh
z{FU<g=zW_Tj{p~^!*k%y1_nLa%&-1iV<{*RbN}Uf`gZU4Zgl4N-*9&XQ{?=InBPlx
zJUI{w&@!I?Is^6xe7t}gGmMe>E}QpCpPZZ=aB9*0Ku#?8{9~o0q~v&RZf52P-i6sf
z;u}z0<vJzsIZ*@h><XaY|90+u=86>H-`y<<>Ih8}JH+*?5eW+T;WAShU+DhUZFFo5
zK2zi{hCw}E*vB&!AFHXU34@6Jp&z*L#Nl{N>R<BwKkVF=$LostY=u%V)HBdM=jUQW
zmW@hj%X?Hlw#P=rNfa{w4!t3Mcp+v$6M0|tt=Z=)DttT3<-*zCS&|?&Lc=jx4QB|^
z8L3XrGf`{r{l4yKP>qnaGq`~TVCMX|5?yV@#NPjh;pzZnVoWZ~hK7c3Q}`=~Z{gwL
zg9e$IyXy~{WK2w6fV{&bFLLYbZ)`^pVXqA+osS+pBISJtTdPfyMG%$2-cnv6?WNDE
z?8eZe41{p7!EG!mrtKpS?EU%J*hiH<5z+qJR5b*1`R;7HaXl(Iszg{+ZRlMFM|h^e
zMU&|m><@U#4U91w9MPI;0(%YCK3WD?UMsofZ$L?*p)#o}oAGSMpGb?9a$)Zc+<5pl
znfFUDCdu~^i+BG>sBaTFKL#LUkJds?n@XEy;7s3L?XZAv5eIYC5erUtclY5+Yjw2(
zjaL>Hw1b0#R1w+WE-~;t!2Zq*)UqmtT6p2&uR(YqVl#ROgqgil?7)V|9}tcvi*#iG
zoy5V_wG;e97X_3I^033oPgETp9ch>5-e*~rbLB)0FhUNo8>IN@${O~(hRkj(6sD`O
zlQ#x@Y;%nacpR+7{b8JPq|?2Cr}Y>*;RkmNx~t@b>4$q(xECO+<}KdG!J+=$#}3WM
zdCToG)(eJHZ6bW9t5~Pj_ARp{y>bL3-QxlKa@R=B;=s>o@MRtsc8O9aB6z2Z(mvn?
zy2L}^4DCO#r2?->am^QyX+43h(zaejH31#N+mnp}SS9jtHFv`5>TFL2j0Txv1L)X7
zy&4L>^`kX_k`8DoLV)tE4CqM&8$~_|0M<+(Eko19HZ93Aby0y=H7k{P%rCakFNxlv
zm5W10tHtObcF?}-KfbhAs#YX{f}5^T!F3p%A`<_OB)RBWQG_#O{<JW}ZMA<f<m0r*
zr@%h8ADz_3n=66VRM<D#6PbEb)Z;eF*eYfPLzICrF;f0KR8&FWW-3Qe5tX5^fc1>;
z7fGjF5LW$&U5~`3snD}f@sg)(Nt8GyZ~SMmc%xBBngCH-0qQLzDg6Ono6GqyICdY*
zHA%;Dii4c!N@)B`qORllj~!BpFX-DBuQ0;JwKmZIenLaE>fY1}^2zYYcIhgVcpU6!
zq326n#NnV$-^6#*uCQnYR?-Uj2g%!0`jF|$&yN<I#BiN}KMx`pU_GW}j=`$QW1ycC
zTrkdojtA&D@FlJlH46#>%&vfUGIjUdi-*ltg(kD+SM7tF;fGn1{st5lyGSm&>4h{H
z#aN46!5=GICxT`3o7bYMuhORo7FQ(yopvz)=m7*>ARR8ZUlfJ|!KVu-Ly3S6-;RyI
zV5!9)CR+hKA8@f!3$QZ<7xzGb33poR%nv@2n6OW<+|QW_ggdt<Sk$q?Elt{hH>&Pt
z?}_b^-~DZhImL88`ql4>8P-_}YQ&d_^=;S+`RUqkF3KoXT$aFe?{fL%W9ASbApSkS
zKVP?kj{;))n%z)}0FS)kdGpQf?Imaky8HWoNJUY<prO&`hseNpKcgVvkr$WM`%#d3
z?eH*}w&QxJ^hLZ2c^Io_?-Y=i5b%$Kk@26djPAGcL9M>SLj=;Zvwc(KX5$D1BzHHK
z@XOiQ(SrYNAOKKe0tVbUa7C-12Mq8BU2~0}XZKXMffcM~lXvaE^+y{Qz92$sx`D`l
z>=+(vX6nAq`gVxGep+%&{FV=y$TOAb_7Rw+(jJp`o!>@sbe+BAl5qcKwG~yc0{1)*
z;c!*`Ao|Yz-;;tlVrqxFZ7s+J`&A(R?4QlwqK1nb@1HL$EYw!tgDZ*}KzNCjgAF2R
zhdDBYf7XAekTGlibQc2Umk1M+pS-8GbDO38Y1V#Za=`=twZZ}npiH-o>Q0IEY;IDL
z)aXfuF8w=ys(J)4wB=bR^H&61BuyQjv3*kjGkWmu0&%|2m0e)y0n*I_`}0442Il4D
z<%i~*>JRII`0gKr9XHpj`HT164&AJ)2l3;|Lv@V)4}Q?mx8A9Klxe~MdQAy1Rnx5D
zuMq3d&1h};(wNjDj#wxpE3c68-c=8dbpE<}%i+J@CK@e+pVmr(gYyzr+-YiRf*Y0K
zU~**O$oAkRAms1wE`a-(=8Pw-LZ^~IMg76(@^Z>_CA;|xA;rKl#N&80r?_ILhYY2Q
zX6AvRK-fm)pO*G~rH{s0FFK-PzoKxZwB#Q8o>D2Pdklru5gY1}{{O{}0DL}RlDOUi
zZ%P?|ap8V*a|7qT@{#5O2u<OqE3FBDiUM&7v<<agRaO(d|3ng08vr#bYSe$f!15&~
z2GOsvu~Dk@@e{}>8dF{yHj4N{NMP&xzuF6D{ZmJFzw6d0mD@Gw*J^*klG^?m_?a+c
zP9ReNx(?ym>SlmO#L7tk)b#}zM0B5I^SW3F7}A_S_St(N{`wKR9jYP*a#ynODUs(%
zKqdnlX^O5azaX1%;^7!FFMe6Vi8+MzCHmdiP4fTNb&h~hIG|d$-nI4@pxI}Gx(E)`
z;;go_oZvUg0TRdlUNx7+upf|CX!3)zLmLfv*#}h<Hj5?6$|3i!ipMr)4#1PDk$H`P
z#^>my*gBK3BcQkZJqvvdnz0H#A^ejm4TL3eYB)W-TWN1dQcL<7bbG@``D+TtW7)as
zwsM<06ATFixeirHlXcKO{DmpO+ONDaIC@YV5`qkPaftiZhwa$8fGY>;eHtbvrt>pk
zQ2>PaqMXDWUbtwKVyL<iygdXI4c(a~#=V%#+&I6n7F<7@`u%?U`1$vlGFQ}JcNY;r
z9sqIJ+yEmf(t@NKcZ2EKnJ3dFTwnoTi4i}<Z$Pp9Nc#5S(w7Y~8BXsVibWXxG%D6x
zruN$>gz<MOdYe7`HPM3|CB`}!0-eEW!p1)Xhdcl*Hg@vg$O6TX3!mjsbomxdl^*Z8
zc$fjO5}0f7hyq%BdeuA=N9Z*~8OYkvI!_XRzt3S}`h($3^=AGn4!4raA>6(fOg42N
z4gX-?>D}TH43ci7T#~#yKQ5J9S@A#lz3KyKI9C?&8ow%@5{xaIu#@Y>px8{;b$>$&
zNX;zoD)ktcN>_UI?hT;=VxGVTnZ-lq%yyP3sacOGj|0K>%)=)bMAvNePz;}P_`i;`
zBn%rtTtfhnlAO>~X^ANUNHGAuG09trvqDMW>PX>4jrlTrAb2ys%}Eu#*?D0eKlLbJ
zA8$F}U7I|G!DL{QcZEB0mFsZ`onph1Z-&P8ZT1hl(4f?qbYD7cO2V*`3wFY9i=k19
z`JqjtrW4<Dr5&c`A^e4W1k8Cf@|*`&e60d5AfCs2q=HP{HZc$?JS5Mk8}T>1h(`O1
zL}&;EjP8uh%@P5yv4cSK8r<v|&uvQ#dT@p@V-SKWz?Aahbni~u!L33BgYJ8c7mvLF
z=mp>hlR>tP)NfSYsQq-@+S8F@uAflUAAieA3fY;(+!#DBdGR=c2Eeg#G^ymeYlj(>
zQ2E)l=F$&Ez2YD|xBL7|WN$6(-SGkq-j7Z+uccEn{slczl>G7wjD<*%JHkRW%^ZuX
z(=AiiFxDZ^Di)@+`Luudf>DpLySrC1qG763UnAjqTjCEJ*EbTJ{$ZMLJ+r!iN|-T3
z20?!=7ybzEC`ap?&11atpaRBBLTfvcZ@nxtqqGf}G*f#4Q*!MUFD;g|PG2P;U93E_
z32k4P_#!M5@~&<rG6;)j8<SD*;@j~6NoC7o>2N7x-p+&K3*(!c$DvK0{)QsWg%$2N
zzyDm#O}z*8xNDWG+Tg#G4S`Ys7Pvt_PeJ7DiB&L-MU2;=(rCf~p87l4>3K)`nA7hr
zh%q&THd}@M&VJ*Bb88XnamArHZGV_wR8Kn31{p~lCMftoeXLDdB)pGtyDZyajXyv6
zyKRBvr??m$(^G!;q}a_T!V2?W@+AhHZpTf>jQ-2gP+-mGe(2VxYV(8H*;y^VCzN6y
zcphtl2~u0VMpmk%Z?R7Uv2UDHN^HBfxVJ15)@h~1jICc6%j>0UmCx?RH%C<H9+fbV
zlX5K?XZPdC57Q$qkiL?ydg{LYRL4i~Q#eJ_ZD83^V5B9~0Jk1VG0)Ie{Uy*=AVNPQ
zks9f-D0qHY=yPDvY;5rYzQ=QW{4U!XY_N~vdVBH4G{fVlqwJY#6|x920B&DAr^A$p
zRANq~BG;qGlSQt|el?_)eBr}l_$%TG_|$a_M76I4^HqPyWs7s}|3X@nxRD{fy84(r
z<;LN5;%KuBzbL+N>G2<&hvKlv0p+)N-GM6xIve8Rw!!<^18vs$_WSymgcwLh#>NS-
z^4J#CQ#>@VV5`zQAZD8J74gdaYP$W;TO~VcDS!3o#FQu@rV!Vx_2EGS&aYpRBe<WF
znA@v!R%JCZWzW4;GTT?<Jo&R{>_(IN-?sgL{9Y4rAw4~P4mv|a%vCSe%Uua46PQeP
z2B%2D({_*|&4zK<b51;lCkrP_DfgYTTj$99OabeaP0FZI>s%})p9qC#;a~&rzrzV&
zu7qo&FO|SO;0Rp4_!tMrU{3{<CDAe+$ud)`<)P#nkNMbieEJ(m^-$0qjvSCv9(^5P
z?C-UPwEx=(<h*7BOG3Uk?oIfZ;lx~R2CQ)JK+gixa(3mZ&?hG+1MISMVqzkh6MRMu
zVVB$J@G&|dJ9>*9ofptY#OqPFw7X}PuolP~C7(`F`bhB|Khdm50{L%eb~rqQ8^-8_
z|2=$}nQLh9{ip?JE#d+MyfPA=0EsSRmTUb}&MNNT*mLhWg?I^Wy0VS-<lq9n@TU-e
zQ%dsSYrw&OJe~XK8wqx+$k7I#`~64t6&FjJZ~gyEozYyO)fV8UVKhMK7JKtXlfMGi
z1`N*lBME7X){*;i9NlL#A{5jiy&QwzV~S6~*~tA>38BiRGM5OD4`wIeuZgjf%C@x+
zC28$#Yjgh(BET{2EFfw?Jo@ve(peIWr9O1&AISdJM;^H+MwFfZ=0K;#Na#PrB5K3!
zpF*tI7<7t3aE>itj?CXM8#OQn`O6Qs6GTSa#d?0n-_U_uIy!0`nbx|QLiktCKm~0U
ztT?_`a1v*B&*<5ce_jxRtb2HNl<{f|i_`l21+$%*4kFD@pQTjq_kG>nPnnsq3f0hG
z0#Ahf4$B~%XV};x_Nm?nns<p+kAWiF(j{8d-r-bS!yB=xV_9h=g`6We)~8G~S}Ams
z|551)F2Pw~Tm9zJ?aRXorn{CTNb^vSaIXK^?;d>4?fJUhC>V4CUT_$o1D)CGx97c4
z{`d)l<z0VlrA$d_<@1_*3#2AaJf5LX=e#nMjO{B-#C9CS;RTxI!4l*i4{2y<fE%qG
zkh1228Ya~NV7+|6*2wI5t>?G1S<xMsICNNW*C6bP3*|D#a<7v@-cE>LLob?XI1v{x
zf&iDVPJqMxD8J^)eGZ6iX(=nkW$%Z?SFS&ZBun|T0X88cVHhGsI!Bv68|xXT&fBhL
z)2EFhK95vJwYC24WWi~6j?s7yOE@QL>=$4`KUI=vuKV!`UY;>u4e&}7TZVXej~73z
z@PSw11|&}iy%W=>1$5Y20xnUA$jHX#=3PKT-P=)H=mC+7nX(7)fLmu^gA4PaRHLq}
z?CcZ3Fn(|d?h_pY)(vHnpr&_pNf^dHAQrZ{I;3%KUi1?&PR}KISo=PX$%KfK`Uw_Q
z>$)=>{X~9m%Nygv0%~Xk@0<nMjYjWg%k~aRtb4Caf<<df&=eXdA=J9~NXPc66f&6D
zFEN>tSP0iVmv7+F(Bk;}Y#)&I030_a?y_=|@VZ7N=hd%op-1T3srJtTsv#$<DpJYN
zyK>%}Ym>o{o0f`^lmzwz?8IuXUg?=6x%`O@JwAQ%LEC}rlc)4i@{R!h7-ZUqT(`$R
z0^HpX*bEu65*~-|0nz>b89+EQTNnasXc7f)2`dEz+xA)RSH`byO(-M-5k6jadQ26F
z)qjy6&au}XPh_wE{R$<t)TJWhpy3fG$(+Bg{!Pz1g5lx$fcL5jF&FIruNI$YNoNqB
z3;##hW9$bV-MU0O#0PEQ;8_bvPdPv<AiKUGlQ{3KptuvlyGbg)-<*m+ALQu2T!!=I
z+MdrS2x3hGwZB9JsUes{!oe)KKp*f;Egv8@i1_XZ2nbHqZu@gRU`ofB@)buEw|k;1
zfSdwz`D1QCts`x6{0Hk75;R6T2BG!EJX~6cE4t-SS>+gR+lxCU1P|YJ?H?W4?>rXy
zzS{ka`Vt=;m5siy%Aim#D?~b<QkatUFMi=WSq!D^&esu}7jHY0Rezga*`MEbBa-km
z!lk)S?f5!Ve7yYK)-*nMeQ&xnLnq}|;`!z)lcOyQFc#{YPZfcLtz1U3-#Iw+%!A$=
zs(|&QAB-$_ZE_}<KWq#M$z4x*j(VD~%BJ2yXZr#TJ-|NYA|Xra+$5&)oKXRnh#mK5
z(9?ZHtQ-3GSG<Z89Ul?iym9xXe~%uM`;1wE`$JgSxnAb*sL&Yo<?SP_R(y-3=u_r{
zDF=7Luh&gQ@}r!#gL;=b^C)hLbPrV_RK*=@W9JzJ_1ptxp1hsg+llQwxxjijIJg{@
zWV!67gwyvgwnl%6L_?B)lp-Jbv@@U;E{G+m-RBNi6L{49bn5NzkDjnnwXpcjcb;<v
z2BIdFS}tIgoZ@g>>p2PL0!lzN0Q?;I-4D|N{+SB`evcJ!%o4e06dW{z%D2D`2P!}j
zii?Z;18h6%n~S5~fq@Ay=rgU<;sWTzv(+@){jtsP(9kg;6IW^W^+7~L%mtm%6@ak@
zJw;?j1KdF60Bqu;0Y~qfpC25T)wm?6_J^}%lRl9GC#P&gbo}3~$P_9-KMwmf!24qd
z0L>Esl!S)(LloV;5f8RdVdV!pOVVJOWoYq%2-S-hFWl=$;gHOR`li18xsG;xp$Lj1
zG!%3tjY7$Zy=jeDIFTXQHD`L?Yr~EM!Gek>bjN?nX}_zfPV7vJy;6NG#Dfi15iVVa
zbhiEt^0|$gSc+@j)ZG<(__A?T&iq71<;}@8RB<|GY4xkNIwW7P9+-1slJY8mPdVBf
z!!Q9tj!~qS`SH2&|HuUQx92h7aX_z?_IGtRHa8hqSmZ$gKK9|!^TC<{yK0%Q?Ck90
zwNB=HbG7A%RbXUp31se9V45`E3_<Jb>v-2&H{}DmtOr2&&E*YU9JbZIn99BY{$K?>
z3+_PU9LanX|8?yh@9DI0q-9$ms<}=xz!1O{O#rizRr2<3fa!bOe?5^GR9w3Nt+pkw
zR!QV{$?J?D%LkTsRnwPYEvxVcz_K2V<@z#)*$-Ua^yeKHTo~=80=QivIeV|Ut2Z`5
zdCgT{Bos>2Sr8YZsFhP>+KKf@O|D?P{WIfj8{a6p)fIZS7}N8kZY2Eg_Vidhdy{yK
z09#V}U>648FE1tzc)JVF?OX(OSfYBe-!-6beUTs^0jY=+pshv-3l$Zdw>Kw4XsD=)
zU?!T|MZsni7>q@h2j<dwU>BM*b~c<t3tpALo}kiVL=-Hy8bCbPo%ul>Wu*(<{_zAR
zR=wWM;NEFrq0`~l#2jW}d8+wYz))=<p6&JdoKs|~*FGQYOK8~GEJyCZIjoBMZT!>x
zi9BGFVGZy$*zflBZNehuO$UQA+%4K+;9n=3Caeqem$i8+xzC{TnS#?(XC|!*Nw+zN
z5b;RzLf~VPHX=PKDM)@891_AHEUfo+rx*^A!Fgyhuhpy^Pp?S7xhcC3-r%WRh;kqG
zz}RTyMvtMJBtAv1OwBm00$Qac(dFnjoAqLuiHOUcO5;0Em!`(JtLAs1{jsBs$$K}(
z0UN^CwPng2R8i26*jA-JoZL@udR-)sYDeo-$Rg(Rx3FbZ0%Q`c4=<2$b(jjg_XYp%
z<o=b&-Iy1ug^N9f4mSb$^fl5xR5bf>Og!3Z&2aM3>d(MVi)wCeHtUc71|FA|_IH1k
z6<E@!7eZIRGd68MY9%ui)rcx*y`?**yRNC>9Y_Y(2jhpn0h1CXjcJhsFP8dU_Ck$P
zQQ)fVGgkY{u3i@GIr^Jw@I&{TuI+B`t`Dw&imU?A0(6}QxTuI3xVS3cbsPX6#0kJf
zX#2&8ADYN*3wy=*d<Ot{V{z2Au*gBeBFzCJDOKR$GftD{{T-myF=zX#DC7rFD-fPf
zwe?Y1*1o&?v<M`y9<W2d++A&bmO){CgzU6_7~`x?W@$#VRf#{^_~!>DSLQdDOoF7^
zBB;h6cRRmwk#SPptyz6nda?Y{`uHI6Zdgpc#><pztIuL9fP3_QBF-+4YF<;qv#j(~
zR$^;?L#Yvt?7}OX8rxSF74ll?-Jyi7E?l&t)hMbIz4F04bb?*z+Zj*rskpe9-1ovz
zioQx@PvD^8xm4X|s#h=P9c;W|`>oF6Oa9^^?BJugp>x_V+=Mk0SAZQ^YgA0Z`!4;3
zC9uk%BHxJ=(Mei_t(JLU{r>OK%+Mbz6iyk3?kxiDhf7dfUp9}Ga1v;tYYvYA0b_g1
z5svWZO({pkUbTx7rpQ)1caa~@u+B}Z8pT#Ht)-H7uGD9&!ui63w0Oqsc8>**s=Eep
z^o2yK%S*|7TWX<+X)bezw_oeTC7=9~Zp^1=64qSAyqBDzBQkBy`1_!d=opTBX<y@_
zHqJC4kC`I22k)70QW+7Gvhf0I5y61C!mu5Q%Vt^u#2mV@X>eH=)7cf68_IlT27p_6
zs@H_7dW@x<6gZ1&rBf1OVv=8h4TZFznvG4ti>*!)dqx+sUsbkafDMrg{3987c{M-^
zQp>M=mw5`tq0&NTZ<dez+=kO%9iKzzaW6o-BMlN%(Z^4k{7nFIX(J~G<B;uN)y1$=
z23JpDKe^39AWhjc`m69qus@YRX<gc>nM`h0v+G^0;!cyU<NqV>t;3@JzII^@^eZTQ
zMF9l?gHGuVMMSy=hDN1Zy7^!L(n>cf%>Y9;42VdB#4vP73@I@T9cP2T=XuY6=RJQs
z*Wm&#E@ppX@4fa~_r2EIYxgwslBx5%ZDR|kclNW?BibUnXR(jO$DY-&KWz`ZwcXh{
z6}@`Ch8hXg_&dra6_z&*?81zDdm+uA?eDj6q%hSN?T6nz<pD_01VtZ@L2OjS;ddJO
zkAZBJIC^nh7q2Ppm{mE2myf!7WCnB*^BaaV+==NkbJ`544AxWfmLQzCs5~WAKyT=~
zZ(43Qp}XF2SSVmSn$gwO^*wByNB|P`xh0go>E1{UEFkz$IWjsqIUfRH;zh6{TF4k%
z3j&Z<U{|ntnaLhddGhrv)C99mBFP6%<2D#cT5EbV`}~Sj4sttr1sSKn*cHREvYN!t
znXnPHRbpF+M#ZHbUxgPdMyA?5^w%qMnWUl_b1ZVko@?w5rKFyE202-Q<JBRdP0={!
zso2a4Zl8|ZkW@{l3C~h)c4{O&2)!vp&N+(^!?$Bi=a-j^DYNenE6K5#;Wjno`vdS#
zb3paQjDB>fHHH&Z5q*Aga&iC^f(xLjz^=8=ylGo<PA!&<SMv~?j?owABhuvpNorQ$
z<RSCRbwD)@$kO@1_SS(b7Y5)XTZB3R><tuXyu19PCtbbtey-Ow_~3~i?q#lQ29HXX
zb?edaYtr`43d`ADHUor%JyN5><6?LJS~-K<N`vUG)vGgMZuB1zsmXpP{m8|+b6q}Y
zOl!fRg7XfSTus^TeU%t3D?$#%vs@gxejtc0tHa#*#Xe~uw)j17fLvqf3n1;{-}$W|
ze%&~Wi<M~&r(uwjJlQi2>ILOwi|H)^7^;~Gpa#)K{{4dkiYt?ZZ1G|F`WFr5s3g3r
zK#E!hPz$*H*6Xxl+MxI>8i<XNp(#PmwCizolgn3oehlkK$+BKI|KxY2mdUzP<XT5>
zdsNK_g;x!xY1we{cF5U^7Q63ta`CJ)9cjpxyeGS29H5*83grb;mR*vxgDjf$X(qOk
z@-Q}EQsEP7B$}~`V5ZQZd>4(FyK1f5`Vck^;*|5;YG>*+Z7{6&l#7(uo4#afa(lwb
z41^3pHCUbME(2Y-p+_eD637$8fU!_;R&Mes(PoEQCQpE{d1eW5u~z!QBbSE?5XgoD
z`ReNGQ51v%MU)Fc%^nLdFdZEo{Q$C{0S4S+zaT|Oi8?M?dFmyYgh>;T^gaCQ)H~4A
z?eU#M{aDJBK7$o!QsZTO;jC9;3yFHq^R(*Cg2#3Ohq}D;U?Gl?5jOq(?S(SIP%|wA
zr$D%keWWO}yd!9r49XTWFQvU17yzgy9blLav(j8Z{YES8(*PX4M-m9ULTFG-z%ikc
z{d;pW9%OF4hy+<4e4w(#ccsx)*IJPHm;*9>@SzkWpytHqIZs7Fo8kL`%@5>6Jhn1Y
zeRf73QjgyKOx-yw-(v-p=K)u~)vqs2{_!M!Z%ycH1YW~pekaw(X@AloddxQa%xV)y
zpIp<BmxOzv&W~`uudsA2mnhG3#!Zv(tarEPv2;1upS|dTUHxwb#6Q_Y*Q#nJh2@6)
zib~VwoR_6m?{TE@z&xt&Qi*Nr7>t^4__H0b7gjG&$@K$M2tAamSt+c&{qMi3b~qo9
z^SAu#>-!<35fWhIGzUFky!lPq{#A;9{6s=xzDBF;St@km0ZM(n0IYuK(*x;XaWmjk
z_m2ZP8d=~FA}f#s8XO+Z>9`8ACiq|f_~?k(eY&v-p9hf>T_Dk*))NBcbFP4xO6omY
zDFzrcJU21ndR1ADBwJCUW!Hl_xpiieFOt3%*TsE4d*n~o4!ZHnr+PW8=oO2psVR%0
z#h;^lxE1-ZDUY8H9>Pl^MRpawG#J^C6TxSon>9-fl+M^@q*v}1FCg@3Hw`6@Hk#)}
zj9$F>2-3;2fbGxqz%B43bz6;96hM>ivo~yqi=Ke2)#&YBM)S$J+1d0n6#<U*K!ZZM
zFi<Bcq1z0NfUAHa&nS?-kPeu}ZCZPtgl!=ARv^G@HBnOzQaWKESM~;t$QhUEe@`m0
z*y1&NmsY2#78tV$&y|8vfXP3Pfon6)VB#BLK7wS<DqSc(d|l`v!?6AuX$PygQHKk2
zmDZjP_A@=hT&z#{#cPE$0@8VDv&hD?xh6n@V^Y$8`qXO~)tyGi9vzgwMnO-Cr%?A5
zq;f7kO^;5CDqJm_brdmR^Rq%*XMK$b(@^YPsng$207aH{ej9^SWA~vk*Y!z*Z`>ei
zw*BK4q_POIbD16Wni$Wj-Lh?tW0Q7+*$O}5*hA`jTp%Ln<v}?B*o4;|BIj0C5P_h6
zT>g{&k>$rkmGBKt?II~q6rm-cHWG0Y#}KFg4Sd`*kxKx3T0Sc9#z5Q1+uEN@)Wx^J
z^+s@7_VfYX@=te^MHGoSTfQqf;q%M2kRZib$G+%)kRAG!j`Ysm&9)y3YkrP+GT_Q1
zycO;09al2&o18+&VB&D%8`I>T7G2M{?#%azU=xsI0r=kMd@bwGJ2!IfjO#WfrKA*q
z)kJ<G`uMSGgxJ={Gd>9XP?lhl<W<~kpE`GPR4xK$Nkqc<?}d$s1s_q!Vk9Filki&q
zr^J38UG`**H~bQQ!Dp1x%jrF5x9N?oI9r2YwhRtyu1Uf|k3_sBSKqgDxnwQ#4AWid
zxoN)$et_X&6=gH<J$cKM#MM>-3;6u$)0b<v7%hP90sgR5Zy%qSsHpTa$=?~X2Z)14
zRLiV%1XMMFgzO5YQ$*bgL3Em-+I2hxuv|U+`pj3&0hXYSfQaXI4#2alawQ`+nYy|p
z)4x>_wm#?}J9e5b&g+&6w$h1Zj54pv5jp$hG{NFOZ19@cG0wQHlzyz~DrssYQ}@TH
z_STGeQLi}L>t_S)2ai`|B)sBJ_J^W*=FK7+f*Waqk}#&l+9o^+Knr{UHO-lYg>`@e
zE2s6@R<n-_Dt*H?odW#eGxNC`*{tlVUB1zv6s!66cTjLJR&?M(PN*WI!RB^YVdpK<
z7l$SCZoc1?JO|L2?A}?yHXHFn(q+3-L8%L`o)q4%E_>zJpy6p-8EpyQ?iPoYt3j}(
zo~!2)@Geg+Jpg2r=YWEOdElVct<6of7*<U_i(mJT!Js9A3a|(_w#SthCjqA=H~a&V
z8;V;&_3GP(hPAE*XYY}*>|&P(lCHVtMKa7Q6p|E%nRcaOx6=yL)l)}F8x(DksJZ08
zs~-q0Hx5bL=8PjJ$C#jFXMI*VI{dcxl0Qv<79ZL|P4+mgw%P=LKFcrv1+L5iAoMvv
zjN0ug*|@pYoEEy&Kv_D|Z!)j^7ql9^0A%{!`S$JG7TQ@XmWhrI!v~Ind1Ed@5ThLt
zh;%6zf6*P4I?Cx&WF(TRchr>Y(H||{-^QWtU8t*n7b>tesz*E%cc6B&FL-ND;#AM6
zG|VMlX)v0g=KU*a<&&@Q1MwocU?t6RMt+x-)<>q-Hk{Ub&doJ}`$s(`+2kIG?6VWY
zzFDabNnmpdSbL6)bWvC8oNKN4km-c>-LOBst(j>N3QF$)cvY`LM%%@am@16%(r0U!
zAt0oum}`osm~)u5rrou2>OsvJ7sOIvwuDAU%`;1bJXo`e75PIMZTh5Ua!z}BC;z%g
zT_fsr@X`vtO&8|z|DMeq;4V}KnHONpH1=z`6vv{=$aa+VN-szu#%7c;;X;ky<pt!K
zqQaMVrV0Gf!R!NKcoD(AkNZg09jG>Q^s#wnxUI0ZNlOt!ql(f&`H*QdQhjabMyGe$
z-@A7ZU*Pg1VtV{g+b69?%_r?%e$>@)w)r+WQ-;h=N%j}7_#bH&SQe-k42x5jw!pdF
z@M>KKKGD7pWj1TWnE?TzfDqrUHPs4+@bfF?n~mashmV<#mKbuUgN`=88`hr2f{NeY
zX6ctE+@|m_IfBq06ZMz?wY>hvdE?ncxwM?9k5cp4c?Cy2sd=F4`FpAMd9Y9W0H2Jw
z<Gl-9e_xd#x@#n^|L^^Qqk^weYl9IulJZFY^RX05HTgz>zrV+S{-Z%473nO^WDh5}
zbCrYYKaCo{{NsKuegI$o?|+UjHxZvTk@E0xFjBIQdt<0GG6FzU$==pZjk@*kwTlGP
ztJ+O*VHy|iLs9n$sy-zvn=+a*KYAboTm?<&@3&zLFd!V2Z0Dey;L5W`VggF3@2Nrs
zDHYnHr{g(g#&YK46#5)6C9g_2&?N9~0en5|!TEbVet6N4fLXSp>)!Fnf%BP*Mmcc6
z#q^jUw<c}H@te}!tNmrZKY0q$iWLixK~Nvs*bXk`UbGv0<^1x1C*Wc8^!P<NA5rgj
z%Bv!*gopBin8*XosD~V<B9B#coz)K(a?aSKLlVhig^FyR)okJ<%f0if?_5dBzaS;t
zrqnxB<~8N!U%MuqnM;$QXH9;?je(`>;l#fTK9f)WL$~lnn;np%V(o_i-+oFzm?rdd
z+!(+au{s#l4GQa1jj0o{HC<%{a9*K1b5b+Lm0hQl?iA)I(uvYOf8I;&%XI|PuK+u+
z8Ura5d0I==jFHIPIi@7~<Lu8cL8+@+YrhI*k~#{M|F-u&Zx1P*Ot{PxiWE))!JfxH
znt;P2%eDrl#I*08H_NdBGu!xn%W133e1ro2Vyx0?<mxgNi;S+bkSnqzvR@p2|8GOs
zdBfk8bn3p!sn*8)2x_ikM!A&l^UgvpCe0EN=yR<}z#7Nn?RqEDq;w=YCT4z(r-Pk~
zt9Oh6b_Y_f7AP*kj?Ft4eMPr)&Ahp7*i(s3+gy*74-#FfQ5~2%e)sK&UM{yaaxWEs
zF>nBn#}=ci$i+zMG1ixR14Hzhdi}v!4hgW}ljR(5<*~~#82>HP=vK_DO16k@2e$HY
zT-~YUakW+Hh)LyGXKe;qXK!y@6qZRf_vS{d8U);4j*gCX2G0}bvDCAanZ)fgMV&J_
zLXsb53`lAXNMcU*KPCAryEHAo6+Fq-BEXuN;>1^E#CDjcorR)}0wj;VY!zd%UbUBD
z_m7g9jXa+3`8tW1x(S(J&7Wn)x~1-(W?#(_NK#_dXxW$kGGWIwO~@*IagvAV`gNFz
z36xRU$z1A}A!Q$@`buaX=Bib}u7&XK_EymEvhfg@$5u~1(=PA@Fffz$(KjuAjDHlZ
zduf~r+dba<e(NK|9D;)I$T=()J1;-2<XT(5QT=iUAA^34ca+)4^7d-5sTSD3db#M`
z?!&m0D1)ltxU{}`w#WAy#~6~~TCNrB5Jd=kLtR`>R~OdfG%Y4ZbB|>Tq}>tWMKjXU
z(!9MFe_ex_c?VwdyQ*n>d%mWK!@zMjVs4B_-&i2r(dn-csOsF@xafD)|G5lvo~Ag9
z1bj3`ZRc(Hyu2vyZCze<{zS~C{@ZhL`JxU)oPc!Yjj<rcpDQBT=M%(5TKkjAI|26g
z!0lX@_yzm>Q4)yd4-|-#8`cH`*l@7qzI0XFQT>2d?ZV8HWltl?Udn&YZ^wEAEcK1Y
zrD(WdXswEb8J*?t3zkowwyPK>(1(q&B`8wu4L#oXhaH8<oVWMe{deP{lyDBFI=9ek
zmKH<1vJS1f_*Vo~l%{=3WKt5jA2z};FEVJqtz<QneN32l4*Cf0#2_n){oLxRkXe-q
zIggVa0f^Gml9b#}?4IPjYG~3~$|A{SbT$i2rM10*B!X5=)?v<v!<Erul23HwnO&|s
z9mBhOUAp9w%D9zulOE2C++JMI&Q&>)6@+#c8P}X}BuyNEN=EAU66)8jx_ZqxHDP|U
zAzGH_gQu1DU2AQstBL-({z-0I78P%booicq^CI%%rAzhtHwXx-hQa7d3D=Zpl({M&
z!`8dvGf$7VLwK~ZUm!Gs?@>;hql;D!m-ff(HN!r>;ervJ4@KxlFv=Y|CeVwIKPLqh
zjsG-gC<xm2)l7Ifug=K{Co7}#$#QqXsPa?-Nr*O#P4Ayy#Y_VRyh`ebox)&*MHlWG
zyyMDp{=t{Xmr?xZe-RKoW(JM^e*K^Sg#K?D{|~f<&R+EY{F?s*PyYL0=>KA@{wH4I
ztpEQHyMGb-|1T>46Wu2O{r`X2eZzD4`+qnG(EnifjqE9><ycr)MynjsLG|}+P_3B_
za5k_h<IJUTu)B_o<$&}&YoDDh1@N^IFMn$zD>HKmc-lA}eV#sjN5QF|1NI^1fh|l_
z;+|z)W~G#@m;W{<=*{E1fCL4KK`ryaS~q)syK!x>g#ysA&)sGNfOUa?2o)IC>z>7U
z|5VMA*vFlNQgF2$f=}FC`svf{F5X@!2i!XkIBdba6&a&|w_&hAPp}^B0gTK5h3nev
zD!r<8|MSE*Mw!>#d$U0-F_J!eAQ!~LxZr=08v((eT8oQ@fz4{}BfY)7Au<2C!p|I*
zNLz<3HEt!5QwRGP0CRoQV{9vs1cINz_1#O(6wt}(_OD@!WJk~14@Au~dsyp^Sw{z_
z)SY>^;Pt2wgP9fa-uEG+OGZA~QP}2|!fBVLnKvESZ#s-`aWgI#BIzd;V4hg7({-`6
z?q@Qc#LCDUQuEGgm+JuH0E9W}ENZY^*NW7yRtE<N%?U?<YKAc2o0nb2T|wpB*h6uu
zGjk(09i<+%M6lZx<NEe-enbzGFbc`9dwb!oz73U~QiPw9qGs2~RNrnmU5tzdfn9*N
z?Bm`#M(YLdo|@}DUO2N|t1_2ijxoM#Z<}X+QQOyjotrZ;aGP+k>igzjsqkM!BO+&z
z%{@gq8+h%qLCJE!kRc`Du7ZMkh>3|cfcLc5(+fmLz+8X;S+;!rwD+F=%w6EkRFsh5
z@u>axQd{L%xz(XE$3%&YLl9F;Y)ZlM%#4*jul%v73oJcRZIZLN>z=attF(YEcZm#h
z)0GH3_3c`7<nO4c`8Pth$?tx3y+NI4lx-n%`Kp)u^QvaZVYzzR_;<|AYxv$xg{UBG
zvrG?5`_h{YuiHI)DQ~ft+_@Yagr*Lm+6luw(f1=Zg{sIK;9NDHP0{PBxtgPXm?h`s
z`rXNat@%!LrUIoneo%wK_en^2_z-YgIu1D)zkYq@uK@XaDzV;AMvjl`4W9bXL~D%b
zoKd1Jh(D>`K-oXWxO54UJpNXh-3xnaWmViNhF`kvV%evUB(2q}7cE0IlGpUcnrrx`
zuFPVzLS)QH#p%KsQwOaF8qioAB{LUlFXDNEkGHb|GL_0{Hzk0ioh@oy!&YQC9{1a<
za*Q~QxS_#JheL}Ce%ZdGa?G@|J0=a4ruYmJrYv7gPX3-6dvavx)7RP26APkw><uwH
zYy9*Hy|@!eHZ}#2I3JnO*w}c63ei>ts&9D-dJ)l*Z;8H70Izor*!B%pCm)xYAGIru
za9)$KAv(NS)+BPU*4N;M#kiEx!|&u)<%{lG6+^9QzesC$tnlF}HJo{2!UP1+eFC{>
zPheXKSpNLem;;Ns<}Kb0c{i2#Qg8o&dS1f>VPT?TT^D1-&t(Ht;c`Gg8$Ao>BMk)-
znm(1<-K8Y|5Ps_=N2PoGPn!>nme%_JV#o`y8o%A`H-aJ}x*&n0u2pO10%bbhYiGG5
zJ5590EXxy(On30{%u_0dSY^U%i#=}D_m?8f{tOpnQ*sUrPElH{7aJeU<Kc7m^+NDE
zr9@GcbjA}&x&F2u8onCq(1(|T3%9D%U5GRB<(5ZN#hzYhUGa^{t0Ix{Ij55jWB#0f
zdDI=nOx>YnXAwNmXjWjun$=W8y`lG}%y)yv8?8|qe2bY7JSH+B_j8KqA~wmMSzw%F
z+L&i4HnzKL>yO@Gp{FLtM41e!&xYEJ>>Vyf9Omb8u^e4zug%c*6ZlGsgm~}`H{uVv
z1)jOBnTc*VDSzz(PjJF#VuSk<Vg8+SC3Hs(C@tde6JYmK9*E_D<bdlloYPrMlBxNR
zf`V$GOjF+aiR{2GhE-z&QF?93>mbLW>B+%zg#Up{_o!oG+3WU>ZuIebJy+mB?x_J<
z_EM4n=Vo>N3I<Jy!N<w<H`DoMWNdG5vEBarJi|=OzBgcZ7Kg#I?pg&nK=8(0p1J&1
z-@Tc@JiW?4{;W41XK);nE{@5BXfoV~q;7!|EAU!8<?r6(Psrm|hUk3za2gH2{Nj5k
zyL$>w50}haQWX2~O4Lg`LmcuxC+IM03ya#J#2QP)WCFK>@jA5@iRQV2@$2}~pFb_0
zGO!dqVJEo)QO>ruyzM_Y=#`1gXsvz5Ru(Gmd{w|Y-xZCp!kt_iEwe{4L7j+S4f-3|
z^h>A-E*9LKUbmHpQ5ncnZThQ7o%0S&v$J*+O@ee^i$CMllbn}w_<Z-h!sspE;22G5
zo9Qi8ag3wR57lrdh6Oue)E+rC!7I~2w?;juza%~nK-iY&nbKw}zbL914DO$D*c`2L
zOZ%1-UDy5Ba8yC=unzJ%g7MimBIPTKYBE>$5Nhu6dd3FoZ1Rz1hcA+u<PjULbi%B1
z!MS(mceDv3CJHr-|LW5tYT$2r7&!oH+f}l$GXGjDwo0aYBq2PqRvbF*?ift(u~<wT
zNl&97@94#T-WOG2z!jJu{_0hUj82GK-_cHhD#YZ}KcUjbjPE}A0yP=OVZlE=1mx7;
zNcoNZ4NUoU1o=@)-Mrwd1(fEB_VcU~+yt@5Zf1sDq?Kmw@1TCxy3y=E*oJ6WW$UQX
z-mpxK_+Q?XlMI{#w-K^q0wp7Bx{iw<^=6M6*8S*n&1(f{bMk&$PgQ+}-d#E>UwxCT
zhj3q0&R$rTq%Vz#sHw18Y9cyc%2K9J8;QSYU*f8TMQ)$I7L4-<-3wJbF0w`f5`teD
z?9<59wD6?}x$@%&Gn1r|k=>->f`i%ma6#0_$Z}sr*u!`4#kC)tcRok)>~&UR4aGH)
zp8UrZ>we&IkaiZgq4q&s2@Z$<<$_^e9SnFy`99nJ?n60G<9RCvGt;<(BfcahE)G2g
zInSM;Br#p;zjxC=yIbd!6*_5miVb8uWKWfgvSh-rr(CX44U;7X=<naRR|^E=WP`}H
zoTGl$9CkFww{;A0NFIGl_Vx<u?RFA!Uay1pN7Ewn5Vb!RgTwpgwg0lI$0C3gn7iQS
z#rdp$;pKbU(endlR((y)SaGJQnO_PDr~R`K`{x!+z7CfJGVj$_$}gkArr#ft=K9P&
z*L9#*hWm=@$qw-LcqzZuswhm_PhVIYK{A$GE#tCJoi4DRN8|{eKW`+ShLR_JrzOIw
zPIMVeu@;Tnbj~ZJD2x4>^bR+n9wAN!LHIz$AJy$>OHXUGuK(rB7CQ=DIeJN2hI!+)
z?~ab~xD_shLdb%y(`wbI_d(1_t#hU;0|TY7D<q#csa^fNK;l?Egp^io<q96ZGK{%>
zm+ZHH$6qD=dC#*J;BWFI<VZg0^P{xax_kqR?Hc&z!uo?(7(@Sk9rS&6#RtAx3?`<Q
z_KsGs7RS<Usg<bspK`3A@s9XOa`KyB{nta$A!$%%gzu5udcJP9n^u8F!SqY?wHvW=
z>iW;KNU}2(BD*9csC>H#>hX>BMA!b=u-iPJ!Nz~V3_{UuDL#i5s+FDF`$v-7vmK~1
z8ytnL*k@IDub5x|8cZ-euiRZKU=wlu)NZo%x|}=^h0d!CoXWl_sGr@AI?AesjtdH|
zlCW~lrp4aqw!#dR$GP_Tz<1#|5tlk`j`JryrULWHKry$DyooVFfFHKn=C2<HVVS08
zQ10_*RZlNny5`dR)fd0)>9RO+_S4bbIZ&KIFMLlK9U2PrOGA&%pAQ<r`-jtns*GlI
zbTTr}m{}iwAIrQGp4Rg13cr<m?Rn&cK=m4Max@Z7gZwbxUc=*U%#?1eS7kT@L+8w!
z95a&)5k9*NYp%6#3c3}aND<JNkQbdD{a2lQ2UkryUAE`sOy}o1Fg98is$X6w@(;8<
z!64iYT_-t9pUNvL4fI0DA-ZB}=ZzoNGBeDXdH9}zGTN%B9vc2H3g%~+s#fu=$9~jc
zXul*E43{YNDyDwZ9CN!}Z9C9#eBiStQqS*`H7aPAQM^l{iRyu9>Qzl!j5~WV?`I_k
z1&Jv;eXTB`E>?SZbab`%jia<I{}>@)l_0}%XU=$e;A?$cUV(NbFVFTuCGU1$gd~d~
z7qV@wv(g5QfjD8GP$*i|<`r>=fiRoZbI;=ynKO6p4*=52?HdL<jg+J4%X7$b$Mgi@
z>jy5HToB(gMmbqtp7QNqH+^btvjtWsvJq*JTEvZpC-#$)iVbF^ZtH%OG-AF_<3ha^
zgK{k+GkkBXto)=<x~;)0ulD3^M_1>-Ni^Sa!>M1xXq|N6EwaJW5z6f}>Z?lySO1*J
zX}QC?#0@H39!biJPbefZpu#%SI*Csmk@uMhB=%^iyM|ZLCl8KK9P&9Ga@Gs1iFWIk
z=ue@JjQ9~6V`KB@mO>urV1kbJ^eBm;jB4C%i%qnb4;RM3R<Gj-NpT(h{vx?SdgAkz
zXVydE6I8;`=gBnOdobrqE9Pl6%z}lYvC$bryQgW6HUj)1p}(@>Mqd@IWKtXyArn6o
zKvg;B^`lf1<sp-X<y!VgU4)x?x0>Vg00|H8=(xPy!r=+8!NYd*n1rZw>O}q_FZedB
z04jg^aeEgY_3_Qbul!=eRo|fJJw_RXhore<=_(c@hqHs<>}a?&&y!1;iRtVq2?%2P
zvcZaxbHe7zOZibuIYb3iK-QY{<KiiD{Yo5(R{I8m2w*C+xD4|IQAr*%M1}VO_OA(b
zR}>aVszK*wX1Yzl=*N#MUm|3X+pBr@S4cjMnDh@$Em(ITQAS^a*}`wJuuzFD@S85o
zEiNgDu8dI^a+3Wy(R*K{di;3K^G1XnVlMRh4bLU=OfxgXFToei`FE-m5E9zh_{6Mi
z$$dq$D7gXqJ6`}wh}|1!yg8HTGh0ryfkz{Zp!7?v5pl+u<h=?;%4h47{~uy>*Aoo$
zcrWq?mc(xInLU2#L3W1h>2A>0DY~+<gc>)`M<b^=*v*l&G@ojq*<U^dZa<5x(m$p~
zWB#N&K%VA5^BoIrovoT_opa~>dUs;__NvXqSVMUYhQmi?U<Zv@>WjA+*}~xu%GNu+
zeh&soL>5P@>kM+U2RkqmKPb1=y~<eBB{Z*abzDCzRSZ4Aprw5^oDO4-Iw@3@MEo@B
zEB!HJccHXBBjNxbi?$g83!Z(|=lYrdgCJ)Qtk%Z&Aa4YWJgc1NL4S`aEw-4muOzWM
zvD7<rYYQi?Y3RR$TjuX59vQ5$A~zVwc$M&TgFAtUZuhQ2tY=9bk!{abOo`~d8k7&U
z=CyyA2ky8?OFMh5EOb<i@U5?-X$#LoaH@vqk8-^y@$Eg6GIPyyh#!691`AGT#@BaQ
zr?Ntwm$%%rm6KFU+d*mHrGb9wAS+D1{vzHPE|3q*d+BdwMgu-uDqo{g<;ay{r`E=-
zFa|kPfzFDn(%A_55x+lk9xqDOA<XS+m(k<2h%avIC$pP&5vG?3j|0YHznMEYykn=}
zzeXB2ur%0E1sSI@7kp=PpR7I{CYQke<|>@*J-gFOle4*a5@cC8nW=e6VM=mmD(^~-
zgjL!nM~7``rS7h1dQA@FkhcTmVy?9jKdZ%KZ_;OFwohU5?}xM$I1tfRMr{LEH~4Da
zz`I@`5W_x`4}R^o?%=~WVF&MSdr@vVEOjH@3UYT&{SXaP(yMw=8-7pC(KBEfhRBLv
z=;o`E72H3#v%db^wbnqx*XWB89N9PX;TfTBft7l8EG@Oz3tMUdX0-Orrl!b@=zDs_
zm96F`R>zx{Tg=TmW}I1BnP05s!8s~Yq~(<ht-C%=tD|4SnzG+bM33uNwz9gfE_=LW
zD2Hx-xJtP6(k#nyYi(<nKh7ihT`R}YLyw*_b0cL67RH@qFi@+eY}+Tvblbb(DUKp3
zGU1k*sMG)>Ua4l}-TMgp5!4zg{xmQ(?Z7{ri2<?59y!t~+owde(rKOwD-|j46zlab
zS=X++BDQO^79y%OcalF`VPMpCs?h+MmHt^Blckf~`Pj99yK3%&sFwKMFl(Yd<3f_H
zF;KbIpD`S!>y*D2(NKvn)NB?lLyNEJwmY)YPkHd{hjO#kj%p&T#+sx@9eCOc3`+!R
z4ZJke>E*gthBjdy<hm2f=rd!rrYNuOCHQph9d@l{EBQ>L`gQpEQ9dq$fEX@TV5a()
zX+#<10P|Va05sD-pbgWSQ5CAlA?ghqBcbg#26{Je>NLkPJxxmyXO@|WHPKD12|tL~
zp>e{$yHTP&Soy?&Kj<I$$23hQp3>XY&{M?h?{Zg+U3tUlsq?z#@%E<iBXQp}anoVZ
zSv;N&qPS8ErUwl%lz+QTelH{W$eZSc+(6n@QZ^P**|c7xFZL9i{omZC`WZM9)11Zy
zZcyjwYR8|5cq+<`2+Mc0_oX;N0|B$wA~QxGG@Wx}|0|vJQQ;Zp7E3efXN)AG%P4~s
zWA_V#G0HUieDeS|{a|0OV0kppbYJ2Q`HzBnPphUHNc*^C*Dtl9%qgK&$CqCDZiSfU
z4PuuSQmW_uOf~hf_9J;lJG3CR;@G3tS5AFuktCHB1RV$nRC_qX-5R7A%0M3tcE8_Y
z$jHhwz~_0cZFM@!y!YJVP6)n74Tgb>^H>48CzN5RB9_jT>ibmM)0L}FOq7I9EAD0l
z3mYT|2gVCOWf!;jr-EZ1-eonEz67?RBl!<(X0?2FR@O)BKz>~W9a9?7P{mi=N7?dM
zy@qoApD2{@uXhy&*?GRTIk;wPIJy4mJHy3*g}6Yga!Mz?#x>bSQ>EdyJDYu9{=_Xs
zr9tQc;yBC6)MNSfZcS_=|F%_c;WDI&Dm*zRca+zqp+r)~<`0%M%c8AE6(@%B{CR9V
zs0P<~2oLr>HYVZvZu)()qz#f!5aF}~dFidrmTQToROUy^+H(DV@g;#jo;w-u+Ih|+
z?6)@$x~b6B(g~iMjPrAgA>Cv1fYcup#>GmXI|lu>REyOq2^buO9;iW=jJ$QO<>uC)
zq^UVjD`t3WZPf4^c%;a4O^k{l>7ZVHEgP4lksFF%8Bwt=v@Uj6Sh&V!@K#4hw<czk
zzuazk&GU<N((7h}_Hq%Ir>Mz;rL4C^INZQ8O|5h1ihYfXk+^Le8`8Qtwf;<<dn%#6
zqf1)2lN4)mWi$$CA<v{M0=DeHoq0IrWt6l_CxpGHGR5-@;GQhov0R`l!j{LMgsk8B
zHreu_Hqs>0S+0P&8)aK$xa^QyvbDu;I-n)jPwY0nJV_()uMr*H#AJ3H{nDmnEHl%T
z;WBr_R5_e@hc2?kX6VCbQm=|eqI{S42UO$rNxyk{@T0*HI#XhbWvRM_Ejlp?@d}Qw
zweX}oTc3Y*PbknmMQ)P=n{&5s*ka^w0pnGEx&D<>ByCvG(3p>BTWlIq&&W$-q`Jla
zvextg<xge~Hib0=e4sArYM$;(sUWM>u`h)Slg{<*W^_I%?@V9)eRxn0oBbJNj9}A;
z%Do*7pot-{rIv!|vL465LcoRPshkF23_Zn828AP#M)fs$fr0Hy&Q>N^63<$d2r^p<
z^T;zV(Af}xcHiH<8_e^qCqrxNC<1cQbm2r75_{4N9ZP$V$OiQq0BZE-&;!`A_M|}O
z;VD#INB*bDD*GIO`uTk!m;8p|x&j#1x5}(uVyaB(j5hWgxJ9eHJzHgdjvyW3PS?6M
z#&Ywgk`>k6Lej!j&DG#vPIvoq>kBRWRb|O;@jLucF&DQ3E)CrZQv|z&#KBI$h0K$-
zU7%aG9U}@OGXf=(wL4flTR*`K<23r;_u2C2=*+68SPLXT+xz9m<aei$p{q9aqE0<J
z5h?XH{=b!L1Zt_q5PG>QphjLOClW53n~-p@TDcEqtL>;ji-T%67O<+sVxPFel+E<;
zEc0aCJ>VchR@h+3r@FMX3>!G@r?ikH^(z-O^&R+fH8qSPeB~t=qiB!tD)mJsnSgNj
zdMF1#94Zl<+WF=b9C3ioCU8k)>uNsaQf)cwJb}vnX~OGf{o`p2a$GK`?fe3=o)-G0
z@=h*M{vk2plpe6{Xs0x{F!nOdDA?+wonYltW$cB%PSU?Q7fn|FfmsFk*rMYT86hJh
zaY!eQC+z0TI=pHhlQdwSjIWZD1Uo*U`f*%8SpmS55ERBKaAve3DqB0ItokCQWe1m|
zxbSI`u~sdgh2}>MY^jqYfAjG0xeREzoLU9Y?#+|VkFG7#4agBsdV|f^jUk19$kOQy
zJ=7jtxIneLD^k9{%~pTtmhYOX7<)XpTQjjV82cfL>JUR!zBfngQ|DWn2Ym=wtSm<6
zdL<1RHk>mSzy8-z{Ebp2j!xxonx7NuanuxArt?G$l*nfQr2w}%x@J2GNv%=wvpaTm
z6pS&*RaV-mEqEOHbW`<z%Vo@;__B*V`m&x1Rdg9wujHyoM9~AKq5`yRr&y0|LE_4Y
zOlM4F>Vw(|P8{kfw3D&ze8$r>FI{prZ9XtfWy&(U!5?-VQ}I5q4A^WcbSNuM`}ttH
zt@YZmW8Rf84;gFM)ibM{c)M0DlFg#Gqm%2p7Xu4$Dj~1eD*)ouq{jfi19uDRo1&{n
znlq`O0gYSQXJFetiX`VrQY6CdY5+)bFZ=ivHWo4z$N%(^)<k)_tdpM`7G~ayhK*9L
zv=ZO$B`F<K4RH-DPYmGm;IhS=Ol7{klMgQ-n`cx#wPN2DP`Cs;a&0eZBE!$Rt#r6k
z%O75=W$Q*bRKJXIuui6W^43a|Uww=$Y8$gdBcml|Hv*};21>_a3OH~&0`+XUh7f+)
z!c9FG&_1s|!n#P<qlfxzlI|)iyQ}DxDMd`bQc_H8OzvE=eu5(7z7D&UIwH$%Z9+k1
zf+s@}k;#TIbFs4SiKhfFX4tLM@DztBz|7oYijGeG3IY>(behs2FfkY=UX$zczU1`5
zx!-Qu=W?Cj+aD=nEOL1?5ZNkl>KWftkmqz9>(jsu5|PA4Bt0-~5^KMU{gbXcsm{a0
z;PIz;EH*v^Rk_Yiwn3{PzqKe<$Q|Mv#>oie)p{tT*Q55jbg*5&_L=yizq)zVp8H~@
zgkM$PdUHWIxu<lej^o-uZV@`SBtq@W%MZV|^ys1@TComrsG)ggk|6s}HZ>{q)hTSV
z7t?kLcRO_H=N*uxcidW-;pFkViQ}zzpmaE~a~f+BF`oat6Z%5=(}W#;KreS|m%wmZ
z1w%kyoW+E7KyZTp(IaQe&e)oh5(U1ElERXM*j2q^vkYsu!}O&;$N-Kz=`Xu@u|c1J
zgT@oex~H5CoPWd%LH}@01P=*!!WWN)B`um$eE@Sf+hrjcJ5^$6Fd_K`MO(MCOcp^&
zdpjee%$a6$+SpC5b3w)~B>3YedKx)}-C;3)#O-6M6s3mduYy@V*-HwsR;R`8^YhqF
zmiAVYGnXc(E&6Higl|Yqr7Of+0B9}5=d{GrZB%I(7@npI81MJ@?_o$!7lRRQU4T^J
z0PETHmRcQ>)Jv?17K~RIERE<m3MKQ%ZQ=2n@2U9tGR)nMcYDU<k9?fD_q`5Dx(zFy
z62Hc*{0h1Z7$)mOp{gR8W+m}jgml!;NzT7C@5<zjOY`?z%?6UDx}jfQr44W`yFz*l
zYggXe3mv7RWnhofPAeUzGK2!5*3De+u&#PzY2>(WRsg9dd!mKVq7}|c_O3nh>6{F7
z+iYr%%m{D!APQnqMAxpvbhToKTIk@R)W=4YDT&$Avyrw$m5yVs6DL3QL?S<FyQiR6
z)7$){G8g7p$IJ!}Kw&lh*Wc-ncX7xwgvy_DXo{t&v_`~^y0IGq$KBAQl+n1go)^Kr
zZG$^eay)X&qa}-1oNcN3KLL!dKtdNYvBd5~XPABBvHDHnPGeZwalPTu5u53ljXws2
zBZ$iOSh?BsEOm4eQ36^ZfRU%veC5Em3qn86d!B1O2HkQ~RDre<7Ea3ZHbf93`8m_~
z8O*U3z*%(WpPx+Om%UWuJS5t@4iBlxT=(%7dhq1Ep{M3)T77M3_|`JjWAEM{6Jhi}
z`UozNjZYZpeX;lNkBz0$3?2#tN644C>y1x{Tadl@RaLuUGIz**jqX!BeZpl+VT|r^
z^Xj7rND01^Y@Sw>vnYy4$RxX_Y?gI9u9M|cdsRyG4=jYl)^_W~A8u>Z30iumwC;1&
zP|)>?qK=zRLgM}$B^5UuP?^R>|4VgO#=_*bQQ=fuPGQQv)d5b$B*py6zn9OyKb&6t
zJWc-LN^EjW_ppE3o0$cU>lx~XHe<opx{sj6#pQQTS{PM5OnFFJe&Wovg#wLl#cwZB
z)=^0$+nd$-HN<ugD29bDVmSQ_ny@3jqQODkE_yU~KPSk%(7lqX9FDdt8tg_9di5BY
zDDB-yxmCyDD4b;f?Bd4H_&)xhr$sKJrI22@dgk}Ou4v5Y)yLm^lNeo>$y-?OOK_z@
z6KdS-^>>~-+tJ>V*bmi7m)%+mUD7=~0iDPf^^SgDyqQ5kF=JzwHNagEQgzKQ%wO1(
zsxT(3T6jWl--&&d=<wU?yUQz@3|Rs_Oxa{HQL+juq+%&$Os`_(8Je*iCZ~ihJY2t8
zg5pLiUVo`%BFSPTV1}_i9pD-E<c&0fIE@~NF~(B#_P!jb6TZH72e#s|hH?+emN<;8
z`}UeDAql7Zk~-kfW7gQ%2TRp5w@}Q+6F|omP_la@qilG?yxVL2ah+4EZiLgKBl9CH
zM@D|#y<0v$hE|wFZY3W}?5aDHWvZt7_mf2--y78(`*W~`eC?K|Pydp5pIqx_g4~<S
zZL@o~y5LjlFx}w57zSj8ee+`@F0q1wKl<rytNO0v;bu#jsud{`oyEV{j&Oa4+t!#J
zH5H@(Oz-6<Iq6KhTrZ!v6F3qPFCF$A>Q{M$+@~}zR8H$4J@)GBp1><RQZ{z9eJWP+
z84-VVd#p9AacDR`bIfh4ABv>2&hHa-`tgQ<B84amR+xjkr`ecnY(D4{cF!T(Z1uRo
zeMXp`>>3mx^F~y(vS*gyY`Uz}ZgN&sI9p_i-#s2H6t6p8;fZD?=PI}E##3f*DrfKT
z6}<TK&Xc_CtV3ni(u+71UX&M;YSXS>1#ZW_vutPG#5gRyoE$QBB&17|Kn`r0#6@r6
zt3wF7Qnp~_4!p7?mtLN<mqxFsvT5b2b7QLxsOm)=-NQC7Nyc}c5XU|4A28nk<eQu<
z@|NYKd!63x!ZtnGsHo_cTrcN%{N{COZ!wm*&b#8!&FqH{t@x46559)R*M0(XI=T8Q
z{3B)dmRX?do)$#!1`ZakD=3I3ztf1O%-XWcmLgJ+zgxO-YbP0{ig4f@uW9)ujALfw
z%0|JB?Yi6pdc9m2+FF)U$@^Rx8O6{gcU!YV&F72W#xJfov2PRAo1x$chm#Q3<Qry3
zt!Zj7-Ht$mcU4>8ZyOR{Z$vz-CSJClGq>&@qZzx&?E_IE39o$8gZ(Iqpm@W++L4~z
zaQ~lof4{$6I$yWnKz+`G7Qu04d9yE-SFbWfG!1l={B*rkFD}GydokQBD_`u*M<chj
zKB_6C0fEohm49QRTDy*Am+KNRk+`@KN}7<s^haEz$c6x0;)tKyHUmV}>g{uDUd+)k
z7|1eN<C+}1Sf;N(C?jER6;@AmtJ7NYIGAX!4i1mqUI|Kg`*uhmz|+(Fm+(jaM^`Rv
zG{>&3ig_tpCv_8JA6_{5huHiWuU9TXp^B+UDKVz2&NPj`L$x)Ua})hDJTC`vFZFYa
zsI2mk<vl+Yl{75uDVe7_OE1YPlHav25d{fRQ3V7wNCc)*HQzp6;J52ocxm<hCwH@Z
zg^L0ZsrQp<3ce1-Ljs#icNGujh;Pd?NRljeW9w^Pt4nw58T*J+;1|2r-yePjqqMb;
zd6t=BVXMI!)wiBPs>U31Ha&W+B(3dTFl*{-k8@>ig^P)kI7wgXd8Us&o@XnADu7On
ziT_c?g?f^yUi^BIV8LbO!{=DRiv)@{w)bmP+$eTe-s*pS`MotQDn22%JnX-xXT$yG
zHNz(s_iNwvmNt_@k^fbFVEwy{w@tpbxVLJ`@~o$xUa`ZrY6jUgCwM+%cF{ptBUnsq
zlODvBdU``L<Iu~WlhV;FR)+0}iw#5#ZGo*S#wITt531>2+F0b2cZd7qiUkCj#2$q-
ztyD(pYQqq-U_v_=yK6tO#VD?#-_N9!p%PEGu|{8EtO?6g$$}zf2=twppZvJl^g<f7
z3j?S_Lu(rw`Q?-kd$?NClI3M^I^<`rDDGD)cHyu==m$SPPexHj+2|fBoK}`G?$PHL
zd-yH#Ng?MtEjEuBR&rvME4ZZu+zSRWF!n~rrNyIrzx#P4PR%b7bOvlAG+1tk2YsNs
zvi&VY?~%MIPwJ^_(9m8r8+v!Xb4{05yb(1tOQ2|tP8Jj`dxWZr7lsP^sYH$2BW}mU
zlFj{wesH}%_+k{OPHdZUPg~gLK7m4o4T9V9*a&yBMSeRrdJ8BMMLQjrOH#ImmY9I0
z3nwO&w~Jn)N3Zfp|CfckJoDt=^QoXi9#Z>pdAc-ryzu%BiwX}*P8iA)u9lHGs<I6n
z8h-9tv3le#(IX&`-{6^|%y#+;3$0n)rs@#Q@pEd=y?$g)j8jf-8E*W9lJJgbWc20l
zX&)T^I8E2k+%+6e%wkz)v`MMJ<*%mCiMd0U6xSQ#Mau0tpiE@KQ?p-bGI4d?islhL
zPq4MaW;`|3hRk{)f1w0=E=755v#5lgmn#i@mDD;EoYv!AqU95oI5Kh{4zVT_?>?Tb
z{KF{rNWqvo;c1i6O~HSTa9lgjS{{DRL276{dC<ADig1-EjXk;8$Xk$TZ2h_m8HfDk
zg7@5!ZT*_RK$W8S;@fVGlDF^9#f5+?Zjm`@n)Gk@Jyt4BqdVufcix!mmnI^y$H3@Q
z?5<sC+~GB6ske5jsFh*N-heQljfWHBi;85iA`zE1)G#DM2?-@jKUfM|TCGDrv&9@a
zZ3_{E9O82v@VSi5-t~9p)`IlEPL~I-Y|vK$T@hZ|#K$9uZRVtLUN%+(T$-A?_+a|g
z>w$hM_kjN>vitvhfB#TG=hlZS|Hh^E%#VF`*VE(f>ea*gPD1}!b#|~}hCP?B<H`oz
z3IEx^KX6}iQ`|qYR`!>D?7L#Os;IcQHWN^rNq9w_MUK!vS@Fd~2^4`JV_{+b;fBHJ
z!NH)Xazi2Kj{-fJO2Ns6=T9G&lBT`v*&NumTMds0BaaOT$e@gSBg`N6sq8udSc%B%
zbRE0AMtQ}jElE)2#m8urFwo@u&#cWdvRi@B;!gTBjvBnU1`c)$$ehF<yiE#{D|0Qw
z9-T7cmQNlX-K1#h8S7iEqfycjr553LY#Ql`CaX@as=FBe0~{g$d7H0)V0LcsS~SaT
zXQR@$1=+nD3&U!fKUqGvx3RH4Re3m5YG~kra{5=pZ55h*Do^!DaIF!5qsg#|H>G^C
z?>=OFf2#b0=`-w$VfjS=WH`{eGsea(qZ`z>+f9CCX3oycX`AL7kTb9tk`nz|s=IJm
zNch)Sb4t3ywWDG;NnZZJqY2*yc5osWcC@UqBtUZt&CUDxrKc2|gs@LIKpXv_vI8r(
zH3{wWuMn(%1TBLNJC-`Ww-yp^YSfbDr1GvHdyqfL$Y#aJcqw#w_zv|ZN{d6=@xccg
z^B!(%XT0f)zsM5&{?HouVuR@EB_XQ3$#{#Rh-_sxRW|<e(w5HAU<7xu$=v+ROVdpL
z$?nP2^eOMA!f71)NJ*H+%uSmbhgx}kJ;;d8@SaHg>5av}w<PU!)=s?N<4!vQW|=(P
zM(F5G&5F51D&&(JQfX<D_)YO%rTmQa{K5|NtOU~Db<1MN+8g-I-^c0MRb(12h6I6J
zfs?#F{cnBZ2g)zK$Jo!BY4xO>{xQ$VZY8!o(ivB&@7K&#p8*DD#2UlG%F)s_4DF@v
z(N)-7{Zgv;$R;eTm|tMw$KW-a>bvbKao_(@mhaO(6>Mt>q8zLEog3Tz$A*4r3c)>Q
zC$qU!1MM9b=<NPri;?elbbQ>0$%Y&6iX`Xe1?;&y@^QXZVp6eG+@QU74e=2%#ZUEn
zww)=v*J00*nNB%Xdme{xRpx6G0<J&mSFbEyX!%O;RNs@Pxs`g#7%077k8<H3#Dp+V
zJ{7u0#!6QA-|Og<?DQDrNjUT1MVl&8-`2`n1L;dIbxY6swIl9DM_WF36Y+9lU9+g0
zi90qNC`!N#lpI|t9Q`AoS5%a3Z#SYR3k`93d|%Do7`IoI?Os7QS0ZSzTIysY`-FV1
z&h$mObkd@i9~4M;!~E(Vx)@Alx#C6LCIz&B{+Bm3rIJ+XQnkP5vdiztEy%RJ&q9!o
za*gugcKETdU~FdLhb-qWeRpxAqG6W`TySJ#+x9|{QJ9Td67zi5o0g1R)AZi8#}&C;
z#z_Z}n(9xrJt0OGe;H;?R{gH-jmpaAk@5VF<RsX(0~<?jhq=Sj-5df^woVbW*HDII
z^5AOTH9L!{!&=AW3@d?6`5yJSj-2)Sm-l0S+uF)#dtMQ|MP@lauY)`}xj{C7#+J!l
zGI&(nm-t^6`#b8dthOPBy}LSwb0&R8V9@qg+w}6~VO{NC*|tfZy<ZuJ7Qk@;0vS&Z
zBHPh;?}E@}w?6<`=7(NyXF(@V7bw!h-NG0vEINKrl#j*dZBc3NcvR*ec@yM^@yGEL
z@p6BSkX_Rn*M(0K6B24lAQZJ%)YtcYV_G#PemcaoTI%!`cU`~nD^4C!UrDSW4-pg<
zV(SS`p+CO|cvC-KwzpHe`SZXNue<|Kt*q0v%VGX6G<aXk^I;R4m=-uB5Oa%6h{DtV
zN2uafaPXU=itl4B{ukb^y{X8ZS}wcV((pRd*33aJrnf&hcz@>gKXDqoZOoiMrC3{9
z#zUJV48;a&2m^Kt+?BZ0ccB>z+v91b!b_Rz?K`|pM?R}UL3i&0yZig4sSK~1n<2G$
zve_ROHp-VaFF)q%E1%uh#8mysgTU9bLEYU<=^2;xjE%=RIL>r5vb!EDo2B|mLpQ*X
zpY7vX0sDn+3^P*k=ghFXoyq(sE!YPD^*9&IoE@CL)|cK=Z?A{^1zC}5rfF{0CfSA@
zDPLbD*W|4>%QgF~uPk4bGLt)Ql|$ZXm2YN!tvWR&m!M-36aCx7EDt9HU}jrPht;~R
zwUZMVx;6_x;|^xdD_?Sx^ioD+V&a^{R#_jYsnKVd-~asLjZ0B>3X)4)_Yi<9)%9xe
ze@L`qpD|LZF+*&v<Ep#DT9ZzRe{BChJh4enR`=cbNv?YF$*_Zj!7iyF85tsMWuW2)
zog{Becz2RGr~uwo`&v4uXggH&)RmDtfpX{GLlHrwyf<LQUf*7O98Yr}xYv|g4NI*m
zhN|Vk^t;=hv3{QZv$*kV!HAltIyph|=*K7|Ez1mN&hz#?>d^EZtH0{SrqJN!aV5%Y
zH(wLAqZK<xNqH+a^=(9$h??(1i6Fx4cfwKe`nI)NBs<=N1(zNjd4D2+^9~|+li;Ol
zNpZLdOpY*tny?4E7<t!F(&uQl7^*Shq@+zVRXwvZ+pyvWS|0y*)H)+~?%tVxZLP-u
zHF#M=Lo^Z{G1~{I-?UoAmIiuLdDOer!=n#lUVQ9iyI+}~Z-_8tTOa$09(S7T>G>cd
z1Ky~KfQvMxwOXA!Ji9dNXK#ISN;Mfxb4=Owp8|NaNAgo0;m^VzFTf}o@I|<wy6U_e
z_RKGJ?*2izGL-rUT^k6z{<~i8R#7Qb_RGVU^j%yC@8@<FeYxH*A$sYF)UfoeRD`$T
z|7K;<fR%~R*Hz{`^AZfqU<fwcii|%-P%7CJ6!Y;sEUo(Rsm$3`=I0xF`p`!YUe$?t
zGlv)@77hfwB7HhlaRDf!GeD)m?C!AsYRITIi0H8(-^-#(Cdgo%3V0U+6u;i*<Vuq+
z-b|K%v7o9bMk{&TwDsT^;I{keRUhn5IEwkuKA9%cwQp7^jwkm&eN%V&&DPldwrMhT
zZ&cS9sBlj;?l`tWHBUnc4b)!ZdZo?m4>u!BRTKIX$rJX8hA5iK^gF+2TL3ZBrgXUE
z_cNB_Iu4%&#iY{!@G%z?;~>tPB(7Er#Kh$HkofklRoC*&`(XzKgTo4Ar*2b<D_6>|
zzQnFsG^`G!q;dIg&J+){#mMJ}h2B}6m1}+8Bb1b+_Au@iK&4Ib?mdi*xJiIwXg$Fp
zFoB=7Ri)l4{+uRxO9gk57(>x<LA#H>THgIJKM&1*ot1A(?&ncvq<?Th!rQ9k=N%O$
zrf(V=Bh+_0=VHs*xEv(Omz^$dj5l7OG6pYUzOuw?gCaLTUS1CmcDO=FXmxbt|KQRH
zKpdiWvZY^S%~J%~#s0k&sQ9&UqUoy^|5vv((O$r>9w2k>JQ5cN=CQxDn7WW+!OxRn
zA1dB^M$*by<c`xi8F%w5r>oOga}XoCbZN|KL;asIdsz%yQ|)oZfiM+pv+loD;Xe_m
z)<>d@V#Tl32Nrn`#mRVhHr<M5S0pBu3W_Bx6&hABQD`_4*B8WlRN^Ekc<a{4{^q;Z
z-!Yb=3d9GM_N~BuW>N0V;ot%7FDhhq{TWUL^khJr2|uCycD$=&=Dl(D@GbJr(%PLV
z7SYGW@Q6U4f6${oe=ZtRc!xgC&8RPZWVE|vXnWC8PLdF~Mjo%v>2OZggkhpj9lYl0
z;;nL(j&}CAkIIZGsgY9)x@H3KeqILw!C73YLBg+Z*E}QhEE`n3TudP&82dV{csrQ0
zntVeQvlh1f($)yHP<T?kxWTdB8<8WbM*=q37y1{U0}4J|aI5s<CBBBClPB2w-+4>i
z*B9l$=>0NYyCTMpzBaa8_&GI|Hf(k6MMu}~&gqaNu_Bf7*23=JxHx7wYq0=tb6#S}
zKLb~`O`DcoynX(>E{CvrjECvQ*OWRNe@UfrPt;h1-tDV?WDi`Z*3C%VleqiitYm2}
zpLMl=1hd_|v%JEar(hn&eQB(+`NywMD>@^qRWvOxB-2D9(qXxKTdOhzjOf?o`}Y5f
zsjq;_YKyl15D^6d1p#TLLsB|LkS^)&kd|%^A|Wl^CEYDu(%m54-QDnS?tSmR-=SkL
z@QZVHtToqM^XyrBezX<BmnNIzSSTsJX1AMd)lxh0eJV^{x${r>y0YCVF3B3*O7sMz
zwR+hMzs=o&s3TEXSl8%^0C{=8uhnzGmtZN<pCAXDvXR6>*F?XySc-tC=^euV#%M3Z
zREz0Z<d5*wlvFySH=juL^_c*$pvaWSCa=!Rq>nsOkB>@7WDUhS`;GtX*>WV2`^OAf
zo>mnl%c1vjFuSnUslPw!F(BcR6c>tucx`}Lb&mfozS|kX!F&-L9uQu^GJrB%7j~o*
zAS$Uh^Gh1A2@;2v{h?VM#u@})0HhS(Rkd;65_C1s7wzfF*v7LoC1{y#{-z06@Htx^
zgtCW4(mX=WRBbR~f)RzTT^vG@l_jbFvyy_b+UZqi_o4a9$7IN3i!Jv0BGwp&?<d~W
zJnAJN9kkd&$FFJ{b;b3!#O6k1ZXaJTH9x^LR)yK#T3@zkE8JHNyZ_Up{5tC(J*%;3
z{G9%bQQL;b7ygeg^v>eP+TGZlh}X!;)q8r^wig2I2q|jIBPvl7VG9+ZB)J;9TVKrk
z@;{SM$x+5s-#>-?IeBfwBK;>&3Dc_7`+Hh^mrN|P7^SEqkNUl=V7<JqvU>F^ajxJ1
z_~Mi25<x+i7n8=SdT1J`ixor=z!V0j2j9Rl3u#}m$PijvyJ4-Or}07k0E!$i;!9t?
zC?~F8yaC8k`|2nEtM2Yty9G+STSbtU>PqjQu8KiqmcazKe<M@=V>4gmQh?k`Qug(f
zlUR3oV|B5;dl8<IC41J>8jVKWKA6ZaJeDdC6<=5B;1ZF0WQt0O@+S;XN_avj8MU>h
z1yiCP=bQ*kU<t{8aJOQ7;_QrOom;`WlXeNJ;pc~kE9cE_jJd0IH)aKs9PZF#)^;KF
zAiq_>vsbj&PENvanbzvvJS`CYU8{nNdvu=N{8({Rr6Nbiz}UF@xeUdx%0L34pW?e)
zfnsFnShoAEfxio6;HnY4H%vLOC1++n5Chb}Xnajbmz6nTg6{!xZQy(<a_mPq^k?5B
zZ*r5P<;6_t=}DVB!<+M;HA0%(R@Ob;F;#X0iMT^)wI-ShYMb?<%f?G<-GDNlV|RS+
z#KID=pCyyyxZWvhKNsBIGwH`6Ej^5TKQt2l)ux!@ZKFqtntiB7ZZ%;c`F7E~ZvWwX
z5s!?bhQNlXqvxM8GPti#bpe&ehqlw_wI*r8)4viWd8zY(9`HE>kozqGnL(*)rQ6Rn
zX}7rcU{teu%r-<}>C}7!V$DMNqp5Ft_`#AXPI?Z8(t~wKn*Ys{Be&n<q63lEETe<t
zV2v2ujWX*E*mIN56vq}Iw!b47#Pi^wQ5@q8J$%D+-sh0*!RGmBW2nl$YqpklXDBx6
z%s!%1zZX)g_`OJ>ER0B}v1a&@I7-Op9nmn6F@e_odFgLmE3H%N352|f)O_-{%F3$U
z^THu;#b#gmc5)|43pmNIuJUnNU%2H&1XL1NN5;sT?*#6<KVJAw$6q0}H@6GY!C~od
zVn8}svRUM?T3l6%13Q`5TbrS3szUpA@fWl!;9x1d>Tk}G+kkc1XfPz{^8N(!LAE?~
zrrV`th54Ufzc_4maj}Ga0D#Ec=V(>)mHMfsn0d<HPo6QT*CzzB=1?|qEN`WcOMiE7
zU;*;FuAW7I>EKr(iz$A2sWFtt`&b|&$MkVsZpyw^G|V(pV@Fd%-lY-9kfp+oau0!W
zLSL?j$!-32hNr;DZOnvSGU4nK@wC0?ItKwYhS8N0un=QYhMQ-J>H6yoxqfJJH}R%H
zV=I&6v^Z<qpL7k$`SxcN*3Zy9>m$k3uzTD7w0eQM@pYpe+u<dG>hw&1zro-3-TW%_
zZ;2TQYSdH6|5r>B5;sFM+~^PykB<bMk<j#bg=T?MqmKJrxr6<Rsya8e<($a^&bXvK
z|JzE8a8TkJ*$dCIY7U2s#7s+SoLjHneMjNoNp!8*BZ3)%gS5w6m9uz3ZS)cFBK~6H
zlWupO?J-1^dmjpA06(FT)ynPx_!w{lfULc6oSnzjTY`p%MS(LWh{d2aw80#lnnY+*
z^BC@+r*L+xeU~?jcN|_iD8sJ+#Q+JRh=^OLiOBSF<%FZ3lpG^j>%&KPI52pZ)EC73
zG<R#&vxuN^_O*gx{ZH9e(3q64pnN>Omeaxum`#nX_H@;vG7iqs@hU^y{cnE|5Enj3
z;D<sS)u6xhGA^q5Xz;6##ovc@Q5Bik^1%{~j{Evm5G^ersD;#P&LZ0-XJ^=tTXf^9
zYpvK>zp+C3tFejKla9-*FJo2GF)?ju-^cf{kvQ3(!>Q7$wSdigqmE~Fb=_8el0W;z
z9CciAM!ovE5gy(zo8yi9bF3$KWs&m&P$7C)6u+3x%TSJL{6IgWvqcQ%YjaF-IxkSR
z#2k;3Mm1cOj^8OzW73=vz_i`Zr7GgW$s^5`cKOn6h)@cGLfEnz@3Abg!PK-bG@79d
z?DBR-#fPd7bdwik$NcDhH?Gy#HcD{)^g@reV+q}OaVVyY&($|MZx65hndckme=t>u
z&tH8>M^b7G?d3Ry`zt3&(x$(L?&@Naz{Jprli76YLs?#J^6z~?gy#oGO+r0Bx-2+=
z3C#HhXUxH|f`z41pm$XJR=tMcS<{8HI!TV;pQ=oESLZeE#+{kIuPAtI1$>2ta9pX?
zd&}9=Y&xrvruYr`xrfIPnc8nShS3gj)4tib=jX@GADq3a%5C+nIef0p-Gel12t=(^
zaFP_GY;CQUx{>9S9k=kc5Vc6DDMF;BlGnpW;(Gia_9o)RvD>{S>|x=`ns83{Cl~-Y
zp|IqTEO-9CFqQ~JBz5OKW6<I<G`~P`Z@_OKh-R7awX1*_j5I5QL1GV168w`*&U^Ef
zF=DoP-N~Spp`c@beF~7j(Z%{btPPV@ey0<bC3iJzM9RH^K~9!ClykUjPsHW=@Y)nQ
z+Qy^)HYzfjZVud3@4x9-#^4~-a^umnG%~t#J(sMhFjLW+uM7|KlaOkjZw_9w|7{mH
zF{{e)cLW;k_c1*@6wFFrDrm_)Kuq-()}y;Q;hDFX`4dJveL=JI1o|{H_}bLiq<iO?
zo|D^&NN=Fs^;_Zpi^JuDo=T3Sg&l2r*;&GB6u!=<BY@JI)pGqPAR%sqZJC3io~XVZ
zy}&p=k(O9Sl7$}jw882-|F>N{U1vKeQ=XbCva6Vp^_O3wWc}5ayj2i!co>U&QcS3m
zbVC0!vux!6-QbOwv$8z%!Tot)Az(s`H?@Vos6J8B(>2~-cNtiZjJL3}w@WR}Zo9D{
zxXe})cCUHeUG!E-gpN`$Q>Lh!<)@ex)BDxfaazYlam2mCo$0PIZ}FVXggRX4Gg-45
zb@<D~Fn_@C7GhTT&c5rtbo<J;=CIwGl&uT`Y^b`jV5)=<$odVe;~;_ms&yfGzMuPc
z27Cw+osTt1sy*JH*!b7SARorCI6uCo*IZi2c(DXdF6<rX=pXiMM@2ocoDcr?ZFl>W
z#JfShd!S5nH6A<bxTk*>y0HO_M{F!$Jakivd?Ryk64jWOgBa5`2lLNZY-t+lw_FCk
zT{YCkMz#cbd3BHjf-MXeoS_$4*&d`b?;U_&hyT&l`N<>#!e2GjrOlmIkF(v^FaAnJ
z=>IoE7tt8cG#688Xy_@jieEU02XIc<)hCz5IxA?KKvGfy5|TzalhrWJ==iF5@ixX;
zDG6u&XLB*-;OM))-t2kIM@BZMhbx$v#y558zclm1Rb~|(=LSYU5Qz&292vy1)pd9F
z<i$we7KN6@;{0sxyYQlZHutyOHfVVR4wMTqn9A2jFSe}Sq!K@KuPmRLanapdGtuck
zfkByw&;<Y~(f-w$h7-FSrOYe_t|5DY`$P3oyIb9$^-<XovbcMwhnwEmy^k!pURY?$
z<A4@WQH>Mse6BCMpA_Ou{a-ry15XTqn}AVDQAm=!SdO};xdgziAOguE1sqhAWcnqd
z_~5eY(kM{E7+9G{jp>jw!cL$%b#%FrlFE|yvAKqBn)3=Bb#N$Ata|DQ_nxAY#2RN|
zF3meCGBPf_>W|+73TQzJ71@drsVo$6Rp`&)6{G3*<7Zh{C|ar+KcC5w8K|t$V_{;*
zj*&h+*D#Ny9(7axA|e)#gTwNd-2ihr=Cwk7V-U*oeZQQL?rPkLG+U^xB}F!;vzz`6
zEd`>So}RjyxvleA5X~Nf`8u5zIY~y!Q$NAP{K|a8gnK@R8aG4`Gyf-Kesp^PK!c)W
z=n3ZLBK^I;WE9=A2u&14sfq~M{|r?LT8sDW8I$nW%UZ|=WWL)P!nyTN*H+i$g#~TN
zJQ!D(oQ~oE;|J#i_-Hygc~$52_dGcffNH=Qsiy%A(vY0byIY$zQ`2xS)EJYmvbi&(
z2rH(4BE6yGNu0Oe(=j@zre39<A=ts0DuJhePAKA0BwcV-R#q-Z94a4yE;Dn?327)0
zh7@>vw-5A5N>I|#kkHZJ_nU}lq+?>em80Avwcdx;bxp;;$lC&@)DGt<Bt1Z#d|92t
z=5PALN|&bn(7y0$tMPjU=9RN^>pEtNe9l${Zo`oy?&Rd=OviUyK(>IER@wTEH#Hf%
zXB?}J;JOqbPphU}b^zmv6YsLmQUf;5QSnzm&AbSYcNj358&?Fphc0MF006#!1J%;a
z^Zz%bo`XMg>HJ=K`KE{mU}bx>SRKOZtPa?NqUnsb`zXP9f;dGE==g@Dy>cttCK;`e
zl|RH-$&G=_=Zz(Wx-BX2LtAGWs=u?xs<18Qezo*Y#9nl*oAJ0k@vci?4MsgVwWa86
zdnPUbAFNf~z0G$k29+6QmEHaIQ$1BvAv)ETP$w8@tp^LXDlEwltYU49dEDim?jFDs
zAjR=r6Kk3OZ9TnsjH{+PI9BkI2d{Rhb==1*P{Mt-!nY~5=>2JcWoP$<dfJ&g;O;>G
zCqZIRovZRUs(eED-d$V|!s82P_^%A)4M`viDk&k_DMMMPC`?lNQgomueBk6(cjQ-(
zhcviI;DS3SW>xC*ku${X*4^V~vbMHQW&OOQ`r<>?6x-V(m=im`-8TwpsInF*@uAQ%
zad^Y+59l;q+bFwSJiYw%sl2>0EyS0L0FPt&i($8Sd}U0nHH-TI_W;N$00pz0;r$&&
zbK`+{P&>#{WO68fb~zf{znr?O4hG^D^zeA!$_mWcS&hxTJ%8}iP_@%TX7u122L(YI
zsN<-Vs5#K1JyIKZcjHGb!$16QS^k)RFj5)N)aWENw))F{;Wvc1{p&__FXE!$nl{LJ
zdy!1UeTDirW-EwGK)LxX24Zhu&kJ_bIp>$6ATPUBWJdS%N+tMif^Rdsm{xZ&ixv?X
zJ<>lJdd`VBJU;?2_x&Dv{)!s2S(NJL>)VLk$~4ET<Dx3tZB{J_NNO9Y3osZdMOhu7
zA_nyrc;n}+<D>P+r?5B|W})0v(ze-Q8ylS31G#?9+F-y%??d)T`l@+r7gNgf7}0>_
z@WHEmfRg|c3mu$po83F|(&Dm>Pu+v4uIz9@*&|RiJTan?wcJHbrPI?V0-M9<Qq}7P
zWwZ6dl7XT|eLGj;BgXNfYZD#J;&;sW9?y7Oz63hi55hFnu8=7(!m2_eCEpV$so%ZK
zY!EWfQ-M&-jg0-m3s7<<Ic{ur$?r`yYQ@Nc!2`_y9!x+)v{y+%9aa8I_nm~K@mk^K
zO~}TuALyUGLq^FGAem*o5TkpiFD?I$+{fp{T<!X*32ka<kb&IwhJ(Y~8}wNZb)w(+
zTsrBwP2@UbzKDuywtTCk0_Yc@$?C7(W`izm&f;-<ROL$<*|LO9p<pksLzhm4C|@6p
z&{*%t(6^R$W&?7xaZ46X`z8w)H;=}B=C!*nd%)LIuv10eS6GN@#!8sUlmExfam(r=
zfVwbna~u;8iKrc|7;#bGDoS3~k_WEre-_x-*|L_l#vaA{5@fQ}5(^n)T$X1!SL`I~
zyae1nL$!nS%+78_aOLq1$KCm<ZuhEn!^UiooE!zi@W6V~3Y|xi;76rOp?o49FM?%Z
zZ3F$*r9L&4Vv~aLJ5W-k)!08-s87q18P&yHybVJ_QgPj5f;}+7EWgtw+J96dbGYns
z?j{ga13eB<j;KRqK#y~k3esfN*%C9*d0~XdwWpl(5p&R}ZNHwcyLbes1vSf0R-B(J
z73u<{64PW9rH<S8E@~c_vbh`*fa^)IULCcxy+~}IbiBpFmX&2>WnKU!JWD&E%SMXa
z)IjY`gwG{8Tl7M3b>(m*gMqZ^q5aHiu{RP9!ChROa2$0rTV7>K`l<H|cMQN7Y+8#=
zmXq+<VR$w+KqK-deMva1CeG#NU4}IWJsOdJY)Off8*};3PD-I(4G}4ww?ISm;_7Au
zz*zT(vCU9>lz@L>AVVZ29seWVn*&ceVdc`YG0%!@f5%D>&hdOhl$fb(WAbm%4`_45
zu}Drka)s`Lo;Wh^4!U4#z{tklUiVm_#2lYJx`#u1h<OY=_J36TV}c(KWj}~jgPBWX
zuE+uc!Mr@d?nzTh5uR5SCzV{kmj_beVRoKI66b}g)2$g<?@xP*faW3pH7lS1rp9wM
zH$m?Jm5&7ryq6cccHyZVNSwfJyxC=AU}9L%|A=lAlAI&=s3f~C;ZV&%PF_>WBZSTE
z(bSh9=(^HCz0NKkH7J3D*(tx>YxMRV3W^G!$<uH(Rx%Enc+q6{RF-FJ+U9K|ZZ7Z#
zqi6MCZpIgXEEN5~yTP)?1ZBzAxuI#$?PW!*h=(`DcMb?YK85AXeGx~cu5SM+4D@Ex
z?ZWE}EE|FaiFGXH-z6tk%Vb#{b?XOCKae>5dcwj}*9ux6_1cTyR!vQZ%FA?u;tFVS
zDhP#hWDk!TT4p-C+W{5uX+R=}b$n_juv#wZBVdGp+-5kyf7@1~(O2C-qZMU`me$E~
z*bCyUaFh8*aJHZ)rq8tIu-`i2Rx^?N#8!<0QA#q=<6-)OlUF*-xhmSvE$6X>O=-F@
z#hy61l$!n#*^xBW^xg*j?7e3soBLxi*52qwYIx2Q2X#pQ<<vjlA9`6NpVd{{lWevg
zPbUjpkJJSh^+;PxMkq}~mHY^<#jf7te=lh!WR+#R@%YVS|JI2nwm%T+=WlOej_2om
zKWTD{a!f4DdNVI=eu|L3<Aq`;Q1L%yk~PSJ65Q>6s&{qN76d5vqdsiaWfI+G<=;j`
zfLJqo&63_tt#1xQ7md$%#Ay&+fbL4KC+EHDa&+S<XG0J!6sjmM5yFC36%nVo{F++3
z%ln1qUz?{MPIAe)zZ3W~hGuIR%s=pfke|l3zf$Cn+rz@~3Qu~zk@aC=oGa2^9_09`
zF0&=E`-x=$X@$;Ej`kdnb*W<4#KsJ_bJ29?X}%dyT;Vu5rLV4-nA_+cPJnG`^M*Wl
z5zpbL1|*xyLYVS_T<_?2J&3<pSF%(c4Y#H3vIIpzxcyaxvm%}JDc4g$pe>Vcg`XkT
zW@JRiu)V-|a(eQYeP3Y5<dCGf<zP0d;DLXh8HhjUnoxIMwdh|nG{b-Y9IUrrEZdz_
z`FFy8%Swe|N%6iw5Sb_Zyfww=*p*FyhX)!DeC-A17=-@>)JDZDr~jL{j{@kRYzU31
zHDOt`=BVeMdA+t?AV)-GxBUZzQ&3Gl3kV*64Mb^OH6+3%ER1X4fd&WHNFIZzrn=Vp
zJ)<9i(f0~a`ylCW=FyWxiYXJ~O2zYym$2S&UY6z-ZBndY7au_S4CZ&r;-H$t%FIl-
z)nSav%v9}Au5+RR@(~A4ge5Zz#r0)SIxd%KZYd0C-IVd0=F3yB-bjlde6=R7F8)Di
z%6f3rKq9DbsGyRno+(od($yaL9T3p~9!-;wF(QPz8*_9eq;g&xz{Ad0NJ-OfKi_Uw
zT5<E1h(V8ly3>l0EGVmT-{7y$E@*@e%?(6<*K#YYfWUCG;jxmyx#R!EGbJH$uX-(3
z%XhB}<c0u&<dyq%^)0vPWf`BZa~2vHjsOMB`gD1EReoTy!_F{gl5aQ(V#C!d5_tIF
zcCtb1NfQ7R^Q|Jy?M0~?H5)wdOIges2_WxpNrZUoeW{VrMv_@5eu96J%jWfMC+I+L
zY{Y5I8R*~=;#D{l3O|G3d^Y{rR<k1MQ<f(q3)50E$Nort<QmEWA#1F9e^{T|P_)?P
zFC{h;O(QJ@uzNGF*eEKqe;Bx&Dd3$0A8OdUjqzvVEda+Uc1(5txSZvJJ6MITf;fBS
z|0@+=;Qq_Q0JSf2Qvh6nt=N(WXEPHxAjloLy4xYXvg9bSGPiXHi8n+nL)SVWQ`vaZ
zRuu2qfvyK1DD3}Qt>Np{7}Gyv)cgBF)yaj?<c=69LmYJm-abDgNUccY0nK*+OIz2m
zzrLqgmx@0F=a1swt=aUNW`E0-p`OHaMFn)Kky_>GrX%diiKTs{I}D^5&|kB1$Zqf{
z$;hFqRHD;c%F-&{JyB6q_tS2Rd<we%2PGwch6jJ1ETI=>8p+BRN`=J%4d_UIti<+w
zrPT3M=>kc^jS&g*c>XeIWeG2#10bJST~CV&f!=3Cr1g6m-*SJ_oFw_~lhp4jS6e+S
zXKst_lCn0|8(ZT))>SLheZg~x`&@szyhxKKtBF7#QI%_IS>I61PhCn~6D|}p&o=1f
z#1V+9i0pPh$V(CS2((PPv&ML<c2mG?O((-T$Dl!@vumKDT?1<-)@Fy9mSE>=8C3j%
z_}hvnaiO99uYAhr#Wn4LZ}hwwe(|OS1Mz^H-ClQZV#8xQp47TO*j}Z%alS+XWE8+v
zIu>6(n0#hj#EAuZeYh|4tsw@M{l6+Il?pnX2Bp-i#WX)Ney`&>EGZeTP6qNJ-LWy#
z9d?Y~T8wsuOK!kPr={uuWfjC&Se@LzEMYe1^*~8}6`=%#eTXEA-hY64vFo9%CJ9wO
z#C75;ird~ZdEeCYeg0#ib!MDWI*nE$;lk?YMytB9U1@LOQR)V88cMq5O+28y7uOSO
z{iLOfWapYbNQWqDCb?b@2iRIF)avi{c?p&tA#W^=Lr?Ma9vbcfo>wvMXw}`D2T#dQ
z_i|V?sUu6R_I!bymH}ur6KdVVy>1?%A0h8;jtou3J>#i8?h_vwDr%wGcW|*Bb~Fse
z?a$-8>)o?i@1mky>2-^*Q;CTgER&>hchSR=!exybS<!E+$g7J291eaAx6gM*s|O9{
zEi`1ME+Sf+C5fSi=X2@ISmuVOPTM<nZ)hX{@W>xM-7XsHKiv}7)FwbLTX=uTH9e4E
zu-DBpBOyWxM6m~f`CN>$evg5^$d$oKH!*z4;rwvV%j(AR1E2d`Nr1VL=d%z$GFJL~
zsX#EBB(;x^Z0F|!_ofg;X-z)!NjN}B;2aJ&`f@w#rv6qh8#W8#@SeO;gNCoiSgY0u
zZa`!vAGhcZ<j+7ysn@al1vGiV*~$f5ESCRG5dWYz5GSA|s^<u%%IjL1{2Iv{ayTwk
zEIR6;8df)bPeoHk++V7>tc-`Fqh}^)t%yY(nLF+Wo|gyQ3UsJP8QE2ZPSA6D;Rtxe
z!}FsQzdX~RQ-%HLh#V;-XO*#i$Gtrg4zAj9b60hB_LEg37|;OPkdgZYCaB-c%zRm1
zR?u_un_^;PZ73t=s!jPhm`(XhPS(n%I4h_Uo9?$=-sJP2VHvzqmfQDQIMn=!H$)4o
z{ME9eCOR*3piiG-Ht-g7E&LNj$0RFBv-fKO$Pc4o{g@XNs=@ibRNN>+jO?Y=p;t$H
zsto0q%}st9OZr5NrE{C|--gZQRPSO~y%YX?lr1f#0ve;PaUUUjdsVhnm(j3N>V3;u
zrF~D%$z20;QAM$5#vIiOirMo{!TH;4{j+YQMh0UqAgccgho@-m%h%Bj2gwTRxpQXS
z<#6bdk$8b!t8r}aY3OM4e!R~<S0=xDvJw&XR*8}!dwC`Z1qHBl&j{ki%)=|MF{2d-
zr3DhWPC0V8h3!@sSAO`0>(#G+#!y4Hxq4ilD*Yke`P628M!{wmC}Ba&`dZXxs0T%^
zH>T(Bt%HW=aCc9c;;G$sPx}l$p3N~Luatms=k;{@5+`=F!t6|;HGa5{V=#5c&>;u<
ze{k6A>LE#|mNBz@A`S%tr$8jz!e-=|oHi2+Rl5u_kFo+37*M!Xsca-|$mYtYV`J-C
z8mp|YSO{1bQVRdfg>=#iYs6@%WzSWtg73By_7ZA;IiD*4cg68iZZ8?Ge%8nVQ-PWJ
zw)2x5<Al*aSM$h`154pY_zJ?Q^60L+Ph9BL7Zi1wHmj>+;$s#}EQLc9@1F0y_)+kw
zk1y%}#@6nirmBvinS_q$GaX|bAQ|15k^U+RA;AcP<CMZ!Z5$q${)b0sR+iF_*`wt~
zfcTf$%3-8B#Aji9%U_Uq(bbw-EE36>$0Yj+MaHASa@2UI+jtfm2gC(+YwCf@!-?wT
zN3nqif$!c)O0tlnA<VKAI5vP!WXkd0%bP6|6Z_HI)z;OnsyF%bo`|BL+sMw24pO-#
z7;W7ha%bZPAMTIRA;kcVss56cWp01avie+I0Y1qKja<5ENcDz+_<wI`@F#BOM%tUv
z&jvZ3v}`1Hdjbe%g<tq8w1|Qj1vno!YgpCfCaezz^Pg^IHSD~j-59OIG5*w+AUT{M
zkrNcSgbsLR9X$)FbqK2u@2gKp5s=RfhKo%S;E4OF7uxXElIVp}y?Tyh64H>Q3EwaC
zdgkp555P766@}fN(R2YNsObc_><5+UYh!EiH2bBzGK_#~An%x+t+cc{DpnFhNeQ%n
zHfIJJ75GJ;En~TNN6<HrQ<Hi37t*<oWpN27;fi0J<UEG>-uu~+Aif70<SX<Gt7+7!
zI`-p!L+AhC`^Od_Z3ie2w1QynHLrw(to)?h@P7m^gV_M1!_D=@qdkU>Hhy_|=}&B{
zK2U>fFiWX^W^D9PZ~mYaB3kdVH7YFNZ1{l>hLwaQl%YoiN}3Pg0T56j@iB@*Vrt09
zJ`Q^#ilWaN)f;}v5m{%14fH1<YOCHifaV_D*Rski%W2p{Dg6>FRgMVs;j|L!pUOi)
z09B7rC#8i*CytnxHvRJdEtlNnl3JV|F=yG>0c7EhDrzN0OOy}gp~^QnTPstqq97JH
zDbzsot)9}iw^esb?<Ec)ODQ2u!rQrPk6Ubdo-Kj@kkG}zKli;cS%Z0}ne-|$`n}u0
z+gG_Y8Pp>I{;@#000WN*bri?dQ~_?%^=5||ra9*j)5}R7zA8pWzr!J7Rh<y+gHrJO
zRcp(M#C^j>HS3a~D>|PRQl8B<(0Se9S|hyI(}Q?pxDVla4-P9fAh}QxQdie4e4R}k
zI}5(^Ruv0_^%o1veN9bgXxcu1vIcnIn}jL)0dmUf-!&rQ<p;ITcdp07eT@bcR+fP{
z-rplJANudhLDP1G>-(l`lh*IiLBr25i173J*}Wk^)JBb(5?|z*6E!s0dF&p!anC>%
zzd>unVDhKC>vSa|NIBtrw_)*qerU$+|1E)g^9SK4(=>K}i;PyX+fYgGUzZpr?_Y|$
zfT(<3ZH(xL;9bSVez1x>U=l<$pN6Mb>p;fa=n)!kKkBTP0R(@WE@)N^<Vl8m7Vl!t
zYzwA%zB~cFbH{bs=g{e?6VMfmdubKfO=BF2a@V^Nfqh7%<#yTMt|#qZ_lH=X8WTLx
zm&UfcB<5^bFqPv`<Vk-td)Xu6u~f>Gs3eU%g}{2aJo{|96FgO-JA!JVZ!nRYD4U_%
z+@&4m?E~Un*YJeq5U!<;?w`_1O=05DgQX;a9sj_bQ~XM3K6ojk-WQDgJLuXDBtX;I
zvqtUd*s=%G7U(@#gK+!{M0bO=9`%~d`&4rN^9;!ZbhX<o^&e_peD?MMApu!@XOy?i
zP`H>1o_nG@WWMHVZBlUv?glEH{dJrRJr(U~`ZMYkzz2oM=iahLiICoKK-ZF-tft+7
zFXSxJ9Prt$lrd3Zai!yk<lSaHbZC`KFgN>K^OaE1(2$eJRA3u~k)Anc0*_4>R7Smh
z9ocfy!eaxbhqhv;k2$Da4x3k=N$dN4bIKckGzj`rD}O#Z7fg4aV}GQKRJvdO2!JU@
z<(}c3U)k&J?L{Ng7O0?=l$DV0vN-d$^+8$#XtT(wI#=iX*-r3~$D5HO_`#rjGa>V-
zE^H%uZ42Fp3f!b(%3V1N)s^MeNo>Ws-Ey+6R#+-bLsr$?73^Y5BHrfK0vAd-O68kj
zTcFR|vJJdf1Z&TUW&&26t|bi~sMwn6@Z3v0DE++z4X(bBQh#Q-LG@qZ$!$eStw1mz
z1pUd8r}p&pbUdGRh-J|Ela>_R{|eEbZ;NWi>aLyks{R_T&A<R=Yqrw+LW};!EWd=6
zmAF3JYY+%Ag@iC7!x23y1KzJkyN6Ub?S>bU88B&H-3bG&DUt^W1TD99#>*QXP+IIW
zAmyFu?$?o)-$@j-(Z=9o9k_q0<Wl>Dcgx3W!}{sNjsOTF@bnaCVJwE1aIZ>DRXc@(
ztUBcf#xv2KApf^n<<H^Wc-&o2F*RRvAAFX+O%?C&&ODSGdC&s12O3?h&5&Uw%1<!b
zGru>eDq>62NCcZMO^}X)4x&$nXSBi(5*I9#Ujj|B_8|dLT57}IwZ!M!qu<Pt*Lk*D
zt_EZ&BYV}Veds4FA2$IF^-K3T+b=7Dpd<Ndb+vV6t&EQ|c=^x|$k^*N5+<4MNp9-7
zx0%P=F({nw;>lh`M}lIZN!wD|*{N#LL?>7b0X)WnA0Vv{_$3H@LJSUG*-|1K&Pai2
z3(<uh9W#?Z=p|q*tTas;k$<*BuARDk|ISnCj7SG)sBVs~@=!QCd@^MyDf2Gw!yCTw
z`aJOoaTNNKMhcWrk&!yOCZ2-p6dx|=DXC~bDU`g;&dJ#%NeVj1#*B{8*4FmmXE9jz
zx_?p+Cs5(838||~SXnW`u&}UR<KXZF8jp2HYq@t_9<NW<IUT1Z`M-IJ_Cw@gJXGX>
z_jLvn4n5Qu(HKG~TxgJyRPFcdIE%L}UQc+NdwlVy46kZQ95S1w0^aiCMrT461!Dm;
z-{Fo?nt$`CLA5bd%YLsRoU}0+6nU+hJjx8_NON@k5So)uh;Jk!>fxVC@tC3CnDgiR
z2p=4Uu#TmZ45alXD#);b#MQB0Kuyz>$fG(H4&<L8cV!=0;<u*TJ|^3I$SBK4yCS@*
z)p79H^h$VoM#jxMhs%z3D_<o7ZN(eN3=9vuY8qAS7G5w_tLp`ZeuKNm8w?4_lxS}o
zSIn}4?sj?aUEpn$3Jrm_0zBg=vhL30FaD*@mZR|OthG>WQv2BD1?aF>#RHHZ`9DAK
z;(#B(voEn1d_NDAK{r#sjmxrRXP50_y_4~d*FIsT_Z=YhiYtnnrrTnpeb6$DwiQna
z^FSiK`i=4iz<^t9x|fM@Vq%aX*6*k&ppQ$poMtUQJ>@69bhueG1a}cVlB@^tH9iXR
z@}D3zQX?Z0CCV@QMgfgk6hIw2VO((xeU`Uog6$^U-;cRLh+M28H~<|yyZ)tD9UCbC
zBqf<ou7D_bXsqI;26)Qks}jzOhqC$awDnuEpopEF4&CMYd#DtgvAuI?`;uyMa)$>B
zofZ{M*IxyJCNLn8mPhq+r25mb^Zkou6Uz?zqQ-6|dftK5p0{2z!NiOL^)Hia(@4zS
z+%_f(Y1b<9*{-z|F7Yeuth1H)z+=``Dylz!#)_&qaR@zmg2Kr2zXg*QMc9Lw8xs%>
zRJYI-Lxxr|B5zbk0CG{_Yab*`vNp2UxHvp#d3F_yj7<6>zHfYN2i+8AAr+y*nmXU_
zJ_@Yh!zE5FHMbk^B_#JwsL<=sd_ik?00sA;siTB}Q3udBpu}74&+Y3lK5a{j3XMkl
z;vT%!Jv8bCs#vC@BLHfO9H55;3;om0l?K9mOp}6i2_Z+bcU>xzohz$Vf2W2=XHgJo
z@`QO<__IR|K7FKalu+DnI5Jp0KcRXm?ceRvkiwrtVn-A*@+9j2lC=;>)}tHpj*kP_
z%vTpOdLlaaZqqc0LkCM#<TPJ1{h)GlxzAH-Lb^9EMBT!KJS)p=w#Gt`05(QQ37uW8
zzD#<<O=*F;<#MNJj|x~n+r2FL!Z&XlUx>@ey_1l@HTgemC{1eI_x4XXb#01A2{K>m
z#Poj2vKa$<WZ1VkLJ713Bg1RfK;x|d5T}|aEDyj2`*J;;=OrY>vt^9zDFi5WSTV%L
z`CIeWE7Rbos<NQ3p=9!fbN-+3EQfd>r(i&v$xug+aHZJ6_lo^5Fk9>*S3E$%1<wvL
znDuiLhs%8}D_^*usy#l)?Do1{J{O&~&5%*0F?fkc8L;E|&J#iC(<fFote1j|>v{bf
zPxq?=x3kmW_o6*H5B^XROF^G-2(bA1b0aGvBwrzY0(&8Gm{ntV<>s!*Fk8dJBFnfT
zd3(#QZG}U6FQNVC43ATP*{!pwyzg(MBYG#V!fECT^4;PjarL4KM6}H~st}VqEy9r>
zdDqGj_>^i{J!-jj5`nBn`d=icbWTm@%4Lu1>Wu?PjrAKF#&FOMkRbRPd&RUtnSzT|
z4_T=M+1kz@VtvPFnFV3#p|^#F{f-ak04;>GdtzrKfJ4s?jP7zrTX1Ts`}WH(=eSr>
zZG2@{3rQ!R3rv)M$y9Kd%UoZ+qMldD=F&<k462oVCxRL=!0?>MY5j}naN0aMbz`U9
zkWIcMS3Yzb@vHOT2~-1Z{M-tDI_mTd4({qWdwP{s@^(LWf0d~(AZdBN_T1$3n}7)c
zMj9jhvdJ5=hpV~3Leh*l>TsAF+bg%79lf%4bMEOws+=CE4SKtI$jigxHpoK#mD8>R
znVyD?uW_yA5%yx?wtG?DRK?g@Z<@1qTW$b#mODpc#*mgPL?$;pzgHcPk>6~@<J$AJ
zjny<a5_r+&c(K&P_&fB3>RmxS4WXp<XMEw93j^4G4_A<qOX>Le<Ix_Ezx0J>w2#lm
z_f)pd0R}pgYhP#kk>bCKVtiH~_gzb>d}JR*E*q6?mKi0>N-Vcutz^BLJ(|=da9Bxn
zs0Ck>dk+aGZ^g8vcCpQ-fPXT`#@}kFNEu`T?yds=2z)$I9LAb0)Aue#LZX>zk0}2-
zL8wu=V6*y)^Yo;O5qG@OVfXQaZI*{~uqQT1w58CdOC;`-r2Ed<c_z|Z+C3ATFhloh
z<vlT1H|%WDzF))@TVsuR8pmr_&hu~;$)~AM(zScyZz3MeB@sd2G7)jvbK_Q{`=LSU
z#?h`i*IS(nnfl`nM1A!YB3_TH%oYRM@}tP;F;J@lja(&X@0HBRIGOQzmmFDpBGtRm
zb*h{i`{s4r`{dPbJI>P?FDqw@8}@*(e-0G8ZE+x3{JdgqJlRfqIc<hJUE-Fp4jE(P
zq|Wu}mh<umbk&}3<l39(F_OvHna8t@OjT-4yeD!^R}pU<J~@qNcDp6_etfY#JSek@
ziMgp<*iz;t(6jNSp4<4BP?;v>f|Y?8UP1C<=G_r=KYF8(cRJB}dS$xg%u+m!Nr0Qb
z^D!%Um6ID{-~6hAT<<(jI_sz)4T;<Qb8&70&B!6Q!ogem_P?E3%w==Z90u&nd$3fK
zlTsyL)n%vQ`uJ?^^34<pTM*<R%#(lVx<V1-hybVDl<B|rp)>v~^Kj76L7vNYaFyw+
z*Dti~^i%qc(O;%QyqpRl!8IDm`BH<!D57GLeEIcRrTXumvjq9etFJ@-xK?&VK|2#l
zckUU|<If(JS2iw^rlWoW`&-ePcXj-c>n4PR<f7d-eG_)yYBCihFnD+@0+R^YguMGd
zCU*5l@-%6szI}U?7Wjy+6O)#HBmjeAqs<RPKQ7by82J_cm+0i6$drfAP09Rx@TX<f
zUVQjdvU0&Ycc0w}4xo@pLe7?a%l}PL7*-dj^BM^mwUqN8E4GX0;4)eTS))iZ4A$kH
z3QrQsY?zH_s=95Q%HE^;ViirNkZrSVWx*rbu8dMm-on4am{fM-DxWyWdpOX!u+(ap
z?%``;F!@Ia^xW<f>3)AazeglE$)g&DetblIe{ylB4b_n5nC8tZC~+6bf~C<O9dLRs
zy=1o}@q7IW7BBh9qLN7}X>^SzC+;$tY-_3}Qu7iz-<sN+eJAnnamz4E!DeK%+QvqU
zjY)xIio&625Hrj=>?7kSXt{*_QdsR|{#Q%jJ3Lm!4}V(!OMoRS|60n>$aF^>8mTbp
zQN#(``$Ck*#aXV<xW;ILkN+-ux}baqzhK5e#-{Q<z0U<2LXqY!Fxsz?lb&aZN^|=6
zFzBz%?AOo4s-TS*;0Qib*6b#Q>Qfu73!QrZ84Jl8<KlrtMji@!tqGp9bVc*YT$*3F
zrY)@{S6?oIa0#cyLYteJNFlHjYU*^V$Q`!RSuljQlLZiInBRP$@6}4_FzhS2di(DP
zHU-EW{PN<Tt=YOFoUPcmjk>GSt>_Jyxzpm?AFm^FpeBAN5fappCd4#u5I*lHJ`GJ~
zx%b6WFBWo*l5yt#IrK>{Q7V`FqftpiOeteYd-d^H$xV$*=il>pVGE}qpmpC%x(P}W
z6Q2wPCpuqjk6G8uVwcZ)Cfjc|6e)HyJnwV48SsCKoVS>6*ON=Db~<^^m+j;T52Ei2
z@9ACaI`eh$Unhq@;z>nA?(~97kB>~K$^X6jt(PET55J?9)oM7%|MjPZ#bq0{&Fv&d
zk%%3V-y5oHOz0n;4N%&c%b3sZ+#L_}B^;HkcOExct68`V!4e76#Qj=9{QWz<^YwhC
zS?UnM)RwYh)SOyfdCl^_6Olr=L~2SY7xKGAm2X)x=x&;sY=15-eyLHW$7iu-4;+c9
z(((l9<XBEfs&G7=U_3LBBrIJVmvz(^3b`$Ma$YyQbW%Zod-1H`cGIZhsbvb+REMH-
z(*5qNz#*5zcG*5OsF!9@)b91~{bYxL;VSH*ub1DRa~DV^cP*OWQ%k4KabYo<4whYx
z6aHj8{1fk_vw&nQ1zoss9C6-amRuBf_{w_ndD`%?ZE&z`z`>KLItm2}s9&FksgB?7
z)4#SlA9w_t)kNu{#ig=Fwq8dX)~I81p9NRl)sPf64hIuVv`5T#b34f*NM-2~lZ-8v
zVr?DO_88o#n$*zg&ei;K+<Z2&t7tGn)XDAg>KOBLs6w}{$kvNVW#4}N?S)VhlUKO1
z8JAzJB+emhVypPf6>n(3NS<hf``Y!OeS$@vsPb}t0%64b$o0T#t!SO>YxPX-O|s2r
zqL}qM)`J?DD3c`HfVXqMThhS&`rX~Os@8&it$3XUTsB1MNJB+WFAc0p6g#rZn>;-O
z=GMXvp443C5@r<Azk8P-EE1L1&DYdXDkUC<i-4fWA-ud$W&`Obap&0g#VI=R$_JMv
zaTk5ILx^;&7rb4uuD&<`u03yykl7LMr9OoDCw!uo4IN{Y`Hn=lblljonB7c@T~u5w
zvxm&>#0lGf7&u@!YciS*UPRqT@Jip#tEPonK}IIU!vi*Xb3@JY$yU&dbW>l`J#rb}
zAt8~3oLUU}c9UeH5IV9+W%+2we$p(#cQ_g|`UcLvwoO!<ta5tSiSFX^vWxFc(~@9u
z*!2WSDe!&Epnli&cwb3I=FP1$GyeXB4EPCx_%1IG4{Lalc?Z_-Z*E*y*OPDGVWXoH
zyAU{|8COoPDs>Ii!0O=dA1Q?O8HP7Dp6~Huw?UD<m)pcBK7HhkVtdi^EXKs5Q&u~I
zm}Me-1blB>7)FnVH^kQYSF6l?{3kYxah_bycVOPC&K+<cSgpGz4iz8PENuHX&Y2vS
zb!mq_Nj7}#vwqpM{fM-Sf%;zM(OmOBck=xJ4}0stK#|lb0)Ew_QRqesRsOqatwXrZ
zYgOpCL$bPp@-!vK@((6!YdeM7-w%8I|6l3fOuiZmlAvr1tKe`{UpaOZiVYlZUFnMs
zXTwd%D>47OB}VmL)qP-ba{r?dmTuIbl^2*W6ENp4y2a~OZB!H|#5(TAHZMQ>@4^2t
z8MMy_kH*rqvS@6f^M!3&(uve*Y})E<j6!qngd9I{MUk*M=L6^!`fFvmf9<_tG9fbH
zEnn6I(10V}nOHw}Zr*yEsfM$V_j16D#jRdXNX=dNr*10Styn8PJJWmjiim$rFko|q
zdkNz&vk^Ua?A=b!u~`lyCr($VPH|5Tjc{*jp(xz3w~9(^mbo#-GMH8BoH3CFmS+T`
zzI=E2vlP-@hOeD>A@XhM%E03PKNde4%jf@EjKzB$vF27LHZTxhRCG=-IB|^{!F3nc
zpUCWX#IrTlph27YKik^vLo@OyEMDi!(!w^gp>~3z#awRNHrvKJ1dnJ3vM0im$JwIG
z?OW5S^*&m&!aTauoea<ONFJaGFn2`;!^J|`X)!U#ThClBO`n(c@1Um;{#rQw?Ec;I
zH|NL1Mf+NXuSe-SD7c;U41-(j#Xo5%gb^+;!_rfHP*$;DXj>60=WO)ih{}pMR{gWM
zM{+##`QC|!y?jO5Sj-TB(Jb}}0fD-+U-8Rwd+8**QVG3H5+_-751FXR3rqtmT;&3A
zlJljnS0_FD-z`cQjHr!EE}RkmC|bP>?kfc4!8_&wUvGX2>7;VbC%NB@<GOo#<g~R@
zfT>r0Fg!9Q7FKR{wn*5kSFcFpgBrhGmU?<V5|srn_;;%nfK4MB*4Ws%12&(ZHu&NU
z@B-k?-UEhWx#{EO<)wA)Y*_lwpDzMNnoOLWoRFSA{n^MzROA*rvI#{1+nIFE^GR!9
zV+F8`Ov?p!Haameak<U9)b336$QETbd8Y9ZKMME|CP}%ltx-HC<+O|HCe!bh?<Gg%
zmR|cb6S+6JJP}#qffob+(!+|7S>p@(d5C~NxQ^Xp7)vKZfYvuyQwPue&GCaxJ*hFU
zq}dwI2x#+z2`ZZ(ul3dJH(Zm7k_ZQnj!5&FTM@(Tnt!|`iAYMK<KyE4{_)LQ?655l
z4x6<N;8_(;`rvbh%cTDUy#E3nRRAtJbP?{um**qQl6eBK4;nwCqB7;Oq(sA1D=i8h
zZW{xxSk=O!(EV~1TT~X<i;d;0ya&I`MwXS8l~-1#1o0wAjv{XvmJJ_F$de1MLnI71
zVi*4S@f5u5fb3P>zY7PpkfHnwE?TP#jvBft0H(u=q7o9Lz&7mjVhr%q&H*mg6%kKA
zXxKdX2&w&+5ET`Duo$PSUO_+Io{*l<g6e;zr4<AtreLurN<CPddA831^Lku9cUKgO
zM>0w=nK|j{q;AK3Y*Y0vmDV;km;?j`ggnk)I)A?P_wy^^B!&_3HzPM0#X_0jm@>OL
zMwn}eaop9-@c@hETpjvERz?ODFR$j|;o*aM`^~Kr4Z?-zalIDh_;2%RT&YJXg<m-n
zoo^01BIDvzm#)gonKOWKT@QWZ=<WG@(OUr@<%ky^^ow%OV9DoA@Ms&E@fHBqdT{D5
z{_&3EgBosbZ%;c(LN;=tyh2$SnV<gt{<aqfI<zXKty@1}CZZocc-$Q)-$%Gy52XpS
zI34NZ_@wU5)=ua$!4$7n!lfhP;&Q=^@LaBhZw;n?2Onp*){6t)E(k8!`Tq9YZZ%RV
zS2Ublu>(P7pyUZXEZL~+Q1<CUV=xh4A)pt<!{*%1HQL&sHY#>@Wgi{NH*Y?Gp&li=
zJ3tJcLwF-9CL;0^lqkzVrPI6PK5#OoQM790qH&ydxiz+%zd%X@SLB0&XEIY|we`0o
zjT9LDb#=$UTpt$z^xnh)totZ8cU0r>t>9CLGmU`{+^4YC%5N|t-MCC(>Oh(U%-;-x
zH)o;RN>VBcimY*KxC4!U)<+}Q9eIicVWb++?I~6koXkW<dV1&LS4t)(S#X8roTwhE
zy03kl&vvLprz$PAV)eqp!^uQ}y<TpGI?NS&p7!y=v0}b*YEi?Ln6a_(Lq_Jb-^t!C
ztd#;U3e$P5*a0sSP+D60uw)@oli_qf`~A5BU^EDP3C)pEP!uK#HCW947F&mm!!4*a
z+#HFE4)0dZ_q#l4>X3+I7VYZlat2H56!-^Hva%|=ySrPn<6mv%6_uM$vsM_7w3o@l
z!$wJ4R$d*#Tt|VyUD1T5OTo-Em&0EEdXh`<=?RRe6uNikxjAToz|0TFFzPYuwFfeF
z%z{N6$8M#&>jN*7#g2~}g!5h~?BuY6D3Aa96D+W@r@<{`RNJg?)$TXYSh;=qlD$`V
zECnpK1i@8cWxy|p1FFPh0O{GYA)@;W;nST-Ij}vzJ%cM06B8Q_#Cd<R9KyGCI-xb8
zb$5iF2c<pACPP3#2ulyAkTz)##KGmTPJi%A^!E17NlE$eV4rB%pOEt@04v9IqHt@j
z9t1}?WrN$R)39`~BIWiT@lw;$N;ZN~%GtKDu>l(*Hz5HZjNhQAKFk#b0W3X*I*ak@
zj{*WLUl%@r@exAfy-0a*gzW7Fs`2v27eAKL%w{YqozM3QT=wfHY}CMxu&7#mvbwgm
z^&O3OaHiVkw4ZBV857f^ZE1O#ERp9Ik;lGDiLUp<rEP#3{HUA8WVWUP>;vHQ7xD9F
z3Usqf1j1ILVP9N^P!O&P0rm%;4$*N~%JaT>w)0m_wf4KE>?j^n5IP~@UNyJt#Q{^t
zb6jSTFx!PDIB>SKDxd=`M5U#bgoJ!{?71jp9=XZ@uW0anRHAXbZZ%-#75%#j1_xzk
zU61=}z|2M@Cl`UqQBA<|pig-)^L~{qxVh35jzs|{qt<_mYXD+q4?YGi?pC$B>qk~t
zgFzeQj~SM()8bRoK)&!D6`%c$sL6PKEHMd55B^9a1c@3ATsP;0z($#ofU9{&gWEN!
zUy(c<FDV0qOkW&JIB6odmR?3w-?UaoSVP;YwK1@n3rlbQj;NOK!~@g6v03l6P_>@F
zrA_T@Mc6YB%QSFVO!|g~Cg4)S($99GY4E}i=G2<fWmQd=1+A@uqO$R<<}3RHo6lid
zWne*LVq)f~lo|dLHc?^7uac8VXlV;(_?{N2lnI56mKyXR1!%*vQNiYQl~q)XTI)~D
zyS+ZoOG#<6aRWp4FwuOso6=$FG(~l?lJTq+pe{px{+i(fZ~|3@$_#_n29k?_r=+M+
z$hb<qbA{nR60IniRMV3t2iLpH^|18MX+kOqk3AHA{rWYUt3V?<3|4Sqy~{Zer;Bo4
zJ1xaqHVh?j*pMEo!ll^1CJ$(a9q?yKCy#qQ!=T0=X>fC?zgP-zMZi1Y_eCQb2PsPY
zUB}nL+P#_pG&9Hpt-HH>_~LMxOcaDpBvjPj>@Q(?3A}EqXgnuBL6qnwc(2_AlT74{
zByofrWe4-XOY{D{-1d0E2=KA2u$<>zU0=@wp>}IDC;SNtZaPh2l_#5?2U<^0&&Y6w
z`01>}{N`Y)fd7vlG9iRK%(|^lE4~c<*>Z%vD|S3&)a^^+Q3K;Op%=nqXonBOvf)e=
z1KDtEEDvlYYTuSru)gi~W>nbb9Mh(Lo0~>+jlg+BqKIgOz~(M`)?(spce)^W`S`cl
z<kF>EdRCSi!NWn~*_aQ}H02`*I->#f4cS#y><=EP&CShVCO_Zb-JTw_pwWOFhlq;*
z(>P91xWF#cxO>wU{8e-g-q|sers6Q@jxdo}zXjQy`<k4LY_)EA;MVP8>G<V?Z%jki
z{1*706pQZu{_*V3Y2~h$N3xX$;M8L<I;P{`RMY)B5>eG+?QiYD1ZvZa+ulo_e^;$l
z64EWG=*$A>AHXGLkSFj3ijY-S&ep7VHU@E5QTU&dTHfQI|2Vq4-goJq1se@K8kzSH
zn&Mc^sUh$uHj@er4L}%QuFrN=OZELgr}UlFKPQMc`2PFec05Z*ItmJkJYY~sX2JGc
z0IWA8;4=zl&aQ{Ha@1ZBTmj63y$ctRx<S8w5fiX~{SY=<W;7tXYM}t4uN%nkHM_s3
zUtwUREi7n*sn&dmTw!5h)X<E5PZVtuhDAnJmUx)`?o>vOT-Mf654{oy$yX<vX_fQt
zYQPqCy)P^>GQ*%J${4JU8gLI_H)TA0mie?|IJu-OFu9#5xK^FLX5-<}kmfm!05hBi
z{@-qQN`b!qBupxq*NxvvVniOiP$-13AUm59L{^y=$QMn?)wRy3ei6^2iXT|SqzQ+d
znK`U<zMR~(x7+zUWTOFaD9@!ol;pgXb8MKsrT1{NTb#iQz?FvVui5?WwI%3AR`k{C
zccBLn@~A$9kW-Mv#Y9DiKoHIXG3fyTf%TH{oA2C9&7x8bjLf*(`K*#|uMdQRSFY3Y
zgy*>Ty=tw!xu55OI!G8G)qakx+?^~Fb~)P_0#SPmIP<#Qo=za4p%wKC6e^5WLy{w-
zU{kQzZVaO9{kEJh1g^l_tI=BHnUe9NsRF*%#{i#i{GPUW*gE3sjgE)9w3>Cw;CTHB
zVCuVS%L)D}$jd|Wy6RR@y5OQLDr;Nq*14!O6o?*H)B*}ffM7{!mrXu6`HBw5v6$K(
zEHqOOa5(I#qu{aT%U@oo1z^*US4?QC<#t{dILy0M%2&G-zBlU6b~)ec?(EE#pVP%-
zH5+XD2ETn1S~pe?@qtV&+qO{+kjoQPeC6B-h}JMJQz-ED=iC{d34n~4&HgfgFrXBZ
zQJQ=8kMpJ_z?U$XnA4&+Hh1)Nxp5jrlH<pCHcMX1Z^}`?J6I9`i7+r;ByPYNlhgI0
z7+6Vw9C&ne6p@gS57ND8*hAz5s6>%{z1(;N*vn@<0QZy45l}%u)RdHiV92<|#Q*}H
zwHe@9?PmnnaC<#}KMsZw*z~S*d}QY0QhT@&mI-x|O|zHP)~nqC?lTn`<Y)V!J}5Xi
z_}bRd;c`|4TsFXWVdCM@{DZR_ZqKHfDy2n42K_N<NE3NP=c^mD{`~oPd$Am{HBpqT
zSf~y?;J3)ByJ2zqNPwF+vy!<dZcj#}nIy|v2f)>u>@Ys8H_evEV3W6ygmCMMQJ-R9
zVCYH3Ff`w6q=odJoSYo@zqeWvlncXo`BEndz_%GIy67H!ZpUJ(E;A4ca;C$>!^Kzr
zf&d<S{0a*zPNT;5Fp$gfV8NRY+L&M2oUSaQ>Z&^(SL?wq18{2K)(rSc%bwc+Bs_o*
zqW9n@OD&A+35&}3*w|QuejBhm2C5K=T#I`d+pY_K-7Z;I_kyrEP*r+%*q_M7Q@KA=
zT{hF9h!OG-gATCpA7-%oWG&v38u(U!M?|BuaszN+tuH<cI$Z9^d&r6b0dgiLCb2eP
z767o24m$xbFsgR1z}DXQhf%Vd;_#%M9u4Q`DW3BgxxKD=|B%<TX8?Rfb4h<PV~r<t
z-gx`}YP$MpsM9?@?r!&wcDH+4?>f?+^|I^j)HZ~$WpmuAOe3$2L9&Rvzon#~OL6Vq
zYqu$E?royWFa~o786lUo@;0m}q?srUX^4rIWoAE5o&6o>%wT@=%<p-g@8@lXFJl<J
zgDq^_+=YV8{m~Qjfb(-@p;8?f7O?$cZez?7_RH1DNzR~$IHWtzyU;qIcVI7t0!!HI
z<?X%m-KKZ{B2|NSKwQ(Ze7M@L)@6Kr9P)igC~B*m>ddjA(?|B2?%Cr}TU#4iP^_=7
z&szyJN`0zvIXm0j&TdSpsi#-b-hKecfv>Kvu5*n}7dKMHV$-8P#Yrc-Y-6RIg(i$W
z2XFHTei9^1S-=>l)r~1qo+LjC;7V#9vHkq~rqbCZ{ME(SBbzcJ6y+*v{`FKRukTs+
z5285U*qGntOP9Rt_;r=b?@MGlHW?GmdJE`?XwT3v*nKiXwITnykM3T^=f51-CmR70
znK29;k}C2H_w!*R2UFw);mmSRidnAAT)chDWu>mqS{Ud5p8>vY^u!l7Hr^;<tHgdV
z-mLwt6&EbaxMKCAok2SyBh}eCIlfKN(3ZwMGYR?4ipl9FAhWh4L@yitT}%JHW1o**
z%~yZExQ^}M&!@LwbMV4je_yz}{#d3mos05V#crnx^n5*Mt_SOR8z`r8i~_yH{=O%~
zR9{F@SoM=aZ6#@vkW92Mt%94YX?EUP4YoCjL0{-qjtbp?zb$&PMg8+HzZ9gQ<<rip
z-B)3<cur}?jB0j+d9we~MDxLe2iKow{jDAwJ2X7JZziEfk&^b6CA)q$bM+tI6fVcd
z)ywOee0lBSF>XCp$DoJo+TXT9pwL(*uCA`UK7#4dMDKD`@QjrT=dV@2J=+>9YfHZ?
zPgr6LyNDSC)iZ`S)|AQkoRJXo0gAjj5M~koJlx&VkagqTKo9utPKCObeypi6=q{hs
zI_gfMw#bIO1yXGMen7%j@&pK+!;?B%T12xkxKc`;ZenWc)0ZxR$bsZGS2luY-Fc4V
zO4&W9@rUC15gkz{y8u1a=TH@*Q^Q5xp1<EQ@95|N4<BO;yBVdI-lSCG{&+w=X%&7s
z@kZKT8vjS(bGAW$C&TY7A&ge`X5opysCBjIQ%%kMD_2U8<qS?#@*~WFfJhI>4uY&y
zjK@5`f2Hijd6;9z!e`Ud^w3@{xXTKnRa^UJhUXrX=U=;4ikl!-E%?SaeYwYs_xXVU
z>%xm0OKvNQ$G$Ls+|o0YsI_q<w~ZtcgER1A{<*oih?|s-jhbf6(~|$%qt)}K>y7#<
zM?;6CAK&RT(d(bX^50!u;KjAj!5NsRGyl)r!jqA;$2T^quTNHU#C09#p#Gf-Az*7@
z$sQg(iK}e?9uXWZ<G5|iWCf<gE1rsvkDRC%WM*YK_34g9*<Q}c5&DV(+099UaQxYK
ztFJF7-c=xGH%18{S=yDd*w&Q-;#rD|cj;#%*nPgZC_+80yc|7p%yH?`HX~<FaNC!j
z{0b!6XQk%=_Z^OjNpr)lFaum;WM6YP)f?f0Xhq6h`TbutDRl3%wq_#m=5^~p()3(4
z6Czym<R2CEFIbe|E-BF7X*^Z8=U;CwmixKCl2tY@G|$sd5A9+c4-PJYc$Tj@;h9O?
z!)}PopPrr`fnH#62Yl9}BcP)k&MsIo*licYn2=Dt>~JGTbXT`ox5I=`v{*!%G+O1W
zTr=KK)rs4^t(z#ThrRh0+S|8#SiRkcm2Cwz%t)oh2R|@@3WS3Vm8ys+{I0X^!!27P
zpaa)87x5p^s&s`-A3Mti#g%DlpM0d!jZr5AGkYYXwIOa+L2Z+X-=^ZY{!HTJ(&D$|
z94F?CQkn$8UY}C=5UrhjjLFPPc{Nso(5c%04wGnZfOisz2#DGM1P68f`r~Kc(356a
zHjyK<Es23?iiY0)en6@4jfmPj0&hKvQ?vlZN36w%5L66jXna72VFuAwcZfgk9J;A`
zWA+Jb)9tW<VM;gXLJ2e<T587_>`$q}wL9CK9x9$xWMySFpzF33r-?-#h=|6^%d6Bm
z9lh1>Un!zTa639X8_q6Dzn7l%5lA-u4I!K%-4g&jhmjuE_*ApY!1ZJxSi}cowIR?o
za{JE$qoSg&D@mCFVMi$h3H7JCwnFQbvkN4c1KP`vIxU@}0DQ&U9=DvkEjNzI*=Beg
zp`;b_QYC7s&>?Ad7lTOkn6|Yr<bgOxnqva-O=Ha?1))O=3{r3B9&|VF#?LZzVO7xw
ztifNLP^MD7y*Doh>UK`Ss~{kv?CtGAaV)vyMgw#xV;HogrLP|me+s9z?<av5Hb(%K
zg?C|@L1HrPJSEzq^<Uq6=1Xldq&n_H@)*nQAX;cpLgw$iM~jXdYK>Mr@d;{T$G(^m
zGS#C(-!mA?A@Rk(kUWmwJA`&o$Yy|B-UyIKJ;=p~!hYgMunh_HlOt{k8Rz@Ev=%Kw
r1>_I}*Q;@pN4nk@#`S3YQ)gW*_V^)@^WEpR<o66$=AjywlbL@2Xq4*@

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 6f29b40158c760098ae4ee4ffe657563465459c0..5946ff1cac6ead8e55ca154d85520e3e7ed5b2d3 100644
GIT binary patch
literal 124119
zcmdSAbySq!7d|=$s37oBL8Mi>m2MQ2?vz%Vp}Rv7L6LAM>CT~R=nw$`0f}K~L~?+k
zJAY@4?_KNuch|c2kLy}4#&LM(J^Spl_p_hpdEcujNfY3a<3S)00$G`7Y7hvH2Ly8E
z)pcC(j=|+2L+}IFSYG-W_zC_TN!I&=A2%Fibetg&s$lHjOA!lOW)R4Ii0rec8Xie&
zQ|=xbuHzkB-f|3$QEwyP{_x%~`0(a>$RJzZC<<9sUmRoL%?W=KL~@G+ZJ^<W-*|9W
z!t(VK39rPi?<dz5NS?}Y-H4KHN?jdC_9X4hr*w6hE~aodn>Ywi5?=a!7ZAu~_yF?X
zYY4q$EB@d2?LNFA{r3@K9L~QlYnbt{>%jhTP$zEq^9u4ahsopbHChx^H!uHrJ0LOR
zC(hqb@IU{5a$$4rYYzN`!IBjg{cf%|qaOyhJxJ_ji0Y!t{B$6~Qt{=B#2Cg+7B*%R
z+xpVM@nH?D*F38AWn`DBlPht}kE7;WFjF3KL24Gk`l({@;d7}ce2I$Sda(J~+1Xz_
z4!m$=xAhJELy8Ca`qV^zets)RcpTgW!bFSPd1S%BbUU~$z1nu7P6)eo5Xh5HUpqRW
z<CRv$;g`}rI^f#h5Z(FCO)lb53O7k39>usN3pz`It8DM?y0}u-sAy@$Ztv{S%U-&4
zDe04Xmzk-ljF2~5+IG2a5otJ%va<5g(NV3yEq!KjaU&fa9Xvup4tERl*>O2}`Fe*O
zLrcrN(4iKjn!5S|2E%S|Z?7%=8h>Gao{67d8`eE#-4h{pEJML>=i+)3VixiS@%Fp9
zbFEj4R2K+}4%=N%FRyLxqWMCt)u0GHFE1~(O)aM-8X^z<yvdJJnQV7M>9*W2!?NDI
zxePvu{m;9h3lgn8bC*5XFQM0*?EQW8nUp)P$SYhTq3Kune{);^Kg*Z@pIo>Cex-&L
z<Z^J^?8-eRoF)J8Q-v*QN+N-XxiqJfo&MChbf-DND;&Wlmu^Sk*Bq7w%w1}!l4{kH
z7#katla+OJ{djvvzxFHgRyT>V?1=M)<ieuodg{^Ad2n}+!+NR?hvMt)pQm$gB*8Vk
zMdoXSuIl#{Uid#~pS#Qk!)q^78CY;ftIpe2l$S@xX!?bRG&G2i+`S9(Jb>#N_!>h)
z!|j6uch`@wo!y3GzL>XdvyoM4Pt1ubK1xNw=URenwBT!Q^>kfrZIZaO@KFtOzSS{=
z&v$BS4yDW5PNSIe6yducrD!`P=`9eX-R4zYZN{HkU)CHG=9S{%;bl2nEa`X{8lXvt
ziNBe0f#bS?x_zVtccYXW-ijt(SR`e7C?K<!kwHvG{%gf@;mra|$;HmjPQMtR%=Z3%
z?3lja%3UHNA~H|gD;kyyOSd|EKlhq@)@?=AO7Qz{t0qu7ebc>97RC@?YsA7^-?%B!
zYRyAFbuKHEq;6Ru3t}2GNz?E8IN>8DQA`qM@&XhUGO+G=9fEiGG|h{SmvDp*7}64{
z@e}uHeq4*nMI-hoqT>_2yHI)~`9-%HOHh*hQ%wDGr&76SaBKMZ_?*NI3q_od;uRVm
z-_UUQn}-@)I<=l&r=bJKuV#&9VYyvdhP=ia+)OHB<`qh&OO}|eJL5k04HJpJt`1po
z^X}K?B8znHkTPC_oRa;mv@zedqBZusJ+rDADb+qlAPm=smE8GlnP$Y3(^n%aAEUDj
zjcT=a?%YP61jX$wL|y#w^|f<Qbbz#&m0kZ+!egG<`MtI~I#?Y~agZ@E$*5d-J?oXF
zSs~wTfL?oUG4V{wa{ou--t!G=SX+9q_>n?$(ToEo@CxMo_P?9>DsR$hHFSOjYE)FP
zNJx>VV!!babMP{hf!uj$&q;ui{Jif(C?>_QL-rEn`QHlvg#N$O=q%2q(}&`r!3@(*
zQ6iz9uWE$Jk*Kp?3RY2J>5X%d+K+0NAs<5jG?)OnpNX~2-C8g57$+(<vQq8en}1xJ
zG@LfC==gr3SbKz%o%A0><@UeWUUt?hiJ_aCs(kg{LSt`7b0eXwyQ?4di(OBSl^oJ?
z<KO2T$OHw)-Pb-tac~NxUIwUFc#VX7qO%>XZ|2Fk1W9`Kryo7RiHlBPk^Mlp>FlLd
z**x={`x2!1-%4<_EfMlC1>7ZTNM`GO*>Yh@rT)thfr!8NjmN=7_YCI3#@&O~k<}R^
z?Y#=Ax%)SCB>IshB?ZC3Rz@rF_&Bz*?BY%uu;D_j-oX%Dh{L}ClO(miRJkC;(97O&
z$FOfDaeE#mkowC|n@8DMPL)^w0{8jk?|ss5{!rxae)(uQu!<rkHq|JDjHY-h?c?cX
zf{iiBOJZXGw;(B6kNnt75sgkr%spA`(|+}s6IOkHwp2_9p<{TLD6GNTW9;o!2r<R~
zcA7CltWXH~aL@YmLD=V=dZ)r_u+u%SuTDL-PFW)iZt_}j)FdKeoy&p*Wc+RGX|U`2
zJ41#1=S3^CuM+AknuWqs^Ttkyy3a{1cBp&556<}O6+ReRVu805pf(!y-1k;I!_==p
ziodR{{DB2O$Mb2Zig;7~GiEn@(&+W7p`#y-G?4w;jn-P)VJ%ONJU+W}2}@gt(_8hH
z(FxMw93L;kbE#23><kR3Xds)6|2CUz_UY4vi}l1~GmKZGxMzXZ#(2@zgu79>0rKa|
zDPKwDGw0F&f(}pW?=wSwLjQu5ytg-T1CQM;bI+;9nFZMu;qI6LY#Ewgf<qljS2~};
zHK&BQ$Zr;h<3VaJ|HVBlwVP@83D$;Nk&ht|`v2_?!lOq^QNLW_eOFg^52wmAnbPs$
zBDgf}WEO;mM!WM0ME}yiP^-M|&0&=C&8E4_rq+b?>O$p~gbek}_?8?$9ynf4RfgL(
zdU43yfYgxx`^tFU`h&cF{1PT%PG#8eeV%97@bso(!xIx7Q{>vIY^nO|7DmwLSu(Z1
zR9neLzT<^YM|;;Lh`-do^x<+nlAAi}^DdvDf*SDZZ|pB;6hBsto8sf;RHi8$%j@T}
z8Hm6qweFv#)!mzk0QjlI5ejKBn!QDa-KzjO<IePFQ+w;zzy<1>Jx$oj#O67B3}o@P
zwg-i$gjZ3o5KN;z>J<AJl);9d1Hm1z&+q;n@Qj%rS~FI*;UBoJyame2Itn=tF{mQ1
z=p2F-&WeG&t{yGRt=J;7Bw~n)#=kASJV(klST@A`VRLUFj(;DP{`tiHLvaqNP<0HQ
z=;1(aHVvo>5C8Y+4doOT*<<ZazU>N+Xv>Z$TRo<4J?g<x%FMH0y9PgAgD^e#7fu0y
z3eTh?y6#(9271w_%-NUa8DSD&S<1P+OxZ3Fh=s}@1gp|Yr680qQE#v(DM`FIhZ!#1
zAvK$zq%k5NfIyV~9fCI#zK5lB`3b2dLlqDEdj(=**LT?<0XyG|3I8PddBW7}kNip(
zvYTo#El<b@+8ad@<~{9t4d7GjE-7!mNB|&f>g077(&G7dHPPJCO35Sc&HboJ-qlLX
z%QrUuOqzPb#n~WHSI;TD;kuSGpB8zGbR=QXpHpUis=q79s(Cm>4pl%PeVKnScUsan
zR;C~R9T!>s7JXVxNEWbD`)ZZ`rIdu|7v`~iYZD~|C;O|7@1yqS(_WK%9ysWH*Y@+I
zA-nxhnTUw0@7wVCFb|S|*;BCo(d?QN5ei|W^NS1;F!*;ucp?}XolwwP(wOb)z2Dtq
zGcttHwepz5g;L%v*a_tlptF@1%T9G$8$cx@R;S*pR~mOJDSGbF2VFC$t+drdRD&a$
z={VuOQnV@*$29lpqXN5t%M0AVrx65eptylAYJWcRJAe*l-zW8_4u8RQ5f&xmQc{+k
zlJ2X?>&eJ@qJ?5QGCE!&qM;Mfdgw%;d%oGT&%fN()SBZiijq!GS^Xj+&i6}i!kR{F
zG=t@8G}^u`z|Ghz;l6>#c$-_y12-X0(0})_rJed{*BK)cowr&bZQV*0NP^jLNBSCH
zxl*~@_h987Nq+COUwaY#(ngLaZUig&27$-j2oh|)iq?F|uGR5a&uw|hz}t+6e0M0c
zpVZn2@?<Ri&sMYqVat)Md>rdwzqUBh8Y^c-{<o83kI%eEBYt>0TO)@!p_Nt_uUYLb
zPQ2}zf1tD{V9M}=uV|szcvs}cY~=3jc$k;*`EA)j#t4zZ(U8rD7FQwB?0@rFe;I_A
zPL{U5Qac2pU3NeOuGo`2Zosoza_H%4wP!)oGt;S;s6+>EM~I`7k*lskj0Z4VTeteV
z)Ix=;en}bHu|=?(gF}J#NRunqh=pk(X=@Ijcz*|9pN^GZ8K2KiEg<)K8wEepR?@9r
za&&vP<FV}-?g9z%w;to89^2LD(#%`A8_2m@(n{X*G(){IsgjZxnIoTW72LXa-@ZbJ
zvBJ4h44u_CzaR^1V52@>%x845D50kByAENZ_?z`L-|9rYHD6Ysb9AfB9+UW5y%XQz
zOGJwD={9%J40iZl8?KejWVNIDIhPrhDL(W?QEzuQ&d7RZz+^nqMU4WcWLRlX+MN2%
zbu~ws+V{m>Zj#H8XlW6lE7*nneAua`MkiYpcqKVSQ+Rpp<6J)T^Xa8yE%r*^?{Vd0
z7~Lk@dN)RDEibq5m1mmjOC4dvn)JhHeMmt0UpNpDYPyhOa593LD~Z!!V!3ec;W9Ne
z7oCDXrU+XZ%c%9Lwp@21gtS!sFL6JCRw1kAq!d0>p_Gz%>lWoawlbWg{Y{JKE6=56
zVSMLSm<gCbv!7jZ9u$i&YKZ=AY?f@;i|5+;8GEvfu5`2&U!8yTXIblQTOq?gYBc8M
zsVRWyZTxL!4z~hZ(`FmJWMI>}RmESOr}e9}U_1@lL}WMLsdWL&LG&-oKPn`zfh3cj
zv7gAEA*5#6KX=}z8SWQ8VBu*oQYL81-o0`e(#`wtCYZj?7#WG2=TXAe<9F({C<V3V
zeMuRq2Yj2q$I*rP{i4C1q|9hY!1h1FfV4hS+P#tM)Xi+3G;bp--pgXo%S^Axzt&gl
zwWfVX5i+2w0vsCT>cMnTtAVgl+gI?*U;bf&C&&Gw?)kcwc`5hyVz8N%$!~{b--E)v
zZ9Zn4oy|K4xbSex%M;NcKQ?a3b@j%y+l-);&m|A`o2HRlmmnMxe_Nj5mP2z>U9qQn
z80A4=PEp9vXLey8^PX_|YVLbC>x#Vj8~J|OA@lE|-vCUr{pN44a?0{^sbpx}&VZ?@
z^V(y)Y~PFLWesbPnW~sthUkJRzz;LQBEs|ysd@>r0QuX~7gS7TLa3J~I<cT&f%5OW
zc@BdW6)}-65?xm4?VWu!jTa=EMEHQHaB$^;FlqkXk2HzWahqem)`w@6mhVCFik_OS
zSzaxhoh2Ww4NM_-fddp-%R~=}=J>lKKlioPNArWl&P(=bggImje$5q)Xtt58&t(?A
zx!j(cZLQ~2E|&OM&#3=Qv!8&f@RR!tbJW1{uX)H%%72#_={V;zityu;tNtQkpe?Ff
z;$oMpG=MZc9&)<b$yKq7T(mtlZqsVyeU*Xj6tX1^Oyc&Or1aMyOIv&-f6B5R0OH6Q
zu`_)qf!#5Jsl5pdqQF{Vm>v}6=SL^#FuUZ+^=V-ZPP!M(_M>^XHvB!h)CeIzmHsAd
zfS!<>(DQ?agLF#5;EuhGqF_BfO;wI`Eq*!~`CuYdfNy$rtE5tuQgs$}E~ORzZORbD
zZb!r&FqFYNb@@?|xUQj9_92KkhAF40kGO%;=T;egre-m!au6oHzX5K+WSN^oCr^X$
zHY=AGlrVBUnK~XMQ&U;#gXI;x>jyzt=v++jFEnm`dd;dJEK~Z3YKf#8nQ`A<a#`47
zp^m!l0OV)F-*y49vzSV0gk`N-Yq&#|CZL)N_GOls^+qR$GG2&X$e(0_)FjU_=fko_
zpZMo4e!htd!ooZQ)zo!!vwC(V!1~{LsY$+^s0o{Y<<c}>`nj*~8$vfZ1qXsk82ii1
zk)YPwq<Ose$k}JAB|#aBj*81IO}meb2p?3$jIVT5+s-e2J<iW-cql1BCo8Mdm+I;*
z!m7uRsru^mYfw=Fti%4?kIKqErC>o^6IJdQoU_pZ*_w4nx86IVgY!k!4~T!GN2jgV
z?&5j*XV3Q^GOYD!wP3{VyEVAOMVh|^Pt^LCU6hK+Q6Vw;#C^E9;F`@T*gObO<MkZF
z6WyF;>}F?D$x1C*w;qx4g(6ijSbXZiB*-Ye8l?n1PB}}c<uvT~JrnN@(aUBoY03h{
zQM)rZaT(Q=oI-w9m4gvYL)E)Rx0|vIBz4sv$ir}5EVUE1K+OO4(dY!n9nzK72(g%_
zgP_5o$M%aB<{Y==`^jHl0(~YeonQrel0W3N85i_nqD&qoy3_FM{N$6wX>z^gU6$Yk
zP1<3&89w`X!}A&2iR|Vo8{IWB=Gxk0n>O?Kk5W_4ZF^>sw;|ossiMCVJo-ab{lW6a
z#hE@s;L|8OOs;l)Wpmc#Sw+q(%}S5MmwWjn@UzbF)o{dg6gaa|?`PMioBZF>p|i6c
z$Ng4Z%J!?qFm>+hKA;RPV^N!-i-yj#s{vfne+rZX78jZr`7pzaR5lLQ3KB!h7z#>)
zgZ&zgq<F6r3}p1`@mQhDxfDKp$Z~D0BlkolcUSAW?nIA`yF9FwIazpNTW7!S!!9FG
z_zr2ZyB>HD<ERE<68@X^xxI!}QhJ16;Q$kzPZ#Kv9ULHD>~`%<(aByMoihpwgpxkr
z?KW_?e73QQogz~1<e+<NT)WwrYSyX=Wt(XfG*C1Q*4-&*A8x=i$V$RA(n>a}2nZnA
zA-sRcUZunan_4<_EupO`&pmLeq-{Q8;hC=*jE^Qn8(!xs$Ni9nDYG^0LUbn!j8oS0
zH8i{Cle)fjeo=OLss=^0w8XQkncqRxN|i;~Un62ZSE!@Qw;p0hINZknaDDNMA7t|H
zgxbOk=-VBBJ6OA^+C~C0IRV*h$6ll?lt067lkxNUC$gKkqT)6oxbB!jl>g^6H0v)O
z6Qc(6wa{kMiMy`Q)%J;CE0s7UuISBP&@`3*)*z*>Z(XXoTGcB-S9@lR9iduLEa|YU
zuThNhwB#)D<DWhqpC4KLW>*G_CR~2LfWe0PzV%N?BPAq<0b_#<32|^$_`gbdZkDO3
z98992F)2l}y;Be7B1GGU*2!z|5fOD&udfE|F$8s$ROdG!5uhKPQHDjw#%=5t4o`~6
zs)aN+WAS_5&42fHV|y>A8G#_o&SgcJJ3~j-qcvpJ6g*yl+5t)X@)za{w7-0mrwzZ`
zT|ojJNT*fmJzjY7h7SEsiybTh0wJOP(~DZHB<tKh(fL-oWm3{7l`X!2{R*enc8Otu
z>NOjpgTMtZ|GS{A41xptLR1PKL67dFst;}SFYe9BCFO{sYxF`<aglk};z6FMZj#=?
z9+ICt6<t~V)ZWje0B}ERde%co!@k{w6BOj-F)d*$<|&7v;QJ9Wjf#5(<Pes-e`@TL
z{N;g++G+p8OC-d13ufkZ$g3RTHf|>*2eJzND$QLLJZAlV&x77USH8uW^2Hac8UvLg
z^?%i{Q$5h`Q(SvSu9Sa~1t~iUJ7CPlD!Q5!y=ce9{bs|%Qjzuqz(MGE$`$fv-?&1z
z`h=0f^!Snp4bY|G_CTGm3;J6S3#iflQM#z2On<+fB5Ye-y33Ao4y6mE<E+rB<(mky
z^%W7_JN>D*cWNs+zq+tc_X7g?r-3}lx7KS;iS;=w#^QS&_=x3LHAYTOQB%t<#kfE!
z&rT!yxXK=`C*-c-p%Mw7yA1hR{daTCzJP60pOab{uK5Jdgg{hYIsO6MdGm#QiFoIw
zHgD>k1t`1wdZOB99bAbK>Xryr{^k}ezG!bD`>ngEbY)pstS-^_@L25W+D;;ja=E$a
zWFMTLLsfHQj}i7I!rpG$F)a`jF8jM0s(K!?BRuxe8l7OJ6b&#Af*}v^IYR_RdR8(h
zj+ODsGC+pFGh<nC_!4sfA+d(Eu5{{DN|q6&t#3;QBnC3#y;k!+#>Jw4y?{gqiN!zu
zum6OR<p%gjzxnsUi8~A|yxj6IQ$BJqJ~0ABdPiqWx4{#B*M7R-;^T#CKLJIdKOq$x
zk@Jx+x`uVDH06Ao0#xLMFaG%Z+ZvTdX==_pl^3Iq7;!iKOpAWagT>z6t{EDwp#)z2
zScFLYEri?!Ik(oeC?y#f==WcI8W!ee)1$`5ber8%M>eK%`TdV|i1sm={Ok47t$uzS
zN(7L@itQ6^a2P<Cp<GT*F8$s#(DejYE}6PaN8d2ZL78Neg040Xw>`1Msp6_m<>lOD
z@7=R%19HLjYwXhvno)+Hs_B=`iyJS+J*KsbS@*p#F<V;5H&;nZfljEMtqyBMUtGjR
z0tzs}`~K6^;+d>$Jw*P{o&M!|f3eI5Q*||uYOP7eF<x8#Gw3-hZaP^}6O&zpmLih1
zy6%*X8b%kKMU|^$y|R8r-e{HBlpw4e%4a|D5hYddRG|`o2r#lA|CHRcHx=VH<h!Z-
zyzae1NIbmZ1aEtk`<WNO#=AjI+CtuZ_Djo9@~PZhzA172F!)<2dc?L4z3^$l&S1oI
zb;Np<VgpMeVc}j8(M{BhE?;BbWK5QfxRK~^#`|g!hd3!WblT*b%Twng^$Atbuuxe2
zbcIpaK<AJ-mQNYN-9t4@*bTJQRdGhw!%QQGc?v%mu8yeI(=<u@TBClPn{*;~YW>mQ
z@`sA?i6Avs|I{B12Tng>C3%FMdQ_=OQG><jgy3KU*VN<mR$VXDe^n(~niK~Zlz7uU
zUjo=&MISZTyXLbdEH2!ye+q7vGqlU4-B;wJRO4`dQ4`G%Y1r>g1fw77p<V5E+n4rZ
z;}Oar4{xin&GZ%R+(pfUd;p8{1CLkN83<yMVv7M$!L0XcmRANJ!i4vy4CS@#Q;Qa0
z*%2_?f`w^^#A$yz*h9SF!H(qzdjkhMq|mmeGqcZ?%Z6mahViwvUO5J=Q)4wx%k~?K
zUWuN;B`H86ptH;7x*1*y#w5q(lAiN}scUwzsep!Df8JLSH&AL1onHOv)_CPbKu539
z3$4eOAnkv-XOdCxJGps3>`=}aJyyO)X`?67hsVpvK0pWCpYJ+bb+T5AN{2(cw@I*f
z(~~?AW7*LxRILd>mU&0fNT1~U?4;7u<9*bTkSBv;;3`V>VDwi>ruF0aUDXmhSQdg&
zA-Zjr2@?Vdxcz7548ZG#R#@#9@NPdM-$TUq%86J!8+r$Dv1MyVZ+CYuYx~=FerqID
zcJN9sc@L^>jz!NXB_dddsLGN>HsKbcZn!+go%@%TEG)%xqWmlOxF)>74Oy__u+3~M
z_MZ^|$PlG}-vpg#y_#)i0e4p5#qrXW6X`VO%Gb8}xLu@GG}c&8eCO^AjD3PmmLHiv
zJDk*1Wu26WKz2P_e*ggTWOy>(-oDnvx9hp)Q!eXGdX=?KlkO0e=|9r)6CBrFmk)IL
zc38%1<=s^yo*~WfWHENxf;;-VdnQ17@OAPte2yjdL0uYeeGNEVrCuR_HjUU5LP)pp
z-|bHW(;2<&>y#9wemntpTnAmP*rAFf$8$a$0Y|BFx+7f?Q?7f_Zg${G@lmX@O37Wm
z(??|JnE9qF0o#3l#??H5!SZrm|2YO*WtkuEhq<>OC2I20y`UoJ`Qh4|*j@2<)$B4;
z{ASt{EDg{{Z`4yZUP-qV!j~+grRG*mNtj<y5|JSN7JW>PT(GD;JlGqqee);-0Bkp@
z7juLnS^vBMDv{y2St|g5KlJgH6IoKQOTnWe47ygkzpu}#*=O)_<FtQ(EVsPBFOThj
z3KH%98v{Cw1Au?re@KHx=wo>Job_P<Pn?UTrJj3E{D*ULWH9&k$$|zeml~{cVLl)K
z0;8c+Jgrk@2;gzPej=v)L`Q*vz<%-xwp=jzH;0-J7wA@fR5#MWe3P*a6VK~qi~?Pk
zoIa(3{%8{O-UWlH)%p_nW_}5iEy`>WZBo_MnK|#;5_D3mYiiYg1tO6C<!(b)S62eR
zeac9QQOW$gX;4s5zVlLF-RwIGVruI372ueFkaVD$PYXbMrw7RRsbNZ8{!;_ju>2gT
zpI`YnMkwi1`t#s$%coynVVQpHARA@qtBT^aURhaw5~R&vkN|zIg{;oHBPSz|Djv$m
zf@CuwI8NA%U}*$^QjIU4nneQLM!qM4a;V|(sxQJemtlPz&S}4vy=wi{a^aH+9>Jz|
z;_;roSE8NGgzk`bht)_^pb|*~_d&Zn?6}%mPp+$*KWZRvx~PBJ9msI5@{Qy43F^xZ
zmiy;{u)cFP2&7nh7S9kE39N>4tWFQNNY4&3r6Zt&S+WT{R<Io9RE~<;FH?YS4zOO&
zw(mrrb*Y>2k?*PHR(ll~u;RSnFqqjI(CP)mk2<Y(;RxVo3V;N#988)rgq1%c$@4@&
z$6j9^FtAtroOj}<lUCIUn~&VZs&iwtjbAt`zP1LkColnif1_wRt$)ORq|w=5J8l`u
z;v&GaR%gdWX5h^nmniJ^Lp`&3L|OP?xn0RtQ?#Xdlsnwa-0F0HYP;OS@?|ceS!B6S
zUFMQ&LKBzLMzLW@;_3*?K>G*-W*q*u(#6$gSfwmU1P8Jp5mf(eJepaZ$M4i5zSmsz
zVA2`8RP)}1Y!CzVPQSQzs+g}n)h!4lP2zC<0LROJTLZ__a|H!D$~O)w4!!jh(mOv(
zHjej3NE^NLN=sD^d^gMZ)+hXdoDeev2uds<9-)+njad#KUh>@q+GQNOdp|zVAiPEE
zEXRLd+TMC(y)ZCW{<+s{D}18()pig3_PsR5YXllkU!_wYxqh;7^Df)xbP?D+n7PZ^
zZ$31Km@F)4Z8eLoJS7{=&%`o3?2O9jp}oY=#*4D=Y>lNnO0eBU--Bp35y+Dpnt5*y
zcF{w}Tb(rYvS2^85O!1Zz@-t`dLIbW?HwH}Wu4S>YX%^V8R@fgR=QM;x=bG}FTM6f
z@-h^BI}sbrct5=|t>cNKHFeP1$--tXcqPq4L{wznah>-Esjrvii?$5TG6!`G`Ej+8
zJ=Ws_D~Yp+H)ShcvZpAJLYh<xY=uyt`IZ$v_n|BUpu*?;cCWIG#@VuEqr*ulmy&MN
zJpx?cy?OC@^lW-uTnBn7**mmh<9(qGK?0w4*O*{kNk0Kw#TCecOc0Alu0s4nS#loB
z$bveLS>Q(s_^?)#%D}=XZWK{vS4>^@xPAHp?N&7Wm_@CgkOr#|#XXgPEcx=fr>Jpr
zBTJMN!z{=1Mo(sRORC@T+odHn`1taed6ChviTT^x_j>qdeYt#v*YCt`FPTTZUMbhm
zF;$c~e}&&Or^9Y4AFvL+9vDp^>ltQ#U65Y&a^1sJD9eJhJ~3-b(W%kT_w@9yZIm2r
zIFwx3Z8No16wBsNBwr2fq-k_}eF>ti`AW&mRe(~r(()4+dMNB190AS?_yBJ)A9~v^
zZw5MHNw}>XPuqPf)42T=VXGW0*8y~V&Z$TU$Dv=Vgsg@ClH3ZC1G6{OCraM@@{=Kj
zW86*p7xh(@fMp$C=@%a<y6}@^y+fGa>wae#td}59T|TVfEz|IXe7@(=xUbN2=Y2-Y
zzSn_39)@a&N^u*~<b%nvxakImvv<TQ@$xEscB7$WAh8<yrC5}FJgwLZ^-(h{4RZ8Z
z)P+k}&pIOYM`h!+$_MByI&@#k)ZU)D^0%S3wkm#LvHFCkzL}Go3r|aX0@J91YGNi3
z%IY}^10zLx3#+T}#ok2gmViqI2KA=ec))7aqT?v+#zM;|?F(&Z8J?*B>RhK8p_PX~
zU`LzELK~bT%_eSiK)IX1QM4N!rCRdO`@Hx>tXoNdL4i_`r30e5N^~|d0a>8)fmyy-
z>e*YZvjdZ`1X1Y?#iR_u6C!Bec_7dX0ejp(!>183<5b}z($lGSr+64)Kw(S4ty^;O
z=;@rFQ5uj#EWgK6#Z6Ve+Pl~2xFJeDnlD+)bZ`km3V(!SudACtmoFE?g59{L!(XBJ
zMm;?}L%-TXt30-h3CMY54yL^~NGvepRq}Tj6l#9)d3t(wfB)_xgn)nUi<ym_caVVr
zFA5$$z5yj}vFSzIxUCjklP^424s)82b|hk`1q~M$;UK;{9<%Y&BQlFB_%^*Pra%xu
z7u)qxr5<|gnD;z#pHi$a*0uY9VrwZP_|56%_ir-QoRhrInPM)Q?+D+%O_zM;>$;z3
zYgDun^akPQGJu-mu$BtkJ#4w;RccSeK2@iP&u=?0JFhO+`(2|nH5D+m`Pw~qR|Yc1
zf{YOw{&s}Hc(&D{De`rpr2!D%^HEwa*Fy|YPNdo~lSP14URYdW6=4F7mrNJ+dOoN@
zNimpd@CvoGrQPDcz#GmHV>YDcqMV!?Uv5|~yhcJ@d*<#u8PNRvh95Y`V9^N)1)NQX
z5{OBs(07DXEaKwg3rkDZW94S#ywGUSVTwV1z|DGM{RLSC1$COw_9sZ1Q+v(pH$|sD
zEzz7G{$#74;@FZVT5=Hi*&sp9=@{N-w`BZsDM>-FzR%2*xzK=Yc9m9EPEJ;}g1k(r
zzy7X=<UN0l=?kSKfs2!srMcb05a;F{Mm4qW1x)q5u$8IuATNZ?R@_9irk%V0`n{E<
z+*#4$p-pO(!TxIa7YWUM>84*`S+Nxl8%j$zGx&2L8X_c%53L!ZWCzE@3qg9L&JJcc
z{r&v`p#%0jUpZBbv~QA1*sTyh3JmTw{G7_6bj-QLA7J{cV`S5J#W)OZ2~rX!_kH`Z
zFZq7l>(Nu$d<)ZDQ>NQ`uhPR)lm-^BY>m4fhG0oy%ka2`$qjfFv7eExg*>Z$EI46l
z`uUG6HgT_B)jvE)v~M~ppV(a<keurXQ#UrwDAa<BoEyJ7)GE@&cJR2fgtplqA-8Df
z=jJlg((W$>)`v{>^Z+*&QLrSi^lfi$Ms5MZDlh50Mj<%3{b6R=mxgQaZkMK<yh!e>
zCGE&dT5~%IGHC_Yr|-|r=wy+!59wrYSfQ2r<z)IJ>Dr~VVk`vx{fs%;qLI*;++0R^
z9{qMOQSB}yc_vhN6%jHr8tqn<wmg{{ZYlzDgEWR>*j-g(X?|(R=*1$OCqfUvji5xn
z)sv&p-II4=-QT)umsbgljd{{cyhXj7#WnE>*ayFleyfnrr^_-ENZvTzu-*#^ak@D1
zVh5?yW5d^FTJVUo#F1=T+w&|nE-n`s#AIO+5fN+)%Ys?$#<dP=lmQ>|zntRW-Mrbb
zx$)q^gSe@V10ot)HYTRSo;UpW?>`A6=dD{q_4yaJTJm(;xXsj;ME+X29Nc%VKp>T=
zFv%ICn?8`XlPbzhe)G2tOW(82QKjo_RzWiGQ)kdl9gh{8Y(i?}@^BL1#cXAS@&SAX
z<U?#(Qd|yc%C&@#Tmp>=R_)PIsU`klWN>@GqrERM1_B95;yXVXftz41er()vrLrlt
zS!Jf*C=bvt!+D)F>d#8OlLcNqy2n9Dl)YEu@Y4f!zTah`TWt&MY+y8TMLK8#H&HP%
zM-ojo?mQm?sPYUj91c25UR#Zsb^qoQGnk{9EVR9~JG@OR9b>^79an2gubkeNj&mc}
zjEfwZ?<h%4bn*4&&#Gt7E$X^gFT_(Qjk)5xHH}){vp#9p_M>)8nn$Da^YWNLJ}X#R
zS$+6W=Q41SB;Igky{q#BfDCe3TBe%3`8txqw+XGReyO@fgmDZc*(270DD)#BoT}Tm
zHWV(L0xNh;YcvEh8>?vKR~PYQSV;nH!JyIrYM7F<K3(FjRf76ngB-81Z#Hk@Q|q-x
zIjK6K^vccP?;5wl!CA<+lZiB38CJToG7L*y_;p8(FiCdkqI&H|Tw*o4H<ba}yWW_v
z+l`Oc@WpF%I_2)3-H;sjoYrb?a{b<u9R~Vt0c)cMrKS0mI&(#mUmIA#9+GCGk%w@^
zB_OAhO%QfpH*0P(ZpV{VOBQnNM%G(_)vYfcy^^Gp<>WBPe{);kg^ih24|tfc1aa=Q
z8Qt+MhcL5+#rzA_ixIFitm2L#nl6eHs%p#v6vcm>oT8KE`EFt^5$|{lb7^yxR&(N{
zR{+nzA+zqKH%A2h`3>dDz$&j?X_c;cK!l<STQO2wm&5kB#)-^OzG?ql^0)6SXLEY%
z<}e4Q_1zmk#e)fB=^DupO8HA-5{8sII9g%)O}emM!1aA-6o&u_iKO6X<9gjHIAFOo
zH3Daca}nLWy)iEPc(L-aYy+Pko6yM~%=kC6nqX^F<Ak?<ZuQ=()lQc>fMseVOXWI(
zT8Z!a4xgisj=W5ti1UUpkCC-7fy6kAsism;L49UAb2}*l7MI!;F)?a$U-7hvb=wgc
z#{MjHHUCV$3+c-q;BEVfn$}SAYcj+P8Umq!CE}q^be)nzWb8EWvpvZ`UjH_RM$uP;
zBnbPN(${vY(~5wJA{Sdl7&kJ5fn>SW6W`^hLV_2<Dto#}kHpgm;0jxfyf-N0vgI}+
z@n~?m8I)5BC+zAK`Dc78aEYi3mBo&Lu?=P5o-Z2(^JTjMHqU~L09Vlu>)*gxIl<Js
zpbCv*G+#EZ_yu!wA)UWyU=Qj(0mbrJu*clg6m~k?SmFc;f$*aq)lO-ziHvwUXu<EI
zy|ekKXre+}J+_FcK5gO061T<bk1o@x7w~r=AB$g{;CgyNSBlQC(5-boUnm9U)sAWQ
z*`|RK>lw)3j#%!H<p!3pr4JMGgYgYMQ$nuc=If!_VIdA-{^x>ZFO+Z~0ZzFRBn>c)
zB3)kJgDLmAr4(;s3JTWDfxI6-RQA@V^fINw21{SH{T)O?fVwd=<13GBYvZxqxWr4&
zZJ5btr@3wRA@hpG;O@${CiJ|SrfOlH*%dk|!)@fPX}GY=GRG7jk)sqx*EpJARB-4)
z>g^xem4&ExkEQc{WTu;M&f0Gi6m)F5aub9Nq@VM9jtOlbY(`Yt6aaS3S8!>&XE>P@
zSQ(6_y$rzl94yG}M4Eml^0Gz>(dSs`mt(dH*#KQ1MefgS2<A&oaUtF2rk$LEtGLnk
z=|O5*lQ=!l1>SH1GOlNm?@GmpflXZps$21H#OM4yRyxn&Fv`c$<IbV6a6ikChi5)!
zRTE&E+>v;1`Z#fK@<*%J*>t_xY{GTBen1LM*X;pP$W?jh9ev7Bh15qVE)KNkwFi}w
zST=C?Wq&UB(XH@~P_i)1^R}r32PxX4Pm(=hrGh|@#~MkFMR?kid<=yjV*O8>3=e1A
zj~C;N_By(&yi68!0aK~r5|Db6YuQ69c0V^q76v9)sof?+8u9HPB%JfWWv?G<fdpCR
zAP1OohNjaue1uaa_Wp?_wbmn0Wx0a=(^=X%0(ht8li+=DkecOGr#0_-hf`_-@diL#
zwSYSoCKPm2Diw82V+{tt$X86<`3=BgdRhMA5Lc+uROEf24OnXbdX;TqVN?~ex3@jG
zba7apQsbJ|lp?I`j4DLFo|6^&9P9TgadgvRZ{vp&(ZMRbLunDHHD-StpV=8ehQoq1
zDbe#hHSP+R?eApqiVV-+z56y(e@HVWedI1e{~V@~Ni0_9@7IAGffX@@r;54H;KcO#
z#(hpn2V(GZv7u@^4PM_v#I?QI8EOqLw`!PL|F`v0R^TOl+SF87g%1g!_$_!o$>+Lv
z6U#$b1~kJn(>^XF*O#VfGqWLPeeI;|o;D3tOg=Sy`kJl_evdT!gmr17euq1h^#>}B
z{vHUV$!`j%#YiAIB^}HBcEuJ6CHcsq$HqQfC)6tXR<JTr9EP>5Xpw%X5T?ZHz>CE?
zxykR}3w`pEO1Vp<iGM>gSc-rrP0}0rp-^|;^H%FqSQg;(l`MqCGl4-8`~Aqs3dGJc
zH|9zACG(_I9bzET)f3SS0P&#8amEh}z}*k>p$?G=sN(EI;Y#hD@hX8R@Kg!^AK&C@
z7kc+*<JrzLUtWt8Je%qsyPbt3BVuaeytwrS0}y|`VovfU2LT?7Ufo8neQ%7A&wTlW
zf=-h}zaOs@!fibZ%kmQ#XOsvb3|*oDb{?S=e^B4{AoGsd&jJ%!SYI(ppNuC&&|6rW
zMmYUi0zdOs((dVHj6h7rOL1{f%1EqIu`alv3*TJ8*#nSTp;a`nDz2b%@{VA4{mMas
z^$=_cn*k-KV)$Udg6O}*tc#MLj_V0IE`Ib8`CeC{Lp!BWp92&`<{v}qxlT@1N2cZG
zm{1fLS-qgUsDw(F+MIBoqc3Dd3$jhff=)4tY#?LH!N!6A+A^?ih1fvxlV1}H6Pt#y
z*SouPlp}8qym^7luvny5Qj+SYk9Ezqj{#X{Wwa(8_(KM}X*-ccUK;PY7|hJ9kppuD
zxBp86j<h(-clGF@>M8}P%*A#_rcP*7gmjl-Lhl$2Cy+IoFyr4$c?=z~2$3j9VW`KX
zIj(J1r6}#m)IeU7*H<Z4h882>z60iu%C${Z#mOpVYxfP+h$X`xq3$<_tZCQX3R<p#
zKm`Amd;~0xs;O2^2I1WZU03P3V=sg^>GQ*5{YphZPA3Fu4}|1|RUXYViylurESj#;
zOHTEXa=bXJ(T{1+N>sce-Bb)<D(e@NbmsA{Mzav_-cmxbmP(q}*N)|wKFnU%v7#5N
zT{2c_$u>7dzt^*G`gKrKqHN!e8O^>Z)X#kGPS|AXrrH2{cMxIldTlyNQ@;Dd!_j_a
zgv)xK73;6v_6~NhCdowOekLZiVLkYWJZv4~+op-|_xp}GHXT>+jPC_EZ6W$2<Gska
zi@!F}rgmqZ%vpsbh2m(fbU&MZvAbVNAvm$eGu+DMrXU9vR2Dq;uVQncJw1w%cQ*_X
zZC{18FIw9)Psc^q`;TxvR0v(C%s6+2u7I9K;5%olb73rdP20oo`wRBTN*w}8$;y?K
ziQx;bg>@QW-raM`bxRO54G>Xe5Ixwt2h_Qv4>NPchN8pw8Fb;{)TrE5J`&+M&TF@M
z+*S#34iA@&5Jar|=j&@#x7>S7c~&r*t5pxsrez2IeqD0(S=n&x^f^|3M1r8=_fHp#
z=TmPrOwn~&dMP?sFh&o)H2x+o*7y=T#jBxt=CgNp8g%S(TxbmJg#U_M<nAfINet{$
zKJZ5rSWF;Tw&U{vG6|z`$kcuI)(Y8YKz?%ahO1Xc1FP78_RadWM}Fwqfp&DW(We89
z0k>AB{Tegi-8;o-_TpCm16-ZyV9io$Om1+np)PqYAO?t!PEP$EsnRZ<a|CB$QyhqE
z&!-U&E&sbygjlCZ7fDnM%;{%$>aunh*i9v|-bLovf;x!ORu=VZ$Eb62@2zRie6xIH
z`K6@^c$pi%kx^*g7>L$8=^8Q}gwHjh28MD&8^9B5HGIG2Ru#0Z;D04();X-Re8Xi!
zZSBlg<oj`kJCW=l1vu+XNZs*Up_rLDf1#89Vk+>UM(OAY5#YU1sJ!ApSw^09mTcEV
zFIV36BMVcm_}#-(`Z|lg4LODW!yb&f^UiJwJ*(?#rC>Kp>B^q>#t5KiLzBxMdboYN
zFX}QBCfC=eT5@A(cYS5AyXnhT#hk_myA&=xz@%~lPn9CoQ8`pAtmfCjp~s3#w#Cx&
zw0YPtZS^5(nKfW!{J}$K4VU55DUb3;Pp$wnVmmoj{LR~CVAtc2^~z8Ty=}BsplJFQ
zkV?qYx6jxo>I@nuDp`6%+F@W5k0|&GK13jZCnxW#v*Gfn6h0Bv$k*(iU0{vVo^-QB
z6c><F7EVn2NJqJE4DB{v8*cQf^g@QHx&$<Tm&$~dgNG0{ANMK?6xzTwr|*U~`5oVw
z8msVr&s`@B)ZGDPksKja<nAD1?)OH`>?OLIV8~Zqu&aUeiZQ8TeoQO5L&N|DWy(pW
zfQl?1ec|~$%=)XLV-pyqztZ;rGYU(qQ1g=LL~+%5y<cb!n_fEBPhcLE{}s3){@Qb&
zWFM>^hrzO5yIKqLTD)x31@mDZaDKR0wbnZfnU7Hsf*SmeO^Mi%DifX$7K}9ko7)wi
zHa3>j_@Z(YE4|nFk3CjyVyP@_2);-LbB^`(7!L&co-8Bl*EKX^e%xMyGXY$ZLd7L>
zCR{nx+<c*m9eDtKy0_EU7-vIqi!#e1m~bTX%5hhZaqY@lqjUrsob#%X>*_WypS<UO
z=i!}itq>hcM?LV)_XL^#D)2V2b3(HJJm(T)C^2~BNJm6NxaLrv>{PX4^O5Xe>m5cp
zTCMl(jP$ZFaJ&S0J53+uS-l`(mlesN?aCcysy@~{xUsv31O<4Iwap&YO(k4qw;_g=
z<pIG_@a`&Hr#E2*Sr7y~Cdu7ew}2^Z<ZGOf3cy8oZN{^vf+LnqSI9iQy;!(-o*9ty
zt4-kUneh<21}#@~Rs-v_`}zw3;kmUkhezn&QtiU^--0bm)#T}AtjPWDjs<@T>ALg&
zP1Jm)ORj|=?ZV_SLbKSZ!&Wt^SuuAo)ICw;yZN5wfWD|MkGqhh*`4_%J*f5V)Qj4n
zBLRQ5w?~IEz&10O1T1Y3Iif#T=j99yW;3tg+Qv^x0S~iq%Hr+sID5r%U-tE-M|0W_
zjn6S+WwP5t-1Amka5smvcRYD;8SxqUYEl2hH$I$ax}KlxtDrjopNhKbf(Ly7rcuK{
zks9Jca_@|jFuJ#Jl277uQcnbhXd9C#XMj<~_%|NP8wNXU{N)QVctlBEuvRG<UBkcd
zk&b!1eYIOESo0GdMp4vxwi(Dp%l6G2!zEkqyDP>x#+Fh;wXuY6f0cbv?nJVI_-Oh?
zv!beILA39t@%Xtvcu-d#xp&aAg=%au+bFUjRnnjoaGoFeo*bW3U18<@LMxVF!c3ms
zeZBDja*J61ej1ava;;6x!NIi^gx{MlI9D+u7d`zazSB!PwKQHow|x6gpf`84OI}&1
zJX-mZp;zoJB*<m<(%+DZf6;v4q`3pxl)>93K7(U0<<pFzn5n=2_CJZ!IN8_Q*B@YC
zE$wZLzFwX;b$j0VsL0nkd2X%l_LY_ZC8|{>?x{)3&{FPW79+ytwhV4k57&!Z-8|%w
zhC((pdhn_R_Z_)z-mi;>b1!W~;)9+yIFF{51WE|(E(>&5>KZ$h47B##sTZ<*rK<fw
zPR_$@)}C#;k+euF^iz{qpy-s9oo!HUcR=ME2?@19fkaq$LyUig!>XH~m*d?#G<k{D
z*A8I&Db%B%6YoSuDkwQV))Lwm<KW<2dSXK<#S*hSLn3kg0XZ4-CLVSjAr}p0joxLJ
zC~7|u*LR)co~Y&j;1s_~7Rny{;_6@_k*sqm4rYV*4|RP~>P@`mC72M5vNzd?xmVe8
zUzuNQ4HH~aL3?pQD9=mI_e_5=bMCc%Z&&sZOmo-W{nAT?jTHAY>H?T2dxWa<;%(wv
z@59P&ZR{?0E2LI`CcihywJ)c{k&1J*V*AT;>a#WchrfOuy<Ev8h;(fmKn<C=D2ty@
zRzw8l_HWts7YnuPIPu+dm1WNKK7Iu?)#%j0t{{UinjksgKd`2^{}I?m+B4QwL%WW4
zZ~Ebj8qFQ)vglGqN5}1Kj}C-oP_>+>Gh8WMlDjDv^v0gXC!pw#)KE6;7e~NrwJus)
z-P<C*Tto;Psd%ZjCb;8^dlWlgmQ~&Pk65B(V;);g`@DJh5WTYY4NVrYZ5$UKbJsoH
zICnc^mZ<#IB-3|I7BR!J4>Rs-bv-)#4g8FXPmGi!>;ol2`ihS(1;|2qQ*p0ey6`Rm
zF~2_GA@jd9VkX9G@?E#u-P`q;`}vsg>sM{>;=6~3FYYI1W;i;RF74fzTjd-eUml)Z
zUA$x>PJr%rP8QbC9X-4Fu7O)fF8c9Sujr*8Lnv#WV5o1CIaP3fN@j8+dp$!~{;o<*
z_f7l7)^}yJ1KKQ}=q01~A<B763KVg76%#jCYJ$61nO{wIU`&aLwY=@@9B<s=$IxY_
zMKT5xq^s|k?|sh4hSlW(jv+Qvu0Mb@k4&FFTGau$ELAeo-Jjd9R*al>Qv$!KjG62|
zz0>;n9(lRdI%Q#jj!U8aP}3oKkycOE2z&fMx^37Z3klvirQkhj^*iM|!-b>%r77iw
zO_fQMO@@d|SF&NvG}Pho1=|Mon&{P$=mRg;%)*8wL4OW-E*nLOa*nbI*xQF(70b0Y
z1g%~9?n%xMM5DxdP);m@G3!N*M*umjOXOhFAdmTPZeMqt8lru-)ziZn03K7l7CyMw
zugNAzAyqDPv3QUDzt~S*7MEHp&wZI$IrP8o*uVev&kHct@$SsX@0~Jt$vgRT8gK}&
zU&ohLPXC$NB<k|vRl3@iQp5+pqSQitgE%zWbgb!09)9R&^ib^Lh?Y>hN3A>~rGnz<
zJk1khDRU}|(HitBy6img;NC<%sUvE8xx$BCy?V`9+vR4EnHhg4NnBioI74{6@csa}
zzh`D<+%@tR)%}T~pX87Z{Mf1!!qWA?`gej$oVKaA1^=2`=vp(nxW8$9wSOajv=It@
z8IYa+fGj8or1}+-A}VSMfq?#$&q{an5^=vD&%~}xhf2Y-Q|s}mWA6%e#t-L^lu!wD
zBj<_by1P}-{F6pB)i^0vf0nrwxrV<oPV2AH>OG>LrxaGg-P#R#<r*_B+B*C#q<O|a
z5#%@ccGj`~JmU#I#a$}p<&#Pe`2I%1ZcPut$32%U&A7Gwd0?5@%svvE=P&-1%aDuQ
zM0eJ3-xHGHz9f?<e+c+N|A01-u8Ld_-&rI%>E1{c{+;8!^sq!d$z1QyE9n-OmX{XQ
z-~uC3?_AuR?h6RUq<xd|LI4wBa9suxdgFcF$6C3dk|@a{<2`rxg}6xPxB7MZ(qmJ4
z(Reuh;go&gL7UR`<;T_6Qs!NT&D6>(2B1|uWfkq6J)xicram7&nUNZDdWJ0<no*Za
zu=)M=zZexJ;+xC!Z2?MRkm#5=brqF>+FG&?P%7;a2UrEmI8lx6XOxhZx-<8iDE0Rt
z@|LBSH>%!T#*>$`G_wv<Q?0j{8E*L1QZrw;Rwsi=F*CRM`td>F=f|f1K@wKaa;#5$
zg1{F!9{evX&_UFY^)hk6LdTect7<|0ElK#m@Kc_t(+xkUzcl~G65+MwGnC9LY}2W6
z|NCnhuaiF1eeEzg2OTi>u#(4V_-Ir|_a}PoYa~b0$Bm)HSPn%Bp<ZaHvhq?}HtVRj
z(A)2m)?ht>60cu3Q5Wy;e?%jrh^W1Z@6fjC7O88fePLwPs@SR#ceFxBN5`Zip^twI
zwRL<+dfJqH<gh|4l1#a}+_x^;$47D&aA$kCDm*4q+p#6d{YrI4DGzU}O!h<BA}M(c
z9_l?R^}RFgN<163V5m~4{lqq#w>3Rq=6-qX9MG>3in;Fe?-Jv&-PEMBF#i-CA9NAq
zu*hN1I{frXJip0!?;8UX4Hb>=nLEudm%4r?8tvegs^l{vEt%NZX@!{MaaQ>@&vMus
zb~hz`wuTjwN4)I2)S9MB-vb<-=x?4?5-gYBSp~bTR^j}j?BYJ0kkVKsuCLJZg6jLn
zk5J!;g9JY@%$V21NLt4m4QSN!C4t&@`nNeb=^9MS;|0&D%wKTh1zk=xH%w}{_ta93
zJlT36_&`SgGK4D86RGL-J4p!S@EU%<a3AQHJ6E{aroj6U>R9n;$<v7D9sf>Rd=-fo
z9!@qn0qG$3m>Y9~kwjma%ZA_n)l~Dj$piG_R6+Vs^K%y`E<er&xoX`owe*)x6IvpJ
zbtH}Y2`_WFp5U9_B`z0dNNMg1b`;=yU4pVV-9Z0$+I&(sUhf$yC%1gELh)8V95*e(
zTHV~uvL`UbZZ!U0CS+S){)UW<d2c(DR?zcO#zUU<2(j>s1N==l*P`a&AJ;`ijqjCQ
zREK0bzNeedcdVUuug)2fex{l;@Dn{B@<~p!Q)lEmQ}a4lduw;;^<?~p=Vl8wA%z0^
zw$BdiT;o%)hvZqHDxb^c-|N31p@=x)-ikURCMMl6?6*tuhq9EoV4lfSSnn1eH1nvi
zpyw_7uQi_49Dbh@87byv9vi#qGe;#+Lds7Ox|^ggB(`&T49aV}W^+B|alE+nP4kdA
zVRM!%X}{pRZVVkFgE{o|jA5^K4pm=JkU5=0Z&IZu2j!6a8t{@6H1#(>`?g@oMnIq=
z?ndX$7-0s2IBLYXei;CZN8g6Rp10A^nJ#8lA`rCv;;9_{%?eIhT4$mmQr~2<GpymU
z(NG>A7dDmLdq>&XJ}Wd=Tl69n3VU{4whw>4z6aMhbzHpsxEAH|q3qE-Ti1$@l*`7u
z#6Vvv3#Q*4{83g5EB1vW>D$hZ$7enEkQYMG>&{;8otuQ$+;qsuw>vklc{XeFgW7O!
zXO_{_u&15^tra>WU%|}Akt3~cK`74my4tPUJ&Aj`CosS&%7Ke_R?Z3U&XTWRr{GZ2
zX~MU|t80oM1;tz+(k6OOmtBJ+iU@*pKIWF_se_iZ>Ai)7P9x7uj)d*gRG41)%|3*;
zJ+$W0#uuq_7MES|VN+D}bQ+C3mMzCkvbFJWoxCU4-Q(%f!+|H)vo3<BQDVq`bnxQ7
zmmGa+-)Al5-(X5&()Zty#h*U(eT>uI-Y^5Fh|)zA?u+1T8kVeoJg|EzA%&ZI(U@}%
zD}0@bu%YGSjbSj)DWu}Q97^uvJ2TV`b2X99DA6dV|9SP+)|O^?n39rt?`$R$3EoIJ
z*;9N9IJCqhhwJCYK$~OLBgV(HQQ%Z-c(0$`=i<18*-QDRyfNNCxHF4P(0-xS@tw+h
z?4BnXPMXRSJ3l{Bb6wvJMbEd?wz_v<unx&p1WZ-sMUm$<lQ~b1GcB+}oz3r2l{>q6
zC=qO_eX!2(7Yqznq=m@rCkchsXXDr}JN%5fd<cC)^KozQ$g{_7(AjxatcyIM-i97R
zbAWl>_}(#W$=%h>G)GylQkPxX@60!5F6Dgj8u-W1I09_n5_yS<iOtuBYhIT((FGl<
zl+Aho|KSDX3Rg*IBDcR6Ur`vZj0%m9LBM}lR@v0M1X0VQN`=^94CdxHt=DD4g;Im`
z+$Vd7;ii@^TaJmQ+j8ScJLm48w^7i$sf|x$Jw3DMqK2V{line!)`-ae!_-#*RlT)e
z9}84K5D951X^?JEk?wAgZjkO!x~03jLAo2HyE_j^cf)^!-uL@v?#w;IjOQ2oiS?|t
z_CBgIy<2ZO4(<UdtI<6i^SLDU&VO<LX|xof^ZJop^8_@=1S5CZShpKs*#l_x2b)Cr
z^@C=_aUDNK<cOaFpwzhMya=0d*uvJcsqDYTbx8av9F>Rt#*TJ*T|rfJ-?Z7$`$q7$
zwB>0xs#9H^RgiU&kFYQa3*AdZVYmjo(cluQIC|$!wF(tK-JT5n>8Y9Fv>f_=U0nb>
z-o!Zu-q4z-t1rBs=`1Jf;YYx)FaDVjRrCwAl|2#DbFMl_(6}7-&AM>n)h&q}4o7c3
zr?=hJPwpq2Dh9eHzSTm!N-p?W(o!#kgg(JVV__Z59StyS3<N$D9p{k0E&2;5`?qyU
zNNb$^IC;O%dg8S`bpXH$m}L?to&9qV_!EmpdzN@=4oWJ0Ej=DPKD6)Xg!@J}dzuFq
zd9)}|k;A>yp^;4EY(;4@q;JQ2-MRZ!7K0NU&o@1x_m9hR9&9W2=-AR0|2oHj5}vh3
z9oka;GqKC_xYxB$k{spO!k}f_Gy%h<exM{RhRhUg-9s(t{gD!G1=_mpoh~^TOWJ-_
zs_BjM5XRh{DV-*~(y<&Z@J3>rnfy?F_oR;2fa5eW65;0qu>NxM4|jaw^cNRy{g~6%
zQ)Bb&z1fT(fu55CpK5wWX^*cUq^zPm$C?uS_3_@+21O-`9S;rv#3ma=J~Ye-XG;w!
zDiY?SQSBWk`<o}uYJ0mqTXW97-QDY&x@cs79>bp;e`BK0$SWme>fpQcOG{p!YFF(c
zq$NxnLa{U3n4ga8!*h14DfR)$flN<dEhpiY7J=Fq$1;)<gWj!lO?Y49@Z0J?{%d(B
z0(oPh30WN-lRndtG5LY;kA8Re<T-1W|16Mu^3|F%X}an)VNPg6ao5CM?{M+y)~ub)
zl>Hy2&KmSpmAe2+1Pbo!!i+%ca$DX7C!0SVSKGY-fo})9vQkFwFXW+*^soM8^MgLj
zE(}L!StMCFHlCZ&9C(|Ja3gD3yS1f<f-=S?*Ky)#VO!j|78{rL1QO#}S(r9t_(D{a
zgp|xv+nS4bZH*-+X1Jr@tD%tk?dDzW#KB!+0&k#YU1)sEQYjP!HsO{MuIRS$frZU_
zb3aZAt(gkDl38(icGW<4qN?beosA6|qoWOGLFA&3z>hTQkoXv0MSE5;CMK)>!({O*
zo9yGYB7%oTX#L405BdI6wjQpSSo+#3a!3rXGP`oZN_^uy{n(1bQfW$=jipX0w@S87
z-gBzE+N{F)0@LdGavfnvf0kZyMyl#ZNk!I1uf&zBMTyX{i--<J#t&bG*v;m5hEPf9
zN499m#|<`@rRrzPl$c=Szs`3zvSK-B4AX-0%~NZ$w9<*!nx(kHR@FDs^&;EOraWcH
zu=;0+uT)A?c_@LC&Y{2Gx%m;3%WR$J=ytqNC&({)%Z;z|i5w-(cg+tEv8+b&chK{3
zU}~$R|MWV<xZWy5?%BLa6#2_Qo|?;XSwl8^o7HdYMQW|P6eTDT5evVRD{5MX>8oSI
zz_h~aC?2d_$ZYl}ASdLtuf5@C>ES_X^5`-H63|B_Aaijnw7*9XnBm~ykHd-bjsR3y
z{wTuH=$)w7?&*KcVFy{Miz}tLB!`sp1=nb)VN_4pS>2*z!xyk*=X(aCf}9_2nSKe!
zg9-*6kIeK^dNwZ{68A}SI-jYPj{T`OBY-m4Q5-FYWw-s3B2m0@DBBHVFep!M?e``;
z{IinFqx$Lc=9aeApcQw*wCMc<JtA%GS&lLjH*X4%$Qu;lM7d+w!Q#sj#vBV~(=DNM
zj*5pJc>pR)XI(#}fd`7+Lr?iL=i%afuDxdH!sQ%^?&SPIUUnY+#D-dDtr&k98SVXh
z_|4(sXY-jh6)ZE8KjS0X@$~wS0As+`enGd70&$+M%(Liq3*s_{<LJVI4SpRP%lF9u
z15i6L3yty_?T1ryY+;HDzo8^z2S=Nwy^|$T)~};}Q&P6f4<n0{bk3}ULe}5JJil%z
zL_S}^<TT%i&Qr%hzLTV)qSg=irt`Ktq6Op0FJGJ<V{thly7LV>G&OhoJ;HHjL2~d~
zL#L)o<55mnb~3-Aj|F28d(fQCGEuP8C2M;|Mu6c`dluqxe%ZhtC?tY?G>ao{?_3Z$
z2RwqgwVLV054$J5t-B>btRpQqidDLHqdx@C*my@@VAsu6h@`~t8SiT&jz*a@jq!Q+
zaHH}%E_0P~8>#WiJw47RO&oXcQcEq3I9$Q_Yed7BB13+$6*wguq<lQxvq2W|l!!=D
zLa8Huer}3xDP?BYqk@rXJ@1k8B!ejmAQ0G#_U3OcV@S;6q+KLmYpuJqjD^eKf?Gbw
zd4r+EYmVL|As(#Qz=fT}7J|kkw+;VtJ;Jkb?qGf|G8oqt-i?)C{AY6Ok_8q4O(_K%
z3^S}_;#;=!3atUv>hAChbtx64Zs?6;sZFoX4DRx}zF`ZZ`(=r4?>7Jw>~AW0jX7T{
ztYAOHz<;r@>jwtyp!y7A{{gkdyrvQTJl3N7M^ALe4irvL@P?%c#!yW%keivDuirNr
zil1H9>@c0>(?odb^GT3QWUd0sO%XA$ZD=~TyM$?dokdAGB}3D8*O9YUR`J`-otIs8
z>2(S|DI`OxtEoT1l>K&PEsq7u*>k>7RPkU?0&qZMN?FzAs52x19pjVH3pU%D134pO
zy~QpU<)`a8OFUhqq}sDyJLzla0AdYQeMy`xLG=9G98AOEJ!E2&5M{(5a<yy(CYX!l
zN<4txWp)Pit^9Q<F=?N>k)S#8SJbR~>OTOEJGOsNYW7*-sZIXXqAENBe6%Hl8=x*i
z)g<g@SfffU-&Whe_4KQ^%@2o);`IUAV1ru6Nsqtb0vQx@0{YU;5-Wt-R4r5GD(W$P
zg|`-w-}9K0smU_rXPBI>W=ASjwYGf2h;(#^`sX9XakK4{##7XAIrg^P`O=m~WL@Q1
zsR3({RAJmIpq`XL2%DCxL8EFb4CjJ9i|A$>Mai32VEUrb#hH6c@v_XUbE?eBfm*q>
zdGdTFhz3xgamh$tMROt=iWo<_B3fXmsZ1!n9Mqt0Ya1M!^0jraDKHOOQDW3nx*FHc
z8>3XHP3bs#Snk`c|L}g)vFSu44;)u#*L3WdM#w#VM6<)*n@Ca8wyH*$RTbSc#cCW-
z*L%F<xKU+C%)gARzqKURgt#zU0!VD#bg9=T69h^OCZCl)J<?(_PP^+qxkk1KI@-k0
zMnelt?I&Sj^&EKoW3$TcS?L*GMwv?MMj|Im+2Rp<VTs)YE@b_9Hz+Z+=IZm^P6Dnf
z?dXWsMZ*&?w9B&Tjy1P}%5Z4aX~V$CJYtYQTM)al@^t{920)GEO9#|s4o~QHX5rE1
zsH-C%j>4;##(c_SKJ1KvQxO;-$}9E(ee=6-SML$Gpf@jt-B~d=Da!TXAzU4!K?}}=
z!Y@Y}h~?F}tRr;yjyX%-7Vs>tJ9FNdAt;@Hp*AXl>*K$8!Bv2HPtN1|lJH3Gwr`K!
z=D=~o9uyJvO$AFWEw#80LGqoPv)MMgw_%>XFEspJ*5-(XFEmEb9={2DAtF*}Q5O+|
zvA}VKG6I88r?uU(jQlE*)jy`&4a5Q)R9{b1gxePVRK<5~@&XQi+iEN28=$HleMqqu
zPMDCqZzGmWR0%uZKb=<?r&#9KXlv^K_N!d@8m?~3rBD;Eb>NHbFP?944cHPVrXzEw
zSo=9z8-R}QJon=U6yyuh*3TKO-@n86l@5`#Ix@|$r2HHInE%DU&MP^&dm?l`h9d;D
zAXX#`uk&Loi&4XEN~!}Z3vK-gYWST${uwz2fXm_+$_O1uif^7ZkUn?rq?X&}77$Uw
zyx`>}MHK?z2Si8zNUJw(gE&oi_aA59Z9}1#yu`!+gEkj?QN6isOr#c_v?Pw)pR5vs
zt6QQq&ptlhx3<V0V*_k~Oq;)do@{(grUa>7`rVU&tr7n;C`~cv2cP{(meA5)9~eF)
z`xa2S6x7>i^V?24RLQfO$~Zk<!$jbWW!tK2z4TI}MtwG4JjwFkl;rlW$6@-E1|o>R
z!fFX?^<uChw{8*0Sk9nLv(E{7^+UL;8>XrHmEWvrugD$hQIemqaF<*jnC18tdk8X)
zPOL*I6tieQ3{L2M%>p!ewRK2n!V|Z!9KixD)`fNY2c+~r(yh(aLoRO(B!dBKwL*0C
zS2G>OtslD1QW!GVxk^%MYFi5JI;gau(wJ6f7G-%$rDVrhcju`=U9|0v=!Zwy#`^{Y
zGG<Cg^vlzss}=>x5Pl<y-oQXfhUGKm`!1DQ_>XH}=E@v1oU&yLmxV<vWj{y3m?)g9
z0~wKJ<_-9;xiAC7S8bCLV{S3?-BaT*Zj(m#<w2;9Hw|WO*Cd=7$oxh|QonwYl2Wy{
zeLPbS(Xyl%H<0RLJ4<_CQPFglXl62d9-m;6pU0V%Tov7Tx~*l|^dcw;@oEC!0YHBE
zt#MDXCs*{zX~8It_t|2;5c+qbX>%s$pbk>xrd$QV4A~$<DJv`Km~V@&3q>wmSJ846
zAid)G9Mfospw7V<CU|qn^lkiZWpmXdtsE7f%mCC3O(Wxc%lsR83Ww!W{T<a!YrxVQ
zA0jRS7(*EAIPQScNIqV@U{_nMsiFD#<7c`*p;En8v(9<XPE`Y+CG*FPkHwZAHsQ}J
zUXuvMC_)t8=I&RNw2EX9)F-=9OjT3P*t+Rly&4?Idcs|Q&M{wDgZGr=KFj|{?k;nc
zfq?7ex<j_scui%JZ(qxg?!|o8);+kt9xWB<<>o=;!sZ-l5>@y?=Qb+d&u{sNL@!mo
z&wcPu7~sCr)os!Crd!pa)Q3uE{cP<o1O*8h7;zB!y~z=g9J90V;;x@4IGz6Jf2Ibs
zA0ukMM<0fZYQ<=G8OfcU!BIplv90`y7--G(3Da9BP&+4rp{Z&>05r7@9aL!b0LlY2
z?&~rgEC`TxIw;lg92l(m?(p_Z<Ro9SQ>(A}zsX{8I>qKxkL}B~UARwOwza<$j{@Vh
z+3o|nNfpCkwM{(0bM*t0lC6UkG2oRrrpK?;+f^YzxqzKjQDD}6s!YP+5fOsFU^>}*
z<w|Xm+-r6MxNyLFpv7~5+4fT))Tm5pp34c_ZEB@c?t<~=g|p*|re0_N#@kn&?>EA?
zeuQNS(vWIPH8xaD&upTss2%^QqJ=o2A$_rt1bjr4Ap?*K@!Tdlg6mh#neP;O#g!{G
zw6xp$1rd+VT3a_)rjPIM2#b&~v%*Mp<BhakeJ#zz%a?EI&)z(4X)%gVyPvVXXubc3
z^REPy#9qDOz6khrXmLN@BVL`)NuB5ReigTUC5jR3nc`P8Zu+(K&bRjGPC~i4t=9Tr
zvo<;^y}^-~U$>zR(lji;_%{aX5IJgU4pt}qh?M{t^7X!Wn5Iz^N1r^6g$2Ka)aJK2
z@aG!`IgGK6MO(wao&_Jfx{Y2x#+|A0ml8ceR2g|}hRxBDvns>JbhJG5=*E=YWhX@I
zy4ZjM5dYXN#vD&(RrjjAfItVRmcubOE}+HH->!&v<-0fZKP*n5=f9G*OPExk#Is(`
zky&o0x>D3rSy8Uh<<ktQJLwE{#Zegh*<inGaDSOEnshu9#suY<?DjymRGV`3ct#?w
zWoq1eiMYU%;d-^JEe&%F?b6^8pdn3O=st-T6U>uy5gzb9`n{f@|GmAaUg+q3|5&Uo
zx8nIQrBZFR#k=>3l3Nu&WfiPCS=-mvI7*GU8@=LN$6Mm9@#<`6#@~U8URSUG0wT<h
zkfMX|O6!STiZK8CN&{XcF>3Ye`U*bFhZc%S|6hwC`E_`BZqg|x0wC5zWtJE%AJdt(
zTP?*<S64o_LtPO4XrONjzA|@=Fchs1^vn*ugO?32IaCdLynYPRjjIKu9)(6Q!+FoF
z_hppNB-lETqr7ynTVl<G-xiljjBV(dpfei4PyO}F3`4na%z~0DM8k*)oT&Fi-wjXQ
zu22$3%9dF%I=c8~B>28VElob1rc&y+ZB79dXP!<@RjL~>#7<}hU#BkoSrnB^o#*-O
z*yb9}b{D*Zk->N(L4ye6C^s*BVT00?b>pC!Uo<eFTxnU5yiUdafDteWS+zwU#t5zw
z_H%{|oi&c(LgH7M@7X%-D=cAOb##=Td|AY4q;sugC@EjspHSbwH1k&|vt+gmg4u5#
zg0QhnuIL&Br6l44vJ*(*q%{pkRc78clrB4C>}ap<1T%^T)XjQ3DtY17sg;|TOVL3U
zd8{TESyo)q(mnVS2<ZJ8I<u5&OqMq56Y^@wuKY<3)n|^LweC+k8<m?V;9RwUv{&c6
z!Lo;Ynw6HuNxiPSK@q{sofdlNCx_+k#D37Z9c5lpYdNq?^$693Cvj``6BIixI_5wE
zj%HjXKyzVJqT^)qKm5?~`5%dL=BovzfqZDoTd&2sGf4KPaCq6?XBX%R78FiBw%s3*
zyG=Z<@)y-Uk>1EKYh0v@Wk5{HYpt7_qqgE^_|qMgwG_k3G|=)5dU-h5lJU6BrU1DE
zY0=cI4h?g(3`msfdV(`eZ~u`S7moqSoxRNUyyycnlkS00(LNa8F3gYQY~tYwh=>D!
zx|Y=zf1uT{LRU}3r7h#*Jpgx>JUp`Uh&f7Wjis$8fe}58#bLMZZ;F|#|6KU;7wf(8
z#fX8}DJqVW2jzT!6qK1sqyq((nDGt2sDe3_i;E1!>*M94wk*A$p%J|124wve6;-U0
z3nde#)>Dc~`%7hHPAn!bH><1LGBZO$65QiT34tyJXo-pYo(lX6*SCL!^~;o&M@9tu
zdsFl4u=SL&X18~YuEkK%qJF0JcaC%;6FS!`t=X(V3CZa5<jntum4C;{Ow9wAG^pqz
z+1@Od2?J9eq?4KJBPpR0phHNpKM&~gZ<fWKea#K?6Td3U+NOI0LYFlIL<1xxjuM*h
zc!fM4+rq-m;$R!1LS6T4b{2R?w)QG&HpXw89xBdRYRARDYh=>`xB-zmf&==WYMw&Q
z#l_nk8HMPL`JPHH-nP!8zDv>g=Zyk#YPdRUnE8FZ`fJnSP+ZKeLFxvCgw$*vCO|xp
zMWoAE6sw75s0A-+tJWsHen6Z5U$n?kqt*+$he-onTn5S_sW`QY7Bx6O!jpwtaY6<}
z;E~2_uO30cN|l~En?Uekn*}2P;A(XaR`)DEcq2DTUR7ywlrsHk51;d=z+bYYVE|xp
za=HvescPevp5URIN0VZ)J-Wa0b4g||+r5`f9lwS3jb7_%t?EGdF0J$2^Z)eE_}``v
z<e7bAC3`_Z$fJ8%rZcvoaX)J(Z-!0L-Qv$5ZhfXV5-vnV*`Gz^9=VE`AS0tN+8E`<
zc{^XQd^3+^t_Xdr^mQcgtk!YGq46P-tcd^XMjmE8Xn-YKvl7CrP1CaS3sF{SL2T9o
zlB@{sU=j65UPNc(bu7>dlE)0~GmJ<G67AzXd0;+N$XeF0un#vk2jQ#jI*nGs>Lu9)
zsC&O^Q^_CzZ)cu9F;yhZU_%+H{CE<ay?oBZ;sL-_f~K*r-@20n%nm1~{@$jfUiAZ+
z&Oihzuw$C~^nYG}`&Tkc0%n~gktCQVI7h?~xmK49rLSPZ*G5I9J!yVaW^S41O=+~-
z*{7^%zld}xM4X3*+YfjIEggM@H9O~w&`q4_Z^nO~s`T~&+EH(VeQ%*XfY*6rajh%e
z#Tg9fs~OJTMs>oRiU|W0SHY0|&E8Hd)vJd%c&_?S^j+qntI5I(yF4ML74O?PQ5KYh
z${VQxtIGBbzXG253&3S?xmxgX>M3zP<880I=tPDzIPO2{C^MGZsRk^OF*fr{EobLf
zNvxf_(pg&+qUKSQcI$$%lhgOAL?t3lmWpyc-5$3qZYCFtn}%xbsO5v30?9V`fBp5P
z=MQe(Mv-8q0zkB8b^c<^53X_+pNHFw2~eo@1UjdNj$wCZwdJ?sUo}Ro3;jpjw;~3L
z7&o)3cBxvmm;-x@Upr?wDI6-f&gR6e6B4U>fHF17VEYPiW=?q6-Pz^J&j)_zTGa97
z=>Qe;esr%cgTEi-X=obX)!IG)XZ0boji(Q}{$-AimO*<Dnsm5*#{COmSb!?x#IMEa
zYIy0{SXmdpYW9{dBU9@^iz(Q=cuEfRzmJ<18ppGH1N#fZF|Z5ov<X7y_ofh3g)L~#
zFN@#pGaPJgrt3l%_2*dpCO$JaP$?trw=xGLT)hAS)8-+R9Hy$4t}F(&FQ2mHgG@-Q
z`bPmFOl4|bOEL;rw@r7vRcMs1hW^*VCfj@8#&gNa|5yV&IwqNe39M3k<3-54>LGec
zRN4jIkxYix4i%OCR>%tub_*6h@-fx^GDC0+tLsoLGffdh^f~{2P`-ewp{LqL49MFE
zd-}{VjmT7dyV{y>4okt5vlR(66Zu-MBG^}2%j7sS89H!WajXfC5}bA|A|HyCt@nS+
zCk>Vl2>NeVXMBHqJhge2ungoz16EcrbuDV@@pC7=IyRocbmpx05<5d9q2YTxI5dP?
zujG=^5Ek}ii-7ty+8br`9ACbsNr`r&mCnQAR>UPdwEmD%s9<mJDwyDc<m4EPhCMXq
zY8Y~HHXp8j^66jCyL(iiqN2aHJUnPHDqtl=2RVKJzC376U+Fa=BT4Oe$(I$)<RbC=
zdpO!09L%`~2A>(8|3gDQd|mUyk0I8AV1iVoDD=@N<S9|e8?#^~i<umip|i9wH@9l)
znR=mz6{iFplK_4Q6(JEKVhXH??_8qBCI>dw%^%fo9Uj)iyeMR0@o)H~jW)!PKmciF
zh(mGd0B;NHZ8+d7KCrlDjWxOL`okk2K)KL%|4`fF&V}P{@VQ7}fMJ4`GsT4d`PJZS
zV5;dkfYS&!Vz{d@H8AWhHdT%0G=)!4W-v>pe5h<+ME|pSFbFmD#c3sH@dH5$IDo#D
zewZK&LYuTEhH*1+hn+XSkN5fRF{gtwHZV(5SeRQR1Hv24%L_QOkb%ruO(D-}yGFiX
zHrv}0rFS5K$enOli`B0ls4<NW$KuKe15wt}x{hUA(l!(lv}Y}uD103q-Ll{vw11XD
z_<#@^8Q?6nR$%%bM&i%2G5=!XOIo@N`GLWDdrr3SwbLb~`te?YbUdfLrcZx=`=kW#
zUWJJndqr`TM^}<_|7p{^fa4Q;FjIhJS=83OEHYPG{Oka8^yCLnXk$t&p|oZlD|I)&
zpC(x3i_bi9kEXj52ExMz@XUETUvKQ~I`lMK+Y=M#*C93Qk^E4!-v6MNT2u_6BY+0X
z66+5Ulgjr0{Cu((YisBd5ICNG<irr!6@V%->MVcbxk32@>vj}?w=8VIsT15*tfDgV
z`8m-#=ipH!aPvLZH#6(Z9cop$+5aX2e{6p@kU=?Jp4GuRp0)U!^nG8PI3SOKR1e)4
z_2t-DvJnQnu}a4BR6h>OQ5(u-Z7nh`R%bS=Wfujdv0LxowY2`<S}G;PGN-wc;0g6G
z2Fz3?lwJQY9H1K0x3lQvqSn>b(9!L_Cdpi|xyEC+EnVN2jE#*m-yb@#n~~c?DyOB2
zD>2_D1qK+kYXkfxbMvwO60S9LRNAGL0a9uz9hXM6x`|S(%1CQ79WdC9zd$^2i-Xgk
zEvc*7uXGrYeMpN9v41(0Dmim8MZz+rV97`lmDn@>jhyu)$a44)?rpg#TwH1@&_h->
znAKBc<Zkkcz~mYnS@iTP=-4c=v7@A&F1yrDgPp;{zbw|TtMmCRw|au(DnovTH2+>t
ztQ#({3|z+~jQ|Ei6fjRl?y`SW5HaS|A$!OiL*f73!H7IL>($Hv7+Ax8W@Qc>miJCJ
zL;-URa91#Q8hkGf5Lw5COF!A`VZn}bOZooDkt>NG%eHc+uw;GzHE`*?oylbd%Ft#w
zR9KbmUb~KT`&dYnwSkgW2+$uWdCxLFHv>T;cf5b@ZU8Xc%-2r9$aP(1Z<GLinA7q9
zcwzr~|8QTJoXT);sZms5323Ii^(Kuih2*~qZc|Xl)1#LLQmB9&G)!6rUS%(2YqR;(
z9&c#0|1{{J5_sRfI(RbyW)Vy&_u9FM#B&h7_wbsnGq(W9f8a(p)`mUgD#n3*Zn}^W
z3<P?PUy(2i&~%9FANU;nFRh_Ce{Dg_iODkLiZ}B!Dyj$QZ!S+ez!;c%Xi>PjYC%iy
z9}kcw0)jo*^L6UVm^7&Fnk#3Hz9_Rw3?PQHs;Xhc3dB}EdHGwZ7mAhh%_H2xaz8CV
zc=Gc*0t=h3xOkEO=-=vv6ZXIB4UhImj_uVuA2mvsT1ud!E^ltw&2ZJql2Oo*Fp|Im
z$0bNy!-%*%PYj4)Oia0-v~m}qda4wGnJqHUlQku0GhJJEX?Ia#%kFOyn*JB=vm@9A
z1>f3)pcmj_fd|<Irws3dgdB=b55rn{kSx~PgyWkn3d1{3PWLU|qhA(%IDCJ7#a{3J
z;<{0x-)lW}r&|PKysD|$=}*CTd`|f&o|B%FKs6AI*=6*mW85Jj^cUS};!6$3MZtcx
zY9R|<G@xttS1?%_O#P7LtQi61lLOoK;x!4#`7R%U3Ye$McxYmpeDB*ch=F7>GTxf-
zdajnXPFWr>s)o+uXK|^a$TqM^y7yn$*%lfb5!jH3d;f)JrtR6dbDJX#s%cv*xx*8k
zU~y!uxz?1Z9Lj&w2o}I3`-+|@@r<JL$N4qX3jVoP>J+NOZOuWeciI~9%%e4a$<jPz
zi7wFldl4v#ug2vV%pN-4J}An@TVF{^|57afK<x$*8NENOZlj=x(97vNThY8RDLFMU
zV?#_cDy5eAjGl`>IV91H4%&eeAg1%VtUM5JmBd60|4>b6jSN2rHY~vK41tFmNQ!O-
z4aTY_tI7ZMJ+=dMj#+hNd_-T&21P!{x99A_!t+Uo&w=kT2y9|*7<T0bt`o|++N9vf
z0wR3EmB=mHoE$KT{I<nn9?7w%Z?Lm?|Ag&=;bW$(5p=0J#1aZ@7b(D%l(2#wDE>6L
zz}rzZBQxw(A}$1Q<*Vhp3D!Yr)MUg&#0YS2-her5>wPsF^6R7~cK<u=8Q-u<72&S_
z_1TC#V~*W{ypBd%$5{oy^P@=<UU*|XKxo&$P{;>!2JLNIRdrHwb;FFgg|Zm{sC?*p
z{M#Y{WTaHYZ0u2U@|p@xOabRc_I1DZ<$*C(`|Mggw_Z%c{KbIN*^|6ixG?0eiXF;K
zC@!m*+NXL^0@09<MzI3|^#B#|w6l%QuwO+Di#rnNa%wwsbxH-1drYQF1>9)BA5dap
zD%!&(K3q13)Vs|g`6rLGTa@_Z7q`6o1QOP-5P7uQ4$w~I8qd(ib#>V|+OOO}*~CzH
zZhe0Tfp|ICf5;IPcBo}&?&=Mv3-kiFi0$L6w|VjHGYvoe`OaB1y$+paCfBsS%~8<t
zRzG;F7wjBT(8p%gay^&+QINN569r(7DdF0g(ExN)%A_4A*55ub4w3ao3J(Dn4SkJ2
z4a{XAwCL4V5U$p`{DPE%)Mwx}P24x=EY`(kD6ME4ZMyf{ie;sjko}<pj;{;5cJg~i
zzzDx(dT{Z5^zE|0ERegpW1jPw88j@n+L?|3o6f_0-;gj_v9VDs)TX6K>ob(;lTng>
z(a|TZExuh55)vk*pnB3YIEch)k9Rxb=kNdB&-T;x3@}Q(Jsx+|FINKSqE!FfouRU@
z=FR|E6&7H;_YX=wmVF#xDqy|bk+)UJ0(Na+^8dAGMJn3cr$Fa~1aN;YcF{8?UbXe=
zIoK{L>U%fRFa6UV(p<@p*LD~8j`^3I&O9}ohd2e7T%Mm5Pwt-3Y4NbsY|vVS(hSd!
zxN9v?pNm%*-}lD-nY|@fd~6{0-D0wySF}q8%JszVld4TR`dkl7>2flJ0)g2PoxV`X
z(B>5tuIrECVX1!3i?@k#k&Hnlbd*0xY_`*Y3HiEFS)#zUh^zjxXAiUyiUBLbfiw5Z
zF*_G?dsZCuqE$mN>Z^wH@LL?L&cZEJ*jIG-?moLBJ!ZADqxN&Ac4!`JHAndt1h}=#
zD@yS9JIxFX<!jL}J$ZTci_Tb?Xom{ysCV(e5v_0EzGcXDefrv3+-%=xMvaD)G&wmL
z0<p9_;yYnqMOSzFBq4!>#bm6%A>)B`l75;+{XmFu^QbEP{#m)epEpJahTBsPy7dac
z>k>M1&TkGy+?(sNB#@9?LV(^`zoN89c^l@O#dA<mj{o&f+YFlRBb25|xxRbfpqOsv
zeGLF|Y3&MI+>QADfuHQ#x3P4`fjo9+j;Hrus(jWT;KYp7iRKJ&q!&rUZRFi@V@%_(
zedcQ#51<|f-;Mg(&R*ookVel|pz`C!xdn#2<~Ez_PjWn+%P=+@7@GFWL41`CH1F>5
zIh4qVF?|<pU8bUK_Qy^9Be+KU%hXNmVI-O$+|5#;Qx2O#f$N6f(a&&~!D|K-8e{D8
z?q;?y$I(tOV9MpVIpSICYZof2;1&&Zy83<7Vn%C-yA%}zuJmz<8LuDg)A;|fq^Z+h
z4fNg?fnGsDr<Ay~&s#CKt?;I1!)RcSAAN0}J^=|O&&tHX6tu@|+B`h?6G4lASzA|W
z$69Gpp~STJ8+^FX{GBC6&J?R`5EUi$(wb-Eqd}<e8Vkdg<c>9b0M*+MMufUdrp7Gm
z0}Uf7V~LL3EU4DL+%~%%{cX_*D}0)InkqZhJKH9TH~Swz|2DgAh>mAtd7e-Gg;?%*
z06|btYujrF&;+tZ^FWH`%bycb6MO+l85%l?)DmS;P`oD)A4NAfBgUoWM#=vh=51~X
zg2wOshX#S!kcR3yaCumY%Q=}uONlA(R3#!2w-LedXJLQPy$3usp>~T_`o6yU)c(U{
z*^w=w$YN>%kVWrLH@j6vg>(bfyB`63U{IoxhPo=QRf)^x^8t&ZTrf35Y2k(ZS}hy<
zfp(|ODmv=HNlk8JP?h$88ZtfIw-XFcM<^|$TIcWHAnG^d)1ppQ#=L>hg&shszG(8U
zzxhU6WRJKE)Q{-l@i^o1SUe`PoULQS-n-8%z~0~9*Yem*+2~b!$I98n0>EKkq}z9m
zfrJA#!VI~tIovDIM(f|DQWkp}CnUes_X+c0EmSh!%UpoL#+|AQ)3e2UM=P$nW_Ra7
zbXhdpM1v14re)ST;>QE?sx8R$>G=`-d!sh6TB-e+&gnP`JU%h4$UV0jTL|x_e%Q4#
zOOh^lzZC&LS7&#4P5(RaO>B<031+<oTY!MihTdcb7nnj-H=}{b47Li~msdS@>p|9q
ziQ20};oq!G*hf(REM2+Fhzo1}phtrm=QC_YD?>UQ>sa;Ag8~Sssc-=ElAtw5@oUTw
z`2CuJGY-%_8`w|;ZuLm@s^6DsofnHG{97MdW0hUESs<mg*2_aS^Bnb-++c@KOo=m%
z84equ%Wahul(lrU1;8AwCpE%~^oAdJf&iBU6!!zU;6jo4y0AXUA?)X@tpLAZ9VeRF
znOrNxva%noPAnWuRDX+1Y7`KUdVg&EC!jQj42TeI>+2<8UrBbpF$5^wC8qtgcC1^c
zrD}Ml*LD57^B9~s%3s0Gim!lwhSgj?K!nRm3f{d7eti!B^(p^#YePlbvQ!9DZna*)
zGW@~eNA;X9stOkq3pcGTq?BYDz$g;^cH=fr2=pj6*6BCE$t0smP@r-<*D9~7@a3Zd
z3*Y+bq~pbAUC`&0wlO`FrK?80g-xWyw@hE43TJ+VdIQphLmx(ao;=CeRex?|pd|{2
zMbV<Q0`OJ-Nwa);$)4472%)mw{iYiraj{~@HG#>I1vjbf*Aco|j276evRYAK*Qj?E
z$KJ;eBoa#*21jvkhS<k6O|A6~gOxn&4NA3%K#zqgI1zL%zcIs@0C);WXgb|Zw;N;2
zr3HvrV&1914Z6N2na2X0*5*LWsn7cS?*&}__&X{~QcX=Xj?axZ*zmFl^PmEc0NCe7
z4A9;C3Z}an+_+d9XPmh1%n1Q3cID5~Bfi3-jTucq3i!^K*Pbh=jN&(q`YS)K;>kN|
zs(Lx|T~q+aAY}#aKa4@_0^=BPagKj|rdA*)EMkZm>6PnR9=8GA{S0glbZ~I2m`PtW
z6Jz63HL*oS_sf1IM+t{1D@Bx5yC5>puXMC^%*J1c_kNuZ=zIYp64blZd`cjg>uVbM
zs~hd{m1~_P3H5Xf^!BFbRAT@iXt|&DN`0WHkyBF0eP(!2|F_uH=6~t}C3pNNAQc^a
z_s_?0NhfMtd$Xl9e)9QOp!o^nPR$$}BnBww7CU^+)4%gid3kxHXUHe2@qk~lqMWkq
z5wpK_%*3G~j<7s0aRy36_@c3_eWRYT)`}tChV@ZGk5ScC^&6n41C{!`{M2+qCYOq&
zars@t)mp-v`<uxY{_6nU)vCU@z-GteGgiW!vNtrQ>OJ^ol@4r|(PPCDsX(@`Dhtm~
z?N+83{jz>?3eS~TS>ae*-QZ4_S6tjbUQ%9MwHAAFV(m8?6VWfL*Aa?_^YN1y@>&M(
zTV)oFv2=vf#5oDllUv3L7Vgq+>s@~4L{(j46DCk4!8GKNp%)sO6x<-?3vy{LK7pL`
z1I^aDp0-+};DT33H-s=E*K~pP$g4JqeMrAN-Y<ZFk|`)W-&h9b5{OmJMb<`O%M2D<
zy^vx|?Bj?=OUALc^Q;eLx`yasEL!HWV*7)PN2T80@Dm9vj#;%Suz~NJs>ZOJyfopQ
z%|jiTo(-sL`3&5Z_JggW(AmO>a3oh}XuSVT0c>bBI~xQkGBM)_lUb!V-)Bjwt65Lf
zWJiJ=WtGS?ZOl-DG+e(Nvk>N{?_y3p6xFq2IGRt_hGCB};Z^H=$NuH+LVWrMx4X7S
zM*G17q${rN2S&uiPIxHY%a$7H*+&!!P!pq8Fp$%Movp`3XSN6f+2;YX*w}wqw{pRT
zMIep%hzNk%EII;q&u8ag?ac=2jYj{qOnv7*+a{zU$r`5J;o!V4Qn)ySMUp3rjDtm_
zpGov|!enJ+^30Gu)LqT)D7cI6*SzZP@3>hNeD7cCZdo|jf|^2ryBK3O6{BFsxR{BK
zptRVh_!$0FE%1K#Slkmapkxj8x@y8)Iyh7Yau$l4-R}b0i7Gm@m=I%wRzZa*bubY!
zWw+n$PE{a)jjVtlC8eaO&FD0+CSK10CWF?#=0rJaQSjz<jlT?3n6S9u2OLv;k5*t@
z8>D&j+vf#LQ%`qjv)XQza<o=@k+>gQl3SbhW8~1PV!cO_{CCL{$Gmn{xcsv*1rUyn
zyerG<D8OC10r72@f)>yp*g>LSJA&!~g_jBe=y_x&<ArFiC>hlRk}Xy66WW6>+PVkV
ziSE}{m+-FK9N2E4P)}D?Gj2RGf&g6fI_787X^*^prnF<iigBaWnr>Z(_LlEQUF!27
z>~c)c&%V+<4J-_Mp1v9MYiq>W^qK<jGp(L^QPBru%s@VeGlJI5WZN9SI^F42cRT}b
zM=vrHYcraXo5QiVg9ip-7xrJKy|MT746xhd0Yha8J?*-?+FZ<h3PP^X%Sy2S=Nio+
z=z`?Og-xq&i{{GY`O+0oxMVx->@H;*W{Q1*nd&KV#$bV9AEMT<ez}wLAp?Y`_FRsw
zqiyi#YnM3%*JBnSg|Cd6)}U%MaA7=dI_IjFBt?k`!T`;_lg7i`VZ{*ytz>wV7zw{4
z*Vcr;%Wk0z6NO-bZlWuaaKD%NM$`*CYuE6)V+t#iG2}CtSx*OM#kiP*uTZgVR_oX-
zn_az~c2bbI!DR+r0G90Mr$+5`rK8MqS;fEc?K)O$Kx@qG|JV*49PBvT=>sG6f%7*~
z`rPyC>agFv#uX|xKQeM&JYl!a_olQy+VSeEJGdf%C>_Us${NDcFkDgEoDBa>__WB;
zeXyt5%Jo3j#^v(0s!~N&+@_p?)5839xh?lXJp57!Q^8=~GW600h|!aHZ@k?ji0fy{
z3%vib!MhC|NA(o<HDE&|ZH#Wm|JhRy&ylX7#IRuQmXU;MdBt?P;fa;C!i4=KwijkI
zcUpT5E4k!%kRaFHuY1H;ub<AsHP@fb=k*Q`|4NC_VZzRd(8-91tsbGYgwA83|M*e$
z`e*%F9s2Ig&3CC=lq!{vl1XP<V_TlH(2K;qYAANinPcJ+M<jL)%5P==sARc3;W@AG
zh13lvb&zO>^QZP_=&E@>Yf<jip+7xj2sB!=Mubh6kJe}fh>+3ZYriNs@Of6Q;6*ic
zQ^?Z-BjL1<Mn;8Wy)6wey$M!X|G3U4de30;yQ_q7LrMrRs?C$pI`uWbbOgBJt*xHd
z?vp?)JL*-IIFfV_Pi;p|lz(SP$(?OA*WA_%QSCl1vnXEqE!=uOSosX&*~?d+bFICD
zKhM12oumw+a-Gw>tW4&^LH;+e<4G*t3Pt#ld5b4Y<;8FrKIM}weybrn>CzD#ET8|m
z(CA}Sm$E*$i)ap*ZrF>&j;}_8FIDI98R^45m98Tv#9u$9r&cYBoG;5&4>sx|ySCUA
z^1MPv0aD(+AcV`+9-d8!3+LVQAK*WU5ue9J_a8Z6<#ySdb}iAr(KBo9{cgq_!2e^a
z(a|93UQExl9GuoUY4JxSW(ScvI#`5!&Ehk%**0fD>c-H8vvBF2=~FyBPK2hiC7vE^
z{HeimCTVYYg0y5ejhWM*Kg044Y(9?+W0YvofTyZ1s{c{oX`f?)aBOA%tK>66fWcz>
zu$n3(ZZ&IDYalnwDc@u8-Y_T3$5=X{-?PF9FKgH!*W<Nu9l^alM92-hvq4={vB@yT
zy?sJxgAUVZ^^7lgI62p^R2PekDj9a=)8**b249Hr@r@)`!0jQpTs|{$Ydf}j@6`C#
zEy&EcZOpNX?<SoNN2&1`4tGb55T7bg(BGx7pMmjxh`{;g4)ah%&&9R}-ra0<p1FF4
zP<z}}@FE_1aAg+ko2%2p4v}Y{ke+@*N~DxrL<Rr%&{Gj};W}bM2`0F4{zyY995YkF
z0K6Ufe`avp6t~9HD7CZ;OTIt4YDn~dI%z`@KJU7>YN|9vXbk7)r|_*mX=&z6B3*7I
z(}lCaUdyT4p2|*(V@{a?$|po_GI7_J^E;YuChvPf<=(JMtL}2CrCo-jd}g?`K07$d
zD#~=Od8jQqqrN7R&9IWec}~^7vzE-0997+9R;GD=paQSLm_$8gHN)%V?Z{wW%MH5C
z&PQho;m=wtla{xdjM@t32&zV<rkCCz1v>UP+t5h{Lepd-r%2c+;_#JiOPS6w<uc!I
zUt6GgIXH-uuKc2!EQ{H$<M?BpD6)gbLq4uZkkUudLt9eVnj-u?{`t$oL?Vjpj4aqj
z=MNcnLO*e?qW5e=t~)7-F4SLR#2nAmg^7aG70e8}DyEXsDERhX-bdj+_UmVA_ixp%
zV~5o>1Z?<)aj06-Dw_KtFsGurvA8XuZ;ahp6t4PvbG!~?rZL`AhG5Cqic(bmuDQ*b
zf6|uuNBS&>j3MLHOX6nVE+^Yt<Uo1Tul<`x*<*$^<@z0)?F&ZXVJh#nI^$ykHiu)P
z^Nk%$r^W92?urklMN{cYC30m@P*7h-CN}3%;9Fg4!KK;m5^^@WA$lWdjFp`UCpeyc
zhxH$F9v_G0!P$Aql{CLsd%G`r_!32_-EJHuM<##T6_WF%W&5<C%nv_GC}u8&5khgg
zQ`CJ-uA_ftW1MvJtOMrK$razti-P&X!XCk+17B@&l<e=}bsRxoz8ub{tT3pQB=Rhs
z?ORwZAE)k_mlmDRj*fkjRQNsp#5<(xM4#zvq#Uib1!pWmJVKLSJtOpO{ct^ntIO$&
z%bJ?rv)YeovRL;4Me#yVn+zp;)G*ZIhy@9)Ss9!QV7CS%#@jJi6-1GRPSe&keQB_F
z|Ki_0{~UolnvtUapBJFA@IJbiFwVXnuWZ1e;_+U%c`TE?6yt9%zPf(qP(kUaQTnLB
z`Nd4JS3h3gq8iwL@!7N(ADq4?<@<E(u5T@mBI;<T?FFif)!aw%sFNg-w8~)S(s7eT
zfiFb-1O&nsw|5xsSfe%?%~&g&TU*km+XNAQ&geHG!Wr;+SvLeyaxgGfw$>UG2_J%B
zpBa?x=*@vc)xdB1B-fV`miLL22aR;jp}tX74N`qYK*HVF(W}g!EJ|@5b=wOya?!iO
z;0Bf8Ys12`#i1i*b$Rd2R<93Qy#gW#2uX6?!7R><9{F_~@t9}rO(IvRmL_`|=LJb^
zkM1Dho}P@l^V1;-Y=hZiCS1z(L7Y$;I_R<edOQ~*heH667h-6QpW_8$%W3t~K2!jY
zTp4<<lo$8!F}Zwk{~PdYvCEP7yWh&U5;7R>X>0Lp?B!KkZN0)PP~dG!U|*#O2a7_4
z{DLkAg`a}r{mE7<NgBU?Lev+6CGguSZ|6%mVQHQfL$7&%w6yq&i}JpY(9|;!6c$H>
zDKq7rSQiZuWOV#U#zY(u-b<-8Io47a8a!(I%|;u%t3ogSI;3awi&szqSQ2lc_MxyY
zKXf&JpH)+1Y|+_2GM{W@TU&>|_tA5&IcO8H-}IYb3w;<86inBHel;@q9+!Yq?_7s5
zS|J*9#KBGVNu%8reVYS3O;AuG?RdFzsNET5h@fEA%#1v*gKX+|e>nEWc;0W|*q?#K
z741P;O=+p$y?q&x)vVSy*c&S|q2EoEv{ZeuUjGd;#st!RLoyng;N#=ttOT!dtHYxt
zo0p7WVoolO=ePHS&bablk%kKqt@YNDaxCD+BVG(NE%a1{%UGpL<s%@Fqjt_$*`?%7
zhKIl0u}z!q$3FklLpLKcb~RaQPw&8FCiiCxSeitOW`CdCL9hcT8k;8E&#1|1H)B}z
z$SKHaUqfIK(2pp3z#j>yamBr3r8%6M3goD`u<3jdz2V7kYm?;&Lxk{Ri>lw_9z~}U
zLEx##ZC7BM2J5PR;$cL&rFrsZtjIEo!#ZXn;&mS{@d%qpk97*%Bn`kb9-PC8M}tl7
zIqBH}tlOS?Nx$Ov?#Lr@SD~}BDu;)Mx$MM&s3<5L;YshF58z2Ij)i_9_>{RsKEoLD
zrFBS@#^VZaFaz1wPF?LAGWb_;aZk+?>h}Ho(06inkKU=L5EUguEVa@sp#D^WD(}+-
z3ivYQ2=m$kb3d2o63zR6yY1QIHdM>Lyz|8+-~9xiAXI1D-9i8yue9`Z!*(c#BdOAd
z^Rl-+%kD>F{QPPw1ME3Mo<aqD<RKps8p*B~Epv0T51)}l<g$2OW3cc&6C!54xBxyi
zbiAxTQpbBOri?e7uM^#_T+!RA_nj#AuI^mTt_3or(iu0LvIVOFh~WJ4U0(R|@-jL)
znq!BdOfEg|s!TeYb`<<5b01G(n)@{d^)y^WmIS$wA1WIg+rY6XcFN()moFgzO*Ted
zynGfX7QdWS3f+gR7BiGXli=Vl$SQqv<K%2t?TC(9Ri%$TGT$8vn)vcB%Wt`DKHVF=
z#t`iM{9LC-BLegpCCC>YU0F7{<C~QYnI68CGqDW0Qt0)wiX^U{5(;@AVNsp}Il?fZ
z91~@KHL7}yv1-B{EXI5*7umk94w&rJv)x;J`I+&GjwD;aaD$)r85|qo{oF^r&$htL
z$C?uo5fl-I`3yy!o^j_bDjI6Z$tZ?RB4=mLA`<dmaB*0VHpq*S&Prk$>m3oe;2PPQ
zxw`A?P)#vWM{y4l8fF^A7!@_G^=!0U?|cv$)z#Ij>+6r(Paf9`iQUOWSXo(p36V!i
zv|PUXc%Lr&9@o~_Pm>Cj_`v}R5H=|?Q1xiTdwd7F@QFBn{M>)uFjE&HUt_yA<NKzV
zmY#ySuqcY;lWi*t+UxeJj+65Ky^Wjp9;do8^>5pXntB@FO9L4vLPLEA`kM=pr5UjU
zNvYi~>>Lb7Z$*AHW>Q}TCf90wJdh?j<>QOq+6s-25?4^+XxyFeDXKjiuIIHxYrJ`X
z5}#1kUzB^{fVu}Qag>bk3(JuBhR=jcPe}NOP0o}{eMti`NOJpxkOAw1@D|I3qF_3o
zT7zr-q>QYCtmXo*YvxUbANvq3L?FW=fHGldh#8-oK16wFXx1-C_uw=E-e4jUk3_s>
zs^gMOkLoWnJpn;=543U>H6&k;L>6gp7IH$4t!>x()<ri~mpvM;7ebb8rADT}v4-3i
z>bf^eY4I-`y8CJLhO&c<Vh&)-4BES$Vb5>=SgRSRsOOn`l9$IK;4?$sPG**)y-wsr
z_VsvVqEu=;fjjO!`_*A9QmOp~@Wxolz(>A+UuL4<^4)o=kke^@>!qIzdEJg7ZF?4L
z{j4E(*`TtM&8u3T){CXj5dOzBBkT2u=PM(@n-ksgdDwYM?$gwxC8cXZPLDtN3NjRA
zs4kfu1<sKA`MsU}^{hs7RXO4?2kKei$?BWpqTRO8(%Ubb$C3gaPm{~rn3%3~$l}O`
zlq>bsao=k0;<-Nl3LW!>naf-rEQV!sxZ`U7nSk9~-?Za_eVvHoqQ`2v!#1Pq%s1nO
zkSO=(Z@C@&9594_4^0iz<?>>FS+NS+?lcd7OsXe@Ep?uRI8=|hWCQs@!=OKaiEGyG
zNY=})=1s^?fr(&lVoK_UUX$Zv)9+DXgf2rR+tqI}WfQTA=7@;Mp5f7mD@_z-$p+{R
z%{;jxj1V|q`ut9?8dV*h7s1K=kP_g|y;qnLyE5F3J97i!89gD7g&k~`daEP@a`!GI
zz{J!)>Eqjn`TDv;A+yp{5O_v;puN!Zh~$@01su>ArM=)EFOICZAW})?<cN6wxNt##
z6z#wrDPf(C`~rNnk5a+%@VL0pICi$rk8PB~-76U1FD;JlFfu%m(=oM6;_J5uBy0|R
zR`y2p&RlF_0HK1C`1q}I?Qk*F8~*C@RU!^aXn3Mnk+@!DK>YnX0+>_n-v}g=X6+qR
zthwg`Iu6TGw?IubThvUaig<9K`{jocMMA04UvJs;xcunk)+2kgosK%}An$j$Tjt6F
zUaq~3kM139C4rou2EI_DIX$J!g>JWDIXi|)N;GXGTuEH(EfI-QBEsv=6i5J~&)#)%
zk#0t<&fY&qIEpYzXo^wuof(6X%&xwXYD_>i&dk2>05w3B0kL-ov|VJV-6b`3Ls{r_
zIpJDosppsZuAY}@Ib(G-SMSWHSQX-8hW|XP$=N3o#h^oirJx`vx^69R^(P;dmrY+i
zKMYyb%lc5rUuls<WG8j1R`t6s&^0AZPb6tG>zfH{>g=<<RR_K@$2DdEEJf8mJ#vTk
z5)zU?U+7T&O~jl367hb)k_yydn=(H+f1{^y<(h*9>%wib_eGfm_A>#2LjB#>m>t*(
zgT8S5CxSgR$usIR`@zD4`s^fxX`A&1u+`n4b%Pyaf<!S5;#ZC|m(nK;{_J|GTT$Lr
zUh@d4T@c!jHJzb9KHvW>Xthem<a&US>E)GCDW4Z}Iz#@_-%nw({E4bzl={Y95eTC>
znsdD<j>y|G=c_X`kfPvTj&7(w9E~daF}x*?&5`(E(W!R25crWG<Y#C|*98`H(#H-O
zTta8(ivCM@P8=L+(?q_(#0$*8RjS{DYy0BPzP0uNUWMg~*Qasf?`96k;Nb2r44D>J
z@R!{|i?xju(C7NfMBNAbCL+2m)r38lO!zMnBoRlky9drGvH8=sKhp!?`Nq!D>r1sf
zueJm-oEFNTKo+h~Y^#HtU6~K8J5DzEKmk!@GF`hNB=PO;bugZ(HNdE=z9V(Mofiaq
zu0g0bL1_XDzXc6E0k?Y_!|u;lSO`eil+8bZlyL9_!YOv?1wsYu+S&}N%%*GpVR}Oo
zV>&BYBL)-M+pL`8FzRcwCs+){71~lq9ZG8#I-N|>o0}!4O(~ePezu_CLY$_@VWm}7
z(Gn4pZ!Q~Ik<dP+2%QLAsp#Nu{U{o7I^jF@@>&4|z?LXhD$M+a&eWU!+x8{p0sls>
zyQfb$IUn2hYY#A+_}05TE&nq#^r!oWv1#IY1WqX?vg6J36IGQ-Ml+`OFfiiyNBkz7
z&R&Dj3fo)O&E@UIey~3U_Hj-KrD9(BSL_f04F4w#<!>gwY%@f=FJTg*pufU3vCoT^
z2Mx6pqV2?E@B0U}->nzW;ps&6Xy+%Ma`ny5#hhF@M8d+9a(sw>U!I?1rsM1Ror|p6
z-h(LP8zC_&#FCJ(A?}#%jl<a}8{iVc${F?0Tn8M5BTJTk({gcG_+fP3=<n$Bn@CF^
zuhA`ATCn21k*AY4dxEYvE}Fy#TKQVXf_dh-O7v_h868ztQ}O5obj|BT0r)!-;n3|%
z)$t0JWjuRQY`gk<I;TcGXr{|Gzi=YH-{utg>O|mPx!WP7p2Q?}I*EmqLMin+BBp&R
zTHz4R-`@{xg;yM}rFiDan_q+cS-|codBjVQ?wr}Lp3IbWm>rx2T<yYsZt1t&@5J4R
zo0RCONAOyIg+d1RTg>+z{wSiX#z9~pttwloRricekcjKB&1J%q<WH4kYhpbTm~Xb&
zsu-}od})#kQfoMDp(H1}&*YXP(SK2ji9#9?Uu*M(;>i=G=@~h4IWuPY>r^dPT&wcg
zY9W90Moh4&8Aq|q1Xfu*0iT8I`(oRw_q07LD_yEIZ00MR0M8P0vLQ7}Fwy4UobzPD
zW3&Y;+ZOMoa%T}}xkZOEN%?x524d1R$Rl>x?D>hb{#f;0%!3@8+ZM<K)z2A~QhxAx
zBmX1G@Vo6?b8GuW)fnI7(b8Y5y%dB~fFWW0G{J&{+B!eD?|u1p>(7U_k%%0wK7$>7
z^}{RO3MLgnf#uaPq#6d3Ln7N1&*8R$SC4vJ){9n16f75!T<|A+>wayeFoDtmHZuDe
zfYao5HV?w-sOEf=zXIL|8}m3@20QD$v+wiCZt$nSG|tnd>US^e9)b^Hd1P23XRInF
z4!>YERmYmD?~#w_&~?eR()zm7(WPBO;y$Lv!NC{e_aU(np5-|h)D=us({hCy*BnM_
z6A1|PT%W<&2IW4EU)~e_ERtCMYQp<uVp<_KT82ee#DdO&4kPG&1V!dVIVl&qCiypN
z8SnjbyTt{k<f?Pd)I?aqKlVQ*lNW6aF%PC#H-4v2Vqi!vOe1r~Y=<aVQ}qJwvhBpj
zF||p#Pg{yitJYe_xUw)^N6(Uk&8upZ;n->wsP2*j?HH=|eb){@<&)nL{~u3Z8C7NT
zy^R3~2uMpFMMCNBP`W#$rMtToM5INeOQc)6ySw|)U5D;C@8I+MziY9Uejx61&&=NY
zioJ&w&CpUE#Q5l7W6k_(H_~K)!g#R$9l0D-_7j|>2PS}Twba4k5!1w2t`i*y3<xX-
zT?NmAMEvR+_+jVg>zw#xf8NB70RkZfqvIf1TRgld+$LZZ@GHQaT*5Y-?T~p7PeQA~
z6C}t}+PE{PkvSuS`+%UclD^#S9CdT^>sO>Adk+dx2@^+WRm#+uoUq}{RmWyhc%yly
z;WiN%wcqPmI~DoxD9?mc|HlXAAXUI3tYrFo@|_CWZ#wwq;Gr>IikU_x_RVnY7X9V+
ztPJufDf_>r?Q>)PDfC(@GX@Ioh^T-$SSD~t-pZgr#g*DCpfgj+#_;y`2B!USt{4&W
zmdRGXujOp0sx)iSai$zcKKc`r^(&jL`=Cwn64{QEBc}cIxlGt|`|`>^(qmNR<_eH2
zL$Rrt&cV^gpP$m(@Z2_K+-z1QR-+t2!O7pW`}7JBaj!Wj16s-P&SUV|T=In>sWBfD
z&2<YZ*qA@iME*#PSJ<vUX2JQgLQVSb#lJ{F{?knR@L?yKXZxQk3k%z1W<?e%7!>5+
zNh~y9t{8kxYzxssKnn{qJU<|wb~t9Eskkn&c3OAP-JS4{H|+Xko9q7QylE87fdd{N
z>H8U3IUk?geUFOMjr^lkGqow5^lBK!d3$CS_OH0mDjoVlhbPZM<=+Nnqk!r%vYFPa
z#D@P-)d64S02JL<sSg^uoi4@%TB=T+dIi-J$8*~zl})QjQZw9-Bl<9()kD&qxlg3g
zhQh&peT~<SIQbWljQ6=-R>Xkz8pha<yV;?2<Kd1WR;G$xF|ic~j2*Z6$V&|_^JKl_
z2I<cm|F^cX)c<$RK*Da&^e9^&H$M*)W%`rif`*TmJF(Y@gY@1Oa*!u;6klQi|FEES
zOxxQkbQ(%$Z!YkXcS!D~22YwAUSt%JC;%KHi2EiT3wKt0E`I2#GWmxqlBUHvrtS5`
zk`mJ5>QsiG1Bv;yb?n$VMfIqt8K?stneA^KZx{g4Vt1&7P^Rl1#HdTA4CCu-@YZyo
zS|j4M4OGi(t~)u`LAbswr>nAV6lz&Zjrj{Q1_B>F{mZp&F~>R0aTc~$%4utI7DP8~
z<XrBI2@yV89SL{HGxzIsJ2A?a9+vL~P2i1R+OQ%46-PNnr!t;d?PgF`#tg1K%ju>b
z_qt)YJ~`$@>**a3Eix4$O-<us+X|1&L{o}Sf|kEru}vjEK9;LY;c1g=a=g8q6rsZ0
z3d~5_&50<+Y}$i%006+Y#UWbvbix9LXodW{xH!Uv`M-^2@$!Gfw8gPQ6@wMfj}T4o
z2zOKo;_*WY@7(vP01sPkJN{#~qMY%$K)y-gT=SIgU6=|@_Du5(+YUE6>76(M_lcR{
zm)kv~qP>*9u=FM-r$Mh6Irl)$o3jx!-nt>3Piwcuium}<d5zR3ycHcYrz88I#_y@p
znJwFqdR8WvYzX@%h!PTJy59x)rIAE5ll#PR65ev$oL_%1MnX%U^|LzZM|;Y!omGGT
z*5n_Jc<U2X+|%n*v_ajv@Z=IKv^tkhaq9EpXd|#9nKN@aI|G)n@RSuNBV6l+o6qTJ
zO=&?|$x5qQ2`^VIOr_nt9s>lp9OBZ5w*dM4fEu=DjyvkzkLoo%51FFy??N>TvvfCm
z*dT1k-cjOko0=;30JU@i$NfX{hffHgN|20<0iY!NLg7kNgSVxQ1nQVxf$D)l-A=SV
zq@sZ&uee(wnAGB$ZsUb^zW)_{2@*GwOaI>aTjeG-f&Zy3UY;<efE#fWlcQ}EL4-yk
zxP@xA`s$ti=}SHM8D}k$$!UU1+{TklzRK#G&y6MQg<ZEYK+pNn;~ZbQrVdz58du)e
z%UGd0e_t5uV+lI#cGo^?Ak%_+zn7<4B|XSWBH?{3j$?Sw<q>MjVa?!3Y;*2Xzfec5
zfr@xQ!l%2#hLhQ&iqCS@toR`HumRdWvr;vGLPq-(2b_%TnnLu=&M;7=M6iexa^@Nj
z+$;LWL;s+dgTaL<{8doVAvG^9yw3-WPwlwVR+QPT>RqDz^)G&aqN^4n?dFn<ilKij
zFa1Np$SL8J+Pag$)5`whQB4Y#?2Ye-gNX1nQ?r%}t;&*(<}{Y7WiZCInWL~rHe`7v
zCW0o(@PiBLCTHW5=G&>xXx?Mp2YCN<S>yHp0)i|BHcDT1V+d$?bsS1El^tdeyAd-r
zW;NQi`RRBPH(rdnPsE<nDPx%eD(r>3b#)5tcgc&_#Jt6WL>JCyWO9imO-dO`%*%lt
zYqYqTmGlhX?qz^DqgQ^i4&I!Z=CQevVaK`t9rJi3>*j@EH5s9X$ql2PH=_ES=3c#b
zY(#iUmP8x5ZKY7pw9XC4v?6t4=y588Icqwb41c#TB}rsMXSlsyop*{6S&K779~)lC
zJwM5lHZs6wRV$<~^N!7u-CCaayKNQvW<57>wx$wR`Q$GrTT&H_prDC_6CZ^6up3nN
zh(CXSO!;cd0Aiz1PuDN;;APm&tVelv-<(#YgmHb-9_NZupFo695+9bdva$`C-UCee
zz3i;=Be^w_(!W~@!_=;h<GJ;s+A7I^YLo@f{~h{qkfL`5ub#!L^v)qiFkdw@0~tWd
zx)`Z)5jAuFx(Rryg<?%)|6v@u#e|i*rYbPRE>Jz2)81+tC`A0T^364pf`6iZURodA
zE6wfmxIThwRhe5E4wzcg3Jkm-)8z1)oFs_W9M}Vz^~~gK|F2}d!B%MZcjGS~o<ol@
ztCYuj_Z$VN!WoFGN`W>A$_K0XE5h32QPXLq+zGFeb5{hP?HNR^^k6(W%I47pos!`x
z!?xx0b^OAM!Bl@hMwiTRJ5B$MpZM?^H%}$AatL@c-FN8Pst!JAWEhwQx_Gg8hy*y`
zTBNWeV_=aHL<swnxjE-uF4Eq3=;SV*={TUFUTzeZinrCx&XziFm~QRvU{^a*b;U+d
zP*;SWs1#Cd%zkk3U6(e<feh@_mcdrd*9R~DrDMNCIxNS-2U$vpofuk#j@%gEm2<WD
z7bsy`HVY<yO?5mb?bE_*2*2?-U^_fOhM{n+G;LWaq)%SOWOVKa`z&Z4L+L=(5DX^~
zCyyR+YL+QnWrLhjDz<pe*sZ+;+Ng(B;#nZ*f^_HAD}+q>)QKwl)W7J4oe)R>lCC}!
zt1N7%+TnZ?LdfDtCjRD+S)<1r@7>9?R8LYcV4Z1d>J2w%#8SJ_eT`$RawIc?lhx?=
z0X?s874hcrHBX_9ZDHMS0be!JHLsd~#|);72g^G)@_DB;g(rVnuw0e0rj3@WX>>uH
z7rg()`g+Q3bIIPP`bGDl2U)qROY_3?GvmnMoC3P=ui~ZqYHz^>r4m057nUB;CGbh+
zlLRCf-z~3<Ob;M?=Q@XG-3e@N92ws_?>!Z4WXGzJQL~r>@ZSD7(3X(JQ~>LdRuJlJ
zfOJS27bh%4vaoiGDDG!${CkIXN*JOI^}D^t2>!E`8yDLUeXHNa&M6P|_p4{@BOzyD
ziVwjd<tM{Hf`jwAL*v!1t<PLtgrJbkp?G&#SEcn1taA!SH#6HcguvmKM)N)|g|QK!
z%20$^JquSq)|59-&qmXxxdutfwT#T}q)p)k1dw1<ho*%m>Dp9OEQEvqEWnCxK|Tz8
z@`P}tL^yt*Kbho6?;PhLzq<O`>R%)?xh$f1T4_;RT0YPgfODlUTRLlQM7u|Lew3Gg
zq=FN{I$p-0mm)B<fOc<;@0R9f5r`w`)dwnpE4Otb;B2~t-BLF4!Y|LA_pCHdMgpD8
zpntImQQ1Y6#E|Z*Yz)noCFrXrHM>SQOIil=aZ08l>|odEJZv^9CZ4wPn11SXtAT}c
zm*zKuU-h6OmmRVZ*1OAV-Sz5v`X2*b7}pk6{^g<+-~Sbp^4|fk7_#USHM6R_@fl7z
zsNEoDWkFMOh4^oN@|px(t&z<>|3rx3wT&o2gl0wVSf0Zd3T!K=W>hlC2Q#a^H;4;y
zYq;P5nXP{c{6Hx#Sp|FxLzbL1-hvm$FL$<&dUCL22>HIRe)vFwUN5Ag!~#;sbX>er
z*I@X9zyc_i_0n5mPpS-qs}3w}vnPD`fY13$d<ui>A47%w38Lf(oa<8qp;TRxYTV}H
zx^p{LnuOCmD<dUBe5%e%5BKBc!!)*{Rn1)J*91y7D*Osf9j^wp<AazmElY!9$g+!*
zObwzg5JcW=G;oB2lGTpu)sLtT>Oqo{uUR<>v>efPc%~oNc=`su4nK>3JBn}We-u^<
z6Plr6r_*+S97A%d3k|!HkmgZ+Ig&{<t<vL{3lUiGxEdRr1KX}ai89gd-s4Y2b6wWi
zbVwexr)Otr#l`ZOj+Y1Lv^*CucoagVm>5(}_6_)(WvKPy4O$*!b(yZ2m5hA6*rTIg
z@@jM;?ce3|r=5)qJ~a=Y@4uXF)X8-Gm+?Nh;RpTax_HJO9O-j(8yA4c9UY$8KZNAA
zmz0QE8I`W0=D~|;V8&auIbY4iuvjClZ9!91Sy_I1+`KSa+G?NYkXni-1@a}x?|86$
zIywrLcR^pr+16MH&(+kxxQd41H)EDWqWZ5RgF3n-)WyRaqAA1G@C3zF-5c;@w2P6V
z!&Q@6934@@!e77RCk3yCPdf=<GoM-We~RYeW1Rma`4<+r$oVU|xTQZ~YCLLcO4Gv3
z@V@)|1F*Si8j=rhnD^)*^K#}eN(vV9qoJe*7yo)cKXF}7R+c|N`E8Q(CQue05L*+7
z`$kU3?~m_N!8W)KUw<1-v9|Wf&?Hz6F{_x%h5LK-4aP%39$!;~*|71+SQ0*2=;nHe
z@F4yqC;sLKhy{qWlO$a2X`kOa9G^fu=D`SiK;IuNEPZj~;xWS=YstG56PeI0#%rok
zoIeImdk9ziMpQPJ>pl`dPH9tGZsW$_%z#dN`PE(3e+!Y8_dn|Q8N1Znx3fqI3XwpP
z_l0V&BPA+=(}CS&dC(k8)iztZ-XgaEJ_t5D`(+@X0@2~<FWVH)CMb*$wsMwR^^56m
zBg4ca(i`<CSSs_v;&u3)&<A>k<f!O*;zA23?gi|Ft+Vgq*v}~iJk3>V?B_Twx`8&q
zh3gZse8yM4kZOpfh?_e>G~kc=B0jD?LsZa_;*bS_rG~5mI`_#8_B~HR(xsQ*Al?6C
ziK}4#Iq4^%gOuCD$@ZTksX}LU!~|W$g17D4`L7YWozZhtt4P0@oaY4N8c~YBz$5K1
z)$zQv9ufc$qRWr7Grc%p%LMAdF!(Vs0qD9U!B93Qboz2>g+eMGBO>{a%KzsD5LECY
zo1*Y!)Lu*0^P%K<hl1c8rTLMIU2JihNYsXC+n)1TJ&x2;vVzZat$zRx^`XTOOrG2F
zUqI5O_+LO03NtV?5v%)#&V}O;=J|!!?u;x;HEGUNEuBVbbgD`xH(dd_!gAoOTf(P~
zr2k@`$D$%x+8*Vyi*iY+?^_nk#<)@y7k~74|DARd0SY#X^IW4;ddk>eJFC;-$Ep^p
zSV68sSmhVA20|E2yBFx%_0A!n8lhqJfh@*sbsJmN5%#Jl&*~fI%%{Wp7?Gba{gvHy
znA5Gck7{w;i29Wh9u!7AfIQfe+3~{Zl+4G7UywGE*K`uQ-uF@dIiwwaP72spG>`!b
z9dOxPSuSbX8gGL^yz1YpHxsbK7#u(%pYnFgFt_obs~RexT5A@Hxh+c!sj?|E9F(!5
zmuk9t9VNBBI2#MJ1-sofRHBP3k$jSLzl$GHL;(@8*=l1^DzUckg;oN~DIfsZ+l;1C
z-3fGb7M`_nri@!CrTe3BU-G*$TCJTy@NwLze~H>Qy~B6K`Bymy4ckbB{>PZGUmJ}-
zLwH>28(jLTuJ!JOWd&v71`kMF+v(q8XJN@7o-lzydL7X(Y_c-c+<1XhGZvBh=+ku$
z*7~5{?DAbijSnNXHxBW&92kv<visN0;uZ)t@5!U}cSllE4i2)%+QYh?PJwHPycqts
zy8jYrWHLK*fh3tf<Fg0V+<XxLVeH68j>_rDE9SJLU^2(cTylD2fFZfP?dIa<|N5k8
zbzvh~Jh)DQZ>PjQRsQAq1ickM&kHBc&_{axuX!2a%#i_k4)iNzPfQ0W`FRtO=CfJc
zY@RuLi3=)05t-Ore`~<wpC`LMu~dgJEYx@W;6CYAf7=V~EXQj5Aw}`@K<~Ycrc8h%
zoywYnxuw0nTjy3J9!yUs=3u4^SVP3K(&aZf;aNAo=EoguZ?=>Xu~A|MWw$Rokv!yi
zgBwfmfJ&whsFoDDm48X2-tK=%<10}%h-CcriBa)(fn)hyX-aq&&?<$<z7k_$aK`n)
zQG7^P7FxV$Xaa=%k3|)++A0rTQ!>i0?&o~j?i#e?;E4gUf_sDO5X+ozi3|TDSJ-dM
zPhw4krhm`g*127O*2<6FVm4;P&JA6}BwYtec&`=wsxQ5Sg<W6g&{<8pF>f1&V8==s
zjpB*<<Pi94wqls!HUK`n+3=S2aN$7rte3B|r~T2>r}sm4F$2FLoAjZa07jI!S}S)@
zDYL2nR$)2qEAH;;O9!v9oT%YtZuu?p<#T$HZQeTy4ri|#w-bDD--oiv%{9Yd{Y(>M
zFw`lWfmAyUacVq|e6WC2-UEux!*1RW0391lE7=eB?dZOA@rTs<X<Th8k3a%m%Kol%
z+R&-;)>h)-fL`K^?~xL_Da}Jc=l4Hq_0{-$up`k^y+ZSo=`S9;+PT^5p_VH5?#D(E
z^)%Y@K7^x%6f(M$Tay}ZlTb3rCbdSWY~Sy@Z}eX&%>R;R{sf=DgpTPMNLQ6ZSAY6D
zx<R1#f!30K#wafC8lNnKTHnTc-`SME<ik{8y~@rst}HaR8Qh3ZeI45q=WzKH;L2=;
zOd3~W;6>UlwWlRSMD$`acqu$vL0@)v5E%Xfo9qMqMP)fj$zh~ow{c{z^EQh+?RDp5
zfuPr~rN4Fny(MThnnGL+4eiZ_lTG|LRC0=Cral)P?7g3A{<ZFTZcj~B*?*2k-5{5L
zeI^lJ-M)ah-z~)ekE3xI<2vp`XuacHH8O0|OZ8QPy}hJ1?%b)s#m5_QxU8CXcg|DL
za|}c(_qAq6NG;G5jzpRQo)Pc@g^d4yqWW18{|=d;KRYFawPXsj?5wQz>Hgi$NNc2~
zELb20>$|Vksh*JzjGsBscgj2$te&6t>b^n^-h45Cx$T5hX)f#LfmAqo*TIV%;mG|O
z<qTdS+udWV@#zJ+y3K=w(yqL8Wy$XfSQ=t$!WcPkgZ}(|0RVP28>vEz>z|KpWROor
z+cE|oJje`)zh)|xqdFhJo!IrYm-l39j3s;p{EqIoZkLKTLearvip-(VlGyu1_-p#6
z&$+@#M-yMw`vx|<ta%X*32PM671QAIi_-}p<cSxmm2LUZrR@Mp8Q;3WH&o24c)}n6
zs2*&4XIP>%c%Up-HPHeM<o8z&_@qRf<m@!xN%h9Nd{4wPx^$nMwW%wC6zO-t_-XgK
z>z`s+S6fe{FxrN4pY}}vM!pizSOT9g7#ipVUh6DQ(0t3ku(PPCTE1?Oz=&~d_-8|W
zrJ`0rU#g~}(hgclr^`$+zkK=9-q8_VfU6S&E~8dam-?@oG=%Q&bKi_%%IK9OJOI$@
zRkX*hUSCR7t)CB&D&~iDONNniHn=&ZrG#TZ!bv)l><vf8Mn>FtfnH*je8(+lGe0=x
zdq1%A`-K{Ez$a_7_twfdr)v(t-1$s;=<POZxi_zm$EIu;{?}QbPE3U2a2#5;(X%Ea
zp6hBrTrD-|mF=a@pW1BenB0I8$o@PV;<g!UoP<)-dD#&7Lpl@q`SlIWB-Fn69fm$<
zIUw*REiCxRWY1?<>%_nr`_Fa}+zCy4T><SQ+t$L5a4EJo@9WyD+mZ(%#UWm1_hV~Y
z5fuFm4>uukmD)!h&Iz7y(?HJZdPjV~{ViozEz&uD-S<yqIk~@#5}k9wcGyA|!npgK
z7SAshGboYa;J|9oBNvYFo+k28a|q@I-w*)<E6WE<^6ngAd<f2S4c}CB?`8jV{ggP|
z<BDvr+nn}>fqUD$^O%JH<JJOIrN3O9@6S2Gpz+n!)jQ7^AC5sQTKo9$kM)D5uA8=R
ze?8!jTtG{hSwDZNGAz2>T%{}eOQ9BXi5?`$x}ah<9C$?2vD#iP=2O#Nkbb(jLemC%
z`x?}DbKSYLGRxOe?}12<Ng*TZ3YU_3+?=LAb8;HfRie-6ci}UzKFkz0%jr}&aCAUm
z_x9Bf!HDEf@D6FHl;o<K6cGZC1L1rEry*cr>L<$r#onYmVt`gV3#Rt&M--)TPudhp
zJU?5Nlc77)NutG*znq%c<wUTvQm8RBGBb(DbX5NH$M3tG+`EhG(dfSTi&?IPoyCag
zoAa_8VE%)Bi%dK^xOmLz>4{W+zJAu&V79cPlzW@k1|;k+y{U`g=$?<@WM~g>D#s!+
zGIG37i{P_#EW=2KWK?!i<YBpr`xnkPZ~XD;HGf2%9IT)&EG)QrdZuTq-?tL~+rrNx
zl!s=50-_c^EoX3c(<;1(O$e5u>KJVOBb4skXJ~BE8anbOj@jes6(Z^%Y@D|WZ|!r8
z!!m<%)tIA2MZ3>JhdC#UE^j}BEsp?<X6&#xkrG=Iqlv&>rI<fqw6^v$E+srS@->Hv
zNp}XGuah7LWd#?+atih0&bL#oLh||98v%AzT(^7$ZK)&eQB~(qrF6gV^0E_`NWoz2
zr(<WB@nChyTU0dAWtRCj=b5Lp^mEHgSHGVpbG;Qx_Rk0QB72>{_ZRvVJQPBoi8*cl
zI`67sKm=qYB%ZmuyT>IauEJn)A^7xWQ$?hOT2-AL9hsf|ySmBa7VJ`)$L86u#}QO@
zmvm~K%q8uo1_sat1qJu^_VPSf|Gl94=I6klURHcRL;#i=jQs(Hb@|P7hEguTu-$9X
zj21mkr;py!joOS*0g<exL@^T*s1AjbD@Nb^Dt@6Sm#5TnXGl@lYN+yioj^jTCTX_9
zLX{_iI&<xIGI!0BI`UQ2ZjhrEuneFY85=V@tVv*ya2<AA9vvOEcXh39Zgw#?-9Bzk
z^}X)_H^iqnCHNhVgJr!qXo53bfxDOs|5X2@t6I@yX`+D!_O4qpGw1J8+8;ArlgEl5
z>Vg{Ve<l5rz;)l=BGxiASp8<0zQD!4;~(DAl>Y4mCb(f^^&8%PvD#!sPme(6v8Str
zdERsH)1N>E{H61ojCLzhOp%h3Mn<NFmik{38-PDlb2LBGHJT&>7-%uBk?nv22UJjs
zY*PahtFORgrjx*)2@X}*hNn;S<e$sC{(L{fUx{QF3jLeOg#n11<muhA5yDdnH{NO+
zrfGEDS(U;wAxsZ}*I4;2?&bBh!^y@#cCu8Ma-EBH*km1ciZ5882Jah}o}n;e&Z3>e
z|F-TB37D{~#KJ21oIk|7fnm**(%$$&+a~!IU3DZq{RT)UE8&YgPS0h`p}U4e=eI9v
zie96*4kGZKjr>4|$!Gp}xuc%Tl}AnqkVD#=*j3n;0$#3aeD%1>hRtOZi06tK9&YWY
z136}LNP5QmEp5OG$!d7|Y=M*w!FxWKu{&D*_fj?~{BVAw(CFSYxD0vlQKEdxMgt|w
z!q!+wyrEv}WNSUN8r_mWelzmLj(EB*M3t);IT+VH16ubLTye-yVO1V}LT9cp7l8Aj
zh2q<_K5Y2?E*EA|#C3(`0$&^WoI^j?`(k5pU7ekrK33bk_D9C-Y-xE^qaChUNX5w+
z`}ONnKY#xjkfN@y<u6RCU-6lPcjfHhK+eX7ixB@2MdKg7{ZcfL@OOQju$9pUw2gF^
zWD0i=d>1Y<{c0LGWp6PGalE<SB1%yHkt`#%hDXkLKyf>20*oCG|I{{=McKhus55n8
zlk2%K`9!2Mzz$H|Z=g7^t}wBU*?cts>f2j~vF(LF+`#9dnv{7ihZx9YurLJ%t^Be4
z6l|Jk&&9bPky}U&L=Uh^k=TUY4b7xbwCT=JRm&NNo#%6=zF*QSqC4jv#Fy?Q_owsd
z!sL-kU8WQjiI^>;ncUY@9B+?Vy1&;+svMHRQgnun`ERCOHt4_)%7F*!3!Seq#{pAs
zzBjmh_e)Ya4P}Ge-6({AsC%cf#{64njwX4xmTGsgH`ciB{Dq4>4sd^2kZ!h|%{E2b
zs%BCFucgrSQ5Qto2jkf@HzTGrAPrAXzce&7gwD?svKVzo$tk`_sSPg9Qw(2j3yi%@
zijG#DSpHxC2X|RlLnyFr55}|a-4wm4A+)cq@`+zwgk}y*>j4~xi9^f@L`&w1vyAW|
z2_D(&sI-$J!_A5K@~!;6giYDjWY(?jb(x*1&0JW7Hr47x%DiGOD#|Rsdj3m`is0sC
z)<+<BiK}L*zk4tlI(BY?7rQdEjQ-6RM{GCNrv+pxnS$Nlrba>W$pkiyhFiyzCP==7
zp@1lRY_piQKW<Km<`hj$^De$pIUCD~^SM^U`SW<!utD2<S<=OkxEc2}@UubAq`(fJ
zHqa+c*!U{6)>@D99e=V7#1~j2mRMGax3*cbEKO1IYMvh4h~-}hWzwwm7I7nl9_ToR
zD?0B&3s5Sotsi_-)rnmd!I4<LQNxbJOkil8#7(7YQX0tFA?bhr{_wnjJsK`G?vvtw
z=W}&z?Bwh`GBH6(MTPV^k<<2gIS{h4vlAQ{i3K9{#%8K7gNBwA4lv4EL8~M6+rH`B
z+D3f&^5|@5Vs*c2v2$@zZ<PxKd8?Vy0A|C^CGevuDP$NJ7^mA~ft1oQzt(Q`Kw6}{
z!6nf?H}&t`$W@`4DKSJg><n!Ik)EHQzrVlVY@wbzib~et0ydB5vG_Glbsce`YSE`|
zZ4(?eHD|Z~>JU>oep6G;iHFZEq?*(0*{$y6<^2RHtJ3(K6)fyjVBn$I1rdNs2*A!?
zIH7`|Y7Jl~xFpiyWX)DYcSd#jUj05^xVHfzVanN5BpYMmhknO=Scu`cI@()0nm-e9
zsb&{DQm3JPWRaGz;U5BH@=LQ0Pnmmqp5HhK1h=-b+yE)-c`|Fn<+ut|vq9v*T%5Tq
zU)=-(usTnuUdYh*0FWjLnGcA`v9WviSRMCXo;o|%xVT~2bB4#voZB;4$IAg=Fv||^
zf=Z3SgTV{_D>-Npk~B_;t~7%1cJy+(*Vlq$74n1Q1RS^ku%C#`dMLI&(A-s7WoE0_
zbTCdlWKHZ$r68$HgZ19%@nXGH(GYz5g9UyEN5>kw<;Tj|avd!#Uo5AJqKO@cQuwE*
zr<q*#K33W;%6WP5U0q#4Yn?;%eDCn|^z@K0NF;Q1b>r<__GbesD%jtMhZ~J$$wKQ~
z!MzP1w6#f)F-Vp`BVb!k&vxYF9p8&wo|Dj!khYGFfHdL2Qu8sY*0wg#VQFMy;;`O}
zr&grh)BNSh_C&sV%|W~Ze_nm_?%rPiU^4HcI5u-^Nl8i729*CTXE-o{{q8JV6-hX~
z!==CuyOuQODJQLZ4BF@rrW2+@1Bwa;WHgRiJb$Y>TC7WpARdeqq-zdL+$}m75AlUQ
zfBx~}C{Tjki!pCbys41<4R7E#et`$y712dh&)L@HwTIQt<%;hv2kkkGvFvVC$FrKm
zBUTe;z0Kr9uT=95(KG0)ScAYHpIBFkzFvdQyb5Cu*V0Op9#7kJH9UK(8K$t1dT{#t
z;Qbr3%c3Cll<l-#M2`P3Jk0=&OF_$|b5Y4eJ!zs1J(3}lIjQ8FVd>IHk=X9>E@4DW
z*6qU<gDOB$y`qYiGM_9UMDX$P5fKwplY|=g#W0(XW+>?mfhKwPla2O{=*ExR3G9|H
znLZ(alV)sV2~L93?j$<s;@O5bk7h~}vY9E=*q5u$@PXBzuXCjq5NOmq#>x9OiMRl2
zawYi5+1Yx2aU(fDKPdqLK@6QnE6CH|!@}sx^fU;<{{h$5vK>H}&dJ@R*;<7id_Q1k
zfC;>c-X1=WUi=|pOb0Cu>$kXmtf2UY$7LxWrT7nB_i{Jm6K?p_FBWm$tgiW`?w=pD
zXE~lcz`koM9?+~N*b*x3x_wUSl3MpUG5}{mxG~SpH$GgzT2}+3`8F&Oa6)8r3{}5B
zBPLdU77?1WZ!*_~-bKW;HFY$P#}sA{z97NZHJO|8aM>(lPFGknwzE@Yysh$*`5?DF
zCXpeE?C53>0j7YEv_=6f6Q5vlbS;Df$sZCF52iz)Kfj_pW~tkBcf2{ZP{~sgGtHk6
zd1_@iRUgx?D=5TIU|QLK`&`F8n$evcq?gus-iyWU-#e3qeW8S`&z?UoQ2F50;)mF6
z#G+QLE9>FGLlYhB=l7UFr&fgp{i{;0(e33?#6AQEr){^m2?Jcs$=w9M=PfBIp-lHb
zIX-^&?AdSW*cE3Su(`Wp=(Wg#aLC98>kb;Vjt^c7e|v$8E267Q27YI2tB{hC5|_)~
z%XIOGqC+1w(G*ZwYB%|ie(nq-cHADNEG;cn$W@_aCMPGK{;U5)Hi0eIE?j8h2Y}xd
z9Hh?9&V8hhi_q!E+zmj*EFd5tlPZuL`9@q|->F>9$?RZEF0E4`or;kYQ-_~_%|O4I
zAVDbiw}QrPhpfE3JRR<pAMM+>2sH4wafep4G&D5a1ZzAFPR`|tMEbxP5`WW`$_C+!
zHMujjNFEAW6vF`*ZX^n%wj74<SA;81aF7c2WOOwBNKB285r+g`&zGF+d`+u)0;;x|
z+{kuaUJ;=jNX`V;%=BODNmDmZ55E0z=?6FtsoHOL9gkq!JNhzdAASo>%~@*Lix&)B
zh=`y3G*xo~B5CR@{kP$>{e4YaYB%`f$Cci_*@j{)2Nm|~m3qJ}azK9-?G^AC6pa|@
zo(8)kWWtWf!M_lVvPE9tIQ+=?dxB6sGgsz)zR8xin&%uo<o`JH=S9qIZ^Dy61SYn!
zu9|HGx}?P4wpPL)<Y)-(IlrUM%Bnfomo$Ld*e6v0o%Kz6c6{yO)Afz@wDffU@V8rn
zQ%3wZ(N`WeHa3!=o3lMb%=_Ez`J8Qy+Y$L(rM2g5uGzjnrogA8m5zv80OeB1_&tWx
z#W8DL_JjR}w>c@FJb7|-e$H^^YG-$7xOu8d6I~8cqmkTPceg}Rax%n%)#ui2vQR59
zAOKNcUmsv2L6=0ZNd&yF?Y4$r*VorqgNPtp96`h$Jdnuwq$8xq`|v}Dfd4~8R##U?
z@bmK{d-oUYp|rF#H#b~glRD1}2JpCey^3xud0xRt|Dx;DWqrU2At=<%EJ=0b&evDN
zTW*dlf=6_C>^<kD7RaU{-h?2aw)pmSC23!=WKY$1t|ymCR68PRIkGnT+_4uG>*}Tl
z*Bu@(JDBU%7_KQfgg%DcTz`Py%;lceDAWsG`uGkE1{^hb*g5W_f?^wu6g8;69tS=E
zv3KZ@PSW5_p{;AAc(=!gp%w$w5fLDtdM}66A$aic6kL8n%_gAFkS@zwGwyWU9Y-Cs
z^@_!53R{6fqH26jO-p~anQw2mq+B^pGn;0;g#loqQUY=u){7VKK|X_AdxONZJ5{_H
zt*+ZLG7=e1%8N}zq-j%swAvM0`zbLz93x-7nBnRIq$o1rLQpC7x${84XVCs=F@hVM
z9T&%LiH}faGw)%BF%RD7&lQ_8lL0sQ-9_I(9eEy&ys7Et&lY(CM%{u@l7PJQGkP&W
z!KS=^;eUnvat$6<>Sqy}HfV`{=E_GdPQ+Jyje8vxDM)~G_VL-IZ%pL+64j!P5>zej
zWkJKk!tD|hQ=YdOB+p+=#JULc&7H6yU%?q3;cRKImCH2Hh#^Vu8d4_k@zpO!-Y+*c
z*$t8@5*I<-XmbR~=<4R6#Oyec1QT@y&|@pgW5Z)3EuaC5<7{^Xnr;)uCngWBT|VQ!
z-B+i`$b3+$(wLbkGEG0}8-ynMKLNx@TV~9_n4B`mL^YFtMA0zF_#|9hI37KEba`{*
zcztRfPR6gP^n=g)nvIN%Y;UnCb)msCmdFxXU7G4+KJrJjX9!Sllw&0C?G>+D_$s2P
z<xP5jy~7;Px>?#pmd_!;!;7wuNlFSGxl8=~5FDG}Oa7R7UX3?-y~^_P5$MDmhojPr
z%qIOWeJ+=gJHtt(0b<V8s$x)6GiH>a<s;N2aP#)^%45O&uP)g932+tnn<xB+mJ~mG
z`S~(!m9*bW^>*h3kLLw|>-8Q+6{IKBFMv@6QSYaTpFS1+fCWYM^DM5dny9U^w$7iZ
ztkkTqs8GwVD6Ft4OougG6p(*k!pFsXYJ^FsPthc(kS9opBJ!aj)sc_-CvOU;Ep}_v
zjX+g$HZqtYf@-TRm0@c1Gf&}@fzGXF*>&g8kfKt`MHk;qO=F4k5!FwLVa}4*uLVL2
z_N>BsWeB%=wwF{RFmzW=>a##Q$yJUj+UV$*BL7X>^ziWCuj$yJt)=NDG4WV|1}n(%
zoq*6(%vE`cj9g~7EOdW+B9Fl1yd&b|#D<UWZY!tN=ymmY^M_QXN}fv4($Xi`Jm~!L
zW`sea<KuG`uSP=&EKUX8j~5Bf!I`&+NMo?>Kp8I96SSBr8mM<a`B-H`+Su4wbD^Q3
z!3-jeSm?_yg@x8fPY8lVa8a<Irj7Ti(Hzwgi%6{D82tSCmx&|Gi;#Fn)m~Iyy(*&(
zXW><}RL4MIL^Te6XVm4>hYKD5R<`yPqGs7{LnedXy}N`qO1oLn<)R{jODL<O83R3Q
z95nCF1>Vqx^i>@kzvT(8Ud#R%AEz)fh>tnd`v>OsT8KRL?{>SSUn^NuOyrj+YeLQO
zT-mu==O0oN^5Z}^2XrJbh(3*$T1B7K{S}Fg`8lAZq|-Z^2Jg`Y%#o4LFX!vi{~W-b
z;6c7+=E(h~GQf^H1lttcJH%2}R#Nu&w<aPrO?1VS8HsldgGX)cTP2)e(kSwZB}XHh
zq%tC^H}sE$(E<&Ij14}$((MDFzSUEfJW`>?!5EZE0KPv_&XRe}%bQ5ARmtRWy5)6~
z!)l|etG%sB!U+42xj#|&yMZg!(bd%y;GTkyC4R%h3N{P%Js`e(tgyfXKdcYVM1e-x
zw-$e7OK05={F?HIa;XCR7m{p_8~sz0;xE3ss+&1IcjUA-nVWHJr=o0fXe#+4ki0v!
zId|`E&#1=`f%Vl@)IbU~r1HREeq@y|G}vR^M~9SDbG)c$%G-0Sz+{2FTq-e2{YhMv
zF)?*9($DCwKn7|pEu<>z4g&sKJX&{e&OK&98)$!uFXPW&T^Mcnot>v<qu*_BE;@2$
z;>m|jiWF(0Z>!B&LtOT#mbb^$VS0b9scU(7*osx7ob|ke!fN6*S7T@mg`o{M6kX7+
z_Rh@iBT5lyFX52bopeh&4o-ZPJ1yuCrD74dkG^qKgHEo``FKm~=s@dpRA|m`J%a+~
zTpgA>Wq-L_4Qpj4y=fCP>*`A8{lm!UWylvKBE(H>>DTr~LOQg)xxWbf?v0f6diAXs
zUl$lQxsASV<rXEc(uz=zQiej&4?Kwnmg!Ic!2f696+c>n);L~Kxk30nL^^`6VvUt(
zqMsrnKB_Ox{TNRmz?&-IoscD)i2rVqK)r6E#!-A%z1Na<5KbM{lcSLK7>`Efw{Rc^
zoo4xWJ)f%p$tWsjqwW_89M+mkhP3r$V8QM$S0ifr8Yg5Y@iQY{Jv$%+XTi^!H?Go!
zKe%7U2peUfmshE+6O~r@3HBHsG&pSaG`-7;$31u;Tkga?qNtxxK&3#}19_hZ&X|sW
zC#J15xZbZOM>fsnY52)RJcc}AZR_UiB401oA}z^LgV5BsGmWeDOX!4srmVIDOWOvc
z(PBqZ7|Q8$`nT$Z=v0$~-KWnbrN6(rzAsp6qWwQFfGC;3@n&EXxJnJ?T@ZK3olNXz
z$tGW^)H6P@`|+<e@{AOv5-B&v#H>C{0i9nM4fkD1dw|YrPx4p+pIcm01leX?gn*Gl
zQ{#d6{6HcmWW;M_s-6GFKyqk0xKwoxggvbIA_4OtEIt$gd9P{q@N1oxw8i%(=!EU+
z-kx|oF1tJxDAlg1;_4Slg|Z`@!_UyC6IpfVXDur9oVT9g#4v1{YZ6IN{IWOyu8N5H
zT&DL&e2R#{nog@Nso_OH$X=se<ZElKleuh&;cs;}rE&_s@Syc{?h`K8J0yf_Dr38x
zNu)Ql)J81axhTuvA-a->6TK>|C_`Lf`gfc-=GqO1A{P1JAuu9dvDrb2WW{i~JH=}@
zbL19I#tR1$-gJQ{l17YZP366Ezt=X)+;IGtS{9H&RII5|KKZzI2>pgmY&Kb-(RhK3
zgtW0xq`bMlxE&rIUQ@+sJ*%?Tp6lP{tA2cYy<6;cIwG#A)01Dwh4v;|DxOLIF`z0Q
zf0`=N355vWSM*b^$^S9~+_6fx8Vw<b?`z`lR1tY9!#`EXIQ2UQ=D*=mbJ>LPbZ3}r
zn;|}Q(8iF@H?Q}S1n}P6sn4O>X1jv{LHoVz2-V||_sK=9OeOX{naM1?Q7=+`t4keG
z>wx8BBOQe#fwgSBPy#*3OX#LYxrTr&yu=xC$0bozw75|sv-qA!E*Q1b_ks<kH!BfA
zSaxZ?#zb|xS2W3I9{k+njw!LN(PwKXQDN7=ICpl3)#*%D*E}(jXy~jBt=j&SAgfSk
zsNx6m?cJG(lk>t*1QIgi&8YXROlfHub46jO_{9v@#QqnN>z!C}6s)SlMfPZ9pSdfS
z38KfsLLcD+8IQXTW-_qQD)*Nrk2<QWE??$`Z=j5if3*XT^I2Ow5Mu$c_z)l*nOjT5
zCKxmf6i<a6UE@2Kq~A7*C*uYNWvI8HG)1uF7O-~J=S#1VP2iuSxATV*f8x3ce#Kan
zeZ#_laZ;*!w8b3gP{FgVF~uc7_o9h(5sRogVp7k$?-(VO^{;auc#x3{EZ=>>?yz*U
zT&S5@%4Q=|&&Y-QRuw5|-kWt>OJ1_p8^U=~lk_5WV0@ng=Z*A$`VvhUYmL2?q8v1g
z*aE8eP~|PCmYEI>Hta7)!hI0>Vi`?azoNKLHV*Xm8nl0hH||-_Rqjh)?C$Q4Wy^!P
zE;FqFqi-)Z2?FRg!ledSEsqP>04Vd104y(Vx;N;LV<w7-f>?LB1Cr$D&!7Lc9?BL$
z)OIA16#}i9{E^G`@XyVBS7%XK*$-~E-gOS>U?a7+w8uq3&J5JbzPVGmJ_=p0l{r0M
zT3KpTsi<80>WctEhF9*l$iwqlw-wyQB5nV0=ns6usu8F5#fR_DVA0VgdY-X+ia=uN
z^<H7xu+|&?M1E{^(Qf=oT24BB_ky*WfK09?A$f~~9wGJ`vK|lI9rVgUCT}>0y-rPx
zYOf#8ot9iGz0*T5+!77p-rEg~*TJimIm=Y*IyjbrJ?=)Q&zbYaC#8mmltkmq4JH|m
zmPU8*Ojb=ae1GO4lPap_!hy%<Id<mhOXGGv$k}3Qb08N|GH1qOg$_C_eC1M1u4~T_
z2bb5?0xDCSCBso(A1-MR_^ad|&hULH2xSw1jiL~j`AG}YK}$mPTw!8E-9KG$!iSgA
z1^@khr!#xYBW9f}osP~L0UuZYEnZ8Ec*cUy@66>_aJKlIks|HwRV=u^GbrD^)vJ?k
zOq|r<n#?#e^)zStgd!FF6$=gD{@CLE7sv7fose%6c|TwHL?w3lOmK-sOZ8;Rrl$7~
z8VL=-u3|m{*4;$AiowYAUG?b@0k?BJyVG6w<DV%{N9xsNE7Yjb9tw-O{<*ottCLN~
z^F0j!=q2RkhwkR&GnFEflKO#=usxoO12)vq{uJO>A9oe1=Hn-FSd&vyB6jC$c<#@#
z6S0|nPECa%6lzwq0@5wI5lAt84fEyZ%AFCAp=4fHa79gx;LzY8)$7+^z;8MklNT)c
z)Z(Y!8P94e2iq?(>^dG1huDKE8_E|il}Y~YUC_zq5UbbaA@t)!W*Ao#*nT5_5J4ql
zxY+33-rl|@a1So)Xl{P5S!I(v*Qi#gN!0PyOgxC%ZB;J^QN6}tyZ8nW^yj2?D?=b3
zGL4O`DlEdrE0k@|j71$;!@Z{ckF%((1U*ymkTIJ!Bm$=P6;;^M@fpO(BS#JPIB48X
zRq|7z!%1qXu+AbW)}jEur9`}ZUB`@_#LFpE8v!qonY=>tt8F<abeV}A^%7l3Ld3#z
zY?L+vx)pmFt1tUg5W${KoNT6YyWe*iQE{`idF&C!HJatAZjZeUX?+$OIIF(4-((D^
zFaC^=7QLF)F%!6kIGv8uo+#!ne@ej6Lu=j(OROP2uWmjUj35dBqKyfkAA43)uW^T<
z3IfmFmmhtJ?<|cw)_lCttjO#R1QzM^0%XJ^IP^%`7Ydi6^M?o%X2xKB=^wc!msH-r
z60}~(@(C@T$q5K?C&BSAQx{BluH!GG-ayL%O^C>BX%_5JquIV#t>@cS6r_%n_{e2K
z;Q+dvZEu2yo4t3CbCj0r_AP~;Tquq#gda7a>N>+!7Z-jpDt`-9cUh^j3>K4I`1ik6
zty3pw4JMCv%b8!I50Oa<2Lgdm$tH*ZmiC=iB^?qH5}5x1+^0&pxXQ}PfIygT^7XYz
z1ysFag~fzd7m1(`A0Z*(>A@lds43tZ!PJlO0u3BM{=Ukd+}oMRKiNB2TWbf{!~x!9
zHbM@CLWifO#K|Mq!cFz{zr@B8EP9=wFE1|_hpw)y94iSquXcv16l&ttXs393nV6Wc
z*+An|^V#6U;#&5PFR!jfY8*|itgIX$zQ9IcF_G7iDIJTz#LWDTt%RYTtV8M$2^kp&
z7x#3uiNJNc&T(^)_CgTux;aP+n6!&gm<PORPY2Y?sY;nEiFn*iYy?%-a>mk*Vb@y3
zM2D`i=Loll7f4EamsCn_H~p8MPqqmORctx2x`-fV<LWeHnH_QT-RLqJK1(k?M4a4z
zI;NWU!0p`;7XLw<FlIHL<)yyAk*!@gsA9>Q!WQ5czstEXCl?vtb72Wb`aSdg>}A4?
zC0j&L7;6|(F~ZWEswGJ5>^pFdIk#Jih(CXrnJO~351USOuo6;(bs;OWP85K@dE~f(
zH(d!(h?;zFuDHuiF=kDzDbkDvVw0(H2Bm||fZgqdjJRG=R}rBy8htX8`vlQ18ay)o
zO#|f->Gai+3D&IjX={m0p8mz!<gIp*u<Is_hweEzB_y;k$&Aei(JXG*$(rS{X|Yc`
z1y!lt+~DImk7~7zp&>!h%>Io_Tb&vgGqT>j{>L|OQ#u_iGz5zvDG^}US)Hku-WF2O
zMyJ?b0yCblxA)@<cz1lXf479U^OJglcV#Tkpzt@;_j@Qbczx7fa`dtF#WoK+?d!!1
z80Fpulp<8C2i6K$7=9hxQBhHWB~~obzEAYPB;$)S><R}wmg3sSM^rL#&7l6)om1WK
z2rC2iHJwf^Zkv$6;;(ngNWQ*sp*fUd@e=HM)qJ&%iHWmJaGUr2TPH_H&-MFFIM{S%
zqZuebn;Ni&_fV(eQj>rzEChNvH+C9s#%T<NZL9tiu3Ru!ipTa2inMo>-Ly7ohAX%!
zV=F67+uc$)JYltPiPO$tK9%iNDBO=VqKz)#vXkh;vaq@O;{R5a#-cp&D**73oo>1&
z`T$^2WM{6=n1J^yl>Wy)XMwDaW*~%39>k6k`}2*BHt{5sdbJGZ1+g-Y`zP-Q!WZ)z
zYOmoF+bpE<;biw0uoJe1rQ=r(2^=OTzk0Z7-7jQJyKp&eNUj1hs6XVWaHppI>ZMx|
zDqd;wB${iqr`oy&Q)TJ6IBYqbY1L`$ZauNE00!`jG<Jt-d1+^?GFLcvRBS&>$;vFb
zpY9<KOOus8Un}U$oi`@pIsS9kqEOXkMWI9}eL?#66fLyHTC2=s^r@4vl2?`Yz+kRw
zjk^~!D0JYw;xqB>e1{#jT~$_zdotbY16;Gyb{6;Ac|5u5%2ip8qH?!<$ey&)Fdglo
zN!z3vP-B*07XMb+_|>J5`-?(+akX<qbT4ivd~xt#7z{GF<BZx`EOn3k&*S1Y2Z*9y
zAq0%abGA(*AkgfL=lTO?90Z>S%aCfmSd-7)<;_7;P(+03U=nvX>daRy^2o!h&D7O{
zgQ4jX!;-2hajGZ~O<HcRPTX%VOEfc)09_+$WJCd^rccAaKLF+fC0^!RTH0$lIXS1j
z=}16`F0Zb(L2IsJx33S|FoVLvj3@F`owi32EiEmLSO_O4C(CW-6A;nxUtnYVMJkSP
z{9=$xVu=qONf*Z<Chil2-wFdorR4A5_cYNfdwVkQ`@5M2PfoCLfp~)ZP9W`4szZ(j
zFe(q&UMd>(K3x<9prMn#mK*hazw(!{@#n=GohiR>_U<mQ?=;z_l$=@<ou)5PHD+e2
z$9F=Z3T4qyMu%KM1dvvy9gnaD!$`6^p0QV5%$&?+wtOa!c#wAu!!W*TFZ3%m(Vm(?
zx1u?Q^u(iyp3@!kNl;X?y_7h;liq7^3&O8hyqnd~kfLl^#*&~=N@}<va3>KyDV?sX
zBHx>RbU$4YTP(Pz^XVo<mS|bGkt8anaP4m~L_n*t-Lx+jL=a#jNRtTaYcSFyk7UY`
zr8Pk(mC>N>6EXx?$nJbcD<ce=;+QM=c`iGq^kPM{^MbCXZ;z$4lna!0D&lg8Lw=Ac
zzqpU;?UhPz$PZFlHqS|kRGb<A-Zf9s=PI@YN%pCOdI=$x!+Q`^N1N{9p!fX^Z@Fpb
z(xYF@Q+o}4V-+m1@XP4j{Hc=jZsy15LsXyORzD<$SxYHc*_k<K(1vSmNPWoIlZfzW
z&8}bkbCYTHb$36rwY9BpXt+8Zlds?bdSr!0sS%2F4Bflqmku>am;#rdF$ISAP}||A
z1_rIeX~Kj&PC2vQNl8iL3n5w6_B)^&ud>y<KBuqNj#FR>j$Uf;INPC>&j6Wiqi6;F
z*|Y8oEgaww0y|4hO)W0&SCLKv;s-a!qNm7Dh70NqOFGSAzsBGo-gJaZz9m%mKX~-%
zph&l2_%?2u%g}0FL}6N32@e&JV&zyVoj;Y7QG&j=m&;Iw(oJZM#6mnTVjh-vV(G@6
z%Iz*y(8Xd>bHFvl^lmYBO0l`&=54f8LGRSNW7OTvy?HHcmwg1&BE3yZRCy6KXPTk=
z8P;mAO$J`z>D$8dCF^R(hO<SM#^7_!%#JpuAcH`Y6`X#35TRais{Pl#C_=`jd1C3O
zH^WyEv7f&Zg00u7wsr^!Ea^O}Vni1B{e2G`gmCH+Drjr$j9_#O3$+7-js#pj?4SQ`
zz5J<!4chNB{v?w^+}ua=%?-8p*l9zK$pL>IBY&Axe0^>*7w<@Nd`3w@{uAV)0+~>|
zza?4D$+J$j;|R7F>aN%L4^}UbA`B?qPG~;TNA)=1)U#Vo(hUM*P(3pFih~6PPXg`%
zrmG`Af2fml(D$%0v5!J1&j8O8N*5fvhYU)&HS;}T@|Bg~Xt#Ian_Y%E0juu!aU$*O
zceml+LxgE2J}jy<prfKCQnM5-Y5qP4*2085n&lUlXY<+vgk`2Kp1Kc-K*lul=kpX-
z?A+&|+3e?^gV-$sN-RwuJ%oXjMc{8nQk{3iX$7a$=q>Kcmqf2V)FmdWy1eZJbzU5_
zI6|%KVQeV)TJY9n963B(7PiKG_pFsyM$4Bh9|3RYK#1X&z!0rdj%SeKmlVAP-B8mw
zrA(U1lar?4Hywgx6KJu!o7}OLx{|YQdyC|w@D!&GNWl-s(<}+yHfFr8lPN_-NbKfa
z`N_R5&tPQjFB;VnDFSmAdHcctrYra<9G9moRWPPSD}fi642IiQ;~#$>1{5kW5w3gE
zS0}{Wc#Z1WSqo;KW!ciI{#?ccB8b{viBd|~k(Z1!MP^p%en0oQZv9u^VSe(})kuU&
zeEji3?Vv&Q`t6BwW-6o|e)-g)WNh`@RZ03aH}D6(%gIpCZ=-xO5ry;L-rARy(Jja|
z+lJ7rJGzL<1!2;kf2@hnVRltyV7M{eWS|syj@CsaN+zD!PsPss7z`N;Q#<ISpek8E
zB|O(TUY@-}7NMm0{U%BuOgF0n>=5T+mS_66t-gtT%-h?VoPmpg(d*mBt1T=rzGQvL
zSPv#Hf*qR_8uaDSx|NlgMmGLDF=vK(*iMIByFifqr;}XPy{5IcFl5XiDT((*`5_IN
zh-dTj+oQvYMd-w@rrzOcIX?sp@MZBL75enUTg8;3XzrcOq1?c<@sf&n$6VBq4GJ^V
zIep!?4JoO0-{+`vpw(N6>6D^~QP>yHsvN>!ZsDbfDCMc#mUuFjaoh@9DJj8*&O!?j
zo`nRxrlb1C57#w1?q!!D9gZkT=Ebiq9ULEi5nFl=MQiFBn+|-b#f^6rM{bjf@LMjk
z-Zk?qM?1*!HRktf=f?m-<9VlObv??tNwM<K?CT)RRgu#E_-eanWGfN(j(_Wl+X74$
zlT@|N(W=4rh46l8-^byae0wbhQM|aP<Yi6fQJtWzijIu#7y`V|Nz6W<taRU1nuuRx
z33~6I3>mmHmx(5rK6^;1C=|Wy+w(uZeH#H(w8{H>Osp*AJnez~r$$a^A&s`e?S)B>
zDx9fPijn}J^sPnb&o~`B!7RLUWeC7YQ5wUp25Z6z;a|C!_XLa@QBx?PM#iEu;EOh_
zX8Zq;)=8+ob?}Mv3}<-;NGBi{a3u4IihA;HU_hjXsEQ|V{}%$JAdz79ZkmYm@wvfZ
z7QI!qed5sW#Jl75oZaHZUQ&5C-iot=kfLymh$7gBNbJCr$m@JtJ*R_K9t56=3x;xQ
zB?}sVpM;^t5P?^bOM4TTu}Du!ibqM}`Ajoy)WBdwS2-*%efrn2Rfks)Y3^m;Xm&ks
zjV+6X!LtU9azla^cu~4bcCIK-%4s(9S0tFYyn5`U^X}SAma@0)Bnh|JVL*kb{9&HT
zQ+VRZw6l%ErQrp`aHN#I^V`1tVJ1)p$yac!hOo&=dA(fbi~#1@mt!^QG3nzn+1Cr9
z;KG*I0&s<@IT^r(h!Rm>w$7Ta;@zhZMgMqz5HIL(fBpYxy2`LBxMqu@qLLDVq;z+~
z0hE-M?(XhR0YzF;8tLxtF6r*>?ykF!?|1+C2nuJ{%&b{!jU-6Q00OzZzd$f27V>L%
z-Mu#=E=KUWP;Z{6K4RDm4hv%SEE~4?WG)*p&nE_)%mJo9J9P#<=^$JLn?=3@zk5cG
zcAC$S9GI=(YN-&&(zSPdG@y2VufryD@&=Gb*A{P?6L;;q8_X4e&OS#zqr)GHd;yf=
zpSRSbE|Hh#Bc<<Lm7-g6H_-aTGQv04f9|nguVEz!DtF>DPuwGp-he!Q$bAFFpOcvT
zUe~$=H-U+PCR2m`1U%1^+}R|d8|qgrx=`m(DTgG5B1Tw9?hI10FN%sW!XBV*1eG4h
zmcFJ7BZ45!$j(0HXzcSbieJ-G;n=z`cxmC|W37w-;{?kR;k~;u2ize~%gAgzxU>yF
zD893|0<!pL3)RxoWgU<Zbmb2{59$6Q8NA#z@_l=En;}oHH6N3?tZe)C0U78P@z8KE
zRfA`IqU9EpW_NlJZ3;%7bz|e@ELx&w$6V?F5ZOM3Ez+0o(cALbZ^p3Q>u1dev+W`6
zXEtVR*dAyD?4RMMxg%+W-V$>;v`l|I($|1Has`p}^mtRI%?dx=T;T$ZZ;0o3PPiXd
z%T&JLQzZ5T>)}&XJg+ef!C<m@%Oe3m4=HQ~2Kc<Lk8be|9A91J)#w&zRruvq$ybzT
z`YQ1tU8`?S1}x<+TBddBkqFKcfec>)0rA6hox(SZ$gxr5vkRwr=VO%oQwJy-Az`7R
zWs(iJm~spF08NEn{6x*kFa<C+lZ7?}F|wJ;n)E1JgS-4@{6o=U+&2Iz*aH$JFBw2N
z(_|@cYQ(eIWv3MJlze<l&MweLnbGAfAqYoJ@FT1Y9<||uppxJO&;zI;zF6XG_#IUh
ze`wNv=dCKbc%3b6!fC>E)z)5WQzTt3g999*uf@1)Q#k=FC)Zlet}}&XZP@t<z$m|?
z&h2ADo+QS%|BH(hogg|bG`KLw#l>~vXDbzJ(r7j)kDp`sIVB1b=^P#tR@~fj^YXdI
zb2%DVSIc^B03DzFt3w`ymh)EF@j4PMipC~_;gk?^DVw!z{(bcXnI=cf$~!|NN{-qL
z%(%ep;JwM0MsD@b=ez*W2nycrGIiQnX-Mj1pw8agWcy^>IA1>YEr7U0Di!HGZr>}R
zgLW0es;}dH<<m#<-<PQie(AOO8{;R(f02MWLvZ-FXG<REI!y=bTwMLwT))2AZP^)a
ze5#W;=^SqReBc7WIvdXg3Ybj<0$Ed3Z7crbtdQKA(U@O7<cXS#DU)lLMB#bOs*Kld
zTt0)v0Kh8oJk6NG%Dwf(Ws3zUy86rwo&gnWMzFX5_!wEFHnyVD@_)^nZc5k{vtIrd
z#46Q($BqvHpwY|1?gItt8`Go7h(S5`8NTXw07uQV+z^x4UtnIkM9ixjtJLvu1Vejf
z6nss14aEm{6<HT}FUXPIt98wET0u)Mcj3q(P+ThO?p!6qh$3gqL3r@(6ThaDY%bEo
zoMcD;V#r#vliq%Dvs}9F_vd}Apau7;|0d3DWhE+qCj32tl}MJHv!nsIIops<{|7tl
z(%WMhh&nZMmK>ZUGN1%Nb024c@hAI&Ea4FBksKLf2oI0eZLK<8wp0?=c&;4W!;{*=
z>dMMzkd2Rcte`dm;ClstMF(@$m3V0%UuQ}L%~pa;3c2^ghl(d7a^972(+l`O@jnAu
zcz)!3nOZ@c2$pikV$7-Mb|qE17ab>mWNmhfo93@(G8FlQ#BfEt4!(>x!b4M*nf6zh
z(~n)g{`aLu1g<~cSKwERq}!snb}3e1SHsnQ`vCesn~uh<348-Mtcmf17y%7-@p?hr
zk>T-wCLvdJMO)`(k1z#EEUCkYYrN)*!Jx?$8H3gP&1K-r>*m}^2qUsPCr9&3@CZt5
z&m>)5K;VzgLpj;3kZkcZ+KTYjJ0g<xv&i!&jt=8C64jV-y;THWt;mWsMHOk5K8NoL
zCgG;`O%LKazV6bgyv^+Jx3TsV2VKjNu5=`=%Uv1mJJ6e*_Pm62#cY@`8s-aAn}yKy
zjkwO^103E{xjCTc`uh1C{ev#9R8<w^cy4fKgCIOQT=klS3WGhAZgusC^vJ}BPG3~-
zO_7h+*rw%d+gdCZz|0k?Ywyx*S-Dqc_Q*k<7F%NP^6hv0A*dC>fsCD1%$n&5F(q|M
zVJ@{v-5}=<|B~Te?Mizhoszfk&$SioaXGlhJT9js-C+dXv$NPeqSuQJE|KTw_C@No
zZPC<9UnC_r0Q$yawTKG_#@RhSTxp@deaUV)-y?}+WN0`Fh$Bo)Oci*5i<;RV+zR_)
zvWZq&R@S7_h8ry`WkpnM(W`}=0=>U!S-{uf$(o%m87yz-M*7h&O-Y<`?fx%ODsQFe
z_l9~iJ-+;`i8h+sr+Mi#RHQvsfh{=CC8dykx*yB?QlQZ%5{u7rHdrh(%`p3jp^T>3
zCYWpzBX?v42cTtkz+<PR5Rmrk4c(=qZ9^eVHnAFGrYQE77&9t?kU4U{rS<icMvijb
zk9qqDo)7v63)mQ%u7Uu6=r&3juKmUi5Rz%`eEi4yCXsN|-$r-m>lAcWNgdy6LqtQo
z`b7#=8H_rH+mIje*7enXc$8=J<v(joSQs<2Ia+m%nsLzzN5^!H21m{gJR$ObX+Y<F
zV2a?<vt0&rKawu#y3WdqX%v;SE(dmNDs<VT-xo4xNGV%mvySN~T`aUOz_(jpUwSfE
z3}0B!WhXF*$pG-bbxX0FH+{C1HfFpr<LCD8EtV9t)o_@ZF9#EU_*Sxl?4!g@*aH9B
zQYg1SohDXHsNc5_S92OEWngV2T2zJE<atP*(ga7>6d@gF9N*=>KvsU3)6#cra31n!
zZ|(Z`=b>#?qaux=>V^Ib&aE8HqnqLwudf0he!r48W%29CrUZF#O4q}GRC+&lfwWQE
z?>qbZ=DSnHVa3Jth}bO4`!nS|r#G;lbLB_#6sW<|AT$uqdUAFqAR^-5-!EZ0p3@2%
z6EFbYfdk~!DZ^NyEZN)~rDA+7j|&Ws+jGht>p+^jp2rBJ)bX#WM6|4kwaLlTf*F2*
z4Gs-EM~a1{3aOu-J9AIVfd)H{0PHGgu!n|hnQ!ZrMTyz#Gm%{$wt{mPE*kXYp&)Gd
zN+pi7_c;$wqx(|5AG;nv@r3ErCFI=)1C`+YcmFlqO+^%i-+~Pf0MkMwga-4<c!2=p
zJjf;%8~Z(nrK|9(s>ZwYuVUNP40T^FAkUb_B4@6@vy!>i37XG%H+ZWyzyP{Suxd^0
zfzc4`pM0a>MmK5+^Td&GoQN27(eXwvJEg^0RppY?z9CjknSA-&*qbZiP0K|wL&f*t
zfvzf&WT-9<Ob#dAk=Fa^<sOo=44)$}cD`Yog?#&y-tEHqsF~!AC+_F5H<3TZdRzv3
z5&KJ=tiFT3E=V;qNCl5ct|6k*htEGo_FlWJGw|*$R+6;S4d1RIFshUOKNo=f+Pznc
zgje!e%imsDGjil}$-ZA8`&^Bz7hL2Ob&PoqJb81F4sHzgbDp#F`X+`<;PT@B{FaO7
zvFUFj&$J>Wtw&IHvG<l^ZqCl=qKrSq7xc`q&d1zC@3dZi8X6|r!d2lUoR}b{B~0w8
zHxdnPg8KxDfX7EyqKgIA;-lo>fCa^b{uTiN@z>XislFd;<R3u38;ygQc`~dZ6uNPl
zn|WXFNt0r~(@~rf?6UC)C!PG&WY}7wPd;sHzg|d;1eU~aNqg&RB6njocqyV&DD+Ar
zGA21xEcZoA$*(2uwt9M2T0{+vh39*x@K$^(b3W{{l9H07tgJ@g0s{ixxPGVMaE|o6
z-63~(cb{(Za1-?>dfjH0cCqwnuG*@5Y6{c4`SCWt<wbxzmBG$vW=(DFJ1ndo0138x
zKZ6~fm@w##AgwTswyV7W;I){fj}nR>wE4kMq=xCrZr0W?pf=-ri;GDhVtr!MA$~v;
z9I`W8!n<_7C!SfBxfWsk`Hsug&Bp2Ct>;{&ze9H@8oFcBbvJ#n{Mhe4lckn~;=4;l
z?r&>u>_0(r^><=ld3~MEyQt~Sr9%AKb=SM17cVinn&ZOXC1gu%wuL10&y|JfDHWY=
z?;a}6Do}3q{9!A^nkp1er4BzMivAMIwoUx1(qcgR$y{ga23~NUSG40GlD@LMk|A3u
zvdev<njL8}zPQO9Q^B@`-2Ke;{lRo6_-0Zr#&FP3Y8aK0E)wTPMpl%>4Xd>oNfEGW
z^H!wY**@s>DV&?JUu8iixP5mwe3Z%71%^ME&qrDD?hAwjVvJ*7*)OxoQ(=U%a=+)S
zh1WLU@cXWZc<8UOrX!DzVQv9$Ze^x3n99k_x$5*?tLk7Kd^`TNg<GZuEvC_rYOB7V
zNC)r_ryG*oHmoq-pox!RmUvppb#`=HTHAZ5w0YvCldS9RqGzdvfwmX;V!mElLKk(?
zFuwT3{|Ze_`a-YOg%TZsLksrmwx{VL`JLtvLN@d17O;;n$ET-%A8u;xwkoiI&?Pe|
z>HAAi(RSXC5Z5~L`N2u2{S=kxXzPfzWn&x9T_;jhYbFmyJdXpJOXS@J;kFIO=P&}a
z513&fx)>DFoLCnY*r2koaACv9)l+qb4viPR4YR@}q~UrsKe0DfqN?CHaGebH6YL%u
zK8{Ep|2N?EQxqDidN-PumP9MwjMaU##tUs-_z8=k{f-Kb&ZWdK@B2E|dDA#6A&I7S
zzTE{9ym#*~yEh|^UcZGWR!IE=06E&-@hnjg%<<BM!9fH4Wd>${gw@IPuTa<|`j>ZZ
zDj}Ex4YH-&VT}4GdKRu%FwC?Kq5bChC<Kln-6N&n$(#wd`dS>5@hi^GKQfYjr~cdu
zx+Rg}(cb}*a=6ha*Ay&Sy5_2p+zf|tRmxmp{Yhl#XG!cd<Otc#F;v1L$R(5nRpKD;
z7cLq!{eTwt&@7SSJ2ZoE*d6!s%`eqm?ZjxJzhq^)wG!$FgwP7qgpaSSFmna@!Csrz
za{RWZ;HUgLlS&hf_xHV9qMy<4ygtN%AfA}y)R&*a*$BYssccaI`Glr2a4~%V_|1uZ
z;r;BE>(sQGDpQe~5qe;vYcE$FSC7L6DXzkZC<$sFS9Eq><fcRRlUCY$EfL%FwB}H#
z8;nf7VyT{Ur0tR~Ld6;WISZ7Pxv$i8Eg>Q5GAZnK<nU&w>pw$d6fsOt&pK!+KZ}Tp
z@{5a8`F0?udHp9RVQBZAoLs5wC-5PMkrIrWe~(6{2k|Jn+M>f4^arJUdgnJqOGWst
zy-BHt06)CkXQ~GU&df}8JWmv4E!Y#<o=t&r)Wzu;TgB-<e9Vu{=9v*e>{e$-g=&Hw
zUEy&_A>d?8P|N>ONSejM*~a9qnj*a7**0PM@g7Nt;7cYEE$3T!y){U&Y3b-Hr<mKN
zrjcCFmg_%M@<pn?F;DAUnSpqf<-*&RB8_@1h_m&#gR6r(E0IN@IoMp35*P1lJUA@c
zbSQ?RVqaeyxpBj+;%{hF%c7MIb>$c<T^nrpe3f-Sdd284z!j=dE9U98r)W#jSN(8%
zDwF@-)Iv4xj_WP$CUD0lqNB-HEQrM3_y7a?P-lCC8uwLp$c&<*E>!@!*l2%0aP6t>
zl?bSgn!&1<_oYD|_81^_34%!RuUkZ^e2_1p{(S{;9NMJZ<dmf>4?u<&B#V|ULjT?L
zzfvQ`gr}9;us?SG#ZgZ<@yYplhH7Pl*~9{%astW3`s34D7~f#y!*Z+d{2a(I#vEh8
zhiY@niwR;j=pP7p%PS4G<(5oH{4s;f4-L;YpiD7FzF7HlKI2;-Hhlx5vx(Dan#Ec;
z7yxnoo!AX_cqx4@Y2NcioQRU@=dicO3x7?6>HF_A3r^mJa?P~%YmJ7iW<*mXg!{LI
zWz3_$dJzNB_0*MNr1kK>W=FtRz5ib9*9=g39;1FT2;2Jr`5%pH;<$O0ozs8c1W5d@
z=y;v{B8eDuWhU`CyZFeZ%daCD5uW~Ngs-~c&APZAZ!JDRPM1VCgara2gXz@V&$gEp
z=!fpqrV4%Ki3nM=2>_;-!8EW-!-wf`^c<jsEQKm~XufGlM*r>q%EhRwPc>k^S6D42
z1M*^cT%0K2ZG7Oc{s{XD;g9=GV&4<LfWS<R4KX0R4<1Z`><micxO?JpyIfy9Tx_hb
zYKNZxMLFPC4-%l`yTis8__0AsOIm>}Qp?{s|HzJHtH_nA#C_e>*Ebjeq%WkqTVNdd
zc>%xN#)64xh(v3_k{MK5*dXCNsXYk0syRu#c0no%z4?vIz&4z12WwT8!ZS3028z&!
z45H2tC06fYJQCPxD-J#rXaVjSj*ubQhM7~Q=8+2^ZZwS!Hg06=XOLMREK93%-!rFU
zK6Lq#OUH8Re+}0iuTpd{1(Nr_>E4YU{x5Y`l;zeKL1qTnS?^#P6W9jJLxAOm1!UKD
zdMd2mNNIV}-8_61ahe^+yJ?`)2P(r~8TiXv)|??Q99v+~9f9=OSsKF=G5H6;uYhfY
zJEh9q)soj(kp?%AmD<?ZrAsET(UUqQu$Z&~+Ix7DwWxLym&0_K0lZq3dF$Ai+?)<n
z6@63+T5ci03$t44UJ9gs04pNdrg4&@?d)ms2K^FO#CLD`PlB~hG5tLr7|Qecw{m}j
zgubc;BhgVIe#GYVrTQ-NM<%*Z%Q3@E%U3ZeCceD<RN*KkR%8L;)vTc>E73`yA`DH+
zBqea#k=LFBw9J#0Fuv5D{*Lk~h^j{!^U(UCovnZ!#B~rr9>hI}199bG>|!YH7)bqF
zKS>W(Kiv6<b>37G_YO)-nZrZ}_NkFYB;!ghZA!v7_-KRX9E|!5=lAtkln}i}CmS9R
z&a!-xXa(jfM{1h?;psldhTye<2mZA~KwQM(dQJfZ0XJ3o8ue^&$oT6((1~@wM5+al
zfOMy$#q-03yUdP%^XkxlH)fLZArbfh+9H~lJvr}w{H(_sf3dBus5EGTZ-(HKu_>d8
zC8c{lLQxx_B(A{o!U7Oq<!!J1g$dw+0XV~sohFzvf9m~olj+SVNMkVLc6xEu=2g;P
z=&mP?C2DY@crWqc0Y6gBCJR6f7ucdFA*BtOP0HFVc&}(T(Eu){YUsibK`Om@D=$=h
zoNUM7;;~pMyd0&@=xr2&!2+xt&j_VVqUDybdCqbKGiK(^P%K4lJL##v?z5H5fl(Zx
zw1*O=_j;9j?(S<0TT{WnF{4C(o*bABc&_|s{o?b7w3yb8D*TJ_LLDoh*^`ix+T2}Q
zfIfFS;9k`@?7a^N2mrE*F_4CaI9mKmTr93L#L0!-heIne$oEsmyuqnWlQrC$)6DH(
z9gH&ofA#Zx-MGu~uxC4zprQ5W%T#MXC+W?fg!$V1bOd@0CgWOA=gMm?z}i%{IcB&b
zB6@4ir+56RjwMH0R-WO`aANJ-5?hZEF}pXYs;7_!){cE9^{HfI8J%GkuvsarLzA}X
zmYvvN10*tM!J$sL8&>;hF)*)?kDOMZwE!`J<?xAXduO-IV-_8^pwMNxI>h%Z{vE-4
zW$hHe;gq618j4Pm8~)ul^Z4KtRbEPGUz8$(m<{3?G;<%Kk>-cJeD54u#w1%?cLf(p
zxBJO4#wSH2>`$SQMo8E{??`|^*y#hRWXfY5`N-H<JAfd_$;m<Iwxp^`xY#=yJ9FW5
zqt9eKN2V#EyAe_T%14aLPnMLL89pvFRGIcMxlhqBU7qUM@Vg;@r({8*RNrb$RFcH=
zW`5%qe11ABLSoiuxe{<+!^7i0@n0kQ&=qX7)iJ)%QVs$~@`Eqj@mdMtD*WFoZ$O9R
z<|aR(j}{H@)hET3X?=d=f84%=Q)*rC`Qb4r=p1)qgqcV@oWCP!47kJ(7Z`@P&xw(u
zy0Tb)0TgY_Vz}cLC+gH^ZXl+(Dx&g#CnAlx9{zNgs4@7xk{B!<UYvZ}s&qF~s?o|F
zjeDg#M(fH73w8wpGdFuk7+lM2FQ67;+TE=PbY1O-O1&FOKb<-ItfYkbrqk^q0SkM$
z+C>i7E=wIZn$7b*VA_P;%NQE0C$dKej{7ly4uRdnu{&J>j_7=P{<NBkjq#wwZhiih
zGNhRMk%R30Qo=U?taYsI!S%{R23(5eKHrXEt@1zA+L=-UM1q7xo8A`h19j8AkIZUN
z0mQ(-V1w+hgF{f&x1{64p8mi5WRZkuDR@gJvcK=eVV?m_Z$w<&oAvci_F^N^$Xi1x
zTR>oS4zRxrfsvlM*lLP>psXww-*M@NI@tTMO8}3VhnB~RJh=ob#iGrngzB0yGphag
zc%wVviF_uf>#u-F;ME@f*2fj}4xjx=Ga?{~NF&Fln%o1B`wTF>X}^NK#hg}D67gT+
z#S<4fa&@marTupG5a6+%Vs(ey0J*W5Mgf5RS7TftK_R~{hPoRFR61hgcY2)%7gFpk
z#shvGpzcWFgmDSsizob5Rk6MO?E;PXGdS#L+q(c9Uq~`is|`hh=HToEm#wq?^64&6
zmL21Y9k^3NPWe-|+#UYgC@Jo9PG;lYaR3F>Y)TY9`(NLYRDc9JV8riAGksTq^Q;d7
z&QQr!8uDXnt6z7=^RJV$fpyW;BehNJtW=69`HtUZ4?qGB^H<(fxxj`7H!nTks_j4|
z6@ays$yWz^E~ArUuuI%dGj1oSQ~dZSpa=oovdfql=i1s)fLSTv;!T7SgHC*K0nI-3
zlt3Oa{UD)CmMt-058CX>80U%}^GlsxoE>U>m`UX!zlY`6wu`$94;LgS)3Ma|ORr=t
z+@0L?bxp=5chn*Q-Y}qgP1<#{*_pJs)Hq-1Y8Mgg>Q2x8*J*`<(%?tZU0$j}`2UxJ
zIH0}xHxKe7`Hf6pXPheCAPfciTkq<+j9<r(0l`6VEiDDHZ@2<pz%p|Kbp3>d2QTtz
z4uPEyOd+JNrgE-Tpj^Bk671T-4b!m+gJWQ*gkA0}XB5W)lR`lUFip;*c!~=QGR@Sf
z%ws7~1Vgik|6Aw=&SKiugruZ_T)FJ&YAXVzB6UI{qUQ?FaJV3t=;xOc%KuJX+7~%}
zly{gS@%*1Zz}po_NQO84z-Y_ed20!A{bDT70>D;zJF2jz2IqW2@d)q$Fp@_9tl<Ep
zl}p7KQ}T~X4nsp`j)tzK-LAYpG~Y~OPE2_qb`xV>$o>xzbLuk_J%)<nHU|>WWO@dn
z@*$t&rT|H?&Ef~RX^#~mPbaNM*|FL2C~!bO%vZ4vm`%`3LZrvWK3vVAB3)mjCPbZk
zU5QcXSqfJRU)&h<=Twu;xbdt%>qEu`6S)XLTes0ZB|n{51KXK7Zw(HhG0|WBNQ(=i
zNkJ;gUqYpI&h%m?ZU>-3+7^a^4KLUOO1&H*y^MsN_<JV%urT11W4O`$w_%?tBr6^R
zco-SO>m2p~)jZ;k85tQF0)ccG6#dJl#dBp#t?#4^Ew267$K^k6mD%ax7}>A?$RHI)
z!LbQy4og}-(jzz4s74a!z03I0`*&w=lKCp75aibY2eS$U^l#v$@x=>UTz4HW4HZ?n
z>InrTB1V_PpM#HPH1fOiB?0!#<k`!_mdFw3#`-I_kXrbCXpa0$M#pzs_I=*7>K`Y2
zgA2h2^WSC!{I&mFIg%7r=7Q}n;IjAYo<d%euNDdK4yf^Xp2-ySLqYtK>G7>&6{rnS
zPqyFg^vY}}L{Xk9hZPe%|D(fNDw_7c?8S1#)#oaV@3BYnkc*q!?(yDbe}8c)ngzOD
zzw*8^TxY-Bhj={;kQa{5O1B1-@t9E7mG2P<A97D1=SxB@i*dO1tI+(`{O(aLexe|C
z5+arvak(V50hsnMKu;r^{-88n$0kH|cT{!`LhZZiP&8+4I%%Il;JYjK*yxZzD@SH_
z)l#`ErHrM~{Pb<Pdu98BnoRn`RL%I)Yrx^6uM`h$k{-U9nCGbb)yaKjz0yVNI2aq)
zm}&4vNRx@n*#$s6uTK_yjEZXkuRf4vFlo%|89=V9+1kfq<%v>3wjBf14GYZFIWs47
zL?LJ%PeeAU&%l9M(=1AiN&jE1>lZ;faS65a7;FR`8@BLyXc50?CKeq?c;*Skjj+^|
z{`AT7VI$qbg91)d>^~YmXnFD&E_5|caI8i~?ywkeeY={dhcr$h^VzoX!Qj<C8xXO|
zI=F5JR5Vx0TKJD+BkiTaa9r*V@m&InMA5x7HkS}U`<|S&F>43v!C4RHXs`s(UX&)^
zD`<J36j2XD#x@9q7m>;3huUlVAx8@zxg_JH#9U9zs+%PL#wIY=SeP-%q8%?Wr+?7{
z@K$enz8laWZnkd@RN$s=@*a&~06X<c0lx5flp-6iOeXjgTC``e<Jsg9Q(w1Kp@pX3
zC^f*sC;Y=kjW%&j!)~I=Jm8kO^j{1#-^U+K-48_nS&l%GK?#)Hd$U;YbOh8tG6Ibl
zBLA*+k!p<xIrm^v_jKL!q~`jl$Z1+XK#F_t41zK-P2Rp-$Q~k6NTnzl(z-(`*;)$v
z;;>$EcglgWcR>l^2=MzG;y-d*&hsUMNfD3hFW_8_^HfZUWy&&Y9}f8miO2zNyQ)VW
zPHn17>H{!*yw)D2G^b`KTwe6(dU)Kz`B`Nj2N1qsM1&fbeWk-Y^U<86u~@jo+kASM
za(^&<(||6IgNvz74vuLwg5yNQ31nRpM+JaQ?tm)Ox|OTg8jyN0+tBps-`ho@FpB=+
zzQmJw5Z0gR^taZG%^_q;03RhXw&&sJmqCr@*02P&wTm6hiWG>#fe?nzS8PzA?`?0#
zx~Y*xnz;QlUU4=9d}Btlc0TZ#Kp`T&cAM(JC7rH55DIZW^;>d0=&eqY&0XKa$bXiz
zWW{Kpj%~~m214@q0WbX5lM^d1<MAtp8Lu5CAX(Hl+!9_(%@6-F@gzYwHj%y2JqO&j
zp=?}r#HK53Vp<Q}!onE_7gRvqV(x>pUfU*eR;;i2IC4c_<|QQnkeLq?C$PMxGpmS#
z`N=NC{!q~f$nKE_&4k6=l~X~TG{#x&_yV-k+UBQY55}_o5c4KLE4Vf{LKdzcS!iS@
zAlnmF(q|~C<K}jAB-<Z*2T8hqW3$%1%3Z-%9?zQL6Gz!ybS>vK?>vdeqdA(2sZ#HF
z@T+e_P*AV|&F=U5Nka&L!VZAL$>G8Jy1oVSKRZ0D5Cb@*h{X6nv$Yk|!g-JHg8YrV
z@K9e3V948FPyk7IsCD84+M5z<FcA@OJ_0LM6HY*S|1Rm5GL_=V;9MCZ%uur3a5z+g
zwAAqfR5s6#&n6F4Lk=v4N9KYQ(7My9BUj5^Aa|QAK=#J>w;=d8PM$m{lWi?87%qoQ
z1Wk2cz^Ya2k94PtD^S9E)Fx^$XRRI_LOBkO5;pBX9wxhr-#E%fOdI~^EPsY^BI%D^
zK82pn;G@b|uKeWV0301O^0whzk??-YS|eX2a1VQhbol~-pyd>)@TMj6`Q(pRUr13A
z92{~i^*RNuke`bnXaj8$hz!i;kNl6bUL_gGsb83wnS1;D-HkTsde>ZVOgthh!K)k#
zu-WI<*T*_Q8C7^d39jjjmBNy{HX)dKRd@Au#{T_0ZO|$|P7tZJK2uD_DJtZ#MewE?
z5H_nAy%1u;<Dp{5|M7Haa$jr32Pk7Sw*8$L(Z5=n*{yVpHDhcm>j7O(EX04c+<^$r
z!@alXFMm<`MHo-!EMTodxjx%kUR$>h;-I8_qv3%@gKPT7O7%y71;{^rz68L(Z0Vo=
z8RrYfRAtvK8$It)6kN;Mz4g%Jls{Rff8+^Oh~&F@we$}n>H%S5ohuZ6!o>Dw;@%4T
z2C!Bd(A3>9FA(SW54T(N16>iQ>~44)tFFGTB+HKb{B5m|2wf!O$eeGyLBUR`2SLhH
zX_Tqd2H;-J;smOq1WNBwJNHFXcLS6U28-1cePm>0Hklv#o{2TG2)zQrCa9&wXsKao
zmu<nxGZ>@dg`TrVZk@1a);Wus)Mo%P*DA0H+-nO_1F0o$UbP4j<*f*Y29F+~T>y&c
z--};xv+&2X;6(fYeU|%&0YNV334ttm6aYPi-8;RrxqCrrlM3y`6Z}9RaXt_O!mC_)
zGN%;K$WUQ0Bw<&i*$Bd4$T7AO(OtK|U*hNoPxDWg<)FB4_ym=vttTm4C0>^Z{{<Vr
ztDd^_?whMkTGP4Nl03ON{>XZIQ?-!*p~8gl=%fK7FB+j-;eh<nb)bg_zwiFR(!GLs
zvEE?>B)K|#60PCgyLZkmE{txM=BTKs6k1IXsQC);Y#@R_5`X^u`TPlLjow!Ok%$fn
zMsF|hYrkJyfJ0{@qn_Sb10N4DED36GAvL-kz7c>jvoXyQSKoR$CW6@0iQjiUqlbx)
zqqJYTs0dIoPulo$qfHiFCv!6612MY$I&ZVSClbeQLw8TD4giws>X@sKrsfUIEQa2s
zi&#5d*tW43Z~aYawmGT(4w)|b(Nl<{ayRkYrO5H$5X+T?=t6lWy9Z$yPm3{JUn2EJ
z)&@t1-I+O5AZ66pS^{<;QBmmYwwzg6S^67t@rcWyZQ8S8`vaNaadZeiPi61$FjSq6
z93*i5$OKH5^Bll8B)o3}m+w()VI?>i;XS>k1ONdUnV8zf#v;GNW2SwjP%hPd2{_Wg
zF{GKv)6=e*!^+v2BgNzL1EA1jefR)u9N>}gxT211^5lH4fa{F=<%|*Ngz0E&dzY6p
zr?{WK0VWT{6$AbK>)Rvg!T3Dbz-uHpIM~|ORu@>(ym;|qox6LEe{YJ4i<7fao|u@2
zC&^|Fs1vj<Vmt(Zz0SwiE~0W2FvBBb4W9c?kv#dI6K>7>pR(n1fAy|^A?=*|wChr+
zW>8h8yoNsYG+lpGlb6r#y_Ui~jq~osJ0aYftB(0OTZ2r=LM=tr+}unRQx%g#TYXhk
z)fl(1`OF&fw?2We;glbJ`rVGMuHg71`uh}chwd&`FX|SLIEOfw+~{qO?1rv5XSPfv
z4i=t9-=9wNSkUzMAJc99aN?{UO%4<V$4ng$&&I)?<{<Yqp&v?G6drtS%f)m;#rgT4
zI1$ct-5l?*cvjt(+rom!Ef^2An>fdamnEg8set%YpxGD?+{i{jQVp3KNajfZeeYMu
z_=$UuKBkkq_FUl6yLq@;0-SW?B0bked~XmNu-Pqv2j-W-QZ#ZY>(!1oo{u+1b@n^_
zhK6LX+8<IdPI0*1<^gS*!TESOu@nY)tadkf++A-bSYdHHnuqK0Pm+zb`y)ez6t#!V
zwi|DOj?d`v{zgzpXl3*A$ZZkebc<8GG&I4$X{|)R7nMvp87NKlPH~THv`;Y__`#=y
zVJ9btk+qDz-;EmYpn&<y$-Ut+FOmPbSe%^rLb2%1;RB^#(qhZX(vs0sVzPPMXyAgK
zo6UNdj$g{c4_@Xgv4yBPl{Mw~1WYqeSKnt(cAtaQ?y2de;w9$7Q;zqI&^|!++@L-D
zYhBB__=o9CviQSNJG+P%e^9`W?}#+-4v2(-sCCv0+HfhHwyCH023#>TK0V&S^e3>>
z53N;OEgAO3&`!_HKrK^sy9v1LDPZA{NSK&VeOtZLD1hti2XCN5iv0Bp#ofc>4FW>H
z=-}<$9U2BkXMY@X2yb!qqH7SaT|mRb3u)E<$jTZGwKnsX=6i@ztG2{IC9DSZBa_h}
zZvVi5N}W9=fatIgUI8{s4X{cBHgum02a`_EFE0%{12MqDZjj`mp`ocXd-AZcu?6FC
z$G~IKVR?jR4y;o<KRi50;>m(Vb>5FEt#0pJOs>h%W|Bo^s2=3zR6{TQjr{EEz!>vU
z>HdB|a2Pq+IYY8TU25MKLgOv-LCX{|VN=~?4<ARbu*^bu2RHXXt9nA?iuWBq3_~K5
zv}uDs%P7|#5p+G>;`~0VwKqMv#EUhnY&bJ_>OyiZ{X$J^=Bxnok?h<$d^4P0h>(os
zu9TXFXwK&Uxd2*a%Fqo*Sq*>x9uM{t?QbU1Ks__dAHN&%exq831Cc;dbTnm_5(zLO
zQ&v?iv)vE_Pf2h{h-A?kC>365gNuGe$`F(WBwc{!ciqH;3G}N_e+$6;#N~Czf`ju<
z)B_F~9ZO3DKs`u2NYMvcud#eyo`cP_w6rb{7gIODDoeS^o%8YW@f<uPUrJHY(CifN
zflE>p&|+-hGBY!$14W=jGUTSolUO*o=7`tL%?*61v&v!)imHOu*1tX9>Evk?@##ft
z=yLb{=!e1&F#(<D1)#M8L!i~UE0gm)eS-tjr!vtTO<`fc-w&2cb^|koDl=rIu;RAG
z%D44fm#d2FF7l>`7&w=m`JU3aJf=E2Yl5n_xNlaB-hlsN&#2nkzEi{x|1ktfV9IV|
zS`5bZu9CuDQ#3cX;qkzp4Ug-&w;2W>73;2u$`P6JY<;S@^rmA!X&Y>nHTdq$f~6h;
zw<p7aYn$tdA{FiZ9}4-)o9kctNxcq9{PfnMoOJ6MN|nNF$a#aIUpm}3mq!4*!~gok
z`KPqltCpF+WgVr7g*1fERxOu2bKYncD)#*%6o}=tBLl>REufeha&VumFo9_}>B152
zOyRK6O~%;;elVClsw-|*w(M#k2mphgm;Ok2Os4bTyt545xq+1gNNTP9j%>IdK<eSW
z!P2jb9mYWAqR#+5;!ZDNIT{radNpBBg||Xs5@%T6_|moKU=eyfeG$_o(MVv)Qs<>M
z<8yO>0YJ34e@xj<KG}RABur!=^!$eD*QLPZ*USx^8hO)T4HbgFuI!n!%vK{O?ULuo
z*4RoZ?>jWOdw<a5pg6k1npVb-(DSBjam+->jTy6>A|xQ;wH6g1Yls}Cb;rVv@3P=d
zQF%_+E4J41dq5XD|FifuJ7-j$Cv0t8?3Zq7-VHEG^lfmlVy<(c9vpgS%Z{WL_C?LJ
zL{&cg>nj)N3*(0(O=nbJfT*`7C*GuE?HJ5-c}W_opgxn3Z9r~UE+i;u4NQ&zV>wqg
zGf<{9U$HQL>%nL!`Q;zb`nkCTYm_ge_uSdpdEKR1&QYa4DvGF^%Y`Z5zRN3eN$4Fq
zdPj~-I+V&(RaFJty7Vk9lgnkfI5=vi83`~VmNCGufd8xNVt2H^noE?u*xG&~NB623
z<6TccrOJ|{7ti*E``k@U{aBtOExK=V@VKD+T$4!l%A7TYQe(@tRI<Ykmn%K*fEFPS
zCzAIfZ35=Cu%Zn5E6If39r4r}#gnz#q0m*$bb)Vc%A8T@6MslqaE|ffxbPmDA7g$U
zA3whf_ADj1WMfmGenOjqWy73iqi?-?Z%uqmPgPfY(rl>_xS$1HC4Pt|)#8HSrMkI#
zz$%!}WKDa*2B0xyhYQOpY*y3d!^1%kO@0Oh^wxTTmk`GNw3l4Cn{CLbIOG~S64lR4
zQ(Z$N1Yaq=*Nu6*K@km;jE(L6$EjDQLMNS=T3_=OX+dN6^QTWRPKWcoGd;9!XVq@(
z-`wUQQ7^1aO{u1ewYI>?R0sY>ItY~D_le>XlBtPIM#|x6c0<Z$_k0#8o6+=!eXvx|
z(aC<t38uzjOX}N!nZo(=Kl}?uCXavB$9K~dbViM7jaiM~Y~&evsA-k!c0Qjg=5zk(
z@CaAxkSpkaHKnH`cRmWo%I*FosVH4kOjKvOfdJlIt!puqfEV{|VA~SB)~4`sC{dW*
zO^<CT%@a(aaC?LdaR~!6wXN3w?&t1&$upP)`e>S?iZwB;xL2@AIP5H)6@ikVj(0sb
zh5>$oi25EyJ(?qC{eLIi^tMR3OVx%Ajb;KSMisLp6hk&v`?uP@!%HM`yb~7}2R;xY
z5)u-F!e)VKzruCqCt|`P;9`~)(7%Hs66&yAP*`|!dfL|A4M#wbqJGeHcgWop{GlC0
z^5GqOWK%BaZ560qS=JfUs10n)M^?2nld?$cDdHRo$BKT+<a5>F`dr_n*T+s<gGL~%
zlClZoUZro+<Y74<tOufq@c<Clck#>Ttv<LYne`vLt&xwL(w0vZkrQ%q4mk#<9pBLR
z!5BZDs?%6#1hOK1YG2gSQL199mr7nu@wR(e<&kS)M%MI56=x#4pSTc$X4VxE9`Teg
z2oee;IuHM5%toKI?HirxyNV*U`e{4#f7dcBg{*1f#Fn-%6(<}QA&i0Q$7h*X8<PA)
z<>tvk3C>1cs9y#Q@G5<|43wb28S}KCc?g)R-(M|%2N4?y2E>q?s+n55?A~rx;r1UY
z0aw6F483*;JYq0V2e(GIt1OUA8&}RI`Z-;`-RzF*WmkdyA}*c|B;~-!ta-SKl9JN>
z?w}gxd~YhOE97kPlQ1-R&)=6i5x|5brv&?|u{sVMbY6JVLbDyWkjtpz1q|-re~oDi
zL?wfX$j(uwq^(Fu8Bx7mhdhNJlF^j82Y(BE-m1mMlgxzB38}iZq#z`P#^e3aK3KLK
zS3l6mUtTbWN%fB@z35@o+!;P>Vl(2WN5geYV$wv+`I=oXvzAl23GlC1Pno_qz{!js
ztt6#!G$qiosDaxJ>DgQsN`n<|vO`K|qpMm$(%*Ef^Z9^gsg4_sd5&s44Ph#9(`g8}
z<BKHMdA|Qku*UccWf>nUm9Ce9XAtIY*zYtC{3RKS*HxW&X~;oFdn6a9NejQ^eEr$R
z$thbVT{w}$#s@0k?W1poVK9|H>1)Ce0@jA#3WQ?Uy2H9BC)uV>fUN{3W%m_>{qFc1
zZ->1}8vv#PlBbMu%6dJ#a{s9d4@fiH?a^pRk@a6%0X3~%@-+}_o`BwZy5bg(yg$;@
z<U!wH4m`L}VcfoHC2QBBEUNGjGpF|6aypMDgp851NXBX#A|hkVZpBdDq5G!&*oo+e
zAUH#d97pcALtG2ntSjh<ZqodPldBVdFk$AxHX1oqE+0TRryWV8UN~qzm_C7oS=Bq?
z67L!qoS6>8%Trb14Y(yfxfbfV`MiJX*-PU3qn~9BuKE3MB*QHyo@v%>a(_>Pzm<`D
z3<5*`9<Hcm`l%#N|JE^?zu3WCI)Ap31II1$bAr>aENWNDiNaK+50yrIsv$o(kDC4K
zUK}||64pz!gI0JPyA}SimF?I#Aj|`|TTI}v0F@wvg-fS+`^xtlU_}9{>`tNKBfW@-
zNYg!&8Sy4?Z74Auh)<N9J$K~+{y}=98KSjf94r>I=sm6y(kZ+ke}(x-M;8i=l;^8c
z7uI?}&MBS95ne6<s*cHh5Y8)ALxj_NU0{wBfIUd!+}-7YoAwg)P@^Er?H39HO{z`N
zSL8yW$)W=ANbEY|2prI*|E%4XKJR?Um=FB)JKQI_5eMn8$$4@oG}_}(q0XS8j832q
z_g&nU@^9kuLwdPVeTTNPJO52)DPuuD(XP@8yUwoM=PT^(UxxZ=WJBfyi;8IQS$$>%
z7InH$B44OC@$0)oOe?XSjL5N-@`e3@IeA%++uFaF^ZI0cEN7zN-kpY)@BU{9CjHn>
zSmphEI`Kr}*%E|Cqu6M3B+|Y1iQ#RD_BC8Zj>>i3FkGQX7(NV*dJT&Al%^YEK3vTX
zU_kZdXf^W!Q9@mDgDJ>+cBYC6fU#Py)3dF1U_Zgp*4B2tlbx~+qD-;{v@Elm8yX4&
z5AWF!-xJW3``*nCCa{MAzc${I8(su5>0>(029B*uP(i?ZtR87xdK{{^R$ha!^s`%N
ze*y6=&B5W4g<|2PY>{h1V#vT-J%bbsVy+ZPWL-<rG{*9d{?Wmn8N=RWv;D#jNLV-7
z;^0IPNx{BGEqH*<b%~NYbBGPCEF{5GCss)o|1VK-wy)*lIvpz5IZxsU(8Ht~UEW%9
zDK%uhmCxr#b7KFis%Aqj*S(<@aPM$K#h3DC{q}$a6tK$YBMh3BOH~!p&wBp;zSx_h
zH%hMrY$?d_2^f;_`t4iY!}(gMs~MO&AyaNB`3eE6<>xzM=nHv#L}rvJ3Wm7=SC2)H
zD-F=V>IN+TfZ*T^_Zy%FoR4!_bb+gO6Pp7(Y{#R;0T8p|_`G;Qr?-1yAqnujPk@sQ
z^gk6new<3?akI6AGK{tyJ-IJTDxaiQWM;ypL2Me^b2X<?3&SN|cjW*%qiOQS&SJqE
zjK49oinwkG1q$_Xxh~}5LBwm82;KEyWL;}ld<S>ww!;pnen|1V7rVHI)-ESiuC|<{
zoK}nNxHpMRs`Z9dXLS#jzsY^C%bB=pUg*K^BTTV#Mb7}K3)wl7&d3P()2d7rLM64!
z+UC0Ps0~73+l`rydRWMY;whdw?tZR(5ialH#;L~>2{i2{Jl&G%it}J@GPs;Ylu)Y;
z<a;h~P0T>Dbh(hgrgN$Wo2!t!+Mb~Qu0XXiugBtOp!;E}GYB&Xy}hl?2c$+!w(C82
zp$1lEn;RPn>!t8G)u2wI2#nY&^%*2rOBW8YrDdW0shDrj^Na9??C)7Z_gdTfn?=CO
z+MKT{p;!dc%yWCRA?RhLNNm+!KGTXRwccYxjfgn1Vt`9LIB?DuacXjYh$hmAksq9S
z6QVkB5GzjciifJQYx`y)bkJ2#zR7%%TBEPN2IW0k-@$eeM>M_jSeZ|EEzGZ@K-Hl#
zm8D5?VXc-%VMRm_4dJ(2SN+82$4sbgLyJ@0^RsV?Z-->E_sBV1$6tI(3ICgPy*I6<
zrgsHH&&)g-e%9Kr+!)#%nzAgVI_uG$AYa_|hb{?gJQK}Ob!|yZBPc-U_AhGzxAm_a
zKfk+BB}XTtPhjq?R};I`)6Dk2h|iDSKRTbt_@K4CU{qgz)AOzhu;My#-U*VEXd4-X
zxP35Eohp<(ilNO>o2jqJzWY)STy=rtN9X+febXu8>WKE~i{kWjVi_5kPl`5c`xtdM
z9$14tJw3G=rK{-|k^$g$k}o}$e>=Z-Cjw;}jPf+cnqna<n^ZE)yy$Cks&%VXD{_XL
zFpU1Wl7Im%ug`t_yS8y=l@$dFks#-P=eF^>h?h1bc?|@7n2LNQ_Mx?d-HDB_!!Dqx
zi=t>vV3vAty$O~JGohYB!7q8aRCV;u(~<hI`93VFqaeQVvPi-D#4-@Ppn#l2(y2~#
zyZQ`_<r3RYUCLR!)QE3liTt_Dv&}qaJe5&9V&{Q^dNjYXFk`23Aw4Nvs%l65cWT#1
zL&b{4Fa|uvB%2X@-TZXcH0#fq6^r$Dx(P-Kepr=uF;cwMdU%ABdsG(vA>}WY*)@^<
z66uhYq1usz=Z4HiC%YCGzh$cLN&m1*vfp{j;+5m!^+zc+=!3f&uxV`8o{u+sX6k>L
z$rCN2YHm&g-btP1mTv$}O`R|m6&HuPJ~iBf@mHV<{}9r3{hfNXmC@xN&9FeN8mY(C
zrl!6gD6tkXngLRfm5lV}&6|dcNi~W<pTa^~fYmVHV1fO{mkz2j*Xv0n_=*lIwP7|^
zGBo8pdEluf`poyU$aUO6qVm;;YNF>R+^1-sxR1|#>(Cp5EC`xo_W9gGTawB{q>j5Z
zImJM^4AD?GmM=72T&pncW%nxh=)8nHzRxnB!-N=@TU}f8WCUt|9yPYUa>)c&2kVkb
z!ux$TsRBIlmMW{yGC9*F^oXufExGP^*V3ZjKz+sksQIO3viHSN5-=~Ml#efvoN&GS
zkaEseX~xKSj>Z0Mxnxh7+x2Vn)(iLe=D^8&z6~n$@3F(aiRygG!^=DsIIKT9(}jnd
zHAZ$Rvkml*Pub;WoVOXk_U$?nNES-9pB0!TqT`5xAEFK^@p~GG9jK*NietN`rJRF;
z6|V^;2DKR&Oh!pSz>nHG0VO-L`E-dtgjz<PZ0WYhYN?5{8ZhWIjPD4%?isfnLH;Dk
zg()Nx%jH1Dm4wu&8KzmxZsHEjor)Wcp4&#$v(6%BZe2$@PaMS=H7ZxC1}XV2o@0AO
z>{=|gYBL-U?sMY4Mxe0{uC8<Fn&=JB?^b&ELERZK_laqniQ`mcG9gD=%kzx`Z$RTK
z1^4aQ48~rYDUSMsIo&W=YtywCqY+k2e>8C1Pj74|+xmk?=+6aY5MG!k`3ReRo@@0Q
z_Uyq<gzPQ;2Ig?pwtu9iM$R9JyqPd?-l4hv+=hQzJ~)(mswxbL&(9^z%o;sAqkoN5
zFLk4uv~#w>nOBYP?q!SlDYLCIkN{fv?}(T8jN@Q&Sde+a&1)sEHlqtc?+A2>9SuC3
z56jnBcJCyCOWREJMe9*W%J%695=}u;Y@;L3m|`L2&dyt~F#NA=2-w-#q0Bot0{2hA
z4F=3Q0zN}OfRA<w7*oS$wMdiiM4$r#{pHRW94NwIc<+w|P|?cSWP0-j+W}q}2Y?nB
zsKd<maBGI$OtTjsfWQH(aW2CKWQ34K`&Y1yU0!j~^9RSz2DIXWZV&TDeL(VF;eG)J
zYIuL}gwNQZF<>FwR8^N&=<m*1Juk6p4I&ZVXfAE{q73{d(2PuYd8)IkbV6xgc|2ad
z)G<2mBZot7d$2xB1Q!0JMHdWs>00diK@ZoYs6DsnIZ&?nD^I?{&5CtVGU#%nAS9A}
zLHMY1gxCyr(r_X_I>)n+GDN-PXG{K+^_7K4H!{1P<$=62d32y^(vPv#N`m2?i-z+F
zbnxs@mh}A_HF||#fXB1_*vrSs8UVkhnGUNVd?=`A*jv-tY`gAm9e(@f-)hR>f&_B|
z`1U}{*6j;dsJZQIr5Vu61gQ&M{$zIXnd<8L0Yo!sJyc;|GsQ4GA1;&t<ZU;L3vvDh
ztXV!Pi9)|zKAlj8LBh!y2j=07mFkIr6MZdE3f}GPzMPjqUqr>zB|3kwc$gTTaOKNG
zkqiKUh)PMWMW)1{Pv1`<9k?`4*=3M+m)<07sv}RUZ=;y>vd{2TvtwR91qb;8iIcmc
zF;j)15dHfwY^1=pMFP#`{VFR)Ec=(2jNq)-9x@}{pT0K*6Tf*MV#O>iUtY$Yc2l7X
zrAkeO<$kReN!_xgXBOw?LeTJ6)$0P}r=L<a3I&hI{KJ(Do;POX?<b`yJe&-|Q6H}I
z7F`*l6B_#mIs^?1{_>S3$Aq_THS0$Qm@;1@gH;V=m5K`tF}@w8K^M~LA8LkN*5Chk
zHVGxV*WYM*Yc{%WL}>y)cQOEI_eU>ZoSo@`T#fTdv%>ft(95yy;2I7khs4DJh9Yu|
z)dz5_*H8OGHDFkFxjBa-5WdTp^&8*c>%(|1UMi&?7&Y9n5%4Ve2S=&x-#&6a#?Qo=
zkxzxkCEr^S3lS1Ryd(Gcj0Thmic)6$!u;XfvZY)5D$T<;FLV=(ly7?}`YW+Ri$ApA
zo;cJT4-v_trGutvB+B4T<C*eZ9KBJMdq_w$Mm~9Py7-##%dX})=L9f9QCH_caN$rd
zh$Z`}%qd^&V$xSsR9vrDi#sGxN9>HvyzLLg_zv$YaW%|zU6$Q5R@2Iun*_GV+&=Ey
zropF=?3Rle0KK798mJjXpHO{5cB?jp1_19qrQpDSClA!)r;rjeVn8vL0PVeNVE%Ih
z2Fsk9L%n=}B1;&IIFaPK?2RPn1dL=*p$<`ELjNI5V!dk!;4gt9S*|JTM9CRix>LKc
zTP}re0t#rJOUzJpp>Uz$?sB!U=Gbp(Pt~^QKS_cw4g@o0^-PWeC+Fh<d}05AlJfxW
zYP(CxSF?L&ut@pLwbnz;)Kx+I8X2C`nhph&>CZ|^Y`WUI;wsh?^fT`3ZL93oen8`{
zsj~tF%Vm{bHwf4mI2WDse^Zsu>XarFeL0$)@Dq>-bR?cIgG1$H1Jg=wdSt%l=8T>q
zrms!c$)_b+aTEDCt`cAg&bc}qnU(Jhzx%j1ACjqPc4Bful|30|w6;P#e#UrkNB-X;
zb*$}td*7T7Y>YwtT?12j$cTu1z@)W5Q+t5L-rm;-Pfbl785S24(**|Si2J@mG@h?1
zm0?UvOXILv6-@3fqYoJ0vCqiP#^D)~CC3@u=<$?7e8uxptDh#>{?$Wsr99#*Slegk
zktPSvykH=v_C0oTC26v`<283xTc;y@$^hw*p@(Q0rzF$Hk5!85P2zL(Yzl~QpkRHG
z;AhDNxj{7PM80CG+8Ein)l@tczRK_8z2NJox+*79N6$M}FSta5gKxh3z%F^O6ihHD
z$A@QT&WpMF16K3gEenIzbdutfL&Ue-^tIdgw(>sbLVmE4K!0Hphs_Gn2QCdj4CXNg
zZW!}5AN(VYGz#|N=_F;a!s77)UeET9EW=miY3wzvb<?ZADBInMF;2wB<E-?AbPLd7
zUCg48>M`i^Y;q-O7NNNPT~H$v^2B_p-{=%I)fOqK)U)Naz@v1seOjBAT>;)c$|I5v
zw8PYWOI=G#sZ6ox9Qj<?<ZcUKXeh>^udg3vTmX#5eu0%1A@I<*4+bXf!1?a0c0xr3
z3t-y?<EK}sKn8nv-~@l?!it^*)*sSR6DQN8uDpYp=D=vutz=*u#IV&hL|=QFXrhfD
zsTN}XNX_^wcTDFZh3mXC;CD()(}c;68XlHk8bwA%ww{GfI5nd}q79Md;)_QftMpuX
zpl<~xr7<}P>^7y<^ePZyjj+&HEmav??ag7mAqk~dJXwKhfD;)O7wr+C$h*WXC8*vY
zs!80&5l8CYOqEP?T2BFU&V_d?LX%4_LEVUz@&v3S36>=SK`Bt)67uakN?5aq?%giv
zaYmBJ{cpS((4Ta+rs@;?0k;nlXYZu6oI4(m+Wj{)9B&LuHRM%c*18=fkZ!kS`56SA
zJMm$~u9wOgbTVLxS+Vx|au77~9wAEyrma^{j&SVmcT4hS{1<x&%OI$4H063g?-_KD
zvUEEGNf%27fQc$p667KwgASb=0GfBDRaJB&9GRjPiVYfVDf(~U)=A)DhrFu1{kG3d
zfsf|e$V+aLHexD%P<9gW75zJLE1@~W-4Axl=jYELVop{>c5;zCSLfkMQDWC>^KU@{
zI|Z@s=<<$EaAj=R)VWb=>}9&HHF}C2Pf8d<t$YkWL;S;R85n;4tZ|kd9i!Y@|H3LQ
zUAN#7nz(N+4&jqKqmX;m<GLSMb)d$J{vG`D;hc#m>b2mksK=uVngoaUS?;t^PBy(~
zm0t&E0?bB0c#<8C$9Z}rBqcP%bu!u_YToqQ0j)>?2B|XBQsFl@ziYE+a^R|Qt1Z8$
z3j}KqroF`bAAc{T;ECtweOjuw9%yS@4R_@+u2j7#pazFDK=hbSt0}RgLoiD!Np&^e
z7=v1IeSm$b|7<!kVtJwuX9SlMG?p$?3)0ZS=w6bNx8_e6<L9y)2-&Q*2?^8R#(6y-
z+^Ew^ul&e({|d9ss$^rEk@w~XQ~3N~pqu`J`Xq3$TPAE%GVXp$ERFwh@YdRLe+fZ|
zkja?<W{Ay7F$4M9-)E57b!7;bm*D;=i3$Tdgp+WE{q>vX)~t4s>-8-ytP6fE;mf_C
z%-UL;GdCEJ4n6w_DmBpVF|too3*QnI56n!KHlTTui(l2rTiy%|YUHt*8v5l4UI-kc
zbUmT(`>=mLE2KcQ;4i3@HdXBrWC*?1RI=3qx^uNM%SfWdzDX5}6i?}auKJtB?8L)l
z*BEfVnH*G7Nc$_K`@7A;{o7C*02B<C0(cwHh@J2ZdYM$J=Hk&qm+%H2pq!N65T!*!
za`OnldmPI~h5_?H{#KKffP-%Nf9$<wSk-wKHGEKkQ3;h$KxCvFm6AB1C?MTkDhNoY
zbSNSUg3{gH(#-)8X^`&j?yhGaocsOqeto~Z*Y!B}%*8zehyVYVd#|<j+Qg=#?bz0C
zPg47J*OhT0C7e=A<$K3Emd2^Cmoz#z7o4ccQ|&ErhT7QO5*)NS_-DELAPPXTWaOz=
zh6An+b6cC3H!jUy^UY>uGc(1g1&OnfFZgvsKR;mE5yT$K5qhronNCc53hK%d0!ukL
zJV)N#zvi7SK?5K?NqWiWtH{>gX{It8LlOelQ(tJ#kgxyKz&_}qz^0)%GJ+Fx&T{vc
zQN&Ipw2>`slYfSPj#Xpb78zmju0B;RbFs6M0kt+zAFRu?VR{@RXW13-y7Xr0Wikng
zp}x#W1Ne*k&2)Sn@Str6S8@{oF5Rlwj3BxF>>{6h7{54@SeW|nFWu=c8i~26iu|du
zSSLH5)@s0gVhFAK%x9|B=U&SyLQF^OLW?P%+5OMmW+l;VPVC@o)=b#^M(=*E?|Sl|
z?_xG98B5j7GV!xJbNFv=Cc9k2*D^;DBcfD~-?q~f*T>U;%Bp?>c-b{m!PyR_uSc77
zilpr7)y3{(1Rb)8b=U(HO2XL>TO9qeqBpX-eU&{z{@BUicvBT=1qrmNh^yqts;I6)
z#}?}w`y63*2ygxwVsA*h!|=ub%IXmwdv?fsE#7eZPxaiTPAZ6Yle24DN92jGqsBO{
zy6BUd+pLAHdoR0YYfPG$cg!Zb-EceoDmRUu+0q{hJ=^J$3gb&6^%poEEDDF1uq*3e
ziU5Uox-o)a(0sY(_ix&H0ktZokM_4#_y}qmF@63Rwk`6%<}2dG70qTnn8MKr6}LGw
zq#Kjcrle%`bJcP6SYtSzGM?PPf7oyI1!qhG63Y5DH~ntv1h%z?k=Cz=6nHE@pmw0)
z;3PB|NE|d%*gIUQu#lp?XV$y0w>hquwtrWJGI8O2mk>w9aptkg|Fi(q2^1f%=2m?W
zPoA17@Ay*XT&n7VTaL5GVt*5WA*2RJVnF2Z<%l<(oSBmnFbx1lOw_QM6Z1{Ffpbp#
zdbkh7g45b-vF3bE#>N8V8<h0dj$NI<K<rshIu9&39*fg$_a<{ZXJ+G%Xp|of2z;fZ
zt=0IxqRuXFE%MJMbtf5)^J8-iEaCY2Ou-4DF#V4+isty)4X?go<eF(beCaxvw~W!g
z)2VCvwiYNjFo=&*<@R$oRqzfOP#}aU@`h{)VvqyV*0f8;Gx`fQ=+8rBW3klsroQo?
zPkUI9)HX2-c$VJ!B74@6lbipszzaB!&=9-}79VQ)sousI<!;LhDZTy5vFAGch$k-&
zd_`&BBG2pM6)jj=U{?&H^llur7E<n3I64!ZpX^*iNQmGzv^C|`>{0BjiG}|5eew0$
zWNvu0HJ2olq>w$rt!?3jqxhw|<GmCJU4)nDE?)klJug4eRc(l7Z)Fdz%Rn}v>7p**
zYc892Y}}fS=GG9jr+e5(6}!!`#B4>zbzx_0P_OreQ}@tV*<dOsy0zrAEwf=zAd!l$
z-^~x^ZCeHFPf4)OdtK(?vdB!iPWzsfk9L8#eUPH;*EJ$`V{vPKy+O@0@+gwpI3jh}
zbTFRG>azv9(#4?k-%V{$Py{&g@pcG!Kkb;MY8l)RcKO+1RyUSS+8$%eV>io4hIh{~
zth*x=iVh;Hqn_{>7_gF`1WV~~0=m-Y9C+5moty8rNw%W3L0SG)F$MQbCq5UmG2`&~
zc;MSx<+{*Fm2Y{IEm(lEe0v(|Z*bh&GaNPa;C<V8;lDg<q-=$*pBT=W&eqHS?wF~j
zy<aBzw7}Iz_MYoyf6VH(<+$VE`5{y0`bdUAov+U^53$wzIZkIOo_f)-r1!aUi67|9
z+rp%t|6%3A;{4f}NV`_OnrFH7{ONrffvCm_BnT4l-NThzaw)#e)N^$ZlDf4)+`43Q
zI<Uaw*t7DSBo|8)hfjE-*SKddN6#iJfS1rFChfp}#AUd1Zgt#p-?YJB@ROzM{@oAz
z?G4#W>_NMKfxWk4BYwEEzspvWXzDLbD52;&Di4`%KII#%m=Nw;*-!nZ)2YHK_@bza
z$x^SK8OwQ_1^aSG-gX6a`EZa2b1Rb;MIl{%F;G%G3_bZ$WkpDcNyo?MdXML842yTQ
zTx?=CZbhnV-H$`Tt?gpe;~nUcnoBNmsIFzfJnvKHOYl}K@0X6?vzfzDG(OHrisoHo
zt4{)g^#412MpB<&9Roq_05w_28-4wd$0ARGgMyQ2`Z75j+V^OM<pylG2uZUs?-lWe
z(ub&=#yr;<>1Yp3jVZsvR?|X0v}xj_e_}poXCh$pQRC**5$}J>bEC_`L$pB&X*gp6
zU5KyyW_B+R+VLXTnVg!H%oP%Nl+(ro;Urw)cPLjRs03!XJwm1`2BGLQ=cbr;%#5Ky
zN8mVPmmsuvlPI4}g(6tpqr#kTY4=X3f=H6{Y7ShMEByZF-iuwbW63yf9S0%6*oRje
zv50y+EB$dke>~(6WU2phUizzZJR2StSN**PfRZV^+-fG)w8D3eDOMi41^}!z+TY&Z
z>@Bd$yD`XgIc-I|J<a+S96jiW^iVu#A{xq`FUzz;7G65O8boN8nlvHXemlF&_RDKn
z6;~!4C9@b(_EhsriuGp=bTI}jr>KYh?#&x15+Z_;Ba59`46DoiOv<ebA2swAY)b8w
zpJhK}>O8qSx-KA^_mIY0LB2ns;I|#44%55<lb015z>w035L*dO9Pz7uv9xN~%Dno+
zlJOhphL}yFtm_{Z6xRJ7^eG*Xg%C2Y_=kPkyxfgoap7@g*B#SYQ6SE|E_DzP6FdVu
znl;upY0Ilp$k=Au^yTq1b)`PHhcQV#ig;L9mdHKYt71(-wl*~tQDl%Q)o3!^*7z@a
zzD65M3drw6!q^}jcTZo1FS>8&#z9iVYMDj%bY28^aU|{j;^HDg^-j36<O}gjH6ueo
z(9!CPZt!zwjQJ#aV~gFl?=xk5DC_foSkf|Z1sx<Mi89tYvDq0II2M`2Y^(o?c>9pM
z#Z3XJJE*4CnQTcF*npP3pa-2C2Hn@!#ZkrS>iy?$8FkIEpQu}De$h>)8NX5k*}|{9
zdv;jDlfOKDqRI)Icf;-gmK*b?rR#?g5vg*veAZEF``#WRX`JEEIw<D!?x?StdcKZb
z3V-UOcgarDe}Bn8o25*0ANrpLU<myq$GDP}WszS*vpGlfs5{_6K_NS@ve1DVqEfRZ
z!pQMk^h^(jey&!>2Puuu;x8Htd@BmrpE(M&0NMLm;Uj4!RMvr%G4Kx22klTl5#tla
z-UBkbSSoInMBT6tc4b+-KR_PFpp)s-Y57Rr%#!Ym7fd}=)xunbv;KGYOZ6lpG2rFv
zQ{~eaMoRBvxzKXAo3hDe{pcn`EF1-CQ&yj^@MPdSMt+fSqdHQi%uzZ?+g~v#bT-9c
z(b0HPWguABn@Zz<^{Vq(i@fvr-PYQgWA0bSY}q2ufIRI?9>2eJa&MY0js9bFv!kE*
z@uRbLU$RT*)%stFfQ-QM@ME=`{%`I-K8T^V_k_s@jHeHA#U%1W!kw?)H%Tf(2U<io
z-m!^#WGI2uqTy$((|Jla-SlVa7jg25t~;l^AMfen{rRy^ol)Z81t@a-t97xRJIfv0
zI+zN1uXHG<rxrIhQ?rf?+~PNN47;znyJrVJmHQe2%t;%Z7fF!RXE=p$Gk!|vNZ0ng
zmh<LIwY~d-5B68@_&YoB9GB3MoFotOZ)EV%o>A<B3Y<*}9YuyB9*hELjo2a5+GkXd
zo#jhiiy-3VlyNR$X7x(o#LJgAJW@J@a-aC9%y+L<DGgN2#m*GlWr`f{D|esF@+|CN
zVM#|r#{d0N!@_tmZZALbOLb|i&zXcbvf*pc6~7anOYUsr_C!hyf}*5F5fBAks)-->
zN{~P)tlW8Dp&tmDGi`5XXVqH*1sS=EW&BxFUO@qEqOe|l%nIeLsn7z0!?^D&C`}%L
z03AHi<cAap6Mj+GG65#fP$XVIvOiuYnK`@Q{p}yVqUuH$P|==LSs>7ShYIT6>2Q&Q
zJG;wHl7ar#QqZE_k`0H`TjFZdaVB-qU~kgZ@zkd4-9V?$1T%Cu{#o2I&FM7^+pl#p
z>@mVS-Mhzq$&S#$4=uI{AODs#AnV9s+zb&tQQ5l3j(Wo$zd|b4)k%z#ss*AQl62uU
zr`=xw>>b?!JY}x(TsxgE>#s5-P$ISaq1%}=Xy4EZ1;MieDj?fv=;=R0AJTlYQC?^c
z_Y}}IhuL5i>Q1J6#>p`gy^A(wuCTEQwF*e9PadqefM#uU``F58`j-%a8SIV0BU9rX
zY?Ju>3O+YKG7k~0B{#f#8JLkJw}=Tm45H0?49{Kjj{BE5E`5uLF1<xkFQaii=lw{I
zD^efeOGhnqR3=?h%m9}g=H8czcLC^#;gLN)CzE5INo$%YElGx4rrP15Dr2xU)yfW5
zv9xJj_UD>Q$B;dv(cQhXKls8AH>N-18){}RJ&OQr8K%&F=vRF`y6wa#JaTbprN?3Y
zp;KT(0jb)3$JH3oZS?c{+QiG>AU{i2D8YkvD~zk@f{H+HfuPR-DKeFp!m@f*RYM~X
zY%aIW>l;ZuL8VzpI#5Q{M@(GF<H7~<v|7(oLAmT+4YY6~Iul=?xHySc1Q<p7e(9Xa
z=1rwIdNj+Z6ze8V;#AQk;OVtCpADsZz>2+;oKj3qkQ>iMBULiPXRlv$7+g0wbIqOq
zQI!9>(axGDPoMezF|q7!c*#cAU;R;Awpbatu*t&xoVz-~1*8=&9*!qGW;lC&ef6=^
zmlgYJ-=7wwbgs9UKejr%vrN~MEI)lat!KeOvoiAan{^R=^;=?j@gK2GW`CZ=z7YHR
ztNG^#9l^8D)!8TF$AlLF>n{T6gsV|#)awlZR+=7eqH|n?)SAdUb&Dc*$GwdVx)b8|
zXeIxhrCuNCTGNekJMRiVd6v?>Xf-?YYdW)s0Ie<;UxV07wxQzIH1T=!?`K{-Rf`hR
z=0}e@ilyJZ`wCgZ`o1Hc^=kLN!Gx6_;nP>I9tUkFe^t4hI9;k}f|ie7V)cTB;}dE;
zcxf1aWyD`ZDf7YJyv~5@%40x5=X;Je9g10&7a9$6cjlw-js7v>5b4-_YGwCMzoW3J
zRSo6O;FdGhoJXX|Q%pI&=W?RH)f<_sFxtI+za6d9h8s!>2kSC;r8J7KaoM<E?XAi9
zzTd4(YF|NlFc9bL^yR$=mRz<vCn=Z7fdwtg$E!g{bjR0-7+<`_y8rRPds1y8RlZI3
z4?lkn*l+xE9)*;6uF2oTkY38eC)Zo&ajz{B;3QqRlEj_LlW8}WI1wWmNQL|-j6|~;
z$Y=5<^iTV9h$1F1)26k6kCvKQ3C=WmY#(-c{?Zq~q&Sh}AkNQF*+>#CptX5B@(-Cr
zOF-NDeb6H6_FV5^FBRo3CP%hI0$6HK^&#w|#(=tpT(u+#nnmv*ZcCq3iZ4us6A30F
z#)3&X>m|s6Cf;Jk-o9Vvl)J}zBk<|b)C1TN&A@h1gg<&KM3@5827P0bk(jV&5_)#F
z5>i5M>mH$GhZG#Bg@1!F#Oi|Fl&Z3w?(^Stgqb?z>*76|If_%86hyG}Dtn4K*@{fH
zh7@jI+{`iz!g6;6&(?^n)>ST%QRRGe>LlMAnc0}EGZ`%p2H|m-jh}FQ|FT+@BPZDS
zPl6yE0O;xI>1QB*f{xZVIE{K5OV*(<3^e2dAbO?)(LZ#z(YBw&6gZf;40SKoL*#t#
zaG?8Y)D{jbLs6HGYG!C`)MC2^p!%B@Vq#|18hE?T<z29@i~{I863*<oC5-@Us6f*L
zq8nYd0YX8m>EF1TnwkrL;xE^I!c7EbMg*B6;~_7zfo5#*1Ixlm5Lhl}MJa^!Pt@g^
z8H2A~*SM{iQ91sD-cFa_H<|Tflt0=BMe#cXfY9ct?Qqr>&TrYwI25*I*)MM*rhASb
zRC4=4WleT+dTKrD2Hl_9V~{O{Z@WMH__PFPd1hT%0VOfWyY+L9p-V6nGR#DpSJ#e@
z$3i{M{N0L@y&P>cjy3Nc%0flu0d1&%08^})XSXs?3k_}o&wuE4e#+f<KFIwm%$kJ9
ziW)6|>ivILf2=u;T;bOmPJy)Ce#v3P<+8u?wo%UUI`v0u$L>&*sGhLob=8e|(b)st
zk7N?#c`v(Fq#jo}++rx9wAvN(Yhm1<IE&@}TT3AH*~=JKl^VA?Z!oj0g6N;LGvECi
zt0QFuZf<TyeOY3%vNu5b1}J7%R~K4~G}Gws5aJugZh*&b(4GWpx4XPHpL==~VHZN-
z1+!-9cc@?jRr@n5t49dvmiAOGDC(p6^XFFp7(U@YUO`2^z=uL5m*YKXF~fsLAs`8`
z9&fZCn1@WSuC{^ygu~%%sFW2eG+#zH>qtwZN7In)uTGD4|Lb$ls#*H7+3V!XeF;DG
z0iXnd<_Gm;%B>SK!g}|op|$D01XqW4$&Xq&5D=v2y=xxAVWzns1sSdN_LAkLV(Ap+
zhRL3F9*iP&iC>4Kkb{vb?lQC~{7p2b((T~R%pwam6(?>~zHUbROXQao^e!?n5x<jr
z2PO)u<^$$+t6;VtKYql4sV?;DKCKpLYHIp==i!G(Z<;Mm4lJ|Q2F>fCc{Hq7vMTKk
zr}e4^xSL48+pB$ZrR&;Zu<B<&zq!;{*XHoZc_bNaf2PVeV7{i8mYSh$VX4jHEu{=a
zhR!nh$f@e%mDN@B{SF$=L?Cfth#z!R%kZh8W0k4&si~JTE$W<NRpf26b*Ma3b-s49
zV`BFqEs;*ADH81%mq38mW=7UHCpn6EeGwk-A<!EX$VP_ZvC<%U{FIA~C-saqnhj_!
zHo&j$RnU;Bao?5v5wdCa6$U(@ff38j&JJ<u(k1T^gY~g0SPBA%Gl7~-qfW^4lWl|v
z@KCDTgV1dSaeBPp5j@g1x3bdeMdA8S4ctH0<m_{r*Ses&nY+B2+Ues5M@kfggh83K
zUg<p<N-Z-FA(Z*-<ZEAzkKK+rh1wzC34aTf2SUg9GoejSIVw-w)7sM!z(-n8d^g5W
z#6Z!_zm~7)Fb>jmV!imC{>Gob+<16-^+#Xy1@o)QD<jqEQg4Mg(WjupJ7@tq=63GF
z1x~ln(MAez-?iZ@tJ;H1ZyG>>4N3KuiSC0t?pd1JmU?ss6}HPE=8-VTkD%9ubkt+H
zDeI<~>8W}jkSWu_!3y-y46oP3#Ki09y6&F~SIl8mS`+DU1!Mbj5W|_d)>|OV-ptE;
z!->h?s(1L0!vGh@oP*o23mY07Jl7NC)3?13q9Ul$>dY3K7CQbf&cDaZmRsS&`BOeV
zFLCpI7w(N{zvyW4y}iAZaLt=MV0Hjq1);NY3+N=#9;}FvkPsXSfvB5QA`?qX5t%ZI
ziiF>T5pB{R=H|zId%P)Gp=k2<?a)i}IQv;~thHs%AmInGdS}x7M&b91>W~;-$jm=i
z1%D3q;9DKj&`O#y<H|1MT=mMTZ65PYJGdgPUAcoWVElV7(tE?0=#M^*lFffYM{vLU
z2`DwUM*gfkWrD&@I4;4hGZ>WK56#E;p@u8`(VJ^vEd`|jJSxX4O23&@a$1Yz)U^fS
zT43ta%nvLX{4!QKQf_<Q9X3qRf+p<f8YmOZD=gFlcY1WkI+MZ_dhvE~ZGY}Wg39~%
z?|=OCDdI#F9SQ!SJukcE{;8_Eri<&`JaJq1?Y#Ku{O@`1zbp77JrXLIk~GV#ERMJ|
zD|Iz*V5$;foM@C37eY;xJj&JO6AsBw(4V7^FvqHx-0CAnl?vF$7(X1g^#eQNK@^Dq
zB`(uAm($(~1}ICh4vQI@%T|U<@ZHr4^gAo;%t80)qnz0$mfz6O5V~U%V<-}8xUsj#
zHOmj9_1SAmTAh?nChBKvmmed`_yb<)US)GDlih|B+m$QnizJY3Mp}eiT(Cl)r~Vif
zeI?`e!xuO}%xQTpaz52)B`S3?`&mW}1L*+MM$yRc+Y5eX{~9MRSMl1d;nYAYN)qV#
zCLkan>=&+8@|cH*gqyqM*v}67%hT<FsrR~ZU|3iqz}FaXBZJdM+p;F`9vBbU*;@)a
zzLKze-#vdzO#`3;9G-rjn*So%34mvnGxn8fHfm)C=uE~)1dEPUn`08^BGdSI<2Txu
zuur@zjylYR1gCyreo^yQhFYjxx}r05RW~dW$^F7H)UV^#wvIZNhRRf4tv-EAP)Yyz
z@c_iSIOyeKwQFf(qX*W4h#olYz>*ql8iTC)Sy+70%?sZ&6GG$iuuSxZ!Z_IN*V!I1
zXaR|;ijE7BJE#P1-MH~x;n96+>W^?fK^QL<%BqzHS=^WQ<hR`^2q<WcfE#4KhrrLr
zM;El~l9625jo+rb7ba2a=p#m+(6!_MRb?D}t0CX6?daXf@3s?T`Ty-cJfMyGa&=^;
zX{A1o-n@J@{6QCX)V4O!H9ea5CBJL%`(+>DN=EbUR#Pw4LACNvgIJf4^-V^lj0Yf^
zKVCbBzQX99RZ!DWxzi<axHk@&N{$-}>bl^kg3+CbK`h2WiVO8pXc<mcqZSB6p#3~t
zrc^K*>(yb<$?rADo@oSb3knK?!(SU{Z2}F-yr4DKLl7`IPa{|LFmvHWQpK%|ND=DS
zhi^4JD)@2vUk1*iUwMBw=?$9s*;&u?_m>Zjj(+B8Z2V?WwZLtsd8*%+Axro#EqU?t
zIwx?gVoKZH7_}Rp4S(XIO3bR+2m^|a)!@E0<RHCEPQEyhM}^+f4<6hgBa1vf0_kvy
zOl&*UA>Tnda)3kV)5Ok~1_q+gBoOWa8MKcaq|_a~x#(a3#6ln=o|>8YsGUp4&Q2hF
zXg6DsoJ<4i4?QTwLKl0Ik&!X8u)G~EW<f#4H`Kn?_U)Z@DAyxzLkCjO(xH|2FdRbJ
z$_mxa4p`8j0&0%dE4TfwQY_MSJkzL+ge<Ory5e7DW#REAO*j+iSye=_5%T7up^uf{
z;gnPqWqL2c9t05c_qpc3+*4C8#AH&V0ttGV?7ILJ0O<(Z=E&bMvsVY9eK6GEZbOaJ
z1311O^@;o(YRvK|R)5j$u{KVv4-p?)Ne0PeGRHJC9K7hdN7Y@Eb)_OBC17QnVZ-^1
zq=JVdXaM7DxL}S!qxJ{9&?p-?=QIn|S)P5uc?FtbhkGdQO=c^nmshm`!L6`k^C4z=
z4Qh|FB{#x8VB0?0%Vvahh{udoT4nvpd!2h1Gfv#$8)cEq3u*l&Z>3f()R&&GU-^g8
z%vnVL#%v@vc2~LC>aC9dFpgAzXmbuh*-YX$f!83IgVGsKr~|o0MC2LmaT(5I2vt9U
z%@NwXPBG)PStIQ)s+srH_RubL0j3;8`7Spm(fd8IH%IZN%<bDhbMH|B)?fUFk(7vz
zVeipqzl=h96j?P9|8-M0`UU<Dw?oc{S4;jFvZVDKU_$LjeU8Z+j=4U_n{SahhGo+A
zTO2q)SQY*D%CN&Ad~<CWBCW(Z`ip`8)!h!Y?QnlRfPTAByL|)yYm@N?qqh+3YtTdb
zl5y*UHZ8699Ixng{WXzq_k8Epu=^(Kyb1u<fVje7X%R;3Mj4cRYp)LAxwA*ok&w^t
zq_83Ei=I^op;pJ=?K~zI`9?!KNsQKbZ3nV-u8jsOn(orhW|Smn*(p>m?I%Ey$|Bz)
zKbip`=4x*pE)B0HKnbPxlObjP1{OeQa0SxYpC8UsKzaQYy;XNRaJbNRS}lswW3BOF
zt8RqZzlw`n-_;cWq7iiO6tKTA{rhnHB!@q@9}f&!OxC;xa)#yIIUtB=nin9^ulCgw
zyo&R6IqVaW=1qSV(JHf{lXY&Bs4bn%OnGCF64}NauiOD1aKLfrPAs#!<_gb$;xnM4
z!uV?|pgmsHFdTXv!2_qckf3Zqb$Vb(hzOjZWiH3X>8VsO(jyH`eIX9(_!z>k#=o6@
zxi+XWQ|&)*x80GT@<cA|y!OW3t$}*3$OQVz{ZY*T1%>os7iJ(dy(s(k@|iTkrR$1c
zzFPfdx6Eb1@i?jRzsfHFPj((Se6RyuD^H1WZ$_{|q!YJ%%h-E1IL)C_J<JhZ(e^y%
zsKmH#J{RX-zyMMDHd8xuVhdgCGs6Px$Z#P3w)(SJp&bMJ=B~<OKlD`5U3qL70GYET
z{k(+7m=3X%AIPx{d(5vPrbC!nl8XMrXFsR<Lm60w>#1YAsfNaI)bM^EogyL4{5V7s
zP}&G@k3+`Q+H(=jZ*ffSs%%&=)Ol?aTNs_%f3WLFew|nilBb5DCIVR~rjyp&xR(>H
z`b>RGs<lvk4xa)oiofQ2KAcm9=yBR(?=hM_n^A^UW5DHBKh|yV56Xc6wnpL<diA02
zaQjE${I$lqy3f=>Rli?RQDF@&ei8EW@>T(;bA$Evx&+2!K|i-o?_Y@$6UmKOjN4^P
zDzKOIbTON%FljJn>XaeX-o8~2S5uZzd#g4V5E2ob9)2u7SH1ovj{b8#UTf<*{s+-p
zgkkSrZJd^~=K5K;AI7kCKYl>5W<9Qgg@VHFcQ$+ogl0DZ^b>KJ{^YotX<leCibbw^
znvR!swv#c9gXOViIxcOF<E&c5s8ngDmD!e}Q(_tz+4)Q<Vu#Vj&CCp>uPbJ&=PnT;
zrpF53?AFy6)DiH<y{lmo!yJryZ6|+Xipi_1M?m{oYBAWI?RfUnQ&ahu$KVoQu^t_{
zE1-J?f#_GF<Ja8&{Z(1gxzcqoU&@V~CnIfpo<?kPwmDs`jdZEAJXpx%(ze$2&t!@6
zu&|`K4Uw%4ydRk;t6mLbR1%e(!Aizu^lzqjxqEM8%&LWDDF2NQ^&*H?GZfPUsMDe1
z{|M@UsKt!?vO4SV1)0T$;5geWgI>!Q!!V=a8TT*lTA#4|92)~oQ+IcPY*Bl?P7t*N
z_Q#=76=mPBQu_*H4D_!MzwLz|UUj-RTu?$qMO9F8yt^D)b=spK^A-%_)r(ES*Pu>w
zr6JaS>Z)kTWyE1XSU?7nPTHu~FM6od9OWn6(CAqI;ro-PhmHngwE-~^hIZLIJuLy$
zZWQZ6(G&y}x!KrwOC^nhL=Fo-+O;YT?PSou);*whlFA;ExyUM{^_=9WZNFAP5@d(V
zy(&7r02rjt%*?Db6mS^+c>?*A9n9@_I+(Ea{;Y1578S$Y_EGVU#zqQKC;I<s0rF5n
zvfmv?M?~00tV@<Ydhw-8lxF;xcu9hTrR9fYjBI1{h~GddRN1VhQKwH4T@TdmsQ*Rn
zQfin-oHHC(MJKIYgnHXXT3&LRO*&K&dUAF(@nV2-(3Ai2<;xtOEU_-gloUZ0T($A@
zy12M_v#ayQL~R6@X~b8N#Kd6?cg<A+S0N8tR4%#bBM<gKnK;HiGn&|%2XDd@g6z*q
zUrV2*cetsgkhFv>H^m@#+6g>n-#p?evo5_RgO6X7t|1z^Pw~eya9F8Wy+x>>7kXlf
zb(#8-%`G3Wxt&T;p~BLAzJw=wU2g7Id|8jRwSy~fG=*6&y4vSQC0aDyC?|FLDlbI<
z`TUfxkj)0%+~Ci=rF{+4EK(=F_T?Y2oz4LmX?di~`cY)tHC$Ylos&-p**k)!LkZY)
zRhghO4Tf5<wl&gJ@7?w9k)2cxmg%9owuY<oLscu}F8O-tl=$R*36{AT&M(hy5B>CA
z&CxLHn>)oWbnE3PB0I5-H~JMjmK^rXn2?k(jca#Fdt23~H)CG#Ofi=AR2a)6(x`|$
z>C@;ZrUzNt?Uzlt>Z8Y!wNTPso7vX86Tj>Vmj^6;DRw^NU_FaJf}MmwJoc)apJ?kB
z9vxAP7s6CZm-n=|{UmRw^4$ICTWlIUFlU8vlw)<i@T9M!CefOBK&TrJXZ5z=()OGH
zyo<O7{@otrG%Y;~a`b9+9T9F>LF9*%Q(L`mZf|69S<U=#7O`pgae`jK+vn#_Z)_^|
zN>=-;k3`T<tL^%x*Ytp*-1g7c%sMTxM|TJ{%f9g9geUUFoD0*gBdU<=PE{onUYGX&
zK|iSls~2`9)JpYc=2mAwz+UvC1H=78c~n4xt9d1jwa6#+eK#E`r>c-#2RWBZOc}?Y
zd#|MrX-4j?_J@Ugl70C?CPSj5Q?<G9ETAn#iDF+#ekA=Q)}-L=@O~YU%gbORFR$e$
zz6zb~jW@%TyL*-Evy-#AYgBz{LaQfIWjDssD$1)1<wfw>e9yL+OovJY&yknV&lS}^
ziLtI>UX=`>lJUvGrm!O(g0ScKx~rn3W>iMeLMm2tEZd*5HCl>M0UQ4jDS;!yO?mQf
zJQ~ln5(*tM4DgX7Wq|<%3A^1Nxp?jL>TEljUPyEek$f{+_ueO(Sk9zq@GYA==^)Nj
zq7sO`FJ^SSaQj-8D(VR`1jgSKOES1kzq^r5zgfS4Xu%o%MDE*<-*{so=*<>MNFs_2
zzd%#u-u!Xu^c&k>kFZpfJy_3EcfR7B+BJ1iF5F7?hlOjvhSy!g8XX*PBQch4G@~~b
zSGKj5GhdzE>TV>Vo4S)qWiDo-@xnriIU(__1fsy-nbP(LoVj^c$2voHTS3jjBS}jv
z>nw_~MMKDxnjelsrFOBfl35jT(a-Z_lPN&B?;7K2>>)FfKPTog0pjkf>FQH&8A-+Z
zBfCbFrms+cy6y7IRZU)PA8ymWVHq6#r;IZ`O`}?|=i}SuWv)oy!zD8EaW0vLlD0sv
z)L7?d8I7Yei9sLl(A3Aa3#95CYjDn{-_N*n%X#z&wi~5b#^_p}(CL;?0KC0Y!RixX
z);sAmkzASJH|+V};jb-?Zx1kX0Xn#t{UwbE(ud>WJGGM0pZXl$vp1Pc{}zV$#qea6
zwtTsEb;IttQz-_(Qpmol%p<!}DV<!^bQ&^l<GL}Ia(jK9y@$PZjA;gueU7Epbey?L
z?BhixcOTP>WvHIxRXAq##AQ<6{@MdudG){FF7OVsQq=EmnNf^@Pklg@%f`ghve~?;
zl3pVv6O*hlhxzE0hH{%cHhT>B%I7+t1E~)^6^klD%q@ttc%4)qT3TMIq<<i2&W5`>
zs_@KdKd4P#!TEJUKzEm-zyn#@{huxxe$_YVJZhz?pZ3QzPnK7CD*98`|EeEU!bVSu
ztn<n>`{n-JSY(xu8Jfe1a|C5>Q7Sp2QNdAUPBYbP&4EoZSm(E!Q5gH99BXO|Q!J0C
zwP$D7_Da@oW6piWAsPAl=*~){`gq$xhcRDtKNtFAOTBO9HL3%=QUsw>Z&vvEupCC*
zzd{}!)M8DyI|j6vGDSSyj4rh+-+%l<S)yD%kaa+b?zo8jr!g7&$2C2*sO$l%w(KRu
zh5J9G*C?18U}6Q4#E1RXxL6+0oQ3e#ElM{&QC5taU5*TGsQ{b3KfJxG8#eLFVS&>`
z6h)$ME?*)feg4m<fOzl%MK90xd+MnfZO4S=?Spm8)MB_9LxRCq(Z4OuegG+4hl1+0
z755VGq1p?=MtCibREA5gt-<AB5Uy}bZ;MqSjhBw{cB6h#nrNi575kmLBQ&y)?TX)K
z9~$>94)pQBJT?@!`%~|hr~4+VU&IxJ54~hMEw*OFqIs-E4B7|W$T$#>y-_74M|Os6
z*3C4lYI|n!Myjn0LwP$=VqLnU<-GaG8rCZFGO_y@7`!(Rt60ISQI_*HUQS=Q3zHHj
zhhho0`O*a5#f>Rk4>zW&s_GsEf?iD9ZRTnC$Y{}Ve+iMbg>9{L@nGR(BwaA!e#x3i
zx4J2rwhQ*<3Ffo0G1Bb`gjiR^B7S33fNb2h0$Xf%IU7YVwXYM$_BeVxWJy)Cl}(2V
z#Q?3!IND#&g=8cRJWd?J!-EqD?E3n8o(ka_v-+*8tTZ&jP~NNJ=vYq8KuMYKnDsh>
zNAWstFZQ$8u5JqFq;bKuL}?i|S=FlHT&I}ci1=)4ot{pq@iZZeA7{rW1@(VeWFN@v
zmn)#me0sJTSJfXDRq*1T%8g;f_aoUD)`FC!m$QGBDlGM%DI*8cbsySXhbJUtiEK-O
z49F0$7hanM*_;Nz#kFVy#bl7XL4b=SS^<J-9qO+8*!hZiNS50RuzCpk_U#))bqU~3
zk_G@AkKgfx2R?^U54|rbr#O%f!o7Hi7XdK~ol;cC<4o3D!l!$#DyGrr0bRG)W*kn+
zsK|C&lHLuWmZ&ffkBvJ^Rg@l5m)_yS+G-dz>{;og>Zw$gisma9lNdA1p&~<tv>7$+
z+Go;eVJo|3(3ls>j*Rp5f-|VgVb}FX4@x-TEi^Ud<pb8L_O$^Yae>$P-Yi@wm+3UF
zw13mM$L<7U*OgPAJSHkNqJ*C@b7XhYiy{N?L3NaCRe=VZQu(`8#=xeeg@pyOa<3^_
zYASmW-S;L-+=k$Et-OJjj&2zWUqYy%#9Ib1N@nj6#J|HIYrkdaCwr6bb*2tgID6xr
zY<^+vtI(P?9-&-+L=@y7lkjVsj<Coed@6D8MF6R&V*iJJZMo;Fm~?jdGc1iAGMVE}
z8ZuJRDbQl8ubkvEQJF;$BllW*jrZ>CB3V|?uWfI^DveaTQ}mYsN%!sE_4gRW=ow+b
z>y}m~B~jb149Gz=8mKd9R6aW?qm=G?dr{{yiBLwd?5^^D`YbY#5I)k+<U}1BCqSP#
zpPocfd^%g5a-jMQ#@pPe_GRyS&jgNuQ&fK4D8R%<D+75;19^#A%F9X(W3I>NX5WC0
z<HTTdfBi6RnS5y|-;=gACUG!Rd3jw+5dF7~t!DOyIH{$MAo|7~@fD%7xkmW)TWTtp
zv#ZJMETmYDJEwA1yj%WL&X&X5zA%@P!=hn0l8B%FW>sB!fe>EBf*IByM0m#VDvY-~
zMWSPm#k8TM{atN^BqqdX!OhC1@om<h7SE2?`=POlB9tP%ULySwTX%U3pf&{nN<yfM
z%*V9>s}72cjQlF1n}cBvh*d!ec*A9f>bMSevS2*PVf=JudHU5e&e_HL>LpI!_F^8V
zv(K|N4tFPQSk3n2SU&5$if{KW@=1K5agQw1xP!zzt9B0s-@C=0GSzyW>a6dNFfxcq
zGyJ5;m<Gnj2Tjp|3f^|NjNn_aecQU8lhEG5Qf5}_l<Zg{m^$NQP`KPL<_KPXMX-M;
zS9X5?FzRIlW2-uggS(aN`mK#)F^}GOkBsEFTm=qwWfkOckITssrK}KD+nS9@%Iq)2
z_fwSXP_0#odX$enE%lL)cYYZMhh6UE{%B&!J+idWqupNQ>d>t<c$7HByg?IEcKxr^
z>1~%Onp;rL?~OSsj&U)Aa?{7%&b8c5HoX!xDpO$DgiF$5Ub&)a?VABmZnNH_#k_`l
zEw<+X{_TLx3q2*J5CDosfQSM4nF?)X`HlNBRb=2mfMR1_hpnU-r!^k%Zfb;|1XGbv
zYBgqJFgKKe`$@sV;s;>gy&%2zz4K7$cz?a4Xl-{61&!x&fgMW$tBmI1VGZz)B!&L)
z;7Ss_yvY?Q1(2muM+}na?%gaP71E6RbJTyle467SzMc4hb9>vFh}-MiMAf5cb8w5;
zd#CMFT&nYMTn8?6NM9y?wvvzxRj#}HHT<IIx{tn0GvuqI%yM*Os<pN8D*PajYfo$_
zJU+N8tW)|gTcDVdP|Nb%(Jm1w;lht)m_r1@`!~FMq0_z@rYE;@Fm2U!rMk9-_}@qo
z<5QJyZ}(OP(R;$9qBSx%H+QXknUzKL#}g><QUvzZ8fmuFlP)eJ(+~b&R{UlKgFE2f
zVhmV0&%JSY>%FznOfbQdAmcJ=v^4?{5GOe`)fhe!2EfqIh1f?J{#P>kE)tq$eb*fa
z47CWJqw09AqTFKg!v_y*))z7TCnd_6N^F%I9{^1cnTC+~+j5|cKqp<Sb3$0y9nVjn
z^w-%V+m^sC3*3*Qr=x(4H$L2&H5$xM?#oi60^_30oY2speU)R|vp9J87rrG`tBZ?j
zan{{kU3W)t5h^wuyy^t{9lhU{xtFZOo0X2MBx!ApdxdfZ>`@UtpXqmOS!rqCHh-Yl
z+4T<g=5?FQIEBcCyEiivy%@E*i<-7?D_soc{LyTv4z-w4<bIja)1nV)BZVb6SS2ft
z^^1cBvAw{*?rb(t%#&_tm)nS;t5SiI2gEb;?lHVv3WGvQn(OJ&5sYCk5RVRyj!TeV
zM{t|R0QHkFY*rx!S7)0g_h~R_Co<k&#x|Pz^#yXmwlkbclfg%(13#X^5#0~JMG>4a
z$|@>aOt_YTc0+&Q^&N-9XuGjaz*6-dh3^1pypU6|7Jx#bz;Mt8EFjy$9!_VFV4y!4
z%%3uaJg=!28v4WE_dROkV0)&44UG5KuRWLX`QrdHo~^)lGM?LY=J9}osHWyg{jXmy
znkBOfL>A^al{^Mb+Q)LhtpdG7g|fIJX`Q)MV1SiA)l()`r2Zt^m9?<fbvOzQ<~dAY
zjTua3cM3J`{3NAb_$n}FY4Djs%st9Lr`oah5R;*-hOy%(o1ttHSRUx4V2jf%#4h}S
zTGvH-p0LN-pW<@dYqw%Q+E?nYe{S%R@d_qUrwTrLHt4fJPbMA!C+I5`ayC@{*7fv^
zg@tKxF){y=k`k(+mQe8aYi?<g03xR!=%tzE<w)IgV5w-tz?{X|FN)>iLpiqUg9ITV
zAwm+8@Wgp{cLZ|t7eT%%vm>B6yKviTaUO9?%(q8#f?;Fkhi2$PD;6Mdx}yMN$aZWg
zBO?P}$bjcg`+k4e_wR8)Ye@t^&#Vlnz}H*f-q?@@yBrS7sb6>JqmWL@I?kKI@7N|)
zBvS{H#0Z2%L^8k*MZMBK&+jfzLIkI=b_u>{PwVob6_!^=t<!cJw}JV*L9-{5=n1?@
zU*{^q;wEHh<!(QzeIq$F9yw;Ht>=}WpIVnsAIA1h#e6Xun5E%F9ZT6d>c_9el4f%4
zyZy_Cy5Um3bK92WP&|jqWqWJv14R#vC;HHba3RBq4#bh?Ci2T?3O#AV7M!cQ5oY8Z
z3j0;ZC1!5#@-Eday=}`GHQHYB8M0eayeWVVsu1o!AKR@(Q4<-hjk;}LqmND0w*-@?
ze)bN2DhBJG6%=}l2P8q6B1gkC`fScp&J^*x8*cEgO9yiQML-C%TEP5ngCBXMu__5M
zF|C+GyRXI#HJ|=b$YFSXjv6;b^uL4ddORNnm}_fmK_eALJ?k`6L}m296$uQwJ`Bz4
z$qs>IW+x~3UOzKzcb_`~rxGlhB~M-7WTwx%ab><;@1LC{|Gu0DNq8^HCCm|3taV>0
z;_YcnmE+86Q~r=ie^vVl=Q79EHi;FtL*{h%c;@It?u<&YdQJl=j#)-mVTmK^D3wxu
zJQwgn+kRv7g`deSd(P?TQ9xi{|EKZfBrL|UX-$2RA=IjeSz65}^^Iql8u$Ywf`BLu
z3?@~4Y1>F&@j>_)E}5cvuIE_@Cu__05Ku|YyPNPL&2@Fp;5AJ4F6n{J#x9<iYq;8_
zBIz8n{m`F?8T=ZM9lF4;GK(F7r;ty!zII0_`(yrz)$Y`9m1k-vhTxtl{^CVrr7Wug
z2Z$aXI&ICz#r-?#dc2Arlj{ez#z_L|CxQGf9=)r{<3xmG^g8p0SSoV;=Nk9Vt!^H@
z?Iw09-Wfur`}p~EPb&y~$HB6-w~7HEA~bjTj?|dZ`LB|Ut{qnLZ+@LWzMV(rZeP57
zJvwv_NcvddwUVLR^g?0EGjMBjQ|{%<O}28aTy3PMr{`N>k$6OIbin-?zyU5cne9yx
zjJ-xh`fMh9>w;=NZd%4S9|Sf9El<NN^SYjt-g7;8#iCQKxj*KV3lTUan0@-*LzX|G
ziY{2&Ls?-IEtgr`w{D99xxt)+WZxxo$SBx~BOH)A?GqdB{r&wa;J_A4UEk1<38dfA
z(Q>XeEON_f5#;`eHBX9O7@wWl``d3WPaMIaCDL&q9$?c#72AZ;tR<G(-nlO4MaMpF
zJ$CTU@0No6HvhYm66>I-h>W8We1%!ss%tjnkQ=X3MAEFgZ$*?iW*1Wk$j5E}O|Vh{
zX9`H_u5c~Rlm;NA`2k%f=nkaIPRCw8Fg)0p3`l|`<VAo9m&V@bu@RAKtkVMJR<NIY
zQ!towI6u=A*kAch8;IK^pz;|YRdPGs4HY;$(Cy9EPzj`!k_Tr*2#E6_{s7TIz!rD5
zaY$AiCsR;c8(0Nd_ptdz(7bWPO8@7QwpjUcPDOsI_G7oRN+2o2s3zdh$`&|UNL)d=
zogL0E!2u{&cq#7VaG@P)@2+5cey@6l`u53qf@Ly%nXc~R>Rn|o=fh>zx*IlA$kC?0
zSpgW7QJ1x3e9rVFdE9TiYD%$)v+Z2ZjEi@-s09CDrpJ#}<(yo*>)9P-o>?FP|6ya;
z#<MkOiukz$_mV&L>NrXVoW+emYLP=Ftx?#hsmf}6vp8X9b~YI<$E9$Cm`?D|MYHBj
zuk#E67CYoc0m~ea1TQ#$9N|m)W^6W+)6Lur*mW8)kbs4E8}TamU#Se>04)vV0OI(_
z9oZOkn;;e740{1u)mAh`&`-J!RSE|vPN#TL^SK@?P0zLNM<hp<%$2+K>hJ}>=3=Up
zvjb_D%k|&8&kOJUOM-Lc(84vi^>CgIz0BhlAfQBtdt8x98gl|{k(}%4|7=+qHQW+j
zRZ9U!*~MCH3g!>2+jPv#k}T?lf50!m7N|Iv-CmV|BnZ!W`lk!9q}n4mCE%f0a=2(t
z&1ZJvpB+Y$PW$n#eWyMJj(!N`=2i`RqstZ|8eWihC>;~rx&@TBqFA_32~TDUJCRxq
z)0=j;Ehw?2XJL_oZrc4vyURuTZKULnEuMhLOTE-09{O{=<KuGgdehL^I{cWiXYr*<
zbk?@ry|&Zh8=_hbEvU*5j9judno;dql0Pzu(XUryiH>(c#-U-bm)%pm?j?bFCy2n!
zF|^OspT}aMYJbf7h$BI*n*%I+)W9?*FX`Mf@tpY1oiq>+hy_G&yiGXVo?qUYJ!ip-
ziHOJmX22NC<%Yr<v~v3&z500t468MOa#j`*`IOXp=qeM<n+(;}ngt~zjxN)_+|*)l
zKXw6&5*!u9%wX+00axh~x{mmv=QA{3f#?F92|{MAeD&=7K@(p4jX1ONMOp|mGvVK6
zXjQzcKBD)IXrIx(d;4LyWCob5Lb#=TH}vv^(Cevb8Zm*V!Eiinu~Om72nsjU&zGN@
z^x0lot?;8hyyD3uVD#<v)D$79;z{7z1q_ojH$9yQOko;$v<_8Fxg=}yW=2vRJ*QoN
z<PfV@u=s0_qR@5jNpdPJ+Ldq#q@$ah(Oy|{iNJS;QX@|`>bCaw+P4pYK&I#7QeLmx
zj{#UDcHR<Hvh)^gpf>0Ud*~ozs|tEAe$m_WbHOUwogn%_2W((8H*#CB<C<TAGi*~{
z=Yt=`CW8qONNCzRgResVJac<)LfRrTSlN;h5hX+5VT>k(svL{X&Y&%42)69CH!{sH
z3x-r0^#gnL{Wb1&^Xo~unicHx%I{?hn3)MX5?mnoaqbY+C&Dl>@bhan6g#JSCK?;g
zcMt^Kuzt{BxtJbmJJsQNWNSrtx;gGDE+Ykxtjhs&NoN}q7#pkOx3LYIwQOz^Vq}=E
zKJd2Va6YgAP3;SjL^&|eM2F^Z-KPV*kb$<@Tzdn?%If7d=~%d=NlU$%EruM74<5*Z
z5eY2!f<n7>?Zcy^xr!J_Zw1aztRXBif?VhtK7MGYp<*zj(hh79bpDJEg)RL;LqkC{
zmqy`w@KmeZ#vig4;Qo5zpJAhy8njvrf=7H`opUi#B()W~4S2YB9z21OuQbwm{`})y
zYj`58DZVLrgn0Om?=R8(keH=g3*?mX%TRjP^jca@^>9OTU*(^6O(hP7_A?#+H@%G+
z7w{iF4@B!Zjg2pGDFigfU5~Si&mW5nyWOS2Uq?7#E6R#2y@3@IiA12&zX*PZe6;uU
zOuO|FWcSydrD3h_wDYa&vHc050E_+VIaf9|HgwE`8RNrc?4Ms*XxJ%)&Yta%;4!od
z1>sMwPiW<O{p`6&Zr<SwYUI84gg7z8xRgl@;wJ?tDS)4$bYjDdBQ)CW(Bv}MJ2yk`
zdX2tx+qUKB!w%6&2i@YIdEzo=TQW_axN8-mw>?6bih~1^h_)`s$vDATmeXuRbxpiK
zxxg#Pia{hkezq5mq_%K&Q7GJCv7BU;y#bgdx|02t8R(?Z=OH}iACMK#tgpvb<+`1p
ziAzcr&RZeG<F&a<{He(Y4{YO=dD^V(>m<eGJ0h)68gak@Z|&~^I@uCT=vMxxpQqeX
zdCwNdK@$7o8lfj4<uHsxr;{*=ihy^4ph=G_l8Br<8-&_M5aL<cv4zlNxKrw?)Kq}y
zJPi$vSisM^I@`Hc5;R>3KmMIoIw}p6xj_B%M6{c5-8M4haqr3&&W<Qg*VfYFFOc!#
zc2-Iw?Zj4QFzR@u#Gs>HWnRtDugj0*-_h&1m#-JL+n=Cv-8Q<XWxmj9kPpDy9h>!-
z|J`(WkSjrsa7inO7sdlD!&}E<c&t;<1V3Xu3gzZk#-aIUmIqPa0JnbEG(gxIb2tis
z=UumPwQ>&Xb3TisUTk}h-_l)2)uAi8v!42|n;ZGqhSK{wJpcgRoE>9Z8Lm2JIjQt!
zM;&XTF6IhR-<0FN4;f4ASoPstq%|@Hw(j7tzrX*UCf*YUTjM$`$5Vgo;8)Up=Ige~
zt?@;Nylil?xjIxt%insvV;I`*7^mQ6e2-#fJ$GH?+xN)YI9wgUoXc*`l!FieORLLA
zDLemp_7J;2&GNael<U@}^X>)WkC5#F$RC;~j`N3rs;}(c^x`4|cS&nY6rWvtObZ5m
zy|%+k_n5^|h4e4;MJ85srYCa@G}@2Dm-B$6+@1f~sm>c;#2PD@Q(;yn7K=j`k;`Mb
zZrCZQQ+<*!GKw?A;gl94vzjt6tznJp$TCMruF0q6qivkVeyp6K9n7fCvBr7m=rDM?
zd%>F_9)HjZ^c%?e=XnlM)*W9FwQOZ4R?N5{YnY_TKdC)wj-0*+&Y$pd4kfYsom0Lu
zwh<DJscN6i>Ph?c*RvlJIhhogaAm?=bIkdB?ZjdqwPYi%$Dpx2N=C&$iR;y=1Wut$
zM(b8UoG2ZVmNWuJ2brkN;mOr#WT6Pj9HE%}Io|5-_<XF=lY6MvzfWa>r5}tXo50_h
z7<p7!C9FVv66XLo|4cf{X0k_z$L}icHO1KC5<{Rfm6DyygtSsk>yC!@XAO={Lf;%Y
z9*Ua^I?rU>tOG=6sdtYU7qxz`C`Tjpj}F1b^gofiKY=Mb6~saihla(jdQkMnr_o1$
z<w+g*cm?zLIMb>uKSL+m=)K6wT>Z>iPJES<k>Pp_ul0RtCz+@};RD-SdpnA=f3}RE
zVNRta?J?@#i_dkZociIu5BE-MjPd?>S9kZeZ4@<J_)dOCSKWQs@8Dz*+{@veoD_D=
z_P$@sNH&rBbq7Z*eA|o6_ctYq_3vjO?-Fl-Ftzu7!_AGqqk(vGHiHbh2rhgqS2+Kt
z1%PjtO&H|-z|F<gC+BkU!|;!5jXCZmlHrg1mM(5z^fx_`?EK$<W%YKy{omJ`gcuj$
z82sN-`b*mX{aXvUE0zC!Ig#*>{lzf?F^%y5^DOQ@P_c6vVL=7(p|{ivSxL=GNJSYx
zUdaToaRdj;pacy!1HOI(KAzCSB~^N{JJz*RCbD>8ePH13F@Oz=?UCwemN(NOJ3@Nt
z#>KanQo<w93xz7k?{uZcE@zG_HyRa)1zck43k9+b<PAVEDogg<b2(f#KicXaw;4BG
z<yajwAiOoPKGS~jnC?vfPGGuIT5XN+?7G@U0ba`h;sjD#igYxFAN%MPkDo-<-ArAL
zdBgq=0uHw}UQ<3BRf#cst=$a8ck=TW&rzfFqmdLr`2J%O^)4`KLrzmkKbPM1AkE2@
zr`Hm5uG!HvwV<Mw#swtzB_C%oO09QoF5Iy_?=a>ibL+es{dde&gleuMgeRKPyKOes
zyEA}xEU!F>tUPJf-oFdinl0&(2Ri-j&dYPOnBvPnWzYWmb5G1*#A|A~R)=#jjr&{Z
z(gJw-c~s0qzB$JlKPX-7d#{rr=~=E`WgY4|mY@9(ALhj)=lH>#-Zk0UzfL38=J+kR
ziS9?uJQ*aJ4js?Mx+=!mEDX&rzA?@QmOTPnT#Vh=Vdh;~q%%5#zowOH)jb9T+tbLs
zH5C(~5108aA{s>URhYoXf7*2$f?jiF3Ps~Ra<nHZ_p6;q5yvfu$$8sy=7G2JclLs4
z*V@f$l)~HeY3S(qDtv{kC1mX0U8ngZ74D;|_UMsbfqL`^Zd8-e;rRJEj*oDkQ02Wd
z0VnBz86DQ2Ypy5F%AwTJe!fb)=lVX;yRn#S_G{)QeN4o>h3Z-2w{yM7)cW>4!E%*E
z&%%n^B8BPp)7j%I)J;Zt^wIT}5=Di}&S$8tsu~lnme+ZNwi1US{{FR7ZBH0PYcyT}
zyCg0pW&4iIbI#v?a)ns@u1_!Hbquz%S9z+6VH8aPXAU|)40g<j?eNXT<X?+>cO7ys
zbMWsszAhTno83i-k&w{LKjVw~wodT9?YF@fvL{Ej0_Qp;aS?6$Bsetlx8D={c)H)F
zpZ&-FdQ(G#DCFJkW{H28$Hv;t;(z^UN(q`FgoLMz%(vYPTUt+=n~4g!{VXE2iCXu*
zHh%E47Ks4)TM4UcbceB|N<z<|vSaXdXh;aH*cQC76`chw5YwmX7GNqo^UBIf)FK7O
zZ?2HT!O7|2hkdNjfQSh4kPReh69dE6#>RIqK8jok_Gx%RXv%9aKD#_Wuae;UX=i6A
zD3Xlb=29}r`A>|R_e+X+Fyegqa{1dOGxL5XLDapcskDE001EcH`(@8imdq4R<Atrm
zX<oNS`G0I#8}4-<5)3PrX{vE%Xt%vO{k!S9;N4;My$o#+NLuyLX0q``5u;ee91A7p
zZ*jW)J#6UJF4i@g-cx$^j0hMmP^+20?+~mujqx)(Q?A9paNLX$5dL%j;tA+kF1{Td
zK<s`#53J}+Ya2(V0<G7+UzU2Np0vx3-2}nFZwa=vues(o-(9=-DNS<pf%9I2jO)+N
zZLg-61xBM^->VB+e-D_FnRjqrnpHK8VV6X?jfldfQ^hy)aJMqu_Tf96zd!6>Q9uS7
zxFvr+GZj^fANRrz{P4^V#L=Pr565VwJXZt2cS7`Jx9aKZ?`meJAyEu<QOp!<Y;Dem
z!lT_e*#*OY?{`*r7!U71<SY@36RnGmc4wDdj=I@+tN6A%P!5rjq+J!M=o=oSanXMm
zE`GO#FdVtEzdoFva&G25B|UC@re0zE?Tp{~dFO3b*k}yQ)T&aH8JP;*DO-CSJQaUo
z^NH6mNO58Ij`j<F#BDaBKKkuuf6s%UTg>@?&tL@N|BLgU5)O^Oe{KuIx%T(Z|NY?q
zI~4zSRiH<}-RJ)&8>Eb{0z<Uiely|S+DJN@2nV`V#o4(MxJPh2CrwAAKPtH~1sZ?v
ze+0sdffHjrSt4|&k?Nb~u08OlfD<M|rUWvvK2oG(AqYYVc>nj^4UC)^xwS840afS)
zx7gk^xo2R)Yh`U6IB24*r<Vl%%(4K!Fap<9$BkO-=+VFNE&?%--P2mTyuO|b#_SpJ
z$=9x4_3tzcs(txAH8pjp*fbMbD7*-m16bOperm?-e<7)BGVMfF4B(vN00BpWb>7O1
z*?%jt#+6&L#<)nmpacoM?FKGz2KY~pMbg7PposnM<A24_P)JG1KIb>i$TrBcTWw?8
z^jpr>gX8lu^P1)$O2En7bvu`F&^9wEW;)v+{)Gx8)7m>m=6a9~1)uMma%tmpmuKx`
zX7<vSkm-kA$?@99d=YTKEVrb!zKvS-#wVa*VJpq&MkIIR-rRstK?=~m01l`dS-yt?
z^t=j8yZiZR85qPYEiLgDJ4pBg6{1Fkz66?aF^#UQu*%AZb$3OLUK0OIIL5Y#FCp?t
z{?VhTQ7m!y!X8C~68BplAsk792mx?Az6xBz_`*xhXC1~hVFgLa$$7*#Wt)SK^00vh
zHm}+ZnVOzPlWPEc_YboCH-7HVf|hPp0-27+n*eR@EjHCao*hgd^5#PJU^G_g5OWOu
z+*lJk>7nJcXp#ELNNMf^?95(9kH|=zcC%lYdaBGIQH|?c8%zXU#DcuIcQZCthKMSr
zt!by`%~wQGqp#!DOp8|>NWG_aEqHi@na+peS4I68X$7?5!H7P1^prZ<+P>|SeCq0o
zS53L{>U6$nLVqH9QY6&HC%Lv}c@oPY5JJa8o@+e+hpMlRifaA-CIqBax<ynvq&q}F
zItQdnYKWn`Q>CQ4JBRKDLAtvcy1U~&-0!`=wO-bme^@Z*oaa2TKe6}rG+w!Y6O>*c
zEWNr}HYor8G@!>$EIr`{+)NV2JqBVlIEQnkde$Nl@JImZUC?Ka6VM25JVgQbcmj~r
zlxngF%-&wWO#mbh*nIE*)3>&<@%)~XETkmmB7%Z~05gh7t+GK*;^6I(eCzgs%krN4
z=g*}*7*FnjK$?CQsvo&xPp<5*deOUt@xi{Y!%Yu!>#LMmemdBSTg;~S`#iVc8?+pg
zbx-r?(Lq_jo;*DxL2kSAyNO-$&Lj~Ff3vAvogsFA+PAS$@gsIr6rV4f(KiOxyqw<g
zer4p$WD*T2F!(Y#R_9*Jqf=T=G6G03;9rnF=rRAws<U<GWougj<n{pstf1hBU6Zd-
zy(*aqCU9YDbghU+0gb?ODgxrrGxfAI0LMqlOO_*OxIEF`Dld&J3EZwFLXMP9WVqrl
z`8X_U$vs%rMVrXtr&B!4UDqH%*Fz$ghU2eJ64&|<$e8X<zknuR?;%+eEyWWI!gN~+
zH(IG&(v3{L6_h&5B^o71)1}kYoq=G5X};t|-hWn|HGB>e;JH|^1cqD3WdkoLv;)!V
z%X3!15Y<zbf7km0Gw*YIj)I2aNt7Dw2xPSg)wDQ4zMO>Cl9ADE(+Ad7OqT?!f!B~U
z2K7RH3a`YQxEB2}#F%VvttH1J8fpQ{h|M&`+XXgPgFX~qkH;n*kK@NROTvFHUYD%I
zQ9Nl}xdVXhae&O50HomN%Q<uZMsKpAZrzDRXln$~;U#9&$H{(yg+YVg6XnI{ilVd~
z66Q_^qkvT|+5;UH`W-+;&Bwhatj=|k(COIJ!bG>aq(410n8ggT#q@r)iZJ8BsisNu
zV5t5-S6zvel5I=EI|RDCsscW?F~B!~>IBc%d_PM7I36h3syr$R*JFookoyy}PaJae
zqFU7Jr2wlP!<@MEvkWA_X;cV^W){riUoHKB*IUV2@sOAM{del)G1VXK7A^bG|Ka~Q
zHE^!!k6*vW=Bg9276C6v7GQbUaM=eEuk-t2=n2%!y`|^u?)^9K%Qfg=lmfciNIKHF
z#geWyiRayv&8#qU2gu}UpnCm9tI$L7HoDCiW4qE;1rqbf$FIApI&q!?c}`o*BV9en
zJS~#{JedMrbmHqKH9#H!bbFV_`qvNw6tvy><)^<3+XS&Z9>V%nn6dBNpG`6cyp9Vt
zgE7{Xs`sh&CtsacgZXiph_WGSkCgOB>i>M(TkIU;7eEu!IzXge+voIFg2ge;vvSSh
z>ACuNvC-I_R2%kcG#C{Dz7gxMkPoxHc3V|fU6Qm5(!c7u-TzE0551o}Q#)_rKHzti
z83a#RYeoI@z3CG)F;vd*Gd<wS9R(GY%}hPK=6!nAY57R|>vJL$d!TiQAX!ZP$?&vg
zoEMn1?|+0d+%Yo{Z<oID;!z~R55`^B_bOi*ApUf6aye}aZvtaBwtA7U&wM#-M?gkk
z;>GyycmSBF?pJu8$)<SW<f8#?0*8Au{{eIZ!M&noNG15;j}oi-wTs)l?*HQBXsPE*
ztR0vs{Z31>Z)eDvB{~()lZvDZ^yf~mjD#vpO%tUUdzE{ERWQm8$-n>aB89+3{(>6C
zF>^4Is?H)k0&0xq)JIxuL{aQPryK>ZDRm*tX44<;#Dz62$I=Ll?Cr(}FZOs!MgINw
zSHQndYS%7A*O$M~Jr-LV%fYl^b#cy@4&DqiEA~G-KUN4=EqPxFx;@ZZMK1}Bdc!(V
zs@OR&I`PjJz|{l%$?u{^vbiGL8`}gb>soCFtq5j$Ni$V}@jbbilM4q6zYjA%9v@h_
zj3p#dcaWHWTSOSAM!pXvwuU0|{g1=Q!35lYS}qAPu6HNxk?%XZsK$0CeDS`KQkDDw
z&9cH1bF8$y+b&ywrq0OgVU9Zn4eW;so2E<mePxmmiiphdi7W?tz|dB_P5eLH!4&RI
zAe?&GSdtQWvBRio5QxFNAMq$v_b7><#C`_Zf#KLXae-IO4OMDngh#nFJ7nB0&7Ozr
zpq`<*N}d<e@(A4QnAfFr(i$ksAijn8#7k@$NT~sStj(TZ)J_UD`S{3^fKg*KW#|HP
zv(48z4_dar%}NG=Rqs??4f5Ejr979GZdWfpf#=Iw?1-IMb(xtT`1O;4bAgIXCK~zR
zw{dc$T%MMPwY|osJy!VPZ$~X9KQO4;V$B&K6bZaw|6Hyurw$0h;65Y)m=-%wFh|wF
z+<bz74X;vAPJ@E_hByZe*r7k!(fP6!%w|A<nQ+ltmQkpdQxVnZREj1dqH%HfrTKD_
zE3xL9I%imbYEXbGhC^~gCbIoEPsNp1;`?5}FxhjsUs{_+$lwKm0LIaOuB!zPycBnr
z?Mu1c@k@M<YWEy`$U?S5;}y+=snX@+3Y^#CEKucTdqOcKi7TdE&)T5&Lr`3Awo@Q~
zZids<7(pw;c%-Z*uqt~=l{qKSpxHW>8hhpB+CUQ6!q3J3ZBuHZZRl^8-J6@(tDNMY
z4e|IzpmJ2yhUn<xk#u3C>+4s$I---O8yA~x66e4tIySX*yN+AEgfjpMrGI{WegE>G
zoBb!CR-h5*;sU)yQv!0lOlsF`0Wv^^k$qy;@_&D#^bgAR0la#;5@EY}vYNOq*JYWo
zHKS?`JbcpPT#k1WXSq{gWoYRFA(BUobZ_tNO3)vTI;-YPCM^Y9a3XNnBVFI-gAV3g
zOXVRB7GpEBma$*AaF}+s7&WVH@32z%o#Y6Zsp+`ebSFQHPUaZdIY1VbfJkM?>K_tW
z59OJ#G$6U$f6n_bJ$pn{JD7Lauk#GPSB4DV>Bxe=n69n(?2c8`<1{e10H#CN(+BJe
zAg`n1TM5r)PEH;@EeA-N({d_OHVIU(#p3~ZLU$*senvGgs;}YTaF`9b%WB?74eacw
z9q6{2KwBO>V%(dW&b_V{O5zTig909&hvFY*R`~U^m-A^+3$QTpo%zr$0lT|WC94vT
zeDB*GU-dZ`$=in}M{=HAa%fX7!R=CsWH@$Y<LLIpW6I9@tR47l0i8F^n)b{`)ko=0
zub{LQcbzgvBYT_OY+lSvtoRXN=He8g+c|Exy9d<Be6=pr^DgJ{hvav;kAejB2@Ba)
z8#zgS*@a7yt&R4yqP(dLmLbZgyXeG1Zl#-0^*k|{j`G2=)Q}a$^DQd~Kl2@5QlHk@
zo|COGQpINjRbv{WqZS@HkVJn6sf?@bl6)*_e%p!C!*LNL_6zXP)Mf=P|3;q%<mQ5Q
zm^|&IMQlfNP8?M>&^rMe%{s2nO-dFa8--rdZs@CB);ZwE!cB|M-(kkGlRJ~Ec{qOU
z;OvqOgf!WjemN$El15L}KhjzoReI^|o~UASu*3HhzDHZRJboxi<5hQ{58}&ldcINN
z)jA$vwbFzNJ{ctls#mB0hwD5FCaZ+vBi}-Eq{x^hWreIcO$?ja(r$Gk`_-MnUTt=n
zMhRHpJYc}OHly`i->;|>uyb)r`O!;5r(zc{63Z>>jqh(YPuR75lN+5CJ-S;Ghz_yx
zqi@{r`4%2((Nf2*d6l3F+A=cFK}&{zcJ`XtKQYeAD`3=C^ne*^c74vwhPXHlmV7OE
zBUB$5cP>sw#J_kBbPL$&nynO@Hbp(ZEuIC2tsjHDXDZ?z%{oi6QE#WWFfx1ViBFs8
zDY3S;iiFR+CUZ#lH?RL!KRT1gvA04C7-J(~r6OUnpS)Q>JV4XO`43dU5(k(LEKZ|#
zY$+MZ2-oO1VkF!qVNBiGLsTF~B@pm;?_DEH(-ch>UJ*RT-^Aq=8Oe1jC^)J82H((q
zV3f^jjo%r0l&Zin8Q$&J^@?&285`NX!FH;wBg)<?l0LLbyoIej4SM`T#L=8QQ-3}H
zpxZ3bNxF%H$3q81{P=aNk&X+J=+ubS_)V$0v|6kF<iw1`aC>Ei)nA0~7t`t$X6`x6
z$)>eXuR3o-fdw{}lA)CFSq`eZq`a%8T3}|(xL7e+j)?&GnAUt37FKwT!2P%=d(=!l
zF(ShI^))&gzOaY_u2}igzBcY_1x1w?xLv;xer3|U)q@v%m2-tf8F0SjRx$hV@JX6x
z)}qG$<CPs&Q0;zGf{W1=>EV*obGedNYsC?8K-~uROsZbm_r^vA0pq^f7Ds=_U-P;t
zt7vrBP<;yJYDb@mYb{YE@pvd678U?u>sn7m#wI!SMw@+W_Q;#tHZjz@_xF5+yKnIb
zN@4rA9_mY*(y!h?%|+tT6AZ(_g;?@Sr>V~u5x*1ky>tnD=3F~3Q~A^w{TmHf2!F^+
z*LoQMzq~6ANZBgQDI`ewZDTg+`CPFE$YTBxobUoDGMV{$1DD7=ZtgbzzQ7$sF^d!@
zE-=wCrc|vjeXz(zH2hRP!G3qw@Nyj&&!<uqsrEphA5yR62i(Fxnb1tcp~tPJlwASu
ziqgHcdkiMCb~rfvZ|=s+UDy(X0@-)DXORLJZq3O)(bBFH#JwWI{W8p1LS!Q*L|RR-
zd?e1n!Qozii|sjIld`@~x{#9cGYn&E=XK}KL{kFD<Hp6cFf`J2cXJceS+4ZA2mx{B
z=mq0~Y@8iUsiKFgiv~wfa7<oO(j1MYe`4o{-31O*tcq4>us1XaB|tq#AxF(eN0aD?
zXIXkZ73bf_enLP!Mze_kB)I$YVR2(d?V%|uiBE(W4s%tIIsYK51|Q%9Zz3@-bcsH7
z-|G*>{~ZwrIrTs(i7g#vRKFfOs|pJ#b4j-QASmERqLc8|-IdoQwI(zOt5R=JgOrT4
z&U}T#{q`Kj&CSg|W1^yvKKGSQ=q4O_1xorY6<$JG=4hsaEHJQbMbDZYKV6ndsCoFQ
z?nj$2)`yP;cGJ^gVfM&hi{R2_RldVy4iO0ONVCC0)N%7_Yin;BxBi!Y{%WmN7@_I2
zR@a-#iaNtxJ_|lul?T689Q!lH*>o)8za_pNVx&?sYdiI6r|lJKH<ud(JF~gfAlN!S
z-so1Z$xnH%DPIRW?^3~6t7zm{zo8XM=2!%kRGWlk<j($gr4(Ix@<OA3B^I2!o4KN(
z)1{zxH3j&oOPyU()g8*_H;x3E^kbjA1}WU{rg^8DpRk(eYd+0GdNv1No}W-|>Al9?
zN&kkT94=B^elMI}$?McUls+o#h__Ggf6aRR7T&YZTTN3Wn|`0JYjXCd;K6*#q@7I$
zp~ocL<4iqz4Q`t$pXb{dwHF&4FK(UZFZ{aytS6SH++4MZ%xLK8dOKC#o*3&*GljCV
z{hN>f%49z0L`awN(a*7q)bMXn)L&W#rS{&tBw=jbk$6-0V%!#)clsvCYJyj9&~@Fz
zA`ovi*@q(5zq<-tvKz(3tZcn-lIU1@Yt4E>=q!ilp7d)qPG<9=Ch-9k^`6g;cMiSS
zLNvT5)j-}N+117fZUF~1!pebU7xP%2+&xAHq=EF1=9$Wz>XGP%j9FX@0Vd1qhr82R
z=GJ!VpJg9(W6Vuqs8?bl1ZfS^f8j`(0-2*g)~5xCEG#ek%<D3AAq(0v(U5CwV!Hzl
ze)c_4Fcw~Ch5?>+?VoV4el>~frD@l!_KelRa`<%XBfjeQjvp-Z;1m2kJqJ#pi>bZ`
z6Xag<zd8}z^i0M1!LK^r_0IBOxLN)}J6ys=u;UdOcJ*z9E8$-$aXhwGF0|)8^fcks
z#G){>PLBMAb}M+3jCSaF>?_agM@gCM*O*?RoMwRqsYS#_sUiEhdRZ`TAv98{x4x$g
zvYwViWESi6R|WnIP9qcz&T9en_6De@<*e7ua7j(d#n&BN8{ReA4nJiKbV&cf5<e+o
zbePy6POQRm)_iFV!ByE^t(Ae(A!C~>g*C@>1I3W)H^CPMLpCvzmL*=NdJ^zqSXcv|
zYd}WNazL?YWp2p!vizjuBk$}j7Dn5aq(Y(L!i67G!rY(#E$Wvix^LliD4AG+;zKq^
zgKz0u;YzVp;XysO?;7();A#WgsbmYLcIqAm&7##*6BdnTilS)g=#|6WTNkn1Qy7xI
zb5X?4;gfPO+&SS0LFO!GJxyj?QLrlRyR^D%eSPq!h=A4dvm`x_8Mp#{s!O}YS#Ep#
zi@V$W^x2mVLpEi{#EV23zD>x-C&%d?;~!%OPlOB(qrz8rHVdURWWtp%>=_-ZUMvM8
ztpv=AukGx*#u~fmyY4-@o!B}Beb_1E?Qckl%k4Z;2PQn*&fzB$BhP;yjdBR<pAAdB
z*^qkg>QI?OBh6IRYs$*Q3N@hr;p<@UY8Kw8?X&IREu7>R5f%D9%Nn6oZW1g~1)tuZ
zCB<Irjm-oh!3Zm(*g_SyktRKVr}GNa_eNgTbu2feR1-V%5I2$0(lT#7e30I=yd4)J
zk?bE-_BlbZd^xi88a}0&mVwb|>Bziz`RU!}?*2f-&e2NxWXG+R^GzAugMJwNw<yeE
za$ht7HOIp)lB*>TxlK39pIF$2qy5b3lcp;V&hc^C743txP@h@UW3zW*VOaOqb1ym&
zi|;M|Jw*Juq3_%1yTptSf9Xzi3n*U1NwEMmj;GNFw9C=_+U*0^uU&Kt+Gc;hdRhAk
z=X6uyrS(U)*0#*ZS0>8Fo<4MDtHlA0?=1Ja#>*W~pk~WY3Xn>mcV9uMjbg0PVX7n|
zzx64xHn>5&c751sx%<y7>Lq@=fScZ-4YvZuWVd$?Jl^#KBRcsqw8Eq40^@KtMJN}j
z*gY-;wc4MI+JLqO<|UE1ry;{G+px`kYn6hf{BHR&Fj`~-%H8ygK0Yq#6noyu-a74D
zdtwX!_e_)DBRD9T-T|9S@nWS!4951JiJOG%VtNdUWKYFxnssu&20L{8`*?dN7^Z>V
zPZ+;e4C1he%Vjg)vBj4sgt+NCx~^8bfU|<28Vp0K(=R$qegY?`g0ixopI@gLWDAd=
z@}!OO*v6M!uqJ%G8|R%iPjvhjcUNS_j?b1yd++SU^02y??|-dE>6*-s5`Ig~c=SfA
zQrneBc7t~!Akb;OvPvbl>32yy34nF$2+VSt;KN$4f8iz7wh>6~hN~i`Vu3$ZD|=4Y
zKNP|@IXpG9V`2$H?EVcOUh4BbPs`UliOHXhK-C%wgR%RI$_^(!hB^CTp+z_=lY1Aq
zaWMtLw&wDDJtu)VJNv=%Wl#z}OCI;@b@%<%qilrrpBol}?%8tPFdzRf;B(jQ{X7Kl
z<@OpPE#Cgm;Kp@?COAsZ7K>Xw<VHhS{O!S&#Y}BlW`&ad2VS78){vasm!P0ex4T8o
ztCtbn;EjxE91RKaGW*ja;0_FTT1nI^?YMf?)z|UjUTyD5jE}rRb~@@VQ9I89c9bo*
zhW{4W{y%%iJG1(FFxANH<B+ck#g5C3!;+N1CpquR3l7iW@5$z#l0Q?(X7kdsRV-g{
zbl2PYblKLB3u;ieC8fsYaT3Fbk*2&tnLcP|X`aH&p~**PTG7CK%Qm-{UaE82zX56;
z{TqK&ei=*g=MFkY7H#H$FlLMrW|h=9*+i~>Npn~p16Sp7HNb73V=~}R{>JH<rP(Td
zlw;48%TL+gQG`y<dmS>6^@q956Z1BcKk)G8J|)^Yg#H^7d`{+zY^U#1k515S!}fyc
zO~zURF~B=EAqo>}>W=$|x+ZPDz>9;yL65tYebZTqPXo(!4IXka4^jOE{LRDT-6woJ
zVeohFLT16Go~ctYuHSBflEXlJlse<wS5yeJL<A~@X(cC{PQU+!VnsO6DQZ<!C!~vw
zefsN{`IaF8?B-~(R5qpd^FzoMQ{&j9O^6*PdcIreZr1poi}|QQ%aa=->JE|0jy)Vu
z5{iFup{~h8Q>XsS@z$x#>r)93JgQf$QcUeH$SN7*-+cIR?dm%6o;!Z8!NF+V`|;dj
zeD2VuhE%X)LM(XJew*dbhS7FtZ~zYv?-;{-f8ao8J1OO3cqG-90^Uo9{FB9cxOSst
zUtexn6|Ru{@Z1~BEadY%Y5V>&0onhe2FKq|nCLe6Dk`8)mM7RvPcYU#`cNN1%oW&|
z06F>3;vZ3o65MO81O_`lyWF+75%HvZ`90M9{Pi5Q{6hQ##FxSGvtXy!aEmf8oPe5N
zONWQJF){6<Z%oZ4$d(oZ;Oxz2J2E$uu*f!tYL_TK@NKON@q*jtQAH&&@jxJ8n!?Iw
z?Gm`3%z@%zsDW2U^LniwbG1~00*IHm9oBrivzs7MgZkY{?Rj2^6L=H@h$|AC#nQQE
zbb_XZ59;^8`fyg8^8N9^H_#Qzp&B&ucqid|_=IIvuukpI{7EEh+@5J7>i;O8?PB_4
zQaM;wdh5^dW&u)@ob`lvD-P114csp3GG;-aD=)1+%x2f)B~k1VPP(yOo5;#aBEcQZ
z%dS6Z_Y#<eShn@4yS8LKO3E&GjYkP5<4QUtcGS$F<VF|VPhvrlk<P2*F2zgbXTSrR
zw=e&^&r~`!gw5cP^zbb+uf}?MIA68jG57Kg|9=NIvATv8d99pF6wu>GWyIX{r4}CC
zLDPh}_ZabvI$W!9L>ax`sB+%PwNHkO*EKv4T<;#tFa8NIZ*(vMrgzyV>{6^M%gCPo
zw>8nZl<gNxWfQH2Zk~%Zh>w431llwUf-nnHkH)D2xn}M=jf8;eg1UMmByGc}fk7fd
zLO<-4eFPAHgHV&s^j(l$%>IP9o-w8pd$Jt~-KWpCEPEV#;+JTn@S%f#Pc8S>7~Zmx
zz<Dvtlgen(3(3ouZcjv9xLvi^Ak-}k*VteYh-x5_C357b|G0TutOGUKHdp@Eyhu1;
z4nytyJYnG8(hZv+d0T_=*c*_>#`D`c`SMC=ARH>YlDj0pm^!sf##aw3aiXDPROavv
z3KG#3dg&zK5;{h&CoRY@xKeA$H&gZmx7>7&AbNLR<D4wybf7<D9#v2{<rAz*HtY>&
zx?#g=s*~?`fK<ln*Vp+$<{Av|wZAl&SiAhi4VAgB#vNwd-NKhr@k{EZjiQk0`ALvM
zT+}p8H3H{Gdj&=;WMQD8`R>}Y!@2Ze>cU-K_rHTIhH3E$9#|#oanaU;=rE9VZ3(se
zEK;f|R!WmRvJ*T%sY)4GgAa(GLgT3HJau}hDvi0mWnGo)i%xd?;>P?jXGA;}djl7~
z%wN2eSUyf(Cy%%Nz=5trS4@GTr;fZ1!r45Mht_OTOg`C9Liz7{x9IBl7+&Dp2aK-1
z+HkD+;*X41*6;zhWgFv49R9d{EY0O`e0cDZVr@&h3>waBf*xhbm->#3K@L%GP1|W|
ze!n~^C_fY$DPkfzsEo)WKX`JI6vR2>e%z1q9K514SNUV}bvTGn17WUjW{JSc<Hxcx
z^z+LsjXYRvFZArF-49>Z{NPXkBZ*rK$w?xhT}^vp`E_ZYZkDwg;cH}F$H39o(XnbQ
z27O8Q4VH~!FpKb?tKF#3f)y)lT?c0V85N{b`N1ETHvZ;_`l=Fy6qrmJQ%AEspyAf<
zqriHh=Y;4_kU$Dn4&z@sV2b;T#P{z^(~RL+p$FY)>wO^%_>hC);!ts`|KAbSB_;;o
zJ@z+#?QM(Wxj0FQxvbQAwQ59mm^(Yc`i>IEb_S1$LO0)ma*Rp3^CjqrjgCH5XxDGs
zZ^b%2Wr8e`*io@Iy+rq*`tvPjIHkJ_s9=m^ag6BcZ(Ku<@4lt`M%$&i4!Do|!}w1|
zdv^c$$*!JkuFSPfM`a-&hIp%K0K;C9re5z-G^_2I$f}4`>pAG_z+*f=rD*6~8z%A-
zi_63$wa(~Uvn+&1QH>`9Ly*U6ksD4oukG<2X_ZDdX3LPpIDlPZ9o#Mp8K$r1i`uM2
zL6q((dQ#?fl!SguZ5J@=LR(Vf_eMr`cS+;<qkF5po4dUA?AD>l)z-SW{L{}5A78sn
z#U!w}&jjqA81Ach)j#H{eU8PG0xL^6+BkezDiP~=%63-tyS+|gZag0S43w1W?^1Sl
zB{JhOS5s;5&bfNT(8JjSB$6_U3Z7p<M5N({$v;iefT8Yk`2e5@;@(2W`D>3mteEC&
zDf+!#SYTsE_Eh@l@$P*9um&dPRkB8pC((!B>>lOyQ4jp^1EaILV79HPT)%q_oV1?t
z*Q}*OL&i?(pN!tLq^8m-+^p$?=GxxOqv@?F=6|xj10fQ+*562+^)x+{YH*t?yb8gn
zv<Sf|#qe&VYKOxfx^J@>hwTtiBd#5Fx9>9acheO4^nNOnp6qh4v&qeFdMw_AOh1(1
zbB)8%v+B$F8YMI#>v^)eM<Eac!{*|25_y{kQtX`%_EKO@g_vp6VwQ*-#W|g4W)}Xs
zTTU~ox#_w=r=_9wNfOon5#!*7g3_MP@~giu{&)kSQkEi>BdOvST1?5E47zn$W}r<~
zP~SQd5C<&jjoh{x-xKHpHGMN!mr*p-TQ=|=w9kU2<C|8JD%nK-J5@~r0Ak7XJ%r&_
z&XOM}v87k0>c&h+`bA%owlWUNGSAN6mIi{zeC!&=RyFC&y3<!~DXet;h-jTpsKft4
z_ah8I8BWlqFX2-q#L1D3Md+G>$v=aNiZm<^GMWqbKN|_aZ-huW3t@dE_o*NT!tS=r
zI9+zW?!y`GXAb308@a`t^<@EVRwuwqC3n2Te=MK+jswgdl4U}uVsuj9Uk@Sjj}#zy
z3uSxcYc>?}ITX#a$%Q$$NHwjFt=31KBm2~3!vS$}QjDcAy1&4l2x6X6NuGw=r}s6%
z>ZmQ))gtuEhorn*Nj%=qOH(cx7g3|Y+pdYtLMFFFZW~VGOl#qur!^QKySap3tiWEt
zGJ&0L<&fssX{`tI$(Dw;_d8CC&hVOpR}n!thdKA%M6{siNI!wLET(*N9afz818UD?
zrjySQOZP3(0bGpb!Oq_fp($H_wn`i7|5eibhhh|)@;80)Q(b+Z7r%(r=5D4gA3uM8
za=G%`<<UnSHf0!Iy0pD?%}_!w(2}nQfU||dIpI|HCq8;6)+3*-V$ev9RS`z$SbZV7
zo)G}I#h9vO|Ap{pF#gEX&K7rMG6Drfkc;HVA5Ai7BO<Pfs0?jaECOBs^{77ANxNY<
z`^-~FT5Lh6hzR(x#(J!9U~Kr!&3@c6q1$2`uPhcdHBM9{8soxhRKK1o|9xx@0xani
zG*a6fM~9+Nptfb<l+m|@6ch0vA~CYF`V*01zV4@qRddb|_XY@r8yj1QF^iIv&-z9=
zya4XTp9*|Z$0Yb_y%?dJ2UJ%3^#OBafb3(HuQkCnThOmlx;Jc<ro?ZwOo~C&Qc4S8
zqp7okOo8+w?AyhFfjdi%@Pz!EyR+QE-SQ_A--(NwfFgV=G#{w)w-y_=gCzEX@%25S
zslqp}$eJv4XR!6Yb6e1_T;T){X(PlwG)h`xpqka$i+k{st=_&p7s{<P7*|at)$t;F
z^WT&w!0KJW%Y0otZ>SZY9|yH0%BTKPFA|S_hZ7d1{cT3?>~EX!lx$jeMK4bMX@wj&
z?%wK=*0;#gzEAfNm<ciVp;o;;@&ZFG@<J7Kb#A`UJQS3cYk!EY%g?{*!k{+>n(z$6
z3PgFW#v&H905x*HH<CJobjE;ohyXw$UCZ;!YFC}SdM%`>s}H1HH<xqTZgS>5<BTiw
zRwG??KQwLY>1L0$=@?{#LS5+U_VN|mVW8&p9KNc{7rNA(6udHKVvDSp6ig55miEU8
z%o3oWo6fLfF#Yy+j_>N8?Me{)wIJryX8wv5i9E=7HA-$|LKdl{@eYo~VRVfb26WeD
zMLPwp)CbpU=N)#{Kb@p`$+kR9A2xS8I^H|OS8ugkY9z1TU62r)B!}s&912@0M$pCB
zpnhxI3b~tCr7j({8lV39Z@|$NAi)akA1``mXTel`gDqj~!r3I*YLSU4qQArCI~z^e
zF%HI)GCW+P)U*ZwY%+FgJ36_IdUcftz`8Fl=8mshP;0eake)wCA9CDVA0Y-x%Dwb-
zxsui_0&`VsxjF));F7z={4y1zhdGbK4sBsXJW$NXWx3v=0zNr!Huq6=OK88M$X<w4
zj3x0ZT5X`{3=#T<nfF&U45UBV8=Iqa@9wK=oqGNd_1Fk_q`|}Q%4cA$Nr4=z+qr`_
z4|56L1gqGEO3Id;sT!!(%n42WX>c9RB^;I%*J<gkB{_HuQ`vf7ohsHeW@|Soj-wWJ
zq&c97s9*X@t#5Q4mzM)#tm{SK6U_kQ39UL?00?Y#Q`vt<hl%{n^Gw5yCG?9<<rHl3
zq+`A)XRjU6@VIYos>?Mux!W3n)0M|2C$}}sItL}`@0e{(N2nr_hiyyCm{?39NA5L>
ztDzJM#j*@l*9<EIjBv_SE*6Ev_LTJsIUF;`^!?fzFJuzA_&@-XcdoeHVN=~>CT+Q+
zT+i$6m%HTFnBBC%EoVp?m1wS{dimk?3L_08$j8&o@1fhR^pfwt`|LuJM`&8}1#}4R
z<m7={fxEB0Lm6dN(QkDrH2ecT<x6mVe%N=vzh*$?k1*3TeODK+TS9&lP74a(;xkp5
zGKT2Z3BDv9e)$(y1@YCsxk;~Y<rnDm*;y{5d+j4csiKmW7W(#YPf(4bBL63M8q2I=
zP5=JM-h@Ih8LsLy=G#q`_929?P?fEze#J?G!`&=a3;7&2;h$JRuB<;0Jwl8y&}6i-
z0Gt7ODY@}mEe9Nr%w5EaZpptFyU3!GH76+lnp|%nmZ{ikdxXT8l)9_SgM#d}qefCU
zVNcI^=lZ;lpkVgqdZD=<-S~`B4sZUa2{*`{t7~L*%wqmlscg0bdKgGm?d)plVLvRG
z>H|GVsJQ4Cg|~~DSwz<A5?aao+>ezse|wTTs+4G;ioxfzeb+ZE4v>GlZu-};NP4Sj
zy*fumT!ua+aS<?Xn%YV=AC&IQC2$?qCoLV!rFPD86pMY!pZmf0VedfiQxgfw&c}*(
zhSK`xf&+32v&qc02oQ;jzaH*H07sm^l5WT9Bn{fJ=f7N)aId^%E_C2<SbB0>_7r^B
zP($!&O1Nnas)y~;GwyPY^dtcY@&~<gaE*_Ky@d!XX?{|@Mo|FY0vW2u0|<_@1p`i=
zYuxBXVo<7UXvh|43Bo3IXR8aHu*I)doDj623`#maB@^#1TYxWkd3FUIzyH<GTVZ;@
z>%f1>dO6C`AeK@KP!gkH_H8!$j4vj-AE75LmP*3{)9=`$n{7@CQ(DP{_zA=F)J=c%
z)+Pl7Rxk^SRFmjQs7IU<e=QQu8Xb8zaQ@AtkCwq;ZjIvlzIK_mggR<=d@wY~swp*2
zd|BMPHtyhz4?Qs`;7w9&>5#Sr-==XB0a%FnznGfI8CXeM+Qy#nwaqv9B-o@%Pa_N3
zj1{;Wg{-{j_H=I>c+y%Ok2au-Za2##N%1Ft{kVB_2y2hqO0`wWETw(W;N-&nT0_D|
z8hP^nO!K<>w6blV|5iukL)X`hxV^hfIW~}f>FDDNzI6oZ*)`tX2;?h21BE#P9iIC9
zZ&g9Oxi4WHeb7L|UE+;(0-LL<t82X-Th0M@>@DQd(a>gB)~&4=uwvy>%dv>B1}WCt
zE~MM*0=JHy*HYlO<yuHRuzo!$jPv6Fwam#!;Uh;&nJ~6G6Q}BO9tv_RG$E5`rrc1W
zeN?AqWJwyyzUaIV|0oOrFaL@CwW@JPG$#OkH#oG$i;8+JhHR#_`=5*lyI*M*mfHQ0
zPT|C-9W4;%I_^txinn-XSgu1bW;Eas85x*uJ|dP$PA)Y=j^9SfB46zfRfHx@@En>+
zcuzZrmLAOlivxRsJ3fueGEmmV8WMegj(<qw4@Qg-5y7AIe>B?Te3l->a*V(>jZ2>3
z)dpdc$+vlEm$<;H{y`+fXt=b?!9e`<xw*$i5aVz(Mac0`b!#RS4syV;yY!XqUm@?M
zO^r6;=qH<AslR{MceiwvF1{o+c2dbs!me`pr!Z7avTj&^hO0exj7H<CwtTwrb^gxV
zHXpp#;AD6amcn8cj<(NZx^ih9^I%r!xU>E8D^t*Lq)MBu27`am{`fqU*g7mc-LutV
zap<@gl0LncQH{bsR{RJz21CB)KyQ&UkQn%sxa`2(vpL+kSpE2R%Pvd&c=Q*DaB8b?
z#Zs#PtwL}Hw-igE+F|%y7^nK}TwV|8)nz$ib@+vl(5`9}><cpT7}U`1LbQWkLeD7!
zZF3P1mFlys(3Gdii}%q5a7U$|olm_rCPHqZ5kQUiRe1Qq>~^PuWBbK@&EWgr$Q;P-
z0{q7(EM)r>e5P-s2$+2T6kMbg1zIqB&nP9QkP?lyLVG`U-kxt3)Tb%jNMMEh^nYuz
z*V9`hex`{Z78bTJRvX^?(XY{|b8<G}SC=i``hgmkXoCpyie@#<qA7aCU!jK$F7h-1
z0>5@RF67P`nsFl<D@<jdB_vwh(jkMHE?W2d3D|;Hr{3O{Gd`fvgrDfTiw#2639YbE
zbV*tW*y!~wv8<SwFzTC}VDkTK-37obb^R?c!{?dOx~62!kIkjNjfe;<onaR!P9jJ$
z5RC3)MQYRCZr#SK)9oONg0=yql?%5+C6a^~I;!&&M>_P;g$@-(t7BIZLV_Fw<$?`B
zhOu3~$YrA)uJ#X+5pNL$Avr<OeG|wAOi*~idf#3S9KLv6`hAmN<Y;GWAnY|BZWl<K
z39(rO$w||#ZJ}R99P9xvjY&7}Bu5?NlB5G~!~qb?HnMTI1PN}i)N-&P@OHeLm&v=o
z#6#8R7RZ*Y@_d^&IINpOK2o-av=lzBtX9zCOm;)m%kuv2pCsWbqY338vK-x*>{?#-
zk<}JO;{Mzy;~u=V`Q>mtY6F*A=kl$2wzR#ah2NonP@yD9{bW}|y%WxzFg!zPI7<}r
z?%jy}4N{3}e{gN63bUYGRSEu4_|7Y}UyVidkPs5{nw_*p$Fr{e`rXEj7*#q#NNKlq
zIvf8qg3d4TBe-zS>(_gi`_-OdCL&QQZ^@fZTroB{3eSV9b%QyO<B70&5r>y}?|te2
z53yx>K5vDC(E-8S&isl3Xaa&7j21>eUlHJ0pE(({Ckm-o8_mzA<OdX*hd7v#v9lC#
zXc@1*6C+*;dl_?p4)FY|>v}!!BtuOB={0|9XM$>PxK@P-8vDaOpmxqDjtYo_+0OEe
zHj_{Y`_cXD+9n)hnb$WC;czsI0!HA3nB^-6=?jY{0;9x{rsfkNy3&4VQ`ngBh`6|M
zE2f*o)nOrDvPSdY&hLFHbPvdh?9(4p%;!uWHQgR<05!{2YcR%fKuUE6<O4ZoW5bm*
zl|L|raAXEuv890AI;68CM(J2T8{H4_lp^@Xj*zUtU}dyv#u=0{gluybXKFnmA6aP>
zVmjh+vwTNlHR2F;-}@qO@;?fZ4n~{Ml;p20T^Xly5wh@)la^LwZXUG>J7@B6xC&gg
za7CT=(e+;m1b<u<5T87Jq{bGMuj2o!Y3PlOqo+n`ne^r4-ZHmx<==iPpS9uvoogsK
zyVycdqZ}<X7{9w#DIsm@rF|gvWY!T*i-z|j!r!%K$gNDZjX9xxjXu(?jpE2a%4s2;
zj^?o}Ol}C--0@HeeaVHp4V|$B8brK_5YCCHciJIFhd$#E=(sJKwK~ntEjfkN=9mPG
zO{ADu#5Fx#Z=xbKAmc$ML#iZ(?-;QUQ+K@v-wp%-XZ(_GMb(lqc8id-Q7j)=<?Y3e
zUtu7`1HIk{t@|M%t50}c73if0VCNa9-08;tr8;L$GT!Xc2xQQm%b^WNE)PoZYOh}*
za#i`zei!l~4+S=MtbD=`XdZQOpFgU^#5;%0?<yT#S!Mi)5Ns#U)KlU|olQ1T;Em01
zxqCMTqm$AF_F(6`Br2;SrhVv`FW2*y2wdoz*+WiTG#VPF@BP~!^!P}}x2VDOLX<*C
z$<8*u(KhDdE)B^?aET#}U(v|nDZbQ`=5H~3tFjbQlRKejf5-BCuu&BC8ff@4(DCtq
z5RFOZd4~US+M5F4x;`PkflJM1!tJ$)IW|BUl*3GAU~(f;E`L6BW#<;^!M3gEc}Kx|
z7VqKC)SgCvR8|fGqQhkO50>Maa;+;4pH~~=W@1RK7V}mWz57P=Ba;*Pr=&G6y5T#t
zLBEIeRzk&=6D%U*6U1yCYv|HG@lVv)ka83nGasK6>H=FrzbdR;?kK&_T4it2@C<uD
z0IIwqT#nXc1S)*Hxu)XmOwmMA<QwBNbZjF5$4Sor4PxyM(0h~J+7ZXbnt0pVsqltZ
z`NBt+2wzLQo50lytIP>3k1YIluax>Gea$Fzktvyq=PTp7iWw?{mOrva4pIyOev<4$
zN%tomJe3DNg0Q#(n>YRXqD)OV=vN0m2S{He{400-i0n<LCNjrK*Y`R{Z_;pfxAIs+
z(u7_xlPZav&M4u*$(JZx#0d#v89Eht=Zbgw^GUF6ojfHREbFlxZycFV1iNK2Bt}Vg
zdk_>qj$aBAt8c|J&YZX3v16kLFy63OK^vfty~@)cp&Hsn4m>Ou@eR~hxT63gS>w(f
zrTUuF)9b$uxZDsxKDE)AKyE9=sA|T+v^Ik4imm|nbGlYrj?>sx{*e%Zsu0BsxP#zE
zRh!-0LDJ$Pf&khBL?iY&(vs@4Z^d(Eo1+(ODRtt@*~ce+x`beZ)!d_}CA}wZ>dPmA
z;OfT9cz3s{u?=umc-l>3Bj(N?HrB~|V`Igf@zs6j+lwU^hS-WescJ$^KKb}*BK++9
zw^kYfXvdx@1`QBKYvq^x`V5UDbBNk9wEbL`#$TR19TK5)U<P?xBX2Y`uPGvYVrdQl
zwtLB-+6TJ511V?~DF!Dn%5DLccW3*VSHP>GdzV~NY4QnBe3vv&sdDygRmw(QIH{?n
zHb3B}lJOuwoW{}rOi~Xlr@@w)R(g5tZceX+Mh+N#n=edGYhGe*^b@3Icm-i$UV1Qa
zgw|ErgLLbs>C#;O^8yr*%@#5)Q1*q2N=kO+nE)|!Hw6~4I#4fh=42G5sr5T8TO3o~
z{V-_^nU!x`BOmjEBfUIkD-pKZv>m_lf8<7}2!Q08xRc|vD=0N*iP|in<X3{p(F-pt
zD;C2D4Uui!_y|7QPJO1ApV94#rIM&?k#M@BHbHFGkrxBSps7mC-PAd{fzT*9C(qyM
z(av_aMKy1g`ppSJ>t@#H-C%3-N#TS1r0Mm1fip^ge<aVXXnTCARNKC&dAhd(|LRL2
z!+hGfzSzfg6{3C)!ZyVseD_AdZ<7Xs9G44A=mt_ZCQt3W@Jqi6>E4X7y`9Oa89)#j
zI@XeoQEvdOH=KV<RD(Njctrf^wGQ)kMbk9!bNmPxWVgQB3ixcXIXBdw?#7Su^SBch
zIqk29r!-0q47|(B{{aVKPJDr6qkUxs1fh$i6ttw(8k>SQ+~X9kzn|mEwF|GP$S6@s
z7bLkbtm`_rkvtF0RPs;BIzWSx1<1iwK=akYPr=B3p2%eP_7~0~F%!-YON}m$>n}Z*
zTl|C^4?fb;CpE1ZAT&3d;S+bHA1)G3WF<v^|4%3mPJZG1heu6aTNcos(PY#$r!Qn4
z>B?dv(jZS=`@bX9(m4?<7`SZc410eW(iUtnwk7>eDBwg$CO#m>i%rxqEp8h7Ml~^|
zL`Gd_+P(;{tsdH2TZxVijJ9@&oj_u=7Im8&|M<7fl>xVQc+jE2mv}r_e*H(2>A{u(
z1{!WlraWNcQc~`Qi^1{*Sx<}aNL?Bm*}!dhA;i`JpgZ;c#mRsz@z9leMJnQ)*102e
zw<DhX=Ao(7M!P@L<9?%Ry1m;)Q2ynA`-*;kX%h>07nAMe!os*st0z<vO6k#r1IXV<
zfiCsy=-F*R31KFzm#e!Lnj@v+Q0k&G@8}Q)ga`MZFENVG{x*NmN=cCe?!6F1k;-X2
z856DcJ2ft-G&hjN%gddAVbruZc<k1du#$=7#5sAF2a2U(2@5e&@R)EaW|9ka4h`{e
zc9~%0TUlK*uo!=d{zNpKk|e!&H$T~XQ&m5cb7?p65`MVyuO7>9O%)kqPr<{|QozVK
zdp_$VV8g7E#N%YdQ8)!f#xr$-<`oB}EQ|biw)i(cpXxk7qwoAyih}z3-O&@A3x#0&
zV-fy+Cf_%UJA7u|a3*!UuMQ!$wl2ajw3Ai%NBnRo7vVEeix0g6tlCbB8j6Z{seY=`
zPhn()fISjWOt7wS+8dC+ydd7|F=j6_X#No5VQ&ha^;LlbIA8aMQ#DnqS&cj`vF{3R
zMP0Afd`OVTKQ!csJs>%|WsRD$tg93$A%r<*pAd)lEcc(oj#I6y3XLs|a~is9`qb<k
z99hY-G0v_`<h9&kT2wHP>2mc-UP$$?5Csw<Jcp&qM4M8^&wGNm6zh9_S$>EO#dLTJ
zHPqt`mv-Mx$kr>@0bHzp-J=1ht<7?%VX2lHY{kPH{QT>Iu1Do;ljJ71L|30}dF~^T
zlFoKp2&@<&pWeTCCtm<E?{xb8?;L+Q8ZVRm1<mI2!g5&$DaVnBA}K8juJ%BMzBz8Y
zt^f6$^Jgv!(;av@4N$GUB=JJi!Md*q--O|=2VYgK=>nh{Z(m$1FFbC}KeNY59x=PE
zN!V2<+1=Xx%XgP&6B217#tVM#t)6~j4^vYT978?E^`KX}fqs6m<Rw`o7CP$MgV)S0
znFn=2&R0PcAZxBJBwkk$6UYyHst<fY@~RI~fa!1T&}_VykyShEQU(Y*=sIn5AJ8?`
zyPV}pw`WGz=e0963Fq<VxP}}#<UAY{g0z<Qs%~Z&su9NMZES2aT}&6>yzIQYP;8%S
z7P~Z>CiYGNmpW-FU!Ay?h<9!r%oPszN4%Nu@tL!mt`+6GyEXe>ZnlF#EaW2sqdp#-
zYyC@4?_+$pdp5U(exE$!$sYLgjNpyO@WX&c1F2ue>c9@UrN-*^3S1+)2t+}MeehQh
z%4NrxJ>CMQ6R7$Z^O@?r8QyL)@Vg*XFhvTyBi`En_SdValA(5@|26yt=J9SMhK`L6
z#@>9=2a-ph81*t5JcG~+b?C7ght#}_?`3p!msB&_XAg6WI~@j68sE?xzzbrvyI9%&
zi2!hJ_^}CP7dB`x9y*%5I6nqeYsv#rwgT#XwA^hse|OUZ6Q5z&r-SypkwjWMmP{|*
z7+E4^v>qu-UH(HZP;0ez)JN0mO)sxrI+~R!TIE^y?K0lLh}SIKUm3F<3Y);+(h!$(
zP8J(30Jl&)9z-V7BkH5^9mPa5N$^yupDrs%co08;cjh-r8(S?F*X+8ND8yheOuLL1
zT|B5LV6)grco3)Vpb!3IQb?ocEBS#N!iSc3U+IEAaxyVo6cOi{3(StdFD?1s-V`Jj
zdu`<Z8Mi3^Kyg|Kc=PfsRid9kw9e48eqsnQfvwp90z@^O8i5J2<AH%Dg-LsZfE&?8
z_dITNbUxf({L5`v9{wX=J6u;xN1x0u&*9A*`BuHX)hzd$d$y1L8p$hLJH6ihyr%#?
zUPgzbYSezhTJ61Lxe63*n|vt*GS0cxH<L`9+fpcRsLA2b`hXtue?d|Bdq1^g0;=1q
zEVHjH`*aHMn(`9Pjbt(wm}E_I*oVGodY4y0;k@mL-EHeob%k$(WNNyS-}TaGPgJdz
z`kIdh)1~tKf)YED<wk~|&R3Z7ZTMW<v!__98HuqajPI*LdeK`x&9|x+Z2*pqiC}pg
zQ2LAp59nBw7mkhGWyE3}fIES*$(g>5CH#6iCUwk(R@MX)pOvUO^h5}@_j*o3WRSzX
zVyYJSD-C75w&C%<1CR*oTkM5qPseu}XohwaO(vBLGZW|e>C%gvTVcnTzTZ`Umg}vv
zUl5p3>ktd{eaA78kSIM^0wEy@$*Xv2(w88_ns_4%I`2ogCdrc81<ienN65(~@L(%a
z!aj?Ha|UW(yCyAiIlls=)?<mGUJA+L3lOWyj*I<YS?9XMI<FzW#c~@Ak7plj*h@cV
z-cARRJb-C1l9F(1V`E&&3VXFa5Q+I@{dC7hGO+I9xn2T>MSa0rIM~NLYiuKX?Eshr
zSYbV1@;T#NMBRzpV_&wRXIWYLXZ>e{qqLxc2Nq8qH&042%HAOy8=tTfdc_xxe2hkd
zEOXuPy3Dj$?PsI*Udok9n~A$LA(63q^H^#Np>dT2Y_m;s*<+Ylvo^u<@sFK6Y=1pp
zig-Y}hROR|Mv$aLMrI%rqo%cN+>mjS(ACb?{tuqq3QiMy7P8;GQH_W)V4&x;<nW;3
z`GlFVpPopI-~z`n_+l-T5YM~1JB_D|6(>}@vm-8*Yu^44C%ho^hL*bD@f%$RhH)1>
z%WBxDF+uO$eL~(u^)h$>xmMgPv_X6lIT0CYyczWHfgKPLvI_f?m60H1e#G=v$cQX#
z<dJ<RC&$l>Okr&Spi@B?9jFO0Sa`%IDQmj>WUi@_;3yzE#2!fwcc~vfY~W#}-6P}0
zfB@4k|JXzY@yIASOC!z;p0m#7Cl<FGsmej<-cpR7E(ORqBMy4&)7#@Kwh6cbH27j3
zplq*pk9sCC-F5!M&p#?9pF>faqWtyglKIJCf~<^!m~bv^4FA;(J@2^I>pqKd$+Oms
zs;19Kj!@3D9Zd;4Cye1D;dAt~K~XhjFV5V&B6=ZNSG-X1EOOW*iwIIUaU=+JnB3nm
z$GN`oWcu99IEjKk^)Da;msSu_R~IA&ScD<R9L#j2$f97EpD)WACwZOhD+2^kQQK-1
z%G+*V21UeU!v?x6c$tQ~A=iGQWR@l`-9sY#l0NrT=k@QQ8-e=_DqE$Whw)N)0Xnrd
z$+G2prx4gv1C!ya>TFW%3*B+Ll^w+2-vHx29m9LJWp~D9OtvNE_RRyN<G$C`KlDWV
z%vUZ5ujx@xP#E`?z4F?X$mADTqXExb(cN}0B3)$_C9;ZH^eiejFZVI5OYP-xjfGD|
zrRC<BQTF#OVO^mZS#wM;>i98acH&(3${+WUx8^uc%@84VqyRd$s~FWO#`oL#K5Chi
ze-lEQ+&Rg)zc2Ft-<XTPf1chlT4le%DOe58MGc8Hb$oC5$#`NDyO*w=qhytAZNHsQ
z^Bz0;e~PFH9bcy4bpfQ`zF&iy1DwqnCfLtZN|i=vVX*5H`85?upUjKEN;-Yd3XLk-
z8eC>4Cr&SJM2M49rQyJ}Swde^GnE!nRq;k~38US!E6*6{p`}sd!5*ATNePCpqsGqP
zAA=b=utgVLlyT@*e-dnXYy|}A-JiBdnDJK&;~0B(JSrqBi{A*nzA>se8pBP&kuv&l
zg}3(yWmE+omSo<*7HbDDf#}QmC@4+k_g(KMop1{~8TkSVfWNFSWOibXUEc6l$s;*6
z-nNCd1BP){r$2O+Fb+&q<H1H{;7nHvb_t!&LU_upun4v<dxvwlh)8l4A~KC?4_+GB
zCxjyo4P;L93S51-E#xf^KTr$C{}?yjEFR-J6OR&p7=+YYS6^4=3>dApzq`UW*_BA;
z83HE0Hjyl;f^V<wO7ITg{d{Q{XhA~$6R<@JPZg3Y@q&Fva~IwYfkMtxz))ANpZQ<#
z`33e{Y^uwp9B_A4e`24gd5e_~*FOng#u}9%#Axz}K0Iai44d&Vnm;NqrhiArWMO}C
zNl*?b*y870x^LE2DY?c9#O$mR&UsyS{v7L&4*Up;J$$uNj1UhHu5cz3%^><IIWLl7
zWN`Y__C@hR1qi9ls?6qu$@{S7)our1x`?c_`+-&{R5?4pm|}P)v<2h}JnQO#%M}v_
z*6mEM4iOH+6ckFNJ`nn41J0NMkJC@xI$R@#W{fHKbVm$29qT}Po2yy|-So;vK5LUl
za{8CNA<a~@NZ`ejH|#u9ajnRAOP=dA2{$)|-KU}Lg`2wr&LAb&!8hx%&*)Bg?9hoK
z^Jt9>N5`>3v75s%m01AWo}6tm%6w85gii+$au-u_565u$);Dk>8opP=i@^oTzv_#c
zJd>|<HAq52G0?sbRuxo*Rq?w`0{+4y>Xw1?4?>?V=Kn;O{#Aj8DG*c(Cln#Bs}<}6
zwqUqL&DymylhpMJ{6R+YEsOawh!7xW6n_6)M|R;I6xg=QplUJK0g$)FXO4>ML;x8J
zTBcM(o)H}4@Bs+mslz2vGqFEN%AMc-)4X#4f7e>m*IrIRUx?(Qww>!O5Ik^Bp8c>t
z$Ql>$I3URjO56fM5RHi@$erVN#+$IKI0mD`KHaZ<eI2iq(UKW#R@S$L^Na_bpCW?l
zGL=z+Zv_BrwvJ*sno?^0N;C)iL~3Iw3Kr|_F8v_qG`N}};Tp|mXNRBPoKq8<B8R_7
z$%r!T-OO~{d})*KPB{M$L&O}U-LtW6MSjPA1waLe)2*?hYOTv2dI!<5veKHZ$4L`>
zcu(mWt3B@}r`3pa_1T}kX`0*-w=L0PMm3$)Pabb4mvta0=yUp6uNS(_s&I!5nzS@<
zda&?W#kUFtwcdvS8j<nZJ@nZ`F8etF?FWfjHL*ru(8(!TttJ^R^guLg_0PPx?NPdU
zLZbm5K(jo5$zN`l&;>0m!iidXLu8%7V^0qdPXzkcgRSbIl(cD8@)g)$Af>BdY-e}E
zXbFgy!P=Pi=w+qsoe88IBa=aWvM#_W+8q<;Pyt!`ZyEVx{)6FCbmT*IpscKxF6_4|
z`gDd&trPYogA-bYZ+QlM#r_VQ*uRjM07QErUko3mJC%gP7XZNNfUnBqq=zRt+CQj>
zo-Bzg$KlHtzYEZCbYfHyX9qE*GK<PZL~NYN2@_*y$MsoCE!@|E0R!F72(vdxRsXNO
zw~mT(3;(_c6bu9m5DCE}At51xgrtgecY{g`NDN&Dpdg?lA>A>gv^0u<l+*yy%{YW~
z*K>`Y^ZedF-}V0SuIG>EELdwi%*?&-eP8k2*N#un+cc-PInv0ePvLOvvh70`BZo7d
zQ&X>%JFoGwzWc!$1FIwLk%_p9rfCuW+Zzv(sxtL>f~(mht}dn!`C6>pTBSrwnl+hg
z!REmM6Tj_OvHM$-bG6=G3!8Kb3RCr~53H@Ne|6=u#q05v3tjWcFfle}39AWidAxkR
zF74IEhzKnGI=#BhC~~cS@wzis-`b9Gbe!fIhVs7Yndn_2BritUCv@?^NH9ID<dPHB
ze?9i2kYw9;Z%RBT_v_BjDjUX=C2vXI;+!+6#5r<!%(*LnQp*rO>r>NI^Uu#9zQOVG
z3Ap}-^>7T7M~3<Ri&`%=_nyZ7a67W;d>;`LaI%;0TPW{^%WK<fGcjE86&dmtReM$|
zBSO|1gKtboUidUQX&YG`R`gnz;)86;Ugg%>`o_1;-u2wDULETKRYo6ya7`_YxYA><
zP=m#Ny8kru`7IQpKoDC!I@)c?cV#VGb5zwc;bLbFVSVpt$INi@k<X1`BB5ciP=fCT
zEM3jR^AmQD&x)#s*1L_k+}cw03_Y5*mKU0d)|qM7YZ80Gd6WE~o0sM-s;Qi4r;qZ*
zYLwNd`I>4>4EUDScIo;u9~M}&B{LqBdEdRoX7z-UVgfh!@PP^9gfMa@z({mudrf$}
zm*Q_1zCLC;vl6uawV^KpwdVMmLo;V=c>WW6M;G5uXBMmBELJVs!Hl8OLdFlbGEQ2b
zgI$LuHDM3WU%in0Vp>{mkvx>P(~y?q@_OaR(vToE+Q8tx>o#TV1<zmcR*#G-5|jHc
z4LzV1H<p&e5So{<V<fGKzWCI#v;|;7oqjr7QRUF7Z6<eayEg2m2wcIR^#U^tq^md_
z%P@Yy)Q`|m7|O$PmFYV#yk`S1{96mOR0vVuP(97j7s>w9#QktU_*^})=Jwien}S<|
zGq8yyWcsHrL$6?GtjBf<?Dc?3goArk8T<N*i9^F(OlA_z7UkpuPBw9CiN^i~)h9{O
z%1PKu6zP?P86NAN^Xj7hoNVDrca<i2BQOhjzkuF)dR#>F(x4XYdcBE>%(4;7f?0V;
zP#UZX*n7@aYIo*TFmuHHM?!hm87(O<*m;Ump{LdyA1Xz+8uo&K8YMGYkx~EHO;9^e
z=ydl>PM(R%c}+48b8h1=X5^wCj<-brcZu^JKKy=shP5m$6uD1CYg?F#Ad8w<{Pq3w
z!l~$|JuonbRV!bfKl-U@IV|YjXQ}lB!fid2w@UKD9y)g>*}zigCzLZ~jpgX1+s_8-
zbGW0grPoDK!D4N#!FUedrweIKi19Q+o~_E~A%b@a&Qo9c4(kYD+S{kCDDQ)HO=TVW
zn8u1P=0jqtKU-PIiTT^7-_~sPR8h&ba(v#~jnXM}57wiwhjrG21h8OOn)~%LO=XL7
zr~R<{Uti=uGdwTkFLllfXQs9M)4i5l^5492bo<YQ69@Og1K6CIltP@*{Cuoc#>|Jm
zJ?PW~ER&uxP&PD>K6Pr0PC%;$rUw1Bh(Fkz;}IphpPm2{F&g%4ubprS3&a<%ULf-f
zp0J=LjJQiOiB%XeG<4ey9j5h#r#P@&qi^QSrJy<(+rP(qMy(asQ$+OA)7k#u%-M`F
zICmn(UG|-Uo%bn*8~3!B1vcg-p~eN%ARDjsVY}#aq|iz^DfN->Y`wVqtDW7|nHAIi
zW#0?1LHDhvJ+bGbY0duMi=kZ|%?C<HO6*<xmbaZNX&x$j2d1*QoE-R1reg4JZVwYo
zF=yNDIQ+J!HRs})9dk^+UihAWot^cyEBzIVvfV6#!-~xypLe?Gtzu`>bnBY4V#};B
zHNy0vIkfO;U2%I$_xoLLk(|q(`+^txpXk@Fk6q`pr$6!Xxlow6hvQ_SJbo1`9L*kK
z`O%GHs_o`*H{zy}KQk`rIZPGAe#rE97u74Df}HgT7VT<>X7Qem<Uy=|Fx6xs5F2ZE
zpHeOEs2LgAnHtROvgEVOXh!$YMh<-poI7-b9Zl;Ao@dFmwcEB2$#rV2ZnxdBdvTuX
z;|Osz=S^$@0qplvorI~@RVZsE^tu1L-Sb=eAJes>x!B&rcB$LCbVd4(i0Ai%6lr3O
z{*lM~|KKIIfJ^(K#FBl$!_>((Y+|x&)t&R|*;~emHnr7Xk{`LvUCZtG=UZpkth!M{
zb3<t31kE7SrPdmYmC@uY8-kSAI`gu91B+-j1$TEEuE-&)y_#*Kp1gTJaM>&M&X6aX
z7omLBFR>@NwLOf)ijs$!xK@GKg7d-W3or@*9=4FchuV`hx9n~*?&UC;@hQ{$QAW?%
ztg%Ei<Va5pYlw^(?BYt#Ts9jfuc)S47IPC1kP}r9JDbRK?|uH!)w4*RZ9#=MUsUoe
zni$C)y*ugq878FKO?Y@5jt1r%r#rqh2bAhHOTlVO-`_(W0p6!*TwnZ~q#Vu^P5PSO
z<4aiZbjFTH_f~ABN3q@2OChvds!1y6egJ#DUuJjH>Ys&`sh>zKFI;Xd6JnTHu+r`7
zgQzS%?j9>G_OT<SbYUXzG5f^e0J)xrU%{pD{w(RQ<#PkXF`gxMjmjU4n6L9p<i?-S
z521GBhYf3Q6ZXi6a*8pjbD~~-sXbqQOeCpFj(baeN=L{!3knJI?Bk@}Cj!1;J@>E%
zumJ|-p}&O1$1!>z8TIY@%s19mVTm3iBb~BDUbopzrT<yv(nuF3j5n2|{hssUjr6YF
zsE1vhb^Web?lBW=2wG;T{${gYHy6JgRUdb9hfCAjXxQULT`~JI%IM56iHY)sE9YPW
z&TKGETXUC+MAT`4^8qo@v+)hxaBfBIQeV&Ad)f@h4pIrFbehLD1R9u$-}7G&s1h_K
zf62GQgl{MlbIx96%~LPJdesx<()GD+Fv*-i5QTg?^5350Z_E$z@pGZCj9*v2cG{Pr
zeR3GNfELoo<+W5?advv#GwQn5p4X5^-`vp{O~ftM3=_O-^V%eXK70>F9qsRZm@t9e
zz%hjbq}54O|5TOfYlr%53N8E8y{h2wP4-f!f7+VU`$IBkcFQLJ(b~@{cvK$0Z+H;s
z6gx!?LBWks!S!L*QUOX+<#^!cCRQOKl!y0RK;Vnm6B<>|b57W}E+@-n=9C{5e+ql4
z)rQ&sJ=xE(w+ajTa|a7aw9K=`EmpR-ru<bI-4EuHKUWCqi}q&`Xm^!+dyW}wT!nK0
zWc$V|SQgeQ1>uNw#g}fI`FCMiTRHACmrI?k09R=LxN(HRCrxAV9RwU@kg7Vz{O_i`
zDBd>;uSs6y6yG+kS-F`gzReeWtk>zfaLmGaBMQ<!EZW4I9piobsn#~(LD7Cl3RixX
zl7KSL9jpBSi>`W{99Oru@M~9LQ%45uCoH`(AaxB^GP%-Mhk=NDG&)xERQ%$7JtO+c
z7@Dsq4sbtnsUiB73l8trv2$@M+LnmTitxDx4h(1AT(u=)e|fin#-PMnH7j(it6QGv
zl`@Q>)CF3@B4IkkRq|%_wsC)+CSRSBcl<`R2;9xf$7pmKc5WIsC#T~l4oW6XdEdq5
z`23{{SL*SLmaE7|J7GCW@xBfn+hBGrQsWs9?1FT(EzftWG}PO4>D1U43MyOQ|2A;u
zg@H^<YjRN~4b_bs`!{a<*GH_NgGho>wb~2^j2htAslVBp`f^ihLx4%WguAZL4=ye$
z=Q6wdBL;C5h1$xnDR<pVdG<xPp-R>rFU6B6s28ccn~kU5|IWo4j#w7DIJeyCJwPtJ
zFns{9tkAIR)?@LbTDA`WaytF$I@kFiXUU*Pep0IM%~M}rKP#?nNb5CF^k}fJukp7l
zTjdH7rK&(%>%7`YBB=|vzI8pYlGM9*=DeCrbaea3IK!>tZxq*-%y|p2)9+73RbUf#
z+v8iF%!yk3u%ft_LYx2DK)}GEV$r3FiH(&ttG{4CKkoYvwlUV_?O*q#h-Ii<;=ljU
z%K7+le*4cU!D4U>J2tQWcUS0%Q_eFQ8FW%gjr>+Ap?UgftLY}KDB?Kv2bPwr4-<0s
z;%BSmw$JRG?tCm(H|G;J-%A*YKbvME<g}ZE()i+zYQf>c_@39gIpxo*>08tR$v*MI
z*7GYi$l<Pkj6?}9lj^r+)V)2gz>#YtosqUdD61CyYp;BVUL{tK|8Vw}?`Mg5YSqn9
zIy<(8<kwv){AbUgqo?p!sF&W=HAKu8w2v5!ZeZ&ySB^z8j-MVV5wWAVg2Ny3eQ4r~
z%71$OHoL_5(nvbGpf)*Sd!!%_y`RfP%Q|51xy3prBWzd`x-}~st+cs>>BeFo`!7@X
zOh=8UD)t?S4xt!@MreA9JvW01@z-N$K5Z1vvH4xqP<+~I3C^~#w*J6U8X53FzRcL0
z#YArJA88NlCzmC`+~>Uv-TPtU+1*jw<!N?F*A2PtQp+41_S~0Mt)%T%r>rzryH#a-
zK5`IaOpt!A_z6~;z(6)nb(dAvv3^ZD@2ZlS?O$s&xn%h3dR=04Q*FZCw%}Vsi&F+x
z9j2!3il*7bM3TZ;@9!|8_dH62oh@uAh3R5ez6KAK*to_IkZLsXoa^|EG+6Z=6Hz&!
zTqP-G)ZUv)i+RA;Bgz;sVCn5AaQf^y3}gAo&FBoiZNHg2+|?g8qkAHsmWBlP?tD4J
zZ^Ved_zn%BQMki#ZpBb+B-dte_z_0(tZ0}!l2rDRo?M*`V_+{jHqlek-M=qxfg;Pa
z+;CIhu)pwSiGZ@lA(n(BGn7KE^Y?oN-0t0Fljr<9xu@)}q*rYdz4~u0z!s;z4BKGO
z@m2OCmgZf#jt{GHs~TbjM|RnSCt&9iz6lys1l)46gT>l!7HT77PQu)myib!q^4a}3
zXZP^XIFWQ+2JW2r>(ANOw%2>>5)zW<6jRblZ=`3s$PfN-V)%&Soc?lEEmB>*pz0y~
zW^4r$&7-@Lt{v_xE2C8{vo;D+dxFtw2{~&o6=aR$QMV3nL=$m|3OQ6iB>Sa(;}3^I
z;=Xi!u$oBnVqRX_r}Zlp`!)e-Y*exVeAo|Wj<K3Mdt55PdQA@&f}<?5@8?9SIYz#f
z)lV-cmlwxI(Z8{Dm8VsfzwILHJHmE5Q>KoA|6cL$I_&_5=L}v5QUz?Vp3M@R6>g?b
zve77u3+3!Ng;Ud$#0Bhf;TF_;jt^Kx>$J_2<vSH(vMlTpEfpv6-O`0v9hGsTs4&ng
zwE8e(FLVEK_L%Czd9)Mb)k8LQHqWigzRdJl(!wRy=n!8|Mdh0Hgt@#O4Jn2UdH9j^
zX-pkhvSZ8ne%rr6s<3cxA-sjA*(=$vIem30f^VPnaP@esf2!2D)eHM+<A*8RH{+nQ
z;b#rhB74`r$YuuXEksmh{u*lA)6o~iw}y}BWc7;RGm-n5PMtFG4LNr_wx34biS%jh
z+df&Hl}QumkF&3<lO`V*e8}KcsMJ;`-;>{9CG1op5n!4HA%*5=Y4AhCG39b519Vco
zM2<Z1eEgiMs>l9&oX3;Jj=hyOq~81RT+9YztGF7gc!P?8bIHX@Wbd!GeVca56BHB1
z9d&<>LiFQ3?KbJKfk97&{7jm6?Z>dl3P-wE?s*fGX4x7z;U8;4oS+d|=6#_G_aY)<
z)|-49N3T@t-7C^Mq&)1_m_!BKQ3|~pM#+Z%NZ#4qzjX<Z^hIFq5E{9C9{D{LLSm?s
zCsMaD>6rK!z-S3)Nb{eQ3nCcIuLOL(|00j;&mn_e#01<TQz@|&_Hm!I>x$LMWYyML
z6tl;psM=aO2Z};u^^Qha4gqWxnkdGpFOdXqwUodr(IDM@NpXYseQpl@Jz3C?{n~_o
zvwe(!wvNoFwXGRSZeB`_kb-`@HUNoR?#I$(|EK~Bh&gQGtywWm;dOF6u41fzXOIzU
zZf?CkrqQw6>9}b*pv8SySH)MNt-Nsd@44%K+Se!i139lr3~b@@!usvs)wYBOG->3e
zULv#d3{9c&x2b22w<(k1eit2^bEdN{t5m;CE+9>^yXT!HTK*Nqg=-ZZQ%>P1=DhZ<
zvKsM)nC3)GZ~18U4g1)7%YR87aROT6l5E>=uNYMq!f}eu+Ib(O0-2-G=mEoNCytCn
zHdc&gn3yWv@3J%R#}>IdUPp1+U*{)dCG$EYMZ3Rq_P%Z9B7j@n;(c+rxj03>@vE9h
zYy2esSs)N@7a%I}rdg01j#foEW4I2NeOjvP$!Z1J^2CzAuRIeK{vD}@E)<L18=@UL
zTK7TQFyefx<ZSPAh}XZ{7`_s1@BID(eUd<=Pi~`4UDgM^KL?2+CJs+Sq&-&j)#*q{
zN$XUSDc+Ko!bY#<3$fL#6`yP9@(eqaRGpkG)wdKac4^Z(Tv)JA%eaG!4psX_^w!9Z
zmVLd&O}2;Os8*tah^SP9`EP4!E<$U&d0F(1-M6fy1VZ0MGDOjj{A1ga@5Z<M{vNIX
z-6M})bBc~RtoN?Ww@(z#s}S2+_hg{3SE$Ky6AmCjT0ORsd8=e@vCv`0VO{pe;lSl&
zZWaYNZvH-BkfvK;Gh%GcZq{jpD~wkwzJ_+4p7K3$n5Ir<C7oZg|CWg@!{ITf$A&Lo
zZfE*sZR718<%FJc@gt^2V%e8Dp9SNYLTvu=c2_&z+oKf~-kOWy_>0XzkX|I<W_NZd
zHXqt4AK2(b8ZY$ua`r`Hyn9;T@-FE$ezR;{vDsJ&d5mFY`u0LDZ@aPIwvqPmdN8&x
zFMIrzT%qJpwSZ=(qa)?j?yQIZ5?>@59@OhgaYe~1W%kbVX16<6{HR*uJaYmm|5YMA
zUJ$KVwf#%SJY!KQb1EgZZ#SiX{wOxs>`Eg#V}Wolk6X?>foIfCvvlueb10R)fmf(h
z-;c(37?uaEpBB1Sr-RPrnEQUdma$*A{_eA)`r6m#+t#60m&IYhu{$Y$v+dEg`xoa}
z5qhsUoQpilfXlU1tkZw{cBhBkggXK1NsJvm32EFu-STl~?eT)X(S%Etyr?~arQ2z@
zoUEiIiNV1^tteG9Ntb;hhF-pu_xW|XuF-8*$)pv-pH$2r5Qx2x*Q~zaLqL+FnWnm0
zFk9HJfCQx5-eIy_9O;Zh$b7N(<nBe%4{ftk9e0^z$r7lZT|QEhKn2A~v8%oBGfTQW
z$USlB>*M3YspU9HM;16z=OuF{$@Ka}UP@P&3V_n|3?D{RynLC*m3+jbIJo3OL`>Y;
zS!YdP`&2_?^`R1Jbygb}S<LjKM-<-<H8O+Zl$6~rSr)f+h9OzKb!ERZSQhIVoXMSW
zsS*t%^7Zv4I!UZ&a3Oc*$n1fZipskuJOKky@U6J0?q}6TFGuaPCH(3R4&6OR-}%Q)
zANrb_CZ8oCVWz|qm3eH?^q3d7)CDag3EWN{5o*3qOiUystlr+JIGr!jZp>oEb8s+Y
z+_=6LJJ%5`u%*5JK~KHsIU#7kI~|X549K&aW-YRwB#?<WOdRZgPBH#UOMgo2cW%LL
z8Xpnvok~<e`QGk+iwTN25uz)0wSJwDsf8!|$cS+4%Q5}|(bG*WFB=!E5fbMkNEzy8
zB}lIf<yfqWe3R8hRna)nE(tNlJN_yPy-GnzZ{$SuN5vtMB;h!GYl-448PfSNMd8_Z
zV&b}+p+^o6P<$?35&Rm||2i!%p;1!6fqM_wvpoYlnFlwfskbLUp;1bpdvw59ySVXh
z%=r6t-zg2byRSCnqj=+cZdY+X{Zbll@uXcLg6VMCyH4FwMTJxI?;-Dg0k8b1vLv10
zxPimFuBfdU+3n+NhadHK3AohtTnw5Rd{DX*J?fWk{AzlxjZrN8+n5!PO$%M=B!bgr
zgBv*G#NL@GqrMyCJ~tHiRV#0*(a>=st(c^SnKJcTpUI3{$mS@?SP}7_yU%@WmrY-3
zTGA)*x2*^x{KZW}?%gYy!BOjN?CKp_bw5gHP7FsCvp@AUE6?|En2(<yMnAr)^<Q)0
z9Mdc!pMd0+hWs9og?h#AIg#E`X^CX?J*&N$loVI8<`iT8OJ19y_S_9Txm^GDq0wvj
zu+$g8u5K*nz{z8JhJQ&Y2%2kzMhvs!iyk@hSpp+=<%`kLMf}!G$4S?6+MfFxSp4{I
zq{AwE765fe&!1n`KFL8jJ0pX-$9Vl_{z)xII4#1$LLT*Y!elovP#<hp<4Jj2LTt__
z7R4ECvtDCuyZc^=0ap|gT2<j~V_&^>sCp*Fc^dZPcjt5D2@pH1M$vyw0pS<*KWGH*
z|6zw1{SSny`hOr)ZfyS>Lbaj>XI=jNZz3HIg#7!mULMXz{`>O(eld}bp<nslT&#U~
zbhH(cI*TB~DjUM5q@?89_I7zCozC~kXMpXKK~d1r-4quWJF#x&Qe}vuprCl-!2Ag{
zyrcz@hKcRkHHka19aBLj%u#L3RipXu-kqloFhvS|Qh`5%<uI}F@nNQ{O-f(9AcR23
ze$B6Zptc?7+=_FSQ>*c`uK*Tgjyyu8ZHGjINL5vJsKO}|@Df%?wyb|LtkxcqI#!MH
zDypOA`$%|oQfCV5onG4CR_whrbtT}-De(#Z6HKcy1nKarhQktxi`RK<yIGD_mi93t
zvZMlZ*Up_pi2d&LJt~gm)Mt^!16eCwm+c&+5V>XU>*k~0+oPy7phNs>xsh1%`8s0r
z%dhU#h=c?#h`WF?)HNjY^Dz1Hg2)4~-vB5V+s_RkXTX2l>r8{nzXtaI-KrY+6np2d
zBa*~q<mGz-EXgdpG*qf?X_?hXN$^R}$SA?*XxBnk4N$|FA~X454nh@%#jzTFvcUVG
z$pN;Z6?Gzs{`>dukW`zQ@z7Q^4UKU4><aV$!}>p_{A>M){}--8m<vpe+#;Vt5B%Zx
z5oOT;QWSPs)RCqn$g&4$eTn0&900mF7#Z&ajJ1za+aNiopjNN2uuv=NR1%qByi&;3
zyIa*mV{X$Sm4JV=senH9w~CQdQtHPh`sM@Eo|T(h8QwoXT2&5xng98-V2=FB2}W?z
zotwRUvyK%%28sQaZz-xXcyGMF%&Dg_zgAMh$IQaQ4q$lsC@xJ21a~^Z_*wn{3A3>C
zFO{|kL>70PQYO9U;;r<IjL1q6VDVa6S?$eyP<95gIUe+US@U##)JpzIzs@RP=^;mv
z1G<L9ejZQos8#^I6A}_`tTd1>6t!`wKYR8mC@9Ez=KBR;UXRRQB*46{ua`K)iXWq&
z`2sLELn)rUfd`YLT_Bu$&M!jjVC6nsl^w{P$|9?;012M!=ZBb`nX%kmn*}TgM&G_B
zU7ou}qa#+JywU(cO6RmfxZ>{PBSJ~IQvl=)>(!}7R2slE+5h=R5-`$`6##l--J22O
zbG$!(+k?uF$!fF`iAu96sL=vX8+wGmL`QIF%b@^Ckq@L$4p;yP=HM{h6eH-6Qt6Dl
zE(^U~R-}Lsd!Ye?or#<BkvHFKh(xh4+95QW_!Kz}D>YC%cg|1Thf$P|Pc0J~h`qCA
zxPgL1CR}^;ZKM_zV*F36T_Ne^D|K1YlkMr#Hnw%(N6c4^R=IMBi~oT}LlgUE5m}6I
zapgoYR2pE0qlQ@kla0cZ+6)NikoYZ6Rrw>l#(*xniOWo6hvx!(Oh8=+=U2|NBbSCW
z159-c>HUPm<KsC48~%|6W=I7dP5`W5OFUYZghV}zUK9^0KP$a}Sm8|cT6@EUgOD}H
ze`4ZshC*~ocsNDP(cXf`JUWwtfq@&mb{30W=v5X67|AfE*4wjFgA}0>d+r1C6v0kb
zLEVvFLG_xP|9%y6A;TiUJz#eO!QePiPawK8@iOSmo9>;}X?{RGI}VS60Rt=6arE-3
zB%_wj&H?xfv+Q!+aV-tM^%DpH^F!dUrOzhs;i1y$T;wDqy?Nzh*^>={ZR6wPDf5V(
zix1&R>n0~BR~{ej_hl+GfHZ97IR_2o5E9ZYw&?69hXbRujlDH}52hLNxF{vA77j0=
z(%Rb<@YLMLO>et7%STdHR-S>{!(S!lwnb<pmti)&s7ID|fzf5{bM@@w-rMm^X!sX^
zD#GL-i~y3<U~a9tS_G_e9-gN4$mc?$2!a)U2nk+VE!DafB+%4}FE9I`cSc>uy@x(K
zq=sI-+jLLfFN~GiY}%!&ufSLwT*NTJ8%T7Ob1^pIQBj$$WA4(xR>c4u7*=%qHD0`Q
ziJ3C&*Qp6ltpcMwA*VS!TA90M12t4?lZkf$EY=3>c5i?DI6s6|0FQ$Sdi%EIj|&9n
z5=aQ?m0Hi9@pD}oGQ%5|98iEQJOST4TSm?M<TVoxTIeVok5r@b+hZ0LH3Uv~M2Gof
z0_OTtl$S4C|A-SprBNAssRCkL&BP?>T!V^^&d2<^BM&^em;L5V4B)Fd^Dp~x<B5)E
z!*8bGvz!AnV!n0IIQ1#euq$l8e*G#k%uG^5u)+!69LW?yCxqJNCtz@1nUIuK0}u*O
zZVl~XpPx7-$8NO!GCk~1jRBEWe{{Id{H8#IK#Yl%Rfbc)EF2CWGK&~yf$XE<iKuOy
zb+iKKN2QrI1n3%&A~?dmUN-Z*Jv$t}=u|a|RJ(*2#~+A2E9kQL@s8~flPrLrKQ#6s
zvgDc9ts+kf)|El{l~vo{TB1B(??~)dFj!_A3o!|w0~#9}Atxb7!gD*{M@DK*p886A
z=jl79QUJ8~T)Tp>CIo_ABDhnYKMB=*{qqw=!%GQm@W4v}7$myUbsv?6k41n)|CFtO
zxO@fRyuj)$HEkf~G^*AiA|gTsNiTTq&PK(E`_?i===qWQE${qF4P%mh+#Fi^`65E=
z&6_u>;F`>`VWL|f0s3<TvOap}@Ebn36Omk&nW`X^80Ar5c}XzwvZ$zt&#EU)ZrFJq
zBMXt3&iD(lR~Imv_`LTv)Ov5lEtL)1O_S2XJBy2v`%C5WD6lHZix(f^^D`<UBK;J-
z8-6@TSw;;lg?RVlv=O2dV3mHIkd!{+iPMmUtd5Tk;fZUk&vo>b+Gvwf^FD|&a6H4^
zgkXU2QW7PH7R<NT_JC(n0-<}X#?xlwSGP4@Bg_(fsxLBY<MTc8BBc{l2@~7Pf{6C%
zV=oQ^AotDLC}Vbhe$Cz8-Jp;V&2qa@=cUqt?E8UaT2ZI#rGb_4pOkz=_8<K8|3}@K
z{i`eaKAx{K_;je&yQ*(?ry6Y0?ROX;opRdnIe-2nSOX#p^CSN5=7>`#8zdOT7(P`Q
znb%WOQ+&@RB=AuWpQFCKJY^P>(U;yY1PRz#7=Cg_r0wkTZ-mF)vDG8h!XCOTk1Yua
zRJksTS?m`tlN)=9z|GeSfQE$YWF-Qliyu@1D3I*-S}+wHfNX|7nZ7jdh+&r9n|MhM
zr1DplT*Z>CXI>j}MKwlXxS5~Zw%maa`xW8kxdb(c8t?rrfZkW-a{Z~Tshq*9d6gze
zptEOOVZHL^S{t`b`^Q_>-6{T%v7f|btgWvj4|W`E$7>DSqPVpobvJ6_%X@yycuXn6
ztQs^-!V=^NV1nggd>VKt=;`wzo<e$35*T@Q-!SSlyc=;I;t|5nn2nvi2YP~87Oa!)
z(r+zk$c<yZ$KIpPy^2A>!RpX&!pQlw{ohr}EiEk(^i^hUmpfnR<8U}s8e~u_;Hg*c
zOa()D%76&5VDGyhySBFWLs#C#rA#Z|01f$K0d7HCLE-Jad-w2gbj4_1DQMYxrMt1s
z-a$rYrcKjZ>iMqZm-xS@N|TEoAFQ{wwh9+`DD+|~kqdlwBZ7_JzTtB{WT9|y7SVN5
zW2X*&umKtYyEyFU>E^ID^I|l}q=Yt*0#oRM84LJMY;JO8Io>3)malvPz>M7L`u;Mu
z4zxah|MJ>w8<55YAz(*UdTsw69jfsxAM;vQgm|b1?hY_DHm8F@61$2~yN*P0V@MgU
z=sU8qvT|~AJ?hG$>GpMpWf~H0cLE!dL5n(WFO61d1EIVRA}~IacyA3_1igNp)vB1U
z&hPG{*oz6G<luYs?X{Y1nf-YF(s22LKb29}_PB2W7^=2T*r?rrqQGbx_-%Nl*I<#E
zUfI)VK(^zrqotIo4nI;-Qt3kx7`3lUE^SLLQI(4Z!9AMWcxyt8`CWj4UlN75xcKJc
zpgO7qV%t?==d2-bE`!f1jMAYuFNqxu;ho<3R~N73SV2ep-ExJ`o@H@yber>i@{nip
z)M}VfC={RD%7lkFsyw4}xoWvqpXG|EJ{SZsIlTc(IHze$#=|grLzWTIVHj9|LXm;T
zuB9ih`Cl_$0K*dzzPec4=}|UZZeP&M;A;>oU@vCjr;`8<ihuGb3TkRJ^rp6SDBb&B
zTsVxVBqg7#sj2Zj`F>uurzwQC6a=c=ejI4<gTup8lP`pjXIkK%?aCMSm+Q6_*e0EY
zk=S@wt)I=`-!pD5`yT59!5;rD7OyP9f8j2iXCt)b*vYOB<mx(oDX+@MH0^q?wcMb*
za3T5Iw+EfONa6UQN@Nc3-IxJWOT9N0%-t&L1BqL&#-lWCN^0*%Z6egTe0C;J89Fpl
z>J^)ZK-krp+X2^Z;Zr=^nGQ>6^Ep_Rj?U1Xx0q{>E_I>BKZgQS2oT}ZlYtC(m?L)I
z1P9-~p&0X|7Sy-7hsU1Kezb^t5nIdKpden*j)tapquR{KXz#R#KXF<7S=fFQ9v-fS
zd^%oRGvWhRAFT0oRjz`-8-E;fq$zrW_#6*&w7{6#;|g0cI>CvG1l!><O`M0Lqoato
z$V&YQDlaeFIOEt3bNah3xg$QI4C247prf}VhBIcWHC>FL^bq~--Gr6)n9vvuH!Y6A
z@Hnb%OCA0+-Tn(?8H7syfU&P38yg$Vou@kJHqfEjQo)yTTf_F+APpkky-UJiFpm4r
zgyZ!}EZ;NXAX(%;8`DR-S-;Lq^wCpWtuBn<40pjlozFY^Ew8&TCVFq_`X6dEb#^L~
zi>!dG@5|QIoKG$5>)(xsM-0$0t>3jjf9jB4;gLX<K*-|VUte1jNp-xT@DX--R%z>W
z9_+4jfGcM^{cR{75{+PY#-&GDgYC9<bo4=x*uaNL5K^F?*GqKkd}`p?wvkH04a`R`
zZ;p8`Yl2dO03#v}?tk{|SyYGlhYoY~d-q;of5Lb~OJ^WW_QMZd*EcX?(f;}S+nY-s
z3wh;y0EH{H9oB*oN`fB1HSDUE?#MQ-mkQH`$`$mUEk*kHUY$SID=?}N89259i(Y*?
z?z6uXv2}@op`fd)OKw<06KRJJ&J&7%2AK$q@8skp4diuVQkpdJgGq82<)hJPkHgJ@
zO_2C<sE2Q_BwxCr#LCXz>NGMq84WJdixXL1$i1mj^D8BoU9=y162gbJwRJWq5pBCz
zY=Y;H+bNq~FboC6#=4y`mG#2G!v_yEJf1>sU}~9lj2Luvc{YMaKJMQ-cORm5AG~I_
z)u!yLAceO137DRQl=;~tz?4-?%*YA*qtEzgjY_sra!~wuw^MD&nnUFFxl}#dF(@36
zoONB`;OMlMQs~VsUMNa2hA&8}{gZk6c-_~YbZ!ZZd#PI+1NG8k9~216Dh7K9-C;b5
zq=$&q69d(et>2>@veqD|qP>}&7Kl%sT8MWY>judg9vjO6(yDxT32ln&Nyj3|a@F*)
z<-(61JrWtlRwgo8HHTg;Rl~0ZnBe%6K}#tpDc#G~*VpBjZ2GbzEcP47#d8Z6yp9jI
zIY#Te_gy+&S`2cSl@dhIBO?Z~JyVUr$mXlAcm;(=StceLqu(c%-a<+!X6LQzUj>c0
z^3)eFn-OKB<@Uzu!8P%DPx4LSdWdv<70u7u{b&~}I$Yzq%~i@RD5wKzKZ2=dvtM7Y
z&c{2%01LJXc{bm18&sDSNZf?uV~@hTVl0|7I<Qm(;#5Dra?l}Xqiq{MK7X2=Tsu+B
zn@M)LW;?-W50~jU4Ebi%sZ%s!3%N!36k-#<HWOUD%MUNS{CGj2WojF_ODzl(40NYT
zgHivA*6-s}oSmCvhv60D!?#_~0@3CW>f0F##`uAW#a$3Uy%~zV;7exZ!=h1D23!zO
z+2GyWwL4SpL^+}!8(o0J`$bJK?^v+`{byP4M<6FF`<{vPs&GmBJ!9yr&8OoI%~#<b
zYruCz%9k)`b-hm}&p{bDO9n@3rh(1;6=e}zPqj3`FLI_Ky{$amhUCKFwelS60|l+2
zVvDc<Qd*#6_iGF(J3E(H>^D*wYnYpxFHztrHg326Y9BC8JUW#X8V=7x1r&(I9wDSQ
zsRs|T?8l+*^2i_ivFdlFJP2Od=-QzYt5g`LHLo8-aJ-Kn8iCoLYj|gR<?QLxz_$zE
z+gi%OXCm-VbV)4?8d@fw{p#1Z*95d)W*GDfKCeVE?e6a{ER~OQIpubkH=Hd4k#wQ~
z&j?t5ODBq)55IWvA_Wc2e1+|;Teo6_k=Zad;j<ko$&Ra4gO@YT!krv96ZmZ+Z=4ew
zP<}HcSl5T_)6};5{_Y<X6NrR2mu|o!EEuXI7k{dPg=j+Pfv0&xc5R(wzliVVO$A6l
zADBQ)%5`^l^PX#PA~SFQsG3ubj~*+;_|jI=5WVk#0|cmouZON94uSo-x~WjBq^ZE)
z6D~wxEbEys5NX3vs1(3>MpYe`4j8SU44gefM%EAiV{>@?sBEazC=122V<2vkgIm`(
z3^28vZH+L?Xcp9u;xa6;?ql!#zViC@YscYu(R<YoT7;fe81Sp}Koce!{|#W`7C+*;
zVdxwl7nh44;$nZq57qhBK^l5eJc!9Ryo+2OGi(SXA9#I%5puoz53xY9tJ&ZuR#4{K
zYZgCB#!DJJLl>rXV(!>xgZQ=XN|K}zar@-4wWw=l9ANq@Igq|5N%AG1{g{aN!D@5s
z_wOp538MU_Ur*wTm(~#I{hI?{y|RR)q*wIh&0I%pW?C8(_#E2z_=x7V`CELM<n`;*
z5Ga>lbjE91uO0NinwT)bcQ15)-$;cMzQiL$M@CA+Cj;>o^Xkl{ndRkW-SgaHVznba
zct-<eZUu!Iye5E%rO&8t0sVj=^*4<?7Yi@4n(x?;E$V(-TA6IHhS<|cnOy0zG?Wf$
zRUQ?|ZJfVQP&YOeg#XNP0|h?NW;ua|gZfGl_87`W<@tC(RekWE>U>gDo^J6&-@^st
zO{n9rL9r>rs+CU>QshI>%IYu-198A4O9S-al+;wtKObFDYRAi3KN{D$c52bpZ<jo#
z-rQif#R3u-Diau0Jn&qip#h)Q5jxkQPyvz*pT_h43^i8t-r|P~5+Sr8s$oQ)7AAhA
zW9+k=j>jB>E^KUSvYx1aP9yG9jn7F*FbN`$AEH#pfDE)k{#gS9|MVki{&R2e67hfD
dqyy&oRN{-X)4ftW{P+gSN+~_azi0B|e*x1IS9<^e

literal 122702
zcmdSAbySpZ*fl!fPf*eTDFKykDQQp<aOfI3RJyxU5mAsX>5`5?x)D*jW9X3X?qT5E
zqxhcn{&~(?-yh$sh4bK=dAQ@cuWMg>@8^@Eyc9kTISvE@!H2$lp$vgscZEQ%y}p43
z{zn&cP!GIey^)c60p7vuRJ_&)yxp{Ysp$ZLPzIoXFhUo1jUbQ*5a^3%s;-F}Q!bvc
zLsaYTF=h3b2g83iJnr67925V*^-J?OFMm956W*+)Uf!v+o;M+F_u$>T99bj3uE)<c
z-rjoo9BT|&G(-Iv@h)sKVPT0LzASb&o4n1i?creGD#Wl%nkb*G@@E|2AGpDR<G*WP
z3r1IrzrTE!QQE=$d+8gSK70M@BODS<xL22Yf$vDJF6)S{b6j1zGU1@-ihk`V6V?5_
zg3M<Af4OnT#wujGW!2IA_YeQ%khhWxj)F#|V}@LDhCF*hp%q=$m7TwBUq*Bpm*Qdh
z1>9|lw0x2Pu4L&WvopT?$u~;AjOaHt^f8^kPY|%-ECjzD)@>X)xEe)!dUE2*k1yP9
zUcRS^X`)9%Ti%9x$jJC|z9o3y5f38wb5<DD^WWi|qad&A&MSr1zoXBIC7={Eovw3B
z4#7zMWcwdHh{|o@wyx7K3_ZA6TYgm?or2WV2d8Ie?t%ouIqB)o%gf6nzJ`U}3TZQ&
z@95~DO`0U>bJi?Ga3kNr;qZ6w-U%n(NLJO>&JPU4@ec^7tfDFHtCN<N?zx{!LPoaw
zn`fsiJ^j%s64^O3Gc%|(`wRwy<rNjhe7$!(w&zcx*M}(77NNZM^ltjBQ}~#rwUWZ9
zZBp~^uwQS&;c$+b!6!ZD<;kZGGWz=ZIMYwI4)!s6ajg>iO{2gEX!RH#K=bnRBf%%H
z+fbN%)94|p6Xx*ky%Q#%ytUbwLZUTr@trBl7ZK``_5W-C{*ML5|J65+8XLr@iw0ph
zxO(7-$MbG*MyetE#Abt#KNyv2g`C+ILyAwdcwi;sA~9)`Np_l-F^{kz-NElhB%-6E
z1?}eEXB$*)zy+w?1Tp(c7F`7)kknsuyqK54C=AoP-jTNA^_7J=If@bz5_hxAzQo6q
zk8L`G818$|=&HN0yqrBd3qsm9?$^lN>J7FJU?Al^%%!24k~Ai4W}V;&-FFV4ZEtOr
z!pdlB{xl1Hk{}mh_+(#sEJlD*Ub9dQ8qsffb*L=W5<FQ&xkz1uO*;#moCLjWskZZ9
z7v@Ly8%DGn?l}@wj;>pF%%l|V)!i90FTnYXqhx4b@b1lX5z0b|s=npR@o}x;;bCaR
zISQpc<bg#^O)a{PS*X+L*0RfGT&mg7(2yMRo2&Gb4a!VDiPbe&k|tD!V%e4-M5<&)
z|N6|=W)bI!_Kf?kdZrT_XTAOGb@Jb{F3;y$AK*+^beR{gIW~9hV;1D?uZ?`1oRTWr
z+WaM0WnQ_)Y|&FA(%|B8m?9ucqAR9MALKIK+G3m5d+A(@l5Lt8IKzx=|H(oBR;_4L
zROlKy`uik@VZ&?d3}Qn)y?ietWR3Od4#?rNEAm|0n5wY&{CDCM(FrlRlVz+{o;Gn=
zmz&N`qyyc}?HG{DpuY*y_sLh`#uwC6r-X}}BJ_=6q?xi_I=TVV^rbelgnW*IFjvE%
z3a{O&5*H3cqm7%AN^&)nNuBUm))w+z!0yhyKWA|KhITM4WpY1n?}$o8=UAy`Z<Xip
z{(fm`r!JI>lxWzlm|plG_`~%12U@>-qJlEm>YLU_lMr8OslN&PDaE(*YPXIluNRz2
z9K-JeV%s<v^YqqhLTjX&Z7yRM9L=YjZF2K+>dgEwA#1_^MxM%_|5RI4hlzUTL2&gp
zcU%{K?JbRE8Ee(ba50#bTRGx<f1O_XCd8ESD(#Q^vQfh$nyuBhZ-z2HU~S4IyOt$J
z7x_znVyPgeuw}gMCI+M^^=~~%#UE{a1dA9l8nt5Yy>TO0y;d~Ru&`(_4SzK_3J)?%
zaCJ|u>hVDHB>u$g*_9rg5NT%%wdcK^v`_&dZi<Iwkng%zWv0xY_<Cl<*Y6)`D@~G;
zm6R0cj#weS$g9d@y7(u)U-49nb=j6k+X^Ob^<qM*{>@a4onKHkLKWuOwW&EwClK*9
zbw#n#1OqbpZ~a)7N#@{v?{7p?&1C1PTrbSGKQ4@-{dMGc=|~8H9A8z-zHu_VwI>yg
zw5xr`)YZaSMR^)hF6<HgmdeY&>&aOtA>Z-;jTYk?oPvoAZ~U}dUsZi^{Q7{iH;P0v
zzjMgujBXEq&t~C01fs6^w+Q!P{X8<w_8#O=vifzi(0iF)fbs~duvcC?eM%-dc5Fl{
z!&|iQlQJfR{3=EIK6x`>Fv5~9{yAjpTsnWc;aNpZo_nEC&Ag1)@!7DwAZ7esS2vE_
zN7K2l5Z|+_0-qYZgNxs%Fzv0FcJsDi-Kh$j=h-78tIOx|+4G4L>h~yzYY4sb7UA--
z<rU8jsFKsXx!J9jh9JKCoE{(@p<kw=;q7)c78eHbmpj^mb+6%Sb<UZX&*!5@O(Nc<
zM;cZ9ms4|4wa5L0QOuxz!LfsHR9_{fTB5#)bYb-b;(Ph*U%1{SB`Zh}bA5qqoXB;u
zVMFv=kqH)Wp-L-o0|{V_WU=O>B9RN!<STE%_;J&&0*Z}`bjZrm$BAM>KboP#$#LQt
zY$3KoOE(|vrUZQ}Q-cCF*=H;ThU<_->?_D(rSr<L*!?s0>?$b)Lh^6DTPXEXil%~|
z<nkcj7nG}Amk3}^v%$u1c>S66nRjvsHiVz>U%)?*N|KuqMtxHj)93I=+ANqDB(WI|
zTozMT&F1fR-6Ehn*c?57hyfva@o#DlmdJ-FB!a}Kuqc|=|JSkVn-nNc$QErW-BVa*
zHFMi&PU68Lb64B2kU9#A%F<6K>z(@jdjYJ@75aKgQL>!C7Ju9~VJ~<;aE<?^AKXpP
zZb~Z2VPd@F(dPMCueA12gtPazXZ~c>HS47_P)P&sY%It*X>#}<d}m67b7W$1(Y)QZ
zYMV3BmSVnVGlDX$W`Q=my-n`f?GS&^X|#v`S+S=`w%cPN$ygQ-Qg!R!gg=hC)X@e^
z6)iN4n2mPZbeL8QS?KB<_mdxg$0U;zl}4JE62KK~{#Rh(k8=9dckD|qP~XXZU0F{<
zhKH|3J>gV!rf^YDYxaBlV)(!Rl^8UR5pST%m+he)t$b~nWc3Wfk7S-!Njn6p+Lsk`
z&SOAy{;j6PvhB*x-@@&$6!*f`b*Sd&NeYgbROp}-)A|UFRS3lTU$Tj25!8B+-JDij
zMYV?;VrQXdIT9Z6b>2*hoQ53oUGCqpw7-sRPY1>L{mHr)=Y~ueE$m)&tq3k;R<tzx
zPdxxBb=i?022*7x!^62j7|}FZ9Fzf#0B0~_p;&vWXTEp@3aUipL@5Nam2nl9{!E>0
z14q3Gv&%}Ms90}Rf?DwiLmo2Xk=pv2;7vHj2%fsFtVaFPFJ~JDXdYviqmcIECswHP
zE5ncf`##0PTIGA0*U!8WIAiOU5wtU@FRy>71mIQ`w#&J=(WTncBU|V-PoZl+VMKno
zpUZWccscK-D4WxDuasH-zLMs&xBpLP2LS`;ab>7-_Un3`WDJO@@W07JKn9I>n6c=X
zw$q@e7CVeP39Hd@5BhNz`SOEGwvnwM%+WG{peQA?ex|;t+{Ux8fDP%?oqx>onCql&
zf8O1t@7pHXc;k{Te$5%2-Kc=ILr`;v9FN?7UnABS0wE_M(fqTB-<v@x=0R38vUs$<
zmS7<4hxKla%vvO?xYjmYGP%%!-F%@O_(A75-B`3#J)9DY*`o8Am*cWbiP}igv4!k8
zsyqU3ymZbenuN-kX&G@E+VL8$QhU)}^51DS&oxM2)m7HtDdwO~$UH}l1#WXn#$LKd
zW70&hSJ5-(WfYxslmBug%L0Yl3$+liJSF<ghoM{s8Z)JpVWVHGj2I8ZGMcR2!#A=#
zSH#LtQm=P(@k3xj;t8%22bH6v_J@IcR%%Z*-nC2$sne&NQIlCz{AW7ZyVS`+D=D-D
zSKr#AS)RJcBBYAZFK!GB+3Mr-p(yN_Ga$KmB%kf<wXbe<vf~XM_}N;dV*19go*Uf$
z-oM+IZ3V;Wpy=v3llp-q+q;QVI_|8B)@(CZQ>^9AY_`xcO0VIqx1gP2zUZa6&8Eh|
zXY&r+o^izQVkRN0+<PduoQf`FA?<2}5QFWOF2wY8RU7%A98IOH4$J>s?y)dWoz5OR
zJ-3%=-MrPeRDGH4m+#!EW)VF$vMf~@EE_W1Y>_gE<U5yhTa_x`3=%V)c-Kli{5tGi
z#-Bu<cTj~xXrpXp6i(Z|v7V9JJDuJz$YCcy14%WzN~E(7$8rM<E60b5wvD*r+1PtI
zHP55_3<}UyNuV|)2bIdcJT-l1A=~4gXp#8(K>0#kD9n{n!9Okm_NDq1d*U<zEAN=t
zk7xDUx>1?pL&#H^pQYSavq)Bd1jYFz;PX(Mx+&XObs=`hLoWNZh|>(8Lg#CTm9{J|
z=C(F8N_iyslhI6(#Q*1q+9X^^U)jIl2&>7{0!NplZBtX!yD8%9ct<=IoHssLnrd97
zUEHH42}Dxe|E$`bTW%#$g-qQJ9%@cw?PPD29$Etg;rg!-2-B7r6woF)ZE33$QB6CV
z8-GS0-D$t8E}8JtW^ztgl`Zl8>8~B>tg*{ci`1(9{ggO1)LvOUSdg-Rz`!-IDHSYZ
zIBll)X<4U%GKnZ1bd3*nr~Q>_lqm$V{q%1c@x{hRt-d(LLRwsoA3Iu(jNt2~MAtke
zqcG_dV|^Jr$=Pu84kkp4>ndvtw~6s6MyAv#*$bBM6{f75G{6q^5=u>~)f&AbY0dVi
zZ6a*Kd{ln@2Psp$$_B0)DY+yT3%yxwTDn<u9C}%lLCN8T6?;;Zl?XLc<PV1OgG^`%
zxa?}?9<msa=>9?q=~Mn9LiuNVATKkh<BMji9((OCw2mMBqApvWP#?0JmHl;0n(w>g
z9FdsEDws7{KeN_`aLdbOjB0N!X!og+V^YW8cl3n#2K+0<&vhJ8j81h`wGAG-xSN`F
zFc0;Hw0f2qeh+cKXZ6{5h%W)gjOpUC``FhZWu*Ty*#k5TIg9pyDJkY%o7q&|N@j%-
zMRh#z16>I53`n~ntMKn_uN0)fmmvSX^z>PO8oBM}4g4L)Tac$gl7CkkqLnnhXHHDb
zrnvaYUK7)k_zp=C0D|qWb{u&irm9z8PKDWyZb&Cp>eb$}R7D`J*$XC$Rf+I$DvPko
z^wYV>@@pPpK4rPuJV>1$BIcUbNv%&VBe#CU0?GZtoDe@y99b;nsyp@w^Hx>}ye^w?
ztWI5{KTPu~W%!87VMX?XuHw7M)!+M`Z<SK3rjL4Eao@Ri&f~LE#}ES%sHk|7gM=0N
ztfE=<GYwf~#Fu^XBaO-+51|*s=;qF$qCLcxx-si7x1(n<6G(ZscU($K=C~>cw;141
zXDmYIly>Msaqw)igNn{djfKPW`l_}BtK2IaPKIkuk~%OTbvLh))Q<F`8{OG??Dk9f
zY{}`vH22R9a6J+w+)tZX?Txwsugj0uTGl>kXe=Nbz<fG*Rf-{cX2UGFFm=rgy?Jjm
zjCf*NXruVGWm-%h%TLKd?^2pK6DAjR=l@NHJA%WR(b6c(de3w5D6T`hSmPJ8hw8Df
zD|537q;vNXy!Ou_rd0n{X8pcSxi&$qE+v)@6DW|e%Z?Ow*|cUefkQ2USdi|2EP`v&
z24w&(^iw@U6)%I}>6_D&>+2e858K;+At0{QSLbHd37{5Q=(BLj$#{du@kFH4@<)OR
z<{fzy05$Ni{LiO0)M{zlO)IKI)(`NslZH1nA$^lw$pn9nf@wVtw|%*dTV9CeIJ~6p
zjb{ED5kR+X@^)*2HSn6xYb-Lp@N9!0xYvKZmyXDjwx3Lf!6In<+H<~td<B}2rw&&*
zf~V$(cx(AwwcB{*`39~!2+6*0L*nBATAeq)<C(i);W^3T<T~BCO?*f)$g30)Rp}Ev
zo-OO*CcnDtmB+-PC|mi0O!hpD8?xP;&ww6Vs#2y`xhAeRl4Hz#jp$ZPs&(_W>YJHP
z;yZ&9#xqZ0v5UDoJM>1@ars?V`{f4;DdyV)Ry9k*NMh3V+fg|w)&@wkU1~AJZ#D;~
z{w}d<&%6)K38DhLa36z~Yre+fe@~Dp57;BK-!_+YWO(QzcBXeh5yyrhHcHRWtl6a0
zE=}c`6d1ZWwB}+1L&T`o`=IolmQ5NXeN0nLel+L5FAq}AP7`r5^{P37yGe<(OwP=b
z@e=!w*KbIW$x4D|zU-zl2(Nh?(20Rp@Eq7iHusGgE>byimrGClFBe0ai(8jCHX{m=
zKDBC_XL?2HUBcbS(*$rzyQ4^_tvBGt(h*O5jx30&C&p6C2?FHuXBwc0H7@)r<ElrI
z1__TIL3jlI?zZnlB0+Y8_nf$Zr89o@mP9%A<BVyG?uNJIUcLAIf29bMb8qI$$$cRz
zgg&1MAwQf|9W!SExw6w~t_|T{SUL<dJDlJHHTqM!V5`MB9BvJua-ZW>QL>(>6o$1P
z8#0wNKRvvUeCjuX1aTQbp%+f;bv%IC-mX-alM+rS3%iUH2T_s%zu?lz#Z8GQ&pMx^
z5K0oPSf^lfKYBwsEiCnL_jb)iH}(R>#!OurM(S-lf<M%(@5DgEG%Iw?<<!RLag{Tx
zl4q}7pgdLjyXJ(bgqX||QA%>|M_NpN{i&-y)pANn^M_MG`hW7hNBh3?2Dg$DB{l#!
zY8CPJ(TmhW-*ol!Suay*bDL+aYt9xuh)9X&etQ7$d)1UxwXUkR#Pg7K-=jN<TR`L-
zF%80OT_d>=UzK*!FYvJNYds>|<!WBC=<DSi&`m^DJ0){f_6S)FR_Sf>o;R=zS<F2J
zx%R0B#8l_%gjWu!4dsB7pJq3;I8U;HDA;lKqJz?n`3K*|w~}>5n(CgOC&llzCPbg2
z^oS{$6pdcTvA%7<3Aw2f5LC;;$~iE|iIJ*w)w543!;~LF_wXGX`S<Yy1qDj0er#%*
z)_R_^f#aEyZ2);%I{V1tPh?e|1pFr*v6dG&l?>C?bhI}%O+^=ZQ`~=>w~M<QOV_8c
z=pSGN&j&;Ww1*0+Z2J6#!Ha8<)ZwdRT-EwY5#SV~N(7@vFIS}`<<xE_ZSG*PUT`=b
z#8)l!FJhRj(2FIEpxx9u=WAbeG?VmhabKj=$N&(6?nuVE3&;y)TL0GLu*NeC1QW~U
ztRSVv<ukvLiS@kl-3)9xXa-dxv)`xCd<bL+8~)|*B*^3>buWJ6bCKd}U2c0s)gITn
z!<|Og%9c<QAVi;(%dCFAP{+6U@fG|D_P>9^S2Enwj4=3elh+=6@L>CHnaQtCdw$Dh
zT@|G-xpY1$sm%$?jQme@MjE^uKi{v%g*>f~yFvne?mG5J*Elb)vXfPX15|5c-}JVg
zdT9S~uI>yA5z(-cWQHap-X0B%x<DFvHHEYvEezc3mmg`|^?jEvF5(4rh8^aA1L0ly
z&>y%x_^k6nSY1>5XcehIHE^6FOiPBxb$ZgSr|xB}O)VDmVTxO`dAnXm^rC@E79Zm4
zaz%0Q)co{rui4DF90Y{Ic&#u@>dB^xo8G6jJagh0()_lNZ)<`TDS!{iL8F7#>yYmj
zS78$Wp(*uiPz~d1I(xD_Cgf@Uuefz|Go2djFuGBq-Rr~21D|sQWCM#pQnLJ<Q=)T#
zC9fJ<MBsr){h@K5QmAO_73%7H!%-~Gj3(+?d4Vjn(2KOUf$LLxQ!Pj8GiD?#4}{yA
zq3ax=-`PAhF6BaOC_$amHk_Wsu^Ri+zsT9&nO<``ec(ho3F9IarYx8ErfnsyUm}N$
z$!50L$AA&VbgbHT@U`hHCCv>vS&AJdIe+^RnJz*}4B4Ui8>Okifc!eU@LSZ{A6L&~
z2P_s@u<><CS}0de4Uc-jNcQ^9nPZ`++1n2nj_Dt!M4*zT?WIk<8M9AAVubfU5fmzj
z?c2oo+Ft)4k(V|1K9Fl478M8YE^Sn4m2o$n+lH%HhB{NpPTGIvn95r8IhFk-g-Z(s
zmEHT{H#BLg>#j2<M9AW?{JPNFC>%?V@3{eWwX$~_29-QM<V64PKKEe_sj05->o?@1
zX12ocu$&qe=$$CNYv0s{g!M1bKtb#Qq9oOi;-c`H#mu0K&*k#~zKvmC_$19`af}O(
z&Dns}*_`@zJ0O1m3|Zs;9V;yzm-T(;lBfzsC(t3Sn^Ts_!RRorSMSlC!aHwMnLo1V
zC;Vly6zp4!7@q>gv@SDLdc`RYU>g0K^{t()$3w?stEzTK?3CleQmZU_{V}0xYkg;q
zx|d7W0@sL^t9Ky2rM7=_EMOa-)GkdP*BtrOiZ}Hq@uGIE$lXR7iDwKp;_TH0-FR#)
zM{bilcWMJ7cM3;gBcm(W<M21-rK#Prkw~$9_LAt%hl1g9>A0i!#uuErpyplr*{erj
z{<t$_a8O^f7wW<Ds%5ZbcxbgWr7^d@v%-V1dYSEX@V=F+OD-$M>za)w|M-(OE+<mg
zMIQj?07pDsl)_%Kr>Tot#-lq-C;%1E%eL8pxsz>A#})Ej^C~!^wJ$2&Iwq^se{Y_d
zuNfo|g-6*p@ut~{@SM7=6Azv$FK%=UsVj6@fB1T+k_bz6Bdli=NFsMzX2t(<^!2lG
zr`rfAAc5lcu!C!mG|++NMo)DHf@NN*$$Li2%M@j1h{`?uz@aE9KE~Y3A>5)}+z><+
z_Pw0-;t5r^bXkWjA6<gl3;-7)Yget4*0;L>q3cLZ7IeSbNu|L!^Zlbc{BRc+F^XL7
z6r)u`ciKHYU0Jx}5Ei{J@R|<t!=wYBHP6*<e60zyHiaRZy$nDcnE%?}k7s<{2`J{_
zl149jD7nLSHcs%m(snsk22|X}Tcy8UJgOG?QCUoTM3oZ*f`7#^6^@ScHG~gtDK>Wc
z<frC@;OlKulV?YnDM&}CHCE)<HCLzZ^x18mx(A($zlr8LycuhJkaF&E{XL!O_P{34
zL8jVnY8P3h6v03p5+taR+!bkg;#%Kt+5b!IG=SEudp(FYyYw3kCPe3|KPZf9HXdnM
z%;k#r95U!avG=Y%IVU(h31Bp^iU*mBg7bCrrjXOtq&1BXbE!?L40l%y#EIGqPFtSq
zu)C#OA{Q~_ao)_{`*F+hQ@W6D`at5N#VjLUH6DmB#lL+PO!ijgHG&0h3O>JgsTjLV
ziZ0OPfWnk(tuWzv^W%dMU~g%ea%YGtmuL@@SfL%lyzuhUk+G=qL*Mgv=|oejZz_%H
z4#iGKsi(uHN&5lU_BYA;s^x9CexqgLDN2|qeg38v^(rfy2jPYD$%5=SyE=*_3Zy7z
z{;VK8tm8c2zI!D=JC^^SZ&uW4MJ&v>^w(x5iDFh|6?ANBZt**$oIjt8pEtM})nvMN
zf-+;zrsi`ASrFYkxKMR4r-ytm{fo7I5g7m$kXZ~wRc?P_#_2HRN+2S-o9E1wG_*Un
z#ii+ryvdXOy)BZ?R8I3kvvO=LFDjwoxqqBCP;x-p-XjcFF+^fWuZTPoZFnX&+E8=8
z@Hv(7I1uJ?T@ouFs5rcf*TIJz+~fi2kw{RBDdOo9F9Qf<+T5)34X?B``Ck5d9}I|a
zkjp424Yu05{g~)XI?yowC}7IF$BOl>H@WA$DSj=|{?s3f$KiRZ6(9iEpn~D$n_Z2=
zQOhh&{oH^++s59t`R4_o99!@4`Y)^BQYlu(MyaEW0o{Z|s75*My)k?Z?UzeDJO42u
zweEVRl+-q!<?IEwb?f=0a?CK<3|9L;$^Sa<%;rZ-I)cd>R#aUGd!Bs@GM#xsB?K{D
z)v{_zNJ)Xi8!xmH_-xd?y&KXqGI%XV9|u#5>2Y#$qM2?;pSdN<HM+s=%m4shP0Sxr
zLq2(PHNmP+HPY`p=QAtZS2{{I>Ht~<nx;Rhws1ORInJB^H6=nxX?plbsD)9ONlQY?
zC8usSOxk{+Y%M9lu#i2neGX1=&4f8^ng6<EMXaGBn!nuA&NUFaXLcFy619s8AopGu
zl3wA}Ydwa?_-KUPK{_9-c&%~e!3Sb32g5=43)`U@d!ipP<m5Qh5{O6#t$-6%zxJw3
z1(V0|8z)l!!At@90A0<A#xPnZvaxJV^Z^@M!_mtg%yrveGD8$-ik%!Aw1$}VBnrfH
zn}jU&rf`&24^KHkAil<xpv}?(EHs&%ai;94-<j-rwC8#6MdVcSJe8`EQcOk9Wj)uE
z9>j+BWl<c*bf%s<-riI5RD4d-A(ORRg>EPopwk#HQjb3nNLmwJqXJDx6i5%fS_57~
zB6)?Z10nEa*my~!#qb0lbH&6B0ZQk?WZ{=sktQ6A1E>-rHt#b7%E`wvsBGb^)HVYA
zWF<{cS!_f-M=3nCMn7iYKPK%g(Lr{#vk0s4Z!2V=32Dv@Dx^HSpou2AuRQR*2S@I+
z<+~_|?+0@y63ql5mm=ftc)HTD0)5n)I(oMEST$0?GOR|c>-GQqbXSD(Ho!8lmnh3Z
zM18qNjPGMPPs3`gF4wcf#Vxb9w>NqBihL|qx^3Ei+pYxZH)bd5#&wgBnr5)qJG^7X
z@b2i<;K=AN=y&lcNaJg?>#;xPnfZPLzxwF9?$PXX(Cu|hG)mG_mr72GRg*=gXg_rg
zV{brl30OL59XIKMa*nUH?KoP#<(JZ7UDzL{-NM8;jtUBK3!;j}SM=OI0w+YbHny&B
z_AvTaYnj{DB{ZT%*xtyX^%|rq2S!}y@9!Tk;!#1ndN`s#(iTp$xV_Ec*Yp5L%T6nO
z&%}^pz5MO~M}C(B6!+OuZm7}Yc%Gw%UZN2%)%p=<YNt>v-9z;;^Fpni_2b1dq4>v4
zkkVgL^wfYH1CwTSYdZvd*^o`3<fPjkoF>>flJ%lNF5+9-hZ%?J^jca*lV1D6%UNrI
z4bb8_j23E%Aw`Y0^$OY+_=Br@4^@O+X+T5O?o+2pAEc?hgpveA-H3-3%&aJ;Q5ZIC
z<R)G1&YwMo;d}2Zn2f`6Pc&_JqJWB>U6H;VF-RrLm1n~eC5WNik*XAyhUF?O__oll
zdf+OT>~(C1GKZ!X`j<Cm%h;$3Rh`T-BhD0HBfnjET)kR@ij-s1DglF+f4I$RcJ9kw
z{NLH?86UOc(-*}A1s?h`dR4M?0%icmM<f){ame>w4BMay#F55RUr#?&0+U^5cwc2k
z*5#oybYRWJ8qpH!Q#9U@POU%dxVF+GakdxN#4H$}cjQ-NzNf<XL`4ho)T=CBTr){N
zTdhFjA(ULqyDnf5%%DuqIbW{!o;OFvSh{y4wUgDyrqs8CDFbsP?1Lev>%%@veJ%Jm
zXFm^}uG2!VnFw$0eFi0VZT3tPO>ndb^XPJslCvatqH|r*Y2!$uhyL(xeg5IEyC`%<
z8YlcxwDVIGs8*^VH-khI1=vu{36i!|N5&eNcSSz?82*)^M+dq)Uqn?tv;l;?PwlM@
zV4PCR2cP`Gu=PI*Wk!Ti$sdIhhrnv+cXvwOV~0}O-0iX;-or6%qmr_CW(<gyzuJC*
zj+RzlO^qN_Sw|=FXnO`;?XYa#8bXPKho^dZf#P*qQw8ODtQiVAggR|Rj9*@1j#_O)
zY$pd}$D#m`>cEAxRK<NoInC(v=a}T2bPy<WH<R*3gC)>sU@vMf*n4k=NUJ&S-9q<8
zbzZGR6EwNERchE-Ih9>{yi^m6ee%mM&zB6t6KZA?C(Zy5d?FHS(Z)8vFmGeF^p#Cv
zq~8n8gqsa_;zkSFBpnrbiUa1<uE#0^unrqE>UpgvQTLR#u0gzWhKcIv=;$K8=HxII
zYJ7VRZ#Y|vn=m&wKiI6CS(~tF9PWzacABuNOmaaai+CLLOHn7eBmjd+bZl&G(u9in
zy158t+M;dpc_=ViO<5mX-gAv=-c50Ms%51%wt-*!3R{aLV15C(kYcS823Ljmf!#v4
zT>c0f+A-my8Uc{RoysNB&-4;5H9#)+&aC@v`RmYnhIBW+YZaANR3V=}V>)rMUY(ZN
zD8I1P8{xRNJly?;y=Fg?nZFKNjSOKbP?0tGX>nh2HC<aW|FXQ<aRKQiSoTEMNI)Ww
zJvU0<)c5(_M^VBB*@8Obx}x6q1QslZwi5>3_IT$SXJp)QXFfvYT=(4ZJ#0(3)HV^{
z3d!=?IHJq5<llxj1+Ea~vu7Zg|BA=MTI<4;V+QO8r!&ULUXy_M_;_yY)bjt-@`qWD
z6(ENr<~Y;i<B=SFz=z|0IIPZV(1NGy(9Mg3gQIL^Rj5{|l{up4A-BKWtBwwoSKoc9
zlcGTl{uM-$a=bofS;4J8NHJzb6nEL#bX{*pf0Wvmyf>D=8+3JBuYAJ3?!DuJDnB)?
zn<zaOSJ2~21E8dT>h4tX$o&)l#~0c%K1n%X<U>_raN<)swmol(E=i~R2o=hJ!}ETt
zc;?E=MwbIze8p`u@ga}#Fo)?aTYlQ@h9GN;wq+1e-XtgscHo7<ZY;y<J1dJ1Du|P1
zfa9#&@;mPSB3Ta++ETQ&Ee2-0^Ye2aySbm-Qd<qj(;if^!1*{k)#WW3L94ABJyB{|
z4$C&)TOKHk7|hL^!y(@2RMR`-L^JTKsfIa(0aIyv^K^pAF_}O7s`HqfHkZ3MCS1kf
z&fa=Jjttl8zFqh5;X{)PpYbxY<w@hNFPvv1jTf0TQXy^T<sGf9%ZU^C_xTk|jk^Vx
zt$>-aa@su~6BCo@-n}JnTf$wIy+g<M_nKFQPbV}$0f>_=TymkZeLH-XGU#}s;c3WK
zB35SPlpRu6eBNepeH^Q{`Ju&ByH}GTS9@#raGjIA%!|49{jC_1`}Q7cZogkcl?`i6
zPfyD?jv1V5#ntj*z@U`)ZVwZ*zGbJ)uyK&VGX*5Xs@NVp($H-^82)Y{#mGrFCVq2D
zMO)iri4{M%W=1fnLnF3`av#BB;1qAplILUVpt{WOL=l_b#22h)Yl#vKdQm3i_V8QF
zxm)nd&z8Gl@b27-3N`vt%_S9V$~31M;F|(xZg^<uWdxY3;kU9x{r4KIG2Ga{FiexQ
zy0fGD`ueypsc+u~XB_j|wcc|dUEQhBd+3#18_pjQun|I;;7evG8J8Z5dP>qLY-I~5
zB)uPb#j{gYW`~ubD|o2xA?X*6N(<&*S#TQ~B^<A~ys(heW4Opbr4~heHK-+7f5W#l
zq?V}o`?nm<XFN4k`#QXqn5lR{`9?w>GNW1Ax@EL~r=+oG7@?rn>9(eTXrA+6ew<$a
znZA@cs1%l1FC(;6I1=KvK5Xy<BImtEe_hgF=aoIK!Q(`?ccul){^>NQS_eqYj&tvs
z_G;sxbU}g$AJe)GggHPpYOf_f;Beow7{q4bIVk!>embj4#$oIG{P}js-fk(7w4zJ2
zEe!W|9hg!EU#mNh-vNk_PIoAJx%u+-;~nz;aov^u#@$8&0?s&Dwmgo<_Om#BYvqY1
z_xXTzlbMrKX>4qanTbj1&D_|sZGK+fS<?}yTc%YzUrm2GDyUvGpgN|Ig+Oa@GV5vJ
zwdq`S%$cZ)x~IQ2_oHK9=AeBF&dA3mBYN?0jeA#pO>n~*Nb|@h=kjO?!8*4PXI6Jf
zk63H<o+27FD@!!O-7elPvc2=6J|6j<B9AFw^`#{r+8VZrEgKtZ(??jr(kYbcl|QX|
zRtYyhv<Px!Jq#;yThs%Z-{p#BX-cSI=tg0^Oy&CQ2hBCnJ?mI5s<De|paG+MojpSQ
zIyC9jXQ?#|w=?cZWP)mynf43V=$9G?)8@*=va_vKnt|y}8;?%K%0xWrMTrSm;x{V0
z!k{6}UMLdalhJiwe%}%jV5|n@;?ONm&1y4EG~iGdDOF%nxv*9H>laz$JQLMD%)TZ~
zWnbT6A9V1Wk|US3a^-%ccH{i2)FD_EK-DVOw~7_8q)Zp~#7?@RzxeQ*F7L_sGVgq@
zE>6&PuK4=VN~wOGTNAD)kTq$c%4<KeRkg=<TF11s(+c;gMFiHjp!j-#-;Y<zXLgke
zCeH+p+d}P1`BIZ!HK)UN2Q^?(^g5n$aYt@e(m`D&?E)AK!2Dh7LA7W6re7j|e=O^;
zbQ}5ml$HIvntCQNyhl;jDHm(%u@<1p0Njgj=2P(cT30VcJmM8ak23c^7w0$^;nl|E
zIm@Qbo%kpO)6~eSCT!?mV*hRoqCES!p4~z&2}{;TTD-_ZNpjG?7e#ev0v730sR!ti
zsBiRg7J5R*80^_-LUmdND?g7`YNp=K%Fml=9w6PNDQ#Gc%hM#O1GU)O_3QP(=f@MO
z2HVkMy(e;czK?U`MqhS{ciwtYt^8n?+#pO{H?#2)wH@&_!MX<Ob#}083tZ-qZR5w}
zj=+Jz{Fv(nHliVWd%8{@SQl`M^7-vv=wBLD;c)Zi>q;}AsoU<bdvdFwP2Bv_sOSAm
z3D6%pz|hA}9N&~GW8<lN8_I<CeRwoX4>&&4PL(TF%b~4R1Y${A6qAo>pO(ziFYF$+
zZ$BFcA?oX2ES}OWPkDDx+o(B?{I{3|@0%lO$*nm!x!x$L3;PyRFDxNd(XFoWCaTEu
zH@ltxGYCX08O%;v`T6sA(>|A8Q}&%~Ykg^cpCvOM(a`X|{_$V^(YQqo__^vtZhQ*u
zle=`+09Le)NiOw$?-2<PwYq|}lDg6gHgGZu_r0@EeO7YsYEDEuDv_XBI4-2#iK#@H
zCB?MnBj}pgBIoIKDZW2M%dEu(Jyh%XXil;l5h1WafM&_?urjdpk##b^%)r3|UP^8t
zft9%(xhzj106OGrK;r&7lvSR7TONOmw3bHgnZbyRlgq7Ob`{QS7YU+BU6Lz+KXm{P
zAim{&WJUuT$K~>lyZvWK<oa4Z95Jk}zq!f?wy97HaH42i9J=)=1Lm!>i+TMOKwT?q
z=n$R;(6yOW=A|>P%YHK4<Fr`Oz0124PR4aw)=`;m6$q%Hx6!qRJf~=mOgA~QZ=dVB
zl*_=bNPud*%GbX4A0=hSxF{!*(?Fp`i|fTF$$&88Jk!f$L)IR#DGZNF6$x)wpw%F+
z9-Fcac#7FNLw>ne)7|lg*_+^2YOh{IfRc_*&@q*l3z&Y-P$mS!Ut5dBw*4o;luUMu
z(8dw#o1gIz!yGdiiMn^oQG?~<mzxC93t9QJe5kG`KqeaY((%Esv6D?;+#U~nhpy}S
z%D@)3w@YZ}71oscbNX(%%c7W*rM6_ks8on`XBaIIgjNuVVWZYKfonPMuEpiC!5d!P
zUZ3C2XGyM8ik*C-r!A_L3kYDv{&2i#62rtXdLXC+rhG(VBvciMnU#AO5Z|mn7EPc~
zfIj8Une)&>whg)oLeoDy2}JkrGRvf<WY4GFOcdR=uvIOYD19g2UDa1@s2#Ld_XZlg
zKYCq?E<gRT3W(@K(SaxF2A}DL-3!q_=AHR4Jw;|<8k=2teCmJVP^1gs{Ea{2H+gK+
zgUjLvrvY{iJ%}D)nxT#JoaET)j*0yBwLGh5Q*%0%lUgXUceS-=ZNPozm5+QASW1$=
z5|H-`n4Ld?XQfI2-xkcQ=jL#fItwH7i{H?Gv?3hAvAGH(33L!l6pV3yiENxodv{GB
zkC8Qc?2VeD?3V~EN3}{j6mhcZC2r<*i6mh*anMFijLG-j2qbXwh~c{tDCj0eG5;-J
zl$VA-3nMgXFc)mcc;zElHL%>&rK;Ks5))D8jV43aM2i~d-0AIT0R^Y$`$u;=p@#h;
z2Fk!At5ehH4YdhE>rSD=88+%7RbWN~&ATQE%FwX0!8CC&*Neq@Wr9Rhy{ReD!MbF5
zolR?R<NjgdV?_gRx6E|GU%IxIhS~g}kPsVMh7GQUHNA2acyrO@6d|7suV(wWX20jX
zM6CngqS<zKlLxstu+*iUgXihn+f9dBk)_-!ohb>~fEMp{Lrxn_l&>xNo5v17=e@n6
z1@X3)>_tIwJJ8yxRRp1#dP*rH@>OBmM;CPWKw}uP`dKKw#wYR6Zpvtv5GLpL7Nq|?
znhDX7{-mI?lTd0FpwQMFl>-@jvBke7{$qVl7Mv=jaL6qiXDGyP*sSd^a8*zTBW>4y
zbX^rFTpML{9ixu{w^V6%*^v}~w!yZ>H$Sl7b3}-*3b>YzaeGBKq&iCx(lnM!EqZ}*
z`@7cl_t?3*KF7V#b^MyoD`+)8Lx`kZO`Wck?t7*Hnw4s{RvA$uX@6Dov}w!Hd}>0b
zZ0__F0A^rwUh`}Z4%`-ye|CG5&hHuQS8G(QO(ucWxnisG;9(Pfui1&0x@ng<x8Y-Z
z9<X<8xBtYXZAZ8qk{5XtYX~pR+o$6*+u`kP7dr@$D#3bB$M`6>v4)^goFT+=utrmC
zxZi5xU~AY1k(7cel8CM&NYJI8K?M>NPS9NT@#EK@%9U@69Ttey5{7IQoFvajacx33
z_4ew3A1RaLVp{|jtF}6)-3r`zUv)%uG_?uLCrEtyh#U1b2T~5Q**&EJ^#;;Ny)AT1
z`oteAXZZwUywo-Si1jkyOn0{ytJijPE$VD>Fyd=e{s&!KWTsDCe(*M1WP7WEAKXiC
zN@|)dV>KX~^k(2j>KFFrq3ykywb>6o>qx*qfW}t~7*FCaVkh{GfT560al-ZZKM)5x
zPoYOQz0OK~I`mA2Q}ZwL!d-d<^CwaIz(TkdI8&g6j51?xXCKNYJXaw|sE@m&KRO`R
zJDld^K@`&`qe49ayrY9X+I@hoLi-e2F6UU{x*D|3vIROc5}c>u{WdL`K<de^yJk`D
zv5Je|<LcNNJ6^eDb&7RraZ(=(iVYtHY6lMhPqcdr)LuSURF*&}n)?*Q-^^V|6X~tB
zTB~y!Kbht_#rm+09#b!;m=GAF4G<x81E#W=Lah}+f9Kq#4-ekvu(6dy$N68<9@=?v
zo|`AbjMfC!gjPomMV0$4Z}#QXim3tGTCnCYwW3#d;wN;S0f>uFy0snx1`e=qvB1lt
zzFq#wi{5O7&8B_Tc5S^?On6=R=ViAkmwAhWONSw+#lA@+VmjJp2)TZL;xJgP-pLO<
z6R=iV*(S!CGi{;mmm_aSew-f%5O&%5V(uZkTatW0h1bQGfkqFxi`UQM_cqv;ry6R?
zlx32WX#ARTIFEZ^s;$osdq`vYv<;|Fp3|4Y8@zTMqA59wj<vv?g#XN`Z;5gyG85?j
zM7PM{O(Vi1vxmkqz8|LdvS@)xNWVkxc;4UCLGv{9dc@ba+!r}O%fr{3F<N1h-Vk{P
zgvFl3#^YAngY_Jilv)+g+)x<LWJlYeUJ(1oiPd-kx4hNaL9Eur9+9R63(wMTvNb0O
zluRqaa3$$3&(2wR9((8Dh!0sBrM?J@w$);&wZ5liY)rjlwX^480bn(2jt-GT2==u)
zeS>Bp85odni)gk~J$v_4?B{jTBLi|qj2goV5qZ3Dui@E4kciCwnk>rPYDtoW?Mp^Z
z0(2Wj!9p|v;{<f9mV{*=>9r=D?qVQm8MYRIV^jc*)VD3G39jl~6hP5m>W!Jnlp{Ok
zYF5samZvI4i2aC2DbstYJVB-KN~h5lQ}GP_h9<?|-UgChcOH9t|1K)u(en5W@R%4K
zm70Pb%K>ziZqLdg9+sTj8NGz_$Ns(LWs8aBP_yF*lBgk@k~cMYFc4vuPB36K9{M&F
z6zyVLxZucK=T%mOT<#?rWZOO8Pm<rf*z*7$Y!)gf0pl5HquO{0N%T>-vF^nJK37G!
z1Kl;qEdIYe)Tx(=7<xS$ZQZ^4ct;wIi~RCkJj!N800$!<paNX%ZNMBQ%uvz!7+^9M
zi}h#2Fhtv(6_@EvA-t_D>==;@z)$usjm@CFt0jXm$>mAFQVkxFNK+z|SFyGH(s`<d
z@AoVgnBk^c4o6&k^ix6W!(uRRX=It`g~$SKOmg>;7x)ckpSv2wya}cTaR`@lX_4<A
zO%N(7s!GMuoMA}-o}CduRmGm3W^=yO(oSIAcsl<_d3u_lo(N+EPr3O2bhNXO2j*ki
z+y&ET20!gJmDCz&nMzQQKMa+B<$Cty0Wq4Ya~ydO)Jy!MkuL!zrbB)zPKNs!iyOiP
zd*{jmp!I8g&y?jM<U~HMB6?9o(pv~Kv`;MZKBXFtg01$Ai=dsU63=oy&sl*h1vZdc
zB)t*K>i5liIUwfiTtOuA&HDu8J6Z|)Oh!I0|G?0D%AlF+d-ul^36S})26%P=W;^8H
z*@?SL7!pnkOJ;QOh!Of@DDZCTx9$~mND1s14l|Q1g|^2YDeh&8O!o94PW=Dl85j7J
zA1J>S6#R|j#6NMmv2!vRJl#o7_@{!5^OV9h0ZYG@Cj%@l>bj4hd$>-U!VST{I7am&
zybnZ)?~mssigd#CzA%-BrT}p-XFheE=+-?SM1{$g%14v!!Bh0JUvfg}CG)=k>8bQe
z^&{GT%PgSAHFRk@f>=2??j9#ax(oxpMU1Bp808?TyMO0p#Z5|dhq`F4022sdo!djV
z^NSd}r&xf;7uTl1k^SbU&@fu?_>ITX%$j*e*%nI3$#MD7SRt=T;ps(Wz#<9J;pN7Z
zlV_AV;`{E5#^hA^`><Q@YHMYf${e+sRV!rbwQa`8hpj4Woie90UoxpBzG`~-{yn48
z+*Ug+ncL-!zD*fl!}PDqNcDDy-s)vH<AGk=DhG}<?v%9c0|p9jQy%Ad4|7>PPx{{}
zR*RF<jB9xo9JO1s^UoVkX-M6AdAc}w>Mt$X8%U%b=HQ%r<LnJ7mO{Thp51u({)9Uc
z3x4Z4`f6~Ulj~!|{&m9_C`ov~j9gFT?ZXLw&!$3HJe1u!D9p|nx7pe|^gSxY!Eu;|
z^xzOOT5~o(ZGoqDl)Q52q<Swa{B9+-mC-{6gs8;9kq_7UnC)kGN@D-qD(urYi>8C&
zl?T{0<L(u34T+!ewvWk`%vg7}gsur4>zT4z_Qx!%d~uGL`}~r?v(Epn*p?B3rSdkH
zhpANT^wy-GXK`62y%<ws&6;vofvnJwJUpLH$e6_30l(pzM}>VfdIqL9&314ezQ6VS
zqgegys^2FKudpw=mk(bJB{33p7%%5Aiz=+O^X9O1-5?#Q#Lu%7#HJ<@7C&$M1bs7t
zUE2P6nt*hdAg(9%I*aY<lZBE(T)Z>)0r?`jH_c)>j4xj|2R(GfQowrP>fTAXV$Q~!
z8GU{{P1q(F8CV_6jWXfLKCVJ4Cg?X<J(YOI6`r6T=D6VD?Y4o76X`&D*;Yr8Q!~#9
zlXh32e7R6#`=(jc@Oz@qmneqa@3BK8YIk^?9+1oIs}(7OiQHoAKUXM1PwCckM^4b3
z%}&9w>Ee`rZlk`Q-ux^Xf;#8N(xs%?iXULAxJHVcy$|8as7zC2CbL@)koB;nV4n(j
zQo~WGoSC+~qaAm-M^=MhbK6f(3Nc3cWkTce#!>sGu6G8Wpoj(aMPdPWZq*TK3qFU`
zZLY&_k_oka#ma6QQ!`wPzvaNjj9vaVmPge!CgpFr8f>L}QqG*)IC5d^RWNt<bb{Us
zPtd>a?W2Daj5U2f>?)RS;ane#<&ex(w(~rkH*DQ9cQ5FF`Ll%t8mbXmAA2`rD`DE%
z>4~5E$C*sYt(*xl8rK1FIYC=l1MB9WMi#r9VkD4E>EYi2ExJS(V`>8-E%PgPI}R9*
zUoFlKf90R?cKEWlLm$P=JE5m1@mkuNLVdiN-^s<`=afqBh+6S-vzsXkPW*jsb(NOn
zP?FR_##eoF-&{YOl)uhIFY04|9#{WAF%R*?I^*77Y<$(XK6-e;`sUcvw?DPfTvk_(
z47*XYBaRK8_U*C3t_k7m*CPV)Y3nH`IG-im36QvN4YsY+3ZADKp4W@LF<X8nr%9|t
z5mKfUEvF7sZTijOlCEm@rEG(QXQ_MQ?r6sGe^^~!H~sW_r+rZ4k}zkz$Yc%Q%P+UU
zc^o|-SHw{c)#uKl<&I~A6J|f&{7>|nJJ&q5k6C~;dYJt>1GB_`f?@7mQ{`uqp1=tV
znpQQDjt{ZONY`1JJeg7-YC|672&b4fHWrQJL@{VUV{^-JvXHmuozPb^1o<eR)ip=M
zL}|%1RnC=`5xI`EZ)F*mlZ%l1C#EtMF(wh*jZ|4hS~9a$$YO_WBju1VGoRLho}{Xz
z+o}uE754r~3CfuZOA0Z!g}rV?ORQSDdGULQWiAor{qq9U#(mD2=^Gi=A<rHejvy|L
zyvDNqs+k4F)G7Df$MQy{_aURXgB<qsTlBIQy&%xM`*%V<cr`}%x0)I@T)lYXvJLnT
zQ2H2(3EBq6i@p`<5Wk|L*Tv7DHBv}&o#s+}b8=Yuo_D!&5bsACEGg^}+KR{wXm>jg
znCbWi3stsnfQTm%WNNme4j<Yn&_XaWws-^r#=8Qy#H}B(Z_wO1PJX?^Hm{vuxBjkW
zAW?$)JXl3++vE~E0V-4-Qx<7~QCw_?;g1Ot7-=xRSZS&8#<fyXyAdKHW=#rO^hs3v
zpMAL8-J0>MRTmRq3wND!mJD;IP2QWllP6PB>{#5=-y7s>m-fDk>Yc^Y!K1fU3b_IE
zkR80_p>$zmwU8=D3fX6Cc3c^QI~4c($QOB2`d=Q6?Sq46w<`KoBDd}x!(Ec2+X5zo
z+PPKhLC<qvwH4l7acO?}rZV{fCu`uw3J)`GASY&p^^xm;$lLtY)DKHPILEU|ff3%Y
zGx|5gbWWV>B9hk!4OAaP1nUW`%I{fdTM+3Wi%UXO!)J<7zU2m;m7k0458e*V6VV&F
zX!uoDC<n=F2h>X`5Cz5*(w5sFV5D+27Vt{fTE}U|alh&(#bPDD=G)&_Y_6WRF}ZDp
z&Q3b=<aK{`@;OiVp8ILMyEvS{KYg*gCP;wY)1pzFltXqJww28t(NcsS9xCE*f_Q=5
z9wJR+-O8jK`coxF_kBej7K;?ceU#1)^+c>FcfjmqbA3jqWyxORo1+U8ns#i19QNFw
zse^vc-%1z$nC|?PYIZGdr%ib1o-tdzlzO%S*u{?%c7Y`uZw?OKL1oE<X$We}i_yQo
z8QRs{G?E#;wQ*)JZ%p_U1~m?H(CzxYwmmkdlT}h&{=%v#EsrNZjeFvq<n}LQ^~q=S
zLOv{Vxmrn6lk3p24R;GRD6&`YEbe*kbA#7X63>{3Pcsw6;(j(9^F6sgd$<rhTID=#
z93`*aS|imsJ2{;r!@`Ea6XQ(kc(}4|!6@Imyb~+Lf3#`+VM@S1x0jX6tVm?LXnAF%
zwr^ye9VMIj?P>0ljE1blydU=IPwqc-pI16>{ZPEpjN~jWG2Io>v#G(~miw{w(thrE
zzy{$!znq)Tl<|0z2*)22evw6uK)k`ZPSw<4D4C_`D2lmk?I$N6eYcSW(ktp+2ungw
z=|#Z5)fWn!xZU)DL0uVM*xXW0(G*GDQaN*fV-i!55eDjdl%b@<#R1lvWAC4FhXGJg
z(&jk@>YueV6qf3W5lY|HarJ<kiQNDBr^}BMD+1?glh<0)IV&+TZiAyU_O14%a8*R&
z$>R{ua=!&X&dJ)XYI^w6@bK!%(o$vHCo_K1&WvMM%+~23mo^?&OPyzFPt~<0^pyFT
zTGhfm7qBA;5*Nh`l9#s1dw&nS%~ufm5vI*hydKQEK);cZ$A8U&Ven{9PUIVT#<KuF
zIIekV2r6HEcHsidJk0ULBa))-%$?1QVFa_!;!PuhRRLW^)0=e|5<a49)N`H7Ip-I}
zM`;SVBM8n{TS1d2?wDTLB5&)ZY>1?ZA=1%UVC_g<oAW0RjB!8?`C5|tc{`>~M!xHm
zputZpkG*KNRFgwRMJ?D*{l>n&Xmgc^&vyn#V{r$Pr)l4RAfl6-ZH!tXJ-(c+*nH6z
z3Qq}Z&n^YekNh6LONQqudQ5EpW`MJ_%w$iTBJOlS?_7$t=?7wSVB~Ok3E>(bYXS<d
zeRIBv-N9-=MfwMhCHTo8WlrW3mHDYgY%DL4Y`z9lWddC*;_w8$>#w>{+}fqWk$4*X
z<mLWkyGXH&^vA<r?|<_U{aVA5@ck32_}Jcj5$8{_6Z~*h^~RqyRa-t0uD;{D>x1IB
zw%qtm+RA&Fl$!;+g&p+)V_)xicA>%HjA(jRS&2x8RRrAHGs)eG(i$TpZh!Li{d%Mx
zcMF5#J%`wo?*r4305BG~o$On*6!_TFNibDwjMi|voZ6jZb=59W6%Zshb$M^yqN-;|
z%`aT;q|mq6^Kv5QyI9v#WaM{Q(T)hc7_EDi5$>lyBjIFr?_Na=o|bTtoVg_d1rYtF
zMTq^1Lh4hd-xn3tXzrf2OMM(IzVENx{*}SnGK2Zgw8m5ZO{=M>67i&9GU@n^NB}Bx
z^)1fzYs(EULeuzElC^&{DO6VIhIiYZ2RxQ84J@|bG2ZiA@7EaT6+j@$86v62eV^yj
zzaB9IV6BM1f;PkV3mbzXW0b77t%&p1Te)N@08PFctrt7*WjumMiR*8f+Jn;&7BOM_
zfb7;q_f#V8WewEstcIMRpw6#MWgT*zkqA5uQ1&`gj&G|;;Dlf=n-N7+d%=CzO5$1M
zT#128DbjC@)X_=S=eO~syGNznA82xGQ|c<ATCuLnGDER@-pmh7m0g1q;e24b;A7^N
zkv}hvU7HI;aFoOyH8$S%ZQ^T~E|z=FU429<xovm+numPsV53s~e&y<z4JUc<0eLA=
zrEh<Ca_eME+{2)0%m?R@Y~3M^$&@Fy+>mE?W55?~Tpm&V5Uma$*cg>;r3l9Ak+L=%
z3~ZvMM%`wSvhhUrKGW9g{zQFTVX~_Jl}p88zm;(I_J)l09LAG|$&~lk^k4N2^9nZ9
z&<4?nu^TM=7%OVSe&l#$b8j3$H+9PY<erHLam1g*I<B#CqyjmJJ73|gG^)D8g3SR6
zxa3?DoWGJI{(>*Wy!%c<CL?cXe2F>h=HxE*x1S5bUM9j4=M4%Rl!=Hh1#M9uz73{u
zxg}GHlQ`H<ajO!Wgps7Y^$0V=4$=L8xVp-ys=96sA|N85B3&xdC?MU^N_U5JcQ;5V
z4bolG-2&1j-Q5imhi=Zf3-tYd++R9m;I6smd}{42)l(kS9Vpr{k%dfZc^$VefNj0h
zOZG$LM%1Ofgzd_Ag_s^+44Ll9X_Y*7Rs>cBs<{x)kkz;Oi>Bg?>I|=7#>4<HMi_bx
z><=`UOC8E|xBnPUmERy|uoA5voS)n3^hY|K#s?;=!QqQ>n)Dn~igJGP=#invXQ-(D
zyb<JRw*?8gzh1S7f0K?tjGGYI%M!-^wr6$L^IZUP)9ZC<v#%`ql0`glOuIR*hgNKI
zQ*|fJEi}F(ySVSAa+VlNH%W>&I}9@lMzuGwy+CJ*+?1h=x;RUexM>_@J3WRsXmCsH
zt}VvN6Tx4w<+<JT;`cTjdV}2<FF2Qi@#%@yr%wh<WbwnE-@vc^JjWN?3eGv$n|x}h
zzv&%f%L+PB9#v8MNJ}=U#pZi_F<b1L9ZmlQo`3_Xn(u_f0xs=Wy`9rPXD~S|*)m}q
z2s9M4=(+OueRS^==z4bzH6&UeCsW?I)b`X;h8TJ9GhC0YsA?u{Wb_Z_!XvwbKcg$^
zVC_d4bcN$MLd%gXEp{&z6T6_T6*TONmE3A2j{4IbaBNOc=<b+-qVIs4|4{s(bWGw0
zV+UGheRj(64P|zx<r!Vv4VD2<@gLHkA$a2YG92r*WXahrt`UP?DRW1c0%Ho3-?8HP
zg*2(EqM4QJr<TY1(av9Lz33?*`Au<GZPpF{*sY-2d;H-TC2fnn2Z1rn34Mh`@VQh4
zCLulIigJo5Uh&Za02TboIl39?Emly!P%-V(wHT)lt1hhTs4YgGL4`a!rs3jr4O@`V
z#Bx8J9=BUDG&Inc43lXg((wN~T>o(&K{Z**Qk%;3x*xcie2W=x(JHafz!1AC*ie*o
zfHR^&RZH4g;huJ7*K*if8LY&L9n@vQrxXoKd$~?Io^{s|lsTMMwy?g4Ku~g-utNrP
z*T-(Ka>?)ze-KHeG1Z6v8KK^tC$c`V&)SURQJzs_<<xThc|mL(J(+vdT48IpP{D{F
ztS6N9ohVvZ$b$;00bWhXb?%GX7E2`#b;yl^;QjMs{+&a8Hx9qCYyh1=L54bu-I~qi
zoS9wflYzVd52QCSeXpCW8PkFPldyd6bG(q){qB0Lu-<pB=MZ%DpO%@=UOYZIRb4sr
z5bql}&wbZ#!BjLem*5rQIOMs{TUvhpFzuq1Yk-?8L0xUGc$DzQ2({K4F9cmCH|hj`
zG8QFl(M!Z?ljCQ9OBdqBZKH%fxC!0CfS`PatdISOyGl?Eb+7rVySSI0U;1m9!C(UW
z$uHw4y0QI#4m?ZmZn*Eyh5yy!W!RqBM7;b!>cUV4lI7Jl8r;KRMs5*PT$>%IR^H2y
zzKYqQNmDwtx3ltODu#KKANkCSTeyF3mhVW;&kZZb%PD;SG<^98%ztf?#g9VT)RZrN
zUWT>P=;5;r%uDC@fy=wo9tx*4+9B6bKiU=<{)7T2bIss=rU%03slCWz6(##mWKCRW
z9a(LMyK@MD!6Lh!f9Rv#8&-S)@Yt7NMzxP~1VSqdl)?*)A>C4M1@mMRPH0ut28Z20
zq`zG4GkAhTB${PRa*Zk*Dl3=rbuk>S&z!b|ori`xx(dTX)Y+mtOJMfbm3YfcH%6AI
z1Yq8^0m#1tX4}dv9c^9A*^XhgQA@@?;`#T0hyUZhb=vRON8IWi#)4VDiEt(E{31l?
zScx`flkB^L&D#x@n2Rt}V;}0ofr0sFYOJo*M`8hs=XkH=NEoRG<BKfehRltYu)0nN
z@x)tGJ_6MNS7`g$)}Q>teNLZr)&^TYgG)J3(E(R`_BxK9&E6#z6sH#^mC=66?9(et
zJ2vA9pdDZCS36Sn)3&!T#(@BCkkn#)$bM=eV2PzLTS-f|08$!9*|?-$05Jgqb|(m(
z(AhH+8;s!lK^RK9Jyi2Bv|;iGmrN7$FJVA*G;PgVQy2?f*BDRiIRM5m{GCxlFu%!a
zx4}3Twf?;JQeoxXy=MlQSAXW1l~HF~KGNjHU9ntL>=24QxgS7v^pkLOR|7<$px<Mf
zrkFYW;?o*Thd{cFh8X`eZcC_snx4ppt<u2>TDZg?-TW}<G6aIypTkrc%IH4N&3&5C
z#OsY@%4pG}0U~}SrT*s268gR2A|(wJbM4jE(hitG52@Oa^3?ZdRaxm9o0g3x5~s{n
zh0s}%C<E*j)&rysTHa8ptwI{&&>Y#sQNttBKKErd05XzhaXPA%l$e+IXS;m6HV~{#
zZja@$SR4D>$A2WNiwTcePPwu(9Bh;+5hdcJwp?{Je$dL%-Rg(DziiPearu^*)qqSI
z>(=@Mc_Q<Dh2AEC>BlLe>0SK4QpbNNsUs@KqP?d(28`{eReMGL-1>>lX(gLrxJ4$G
zV$>`w1S@l7h0vHM@Gv5+PS+6FGwbg{KfhBTi<by}`;ol|N@oENN*=S!x~zUhnFiUk
zP-7R4G@UU;r+E&iWXkFm=EiUMei@nyaDBQwi<z1g)-}(RxP>|rvG;166U>BZLxNMj
zZ=eBZlw{G{F9bD5agSoVd78YViaFIAdZ!1>1U_)^WD83;`bFNgLB6}%Jn8s_zojqM
zQoy{pc11aJp}9n=_=l!iS2KZT7Nn9Q;RfHHpwi3Q(G$73)38xUO>uth?^O#T{8I!g
z5jdqNETB$!)QrqI9?qdap7@{^;W-|6^vibl=G<<8=n9F3Le1iG_Z3sj#-99P)mx2Q
zT182gsP)bY=Z>fEGd$`|R%g;mC@coQnz_4joz955AtTefU%pjJCiQT91Lz=t7w<1R
zSraufrM76V#9eSRTbjMYYSgXOoh?whnihRTbz6R3d+_J6m`-r9yqm{!zRrR3ZK!YT
zM!TtIIqqlztL<e6UWA#4IVGS`B^;VS<8-;^+;y^~2EdY>x6BJ5`eJ;{hD+Q_6qTQ6
zgacmFy@?vS=dW0YO;2Pi*A)n_Y+Nf=?;kjH_Xj`}pi>S@-uR!a6vKEzLJGLMMLhN9
zE1-mzhwwl#l#z)O)v%{pci#NSSrvM#wa{5YclL_HgU-TxVeY#3nC(K{9X55A%_4XF
zJB8Ky&%UsPGjTm{T$`n3$SWYJh`!GpQpA?kh!NoHVR&6rxHRc`vb>cqLcg{1D1jq9
z3+DbO!8S<~6TP|Fxg<?Gp}NK##@;KF*Ux^u^X%lbPPj|ZlEc4IQe?<iUzt<WV@XZU
zj->J_#UlJ#r+&CQ%lr?*B8e;++npW}lz7Id{FNsTzgVNqA>uDR77YPQl~Yq0vtrXY
z(w4Mzi9hZ2VXNIYZvLHwRl|4%jh{?fJ(=;JG#WyqN%&=R4kKWL>A!eP7STl5YP3bz
zZ+$=gcKis-$?xw{=&$SZC1^jL^-^YklB&Jeg?sArCdjC8SAvQf{h!9^;9ssLIjXhS
zO?_sxPDWji9h&M>bQkwu@m|tjc7H8lTWB2!+@_<=hJ|Dn$E)QOQJj7x#J44IyXt>m
zD~zh9#N|T~f!7dng1NcHPJL0%uz8jOpVn1cP(s;A?e~RPz%rYS^m}_L<uCsF#FPRo
zW7>Y@iv{<S6fUm@i1|xaZQfdx>k>Ae&+1y_+|Hzc=}-0*y)D?d$@}4NNXXQ^hF_F?
zV}*(<v`}kR4z#8H#kMu|yUzF2ljVO(#>7B%pWC)#Cl7rhIFc-Smu!mMm+HdDpz;ab
zG`5)EMR^^%Gt}%|AdDA(%G2u^4T?LMLQA@V@MN$OW-%3nUnOw7c#LQ;vGMclNC${%
zDDmvLQ4-hBRXz#B17)Afr!zygrm$s?nBKo{%N*!RtTRjfys;s&=Kf;muihr+Oya^l
zjx1M;${DA6agxj!44O=&16M>lw;`o?Pj*Jyld5=pHCALG%VM(C@`5sykW`CH2$HdE
z?)Qx9rYWSe!N3&&m-u)<|G(*v_g|g!nW^vLRXWl7V2AB($*TIzi1&Z1&!%4~o?w4e
zT%Ps>js|%Ud$Zi`;6d8!rrmPWD-NQWa+AX38)-__YM=P=4(G@3Qgaxk^A%1AvF3~-
zXI~Uu;%m4$L?f=h9U!|al0(DSXqxwR!;6|3F(dj`K?n=z>7Up2g7}SRQu6lr^d}Bv
z6iNzAt>5_+V;dRq!#qW!pZm#r^DQ1ZpPVYC<j}|_8hcGRhn(POl12v%4j9qPYN+#>
zXJMP-sDTG6zJRHG_s~{cjy5mWsQw;Qp9!FHv6F1wH+x~(bw7?q$fAWMYP7u{N|6Y{
z5HqFKj5tb%dZ+8y;iOgKCRWsOY}NZNn_f_pO$<25V{fDv6bPFe=wEROp&d^#y_Ztb
z+p)l?JTui>Tmv|Z<FzNd;7##p60h|>#@8(s_URE*e#-P?ks({Xa3<HgUNZc@dmTjm
zvBzL1h&RIH^)om>SwBEcXwBwxCu8Jf!ufmBm1WU8Yz`t9<F5K7eEaZbS`=Ud0N<J7
za0ZS@r5Dk*w{~`{sjHN6!XMymEpFK_dPCdyf?rZtU^L!tCh$o04TxPrpQzt(Gt`v5
zp`%K=nh|B%V96$XkKY{#wxA5Ad6D}GT5|R5dqVjEw7uhCyWB2mWu;l!T&aOy^acuf
z`+#D#&GY)3O39qRm=wn7pL?NstNp(sK_v4}*1lcUl2Iz^Sjc3(HEaUhtTs;Nwy}H+
zba#k4TUq_$rrVmZhw}|;d4^p4)yeA@`7wJ!v-mGW5xpa?Y+d^K2K+oF0Hu=Xk&E&O
z%s6>O_$%2O^~A(Z972WgZ5EHO6-p;=aWFBd_}#`03l98JeAKE-=149|+)ymg6+ZIZ
zaPwi3&{IhP3f*%YX|=N#;!|_1(1m}3<j`1_u)6pae#m?%1d!xM77fCg@=)kx;!M5{
z;N`DPb_jkDg&3~4vMOJsHa(|rEcwKI?|IoN62ZDP?rMj>GQUz6rO<gv?oWQzNoqwB
zfhKnl?Ds5eXbw?2pUDYHLJx>fac5I3eT9T<SEujJZgwWo^KWc`pY(-4@1@DC+E~;k
zGv|U4la*V4BwX+>$1AL+>_x+gr@LH)-QQ;TCEemaRS8?VqnbPOyt4Jz+1%EX4c!_3
zoiz&gZLa)F-JkCUT?8kmNy^tR;4=5)J^Byp7u4GhA#L<E4mwwhjttx0S_fO)P>g;J
zI9pQ+mypIF#FEaZ9LIPjG^Q7qmLw6@)OpkYuFK$G``Y5q52RMU9x}u|aS>iPWAL$1
zwpLj!U|4WBq`#gPFrwXKyIFTwaZIuJYEsf)6^nyli7%j`fTEpBE2vdsu6UDoI(P?h
zHi?OW#&j0AI<|pN&RquQ<*s$(*OPH#&>v}Sz-U<A=sly}*4|}dq_}%JP@v^_HGn59
zq4Lg^S1p)NYw-k#Rja(blG-yUTvr#*I4my|zINR>d;v_QWo{;7dXT8IU@*#5GF{#n
zxBTGkooIuUoWcwcyoB*O^XlD#epQU~ze_^*DECVk5p@aCoY^hu<#amLp5x|)E8W?v
z@?<N`CzbGA#s-IpNVWU62a(Lqraez!03x7>r^~l3b4y(V^Oif<oK@?RS-*Jm3|tmA
zcp^;=%n7ZPkrMe3^@~Yg_wuK~!W&IhFLEp_Bm#Ft4N8>11T|3rfH8Fm!5dS=)?gF{
zIOAs4_1o+<Lkv30+|6NmfdyJdL6PA(?<71uv=OEQ8Ln4h_6PI<_N+!mL0<C0a5e$!
zFS>~TuMY<SLcOx)d7jBqV{p*QD$%pd`k7eMJzLtQ(QC>aj86VhBijCp-dKz~SP5pP
z3M_1i#$r0Hpn&}JV({ilXHhJqO|8l0A&%#usnic?`r|A$lw{s7f4<XJ|6=&kwzw#z
zV$rHE5Q4-FX5longv!^`6+<{PMTKs0A`d;!R0<H9x#}aCxd#p10T`>X>kaJn$6A1(
z&=II<seKw*hFsb^({G}?`v4?VQ|$H-IkkW!!_lGF8qkzOlw?20MZA7G+4Rr9TWe(E
zRcwG(2<kB*Rp4$_WnliNVAXy5acju1b#j5xgd)c27<v3q$}N0%YW~e`=&uc>joF~0
zL<Gm%b7QcS{l76Ivz_+ZCaJ=>BzbPjtoQvzbtUbMS^SJC=*(A#%#~jrFKW8<cP_(Z
zjmK!6HZ8#TuYm9pd}2F5L_uzBo){-_4Ff!qC^zqS&YRV~tc|S9R2o~&dLLp^pP99V
z5}dPzfw_hamFkANK;Ysnw<l1r*Pa<C-5}}gtp&;ESvZA^Na+lX!!H3dq}FmFAo=Sm
z^D)J)$rQ^QV7ih|8m(_h)WEZbgHF}GTy_%AF!c{NB>(4<)_ps{41jYwPoB=;0?#u;
zI{)4J4lPrRL7!%s@3N!B#+0M$$2nVnVBH5YzNQOdsCZ9dHS!W^fh)$V?3EIi{9Zm#
zFBjS4hn(Bkq3*bcoTR^{#IrMAUY@_ACY%3yo(-8>1`_t;EY)%R8mD<iAf-WP#qi4P
zr~zsg0Gt_251A<FwTYE2f!IwK+MSJ;IQ~M4*BU)*=E-1Mz}x&`3OjPDE~N+$E}6gj
z>DKfAI=+znTFisbn;VN6Tqd@rw0qpS)P1dKqmFeOSzsMhXlb<WDp_RG=J@t}d2eTb
zI#tIGBnZ3y28YC+o?(BKumRoH%;Vg5#c6K}!UI=ql74AeAK10_&15KSYzKm1Lfmy{
zGA^2Mk8NPyZ@P{&R&|$<gzJ<M;~k(k>F&&6MrZ2}&Y!hHMD#U{gO2VSv>o<;4dC@3
zQxk80By%z2KJ#~Lg>7>a<sFwZzWm<6B|p=YH0Ozqj(lry7Q;3%p>3=m@BmX035G%y
z^?0`-x$NA*LMEk-#mRC~VX~Sn_O7!C8ovAoP~Nr9bf9}Bl&kM@t5%bkfd&Sm5*sYZ
zw$UcqHs~UX`)`iz-j8P~;GQDU?iN~YZ?_E1QmpV+TZ+U|szzirFh7p|o=`rAy_w$_
z-ZnE{+);bqx;XwLRCdpRwgX4`kvkOVROFk$=S?flescbjWb~}mTp&J2Zt7r6G;s8o
zu3k|?_LH>>Pk?JCeNEZt9!6G%ix_Jn=`17j$63IqvUribCMy#cSR{rKX^ZuP7@@hg
zejEAO-<d|<6Z-DM-<cMwi75d8k+T*%SUB?dN3yEpE3GEQyn?D-=q5*_<_&ho_GO$3
z`*XH^9`bi;y>shb@i5spawPMjP1A$Ao7f-BvtkyZhX|m9llyaMh{V2AZRt4Io$0R1
zreMJT#5sKcMNb;5U0A%w%-R1rWjM>@HJtCDDKD~<-;lZ#d}0R)pKsHb#FQ_52M88=
zuo5iVB5Bxf*3s$lw>kGvEL>>(h8^Ab&Rc@4ws~GqEJ81Q2lNiUOsX8H5@tef9*6k7
zjgO#mV>?BB!XY5C_{AvQgEbg#X*bjn1Q+6>NN-O{0IyjlUh{FaO#PpE^`Dp%Bu$AW
zes%o-F7)Pr6{yb%vCT$n2j`wy)%kHyRN)o;rEw$$mc>KJ+0&{+%Ews=0lgCpCmW6o
z6*fOtoU&vkmZn_ej$0WDlijF#-)(edL8^2+GTwrC&hz(%lz?`J-dudBP|8z^HLyL&
z`3e_z(*8xV=JwlNY5C^;x1jtUN)Ubcos=#gtyjKtt0Et}&P1hwN>=k<7GU7>kTnPh
z{YX*v0aYxQqB+(X%@P8f(c&6{#_zmG)nQvaPTG;WL|i`cM+XLD1~EX{e0jMvT}>@l
z8M!@3*E!Mey_U&q<F*%xi*`lcaSMS6G$E31GxJo2(A*~>jClWBK%y@mZ!}orH2U07
ziH3bX4txa;Ydbz33e@jiH%okLZ4zy5!;Y0f;?|kzYNDXV^0%;yXYLKegK3{mh|ebt
zx;QV-a$vjbaAqFMUnky4RWbtO3ET7RC731cyKZ60uy{UWhd!)3Uqm*vMCaI4fNFqE
zpeEhBPsO^vaN+KB1Y<Myw;tTD55b)g<pk?%N2yZh%kw*^+Bk?~>#{0lM^Fg|e8sX#
z+sxwZBp^EI3B!aP=71YveR#eW15PimUQJ}Nl`O#ZH9Rgoge)@A#yy)$bga6DYLE)}
z)Pw{$^%Jx*m5e1r`*X*=$AyhB(mNArIm-hW5?_GgU^f4NzM?WSn=1?q|2xOSS@Sg<
z0(PjuS%H{ze7BbO-3bm?{mvn{!-(JBeKb$kZsBwsh4?=MB1B43oKr@1<Mfs?AcLJZ
z=Hvnv$zC(+qXW^JD$bG&GlBTv09F{d<~GDtwah%}jGwsgfa;#{l&6CFq`HfX6>KAx
z?NAqYIn|Pa#(+YFfs=q}XS{?nGWBN!apsrVTG9K%p1u0lpo&G)n?1!JL3Q~lMf=z1
z8jVB;aHJLFy=y}d9f(8iNS^sZmf)iD2Wo7U4D=qu4R|2PP%6CbS)Vhf&!bS3vwBz0
zEQ$$Ek-zOXaGGw4sgih|Lo0OyKww2$Ew2M~KLB`6@JVL!6n~)J9l8?npB8k3!ghYp
zT>Di-BzBk<v<*1rsj6RmhQ^!dAM~|Vf@}i@gz^-J-36b{i#)Wqqq=VhMEw73xeU2-
z8xyvzB2Bg774_2@8#qAP5=hLLWRo;MSS!LSkYTa8aICIv`iB&_2krc!1aD{z968FR
zr4-4M(cancGWDVmweM9Fy+5UIXB2<$#-m{>=ZDu1q*EM0qXmort`szP7ja33uT3+2
zHv^+ln=$GR^W48<!3f2z(|Zk)V%U>Y6TvPBhBbtnCo@(fQYme@tT)SbwMAqNonPM{
zr5HP%?M?70-1pIM8~^&5?^e_={u{t>fiWYk5)~R!5g;Lw8PS)!2)EUAF!Qwv4^8+k
zes?ffol!Chzck|h2ndsLDl?fylxe`fbUw^m>oBxBNd#MJ?NPu>F6oOJ9f14R?VqFg
z9VZpV3+Q>pP|~vU^Y*k7;6;brthFW0+ay(}h2EJ-VAy_%r=m%@7pMg5|94VJQKDR5
z+`B_1NgW1}pVVJ!k!N*t)rTD2ykWw|zd=G)w=a}b8>|!IrC`{7LLGt#l8<SFQ6k_l
zH>)2qBfb|Jy=E<Wxi`k4NOv9gNs9J!-=aahysnz)6V5ieF!xXH5^syS{TUAYv78X4
z8h4lKT+9(ai8OBZZD0WbN*HL9cL60ZZ;9k<?|Yq5;DDog`Tbv?;`Dj!IY?7}n>~J(
zHO<f;IytUrt+OWw_eACW_8*@RVHTcqv;b|xQd`u|3qjAGs9a2}16t$vKqS>eT)jF5
zwBU4>ir2eZP0!1&K3ABUiK)os6}n$Ag2|7^2E*Fm@_YdaH;!d?glEF8zJrp8G&+Y*
zQpABtuQ-1WZ?7<eo~wR81DT@#iMWC}>(gCf009D_ZR&QJ>uwYfnhs6c9_(;Z^@nx<
zDJoBD-f8?WN0H>EK`%-nk$c=}A0Hu?Pa>1lc?)X%(QgA_cBpAV1HSy&C7+y%0Vkt!
zy_B=HWA~(Cn+Sx}p1lY(?TlCBWY{Y+2aNFzFGJ8CjR&Asf7RswhQTY2s#!%IQ0&}T
zh*WpNB>=X?cye+nT?3wj;EB*tDp1E1dtgT?LATVDG<rc?2A+P|(zLYl+(<Dbg)b79
z38nyYh^~vXLxoe-HoNuQpw^cFa{!r);X>ss(Kg?vBEYNyexH8rmg+P4=lTnZ{i7ZH
zb|#`#-<xjF-aDj70xP)m@^qQ-!UTo=(3N3p$CJe9J{R{*E4E_47S3j+O*+w7`RI1_
zClXV6e=F?Uw9Q@MrVc~XHxc$R^3Rj_D>g-R&2|Wf#3y0zPwc8e!=Kjxyah>sCn_fn
zk&63JPi&TVCyA$oxFLiv4=DoqdyVFms?R?i@f8O*NHP)6o|Eqo!@mFA%lXK2g$Yzw
z{;ab0o)*;T&ZIvhW+25(CrAlse_Ig^Nz1_eVZtTmSrYt9t6gsL$=vr9KR{}ey?#~W
zd0Pgq$EIGSZI!&DJhg(~I`DFX8brdoogWC#cL&%@MdD)n4SyC)H}lB6myP%<-~JbY
z-SLs_&NzqUaFzHbTNb|Bv*H@pSCD`4MhicqjJ^%<+D=}Mu>Cd{R(5*_JN(Cf&%+il
zRpm33QmK;W58tU6$^wxtt2|SdZhZ7Pc6Bk^HQEnK?k^ya1)!%|Cf8j@eKZ{S4->p9
zrg7L0rFDTp#XO5?X>%I0u|+I53-&8q$&ebnds=M@ce$jEX;tMDt0J`oBrdZialy1d
z!ry38mJ(Cuyp)L|<`d$J|5x4|ul{!t2_ZomK`S9h2)H4X@^4{Ah4RL5^joKI2gaWx
z`8<QuH(iJv7+U!&bY`6XtBJSYF<n!VOK!y->^Oib87xdn`5{kMduPA7lnfKk8FGR>
zm{~Fz;YRP@Udp)Ep)VVu7092}KG6?fGGy^klK4Z|a)V5+s*^tmt%BKuQcue6NgV*b
z8FEs^0W?Ody>*GDc8`;UDGWxMW9&sarObjN3S*ec^R|KT)Hor1K0F}$sW~D8Mfr`D
zjb|Mi+OFO8&wM5mS|YPts@@Kxp-ky~@c{*m?}%O@i2|n4hTnrEAb&=7u{Z0Sq(HC_
zTpsojX8E^8EWWEl!5&+&+@dkyihY%1RdSJdlyx2jugo!JXBU4zn#_g&jwbp~`W|?T
zb4S=@ctT!>VK^b5hqkGDsdd0R7av`pBW3frVzybNj@<+V3umyB7(!eLDEp~@?b@g)
zK8w$Zyd!2nZWZ@1dDTGfn_wpe{%!j7jN`o|nTA6V$Ux0aJ;`7t?f}k_HO7GUZ*xrP
z>dN#vM8JH2>UI}{lSxstnLZsUeEt6QclN9PcOY3EcXaaFCHV2amuN{}Lk+8O*K^8`
z?15^{Mv=#B<on3YZ-?ruL>U8I!!h6hqen?Ttn7CFV@g9M&Kpo0LI(PAk%DFcrFggg
zAn&9G$4iK(0(MjcPAcDA=%YTxLW4DKIXwGCJEoW;492@#vE}l&9KRWe&0%)`LFQ>O
zq6cjCfTYZK-v?)9#GMipGO#bH2Blszy3n9pH>KrYaLsDn#)CSbz1%n?AIuz`sJQfl
zd|R-4V#G$@<Jtb4>p_?9`jX$UQCj|+$UBx#1dX|A*Db7rAP6sTCR0f*Or4DX5+)UC
zn(i;InbcvzZ5e!SP}lMl`G6j|>VfE=SJ@BFaAb%y`_8ZmK7xS+7G+oKLQhoJ6dA}b
zlpp4m1kB>WdjSOJg5HD`)?B7M<Ek^ZlS|7Z<*@tiPHL3wGxqp-%-zY&6nls<xW`oB
zj6iWpv8Vi?y9;V+;{z;?NEe;cEH|RO*`W`8g>Qi?EC7s}7>9lG^-sK2#6HF7Flzzn
zd<m`Txfu{6)oWv1OdFlh-rEvbx@cM<;2_|vfg9VH07?m!IOnIUJT1_)NUwS4#BYi}
z0!xf<TI9aFXoS+TKfA-b6+egrHaGRr`ZZW#GIKv)hn6{d1_2oroOzmbKJY$Qc3H!r
z6<VK{cTk?xX(Lj;B;$&)#BLh@;rzq}p&-^>9EjP<>@JL}-Q41^>vy1=1%!w>Mwu|m
z<D-HY;-iEE)(tL4Uobm$42%aON*qeJFCooFzs>}kp%|<c$NkK>3Xh$BAiE7?V7u_M
ziqR{55ruzeK)-;e`NUU>zh&@ylt7tV%^DHd388lMu|y4M{ji2jUZ6HNcV=-`q@o%A
z$!uEYU17PC|MQu0`)JF~w<HdJe2u#fiwv$mM+Ub8grgZ}wXSP3EtkIo+%>Em7Xi}N
z)Q@Gwdm`LUC`2oCK?)cl43)-+Gq6Xz-^&kt0mvj^?k4{eJHCz<{O7x{am6p}fEg=a
z`!J7CK!Chz>B<jI1(EtLkq73HKnPnPU0&OGGt{zaI#}h3nY9r+Ok|1eXoXSPc{DK=
zJ?*b{P5rN`$qNedR&%moNSP6!oEcX0=y&H_HJF97%F<v->maFy@Hi_l_?zC@uj5|k
z&6(Rl`!`>^mX`4P=GG|=p44@op+82PN2b%+KU!6ucC|6M)eeO{b?A2nlBoggYhnvO
zBIIEkq2teq{1G_Qi!{2EDl$5rNQptW(KQ1brE>EQ>VZ`8)^0~QTP-$^{^$}gGhole
z;*RH@^%W(v9SYz^?6l72%@kBPlw~e{i2y;;-P>J*6MN?VK)T+L%`H!o2cK7TS-mn=
zmB_=~KeK>L2iVZydz{0{L^ox$_8X%Y?6PP9oPb*xsY)WEKZ+>1a|6fYL%Q^P>uC<8
z0Z+u}_;m$cGY!}Gw~+cR_FX!8?5MIU@37&?otZqHdwNm*4^T5rhdJJ$`Yd==EPm*m
zKjx;kA3d))A)pV_4q8N5I#A*qn%pj7t6oxP>1v!0iO=GWBw6r{ZZM48EVISX8}VrZ
z`TeQm{StYs7%l5N@Cg<#KJ*}%_oU4Hh)X$dDQjWLKoTU`z%6Fsu15V0;>;_X2q64;
zw`&6Y!NNW<Oc8Yl{-i0VFE=<U*T>D3jYqL^HsGa!X`n$e07&~!V{5z!VDI9e&OA<A
zUTPU=o|%quwp>7IhSFJ*a_hK9u7FsGh!S^et-#{vJ)N#ORCbP};ooOUO6es9ciD&S
zJ0>j<ZNk_|dy79w+=jnUUw~1qoqj(V_Qd{92BYWh5=8NavhcXIYmuhK38vzp(5=o0
zz72oYyT#Pfw4Q^8vFI}4ZnIt^g4LJ69cDVXd_wHW3vvsDP?IjKu4GNu(q-w=XQA{n
zz6d{d$E{LH;I5$+WtWGOW$q+Qa}W@M0D9x>t`kt>9tPJTm^)*2HO`LCtTWGEMt=H1
z9MT1H#vl;3*1o4bTHtdBii9)MhvR_5^P07($I>!18$^3kJt{z8jqu8i;FQ^Z&kZ?b
zhtc+tT&fC)e<qYlVcTrgyXH}M|5~~DWs5w0HG#X*%F{Zskf6rn`G^yv%c1b1=R105
zI9~jhOh+yLY!-JqNDyc!!tX90dLfWlwxrb`@xK2X0+w0}XTB>KrtXDv)-S`~z_L?!
z4W+00%Afwmu5)zvdWG%{3ES)zZgO=Buu4HUcqrvEwB{mU$RnUWCDLxe+c9Q{&3qMo
z8COI>0A0nVi`>*sVsvv)Da@4zYI526`*az%<lz{FNJ8Quus;ZF7-3+S+Uhf82dP(`
znq<A3He0ol518#1rcQJx$%ilCK^AX`RSfKq8T2lRvC`?s1kzFK0O?akG2}|D%NJQ9
z;Y9VQ8XC-^SqXd~fekC%M7g$mw*7496FZ>@q?^cfXhw(>k{p_uN|2ax9n<bZb9bmY
z*N~RCpP2xt#r<QNM=46mX+WW0ivkDs4%m!#^;(^nU-j%(W4kV&S3yp4q$wph5Bh6o
zKZBqV8K(ozrc_Gi^|yht(>t{HN>J>d5G7$#hWCf{vq#X8)7JsPba}1A&*jHC1L>pS
zYe++8AkarXsJpl2V_U-qPFp%E=Bq767_WLsK`|NWYm{B)r*pwYcu-zMG#W3@eljvH
ze{a?>X6{M**I=XxH^8S3sZc85M;=ePo^%dyhkvoX<McW^`$K^WEK;IQPvAUmH$m^z
z9YJUc$0QM6+rU~QBVL+Iw|o~Qlc=)HtAm2D(O_qsRlDfr{6nOW0w_z8qf3a75?I15
z+7Q@JE*<VHPm5L$8_Y8;mVviJi4o`GnZQ4xluzF2_2^z^pnM_fcHM6wL)-*^{5y1B
z=%@wvFCxtl`<L=d__5K-lZU?Gt`^nRH$qHc$P4B|ZXaD+XFn8UvbQ`C&g#TunhP#;
z!5!#QX-oxMAOk6+i6ZRttWJ%E$yhDJRe-eagnv&%!&vdv(1Dv;&u~56x^qC30hYZw
z80kNl3LY_vUGti;F5@ww0JRDX=WdB9Df>_4_TyQ~Q?+-}a+szr-E98a3L{mo$qck%
zuNy1|Dk(a>BTUrafudP2slEbDTi(FLY@YA{Z>mDBXzErAxKV+>WC0_jud%j}DGn-;
zF+NLOJy*D6m>U1G)WdIZq>^ilv~3Za=Vsyqzr3)~8Qor=fM8654-DXUR*rDNfiEG%
zYSf=F)+=7H*}W3cJ+>3VfgG<Kn+F7y*h(1d2ieApDisJQsf87puAb{^ua2xCG<bLp
z+Rxpp2LZ5cn8wS4aLZX2MPBdpRI32$nZ7o7cufuJ>|3$aecKYmQB;!=xC#*LDG$u0
zD(UQLL2|V;tV2U>@9X)cQY(+od}rYI#FV%c6q)>9o|xS%H;p-)K+{#VCKf~9>D{co
zsIbvkZE>!8ss=A~Ds2cyfp($7AC5)|Tl~I=%8s9yLsRFJijxX~y~jtpE7b2;5<n+^
zoP`fU$?C4?mM6`>f=Vi~d?aSE-LXt;nZXVwnx;*TeUIV}Bl_<|$-*tQwA83QGRE>#
zb(~9#LnO!~v7zruWUw933f9z4nQd-tZyN6en2epjG1Os1eUY3pm`^{L8cZ&-zk0+a
zmV7;eJ?)6^k`6t_#)d)~(J>>16<{L(V{P?Sf(>=h0K#X8^QE`m*#cU@j%YanuWu2%
zXgU6wLe+Vl+jweGJt>gUbzRVKji<6(9__yMWu6X~op+LTMH9F6?pSClfl8FXH@DQU
z7oe6P$7*I4E^1|h|9D{k1}6Zy@kbqP&M+Iu@A7B!NH)i1QAVQ;7Bkf^GyfSf%mJy}
zb#WP!yX+T8=Rh$qB<#1jH|H<WXGdccY`1#%1dCB~@^dqW87vwa@azZe_&3oY@~#Ke
z<J6$ya%q2OSLfr}nvSMgcKe+tHEjOBGW+&U5aicp0~G{s+~D9CX++$R`kPxgAKKE0
zR4NDo89dV7fzz6k^xLd`!ZWcP`t@xg%97zeloWQ~OfAUTO}vD!_;r0Zvq8$d|8QFU
z%43~HK6tRZ^LOBMKx6SnIk#kDKEUi5CvqS+w>;8zTfSqUxN(90c<H9w_Nfy5fWg5G
z`$wY-lZK)KULJSD&#NOu(VbuZloK9BqDMHsBV3&!oKl7JgrMjqEq!eBDBu=3XULn_
zkqBGnqm&2Pry8B1$w*9+PLO*PQImmtjA#MPw89_^s`6n4CN0<}2X;PYFj0gGEW)DW
z<{^y0^@gJa8{(Lo3j{Ef5x`!juNh<N?_aY%V>*P7JO;0k84I@*|7aRLLH*o8r2KH&
zawtZ@haW!O;i=AS>3a=BGvzXm@~9hMS{ANbW-wp-B}m@x$kVvocKJWO?0xlqzzl45
zB_iMK{x)Yvm3rT@mHrw+89})PBS)j$kVtE46p%&&Xv(jd;XF+~QG-QMtAxx0LLq3O
z(lT7)C}Z_5u_Kk|+l;zx$4w+j;^45QLquX}>X)cYkT@IO#Sali7@l1~YW7rLMYaPm
zrAJ9qVN>Q50l+<W&*7ckb%x8Rfnj=qU0h5|mH5Mk8{z&)5LUc!MFjm*Hb;W?At6{h
zqmgW;^*EVI(!$MSP=&r&0rcI7HYSJrldB^=6U76q3y)#L?jJd<Vh&1T??M#9ncjE%
zYCx8&)P8T;NZsS;nPj1R;LQ(#yS<2DYYN;Jqpg)|%wjG$uGP&o{lLB(pi<{7SFW~i
z^=l7-9v%p!QEn0B8ouakb`W}SXspd%#u(LDM%S`xXuTDRZj0UQ4qzXQMb%ubjt*NM
zFLA9OcVTunK!0@<D*!YRbSS9>!0XJMA~IK0<vI?4kc-r$H_W8{6SgOv8y)XcLXkj3
z^>AOI0n~YU&iv`IX7Ed+)^Zw_Jb*>`vu~4Yd!+y>7Y9S7nid8R4QseY56hpyXS+#^
zS#YIzl`gO)v@P`X#Z$h(KplaWeRJ&HNe?VuELQ{q384=z5VD8-sNNqw@^ftKe;2=o
zvc35duJ``a%Vw9<t}X$CRQCzv?!E-~iE3q&IseY#(D1ECa1PKD%4|{hhE=X=ZS-2I
zYneQSIVe<sxtic$swhqNS<Ow;i_N6Ma|fI6XW^yWwsxUsa~DhB@MrB4r%04|9lz1|
z^CJ`?bF~)@?!76UAHt7pl}u?lY1kC==c#^J)WB%QhT7ar(KW_7P@pmnO_{sJqz;LJ
zo)79T3Gj}`%^@BH{qmN9Er{nN!i2IK!x_ISYC~28A+BFFic3vjOk`p`W3A3)+kAtp
z^h*J9+E0$a>6sMJp_9GXMmo$sQPOEI7obH{@Wyd({6NR_LPW2>Tf@tUHEtfh^CFXu
zQXfi6A{Q0!dHJANXv_=U*UtEp$BcGp;dSZ4PGT_D?#;Fg91@ocs)_k6X8S!d?vHqL
zw*&Z_>4)7fZ{HM#wT-iWe<oE@R_C9asP%$0MI|WG^(Aub?%MCbIMvz3+csY5yzw>i
zuJZL+R|W!_qELp@B^w7iCwyt^HU!e4sqrNBLsKI0RO7bXF`V-KRfa88#-asDnF$mY
zHyZ+jNbgsJbN&@!Y9Z!T=p5<@T=1w7?)w>#F}gzHTD%E;S|8#R3~DTskOZ&l>Phnk
z{7b=y@?jVIta}6nB=Jgl%qMXN@Amw;kd&>?99mlXtKSobn75-w4zW*_d2Z7r&7EeB
z@2K>VJ3!eYgLMfG&9MYaT>V+Jqb6wj-)@A#+;VPKrzy8A_2*TK9utRnEjLE3y(rtg
ztd+pgbI+BZyIFa`#POCP2r6N-3sEdFVZoa^o4;L4nvsysv8t0G9`g0R_Yu@Jzu)gf
zF1A5(M@;0^L~3r{=GBGBHob?xtA~1+D;&#!rcL*!>HI178S8ULDR0q^?Kq)A7f(u|
z>42&;%lC>;&o>DQVNSHz6#+xVZ}H9`hxVP!yS$i2(>2_P+afAwiz#~KX8G=jL4Ipf
zi^^xhbUlw5zY#kkG(d{DhhysKB2IjAtk0e-ce*wfT8~*s@Ml_P-40REC-#{^^_&|f
z8fNw`T$no_BSf%@la&w7pK3Nuu1z~VtfHV}PxM7YPJG#Vb!<x1XAaI*p1RtNJ33A)
zwL@7(b%!I%Ij`_`OEJ{)7GDGdhurfJm!s<{*3szkEYEdh&xj4~9)|l4J5~H+r}qyY
z?6TLMKFV<<GX0LbLqiE@ndO{MhmJv;;VOISiGsNQk?JHBPh$HY41Zgb^TD4Nb&Q9j
z@kVyCd6IWfYZCp_{ksa3YFBqKWWURORGyqY2-(t?BWc;Qe4W<+;pltCrx_L3*Ay$A
z4&qQVa_XwULd@9<b4AW-cSn*u<XAb)q>v7sPH7o_O%p<}a5^mMDqo$Qt0>cAzrLzo
zZS)&d#fntT1aCVH8@9Hy6g?g+J!)tvjOYH+PIOG!FV<rA{_}f9=_K3lzWrLIH$?=~
z2is26y^nt`doKsOo2n+G%(|mSb#HNmc|0phj#RL|)!0-sOs}$iy?>sJ+l$6>YF$+p
z*Vq1T>WgSr_UzQ8*&l(TiIKhPN%(KcO|=Ydrr{2xbqw1TB4xHVOG7=57pIqsgvwXc
zhEiRhhj6?*_?^2|MnVg<H^Uh3e)NM#)6Y>5?wrmIdr*3j27Ia|?uqaj-2yrX^PK+b
zmQ6B#HucUvC#$ut?Njc3q)tyZqN0Y-TQRNWNgP#7P*rQ$yIs<n!JuvYVoL-qrqq4k
z=jM!`;5{bG@*<1xH4gVzHLuO+1;jhz_f42gOXJRjQ>yhvvxXBqtL;6Ll&5~}x+rAO
zYVdr5B-hsAazdc~pgp@L-gVRUMN;>A`x%72?DpDLBd<)k52tc+tMbY9pf88W7&Li;
zuWJv&`=P*_Xqw&?#aLt_+(A#mzR$A+Z+DjZn>LJCs(^3U{>+>qp1thwsYyu8H)6~k
ztjSA4b+<=<=#!#u#XinpS&mniI2eDGluJ+(|F8r9^TKy_PU_>y+bu$M_9{|hpT>;o
zu_ANEO2$!`Q^YCkI{IR~Y**{*@pWwD19dld(>?o>2AIKBKdxVc*fAwDeE6?(-}f#x
zvB*y2T+TYlSI#ZFa&Ihne5Z{}SnN$X=SDD_SXv4_A*`X6-$+5_v_pANe_jx76Yg!F
zHyd3^k_?*_Gu+>-&NnD`xz`oy*kmgtFRux-0xG;AgQZ&NNTCUYc&h`GVDa^>^mEe%
zb-}CjLupK@|FQt1anvEAIzoshRrpGtQgw^$;T{-amE82vfs(q{Lmqk<&ua)bF8tZb
zLzC0l>fx${LxM^;`I1|WxhompS{i*7)tyHW?U+W~LKQu*7$)MRz?fY{c%X{$Mj!UP
z9CM^N=%M8q!+OH8<q~tw@XRWXGs4}yk4*~koE_?XR3hW#cWvE8!T9@=_~ul?ZPJ2$
zaBHwzPf(B5p*{FGU~H~8kuhhXX_0+ONmSv5ynebIg0Ah@2U8sCP;9xll&^-$$GGWF
zmI$!g*XW2zca*8SCWv1;J3bInxs;EJT3uQ7SBfK%@!=RYP{v8((kQn3c|TqxgrEOn
z8QD2~@#2O0Hocjmri^P*`A$T7`SZ*fr>wT8O5q(hlSZkc0%X*>w<gwsrefo#R^Xdb
z^3Rhw2-WdpoUBhC<?om9T<?u@gm*9U%9tj0a)iFUs>u?9RWmngY8O7F@r>Tg-UP`J
zyt$>HNdrXnH^FlSD+$KAH^J&~NnH4lsr8iZ7VeD8Hy8y!{auESz0*(JK2u^>-~}Pd
zn_%T1_jodO*=okc=g?j-@pFH+)PHDYp~?~pEk=WsMAV90jJRG&oTPZjg_z^t5+<LY
ze6Zz$tsIprHyhD^em}0>rm24W=jz6i2T~!ZqJrnr(3mGst|}zdfq01Ws+jxxQdNX#
z7JsHOaais&zrx@ry<+3dh17yE6ZUtL?7hxmt_AYcwZz4?3MBK1R~+<;xb!guR#vu!
z=MS|M#_*Z3D^P+uf<^VO7s-RVmFE{vt!|ZWiC7cyq3bCfW`+Bs6S{^h3AvP1T`l9o
ze`J0u1#ZmgUMXD9KUL_`%B;Gb(qJ)R#*W0R5-k$vp9sz?D1H^0b>=_87F$t=KM;{O
z@=;#}5#2&EcT56j=TfAqmIQP7M4tMR)|xwa=eTM87oEqL3A2g)@uk{AlFl=JCbUln
zGrwC};FTsfSiiKWo;k>R76u}sUcKj5Ey6Amwg}i!$x45|TC3+LSx}$f)cy=jNk#^q
z2?`O@_497)Jr6H+;H6>+*g_S9&k`<j6!TO?3nag)=D83?RGH5yNJY-7?TEtqvAU~&
zO}d}u+(SBF3{9ErSVSrLtY2wltSRbqozLMw?5(Lt!WlH(dWdG-UJ^=XEW*{n?)SMY
zZ6x>b(Gj`L2V{<gnV?eQB78zE-Q!0m5Ae6A)@iy1Xr^kWwH$>bBShuM9L^ka6P*1g
zNq?)U{4O?fGGrGe(d6rg5XMUwgcb)cgLU!-@+pm}23Pv|T#B|wCH`1|q^=9*RGCnw
zg_2fr2~_rHEfJ>m<@Tw*5fi@M4ANLVcE6@ZDy+AshmwW{A<E6W!TtDhK5caSO*c(D
zvt+pnSQYL1HcrzP6AjliLh+Yyxy&5uZmq*J>@tesd=C8h(<sJfFdMAouK8vOSiD!H
z(!3Tql6~k)MxN-C3;W7HuBrWpStJ?-CzDij8$8A2?|3YJBDzGRI>eQ?K1rHGFyE&s
z`ZFaYqCCz+w(+n)^V+Y^)?QYDnpu*VZ;IVTTE-^kW{g(HV@9w~D)A0#ur@IvrpCat
zbWKj>pE2HU)ghhgQqqxMuSxD|5FI`~Hw_+YXCI%`F?%&fq~j{$hxAeY4|;vF*e)v$
z1q+d0e<Dz0dSpG)WV}m=*o^6MaDPkmykVt^)L1Oanm@1zj$_W7-t)PqV7lQC@7}O9
z+}=mdSg@$vyg<SN_Y~iq#vFO@P3hieCXTOv9E4Q5J_V|@FEc|h&u7uUJCYb4;e;(u
zW}k87nEN{bV23Sfb{(q7Dp`>?22jREdnhCt>TjQSex*bf9mpcUS#jPK^2P#|vL!P(
zJtkp2Kvf=U**QEPKeR7nU%hy}Dy-PRLlQI{Eaim3@x;?VJd~6!ZmXz}-Y4qT)5>YQ
zK551ys>IUMXV#EcaHpkT_W^5@BA=7*MR;O)>aQ^bC%+ko{-l{>R0CZy83dP%;qKXy
z&KHD4-W6d4St5bU!NmN8<~v4vDy7Os%s*`-W4Lm0g8ceP=yZ>}dZRo=>RW!oZEX~b
zKUN|ucu>9GlJR6^vi*BG<RRu}U*#X3`%u@<gx|RCV9I1rsn8a>ts9;}OW|E{EQPr$
z{I#WMNq>y~M*Ut7X8|qRcI!o*JDi|utq;q2xkqaYaS6K_OHyThHr+(J@C;rrc}i4t
zPByVse2yXPhVkp8(9oBarecX#Kk-kKI37l;cBDr`q&1T?VCb~iIv(!W72O{JT)t5_
zuSB}`nC|OQ@)XV*v05D)#Jg=JRy|9w;j_VM?&c1!_90$AgAtRujM>V^#8fOru)o>k
zQiS0qei0Q)z4nJMbsTOgRsO?k9mJ>A?wwWgV90f*R2WT)60_yB(J)t5cxfjJHK<dy
z;P<2B(GT{;B?da9BAegq(edNPA1|0<xHr5L-+yp1+jjBcq7gOtD3hrsjwGkQn3$dJ
zPJgv7`d3K4=r?uA?#HaFOMiN|a2znl8aGrH3Z~ed>9->MDv3@WUA4+$_r>{iOCReg
z0#X6s2S#%>WIvOVvrLm=BdfD)Cr1Px_be5KMe59f`N$aWLrnm}9dlV8MTrZEK2_!Q
zSS6I~qHUe-3gJ1Q^spFR`L2BVhc1^{)t|TOI|~9y@v(A2m4jW=t%SwTYerfF#vl?{
z(n{fs;bEhCoz!laY9{&Z7RO3VgJy~STLa9*rn<=TFLq!=KH9dqb<4A2KTTnIq2!4%
z6@k&DKW3?)i3oKi3=Fm-b`(gKdanMii<MZfm!daHZy$M2+skntbAdf~xjTlQ`JsR_
z87lFmZ{W%XV|~N@`V#o`<!(>9(d+1}^+Hz&D;(0Ga&Up=kMW&+c@r0@%SSN9+uxEO
zbz!aMRn?Y^g{Tb>B2gh@9-2kG50WFq1=W$(W~2pB)`z<Kcs?nF0zsNZMWKhP3^$jY
zkBo3cCjyEJQ$X_Jdxd?8UYV8AI_K`yI?9+Tnxh*Rf@`cQv3HHK^6|!lK61<k@_zDC
zjay-XYNC^m_N|_yBGn-y7G`Qq%6`*&!dbEvfUNCapR&^$$R+RKod^}CGKP-%cxK4v
zQT5WT_d(@iU#;wf4GgCQ*cxP%jke-+AJ`_?lR5{5(C^?9Qnr@Z)01i?jI5QZwwW=N
zZS*7d@IvuWWCkWm3i<Pd+%j?eww-rx3?w4kDenu0?Bl;XQ-)eC8>1xNt10)$h&ydz
zr+}HmMGE@Vt0)y3(4p>Xu#vq<K5}_RKt<E{xntIS`gBw`Y9mX%++4=t^=zZ3*Ej?t
zGlMBi^ZZFVtWRx+XtnCYrqSRU9Z6S-1bsY8;>B&pS$hc#SHSsD$owrfBVHHF(I)3@
zz~!k*<rxQn>dB?VRg))N4V>O%Q;|*&?bG(s@K`#g&F2WF*z0URA66SM1UgREh<+&q
z6+O&eXKl4UOY546V~AeVY}V-1URzckysFl;d>+dYj$iYHwTV*o-1AZC0r<|E=EhbD
zNlP+m+<_-XrfsPYL~c2+t*6~I_R2kjwWXZai4t80^)TeIUQcY8?&hwQ^(@WORuArx
zq@jQz>y}}go-y@O1a_K2+3{=@?g>^1apqlBwMW0KP`M*&5^Ga4%5XCnuB{t9ztrry
zniM?CCC$k8KFbz`l@~CZbT)E_c%5mR-F!wlB-nh~1m3fmo_zSoEUcjdXPwNjTZ&VV
z?@9)aa4zZR_B7{~iwJL`?l0`0h)y=J)=lNZ9x1{bcZzxs=P%qr&;YHqke^q+nZxYT
zU@V?>Cy!UTxMzO(xz})wvOASA09~EjMNXR!?~=}Awy$htay6uTZTai##VLucn*iZ_
z&XdZ#)Xs2gCcmEY$4L#XTko&*TU{k}dMD+&B^J77VD*ULH*PYnR@R>RPV1#H(fI*N
zNsre9Z#2B;aJ6uaMJ~XOqSp9w(+NMWKP{#VSt7l^^xa(3mWl^#lU1)Pr={X>so3Z2
z-I6&n5~H4CbA(P769P#4T3R^o70-xdYZ-J6%!>r0=GM~+NIckGMv{7>cFb1A4ZVt&
zzkjjaB-3BoE72@({D%4L4f;Et3&9f;X{}eonZG&8qDSjP$2AJ%vog*b1h$<_=^Q*N
zgz*kz@I^FD<`DaTkv`a>&qsFOuSY?4%t7jKEGo4v->i0ofOp=vO<nfmOPX<fG-xG%
zj|nx&lr4C)GaOj#ctYx8Mj~%7p07yeTr={?=;SS0AVcLS<;tX|_(-v?5>t!o=R*fF
zFgep?m!D0A4p*F6aTq)U_u(~-He;T3?>l7?LkPdo`Jo*y^od6Vt0HWUvycYJXzUG}
z50~9%)F*C#rd-sZtuOOxt_+pW?`l<sM7akjJ8WnTbuM&x=XxjKaq5TmkDK0mPrNps
zWEjd%OB9(8F&*v2tlqG8nPm8TQ&tjQ-qWlM6d=#&FsNCb1_@@FOyvrKSqyjTLc2IU
z8x4B-V5#T4q{knX$D_8xI5Ep>+h1HuS&3T~&;qwrv8P@I&+<Awtr!-ZIAl5+v7uMM
zwMIQ?Zc=y@x__E`?JOx{z=L1JJ&^JWEwI182|CFyNBOZI9$Lr3mq$P}Z2SBfH-$_d
z-4A+3cz10ilMr-~^AB8KW-^Cow8ml+J!x#@&LZ7kOFaK2A)Y0J={}1lxafee?x(+&
zGLQiXTv0N*lBc8FYlPvggv~Al-CB5eKB|jZn#gQs!BSb93`)kPQ2%?5r1W*;68%mJ
z`PrioMH!Z{LpX;&7n__pi|9TPB<g1vN0*}Ah!zU=%}tmbm|_OI{=@(uJjdot*0|*?
zSI(sz)`jdMP^+cVD9LL`VUb-SosLQ?%ucdvUR_n&+I3C#MC>SB%f~2<u(hBRu>ItG
zHpjjq`c=ECr`)y)mSj$2Z39n{71kr0cfU*eo2k!VMUZ@HH}g%=iI57~zL-{&J74M~
zqdOl;Kq5w@^X`)l6~tU-6ZP9Ul-(DS@-Xxe7QCpkxep`7PO>`Bsrvfbl(n132Z>-4
z>6Q3%C`wH(T(65|mv{Gsx2M(n3`k9?_AHz7|G?^;CSj@oFTS?!)dpjQs{d{4LdUZj
zIH}g_%2$<YFHC59CE$Qygb-ruuCy0vm@DOT{k&fcFcyj$qF1*|#(@>6^aTAq?qlS>
zxqSae)K>sS{rrC;jg*3bv?5*79V!CSCEeZKjUXtkG}7H1-QC^YDGf(AJd5Aoe;&u1
zaTtfY+t2R%{i@wfvN>@mt$Q!(Yz%l^jE7J=9G&cq&54txi}t^P9`yOYZw6*cWAHow
z_->}JQK{wC0>6))YkQetTr|5gY)&J}XRlVV!is;k>Q!l!g^J{4^<OoidFVMf!j?4w
z#ys<|NqaFc@AFanbh|0{88QAzxGj7IW_i9g1lJ49*N9j<m=q`pM<=4qMTH%MOTe{i
zwQpxz?}6y-RV&sQW>S+u<RIy{@@hg)(x&)iXLJ#L<;!`Y$Eq|U&D!vV4bQ0nEqxE;
z;q#Yc>}9&1fkqi;7BAKR(o;FDDoI9nNmK#Rx`J>tsa0k<!1=p(k|)U1@d<hMN(=>*
zN3ZC&o7~WLOFY|b`ghzz4etLsA1si~dY%z;)sY3TB0Z9)Z;>b5l*UPA?mxoGMg&b~
z%pj<%f*KqbN@+9M7OM5);Ca)9PSpo#gbrNYx#;1h>jdr!2fXzq?~<iRs;ol^gH`<$
zIkzh0%BiptMtzh+E?DIhWHufopi%Y+lqPN5H%HcNG6kw*utEcw(Ok}yzV6O}tS9>Z
z7x4j<h|BxARCsH<eQR-VKc30iDHUCv>tI0}=w%X24N^s#@U*SV4`;w=F@;=T3deVg
z@n2ElqY$c|{{3|EexWOKn6!xXc=EKk)#46{36!%|pmq(lR{j3PE#^+$FVjBMyOd9!
zV)6j0V2`zocUDxC?MTZ9>+Qc7UKgoyG`+Jz!YWQ1who49-BTSbdQ%2&OYFHrSd?;b
zyrsrVZZk1R)d(}VsR%R4NI=PQTo(Orp~C&IWRrtB3g*XKt7~tRBY|hpbq|&Au$>=1
z9M6)sxmC2?^CZ5uT;Hd#xIdGjU}j^7r-b)irwCg|7NlltlQGWKi0Jjjx3Af(tltqZ
zOjVp0Qz7k6?-l?^X0&Yjm7Amsp)n*_bgOEg-9I^UiAwhrZ~ykdmBYIt>Mn}N%KfrQ
zEmDElYNt0FGZqVUSz?hpIHcH{*mpfrVHBG!@`()^rmp(+YV!pB=yyy9aPpTFYa1kA
zAid=%N4DLa*bl-&KURl-#S(NMUdu$pm&60M6Y1`KgY;l2WWK-Lx`-0~!uq-Vu${pm
zk^}y!UE?LvN=upYv!4yp%?&kEL@EuTS;NC{!3ZL^beDxc4%Y(OJ#<94?qa@Oce56H
zoEtwMx5EdJy7FS#)WVvgwVsJs<mOu=dSp@Os}G-0mR>%<J6Xs*p8v~F|4BoX{`j<i
z^5>rl5+bQDu`wGt#FSQ@2|%nS=Zuf*d_>gg5wx=2hyH2wlW?N#)ih>kSd<pS)ugRe
z@6<$PZC~x2N>wahn@ka#d#SzU1hlZLgB=JHA<%42ywN<ib0jF83rs35DoAhOs|BjD
zjfY=Q3=Jzy)S(Z{YS&5c|43_yBR(bR(qo>+pS})>Hnd2&eFFnW(uk9!&k&xYPEYT7
z=>9Z@i|K?J+97Y#w=#nx>$i@d9}N8TF3bQJg!(<L;X`|hY>cvoxupJ5B4SHxv~^?x
zkvio5*%!dPfC8I7W%l}Zs=^KO{!i=%f48c@9@6G`r@b81r6Vh{UXf_R#+a#Lo<yT!
zA1L;>c$Ni@gsr15`+DxmqoIa&_kQ#&ZQWupXkJDdN$-yS4O}tqe-6YSaK;w{L`HZ^
zr~858y{$*TvJyn?Pi$XZxK&mI*N?+UB&?}-`6zIfUVSG)V1*C&<<8)xr`q#V;_MwE
z$OPy!$6v3{9x*FTu+T(ywv}yd>5V9l7!LG<@%o$;9;kebw$o8U6$9jMz46EUkJ5T>
zMF&WNo`fW=0_5EfU0@FBGA9(yg{7r`mr9W#<1`f1<xk%hF#J$-5MIRFvbyPR13Zds
zs)vr=cRl$Bxsz%)lGi_>B+jyN$=!3#xYKh24Kjpew~j3a)W(b5`qW<p^(LKQ<kw?L
zJr9Lo(A@0MoESq{xsz*HBoW_9q}-E~_o9XSx39Q`$R2G!e_x4M8Kek#{3Ir=TQn#-
z@FyhW2P=KQZ_0R71nLwuOItt6FV&CptD2^Q=?Uk||J)lk`~QvIFYoymU5$@#Xe<95
zoMxAU_lv?ao^AluddD;P$$)v9jB5c^3wL81{dhL4r3b;$fbg~#d5yGZJP?OK{gy~x
z8E%vw#SXG_@Kzndm?A|&(E_vkw_|r3@gXCO4*2(2xaIy*<&U|QMUgqEk<Bh>NSssv
zaSi;z6qS#Qm)+x;rE-RDse6j?*ANXeX&+u0P!ebk;#AQap%N}V68jKldj({;^Y2CF
ze|_>9_mlXzYYdE&#P;f&Tw%c!?t4B4ydi6>Dv!Wfj~%f}7_=L*i~Ad*JpugS*{@>t
z3H9Sgzva$@<=A+^#Cj*lE}B>A9*kZYRbSJp;*EAAZaCq^z~|c-&t*)h=1%O9^8Aq>
zW0qa!=TC$z`u6o7t6!L(9kc`Fja2wRCi+AK*390p|1%9Kr<`1r4<8;(aRnc^yQ{sV
z#*(iqwgr-Zg>kO0&0`fMVc)b#mdPA!I1K2W3=_ep_J8+UIN8;#q&zDY#*K^z3;LTH
zg0uh3(kOpgNxe_avhx@a3M4r4{`eH`#l@O4=*lKI36XA%<U4~RDe5YR50pyws&jMf
z0l454W3cFcc{wmNH&1*U*;Yq6;3EicsRHa@UMo>(DV{4r-<yh67aNAbt@3Y}jM2)*
zj=VYj&JP*;3j;)fRet0HhhV8&{jH6Upd7xa-w?)NSRDLi+%>tNQuKF}{n2n~;_^$N
zm_c{L5AKQ+O)@uW^3-#nDEl%cphazDgLg(3bH|gD`-4i01Bi6rPu1z=SiI@<nq?<y
z)JwJ%uSV9_F^0A#vsNNtR;nqS-p`~fn`hsjSOZBy=y-~6CLZ&?AvKl?ewNNq^90?d
z;dA`20~Hh)OY7afCr^%-%6)$QUl%tCD2B*{JdrUREL$OUqhhzvqLh2Ff#C$P1?Jce
zV*B^A9I2YrLJzLIa`(I%u^9tnm{fG<+8K*21u+E4j)8WF-tH{n*|Ld>J_V2d@4iVP
zzN4xp`DniN`Ahe24r`I8f1EKSE?mRqR1s(0=aN)bU4}u8Z%E&lfqq!!SsD=q=l#oW
zWh}#c5ta}CY~yu$_o9k<TR~r7=noj=BZ0J!)6y>CC>@Nhzp)81CzDqYCe+x+y{<7!
zQLD=i4}^iMzOfQi%~i8jBvEyVe7;)cd`MMjOqZ~ZJt(TG#cYZs;}=d`VPDz!&YAdg
zy&Q@97c&bo2Dhr>N>^aL8Wp-k6Kg|^E-&|(Ec*PVYGW<-h*yVKa;!7%2Nxmz%Xv>7
zy5{0P0i;GU(Q~2Zx|0UYO^+TxklziL@D#HMug{{nZtK>^S$vfIGEhti?KG+0+Q}Hw
zqFS1h7O7F;j0_=^(f05sIm-D~a}~e<eZ}kd;^+)Fhlzq0!>x_o7AC;<W%-RmqhU>$
z;hBJG0jp{_@8B4e06DF%0g<tEByR<tAdX2jbpckrqMZy-oU0A36D~4>>b(i~i58>p
zOH~T%Nb05bWvT=3fp5y>?mg2ps}LWB7n3%YYir|7mT3>~c}N=OJ&zc2@@i1}aK~bk
zz;rKt!5)2d3PBIGSE4VUwH0+`<!=X=tD{G&zY`anqQ>j7sfYS<(ld#StvMUiu13vd
zseF0-`@ft08<*_=jmrShLee%@9Vf9UCq0UD5p6GnJT3t+#)4vrVZi-}c-C8mC*^L_
zd5z!m1pu_(txTT!C<^?#3W8UQE3a1|_ZCorysun!NV0M9AB4rYFN&wATHf6{lb3mS
z2Uu!ru!|2bin?$t-K#lTtMw1fegnSes5GVYQ+&I_Q5@^v3CZ9%38LZ2ZhB762G&Lt
zn55HAcFuw-{X%L>s1e$xc&H376;i^~u*~4wbsH9nS^pX&cWj`q2B)|;-j1^W6;O=w
z5%(V&EneFD6H57f{FJ(v=5x{H%I;PeFf|?dSsH7Z@k6$T<X#f+q8EFS)Lgqy+!F01
z4r!8xx-@fK64c^QxD6!_rP4#-K`Fj_=WAmFslBNBr=TfB{!^&m{)yi0FmiKk=j{g!
z;{3E`n?l*!kXLw?im)MmtvqMw-1XSWOg=8AAN0!z0(UGS*G*vvZw(I?dkJ=RlK>7c
z-X2zW@w2{*;4&Ei9O@@a?~+Q>UwP_M4rwvwRuU9UOU%6-ZbDG!s@-J5p}<QPR2IZk
z!`zXCxc>kqn}NMW`ZIc<bPEi86Q2LF;0Ev4wB{ST2-|1^Oixg30+ee~6g8EW)$yg=
zo3%`V$A61wwb>KYnAw%w0cZ=<O{W{TqPdor|B&E&qeQ!S(^KL8G!grc0U?hTf*`)U
z$jh!*J_z%BZZYvIHp~~rVOP~CP1}aR@b}2@?GfV>lbf<#xC2-Dszz1_(LFQK-R#hn
zcsOVH@i61xX~!tGVU3v<wo>?w>|G*lMuY4yX!Trmv;p73!=JRV-Ti9eaHWo+<xg{~
zil5ICO->&&yCLm*9IA>)IH^qh;fBM*n-doqeV7hrCP&Ygc$ie*6hy2kAnYeJ$Itky
zKM6!44NQ4%9i=92*EHPpo)_*Bmn6zwI*<U2VsV;R8DO<i^F((qk#b_4_8S2R5A5Y+
zpQ)ycG1R8t-?&`sWSuJD#Y0x~*1J(rOO$DB#}+78c$<#j&}lscributB9ZX?w1MR;
zDjTM8W5J+`wB?F_Dl%90*gGD#p~@U4Ph+N`O#UA+HtOu{k1jcs%j`(>v2-0-?ym|{
zGV~k$1l9+8*e6SS1w1>f@xbLPnno&blX0V=<Ru9psxJ670PR%GT-`qF_xr9UX=%N>
zh?=HeJWBXaHRJw)7Uc+GF)i;+<1TSaY%j3J)*|BWI`u0rNZE6-&58mt<NQJ5g#)8_
zDzxpc{tDe5#)P#{BFS=?h^y0AR3Hs_Q?LiRX4Gfr@ZHVCr?7B^L#<_Rk^q?#6Lk60
zt1VC*W%WX8_v{cJcjzwaB%KA>_PhJ16+{_|UiYR2ag}I4aF(e1UrX8I+F{c6+3upQ
zCPf1Fq8|=_TOjkR!xnJ{>)ZQ_*S{3<!1WA8-mVeZW%t3+h2D?OhGex%$xIZ;bG2G7
z&1ycGEGki9UhS$6SHsSkq|$%?szw^xasC@W+$ouUdN7d4w()14E}3OeOL0fdA-7)E
zmSNLW`BCQ^fSvZ(9g%sEc*87jLVQE6+^oASILeG$S-m?oMDNF}shTs$ofjRUUPxEq
z*j+5>Q9+*K$SBT|3F@%OH|arO0fNa<uXgL~;GBIM_3bsC`0VBU>G=|=&qR%3CaGqy
zaqjyA(evC&%cUI)FE8k0)W+c#L6Ad1JFVraMeWP&7W)XbG~V78H}aru<~w^CZHR+c
zjT1_F>q2mv<6TJZ+>OwpJj&E_ySBmTQ%$^pQ<&Jz!I;@e!Tk7J23WRci(R3Zp|V)<
zsAH@9otq-NUv&c@3NYKZJv}#&<<ELvOC8Ro<PZOs1-N;<M|7ffNyBMa62XsuxV%+F
zsBrLo{M=f9&??mrT(Oelb_(S;X8(pS{Qst3&SXp2=ZifU(@p51*e0{LxnkUE<+`jv
z<{t4q)S6IUUjYxC(E~CMvnste;UEBFH+6HMLcgyHMN_4!TAk*Jv%<cep7UtGZI1NX
zu8<GVbwZ-}W*+)h0J_YPs${z|O-!U--#qpAiTi*e0Kz{3`UwK%cE+y|@Q*F7-DqLj
zlaqWzluz5Wj3VRx!#%}ab@3NSkYy^rinC613gwWurZ4vJI0ZyQ9yB1Y{m&(%MZmQC
zDwe*CgT~tSQNQ!2JSp#$_U$hD(G&1r;$!)BsDMbj`CLmZclo=y5$fF$dFD3C(PBWA
zQV2ma5zD%Q>iOLJzGJrx2DGy(tIi>-zF1Caf-27{U+e@7cFWYK?)Kt8h2zEGZ}H9=
zOp=aw9Y|GMt*G5ma}|!g#XiPiEu(tAJe4{d!3w(?Ss+Mf%Fp#DegAVz^Ju~yt0R>`
z2u_y8^<NSI%0v7TSR)18E2esTLZU^0<f?Yv-JJ2JqfTr*-_h}hsd)O$FrIf|WnOtT
zowLUYvRP<AlHicRV?-=d!k&N=NL-Yx9W;I2MXfz<eOW=cR5W#A+iXR}`&J5n^h#Fc
z-i*Rr5#;u4re9v(N|%LYP3arz7R9A-{*FyVUtCyCp6mai(V>M7sX(L(9UI=+S4+V3
zwjz70sn&KZ@3PBepvKHn-+}y|vgNM^INIm+qc`R>_wxrk)yG3foRpCZ6I!}Cqnxc6
z1JP=H3{0WBGc&FNKi1N}LE`w&GJE@nf=UK2d&m<prRT#>hVQ0GQDvy!GDNCAuNqH&
zUf$c<q(_Zd{g%Am6|2CglzD+5jy^gsm?)i~t4Te?-AM{@Gz56KS1L?#*D6_5J5KEG
zCj1&?mO%3E9q{QTlWy67T3KnOfBJa;tyf20<np|<^zZ(`w-s|yQ>&K%bQV{dx*TyV
zFdHNCh`M9YdnXBMh5g+qCe53`(;6AF;CNCb&h5KOjV_RB#ZPzmEnf>gy1FN=0y`(P
zjER4sPA<)Rt5|V2UX6bL&dKtA8n3ckWr$al!6Kn>WFF%XxlNCy-2JdlEy3)-QS#x$
z)=;lW!4Ud^$Ri%D*KL!8Ut0-bK<0-#M#~axL;HO^&n13hV-eBhQU{O@xM~lO;b+=s
zVCH#tj?fP@d%1s0>k(fT{AaJR&yt%{Mg~eHaFlvn+v(}+_xAK$nP+dSC34xO=Y7I@
zLOF_EtOKz~xPskt)p(`eqa5_hIc=ADE7lP5lmXnHyswl$VA-71+yvPYTV+Cp^^Nc6
zi2%;09E$H9vYsw+97-X#xQ_VFpuM)T8yxJ6`cF|~0Yk6aC|&<@@%8mdl9CiMPBV{^
z>gye2^>p+WY4moF)mS#5NPaMQ46Ed@1_%L@-YX2}{o`lx5P7$cMWbuGg#H-`q|rBI
zLBq*ysqbN1CYK`<9$1sZGTf)}#jWHijR6dPhcnOlhK6$iuStBhqstFWs0jjm_YoDA
zux$CLcuY&psGE{1f*mA#g(tzPCpLS6r){ct+`oQHVDN8vZ0R0wjqr?H*3h71-yn@A
z!XK=%B37OvHc8!=hF7Q9wl?M;LPF_v@1Gg!)<9V17I0~!^qni>zMtDZ;Y$+`mjPYk
zPBW6+C7e`nDD|UXPfw3Y(BYHZVM|=<Y`Zb_xFnQv`vbdP2eq5CgC;!K_?|st8W?XN
z8IX3n@eBkZW-Ew+tsROPmb&-72_Omo^}Tg2?AXt)w$hG-Kv;glSh2^RC_xz2bw_Py
z8ir~0Kcq?(_f|2QGk?3Xb>{!|D)NiJIBf=?>${MIb7alV3!oEI5AY%Ss1>Abh8~#k
zHV|rdT~?ZPu(rhC=S2Aj9EJzu&ef#~%1bWo&>L1%nY=RA!;bbpbo72ENB5q%!XbV8
zD+$PYI4)@uB*nZP4$0Wk^qalxH+o8&u?i^0UqPRk|I<gS7P|)qotcV<)XKnvz4iuM
z*ZuRvKz>zGAuu*=95ggEpM`~=y}-s6a;We~b7?d?gSXbzeY;ram{Cztabvt#@0{a)
zc{npa&w26pU(Zc~n&P)Ww<uZeY6C%5@zvI=RHXvjV@Jre$zmOsTg?a%2hQ9LUIhe+
zpNtLb<uk3}ZU+X}TV^OqiC}Z(Mwh{(ujVww=2O#Ddl5$m0lhnrVB{q|ZeQ?u)L&fv
zqcxxMSUOrlkwIKzJZjWuq0ptFWs<O03W=K3C`%$=6FrNGMD>$06b!*TtKB066*MFH
zB@^_g=5*<?yr$`8?66>2U)u`<cz%_cJ#DSfe!AKlE<7og*hJBL@jX~k1oQ#|$(_M?
z<xU4`7^DIrT9EGcb}%)*)H&{nEH`=XjOTz$iF9-Sh-qGC8<eUry+(87d^O!5A_E<Y
zXU*ZMoi8(#sIp#`{@@c@S@1L_Zy7)CrM9zHT+~&EetcpMkdU?UT?>zS0}eX&FOH$I
zUM<o)?G}5PQnWZgZ$S4VTKej0HO8U&&clYsStA7JWHpl+pKh3VSft@Ff7qSh)In{*
zuaS_X+fc7?PNp7FYnk>#PM3=Aye%L;$FXZ`dX&<MJ^6|SjBX69tdRo)lHl_Ub(4Cd
zwG0bB4n)y7%2d&HRHlz&H=yFu=W1y+xRjDR|C{%o3I%fJ6LG<zI!xMeHvOC}#U3--
z$g|P1X7HTE@N+^kn45xBgW3nDQEKf@L&_I<xx01kuA#r8_^V}#z0m0?Mm_@~$JHbd
zmZ{lIEjgUgkc$Rj0`Ln0335Y1s_f0Ku5;-bsDJ^3#Ysvj$o`|O0gGsX2%O?bmM){N
zhIwG2*tkVIX=zWmN|aP!d$2anH!P=bf9$xm-;1_6Q_Hrp9<c=PhF#21oV)=@6q{&t
zboANoWUzjBnC6b`pCnucH5K;d=$Z?5H{Z*v<Mqk@Bpyez{eR2)Z7~lR;Hb^mmQk$j
zeMEo6D`xbE0oe8B<*2>Y7bTRUVu#XbzQ^AH<Ztf;&Vc6Z_un!TwertJg9xX7kHQ{P
z04@dSE<0a|$kzykp6u`#+Q{2`%MORbJjn=WDjS_+jwi}{*+cCnMeIAZR7cRY#Tr96
zbAGQ@<y{N572H#(C2sVjxXU%4U@ual>#VG2*k0HclrQhjRe>~57-CdYE>nA;JlF_O
zgEXtOd}Fpwh|{y9*&U@n-8Om=Hna$7l9P7>N68bcoem2^qz`u;c$VCc_^s*>IDCD5
zN6acg47snbucxnXZKc(ZUb~413F!wNQRRzR<qRqpp--Q_NPeYs95G>YtiXNh)HN6W
z%Of!vbb31LsuXg!;=U|5lONGJZ$DbgAlt*^p}!(nHK5t=`7u}v^~^tcRGy=LMB)iO
zuou3h0*)ERp>$L?4$bYV;$sJ8YKR6SGuBcyzthm*mY>Q=@;bfSJ}U*yQRcS<*wkvI
z?_9p+B4Q&c)cyz8QPzX{i@d?yb$ags0xGP_{J{>mwF<O{s*l`&76<Jtph@e+Cj21}
zHT%bW5~odlE$0}3TwXT2kzxyaK4FH+3*yln1|CsfwA<&YGswjpO$W=-m^eo10nR29
zYSV+NdvB9wLH=sS0at{5CHRN^r8)QsAq@-o&H|p$R;0=P+lzyG6Q(rrthUZh1Ux)E
zaPK%ir`5tHhTKoO;+H{onkH@EQ6~F4JJE?IpPB`-E&A)Tj@MZvTlqvTe}6{2B3^^D
zuW!SCDSpB<q9xYm&qW>Bw3WWHH`MoYIg`cZ>;(~{NXGcc`p`#Z2XBzB)U-N6r9{1R
z>8g^Zr|vZeO~We`PXjIqWWY!dUA4sy!_NF%HCIP%wpY_6RsQBHVm11C>L}4&u@KDF
z{_>)Zx29|`J^!vIjPL5fx++CcJ3ork6c&J|*80Xe%UijqtTCAc>H<Ho2g`Xa|3`hG
zp?+CGByQ@{UTVhr_EI16!nJ7?F<K;om+X*rEbq<SRJS5EP{V;7yZYf}2hQx2SU$;A
zQYY}vtKQYITlqY7LU5M}ITaP$NQP)QB0o)&_oLVT{(eb$`8x)w>v@;`8RlfyzxZwK
z?Y2%%*)k~t9%rMXllBs-8HI%*DPA{$$LrnLuV4Sn$arTtUuF1rAbx+betOsI%#A-7
zmqD*Dist)|ANtnTGP1HS-id~7M9K){g<X4q&NF%I09|RN3?0{Dg5DlAvl{ljQme8E
zY;5E|JUnzbn2QFLxZGm4%#E0Hw1$AcmDm37!0zT`fikDl<<rD|p-c=g=cl^QzO(63
z6gBT(JzygwJBE0m+xk?`wc2P?%Tb!`3Yfyo88JnaEr-a>)N@fAQab&eZuWD{1F*CO
zL$`u-GUFJo;-i;1MsUiNMtp~r5f|Sg*1|5=6u`XRARxsF`Ye6oF%6~AL8KWz9U6Tc
z^N)zcPZXn%R_@3}8LxcmF+flVv$Dm2sdT$^nC44=kz+JaER_00?*8P3#1umy@^6G?
zRKls%&>OtOA)vG)3ffzpO}kU2rGigOHvm;4i(@!T62KC5-s-{xS&m|CvvX}zI``Xv
z$%bkb`qyZPXozq|<?)cFn;@ryIcE0@i-^cb8ag`t%fm&!>y1cw1O(fSp5HvT8ciPD
zO-)S;o|lVv7xUHzMn>{lTGF3B!KQd#;5WLR>kY&)Z=c7~wx7jjNyM(LueWyuV#f=5
z3-tE(epOP6`2D*TG!>V-<8Blblv?U<ZVv+NhKcz2_}~5gZEp{%S-l@TA|fJk3kx%I
zbAQ4iVL-uk_sm8EuSG&$%WW}ZOJ#tLla`i7uUZE01D;-FWaRiv!c&L*Ub72=Ikc!5
zcRF?@Ea}q(YB+q8Sn1Jq1?TZ51vK2Qbxb#*?$N(%tSQ>8=pq|%bw2XkMFk{9^6M`*
zl19(X`HfagHPrf?knsb8SUstqu9HOYK#XFk@r5_8qe;7JV$S0Tldy*?0`@>NI<si1
z?_N1=Vw`wAys0G0@rb^-vR2)PDnqML;#v=o+$kzohL#3as5&C3Zy~OG0$JZX95cta
zwJJzw;mYV%uJvuI#_Dh+LZEiA<T1eT3L+#STB-nhbwKF0NNGGv0uSa10*H%?XPn*c
z<fO28-kcF=zo(^5H=iyxFff>y`8#jjY<RXk8WkNqSyyD(AH(dhEjwUzN}-*?YK#Gz
zSy4HuIy;+maA#YIZhOzbfJMaB=k*fq)`Z``<<`By(^?DVTuMz%^(iS~61d*L29HZj
zMh4>8IA=xlRI7d#C`#+`h$<hzIWtIF>w4N%2Pq-Q2>YmYUO<%O<m7LufAv2p?`Zd1
z2eMYau~;kP<CWn~lB0CJ5Nr97xjjJl#=+hykJ+^%shieuuvrUEo7eEWqvveiYA-TO
z)qrYP>DpL#YbwNGev2;Jd?gQg4;Yp1B?vCNTy-n~h5C7w90x2-@9&M5(kD(m4_|vk
z7fx&x_2#j-2#|R3`2k&<S0MSDvdifon4>b;yw|rwY;A6zKkCs8@WB_9MVH~k>OHIa
zOOy2##K;2CHh`y+KB{>=dI73ZrTanUbbf3`db-QStZ`sysHN<uvxDkok&IF3Am_5I
ztgO$hMRt0+Z(Lm5F3Sr<zi+pP^+yqVIayi9EpV7HE32!m3$^x7BSu18IK4MpZB6Qj
zgo&uLHIziJR<WJ#1tzk~erb39Bfr%G*V@|J$liI}pb<FIsJ6D&Vy5I<ZzR>UyCC7a
zApoz!Ncf+B$dOL^85o#WQ}eofR%5Xn%<t>fAEZCS!it?k(rst%!ZL(|aUC(9AyL17
zzjfSAT3x3doE5++^<S;Z10A?U@7p|Vv{=2x%QfWt#@w02{@x<ds8RpH56eG9*GTpU
zpWITF2o$=qw1Rq;wHfss@{Te(UF~N?G>E8xj4;~{p#c?Mdk0!TtZG{fwi13YB;!;6
z%q?V%+2@33<E9J?5(ion8$bpHdHCqwiMAw*`7l3%E`+vf8Y0z~e%|wQ*kgVvO~9)r
zRcbh4?GnZ>VMDERr>^vFbkr&NpoGxz>VpRg4B*_7H)nEpb8O-8chRZ#6~?;@-X&|%
zDKpI`gdR5Hd4!AI*lIj<+B2^Rqu(<+Z3-D7q)O&<9C~Bw9I1)UqJtT#gwOBBCc}UJ
ziqz}sEqFlAORhh_3(2RY<%3n-4&~D;01f|>tn9DJ`@`ktN8dk9PEJk^M@vbbw-=)s
zqL>&Yd_B991!7WC6l`a0&6VXwgRm{*<B<wE(yObstK<b37#I$FQ=yJ~Q_OZ7U#1FG
z+xy=)4`hl(GMZfOmk%qw_XaS4@BVBYuc3I<1bPCF)?aCz6q~s*3E>)kk6LQbiv?Pa
zs~jnVT16^&dQX+}JWP%)OyvFOoByDFRLy2dnN5aqJJ0T(V{vnxd7Gv*vGqac`rdNp
z)}r62se^cJ03>4d;B!{M&catNjSvvT5qe}#7FBLuDvYSNdEh=VQxWA<O3G+=w8;53
z{YlX9zGE;J&>F}15pdIH<3~W!M)TQ9-eERLD)pJ)9c8k_t9hURt0BZi7mHP?!A+A2
zBo6i~LcC734%MObFF7mxJu*%xx~RmUT_{)7C@}_~j1WGKt}a}kM@l~2$%!34fBX4e
z0>3*406_XzP*jO!P?MOfm-w=!lf-RonBWl6-Y_$l4n2T1(?mdJF}~yAaXQF3`KSek
z7y&|vhWK|d0jb6PsL@$&?99Wg@=vUm8;Z~O@83})vT$&4=rw9y_%LeLK_?VsCh?s;
z*_0AN**9V9gYoBgzhE;9Nk5+9^K}OA9*Y0tss4Ac7px4Z$`)9Vma1lDjcebtvobav
zhIm{{Te)3eNFPZ*B_qLx3zG^0jXOT%L!ys14xh{Os@<Oj3v*|O5g?n)z%x1~W$dPi
za@edD>4=Wx?Kv_$Wj*)2k5>PxDgft9Mp_RM*gIgLYHnO#j8#&6BUWP@ws-LNcM3Mx
zZSa<LtNN)hA0A=P6ZwymBtcP5->r249L^5ye+DLO<?d7wVSzFvJ2Ue;n1Q9vM_N@@
zi+wledj;4L>eW_RiUmrA8nqD#)=jltO^3Bx$voAH6dW8?bpy|+#WwSKDp`#NRsyM#
zYf`N&C+#M&cPS(DZYs--hu8KC8<4($TJ&%|E+ZJ4|9fwdQz`M{;^Ge4j=A8vl_VL@
z;(xj9odO<5f`I4E3mhE(>T2hVE(QyId*5o~;pG1t$I4#4H2P+EjmmRK<E`@{4-80Q
z5RxckkIwD=BpCUf`wTCZ%A`01M+EJrFM!JV8K0S;sNuCZ3j6|rN`fLnjX8~=dH{$_
z7nMG=@)h}oDfy+FvVYm#n2ZJc4U)PpK57?t>nRDm;N@nA20xu=dJalsTgx{%d@h?e
zk9S*wr7HB~l$0z0gb3X3v%*+4o-5he*+I`5&ahblZLYKuDpJD6!4cNeBcqZ@Zdj>2
z-3n)UjiwY0E=^Xwp3Y2CRaKqr2T-xH4%3`nDG^2?TQW@|mazfU?|}r)@TYn{-O$+R
zdb5+`3Rds&<)yXDN*ks8xi<2sg{q_xaH17{xy6*-=>aBYi+MuE#RD6fbazu%tU_;)
z@`Z|%CU~q|c$?_J=3H3>ey(04DZEwb)gCvtb5O+oCy-A7UM@5ct%Z}NM9#Q?6F(%g
zGbx+;9Uh2Nh9xtCjTtW{Xd1Lo()Q#&?y6!i&I?<~;|83F^L_I8*-%RS+jHKpj-*>U
ztF3hs#nUvN<)wm<v~?g0cZ{F<J+?T=W(-_?Jy9qF(L<0>eFhXFU*77^TPD*!sl|z9
zcOb5poDI^P<)h!hcmrwZ*i5cjthM4o4TyM>6xnrz)0+$MMi_^&0-}>!#S`!X=nG1L
zFL{lTYGVO^*O>lFi&Lalxdvtt1s7MGYPpdvOT0>r&09>tJDbowmGL8Rcj{Q>+o2S{
z{8G`QPwSnkS(^1u9+!*G9{1<dnc3NPXWMd0MQWYZ%U=D3s^xL)=D5q=kNlTQ?$%%v
zl1k1fD><a$MEqR%>}C(bIm-jKax2S;kf;l)w~j{Nj^IUUoYetXR`kP3w<V)`1smT5
zf$UYRgm7tc(sD7y;U)2_>`kq+I<<ngAm;-78S>S}d5v$#d+$xRn4LUogEEtj<D~uu
zYh5B=WS?F&oz+^a_(e?Dim5)O+pwk%faiRdK25xkWWt%w3um+hBDj1?ih~wL{W)|q
z=Zk86!HmiAuG7=zBin4>FRl7(E=gN@`zQ0Q7X2DVx??|yaKm%YkZ1>y$WqdjD}f}y
ze3MG=s$o4%FHPO{dH&1ld!rCDCr8`&y?U!Nb#VJPP7Haafs+~3!&|_}FUGzdKRs+U
zIdF8m%#&Xw*%sEa<JiD&=?`)=lkeuXgWR|Vk*Kk)pynrvQ**A9P#8XYf?~Ae*-{0`
zDi-rqmh6-+3i9$lfBwuZB`u#F2*jp>Vf^!TVE72~K7!MFX~5?hoJG+2-Doj@*r5-Z
z9{Qcv&IF+WP^b-{SorK_|GFiq2lUzF&2)1dha~~1)w$<0LD{3*C`+8n9_4N`)s?+!
z>QgU2Y539OI(d`_gqA<1{Qsj4B;7D+;3HnVc?SEP{7l+sL>M88%r}U*^V4yV8ze>Z
z<(}}XQMw>;h_CojBledspWb}<<aQK)b6+Y1BNpE)`Nij6V3?1H<#xYZCcf8TkjES%
zR4|rWB0ENI0%O(mc813-m3D}Nlj2WQTvFnqO<b;E_{e?WN3@&ib{6*?##iN=<b+E%
z{t*i&u#ZJVZ+c!5s0!ub%J+>V^{lroQ8Vs`l9||ul3Z;Bsu^-DVNkdV7d}pttvMy)
z3ffeBAF>+hK(vVZRpe{$ztLq}dfV)6a|Wiv<1uzcATqlduUk<Vv3B8R%VSmQ^lpp%
zGffTQH@&@BC+xpuB~1+ksFe39><*sYNY$qZ42Rcj?O7Rh<Umi~*%xcu55<i1A~aaG
zs=bBvmP0VX4yrAAA2jSfWUw+YdgkTi66#sKy%;uOA8133)D;KqoJb>rfxW3<&Asl8
zzMZ*iW%=4^NM>J?gk9nE!Vo^F!ln{_FwjG<Dw*79Pppk(ckrh(N$1$qcNk-0+AsZv
z=LM*>c9M6k{sqxoqTp14%H%BHTv(R(*=6UE(L@i1y!uTpe~dKMCTE<3wJe2wE^RQ-
zI2qxtC^xrq4UhVU<@%Q_xhvh8O{S;6uUK3=shydr-wKe6MUd|>W{jH9#smjj-<Li7
z0n0ZD3Wd_9golI_svORgu1gJH-`ptCzt^m*nEIwbnz9!e?*&#r-_3T0(clLAG?%%a
zu5N1xKC24=*)S^AR>b2uGLT41F~!JBE{TYVH`|pwQ;khOS@$s0Dk{?2q45r{ZwNTT
zs*P~-`=1BSGN<%33;Z3CQ5Opq(`)m^D=lLcI{2LTai-Y@dw#1J_c0La(|z~(%~=YY
z+e|cqAQe{mIXV7TWc=Cb8Di}3KM7I2)-oP6zF<^zf=IvD`;|pt+Vp0Xa0RXr$J3|q
zq_jc=C5LkY7Uqhv{>Zb_Y2C$K@f){lp{(EOR8+@t?WR0uFQG*^*nQ>XRU1|jN8aUe
z%!V5iUD1e?xa=9QIqdnaEZ@_jlY4jki)6@!jI#Jj?B_H0@>v0Asq0er?74?|^u2*r
zQqK0euP!a>@b5ySvwVr;sf(g2cD1hyhjGr_Kswg9UC<i)*2Y^uZX50}I!aG=N5>xV
z?oUVF<gHk^#M)?zR_D;<=oK@_q{n^s@;)(t&U(^;hbzP3M`9Hn+fbu|7L3&^m^ip5
zmU7(EE%Sn<&bacvns_RHUBbEj&#TwFs@Js;F>(C;{;s~E?7lZ1I?Wg=b82a`(FozU
zzRC<<&ZcL2i67>5lgv!$!BZEz3<FiZM6cF<Ye=!`7I2R^W_^*YN^5okbQ!^Wht9_z
zXtmenHAO`X>s_Jxok2MCDkVOk0F6~!YiD$u>Fc-8l<KS2+EH|e5w8QqW$St?g|jXt
zKh8XnN;>h4pkT_=G=dx9L>yO8P%!5=%DW)=+&5pjuSr)B5YIp+Ccij;F-rhWCH?M~
z@cy!~G=&k<Q{L1Ducfc+H%7bm*0*YzmAMLj6r`cj2kqbw-g}mHV@X>Bq(6jhrn_Jt
z#tF1A3;AkL$-Z#(z~MUh6c-z4itEpwF8{RXMEs=qwQtOLD{#fine$2j9?|V9$pV7y
z?Zf^DmjWCWx|y3^@$?5iwbKCbY#Ppkkq>X?S~-b7e}I&&F#beaqq<w!(=o9SI$Q3Z
zIXMoRJK83ur(*Wo93IANUGzGHYW}>c&WF)v<+ZhJ!i9->K4it!Rt$N}bs`i<V?f@X
zCz0>hnN?gz%0t4{2w)-Bq(Uz&W$S27sR~*tj->Dy!__jAa|PdY?i0KZ-uDmcq7lOy
zMw=4L8RK!)RtYpgET^hB1AH$Ri1$%Wc|Q!(`qslR9IW1RK;=;#wpAayhRKOS@|*}6
z+!0r&xnD5Gw#=#B>82w!HNq3{^QREV8wTR2<GI%f;yyDY_Hb4$G%i%6v8jyu%kK;2
ze8Z~>o_@rk0dQls%7+E?kCqxUGc#q?)YRmfl_1j!l5)~a|Ca@jm2Cmk37^}JT<{Jm
z)ajNX5}IGNJTf91!>B1P_q)bjqF~bQ?*P8r#eOf4?w)~8)v~<&7~?1iXj0Io*ZmE%
z=ZypKB5COAADn+G6;N<;#sX=O#dwHF;PLKM{pR{=J)ABEu;IRpaDhHB8)?@J+!~Yb
z?mRAzmWLH&_!XIzgyQuRudlDsUNiX=Yu1<9u8Xj+veJclgAMJ!fB!}!V9QPSG|9Df
zek1-@?&#T4Vq#*@rc%)c(pRxMJk15tL9jOUEw|cAROf9kAvXzAgxs!G23O|1whot6
zZsWGn^vmkq@1l3rKzAlwx)YXUxo|RXx7K#5`0hOa;U}vpvzSFfdA=2r+4n)xqBrXv
zWi_F9NmFFyQ!iEz$$AQW1edPjeJd8s5=4u6JK_Z~{C==|OksGS7oN_IkqB5K4aE?t
zHmtj$1iJTt<r}DDB*SStV2OU`!Zw5bfP>tUd`;LZ{-vPldcQRNtaXii#?xGChj!=N
zatDuesOHXX+fTEEoMZ`u`J955K3IY0zD&mLRpUM4@P6tu_B*m@w@r`pQ0A#pbnAyt
zUS=3*P#@K2WYm^FZwcG`8Ai+<3%wlQ8v0WxRx47!zL|1~QWC#f4WgdZq2`CK8*(N3
zC8SV+S)*Zjx-^WA=S;9a9FT$?ih6Q593ZSLtESGfF=%U)6eagGZ3xnV><XPThm1nB
z1^;kZ1|=mC<Kp5{zj{tjPj3S>UBGPaKqbK0d(sjs%)x!D1wh-#kk5<bvLy=(3w!#X
z3=9CStb+34lgnOjGMX+veR+AgwUy#seD$cT?I{ez-Kwf8>Q_NQ4+{$m+ki&_)<en7
zeI4Tk3kz%YJeN_m>^l%IbuBG5OT8Hx8P9gce*pEkcSgv#$`+PL&?^xXBvKif;h=SE
zb8`t36Y3wBq=+rOy?I?NbzeCppF_L4yX_zDt`kT&hsN*-3Elfor%6gnOZVrikFsaa
zAL1_<#l{w8Amil+3bu4!!@bL^3pJH;d-FW5SP%R1Fi!WEm=hGV<>7L5o=)JJ3aNO4
zE{<=-#a%@fc|P9Nu<jJyu}5!j<=qz1#*pLj<hd%%QjkA1kLK?dF6N)v_%Tb##^%oO
zLw3>Rip1N8AX9;l4J0Oa@N!ZXIstKfdbh^rpAVMYJHRbzJrYBRoOHJPAz;I=^N*KW
zsng{~H<*(g=Vk}(xiTrQ9<QN%cYnysxC#_9{ogaTK*T`VXVz`4AdsN4NU2wo<hzI>
ztza%2@J*68prXSeQmF}C+lO7Bg22UwcfG7;m5<_#A04~0XInr;bdgQeY-xS8EUGdc
z#~8%FZq-j6Z=bg<AybVK%9U<H6yU+6Qz&sNF>JdWH%ojoIyKwWQ1OeMzVUfVS8y54
z<mk5W{N>sQOiYhE_$ABxd~MNxwdNyuenzWp{*zVl0-ikDp680u!@S(wws%*ycX#;@
znzx_`lv~bofTwN7ujGTzYE<H|GsaV2usC6RaWEf?fQqYQt$nXSrKP2n$md$t<s`qz
zl!s_t?{q*r)<vhF$`HBRaJ3ekjV2nA$tydS7n~`go%UP%HA7t(M3GeM8Ncb2?JH&J
z;&)E-V!lU5R)6KhtA$iQq8Lw+!Xi;fnyA%FYZR@lQf}i>*?L@)AT~dgxFVTmg5DbZ
zVSa{kMZRCV7Ta^>$Uz#y-SdmHOZULW+{kKS=*!E@{oUtjh(_iQ3d15VHok~Rvc!gU
zjHc(<Hn)Q__WoMGLJhxqove8;6X~gIjz}}AEDLl>OKryshx%1b7nfvTN60<qAVjQ$
z0#3v=SjS62{bj?%?y5ZV{_wZ+JlWLN(eZfOQ|(kSXi+ut67`liBHo2aU|lh$%e0lZ
zhox=}UVf<Q=VIQ^;Zy~~B1Pt+-t#oLl-EIH43*5Cm`=vF7QKvZIia8B20Ied^xeQQ
zw=>YG_a@#Au-;k0_7jWf4VE3;W81<>%bv4PQQBALwpL?5QBbuy%A<9Uv_t*wT<)r;
zRRkOb5awA|Z5a7RR3^Nvbl*3<7=xq9()(0-p7%>WAS)CH)8hob^<Dksp~@HV$j6AY
zOG<u$r~X6`pM*D1*uQ-F0#E^|xw$!5#K<Tpx*b1Xh7fUMDNp~w76#W<0;2eQb#)b}
zLBD_m_d7XRT1Dk`INu5Ku)r<7wDe?b2drG%TXy#7gZXOU<gj26aeV-81P2F)i<=u3
z5s{>e3)lVaB^5n=VV@VJWLyi7-@1VV@q|c`Fo@nzQ6X`1a<;X#y%zA`((ejs1%Gn&
zjl*ndqs5cm54lOa?`)?a+QTQC{j-(kc&@IlY}QMOB>ZknUUx15n52vnkf{tr=3X^W
zo&+DR-hKP_?VSIWScFx%d9@$XMPkS&x!Z*2u<p{h&z25y41F#!aGJLYbHB~C3mG&_
zDSfA+mnO5i&2z(qb{$ZscyhL7TqJR+f~$!u>dR|N-v6l*vQCoCOE2`LnU{m{<9nO^
z3tUTZ7zhKA&cLqxjq_9e(Y$|=fTynIw?-KK(#Mpog7;}7Rw~APaDPew2S_(Mxs2@;
z%6p-0^U9=nxeOxvf+rgf&N!^Ax8Sd!5^b@+&E=fbhLhBHBn!!RPT+%}L(L`$827p9
zF0xX-QK-|dTf!YAQ42)MjP1bR@gRJi=QD7SX_&kE_uHM0s$)AcBJ2TB7_t2%h}gjM
z9O?4U<BJBl@n3)W$`6<Kuk?s>ruRjgH)!c=X^X|EB!0IRH70*tD0AjT943)if3uv=
z8x2}WW_^gCAKd@4Ad(3jgN2?y{QZxC9Df2{Q$WZuHwVK#1YK>HA_$I*#0Hv<E^z)3
zA?rs++H#oHl-Xh9Q&TYjZ?JlK9AK4a))~tCTu$UERBL>B`}Xa&!<th&pdcAvm6Fjj
zWT{K4tGgE$7gMc;$;Ik;vqUtXweXtwh~2ZH5K!L)H|d?kW!}iy9=2&n?&SuevsJyN
zdyDQED82xziP`sDqDO4tg=)N(q9m4aZWv<;Z}7x6T{NQ0pwoVzd1ZF81Y1!G?`Jj1
z^qvw7hg{Eqb9P;_7Wd6<S0<ek{b41k_^W2+8iM!_1AnnK7D%N)ccyUd<Z`K7+uf7a
zaY33xKMHY2ZD+Mr0MAFw$WD8F>ZJw!?ma&{IHRPjO&-}P->vO~kUw{OyoY=SWa9M-
z0@P|cq_=tC&GCk?URC0o)fVm(1$$}+>ekkN98N(QwezD7pt9JQ|5e@7<#wxswseP&
zZ`H1Te)J3hw5O*wLQKx-Srua#mwWwanX<ZXfKwZ1p<z+=yC+f*LB7~1$v6s?5zi_k
zYaztK(i6V~KZASD($OJ@VU=z8PIbOs&vgC6N4I3|0E$V!Ap0%T0H4Ai?3vMWaleh_
zS2|xAEvFFP-fdNl*$Eq^B?MM~jt;M6T3>P@HtdL%d9$h;VE?sWe+zpIHsu%Y=Z1@4
zJNhusAyQq&yN<?bJ|ap{Sy5AacoaC1sBV9*C{QL2iD+J3IX=%tMmDb?_-aVIzkeY6
zS`)1XT;JwCYP^elBUvN>r}6Kzi$y&2?o$vjYByONGE#`u1=9?$-&SkACmBoHo6FJ(
z8&|P)uSe0`KBJV9hUEI)=h^WYxf93dB=8D%-iE6}xK9!9LzM7t4aD)LwL~lV8F6m)
z!&>8(IuA0ZcjQzi?#~TQxc~P~yGgTf4N`yDn4M$60Cf`aYkYYZ2y&YiNoFag;Ph+Z
z4BD>6oKk#Iszqt(<jqY4mVU`2dIiNL<3V-_2n`+m+nb~H4og8{@d=#Yn>n&8+pCh1
zI^J`izmHkD#EMgK7li0{KIDdwJTUZyiwUPZu54_6vUPg4k(>4#MYRDx->llc`WTt|
zaUeoBz?^Y?j$OIAduVhoV6WI8h4uB`*{`CmuHK>C{l&G!Pk`%)FNBR}c9P3QY5Mq&
z!HUY@d9<sIQIbWazD+8rVBY6(JYm_|%e^DQm+h|+FzBGdJ~ws^4Cu#?+Q890&<U1^
z6sg$8`8@JVn(b(6ZsUUm*q?6v$k%sGoFgWfA}OaNAWkP?8^(k9f{i+a!>?Udjs0gH
z&(nkgf8ly$tRl!K3n+>2o<x$UIO3YyWi|B)+1{<H;D<S7!S;p;8+H^%L1(H{-CX4Z
zUU5eMGZsEIH+@re`xj!6ar?7s;{9u*cR<shPO-slCOcKI?*UsdoaSk&1~rAQDt<e-
zI(!I?cjO3jVmrCSigd|}{-}>_#Vi<&`PI{t)AUG2O%f}y&+i+hB(4(+WbR<Wk^t#C
ztwc6ehB|<cy(aDjksw8wX8EDz09)OhDZ#$X#+;uXpQaSDfe4<jxM-wwfF`P1WdNL@
zh0$Z1R4&-oh>7D9jH4Es+)$vpp*cvlJDgF&X4%n#pRoh|a7o}7ZRM3b!!;|1tAI?Q
zHogmu(OzCUNSJI2=aKyt(_bZ{Y-`m1doR8(CjE0j{uf~p4;K96UO*H4fQG|-R&+eQ
z{5~|5MzQ>|UXZ)Ogl;$D#K-lTd(7l)gVHbpNFKB3KWtA=$=9?oM?nubJ;?oUtiKu?
z_$WtYx=p(2P~SG)24DFFn+h5rKKB<cx(Qr<4BmcEHj&sKX_evAbWmFIHJvx=ol&lH
z{w5^12;Zls%1mBKrEM<$@tsw->C5fcE6@D+-gMww^g%jVh!cl?6yd)*8d^KLjES&e
z3Xro;DP&HlK5vo5V3>P8&$23)GIt=8jEJ$7UQv~4n4)gJtUW&j^bK7Jkps8}Hy!os
z(t1%XQQR2Z8cu7;znx!al2P={T=C?y#o;k=p4SNwSS@`d$OFHSqp=yy$*6VROGmT*
za$*mX@n=hcx{p?8v%<R8j$Uixkob_Ixcx$!Ocq79&>!2igeIxs-E3jfxgaBPm|+KR
zczQ+gO{JA?9yt?VTyChSu5!$FJx7GZ+x!<u&Qu<s)Fz6<d49R9!9IyW=>s0e2<_C?
z+Xsjb7i;`t;WbSdn=Nx+i-~xLW!43ZHju(!@?^)h&Vx9H;q`>|7Xf>GdWMyUuC;r{
zp5q&C+KHd`XUaU>R4gjzW3YAn9KYULhjo8a?caMsLJ<VNq?35kVh!&z>!U!)h@_GM
zuJh~4uuo|%ot^$ZJ}^8E+aEyo!q(9;u6kH3!a%DjjlpCtjl0))$mV$D2>vY5!hzIu
z>oVeX?MtwK$`{eG&=0^@zt9X~0O)M(I<-{naFPpzaZRfTrZV+(^D<96cHu0+Y&9JA
zYTwMT9hty=SbPpkOp70f;9!pvUMjWf#|4B&gW1#5_=;^sdisJP>ixSfTtD!vMVndh
z4^LsI)eL?<B*SN^@ksB7S9)%$wiF>cv`ls<EnusXeHhQfHH5t|`l#_T)X(GiXm0N6
z{7Cj*+w+&=d++zfwQrMGRD#1rn#p}d!ezdo79OC}t;#8Dc=DS}3VuJWC)4#L2q_XC
zxe-UP7;5PnjTsr`FR;v~rQMwF7NFuk%y+qOn0@X+$BlR)zCHehPkOUmbc?&zjx>Lb
zh2-WW-i+0?UMlRxv!&&T&$44xxUF3&SRWxddmG=Xvx<w}$`5J%o$eMCRmjH$QHFE8
z?j4cm9UC^6oB=PjN+@v-!j)AXZ^&T$d>gV{JMT`mHEcOQs*Ad0;J7HCXswx`+%I&l
ze_<R5-0E2)&J9xJgO2K#i9r6XE_pbk3z#E-59JCbDKAff+jhGA_@F#>7M)q2^yy=V
zIp*=6eu_Qx63_PXD8ziculyCALPTs|v41!WnXG;f!?k?-<l`n1SN$r}E;|iba8*vi
zDqVpiREsFeWamT$-pXro+F=QZQ!f`RO)>4~@ne`5=AJd}z?7%gI#35#rW70XB&_FK
z94@6ZCv}}~UX3k6TV1a*56&&Ph4%ZwL1QZO^CQ-CJ+uilMfCVw_H5=xcgTR6F}uCi
z;*2iHf-1toyeua%V|rFHpSB2Pthp(o<YdMc2>Wq4|C9c!kmaDyAU>ezT^{F>i|7-`
zFUz}LG+)x+2nUWVy`1C5Z?Y>@Y6x?1yWtar<yG!BOL%0yHGa8(?ig_Chlbh|z}<d_
z0o_%7>NlX7x$*N?^!@cpf<-1+_zf;#(6!>btFN@hrndUx$FB-5;2YQP(|M;<$O~k9
zAk=EZJCmOJi+Y+bQDuFC-%prrd+aPdUJgDsJ5cbfMQluFXZ3o~x_DR&buuquSs`+?
z_jhe&;tx_NxEQVQ6V#-<;55Rc_}6kW=X>@q)z1$X6P6m?LBJ+9!OP0hQVK|wJ2R#4
zs;W4Rog5Aq;z1auu&@}(2v}JD&CP<$%*=mD(#gfpC4-Ycp2hRPI+{L95D0&>zV_B{
zs$V8}{iybyW~f#idcTOmy5^*qr)y`n-{)ta{c}D-kPnlW6<xjb1KeYn|2FI&_*-md
zbIHoV@%3W6jp-yZ{ewo1u4q3SM(ys8y{p<);2iD(=8e!&HjESh9uF5@r3FOGp9h=I
zM0F+>a?BF1=v@prfeXM;7H2HnzKPXdg{=$^5f(}dNR~l=%fvTRa(+k_{>URCN$tp1
zy(R(ccNdMJtpTso5+cX$m1k4$QhBVqov{8f!>?LzRi#*$&M3d5(kAP>kHEo>s?-S1
zn^hgO(S$I+SoFJ`O51p_mvYU~AW!Gg`|(92fs+0PQl#y<?T^M;=&`FDPZp({j?{Ax
zdlXa&3d~1mvuIUl^?cLO0~y$_ZO~7Ewa_T%%;{2xuT~i|NXpo(Rb7o$IxmCWyhVRC
z6EvVH<Xb<@R_XtU9a(p9<-q{fod}xze>kcxX@F#9vDThyzS_FGU1;L}@$?l?Ren#`
z2m;a|2-1Q|cSwV@bax9#cQ=BBq#)gmba!`mNq2WQeCPW6zn^QlmWpuiJ<psuGkfpZ
z2MXx@*v7`jqGDof*E$iTli2xfY#0GDb8);(45gx?0-DJ$0QZxVlY7&qSEW7#7%<tq
zK?b{W@Yv?ORYb>Z%!$R%<ZsQS`B3TIq9YsoVT|(EG^H(^!}*r>CACdyzOS}{O}&10
zHP#>mDrE>FCc7`}eeQd{7IWJF$Kk|@lN%p_{QSgxD(ZB(DKJ{=a<;&&<zC^XFxg$;
zN>o{+g>EC7YjR4{@{^i{0yP&tnfvo=BHAY5w8Pmif0O9Cwc`gqo_G=r{uE80VtzOj
zeTDb+^Dc&;&;3%OP-390oR$3?t{i7tT9CJGF`j2F@<ir41>pdAH-??60Mn&mt=_Jl
z<ez-aH2oF?z2CP<g}t>DgjWs(FdoRUo9kbx^XQ>nW{jM_w>(ZW8?`uta^+(q2x1}a
zz3&ocoYgTDcPQl)KDfRmb?1UR8C4mXR&{b1VZ-li|Lyta_iMb5tuI7AiD+f40mh}k
z!gRU!r>LhhE;v}^g0FAC{QV6(Cq5iO1cq<ehTD7wC}?Tl@?29UI3TcihB%~sC<%Pu
za)f>vk$TNWSL-%pY;{tx-d}dE$@yBvup-Fr!u;bVG4j|jg1@tTg3e&Xvg72}wEv#;
z_+i>;VqJTsazVCeI1vGtV@L{*TRBigMD_Rc$pT$6DJkh+(079l4gh!r5)u-?*^`jd
z{@tff*xsyW(+sW`#&)O`(?ALeIOt8C=R2JBZY2*5OTlZeQ@-hIbUb7pSMeZ;((1Bq
zdCG0VLKRmYQP0J;t2#5;*!<7wtp$cMc(;DK`#WN4Ho#qd`%CbAsB}S_Kwj1ivmoxt
zo-f59!;28{B7scoHU@nBM5^V6toiqrMu30>m@x0l(FJYESCasB&5{=vrX4vmAGo@9
zFa#T=%%-xST1P5Ap-w9|b2?AYFm`V0{AJl<K$xDs(LiB`P9XBAY1Esj($ygyq3-wI
z$*(y%eQS!|3RdF{zfm+;;&kOaZ|zYC8U8c1e{qmf^W6jKNg#JDXujSzxpi6EfNI<9
zUHaP$ME;VV@$R6f65cN*fXzDBi8RoW)*ri(q3ow>*&>J<9fQT257gDHM$ABbmxl#~
zlnv|qEj+si(|6Biaw=HU>6g?oeKt$WgG8h<>e{O5$9_glS}(#phcwwy7Xd#1M>l~g
z%+&93s)y$(%xy!jV;qaMe!UhNbvYATNN_nDuSPR#gbN1L1lPGJ9T7XZ=ZZ0<HXz-E
z98CcOf*nZD0R{R`5w9n#W<nTKj@Cqz|KG1o+UAac+vn!go_*6(7$$z$SV;OllZ`r@
zefDp8?4_02)*f1Gc{B=^ToBf);uDRPSc#u-uDM*F<y$I#bUtSKnEi3bVQqxM&F$Wm
zC{%G-Ts)|(vb?=tJ_fNNu?hdizG?*Mu@y9FjvTFRZQHwZ^cGm5834x+M!?YyG++B}
z6rlgWs1U7CF3}z;)9-=0dAdI?K>4Mrs#>bs1`VX}fgvFVv*m`82`mWI)YQ$7R~jH5
zp;YVbO?(kf#3L00un_Bcu=zp4`1}00EcECQa)AW@rNw$B=UX-1)k~B;>D0XU#v3{V
zxKr71C70gl+H+4BqMozEyH)VL;$M6v*p^AnP4tC73o$gM;#*onk>3t}ocw5*L1bv*
z+wcy$GJQCTaOC~w!6sgf)w)4X$9W3*sWUeioL*adGG>9dwKUlt8&4c%Y}&a}6$*{-
zjh0iT*m>@U`N1<83<73tVGA<~l;{(;Y*ltA);ZvCzi_c+mk!<R%z+Ahz-K-m-6ps*
zp$1$*B7lG8;BSd(ED8Zvl>)yl**0CL2k$5#a+!7MEsOe48_$f7`rC;YX&7g5vU_gH
zUs+Ia=X~LnoOCBtELu+x&CQWz6&iuhtHJ70a&i*ig$;8#|8zW5qG8E97tT|+O!kWw
zT1`%ZV9Vp8?cRnWV#1!w?1`IZR?l182Bnv|4ft~UjZ8cXD=i2g4#JBsc7K@7@9M2_
zn0|v_`jnL-k;R^i?pCw6&m{3*>ZnPb<vV=>PwdSAKR-WfH5ziGXT!pA%kwZ0=RpB}
zK%;Bq`f%Q;8$&#Dauk3b3>V9oZ_l=z$G`pt{#{uga?v$#;FA(-$FiEdLTgYZoZ!ve
z@b4rE`N1>mZH|(j&Q=6;Ygu(gTUi;Bly4$F;maWzN;J4z<<wm>bUrTj6mnA&yn6BN
z&t&d=C=Oz{eMa;OaWMh)mB{dDMhhCwiH8zMDmrK)o+oeLFC`0^PQE#l|7RFBJcXg6
z;{<xmMTOsaxS2h_eRoecbiwQbd&Mb3e60Lf)E}G6^LJ@RTnP>qO3Q*QK}Eql0JQtI
z5+Y~NnfvQ^9KW3Mi%((I?5vKNCZ%4XxDC%yM#D1^CA1%~ey2t%=zHTp`T4YZw!OSU
z$|+0}Y7ApK8_j2BcdcK+{fiZw6DmopLYU{=a>9K2*G1L0NI3eKZ)SPloZ4ea7qL@S
zVF2{e5ihj=tD4|?4_+*R?5cIVP`-6!HF(lwqXp(cf;KBFSt>%L0Ja|7F7Vw+gRgP^
zZt7Y#3ocF0jvkMdjps(qrl39(*Aq78S}w6Kpzy%_1hfSgjFYVbCj0;(WI-{}-kz_Y
zXLqd|Jv@EY-!F8iE<r6(SQIjnZ^?j#$gRQJRiUKtyMbzmzGaJWNVY%6J+fo`Ap;L@
zslU4~44cPHn!@-CRamkwZdh{W$NWJid-X+H9rqaQ$u>8w<|JdOf1#2`!sO;X)4%7s
zm<||3P0jq&%Z5KYcLBHfGffc{6%{UU<lvJoo*&>x8>MdMNgdOuT<9Gi8Mx|1D)t1G
zBi5E~)6A+cQS|rsbdRBeLowmn*jJ?>M`TLZvDR@Y@zh~s#d2Yrr%73KtuOdPa{Xt1
zbvn*aWrLGyQ}KbN!_sOi_9`^1jnN%{{cXK=95~l^DxpQe9@$P+iUxCA38w_Ck;e}I
zMKwwUXa|Ap-C(mn8mdsECJc;RHO^h(vspHpESian7mvg(Ra<B<heJxGdX>6cmmojq
z^$J)koAf`8GH4U`ky)IP1TH_5b&A*9pN!?$YF<MiF{eF`AKZUC(!{8>N}9sx;^M~g
zT9yL%H*ukqY;Xg-mbzC{Yvhs?R88=+t3Cq6ehf4u){Ldg^Iey7_g#KpUs?ZYwsCkZ
z@j0_+sOIdNxaQk8PEk%CO&XE(O5@RN`u}i-Bg1cKClChr&5I?Y6c>LW-yCiBxEG)A
z?&>=2#FSBdc=p55{-fG#zboHm>+D=x7eEPF`AiTwt1CxBa*FA_;&w%wAv0=Jfs~zn
z18+*E{iP&7iq-s<F3{iap=N&=a6I)DQU2OqG#$2LiXg0BtBx38*Vp`O_-+OV!fJF6
zd8K#%yiRbk@*VGgeK@8`G4WlBc0WUh#;1}gq0-9VS9A!sQRO|R>S=ooR@SDEA9Q<C
zdPPB8p;Vr?`I{u#&u%QS6K1H^j_VjA56eQFVfz{lr&Hg-ULv^nzyT~MNlMkRam$}F
z{I^T#<GC7FmH{C8pR7@hFLc!g66VMJe&yz<)l92$#i@>{{P$4vF~Pc^0-{>4!@taA
zf(+C!fM$?P;f@#kO*U4j#tev#pFe+snT8)gA*@u0gMrZoW=W}@p25>sqzNbDnOo?H
z0&9FEhnNj}c)Z^q(Z!{!#>9QB(-0Q((U)j$c}7WEnT(Yk^!GrIK<+uBCdcOxM2vi$
z&)s|dgIH**!)HK~C%F|ZlW*m8S*N2(c%ACLrn)9wN{9j!{2j!34An6X9<iFRX^*{@
ztSq1-A1+<%vI3&v(E*f{s}gtUGWe!+Ss3cbv0}2XT%rG`1^A-Zp%L6C)^D&9;(zK=
zZdl^@w}C_6odTW?TTsGnor`j;JwXv_E__Gb1k;U86lk4rW?Gkbt-YqUE7Ui)DgJuf
z{fnKgti>}U541Mpev<2{YxFF)sM*|f6$-3Et%j={-NTZWr~c7&L6iAvT9E9KiMVfa
zRxH6J<!q@Q3^w*mz(kq8sB)odThJToo$>78|HV}Yq$8!~GQ)@qqGM?Gt$;0ML;*P-
zl*S7WFi{^cEnL0gUme^6l}8(Ct%??$Q6RJX6cH&m#bB@gtljb?d%E%sHCiA&n`aAE
zKkQpGuTKX<-D4Ae#o?%bs!h6Y4U}8XnqD38P9Q&(NNu()ni0^h_I*1M)3B~ISa$NV
zF}?I$?@@)Pu_)TPZ}-UH?uy0X&t@)$;W>FU>{XcXG*AIVisEeLGc{U}Un(X^)!<E=
z<;V~Hy4pKe^ebzt^_kfJHkKt)1-$bMa|;Vlj37Z|!vFL+NJvQQ;Jo_=aH7Anvx^rC
zV1;nG9BIJhM-$}<H6(ijEv7xm*J;7E>P@HhcWr@q(Ktb!oa#KjM&V7ovr!*d(Q&SX
zZQOS)t;>AEebe&3*Gy?>w3@vbor*5}q3Gz|Ts^my@?C28@xYp5y4~w*DAGHTt&Sul
zs1WCExl=!hHy4~d`#L&5@@tWaE-_}?{mF&`I6A!4zx(;LEZrviTxfRms0P$a9zZZ3
z&^~aS`op_iwvHE^sBm6|CcS+q2`I2wV#(rIB*K3Mi+3Y~Y><ykyiulkvqp-D1bf}Y
zhL9aSYxVH^zjYYQ8Yay}^z-5Og|h)uV`t~*VB6@8r0^WfRs?`GiHMEu0YeqIpUmGP
zzW-(s1Nknyolh=M$=%m}1gwp(VH3x@3<Q+M_Et;mommmjChc7+`I3nvSl6=QL`sbA
z7g)c#sN2aUPQS%QyitMrOU+T{j>e(FlyAYiG6iqKOo1FJp*M&7H<8xomy+DRt}s)h
zC#rZAavX0^8VO;+Vs{-aIe-EYpb1wP`PQ%C(=w4lu0#~_zl}78vevYqL`}zp+#kUv
zH7#~BCgRfevnrBHz!Ikhq+*s;mVI4#z_GTJKhH7}TzV@yplz{g`;%QXz&0P~t2u9I
z5tf2rtHMaELf)rYZJwL_!x2G$+hK7_wba-5^OMEgXpU6kd>IGi!RN;dw0Rs$bp!vG
z#q&?CI5>KSh(gYRg?wmc7%%%2+w$E1ro0f#hJw8FW(Cp7<C~9puk#qi07d>C`)YKg
z$uYTBwT;5cUZVJQ-4psM{i2Scg5S7L*?Ef51pGi}r`~bT&(8-?PI6ami|Pjbxv_1h
z+%^MVuDICZNd8S+5p=+~gFOkLFhBiz+)+jr9&<(Abp`Q^p4dA}pCvz=Xg$yVk=8Ee
zuDX+okx7=Z-S`-XhrqL#=JfuRH9F>JY?F6oiZ2bFz6r#mlGEdKd$EP)kPM;Bh52Pi
zg4Q%X^!YX#r0LfAve9n2q|7VnHG;g$U&>_Vjx|OCF^JjGk@tM=Nx#N;_Al7~w^BI;
z&oVWTgsdoog^k#*<LQmvGzSd(s9wuxqE}HBj223~!zkei(aH6mXBQV^TDVP)_)4<S
zbk#O$wohLN;Yz09ocUj(i#DE96v6)}f2^1S<O#pKT=b#F<#B6kyuLBPhQ>Mr(15%~
z?euPXU0;jWgb&V?^h~7u4i3-`OMFiCkBHdY*zAwWF0lj0)z#qgm%i@=?7tLDG{Ac<
zR)CuT?A5^{LFioA_*^7BpNr2Y-}b+!x9Uv#3PLV{R4-9>djXwr`pWq0q~m{^BI}L>
zrV)Iuah?*j*v~mMZpo8zAAs1EjCBw3)`YZoZQ*YyS=&=3(OZG_PtETJ5XDebLGCIW
z9qybP9Xf227TfZ}8r?^!KZ5D-MF6$`26JrSqG!;>-vX>}Tj`aGYdgC>GO7}Qg<4x`
zpI_12`0y=F$?V<V>^reF<gq0L)a_p`BvoxM!P@9F@G%BTuir;lpAc+2C6ll*{?0gn
zHcvjZxpK@jEUA8Zb*yWzdrgK2%kRrI{!N!cESK(e%{mKEZUPHlQ4@=wz;UoS8QK^V
z!(W40=KL`2zt5<+4Ph|ApP{f2A;rR?^5hK-H72IW?BBO)9g2~Aa(dsIRgDb{)kP7<
z0`5#H0W4fdh@ALGN-w>Yyb=~D@FeFwIVuiEP&}}{ZNx9B5SEr#<N#Jd9Syl?;*cT=
z`^Evav&2p}+47${1r%2vdo-;DQ>GBwACzAkV2%+e;?ibR*fM>sfl8<z5AuFKuu=f|
zC@gN>9|=WvS9bmFy}<VN2i*1q`8T`Uzx$dJ{1QvxQ;IxF&7oMom`Ro7R;K-E58=kb
z#%6b^drRQmIIlJrm8udwd@<aA|7XOtDrQ=lEwt;yribcW9Es{bQ6_jDlrPEdU&Dig
zGrgXlz`K|fzJ|QGeGjlh+xQ!~*lEcnd=@Xk#$$B6^+op?^+)Kg@Hr7bnT4p5X8-ar
z&FVujGuHNky4ghAVtiVg9`fVJdnd_KHl8agh$IQ{I%mb-7B7<AkgOnFqsv!zOV8`0
z!%b=JO$SqjfldRXs_b$8YCLd)M=4)Ad4RqH11Zli$0>&F?hm`jzl8vACzj!!{Hn2$
zo6})WNN()1kCoHuND@rLZCVtA|1!=a0Z6-@T-t;9mIy#&pLU2Sk9K04)4bD=I*leJ
zogw;Qd&sYTW!d9#ER(NPYCALIRch6yv%{i(3gh<iX7+m~M1!p;4tT|$*u&MyJs9L9
zjsyUGGOjNrsF-LoCi1F^Y}^TE5Yc#-&^oL7qV`g}rzVMp;mfhz*|xCvsN@4Mcl-F&
z(!k`!K0z;#T(79oU6?4|he>bxD(f8AoHpD!uk7p!=(!VZ<hM=zuybX>fod7HjTD09
zRG^AJ`U<fAhud>%p9WfuYJ=yeM`t&D-gtk=7Z)wS#-AR#Bi^n~f<l4~Iq`vSO!sU0
z<KQZl4tKTPdMpOW{-N+pK7)gYJeK37?oyOp`v=ZV>er}An9s*ReU%^pznw__M}B|B
zfS0Br=*s+s2A`T35H5uzUpLNP&1#mvmRGww;4Y`n_O19BVuc99)4F)K<xgo?%0al8
zposchp4|BxC$3Z%56qcdGM~7P=X<>Z=ja{J?1??Ib>XIH6W;`sF{WJ2rLI{<sq82j
zX~_D3W5@FDNubOjC#T9QoM0QXt)E7pG0#!|2gXM3#rxH9fw^Cot}`H<F|da0!}{D^
z2iNH)WugTObC_jEZxs!J0k4-%{4dx2UE5{!82S?<8%bk$A9cpRzDAjwKEr9jY}3L=
z0y<mw1#Hgx1%d_G(sFOh;u9CQxY^e<Z<6NilYmh1YfAE`-++4a^mvfKSU5aF<O^u~
zn^Obl^djMJE2ig(0>w)XNO$v5#v`=86S?MUnIBB(yds>?a92AqlRxC-NJfLV1NZE`
zcQ$Z3M=iy6<>q(=nC%#H!I=IV)Jj)XXI$^45o8#@+kMS{IKrg0Ar4ll@wqbaRAtl~
zk;&Y>5v%lEGr1daPp<KLipJbt-**d)G++V+*!tQURScK?j_%PBgizQGvjxGNG_Ogt
zkc>kE-N0I=*oriyS5Cht|7IXHHM~CSu66qKV?*GgBVr*ILoV%)hM7h}ScXZYB{LXo
zL}}o;4+J3o2g-{R`)bVV@L;5;tGRK$yVP6}&FrR-@FiD-D_um(lA8RrzJ=~x%b9na
zy`=@!!61v}W+l_jXGv!+>#ZF)xvlmlKQ=8Qgn+NKjhOI;B9gT^@Se5Ma<I--S0v`<
zshs$^knj{kkr0ayy7mN5s!zdrE4W0h!ivublP)4Ywx=sf@hUlVEi-#y_tT-0f72;3
zDLvWb^{<b4(*TLTyu2(@Ri=vGI?{&hf}^*9aR+-?F3QlkCiL`Dko;D_NyWD0n~lNg
zRsQ_0z4@bkw5n%1N2MK;B^MZ;xGEredT3b=0o`opVaSJ7-H<<o+|-{51O?SJTA!=n
zaHxTR?z8DdbUnT?%*MfCcN~2xMtpv%rZ=F&M%8}X`JISkZRG88lO+9ucbmLbMPYTm
zQgs*5_-sxx7~&%d!{nmG2@1wSv9{^16r1H{AZ9oq_V`Eon0mivX4++2XI;8?K*%<v
zB7?6nJQTeC&2nghXw>xO>LQOIktqVL-EC~4E8c=!dVD>5mu5+Q$mpm+7HFW$s;kLp
zX^WZzNaK4CQf<MQ>1xoXj}Nq+6Fp=}okAdDK!hoO7rdCddl|$P4zE&K>y?0^8FH!n
zfcGuOKynT@bf{OWatL?-i_LjfHcDyqUAgl${lW+Alrzka<RP11VZ(!5i|)Pt8gxf2
z;;N&)(wj`81PHrso1yb24Z$$%5Fcri>&jO~ch48vp+4qsNHX1+4h|7e@}s^3i5+IM
z7s7IB47$oonH2vnv0vI^2w|kO=EL5((j8m&1l&pSH|x|aXC<3vf{=%o#U-vTP4}2H
zvS#oA>w3qsdZF<9Mz8aoKRioM#}X=%AG8~_gb&YR{8STjgpACZ?}@f?rFzQ4fkgB3
z`SIP7-#j3yfgR7j+uKOoIFHCG9Bt?CGQ8lD_Mt>JX2%0nJ1c>ILGr3>UizmWyyPFz
zV>5WY;i#nE_5RZvadJ(rraN!d+L3`;$r*e#wZP98YahbsW`##brK^Z5ege4V3%3pb
z<lN!xpDRL?%O8HRKbIZgiqt_ne?aVpR|!fwlgSK5{O6{)i@1W5)VHN7LG;j?MQN%q
zDiS1u1|4&p^US(;6!pAwE&Q-fB{tShYh}yd${s6N2_7xT>z5-}YCc44F8&pXBJY#`
z>Fwb1j;mCM;XPkmh#e+o@qdD+qqlfPF#*f>zdA!CMQTu2w(kSJDs(JvPK$r6sCOdk
z9Pni<U_0x7fUUf?5=mXTsxg_2G~lc-I=Q>OEwfzYgi7(aQe$Fb8f-#<go)`1OSUOc
z^ZS4UF9UnFfo*RKu{BzMo7#hunM8@k-htJEYjZzW82?kAhL#vN5t?fg1)-jJ(wH;?
z$#B%0m`EbGDf9J)-#7kM8+Dgf%v-^^CR2PbC5m8$a9Y2cvqz)hQUPue^0Bp-P8q>u
zf$=ZUjRGha!eePKobmGHAa@UqV7`_B2Oh!T_gYeO!h@X?T{NP53vA5i>mcMKaqvFJ
zc??dc4+%neMOF1>Jn=ZPx>t@u{VPwl3;0(-6lkZkKqW8z!c&QyWQgO5{`m2upXFlx
zDVP#Do7S*`;22xO$r3_BFM;@)!t^FH9iSUqM389su;>U%q_1uB)0OW&de?;RFw@qI
z=dxy-mW6hI%CX4(Neh)9DQ@Z35{fuJa9#29=RM3!xRxuO)CpTt3)e3I7ed(8<`ypX
zXn)XCf*>iisRZ7ZZ}7<*?uatR79O~9&$ih88yRCV4g?UHZO8H}0jdb>lYIoH=l<yn
z-NKY}*B_TVUpDBV9fc=npOTd1w)@;}O0tO3la*RNzv#Z$wK&q7{!%biZaPH;EPy&A
z-iv}O(cZs%2f=L^fp^FpK;Aa%!Kmye#CZV><bhHQ9UZ+Dh}K<pvtx}XckW`B0FRaJ
zbiDLt6W86{y|lFU>DJ_Mu4=()+2hUg%>E@r2M-yV0T^G}l<DqRCj1EZIWkb5)|qH2
zTnQH`SD4MvL8d2%ley|#FR6%#h=hcMPWq`!(DCrp3-n#-8333G#wl{{2(z=Z8!yHs
z;7Uqjmw!RJ9H$o!Vz(b;tG~TsnGGC-aWHv<DY1L0Ya(}lhWut8Lx0M|A?H#*Bj}N3
zpbXPPF)7?FX?*TAz^H)^W6zCZD_wSF<>-fRLj)!gBpxE|XE^ZiMZb!WHMu8_H<=uY
zXHk0NE^?js*0-ou87*6&otRdsgQ3FN7O!V1K0doGWTW9Ey`zOX14F|_tU_1}i9|5`
zspEdyCB+vW6hulz6%@{U7YGEo5c)l6+cIR)q)NR~{ZQ&r3rqvN<+P^&>h1cxV^Ca`
z8V+FB+wZm?EjE<ZwA>bV%^eI3#xuu*2}h~J63x2y4B=qaN@EOSuLnk;DQ@1Z6?2fy
z;QtqH)))2|NNV3+(R>T$1tL5ukGXbv1XDsR6HCtwNqBhMmBg3MT-Z6HZO>ipHL)wk
zbbp>OdWSHG&%l-m!AsFD0Q9AEwO6%&^hERewkWS*--J#DAnMpFp3s3TLF96K=ZFTo
ze2`&PuzIJXtz-7084~3ny~3Fd*f7eS^tNJ?T{73YFc(Jm2R&Kl7HQO!1J|3mDl<`g
zdlsr_#AInv6BA0nX%8lGRE%@2Zx1E*;OMyXgOAVdeDV(LtsQ&Lv$HeE;4M{j5SZOW
zL_&J8Ggn0oZfhi6NlZ!tu3{z5Kq_6HaSXmO0s!=@hlfMyf&sDxQ{ZYppf3VgA}zb)
z!COpnX_kTJO5-t~&4Jk82?;^!Ma6#ufSh`DWu+D0s*zt)Gl>K3R;%LxxA8e-5K<S7
zH}x(}y#VEY5N2?iEKC^LLq88WZ=g%1x#n&40s+OB5zNE-`k>aM1T57EWZ^mS89gnk
zTGjUVykM9OsG;p&K!w4<30`vY$axVOZpZHT*2%jJJ^rS?POp?7zm0gh@L1*bWuoPQ
z?jK-Hre-iEP(u5KAC`=Ox%h7#LAA0JU1<+B=T@6I8J=17W+*8+aHKCgE$bzD!@SDy
z-p*Ji6xc>9IcKmy>s@Iv9|Hv82Qdy{l_1juz}^Rt$;b`(-TMazgUln`zRCrLdy{z(
zh}9_7qk8SuEc;D9RoSnYnC@`CCsoEzIy%ci+X8}u9ALq~KtZ8SOUliSGc*j0)Lbl>
zRarh`HZn3IazEh*l#>;h*Y9cfL!#EY);79=vYIK@8d}W$m*-rt5TWm#-hoEuZz_x+
zkJ(JPIyqS<G;{}-;#`*Dv15|98~T+2zk)_Ywp+GPl=;rYE7$t(Ej;X0>7QW`y1q(^
z+Jb}sZw$h5HO4*+zaZx{JB^YYrJ2#wmxHT`HTc8iG^_W<>#r=8Wi$LXn82KlqZ76$
zw$=7Q_xq5<XXv5&m&)@_KxP9@e##&!Z^<{#s`xgomgp+4b`4n&u3A#$X#ZQ`#prUR
zPMrJD19PdPf{0|=jnB%SPZlUA+KI7LOiYmgx5uj-21#ga*S_)TZl4X-$?9TM5RK7c
zPw@#%W+Xy4vd%Vpa6w^`N!C)<x=Dd4Fz4IN*xR!K`rG}I7RaB$((M8B3>0r*KiLOm
z_0ixb9OSk0-|OkS{`6b^{aIlE@7ya#usWgjWQ7XcW~g0HjD`3&0byb|aK(}w=Z5j+
zk|l~SPkNQ_tEXz4A9~s+p7=L(FXuS;-7zFXf5321&|qD-ep9Uylz$U@dfr(_T01|9
zy;bQ)Ki?5YMmh@q0!<*)>ZA<yFpdG#2Vjn6tuGa$W!t>!k(vhUN7x>vDbjO{V#7Mz
zNw8|o8u$O8BBvq|=`bjR*r<esnJV6eWGk!q5=Q-QT|>jJN|OmlnkrB(B4uR6NG{&v
zU@A83A5P)HH#6H8C!X;s0-IVSl$?p_wfAaAfXeK*(KLQ2AYK5s<kWrY2IKS?cz8jX
znfhJ4?Ck7Nz!y<}l6QgnfAGfM7tprf{HXdd$)xH$dxHw5_+(U|-m=?CMh}5Q?)3Qe
zE6yg0KCsDd7+$%X^?ovjD8kx5qxIrky3HdW1vVw>Pwnqq7J40$18I1m?op%ej;f%e
z`Fm=N9glPl&d`{t5_cJ80n@623s1|ppQIguvV~wculc#(T+$bUJl9T`xbyvg<oSr=
zX(=OVb~c3O_Z_8dUTsorcw*+d#n2LL1V-YoU#%N`zaf}3fSj|zM|*Q=G%+z@rAE%G
z<X$|zYyW3z4=f3pY=aFctDu45PyO9hxyzSLGh|`|gM%zCI!hkcT5~m4BVfu)+Q8t0
zu&^-MaYxfy^-E_7aq%ybk{fGX!J3WEr;g_Tn$f*f?eH9-k2&jFv4|~XKkG~VYZZ~L
zx<2RK_&2b$)B>ow$h}j*i7jp#I;uvxS(cI-3Bw%)V-`WX$Rppv({f*I*n;ClsKrY1
zrxgRCs$97xhH6SfzwC#{qjA&3xjIg!v!=c^XAc6jo^#~o;V{WKF9aZ%K=Zu&AZ}Kn
zlJ)Ow0piU#l4POyve46<uH@<un*2o*j9C)ocE|(9yu7?L+Kq9Mk*}pwxcdN3fPjd&
z2JFbvO(t?`I;X%S6O?M%*Dfi{Y^AGO_u1O4-)3#|X5W|kH@3E*SB^Z*4)yh~KGM;J
zK}>8kG&H7*wIp0!UAu(LT$_lZ`;mXEEO$XrshNGXD$_TBL;5>EKLP44$n4CkSG9SI
zGdf;F01J-?zASj5IYCR1oU}WPS#gPFQYQ8fY!ULlK79LyCeQU_73y?OJGI;|Z8!_z
zPlv<w)5>+?fVd=AIeIf6w<P~EDpW7e8<&<pZv}{gD2ED2;|gag&VwBQLXN~JmROc^
z!;gticND&L7O`j4YQa}d4b-0KK~*Q;LVCZ^6>e1sRO10<Qfj9!s)#28?z3f4C~@Y0
z$E1DFF5aa7H|-RH#)S6oSsUt?=@a|652X~WC)Qx>n;>Nja2*JM-T-vr3J5Xa+JRQ2
zJOEGZZmrM&auQ4_8279QPZxw|gzv4kn9nz_(C<No;&d!@B|;OtG6V@3${i%25Ij~y
z85tQcTPR1-Nz67oJDYPftW_ip-3Ur2U>XN$STZs)@+~wpHNin_7YBT^z|?7e@G2w`
z$TtB^W~r)kKCU|<O}RGOcKbxG-W7*xH1vJNsw$oNkar2D?wQ!lQDj_}c?Oz0F0zz`
z7`M&$g@R}=J5S4Vn@Uq!<m=a)=%$%@w8++?VEp)2D<b-n7*FgM0J*-h2LMRbd{{Pg
zcQDN=^Guiw>s=Fjq+BZR&v9{tfk?e^F^twnYQnIM_H=^3zu=C$+t$eeZ(*Kv5O>ot
z@PkLQg}DF!l#M+C=_`kQ;m{}j(~DSnQzF-8CmXoZSvi>4qml4#mn_%lQcSw7P@fIz
zmyCOVkw8Jh6bXm!?Nn|-!HuL2ki6`kV_;ys@c`~ti)ZfI+C*R=?+p|%^+;`-FEZCD
z4sGktKWi}p?5>^feGAT%n70!!HUz>v9CUIEz(A?0DmLet;UH;-PyZ_*lBG)~Fk9Hf
zqdxEKQSB{0y+xGyglUK(z45cY@wofv!DJoI$_PuZT=3PhgE#p@r`2A{T6MwAq7l`i
z^25=Y*uWhNlf~%fe(JJet7FK}>t;E6HrdbaZ{}}vK?@jPy$K`GbI~r~N6~gNC4A$A
z+(?LY{#w`T)APtua*wc52^zukATlycN0@hD(5nn@l46sXhDi1JXv4+AN~VeF`-$>m
z2aRJ^mUHC<!-8+Y@i}?%ZkBb&e`t;ZDM;@*=a0jEKa`Cir`<H_&}$QYYu@G<iN|eq
z*>X!M@fGhF3cuQ(tvY@^*U*H5@b7}2!2Pj5FCJ^FJU#@B(=oc<nj(g>9?#UD-B6Mh
zFNgSwbcl{{hpHP6csUoH152cGATR<JdlDtKY>7i4^8A7R2%5oqrIqO(46ZSOLlz<d
zZv{mLFwUcFY5{|DkO^t<?;zZjzb)_V7(+GRZs8-p!Rmy8hIYPL4R`^WiG@caqOpia
zCnlBw&xG+Ua%X2JI=1pvf8AV)M2+PlR2+-R>-+0tdaV#z9-d^tl`&h)RpRl5v5_{g
zWzSa_p#h&9*ff<AZ9u$YSE+}JWae3CHx}qX0SMe+t@G!Vv>343`{Q&QYez9mt%%rA
zABo11Jjqaw8}iD#mViA8X17Z`Rf}HKc(rga<nnGlciJNAjXv{mmXdSffDv5`u2I#f
zX_(#J*LWM{bNiP%<mC#?3~%twCg0Dz9}Unu4&Az%5wUf*Bfts!_}h*VDOw$ZpburJ
zkKM(*BCuB;*8OKWiax^=sy@^{qrDol{nS_9iHZE+$*6-&_&KgdCHo#LUUKknU2M%K
zs~+e<msK@ap&v6k<0?PH!;QeqSo9JxXr=TP53*ymNoi?AA+AX9M9yjBk28g83S0N0
z#2!(9|Eex|-djL^v*4iscdCyPF;T<4$y`o!fO0(eTjq;O$Qcz3j0^ifA7aiM3v`R)
z1uFD_g;UgP2mogxqh9+D@%6=>176>2U>QcUm;$j<V0J38c6c<)KI#ZSo97|~?nfE?
z8`-h$rblZ0ZE~6N*sN|8=qV`vVk9R*5`m|*ItxuyB1hQ={o&b4z7^foSrTEtW3^9h
z&gfNaq^o#i<R^z3k`#to%ab&df4*JtYo=cs(t7rY;#&6j!Ap<)zEB6A5?FOXjK&_g
zd1f!#?CXD2>CN0Ka(b3|k5MaE(44;9wmWA?7&3VCrs7xtPvLn%_`dGk#qQj(GAwM+
z&4pA3In1PVfe%<Q#n9^vgTfmD6%~L$E4#aW-Y@inbEf`J3jh<xW<_LUW3zL)Nv`on
zkqTIOLYyIglS!@RBw7l9Z`s-siKWwml!miF-TQcVNnN1>@B-jzvtHbA)IKnP^t<mz
zn{MmZTQfuBWa%Cd>#B{;jJ|Lv0wpt<6T3UWI2qUtuK=?sLsPfulwk4fufRxpFo87~
zxGMu$1sK2NzWS?A!0r4iH^sF-EZlxw{~G8$5~zLUyIKU4`X(LW16t(2NEA4C{{m|w
zr&B(nLuS*#KQ5gFAC0)Yl2f^h@8|aUaE_S0j)NNCg|&sIQ2ryM{Sn=pr6N1K9E#k}
zYH34b{v(zWVe1>pr#6+vD#Lk^e#7HItC}PRc&%zEyy~wU<q+qh2_&30ZR+80gOQW?
zT>V4rVi~aUZLY6ByeovgBRg{B3F|T^p{4XYILF1>+E`!P3^F<=bFR7YgR5ncg?Nvy
zxy}ssB!(Wh^bB@M#8;n?H-`sr-`C!V_n_@gM?dpEy^SdH*`x161f8NkDxrFTXSIsi
zL=FnjMR#_0PdlfADb*<LpE|fe5DmsAae3XHL)X>q=zybL2r*v@n3Pm;exC9(l8aeH
zY6Pbq)uN-VZT?g<sKOkFMIgKUMMb;=v!~uMS|UYnKeg9$3(m?xUhcV|$IkVGmh8|J
zB~)?q(H|C58cILiYVSFJfyFv#sD(sjcawNG1^RM%I_j<|<u^D+W{yeirr|K3ZAt7e
zWUGS*OKILrYy14Z-E^{bkac~vyo~IV^gKsi^p;$OP`zpR&VB#O*%Ri&?d|agi=Yl!
z*{5f+?{aNI;X7hI_m;wkx{rmsg?78<$qhmJ1=o>@%AV7eDiS(#%VVKL`f%yzrK>Lg
z-AVppw^E%foS?0i%fHmU2UYp#ukczO7M+2?%GQB^cQ{*9#+=<XEl;2+;lk)Hq8?@g
zjx!Gr5AXtnZ8czu9v&?!R1yU7yz3|{+u<os3LZ(f-8U7|WuDvgaeAC6*ilPcc%7m%
z4|PU@!JK!@;j|ShrDjW<Pp4hyEXP>thMJ+7Ftl8Lw!k!zV8HJc+y3u`2!ziCM;6qL
z4NBLJNsDfpxJDA+okZcF+IZ%aoSm%09nBLP5FNj@CO?%jV}2tD82E_9#4^FprIPI@
zm@Y>7wr8_7fkibD&grJ3_TAr9l&)_+ZfufP75!M7SSR~YE(B_I8g*@AC|ca<dB4Y@
z-#5Kovt@TW`LK<evN1BEVt0MfV?wl8uiR4>M{Ax?l|-9b2z+HJE9~DdL=g=Z<Mn=O
zKjmhU&MI@cLUi<KNZ~Kx{e02Hv%gaK?+QOWHW?KjwCm%Zj>mf?KGneeC=fe5t{1v}
zksrOsZAsp}OY1*X8M!%}op`d+1PCa^=@t07)z;TD0cY0+VnJZkcWE+$mZ*ArYz5wf
zxSqW|l3_$eOAQu-TGZOMiV-EJR;vnHx9xUA_-hZHDX*HjE_->z#$bga`jhT$0Zr5+
zsV3v;W~Htk+{vFq6)LS~NY*k6`*SQCiMmR7e-tYT(Y028DYMlGZyyP)4%6JMA=5~{
zK(a1i<6h7-={dRn+3!N8>hhvl=DQAk>#4i7EL;`7s^{qs|40YAWZ3ZMaEYD~>Y|S#
ztN~9_QKZLHG_4T?H(NBZ*E$=C0bg;kx%{u3><h+hLWJQwJy)66W{&99?CE7)hN<?<
zlH_0UWy#*S;SG>^4RzhvA-fv>sI>TJPm3-PrFShn4e|6FKLp^564w*3SfgDe;;_xq
zvIA=bHtkmyJjQs*I7!DHUvLR{h8(zkIVHLz7R%}QEMS(6vgjq%9!`N=8|~7G7#8wn
z&)4dc{g&p{!IUhj@Uh(|jcCuJ#PMjgf-1wx)&m8og-?1E`QmfjFNLZ<R64oU%D&{w
znS5f0I*%k7Xq9qu>?p{#@)KlHhqrWLf}_FSWDhfSu0m-ff-9n?ID+0`WyrvF$A-1I
z5VyI|{iIFnZsKJ$6qMmm#9s32xW_XK%Zbaab%EmTno4smQa^`CeVZEw1Kq3NV}taV
zEi02@{ciQq<*KT@hRyN|7CFA*Rz8fhR63Q#@u*L;csbQDD)7}_RsKbGUNyU#|AOlW
zdT2NZCvY-~<|^Fn@?0<20UVJMK0i&k1DpiFiU<x0Kr>+IW4XA_e5o<MwwC><`QEaj
z8*FzQ4`%!!cfKpZY_aD>snz(EakXqETiG_dHtSsZjYrEtNrpsf!lRvo&eqstg=QD>
z7m-M&DOs>LCm#rgvQ2~}IG3EvqC4Q5xuM~c-w(?8)5e;45_m7ux38UpqhR7XRf#@n
z)KXW}a8>cmiTc}%)4W;61Y<o{Ars=IR@AmnmnBuX%gc`u!3f)FzFdt(SPBK%&_@W~
z+rEs6gEX#33gF%bDgxCra(J&>g?N&OC!KKNDo)o{w^eW#sO8<)$m`TUv!4H)Tm=I0
zeZWEh#~VO6y^x)I|HI$^3|P|YGQ1V<0S<7X5C}_B!XJD!eLA>Y3F4m<&uj!K@ZnGi
zjB9GaPbu~)cy2%!<Y)Y`x0<JZ<)!jMrp9NQ4r?2yjPq)nDnN8`(bqEkLg6l-7e{4o
zyINdL<i?4+us$%j1|Q9xDShtiAFIS~$cdcll*lc~e+Elgl`8rR$Dnmm{jO!w<O^z@
zwp6)c3h4e?#-At3ziny3T{uvCtkWD)gR!cFg@1FpfA!m`FGy|3F(xLi{VBDEN_de{
zkCpmD$0CsU!c5?4lc3r|`}ROxKjJ$I@k56Dx#=JIjiS&`uGwB?UG4RMQ=`=;|67e$
z{DfG}FZ@yPmFHYn$Fn6MC%hqIBd`J8SP$o;d-Mc=Gw}QAE#?3{9HywKXm=Q4!5|Cd
zZWHl?XP4meQ|Eyo4JO4Nepr2V=+fgJsVUTOm&+Hf+`O-BLr>JXC3z7w=mp!9ihOBW
zf4XO{-0krGP5JKhi#Xha4I%v760111t+9%K%kxV-(a`btP6zy|E~h2Jqs0Jz7@%v7
zdpud>sHBr#;4|Y+Z<<rV<>PxFD>zV1-8CBVowYwN=`$hnDLBRkn}szBgY`tYwo6pz
z>_g1qh?ME180uk-6@f({%s2h6!?duXB?(7UxP^mknl+=)HqTMbo?R`3EFy_&I{*K&
zNXA&%J*k;W2q9NYn=kzNUO@@OAVHaZ3%FJ2%~hIIT52_brmh0a&^R6~^Z_JJdyWpW
zj!<7D*hOu$BgUE1eg|BBuXpu{eJ_mbmzCP+EmP{1EeI~rf*}u2bLl~!ZmtOoX)-Bg
zlZZz-*!*+h+GOrcYPWkM{oZZDYRx){=b}n{Kc~U6n~lX=$?CZ^e(@(m#fGgX8~i5q
z{2#0Yi60BtVuLM2X<}cdf-|5$Sa4@y30hq$yY8z~ccaB<z0+HKVM)l=dB<K|uKGpQ
z>$3|HobrNGt;j<%*g1Y~73>vU5z346(fF@hj=ak;^gC==sCJH#mXki}EqW&S*Oc<c
z{NG~yseq>c3~=r^U@;DO23TPFp2X#Z3d%?@x(8T<&e=Nurft)7&_4nH$?z~jE)YOC
z1z&+pWjP>n5RWUOG+&#r;89Xhm2-~xe`CQdAh!3dJxfXeM$NtJw(psBrX;4!Kk4(+
zeYXr*^Q6ChTpG(<6K*pVhZS;b^_L*H^<d*k0mfUgpI|bK7rA`&TaMp@4>)`zO-j#$
zA3W<b-Gz{PC53QkhZ*7I*u~Y0R(f>p@Vk1*+4(6rX)dC*Z28*Oc3%#ZOZvl<LYWS&
znbCYCHNxn;wai0Yev>!seemKzUDa6K{3fBRd!PH$72YEIi{peBnAMr6ZD{r?Vh5h*
z8>Ckx>}|YotN6m}9J!?>IIx&(yAxGDHrCx!oXO=51J3`*mQ9qE{$+&BF$$Hy#hqf)
z-=uuO(BexHy=Lih1xJr9W5$Kn_v*S#@#4$Zh$MKqoU|v}=~JdAQ~3}>(EJn6CT*R$
zcHmC^H%EUco;jesUEt|<#7muU9S~p-_sh@4wvymx7Y+^%W#Y{AH#oCb>}(<;9aWd2
zsw{;C^ZkL{A0H=S4LfRLcx0*NU%!8x>{5T!wqP!Wv4C@gYBgulXL@Th89mfP)RLR%
zfSjY&qq~V83=48t*sywSK7&%?x`7VGF0uK(ey;r2IGL^A4va83^eXdfYLx4m_2qg!
zGXlzcn`hy9T^U%BS}fH=?r<jx)6~6nag<&I_`lLspF{#%dPg##7;vTHs7xl_;|0t$
zKGReD0r}>#RhJ<uevtHsvHM<js|KNOn{Mpae$bVxws`UYjJvg?qZ^kAY$Uvo!Fd~e
zHCJ)0>aR0da#%@NOegWUu-+DA%CGKxo&TDXZH0Vr^H<NZcJ@OICp)pkN2GZ*8tew5
z9`g<}X-Hu&@M^496BAb34cz+z@uUUj`m>F?8o)Gpy(hfngcta=lc&1-C>L~rUpeb6
zA4Em*>4hzXXjB>SLIAZhxyPyIB&$g-`AxxOOlUogzo;Hn2Mrsk?_MUd_6W+zkv!6l
z1XIux1GE{=HD~X$oZRM6cLIZ*ItJq!v7Kua)ro(73&2&`6-O?4*OKt8U8Tz2u+5IJ
ze`V)5Vq;79?JEt1$npMnnNq&=rg_Xy4E57ZPW!$Msj=lYece678J)5wlePD*8RIO2
zBMC;{;gFU0X=pL>D}e=EV)<}3wMR^{Is0;C&(cuI+InEU;ip7U*RZp5rj_UQ1lKO%
zhWd}wp`2AvfNKX#L-9iAx(ir3S&QngpH*Zu?QCrYGV2_*(LkeermYevQapy`+;V@!
zRbE~m7#dntM-DOD0gP$h2`}ZzC*EvWZFm&a1d;RiB!#++s^UnEsCWTpdwhnh5w<&6
zW;sutyU7TK$JUAp&S%iO<n+I4SJC&E?&FU#;cfS8+t@khKV_b~Vole;Xz;2b1e8Y$
zWL9}G(Ewr{YVK-ZHgmba=$goAXfpN7lG8&8zLK|ZLOW(&R_Yg!jb2Q5Jq^;zEJ=s=
zz^|^@A3ijr2Ab!Y6Ab##Hx{AU=RlPs-vrcrE-<T|NmKDgqth<lNbU6{$Si-ATqUlJ
zG(@VQbfa#Jw<JJ$LNut1y;E_@{sMz`)5-;gAt5AAl8k_mFuT>wmpLIM+rxR&NbYSY
z9DpOjP|wtj2lL&YQ&UrQ)K{iVtJpcTxSmHGt*w78oeHqnhUO$xSuOK=BN{Df*JZJQ
z@&Hf~Fu>}VSaA+h6%+?-;11W$el~0h*u5>z+79A+7XbxAW5r)qJ~x_wqr&tY@Dhzm
z-hq3sM2219!hvd4x<#Ay_%9wjxo8NbpEV6mdr{I5K`abTu(kd4YxJ@5y)ta1`4;Vg
zggibyMvZs9xc+z4e!O1fn;iYj?PHoZ?~^mCV*j4B(SIo9YJE%%=@}aF_emr4yF6?~
zZ1x-41!%{<QATk?R_FdqTiEPWig;#@)G`Gs%@Q2`#HzH{KC~g8t_*hAuyp0#Qgu6t
z-XkY^i(`xJ(KNO*ZL2?ZxK@+{<3Bs~V$1}7L3~bFl4AVOoK`o#K1B!p`@JI^fx!Kt
zVV8Z?%PTbT2y0DrWV3v89jTJ<(nS3;zTeT^l8^!-Z_mExan~6j7B{mgFP}O3a@yX^
zB%f^J28RB5q|rv|Z~ol5LN|rF%(TrmJ)#PYYU*752KkD5Ieco0pYp-s@<+t?6YbZK
zUTZ(1jJxbcDTxQaxyV1Uh1>#=f}V>nR1*-d*$O;<4Zx||5>@&7?_ku-@%5?iJ?L6x
zx85&uyxl?Kqz;-!8s;yH-wCkFv+s83Tp2D5)e>s+m*`)go(n4Q{*)~!PbAf$f&Mq$
z;qV#<a<<ooJ#g)w`%9-!FClwmq${Aa{Sle?0l6uKMx(m09Z0>ty-1tcm$3Rh+aM6w
zy)t>F|L}AaF1~ugces+`eb%+ua_;1F-JV{9#(~46#_oBAQlmSe_3VNbnkiDZ@EBlM
zg?3`WL@>;s>7ctapP5gPP<xstzu+L>abaz^W$6%X!8+4+k1@I!98zKOuh`u>W=$J8
z?}+lbnK6pS?WxkMY~@0bHOM*iHs&t*KD$1YsWD+GE1Df}bl(%b=3!ypge;*jR?j|W
z)=n6WYG>=PRedTLeRKRlpt#JjM~w2vbMniy=PWn5S^c(8|B@()Q0AyHVxv;sxS5FS
zFi6ap!F_{PRvsl-PbpTggl3FLGyJfag6|Y%0GM!~Q&}~9)`-oqQ;8f*n2ANz!imI}
zHCb3{#2>%Q!H28J@bE}(J*w=<7}BS^QNz}v(j5O58pD!RM!vvbMS!F?{<jy=dT;+C
z>v9ch@}h?POICH={<86u(y4ytwwCo(YgDLRa`}?@jLXp)k=3z5{HHGgyFKZ({>0Aa
z;sJSLP~c^Kn}51RcUXC5fxJ2=X#1kkws0$WD!DfxHe8@M0_3CDwi*UQSCFfRK`1Yv
zTLwaxfa|G5M3oOoXc>hm@lekSJfO%~xhD|4rHG${HsKPx=AXhW7w#@(vwQD1ah*3Q
zK5QyA3knc5x(mhE<@+p+EQoa;CKqsbw|(@AjCU6^2_2gcn#~`^bFO~_h0mv{UY{l-
z2pSZC1!&&p=wKS^mB)s+A1M6N`(r?kRM_%z9ksn!($LY?42O;^Y5nh<$7PJ0(gm+2
z_sy)GctIqq$;qJJ-*`2bv1b7cNv`6-PFg1Bo#=an@Lk128$~?UYxk26;8|1il%?{%
zY|)kt+I-W;*4Aj?{KTM1LHhLz4qT&q>LoeVzrFjzI0{b-O&bKX9Kmgofynmbi>hN1
zH{l=nl}_*Pt>M7V$^r#y!=@9@&h^^jO|@nJc%P2`QX=d+8YNPiu-j+MBKegD#g~Af
zk$eG#CyL@E<CfH$u2m4~WiUl#*pZ-JrDA{PEz7ejhEp5&%SUJa^S=J<&cV_L+lg{e
zd`hA!Jtcb}dW!iO;;fLd8Sf)%hl$&qxsx39`xHj5S)tL#VL`i9tRxGCJyi#s@K520
z?E8WrlA_7vCp_ZSKg%Q5f3QKOzO#lPmE0<a5*r-db-8BO20cvVdX#YY)&gE==7ycq
z35V#`k+h}hlII@V(m!0f9mgnYy>j$>B8&mvsu0n^0Guywm55Xcj=#<PgkUzS@BhkO
zf<(kd`lgFWEo@rmw^Z?$`s6@$X9NgX2Ulc(*QsD5kyjJR1xY9KSOiaX;G=F^f8CG3
zFK$J$o+v%4JR}|(S7{~#r<P<c(uzHQoeiES{2Xe?R`D}$(6q3Mg=R(HAh;WRCFA>_
zeEP)_Wi<LP{p+36?@`oOc*-_=b#A@ORTqY|#fjH1W1}2VysT!^lqvmI0-&Ew-;YmW
z5-1WZzgF?;oACw-FSEt#XJ)~5nZ*&+Fd;ia0bLC7_O{rC2T$GR&$menNoF1d;4nFI
zzsBGUKj_K{kF^7se~#4dX1B>DXZ!JiP>In|Yk^;h()Ne{g~o$9)yXhW%M1rsfDO`5
z-A|iEQG>3|PF;j;j**4+2NclrBo+I=r<HmF!@FNtgP$knML_!g_BGTJ052xkf>t!z
z&E7B{%YP#C#qhZlAJNySY&;`~riY%+9iRC^9ou4g;rQdUNLLdbz;$+hzf5H3Bv{4-
zvJ#o7FHMK>PV6$?eMI|Lj?TGq1(@b3{^+qM<<7wbb$HV3Zy}NXkCI&?3lBU9#`8Wo
ztF*1C9*>EOlR*)wyn;5`Bt};sB0P9~+Yh)k8y8Q^mwsR;3yiY&cPan3-2?Oc8sI?W
zqrNeSz6g4BTOGvv$cV^Pu*2sDgHT5s`D>mOG%3))ue2|pJv%(N;4vlh0nRflpPRf&
z+~ihn!f&Iv@u>W)xCr`K3F526#ZZ4ywX(9~Jw5hRXL`wLc*^=;-uzMOBX!I<F+P3}
z4Q|c|RJXpGq{g3htPRTe;9n72a9CDs2QC?i>d5`VW9vT{WiOj|f_y^ujx^Bwioa^l
zYwy5elY?n4MO6$r7QFrwCplN+k@($Ny>tKm|LR*3b3KLcGaVfr+AbiK{edNKLQM`p
z^2>vj)s?4nshRF`hvJ6uIEY%j%w=y0Jk%-DyY@lZ$T3u5@1C0RwamP;IoE%*u`l*i
zE?Md!gcKcpe!2|QkC8<Vq1>X{Lx~jl<VaD%ZH3WhKPtb1Y{U*OWB8r)-!2hYNHaHI
zfZ%8Re-uG0iXeh?&|gVAU5)(NM9TU-XA|}JaQMdl?s$p5rl`z|5D!4OpeZ*KA^v#9
zW=n5a=Gpz4mx#fy!YwjCv|##=iH8I`R)Vfqm+(4X8Pt5#QVj`;JFeHR*@g`MJ?rt{
z%iq~;BclJ+^(1&OlqcW@CkHCqFDi!X0a)3uj6-2&LA_Lh-TYDrZkD!YziV}qE1qx9
z!*Ke@V(CN){^s3lBC{8QkI^MP>aBxcZb<PgFUhvQk}Y#{C%>n5Qo-i~Nz!or-fo&t
zl#6M_{)MLnjLGW*D?{}AS0RYqnel7{!*b1qdeyDrlu$y*M_+pm{@XABt^N>PhydPA
zzQwaKG^vIy?-4SMlwU*QG~MU#2)w)qpHsJw^Od}2yH^fUK6u7oeEb#+3%Dn)0?L$6
zGi_#i0H;4cn4pQjfF=6Q`_l1Mv3=~7)6(G9EVUS45@G?pK#0y^<REK%(gYOHb6Lu=
z9@u4#|Ji+9(4t9vs5+aApCz)ddMI?Zp-LJgXP}BJ<z<@^OZhL{HT7s%tljr9B0mw4
zg0Aq4&M?!#roAYSoL{TMnCl%LfM9!n)JpduLrw$fyh$p!mQ%DzRJ`n0wME#mLpbrP
zd_S%0y;q11jlD!`>Lo7l45&Vp-#qj7n6tFirtJkcsQmHPlhQMn_}Cq6$B0X%BP3NM
z$GkD*l{Wro)hkz5MtSHr4MUp-!Y}zRs>>8}SI4sF9{aKZ4Q}X@tjN+E9g{+vRf-CU
z^c@S##+9hZPuGGd+NJ-IkD;-%2H&oILLW+w4Lk|18QnZ@d==)yBVwZ3>wL*Ms$Gd&
zTJGNyUwQa-JwGH3Zb+acm!m72oj{k28PlA91flu=*n7*Us=6>-6j2NiQ6!{AL8KK4
zX%qng3F$7SJ2xpH0xA-c(n|NHyAhG@Zt3o>^KSGz=bka{kNe~Ny5lfD9N&P&+H=h{
z=R4nc-shQ(5Gk7VLkQD`RaAweZ3U$>7e}{vyZ);2dsVJ-Xf^IFMT_y9<oD_tJ7lHF
zN*(DihKEXN(6}4|EZKc4_|3Ac44_!>v-}Y4PU|}IjhHyLAY1!u8FC)Pon8H@;UuCu
zeICnJJXBWaJ|B5<i`78rh1>qs3-8kqbS!oK`jab{vXzRrMnwz=debyE&+8>OIbz<o
zs8zZwUsR~faVr-ew8$fTh3VeTUEQueG3@t<5ncF_oA-Th*U_4&ALaqsrvie>V<}K!
z1|Tb^>*+Cew4jT@IsrDoe<4HBwZ=$b9Ek_J!I$=MEavHAK11{|{;z|^^<kmnQ5=Xk
z+-^km257szDaoE;F%i3hl|N6~bV9-Jo0n`X82h|}p0UZ#>_<yeCZl{~rBmL((Ue{(
z-$+r@<S6-qx5C=b&;7G+4mYo76o2OtQRfMs`<*(?$rX4pk}0h1*NsZ*s3SgSOd3Uf
z`oIisrZKZG%1^^Rr}Tr$&n|u8y;@No!uQKmy{7qfxO<e$CGIB^MXbeYU*aX3N4qI#
zc#>oVW8a@!6N3cIysd*p_q}49d-y6P)jw}u?G7WGo<(~n*-=AAZgyk*f$VX&!@D~i
zK2G{PqSIvRBo7XRe?cxG^<XrGv&Z`~GZ=&q+3M3BS^GuCZKVr7pES!TDkT&u=_Zzc
z%=IOEcdUGUXFzIR&VP|ezx$=Qm^(q+%{1gY@1_qW&!3PH`bAUI&*bLrE8)kjt%wS|
zy}6Wh_Nt4*donVwNf85V-*$w>m1v4Zk3-^U^_9h?ns}@SId8_eKwM39l44<9#4&NI
z?uu>P_|M7xORA6GsGOUskHp#E7#ZGKX331KOkX&=aZz8Qmz&dgqVew4&N?Si)T<h+
zIfcqWMr?v-w#Q{x{qZ;<EW=iitg`E%qH?>!7cQV8gSlb6XYHtu&Pn)$wvF*;Z~w(L
zh0_l^zW1QqkngB2m^`e-^dR9O2}j8w!Ap#Mgh%l6E?n6(#&vYmTnNVUw|l;6{D~ZH
z4_E&*t=95WTKY>m@(N!K++#_=0Qr(MI@+WeNe<g@Wb?)KXp!A}hEv+O?Bvs%A)fSV
zw#7$Iw|XR$i{nIJ!cwF^H+#e-#UW)chivXdaMNQ^DJqUss8l_$$hk8WroX@9ivi2v
zAi^-n4u1O1MKIZ-&R`vV4N9S?fboNYoU>r$B+1Yh)dp<CdATgn62tH&FaN$>E$YW&
zZ~6jU%VuHRW=9*yTasCAFUC6MnRNd;>(~uC+EXowO}kGr1vGvBtm5xosXi2^TH}BA
zhr-*So2Ds?R)LzC8Hbsf*@lZOj9K$n&5u8Uqksn{VtIQNpp$=mm@6M5&LB)jVYdKX
z_y^EqAcplw=jT;aGeLO;eM~SeIe8n5Ra9L3K4|g+Dm7219urU`2ao}tAFtvEp>M$a
zZh&H9j`f`6=g*&kP<awm8udXW5YSFxJ|wqq*Fhfrsmmejg$pw@&Od>d0i*?U9d{PH
z9)ot>Y-)1MJ-W2xH|9A?093;OIZSOR&O;)i6tUI@K(<CG%6O=#AEi@deo_u<NUvnG
zcUMFm@luJVv>wZ%B>!OnzQ}Jo7omk;)(y?_elz3ZD6TW|F_Y=<!BiX?<<DI=(@hw{
zM#<%%MED}}{nu74BV0>*yMAMLMP&aIlYvJl06r9b=0S|pb4sr*>QjxD7Qila%|Ng6
zRn*h3KnEUN!V6G2C<2{s#hzu`E-3(9T!tyB)8MDpBNTxT&^=4f@<A%N@ogL+@=Zod
z3tC*fZ;<_gK6#)R_TApzw!)ZhIvj^wuZI031|Nu*uoT7eI0O#n=|2JaDH&N=(1v2J
zKo%E+X4D;avk@h|SQt>A<Bkyunhwx&cOM_|k}3ILL*IINHLxd#OqOZjA$x8C>5=Z#
z^VfsiO!)LJJ~X87DY+-yc`eS!b&tXYvnNOs2*XC_;&AJF`<u=L{^Bzzm+u*NYwJ!C
zFb|{KZe8+SMB4AnhwH~&&;r#t{IS^R=vJUNu&}m{H}20>Y<B^XR1n+r`1tYa>G75n
zCMM>i$X2MAZ4BzizE)Fv1>H-Ht*x2c`;fex$Y~&-f&~TYgA4>%0H@GtjbMZM2!THa
zHl*jkrUvStkk3U?BLS1q*^e|KZXdwFz?9t2S7F?mZM{!VkCiw03a$j2exR->8EF~3
zcKZPehzE@SO+d&FXrnv6!XVHf1)0n;#M<l+Z}K$D8@nHs2d8>(zHRynz9xy7z|mdM
zR}XbjZhEEeK{QKK7by3To!(=Fw7Mw4Q&8nG)ikXpf0W1Mg-3KZS?LDCJXPW!++|6r
zY5V75T?lUs3-;|}XGpj{v50mmIh{PM_3&X@f5BsroA@M|{ARM&69xU7-_dQ?J=mN<
z;{iSrUCi$wxricz-{0TQcG@$GijEdvvqm818fjvGzGd0i?v~mbMU00;o;cM5dE7pL
zn49s95i3I&)d0?F%GUt~se3#;QLyLw_et||bM0qBfF1xqEMx!%u4wP<{0hjQbs&B$
zYT+YfQkwu_W#-mIhpnoiQQP?rcObT&nwx8Y+JmT=*gaO({2`v@TY0$+>6E`w4$e&u
zJji3K*kF2IznCHJAu2M_5a`)2e5pF>p;C{~!QSGE6MFR*ndxX7JOECZ+-EHF`mvRA
zpeyj#m%4h?-2N~;d{NuG<O}uVO(sRY;Q}T=7o#}J(V&+^e)SXps2A4f<NZTRT)+ar
zI?#A`FvC(mnm!5&d}ydFQP)(}#xD|l0(#_jJFoq^7Is^l5h#sfXw7&0(EVH{uMCXo
zXrS}JAp5bRc6c~)F2?DGTL`^kpK41JInqmOC;jyI*ZWEg^4rY$GH>4|jJ`oFYsDo#
zy}OW+u5Afw$`~!oT2~=iPoL;^!0#Bs_=Qokp0uuqPTJ_6z=H<Wxa@a9=jkjcmB(8%
zhueHL&gWI~>k{c5tT?5y;sMw>YQe*!gb?UnkdcuQQ@wKR!oCnfR0ssBvq1i)=xN!i
zH)B)}!<2JeMq#n+kHBKpfrhukY`DHsp$QT6s!@T))e}Gp6=)2Yz%D`4ZOi0R#B2up
zmt}(z&)T2xz!}KXm#V1GSl%$Ysz7KTA0JO|5^R!Q`j8k4^-#tfcZn;CNF2hf?>3Q*
znO>1(+xP!eb=&#oH6*EL9VgItbR_Lb*v4l6i`&J>FHmDSN7K47#q?bP|I71Uz9X4r
z7S|D#A{MaQU%GTD8}O@48s#5jV=2)1fkz&Vx$4-e#04zSg~X3ei2<tNX#^L#jSBF`
zR-wxSaC;8008vFR5EHyCklejdi^aSShe+tGay0NVxv^lS$kXuhlDR#Qh>+u&Z9+E(
z;8j}Dw-B&W&s12!WEW3TQ9Mn>LsXDW7Gy`@11?RbZb^G$=B1em0j;-Gzk1xixRm(X
zs&W^Jos*ZT`>vjQ&8wH~R}I{n*W~H&0ojGF)fp$`#zM3y=H&HRuhP+h7OR;Nq#J$e
zWoG&px<jg5Q`1HTw6yMm35f@#S(l^T$HsjOaPTz%H@jh91`+52rF0$~?C+zw;yW7u
z5&N+t_PX$1czC$(pMH@R<3!$z1JV<|{{D0D679}VMcu`rP$@CThrC_2rR3@{e(Tv?
z$%f5Q5j*y%U!tsUG(^^xJGR~@g8|?4)o)_02j%jK?5{*Ghp938g1CSiST-~eTx~ko
znk!i;9Hxu194fKapFj!?ECxy@-{<0rfMYffGUzD#_v*pzwel@V-`%wyQ3o}cMwO{I
zV>Urnz0)D575K_f9zrAIwlm58hTXVGfDNGR-4;N(d`KbmdVkN){L@83?CYBae9~W-
zTX+S#lIreYVWXneu!MhWLjg3si+Fj1$HxxI^`hS|+?tSXCJJ3*u9+l_$Tt)F8IWZg
zs4oFaqyvVyyF4fevWJ?kCv0fY){|jz3_L+-DBK+|<ZuB?;!X%1Ofkk?O3EL?Q&(=t
z^*vI~eR~ah(x`cO%7<JI_ttP8&VjaxYF34yf&#%4{kBWN^}|MSOgijThA2VOKVDX=
z)uNEcn4f=rbhfT8k?v)H_~Ln4=0<F;>OXkG)g>F^L7${sj3Gi_^<&Vi8T%Z#qApTI
zp@n=N)^xpDkfbfHr#Qs=DwZ!Am3el>7ddMKP1RJ{WOO)suMjY)y_A%cTsZOc^dtq0
z8vHyL)EZUhHvbne_AzmY&~l3&j4w(Xj(z!Ah1Xf;6*~lI(C0J*?E()#G(!6DVl1Ca
z1jr1Qy~6Sa(q!^?``T!wwoAXDGX6NS6;$j&(Rg93g8Q<EJDlI3@?aD!qW;t5r9%4z
zD&ol!8BC?-Er}S^uK5snhch+2u;Jel8h9}^Uwtw5)x?(57^S%_@!LBSXU~HRIOyIK
zRyX-PI%2)kdfonjPi5~*N>u{ArEcFmKr${@r@W2_xtlDTV{zt+7b+^mz|O;m!?P}s
zI0Z+z?2;M`C!9rBEOF1<%S(Z)toR;Ef$u4_&BeN&ast*`)ALOho*(kt|Bf;xp&%ks
z9?C0DTD+$x_?4&}MsD`7{y?b7E065-Qs~czP4xM^|0?3OdgQI;L??A`VeMd+u}pTl
zJetdA5$xAexUB5vBb5CIoyX-WAJ3;IlSvfLwt32fnDvZaO<lbikH5j6LUYDGM64c^
zb3|Xh)aptQ9kJ<_fQlnZaHlH~3v3b#H)1tGv+7+z+YCx33J%yW<h&6R6IxJ~iiwFq
zsWCZh&E6s;^oI5{OCEq8f<*FQg%by$my_T(a5Ip@ZM-eTQv<3^O$DzypkEviE%iQx
zPYgQl0*@zGv>(WmqWm||@^DK}H_mKJht(PE7+GTs_?Ki$#l$qJ`zTLRb-`sM#P_<#
zN{yJ#d}MLI5l14<`mMHA(R)YFFaLV4vt3EfYB(Cq?h?0tZ5PS+@!y7fK9+fG@7$Sw
zywQmS;013|{haoO!;GZjm4~eE8^~R3z#3k6+<B*BbHnue7y4a5Y4tkB!eMd~woR#k
zssxCWDmYYus{tot!K#px?$_KV6mc+2PwuY`K%oMHNq>KR(yqBC;@mC#(S})0;Nyu+
z!u|-css=n&uaovfY>%#iv9bj~fYjj2Y%AT-^m_Wo&MF7J2KL$te5!mWzR_F#+6)=R
z)#Zo$ScYSdKb+oPqf8kt7+*p`u)&!!{u&hd#6zPZcJ<9=tLdUqDl?v+GqQxgC7Oko
z7T-e-_SxuAmze-7U<=)wymmN07--qbOtdyOYFE2mSRN_KW2!}+W(r&8Bp*?D&8a67
z&u{H>D-+H;-y&_V%X7FrvxzhUlwy?UZM`4_EeUCFNAs0ZM*&%~`JMnXw>O;G2P}<8
zep{{lNu9QG{5q#Q0v^oke^rft?le`T5{`=CBpl60Frq*BxdX|JJB$Qaz^9y8KMK@v
zD6!z4#i|=S0ODaW?4p7dgvMOHjxHX`pbc=sT}hP3_fS0X-r_2G)9!ogzJ>#KBDWOn
z(e#sie#<G!SzaCq+SOfS^zj#U!R$Auu8@uV$fO_n10bJTLNYCxtHqqz#;1<&NI9_b
z8ckgO<(;5T#0EG|4%4BZu<5P|0C^OujSWIDI>uS_5!bU5z~Nbe1`)(*5dKht(+7&V
z!($7mIiEp=p^&}bKe@NulHuaU|3u4_xGl!fQ7-RlUE<W-!GUM{BcfM;g44L+`CzCC
zA)T)Eo!Qy>Hf#Ua;*Gn;Nqv=j|28W4!XAL}f@6_>G=|T`F}!wl90LPm1G=eYl$2gV
zmnd|r{F?)jxyjl9j<X*bF-Ng4o=W>v@gL<vyriTxuZja8<oZ9~39y=zTauq5Zwg<o
z<-H)?-KTLua$S<!!d9!H-<T6I@wU|(@9*Jw)v_+z$sGZon>Y+QY50C6j+6!T={@)a
z1e*{z4*tcYKVMm!2R^LT56-t=@ulS(Kg@{3f9`$_+;#ef@6|723B7JTE*1}IcsWGc
z0PvrVS|V~&jU%eag^5q0Xc?}qmj4Qqu;n`E2VK6R6pBO5%1TQ^&C;_Lr3`YE#H#i-
zHfSqO!ZhDaLQ8ZJLJY?D!t;>Fm2mYmCc}A)yJbAeGcgr#n=i`FdW9)7A+c2#YV(^V
z&aT6vp$zA?%Xnt*FaG^zJKZS?_wPAY@?&>2l>%BC`Tf5}N4Ud+Uix#sPv9Se;AM@8
z-lZL&-3cw@g@$fWg1H!gDxM!#l&+V#9N9S~qfCliYWHuAOQWQ}yojnSpkSo*?%Ba1
zcgW|Dl<H7#vM5(QkSFQaDGb<dptO*Fh9LQzON_t0?WHLrdN}ncn!LF6Q0d<+QxcQG
zDiahG1bO%?-&0bqMoV51kdu=$bl(H2_(Y5`V3+N3=`|`td_kF|w3*u@!AE?=BH(;|
zBEhlbTV)+gB|@{mWPkhSU?5nq4h~DCWecR|ARR?HfT2dWCZc)YG^8p#OiEtb{%=+i
zK#+oYPVRf5s%F|iYL1qWoP7JvovYQ59cTmh0J9q6)fzaW&<4s7<_nbooX<mWp=RB;
z1RPe&pP4qKmh^!BMtjQ1sf<s*C(~GKs^M2iqkzJ>{576{n6F)x)q@Ta&C&{|z16N2
zz^@=}oJNW*A%^$@>eZHTjD?MzP6tU1p_Wq0*FVSin101h^~5&39CM2L^2AW7C3F#i
z7l*4-VTFN(@nG0vCQ?jCMcG8M`Ok=Z5S1y{tEL*+>BPB#n4o_j`qB9PsQx<mK6y4r
z>Bvf*foyHG{ey!IP@LO<^hOZ8NXH|5Sj{}^)fdnA44v*{)u*{u97nEWO^quJk4Tyc
z{;0W(UR$D&@PB-9<C@u!fb!?15a=vPkqeNyL{Qzt)`#{Kxm}3ZLpm^z*DgDQ>aAaV
z`mV*}Nwt!rNa1`ctCxhOev96H^`*5r|BcH_e-9m*oUHHZlX0Q^U^X(}WH1m#h*4f%
z-Zk9tg{Tf%Nj?a_dE@^aX5t4{L!k2EreXp)XZ#4nuG-`i@t*gdBoG8ferp}0V9!}g
zX0&n0<^U94bxq@quwfsP_rgm8Tl0NXYuq<JIowPt;u~mth4#MGv)5q1(lb)|cG<_;
z4CfgyJuZWpjPkZTUkjGlp8NW!vJKQ!CU<FZa2-Sxm$v>Dl&^Mb!8Y)0I>Jwrql>Zs
zbL|(55Ds2iCZ=EG<Kv5`UeB*;`|jIbQ9Ayke+Tn(NZjy!#_vH_o*tWi)(yR&pz$;J
zk*E5&+1;5ATH$wx42~2<UZQGhh#~AR3Yg{Mvf45qZJ$BnyvAf(Vg!XUZ~8$H6b^oE
z%;>nOoE}cYe(guSj;;gqi|#=O>vf!)>`7Z&KCxdNml;mZ%!_kl|4u9Ra%%?-@i~CQ
zZmmW&-XpD-)pRI0lu=d8Qt|3*WZ)i|erI0n!}?K?h&mCDXB_RD_3CIiBKD7N@~xqn
zs)M<s!fI%*l{3jW`&n<By7hbPsnfi+DKTklrVCmg#%vTi*2rOxmRv;InF;c2f-tdo
zY4U1!-Q?NxCPKtBL&DA>3=PwZYlYystMZ@tnrdcbx%mCheT_oAo0o4~Mpo<+!?#NG
zDp-yJ`6$KTwbx>`SiV8^kY}V^&*imC{v%%W=R2tL({_zHYfDXNYGQi!)i~jrf?i8y
zI2xgkQDq*+{_^$Z_Fpu`w9ypH>ND#J{5SG;y2q=9;>$J6t&oG3rH-}d2x0bJ{kBzJ
zjV;oDQ0eEaPr<;PLFQ|{je8CC-Lt?4+f&u`T~5SEK0|RUiPx1LGkqGi@<D>Ap5>^b
z7^p+o1|=BSl=AdMrcClqSUrYn!5^!%@zjp%-GN#g4p5dQI#~EXcf_yyO0TbcFKDyX
zCzCSey>gkn;S#3Q75Dnup#dSSoE|e*lmmjYuGmTE(c}w2UrQ{1E4sl-XJ4+{s<(O?
zjU-sbt~O!KSsKQ+Yk$ouLx+#3Ji6MsxBH>_vi;*5V-g=c2DJ`VDR*QZGw|W~uX+@O
zS!``-^>Cd1q$xHG@qSvlbJg$@?afup3g?XGEjxaLg%&8UM{(OUmEQU9Z7ixsGvq3$
zUiP?vCL*F7t#S1=N87S!+#`BT<yV>EZHEFyM>UAKkD6t%?#$yfSTdi8JD+k{^t)0d
z3MJnJ2+#8|REC`Esd2Sbddtl!6KTwH>U)<FlzL4Uiw_(WNcf>xt~+JNzr<ILHdE*S
ztaAN;2mt?3%o=F=FZn{bDC<FJ|Ad7_tY{dz!`;Hc#mKWhl6puIoz5I8pPG96(qggm
z;XMpxho?@ZLordsk5lAq%Fo#Et<C>(k1fKjvg3!^ur`mnG}q5`PiE6Y;ZALP9Q2_p
zy-&AGE>a`cB*_bB2I_C~@o)9F`}}8njAJA)#u5^!7+jaS^#QY4G&ME#I-PzMU7I};
zl_6$lFWfE!qRhcSKRDRv=5&&BCy@ob-(qh^(NoRogyPu?Wg4nS&2yzusEh1e^znSZ
z?0g-ge%@uI<EhG?J||qlV%^{WtG_6A7is_BglA!~<J;fAcTa4XUM>M*v4YPW+O?Nd
zyKX)IC>DO5nu3pW;Kxp}N1Dki%8W#r$>kpngM#L=cVu&38|G*kHwOg9;sn(gsFMD-
zes*Rfl-_2-jZ6k}Jv%x&yd;ILcD^h*3{@+Q(NNo|j(GTKa#nw8X8rlVr}Wq>cNhaq
z-)Jjn4F4OC<!FxK<+an3qvisGkDou+ny_Dl9?A=>#1fuIVC2-))YdmPYFb-cLs_Bd
zrUIqY5Z^jZi{z>ucPn3uG&IP%ZS$?}ib`H3Iq<{?4E=`go%bWTXvh7A{>2FrVuR$C
zf9w2t4_l-v>v6PjJT2q8u#nIouEtf$<p&A+013bHikszb`=!Ir{LX28NzGb?9ZBYR
zbOpX)by(Uo+Ai@QRdcY|6NVKgMgy{bE{)enhNC=x)Dxr_+nIM-9BvyzGV7(NXb!>v
zN<|S0$ordqS%Afh%{V@hd+UYKhz2KfBw60=T6O@CK&u#LWEUlEd+FUPx!UFgG46c%
zt3+nwm;SR1Jw%au>k3qN?^cgddfu#MI^^O1Qar=`_~aWp8Si&NMimAnx3g%se4ScY
zX}Lbz*9rS?WNL_jyQkGV%wO#qG*P!(5SbWb=w{e(VScOj5v!lw*_!*m4G(jSYPMF$
zQWJ&dKFD=e_x6TVR|`N5p?;#qV`H~)lI|&G)`gwT8(EO>fqqi$M|aQ+z}s*kRRG+^
z`t@z?^(CB~<AMXo{gw9#({3v?ymjYH)?T3o*utH|wv<vSg^wzWoBoxkh-RbJ5(=E_
zYme5gGTt@#2(3Hvs4?8ru^^1|vPB-f{rk`<Zy>&c+6RiI2NxHYj&uvM;DB8w;xHox
zrNEm0{-T&Xyjs5HvAD_4^!zVDEwTZGmf%c*tDl1iWw_-T0GPk=IIut}@s~lq<&HVL
z9^vG#74ru@TDm#;ihec8$&_WE$9XL|*V~^eMSRZlj`?aGQ`6d>cZy38=LKn;y?2j7
z0V82Sb>$budp1L@Lb-B}PV{Ex>JN1m_hP?iX<v6*dn_$#Zg~FsEnS+Mo8N%mB`WBk
zqN17@KGZAq`}S)h>~taq;b+-zT(+>)<Nsl7^PwK;KIjU40yTo>P)6+N(q0`Moerad
zD1O(<*Q8~X?Q@QDG*9e-515G6;ASL;d6u2YyLWtR6?^2U4sHtdQkzttVAue1XP@>a
zx|{q%%-?x$%x$eNVa(6pxMeo6a#e_+So&&N?t~m4+UZJ!_1W@<eL;YfO?`nl#2#<P
zD}HX$ZavL6@$Ryl^mzhx;Y*wz09GQJe9O1oB3<a-cHhbRXzBicNTuS{nVFeR&;$oP
zUQgp7g%K8`8}7F}nk3$?Ut4@-(SNjHwI*kG#uJ`kI^F@A!H|n4^tw*eB`~Pk4y98K
z&ha~bqXBc)ylFuQD^$*zaTTTb$}wi|0lXwS?2*mBE8|tJ3%kF3NfN8+A=lQqH!AiL
zxodK0&*>d%mTWKbVEH!#M;pyRY|I;T8l!#26gFGCPLG>*Q}Y?R`$tk8hHsfq_tzCA
zS3Y`=Glh~Qccolg=#AlP$B9>#OFxv7Lmh6<uX5)pF@T7@7Z|8Xcs-YSj_A3H(r6Z*
zI<d|aKGy7}N_jF|<?08-1Obo(?0}SpW0T6&;lYaPGW~XP(8I<QKx$MdmDw>G6%hLh
z%WAuGfD);V6R4*{5vy~I_8P=IDE6wuzPb@UU8-aRwFTwl@a3LYq_%$cm-ai-RPXTB
zNRV(zP-fuvuzeAkJ)>|~;EX!lx}!YTP5Zm0uD%)He^%n^DiG7-KtYrCuVEQIK3qbA
zXzESHwb1MKcFF~i>G#}o4LxqbyK%}Mf+A-xFz)ZuPTDO3Ic{OwWiOuCrLUvO@h(t0
zNgRHn*CLRab$Bx?D=TprRFVr(al&gx-#+&~-uv^%{(K|5BV;xdB!YnFN-U-)#;EWo
z`HHZa38gVl28`yy#*6E>iV&fHvd(RWfgBW_U>9DPnDyp+F*mEYbcby19TuWW8blfz
zpdPs#trG%VN3f!c%gPi>gnJ=9+-X#RO8D!VA=C6tLBv%hGc+TEK{A%F74lBvOMRJe
z0r2Bz*{DHD3L2p(VM~i3`+--AkZ*tDA{0OOkGI<4qGKSdnfaq=x18=D85me=-Oeil
z@Nwv^f|UeJUuk0Pc@KC|+>Z9pnrG0TMb_5VmTxv10mUtg^@$p*H`tdi8uw*PqDrF-
z4A(0(r2eAMC!q;BgnY_T8`WQ|IB@^|eeojs5Qd>Y6YGkYhK7cyIK-0d3DFnu-k<4h
zQg1TNLBYQP@H&iKo)h!SZ>ugUGbE7niW?m(=Q8X4nV9~lnD-!m8SiuDr?L~j^9&cP
z!8l1$KId<O-Zv(Dc|M1dt>)5ssZ71>2~Zr^+?hfViNRwGaW*$C>^Q<q{(@%ofkIP~
z3dfx^ixHs9f}$L>3Kh1j$nRvZ5&PabjjcGkVO}~i{`6h4VYkGUc#I|nX?8rKdx8R<
z9$3%hqOR<{zVTV)nZ;cui-{Lc`3Z}qy}sYOAuxECrD-fUmp0#~opowY#K3<5-83jL
z$)zf*xjDeOs(ni|BB^zhCN}xI0OV|YKHj?DR?=!zAQlLq{T@hMuZ)%z%@+a>5L9m@
z4x!w}2F+^tczD743&&fn>QC3o9kx*PLinxJoP70F&IeW!QBPmPHlg477Wg)A5SlC=
zN`$k_L@nLM#f5Ny6>cQ~o%o+WetcNE0#!T)yJO&MV}0}Mdh55xf>4?0f~FMlsP@~%
zJQKV&^U|HVn&Y_`k3cb=DeJNujTmoWLWnG((O%rs@Vu?hqpK@R@p_7d@@uZ+-+wD$
zH}Oho2>YGx+vF6ssK&R*)*r1kX-RtwB$fWDaeX+OcB4LG^CB92(Ro7_Njb15rCTd`
ztc-2mKK}Y;WZVW*qAQ#yhY*~y*)I0rUL5=K<@yOcHRg>Q3Q$34;yHyGVs+lP0I^U}
z;N6&G_*K4Ed6W-bTLmVAGLW5-ef8?gvRxTy4<mu1>$Q2+iCxKB|H<AsQiR;$1=Mt<
zfmJRfHWtZgJv$$JcAOmJFiXU{`TIU5Ha6pqJwzy8Nw3211vMr>#p?x>d}S;wvKa_`
zfVLn3s+fdJ4?7Y>Z@0$qsw!$YBpUT)(8hPBw6wf}o~c*BK9dZVHJMV-E2>+pF9@a(
z8yj0*M<<S|LGuEp;(QEJ-eb^&sdOae^?k*#i`5<jrgoVTh0^r=N)ID@9~{_=W_0CV
z`VR}BE|{2fK?a+Gw)q}eV~DP7=`O=q2ColC`WBfB>qN=$+ilw6=E1S?LW7!t_Tb1E
zMk1;L%s>?kaW<C(=g8o1%%)=foyGR8@7|^nBYZjo5ae>wD*>NNwoos?&C18$Ka8r^
z$R=ENY9FIlM!Hp)%Yo$?)O$Dg_7Fl2iCys`IPlsH0XrJO=TeG(vE4Mi&|>nrXcr_N
z-*twjP?sDorn9*mZtKM6@>~A-04#>|6ciM|*(OOzdFC|kRFAJ{00nbNQ!d2ON|7;i
z*$JKPZf$i#X)zw0js<A4>x*DBwi&hU4>8IBSV0_q)lxrF5wHfI&9*i+Uc>ujGwe#x
znd1c+jQXtf*YRGLv56Bj-yH}+7Nrle1PM^m3)@+h$7EWo8HS(5H76RDtoP$rL36c*
zEhie!hgO~5@&^8%%{_wOvy2O?4-Hp!yXYAckBG$tm&EEwUQVe#&2r7pk*iZBsnW<|
zUUMk>8B#g7)j;FJdh|8b&tflPsEaVC%gHp+l`YULm1Y#5nirqZWmMGhv9QrRQqXm^
zmqt)SzRUeK{?1L*%%U|dKp`MdE^DmJbxtw9jq{BL-PqWeBj66GQ-7Z3n^ztrQ7ypI
z(=YiI`$+rwCq_A}BW(XWkN-qj6iL5&)xznZsm#08@`PH0A5#0GUG5jJi18{+&LaBN
z5u==wSx-tqsXcYNVMG=d7#Yjkmz^_J>_&XCIQ+-i^PAb)b!zH2&~S3>yGU0KZG-S8
z=_|(jZmnOwBhEK4O+AW7D`8B-hSl@c_h@G{DE7Bmo~BJqG{0#_V1HGJb49|O>dW|e
zGV9QvcT=6du@*%46os<j%Car}zzL2Fl|SXiL_MNK2{9Y#Q8M1N?o(ccXJ)_P8O1y!
zlkvSuJV9md2W*BEAhCqqp<MFIY^>bjQ@{CmrFx(Z6klWZMqH1lC}tvL13^qm@z0L7
zwsv4bR8*9VKRzKLjc6(4gPPSEz$Mu3k(EjX3l0#Cz{a=mADs$q1vDSB_$w=;o}9-%
z`}p{nO4Ofs_ghzYH`(owY#V3T-9o%PP_nY=l~*h&B}F6GI@?@PIjfhU9Z+?QFX*BA
zZbQ?z5u#yC-MiC0HeO`Z%|Xv(Z?7OzO7P{V<<<z_u-YQNMyUf#MNVr1eQmHKt0;ff
zzU@NWx#3d7#U_5bt1rr5(4K0U5$dcP&6(lm=Ekz)X&VBNfmC4jJOmI!yYe$0K*ET2
zK@~)OHAmus?5F<ZgoKxeniDF4T((VBEJFqi4jT%h>7SUBvQ?`!Ii%@*fsgPt(8_eG
znpY%3VUL!ZyUgHzQElt^Y3a47WJ`x?k~?=&pfV^L2=(n=;IDEyJ(!zYb}q78QGxpG
zW>o0)JiS)Btt1mPp4d^n{<WYv&eO+fsd7bMkQdp?6ZJ`5E7TtAy0>u7jb~Wa2UExg
z7~5r>RvaJm7mK|OU`-0gTa1s!;+UWItJCGCK)q;b?Ze#_dQq;z6`JH<q9bL`39S8q
z#ZNH{txNf`$1|u-g;Ke|<_@GP8`a2<E8?M7Jmazx#q=1>&&Oq_;7?~YRbuvpO)f^K
z%6qiC(*5Du?diF>WH8#2$x^Wj&z`xPutzT29dRCcIM9V%DG!YI7S>+!oSuG*O~H2@
z<yHhW*h2sod_FzdXrM?39(VlLm~LXL25RM!YrISaSA!ixt%O#_>o18`)P}e`kaZ$x
z2(f?pEK6CnN_9!e*GWX~VD{Q0F+@)om1%_SWN$6%vE0Oy0ojC!gVVo3<`ZnEqgc`~
zSTao}>L-;_Ca#Uu4IZ;#U_Jm|qIn!-z$uV{Mou+8O<<X%5(TG*bL}xyTPzRCXF5}5
zX+*nV=SeSG1I-JQsN-BT0*Zeeso=tI?(Q;SmSZVCa>~WLW>38Uz@9ERY;9m4NAG=u
zQ{o{Xp9ZSVDHL+yd*JZi5kZ4zx)HZRV`#s80ofGesKTGtYp%<o)JTCvVmeRzFxMD9
zHtM=Thh@UG11!{Yq}V?Q5`~t&ucKz&v3cZSqYAtJ!xV$L0u6}cG&qWuAitI2Mkk&O
z%d&Jix7lWYT^k|*%~+S+aH<7x&C&u64!0NXi#kHk6C5J%+#U{a4)E^e@mUbc>ej7n
zfeENr1+XNhUQW-w#)gLEpTpyhON{+^bHexp>tcSVG+>SsK~f|nAmBS}Xyf|N_(OU6
z{#I62tWc?9g9Uw098J2BdX;anxTB_Vc1OQmYQ$x_b}JuY6(dcN`r0iwunXQ?cS#QF
zk<gf)SWUkyFz9?vV|S8-Z7mXJ_*!x?$5iFu)0?k<U~A0mvd4Lv5<ifN$$$#D*r8_e
zf)OZK60;fvwCLqI?yY9RBQ%*6HS6U;hE@{V-7^9ln(wo+%As07s8S)aj^HrY=zL(R
zkS-5SdLOV1IZI-BgT)cE;z?31H2K=97ep1!X)O%&e2j%ZqG_yjh*THCUp-(0B%3b@
zM<rhOh4s72jefp&<nyNW))EN}f{%ThxS};Sc|`S*$eC%LK)GQmvZmL1GNtIo>WAv`
z7f?@Dk8$aeLCt%N#lFl)J0=H_5~AdulPp7)1MjFasxo!&KeNU=Unc2(H=}$owzt0n
z#1*$4)DWY#feAqW`ZCZ4g3{*k@v7R{BLG_Uf_9JTSh+IrCeDoz+`b(T9HvG<)3mvn
z)=?0bxVfCyz6i*oLntY*+RS(0`a8`3@czuSH{U6UIy{>~=<%U2yF37sXavzTzQS7K
zL>ICN5F7>IMlvd83p9r@_jksNsB`kT(WRs`mgTM7SPnF?BPthL-I8YYZ!IyT5>+@V
zd4rC6I66;&3NWgd%E7zf+Mnx>;9ycwb#;_q_~g{inRS<0)|<Vxn>pynaM0B|?Jl!;
zJMUhUhRcpThx|LpcYL?5ZHD7K|M?u&jT=DUSH|kF`L_P&&)){DX%Iw4JhhA~G#QKo
zkb>CZ_yhgb>6w{C6agvFYxPv@jislgJS6fJW-}R(hW+HmZZh!w<HwJ2Q`nb1-kqHw
zB%-<GVI6|ZP1s^T5(#KgV}NHm0{Pddy}4Q)HNAZi+#J}74FT6wZw8)9Dv*Rl#?5q~
ztt&n&5VSFBe@xrBY~U)b5l%omp7h=E$6T~XU~sVJDJLd1r`^EyENk~F^z#E^Bu+M2
za;VrcUf7>J6IppEW;xEAT)Y+&<nM}o(El-*npqa!``aRJ1vJzmhqOr_Hpj@sY*RZM
zVs%7@qReKYYiq0!44sVc<kC_)v^2|rEBZEP7?VTU7)m&)bCm<-ikdD3%X7-hwm*(!
z=X0Bxy20gOc8sEOaB^De6rtcXmws2t#pBscZy&~>95*m9fLo{D*!i&pM34?a`!6~g
zQvhEMss5(Hvz59?;bB4ET)**n@w5Xbay)8sgr`w&!657TV!xb#Qh@cF^TkUCbjlwp
z_y1s-!buJRhAdd~nM`!usG;PMFE4~%^mLjQjV6I}BMeZ744X+aT@-UK$}N5nl&W-e
zS?-1;Mmczw#60#EomIfS#;VtHM<SLF#^5J!VPS##b34^f+#45C#&oh4=ZmwcL~}Ve
z2=nUwE-yYl;R?35p2~X6f^jGs*+oie>cD8((jBLH<r4umYV}E{78rI51GJ|HV&jnL
z=*;bHoA6dWl=+4uu?w7{=@l2z!j!b0z`#IO!1saoAWh&S6c`*9mJF~UdC0Q`QK6y=
zpy-O${lgQ5X-~+Y^nH2RMAVOL59QPVN2EG%As_)6mH-Xs>43@zq{<sqOG=zUOY?SI
z0WJh~6|N{%yspr}(CdbR5vciV=_M2A`Dw7JFgJ`0X0Thxv)a8PMISI56;RXsoe8nS
zIDQ8;e6%9yX-|6{f7#va8k~m!GSn3k5ovB{5P?JX^z@YYsl|6V^KDlO`WN9CL?FVS
z;v}Pw!4UA(fgjE5`SZ3F5h7AjXG!KS5c2eeFQCZ@UG~T#x6Yf|>=Ep$iOP+RH3`+Q
z-riZ*`zgT^(RSuh+E8`5iJz+D7F8JAtHOaq#a;11h5E4}@q!59&=u$NVbP@m($#dd
zR33#41aBPDQ;Bex8$p17xYMs$0O5ebnt~3z*+_O%y3!FaEV%v)0N3oC0Xxa{gv^aD
zO?z&3ackdKrc%Puu_(VH*u@HG=H=NSZ6aPuT>yR0>Mt`wo)La3X~8iU>C4O>+q=&e
z8f0Y*OP{}0NtoMxIn09B$s%?!rli1T=X@bfPy+QS63FK?VFd>D8p&H(<<7RpA}?IL
z{9C}MR;)GEZ4d0HP#CN+F`i%j{fZhIeuj_%S!uCBOUu^q*H|-f3edRgJH}WMb~n0-
z`Cwuj-SF_1;q(pK$C`}DPrW<L<LYTK#htvv1|7zGTpD3XR4?0?LwEH;NVjK)6Jn*r
zam<4Lbjg<~t79KoBsd?tM3?2A&leigAGz*ZWhbtmetQNw17k#F>uif0TG~^OuxA%F
ziv@2zeyy^3s5WzEBX+nD+`}MkgnQT1Q9N8>;bm4nw|MTbE~`U&e+o&@($^)@bSv@H
zlZjPKQ)$(4|5G``$iPI21(^@X9xf%!t*<TYBE{v5b+<OfjI3P<)z3CZH?cR7Dc*m_
z@x5i94*R>z0&7N*%8mI1em>;{@bRUG+@Ey0FI)Rv1BR~5LxhaK_Z7MX4f~Q=!-0};
zVTO0gmU#+B&Bp~?B~unT_;q}zl}3G4zFGG36lvo>L`&9gpWj_Th;jl$dg0OsjD@sL
zj-|9@ulbfAvSQqJ!ko0>n^@Q6UXEJynrR?k(c85*%ui)9A@*d$%8qEaY$9;?$Yh0e
z(wvW9VP-1I?hLU879styR&V{iZ@q<b^$`dj;_VEbltHCmhKxoO&}*UTV(1eK*Dv>)
z;HE}M+@p6+o8bDfiH#MISbeXtlRt{=`T4>S(4EV5gCla>w$buZ`QLd!ljfS!6g&5o
z;0Gx^Q7Sc&!NJpMT>pNJraLWHO@fR2=l}7O7e1H&{dQgYf8J-}$5s3DPeDVg7QD57
ze&m54>3wek<KM+NLe>9$4*6#g68-zd6I@h=h2~~dZ>DG`77Cuu+oC~yIqnzkY{tJQ
zbS)6N=a*Gqgj-fq2iQz}U~f~)s;DxT|J4|5ZHASmF4%fxCk@s^ZnV&MM2_<O^7R)m
zE@}6cK?jxXMHAETB4NA8QA^$XuBm>=DXUy%Wp9>z;MFpX$rCfpT8cBK8KFGuxVwjc
zZYt4uE?;0qlMC3M(%)_sl(lgjx1{~8T3Qg(Q&ot!x>a-$D?wL30qJi%Cq-hTGfX*p
zy7992bJkK!?fLV&HE0E~p{1Z5gqK7(jF<UY6bWJ*3X2#jRs5OEqVn8-$sr^bYa5((
zRd%tH%Efk#yGgRGc-!;*Y((>a4s(P*(V$-?y;<`j)jEYWH3U!QGtR3^cM=J85-JOV
zEm`EZV)m7Chv_So&YyK62PVF{-tSkvaX&LjqRNAXaJ|5yAKNq}R42))J=L1teJh=n
zy2ErCl7Fv1IKTdM|LxbQ+R@j3V`aKCOVp#|0)hFvQ<~~ZMCVtf7@@B6WN^UX$~N-8
zpHPveaByXjoXPWO9wG}X85VWLyMHf~IlmAWLGhIuS)G#;(fBD$$s?ipxo***tx5B<
zu-+@@*QJG8IosU{e9Img5)?GG%F>ki{d@H&wH&b>Cwoq=N=EMRKt~!^-k^z47^6Zn
z=NVIG{%w6y!o@>QP3Z(A746Ae%<()ktvv3^Q=Z*k)yXXZt8(FW=aex_S0+3>yx-qh
zwiQKsx56Z(9n{^Z1~b_42OOB>bP1i#&ieb&^9C8xdsRBALn7*RzDBo3y3_f(6-7bX
z3ev+P>cS<;49A-r2=+*McGZ~!%IkZ%%9**T%kAFBBY8-y2ZZGt{AK7*`Dwz|lsW$7
zh)|ZbJ`m?TUH3?9Fv=+|yCioH0<qCHr%#I$I%{WV#NVOugXnt59FDs}dWaig%Ac$D
z&9ry#3IHT~l_`GYSQdRm{duBxn0g6k$>!tmY13)dIIbsz-*jm$$_+Wfopo7>n|Xwl
zDu(>9^wrk&nCsP8ldAZ~^TJf#y{n{Z?C<Yy4XKXKP%!Sau9;t94DxSm#jlBGQ1EFv
z&^tId%6nfF(cIe1s8ZkDZU79>zlR${_{R8FnK1C@VSK=Y@>${FG6sfx+Bl_)i_6T6
zkhkP_x+!i_G_@3k?w1Xo>fwQTu2kb8Gk2bFbv3%5zdr$a)b?ePyU+6KYbJ@<r-QD_
zj7=OH7R{l^L}J@I{9CtYO3}|ZfItXp5i|8|%#!<IX_C^P-PI(u{lYelnTor!_SY_i
zV8IS69?-3Gq|bjR0j|LIRBA3x_Mb1?PPxZ|7kf`mPqUc+CX3{VU@K4P2nQ9JML1WL
zf6?fI<2o>*E4G|nD1wxk=sNNCVZ|o)E;5_z@AjL!KY`(f*^zB+J^E-P){nPu{0rhk
zVdLypESzZ)+*A?+G&93Jt|N>_oya=sG=lS|v!^phfJKs*_oh^Qm?HUc*8fnc#3v{r
zu~8=ZW`k>i3)z+E$Kt9_5_}EeoaQ@yfbBTNS8&>(yUIxZ_Xd~HHPBK7;7Ejqsj4(K
ziqkY-P3ABMmxz46%mK@>cYL2Kmc(|*Ei2HCdttl(od^FHsO{bYhr6PGiL#3bcGr&`
zCtIrqZQU5juu(l(i)d+~e53D4qpFR=?G2-efBy<c5dJFcW9M|{e2gjHXr6DG`1zxw
z$zO`h`1^nVUkooEyo0}A_77q3f4}}uSoHsU<NtZM{U4Z$nUz)a3GykNK3l}K5HEGR
znc3;4=_Pve_|DOpYTp(NJ!MRWKq^g5O(F+AIb%pj2a>j9$Z0C17tEJTtW5nc>yX}Z
zaIlH99`@`qBJ3hZQ)_e3D1`A4au=5n7pqE|Wr42D^i=Ec5?y9E>PX&tgYkbC$lqPT
z^MBwy{=a)a|9`x3`KY=ZJrLm=LNyg?8bNkzW0^fYa-JVPq(G{I_>q>IN`cYLD(*j1
zRjr%$syg!P*B?%+CCLLjqN1WIfGdC+cN(NQi~s@3LhNgn^7k}ALsLs{x>OE9AJi|T
zL8uv@j4ZRGL)yW?q0xlh7{cxd9*101aSPJG3y|EiL6rQnY!mr(-5wlIpDIc5pX5V@
zAPwTQCIi+ctw+Bh1cKXjKna&dG~$WAUxq?@Wcq(@>zsVJ+nx@W-SR`x+~Jj`mKGap
zw*PKfZghKn$fPi?xw5j-ECUKt=U-~^P1_R{-ypH}G7wY%rq|Y7Yu>c#L3dxSSb3{i
z5#h-h@<&c`PV?$2E**B{imRq9dDvp>X&7^rx^t`?LI}CXgOgTJ^=@ayE(WJpeRAo2
zt6rF5f${@)pGn&bfmCDSat`3wn4dLAergIOJT72Ilb0>HY7d_QSw99*$K{*`CdXr_
zcJDxmKz;oiMCpkM3B?I~i1pS+lpT?DAIw0{D@zy9)oyD?Ag-lQSvgevR95g2Dk1PG
zYZjmmhz*-t#g+cFZjBx&CoTOQwCNCG?wG-vUVCSAs*h%wz|brUU6Qkuf4Ab^s@DxF
zrXdj#X(VRF&mg5?gAo01ig4u}F_5EX{h+o0<%?wCj-GeFLjJJV0h)?HIH%je9OOrX
zIt_yT{r%I$DyGZ#)h0P1#J{eg!BSjy6GtmYrNC@YBXcB2kS4JG;e8H8+vkG#>Krl`
znU)fgZ10b%>y!*G@5b?A<8L_sZfWc_k1Tq)Do~CqE32C~h(|=8v?~#ja;BSuiG^vk
z@x{VxW!bu>xtd<kR&o25xQF|kAe(*|#2ov)m~+Q%)ms<J!&1nZc2CER3J6Zxp#6lt
z*VqH{V%LvvT$D4EXP+n-UU7tkegX9A+O7=ALx!Je2TR4oBm>pvyX7|wK!Jm!BYG;T
zpCQzmG79|sh%jlx0bNBi$#Vm1%34|$B}U(F(&y@WtZd#Alb*1>jq6WzvSDLuN))fS
zRw3fTZCCsgpQ~L>D>SisW5YPe-~S7fBOj?0U%aB;yZCkj=GX3t)BeIl73DUnv11Vy
z?ds%Yfr``z`olKakM(!-3RHYzu*SxO!<}E5WnEo;o1>!J)*gn8MP7fw*{fovq48Qu
zoc|p%YuC!kdS+#Xkz$SRXqPzf3!%eVY*0{qR1~SGjDUbZ;vrerTNe4WY9hBK`NmiQ
z&&gG26h>9uA$Q?N_@w&Uksn!qz%^#$_RcpfFBgYX+%F4rDCA{J_dYt9my^t5r-|v?
zJNi@bu3bZYYa0~qs@V>;rsFqvCPG%T9=Yxij~>l(H?=j(W;Z58s*0qW$6q#*zOQsy
zlqNT~;Lq>JC&g#qD>X{h#<SEiQ~iBADlTSOA{Ne9R`o@>61P3%q!~;+s*gP^v-V<I
zeCw4zwyodvgF~Y;`F!J8OEk?3GS%}2EIArjyTvQYj8|h$i1Tbh^cYm)_Je44w$hm!
z+vRu_%OzB^rTryN_(i*Zw~?`BIT3mu5o*p@>qmw*6*pxdPQP@MEU_=v6peT1OuyxI
zom&YjeS1EP3GK9=T!pgHtfDbTae9ReX(-L2vS+pG)z^|`x>@%<!$M#%U!%DflU2>>
z9*;0i>pq$<RA9{CjT!E#lw`~tDHQZC-+R_6awk2go@9#EXvlk$jhM}8y}&U~6Ny_l
z`kd|6gNlwNE5>r$fcFv)D)#GNwQ5vu;woG}9|6znkZbl>b)(QRd!7h>q>VnwG?9?Z
zcl@FuFvwOloxm|(lbJ|-u%YywML;EmqsKiD^qm=V@ttNbUwk#gnZTe}<dv~^itJ_T
zXD1^3*hb{)i`GZ%M&H}b@7(oB5E(xcpbZblSA2N>aqIT-NkVm~9Jp2D8xR>G<3{Dq
zx5$qmX_^kG(6BVt>ia7%pC*`xGAc^Ax=Ijr6t$=}_F!Rll__ebYLA~~=qI~oxvO(2
zMn<ci9ZgAbSw0ZrU^h?W9?ECfI_MYX=v&`4WqmU&S;0A{Dp7hihM9tU{?T$6@_HC+
znu9BL%9(vzr;23cY`^#Z!vgR#-3abgtYQwL;_YNpU~3u_L`eG>6d+~Ahi~679Im;f
z7^;}+k~57E?b_lk3vv3Q?_?A5tV^+qG03-{;r#v<<Zyp;Ik{j02BnEr7B{-7Iew0p
zfnGmZ>G$qj7WH@&VK}_-mhsrSg6jo`?Q7eR{>kp-f~^j1HXI!JvY~<|UP(ENfYLsn
zo>&^(0=CodI*}ojt$*YC6uz`+ObxWQoWYeL#Bp)$$w?0@>(A7j(q5lnsug~_TKq(L
z$0L8TFe=+rl7($0Pyf-%yuWn=v}3Xh%x*d|Px4?N;}hMP*jBow-|+I4gsmfklm#7|
zM4;OC0ESqgWKzOY-oMdIN(98@c(*5(=tY~DvY`T9SRxRJH$S(utrT;&8o8cBZ{e7+
zjGjkD`+IzR;crz*pci71?{{1^gKFQuw_%}IEg_hDKIUOa6qw-1W?UMVFNQoFJaTx%
zIJ%I7`<cAVBDvvjG$MZ+hKIXj)UTkIE>G`AYHbuHbOlE=(6PvT>3teq)s0RG{OPr1
zc@e3n1)uwkii|C5Z-M3Nsj#IdRJeLR^J%yce^3+K)XxMaocRoEtUkug%iTGtv_Cl9
z2h$y`Ca;n>yEGN=u3q}EU8j?97!l?a<>0it1m?}Lz&Jho``!QjR;8Iz=4=_Vs#WAX
z3H8^!o7sL@)1OIa)yIjL6-!d;b6iiq@oR)#E|Ar7#gTe$I`rW9dY&?Kh6;@+G50b-
zSl{3LRvH0nSesA6w}&q7i3Xk?Ox;_N5~c4xIP)D&LfS1WtOVTjC9^XY3RW_#XZPvj
z?N>`oh*BHfJ!x;sQxbPEb8k&LB4AQ@u2Pz{G9;vk_?%fS_BTjPh9K2Kqex>=+#kF0
zb*jTbsnx6+Be(N*+?<jLdqwsp4)4Y==7BxQ&dR>NJbl71>XbsO@Xa-m9?H_xK2q)<
z#+Bj8Q{l>lyfmCNiX1H`&#-W)Qj{x74^3KJnVuz{Piu7$B+ZJybC8#7-Zv}Ppe=DA
zetP7DeRQgy$>*p>5V;aH)wQ|BW<5t<FBzNK-)}0HT~#$(u^#1^Z@QAyG)mvmF2!GQ
z7*|^OrPC`Yxo3BZdFQyPU??n5WqII3zKw)h*|SXk)91}N6ta=Oe7Bnu6!FMOm%W<)
zJ|oSB9pVBS9fx`R>owo~u+n^!AC|E#9dIC^UajYTM9)*I7d4d+sAr}h^08Gs++OL;
z%1q*Af1XLqZehK%icdrUoVSu)`|XuISNM-pIQmVyBA3`xHKYULj4n0aU3j5T=%@Eq
zEbxnlhU0vxyW>gC5#I9J+Ss4JpD%J3lCX3L^<F6tC$%5#1ZPPUeWyR<hY<!#3kh+<
z#(h`&3e?Ny{oc9Q+9vz4+_`R+QMkDdZ~K?vp7Bo!j`^Ti*WUO1r#)?g-nRFaTp3AJ
zdXUH^=C!mVB-!_GDGxam{^m1zuwe(tU~O-8e<VTn!*^J(o~(>xDNC+3-YtSO;tLl(
zloEENNRn0X9FwEIX9)4;-->-UVJ}<s@@^jQpI+TOe_`j@QfVe(W@aHTPg)j+meE9e
zQcVg9HouP+h`)Kc@0jpZZsOHuMaEn1_9y!du|vhKRj~J)^zu^8?2EaCIeK@(5b{L|
zj5GCg^Exb6R@-C86s+dGMr8cf!ki8Vo9x8$_a6zUtK@h5uHi^C_ugG)%99b3Z10eA
z&FVL|RJGaPHcZsjqb(QCCKh>-;yNcuJW!AaR$)Z_0cx1l*HFpTn!e?M#AN>PEZed@
zyfRa;u#7{+qR94`blC803zd_L<0U~PX&KHf1U}^N=3HDVY)UfpIo9ceHKo5Vm-8_0
zzlkO|&XL46>hDXI@|CIbBO8%3oe7KxN$jr-kH>kp|M~y}quxxj;!Sepp-U?I$?^8j
zR*r>6RWrRe{HSr9rz)hM0hmqWz%f+gSSN`pRkzh*wDW<m3kjQa9dQlIORkuZ%n%X3
zg-CU$0_xN6SuS?v!q_FENL~wMxv%iv*aZ$Q=f(U|U(&?s?dzg4XAgXBtef=~WjJKM
zx?gBwdMN!|#fyNNv@2|QZ{9#A@bVWI$+HrNYz^0SJ`Tk-4b>?eqxBDAw$nC&$K@4l
z<u1nR;~A+c{g{d=u}B%Kt^E57UkH}@Q%cJA7DB@FSP}I`E_;8rjBd$~kItN*3{4*)
z6HfPgs++GT9|H@2Vk)x;St^4lGNzsyS9#QP?Ygwd3sc@cYK8IDLF8ATu!Yfs@g-P0
zmBieoDrRa37HT0=^8K*Q7QKS2INc5X8kuH0_u4rhjEqZR8(m6A)EyG{SEP?A6}>jk
z7)~jwj7IikFt}|0d}Ee2R`F+-LV7&;7acdV6i30wzG1~Zg^H>It*d)mvf#vx_S{pM
za<c3GUEiEfHTltz=QKf~f?Ju9<V8uUBkBbRIM|nzLDAqd-|2xO%7tP(IVm`9Q)Ml8
zS<;zY)g+=C?aD1t7<WmJd+)*ObRbs2ZoHaEUmTb{XtKzq6AXji`Iwd*f*6-!7&$4s
z8|1ICII6CRL;Ntl<~vapQ{T2GfqL9$djD=iLt>UmeBHSj!<L!oomPo{0xZN?mgP%q
z7yfNwUc6z+RApu2<63dC_O{9w_~#R=<_;69#=yxd_=fLLPT1pFo;h*KuB}nX;fS90
zj|jxp&gR~3U$dr5reLz9yjEgJJC<fkPh&71+l|y8B_2rG6r_1-I~i{GE9Bl1by-5g
z+0j>Mi5XVymDtxw@}s6%>Qi9c)>&JnTNyHk@glM+WiCdrcb!hrY|loS)90UO+ue-8
z1iSfEoUAzO&dx4Y-Z1USdp774VV5GS>W}yoGN{v;{zPA1z%wKWf++5-N~p0`rWJvI
z(8TjAK6)r!_AS}W?40ok%@Z|sokZLD9$J^f=6k*CYzxupPGMXljBi=*Y(G1zWFhi(
zG(W`=s_;KIejya71Zx1qUCT4{)S_>g*3?!lg>8L98V!1mdyVC_a*ET+#a5DcNP_j*
z(<L$^O0$%Mtb8YRAK~$?S}SDO=sJW9!G*d7gWCe$g=q_GnpuyDd#?78nd$@^vl~yw
zE~%?gq87pTxSQ~VSSy^4ue<@T1gn)T<raL{a~GH1vS%)d`ufHvF)mCQ*_#ok%{ro8
zzO9j=4j7A?>fH%HoGX06`%8ozfl|J95LLsQxi8OJ<Ecff%m(IdBv5yE6Zb&9#(C+u
zczv@Yu7YUt^lXHAZR(5UhRsW!CtkxLKlCj{(OsN-7+B@(%G9q&McD`$hG<M~wGB38
z&2xq+RXFma)`J@joVaL&|AVNn4vVT=`$ka|5DV#gloF6;=ui-l?v5d(yJILpKuH1V
zF3F*X?v&1<ySqET#d*(jzCZNh^1_+D_FC({f3?>Z@3TH;faA!X-WGeR&@~>&U1gMQ
z4!(f!@o0oA<|WXh-RABWPkCN{AJ%vAUJiP8;X;J7zO*#L=^Hvl;QAj|%d2DmzOQMn
z*{jEa4P@=cHe;$Zvu};C7*dsbN3O{&Bd=a<$e{Jdy(5GM!YUyL0E&bpCe{X0XjVJv
z%!xb?nY$4fwH|#!WY0#U_JArPP8d0y{niIrAhyI3jD!1zLQK4yHQE`br(DjSEH1C-
zSZ?lE^9~giSuYKLJ;jx=vcl8;>^ZFhRub*Hc~N{+^4>jQ)DptW&p3UQ`t2KcLL!_u
zP|0Jq-8NL=ywX=WVvYFzW-2~@RO{tmI<*s5VH!M}cvp8jU8>?(Q#Z~GCxVO!jbYvY
z`Q2G_J$IVScq`T4cdnTwFR$c=+qu4yIc_+3xww%v@-Ek-YOPA~d6MKvqfydW6E=0=
zOfsa#aqFn9p?k`BL`&sl*gyT2KK)>^NzG?6B(!pmD?;8ru@@{wtt&Qq%RKgk4;Sfs
zJ6j_|LL^fg$vYa|Yvy0-&m=dw^pzb>%#cF`Euv*>k^u^JUwf7F7Keu@#F(+tG?lb7
zM^mINNci2=VG8miO+RI8_FXizlI-q>CW`}BA?f!1#&Gowa_pXPiphVKO}T`aruad;
zS9S~re&(CQxm`7Do|6q`n~-hZ?0F3I*yxW?{t=E{<%$SP-3`4FeImo{+Jts_zMk8T
zHS*zuA^<0eeSLD8izaY^u<YQg)4X2PyCL_x^}-jT!S)yKj*+lAH&N)#DNL?t3XnO6
zJiMI0DIZtVK5+kzN}YQ_rX+5e#II5N8Oo^j$H1ztz>;2{fzxt|Dp-k!;<G2QK+f-J
z-pu8dje^GwrrkN<&5$?q|D0Hie8e&X_{Jy1z|dr6aB0$!TkTG}R&vj3I<3(1+F$*<
z|HfI1=&}(@4s1&q=SoNMb<5@PXt<QkCD-k*cFozU>7{&rM?tmKhe(W^g%L6~Ua^Mp
z=jx!5dV~l7JNIh%+as|drD75-CBplgm*TAy>Qw&`r_5%~V`hK)5oKmpo_nIY>q-}$
zy~C2<Aq##02~o7O`jX0R>fCb@xn%0((KY>~k%3u02}plZx+v12_5I57TAs9ySJK{~
z)LEGs<=kuKQqumOAu~_&fh-ND&0#362~5yixq($pJ2KxwXzf*~S@$<*v7v#eUP!$W
z2k2gOjtNN1vFo}g$IV>=_Du9ze4>1rGPkVaciWywmkxGU7ZdA#y5m8a2IHO~C3WY2
zKABuxY-(}{GM?=g@COsF??K<zWSmgHKA8?cs5=~2UiKv9O=(mpMpb>hM}R0$DBkQ<
z*H&NN)y9HTmhsF%gu2sp216mVLUXCMlwUfWtw`cRNKeMjuAkI(qr32AVRgqf->$><
zWKpWKdw6Zisk>WeAb5tc(2MpzRTd}JHj@EGe8pJ41TEN8PHbbtHAP<oUVo*0Vl|ZA
zYONB^Ej&6HTJd~n%N>^6s5i)0C^K3;17qNpQPxf>aNZWXziY6Y>Ph=O64xUAsrwL8
zF}H*GUH#=j)F+IC6XYe|mHkv+A(sPG%VJ%k&=lOW*dUSK$^y?p(}{3=EVTO;JNlwQ
z%O_5>+p=QoTg4pm^t%vVf_{&M(250{>1sipcz6Gpej<<QMWPV0;ybviXW@W%jx0|3
zuls68(`7ZLZA>|ehbPt8#n?HG`#8f~HUKE55R<ts7j(fji|shKgw{q;5Ql{hX48<9
zIaqv)R0VjSEyL&qrJ=!ABxI}GtzK)0wR`M6p8)jiHr*4kq4ghiLg~Q+#(7<U(01LA
zEE2_~hWAh5kW5c#toR_&=sz0=48`~2L@Yfn7HkEwvKn3cN;Q=zIX)#Qf!G0bNK7k|
zjL8SLnYjmGLzc})5v}MS4>{$HZTcfTe{a?Xl)H9F-xRgWb+2(e8`!a6n6^xn{naT7
zonw;)tvGzfQl*eu1iF7N;;@-5ZTtgVhlt~DHpk@&EppWkO>C*!X%MrHAsYPN7o8M!
z4pY@!!n^>J&|ukL4auPL4hc0}|9Kb>A6c{HbP+$hD^m?W9)SztEU&4^87PZm?~xS%
zZ0~C1p1?m>pC`Fi-=JH{KUFS)5E49Yo^eu^{C2;)yf8jBVmikN6LFYaj9QqIlIETj
zY$z-S3}6!2s6Ejd%l`?$gxRJySwCzV%$zzxKg+GQ_PAFHwfj1351_ddy~5papu;4q
zx6{(K4@7H6Y7Mm%@32)0+sF1n2%lYUUp~M%J5x}$ESQbsxvTBy(?Ez&_i?h<#EoQ$
z#%>W$)dd!q<7Xako4a3bEwUu<bzVzS(Z^p`1L71{3QI0XKJp1MGSu29j-M%DblM(8
zLi-h@zX_{28fQiO&$<7kE#Qqq=H(T(2S-IWrTBYZX!XNY^E3ljyw|?QVjiD|ueiES
zm;$y6ak^&BRbjn;t@NMiw|@LkyRk)5V?Tmo!#629xk9Pg7oVceb_K_%L;XDnL@(Ot
z3nVTWiwzXR<NW-7=Z%W;ILOYVd<08{PD$cNfIW~mIyVRok4{HRRqW}<@O3@vh2uTz
z*kOGxq4B3n;8}~nIz~oh*G$9J)mV1*PYYc5+-wYmKt!lAG1s3_GWXQG>vDOskVM&f
z<Y0Ld1K^BNhUQZ_NzlGQAA!3+Zh1<m)tY}O3+py#dFDmN9Cy<8poE1wK)AZKi<tDd
z&YZnum4@|abh)0wW`KAhIz%9Ksqdriaa_eeU8@B#d$ctsR+w!x#co;upP++RiCzSx
z(Pw#i&1@qIj=^%V?zbw{%_p4~x3Kv}P1NRR7J_f#*CdC9{B{h2$4Mf5*Lb*O`smz+
z2vTNXCpA$iyw`W9U7AFqv0Iz~xlEU#6uXIMQ%r31vJJI^ZQ%!Nm8+Y?s+Fk2S#IgS
z%5?UO#F5k}(ltWI*i006U8+ZdWsmk{cevd?KaruzHJSNq8@lonh$QLfjE2U5tCZ>X
z^GgLz<(k-NP;z;nCOnjo8>bU{P+#wAay~-{S*CJ|)HKnYLImDk&uazWeD%90Y5fJN
z=|(E~VS_rdKavZF4i62ky2%_Zw}v(DHv!mQm4(LlYyD#tfV-q$_XhR5SmTimV(GT<
z*_%9mtn@daeZ;?L#lv*AV2@|9vwidRaL^#E&zIkwP{gFGoqp1hsVT=M0byA|UbH*>
zzAMM(-9yxoYUN5t%B=O!8CU?^c{<EmRXJWtg*}^b$%Va+4h}xHOn_2KV{s`@tC3FZ
zm2Rr~)$G)y+dtZc9()>f6&5NY(&h3;hXzde%RoFbJlQl=o#Q8TpBk`*M-}9%e@2d7
zqxYKt8r=)66A9$77#$qWiVTm<<W5aBeM6A?AlhN5x6U<NYU%Lq?SHhcc@F@fo^aW$
z22tbyF?CoUC&Y$@5;1Pk(ch=Q`2LHwwBK*vqYOi>%1~2biSDkb<?0Sg72AN@>ADKl
zMi>iaTjmjG`-fXduC!jf@Yd8pe`U?H&^J>b)Qfx5P7e2B(^~&|dL+fWXd1opySd|)
z3v4#ZZ^3-rROO=My&|We*abYgse9#!vs<r}i=%PkYNNFUx-6)Fy=S)U?Sj!^ulEcj
z>aO)wCaW|L=A*pZ3?p`4&t&DU%DV^OPY_D<b2{g5829@Mp!|_$hb_d`{wuKEB0D=o
z9a>uKUGkKx4+1O+?S-9UYbDO$svOhBH8%&6Ybq_cA#$9nRY%|Nl^4T%NUD(MjJbx6
zNjnuVdXYD_4wxxu={{YMc!P?mK>nz?TwPQ?CTlO!`rAqQP&`8X?B79>Bz6n>V{j$O
z;;DBn%@krCEG$~=GBa^)4I*p$#S?5cs~>u`i{;U&u>nQ-%h1`a18Q*tgVl_D|FliK
zYq^aFE4=dN37gXjE&Xhnny7T5vZ->EpwB~Oku-_cFpaVh>`XB_23NYj)bfn9j@vG3
zbaxWoGfg}JV%hiM-#nYo-~CQ^bJq5%>qh28k)6pkXU4O+;GIhq;M1sS01?pr^hubW
z&~jr&fh1nIrbhDqws~PYb39Y@;w@Dx?2SvWXt<H)x;NVMn9LrZ=QWxt2rkl4QX~y=
zbVucx%zTlmwW<VP;xcHXPx!gB|Lf4CK6Ru~aaVuMwYzgP89%MqrlMfw63B|qdx3!6
zva!M<je||v-BwBIk$n)u%pYrTMCd39Hgv2wV>xY0QiRd>>dAXPMrf5`DLdJZu^%+U
z$Ja*C%@#RwNaY$19!UiNg51t5azKegHZX3<*7K0lFw<$&^k@79PbuXs5)B~CCRKs1
z1m|?sRIaI;c6Z*noYTkC=t9{|beS^&A4WR{p1Wfh|0?elS`lt%8yOan4(y0BXZ?gp
zc9Z-)D&9h>uvZ2JJB>dh2#*G{)b~58@2@N!xCxPQ@mUs6u*-IIyk{>R>(Lx{@;S&{
zKhhKZt~)yz%%H&XSTL{%7EmpVn*j6qS;xy3<3kr)JmT)$M(WbAd^VN?+CS&wpG4{?
z5xbTX;7G!iHb~}?3pyr_*4o<6Ibr7V>OJtW;Cfj&X4H?A;;Y^YezI2><(_m8!%}F0
zB&NCD8vADEdXa9j$SL{0f6AKk%uJS4RU-@Ol1t-1X0FAbOr?^?)jRlm2-qvOXuC_@
z&D!`-30!;CG8Vb0dleQI?2gD`j={%XTrC0H<PaV*9qT3%FGHDQ^EvUzvS_5#!tAEj
zBfWmQIBWWW6%dnMV9&pVi(nDB4KB^y?LOf<S)eYN*vZAlPYH{57y(2#S6HeeK(8i)
ze1qVUhW`_p`E;(2$wu`_<WE=g{eS9C_I>b(ufRygdgh`{I#bETd%Bdr%IZ|@4)-L&
zD8+P}KCwjYBY<%!U+8FoaNA<PA@uw*1~CJzV@TEWo!mjrhUsu+7@?>p=Xb<}dA06!
zpn6%;t5t^rqEyee@?oz(fM_&A!@nY?5G#t|^;58q_tG^9y8p=Kt^d8W?4VV_kXc|h
z*e+GAzqkeD6S=ySe}cECKXI^G<*pYEq!r2m^%|G-%fRnojpOdiV0DDMf4!S)ndLh0
z^pCD_D};U88?vvxA6|f9Jv4E0>u#c^dq2AX{%z?lwz4Md`^zTOM@~*o^aC2HyP4uN
z@*Lt?#({4p7k76*MW8fcFYa2h9(jCe#{9}6M)gU3bkAisItj6&y4w=xK1w<~9?D+O
z#}Tob1Kj3kq04J2M+rd^I-`r3+s`rd#;D2C+<tIqPFWvwhI-uMR6Ty|JfV6dIHmmT
z=;+TkcyBMvo{!!~YOML<R+5P*gmmrWKG&_wp2a1?#JwSl@&6uaf0^b-I_voW(*`eN
zZF%4<xPr@Jg?48~lS@FKaPD+Zm_z0)zEcZ@2wT6M{ZxPJyPP>O$3vKcLj7E2tP!q<
zib3$~n&`79I=5jTF~k)iN$F@Y4)dxeP;$}fd;%DFhRcoESFnlfZK|e3clH4%%3SjE
zw{JP{IAAbtw!y@pTZ3+P?K_uOronzjlV!_83d1hx3L1pbluoJkIAzhnkbZ=T@IH1Z
zQx=SRppj#X{qT}8(EfB!&nU1Ij^#^9%g^a5Uw9<e%{U*3o%JAw&p|xXrazPztudjc
zgJv{cIbS_pceH5;cHI1TL8<+Pn-M#YOkz`Oqt~3@@zQ=7y5xzKD{B>|PKxZnC^Gg1
z%yHpxIFzNe;z1iFWl$1pGI?q$GHz@(MMtD3Nz~Aiw9wI5XkB6086iY;-4$LV>l*52
zZsDAxg|e2M&>o*S85RL;jBz0=*;&pp4Yu3dRPyU{O>SP)yM&FcflodZD=a8;va7NX
zF7t^cP>ADLm>)U#K*0WzRY>UhqvZ6lo{C&ZpXcD8rEfuZ^ZK@!7|s<Eus%MnNW7`X
z^IB2UzC$p;Jfp^V%HP_z+iRQSeH5`|Xqk=g{`Sl6%<l5v17(i?LDzn}`9maKcQOlV
zOV1wijbD2^iPsN}eN?HOXnm<IYy;c6oXM!OfbY82AinS1!{Xuk6t`E=v1HO*RbhL)
z3q7w(6v@IXXD*(j?fovZarUEJnSN#Z_i}l$wKhat!fg=3wJ-J`jzWHWG~X?C_iqL9
zd6IkMd7cws?Snyn*dD{iBqvX_2*P^ioMi{QC1!r=Wi@8;_2>Ybdn!MXgXr--?C)b`
z?P~QMyKh62<q8}-|D>n9WYyQ#CWfR9jmsH0L^4uj9k$oEKnS1Ar>9eMy(S6#Cf})#
zdbJglc((@Ke(}7iXW$o~LOT<(!6Y)igex|_*tLg=)NFznA<~jkVH6P=omDEOD9;kX
zfBhy>z1IoV^$1*BjFQ~Z=*WQA1t0g(*1QKeJoAso;bYx{1MaCeZ*T_pf}P}Ykdh{|
z-%s@Ym-KWCN`oK9jO4$%A|>$8EPmA?$arvS73ffY!B|a*-F@6&|2ciI$ax)Qc5f6N
zn*{m@C5DLLP&j$_!}^%d>&O6fGB-g<@qj+|f9HnuL-;&ST8KG1Z)b=?)&ZhacVhn&
zBsX+R->hYGP)LsUO^<eM&~ZW+$3unpZ$5Y|L8iBN)Lyy9W>xH^$|{|RYh&%Z%$v;e
zIlsXsCs*$Cc%q=MpJ>6}8}9?pq{ItXAgHzFL+7eOawmfc4AQRgeT|6wS?9`B8?h^M
zM0T-k2M*I5zK%Z>_h&~ErcYN=VYw8ViLx~Y6*deK5WPyaW_^3xIy&=#&$C!v|0@m-
zv+j-ZX(X~0!of^M-H<YL-f&8mk)81fZ0vrc{)4V0AEbI`-l%bm6X&ncki^e9vfxd&
z7uU(i$o4gyH;0`9c_^xE8PH&H6a4pY+;~#(w0Jk=WEESnq4XmJ0x+=+y9@L9sGg_$
zg{Pb-!y0TW$mbS+ySnv9$s`=w`Rn$1z&0@6TWyx{xzthhk;jB<C8)BP|AT@#W|7Te
zasxdA)6v<<I3Fx+5j$PGh~`zzJN5^M_2IQxcc0<z(2A1@ID+l}&IZ~=6Whm-6XjPu
z;6O<}-Si9ox!an<PO??SQYWmyVrzF{*+P|Ad_ufEaot3$xwEb6c>6jr>bhmsY$3bs
zx6=R$_2O&%^M3AG0efNva{XJ^Gx|x0sp?8XkCfEY>r=B2l2Tka$zP`0JP&SSUmgZX
zl(ma~H9QzO<qOg-jonUrSAD@@)W&1ANld=(ijPO0Uc|<}I6SMBr|F*COO!adXrM18
zZiw6MxCdrt(D>ZiWrNRjQAhNrdhhVm8H>k_G6O}Hsf9v@{DI24!Q{;W-c<8D+4mAT
zdp!SU0mSzdOdX#GmhWWS^E0~NVWWhW8wb?I(g-B;iH;f2O%Ra&tPY3Sb8<(`(v9wS
z9w?|163s}ad1yv8B3BVmTscj7K`Ta?c1X^p$86`9oUfneNKwpr9TU7Y*o|tCCzZ~(
zlTIj4SXoz3)-{)GP-eF#AK+oyjS_P=w`ISFGgF#E?zRP<HhO#T<=&5qNx%y2f-U89
z=3-mP)jks^s1i~gs5wR^oi$qA>&T94QmrBgM0aAh?lT3|=)tpp&s&<NFX@bX>F^H=
z&Ck-Yf+esyiv0Lp&yK{eMn7;lg`J3o9Mp#p7NR`xsDI>f2T>5X@FvF-BxvxUdv0iW
ze0tb8Jc&V`89u&E{9Lr-bz~&gM{*}8<=QJM_>Eg0AGLrJ$2*k|nNxIyc3j+kv9K4+
zTk~?RV`r^neQ0L-`r(I%T=BoY$m+46i2Hi4`Mp?)UrWv@xqECjhRW32Km)!*>9d)U
zo=#(_h5F&?E%|k|1JCf)8T_Rk4R;(Jb8(ry<MQ$y)lqYw`!(t7F0g>i>v5&^pYU)a
z?OKq{3D@DT-?5fX;1NhUstpZ;hSV}lC7Ufjc+;<xFW?%a(h-u&B<&&hDz>WXe%DOr
z(OEfJ_LoWVaN;EY{R@c))jP|{{RMA6y<Eya?9~23a>cEN(FK#1yX%d~<Y)_KMlFY!
z(Vg_P?^u}88S&OI^et%U#13q)Q0fYZK6(XW*VXY(kP-CmO*tLbM#VJ{g_4M-5AHkX
z8LFxjZ`QEjiiq?)HE!}3l_^?Sw~0d*Lf@|T&%i)sw0CqGhUp8ucMo;^E?SgsK3wJB
zbO{bAicdD(ff{TCb&bcy?Kj}|syfsDkm``$A6oC~{`KGe_ubk0i?_-T?jnIod;PSg
zdaPshZ|VibK76y~><JC8KG_;lNwc$aFYytROx&$x-Pg!w2O>_}x4$9Lg-SoncGkKT
zgq|EVJh)*RBEQf#Gimaj^Dsm<wbWLvJLxINPrM?Pd)qhA=oKPGSh@|5RmTpWB<6M^
z^?F-B3rQM-`(3&ofyqj!k+ac8f%T)!vG<*!UNPmFJwYp+Zj2l`j=!skDE{_9ru@+8
zQ^QbXfoU)rwH<2+wDz;*rw6!y{#3Mi)>x9GB=T+QTA2pa<<R*5k<yuo+`6`=`rbJ1
z3_tOls;Syn+`<L@V|i6X@1}06biov#41Vqpg~B$<oe}C%dam4tetyTfdul#(N`o)w
z(6&Q^40{*I)v;@@yal@!O`Ssh&e`W%(y7N`>@iN%GWphSE8m11Sfo|@T~!H)BH|*2
z<;2k2U8#8Qq)yT1Zr!%bvXeP5WgZuEyjOjFf0Fe*xI17xn7dOce0lcK-gcu4*?#JX
z*c5%hR#LGdB0^Q%NuxyNoLWW&4ml(tn5`l=nV5J4eL>AYANe{m&h_+^&1&AaCS3@=
zs$?nHzbm;ukr<SqsZ7fG7dfsvR8o?&w=G%Ijzut;%GX6WBKi1l4?TQqYCP3(tgoXF
zEmQE*C%M$oCFagdBGDPw(CH8K=pK5sD0Gb4x*fq^-m#PvF|jxIk48mLKAG2tUGd4%
zzF{k?Am^eAsT^S=tRAb`ah6q8^|g7^INt6NWU(!3>cRh3jHY#MIc?O~y0oluZt#JM
zL?DKprQ&D@a(o=4<6T7fAc{8oC?xkRU+RJr-2?}FTGpxG>4LL3(tHsoT3wwsy}aK$
zO#Wvm9P6!Er{>R)R9`0_^^-HJml2AtQbfo>zVP7D|Dxi>IU#T#N51ux^mdp_N}0|?
zYpIb{v#aL?)kWgCt5@Wn5a;TF)i>PYHzEcVrp!EXvye1(+IjW{!(1{*jBy}oo?eDR
z#W9qM-K?k85IGf-drca7=8^~eXB2ZOL^O*v6Ki^?%8ByyWRw8r!~{-ABz68t&Y4@*
z;ZctHQsv=MS;2}bZo3FRZDb_>6yVz^<D-XaUUv<Sa$=n?guOY)Iq(#OeAKD0x(hW7
zAsmYTJBj#C%v*(57M%MWotDn^(k&!r`3;MmolOa%siQae!OrBosY(sf#3@jJLVW6X
z?KMWnYrCWKQ~Q@j*x_P*-7Q~oT6>4Qo9btICrb$0o${Tfy6AUWCPq(pWi(%yBI1HO
z9$s8Ee$oq%06n*{v)zv>DXEszl~an>Oqro^<Zn#c;%oWCLplmN{ZhMe>{r5{MQ}SW
zF2>{I7ELUKtr%ZiZ(BC-3)A9kJod#`Eb$&6Y$qJ>2xY+x94C6!wlZS>+Rl!dmF@l_
z7)UM492`F7RMzc_T=3EE9Vwp>i>pXbn-OIF>AJEWJe<C&s(J7$9C~FF-ql__SQ3CQ
zp1e2yTT%W9)4G^x3>=K>Ehs*+DV}VhAt8|#9xhhkV<+C%jha&IFB&|FgNtX!w9NZS
zSGSddGF|q)wUcc|a55R6`lY_!SH-QW87yh`QO(e9%B*aB3H_Tj)!QBuSoy#Gi8qF#
ziH`IsU@nnQo*xbdcvcdVD41v_kHXLA2TknFCznb43QYM8hy;BpDCe%NI`kHuZKp_^
zSy|{>pHNbj(tq%db|OoWL6wPmPV~bkmIi#%Q76rN91PyX#NCPOy!7mh?`sV=-&SC!
za{2N~oF4nrE*=xrSfWH`^&Y1tRA;+ta1RoT{LZd!atay(XwA^)i_~n@RZ{e2RY}dv
z9d%BwL+y&^LB{KW4>^-HFI^f8D2Wg*HM6dvaSvZcC3V)4f6H&MP1HlEoAIbfU}{^;
z#bs$!v;t)OE;67y$0P=yV8PG|`nDb>;i<em(_fV_>A6TMUv8ph)(}8+Ry8q|uK4~(
zPY+dkQIT?<c6ajBWhAd1e?h_b@4ndKzk5)u80$K?RM@3aRq)^)by`|ncpg4>KRdyG
z&N|}+6hc#`|8uJB{L&xye4b1IlQsnM&MmG&#8p@{Vq=;0|Nb)@$M)cf8C~6)6Nht6
zl`S7blATzm{Xw>hgOB)c0lSsG|DD_UB$>jw%{@9Msp<O*fz@^2K$LSHE~eJ8c#+d`
z)%ciS;n6a(dapGK=5(O4PeGM+c08BG<z*Er#A?~~tB0e40%VREnDQlaF=q#+=5l~R
z*f=!)qlq7DeNpZOxH$oew4hys<=hsw&Urfhqj}4J?!w}WXRUeum%i?-cu%|h-V-Q|
z+L*Lfop=MsEM-Ls+X2q}VyjLv+WEwko`Jcfyd|ko_pY40A?`5$jkndP#r(h+RI#cF
z`Rn=4g~yU1yB4PYk3x?Az3GmCfo_zMigLY}<QMb_B%oA~mlpk!pfLfLIPYeQfjwrG
z-r(dOg+<o-CodCW#lw!&Hq_O*Dn|hrqtD2_`e-Tk<t(4|@7D-I4Qy<*EA|w9J{3)R
z`oD}cZ{M*7&Tyo0l45sgT!rqN?Yk(4l8Q(=vQG#LKieIU-ZJg$xI9~$C~P3-WRCsY
z^P%=w6FpX2_V2>F>YK(I1uC;(lletKXTgV4g`!MVE+<_Q(_%s!kO$6g^C)m;)Ts)C
zn?Zf^T0-r-M|*!tY|s!^StsWaLTDui<2GQ3MwX#{u2!m*TOt&o1|&&KBXiN~PFltO
zc%D+mR=LFB<1+#F&9?=cw=_WiT|go55~pSMhcRxxRzDe!WASw&2Q6Rdm99=r`8Gdx
zYrEM3LVVw{R{4??ixvj`Z}k@BJtqy?j&U}g`SK;Bzo+7dv~<&4=QqYn`>cx5%Z%QO
zYN$+9VnZv;KjdI|6|VX7r|!`cJF*`#e@OD->vQE;JTl3tZC2esdwQB>bG{mOl-EwC
z29I**SUH+@M2^_ca#?@3(0#>QZm`x{f{G10sT|9a&4E)(GaK$c$sRGCtmZxf1Td27
z7w)OTf#{d_du0558+l{<NtXfz^cN&{B#56-G66p_b_?B;OfyYMu_aA;jDjd!x1mxx
z>^MmCM>*!+Gr7Wo?WckBy*K_7?&JXs-OAzer((A~2l|MRrG@n`DtCs&b=G08^4Lol
zSqQOFQD=8Xe@0I}HonQfN$lIH`jyb7arL5<<6RtSx5|QnR$LK$o0I?_U$%en8z_<d
zn7qBAs8@=vyVV~!UHtUGO2Nc5r=<yGa{mZjWzcBCqn1{2h3A&A@_Ns=ehmtKMlcKy
z9F2LE$aB~gDo}deOgXzTL>gMsY9vLmyj*gw`M%xHFKBZ}&8(-f+=7XfeO2C+FuuT-
zxg^ypxI@m2&CT(e4J9JCIN&-OKIO7dLpEs4#ygwKwdb^x?a)#aJyk))K<)41FH>}U
zEezzA4jIOVMr@xaqs}2tEisl_E>{(8*QB=JVo9bJzZSSt)-Z&{s$G+~KDNWg_H)>D
z#xwlyF^w-5Ewd*a&jd$Ox3AqqdShkmZ87wTu564qzl(?0WIl@&C$&N*EDci+^$aGt
ze|pP%w)4W6Ee^)H)z#~1Pb(WoQ49#lRjXXj&aBuPm1;1_wknS!=aH+u2r4L9jcub*
z9ZyW4VlIVXeSMw24wVIR!HHE@y=eP}CW*1RP9v*@x~=`>=WB30gk*op7G2>Mam#S$
zPwfZvawcpE65^@wJl1c+)GSplOXAp37&=8(s3FqHfB#8FoBsQkq9z-^!PW8@J=-m!
z>IG@y^TztW#tA*W<E#eH_8d(LOAT`xB2eQyLYH^?b~<TjSM|hcPqY$ZNAeDr7cN~&
z<xnL0w3a`9U7-}EwWE6b(@OerI*Lc~g=wbEo^;ZW%vb_Eg`x9z`}V;dVs!845q$nA
z&Nj<G#8qmzrzQ`FvculwD)Khik>wKB+ku{h=S22TVdZxt{}*jDgP+V2<a3sT(U(wS
zs?Lf=T|1R7Rux|YI298gci5yq?9jf;MbsKNck?4mzO!Tj^}<Gn4Bk0od-8$)!oV7e
z`+~&ln_!dO+->9S{4c`(1Iz5a?be{TK7YY5GPSO9HlS@ltXOm0O;?4lB8$_>DqsJL
zYu?gQqifsD^=OoP5B&Jz>f%!L_55Q`aBfA)ba#7P?~>@Eo^Ice7Hs^r^C3fjnQMR&
z?Hezk7v;(cZR~Dn!yPf^;nF)X6UFp||IjhMAV>A7;X;}coz4;mNQmPvPI*C}oV8`d
z9hn+ib*~<f7qLPXu7jV}Yu9KS)%yF8%eq=|1e2w*b~@4ldU#6;iE%V=P~OQ^Ra2MM
zeDs?tZ7i5X1Xi($BmmzF7S2!5N<X50{ko=;y`;`|xR2ud5QtJ4!Bd>^zZBs%C=6MD
zm_252L0^%5Z#<T*yK*E?YqGHeq$U{WF9Al)+*3{hUtdYhh8@G;^+7E}VJ`jIXvP8C
zp|z2bGezMx|3@)ONZ?+^kg>D=oGv>QvttVBs}?FbL~u7t9<gt2OTkSSU%XZROF5;E
zw1AzCQOjVoN&mgq9@H6Eh-8J}URvA^e4Gs-^dKION0%Z}f+ckJ5*_V$X@d$zdbMbF
zh#8nWb7~pKy7Ow2ho|_~6n}49z`Hu`=Q)@nJSP$ztIupwTQC+o?^C7ye+3p&Ng!^W
z2?Nbdz;e>ND(tO<-}A<7LI)PB6-$H8GgMbpCbNSfQ)-soKOGI%gjoCHS17I{h1#}f
z%3y!}sY*w^5Eia&uy>h}De`kn<m!%G$Iq7vx^K0=A=lS)Q=@*5-6d42`n1X)I#8lF
zC3ru)o%KZV685=TknN;0i3HH`>qgNx0Gku)T@&+|izD0lZ%&+0YKJr35fL>NmuWIi
zwHEjK(o!CBi#9m^04WHtayuH@*Ees5U~PmR=-ki^5lh+Xu9L!nZu>hshz(a2__%wS
z(DT$F(?YPhnalCuoibHBz-quwZ!v>C50g{PKH?4OXdhncM-qpGWNNFixI5zXCj|RG
z-dgKz?;8piZo<Q_GQ`&1z~OjVtMu6rSAW!U)t~k4xN1zP?(EgL#)dXQt|Beh$qh$L
ze2n!whHsYL$~K(UhcJv^N{X#hpjof4CW84&BIW<RadWk4est5`+nSZhnj6Nba1*~7
zgj_9->LplRladt25U`w;UpjU2Ne_l>t@X8c^#u85dE*0dL{4pjdkTp0I^&zcANcVC
zd=NNWU6c23tht}U^Rr)%3M(1GE%izZf1PAGQ!<$c*<;I3ePN-MScOl}wn3+|<?IKE
zh)Qfh#HjW>!-}1dYY2`8yvNXq4i2mOy0%-LkfhD8e((Z?5n&cw7n`H)E@?jcqy4$#
zGVYtAlVut>^BjEc^G|WW1!U@8FQL>svV-dBEt0<AgX3uy{8Li$9XA(&h61*>IZ?_O
zCzzR@%^^~+<Z^R3o8{KeA-frf!H3}Oo%f14&Ms#U9(q9gfJ$Ngfhr_B-#o3fgzHsw
z<Jz|@lZJCE(z)t*Ai7P%KX9F>ZnT?<3JK9kg)mMao_MKgaTn=MTLZRsabW&b%2TZi
zOk*)XC1b;^U;cJd0`N2E&B$brL~|`vZ9$gVqg^`NY!=hp!!l3nr(t5%Duf)T`}5-!
zn8u46O#q^(-_$3Fa_oIQSuy+mCnG#2zN76o+o*O`pT36Epx$6Iuy8U`VSFJTrCzI8
zxfP${Bj8vnQ6HL-eIffnl9zXNA>Z!u*Tzzg9#K+ZCr5Oj96xiM`=PfwE5vqTKm*u)
zs_gq1h{5Z*hW-4$TL>Xmpwsq49qBV^WLyOMibeG9bh;Cn13VJM^BOKN$T(EWwHv<O
zr3tvWZCQ?LRrO_#umPs`R7%WXB{NTQs?5i04!miP#m`XtgzI)^jLm+`(}(@P>)_>?
zzj+DOZs6gg@k**lPY-RJ$4up7LVXkz#gzLLByns7=@ZE57epNC573-px>@={#ktQ8
zO}whYUj5wI_`1BTmuGZX!;nzrNaA&Xs1yYnnoTg#;b3vZb(srY3&MQ5>`=!9YaCtb
zge<GfojfX`=I?B2lX@tlX2k=q)cRR&vt4BaXCY|N@2ek@l4(F~)^C}*3rebTOss7d
zBpg>A9_;nClaV%<4a0VO<s1i$3w~bL@E4(`jC5&=?_*xg_jBC99^9KctmtCqoQ?_1
zGu6sJ=3LXtl5M027eO(K?I6<ywpA3gvM1C?==>o6^s;DIIjVLJP7bz{UWzOe*jRH(
z&kfDy#gd4aCD$%E><dNboiNoZEm5Hb+AUoYDrzobS1&#QU8ok~_@*$J$Oln%8oWrm
zWN0PR3YBo#xFWmI@34F>(q_$+d+6E@xY6~&YM#B>Y;9M4VOzTm9|Nf;^^-X6@iAlV
z3e$z<FF+t<5D|Vwn(P%yn%BJdpUF9C;c!E^k^s4T9#o`j99K27d_)x&D|HY!4xGH0
z3AyM0zmxyZqVIW-4Gjy1J}Rnu4JR@;xo_7p^F~;BtgY<W@zT`9J1+vZ^P2aMA6l7>
z!z!?FzC|;#W@}AV$7l!VAWbm>0zm6a4fD~0Uv4Toh8A?RoA8lFB-m;s%r1@4StLin
zZz7I=6<N>UB@?($#NZE4cYXr^v2X52iiJ?aR7kW{HF9Dm_%IFEuG(R_&XVL}!JbLr
zNg_D|kN-QRja*Ig%-29KCbgIds;IG_AML77oSxx~_|>{_l#27Z_OY<=_YUfL6ANFM
z+K%oEItdSw=}@j2?qa}~TLz0;SC+#S6u$@ha(OtGKRY=2Q)9#zWR>%Fx^pJjO6G|j
zSG6lqvszs2A!vfmCdW@5llWV9SwDTIq6!O-^aaK4Br)M0;iFC=;K<N7e&Un@^~VpP
z{1BgHcB_?S)F(g1#NJXjppXlpn)lGJPv@f-1`$a4Qy~jeHEz$+%KM3b^E;`?4_6VB
z+%!@E0WG8b`c7&IVWU58v&mM|kM=6^x1_owd&uZ0o3(@P$0_z_#$zJlEdSL%TN@I<
z{P3x$YD&6FKTjG}h7@eMQmX_GDSboV*l=Km)KO*_psZ_IE-G6Ztl863oz_!2TTxIw
zoHy0|w6puNQavh3#?Z20bm_w2L{IC3X?gRJ^s0!TDiKkHQigj=-f~Y*$;HSmX_dMv
z3edO$3VzK8e~}#G4tcOIUxT7;s&aGqGIQypz1r(*<(~uu6>Za9GhN{X4{q=6mr+hj
z5W@PBHv#)<VcB<1Izup*4hx<{{;tYFFAODX*;?ut5%XN}WGvIFi5ohboS`d^!j;^d
zi4d!U&F<y=tMOJ!0A*yCi(E+P*>7<SadFyUfKsX0C?n4kGyx<wS=R9{jtU;fP^3kX
z3=Tet3lxV_(Xjf&e5WAZF}HED&)@HKcZi8{KZL5L4VjTv&<apX@TnkImhoODM~VCd
zN6FUlvE#13H;wsHK`-GJ*`LXd(N*wMj&cl{vRyxVI%)Dg(aK=W6o@$N$z91co3yf#
zr@EPxG#0vg@<vN<qJs4e<qs~{AuRRuf7rX$Ow|L2aLzyjdJCb8Z#}jLUXD#^#qib)
z2R#heUaKZQ?cHLjo!oozh?CLCprgAn>4t_+{{o)R7C>&BwiEY#r~>0zl$QrXtyg;i
zfi!sGIqv9tetkyp*tY^9CtEFVDPP*&vo}c4Qj($b)eslu#8g&P0X|rC`<MB4!t<Aw
zQD3-*to2dMH+eCRyVanL*~kvG(~jv3*Ax305)zaj@8e}GekUu~6>eI5dHgc=cPnFj
zMKRa@XJ1=DqeGtUjDMqk@`9qVze4?nOH~8tSzY4j*_Kxc4S%}&J7&AYw`?R>Sr0KV
z0jsU9B9tfl$yN0xN09@&%E`90f|~O~ZOE>i3Bl7lG%Q+Tvvs*@IZdq~Y&rIjP_s6f
zyF_yfCN`#jXjzF6A_{rP{gpL^{-#fR{JVbuf%-627i~c;qx8OMu-=AT#D2GRxCKq+
zGH52}nwOl#vOyGVm<#&lhd_Rr7ZF*t=m4HM{_0w9kQXjced{jfPA-Tj{^jl?nE?{-
zd@G++vzhs!jS~?(ZltSuArH>|R`sL|0(#!Iq?F0ts^A=2Qkkt|^Z0;QWr8Md0&n0E
zM?+DXc}t4cLgr2lo0T%$Mn<bj+<y&fxagl;bi2;1NZAJdN17F?%w$cKj)i$^liixj
zg@3Zi;rKHkY%9V&y0_AI>tCt0(Am_0rP@cCsgt5OR1gM@Jn(I6tG$`hi{n3ilU%6Y
zl9_G1si855OeY0G#lYX$koDy^9<wFEj{OZiWikyE5DBntc(^826|&`RIMrc0)0>k@
zj#AZEW{h|*=hsjhcZuW>#GvsRILP`tY|2tE0bU7i4HcD$h;vP~A}4>fnbDiZrtI<f
zyy%nEQDaMfnTRXJ>ssv`*?}ArY!(bmE|TQzGtT|Hvuuf@KZwn&0<F+MqsFOH{Qc7p
zQF}9@%9aehBqZxgD`Fk41)Lf&cAH&FhKy<=EuzUf3)NVF@6ujbj8(q%LqppR4id>6
z{H3MXYR7Xnco~Z0=EMy=?AY`1$$^>S#NDJ_@YBnru+EAsAM|r~F2>zS8EHA@HdDye
zyL4uDwhW8(NsHI-|DC{f3FZBOE4|OrA^H6FF;MXC!p>raN7PiPOx0nGj0h_!f7b&1
z>5_NMfK_U7yt_DdHrbBQaI#~g3JF$ek$e9JEamzxW)+|~DnLCYbZwX|G!tF&?07CY
z_M4~I@1UrWB0@UmymLtyhD!)0QebHJY1Xx8v67Be3EE0lFyIs7TLa!(iY^8^+~`*I
zwGG1?uRq!+a;7w5HhCFA&)WO0ek|cJ>KYpV1(6MAVEux!WECj!x6=T*V&+Eb9M^D0
z_RszhCA$Zczv0Q&*0v`o<oV+@$#dTQ!xi7MwAUiylKZRa#vV93Y#YOw`=66_6$~4x
zx!%6BvN!e5(d|{iH1h(8A0=j>)PFRl0}X4wZO?Pj+Wted!mx4LT0r-lkocb5mbkJI
zkE+I!o|UMeI`^GsZS$SF@?M8Oj|FDnKg=92=<Mv`oJ|{U$*b)b|N6?}K8B_U@!hbW
z2XuGpYN$RkY;Qld(7L&7beg)Ie@W%GkG{dh1?FnsGWztH*}BgTecmSI;}<=ahDV7;
z8yA5+*DD;-n$^u@tWc!-Xj(#ICM6KNOBvsPhx+s#Y3x51RFE%bW?n^vSi}zHGbfE!
z6rzQ6bs1jxc7FhI5HOynK~nRc^zjjDsuV<#|GsahvheF;KpNWZx5QEL@J?rcyyIlP
z$M5Bm_?>g<eghLTJLSs^zb#WB08n-Oky%bp%FHvkdx}m85lHT=gDozsXWP~GcF62E
zFyhK;$~R}#)!2NUd?dntTu1+Z*&MLb=tN)W4xhgokBKSL*>PQE&?+eP%#`oUE=YXf
zY<Blb5veNdoCVxo+yqX{MsFj-^`KtH4|=WJF2oJaKS7Qv0cJ~k1hX&r1u`_3sf-BB
z`9+0cg4qFoBJ`inT(9^R<%jJ<4ptOo^imW{0NEAQKh|jDx%-Q58xUZkK8FnqV>ojQ
zk3(zcs9(%BaXlYN#u#TFqt=R|+rvNYBDD~c+EILcA)!lzMX-AMLL#bY&IO4#z@;9X
z*xnD%S1*hJ*}VDt6@{PE)w@S<C%%3o94g-odlLF!|N4h+WIG?fHHu?YRQ}UCN!f`|
z3JBzC-Uf8fd!-u03QjiD3m$8|&<6toNv(ao8Witj=kGbUQdSbXf+ibLD0tK!gd_gx
z!&z-@y8mVYs26n&IyZ&<T^92S^E5ko?50&wW~V)427W8Lj-LGk5=66`BL#xCcq#P=
z9H1@9fM^pCOZqOB-^}t`<9l(mhr;8r|NHjvr*H35MK(}i3k>FN54+Ln>z@j+gm_@t
z7ryr$>K}scjr$4sJx3A8=GCuuLQYTt=iVgQ{Tm1&9TYvN9jbRgY(_rcKh@j&T?|~z
zqk}oR3D?DpzVkV4&&MFi0&IGgtNCI=%9nP5f%iUsRa@Jm^UDG1l*qpGF;BHtprEeH
z_DAwXbeKwgndzJykBLn80}`}8n7Wwb!7T0w9qDp*-k}N(6a{umYAT*AG^0nJzS^tN
zDMx^+N7X?TjJ87=`9mMxCspcR4RY)6*}vtAG;-QTouBE9u(u4aSVu;K3S|Bwa6Upm
z%al#pQ1{i+s730R%(VJaRtq9!ZNa4tT_I~H`+wMZ{7vE=Wz&>ZVE1vOQl3*=E(jJ8
zCg!RDH)2RtLju!KCea(j&zdjc6f3L9^#l(r?tY-`bE<0`jb@-jOkowv%XmycGG4~R
zVFm%W`t{*Dbx4Xv4d?yp=gt41=CIE+XEC17TJJeE$~iSDn=jA$H|v20RZ#0-?WR(t
z^s1qGB2)P7><Nu>jt!zB-3eze?>{^q{AI&&4@~e}9{ZZ}$i?Yg=^J~-jqclB+>1OQ
zBxvmA<e`zz>ev^)0<HqcFr$*@li4Cmz&6?6`sIOAd$G0mJDavJ$JE5MD~_vF7DSOa
zU83-ucCx;Rx>c>`n_u1{qCkyR1*xmOZW)Jmf0556##7Hhe^!N<e(_cMTMcfy2jO@$
zu5<`7X5HUv@(%m<yVJ?;b7Y>Vo+T;Yt!7*7_YB&O0Te)P50=aMP-+D%#4}f(bp`K^
z`fi|JrVbh7*EzC!Mj}-Ux{`jpzyHSRMZM|@<33<SYz9Hy*O>@|(NKN8FGL@@*v&1N
zQbJo;SvKnQAkp%};EE}X48voRQcHVNA47G_jsF%9g*}7T{VcnFn0SB!3=hfKqC`&y
z?Kat*<wlicCAHk8Hn|ppv2>zAwF!G@eH`%PltR)+jnB`=X|yNZQG`%Yl^gCDo+Ho0
z=C#jVXI#@F6!1X_so+u3a$Qjd3VuE*>$6P3rfUI}Ur7oVp|79xCsNapV>1$&MFr#k
zki!|fPzHj=HJ9E?QspR$j7+=Z<~G7!g@2)Qw#2=B*0!D5{hom)H9Go93*jZm$iQZ}
z^~_A05#yYK%`J%b){KlIy*ZEVUP+fMi6?dK1jxV9mPxz7$1oT^_?TGiGgYy4g_k6p
zqqq->I?yG&SMV6ZloStGZ@|2*l045q?scO}OSX4SCoIh|G)_)yuHj>j7TgCwM6w(!
z)zwGN-6XNF&MQ?ma@S<;)G}F>a};e4{e7LQDqluNNQ4Tg9BhZU1DQivHB~cP{+^ue
zdKN71+=vbUQJ+We{a$1wU3S-AVMFtezQp(PInF?Y@XMbDkR}ELT@~z?jq}f>6P1a&
zdF<h8s$Zu3JhpGIu@i6dAcXar2;9f}GmW$3aWVTl8#<HM2Y5if1TJi)U_5{&wXpg3
z7PqQRKwWW#xg37ZTV2M%JzV-j&)sokYhyUJMy9Lxwe|L|aU7$1yLW^UqM$?y<U`j=
z*Y=^{T+l*a-LE>0n`cIY%{8BtqOtDdsjN@~`!R;khK7SK8{*m%cE9LyHgI<3JjBq!
zuiyXMDj9m~bnACd!#b%m+)0kcny~<Jan(vkK0qKrj!Xo(wM}Qbv)IaZ_vh>KD%b`g
z;!FHc1e#aX47aX@rc2#y7Z}ZbiWP-lqe?wA7UN%`#X*Q&4g0LkFTLY|Gi{2mwixky
zsUolye?Yzaj!q`AKDaC)FqkzS%A^e%$4uulvvad+4GBNs#hLF9TUTW#;c1Gotd<bu
z<uiXu?^a6yG2-W7VQ~RQbCBb9dS)rPWfU&Z^FJFH*ik}m4CBkOh)T}7Ws36C&~nA$
zh7$AW_jZWcgl6H<gY`A7WDpR2r1gTFo9nSmR+kyRMtr=TMdv#zk7F8;AT~OvlEqEN
zYy-(EC=XG7aIjcc_nMnLMpav;m^?^sdN^;m{5jjxNOyt9Uy&x?l*Uzg7U)2~vH_e}
z%nHfON#~O`G|a%mwWfq5ww6%j0C_$r)#u{Y*C=oMYGj3uA|q=UwEEBrbLkf{PDwMx
zp@8!9R=ZGBp`n35RcqV7xv47ca`7<`nyTQbSgR`_c`DJrTrrPn&P)22mt+MSr5a+@
zcVf^mY_vM185q1YgDM*9F#o<ck$AtTL@?E>a$B>HIxB^NE_z{M`EP-YKw6kT3{e<g
z^%I@WsG%h~Z|v{@dbfWxVR!9zN{nTme^88Q0(S8CX<iV&`=z4!@Z*#SmfG>AMA`8{
z!6~QHRI1%tXHT#GFBw#~9rNSewVY9URR9L#LmnTlez(C1F<MzE7T|ouqk!EGu>iOo
zJDz<J7%$^QZzL^0HK8iBsQxPuv<SbhU9A5>a}`vZ?1Xf1av+%BX=D7f#--dzJ+zW6
z{bK(#cZQQcb0<{uDO}kt+E7YN{e}Yy*X{_7uXr8#TQRlv8#ujzzLsbo$tp8gCwvl)
zP)XDO(%bbXP?Qf8=Uhu!^v9WN$7TaAA=*z3<oA-emm<X}^NboKRNZ)G7_URHqYGXp
zMoQT$pIW#BONb@a$f@4V5X}#AzI>x>B}LeOl<s%b>?BYEY}=tyJ5I+ADl27>xid$4
zL$zJd@#u-VP3$VSXCvn%1b8$YrAS+LiNZecc>YXW2ulc%sfA0C!+88IT#nqPlq)sV
zqHZ~!@Blhxvmo*aq}|3<*R_Rr46XVz1w0IvikEt{52?WbDuJKryF3!l45UQA-4kb{
z!&XA-HmjYbUi78&IWLdvN)t??kjJ**Lte$Iap?C*N#ajrl7t@b^sX0Z$ttGfp48bN
zxu4|B%=wNoNOQ)-+AJR8fL%;GV!#hF?`@Q{m*=iy*=#L|AhNu^v(og(YynsZ!AbNQ
zBc`+e(*5uJ_u57kFxDg0gVU>GwUb5Wx~9tQBcZMZr|R*=IWz_~qX@!|(boIPUX)h=
zCMQxlsE*&`8BkD!KFPgWYbLLJgX8O)LvuT;&1UOVSW>^4sHY$v;XTWa;7QiF`XM2a
z=ST$Q;5c<VWkw16DV9)U;{y)pAar#e=o+i(I(&^ZFfnzt6LwU0eBPw1doPy(<&cid
zYI@sDxs>1^e#9}ayZN1`LALkM@*PNF1Rzac+uDl#?QQEBc*-GRN8CeeS)G}3Tou^W
zMO9!<UbYEt6)Sm-y$D<xn~<Q*_s17STAW@Y=fUO<jlJ*Zi3t`Ku{`dyc3Co%eANEu
zBWxu{ht|av=}A65eNyYY72_j^YZ`aR8rit*({cnAgg?3UI0%Do#HwDj96_`;$`5Wt
z-jDw|dAlJ`&Y+iz0(6Bb_N6pO++Wxtt(cVM)7)80YrUIGIMdZb4GfJxyMB60NsA)_
zNvO#wW(JybnovZX#i3(XZ8OM71)8b2N_7>=Nx4*UF0CwTZn-G7&31+JKY*`e59BJ}
zOxHR_PV)e*QO1mpH?Md?_2Q(TG5-RgKpR=MZz&7EtVS3Znfy%5T+t%SDi~QgvQTzR
zKEj~0v{c~_QPZjwU0GVcApMu(yhKm}=-aOr9}jeuFFn0NsIFu4k1a>_5Nw5&L$4c=
zv(LSNWWbw$EMuPoAq+n}v@R<zu(7RRsEUdt|1wY_ry@nUx*Bz=<`{gtrD>^MGp&B6
zbUV$ZQp)>1n^dLd@C?C_v{yTIIl?Y2D7};Hd+2na>k>+vE}w^9>m0k2C2a$t3IW{p
zLvIg2pDCi4o@h?S{&h-7fu#R|2QD!{yqlXxekWYJ3BHa%x3(FoNd~Dzl-dMDA%(@_
zMrTxG1bBobr7k{jb}2#XOWJ4npRg?dS@)600!~hEFdkDcy_Pj&dv7Ku1Yu%-;MEQp
z@7bEm|CvR7AAELBi)@&^`HF{Q^Og(8%+3zpDU{mKZm>7jS^0<xGVF<ffs`JwPppee
zk?QQrcO7E+*7<*=TO)6=OqXpu@7SjqfuF4O<T<m^hHdlW5~BUO-klnx2JV~Jo@A0&
zDg2u0-G5~ShaAfQehfLGl}rp^UKBWy5$af|U76fm8V3^$rSeU`&MG`~c@C5U<(rWx
zpz9{avA<>J2wf_kjG#e=B}Dgwyvp_m`i<#)6Oz9DCP<1F5l|z`wL|}D$oRbH0e%2s
zA3fQta-bHBEIezQf4+k<?o1y9(olOI?3!3Y@Ij_B{U`8~l6Ys=`JGL{FY+v4Y{#T)
zwx)!@KsR3Al@>a41%YI%$i(FSN*6NblVQfEfuUKYHAKm08-4XO`c+K&gx^A?_1gY6
zNgi1Rq&}>lD*CA`!{=$o%J={K|9!o`_jOPN642jrF+vOZ9eqV@Sr{GwGD%9r_R&l-
zdR6Mc!EQ-ak{Cvwc0x!E#+mHXr~B2oF=Lhw5JLt&?i4bVqmV~wgQcM|A&F_gJP)g{
zBa5TR1aZX{9w!Pww3GXihVw#!bsHL*_(;d~uHJ!*?}0NX_fUlcO7R5Ohv(?Xyimc&
z0GZHaG};DH3}5dJgT28gqSEUeYC0ZjQzkDKN`FN_Kkkjq=>e1pJk)V^n>SZv1RWAP
z5ARb_pw>UsOu(NS4;2==wAsiJjoqC4z1$;tS$Far1Zt}0vi|w++};uV$0Y8LqTfHW
z{Ag*z%v!$lDKkecmxYESbj@iOb^ao(g}ESup&<3!12SHA5-&i}C^P*!m%-&zSL{&T
z=Mns{;AL-G>yC}@>e=iIJ3#WygBKfkp0Iam=UnT=_nr)^%c54yf{)ql-5=ncozMS8
z7Yq+KoAx+OUDA}2pl%%V|F$zZ>q`(2O6vK>>+CtnlDMgB@Kpaa><7TfV1J1MAvUtT
zj>Yf|NG0e0zw*8_D5`A>vlYQBh!_9`1w@jPGl(RK0+MqE$w6|?dM}_TpiPpjl0$>2
zfMgH>Ns>XyK|q?&<P1IQ=zVYI=TuGA)KpDZSs>kg&OUpu@U5`-SM3oukWf6}lTrt5
ztkCoOY52!l@2?92PI%|+9$TcEekwe)4{9G7eazi0?V2`Mq0T``y3nE<=&xq<_QAS*
zx+@d&9Bt=+oc-|#;!KhwH^y7OcfWwm8lp9UT%oMmrNd40KY86cnIF7&U-fXC{=h=l
zM<r1!FVyUXQ`FKCWEgVH-)>e6Wu@RN!<p|y-c0iG<M!TX7_j#)r=p^MzZ*HkEy({f
zGBA8^<8zjMxfnS`JHu7)*AU38gxWYeiEh7xvqEJvdR{;R;+KG-a6+sG#*2D#qrNX7
zaGjm_dht6o+(4gHxmP4Lf?q5tD0v=qGDFLPc0npcv@27Rkcz7B8BW5P2B_Fn$0brd
zGx4VaB&|8J{cd(Y(!R?TGIMD^YA?gzoQvz8hvwOpjvQY`>Tp}~(Nk6yok={>B|-(`
zdu$=|BW2U2zju7h%+du_R9&L@<7}UV|AW-Bz_@o*`hMV1W7Q9ux_ESK(BAs#AH|sg
zYEsCTDb8o%&&R%c9tWg3!G}ay8aoz+u+kP#V{`0%1(iP!lQOh+Lebm768l{lqmzL;
zg;Th0HdwimAgH<P>HP{U^2BKNF|+$ETcMC*K5VnNSj^y>3MKbLhDT@hOCNQ7mKskK
zoqEI0vwGY+i}G;uW`I5=6E>ZV&b;fyimZ57dUmmxr;Fw5$8Tm{QB!w!k4Hh@L7}+M
zu8H!&n+<4-fz~NDXja<PtA(^-TE@K=GD*k-UMaQaOQ)3x%}TuGFxQbkXn$ayDzduw
zoCVtS%nvI$sw;5Zc6)?z%{g=P8~fwZ&<qj@3QE}3VWxKA<(E2B7Y9Z2Bc|Y!@7|fJ
zUhC+vA8i?3NnI@~HShXy{ak95+K}eIKKDs@lvGCERQVd&?MOLkU`FxX=oCuiHvfE-
z7B#L_RW^>hHR`dxuDxHgcg{C76y;Gk-*#TYnR@sK#GO(aPT~zu6&!16RrLIrQQ}TT
zDTl{b9XSfpJs`z4o#<R^(KLR9+Wql^xB3kkCM6a2CNN~Yeo<hz01b<u5^bG^a_JrB
ztHYt6oF?6K?7O?*lUBMwIjDeThlQP8EmEc9QWeEuxOvqsCZx8%`K4DDK6bt4^?4}5
z)%s7}PP-<$8l-xZ_Do4R<2ZH!Afwyfg8bh47pgfrPK?Epqxm-%2|AIgD(C>Z3o#qa
z^it)8*5j{N6#MO=_OCDIck@hz?{odCM3~@r&}^}D@bUPAmc0OuYcHltr--N?nvE}z
z210g;0#w%R*|Un5FPpxt-h;v|D9x&?yVY3A4%jNSj1nC>k&a8IoQ%cq&MR3k5FPNp
zhA17%#<u+DlT#ZX<z6WZmwjPq1QLJi*@RCVRGkNpo=QwUE)MTm*-HSIJiVI{llv*u
zc<b=O#WqN|BYmzxetV%JsTRhXd0OQ0NBEdGFX@}*ns<EVrh{13^KP=Tf9V)K8#?YB
zvdi_A>!ze;8GUX=#a>`q<fn=C^$i|g-p{${_pZ}*4QBOfsVy7p>tEa33F_P0h%a4=
z@#n$V?*ZGFu<tsf!%$JpgysdQQ6}^xhF)QHoCeCRq6eKjhc)=?Od>j1sGRcH43nGC
z+%dqFY!eY~j89voRJxV+Ug5H_u@>Wf3XLb?tI032tgSZo+6JJQzu3vRmkOShc^lg>
zx}dq{>R!{r7r@Q_qcohQ&fsew#&A^`Z-O$qr(*SOYR0iDD5DomQpU_&J^Ds~tBe%s
zvIUv;o8XkfsJ*BR($3mSmmL%KUlze}=r==7oh{P!UD<=VUDs-EH(_h$eWnT6=CS9}
zG{HVr8V)@w7#^sZ96C!%j3<jw!M4(x=DqTRpE&IfllRYWkKI{Qncw!?iAhONJ;DVy
zj`h5JLZO6p4D$XPObZ6AZyS;%Q@mJBb{aN@(bHP4<0Ypbnb|vbObeM<^eh&ycGVIb
z22Nggy_E-tT39$+g+?c{b)BVC1m|ThrZX{kV)G%5Bm_>dt(0Dr{?Tr%))EGT?2eci
z#Mmy+Ka*O+?S>}<KuPcPNvY_|c`w2Ek7h#g#RR01DJyo`|L`im8FaG>7%U%}l2cwj
zEqQeakN22>e$`ToFa*d%t^QS%nGGEwAeHoxV)F1TmWPcXSH?7mzCZ|U9kmTX6{9%E
z@}uuQxVl1RX^FBx{`B3C{AKB)ou%LPAkAIE&)3x1kVeaKh~Obi@Dr%APUa+-PHyC+
znSrf%ToigC#;Ec_dd>$v*dDbolr0cHP$+a@s(nmb>gAfREdi-UA3ehhsc0#v$H0tx
z^85$E_}u)x-vH7G+1rba{Xu~%`+BubPNG5&u2A#;oS(IDe6U_$JUGzQ@#Pgg%`ji)
z-^T^m<90NkV!OYvoxatKrBgGcO$JulTSu3y)6tVPLa^V_H=sMDN<0cYOGcHKIV5L<
zbRUY}E8tow|JlJVdEl?5`EAZx#>OfpCGlx^h6U!}(tn&r9HWnH>Fi}nSNa=|VKYgD
zVgwX(b4qefgA3pBx0ljttz@T<)7H~ktbHF5;k>B%BwTinFWs00r*20_)=L!sZ>YEC
z%y`B`Jqz}f>A$lUtLjQ&s{6Ga+|Nc%EFx&v0S(POrOvqEEn>a3iMbW6N~u)?13%1u
z4+=mcDtCU>%&C^;%h>dk)_#YnvQkb$LiqR*d2H^JZ*2|3qu1E@K2#>9{hO*4jg8d~
ze-f?@Z37nbf%l6404#E90fNA4q>*Lgp&?Ip2^}Zp+$odFju=yp%q>2y^8sQ@L8^v-
zVfIgbJZa|dBZJw8j2<&ROTsy2YKhG)y{5(-?}AF-=xup9O*IWETu8hpH`~9c)8|tw
zvB?``FJD-vU;3i!8m#SF&`{KU+u^@78%K11<qJ?pHGkomce&p3_U*zk_9}buyt16v
zf=9YF=RaP@q}8i{M;!x1<t?UvVbxWOgP<I8svc@Fi*9~Whis4vI0@~<USaZXkm-Te
zFTbq~$mF5qJeM^R`u@9A=)y?~3aDmqKiD=ZYH8x@?@9*zYVP+en*v>h?7M@~<5DVb
zyswyurkc^0*l3BFpR;T`=5tb7DsR+QS7M){s!MR(giQI=JFd|G6)aUeY4)&*4>sj&
zk%#(BCSIZ9asMpbPfc2k6Z=OoBhQs?u{f+mf|nYiYvwJi>;G&SCc6x=6S2weihNP?
z7vAsXCko`qnufoBY7_k3Ho9Br?M)0-2?;#1Cwt}CGBo#5-qC(Hl)Jw%Gp;Rts}WFC
z3AM50b#3(GuMdwFW4&7VG<;AgL%zGah87?x?25lBps+^7#LQ<g;)JRV*jIRdc2@ev
zzYCJf0k<DXsql-0%_LLM$ci>6;8DA2-+$cXb>gh^FKI`6R;UC(y64tEL11kMt0hRl
z*d3p%0sNM%(Ar-7ehd=il}8m3KdVL44(qm8L`C!0&G5T{*63En>x?owC6d>_a(;cw
zFR89&S)fgML1t>cpp0|1%-(po+<@bgDMkNh5(*i|%ymWmQ=u1SA*fW2<#-|~b@F{E
zV~iEAdr+`h_7`@^vY6HYiJ`bfYMZo;4l$M)Zuj{Pcc&`ZKBuazbG;|GG;P&h#2<CJ
zNBl)PkbLWo$3$Cc8z+0d$5qYV-oVqIedy>o!=Jq$F10xP3<>hQ^qb#~TA0`*Rzsuh
zIH-Qh{+h=EgLiP-i_Q7ntCshYJ_Lk{YVYoIswEr9%G({W1*_)_Vzz!vF+?O+sRbYM
zDpaYav3fGT?b9xdvkMp4PsjH@iH={qAM_|l%>JM8{Z47#qE#|FN$I{ltf0Zpx{^I`
zCIkogOQz0q_IG5G(+Inl^?%xsmEHFVP734P(~CJSmGk9HI2QJOdn|1CR<`x`DH~H$
zRWDary{bz693oCS3M&7)s5|Q`A}gD3T0l9Bewo)#aN094CraSe#oXE`jG^6zscD}(
zuBN!6KzyK(V}#$KT`-i0=vV$b?%B(i7sEw!yw8M}@$mTBVaX#cYSii%E5Dx`<~<-i
zIu6DcH7yGL3Kn$g!8?w=l{**y;ImY$=FeU-Ulkmkiiqi$T(Ky(cDcz!OYbOhItvg$
ztbSfRxWwYg%pxCXC|KOvQ+4_3a9L!TrKSB<LN07xv$^^HVliF+jQ8^%bTEtkY`o>N
zM0`=>{Cle_VMzi`g^!G@Z2L)%JaiAcD9EqadRo37x1*)XtHV2m1qp4d1b(7dFF<R?
z#=9DYf3t44zG6>BleOzK#=jqSOzl2{BK+Q-R`{FzO-+ovdbLe7&;ae>DU)H}H6U`s
z2pzi>Q@nU`OSy`&b8G3#nxEIS4@E_Uhd7IBR6Qy*G^|a9iqI(DVZ;P;Is5(~i6(P9
zzw#^vIV<B=`l`sd*T^%zE>mCq+`hbaua_cS4<qW;xIL4yV&6sBpVd8iy?_BF-pPG7
zH2jny&F|As4~Yo5stsfW;uy2$f9m)bJs%};=XW0dIPAXnenTZkhmpx)HXgG-_w_=m
zu)j`Jh-B##`}`BBGyGV1p571Pjwv7i<!aNGY#=F47jQ0GePNbmof->L)?J@3AyzAz
zM6ZDDS+NnI=*l6l8cV7P2t8gmf2@+;wPr$R`yuc9jQU(6-*6DOviotVhS+d<UppQb
zM7liC_xTI*KyZMsEKHT#sZ+n=#cQ_Knw1TLCSNdnENEr<_PV>3uIYs(3EF3#81gEb
zirtz>Ou44);~GDR`+iHh*VwhBvcO$_?0aQsQrJIVm)CBJvFI+IqZ@cUHeq}42ItYP
z*@Y$B8O7M(66<@a|J?4AW6IH~EK%VHpBNYRlv`OnbRQm#&esjUeGo0L>3nt0+va{(
zCs(&J0P(FmUfAJZ*@|hrcX82`+8GZ<Zm}}Tgk_?Vm52DFx<95Fm)flh9|eBl4kaYQ
z?lmcGei_}<kvR3|M!V-hnR(%Kys^=$gmdZbgU*9aFMLBs@G~O>2Af+ZvSHziPhGtS
z6r1-QhZmjgcf7+rEf(VytLHgAqn{ReRMXpiE=`T~=}L__d#*RRUA$*{3tyD{?$6a%
zeaWkD&IH?t7R|kU=@;B(BDVj7s6sTYeb}j0=wP!xXVLOX7^T>rA>3xjt#{w*C@OV-
zKEX84Q-&dMvwbSUMVQln#0d_ZYWOw6s8Dw>5&)+V4(a?p=Lm)3zYdqr(DsOpm%yKv
zm6x#{|0Wg@jAwW#>{j?W!?*5E0>)prZ~#vA(k*Gp_z^8yII_Q;*VnPtkUnmDL&49r
zn>2#>OvUtaa8Frev*$`K-TMz)-y*aHXvqGIpY%M7zI273_w#}HAI`R;={Bre^>@eO
zSDy|GMs+I>8DR>p3TeC;E9w{(eW)l=^e~ZyQK6GsXz(-3v0>xBk&WwqWr8dAqhRm`
zW>|(F5e4nE`MQHGozLS0$a5|{E%6>SllpTre(pjVFcGwJ3Nm-KWXP*?y9IQOvoh;y
zi=r>q^*i8=vT5wX`GbejcxMdc<sGoAhMT)I7X7)z@<oED@Z%LqX#VOWf%jB$aq|n5
zmoS1CB0nCPC4r2qYsq9JMf^FwK7{DW<MN16r*b3CgaMxX4v#7>|0Ux!0#DsT$t=8!
zqTvpm+F&9dt4Bru4OYd}VyBE1i7L6~=&bAm;yDbLVtg967_GLz(aS;-c^=cNG};*U
zu*`4j%AQLBQBi5aZ<x*D1Qs&gW;pYS{$jM9qWCWC>EY=l={jv10Q84zukgOc++xgK
zrzLETeN0Z4k7VTIZ(k;T<lEywP)zcIv2SzL$aYdRVfQe24OdsYMMxW+)4xa=5*7}_
zUI$|8)YONb&sj-W*{i1?Cdu;mE)VT+x(5PJv9ip2BhTX#$Mgp&jCl$!YNhV{q!rxq
z@G#i^e5=xV>48twUpJ@+boD}ql=$yEl_!(oUNBOuRfarU#hqfz`>tL1fC5ZXY3?eq
zIQ~xRH-r0bvk?&yKPx(7atKzH$*Z-=sj2gUF;<TzqfNA-K}SAU;=X2RXeja7v$35{
z#l7S_FJGcF`Su70qTJhj&7&xLdWffYv~nVhIh}nk{c`zgJtG=>T$uTj%tM%6<s_T0
zC@s#t!Ks|>I(PT9%scda3&VRL%FMK9Z#^}ktEYGK$cO(LZV0wk=Z2uFmb}DGge>V{
zW=wx`+>oTtN=Zx4=E~#fY0iZ&X&)An)C;Y0sZsoJ6eK*gyRzc0Y2=$I!P#r}-SNGy
zYu#a?ue(BT8l8$R-?PomyguHu(hrLx^4w^(B1f!qEmO}Wo4>;!8LdUnl6DbDc=rrZ
zkzOq(IWGFSh!Lz(CVIxdE6~VY-F4dJ@^ytU^{uXptEAgX0mpT_ljq@d#Nz#v=ZsEJ
ziyw2Hwy*Ct5d&8Ru{XQ7z@VUFr|26{5aWrwj2n%O4&3yA#GU#EvB_&9BZ?m~e^+{I
zTEpS9-#-+$ZB)N!(nj-pb$2<i{yy*(g`(z$4~8Aut&5>E=W9m`J4UP7#)GVe%{e4|
z?r~%!VIPe+7cBVvxvFi*O`k{9ZWcohp160(9yhx3zz9^oy;Ss2EG_F6+mIr9JYctu
zoSIzAkOzSZ;E+SqR!<7peq$gj2%Z-Px6x1f>|$)}OZ|miFt))XYo$)dE{Sf>jQpIY
zPmS@cHjDg+g49$uSY`1qepJJ`aPgydlBD%@2ZLJQXJG-Ws-y?{hLOwH3a%QIYdu@#
zrl=efNlMM=f+Np97uO9u;<K^hJ6p4F+PC?gOWyG4U%seGX=Ln!HyLG=@T<=$18Yr!
z4L<6n(}=%?j<r0O!>%gNfkL~TJZGaEk^5(U4?ZL;<niTPLczxFSaTt}!$}{^NUbJY
z8ROyzgS_YTl0G3JQNC|qHg)%Hl?<m}#p5M3U~mUs{(17rzu|rc<V!BLGWc5Rp1R@z
z_1_mob)(c*_ir7%8^J`WIs1sxQl#Bxkk62Nf0|IA<}{_-{9OLK*Fe%S`l)2;(IZNh
z$8ix|=?RRXax!`;r*zgo8L@VJi)W;2Xm}qUZkY0y4l5_SmbpsQ=On?n&gP1Uqr}A}
z{faRs1&xefw$J+N?PYUR*{<_zZNE5Qkby!SlH>0G86T=%Uj4u1tkG}bOzb}|I@yN*
zkh6+MJpK=VT>$xiDd2W*0)NS$*Qz@Czb+LunQ`;KtgrvyU-)F<xygdseJoq*ZTDvE
zmE05AcB~Q-rg!h&1>!KBmPesr`Lv9T*cFJ+or~{mo~S?6H#dL7c<biPuhWe0&@D6S
zcke!8M2JGqg*PI`3%&&(kdl%@1kBa!4UlkweN)h|wasm^Gb=qj3?L*TV!a*`J#^eL
z>()B!rl3)C;8<bPMGJs1*Xoc&i}2;km%~-w#oH@m4xZ%w=ll5gUK7DJ60>h9Y+*^M
z3f0$*CQBRY60;XLT+cR6l1WU85iqW$oydZJ8Vf^&$(=g5G&10ZFA+HzVqp1-saO&<
zIe)yFBVpGRL4nRBB%#dp-&vt#IcZk9JObwrpQe4m7)iw^*Cl>X*z^?De;Np~bmZj1
z;P72^ObjN9;5F?liQw9C-+3CK;^~q=$mT(yLnNlS$^TE~tBqT^@1w{mxaq>ef{6ci
zA!A<P!8S1|X{qo9Q~_{vFI3^+gnfaB!&(@F()Yp_LiWw@$49KJtZ4Hq*x}NG0asK%
z0gGx{|I<r=IxvRiI+m81@D#h3{ukkA@A-F}sQ(R3IiUlm!WKA=eo`;4JNC=6u>O{k
zr;;i%pne0j)!7=ug3yQoHQO(sN3$O+OmwPVlN$@z=!|&!v_yn9L}b#fg5PWDdpos%
z3yrsJZ>B<N%3qR!=1JKa8XB8>^XZHAj|ke@+codsf9|)lf`Pd#kbOp}BY$d=1r&3a
zqmO^qtBUJ3Q3#$#S(7gKu!NGsKnuGM<nK(3j5q1&@3zNtXiB1-vrxe<c{?ObKy#<p
zu7K*l%$cC1D!!N5+uN&XoSv3e1mEB=tu-PO-+diMM(3n1g*yI6l?in?E=?wuIpS2+
ziM@itzRXfeuwR>QBL3^I?wM$bqQ&AmlZJ){R!$&;MHr)I6$=<n<}y5b^k{#m?pS+j
z8JX&jc6}h;ep^`W@K%)GLL-@kmVrSMxZ$nMKGg#E`ESV8_BQ5`ufw2!0e|&?=a6yE
zTxYT&&1Fe~7V2u$VOnRi5`73>*b^H|$z=izc)-v`K}T2o;>CHYus8MflBEwvbn1?`
zif{W}4q?2-!=sM&Sj3dKO}Xe@C6Y|lukkBov}=A&g&nV{I!QnPo|an3sSu<fUnzkT
zof#WT&Sr6P_0mVgeO}fFdcsT+2RpYfq#<puvh%Fk;}tGW5S=`HVwdD5l8VmUULHxf
zlWkfTgw7lt9TjOOtPlQebPvDsD`QOSOwb-sdoWBPk_;>l#s}R_RgsmI&3iafrqgml
zP5x}7TDmx#z8y|R$LW3F-ael(50ciN@gyPm>8eneafH@yAt#+xOw3q0iC;Z|)7Ya=
zSqS+M@7YGI4+^X1<y8iz24Hit(9k1-(;qz+G?s@;mCVh}u{{~#a#7dxr+P$ChwbaL
z?Mw>Z<Gw67DwG7DCsr#fI~yA*G>$MK=6wII2UJ@yxWN5|9JptKh{vx<<NOmS`xtY#
z2p?1sww<f4zo59*U}|}+(zA~azbil3M{xKRnCFY-=>oNSdU~iOf~u{%(u`L=-C6l0
zt(kp4Z^Q+KrO4^a19?zFi|<Yn2-x?3ot8mihktyqDK%?28MxMTIje|_grpA^8WCtX
z)!Z=l9N34yOi)2={QO#Vd!5exIjVF5cD*uT=3f^)Seck)KiG5)4w(jHhD`jG>+nZ7
zOce3!C~`_lMfA6n+Zk|ehmKb~m5W7H<g~O!pFc|>w?U{L<2ekofs-&-x7dV|g~%)p
z$P4eGK^SE#NOto;=mUW--KwYyR1AdD&E0+AldA;@`oyXbxWGf5)=K@$Brw#;hQ9~i
zNS_uvTy0cTRMgPXp=2>Qc_`DWE=U_Kf3&~VceFEstKk6h$?{P*gmZdx+#hGx@R-mK
zgs?Dzxc<SJGiT78*Luk6@rA}!I_`5F>Y&mlL3>uE&J!jlPMiR_QoMh^f8uysh>3~G
zZe_H>zBlvffa?i<mlR7<R9jD_r-So}YUc^@XMLz#whI?7jJIdVMY4K#yr7q6RFIFl
z_Og98KgI}!O?dwNtd+Gjh`!57rg}08a&lXkB6KDYX5QlDQ~~7#6bqtKd9*#^IT5&%
z_xiOY2_^eGVGdL<<N{C>H*Y`fP2{rywEeos=un_7TRA!Y4l_G@hkwP-e;h%qKv<X*
z804cY1C{H^Xl}d7p!@T+*+9nQ?pROOKSFb+?EI}scOv&8%%R$&#&9yFi>7%e5AUg|
zy#QiRCIu1i6{8A^6cwAiHPqp=k>$a*IN&g83?uy@Op0QQ4fwl-`#Et-?C9K@50>!~
z3NHtvn*G|=);owXWKwW-b*0QXp_ZUH*vEd$CCDB)LD8A#Xhd#a59y;q9a|3-8G}L6
zDbT_6zm+8^m_fahY~8!v+t<f-`1dM23!?&PBIe^?L?KD=xb<9Cz+eI^wGRY$+E6X6
zsfGCfj#7kUlTl&F*gMsr%+XH!K8o@RzwL|9jg6&teasvt)p|gX2`h^EVBSasdZ>z4
zObsCo!SC+QL<1j7OH`@w8B~FcynF#rqM|cd(gQSrIuJa5yvNkWm-+PIA~1PM)1@aR
z$GGoP{(WLH{9Eds9B>%ABFwN#T-B>5O&f>~yFrfxESjHdwDTA(rTMQ+5nP=NfB)eF
z=mxM!NJ%{g@W8Hes-AWJTWTLewCLNVr6uB)v()_e_^dl7L8D5e9$yZbXlrW&hk`D1
z8q0^}D1aw~A?w0S30QtTZ|uKZT1`c8s0ZZZIv9g2uG(FxmM9ucaE*x1>B*2UP)-)q
zX(2FsM8T$GJ6;VmV=k}Le5TJJll!WDwzoO*YAH+uP{Ou?w!#)4#9m;^$Mb<c(17Mb
zNc!vli9?&fod0?KU&NMRt1{$gU4(zS2!Hh7aVnVEt)aQ&<r7RtMMVWh8ODP>g&^VF
z{_z_Bt<NM}H^s$G;Gwne-Fss2(e@1(S!|Cyoi<E()V13;`uH(X2J`p(V%ulYpmpol
zZr{E=uHm-7xghZB+s^Vt-9(l1Hf|!gb?q`<{BUg#Nh9FN%{<|UngleX+b(@CG5BDu
zoHzoMv9WP+S#YZbm@hOVhyHt+OU(Ck7hjP|DInBW94Z=hvE*e;3CJL0e}nk8o2;xF
zK#dEQLmUDB@M67-p!+r-FlIHgPG)AWO?+o55R?YyPD@fU1cutTfPfz4=`ltX8sN1l
zbAGen;idn4^=NCTCP7nEa{y40M)}pSH>;~Iq2b{L0I>FFxq|hFF8mHKrWqYjs7aQH
zXU{x;e!ZZe0j{hqwrXI?Lu9MrG~F-Vy8Sq0!oV*#H8m|vh12PP!OR+2`CiiK{}}_O
zNujqk@POZ<`K+;frvNbti6#KFEd(t9#cx6fIyDtDP@wAouDWt6>?%2Avf)ymOW)Ns
zfyL>;gSTCn1B1Q;AqEENF42wW0B3UH{jFa;(jc{OL4~~aeed^a!3RMmbAJ7@6>u8W
z13QVRig(SFp;LwIxo~MP85>UC(_hZqzIM)Yp|{E37@TbNo?=j%?_8pib?%A<GO5@*
zQHa=PeRss~9XPA0sYP~H0Y(9CR27w+5B*yQJGi0aqk}DAA-4Vb^`+vydusRY-Ai2N
z5}%51ndzza-Po)rAd97rNl7XE3ACa+qwd4izV3OCS08|_1oN&?QBEW3qjs8%w*N69
zNq@1aMBw%hn}{b*a@(YfGzBVxmHWW37Vtm#@<hV|Az2=)@?OkJFzGFOI07uK^TZqA
z0I|rFgW76i_Yv&`(2gA_NDy=$XHRHnN@&lTo}Sj2QG6e}>La@DBbrrI)vvISfAG5~
zVOoCybn!KLa%Xm$n7<-g@ZoSKg693d0qRE{$*Qd--Js+_XxKR#gY2&ddF2y9Rrf3W
zOE%`a3c#TPS8}1o_{G-NR%%fnml_FlIkZN94kc+P&$#R^l8XE>QKRf1yCiBcKaWJW
zod<Wkmb$;bWYC7Ww2RNr|1-@2gox(doB_1EW3tly=I=B=?5&3rwMS^e2fn<#y!?P$
z6f|O(l|Q;nNzq6IK@v_0ye^nuMP=oI-`?*JrL<j^J2*N`V2Oh94Z-~UT=UU_JP??B
z--K9~3ZqI~?E!0w1w7b(9sWQ9EWmxZ$~c3JyPI1qEK)lDa30M`{&H~;Q#IXUtZQ$9
z7e83~01Kjqh!4Kfh;w<gLbIf#1f3bfqKcisAA47furM>*1Ns85mx-I6&{S7{S$n+i
zFy=E8X;AKzibkW`9D@h6bzAwB0|6C|FV^nOT}{l^$J)$wB$jUfUWjyapdW~xS!<ya
zVv5_d>v<b`+jre^13-lkA1^QG?XK8cn)el~!65r?Fa02Hfkc4&_jl&mj2}QNtejMp
zl{w-pc2>q_ziDb|S%RdD-C6n7laYXb9gMcA4>^&|xDj<{<)(KBKsQ`XI(Q`Qg`jWg
z61dcvm-4ib%XP58i(EU$$HrDL>Z_6>JVmp!+DHZv*RoE#%oGxzRCIJW8r^a6aR@bc
zwrl)%baubn&+s5xQu~+zG6a$p<32N_059!fTv<h39Ub6&pYem2DzzU_Tw7aHR8nH|
zJ_qs$;fn$q%u^ikP)!^sJLy`u#NplEP)`fCJL<!L#&oAzClsTuaIF2XN!J!uwF0VW
zTHl`+6ncTqJD*?vVFBjoj{DqwW6uHY*>TECmoiQe5Zyz2t&Hl=ZC6)|;?yg^eyYJ_
zuyyb`k5?1(Y3X{>A~lmqB_x$08Cq1eB7y-71~wZ(Tx`YFZ`Cv8ZUMfzV-1EOA~KSr
z{8VR0$7a#aez!Ckym23>tjE(|zrIdLNZ2~%@^#7u+{1X)a-L=mYlndE`cI`f&JrXL
zZ8?Hz^8+I+;J0a8>M*2&e32at4yb-~CdpzcB?X1HU%E~%=IEeL1D^vpbhZYcE&{FH
zJL{PWF|4U9f`U44-n@y6%FD-!BKEmH;glW?{Ok{&nwp{p1VThgoe3zhoL5=X&d$zn
zCQ@h<3=o`C|NP1ux}e>=kPsMO=H%=GH8J8GvYN253dIiSni?;Z2_`Gkh<K<0&IPoj
zwesq+iBE&S%;GF)hMVxU1hHDq{=wr&us$9mlapo&fK@o%N1B>kN6H?urM7UNMcA+n
z%bhSDM%r5Qf7bpVWbNxI_|8g$QXnze*1QB>zc=*lQbx2mt3Xj%8($-LzJT@A=TKd(
zl|fydm5GQ~+4?p+R!s;BkPQ6oeDbZ@VsDMNn+9%a=Z!pg@Sr-OZg-j_G28ct1YDM4
zd>))PTPJ9<$Li!=1&u1N<+f3s_hn8lF7E(liv{A3cG_B7UjnvjnOU>eD|7r|wM<X|
z4}*yU4}hChI!a@->q-+VMI&;sdKt-Xy*2<BR+EqlAan%ubh+jbeAVFhz@XKJ({baq
z;3jdi=u#b}6kuxtNkL~eJ-fUJchiR1prEDYguDs%YuB)z$ZO^S<VFx@_;GRRzt8L~
z0i2h=Y3Y2KP2U00Zb`d8_cIAtlf}&fzDTG<0tJAGaf~$wD^<e~aYT;JgS+_XJW=cB
z%m&dk#LU(BqrRAj#?>khTSW?IZ+k7hG69@s&x|oHEe$SU5EV5F+~0i3Sc5-4*xw#;
zZl{@F^VyvIn@mS!sENhw4%m;*P^4Z(4aF*e9ddO+N1ND!VgxRO-PVFkl=nmJKK?l1
z1Bd80@(PdC#$4xylaO`CD_-}S!}a#@c2L_S@xV$zGQ|*4;kF=91DL6)s2E}0ndGV8
zWFJi{wh7bb7mf0)8rz2gtbkcP3XFgv8u1u&GNzqL0vjf^JLCSPt}}A8^Yi|6BMW^w
z0>Q@tfDVj%6xvyPfo&YT@a=0+imHw}+Q_rbfQwr<SphE=&#pgxBnIg$8etbnh!6TI
zJr_YoSXtZreFqEmMZgX$pF*@G;L2|wVgZ(We+(Ctv%Bk4>i9zgQNLLdupl>X+~_ZN
zHie{6TU95R)nbR@nz9Zf@C*a~xSIJRoX|voCIl8p;?DdE#77X3MLd751t@E)UsH8q
zcWowaE*a(m!fp@HZ#GfU(L3^BWFYMzr~pJ%L*NT`ct^m9;$CI|CxA{sKsBT|YN9zr
zL=5_}mEqzg*{bO#abvoQ@GGL3aZS!OTk+**Yx`B74u7v3@Pr`W?&}MBbmrT)H<VL^
zdk*{|t%&W<iy167EruuBS{zbETYiWOG^D1a%yt^9R6xTdWaQ@FqzaoFI}Lc~K1Tc;
z%trlX$j+le?ESHMIsD!$qr}d2$Pze>iO9;y4SqdRP8T<JuHCZ^Ef=!yzn7PnXAdK+
zD6c!-UmTx)2UVJe;R^ShptWbFT6Y%XSa1E!wdG!tG+sA=6ex;KjIDj2KLgk#>o8cj
zMoUF|;ez_gr)SaUxzyj2jXHUE{#g;dkO2UUl1DMg$pvEm+czPG2RDu6D$mi07lMaw
zpK{3sqlsi>AmdbYfk7@;A%;02#r;?J8;DnW_qP_eqJs~mlZBiIAx!{rY$k{uk_P#0
z2FYj(0kr6dB$!`8LN$Kd?(3bYcXZPO)}N}RiRQt3_MiVb<2U)}bbqdT25~vrnKL~Q
zjv%=b<Z}R8kSGzrr+{_G73EYBZ3w9$1JX@_yHn8r^B8jmrU5W|Br^Gcdkhzcigm%a
zBjy)yG;(cNo$*W4P0h{mcT_>U;JfslX<~*jz=4LpPdivHoQ1V`{f4^(-N+2TVM`!?
z>aRKt={_D<i{WXR+#-1XJjj5dGlAV53tC(bc}(zB{(5f_u<cp&>60ELRo0HZVVoEe
z#Cq*QVF!NnfAU-$T;K}cdsch2Z3symNce&>!(4y)A_H$J!}Fmt5xWV>GL6*ZL!2Lw
z{`uzUsFB#wW}hFVeb{Jh%0~ojzNLsLB838e(Q8OBfKvMM<%=yyxBWnV+@JbJdJ_T_
zQMs;wrI9dvtCc0)7BmuBPXH3duU=gty(X+S(-J-O<N^aZIk}~$7JR-whbr4HNhchB
p{rxh!9siq03i;3fdliXeiH8R0?&!bLuOs7AkX4Z>k}`YzKLF3wXUYHo

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 aa4af5c283685dce5d0a0f590641a17fd45930e2..2922de58319c6cc1796b4452aab87222c3b3bc73 100644
GIT binary patch
delta 108934
zcmY&fbyQVPxFtjs6huN&1QZaYyG2R_LAtw3q??05ry||m-Q6JF-QC^&=Ays%*87jO
zxc8npGhghzzdg0>$l=Y%`8!-t$AQb-*(%{&{;9BFJrdQzOyt50M`40d>f@qfJ(<Cw
z51polrQ+S<)TindLKsMF--2txwj5$-iPL8<9-<&WDc-yuJ>9gL9d+SZ<v8M0om#Ej
z-i9m3BX(S!MLA&H)8PG@ZZ%0s+f`-K<IOiy|ECKr?q+u&dw25HZ91hLw387KqjpT8
z&cAB@<8VBZZydwkCwpqDDz}P@gE7?reFm3dOQQd>E;ZMsy=VHUyzkPlx^8keZE}~6
z<j0eDLuk@-*zX>IKW`m)l$lqRRSBh5O9%50?PC@NV<yquG&*tQLLwzTmkD`sR{7f|
z;XJin6G0de=T}jE;8x)RxeyZAHZ%`JEG!@t{%Z2BDh{m9)|JSOX+2kqW16c;qHGb;
zC7+G2NJF*_#bI#}yZxo`seV(;btz2E=@S+b4<gwH^QNb}j=&?ouJ+FNFFCHT=A{h6
zFE5Fm%k;QrYfey-o@YgtBBNsxquZHwb&FP8{{9wi%}g%O+47pU1EM%tQ1@fTU$3Kv
zy$$34y-ib<wf=g~wCdC)%VSP|ffQ*4c41rNm-A^ijfzPL=27y)L3-y7>>LAy{4Nq2
zH61hN$WL8dpT&PlGty9g9Q9qThIQ{&`I(_F>O;Pg(7BIHf5a<$QUfCwwx%)NwA46<
z>(zeKzJkbt8<_MlaPd77B4}z6(+0~YQP&7&Y6RGr*6|HWd*(OxL%0vuo!^lqovn=<
z7{AzzOv+1Z>ysc-taYLzxqQ~I-S;SFL6!H4$^BgE(b^Ea3|SIYk2LH=LZtR3bMMaP
z9k;e)*=NzoR{!_DydrB~Uty!iQ%0XHKK6lpQNlT$4RZ{U>0OQn;RNGmy+7onHZl3G
z9bQOmO-;_oO(A{S#G4zx!Sm?h6swVM>c@G8VWq{kLf%@<X}$MdqE6ObNQj%`lnFtg
zDgV^4ZLz4QzxZ@B1;;KyH_dW%RiSUQga#Kkg2`#v*ma`z5Z*xTaK_=_;nDF?dHyDZ
z%J?)--riM0`4M>A`!}bxPr-@>w@HxHVd3>)KQ2u=<IH%#cWrYW<E)LK*<*|5R#%9`
z{!m2lo_T!kcOG*-hQf=dIc}kkxpxWaN7l!Q_fDKjh8lNn4Kxjf14T_THdUmDhDQgp
zSUiopFRf^hdt;j>hCOS|cbl3!pcK`^4OU{oywTEl`Aw$fD&hzJKDjI6CSe{ns`fg~
zi&!gqh35TF4j6+wc*<U!l3&&m2&?Nq#k@RC(ffWe9G(FdNn-uqa!K6KX4IXz&h?5X
zNKagO+=5(Y6AgWLInuy4MM48tw@NE;3GfSzMjmB%8M~g+)<a|G8i|xBvD>k!QlsA3
zky+9i4yFrKel*eVJG-=pXD_o}cOHm|OBysxJWfA;v=>$U9I?p<J0=F-F0Caf0t@>Y
zlg@dEaYx}v<l&Se?2sL&UX}Kki4FMIU-pRl^7fWWP6QLW{yG8cy6zA|e<1_gSICK}
z7Y^<PoG2v7FDv^S%ibOK8b8XHTfSbOyKZEq%b`$=i0ci$(V!*1iHJTLyL~K;HK0xV
zMSqfTq(&nytIf5gz171d4)QT~>5UY-AwEDsVYg+t7or>}-^ADd+T&{hQtkINMEccJ
z7P*L63HhCA5!=tS{*^UuAXj6wnYRzgTX6zRp(8$Famx?b28vbdo9hKTN$7YpDNIVQ
zF-$L$d5~Rkg@!7&v7Rl>;3=s7mh5WbA-vIh6UBS7tnZBLGEuBOIi-Ya#LQxc2wOd6
zFkvA%ENDoY`KsyTiECq@SD!ca-czmgy6t!d-^a!*q@yha4sRk6AwIdJDLmHI8RDGh
zP&G9(rr(5AN=?v}g&HdMb;2hd)(F#FFAdlW^r_eu7G6_Qy-U)ts^c|%%ntwdyTqb(
zKtdivLrpiP>O@`mHU)x%h30MMhus8CW8G%r!P&*MZLagPvf&?b%3^03C2+VA<6l1r
zsmn7`qQJ@P>pSmt1&b50V)2x0J%(W6ZcP1p59M#T-xePlcm2$(ovV+qGCksJ{mNwE
zEg?ZpO><ub)rvcMqrc9KAa-pgg#7bomY5y2lNer9x1utQlm$ca?1HS($Yg@A_dMMD
zQ;Y^P-{&=A36Y!~$brgOU}Y|op-4N!Q&9{tG$>HHj*TU~&5_<efa2l(*4Nu6nncEM
zzI==CRPdG*6Oh8%7@L^eTyKv@+r~XsEK?q|6oeHNy<mjAyEu80@+m|1tmLRAFNF?H
zsHJ6TX-V_*XAhrCVtM)C1!=Jf^l;_T9P2ljCJN;k?+taW{ju$tEG_@gP}UE$V}@i}
za*2M-R)hBv`v_(E3_VxYZx#02os8OdaYBr*DKSw{3;2EpJ4mUj_g@uYPk!fRIRG+v
z9jjwmJ_k(>8*B4(+o$4*4{ZZ-B~!iI@d!bUx#08r^#4{U_?W}y9EBj@tZ27ODe-CW
z``(Vmo&f}O1B#1MqwYVsYKBY7X{m#6%AoY*m-;uAK0N#l$x~hN`Q^AFAKbszKDXK4
z#v7@?B=&c>k4e__AtxAgooG3eu`fe$f0|#S`1pX{gUG{U`lf4rx)Pfyj>z9LSO<H0
zys2$GHJSK}t+I8B&UC=M-1i*HHy=OJ4;cQD2z;G!BA<xsaOoOtBGs#V_f{QHki<DG
zcAuatzSn*EJ+(Kn$;5o`Tx0V=rL%jopLkz5W>lodt@{&R<$Z!~^4hK+bamvkEG4Nj
zwQ14S&d=jG2zm|T43=kHn$uFH6jl8|-Wj`)p`iGd;}EocE7>d{+JCuwQ!Tz)!mD~h
z&Dtc)aNAp=>h1liF8@hMEeUkL6$cCFS#VHjPiEce;97<EqiEjX?CNT&>OoHtI}KvE
zg?H$(R1*9<AAdD{2d7?aHkD5$J?Q-*rLnA*!o317S@6CB)yVc}vXDz}=KGJZcS&q*
zN4qQEUdV<`Qc$Z)ZP3(}bC>Dn{PZCUsJ1`c1rTM8d5T-ylOdrvii0C^+HH-lJf|D|
z(e@XtJ&_Vzd95rlH)^>pE&cPuZ0|ce-la&()HNA$Px5`&Ny^2=jr^Jz6c!?G`bV9X
z_}sm`@7lTIGosVCZ~M0wotC>R-PDxvlLUhgY8@ZLrL_k3r*8^wFmHeV8Nrh-i?wVJ
z{N<Q(R5E*hVt*DIb7Ayy?DcE9jY6^?&T*!j@ND-*ygVEU6!S*sdT(otmC9N<ed`|#
zfjYkT#j^Q*QpUm?YW**sC(-3t9m6`*1v#V2X}B%!4j|i6>+4UpGZ}gA&K94H>u*_o
z9`sn`8jod_$@1{fh~bRP19~c4#A~j#K7<<rRVJ5ZDfhxnI_4OjpP35xR7rsP^8US8
z(rPg}yGvHNlFG}J+2j_nykDUq?jd3_ur6HYT{pimSeIOm8ZWJ-`#I3y(qQfMgx67W
zIv^qAF#vLd3(M_>o&BS7S;GStA+PUg)OfUx{ff=|oO(NnoIb1dNO%R!t~NK%+z|Q@
zlj^d5afR|c&IR6NbY-fmC$*be#^joIEmJ?azc1H<3>;JR%+*f`2A_i>jv_r)xLsbB
zOsc`b_Htexh7rRDTt|oz?QGUQ7=`8-9P0&-OlQ&9kT0ovKHY_a{=W=ffPdiar8~8X
z3U9yFGF(h}!QgmtI#UB>G880ko-G&SG6qk5gT=GwZ@(_&pPwtN!w<+S&d=GI*Sg|Q
z1GUuB#$8s+FCJ;`9R=g6a9Z)u#!6YOPTTd2_13+7LxoRfg?wKUy*Z~P=<-Cf#PURc
z?(&A6Wk*7tAD&3vDQW8TIQ?w<+9+L?D)%gJPPsnQ=c7DUz#GUZAo>Uwq-exgoq5TU
zB1R7@Yc+#5Zi^5S;n~eY9~4U}X1@Ii3kmbcfLP^jip4g0nb5xp8~d+oK6f+XyOu7r
zvil>>J;|eX<4uJPFh+CIHl_Bj;E2<<xZyBeB<C^*YK(^u!xe`UR+k>*;;Q_1%B`rO
z;DI!ze|*U4UD0WK=<h>iwN+Bie#31KiqOa2c7psbrm{OWj~;b{wvQ%5z^466Mls*k
zrDByArI9pA3Jb|CsWWc1qyONTmzMU=@*bVeLv&^2bdaUs7;{_2J9nG<CHo7!*<NGt
zZ(UdNIYfvqM?NDsV1pp0poK&PoN8w1r~dPu;mEYRuKE4tU)1GgjDw5cx+B9MK3*~C
zTz?7BqICaGdtTot?&0)`*~{r_rDIX~zU+`X#6HI}idSY&ss79)5&2iuxFu4GUmZj-
zw$+sHsbC2uXoc;A15};2dDLxc#8jvownvd=B4TmHA5mk>0;fJr0VPl>2jqs=-@IxP
z4Sf9CraIBko4Y82#!A_WMiZNS#3p@{%j+onO{nSk%?la2;DoR6o4glKW-nr!HF&#L
zlxlwv-8?jt;?G#L8qj91HGb+;yCdQ%fe%pOHw8~L*Lo#XSTf4eap?XnO_4c#3+Qiy
zV`X+HLlfSRL-MrtrNrIKWMbnM3ojQhjxsy2sFy6Q9~mhr;i|Ap*EQkBEHZDh`S*RF
z#k5{Fj?h7%br1F1=K8p5!#%;<o?)~5D+0e|d~<aF_-Ke@q=}S~!8=$hjG^o>esOL0
zi#gM;CcZ!tQhPFt27&_H<Awr$1HT;AKAe@{kf#z|El|yp9y!7*&oQ2p1^ff1U_g>W
zK(%nVU60&tLY*@-YngYCV_J`D$)16Nbf`5JmNd*42a?%v&>qv%m%d;l`}%uN=A}A1
zoKQp8Sh%^mLxwQ)`3?hR`u>a0%<)<#19-5H>CzS(7icR7KU222TvCNm<@k2TAf(?%
zgEcPf(X?IL-1(M99~`%Yz#9%UOm1q7_Rc<_0Br)*^S&Y{%ZLLJh|1{|#D_d*h~S&u
z3knqLgqvCgw<AtH!*);7W!at185WANR42;3Iz=E7nhS2D{?%I+kre60v3zRY>r|;S
zhjwmER0_}uEG{-!q!}3tpc48#@5u_^jGq|Pvo2BX0{8dFiJ=0vHD<71JNT=nI!%K-
z&vhn($qLGfv%iYp;nd8y|6WZ=R-4qA2l36;rUF$!NRgdrb@Ec?docqA9c_5<Hqz^g
z@;x)jZR$%~C7X8G_L3Sk@_L(!GmhK2Z7p(~BTX+L#sXc6+>patbuJhGhrHLbEorw^
zQ-0`VJgh&aqAb#nA2SqKHh|#v&@a73OP1DJ-})&3;T}VKdr4}&lZul$9jnU~hEj#Y
zJr56z;9u{ACC~!o<(a=&q&m1r4xMrYNy4h&f)>SpnUze5Vicid*tHe^%bnam17d;8
zh=`E;8QC*XSP0h_t_W7O?d0=Ov}RLzDhHq=n@1l%jEHu2Wm;!@@?Yc_=qk3F7_1MK
zPgN!ER@*9-oR%;#u{P$Vn14)z1^NRivgCRgm3<)PFH41O8yfmjFB!Bh>fwPGbW9K9
zTkDKJ+nHQfXLszrtaXt*h6uNID_2(5zL<1PZ&Lb%{`S)Rm5Uq{{M@%R9ZrYZAX_bR
zVW^zT@xE|Mu8z5p)I4EcTZ`4+76&ONrfBz{=9V5DNdp7N663G_4%-v`j!`*6?Jn}|
zOdV7X+gop_P9}`7-GdXf090`=|1Im|<7-h^kWi7vrREZu&>yBD1pSTGZ_b%8Ap9d;
ze@7zVWolaRbGwI|oZZ#|!8k6tFK2QZ4L<JMW_6N7?|nHHD&G27uEtI{PS->gn<WX?
z=UmETeaZk5$Kt@(-STn+63~TGJcikrQ~28o&d_4NdZ#eKWt5E=Eo~_G3}LYnarI2i
z-p+xLNC|-YWwO>sOCJ}t`8SlgZgZA*H?==Jn|0azAXS_!6ep4+EmiiL)Y+);@IbNH
zh}xIcjGTm~W{6>>wUv7|4}F)1C$sfd^KR$&^Zx^%ESr;+*k2R01Y>7pGZ4(1*FV16
zpD4}E9I`$*h=+GBU6D+r+e$8{-_4zd3YSB0tiE<7Zvq~3W?w#QQ^QYWAnxz?N5Yr<
z6(cU-Q`fVPDVWahgZvuW^%5K84Idv7$Vnri;0bNc67haGF1FUNTj98+!0CUD5y+b4
z#al+?)-HfN%a*qkg~xq3?tBtQ%aPvISb$;Uf~>VZ_Q1d373c&2w_)1YLIjhN_X@8C
zUzi8rwx(gm3TQ~45>3cuC}iP)R+-rIStyI<UCzk14m5BKXUX(*Z7TaA4mtTL5yZH~
zNM-<i5Vt&t5tvksJZc<jeOwaD%VER-Dzo@hii>lR1ykMDMALzhIeoSjMa%)_UElnt
zu%z2cTJLs5L$jk#3U^f^k0O1(O10FD#Qppdu_t4~z{UV1mPnetk%FR@@m7!WA=!86
zH*W`L@;0$nw*n;L*EVEq`zQ0<=~NIUzAOZeih4gtO@HCZ*?rM};!W2<87fIeM;&1`
z$w#uZYtB)9r1bgTjVNGXC!wMC>>M*pX#mAvrPhc*_Ra{PL=+R1z>Vr(i*3iL0-iaR
zK+I9&K#n?U{*sw0@|cOQIUc9@`Ir2{M)1H(iMaX9TNHeom~@$Ft4Xog{`9uWX}hhF
zMGE_kV27n=I13}%lxD&`bS1rclDBs4HF&)dIYppq@<?{w{6?btxO6-pxcNKjuH`~d
zOl2x2?9U(eLo60ni|fy)+dIW!WnRI)wzT}oZ3tb0p>G7Mv0#O*w0xO4dELGG2%4fl
z@2i#P&w<otMYe}NWXwXHTBEd54yu5VWDfbf51$Yc2MlAmr0$fjN*IzcaZT`?W7njb
zHr4-$26`ae*+GiaGmfG|=m3}11Z{r*VR?yjH2)x?#WCKem3e!|3lHXX4)+G)``Ytk
zUPNCh99-H1m&-4a7Z)vb;2uou!H;K1VaF1?8q<I^$c#3`KD@q|vf8THWV2r>>TRcc
zlf0ddbQE2^AM06;r^4B9(e?C7ecY|KHM|T`?Fe0j$L^7WRYvIDfB5k0nRTPpn3D2S
z+cS!u?P!GonNmUkJqcCLpYCfR8?DUu3~32uvGg9`1A!oifHfXSDiFjKzE!+t+555b
zd3{1$EQ01i5I?)JN(jeD>(teV4N{!zpdrIq5+iEvZCsWlR9Sa~oGzO_Z0uNS?9jVk
z1gSvX<0hjP#zX$k>8`!MCMZ>m6nrs#Kg(vTqpRN*IB8dSDf_8nMXn6UIxtBA<jqx$
z;e!gihs&`9#I`|XovjJd8H!gMFa4M!8QXA7O8l;`#9%ldfceN{hpegIboDGPYnK}Z
zxLA!n7Jw?2|GxOU-$5`k4FRzN@zdx`fDN&W8vr-_o3pC`$AJ*2UXAwqLA`aJNvCiX
zrek2#-m1hz&oYoRYi7e<yl1ubh6aed9U~lOUYupuvF;{j79?ROG}>)fZIoS|iB^N;
z&MfOKPPqlkXAiYyqyh9uINs}}icZSvg^N6^@sQ724jzVnLmaf691$&1Wr+rwf@RNK
zMRGK3`HPtKZFj7lPn_U*6;G%?aG;(IZ%4J$4GyQ_Wk?x}ahA~(kU+Cm&K6R<t9C98
zJ9JJmvO%3w`#N}IwbMMS*&JVjP}v#_PE93!zYbDICpkq?7>+q`uPZ6NIWuu_U%i85
zSEHhU+e5JK2){m%{TdmktEqn?(gZ$B@l@R<AkWc#?U1Woyq2F|f)nqd!UJ^WwuCs5
zA`4+mz)1UsYM!^%soBro6re}`6mD9^7Bf<8bD#3NcXMO=WOCJJe&D7B)Nfc3h>o&&
zsm9k3VP-YPp7apjBsHFfR?{#ofKZWsw-z$FDzsvQqw#e1`;apUuV4E;H7U*fkGa#S
z)F)Zljqu@1^j{i^UJ9q8;mTDT#Q{!P7-K_b4IpS9)=KY}Vh{5t<@KIO@~*V+O#hMa
z*FJ5W*}6*PN<fBQ0Q_^V^!k<Opc+@}I!)0S{j{&l<_v$1^cfJ?sxOx?Pt|nBp<Eug
zVn=6*3cH;fh{m%A5Zxl}UiIRVWOiCuYuwLIKgWD=*CcNQCLCB+j?jH!;`UXs3q=BQ
z9mUbyX-WVK6YI)0MeKq<M&M{>Ox3NfmUd%H3g2Esp~kuGqW;r7BC`F70>(C#h=_u{
z<!88)RjV#Kt1Hj)Z@VBRKrg<S@xEOp^zYiXI5{D^xRMBnu6x@XO0$Ds1R4X}^X=FH
z2O?Q2?uh_WmW@Y~2na^g=?OapnL*wOBv~_2t#fGG5&N6wQxCWVoyQUw+fK~}BPHXA
zJ%KWK*(%Z5(RlmreY?rla+*~@D(h#!2Lu~KlA=+n#JTbcq07@d<;x5D*F2jMS2Xon
z%7QgkuG2&cuA4<qkm5+FUV8%$078EMi)FILB<qpMB)*u;V5G!pEvS8m<N*dFWBsr#
z$-0@5wAUsbcW_9TGw$;9H{4JUe1?2vpgY==KD>PSblm^U9w?PxPBy;9)@r3#T0^FN
zusD0@?e(Ulp5W<oZ)J_|F-}n(IfMCPqSb%?d}zrqrV8;%bL4LI%8lW0zCi%*4?g{p
zmIYe?5fMm->vP*JK>^!F8|rye7IY#)GTFn@X|5D8c8aqrWAUgD6#^_7NTpG#@0uU>
zi+?x}e*qxd<{K9X#ydZzNxuR!h}1^}9lxnKAJFo$Jch%hUj5L<SZaB$&;@nSy?IGP
zjS3i`Y-WcIJyS)AgzD?fq5JnQE{;sc6NgC4Xcq<tzGgZIlb_lP-#sYN^k1Jn3;cxT
z?iWq?bkJN|OZ!_*$SY$ar&Qht_}_m8X0`}HNFf0!G#$wGef4v>CKry5>2g%XEg>3d
z<@sAS&xJhX&is7L`N;e~=f%iNz5h@r3In(!^5FyoNtrFX<}l#nl}Y;Vr9{FLDYqVT
zc5hrGkK1fuwX-qps1B}e4(3##H&t3|o94yh5*QoP4_0wZPQV3)5UQ^$LS;Hr0q2sA
z#w-CNBh4wVKFJ1suWc1PwUPcXV)0rpM&ze&4$Y^Q-aoCCQ&yG-he`!bbWs6Bt83@J
zx155#GXuYFwkPN-4bZQY&ps5~X@!Ar{;8?liSJ0sBCfWdf_R2=`;g<ABBSHiMFh&@
z3#N5W2sv(LcMckKuDp=sh2|Y%w`OuYif)L}DJTeYa^CwrAf=<m1C29Ar{%{Me0)-w
zLdGbUy;DK|Ev`k_iPWfF{?4ZAloiDc^fgRH?Txtj`*O$MjaP3#K3StCQ#PwkJ$~Cp
zKC>Koa9N93ecEZcAT13KI9q+|I3gmM6~I!W5|yn`N6n>Tg&py<OH|QWiNapyW~Pek
z-3gY2|EDkTgOJ1H*+nh1vMR7P_S>yT&)DR`_4gi*VVGEW_cWLA6~H)NfHn8Wx%6ao
zhDEULI~;X5JW;lh&XRq!(HE;>yR*{KuXlFjB$CDsw18y6dosQK1zm{E9+34$z3{Eh
z526dvRRnV_R~$BkQ8@Z^0jC&kZc}CuU*@A8eCW?tCJI9AkL`uh*!RQGgMZqG0M($J
zi*v@xRLX+E{I>iz=pYE6@`k@e9z}l>XadQ0dwERG#})F2Nrk;^Q3frRN2KkwdEv@;
zyS33$OXT&3u*ex=Wr1$n)nGW_oot4{Y7dj;35D+J$^w@mXJBvb+v|Ydn=Bx5jWE@Z
zXCd=wObf=1+9fg+EMy%=h1Xt<;&)tqmhpTnlp}57+Qtobl!{wJs`%c`F)N(8KcW1}
z=uzC-h&!jPEto_KtFF1mSkkmJ{G_O|OUkX;sq!agHTQ#X@7IIQK=~)*4BiOrCYU_l
zuU8@SoVB~+;=65J=vHB1E=kypa@u#&<B{+uzErSyR^K<|;Sm$3Gq&Y>b_9iWMH6~-
zM|KwYlKjIoiLC$CGP0KpJTVk2&ca}qh(sWfthUuDxihMdG5XiIXR&<vI;c?XnF&wS
zRUn_6>297V4_oZ+sOMEQ^9dauzlD~%`wFtzncG?ZcV<f)y#+T{y#S^G!?UIH0yx5p
zb)!OQQmSH!q4AvP>HXKj_>YMLol4(X5!;&Kvn*vi5B%xG6pP5m?Oj@)e-(`FUe7o4
zg5&BX4gSz7Ufs?q;mv3QnR69nf$Bj%^2qrXm-eIMwB5s!Sa>txuB7Q;4Jk$0X0P$M
zU%!4T-4Nc<Zex!B66+nW+BHQ69v<@+cWt9aSed2KyU~Yr@OwSIoi{d`avzo1*M*ds
z(<K+=AX}O>(VgzI%N{T{e7bj;6+O#A%%r0;(tGv9|E2~NnohcQq%VLp!0C<i<}tO;
zr*(wPh9q^4)#=`&M5(Jj=6tLg-N~{mRjAvcTRpY$l%Oz-N8^}`k9xV*%E$MBqf#3o
z49(fo-yO-s!1+t>$O{J~r0ObbJh;G`a(uD0R_u-fqr8lvUHZu6BkPsheFgWGyN)xY
z537&>fBmo*OxD*wSUibYBZwSuY&PI*`C`IwFO0mV{Q0xC$|jL4c7f?|T>L`46LDo0
z)3*yHI;P?C-4!t@Wu}x=f98efh)u83$734jODl@@Y_ptZZ2<G0t>yIY=Wz-O#%U^*
z<oaT|ECy7CDwFI)8gkyPCsnmuE&bS-pI$>PZyar;b2yWC_zUS7v@R)#2LZIPNVpuQ
zuVHw0;tV*;ftn8LVNtoG=_`UXC#x2S15`@WI4)H5G^f*<>%}=OFyHucjQ2K1UvIaS
zHu!^36vkmcOsNYs(_LDU<@SKF3qhKfp4oRv$4zP*^C&St-OFRUP16*z*wr4du|Ew5
zTKcRh$MY|<d0l*zw56<sSC+i6#7^O|#VEdU`Nrz(=AT8UUwYifME??S&!s&Xy9st{
zlNmpa8OOz1_a<myZL>&wBC7cFSWpDD0tSHbjcB(&DqLLN#Na0|l#;SB{VYx|vhxXx
z?diEXyU02+oL8ZNPH*zlSLf>c^PcTp{32EYrI6y*F-pc{wxKoWRVu5ERMxRrtW@*s
zVj(2BqtDa7;%1|LvZk+{aLjq-2UwdYcjaQNm<w{htN6B6JTLdC;ZMWsFDdFNxPwf_
zsQ*PQDXa&8ExM-0Zc7n0mo={8+BZhU6o3Rm#Ur{*0@1bhQ)$3>kKc3%Pdce*3QtGh
zR6On^CZewn-9@>K>hM#ElT?PRI3DW{gR4rQg65d+@wK+>ndp+MRWS)m=ByN~@N#dD
z##q!iX@7}3LBGWY62aBM$YwR~l?9I3MB@qO*O!9#em=#pHt=)ze(1tgvI>T+5S1(z
zpXCjLvCHEgbsnmaB*0Q~k#pzwDh~fP@Mbc8Uhlhr<tEksC3X?3%s!MSOXwmX<3|(o
z{=IKPetP-{HeJp1#=-RO_BXEE*onV>c}sqT&36n`RHlHwGig#*4X{%<hb=QxTLAg`
z%16?tp!tZkb&vmqDo;*8GgoNm3?u~x30FrQ6lEb8-ukE3l$JDf&(t>VW=vfn3=|bI
zl${Fp)~8EXM9f~CdwmqmktMu?027I}F3QjD`W9zr54F+9cymMLb;bK0`fy82t2#=Z
zDAi>dvQ%I-7FPLrCnZ}nGzF|oNiX;yVjWB!Sy^X;J?N)$N@DDH#A8B2D(p1QnMoyw
zijaU;uyd3m80wMSy!-XUM$nw${?y<FrUWabXfMzBwY^VjXlzcVlFqVBkdk_%TK031
zB<y{6S8}?dZ%G86|K^C+KoI|t$zWxw43^xV)<!~}lE|s3{g8kVD{2oYxWpz!=WjQ<
z{|_?R3o_f<AiY=@ej(vswMn=NX3a<yZ2S9PMl(FkY*4`1$109)$=GyDZB&|(u^gYE
z^8}Nk0A%k?^L{mbzU8%-(006hh9Mv%ed&gb!RDHuR}-wVlL;u`4A+#?G6(L2kaVey
zNQt+w`U9GuKm7J2!V|!zDSdh@8o!UggBDyx)t;4EU1rUpHXGhLJf_G@KEk|lg1|$b
zqqleT)}x18U^j!ykarL94IzHe&)}wH^SY$YO>-H*TLv?Hsp#XU`DSJ-rLh_%ryZfP
zJhi`Q?SUGMc7ca3VDwi?2|%Ev{vvesNk8PW_K^y53ifrvNtDj8pU86xSy8}S<h}ic
zWHOpWiPC*h-GX_TJLs?_Szh+sJH|=HbmL-SV=!){iM4-;>ds66yac>jV^d<*&&#j?
z!~r_;0?Q4OPw_-#Es~U|j^jD}7p+qSltV@v9}EjeGN2CcLHl3Y+L~%QEdTC_=>Hx8
zz@aO=Y&dR<bv5Fjh2OOxP@=TS*S>ePrRqI7Nb-(SVbN8x0CL>?sSEO_jN^3%BO)^q
znzuUIax3;F%5lPK6!dFNnBfycuYe%$x+@8};rVX#O$~kr?VpT}$$K$NJ!6md7>Yhp
zL~w<$-6Ug)6*eaV?~Y^b!#}tl7?cphxu<*jSvO6n=rjs`Y08g!lsl)?%Z){*+~R<4
zgL90@eBsQvr`Xx@;J{&ZRS(XZum<;Zgsg*Sv|~I<J<m)?L#L~OX?4toD!p<53;oZx
zF!BXclco|kxqG=T1VWICYOcfWm4kh081eOFX~xQ5h5I?-e}(HE=p}r093cT8tAWO)
zzZuOs#eP1@W5<;$6lc)e$O{|@xMEB)tlg1czsrqt=@KU^;JAorNn!M;yE5;bXuGdY
z4js4HX###KszH~#aRdk!pdP6}OkiV}Uw~rWx6Uhx*9)F?66R4VgVLk5IaQi09FvkI
z$y6J6k-0Skvv)O<3Dy0Rsh-GnHunYb9P-XDo8#;QY)g2xFDhXkD+;_p`8lHi(;%Z*
zGW~g9{_f~}yIFk#+2<R;$6%_<-^5AF;(}^t%HxXQh9jW=EkLM(9I8pR{JiQ6Y8IGz
zw)I!qV&%DiR%SE)vv2;4(R73`_s3OtYDBk%(EzGK9DFzz^IIKL{?Eb)04I1on0YVU
zs0SCjIq18aD&E>xAMdgI7IuJ%#{V1Rx~Az`EyZkjgM)$Sp*eF$!=HM#&$5RKfDTxF
zIfPjaaT~%3<98$W6q`zkih{9O>z^t3w)-(Q0ZHbDKA;$Xj=l;OiYleIyBeeFk{-I3
zSDAZ{PO14<&Sd!QDyKH>4#~<wtaMwc(NK_N9E4UET8qvuc2^UT1^$a7%8(C2yCXEf
zDey|Gzkllycgu0P{nNogP&1&0;hun<+V1tCAp#ek8o+32DQMl{H8)3|hE!UfkB^vi
zgW0ns19u{oG-iogL4rq_Fq%QfqQG7qZev2xs<)^p2^j^l2sQOaRe4{XT9k4L*&U!x
z<or)q<BKoIqV`u)GvV|{slZXOd>j4KT_8mVPlX*JC<uy`Usp`DW#}7Ct1`E$ykyN=
zm;Kv1fBd5#f;qb8M(}U7)b&$zuvhfpG(T%g&sX>)eqmt^@cMC2dcb^PW6dqC2op?B
zLwd4OlMVE&UXgp84KXv{zbGG@SdbgVMsztpzB`!(EwZ?|yWbbgS(2GWbc2A#1uF4R
zdu%sHwB^~ygdxAX!c0bs+HcwW+uQBb%y-rs26-wEJA%H3<KxM3wr$8~4<EFB6?ijQ
z;+gb}Ryuq5X0tLS<Ms7q$$U3Pli_ea|6{$^caNa@FO3uBji=+`0b2%>pqJFI5SS(%
zE0LtJ5_`>%Ax+b`*i)ihIjohNT90!m8`d}BxK3FW*;(J|n2G|o5Fkv^Rj%KyM$Pw*
zJw7_~*)R%l+V;`ZA<td;e`)Q}s6Qu|hgp2)rR&JZLwkVX?~gbd#r_1t4f_`_J~YAw
z0Jojl^cVpiV2Lb)fw3RueE-th3sM%%kzd?gZ^-CWcrboSvn}kBArrlZ%GLh9+6SCA
z?Z+rUvOc@QOG2E1X7FFv@Nnk`Z$HBHa_^MGI%FyoBZgIuA{(@M{!Lt76I!dbz{Q%h
z*uK|o8^ts)f(&>sqob?2w|FM{5B53A?N6&W=PDbk#dDu)w2$XlDb_%@FAQ?k3?`=n
zVU0PV>kjA(I;5L$!o!020ZGTf^^gL8%ECZx3)sEvy{p74(>szx{T+GQDjUpuRnJ%e
zK`r=VSRWMl&4@{RX1YZUE(8Ub368>xL!Sqa0I!tR#es%WrInq#Rkj306Yb&BRPUem
z&ZOl&C=l<6AI>K7qyi8<O-<bZyTJBGQJCL~>?GfjlHzge;6zkZumn}yZ5t)sf7_;<
z1<ZG6C58c|<!*J)9+R+<B}`eB0JT_&HQ;FMNp_d`xoM--DdFv5fZCPA9)0Ra?j!rt
z2SDjxR&J{~e&G0V+~-41n~5(}NAZ%eaEe1dZL6AFxaa2gRpYw}-hCvR`_MhAVV70B
z$cety;}~|Q<2h~nd**M>_u+N##%m<jul}kY52UiT&#ciTSVsU9Gyui(rDt6AZZwc0
zPn?GX_u`L%bJ4<n{#p949(2^*(XBIL4#wu!%XRdn)|sE608pok@j0U!6G`>oXecFA
z4D0o$<}|;gSRB!Q0Gh5`LIoIbR@PlT^GIdAs<b11)<l;1`mVw}kW%{BE}xV{)+%?8
z6W0M^%L9~o)9Yj&&#(Xz+9SA$g~r>*vRJ^o_oUi_b92W0Eg-^=9zPo6F{k~wQ2|X`
z`2&7W%0NQ|x|T)ytW+9z)QKx1CiOE`V$?ZD)oKZbOdokEF#KYo3W;PTTMaW=1=dPk
zeUU@>U4RQf3(oL(^EYhdf&@nXCAWczRzpch9sNa7uM)I`JBr}Cs$9PdqE|<I`CIH)
z*oY2mDS0*$N@#3riZQd|OAUzGc3Cus8SIjBF|B)~Bz63NucbWS2_8gO1f!`}PXX1Y
zt|9J2qjh%HEuWO&upSG1Yy7r$o85ex7f*5$<3|!cy}5gQX4uw0Q!+YIiE?)&`s`wB
z<}!DG4r(dRilOfTaN-I(nm$0{SjUA3@DCKNh)-PhU#CHes<{ia^%?op>9WSr_;z`+
z0e}?4zw(S4POewkob3@&5MDVMt1NlROXHuOocz!2^iGyxyy0(5Jg2hRGF3X9ce1;I
z-YIhe1Z&UEIaAd8SqV%&Mc+%djWs^xK}`CvtZytcfI`apDg~0ZprWRIg?*;M8=_^r
zDEvW?e*bmGkkyxTYwo)Qlq&f*&n0<o41_m!?UX}j@UE07R5cwCv-~C=+^1xkT<h!(
z$`NDuPYTOC*e29OVs|h`mgjG}BuanFQ(w*@4z=7kzy|8)2NXRjY@oP-wK_n0doN}j
zwe3?zfyq(3H$$@&DSB?}n4mc&IWbLs(bPmBO87C}Q~?hVsFKo}(0U+gEO*QVEmIC0
zx7(H{z>)HY?oB<=?NfJ-amQ+j5GP7ny2>rNc!-4n58vX%Yc6%ej7AT9LBjIbg@-5F
z%*=3iIJ|*q5)2cRayA#C%>L1~k;drasC~)_ZD0zE(OFJzft%BosU3MT+|u8Q?*`nb
zAKEHiBqd5mLIUK7Xr~?+EEX&h^4fh3U(#hPwEB`?n)uZh6*K^?_ibon%HgBdk#}5S
zSom)P?h`frKowYXj>fit3Ad~<3VbQ)<>6jG1ZUJLAwRsjo7QJ^;+hA2{MhOuQdlE|
z?9G1Ibr?T`zKkAwH46e1TWEhMA6Yr7GxK8~=wZTXPB0F@#YeX|))G{e+me!Mu`*%f
zI%@~VB_@^{=;{-O{P-=HqiWA3z00D!s$4bK&$)^TUs|fl^`v<Dg?s6#hWVIa=l$ad
zzJ18yO8H{sZK|Hh%1%#$)v9Rab2owSJ)CXcb<Pjmz@}zWvBWT+enEZ#R0H1YnSKrl
zAkwvCq~+5yz?hlwuPo98ss8HQPRr8gd={f6XT{7tYIO}BNjKG>uQ1L1aa&a+AGHm=
zzo91sx#1-e__wV~X{jJ0;O*~2s>RAQR3*=kx%?VHT?CGAZQak3Qv7Im-U{Blb>H&^
z_<qLVw-3GID`c6wSH&A8vtwg<h-$ySX|7ar`{I+OrtCGE&KKh&Tyl9@;2+p3Hq;il
zS=lJSR;_mZ2CVJ-BY)<dIp-^YD^Cu~J&(<FWjYSx&!U{ue{@Y0?+;<K<9(||89<rW
zCMWzra&AbKM|b`cG`j5bJ+<uaE^B*i;>W;V;diS|LU0`jKexdI*L%`CH<V7@8rlQ9
z9FQUmlZ2>6!hog3=%V_Rgh`mM+s_~_gQdy=*~YLJY-)wAc1aJCx{yj2z)}eG<pNci
z;(3jg7tBpl&{%l6*yW$@_w{3qQfcfM`~-bC*nMR(PLGaEdh^gs(23{I>1xT#UJ%Qe
zDB$y)klMI_k!T+r;NJ&@zeLgagarE>bCI&OPcny1Ic=l;iF<>18)4yRzm$$8B(=jQ
zK9x#|Vm4X29Ah`5HIy#J5);tlb84Emj%EB9$sL-CR%Q0^C`8@l%Z<->OKc>Cd-V8m
zfADD&r0nX-`{_Xa$*nE0TR7HTlK3RfeX7G$i_`e(uG7TW_Uze;1)x*k*r0np@6yO;
z2yZrgFLPyNRlDZ!ipB`6@i8|!t)udv8qn1hQc$AIH*D@}F#P;B73Bv2bBeiykX!)5
z0<<+N4|C?r-Z%a&daUV&n7+%fYs}hJh4Qlk*b%IjvXE}uLm!{JyB|613S80s(9~b;
z(ZN)hE@ICFbRkZ$dw*6vm7ed;xzoKAMy%PYC-z<}ZNB{B61EW@FB_JXQ!h#zOuZ5b
z^-qdX(7CEp|CD>GZsWLk8K+nb8KJ84s3!oc6cu+}Ic1zbQ)Eztbxh46UxJA5M8T%x
zaP2ks>G?Z2<wrMOE7!9Ek~*p@XTUB&sKy(6*69iCB=gOQdFeU|6*%}64i1!M7lcva
z;j&br_GimWE83cxpR#TqkY>iJ$9+AmjtveLkdQ!wTUl8t*<f0_IzNOSK6<2keR0eZ
z<MSjjQZ`pK28J|lWC|^<)-PwQv4lWYyubK33H2_}y2XCQ0Q12?s{?XSr~=$o7ryJ&
z&DiCfLP|X|@q>ySPE7U}?M~0^;DOSi5~vGUy7-%zbDQ4TH#618wVS2~eCpCuECT}P
zjRf~OW~$0j`|2UATC@AzT^tZu^g|8T7HHpm1tb8*Eh-6Y%s7XRxLb87cV~XZy<ZCv
z3dj)DP8RwvSDo_4o9=e^Pxu%(9Sf?c{`pEQETtNxl5ozpDDY76nhK6MkVs~+Ma}Ir
z?zwwJ0Yd_E=o1G=nHF+tlr|_zP=CM%+!atV@qw!N!}9h96nsBr3<0<n5Gue|YXBtf
zshQwf^PQHW=|SKtW?WdTz{0}jiTC?|_A9B+*E4ba1@gW*Dq@u%6}@9!&+XF6OK{&+
zQHa@2)0~+QA+@yV1w8ilT8erc)sL$(a9BbUuE-hYM1<NVeD-n1OSu<vTkll(6%?wz
zG6oc;RN8Y4xaH}KzGMVshKRL>uGTrtKNn;-V5p>aQxWJU1J()<9d5+rAnZ~}L|7eP
zR!J5*kVL@!9H#9w4dC;>^F3OMw`*y`T#stZ;a6Kjmfd3hoJ~9KE@1X^Z|<nXWH?ku
zh+R>s;0Wr@09v{O%x%`eBAG=)s@aW|dq&sRFyRwt-jZ(QlSr@E9tcbw1NuwO;7}<~
zRZD6;Xcxn@cKZXsa`LU)a*exWVf}WY;n!b#e8;P@csI<jMz}S0{dO0-?vPfXQkqUH
zC(2>PUT+;Twsk9&+|NnubDr~5gxHzW_c%{tw~emG#B^7e_m?LM>jKJ6Ph)=xG;k^k
z7(}7ueNW-7p1WV)6_Q28drQJ<vb+mdOvp(f6IF>FySp{~{eDdAuh@{5vM5#s8%OLF
zFj7Y}=+V|!L8mq_U!dLA>D-c6MLi=D>JQ~iqv!KwQZXpHF6z-u6P}-NhF$<vcMj#g
zT9uCqi)1ZSdA3rw#|_*gTN;cAw;B|`VsO+}&VEpfj4|Y)hMmy0PE1926GK0;Vx|$h
z%1wdgn#1BX7O-j=PF(t<^o}yoCoYD+g6mSdMs_ds{X36)VD$7QtoK+(9B-OaQnIsb
zc+?mm$7lnT<qTcaG&RJ<>YrxBOSdh7ku=TRy$g?2`^%3b92(%Fnx)aL_6YI5!3}C}
z*hR>BiD$s?tH3l>Yjw}hrTK4u0jH(6xOiiC!WEYiu{>~BY&}M4?du8!^3NJ!TiUIY
z>b^z;knqn9yO`v{;?YKkI8fde0aHldyi&j$Vo5=0?>-tGQB!-}OyjTt9FBy4AFj&J
zu!#2nhZT@Gy{7uZpEHs+oL_N<cII|H8~Bpc9`W*9FMYYrC4NoX<P1JX^hHmPxZ)X0
zSm0G7_ANGmrr$vS4p*#*r|F_#dKEq9>r{@XaUennEUVcu`|z<n5-d?yd{Z|LW+Qg^
zR8YxhmjkoSl2#x}YYUt#Kny>C8j!^(bSEL;|GQW8yK|!H6%u_JjsP`wT7V=ZYwS{J
zdHA7b2ihmCKtdWaVR_=>zvKBVLb*y?B@Y_|qd_&mJE0(Jvo6<QdBtK&`+*|9n;nue
z(-lPZ!4}Tk&%pc>J?6#fFb^~$u+TSth-Feze);HajzPQ?A()u}@=N-1fsvv8bqMcS
zCgaI{B*eKDu6u|&s8{Mrl7K+;n8pY+E;oVQbsZ%H#z_Dl62NsNJKM-AkpF^v_fPI2
zjYcBS)i1_y4dgg5fhiXewe9Uk=P{CnCbVgm+zclk*yi;^JWlaxxoR3GlLXHAcmC2-
z-nBK?Ddk5bth#hg5=dfy)ylwo|FEp;0Amqce&%4dv;}i(X!v;LPtY#Le#@$w-cI9_
zkdk1)nS^!ZAZn_m;l0La5F~#&EjQ;cDk=;R$7faP<eW#RzusBvYD>RD<RNUtd<wb!
z8Z9f#-ee{tn`iv#?ad-T2VlzH4?*zbx5jMqTg#qBVZ;tf?aiJ<q!cuf;Od~?8$uw1
zH)m*#Rw;)}xqkdy7ifY>U}eP{MIT$S*k%EYaU>W&n^RJ?4K>}s^qQ!;#&|ej8lrkb
zMM}<r1aQQoZmLt>ibKDrCbJOJo$s;F{U0o&Et8C4%ha_%0z<}7(B`6RZLyzC+TK=%
zI%Wl_j+6Fm+rR<Qd8g90O4DBm+nKNuJE<59>dQ2$CiBeord)eo<3)gxepdCjcL3_s
zm)X%AgDq04KtHNHjP1;>xU3)s`ZusMS?`@>uFFClG(1&HFKRQdci(Nh{PhlC|Bbo@
z3|l6yD6&<$r$58rE9>I`?gU1@1Hc3XuIpg?r1Xyhb8t{~_U{j?A#rL?n?{=!b4^%R
z^k3;@ej!pp^{vWPnpO7nV?<N_@W(Sq{Qc`62sC53Zcz;O5nnHE&CY9JD~J9qeV6a`
zo%wV?yC>L*>3N7UF<_uuJd<ILnlsz0rY2I}tau^1$qe=2{>8*;v_toF@3kSeyNp&?
z1ullDvoStCq6;u%tj%r1F5-MlQ=p=jk%`Z|t@`8?_Y>CA1%>|VVfOcs)M^{s>T>kW
z_tvE)As;qodd;_0rCs=3Fb>m!JvcGt*EOX&phF98auUZ_QFnE}pnmlfPT{f&{bknb
z8r)muEu@&yavGWyQaBqij<f8593tZo6f;3u%HPFCpWG1J_Y~f!_P*vxPk}!w_)f8T
zWqxWd2(J1qxUrnIG5MU&GEsGP&o#!<RXCqP29}>&LpXbZSe#lj8_yrMpFaF5PtDP^
zoyiiAz?JOIROt*bAbYE5%{*?@AFdo$8(hZyxo-7yop6b$vOEszjdE|{v=gvqC~bs;
zseCnyC429gO`aT&{o^AspFWF>f+d4AZ9`+LTT#9DRfn;c=2z9QdD22Z%##fKw+k4+
z;q(V~rFNC53XYz+ptVSjFMETzrLQ@%dJAhHu@}=y>lpyw1?yuL>cqUid&&JoG<Ijw
z1_IY6-1A1vUmk4XM4mXvtP&>CZ3*_{Xeqa%El7Nf(#-CEVr7Bef>dF5%ubUBsDU);
zw=Y)tX$2-RRk=}k1pOIC5N3dS`GYXV26;rQytj&?m+)4X7v-#*o)IxH(=ia-$@Vk(
zfQcqC+y6U}=w2#?b-~xg47elXY3VdoV%9raDv#;niVTML3-3;??AMEcqyFm$cY$jO
zcGeA_3^^j;BGgS0d;(a)QMuSzp7T#%2>L!ty$KHtbdWG6TNyr&=Hqy~&Cnl0BA7P%
zFhB}u*3dYO-NG=rN&xcS)wev+-VTHi6z4bb84m3|N2DpbZ5J8VFe-ZW3>D_Fx{2K9
z&zb0-OtYbu-#U*<FPb#$Gs^C<NbE7K!2Ih0m-_4dS-|)i5!G3bZeR(@vM{uA0`qD5
zSBk&Q$*V9iwa|JFpc?F&Wvz0Zw=c#|({K(VvhHae^Qh~oBIg=c%zZ7-_kbkU7<i(r
zrc)~S>_(%%tz*p*Bht|}{8Nv%t!Q^f2?pXKMYX^|O5}D}#fz+|J4%}_n-Z4@`g_<f
zgxv$m2fgl5?qee+LeS*TxW%ny1?dNukR`Z5QFE*VZhYdcsNUz}^6E@8TNED3YEVl@
z{{=>AHOeKrrx`B-XbF0c>scs$Z~zhO+t}v_pL7c4q=Q|O$&C>PBmYvDf#;t0W88kN
zei{_E-GrGaD)J3T{3|8i9Ivjp-UQZM2AF4VA+X<wMMBoQ(*rG3PNi5+o;=a>a(nT>
zYtILT3Wb#RcZOS{SNU5#%BJq$HK(aGztvdXcWM+OtDzqUji_dW%<c%Qm*;2ur4Bn*
zy=t#GPTL&@7JDpNoL`-=vRWID6hBR5?#W)xvWIq@uhG?}G`;A`L-L_ECQWrL5i+^~
z%O%4R$xG<>T>Zo;?U?kDOEnEDoB8rmD&CMO%LDhqBTs3&U}qM;HJK<2E9+#lru!jJ
zQec&&xkkaJ*6waFnPg<c>EXPOeJL5!+ET~CL*XG))<p1%JfmwUlzENY<N0FX+?rfq
zcfx%2O*i_Qb>-7CExpeY@2B6wp<UgO3n5iN%K?X5C*v<gh=eje<u4_ne=ud!>U)cY
zf`nm`_AA1^U}{pd+By7B44;;nQG3<(hgibRkX*GrYg-R5@%MBu_x8285BuS(H*6|R
zET7d+WZ;{5bfpZcWPMT^t3P=OM~;psyTeYb+j8l&-?vNIC0p#|fU)g)t*#9|U88RT
z#olN>^RKLx<p0b$Drwq0GZFc-#%U%iUu>vU3o+Jk<VRPtFQLY*L&fMQ4z}f78OEBm
zGVj*xGlI^8;qu%!)F(TO*A<6Sij-RKUR0|TEiKW$)o}_RhI4cb`t?&Z&pXbN=k@vw
zlBy4F1AlS0LfFt+aP-8*$BI$9fhDWCQ-}>7m(LK+IK<Be1GcwK{7+#{xn=>)nG1hl
zBz?eD7HREZP}IO8xIJysuh)f>O;mkmHfbBWGoxaB`Ef8!J5~0VGfVVz-4$|9YW*B;
z9+5_8LftGpR`p~V`bx_0ABIondM9g~q=KS)Ms}TkMLjpzvt?Z9ExPtJ=*NMu8JANU
zfF(S1Y^0-+keKb+Taj-0?}kig2H!<KU$ATod5+#7xFawzG2zRh)3djF^2bZEy!q^R
z_W5!E;d>op-ls^6sZ+5Ho_(D7rs@9qm3zvkli8!5I(k|S{lmc(_IqsKM+c33+4m+N
zX!q{K*VYk}#ns$CG;#Y;%&dL`eTaK7?4N_Q{U}={OS(R4IH8DOKO^H&+_Bm#VIy|;
zcMWA^>P@OCq}J=ms*g2{_9`b5-<{lCFnWifVPGjnX-V4UG{3*2Wji#YcejYq%Yl^5
zt9F14x9xfR)!ppeI6|3AE@`_H6+S%wOS>9{9<{!SUdu%<uTa7#=*uhX*8~Q(q-68&
zGOnK)9e69U`J2t2zDuJ?w5j2AzS-zwT#UTg_!Mh8<5wI$m>n_kt^7|+N?!b+1&cG1
zNvb=zE7Cbf5vM7O=UsZ8CV`3X@SK2x_&EV%S+i8tt9`SpN5@?LiRtNHMs>Gxxi-Ys
zOkue%>f+rZbC6|<h9Tc&bw1ttH-t5tRwU%1CCKOs?|v+gzNn8ABKG4?_>}K-qxmi@
zP$)P~ZG&_C_2}5aG*k2>6S0Alle`Nfh6$>#6PCq#>g_SQQbk2y6XL{gD``fgJzJZ-
zVofVUb}q62g%5tq*o2BvakF^t5@~vufxQ-{#}ruDD<PE<`U8+2tIKqRnCDIX;CRGx
zaH|$=k{xU}_n-W{2fORM@229kM-NWqjKvA3Y=cDjSSsg6#(u=8w<R3cmXMm~=U|4C
zjJ$ZPn{>v@n=5sBzr3~?4NuGMGvxzD%)a$vtQCWwmNefMH~9=&$~k}J5e7=6CCMD7
zZk?}Cf7h0VfF;-R;NE@Pt+%!tHmqH1e^!*#%JU(H@z+v@z9_E;mVzVffBsS9D>^xB
z8{=K&K(QRaJybPcspG0_o~UZ^p-3_r&_STGGHF5^xivcIvX@;Mc{5Y!Kjqvc!p!%_
zr&xK#eC)@&gk*<2N&gLj4B7sx@z;Q#)iwN{mwO)vm6|Cs7;&@8%SVsxA$9WQb{7oC
z1zErP-8l8AZ~af_2hxineYC5S``}+6kJ|n3U7Y3HjXW~8cbM5#gNuKgG#}=#@;2uO
zzlZ-?-FM}O-;XzY)D&!b`awv{1qSjpOPlq8G*IHD4f3)Xd>_F1K}(&SGbXgU>mCq~
zdTSw}yh3eNq6JdBfpkYWr@DxNXN%PTnF+oIW^G>2Ai=26gWz8xqp-q&m}`He(=Csk
z4xjbuQ_=DZ@XdmiITB9fFy1QXKzR~X<!U#yUH|0|stBv6G)q^t?rfFPiGxPwCg!=)
z`z`rlo!7%Q)TmZC&>5|o$~}ldYyN4Xgc<q8&`K1;A;j{1^YW76dN_fbXdtnx*0)%B
zgqZi6bVj1%-r=&(rhI5=u07kD_)x#xmX{a3;Us!x-yf@&$tm-AXY8f9@WQ8ygXLVt
zVqB%f{x7)b@;Zicw)-nzB^1@^P0{%(7T6JsC`hiXZM4_NjFzQTaO^_7vtb2@acwgb
z<(~IKw0gaRFZ!!U0_L(=Ym-WQ=r^Ad1pJUn@0X3JEa@#X7irqKajw=RF+qP$?!0W@
zs1Sg>!!0rbX4zcvbh|vasv#AvcM}lt4C88nJ4;1>fJW^jH-ouF?2Wb}2`4xS)mrUO
zPQ+|%v~QH(8ZJQ&Gf`Y|`G>3R)Rcm0gNdIg>1C)gRu&rfcBaq1yyZ(;l%~Xbv@!Iv
zHn2=j#t;|w<b?q};AI7$+qbqQXU}KlQ7&2X*8fQVKQw&>R8-yjH6frNp(q`%5<`b{
zD=J8L$A}0>cgIz_5s+?ak&^E2F6r)WhUPo?`+u&*S}b*7=AP%Cv!A{9v#%1#8CquS
z%CiWLSJsEP%`??2{B^zukIm%3)k>^%TMhqkK+#|w;-q{K&!METpj*N<lnrs)veQPl
zbNx#AkAEabr!N7bb!*+sp;O%zXbjcws7eN{hblKkzwGG2Nrq|Ae=b6Qt4H|<5K^r)
zH}hyDlWM=I7&#u6(j42-Dg*?Ko*eyhvZ7U-9|%)PkzP5^?D;(febspa^OumN1v0s#
zxVVY4>j*P-Y;wNV;+A~VGW5?|uizu$IsT!lswyKhb9SFCIIEwrPIWRpCp8sGhR>8W
z&@bpLI{Yh|j}ILt*1kUHk_Tfw#Sz8VhM3|mdb++*^{Hx=97;xUOC&Jy+fz1=ed9jK
zyBn!E9lOQrVDa}ye_wp`xr^Q#WZ$_e1?$UzX|u#%D!eS)q#b*Zrsxu?b)WKRnfC*v
z-{dlmI|Z%nPd0SyBmCM-Dr-2h99i2>_A>8kQb*F{8LCdn->L|%4+R8VqN%meVlzbr
zc|5REHWM66Ok|i#2|#mkJ$y4S>^4{-7ho>!9%@%nYm^hX4xiEIC-?F3c^0MEqAPfg
z*>iqfICi_AfXc@Q@vW?_sXBZ>?->|qU={`ku=vwjF~yl!_NzCUY*uz7REzDTcA26t
zI#!=#51Lqdd{p`m(M465^DyUn`@h1}`@w!PbMx-sgj!Fd;)wapheE<j+V7?_7>8$e
z_7y5`Re0)63YdSLW7s6A&ihrPbQH)dy)(4H!Qtjc4->72;DHTU@Y?z33U1e9CMr=4
zdp|UpHIr~MMHTiQUF?-5>?;M?W%6)(-wwR^g|R@kN=Oiq6yukkpMNW;5>0w1_O%v{
z*BH`(Wy35u(|vgjZF^B>DhB+))WKDuCwTF0_4|Xua3<h3D=Dd&*CWdBoOzJWNAZ;3
zZe?xdr(&K0B-AM3s$h-ggZ9Nja8fmI4EN+MnS%Aj#RZaj?D@U{7$1jrXZjkhaEta=
zj1^yTAiIZ5y@pd(`;~qij3m$9UpKN!(PileTR?aufmcXL;ZU(bwl-?xG)Tp`(s_R2
zT#5QU>8Z+MePCEpSWQAxhiu+R@!7%))t$+KGyM(-nCRYTkLYD7FYPbN#w`4I>FAVL
zj!mk}E39;EJrYL1(^V`rMkxT{hA}|1#({x!odX>+&17gQ0!`bl%Vgo-;fMWCPu<U*
zE7huOGw2yUP43m*7yWF9-1qO%I2)A8V<$&HwLY-13J!f+D>IF=s{u4QPb8npZmhp6
zD+J>q<cRv!J(@!h)$%)d;Es=v52Se@UAleq@oB4j*z=!BnPx<-sNl*9{J;CIk>bL(
zLl(FmKG-QfMUv#;0-Ih=Z+@3V(CW%6<NcT?>zzjX{pKH}$GW_>AaB_4!X71{(kUR8
zLo0XIFlXqnEpZOzbr649$lmgsC1`WSyD))q{HI(5m^m%;I$TBm85E}TG4;(ZB3ZC)
z+SY@1X^G~w2>p+!be=WU&=n&ZYkKZb%JJXJ9bDT^e*m>ad`M#N?j9UEFe<vWf#6s>
z`HOj~6dNPa+=j|KF(IE9UqLU5&+{yWs(ECw4O3fS)5q!s1}-f7jwU3}A3F65JJ=gI
z=7WSZ@31(~D&xG$h~rB|Ulzh)N0;{X8pVNg<`m1@Jig1@Yc7x%0XT!aa~?$e)0Y|I
zvU2<#q4<uSF;nCNfOPByLL2ie#Cu2Y5{87RZpmWa#@LvQNE292wihhJE)bEAIDb>f
zo7@R$R(uBdHgWz6Vl1O{?aC<(-L?a*9Q@(yfxr3Pb+=w0(&Puz?HSj8>-PA#V;Ntc
zh>M*9BoJ`kslz)xd8qE?Vo;t^;=;Ut%^UMFpJ4p!M-n><FS!<r3YnAWND8UZA&9T)
zP^#RQGIOlc@%$zou{rmuHHSzh9{^C~WD4y7e#!1KeO~A)TAQ$nGY01_Kvn5=3^y}-
zOt>ZO*iQ@cnItj^-YA@L+oq20%FuG^Z?YX6`JxOJD{uHV?D(^;&tNIHXMVSYA-ic6
zFiyuoIZQPsVal_t#$k2Uz3G8DrEGO!YqBt|90)@B=M2rEekBIKzk2WZlR_0rpigI8
zW$MsW^`JqoVKP}6dWzVrh14F(gzJOhj4R$t*@kzo2W_=O!m^jID8ZT+jGEiwn8>Dv
z;U2Cd@SYb><?yoZz@@iN@k?{?KjgXChioms;rzMILH8U}*6iG~w%7Z}F?HlgwHuRW
zzHf!gDP76vz`!;)o5`8n#xp}f?sJ$+umB_-^8H%{jUzH>D{`h6?JuQ;q?WTqUNZ8;
z2uEiBSyq2{q5t~<rjKWZD*|Pqa~LtS*)3u%rDuWOp}GxQXWWyD4OLA0__;9ovI>1o
z{2P0w69?yk8Yg%~c%4hEpqRdU^$M==%`Wb<XhXJo?%o_+^F&LzJrpPPMqAyYB8Xma
zSN4-8zm9OcmF-F`06Jg`7X8-)y`#~rbY`?2H*Xy74>4*dZ?3sjGAPiu*!|!)ewSw9
zH!cB8t^+V7dO8C`3$XW&z}+8xt!sEp1E74>C+RAJrl$G%2Ma-JaBi1B6&MOSa7tbG
zYDToGO|z5{!u+gM?d_CFh&_QZP0HHQ=JSc|K=jnv!Q_Xz<Qsg+%)tJ19{KDWQ-jcQ
z)nC1jwvLJO;n(3W^P|_ZWnujERod*T(P}mJP13p4YbUBN|4P%3SM!@csadOg4YIrn
zaL!8IyeSxFq|#jwt<z&@XW7kx6G!>Khnhxhzo+SSDkXJa(>B%pf%XRVvaWAon2a2D
z@NvHm9l8>-KbcSG3Xr|rJNpLKC%cqB4D{E-4*yx8XR!Y7#*;9#V4-AIfB3qCCt7b^
zb!5T*IVKG=hKHZ{Zu$Ktght98HC1o4ts1;F8+o0m%<Xo_8xVs_TADObA~112hW`cA
zCYJLvUKqq<@x>|_H;A|eo+i;p(wZ_4tdTkZ_cFNBB!%LW)-Noiq|4_iGm+o1pZ)!t
z$vJ)qc8j5B&!4>dNT7SlAH~>SO|IJzchGQAc5<~7)DWYXF$Pc+U$x6Cy&e(JMlsmd
z0W4kKx7Sy;ye&`ngF4g5=Ad`d74I;;qkR~PFu)gX*UtA%DQu=3@(o$(8*Mx<v10N<
zN4|CASSphY8^iM<!QQ_Wq(qu^<w}bVS!84!9DnK28)y0`pjWg0kB#TE`#q@=m-?Rx
zMLFVPn<xB7vMi`;Cv;)mz0)W|6HT0R^sfWA*QEIcC%wHfExPv$&?ymSa09KI^PXR6
zea?P2TV^iSjCRK2<|D(ALM<ZT18$j*GgzqI<jXCM?ZXj}NYaOvE#%hsd!JXmFB2!v
zgnXRi981SGOIwp?o2pr-?(ei{6I+kF4fiQdrn;<ku;uYuT|N$-!v=R2cKVXpoGZzi
zD$aGawQUqp+f`qn3KA!@%Z^+YCDg$qZPZta2U>zKRo8WYTbVaF>t{?KS6k7wgVeq1
zT$xd(u~F@6FK!ZD$!^lAGAzh%Nf0ko2?&Y~NP>lPQ=DX6V;W#p?~7<DeBZxz;y=B1
z(-1Ov>Yyt2cb}`+!1c`L1wVTH#9OU(uqPntA~2#n3u6jj3a~;0EA0O(E!<ZL_l=)|
zgzD}*?&}+vIPt7s5QIljHvPoa16RCXEc#N#0o-O-bL283@mz65Umh7b^se=WPV!!H
zHx1<p6L9yh7|?{1xO;*=8HsmdicNolv391cWQbk(v75Heq`)?#*e#nMjaG&2!GHpp
z?j^+@7ev|8&xOQMM#E=ZOsy?wRibJsBqBoK&gKiw>Zr-P?z&6Uv^jjsH2G-V><Vh|
zJ4W5i9vYRq4@|xFbIy0=Nxx+G0Y1oxs06=|5=t3Y>6fYRA%_7<>7y~Wa<^%6EK;Kx
z)3T<iAyRXN7oe#G6)vTaCY-i5UyEyg%K?S`Y!-a!c-Py-s5WR)Mi=$!EcMli*=(Tu
zyj$CMRwJ9E-seQF=wzev^90(xY4TXs7gh{P@OWgRiDJ|e@en%u&zE)OCiBQQqLu@H
zdS3l#9YN6)2rq&8n2U!Ay^56lM>Q|fKDy6?`uP+HsWuCv#7yQo*7+2QY4vUpiHOYg
zh0u;auYzH|BA`2R=7fj`53LQE;a$VCu2rCQ{$W}n*Z+KDx#f6w`p6;0$ofxNzPuYQ
zmKLLwhdcBbY%ojP28<fe`9te)fq<ZEIBBNnn`vKcqN!DSUALo0y^p0o><N0HAkDYW
zGCoVUOwZ;k*>!cG5b%Hl|1TDg`B2Bmxv@N*)yXgMutUG9r`10?XOI2EX4=a9xhnG0
z%Zu80Hu&z3SJ%D{i`%)jLgFq3@k#q7YVrim`s}MiI%Dm0em&dz<pFY#{I_g)mLw=M
z^lfx{YC=P6Bwq$^|7`cI{Kws^XU1I)Vq_RmDDeW(vz44LWQ$-vP|}kQntwO0zn>@d
zeO{EmBwkE>plWtYe;qC6`!+F}shVwkX7(;$h<Sc?D0~1;+zlKZt(ld)1vJzV5pe-|
zb*LT`77v`9l<BhkG)gqNG+vzvVRDhk>QqT8y9qj@`8d^7%ZoHp8VS_cAc&KGc*k|4
zkit?h+YI;9paz1UJ=V<1zCG5~jdukygX19ub3aAKv`(`B<4HQH{9rSpuMnSGY;({&
zR3mK25F~K<2azLVGpZm9uVQ#KCux|aeDo<B=2Sd_pJ8ZVee1$mOrNWsqvBLkDL+0y
zXsa@I?eskX{@O;2)=7N>CX?rH5{;xSS9_N!d*7)-FQn5<>b>q;RSK)PNj{_;=3T~_
zn5wAyuwYX6Xg|@k)T|gnuDV(VvW|4Ie8nNa5`3Gx)X!GTV^d3rsi91g6&G{BSt~C{
zWjk|S<@r=?Eo89r<4%(;GdNKjX;js*p}sGl%!<Qm#pO_#CM!FEmRAoCrllpi^|uN%
z1P~LW9Mq?$uN@FMt-gKL*<06u_@@_n=YH_Km#=dk?8#mh;;kFDs#O-wJa+F>!_CI@
zV7_?qJu$qiohps2!7dK;yuCjpYjqMIVg8m`Us$6}c+9}l@Rj_@lN|x)`0apz6~bpr
zn@jInw#J<KGtlN(eB}-@<mnEqY+gbvT*UdRUW&%o(}hcy4HTiP4Fvr<VTQ}$Y%J}q
z0KtdWg>a2&>ugM9u;*`&zcny)TIwB{e8kpws#7b+wE~z@7$Z*@^WvhnSTc;3k60$0
zfYk^_GL|D-LC{pmPjrjaJJ}D~sc+n6MN8i@H?ORCmW2P8wGs$UjQji*WfV0GjGg>^
zi7){vU@kMkTHc<ag=3?o`XbU3kZSPodotcjsaJz^#H76f^IjDO-L0c%__}2NF(r$S
zCbTjQ3^<lk46YCMNkB@9q}ezPufOqIzrj*f8L!Ly&E5BAl9==g9ah*6xf!^>W-zAl
zTyc2IX6u`tv$wNC+R_tM$ByR5n-L;bwstEmx9-mA;*LzCTd%Inwd?qM9QJ80Okx+C
z=7nBISey*~3ceg(YI>3q-(D(m`>DxSW`%H0=tZ6~-Nr}(cFL_0+c)8{?wt|o+zj6s
z7~!W+FLIHzdqm=Ivur27F8*(2->Ja<Z#M!+5vqQGQEZ2xD4>LkpAy`JY>8(xzuekr
zK(aG`^0^z2lIWPn?9&RaE#2s|Z%LRGw!ZBS3x>#=<kV|#ozT%Rl;UbEP#Y}gn<8-M
z84)WIKnTd%|1t34gG`IGK`4AgSJ`v|O<;U0HkY|eB8+D7TDX8hrz>lajoZhU-TDA4
z7^39FGa$|+tS*0OxZkCujjg4d_F#YP;5m|=se%LL{Ve0*E-uq#H8~Q-0RK79c8Jt#
zyl96#5KH>~iDRf~$M_S{Dlddb@_;9Hg2qMH1CyxmbC)`CM_QK45FlO9_x8Lq*SM{2
zMFQ|AGLj@DKPvN$gs{SCzKEOmo0I|QKMoiMQ{MtR;g>z6J%W!}gD@kx&#AL1%|KMt
zDBDMZHwXvwxY&s+w7ZWoXQ0g&YGS)8EojIsql_Y7DIz)XXZTpVgZl)O5&Qpvv#wNG
za*b+NT%ce)#TzB``Ey-H#Y(0wYB+2##clSrp~jWwokp3AcBS{gEp)AwLPWBI>R#8l
z#nMq6B#>dcfY|<A7??h=lG{shN?T~ZZ#d|1Z^}GY63LcNPZR5W?Rh2J(LT8!>cAz6
zKK%1e`laKI^m$;|g<Th=c3{BrIAWfji|^F!%0XK98gSdaS#M$-i%adf&WWB$)sx7M
z7LQf(^A6@Zk0?$<YiopqtLcY3BK9Pv7_6e`y)XZ(oKM2P<?=;O>i7<hpSMt5UtFh7
zDai%Sw2_<aGRc_K&irJ5^j7koc4*N0=vtxil&xYF0%Kt!UJhx4?D~svt<%JHwJ}zU
zk-ML4wigA(rR(2A7OLhHJ0BjDkMEh{=L<MM6Y`FR`qd3ImX>*T*Njho@@NkFl$s!*
zPQ*T``Gw@2p~7@FNuTLB!`<aQyu(wGN6SvVAH{hmQC(s7rpLuzB1BlC^>U!Q&O`M`
zMKs18sj)Qej@{l#WzKTC{CF}2N-t7qI&r!DJTx0o(?+Ju_7aKq!B2U5;?tHeGZ7Mn
zH3wPUdZ;;dlY?&b%NM=_tts5An}V*{PlgOwTaF~}5n(tVmseWuf7k{5?niY9iEeW3
za2W@HeLL`L68@y}-b#y1`cA$1pKAX^oWBGFH05KPc8?QIng72cM(eFu_x?^GvZli7
z%v?0au|jb_paKLs{v6Mc>q$H@gZ5%8ka_F%_`B3vddftQ2+vs8E(5T0dV6D`bk@U#
zDfeq?OSdajewqUXW+IK7J@EdbC7sep+%zbC@&gN_S=M7h5~OCug808(hF?gau^~(@
zym|*45i!WIm&4y*iO>B@8xf@YKY08UU<^6S2A?P&G;uYV6=-J98NDze$tEO!Ug)i^
z;mZ~<9WK*y8AW6~78#q|uQh-75EHH1x-2zdF_L|#pLli}`I&K9t)MLdxHoCI$6#+8
zs9>yNSwb*{h7me8w$0_05r3F?{u={}zLob4*8B{=M$b~Njo5A}D1d;|l-3}YK0-iT
zngpT(9BxOUl$D(HQe*X$MT1~TUH1h%Xm`RVBobFHMANS<z2eQZ0)t8^$vHW4(@kac
z_s)rnhUlq!{m6E~=c9=Lxs`j&N0!Dap_^kYXtt7^F_WM-!rtlv6>n<6{{5Iz>d-;=
z_<SlogJ%Uwsy?4s4o~S@9(KGRm~+X;9L4nXncUFbWr!Z{^mJ3a(>awU+@=YSm?ZYu
zBkK#n=`1eyDPcpW;Qy6h;{Sn@QE&iD3A^EPWJb6L`IT4>u>qY!Quj^vvgE|8ztZM!
zOim>Dxy_q-sB~+L4ppDSZ|i>ip3!z{C-)4A-K-B1D79O0`Setza9aLT$ruKNV@-WT
zCMz6bJf;7sif4Ie`bU-!?a#5z*wTXv#p!DPiQ|13zEx52YH#`e<h2aX@bXJ39oXWM
z$6xn3U$Fftw)6yL-nURHHa^VxvK^Ihko4<v5T?%eqeM1W^i9GG+u|aJPhHXG2DJI(
zGD*_da&k7qKQ%b=X~JvPe?ns>l4kJOgf|mMVH1~cS;3g`iSf8EF2)HXJosn#J)C3d
zXx}v5bwJQT7&5zQF&H3}((Tr0Gkp<NP*9y)kBnf=MBNEHE7{BYIznwQjkSmEh&r)5
z!#JOW$?b0T)%MN5{zC9W2Kv3Aq|lWN;!#3U<Q6Y-NBAlrBjFg_J0ED!_MCGkFkcHB
z$Turl9pP4+NNS=n;Vw74NN;Nd+i<DGf5Z>E5<NFA17`v14@{yJeK+y9Zj16HQsoxw
zyK{AWQ=bo`^N~2W`Eo~}qG#rg8&ku$J-Xt({J`DTKJ+)MwH!MS!)l=>PEf!Yb6wU_
z4>VWj+>i(QJgpFbzCl%G_|9<w8QNhBCg+ksRY5s$TK2u>xYgGtm!#|>bJ!C!N-tO>
z!3(xb-xJ#nt16?y<_~+}Hum6lraqFM6`Q^|Uom&(q|ny+=q7B!_2}6ZVMBWWHe9i;
z?HDwop=IejK)U=Bk?zk|7uO>m9g&js7!wVmO4;rD&Ere@JUvGf@MKwVBE@f4r#>b{
zw8<jn`zzbyerc~5tH^qmLrDL)uEm_(er8~IjaYn5Qflvs2RkT7B7L4g&|JlI<85YC
zg0yj}&Pd>XKvAw5-QMwU3QPe_NHc}#9#>yw=#S7QCL*X(?k;#h{WbdB1XV&s(2&kF
zcdmxlfbK1Bl2^P)M0#&?e-eydh@52F%a)!?ut%{6U66^_Ve#dVM3=Iyebw$NN{BDB
zp14As*7C?;QP;$tn1Mxz-145bV39XKVkfS(I=Q>3?!TV!uWQxA%i>vYb;U-Dp#c7~
zeVu7lTYCeLQuWy5C_oWk7i#-U$&d2-wLIKQN@hmpwH3TsP3Gfl8dWB(6SA@-jewCy
z6DukA;cYGagIsOr;c1Gk+s(S8Ied1+3qD8swnhv)I8;uVJYaK^EI<&WU`>9Y|2&jz
zbZ{cM^uP`$QTYCUNSRXrsd!mrKyU5Os$0nt+{Tf%rGKTSJbxEb3ZcIv?zP>W5{!V9
z{wbdak^huW>oA~-ANBc<-}5A-Vcvfbt@NK16~Wk6$H-*B>l72gkyqY8xDgqa@bBCn
zy7Z|of4=<zd9=jm+>NERbtZfvL{^d(CPImb&u8bT_%)a{V_Lej^o=HwF|_V6I4j&H
z8Yt7D<lVMV|A*y1l*^i*p5|(3RPs$I&5lGCFnfN{X-Iw(F#Bo!=jpU#-q3K)4c*#;
z0){OkM!&dt`pD+vY6|$?U!Ry7dp-RxL?Q*cJs_1y0>lkT8~a8kF$3C}c%%Y1<QIlA
z9k)W6wtwiG?2*m=@xa^3U#3`#n~60PI)<JQVdKzWUr@e_Ubi<BPYL=9{PEwxV3FaG
zUViELWwmu11W%={l3vHj|BDaB<@%q-j);xSXGRrp`1r8{2O;o-fnJzPr9nP?sHQEk
z55!ta)GcRvlTOn+CQa4Q?JZkO)7uMS*(T!{H;uetW*}?HpDsKQrhlf+=LAJi5Z-6%
z;BeE#B%MP#US1m{WXi?6K^S6gCh;CkUET;&lZ!nn0Y=8lQuDVd<>m|&UjJA^)DQny
zLhl?!GbYQ?j@}W~+LD=<Uu*u(FwKS1iqtgUV?PFK0pCg@(w?DX!T=eG7mY1m3;bn=
zHV!O~BDAyF*G$hKOf;|*t8lgS7Yyv-=RcH{4Go>DDi@Zwjy{A*7n)#llKgynvfink
z-`vv{XTq{7A|7SiplbHc-JqwlYq9zC%hQozt3PF#<tePP<<GC%zumvNcWD1>p72xx
zD|$i{G1;A;?=x}lp9xzo+Alj2KO$tTjqA;jx)e@=_Ge@z3ff_8*9jnubn7wLTpero
zKI*?6CwOyxex@~{rA0bwWKA3x>>X5jPAU?p8kr(v;X(!kq?{i1uKBa{C%sw?;UQ^y
zpVwAq&f;T&IPT@hd?Q!+`GJ9r3GYKYesWN~hdu3mW|&wo*#4PoZ%1Dg(Er$vCX+{@
zf;DsblsOh2ig&H?M;^>~Eg{{}AnV&RB;mqn+yW<&-%B;SX7?1ZF^AeU+_gjLHBE@T
z`_K6Q8$>*nZhh>nG<mu~pR8{<%iAt%&zicVcg%kq6D>d;JuGJqqln{WyV3W+bE3EZ
z<cnSJ5R4n?pz<5OUmfQ%Wa=?r{uW;Itg!{fEx)n-@heU%dcTtsv)j3<gjLrw_QN*9
z^K+kDgR^Pi$63pt{{TdIWy{0bp(}h_Lq>k{@Fx7U_W(({{dF|M7WsAOqxtA`xA8sp
zdwze`|IQRK0e}4Q_A&(UP#hd^%Cm-{l%K;o!6B<-ar1xHTP2_yR;C|=jk0je_vv2e
zZUu13PyJhE*R>8pmM0tiCu)A5j4gCc8tiY}tgVY+7k0FtjT`@Vh7s~F_}9<8o%U5R
zMX9&<!#Q3<*j6UTX59Itll;>1s*`LOP^j`wX?T5pfz_*B?n&IDyC{;BQrqz69*r%9
zzx>yMi7L}WU?G8NGNleC*EIL+i_yxL`jAJdLql~s9|F?lUCWHyfEe7r6Z^IYd7h{&
zE`w40L#-Q;h$Bz@=x+s+X|NG5JW1zu=aNjR^ytREbb0gjbT_~(A4k@xDP<_7z3B7*
z_Y!;w&0x14dWs`1_Gc?HAoFaVnVP6fvJC<ws1Nr9$fX1v)6X`0&nA%o`Sq)}+^;`z
z^`3i{mcn4gBHF<+Pi)+}qxF+w^p*+uau=$1>nlUX=a>vgQU9>KJOy^lmR)`i&voHA
zG2qEOS8u|j{IZ)NcQnUPg~Uz!^9JTid8^a1*lGJaQ4kpy*75X;TZv(F=#@9*lTguI
z{b0j#qH|t=Owy+nMRW8K=PyACgNvp8%-}ow;#)mwDHzcP_UVpA1KhrGl}w%FZTA_q
zZrla-;yNpvX{+h0s)YN3+<Rm&Df3VL-^<6nLX5CpKeaDmF{;XV2AE~KXPCkp^A)Xz
zmS<bv%6!nm1t4nXGwCXbCw<I>_ZmTj1pc|O_cC)=tf5eDk3UDx5Y?$6%Y%C1t}sja
z^;>~hRl@eaC}k1Ap>y2bI~+OK@8Iwls%DJ8HfvUO)84y5t|Tb-jUCa-*;|!pINOyp
zcNOOM-CKrjN7l^F=y~8f4WIU}uxyA6ogsYT@!%)EWYPWE0iAb!UX5SKyZQMpyFfd0
zy;bg{Yj!S~t&rW$s*0luYvB`)*RM3K!3SFtHs@y4piEvArlyXX*G}*AxxYN#C;Grp
zx2v;is^{?xNgsnq<Y;EpL$fB?x`7GB7A-grTWM=UJnQc6<`<HQdW&D#LeC}Lj!Lg?
zU9?0{^h{Wh4+O5<;lqf3C*F%+iQw3AV#pzWx84*(cnUw}QLZNt7Q!GV(cPPh6B|R7
zG~3!-Y4t<BRLG$JIE@hM>XCfqCcsbbcSrH|`l62W;lX-TL6VfY@a=e;calgUYYSW6
zH{oxl{xEGl6LNV5^-^cXyFC$klVfWy^aV4^fhtG{i?`-j{4E0gYv}~XPO=EmS608e
zU}ySMN3QqpPZEyegsmq!u*Aj(V+`qn(n;!n<A&_=SzlmU4qW7FgR%pH`0#pUE=bL4
zN;NqULy9G4`#CyBR{sm)=(<|0g-e?24h+PMCp&Ax3qIO>zM$0}*;>d8!i$o+Y7g0%
zgsL+aw^e`o2gMKxtsAi^r6#BnPLZm^YXXF-<FMt`;|?!U?ur(zzO14%+}GB(e{H}+
z1vL6+n9*+M!azzaVnu6D8XM1ImJ$^X=0nTo04!J^k7cdu8iPKDpd2Fl#b;{mkGrMO
z8M-10Yy6;K99W5(<ILxmznJzwfnCc|Nx@8osb33Y@(1q@-Y_4~8|_XS;v7WaB3Zf#
z)Qo2Bg@!UE9Xw&+;J~@&_Ds(M?LoBr&QCdW!<%(;x=xw(I*);&nm+x^ayYoYFjn|b
z*cY-jwR+7i1~lo14=}ibD$JIF7i3jkgJ>(eY`cmGAFGf2qrc^aYB?m|yz;t;dl~#S
zA0@ex$??4qWKdUa#X>5U;$0~K!>N56jZXWnx!|+gjP`}?jGsZMXM<Wh%lUH|g(tOJ
z7$Lho6@MC@S391rY6>UbVZEQSp{~8hQ7?l3`@c<20;pkT@xoEn9`$(1p-Y+i2jVA5
z0aWC8QF(VzAfo4qm2kq})9qrzZtCAmk}3aTo^ZW`?lw?e$}Qd-F45fHloA6hEp{|!
zH8?A5!^<NS$}wGUbkR|EZ!WCW(Vh@=X8;VT<ZL@k>K0hCzR{iycIVdmi57+tR>pxx
zJG8eC7AYo_Th933cS=*k5&EXt2m2ZA+Jfg0DbNEe%0^7MfUxsvdS}iodCUh0W^;S9
zLB_0i*}Ycvy7h_;Y%>Ez3Tbx(Z_rTdhBobb8G!B?EStT&ZP2#fIbYd@sneU5!t-8}
z8Sq0oCQ4GawzoH*U+$7{c}wvfNIvMT!&I|9MQr3R9WNFV5!pNr&A6vG60+r>mf7MJ
z%?8A9SC87nQ-^1Q%&9Ti)=8zd-!2^z>q*+OUhS#flS}@Aze*+k1_lOBRN2Bz#|vl#
z1rsMGChq#4=>#6H52&erxIdDe8i51o(_IC}#m7AqkI72n*=AFq5CP&jQVwu-1RF3v
zl-@CuIgfgtNWWTtA5C-Uz`-n^NrT&%49=?61UN-n8-(^t$APxterzp52>5tcGTML-
z_i&BheX4wMng`DDMKtHRD`Jzdx{5UB3o4shbEL(A5vT2~&fD>YRK*0um)Qu^`Tmu@
z{q-Wm3O$fwf)#U{`|Ai)&-R|J7=r1>0H6mVfpt6oG$9WG=byYGUHEt*zVj~sxz?um
z0ZCp7X>ZV{*+r@^;4vd{vzZQsY~95{G5)fWp4H`-mqqPcYdAp##g~n5>d6l^bE(XN
zD7&REc5%T<MO+&c7cmhejx89bL3rP_Q{@g&Bz-(zB&Dq6jr%TsG1YeqrajyL=5^`Q
zZ*e)a)(*+1ia2C*XxJ{`$@!JCX}H~AM!dU1#izCF{YmL>{cu{_+f~I7|E{h#Kx4-v
z8{xqo1w2^XW}o|-S^%uBilavPqAv6`IbZYDZ&x0-S?!Q1pouSlyOO<B$_QYVPSAz&
zx^YL;_O>RN10~EkOXVS}y>lS2UPI^zcjiSg$~Vu+(uj7DATisMW*V<}o`(lU9cm+!
zWCZ<`AcS|(ijLme2=LQbUqMKeQc2c5JHcx2(VhBa0Lica-4-E3wMaQbj;U04u5`b6
zhx|ZcqQQfA0cR=o?8YCzzis2UQGJVT3ee4xi=s^&BBn~R?MIC7!6gSY8&~0hYKj9~
zZA&yat$sYC2S-1}^YC<?tFcaRt{Dj6G`vn?5EQ%Ou^MfAUx~5m%F55mOvX_D<K+C&
z9zr_5?%#Uz>F9x_iwl2W0uM|;;A(B()6?@ICT27EGn#O(ZM0JTEGWd~@NMW0UM!3N
zatPqW4Gkp)lc2<IF3k!ifD(hr2g8a{FT-Qtjb+|WM_8DL4vuY;1&L+6mUK+RUo#cq
z(Wx(IPWLlk+1ftV1=ig`z_SBWH85igm)L^zoR+zE5*`eU-XE)l6VD?Pf5Hzqn_LBO
zv3R2-v?T+non+{^`7|0n{<5{B_pAKz%!@J=G_#Q2(Rac)!Ik#_=!2Nx!Dr7GS9gN#
zpWjJa9bca1d`#?LOI5^+iBLRD1)m+sYOR+znpvwIeDOLTQ3J;uFF(IxiJ?q9kELt^
zxA~W`U(Y=}JXEp@Tn&ad?Vb@3M9<0#o=7JzBni2m^{rThkZ?^@SW+e@i~TLQS0Nzp
zN-W^UG;vb7y}~N%4mM$7LCP6zuQMMuDZ5mxH^G;7|5&j9R_XIhh76_;invw+CPZGJ
zdhX>t^?)+bf-I&M<-+Qd<OR<__+~+k8<oiG_&nwFni@61P}h~UQ*+&>Z7t-SC>wKt
zJ0E=pF@M*`10~kW?E<<%K|xe3EKsQX)!E?<VjlINgA!u1T4;17&%#55bOCAb?gr7f
zYS0rcVe`K^MowTXi&K<P$U?^~tANcYVl94TAd~U)M(O<fwT`5gDb*+aR%BaiYdCAe
z6nk1zV{W;wGRf~aBw$HV`g>({v7Bzdn5?i`*ZpHWY!C1R!pLC^5PLXCwiQfF%voHO
z7$<15dxZ0<>!P!_&tN2+jho`ps-iZTgfTHtlNYwvBs;QXoU&B>-<&1~IrGl)CU?>D
ziu0Ig&+i<+SYKRejmt5;)Hl-l7A95o{=1fU48kjNh5+QV0w3eo@89wIlZ9Lx-saj|
z-`rrw?@J0^gJpq)9QS7WR>eX{;;LQ$ZL}Dn>G{lE#KczoSU@m$d2vmEy}k72_|CH1
zcJV<GpF+xv`y6TVF$xw<V(){>{qz^xgfWV|Q5waeocz3zKAHw%eI|3yaeKW%Ih_MW
z{`!wK$NQ;iap{ygha{a2xr6O2X5Oi(R}Qa5=C;NlR4LmQ6QcQhz06YWwkPI}iH7Q`
zf0z{GE99`Nw}NX}JC%=xuj_`4oBhYMh5G<8&YIZm?V-CGx}D|5oHziEv!|dPSiW-p
zI*LH(8?{>B6dzZ-s!u0Hm<rwfQZrHq$|AMM&B;jLN`Yp+VS6yK;MGovldEgGV(#E3
zWCz@Cc$Aa_hszxrH4e<t(FlZIB_1Lk6o9a>uy&;tG7?AJBaQoA(GwQWZMeBtS41U|
zl9NQ1OL=3G>s_!nIQ;1PcjMp3!(3MEq`>D=@LAnS=I*Y`{^Sk=-G0jnj>xvZaRDDW
zf*?F?9@mJ<^gl`&KPK$m1fG_D=od-=C;l{Gxo@&O#B5xAg}ZKlz)I~NP%;*=kH9tt
zj9TiIhtD$D9BvDlX6s^}N46{VjkPW80`*`RkZd>9pE1!&9lvq+%lD9wjegg8Fm40N
zGCp78y3|3a2eGH$mT_2G052S^7vMAw2CR3Lp7HlwWC5!AGfHV6H;7qhij&7UfBV-?
zT=a`xjXDAc+NP1`PO&!29`0EN{sAxXL1@ItI2>}f<zr&n32pKQx1pAnmYKPEu0{!M
zohzb#`~{}<d|WEs`0nM|uB$B{M~P=T|2<fIL4gwK`*;|3z)jm6t8|&MfnQLRLmd38
zK#syiV<XZ^IQkNLH!oiH(F6a+f4+zBVC@laT!3cgCntmj!X5%tMD6H>1l@3x<#@>U
zJ3nD(?8m!DKQ=w{Qu7zZHw7IMg{!9f$O(A7cu@bFP!Q3JOOj04*xkTfHf@<{mmi2O
zvR!xH*=4^=NX>gZ{xs4Fe1ExO0eYEvqV_Rg9G1)5T}b(Ix6H~rd`!<;$*DV}wbU-z
z>5l5=;fSKTM(pvuE<<H?7s$Q$@!96bfZJ=FT}}Ib1%vIuQubSlJu4>Em#>1iS2S77
zWSaY@;^)jMIn2M_Hxxr(IJN6so_}3fSip2wy51N{k4;F>x3G}4E4`g>FEJW;W!OuE
zf`U>66&I(e<f*zLW*vFfqhGxP=>!dTw*mt~2r^w=U3Co&g$fHDv0$R6a4PY~doUiJ
zm`tgdS5^zJ?_XS0)HpqzU@=>L*n%rE6MUhnprF8m*gQWRo+nl2>~C%sRV~o&YW#+p
ztB{UPKtRCD%RBg6gToh>vOk{7wA+5h+Zc)yY!dj?>Yva9cJEjA|IT?l+{Eho`qE!&
z7^VE~8lGwO2YSp_4Epaf!f4!dartZBM5BLQoUI{I!E79D$8n+%Fg10L3qZ|X-T!#V
z#hW4nEC`$9^+}tkoXu?$>X&i}p;<W(#U3;Ya5dHkhZo+e9;moG&G>9&d``ZCJ~N<^
z$mHg&JbyN5bo31P%1;AsS+~6`iK$}AM-Oymh}PuN#NIa0r&!Yn*N>{u*7D+E*>W+3
zjf6$B6gn$_J|@BUbo1}a%Ij^BRG>{~)pEYOYY7heuG-JvexZ4Ojq=6b?Z6G5xVs}w
znSm!N`0_{hU1mNL6BEN|jBGX#$%?*%3k6b8j@SCqb?aPu<GF(J)CwaaBU{qm`Z-<g
zRvez4t&Qd?8Jtx)>}t-;%s2r~Mx*!>nzgm{%IfM9V&bkak=qX+K0E|R95oM5#i)mB
zp>9ZC9^-(>ogj2`N~l(-n+dkyi3xRXi<!>OaH_+@!z!w8F5AK?kYQqRaq;=Z#Za~!
z<?ZD*D+|0{`pp|;Vq#)40f%7c<F%m@Be`n3O~o~K{HT_dN4ju0Ts2=4t?^*7MM7Gd
z^YnjDJfz?iE-1M4$)44F%_UD%`BM@(-4qOBAqkS@wZBsX4M?xM5%tFF!wWRt&mSYy
zFs{i_8^rbX4WO=OL`SMJkqaBcxU0wjOo);6(y6IrHI`02-y7FNlw2$HU9$dOHFhR<
zN}UU{IIFhzrNd=Les|-IT54hT^oZAJFH?K}UD@awcmur@7qnmO>4b$Woxty4s>fmc
zWL;Wja2q>B_p6hyS`FR7Om4noT2xfn&5a-e^RX5?B!wFJa#57qoaFENtcN=Tk32x}
zo%B(X@!aN^jrHzV{(*t7BpPhjdiBl^7HR5}!^7W{820&xgfJKPu{K<bhEWK0c2$-5
z-sb=Pt?@QnHfit{E_3$@++tbavKyS~{|;==9ocj`>VgFX1?BxX_ckU<S_lEJKWqSJ
zB9tYQAf~BF0zk|4K?`mopRI3fY^>rq<8J%?Jes$Yd{SuOr;B9wONH2$8*I5f0wwy7
zcJ}fNvBh?L#8D^6skN6vbD^MB!qCxuA|33_rzT0cOB9tS=fUjX+^pS~LJ58FLb?Na
zwZ@_YNk)C&-Y%FFQ0nREu1Px8@FPJ=&|OfiXASIu?a^`E?YPLLDfsa%KreL}qf8uS
zny!mkj$w(JI?3#pJ34H<-T1?GFr*Rdr~;3yppLOwSmrpGw*Sq{tj@nkWujL6=wJ^G
z2%Mgd;v-pl?v>xMsbm}5AGl8j)(j!LDxPF2H;?3QulEc;qSP;Vb@2lR@{bQYFHT0}
zELp~XsRR}4cRZz_&`tB+IN2Ow=itz9@<an85ou#4T>}F!PR<iUV(_6veR0TL#!u7J
z)16nN)R3&~>^=|ueBTN<e=_pKgaj#dbs}<st#uh*s|6&f7?!&9l%5#Y<G)&#v)5IQ
z`*ZK!z5AY!5Tu&-K7Hsrl$h9iwmVG>X0WReBhRec+uE!zPmGaBxJ=%C`9c>L7x$%X
zD#2a5%EpIADtfTedfBFfPP%2ok?i*N)&K_?qqC<+f2A`V9g#N@b&w@-e@-tUmzj<0
z-nb?j37tkd5Vv$-PVq%nF~)&0%*!Dkl}5jttHu4j{sYT;@BkYXvfpxF4RJ1?bg-DX
zi!!AlH4iT+({3~Zj(8Qn1B}>RdH6viA_9)vgqJRxux1W1X3{U4ABNL7ftM?LTc?`@
z(5{juhS46c%8*Q#%k^`;=X4LGo=G|Gr&)(=Wpdx0WHu=yj#lz1*S<G!5~y~({hnTb
zRv1KrX^)qJ3#A!b(PDii(bfGO%NgoBh8i4n_~N6!ao5`B2+>7L*gbl3bC-bgR7c^D
z79nFq32RVfraTO*bVEuue%|~Ocdo6eeNA{D(C&g{E+-pJ;HbG+O8_kc(@GF){Q~$x
zYV7F=%1(9(y#InFJE2%wwopn^2)Ue;$JYb@!`cHhv?l}vVpdjc9buG4BUv)qH4fPD
z`Dx1AP)ZRBBp3|FqEqdip1y2esHLenT<dIGZZRuaVKF;V>&y{JEAtb&yR9fA6>!+;
zTP<w3jdz@LUN0Z->+Q92aLCf{2<3?^TGb2WwPeOBHW_0u8O^a4rv$5aesy)c1O>51
z3yA5F(6SEw_YX2yRIqbOHyo&=^)QxNqkmMghE!U=hVJrJSP4_P2z!+cYr_ll=2T)m
z-oRVzhPVJ;s7VI{l}Asr39q#?BWkR+`D_Z*#lxPbq#xAO*KYa;6!?^1p7M~f?`YmT
z0$)N&9eK0WAAK;-vKwdepi9kSxX!3tmIEyoE1{8Ly>Rfdt&zwuuo)1TaM{iiw<T+E
zY)q6s3RDicebJ5^qc`PzzW3^PpjV}vC#q?Q!`5e2+2sGabY6I~6>F3j-d{-2I9p!1
zo0%~M5waQ{EHo+SsfK{&A`y#_<;UdwVA@-$#3Gq|)sl}*o(YhM<y`GW`8XC9mPV};
zI}(RsZzYwn3uC5YS8D)1vVU-JEwe^^N(wkiu2EluwILRAv5T{g@ylI0IX!Lbl@@uu
z2u=dp5QW&#Au1=t?zB-i@BgrurrOfd>s&UER>NYF%Tv~lDB-HVHw%EjbhNDR2OeO<
z@k9JXsMaP5G=rmt41!5X#u_}|3GBC(<6LSgK=-B{5QM?VaqxTM0n43$xMQJ`W0ljV
z()j1nNU9|$Wr~nFac9R^REk*G=f9+d#Z)gnN7T}R_SGhu-ww2*zo%Qnza8=!oJ032
z?%A!?AY%|ERLpfvNBA>cLRzY$t3R4#4w?g_+RsEBE3QD>B`P=y`pAB;C3kjrzhq(x
zrV<Y=G8&*L2lKuhFRH+ICgTNtzSyK2Q)Q+B?%2eyGJ4IQW8va9XsjWfNUSdLms!>x
zqNj_6ROefmAKD!9j0Ob;TGU7X_~G3b;`f!Eot+1|>lfIMt(~jOT`Daq`UrJJ(9|9@
zV?8G0{T|0*bhH%2I{Zt8(PZqG4`>i2mI>O~`J1MHzQ<B(zweG79?F9Y11#KM>waCA
zIo_psk$i2nLxuHJ<1+R2Z9XM?7qgIhMIP7$vNyWA1`3_Ply&hrnZeBIYR1z1v|15&
z9*UKGc=V4D0EwdF7BeA{-%P#tra(Er&Z~=~BP*NF^4Xy)YpZEBhfNbx)t<N(;Me{L
z%Fk!sovz3R&1x-T&K30hEZUWgzcotE4?8H4oz~)vK1fTqc5Uv>#~vOZGuo3%W=3e1
znOHA2qp|4LI)Mc6eo7T}^)K=l!=KP9MmNI3?&eXis_nNSk{Q1=>F2OUfD(<TO6lPq
z7c`)3X~AB<X3E0CQhVGlP<y$R=LGDff>wIZaxYoEbN+y-QlOEq+h_|)5EJuGPA#qP
z-5YS!bH)$?g4Vd3cKovNke6V7W=UzJsJtjOJ`d}K{r1PZuO3CR1O00UleQq>GBPqH
z%nypB0@_mU`nKtCHUf<VTf9G+ih=#xk6tG2ot>P$%{Ra}YeVBr@jhmG>B9C3rym|h
zTak(c;X3kjn@X;)^~p0}7usY3_ikw!4UK@5$P=aavX;>vHcT+BX`8}dW1Ie9_9%$G
z7z}9XS?Y&`<uTsB6LUS!);0I<Ka!z08&%-VWXr27duK=DW20Q>uI{Q7crf*iflsqy
z^gL>|k|{1Tg-U#yl*+LrsIkm*x_kJekpT_j4`3z5<AFidVRx!Gj>8Ys4Et@v&a9)G
z41%=raG-BaARNz=NWx4sN3+aG;n%2YI<t!8(66b{y@l*wz{~2oU)KiF`hWQS;q(5A
z;mq;XG9Ke{0iN4@p_;C2#QsI{`iw0;&wzXXF}XnW@83*+^R+s_GTF|GCpp_5xn<_D
zB*VQV=f`SN8+cQW&dv-shclUUYhy8RDL|M)<KAqI#-GoRzy@rvGf~W>RUQn`?dEBv
zr^`w50=bBX+gaq7VJJfatvu|%hx-K>;R@35lIYWpQ2eCt=iP|>-acck=`hM_`=%Oi
z0>&gP%aHSx&uiury&_jQR;(|VqV`EiiHCDco{QOUN_%zlAqE-DaCEABgYCEZ>^3pu
z_-Z$f=$1qa4Q0;D^<9%9#=qgdjjMKH){;%4W%qmTyZih1c?P7b<FWA<4VREGKkT^R
zK4D*{B-)v@0x#yrNjJi%aNfH({0VlR@><G%L(62pYQA$&4KKbe>`C)*YEIl4@|;}M
zx~=M+>OtqZli}~1urZ9V177a4sQrXEqLG;jbP>B<#f{6QwRaKJ<;nalGuR2d&O%5k
zI9Mmqtb4Kz(NN#TFLTpgY3ZXRr8If@naO}uaXq|9WE}q5fLQksoBp|ICzsv5Y)`+c
zvyK31KhEcEQvT&gp8q}cJW$@Bt1BTS5b(HmDtI(^h1gg6rMwsP<Cijn#e6;XLHEy3
zECY28!Zs@pPq)X%8UC*4=~&KIXCeevPoJ;%fS5mejXw}IfHfJQ1ME%U3FEU}W3yTL
z;DXqb0g-gaZe~87X^02=p!X>0sUMU1Yf^H*QWO#qv2`9)BCPO$S5&Az-(Fbgh7>s#
z2FFF=Z_-=0`k>)C9OIedkQ?qUQa#>1XCJvdeKIDspV|kLW&b#5*(xY=6jSzqmC=gr
zTPYLtP$|OtMRMIO;lwudEleb*Z+;%zRh3)~JRgsC*zvD+d5Xs$wr^2SMy($3D6w&Q
z$J$kfx7MXTq%BbIyUw515HzDR-#$0COJs@)Oo_g20okElin)y+kn0_6UUp4AeERAI
zwNEAEqw6|T+`Aot#)$zDed%$QefRmkudWDVi1*WI$$DbQ6XgmtH4Zj~&VYvPfY06;
zMf=xif%dyiD|12@^-|ex>lTXn3^Dr;S~k@;ay0BWv1EH5P0^0!{o-=#+N*o4Rq4Mn
z%2{PI(LvI!(QW=9qU1l$s8Xy)W_-@^%rLo;8V!68%IoKASr@SP^WGqpUpR%^Yk9n&
z2Z9%IL>DSNQM#z2NWvM%ud?5NG0%FLFwbqSx4IGTrd}*sZ_87lsv>)k9N-hpU%MIZ
z9`b9|EpbalqujAUupk)+t6GRpR-Y>c>t9g_clP%7&dGMAbuQ+DWgSX=<E7mIt#(K4
z?3#?}UW?4m&I-A@_Qi8)!eoP<GeXgIH`|2`M;GfU+hq}IaO<^>#>TIUEq)xJ5RkYn
z=X#6uS}<4CMfctXp})by^Z}Q-zGUssR^yAuA=0^{HqWBs!&*nT#(Z~)yQwWG5(lD^
zMq+F9IAXlsPH^7BnELE;0`W1Uh`0Y9x6Jpf$dL%V8CbG;d14Uvlq}TO0K%QHIpRb{
z_?5U$Wr(VUlRUkRCD3oTSyzZn^xsk<BjNSb!(m;Z$ZQh94G<=J-|;B&A`^o9Lc?$J
zmopefHaV>^U9Shy*?7EJ+S0u8YOO|lB3|ja6UC7>pF3Yj7WmE>wB<Y@BCShd#db(K
z&b}FlToDKijO!ZTc#Gkz2bIRm%K5C~%995z`qA`MK7HEUJw$b!co*qn9|x-349Cx*
zz~+^r($J<A8>T)LmFC5&&f*J+r&PDvy@d(oo$G_;5A7JyL;0vJMpTMmPgMm!PtM(F
z*SUrzDbX%{(a(1dM%$3IdYJD)ChjEt``Y*Y{8Ra9;q_?3Ci<ciWC(ZQ-O$Pv<+<s(
zybI?A-N1#w-@Kj_0f(k`wl8~$$7KU@u}?!gKI$2MuS>=J|7HTQ5fVO_lh)`-TkoiM
zY0J=H54m>fm#}d9E^8EWk?YsTH)wD09xRe`8Fc^H0V<dc9CraA<4>T4<vXhf$oL+c
zciz@&lt6~8Zu7mp!Ez!uKkT=`!v_~d!Ko}(=W@y!IyO4`qd!UDcpkc2+}|Hafo{7%
zHE|}xcrF8vn|-jTsi_C=%L{*S05;OXrH-UJNO?-y`8+jytfu{EcGiEj6&~lV#CbSm
z&BJp^fAwV-uDVXhB5RP^vs9H$_yQf4Sw2E)r6?Gzx>@_bt5{Fip3dp!frmZ$HN>QW
zTRo+_Dd!a*%<+wE1edj^PgH6ac?G5Ku*b~12*nXy%R%Q4(lV>d*7z$c1t~(;8pOgj
z-F#J8N4r<)J&j=oP4AizLAQpp;rbU8;t|y7{Z}DEx9d%_WX2ved}%RCHiRpfOLI_L
zxJUlO_tF~B2`?4lO@2#Uechh7&ye-?XTB<}`6st(8n-+Z58#h%FRJ*4@q?dTyKX#O
zzSD?XXE$+<%8s+Nf2h{nkQR?YkBqhdO|fl!cWS{b!O_&eij?OVFwJzoLq1mXJf@Gb
z1xT#*wJZ|r@5ghhiMX{gF<m2_P6ZIknv5$MsrYnCdr#MFRzF+A)6p)PDlGXa6xt}i
z7nz;!+6YV5tv7~O1;Y*%#u(h>n=tk=n`xPE*+3wG;|ryw8ae)zsZVWl-GP?E78t|~
z=kr3&A~N~7UW1$@s=2>a?f-zY;=kIy>+9=SE>nVzP;x4OY)eZ^cN%V}C7_q=?BTZS
z{X<zYq=5b@+_{;muuNI6Rr#f?2Q~?67!1{BU<&Y2irKQ{{V5{2ub10H4i65T4CaYB
z3_n|r7s6BYT70dy|1$rOG&mbC`SZs=C}?@6%J%xGhqX6Z=qkz?;AybStj56acNtso
z`W?YIY6(z1R!gnWqs!Cn81lU8lP5-^q9{O~`Dxe}Z#G|_l#`RQ(*e&_wps0d9?xkU
zAyh9J#X!<Vx6wf=chD|8j#Xeb**uSs!FzBs4K2Pb8Mg!kKr{gr&ca68T=(|h{!C-O
zBRthm3Nu~@&$gPgXI2@aZ)zx?NZeI5c#3}GEcks{!KwujH+dh)Px|J)a3rpCl4wHz
z_3a77Lc1{Me!+coxY{V%=kOAnEZl<$3v+|Vl>$T^JLABe7w|xtqQmfGS5fy&R7+#Z
zzY+`6nFXKg_mmqkQ_oS!G~B(BHT+u~gTHNPQZHz-^>{u8`D6E1*#qHT0c(L0NTgR<
zW+BDTiW=JAzCd}TPQR}-+7-^OQ?tT%eU$W3?5^yAPcibaGHZ@W`kXac@aktPdMq4x
zeKV`+cpjxbGASE^=fWs896$KrN96F<7p({N^zWl3-;d{VeCS+Ol>r>g%$u84hO>sf
zo_vx~8swn;rZ{!Us;XDNmF83s^NyR+j|d5QpP)iL6knoOgr1;E?dyLs>)rVVf7bbv
z>HZ{<k4+SEjRv%@VuOc!AOUk=RTUpPIyx}&JbwJR2++oW1q8T`$~6VBWJ}A-k=uza
zzSvYkLP>xS)-y2h5lw6C?DX%BVqD(e$L*{$jA7A<Uj{C83IGL}u8sk<xVX9s`NYK$
zWBR;w6?S!X1qikc@CVK(8$*o8PA)F>-{p1cT)2P^0yweHDJh@{q4YR^Ar<}QmM64+
z=>acvD+rQ}+a@RZt`2!PIMP9d00s>m$<%b)4S!p-X^_5&^x3mk(7C|RojjuIwJI#?
zj&8Ck0e7%;ez?r~y8uGiOx6HUBOFD|e%Y0OZtrr9oA7d#b85bKDh?g>FA|$xn~7qw
zD@@VpN?N@b)4&DY&cQ3*A%52^<7=^jf;!RBc>nASPTz%3*S6jAEh@-`3Tp0t*#4uh
z&V+}jY&R{PG*^-)QnjGpKK2!rmvxU?XG$<b`UQ%DMb?LwRj7^1L;a8R-t1MfI&j}3
zs9rYBse9~j_C;~od8__Enyvz>s;=82DJ>x#2Hh>43X;+#-O}A~qy+?&l<w}5ZV*Ab
zyHmOw-p22LV_e1%<;p!f)?RC_Ip=12ANls;6Z<iVW_lLVA91uep?G^@RsQHg?vuzQ
z1c&|Bkf0y&31x1#AFh7N5mXux(n*T`RU_&B_{pBjzU$({daHa}-@^nBA$jTaJ;Kt|
zxWz8FgQ~R+TVvJ6WL`EZhduO)PA~=Bn!C<ZC0<v-K*CF)jQthW-zj3v`Kg=aZ~y$7
z<rbErzW!-8eXV9xiWv=!3b*zT>ObP!6XbtjQPjxxNas*QBojQvW*`%crc22_S&(=J
zFY;_bEK?%F1i*4`;e=e41OVIqb9OT^F_9z$rL1&0=#Di2baZ#I1tJIYUTla!An3_$
zF;PTY)-dA~RtX9P2Cdp>p`JR6wY9ajM=M>P-N*=Hx$vbC5ttyC@mLl+n<EbZ9g0%V
zhwY#^K$iMDZaDkXYI?Qh{@Oyj9SeS$(Q)IKEEB?fTe_=?)OcS59pf9O6ek*mVu58=
zR*;uT4Pv2`N<*IZY~X-`l@_}!%5#pIB7E&+|5`fR;Amj}8eJY0cyCZr&+s<*6vsAM
zrB-QNnl3)UR#I{xsaWmPAbb0^SMrF?T-BnvC>+jJ1IzY6P@+_8+&a8wXl?wVmsmSA
zUswCCf^1)}XlW%`FLLDA8gnH7KB$P4PrPJJ`{>!V4it!fymvmWlB9t>CmBoc<)Py*
z@Ey0j#`J+%DPQNOOv=e3GOA(e143P?1)tj9K=(YFX8pIMgkQaPK~;%j5qOqfieaPk
zq%zf>+cUY`1T^ZkUZ%}9dZ*?S{@JgC3@y!`h8}d_$E`2)==YjV(kS`18mS)(H;YST
z*ts~b9709gm#)$B^jW>x(ss5RjsQYk62?7kq)<CSr2I4Z<XglAjqHBS>;X1|*XR6l
zR(gqutXT>4Y7Yi5$zYJ9qN6*vM{}=02YY*>n0UNErDI?K0}c)jWTAQT>9E=j&c<CK
z&jE>p&1!}N?L@=>W4b(x<CZ*t1%AirdPBj9iNt|{fxwvD?-9NtA}UZV^Z~R695S*2
zkbG=US5ku{rfq2G8Gxwx`T5P}>p1~PwX{23DU&KFFzF_Ev&){5nfX|1?#)zp^z<N$
ziHRj8C7F$6Ap;biih=@OT6zyJdV*38`~xBe$$Ljfj)(gjW>Q|qR)GJIfe$Mc$0YlR
zCsr7vVv-AXj*co7Uc%2m4~bE3Y;05qxR`P8+MPrwh!ZKfHGXmVqi_MDHy260W7}<a
zp^4MZf6=swImeV7+jpwEA7JA(G~{O`t?SZ?e+?WCiv3Ix>|nmT`O7TKXz4MtpH?1q
z9;Hq!!pJug@wO5Ibq24u`bEsk*)uy5grQDCYpU|SQ$ivYuL}=T;{s)P0e6a2kJqGR
zd=fZ$CZUd8;a7<@tbZ#9+%O)pbhI0#aZT4UTGmr~poJ(QwCBgfB*F31!EXaZMIv&-
z;^%Vl-;ti!Xhhy=nJwkbi&}r+boSkgn05x<^MY;;HPcYGsY%7`L!&3t*|_C&B#xR=
zo}TR5xa+_|<2zFF1OA{3vtz6nDR-9RQcC-S1q7G%2$GWT;G$@+kJWmjrTev0pPV;+
zZ0ahK5+CS=81I7CJuWgeNl^;L);F`NziYK6pB4GmN}iebkqkvhWlNbml7Mw1;+m+T
zskySb83d)KlKUZHPVCHqV#OH1{S4}5!aF=NGBRe<6_ku0e~!`OSy)(HpKgn#3VMw#
zba!-!?NysOhK+)x6~Mwn*`dw>8lr6@v~3;$eo#w!A49J;P_okGd1YU6kN{3A&}jo5
zg3g(G+u{+1r|rPzjpiwgTB$g1|LO*1DHQI#Emt<cifr>MJB{yHe7q4qqBky)7MFGX
zyhC0wT8crYT0LC{6I)U7DKcstzO2JAYoPg0TZ#l}kv1MGSLVMv<sKu2XI@u4^`6uO
zF;c#4u`_DwLMr2U1HbG}XE?>byZGX_c4xx#<;X0}6$^8b3iULj;v<KJyyBuoHieLo
z_(O2h-}MtY7CgLMpPwJcSe+jH9vd_F!R@0@{*yT0g<Xf26#Pk@(YSeL$v3_}e+$gv
z)NpDE9H|%KnT5yMQg)LC<450ES8QKOVrbR(b^NBmj*POibIgh^q-V6adJH(O^>v>}
zV+)%EAawHH3vJ%Y`z<@?n;RT#JV?&SliyjD3)jIpL$bQSG4CUi(1DuTHQwX)2CwA%
zgr0o;>hyGdXWtr-iq*?7KW+DT!62^F`D;Yy4qjYNhGCb5t(t;?SDaiQ*`_r!cXD9*
z?L|+L%u;z|*$;wbZ07o*n=2T7Rvw|24g-U=UQrLy8)t!FPJVH}u54(*;LZMg+?%9R
z+prXV=Lk;EUGrFGiT>mDA2X#j`>bi*o5?<~Bwr<;O1ov+L>oQGxlD8%Rl9oA<D&e%
zbTy^+XRoYyuSUmO!y=Jx$G+4U!>Ag~sx4$JGx_0+xAZaf1xU&Ms7Gj^$Ho4(hZH!P
z^xbbx)Fan0my2VXpl6MOOsk`_vrevo`4p15o2EWV5wAACoW`2D4|@Bfr3j{!E)z8|
zg}`!CYxCdnjGWuKISaZ!&H1MF+9{u6d)lGub42DJPVE22Yk|Ls+I4<GK0ErwQyXeE
zBo9!3Qu1s(F_{?99E?gKk}z>cvEJ`GB}hKxt>fX}E`<oQ$Gj@?PVqOD#YCl0b|T13
zXSdn48oh!Jr_^y#Xben0oK+Z`ee8~qI}_@hnbRu1N=g1u|7*ErrN=OG_Ntb53Y_kk
z<%COVV$W@i+|NA^{EjJ8ib6>8gEr?zyXCr7qMg>yPiNj`i_wy*vZzWvi}ESX&K4hU
zP&h&^;)Cp?_e_Cn=_M1EogD!o(TcfEM5>G1%*0Wq*gPf+yFzLE)Y;X^u!{1sw>Q#U
z%ODgKQNEKT)TX}-02%`;YdJepQ^5k&01R?(;XA44PI8@s;bDe*%b6h4h~05Hz{0{0
z3NB~0#yB54KhK#*^7gzHA(i=x(xpiHI5iOXBI=}n1Ez&$(cXC4<^Gla8_bn01??R-
zvUzTpZwDwRqRTN}drVZq4)}Z2?$L-?QjH-EhSmuXwj`xw-(p2rPvHY4ui*0Z8_Tta
zSLEXFj#%hIigV@IHx7(%$C~JMe_AEm7$<`ZT2tQ5*PhUzk<q7r!KnG68@XDb3Q=vU
ze6omU>=^&nU`tCTIHF%(7J3Au41K^8C|=PZ0*8b0OWJ3tPk!=JbTjuvX#3H{oZ==s
z7sN4kgxn#myPBK-T$)6MjeMc~j??B$Ra9=U&uF9mtWT@PCu>akDio!4;zyB8um={_
zT<8})P>tfc*qPWLq&VK>={NWELL((5Q|g-EY}t_&D;;+_67O!Av<mrx#NUCBpIyY`
zxHdh5CvG45wbO|5$;APV?Smj>IsGf>IX#Q1JXCyNDy@>kZq={3VLc~PaNLtf^r1vS
zGbiNRVYk1(kV$`fN3^|zv3rZ4r{ckQ%HrdQzj!LToOu^{2L9ka^b0D*gAvKS7l~BX
z9K>-nxI#Tawn;T<9%toZ^H(ZAMmd|A!ABNOobkTGkbN(izIrK%cC+{@0|y(_9x_M1
zR@Hh<rMKn9crxd{h~aO$y4fA?MR`9YiLQ)NsFKYHhUr7m=bJdG$fheSoLo8J9N@v|
z5u9xLK0_`6d;aI91}ctvFuY)5VPi?tD$h8sBOh8OdY9eN<q8~V{bqhneDs|8?af3I
z66m-G<qF>L^gg(lzttj<mAGpi1Q$1C$@s-Gv5X=)U$)f~l?~%_P5Rrdr^S+DzIwII
zX~+e)c^A(1`t@L4SlCxHjZu6yN^X+kyDNSvza2?nW2HD+2sNIcY)$8XCLSZ<j~++D
zd_7GdQo2%>S8+Ra_Yv;rgs;lD{(D;6H=CE$J7agw@PN@Yxy232Kn@btkbH}Y$b^_D
zR-vs)D9k#UVa+T*8I)@%DHxKNqH>S+e1&@Ydg;<C@j-0xpDc7g`};o9pF*<MqU=lY
zi$?TQownIbNCK0Mo%5zcD_vgC!Xh*`mpX<{rG0P^laG(IKn@&~7GuYBAHwX7dB|O3
zmPQK?i$=v(dB_$8fBDhR(r&DEc9W0Mex?bSZ2Q&jn<ATdcu^uy#*2&SZYC%xgBTLF
zOh81)mge+_KPbo4<oKvIn54lSJtUzgIE*c}6oZ2k`mE`EEU}rv!t?DBr(E8Y+)d!c
zTIIJkBpa%yY_VC8nt&qGs!^vtDZH%z`!dUTI9JvI)C2Y;BfX_mWbd?UU0%X6b&5HC
z?2m9@AcBp`S1zD9qf;64e5e__Oqsv3>~*~2nd~Dhs&f5Qb^F9f>N1u&C`Myoa?6Q!
zUDF>@tv0VX+2xr!Cg8Og=H2n8+YRoVY4;_QCEDcSVC9U-$@E^kp7%HgLt@+}5Y*k;
z{!GyEn-CLa>)0&AbW<GIWU66Tf`d{EZR9n^Anuf{q~B6~jmJAfmL~g5apctSvoyQ3
zn3rh|BV&7uTv93Lj|-#c_DdnTL8TYW7|`U;r&aboVarAg%^zB(PhhsjizJ)7gJq>w
zGQVf8Hy}3_>&a-?_AiIS(!b6-pn;OY#Ezt&fjua0uSpad==IwD6y@Kx!ZmRJa9E<<
z5c}1$fOTv*Uy0iLX7A1LsM@myJK)B(edj*vI2MtldBzz|##u)HH}m{H|7aN^(HVRG
z8M&#x&`G*k-ZgAzAq=d-sCrg3x>;dTba!KuJUxRKispE_KW!vFy_w_>f?Bq005mlt
zTEbPOy4uodAA2Z4Nr2V5{6~%8<burv{r1(x@_yH4hOxu_TsbP$V&L||aq;9|UNX4E
zj&hDStKEY_f1G{<f5(ctIn9E)Pc>LpF>V#Gk~l6kp2~``GQ%VPm?(8gm;Bs>k^C2W
zpWhPt4EvrS2ca$NSTC*Z>L3(UC?Ky=4JXvF^)c5oU)0$>68bPPf+XLD;#oAl$U})r
zh3G3~aB7-tSJmUO$ZRqg71f|(sTUdN-ZNRi1t2a@k1!{RaD^sILP*^4o?>#@$+Pk5
z(r*ToWke>3ZBzCM#Y1&_tR!JMUz!9Koh?h-!V##cSifVeQup-Cr|cirdE9Rr<$c3B
zr|etW6h^$eMv9S&>8sTCqmXn=<x-N{zn<<<8T)0fm8t3czD3Pne^R5~l$^6S##la4
zl_)v$UmWy3yxxME-igiEJLP|@VQOw}X7-%#?DPjPghH_<XP-`i^&@c&B&J<cmm0>P
zkFf)guceLt#A^(NXa1r&0FV2+eN9XFDR?(!{>Sg7QS-gHL+)?d?8K4S#8Q*u=-cBQ
z2U6_mhx8k3zelh-gkig)XKmm2&u-wb>+do?eV^qHO|!{t&(aSrPaZ57-cRh=cg=Jj
z|D|+Et#GS|$i`{`7dQ6<#S@93p|B8L920^Xw%#1$d&Lu`I5-8G9E2xjo@*{wOavDi
z9f+sa6YvHj`fp+8{<H+Z!}#>Du))yMMlGj0roEG0yuEPi-c4nd)sg?Jm6S9mqjC~^
znvBH)Q4@UbY4TXnc)t)`qFz8ziF6hu^E`yy#s^Ar=M&HF_zhE=Dm6+pep#Z_%^yTw
z?--BXE%!N;_EuT0vRBg-=W6r!`?k)QLga|Beuv9;o1i6yPK5BT)~eYw<;**OcuO=t
zp?w1*qu=%AhW5q2(0op*hJ=}Nl@?>6oc)Cx<b8hl4w;ROov`_^)YQwvp$rs1=a;EL
z1}66FCA>ZJxgiN{gX_ZSOWoOjz^)v5@&3QWblGXd!}kJ^8A!No0Y${^dmO;6v7bE?
z24pIHCaooaPiJLi*#caWh|BWn=j3Ekz|9f`a3bhz?XKC1iz5Q8@s$9ijY&_9>MNfo
zyEX3D%$V?<h<so`_pigoqK#x2Ca`LDmm-P#DQmRTNz@s%nDI_l*Y0MEFH7kTbA3%Y
zkU~vau9<O>EhP%=`EJPPCu&Ou{DNeZQ{3JRcsXv&W?WAFnmn{4#XpjgMxqO89ce?P
zB7gieu}TqR?C+c6*C$zrLL2@B|9qmR_ragULk=LjS;Ci;Pv7Nqmg-_=c+KB}8-H+7
zHlUO0IiU*$qL4PV^@%5SXYApBKDX?r6F8G*j9fL^z0UC(re0ia4CN20E%WMy|N0dT
z#51(UR~KvjeUx_2)P&yk2q(TTr8CqjslHwpo$R3VlIzx3bmTEX$=@T{m%BgOzq&4G
z+id$pH}#Av^<vM++{2^%I!J+laHA=0I!nc3Ev&h#5}$JHOD<&<t*BH&q#H5UCePUP
zM64BH1o!nQ?04~Pp^$_H7J>2*eoo4l^^M-Lz*ls14Z&l+zpBa)``>(VqkjA*|5)P}
z`~>ZIh@tLA8A*)>V!Bupa6i*5>7Xk1v=2jE8@d-E6`UCAi@(e=+OU}}d6)KI($c<@
z_U+hxRu@%6Lu0H&n;3$izhn6wfHB6>w~h%9Zny=&GLPpa3#k7#z{OewgoNJ#S1(&C
zrq*c(4~F=SbruklT%##jL{caK!9erFy@%l4DKz?qG<>4seBX8SB`y&*c0O(Xj~@ve
zH{C_@WfYPt5sb5!c0Kl>4g?*3T5Y7Dl^E7ad+~T|X{{-AvPE8BshHsJ=`UaNx)_O;
zk`WMX-FvIJi}S`4+OzeIjd4saYPau{1E}bF@Gk7j6!mxzC1dL!6oBXYE2}5W{W+0@
zs@HST|CS%lzMN85U*oaSbmwJCZutHxy8~*TIXi*Bs!hE#G&GmHtRJ^WFe6)uiXsHi
zZ`}%3>)f+~>Cc7nBw{j%bMN~~8W|0220`z1yf#CXQc^~<JYm~n^85FA%EbxoVIuy%
zb*XiKnWvIIJCZLXe<TpFr<=Mo?~ENmO;8jC4ae7O;Jb@;vBqQi>FKy-`j`Bquc4Ab
zJHdx6=rCepa3<#m-ET+*z5Nt6t=4_~(6kS!k0g7J@q0CU$(?`M^92r33rP^tft(Fa
zvs!N;4vvL$ph4FWBx+rPJ3dA>@D}D~aar&@+jFb0vghY&s~r`Z2iv;K?M|UvV>~1#
zoyZnFV~Pio=d3Y>N|~d}ni-9GMDo4;G0!u_<!5K5|G?-*$H%;@R}8KT;6Y*h;6Cmn
zcACISoWR%Nk}`u-dB=K-*A7}QhdYygI#=iO-4V2-m8n|64OIo_lUq7;r!6wi2S<{0
z4-?Oinhh04N4_Z^6FcEu-L^ol8*da9l~}pB1}d3b4$s*wV=qJtXlZFd@PN^?vI_eo
zh=5c&3Yebl)0L*vpc$Ho@$Q*FY{tcd50~ZC=av>B80h}WGzA$OlnQuET3-MdsaB%3
z44P)ypyDJTCI&n!SfV$vHZ?k6*`)*hzOTG37%8!tnoI{j$h}YfhWvq)M{X)P>g^oQ
zRXeu_J-y(48dqk6bGpoMyPb4Z8dQ8|j#cZ(3evG&p&Lxw8*7?5SvUQ!@xiRJJC|$)
z(x%lo-1j~;l9JL<Hwxy7a|k>lAG-qmPmZ)|Nuv*rtQZ+J_?*$GOOaR%=iSe>Yrh<?
z71m_O0R{pzm=i`g(yZMa^P7iNp60S{bMw@4!e94ZyZaXob#Pr0OdO65!KQ(Y#HKs>
zu~*!Hy-JpFyn!{pW&;QTZ%+QSJYW%$Ghc{!2j>T~_FIhU^?cEv(pWEJ(S146bR<!A
zUf8w!z*V-fgtb2({&}{C`76jlGG*sAzuxaE(ZxV?`4f#2C0a3sOLx1~A4q(YelQw~
z*uE#Yc=d`#BkTP_$KKwQfEBjKpswD)wJb@UeN3NWsZJ(9nztu22z9H>zaJzrm(I3H
z$E~9ap}T+G%=#&Ci+9X0eRpifq~pyc7iXc|PNF4S=Slq|?-D-A(#r9N%s*&DgyyL`
z1a&ol<9^qQx%Bp7ib-i>y8T(9YH<dbIhZlP!@Ai4l*CCxf8Q6l&L!vPeS{pJHQ;U>
zIJx~BB3^$9i2%OoAu9s|%Ji$W1;vXvsv9Pc34vj~ckkYi#2f}XB|jCOG&_3;FRy&2
zmPS}q8s@9l>*y{H^KcB-7@MnZ1P~#@!r#4P5d?Cs4+Ih*rNp;JI-=nX#LbR;b?zMP
z;4(Rzh?lJuS_UG-E$bRR(>(Hqh!F)VaxnMIlzhb()EEKj7PZj)Pb;(Gdsl$x(r#d&
z?y*HvLoP`1F8>h%3M){_*2zY(;e8u6^-+Vh?gKN`*?v#h4dzn}RBGV=hCz&q8p6Y)
z*kl>+N$;$C{^D2?vh=zosY?`jkCOg<_=?I^#GAdW&IUc<X5xh)PdsXk-_y`wN&$A9
zH%K0h($Q)Dp_Th~&%f<1igY=7?k35!_I(ag;JKiK0D;AEm#;`ToKs8EOgZ{JSo)8N
z7p^GCl*t4=FQ?bcA!QkU-Q}Cx=GfRlIs-Bx@1nRVQ1vtr(BLW`G01{lYr~71{*2~h
z5sh)^^KrM7+T5CgkK4;Ft<(-?m6zYL;|U+D4S_(<NATif2}ARGFWzPO0w>V*qkUS$
z%U0aiwlK2v4=(Q^4wv!(+7a>yY=C$Vxs~k3Vye1Bu9f|oH>s~bN=p-|fO$tNRX9I*
zAuRwfF>#7_+27-1&U%Sck9w{jwDIGTk_LtUmwm_%dH7CV)}eg!2Puw~)d@-J?Hua6
zRjmREOUt*};2Ig<H1dHD54s_PLtYi!ZW#D{eDj;w{;+sh_}yd=sPIB4QS`2ij5N^3
z+S^RJn4=?_bOp?fT{^j@dy@6Lgd<}Oy63rY*7kpYp|@uk{{G#Y@ui9MIiTv;RTrqo
z@2J;l3hP{wu4Q!)BVs>`^J$7v33sXH7Hkr?yKWF4xN0nUEIPq!0Sy3BLf2cb?{5x-
z9{C(Hr8-T=I>Ud{{^JrrBFZE!UN@D`-M|lJ?(U6^nbPNYcoe!V7CuhZT4q&ZVIx*n
zkW>}lF-f-8NlUTTvjyCT*4f*tUGj6?+PBsgJF&E?(d6~c%O}6W#x$$gZZ@~%q!!>#
zUJE34h=rY_!ARzR0(NBVEmjLsj-I07hB9GcS49eNzs|M-gP_n*Yu3Nhs1eDz7y?P;
zS)w^_d$hbCqMgS_ITa<JCHVrVxeL{92le06_+P~D{~2g88cH)QygWZxOx>9*-Q?wB
zVfnVu<Y5ZkkE`h#+blMDyzyud@k4YEQvgHY3aK11gjFUpYJC_fu3+RPwWeC$*b?qi
zq&$0`@&ekDHBrUuz-P)h|E9mo2VwOrJk_jgt;yE+<RREbo)7WKr4&hz*y6KbItvV=
z;GpAOtRq$~VPMvAX2OW3+w(({3+$FEh&jso;&mT$xM_{Ba?&O24V183op1BZ;qIhb
zF^!jV?GK$K&6d85yze@0x;fhC`_|&JH!Ro;I|R+u1UA~OjCU<_92-4=VtC_n573Hw
zynYULikb`rdsV|zh<GA8(O>IXl?0{y`7asrUD$5&tx?DnuQD6<oo{f70+|lzvTP4$
zB7yI6-y&a~w*VGinfsY}Ea<v6>4Mzo(@2)&`6lRq_Al%YrTaB^;65%7NGT8z(GeJ6
z2>rD5gUH)nVD3>}Utql-$`<h|SC5z9$}%+>c|-+{gd#v&Gyj8~**2^=55JlReh@u!
z*&&&!%|g4)?;4UMo@D#|F8k!g-np+ooo=UWci}D3g9bFuyLCuWgyRCwPDtx_iiA>S
zl0@Cb5k2)VbzRpQKTkp3CEucvcgVgsGkx_r@!s5D?UwmFuhc+rRs>c$5UoA);>Gh-
zN~;Rkh{`f)`(+EmJNdxU{Zj5H7e}!|GGKJ*tHD=|AItgeT2P=c{X<&lEe^upkox{Y
zORV-Defji&wn^;}<R)BJGpC{z&S0E4Y!*mtY-~Vp#|)aO+D7e}2iQ!x)hriUN&{sX
z;H?&YtO*BfnB|?F!UGmSrSZR*${XZ0v^J-s&Vj||OkL7^WeI)Ft|J7E^sEOM1QCf2
zyjxo8^SS=-_jxqb7xq&7f-2L98!DfF$@zD8j@)ks58}c7=^7Le>K0BHE@xA=#!t(~
zVYOv|5CKTsY(7afnEvEq|73b>+uptgLcXl?900XX(bqH^<RANC?Y&<<Sxl5$9(eZ~
zGQe!)P~8qyht8!KShHkVs6SzR0xkGmqSbR0rEZeK<G4G;y@?h0Y@5B&M)n2vfm^B`
zFa5(wPQn56KQR17BOVC9us^`XFD?f<5D-5|fJ(@oRIS=v1u#jwK*yg2+Kw#+W$tL5
zgDJ1e{;lYW+G8wnT;Et~t@3VKpN*x47yW`LKzBg#=f+5~Z)@=bLm3q=@6CYj+8}8|
z>jXiLn}`A@J~2D%w^^>Ss@I45G=_0ZgATPHg0VdpvvB~PaqB2u3BM8p!Q-~fJz0uq
z)795EPuP)<p%^xu(27(VT{<zauP&>9ccNnT&6^Gt5e4y~{IkR`@6UHbQ6g5F3YDHb
zwK4E5@cF50F`qfp%h+M5KN9Ca!*2$c!4bfAx%%_l4`iWv|2)StztadYR%!yc0AB>L
z=Gc$+{|&>_07V~TIp{eGf@}f@a4Js)k=%Nrv1d3_g2(-o=8g5-3qipYz(pPb?CUU9
ztA8C4ugiLE;%2jH5(9R5v?h#(*K1w=_ED5=15a@m5rsVw0oL5#I>D%`+v=N!C>cpm
zxz?kt=qD}ueGVs=hfTiqB88?8U8B(g)1|%rm_#sL4Z@rm&qt3Kkw<Gr={8uSFtTP-
zF5AcYTJIAb+vTq_|4Fw-E~R_Q@s?r}&PI$-dP_sA^5jXL2`Z@;fU?-##)<kbm|=2$
z{j$7tW9UjES?}I$fT`0K)5#dspRrOuhf%^Y<oZf1ZD*S68NulwEJUPMh;pH*f2hJu
zns~vf2o>n9i(MY!DF%kb!?mOVK2yJ({D0|@Yq8Fsk4K<qAI+?ra=iYhCA<R}gdfpn
zki4P>V&vpyzQ$223FK)z)Ft3>j>NaNXZs%YLSfsM_6f$@_6`n;wb>MCYi3<G70{1X
z00l12*Ec;HVa#RB8vCPR7DkCt35GEQGWoAN3yhQLAP#}Pc)~L3!wNL(+?plKgEjkG
z!TM&&?3G%t1kJK%qvs#UAN#eK*akY`?=-IBV|3H)6t83P4=!i)F1i!d1eP9s1%5$4
zZq{MI2~(Z)9s+5X?cbU@M-n0AhmSSOf9U1A$^P<w&ZV3p6tkz|e;*Z90R?3U=ePF{
z`^c_B?KN5&)!~MgMx7IqLMP)kCmPKzLSGvY){ECHl-o3$#R;+bRehl(&#;V9fR7st
zm)UF$uVun|Q-IC!_|gvcD1MAU7qLE~ZM9_cW?IE`nGZU>0A98)?bRA9-s(8@_R3pb
zy(7xRg+H?ra!`d9Kws92?R5LTP;A=!+A;-M=NrzK7R>v|32)9T{T}-g_yA$~{X373
z6UWBpiwnwhDr}1^pdQ80X@tiMTq+df!zF&5XCi$ZuO6;b2Z({=RjJYuk56tifQDzi
z*b_befbGDBC--ZEd!qCAmr~xJhNq^fD@GQX!*()ZP?!e4C(Yj9u?>K))Vb~W%#ppq
z-`kncS3F2@jT_1^(EQ(G*n8{y7zA7o#=%nz^RU-qleGUFHLWAS?aa(Gg9+h%%peJ8
zCeDGZIWnX8n|$7U?ZX;63c1BR72T&N8-a6;4=|FEW9|zzdE3(Li$s@;C`Pl}5W5z7
zowTbA!!L31S{`osKZPWK&a1F6uf4}2+TJcX*yk_x(~TKBHymZcmMWI`f*dyip80pr
zWICx}4pfLo(&!qmZ90fWLIxU{u+FiuwR}9oweBjK6Q|?c<v09*2n7)G9N9h1q&mB(
zz1%_jy9KXyLqk?hu3!CaDZG&2LRIl<P;O!(+2H@yJ&oC`Hu}dH@w+Dilo|P;3j~v`
zsKe6B5z&<jpI(L>;ih=Eoo0v%8^A9iWkw|V>63a(6@j%i8ZOEsCone7d&WMX`|q}A
zO5&=W<-qd_n}mRTS}XI(4Dc}!+)2dX-~{uQI*^`wqMSn9|DEXVJ4$-U3xH&gK`!_x
zNUO75hSk^Cf3K)R&-)wDxwYkA{@Z-%kp(i;3{-*2+1U6Vd?Z$uEHs}{mZLc+Cl3}^
zMcz`7y$g|V%Y-+fqH-4+*m)(1ebhti-_@xHu7XX>^qWEc?V^3|B&)V9I32~aVAP@s
zv9H2%lSjQ-g#LwiuzFAPg|Qobe0(@8r@obWs_Hz(J^u??N6}Z%{!~wFR4xzs<w{~g
z_)o5?f^BKQt~XQ9u8`}Nl#(|5eSkXUHj%pJ@2}Q;U|=e7Q(qm=Zr1uSYPSi`KCZ=i
z-{fYy612fwmp#_!LWfJ_tPr6pYe@iWzrkHjR2v%>S$L|}8=CxyUK@e+QP-hoaJpfA
zY436G>3F7|l9`to8#?$n43A;xW%h;IE)cRRm+B<1t(k^eJW@!2Z12mg*1r=w7C>zc
z`Z!T)chUw@A-C_43~kxrl;FPfytwv$MSl*iB)s?I$cJ>o5Eea)RVwUnh3Mn&4<I8G
z;~w2I7XOtEJ>#dy@rNv7UuwDW_NMwPnr`Apa`8I^x6`^S$4jTOC;}PPJ3+unnMYr<
zMwFvUgMAePFNkJ50TKVG`NXbUZPgzw*u?tAiw}G9`V>@nvAYYX_jZTre&;yaQlA-u
zF+CS{A*)+T?a-wSaKWlPTBq^nWQ_5e%@vr-G+%X!$L#fS%hhZNwDCdrx4UcBfa1Vo
zw*p^Ma$rLl_Bak%G%=^w#$1VVN$_w8ppoHRBHt^z?{uxKNR{*@Km~IeP$Rv?nvL&j
z$v?^0MQU~fi9^-pGBpSv3{ZL?Bjl-xOE=wEE36WT!bH{DlAfh2Z3GA<JWi`rJHNt*
z7bSOkii0q(T<6~Uijw<;VL8&Gkn{MgO83mNQ0@l5a4_Ts7<F&Y&YLNYW$HN+$M^Ql
z0h;5-T_DZjCGtCchg>#toV^g<4Fk4WZub_4X`1}d&v5ck=UrcItf6Y?hcnLnd*3Bo
zsfM~`?Z#SfzsI!naml^u6JEpt{S+<_4*=P|s(aj$jg!ieZx09^*4C`Cl+@*sn;Frw
zb*1)wlfGK)qe&NtdZ-!TM47QW^Q-xajq$4H1}m+vmE=FURR?IS80RehBlj3l{tt$U
z+UEl)zwf?b-H;rM-oDK>&nE#lz`{KlHUw`-C1of?Ne*pJm_x8-4_N5ZmwbL$(Jpkz
zW%VnZnul@k4Cx-RZfbhCPy;6`OGf=jzCn_Sx;0-NRMAHOhkxXaiSc=V1(i#Y7yR&c
z!iR5^Z5bXgUvE!wa1n+!|3y^)c6!jy$o7EFkG8WuDU)*i$k47=D!IVtxV6SzjV)ek
z-%EQ#n(<;w<^Xx2wY7D3Pby}tko%El;O{aHWF%9rxBf|UcTE{dQ~eR&VVmiY1?D7G
z=(btBv1xl7J-*%7c~O7-kc?=EaJzzzKDNbP0F%(r+6h`XdO)yX$?L6Hd_rdy2~iAt
zUCWS(i}!4@r8l|*+jc6JKS*?%4p(w4V)vK2ZAc$~-_D&pw*2&!xJg1rEH}HQelT1F
zri|2>rkgo{NS*9sr70qX_Z1U98xgfcvQ_tD+5OdXDfgjVw^7(orGJyzPN@<<0J;RU
z^O=};o438cR;g=2NXWZ{Tzxy*dA#>{l_ra_CzfnJv2IXThtk2PvjBL1@=>cWJ2WJ}
zIQl4VX#e<{PSB^P@IGdiO@^k_?O|wtf627FsT=FI+0CM)Lz(V*CLVK)&{YA&rMiVP
zi7sE=_4CWW7gq^;ABCNhc)HU8NVL%Qnt#1pUYiQ>(1E9n6WU1iLaN1e%w2+I^q)&k
zQs*?u(#q$(!>~k)!b!fW`wlNwD&~m@r&+Uh5<&Fbwdilca0_#FV{b<>#pOQl+kAzd
z#1dbgwlY>UV3Emn%a8Z^Xh!+*nHNwoUuEVAc~ao06;a8?%SD7NSYBr_Vj@8zBVS5`
z3C}6-z3N$^x;CWfs%rg*fY32Awy)4Rsq1KZiz+BSx<DrifKM&k^6;eF0&TPTO1Zmx
zot<yJKK#q$ZNlF1x%8kGRNJbS+Fux>=%pbM(EK=CL<GGQ2a@v^E57$qvH_8JLnH4R
z^n~mX@ReSzuAshqegNcA>tWtq>#;Zq6`X<4;JSA)`@?Kn_wNM^05g^wb+Q1|UVmAD
zz#V;{Q)ELrUI+Q9oX@n&((KN&D7gqKa@K=FpMZiGpZH+pMCx1r)EazvDzU{sfa&(-
zcY1{Y59;|_(Uw1D)Ue=ya+V?c5T9XPDq#I@Q|q(<%y*2O%QsZSB~Pg@?9Uxkf1*}M
zRdvb`92LNWh=PFkOI+E}*~niPy~J}kxd&bF1v`Bx5S$!uhHKkg<LV^#^~tcYu`M_f
z_>zIJc=NB~KbWqR`2Qth_@`&|Se%a(Tw|Chra}zj0XL!T>+YAJaUb_av%ySt`<mYn
zjt2|<*5(FdZvz^N0`XlAVeXs9-o&u}_o|Hny)3;=4QRwcGQ8yvkYs%6sLW9+SC*`e
zhX>`vhuihH`<KF(f-v>EpG831hT>-|-tGC>i`cIOfx+JwCYQ(KLz797Ok|ABf+ARZ
z;Z+6w_wm7dFyAb1eq_0%b3Bj3nTUwtxF~LmU)=10g7yE6%z_(W4LT-p2LKtIBh6;d
z{vhwtC!54gL8`#OuZM+g8*)~xY%F?Pw*e$A&Uc*e7@l4OX_gIO#!dqfGz=Uf+6y)|
z#l@h1yG@iC?6#}c*(E)+?5ey4FC9icm~d@CQQjyHN<S8t<(m&bF?#sp;_H+MNk~wa
zE(%9d)H(Sxz02AN1YCGPH7g028&`9+c`jtbJivSz_B>N!s6bV=uLAAG3oN-`t=aHj
zWOCg>>vDJAFKrz<QEA``C;+Pvq#b=^W2v9Jpku4<b3i=_K+0btN6IyL$wJ7-E-eTg
z;NaME%ME+sL-_cbw#yJiAn<`*W$&k5J}W(Wlc(7kPid8)Um)U<W0KAPV>0Ur7!^-;
z=p~(GGMSpMJoi=;|D$>Vl<0txiR*oPBq0?|(+1R!pXO>=fByV=bw9kcL1mqa^SA-M
z^;ubu0}JD=C8c20l-vfkwn$o%V^f&OGr#0pciXYUy`W`;WNX|UPq((+>bG45zuip<
zgoLNNYs}|*5EBv+3{@+L)pMDD0laOAVb-_5y7aB|MHD2EEEyRW9lsFK?``xJ`RKBM
zs_ND1Dssd>Oca#!b0j1z977)He*Y4c0IgW5uB@2CCzsWq#QLghU+3$NhcL7dwbm^?
z<4T3NCA%?q{_^#C8XFszd*8VOyrvg)8WiYa00I87^aRvLznU7Ck{g`{=P(3RyfTjq
zYvAj3czT)zbQ@@dZ16}yrY^sd`CP(52h^m-lt#6P0CBL%d#36WPVvW@fqSh;3g_La
zHqa5}0iId^_0JOT?`}px^9;g2A79r3JBk~$&K}V;&{6dd3|v`TdkZilLS9E2padEI
z%j!iAcH7SK@{#C6cvTfgAR1xD`F*5o(A&rPL&~B{Jw@ak70c*RF5!Q#urXwVyY?6M
zHQe4TyAioWxJpd+h!q)W_SRWByQ|&6MG_zq158`KxB52(+9}MSpk#7(uASOeZ(@4P
z?b<&c0H!3_Tk#_w4~5$h^T}7TPtYz3MJD_CfKKPq;=`edu`y7#!O10YAGJ>cM6-9i
zP%WOvfogbon3=*tWFR|?fYles85<9Q^s&_8?%F9wF1a@thx+jBtp9%Dq^<#^+7%TQ
zI?#`U{L@p%N7*+Jz3Kk&0Gi6(`38L8auQC$;~UTw>cVL5z_jRfK?;0vZctYb7Fz~K
zvZXcZ9ch6Y-VjI?O8KotpH!JW?mYM>Aycz|h=;f9Kj6I{T>h(<x>^$!hW1hnp@;b_
zUZr5kh~X9scGbGAY{x!z1i@ul3nIyCiT)iHqq`O#|HBMd2dtG~izm0WMm(u<3E~|G
z;=?Apu84hiVJb)%SeeTig@a^#0t}bWqK=H@zQg>NE4jw)2|!Oia~Vr*b9@+$q-0^`
z=u0_p0zw{gYPqD=?(U$^pJ7HuM?vZb%rvloG}mR(D{eLU2s{zQ2dqunz>g^9%6(5r
z2$9Nu1yGN7p?Kd@Qsn#)(ceo+!KML4&ilyd<U_C1Ev!fiW2oxWD3!-P5R@f!1YxfJ
z{o4x|;61;8zhGyN0)86`1<JI*;tSBh6Lki%mGc#g04e71&=xGRia~v#1W?!1{AWJ|
z2X}S_eQRL{M5r5k6P?EaT#yLons`M0?p|O;vb9MCJ=oj)X722*YuNN;upq9Z><pu*
zizrb<AnwqcfBk--*N&Vk$gdd7krB}<^mk|x!tdazU3d#Y<;c$Z4hg~<Ps%oZeV%ZQ
z3J+!~ad9-u(Gb85RVf@9X?Sbrwi;*>3@xj@=f9YGK5E7EFAl?}%Qrl5!+u7Q3+L@O
zG&YbyCbsZ3SOrS6E<nB{U(R0#(k2)nPB;uWN65*^0X|;rK)0ZFE-58tFjb}x?C4ew
z4kE4`j%KE&hFTy9EV4_Wgwg;y77(O=-`|f;pwv~XyWU>10&Hmj3h%$GYmq2W&i4ha
zL?C##oURB+OMB59O`ADdheph`8&f(xHFXGN_LUZsOn)a!2Q%J<+Rt5h<$~}J6k2Bj
z*^gnS(Fts#V)N??g9j0AEY$ww>O>kbO6D&JRYmKGtKR^m+}q7Vg#NC7jc)!X-~LY^
zCTUu@fU7$6Ip!sWiRk+flF?~(byv)D3805Mz841C{9|-_z`sgq=`)zkXFtZLQx_9J
zqOn0Z|Me{aAtD+S<EO8OLE=UKDD~~AEeGgE)aKmypV+pph43#sN6%s~<PubHO;Kfh
zpHe>l^<MTYXze`CvcHS8Ovdf6X1}_4XuyF7+<w6DmYSk*0{&psDZfOkP87%w$7UWw
zQJyj#u!cJm1x%4taFy1hI{-kiy<WB@rotg%z#LASrZ#<>^XhGD^Ra2Z3aoS3tZ>+r
zYrfhbP9C6SesoFHyV#$9)s_f1obm5hqr3+d>spu!=1ZkNWG8&Dpchnrc${8#{n-`H
zmKP<*d$KnfdWR~2U^S12rxpmvg!$~zmC;cx>&0mSs4{ez4~1xg6VIa2bai~9$_WSv
zSdK?O9HY_Fa)yb&76;^{43opqH+{u0r1NWUZ4}rKtD<q5U-Rp`ahh$&D0{+&NrdS~
zJY4Z_zLBTdn$BW{-eVVe=FP7nmT4`#x#4|`e&BV$fMh`rW1#4y%W}C*63)fNJD43D
zPReIsY1y~b=J#o#k>@d1xA`IEDc6;LKdMr627w5Oq}DEvIHrl7Jxf#I2TkZFX}~?O
zF8tHTTkm48``ggcqiUYGT-I<98|hRE^o)&TydaF&4|s^pZYMe7n}^NMkaZqMiRJ}R
zy#E?3<oBo79~S7Ru8l*H-iKq!{xGN&<W^sIq~C>-7P%IS28fTi%(jEMGig-_eW!sD
zNiRh97iaJNX55;;#ZUwbvTxeb(x}UA-0NT!OmBFRAQ|!u->68ku~%MN_A<}8|8QA>
zoY!)b%gU?3CCRo_TnASU6?d$c-OCD{0nD0Qlb+Cjn{v5$<;VzY+T!~#Z+{fH6{YDC
zez&+t$Ks6X_w(|XFJFR(2bc6+i~zhE<OAR5A+A@zAHb`<qvMHuzOs>#5f~T8C)Jzy
zib@iWRk!NAJUr`!?83^*(a^vqn|?byopN3)P?Xf!uLIkR9^TZBo0Vz`S0yDSx5K3;
zZEbDvNJuT=Vvp_1c4BHt>>jD?Mv{XKTTuJ$qZWoY&>pVJzK>d54wsf2aBxM4z&;L4
z6+mM`2rOl4{`MNAn^&3|a(}N~UDtR#PkNLrdF`-6ezad6>lH&HHM!>{Lv=L4w6ti+
z52h(%q*g^&i*NiyV7ta%aFdtb+hK^e-8O*kW`$#?EdVyh7ypOUxYBbL>Y#t#6RBTM
zJ`bFu7R$~6@U54BcDX@U$PqAX`IeB-+ZBo*F9Zn!r4&%hB!ke=9H~|b*~cw{TtP>j
zh7rn*@VMZh=XFbH^h((3m(<cCfe8;UJr|r<(}IPABNX!HH!(4>oT&=_^5x5JzyyWl
zD*<iHP_>p06oxi>No=qTbHT4uS67$Mc#F(0AaHbg>Zv~m&W2u~<O60g9Kb@Uu8-<*
zRokQjim3zR8dGh&Prgsm5~&Lg8I-lrtQ3gC70@xX^DG@6SRz-JO9Su1IG~ZgUuaSj
z{8ct$FRFnfjzXYAg!a5|wl;o>x^SmiCX&Ki^QY)N&_%j^=aN06#jwuVZL$5Hsa67{
zY|<UqUk|29X*p57YiZTksqa#oZZK>LZeh9aWux}zMWxfDo8DB2d_&KG*o=7Og+<_z
zX#S+8Wu#duz+=6|eTRi*^yTUAXEaqZK0m8=Rd}~D8Z=sjtn>9;BDv%THr=jm6B{q)
z9HNG(at{P-o*X=1sV>`k)h8Iy3!~e7MMO&chQT}h(Z1{1iqB|;H7TrTufx_aS=dsE
z1l$fij*1uWKXI~@86rvNKx&p47#JK@GtnR!vtQ{%#$(p`s#9vZ(HDR5)no0u1k6;u
zQ(>2b*;pPHOoQ`YuS*~B)p%kC%*L7xL+qIVQ2}9T^n*~gZi_d->B+*xGpjP-R#sR0
z1(Ml{-=Z`EBR9vtW5J*uiiM4>xHJWlH}HRV0GT8(P&JZ-vR(s!7xk_W9_Mt#%m}Tf
z=rJ0lBEg5pLM)tU6N~<SN46YT&cE7C<MgJGFOMrK-cMHo9&(`ku#Z@}zE*hnzzH`o
z2FcsPQ4t&RY*JrZI$mK@8^w{)X*@@Y!IuwOwZEqeC)0A+tBzUrxQU*l<-^aNUE=D^
zIEZO*ZZqye97O8G$b+e$r#JOKqQO(QFT+n^qqDzS>cS~wM0{gP{nPUl7%w@$VU+mK
z|E9HEA>Y+z$^oXa&Vf8%)2iH%LC7v)Re|-f9oW2wqU&1A;j}+TC~s1~P%IHa22-Td
zlmMKMaOrb`f{s%JDN~WjfGDj$m^8WD^m`VRMS;ht2`>!9MIoMZjRWg)7(cJHylQ|q
z(yMNWAm+V~%hpgj5Gs<(r&Iy=8-2s5i}d|Q^`%Agrbj!TG<JfpTes8N5!m$Xppk>~
z9V!+!;ZGK}{F2*$_+E>~C-WY_-jbmsYZ~9ih3UB(ufK8nxL#4XCvmKk#ISje0qJZ|
zCO18RBzeQ_g4~5Xx3<#%U}6AA`B1`kq5G;-f4=_NdB0lYP!s)64H`xQJZ=9^rpryX
z>Ym6LzgE8lzPT29cUkcK)O@=AEBrs(?@m86^%^Eev#HOc<uwH5PDq6H85b7w&dY2#
zE(a2ss+E`3YUK*?CxMe3z{n0V06BzKi?_fgor_;vD@das_bkkO!y;dBTStc<C`I6g
zvLtbU_ltzRz5T!)SWo@|0jwV9mLNPFOxJuG#V*Lu`+sAC9GFKZNwX80FE>|LN5JbA
zZV0ew<dS(~M%i8h%WTktO&d3V6!leGOmgjLX+qWJ2(4L?QICspz`zn%8p?nl2HN1~
zS9YER@wQgRZQ$Jd1tjD!TR?0Kc0l{DUx?M3$(vGcE-f(U`}2Q*PZeWwB$y0^zo6QG
zFZn?Yr2%6aTC1w@Nq?^YjohE2Bi`uCAI0lk_Zd-R31gF@P3YcSTUvwl++Ypr=Z|nY
z+UWQA6<?H7@etq{?(mZBIK__Yq%7e0Tq~mPU5@I<EV4QkPtQw}6?nWbuK-es2#3Eb
zl-9qx4Gj(bQsbx@%_nm#Ds#AJX5-`j-Gt!ps+VtNss84Ubf)PDFL6*i>*fQo`gAUF
zbrb)|Jhf7(ss5pz;2cEox(u?s-Ki?pEj^8FmUeaJ1|`Lxi+f;>xuNkUk>8CCcryY)
z+~=2Kfoly9Kn{6ub>D@rYqURC=Xp9T@%WWDZf<F5IaO<CIPi7uIlZdw{u~c*e+KGD
z$&w%FL>zhmJT*y`zB%992IB`jg<Vqi$YL2Z+JTkbao7b^R_W?<k!aK0Gp$o=tKZD-
za$sv`*E22Xkh7%oXvW>?_lnD-{0aXtjZT3|Z-&V<YT|P`KJnA!9}iejKj59Ymh!kL
z$F%@szRyFXaw==~;w$@sI?<ON3LtNLfK^?;V}3(T4P3_c&Qi?&3yJq@tKdGx)B)30
zByz}(Vuhq%wM5_cZsxR9J1W+@IoidZLZxukzyKT4Y3%ETL~EV-2t7W2RBb$?8gzE+
zGp}RTZB7D1atMq|DnJjrM5iePxLq=8RxJhK>*)Rd_Ht@^J!L3OG5-fR`bry*QRr2R
z^uP%Gudq$#bpi%n^uY4MQ4VTZ-~&nZ2R*$PKrCwVaDNAU3DPEaScx9A31>iB+1e@$
z7HwyR@gHv2!x!#NDd0O6^D!Dw0sCqu-#EOt8+@0ocmxb;%Yoz&RPR#2eh|1hIvy+}
z*DSb)0@Ap&PYnzn^Wnm$5;Upj4;U(fO0%2^kP`5atagNW!I-=)*9r|Y{y2oyUaBpT
zt*9{Df*ny~ExCTY9{RkypT{HC!hDz+^^ck}8s=uB=|DXn-*?=o0)`<I3-OEW-PKwi
zzIC^nM5Xh+{3`}WyhMaIY%dg0!?~8Mkj9gueVqln?FToL_K^rRG;ST1U#Vqg_WT}E
zEN^^+UZ@InOpbQmcU-X$g!yaN^;S=*V=F1r0z0j7l;?TNYA+WrKKz@sSJ3mOrl|%!
zj8HC1gN1tCOInJgRMGYm(g)YVSkL1TTO#FRMR&;!kSqBJdR^ldrtZx@%K1UH6wl{k
z{r%z-P-hRsTwZ|{oeuo^PC*_0+fJ4T!VeSgm?}3U<a4%6xmeV8#k;5g3lg-Od1lYg
z&%0)Ttbeq9`tc|YU?)J6c{ulowYl;0U*pFkP)F0TE@>Ruv@>Zu)Jj*&g;72{ersju
zC&&heCSJyxDev-{^t06ZNy@fnz7|i(uTHwI@d7t|(@^Imw9lTVW7lyPst{L+sW7{X
zG2bXmVy50CV{LS-2Q_}!h?uMGyS$al1Hp2QS{|O|zguSR&6E^0y9}rkHbytYhFQi&
zYxOP0$unQKrjxM*8Vn0r?~2D_eh_Sw-o&6I!3N8{|Lp`h2??ca&jlZSxac@3p-=he
zi}!Y!WOw%muj4br6xSq>|A|<l^0wCze+L!N(%-E|I2EK1S-K5BIZAZ;XSPxF@j~%k
z1^D@a5h<*2>EveLQ2#-Pt*WZ(fO8Dp#GQ07umKAiydQnZ&v^$IIU4hdu7t>J4RO7a
z89$`oSE(krCAp7U=&mnP_Tv<<1>zpc_8;Ti;h_={uk%zvILK5ft+{Yp5|Or;koV1}
zq|N!{M2-!4vma~&Nw7BG6peQ0I9zNP*STeX)a<K0mHSNX^^zM4Pv`UHE&?Gq>$UN-
z1H=M#XbFNSItJ9Io}+%aZ2v)Ld(34-xHb3c;G{ej2y}iiT>I+YtKx-pEg)*T3L|)A
zBZzC^`BOMh@I$_@m$4YsTYSc&7IU1M>brQu+suqM5>hYMa-Gbac;)k>Cnby}Y!y1J
zNVVRtuXH8$UE4A0S}wHvs)@5UO4ML;TRbKPUn=@Ly5ct*oR_akw2P8Mo>Bf<L0ACB
zeHC`A;=-TmmNqxlmM$AiZ-3e>E{tq{`}VC>nz_k;7&7bZ>Iyh$d01HKemu|hG>>k`
zzup<piu)<hehtqnVz577jD>a2RJ_%43Y%3M0Vd8po}BmS;-|YQE=!hovo%U6jl>x;
z%!h%w!tdE)dSh0<&r}}^$^|)l-i{8IaS|aavlPpUC6N8>&QQ#EEzsWK4{|ggdFq01
z8tEGa>8-1%<0Ffw`Qh^)lvWvD9=u&w7O)HrOns#<&7t)kE{rWt_r|$vu+m>EI5>T<
z;lSbiFK2djCkwQtDMx#}(Erq*s>5Zn*7lexLN=+lAQ*>WrcUX5c+~fpu1fMj<CFA`
zyj?3jJ<4^SKZT`sMu<gW6x)O9Iyz-v*yEVkA)8a_k)I-u`%Cnx9VR=`*}7Hh=O;X~
zz&BVeeYFKx?<MU4A7V`o{LJCsU^j!?H@42)Pu)>ZMZmr93RJsSo4AJfruZ(Y#{t=4
z(W<a+XQ3*EY03DW(*)O-1}p@GO<N8a{{FbGU*p^_TDwQ#?!%Q#it*wGo=P*r9YXt+
zVr4zKWs^lF=<d~)Pln?U9gL15;L5IfDL%R7I8d9fopft}=ud@C^i(|4p5!C(r7oju
zDVOqKW-aPcQJ~0keK*6r^&(t}5A!(*spMkhJ7$_Cc(F+-14^T&G+px&bNcTLUFWD>
z?C_%<ESFFm{&xJJGdpt+9Q^pZSqaKT6$l?TGwRD=p7PvcQFd&-eD|6_gXiJh4jpgn
z$8mjUZDl4RJgdtvx0YM;g`R(RDL#GKO9D0}a9oS<^MiwV^n8?*?3J2xt2TRzhGw+f
z9su(Q-d0<vg>F<p58k4pqFa-tsqa?7yIsq+(GahjFVaV{T67Uyp<>N}q7m3n%7@~d
zt|R!FZNG62nj_lue!J9G`oTF!gux4Qvcs0-;tNQ=KD?rQDxLnR|8KwmUJ9Q%1;)Aa
z;hhurGu_iwm|H?PV(XQS-Oiwnpy%9Gx;I?CXo}BSnz;Bkw5trGEB&s{9Ig+4mMF~@
z-B-t+o>x5g*J_59or9J#>!lW#+Veg?@jf)+baWRGGZ~<w<*aViZeFZTB)^|2?t7+l
zcwnfIXR=SQIVNUo+R?D+DIq^Kov}y#FQECm(x^AK?(FXN0awf?*#e><qrBR0H~`nV
z^B968(e3VR^59cX86cM#F!!o{i;l(xDfe68x3IO)1TA59fdht&2`8>t9&fN<8?6+#
z@}U96wVF=Z<rSrXAFXaFo$cj`cEh3OA~oy6M7sNHirhgfpGe0;Z*I$o^rQx2;)PTl
zGuA5a7?@@^4_`%s`>iCl4prUXbd|iXJw4Fzv^vGJS8^X@(+f&%d>Z=hxAC)C1Z!Va
zozvWELHiqdftY_kYj$mP1c9K<uy9ZL;<Q(#`OYC%aV%N2!7SjuL0}(FyVs7MHW;DB
z^0$k+xGLUgOTRVFSxbV{_Zn_<9zzl(1h6~CVitQ*+W#GwqF+eBi4S|aH~S{{tR0~6
zz~AIj0AT?zJ5E3Xx8~Vm(1il*a`mBM5-=G4z}W`)rNrtufp_m8?S*Ui{*r#;R%1l-
zRUF|p9E|!^cWMBJE{h4Sjlev)VF*S}P{$YxHv#6hOEm@?EdFS;?@lQ7k}WJJ-F3O?
zJqf5c@o9_OlF5wEqJyVCaY#kX4d?28>}v^4o42D!<&x<59d)>ho<vFtsXyHRA7O6+
zlvVphk78gTVNjxg3QBiKmjV)kbhmVOb3{~BB&54LrMpz5yQI6jLF(>Tzu$lUb7$_{
zZ$4)nd3nz{&v|0)z1Lo=a@TC@h3A&#<FWb1G1p+-#V&E-7OF6D?5;-dqA!w)r{%px
zg!o0Td+{+v$S`xRtv~6YB$2t>M;&T5lt~*I>g{{=Qn~dCe$l=>{uYa{CF!Q4_W2S^
zbcIuM5#fKs!JPIp*Y`ez7LcZyN6wE)`A;^V#p+!_FbV6boOjYSt1E>4-A@j!-+FmD
zZ&V-H3=5<7Bs4>Mt%r|!HH`0ejgIXK(DBywhK^h}ey(0CXnuZjBi7{Jy?H;6Pwaau
z%yTOW>Rcq~jX}Xfb+IMld4o=LuZe7a)%K-5Wf84<6K1-FZ|GXms@I2;-gd|kW;-c)
zjkY2k8J)_Msj*mjFgNk&wLgBl%4Q4YGP*oH>pp~J)D2zOn&Orka(la;?cHXa_kZ>!
zcuD%}J%N`U8m~R)yr^09$R*ZoE#cEWvBn?{!rLu+(6>rdF!w5;pu6|WOu9l>F^`+0
zZ@agbk?fCxjktdc<|a<+ZF3K$avS{`fI_Yyk}vl9H90U2;e4N2S&<MBfE<kQDz6eI
zL|Z6nPe{Onl+KMH-RHu>)*!mXvr%a)mMoP<2H<FxCr1iDy48AtlZ|k9mKS_AXfgjF
zt{Fs_S7t9R78w8V^lA}dIjAB)r=iB{>LcDC)rEhOyuWLgP|4Gelweh>mT3R*6euCV
zm=<_(4LS=5zlf;cCr+1ZjT;9`og!ON6UFRnZ#v$^4Ps>}=r2U8ta95&0>&A#Ke;@p
ztJnR+P}|Z}LLlYyu2FZir@uqI`KgOApVa1%mHv<D?;MmlOzN!neK<rzrNTC*tsm1t
z5x+^zy5*vE>=7+urzqb(Hsy4?-jHmu>ziP1rN2nD9D8KM(Qh_euZ>9~r*~p`SUMzL
zg3fLBT-E9i_X8>`Uw%H(!i}$T3?393N~M|Isa?;_a<V&8@7AMpsn#6kzAp1Cq-FVu
zgM)*7j%rAzLe{)<ZDFCU)D3T_`V{&Vn+-uNg4t;$r&JR0v)*4-8Hc#&Vk6`U1EQnN
zwwbcq&jjhE!fwVWUX`838(O!ZA!`Y}F-R8TyWOzr85y=pNRTTUsPH8&>(X$EfxyVX
z8(G4xWjl_uG#q#9>A5tGF|ML4wX1}l-A3eeEO#hek6kh38aSn9bJSWsl>5hse@dtf
z7U|1T3${gA^7TC<a)m`{?%})9jh5j>tuV0>`9iJHD`>bXq7#j;WrduXe*971Eghnq
zhz%v-p-1Ng=x;RBo*{8~d++r1@n?1r7Z)b{0XlTmoRYftp6_qs)Ew&`ndB&%e$_?Q
z5{G~g55>G6V`FW1nPEM)kZbH?@|e*w9@BFpAmDMwsLzDegvG~07(ym1_CoWoXB}5B
z-QaV$;<aZG6#VmOanBiXVkY#(_ZXdh770KBg%+%}aGmG`v-+7$2WX*up}YUq9MpgG
zV0Jo>8?q7|LD3_>un_cknzLt=gM4!$!V*ItihNlt;_5SDV`qO{upaLuu!!vaiB9~a
znv_4RKQqrr<a;>=%W!Av0e1BD$ISG8!Mm7lyu9fO`H!kq`Gjg6%&8VEV(tbVcrAN)
zw1{M*8%#D(F=Z<4{wVp#uLKGBtIvdn+&IGP0$u^xL9Dec`EA<OVkkw$oZfe?s0fdX
zd#|v&ugGozv+HgMA#5Jhh?8qVNl&Nd`S#GM72Tm~x_qvBY#us4gLLlmMoRb4wRVn)
z)yDD8)~u#HP94rGUKgr0{eLNXwPcepObe#<FFiQ6F+8fJ1pfwhcJ_3MDV&n;E`9Ia
z&T3t!Bde=V-S=F&j99l-ls+=&w&6M<P_YgS3=HRX%pKzGhG9d#gB`<8RbDo()84iS
ztEiNxQW8~&irw^?{?Bl^$$n5Y?vE-+<iNM~6QfbwKFDA>7}qg<*0U%7D0gD=V3JHy
zrTp{tOMh6Ln*<6C?f5?aDKc&VgR=Ui5%GlXKCQbVlW`+xJl(1GU->3UF|hP{wnOe?
zb02T`QVS&Uh)yAcmfz;AZptZowEh&l<m5Vuc7K(&i?2>7av(dwEV=^A-z9=%(FDMl
zv2qd@_POy(nMms6Ka*?gBBKqjUkp(BbVX1f7k6bV`toSqc$ulBgi)1dL33j_!Z|vX
zElWbC83QndOErm_ni|z6->jYEUSaULk5HBzpTuvImNul(^sD!Gj4ESeV@`BUZf@??
zoI{d@T2><8T|a`vFElK*o>1hy5U|iT!i^pU(BaR)22NdCoJXXkv^JoruI%3>F@v>-
zvMQrLCN9KozE>o9Rezz6`bF<ECyWzf(P2Mj18^><^Y;49p~nSToa`8n3=_FLdTa-W
z*Ut^=kRjZ3!)eK(fY%>oT!z^z<qq4tyNzZGeWSvw+)`{ZJn%#pXR>hrxUGdOP%2#2
zq58()H>*+VbY4mN+l3=sYa&iW*plUQ27X_fJ&pGDRe!IrOOkP){)F|k9yGJ((1t}{
z?_+1@>8`+=So@9cIx{Jz)qC-=$&~VKDQs+>fn>xw_F1e^@8@=JY?=iN6VJhGtK}iY
zdBw&|e~${1o698b*)eZ#;(TF#DykL7TR6R*SE=MvWebU>U_2jWq%AAX-4pxLmyC-v
zk#%VNn~s-##UmarXMdPqXRP7MHo@>6<*?tn?JpJnv;EM?G)^dEWw0ivV5?mr#Yj>8
z7qaT^=}F9u%jLE|tQCIPKS<a|?NmtQ)=&6H;f*-<IiDy!QOf4i{3~u_mzT?|$a&sM
zSxP@y-#^VyH%ze^=NT$t`)R5p7i=sDDe$c)B}8mLTpdcy-aa8bbz;z;X}lMpqpLN-
z_Vht(-6qM{c8oPIq`j*~)k7|>HL&*j?I4c#qoq^Ce3y$d)f)dVv;b>`fKK%%G$HJq
zsDiW?T1Q6*XKC}Q#>3|--wxvBs;bS<jm#!zD#Ue^G@9m=YW5k^NbI(sY%AWs#DAh3
zaddoBh3^b&m6d4D=AiiLXVOXU6{{zAxz*G{YnXED5=P$oGoFt0ry~#>dGPAbTj=<h
zA92^c{nPa;NaAS{h*6X<?;P-Lo_$+*mGQ<hkBY_5=b7QVh+6^1lNlxE>C2|mUsP_G
zXyoO}Zu+;rMJM>IwaHJ26P0M?U^h8t=snNf*53T-oR{VKC;pxL$+)&Ddrg?~aT*sV
zgTDSCZ6cZa4;LYUG6on`=;ghpA}$@UHQ%nTue&D^O7*_>X_gSnLE=WX3>Ay8Dc2ob
zhfqt$Rcbi;MWwL!vm-eKBRDIUiL!lZRdqT)GWYHruf)G4O_iekCTbBX)rk6zN~!D%
zdm28m$uU?}a(6O03KnMhPYNuN?gKhOZ=oN-_V4XPfPcxJMq!p9y9aPOTE>9PTdUT3
z0yihDhRLsmZ%#++ep@Y?ig0xEvE%*Bp@j>|Y}OZRV$n-AovJ6~05@J!m-cq-z<1BD
z@G)k5jH&n8Z|l;Z?j+%z|0_D+!9~}6_>R)YV+;xCqEQM-)WNy)XK{F>&Q#$7a=7Rn
z*L6>_is$qKTe2<1_+75&RNEf5c;nZ-b~27OZhpm(6AFpQqITLSH@w@G;O^OY=&_)s
z?)O=7SkG>o-XqY@1Y!j=Qn%Xq*wAMK@x3b%do!wYQS@`c@0(NK)%Uap#J9Urc-?ZC
zc*~SF`67{-yS(C#A-#ry++>1mD^mS{kA%eRXw#kA$@zPSveOzAA?CGq{X5<{6KKzK
zeu=m;@9R<CHF%Md7a6Q#LD>pW^X6;uVyPvD=sIim>_ViP`SjTXt{_2TQvLl_GD&`!
zzrpmw=MnM-_$pvu<=y_L7Kiel*D0ej)4`wIuC9=Kh8EVal-T!{5mnmStM#Hkyk+tV
z_{j)pg&cXV|B6+_7cgDMo?8vRIFTdC;e0PQ^^$g0_vYAa>D|$#ZlRfTm&y17rn1B(
zF?L7q$c|`k#4Dgx$w)aTb1A>g4%)e))75DDikvC+Iej%|t#JLkk<l!My?R}*B9+v5
z+Z!^mfMAUi^OfsCh)LZRdAPHm)&XfUa_peU9aqVgtr!yD;Nvtu!+H6`JL<xq)*-<g
zIO*CrCG0ojQfPAtVlRqp+S%*Ojm$V_nnyh%ZG_BA?Yu`@ltiuW94YdBTH@PwrcHDU
z_GG5#b7FPL_jgJKO$BlkS=Qb!98DBHOJ@7)ApY*=BZ5K%qbi6(x}@o?zN$H2p$F|x
zXv6^*!>~oz{}Ng6`ieLZ^C#fOkWd{KYzh_cy=7@$<?)zKpxM00Gg8bVU1Z+gdV^nY
zCE-CqZt{<iUG<q&Gqzpxkq>y5{*4Xw30-Ae)v;jO>ls8uR2dQWT1}cGuBBs!B1d&{
z%>HDy_6R}S-S-@b6(^k;D-CwQIyWEEPLjo_?^OI#>ia?hrz`!p^QzO*W39^}EzX6O
z_4nCHFkhcB+`Gni4|_z<#&>6DU-!BHdKIq`+ImMv1D~H)46&9k64UK}FrCE4pFgu`
zU2x{t%GS}){q=%>vEHz4=c}>2w_*$A{~`fO<hYG`lU}tb{ijbMYza1%F`>WKidfTE
z(~DFYMe#(@avOfCJ}A5z6cu!5CAIoeou6Lg5!oSSSW+Tdx@NaOUyfRLigA!`!xyO+
z<PvFK%N^pKdbXEe^NZ$}`}n3TiQDaOWpSOlCWX;`$-&UetQ*2D`+5O9&Z?htavM!%
zB5mT;QAtE@tDk%%pZvqBvfSR^fp>Am&ZzSZ_FDN}($0hcM#!E9M1;K8;<SVYB9S1X
ze!8B)b4YX@n*VYoPfNQ4p-=H|HF7XA82=Gtg6C$eRvTMD82=wu#1-IUgHpmJN<L2E
zp+n25HryemX$u1D{A`3dw~?%UY3qe8GYiXwZ%^|N!vnIRzrp8mT+-_uy@-c3@wNpr
zRCN5GD+1K_IU6V_CX~>}E03I=yGu2qq6gms7K<@I^I48)>F<_NaeMhMll4$idmydj
z_it~<k}xQ529>Hg+*&I^&rNLdlj#^6J5k2r@ak^L|7eRub0D6kOib{=C#_*=&N-2+
zc?yNfr_$A<1tZL(0}*k`4qt363`$%WgM_slhuod3HF8{~{_WnV@?Tcn3Vj3G!*|~>
zlte{&G}K>w6n}45ZPQ$tL7#Hr=A_)X&=9YaY!8ca(bUtL(QWST>(`zAwsy94SLLKP
zMb42lD|#UuO;Q}k`>STnWw94jy?Pji<Q-$lv`BmPlm&`0hTOFoQfPeMMo<f1%~v^e
zDw{E~Ji43zzdK-LDzr|5VKxP#1H@nV?L;^NjWoXrq{I)cJgj~W@~A9*cbPupD4q-0
z=_dBD><@`Uo@Z`lNc`1ANYr&?OY+=-4a9h!Fp?K_l@S$3=C(%)ejt&G>^}Q`?eecn
z%BRd{By^E9_q?auZ{S8>b4lv&6#Z|ITP{*<V!&%Pu}br6`8E6^<(koF#Nx*gZsw6f
zyh=_ae(RcGzC$Npg|xF$MDcMtLAJ@oIQBPFAqoC@+iX?Lv?gc>O_b1828&+}z53hS
zC))>SLNhqaeS$-WOCOdxWnNuW7faJA&g|RDd#!u3)kr0cJ;c2Q%6W6GxL3$d?q9E~
z^CPAWkNIz!@kA%r(Y(8flTI;GlB&{9WW|TaE{w-7P(e~tcDe5Ak60Q-WT@Z)BYE)D
z4nZOd(d3>90x2?i+W5V<wO-5y>I!x)ml!%UUTh9+Ui4?$&6n!i&r8OA=J*qrzZsX9
zw(??~Hv<&;TROiE|N3y?&70=7)9|>75tqN|17%sTNGn&dbCrTb>l4bqtNSOCoR(JT
zyo2(5d<N>rMqTk2AQfZ)`BhrgqK`n8Q{hI0AGSR~9gj`|L+SEvTf<pKtR^Yf2$jFI
z&a+@`ZgG|cN-d<mMCEjwAG?sY^qDw&?7P`<@dDtJ{)2BWiq}<o@eB3?iLANup`E6l
zx!RABA~^=lD|#{`zv~Rcrj%v7gCqxkc;C7iAs9E1<5ci+C2>RT?Fajn?@7Ky-;oB^
zaCa$ZI=ucfCuM6qz(@GC`<eYW2L8^k#8q&cq1C7ehEh&<sijuUpJkf;)uCtLj0J@r
zv(d7z*Y7-eRqbLQ>$=WKwfp&MWdDsSu9)>$1~>wR2P-r+LaUD|Y=(}%Tsu5-nL;m>
zFy2(WcKNaFSf_vShLi$E0D|*XU#--d?a38Y7fSSzX`;!i?PKM3ET8jA%^9qW;L(*K
z>$Qlvxvr9u@}VSPVbP7c%KD1bcq%I07ss_#L-ir^U#&G0eEt4BmBxHevxL48Yz&~S
zUi21_E!;qrHC-di?l%74XTpgoE8{>pRY3vtqaWh;VIY<<0|jqA%rEe>dO7bw$+&D}
zd&Zx?C*7K?#B4}+t*<p0tV?wb4IANTB3-^O?)7W<ItM2w_>$SyJSBQ*E(<RlCvA$i
zV(PVJXA^o{;MS@AA=wp|BLNzmO<Oe4doQ}a58-m$#6cQ8Zw<4hcQW*zdzVIK6-y%t
zeL181X$4G)1<W+1W%5JEB+}C>{AaIWd%)#5DXx*8o?dK1Vf_85tWDRV?fb=6O0i_e
zs|%E@chCGfBYekFRPGGaaN|Lh^-dryCrF46oiFiR@i=4>)G^fArCMA}*=5R|xfom0
zf;rnH2_+nU-G_WPL|t8LqE7IUOI$^z`&54y^haiD5QE1i=CXp6Nc!^SJEj8W3$9no
zP3js+BFpM`&NhG8%Od4s6E*Hfz()jfaJqdUgI<y98ZTR|H=B2l_TI%jSc*GX>${X7
zJ25wAS)&K_2l%x)8l2c%F!O<2^2#x}L!|e{U|mC$%sWk_8Y6@_A=P$@dp51)hP)k4
zz*x-x$OMyp4HD|UpB~qxxq1HZK3M!kR&27}(PBB|F-z&@`0}XbnKO=KD(?Yi-tLmG
zNDe!1<L%jnxrTM;xNnE0?QNRJH$ELtYip_3%p8|mHnx3bc-1&ji(3d~3z(facYyA|
z(uwdS;W|cD{x~kK&ks20^)n$<KsVvgi?)rOA`^L4>0l1bU`;Nu8#)&h&tGdn#?H7M
ze8TWwF~l?9;|mvA?9Vd8#+=^hP5bo4hvG3Eu@-~0i9+^^ICi?F{sQz|KfeK~K@%Z*
z8lTYarnZPy>i7hj=LR=j=EM`wHQ$KzXCejm^H-y5>>^%zdUIMmdDm3N(Zj;BwYQrr
zrH}l9=D1FL|H;)?b#E;XuA{0T*e;RX?k+o`n=&Qnq*J55FLZC;wzjrO4lI2NWy4ui
zXhQI62+D>i4Em9IeIX7pfXeseMmlYPy0Z0wg+_~@B!0w%y8dXH_2ny9knyMHJtXp#
zjvFNo7&us21WpX<W!h`FfW9O%NN)<9zClaHCVsspzb^4~rjQ_Mr-X$cPQ<9K>kY#>
zGR2v!nTq`<GZPe}(b0t4%Ay~52D*nc{>>)uU<jnE560zcoO7r-qO2Z|YV-eg_W{aq
zA+*{9tzEDC{tYA$NXMXh2qdVVjg3jo3ns0=&e=Mg@n<pn%?O#%h?O%I!;WuYKnaz*
z4p@IENj4xRPwx$CHMX|WbcTcU_Y82`L1H>EJiG~{pHO1d<sPS8kZidMUnZP4SO~;=
zkgFF4i#ed9p&A82VVp_#=e1>|KP!yJqBE8k@?<zbnXK>ZWEeaJN}=0czY<E`-cS-4
z;V7R$BN(vUEs><JpzIR^yvkdIgda8Y(q2Zfe<5P~eBsg+ZN2Zq%&9+VTV@o7vo$C2
zV48^s#pL7^B+$e!RNi{gM_p()X1=r08$~>$V`l#^NI-s`0yl96WYINM?j)918Vj1q
zGZq1rtP==)>Y3=8hK7dVYTF7lg6o)=$$E`d6w+{bT+2~x1%BN|`Ir$av-R|I5FT>s
zT!Y(w_wF5vD{Q+tDX3ax^pz%6D(zWTe}O?e5ol^VcDh3*rdm@Pq|kuPFv)umgYdaW
z_NGa*SdQ~dPfy$61?;RKri_Gy!21Jwa?Qd%oKpEP_$Xon_>O}{9_ImcDxge3fqUcH
z$p@}`+TbaH5~2Wl)Rg>}Jqbx)mESY^SU7_POE3l`E_fSYgV#}Y;HRb{J~3^bHnsJD
z&<67gsa%N`H-cqqA)r1`ef*P>>16vzqU_&Mo@fUa5E?^93&xu9iX`!hr}y%3tM`XZ
z(P&i)$UBcMCu-WEX6uJyeW;aGEz|(bu@J0FQ1oahR=$vyZq#OkLeo#+5j3~7Bm%h|
z^qgj0Pr!Mb$?ebz`usoDYu6dE_8>{3I1cNh%|NBi(5mH2NJx0Cudl;M*aUQT@K-OL
z@TgQ?qn+=FX`OEHfqbSO%r^4Jk5Qihi?~OR-a=c}d?Y8g#V7}ihs1jfJX9lvY7Tyk
zRAX}xMTTMy5s=X4fTv7ngB|)YVgd$+dF};(;b$oTHh&}O@`O_4R4n$zR}Rm(%)Y#(
z^<r*PxHMO&#_`iY@OkET@&1+F123Hrft<H@#2`@{*`KLXymRaC9wtOt$bTW-;+Ieh
z8a>fwF$FtwXkVXvbWHmfB~0)<Xap_>G?Hz#Kb@H6+6M;L1*e_%ryEpnQj?Pzf%IBc
zUCoJr)PbL4PWklw{I7JGq^^}cl-t?~tiN)D^YQK+wWmO)j#6(J1}i+YZ$M6WRb~1;
zG9s6!$vfW}r!{LUDK0)8<2Zid0S8C;lb4aRqlTuY5;>~HhqFP_>q8V=T#>*9<)R=s
zuf(^o07-TjjRt5pfMsKScH(S;0Flv%<bMO>1xDvEFi_K5BkJ?MOm~8JBS>g!;l*5s
zpvSjtrfkcE6W4=%G=9c-TxW{+^|@+RYNFmBur}?i^OrfV|6NgA-zx1_LJOuWJj=ff
zG3n~*wfiez68U}rdgaAgPA{kNDrdB+!)c#`(Hijc!$huJl1r5=Pb+gRg`^_U8`xkL
zTu*TL4j**t&c&RS!%Ct4u7Yben5%&Xo}>%0(-wj#n_M)M2dVVKeQX8>hH<AEEHt1R
zD4H<sxK?G?&nmeCKoZfraC+lqA9ShsG@_22b@r*Hv-=!(F8a_@6*{bx)$OWI>ct_h
z-!RhGFY9FroMSMOT@(tMa-&AG#9Gy=Xa>i4)MBs!&B<5nvUc-VB^3XpySPLSt=C^>
zsNTH|q{i3qG@QaJZ(z1Y4_r8pdFf*KVGZoMoPmV5^Tm{rQi%q%_2*qzp<!yy`S{I+
zON87Gw0YJ0L{U#I4PYNELM+B2!8HTVRi|*K(f`t@t;PFtIc1=jI@ag`MfevCN(&5~
zIGurqe;8Z1`lLCu6~+1nwkzBg)`J_IwvaA1e+lO>Q19aN@3m##*cW!vn>NJ(obUL3
zQj!_3if3tz`QQ=yzZv_n2M^fkX6i4!@kvszwy+`sc29D1bIZ!hRi2dVrYq%<0Cx&y
z2zuv^7kE9O@t>ctpurnzE|!MvCmz>&1J`CpuIrID6nC7**XQOMz>nH#A;$3#1dfLC
z^{%49N%Do1KNf)L?5s*ScVXg&ra%rO5DU?QX_V+Em>lt)S|ge@F#f(PK}3}mPAjLy
z%W+srpDQ&>7^@w+KUX^b9;tIJuI~+nJc?f&x0cPP|LXpqlcD{f<=52Zor}cM7jrye
zpw#BJT|=P5ODqUY8F-ZOk{_yiV;|;4<T)A;<*AioCv_`N4uU8l29JXWu<m~@_h*So
zOD_Ox5%#io2n;eakSVDZvzuTO^Re?7A0HRk8rB-6DjKJ1nSad0v=w`pa%;!KxD(9)
z*XhTPA0Q3zwxJevya(<#58Jo=x)JhnrF%~2M(6RwJ&C_(?=>AZ@IpuudXE82oM;I)
zx{tRGg;nB`IOs0ZnggE|tu2O|;LW#hj1~weo+SD6MJ{~l6(w=LeCT+qBye_e@~_h;
zOcm%K(uN*GV0;*KM5`p9!nDDz12X1kw>Id@xNyF|(cbzWA3365S2d`ajtI(a2F*OE
z+mrno`L(`7kD^Wni?=0;B;C+i**b*ZpM;E;<%a0-@xu1~hwl*76enSa1&ZoP0u~){
zsQQPaJwds)yP=hq^%1X^2Xb1W<w1hHGjVfgXXdLY$rRAVqM_T(vi21J4YYL4bJqNi
z=L^=9{vZJmte9yB6yP^NS%2gU$2T`h{*fH|20`WruR<Eu@`bq**wdr3s*z{1vbTNk
z=`IF0jC@Wchev=;E|kqC=WP<fRK*+ZV?O5}HY^<2yZU^0+i3k_xDCty@oeASd;(P2
zBIXfh*1Px`m!B~WICBMkS?=|LR07}mi7hKp(86=opE2xMI}4qLVDnj5+t7W!y}79k
z5fmD$anC#0o4<npaI`G~YNR0c0B%W9(CV$bfGVlMzW4)r*SeR2g;z1JTtazCLze^k
ze5g_`$^!=t&b?U}uE%tASg-zE0%ZE$pYP9NEIu}@AD;7*9ke2o{?Bmk!AS*WwRhg$
zKrOuoRVwfZPX;E2a7z;EVeJvE9C!iGsVY6v1U-bW?s(}pmRZxB*S!4Sa@B#T2S!@N
zCzS5f*sV*~Rh=q76zq$5V2b?4zoeyEpU%eUxIX%bfJUMxMkNW1mal66YlE8S_Sl|n
z22@r6`iH=l4BRM#9z}faFlYrsIYMomvY8IFA&w5#n&a*V8bIP~f#+!lgUOxywrN&w
zji2#oI2_mS3knLtV37|VUcn-o-KC|^4E$(J$NcZ4c>I|6*^pIHU0&TPVe`_`eB<qZ
zw;fLR;K3WOtJtr=yq+WQ8bj<GeI?k7jIdfBklhTrUMu<|D{X2=;`ibdQrO8yqnJ=Y
zUS3u<Dt7{&=VqQW3e<S;&M4q?7YlmQoi0NL6-(HpP#A|wF7@X9`;rjNe~xJdbTL$6
zPeUg1B_1|dVB&&5$s)L*fL#{KcEr$d;Zb8@e!dP!P672H_!gPSB=m-<<~)JB1Miav
zI8%{XlZZ)JZ@@pj$8Hu66PbgXduDR7PS2nB7w~`&0l)(xmukQ{<hi7&%w6AKx@}5N
zV-&O6BYtTFygehb=7-l?1lApZHq?qqead&aJxU!g4zS2XdU|{E3rm6Wt;S8PBdjt^
zYo|K55L{i&YvRlsEC#Io*PEh9TwPGjW<|v5EP=*5b0)JBZ7%5&bB)<CSr+Es_9x_5
zJKtH6(fbv(Nm4tx$fgzW`<I%CnV*Z#&poCU$r}uch`9a+m4eIJRrLs8+I1;lX0f)j
zdERhZXi`AwzI694A(r^pIw%6S3|%vokdSU17{FJf(dwWR53LZb=jA|pdwOi>Ssd-{
z8^*_LSbL}gp=%RHxD8N)$(p>l0xVCkd(npV1HkOm)!XL^^KSfc>HPuVl>_R10l`g9
zK*r5^n*0d?0Ov5hzX7$z<NWl%)dj&ViG#%~BWt42LLVy<URJO^_C17<jYfgCXY*Y!
z?)=#aa>M7T`9m@mnog!WwSOWZBfq`XG-YQt+I4dOFP;9K)3x%-N))&pcUhtHIGQ7X
z?{)+#wkVuOx$Eol*~(vm-`<_4#b05!^j7>aQiX?;dz)(W2?a$jz$f6dNIa^7yU;24
zppf+nh^Q9tCf+VeZn9ZBG_*@1WdY*-7&U!4t6Gbj&{4s2Iz<a1UH!lnE|+H#j_H$V
z*z7mc-WN81<B$EJW4Od@Og$6p@)liBq`CjUBQ?AUa^`T{p&*@U+N0;6@||!rle%9(
zP-FYK5?ImNfLVY4K5<P42fqlI(R1o&Jm4A1ZZX>Q3-P^27vl%FXN?1g*f)!q5X*s4
zM&2%*!3$0`rVB#n!{-odR!F*tQms2sU7cWiZ(JOohQn&@#YNxMaYy7zUCR?lpHM4d
zNw8b$*Ii}lUfN4tp!$nkP%$KmT@FZn9(wuEiYzB$A^L;B?whNN3N{H(c)2aIU;`e(
z_wV0VE-||c_n8EJETK}U1Y^_OnxOQ|`ZOl;OTbEZ>(YJkHyZg);9v$Y3LP7((#6-1
zFV;&j&5KDvTl^)Zk^rgFJv=)G|1BR_WpyQ}%J|mHE!%(5K59q*=H%%JSKu6hwasAl
zH4zTu%_C$X4kexkn`62AaY;%q3Rv50*;(r}o9fkH%T!sIjMF_i<NW&0WZ#+bd(U9w
zlxCetK1?(v{^Pg7yLN=C*RLGP%-Tnstm+t`sDgYWWuGPsr-VCY{++~tXMX?__4N2)
z!R)C~*Mm+bM~38q+FIrvDl#%J1td#Rg^L(~)FbI<+>W|`Y&*3fTi3^!qVep*!*k-E
zJz)4bEzth5<0Y-y<hjJ{K)FfC?MJ*&7jZLv^%?{Ic-qsj&)9@o*l3eu3RySq(*8H{
z(>kxWga=S^^o48F(}9KVj)+iFclb^<w32tM)t>I(gMP`>6gdR7V{Y%B-RP9bkA9lR
zaCCdjaZp<h*KYG|s>OMU%*Xb(L8VgyJmkjt{pl)PXKcWG)igRGby$;OkIlMlaz|-y
zO@(AY-LSJ6Df3vtKbvcb!2e%<_<6+Zt&4eug)Bg0Dz`X*;~U)I(4Yr-4j0*=KLH8Q
z-$>6AYMjuC_M8~F*e#k|Cp<dyj&Z&Z7|pHlgP_MHZoU?vG|FVjHVDlSeGw@k=DbLR
zSG2{-D_D#%>{@RrW5#duB@BAlctc?pqagD~=0+HB*gJ0%5L^{7Fn9up5RDZAw6wZ=
zq@;0jlP{&EJBJ2;{hB5wL*b8)Lj79@3k${@H$wvIO5({-6@h!D78fS9YIxdB1}3)@
z=^!6NY139VQYt1c_TEHlG8p0illVCOq=N?`>#5@ScisP3p(DknqOV@vL<7VrpDBO+
z4P+Yf^;_}Lfcn=6^1)~vu=F}=AVMmz1MWZa@cCDT&E40p^GM!oi;AOx8yXaI$jPbi
z#g@!dgc5_|N&SiP|G5CJk8}heJJXJRkg2mRxr%*1ypY$RjSywj3Kv)cJI0PR@E`yC
z-7@4<OLwEo-QJH;oIhk_{o?a9iE*^}LGsOpxz%*_dZCK(6K%v;98-xxG|0hU{KbPq
zGY*7=_gD8vaHM_`85i`w0X=@(V7LkC7d}H*p!fi(kpPX9lr&X3u0xd6{kf{DYR8%*
zt5FxKO+^_snVZ7ujb_8#kB=N0rS{KkF>=<&h1AidrDM%TB7qB^<X2l8Qh<v?&CkGq
zz&rlcB*}R3NpAj?x|b=X?hTdrPv|6d#i`@YOFltal6-bMME&4tj*6EcF*oCS28DIG
z+?Ri;Tv*k0`A(jLKYoyP3J;(ew1o=-bq?Yf3N|*jg?&&mEO6S=?K5F<#l?8|{8kL4
z*>3F+*leE&&s`zF>T|ei$&ECz3dmoPS-7p7Z_YhoHc?L_&C@X|LuUQ0u+U<9+EWLZ
zZ7UV_9QPgYJ5q$q4ST)NX;>_{*({Wkr)bU(gMgm;i0QF#c|zhPd3$-yg;IMfD&L?5
z2CMB8bQEh?IIre=KqFUVEb1Ua&CY9CaK{AO8%pUw20xlZ!tQd~kb^NioNgdDX=;N!
zd-#sAA@rFb&r?@=<GT^!-D~e#x<-jv&m(jrlHZ#%5e|&KUYgIj@5|9>$&xxVBeVOy
z<mZnZ%gPdQPe(ai$xI#903TzkWp=(;lA>It(z*l}FVCl8w0c7W=JB$Eo!gA*k1n62
z6?@$(3%+8CSl+CB>bJlb(VF2C7S1`1E>CTnO7vgo`Hs>Yf1_S4*=zXss{|qRix)2@
zm(jeVRPg@M!R$0oA#Qzujwp{_Q+1xDU%yy3@|`7w(PNBox$LGJdvuy%+atLwy3?Nx
zmyGS=G@U9c>T;E@Nl<y9dEE-txI!j*m-$-p%X*}?Wa4<=HKlCD;>ONO{Ro>4fuo=?
zfy~(TGo>2_{las}1|6NX-?UFZOfUSas=T}TJ(E^*@^sP*C!Ein3zAn1-R^MIlm+(o
z-aVY6E#e}E+DPXyK@LM^DREmK4qFT-F43<dg^3WS+(a`gM~C?z1m09E28PcCBn^;m
zNA`=YdEc8ypw#n|#@&U0!|nKMQrpoHfxMr|+}?0gj!~oWlBcy4ue<|$RpGW8_WKWb
zy#JaJ;e9F^Ze84HBt_bQrU?QR2A|fUpR!?Z>5uR828y~Q&8~?&C+?Avx-VSXmOluh
zO01};D_#fU>UV@wkQhRr9&$4d-SVBl3!5o!eM0SV=igR$py`plBV1Ga+z_}Rvv?C|
z+HP4~^38e!7J|!V+HS=daQeWs`)<UFZ;AY!QmFgT3&F<%s51<Y4+{s#xlUB^Ml6o!
zSyjY5Yu<`!3$5*Hbg@a4QLx~&UOIU}v4To?{5tZGapgYHy#Acy%3A8ze~8+gzW&?(
z2I#&xbi@i1Bj*>v?&<KlP@ejG7y7R5k)r3WM*?riJRc^&1T<8|{NMFca-ZX?(9oMZ
z@Jj&Gi^lNLJIbt!&wjj+6qCW1%UaZSiPz)<#xXlFS6j3UnZ6*k47mdl7R@-~5>&Fc
z!n8Kh|8w2P%iVf1{R%0!{s*TjMXF7Md%yE9-Dv!hnpmoouvK~=Lzdm;D15*1#;dc#
z?j#;J_Yh!B^%h8(-?{WJ_X8x8mx>IVfW2qU1~U~}KS#ixTmY~s#8;H3?S_0k2{otH
z)UTJ*L2{c8wru>ITdxo(n^yHzQamGB`RT~*n9sS%2qePhvcOoD)JQG#SfjDe(1O`j
zrP%(c5u58?t{JQ2Q-)@IlFdB@Z149`1pnu#%V5f?b3HqO-qAn1F76!;(9FkvlznK<
zNC<si`p~e)S-Q8_{Uh2Aj(V-9Gt(hh6DP;v_(+65?`YElYCcVjs>I(wyCpp6yYt0$
z`T7?Tvg@EiuWv9tGvyict4OOA^-B?zbM0>K$CSsS4UPY=i>uXooJQ|sID?gmMgFw!
zy{8@juB93yHPodXzgH$weuZyEO7fv(vTVktjeSFFe-%G2@_NZm+y~uO?ZfNy^QN8_
zQ?>tK3kmK)!y{7&QRg5c>&sTzZIyX{Gm*osnkU4!Ec@+Md|`(#7>>FaI{J{|5j1=^
zVt{Y3QB_hsS4{UG1i`^5G}w%VCPpQn50HG;V<bdFMKiCm`kmLafM{xC^L=Clxu?Wp
zNPqp|a~&dutW2|Lu@4mPFL8nsy;_%?Dl4Z}sDv4<5hLYdj0C0J?-3(iHJidi+xGM9
zXjqx@`L7KAcQA@a4$%>vN#rOl;5tF9Zzkmg75Q+B_9&gS?f|@Cv<w%kj$?o6_kXxa
zFh%|pnSAms5BqUL=b<~ITB<u%bNY><qw3;DEC*VwS?{n=#Q&L2#`u=qR$xH<h>Yyh
zmoG(jMzA?YWF?Wobh&2ia@g=QPCDX>vMuqOFqgd|Z3m(7g$xJtO$SG>+L=kWOk=^b
za;(On^y!sngYz%xqW(vR_;_1y0CT+R@aRYd;YtR6!p~y{p2p#7B<Ym%c4YapJEe&m
zW|#RYujQ-k8jpS6$FkL`fAFrjjEjqTZ74zV(1z2^hJ?>jxk_dElWMr*g*SJhDjoA{
z+7jpA(0{}5suutZ0xfrB5X3Gn8mX(Rr^{#DCgyfXcH9_ua*nXZi@tf_DSL!m4vUTL
zfT}cfYoLLYVl^u^RL~F7mseUBW`1oh8*0Tz-tx3|p-bsfBdms<+f$=7eR8fj%(6yt
zyBR}9GLLV_o8Rxgeq;9x)M%Y**`0>*YxE?aR1sh^MPRP}_FSm2qd$86FP5UzfxQc|
z8Z6)i3d!^B`-KQJO|Gu460#Uz0tQcVJK9c(PPrMXvxSf4IC|vY_e}@o<$3MewRp%k
zLBet-);i98|1Kmp^uhjP(as?E4CdQ=LV%B{n0MbX318iaexH|Hw|3EE9~ytv9)6?3
zkX9n_h}t5-tDeiNoIXpE$x_+rcpW*Ic|D<-5!q*5{eterv_Whq@|y^2rlYQbmBhR+
zAjHyN^l^2{zmDol^lWS)&_U1Sw5biZO{rDQMt8M6>0s6!Gu$~eEtPaGU&$lYppt+o
zerl(aKMg(#t~7PS8By7Wow2`%hO9yLhS}Q8G`7<#{kg%W!RyyuP<H?Ub1h`?zmEJ%
zb+FEif#olj4BQ}217EkVks;LEe*zhr5BHv!efnTv@oECE%Isn9weGuC<k#ubE_9W-
zunp86(D$EGD|5O&++J9k{yvUwOIT>c#J@hCRivTpjHl`<gePj%FA#`LpCV}WwgBtP
zdT*4lp4Qg^%)e^~nf45ej1*t!h~c!Ey+%Km*<<mWnYw@D!g)l~q{iH>%;Ni!K52%w
zyMr0gI46GKxVbrES&ORAsZqK18W#^o(=4=p&(Cq@!%4h-ME=xnu`7Do0>17D&gaF^
za+WJz4+nok95QdVF;)rWU`PuxB^o-qlsLUpG4+^6yvJN$V2Qtr|EWG>a-^iYJ?V+L
zGJx(|LCE{^{kPi`uC9vG5l!ihu1qsvlhHZu?k%d2V}9S`Nsy!q`t3S_#cIdoYagx5
z9(>d#Xz&uf)tB@@Io}gK-<I#*?^hKVztM~z5hovvU%Z|yuuRT!d`$M_t=NNR$?(<9
z%DfeI!zg6PHOKKu(Dj|y?UhA#JNS3`8|Sz?m*-Mq%~e+#TsLH+T<m#YQK6xsO=^D7
zS%f6>Ai!^h3`N(y`|L^{^#m4`lccID8tdokczUX){iQ$bkJm~~A<i`Eim!j1Nn{l>
za%mi<{E9cJM*z9~+OPuT?SsR@!m<@jJug@yAa%3t7e=d+ND=4OGxqp`fM=%S$1VtY
zOuj#RwtF0!GwiS#;1|b5hiFA5(kRG=8N8I|=Nj(SzxcQ(#rO`Ln_l*GU|^Bci~+Hq
zkJ9u|VdhPuL?RiP<?+ObM{m7&HJ+gEogJjY`k{V-hbyDO>)l>%R0htJN;q<=NRPGo
z*%2G>b`zN@^nX(+E_*;Wlv>Hwb^9#*>xk>7R3t0ahST?V*SEJTHgO>t=iCSfjJ+c#
znw3lqAK&s&W(rBg7FV^^;2M#0YkNiB0o%JuN`gl}zO#N)Q}Qj_Gw>zmc_EFtm={%P
z_h=Ly4Nah2m+(WZ`@V{p>@hbo9;9XUI=|7#xmN<?Q+CpX$MH)jJUTdjA_$I*q>gG&
zg;vcQ>oUcVfEhAl$0Ff5tjg0$H#7&Yas9%f;OOXVu=O#9lq2MwL(SNuz;vy&YG>7<
za@-Y@Dt_gET!IZ=Hju#qSzD9dRNrxrV{?52gJsaoKtkFLdO^=>+gI!N=t74pbRc1o
z@~HzL2b*(w&<@&bWvIn<;rHqBa)j-ah|jHzh}Gx;xDP=jBmtN1M^ThxqC!GaK$~r`
z{wk?Fyir_2!u0TP!ZBzofpUplf^GQw5v9%8sBh5Y!?98J?WI~(=h6WIDd`5XZi5yc
z&daTqRkG3Nd4~cwE;{?;q_%>8cCw!@5p`E*r}+-yhlH2M`=+g)UY~j0RXy0FGDU?m
z^^!zBW8m11@WAUud()F|zoKY9R3JRn-7a749CzDyN`FtLa7D!QXu*pOr1I=H#>d(f
z@aY*DUqP9+AJA$sq{-~pkm2IGopI*{VkqhwL+AX0`>7DGX<=bu2s{xaJB^@PurM}n
z#X&9xrZ}j<gXD3D;|SEFi=n9QzCUERJXTo>DpwdMb{;w~&A1_Xjdhoq+>)MZLGDB%
zl(rl4^=v@6dE`$eCY2`V&$}Ov>bKdc@tCZ;fV_<6khw!4)dz`uTs%DZ<~!0nUpG!5
zPh<+|<47Lo0>BPwV0D1U5qGzYcaj(O;PL#tJj?vt_O>=VTQ;;dBQ3LR%@}5d!mzwo
z;@}!Ta%#r+M`?uJWxZphzse!!c4ts<VCHoe!<8NM8}=uCOKzoVwD|T$zXixFkZP5u
z4n|swq?xF3M2p>nIUyFPtGd@(nxJYeZ`v)2@1r#C6xKuL&yyIzVz~SRho6>llGk?r
ziwE>_pkKY3g>s-LL13I<Vg#UWHa$IkXM6_rjIdRN!N10Xvvr2Y=^tjyXP^}(16m`j
zI^W;jp2(~BLawePc$4&mGwFd)+GCiMCP0j+fI7zBZm%@zN3zPw-2_V2TfgqPKu!~;
z{O(d(TqYF2F|e@}9UMv^jh2qe>8(({0(&7StG-myunz$}qX$|2nILIk0_B^Eoepkz
z14err6X#Wm1qRdm_*e@b&pKYt-#|?JwM$2K{Pc8nM>Vj~?pPCj(2U*AA;LNHuTkHO
zk}Y6KanR{8S4ert)I+TFB8Dtj$A0;kUoms<$B(_H<2(Abb&X0GEQWK`{lu)wdbKgB
zGilNiSYF9z3dx94*!rhX7lpuEQ%NBsj_K#&N`1Doy<A_zl`#&8P{R%UQ<#lnfz-c>
z9E|VR#(&1hyqWa-w{i->x5;!?`{h0{sAfd+xmU=_$?c5s0ZBou2dY&r-J$8|AR>|m
z1?OZ)Da8@2gL8~J<cS0P{r!RHkpf_$*`&B-WJFC#(>dh~j;p{4Dlt?w|A|y|dQ;O2
z6<YkGqobu*#C5%!*m3lsSOXD1Cnu+6!v-Xd?4W!~Lr(r2-$$5_k1tcLG#B++#X~wa
zYEhN(l>s$N1<;Oy`ni6-GVbl$i8b!W%K&Xd$=y%38tKQRrEOg6Mxbn3HIrDR&f(Z#
z31wu|@CG%tbPTNJ)ua;CMnCc{eK<yp(6t-URrd%~)K;*@e`ZaO+;yvt&f`~qhOsqp
z-hM1r9qS@u=(FW>kP}hZkF?Q<^F%$;kQ4h;kg{uWXNjMwhfX1-GBJh1RAQniP&1oB
z;b5?*U9IozYh%&r=2VaCP1G=>J%&^mh1lce>SJk-hhG<boijkUZe_}wG!#j`c-)IB
z@dqZmZd8{(pYgxTl3FL7I=_F91o&M-&z|j!X+mVNvt2Xo@@FaS#)%8izaEQ$?hxI!
za(UE*1~iJBKpFazDVGwcvykJRgh#!HBk%0=c!4->B)y!G)O}ZGe9OFaY#z1O{ru?9
z-<v&g9+2p!Uo}XMP)uTd!*?ram#D}nV9n2vr+C#`YSGJ^=O5O~NYB-|n^L=yA^oIz
zdgU>%k~EQw+GpzIm@mSzjaExPQe_W5b2%&*{oy+gl#1h2+vdhVJ*-t_K)_8gS;b#>
z!x~$}`EAF;!(`Xzjmvc`av?Qj3Y~D7Dut<l9_14AR}i2@Ob(Ws4LuU8SkBDCsNUB)
z|D*J2J~$c96a2$HC{~HdLWb*W<M1~fFY=}NC`)<=;*nCz!60eA4A}c=C@EjCN4<pq
z;cy8|MwPf++X}v~U9f~vn>`qu2W~%(5=uJfIXW$|R=%#VCXPRRtA_ht|2q9u<jwVy
zDuWolL!DgO`thE7zR8gngew_)lw5KW<Qt@Im2w9}cj!mSsMsG0;z&A)7QFbyVxK~P
zOE{`=5FCp{E11=<>Pj0OdY{t_+oPWTjYJ~Uq5_j=gDN$M8o$P51owL5-n{uLyXM<^
z51?s<3|V2QPw`ozZ^5xgj)+O~?#X{@EEzIG_uzf_P_Sph(^T*gnM4iqunC>hmZI2x
zdGi{+)%q0^;jL*OhD>14xIp7%5ztywQd8&6XG2GH_eT{A4QBS=JcZK*ka-CC<0nrt
zp(yePexYOJ$O7tLV8Zy@neHyAaH_7p96}3^>t$qh=~)2%>6v_0*$~|t>~~HQ(hDbh
z+bMaqGU6^G328Okd!^#7O>HWtBMF;Y>gWY8h*U%|cDjomw-)J0ck>HfL_II125=R^
z$ce{Jk|buLe;3OS?X#i+ZL)jju*riAs8ArQ?YyK|pbnukjf+Hq%^sMd2HsnJ^7t{H
zi@6Lk{7n-Bk6J!JzRJa6V?2jR?^jc2r!?q&#sD8GI3(n8S})ZP#*Puvw(L=oO>s_M
zRvR^jXfAtk;Gl6aOaj{?i0^RHtLpp=2rqpwvOSdO0(VPoMzGC!tzKmkP}<W^e;!6M
zNy+U$y>gQ+sL{$#tdr|uX4z}_K4GJX%#7dCC~&@*&3Ug{?9X9ioCNz7)MM4HRXD6=
ze*Acic;ru3pr0kf7RLyF6&=;Wo#oiTk?*^{Psanu*M8?e13i`Ho)jP84=nB<MY5Yy
zFMW@Sp;P`=x3&}d9IVX9#ZaBrU-MK7UVkEAS}UCxc=>cJ87Lof=0#6VtC6=uZQ_q7
zLy{qPySO>^qhcku5`#Al+LA#&G>*h+(zA#9lEB@kj{VEgURgAA*@1K!S+)*M(WNFY
zJ)biCu?=EoUX!h4r0T1j-LcDhppxvIHP}Ka)TxDf3W24<Au}RQn`F!%8U1pkAyc$x
zjcYiT(Mfu0hzz*S*exJ2j-f&!!g49%c*PcDs@~YdJwUbHJz9v92#j^xZkCMz4uwTv
zGOz=b0Dv4>Ow=N^sOp|<L}giS`8<(W5Bw+%0$bdr$kF53nHjmsHOuO~P}k2p@v=9Z
zQlif&wA}YXpu4~x;0B#mjXE&e`k{Uo5*nJSQivD;MN2hB<>DJ=i<z2J&a3SjW}4@N
zdr>xxG-j~|^c{ER$r>&%J{NimC|l#+4b)RjvIF>(1DI#$b}W<=!z4ugD>3?KQdIG3
zZ?y1d)WsV|KF*gI1BtGR4+dWh@s3)@CnnB;W1_{PG|O}_cNm@m^G87M64Lhp>s}w|
zR?0?{l7p=8ZjS_mSm1Dpc@|)XIrAPsUJz0}tUlRpT`0CP8qAGJj&8JrJ<NtHCNLS2
zpD5ddBlaQ|GJ?>2yqI<aIT%)Xr~pVX5u@GV{2Cr^`zm45U_%(+DC8vPGu><w9);~f
z#^N+jpnUTaVo9X;)4R*U@3sD<9-|)Z%?TuVSq(ex7h6tf!i3A&WufTb%KcPu<c;s~
zRcCSX!;mVUb3g&JEGuXrq3}cxp2kPeVbToT1=^uR;Cjum#nJI`*svRF)PaJS01O^E
z5c}fWC;`l93fy-Tj+vWN$TSu9nT<dUW4l3kcMm{~3H%H_mgRL*K-COv9vLu9swHL-
zs6QK{IQ%EC0BzG1yD*@adMJS=Sn6YmP@|4$PUsH&1X8`z9a9&ApE+yXRX2Ckeqlqj
zU4VO15IQJ1zBJ4lNG>Xsqd!G(@9*2QglKFKln`;@jhrG`wcXj>#l=gFN+E8#gU$>3
z`qBOMdxg6@ha{>ewHI1iUUUORAWKQhz5iKA90MigV%6q0oSgLiM^M@=c0YFTCFR$I
zuMF!qQGgT%o5QNI`DmF8$R>Z?J$PamJ3BX*BF%U7d}Sb~*Zp8D4YDyRIf@E-ni@Df
zn{N=P(PaT(tlFW&_dyv(yB|(@n~av_gE^I51D+D}3<wTaITw<+Z3^Oy-#f5s#TbX&
z-g3!^MK46v+hdAZigfM}^EkIAT$of>pnh%yae{0WZ`!G#SWJ5;b4=!KjZG2r5HcGF
zbOkqq=7lCaOMXNWH4*|5-I<624J3c0fWpuP?28<%3EnLDLqmFg^T=CX(cza$N*pKE
zk0zB1kQ?GfJ?YbNrzCG0{)sPlSw6e1!!Tl0n?t2pnCQpgpprDt*1K4z=}&H@a~1`P
zutFYny5~E)0-bzuoSmS8zg)G{A_vf_#x9T8_Wu6B4;+MF-RPCYKCgqirbh`7MW8po
zpw3FE#rWrFjk~KmOqE<I^!WQGYI&L+^>Ot3g{LcjK<z>Ebu3G6yKM_Sv6zzN>tyN=
zWxqQKP|x&=Ix#VEcP>oJq&1X|2>3Yw=bl*VFSRv~0Td9Dk@-r*VHIC<x~~SXJ*VwG
zQmNkz8%`0_iH6Kd+}7^BVlTtpOifG6f<m<OIWdscpp0Y!YHBE65xnsoi+dba+Me87
z7s{%}e&@S))W3CVUiz-pHow67sy5+SeH{McaQ3ISC#NZUVw?}n#pn)QW2;JVP&fC6
zIRlPQB5sFl5D`lTs^((~gre1=l!l#M5zftiz(`LlHzW##6(aCSN%H_vYJ$d$5}<r?
zC>B4fJ=yjdF70%l;Daa<=8$ygCn31mjtCafjEsyx8;4PDkCQS@<kWr)_C%tm7SX6>
zQekha*gLeguS-XF)ymmOqtuU!)$9SWc;^jcAM+XHuCaMwrl?4!bR#x;?DWb^()p-W
zOtATjxUg^hPO_n_yx;)iL$Yl9toaw}5zw68HR#H`cyZ0s;zG1knx@B*et)ip8o=ZN
zW!mk7gTdlK%`9*!&~O~HcL8}i$Uk6A$<{$om|Zmk6crlQ-4l)LUS3|EEAN@$Tt<{q
zFTKJBG+w=5=g!$kN)ouAhG+Rk&S9xIx`vMdr$ViU@BP6RF{jm=?_ORuZ)c}JG!K2)
zZ6TpzQ}fDO%em5@=;O%y_s;a?0h6N!R_Diu42!Xf?d@$@;N$iIGvgG*2!O~>lI7US
z7<Lhrj#(gf1a^0wEdpC@lw*~}7#HeBr0gwvlo>{C)4oW0(@L)el;Gw|uh+9K)s7^;
zqa7Q`N?Tv-UVED+dim-~RJU|<tM1n)t6}r8!Jm8Zrp-!rSEY=1?y_Ym$jy+t#T)Hi
zK;4%1ZIHw11#F!LIHh&UY7uZ}B(H0U%fb3fpYO~_N6MQ6U~)S*Ubq0eAE<xOFyuuW
z1NKmlz!u=#ZzC>Un&Nxf^r9nBWHn%%_d!e>?cwY!t#OZz$_Mihsn?ksA;hfKhh2|O
zzn7=lOStdP&TQ{*2afYU;H`c(93pGg6}1%X@j2+MT4gvO`$K=$(lJ4tQhZ#TsUtqz
z9s*F~rMd~1E+InkhyX(|TQLh>D~;R*%7rR;DNlF#O;ig%h<GrI86rGXr%tk&cPRTY
z1HM<K6k;z~RTPm4Gh-ed9xg7qu6Ol1Vz#M-s6BT)SWQq3p;;M!I?+?VN0GAEEE}BH
zM+rgS*VT?_E9aev=m4djItz`xT2X{l=EXy#?Jtx=+q_&Meu+}<%zdo3FRwQwlF_y#
zl};n|nd?k)b;d%>ox3EdXEC{bO`fIcF0WH=xA$g<k(r;blDTj9GW^j!6Ego@_;lj)
z6dFJO-g=)nnMJh<(1;nf-2Z($CEevJw3g`42@i?5odPtg&o$Th91uxwsh?DQoGQ2K
z>oXNI5{ddhQVo0_`~JJCnysi4i)eV^3Ne37y!lm}Wv3IWqvJoBs`AOI0~vJuQSE!&
zm;X+@SK_1in|u&tI#)ZKQ_;|5y;u<IQM0U`%MC;ekA~%Px@gqz??=B^4ynCY4FM1I
z4fQ2jmPk9W6Hw!ih8CY~t5$d1!M#y7$8%Bh?{@;Qjr<<g^Y6~l&{|l-6gPi0Hdba~
z|Mz17rgGd(Aoo`MU;omb_;<>|CnP_=`MVwd|Na$q{r`^-w~X|nqt(s*WW<E7%a!Ls
zi7+v!l|H_DtsD+hLptm)PDPXa2v?%_T=T|RIiEjE=A=<tGCIlFvGBpb8N?zUUaDWo
zQ^=~Y=aGN-_r`(atcgY`Y}nho_w(F6u@gZw9;E5ax-tFI`n=oD$)`Wq?Dgr->VnqZ
zP-pG(Kvu=#sVIr!`vC1uJgmQ8An66y6K_r^<Jfav7dGd1dV#SWYLP>|mq{*s7>(%K
zgccnZ?Y1+$)cR97us)u|$xs~+v)(z?>hG^>6PXoUK}&hS$)+&Kd+H{Il#TE_j~o@M
z$T2yyQVJAuPmZFNh${`q#T@kfeWzl7X~B+}X{m?c->^%*M_s($YoiK;8vVr0zb9K`
zrF1=V(|%M(UHwXHSyM)IMtZ5k+Onz198!;{|6ZZN16vhB{=;a#U`|#zYS)gqc|>;C
zhc56;O(GMS&wCujzS7taq<N{ozo6{_({$2c#;~o0vy7W7?ZCqN$VJS{_RmHQr(Wtk
znoEB_HHUS90j*5NB~@c*(6Z^eabB+SwkMa!i(7oE<>uaC6F#Q&;O_@1m99BmsPkoC
zo5<UhbnW<M!cL&6OBbl>-E2&#TCInKjV%B7jS|=SxG}{FM%Kh7BzElXKAsN=nOnNu
zg87UyN}02uP@f})B=<_}tQ~0kcRFy)($?=xH8qz-dr^Dj637oJ1=_LlX{wyGY1(wu
zvX{-|*H&?M4lQ;qSy4;ykkT8L*AhUI@U;xnpRA-&PRBd)9?}AV(b$N0!SDN}>aA5x
z$LF?MHz03&A6wKoO#b!I(2&_UX0Yw$s>i`GF*#sqFC8YM%QrL>OP|nY%@JdBb90&j
z+xyw$BMKFh^_92nSe7T}su!XIV{>Ue40t~@l9We(cu3KC^+ihuJ-6lXl=DgK8%(c*
z+hLyBK2_)Z{||9*9aZJlwGU&WfO-TG=_B1pcdB%EBi$eko464LL_nmaK{_^DVv`$C
z>6YAdcX!9P@I3GHeE<A@;~VdIzww?Sjx)~TzV8(?u6fNl*WwXHj%}1^LG9h-H%T92
z-e6Vge!Ly4Br56$GCYF%62W}u7ee~t@kXsmFr8{ZETe$Kjl1=z#9|~SIcATh3vC{S
zyxkLItsKFJnKSV{f%YehbUqn=7KX1mYZbVqMmiFKKm;+qqva#{j9T)qs8Hu7PYh!8
zr!^BGDX}nbpPCBCN5w%^`=Ozz?;oamrDPT30DI+)^7PVA(PrJf6k%p&-cqdtl)KsL
z)ECfsv%CB4C9P7*4&Flyd$oEUm+qg3rI7^?)l>?FwDlK(YoZ@mbCJKcZ~76x)-23@
z>S`@Z%bqnsPfbq8tvVX>%{qo5->Qpfli?2<!HDL#je>xvy@wri@*VG5wqc&^exzEN
zG!W`gyJGqCHl_0Gw<TVjRYUYN3@SQ_R9!;a=Qm<gVtoa&K7M&4<m)VQL~!rV@3ud<
zmUIq7a&jNG-QXnoGhsnG=8vqWO^mpvg!PJmOp8kK&`W>!E=z;<&jT++ZW=%sba%js
zlS3`xkK*E<uXOqJ9sBgrd(99~SxDnOk}+}XXmwfbdj8(^WsqB3zTThb-l$BbfZ??l
zcokLYs8dk5v{a&Fcts3!Tx6{A6ZYz5>iB4r%ViV0rPd-d<#KMGdvl$%U9B#%Bn#^P
zq02E|-1%dL&}H^eo_>@_Z(_g2D&!2Qv?owm$*a`uo%`5TBNSPMoLlSw$-XC=_2;`k
zLV+r!MZdm{X)`{3#<`eMQ)auT0iP<>EY4BvuzHrS@?5;jW>78sDf9f6&7Mo%KTny0
zr>1dt4t9FE;JT{`L(D?@e|RbkI@)1V=pU>0e>B4XU+p;f)c?VZ|KHyL?Eio3x&Nuh
z{`kND?%@9i1_Ax=PX8bO|35H~|L!!P|JB|9gFpWlcOOFz`oF8k|L;5VNB;lcdvJ}q
zDpgffBCjJE5UALI=1LhkIW<emT)^(-gL}>kKzx(J^7xN&!N3^L9={FFgih)eIvS7K
zS_2niUiW>1GbOLPJgh*=NQn-dEcc%@0Ql_!@Y}!=gCnq2Kyp*P26D3t2ncBF=#X`*
z(Z8lBKr<>5bc@O-^Q*g0dgkPGfQ;>*f8t^i$xxu{AGVs8Gf<>2pq{HFrcs~{UIttf
zVAG@iq;&g0l6m1Rub>|DzaFbB0$(KP2d5WaQ?UGAhlN1}|AkV~`rObBME^>sz~RBc
z!5iv-Ju*$Ukd3+3v_4*?8VN`mW(c~%PO+&Ni;0X>nYIfwn!mWJN-rLH#x4l_^Nh>o
zh0uay6^d}l`|QDvvr|=vSM8GLTWx`(8i!{9dAL_0uoZHiBRb6EZ?goV??^>@uqtor
zuV2(Vb|(fyC9woay2O6FD<$_lHr#5O!^v5tL6-OUbr0a`1Fk7Z?3Ad4gnV#2bj85Z
z{W`0aYXH(#G8SWY<bqGfxE>#$!g9mx6+a^GID}F_cjX$P+%gk2mk7*}6(#(&bze$F
zLn9VoA3^=QGRo#`lUD@gZ#q>0AvqFqj}BUDzZkuh)1L>jMWFj$`I=q<8N=z7$bT)y
z59am*EQ5v@oS<yjZuq+x*tx!2j|7!6*FbIt<e&@09i~8L0>o|J-rnQttJF(POIST~
zF8pRYyOB-(0bfPS8`9MU+^j~1;QYR;W!;%MjO<Ku$O1Kuts%G?R3Kxf7^h{d(9pWp
zuSC=nRBE-NbG#&BH(aAgi;K75!mrLpp$E#K)eu9>_+(G!7YS#umJECDuwo-{;%ZRg
zS~^T%k(~pn8dm0~lzF$Na0JPJcm;PfG)-O!0OSL4Xu*fr4@RdK7O1hW0&lb<ort%I
z0@Fdmg5}BX0%Q#+O^MKF%JT$evCO)w;MfHgmvPWc3Gk62=*Hb-I*D{TKJl#nz~1mB
zeZR~F1e>j6)(0+pW*UJLMOR%H(hhQo=^!M&*+aO~MuoIdGDJ+ehgVlv`UW$)Bnx#E
ztrsSR8;*H<o4WSz%?#)D!_3+%PQ5R0nwb%t_d;|9adCIu*7Dho#?4m0pak2_3*<40
zoed@d4xM!VIX1?S7~a)@|MQ+}<@|pY6GWAeJ{;lTSXf(P#8&P$V-}*MMC4?VUlrA@
zXJzGdfUKkN<ig<w7uCODoIE()E;7^4&yW4UK`OCJBm1TW#-=6<bZa3HssVJgdLp+S
zC&<qlbYc*!3&Z3tT2O__lAqft^1S4@^mS_jS3d;NH!fyGS*@Jzy-KQ^s#Uac+w9{~
z6v{Y*8D=~R6k;;OL%6sHjdWc7_Hn3#4}{{}8MCdfQnqsq&&#%c{<B1iKP56hD=6H0
zGv(scycf4iuUwD(A$dLoUNE=P_eUfB?nl@79gV-O0Nw4d#dwSHa;!&{Bo|fgsYQe*
zWsxeXgUmEKb6zg$Q8M_u$?nb<{IEP5$*$olNlJo?qPBXL-*v-U7yGNSyQTUmjFU9o
z@i{Z(m#I`%T8U4ICW_${=X;3Vs#3>W$yK986`Bsei3^6Hl_|W*!Cuvz{qf`q!_~-t
z2GSN?U<L=}Ht>qcK5*IEAj;EO2=uglsPV&AprOf|!e(Wp=TdJNE&Cl_F{$>ZGO?q>
zxc5ikLD_7|JL{*~${pMegD<yyzZ^Yo?r1xACEg!y44=0_TI<xW?VE>#dwFWA#BI52
ze`mdi7AHnEe{v@poI8EWX4ZDQ&-hPRYp!O0b8`Hx`$&nI^=Mh%^$9MO*p<s|;lkI0
z<@|0WRNI>oQuun^#OL+fFLmT^GZ--MtQG`>?C*`bX6adtM8?MFVbY)|bBvautJU;6
zR~v$p(M`6&{id*5vxxm|FLJt>^^rb^@yQ=W|7uWW)%$jRn2lLCwsh1ESYo0VZ$#6V
z>Od(#z6N|Wi{ZhGEI8)F)v+d7wVzXlTSuD)>u}TjyLVQSQ#4n~x1c_bsJh=syD>d(
zfr?9q&2WBSelefbp+3?u;wNXPYu4!BY35;Eo_pN2d_LLD59$z4@&SEu?SB@;s<eN4
zu{6Axf6I$?F|}~)=$KYncWGBlwX);JjIZ*GtlDR7P@}Z<)2P#Q8uymSboXw187sj#
zjWn#t`3yIFM=r;$MCueaW6bB-^$%qcK*z)Ak&%(X1#$_BaW|rgMc@XE(yNB`3c)Sk
zrB*fTKXbPv_n`(Mn*o#?GmpBa8|kNz{_^)hYCIHj@}NO$hqw2qn6)OdM1x%9y8NZU
zEcsi7=RZa<F5IcgpTN&1DKLGa<um=SDMjQjaebl!pTYLXOd-kenv61bhn`-O$jowe
zh8jVQO_Pc<1lvDe=Q7xjKvX!`LN$8hZLZTB$gjiX^#xaRaO11E7WVg1!P1fQ*)1$e
zkxvr8Oz&#_`Cghg#q^yoihc+L6+IW@*g~<&hIVA8sEFb96Bza5Ei4`2zm={>W`eq)
zxR!qwU;*z2tDiRrB(NS~so2Wo_m%2ndmhmyM^LH5m41?D#dvC-yNVn^wA3|%L+?;6
zEWZ|t|9nmK<kb)CG*SHo?udz*-+c{|`b10U3ruQh{X{Oum+EGQd!xuF^IJJn;Uw->
z_z2(`u6`J4*P;fBm0$8`{3{g01M$tBRWCD4>AowwVU<}naQ?yS*~HmAAb1Vnx>W^b
z>|N*<Npv}6R}i{x>r-VW8Qc<&bPiVl%(rYm4H&eONx(m|7koZ_M=9;yM0GyQl=`CN
zDs7RRSM3{Hihj8YaCWfe<~96Nbf8h(`-oeZQjyI?S{bTVD@}Emp%X;}zJ}D&M6P4Z
z_V9n=oM=`+ZeYz@l&&B#H{+-#1fV0$4QVNcp@XZj@7jhA6X`PP>EHo)ma=jDASmsg
zaPAuXq$rF2u?49}lxMV}&H0k6p-FF2DVZTqOqXv_5SyiSU7m-5@j?JhI-vR>pM_`=
z5su5j?m8GvsN=od`Wn&Z#((guBnIX^qjKV*%e8l6p~l8JkE^Cz4xz4EUs@Z~#9s2k
z^w8b{vMt@&dix1r2Gx$uj}j`5{8rWA-+Cl3k8sugiP}0t{Rn>CsomN7c})&dXId<e
zT1~B8E8s3=OZ0|Iio1b|d3Lt#SYB*Y_VR)H+LQGygf^>Sfr31dXRcEZ*8X09IHE4s
z?ZK*z-8v-Abu!N=RAC$7MoYCQa3rcy0#>j<BGo&OI_Pu0W7x9`!A*P+a(bmX{h8$(
zn(N?KbY#&ExjK=4r95``fb+D!?bP>3*r(nrRoure|1nYi8^gre{=PC+Ha^z<jlQyA
z8er<LS+Gi2Pd*7*(QAj#{d=iU)+1l&YXOfY9GzG&RA8<2<=abt(+C$qYIVu(q!=n7
zRGjmtHV3EmTbkeN>n%bs|BN786v(I4W_5MYscW@LYL8(qxkll@mTtMVs-E}c2EMqz
ztA0&+b%vz-f$gC9(~ze3gR!KbYHjivyVgsGYUR|+{gA~%JviW0S3IEhDQ4nTea8U%
z<=+bL5o_?%-d?i8Yi1e)<165}{3x)&di@sWv9wGjo`1`a59;GM@~l8qt|}#pk-mAS
z@s-5tg4oeA_@U!qlr6r0---W$dho|D^@y(ig@1HD3h1vAg$66D`ol+>3)^oM<Ba)1
z?PC68`~_%*{m-{V`M|6&;{1%BojPv`cj;H@tR1c^+x{Uv^IyHneZ8Smo&Hh)=`bv6
zQ0U*iB&U-wnSAly{Gv{;MIlu~M<N^WLjkw%=GxW5h=vg1)jST1E9@Pas%;kgbE{lB
zl|~s4o0{lXY(G(m`>Rtj$kdLO^&vM#Av!_2R#41&JqW$)A!MzG!^aK=a-zztTe`g~
z>?R@6`XXHfI;TqsC_d-rrP%6Vb_`}dlh-;>zU_^p{jP7+PUpyXhDDu)Oj+ipmfh{@
z38hXk`Re(ZJt^LDLOPsBN=P$VW0_9|XKvNkl_u<^CA$-K-fP@G7tk2Rozs(*okWK|
zyo&M8X5Z_9yioyKsw9Kc@SAZ;ZZO19A%co4_pfvTBDChwSL=K-T&9pr6Z$+Ix8vNp
zi8@*0qYbock07GubSe`~R;f6tQDCv<he^s$m~+@b)}J@e5t(VgW#Z!ep5IzGtAqjb
zE*%%6EM{c)GWr&Tc`z88%#TZsn9N8Sgd8?=&{sb7K?O&J*EWCZYn&@Divc(jfoPyG
z!ai43HjF`nuCqt;{tDqJ5v1dMxQ-4btvP6q&PynEsjBnZ;TsI8`_;ejhtm86Og*5A
zGw?+E8u|8~OwYmZYZ_+NByvX|kTXA{xOHoQJs*k;tgy?A7dWyUK5v3q93Fj=m->y)
zkkE@IMmNz=<?h_NQU*8^_IT!WzB<L)5V`BQ^y>I5Pe<x|wMM;ohCq2gKrt&}>TCN$
zog*@G(!GGmI#N1FK^pZKxI#|RppV)`wRSY&cWFWXWyYPCA5v2Rnp??+ArHKl4v8Rc
zsQ4)fZ=piCaQf5~Ap+#Hgw+1zv(l(gg8CR3(UPg>`ytm@3hVMx$7zWftb}si+atZF
z(iNP!1q<#|`ODLP%j=fNA>@-CSFl(p0Zl*pz(<bIt%m=#=3q&ULmR>7rRtY1Nmkd5
zT!&Kw#)77AZS|e#%v7lM6@(v0>L@L}Dzb=wEjy+`+^2`xYr6q8S89};ucjU?XFO+4
zWB?s*7X}{EZFzZdxyIKL;wSKj7fhBB&WIp~Xuun-c=QX*Bk)7K?UJ`&bXA|+JpmPF
z3yAvOpHZ~3#eF490jxQ6poob}{3U0`)B)Ds0rM{MVhuv~t_fAQe%W8X64p2rIX#SZ
zH>pedmXV{dG8A~QZ;bxTusZ4h-e2KmkTr=kE?XLQ1OUc7Z|MH@INq{ebB?9gwVk%G
z>P?49t>gmy`W&mhRvB2Co$N?y>A`ArUY3&4$h-^x!C~PRmGm59Gyfh`Zydh!w!qvP
zQs+}yiH|3`STS{v(=sO&^i)Z`I4=<Qwu9AJpecfCstTWm2rm!na~G&@bzeS59I;)0
zH*KAA`VIyUDCq2uWs0H-TmPNiI43w%lx3??V9sioUbt@S9;2CkQE3imUw7jw(h@_X
z^R&OQp_pP(Ek37xW}8e1xx`xU0Ek2(hs<*F75@2!ET;tW%bSVb%DTP{w=Uh<<}Um{
zW<dt~n(FWErj&H0V^LtvU;x7RHn`>f-hn#WT9XcBGC(lkW)ffqc63QL@^k>MzK)-}
zRGn?^dIHmRJ$LzK)}K9tMy^pW2)Y$5>sFLgt;8YM^op)m5W;wC^@lJf6^!WTsL(d0
z21Rol6JFEv_pi~-xNn1eNN;K7-|v9A#CuB9@S^O>qfkJGanC8$y-3*2v!@C#PIJrR
zdQDz%T<=AOs)Z9QP6LO|_i7l*fA(ch`Rq8eBZJ)9$vU4Z&-lJBP7>o$@tE^^Wrs=p
z&pe33Vt&a^JuN(3c5YtSK4<0UpAwtwnT?g37?p~8fCpJ0N!;so33>^b-hoW>2ZMzq
z3qzi>QIT9`y~Z2DFg1YXp6KUAjt(JmF7^4_#yY$q9w%Osg#ERf7btW1=f+eOzpiZl
zUiAa5D_@Y!lqhZn@FtRq+H4pb1zo(7$-fjNMBwG+IS~Z{+mx?u+H`Y_kIKw+F_e%*
zt&pgbVHxp;hn}N<q1cJMX4PIx+U~B=yLbH23HhP6LFXtqH5u)ZBX*5kiu3g{D3O8n
zq_KDwXnt`SG#P(UtFbjs-SA>Jxt}h(I#3USU;o%nQePEcfs`dA#BqY1EGMOQsZ9kU
zQIJ`Lv!=pS4!$Aib_C<vFaOCwfkn5iJKJkHBn4?shuv}=SM;FJey>S*t97l%TMf8M
zx8%5cibCsI0w}AqJF}M`|1ND~f_=H{1wcDd-?zL?yCErIC}6%ufi{K5+|&Ja)2C72
z<?n7Okiy%%xIzFl>oTdNRW7W1HyIFe1y8l5gJx&T7@v!E>EA;-r&h#uIRD9jwcomF
z@S_oUvXw(kLv%uN4V84v`<sK5*$MeUw>!o<zRl6+onExE*@?Jlh0C8yZTTF%&kiNa
z13iBDHYxoU-MMqySJsfN0_(6SIk~=m29$6e+YY0%D+f63cpv9*wJ2InYzkSQ|A1Vs
z>R~GxErC@R!ZjV&Q7wYtGL~SSJESo8P@S>BNhyuJq}27rN~o`Z%PZgzu48(iQp@6}
z^#|3XJh{+M$HalcYdopr$)o}Pzf5cEZ+&Oq-sY&C$V^86_X9#;)KJjvHj0qXDe}D~
zlr3d^o6%(P^;%o-v0TL2x$FVOuY<AlhTYN;0InPQd|tGF9IJL6^gkfp@2`7LG?FeK
z+@kHfU#YULo&StzwvU{%GgGo-QK!x|(RbASH^}U_)M8ReYgX9K6!l%p2Jr5?9!uEK
zFW20*dc%4?D|<suB?+%J1{gK<^HR9-K^WWSz8Q$;2=g70VHWhznn>}GF%0z<zg6wV
z(kCboPNH0s9!}59p&Ie-2CPWcYofm_-5Xo03$||i$w0SCBh_v+yT|uZUH2i$@7;=)
zj)G96oh8S<vfX%%VkBppn7+_dRqg8Q4@&|)M$ZyzZ4i)wR=LVqROH}EG3~($MBZOp
zTg2jG3q<kMwN8cn>bO*b<y3#~kW2}H+EKOOz7BnXd|f%Ni#-8TDIMqY4=MW9;`QWw
z+4idD%ZCyVqK$i0>J>B@exT7|M|JDf;8Z=K<L)yZS0;^u&rSIa`06s8-D5V~QjtY=
zy1QS7AzS;+acNecgD~M!=a9e(E9GwnsQZM3r^X!*J++xu`-@-FtNk(cbPvoJU2iB+
zNzN+fZSAo#fSmoKGRtq*4iJC~EF5?0JoN>su399*U~5uadt{_~>0n0(X|nWsNxH4|
zcT7X7l8$*czk%}WwXK?$SN6S`*L6macJiq&Vl%-@)-d4o)ViyCg6MeFhGjz2op`&i
zV6@Ub6<Bl{!zo4wq^%fm1@=}-nl^VQia<Qy@d8Ka#SEz(-`j0}=TH(}QGk20ZKtv_
zJLMiY^oz@GlB8|k`A`#S8Zl63ZPry^TW_%3CuXHRl2GQENVPH|;NZOR5K8=55f=x_
zO5`K=kAG#gx4J#?3f*$s>i54Dv|xf<y)b%IQ}Khn5Cj4TTjSESDk=x<<TjstYM-yH
zP(yTl9I8hb1!%MWnrGMI^+!ET0dDhgQ25=}NcF&aNNVe7l|}b4tU-rLXtX?Qa=G{Q
z;kxa_VY`h?50eKa^j*&<g*1c1%F3EdTYAraqd?}y<?H;NRiZ)8rC8zJV%SQ}0^@$e
zr<m8(MP<xW)$$_e5vm;oJcmOYVXWi!&ou$Cm>PG1!_c21jGnFT9)HA6A;%-kYyGZ7
z7?^Hkr?VVJrvItzB<XeE@@KmP@wkV;X@Vr6^%K=vDXv(E059l~`j`6DYa9GrgjFrJ
zUwhsL-JCKoO%9w{3d+^b|NLC0pp)HdeOe6b{=MHoc$Pz$$3k2;ylk!pmJ%xpR4j81
zD$lz6=>M?XN(><17Ioq}0b?F1({`o&wZF+Vfu;8t_WY$T5Bk8>dUZKZ_4IA`xOmT)
z(=?Rk3+4pwM+UIRCgdVw3z(L*CKgg^=|}ZPMr;L4q_+%U>VcV<y4>pZFl2};e9*Xe
ziDzHqLwzc^<@R7GbR-}}VC_i6<=my)Tyw74mBB4JgTH)GeJ~>?DIvedH{Ni4t*f)g
zynl8F(E%FERs`JE0#`(qhex8Z4dx(D93P}>5X_fD7Dz^QzNc!f;PR%dn&XIajy1lI
z5bOh}wGcYqbAY+6U}afY7;Yy#@rotW|Ncsm*?@)Ip}<`5stlBsRKCLhgL%i_`4PN`
z4|evArmFp2^)elf%hE`(5>rrgZd2&Vj+60YPr`?rFNazzgeD*Vsm?b@hunTpPu>5f
zD?3}B6!UYLejqjg)=`Q$I0!kL)!>Fb-UMrWg0|_;CV)*gR_#$qv$}04WyaK8?e}c;
zE@Xo{5Ot25c~3^X%l5aGxlWH|FWhI?FC6zP*-82Ah;gJ_R8=?u?c{oCU441cFK)(9
zvN}<Ye2O~E2T=lqsgEv#*Fh_uch1gi&Q}ktciS{!HV1b%_><{c3`N!+HvAqs?kB2n
zf)jb42V5Hoh7Y&b;v5mV&QZ-|b$>!qzggobt{<;JQ8m(Xj^@sYw~NVy4mDYhQNTT9
zK1tpK+S#&&A7Y>|i(`t9v3%nT6#!Xc!Am=P<<^ks`)U|8AyLb5<|XfA4l3cD20@2H
z6xBh#wmvYj)MQJC5(iw?J5JQe@(~R)Uk=Ehd;J<IJvCW#Q^(mxb8(r-2r7;-DCZMD
z&fW-mXalX^AIw*&YBY&gj>`^1ZgsS+@+?xrC#rPucx?L`A0$#O+};3N?P-Ldx^sLA
zt!9b2uN7bNc;?Tu&sDqGAy41CbFTKcU59Wf+NVL+KfvyYUwFqvN6;_vLBoW@@Pg)t
zWSL)ErT$eEL&pZZ@+&t|zI>b-{aWkhXv>4ms%rJ{A>c_sI9um^on#ppDdPl7Gf-V(
zldf2?+Iu}~5M4q)vK;l0){+WuCYWhPvmC$Oyfu(({GmKm=65@(z#iOV95($0fYhRV
zw}8R`TbQQWM82erqIe|t=9v4)9aQ8Ke6no(Ers8yUPmi~4%t>Ci_H8{vH6Xma@RdJ
zr^mSpgACPhFP^YxylJlrvN{T2y_%MXUo8%nCVdiWq4P&whF3YC8=L57Ky2Lb>xn9#
zOs#H3Uza|Ao?}?#)$>0Xc-r$e$}gBi#78I61@NsMCJUb*Cw}?5Rc-=xsdCXt8L$LK
z1JTUZs|>fcl9iE;nXhZHKg+ufpDboH{6P%3ydh7g=FTgj>ZGzlRb7r$2y{?>Rq5b|
zkk>2|&^a>2xIbT_qh@ly$?`nM=*j3_IUN%e01uQAmM9v3jMeT?%Vc1m+M+(9kYjTT
zkazkX$9sB~q4o8W_Frk;<p`SJ@+TLGxV=v9Abn*2*i<0d$>7_W+N56o)}LhEbf3{3
zR;5pXz_tPbKthFeSUEXAR2UTus;iH4WnNW+y3AZ*&2r5tMuB#4Yae+N!(+ahF>-;m
zQ|sYzHm4RrDCMm-6{$!fLyl2p*(V<Nl<l-xxMgzf`J-#`1wP(z#jOUXq4iA5w_$Z~
z{|B=6VEO?X-xKxyhbVOtUqObw_xdPLVd_?S`^-#aW(K%Q7P3oWz~6<!v7$Cytk*#p
zUG5V9urWnR!>t@g8dQFQI@ZB-<y%5NEuyaW>!;>A#E0vHp=R|qm!~4G$EHc(w%8|6
zpX9DaL{w+1%RX2h)wWwhJA0;pjMM5My(-f9>|wIJVpP9XD@7Z~i*SAW#+hbgA>@wU
zczwxNom2=-7OR%}UcJQTKF@1#+>ud`SNBj-G6)IOt#;E0;Ya?{zl!nW4vP4$j2MW-
z(b)><5Ajpi7)Yg?q4U#?%XTMT*Xc%SQeH=uDHeloI_eBI>lg3AXO==f>(=X(o!fH5
zd5H@CV0GEaAosR(%E(NOJ0AoL3=FWKhxdNRmvO>H998V}m#?aiuSROtwrLO*@8+hc
zT;0=|UCTt@J#}|SKNxoS^_6jt&D1`$>PMghxJ)ZV!?L%!#)ioAN~!~!Q+}+hwS2Ft
z=%RmIlB|PeGiga!eeU+@kb`Pvzh=>HD{X32QjE`<g9JOhTekU=XiX)WI6NX!s9|sF
zsu6`>aB>hsKsZ$(nS9~q(4R0|=JJ__YDawO;VCK37X!UtgGqW+qDMnqJYZUL0dfor
zS}IsFNS!Tyl=bD+xND)s&CNu4-Qzb+*6s7Vx=<-6`>)R5e(QbgShBJs&<ySws~*LV
zu@GkL#hH!S@#(79IoUgl4)vwbg<LDVp0LfdQtNA!riQ{RTaw4cj^4G791S(McNY2;
zd)qT+kl|4!CC@~zUj8yMc7U75%24qpCMxUXrAe_Ynag0MULrp`aNC(rW`u7Km9;$}
zTbf#yF{W@hKNUD*(L){Z;N9xtYG-DoWa1C}zWhwsu;QJgP|aS`I_*^y<eG9v?62E@
z{#-tI<*)fyf|j)c3m$=Q<H;J&i)b0oY7>F%vfZ45M8-h%L{WUTR=#Jct>iNiB%c(G
zM_W>N#OE#R684_l<>yx9;NUh_$`N%7OfnhkdX`;2<lt(m=Jqh1P}_q~V6gh93c`<j
zUD*BIU<7(!VLO)oiETCX^uZ5Cyk<1v)kLga=Y=aF1Cof=nZsb6r1cp@t}|?-@uc8g
zqrbRYQ{zrUz+2-z?614Mv?xwrCc~b1c<C0axk%gSOG)0pQl(&y_gS7U!btl^bJsA2
zb#(sjcY@n8Gc8ZB_+`vIUJnbl$V_z)`+CY7m!KeLa03{f9j75E`Gb0%TIl7Q=V#qE
zk)C2j-Gcqaq%VC?Z#g;ch~{bF#YQy6uJxXK)BCqpdW;J41%;8LT3U;K9;Be_un9cp
zqk8D+=(FMj?y%l<eN$HdsIHv5qDWcK_l7{m-0`LOGp}y_>Ba>qjrxPT88z<kBVJ6_
z0Jg*T0cvU*#@26sKpe@V9O+r)$$Cll)S#fgeV8Ee2Wxs9UtB`BSBIw4-<;7GslFff
zMNl)NL#h@Y>mdkwde)Zy(N6^Y?r)+~c?`IR!1rA2g~uG4cN>~sl&A1F^nSm0>3d~&
z*LC$cEr<1<|KI=o;rtC98UFBp`)D4YfMYwXf)Bd53;ir&(Mu7E+PZs`5f?NDpJ(wy
z7C*+*JuS9r=Pj2fdY>G8U&MCoCx>huqEupQ&qU(Z_hr+23s`3FmvZk;3cz&w(Zg&N
z*3FX}r8z006cO8!9#43j_lN}_{&+<2OPZDb=V13|7Pt#-1~gHXh=#O3$-o!m6*j6)
z_n9V_hoJaa^)~(Ah5m<S<+D`^r1!jbl^2gdx7)GX9=E}jYhL$dvYp-AlHzRf^!WzW
zrG5UC@e-#xL9ZKro*ZO(T$?Aw85m!oS)3Tyz$f**L6p0kbC|(1f$Hw#^sP>lsgsar
z9i~jgY#;GBa$ZoT9b8XPjv~D@1dse+<0q023TOL;4v`7)-`OK@WM{}X*y9I>6@B%@
z2D`d2$dddK2BSf^t$Z+4cOu|Kx3)AV#QEQ>Iv<dmTz~q?QJW9%sc++^^U@ku_X#o2
zvy(|ou&?aHHGJqr!RDL`81DKiI7rjdOI4N3YV4B<2TiH*fRfw;9A2WQV)a!HgEmkS
zl(AtU{`Gu@Haa?P&p!MOMM}cY5V3b2n$+%M&SB>pwjQ-MwBO8oW}DJPcV%Us9bDf0
z94IN+x`O>*`+5)d*L?#qOA2dbLVT)EQ&+TF;Hihxo73B<LinmJ&p=C5RB86KseSXO
z3(ONUaD(5@d;UA@-%b8z^SlEI)^ELi&$FlFw2A9wGPmR6G&|CZM1QxN8%e4o9Zgrt
z`&wtm)3n=uH_VxxJ|KXW3$cB*zq`k8!rIcbHyd7`c2{@<9Yy{A)Y#rKw=^t5u32E>
zy|$3~GmI&_$>z@-7Ov$ZWpBxFr;oO!hV*Ndw@hxx+l>*99Yq=%yvvJ#9<zQXR<&8L
z%&%^b(B?xq9>03ueB=0dfKP!Y>+dKTPw(Y7$;BZWS8WmB-(D}=aqDgK+hs-0s@#$s
zs@!tVn&vjfzNr)t9v%t;Q|nqT-Pvh8Dw&$DmUrwc1%a{Q_48X#COm7^8psi-e?b00
z2X5Xo&}#UvS^`L|ImZw;DmJ@RzDln<jc4+8p308Yn1*dUFOh{aN3tE&k1}p%-t%QD
zrcfL{QrDP@IA-+7;?%N)6IQA46&6ybYd7NMahfJis4MC*#v60XA91%}r5;M;Ph_Z7
zh0mks*KL#S{;{v%?+-yCQrlSKcq4>PpKF&`r)?mcuKeBxA~kX}5UmGuk`ysgUn{)v
zJ)4jYw`n4zv@V-%Do00(xmSst4@ok`wU1;BxVB%n-kB6~!!mjKa;VD-6OWQI*{+;5
zDLgFT4p>4Q!-|0Y*+>{^&*ZR2v%_exc$Tqsy>RGe_SxXMq{zihaL#0>-w>)VOSQH+
z7vj!~5!pIg|3vilqN{@6WUjx@fx0nu+|@584l!U8DqfXDuj}SmnNL#`U;MbqYysJP
z<vIV%Dd1-Ct6!{f6dnuH#W46h&^#u=$G=)~dY;EfDdmHURE`35=43k<kN39@FqlZ$
za>cSZO>yr}Jy?-^bYvdgu^@PIfSbIt2WePtc321?E!^IS*0$XEI#|R7)a{j#75&yl
zv5Lid67`g-ih;uS0uGT!#C~G8WSCuC+)SHt(=nJvP{{7ggZi$qj#yPajc(!zevVqZ
zt_K*do<*vd=9qWC=&CjP$JlK4;{4mgeee>a^0<>8>|J@tiXps)#<qb>CW2M%s=1ww
zG-`(C<D$1OhBo+35=X|0=Ne8bheu0Rch#%m$BCpoCd$_ITWNQ@v$Hw&x4OLFp5xF2
zN}Qi;>fT0fBb*h?X_`NHA3a#4p-EEgNZZ>LIhir$&5FT-@9f+O5UgGLdyZ!l`0E@^
zQQe`sz3gr_@eGs7$S=^xfQaX*2V2@ptEzUu9ug|rFqU9rX}eB;8F~hUOVWlDof}ij
zj1}{V&+8$ZX9o*KsZ&{b1Y^uOp=($gxTwTkP7pVEk$>R{eyPjRq|VY_-y@M*+gde4
z!|@5*9a;BaMLI2Y!?)FtdKq}q^!tW25sE!YEx%f~HE<82@}b*B+=z?nXkY;^Q^@43
z)sGHPiRo(Xv+sN`hTTqa7Aowz-D;KxI4=dgp8zxV^l)Wz+7V-NYvc}44{kD%wwT6O
za5gZ!`f!q4Z+6mzU5kVV&4FWT@J1=o#h|oYNiVTkTAE5P<nDDbF?m!-64agoUBu+3
zX)fe{3w=Be4W&s*|Cfup=l>S*QGV!l`h6->bEg$^PGRn$(A}fnnp91Hff9;YEM0|^
z#apCY1~+bAc;7#nAvX6urKY9K_=++kILyx8KXPelXvck-7ajf0Sb-|*i<Z`HQEVQb
zp}I2~(hKi3AzBe0P28cPOf)0HnVCMmDhT?#2N3YS!{$2#oJ&JO!ZbG4kzVNi?@%U;
zOx*iDVPtI;JtC?WSKyS|8*j`-ZPw-G*FFw)YthV&C~SD1wMzSrA%nEvV-9_+N{DVe
zn13=V!1s0&>i0C7GV*;CZrswP<u)Fgl~v0;kx(^jY%dWnE+#JaI0!G2E?)*l)B%|$
zedP<l%8HyUK$k8ISQ#Yq7B%ylIq}`0rf#Le6<$F{Rn%Z|IG-7G=a{{4n{!L*>{^~0
zCv|7UrjnAPXIb<x#-e|npgv|tx_)hS>&k};gHaN}dJ5taG%4K*T+O4}>XEYx)1tNA
z#G+}XlovumypdlTx4aH_P_YU1(3nE17+hoTRS>t<VDV}V_hMjR=R?~T$(-(zg8{-(
zP6q=@^MiLU>C_pjYpPaji^(lhGHQ!=*5RWroAqxxq~*I7a|G@VX!B=qm(w(#BlL-5
zlOCC!t(2lT>kiqe{_UfFHN3o5z568h`)%bo$I-F~{M<pEQD|(&&ZKLh(xU_m>up*2
z5aHJ5^p6OHu!%)0U##t9ds{0Vw?U0IWl=^dt714w_f;UyFQxE#VC(oII(&^Tit$iY
z`|c_w&cnpShLqi+G#NGpv4t?3qM;9&+%1$Ww_f~0*Zkb8o{%zhm=t!3{8*W=caztJ
ze*Lqf?=ybCBM9bZbX*PB>uIYwn3@PRWlvfC>&L8ELQD|=qQJv+>s84}d-4Ibc5j3#
zJc%(TCK8P-jQs&DYOzo-+B>lHMLkoHl1Ugkt^WzY<3urOc$uB64Q27q$*vI?Wneh^
zvdaNc>mrGg$)ccSeWZHp)+yJK<&XG$s}BeQNeDM`*kd+x==Rpk@9q4`X8nxEbq3xp
zt4u*G0Xv)fd0fJV?)x55q5b_Cgs~D5^G}{qXwtpt<Cr;7bfR1DKGQX%yg+F?2}gso
ziz$9G{ukEUezdc7rHS}^98+35A*O(N!<Kk3{I;c$#Y=!p>9t2(X*|SKqugDNprX*t
z$(ZqP6u-WiNOyIKrK|OJc7HB6NU7u-=CL*y%--MVimt0|V`VX#=rtLal73$C{o|s7
z!foWfg=yTI%m=ygSV~IY^47-s>KvE4TZV}8UWr$s#MhSEv=hQI?CT|po<98`#COZj
zRkNs+hvnVzxp-$7&Q59ZOxCohw=J}Zx~md=8xqhhiBy&DYHON;kC3JYH8&1Cle{Ts
zyVBAa>pLGE^Ef45UdQo1*b_}`=bI$!j>Rt!Y)enoeZYwFcC^LKpIZOuAI%p7dEY!T
zTA>Paf8vDA%OWz`tXO%JiZf>xJCPUbGnuU8VJ*Cr_@;wQ(@{?^uKaok@}HOM;<M|0
zg_g1{qxtvp^*qN#`TxF+mCsFH%FSl)Vpy*3vpi}E=XQ&Y#aSLs?{&@%0V3!u5oQ@#
zH;*MRWC+f+NEV2H<qy!a=fWr+YA|?yw5WH0n>l_o+m~2X-7sTS*db0;>z-d&@S)0G
zV61Sq3#UFox{0e^>AAVh7K8U~d{%{~;HMteW4uMDyl^7E*s6%<cU7$5M{bVVgc#gm
zD1zy$9(`l;A&BB<qM?d^@Vmm^w#(H$4W*t3`2TTon2J;$0YRU1;EUEOxRMPNQr+#@
zX_>briEX^~^a*hX?h}kP!xd^YN%A#DA@jJ{!DFV3%;+ta4^H$#A%%S^I9V!s%$<ar
zYh5NAV{cs0pU|DS^5hEK+(*8%FmPU@tpMpSL&@-nq?p$L0e>DED%;(jZraJPTIl`m
zbOU(BRz`eN@LDO2>?ionFN7n&o|m+@f9GKP+0Y(&nX-92GU@3%FGKCQfDg4=Ym2SF
zvAbhgFL4Lo>0uMC6EkgiXe2JcdT)_0a)N3Yll7A#WHy;l%06NFXDEFRuP*BqCgwR#
zHCOU(olP%b(J@Loh(r|32YH8dG#<ZSyi{oGex6fE0vuFrOB*XfG+kv7<}jwE_2p~)
zYYjXSRShe84CaXm;(?-XP)Vf8Ox*pj;OBfC*uk&-6EkilpPw)H3uBnPY>?>LFaSo0
zf*<+8*pvFH#A#3QX-dDF!+((XBie{*d@?KsWP+<?_|DE~eN}9<p3S)5XVT;C;&>Ok
ziB;k5B70$VV(|bA^NG{3m3g(4QJ*iv=l(=HAS_=03<WPO+>#`mUijIr{IZybODs#n
znU9e;@0EX0)SOQu<kO0em2$NyIYc5Tm!+aTw`cM=?!*;sPf`d!UL4_*+nTB{xZu~)
zRo2px@i2c{SDx_dWT)}T(<Gc37bk1@#wZ{Ewl9Jubm!>xO}*|u)TD!Q2cR8UDhxA*
zFdFCaq3_$K_jHRSRr6G=?3Rp;x(j}i#`hNs(cEPR8-#XpV&mp)9PBQV@ZLJSQ7Gi{
z;bUiW{VbUQ(n?>lK7shCYIFo3J}Ejniu!NglnV?Bw#F8b<Ich?EFW4Lae=D+<)B#G
z5}Zra($S>7wlvAY@~W}<aYRHD&}&$@DiegD;zwYBg%2+{EZ&UOS8BC@5EdgTwa6mV
za!xL@!;eNNE<3D4jaXREmch20>vMjtu#cTguu7hS8ZqYlt*X_f*Z<-#Zi{&w2j50)
z|1g_h;8@5|<i+>A<A)*XC2<Vv|J8k$(Zr-gw&Pw`tEq(xNzls%J`$qtQgHuJ$#MJ$
z%<h>RE6)0}k7^V~08lw)Mdlol7*E8k<=()id>MG=(c4f<2-t<z7#UnFtZN3OEmQ4b
zGxhAr+!lx~pDEhNmdJE0<Ntoj$Z?@5!;ReIv461~=1xzKrmDM{OiH*p7m%9#09bI-
z*l4G`RU)&ePuTZLSx+FhJ&n)Mo>?-nUf42j=c6&V?-%Acl-{+TzEeg+68g8&s$!@!
z&b=*{Ov7vQ=<nd;>veUTfg#3~mc&<^@w~51EVbT$Q9kv1pl)MS$16S{{b%xxVMmVG
z9?bp4oRArm*Zvc%7%XBFlfW}rFESn*%p#ipNh84%r@vVuIH0yXl<PcIeSp4s!?7|I
z0lqR-90yAF%g|s__#?8$(kX96Es3Z1?*w?PR^hTEP~Z|IdMQ75^7taI=8lp0;eD6G
zS1kZ{U$o1P%ZDu{VF<Fhd>r};1~KBa$~k7=$j$Jxrq(*i=%<*D2xQ%;RGuu{aF|%k
zwA`=1^#J<_YN7sHlVRtrOQELxvc5$R+u9z2j*h-bi^c@FN1w!+IdiKUmKu@d@^Xtx
z-namU&4kyUfdeFokz_6JUvTDfV6p_3Y?#|-;G)Ltty40L)O$ky4JNm@y?9GZenA>~
zu)DVw>t3s9aNqcR=*^eLbAcS9wd>${51n2k{7DhtgjDzSS(p>MfUw@S5{li<!K`h%
zJ9*fBhGHWY{yaVHuUmQ1Rr&&Z)^Q3tYngQ;%pceIJW<a3R5d2Ep59A!x@MT!hilID
zkJ2@2x>GEE`9D(5S95W5ob9+#Ya{<RGhya@_%@v|uVWxKkNvFB95k-CBVf)CS>x3U
zd)^eWpK1#%;HipCmVIT;4YtU8w1QT3v`myHFMjc?(opfRxp8kX0*~aQY-UTotEpyf
zq=@_OEBIWc0Pk;(v<Ep9h!^G;+z)A~jF_T!>=E{;Cr@$@kcaNon0WCq4Z4yFZ|Q`C
zg>8oBtg@tshup$U*$KNy)}`B^?wP$W*yo%sqVSfYj#+1dJhi;<Y&<R1u_gCP1941$
z<|Dm555?h2`7CF~bN4%1f3^PhyI%{l3?uBXWkj6sW7|+5b``?kj2GahK8OHJ>F!6m
zP#m0R=!_a52FU_O)~1bsJT0FJH7cxkc;BL&H|XuVcoU<4wfNIHS|9TAm^n)YXVMDx
zrZbc8ujjiq)Wp3Ph}6;R{Do(s?RkP_vdBEY6lXZyHuI8Lu*J(r9FYPpALpwNl4P&2
zu^{OxLztqe)6>4xe=9OEx9!Ze*Wdcd^dD#Fm2NaPfh~^5z`$rSe*T)u%#5Qri@e1`
zE|0pZ!02x%X7^Q4c~U%A4EIAfl7|nLs<wPz!8N8@w9)pxCL9>@k&VrbUxid#iqqVA
zk`vz`%PP$J8+)17`$@U}N{$OJ4SS%7(_a_4>l-55^QP}xIN0^hdzM`@lC&#&q?S}s
zpf>w3DS;1PM3<O-cX#EaeT~0m`>ji+;Tx}LYp%yo{0I;d3ezFA><T+=n_nt7agGjV
z2MY6W_I4;(>isA@seQZj2>;SLqj4q|=EnAGxG&9{5bv}daUV5Csb$gC2MnB>99hTM
z#C!T!O($xBpv7(lxiw9%-M1aTJf-~bA8Ya^D0#I-jOxkt%#ZCEN%>7J)#=!b*YxKY
z8(Bl`G3$DDI90!=9i2KlJU2jWXFwv5Jv3MZ{5^`(o(W%ux@SuuwB%h+5+LrHZb{O^
z)wb7{BLF_M35$}+La~|QVmLoa{;k7W#WI!hh+2>Xn*$F|*xNg{d-#2vxaZDUDxi_I
zxnOi64@lFvs&wIiEUkY?-U@h=FVYZ|;L$`l$$e~xZIZ2kMTO6uWNe|p*|Khkh`+4I
z$7jIr*$|M+l4&ZyYY}mwruz?*GWy5VGwC6(FNr$!tS&X)SL(!Bt4JFM<S1!W%dD+^
zbNdA*U|KfGddhqmVq~cN<QU8@DkJs}-xg>J+~3xy_<Ip@+XyA1V`6SnqN#~fg;ywV
z_2j6ezX7%fa77;-g_af^EE+zynSAGBTXfPBoU$GL#xXfDdE>X<*hW;#vRuhTcYEu{
z`sN3;x%u7scBG7e^H}hsd0gu`OXsh@rfH$Z5b+&|&Z3KL5K${DL$hd0mF3LlLtKDm
z6RgJEAZL1^QiC}Ry^c1s-Nj_!U%p7)vz{?LLuIOrdU;Xac$#$2$aQ`Y79M4&{BwZ0
z#$oUy*_7&8{;0Kse4-q-JI`Sg&35~{3_9yPZEv0|oi5@a#qV?GOp`3&#=m9G<ee=M
z{s-=on2}f^Un@Ons}@GrW~@s$S079YxDnmoh8nPYjMFj%UxFYf9IAR2qtm7M@{A+P
zH3Sw8%(`qHao;1b`g4Z9FSP?ulXhEKCmiHiRZ*0>_Kn-!5mC9T@gFsF)lijJbfD|q
za$@{_m1-HnpWxs7M5X}362ZTM<F(Jrd}z%B?-oAeyWIdAXhS6@X5A5ia8@4u07UnY
zK!hW!(9R%F?mJEYTPyu2-?oV{=Jo#M@`s1fZ42Jto#gu!zrQ;)L^T5NUKkpt#clke
z?{3}Y76>0_-zBIQVQau&tILXiuS6x)e<r!}W@TG!7+oT*V+W9=7N0fUT@`u)`&Z7C
zd|5-od{f?!eh3KpuO&m{wG)7_o3k^K2b`0k!!ST^>xj5CjRLmFwpLa|+SO>-IT5UC
zC~RD`wEdtb*Jj_K_T9iNE>5NjxRnS{fcZn*Gi=34bxwC=O(+3pQ0h)Tk9^<KhRFTx
zNn)YWw_FOlMH{JNsSzWWs@WoBVQ1eR8?g5s`TdQR7L^IaV^2$iSiI|B5yDDm`YW6C
z+=Gq8#+jp><s3smc7pL`mJ0n6bo`XV!~`%;{WgOz1*$<``5RaZPS@dVi`)9kEDa8e
zeJS$?WVxkX06%<*O>C)rS_g~}oI@kAWvs#skTGo|Bj;r%7VpQVupbU-*lNFcB&(`=
z9jHUQvHDa`pRU+CbMT}p6PFhYQeb&8%5rNW|Feacd>}96@%JUa(~~OkpGluL)@pV3
z>@&`Nme*$+V?MlM(W~!m<gooZ{>|8)Cg%;vYK#}1+M%eTq`snMRgavS>Tb#N3DtU|
z4jnKIuvR}N=LH-1$l3M##~zkLfb}k28~22>O*xEw7SGG9&j~UJcEm8$wE;A*IAA7b
zM(fL(sZ32HqhkWXg4tNz9uC-$P{xNI!EWE$-s{#!HgE73@C{vr%Vh>6GXrzQO3o`Y
zy=8N}JfN*Wn-i&}^-9I)v4v{?3B~-)pB-1T!TOkVksXeuWz*+#2~gOe!t2+A2_%Na
zSASz{9W4JElFlgR=AO-wohiF?rACknc3)QB{<gO(&C2qei892@C?Zs^KIYPf$BB|B
zJY6^?2|9nApsVdf#gz0xZh|pPx4Oppw5F+S_w5n@ETq1M4JHDI#WSPw-%6ZYK~h1+
zjU};2a__k4qD;3s0rDxne_X>gH&78`wWePGV3fqJu&U0Oo~69JagpG~F`^6S!V9}T
zz<`F6GNf!47Yu~&uejFnmn~KYYcgO5UxSp(BxV2cOn`;!Awr`-yUtw%W6#UI0G(8p
zXJ6wqC02W8X-(?w!+WrSKCZ0(&dC4SFRQ@Q7=44=FnkdL$v#5rtXF1B-+@gqx1W*)
zOUIk=MUJ><G3jKJ_<Ul2Doy@B0Q*HFjm@uH%E=gs1Nn#Oo;8XaH08TX#xNRzm)*Tt
z>K!x*`VAMAc9_<0K333{^K(Bb=y*ZS+KKa^t+j))zVbG)P}G$C%xsq>1`W?Fku)`g
zOM~2Bn(JRW_H1F1uo<benqZTEFia0WAeZc~yTeFP^mbY0>rhXe?IhL@z~Llwxv#O0
zP7%CZF~VhMk6?WAjBxyf3$|m<nT3z<p{?6HI2ivi$Nc;Dyf*M^q9D&lp3D!}tg<-#
z2pOu*hd9$?KQAz@1-EHmt~^B;YWQFvWYf6CU$A1K!U&BsF;^ZpW-##+{z_=IId`E?
zN@cM^JX$li`iWUv%}$xd#UA{@k=grXQ~X(tc=yiuS7rY2sVO+&^C?!Nwnex%(ui`c
zX!6a3`i98&uZdr`hZ-GG+*UU=%tv`X6@(QelkPx8^I7D+v*ERvys=+9q2k8THNvyX
z%I%g!L?1ZImjA3SK2pZTng{^ZvhoU>-za-@cker(Dtgj#RNTy@<U$$&Q`mNbiO!cT
z4kxe%^J|(cRT+KE0nRqiV&vdhRETfwm}@b%|2FlA??3$8dum+NuUn2>&mRNj`P}_E
zY&EAQu>i`hZ_Ld88av4{{+p_@x{8bCr-||38Brm4v{5Ses@%ei#F7(@VGpuZ^626^
zin-Xb18VIuPFFQt!@jAiAbnFnQe84>a*ngVtcSVpQU>NB$FzNQ6k1mp7G}9-c^y2&
z&CRZ?Z@ITD4JLwftDUo|bDPc$^gZh}HGNTkLCvh`k29a*Q)*gkED*_LfyDNX))_Zy
zA_KcZ^KfHvu^sG_g5Kx+h-v1yj$nD&Yks#Pf3#sU_USpd_g{Nnt|lNF0tQx!nI&C|
zrDbFs2SBaskZ&x5P@-b+uTAg{cY<G)v1;`XW|-vUJ+6AKK>ip9q{ioeG(k#0jM`Y3
z8YkN$7e9Z-*&Gi}7fU8SpbU7Y)~?7}`$JQ)XK{(N=J;%xEwZ$fo`tdHw`d?To^N8T
zY&PaTXx+@e<Z>yfXC?CLdAC?slM>Z}%*+g;`H7tnNfdI)6NRRcf3tOhO;Y(p#@PI6
z)$7;1TG}%Uvz=eMlb_)ZNJG)4|5mQ!AiVnqGj#>Ig;a>jXO9>=?xJ056GN)Z)!)Dj
z3{>Uh9agGNFko?smB}`PY!h+{o^S3?R)*qW=yi5kjpN@W!0y_45*=-Mfhx>5C@l4C
z(D0b->imttNJ)vcFv^2!XQ0bXP@F?VgjDf;X*y`kB)UHd9f!)n=UV+hR@0L{_j2Jm
z=bWgDI_|j};mL`6<9K&hpa00a8$fSz(Y!_K&hkBW!_KB==FRiMwejOQ?(avrqT_?v
z5Q-^3)}m-RrJ=zog3)gMm*6<6O#)(U=j?;6j<QQ1@_n^`1A-4;uK+!D*8TejM!cE~
zR3)x#BUg-y%v~qx`y{(vicAlvWtPz4HfGdhxbcFGz0Zb4CgxM8zjtImhvRqT&#s^8
z&5TC}MxYK83J}jj3)*Y?$UTJ><ukG{(lYZ91gHc9EgM=imm>NqRgov-t=^Nx)?UyZ
zY{a~zuHmXr?6o;yxu`o<Ohini(Uk)oo)y)#$0P_dA*r&u!om`5lRo7;%7xiXP5~mz
zy^&MmKu?}Xxq8+Y#3+D$jfjW~iVwbD<W>$7I-^lP@-K4Mq>GEvQ1gHFO?pUokiGjV
zRp6rT{lBMsJG~fq&&n-;3|QRCGtHNjm=GxSM_f6DNd$-Fi;$E!`^f^nds#%9Ha!nq
zUiRt1%4ackl@}o5^fP~un~|qdFyge!36b+PUxvid(X-Yc#LF>cfix1${$|_UOH31U
zy?R)+{li0yM$N7qw*2rsxWxr`=1r;A&UrLx&B$Y?a)~OeN0OKlZX3T;8-j3`mvj-A
z>+r%aQdH_$(Cuh<RQHt9V2g=`Ro;O!wyLTOBLm=bP)H&ba>6F+)|htP;PD#3^E`fv
zHwI`@)w>iwItrRCE;kdxW)qB)j`JP8^{neow=7M<-pNTAi#=vQRx20G6olu**Vaz@
zjLQP91?02-s_98aDEm`ZnqfD)Y&9n-H`vs+Y(g47NjoR*1{ollX2K4?re+*Z?znp|
zf{TOcL>$a9%NwS;7l)<}P9Ea^drd!Nt2s+&0&-huv`p*)iHy|k>6lwBmX{6e<$B@I
z>rpp#@$mfhlL<jn8YZn|D2{37)41=)<J3uY7coe6;=3E+gJoO2<?^-7LJ+8d&`*Kl
z^T>VLV^wpM85mZk=UYWtAA#qK+1cl{8qEwYcL;QZgcAY7hc1{}J<Y|O#U=8f*^JiE
z69>UMdtP+fQ(xC+O*M_(!;*zE-*c-vt@?<9U}?^ZlFS@0j#F;;9Dq!uwMWiNoO#50
z>N`7o>L_xQyuwCn_#i;trK>qnNPBIJpN<;3D!?!;Sqc7HUb^+;U1g)h=L{=DUD5&N
ziE7R^-d@j>#sz{mww50&LY%{TT8`E~-v&O!$>yL$Sk3wXer9#*mw-UaQsY)5?O*%8
zzC+N*zE6gEc|~+!IaaNF!HEQUyg<|i*~>}aaIT<-s#+@IkS?azT@ToAKny7?nkUu!
zzV~F*n7JB`0_jzJq>jEty%_Ta+m|-p=r12s;Mt(ysLCg3<+w$suKq6xkV^phIS@`|
zR0I)&e*=hRdm`Y)O~|kz<SZNrdZtR&>|1eMiMw}Q?0YNaZ~i+hzb*Fv^!C<KQGVaQ
z_!t<30RjR7Dj*05($XjrLwARObSd4BqJol2h;(<1NS7ceNQuONG>Axd_ivBB-=FXI
z{&Cmu-n;I)vtX@R`Y_Kk=j^lhYw!I!=VfS&hglX2LS{N6iJBCbb+zad6GU*~0yUw#
zy{hs^s*>CBuFGJ+=Pnt%y$Ry&`Jfw{+4`sNM8>t0QWZ+m6HB@f6*#t)_=ha7^NlJi
zYoi&-8AoPmK0jbrY$H56;!w)LK2J+q8Q!^FIvh0KsaQMqGdR|ZckgUG3q9ld?9-=m
zjSyN8wsiBZ<}YZ>+N#w3OQ?L|SURjq*iV<}ogF#7Z4UDSy)Qj=Pxz|vICAx^_ibz!
zU&=x&<e((h>!QO`$TJvzoetl%`}bgiS@lxyJBhsQv2*i?^ppawUTSI-1$LO+w*p;^
z!W|E86h`3+hUa&GCmoI!#)7aLjtAS_eRrALtK>bU7A4u)x`>b7#f?HDHfPumm$@LY
zmA?&_Wf9iYJESS7kfgTOCW6}3i=y~jGZpUp-jf1H)`qpTw2P&JHxh!vR9M^FEGCl=
z_CsengTvn_vl0cS$#Na=Zp++aGlMvpAt^~XuXVIrw>1EKiVUx9<fi+VFVGdiF;66S
z_8vG6=6LX2NO<!FlN^PszBnTKjs3ZOJ0^>{I;}ctMJ4Z(6!%c){QcH+r%m6!kRXnX
ze+DfNjPpVpovqGKBj+irvCs9_d+!^VG*`&N?dj<m;*Uz~W$(Vwvr+p`gXOA_(x3U4
zL#DAL?B?d@zAzBKeS^dIf|)^+n8$R}i@!ajfO11+5o|r)I-{++Q;sn!nWr6BhH>Pm
zK@5keuQ-L2O|)qrZcQ39lj4iajRuAT$ESC6E95W72bNxMFzu&JNZxsQ%~mpQv1(OF
znjfiTM*#cn#B%SOaHhFJBw*R@ro<Ok5zn(vzI3$pQs`=2GjFSY5m(_5slcXiUx??K
zgm-11^rCNu&$R5_r+V%lWT^+<FQM4;V{j=vCOp?M2$SH0AFbB%mWm@zrq`ZyDJ9$-
zN-<rP6{mgd3Uo{)UFQN)gL#@IzIeG_Q^P#w)vNimQ|u%cMCfikfIQ6SfaO6rJ9H6J
zZ$T-=&SjR9z{|m<foGQ&);fA9`Ib6oP+o4ctazjYeBRyZa>-){CYi_FeA=>fCK4_S
zoYW;CfJ!l4@vKysUySQK;Uqp@M+nWUidT?kahH6bzTDWRD{b(YgZ&g9@#e2o$PB%4
zLz}J(g=u8R{Rc{+Jtm+k9giv`ajY3ro9)N|I~_<stp4j+$XTWbhvKvPi`;T`m4bh<
zi8->!jx1c1>*+3<UKnqj%(EF0@i_j3?}nmH%!&U>8BYqoghIU7Wh%<9<XNnjEFE2l
zkie_E29=r-zQ0S>`>I@ihQ1L;$@?Pa_m&a_2BF)}xwgixoQq|p<#`Uh{@z@J8E74x
z6Xv`Ke=M?NO$}AmZi?FZ4b9uU+`sOt;ID<m+bJC{O#2Yltm~qzqK!_%oIi0~a5@M%
z=u6ite;m%%u6O5-cS`T`p5eTpAT2(-lih=*^gN7Fmb6F)*V*3|Che&wki8SjUrU9d
z@Xa^&atzXYNcJhR@7L#){hkdvnE5jG21TIw9EzuC^Z8+*Ubwj|8KFkvMdYV$xXT)P
zQveGl;hmd^u=mHi6PMJ!Mi;v?Lh}F`q)wZdn&%MZ=pH$~XH3s}7d^O@qLe)=90FCe
z{u+N+)PW@x{DBBftg>|;v=Z0MgnVW4&05~-Lx+HO`kfC!P1``0HK4_n-r}1?R-Twk
zQIT9mpZuX3(zY}l%zk>kK$B=1fP7pWKc1^BK`*V~r!8IA=^)yn3OWVeUNvTQs396p
zMHJq7hpoZy>uihAS8c>`(;{U^0g9|+CfrF+BL9*<pL?PKmYq&1uJ#=^=Q&#{DtRu$
z`GjFq;z%h2yVK$mN;mU<hxo!Z&AjCDyf~O4&JyRqQ^6!-79I9<r;aTHnKf0Z*FY6p
z|M_|F+Nk*TH@JvvFoc|}2-&tmzmOo@cO;4G;wt3knq1aQoCyjIP<$}6h}8Uuxm@Hk
z^pZ07wKHUB`$A4v7p_p>p>}@J&ZsySWVkudy3J!PDT%)6t$~3=(=6U3>nX#OxVzQX
zU6jbX9qIf@!kpkG>@#igQ^{(%6t$M85e=+XLF=dO`=<^oUWK*Dc$8VUR)xx8DQd^J
z&qC1#QaA}laksVQR~`(Ap~)J-bgjz?O-F|}()15WFTH*fg(+%Z`C@4N@If7<{*v}T
z6zsGszPUZQUJ*B=yS2O(Np)E|{mQ}ant-fx^&LCz@ZI(6+3#PDSo1#OXQ}%3Sj~zj
zF;1jI_H>Bx)$qdcXRd{qrqGjH4Sca>uOlMcEfCs~MFZEF$iWJ&cSw;@JriKVKn)B7
zrhJo@Aw)v&23IW#Xh0gl6ahxmQ#&q_oD{iDgbl-8zUxCm3K9mE--Z4wkS5>6)j`GD
zbSnaY!{d{%h@;EohHiN}e}Tev?9J@wO^sH;=x^W57RKptlpuvrHjiMgf*0&HLPO;V
zNL>9U*EInX|I?>epcHsLQbvo64DCPK_u<AW^NZCO%nTzbS>f{iGf>D7hv@v)%q(ul
zjzh!DOrS}t+L8b`hkTWI<HklKV-p-4uUF8b?Kx8&hXxof4dzIP#U=(f0l{z`y?cq+
zId@n(;7_dblnRuX`>9D3m*C$JpwPgSbc9ee@2OAX&#vyxVQy!5@<H=aabBSArL!*o
z+jn|&$8uG2D_+0-lsdS1+v<Q9pP6}jeyQmxy9l<w78X?QM(k_U==p`v;6jd3{WGcE
zw8#f2X9P?}@(ciET3+l_b=oz#rTM{eW@+YAIQ4*vj`AnPLD$(UNJn@7s{=!?pu*&9
zQp8~vRZgUV9r|6BExcuUafWOX7v4MZ4B9n5V8Tu0mc`V9dY>iO_HxsZCdFvcX=sc&
z8CHa7o}-!YP%sF#R?AR&zhFi<tXW{v*?gfbD0pYDrfg<dVegdNR;*s~gXWA|U^`yD
z7;<M!5hgpe2%+OTp%ZSXQr|V~X&t<0d?1M;Tc7odR9@Kdur@KVY1<zeMs0fypAt4>
z{+FF_^^Cl@&4(L87jzZlkB!aLxp*)<77Kq6Jy3~P>u2yAc<9<Ijn+KIz{3{L*Wjv=
zm(6pwsZ0wz=_YXdNjUnZji~qAvo%gp&);!bAtx8Ssww!8i-5X?O7^b$d7kecX+8!l
z3+4zJVuHZ(gmBkfd2<9g=e=>0%(+Jl&Q+f$<PU`G7>Jj%&O)eX)v*(%wh|Ujgn)3}
z?(`Zo{Ag(lie-7rxiQi4!(*1Pl`ch#=Hd(gNCqG$X;DzT(bj34Y_pMJ=0U5dB6M^<
zUNi1cz5J2nZfmeCl6X|}6fd{W;m);J(CYXR^X~h=r*l%Os;kT`6$#-|lIme<HzGMb
ze?9VX8}Ap`6G0XPX58J^o$MP)76!YT8iLKYXnsR@2wnK$UnnQ($#^=hy`4#w3D=^o
zw2CY!lxu`miF>U%x$!E#78yd;T*`HgwMSXxSd!*o>3z&@{4vMT-)1t*OEkh?JRu(U
zcG9@Y%F+`{-(ntj(BRF<aiCZXIX8baZ=zFPh25);rl2aa8xe%c4@u(w4qvTaz*v;*
zMJPp$o0tRxxo-xlYi6dZI@zcsM*`&S$6ewD@9H_>Z#6DLKRNLbhIH`K^?kiMOR#o-
zgaGUE{ejaNdxdDViM9GJm5}EzOstLD(s~b}9$(^2U?Aug2UH)Znu>Ds1pUViJU=4C
z5fNVg%IW)h+PzQTRWs_^NCL#N=<bs0>-q<?>lf@h8iOZNEfG%69D@}|oNhQa7<$zp
zMqHu`KO~IJ%z{@>`EgP#RM%|pMrY0p7x**}%%Phh<;AtWf!gKFm-XUgJI4W;N3`l2
zWx2UgP4j9sX48gh*H>s+lW$>5{GJ;E;OFT{$%Z>K!-I)bcQ=3Mnwyz?f1){JzXcl;
zghcJQ*}aZPAd=Hzvfg#O+OC=*mR2&O=N7g%BBLJlekhFNWy~qi-zZrxDls}*KMU>7
z2iVEn`9S$GFP%neTb&>MoGl%(Q-ZE_?OpfQ_&jx?@=8oWu6P%Hw)c0XrgGIl|Kp9&
zQ5<t|O(XBOhYBY!V(*(x4h2AL!PQygXH==;F4fkA@@2~XM8r=aE8Uy%Hp0~R9_k4J
zQM1e#)-1&!-Ak#{o912}G}5!mfxAAH8`Znqi=N4;MRdU1iK}s;hN?v>+ZybU$WmjX
z7@HeT>LPUI3TvKTn)wFJZ5;aexG>0LHS(YGn{U#2n4m?$7cwtY((6C9sbMrz?Wb&0
z5$(w+&6Jv6GZ_UJ^*Hs^#zGF>#@^5Gdg8A1SUJ~_nEX>(QDTcCxk%N4{d~VoUhd5D
zX4;Eb-S`dlrMF@EXnYS{zC|~?B|09yvJqP^eSQAz9Yz$C%!x0OOKBL094;<8EpHSQ
zRp0m4e)n!j=S{q>nw*v#VWH(rIU*?V{ZkW_W|be-=9F}6=atOR_9F~oWD@%D{K_We
zZ`W(epRfO+<2)|8-7r%8J42V4drRNg=wh}yRWi2e2hA|WV5%T<WI)#Q``0H(l$axn
zH&)Z);b5?{D^>jcKi)s+B_wN$xeEZ)lE30o<lCP%=CWL#p7F8Yg)MF({MfCs!(7+9
z;q#lNh1VJ3pyBF$N|qjbVJY5Vs;Ny86-S5JmzVx4_4k|@KZyFcj8WGq=(lOfJ4-B&
zr3$)vX@HN5<rXF@fXilGE8~>kuSm@#C1wtru_5`Opy{q0JbfJ=8(>mKYN*jFs>TVn
z^Jj}l5lzm_BhNF7R8yyv$pvKzPMri2@&kX#4ad15tiuA(poH>RSF)PSOt4e6c`4&l
z3e1sR-v)CHHk_$id+U$u5(IjRD5zSt><t#r3Pfr23=I`{T<Okm3(0BfT;u<5^@>Jf
zTw$+>lM6z}Yh{%H%JqUz1VcGrtc(y1$8?4#f$_IcmR}gkb!QW&RkE=_LjC!`6g&3I
zPSqisS?YCxX=;IA3)9Wcb#3R6!-Y8e_^L3;Ep_iver%D@b0tx>Qf!B{>d<Jq>w30r
z-1^VPQ37noPh^rmCm1I1R}0T(oM=bzFExPJg|~UWw0}wvSvY=qgyD=}D^iIFUHs4J
zQYzpuVExeN=RZGi)^mM0<K)s)mH0Y>8zmH9jG?!k9RKv~wPT#DJmU|n_R?d>odMHY
zjE@(%(#n78XeR<RT+!nM_@_&<M~MiHKAmKGf_H6~(Ry<HSY3aQQ^Nd(Bv(-h2Yy(U
zTG7##@-ztkuC?}tSG=sexenEVX4MpM{cvsd{NUaX*N#LLB!fFBXKiwR3ue98dv;*w
zrUV~87fNG(*uwKUE!{GB$GbxW=mrqUPA<tM`i2K@8{Pl6pqhM>N4`_)u0_(U3SKNm
zYHrw{@`oDM=XcUENXnef`VIbES4V_9-snH0!YZ2d>uPu9JF{@g$nb}!2uG{XYNwmV
zNy~idt@mBD)!`c_q%{!t6}T2I(@#pR7=16<RkSK)0(jVj1dBmW<$9lUUGj51d;9@x
zd}3B-uGs_b=PAhOz82^g_qN(ij)#;5a*ao8-=CTU-(M1^SMYGEtG=BGLCVZLw$>EY
z)ifH@UeIz)6tROlO|6P$F(jM@7Qqc-e$N}%Lgh^sevG#fS_D5m`Ej%|{u~G=T5}2i
zi``EqTmDJ6s<*y!qi}pw4a;Z)@A**qF;EtO>=KFzz{#Pp)t#7|Z2<}zu<T}d6aZ4!
zK1U=USX?#HIr`|a(HTKQ#d+q*3r2CJx&Efi%nsm$txVS)9!1lq$EP#E?j?rIyr4vO
ziD|6=z4V{q-BYM{JZ=wE6%<jBj9>o2@ft|iZ~diauu?Bnn$6AGP$h<RU=evaQ_~IM
z<{PoB@3V3i+K$F*l=L(dj`~K8_o6Q#)wX&4!p>KuXyQLZdcFS0-Y+qq=;^l*9QU*9
z;<NJIbrVQf<u&dedp6?X?@i7oGA;h^1mE56gPkwqPp1bzQXd;Jh=8p(o?CY#X8qEc
zxnyN<3@TIttsrHp&g@UhuMfSpw^>mM!=5kaPu&OyVu9j<h?pgk?rSa?oL6$`^TGO$
zv8T8Y0gP%<eM0+3d(T#0?GsZzt$1}7Ov~fE+4i!|dTlh6o<xM)-!8Z1ohnOo-6!O8
zSUa=6VPv5+<#}6te1ehY#Mr?^CSaAmu}#PG)QpfBjO+UXHLSF^T=PayL+3z@v!?y~
z@qzYJp=*6e<ebe%n&80vTkOHK-D8_a4*^j()#`zNW*r{Z<L7b2KOrRSRQi~g&t&H1
zT`%GH=OghMx<wD;=9-(gEF1q5V2<o?efHafWIj(|+ca$Di7uMDKq_NeGmuUig(HrK
z1qJ}k>epH(O&^75g(CZn%|*1a5pvOl&knQIC<BetqOB{x#PqJ`3k_-Cp6m1|sp=zy
z);a7zxlD0+xpAB%FQi&$skDhi)y-+p0M_AX?ectnaMK|Ret;}g$AMKAG+G}1vTfd@
zMD>K>bG@-x$EepR=*xLv63Cngtj0THRc|P3L4Boh0r(F2nu3u!NV_k|H7FB({?rmc
zYv83|UY(wOQ`XQEmsodKo&UfkP%z>+<ih1=zIo*y<dpLpQ5<|6*{*}gv;54C_rRDJ
z%Fq9C$B5n~V?)sA?6TL?x$53@k{9Xe)%z2eEti!48la1R1_(IuHU@XjL%k;EYcAm~
z;z%;3(rq2AnT>^F+Noj9LwO5amXo2kJ)@hZGF(=8|N3}n1*WS+U6w7wyvrLa&}&%p
zRHi9syeNByP^;<<@Qx;oty|x|pnYI#SFlycTVj{5Tew~lb?jZu-c{JOb6sBQXQdLG
zIV1#<d80CSCGfzURufq{mOr)dASfqCM~0H}cRtp9%L_Y7Eu<w!<rl&=>wxO9J+@6D
zNBD4iu-}1>{{>N#!Vt#tXbNY>H`Zs-_wHWv?zO>T&9Bb;D5qPDQIDxym;0mFwBE(M
ze|+e<p<^qWqf7qr-2uHRK3+KA(y#Ug#;><nFgzl#jQh@0o6o)-RzsK~;kbU@2H$7v
z0*wfBiB9`z7L60{A1WEbP6w)?Cf||a<<i36W9dpEQulc`F8v2_9y-!nIiDB}J_=RM
z#hyHUdJUBMI89q3P771l+}v97_^w($IaESL=&?o(4F%NnK--IpJ|o+M0}o7$<;>6&
zMikNWFGCoR8yFOdR+IeO!SXF$vJJ(`x9VTre-AEcdrqH0`IPboZEht$3#U`~o$)J5
z^AU91hdEZth}-Yp+VAH+@*cV}AG@6u+xST~S?*}taWx>2gX8vtm#&Ub(jGH1eU&UH
zW2JnyDO}|T35k?}JD%Su&X34Ae*&e*`d!6?kFxE}Ncs+4(Iemj48%2U&(2Gb#eeVN
zaN*=LDNZ~O>+caS;KbysKrcIZeD(@jS~6K>9D2ciD|E24(G=(!L9@Lt@(D!UzWZ-r
z(qYdTYygJ+wZ~A9Ks9dfeNT@j`1acj920Sr;aAv&6T)yKRV8=jv9r<I{?^&$O)KQP
z*59(*TO}Rkr;si*cvnwP&4d6q#)>Bm=+oRB(ZU=n@Cn9t#`UXw2ByG&lSjtRSHMnQ
zUuay2nM-{W;th_`{ekU(>ZnlJ((NKC!U<r^S#<I9@}#SvC9p?{goH%5Ik`RRxQJ=!
zp+i=ka$J>5mEJ@27*bz*SKFO`Hizgg$hp!i+dMy=Uz|rME-p@JeUplDU+}A6P|uFL
zrRB}^^mLS#j*hX7jhASKI(s5=$PY5pCLp<W-h<iq>JZ)aPlICDKY7~YHh1kUoZol;
zbs3+UWdmBS(V}Y?VPlKIQsKE5ek+Ho>d<FM?Y*qX$T)&{I#Cv#I`1{jcdhj%>(4-E
zDj)(B=%QDvn$uK(P}$(C&9(g4xuS#1>rUPe<aOppgL+_~Y`Z)XML{_ZxAU_qb;9aa
z#RJHdTlEE2ji<z5r(U)av*qTniW>1*e%;<aog-}v$|yU@wZj1Q`PxU>y_bRF@>?O{
zGl2;rWnz_QxOxkCbD_P5Uw|+MoB|z_AR?lj(bspYMNI?>TwFYY*A%yA4CkWDj7fo7
zF%iG`$zv;AT}N___U-uZF|qvtQhDB5+}^SmgHVl8lK=Z=33&#a{d|{ilN_BUroq1<
zDandV@Vxzah>y^UwBQ)@G4Y5vQ2Sw10WdR`+os>F+m04=nh?!j{#qCE6aY@)(!PXo
zcQ?BA+ox|W0Z>-NDY}RFe3Yg?3PQ>OvaIgRGiu$oJ|$J(z>h4P%C4wjijaxvAG|m=
z`Q>}kHzJ*defvw{G5BJPQAdcgXkBN5Jc;BRb)81o99S^*>kUw2eWR0)0poPt$y!n|
z)a>$z#;MO*H&TP2QBQSqO1HLqGx3h>@y-2OYFZ2>fAcoBHNla*JUZ;2jsgu?#*OlU
z!Ol#56m-Mw*3WZ57jl1pmB4FdJ^V`bC7J2=%V9;_R4eXx_&Ym0qTuXoW91Hyq9YA#
zhq9D;ljO#P1!5LZ70H0^?}vS}#x@o|E_IXqi$=E<ai}@j_-Z7l$#;|)A?2fvrd4O>
z$qJ5CrPPePVnRBT34foG0=A4s$spiPzhv9R<m~aaw)918J?i<;(g+pL=wtbF!%?Zx
zGx;JCzFUWnN{S|48Hq3M+<BXk5wz6*Yy}ba9MUP-xt;&f?K^BiI=0ZEEY2St9j2yr
zA0Wr#xpTpbivPlm@md*}SPz!m*CZ6Qia?xkkYE7oC|-Zm)H$$wS5f}L>_hd)*UWjZ
z`6@B@+$X8M@#_(Z4PnC``i?glq5hU9WdB8M&ynkp4$eApBAfwVsLCFV{bXqQ<$9QE
zaQ)49h%^biwVR8ALOIMw6Wdb?VQ7X6#MhHAlb?~Oyee_KE+tF4NcsMS7m1?ISW=f;
z*7|e42g*w1KJNowQYmYspTQ@VusS0jL6*iqUo0lw@$mQIVO<-s>XYKw=4!)^@TrP6
z;jOl*5sDjJqj3no`SM1k$rvh>=2&25z9po%C$};ZbJRpL($Z8j8PM|e)>A({#hfjB
zr{^7g6nAfKgxYD0`>f)}3wYg-O}=|Khy1dZ`U>v!qdJ!UZ=)win1#7i%Y?R4c8?}G
zyi4qA2)ongbned~l}_IiE1i4WZFOeC#;)d%R>+1MAc(B~lDC*{_@*{B6%&@W$_X@l
z59XfAY!q!wKTbC_&6OJDH4vp`sqT34O?~ai0t;J6SXg_Wu*K(GCB@y;jZbnpMGY;b
z!t<-yJi>`yz5U+YOtgqF&j)>Lylpbt<T+0He6&c4xhAG>K-}MX<Chfol6|Pvwt5Y#
z%()1ppux1$ih?w~Jv&<`-0Qw~3{SQaoV_z}9ZlJp>8jp%Lg-%YH)`zFJ6f%XaMm!R
zVz#`S){9+t&u$HL*Gmyib<;nRCd)EGV-jaGKYZYxpQ&Y;k<PKR**I<c5!4BneiYn)
ztQltGmz*2Rhiw_O?{qAziJW5kzQC^7-H7N{-I6{R(GF)>J>FwNchQaJYZxT2rR-_N
zdfh)dbT6wHpPL!E_AFQ6y3*Xv-p&k$&OWl0dm}1R^s8kI`EV1mIp3Z0%pc3iER(v|
zq#-VY8<b_XOK!^bCzh&ng`00yCQq8~u4hn@T9qYkT^8>$sj0Lcc#<2r^k}+t3Nh4h
zF)-Nh$78Ff?%P+^EVayUw`_76U5zAHkK5gF@{{ylDUn07k2o_D<bARhcIv+Uh1)iT
z^z^xz=53zJcf?C?Q=2wKWoskpYc^}ce@(TLzrsC1(g&vHly{$nJxR%rp0S-h*L7OF
zAUxDauV>}F%Th`*$E^s5)wGcvV?>!J{Umc^lUcWUhfsEwAJIm0N0oE8d7#sE#;xk6
z{5y}Ny+wD`nj9UtXZ+@HzgAaoDGB23>{*YGCTdtHy&jSIky1s*t|{qrx1#A?7}@+l
zuJv!EAn&+;0;QmixaU@0bRF)Al$w;sBse4_`fE2ns%#u|h&pU-5uy?HMgs=%t{`^u
zX7#6-rQ_Ia%lqXTA9v4}SoX?|&WmQ1a*6g7iRN{0nlJ_g-*0<-$?9>J57E&)3(?NH
zZ2Hvi8N~*_>G;!&;g8=S+o;(Xmnh!808ImpqL*h5_M(KX_}0Yb^`nx6XEFmy1XY!i
zI$Rko$M!pT7IRHoT<)wR^V@|UE<;*@VYuU?#D*JO>BcptovdaWPwouZP<DT`eX(ar
zLX(jopIR)!p#I8pbGC&yKle6+l6!tE4|br;u&<leD=k{~V{h?BEVu2!4ZNM5*|4z(
zlVoJGZZ097`2N#hrE7#9`$}>cGL5-N$g$)n9%dYFB1c`-k-6*07Itm}e$_kdG#@@n
z;v~2tkPts{*nqB*jh(}K-k9BGsnU)4BiuPliC_Il@X`9aE-o%?DtZDxctp6bZu+-Q
zVpxtm*qbc&{JGviq*C+zsL;K&+Q7b_8&t)6k3MWBetT_n*-EYU<F1PZ#z=Fc$F@+j
zb+n^n_N!Pv(tU?j?#Sa|01btEu81}%hJlJWBEsSRo`UIDv&&Y*7YlxbQ1k5P7{F>Q
z^YWHq`+M$a))jabSz&jTFegik8ksFDEBF1dh|Ma@-0>3mz^q0Ot9c7y9+zVLD|iCo
z(YLA+F}}}ye~EM8mtT#NGRz^r5!6RO(%8x^kC;wPvV0$9qov}yi#x1X%sI7=fG@L^
z0Zm@>UdPglaM0dcfMu9mDctw?etrnbjl1FYCF08uoC$F@wM==*ZJdmBL@6vHE?Rx&
zqsoVGo1NuG4KiDk&EJ0gg7ayJbgF$)eMH=pq!i=BhWiwi$BD32`np6_IZ>yaW>~3^
zK$pK~BqtqbTCQK%>LDrmmrQUzn2Lq|w%9Anrick9wbzDU5@|%Q$h@I<wp;Pf37AsG
zk5`aCa(HvNf{s&2Xm%Q1G%w2jKB;2mQOUa`-xY)*%XMre^O!^yd+Q4HS>^sbFbbu3
z|0%hhi=m-@UrnrtVGSZ_xY(b`u`sBXmH4ZTHD4qy#cyLaduwtb=Z@(@R6=w%2fKWC
z%Pn0~>B@ZDAUo?m9q$;4Y@-KBacogxTt&skN7vf>4~ux0_CV0>Fe0I1!bRA*>9XY7
zpVORBW4?vEX9Fr+zdk^Bty%LjjcOO}I3^xFuEE!5ldm><jr8TeONflkw5IXC$5Lj4
zoxDg?VpM<Gn|t1aD9|!_xlW8C%*&QnUqe(a90EwpO6of1+OtdZQ<+m!?R}|S=DOa0
zj|E39Uj-X%gTUZyTUIcD2#B9&HDp-39(uIJ?#aEC8IMnJwlg)=F86IvNq-vONY|#}
zR<2?ZawDL2g4a19h^egRnq}Lh?IolsAD>{kPnV$fdv)X}R_^Z^^e8tW{A#B0qYR<E
zMkAwLizWr(^>$`{y#fLOyTgtur;a|BhnM|+i)?DXc=24#v1`C|@@@X^ygx0LwKzef
z#`68Wg-rsHQj^p(-a`AQz|SAW<+lwD%_0GR4p&4y5qN%_UB-g)VgVp)$&1O!F>orK
zF?hb*b2My@+)J2DE~lxK$x6@WGG?a=?t@Qz&f=kqJTVo#Zi}fqVQEpj`svGcb%7#>
zp`GnKIr`Sr9bdnD2EJ8;4$Yre&t29Jek#t^)TtZYqdi&Vnaqzn42CZtTQ5@P{AXiY
z*|6ywbIqeoQrX)5xIh@I5-Jl$iN#>f(;ft#a;~laxz58r`5B#mutUysK0kACpi!+n
zv37kvqdeYoz5l%EZ0DOY4WoPi+)V-|BcqkA;{M!-k9cC^QBt3%Zq_aI9=GRU2=PLU
z#C?mOV{_}v?FgM3<~W_^obOVpT{u0FoqM#6pYphmfTU!WyM?grNNifTIVUNN4kqXN
zUxP|Imr|#DS0PjNoPWdYWl}SNc=Dmm(jOlyZ<S)gLKD9?4tqwtO(^j47gru1w{o>~
zblz<+<@8~(5@1gm9Z0d5J=+Z`fZeomf`S1cw}vdsm~zmU5ZGrmm~%Q$Pb>4BIhUh`
zl;6pit0j3;E~mO$H^r3`VwvDA;`VcnGAE>Ue`(HlhSOcdjeo7o`D~m*6!YPfUJu@H
zehg7Hn^>4!_ZP?H5XbnbwWEqqINhJTT(7_wql=oQ!LKHbh>gt!CB0k|wRecmp#*I=
z6H=UDSN%cv)gS>re$Mr<N#-?zZMQ~Cw8}Hok<66EWw+AM;&rDI%Yn46gJk0?&cED)
zTUs%7a}2XLGP|4oU8TeD-Sq4Dw}p?PnVl}>D80~V660y9ec5ru&(Dv2_v<I=rAuF(
zp9W^0SeCow#=^wJBuzxg{a!%wMaA*sc_B)O=evy6wYABqDaP1P#TS(?2?@gKA*`tj
ze0vpF<b7Y_{@28&Y)MjqD~msa6NlMigPoabw3XkCcG#8OIh*&(p<<{5kCv8pXLm<C
z?u>@R0q^xzP>#v@93i*twX?I+WuH3kR_nLF@3phL`?bkGw9i?4%w6PYYI5=>`_jBQ
zQjwH&wconaQ@U!HEfL?R7jKMWn@H_0Z>;>sj|rYTahD2)+D30!-nQZ0QCqIuPkwBI
z?h2V;WQms8-T91$BQB?9ud~FRCY>mzV2;UbFuSaWuxoDXuJ}zojS`%i3-Sx@!?V|1
z-_VE+l|L2E?Z(q^d#K~I0r^$5pYlovF|tf3GPSdhERJ2ErA^dPXMZq;ue7yNdnN1m
zi7FR3QB$YVzBVrM-5?XkMwUv^2WQjRUhpK@_0RtrdR8FQ@^SUCu=l5C`?6Evf{m3B
zebU7#?8d9Gv^NI@1vwV~T-boS@K?nOOj*LAQ9fMCzPtSeLGNu$j(A%Uk0@;oxrwVc
z{S|{0qXsvfb=_M)uR!p$erjqzf2H&I&p@3=wi9o-<+ZBy!q$5aj&iSpw8?*WAWARH
zgxuhmOzYgw#U_<jc5%JwX>SKVexK|TE!$7C7S+ze!6F(#NxrQGRdGv*JUIQ4u^Fq5
z4<TaI7ao+&%^}rEXED_U?0p4&PeMhzlSJ=ZKdic}kIj3Pa?dw%$!S(eGUMYgB25b_
z0-I@xVePlq2aXiF674DNHg5j?s&&tl#_)2qYU;#px&21BAGo9^#0o0nz?A(aEo;#M
zmRitaAg!t=)i#zAk^9I0RX#b+u0#YV`RucT&gsKKokI6qvi#8`!*euUp!o7@X)3dk
zh$_dms+7`2o!vRm63%qa3;8FgoyU!q<S~~jzxCPVYc(EP)c>evIbXmtV4vigmdvH=
zp=67a-I<yB(3i_diy*lZb@#sIu_BXGljyMHvwcyaiHTocJxH(TGN(u1O-?3bd>W~D
z&VQAQlO8Q8_QdPnJywO2ETq4uSVXy0KM_&2n3?1o+yB1fECaqMOpX0|{ZH=qI-i=u
zSBdb$6dxpEl~E{z0N0GaLxLJhhUx$N{)vAI=nVXQ(I5fG+5dU(#)Mz}&)fgm)&KV|
z#?3$OMe~3EqyK*J|Ia7m?*Bgw$p7(k80MaE%b`W~m1kyVzzLGxmRGH)r~pZ9oT?}G
z@q=Vv-{YiYWHMpLj~{<nw8D4W{%v9+JA#HQ{_$iHQBjQO)s2nFVeoAi^Wen#$izfU
z9Y7F2Bj+nxOziB6U_xP+E?s)~m{RK&AD^a=kI(AH<U|Rd<DDyT<awCdoQUq?kHUu8
zGpHPj+NaD{Sy|--T^D=4#S7+#_%a*E+@S5b8r90s&CQUlQFHAXPXLlYP|fSc*UddF
zai!t*cEfqM4iviDT<meR`4hE$=TBxu*vcvKwcqW4J(DOY%sN>vpxi2^5<C?G$!>+w
zw}bY0ynyqEI~P&8wj<TrTSJba^p>ArUc{-yfx#y6UKa5xS5Z;<H8GJ&M8&H`eOm0+
zZO=`c8qhTgO9O>-#NgTK^QG|j_Vu^_N+AKPRl4N%7nP#w#FgQo{lBmC5>NjXH5R&r
z_peU!|8OxyhGmR~KkCWARepZ$OojL`dJvRX0SQ$y#^Vo4=I3n=cIPl*X@0PhExR+|
z9}y+0`H>o3JUl!&mY7<=>Ak?YCpu$43QYt&*0bodwTsN~2?&aCk9Y6C-CULXpGW#%
zF&(dbI67TJyT+qr#BCzjVNedF^=v^?D)~OjDi$$m3bCB&N|%wA{tgmwJz2_h8#CQf
zeFmQS*Xii;Lry21x^v}R@D=`5-+>XYwRTW-*W+P4uB7`eBO4^?y!RH$28zr>KNJ;_
zQc&Pj&n=h7J_Q5>d=R*fV!!*8{7E0DysARxTD2|8jkcS6V`nPSdU2@Yi=Co^f`*=6
z90E$?#gE4C?d+~<;4hz=@~l-#bDo3U_b0_{#~<BU|6vk(^XAR$9}CUR%`yrKy-|if
z*&t@FiJLU0YcOe<#3+NMk9&*VM0`WItfb_tFCGE6{fH36+u3n{+TKV_gX||yP+<DA
zU*FK_RJ#>oFCOc;ASx>Ay@Az4OczC6=HK0rkD+h4-BV)SM=5uzPA1W7^;K9H&eGEu
z6Ws1C43vapFqm&~{Hb38$ywOg6qag_J{FY!#OJXa`Y~IMn!WRT3~y}1hSvc9*nZzD
z5=Ptupu4J;KIZ=Tp@nzqRLgBb00TD#1ay$y?!?PA1Y%*AXhG?(5;@$RGc6m;CpwYT
z$f{qd7RGJba>;qN2MtRLECK`=`@k5mAH6-%D)%4&aU63yn?nh+Kc)5l!PB%NM@!D1
zr^MXlx_jzh5HP&5KXJ_9Gq?QYuV2z(I;A$+R<Q}fo-$#e{%6HXj#~O-z+-$FmR6H3
zkzr}pm>Q(_CpXHf3bf%_7#XF;$HxOgLQ25~DUC1tRk|**p$FiMm~2~O5)uY9m}gd@
z9;%vr^w!C-jkoNE1(@1H9hexa?@3Pw%Iq=LID)o-6J)y@O(&F|adL91DJfk=TMd;L
zZqE0o;t`Np43^sBpF5`km2)1a>F=CU$Q%@F41IRe>1TR%?I(p#jG5k0h|Bu?`F_pL
z?^uuBnM_g=5;_1_@Rg?LFtxoP&DWEo&a$OOgi?wp3}^Ux%q^SNXZtRuW>Y76PtavP
z%RJHh?an%i-C?ZmnDF{{@qm{vzk|X(kN57n%HS~AaLjF^d+mNFlyZOl`h{44L}gf7
zsOX{A#w->V+LIWU*`Av?zcPv(c^3{od-lv3jNBT`dUa)`kck>>?u-G-0HCDwd7<Se
zFh~Wbh$z3_CWI$sx_0euccx;WqUb?dpCO`;><y0Qh2SUo{hETq#Jjq?;US!YFLT^0
z8+M7kckf<AB9e#!TQcA>Yx@oph^=rm2URKT&(Em54?YJTXqCPK{hN!&4Vb~5)E@3y
zzZxtmDvFGVNXyG(#bOLceIKH#iNF;Z<XVj6IpUo@or@eDT48e{7i*7v@^#A;K>3fg
z_@vvb6lGrU-NK%mJ;Sb}3nSzgE+owrR4Mc${r$UR&(3p$Oj^y_H$e>vya$El<#*8L
zQtKy{MOvfS`+_LAF2T~)G;o)=rCS;S>%hUm0gM-e9Ihqml-b3beT&P65zwM07o%>v
zZjDk3Fl&#gAgHTyS@>{==t&GQCFSf04X=$>w(Tjx%U96?Zp%p^ZW#RQxKdRbl9tBA
z!=vs}6^|EzVuuKnNx@|%9hM;<n}HKy?C4PRGh(`PM+>J*xmY>JHaR(I3G0w76K{+N
z5QP{;0GYCkq9R1FCov9-eN0qvAFJ50u?Jf&oxV1x>ekgssAY)zpELE5uOpxqa#ITH
zDYjI7;qM<ckYPXi5efTNu=lA;3Y9|z!^qTdc(BLt3Zn+A2vEFyEN>)zyoh&Y4=-dB
zUj|MrvjNx%Bxiq14@tyVSUGu0>qImSw_9vfi#sMAu+leBBf$V@BCg-MB~Qg?7eNn#
zcz;%yN~c$h1YX>zF1TU(Z1r4?)eTer6Lo`#&0wiKD4^m*{!&s>@<LA39haDMq-SJ|
z+BrSWW8EtqmXey<dh6r~G4-T6f;tw{hoG<&$1m_`><Q#8!)1_2SPoY@@2({tv7#$~
z2C(#`p-FO$kG~hI^*OjiOq>FrVKGwe-qPM4>Eu%U4HP@Mt$SI~%%vzQ)Y1w94p}x#
zGUSToV`7xXtCug8Ag<D*LxnfrZ7vKlW_I74nV))PUwfcYWZv=ivo+q67^8P`r%u!n
zz_{p4G=;L4$-|o7s3%}wCbfYP=6ATSnjv?Sm%5>1v3%m3$opdkOmuYO6OF;)^j|=M
z>h|sLKkuXR9Cp_&g}29^AuV6OG8M_gv*H||<nr+>7l^%X`NeQ1GMF!nVGm46B-lk`
zFjX$hBghgyVXNJD7zsLNzzPsV_AP7Y)+mB$@ITK2Xak0#dwIM8r`VV~aqN9vibNpp
z>uxLzYKDsJW?>(rwA#iWS-@J?DX{`US0pSVJUsR5S83<@AIh*~Yrvvyo$sRzeE%CR
z1?T?14gvnR@Pab(KmX___ySJTuR)O>eoE~}?VPV*KpB&4YvN5<n6dK}c%mdpNHjn{
zaiGF6)4**U?@ME2j!}ajJ|Ur+np(u}=0dJNjfeE($B%JKPOiIHz$%za9tsEI=loOa
zy&syHRR$tq;f#s}^-c-)Aji9CUF5m7xK=8<wX)neMF)4N4#!{t35hL4L`P@8p_8%#
zJxYd6XocPc1GnzY>l9q7rbEYwA|FAaQnj07?s>}b!tR;kJLm-!wfDCMaAKB9#5k;^
zbnounrylFm$@*0;+JDm%rE~b8|Bgy}4)ZsvKeVy7)}4#_m~)X%G7BAk0|!JcV_KS<
zJ3CXQO1$@6ZW&Z-9Xoc6Ec1^-=8`xL4@E3DQrVZ6k&uvJ0lgUS5j7}DOTPrq+VjCw
zcG#&~o`?F++2>cuv>~yqRn^kU0^}eQ2434BrCgVSAK~ZU-JS?_ZoO5;#LO)76oerL
zOY?EwlYyQ-!gC%pVTTCp($mtU!$8kC{ibTh^oLe9)klxsdu+_6myP=9_V@QswTHcV
zlbkLaWz`aKvvM{k_tM3S$$*<n?1ptn7!}mKysDep_jIs*d<+Z{iXuB{U}X95|IUj;
zh&Fr8dYD3@i1%Fx;<zg%gDKNrzB~io-LS!rNEI-W=f`oJuLgnSRLNimTG#by<^Bbi
z68q5+Z{5Akh2fn5ADm9`npxI7Kh91|sObQ{igEA`LNJ(HS~4^}V1u!PKUPstP$(R9
zT^h-UfIkNqxEuMNEKW*Jo>^B{N8FTHHTJf3vMW!!$f-kk>ymYmQNt;m<}ZY^;@$#%
z-8?b30>~nGEWRi8c!Q;2K@A!wu(>hcuR7|r77U8OEDkL^Im{T0WsJ^;hp+qyDOB>>
z!I9wRuV#O>b)eWXe4xZy(<}ihBEt@?w-C4tT{uV%deIVrWZBkz1v*tO`B%>Q!w(?e
zz!H`gmCut7&65tKRSRB$UFw@S!$&FBK@3JI1+|;0MzQJBqj9E~Ebh-dJQc_9$PLJX
z_)F8ME(lSdI(J1Y<@J^)Kt#J&KNl--7_LCSikXL}<eqz=%@5@L-GjX?%m$r%?(VM7
z_P9R-e(Dq$o$}G)f$SyIK!Ca94gOI~iwQt+xkU*GXmxe9rCmiiQWz)wP^bE51u_VM
zI_%<tM!wF>N)v70@AlXZT`^X98JPe8GfZe4J`&kq7RNb9SbY$PAu|I}?VcyckQ4N1
zoa{A#)|LfFkg*1k(K+8=<P83?LXQpVQXqUGzj%`Y#slJC0q5Rjcm{~PN7LZuz^OZj
zUI5}k4`dWh0B2<dbYJQf8h?>Ubd}EzncM48RlLc~tp+Y&ac=I>onq%XjW3@+ONWI}
z3ueHO6!ahy@G9}zaX_1s&j@ZkW@cjgjzcdZ-n$lG0!YD7^<Pb}SNtN5&d3JJL%;W6
zhv*S&H+SFxq!A?p-*9Qt^5q-gB5u<7Y#XVnyh_Z<VshNyUimFW>-`X#sBvY9ptbxi
ztnK~#+iN}9s-~rm0`b3-?mm-4o5k`XwledIa&oM1O~8`VH&vCC9svGo9$o~yB=_m6
zZSAgdwO0*uUu)-!8u(rJ^u*#`f2pnRi^Zw7Xs$&O8n5L~fg&FT3NXWVeBE7LG7D1x
z^%h;gqX23RDA0B9`!eGcQ&?EI=#s8K5ijUkxwSM@QPk#`8}dgj3lYNmvskhrxshkM
zaRa96E+i-N^4OsYM?m*?cx(sM;Ll}bWSC3N`!^(4I8Mog89?g5IaL$+O@U~ojjKai
zK+xXCogjVKxliAe?NEc;xcDpIsF$jl=wxBPjmPpQP)`Nx$yK7o=2sRh!5}y{GekvH
zvU74GDm9*4Z|QA^D@#5fLVmcFHN;`kn5CMjVEQ|n$66zrf;pFS3TN_e#JKWElANaA
z;}<yW-PR`erKxESt7(XRZ`5$<%9VM$B99F#nBEATlk)QN&K+AGquyIL0X%_MTZEDi
zva`3>YF7q7EWEtE-t&;Bv><IfKKt8izu7*xul?p6q!#tj!{s)9=V=Q-!<`ADb^I5*
zvbhk!#n(<QAA5a$eI1s%@v+X{h+)rB4XmWfrK7EoEBrr4XdpAJIQabeGt^kKhImQU
zqyG%B(5rtU2svO8pM%|O!&;x+Im097Zf+coMR4P^RI`;pZ~680H36Uf^7{Jvp{96U
zp}!>sa<9ZmQZ?|AT@Ilo{Wdsg3yE1VAUBo4>g9TZ#f$mm_2ZAe+LakT+KW-r@8jd<
zE*>&Dar`(vS|PLO^G8R5rq0gJ$xV~Xeh^NRf~ol44i;Qz{v#*$gwY5&3FW793jmon
zZb;%*gn`$J1lpXyG|^|*vd{&qF2qHcs$vVPw;_;uu~$RGVD!Vg3~iGwQ;Axb6fxUB
zVMwjz`B{>oLD9o)lS1z|Z{92};6QzbrNJ`Twp{d<`}Fr@2QFC83&T~KVLbp!i1z}1
z_Y$2sgM2?t>y>ok^f@LjF36Kr*`iwGjmGO=kUJ6Osp$8WabCZEeQ0pXGmhS}DTEsP
zr%Sp_hAU5qrC%Z=t1NSLa0plc_ZRKC7p^Fh{b6q(!ddHZEjV0&z<Rk73UPOm!B)^+
z)$VIfQ}M3Mkc3{lcI`vI%@;`@4Wy~(n0j_)fxRiWW!L8#VdQY%VQ_J<3^Vj)<eNHo
z-);DkPk!QRcTZ&8XVRmF1_5EQl(JvFDn=(20wO<0hkH6T9?p?NQxG<a!GEi4L8@)l
zm4;qKj(iI99jCHy#rT(S`=6)nh5XH+ISreOyWR8W&$A$eAdq{B>`K?}ugYdG=&?mD
zLQrzp6L39N3WW)EeQCIT(z#?@y-;SlzbMndbAcz{phhP<JG-T|)v+VrDkcMz>?2>l
zPP!)UXVIIdy*PnWb5FrQzVfApnu;p7Ih@fNunnXVuA_@i0yxJCq1^%^5xJVy0s_uJ
zWak&j))}SD?rfNJG^|GrW8)Nh%dxtr3}m}jJy~x&_r3`(^~_>(BV%LdM9Kg$SwTJv
zKw&}RmX5|-00zx}mQ_^Mdu<8zzK#4KC{y&;3`3x6PZS+JxPzPd{vxSor--@&NLgAI
z7=?s}YJ_bExYLo$OijfH@Ga|}{Q-^)Hbdpgp+c)KjZI9jfaIZeP?^d8zTXC)h)6w@
zM!5K4Gw~LO>!^1bfZ$bPZwV=>xo77^9v$t0R=ElwancJHt~T6;UYwV=H%vqZjj+cw
zE+zqZYiZ|mUxgihm!>xd?KJ@>RfQ@b@d8kYk`JSY6#D?TEFb3$+qs5b5q4iqgGYdi
ze~FX<9FdRVMnjqsBjoNN!Dh$-Z=v2tuEy#E7B%!`Z0+C@gd1J$K0U|aKS>@gjzHmJ
zI5Zrpk1*D$k#L>`FRun<pb_-_HVmV0PeHP3(u9PBrGec0O(7vYKcjNYC$M4d{3%>i
zaXbFU-*wR-`;Xh>_>0;OTu0n9HS)BQ-@kv)Tu}@2ItSmG*%7Jn*ob)dE(@Y=ROL}H
z-^e5KbFw6m;6eg18;b0kq!$_Fr5G>o_<U2x_LY4Ac{0#+mz4{BT(G%-3|a%EU3@d1
zoSd9qD(L#$?M;gHn3F*64V>kVefm>8Ls=z4zP<3HQ2gxKvk(&FT^5J30XED*4Eiv*
z0H8Qsz<D<9)3ejo;KU$=d~)Bra_3P}S65fD!=%i?-sn*;pq>&~5a5Dyl~W~Wzb8N2
z-I&WfL=N_HWuw?8H%%}59qvzw=7M{S;djjVxc!rWTF^yKztXAm9ft`Hk7XIuc$m`A
zUcQ_KdAP+VUp(A(yf#0Q>g(@8J|jyf8T9@0D-u#l${g>#P5t$fA{ljq(VPn~5-`G+
zQ1S8D8g?y#2ZCA6g8NVky5u2%+i~TUEnhbd4vzHA?jm!=g~2jKTonUS=Mun*%+Uh4
zrc4$KVUfj;li>4imAbE)ZC$s6C9P3p*3PQ%yam2p6#O!xarHI3>J`aA65Jdn_}|3U
zs+1-G%JCT0pWLeT+?X}h$7#j=WABJqq;8?{i^G*rWEvKYvVua0Tz3TE{+1RM0K)=M
z&w;Y7bQmrmEiXR`B*8sO!Yh7UeO74_>RRicNff7FF(xpT#e=1pwY^go&xT8$RQiXG
z2ymcK$4;W}-&MPa>nyy#|L|cNWKsoCUC{ZHf-GwxLAVS`EE4Jl?|AKove-Xv-^<m=
z?}Cc8^Y1na2}#Kuh&7bL9>suVuR{#YO~th_XxeYfLp}wg@#i)6lS2acV7S-+UK_*f
e|4q~5koNqeT5rsi<9`(F(Gm*x3+_Ge{r><VGawBB

delta 109618
zcmX_HWmuL?6MYaB6#)fQBt@hJM7ouf?vU>8?i)qAq#L9gq&uX$ySuxaZ$aPhpVuXD
z@6OK7oH^%gL?c2?8^Z5x7HIw4VfJ)GdA8t$m)rEacUfs_nT@}pT$EnJd!w;I+G4dx
zq^>Z%NWCTHa&w=zPeh)@Y*wFCoE%)hA1QF~%-#EL9e1;|;?#ZNxV_T9U~W5eGSLU;
zlHIr)eIuh7AzbIp-xxFHUXr{U7O_xU>+R(RU*kHRxa7jnel0F;JPk>*A$UYlRRwh`
z`7AEbLA$cO{N{*HjLWyz<~xI(><`vg0~w;HZ|IJ2I%KxBiGIvzEly3@a!lDmkMVyz
zOEmuZFz*@XBm5tyJsVRMFJw_sJ(((FiKd9wV~LkHtznlFMY*~ChvpgzKO$%*RvE3I
zKS!t~HtnLd97lz{$H9W;CX-pThjVp{n-)`fgG|+^+{!az<XBuSSU9wHC-}sQQWley
zlcc$*u(NaY55lffewRhahB^0$ve+b%JbB1VG-$x|jC_R%BT`NxB34L$k^ilR2+d9D
z%$atFb&cH_Pcyp|+XU{UsADy2eqqtOv;ODa2Q+;VW2>Qj3i8~L<0`HYE#?>`3-*w%
zk^G1J$L)5m4|uJ1nhS6$A59$)`c;a2N|BId=2Nz(4J5v4luA<jbWuVlrqyIu&Nz^2
zMj)b8(;1iN@yx-P_?K8-6Kvc+#)HfCc=x7+vR6;gleIkTr*L5_e|dZAiNfsq0D?>4
zRON8J(xkGyHH6%t-GxuWdQSt<;nKY``D>33HEoTRo3Tjy3lWppmL&~xyZNLCKloa9
zuO&qXIfsf{dh1-CE3p)+kKooHQ=}gCezjjJw_W8E@HT%vSxmD`L^PzBa=VAj#Qk_h
zx4Xl!tvfUraYyco3VQAkwtFQ=9{)+4@%kc5R}k`vWY_189Yt1ME^C4FWGjUal^0$i
zemQUxio<AYU6xXg_>2BSqFNd;2KCBQq;Y@FGz!A)OUn0MLUwwMza5mWjY31h_E#?h
z9zO}F>&qfQ>0uJ|an748DM!>I4|QNYpm9=}{vPN>Xd#=k?6JkNZ8lWA)5VAP2W;bj
zg##jBIWtdLIG~wlthS=utvw07UGO@x-tH5SUYwpctf2x;uO#Cdz>oi$Q6XeE6m4>S
z;r6p@2kk|{w{sEys=d~TDDooIaav>B1DE_<Q74ap8^Kbe&1;Ca`Q>3;LD;#omaq~R
zyeqRs`Y<t+y;K)jI9hS>an`RoVyq03CdqZqN;q_~L)0(EAE-I=-;Ef>%q=q3lj!l2
z<>L_IS3`Dqg_8DxIHf+n>A}~{EA4k~Or%1%v@Z01SVO8Qx4GZzL2LUr&@ko;+k>OQ
zSG{WjV+TvB^~cVBi38Krsbz0`8pw&vwI>{f{QSmfbzaiHEUvYGrFsc{?-56LbH~9;
z(+g+|^_I>UWTmxxkt#!}xwu%czT-qErPV~Jqg`OS^>FYZ@2Tk5fIzdQsBX$G`#$?q
zLxEy>?HswUT++iz2XVNfwbn+hLS=QsXE<NVUs{et_>Ni_Ui?PO2qa!zT<8j?Vuw)3
zUVsI2Yw?$NKY+pB!1y8V&r<xZx5*EJy%C4?(bFWY_J@3msbiNVeb?T?d;(rukMGUQ
z8y<XuZPv8XE{!P`zcR<eSBjgPqZey@r*!&q4ox>)YG%}YTn%CwNV+WRT-D|nk@9m4
zhD5LfX9daFr8TXkxa!3h=)FZNg>p;`edK&T*LiCCOP$RwP*P(`s<efhRdm8PxQ%Hj
zKzyn)F#2Uv2?|U8O=Ui)UH&Cyv&)P|de6-gDwz%kZ0Vh5`DE1ZUrm&1O<XcX=+YM3
zz~3GMNjP=^(cR^3<X~m7edCN&ZM#*8_qghrKwbL^>;o-5pN2M<n#hxie)V651|Gy0
zrw4t>P?}j`X;xF_jnP^6)GGRe5p5K;wb>M#INsQ}+0&>sMZqu0Wku+IYFk(5zv*@v
zVW1fFvUnskKkYdVFjbpDEx|*XFUNPy%&NNhUheg{w#P9VuViP|RYexf&YU8Uc;ud{
z`W%ex1t@F3NMNIKa=Y7T0Uw<fA3Yr`p#Sm+0#zsLzpB%CGtgK0+N1W=_bRQYfO?kX
zLNp$qo-TCKmvJ-a{hqmA-7n-VyJT6EC<og?>+;f3hxMnhT=r|}_9W`Uri9Nd61fka
zqd)(A@{%GmsY*Z+-R-AOIQ7T;#y^Elq2}dhhjaBcgg+EBFV!?q?iC9(A#x!=2by@$
zFfMavN!`MIl`0{1-iO|aVm;HyweCo-$IA&c@YcWTlwd!<&6>0SuhS@9itKXYbZ({@
z#YG|~?DEp_Jq3lB*-1mYz9gsp*fIQa$x&8JN9QmOSP(9bf&tR>=B48kZ=ZZ?Jx>AY
z;?QL6l5N<#WF0ApPw#{RQ4);^d9|ex?h4kcq*9!fTld-V_DIte{`(^LgV&fCgVv8<
zm{-esWMW2HF_@DuD}`FDW$|vW@Jhga(YQ_FviN_goER2UW$T6H0YB<+{*=0nRC9TG
zUe(AgJPQ$iyi`LdV4Kz|Pj*5}MotJ4cv1*`ZQ7oF)?ZLS+NUp`A%*c8Q*5R$FmS43
z|2kMyRa;###qNoZPh_b+5|;n{XR$GyW_qOFiK?2aF0rA6SL4N6$C3-X=4QF6ER{E9
z<Dpv!G2zb@mJ}JxZAa#1(1(|`L6R<NHHTU`WcB)1nQ`P=>Go*eUAWAjQ0)F)<=D>u
zy{hZp9mBN}uCg@1G4kAT50!L`=~w<cb~aa+1GTD~gS?`c%X`7>H$guR563Cb{Pz)k
zY;BJfG1-rvlAml6QnR_D=DZ}3Bp-AqLJtoyCuINTW-y|`aV|c6ZdN{RQC)iZYLnMr
z>kpnmIr=M6asJ2pTpGTRP!36-K7V_w=FrTA?)jmnduNQWv6>p)^x9QYcsv#C@!UmL
zNE)bLgRQDkWG`@{=ra8K4-+Lab`1u927Nc|PvpuB*=RQCuMlroClS6aB#FaqPn7TW
z=b+mV2qSW=@cf?m_WdT-?7)iFdMT!NsHBFJK0H7GAJw6r=T7=)uZ}6^!l$6Z1MUx4
zE1k(nHihGqH!lI0SiS7e(p+2ypVJ$(!$x9Z`(`nKi}6~)B~dIQTEKA>mlF5d>&LE>
zV~sI<r7PE?)BS@ddCL^<DN|RnXdab2K7fU5qx2QEXs#Ep26(v<q)UaoTne;2dn6l>
zH`ku8G6q?_jd&<3^kz28&2}X3T)F5Zc*Xtc{LxN$AcmPs)geab)?ww10rB;xzR>xl
zd0#JH$u4P`?%&<9jL!Hm9elFtGrwY$nTiPAxZ@NN>wb#hWEHaB?OmRL{BKj(4n*R+
zhrzO=@x<F(mu=ID*9qfu$90~@R{zt`3dqnzw)orgW2Eu^<+*a-vqdscVBU}txU$ns
zFAW|oYbyl?Me=N{sJta$c!gE^(l^4wJuVn(nz9l5YVqn*N_}I0pnGhb>MWMS8iNx{
z^RjhJ<xBpJSE-3vbv3!nvHsRnA_R;3FKAPT5AF15T&G7xle$IK)z>T!HqPCfm!Uxe
zY8Tv|%gzRi9*vD@@9Zkp@mj{}?!520#zdG%IMH-GX%usV%O^+e;_@LaYZMl6wj25R
zLto>Ri4J#MGkCcg|LnB#SEtJQ|E~4{twQHZn!hb0lZ=Ji?G1ib%!OHX_m7iLa#FNt
z>`M3RxO#F@yF5`oe5g8DmxlZzk=TfGe;w_Pp`*Xr+ugKF)>@hLF(*$xC>kLlA*;_Y
zV{osrK%cOu*uh-bW{FUv5mnXSY9IQ+-Q0b%m`lCDaTVhkAt&@c7K5JVQNVD+P6a}i
zUlt12^y9jeuT<dTu3heDy{O4UXv*8^1USx&bXF(#er~4X??hAEGiYE@r&-O&xO?$4
zynb?XyJ>OZ!cZSUU)amSn+ZM=itaoq$Nf-^m~mP(_~IxOV*k7ZvcDIW?H@#$gr{*n
z%`jZr-W{XeJhg3tmnIv38to_Rkq<iCe%lT-s2CW!+qIm-O-QJwEg8Tr$)KvKFFhwG
z*y4U@2kD`a72`(-;1aMBrB3F9ls{dl!<QnXb4oR?YcJPxTcQ3~QU|p&cW$wI3JPn?
zjUT=6cY4On-I7T11jZ-B+K@u<V&XzhL9s3&z8URx6|z`@L`G|SN7f$!^6hG~fWDM+
z6g}}jn`h?F0B(8V?Spa=U0aL!y`(6@tw%H!N(tBT8z#N&n`Tu1RZsT^ZhXHtG(WsV
zjw`~v6j?$`m$Gy`M0`q@>R4!pMf&yYkcuW@*Vi*F*cPvoT!}CzA(Oq;Ep!D?rY#<l
z-Y#2t>ExG4Lc7NPJgs?(j4dGFjcuzXBnS1{)M`rCe})`@!H0E~Bk1T5PfK7talXkK
z$;X8TV-{s`OYZpT&M>>JGSspsG!yoNCs+d7#*~7khPB}-;T@x`o>n`x9x6%h@03{^
zmzkWovCrX@^!AKeR&*N842RIIA*P>+mrti{HEF5*FsuWzUfwm5ujZ{$x4@DcNdaPn
zxhy-S&R8k9??;S5WqrQ~7%6d;n^jswKO{gi>+ka=D4UxD1EYO1LTUV%IWlbG!r*nD
zInOG5r&6M{5t?<|Ey#!0US$iA_26SF-=;9{*I=|bB6CK<T94guejXcB@8#(u$86iD
zHOXLGg4d9M7yph-lRz9FE8ps*soab~$Y`=R6dj-(e#KP?_lZT&w*szBTzg|A9wySp
zn6Y+u0=TLJ+pV1j#92dwh6JAYk4ssE+0!;IN#yY>jqxV4!&P`PB7W2#ixc0lconq<
zzf(cF&-15#FeFdeu|@_mzC%I!-v2<U{^<084lI3#p-8}P{Q_xWeokwX&8k8MOPb@l
z7(yeN5It$0Gd=%pxm(|y^B{6B;lA>tw8DUs+T3_hN<g0g-vWI*^@NoyH0nIZ9fX3#
z`O@H(pmJ2AP3}cg&BhLsO>M$2iMjQyp49*<3$-0}O58IqfIc-F<1S=<1yY*J`B5AZ
zx_>!m6)@6}zXGkmbbbFte}TnssI(y4Z=&1{KqvAUd)JHxkNv$JsDSts9mLD{^@_eu
z%R)7tdUKM7X;De8o50=O(~tXaYjQ&7ANAo0vazwGNdx?{M8DRA46c(1nuIUR!~Hfh
zP+n4`&t=za_Ap;=A5y@qN~2WCD6z}VtF@LzL(PoYaDE2GDwd#*G#^2Eu>jM}d^VY5
zS*0jIC1~|UNqR#LSDZktO>fL-Z`C+je?8+RMV77P)GF-rmXfZxo7UFk7FMUCOcR&>
z`uS`3(R_b@Yj;n}ShIVxac+_R;Ikto<VCdG{41OJ?`l0H_2%Rip~OT)NLNqd*D!_<
zKiP|jScRmP`c5jsXtn93yR8=D<_G#s%EL^rW->K!H3|#tKAM|nfz|%ByL$aP{9ycR
zC2iJj=4y1FlJS(&)_Wo0U6Ta>XKw2)=irlnb!qTTsDPVmF&*jH8VR#eXbCMzCqG>J
zQC<{_Vtnm~(sgn#lSxT<h`2=|*N%}z3D07K)jN@Zg6u~~);CZSlgKHW2b*6gl)fyZ
z-EjLB*k-h1R#cK{=?5amWV-ymudOXvXJvL-Rg_Cs_PoT%FIH<xzR%vZT6)Y$rh}%P
z!eI+Lo9twK4c!fG?s-g^t=EqtOE9m1xj}cRHyF2uq7S1z?=5uK9ELu#OW)E7oP!@k
zU+DzV(6kAaxfoT4*{`&DvMGe^*e?{8oZ1fRNWwH5V{+u*@|4?&;#8R^36Ce}NzR@-
zkv~b7(iMDC{`&@t(}wCT2hDYf9+F>|-lP_*bAz^;_STb6=SOx?6_v8TY2AeR0;mB$
zz+|zsZqx!{qz^sa?2Ri&i)+jipJFcS>S%g!I^(n<C~gU|mn%m?{DxlmbfS&?z2;tO
z&im=2$YkHn3IBXrx~0f(S9TYB`Nh`=h<6?TBKE(u9Cfia0`u^j^TAXyI@=@>BeIT>
z1SPA`)bLmC$jg9cdFJ4ZFdkH|AOm&CNbyeEnM>rVC-%*5EZKD&H6JNbGf?AG5rh@3
zJ_9)E*onqFX3zlKufcRgV~TSPwHa9PH}}g<+aM}Dd`d%<IfR;}{XUgd=4*$l@c_bO
zK)a;+V_1(=NPyItUwE*|B5f@e8EPR3ZhzBwIVq;3!4nIKFFwfl%bgHC7+};d_TDIC
zu>}PXvH2x@yCyHnlAI~Y0Z<rJBX1E|Z~2^=mTQm}=$gcRIa(Ikot@jv*W8Uf&7D`C
zWrKs!9wp^;J3i&H2>S3``!w@ks%$-fAUd4sFK9cDmq)*#QP$Ed&9xsQzN<_BAiekk
z?%I&69aNixZ@4o;yTgJ*5Zn&saImbs1w)g%o_4d#aq+}#&WdV4RxHlrwY493r~WUm
zl>5kgB5T?2{qit56E?XSb~Z#}?cSSF!3T-kKkv9+jQz!fkDWa2E4yp`q*S-@&ClPp
zwadK)X=N`D5x@+_D6g=vH1F6Fl^86!8^1>n1$yz1mYw&iZE{X3AXjW%w~hC|eT(ee
zlhIZ9waA%ZjygX({tB1ZN_7K{7z^uC;UfQJP>&MfQfL4-_^92&+F^_a^Fpkr(tpZ+
zlQF>Q^f3=1p=}ZF=L>2?<I0}AonNIStpbaO!{uQ8VSLNZjrA+U2kq$|N15Wc_Z6D2
z6;YaaaCdcCSHWDjYrCV2zMO12-@`zLlQ$6d?DxJMdTiKh4L(_2LM`D7{q+M_%=AZ(
z&BcyT{{X<X8GNoylF2%=Yc^Q&S2#Ymd^)HfMEO8WK1MY9IbJVxu41G3*hqcuQwv}1
zubm<zz>}d`SYyHBIiM9H9c4%jpJ1hv;3M_p0D2S}OYLXdHLe|O*gbW<5uiy@Z&ts_
zj)~rv?R~8L_1x*f_Vx6sYJD*5zKLyk)nSA{3tya~VCUG@ppCg^f8DhWw!`r{UEJxZ
zq_=>;>IZ9l0Ai{Z@QqW4DJ|{HNw2Al>%=tLv1X_=I`ZoTMo2(DK#%bMC4`x%Bkh_V
zVEMcl58-ue93@K93?Hl(5n?Nyv3fm6&|O<<I}9YyFY5Nj9ID3JilV(GRcdQ-d6jt$
z-83X~ADZuaw7?>m9Z9-N=$HJ93yZ3pmHybwK9xstsGW4*xLMg|@l@`)K_+{lUmV45
zG%;y*IjI84S}eCqD@GkBpBzA{nug*Dc3{IFW`~p_<QR9aj&U}bUn?e*MGcR&MC%}+
zB!3`ws|9q>8<vdve6T9UD#v?dq27d4!hg7vNNuuZ(ljp9;w8nTxT8H*`U`G0UO4jt
z(k!`!rWUr_Sx?p0+LdGKcUSWgeQ&G9ZT;K+cJ7ay?Qo;y>*xDO`3Bj;8B<hEWFJSP
zAeb`$!Us<AiQG3($iZY(KgnHtWo%IYhbJD)vph2Dj0!8ofM-2sq0hA%N2T0U>Qq6*
zO-%j}ZcGt%&=j(=&DavA(c&d4*fKm7Zk@!QYv~6LSa190BNz=(WN(zA?XuB9sW8K7
z?{9&R+2XH|iq+1Lx+d~Xcau9LXYTE8@2iNBjn?JqsR%oL14WxZfT;`cOP(kf7yb*y
zDLgR~XRj;Sclh+o;0}oWZ^6Hf@7KZ`*h(JH{OE7A9FNXyOTV}+yafJIFEE7{?Diam
z`oFpjz-eAEppI|%-^kWqt%Z0@b$MS$K><K>^{ELz#*%d&^`5Oa#7CG-XOWu3N^d{f
z;vhiSEzMuM_#e%PBKADQzOe68Iti#A0sJlHb(JsC(IiQ$$ox*8N5NI_w&G~%58u&3
z^ySo`_|Q-7Lkls}^Z|Z_QD|k_n9N4h1Qph5H}#_15Oa$ti#1%Y_9;U8;5m-!GJ_@2
zL-*#(3LUDZl$27OLi-swkj%T=bPp^PY^s;{3|TNM&vQWYAj1zQt2N>g?!~|?xN1PL
zEs_=8@7EYDPRC<m-{3j@w$tj##wTtpKQ?D6>u;{#flc3exV%`wAfAO@6MKzn!gk)O
zv?2%9STMkh@Zx0WLMv}nqMuucWei?cIiOT9_ha9MsZ3??`tco!`|JiEr#jkfy|StE
zX3NHTw8nK8eg@Tfc4ET6_ygUNA0IB5Tu??bz#r1o(&k+)*s--aqjIb^qiK`%4{}`6
zEnhHt|6cB}zR-7PWF@y8(c62>20DKVXwtG$%xoJ!qA>rm^HVejV+4<$*fi1T;Zol*
zXVyosJpX_-HozPz<1`21T0|Lr`&?u&&V8de)Y>^3A#Kv5Dz_ridA=l@%Y*%ndS11B
z0inpGg|A1bE>FCna=E*;i*VLaZp>hFE6>Jn*&3ITZd;%~g&@Ya><ixxw12X5OG*BC
zc0+{pk2L6PYj)2|tEpoN2++FNL)%1R5*8A85?Nl(D0f@DJL{0QUXLc=3@<MiD{U)f
z4C%8+vIa6OwCpt;<7A&Z;03m|Exve}D3g)@fiA&K2R;sq)svd7B0)xWQ&4jR6^-;g
zW%E!a3pi%!OsKEi%;q-H-jkh|AWi>n)g5%l#Acn4HHsse>e%PulO+|klFX1|9lpTl
zJy~r|uvi&;iNOtKS<sR&uwKt;n~3V*nYi1maf6h6(d|G@^fE_;ubNlsNLkIv)nZ}$
zof0#ru2j)KXh8P)-#lF0IMnUm<_bOhVAtz1PMIPPNFpIDrtf+~md_N`?H|Il>2syK
zwe%N`^I~qKq$D_bjVVF!(xs#){&a4hvpz13Or!kpkv94EoQ%vj2&f6r(&KbAtvz(!
z-Syda;(zhHdBh_Qa49hp4I()W4YpCX_(r+Tdi5`AvWxUgh=L;5oKi{eIjotNw^Wga
z_QR(fiG%?eCtM@`vjbL7->kjYdM2y{e$IKZBH~}#I^VKz;OQt}y%!c1ADTNUy&I~m
z0{@CYjqpO^=*Rct)sJf6H(Bk12D3M#M`+#M4-J<(TDO&6&y=F1Hb=>4AIjt&Tf>0f
zBf|uZ_+c}cUk87a_bbeyc>J0_<mHM+I`se)S5#m(X-w1O+S)aZo4%~0siQC#ahdqm
zdAisGAULx{#noEmK8vCfZLyYP-;Dt*fCc(DZ&RreUhrRxQPKgamg-y`(y-DE|4RZ(
z6;Li{g2aIGYXf4;41U@v3$%!imvikZ8t>%J=Wvw0lzGzv03g_4P||nOPT{7;`p-Lx
zW@?pitMdy^6C1HkRbv>YApqTE>@W!3*m2s-be^cpZi(Z8&<0WBB_xx)BG5p*LM43y
z=xW$%|702`e(ly&&jd%#8+-~J7~tnN6{|md%ujy95;+(`4rNakhKD*F_jlyI5L9-4
zzH`2P{ud~Td;FUj=HZGcK6skNJMt=jh8w2r+Qr((J-TvFpe`(D;!UH#U+aHfe|_z}
zIuQN(WG@X6pMj!89Il@qpm1jK>xX(L#Io!S_$w(g7hunPGg%KMrXY(bN`HpQ$P(`0
zG79bH#@_D8)0Ja&YL<eueM!C-6XtHcoh<8r;n&>z^E1xGLCZHTMWzbWooOoY>wEL~
z1;JDU{T*ncIx8&x6hr!U3m01N$EseqJT&6#tLmYRiL&I%xt_^m;IUf+(|r9{E&DtW
zT`VJyF%71jg1_NpTK|l@OB^Lh=pugdZl>cdX0|=X)>wA_0(R-lrL(^~c|g0qKUf=8
z2?G<JVztA-N{a9yq=kS<gH<C}>lI1voRQLQLI<eyL9!Ps*XKXJZzc=0kbC?8;+TQW
z9#Io?Tt!_IuvJR-{P`A9IPvAtOqy(t79G)nZeHA=%^*~~ceOkO^%1}^RvY0qorQ9|
z5e8IO<M?Y>6}$GJQcLaGg6QqGG%Q#iR+S*!yc${P-%Nn$Y~XymwsiZq^mx0GnrwO%
z78h4&%)1@^-QA@tc<%;*3ZefJ@aJNu5N_lNb&5QNQnM4+gvZKM|7k_6Jq5zY#Vr_Z
zstn|KykbxP0J{=lXj<vuAucP;Zw{Wb*P9B=m`40+Df6xjRf6h`aA0(yvD>XPZ;8~U
zbv*5~g&|+>GM4L!hei=Ecgl2M9z6$S9iEVFrbPZp<*{(|uH23Xz@3^mHUWtPVN$Ge
z;*ZX+q$LPMdn?Qc;&ykeTIL{J*3J<us8rQZyW^*g|NL;(j#8Q!7Dk`*A@MESBOoC7
zn(c%D4C!RbNq47(^It_lXywbgdBesM4Rl_a^NoFa<E}O8HCfp7rP*06a(1S$Pbw?l
zMzi0S8tl4mHz%-=Dg$2;1<qL$Dl_AjdthMw%5lL)>uzIg2?ZSNbOG)SbZ+en;;Aow
ztX~t(DRI7b*Q6wU7C&7(lGWW@mQN@{XzSOL+&fMe1)Q%nMEx^yvZ%6ISI~!X|KJ_v
z=FLMEa`NGB`xAc!SM-;!4u%VsV62Ce`q@fR(wWUKoze!<4;Upr(tO>$h!6F?>o@5g
zceoSak3E+!J#BLrTP#2#N}E~@IBK8kvu8B!x-h51|7PXkFJ<9&PEA9v-7z^<WYizA
zNX|F*p~It)#kf|BOV5mb<h!%Z!Ucs1J#uMRylVj8Rb_XQM8@fP3Mr^c(U-oAB~F8p
zfVm+ir=AbX_Z9)i`PtfeI-Gz*lru#a`g++P!ff?rsv7j{=3^9&U!Y;3)}Q37Y14G?
zT*uT;4k|Q<p)Z-AtqRb0AI?eCJ&7?g8e@l$l__ne9ZpW*Jv{kE{!H0?IFb*m#jI2$
zd$Gkwbqli@6iNbP<5kYE@4Bni&wuLJp3A7++lZiki6~2ZRLLxeg`s1+G3`WZzTsLM
zpz=y*t!rf<i7nKvAllyE>k9+oKN~anwhfh|<rivT^n@qS*r!?G_hYrQ!19R8k>4NK
z$eS)UI~B#`;&P7L!vx4&?YLbU5Od!=xj0nUMjsi3r|JqBk+kv?5iEChEo%Y7paxnZ
zv@bo*3WOYLRcKAk=Uc-O?X=cTZmS_0RP(2;*hGD$@1aHIdHRsMRu9SgJD}RZ&WjBi
z0`i1grslRCUYfAWml$~ht=E1;S6cxkU>L=MP8Z4HyC`)9A49dpUpnDM|Lomq+kV1k
zx50G*_YUwx3J1>VgAW-FhF)klL(T7F=|jWuaJz$q6-F^_V@@Lml3QYGn3@5db{$(m
zttSSZml`N+C+OP%6(WM(oKf`jWInn}VeFY{NN+3oVzOd6Q}d^s<ph};*6kUe$5-d7
zmf>4?*W{o@o`L+e)f#L7w1L}Ji|ZHfCqxuvr^^_iSv<7dYP8^b2+3rUI-gG30n`CX
zoNNZK&CF*-B>9Gf!j->sktf1;N7R3NXiN#)a?oEIOc$_w>eoeGY>iJ!;2+fA&pc9f
zxR>6;FpyfXxb(`4pC5=pv57&bid$QIrsh=}%9SscRE?T8_ZA}9oUs9a3#jPes6a1|
zEE?zt2M2H&Ibn9V+%?K^m%Hn<mfoBIZavmp3(+qw8=sh)Ta{Zok@_r79a(NC$4uW?
z*P1L){8jnp%u`e_PXhOE#Y;4HlE1^N&+I$M-%mu<^^R7xm;3rsz*?!u>MQ)|Y)!;6
zq<}CmKes5iV8JA>pkbi)*-RXIo~oUio_^Urd-JwffCrwK7?Y7nnp9M{s?-7zF*8Q+
z78)i}rbovQ|7{Z~OcUcUv7e!y#XdHn<}Wk}0dJzJr7eS#B3-!~f$=hDqBZ|FV)DwO
zSmux#_(b9PQN%|_68MSItSwX?x`M)j#!oq^Z^-l8%lGqEOriYZ9COOQd-(KU)%Mih
z-i9SBBGv-|q543RkXKCcr(0flnzF;_aB>~u9P8EETgsM}8FLB>mgtVXJwp!Xd*Dc4
z3dOY+a%Gm_<)>qTmL$Ah{u~!gP-LC%Jx<kOcc}RmBTRU5Lt57E?nV|1+|K<#{{_PP
zI(3;O0_?45?gO4a7E#Sz=GzV|M`0x$UK*bBt0jx33>z<?xic)4Ek5NWr1)v{>qgJ9
zv^9n+DtRQkSLA8Unky?bOB*Ju*X>y7;bFtpYl)aow@(93N^8*FT7ySsW@6WeFBI-r
z29^I`34r47LnZ;|$-HszmM2y0mq;jE-&&r$1u|d2#YyJyd8T@OT704DmOVO8g4WH5
z-C}T-CDsjm+*b4K>=<0-xpUbCa^h5L*h7)EJ2L_B5}>?hy16$+^uD3BF$hG;R@5;}
zA~v_g<g-*?u+tSBW32%`p&OI^k|%R<AQ;b1A$z;|jis+f#eeTB(LeH%^$iH%K|C%>
zUA~ISFEDMJKSp;0MKp<oDmLx>d|<L6>0xGeRa11y?!@#}`k*^7diW!2yh@}lB(<rM
zH~IqD8I1F3v(Bw`G(gDL*=SEUc=HGY=FtbhyxS>WBTG5^kn9KoG}=AL;js%!e5BY)
z3Er*H{i8ShA6mzK1Lnl#K4rGH4o|HfS=Y86fpoDL5m|xRC_P1%J>sKS>@&^%M&K7w
zzg1%oT9#mzuS@UuSSOKV8!W7I&4ib#r&BN0trc#vzE+gf@^;XiI|w!_E?p3qCHp>H
z^DPk?kRjQQ=RgrXI?@sIIHps~xNWtK)dc^*JA$V`%6!X6>cE|hgKro+5(mV{ss;TP
zK(B8q55^;2gsB>-b*C~Fnih~kZkyGI^YbPoq!dj<6Y<(?@mZEtN=_V??jbqTvHMFw
z>y)CC8oAcihEX(#f;2RoHV~EiNjQ9)FA9ZCRpM$HW|4c&0n1NZ3wJQ<pIM&^7TDi;
zLr~Bg#wkDD<&P8<jfRFbw2x2aSR65PyuDI)5@4=>V2D|S<K^UP9U;#ouJ}{fx}Q+J
z$B6j4UReyiifz5ntowb`pheb}LYdx1{sOEBmmTZSu$^uPVu>wedHF0jh*N@>sJp}}
z$&TW1AXg8~->KR&&|=w)bkl^s%h*lfpa$exbAJXkrX=~3f?$9X_@!ET0YP$a*Yc*u
zk;aG&%rC``2zOU&=>G%28%Yf$jviaTTeHXZems*EQ%oV`<T#P<KRjG@G)hnfeX#sM
zgxSTGTjWShN~*cEq^7?1!(zeMeubKrzGceh*@(#HpLt=02o5txIvG-ZR~siA&M->t
zwPB<8ry4i_<a;4bIZfnmMbpnE>BKeH`pvNAO56r&e)^v^t`S}!MWQcH2@A9@E^fp3
zEZ;={vIqpo5D=-5sg3Eca24{3l3n14`c=|V-L!}J95(7u^f@J0-U;Z!W?PMTKX`ei
z`ScQzKhO>Do=1oaIoLA7C8%MuTKJb0(H0yWn3d!w1V&@IGMD5@lK21BNWTBe;cNJ)
z^2&kEsp^uB^bqnrEriBMIVNeQaG-9%_#>g$(%ppXFA2D~?9Yl5!&S1zDS2*JpD6R?
zpDSpP-+x#J7J|Mk#U#tGMh^?q)(-WTMtAGjdkor0U|6JEWg!}pk`ho(v&@x>327Jt
z1kKFerRw~m{@4g+H!#s)yl&6?dom|jXZQco!0Yd3c$HzA5V2CGc`hI`VYH^1r~0!I
zAyVurm2tvwEuO-{)l2qS?#iswm9MCf@qoLAKhz*wPd?Uv<^)d@*mba(@VsTxE=H^j
zEG!;c8TJ??@%5MRFUV)!#h+6u>|X@9V+V81G<=^Tfix#%pm>6K@$j5b0$shw7(o?s
zykWP&$(2N3zA15kP@9*Mo_rsM&~prIN<13+!oV&B#M78+d%r^2yuBrNVMi7xvW7~_
z;k+7$*~#KB%~W!&<pzN;Ko-spry~L*<7!dUVlpqFTS#hmI9hiuQ1*X7R5g+GF9anA
z<a7R(;K~1l0`6ESSte`GV6~~uC;ZXG1szU{bFOUM=mATsNR4kszWX+X{G^#7=)D2Z
z;p2&Wt{*;p{A1H=a7&ld7>qzDQs^J&yUpx3Sl_yiQ0sR)E{n$0O$4wd%KFjiDX2aZ
zb-W}!b8`dwRdQFuf$71TMaLg2P=6KKOM++=R`Ts<AUF|~^7fK5yjjqf1w;P|8Qx+e
z)^K+>i8EwnVvXvXK$BawAdtjzEFJa4afye!VY5wYTsw?1;^;(F=RtrGl4GLNKE1#%
z!n<nA=~9A)K2bQ4xi#NSOAf20Q)?`HtH_mT$fg)%^Ao?NytNCqQN2$FE&vGQ-1yCr
z{9!FH8melldx`F3_{O|)&l1I}PfUDGZ-C77;6bri`E61BkxufTE08%$c!Bq{)Ywf#
zp>xK^L-J>=)SO&w$&ZE(;-9ba-2L2`@I>v;{THB117Zl9@ke<Mi|Y)tFAxab0E5R6
zRWneMD1o)RA7lbO_VLd(^`j%r`~8#Jf+I0K`3G`<BkzXVj;nyx2qT#N&T9+cWp}2*
z72I3z-`V+G{5LTFwb5r*<0^Pv4Ncu)F(Fxou`}NWThVO%NLxkJPhP=$Jyi|>zBZ4S
zPVVwb*L*XR-#M2r&#_?D%BN_O2H^rjtI+zE=)`jiQc?=7VeimT?;KJe1|vdBx);F7
zpf`4NB+!(J<hy!QACD0|vDXn-4VEz^7J9dY#LR!pIsQ5=S8){048-Nc$9Ky?o~scf
zQ#b_u*u4V$`3+V&iDqLgvF3u%aP>e0Fv|eiDt{yRjMik<tOa7aW;Fe|=qtoJ;rAy)
z{8jb3Vc(8)r{UzNBkj+u$6oeJ>6FYoYI~_&=~DXR0bj)u0KFIwT*NsKYh4pb0I*GY
zn2CP3o}tgkU+>h;XI635p)X`*t{ck&IJB<jk`y0b`jLODRa4dUj`YKN$lDC`Mah_W
zJ2p-?C@-9NVj23xhHz22JcDVA?UKqyw5F8wtvLkPT4pbE-3$?*s*ktM*{@_~0wM98
zmgf9qGQu$BwEc?Sq}4s*;p(=Mx;K{I-{k{%^Z%yF*MjeV4ULnp2KDI-rX8XUvSfC@
z^lXnRd`%>$ct^bSByvv;eMLuI-w^8L%5faK7~_jMH|BQ(TmvxyB`{E!=?!NI&{fdw
zAco)=%S3<A?SYbTog(F!)pz>b8-1UuOGsoU(cWUWN8(+Y5B^%9%kl;-e(Q8On9N-6
zB>M`PBO>maTh}bGS1yqb0gF{<<6!#(c!FTEA#G#Xk{8^L^75hkkd!p1Jo9d?eP;b+
zw=3XfPzU<;W*4QF?hJ6x67BzyK%A2!VD2+*=H0Q4P5edjgv>P)qLsnIfD_9ckYSRQ
z*0o}W4G!6f$^%o}9hCB1|2M5m@wYB@R9#Pf6j83?Rx>mEnk1A`4<K-yD;pIu7Sz5G
z;G4P63drtF_Sf~m1sGFq?+2t>AR~eCL3gI;aYKbAXbj{}59R)xiFzlRva)`s;Ddks
z2#J7_&Mz@Y9jr@lZp;Ngmic3u2Eu!xtdjnN&r))Vo!w}E3=vto%Uel;`e{DUeyAIo
z_7|D%!Wd3LV{=p(E_YI-bBEl)p#%kndL|Bl8@U=w&Ov|-W2p2tLK2sZ*)oJ{?eWbs
z&&4`fR5t;ZI=Z>b;xF{LY|UUi@)8s#WTC<%R=~onET>M_Z~z>vZ*{1aQTUykMf^W6
zcXeZ0%0aJzi8%%JW3g*1VR3YzLMH|V*%^$PYhOY8<ne4QUKs@LGWNJU;&O#IZwj5S
zL(`L!6>|9qiP3EFV+f;5`|Atg;)SAGfBA16Ripu2TXnWz4hWwqa|(Pm*8$nweSkuN
zw0JOG)?%x+s`?b^(BvUIiKTZvRppI#cl!*m?to^dDS_HjPq_=)NRi&8=Y!z*Kedpb
z3Wbz>O;}tUHmid_&{zRGo@x`<=z{>KJ2&>LLSR6UoJtgiN`O4ivMXGD5Q4rGglUBe
zrVUnXoUF7meN<eu4i=6aRe-8Wffm;~yE$ZF$cS1;p31R%z8Y9{-Ad9ayof?`dqr(`
z|26f{61oqbeR!RGY0yA|1XgeQL9lHJmY=H*g3x;$Qi^V3ma~ufE7x!5XHgbY_-LHg
zPHtt^?tm3(+qmzE>Tx&w1~NM=un-)KA@HkZv`&v`uo%#J_>_6y@2No+XRAFwl-as}
zf3>i6>Lg*0I<$0?Js|V0567ZE7!Ke_kVLQA+KL#>WV)(?FlY6Q+?qcLn>(}fX`soy
za`qB_mABj8zXNFY#;B9&`r&%c0NK4iKNZAU8+%^n16M5UR>B%xdkj>&y20@mJPHEF
z_OyFy4rP{_Olg$yi8m8F)CQ9ao%ZWq%!N@lx+@~Lu_bSj-I$t{wv^9}J%x8si)E3~
zu?roB8cGSnQd8e4jtYLZMopFJwqNpkGVPLE6alBiqNXWl<7_48u==6L@tE*I6HmlM
z#lT_0O8O4VeK2%kU`9&<7b19akobMXg3$B@a5ZTU^l2Dgs)lV6yuTgn5c~zi(U-Ox
z8j-vcdlR@2rX3H2oBeSVCYDx(D;*q+!XYso5)$;qSHGy~!=&YI%$f+BLc&7ssTvyk
zeyhT}9`N?|b}2adep(8z)7`(5b0UWRAdOdPZLhNjAQUj_lqCN=sC<c9aONb2UO|5{
z*ZV4d$|??ca@7i#e*+hdcFEBylOAF2^z{+~$&*0*jmM{UAAyfuXmuuBD^HjYlK%E^
zeo{r2#%4(07wM_oXc}?PX>r{wvRPV?L`0bp!<^pQslA=~sgK)D8wJ*_u6uLra6bm6
zq5e;>2T@*yhJ}>4r6OBRltyjMCw_5UNe`5ztPHBj2JepWkU;+6$H64;hp6?P8N>XM
z$YU4w`$Dmm*}wTucRe~aS96<&9O%}DOd!?0R0`V@=RnGZnheXTk|Ix_)CC8dhq58E
zC);dP<geQgO*`F=-7ui$#>==)H-FNQ-#!H`7yB!9V55(RX@23t={!~XidZoP?UN^n
z;<-B(hT7UPDT?#g;VNP-fq~+&=LDYV&JV9VCim=}o!QUM&R}x#ko<5)&rigc@86J+
zcphzxj98O$dSXfx3$NeS_PzKZ1+hApQXOEn15@t@)8-l$dK<W>zCycL%)!9P2QH5(
z?dH85%A2KJi|T`cCqVtEZyxdhF2%`m@N`0AZp`>;gZ4vUL217&1#QbYHnBRAb=1qb
z@_yxk&nqE3yr#k&h3Wiq#NOtyKrT@>P6itW@EAw4e??_i<udw=T(FjS))B+{NP%Xv
z`d4MHBX}b~yMFnOYnsVDzZ_@AaID!|$zfZ(*k&2H#U!0Behq~E2G)69A4IYj@u3(Z
zv+LooQ`O7h;_1l;wkv!2auHFeH4tfE8oe~793;+oynznvC;fR!(aZyrNHw-zE46aK
zvz2-hVpa5d?+;FNB@CS1>)ml&Ol!xO#zRY9=tDg4Iyp$)^;_<)TOAja7c#?Hmcrh3
zOE9}6tR#nGV8}RHdGl?|SvnRO4o%f_jEH9sg|X1TCyUCXH{=E2IzGHq71G|4pSzXZ
z=Q*qh0MD<TE5|Rvt6i=E)X7^y5cVE#JV3;l@<lvf(p*oVYlewd+j#@r=3{p0xlx2a
z05+fQ<;6zGG54P=QAEheI6^WCHY)a_Up)j;w3|^l0wbekT|;Fio5~6r8v7H!j}_Wt
z09oRNw;&L*b9t^HQqoSi*}+h9IvEWu#<9`go5gD|>`OCjU_T>6Pa6nD1$uG#J(*ce
zKs}CF4AS$6N0b~ofP0|Jp$!+e#VgAwW0an7(YORa_TplbARtV2kYvbiv&YQjU-Nz;
zA*R@QJJtag#F~<9*OWh?EpJD>G!lsQXfJKaKN^_JnRyb1exCd`*H>J#!^W~s9OLQ<
zmX&K=G3dBqp5Irh^b!`<`l3O_ORFcfX158*#WLBk%%L%IS!w2LXL_9_6Ijt}>Q{g=
z4fVG<+7@2OU1;m<TG|kf&Z?{uKxD*gz8=u|{oBL^m}4KmIxf&M23q?y<es={j3p@%
z(y^9fF?Man0Pv%!3Ellrfv)cy!$FW!boe-Zs*cS|){CbpO=m!th;mUO3I+!E?bBpI
zkF$+CrTxBwpddZ8%f5|_eDZsKe+wCT?$piHaHBYxr&g96$3#_=Gn_d5-UsMtC#03^
zPUdBOqZ#}L{|EGpQ>rya#hK^xpA<b&u<c=TaB|Hyv(eb6b7t1DPiwX!K*uS0fZNhL
zij8-ZVL!3~9D0<S3iTij-rF8WPB?2hZ~bY2b*>JI?AY<r8Tf|#(SacYH8$qEhh&p%
z!F1k7bBi%F?C-#YD(oQC&k_JkIpyfPG#kh`bI=uWe0t5jqu<-R&tdY&F`pN@Ma^uR
zxDz?)O?AhDV@?mOCY=y8mip2CGLYPu+LC$Pp5c{v0ISiUJ{NfB6)-bvGj_6t0e=2B
z!y+xfN|{3r6>~?_NxF;c?a4eU=6k!Micd~ouMka{=}#hoK50qFvzW|-idPYx-B<u|
z%!4O1t>a&J`g@D#v&Id^{o+T$;I)5AY?O-D06wLRZif$O9``sKR+mOHnFa5#cKSbF
zo7*j$O==i+pv#sByq#yroGJA0N`LK^jKFhpk&;lRexAU`&_Cp;YMmU$xp-*g<^5z%
zT@kx$a_Yv44R{ul64W5Zh{*nYR0Wz#2xx`)Kz>1WbZ7g6FN8_^X-DzI$5p30VQ6jl
z5XX1C?#m7<kRN}W8w<z{^2|#NWIFCjOK(0#hO28ZVdPDfIjt>=1LH-^G;=_*k4G8$
z@A)tcF2C2nL(9K0S7?$;d!ox0ll~KF6B#*!21(h)qtYq?(#Z+~5*V}d^brH_)A0K;
z+IEz^j6=Ew4$Vhve%~sLoyJT=TS<W>`bc5w>{n9d9{4M#zu5p>efyUrV}16IDu2?}
z+V-F>Rk?#~sV_oaLv81?NV_!fQUX2WV&o`2SrcYBl}V3s>nyxI$Jh{aLCBSXlHx^d
zV(;W1WUCh_jrwzP!x6qBA{i2K7d!y3i5;KIgo2+34wQGyOy$FvRo;ouB13$e2hUX|
zgv;zN7Qg_fW=saw-qz!p!pMP!My><iHO}`Zw@@gAPTC3^R|RaKvIq*Y$I%9!p%I(D
zoqww)-uEs!O~X+0M>JH;1~lc3LGLMujFbWlen`M(;!BZv%wB%x8?-S^gL(_x{(E31
zj?W+x%aEaN@t|&WG7faVm<sdP3>W@46=BZRfr>*!-F|kk{(?f2a_b~DBNKIrCA&=_
z;r7^-&z1|tm3!whO@1peVX=}+isxkSDHDE|VL*a>ClksH`KIZ+-;^0<ESQRIIBYb2
z$qWF$Q#bo`v66tB)^aFLqipq!N_u6_mpg9g<KK9*))nmUGdfa+q%Ud$hiZOlLF2$!
zBp?p+KjbKCNNVVQ0p?-1ay{wXERwwU9`WHL6eFAODcYN*7FdsEWJ(natJxt@`MIr9
zZ_n=TIXT69kIU`Q!@}m(jr{{$aki9VsZSwt=Yc#E2D3__PcSG~V&&~{kb3)rISKF{
z;y?uuF!BT=r9TqO)Nx%ER?paUdx5r34Lq735JZKKYF6GG>^N`prjgvbO?0Jbz1Cjq
z`aT8ukW?6eqD<Lc3@J?;jYUMm4_)23Rb2SqmTi2pzi$>A;Ym*C3E5Ar=o~mjrjC<s
zH3$eE+HBUv>@D0vb515B3W`=zk~cKltbo86sVo7nVax{-;Y@9+jy@GU>=prL)%Xzy
zf2@!5oN(z-v}@1$7|So2jIy6m{vzUHSyHR$ZX6JmqPn=)2c2L^ljwoz!H~Tgc2LN+
zm)oa)4_dn1G{PC|2I!!l)apn{{OmfLinrM74OrA6oTTptC<E9bQ_N1h(ucJ0xM&tS
z;sQN_C6Q#6*hBAxl(*^{fVOU$JU%A3BcE?}*lsqE3I~sAZarApqs7sSo!JxqkRI^W
zZD{cB{ebeAixKhy)xM%J<FmyjR#^g-dmC5E-#2=v{HD1l>ozjDj(^1i+&94GSXFT7
z0DR!Yux9NNT0@p+(e@M{=nGRmQ27CBx-oX9t=T2XklEf>me+4~fgg>Uw(KC|Fv-Dk
z#pejii@nh8<(0}nCDaVaE-}lLi15PiWd=UW5&E=^Zri*Hil!a7X)i*<p9{)MeZUp*
zr?tql>97J1#VSg4y8{mr?l%$7YwP`4h_-naBUVzIouWYi{F{ySf6(9w(Dn;af%N<d
zn2a2sqt}b-aGn|e;vk5o)1Fu|fE^l*@aD<yFjQa4tNoIJ&>Xg@z)K_`xti&Y0U9vb
zMZirmDVLF9>sz@n8~7h^q@E}fT-x;y?#Zz8t<I>--OSs=fJIz)drmwHAsg=YX|3m|
zoVtj~pUM*-`xP&y!X@pR3eArEPec@#LBQ7yw2Y~Zkt72)S;hCAB*W{T?d{q<#k$=p
zKAO<Hj?bz6g7$UHk#Vei-$bpFg7fqh1GP>`U)hFj^+99Iv^h=uZR{Ie|LRkX@Nd(8
zx!~_~E$ixH0LvD@XmG-nqMSzQ<yzTb901RInSh}|C3}V&Bs`XOPb0U*#wgY@`xUxM
z?hH3jpCxi10&kv0ih&e54zyzZ9;T{#VO@%&p>5dtvm58D&FZ69#}HQsoQae@u8Rc=
z#{*UHVlKU%iQ7o1XB&{N?H3*a&%>fb^qt{u-z}+irhz)xU^EY`zl_C4@RYdcfyDGX
zvwh25Wnqy#9@@fsVm?`7JA?{|h`=b`kQ@95`zw23)@K7T@O|*6>oc%L>_MATHgZEl
zcy)UxIGb!C)elZIJ?Ez#dV~ZeCuB-!`ML|Ti-gGe$_va<6m^Xo;Yb*t2NH6HZFW`H
za)?1u_reiL*6r8=dR(V1x=#*pUhP~t_MBKeU1#LTSCQa`t&}<n^b13duS!ylt4ltu
z+1U=eTj{p+K$XX)V><|w<D!o97pZSwdyRM@Pk&2@!&Oo!N&iMNRH!*2!md<~>0%~~
z^|3ORq2#39@N%SWlra_WfOuE?{`3yiI_-9_Yii*XH5aBQ=V6_F#L~Is>SMNt>&h*E
zEu2Bj?}z+$+Z)fy*H3RG*$j!B^wIY>kG}674#lnH-iJOvyZ>1%nlnRll3~W|yj{8z
zTb^3g`U4VkwUTy_Kx%^U0<Pp5#{p}z4ALIr)8(8LjwC6vj!Ad_R3c@hN!X*)10{hc
z2V^)aM+6=Ge&sGwZ`B$Od)f^d3_5k+p+3g1L2MQ9Pd~H{BItQ;y(w#1TzqD0JLaQN
z-IqlylbQ;}czO{NGN6<q2d^@)!Lr<y;v*&>^wQ=3(U}yes3}E5qF}o&$9z~9W2ON<
za<F-Q%^P!t<|QO@@pxmEajoZLa3#!2X}ou(aDjH~h}6yQ^<<4Dj7v#r8jp2W&rK=S
zS{QR6Ma!Zfh~UxGiB-x4xq$bh(NkoBl<z%!$Xrk!=F6Qd<O$KUgSGad)z&3L%+cu3
z>dMf_$p^Y7It_i)4Ypa*G!it07_<(!rIK-qxr=%mYjYwS?nq@=Hj(5W;W8BvzZo1H
zRD)R8NuAXD)2E#BHo$MHFu1?Vg6KX|hGiY&sEXEoeY#%OdqBVuYT$`EoWs|X;n7P6
z6{s)N@{O9$OO<b-P|CXqkrC1{N1)Z#x+j*2En#SIIWb-I@VWd5@Id-p7!&l&4+rN|
zb@$VggO7|=hA_L=USuZ32O^Ao&bU7klReM!0-EX@%Q{P#<)cOiso_J)Yy<x2Y?_eu
zOct3?|MWMxMnZl3G150x#GkC?^sQ-OG4y$3+IPn~FHig|(9TQBz2&!w-slqBCw?W3
zBm_iVHUq)wC}01um#?gQ@aH>uLoY~Q5Z*U*^!TyCr?;rubKx52RFC<Flr5H_hFthZ
z?$(Se{t=$a;x<2RvP78WM(ZdM5`UFLc$=VeX4?m8qQEbwd$A0CAsvq=_h0w@VUdEi
zvq(ti2l3Fp)L6Z=VUQ&{T9bASd~z2+R;9Op$ZwOir#$A;aoK3}{->6<=-&Q3MG8qJ
zG^}x9ImEi`Vyw6_k_o4CX6DTceOy$pAL(@z*rzq*k^BrDIJL=TE<dXg&uqp%g?NJ}
z`q@1rXf3iBBZ*!MYr~m@YkRWLSYcgo<8O^68!<9)^Sa~Qloz$%eJD|Ql`8d9m&Iv`
z?3>!$YUF6Y^<?`;ovpB@+^-y%R<WG?{ErcDE-NLR;%#P3#za1=;Zqtl&(B}!r(8Jr
z$$!tcA~0?yo^@DOMB<jS1BWJeGA0GC1F=j1q3HyzH==p_R&emkjsz(F+-zY{UkC|l
zjLC++LS9~!bbg=W+=1q+ot=>JSje|r%FuOkIOpxL*L(zAm-k$mUZTp=e3l9Fh$gRf
zqJ5Ae@h4El-Q4U%QNVk@V=Fimwe1g{z3Ij#5wL%Z9WOnlZQ4}2z2nc6-~1O%rWu-*
zGK&++V+sUusHIIXhuRR+s-?i0@9q2olYRarx5!8s#VEVZONopSmbtW1_Cox$YYb7t
zbgHwj{3M}Iyq^^C`yk<4&Wsqwi^tF6(1??TS{e;~4;HMIBqb^rYW8O;#F#2aW3n+R
zgv@j<!wwgh*NzsGn|_gyCV~^mUVK#Vf8)M0V#HJcsgt;suL<EWM-1UXa@t!{7u!_4
zBULx=7o#dNak6i8r%i~=g_VZ!&cmzj4e8Kw^a!9ok=fz!<bInnRA;B}hacLmN2kjU
zBR|>7>)y@O{KA4)0x_Yn;owki-MAa6l-<TgBJmz+d&m8hpvz74`P|dzzxRwvoaO{3
zm*AnYv2k<#HuadC_wh%RFflpv=|A%SnF$m-@eq&`2!pAUa_nng6={CdJ`XuoDla{b
zNy?w9_`T-S(>L6^Sij9=7(K@e%5ZYC;x7^o#1x6o&N8T_R9g+u6n9Bax~Tdnbrds`
zx&E5EVOJOU%v}L~BSmhuArIXiO`OGv4o8KpY{J8A?(c#d)bHGJbb2BS4|ho*tF!BA
z<w^NkcaeA@0y1hZ)0^TS7kSkTA?PU*dtV%c%VSK4gwqd`Ho%R~zPZ$}x2vEr*m|fA
zHfhte^SSdfP1}?J!QrJawkRG|&yyBPagtG+ZB~g4Y~q0Dq23#i+W7gh9K-hKPPbTW
zq{rfV6Frr6zp>t>q-m@T!#P>-6}RLY4ZlJ%g7;T^gip*-J3N@PHIqvBtpn0*-q{#+
zZjVQ-TrpKU!JggQP*9Kz4&ZyLF~0g7V?EO*_!m)3m1QJ&(reGeqMnfEtmrzqzSsRH
z37Zj*TinbnCqsb+O4M+OVr1HFiYh*hW8&NOkG#JV@tD=w|E^ICeqeu1T{NtyaaRu>
z4P{t6a-xBN-T!Qq5EuxPtFTJ%XAC=Hxo1lHQQ&_(ePvixUDqy(ibzQ}NOyM&NOwqw
z(%oGPB@}6pE&(a&?v`$(Q@XqROnlz&{L)J|?7in&bBw#jTv_K;&f9j5^LS6&+pdh(
zEIzlj1CviAPxVctWQ0`KRD1$Ut*xyfF^A4x4cLl?buK0vUmzC(M8vj>ejX7Uv8edi
zKv~+3(J>4ZF7<2A>TH?5_3od)1b$$oC5l^^V3{EAn{CJH=BQ)F+7JfSmPE@$Dlv(D
z<~m*AFi`5Sl&74`D3ZZ{yuVP>-m;nISbQz=sF~3ycb{-7>ik{uN60r5TeLPMS=Pe^
z87rHTTU%?i5Bihx$$Iv?_QsVaeC_*d&-9cO=a|ls$*~_X#=3*QDPBh>hXlvGclK9j
zz?+^eciguUu@x&jaf0a*WkgL&e4I^n4h_vSwU)DGLU~bf)H&CEsQKkz)%>hz2pW)_
zySw|(oE)JY=CSNZeV@XzIHivtNjhQ*8mSvJFi@a@XY};6BFd_#h?%IVl{b^3_l}1*
ztRx&rd9tw9cf9ZMB)V$YQNapm<=xpVf9U!Y;MixaqT5v}(T8<vHl5Kw84T3UvlRSj
zjXJxe)F8Q-mC;7mU4!VTv|!8ZU*J!eVUD6EL+qG{uSw2(QbY?II@OFSa41IGe?Mh4
zL)hP?$(3nPAcNe{4Sq%*To`=sJKAvLLjAgzda`YVR<g3c`1PC}%d3{lKZ;&jO+~QI
zs$TFY3!WyVbEZN|G%B<-w6qyNlkv{mPKt9sA3VJ3ct|7Y9vGltVk*uSRn`<1ZY4W{
z|9rTN3a`f@wO<&2*ASJ<UB8uOcK}<b3X!6Atx-@ZQ4@K6jDE8~#nc#u_G7)+@3xKf
z_|+@siOKvDJuA&sR$OVZp2mdo?T5vm^fW%q3w?HXeFp*-MbvrM*bZ44PKcqj1oauq
zXl2I^ETex^+_8?!9UfTjZf<ljFT$Gx0~;pvU6li#1}F!qlIER$_EMO=8_MhG{RKhk
zOighZ5khbqxj%Bhf<6~=b|y2VB3F9>J^MO@@YL?$;&Tm_o)xC|19aILu1YJW6<qXX
z1|#^*OAb+TxAC8ag`uB5wU#4`T2mw@4)OYHi3UYA%<-?{;dQ62{A^To4iF1KY5Hwp
zX2w6sW$*usj+z?gZGIx8zmAP{d;yiB@A4A*KCI>cG&X*c#bGhOYqhnr))Q}&Vs2zx
zd>pT4k`0PMbP`c`YF^4r>DteZrys0M)b?(4@u#)Np1+}D$nG_q7jmg`-pTda#gL|G
zG94_tTQv{UX^<3|(fuw>!J3Bw7hHFGYG-PcgIr-bTGM;<o@<sjtGpwBZ3>!a5CFRY
ztMa~%j*%w4<Ohr#gB<zT{<GrwMn{b3g!gNuHm1g~SqBL-E@9tfq^CTVG58eV>_|`1
zq_%M1ZiUif)6sc&x*tl%$yYcOsu#`<^bJ7~{dv}1@7%5(PL3mb)+)2bD;3#{Fufjz
zz-__Bg-tzxZDwvR5L%*|0i~ytL^XUq{EYqp3lFMHo67Ozh@q@@V{FZL449v+NT&mY
zk(S$0NvSPs-PDW{;zxVyg&}(r0if*u2#Zt-Sn(KlB_m(PI+{dY*=Ez*yId|NdfHOn
zls{w0uRt#+Tas@mZ#8G^z=k%=q1?>>cJ9t_V?A5>MkTKJu=U#_^fIXC`FF-wS%y}j
z*<mctr)lP-C0FAFJg<z0^y(@tdPZFrett?RoB?ZAe4TZJ3!T};bkg(=k<yUDkO>HE
zPpcNH6&&W?Y-~!Z(zDHY<O#@2+Yn(NkS}^Rs}>^4I?sY~pLSc9h;MHX)19Skg>=d?
zJVs8#6<#u14POrg)~s1q)~Or@p#}lmdfxRRZi|0#!GD_yt>LlkDz<}Vzr37$PrtZ_
z;Hz8i!@hZotcJu(_UH29+kS$p&1>A606Ms4)4OzthsYyA*U@z3cTefu(M`4az%OsO
zmqKT7Mb`T2J0td>F<x#}!XTQh=x*H)kp;aId4M<<3a0k@<Q^*7ci40b?aQKU?WuOT
zvtPWZu%0VvJEwEGIUZr6qfw$}c0K)JAD%bGxpmhqVq1#~|66COV-0rEf&20)U_rmr
z{s?8+TArahkLN}D2>{6Yu0!Vy!#QcTi}9ow<4~a^lupg~Ge^Oay`#U{Vr}||AQolk
zGw3&k9oZhEP4@Qq;Jn)!@2jk+rUg9??c*P4;MP6iy6<QRO@eu30-i(zTiYMPo3cCo
zTt;vcb{kMzqfW4=mLR+bI`t`D&llGCTpnua2>)8u*h>}({m0gB5uu!C)O)P2*))r7
z?ak&Pf0jA>U0{lH?@rPIyJffy3SIX$Wwsh&AejGx=DmTOB9t;%=>OZHdZfJ()26Cw
z4qiw~w%ew1xxw8-z|~g2t-J|;0SP~sqw{1l<;h)yG`Y^q0}{YxF~o24$N{v(_-3$x
zKh@l(Ec7_`@qMj1<9oLm-?{qr5cikz$xVr<dYC$KHTKO@EKw|-98D-~%V=Yb#c1$p
zm>y!DR|v;j(#XTszn@@|7msPAIkp?H){GtOUALCKF1pZug{zV-HGYCsq>kz5_Q5mm
z&?SnFjT@>qeHtlA-XK&0Rd(eTx#M%P*sf5Xva?@0ez3h2NBQ_8zmKd(7Y2?@RPfNM
zCq~LJVYq<>?B3Hv!5yfz-~GS)4Ld;c@DcZmAZhKm#>aSGWLhhJ@nIV@n9ubL@Md=o
z-8!8x!E^EmuGiE=OsMy5b5L84?K7fJyyS(+@gY_)86A-Bh{wWjOAuOo{x>NIeN1n6
z&wZ?AFgD2BLn%J4H4?L35d&N8*F$AyU(l;L5nXgcK>~<3cdlxq=*v;-v3{iV+Ucd8
z&>3Ho=8~0Vo(}p2iT=p?xIVQKHBRXVRC}7dm~W-ovb*!>398b!<L&;t3Ho_W(qiyk
z>Wv~c9{lU#ffYVMHnjY5Dlbr4I{QNemZ~qeZUdyx7?i2qj4ymfkB@OD8!ZDA(D)fV
zhf)pi(?Izz&TCy#4`i+IX?&GWvSV&m(qM{ha_trgjK9A;CqPklr#<=~A42*cg6ANo
zV+!B0oOn8XwO5EUX0KFR4B+tDW?G4zbr&d86T3nhHaAbK8xmOEV4l~yPCv-O0sfs=
zkpgHN&!H~%Hw4EkhdCTxc1tM3A%y8$NA%*4+F&-<HU$SeDTU*5725rust)E4T*+6K
zukajrv`u_Ir5D-o48(Zd&RT$aaEMp=3~+TiOhZ3;TLH<-MTq`R^JYDb;X5^kB)Kx<
zZmQ^T)fP1-1?D+jZ{O<TnqLsoeDFI`3JS7dw#g!)fUo+39y2pm3DDhhEj&CgpHmEd
zxMX#6N?R=-LD=fub&XSNJU-r%(-baCezo@)nAj&hw)syqrt9$;7*vbl>7STfR+#wa
zv$06^&`(!o24-OVgN1I_|Kr>i-P_)9Qa#-*)0XT2t<lcw)BgE!mYW6Rbv7wP{e3OR
zq$%a}#Y!FIO2I|;$y@)%Zhh#YTPoPyV7C8fQnJ>~3%KS3abbD91Ek#{AR)fz?Zp0+
zTE$tvK9C(0O&0XAYl<Zwb@E<UT*+s`&^ku?Is9aJ*+tU%Gn=Hqsz6Fe{sAM6a&7XI
zpU5M~H-D13%0?C}RX(O^ek;Pr7|U`vtWTu@)&VrS$M?i+1j54O4fNmF{|)%(TvG)~
z!1599Mn^#H9CvLXVL3iJei>YQ-I<-G5WM2?ei^0$&vo1vb-(Ldjul6z%8Gc;V?;tO
zn<YsndJ2V0?<if3-@OQNX0_`IVeh&|yBEUdPnv(fmDs?2cVAOTrGQF=tsvp4{bI6M
zeG9rr`#a0HHxePOS!PYeb9FUN{PR(~c-hAGMy>p@g8Ac>ot&D(;5jE{oZ|optWi@%
zMRH5NQT!$5hPqyd(r}=Qc(kn}Ky!!FZg10}h)_2yPuDS|5HCH;RNBYomFo&)BoTxO
zL*^}jrQqQQ-{_G`xDwnlF_)|&)3gHJrKy)J9ik52as)VP3#J#zXe0I$Mw**NC#Aks
ze%Ios)rhpWrwQ1zR;^mH*^K?KQ!O(fP>_!v@0`Yp$qZq{Rjt&VAFz52rLGJ#HIC!u
z@aojy3cKhDdSsthY(7wgrIPWKdeG;M2oqeFGHH2aj}>nC6_s*)iz0OxHGpc+eZECy
zjQS^v=964DRDr*HwaMRhuH{l=ym}4>E<6>{0jyph8$H0e7`R9d{+2N$#K$r@-a0#j
zM&~W?N?Q8sU^nR-#!qcgqncBhe69vp-(7oC67DPZPEla~>Rzm4#u`%xXGW6T>84B*
zxkLlS-OC&+=HLCg)B#ccugk)M)0tk@I3QVjJNBnE{9=!_E*AvFzh@*u26KWoMv`0>
zm)}<@g|EbD7P%^!TTIpJrFm&M<KUW!W#8AZy~_{XZ)8u;@-Y^JMf4?-galK1zqi6(
zecVAlPvLtcplI#)i{iAA!$netofo@<`SW#eR9HydA^KA^7?*)-H5&2yZ12Z~4AVvQ
z&W2y8ghPqx{8NA;Nsgn<@$;eI@!fM9lHK35`~HMfoJK=r-s}~prH;hXKpxE789i@p
zPMWSJ=U9K*#iX0<cw(a9ETl&{keu@JdX4HHi13RjmrKwUZ4(z6vAJPxe%~7fS3x4q
zXRZ*}q$!5^bXb6Y21(55_(+lT%Bh2kk>`{g_@{5rivQh<OwS_!V*a}oBit?kRzmZu
zW0WL+pA=)JQO!PrmiFt~@CvY?tLxAsFODT=8wyIar0=p?YFO)m;#1mVts`w2(kzYL
zS`&^R+ba~PX&Kk0RoPiw_apQ*Apgo*+)x5SW)9}u(|CItn(kFCAor-zaEb<xT9Oo@
zO^V2Um6ZiXwY7-(ycyL{B)5CmPi7{+kl)P5tY{Iv?87W&ZBuoqthEz{hPYZaQFxA@
z@|8wNTeaJblxd1x_S*7w>+QJsk`8C>$$!p^$%uwkCg39N?hSvR94v-|+@A;hG%;EF
zVDvRLP&}sdm#M+Cmc3O|{q1*;o9|gV@7SP%-6N(rOz!LLdHz;(wxrk4eEWM?q`Pf@
zZ-gXyjfUouf${q_Zaq#=spV1#?8_MiTpW;i67dpah$Qk<qY|QnvzvG%$Gl5!^<*!K
za{A|b^%2Ht*tk$&kf0PQ{l!FH@i8U)_3Jxt>A~*Y)W6SS;x~6z#(&%cR$)ASl{J~1
zZnC!M_h9_vJ+HXhLb*5bBUJwUMiu-2u8ZR2Zyeu-e!llabWG(H>-imxEczyS37UZE
zOKuiQF01?j3|u2AVR?UD8YNt9yb4EEPnbkPR0KL~_KKP9>A}Fj`e<$<f=)$s?dEP^
zZSw0_11;zo0y;X{L3V~B23gFnhoc>zJAH%|@Y%U+y6pdYy+%DcBF{4Yked@6T8o>K
z#FZ(R8Q5+7Yw|@EJI9k>8`n+E3WLj(zY(bTn9@XBTb~66paaHiAM}(?1wE0MZ)Ell
ztRqA=ZtI(Xhw~;zvPK1oi!172{zq<p=Z)lltIvrBulEQ~a-Xb=i{or>M@548<7X@O
zim_tHBs{{GzU60qNig-eX)SR(oF!|QQW(-}=(A?MYYhwKeC2wCp)q;e@Kay1OgBp0
zPShS}hm&t_S!d4x8CtruugKkiGDd3{C)CJ)W<_E-qB=^TR2sla2EFoPPNfC-^NSx7
ztK*)>XB&T^(t!q~q+CO*{tdD(M=lRdW_gE~o@`W9RjJ%!7ETe{twKC+mG?uJbcMam
zm#*wLY<Lm4?gkR?;Ap66Jq+;k-nB)X3-)_j)7w;WVusN~OWRkKt(tg2Pwh3Qr*Lu~
z_&BK3PqRP&Iw1QGj?KuEyFGur0bfW`pm})Il&}4kviY*#(c}12YXvlRfWDu<Wm34<
zMm(x5^>`3ocyNC%)<F?(VfklqJOTNfb82%(B}3JLk1m)@sdP8=tI^uRofWUc_p>xe
zd8;G(G4@?@r4L0qaS<lypx<D@mYR>E!#YwK;mrUAHUeW`;WU5{h%?I{KD;jPYH9=8
zdZe17M|xZi95_aagx5?v2CZSFb@s{Tp5_q_NSQa>tqfUldp(08>$~QzAP7mw<ePcn
z45$)0+tG{CbWf?5e7#i7g(d?&LN5I&VvVoR^li_egT7s7{Gj?}2_+vayqTC{8N>(z
z_prHo`^_D*{O_yi7R}xpgD2aQ!63{crXcF~1W9LW+3ZV7`V)ts@0Yt)@|~UKe79KI
z3dsgJ5R6a*M^N?6g_v^dO>T7@RiDeg$wuhQ1!vnLH81H8>Tb_qyc_t(@kIZ-yQ)y(
zq5p0#l1DYFFDTmolCfpP8wV|KJG^DlmF)wpCQT%u%s5}_oGst<&RCz%AUy&Yz~cZ!
zJo;oRgP5x5cT$-hDml<+nEv3d_UsEY!1A#Q?r~f<=W6DBo=*o^4O%AwfLht*zzhEv
zJj_-J6{M6;n(<5cC`=tei!T@MpPz7AEvtEvLSENOwLN*LgtKg>cjBOv%deyoXBOHP
zT?^Mx0b3$IxSoZJHN!F=YprU4j{4`uX+m<71Eww+U4uA7>xkjy3c@{yU#CEJ6oBE=
z%|d^*rtBLQAHI?WH)ki5t0b%>;cc!MX~uQ`=F>U4`Gf`AQ~;gc!|2ShVfzvhF>MS4
zV-FC+p|H63C{V{pQ&wB`(x**~jd(X$_0omnAL>q?-(t8$7T*rB7+)3tCZ{MjTfTJY
zg0f_&V)y0-E=>I%i3P6({mTtky2KWLzMAQ&;oeHUtegLlGV%A?6WWhU@6UQ>y(jf(
z;XuH{iSX8cUkABZuz%7Pg8i^`Ir)pPLCY4{GMC%M5cBA$!Qxlp{!{b2->duHCcM12
z!-cR(FV4kPC=Gjun(bCaBm-t^32-~&#e!*Q8MDteiMN0+MMdd)5ZzO0Hs<`wI|uHE
zsX<>S6n=QNnAA5PSRO-Rd%7(Q?zkLivN=dOFPFLK3nupZa<dI73|m)-h`LOEf0cAr
zl$4Cm>S;`VJWoZ&L&4i*ehI<UIq$CkUU~R<#RuvxkbxQKPx^!Tb|d96Dnc-CoE}J(
zGyUi)g<!Z8iZ{-)?z)Xx)fVw(+<a1AHuld7-G|7Me$;(Py&Y@(gC1Rfv1JaKF*k-%
z!AT%>zS9mqsIpqQfc06kX$LG5cE$}oEoGo0Gl!PYP)kpcn1J_1%k9w$ZL&MlKLPu&
z;r0KDBMGow^o86|2pK>yDtRCjJzsS{3Ah_eY1ffyG5xDejYsCAxi82Hy`$Jz-JoeE
zP*OONH^tyQ$06cHV&!I{J`bmicj<pa58^;i)-ahJHQH2IUEla)&=SYK&HG$c_Bwuw
zGYc^1ZVu;!avL&rLxWyxp^M5=ZVipsgbgX>5cE}9DPz<HcoRJ5JCe$VjDP?9M2)74
zk7?lhOAiqXgrx>C%;)TiO-^ZL?MY@iG)O%7XLi{^K%h*hx(;`kb+uUt+21durs>@{
zEPL{^OJB*H52lpF^H84)-nVu|uR~E?eXU7kPX#2)e-p<l>nOL4+7RMS56i7JSz8xg
zS$-fkLW3tRx&-O<3WwQ;!*lX}M5N2THi{->?xAQp2O&lVrnK}~?;vp4X@mlPV1;GA
zLkRS*k-n@<`9Jb&vAx<II3y%Y8MEB%!+O6%306?m1uiQp_ebHb=HH(Anm#9+k>sp{
zhO4pVLWZV0Q-nvW$ZEuYU*32qo1xEA4u52}BXVjL!)La>x!<K8cF;j^_XTCu0H7Hz
zS=ELXlb^BfOfwZ}uN?L|A&1#SeB3h2efp5o(^>%kZe!0W$yf^J1f@`Orf(`yO4FOT
zwKYqPC$P<8Pc-U)QQS$q17pT^j_njePggC91~0gxmXT%I7AX3l7GJSY{{eY0Hg+|;
ztJL(PWt)7#JD|Fhdn@Tt3eRRcUT%D%mVSjhy6|O40;yOhM5GWpxp^XFLqQIpUsw=B
zqG}yP1*g4WGafSc`DxRA{_XT(7#5aQ0tc+M9MXC!e|x;{P0&C;ZFwUaHwLcXP~gC`
z-xR}b9SN0bJ~|&8_5bT#8$r4M8HXHJV4;z{1kz9uupVtT(xJOhzZ*rXq*IaRFgD;T
z9Yd^WQgxVbRsX8(FCahv;5D>=dwb8fMtQ3Dj^fpGd=z3a%E&`8{4;Ff#I-+(cc@CB
z!}z1TK=Cq2I{ShQJM7RBw-(Dxt<VngzFRHTyF8Aur@I6|!|k=2{oNDko7QAH9CuP{
zfA*TsgsT&nhr5XvT~oinfeMr$MicD<@%eB5B?N{K><`g$N~_eX=RieR>qK)ROWt8A
zjpy_DFj-Nob(5KADGQJ``~pvKa~Of){mx3S6<v&U?RA%Qoi}d^5EJb@?vj8M#2}wa
zR^qhd`yat6Q!)J7ye+YM#elK;lQdG&&2|_jwn0F%5GP>980d9R;-UE>2LyaUD=mBl
zpL)9>@XJ@eR8x|~#psqdfDc6QaIYQMui0{5l61KcVib{t(3nhU$HLP2Rweppvd(9;
z=qY!y&nNIfFo;$=19d}!!uk^0d)$w}0cc;DaXa{#k3f;1EJd4r<M@~U<#jQ1X?@^h
zld9J>@MkA=3aSFSgT9<wDA9hQUazK!Fjt$@VE2#46r|WSav8eO(K1G4@j#shTwxt)
zTF!No@CH4w|I&XTfI?7eoiSrzr9e74)={OYPx_S8j+#4uiEt)vc41&5umw}7%xc4?
z>B^Lj1=|(X(S8zrb|yZF&2$c@(`855rfsLm+jIzm{kthByQYk;u)N6<@;9Y?%JbZ-
z7G4jgR<*bEpn2)y4(`fn{J@A@;&t7DrO1W%-S2bxSghOo9@QBx>jL0M?L{U>wqBLs
zYwYZ-80+@nuAl&Rm=3;xM1~=JjYsIYTwfaeF2KFry_jCj&iFz$iyiBhTvJ|qo#?Z5
z(DW4Z!C5Jy^}&hqz5FA|dA8K?KjroCyGA@tIgmgewHoAPl-02;IxtKUnWXQVl?M-S
z+^OZB81R|mc;Qi9_>f_kEQ4u7$)0FwYonz^_6%HJcA+%@(P|Y7Lvb<-iFKe1+fP|c
z*VOiLrB?KIF-OKnm!Xr$_5@b5Hvo!Xa$S;^_R!1?3}%N#lE}x#o*p~p07y+s;c`%q
zs$O@;J6JeA6Mhpyznz(9_|8p-0L+zU?bwSPq{?p2N5g&+zAc-L+bOM<8X6R*jQMEr
zBV65^D;pVKM1`u0wu5qvk!xicSjBFd+KS_gT7`uSJZOYaU+AnoCZ*dGuw~*OuDJ#A
zd6GqiTM1BV-|1M;f-{#iaFP^<X1*=x5xI@Iq9gqJ$2<;2(En>aZoPo?;sUQ35O0r(
z^f77I{e|Z(A_{lY^%t6(F^mf_@h%solTrKDIit=NX@7x@JtJRxSHZruvL;2$bS48S
z@$cTB=Qk*z!Sjm88j%Nrz;>__#h@4yimQivjk@(so8!vhywl%@gUVlShdzzkwHJ|}
z!jMI0u9$zOD8v~4Lvq4yL_z=Ks+HeTUCh1vlQS}C!(((ccsrvlY!(*f?S<uIs#U(H
z>8`t|qDDrv1RTCZ8y&T)fJ$02o9I9gCWa_S>mu(Be$czmMX_pYH(1!{(aHTO74@Kn
zgE`$24ld6;)9gvkfR2ve<T0saa!5SDngQou-4onB+-aRJd;IEUtb6b&oC&#3j7>=D
z8vuMWC&w_OI6S|Rkn;L^h?oA~uGBO5pNc3_Xke`dX-O|l%!km1V<tp`CmEoW*In2q
zNTI%8kXP$19A%LsUX<6&I3Q<n-%<cjFSOzF=Po_2(Lb4SIeg7theUyOzXrk@^aDxR
zXr8jcotchcVXsESMM@Hqk+IX3b}vREkQB_g`ucqm28XW+9w$~N(A+hEF{K~(>}A#J
ze4N6NH-I^?8-AB}zQjJlt>7#hq2oGTLdWasqxCysSi8znHFH_Z@=<)$>e2kAyTsJ1
z?)XY6pN}M#$d{#8m+()EbVSot1&=-GCOyC=YM;Hk`E~2_ZyyQgGXDQ>I$C8f3Orw*
z$Pw!A(v_B`CRkX@#Q5QnsSuGI0U?p+eE+bo`1!d`XwaLJK_o<mxv?;w{ZStn-rO3y
zshwcx!=GjeV)CF}B&1whlh38GSTyceI@0na6jWcJu#^w8QjkgW^(*}b*0MJHSq}4l
z;Cp<n_e#u3gZ?i|07Er|YH^@&TiFiC6D^ZJR`mD&RSF@<9@rN!TCBKWJCw2v=SZf%
zp(G0){%NGPKR*N{%=qDaxW#6Ut=3(I^mQ?Ea^|BOVjlM`Rv2Cf9&4+DlF`Y`lAr#Y
zA@qzCIVA?wOwCvOD|`3en(X%cGl_|7VY4j@$d00-gPjz1T+J}(d%Z)KFBG4`*@`{M
znzXkryMT6i7PPBA1GE6My}3R7GmK}QcV=#Y>92<cUh5JUJaTgnigxwM-cK&KLWycq
zg3A8{8T9{%|Fd+pJVPUYg%$DSI<ElTrD~kr^^6H_OVMKGpNhj>edYa{xv<_1wRq_5
zm6>#A0v35%vr<943J$A0yE_x<wZV>1tu;b$@R6J8EA>pn2$*6{^X@OH4CXpryOGa-
zNq?6HAH0P6&FkiFGn3-kmf8|#m%$X>{maHi{^5weAjG+lt`adb{IU0zX+?yZ$SuB{
z3z$^`o_oB08~s!?0<ctuWF}W`EG^^Dz2!nBW+*n+OZP|LH}T>B-oAt5fyj~}-4hc-
zOPeCKkk8v;U-S&7<^n(g>>L>LPY~2Q&EE(q>dZU{_TMGAf7#d@<=A@LV2QHPW>aD~
zNKY>t(np|888bRG8(iu0MHAg2K|w2@%h79bKoWSCE1HXPg~#L^Ye`TQ^GWy_U^Le?
z+{X#kaC#H5;d5@vPvfGLH6RDU_~v5&qs|MN$F+`fUBC#g|F6cWN~Sr({09#o&aS$b
zVETD|u7wHbduyVj`!ak_B+1c0wJhj&IY8bGFN($b*yTeau%wk$ovDc7xSy5><HhOV
z7H}pr3tiHH#burxSvsT>)tt}4m%ERidmbT0UQt;InOqz`DcGAtXGlN<u^@2d&T;fG
zaNlJ2{{%08Ciq+hDQ(?7`HD0xdOI%4t*Ob{V0hy&8`#zH$9A0nj8)<+7O-G1;<N*P
z#veNuFsd1zR^hTS`n&XUN(&v$KKPz(BO)d4!9Ox!+kt{M@l*MR1JJ@Q=919)moUZI
zoghzoFnFC++=Xo|0H4790)-A?P#4AUF!^c~8!NUxH2e#S7;WU?{PdgU@^U&tLmMR)
zwy?ApA7-e}OjFgX^}R1xRxX^)`i!gyhJ*J+Z&s`g&ooBWp0ej#4ibBN+I<-NAnRNK
z+_yNUNozOGPyd+U;V8%g{0C8>6us1aCs*5^f%>h6(u>y}uL63#9fQB&mKrjdyI;v}
zj{AYIoN-kpJIm6EQacR|<LaU@Z1<47l7M7Zl;eZ!n*1dVa4HVPi;31a)2_NaK<tEN
zbaZ(${^RA;FSQjXRWdFY<ehLpyc!ojx}+deSKo!OWHNKGuXgru=s-f}!M*jc5vg{7
zHFZ5b-1`Y{7OMhP*ZlNfI%!(Aap64htoR^S%xKnQglUr*R)l;bft&%*MI347<%f>q
zm~Tpoy247Rq8p5FfXu}-<8iDziQJuq3EfstOl}NslDI9%9Q!nFUpUNiRaCDG_;^bW
zF@3l^Z;a7;Q~E9sGCIC**HMtmG%U|XH2b5q*2F;r+-yK&kN^VOUxLM%XnCLIL%3&s
zbrF+}V`g!Drk^woF_r|9{a0lF#{pkkkOaMJ1H`4HXY{k~5-Q?cT@pK{iASf2$;e)@
zb#kb}@A-k4EZb(yk7?$a;=|RI*dU>Iqzg?Y)>kvVp8`*K_=0ZPX}`B<jI*;-?mgJ=
zfKArNM8Eo_aQ2DDpK89MY-6G0d-4>*-OI8@F6qB}#Bp)6RYj#N+9q-mjL{&{MF_ZO
zsECrz5+{h@lpa_TgOKE<(WT;xn+k^NJKpAYov?3hvz_0H`L6gm5o-ApS#cp0Xgf3^
z40#9TPG+;PTly4B^(_BD5JdNeJ6gjM1Y=~9w?QP7XbxMHLC)uFUcybst9B1t!N+d`
z3IFx?_;(vZ)o3x1noFGf>)6ajvqV*TC|yNiJkY>pL!0KILSF?3qzH&&TbQ-Et*J_0
z6Via}Ri&#sjG+D<A4N=!Fa`9FO8-Fq|MirWi*>an<2gIHR*ShBc!gR+V5^5ygne&0
z8)G*ixX9Vs4t+*NIg+g;dfxwlMp$@^{q^%GUIs9nLV)AfBi+<d<j%m;crOQ(fTaae
zvx_xjatE-iRyS=iU)k_WIoaB6N1o~w6!gw_&LKfb)Zk$ievL@pJ3Q~NHDF|q%z_F(
z2}*5k5xhjCt+wtmk(*}SK>%s2+D*PRU42e6kOBn@OJts-h!KtBk_^S6blOTOymB}=
z2DUMM@}XA15pd*SeZ^9|R1-0;^?_m{Yq|1CInRqZR?!=Xnle=jd=e?MdG8S>r1hb2
z6aj6e48;<wjeUn%3rh<?MXdm`7BN$PHK1;5G&7>2BPS(#l~%=)ph+)?{w_0Shue(~
zk4qaLnNdqpccpj!GM6TfER!3<=?O?k0_rGR$PzeKEMaLWQzYJHSMvEH|6k7%c|Sfy
zYDqysLBQvNsIRa8+0Rc(QWAQbcYDxP-xcuX%cQF0ziP{G3Kr+^;mA`^yEYK`7Aza9
zxa-OMevRo*CSF{Br~W=S*h@;n2M5D|gd|AF@||ZzPfTXzh_}ReWOBr0Cywp%<BkiK
z&GnZObUJvxQZ#En-pwSQH-CmHxdc8HtbT>_4y-s?>T+FFH5bl>_Hu5Y#19upWtr3o
z%4#zj%C5Zda-@vFtUn*_%e?I<xUM#rYYAUG0j}R;j|D}7G^Up&R~v<}y$)mDI4Z^=
zIHO_IzZRGZi?Vn_n}aE|Ar!mk)<s<l*^#w03!H0I)qQML=As}lZn{`|0v^rsMm_zv
z-;cy#)asfPDC%=g48_3qujYt(f0NL7*KI@~86K|i*~ap<Pt0{k_yuDzLo1ZAR<o_6
zC)gSu_fKQ5J*IFs$)csDx(7+F7v%7Ow!9-ycQl_A4f0Yde4X1>bbk>QJ-jV`xywoW
zawT#tS$UCy-#Q8a0Up`^j9S%DayQ$#8b~-tK1EVeGB;@HA8B+80xJ9Q&X<FerQz0s
zX$S^1rI+EsRFrHx!bz0!VTgw|13#M@OY<}JSY<$C=s8ReIhgi)e}i1I(XR<w2bmDv
zl`y`xqvN%^=aI+IzFSS6S~U6_&OE1(eR&s)4i|k_D$gb!^zMM9YE_FxXlmqB<|0EI
z6Lq|tL|EyA&KvWf$sWlDWZyuY#~3Wm>Yu<#PbYQrPBvgAe}r*ztWdX=*k_p-ZTr*v
z*|+9d+sd-JnSSL8+3n(1epc`cwX?+XvB^wLT}P2<<(iZzt)&|o6-HCQ*xmB=_0{%f
zhf={e3aSxXZcJiU^ng~f)2gv;Pq(DVZhq;CaxC>%j$TMy(CRJDr%2x{3SmFql8W;N
z1_rt6#Y8ZlK7C@h%wCA50%5lq^afm9+zN~FZ^_9I&{s!S*RJ__Ai))GFC_nm(t=*N
zRMDsMM|MP0MX4)FifG}nCSb^*M@tV(6n;%efAU8h2{N6@U0TJ)KudD^^0|$)U?<Vg
z;zRC4yjVxarn%XN>`axcsohZN>a>`+d<SBX*iity$fE|#0Dxoy1&siqsDSP^4lapG
zW`l^A1s)8t_%>nq7aknk*RSK=8fYa-9;#*fkE#Cf7>oq_CtZq$JUc7T3$~+kE(Giw
z<3XrK9UO;CNx~4i9EDaOD>qWOwg~@g=elZaQdE?df`y@ww+#)jVqeS2qEhD_HDp5G
z-)ntwJ$-%cQ&TwD*w|W?76IYmC<_Y<Q?s*^iFNxNQhBP%szDuc0odx5RBmxh`uA^Y
zcXoF}nwy1UV`Ik~asNkyKw_3#8ydgeKtQTR#AjoJS;gD-wo<0Xpng(lK`r#d+csnw
z?F`F03Rs{!JMXXTFWWrz$$uH9RTyP;aWpG>@#M)rzZi_Hqj*7-?QQ#{^(ja5TZ`T`
z?y*AcHSW~dHVa-Cw6`*h=#y0-)xy!;+iURqD=GoE-QQAh2r#;otnBZ_MJUa<FZM}^
z`5#L0aH6}4b_{>hY;a*8Zcm2%(yDFdX`|}@6{Uu;pUO0>dEYC*W2~g5iKM<!uC11F
zkw#^XR0n}Wamkut*mMb#LlC!?cHb(P!Xa`z=yQ@REyy`}6Yk|S5u=ndb3A!-lulJO
z<_sa`dm2t~9_5(>wF-(g#1;*blBXo&!v}5~h(T|yZ3!!&Tv+fB16`4TpZNEulTUNf
zvg8y3=)l6?-H!iLEB3g#@>CD51QDV=H>m5kH#U7jcWw-uHtA($Yp~E|4CIk1XSRnp
z2Wjca>r1|ROl(iS#R6CHM@h+RCMKpL{pR}u(L84Y0s=@?>p<G==)LFXi_TDd6d#+J
z3Ug<a{|U@<EEAJd9Bk2S_V7*>)f5z5;Q={ArPs^G-!L8$LX%$Q7_3a$UEu{LCVGw4
zVG_Glyi*q)+?vhW!ROzZjtJe4QLZ5lult7a3b2R_lDH)ULtVo3_euU0>|mJS_W|53
z^%XS4Q%B^Pw2~fox$>6;hzVC-V|E}wMQklMZh<32U;HD27^GQ`))|ae`Sm==V5Yoo
z$qydr4Gj)b`#sJu0STZZs4n#rHGimiCiu%q6{SO6N7l{&4WwJ9CJR*pB2$S&33fqh
zU%F-Q8D3&us_H|rZw=VvqF-jzO*KINd&&KHr?^>AavwxAOWckOULfK5S5|U3)DmPV
zOWD{ky6(@Bu(M-;mQ_)iwl`oW>^AdDpw-gSvTq=X>q$gVOvHb3+vaEnuC{!#!AYf^
z6UQpzVrlsAT>=opa%<Ei1I%NjrT-b3HG{I_3A~iFiNYlVz$&KuP!F3`J<(t+$=#HL
z0D&)yE}zh!JM%xeBtS7p{u)Zg!UreAAtOi-^o^qNhp$bcmVsi&_3!bmcbmrI{SV*1
z_sb$yvg7`#dE%{;Y{$XTsVEcueU6`TPRisW^rg$(*AFw>LSVWDW41XtD6}@AcQy#T
zIY9UR@8(2kBbf`JPjs+4Ddr!Q;2d=cSrJ~r%QW^ieU-d#a7Y2zzVz<9&}>t(RFXJS
zA7Q=9iOHsgxji>+AlL=6<F^_07bfKH9UX`^d+!S?i!9513L7;_XX6GYZJE>E4J*5*
z9T;v~-5&SPwa?|!7dRof=g$Q=6BJ1!e-syoH#G?|GD3`qMEsZLU#MRW>8z=b=eCr5
zY$v)+j){3MFY~WX_{X)!t#pD2xrXN!tuB`}6${^&k<e{N>thf+-;}k*mPif>WgegC
z93PASh@_)A*-VAK=oSMEP#t;stbv(k9&bJ(5YzAR_5>>%*=N)+k%`~&>N$c!nu;Z2
z0HoYAr(z0&vsmuw9H{<i_zoM+AcCV}aRITKq-ySrvLW2sHwp*&k?fh7nuJ8=vUGe1
zc~|c5rTSr675Sz(w;-|C?T1fA-+oi({5@$^6(O}5nV3jKK<_NEe+#Ddo!SS+o~c+{
z6NDmeZSA&%&-2-PA6que{3HPpJSXVRiMLBcq>oTvLMD1dmp4*dwSSpS*n%+FAmH=!
z`>YLDui_cUFbl05tO2`<lD{CA71lMUW*Jg4R&Q13KnMn{UV!6C2T(yot%3hraBxsq
z*o&6-_DpGoZ~8TMPvqp}*f}_Yf`eiIUZ3sArto4)NJyNXo*tf`uji!sct9mp-)S8j
z9j{J?gm8JCsMXZeHa9nie+a`axNi^+rU?}vbAtrC)pS|FNT$Tf=H}PfSgeVOiPQc0
zq%`RMO1K{-I4H=yL?JzW4dTqTNEQ<l+np-)dwemd`QXLk;^Gxn(@{OqROO~asnUd8
z=^l^OCdGBZ{qUp}@`KCwbo=j0Z^*;)@o8}tgo~vA0HR7Q;mo0DGtt=iABO6*{5%#<
z2}A>{=)j-Y7%<FvwoH}MPgJV`mH7bDEZf0xM~%3eTlWaJLukqI@A^C#5f7%CUXT_v
zeD!}O&Y<zmUJ8EEb+3h~vF?MGu=l+9dNu)`fPeZKv}4al5s>bPt*D`+W0Q8<CKA}6
zpa6o?;v5buE#1OnTy)J%g|$V<`PWV>JMw}USlDa^_ml7DYgKI0G>gN;wN?Dm2KE1%
zIa$?Or>zDX-UBk%o+Cx(;pw4T_i!Q&u*brV;gk591LmGu_n7T8aDgqG;AoU+ubh4}
z%R)$q7NoMk5smw-di_ul6B~Q&Jn{sfbg+&Xn3(B`)Ebo*m`M1XMm<p!?<6Fmlal(T
z%T2HT`k|CmRjv2N&|!vc?d-^;@H)GljjK;LdGiGY1+8vupkrZmGBw|Rj*G*FeS!Ff
zsrdFm#^qokw6>OqOg!Z8LGuG8FK=QS)Fy(4h6Z{f4EPtps;a7R3Ps|DoWu||5|SsL
zo}Ph-xNVEjgPf5Oc|bsbk%@`J(XyCYp;m|Q6WC4gK?1sDb8*L0|5lcl8*U7yz{sZZ
zC#<fnZZG)%gFtPLT@!Nz6p!UV@Z|h$hJg@L-gnH#y5WVv7cdAbGo{dQOl>4dv-hc>
z^xsMWxM^(EEM*~zN4=qm*PF<j(0am$7c=kOc&1Xh;FUGU77q_#NUinz1Ml~9D56WV
zZBh6O>A8tr{6JteZr10@MxSvmqy?}{FhiK0XQ0Uje;BDww_q_IQc*ykU3i~4cV4x=
z4m_s%@6<6NGqsTZ8hJ4=)$a4kfwzr{?(MCnaQ9k@fk#52Z*!0kcOspzG(q!pu`IZz
zz$)f?et_x2l#4#&f>bBC#)3qGKmYM?w^FGrWvvRcC%ypWu&}U<v+jSxq>JZwcStBU
zH8B}2GnPR^>(#UUd$9=RKo9q;>w<cs4<9|xccF9utf=qn_uB;e;+RbTwtT`2%PjiU
znWsv-HBpEUt|bRrnxep%{2?Vd`S+`?zP{i91nj8P)Ud`z0ilOm5a!e+6BKNk^I2=O
ze{?UvG?O}TXQy07iP;DVv7Bvk92_nL!K$wb@R3DJUfT3=vRZB;yfsET7)}j9hxEvD
za=)g}o->~v2057W=_$w=f-h)7Botbe{UWL9GPyGhS%GD_?D@h=Q!qklg?O4Un=Y4i
zW7%g+zoVpu@4l1<MeA!^{&uN-{@m*XbcB}U)Rgz~`+`;<yQ;wEHQotQ@n4{Y&QBjn
zO2=lT5eCMv?Fr-K04;+M)ZBj_7o~zzV$209vZ#V~(m;9kidG+1izglsj4cK4U#C%q
z*0@pO9ymXgaGuh37<=)kKjpJ|H|%mJL=Y5~mEK6GZQ}hYyJ04dAN85A+g5)P*Fo#h
z+4i_0z%tL^;M8eI>dZz+jaYFlCkjXe1PB8}goTBl04R^r`xXM5jK$~nlIlD>ICv$L
zYXPQZYAW7&X9AU&_~C<#ot_>Mc(K#I&h_-0H*b6yw5#|z<-uk+XuMePYepgBkB^9W
zeSCbZNDUreh#j4n*c(YEfx}^;nsrgmyDTd!yShD|55sOf^S!0zZCP0vcpXaT*Kgkp
z+5?f;!RLWSSze8cW~kD}Y&(sOjTP9vs^J|&+{k<MX*@iRr%lsX0BSWQ44aX@Jr~@0
zZGl7~ee7^DJT@xAi1orN4-*ShSx5C+qYyxA4eDbGb-M%m%U^=8DESwq0Z!FH_!mUm
zjuDTq{)m3UX=C8RXWL}9ljp*3=t}&d?+miLPh|=HNH_!L7MyQxHtiR5>O#O2OvRc*
z9rx5I*rs&e-rqpn$jHPIG=AUNBkCt$O2H~c0M->m2MMoX(+YLkQD$8Zt>ezY)=WR=
z_N(6e2))B-LRcT(fj35SflL0_o%|fp$$2TQ)q8AV5VW^`Ui-wYObO<PGsqsZ>H#3X
z$N17!fY_!*_tv_Zg0{A0!RZ+_Jq-yj8G3U!&R|E};GlhJnZaxvo7wVeu1g5hXYfeT
zg@-F_AECc+zFAE~-nz+VT3v_$#Q-VDPJkn{k&6sC{m*^?dbcBZgF_~8hDD5#Dpc$4
zZf0iog9PoGiCr@soI~D}nv`_&X36Zs2gQ6rQJ<3mu4F#fFzeYWfgQ!M^AQQ+Qm3ua
zuKW4k-f#Q!^?w>YUF)2-K36(Vb5Z#D`PsX=LWdW_!^2b)rwp{TZSCzV!g90m@i@D?
zyF}(}KDTZYMY^Cv!UHbRHZ^q_G^<LD0h%YikdV;!WbqIL&YU84_g?Lev|4DurPh_*
z&kPr&WxJ;WeY|sI1l2bpBBBeeU>tP=fK7X6hP!?&RMb!9<&Ief7^z_vOe6owEO2<Y
z-+tI4RfpBdKmSzo(0C6AJFd;1pB=J27g<u7?`HZ|$jE``hBPDIr$24d8->+=@#A_5
zZ~GQBdo;$uvYyYSx6zxpXz|_w`_};hlAB%F=d{b~){~2>6}8Ge#D+=e!F{|a;7Fj}
z@h<%RV+23zUGM&(wk)5=x5;w?qtT{z`a_|v+RJwZKw-Ntl=3goxgVu+eNw^Bu5mSj
zA<w8OyMZ-O3zYN3h(0g3OYY)*!Im2z2}qL&Lqt`alE|j$k}<v=Os>9j07Kj9i42-$
z{y%=a@i^T)>TaeYCol22^8hEO1&4+r5%N0yT^~ro4C4YDmlfOU&+zbvfkY0J`~8iL
zi|!rJIScfgiF`YPvEvfh14hx%0q}V7hMs;m60h$0^XDj0^$c1Sf7?-n3Yhny-rOSY
zJ#ejLeRy5v8#|l>5_8TO#3ATmJLJEV*iryyO#D|_()7q$^O#O^wMhcs?%&7QlO9-L
zy6AFizWUN-^ck=&mL0(HBIf31-K!djiKBI&@KcetClfvDQB-JS-cc;HzEwe&AZw=$
z9j_R?b39$;b&kBi60e6A<#mg)DS^8D?tKPPYT?zxRQ*9ob`h8F%Hr#no_@Ztua#Pn
zsK3tEU8Fi&j_+IKPfS#4SrQhGPatcXrZ|H-^b`Y|^vyT=X-}%FyHAJZF{mOKV<+xJ
zY1ylDj*Cg2=xv@u?%|kyLRS$K2lC6`n}uA715i^Uz7~ZcyTn3RZuND4)n;pzQ!~G<
z>kh>jB5WLyvjna9Ou~nKeZxtE>u5-p{7Ws%PaAzVUqjs~9>Dox4p&##FajPCEiJsJ
zrY4<wmr%!zLHp(3L2dr4wT>IPy7io2kO?zDYwI`%wXmmLe{Oyrf%xX?3Nx0+_x7iQ
zaLSD9qIX=lAauQ{?`(-cgx#+g*S8ih)X~xL@$!`NgOQQZ(ol|ULdKEDMdCLVb@e!~
zQP{jL?E;bTJC{3xaajzWfgRB`J&h|PBNN)mkSnqG%(mQQTY>3+0+$Lfe3qkBN`@!t
zT$lT2-V*yxts2SspNnkv7&)153&&*>WpNcj<Yf*HJ>7$ht`69#0unT0g2Wf68)`2+
z)@H(=m9P#|k#dCc*EYx+XP-J7J6%|byDA)cHx*d*54hEfda{<6DGJ`TPqnoC_?!u3
z61Q?3T-=c=YwejT>poNPOc~`oa&GQ;3Ypl$E@B8~xZG61z|fF|i^;=Y%hU4)?IIn@
z$_gkiXKQV3_4oI`1rO$#4DTHnU{JXlO*RE5H@P1_;GZ)xGSXh8)5>+gGh0nDC$gKj
zWrzi9RND{=3JL}cX=-V)8uwyQ%EtSKhRR*RVvmu2dj|hm(Yy!qmDd3rD&Czx7^UkE
z4Jj!$kSniq9v&ab+dTA-lgv|fxKM&{J}oWrOUbsqRgbe!i*>6T>sZCSYv<(=NXft0
zN5>^r5O?7OxsRHu$(BlSq|k-Np`N~#nZ_Icj?RdP#K5dl77lCc!(0{ZF+wIWvr}}=
zG=4Efw`U1FJls>N>5baxy!G|zSy>)2bii3ph(x8q#idT~M<@`WW1(v%DuHSvl@hAo
zQ^#C>;eSKU`Yrw^xhx`5u8na(j*KoMtx`%$*;g6)){T2_(iQaXn7&dDuAHV!*unRE
zy<_32vT(YDgi&b|+Kl0&gOLeJ_h#`w3kZJ2-m%}yD64UHUSBUa(gTIRVFX}_O3TUh
zNBjj8tNug|gj$`!i9&5Cf!$oWbT)U?Rw<QmON(fzNH}Y2r2a=OH$qKaU75)Mo?4Mk
z&q`N#t>*>f7ZhR>fGfaH2Wys_l)CI`0nw;%<nGn45x38-=eyGZ*duebj`~xa;4Pbo
zx;jfgoM2U@@hNyoZ#xd*Z|(fauf^6!B>tT59ZKr)S`WKt3)-TeFHLgpl0T%TT(8}5
zuE?se9gW8oeKAs{G{R9qo20^Zt6Aa9mt!+WUi_6wH;nT=&Uh_Z+UbFI(dyaV;mfAQ
zih#yA#%Sf=ui8OI`qXwliYXAKHC-CB^5UdUi5kPJ*&RW~b)STWVXt-r+jcP^tcr6F
zkIP{GQ~`Jq(EGeay!Qo1DDH2~lv(9zZB)K{9tB<9Y32)J@^kVeh-p`qUajJ0prfaG
z3PQgN`ScO-mAAH+vBdIJ%PvMSV|S-II`~|bFiPJR%Y<%+>`Vk)c$~deP`hq(1F1y3
z3AJZnW|Le17GLyi(ept_LM-k>3FNWA-$C^~-$=p@vcTJ_tsN_;(`Nmb4AnC=><sxh
zQJ@KvECe-8J+sx&z~Qi*psykZs7+DvbqnwY-@C8ZpCquEn*GTR)hZR_o8e$)R$O6Q
z<G|d201hW0B(!&O`ssPGPp@0|J<aDj$ox+>i%CBYfP*T_mUIm`cnXI;4|f*<?M?3?
zxrtxpp`fG@XDRnA@9|x&A0QcRtdovJ8ElLaoot5=vz3ujkp=zCB^#!2`oAzi{kwU-
zj&+rsX@_e>;){d!+atq{&8*4ceT2C9iqp6@^5%ML%*5&G58w<e@Z7F+sh(uF%lJ$|
zAA+-^Ooia}f@>$~rOldNN)m`8#$Jf=<HOo*0DS+M)0X0FP@)OtrsdA-l+?lTZZjMH
zprzTfq)YE1A2r##c)bNZ<C$=s^uPHWI~on%R1N`8!7N-(;`tbOu7#{ZoPR&PO;@Mm
zzWuX&WWtYkrRZK)=ZYrC_{W`-^9?uDAueeHZ<e+=Ow)^j4B$y1CkG9W$D*ag4rYC&
ze*D4y$?)xVaaPxU>TDET4U0v>`NY4($G)(w>0VP6K{G5wFOdfhhIA_h)rbXq$JU_7
z<KLf|=?x&eCedo(Z$~Glh$<SOQ4tH840ZJF*19FIr0|vP`dti9G6>LUp^`2yyL;Y~
z4lXvquEvROZkmNdjX^BD$}~#mTRMp0<bUnC|2$}Ql<9hEJY2DBapdZQk1&_`8l-z$
zBNRJ1X&b%mkpnT3$*MChRu=p#CBhxwaDTsdP8nZNXssL;?y&sFfEJzs$s!<}?HkUj
z;)j*a&<cR&rZ#LB8Z<Ab2KE-4|KzI|>s`8?tPfZ~RPy=n^VKddH`1W>z3Rn@O-Vhu
z4oy4;28L|;l->&SKQL2`o?I%~vhE}|Xy2%$*?}EE(~bG6^RC7}luDjtd?M>(H7@%u
zE0S*J213<_Nsj%&s_h&<cNzts%gV!>@4Ds9ZHmQSXNUp4h9eVuUZoZcFzwruYdf)e
zWAp~lm|>AfYnAVfmm{MP59Vvm7}1jrwK~+i^bv(kPDiZPNG0z0MCN*D#}~C@;`Lw6
z?@55(+-R2o?Jwj!Bp+y84svHBZ|02pu$DXGT<-CQU0xc{M<3<I@f~=wch-DG4GGem
zzV+JU71FOC<brJfs%?r+3|%4;?s1*ux28juf~PkojP4FRgX<>!<|}fGyp<7HM-(@D
zUQZndbxkJkUN>np#wMkwTU~l-1>K>T%|~_2x*jJL3q{Z2k*cg@tEy`8KFlG#%Py32
zEsm0K18rCE#*sqkV8#!eMHHniDtKWNsEf+=w(NNI)_aF>QMaaTCbO&d><|@h!y5|D
z3bIcy^oq0*e3{i1xA;0H!Z$zbx#Mir4^M_JQd`-%Z=owm<#F6N#}Q$Fnm=@mh2u^i
z6{6i7EY?v#MBKlbCT*C1eQYEh@-`QVj+(~A2?2*UrSN}~KXkF4!0=88*}M20KVQeg
zG3J)yi1{k^YOXuH{(Ug3?LUpX7|M4)S;rJSKcw~%@WrH4V|LyiBeh&yUt44O*yi8%
za4=VU(wsW^=a1px-&U*nx)@-7Ji-vIY8%~k4QFR(Tz0e9v*1*#uC6bDjz<HTE3P(e
z*^E{m0DNU-Wi#-9rSO}k(OV<N8Zc_Z1mg(Zjc=c4x;np;lM{9`guKE1fUl6;Hr&7~
zV50Y9%lEk$rRx(Xr%<yg>lcm9=j{a+-tu}ErgPyk_wU;!-JGp?-4mJA3On9PBn^1i
zmuE|Ckb)zac(t?;ix1i=OWQY>S6e_|xc0eYFi+5A;@Mu#0t&>&4i6~ABaN|Slby(h
z8jW?2?_!aV1eH~pVsoDEe@j*EJ#AJiJ?pPkz5wo0Ld7XIxRsQ<+LD7a0}pFm%fY)u
z2z_>Q*;!A<WqIj|ZdjEjFjUd37PF(F|DHri=``(*5Zp$v+q{m({faNfoEbS=Hw?^M
z*PG@wryI$5l^O1NoCBwKIKS7RD8$t{*=!HCuVO$xHOt!TI591!(Hq_UeJhO%0;QY`
zwf*f;DfL(GlSK5Vr`*gnOhT>}>P8OOZxytA$*Ti(@YxWDepOvApcr(!p7eN|Jd+jk
zjEaAEP`nQApO)D(-5PceYJn6>EWYiA^h!)VxYOl0`MtM!Ms3Sk$#qNm%Yq9Qqw!d0
zJE+j>w3cFp=p5s8kjFm#t?7XI3T!8Js~}8AGBUy4*(9sl<RSTF(Ue6UnPNREZk8aI
zQ2ac0Ud>0PMbB7IPw)0}CH#>w=H^lZe*fy>9N_(*+1U<zGaMf0yFEY$dR4tMTWuS3
z$7wSc4=`(VYAW*TWI?0z4%Wf~v~YR74(aa<r^6y6BM({En~h|&by?3;eC+*B^X$co
zs5@B!s8&Z_SsCsB(R39+Rc&7%L_$$f5Reur>68uu3F+>T?(RHFHxKC+knXxjOLup7
zcQ<^S_x|6E&ggKx-h0kId$0AY1&|)SrDk6mt-8x<&-H;+?aRwKr-e1JX_3Ew|BFF7
zOM&X^*RS9es3Amr4Gc<(CLj8vqTWtdSxJ{0_DwcId^R>VzQvHEBP*NL{BOOEFyB2q
zh|N+X>m6cW&f!z5W#-C7(}|*6hlQv7gLR+-Pz5IplZSkbr{fcG<S`{JHkf1jOojYA
zRnC%?Swtm5&SNhUOs$HNI2CfbD;(NDpFWS8kzy&@Xc+Uzn1S0vN=1=usCpXOmHXN@
zHh%8*<GuizkE0mDm$37lAA!%$K11+GHKT@t2SuXap9ROFzrohly?IF?Rl2oml81|D
zV{go7svehHkt(cMz|K@bgmvu-Mzd<`JPTy|n!Tm{PhzZy^hQF#!5pR_rWJd*@nU!0
zellEKl?m!a3yGNdLkri+P;6J6+x_#IeZ*Y+RP;EfB`WYo*}mbQ@hyH)lh8chDtv0O
zN@$TncM5;K$%agQ>p7x%=nD?^KlCpPXgT(vNMChWKK@EDt4$ZZPe&VLYvn6yH+unT
z^~yJc7<0H~1$JDk&k<W1wH5Svwx58ahb(e|KdZ10JDc8`VKuG$i1G*hfX3>#PQj#I
z^LVC`RoROX{};`mLOXxz3o+d6W-^hvTs-wXJevyM+}<p49`?a5`v~$yfWeO(98?CN
zZUN_zU!G!qI-pGeqz;^U;Cuq$(&6dp>fav-zB3_sY+<fv+n6&OQIztXU0phthfAOp
zA&w9o93Rh;%@Dr4yo9@|G@K|<m0vvvE3H7Y7Iyd$O9yEJ&H&Zb)m4=AXY+k;U*8dU
zOTYzCFfkohdzy~rofZ3RXGtbLvfoVc1&5asl9DE4d6@aiv3FMZpe0!Z;B&E7JpiNW
z-0#j#PmO2)Fe(>mzylcV&uDASp>7m)6COUkgpAB!<2_HUnD)1Ci*^Ipx2P{(tlnOo
zKo4q~2aPTBi2TlXlu=^55GUq>+bU7q<S6CHxoi7)Me-k<?;Ja95otd#2YrJceNmuc
zi(lRn#?Ib2h>HCpA|5R1QKOSit=HDn@>J;UUE8X`a}>AUgccFTkr4xr1G=`bX<8|?
z4Zu-?CTQ1u-EvaFymoh1qv6Vp2M#iGvu}rmI0Tp_#Fy2XA;W>w4y#%~HfvAYAW<sm
z0}Xi$l^;^U{#13ZLjZE^xQ@okH8|mfd$kpGXCB8kgUK%6BiztV=6v(@W6Nr6Y#m=k
zz9QJDMO7Y*?V9|wkuN7s(W_C9<tx|aWT8IT+n?16doUm!(XmV2H#ReYj#glMt>hH3
zCO6nivxPJGe{3<q-g>_|{i7_4xq~%t;PnJ2_<6bhO%eKx>u7Et5fSuVBUgX1|Bj|%
zS;Ug^z#kNS!h1ZuKl42wrCV<O6Pa}X7#r$$7I9o<`eJFF>&*VhAT^A>y`dJNGg>kr
z^4sSOKv!Wr)5D#xg{W-s9&09N43GP(joY&cZ8BIqlfiTGSUTEs7d<^acMp$T*^En}
zdXaF#zN3{6Mm9EOE~W3e^4VqJ<iuvpudlO#S<WSZb5T#G`y!D_G0*z$`b=)-8NF~d
zdL3w&GNh6@AFi{V*TqYMgL@wDy`7Cs8h}`LXR3;!lDnIOM(q@9@DUI-pZ)0wU3^Vr
z%93VS{!=6qGh2+-YIp7$J^7!%DZ#NU_-Th_KvP|Q)JiGg;1EAvm#gNXg132M!^{Sr
zKx#C?acdugs^vxEw_;@S+;gR)6UzDFIP;Q>F<mN7yH9iD<Aq=x)@xKcjHUKPTc7c6
zOobpArDXB66ia#e+I|K@|9oAh5>avb0qY&Y-;i}lMMBLxZSBfa0rF^Do>h1gRkB!Q
zoXIRqb;G_d+T<(XCL34SBJgSVq*@j`bB@tUr~Tz4kEz}$UcBB|D5%B!URcf1f#nhI
zw)Ycz?Rj9(a!xqWCy9{w)p41^m0Sy*_DM+iOthz0x8+FQ^epXb<3S%$D06-$HWKId
zSU}21`=f94_zqdT{s&`?oyfREVh~KBJJyJ9P=5@`^W=<zg|hk(e~s$xrEmw>KE(gr
zp0PddZY|S6@_cl^QoQTM7(dzObp;Zw%a09v^*KGUx!bWaOs(xbqbObF36)KQ6pr6h
z#vAam8P}PTEF{(Y6iN8MOqJHVRNotQrF|q6mJ}mnV%ov&RfU=3wD_ZS^!2@DW@fIa
zs2Cg>$?&|peu2jt0>GKbxVTR6Ku5%s&HwoFVQD3tZyhj>bvOI9lT%Z~(PBr3hl>CN
zA^q?{AX5$jV0iBV|KtrJqcNF|eoRVEju6Xy_wL=|;^OCULLQY0<2R36Uy~PN+c~rS
zrp)rnifX;<r+T-G_U`WIxe8!b4kiJYqyEuyn~$HLu5r%1*YH{(IC_~(hDpJTuRuM&
zIZ;fszP=8khL%=WTMw670AGc+w6rvumpT1Hg8zi7O^gZx;7Wl;HM)_}nKC-o)1@&5
z27=isH&XP#>x-nMdJ%L-54IaRO&ygRN8BJO=Xg*PMNq_!%(dY#Yid^B;Aj=o*$rsx
zEM!vCxEm!Qk<iq7?YKGD*O*@Dq-BdJQy|0UP?zJ1iSE=ZX*~JPQng716ww#=t{tke
zF#8&Mlp*7V)wAN{=xmwS_-_JZzBCWI8YJ=@(r^JOoo!Cj5;EGA#Rae73jzh*8~8%?
zWm9c$6R2=GAe^3AD%q$#_XO-AVDlqkWX`wlZus&+M<M%_JDVemA}y_0Xy&5YyHz>$
zLf_fvkHujSUe%P-6_Y%Y!YoT|Qr%;#@BxYx+tOCPI|u}MqRz_2-B%rNH0rSfoOmYC
zXTo<cv0f4o#KCkwWm{gJYz3CPyDpnYY>4J%wsFnXG8vts<i+JG;bwjw;jFB_o*8hj
zG$a64zkV<A2^H%tT=c&h+Tl^VNzMB7pGBP5Jmc6?{5}!%R1p2U;Y8!$;x4a2YrcS?
zByvCeA$(&w{QFts!#x-HT|4d3sn1m7up<m`^KQ)G(o#lHuNVMknaXNqaq-u&8p~R!
zytH&Uv7jG=et;`Nxes=}#<O`mHcM_N;!mcdIo5BB8ylfgCrN<(AR!^CsHy3iYg8+l
zfP1`lg!&5U?|7z1^>ry^*QHHxC*0|6+d--jg4aY0S5Zapi5o>_nh>vRrQ3TKfWlN(
z&iGNJ><_mb6O;36Oi5ZhdDAGS%m$V{{Jio#t=DW`L~r94>$MK2b5+bLf8YI0JmEb?
zbNMci`#UhSa+;(7wZbl^AKXa}w;NXGj~q>?G6PL12q){-w1>P<ZNK(v+=2t>j725H
zTl$BeAmicHu%3MTmZ&sZSaj_|>r9lZ=2XR=qG&NRF}^Nr-KK?_HZ=5GUw^R7RWU?=
zEfF@^-@v&+Bx23O{(JN({b!5~HGIuOf>?z8(w$o&so6scZq~4<Q~NwMRk@Mz7|RsN
z$1SXKcXu?bXB?VM!j$JTy#9yZoX;(#g_qYySO_CADWtlTajc+lb93CsleYfIR`G#6
z>_4XA@R~~5t18E9OfI*Ap=l;ZHsjcy9;F%YJryyOCM0-rOh%%QWN5sDi;kK4E3m}p
z`Rl9GSw&`6Dktr+kWduyuVhrjX&QmC$AhOwa9_xvw;moMZ4|!>Gkq)huYN`$-ULdC
z(qyjf5HHHU`!!{E`a7gp6QGvHN>#5G!inVf-+i{5NLKjJG~W5o;P2Jv^|Eq@@miV)
z8#y(&6IEY6G}We<=?Jh((@{`x;dvl<kQx*rX_QTm<uTb!egi;A!I0;FY8{`6)OK+e
zZpigibVT#H*CFo6WNL?)!9Cwnv|<S?x=doH3%k(T)s3Tw!B<YTXs7(pnZ;Ta>y((%
z5WZyDQ)~Us_;;+bNRyXoo|GB&SAL=dPVDWUZwckEZiZWT<PLe|E*vPiQ*1?!9|mHA
zct!H$K1Lg8ol4K$MQz~!7vF$_DZK(c)2e-rOG$>(oZmIxoCJ#HuhPyTDW#RucdPp3
z5wVd!G#$>c;?Sv5l7&1yh5|+_Y%dfCMfg+mM7uh_-AtYA*1G%d@DqQbApZalP-&au
zJqda%dL)>MZ)uSHmG=74q*@%fp%N0VNHF1T+PcGCPjlMtD)htr^!*Xr_s|k9w+L)a
zL#VoLHzlPuV3Y+{MLlmWTqBp1lZQZFu!SjfK2%&vXD7aVx6w9Q9gPq@GRg<05Cl#(
zts;l5wLPK67}`O38~$1^Gue!doS#|?PA$!oj@S79MOTOKsS<$t5O~XSbu9YSEkm~^
z@(?X>SwA{`IRC1Pt}c}dt^JM8pLMwm0Kl(4C+DPU+{o-$^gM0d=14PMykn+H-hV6%
zzs!%zb#fE|U!a6x?Qk!~a5x~1d9`nPfq`TF<ku+Xx^H*>;%u_#+x|_zfmk>>MNwU*
zaCteKTv-B-x#37rkSn%uzN<Ii4h4vRrM#{7Z6JhOva#`jI@(Gvh7B52Dq~bh{EjS(
z2{XeAm>C)VR_74_?8PVZx2|>=XI*wnGrYAOovWf+_;mm5bC1s45^KL`J^m{Ba6*K|
zdp{c2=|3?Ta<sITZLM!&BLWpU#5CR}Wxzu5OsTjl|4cX<-W^3Iy+X|nj`W_s(+CXs
zKrJnGa-CNQ`6NZgc05{qR>lcH5XqA(W|o2%r(N<No!?@#WW>FLrEvIrV4AIm;j|rA
zO-Cmb*q2qdjSj-4=1}}zU;S2~tNZ6;HsM+-$%XE`2qxJFx|I0*@sg|2_-5G=018av
z;`~gC?wc!N*!S~2+DY-KFqKJ7C2!0aKd9;5#b1Z4=9`C;YXJj-bPt8uu00pXNPP_`
zgT8F36@8-~0Bqt6USB0F{40V)5m+O6jLT8TNzqE_U8sv*6yM2NaG^pEj8VILs(o2o
z!nZUwTAj<qIW2dN?EmwhXkXuy6tz3^It;ka1M6CuFiEgzN1KVJKokx7w*WH&KWM@!
zE!5-FX=<2OPiNoC?us?*Y(vMn8$SrTvX#<mY?|Imjp0nFGP0wO*`Mw6o4L457h<jq
zqu^zZ6eLC)8XHQ>s8uh$QFS|I$FoGvNnTWZf@igW@GI77t9Wkbv$}XNc=&A(Y}ga`
zk&b0Y$W6~MA`;rWM8ZL2n#BGXB-kPwBMCm6#MKHt3~6>FQK@fegwC~eX1D-yr9Cy*
zx;8SEkVE%g_Z;w@af%Ql4Y5#2O*`9O)a-jTHPYrNZ)?kbb;xLDHNSS~?H{655y#AC
zRl4i#G&Dx(dBj3sVNt_s)*UlqH04PjP#U_O@4yWZJdL1$^q_dkJ`HHJ@mab_>^>61
z;f%{84<o0YdclJ`k{yNtEJem&Nk4WqHoQXeKh7`a+pt<l8>2`)>gXV10M%gh1;!g}
z=?9L5B{x2lc!lmSRempQ3cZaK)$J>ep#yRQBfI4#J*Bxnec;>T9M8G)ZjJ;KQAmES
zjT4G|+ys{tm-nIlq6l;aFBu#Y?j|nVRZ$>i=&lF_`{?Ls6d+25N^}J4>gpJnnMdr6
z!HpM4p>sIOk;hxOsdBb<A*rkriKz~!;NNg|x8oZtC;$)b4dNHdx)<a~NFm#!5sLm{
zC5J!JP-m*B?1+X;wr-d05C|+NG2HB_OD&|d2t690`xv__%N2dY?snce>vER{iGUwr
zteVOATfm3^R@!NP-Y$R#n||aVu^RvSUfo_SuARG7ge_)ozGv9Xvu@<~54GT59?;Pv
ziy*N-ML=?w9yvc(pRi=C*AnO-&T+NLXvH!5@x;>+326n7K(Hkttip2gSMN+?o?$p{
z(;(!6s^Ywbn}J)So7JwXYuB$$u3Yp>b&m60u8W?4yy^7x;9NNla<SG58XuJ99OGfw
zT7UU;e6A)q@CQ5zm*dA$3sXb#WJC9^WdCquUA?|*EzMkuqgy9-jQEfS2h*V{DOxGM
z=SHgyBn7<gQ5C%sto=z^%cApW*QdHYTjh{D03^IMcfP3z2>fB!POT3R$X%_Uuh$aO
zR9C!2L>zHxkJVpaWvJ2N@!TB5jeH1^iTS6IeHPs85HpQbP`F<n_X@SsYd!xZ_mhp0
zHuTj9F9kc(a{!MO?2I9b$VYBW$8T*j{VHO9MCLA=1l#xc+Kb)JBT}N<wB>mr^iQt;
z9f(wnp%l(l=2HTemX_5wIPBK%8*h(;bGE2aiaCI_$m>XNlVwGm&cdBY92u=QFOQdp
zkC`K31bT_^t$*_km1Gw-RdKP%3!{^+kz3n6&O$SVjBGo}=&HV$9-Nwu*Crg*?vqt)
zQjRb(SnB-*RsUR#WCoQIgikmvo%yDP5T^Kt-4rF6^<cH)LDH~DttUK^{OwGG)S0Qo
z=D05(8ousjiXcAzaFumM5|!fu+7g>K6moiYdAyg_28%Q?V|LmV6{fq4D7^`{H938q
zhl`V2m%7+WI?rc|{{3pDLm}bmWDi4$B(-T#88OTCvy2M@osI8WkZ6zfE#ASb-drJE
z1t~gfZ_@b_62-ZT6=A#9j##UH&Gy;^g1-FRdrd0fHBGBUBUYx0XS9DElNp5-U*b&Z
zM97<7=;7e_YBN{vxH4JLfi9F#p!Vg=r0Y8Ver58N`QUfm8Mg_$Ou+g)@*&TV!%9n~
ztC%!1AkD_Mu)okZ5J`RY8`6%ha*IHh15y7sTugShmwD0H+7*X14zeN^fQfLfk)Ig$
z;^(?J*&k>eF3HjhT*Cz#L6Y@{&(`{H=S(?{r5E-$^B8ce@*_z$L|$giweJ;h*uDto
ztc;&6cbj@8_Xt;l7TaVr5mnPGhF1Mil9s?~aSaux_UJRs&dyrf*$E!-L0{aQod<t^
z#jsDg%J_D3zQ@kaZhbJ%BMAh$0%Ri%U-du@C1I&V<&W+3{=CT`8Yb>yn<~gxK|;dl
zrL#A~SR>y>(&y7%oDfZx&zI<h{;u?bLH)^oeubuh5|&5R!HVsdx8S<Ovsu~)?j3Nu
zL#<YB7kWk6Jz@7eJqoXy1+>En5L|Wh@_XpPWQGG9=fFK|keyAEosi^)&;LMhBkHPl
zQ^t~b@RZ4Q;SJol)J?@G;));4tS7FYz~yHOe}^R3vjAxs6TXHaM*Djs?60a*5(I<c
zMSTe|4MS$@m+4oX6_p&vi*E#-h)6$CKq8rYN;eBS!)dqI3w9kL=}KhPlykqbNP;OE
z1;aF?Qul7yv2e479pB`d*xwmpUo2y6t}eA_BizJJ-jBY79FuB4zr@gEz1UL9il`h5
z4V`5CMFGy;-no+a&DK`*8TyA+L41q6k{wSJy>lWmlB&}wU9oTR121_wJdp@;p(xsN
zLoBZ6_Xg@EF4UkRd6BbIAZL4_I%megRIDj(dwd*M<1Q2^FL|Na(ENr-JE^blvQ`;8
z`e!X-Qm@3+IDO;%Wx*H}8K-pbyZ`S^e(i%TxbkdXK3fVXGM*I_6nww+Ihd<;>gw)p
z9vDCayaW>y6UvJhzcp(e;h?zqc$=+Zva<4WAl#7Bv1Mmti#^|)iOg6gQUfe@&3Xu7
z*Po)~E{U-9#b#EGs|^gM$c$j~#K?%&P>+SO3w5-g83<d`1-Ws&*SL?3#p8v8Y4mS;
zh&a0q5bt9~7WBsSsM?*pBFtyA*&fo{=9nzEJEwUe7H{})w+_()G|kYNEyE8tIF%zI
z;V?cH8PsZ5iB(aP<1*6Wb9N2wnZgcXzJ5!kt{BiljW|q0OaJEk<x{^s-p)w`kuJrL
zmsfXPrxvQNTW`QVWkpmk4GLgFQnC@-IjjaADBh=_;KNLZp+3{eDdPO}pXAO_mihg`
zlq-z(7|%nfm>4;cYB!Y)nIJdHgCETpX2=W{*Ux;Jw^Q=~%em(|Kbnt-T%~*)vOgPv
zBZyqTnj?Ph3CsuYm;zRJ&9BWDdRAK~oTHND^k+_IH9Q*}lMT&SSFi~L?Ub=uJs2}a
zYx3>bEc+j<60v-P(PM;GsKzDi%|u<5u*98q>7dfW%3MmfjsC2fXiRm4{grA*X09jR
zUDV6?l<=U&I|V*Y6N-T}+UPRJ>+A~iU-!$Q2~qzfbc4qxBfGE<jX*7yvRZu+A-sLG
zl4av|k1MI(p?1>Ij%_tH194BCNBN!~Q%lhTXX<bC?^AovnR4w=(f@TovHa<ykhQM~
zpceoYdUSNuG&7S3ggnHw>u65_-3}x?>>k%Fp}3!#Cnm5hEG&Rd++?zZ7$_&<c4w<2
z0rq3UBvDBjEtc@fto?Lr1c%$%TF5_x^=b$KlRzLd?aNm~VR9x-&C&o-L;%eI<eo^o
z=$;GmwF>m1lA2c;HTHp0B0f#w1d<9N9+DoucrHXZEk~L9a(f_Yh$JFLXWGri*4D1m
zjgv}3S-EO*fNp7x9pHBuvA4~Ut)y_<>Z-#9c`sppe!n-?(G#M2w@R#sZI64NytJH2
zSW9&e2>ihxcxSk_`i<F`J*bYmnJ?*=5|8xv(IoslTJ^w&6gpW*osF0}4U7J>tt;o_
z+lJX;!!i8a8k|o<`XnjIFgQfa$&*a|X2_Ss%A&7N^=hlbfKWDHhAO_+69W}yPR1r3
z28+R>%~x(4>wbdDqq)pf{)Yf>U@hPyz#v_j2p)?Juvs)7v3>O-NjGM8z-9b;H9?H^
zRqH=30mQG66j&5eg>Sm}&D1MJK}^AwGxkF|idZM4rJmQ8?$mxF+sKQF;~2=A#NI{G
zc=uJ#l@o3srf6NBD}agMA!logA||>lp~608%K^BDz`5Gvky;FKjmsCu*h=}V_N1(S
zNZig&1=9vR&2n<N<?ghg^Z!$|UMyGKL=pHE3Q@Zxp(t8_`>P=xd;aQE2^*uZG8GvU
zZ8GXo1h=Rd(%-KB7zbwFjOzSgwWB-b6yU0GWW7eZ79ADqYi`HLepheI2&$)JWwtSF
zv72H{Hp`QxA^Ez6S}?1C@2Yx(h}XsIB<E<9iEZtiD`PYSYrmarvB8xj)Qn^4=7t_J
z?YODw>g^V|Iz{s4r)+fcna~zC1k!X)vhg)+Esx=TGWy>M5fS{t!bouK?d_X+vJ($N
zs(-ABAK}4{Fak`1p!X>B0qqdN-ef5P5fKsGJ@9U!W00z;snKXR-~)@A<$PTi;AUF_
z-jECq4J|!tfCT+-x5g9MtcoC(M<==#KG-?@J=~acEX=ShrrKf58-dTOMI}1hph{+t
zxCRbycV2UpenskzC}aK;S7#S67Lu&T@xTB`EM@}(7Ns<98>fM=?DY6txBVY}Dhg}&
zePmE#300|unF1MPhyl~|Xioc)zYeJBVrw9IK|xq=LRT9X<n*ZS>A15jq2Fw>+tV;r
zcuZ*rm6B5~NkniuS;S#qP8AMN{GdKebr+#xb6{Zg$Cp5{KT8<d^uFyVNf0pw0Ns|s
z{J>6Ubpk<zI%|pjh9%a0c_>i3filUZC>bESbcBQj_vX(Hh5ayJ=WYwzI|mbLrO?d7
zy!#-EfkkWrZ*BYD#_4xkJ3|xsWT|Kb5}-_`I&)iJrv+8UToQhdV)@j3YaG&e%xqZD
zvaH05FjiJFSQR7N%HeG_{@K||_FgT!-3b;Kf_JLf1n7iUY8?#)E{Zbe-Cd@n<`0np
znLlP0f}QDKPC`VO$uM;{>sj;w@vT)u4lAYxKQeNk-4mld3cd0I5U2lliUnmJS=9kR
zkH&p>S?-!>roaC|`<*V3bA-HOd;%veJ<yz*$_HtA1M7>-)Ko&^$;fKpE&CRMB@fEl
zQslekpOK7gtm%H0pDg<;Ta9<vd?FeMnf+fHu9q}D75aky_6@(izz4|B)m5Rg;%|o3
z9uoyWQAhg|nU3bV*`u*ihK<V>hGxB$zlxO>YNN#+`yASlDbc>#G7J$%&-C?;^z<x+
zlefn{IgX@f6~26wS8DUQhvN~WIpgDr3y4w0kdw1dE2PQNX0w?xR+e3Y8jp8tAvVY7
zDJ+RoT*jFlXKHG+0;<de8aZHk(H<g>2?P^i7eWd8W*IY30EOk9W32o0Cj%qHFHlW^
zqDh>f^hP^eZv`$#Pe*D0a$6<Nj*B%VVDeH9qwXJwxkOwZt%l|7JqHg#<F9Y&=;&Qt
zJuUss_{wf>K)D&qRfH`hpUSTo8p1F-M9fWB)Va}uIsnODos+mscK|9LK#>0nZ+@DF
zKJ4su0e19<0}bCd7z1I49**jQEf3)VF2}2Z!nzm$C<kkE5Zr0-94p#?6d;qSRZ}!o
z&R5&_{5gD_otJ3(Z91u1%adrEhF3o)TLfhNbrT4hRAkB}jOR)Myjc6PL5s~13&H|x
z9eYX|yM1JQNH|OWoHHxh_sfSgFrEAif465&`FP|aB3|oiLBw#)nJ69?cevr}ifzXs
z6B%ZM<62`(@2P#Ds3eIr9kkzWWb4{wq<M8Af2gV~v;9BZC`W@OWpT{K@TsVn*z>Hc
zEHaohv>u!;#h4)i44|SA{9|3rT#$ON71Yy7<D|_fy;GpbNd53pnBRE_Tz>ZDl@9OD
zoZB~ben<-CW$7L(!eGifz5L5{FWID0TxTN8Q`xqH)q{`+GUbnd+%E`YpF9Z&l76pD
z*8x1XNMT|2)@EK2xTuZwFA!7tOL5{AJ~7GF60ME!0yZ)z2^!PAI>m1@FkLREYed_T
zYnb>iD-i|rv4*RfcMS&GCHNAz3riq1)YQGXyLo8*aG-8guXxFR5>KB|R7BP9_HQ<?
z9Q)MNlSLv7_5Frhz2of96BV=`YSr%*-lpF0yH`3AfIUWLcDklAU(sN_gt%@ltzIlx
z2@9$Vw`_l_N%&c@+vwU_CXE~6vzj3m5byK>5mq38%F054PFQfJYHcT?#U@8CQ-jfT
zy1ZC(#Ww0EN{IA(5xQ(o##BBjm^9x`O4I><SG1UATRZXCaJn!Vod+@Cut9Pz`3R6l
z9<O%!KAQ)wFMmKM%QboqRc`MuG(rP$jC#)>9)O*K+Wi)AG(h|?0~`BSX55UIXG2So
z7Z-0!`5|aBM55s3>DV>}6>YLv#1MF9fALF-gtmV~N%)5%mf53?%5$m`8%~`>&~-E~
z(BA?RplDzFx`^3{1HFmSj=}@<fr3V*;o<#E!*%+o@5MB$sgHzuB9xI94s<7i+I_yg
zmhG*_A2|@&Hq<8LVtLW;)fPRF`r}2uxO2+|`Z&HC+X8yE|Kw@_$^de;)kANQTaQ!r
zha}Dx`U7c0_bNKmwu=uyT--$T=!f@2A;HVx3(y%D5yGu?W2HX63fs2?kPiNfXJpE0
zpGACIJ!By|kuLx+d~{ecj4Oi>vi~(|rC`-sEHv<d8hs61mDTn2Vc^LV4#V#Pr&vkV
zK<fR)<&hq%#dIsEy@0?3=tlq@MsxcB7&0v0{fXk#td>V1^t=oJC6THuT0qNx)M7w{
zRX?Uy=!vCy&j0L%<%}>5q(rX3$>HMiC;}64cW0a!F_1E|YG*Cd*8h2sINrx`!R`g>
zPn$EX_kDax&5uN4wri#C4I2nPimk%t>epOEYfO|lvsb3>N5n3MXZWp|C^th`v)%I7
zY9k^!Q5=w~?h7!o%3zh%J<SVYQUjy%%OH9_%_F_xVFS?r9s5-43Fl&FWI2H6Bs%aJ
zvX@{rCZ33(Sw7Q6L_%Z>Qz{<Z3)7LjPFAHXa{FWJ4M2UF-5xLu6{e{j(fv`d4r~36
z`UqyN{UqX%86u8~udoXbdd2T7nCnE{_!N{x;1b?ZtJVVlz@u>^Kc5;<e)ij=TiRW7
z<AsoVxzz%11?YBL`uai()vL;Y$sIJtd0PyiO7>x!3`z557#v#zUs5A5gzn@VyO~7Q
zW`$Mw+7%qVhlpQ3qO38mUGQ-%uZR<%+R=sJHUnseENszSGXpn!fv~mY+96O(^w$EL
z9V#{gbw69iys-EOo%=Skr|lFm9PJPfKk;r2kyo-W$HCdYs_sl}c=NmSyC-G9lG&gs
z4rDU2`nFZy<BRC*W-1o{E;ht%Z81S7O5=n)TI=t$pL%HGF#DtNKdW=PBvo}eX0j-L
z%FE~Q<4=eL$hVYVe&NNDAp*y6GiyLdj;-0BP6~P(TN;*-1Q=_`-f(X34D`S4hTgv&
zqYe(}K=657qX5+gXi{18T5`z*>&-#p{v;0Xdnn7@VQbiz-y4b1c<>iQ_Aif;R<}z-
zbLn~e4(O(k_Bj9UO*%(corMP)bnbhDo(XWGbt>N^bL_*ev^EA1{DGj}=wXqUJ@=1u
zd<rZodX5Vks?I_H67*iR6Q{tvov9gSkIv6>cD9?1cG~gi0ULxO43-a!a4{Z-gh(|t
zZ7`q|54p2>29#~?L+Lp3Jae)Rz3yE8+vfWtF9BEX-{VQa#LK_-PWyKkPwhy7B|(u8
z&!hf<iozxOZ!_Q}*C0GId3h{}cgngqSq`Sq34f{<slLGw)WQ2<WLP2WNX7VFBwFfc
z@efN?XbU1)hB&VtPYAHU+T3*F<v{`iC^|u}L`+8J3na!i=ew%?DLmqzKf`r(b!jwu
zCKqYe#+Zy5KK`RYp9{AOeSs>v*55{hjxDU?^L4>>afmr^$6$uq)k4-?`ySM4pnnWl
z2Z`;WJu+b$luXbk^=d@(iJO{qz$O#l#hSQ*ueD8>%m6eDE}JMoXb12!W>?bs`b3PM
zaT37}>XSI{B#sU~bb>D1y)E2Q-7zvHu?O@|n=`|;tsc9a(3^!DewNZwLnZNICIYn2
zZvMFLgB8z!Lf^g02C|E<g5KCW*iRny(d2=kQD%lXsYEAhYS?PqVEXVdPW=H&`GBUd
zbXLoGapP~l=W7IyJnkz?$v=3p-!B%_*<a(UEpv2t_UX;H<)2P9!~($_5tz+mCx0ep
z{qOb<%5#5vAdWBBK9#a}a40P+Yi??SM@B{l(8M4_tMgmfph~yxxf;w|b$8jO_=$n2
zjF7)f?&{h)5Z7X5v@KZM`r`Wvz<-DRA=GjM5A?oPis(Pqw|;y#-hxwa-hGjCWCe)J
z#~-V^_@KEcxj*OOWp@oat2&6t2a|PAQ5uNYglB)bBLJjf_4sxuT1}4jx10&0h(cK+
z8YFP#*g_6CuVeWRn5kw45+^4Zk!!<BgO94GH_b}b;Fuo+#^fl`ii%3E%_hS+K1F(z
z97iSLSEPFSk?R}SP_u@O-8(sh;T4W^Xu{&+!P>mevAGCh+D1CpL1RF4d8{5=m1Ab2
zwY~y>SW?(*<;hl!fz=)be>bRf!VOWSXrX~_M^u4OM1w^S!2`ItTdLKoj$GchcjHwZ
zg$I%-r~O*25c-{E15QD*iCgD`!>_k`y1Aw%O=(vPL$hN-nRlxb1Xhw?G`O+%5z9U;
z@`?~~f1qW<&i(QPJ)ozv_2DZ%^{v&>{!%19$}1re^~C5@&*m>SB>AE)E|AONY1Xf>
zFeyY45`~J{I&KhiA$K%)RGFoiFBLejvA+H@F%flj<8uY|P_I8KHr?wtZ<_ug7Np{g
zVsgPe!OGO6@H$LaNRh^WT=33a07Gpc<VXVR<I;T$mu~LmxBlv$Udk+w9;7oB22Yy#
zGYxIiB(Gn2cgjUi82)BLj@4KXc$;#Gh_OYb566lPgp75qFwW^|nwv#USLaVF!DEo!
z3jas8lvb>nhD0NIa5MdYia<P+goG<Ga>D9cI(3G;<P8&H+k@)n4v-c)yz=8--R+wJ
zXN7z%y<{G^PVuQ&rS1>>4GX7eH(rFG)8db?FqocpA8w?FK(`YTo%=QZJR?7&r8}2s
zWTuv*eEhbrvZO;J*j3K_rh&FTp3CrU(fOzD|C9^Bou(V`Z`$NxyV=w)LL*46wGTNB
z5V6raO-*0g*Cy5qO$`FwER0+~U8L(siZGfg=ZrpS21KFeSdqrvXceqgg?vS8QZGs@
z6B3*7m@t&wnjJ+*{d@CB<^l>BP~cxg)bY4bQMom@2YY6MYsT{P@kPKmUvN@a6euC6
z0(-KXa(;KyKmJdBc#1#|+bP}PIa`c&V1LlvwR4YO;~I)QGz1a}z8M;eI#5${{$Wc~
zFV8ZN+T|7h5HXl7War9yu`2eL2Rheo=uQ^V0sL`#eug(VhQJ+qWOZq4k2U%JWBr<Y
z$o?CW@!9h@15{){HumH_7Z=`QG#w>rol80fI(*3|rjN>e=zrS>zXQyDxU7ufen9t6
z+t5$ZL;Kv|Z1FkUy~fY>>4sq&Q@knMd*x+}gJYRScG9j8>xFyTO@yh*P`W1%XtOQ^
zE=c~{HR`zOX_}~GdpH0Db#rmhF)$DgB(_mxG+OfSKW>iYmwtGB-F9bhKQR^YkKSGL
zytqUi6J5C?le4z2&hB-oNfN@2*5z+tge$V!sXkNa!G-~F)ski#J!;Gq#8J7#QKK=S
ze~N3pP7<Oi=$*7xU0gojql^TlXGcQZQ@9){x4iZS=~{XJ6F#`Q5I>s-pJb#+lPg_J
z)#YCa@I1>AW$vxZ;(Iqbt@exiMk`ab__)Eoa!L@0`df2Fcx_r5%)ua{yGiy|_6U#$
z*pD6h-|60v2HJo7)nZ7Z?80$VPzj&A_r6Ej<t@fZ@y51Nu`vC?y2S3#OA4-;=RqaQ
zij@^`i#oQ?CCv-Q_9*~$Z10MsL{8pGWgYRZ)+j8yj*w?BBQMjS`sOjK2mJEs{<5+#
z85tP~{HYW?JoSf0goK37&dvcQMc|?}<B6v!l<9ZSty@61{!t+>;DF#Y*TmEdP;Wyf
zIJ3dh-vC`*z~}iB6s))RLpP>m)J7owu^ECzP!AB58q-9|P7LD%1r&N&8R!B?$teRH
zU*Q|>4XyOzaV1b9T$`Laf1(6z6lLQ>TYF^ob0dT?7(+%v=^I<$pNFF?Pg+P&W?|l`
zPJ7V$oEA?Uo8yb?nquK)waCX#d%H!10}EFeI1xtoH|!_%j^747Pi;RiHUrBOaUw%6
zps%xa`tx1Hw1^-P{9t;63aRYBCp5ovy)k#w&5TteCb?5`Qqmvmr6(|$)xIe1<2C8*
z>^D4ZZriP~egjU_%l4OXtK<LpL5c;=u=j$xN~~aKIJ|4i$G@nRIzi*pnVHZYhRF&>
zpJ&K?+`Mj6PO95S`wToYyl1xAilL^P#wM7so|qnvvrR@mjq^1DXVQ^-wYJRfZsy6K
z<RkqH-+-d#^zK^S>%)iA!HE@SdNjYW{NUt>rN(Jra<cM$CAh}R*~M5$)2-T>;=86@
zLvw#~y4Q!|cXK?6ptUX`VF);wh5=NrK5>WIY8L5NY*j3I3pF!PSpyT$t;6E$e5-jK
z;gc-WGrLbr35kq0qh;kSz@qn3<i7#a?@=(t^H0eQ%0a+Ua|aNhV|P6cqFoBjlA*#W
z-=uKV{1f1v3q6Y!Mq8_*+cP;$U8Vb8ok9n|JX^bFlzBTV2?0~wdc(f;b#IyI^G}B}
z2OKVQL>?9vz%I~z-{fI@AhV(k=89?NMiohg;S=$G=c{%2ERxsOZ-*WCXBGP;L8T#J
zw9|Js%29#>eyh`p>`URdDM(*zfEj3?fK52d!Ojw@XZA95gk8>}rEe5W{WdaC{Je^p
z9TblPo*z<5>O>dpqMWgg_!4WNCy<-#4ocgbHf3)vyZHkV{RRME!?L)7n(G$ctw@SM
zQ;7Tq{H6IJ94Y;;{{M1Pqm>0rsmwrN_u=dnOZ+XoqJPM*{<voqH}*){#hT&TGS-j~
zV7At>j-DRe-CUp&v6<&g2mCvM!&#Gdsx&xxgC5|r%~uiLCJJqNqZVMK44;E}OF>Ku
zK!8S|a}6L46Q5llxI><D+hPLwP%VNtjDwkuxi=AT8F=G*Mh{vbYkwS{x4Y;BTal*%
zTP#-r%!5GV4hz-$C2j$`fDQuVpJ%23y3)p61yWNg4uU!ncyG_bV_#dW9RutSL@;eA
zYeTT7v-BR`M%-djOMCVH@RydeIbyQw>6T~P65CVBQM>p^2b`aUEmOK@sD8u);W?-k
zfI1n}_*-p=RF4~_K<3Z*<d!Nt)+H8OnuFU@X08G4MaNA@5a47kuQB_sW}GfPPq4W8
zE`5ra8|7SNeBUbw1EZW05O*JzDo~BtUp$@$hur|B(*ZIXF!|0L%(U-46Kpi!dW1Q-
z-O*dh=G=lUvSmKTmD;|9<8v_fRvs(sOW9HwNtr6Fmwj?Tds8Se)9?|wNMi+XXWMEC
zCy;Vgm}n2fwj6MF#5By6f(Bs@K&mLOTz*6-@s1T1OfuNvf|?J|x@El^NMe|<T3j5-
znb4WQo=$z;aLc+or5kL*{5-6B=2yT~0*-8f>^`mlQAuit!~$-PNVMcft_?sfkM(Uu
zo0I&%WeN$n{uzCvChoP6Cx+@2M~L}icM*I&%EPxOCaLw(%~4KAI9y4O&)?AKs)<Vl
zWdWCCdb%J~2%-QG<0XU@6h`p~T6%tgWdr2*s6^RWS*A5^|Gu@st;6AKU~&yHQP9qe
zsC^)V$AVe;0w>fFEO~VulgIWb+yfa%et$Palj-42K+4qP;R&Cb0J=K)dsU1v$O(AF
z;Cd}ao**Qb(c^qG_%(n{n*e{<0wcf;-BTQF>4_XmmE>2aGMMhNt59tW1tOR(mzqN@
zcrt5Hi3a+IjaaTYrjHB*mlSPG16!_KG}B;Y*Cfvdb)AOhVQ5&|Q1{DT9>}+YasZl;
z=!g1s0P~GoJpFf}YF}5+vi<(Ra7ANme+~KHDap;^7r+e8D$l#(YK<~KwtGm#=Y=`P
zHCz}&qJ(q}u#9(`#v^~He-6}f|KZSix=?AspQ6$KhbZ=N<A)%o2TF11?5qZ$jTSE7
zX(5Humz~eM9h81+dIAUB<ITVLy;-38$L5{inGBHQIN!foZpYH(>FG#;j_%Ph8SI1*
zq+z{xkkJ7$8h`W03)k?9D^nTRXShCo1-^5lU)uRtrX3nuj9<TM)Zl$jjvxlEpS-z&
z9Hg_$+fl?<nrFlm^l0(0_QrDsQxdKQp=47<U~N(^;&B&kB)DOSfx2?6EWZAaNRbl#
zaJyx!!YnA*K*0Rd$?Yshp=<7mVs9^d>84wQ#vNPZ!OTEiuyVZ95tRdNU+I5b2>_(P
z?z>9$7VZ)GIhF>QFb-2zCI412hud9|JvhLY2h6}jz!@`szU^g6$?OdFkyIC#!JHrn
z^Ui5^V`U*mexWHv>E$RMJ246-{|bq-*=SDF{#-459J|ej($dm|dEdt}M8P$FiUatb
znt}Y$&opG-LkXKg1P*8^<X2CAEWLpPvUZt<JMKB)YT>WGrYsJ2dLjdcgBNXmSK>eq
z6=ot(Q!Ew}6Go~MBr>J@s~ivu<)b!0x6RiBd>OSy%(g#z-pKC0bp5*}vgDJUwWKu@
zZ`Bb(YCK0ZzcW`ZNA83QI1ZYW+gDCX_gxfsH_Q4);@{jqbO^Le65$6~*}o7*LITJl
z4~1vkzz-L@f1?}#2K9a?EO~Kw4lZF*rM$*yLoTv*U|VD4M<&6{PlzJ+WXP&V_M;-D
zEm_p+YF8xa(sp-SCGxN(0RRdrK5ibKcs46SU~K{cgC`)V4gK}&&me(C7Ieb*kHe^a
zCioQuz>3Nuk7CnYH#YsbVC3!|F<`W|`W_x%BDANHh=N%j4%DRGW6~}{ygdEDV7s>8
zb7VpUfP83ee@BsQY*-fvymymtULJ8(y*|C5qPwq)Q>VUv@410hbYBxi@AtB4wlgPP
zP%%C-nU+r~;qzhw<`>JU?+Zjdf9lMip|KT_xz3Q^r`(%FWX%0buR41G|2^=UJpDV6
zJe^kDzFT-Ci3{u>eu-!TQyxoeF|^{+(jw$gq>iqxfEHus6kHe)Uzh~B-tFa)5HECd
zZUG(@o}{Ft!HJ1%0J`+XGyD7aEH|Rn*3^86>=KodS|7>~aR;$#>>M1#kzHV!G}4u7
z&>8i_(Aw`!W1^y>zIgf4WHjf^NVYTuFo<<}iswUsXeQvj-Xzh`&;ak-(bWY70y-=e
zFkx6zT33AM`&?9c`DTlUR?q9bg95w)Fh8l3T=<oep6BFVuaA-9N`@qC#v0a?w$%yi
zu?)3-(L10$M|z?g_&n(2^Pul2UiNYEH+TGjUpPO;+!eH`t<2h;b6hZO_ks!QvTtEl
z3sMjFccoQT9qsLY^Yw1!o_CPz3l1miXeCwMm{nIEb4<_T^!$9YhH0#=tu2W-hP2i@
z%qzbyD8#K8?`cU8;7*=@MEytIoAndLdLzzTM7>a?&>HszK=y{{4aV;E$qIGxNZXNQ
z*miW+-}`FTW`_nwys;c3u*u(7nUc?aO%)wv3WF(>R~GK(q7eJMPX7rdqg1z^P0myl
zt7eQQzS@w3Kd3b6>g?lEg;w8rb2;Wre#L5PZuWvlKzd9d0dANr;PC&k2mE)^N=jG&
zDeJqaOaCYl&*Tg2Fl%nSA3p|w*pi0(YcqFu_rv4k8u~1$Y!Yhf;Am=PCXiR+-f`;6
z<656Wr&+VOvJwd$Oak2nyVE{C2q___qk9EnA>2Dq^C2^acg`1cPO2r}_!~CB8QlGt
z`2buILqkLCocAXwpO~2=!6mfY987cf0O(3f3XdE3@0I~r9G955z^0|sja&X_M-hk0
z9kKYe01ENnq5fCjH>09#JA%G`Mf_D$ulb`Rv;x9qI(0=KKA~WgFSj>GeYrB{;g3^h
z!!@T=WVG47CovECG&VPXomc~dD!9UOOKo!3v)SGqauD?bsBOiUz9DUA6>alj=taBd
z1)tJBpQt+}n`U3FV;82=9{wIF{6`<~e;<*~*d4$9Tzzg0x2ZS-?ixA$e9P(ij1@u&
zWWCh<Bz&RXEf&Dn-wX_d@|6mY3Dirz!2?M`OG}G4h+Y8x6ThZ!?->~*K#;~i@$p|{
z=`_Lmr?UsskUE%wdn7$Qo#EY!DT9%^`l(tcdJu~NTAkOxbY^Hsj^zF0^wb`N4ab8t
zuH0^S5EK)Ph>G>i+`Jc9Gesdl0)1luZ`ze99vjDMfx8XSYT@vCazEym07u>B)m21v
zw69e53U~njR9C-w_pURTzCmS~L*YwrbCI$^&LL*qL;pG|SWVI-$Y#`eD(V>MPX1h5
zIFs=SQ5-Sa`scKgXCT4AnTFadeR<{C9~{X8&Z6IjTPoJaFHWFa>7Q{Cs^3z5DIVIJ
zdPC5rU5-FeA7{*RDI|>WK&ld~C-Zikkh3|v@vbRaz_L~i=7!d;Mr(C?&~q9=A7z|r
zz{?cW{QtoOB(Z96mkS|uL)_0Z;5<1Pw<qlBlxe_c12@4qIG`Hw^9t3gwIDqcbD!W!
z%-91`=%1(<zvy^XA&KHIoc5dFb8{&{YKy^CnSqLm$|_fb$BiQ#NCHfDe>S)Z1nH6q
zKHlC4KmZaq$7}>5Fxx|LytRrq3ajRI$MO^}3h(>IK*kBsmpN*eowo*JzyY^voNy@a
z8jls&<}!ps<Cu*AtOO#`BDO}d5njKBUVqQdHkxN7At3=lSsiQ(?vcgP5)v-}UvD?B
zVLE?O-8nK639_?*`>`Fo?;rOoCSV<-btJ@mQ(p48yg|ozVD%b&!+>mHAxf7vqxW2J
z`fib?Q#V~9dzAeApR|ecy<<hV>xza4N2>2Bv<XKy4G88a&)Vkt{nEhvF>6LT!ceic
zzt{Vx#XG_oRIk8|dmS4iC1Z}SE?8W$)BS0l&pL|Fo36~#8e;YFdBRX4SgX4c9Tn4_
zztFEUV*|lWb8s>K;n%E7!p}9dl(JPKSKcQr1OE`by6oGDk`fVj?=`3k1Tn%t=vCz?
zV~cR9hlvQkiH{0_$`z+fTabj?tnVz*X}8DPwzR8)pFQ5N019z;x$|xXH|7!Bjxsi8
zU#{0q7yE*eho+TkAgZ;a&gJAWnLyC*MVFI1*Z|<_&Zo_;Quv$?--9)e#(Vi*+x9{W
z{CMf{?(B>g`jR^v^~TA5D`RqstN^>sF!1onL8%k<K<<V7xzY3Pn9M2or}~5=NXflI
zFRk9)+lvRM%7VuQ0hiOhK1laLBQw;Q$Y&A_L;@dh8Q!I@mI6-u2I(Kwb~OpUR1Rr-
z+vfwY<8)VIYBN4p7gDg>MsLs$d$!@ywYQDTRbpjpLxE!mH^;V-!W9Ktp`0+*3dq)0
zf0UWf9RC;}K0{<UOV~lGqjh+Ff^P~7*H5_6a2r+dIX;E5ts3VXIrJB{az9#HsWWl-
z1K-``BSu}qQ2*33CoVj0Kx$fh*U$jn8mgZ5N;G+BfTYCl4djt4>O=jAqcJMX;?vz9
z{IZwbY`-KXUM`;KdH&^VL<Zji0uRP&*?{WX*QrYPFTL&b*i~Yok%i&zIc{3E_{1HM
zGv$(ra9^Lm;^eyrT(}IZtaOJIY+r$;Rbz7!xErYtubRA(_4M^y0q&4o1wRF>a6J{J
zQ@Qd}ne_L`b#!9;S!N#Br<4_v11VDnSJEjo*X6=O?KHctib!7%q37U3pouCEfHad5
z_zCm_roC@l;_na_!ZspFI^Zu&PxF0Yc22hGdXep$n7*|*@gWxTQH3oj*`CCo-0VIT
zs`>JeATtkVz}s+G=^|HHsjuRx3dz3XaU(P;@&T78P_MY*GuwokP}qvGPXBNNPy9><
zvnnSKN#k|QKX`0V&@S_tZ`!0vT>0(6dq=HmCH)yM8XmipdS#NFz0FSO=9JC>dR&hv
zN6t2VQ(w;AfsJaJ)W8ru#TlDRyvM)Bl}#Bu*w(hT%?k^u>Q$C&JdGeW2afl0(fj85
zi-=Jy8S}drE_jJ|Hxy1M#p$f@%yjzn;GA<S0;d{8U0l};j;?pC=I;P62R0oBaIxG^
zXr#=|X+Af+91X(2j=ke5l*<%Lv#@1zuJyP%7YtUNJj;85M%I)FoQdFiKSM<&0r6fs
zFDSw1RNxen1>Ts?h%}sMY}(gPe?^s84eqfc6@Wmz{-W|{ZRX?SlVgbA8}=*X7s?@2
zHHWVIV74J+mEkIiWK)BQoHJ_&+Oj*tg%$JTqqed3gfs6O9V@kC^~Lt8)7AQM+CPa2
z-CU>;Rav2TC1x;rL+Fz_*dG14X7I-&_Iz)8Mw11d3fz}HSrW~yojI)@zK@4*>q=Lx
zkR2RMn@)S)WMB-X>;+O?y_!O61tcb!DaYurtpGP(tc|7*6?JM3sSkE5*Cuh@JtoJc
ze-yg+@%P8(aMS@FN!3zaL=zK}BwmjMkODaj#QBE(Nne3LPb}%c&bqTGA}Mi`HgNWf
znWY@;{j+D!u9o>i;=;vTket6i(-M+ZY4qfgOkio6oqd!=fk-YJC&+1FVjnngjeyBq
zB9ZlFveX3-zk<uiZLjWr#}1w0I_#P=vJD?+hVyx|nUj$2OQm^$lEIVxicRxgrQ@vj
zYX$Jwd=HN){#YTsy2lQb`QFtLF+%mV8FXuSFG(sB+6K-)=PRAuwfD_Z>Kkba$zHBb
zEWpPtl%u{fP=SURGUU}~%`GOKwpm2_Q|Qa~4vUiGkG6Zj5<9-FDRg&ljr9o$5t!-t
z{hR5&*gItU9@|w2#P`o<oiIy^@reAlHZM)aO5o{$(}SDo`2rqW&)b$?AA8RR8VDfc
zjWA4(db!8IFgTEd@MIALBzzGP8UnlD6XhfbW<G^}X@mho?gEYn!@dMY%UO1*RNg)>
zMAX5ZGxw{tShDSfbRoxOcTiZ95tf52p}%}*Zm714f8pU0*sQcn7^z1Pz8+>}XOCox
zV}p1yVHp{j2n%_6`72Q2JpSgeTkrLb3j^ywmhzt48Ra`YXVuDKPYW2XWC%?}o3Uua
zem0n~GC#DMz<jiH4#rUK*`Xg*7x=%Mq#GFx9^U+9E{~fp!o1PO7a&l_3q#gkRFjbS
zv0~1>E|?|Nv1x=WKR2BIm$l8Ns(i^;`0T5gxL)={;FFnK53&%{Hxnl7D$z!BX0ayt
zdZB)~7G(?H*R@j9-A{*6Vx%DtG?K!YDNl;r#!F>ki-%wX2)UGN7nh4pdo6op<6VOZ
z#Y^ZqRQsMcJk*zCrr|PbsNOBwBn+%+a0DD>DCJ(RM#sWA8yzgwaV?4Ib#2)t1vLmU
zUvR!K*1m0TK?{nE!~m(8q}<#|gM%`jcW26QSrYMb-`uRMtl*RiAeAOi7!!`Oz;A-M
zQ#erBQOVwD)t4<~SWQ<10=_9KM;iEnzwe#|Q!X;g>gs+t*%&}BW3$RHh>PC(LFz0L
z@~#OqLO*u3PnTN!WwCAZjrS1|+3hz6|GVTzgE((}d46Ow3>j}~c0>bO%lev+{zz!!
zd@>=y`aN{ZIr557?=mZyyUG&vVXitnbs&j?oWpnCLF0(%lb1ACB`8TxH0n9wc6nv3
zZ60PfkjI_q51qfRodefq$={V)+jbn8h2-PekW=ugobQ_Oh~Xzz;JlGKEK7B+R7M*<
zbyY&v)}7Txe)q={PmtAeBe~vx65SPX)Rg=c<eC!P6MJk<U2NMc^LR}KxWlxzrjo-X
zaqZ~NSu?5*h&7duL&Kc3s@tOwM21XRXq;~f<rXlNhirRT9-Bh25j=Sy6?y0zACCc{
zSWgzWwz{S(&Ay6>b#tYpr1XQI$8p$^gPgw1lpbYmk2a7QTy3)=JU@2<>I5ht00qP<
zs;|yIPuKyT=~v)UEu1XTLCDR`{p{$mcz9^taJd}pPKFO!QD^Wpz#Grid;&qvZQmGv
zeSA(#Obo)6VhRdqTrc(|f$v|OB^>!pIg<%B=OT#0If+pP>FM!c&rVayA2Mb^+%IxE
z5CV0kOxl9*<F^fmA}4*m{2U&3&Qimqu{HghfM*gv4I_(GW7}}nf7=04&!{+jM`;Mc
zunbvBsy3!LixkQk2C(zv^%UWnErl`Jl~fcI8XOJAxU(Ok1=_1^PYl7|6vt*m62_G@
z8a)z23JTE2Ivom?LJj1}8KGIc-DliwQziLd+B!0n@@=59s+N3D4_6i|zxb|<@}VD(
zyS0z=9~RS3W_&49H-QJGpIAb)Aj{QQPoCgecXq*IBdx!bynNgKPiQ=yp6qes^3;9B
z{@^sW`IAf>&Ur0bnSna}%QV<hrNIdX)YJwbio+K+Q|n}vy8)t>n?;B`6gH25wK;|=
z^<oL~Cqf5|>HyzE?_ioBs(EB0dw-?<mYjf~;L*iJ$H>S@!+9tkTYtW)h8)@lLxbGI
zrw3&QJ;#ItiEPD#(Y4hyw6rILseIb@w7ic`q|KW|`F|SouI{XwD@;Enn{sSem@6fj
zu1`H%h)arlaHJ-FBVUM=Z@a&&_Q@z|(g$jN65c^R-#1ajofn^I1B%enz(9~owJ{Bq
zv$7e1%`;jJL-N0cS(7(4ZM#~_@gn62k%G>5_%=4WZc#Rs|7>TtyDK7Z-SD^m7K*s&
zYH^9#?XlDAi34&SUH#VKi}|IteNF0{x8HZ&r@)syK@R~UPU^CHu}s~5D2b!)s2z_6
zB((vH0tP;Q2neWtiOcd!B%JUC4ijmymX!$&hyn%AcDYUq5*R=6<;$0+Af?>d#pP%_
zznBSRG5sG|Zyi--`n?YmqbP`u0tO0<3esKDiUQINf|PV4-8?7?N{4_nlF|awaOf`S
z?(XjT?W3RX`_DUTW{qpjaCn~k-1olszV@}R%U66$<z<#qJ}ld~%Z}u22DK)rjK6q$
z*MK@Cgqd-N8GI0h+u37gXRjX^2#0Wl*7=9}{Pzm7hu*);It@(lL2Sg5X&o!i>oVzx
zVkg__pIsY@SXcYBMwi-m&ibi)HJ%cEJg*Gs=nO3(9|Bi6WzrOSbCx6~II?~y7TCbS
zq0cj!8gFl+JBYqz&2c-wfT|dk4Po;pHMQ>I-nXwa_)TNcmYMXpQj@9+<aQgk^SY2B
zza(9TM#^`z>_p-^I?;6WhN^v-W|*4$Pv+I`L<x2*$p*$oG*2kkeE$69As?#`u_B3~
z+z7r1GzLi;<EhlH1WY=a(p_z)R>;jwvD;fN=21kF?B<F7Y-6?FcRv??k6J`jTH@so
zWY2WQ|9gH7m-?6MM1R)d{GKdZHnMiS`9uW^$^%;4IH-4-L6f)~e^~4h8{4m@8Q@Fq
zt_<Y#78q`{-@Kg!ous4N-nT`vlvoXa{|7z^3U95c#M`F-l)?XOfWYE5Y1FrGU(+pw
zZ}RA%OX3^e)moD?nutMjhx(|f7^<=w>t|8nw4a;n(Q03~cLfiEMj}Xv3nFRlZ*|e)
zMp}_CGfXsVl&q)~OonE6D5n2wNgLK`f6Gv<{(IZZ$9Z4*LGR&6#e5yF2Fl;<1-CKM
zRJd=)@wcYO`$Px$mE{D)q+N9#8~fH>NCCDFn6G-4ii5uT7L;%)361J2shl#Z>8>A3
zo)S3(f8E_Z@h?JP92Rag=2Ck`-(PwNEs@*GZR>t>tsZ90gpJXXjpDKpg2$z{Q)3UY
zbITU|DbW%7iY%|Zj!uf--3G{M^TsCPV$LbA_?3&c#qhk2h?tqxl3p-~WcumK6Pq@p
zG}J?spaw-M6wBf^Hf(|F54h|<nZG6TWM#cj8Jb0wj}2YrAHm6Ow%=9zw-e#@NczQw
zu___Cq>ypBLuu%0K~s{hzghtKx%sV$(-Kb*Bg#F?=rq)nP-oD|3fuSNd+M~#OKC3L
z(?68fW<6><M~pGX8%MqAz}ac=l+@+Dx^i|XZFLl<RQ}0|v|hkW^J>BUK_WZ9-1ewE
zo>hd{Vx9jP6mA&nb{xN~tdb&2xSCZ|)Jj4=Z}>%JUMp>itA6Nf)c;XsjK`Ad^zsrn
zNyQrh)%)KmQj#|3Psv`}lRGEAX3YxIFlXp3%xzG}py1}ZG|@yQ^-*p5WfR4G)<l!V
zhI9(}m?jg$esw9?mexNn36qGqULYnEMU3<rEoud1rClwO?|JIH!m*H$aA@3TeA|4e
zA)2$1sikbe<Ea^X(LYm0W+h)@JXq{#<ttU*SNQ0l1*<v&y?7u&{-YbN_Rfmt7<;XP
z){O_Qm#&_vju@Wu^WzmN?+?kz6d)<#{3!q(Ah&+mG%_?WzW(@J7WUM8q!V&6xK$29
z34h<awpwoGcTXgMz=bC^J7i)Z)t-{2)V_<S{`&Io*u;g7{Qov3A_&SU6SSX_-V<u4
zQ8ZMk)To}ZG3^{cvfo1LBe7|$l){2<&&sOe#&I`PtbS*0gm5s$$3Whcp6-yAE;``)
z3Cq9*IWE&r6c3r5p=1b>u6TIBJy+)|;dXSwn}3FNw6x3BQCFw^m9_FcSm`>&-Vi&?
zdb4}6DbMgytr5)K0=R$aCyBgh^V_ael*6cOqT=V7Z~VS_-zJLX&lQ*OCj6FP2*eB@
z%}(tVOM$jus^b<b_5bAPU+Oo$F8)e4O!Gflplw!KFIb;1BkcMCuIeoue<J*qhuhTW
zq&;%PFvl}K0}atTBj;ce(c;Vt8sxY!P>i63Q)8sOG;&0829yn5U_k?NJdn6{@paVD
z^u1GIzj#Ew2y3MtzQWiXWt_?Z;mrwsNv6Am<A>|W13X+BE8l?=hZ#eIR?ooJ2e_58
z$TFAy@*cp9|F~Z`TRvSDyye`!Qo3VG%lK&h33KCxZW<K9)87Ovjf{L(`!5{`qN3TE
zS+ra5UNBSPTAxa5C-nWCE*Z}x9Sl#oL~f61D&gyxV~d9+Qu)efxo@-bUVH9JP_Lzx
z_RE0|y+{)waWV0kqX2;!!8%)~e!Hn&1ksqb?R;`TVUMZsjK_e|4xg-Ubv@?y(w=W5
zUqG!vClQZ-0#!NJ1x_L9L85;aJ}L))VW(paS7IFt4Ml9_^9^p*pwH{KN=w%K0W+$N
zC3Hy-;Ww5EZV?iY2ZSh!ArbvFsIG1`rWlW=3na5z0lmTanV+ACz&%IA9iAu>aEp+z
z4w{LyY}v5c8>2cvCyO!;z`urqO<lK6hP$Eh+1a_mwOAYGfB+DBB>cXpm1e$s?Neb9
z6Z_l<^9AN)eO6g9A%ev8)!Zfg^u_tc&3CMk`s=pd>h!!OZ3q)9$9${Kxhcvfxo!sI
z)Q;iXrS}Fu5twt>k<B3-veWV^{GmvrP!)%Lye*h`OhH^b)Dm1Ne61!?%#LKdAAdPK
z!_7=6vfP3C`9RImv|Nbfi@4g3Ux$4EC{)4bqT5{emdY`x3uhIT0M>Zt#>w@~#&hha
zDyly(>K{6S1xXKtYaWfepVO3D0Xwd}sHwlTM*#Q>dbM~-&>`~e+iCy-->+_80f=v|
zXu=I}bJ`rgoE%1iClXD{!EXU?Ne1+T3aFNCaRPCG8+2zVcaeERjm@i`LZ=}n=EU7j
zlEISU+$guFh-vL?lVa3IL!O$vQ{CC>V|rl~uwKqL$u-^6y>!~hZ8zUrg3ybpsS?+6
ztW6H&S1T0czMb(6@R9U%DlXk@{VGaT<@3f4b|HD`qu|cK3PFbkS%LAv0<r?J`t*wv
z_hEm3q5X)X1@3F!HM1P_OEVf}gVrVCp}EW^ID56Pvh<_k1em#^N)=NkH?O#znK0U=
zJjo>9H2rex?7(R(|LSAjWrQH=ThplReLS;pII@(L`r%)W&q~H&*}fV_-|ov`E_0?N
z6nZ<%SeZa2mUfK3Q_)*X4Yt;$yG&ZoG?qRAwF5x#A1+tNAv@XxS@YR5$EEIM2WMyH
z#Rt%*2nY(g34MxCRa^MU?IYVxehk!NH>xW5T0OHIl5vzE&V%7Lw5OcAwt(XDJn+2s
z=pmm+_5H#wi#2_yM;~(S#9iDcnp9gg<L6lxNu;11v^=7*HF;{O)ghslu1MciuCdtt
zg@{a|`c=kUVlC>I8p<20$eSU-kvZzRax6a5F>?`d7h3aM1g}A4?Y!s4(pq0$(_s<8
zM&2?~^&EY6h6_#aC-oKN&+X|;z7mx-IMQ@(B}{xoKbnW{k-(JfcjvnPrEoE-4g=_=
zDQ2krGUHL<R)25Cadc=kvu}1>I*_76zrf<hdVKH3^3aIY?x<sjr#ij-wZF?%7qakv
z>^_&>>g(g(rP_<LiHj{zq9PpB-<(9M+hX~mxB-A!90Vu4N8s@w&3{CmqlR8*%y;Li
zPn(U5dVl^(5-|&BY2322PLBM<RU{`@V)Hoom>o}g*r?Z7-_JFskQ6^+f5&PcL{`sD
z24pi76xcayN1Hx>`${O6Z{xVuxW7^xo~T(r-jc4c<CN>H%&TU+wH_djJhkKhY>AM#
zS5}Sp67sMTOw09%HZ>hC@RqAG9r!V`S*0nOuBuaFLAb+SF4NJ`R!esrTj_h^_u6Hd
zZshbdf<yat#LU<51S->m8_zD4cPG7#m?<89-BVv5aRw<L0igtdi|Z!&YsCU!@4H8}
z^HjAUGKdua==2-en=|Be%O+>=K>F`7{M^`sP4A|hp+pb*$o6hWbLiPQQE0Y0Rgi*U
zFmw%>XlKMC5|UC=cR+WtyWG*byu78blnSK0LMZXZlb#%ey^4_%Y2YIZB;(1OQi~m(
zNUM%dowQY+HhTN}!kVY7TI-p)yBmuc?>siZl99+Ib6h1k<4|vN26h$esay90j+pcY
z1GP_hC$i}QOxB)Z)+{7DSa>b-Z_zrrmW*638*8%O+VHN?nD@;wZuuf%d<+}C!-LW<
zZ_M`HPKdlZV|Cz1_G7r)_Xb~2ZzSnx>%Qb$Aa-yr&@xQ*|FvVNFl~v&;EC~jtl&`h
zoh%ZmoZ|^OBPHXf;Itm1QpZt`+a!or(}*P-K&!^L=jXip#`-?NN)o)=x<wMjhT3qL
zOqThtEY{p{_lLSqXTZ_TyH;ruIry^tmtX9&4=n9kc@4_1kF!v3+si)NcNTG^=N+=C
zcR3#zv%V9M9mwNE<D-!izAYulJ4M>Ejy!>(#@=l8C@9(QIE>cRJj(g_LSJ9Mq(QaZ
zff<^h<p<-=dj6K+y2F3A+w*7ymOkX`SA}_InyBWP=xDcRHY(hKggH%?7Ap6)bW5ye
zo?sMYiWjjI{>Pq0zoV@mHa*c-y*A8LtbYGL9_GO0p11os+`GwNl)vrc-~Ww5-ePO9
zc_#!$2E8x&B8sY0Fls1qJmHWzH%Wcu+*7cVYx!~lB}rtmVpUI@vBJy@uzBNlliy88
zo_&-_Iu)+0<<{%__rp!*Sf5E|tXi5k)GIfcy!<Mu#5IxQP3d#0PKe<YbN%dbST6w)
zZ6aaZMNfA(>7;}hBQ0%4li&MK$iD%VBHAa*zy-$@O;Ia%2*Sp>&IGm@rlzJeTKec%
z$5t9m<PeGb+BH=+ZKs|NFH6`%DUtd>%{x&p?I)GN&?DCl91rhuyUQ=;vYw<3?lk*<
zAGmWm)f`V{$Tl}|{Ga`UqrfMB!sf@Oy;tN8w1UvM)pXq>%FfQ7L&kr$!{CSIOaqaF
zge;-k>6VDEpC2gnPpn0Uikj&v49Pp_Qc*4D*%-1B_v#Irit=*pI=O0ezf9~?8aA7m
z`%Z;>e&@yD@<M0sPW7m-&toon5@_!%_x-5O&r)T9Dj<DOp~Jgb<T(i?C5AaUML3U=
zu9Pn~do&ZxRAmO@oiIs2*rrs8;o|Y3<lY@=W8uqvYo=+M`A=M}+(hh@<}?*myNsVc
z#TXjMw&j|pdZuf|0_D>3^?gH{rHOGUg*bD3GTjGi+QC9RIzxeavp{BS+<)`i8)z{y
zanhWO<TOQFMz5}kpGFc8H3Digu+l|rBpaI1<l``^hb-9yT9@U+_oSkt@}tj~?Zy<|
z8(Bv-*szuMQ;}c3G(X;R`fXgWt)l#ya+DxsQXTB5z?BFR!HiM%E;V)aqfX>pelnMm
zqB<j-Gr@8DR=Mih3l$+sQgvK0jStEhUA%G@En?HN$ucPnXP+J$bS^lLzf7F!_It%J
z_E{GZsk)kSwoSWnowVts)+@|eXkq4u=19Lwz7xGjj!43@2zfT#ce7697-u7e)Ll5I
zGu3y#mCoKdWl(E#mb!G;gze+#XYyHL-Telg%F~wFx6tRcVM=M3D@(3?#GbKoy{c;^
z=$g4qQJW<+p{#k>@ud&Azny2W)z>Pt<`OF+nOR&=c5E#z4TjHb)CPUNkHMC+pfPDu
zoi9aFqhoHl3!@e*#bM6T`nD^t<kD^}?m9sOA56`HCtX{X*3%5Akvl?R3@7;?gZ951
zInd+EsD0pMXJ==uHZe8z6{SEm9&u(V(H6{na;V{BzQB&*cO{Om3N0$R8U+p2%LqR4
zKUH9bp=>Cr1-R2F-B?M~Re{sbQaP!y74$lh>~m0@pbM*4iBsmnPu|NiL+5ht!1@ll
zN-8`$%3sP4u*_1*%1gbFDKP8(7YfswTer9@StI+|)5EDdw(fJ5xN%;!CLoAhAm_(R
zq{5xB7_K=HcdL%4a!|-!Kq}iKC)eh^GyK@$TKKqo4a;0O|6H3Y;{(I?k_GZ}jL{uG
z&g~^W&IwC<Zo-~P5Ao}!25@j0E5O#*JN@s(KxSXxth0c=lY4Tz=SSbJ$w(wxXY1?}
zsI^<Draa9Bm&-Cc5<GeNQYU_St*CReH~D+TWXNSrI_nS0nEi-a@mGQz$q4PgX&?jJ
z_!*5et{^W*Y%x1HAchhhoR5~h(OtKO{!`LxkjSuby}+iz%`dZ`-bFOD6O>t(M@HUe
zcEw{s=D#FccH76ve&O_+^5W~(i2S$8OQ-7!&~yjziL?~TDa_5Spqa19SeeiYJZ!SW
z@U%nLUz?spWDGFRrN(uo_nl6T`4J>)>Z;*MFAooCCn+UjL3)A)j+0%Tt>*W_jpk;X
zz66Tf#GLY5+LQT7f=$NLO|>$Px}>DM#n~a2`p;5QQ(hqkOx?zJ`^uCrb^hFK*mu3{
zZs8yIU$XBBrpwy>XQfh&CgXf~>Xv-JtPzn1nOUz1T)yNNnTH!i<KKC~ML^(_>P(`_
zX)2R)_kY(iWhb1x7y*PI+H3Ti2B7hbfo{+9QD)eG`d9kH3ve;ro=%3tV0wR_7uurV
z=~GZ)@Rjmt-mUuSx%b)EM`v+CGts#Pb*Q?%{?fNV^}5KYR+y$%8ghcHIb+4$=VXSe
zs;h2G*!>#A|B(rV3diof`w)PrzlVkn-{Z?$P|A6a^=81vbQZb(T~nR2CL!V8R-eQ*
zu{kqgyBujv$E#5D`sy?S$w56G72(9kXg(%s9Ol0;QjlDjID;Lj*IZ!K8wlUT&CuwY
z`(^v#2=`Pa#V)QuEVoZ`1T`7zX~J*2{`q+cGOh=g92{57Z?Uk5v3G>>e32Nm{@2~&
zoOL-j^CrXJXzsui`|TeKSu2zFSMB#6)`h&ZU}CbI(uQOlYa`pnI!kdH)j*6de!QPl
zqN51N&+Gu-3EVwoeDf(!=;Bze5^XL8Av10JJSquwv4Z<zos6!hvS|vZ{AOZ7>{hI{
z+~X?6cikymow@;9sk3J5Gd|2hEQq-VtvA~yGI?Bn)wz--K|GQn{6tsr{{2dk<<rzB
zrk1x@S?(1Y;h;FpvHrDsyd5N}${6{SG~8HG1!4Zm%ySTwO*?dl^u8bitMLdaf)OSe
z@i8&RENG4LnoI1s27L)3;&pCP^p3SeH>xbgr$RE0J*qe)wz|v%XL_wK;pN%im;YDV
zEA>$|EwD}!@I_p&Tn~uP$|peQceV(iF-H}pSetWXNJ3|t^7*g3<5_q;!q-JEp_^%L
zCLGq>PYQqUQzlw;86fRXAglKhOP-zy7iKBe@p{yF=(^zs&fUKg?dUx@GL@IPREh61
zr-~S0@W07r(QN+S65q3O<kV-3A?(mg$<G%9t*f1J)Z<YxQTm?&8nUlRT)y;Io;Pd0
zo$AdmN;Xo&>CmibxIHm*!}|DBy1q$kKiZ^zJXWAXGrravVPCyorexSHz`MnevoY8z
z{>CmSrH9#h74dPUe`eP{5!y463gFQ{!u7w82l`HJjvqvP%8aj!ifF#mc)GOR$c@Gc
zGOa1~&+TovIu!om7+d5{+8inbKnZC-e%y%RkYnq^vN1Of>;mrI%=evsQPEdQUXBRk
z)0!h60^TzUR|qDmvOBu2yIlXMnL1{XVFVbS@&MQqt0YkU8xh?G@QRvmYqp|kQs4H-
zNr>Zjaj?F%csh?=1!Z3=ZzE}eNmpAgjlrYGS4(e85flyU`73HPdNBS=gm^J9upbp5
z7Ny0#y4&ZE(M@TA7=CdtRd&O<*;nj{wRJE{KA`5hoqg}jQ&TX1S7zTAI9REND#FpI
zm4V~I-dh>3{ZKpp1LJ9>`<0g_n73Cfw(xQFCy&kG`MO)%FDc(QkzglIkeA!4xGAKi
zz=o@3q1W6QGBI~#XC}_;53t#ZfonUDkK1&(I<nO<{sO%XGz!!WC$+qgtTgKXL2o#<
zIU(N%68j9lLHG4^!lyQw%>!L*0%*oRC$@>YI6ooEd#j>!kCUA`4W^D^Qnnw5+Y?G}
z3GalT8@)ynY{Y=6P%CGi+m3CPScX)jN(tCr)o{P#Y?FwTah%$ZAD9ZPcYZ2%V5l3b
z+=|W(3d2O!hVyK;%fK@0_I>5l0qey!l}t(BC56@!HS1N)0Y6se+*Ll?6i_%gk&0=K
zx={t)&YpoWKNYw4r}rMz1bu$x|MKPhN}_DqB0v&GYvtQ5pB>6dw33wa4`$Q6oU!WP
zWlvX4l}zL{H->YqaNK&w{86*>^29*4rw3nRxiiA-IH+YxM<>Py<Thr33ZD15u5oid
zp-3L(CLx^BiGhxmHs2|dk)1B_$7wSP6%q^N|BF7`0ITvK=chQa(SWvFR+goujY>);
zzxN=S7oID-mnci*jFCM-Q_fo*(lI<5T|g#@qt)3gzlwWhR@G+zvbL@9)2<vu(xjAg
zhfcBW8_!kUWgE6EIs8Oz4iv4_3$N%2+0`todg`3bZ>w8;=}%|*7J$JqN*WW<S3v-Y
zfEp^~nS>aP64Vbrf5odyQm$Tv{#(StkrGxw1OoX<`np=?+S@XniBv$usV3oJD%Qb)
zyO4|2=Kbe4MNLg@b!pz%NM~`$)RIQ3;PP=%re;8UoU5Q5n0h_0Rtu<Tk4e%^A0D1m
z8Vj-`^<(O9X;=E+eU#9g3KSrRYxR_R79F~8Fu&xE&b}EPE1UD*Bi}S1!rBO<Z)$47
zs7mr3IFypRg;AJ7w8_KLdPYJK%A@31cgIqSIMkbiJ-sId5pGtB9FB^(6&Co)MXvjC
z=wwBBNChAEa8bK6y`kT)yZ4DuX&Ndj5fNP@8LN7{{}=p<tSB6GbmGSQGysl|fIAf9
zv$Qm$eddVK5OezA;8&jhnT~j>=yJlArB>x)0QF%FA{8^b6ExISdo$joAE)m3_hZzh
zB8pWTRH>e3q0!$$DK>1C-!IH+DVF@-cl>Wa8T=kI@_YY%{5Q0wrR^GzxOg_jC<Qeu
z0qsm)4h=7vGxV&D6={u~0%B%=+00vvwoRMyd`4NPR&hw}@B2EV1PE~W0)(y_Zz|;%
zRQC>k69@4S=mi<F>BVXS!wkV-@b7DRkoc%ru~Q?)nSY<7Vzfo$gzMpBkNcv{z*Mbo
zOPW@$`yZrO*OTNTgb8Su0&b|-ry^xM;F0XB`B5onntfNsbl9h^5IS8f*<pLY=XsCz
zd{RjeYb`nkO&t($u%E%%Fqg0x@Gk6-dfC{xvz0?@xEq&?9A1W_*k5VHP7FOzVKJrK
zUNsh99o=w$u36oEmORn0vDSy>ZWf*sM%%GYJGFT2veQBx!QX9gMb=KvL;1=>Qh7$3
zrMAcyBHw22G{qmUC)1DxuDq+mmwG8A-<ZbTzS8H4k?-;=i{Ig!Z2rRco|Su}@eG<a
z<^g46XNY~O3+~4T%FV5tnjD&?f&mAq`O_ZM*H%W;nhtAAZFId5jQmrqt?U^wmBP7U
z$0M5Rq8TH(G7fxW?(m0#^+tX1{6PDdcAj|x7j8F7-%qabH!UE;*qC>ROLT;!W&hp?
z4`aFWKKyys&H|GcJEiN<{$J)A<qZ(=F#&4U5YdzC*qckVjG@%ePoM5Ublr`1q`HiK
zvC<c+h<!}>ZM0pnb39A{XMe6NR{i%1zHcOZfV<B_A%ErslRLz{HVq36l1kCEZhae-
zW1^i}%kL90T=<vu6hcpVx_|V!Y`2JGl;<BXQLCsV`{(|*d2K;w#t@#|^$k*#zDWJ|
z+YOQiJ3Tais-1HJ?Ow$Ob|KK)U~<^h0u-;aFJrEZNAj}>a2`JWJI$O7%!{eDBFHlx
zE~;cf0xPuG0x~V-qBEh*HHT969o?5j3^(4T1jPDg(rH9+dC|qh4)Jj_7drOb+LOn$
zC~)w!D7E_!c}(jQZqV^WK9%-Yg;L^_*VUQdKl&iq`s_L&rXlqh)+XqU9r|=ua%4GX
zn?>C^a1BHg6r?1h-k}K!g&PmEKJQ!<Nrq={B3>>`kVaOISSUKxsQ%69OKiAFd~Va%
zK#>g`>dhW$0n5$F=fKCkdgY21oPSA)NVYG4zioDlGm0iC^Y+v5RcG`pZJ)u9Q<ZS|
z5pU1%5WqKD%egiwPIjS9U=C}yc(icYPQAnvnsXTf(VoKUOBW=(3O+>}<fr}gP6L*6
zS+`@8*9CGFt(|(ag<dPL32cn2>yDVW^G2M^-(}fzx|})EVp<uZyNK)w?Q@Y3h=&XG
zquX1g*$j~WwLA{F*7u_?^UQ%8$m7676wVRgyye1!YT}1>qyB?Eo!R*qz}VxSNLFRI
zRd})K{J|Nweu2++Eoi6#Et!CT0NZR;E1l<b_3Hid$$)vL5o}Mc_kK5_OPsL8Dmeru
zfqK$s-;rfd?f?jAXy6sRdd1}&Xn(65j83OfiBWU_MS@fSE-7tu?V1&9)Qor>M%Woi
z+oA!lL;qMP&?jHty+;rdT=V6}T{t^6o<85Ho}Dz1W}X~gy!%<0zAEpq=r(k82P4kb
zSUIJ%I6WG5lB2$`ppG!FADs&oBF<IT2tcyT>nCJ~ho2e6JEL`aS^WUy{0@x{ePLB?
zIk_z#UwxfdhU3Po>+O6hXH#sM?d`2htvIRae)uA&sP^BTl?@j)LgisO+OIrHubIk(
zD@pk)uyRszv4;Ho_|ncxT?cH?5<=veIGLvg4kmQoHC_*O#r>OBTn!CG0-u4NLW#rH
zqh1a#>UEikN3bL56<9H)O&m&|oLYK(wUxbgm&qMGy3j_rEfJ?LQebz?&DPCibQ6@r
z=d^5mVu&$ZR*Je2e%8=G<K{AVze*W}d8x3VC{K5|X$lSa`xZz~AlPqZyZ#Y*`Mavx
zU*BJ`^+9TK$YWc?)a>bfyU@>Xa*Gvcx%;U$Vt-{Shc5w?uukya#jPy>06wC@!pJOq
zG89`d=keb2jIOVa{x^c*lETqPYCp8UIkMa1?pMF2H&*{!aO4Jb3K=hUXz-2uT|Va%
z{&`Pn0*~sdrl809A!YQYlRd{;N^K`btoXi`m+bM7uNI7~#=C~#1uG55gtQ~+(+`Z$
z^JWxyY|BMHzw(gg8F0EyqqlBW-@IwMx3|z#Vht!0B^Nush%*nMOwTI|vm6RT?<nc(
zhw_by5!lc_aKFxO@|D(~hEn*o;G^oPsrn}ln31J@`X}M6Pwn@J(Y!tYoue!*lk{|v
zNkan>4^{T1ffS;O3RUg5#;VyzeAk``3rpPN_?UZk{FC;-z!*$)f4e!0=j!XosHxS-
zXVBl?bSW!DgVWI8{`-=}VAF@zDf(e_nJoOj66c(C#Yv?n_gWv_1vWjfj{+nj*dBVz
zFI^p{RnAt}Cxo=}%k&g7{lUuzij!*@t7*K!3Dfy2_5DA8QKjf@S?Rp2pP*K(OffGo
zT2?9>oe`jcy?1ZCfVH;FF8lBN1|;}j6R2Fan<<A7hcb@WU#Fys??g{lnVp`V!dwL<
z5Rx)}bA^4>$T5IFf*f45yE_J`U%s=qUjS=mz1>Bmiw$}wzflXFaRyxpLaXByQ804^
z)W~~4NQJIb)%ZAn&e#_MN|>{Fq2YMhz`&sQV6`<5Xn-yY*;KeZcB@+8L2-C|oG`Fr
zG*U>9wl4&$u%qL!1+~1qJRqX|3k-^d$!d?6!ot8}B?3p=YMza9k5q%D18GM@Rlx~;
zFMV|Q=IiH+(bRn(*9@9|PzijOOMAz}Dx}`Td7$FCdRJN}?SIW5QVM8POmEM%{;pyv
z<9(s>Ly;-ra$7kdaweqkSvdn}9wLxiw)v3tQ7&kMH-h#CM5fm8%ulQTtb#$%H=r$|
z7Cb(DxPpt|PQyr5Lt`TjBV$gv&O!fp-^$+N#?Y1H@v~eOR@SAlGENwUXu0Fu8OsNs
z1OFT2buxb6qso1J>edpp-8)QE%$kq^i3uJ_1?9pu;j)9@lOkERQF_&(v{3R6AUC}k
zK|7Zpovx3$bc#Ho*a@}3SX0idc3OX=hTJ|E%p!AZfIo9*zOEfWQGOPaY964RHoq09
z9-#F1@|JZ3=lYnl_zLwStb)Qze#>zVH@bl0gO;p>5s}GBEqGGsg#tDX45Bs|BEzG*
zj~+7w_O_Ro7fxdU86`bEzDM;l;17z2(LIFO8JNEbjE=sBp!sexgcOWZ&O7sxghI_H
zOA{Ayl1cC*o!?)0{m`i3c5`#n)zkalv-H$@{u+oQgDcnev=5_FQwhw2<O0<>Au$oO
zRa<zxz}!m*dYGV_RsvikA&qTHPdct!@cs%$vN;UU<iMWdWTL{2aKhF+KHRBoW=&c_
z@+u1oIoa!nT5dH-oyPj2>YEqewC&^Reo$uRU;Ry-rP}^}bH$LIIC8)`r%fI{3|)_<
z*+VCE700$k#k*Y&&;Lo?xGx0-#U(tS?QlEVGy+N~3?Mj7yUgK%N0cFqfB^xuBuCoH
ziY4TUB#5uPhk1&&^&yM;;b9~axK98HJFa9Ey#{C(J<!Ax`$qNq&s%sfrEv;cijLM9
zj-8&(z$Jt?m&;OR2g+;>_`#$>u2RyW*PSGKz-M4)*6xmpJqS&V7#@55=3t6T`>w;y
zxEk=Z(xEgQ{l6?$tFyi;0Q;!gc~5x$Fzz2Mxs}_|cM-nerY|~AnVodUXWV5UM=bVT
zgN9!#W8ioOXD-Vk)|SzdjcNB?AexKu)Bd?-AI6#&Sxj`eX~|=n+_ToMYiI})P@;w*
zILvo21##<u>;E?ZfvE-brsT-Z^RpAoyUf}zQM2X@KtF#X6@8@&f%CCLLfcwGm<;7C
zMrdB%v8Bwsvg^Bd?@*Bsh?3Eb0aAHb^x}b}haTv-x#fI4FuP|4a6ssk%d`$JqY7yz
zBa9WlK~*;M#)(CHN9y!Z;!^@^Jof|7>rPGiZ~)K`ej1gPWw!63tA6A`p*JyPV4&P#
zbt4&JgY*tP&H>W{?chgyk5&H~xWAQH5d3nU%l;Yz6G>!lhi~7TUE^~00yFz?WB!un
zA%53m!r>HPVY<MGSTwI=-snOwPb1K?a_u)Xveilip@W37E(VfYTE~=;>^#~TxqX{R
zH=F9lBO^L^n9&u$27VZuyg*C^9|21+-*#~4Tc}nd8(<0hm0+p4QaA}Ib>G_BVoyuP
zpxfV9_>9w9Ur%xJ{3Wr&P#B+#C#E*bPf+t29xl~>qAa+-zpqx|6bf@aJHUQ~0TWUT
zFbi{pCj^h(TQDBfIzY+IP242qKL6x7p7n;XF~!Y0$Q_(#p)`u#(X6BSA4(rA{BgX8
z7>dj1mHO$cA>a>Ny;qCx3c_Rd(bK)=`zie;xU8hj?^ou%!Ek8Ly=<JAQdpeBc`E;R
zbG#JYT^(XQc@Ko#OWdJXR4xOjO>xpw|3)OZEYm7w{zp&m2LqA$@1_KaIZT_{kO)?=
zx4QiH?c1Z_g=`j$GCMPrUpMaY)f+b$`_!yLib}qWe;_9(#{gp@nQUt~*SA>p+p2&o
zE*ZsP3Hz$tuOFhun`T;EC0OQHudaakS=uHUJk9j;6a76^4F09T%U*)Z3Pj22CHKe*
zEn*nM{-P=g+rKO(kXT?m)JBy38wrek+@I@DHl@dgTeOXh#o!ye%W3`~RXR=|T#y#`
zoKUCRtuAXIlXSHTCU$6DuEP^t(wacIegadgNem2#K8(L4MLXu;roupPjB5{`2BJ;l
z`iysiS0Pr;@x@T8Mu@XP*ixKDSJI!o9w8KbF%^{3I(nvUt&+rHTpcEoe20}U_)l%8
z$&Kzse2ciNJW2OwgyG}A$=~~okAPqh!Soi3uKcn!@3S{8M8!DfwMXRWwZvmA-+VT&
zqBK@&i=C%jI^H@mavyD}x!Ft^H@Uv9G5xK?)(0&~Vq;Uj;Q*xWZSBrTgnb3Yf$k^8
z36QzFogIfdjDF~#;NrT6@%V8f)(KGRit4uFgyj+-6{?e;14bE3Uf$xmDxQvCN}Hub
z0dZ+*KRG>XNXYl^L{DG*Zx5*WlJIaiQy9Iw)pi5>1@*oocihM2GA0EYqKlJ>i{D`h
zzHfw%fIRxT#0sI{G#`DI+B#0xp~ZyOHX6*;!31>~WfXs5VPRV|7Xc`ZO2Z_EIGA~k
zgE^iJgIaMF@UFM^_Sy%Ezy=!Q6e#|e*7N7UW2%+0^4c@l6kN{xrW(y30|WmBBXDtO
zH-ai(3cTGMcX)qR{zhOlw=Io9M|5)rB?6NejWAdAjnCQcOU_Ks;#8$7Eca(Io5!h#
zm1QxKB|3e%?fy3%%Qhnw{&xz!Y6l0=Mv^n&!<Xx9UL8aCZT0&8r5xN#c_CFi7S#b$
zR4SKYl)+Z>*4|r~0+~t-;HTMjfbRvI-CA5bfed#7VUxvvUCk4@L8f7d77d&~*)hqf
zXvkE_2f8;S`k&v1Y|s#d3iNByBYPl}fX~+S#zsJD>VhjX$g;M9yH~3#0~mS51IzgV
zmsL`g)eRu*>%su@rv~T$FZcs2f<fo&;}D`OsKmt#RPPoN@&8@xZVU_v!{GxS;aX4p
zsh3#gcEq>6U^6wfXb^d0jkdUxFmLvA9Zo3rIvt*$Y~V@6^V7lY{T5{DeM$L<IXOAm
zi1FcDf)gJC`|Mvr!jGRm)w`a-_Y4=_RClttL=1N*s7)qx8KUD<YpV~8#koTYSZco!
zn-d3_0SZB(crQ+O28V{e53Il~E+~7ha%5_3><y$Nc*#20!!St=acpV0fEE(h*^SDJ
zhVE_%xCwG;XirUe<0QBwV5ML_=N`K;Aw=alKhE(VRkUZbj)t@j`FVNUVY+wfGp|LB
zq2HTAk=~$hE$eG9dJwl6ky_=R@6D!31A`_YEqHVu{r6~Q`c->Vd}#~`zHwiY#flPe
zjMdgl{tN|P+4_|ZPK=O}xa<!Z(5V#%w)x5aWTrWZPrlu~7k@(eWvJBF1kxT6NZM$O
z8ivqMK9HlCwmdYRp;6%kNg~?1JhWK4%x-NEJOGd&vtDC%kIVSlzvG`%_`eCW?lLYw
zyGj1(h@Xzv<u`4-r*I$=nt}`#&C-T3-Pw!_Sv7$_5iV^ZI)<0;o@>TvE>@y8#*Y=X
z#FuLu<dm!??c8pI<$k}~+OS3>j+P8SZnJ6N;+_yx4qu)BPLBxk`k9&{rjU%1Zd4qQ
z!m)wmg9LWjEfNyHP(OTd@rnl<LUa@ZA%sM^P^%Xl4@n>1Ag_-4`t>EGqkted{5yp{
zEi5K-BcYK%7QRdLqHc^_Ei<cyaKRYVeJ8x@vq~T1oAl8Yo^4IyT{}Ca)dm_#LKjmK
zjBBnGPb2y|9uYxdi?f7S2?;}lR+lRA?kw@!63@hKbdGX1#?v?Izx#Xac-xKpGn<n{
zL%xAyyq>hsXPlH8lT8|L!Hw~Xo!~NXg3WxLi&bzO_Z{5JYkSk=yJlUqTp<h({W2nm
z0!7KiCG^*S^h1G0V|@tUJ;$~6Txb>v=O1#_T<*#mv^cT9^iQ|WvJE^SQ%xd&wT`CV
zoI8Jcf3HUeLkCcEXG@%H4->4N&(AknPrGyO?5CvAjh4n#v!Y(64um_#{(Vd!Oqc>W
zjL&NJ3goajdG=L1J3BL>${Nk$^H}AZHF)SP02^7$*tli7-UmS|1OVkC^Y7CgMDUtk
zDZG9A)Q;oZ*xLcIG2~<}BNMmuGmN^0coM36Edy~<<Mj9qi!Hal{YKsPd4*+RDMIi;
z4FKJ6V$QmSzB@k*ELuC<30!{KsH83Yng2#_l_i;}tK)D=XZd1b1Q=DfURhFFx?PD}
zoCQEoh1#V7`2Vy{-0&74T1ZEK^tR?H;bdqQVoCleug>*`r=~&gex;22CxQkY`6`CO
z5<;P=tPSKpYI?te5hv#AjC&MWt8@A$wet@<bqtJa4x$SGeaD<sXZ^PZhdg2zJZ?8&
zRA`43LVj@-$ANiLSQ&z?>Ey)S8qS#WcL6RDcV*#`Rk8&qwMiUjkF`NT@~XLo=9jCL
z{%uIwdyuu;=xHqIkX&0E&23`dkLIn<-FeRT?&nO1zZs09!89srwVTxaMtBZYO2OZo
z9?DeNPZoK0P=l5?hJ{^>esHpI6%>-)K3_#9Cj$>o(Nm@iG4?c0V(|Je-}AffQ5Y*&
zYE9RWp;=l2vBy#G_r{i%Red6r!l=A!U}WIB(5RMt6VY^cHz0todAH1j57re;PX+c#
z{?Sr}hFYKAiPT;}3h71XxT_bZ${zCIBFU&n-4O4q;E=P0j4a{HXDR>T^3x!X{<}`N
zw}5qtEcz)15hrf3ObRxkOkhR$DxYdR{cj<0`|jBTk?%JKZx25)i`;fo4-dBwv5wo3
z6z!RZe<yyqUuZnnq0UWmKjSAZk$3M64EQVF)XJf1Wo0E}cXwB<$jj4{223A=zeYwz
z5}a$0o|+H;YVMlDz<8$^u#2pImuin%?z2D*d?;Ai>+7gEHeMK4Tf27IV|~fSM}zD*
zepxZgs@*M%BPz^vh~@?-=U)1ojHbPkvXV553hramMh5z^QuW8{s~WQ?^e@NI3#i2&
zhxr;B1_r6%uR%c%p=1@6133X}tntIc!)SL-Fwb33D|*ynj)j3#k>c0Xuo;ZLjuCzJ
zd%FG2UT2q%x36C)`6tE9A4;|>KMYRnaIey7PV&zL2ix#d2uadjrmfP=%Rt$Ee;AyU
z7$j@AhmHQVRL&1bH^*&yB!|l#^U0~<mjh-MO1`()^#XvMVDc=D3RSp%;>nF5%5g1#
z7{)j(0)io<6^f0`o+-j9Iz%QX77WwyLC4(T?=*rHnj0I}OKFbu4=kVXSFiu?<$P*W
z+s55a*0>tMS-h2t^yxdTjxC)4z7&}AGKO2abS1mY${d*u5_zD09fR<dm1^m}r-)&F
zhx+i(3wM8!xS+i}{D<#rw)Z+vA1ICxn)^jWy`S17M1v~>?1MT4LIUIIac(SinvE2!
zZQhLGLB`i3(XWS*>``m5X}3Dayx=h376Euw9C;ex8N>o^<>&#M<dl^8Q8!U2Fxg->
zL?j#=;|MEFNdN!11bM@zkws2fs3VgIrL|%De2NwC6Iz_h=}4y^$!^=2ecJQa-$K$q
zX5ixE23KqS-rD&BKVem7+J02S?Zz(~tT317Cz><oV&$E3ejIf>Q$5FZ>SoEhg-Ve>
z?2~-qy|u@cK5yC34_dYNly{O|6Xg;cl_}Z{Nlgl**eekRv9e+?0lWseMh~z*=@PpZ
z7p*Xm|D>IzKCVn1q#7>d&IBiI$Hp2HC<%qSx`vwPQt>j2@^nP^DEovPH*{C-8g9fo
zZ3MmD&a<XhR1@Ik<L&+a7Tb*boy<kFpKMf1sQ2hV_smpq(witwdYO|Y@$=qqIR-Qj
zxm1D`D1Sy1D_>D^l$NefO*OHQh_)l;ONkK)K$a5YkUpFMf%NA^wlr3*=oi3Puym-R
zovYON4l4n2&upfHOi)!zMMOm4UcE}aee8)LL+Ha{-jBUeo&$M9L7JNAjKzBVh>22o
zV<m@^ucTzp?$`;N-p+N`!cBWK9+#JuC=QL#mOCzq^lB5wV>XE#Bu{^I7H)I{hsK^T
zEr5?XSJjzC?&15nzVEWN8HMe6s(ZVsnu|UYcUs%Eeh2FN?ILS$GTy%XO|)hjydUk;
zm()JOeo*yxOkj&|jo(esf}i*0WgdOvKPN#fwsv5W{&&2&dAeFj9^@-h#dz^A(d*<f
z9{mFz*srq0qo3a)2f%qGf_5}C373D|Y00y2nCsN703W|A%)`s(z{G;O7Skks`j{zC
z`NpNAlch#ioC1nviSM1Ac|*H3-UWH0PL{v<Smw8QD@HRWTyhxQG&8j<EY8@AD^tdX
z*p-WEMK6oC(<=V9cP(8hdhbd!Y1Cb2J#Vyq8i%L;7`@I_B@YnNO{s58&p~Bm;Bafk
zXYT<9RVJR1QR%{7QK(Vw;u%BHVrbOx9YghA83D~Rb_aLWPYS&aXE<!;^F<Ug^erE(
zP$;Lh?>KrPHXAuUwlqtE?>h@biOZ~Fl!T&{Jbp6q4X<=$6l#{*Mvcd?N}oRo3A-?G
zH_uiocs65`qy7T%*{BfLSYIdRIlFH*io{p3W1*icx^oJ110aDc5X6wTKppAl#$sc9
z+N(a*BqSlr!Ngx)ZZ4$9v&|0hDx1e&q$j&!@47wmDCVhk;kWdcDm$-Xqy7XaMpUY+
zeWJ6kp=r@1Hbk4l<dCeD&2~g8bZ4)tGsmxN`M48NQOeLbHTEQ5+96f%r3z3fQ|{ql
zdl^zX`Y5VBB^(t?9o4S%V+RxcP#rU0DTp!RlOpJEyPmEHLIhrFe`6sht8@>4=b)#@
z>`k+2bN2!sV(;Kb6$Jqt<zn~C({3d_pE_-s*91kXR;*wmE!3i2pwY##Etr>LK9VMk
z^q_hIPOhhmy$d{dTA*jYf>*YDx7GeM@$4pLA4*JbZ0x5-(DH-f@?7G}I7YRovthf#
zS3$?<$HDkou05oi^|S5HK>jYwsajEBw4~#o+@OL3S2U2BGCbExR1)VE$=P4$_om?8
z%#(_fCMJ#~!S~-r$M*Uz__FTb=j=ufEl@q>9wKZTlQ)kdcoo_wH2F!_xkG}}oA8>s
z7Ot(@<frg*$SLA7-M8;*Xb(>EisEGG_PP^1yFQ<S6UEU@n=xQA)gWm+5bt)=`OkOv
zl{^)v9Lg@PMlfVI>%WPi)|!HiLm{eeqEwU-7deNe!1=S>Us#CtTrlqgf_+LNOPY-h
z<|nlqtJ%+s#?mdq=?7}?qkg$fxRHGJ=`@n(`IYBnYnX#mW+%JiWQ#X?Fz?|-E?P8l
zbc#w;vaQH-s(|CHt1N(C6Zc>p<tpdh#V)m#T|GU$xd&w2Hm}T&Ru9bzREY=(5`j(3
zxB!0Lr9<-9F=lCx5$(@P6>9Edu8Me@16?0?xCzLlt9%4^cJ{k}peoqYEi8siIVi{}
zY1tpY2yF=uZ>}(`3`?Aup3`UgkzN_6UZ7ZZ=)zAxhWLg3`Ssm3+3rP9J8-qq&L+RD
z_58GK(fdPo|0`SvEN|$7$>sUUxQRTQzm18^*Ym$mU_tI)!`(hm4y<$~Bh*!;V<Em~
zm2ce~7`;#!vA}`;#A0u`+-6k>+6Crg+<jSZHd|#e<Cz3MU{Q#6HB;Li*ro-v+&TFt
zI?GBXzD+3uMmtuQDHlpe+N|%Re_ZX8I;#VE2HE6`CuCWzPL|4>APC=V$TshoU9A;}
z1Ox;?Gjh&ip^XBgI8495SS+|d&cC}!soi^2TF}XJzrHR_@R_?H*41hmZni9y7blOU
zjK`L!_A-=Ho6a%?NGL>&<q|y;#r}joLe{cy!$Z!*u5SdJ92VoFvFn&sZ%Yf!DMHH1
z!XDgE))r1_ag6lt?PT=wr)kxf??kSzt)d=~zWb^{GHkNGTo74y@lzdhBl^sXLe$lK
zaq>qu`YkZ(+or_XVDgoUhDHp8&-~YkF}&L*M&5XsuOub;O1S>i6A3zgmH&Eozac9)
zX^O9uCy%#^!PTPlIT04tugTxCYrOa6llc3ieS5t>%ctLy6D=w-Uz)_lME-nw{S>B!
zp`l?1WFwKM_DyFjC1obv^J}WsEO{~u!{mO)*2z+o>M}Xk*ZOje)@koz7;GKv_tPHS
z2qWj)J&6-*YohBtVooz6Mma=~8`ZqH(KKs5aETv2`&ruC0@VT-H}@kG(10?t+pK6T
z1)%$tE)lXm02>fOGO~@;TbN~`O+3ZFezDazG_;s<hzA=FWUJq*#i}AQ_WZb*C>yY}
z6~NzBQF;D7?szP%gFRh+hu5bs{f<&*A4CrtHmX-~0%y8c74Qf<W?Y?&n)}U;O-8Qo
z#T-P<?`<TDq4)=QHx#-rjh*dmaEIDF$L1N!JSnqbcyEkAWOJX^%O7Gx&Rt;8TFUDN
zSx0_5ATqiv5xavr0Y;<6mg%o^wXZ5)ed-ArkcQO~P8xK7wzsxq0fbjfl_Kd&vz{Av
zCv)8m0y(5az|)(wa|xBXF#Z#DN;0@n!L~>mBi$*pQBt{-9J)dn7Zz)A;uqAlygh>j
z{+BvMA5PD9TTNyZDW3lLhxRgV!<b=jW-ep8ZAIzu16yGdR7@7xKB-K9CC7`3I@5yX
zj2LIz+9K;;Cu*7IM*Rvsan8cv%X1lC!$(|Hw2SXxG|_pZLK)seseYKGcRlvNcEKlu
zeX`wSna3^4+}Xx|d3-l8(`o=p)V?G{lmJW*I);+(s2r@1>9Ir*l8~rBX|S80u@tIj
z@!;g+QwQ9r2c{EphB(;S6*o{P+2lr(Rqu@<{j~i04+hsb@qXYkG_||pT(ScS&l4Fh
zh7vp2YiwTh$tdpc$wm$(d``kUUUB%iU*8}a966Ake3XMq82%m_qsA9li98b!$cjA8
zCOz5IIV;u<6{ge3R#vi*=>+lLAA*4&%6W;R!m=xvEmf`P)L;lOI4P)Nz1K^cx25|U
zY_E5Y)YR0=qb2ze(r7Ai*=>6~G=CV+n0Hq`VfkmkFi~}FU2&vrqO0(dj*U%i(Q<dP
z4_bm^Z*{0gwC_&LkSWqpVuhmfU2AKc@%EmJ83mNXx@HWT0wy&ducALVQ^_Q|5Ju6Q
zJGKBzz2boR^|M#m#P@Zxk;N<(0i%BJQ1p|D`+r(6R<1NwtY9*XCtV#DS{K%7zJp5`
z3~(9BdC8HlJKsy=<Xj|cyh4Q4extRoV{twO{TWDs7pct1@m{nHHN4K`YmO*Mbb<<<
z0L5H899vi8pz_lk)anQOl9p1vLC$oUL_dx*HdTNX;vl|cg4{@$OrQH1lpf|2DD}#-
z{U~w}Hjz?Axs-sZqb?iq&NT<<KgL;?8~4%h0n+8qa1~=)f#T__IDgV)>Hs95cbLz)
zfRR^FHrH}{pN%?GTWGAiJdOC{C$2Jy_*dvH);i(*W}zc)pe3fvc3aK&7K571yAq7P
zYgc6#!rF;uH#SyT(D{Lv$}hv9`HmRc%J}8w<K7w*&$4h6n-h64KC<L7@$s$@Y5p{5
z8{N@K_6NOxZx;4KR<Zzu*6NVB93b^3u2>d5@6pk0{*Q5SnjKpZBS@ftQ42}+@`3=G
z+<3Cvi;xvdGmcvqk*>SpOfj^46q{=``E3@nTT@}bvNrW%Z?p=t^d!vxV;MXRmAolO
z6f{;SMVB|@(zdsA6c+ha$M#CGX!`mlkc%u(9pNsm%Rt0LMT44k<(@sOQat#9UN(s*
zA)YtL-0Ihq-{cp{CJM7eNQg;DD7<|6G1h6Bx-VP(&65yOfGzB$%G%l_BxAT00f!IG
z$;knzHv<a1XZP;iJ3d~?fwRO07_Zvd0_XnRjIfZ95!}k@vWcFcA=p=9Z9uPH$~Mu6
zs06Y*9ITC`0x*#Q_vQBfKGa7AfPGP|jaTr%GYn-yUuOxzr6Rxwb5K+hvl|<NLI)g4
zWt2`au2<(|1@Bu6Xv1|_x<wThYZi`cN=W!7wqj1!u+ZN&8%AU(JECPJv=boEC|1Y`
zjSUtDxm1)t;P7wP??JM48VeR)7AMS1&+4(HIlZEQ|C}p`)yiY~09Fy>+16m0eGW9%
z603gBr;jIC1RcL)b^77hnX7ux{tm?S#Gx{=dl!1Nv$v-R8e-@$YBu=uc?`F0Oa7P_
znHv?nK+t0m4a<|h{>*~cbSzQuLndLM31?gHs-l9Z1Si29M1k3e61vS?VgcapFv_(o
zMLI4UP`_4Fjs-_sH#Z$kVteR<5OZ1ScD$C`S3HHTZ8thU&edtWT`Vn{r<&z7A7QWy
zbA6hYC(y3nh4%hZekXqqNTLwlx)p3aVcp3Wo^b}EB<$An67A7k^3W#j{f02>Y(MBE
zf{GFU#*I#{{oP%<-V*o*uV-I&r{%s)l2xSN-p{zgu~<_t8PBi)OC(apL5y6u8q22I
z1vPTTMJG`2ApJ$HVYJ)R`o`fPbyfOZ_C>|Jjj?6x7hWy`KmO!|PX_>NqY!0RJ?Jd>
zuKHHNWJvk4#~Uex0$GCYNbe~u-%@Ja{JqU56ckGV0_TC1m6hl(c=24sk2o0`(Z(RZ
z2E|{=G~fEqx3!>d^N`(iPzLWdoeW&6y7@!D^jdH4t&J$lr$|t5&k*yw6q6#SmXc7~
z0OqD3EbIvfGsBD2=@nz;X&(ANg)(YxHY;u*6$i2W=Z88G5>H&u_6E>p{ps16UhZo`
zQqpW_M$W3r#sd{4lGVW9ykbx35)MwQA=@%YD5;iMC9hXrxcQOrC}Cf|R;*2}qCCCn
zUgUDL&22sJo3tv8$<Y40-6GAit*tG31~%=}(gzp}b;kKiS!4z)St|K@%`L&HNY@Y2
z0+V!=WUWii&a}*YFL0)(3wTeH+|(;&E3+qlG&&vher*(Knbjx`&;N9v)2MFt1`+a5
z<+A8nhC)BSB27kKTuwLL?N4#i-P@ZX!^Icwmn9tJi#RM&R2VCNw7h>`zccx#dJ#fn
z7ou{Xh|N})8(VWeoPIE&;yy+`c6WEruC1+o-o{M={2BbOyFuWSU8mvN(ebf{<ptpV
z45CiQOQ|^}<i;AMc5CVARf#MoBqB-%#n0y2+7|%s%9U8nS*@2Xr9j$#=2XTJQwA3$
zdJoRItj*0OKYZc22of*7@Xh`Jy|N7OF1)l%3;lHiqyf0HZ+b!8Gx~&=mscT2gHJO0
zDf571C;y3Q_6QahR&S1`0KH1yOE^*8ke09+btl#Em&;W{frxH1Xu2H<$?XT@egFPl
z*$g@Io$x2&;YP1?jcM-Ud;4)TtTs0@mCej)^Lx>bo=A&j&N#}vxqDaLd9UA-9yy*q
zs2{#h`d!FEDiFa$wxnV&@->cWoGUhQB00Jb{nr!|)t+VBDMd;81bi|XUB*|fa1Jnc
z*jAyBJuD?w&M_a*xr7cBPy0c~lsxT~$;Nq<@-qtc&wSsZEfl(_`_kUto^Pv;AQHlR
z7=wB`dS$#%AOyRLcE7lR@F^snQ%ZabTnJ4I@4rt9fsbfvodOhKexu(GGP2u7qv4uv
zT%98)dn>7A$f+z$%+tTq4Ckm^>B|@(n#$%IF0x=0RV}qiYrR~4eL~l)${mxdF`ThC
zIu15F%Yp;Yar_x@+<hG04%^VbF4JIdxn$UmePFECxbKK~mIcqwnfLzSp#OFBz1sMm
zob1Z}dQYG|WwUX<x^;9(#=vrNj(*d~n7Y7znY`w~YNP=G3+9(SoCdOz{chlQ#|!V-
za&suzR~(m42Qv)rz5cV?{E7YzJVSLv$LiW^&i_iHJ*E>4=a4QVoY%Mo8wX-$I-IKX
zH${kf&~Y|whuETxp?r~G8=zs#JDWpu)SE4ZTro|KqH_eBQR87$WTfgO3U=^A(UO{t
z572n3j782J3k<sip%(;K6#UJ_ntX$l_w^#bIGz_yIjA>;57f5@fNfk>9VidH-~Psp
zf(s^^ofium+xBFwhzGq(5Uu7gv*Xz!Azaw_-OugQjOr(`VwCBtG||9BK}&c~f?w3$
z+=jB|V`^@GjR1+GSC>TvcGtdR;^Xd%cfqo-I@`;~IThIQ9{UO-$Byqi{uDxAeU1Cl
zMT2j3k``vf(TRHo&peu)nUVit`<5gbE>&E7{5~khto0EJ@$P6YYjRP>ie0&&6OK97
zFEsr88r>II1LX?o3cC?q!imhCBR+YZl5nY!tLMH~eU28oH11CN@<ci|6Wufb>WjhS
zBJrq7b=1}948YFN(DP8ffp{?JEh!Y4tGmM6DzCZ}fq)R-Ua!0|T{DyO{;Hrw%jGI$
zPsNPGFXr5jd?{o-Mlr)>z4dwalco{ziGw9L_cLB(q0x}Ak8mP~qSQ-E#C<8Bocgj9
ze^`!QyDYh3>qT?yVA7G!K5UsZ86Jo}2N>SU3mq{OqN(;5Z<PfMu4jc&uqvvT+E}&-
zk3f+RJ@OCaM(3wT!Q@fxVXTQr5f!U4;|337_=ho2Hd}CBPuA)i`rwgI`H?6kw$|$g
zIuAt$CD_UdaG;t^Ia-YiTJ-bPa|OO-0L)ISm-F4yXzA17;A?ym?p<&9B8Xzgvmw$}
zsB>s>_d<3*O?BxRb&=%+<{JJDWq>JrE8SU*FS5)GKVN1;eq>lL5bS|Hqhq0Hd(&_G
z*7Gm`l^Vts4#$N5A8&6RROR>mivtQMAu0j_0-}_3cdCFi(hVZrNasTcC?(R}dFTe|
z66ri3-QC@AH~N0x`TYL)&i&4v`~5M)%$ak}^PK(cz4nUNdhNByUp49tMiZ9yB^<Yh
zzBu|L*jBaD3EZ!!sNQ{dxfwpoz7pOxoGHN~)7MVfrj3eZbOWh8GBMN#&tz?F%Pgvw
za22y<o8}e(H&8MZsK#^G30n1v?zshAVUX`57i)m7j@3Q;RLe~w$kx2E`1>^-rZPZX
z-p;8hKFZyCR`0P09EnC6v6rI3!@zVOHQvEmoG<_zoZEggS|sr4dkuXAswLp$(2D&i
z6nisvFt+LzMEEY$eWr}O??U;qRKhqWW?d`S*K6xH^3k-&k$%Ww`*MB>j!uvE?P3W+
zdn_@M&U<Zm`lYJnPRBzfs_0PlRYw9kPPIc0NciOdDoevt7;Q8tURQ2BB=YFdBkoU&
zU+T}63B?8v`UJFT7pCkd%tO<^@t?l~Whtt+1>7&I0AE0z_9Na1k;5dZ*!fbFwk?R5
zkP>Xxn4BCP7oBbZLybs29ver!5d!N6?O&bD(7b&4hEh7dA0#fyoR3V1cpYUy2{yGN
z9nafPIG!9lradtaz1X@!K1}n<7Dh}VHf~qk;M!jOyC54#VXn20j&OVX#^l-Cn)AOt
zN3XmLUk1q1)0o_RGEX@1QurdqbwA6&v}bfI><Ohz;^v|LTZn;fh|zeE%new$&}%#k
zZZ#E^XmCvy@NMnZhs8h+F{;NP3~mAkcE$FfhXb#xO89-ft5Y4QtrMDVzcp?Q{4%b%
zG5|Kc0$CnQP|XWRf)g-XwB(LYY{6Clu32b$RTXCtJ`2UDuYJL<Ix-Zd7)h9FX(pIo
zNcC|Gq7|v(<HuxTidSx1>sj$xV(t`J!Yek$ZDRoo3dJ_h#9-??S#HJF0GlcdX$;l2
zCAnR?kT?L{fLdSKpz4JY5W(<BTO|(u-5AMf<*|)61Nqb7q;iX8*_fd)HiPb`m`@p#
z7yWxhK_zW6F~Cj?BjW@0N%t$D1(RXZGr%r@gJk#>SWuXJRxi#8n^C_Y$chf%CJHzz
zbrWdRo)^~c`%5SC<`gzutAUTYOM3vN*CrScK(oP)Fa};N(0Gak6-n;rIl#MBG&_pC
z>vx?CSf97=t5sR00F9pkVl%E17X<zO41N4O`vQANCoEjQyrU=Nq-#8*+-{RdG_=HQ
zEa^DHSKI9NDLrnC`;$GD@>EiaiWrGzF4J%W^V2k;+Kg5fzXBse>4xOpE%Y)yIlTN!
z*W~WC4gwB~_-W@2I!Z-%ashz`n&vLbM8{N>_Jfm?QNVv=1517TY$9pz!GMP*ETM*c
zV2r3)SrrCv*g=^<K&o&%f-8eqCXN{J&U}1)08|D8Bq{<X-OnJ%hJ}QL#2*7i1|pzJ
zP?10$-0>|SMZ(m|VZINDU_c+=LOXWzY1SBi-Lq5GHd!cG#CdQ09xVMR1kwV;Ea_9`
zn>&q5uf4s?R^CEY(&U=5!bzzRK}<+w0c(IToJP6#;JQ<2ti~%*`iJZy6Frw3>6t(O
zAMKMN3e*M&7TD`RCi{HOkECoN%6fdSi_^&w^%qZa3GEu*tsGPYggF3Lwh*MFa%7VD
zr3)^_LH>`%YT`Bl?J@)65wu~ytgNg=7igOcASO1|+cjk!Y5>nc=RGd}2?8+FdY1`}
ztv!vwOew;iXr_hOEwYZODNPyPy|?v3)ww&xA5RW%JL}|tB%hOSjok6;Zoq`;2f|^p
zBa6_PY=5*N5vn8-53C=~qcLTSGIQA-f`#7>`a+97dPM#bvO(ez3Z%~)p@P48y|$NR
zWN$&K<%HW%3j{SV;9Gu*QWZ1AqM=(0eEpglh=I`U<bj=4FNe1sNg*2iYygyU1YzbU
zx*3nbrh?nNrW7#yeP`qC?V=3zr>O?r(J8lZ)Xa>)N<HWyuvCpr;<2HgM92d)<Zl;2
zReM)R_joFCbBe`fCgcQ;ln~1i(NI|#>6aw8=#$!jdsVcf)5{qZ`8O>UE)9yu4%|jv
z06~34PA&|1Tb<qA3P95v^d;tlGRR$1b#E)Oso}c1x`)8{tb>$Fi3)9K`r)B12zM2T
zxb+$=lQ^wY14YtUy590{qB%l%U<ZeXKM@<_*%lTSUYl~5frAn@c_4(NW@eT<TG#K(
z?CS2O9UlIKFzUyCw7%{uf@$}hUTa1h!x-IEvDNs>?!{|1YDKvjLd##JK>#fv-kOq+
zQB{pCvX}q<^Q)NDP<Ri|q5<RTUVUA_FzW55YUBdh3E*1avps-x%GHj35?bEw;r-p+
z5C92RR#tA;9!Qguk!1re&#mj9ph>a+LAkL;r^zdR0R)sFtwrNNet2@ypi38ccCa$R
z^2~$zmk#4<!Xqq+(CeKY__!N=h3twbIps#m%uP619si+_le6RE#BrO@9e|h=?wk<!
zpSYmoQ9yC|N>?kD<N!kL-{q;v{ZL-?I^j<eUC<ck@#9o5(`z|diE?$$N1z96CZIYW
z?aXKcS`0X8bkn;zCD&cx;Q(T0u6B-T$B&8TKyssOYQY9^@x3D}GBsKD^_pkVJtLOx
zhfQyxngex-;L27&kbIOXx-C1oB9<Mr0lTrS-hyp;UtY19ZU%f}d3yLewBcw;z(pZM
zm~?(<2yQw3I-d_GXSv+ecrfiHn!+FLi%^gXBs7-W4hGoZpm+x@0B+Yo{ZJ)<BC4hq
zg_qLI^0}QuaWBqJPZ!624h_j$Sd6vA^EmV-&LM<<D7-W*N?e^#w}y<3*ZD8-_|8Vf
z6C{6x7*6~iW)0FV72<$6l;^&Rtvd{1o|_$JBV9FDr%BOBLD#C)_)~|6C5r1TCmiNM
z{Bef;aR$Xk=LK~pd<&xaUpO;<!gJR~N3|jfqoNYN;odGcLL<|rnr*B7nD0AfbR{5r
zc}lr_ZxE>#fDZ>r5gkRx!eKd4^|2|14%_fio)eh<J7Knb6(JnA+K7<E0&na4<B#n`
zKm?RXOdQL&h*)=-B=m`<HR~&u{Z{ID1dZwHMc8f33bu{H5dt)&!lpo4&o5r8^5nat
z5Xg8@Z_Ur|-y-#WN$zrpXRgtI0w#o6j67;gC|SV?fW${Tj$xVIH>x(ug#*2%D`f{}
z-8kj|g~^55;|}MOT@>o%+Fk?=6&rutio4FJSF2T+xh~h>34YeY_VM><m&6rM*(mzS
zbUvu*^5l%^kE%p?f`_DqkRk2!3DMc*RJM+7lawepP@y%(#$QZQ0RRjz#k}O75S!@}
zlb;suk{|^MudD=l^!)59$1QX$Wd+H+T0TBSy7W*rc>YU<u#RigyMJa=R`MC{F8{@5
zDVOQib=;U4y)2JdEZ`ugi6hO<gN+5CLy*l^DgFi!?RPjxmwHTw3>~*sX&<EgxjuQ=
zMy2rr=x*+GB1DMc&wB_6Pb9D9jsFfDa1CNX*5;$GZo$;Oe?FESt2Y_{mrOtl6h~JC
z%}O@+Kc6D_v_8iC=k;tV0BQfeZWMmB`^Oc!1;<~^{+)IN1U%tJ^nbog5BYoG{qtJm
z3F5zlsiJZ)1aFUGgB)md>)V8{G9hpi#iiNjiqzU#AoeSj_!HzH(0>6}i#J{a&$KPI
zO|8s2OuDUA%UXM>QSj(TN*(c*U6$Ht?nXM=p8<K$fH8RJ+LF>q^||du)SubJ<U#BJ
zuzhb80z*FS(g}INKxV2f8K22;CO;H)nd4$vCrl!6_Sx&sWr7CRpJ7@d->pTEm3+~g
zAV0HLXsqq{xzojIz53YvP-n;&wz=XWTmyDc50{n7j@cv(M6V_jb1%=gj&5M@k4&Ut
ze=t3SDl~T0J07|r(V?l37DEFJP$_<tJ4YY$?I&whrAQT6DTh*@I+`lP;f`~B`f@&;
zo4+t%IDD|Q#~o)ieLv#wvn#&_v)DE}fF&@}B%fs`n!kR2HSLaed|bcGg-8*$JPDWg
zCV-FTM62AA1jgv;H~!9Z$gNmf6$M>j8C}QPL9qFiteuPF5I^-MZIdYs6gTUYEt7}O
zvK4FWH@@|#2NcWwxombFtZHKuMUVP(F#^cOCF10e_V8?ad)ql}NGde_WTnT!$o$VY
zlg&uMN~4N*{TkMuu>aNGqItq*p|a?H8?QgqCNZCM5M=uL^+DdlKi`JJ$-#uq3f2UZ
zD61J{ax0>TMb#Alx~js*SXV3kk;fj#PQtm^_;&#XRp+-iku($y3^L{9<r6eTzE{We
z^wOsc7rk=9&ZP|yX3vFpz#C)oRBtX$D)sa`)IP7Y<dm+EiIsPv{?_cv;0MhOes%Ym
zeJ$0RoDJT&I_iO($3*-_Xmh*g&#d@r4qci+eD~ZtEZ36-1hdo)l+VRR@@Qkmo>fYA
zRsFoU>C#3AM_nQ!A`~adaiZIFKO7h&2fm!8dlAr<3XH5_)O$T*+<;OED!Y<ACH+Uc
z^Qk~J<1ERPga;I&Yi%Dl)M~AR3Pn-|?gk0IJACxT(70L2zN-gWK*nLMvERX-3sT9L
zBZD;&qxQnp%@U#scvKAE!2=;3Kw?H@kp1~+7x(*Z%GAp%9z4FjlM7GHXSo)NxIwww
zgxR76?K%08m0QfRq-yt=<sh#V1%AM;Yc{T&Sg0)pSKGET{n2|qvbX$~-H}y+ff!<;
zkl`ct&NpMA$4bP3hLr;csNnPIldu(qgi<6k3o`)b+AyX!G}N4(rB#}DjM6hRZUHyq
z2d5q+8Bu!#J2#-rx=q3U(ZK<YBJ>8dukP##!r}HT2~4s=uu4XOH<O7GzD(2KF<gvG
z?R^(YptXtpSDa>U#mTdo8d8w$`eFAZF*<!~z%XF4@9V+Waa=bgIJJ^&sTv!R(qCgo
zrT$0y0GWYW%-L;f`NhfJ`T3>|$@)mSLjKHOy(0csu0@@XWqU%z&o|zF&b%tk>R;*O
zaeOPV#gjgZ0P(pv+~{Wi6})hqTe;JLfK54HZd+NjU|k<u$~|hCzJtu~F!z0dpxUt-
zDAP=Vlf2HebKH*^f5zPi2O3R~sjAu-Tgt`0%q;XaUnYvnpuFg_=sMA!nYnfoSqSo^
zmnp!L-ud>dV49<hq1x-O`jPnLV2_9(tAts-VDE#1ymtL{XAQcjLuFlQ{9u>!UP{wM
zNj|zE3H)j*J(+1w$2&UoZ?dql7upOAJ&xC^Q*IKVnZqC71VIA<=*lN)JmY19O)*1<
z*OO=Gx&yj&e8dGA?JYLt2kbG2Z)c{Rmjt|V9;?(ZqW_&D`maD;`l|xl0}7w3VEghW
zvq3c{CKun+{DT9+;0E$WRv~J`1sU{o>YWoqi<@$P#~lDRw8!K5wYikr-y<?t<EnY`
zXEtO-iaH4XyaGG{Dxg;W{#m^TQ~USt|7Z-L|Bp5JFO0$e)t&!)H-`Q%#{A#A^MCKg
z;P?M4)A;|(WI+F6bN>gY`M;cuKlA_J@Du*0CIh1XpStsZ@5cYg-~Vx+{J)gB{}eg!
z`~T^V{lA@zC19<et%2m)ZBzsZw*>&Vp&}&}TI!6D2gj?1>s%ZlkZ|}lM*gpt*b*M&
z!SU*q*lkxOI~RaB-y$ynG{^=m3<!Aa^FZ)`P5svmMnGt%W<e}$(vbyx-rIyOJV#y>
z<hP^Z;^Kgfz6B{jxGbh#VKxBH&Dz~uo$%CX{3C}&>M^R`qcbzw3{c%+D9B$Fg1`<x
zNCY1P#mU(j1`ym^9EZ$!d$JlFEdBJU{GaP|8kn(nwIpskk%$4N0EnHfKC%7p&o(G#
z%hmF^+QCUp>AZgj`7EtFugL{adWplyZqb~Hi3^xH3-BR8KZg}Jb|{W+<tEDSXTeX3
zIN&z0<&E52#qGOQ*N@R(tW<tY3fs`^-Qer<-WXrHIcF#A^V^up7)8#7Q&esk#NI9N
z8AW~<>d5eZS##s6VXK4#JA;m)s+Ju=f5`69E;xh;V0y98bdcNDF1-R=g+p+hRJ&YC
zT-<+tVSz|4iNz7h%;W+dM0<S9E1S&HTtVaSG2iDlAIplT?>T=BD51s8P^Imqzmp*-
z;{2mjLhx1d&i!|xV`h<KX7={@R3i93jyFj&pP39he*My6c##}Pb6|r7!aKl_Hopzv
z0r+F`yC|hp;-3XZ@S_@HaWrR-1D(v$!BLiM=ytBbnNZJOa3&&Nt<oZ4=L$f2Ap=I?
z;N)<cIGweHhO#4gU~2EUIE-HUR%DfulCrf%()_U+Wk8t_7nhLF*k4klEGzj@X`Fvc
zYthvGu4-NpZ%5$2aArlyCe?K_mO~`^ZB6pE7Z@2eDl*I|Uu}IPGXm-A@)^YYV)H}Y
zsfHEHZl_gUS9R;4&g_|0LaA7ZYW^jW56!i~>>0wega%2$(X{}mhP+~lk~W|M(;~cO
zcg$KoO_ro7MJxn~d~mG03^3%%0q_cP46i-4`_;-#G5}F&X~Ghm^#Pt#{->o*u9C&a
z5lc%;n~%jY<IKb*q@XbHn3F*whEeh+8HNVy2f_hI+wo5RQj!4RMAz<-z3|069BFwo
zT&*AQxS-$tuV)vf9mmV7XWM^3%FHfp*?TTFb6GBFATb>!f<HCMx>_N98zx6R=;#U!
z4Ct~gVp=((yB&M=?6yUCN6X@(|Ey$W9<9EC-Y;mVOCBXPenSnoc)hWaT9=w`QxZ}V
zfG$_VJK424Vfv0)YvBqqF@XV(n_LXE_RCned>7WC07EiF2v(+m?r5Spfc#m0h53ZZ
z=I`R#XheHXS8lm0%G#7SS<(_HV)>n7M4b|fbz~%I;X0AQ>F+k?%a>+Jr}hSEoU1M%
zqVvTZ%Eh&2mMNmS6}t_#!>d=~Qhgz{L2_AWa=8dOsfNcV%<S^x3F&!9_s^}{sC(2)
zrIUh)GXM){ZgcRjhk{OJgSfLaX8~9Vhn+NqoMyPhR($6xt4wxoYx?9AS?}oXq_>-j
zm%D!A*nC2ZnM<GVA5qVn69dshL&d8`bnY8=PB8_2_?Ag@mzrq;S1&FC5K{&rm^h+S
z?dJ6R_s+Z$v4rFYn^b(vOD_|GgW#tZJ2YeexOH8x<cHL5MH-~Tz`?~_;Q(A`0O8(u
zK;(&+;w-GWzHqj$;sLE=v))gG)GCTt0?oT5)(Ddeqn_<L$oVo&<`eyg+QiFpbHBv1
zO$mJ7$@H6%DB|4j4plB+zr80=lZyQ<i(;9kbzu>?oTn!L3HfgT7^Pr<G%Yc=4d@)h
zd8#acYv5^xS2T2>?VkVZchzS9@c==nI^rl{o=)hr85MHO;nhBlT8W+d9<Ev^C{AT`
z8jhu34vp!#R8$NzWzSMpQd3TJ*^zr6o({`Af#zQC;h_CdZ5moJs;Q&^3^X0l>$3-q
zR0>fZJd3uR06ZsvI-8Iy63m7kC*B*Cp-~oF;!V#o&FOdPP+%03841@C$Wu0O&E=UW
zud}eXxN-%h!h-{%EYL!KC{ekh5+P*}LlNO+i}BM*JlfcP0-L8V{;a29jI4pdu2!!>
z=_??%GdBh8U2UGt=W{y|W2krDlty7u`3DQhnUidBJt~-Lm*Fg^lH0D)hYU$D?$bC^
zVtp1vS@T<>(O+Ww3o|UHb$8EVqN5K1+C+jTimU<cbS>M(Y~*WNhF!=KMc=aCo>#HK
zayk{)&V(nuVvu%zl4}e7TS#+l`na-c;i(#RY`Vy`|1?RVW_xh1;uLo<&Er2a6OOO{
z355aSLdBdNKs(*jPxAbu+!ELh!^nuZh?YD$#T#B`y)4C|=eMZQ*j(7n_=}+<T>E(7
z3DV!k3k3`eC=`irbBC6y<{QiX`^P6sNWB#Cef;-W$7aUbb~J5WYp>k?ia0ay-396S
z6BEZ!*OV^6dHn5`(XOHEdJT`hjP{sB@bt!GZl_Kz6ElDD6sj-~GA+(;8OtRnB|V9H
zrT-g$UnEN+6m}OC;Piuq>ly9%2_GQH5bq-Uu~|F!uRl+eEctR4IV4LxmMwyo-B>x)
zjmW%1t0Y<eAoD>s?9lXmR)uR_UHX*RagRsEc>30gzOBtrP!J(kf#9FZWiub{BK?wY
zh_m>`ui}`wn$Q}TakR(m<e04<Hxm3RFZ|$eSGLc7D_t`8aHTrcp^O635-(egG>|Ga
zl>aN$i=N;)6H%r{#NgD5-HDzxN3H|hG467mMeASr>lAuNM~jgobr~KIM7uH4$ceM!
z)q4-q8ZV294MsPea27HD9mmM+IKrYj(j7!G!pQI`h$E!4@Cv+Vdf#>FDx_$5{KTlQ
zd`eAQrv}0E(?}gA=@U343W<6Xd(>rd;|{mqzv6zwH&H1+I+E{SDEsG@v-`L6xiPJ7
zIk0-|rY}+PYqwi#EveNsDO^9yATg95GpeTMJ-QZA7TOY$f5UIN<poL-<pA#Tj5@sZ
zm1BwGm{I;$1GIIng5%N2+={P{dID~gFMXiDhA~Q{5q%am?V?u&5I{QhP+{jakHhm1
z#xipDAuiQ=0cErz&U#UY^68hgbw>wKd(Ow9pS`apI{=UDuwA_{$E^irFhy@|E@3b*
zY;dh?GANCuB+ACoD65P`Ov+@WfAFg%^{<(RMf5-TI|6;%)EF`X)zoAVB;v-p&Qxxy
zqW!TxF^6SiKErM=wD3J_ll<&yY1E4^;(hnsECC1cR^y7mbGfRWS76VcX5;M04Bwvl
z_y(zy0kvpsAFj|urpPJ=N^o~i+u(<irNzbhp8SyC;b)e-o}QL2ApSFnRY)MlK;CcM
z(j0DBkUC#3T!PNcTSW4wN|l#CzT!GTMLx|{zOX-+iRxV+|B(SAVeDKwzFJ9&I*w0O
zB2#ctZ3ZnEU?G`mJGSCTun(i>!uSDi+p%xk<RR{ItI6N3k_+NhS)F<<t$z_!+FV5d
zIf;vlDZbNjLOwlxxUMA(@;`?^`$-|^kE;*$qkH^=zeT}K<c(xQYWbvk>HkLAe_~o;
zU0WM@04ORxz_7J{Mc+Qu@`mlr>v&DT_yvDdcYs(V-~DaIer0nQjcY!cXD%8{#mz6I
zl&hE`BxItRRpC6-uSG<_7EV~2Y$GNv1s4^#$wW3CeZ@>KowEkvf8rp(ncYYO39F5q
zYrsqtwlR!SrEa63SOf4|nQObWa6C`e0bVH1li9*pQquL2it=r<86{75_g?xxV2TbZ
zm{cl*jg#$B3GKdqaMr<Qs`BFW{#CVk&Fz^Ny7l3a)(>ys`R1mS3aaOT<?npxopWnS
z^;PpB<;kcI>@?(jn&_Z>TXRsHLcXprfcf$5jK`YA;F0wAJ35?KvZ~KdD}_5*<7}oH
zEOz*>c2CgG9G%u~w(BMOrffI}FK#FPPaNx^t6MsDxcsg3XxyQ+#4>xE&MyCXIW?u?
zVK_$UJadMo*x*d@xZPoubNxOn5#I`mn#SKN0uE5t$d2SA!@isr`93OkrdJof_DdPk
zLUM1QSN(6q`$AJ$((_j`^qIJEu+uC)Y7N*pVD;C0!?3Z~ldjnKZ9T;{hnc{Q_cCTO
zoj4+$@J<H7bZA3OxZZc^baBG@h#%CzON!0EB@=<vgyK~Zte2kTcKKwXcYpBQ#OqCo
z+p}8uyg?;FFjBi*s_HA#`@Vu$IR%BR&lS|LDO*G-59j(Cr~wH_Akb;<$7B^fcqt{q
zn|atR`KnvpxxQlf_f4k=;<g;QOpdUJEu^@X1vW-HMc<7on8euQO{0}3J9;2z05|Vz
zNFP4LSX+L+27Qf{^MTCgS9a(Xk2>q2cWlP%_lc%)bK(-+r8oHF(|zI1@p3%3V@)=i
zjMoIGZiKbzv|xR>*YptwAzNt_W2IC#MB7@u&(}tD#`rC$6<Lh|A;#J8xb|^wWm%D0
zzKOtk%%)XkP6KPX*tXTlOl{C$k;&lcd1#yvOC)rifiH1Q8{;MM-DagP_vSN4Ny}7|
zVrc`(%>#!UP}cg;;7*RrmtIQBje?u{g4Xcx&{t1yM>G5I4}MnsRm}Or_~EAD3EWxE
zJjp9b11{lQ=ZTSM;e6#sg5!Q{wuhsJc6Cyt3U*lE17b?nkV`b{lny$f!8A59F&PNr
zmx_e)C)7uzPAwu|d|XJM@XGMZ)rX8^A6lRoD@f&NIuQj04qn)-;2Fl;{xAY6Fi{Ao
zam+V{<XIVFd^BP`NN6)cMA~E94<g2N2h>Xqc<LMFI!(&gy^$OApgqT)P5q~68wH(m
zeA*fMdPbY0EJSMB#<=dsd6^8*nfyMJc$x_g?_^zT&vQOa0MZTlHC!Gylz;o9O)`i9
zLc$pWJp6b?(~^d#8X2oZm>3b0m}*59b=|lDAN^^Tu78)XxfuJ#@(*Hq3U9PNU#s5?
zYF;kTTBw@8PO7^I4u+i~ab8bJ?bJ~Y_X}~qdxVaLD<~xUM7a0{+SGjVNcO$L-6yU7
zNdBo*&vg)mA7-39AoV$%b}1RVx?+-`8Z)c-OmSlKz^`(%HpWT+<k|ML!wOR6;f&Pq
z>XDWFz0=<+S^~e33oz4DOZ*kpE7`X1{!=mJbv)bSS)A8jL~O~pd_Y^Zax1l>fGr{?
z?$?Hv{^DJ~?a83bJy4zOV!x8jB&{&%2tPY(eGa`>ZSd0peGD^ZZsm>w;guOiU{Ze{
ziN{RkPLYkszsKBraqjP;k-mM&o%kUyW@51Ax>qfm|1mSm=FY|oS#`_b(>5zGA1h@H
zlslk?(6P+s3t5+$@}h&ua)qN*Y-alW@xI$TLEwb8HB%S@XpzpE8vo+smYh0Cm_S6~
z`pcTx!E8S@6*U;USx%h`#lo3DWUK+CjprGk`od}iHe10LxI(9F0>%I;senW4!fB7L
z@#sER$?D-eDcVrUb7rl}^ksw~ZgxwNgZ>Z^BFdWN5Me0)B?Y<9<^1Cp-d?L?nQYL@
zf>pn?<N1rJ?95DG2IZS3%ZZAx_yq5oQkH@5bG&@IoFFM*3S$tEktu&KCo?r!Yr?67
z=A@{DaG8+oQ?vHzK^q%PhCwEG*{XOSSPH_7AX%HLw+S^AO+v>fAPPU)O=iq}0z0Jc
zH>FwNTXFe132LeT8S|d@b3{msl>QUw1DSL^_A4)=Ys`6(3dK7`I__91Zns~G<Mw}c
z)99kzy(v~8W*3gcyuUA1zmp6fWcxvmuI-#S^8FF}H=BaiEJLdn8tkt!GArFf6?^?3
zaPw3H(!B3DVAlG|B)LCBMfF3u2t-3uAQW4oycs~ajJVUqGHdj84I)ynYUzLY^chC*
zPo%PrrL2=P61+!Gf2FjN*=?VX4}9-MFvdYB`|58N-=R61>gBDlwX*o>LR>>b_ay$G
z)Ag5l?>v;h1gPjL;W|jV`2hn6gLWOfEcmGjhlTD`D;5=gdGl3_m+wTLU*(u(rvCB;
z-$Yr3N&IB$i*tzgjCODBC)i96`r)WAG@4ov_Z5{)dggTPTc4%1-1%lTbL4$`xJF5E
z_Jy#o>cn9GXuO>I6pgxZd|D!*g_<v2xq17wtuXg;;hDGPjavJUlbIBU33h{vt>3ly
zp8Wia&uG0~5Ixz@S!epf7D;l?R%wO)H$E%DuL3X78dw`bAt7s&!QVBZ`C{v%=~Knc
zGmzz0Q|e4l|0J!Dppk^fp(zsSsd+or-3$NZT3<mD%NqCos%VX&a+Vun&U@<Uo9Me1
zN&?duo8uYZ4gP6!@o?U(hV^=drWNWBj{b<@WcQfMN0^$^3Bq2TLnpuv`WDd~B}TT{
zJ=?hgYUs}e^f7%BgaUqYg~m%3wkKxhl{~(TpWS}<^x0i$3gSC`IbN3XOqZb@wKGbk
z0;Hahp|q%rXN39c-f#KMuR-z*1sEPKFGvUTE#A~2UUzsGhUSRpCf^M89ki9o*z_fP
z9vyEb!<kbdZT^cTj#IKn6^pr_FwJ@A>Zw-eA^8&me2<p4UHuX?X?{Bd6ciyROpe2i
zpqhZM`<ORE9}DVy{S4gR^`z)>D&!$;SDe15%d>yeM;ydigFcaeb+LCoI@9bg`Sg*2
zI^DD`QC_b|?f_$6-%^l(*x!x$`JrK(502fV*D7ROJboVvrQSD9#l6;%rd(B(vBHCr
zBT#GM4~Q+qBc$7>b|ekE=3Zh3%genNnxRaRZ<>0bwY#RzNY-j=g{NS@9@aCrWj{OD
zSn-l+lHV?N3)LAn(A5oww#{ukn_(2)-wD1ZnEM`a_8X6o00dMIR(EvU_2&hHcE`Qr
z@0{G6jpJrGlSGeDO1ZPUHRVU&c|&Lh^`pewL0;l~my!1yb3ckq*wx};i{?i%jcVj;
zb#oOk=j0#B5N}d74m5d=$?%S6AJl}{8I6QLy1K|2@71PMpYcRI<-IR%7YHPST)sx-
z@8%}GMki~eX%+sd!^pT@hp)GU*Lc)KoxHB1t#|T-V0MJjG5r;sn+U3G;0oIBq%N(&
zaIs?+*4JPEzEe5jSs11twz+&qe!?rwSO3Ohenl#dZeoY_;-pb1GAX+2REe>_nLEwm
z0b?=F;UQ=RYt_cW5N$rq${DSrO@hNpn9|Gj09PxG=cPi(C;~q*ahc_KYYEr2Xt&!w
z6YrFUqFBJ1H7v0pBN&QobSXxtwVHUxtEt&HNXsB;IR!C*1JOlfL`sILl?Q{ir+fiG
z>?x-Bmh-x0{&~38F;YaJxtMbuHubKi-C!ac*gflg<N1AbDOS}aIb2+>WxD!xJ;#*K
z3|l<!P&0BBM;v`~tvvCveviO=?(yr%#XnpENMTM?NqT?t>Djpgjx7-ZlJco~WvS8~
z22t<Im*h)JwGDd4byCo|;Y;IcCu+v)WK6_yo)^ib96lKt-4|{_M;F?y8lW*Z#}(1D
zGBuSXbOsWo41sPY1k;kau{W|QE1kUwjtkQ@eI2e0WCL>_Xr0fHL3i+460Czz1@<Z~
zXo-!H2a#W@bzs1gL_1E<(IX-W^IySTiMZ?NQ}wG`b1z`LiV}7$cYoPP_Q=_@nnvjZ
zV{~uGGl3l%98SW=NOSB4ucx1(A@=Gm<Ebb-o;U04W{@SPi!ZlQPB%|*<Q}2FX4|?T
zSvO~QSfVaxqVFu}o?<cfK$X^l(j=rtS0gXawkN7Md77qQR2gD*^F2WIPZ@t&#F~uM
z(4gP6aG_XXS)SO$S98!_ma}4Wz5)-bb1u0&8r`+E@!lEInIt|X#SHCIlroluD8rj7
za~SQh^154RCIaA=4PrOJM7=wR|18|+qFCAzidNzDypqszm*Z?G6w&R<Y$rR1ux$VN
znhN^`cCd|Vz_MMe8MmW~xrXTn{5$gTfLgdUXlp>H**kxK!F;Dnn!%M_#{6(sF`#X1
zJy!6P_B$}BCMG7JXdma+R_(klJ2B^7e(C;}_G9aD$JgmJeviYx$n+%(t7LDvX?`wP
z9rir+(Pnbp-!OxC2!JITCXw|#7QgDyVVK>XZ{9eW?iqS%!6m~|Z#DTlmvUmcTlW%A
zyIA)C_w<F>)qqLkkPuy_edi8$r8b*K<#t0<W>eFOdGEiTi5casY`zc;lh4kBDWSv0
zLhD>$^!!$N;Zw!dc1X3(b&2yO%sB5Xwv1OG&gml71h>8W$44Kt==`vf;=l1-CMbCD
z7RUGaeH-p@<P8aV7o>%iaHSyFL#*lgJj096+B!<Z1MhijUz8L{w~m9^Gx>(AM9hn0
zc?Bi^g13GmX-#7tUsrb-MYP>;7$*52i{|r&e|o*~>t3X^djuE)?UlnzQ~PZSE}DQ^
zX#O#h+m#UkLG$ePRMnBcyRFg4p{vYoMbbakmFSNWA*16^Howa~F{2u?{*sTOp;tYJ
z^5=c4iu#{BF@ZKc46lkyUtT;34~GRM|KxO_4^inIX%XGH>o|0<KP>W5Bdzhga3CYy
z)K<dXbt~e4FDO6`$eeAEMF$~t63jOv|LC8ovyOcCBkbWy$h1y&<#IkbE5XMy_%XdS
zbNp)Qwz^u-@u3JFLPEV~TPne)c}~KmN%+jypTqad!}$2tkDM?15;8uHSzD^V$r?=)
z5N$uywkat1{*RCF#YU9b(T6jD`2hWNZU-}Oyd((t1bWNR%EIR2+vo<8s7llJqWZe(
zYD<!MBDxDI-!nqx%D$Wu=o42&)``<Nr#odklaBUhCiNb8-_Q>^EQy<`KOPe{p^TFv
z4vIHokcf9c81`1|+r36iZzsoJLAWO?8*@Th@2$67H+6Y$>L$u_4wSN!_9pd}#|46|
z3qBy)6+-XbY@PWm>D0PUAj{{V@g8-;0|e0FDl3jKs?xHH#$j6e=D{&TX8DE+tpxjJ
zOKd`oSZDW(hV;AIGFY*9QEfCU3j&|@CncJzP4M+bwunamFk*^(K$KZki&1A?%t$FJ
z8te0!5!iQ%X*<8Rq%?^Rur#M@NkNyzrW|6b(Abm6$oZ8uRrS;l!yebRUoD*KvtIpO
z`-jaB_8UHs_Imlls59xVJx?GF5vL5M{wze-`0V~Q(58-z^kvx2NPd=+3YF!I1H((y
zL(8HW_V?d`8}r^e`6=ws<PnSd7ClmO+}GJ)rQEiOsf5%LW==IE^R4Wbnwp0m^bno8
z%CPHDO3UQg2v54+O8lj$D5fLN2dbQuRG2>yR$g`&dh8XbCrd`4Y4_TVaI;e8!XCJr
zjft<HT8Q#Z)SA4bRd%Z|RbSbXri-0RhRc;!-aAB@*w`h==5x@b!)WZeG{y&akP?Q!
z5;EgJwHiug=zp)E^fQ*IMx_+`h(`pfwvH`+WyQiQb~Y(CUaLJF<1P0Xy&p`l>yVmU
z74LO=_qdI6_xhVeb&4|9w96{uxQMVCpY$ht_7=*zro^h#u$yclxDfrQyk^0IiV3wP
zbi&8tuGVS#aeFaiAT5u0rD0W8?lFTa_cu6?ed494A>Ap@cWni6cToS|MR&n*<$`>2
zun=p9-|>cAbw%|*4fxxa5?wtDBd;t=u8acFoG~#ybCHr0qMR{h)$Z`1&yLbQw%2EL
zUS}m1U<_Zw67Wd#epBaGQ&VMSUReNy9Fu*y3GP`TweN`E8ac>HRyB<rdG27=S2$tc
zNHSJ|>Il&s6$!x)+?-o?QR1MgO~b^CO>_+BJEvStyNs(VsNXH9iwqG;s6W{7)Al`Q
z$d7(+S>Gn#Vz~^WYu1UA8m#HQ)bb4-!xXf^bDlAKNyy;N?9S3x@2{pyD{CD%3EU;_
zJN&<u5TY)|*n8)6-{mI1Da~X>XCnCw-#Ef>Gl$S}D+#@w%y~bkd{GPR8+D9#zSA1n
zBv8F;LxK7@c<B1t?`*I9{^W2?R@;zT-A77?+M@@;8Tv4hT`;~4)VTu6!uQU|W*vSR
z_?H#tsD#HPvvE;cIz~y;xZe&|3ivQw%$XJ8+*Sjy1^kBy`MlC7J?M5tbqGoZvQAZ>
z??c*jv)`s_R1c2hj2PySM<YQ8hu;oa@=EXTlhUuF2Jv4%T>rFBkfGF=$*U6Cv>z0w
zTIoPCBlEPceCkKy&e}gUnLif|W?Ts9fW2}V`nbU9LoM2N#nESmnnvuNl(3VIS$YpY
zZ!{$-Q&jnh)ID&9wU5S0Z)6k2^bq1_H_IMZ2}vz+%-2V#e}^=?7v~38XR#l@bCQEp
z*I*zd?sK$&K-=X}YtnpqwFV!#Ogm31IFB$CIMYCO@bhW(Dq!S(mO3dN`K3P@NvZO4
zGhV{nG!+8B$~jA)+j;fa^_xpqU|rs?n0+WQZ<kN^ZZk%Ztlj8!j^ndGcnFp6DKf8`
zWjmQ5Ea!0XyC3wU#9_HRrQW)<b3T!~A4IF@SM`AND_d5MIyrL}*WPp5U17s^lKIT;
zn$NL6)i{5u&hG<uR`%?6HsdKQ;GC!V=+N}ZHrmVP*{)tjZm+?8G!aRsFzrh7N~8EK
zCZ<WwIX0Zht-0K7AJjM))OeJGKfoVsW&XT-{oX=alW^BL-$m8>Srje7N`q0`GF_na
z0v+&~u~Q0~zA)ai;kqIa{?Tkc$Jl@#crUFeJ)E>qT1i0iUwc)ReH#mk=rTU7av8VP
zl^;@AUOYUe{#ZcD=f>2%ea~IGTl;`ADeoL^c3@g#_W{xwr4La;Bq392F#7#%ttj7v
zqN(Z9(&&oKY=XnQSuue}C<t--i|0PU#?LYByXXobpSI<{#aPB##Z&U1tJFCbz2cpI
zra^e>{Asu`;nA)*4Vd|378(vqwwf*(c?#I<NTF&GeY@ENw+Y7X2ou%7-125Ob|pq`
z<r*F3I`yr;MA@ECh&sFR<hS-`Z>6rPrqLMla}d>(SoCQ{ES<Z%;QThvrv6-rEoqF&
zara|CVU_#kTgBj(!1O#b3&@6m&6$sPOhc103Z>wCDeiWF*NPf={HF)UF0C0^aZ?#U
zk9soHrsBS@VShMnG-Op#j#5SO(7+yn1C^Z!;_W_VsAq2IgJeN38HWZI&8v5Za;%fs
z8D7`RRYr5)Gk1+xY}wduS3&Mm-lf^U6U|u&JcG9{Rm=4W8!Ts1O7OIe^)=llt(2pO
z30MiV5*k7-Lqnb8Ojrr(o##|vUtKAjea5$zH5Z~NL{y1K(<YnWTx`X{hO~de$Hq44
zX)-07M~CE*skM3Y#0gb&)e$9m(b!Nd8cY!0FmT)$T}M~WD54u;qIrG)5E*3M9D!)%
zK=yRDDTXB4+?kheK~L~<ic=*)f9$C~8zUEreNJqZ1$}RL_-GRRa_dI}eBOXSN0_2p
ztcW&j6ro$j*HTXK4w$$dh+a$Xri_}}$z=~~z3qT5p9WO9&EDJ()>2SVw7$70?PgEn
zgh-Le$*pXtFp5hoc4nVH#SyVq#mBDK(5YB!DK!APBBjJG$r$m0(lBH2v209jBS2CJ
zG1P8G18?SO#Jgw|mM;Wpx+*7A3ryXu?}y>J+SpV4Qy39?IG+~+(5IlqAmN#uAir4H
zV~i1#9b|8BD%w~2$NM-2alWXhrRzZ*pHI)4*3BGk1<X2??UefT5FX&P1o+K%>4EU$
zY&#-qb>N4hxBMpuyANG{EPGaViUIk2Wy%qn*?E-7@$c{Nv^TZBlvR2Iym6Jo{Q!gZ
z-$aD%DXkD$`P1LtQiQ$yZ2GpGpQ}RTRa%<`Mq<r?2sthgF0lE3lE<&!X>VCfqxLZO
z??r%?8{b^$nAXRtyU6VOOPpSg&TJ&~hS>y;{qpv1ns|@wL^4yj-WKbBK;YpfyxT2w
z%C4Az{_){aJ~pBTcH-5c4a|;vbv+GfY6R?<G^hvR)M;pW2+xp5*i4NZu-b`T<8?H`
zfzF#GLcW&1teVq)Ku#WGmO8YdTDxs|G+%6hX_yo{*pB6VmYSP0_Z53|?s1X8T$S1(
zeLwzBm4;-Z<dS?oBGuD&wO~U7_W#5pOMb;Z40_H)6QJ3yt4Lu%LWl>4kRs~NU2wau
zr&>)7OlpyU1Jp6X#ZtYob#e-VU7YMg>(BK34>vI$sQt>9^K9uBM^F<M+1=`500>w{
zo4kMUy_ePlNaq4q&qPDzXBQS28SejZ%w4c^@bEJE^2+D8oS;6w0QV*ZzvJe2l#8Uj
zA?9)>s&_pzxYBA7s^_ZEi=?Z4!jFZ7Dp?(Ki+Z5rMv4!Q>nHqff4`X$1szxUdbIP3
z6S*Y>>ZA4eEqaBqnq=k@fz^B5e7x2k`h*;N-T9Tn_*u0u|K>rt2+|83Y(){>btatW
zNmg&^h)3H{z2X%cIkV7?adR3FW2;*}D?Qvsf|h09YYFQb;7YxLyfy;M*WZfytLJw+
z^DCC2+8zlt)Py)S<S#Q}vq8{wnJV8g^d{uTbcGeXS&DSm4y}vBBGWC5!Y<bJs$ESH
zDb+h_3;hBy5W~+L`O;l?pjx7_1megEA|=I)>dpR^^-_c6^L4welG%7?Sjz^(n-~8$
zm-Ee}__ePgRxdl3Ky>V*{c^fHN1v%U?%8~RnypLF(Fx6n@t^+su{NZ_?({-gYvjJ4
zE{KfKB?c9_d<9BKOG66==*~`a+Q&}TM3s;oWOk1ZXB+X7cDHC49^pU97JqF<<FSNK
zy>K{6`-QFzfG6%m<JdJiiNZ~q9w1oC4Vtc3>f3-=qjDS1>pw40%xKk^_8EiLImoUP
zn6XdstR?7`$qDNvd@|OGm}HP0h}ftdYhie0wtgdtvm33;Z~R{6XRBw<wPXn720hT4
zUCLf(gnmELGDjw=Q($z}d~g|pGLzLb{tRc*d?BZk)^+fH?2nveI|mYkOnNm9!)J9`
zG9p4yJsPjWUd6_60_Taq^+91r<;{5A=<nWO5{BgC$x*EU7S{&%Hw%G71L%`Pzv!st
zGMMQRO;6&jp^5cYwsG@wg1>cvL4o?+cPK!QJm=z3H-B3_Hu@dqi>>#exLm2tz5P97
zN4y~46gOv<G8xN#|9~+Nw5Vy({~+Ll=B}?n$QS-akdSzWW3Gaxa2EV0C#Q$+jNV8|
znov6lx3S)TkSQ9cG&!70sg`foK!_6w%hRsu)2u=_N=}6hE>64}<FaeNZ{^Ftox5en
zMQxPuKBq+LrN-Qd)jgh%DZH2u*ZfnGqsu0F)r3DM9UpI3aV~HcWbLp#7rcUA4GU@H
zbQZ8bUFn~kI()7B-K2miC&zADyt`BB!taNQ^3K?k1+I=?T^wijX;U-b*W=SiO1~(p
zSrm*L&717I1#p=>@5Y%I=ZcMgoweMvxv}#iBvhX_>$`m-&jKGMdj3E2)7th1srEIi
zSM%eKA%jCd+ec-VGWa8xAzZcGZw*m+qqbXSH_r%WH`bb3-dLE0*PF1=@@-vm51KT#
zJaYTG7T7Z`_{nW_bzShN{qfK0kz*(S$<)-L@!`3Jds74tChyE<EXQhVXsqOqdz9C$
zY=!f$t8*3{bDHPSxlt>lTJ;aRSVo0M5w;r#ew4G3>g=p1L#jp2$bhK(EkVrW*avsd
z`rj)#E9RlBAq(I`zpyaX2161szaUC`ytwfupokIWXjLRTDd8JQ61gWOvpKrdUA%)_
z_{$rnjP*TN)S%As%`Nm%RL=!`#27CV1TVKOT2-q^P(I#$y3A*1<VH%|o}GLN=kD2Q
zKRdNQCFF~hVIFQ5U{%e{C9jz8-Msz-u{EChw<^L~#QLN&7wo)Q(@nv5E5{ZExGu6X
zDAvi}iUWl_znUtj-Kjw%&3U#DQ@|WTFjO>12{N22A^d5oFtPHg`A*>2pbs3A$hjqO
z;d6A=PTI?6Y15&n3#X3G(XKrG?S8}Y8g`hH`VfISOaYb6*LlZWTSg|dq=fKjvixhA
zMN@I-y2<r)P0|!-Q=_M!HemGS9$LD(E$bcnjV}RVMC}kR@|mN2v=Z79o5^>6b$kI5
zP*3S!;0%xA{;of|*R;;;#SwI{oZ9wGN;;=<icOQ};3uEMP>tq<4QGvS`8{++h|oGq
zua1b;*J0uL)LSJzvsY$4eq;Bk;|`XTX1B75tG_Y4$sKrz^UPLy*&>HJA-J9Y@{8rj
zwQzYsX72^GGtrmTa4DYrjf3Upgwa96EkpPU<=hb<F(}shCIjrU0l{?deDBz?D4UjB
z`PIBkDQrM%{27HfyJqksQ1GB%9xD94H<_u}c&T^FA%pxyYR|#z{Tw6fTYb@Wj1U#T
zipTzNeQg)3ZLKPbo3EJ_+^+SL|A$Cr_JLA#gMeBPikks>`08fYBHMJO)Oz)b+GpPq
z4tlNfgUSUz)P+4(bWEl*`-6U25Wr^J-`S+1k<+?kj00L(t9_0k-jL+-frKU1FYhGX
z9P?xx=`p+fsNU-j8a)_Xow&DqT;k?#D0EOHt2;CZuih|fpA0;Cy|p6zj|bR_EQ3@#
zU)$BqFk8Vfqr2ewW}PIqCRaEzRYOt4&{G=uEyv2i=yu&@xeLP$p))oMvC~7!cJu{Y
z;@{-~XcgwKWDy(JtfAegy+_U*sO6_>q71tZC<(;OG#?B}H)-UcnR;w3v-?f-va(X3
zsL8IqgFpkSoj-^mG%lKHDDiLtj%C}Qe5e&UgIh^zro~^r*iJkqCZW~r^^Mt?n`~tJ
zQLHSpd|4|;XEHK@$Nx??pat=Bl^}_Loy;lcYcWi+$lO^+R|sG7t0!^&u6Wrv$?8-=
zP=__j$z}mN`$|iT)Z7XN0M<<~s5!^G@Wx$oJS-x#;PX@OlaY>+D4IJZqX$t+s8ecv
zO`Yx7VtTl?f0&-l<c&Smm4Mqlu+AC!`GTAqXB9=O`rtvB6aZl!CVfX#N7qUASt9S%
zLtxJtuQ|aahV)80W)r7WE3Ct~+xfQc?Gh9Y9PznCyIyz_q8=P%_7uIs+72plYicI)
zku`tLO!X@@w!gt_>L<umWM!90(pij7$k1uvt>zIdDl{d&7PrS`TFBDxR<H9FUeL}y
z;q3Km;`?#X-aEDaYyx{WzfNR|(D=5k`X4LYTLGcJ<ffoky^kNGadjyHOEj`^a7@o}
zzMKwkj|J`~t2V{nUZ*x^Y-n>Q2Zi>-Z<45vF4meS7J1Xf*gv=4d-9=TP7)WC2uQ){
ziHHz8mUdM`(Nv@Ze5@keoG%pb^zJ<x-igYez~?{11Z}g+jy*(U;Ut32lL!q4P0C`M
zUmz}#gPQH7-<M8RH6tX8t`@`h_|=NhTFy0aYJ+GLg+@^0n&F7BitQL;qg33t7l&17
z&HVPs$HmtxqEjZH;(s*?p4*ufy1#Q=y0=NM&{SLV;=|kQwf6CyK#$MNII~lV_hg7=
z17hFbuzj+!dd~Y?FoN}z9w)IN&du~Y$zU?{)1%N2m7h4%`|2Yqrrv(Da1m$uwGBLE
z3;ChCe`(mum**#JV|pq{)4Xga*wi<_Jej^fTr|7lbb=eXcQp?Td{;j|W*XD8I<Wwt
zm;85?0<YSub%S>tzYtt=aWe?bg>cV+hWNQqx&l1v*`GR33H5(qwB>)G?yq_7(+r1H
zG2%FVz9l)wUsyghyL#5=1@byf*8zpuXKd&#TXVCfEhoAL<as?!J#9XuQ70CcV*ezZ
zKXcM-Y6J=2o4|>Rvgt}6RkU&tL~#9Gd_7B_?Xecr^pt=7uGG}2>4_Hc)$U82=xT&o
z_dZRNm!)$0pX-OD^+{vZqp3R~OfAbh3#ke^)aCIy+S=PS)3y{#)sZ3JJARF3!DhGZ
zgcwV`&o%``F)`3R#}b}5M|58zo%3DAk1?fsc~!x>^piQVXDv*zqFrJ--{mvt<Ou0s
z^sc2dTMp4A=lKE@x?oN2=7o*rB#V+cSRllrlZlq-dk1V1aWyRyTMAH7Wi3i@)BqV-
z^nfy)gja)IKXv?+4*mE^44-?LuJ~M*Ey6P2jbvSq7C5z}82Y048E%%UfOPQ#e(nd2
z*c@27920E6N29VVs_|h*j5J!BG^S5IuojpU)}3rurQY6+rEf@vn>g)Nd|z*C|5^c`
zzhWYq7q@n>TMDxN*uxH8*^F8wM)N7VRn`uQBQNy8pFXhlT~TY47+58`J=~}}l~^~@
zw?}LA_*ttrht{{0T9F9%;$zd|p9m}qLWZ+r>VDc*<o}kre?9z1DaMpoJwp9P2@_3e
z7tQs_r+qY^iezyuO{;Q1nOkATQ-fVDYZp{ZpYM$VQIz|3(Q2f>yEFMHrNqq>BnIN2
zZK{yvJi-KNTR3-uv)<KNB=rLP@%MsVC7#_+dcZq2xGw8<JwW*A<oOUUdc1evr*@J^
zN7uCcA;Ij*l5vHTuwBHs)nIm`v--i7sj+O+c?zkHhu8e$XFVEeBj3LR^^c?H!Y2|@
zSfl{SaY=Hc!BIdb4L-F>SY?h<?o!Ee4?N(kqljW+-&))j*6LF%a=7cBS!yEVXnaTo
z!k_HG8LoCk!T!h2cU)bcu`dnLvVOk*Q6$d%dOdpJEebMB?rE5=!ozX7jZd=ypy6J8
z`;A{(%sxg24$kDoG{zyy9bd!P7Dc7uOCsp$JsH+4k;0Yf69YLdRsFMi!-4dg_U!Sg
zo5iM;uaT%urWm^)kT$8)&BLe?8+vSUYyU;`(ZlXKwmE$dkLVGlQA)8MXg*b<J{ThK
z?A=XwK6GnVM(@`~P^~?YO3Gok!1V5A&l41&M~khN*HUaVMXxuW>=@JO>4h{+ns~5K
z9aA!nj$~UjW(N!qB(<vE=_AI;%Dr!L*ci#h>#I~Kx6hwT^@&Pr5^6b)RsliUhWYt!
z_n65?$@HbBDfz2H{Lj_008+k?h#Z;NY&tQ_>HZ`dgqF_lCJW!qw<P6|ZK@KmQNaH)
zNp01y1HhS!-a}`HF+#jPDD#cqFxz0Vi9jD;!B5r|9VT8UP|%TmBqrf^Bj3uxT+{w5
z?7gGI&x$dOb6?cbd*cJeqaB|_8qRLb_4dhYwaj$68e}Ww>R-fTSvgpHzG>S%MP!q)
z=`@)wee1}Bi?RDFB<`uplwfI1_n8wpzSCL%jJAH3`i8I0>Vl|=H1xnmxuC@Tru2D#
zd)!hOB`j~7WC)Ih922qZAfDq9h1FuIdg|g*Y_|Wlc3m}F_Fv=gBX(>2Pc5dLK{$`m
z@|w19gjb$Kz-X%pFgNU<1sPA?S9?`h^*MT20mtkak%c;?8fIY7U7hQ70Rn6LikKfp
z#p6#Op-e;UksI|C4^t<p^0Q~7%qOJSW9rP1f5Z|fCbmR2atc=Eq7C9>)E2Mugh@SS
zDSg-5tF*-8>o?9GoF)R_MCR8zlBnx(A2Z<{StC-w4>qZDHoNw$15xTTDkaa#JaL>>
zjd(Py9dCz^ibc%*`25@V8h?7DZeRxYX&+>bSJ6O^r_FMa>bh8THi>aor?e1qa6**x
zaj7rKv44{I8a{X}UaiKswb1pv)yK}|I5av!|9E%etw_mP3qByFD6oCqnaTBnK`YIB
z%xu!STxP#~9CYXE`HJd1R=-%WyD@q{=!sRcjrDeQ<K(`vd3|-M;JbI=H|z6;`W>-4
z3~3h1!GT@xx3_=Tw7}^fHj1`dF<R!m7AU<U#6YRPbPIB}mI)WOVeY8f@q$C+)1ogo
zCQFdx_#Km$Kcp0qll<ILJb?>1t&cA%=9-q%qh=DzpDZT67zfa0B$nYIb@R_1hJT&n
zl^~FL?F|@5_#NR{;gE_5XdBUTkdpbBNMVIJfxbn>>FP=Z^_^0`LbT`OuqcKqlMbq1
zS;^^7v}L}D#HE$n69*^~j`03)&%_>GjeEkcb%y>%igxnwYKCx=WTo}G!LZnVlsL??
zPwj_>`QaXmwoN6e+f@^3Vrw@o%$7TzNTy+$=Ubbcl4=g_N7I%22iz@QPavFDsa#pa
z=|XwQlYC?gG*8MveLOXn<bYTDXPG^Lt22et2a4)JOJW5WvO;rx+%2y^=J+n*uYY^x
z*=r?}j^=O5X2MjOv36OaPx$Qtq3-dn`v0o!Euf<Oy1wB-1Qbv#K#)*Ey1T&uM!LJC
zyE`tBQW0s8M!LI81OcUnM!KYiE@^nr_`kjH=Uv}=-uGGGw>}o?xMt?M&UMZ{d;fNv
z{d*Zu;ryEsl`Zy^^2v14`G`%BFt>&5s;HT5t)P^c+qCicTzeXDa^`qGi2(`nnKJ8L
zUSicZy)3P{l%65@{MqYdf#YDi|MS=8u&nI6ak0fn2A(QPt*StQ7j^cubnlK2a5V$Q
zkK4p3EoUolhRV!b{cTJjAbje36vt@1Jh&8n6`xVwCbQ$<x)L~Pvy?Z_vpA3R_-IQ1
zukc4Y8wgY<5kkTDqm;Rg7zEHiZH_F?1V=?aGEl9ayldF=*v59H^+lhi+0xCkke`6g
zW$b9x=IWOfAnVJ^%~$-2loTKzvnhG?<Xd)v?sZ1$$%k&`M^n!7XDj%GiC&(z+UMr`
zW%0r2oUWaBw7NhMijytA_rzpXh)vZIzsBf7ie!xOnIpts_&p$r-mnf6o!4UIjqMm#
zfo4xmN6Mt9m^9thhKM>%i<W}Q%V!>#*BgdoH6DK_O0mpim#X<yFo;<<VqwoKLi5%x
zu=_&UdsV4B-D&4d)*1=Cm}bWdQ;tVXmy{!cCK%7~@MH3MnIq4;c#U5V@bAuTzKlwH
zT)aLL3Xmx@c}N+Uq1rt~LvDKd9B&VRP4&i0TSW&4TDp>!9|LIQu6U=Fw%tRNK|=-B
zmLX?F7tN(dDFOo$4tx@kQ7oTeEV+@yC9|G9h>s*+w<le9I*X^NjE$qzw58K(czT~b
ze>mLq$lt>$DiZf^k~spD`|+;O7Zm(pp@S6s=q%UkQ-0`SRI~5xZ>D}CW#7Xd7WWNZ
zdMazosA11Yeb|WMcAuSNMBG;?LnotEaR?_gMvd4qU_vv0@j4+>tw0(>0r0Y14CcXv
zhu9ZOL`8zscbF-0H%XE^3vXJN&W9TnuSvMZer?o`$Hik{<n%6a`t`9=y|^edgGfds
z#8aG8DQ#U{#c$QVQ9v&)wGgqaYpCCu%DpSdWAP&6K<&)iU|R^QYHWPM(Y5&4!-Mq{
z$w1_N7_;I@|8FFJ4|u$eeLn{7Vim<E&<{DkM9b}e{w%Lc7}quMX)*oKu5^d6cCt0j
z#K`KQ5k5h9WW>v~RF=DtpPw^XT}?ni#Sj?1R28yqsH3nxau0iOf3JUJ_|EcnSXi(C
zaj3I`rXHT*s4*p3n3B-Sh**JjQW<W3*y+cS_Po0#5oLHRVfDsXg4)_2tceZgMHcHE
z9rHY)27bO&Zo-CPpO3^(=G+Ou@kWP`_dj|Nozo*RtdK~e@EKH0vlVE;u~J+pK5{ye
zc5QR(F#51`UqJG2o}n}jSb4sj5wjbIN0eMVYC`-z>0}lWw}syZPrsB;7x1RJb!QG&
z4XgD+MwS`RO#6wjV9V^n?ja^RDyF;O3{{w{Fu!7;x0rS8w^*AkLk9(yogl|D??a(J
z=KB^BHN5!w0aLZGCZH_rwhWV2I5sLkET?wa)y1T>9$oMj7u!g3TUXGoO~3v{O=a#E
zn5``UOF>B%Dz91wK62_V?jC!_`;g2G^~#JC_7*CS{Zu`%DDIQZ_Kyax`^LKv+ayb)
z#%+FXOVyPk)rh+mlo{!Lo~j&uRWHq^C7>%{8tk0xMB`Ml-c0^B(j4>oxKZfu6^nA4
zajw+6OD3zw=)MkA$XEs))+K?ndh**sF4+@{(jtqxlFY$+<@%1g(?_kLOH4rlqQB^>
zEUb@j|HBdrM1wf$U9{wrzUJo5$%3?D=0(GO$9r?t$rf2h_ekCNo;{LVg_hBogj(PQ
z$XI5$e_~CtHmg~%hjf!|@Na_9h6^*?CNk*#<g8aB2sJC@mndGi?)!144A)woc+UEy
zoUY!V_~047J!V^)J3Dcj+i~)fu?Bmg+s>|kFvHo&zn7t)fat1St&;TU&i3U_2KH7^
zt*8Xc>c*`I8<6YY`H`;&t$`}SDIshAx|i9O!zpzAom$4cTLq8-L&!tK4}4OA1g{1u
zCmh3uXJ;Mb#T|Jiw~^u9p2^(giCL<m9@NaTvYdEDTGtHRz6jI*tgrdryrGF+s?>Y<
z(Gn@FTQXu++o+>;=+3A3`ixiJbL-T_%v4V#l#!b$S!|Jp)$IH_)U0tv=x$8i-=pKL
zJL&0{dVeykgVp}>S^9Qes)kgYAuDGR-!y1FL_$iU@A_o*7BV<nV|S-TS3RNlO~ZU~
zzUr>=jD<kav!<gK-D6g<%?yr~;%7;uq-!(vBYeD+fA{ALoye~YeIW+ko)t!IBBks&
z1X<4t%Y@e<{L6fCa{;;jl*mZ+JT<Xg+feJ-kU;d8rEKJ(;%zcbp4`04Aj9rvw;Zgj
zqE^s8z4gJDtU2Fh{iX*x3B6zHAbj>yCrFMNoA$6}3%QYNav0whKiDoPr!#ytub#IV
z(s_}Mtf{sfyk0bVzRjY6ficBzOrMBHs@loQehm}C$_uQNjZwR`?l{-L*{oHQ(i@Cd
zu@DNcv>p6rb>(00YB(_Ayd#y3ojaZ`kM(E|=C(XK|1|O3c2%ZuybMRx^y{i!Rew?y
zZm#>a8XdkCqbU0^yoGms<QcPS*WpY3-#P-bd{h>lifA73w@fC=i-@R19hQ((xg{`F
z|9Y+qae-RD%Fy@hz{j5LJ^?8g=Z#G(Z9(Q7>}rBJ<U&tNv5wXHBBOeZDNDwTTjHC@
zd-(dK_tCu1hg%_<hrYFn>hbV_?Qy}YH9{*PuV|K3sKZDj<o-EerL)tJxSVJrR5fnR
zXXMq*Cg^WqBEiJWC!WvgDpVF>WP&T5kwEmSW7U;G1xBCQ|BaI(jCS7dzkYG$!+Gmw
zkPZgD{`%St+Zz9OMW6l0*1qb<qxD=~?kLc>wLJWE&V_?rP`8L=c!!J2DKOL3t5tW@
zM5Mgtv?H(dY3SzZNh;YC2A7Ql3&@IQSHE1G|NbF0md*VukIC7XMHMLefWka>&C{_T
zXffnq`$%HxcTnObSZnHPw3oh9Qnss!!RlGZJ$gBEYwL}sq{WMejH{qpi#H|W$J}m7
zWNf0T%ZD{Qp|m$Wz*$1;F85hi|Af25RcU$erBlFKdwY=XOyXlb^6$lpM3L3U0)!u<
z=N2};No7dHTP<v?ey^Ht$cNJAR^B0PEwfZw_Hg@Ku)>5CZ7bdxZplBwQA!&=NVhdn
zK^o}ZgN5CC1l~Bq!~<E{`<LJOPAK11HNOg?7Ye&wcN0(T{t?2xRDk>A=Rf7OcMW!4
znu5B40}zUkUNggU%f|2;zWXLgqkNux3xZ7i;~NoSjVjh&k-5;DhU6v#AHrJ7bQ@4*
zU~>NH#>g_e?OPUq#62`OgW;l3yWzikD8L$|ocnMHX>y_t^`}YoO|VvN#A0(XV+@X7
zJ8iDtRXgBlkU|xKH|?ijAAhjDN!i9&pf)i?*!JPK*%XP}dc*dK@j7=!?-#c%^Q2@F
z6}Z1U^kWp`4#NTK8m;m$=*fsvyf!#)wZ9^KkwnVYsXV&;;dVB<qN#(cFWg#R|M6O#
z`-c?%UaLEd-^YR*Fm;P`8}DXZD;+)Nz*02<ER*WTgZn0leclUfRcs<lM&q}osH4Ar
z;DtS%Vw>2Ya8DK4kFNi%6*@jQN?W8bmOCLj1f}WM*W`Z*=m1#<W#@Ru{JA1SdBgTz
zeV9qdb45N-I*W?h9C9Cm>{eC7)^&fhr=7X{-$vErZ$d`*=N5aVGHfesW-Br?8{f=F
z&8DT7V}oL1i3Ckmbp8o+6W2G7O?{5f$W2)E%&SkFryK5VtY>sF9_*=zu%4)weoulx
zS%GCv`IR#!NYU5U`ymSc<zC;g=LT7bBtm2SnDct)29Mp>oSapAyUTultbz}4clHSN
z;;s43FL|P4)pA?5MwU6e*Uh#*&b#Bk^_j15U$sCYIu4wimvZ1GU|xm|r^uh?&M7=E
zr=K79w<~<W;>vTGAlz%U5BYcAKLI&p74F4MZs&^d-hs4oa3r8&ZSG0zWy24veU=pO
z_p3X0_`F_wuy5+&)D|!wX*FX`U%2b6hAp4}{h*`mZXy3-m3YMcu%yDWsqo}L<dn*4
zUAht$h`hQ~^u}2oOZ3VW_+k8@_Z*6xv(&^8^oHH*7DLwv3<ULmi6u)71|6RCKO381
zx_s#bcdaDdZ#9ykLrZgWX~q|4+nnn8i|@38*g*-MGr5qvsBtX!H(&Ztb^$aAEG(a-
zV{EjUSSERw>R#>o578SWtLr)C=0)?{*OexG#oMB72_p0q!TG8$N1HSSY4XO)&4XR_
zf!9g7c3yE_=ZlJf;t(C^!G(!Af2V!mElS8k-r*1^77d9J+N!7x5x7dc$)W629Qyf)
z+_oJ4O7?chr6$(QN7ko<)%K@n&uMHyL_(v?J!Cwa=V0S2D%zr2px~^qJ&O(*J&%9a
zVy8;oBc~d|-5$HJwAHta2Z9G%HZC8CM3|JI_!`%M0^iuGq^c@oW>Dq#)k>*IwYxgZ
zWPp?E1rzM(7g~&tM`NK)kV6$&^>P7shs4+w&&E?-A%)^JwEYy%1Osp8U<>rdM>vak
z1%X|CxGluK8Pq$=Y;Gs4(WCuMDx6T`S@rlO0A$-2J5ksUBIDW^-{Z2|{Ll1(YX%jk
z&^6r?mU|QYJ8#ZLjl7oa(_YuB<zjVXm&WWTUklw`>4A;6U@NirRGynvj<t@st!mWT
z<f(6%!nVrq2e<>{Nzp|!26mSA5v+}RdMY%(%PuSiT7T`;kL4=jJ=&iexF~&0#6w6(
z0BW}IlciaFU5i!LoO(b}Ag`G+9fZ&Dy0|J`NMHU=AaxcpINYQ0tS!u9afUW#+J;F^
zqE=!=+y~U>aZ4-zMWkjVppJl-1yqU^at8%X5PU<!*RSIT1!e5}IZHjz{5Zc1aH_N4
zd>*@=j}dM487(NN<&pxG|9*S9EcHxK@RvA$m02HClr9ZR6ZCK<#Gh!le~|QAk_wEY
zLJ?!`Lw$`!`^1{<;<)IoVdAEg7_I7j*w0(s%rhAM(}d4-K;edjgUf9Nu|rdC&xQs^
ztIHK|7=ak1<vVt?Vl{D@e`2)iv`pW{&R5C1ZWum1G9F(iC*1LK9aEF5C@V^DDY;$h
z)A=Ef@;1Hl5=7icq7qaSc0YP`*u7gxOc5;pbDn8b$Zkx3b~22FnnOfyM7^0!eMCE%
zsN|JJBPouorrCV)o8=e}=TJq4;XYZ-D%#oU2>L|e`TEQ>2GW^y(ZGQJ?DsQpzjpa#
zmqbcN>FWYZWLmtyb-n^mh8*GrDNFyd2r<9Hcy1_WG$Pitq`skc=i*yjeDg1wY@dct
zgZT0E`XtD70$N9B-*ftyX{|i0G|zQaugjh&?a}nQ_QfRYN53~w36L1+8#NhZJN<Xm
zliT0RXt{wvuk#EuJ%b^L+ad*ib-^Y(>z>{b=I@)B_5?AeT78R$e{1$?D<yp)pyKGp
zJ0y0P@)|5<<@D6!e`I@*MHG8G)#sFWFCNJ6p94%1l5k`_`>~LEcK0$et4HSq2d<&<
z@bxRdDC*luB&4gug+$vTdw0~FkopQ=K-u?q26|PoE)K$^av&d$HM=chpcFf6d|4AT
z1HZ5Lgjpy#*tFMVeLf@F6jI4&{H1jL?9^d0Mve6%0?Xbity0ummsse1rwvgTNE_^R
zk6QmCj2qVY)!W@0${^snFqD%@^$k&Od4_mr>AIi`{O_ZQwvX4;qi#)NE1T@(QA{fb
zo$Scpa1uVTeZov4$t!DMP6sTBNhjcGTiJ<A-356i>7*+zFjOcPPOF4wPg$h;ynBaL
zw|UW}z<;62+l;f(ZBlZJ)QQPat?>0pHWRqYZ6cbIh{O;a|5F#^o|XclnUWHE8V9tz
zi^kuj`}-4J=w2kK<6|oPI?44t07c6e75LepjmDVpMrB~upUT(Kg#nvdyH_szyI1S)
zW`aun{`hx8>v;p~D&pBpVdAOu(#AbYaS6p$@d=i=RKXghH{$tHKF+O4YT7Vk`hQ($
z`RSM;j&hGdiB=iQB`*k??ASSH8SZCSetc_I+Kh}>baWUISEn?>4$2oE+}mTE$ud@d
z2sQ5$z(5RQLrU_oz!zWDhzKW}aj5a@-o4m8C+DKx?|lr+26EaYyK6)shCK)lLW6?K
zqAd;~P?VZu5|nbo=|~yJBtk8#++U}1ZW~wlX)^^^ZqLu2SKzcFG44)^o#U+V)Km=~
zsPFfG2Pff8k(80SMVi2sr4+Kvjr9%jU}L@eSD7oD`-aZ1xajbdL<xHhi0H|sPXN~S
zPdWixUQ)f*nP*z)0rjtG8J}=`Ns|pa{eg%`p2f*Ih;Vtm^%n<y<8zt6l+!ymjZt|>
z?&)eB;2Uf2$Cli8uwZJ8-?8zMPyqc0tXMPuRX*_G=ZuWQdw>eiwbnufP#Uu3Roxfd
zi07w>5YQ%HIixrguAdO}Q{idrvrF9+?V&7=F{d>vHIZ8>0urJ00(Y&%`VnqhT?|Dc
zu0VW&zSU6H8`JekNjY}+NIA48mfGa9__iHX-u3<z)HWa@LcCKc|03yDU2ytyC8?$X
z1Q0Nv>sI9pSRj-Z+*M$1^$qb}0>iW0Mo`0J)5%UY*vZKCQ#UQrDQoG^G+lQehD|be
z2wNZqAZ=^9V8Qz#NgCZhB}xOAo&*>_yZCVQ+Qx8db|mM>+#;!?mheN!vJQP}#wj?B
zYQ^UiFk5X)^OQQY@Xa;B>8ZCjaP>uVK2G%)U|s6!u6QQ=YZ7-;snB%ob={!adPPOu
z+xA%&Bc58Ms#!z{n2CK<){pSp^O+`<8@OGJ*gK*{L;XP7r21&!Mmi;hvfa7lyNj>(
zi*=TU?XPtS0ZEX8#iCuArctlz5?8o<5a`43WA1aQP9p8%%DUt`&6Xb3Kz_-h$3GEc
zM&scV_O}z6H($p*&lf&8xwL_J^*}mAq;R$c<RqR53r!m(g4(N&@qq~7>VNNcof0oC
zIcNP_GZmU=|6Ul|AZtPq;}K)Ym1Z#7&ord&jQ1wj(?oDmCC8pbVBL^-#a(GT1$CYT
zX3N5CJEm$(M#gC47JT|ZxviPPOGs#Ks99%GX6G7rt^^MRPA&b6#EbEm6qrcD<3|So
z;$y~h%IM2GRW_?5&#liBb1Wz(Eey#X;^MANcN;ZrR4YV)vH}GW)~r#cdHu3X2T8^p
z6{G7#6s^|VD-<f)x_#g2l<F)eJn3sEzv7UYV{`DgsOjk2(709M<6@_Gi6J(Y7-}~@
zOUN;2)BHJq3&bT|;ZHR5{uW6QcdTiyeg>1vgO>Swz>g+!y?7iY0~~h=Q2W9wkK2H4
z{fMT7xc5CS(7X>p5Ul!e;O!QXQxPblE_^dpeawyTgt<lRc1V3c^WbRjH7Eq9Z*Y7W
zWSu<xa4u09)}hN&de@W&MuCn_^s@>F0$FogjcRAbBDEu)Lp{n+amPK!O4jyjdr@yp
zzp^o1kE3UBcMLU=t9uOGwhAE&poW`6pye&=>MF+FV&vKfG_)NrDfNE*VjPN`(zkvr
zO+83b0mRXhzljnk2$iL>UouW!1qVWvhg%MK`?ZMjdP;&cc42}p(7t@qdS1x->Re0W
zVn;q9$KvPyE`5hkwCqlU+~IwYyg@Qy*fUQ%PK%~fwi?*pJ-2$E=3x5dJn3CS0S?Qo
zz2~Cb&b8mCO|AtW`U2?y#MRV$D)x#`N3&YrE^I4QMC|ME*%q282g}XK?xO9v#TVc6
ze+B6Zt<IJ`Z|Y4UgY0p@A=Of5tjp>rZC52E$24JgL_|_ZgnLdeSYCsJ!S>s|`f^r{
zT@%UEck53+&Hk8$V|b8T2g$0nt>$gpakhV)tN)vk(?o6eUXYbi@*B6t*zsxXYOHu+
zgD=f{ZCvFA2F7k5`eF|^IaiuUjqU&*JHv=iEHhN7T%`7Og%ks1Q;hb0vc+z?#`5v;
zvtI?8j=Z)dr)f>F>UN#c3Rvzn1P2*cc7fCgy1WCmvEa~Q?!8<0!=8D?8w(Kow&#HZ
zgPiiJ8s?6Bm8>Na>#8UNhHK2V({)z`^XolKeING1E~=K!L_Rq|i|V?##7BC9XAg*>
zP|V%~UKNiv#Yg-7jpSkC83r?sa`W>@i<9Onv~C`2lkpCjI}6vMCVguHm(ICbm_D=8
zCxI+22!0J!@XP&6)zvDa<d;Y0`5O&>V+ze8g3Iqgrg-ss2j6|W5<IO$zM+tJ$HzMI
z+>73RGfuCQbB+U>G0fMmppnaU$KE&|US(VaNkA1JsPZlI>asj``&sr-o3a5!ubSO{
z(;cFvkmX$aRi|>eEw7qEMneuge4h6)Zo76fl96p^;*IUFxF?GW=3E~7juO9e{E7XO
zqHEdTTH=1p6$akmL3syzM<OdGgK=p*9if2q;V|RF@(=V<iXksZW+v0Y?eK<I;``u4
z29YN2ujy82pw3`n8f&{#J}>x(#C+5X+QE@8&p+Q@CJUxxdzgRN<MvCecze9<6ut2+
z2<blUyV7iOYveorS*p5TOMAm{9--4PUb-#1DEl{E$SlTvj*x!a;piQ<buQq@Kh}J}
zZY&|C+)@NJ5gz;wl;qj5ljrg<lrjMIuGMHvt<T1}SJ1tLm~iSnYU}q_1Z?-eV^wHW
zx{x+HOJfM0or+1}?Kr;~BtDcp9zCBBbqpIVIyN6HcyS#wsmf8mfDOImDL5WCpfvkg
z-#OM}xG}iILz(|45eX;#YD3A5>Ib6lZqrC$nTT9Kw0Kjj$6j(*N#L9f3NzkW#T}Y#
z!!t*c9djyJAMGlIC7$u+`Pzz8Z_oxz2j(kFq;W`_qr0oH?$XW{X(2Y%Zt>0rAEy3F
z8`+x<h;y^P9BbUaxc4avHk*}?u+);<aaCQ?8Duc<MBAXd=I;mcc8rM!`Se$>KbzV#
z1w!A`U|Wg8A$wVC`M+&h^gv~FaUI(CJY_>GCg1nOQ#v}Iu39VTZ+O2$Ws4&+ja2k%
z;zH$MycTo6Z+yZ>fDARqT=~aPfxoAnYrml+(Y0x!`l;m)*Yj<EKCcEFA+F%}Dwst#
zQ*3Oc>(69U>YQsw7HjKxE9Ri~6;yUu_rl@{fNZ%ETQZ=~p$8m8=J;LjM|ePCH0HA2
zHBchFGQAzg>bk@dZQJE9<L*4cXI3@k)2kv)=1B-v3vuy+1qO~3Dn9Z)SN2`&yHRza
zQjU$hO31oQ`Z07@Jm~9=yq%-31&^Jx{^phth>snPLIl6<`z@e&k6EE9kC!5-69ZfV
zE?>K(4_9z6XLK7`EvUj~1t0Jbnig~tbIpM=Xj?>LNV6emM38qXFlMBqOGrbr)ntCB
zWX2f_MB2GY$J)1QHapq}PR|M=ajKE)SmZ6XlP;WHjejeS6`w}jv=_5;c3CbUOj+7r
zNio@l^wv*oE@m^AegPQp5fp-gr*Su0mC<(BplRqN1{MOn!3BlbM<!kGrACN($TEX|
zLZBknM1QOgoM6l&SggNs-OY2i&;JMo><T9v;x;8-OLliZS5tcK<|{&)$Ylr=F@dST
zm0EjNR$l06PEzpl){e)c80zLCQOMYCMa4fvl?i)f;1Rz^p6||0;9eocPX_LR3=H$z
zz@y9V`l|QKV|gV}DRG_PdF&XuVh*Ofz$xmRI6I~Q+Z>qB_HB{BrnrWF49cP1k3OHB
zfSMK;@(@aDt&y6K*tfI6!OP;!QGB@?se+lUcncAU0T-XtV}7x496CsYPgzwBkpCH7
z{rWpi_ox9h60;w+s9T?~tObAiPSu&(THTy}-obrH%fiq(-oEsvZ`As2VhG2T?RN8G
zM*iMa*caQbOu)Bv?ekll01BP(^oRNi%9-P|>2F&RDr1<lRfXy<&dz0LI`5gi#9j{f
zmY3H*5ED~?Uf{BEDrkns%3lk3plI@x<LF4Dhmj3KvLU`_{lx9pSL3ZZQ(;F__IP(h
zykpQ&lXTmeu8A|<;o!T%jfK$f4sSBQi<+ws?Qd_y3CkK8-`^RHsC_EqG*;DSap6H6
z7_b_|z&v(KO`Ex4ni-Cbjh%nEen>!5?)aXk>rAL#9*RIy4AP+~uNa|!ntgilr;kt=
z#az2PL#aUJVhb86Uksf@gp$2XLayPus>MWY*+>@J%7eSanvD@F`97v6#e?40gj@GN
zAxi3IGfpdgsZlLr+$gW<CJpfj_zYWKoz@7?HfmtFNaT438RY8ad`#d!*PP~ddaD-Z
z`?3uhEnj1yygr0oFcv{a*CZM_Oi_Qy)QmNptttb#_mDG(dx$GHQ3h*u+0jyPDu+IM
zW|7T{y2Xxyk1C<>DFX+O8qdWE3-P%~*qX?rFG_yzGrK9W?)GM~^=2_7e*b{o?<(MG
zjx4avT@gC61X+=SZp^O3$rPdYJdv_$B~XU9yJEJIw}A7($%*>BXVAg%(vg`G7Zg_g
zOY(M2(7}f8MLOz!EuCy{2$x<7IVg`I+%(TVlEfrTWWtNGo(x28s-@oi6zTIlVur6!
zN8q;Q+=0Zw<zkq{!O{g&-lJT}v5?7JYwaYT8BgVj2(3mpO%*RE<w(xu_h?+|5a}E#
zsA$gam^bK}iPjN(U*n7~G0V(!B2?dqS@5SN+w8X+y3zKhJw%~iNV)cTn=4v#AH21M
zY~wZ^I6IYdSh(&c#q_vl;O*e2jXUYwn-e-K&ge00xX(FkNY&0$dGH=UFVCNCevfJ9
z&3k$`)Cuu+o=cN_?$(=Y=!f4SCLXD5x?Z8jb-g&-M&8+#)LPby3!FZ7&qBxka1S&G
z`4SpWiC7l@&d!p$pvV$&4u6I))YW-D7!W(SJ@I|s3AwzYE3a$7wZe~K@zYxELN(5L
z<MC#1l<w6@Yik~5;9!L(m6FG0pMbjlE#dXAkHqu5y9pFuRUbgX@4rAl?3)$$jT#jN
zV-N*j5;vbGeg0caklzB%1O92_I`Xn5L{ahghRQ;HN!C-p!NuLkRC@!`sXH_}WLcCn
za^|C*-4;2LY46W3&JRj*Omp3%Xw5g_YkAX()%#Oe*+$)NcEQn0!^7*v1K>~I#cMwg
z6le0)<q)J~@qn2AVm<fZb$;|p{`q62LA4ws=_|O2pOl;|Nl66y=?J_DB>?^XcX!es
z$kBRQGSF=Zif>&eJTdQj8Tq95#3+8s1i!f%XE;5UZQyiaDIVFMpB0@%Z8)g9w%Im_
zILVTrn3x{y7#jSbGf3{*)k8P9EfLyZc(xs50BO%*k&B`aourRyo(<wy5Z!0;wpXk-
z3^@?1k5-khH)xhRJ&isP5qX8!j?q{h<cNBTjGv^@RQj_CI7Pj9?vcG@jswv(s*^{9
z{h#t(hxuP1rz9nyF6FM2rgpt|SjdRmMVc0iv${h{+l@Ky48ke{!E1xTYl`7|yQ`3M
zgH{*)p7gw(f>q%qzj1j7n;z|)y}6esb+6BJ==@7fmD^?<`z#-0lcPHJZe04ctZQ-6
zO9x(dgKO!d9!pzv^}ox7M+t7Drsx(f8jIj<5=8&n6o=x<PJZZ&&}3F<R;t6H5E=_I
zufxFLbaH!pppEC_0G3|i(B5`rCA4mr&_}`)ZtxJLTQC8kh(g@a5YtUHdRo?gWTc-*
zj=OU7jE_-08Ifm&@VP^>Dit*Y*Tvhtum@cT)(3>7MFBR`oTq!2#E$JSjAONZ+dn-g
z({Qipx{(Mu=etpI=bk2WRly@a8mAQd=!r5}JQ~wRtt<`A$GEmNNM6<^#mWy28|PMx
zh`|FpU1oE~h?GP@I6@|cYKSvN%}h-)T;|fJ=lY)(?p2|_gk;P$4Q`9i2vr#nEZ((K
zE9!1oxu(zjE$7g0w*(UfeclPzFX%>12P{>r1t+X_^=bPfnvId|jp*l*!;SDL<k3>Q
z!`^<Cm0@Ho1l7Pr$OLm2bL6@S?&N~c68G0diR;YhJ@@z>P96gK)uVpWuy44{?<Owy
z;uG8htR(L3A<`Hsiv?Aa4peZViOcMu{C#F11tw6rG>g}#+@7SVtF0TXRB?G3@$nOl
zJSvWt71}}*RU=YY+swB%M~3SEX`Kq2i-Ev}4RCJsS9yD^XB<Atv}Mow%%-xw8t{A~
zB_mgSp4bk)lE5^GW3;?%(YnuKu^C$w`NUkll9NgkdGE=~>ykDV$0S)sa0ZS_>up5+
z@-EHcfprb?$K+Gignto>tb~rp$#)4+1`*ma@-y;HtE?Lh82m7{Kst{dPeJe3OtG<@
zM>6Cd1UFbe%j9X0;#kP)PLvEqe~+#QJ+~mE(wUV`n3nNqZ2F)N2{jy}3-R0fjqv<4
zF)-Z#U&1Whweja_M_bgSRZ>?Cdh9Hv{w%Vu3gl?n1+~LIHe{(<6e6;le+N}xHR>TJ
zBu{6H6({?<@K~uXxW%F6M><9USgWB~<d(Y=EXYjNKNAjf6S<}vy$cPIt5)RFu2tM0
zEWA)0aL;bElzmll&e+C!Fk=YGE4#UE5@;d{r^{8YPEQ}@<;i$tXFTvg%ZQr{>hfaO
zjXVhc!Xj@YCdQ<4^Uq~{gTQc&D?hv&g;KmV2SO<`p{33&ELTabat5+q&rJDK%lrey
z{B!MW6A`=A;`PZKS<7O)%kvqm_D4!A^eG?bpXp4f6W_~-8Xxy?9Dkp{^k?6Jop&#!
z@UwQ4z&8f-(RJdZxcFD&6>1(<Im3*S$m!bSSy7R-=;sp=44cA~@b~6Z>LFwo<RImp
z*A9)0H629d9BhV(z{|u2=RmfB2EM2%sh2YRrTJ1k>|;~iGWI{cv_<`2w1i;mOo6EN
z<tEpK<Mf!vp`KI~wR37Edxms_DjQn@`#G_Y*WyDi?3-n2itFBGpxLeXm+uSi&IL%G
z&x%&z+}koEc}dzYB{=gVPXEjZ1nXzLyQ3P>y%agLR$2dno<pa2-^Np_RdhO%Ro<LN
z)?#Q&h(9)-u3;#zcGb+eIk&3*YWX#hxPAUwy|2%DNMt;Xbd8ou(dAs(V*4Y)R6~U-
zM>eXen-T0A{7&b!6m9wA<jil^F1H1#Y=)v6cXmG{7$&rbVcB}=KJT?p1T?I=tkX}!
zu*gm8rz5H2x^;>Wj`J5i6)BJ0bbYDfvlBa?p|1#co?TUJU!BGh)=qCJBD7${WY=cs
z>1k=@U0kY1jFmOQr+vP5a)z2^uCuJ`EIS0ZDcq;lDCUQJ<%~@6Z3FaLS}pL+mCr5R
z899_lwW@P{pxk?==<eDX(CLPXTl0&yUk-I{`?ukd{wo+sfnV;2XHu9(yjM1ICdz&J
z8wd3G2Atp_G1|PrZBLF^$Q7@deY)XRKnqywA|fJG;}ndy=JX4saz&(-TpD!z{QUOz
zOEg6vJ;GOMj8H;nL6(R3FV}v0h+!`>l1Gs<bbsdhgw@`@A?YhURp{Ox)`S&-+BJr{
zX&Z#$SJ0Ck&CicO?K(ppL==Jl-~koUhxtQeaM@F1PLh%_{hCSqJ98}HV5BaO(UTRo
zD?N4f+}74-@OF2<BC@8S%5kf`k_As4*o@y7QI%gSD>=TfZ6R;wHw<|Dad2W>5s?$M
z-Ny55_x**?i;?q8Xf)a>O%wn_nTntEWB&Q}Cq(q|Ic*45-Yena)wrFYPQNgL_4N-y
z<OrPR;$Q1UK9@f1pEAGZ5CzS<3+^<+;Ormrzr^{v?uek)2rKyb>};%{(btbbw5`*Y
zPaZw!H&R0o&t5V1S8>ObxM!ba4oQ8XA=b;l;4y7K80=5~f3xr#Zhpl5jP#12-Xltu
zFB2iB+LQs2L{Np2IvL+m`KQ&WZ_o!{!rUiNij&rE$7Jw({~4UB-O9=*O7;UXPDh7h
zBUbkR=#GCb=kRh&%Bi!oAX7T$pl1t5%g%a*h=>F9@V@e#-F`s)Kv_!V+CTSCdjcA!
zTJ$GjZ7E&#C1(n`4_=5!iY%-8_;yFflE&=@ry3E|9t?lGS#Ljw@2tf5&s`dIFEmE)
zn!(;!b9<kTt#F=XsU7~PnS~GeL^Fp(V@gQr5W@$!f2_r_lMEhd4Cdcg)bzt`>+k-O
zm>+EO(E3p3U;EFO5OsI;jPN^5?jAt|-p#6zW~I9M51OoBh1ipN<3$C!Ngg?;E=cuY
z)PTWH+LOoN8-A$<W_zUt04ly2wVKH{o~2&P@eg)7KQPC@z|41#`nM%(TAb`=#VwHf
z*DMkO5-1CW(%x)v(8J;dbk(dH_s?a0G2i-*Nu1y{%pmU}Z`DjPd75Nfg%bY|q5Y0@
z(0?)(&rwcD@GA1nBc51DKdl}A>#LW1e36;`Z*PP!_{B`g=4_0QM^aJps$b5}qXZa#
ze6xIInZ6Ky|Ky}hwT#6vMVKkBn;LHM;J%?fS82lHFdaq!6x1a`^$iI~mW1D{W&e3R
z&|5PLG?o45b&J%0D751KL!rg;9||qq|8O=m{)a)!jqcxU9y^Nvp=0@fT=<Xt`TqZL
zWz_ip?-z#tUyS*GcIE%>!f2|5b44irSYs0tO&ti+!q+?jrAg)G!vXVp5i=Hn11||Z
zJ$)n<ifkPbhG_AMpSFpk!oa}L(7svt%cVg@U7Z$=ed7jb3R+Q1q5fmZ;fGBP1K^Jg
zGEmpK&8Q6C2GRou9LrU;#Og(9X`(<SV&9nCtYUHGPj=u>j#S>B(c1c1)`q+X!G_r{
zuc?SBD@VI*P2~YDP)!|di?`+pjSBu#HZ{^i(U~5HZ*WIx-5GbL<|d_w4gz?C40gWa
zx%I-}jV7YzjV8Xzr|w@7h>{s0GQgi;?R*nv0eUnDqNDkSo$aSy6#|w_cqEFbb!)mo
zOZzUYz<Q=p!*$lXhRVF>6U64R_k)J1_=c!AtH*&^R!&YgP{D9GTs00ojZ7WT-_CgL
z%fJUK5W)rXZNaGT>$S4}L;nMu;Qy0o{hyYpi&bW*vi=4v7s3Pl?@p%&=5R&8lJJ0-
z*uwrLtS~k%E+-Xm4Mq-U!tYmEjSnuohYe#jxNgU`%>$^8oTw@oS%T$av9q%~ogdr7
z6+z>FKnQOPzUy`K-;6eHblGO8L7;yAyAAUn{u{SESOLK91%-NFbW56kYi3!S^d;p1
z*4<HsmoQ`^V9P4fu2TdQ!(HFM%L4{m=XQLxRY0<-YGqY8vV@k4E}NqS$Sj0PssR&Z
zhRv6Y-<qgdHAu*gjEXuuSxL1-qzH9&cPoL<dmL>Qf!SnWW`_EZ_t1`SfXyo0%_&o-
zUF=wmHs>ilG%P{xoe7u(V(J9}t0*A)d?hVCJeVbgf2T!?u7xmZ{qWg4K)gHbKn+9U
z<P7C?l6gu4zTzxqIVmY6P0a-1lVA6z9T7sr!d|s6@vrSrH!`5j?Qb<5=>mplXVo~g
zG%}5Hqjv<L*I%KpFHGosr>%aYy!!zONnv_A1u37yi|#m<n)(+oXDU`!`KuEIL_d`p
z5%C!`{qJJM1w^d{w;9m&v+zXDQ8dXYI`R8*f}i`61sEPb7UkpP(`@#tyLbQo;i#7F
z+MpDR3ZQ>2VkQ(t?<8r2d8MS1%LT$N*8vSPq}k|J_J?TYm87J+q~u%NHhF#hw24ad
zjM=NpqYpC}kKrg*5URBa%aRzd;i@0PoJ|0MOl0QTP)>p8(H1r?ZZz=TrfSQc$|mJ9
z>(om|f(b8Tyzd+Kcm%9GZ+9fM6cs)q>&M$iuoyNOB|zR8L)Coi`tcp$UqPv<EN~RV
zbDcF{U8aHaJ1$Enk6^U8{!zQ<t(-D|Y*jVV2@pDHY2WfawE_H^T?m94;9Lb`G<4>J
zA>RUl9Lfo!GQYaKSgHnFCxXfuFxd7Xugx_a+dP4#)&M?f6u<K;urx+Sne|R<=>^60
zuK=|u@V_6b#$l!1HZ}!~FYP2F6L@Tkg*<<o{fMMSQ2@8Lwhm`V;DwU#mA4M?!@Rm_
zQoShP^(V`T?x1hSCpx8UT0jE}NN?*?Jdkwlot>!7dvbDedN3r1BESJ`Br_Z;=&<3Z
zZAS0hV8Y1tk}snSC5?*nmp0S&Yw@Njkb$sJOE3wqohY90aT?%_<@7ms>rD}Am~ugh
zG2pb&u4zZhRomO&=K}A*2LT^<37}Wj(7x+i?*jHMlXk7Vp7)_jUUyq-tJ8W>jm7cK
z{NB+~4IqbX4!8?eTh+`gck9cQ&*9pT1b~a|xX-}|U&+dj0$QF!2$0RLj;QI4kY623
z6i8vnOKi9_P|<O4C{>suMnAC`I(0ME&<YA_0|W?e6SLeCpOGaMmoE`U24pa<9Ur~4
z3+OT$IQSpoJo8)zP$vg~MmnxekcCAKh$dLj4}i?aO3WDMbqlD5R26pH(Qz=boNhb%
zRkqNq?$NYsD!0v4F4(+f{rBc5Ux|;`PTL%=kI8|Z5@)JsXIENWw@kdWw6vD&Ix_%B
zBmkQ_W5Fv7{jzW0UclQaPtlH30P%-qUlQN@(9jVnmL?Q0Hyzjr;6YKsdwF^E#d5JD
z6s&Fv1tsM_8-c<SCjU_tD$;bnHU~sO14vi>y}Nhk$H}-Xl<Ta6aEU13{BApGNwTKN
z!x(8ka5=d#K;&awF|ICpb4Ltt2_<ltexzD!C)88UkxhQ^@F8s*>jMIUY_P_<t~2gA
z*f=;<cJt3d3NYqyP)+KuQHRi?Y@YT_-+G7NW3-8!=885@BH*<~xcqoy1*8kiAvKEX
z-N;9#1q;Q19yttHxnD|1#7&w20`<uEV;A~slrU5f0!Cj7UHQNl7*u@mZen9A{-H$D
z(9l?_{((TSjF9=-+Sh5LlmM0a&r$;dHS`v2>e)Fq+>BOk>a!^u*W}^rEj3_|L85#>
z#b(PSa>FC}osmZNTri~K53dAjw7!xepot@go4q|Tu9qgE%~=A9nQWk(fIU#Ox3>od
zs^uE>x?^Tmmddf;4q!K;f?H2i=5gM@^i2?L!D*oZo62&&HIUzK&KC&P${uhNtfy)f
z@PrP+ss1RX5CSHskp8|e2Iei|%$*%O>z}1!A%rZdVq$)o3j_RmJ||hr2%sOb2=vkC
zz}cczO8DT<GDD`~>%J;rs|;6LYrzK%!-TQ{I~70yaC#q+U6i5+jO>wa#V~mu+^8S=
zu;Vnq0STz6@&>t+`JD%WQtnhMMI(3m_5H3)DkR)A?1XFDuvtCn)`oeUwy|L~?Q^yn
znF&NIFd#r3Wv`8ljo19J$bhfs5@?MYD$sl!ZwN(;KqK_I3e@z4KH{+eCTZde^aZT?
zNTa(W&_F7HL~DRLPP;XPttA+EQfBS%@lS!lays5I094O;uG=%9|A<HBAJE_qV=T2A
zXYWlF(L?dBXM_n`0Nuv8?8IN?bK&8AvK049Mn=WQr}?#^;Yg7dkB;YN6|(=c{|NQ9
zoxA@9Uw-ocSJVae-{G(S@?jvAfAA-WZvTIRxxV!LPOITo+=<{9(2UtQ^9kDmflN80
z6cU<7IshZ!%MU_fFc^K6<BBYbZyaT8qMph;xCh-!udU_1MZznO(tp5WIq%Gg#b=v4
z=Q;Or%J3!2@+J4tA4Mc4uHU52kx5kbA94QzIMzM;dI6e3j9+e?wGonsfjjKT0ALCJ
zMdnORX@`;ur~*hrU$G6uh2Lf46SvjRG>r-qc%<<096Dg?`CQy>b?4Q!*Kts{u|4&n
z6kK1c{1*6j8Y2p}%Lld1%g;7@S6PlR7fAi_1eD)^JqL^41@_dvb-JZRwCQ}Wo0!Ao
zgJz}KXJEkNv-8f+y+}oTv?pt9$M%-G@kvOm>{0d)Hs^`*CPaw<s7Wvt0QO?K-kG6l
zmK8TJIG6zfsudWkp<6st@?-zduzdrX2{xzp{{4F=Fc)yeRAKL~fQFHX)YOurSxAp(
z+9_&bVc`KF#|E6;Yt3g{N)8SVS)DfH<?R)7svaKoROY~*CJ1}iGdCZ<26`$Fy1G1_
zoW0C$Y;1HKjZ981ieXT*I6YXaKi`A8@$cWy2oJyGyxK3Olr23_@4Vh<Gc%ST{AF!z
zErHE22rO+Hc>i#9;5kC%s<FJhysv<4h_eciZ2_^DWn*JQ*?wS>vcQ%>c(sjbJB9)|
z=m>qZ!ncJYKp(e44rYa{lmlqshGaO22aN1|mbqKjsJTHx0^)$8*jMrV{POI?7B~T+
zhYemQ4yY>w<(<Vm+vHi}vOQD8GwY=UG^6tC*XhrmJ?kQYe!tPOsnObBL5u;)be((|
zMa4+)b1<wVw+0y6negwV^v>JSva^>@c>|C#GSlRVQpPB&lIs0Co`QlR-yv#USq4hc
z1JClh9Xs?PTe@=M=46c}U=#P44<tOeIGQbRT<OCfHLRI*9+KfP2m3dRS+CkEK9)%<
zTeryrb)%S}wdkQWWd#L=JQwP!XuF1Khlc4aW`VJEr;#R<MEUaQUuOY4CG7}=96%P)
zR464Ee48}iB^8H|c^LX0LD?M;_N5wZ{#&H{mD1rymf)sL_qS4*XlNt=?fD=uIt9wM
z?=mx~ZjlMr&V>CcTH8O{YS@mwDUOH*LIHT^<8B&Deag-QpNzZdil{d3CIFB{5iEVo
zsG`ch#!&;Db{Nc95D-_VZ2sPyT;J^^A6tfc6A?s@V`+qhbinFIP`wlvN1{0BBPb<)
zIq@$6XG>l_4e-<7B4U3LS!us0&ZJSEIVfz}mqZ6a5R-bTC@?zUk;^AHPcP55g_K_L
zCd*E{uO#FDW^?!@d2u#-H42b6xXK*SlZ3>?0`M3*?n^O|@BP06Gp-HoY!)+A+54n=
z?ZIEZctOj?HsI)!ESpnpGo62Y>;_j97Y}MaUI=jr(*uTFwM3UG&n=?4Vw)zo?K8R0
zv6z~+ec))JMg|a=SiWYk#o1+)L@tjrR?5ktbA@d4*wc7EN0XvyRPdq2@)P)b$?bSP
zSO#wF12$0)9Ey4xzuk0{5}%Ud)8_-=menivjP&%Lp&U82pM6UX@oho3IH=<RRUTu7
z2sL$<GAY>KEH&i8Z3aLrr-!!sKR;iAG9VR%FB-Zud^wb7M%^Y6sBpJp)5&UUmHL#V
zBoaAtJi(n74063kD`O@V@qjOpp(b^)Gt6^h8Uo;HshW<@HYS-cuG{!26_wjWjO$0&
zt*N!~FF+8mZMCeY{tRA=^$kh`tsjo*sR05(U~g5Cdl?d876a)o>o0y!jsU0tDF0}=
zxT>avlVo!Um^7*29>8!gBps>Jrw}i!<)p?c3))&4<7E9=Am&R|_SaD5H`7#aRRfWj
z-TnO`bUZ--_p2;aVor0dtDJqnGUu7}Cd4$baC57oW8#KWSx?tj^;|YK3L<i+Zr*>I
z3-%WNC<k<Tj%}G+jQnS$iGchG=8LW4v6iLT<Wc3|=!jAle-tK{MKpkZg4UxYdL_rl
z#}3b-e((q$%bra~GoJVWJONFEkOQFrK(K67F%EvC!^7wJegjVhf0WC7Fw=2V=QZ$U
z@T&#v7XoHH)|o=dgfy)ut5tyOl0yLK$|r3aWE6M3if#vI46au`vreAX3qJiaTA;>)
zIJ}wwVl6mXGwb7-b@8pUlYx^HsvP82QAwTzf&p`j`s0DZ4)xsJYEgDx%pY?N4?6aj
z9#RGbe7ewua}$96k(qZX#aF+*yAL$<9*L`q8!*yKPW*^7eg+(6mipE7S7CsUSqZ3-
zn1qCU08L^hm+Mwi_CJ^jkpmx<J8`pdax8Fy|I6LGcVi&3M<gVSaECf=<KM}oBsGAr
zaL@<%U_gUcf!D7lRbD#Vpwhr|G7rB33(EY-;NknQu(7E+$HTZU=PwoQ<$oif%L?U&
z32X#zTdry%5U8U!t<$GnUHe*hhd9?a>I+v<hB4Rnj~S|xP7xfhxaj4Xjh-NU^axrX
zzXX`XC}G+C3e1R_TQW*FRX(kmENb#RdKC%0AT1NUA5-;Tzkc;tNp^LbYe8FDUhV=h
zmrWN>3w7X3xhJC+43C}^ef;$4A>w!477cu&%uu3e7-ct)vH#dpCV+QRHqXz`70-Zl
zEsbC9wBcE9@Xtb5(8a4G+u7#n3LW+_i+52pa`ZerYLS(0yRWD1J4pek3T<s`TUuFB
zv-o9dElnyPWwCy)Dgl6_;3T=%j#!rU(dMMu+G|@=hIq?|;D(~+D14KqnF3w9Bft)A
zlG?<%eR~wZ0BX2x(_t(SJUz-WKIqa5SlxmtZwQPmf!{fQ<?4&B)%u3Jh19YE69Q<g
zLsV#?VsJu5Yo&%CjA0~4jvPMdx!shL>T{Ym<pPWkz%g=J-pNfcEpc)2UVK!Fhx75D
z+kf!6I9-cT(;Walt@FXE0+4KvWmPU2rMkL0l;`#KE+r=z0>a)9>i9r6GzMIjU*ocp
zlKqwDLzSLKHshNvh$ZGYDWF?X9}i(^cGLZ85k@9-Fdt0rA?{nR&3pl-keF59S5Z-s
z3jRF!Hd95v8i1felde<t!Km%^=FK&PwSY4+nt)L~4Il-b{&ZvfwW+#_KR8V*3nx)U
z#X1?9MY;iFCpd|+G+5+bb9Jd8?DczNJj0+Z$h<d!E9UooQqs+FpWSxCnD}@Nfj{Jj
zx*!q`bWxf4&{Ltk@02Js2bKgtl7;x@7GUUrWFseQN&tE6Jl}G0!`2{x!HU@>+}0DT
zz=hjAzXEDf9_2KVKqMfUX7n)V+O=ySfRxqjeTGEYC9qHFi)?|14`PA!Ac%oaMz*z<
z0hA5O$WX&CFE2gLHfy>`EGxzZDJA5R`LhB7t_KDMdDv(H(%Nj<<oxt>nO5C67z{lh
z{z6Qd0ysED^f@Kbz<&XVavwXyv^QbtrauBYR7&UWvvt;S9#J7?H+oC;W8The;oG~E
zvmd~f2Wo#9aOcK^ho^xRg-4#BcQrEtCPSqfn`tUIK$wak4uRXYljKjtVKR)|1QGyi
zL_-C@a;rPZA#I?k3hb^(U8g<+TSrSzpD$jFdMJ;@P#*fJ*<1z6EGRQG*Li((X$#t{
z8Wa3#;Qq$nzsh-C_2PHUmm-%U!h9-R;W5Ox#p$_foWVoQItSzUHMYLdEGfbw&8jXA
zc0me?^eGq9usbXY4!5<L)p{S`f-$(2{Y#S};yFw!4LfgB)6k$C;E6iNoZs8CCBP<)
zTX6$;D+>WIJ+?F7#`XH!4Pg6IDiEKa2P46D(QI%jy4($eRsqNvxx6?74ooI+6Z0*{
ziitVR!ZGp43R8qURVToxJXSvkbDH%(?oH&816H>TL<NT{OyvN~7DI(OiGe>ozx(9L
zlZkTUkXS~Iboc#bI>0dub{=vT`tqgY15pMll0rf84Cu^;IY-l&1i>#{Pk4FN!3+<A
z`R8)o(nf`WZU;a2$GkNNR4<o7>-CefGf#qVAz@)GHK|~A%mH9e5b~@B7y+faM!yde
zfte6}Zt#h^?J0PLECb{7B(N9oIxYoD2*fILi**_xmpou)BQpVU@%rg{Nlh&U_-M-+
z>r0f*MQST3L;y^%aRLzZMG*ta4)47F<3Q!wTaFeE=v|$vU4U2*h)BhmCjjUH3?>Re
z4+9U4nwlCODIP{vw7O~*^5?6K_Z@*@gr#*P^Eryk3GN1@xGg~MUQL1rLB%edmGQy#
z`*;4>d06aQ5Xt~B4Q1B<@j)61C{bAn`9&1g2J*bN7$!0rr{{wTTGIopSXNd>%xjm{
z5k_7D3=}ObNSpj2kS{U$lx?o5^G^~N7?M`DKMx1)1#_c={%2_l)QA7?$%Oxz%Za+y
kzdcS`)a(E4%7<5I)>j0sES@jcQ18KC%DyOmrvK)D0X&BX6#xJL

diff --git a/dev/articles/lazy_tensor.html b/dev/articles/lazy_tensor.html
index c59a49d7..c2c7967c 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.563486.x.output</span></span>
+<span><span class="co">#&gt; * pointer: nop.09428b.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.563486.x.output</span></span>
+<span><span class="co">#&gt; * pointer: nop.09428b.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.563486.x        &lt;list&gt;   poly.x             </span></span>
-<span><span class="co">#&gt;        poly.x &lt;&lt;UNTRAINED&gt;&gt;          nop.563486.x</span></span></code></pre></div>
+<span><span class="co">#&gt;  nop.09428b.x        &lt;list&gt;   poly.x             </span></span>
+<span><span class="co">#&gt;        poly.x &lt;&lt;UNTRAINED&gt;&gt;          nop.09428b.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 a84a9cac..540cf350 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: 2024-10-29T11:35Z
+last_built: 2024-11-08T09:47Z
 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 7ddde268..e032eba3 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.6c4b31.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.1567d7.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.6c4b31.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.1567d7.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 18536efe..cb5a699d 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/Rtmpmt7gXp/file657f35f35d4b, freq=1</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * Parameters: path=/tmp/RtmpO2o111/file284f69a24b0c, 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 3a347f5f..7a32c29a 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.fe081b.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.d10b3c.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.d2a8ff.x.output</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> * pointer: nop.4305b8.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 1231531f..8c12c050 100644
--- a/dev/reference/materialize.html
+++ b/dev/reference/materialize.html
@@ -132,86 +132,87 @@ <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> -1.3420 -0.7548 -0.2027</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8642 -0.6203 -1.1163</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1052  0.5600 -0.3974</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3943 -0.0705 -1.7041</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.5337  0.1588 -0.8624</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4815 -0.1104 -0.3846</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.3301  1.1064  0.2867</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1371 -1.4983  0.6197</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3968  1.8850  0.5676</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4147  1.3346 -1.3396</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2337  0.5821 -0.4541</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0929 -0.0327 -0.0007</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7647  1.1990  1.1332</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2069  0.9962 -0.2510</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4641 -0.2855 -0.6539</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2911  0.3456 -0.2388</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2894  0.6156  1.3607</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2922 -2.0550  1.3359</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3164  0.3312 -1.1026</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4172 -0.1749  1.6354</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> -1.3420</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7548</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2027</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2337</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5821</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4541</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>  0.8642</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6203</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1163</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> -9.2944</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -3.2728</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0712</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> -0.1052</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5600</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3974</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7647</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1990</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1332</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> -0.3943</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0705</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.7041</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2069</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9962</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2510</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>  1.5337</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1588</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8624</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4641</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2855</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6539</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>  1.4815</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> -0.3846</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2911</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3456</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2388</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>  2.3301</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1064</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2867</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2894</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6156</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3607</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> -1.1371</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4983</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6197</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2922</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.0550</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3359</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>  0.3968</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8850</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5676</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3164</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3312</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1026</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>  1.4147</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3346</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.3396</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4172</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1749</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6354</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 +220,185 @@ <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> -1.3420 -0.7548 -0.2027</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8642 -0.6203 -1.1163</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1052  0.5600 -0.3974</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3943 -0.0705 -1.7041</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.5337  0.1588 -0.8624</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4815 -0.1104 -0.3846</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  2.3301  1.1064  0.2867</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1371 -1.4983  0.6197</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3968  1.8850  0.5676</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4147  1.3346 -1.3396</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2337  0.5821 -0.4541</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0929 -0.0327 -0.0007</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7647  1.1990  1.1332</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2069  0.9962 -0.2510</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4641 -0.2855 -0.6539</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2911  0.3456 -0.2388</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2894  0.6156  1.3607</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2922 -2.0550  1.3359</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3164  0.3312 -1.1026</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4172 -0.1749  1.6354</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> -0.1006  1.3168  0.1388  0.5232</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2336 -0.0717 -1.0036  0.6683</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7921  1.4544  0.1958 -0.2346</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6385 -0.0318  0.6482 -0.1380</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3797  0.7916 -0.2373 -1.0011</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3849 -0.2163 -0.0237 -0.0863</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4032  0.1983 -1.6133 -2.2197</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2452  0.7073  0.3545  0.0833</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8829  0.2502 -0.3007 -0.1676</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.9140 -0.7256 -0.7844 -1.4341</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3209 -0.9323 -0.4262  0.4701</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6289 -1.5827  1.4104 -0.9415</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6591 -1.3246 -0.3064  0.8200</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0953 -0.7571  2.1173 -0.8233</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2794 -1.1853 -1.6339  0.4039</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4075  1.4175 -0.1898 -0.2829</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2797 -0.2720  0.2012  0.6381</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1676  2.0197 -0.6833  1.2002</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.7206 -0.0778  0.1583 -1.5801</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1183 -0.4630 -0.5408 -0.7640</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> -1.3420</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7548</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2027</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2337</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5821</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4541</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>  0.8642</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6203</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1163</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> -9.2944</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -3.2728</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0712</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> -0.1052</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5600</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3974</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7647</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1990</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1332</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> -0.3943</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0705</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.7041</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2069</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9962</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2510</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>  1.5337</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1588</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8624</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4641</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2855</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6539</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>  1.4815</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> -0.3846</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2911</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3456</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2388</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>  2.3301</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.1064</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2867</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.2894</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6156</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3607</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> -1.1371</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4983</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6197</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2922</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.0550</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3359</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>  0.3968</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.8850</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5676</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3164</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3312</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.1026</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>  1.4147</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3346</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.3396</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4172</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1749</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6354</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> -0.1006</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.3168</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1388</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5232</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3209</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9323</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4262</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4701</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>  0.2336</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0717</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0036</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6683</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6289</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.5827</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4104</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.9415</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>  0.7921</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4544</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1958</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2346</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6591</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.3246</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3064</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.8200</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>  0.6385</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0318</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6482</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1380</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0953</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7571</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  2.1173</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.8233</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>  1.3797</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7916</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2373</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.0011</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> -1.1853</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.6339</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4039</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>  0.3849</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2163</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0237</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0863</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4075</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4175</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1898</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2829</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>  0.4032</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1983</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.6133</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -2.2197</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2797</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2720</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2012</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6381</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>  0.2452</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.7073</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3545</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0833</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1676</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  2.0197</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.6833</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.2002</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> -0.8829</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2502</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.3007</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1676</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.7206</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0778</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1583</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -1.5801</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>  1.9140</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7256</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7844</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -1.4341</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1183</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.4630</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5408</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.7640</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_learners.torchvision.html b/dev/reference/mlr_learners.torchvision.html
index f81e4292..4717a642 100644
--- a/dev/reference/mlr_learners.torchvision.html
+++ b/dev/reference/mlr_learners.torchvision.html
@@ -133,9 +133,9 @@ <h4 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 Sandler, Mark, Howard, Andrew, Zhu, Menglong, Zhmoginov, Andrey, Chen, Liang-Chieh (2018).
 “Mobilenetv2: Inverted residuals and linear bottlenecks.”
 In <em>Proceedings of the IEEE conference on computer vision and pattern recognition</em>, 4510–4520.
-He, Kaiming, Zhang, Xiangyu, Ren, Shaoqing, Sun, Jian (2016).
-“Deep residual learning for image recognition.”
-In <em>Proceedings of the IEEE conference on computer vision and pattern recognition</em>, 770–778.
+He, Kaiming, Zhang, Xiangyu, Ren, Shaoqing, Sun, Jian (2016 ).
+“Deep residual learning for image recognition .”
+In <em>Proceedings of the IEEE conference on computer vision and pattern recognition </em>, 770–778 .
 Simonyan, Karen, Zisserman, Andrew (2014).
 “Very deep convolutional networks for large-scale image recognition.”
 <em>arXiv preprint arXiv:1409.1556</em>.</p></dd>
diff --git a/dev/reference/mlr_learners_torch_image.html b/dev/reference/mlr_learners_torch_image.html
index 3189c0a0..56812433 100644
--- a/dev/reference/mlr_learners_torch_image.html
+++ b/dev/reference/mlr_learners_torch_image.html
@@ -109,7 +109,7 @@ <h4 id="usage">Usage<a class="anchor" aria-label="anchor" href="#usage"></a></h4
 <span>  optimizer <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  loss <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  callbacks <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="op">)</span>,</span>
-<span>  packages <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="st">"torchvision"</span>, <span class="st">"magick"</span><span class="op">)</span>,</span>
+<span>  packages <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="st">"torchvision"</span><span class="op">)</span>,</span>
 <span>  <span class="va">man</span>,</span>
 <span>  properties <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  predict_types <span class="op">=</span> <span class="cn">NULL</span></span>
diff --git a/dev/reference/mlr_learners_torch_model.html b/dev/reference/mlr_learners_torch_model.html
index 9dcd7ab2..710171ea 100644
--- a/dev/reference/mlr_learners_torch_model.html
+++ b/dev/reference/mlr_learners_torch_model.html
@@ -243,14 +243,14 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 <span class="r-in"><span><span class="va">learner</span><span class="op">$</span><span class="fu">train</span><span class="op">(</span><span class="va">task</span>, <span class="va">ids</span><span class="op">$</span><span class="va">train</span><span class="op">)</span></span></span>
 <span class="r-in"><span><span class="va">learner</span><span class="op">$</span><span class="fu">predict</span><span class="op">(</span><span class="va">task</span>, <span class="va">ids</span><span class="op">$</span><span class="va">test</span><span class="op">)</span></span></span>
 <span class="r-out co"><span class="r-pr">#&gt;</span> &lt;PredictionClassif&gt; for 50 observations:</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  row_ids     truth response</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>        3    setosa   setosa</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>        9    setosa   setosa</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>       10    setosa   setosa</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>      ---       ---      ---</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>      147 virginica   setosa</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>      149 virginica   setosa</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>      150 virginica   setosa</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  row_ids     truth   response</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>        3    setosa versicolor</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>        9    setosa versicolor</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>       10    setosa versicolor</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>      ---       ---        ---</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>      147 virginica versicolor</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>      149 virginica versicolor</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>      150 virginica versicolor</span>
 </code></pre></div>
     </div>
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
diff --git a/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html b/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html
index 646bf68b..b0b6d3e7 100644
--- a/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html
+++ b/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html
@@ -81,7 +81,7 @@ <h2 id="credit">Credit<a class="anchor" aria-label="anchor" href="#credit"></a><
     <h2 id="parameters">Parameters<a class="anchor" aria-label="anchor" href="#parameters"></a></h2>
 
 
-<ul><li><p><code>output_size</code> :: <code><a href="https://rdrr.io/r/base/integer.html" class="external-link">integer()</a></code><br>
+<ul><li><p><code>output_size</code> :: <code>integer(1)</code><br>
 The target output size. A single number.</p></li>
 </ul></div>
     <div class="section level2">
diff --git a/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html b/dev/reference/mlr_pipeops_torch_ingress_ltnsr.html
index 3480a19b..e7cae3b1 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>   0.1683  0.9521 -0.3007 -0.6531 -0.5466 -0.2548</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.8280  0.7439 -1.1441 -0.2175  0.6100 -1.8303</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.6169  1.3783  0.2681 -0.2965 -0.6230  0.1022</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.4736  0.9308  0.0779  0.6800 -0.6026  0.1194</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.7146  0.7433 -0.7192  0.7677  0.5538  0.4407</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3748  0.1611 -0.8856 -0.3067 -0.1230  0.5929</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.2125 -0.3882 -0.2121  0.1723  0.0384  0.4917</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.6026 -0.0511  0.8008  0.5735  1.5137  1.0633</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.2825 -0.0818 -0.7123 -0.4431  0.5773 -0.0926</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.2583  0.0694  0.5027  0.7375 -0.0705 -0.2815</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.4183  1.7796  0.2513 -1.4076  1.3036 -0.8148</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.2644 -0.3981  0.8998 -0.2581 -0.5860  0.3477</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>  -0.1821 -0.2982  1.1122  0.6025 -0.0759 -0.8357</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1558 -0.9088 -0.7307  0.6930 -0.1729  0.8293</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.4416 -0.2946  0.4691  0.5564 -0.7042 -1.1576</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.5102  0.1238 -0.2319  0.7806 -1.2194 -1.4314</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.5757  0.0112 -1.0029  0.4919  0.4147 -0.0619</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   1.1526  0.4710 -0.9762 -1.2955 -0.1328  0.0883</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.7151  0.2609  0.0981 -0.1435 -0.1861  1.0630</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.4236 -0.0667 -0.5166 -0.0989 -0.6529 -0.2784</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3545  0.3712  0.1984 -0.1119 -1.0725  0.2314</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.0875  0.5762  0.1458  0.9248  0.1061  0.9214</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.2276  1.0706 -0.1771  0.3537  0.8959  0.9463</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.2926 -0.6330  0.4849 -0.1984  1.0717  0.2909</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>  -0.2915 -0.8742  0.2805 -0.3131 -0.0847  0.3623</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.8970  1.4281  1.1198 -0.4452 -0.3298 -0.7621</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.6495  0.3189  0.3577  0.3509  0.4461  0.3858</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.2846 -1.6956 -0.5109  1.1558  0.6108  0.1983</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3220  0.4607  0.1320  0.0521 -1.6308 -0.7962</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.9437  1.8788 -0.5748  0.0037 -0.1264 -0.8996</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   1.2679 -0.7824 -0.5283 -0.2615 -0.5069  0.0035</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.3564 -0.2301  1.1684 -0.1398  0.2603 -0.6686</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.0015 -0.1906  0.3344 -0.8316 -1.3696  0.6962</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.3337 -0.2110  0.5441 -0.6840  0.1694 -0.5514</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.1862  0.9605  0.0916  0.1290 -1.8815 -1.3006</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   0.1662 -0.0681  0.1811 -0.6044 -0.8783 -0.3254</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>   0.2720  0.9749 -1.7862  0.7320 -0.4558 -0.2841</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.6081  0.0487  0.1334 -0.1368 -0.8086 -0.4014</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.7191  0.0231 -0.3324 -0.8849 -0.8758 -0.2848</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  -1.2964  0.1189  0.2938  0.0605 -0.3840 -0.5428</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   0.3190  0.1164 -0.5326  1.0797  0.4521 -0.6165</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.5820  0.8655  0.3351 -0.5340  0.3202  0.0760</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.5524  0.1915  0.4468 -0.0970 -0.1182 -0.5823</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   1.6456  0.5373  0.6732 -0.2632 -1.3037 -0.0948</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  -0.1056 -0.2319 -0.1575  0.0060  0.3981 -0.5047</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   1.1163  0.8038  0.6666 -0.0219 -0.1799 -0.3924</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 f324daf3..0b79ed88 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> -0.0813 -0.0115 -0.3071 -0.2139</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.6080  0.2067  0.9018  0.4727</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.5266 -0.1952 -0.5947 -0.2588</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.4871  0.1782  0.5019  0.1886</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2023 -0.0594 -0.2875 -0.1301</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2849 -0.1188 -0.2144 -0.0584</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>  0.1392  0.0041  1.6147  0.7540</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   3.0754  0.3983  5.4061  1.4248</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>   3.0210  0.4019  4.2777  0.8787</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  1.6796  0.2009  3.1383  0.7971</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   1.3630  0.1358  2.6025  0.5677</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>   2.0308  0.2634  3.2504  0.7642</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 +294,18 @@ <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> -0.0619</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1498</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0879</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>  7.2249</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -4.9365</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -2.2885</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>  0.6434</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  1.4496</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.9899</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> 0.0001 *</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  8.7081</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  7.3037</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  8.2450</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>
diff --git a/dev/reference/mlr_pipeops_torch_model_regr.html b/dev/reference/mlr_pipeops_torch_model_regr.html
index fc4ca703..40a98464 100644
--- a/dev/reference/mlr_pipeops_torch_model_regr.html
+++ b/dev/reference/mlr_pipeops_torch_model_regr.html
@@ -274,14 +274,16 @@ <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> -8.7515 -70.6545 -155.1045 -6165.2896 -82.2042 -85.4106 -3880.7810 -405.2719 -7.2561 -78.5860</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 1 to 8-11.2135 -85.3135 -177.3723 -6927.4478 -95.3926 -100.3730 -4563.8008 -463.1545</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 -8.4023 -89.0784</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 4.7060e+00  2.7571e+02  1.3329e+03  2.1144e+06  3.7363e+02  4.0293e+02</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 1 to 6 8.0234e+00  4.0437e+02  1.7375e+03  2.6535e+06  5.0269e+02  5.5943e+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 8.3214e+05  9.1789e+03  3.2087e+00  3.4837e+02</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> Columns 7 to 10 1.1598e+06  1.1890e+04  4.1586e+00  4.4203e+02</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>
@@ -291,12 +293,12 @@ <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> -23.3342</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -26.8348</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>  30.1836</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  39.7718</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>
@@ -333,7 +335,7 @@ <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] 0.039</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>
diff --git a/dev/reference/nn_graph.html b/dev/reference/nn_graph.html
index e7c11748..4dbaab6e 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>  0.2515</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0529</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0081</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2387</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0763</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0844</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2905</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1252</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0067</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1168</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0734</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.3246</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1031</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.5290</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span> -0.0672</span>
-<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0823</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0954</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1976</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2559</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.2567</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1011</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1625</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1907</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.2813</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1443</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1155</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0968</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span> -0.1137</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1292</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.0884</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1634</span>
+<span class="r-out co"><span class="r-pr">#&gt;</span>  0.1503</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 c9871a54..2c939dd3 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: 0x55f6ed50fb38> ##   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$generator #> <nn_l1_loss> object generator #>   Inherits from: <inherit> #>   Public: #>     .classes: nn_l1_loss nn_loss nn_module #>     initialize: function (reduction = \"mean\")  #>     forward: function (input, target)  #>     clone: function (deep = FALSE, ..., replace_values = TRUE)  #>   Private: #>     .__clone_r6__: function (deep = FALSE)  #>   Parent env: <environment: 0x556c58224810> #>   Locked objects: FALSE #>   Locked class: FALSE #>   Portable: TRUE mlp_l1 = lrn(\"regr.mlp\", loss = l1) mlp_l1$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 #>  0.9197  0.6295 -0.9055 #> -2.5884  0.7595  1.2294 #> [ CPUFloatType{2,3} ] module_1 = nn_linear(in_features = 3, out_features = 4, bias = TRUE) output = module_1(input) output #> torch_tensor #>  0.6356 -0.0022  0.3491 -1.0918 #> -1.5718  1.8213 -1.6298  0.2341 #> [ 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 #>  0.2569  0.4082  0.3350 #>  0.3488  0.3890  0.2623 #> [ 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 #>  0.6356 -0.0022  0.3491 -1.0918 #> -1.5718  1.8213 -1.6298  0.2341 #> [ 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 #>  0.2569  0.4082  0.3350 #>  0.3488  0.3890  0.2623 #> [ 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 #> $module_list #> An `nn_module` containing 31 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • 0: <nn_linear> #16 parameters #> • 1: <nn_sigmoid> #0 parameters #> • 2: <nn_linear> #15 parameters #> • 3: <nn_softmax> #0 parameters graph_module(input) #> torch_tensor #>  0.2569  0.4082  0.3350 #>  0.3488  0.3890  0.2623 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"modeldescriptor","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"ModelDescriptor","title":"Internals","text":"PipeOpModule represents nn_module fixed specific tensor shape hyperparameters. constructing neural network using operators, one take care output shape operations match input shapes following operations. complete Graph matching PipeOpModules can constructed using operators mostly inherit PipeOpTorch, making use ModelDescriptor class. ModelDescriptor class contains Graph (mostly) PipeOpModules information. PipeOpTorch transforms ModelDescriptor adds PipeOpModules Graph. ModelDescriptors always build Graph specific Task. easiest way initialize proper ModelDescriptor use appropriate PipeOpTorchIngress given datatype. use PipeOpTorchIngressNumeric, used numeric data. ModelDescriptor S3 object contains Graph, information generate data ($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 #>  0.2935  0.0055  0.1483  0.6004 #>  0.3216  2.2073 -0.5215 -0.5771 #> [ 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 #>  1.7664  4.3770 -3.4375 -0.1343 #>  1.7621  4.0245 -3.1681 -0.1304 #>  1.6213  4.0500 -3.1483 -0.0796 #> [ 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 #>  0.2308  0.4097  0.3595 #>  0.1968  0.3787  0.4245 #> [ 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 #> 0.01 * #> -8.6465 #>  -9.1034 #> [ CPUFloatType{2,1} ][ grad_fn = <AddmmBackward0> ] #>  #> $nn_softmax.output #> torch_tensor #>  0.3961  0.4251  0.1788 #>  0.3951  0.4239  0.1810 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ] iris_module$graph$pipeops$linear1$.result #> $output #> torch_tensor #>  2.1381 -1.4562 -2.3068 -3.0385 #>  2.0602 -1.6112 -2.1434 -2.9874 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$linear3$.result #> $output #> torch_tensor #> -0.4947 -0.6350  0.1133 -0.4768  0.0452 #> -0.4947 -0.6350  0.1133 -0.4768  0.0452 #> [ CPUFloatType{2,5} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$nn_merge_cat$.result #> $output #> torch_tensor #>  2.1381 -1.4562 -2.3068 -3.0385 -0.4947 -0.6350  0.1133 -0.4768  0.0452 #>  2.0602 -1.6112 -2.1434 -2.9874 -0.4947 -0.6350  0.1133 -0.4768  0.0452 #> [ 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$ingress,   optimizer = adam,   callbacks = history,   loss = xe )  lr_sequential #> <LearnerTorchModel[classif]:classif.model>: Torch Model #> * Model: - #> * Parameters: device=auto, num_threads=1, num_interop_threads=1, #>   seed=random, eval_freq=1, measures_train=<list>, #>   measures_valid=<list>, patience=0, min_delta=0, opt.lr=0.02 #> * Validate: NULL #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response], prob #> * Feature Types: logical, integer, numeric, character, factor, ordered, #>   POSIXct, lazy_tensor #> * Properties: featureless, hotstart_backward, hotstart_forward, #>   importance, internal_tuning, loglik, marshal, missings, multiclass, #>   oob_error, selected_features, twoclass, validation, weights #> * Optimizer: adam #> * Loss: cross_entropy #> * Callbacks: history lr_sequential$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.563486.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.563486.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.563486.x        <list>   poly.x              #>        poly.x <<UNTRAINED>>          nop.563486.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.6c4b31.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.6c4b31.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/Rtmpmt7gXp/file657f35f35d4b, 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(36)> #>                          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:              activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:         activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:                   shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:                  opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 29:               opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 30:                 opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 31:        opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 32:             opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 33:          loss.reduction ParamFct    NA    NA       2           mean #> 34:      cb.checkpoint.path ParamUty    NA    NA     Inf <NoDefault[0]> #> 35:      cb.checkpoint.freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 36: 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: <nn_relu[1]> #> 26:    <list[0]> #> 27:       [NULL] #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [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(33)> #>                      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:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:              opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 29:           opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 30:             opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 31:    opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 32:         opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 33:      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: <nn_relu[1]> #> 26:    <list[0]> #> 27:       [NULL] #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [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(33)> #>                      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:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:              opt.lr ParamDbl     0   Inf     Inf <NoDefault[0]> #> 29:        opt.momentum ParamDbl     0     1     Inf              0 #> 30:       opt.dampening ParamDbl     0     1     Inf              0 #> 31:    opt.weight_decay ParamDbl     0     1     Inf              0 #> 32:        opt.nesterov ParamLgl    NA    NA       2          FALSE #> 33:      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: <nn_relu[1]> #> 26:    <list[0]> #> 27:       [NULL] #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [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.fe081b.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$iris) )() as_data_descriptor(ds2) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,4)] #> * input_map: (x) -> Graph #> * pointer: nop.d2a8ff.x.output #> * shape: [(NA,4)]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to Lazy Tensor — as_lazy_tensor","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"Convert object lazy_tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"as_lazy_tensor(x, ...)  # S3 method for class 'dataset' as_lazy_tensor(x, dataset_shapes = NULL, ids = NULL, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"x () Object convert lazy_tensor ... () Additional arguments passed method. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. ids (integer()) ids include lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"iris_ds = dataset(\"iris\",   initialize = function() {     self$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 #> -1.3420 -0.7548 -0.2027 #>  0.8642 -0.6203 -1.1163 #> -0.1052  0.5600 -0.3974 #> -0.3943 -0.0705 -1.7041 #>  1.5337  0.1588 -0.8624 #>  1.4815 -0.1104 -0.3846 #>  2.3301  1.1064  0.2867 #> -1.1371 -1.4983  0.6197 #>  0.3968  1.8850  0.5676 #>  1.4147  1.3346 -1.3396 #> [ CPUFloatType{10,3} ] materialize(lt1, rbind = FALSE) #> [[1]] #> torch_tensor #> -1.3420 #> -0.7548 #> -0.2027 #> [ CPUFloatType{3} ] #>  #> [[2]] #> torch_tensor #>  0.8642 #> -0.6203 #> -1.1163 #> [ CPUFloatType{3} ] #>  #> [[3]] #> torch_tensor #> -0.1052 #>  0.5600 #> -0.3974 #> [ CPUFloatType{3} ] #>  #> [[4]] #> torch_tensor #> -0.3943 #> -0.0705 #> -1.7041 #> [ CPUFloatType{3} ] #>  #> [[5]] #> torch_tensor #>  1.5337 #>  0.1588 #> -0.8624 #> [ CPUFloatType{3} ] #>  #> [[6]] #> torch_tensor #>  1.4815 #> -0.1104 #> -0.3846 #> [ CPUFloatType{3} ] #>  #> [[7]] #> torch_tensor #>  2.3301 #>  1.1064 #>  0.2867 #> [ CPUFloatType{3} ] #>  #> [[8]] #> torch_tensor #> -1.1371 #> -1.4983 #>  0.6197 #> [ CPUFloatType{3} ] #>  #> [[9]] #> torch_tensor #>  0.3968 #>  1.8850 #>  0.5676 #> [ CPUFloatType{3} ] #>  #> [[10]] #> torch_tensor #>  1.4147 #>  1.3346 #> -1.3396 #> [ 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 #> -1.3420 -0.7548 -0.2027 #>  0.8642 -0.6203 -1.1163 #> -0.1052  0.5600 -0.3974 #> -0.3943 -0.0705 -1.7041 #>  1.5337  0.1588 -0.8624 #>  1.4815 -0.1104 -0.3846 #>  2.3301  1.1064  0.2867 #> -1.1371 -1.4983  0.6197 #>  0.3968  1.8850  0.5676 #>  1.4147  1.3346 -1.3396 #> [ CPUFloatType{10,3} ] #>  #> $lt2 #> torch_tensor #> -0.1006  1.3168  0.1388  0.5232 #>  0.2336 -0.0717 -1.0036  0.6683 #>  0.7921  1.4544  0.1958 -0.2346 #>  0.6385 -0.0318  0.6482 -0.1380 #>  1.3797  0.7916 -0.2373 -1.0011 #>  0.3849 -0.2163 -0.0237 -0.0863 #>  0.4032  0.1983 -1.6133 -2.2197 #>  0.2452  0.7073  0.3545  0.0833 #> -0.8829  0.2502 -0.3007 -0.1676 #>  1.9140 -0.7256 -0.7844 -1.4341 #> [ CPUFloatType{10,4} ] #>  materialize(d, rbind = FALSE) #> $lt1 #> $lt1[[1]] #> torch_tensor #> -1.3420 #> -0.7548 #> -0.2027 #> [ CPUFloatType{3} ] #>  #> $lt1[[2]] #> torch_tensor #>  0.8642 #> -0.6203 #> -1.1163 #> [ CPUFloatType{3} ] #>  #> $lt1[[3]] #> torch_tensor #> -0.1052 #>  0.5600 #> -0.3974 #> [ CPUFloatType{3} ] #>  #> $lt1[[4]] #> torch_tensor #> -0.3943 #> -0.0705 #> -1.7041 #> [ CPUFloatType{3} ] #>  #> $lt1[[5]] #> torch_tensor #>  1.5337 #>  0.1588 #> -0.8624 #> [ CPUFloatType{3} ] #>  #> $lt1[[6]] #> torch_tensor #>  1.4815 #> -0.1104 #> -0.3846 #> [ CPUFloatType{3} ] #>  #> $lt1[[7]] #> torch_tensor #>  2.3301 #>  1.1064 #>  0.2867 #> [ CPUFloatType{3} ] #>  #> $lt1[[8]] #> torch_tensor #> -1.1371 #> -1.4983 #>  0.6197 #> [ CPUFloatType{3} ] #>  #> $lt1[[9]] #> torch_tensor #>  0.3968 #>  1.8850 #>  0.5676 #> [ CPUFloatType{3} ] #>  #> $lt1[[10]] #> torch_tensor #>  1.4147 #>  1.3346 #> -1.3396 #> [ CPUFloatType{3} ] #>  #>  #> $lt2 #> $lt2[[1]] #> torch_tensor #> -0.1006 #>  1.3168 #>  0.1388 #>  0.5232 #> [ CPUFloatType{4} ] #>  #> $lt2[[2]] #> torch_tensor #>  0.2336 #> -0.0717 #> -1.0036 #>  0.6683 #> [ CPUFloatType{4} ] #>  #> $lt2[[3]] #> torch_tensor #>  0.7921 #>  1.4544 #>  0.1958 #> -0.2346 #> [ CPUFloatType{4} ] #>  #> $lt2[[4]] #> torch_tensor #>  0.6385 #> -0.0318 #>  0.6482 #> -0.1380 #> [ CPUFloatType{4} ] #>  #> $lt2[[5]] #> torch_tensor #>  1.3797 #>  0.7916 #> -0.2373 #> -1.0011 #> [ CPUFloatType{4} ] #>  #> $lt2[[6]] #> torch_tensor #>  0.3849 #> -0.2163 #> -0.0237 #> -0.0863 #> [ CPUFloatType{4} ] #>  #> $lt2[[7]] #> torch_tensor #>  0.4032 #>  0.1983 #> -1.6133 #> -2.2197 #> [ CPUFloatType{4} ] #>  #> $lt2[[8]] #> torch_tensor #>  0.2452 #>  0.7073 #>  0.3545 #>  0.0833 #> [ CPUFloatType{4} ] #>  #> $lt2[[9]] #> torch_tensor #> -0.8829 #>  0.2502 #> -0.3007 #> -0.1676 #> [ CPUFloatType{4} ] #>  #> $lt2[[10]] #> torch_tensor #>  1.9140 #> -0.7256 #> -0.7844 #> -1.4341 #> [ 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. 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 :: integer(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-tuning","dir":"Reference","previous_headings":"","what":"Early Stopping and 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 = c(\"torchvision\", \"magick\"),   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 $shapes_out().","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"credit","dir":"Reference","previous_headings":"","what":"Credit","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Part documentation copied adapted documentation torch.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"parameters","dir":"Reference","previous_headings":"","what":"Parameters","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"output_size :: integer() target output size. single number.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"internals","dir":"Reference","previous_headings":"","what":"Internals","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"Calls nn_adaptive_avg_pool1d() training.","code":""},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"super-classes","dir":"Reference","previous_headings":"","what":"Super classes","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp -> mlr3torch::PipeOpTorch -> mlr3torch::PipeOpTorchAdaptiveAvgPool -> PipeOpTorchAdaptiveAvgPool1D","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/mlr_pipeops_nn_adaptive_avg_pool1d.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"1D Adaptive Average Pooling — mlr_pipeops_nn_adaptive_avg_pool1d","text":"mlr3pipelines::PipeOp$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,.,.) =  #>   0.1683  0.9521 -0.3007 -0.6531 -0.5466 -0.2548 #>   0.8280  0.7439 -1.1441 -0.2175  0.6100 -1.8303 #>   0.6169  1.3783  0.2681 -0.2965 -0.6230  0.1022 #>  -0.4736  0.9308  0.0779  0.6800 -0.6026  0.1194 #>  -0.7146  0.7433 -0.7192  0.7677  0.5538  0.4407 #>   0.3748  0.1611 -0.8856 -0.3067 -0.1230  0.5929 #>  #> (2,1,.,.) =  #>  -0.1821 -0.2982  1.1122  0.6025 -0.0759 -0.8357 #>  -0.1558 -0.9088 -0.7307  0.6930 -0.1729  0.8293 #>   0.4416 -0.2946  0.4691  0.5564 -0.7042 -1.1576 #>   0.5102  0.1238 -0.2319  0.7806 -1.2194 -1.4314 #>   1.5757  0.0112 -1.0029  0.4919  0.4147 -0.0619 #>   1.1526  0.4710 -0.9762 -1.2955 -0.1328  0.0883 #>  #> (1,2,.,.) =  #>  -0.2915 -0.8742  0.2805 -0.3131 -0.0847  0.3623 #>   0.8970  1.4281  1.1198 -0.4452 -0.3298 -0.7621 #>  -0.6495  0.3189  0.3577  0.3509  0.4461  0.3858 #>  -0.2846 -1.6956 -0.5109  1.1558  0.6108  0.1983 #>   0.3220  0.4607  0.1320  0.0521 -1.6308 -0.7962 #>   0.9437  1.8788 -0.5748  0.0037 -0.1264 -0.8996 #>  #> (2,2,.,.) =  #>   0.2720  0.9749 -1.7862  0.7320 -0.4558 -0.2841 #>   0.6081  0.0487  0.1334 -0.1368 -0.8086 -0.4014 #>   0.7191  0.0231 -0.3324 -0.8849 -0.8758 -0.2848 #>  -1.2964  0.1189  0.2938  0.0605 -0.3840 -0.5428 #>   0.3190  0.1164 -0.5326  1.0797  0.4521 -0.6165 #> ... [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 #> -0.0813 -0.0115 -0.3071 -0.2139 #>  0.6080  0.2067  0.9018  0.4727 #> -0.5266 -0.1952 -0.5947 -0.2588 #> [ CPUFloatType{3,4} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> 0.01 * #>  0.1392  0.0041  1.6147  0.7540 #>   3.0754  0.3983  5.4061  1.4248 #>   3.0210  0.4019  4.2777  0.8787 #> [ CPUFloatType{3,4} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$step #> [1] 3 #>  #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -0.0619 #>  0.1498 #> -0.0879 #> [ CPUFloatType{3} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #> 0.001 * #>  0.6434 #>  1.4496 #>  0.9899 #> [ 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] 0.053 #>  #> $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] \"34a6beb27c3181c3\" #>  #> $feature_names #> [1] \"Petal.Length\" \"Petal.Width\"  \"Sepal.Length\" \"Sepal.Width\"  #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.21.1’ #>  #> $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 #> -8.7515 -70.6545 -155.1045 -6165.2896 -82.2042 -85.4106 -3880.7810 -405.2719 -7.2561 -78.5860 #> [ CPUFloatType{1,10} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> Columns 1 to 6 4.7060e+00  2.7571e+02  1.3329e+03  2.1144e+06  3.7363e+02  4.0293e+02 #>  #> Columns 7 to 10 8.3214e+05  9.1789e+03  3.2087e+00  3.4837e+02 #> [ CPUFloatType{1,10} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$step #> [1] 2 #>  #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -23.3342 #> [ CPUFloatType{1} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #>  30.1836 #> [ 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] 0.039 #>  #> $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] \"6fbb93eed42adb31\" #>  #> $feature_names #>  [1] \"am\"   \"carb\" \"cyl\"  \"disp\" \"drat\" \"gear\" \"hp\"   \"qsec\" \"vs\"   \"wt\"   #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.21.1’ #>  #> $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 #>  0.2515 #> -0.0529 #>  0.0081 #>  0.2387 #> -0.0763 #>  0.0844 #>  0.2905 #> -0.1252 #> -0.0067 #>  0.1168 #> -0.0734 #>  0.3246 #> -0.1031 #>  0.5290 #> -0.0672 #>  0.0823 #> [ 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: 0x55dfb6df1b30> ##   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$generator #> <nn_l1_loss> object generator #>   Inherits from: <inherit> #>   Public: #>     .classes: nn_l1_loss nn_loss nn_module #>     initialize: function (reduction = \"mean\")  #>     forward: function (input, target)  #>     clone: function (deep = FALSE, ..., replace_values = TRUE)  #>   Private: #>     .__clone_r6__: function (deep = FALSE)  #>   Parent env: <environment: 0x5584a92d3ea8> #>   Locked objects: FALSE #>   Locked class: FALSE #>   Portable: TRUE mlp_l1 = lrn(\"regr.mlp\", loss = l1) mlp_l1$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 #> -1.3766 -0.5136  0.3212 #> -0.1381  0.5962  0.2744 #> [ CPUFloatType{2,3} ] module_1 = nn_linear(in_features = 3, out_features = 4, bias = TRUE) output = module_1(input) output #> torch_tensor #>  0.2026 -0.6605  0.1249  0.6521 #>  0.4501 -0.2476  0.1827  0.3494 #> [ 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 #>  0.3464  0.1966  0.4570 #>  0.3344  0.1942  0.4714 #> [ 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 #>  0.2026 -0.6605  0.1249  0.6521 #>  0.4501 -0.2476  0.1827  0.3494 #> [ 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 #>  0.3464  0.1966  0.4570 #>  0.3344  0.1942  0.4714 #> [ 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 #> $module_list #> An `nn_module` containing 31 parameters. #>  #> ── Modules ───────────────────────────────────────────────────────────────────── #> • 0: <nn_linear> #16 parameters #> • 1: <nn_sigmoid> #0 parameters #> • 2: <nn_linear> #15 parameters #> • 3: <nn_softmax> #0 parameters graph_module(input) #> torch_tensor #>  0.3464  0.1966  0.4570 #>  0.3344  0.1942  0.4714 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ]"},{"path":[]},{"path":"https://mlr3torch.mlr-org.com/dev/articles/internals_pipeop_torch.html","id":"modeldescriptor","dir":"Articles","previous_headings":"Building Torch Models for Tasks using PipeOpTorch","what":"ModelDescriptor","title":"Internals","text":"PipeOpModule represents nn_module fixed specific tensor shape hyperparameters. constructing neural network using operators, one take care output shape operations match input shapes following operations. complete Graph matching PipeOpModules can constructed using operators mostly inherit PipeOpTorch, making use ModelDescriptor class. ModelDescriptor class contains Graph (mostly) PipeOpModules information. PipeOpTorch transforms ModelDescriptor adds PipeOpModules Graph. ModelDescriptors always build Graph specific Task. easiest way initialize proper ModelDescriptor use appropriate PipeOpTorchIngress given datatype. use PipeOpTorchIngressNumeric, used numeric data. ModelDescriptor S3 object contains Graph, information generate data ($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 #>  0.2228 -0.2205 -0.0451  0.1615 #>  0.2096  0.2013 -0.4630 -0.3194 #> [ 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 #>  2.5036  3.6613 -0.2827 -0.3227 #>  2.2548  3.3180 -0.4614 -0.4464 #>  2.2830  3.3477 -0.3081 -0.3362 #> [ 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 #>  0.2415  0.2577  0.5008 #>  0.2506  0.2512  0.4982 #> [ 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 #> -0.3686 #> -0.3300 #> [ CPUFloatType{2,1} ][ grad_fn = <AddmmBackward0> ] #>  #> $nn_softmax.output #> torch_tensor #>  0.2793  0.2349  0.4858 #>  0.2875  0.2328  0.4798 #> [ CPUFloatType{2,3} ][ grad_fn = <SoftmaxBackward0> ] iris_module$graph$pipeops$linear1$.result #> $output #> torch_tensor #>  0.7996 -2.4717 -1.8110 -2.8502 #>  0.5982 -2.1502 -1.6179 -2.5826 #> [ CPUFloatType{2,4} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$linear3$.result #> $output #> torch_tensor #>  0.2892  0.8388  0.0890 -0.1588  0.4280 #>  0.2892  0.8388  0.0890 -0.1588  0.4280 #> [ CPUFloatType{2,5} ][ grad_fn = <AddmmBackward0> ] iris_module$graph$pipeops$nn_merge_cat$.result #> $output #> torch_tensor #>  0.7996 -2.4717 -1.8110 -2.8502  0.2892  0.8388  0.0890 -0.1588  0.4280 #>  0.5982 -2.1502 -1.6179 -2.5826  0.2892  0.8388  0.0890 -0.1588  0.4280 #> [ 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$ingress,   optimizer = adam,   callbacks = history,   loss = xe )  lr_sequential #> <LearnerTorchModel[classif]:classif.model>: Torch Model #> * Model: - #> * Parameters: device=auto, num_threads=1, num_interop_threads=1, #>   seed=random, eval_freq=1, measures_train=<list>, #>   measures_valid=<list>, patience=0, min_delta=0, opt.lr=0.02 #> * Validate: NULL #> * Packages: mlr3, mlr3torch, torch #> * Predict Types:  [response], prob #> * Feature Types: logical, integer, numeric, character, factor, ordered, #>   POSIXct, lazy_tensor #> * Properties: featureless, hotstart_backward, hotstart_forward, #>   importance, internal_tuning, loglik, marshal, missings, multiclass, #>   oob_error, selected_features, twoclass, validation, weights #> * Optimizer: adam #> * Loss: cross_entropy #> * Callbacks: history lr_sequential$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.09428b.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.09428b.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.09428b.x        <list>   poly.x              #>        poly.x <<UNTRAINED>>          nop.09428b.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.1567d7.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.1567d7.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/RtmpO2o111/file284f69a24b0c, 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(36)> #>                          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:              activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:         activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:                   shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:                  opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 29:               opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 30:                 opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 31:        opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 32:             opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 33:          loss.reduction ParamFct    NA    NA       2           mean #> 34:      cb.checkpoint.path ParamUty    NA    NA     Inf <NoDefault[0]> #> 35:      cb.checkpoint.freq ParamInt 1e+00   Inf     Inf <NoDefault[0]> #> 36: 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: <nn_relu[1]> #> 26:    <list[0]> #> 27:       [NULL] #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [NULL] #> 34:       [NULL] #> 35:       [NULL] #> 36:       [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(33)> #>                      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:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:              opt.lr ParamDbl 0e+00   Inf     Inf          0.001 #> 29:           opt.betas ParamUty    NA    NA     Inf    0.900,0.999 #> 30:             opt.eps ParamDbl 1e-16 1e-04     Inf          1e-08 #> 31:    opt.weight_decay ParamDbl 0e+00 1e+00     Inf              0 #> 32:         opt.amsgrad ParamLgl    NA    NA       2          FALSE #> 33:      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: <nn_relu[1]> #> 26:    <list[0]> #> 27:       [NULL] #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [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(33)> #>                      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:          activation ParamUty    NA    NA     Inf <NoDefault[0]> #> 26:     activation_args ParamUty    NA    NA     Inf <NoDefault[0]> #> 27:               shape ParamUty    NA    NA     Inf <NoDefault[0]> #> 28:              opt.lr ParamDbl     0   Inf     Inf <NoDefault[0]> #> 29:        opt.momentum ParamDbl     0     1     Inf              0 #> 30:       opt.dampening ParamDbl     0     1     Inf              0 #> 31:    opt.weight_decay ParamDbl     0     1     Inf              0 #> 32:        opt.nesterov ParamLgl    NA    NA       2          FALSE #> 33:      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: <nn_relu[1]> #> 26:    <list[0]> #> 27:       [NULL] #> 28:       [NULL] #> 29:       [NULL] #> 30:       [NULL] #> 31:       [NULL] #> 32:       [NULL] #> 33:       [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.d10b3c.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$iris) )() as_data_descriptor(ds2) #> <DataDescriptor: 1 ops> #> * dataset_shapes: [x: (NA,4)] #> * input_map: (x) -> Graph #> * pointer: nop.4305b8.x.output #> * shape: [(NA,4)]"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert to Lazy Tensor — as_lazy_tensor","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"Convert object lazy_tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"as_lazy_tensor(x, ...)  # S3 method for class 'dataset' as_lazy_tensor(x, dataset_shapes = NULL, ids = NULL, ...)"},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"x () Object convert lazy_tensor ... () Additional arguments passed method. dataset_shapes (named list() (integer() NULL)) shapes output. Names elements list returned dataset. shape NULL (unknown, e.g. images different sizes) first dimension must NA indicate batch dimension. ids (integer()) ids include lazy tensor.","code":""},{"path":"https://mlr3torch.mlr-org.com/dev/reference/as_lazy_tensor.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert to Lazy Tensor — as_lazy_tensor","text":"","code":"iris_ds = dataset(\"iris\",   initialize = function() {     self$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 #>  0.2337  0.5821 -0.4541 #> -0.0929 -0.0327 -0.0007 #>  0.7647  1.1990  1.1332 #> -1.2069  0.9962 -0.2510 #>  0.4641 -0.2855 -0.6539 #> -0.2911  0.3456 -0.2388 #> -1.2894  0.6156  1.3607 #> -0.2922 -2.0550  1.3359 #> -0.3164  0.3312 -1.1026 #> -1.4172 -0.1749  1.6354 #> [ CPUFloatType{10,3} ] materialize(lt1, rbind = FALSE) #> [[1]] #> torch_tensor #>  0.2337 #>  0.5821 #> -0.4541 #> [ CPUFloatType{3} ] #>  #> [[2]] #> torch_tensor #> 0.01 * #> -9.2944 #> -3.2728 #> -0.0712 #> [ CPUFloatType{3} ] #>  #> [[3]] #> torch_tensor #>  0.7647 #>  1.1990 #>  1.1332 #> [ CPUFloatType{3} ] #>  #> [[4]] #> torch_tensor #> -1.2069 #>  0.9962 #> -0.2510 #> [ CPUFloatType{3} ] #>  #> [[5]] #> torch_tensor #>  0.4641 #> -0.2855 #> -0.6539 #> [ CPUFloatType{3} ] #>  #> [[6]] #> torch_tensor #> -0.2911 #>  0.3456 #> -0.2388 #> [ CPUFloatType{3} ] #>  #> [[7]] #> torch_tensor #> -1.2894 #>  0.6156 #>  1.3607 #> [ CPUFloatType{3} ] #>  #> [[8]] #> torch_tensor #> -0.2922 #> -2.0550 #>  1.3359 #> [ CPUFloatType{3} ] #>  #> [[9]] #> torch_tensor #> -0.3164 #>  0.3312 #> -1.1026 #> [ CPUFloatType{3} ] #>  #> [[10]] #> torch_tensor #> -1.4172 #> -0.1749 #>  1.6354 #> [ 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 #>  0.2337  0.5821 -0.4541 #> -0.0929 -0.0327 -0.0007 #>  0.7647  1.1990  1.1332 #> -1.2069  0.9962 -0.2510 #>  0.4641 -0.2855 -0.6539 #> -0.2911  0.3456 -0.2388 #> -1.2894  0.6156  1.3607 #> -0.2922 -2.0550  1.3359 #> -0.3164  0.3312 -1.1026 #> -1.4172 -0.1749  1.6354 #> [ CPUFloatType{10,3} ] #>  #> $lt2 #> torch_tensor #> -0.3209 -0.9323 -0.4262  0.4701 #> -0.6289 -1.5827  1.4104 -0.9415 #> -0.6591 -1.3246 -0.3064  0.8200 #> -1.0953 -0.7571  2.1173 -0.8233 #>  0.2794 -1.1853 -1.6339  0.4039 #>  1.4075  1.4175 -0.1898 -0.2829 #> -0.2797 -0.2720  0.2012  0.6381 #> -0.1676  2.0197 -0.6833  1.2002 #>  1.7206 -0.0778  0.1583 -1.5801 #>  0.1183 -0.4630 -0.5408 -0.7640 #> [ CPUFloatType{10,4} ] #>  materialize(d, rbind = FALSE) #> $lt1 #> $lt1[[1]] #> torch_tensor #>  0.2337 #>  0.5821 #> -0.4541 #> [ CPUFloatType{3} ] #>  #> $lt1[[2]] #> torch_tensor #> 0.01 * #> -9.2944 #> -3.2728 #> -0.0712 #> [ CPUFloatType{3} ] #>  #> $lt1[[3]] #> torch_tensor #>  0.7647 #>  1.1990 #>  1.1332 #> [ CPUFloatType{3} ] #>  #> $lt1[[4]] #> torch_tensor #> -1.2069 #>  0.9962 #> -0.2510 #> [ CPUFloatType{3} ] #>  #> $lt1[[5]] #> torch_tensor #>  0.4641 #> -0.2855 #> -0.6539 #> [ CPUFloatType{3} ] #>  #> $lt1[[6]] #> torch_tensor #> -0.2911 #>  0.3456 #> -0.2388 #> [ CPUFloatType{3} ] #>  #> $lt1[[7]] #> torch_tensor #> -1.2894 #>  0.6156 #>  1.3607 #> [ CPUFloatType{3} ] #>  #> $lt1[[8]] #> torch_tensor #> -0.2922 #> -2.0550 #>  1.3359 #> [ CPUFloatType{3} ] #>  #> $lt1[[9]] #> torch_tensor #> -0.3164 #>  0.3312 #> -1.1026 #> [ CPUFloatType{3} ] #>  #> $lt1[[10]] #> torch_tensor #> -1.4172 #> -0.1749 #>  1.6354 #> [ CPUFloatType{3} ] #>  #>  #> $lt2 #> $lt2[[1]] #> torch_tensor #> -0.3209 #> -0.9323 #> -0.4262 #>  0.4701 #> [ CPUFloatType{4} ] #>  #> $lt2[[2]] #> torch_tensor #> -0.6289 #> -1.5827 #>  1.4104 #> -0.9415 #> [ CPUFloatType{4} ] #>  #> $lt2[[3]] #> torch_tensor #> -0.6591 #> -1.3246 #> -0.3064 #>  0.8200 #> [ CPUFloatType{4} ] #>  #> $lt2[[4]] #> torch_tensor #> -1.0953 #> -0.7571 #>  2.1173 #> -0.8233 #> [ CPUFloatType{4} ] #>  #> $lt2[[5]] #> torch_tensor #>  0.2794 #> -1.1853 #> -1.6339 #>  0.4039 #> [ CPUFloatType{4} ] #>  #> $lt2[[6]] #> torch_tensor #>  1.4075 #>  1.4175 #> -0.1898 #> -0.2829 #> [ CPUFloatType{4} ] #>  #> $lt2[[7]] #> torch_tensor #> -0.2797 #> -0.2720 #>  0.2012 #>  0.6381 #> [ CPUFloatType{4} ] #>  #> $lt2[[8]] #> torch_tensor #> -0.1676 #>  2.0197 #> -0.6833 #>  1.2002 #> [ CPUFloatType{4} ] #>  #> $lt2[[9]] #> torch_tensor #>  1.7206 #> -0.0778 #>  0.1583 #> -1.5801 #> [ CPUFloatType{4} ] #>  #> $lt2[[10]] #> torch_tensor #>  0.1183 #> -0.4630 #> -0.5408 #> -0.7640 #> [ 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. 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 :: integer(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-tuning","dir":"Reference","previous_headings":"","what":"Early Stopping and 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 = c(\"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 versicolor #>        9    setosa versicolor #>       10    setosa versicolor #>      ---       ---        --- #>      147 virginica versicolor #>      149 virginica versicolor #>      150 virginica versicolor"},{"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,.,.) =  #>   0.2125 -0.3882 -0.2121  0.1723  0.0384  0.4917 #>  -0.6026 -0.0511  0.8008  0.5735  1.5137  1.0633 #>  -0.2825 -0.0818 -0.7123 -0.4431  0.5773 -0.0926 #>  -0.2583  0.0694  0.5027  0.7375 -0.0705 -0.2815 #>   0.4183  1.7796  0.2513 -1.4076  1.3036 -0.8148 #>  -0.2644 -0.3981  0.8998 -0.2581 -0.5860  0.3477 #>  #> (2,1,.,.) =  #>   0.7151  0.2609  0.0981 -0.1435 -0.1861  1.0630 #>  -0.4236 -0.0667 -0.5166 -0.0989 -0.6529 -0.2784 #>   0.3545  0.3712  0.1984 -0.1119 -1.0725  0.2314 #>  -0.0875  0.5762  0.1458  0.9248  0.1061  0.9214 #>   0.2276  1.0706 -0.1771  0.3537  0.8959  0.9463 #>  -0.2926 -0.6330  0.4849 -0.1984  1.0717  0.2909 #>  #> (1,2,.,.) =  #>   1.2679 -0.7824 -0.5283 -0.2615 -0.5069  0.0035 #>  -0.3564 -0.2301  1.1684 -0.1398  0.2603 -0.6686 #>   0.0015 -0.1906  0.3344 -0.8316 -1.3696  0.6962 #>  -0.3337 -0.2110  0.5441 -0.6840  0.1694 -0.5514 #>   0.1862  0.9605  0.0916  0.1290 -1.8815 -1.3006 #>   0.1662 -0.0681  0.1811 -0.6044 -0.8783 -0.3254 #>  #> (2,2,.,.) =  #>  -0.5820  0.8655  0.3351 -0.5340  0.3202  0.0760 #>  -0.5524  0.1915  0.4468 -0.0970 -0.1182 -0.5823 #>   1.6456  0.5373  0.6732 -0.2632 -1.3037 -0.0948 #>  -0.1056 -0.2319 -0.1575  0.0060  0.3981 -0.5047 #>   1.1163  0.8038  0.6666 -0.0219 -0.1799 -0.3924 #> ... [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 #>  0.4871  0.1782  0.5019  0.1886 #> -0.2023 -0.0594 -0.2875 -0.1301 #> -0.2849 -0.1188 -0.2144 -0.0584 #> [ CPUFloatType{3,4} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> 0.01 * #>  1.6796  0.2009  3.1383  0.7971 #>   1.3630  0.1358  2.6025  0.5677 #>   2.0308  0.2634  3.2504  0.7642 #> [ CPUFloatType{3,4} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$step #> [1] 3 #>  #> $optimizer$state$`2`$exp_avg #> torch_tensor #> 0.01 * #>  7.2249 #> -4.9365 #> -2.2885 #> [ CPUFloatType{3} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #> 0.0001 * #>  8.7081 #>  7.3037 #>  8.2450 #> [ 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] 0.053 #>  #> $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] \"34a6beb27c3181c3\" #>  #> $feature_names #> [1] \"Petal.Length\" \"Petal.Width\"  \"Sepal.Length\" \"Sepal.Width\"  #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.21.1’ #>  #> $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 #> Columns 1 to 8-11.2135 -85.3135 -177.3723 -6927.4478 -95.3926 -100.3730 -4563.8008 -463.1545 #>  #> Columns 9 to 10 -8.4023 -89.0784 #> [ CPUFloatType{1,10} ] #>  #> $optimizer$state$`1`$exp_avg_sq #> torch_tensor #> Columns 1 to 6 8.0234e+00  4.0437e+02  1.7375e+03  2.6535e+06  5.0269e+02  5.5943e+02 #>  #> Columns 7 to 10 1.1598e+06  1.1890e+04  4.1586e+00  4.4203e+02 #> [ CPUFloatType{1,10} ] #>  #>  #> $optimizer$state$`2` #> $optimizer$state$`2`$step #> [1] 2 #>  #> $optimizer$state$`2`$exp_avg #> torch_tensor #> -26.8348 #> [ CPUFloatType{1} ] #>  #> $optimizer$state$`2`$exp_avg_sq #> torch_tensor #>  39.7718 #> [ 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] \"6fbb93eed42adb31\" #>  #> $feature_names #>  [1] \"am\"   \"carb\" \"cyl\"  \"disp\" \"drat\" \"gear\" \"hp\"   \"qsec\" \"vs\"   \"wt\"   #>  #> $validate #> NULL #>  #> $mlr3_version #> [1] ‘0.21.1’ #>  #> $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 #>  0.0954 #>  0.1976 #>  0.2559 #>  0.2567 #>  0.1011 #>  0.1625 #>  0.1907 #> -0.2813 #>  0.1443 #>  0.1155 #>  0.0968 #> -0.1137 #>  0.1292 #>  0.0884 #>  0.1634 #>  0.1503 #> [ 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":""}]